Be eo FROM THE GRAY HERBARIUM WX F HARVARD UNIVERSITY | Edited by | ie Reed C. Rollins and Robert C. Foster == NO. CXCII THE EVOLUTION AND SYSTEMATICS OF — _ -—s LEAVENWORTHIA (CRUCIFERAE) : By : Reep C, ROLLINS soe 2 TRE TAXONOMY AND DISTRIBUTION OF BAJA ¢ CALIFORNIA, MEXICO CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY dited by Reed C. Rollins and Robert C. Foster NO. CXCII THE EVOLUTION AND SYSTEMATICS OF LEAVENWORTHIA (CRUCIFERAE) BY REED C. ROLLINS THE TAXONOMY AND DISTRIBUTION OF THE ZYGOPHYLLACEAE OF BAJA CALIFORNIA, MEXICO BY DUNCAN M. PORTER Published by THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U.S. A. Issued June 17, 1963 THE EVOLUTION AND SYSTEMATICS OF LEAVENWORTHIA (CRUCIFERAE)' REED C. ROLLINS Different kinds of higher plants differ markedly in their sensitiveness to the various features of their environment, as do organisms in general. Some species, for example, exist in a wide variety of fluctuating sites, characterized by di- verse conditions, without any obvious difficulty or radical! responsive reaction. Others, at the opposite extreme, are very sensitive to every facet of their environment and exist only under a strictly limited pattern of conditions. Undoubt- edly, nearly all imaginable combinations of intermediates occur as well. For certain types of evolutionary studies, plants closely restricted (and therefore adapted) to a rela- tively uniform set of environmental conditions are highly desirable. This is particularly true where several different evolutionary units have developed but all have retained their adaptation to the same set of edaphic conditions. Such a situation permits an examination of evolutionary patterns and trends under circumstances where the effects of certain aspects of the environment are minimal and the selective effects of other aspects are more clearly defined and readily seen. We have been fortunate in selecting a genus with many desirable attributes for detailed evolutionary and systematic studies and the present paper undertakes to set forth our findings. Leavenworthia was initially selected for detailed study because of certain obvious contrasts with Lesquerella, an- other genus of the Cruciferae, which has a number of species growing in the same general area. The most important of ‘This investigation was supported by grants G-5849 and G-15850 from The National Science Foundation. I am indebted to Dr. Otto T. Solbrig for some of the statistical calculations, for some cytological observations and for other assistance, and to Mrs. Rainer Weiss for technical assistance through much of this investigation. Mrs. Peter Ray, Mrs. Winslow Briggs, Mrs. William Allderdice and Miss Phyllis Bartlett prepared and examined material for chromosome counts dur- of several populations. 4 REED C. ROLLINS these contrasts was that the four species of Leavenworthia of central Tennessee (cf. Map 2) are sympatric in their occurrence, often being present in exactly the same habitat, whereas the species of Lesquerella of the same area are allo- patric in their distribution (Rollins, 1955). A second im- portant contrast was that the Central Basin species of Lesquerella hybridized readily (Rollins, 1957) while those thia showed no evidence of natural hybridiza- tion even though the plants of several species grew in close proximity to each other. It was assumed and subsequently shown experimentally that well established genetic barriers exist between the Tennessee species of Leavenworthia. The auriculate species of Lesquerella of Tennessee have a uni- form chromosome number, n= 8. At the outset of this study, on the other hand, two chromosome numbers were known in Leavenworthia, n= 11 and n= 15, One of our major objectives has been to see, if possible, what the conditions and circumstances were that brought about speciation in these two genera of the same family, in the same area but with such different genetic, cytological and geographical characteristics. Furthermore, we were in- terested in speciation within Leavenworthia itself. It was hoped that some insight into the conditions favoring specia- tion would be gained and that the pattern of evolution within the genus could be satisfactorily elucidated. From the work of Baldwin (1945) it was known that plants of Leavenworthia stylosa are self-incompatible and that this contrasts with the situation of self-compatibility found in plants of L. exigua, L. torulosa and L. uniflora. Our desire to exploit this fact to determine whether self- incompatibility is positively correlated with a higher intra- population variability than is found in populations where self-compatibility prevails has not been fully realized. How- ever, we discovered populations characterized by self-com- patibility in otherwise self-incompatible species together with a neat correlation of anther position and pollen dis- charge. These facts are brought out below. GENERAL OBSERVATIONS Leavenworthia is a small distinctive genus of the Cruci- ferae which has an unusually close adaptation to cedar LEAVENWORTHIA 5 glades exhibiting certain physical characteristics. The most important of these characteristics appears to be a combina- tion of shallow soil depth and dolomitic limestone thinly bedded in such a way as to maintain a temporary high mois- ture content at or very near the surface. Often the lime- stone is horizontally bedded and is covered with but a few inches of soil or it may be completely barren of soil in some spots. Such situations are extremely wet during late winter and early spring and become very dry during the summer months. Thus, a set of extremes of moisture stress is pre- sented to the plants growing there. The severity of these conditions is brought out by Freeman (1933). These areas vary as to size and may be as small as a few square meters or as large as a square kilometer or more. Quarterman (1950) characterizes such areas in Tennessee as open glades. The high moisture required by Leavenworthia is provided at some sites by slightly tilted limestone strata that bring water to the surface. This type of habitat is common in the ozark region of Missouri and is found at the one location where if worthia occurs in Texas, at San Augustine. An ex- cellent chink tye including photographs, of some dolomitic glades of east-central Missouri is given by Erickson, Bren- ner and Wraight (1942). It is clear that the glades of all areas where L thia grows are not completely com- parable, particularly as to slope and relief. Those of Ten- neessee, Alabama and Oklahoma occupy relatively flat areas, while many of the others are on slopes of varying degrees of steepness. Most of the Lea vorthia glades appear to be on dolomitic limestone of Ordovician age. Harper (1926, p. 49) suggests that there are no true cedar glades in north- ern Alabama but he may not have considered the principal Leavenworthia areas since he specifically mentioned only the Coosa Valley. It should be emphasized that high soil moisture content in late winter and early spring appears to be an important feature of suitable sites for the abundant occurrence of Leavenworthia. There are many places in Alabama, Ten- nessee, Kentucky and Missouri where all other conditions, except that of soil moisture, appear to be met but the plants do not occur at these sites. Every species of the genus (with slight deviations in L. uniflora) seems to require approxi- 6 REED C. ROLLINS mately the same set of conditions, whether it be in Kentucky, Tennessee, Alabama, southeastern Oklahoma or eastern Texas. This leads to the view that certain of the physiologi- cal requirements are genus-wide and therefore must have been rather solidly fixed into its genetic make-up early in its evolutionary development. This close adaptation to a rather narrow and rigid set of edaphic conditions appears to have been an important factor influencing the direction and extent of evolution within the genus. All species of Leav thia bloom early in the spring. Known flowering dates range from late February into May. The time of flowering varies from year to year, depending upon temperature, moisture and the severity of winter freez- ing. The plants are classed as winter annuals. Seed germ- ination takes place from September onward and the individ- ual plants persist through the winter months as rosettes of radical leaves. The most recent study of Leavenworthia, that of Baldwin (1945), was primarily cytogeographic. In it, he concluded that the genus consisted of four well defined species. Since then, I have shown (1956) that L. awrea is confined to re- stricted areas in southeastern Oklahoma and extreme east- ern Texas, and that the plants of Tennessee previously called L. aurea really represent a fifth species, L. exigua. As shown below, there are two additional species, L. crassa and L. alabamica and several varieties heretofore unrecog- nized. As known at present, the genus consists of eleven taxa. A taxonomic treatment of the genus is given below. GENERIC RELATIONSHIPS The direct relationships of L thia to other genera of the Cruciferae are not established. From such treatments of the family as those of von Hayek (1911) and Schulz (1936), where Leavenworthia and Cardamine are placed next to or close to each other in the tribe Arabideae, it is usually inferred that these two genera are closely related. The same conclusion might be drawn from the fact that Michaux (1803) described the first known species of Leaven- worthia in the genus Cardamine. However, I cannot see any real evidence from the morphology of Leavenworthia to suggest a closer link to Cardamine than to several other LEAVENWORTHIA 7 possible genera. The fruits of Cardamine are singularly different from those of Leavenworthia and the seeds of the latter are unlike any found in Cardamine. The straight or nearly straight embryo of Leavenworthia is unusual in the Cruciferae. Leavenworthia does not tie in directly with Cardamine, or any other American genus of the Cruciferae, for that matter, in a way that would provide significant in- formation about the origin of any of the present day species. Because of the lack of closely related genera, the present study has dealt only with the species of Leavenworthia and only in a casual way with intergeneric considerations. Per- haps the suggestion of Torrey (1837, p. 94) that “upon the whole I consider it more nearly related to Selenia than to any other known genus” is still pertinent. MATERIALS AND METHODS The present study of Leav thia was begun in 1955. At that time, we started collecting herbarium specimens and taking notes on populations in the field. Additional collec- tions, including cytological fixations for chromosome number determination, were made in the spring of 1956. The fol- lowing year, both flowering and fruiting materials were collected, together with packets of seeds. Intensive work on the genus, involving greenhouse and garden cultures, was started in the winter of 1959. Several approaches have been focused toward providing a maximum understanding of the genus. These included detailed studies of wild populations ; random seed sampling of these same populations to produce cultures of plants in the greenhouse and experimental gar- den, where careful comparative measurements were made; making numerous interspecific pollinations together with an analysis of the fertility of resulting hybrids through pollen analysis and/or progeny testing; and study of herbarium specimens to provide a historical and geographical basis to the overall findings. In establishing living cultures, seeds were germinated in petri dishes and the young seedlings transferred to three- inch clay pots or to planting bands in flats. In the green- house, the plants were brought to maturity in three-or-four- inch pots. The planting bands were used to transfer young plants to the experimental garden. In general, growth was 8 REED C. ROLLINS satisfactory, although not outstanding, in Cambridge soils. Vermiculite, leaf-mold or peat moss was added to soil to make up the greenhouse mixture. Self-compatibility tests were made by hand selfing on both isolated and bagged plants. Seed-set or lack of it was used as the criterion as to whether the plants were self- compatible or not. Field sampling was done in an arbitrary manner. The same procedure was followed in taking the flowers and fruits that provided the measurement data of both Tables 3 and 4. This consisted of walking in a straight direction across a population. At each step, a single fully opened flower was picked from the plant nearest the toe of the foot. Similarly, in taking fruits, a mature, fully filled, green fruit, just ap- proaching the onset of ripening, was picked. The flowers were fixed immediately in FAA for later measurement with the use of a binocular dissecting scope. The fruits were measured as soon as they were accumulated, before any appreciable wilting had begun. Under these circumstances, the fruits were turgid and it was efficient to use a thickness gauge for width, thickness and style length measurements. The silique length measurement was taken with a millimeter rule. In sizing up a population for sampling, the most import- ant consideration was always the appropriateness of the maturity stage of the material to be gathered. The next step was to determine the approximate extent of more or less contiguous plants and to visually mark the area into a circle. Each sampling originated at a different point on the circumference of the circle and extended to an opposite point on the same circumference. Sometimes several such marches, always originating at a different point, were re- quired to produce the total sample for a given population. By following this procedure, a particular area of a given population was sampled only once. CYTOLOGY Baldwin’s (l.c.) extensive reports of chromosome num- bers of four species of Leavenworthia have been confirmed by our findings. As shown in Table I, where only new counts are reported, L. uniflora, L. torulosa and L. stylosa have 15 LEAVEN WORTHIA TABLE 1. CHROMOSOME COUNTS OF LEAVENWORTHIA’ SPECIES LOCATION ca ALABAMA, — Franklin Co.: 9 mi. e. Russellville 7 mi. e. Russellville Lawrence Co.: 6 mi. w. Moulton 3 mi. s. Hatton 4 mi. w. Moulton 14 mi, e. Landersville mi. s. w. Moulton var. brachystyla ALABAM 5 mi. w. Falkville L. aurea OKLAHOMA. —Choctow Co.: 1 mi. w. Ft. Towson McCurtain Co.: 4.7 mi. n. Idabel 7.3 mi. n. w. Idabel ALABAMA. — Morgan Co.: 14 mi. n. Danville 14 mi. n. Danville 2 mi. e. McKendry 2.5 mi. w. Danville 5 mi. w. Falkville Near Harris Creek var. elongata ALABAMA. — Morgan Co.: 6 mi. sw. Falkville ESSEE. — Marshall Co.: 1 mi. n. Chapel Hill 2 mi. s. icici Hill Wilson Co. 3 mi. s. Gladeville GEORGIA. — Catoosa Co.: 1 mi. e. Ft. Oglethorpe COLLECTOR & NUMBER Rollins 55186° Rollins & Chambers 57135 Rollins & Channell 5631 Rollins & Channell 5642 Rollins & Chambers 5712 Rollins & Chambers 5713 Rollins 5929 Rollins 5926 Rollins 5932 Rollins 5582 Rollins & Chambers 5774 Rollins & Chambers 5766 Rollins & Chambers 5771 24 ie Rollins & Chambers 5721 Rollins & Chambers 57141 Rollins 5925 Rollins 5931 Rollins et al 6015 Rollins 6122b Rollins & Chambers 57139 Rollins et al. 6021 Rollins & Chambers 5731 Rollins & Chambers 57143 Rollins & Chambers 57149 Rollins & Chambers 5704 11 2Voucher specimens with corresponding collector and collector’s number are deposited in the Gray Herbarium. 10 REED C. ROLLINS Table 1. (Continued) SPECIES LOCATION Walker Co.: 7 mi. s. Ft. Oglethorpe var. lutea ALABAMA. — Jefferson Co.: mi. n. e. McCalla var. laciniata KENTUCKY. — Bullit Co.: 6 mi. e. of Sheperdsville L. stylosa TENNESSEE. — Davidson Co.: 3 mi. n. Lebanon 3 mi. s. e. Lebanon 5 mi. e. Lebanon L. torulosa KENTUCKY. — Warren Co.: 12 mi. n. U.S. 68 on State Rt. 1083 TENNESSEE. — Bradley Co.: 5 mi. e. Cleveland Davidson Co.: 14 mi. s. e. Nashville Giles Co.: 3.5 mi. s. w. Pulaski Marshall Co.: 7 mi. n. Lewisburg 4 mi. s. w. Cornersville 2 mi. s. Chapel Hill 12 mi. w. Lebanon Watertown 2 mi. s. Norene L. uniflora TENNESSEE. — Bedford Co.: ville 1.5 mi, s. Lewisburg COLLECTOR & NUMBER Rollins & Chambers 5705 Rollins & Channell 5643 Rollins 6201 Rollins 55161 Rollins & Channell 5615 Rollins & Channel] 5611 Rollins & Channell 5651 Rollins 5902 Rollins & A 5610 Rollins 591 Rollins et ey 6001 Rollins & Channell 5906 Rollins 5901 Rollins & Channell 5620 Rollins & Chambers 5723 Rollins & Channell 5625 Rollins & Chambers 57142 Rollins & Channell 5616 Rollins & Chambers 57145 Rollins & Channell 5908 Rollins & Channell 5911 Rollins & Channell 5912 Rollins & Channell 5660 Rollins & Chambers 5726 LEAVENWORTHIA ll Table 1. (Continued) SPECIES LOCATION COLLECTOR & NUMBER n Rutherford Co.: mi. s. Murfreesboro Rollins 55298 15 13 mi. e. Murfreesboro Rollins & Channell 5651 15 pairs of chromosomes. The counts for L. exigua aren = 11. This is the same as Baldwin’s number for the plants of this species although the number was reported by him under the name of L. aurea. Both Le thia crassa and L. alabamica have 11 pairs of chromosomes. Baldwin did not investigate material of either of these species nor did he have true Leavenworthia aurea, which usually has 24 pairs of chromosomes. The chromosomes of Leavenworthia are relatively small and difficult to handle, particularly where the number is rather high, as in L. aurea. No less than forty-seven differ- ent plants, representing several independent accessions, were studied critically in determining the chromosome number of the latter species. We are convinced that n = 24 is the usual number in this species but the possibility that some plants have n = 23 or n = 22 could not be wholly ruled out. A very different picture of chromosome number relation- ships in Leavenworthia from that given by Baldwin (I. c.) emerges from our studies. His assumption that the three species (L. stylosa, L. torulosa, L. uniflora), with 15 pairs of chromosomes, are hexaploids must be seriously questioned in view of the fact that three other species (L. alabamica, L. crassa, L. exigua) have 11 pairs. Actually, the presence of n= 11, n= 15 andn = 24 in different species of a genus as small and closely knit morphologically and physiologically as L thia is difficult to explain on the basis of ordi- nary polyploid relationships. In my estimation, it is not profitable to speculate as to the fundamental number in this genus or to guess as to what possible ploidy level we may be dealing with in any given species. The facts are that excel- lent pairing of chromosomes is regularly found in L. alabam- ica, L. crassa and L. exigua, the three species with n = 11. A similar statement cannot be made concerning the three n= 15 species, or L. aurea, n= 24. Rather frequent ir- regularities were found in L. stylosa and were particularly prevalent in L. aurea where “multivalent tendencies” were 12 REED C. ROLLINS repeatedly seen in late prophase stages of meiosis. How- ever, these irregularities were probably of a transient nature because the expected effects of the observed irregularities, such as a reduction in the quality of pollen, did not show up in any of the species. All seven species behave repro- ductively as diploids. On other grounds as well as chromo- some number, it is suggested that L. aurea is a polyploid. This much seems certain. As mentioned above, the chromosomes of L worthia are small. We could not work with the morphology satis- factorily but two sizes of chromosomes are evident in the meiotic preparations of certain species. This is particularly noticeable in L. stylosa (as was also indicated by Baldwin 1.c.) where three of the fifteen chromosomes are consider- ably smaller than the other twelve. One of the three small chromosomes is somewhat smaller than the other two. Sim- ilarly, L. torulosa has three chromosomes smaller than the other twelve, with one of the three smaller than the other two. In L. alabamica, one of the eleven chromosomes is smaller than the other ten and in L. exigua there is one smaller chromosome as compared to the other ten of its chromosome complement. Preparations of other species did not show similar size differences. However, it should be clearly stated that this aspect of the study was tangential to the main emphasis and was not pursued intensively for its own sake, Some items of evidence from cytology bear directly on considerations of the direction in which the chromosome number shift may have taken place in the genus. These are the small chromosomes, more numerous in the n = 15 spe- cies than in the n= 11 species, the more frequent irregu- larities, including “stickiness,” in the n= 15 species as compared to the n = 11 species, and the suggestion of eleven homologous chromosomes between the eleven-paired L. exi- gua and the fifteen-paired L. stylosa (see below under inter- specific hybridization). We interpret these as minor indi- cations that the chromosome number shift may have been in the direction of the higher number. One thing seems un- equivocal, the shift from n = 11 to n = 15, or the reverse if that were the case, was successful only once during the evo- lutionary development of the genus. However, there is no LEAVENWORTHIA 13 conclusive evidence to support the idea that the direction of evolution has been from the L. crassa — L. alabamica complex toward the higher chromosome numbered species. It is quite possible that the presently known chromosome numbers are not the same as those of the progenitors of the modern species. If this latter is the case, then we must as- sume that extinction of species played a role in permitting the emergence of the rather unusual chromosome number relationships now to be found in the genus. As suggested above, the high chromosome number of n — 24 found in L. aurea suggests a polyploid origin for that species. In view of the close morphological resemblance between L. exigua var. lutea and L. aurea, it is not unlikely that var. lutea or a similar plant figured in the origin of L. aurea in some manner. There is no evidence that var. lutea itself produced L. awrea by chromosome doubling and the addition of extra chromosomes and this is not suggested but it may be involved as one of the taxa in such a process. Grant (1956) has made the point that polyploidy in annuals or short-lived perennials is frequently associated with auto- gamy, particularly where hybridization occurs. Autogamy is present in both var. lutea and L. aurea. INTERSPECIFIC HYBRIDIZATION In marked contrast to the situation in the auriculate- leaved species of Lesquerella of the same area (Rollins, 1957), the species of Leavenworthia of the Central Basin of Tennessee do not hybridize. The four species, L. exigua, L. stylosa, L. torulosa and L. uniflora, occur on the same site in numerous places (cf. Map 2), often within inches of each other. I have repeatedly examined the plants of all four species where they are intimately intermingled and in no instance have I been able to detect evidences of inter- specific hybridization. Results of attempted crossings be- tween these four species were similarly negative, with one exception, as shown in Fig. 1. The broken lines of this cros- sing polygon indicate that no hybrids were formed from reciprocal pollinations of L. stylosa, L. torulosa and L. uniflora. However, when pollen of L. exigua was carried to the flower stigmas of L. stylosa, some hybrids resulted. These hybrid plants, though vegetatively vigorous, were 14 REED C. ROLLINS Fic. 1. Diagram showing attempted crosses between the species of Leavenworthia. rosses were attempted reciprocally. i i f each i ut these proved to be sterile. The thick solid line indicates fertile successful in producing F, progenies, found to be pollen sterile and would not produce viable seeds in attempted back crosses to the parental species. In other crosses, L. crassa < L. aurea; L. stylosa X L. aurea; L. alabamica X L. awrea; and L. exigua X L. aurea, vege- tatively vigorous hybrids were produced but, in each case they were pollen sterile and both intercrossing within hy- brid families and back crossing to the parental species were unsuccessful. With the exceptions of the L. crassa X L. aurea and L. exigua X L, stylosa crosses, it should be noted that in all the rest, hybrids were produced only when pollen of the polyploid L. awrea was carried to the other species. The unidirectional success with polyploid pollen is not un- expected in view of the frequency of similar results in other plants. A cytological analysis of various crosses involving L. LEAVEN WORTHIA 15 curea revealed marked signs of hybridity and sterility. In crosses of L. alabamica and L. aurea, there was considerable fluctuation in the amount and degree of irregularity from one preparation to the next. Although pairing was irregular and the number of pairs was difficult to determine because of the small size of the chromosomes, the number of pairs centered around a mode of eleven. Univalents were always conspicuous and their number fluctuated from 5 to more than 11. These sometimes occurred within the area of the metaphase plate but frequently they were well outside of it. There was evidence of multivalent associations of the chromosomes but these could not be interpreted with cer- tainty as to whether they were trivalents, tetravalents, etc. Both tetrads and pollen were found to be irregular. On the basis of 500 observations of the appropriate stage of L. alabamica X L. aurea F, hybrids, the following results were obtained: monads .8 to 1.6%; diads 0-6%; triads 1.2 to 2.4% and tetrads 85 to 90%. Similar overall cytological results were obtained from an analysis of F, hybrids between L. crassa and L. aurea, The only additional observation was that a more marked asym- metrical distribution of the chromosomes appeared to take place. However, it is probable that this does not signify a real difference between the patterns of irregularity of the two interspecific crosses. In both crosses the tendency to form about eleven bivalents suggests a loose homology be- tween the genomes of L. alabamica and L. crassa on the one hand and a portion of the genome of L. aurea on the other. No definite cytologically readable figures could be found in hybrids of two different crosses between L. aurea and L. stylosa. In a few instances univalents were seen but in most fixations the pollen mother cells were shrunken and devoid of contents. In many flowers, the anthers were not fully developed and in some instances the anthers were com- pletely abortive. In the crosses between L. aurea and L. exigua, appropriate stages of meiosis rarely occurred. Mei- otic irregularity was evident in the one collection that yield- ed good stages where univalents, bivalents and multivalents were observed. The cytological picture in F, hybrids of crosses between L. exigua and L. stylosa is one of irregularity. Univalents 16 REED C. ROLLINS and multivalents are nearly always present. Three, four cr sometimes more univalents are usual. An inversion bridge with fragment was seen quite often. In one p.mce., eleven bivalents and four univalents were seen. This pro- vides the thinnest of evidence that eleven of the fifteen chromosomes of L. stylosa have remote homologues in L. exigua. The F, hybrids of crosses between L. crassa and L. ala- bamica are cytologically in complete contrast with the other interspecific hybrids. The meiotic process is not wholly devoid of irregularities, but these are rather minimal and seem to pertain to timing rather than to chromosome in- compatibility. Eleven bivalents are regularly formed and although some “stickiness” and “lagging” occasionally oc- cur, these features are relatively infrequent. From all points of view, the one successful interspecific cross was that of L. alabamica X L. crassa. In this case, fertile F, hybrids were produced reciprocally in several com- binations involving collections from different localities. Fur- thermore, subsequent generations produced good pollen and viable seed. L. alabamica and L. crassa are strikingly dif- ferent, particularly with respect to the morphology of their fruits. These differences, as brought out below where the species are considered from the taxonomic point of view, characterize the species throughout their geographic ranges and the only “break-down” of these distinctive features is in a few small zones where L. crassa var. crassa and L. ala- bamica var. brachystyla are in contact, and several other areas where L. crassa var. elongata and L. alabamica var. brachystyla come together in cultivated fields. Under pre- sent conditions, where cultivated fields and pastures have largely replaced the cedar glades. resulting in the destruc- tion of the original vegetation, natural interspecific hybrid- ization is actively taking place and hybrid swarms are to be found here and there in the area, ranging from a few miles south to a few miles east of McKendry, Morgan County, Alabama. In addition to the experimental production of hybrids, in June 1959 four plants of L. crassa var. crassa and four plants of L. alabamica var. alabamica were placed inter- spaced in a single row in my home garden in Wellesley, LEAVENWORTHIA 17 Massachusetts, completely isolated from all other Leaven- worthia plants. Lively insect visitation among the eight plants was observed during the flowering period. The seeds from these eight plants were sown at random in an area 1 x 2 meters in size. The seeds were permitted to germinate naturally, which occurred during September and October, 1959. From that time until the present, March 1963, the same plot has been devoted to the natural growth of Leaven- worthia. Numerous vigorous plants have resulted from the seed shed naturally each year. Other than weeding and oc- casional watering, no special care has been given to the “vol- unteer” plants that have come up each fall. These “rosette” plants have over-wintered very well and have produced vigorously flowering plants each spring. The first progeny of the initia] eight plants consisted of a mixture of F, hybrids and the parental species L. crassa and L. alabamica. The numbers of plants of each species exceeded those of the interspecific hybrids in this generation. However, in each succeeding generation, the relative propor- tion of hybrids has increased. After the collection and anal- ysis of additional data from this “natural” experiment, the information will be separately presented. Three facts of interest are: (1) interspecific hybridization between L. crassa and L. alabamica under quasi-natural conditions pro- duces a hybrid swarm; (2) some species of Lea thia are capable of surviving at least for several years without any substantial nrotection or care, far outside of their natu- ral geographical range; (3) these particular species and their hybrid progenies are very poor competitors with other plants. The latter point was suspected for Leavenworthia from field observations. In the plot described above, it was tested bv leaving one corner unweeded during the spring of 1962. The principal weed was Poa annua. Well established plants of L worthia were killed out completely by Poa in the entire area where it was permitted to grow unchecked. Over thirty Le ‘thia plants died in less than two months in the area invaded by Poa annua, while over three hundred plants grew normally in the rest of the plot. THE BREEDING AND REPRODUCTIVE SYSTEMS Self-incompatibility predominates in L. stylosa, L. crassa and L. alabamica. The other species are self-compatible and 18 REED C. ROLLINS vary somewhat as to the amount of autogamy that takes place. All of the populations of L. stylosa tested were self- incompatible and this result agrees with the work of Bald- win (1. c.). L. crassa var. crassa and L. alabamica var. alabamica are also largely self-incompatible insofar as they have been investigated by me. However, there is an indi- cation from the work of David Lloyd that some populations of L. alabamica var. alabamica are at least partially self- compatible. Of considerable significance is the fact that the tested populations of both L. crassa var. elongata and L. alabamica var. brachystyla are highly self-compatible. From having worked with growing material produced from seeds of a number of wild populations, we have no reason to doubt that self-compatibility largely characterizes these two va- rieties. The shift from self-incompatibility to self-compati- bility, accompanying the evolution of distinct varieties in both species, is significant in itself but it is particularly so when considered together with the fact that such a shift is a general feature of the genus. The switch in breeding sys- tem has occurred in the 15-paired chromosome group and in the 11-paired group. Within the 11-paired chromosome group of species, it has taken place within two distinct spe- cies and there is a strong probability that it occurred inde- pendently of these in the line leading to L. exigua and L. aurea. If the development of self-compatibility in the spe- cies L. crassa and L. alabamica took place relatively more recently than in the two main evolutionary lines of the genus, as we have good reason to believe, then self-com- patibility came about independently four different times. Leaving out all supposition, the evidence is incontrovertable that self-compatibility arose independently at least three times. This event probably took place many more than four times during the course of the evolution of the genus to its present state. In his survey of self-incompatibility in the Cruciferae, Bateman (1955) was impressed by the fact that both self- incompatibility and self-compatibility occur in most tribes of the family and in nearly all genera in which as many as four species had been tested. He does not cite any instance where both systems occur naturally within the same species, as is the case in L. alabamica and L. crassa. There is no LEAVENWORTHIA 19 reason known to me why the compatibility system should not differ within species as well as between species of the same genus, the latter being a well known situation in the Cruciferae. Since this is an evolutionary step that is fre- quently taken, it is perhaps surprising that it has not been previously emphasized as an intraspecific phenomenon, The question of direction of shift of the compatibility type, from self-incompatibility or the reverse, is a pertinent one. There is no intrinsic reason why the shift should not go from self-compatibility to self-incompatibility and this probably cecurs, if only rarely. However, the ease with which the shift might occur in one direction or the other is wholly dif- ferent. Most of the available data concerning the Cruciferae indicate that a relatively complex multi-allelic genetic sys- tem is involved in producing self-incompatibility. This re- quires the development of a genetic system that actually prevents delivery of the pollen content (i.e. the male gametes) to the female gametophyte. In a sporophytic sys- tem, such as that widely found in the Cruciferae, incom- patible pollen is inhibited before pollen tube penetration of the stigma. The release of such a polygenic system provid- ing positive prevention of autogamy might require only a single mutation. Thus, while the development of a complex of interacting genes could evolve only gradually and prob- ably over a fairly long time span, perhaps with several steps before it resulted in obligate self-incompatibility, the shift from self-incompatibility to self-compatibility could be rel- atively easy and require only a short period. The wide- spread presence of self-incompatibility in the family argues for its early evolutionary development, perhaps even its presence as a characteristic of the progenitors of the family. On these grounds, added to those relating to the particular nature of the species themselves, we believe self-incompati- bility preceded self-compatibility both within and between species of L thia. Anther position. — One of the very interesting findings in the present study has been that a general correlation exists between an extrorse position of the anthers of the two sets of paired stamens and self-incompatibility of the particular taxon on the one hand and a partially introrse or introrse position and self-compatibility on the other. Against these 20 REED C. ROLLINS changes of position of the paired stamens, the single short stamens in the Leavenworthia flower maintain their anthers in the introrse position. In the self-incompatible populations of L. alabamica, L. crassa and L. stylosa, the anthers of the paired stamens are essentially extrorse and the pollen is delivered away from the stigmatic surface. A contrastingly different pattern is characteristic of the self-compatible spe- cies where the anthers of the paired stamens are either par- tially or wholly introrse. Here, the pollen is delivered to- ward-the stigmatic surface, enabling a high degree of auto- gamy to occur. In Plate 1, two views of the upper part of the paired stamens, together with the stigma and a small portion of the style of ten different taxa, are shown. Ex- amples of an extrorse position are seen in Plate 1, fig. la and 1b of L. stylosa, and in fig. 9a and 9b of L. alabamica as well as in other figures of the plate where the extrorse position is less definite. Examples of introrse anthers are fig. 2a and 2b of L. awrea and 8a and 8b of L. uniflora in the same plate. The anther positions of both paired and single stamens and their position in relation to the stigma are illustrated in Plates 2 and 3. An analysis of the anther position in a number of wild populations and growing cultures show that a range from extrorse to introrse occurs in the genus. Five diagrammatic patterns, as indicated in Fig. 2, have been developed to rep- resent the principal pattern types seen in the living material. Table 2 gives a record of the stamen pattern and the com- patibility type in a number of populations and cultures. These data indicate a strong tendency for the anther pattern to shift from extrorse to introrse concomitantly with a shift from self-incompatibility to self-compatibility. These cor- related changes have occurred during speciation in the genus and apparently are continuing to occur as a part of the evo- Iutionary process during the present period. This is shown by the fact that there are self-compatible populations with- in species that are otherwise mostly self-incompatible. Such is the case in both L. alabamica and L. crassa. In Table 2, it may be seen that stamen patterns A and B tend to char- acterize the self-incompatible L. alabamica var. alabamica and L. crassa var. crassa but that patterns C and D are characteristic of the self-compatible L. alabamica var. LEAVEN WORTHIA 21 on NOY Rog aT mTNE Tra UE PLATE 1. Anthers in relation to stigma. The figu show one anther from each of the two stamen pairs. anthers are to be seen in these. di oe a single pair. Fig. la, b, L. stylosa; Fi var. crassa; Fig. 4a, b, L. alabamica var. brachystyla 6a, b, L. crassa var. elongata; Fig. 7a, b, L. torulosa g. 9a, L. alabamica; Fig. 10a, b, L. exigua var. lutea. All figures X 10. Daninwe en. 22 REED C. ROLLINS brachystyla and L. crassa var. elongata. It should be noted that some variability is present in each population. Pat- terns C, D and E are found in the self-compatible species. The differences in anther orientation apparently are due to rotations of the filaments at the point of insertion on the receptacle of the flower. There is no evidence of twisting of the filaments or of a change in orientation of the anther with respect to the filament. Flower Size. — Early in the present study it was observed that there are substantial differences in the flower size of different species of Leavenworthia. In some instances, these differences are of considerable magnitude, as between L. uniflora and L. stylosa. In comparing other species, for ex- ample L. crassa, L. alabamica and L. stylosa, the flowers proved to be similar in size. The significance of these dif- ferences became evident when a correlation between large flowers and self-incompatibility was realized. Since the self- incompatible species are obligate outcrossers, they are wholly dependent upon insect pollination for survival and large flowers are a distinct advantage. Whenever a popula- tion becomes self-compatible, insect attraction is no longer vital and the maintenance of large flowers is not especially favored by selection. Table 3 presents measurement data for petal length, petal width and sepal length for a number of populations of nine taxa. The site where each population grew or was grown is given because the flower size is some- what influenced by conditions during growth. This may be seen by a comparison of measurements on the same acces- sion, No. 57142, grown both in the greenhouse and in the experimental garden. However, such differences due to the influence of local factors of the environment do not tran- scend the larger differences in flower size distinguishing some of the species. The range in flower size in Leavenworthia is, graphically shown in Fig. 3, in which the mean petal length and width of selected populations are represented by ideographs and the mean sepal length by an adjacent vertical bar. Fig. 3 was prepared from data in Table 3. It is quickly evident from visual comparisons within Fig. 3 that the largely self- incompatible L. alabamica, L. crassa and L. stylosa are large-flowered as compared to the self-compatible L. awrea, LEAVENWORTHIA 23 PLA 1b, eran le, stamens; 1d, petal. Fig. 2, L. ext and stigma; 2c, cima; 2d, petal. style and stigma; 3c, stamens; 3d, e, petals. F 4b, style and stigma; 4c, petal; 4d, stamens. F different petal stance; 5c, style and stigma; 5d, sta at ae and petals X 2; styles and stigmas X 10; stamens X 3. Drawings by Ruth Hsu Chen. 24 REED C. ROLLINS E 3. Flowers and flower parts of Leavenworthia. Fig. 6, L. uniflora; 6a, flower; 6b, stamens. Fig. 7, L. exigua var. exigua; 7a, flower; 7b, style and stigma; Td, T H . ic, stamens; petal. Fig. 8, L. aurea; 8a, flower; 8b, style and stigma; 8c, stamens; 8d, p ig. 9, L. alabamica; 9a, flower; 9b, style and a; 9e, st ns; 9d petal. Fig. 10, L 0a, flower; 10b, style and stigma; 10c, stamens; 10d, petal Fig. 1 . alabamica var. brachystyla; 11a, flower; 11b, style and stigma; 1le stamens; lld, petal. Fig. 12, L. torulosa; 12a, flower; 12b, style and stigma; 12c, stamens; 12d, petal. All flowers and petals X2; styles and stigmas X10; stamens X 3. Drawings by Ruth Hsu Chen. LEAVENWORTHIA 25 TABLE 2. STAMEN PATTERN AND SELF-COMPATIBILITY en OF POPULATIONS OR CULTURES OF LEAVENWOR STAMEN PATTERN SPECIES CULTURE OR SEED OF FLOWERS SELF- POPULATION NO. SOURCE A B C D E_ COMPATIBLE L. stylosa 59-135 Tennessee x — 59-141 4 x = 55161 us x nt 6009 " x nt 6030 ni xi nt L, alabamica var. alabamica 59-29 Alabama x aoe 59-117 e x = 59-173 ‘i x ar 59-113 % x mrt 59-114 s x “i 5929 e x & nt var. brachystyla 59-115 ‘Alabama qa 2 5926 * nt 59382 fi oe nt 6019 sf not observed * L. erassa var. crassa 59-118 Alabama 5 ek. = 59-120 id x a7: 59-121 " be = 59-122 . x = 5923 x nt 931 os x nt var. elongata 59-119 Alabama a + L. torulosa 59-20 Tennessee x + 59-33 cf x 2 8 59-61 ‘n x . 59-152 « oa + 59-179 S x = 59-181 % x rh 5918 . eae 7 nt 5920 8 x nt L. aurea 59-1 Oklahoma x ris 59-2 ° 2.88 6135... 50 G50. 5372076 3:20 455 . 400 . OTT 512. 59D. 304 . 95. 398 . 509 . 212 . 619... 440. SPEAL LENGTH SITE .071 greenhouse .046 garden 075 garden .066 roadside .068 pasture .086 pasture .049 field .058 glade .086 near brook .052 pasture .043 field .065 field present a sizeable and well defined target for insect attrac- tion and landing. This point is well illustrated in Plate 2, fig. 5b, which also shows the widely spreading sepals char- acteristic of the flowers of these species in full anthesis. Other species, such as L. aurea and L. uniflora, have the petals less well differentiated into blade and claw and there is a marked tendency for the petals to remain in an ascend- ing position over their full length even at full anthesis. Plate 3, fig. 6a, shows a common petal stance found in L. uniflora populations of Tennessee and Alabama. The easy movements of flower parts to a closed position insures self-pollination, at least in those flowers with in- trorse or partially introrse anthers, because the dehiscing part is mechanically pushed against the stigma. This is of considerable importance to species that are also self-com- patible because it permits autogamy to occur under circum- stances where cross-pollination may be inhibited or com- pletely absent due to the lack of insect visitation of the flowers. As a general observation, it is more than usually evident that flower behavior in Leavenworthia is closely tied into the requirements of a particular species for insect pol- lination. For this reason, there are a large number of flower characteristics insuring insect visitation in the self-incom- patible species and many of these features are modified or lost in the self-compatible species. Observations on material grown in the greenhouse indi- cate that pollen ripens during the late stages of bud devel- LEAVENWORTHIA 31 opment and tends to be shed on the first day the flowers open. Ordinarily, the flowers will open a second day even though pollination and fertilization have occurred on the first day. If compatible pollen is withheld from the flowers, they will often open a third day, provided temperature and moisture conditions are favorable for good plant vigor. Pollen Quality. — A broad survey of the pollens, including samples from several populations of each species and varie- ty, revealed only that the pollen for each population was relatively uniform in size and nearly all the pollen grains stained deeply with methylene blue, indicating that very few grains (less than 1% in most populations sampled) were empty of contents. From the morphology of the pol- len alone, deficiencies that might adversely affect the re- productive apparatus could not be detected. In general in interplant pollinations within the same accession, the re- sulting seed set was good. Also, the seed-set was good upon self-pollination in self-compatible types. Although consid- erable data relating to pollen quality were assembled, I can- not see that this has any real bearing upon the problems at hand as long as they revealed only that the quality was uni- formly good. For this reason, the raw data are not presented. If anything but uniform results had been obtained, the data would then have been pertinent to our study. Style Length.— There are elements of an inverse relation- ship between style length and silique length in Leavenwor- thia. This could be stated as, “the longer the style, the shorter the silique,” or conversely, “the longer the silique, the shorter the style.” However, these generalizations are only relatively applicable and do not always hold. It turns out to be a significant matter only because the long-styled, short siliques are more prevalent in the self-incompatible taxa than in those with a self-compatible breeding system. The nature of the differences may be seen by comparing fig. 1, 3 and 4 of Plate 5, photographs of fruits of self-incom- patible taxa, with fig. 2, 5, 6 and 7 of the same plate. The latter are photographs of fruits of self-incompatible taxa, to the same scale. The fact that relatively long-styled siliques more or less characterize the self-incompatible species may be interpreted as suggesting that the progenitors of Leavenworthia were 39 REED C, ROLLINS also long-styled. This suggestion follows because the self- incompatible species are thought to be primitive in the ge- nus. However, the point cannot be pursued because no prob- able progenitors of L thia are known and no fossil record exists. An alternative suggestion as to the reason for a correla- tion of long styles with self-incompatibility lies in a consid- eration of pollinating efficiency. The complete dependence of the self-incompatible plants upon insects for pollen trans- mission places them in a vulnerable position with respect to pollination. If too few insects are active to insure an ade- quate supply, some of the ovules may not be fertilized. A frequent low pollen supply would set up a selective system favoring plants in which the available pollen was most effec- tively utilized in seed production. This might be expected to result in shorter siliques with fewer ovules. The difficulty with this interpretation is that it puts the short-fruited, jong-styled species in the position of being the derived spe- cies while most of the evidence points to them as being a- mong the primitive group. There is good evidence that the concomitant occurrence of short styles, long siliques and self-compatibility is not wholly fortuitous. This is seen in both Leavenworthia crassa and L. alabamica, where the self-incompatible var. crassa and var. alabamica are relatively long-styled, whereas the self- compatible var. elongata and var. brachystyla have shorter styles. The general trend in the genus is shown graphically in Fig. 4. Silique Length. — In Table 4, means of four measurements are given for the siliques of a number of populations of all species. These are length of the fertile part of the silique, excluding the style, length of style, width of silique and thickness of the silique. The length of the silique in a gen- eral way is correlated with the number of ovules produced and the possible number of seeds per silique. In Plate 5, fig. 3 and 6, the valves were removed from the siliques on the extreme right. The funiculi, slightly projecting inward from the replum may be counted to determine the number of ovules produced in each silique. In the self-incompatible L thia crassa, Plate 5, fig. 3, six funiculi may be seen, while in the self-compatible L. wniflora, fig. 6 of the LEAVENWORTHIA De ” q E = N = . = ™ 2 : ©) PLATE 4. Tennessee Siliques of Leavenworthia. Fig. 1-3, three fruit types of L. stylosa of Fig. Fig. -~ ~ 2 © . “Saas —_— ihe nw. Murfreesboro; Fig. 2, near Hurricane Creek 5 mi. east of eiciettis Fig. 4-5, two fruit types of L. torulosa of | Stes fig. 4, 1.5 mi. east of Eagleville: fig. 5, ar mi. ne. Shelbyville. The siliques of each hides were selected to show variation in size within a single population. 34 REED C. ROLLINS 3 z 2 PLATE 5. Siliques of Leavenworthia. Fig. 1, L pa astounds var. brachystyla. Fig. valves removed. The siliques of each figure selected to show variation in size within a si ingle population. . alabamica var. alabamica. Fig 3-4, L. crassa var. crassa; in fig. 3, the silique at right has had the valves removed. Fig. 5, 6 i i in as L. exigua. Fig. 6-7, L. wniflora; LEAVENWORTHIA 39 same plate, seventeen funiculi may be counted. Comparing the data of Table 5, it will be seen that the number of seeds produced per silique is very significantly lower in L. crassa, with means of 3.3 and 6.1 in the two populations counted, than in L. uniflora, where a low mean of 9.6 and a high of 16.4 were found in different populations. Other species are intermediate between these two in numbers of seeds per silique. With respect to silique length, there appear to be both 5998 S913 S098 Leavenworthia uniflora Leavenworthio torulosa Ht n=15 n= 59104 : Leavenworthia stylosa Leavenworthia aurec n=15 venworthia cra oagote Leavenworthia exigua Ay HH I Va Leavenworth alabam ziti Fic. 4. Ideographs drawn to scale from the means of four measurements of siliques 7 te specimens deposited in herbaria. The vous number of each taxon is giv below the n 36 REED C. ROLLINS within species trends and trends between species. Siliques of a certain length group are sometimes characteristic of a population or a series of populations of a particular area. In L thia stylosa a shorter fruited phase also is yellow-flowered and is found in the northwestern portion of the species range. For example, this phase is illustrated by populations 59104 and 6142, Table 4, where the mean silique lengths are 11.8 and 12.6 mm. respectively. A much longer fruited phase with white to light lavender flowers is found in the southern part of the species range. Here, populations 59114 and 61167, with respective mean silique lengths of 26.9 and 22 mm., are illustrative. Similar but less pro- nounced differences show up between L. crassa var. crassa and L. crassa var. elongata. Differences of the same order of magnitude are present in L. torulosa, the high mean of 31.2 mm. being nearly twice that of the low mean of 16.2 mm., but the other self-compatible species are more uniform than L. torulosa, as shown by the data of Table 4. Within the three 15-paired chromosome species, trends are evident towards longer siliques with more seeds per silique. The shortest siliques with fewer seeds characterize certain populations of L. stylosa and although the range in both of these characteristics is quite wide in this species itself, a general trend towards longer siliques and more seeds in L. torulosa culminates in the largest seed producer, L. uniflora. The mean number of seeds produced per silique in 4 populations of L. stylosa was 5.9 + 2.4, in 4 populations of L. torulosa, 7.7 + 1.9, and in 6 populations of L. uniflora, 13.8 + 2.2. The differences between L. stylosa and L. toru- iosa apparently only mark a trend but the jump in number of seeds per silique in L. uniflora is of undoubted signifi- cance. A clear trend in silique length is less evident among the 11-paired chromosome species than in the 15-paired group. However, the trend towards reduction in style length is un- mistakable and the combination of this with an increase in silique length appears to be a significant step within both EL. crassa and L, alabamica, because it also coincides with the change from self-incompatibility to self-compatibility in these two species. This and the general situation on silique size are graphically illustrated in Fig. 4. LEAVENWORTHIA 37 TABLE 4. SILIQUE AND STYLE MEASUREMENTS OF SAMPLES FROM WILD POPULATIONS OF LEAVENWORTHIA SPECIES & NO, OF WIDTH THICKNESS LENGTH STYLE LENGTH POP. NO, FRUITS m Es V2 ey W/E z 8 VE ros Vz L. aurea 6152 50 4.84 .454 .064 2.56 .254 .086 16.10 2.235 .816 2.29 .299 .042 5972 56 4.81 .830 .111 2.67 .690 .092 17.54 2.40 320 2.21 .265 .035 5973 56 4.31 .265 .085 2.56 .233 .031 20.00 2.58 344 2.60 .271 .036 5974 56 4.67 .225 .030 2.66 855 .047 22.29 2.74 .366 3.05 .294 .039 59109 65 4.10 .245 030 1.94 .198 .024 12.65 1.653 .205 2.06 .239 .029 6143 50 3.72 .538 .076 =: 1.17 .181 .026 14.62 1.782 .252 2.06 .174 .025 5997 63 4.67 .255 .082 2.83 .281 .029 17.88 1.3825 .166 1.62 .256 .032 66 4.81 .288 .085 1.85 177 .021 1448 2.838 286 1.50 .224 .027 L. alabamica 5992 66 3.52 .298 .037 1.90 .209 .026 16.80 2.862 352 3.72 .565 .070 5978 61 3.75 .823 041 1.74 .226 .029 17.25 2.660 .341 3.72 .514 .066 5980 67 3.77 .329 .040 1.91 .221 .026 18.04 2.894 .353 3.50 .319 .390 6129* 50 3.62 .815 .045 1.41 .148 .021 19.84 2.335 .830 3.71 531 .075 3.94 .215 .030 2.11 .490 .069 13.52 3.899 .551 2.63 .215 .030 6014* 70 3.89 .282 .027 2.17 .144 .017 14.56 1.740 .207 2.47 .302 .036 61172* 50 4.00 .165 .023 2.29 .138 .020 17.00 1.604 .227 2.24 .188 .027 5990* 68 4.09 .174 .021 2.29 .238 .028 18.60 1.881 .228 2.31 .251 .030 61170* 50 3.85 .273 .089 2.15 .217 .031 19.38 2.276 322 2.58 .547 .077 5989 66 4.67 .339 .042 3.37 .430 .053 6.17 .833 .102 4.78 .554 .068 61173 50 84.97 .3835 .047 3.75 .338 .048 6.67 .118 102. 4.56 .5381 .075 5991 35 4.91 .497 .084 3.27 .409 .069 6.86 .944 .160 4.97 .547 .092 5983* 65 4.47 .405 .050 2.99 388 .042 7.74 1.166 .144 3.90 .404 .050 6015* 70 4.55 .481 .057 38.71 .408 .049 8.00 1.077 .129 4.49 .250 .080 5986* 64 4.73 .3813 .0389 3.88 .500 .063 8.59 1.137 .142 4.91 .582 .073 5964*. .37 . 4.33 .407..067.. 2.75 345 .067. .. 10.08 1.344.221 3.61 .524 .086 var. elongata 5985 ol AD2 262 O4T 3.62. 258 046 9.68 1.194 .214 2.84 .348 .063 L. stylosa 59104 66 -8.91 34) .042....2.96:.312 .038 11.83 1.878.231... 5.40..709 .087 6142 DO o.4e Ae? 060 . 222 2b2 051... 12.62.2407 3be...:.5.26 921 .130 ean, standard raver and standard error are given for each measurement in each population. A population number marked with an asterisk indicates that the population occurs in a Y eabetvaaea field. 38 REED C. ROLLINS Table 4. (Continued) SPECIES & NO, OF WIDTH THICKNESS LENGTH STYLE LENGTH POP. NO. FRUITS mm mm mm Bkiig Be Bi Re va tt. Brig y...rf8 Ee ye 6030 70 3.58 .3818 .038 2.37 .238 .028 12.95 1.861 .222 4.57 .723 .086 59105 25 4.36 .269 .054 3.06 .303 .061 14.28 3.348 .670 5.29 .719 .144 59107 47 3.65 .256 .037 2.82 .270 .089 16.72 3.174 .463 5.11 .8385 .122 61164 50 «3.83 .869 052 2.71 .257 086 16.88 3.532 .500 5.04 .986 .132 59117 47 3.55 .294 .043 2.41 .314 .046 19.17 1.110 .162 4.58 .892 .130 59118 47 3.72 .298 .043 2.64 .289 0385 20.15 1.552 .226 5.13 .358 .052 59115 47 3.44 .400 .058 2.48 103 .015 20.30 2.843 .415 5.01 .939 .187 61167* 50 3.76 .362 .051 2.89 .810 .044 22.02 2.903 .411 5.11 .814 .115 59114* 45 3.49 .363 .054 2.69 .276 .041 26.98 4.120 .614 5.28 .539 .080 5998 64 3.60 .242 .030 2.56 .227 .028 16.28 1.035 129 2.81 .493 .062 59111 37 2.85 .170 .028 2.42 404 .066 16.43 2.328 383 3.64 .321 .053 59102 50 3.13 .202 .029 2.81 .276 039 19.84 2.454 847 3.86 .416 .059 59101 64 3.29 .272 0384 2.53 .215 027 21.53 3.348 .419 3.48 .428 .054 59116 47 3.22 .227 .033 2.63 .191 .028 22.51 2.726 .397 3.30 .374 .055 59106 33 3.49 .297 052 =—.2.83 302 .058 23.73 3.792 .660 3.03 .576 .100 61166* 50 2.84 .249 035 2.85 .162 023 27.06 2.535 .859 3.60 .843 .049 59112* 25 3,02 .160 .0382 2.08 .236 .047 28.80 2.848 .570 3.40 .137 .027 59113* 39 2.86 .250 .040 2.61 .887 .142 31.26 3.109 .498 2.86 .476 .076 61161 50 4.19 .282 .040 2.78 294 .042 18.32 2.944 .416 2.31 .3820 .045 61163 50 4.24 .394 .056 2.88 .257 .036 21.72 3.603 .510 1.93 .215 .030 59110 49 4.15 .267 .088 2.57 .197 .028 22.67 3.191 .456 1.53 .187 .020 59103 213.91 .368 080 2.61 271 .059 25.48 3.109 .678 1.15 .746 .163 5982 66 4.87 .261 .082 2.86 .189 .023 25.57 2.720 835 © 1.18 .228 .027 6101* 50 38.97 325 .046 2.58 .239 034 20.70 2.660 .876 2.22 .254 .036 6126* 50 4.03 815 045 2.59 287 084 23.17 3.185 .450 2.03 .222 .031 5993* 64 4.14 310 .089 3.09 .274 034 25.34 2.808 .351 1.49 .295 .037 6135* 50 4.41 .451 .064 2.93 .239 034 28.54 2.643 373 1.12 .225 .032 5999* 39 3.81 365 058 2.59 .214 034 29.31 3.163 .506 1.01 .281 .045 The interpretation of the silique length data given in Table 4 is open to some difficulties unless care is taken to discriminate between those populations growing in culti- vated fields, marked with an asterisk after the population number, and those growing in less disturbed sites. Growth is nearly always more luxuriant in the old cultivated fields than in the more natural sites. The effect on silique length may be judged by comparing the data of population 61169 with population 61170 of L. alabamica var. brachystyla. Population 61169 occurs in a natural cedar glade just uphill from an old cultivated field where population 61170 is pres- ent. Obviously seeds from the glade population provided the LEAVEN WORTHIA 39 TABLE 5, SEED NUMBER PER SILIQUE 424 WILD POPULATIONS OF LEAVENWORTHIA Zz 8 Range SPECIES & POP. NO M ite aLor M U&A T M Sven. L. alabamica var. alabamica var. brachystyla 5990 8.34 0.36 8.70 1.66 0.66 1.58 5-11 0-3 6-12 6014 5.56 1:20 6:76 ' A 9P1108 132 1-10 0-6 4-10 6106 8.64 0.14 8.78 1.76 0.85 1.89 6-12 0-2 6-12 61169 5.24 0.48 5.72 1.20 0.77 1.21 8-10 0-38 4-10 61170 8.92 0.68 9.60 1.22 0.87 1.41 6-12 0-3 6-12 L. aurea 5973 7.86 0:94 '8:30™ PGT U9" 1:50 4-11 0-3 a-L1 6152 5:62-1.805 6925 D277 807 Bt 8-10 0-5 1 L. crassa 5994 6:.18:.0.32...6.80 o-L.32) 0.56. 1225 4-10 0-2 4-10 61173 3.00; O34 9550) Gad 0210-406 0-8. 0-2 2-8 5997 5.80 b.68 - 7.48... 1.92 41.27 1.64 2-10 0-4 4-11 59108 fe 096. 6.42 "* 142° 625. 1.26 1-9 0-7 4-9 59109 492° 0.88'° 5:80" °1:31''0.85 1:41 2.8?) 0-3: 4-10 6143 6.84 0.34 7.18 1.36 0.561.389 5-100-2 5-10 var. lutea 5995 6.04 0.86 6.90 1.64 0.88 1.43 2-10 0-4 4-10 L. stylosa 6030 381, 74-.b12 199 be 3.64 0-9 0-5 3-9 6142 4.78 0.46 5.24 1.40 0.84 1.53 28 0-4 338 61164 5.78 1.10 6.88 1.83 0.99 2.12 2-10 0-3 «4-18 61167 9,90 1.42 11.82 1.93 1.70 1.83 5-13 0-8 6-16 L. torulosa 69101 7.12 0.72 7.84 1.66 0.81 1.82 4-11 0-3 4-14 59111 6.56 0.52 7.08 1.42 0.81 1.65 4-11 0-4 4-12 61166 13.02. 0.24. 113.26 1.68 0.438 1.67 7-15 O-1 8-15 5998 6.34 0.84 7.18 1.48 0.88 1.45 4-10 0-3. 5-11 L. uniflora 5982 13.68 1.98 15.68 2.45 1.54 2.39 8-18 0-5 9-20 6101 13.90 1.08 14.98 2.29 1.05 2.30 9-19 0-3 9-22 6126 16.04 1.836 17.40 2.85 1.95 2.85 11-25 0-10 13-25 6135 16.40 1.46 17.86 2.89 1.63 2.91 4-21 0-7. 12-22 61161 9.62 2.42 12.04 2.61 1.27 3.00 4-15 0-7 17-20 61163 13.42 3.06 16.48 3.04 2.18 3.80 8-20 0-10 10-25 ®‘Mean, standard deviation and range for a random sample of 50 siliques from each popeteci! are given. M=matuyre seeds; U & A = unfertilized and aborted ovules; and T = total, all seeds plus all ovules. 40 REED C. ROLLINS source materials for the field population and probably con- tinues at present as a seed source. For this reason, we may assume that both populations have roughly the same genetic ingredients. But compare the mean silique length, 13.5 mm. +3.89 mm. of the glade population with 19.3 mm. + 2.27 mm. found in the field population. These data were taken at the same time and under the same conditions and it must be as- sumed that the differences are largely, if not wholly, due to the differences between the glade site and the cultivated field. It is interesting to note that the means for silique width and thickness, and style length, do not differ signifi- cantly in the same material. It should be understood that a number of factors favorable to plant growth might be present in the cultivated fields including such items as the use of fertilizer. No attempt was made to ascertain the nature of the cultural practices in the fields where sampling was done. Silique Width and Thickness. — Measurements of width; the dimension from replum margin across the valve and parallel to the septum to the opposite replum margin (cf. Plate 5. fig. 3 and 6), and of thickness; the dimension from the back of one valve to the back of another in a line per- pendicular to the septum, were made with a thickness gauge. The plunger of the gauge could be quickly put into position and the dimension was read directly from a large dial grad- uated to hundredths of a millimeter. The most important information derived from these meas- urements bears upon speciation and does not elucidate any particular aspect of the reproductive system. There are some differences in silique width between species but these are minimal and can be matched by similar differences be- tween populations in such species as L. exigua. However, there are significant differences in silique thickness between such species as L. crassa and L. alabamica. The siliques of L. crassa are hard and the valves are thick walled. This feature is not seen on herbarium specimens after the siliques have become dry. On the other hand, the siliques of L. ala- bamica are flat and non-fleshy. These features are reflected in the thickness dimensions given in Table 4. LEAVENWORTHIA 41 DISCUSSION AND CONCLUSIONS RELATIVE TO THE BREEDING AND REPRODUCTIVE SYSTEMS An appropriate emphasis on the nature of the breeding and reproductive systems in Leavenworthia stresses the congruence of certain features with recognizable evolution- ary changes that have taken place within the genus. The primary change in the breeding system itself has been from self-incompatibility to self-compatibility. It is important to point up the fact that this has taken place within two of the seven species of Leavenworthia and is a major trend within the genus as a whole. The most significant direct result of this change is the emancipation of a species or of a popula- tion, as the case may be, from its complete dependence upon insect pollination. The concomitant changes in the repro- ductive system may be attributed to new selective pressures developed because of the changed circumstances, or to the release of pressures that tended to maintain, under a self- incompatible breeding mechanism, a balanced situation in- volving reproductive structures, insect vectors and breeding system. The evolutionary trends affecting the reproductive system that appear to be associated with the change in breed- ing system are the following: (1) Extrorse to introrse anthers. (5) Increase in stigma size. (2) Reduction in flower size. (6) Decrease in anther size. (3) Flaring to erect petals. (7) Reduction in style length. (4) Odoriferous to non-odoriferous (8) Increase in ovule number. flowers. These trends represent a remarkable set of correlations related to the shift from self-incompatibility to self-compati- bility. The fact that this change has occurred independently at least three times (actually, the evidence points to a mini- mum of four times) indicates that powerful selective forces have been favoring self-compatibility over self-incompati- bility during the recent evolutionary history of Leavenwor- thia. These same selective forces are probably continuing to act on the populations of Leavenworthia that today are self-incompatible. Given the proposition that selective forces repeatedly capi- talize on autogamous Leavenworthia plants and that they tend to fix the mechanism of release from obligate self-in- compatibility into the genetic structure of the Leavenworthia 42 REED C. ROLLINS populations, the question as to the relationship of such forces to the results produced should be asked. The matter is ob- viously complex and difficult to decipher. I shall attempt to deal with it first on a specifle level and then on a general level. ‘The first assumption one has to make is that it is of con- siderable selective advantage to the populations of Leaven- worthia to become self-compatible. This is in addition to the obvious fact that self-compatibility is of tremendous ad- vantage for purposes of migration. Looking into the most vulnerable places in the reproductive cycle for clues as to why self-compatibility is of importance, we are led to look at pollination as a possible critical stage. If for any reason the supply of transported pollen is diminished or absent, this is critical to the survival of the self-incompatible popu- lations. On the other hand, interplant transport of pollen is not a factor in the survival of self-compatible, self-pollina- ting populations. Since it is the transported pollen that ap- pears to be critical, we look to the pollinating agent for fur- ther clues as to the reasons an unfavorable situation might exist. In the case of Leavenworthia there is evidence that in- sects are not doing an efficient job of pollen transport. The evidence is largely observational and rests on the fact that in five different early spring seasons in Tennessee and Ala- bama, populations of self-incompatible I enworthias were in full bloom on many days in which insect visitation of the flowers was minimal. Had it not been for the observation at the beginning of my study that many early season plants set no fruit, I would have been tempted to assume that Leavenworthia was not insect pollinated, because of the general lack of insect activity on the early season flowers. It is clear that Leavenworthia does flower when the nights in the areas where it rows are still cool to cold and that there are numerous cold, windy or wet days when insect activity is at a minimum. It would appear that Leavenworthia as a genus has some- how gotten “out of phase” with its environment as regards a critical part of its reproductive evcle, a part upon which it was highly dependent as long as i i patible. The part, of course, was that associated with inter- LEAVENWORTHIA 43 plant pollen transport by insects. The primitive species of Leavenworthia are still insect-dependent and it is a fair presumption that the whole genus was at one time self-in- compatible and hence insect dependent. How and why would populations of plants of a large proportion of a genus get “out of phase” with such an important factor of the envi- ronment relating as it does so closely to their success and even to their survival? A theoretical explanation, not without basis of fact and observation, requires the assumption that all species of e vorthia are clesely adapted physiologically to a par- ticular set of edaphic conditions. These conditions, extra- polated from the particular sites where Leavenworthia now grows, are thought to include especially shallow soil depth over dolomitic limestone, bedded in such a way as to main- tain at least temporary high moisture content near the soil surface during the wet part of the growing season. These conditions were present at all natural sites where I have observed vigorous growth of plants of the genus. High moisture, even standing water, appeared to be an essential element of the habitat. We interpret this as evidence that soil moisture was probably a feature of the erly environ- ment of Leavenworthia and that a close adaptation to a wet habitat has persisted as one of the requirements of these plants for survival. The requirement is genus-wide and it has not been bridged effectively during the evolution- ary history of the gens down to the present. This argues for continuous high soil moisture during certain phases of the growth cycle as a limiting factor in relation to potential lines of evolutionary deve'opment that would necessitate bridging of that requirement. For example, a development that would include a dry habitat. If we now look at the hieh soil-moisture requirement of L thia in terms of the recent geological and climato- logical history of the area where the genus now grows, we must conclude that the long term trend has of necessity had to be towards earlier growth and flowerin~. This is so because the area has gotten progressively dryer and warmer. The annual period when high moisture is available, combined with other conditions favorable for plant growth, has continuously shifted towards an earlier part of the 44 REED C. ROLLINS season. It would appear that, because of its requirement for moisture, Leavenworthia in its growth and flowering period has followed this long term climatic trend. In doing so, because of the rigorous natural selection operative as part of the edaphic environment, the several components of Leavenworthia, perhaps at different times and in different places, must have reached a point in their overall evolution where the selection pressure from the habitat was of greater survival importance than the benefits of heterogamy pro- vided by cross pollination. From this point onward, a dis- parity between life cycle and the role of insects in it could readily develop. Whenever the situation reached a point where flowering was critically out of phase with the pres- ence of efficient insect vectors, an immediate premium was placed on any mutant plant that circumvented the need for insects without adversely altering other features of survival value. One common mechanism that circumvents the de- pendency of self-incompatible plants on insects is a shift to self-compatibility and, if necessary, to autogamy as well. It should be clear that the insects were not necessarily sub- jected to the same set of selective factors as the plants. There is no indication that they are specifically dependent upon Leavenworthia in any way. We see no reason why they should evolve to follow the same seasonal pattern as the plants. The main step in the emancipation of self-incompatible populations of Le 7 thia from their dependency upon insects is the shift to self-compatibility. However, this is essentially an enabling step, permitting self-fertilization but not insuring it. Further modifications have evolved to con- sistently bring about self-pollination at least as a backup to whatever cross-pollination occurs. The evolutionary modi- fications of the reproductive apparatus relate first to the pol- linating process and secondly to the exploitation of an in- creased and consistently abundant source of male gametes through their utilization to increase the number of fertilized ovules in each silique. Perhaps the most important floral modifications related to the insuring of self-pollination is the shift from extrorse anthers to partially or wholly introrse anthers. This step places the anthers in the appropriate position for self-pol- LEAVEN WORTHIA 45 lination but still would not insure it if the anthers were at some distance laterally from the stigma. Two further modi- fications are seen in the self-compatible types. One is the tendency for the anthers to be in contact with the stigma, as is characteristic of L. uniflora (Plate 1, fig. 8). The other is the tendency for the petals and sepals to be more erect and closer to the stigma in the self-compatible groups than in the self-incompatible groups. This, coupled with a high degree of sensitivity of the flowers to light intensity with respect to opening and closing, provides a means for mechan- ically pushing the introrse anthers into the stigma, thus in- suring pollen deposit. On dull or rainy days, the flowers may not open at all and the pollen-charged anthers atop the elongating filaments are guided by the closed petals directly into the stigma. The rather strong evolutionary trend to- ward a reduction in flower size, showing up independently in both chromosome groups of the genus, may be related to these aspects of the pollinating process. A reduced diameter in the total flower size, coupled with an increase in stigma size, places the anthers and the outer floral parts closer to the outer portion of the stigmatic surface than is the case in the large flowered types. This enhances the chances of self-pollination. The reduction in flower size may be viewed in a different way, also. It may be argued that large flowers with an open stance to the petals would be selectively maintained and en- hanced by the self-incompatible groups because of their obligate position with respect to insect visitation. As soon as their dependency upon insects for pollen transport was bridged by self-compatibility, the selective pressures main- taining the large flower size would be released. Under these circumstances, flower size would tend to adjust towards a smaller size to the extent that they were previously main- tained at the large size by positive selective pressures. It does appear probable, as suggested above, that the efficiency of insect transport of pollen has been of critical importance in the evolutionary history of Leavenworthia. However, there is no reason to reject one hypothesis and accept the other. It is probable that the release of selective pressures maintaining large flowers and the development of new ones favoring an even smaller flower than would have developed 46 REED C. ROLLINS otherwise were both operational in producing the smaller flowers we now see in L. uniflora and L. exigua, for example. The change from odoriferous to non-odoriferous flowers is probably due to the release of positive selective pres- sures tied up with the maintenance of an efficient pollen transport system by the attraction of insects. I have no evidence on this point except that the change has occurred. In L. torulosa some populations have odoriferous flowers and others do not. In the genus as a whole, it is significant that the non-odoriferous species are self-compatible and the odoriferous species are self-incompatible. L. torulosa is ex- ceptional in having some odoriferous populations. I inter- pret the situation there as indicating that neither odorifer- ous nor non-odoriferous flowers are strongly selected and that the odoriferous condition is probably a remnant from the past evolutionary history of L. torulosa. In considering anther size, it is clear that the largest anthers are present in the largest flowers and that reduction in size parallels a reduction in petal size and flower diameter but there is no similar reduction in sepal length or silique size. Furthermore, there is no overall reduction in pollen size. I have not determined whether fewer pollen grains are produced by the smaller anthers or not. A reduction in style length and an increase in ovule num- ber, along with an increase in the length of the fertile part of the fruit, appear to me to relate to an exploitation by Leavenworthia of an increased reliable supply of male gametes inherent in self-compatibility and self-pollination. Again, the selective pressures maintaining a particular bal- ance between ovule number and pollen supply were undoubt- edly drastically altered by a change in breeding system from self-incompatibility to self-compatibility. Amone all the species of Leavenworthia, L. uniflora is the largest seed producer and it is also the most consistently autogamous. This species produces the most ovules per silique and the most siliques per plant under normal growing conditions. This results in more seeds per plant, on the average, than are produced by the plants of any other species of Leaven- worthia. The fertile part of the silique is longer and the style is shorter in L. uniflora than in any of the other spe- cles. Our assessment of the reproductive apparatus as 4 LEAVEN WORTHIA AT whole, species by species, inevitably places L. uniflora at the top of the list as the most efficient seed producing species in the genus. In closing this discussion, one point requires clarification or it might otherwise be misunderstood. This point relates to what might be considered by some to be a paradoxical situation in Leavenworthia in which all evolutionary paths seem to go from outbreeding towards inbreeding and are, in effect, leading the genus to evolutionary suicide. The latter does not appear to be wholly the case or at least the vulnera- bility resulting from inbreeding is slowed down somewhat in Leav thia. An important point is that although self-compatibility does permit self-pollination to effectively bring about self-fertilization, this is not an obligate process. Cross-pollination and therefore cross-fertilization may also occur and probably do, in most populations of Leavenwor- thia. In this way a sufficient degree of heterozygosity is maintained within and between populations to protect them from sudden crises occurring in their environment that might otherwise wipe them out. In annual plants such as Leavenworthia, a given well adapted population may build very quickly with an insured seed supply such as autogamy might provide. On the other hand, they are very vulnerable to any weak link in their reproductive cycle as compared to perennial plants, as is evident from the case at hand. GROWTH CYCLE The natural growth cycle from seed to flowering of all species of Leavenworthia is considerably longer than that required under artificial conditions. Seeds germinate in Sep- tember or October producing a rosette of leaves (Plate 6) that persists in a semidormant form through the cold months of winter. Ordinarily, very little growth occurs after the last of November until the late weeks of February or the early part of March. Once temperatures are favorable, growth proceeds rapidly and flowers are produced in a few weeks time. In the field in Tennessee, we noted that L. uni- flora and L. exigua tended to flower in advance of L. stylosa and L. torulosa in places where all four species occurred to- gether. The question as to whether there were real differ- ences in the length of the growth cycle between the species oe REED C. ROLLINS L. crassa var, crassd. Fig. 4, L. aurea. 4.6 xX 2/8; Mature rosette leaves of Leavenworthia. Fig. a. Fy 6. Fig. 2, L. alabamica var. alabamica. Fig. 3, L. exigua var. exigu ‘ig. 5, L. torulosa. Fig. 6, L. stylosa. Fig. 7, L. uniflora. Fig. 1, 2, T MS 2/2. LEAVENWORTHIA 49 was a natural one. We could not accurately check the growth cycles of the species under natural conditions but we did keep records on first flowering dates in greenhouse cultures. The pooled results are given in Table 6. TABLE 6. SUMMARY OF GROWTH PERIOD OF LEAVENWORTHIA, 1959-60 t GERMI- SPECIES CULTURES NATION lst FLOWER NO. DAYS NO. DAYS No. ate ate ange ve. L. aurea 23 11- 4-59 3- 1-60 98-130 109 L. exigua var. lutea 2 11-17-59 3-15-60 103-119 111 L. exigua var. exigua 26 11- 4-59 3- 1-60 107-135 119 L. uniflora 8 11-17-59 3-30-60 101-142 119 L. crassa 5 11- 4-59 8-11-60 119-134 125 L. alabamica 7 11- 4-59 3-23-60 112-161 128 L. stylosa 3 11-17-59 4-12-60 124-152 141 L. torulosa 6 11- 4-59 3- 1-60 120-169 141 The lowest average number of days from seed to first flowering was found to be in L. aurea. This was somewhat surprising to me because in my previous experience with polyploids, particularly in Parthenium (Rollins, 1950), the growth cycle was slower in the high polyploids than in the lower chromosome plants. This was generally true both within species and between species of Parthenium insofar as the point was investigated. In the case of L. aurea (n = 24), the opposite situation seems to obtain. Although poly- ploid, it has a short growth cycle. This species is roughly comparable to L. exigua var. lutea (n = 11) in its growth cycle. These two taxa are the most southerly of the genus and appear to be related although they possess very differ- ent chromosome numbers. The effect of latitude cannot be readily assessed but it is clear that latitude as it effects cli- mate and in turn the growth cycle is not alone responsible for the short cycle of L. aurea and L. exigua var. lutea. This is shown by the fact that even though the ranges of L. exigua var. exigua and L. uniflora are north of L. crassa and L. alabamica, they have a shorter growth cycle. If latitude were the prime factor, var. exigua and L. uniflora would be ex- pected to have a longer growth cycle than L. crassa and L. alabamica. It is of significance that all self-compatible species, with 50 REED C. ROLLINS the notable exception of L. torulosa, show shorter growth cycles than the self-incompatible species. In the self-com- patible species, earlier flowering, which is associated with the shorter cycle, is possible because these species are not hampered by coming into flower before insects are available to transfer pollen. By contrast, reproduction in the self- incompatible species would be seriously affected by the lack of insect vectors. The situation in L. torulosa, which is self-compatible but with a long growth cycle, appears to be more complex than in the other species. The wide ranging growth cycle from a minimum of 120 days to a maximum of 169 days suggests possible wide genotypic differences. Otherwise, L. torulosa is closely related to L. stylosa, from which it presumably arose, and it might be expected to have the length of its growth cycle correspond somewhat to that of L. stylosa. GEOGRAPHICAL DISTRIBUTION The remarkable adaptation of Leavenworthia to calca- reous cedar glades makes locally disjunct distributions al- most inevitable. These are present in nearly every species. In addition, there are some substantial disjunctions in sever- al of the species. It is probable that these are the product of a relatively old dispersal that is presumed to have been char- acterized by a more continuous distribution than that of the present (Map 1). However, it is also probable that there has long been an element of habitat disjunction to be encoun- tered in connection with the dispersal and migration of any given species of Leavenworthia, This in itself would place a premium upon successful dispersal by a single seed and this very likely was an important factor in connection with the repeated development of self-compatibility within the genus. There is an interesting geographic parallel to the evolu- tionary picture in the genus. What is presumed to be the most basic species, L. crassa, is also the most restricted in its distribution. The next most restricted is L. alabamica, which is most closely related to L. crassa (cf. Map 3). Both occur only in northern Alabama. Then L. stylosa follows in size of geographic area and it is also the most basic species of the n = 15 chromosome group. In these three species we LEAVENWORTHIA 51 @& Leavenworthic clabemice AL. curee © AL erosse * L ‘gue * L exigue vor. utes * L. styfose L. tereloss a \ wsifiors Map 1. Generalized distributions of the species of Leavenworthia. have accounted for all that are self-incompatible. The self- compatible species, as expected, are all more widespread than the self-incompatible species. In the n = 15 group, the most widespread is L. uniflora. L. torulosa has a much smaller total range than L. uniflora but it exceeds that of L. stylosa to the north where it reaches southern Kentucky, to the southeast where it is beyond the Tennesseee River and to the south where populations are found nearly to the southern boundary of Tennessee (Map 2). The geographical range of L. uniflora is marked by dis- junctions particularly to the north and east of the largest area of Leavenworthia concentration (species and individ- uals) in the Central Basin of Tennessee. The one locality in Indiana, the few localities in northern Kentucky, and the few in southwestern Ohio, as well as those in eastern Ten- nessee are examples of substantial disjunctions in the total range. The species is presumed to have had its origin in or 52 REED C. ROLLINS ah NY / Fas : J\\ ™ SA Nf Sf OD * ) » OOe%,* * Oo * te a*ky cou t | O - x *o* | at * * «® ¥ ging ca t * at a * ° x e® > } * ° ao ae ag) county { Leavenworthia exigud Leavenworthia stylose Laqvenwarthia = terulosa Le) * a ® Leavenworthia uniflora re i 2. Detailed et of yaa species of Leavenworthia as they occur in e ; ral Basin of Tennessee. outer boundary of interruptedly exposed Ordovi- eian Hbestons has we ‘roughly Peete by a solid line adapted f map of middle Tennessee by C. W. Wilson, Jr. (Bu P 1. “9 Div, Geol. 48: pl. 2. 1940). ; Tenn mentors near the Central Basin of Tennessee because that is the only area where stock from which it could have arisen exists at the present time. From this general area, it has moved in all directions except directly to the west where no suitable habitats are present. The comparatively large geographical range of L. uniflora in an area that has largely been available for modern plant occupancy for a very long time argues for a relatively old dispersal for L. uniflora. Such an ancient LEAVENWORTHIA 53 dispersal, together with the greatest seed output of any of the species and the development of a high degree of auto- gamy, adequately explains the large geographical range of this species as compared to that of the other species of the genus. Area of Origin. — Putting together what we know about evolution within Leavenworthia, the evidence points to northern Alabama as the area of origin for the present day species. This area also appears to be the center from which the n= 11 group of species radiated. L. exigua is wholly to the north, east and south of this area. Morphologically this species ties in very closely with the self-compatible L. alabamica var. brachystyla, which is present in northern Alabama. The polyploid L. awrea, of Oklahoma and Texas, shows close morphological ties with L. exigua var. lutea of Alabama. This connection suggests a much more southerly migration route westward for L. aurea than that taken by L. uniflora in reaching northwestern Arkansas. If L. aurea or its progenitors had migrated along a more northerly arc, one would expect to find some remnant populations there at the present time. Certainly, L. uniflora persists happily in many places in the Ozarkian region. The n= 15 group of species shows a second pattern of migration based geographically on the Central Basin of Ten- nessee. If n = 11 or a number close to this is the basic stock out of which an L. stylosa type arose, this chromosome shift must have taken place in Tennessee. Once the n = 15 line was established it predominated over the earlier stock and then gave rise to the species of this group now present in the Central Basin area. These suggestions as to the area of origin and the subse- ouent migrations of the species are radically different from those of Baldwin (1945). However, it is well to point out that he did not know anything about most of the Alabama populations at the time of his study. Coincidence with Limestone. — The detailed mapping of Lceavenworthia in the Central Basin of Tennessee has been carried out over a period of years at intermittent times when I have beon in the field in that area. This was not done systematically over the entire land surface but each 54 REED C, ROLLINS time a population was examined, specimens were taken, notes on variability were recorded and the precise location of the population was determined and recorded. These rec- ords for the four species of Leavenworthia growing in the Central Basin have been the basis for Map 2. In addition to the symbols indicating the location of different populations, a line roughly marking the outer boundary of Ordovician limestone, taken from a progress geologic map of middle Tennessee by C. W. Wilson, Jr., was traced. The correspond- ence of Leavenworthia distribution and the limestone is - quite remarkable when the fact that a systematic mapping of Leavenworthia over the entire area has not been done is taken into account. It will be noted that the most restricted distribution is that of L. stylosa. This is the total distribu- tion of that species as we now know it. The most widespread species within the Central Basin is L. torulosa. All four species occurring in the Central Basin, L. exigua, L. stylosa, L. torulosa and L. uniflora, frequently are present at the same site. Within glades, there are slight preferences by the species. If there are very wet sites and slightly dryer ones, L. torulosa is most apt to be found in the wettest part of NORTHERN ALABAMA Map 3. Detailed distribution of Leave nworthia in Alaba i iflora excepted. For the distribution of L. un git hats ba say i a iflora, see Map 1, LEAVENWORTHIA 55 the glade. L. stylosa also occupies very wet sites in some places but, in competition with L. torulosa, it is likely to be more abundant in slightly dryer sites. Both L. exigua and L. uniflora occupy dryer places in competition with the other two species. They also flower earlier and the sites where they flourish may be quite wet when they are at their flower- ing peak. Later these become dryer. The soil is likely to be firm and merely moist at the time the spots where L. toru- losa and L. stylosa grow are still very wet. Sympatry and Allopatry.— The area where four of the seven species of Leavenworthia are sympatric is in the Cen- tral Basin of Tennessee (cf. Map. 2). This is also the cen- ter of the distributional area of the genus as a whole and the area where more individuals are to be found than in any other. However, the ranges of these species are not wholly coincident even in the Central Basin. L. stylosa, in particu- lar, does not occur with the other species in the southwest- ern portion of the Basin and L. torulosa is the only species I have been able to find in the extreme southern and south- western part. The fact of sympatry cannot be disputed but we have no information on whether this represents a pri- mary or secondary distributional development. It seems most probable that L. stylosa originated in the Central Basin since it does not now occur outside. It is also possible that both L. torulosa and L. uniflora originated in the same general area although both occur outside of the Basin and may have evolved completely away from the ter- ritory of L. stylosa. In that case, their present sympatry with it is a secondary development. However, the geological history of the Central Basin does not preclude the possibility of sufficient isolation for speciation to have taken place a- round the periphery of what was earlier a limestone dome but is now the Basin. In view of the high affinity of Leaven- worthia for wet habitats and limestone, it is likely isolation areas could develop near spring heads or stream basins where the limestone was uncovered. These could easily have been separated by some distance with rock mantle of a very different sort providing partial or complete isolation for a given population. The presence of L. exigua in a sympatric alliance with L. stylosa, L. torulosa and L. uniflora in the Central Basin 56 REED C, ROLLINS is unquestionably a secondary development. The origin of L. exigua is certainly traceable to the Alabama species of re . y fF OFGOWs. The complete geographic separation of L. awrea (Map 1) from all other species of the genus, combined with the fact that this species is also polyploid is of interest. But it is of even greater interest to consider the present distribution of this species itself as a possible model for the type of geographic isolation that might have obtained during an earlier period when primary speciation within Leavenwor- thia was taking place. The Texas population is certainly effectively isolated from the population of Oklahoma and it is the uniqueness of the habitat that is the real basis for this isolation. Given such a situation, even with the populations closer together geographically, the isolation necessary to en- able speciation to occur is provided. Similarly isolated pop- ulations occur in L. torulosa, L. uniflora and L. exigua. Other allopatric distribution patterns are provided by the Alabama species of Leavenworthia (Map 3). L. alabamica and L. crassa are essentially allopatric although the ranges of their self-compatible varieties overlap geographically. However, where this geographic overlap does occur, the populations of the different varieties are associated with different glades and in general they are separated spatially. In exceptional situations where the different taxa meet, they. hybridize as described above under the heading of inter- specific hybridization. EVOLUTION WITHIN THE GENUS Many items with evolutionary implications have been dealt with in the preceding pages since they are intimately tied up with the breeding and reproductive systems. How- ever, it now seems desirable to look at the evolutionary pic- ture in Leavenworthia as a whole. The first matter to be considered is that of direction. Where did the evolutionary developments we now see in the genus begin and where have they led? In contemporary materials, where time sequences cannot be readily established, the direction in which the evolutionary record is read becomes a difficult matter. In many studies the best that can be done is to extrapolate from the general morphology of the groups under study and LEAVEN WORTHIA 57 by comparative procedures try to arrive at what appears to be a reasonable evolutionary sequence. In Leavenworthia, where we have managed more intensive studies than can sometimes be made, we have a number of lines of evidence to draw upon. Two evolutionary lines in Leavenworthia are clearly marked by chromosome number. The three species with n — 15 have a well established starting point in L. stylosa. The strongest evidence for this lies in the fact that it is the only self-incompatible species of the group. The large odori- ferous flowers, with corollas ranging in color from yellow with an orange center through lavender to white with a yellow center, are also regarded as basic in the genus. The flower color of L. torulosa (white to light lavendér) and L. uniflora (white) are represented in L. stylosa so that one does not have to hypothesize an outside source. However, if it were assumed that either L. torulosa or L. uniflora were basic in the group, such a hypothesis would be necessary be- cause yellow flowers are not characteristic of any popula- tions in either of these species. It is a logical trend from large insect pollinated flowers toward smaller flowers where insects play a lesser role in pollination, and autogamy is of increasing importance. It is clear, in my estimation, that speciation proceeded from L. stylosa to L. torulosa and L. uniflora. L. uniflora is the most highly evolved species of the three and it may have arisen from L. torulosa or possi- bly from an ancestor common to both L. torulosa and L. uniflora. The second evolutionary line is characterized by the chromosome number n = 11. This is a group of three spe- cies with two of the three having self-incompatible compo- nents. The obviously derived species of the three is the wholly self-compatible L. exigua. Both L. alabamica and L. crassa are candidates for the role of progenitor of the group and the evidence is inconclusive as to which it might have been. If one examines these species from the point of view of which one possesses most completely the characters that ultimately show up in segregated form in the derived species, L. exigua, then L. crassa is favored. L. crassa is both white and yellow flowered and these colors show up in L. exigua, white being characteristic of var. erigua and yellow being 58 REED C. ROLLINS characteristic of var. lutea. Also, L. crassa could be the source of flower color in L. alabamica. A further test of L. crassa in the role of being basic in the group is provided if the assumption is made that the n = 11 cytological group preceded the group with n = 15. This appears to be a bet- ter possibility than the other way around, as I have sug- gested in the discussion under the heading of cytology a- bove. Under this assumption and with L. stylosa rather definitely placed in a basic position in the n = 15 group, we may then examine the morphological relationship be- tween L. stylosa as the derived species and L. crassa as the possible primal species. Here, the situation is satisfactory. All of the distinctive features of L. stylosa have their coun- terpart in L. crassa. These include flower color, fruit type, style length, seed markings, and many other features. Evidence derived from attempted interspecific crosses is of very limited value as regards interspecific relationships in Leavenworthia. L. crassa and L. alabamica were found to be highly interfertile but all of the other crosses gave either negative results or the F, hybrids produced were com- pletely sterile. The polyploid L. awrea produced F, hybrids reciprocally with L. crassa but in this case as in the others the hybrids were sterile (cf. Fig. 1). In summing up the evolutionary picture as we see it in L nworthia, we would place L. crassa in the role of the basic species from which L. alabamica and L. exigua form one line of development. The polyploid L. awrea was proba- bly derived from these components of the genus. In parti- cular, it would seem to have involved plants like L. exigua var. lutea in some way. The chromosome jump from L. crassa to L. stylosa is hard to understand and there is little or no evidence bearing on this problem. We have no alter- native other than to assume that such a change did occur and it follows that the rest of the evolutionary history of the n = 15 group is reasonable. Incipient Speciation. — Three different types of trends of intraspecific differentiation are readily seen in Leaven- worthia, These include a continuous type evident in L. sty- losa, a discontinuous geographically disjunct type found in L. torulosa and L. exigua, and a discontinuous but geograph- ically contiguous type present in L. crassa and L. alabamica. LEAVENWORTHIA 59 Although there is geographic disjunction in the other two species of the genus, L. aurea and L. uniflora, I have not detected any notable morphological distinctions between geo- graphically isolated populations. The continuous morphological trends in L. stylosa may well be a secondary situation developed from previously iso- lated portions of the species. The present situation is that three identifiable morphological types are present at three corners of a geographic triangle. The characteristics of these types are given below under the taxonomic treatment of the species. From the areas where each type prevails toward other areas of the triangle, there appears to be a more or less continuous modification of the distinctive fea- tures of one type toward that of the other. The situation in L. stylosa does not appear to be one in which there is at present any incipient speciation but rather one where specia- tion was well under way at some time in the past and the trend is now reversed. The previously distinctive parts of the species are now apparently in the process of fusion. Such a history for L. stylosa would also account for the tre- mendously wide range of variation now present in that species. In both L. torulosa and L. exigua, there are wholly iso- lated populations that show considerable morphological di- vergence from the semicontinuous populations making up the main distributional areas of each of these species. A good example of a possible incipient species is the yellow flowered L. exigua, isolated from the rest of the species and occurring only in a limited area of north central Alabama. The Georgia populations of L. exigua are distinctive. In greenhouse and experimental garden cultures, plants of Georgia origin can readily be distinguished from those of Tennessee. Similar divergent isolated populations of L. toru- losa are to be found in Kentucky and east of the Tennessee River in Tennessee. This type of isolation and ultimate di- vergence is to be expected in the usual speciation process and therefore it is not at all unique. But it should also be noted that isolation alone does not assure divergence in Leavenworthia. The best example is in L. uniflora. In that species, there are many glade isolated populations but the type of concomitant divergence easily seen in L. exigua and 60 REED C, ROLLINS L. torulosa does not show up. There are minor geographic trends in L. uniflora but these do not appear to be associated with any easily detected elements of speciation. Perhaps the most interesting type of incipient speciation in L thia is that found in L. crassa and L. alabam- ica. In these species, where most of the populations in each are self-incompatible. there are certain populations on the geographic margins of their ranges that have become self- compatible and at the same time morphologically distinct. We have given these latter populations nomenclatural status because they are readily distinguishable from the bulk of L. crassa and L. alabamica in each case. Detailed descrip- tions of these taxa are given below. The geographic situation differs in the two species in that the self-compatible populations of L. alabamica var. brachy- styla are not in geographic contact with the self-incompat- ible L. alabamica var. alabamica whereas the self-compatible populations of L. crassa var. elongata are in the same geo- graphic area as the self-incompatible L. crassa var. crassa. They are in different glades but the components of these glade systems are not wholly isolated. Evidently, evolu- tionary divergence has proceded in these instances without the benefit of full geographic isolation. The possibility of this happening is provided by a combination of the switch- over in breeding system and the semi-isolation of the glades. Although there is a difference in degree of isolation in each case, neither of these divergent varieties can be said to be sufficiently geographically isolated to insure genetic isolation over a long time-span. The fact that some and of L. alabamica g out notable morphol populations of L. crassa var. crassa ever, the potentiality LEAVENWORTHIA 61 change in breeding system and its operation against a back- ground of partial isolation to set the speciation process into full motion. The general importance of a situation such as exists in L. crassa and L. alabamica in the process of speciation can only be suggested to be greater than is at present recognized. Too few plant groups have been carefully examined for the role of the breeding mechanism in the initial stages of speci- ation to be properly evaluated on a broad basis. TAXONOMIC TREATMENT Leavenworthia Torrey, Ann. Lyc. Nat. Hist. 4: 87. 1837 Herbaceous glabrous annuals; early leaves simple with a slender aes later leaves progressively more pinnately lobed, mature leaves sette fo ormin 1g and lyrately pinnatifid; early flowers borne on erect Herren les originating at the center of the rosette of leaves; later flowers, if present, occurring in loose racemes borne by true stems originating in the axils of the rosette leaves; pedicels slender, straight, divaricately ascending; sepals nearly equal, greenish, non-saccate, spreading at right angles to erect during full anthesis; petals obovate to lingulate, truncate to deeply emarginate; corolla white with yellow center, yellow with orange center, lavender with orange center, or several less defined combinations of these colors; stamens strongly tetradynamous, single stamens with anthers always introrse, paired stamens with anthers extrorse, introrse, or between these extremes; siliques with a very short gynophore, nearly cacti flattened parallel to the septum to near- ly wae or globose, fleshy in some species; funiculus free; septum with a central nerve; seeds in a single row in the silique, flattened, winged to nearly wingleas: orbicular to longer than broad; seed coat with promi- nent reticulum of areolae; radicle of embryo short, straight to slightly appressed along margins of cotyledons; cotyledons orbicular. TYPE SPECIES: L. aurea Torr. KEY TO THE SPECIES A. Petals emarginate, yellow, white or lavender, 7-15 mm. long; leaf- lobes entire to shallowly dentate, terminal lobe markedly larger than distal lateral lobes; radicle of embryo usually straight or bent in tip, not adpressed to edges of cotyledons except in L torulos B. Sige, ‘sak torulose or only slightly so in L. stylosa; petals yel- low, white or lavender; areolae of seed coat smaller over em- bryo area than elsewhere; wing-like area of seed well developed. C. Petals 10-16 mm. long, deeply emarginate; styles 3-7 mm. long; siliques either thick or flat. D. Siliques thick, fleshy; styles 3-7 mm. long (sometimes shorter crassa var. aging At petals yellow, white or — E. Siliques 12-25 mm. long, 3-4 mm. wide; radicle of seed e 62 REED C. ROLLINS bryo slightly curved or sometimes straight; seed slightly elongated to nearly orbicular, cleft at one side of long axis 5. L. stylosa. E. Siliques 6-12 mm. long, 4-5 mm. wide; radicle of seed embryo short and straight; seeds orbicular or slightly longer than broad, cleft at the basal end 1. L. crassa. D. Siliques thin, flat; styles 3-5 mm. long (shorter in var. brachy- styla), petals white to lavender ...........:000+ 2. L. alabamica. C. Petals 7-10 mm. long, shallowly emarginate; styles 1-38 mm. long; siliques flat F. Siliques thin, not margined; terminal leaf-lobes angular; plants of Kentucky, Tennessee, Georgia and Alabama ........ 3. L. exigua. F. Siliques thick, margined; terminal leaf-lobes arenes plants of southeastern Oklahoma and eastern Texas aurea. B. Siliques conspicuously torulose even when young, ati white to ng-like area of seed very narrow or absent ...... 6. L. torulosa. A. sas entire, white, less than 7 mm. long; leaf-lobes deeply den- tate, terminal lobe only slightly larger than distal lateral lobes, radicle of embryo adpressed to edges of cotyledons ............:sesseee* 7. L. uniflora 1. Leavenworthia crassa Rollins, sp. nov. Glabrous winter annual; true stems present only in well developed plants, arising in the axils of basal leaves, decumbent, unbranched, 1-4 dm. long, always bearing a loose raceme of long-pedicelled flowers; early leaves usually entire or shallowly dentate; leaves of fully grown plants homemorphic, mostly basal, lyrately pinnatifid, 3-8 cm. long, terminal lobe 0.5-2 em. wide; lobes dentate, mostly paired, variable, terminal lobe conspicuously larger than the lateral lobes; early season flowers scapose, scapes 4-8 cm. long, later racemose flowers on slender pedicels 4-7 cm. long; sepals linear-cblong, 5-6.5 mm. long, spreading at right angles at full anthesis; petals obovate to broadly spatulate, deeply emarginate, 9-14 mm. long, 5-8 mm. wide, yellow with orange claw or white with yellow claw or with some intermediate combinations, blade spreading at right angles at full anthesis; siliques firm, globose to oblong, fleshy, 6-11 (-12) mm. long, 3.5-6 mm. wide, 2.5-5 mm. thick when fresh; styles 2.5-6 mm. long; seeds 2-8 per silique, dark brown, winged, nearly orbicular, 3-4 mm. in diameter; embryo straight or nearly so; areolae of seed-coat larger and more conspicuous on the wings and at base of wings than over the area of the seed occupied by the embryo, n= 11. Herba annua; foliis radicalibus rosulatis petiolatis lyrato-pinnatifidis lobatis glabris 3-8 em. longis; floribus seapiformibus vel cauliformibus; scapis axillaribus tennibes glabris 4-8 em. longis nudis 1-floribus; caulibus 1-3 foliatis axillaribus decumbentibus 1-4 dm. longis; sepalis lineari-oblongis 5-6.5 mm. longis; petalis obovatis vel late spathulatis emarginatis aureis vel candidis 9-13 mm. longis; siliques crassis gl0- LEAVENWORTHIA 63 bosis vel oblongis 6-11 (-12) mm. longis, 3.5-6 mm. latis; stylis 2.5-6 mm. longis; seminibus orbicularibus 3-4 mm. in diametro uniseriatis; cotyledonibus planis. Type in the Gray Herbarium collected on thin soil over limestone, 1 mi. east of McKendry, Morgan Co., Alabama, April 4, 1959, Reed C. Rollins 5923. KEY TO THE VARIETIES Siliques 6-10 mm. long; styles 3-6 mm. long; petals white to yellow, 10-13 mm. long la, var. crass. Siliques 8-12 mm. long; styles 2.5-3.5 mm. long; petals yellow, 9-11 mm. long 1b. var. elongata. la. L. crassa var. crassa Plate 5, fig. 4&5; Plates7& 8 DISTRIBUTION: southeastern Lawrence and southwestern Morgan Counties, Alabama. (Map 3). LAWRENCE COUNTY: 2.5 mi. west of Dan- ville, Rollins 5931 (GH); 8 mi. east of Wren, Rollins and Chambers 5713; 57139 (GH); 4 mi. southwest of Danville, Rollins 6125 (GH); 1% mi. southeast of Speake, Lloyd 204 (GH). MORGAN COUNTY: 4% mi. north of Danville, Rollins 5983 (GH); same general locality, Rollins and Chambers 5721; 57141; Rollins and Channell 5628 (GH); 1 mi. east of McKendry, Rollins 5991 (GH, topotype); 1 mi. northeast of McKendry, Rollins 5989; 61173 (GH); 3 mi. southeast of McKendry, Rollins 5927 (GH); 5 mi. west of Falkville, Rollins, Solbrig, Hilferty and Lloyd 6015 (GH) ; 3 mi. southwest of Massey, Lloyd 268 (GH); 1% mi. southeast of Massey, Lloyd 299 (GH); 1 mi. southwest of Lebanon, Lloyd 252 (GH). 1b. L. crassa var. elongata Rollins, var. nov. Herba annua; siliquis 8-12 mm. longis, crassis; stylis 2.5-3.5 mm. longis; petalis aureis, 9-11 mm. longis. e in the Gray Herbarium collected in a pasture near limestone uarry, 4 mi. south-southwest of Lebanon, Morgan County, Alabama, April 6, 1961, Reed C. Rollins 6112. DISTRIBUTION: southwestern Morgan County, Alabama. (Map 3). MORGAN Co.: near limestone sink, 6 mi. southwest of Falkville, Rollins, Solbrig, Hilferty and Lloyd 6021 (GH) ; old cornfield, 1.5 mi. southeast of Massey, Lloyd 299 (GH); 2 mi. southwest of Lebanon, Lloyd 277 (GH) ; 8 mi. west of Falkville, Rollins 5985 (GH). The great abundance of plants of Leavenworthia crassa in the system of cedar glades and adjacent fields and pas- tures in southwestern Morgan County, Alabama, makes it difficult to understand how this species could have been overlooked for so many years. When in full flower, it is the most conspicuous plant present in many localities. There is abundant evidence that the species is made up of a highly complex series of populations, each deviating to some degree 64 REED C. ROLLINS from the other. Apparently the presence of isolated or semi- isolated glades has permitted the evolutionary development of more or less distinctive populations in association with each of these glades. It is difficult to reconstruct the primitive situation because of the widespread destruction of the original vegetation through land use. However, it is fairly obvious that the present local distribution patterns of the populations are not the same as they were prior to the agricultural and do- mestic use of the land by man. Fortunately, there are rem- nants of populations remaining in many glade sites that have not been too drastically altered. These we have been able to identify as the well-springs of a number of different population types now more widely spread into fields and pastures. In some places, plants of these differing popula- tion types are actively intercrossing to produce a broader range of recombinant characteristics than may be seen in the individual glade sites themselves. Considering the species as a whole, there is a wide range in flower color. In some populations the flowers are white or light lavender with a yellow center. Others are yellow with an orange center. In most white or lavender-flowered popu- lations, the admixture of yellow-flowered plants is consider- able. In at least one population, 5 mi. west of Falkville, which was sampled in two different years with the same re- sults, yellow and white flowers were present in approxi- mately equal numbers. On the other end of the scale, yellow- flowered populations usually have few to many white-or light lavender-flowered plants present. The one exception, so far as we know, is var. elongata. Here the flowers are yellow. The yellow portion of the petal blade of var. crassa is not wholly consistent. Usually the major part of the blade, in- cluding the deeply notched extremity, is yellow. But there are some plants with white or cream-colored bands on the blades, tips of blades, etc. It appears that petal color is in- herited in a complex way in this species. Within L. crassa, there is a complete range from self- incompatibility to self-compatibility in the breeding system. Some populations are predominantly self-inecompatible with partial self-compatibility present while others are predom- sey LEAVENWORTHIA 65 inantly self-compatible with some plants being self-incom- patible.* The populations of var. elongata so far tested are all self-compatible. Furthermore, the anthers are in an introrse position. It seems quite certain that the develop- ment of self-compatibility and its predominance in some populations is part of a general evolutionary trend in Lea- venworthia and in L. crassa specifically this has contributed to the differentiation found within the species. he development of facultative autogamy accompanying the inversion of the paired anthers to an introrse position and a reduction in flower size have taken place in var. elongata. In one small patch of plants of this variety (Pop. 5985) found in the middle of a large population of L. ala- bamica var. brachystyla there was no evidence of inter- specific hybridization. Progeny tests of plants of var. elong- ata and of immediately adjacent plants of var. brachystyla did not reveal any hybrids. The progenies from plants of both varieties were unusually uniform. These results were rather unexpected because in other localities where L. crassa var. elongata and L. alabamica var. brachystyla come to- gether, hybridization does occur and heterogeneity is found among the plants where the two varieties are in contact. A probable interpretation of this situation is that both var. elongata and var. brachystyla are largely autogamous and that within species pollination is reinforced to some extent by the slightly earlier flowering of var. elongata. The net result is the presence of a uniform colony of var. elongata in the midst of var. brachystyla with no evident gene inter- change between the two species. The fruit type of L. crassa is most similar to that of the yellow-flowered L. stylosa found in Davidson County, Ten- nessee. It is fleshy and hard when fresh. The measure- ments given in Table 4 indicate the extent of this thick, fleshy development in relation to other dimensions of the siliques. In var. crassa, when it is growing on glade sites or in relatively unfavorable places, the siliques are often nearly globose. In more favorable places, such as fields where fer- tilizers may have been used, or in deeper soils, the plants be- come larger (over a foot in diameter) and the siliques are "Mr. David Lloyd is making a detailed study of the breeding mechanism of L. crassa, The results will be presented at a later time. 66 REED C. ROLLINS 7. Specimens showing some PLATE of the variation in flowering material of a population (5991) of L. crassa var. crassa. LEAVENWORTHIA 67 Oo fF ete §F ve §& £ 1 Specimens showing some of the variation in fruiting material of a Population (5989) of L. crassa var. crassa 68 REED C. ROLLINS also larger and more elongate. However, in the latter, the style is also relatively long. In var. elongata, the siliques are even more elongate and the styles are considerably shorter. There is a very considerable response of the plants of L. crassa to favorable growing sites. This is shown in Plate 8 where the smaller, few-flowered plants were collected from nearly bare rock and the more luxuriant plants were taken from adjacent deeper soils. The range from a single-flow- ered plant to a many-flowered plant is illustrated in the plate and, in general, this reflects the range of the site from un- favorable to favorable. The larger plant at upper left and lower right show the more elongate fruits mentioned above. The specimens shown in Plate 7 are from the type station of the species which is a relatively favorable growth site. Aside from the difference in chromosome number between L. crassa (n = 11) and L. stylosa (n = 15), the two species show a very different and distinctive variation pattern within themselves. If fruit shape and style length alone are considered, it may be readily seen from the data in Table 4 that while the shortest fruits of L. stylosa and the longest fruits of L. crassa-are somewhat comparable in size and shape, the variation trends in L. stylosa are all from these dimensions towards longer fruits with shorter styles and those of L. crassa are towards shorter fruits with longer styles. It is not practical at this time to try to reflect taxonomi- cally and nomenclaturally the full complexity shown by the different populations of L. crassa. The comparative differ- ences observed, as indicated above, are thought to be (as with L. stylosa) associated with the development of different populations systems in isolated or semi-isolated glades, each major glade or glade system in effect producing a recog- nizably different population or group of populations. 2. Leavenworthia alabamica Rollins, sp. nov. Glabrous winter annual; true stems present only in well developed plants, these arising in the axils cf the basal leaves, 1-2.5 dm. long, decumbent; early leaves with a relatively long petiole and nearly entire terminal lobe, orbicular to broadly ovate in shape, leaves of fully grown plants mostly basal, lyrately pinnatifid, 3-10 cm. long, lobes subentire to dentate, terminal lobe markedly larger than the lateral mostly paired variable lobes; early flowers scapose, scapes 4-10 cm. LEAVENWORTHIA 69 long; later racemose flowers on slender pedicels up to 5 cm. long; sepals narrowly oblong, 5-7 mm. long, spreading at right angles during full anthesis; petals spatulate, deeply emarginate, 10-14 mm. long, 5.5-7.5 mm, wide, blade white to light lavender, sometimes deeper lavender, spreading at right angles during anthesis, claw yellow to orange- yellow; siliques narrowly oblong, strongly flattened parallel to septum, n-fleshy, acute to rounded at base and apex, 1.5-2.5 em. long, 3-4.5 mm. wide, 1.5-2.5 mm, thick; styles 1.5-5.5 mm. long; seeds dark brown, nearly orbicular, winged, 3-4 mm, in diameter; embryo straight, areo- lae of seed-coat larger and more conspicuous on the wings and at base of wings than towards center of seed 11, Herba annua; foliis radicalibus rosulatis lyrato-pinnatifidis lobatis glabris 3-10 cm. longis; floribus scapiformibus vel cauliformibus; cauli- bus axillaribus decumbentibus 1-2.5 dm. longis; scapis tenuibus ‘evectis 4-10 em. longis; sepalis lineari-oblongis 5-7 mm. longis; petalis spa- thulatis pyres siliquis anguste oblongis compressis 1.5-2.5 em longis, 3-4.5 mm. latis non crassis; stylis 1.5-5.5 mm. longis; seminibus uae ete a orbicularibus 3-4 mm. in Sigetee cotyledoni- bus plan Type in or Gray Herbarium collected in a cedar glade, 1 mi. north of Cherry Hill Cressing, about 8 mi. east of Russellville, Franklin County, Alabama, April 28, 1959, Reed C. Rollins 5980. Y TO THE VARIETIES Styles 2.5-5.5 mm. long; siliques acute above and below 2a. var. alabamica. Styles 1.5-3 mm. long; siliques rounded above and below .-...-.cscerere 2b. var. brachystyla. 2a. L. alabamica var. alabamica Plate 9 DISTRIBUTION: Colbert, Franklin and Lawrence Counties, Alabama Map 3). COLBERT CO.: 5 mi. west of Tuscumbia, Rollins, Cooley and Brass 55102 (GH); about 6 mi. west of Tuscumbia, Rollins and Chan- nell 5641 (GH); Red Hill, 13 mi. west of Tuscumbia, Rollins, Cooley and Brass 55106 (GH); 4 mi. southwest of Tuscumbia, Rollins, Solbrig, Hilferty and Lloyd 6024 (GH). FRANKLIN CO.: 5 mi. east of Russeli- Ville, Rollins 5978 (GH); 7 mi. east of Russellville, Rollins 5549; Rol- lins and Chambers 57135 (GH); 1 mi. south of Newburg, Rollins, east of Landersville, Rollins and Chambers 5713; 57187 (GH); 5 mi. northeast of Landersville, Rollins 6128 (GH); 4 mi. west of Mculton, Rollins 5543; Rollins and Chambers 5712 (GH); 3 mi. south of Hatton, Rollins and Channell 5642 (GH); western border of Lawrence Co., Rol- lins and Channell 5633 (GH); 2 mi. west of Mt. Hope, Lloyd 923 (GH). 2b. L. alabamica var. brachystyla Rollins, var. nov. Herba annua; siliquis compressis enervatis obtusis 1.5-2.5 em. longis; stylis 1.5-3 mm. longis. 70 REED C. ROLLINS Specimens showing characteristics of plants and some of the variation PLATE § ina ES (5980) of L. alabamica. LEAVENWORTHIA 71 Type in the Gray Herbarium collected in a pasture in a cedar glade area, 4 mi. southwest of Falkville, Morgan County, Alabama, April 26, 1960, Reed C. Rollins, Otto T. Solbrig, Frank Hilferty and David Lloyd 6019. DISTRIBUTION: scuthern Morgan County, Alabama (Map 3). Mor- GAN CO. mi. south southeast of McKendry, Rollins 61169, 61170 (GH); 3 mi. southwest of Falkville, Lloyd 685; Rollins 5926; 5984 (GH); 3 mi. south southeast of Massey, Lloyd 275, 288, 290 (GH); 0.8 mi. east of Lebanon, Rollins 61172 (GH) ; about 5 mi. west cf Falkville, Rollins 5932, 5990 (GH). The relatively long styles on the fruits of L. alabamica naturally led early botanists handling limited and scrappy material to refer this to L. stylosa. The same interpretation has come down to the recent past as indicated by the distri- bution map (fig. 8) of Baldwin (1945). Unfortunately, Baldwin did not have growing material of L. alabamica as he did of several of the other species or he surely would have discovered that the chromosome number is n = 11 in- stead of n = 15, as he found in L. stylosa. In general, the style length is considerably shorter in L. alabamica than in L. stylosa but a more important difference is in the fleshi- ness of the silique. This characteristic is not observable in herbarium specimens. However, in fresh plants, the siliques of L. stylosa are thick, fleshy and firm, whereas in L. ala- bamica they are relatively thin and non-fleshy. In this re- spect the siliques of L. alabamica are more like those of L. exigua than any other species of the genus. The measure- ments given in Table 4, made on fresh material, show the basis for the above statement. The pattern of differentiation within L. alabamica is per- haps less complex than that found in L. crassa but there are some marked trends that are readily observable. The most striking is in the breeding mechanism. In most of the popu- lations of var. alabamica tested, including some from both Franklin and Lawrence Counties, self-incompatibility pre- dominates. On the other hand, self-compatibility is the rule in var. brachystyla. The results of self-compatibility tests are given in Table 2. A more recent result, found by David Lloyd, shows at least one population of var. alabamica to be somewhat self-compatible. Undoubtedly a more exhaustive study of this species as a whole would show a more complex situation than is at present known. Our interpretation is that there is an evident evolutionary trend from self-in- 12 REED C. ROLLINS compatibility towards self-compatibility presently going on within the species. This trend has culminated in a high de- gree of self-compatibility in var. brachystyla, where a paral- lel change of the anthers of the paired stamens to the in- trorse position has produced the appropriate circumstances for facultative autogamy to occur. Our naming of the self- compatible populations of Morgan County as var. brachy- styla is a recognition of the culmination of this trend but should not be assumed to indicate that we are not aware of intermediate stages in the trend itself, iia aad in the most easterly populations of var. alabamica The most northerly populations of var. alabamica in Col- bert County are distinguishable from the plants of the Franklin County area which are in turn slightly different from those of Lawrence County. Again, as in L. stylosa and L. crassa, there appears to have been prolonged isolation or semi-isolation between the major glade systems in the region gue each of these has spawned a series of populations of rthia slightly different from the others. In the ter- ss of L. alabamica, one can distinguish four such glade systems, each with several to many individual glades that are semi-isolated in themselves. The most disrupted area and the one with the smallest glades is the glade system west and southwest of Tuscumbia in Colbert County. These are only one level up from the Tennessee River and are at a lower elevation than the glade system centered just to the east and southeast of Russellville in Franklin County. The latter is toward the Bankhead National Forest which oc- cupies a low plateau to the south. Extending from this plateau towards the Tennessee River in scallop-like fashion are low ridges that cover the underlying limestone with a mantle of soil. One such ridge separates the Russellville glade-system from the one centering on Landersville in Lawrence County. A similar low relief exists between the Lawrence County glades and those of Morgan County where var. brachystyla abounds. Flower color in the various populations of L. alabamica does not show the kind of variation found in L. crassa and L. stylosa. In fact, yellow-flowered plants are extremely rare, having been seen in only two populations and then only a plant or two. It is probable that these observed cases rep- 2] LEAVENWORTHIA 73 resent mutants that have arisen sporadically and rarely. A comparable situation is present in L. torulosa where, in only one of the numerous populations I have examined, there were less than a half-dozen yellow-flowered plants. In many populations of L. alabamica, the predominating flower color is white with a yellow center. In these, the color changes to light lavender as the flowers close in response to the diminished light of late afternoon or of deeply overcast skies. This is due to the fact that the upper surface of the petal limb is white and the lower surface is light lavender. When the petals are fully extended, the white surface is presented but when the petal closes to the erect position, the lavender back of the petal becomes visible. In a few popu- lations the lavender is somewhat more intense than in the majority. An interesting situation was observed in several populations where the upper surface of the petal limb was light lavender and the lower surface was white. The fact that whole populations. consisting of hundreds of thousands of plants, bore a similar petal-color pattern suggests that some adaptive significance to it must exist but I am not able to suggest what this might be. There were no easily seen differences in the type of habitat or in the geographic con- tinuity of one petal-color pattern as compared to the other. Adding to the puzzle is the fact that the same situation occurs in L. torulosa, except that a light lavender upper surface of the petal limb with a white lower surface pre- dominates in most of the populations and in only a few populations is the upper surface white and the lower light lavender. Here the predominant color pattern is reversed from that of L. alabamica. T,, alabamica < L. crassa Of all attempted crosses between species of Leavenwor- thia, the only really successful one involved the two species above. Also, where L. alabamica var. brachystyla comes into contact with L. crassa in Morgan County, hybridization occurs. Insofar as our present knowledge goes, natural inter- Specific hybridization occurs only in that area. Both L. crassa var. crassa and var. elongata appear to be involved in crosses with L. alabamica var. brachystyla but truly hybrid Swarms are rare. What we have seen most frequently is a thin band of 74 REED C. ROLLINS what appear to be F, hybrids along the contact line between the two species. In general the species themselves remain unadulterated at only a short distance of a few meters away from the line of contact. At the present time we do not know whether genetic factors tend to reduce or eliminate potential introgressants or whether the particular areas of contact studied are relatively new and the opportunity for a more normal hybrid situation has not had time to develop. The sites where hybridity appears to be most prevalent are fields or disturbed pastures. We have not seen any evidence of interspecific hybridization on undisturbed land. Some herbarium collections of hybrids of L. alabamica X L. crassa are the following, all from Morgan County, Alabama: : 1. L. alabamica var. brachystyla < L. crassa var. crassa. — 134 mi. southeast of Massey, Lloyd 917 (GH); 3 mi. southeast of Massey, Lloyd 703 (GH); 1% mi. east of Massey, Lloyd 29 (GH); near Harrin Creek, 2 mi. south of McKendry, Rollins 6122 (GH )s ‘ 2. L. alabamica var. brachystyla * L. crassa var. elongata. — 14 mi. southwest of Lebanon, Lloyd 254 ; 255 (GH); 1% mi. south of Lebanon, Rollins 6110 (GH). 3. Leavenworthia exigua Rollins, Rhodora 58:75. 1956 Glabrous winter annual; early leaves with a remote shallowly den- tate terminal lobe, leaves of fully grown plants lyrately pinnatifid, 1.5-5 (6) cm. long, lobes shallowly dentate, terminal lobe markedly on flowers scapose, scapes 3-6 cm. long, later flowers often borne on lateral decumbent branches; young buds pendant, older buds erect; sepals nonsaccate, linear-oblong divergent to erect during anthesis, greenish to light lavender, 3.5-5.5 ight lavender, upper claw yellow- ish; petals yellow in var. lutea; siliques strongly flattened parallel to septum, non-fleshy, obtuse above and below, oblong, 1-2 em. long, 3.5- 5.5 mm. wide, 1.5-2.5 mm. thick when fresh; style 1-3 mm. long; gyno- phore evident, nearly 1 mm. long; seeds 4-10 per silique, winged, nearly flat, irregularly orbicular, 3-4 mm. in diameter, areolae of seed-coat reticulum much smaller over embryo than in wing area; radicle straight; cotyledons orbicular. n = Ti; KEY TO THE VARIETIES Petals white to light lavender, plants of Kentucky, Tennessee and orgia. Styles 1-2 mm. long, sepals light lavender, plants of Tennessee and PUREED | steseaternesrarontbegnerowsvetsancoretersiapst i, ene WEE a. var. exigud. Styles 2-3 mm. long, sepals green, plants of Kentucky Nee Bi ateicl i lr hLo eee Cee eeeeereereeeweaee rents esevaveblvuigicea tides, & b. var. laciniata. Petals yellow, plants. of Alebams sic iciicccsbicuis 3c. var. lutea. LEAVENWORTHIA 75 3a. L. exigua var. exigua Plates 10 & 11 Type in the Gray Herbarium, collected in shallow soil of glade, 1 mi. north of the North Fork of the Duck River, southeast of Unionville, Bedford County, Tennessee, April 3, 1955, Reed C. Rollins 5535. DISTRIBUTION: Central Basin of Tennessee; northwest Georgia (Map 1). Tennessee. BEDFORD Co.: west of Shelbyville, Sharp, Felix and Adams 11244 (GH); 1 mi. southeast of El Bethel, Rollins 59100 (GH). DAVIDSON Co.: west of Couchville Pike, Svenson 7715 (GH, mix- ture with L. uniflora) ; 1.5 mi. east of Una, Rollins 6139 (GH) ; 2.4 mi. north of Lavergne, Rollins and Chambers 5744 (GH); south of Nash- ville, Shanks and Sharp 383 (GA, GH, MICH, NY, US). MARSHALL CO.: 6 mi. southeast of Eagleville, Rollins 6133 (GH); 2 mi. south of Chapel Hill, Rollins and Chambers 5729 (GH); 1.5 mi. north of Chapel Hill, Rollins 59109 (GH). MAURY CO.: 4 mi. west of Pottsville, Rollins, Sol- brig, Hilferty and Lloyd 6027 (GH); 6 mi. north of Culleoka, Rollins 5328 (GH); 3 mi. northeast of Chapel Hill, Rollins and Channell 5663 (GH); 1 mi. north of Rally Hill, Rollins 5919 (GH). RUTHERFORD CO.: 5 mi. north of Jerusalem Church, Rollins 6142 (GH); 2 mi. south of Lavergne, Rollins and Chambers 5742 (GH); 3 mi. south of Lavergne, Rollins and Channell 5617 (GH). WILLIAMSON Co.: 3 mi. east of Nolens- ville, Rollins and Channell 5619 (GH); Rocky Spring road, 2 mi. east of Nolensville, Rollins 5512 (GH). WILSON CO.: 2 mi. west of Lebanon, Rollins, Solbrig, Hilferty and Lloyd 6007 (GH); 3 mi. south of Glade- ville, Rollins and Chambers 57149 (GH); Lebanon, Lloyd 413 (GH). Georgia. CATOOSA Co.: 2 mi. east northeast of Fort Oglethorpe, Dun- can, McDowell and Runyon 12124 (GA, GH); 1.9 mi. east of Fort Ogle- thorpe, Duncan and McDowell 12235 (GA, GH) ; 1 mi. east of Fort Ogle- thorpe, Rollins and Chambers 5704 (GH) ; 2 mi. east of Fort Oglethorpe, Rollins and Channell 5650 (GH); eastern entrance to Chickamauga — Chattanooga National Military Park, Rollins and Chambers 5702 (GH); same locality, Rollins 5997 (GH). WALKER CO.: 11 mi. north of LaFayette, Rollins 5996 (GH); same locality, Rollins and Chambers 5705 (GH). 3b. Leavenworthia exigua var. laciniata Rollins, var. nov. Herba annua, sepalis viridis, siliquis 1.5-2 cm. longis, 3-4 mm. latis, Stylis 2- : is. a figs song re flowers white with light yellow sotto paired stamens introrse; thin soil over limestone, cedar glade, south o the Salt River, 6 mi. east of Sheperdsville, Ridge Road, 2 mi. north - state route No. 480, Bullitt County, Kentucky, April 1, 1962, Reed C. Rollins 6201. : pion! Bullitt County, Kentucky (Map 1). Open field, Ridge Road, March 27, 1954, H. A. Korfhage 2 (Univ. of Louisville). 3c. L. exigua var. lutea Rollins, var. nov. Plate 12 is iuteis, siliquis planis. Be revea ee towel Iden yellow, pasture among Type in the Gray Herbarium; flowers go 76 REED C, ROLLINS PLATE 10. Specimens showing characteristics of plants and v: ariation in a popula- tion (5744) of L. exigua var. exigua of Tennessee Sp LEAVENWORTHIA ~l ~] ecimens showing characteristics of plants and some of the variation ina Rae 7s (5650) of L. exigua var. exigua of Georgia. 78 REED C. ROLLINS cedars, 1 mi. northwest of McCalla, Jefferson County, Alabama, April 8, 1956, R. C. Rollins and R. B. Channell 5643. DISTRIBUTION: Jefferson County, Alabama (Map 3). Western edge of Bessemer, Rollins 5933 (GH); Bessemer, E. J. Palmer 38975 (GH, us); 1 km. north of McCalla, R. T. and E. R. Clausen 5760 (NY, UC); 1 mi. northeast of McCalla, G. Cooley and F. Harper 4524 (GH); 3/4 mi. northeast of McCalla, R. M. Harper 3834 (GH, NY, US). The numerous populations of L. exigua in the Central Basin of Tennessee and those of northwestern Georgia vary little in flower color. The main portion of the petal is most frequently white but may be light lavender in many popu- lations. A yellowish band on the upper claw of each petal gives the flower a yellow center. All yellow-flowered popu- lations are found in the Bessemer and McCalla areas of Jefferson County in north central Alabama. These we have designated var. luted Ever since the establishment of the thia by Torrey (1837), who referred ma- terial from Jefferson County, Alabama, to L. aurea, the presence of these yellow-flowered populations has been one of the main bases for a misunderstanding of L. awrea. These plants do resemble those of L. aurea in many respects but the chromosome number is n = 11 instead of n = 24. Ac- tually, the thin, similar siliques and angular leaf-lobes, to- gether with the same chromosome number, show unequivo- cally that, although yellow-flowered, these Alabama plants belong with L. exigua. At the time I described L. exigua (1956), I did not have chromosome counts of L. awrea, or of L. exigua var. lutea. The fact that L. aurea is a polyploid (n = 24) firmly sup- ports the separation of L. exigua as a distinct species and permits a more reasonable interpretation of the geography of the two species. As information concerning L. exigua has been. gradually accumulated, it has become clear that this species is in many respects the evolutionary counterpart of L, uniflora but in the eleven paired chromosome series in- stead of the fifteen paired series. Next to L. uniflora, L. exigua has the widest geographic range of any other species. Also, it follows close to L. uniflora in not only being self- compatible but in being autogamous as well. The flowers are the smallest in the eleven paired chromosome series and they lack fragrance, two features also characteristic of L-. uniflora. LEAVENWORTHIA 79 PLate 12. Specimens showing characteristics of plants and some of the variation 33) of Os in a population (59: L. exigua var. lutea 80 REED C. ROLLINS The isolation and differentiation of populations outside of the Central Basin of Tennessee are an interesting feature of L. exigua. The main area of distribution, as shown in Map 2, is made up of a series of populations on appropriate cedar glades that occur in a complex pattern nearly the full length of the Central Basin. About one hundred and fifty miles to the north, with no known stations of L. exigua in between, a rather extensive“population of plants with cer- tain distinguishing characteristics is to be seen. This area in Bullitt County, Kentucky, is in need of further explora- tion, for my attempt at field work there in early April, 1962, was under very unfavorable weather conditions. Plants were brought back to the greenhouse where they were grownh to maturity and a chromosome count of n = 11 was obtained. Beyond that, the plants were not worked with intensively as was the case with representatives of other populations of the species. This Kentucky population is distinctive on the basis of longer styles, green instead of light lavender sepals and more sharply divided leaves as compared to var. exigua and we have named it var. laciniata. To the south of var. exigua, isolated by about a hundred miles distance, is var. lutea centered in Jefferson County, Alabama. Several populations of this yellow-flowered va- riety have been studied and they were found to be strongly self-compatible and self pollinating. Here, also the yellow flower color is a fixed feature, not present elsewhere in the species. The populations of var. exigua in northwestern Georgia are isolated from the main distributional area of this variety in Tennessee and slight differences in growth habit and other characteristics are noticeable in them. It is evident that L. exigua as a whole shows a considerable range of variation, with the most extreme forms isolated at some distance from the main distributional area. Without doubt, the self-compatible and at least facultatively self pollinating breeding system has been important in enabling this diver- sity to develop. Some comment should be made concerning a specimen of var. exigua in the University of Michigan Herbarium. This specimen is from the private herbarium of Geo. L. Ames; M. D., and was collected in Tennessee in March, 1855. The locality given appears to be Decatur County but the writing eke iO ete PEE ES LEAVENWORTHIA 81 is not as clear as one would like it to be. If, as is possible, this specimen did come from Decatur County, the range of var. exigua is considerably west of that shown on my map (Map 2). We have not seen any other specimens from west of the Tennessee River and I do not know whether suitable glade sites occur there. 4, Leavenworthia aurea Torrey, Ann. Lyc. Nat. Hist. N, ¥. 43.88. 1837 Plate 13 Glabrous winter annual; early leaves with nearly orbicular and entire terminal lobes, later leaves long-petioled, thick, lyrately pin- natifid but usually with only a few lateral lobes, terminal lobe much larger than lateral, lobes entire to shallowly dentate, mature leaves 2-8 cm. long, 1-1.5 em. wide; early and mid-season flowers scapose, scapes 3-9 cm. long; later flowers usually borne in a raceme on lateral decumbent branches, pedicels relatively thick, 3-6 em. long; sepals greenish turning light lavender upon drying, non-saccate, narrowly oblong, 4-5 mm. long; petals lemon yellow to orange yellow, changing to light lavender upon drying, narrowly lingulate, very shallowly emarginate, 7-10 mm. long, 3.5-5 mm. wide; anthers of paired stamens introrse; siliques strongly flattened parallel to septum, thickish, erect, 1.5-3 cm. long, 4-5.5 mm. wide, 2-3 mm. thick when fresh; styles 2-3.5 mm. long; gynophore evident, nearly 1 mm. long; seeds 5-11 per silique, dark brown, nearly orbicular, flattened, flat on inner face, low dome- shaped on outer face, narrowly winged, 3.5-4.5 mm. in diameter, areolae of seed-coat reticulum much smaller over embryo than in wing area; radicle of embryo straight; cotyledons nearly orbicular. n = 24. Type in the New York Botanical Garden — near Ft. Tow- son, Oklahoma, M. C. Leavenworth s.n. Isotype G DISTRIBUTION: extreme southeastern Oklahoma and in the vicinity ‘of San Augustine, Texas (Map 1). Oklahoma. cHocTaw co.: 1 mi. west of Ft. Towson, Rollins 6152 (GH); 9 mi. east of Hugo, Rollins 5972 (GH); 5 mi. west of Ft. Towson, G. T. Robbins 2872 (uc); same locality, McVaugh 7618 (GH, MICH). MC CURTAIN CO.: 4 mi. north of Idabel, Rollins 5974 (GH); 3.2 mi. west of Idabel, Rollins and Cham- bers 5769 (GH) ; 9 mi. west of Idabel, Rollins 5973 (GH). Texas. With- out locality, M. C. Leavenworth (NY); 1 mi. east of San Augustine, SAN AUGUSTINE CO., D. S. and H. B. Correll 23543 (GH, LL) ; near same locality, D. S. and H. B. Correll 24817 (GH, LL). I have deliberately refrained from repeating the speci- men citations given in my previous paper (Rollins, 1956) except for the type. The type sheet at the New York Botani- cal Garden apparently has material collected by Leaven- worth both in Texas and near Ft. Towson, Oklahoma, on it, although one cannot be completely certain of this. We do know from Torrey’s own words that he received material REED C. ROLLINS 82 of plants anc icteristices PLATE 13, Specimens showing chars in a population (5771) of L. aurea. LEAVENWORTHIA 83 collected by Leavenworth from C. W. Short and also direct from Leavenworth. His statement (op. cit.) is as follows, “For my first knowledge of the L. aurea I am indebted to my friend Dr. Short of Kentucky, who shared with me the specimens he received from Dr. Leavenworth, its discoverer. Subsequently I received some excellent specimens from Dr. L. himself.” It is interesting that Leavenworth managed to collect Leavenworthia aurea from the only areas in which it is known to occur today. The Texas location is very re- stricted in size, being limited so far as we know to San Augustine, where E. J. Palmer collected it in 1915 and again in 1918, and to an area just east of San Augustine where the Corrells made collections in 1961 and 1962. I have searched for the species at San Augustine several times and on the last occasion I found the characteristic wet limestone pave- ment with Nostoc, Sedwm, and Opuntia present, all usual associates of Leavenworthia but it was too late in the season to find L. aurea. Leavenworth must have collected his Texas specimens at or near San Augustine. It is known from a letter from Leavenworth to Torrey (cf. McVaugh, 1947) that he collected plants at St. (now San) Augustine in Texas on at least one occasion. While stationed at Fort Jesup and nearby Camp Sabine in western Louisiana, Leavenworth traveled into Texas several times. We had one lot of seed collected by Correll from the Texas locality but we have not had any growing plants because the seed would not germinate. For this reason, our chromo- some count for the species is based on collections from south- eastern Oklahoma. The fact that this number, n = 24, does not show a strict polvploid relationship to the known counts for other species of Leavenworthia, makes it highly desira- ble to obtain counts from plants of the isolated Texas popu- lation. Such information might provide a clue as to the ori- gin of L. aurea. As it is, there is only morphological evi- dence on which to base a suggestion as to its possible rela- tionship. In this respect L. awrea is nearest to L. exigua var. lutea which is the only yellow-flowered population of that species. Var. lutea, with thinner siliques, shorter styles and a chromosome count of n = 11, is also highly restricted geographically, being confined to cedar glades in Jefferson County, Alabama. 84 REED C. ROLLINS The thick leaves, rather stout pedicels or scapes and gen- eral succulence of L. aurea are characteristics often found in polyploids. The relatively high chromosome number of this species is surely the result of polyploidy and it seems quite certain that plants of the L. exigua var. lutea type entered into the combination that ultimately produced L. aurea. Whether another species or some other divergent type of L. exigua also figured in that combination is a matter of pure conjecture. Considering L. aurea from the stand- point of its present remote geographic position, far to the west of what must have been the center of origin and dif- ferentiation of the genus, the question of its route of mi- gration inevitably arises. Two items bear on this question, one the position of the similar L. exigua var. lutea, itself isolated, south of the main center of Leavent thia, the other the presence of the dissimilar L. uniflora extending to the north and west of this center. L. uniflora certainly migrated to the Ozarkian glades of Missouri, where it oc- curs in great abundance, and it seems probable that if L. “urea or its direct progenitor spread westward along the same route, some remnant would be there now. However, this is not the case, unless the plant has been completely overlooked. On the other hand, if the progenitor or pro- genitors of L. aurea migrated along a more southerly route, through Alabama, Mississippi and Louisiana, the only suit- able habitats known to me remaining today are the glades in labama where L. exigua var. lutea occurs. The develop- ment of the Mississippi River embayment region could easily have extinguished any habitats suitable for Leavenworthia present in that area at an earlier geological time. 5. Leavenwo-thia stylosa Gray. Bot. Gaz. 5:26. 1880 Plate 14 Glabrous winter annual; leaves of fully grown plants homomorphic, Mostly basal, lyrately pinnatifid, petiolate, 2-10 cm. long, lobes mostly dentate, sometimes more deeply so, terminal lobe markedly ew- to Several-flowered racemes ; including the pedicels; re) ; non-saceate, 4-8 mm, long, Spreading at right angles at full anthesis; petals obovate to broadly Spatulate, deeply emarginate, white, yellow or lavender on blade, LEAVENWORTHIA 85 yellow to orange on claw, usually with a light horizontal band between blade and claw, 9-15 mm. long, 5.9 mm. wide, blade spreading at right angles during full anthesis; stamens strongly tetradynamous, single anthers introrse, paired anthers extrorse; siliques oblong to linear, obtuse above and below, nerveless, fleshy, 1-3 cm. long, 2.5-4.5 mm. wide, 2-4 mm. thick when fresh; gynophores evident, ca. 1 mm. long; styles 3-7 mm. long; seeds dark brown, winged, broadly oblong to sub- orbicular, 3-4.5 mm. long, 2-3.5 mm. broad, wings thick, less than 0.25 wide, seed-coat shallowly reticulate, cleft with crest at one side of basal end; embryo nearly straight with radicle often slightly bent towards an accumbent position. n = 15. Type in the Gray Herbarium, cedar barrens at Lavergne, Tenn., June 2, 1879, A. Gattinger s.n. Isotypes GH, NY DISTRIBUTION: Central Basin of Tennessee ‘{ilep 2). BEDFORD CO,: 3 mi. north of Deason, D. G. Lloyd 350 (GH); wet field along country road leading west from Deason, Sharp and Sherman 23002 (Gu); 3 mi, south of Unionville, Rollins 5534 (GH, US). DAVIDSON CO.: 1.5 mi. east of Una, Rollins 6137 (GH); 2 mi. northeast of Lavergne, Rollins 55161 (GH, US); near Mt. View school, Svenson 10237 (GH); cedar barrens south of Nashville, Sharp and Shanks 381 (GA, GH, MICH, TENN); near Hurricane Creek, about 3 mi. northeast of Lavergne, men 59105 (GH). RUTHERFORD CO.: 4 mi. east of Lascassas, ae 2 (GH); near Murfreesboro, E. J. Palmer 35487 (GH); 1 mi. east a ‘cok Rollins 61167 (GH); south of Murfreesboro, ‘Shanks and Sharp 386 (GH); 13 mi. southeast of Murfreesboro, Rollins and Chan- mell 5652 (GH); 5 mi. southeast of Kirkland, Rollins 5523 (GH, US); 1.5 mi. northeast of Readyville, Rollins and De Selm 55123 (GH, US); 3 mi. northwest of Murfreesboro, Rollins, Solbrig, Hilferty and Lloyd 6030 (GH); 4.5 mi. northeast of Stewart Air Base, extreme northeast- ern Rutherford County, Rollins 6142 (GH); 2 mi. north of Walterhill, Rollins and Chambers 5739 (GH). SMITH CO.: west of Carthage, Sharp and Sherman 22740 (GH). WILSON CO.: 5.8 mi. west of Gladeville, Rollins and Chambers 5750 (GH); 2 mi. north of Lebanon, R. C. and D. Rollins 5206 (GH); Vesta, Svenson 7750 (GH); 10 mi. south of Hartsville, Rollins 5303 (GH, US); 3 mi. west of Lebanon, Rollins and Bold 55140 (GH, US); 3 mi. west of Norene, Rollins and Channell 5913 (GH); 10 mi. south of Lebanon, Rollins, Solbrig, Hilferty and Lloyd 6009 (GH). The type material of L. stylosa is yellow-flowered and the species was so described by Gray. It is true that exclusively vellow-flowered populations occur near Lavergne where Gat- tinger first collected the species. Extending north and east there is an extensive area in southeastern Davidson County and in adjacent Rutherford and Wilson Counties where hundreds of thousands of yellow-flowered plants make up the populations. Features correlated with the yellow-flow- ered plants of that area are a fleshy short fruit and a rela- tively long style (cf. Table 4). There are also exclusively yel- 86 REED C. ROLLINS Specimens Showing char acteristics of plants and some of the variation bn (5746) of L, stylo PLATE 14 ina populatic LEAVENWORTHIA 87 low-flowered populations in the vicinity of Lebanon, Wilson County, but these plants have elongated fruits much like those of white to lavender-flowered plants of the same gen- eral area. Over much of the eastern part of the range of L. stylosa, the populations are made up of predominantly white or lavender-flowered plants. The observation made by Baldwin (1945, p. 371) that “yellow is much the commonest color,” in L. stylosa does not hold. Actually, off-white to light lavender is the most frequent flower-color, considering the species as a whole. It is interesting that in many popu- lations there is a wide range of intensity of the lavender color from plant to plant. There is less of a range of inten- sity of the yellow color in yellow-flowered populations. Frequently seen are one or a few yellow-flowered plants among thousands of white-flowered individuals as well as the reverse situation with sporadic white- or lavender-flow- ered plants showing up among thousands of yellow-flowered individuals. My assumption from field observations alone is that these are mutant forms that arise here and there in the populations of which they are a part. The result of this easy mutational switch from one flower-color to another is extensively seen in the northeastern, eastern and southern parts of the range of L. stylosa but I have never seen evi- dence of it in the exclusively yellow-flowered populations of southeastern Davidson County and the area immediately adjacent. What is apparently a different kind of admixture of yel- low- and white-flowered plants occurs west of Gladeville in Wilson County. In this area, there is a gradual transition from predominantly white- to lavender-flowered populations in the area around Gladeville to predominantly yellow-flow- ered populations about six miles to the west. In this dis- tance, it appears that a gradual intermixing of the two major color forms is actively taking place. From field observations and from careful examination of greenhouse and garden materials, it is clearly evident that L. stylosa is indeed a highly variable species as Bald- win (1. c.) has pointed out. The question as to whether this variation is at random throughout the species or whether there are correlations of characteristics and/or local geo- graphic areas was not commented on by him. As we see it, 88 REED C. ROLLINS the variation present is not at random. There appear to be three areas within-the total range of the species which re- fiect separate trends within its overall variation pattern. It is suggested that these areas coincide with extensive glade systems, probably somewhat separated from each other for a considerable period of time, which allowed the evolution- ary development of recognizable morphological types. The glade system centering in southeastern Davidson County and bounded on the east by Stones River has ex- clusively yellow-flowered populations. Correlated features are relatively short, thick fruits with long styles. In the Murfreesboro area, particularly to the east, south and south- west, a white to light-lavender flowered form is correlated with relatively long, narrow and somewhat torulose fruits with shorter styles. Again, the assumption is that these features developed in a glade system long isolated from the Davidson County series. A third distinctive development in L. stylosa seems to have centered in Wilson County from the Cedars of Lebanon State Park northward. Flower color is not a distinguishing feature here because both white and yellow flowered types occur. However, the fruits are some- what intermediate between the Davidson County and the Murfreesboro types, being neither as long as the latter nor as short and fleshy as the former. 6. Leavenworthia torulesa Gray. Bot. Gaz. 5:26. 1880 Plate 4, fig. 4 & 5 Beh a — narrowly oblong, 3.5-5.5 mm. long; petals emat- der, si cat tpi mm. long, 3.5-6 mm. wide., white to light laven- ie ots eep lavender or yellow; siliques linear, strongly tor- when fresh; eviec oy 1.5-3 em, long, 2.5-4 mm. wide, 2-3.5 mm. thick seed nearly wingless : mm long; gynophore evident, ca. 1 mm. long; seed-coat feteniaireaite'a on — 2.5-3.5 mm. long, areclae of bryo slightly bent. n = tg niform in size throughout; radicle of em * LEAVENWORTHIA 89 Type in the Gray Herbarium, nae ee in the “barrens of Kentucky” by C. W. Short, 1840. Isotypes GH DISTRIBUTION: southern Remtacay tis southern Tennessee (Map Kentucky. “on wet rocks in the west of Kentucky,” C. W. Short en (GH); cedar glade situation, 12 mi. north of U.S. Highway 68 on state route 1083, Warren County, April’ 1, 1959, R. C. Rollins and R. B. Channell 5906 (GH, VDB). Tennessee. BEDFORD CO.: 1 mi. south of Unionville, Rollins 5533 (GH, US) ; 6 mi. southeast of Eagleville, Rollins and Chambers 5734 (GH); 1 mi. southeast of El Bethel, Rollins 59101 (GH). BRADLEY CO.: in small cedar barren east of Cleveland, Sharp, Felix and Adams 10566 (GH, UT); 1.5 mi. east of Cleveland, Rollins 5901; 5998; Rollins and Chambers 5701 (GH). DAVIDSON CO.: 1 mi. west of Bordeaux, R. C. and D. Rollins 5210 (GA, GH, US); wet lime- stone flats about Nashville, March 19, 1880, Gattinger s.n. (GH); 1.5 mi. east of Una, Rollins 6138 (GH); near Mill Creek, 14 mi. south of Nashville, Rollins and Channell 5620 (GH). GILES CO.: 3.5 mi. south- west of Pulaski, Rollins and Chambers 5723 (GH); 4 mi. south of Aspen Hill, Rollins and Channell 5627 (GH). LINCOLN CO.: southeast of Fayetteville, Sharp, Felix and Adams 10884 (GH); west of Harms, Sharp, Felix and Adams 10851 (GH). MARSHALL CO.: 4 mi. southwest of Cornersville, Rollins and Channell 5626 (GH); 5 mi. north of Lewis- burg, Rollins 5922 (GH); 2 mi. south of Chapel Hill, Rollins and Cham- bers 5728; 57142 (GH); 1 mi. north of Chapel Hill, Rollins and Cham- bers 5732 (GH). MAURY CO.: 5 mi. west of Columbia, Rollins 55109 GH); 6 mi. north of Culleoka, Sharp, Felix and Adams 11134 (GH); Rally Hill, Rollins 5920 (GH). RUTHERFORD CO.: 2 mi. south of Eagle- ville, Rollins 5531 (GH, Us); Stones River Military Park, Sharp, and Shanks 452 (GA, GH, MICH, US); 2 mi. south of Lavergne, Rollins and Chambers 5741 (GH); 4 mi. southwest of Murfreesboro, Rollins and Channell 5656 (GH); 1 mi. east of Eagleville, Rollins 61166 (GH). SUMNER Co.: 4 mi. east of Gallatin, cus M. Kriebel 9373 (PUL); Gallatin, Deam 61330 (MICH). WILLI co.: Kirkland, Rollins 53144; 55113 (GH); 1 mi. north cf Calléae cede: R. C. and D. Rollins 5215a (GH); 1 mi. north of Nolensville, Rollins 6150 (GH). WILSON co.: 4 mi. southwest of Lebanon, Rollins and Chambers 5752 (Gu); 3 mi. northwest of Gladeville, D. G. Lloyd 285 (GH); 2 mi. north of Green Hill, eyo 5311 (GH); 12 mi. west of Lebancn, Rollins and Channell 5908 (GH). Unlike Leavenworthia stylosa, all populations of L. toru- losa so far tested have been found to be self-compatible. In the greenhouse, without any manipulation of the flowers. self-pollination will only rarely take place. However, the plants are highly self-compatible and it takes only a weak pinch of the flower between the forefinger and thumb to bring about adequate pollination resulting in full seed set. Insect visitation to the flowers under natural conditions is frequent and it is likely that a complex mixture of out- crossing and selfing occurs in any given population. 90 REED C. ROLLINS The stance of the flower parts is very much affected by the brightness or dullness of the day. On bright days, the sepals are widely spreading and the limb of each petal is rigidly bent at right angles to the erect claw, providing a maximum target for a flying insect. On dull, overcast days, the sepals and petals remain nearly erect, spreading only slightly from a closed position. On such overcast days, which are usually quite cool during the flowering season of Leavenworthia, insect activity is at a minimum, if it is not wholly absent. Under these conditions, autogamy takes place following self-pollination. The position of the sepals and petals is such that the anthers are guided into the stigma as the filaments elongate. In the Species as a whole, it appears that both selfing and outcrossing are importantly involved in the breeding system. Flower color in L. torulosa ranges from white to deep lavender with the center of the flower being predominantly yellow. In some flowers the center tends in the direction of Orange. The most prevalent color of the outer petals is light lavender on either the upper or the lower surface with white on the opposite surface. In some populations the upper petal surfaces are predominantly white with the lower surfaces light lavender. In other populations the upper surfaces are light lavender with the lower surfaces white. Yellow as a color of the petal blade is very rare in L. torulosa. In fact out of over a hundred populations examined, only one had a few yellow-flowered plants in one spot. Quite in contrast to the flower-color situation in L. stylosa, where yellow, white, light lavender and deep lavender occur in many com- binations in a very complex way, yellow as a flower color Is essentially absent in L. torulosa. ; in still others, there is LEAVENWORTHIA 91 lations are most apt to be peripheral to the main range of the species but this point needs to be systematically checked. From the study of collected specimens and from field ob- servations it is evident that the peripheral populations (e. g., Warren Co., Kentucky; Bradley Co., Tennessee) are also the most divergent morphologically from the popula- tions near the distributional center of the species. The geographical range of L. torulosa is somewhat greater than that given by Baldwin (1945) but I agree with him that there is no solid evidence in support of its occurrence in Louisiana, Missouri or Arkansas, as single questionable specimens in herbaria might suggest. The center of devel- opment is undoubtedly in the Central Basin of Tennessee. In Kentucky, we found it in only one place and the popula- tion is very limited in size. I searched for the species in Kentucky in three different years before finding it and I have not seen any specimens from that state collected in the interim between those of Short in the 1840’s and the small population we found in 1959. At the southern ex- tremity of its range, the species is abundant in a cedar glade in Giles County, Tennessee, only a few miles north of the Alabama line. The most southeasterly population, in Brad- ley County, only a few miles north of the Georgia line, is relatively small by comparison with populations in the Cen- tral Basin of Tennessee and the plants resemble those of Kentucky in a number of features. Baldwin (1. ¢., p. 370) suggested that L. torulosa appar- ently intergrades with L. stylosa but I have seen no con- vincing evidence that it does. We have not been able to ob- tain hybrids between these two species, even though they have the same chromosome number, and I find that the seeds of L. torulosa are sharply distinct from those of L. stylosa, in addition to the other character differences that are usual- ly emphasized. The flowers of L. torulosa are always much smaller than those of L. stylosa (cf. Table 3). In fruit, the torulose siliques, shorter styles and wingless seeds satis- factorily characterize L. torulosa. There is a slight tendency towards torulosity of fruits in some populations of L. stylosa and this may have been the source of some confusion in the past 92 REED C. ROLLINS 7. Leavenworthia uniflora (Michx.) Britton, Mem. Torrey Club 5:171. 1894 Plate 15 Based on Cardamine uniflora Michx., Fl. Bor.-Am. 2:29. 1803. Type at P; photo of type at GH. Leavenworthia Michauxii Torrey, Ann. Lyc. Nat. Hist. N.Y. 4:89. 1837. Based on the same type as Cardamine uniflora Michx. Glabrous winter annual; early leaves with a single remote terminal dentate lobe, later leaves deeply pinnatifid with up to 9 pairs of sub- equal acutely dentate lobes, gradually increasing in size from petiole towards apex, terminal lobe only slightly larger than adjacent pair, leaves 2-10 em. long, 5-10 mm. wide, forming a rosette during first portion of plant growth; flowers mostly scapose, true stems with loose racemes of flowers formed only under favorable growing conditions and late in the particular plant growth cycle; sepals non-saccate, ob- long, greenish at first, purplish later or upon drying, 3.5-5 mm. long; petals white, narrowly lingulate, truncate, usually erect, 5-7 mm. long; 2.5-3.5 mm. wide; anthers of paired stamens introrse, small; siliques thick and fieshy, erect, 1.5-3 em. long, 3-5 mm. wide, 2.5-4 mm. thick when fresh; styles 1.5-3 mm. long; gynophore evident, ca. 1 mm. long; seeds dark Sowa, widely wing-margined, nearly orbicular, flattened, 3-4 mm. in diameter; areolae of seed-coat reticulum smaller over embryo than in wing area; radicle of embryo loosely adpressed along margin of cotyledons, Spproaching the accumbent position but scarcely accumbent; cotyledons nearly orbicular.,n = 15 DISTRIBUTION: : northern Alabama and arthwreatenn Georgia to south- western Ohio and Me to southern Missouri and northwestern Arkan- sas. Alabama. JACKSON CO.: 5 mi. West of Scottsboro, Rollins 55196 (GH); Larkinsville, Rollins and Channell 5649 (GH). LAWRENCE ©0.: 1.5 mi. west of Wren, Rollins 5982 (GH); 8 mi. east of Wren, Rollins and Chambers 5718, 57138 (GH); 3 mi. east of Speake, Lloyd 258 (GH); 5 mi. southwest of Danville, Rollins 6126 (GH). MADISON C0.: Smither’s Mt., Harper 3555 (GH, US) ; southwest slope of Monte Sano, near Huntsville, Rollins and Channell 5648 (GH). MORGAN CO.: near Danville, Rollins 55190 (GH, US); 4 mi. southwest of Lebanon, Rollins 6113 °(GH) ; 2 mi. south of McKendry, Rollins 6123 (GH); 5 mi. south- west of Massey, Lloyd 264 (GH) ; 1/2 mi. west of Falkville, Rollins and Chambers 5709 (GH) ; fore: 2 mi. south of Falkville, Lloyd 680 (GH). Tenn FORD CO.: 6 mi. south of Eagleville, Rollins and Chan- nell 5660 (GH); 1 mi. pon of El Bethel, Rollins 5999 (GH) ; southwest of Ray Chapel, Sharp, Felix and Adams 11160 (GH). KNOX Co.: Knox- ville, Ruth 358 (GH, MICH); Sequoia Hills, Knoxville, Drew et al 413 (GA). MARSHALL Co.: 1 mi. north of Chapel Hill, Rollins and Cham- bers 5733, Rollins 5317 (GH); 1.5 mi. south of Lewisburg, Rollins and Chambers 5726 (GH). MAURY CO.: 2 mi. north of Rally Hill, Rollins 5917 (GH); 21 mi. north of Lewisburg, Rollins and Channell 5622 ad east of Columbia, Sharp, Felix and Adams 11084 (GH). RUTHER- D CO.: 4 mi. east of Lascassas, Rollins 5903 (GH); 2 mi. south o le Rollins and Chambers 5740 (GH); 13 mi. southeast of Mur- LEAVEN WORTHIA 93 freesboro, Rollins and spre 5651 (GH); northeast of Lavergne, Sharp and Shanks 380 (GA, GH, MICH, TENN, US); 1.5 mi. ice of Eagleville, R. C. and D. Rollie 5220 (GA, GH). WILSON CO.: 3 mi. east of Lebanon, Rollins, Solbrig, Hilferty and Lloyd 6003 (GH); litt of Lebanon State Park, Sharp and Sherman 22741a (GH). Georgia. ALKER CO.: 5.5 mi. south of Chickamauga, Pyron’and McVaugh 2701 (GA). Ohio. ADAMS CO.: Mineral Springs region,’ May 2, 1932, E. L. Brown sm. (0S); Lynx, April 2, 1938, Floyd Chapman s.n. (0S). HIGH- LAND CO.: Brush Township, Bartley and Pontius 589 (NY, OS, US); PIKE CO.: Sunfish Creek near Byton, June 7, 1931, Bartley and Pontius s.n. (0S). Indiana. CLARK CO.: along Lick Creek, 1 mi. east of Charles- ton, Deam 24531 (GH); Charleston, 1877, C. R. Barnes s.n. (GH, PUL, US); same locality, 1878, J. M. Coulter s.n. (GH). Kentucky. Barrens of western Kentucky (probably near Bowling Green) C. W. Short s.n. (GH, NY, US). BARREN CO.: Glasgow Junction, “sinkhole” northwest of town, John Hussey s.n. (PUL). CALDWELL CO.: Princeton, A. M. Har- vill (Baldwin 2551, MICH). CHRISTIAN CO.: 2-3 mi. west of Hopkins- ville, Hardin 378 (GH, MICH, US). LOGAN CO.: 3 mi. southwest of Rus- peace Rollins and Channel 5905 (GH). SHELBY ‘Co.: Shelbyville, . B. Flint s.n. (GH). SIMPSON CO.: near Tenn. line, Baldwin 2554 pesiccltg Arkansas. BENTON Co.: near Eureka Springs, Canby 7 (GH) ; same locality, B. F. Bush 1531 (GH, US). CARROLL COo.: near Elk Ranch, Palmer 29828 (NY). Missouri. CHRISTIAN CO.: near Bull Creek, Rol- lins 61160 (GH). DALLAS CO.: barrens, Canby 8 (GH). FRANKLIN CO.: Gray Summit, M. Ownbey 1179 (GH). JEFFERSON CO.: 1887, H. Eggert s.n. (GH, US). LACLEDE CO.: near Hazel Greene, Steyermark 8500 (Us). MADISON CO.: mine La Motte, Palmer 39183 (GH). OZARK CO.: 10 mi. west of Theodosia, Rollins 61163 ne H). ST. LOUIS CO.: 1883, G. W. Letterman s.n. (GH, US). TANEY CO.: 6 mi. west of Forsyth, Rollins 61161 (GH) ; 7 mi. west of Rueter, Rollins 61162 (GH). In areas where Leavenworthia uniflora occurs with other species of the genus, it is earliest to flower. In some seasons it is as much as two or three weeks ahead of the others (cf. Table 6). This is undoubtedly of significance because L. uniflora approaches being fully autogamous and may be wholly so under some circumstances. Thus, as a species it May exist wholly independent of insect pollination and is ‘not at all limited as to earliness of flowering by the lack of insect’ ‘vectors. On the assumption that selection pressure has operated to favor earliness because of the preadaptation of all species of Leavenworthia to wet or at least moist situ- ations which are most likely in very early spring, L. uniflora has distinct advantages over the less fully autogamous spe- cies and particularly over those that are self-incompatible. The comparatively wide geographical distribution of L. uni- flora appears to be attributable to the development of auto- { 94 REED C. ROLLINS PLaTE 15. Specimens showing characteris stics of plants and some of the variation iflora. in a Tennessee population (5622) of L. uniflora LEAVENWORTHIA 95 gamy allowing distant single-seed dispersal along with the species concomitant emancipation from dependence on in- sect pollination and its increased seed production over that of other species. The flowers of L. uniflora are the smallest in the genus and the petals are always white on the blade, tending to- wards a light yellowish green on the claw. It is difficult to understand the early confusion of Torrey (1837) in particu- lar, and to some extent that of Gray (1880), as to flower color, because even upon drying the whiteness of the petals is retained. However, both were working from fragmentary material as shown by their preserved specimens in the New York Botanical Garden and the Gray Herbarium. In most of the populations I have observed, the petals tend to be nearly erect or at least ascending even when there is ample sunshine and the other nearby species of L thia have the limb of the petals expanded to a position at right angles to the claw. In this respect, L. aurea and L. uniflora are most alike. It is evident that there is variation of limb po- sition in L. uniflora and in different populations the extent of flower opening is different. Pictures of plants of a Mis- souri population show the petals widely extended. It should be re-emphasized that flower expansion in all species of Leavenworthia is very strongly influenced by light intensity. The flowers will close in less than an hour if the sky becomes cloudy following a sunny period. In the self-compatible species, such as L. uniflora, where the anthers are introrse, the closing of the flower insures self-pollination. The an- thers of L. uniflora are relatively small and under ordinary conditions are often in direct contact with the rather broad stigma. The siliques of L. uniflora are characteristically thick and the valves are smooth. The style is relatively short but the length differs significantly in different populations. Both the thickness, which is due to a fleshiness of the valve walls, and the style length are altered when specimens are dried. The styles are always longer in dried than in fresh material because of the contraction of tissue at and below the base of the style. Also, the siliques are less thick on prepared specimens than on live plants because the fleshiness is com- pletely lost. As with other species of the genus, the length 96 REED C. ROLLINS cf the siliaues may be strongly influenced by the size and vigor of the plant. Plants growing in especially favorable sites produce large numbers of relatively long fruits. The tendency towards scapose flowers is more strongly expressed in this species than in any of the others. Many populations are made up wholly of plants with scapose flow- ers. However, under very favorable conditions of growth, and late in the growth cycle of the plant, it will produce true branches with racemes of flowers. At present, we do not know what the factor or factors are that cause a shift from the one condition to the other. It is probable that the race- mose inflorescence is the more primitive since this type of inflorescence characterizes the Cruciferae as a whole. thia uniflora is obviously the most successful species of the genus from the point of view of area occupied (see Map 1). It has also become more successfully adapted to a wider range of moisture conditions, particularly to- wards the dry side, than any of the other species. This may help to account for its broader distribution. We have found L. uniflora in several glades that, although moist or even wet at times, are too well drained to permit standing water, as is characteristic of most of the glades where other species of Leavenworthia grow. L. uniflora readily invades old fields in areas where cultivated lands extend into, or displace the pre-agricultural glade sites. These fields are often very wet, poorly drained, with a high retained moisture content and stand idle long after better drained farmlands of the area are being actively worked. From the evolutionary viewpoint, the evidence points to L. uniflora as the most highly evolved of the 15-paired chromosome group of species. The self-compatible breeding system, tending towards autogamy, small flower size, in- trorse anther position, ascending petal stance, fixed petal color, lack of fragrance and tendency towards scapose in- florescence type, are all derived features when assessed in terms of those possessed by L. stylosa. As indicated above where breeding systems are discussed, self-incompatibility most frequently precedes self-compatibilitv and heterogamy precedes autogamy. L. stylosa is self-incompatible and therefore obligately heterogamous end insect pollinated. It has relatively large flowers, the anther position is extrorse, LEAVENWORTHIA 97 the petals are sharply spreading, the petal color ranges from white with yellow claw through yellow with orange claw to lavender with orange claw. The flowers are heavily fragrant and there is a strong tendency towards racemose inflores- cences on older plants. Thus, the more primitive conditions, mostly associated with insect pollination, are found in L. etylosa while in comparable characteristics, the derived con- dition is found in L. uniflora. The changes could have come about in association with selection pressures favoring a shift from the interdependence of obligate heterogamy and insect pollination towards autogamy and emancipation from the necessity of insect transfer of pollen Individual populations of L. uniflora are usually relatively uniform except for size and the features associated with size that may be strongly influenced by uneveness in the habitat or the progression of a particular season in relation to the habitat. Even the species as a whole is quite uniform but there are definite differences between populations. This is shown in Table 4, which gives differences in silique dimen- sions. I agree with Baldwin (l.c.) that L. uniflora is the most distinetive of the species of Leavenworthia. It is readily distinguishable from the other species because of the small entire petals and highly dissected leaves. The position of the short radicle, adpressed to the edges of the cotyledons of the embryo, is closer to what is usually found in the Cruci- ferae than is the straight or only slightly bent radicle found in other species of the genus. LITERATURE CITED BALDWIN, J. T. JR. 1945. Chromosomes of Cruciferae II. Cytogeog- raphy of Leavenworthia. Bull. Torr. Bot. Club 72: 367-378. BATEMAN, A. J. 1955. Self-incompatibility Systems in Angiosperms III. Cruciferae. Heredity 9: 53-68. ERICKSON, RALPH O., Louis G. BRENNER AND JOSEPH WRAIGHT. 1942. Dolomitic Glades of East-Central Missouri. Ann. Mo. Bot. Gard. 101. FREEMAN, C. P. 1933. Ecology of the Cedar Glade Vegetation near Nashville, Tennessee. Jour. Tenn. Acad. 8: 141-228. GRANT, VERNE. 1956. The Influence of Breeding Habit on the Out- come of Natural Hybridization in Plants. Am. Nat. 90: 319-322. Gray, A. 1880. The Genus Leavenworthia. Bot. Gaz. 5: 25-27. Harper, R. M. 1926. The Cedar Glades of Middle Tennessee. Ecology 7: 48-54 98 REED C. ROLLINS VON HaYEK, A. 1911. Entwurf eines Cruciferen-Systems auf phylo- genetischer Grundlage. Beih. Bot. Centralbl. 27: abt. 1, heft 2. 127-335. MCVAUGH, RoGERS. 1947. The Travels and Botanical Collecting of Dr. Melines Conkling Leavenworth. Field and Laboratory 15: 57- 70 MICHAUX, A. 1808. Flora Boreali-Americana 2: 29. QUARTERMAN, ELsIE. 1950. Ecology of Cedar Glades I. harps of Glade Flora in Tennessee. Bull. Torr. Bot. Club 77: 1-9. ROLLINS, REED C. 1950. The Guayule Rubber Plant and “i Relatives. Contr. Gray Herb. 172: 1-73. 1955. The Auriculate-leaved Species of Lesque- rella (Cruciferae). Rhodora 57: 241-264. 19 The Problem of Leavenworthia aurea. Rho- dora 58:73- 76. 1957. Interspecific wi i in Lesquerella (Criitiferas) Soil Gray Herb. 181: 1-40. SCHULZ, O. E. 1936. Cruciferae. erties 17b: 227-658. SvENSON, H. K. 1941. Notes on the Tennessee Flora. Jour. Tenn. Acad. Sci. 16: 133 and pl. V, fig. 1-4. TORREY, JOHN. 1837. An Account of Several New Genera and Species of North American Plants. Ann. Lyc. Nat. Hist. 4: 80-94. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 99 THE TAXONOMY AND DISTRIBUTION OF THE ZYGOPHYLLACEAE OF BAJA CALIFORNIA, MEXICO DUNCAN M. PORTER Although the Zygophyllaceae is a rather small family, it contains many important desert plants. The arid Mexican region of Baja California is an excellent area in which to con- sider the family, as contained within this region are prob- ably more taxa of Zygophyllaceae than in any other area of comparable size in the world. The purpose of this study, therefore, has been to ascertain the taxonomic and geograph- ic limits of the family within Baja California. This has been accomplished through herbarium research, a review of the pertinent literature, and field observations. Collections were made in Baja California from January through March, 1960, while I served in the capacity of Field Biologist for the Cali- fornia Academy of Sciences, San Francisco. This trip was _ supported financially by a grant from the Belvedere Scien- tific Fund of San Francisco. The assistance of both institu- tions is gratefully acknowledged. The first set of plants from these collections is deposited in the Dudley Herbarium of Stanford University (DS)." Duplicate sets are deposited in the Herbarium of the Cali- fornia Academy of Sciences (CAS) ; Herbario Nacional del Instituto de Biologia de la Universidad Nacional de México (MEXU) ; Conservatoire et Jardin Botaniques, Geneva (Gq); Herbarium of the University of California, Berkeley (UC) ; Herbarium of the University of Arizona, Tucson (ARIZ) ; Herbarium of the University of Texas, Austin (TEX) ; and the United States National Museum, Smithsonian Institu- tion (US). This investigation was accomplished under the helpful guidance of Dr. Ira L. Wiggins, former Director of the Natural History Museum, Stanford University. My grati- tude is extended to Dr. Wiggins, to the staff of the Dudley Herbarium, and to the herbarium curators, both those in whose herbaria I have had the pleasure of working, and those who have courteously loaned material to be examined, herbarium abbreviations given here and throughout the remainder of the paper refer to those listed by Lanjouw and Stafleu (1959). 100 DUNCAN M. PORTER for the assistance offered during the investigation. Special thanks are due Dr. Robert C. Foster of the Gray Herbarium, Harvard University, who generously supplied the Latin de- scription for the new species of Fagonia, and to Dr. Reid V. Moran of the San Diego Museum of Natural History, who provided cytological material of Fagonia. NOTES ON THE GEOGRAPHY, CLIMATE, AND VEGETATION OF BAJA CALIFORNIA : Baja California is a rugged mass of mountains and ar- royo-dissected mesas, with extensive flat, sandy plains in the northeast and along a large part of its Pacific Coast. The highest mountains are found in the north, where the Sierra Juarez and the Sierra San Pedro Martir, southern exten- sions of the Laguna Mountains of California, extend along the center of the peninsula from the International Boundary to a latitude of approximately 30° N. The tallest of these peaks, El Picacho del Diablo, reaches a height of just over 10,000 feet. The next highest range is encountered in the Sierra Victoria and the Sierra Laguna of the Cape Region at the southern end of the peninsula, which reach a height of nearly 7,000 feet. The area between the southern end of the Sierra San Pedro MArtir and the northern end of the Cape Region is composed of a series of low, rugged ranges for the most part paralleling, and in close proximity to, the eastern shore of the peninsula. Along the eastern side of these ranges there is a tremendous fault scarp that crosses over to the Pacific Coast just north of the Cape Region. This escarpment is strongly marked in the north and south, but it 1s not so pronounced in the center of the peninsula. The Sierra Judrez and Sierra San Pedro Martir rise abruptly out of the San Felipe Desert and slope gradually to the west- ward. North of the Cape Region, the Sierra Giganta arises abruptly out of the Gulf of California and slopes gently west- ward to the wide Magdalena Plain, which fronts on the Pacific Coast. To the north, a series of rugged mountains extending completely across the peninsula separates the Magdalena Plain from the Vizcaino Desert, the largest ex- panse of flat land on the peninsula, which also faces on the Pacific. Another series of low ranges separates the Vizcaino Desert from the southern end of the Sierra San Pedro Martir. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 101 The peninsula is visited in the north by winter rains that may fall from December to March and in the south by sum- mer rains occurring from July to September, with dry per- iods in between. These wet seasons may overlap to some extent in central Baja California, but they are less likely to occur here with as much regularity as in the extreme north or south. Both rainfall and temperature data for Baja Cali- fornia are meager and cover such short periods that gen- eralizations cannot be based upon them. It is known, how- ever, that temperatures of 90° F and above may be encoun- tered in any month at certain localities, and that other areas have gone as long as seven years without rain. There are three distinct phytogeographic areas in Baja California (Wiggins, 1960): the Californian, the Cape Re- _ gion, and the Central Desert (Fig. 1). The Californian flora occupies the mountains and foothills west of the San Felipe Desert to the Pacific Ocean. It extends from the Interna- tional Boundary to the southern end of the Sierra San Pedro Martir and is essentially a southward extension of the South- ern California flora. The Cape Region flora occupies the Sierra Giganta and most of the region south of La Paz. This flora is most closely allied with the subtropical vegetation to the southeast on the Mexican mainland. The Central Desert flora is found in the more arid region between the first two areas. It also occurs to the east of the Californian flora in the north and in two isolated spots in the Cape Region. This desert flora occupies the western part of the Sonoran Desert, which continues northward into California and eastward in- to Arizona and Sonora. The Central Desert flora has been further divided into four subregions (Shreve, 1951). In the northeast, the Lower Colorado Valley lies in the rain shadow of the Sierra Juarez and the Sierra San Pedro Martir, extending from the Inter- national Boundary to Bahia de los Angeles. The Central Gulf Coast continues southward from Bahia de los Angeles. It lies along the eastern side of the drainage divide of the peninsula, continuing with few interruptions to San José del Cabo. The many barren islands in the Gulf of California fall within this subdivision. South of the Californian flora and west of the peninsular divide is the Vizcaino Region. This region stretches along the Pacific Coast to Punta Pe- 102 DUNCAN M. PORTER T | me 0 ume + ges ae ce ome — h4* lige rr 30° A. Californian Region -2¢ B. Cape Region C. Central Desert Region 1. Lower Colorado Valley 2. Central Gulf Coast 3. Vizcaino Region | 2g 4. Magdalena Region PR 100 miles Fig. 1. Map showing the phytogeographie regions of Baja California. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 103 quena, where it is replaced by the Magdalena Region. The Magdalena Region continues west of the drainage divide and the Sierra Giganta to the Cape Region, where it in turn is replaced by the more mesic Cape Region flora. The plants dealt with in this study are essentially mem- bers of the Central Desert flora, although several are found within the Cape Region flora, and at least one is reported to occur in the Californian flora. There is a wealth of literature available pertaining to the vegetation and flora of Baja California. The following ref- erences deal with the more general aspects of the vegetation: T.S. Brandegee (1891), Goldman (1916), Johnson (1958), Johnston (1924), Shreve (1926, 1936, 1937, 1951), and Wiggins (1960). TAXONOMIC TREATMENT The Zygophyllaceae is a family of about 25 genera and 250 species, found mainly in the arid tropics and subtropics of both Old and New Worlds. The family is represented in Baja California by 6 genera, 15 species, and 4 varieties. The order in which the taxa involved are discussed is strictly a matter of convenience, following the order in which they fall in the artificial keys. Phylogenetic lines of development have not been investigated, as the present study is exploratory and preliminary to more searching cyto- taxonomic investigations. Generic descriptions, with the exception of Guaiacum and Larrea, apply only to those species represented in Baja Cali- fornia. KEY TO GENERA Upright divaricate woody shrubs Leaves alternate, entire or odd-pinnate; flowers white, drying to yellow, crepe-like 1. VISCAINOA. Leaves opposite, even-pinnate; flowers blue or yellow, blades of the petals twisted like a propeller Flowers blue; leaves of two free leafiets subsessile on the petiole; stipules caduccus; filaments not appendaged; capsule tomen- tose 2. GUAIACUM. Flowers bright yellow; leaves of two leaflets connate at the base; stipules persistent; filaments appendaged at the base: capsule pilose . LARREA Annual or perennial prestrate to upright herbs or suffrutescent sub- shrubs 104 DUNCAN M. PORTER Flowers red-purple; leaves digitately 3-7-foliolate; fruit a glutinous 5-lobed loculicidally dehiscent capsule 4, FAGONIA. Flowers yellow or orange; leaves odd-pinnate; fruit dividing into 5-12 indehiscent nutlets Nutlets spiney, 3-5-seeded, as many as the petals; beak of the fruit falling with the nutlets; branches prostrate to upright ............ . TRIBULUS. Nutlets tuberculate, 1-seeded, twice as many as the petals; beak f the fruit persisting after the nutlets fall; branches decum- bent 6. KALLSTROEMIA. 1. Viscainoa Greene, Pittonia 1:163. 1888 Divaricately-branched leafy shrubs 2-3(-6) m. high, branches crooked at the nodes; younger branches green, tomentose; bark on older branches gray, glabrate except at the nodes; flowering branches few-many-flowered, leaves crowded; herbage usually canescent (young- er parts especially so), tomentose to lanate. Stipules subulate to linear-lanceolate, 2-13 mm. long, caducous. Leaves alternate, entire or odd-pinnate with 3-5 leaflets. Flowers 2-3 em. in diameter; peduncles 1-several-flowered, opposite the leaves, to ca. 2 em. long; pedicels re- flexed, tomentose, 6-20 mm. long. Sepals 4-6, free, tomentose (pilose at base of inner surface), obovate to elliptic, apex obtuse to acute, deciducus. Petals 4-6, white, crepe-like, yellowing with age, obovate, short-clawed, strongly veined. Stamens 8-12, twice the number of the petals, inserted on the inconspicuous disc; filaments subulate to dilated at the base; anthers sub-basifixed, sagittate, 2-4 mm. long. Ovary 3-5- lobed, 3-5-loculed, pilose, on a short gynophore ca. 1 mm. long; ovules 2 per locule, placentation axile; style ca. 2 mm. long, persisting to form a beak on the fruit; stigma with as many lobes as locules in the ovary. Fruit an inflated, 3-5-lobed, 3-5-loculed, tomentose, slightly placentae, remaining attached below the beak to the upper part of the axis; beak 2-10 mm. long; fruiting pedicels reflexed, 5-22 mm. long. Seeds sticky, black, obovoid, 4-8 mm, long, 2 (rarely 1) per locule. Viscainoa is a genus containing a single species, with two varieties. It is found mainly on the peninsula and Gulf islands of Baja California, but also on Cedros Island, the Sonoran islands of San Esteban and Tibur6én, and on the Sonoran mainland near Guaymas at Bahia San Carlos and Miramar. Viscainoa is morphologically similar, and probably closely related, to Morkillia Rose & Painter (Chitonia Ses. & Moc. ex DC.), a genus known from Tamaulipas, Hidalgo, Puebla, and Oaxaca on the mainland of Mexico. Katherine Brande- gee, in fact, stated (1888, p. 230) that, “Its true place is certainly in Zygophyllaceae between Guaiacum and Chitonia, ZYGOPHYLLACEAE OF BAJA CALIFORNIA 105 and probably a slight modification of the generic character will admit it into the latter genus.” Viscainoa geniculata (Kell.) soos Sprig “8 163. 1888 Staphylea geniculata Kell. Proc. Calif. A Chitonia simplicifolia Wats. in Orcutt, West spiny en 2:58. 1886. nom. nud. KEY TO THE VARIETIES Leaves simple, usually entire ............ la. V. geniculata var. geniculata. Leaves odd-pinnate; leaflets 3-5, entire b. V. geniculata var. pinnata. la. V. geniculata var. geniculata Stipules tomentose, 2-10 mm. long; leaves petiolate or subsessile, simple (occasionally deeply lobed to pinnate with 2-3 leaflets), usually entire, coriaceous, tomentose to lanate, green to yellow-green, ovate or obovate to pirate occasionally inequilateral, 5-55 mm. wide, 1-8.5 cm. long, base cuneate, apex obtuse to mucronate or retuse (rarely obcord- ate); petiole short, to 6 (rarely to 10) mm. long; flowers to 3 cm, in diameter; sepals 4-5, 2-5 mm. wide, 6-10 mm. long; petals 4-6, to 11 mm. wide and 15 mm. long; stamens 4-6 mm. long; capsule to 2 cm. wide and 3.5 cm. long, beak 2-8 mm. long; fruiting pedicel 10-22 mm. ong. hoe salt from October to May and fruiting from December to ne. TYPE LOCALITY: Isla Cedros. “It was found growing in sandy ra- vines, near the sea-shore, on the east side of the Bay of San Sebastian Biscayno [Bahia Vizcaino], nearly opposite the guano island of Elide” (Kellogg, 1863, p. 22). This collection was made by J. A. Veatch in The holotype was deposited in the Herbarium of the California Academy of Sciences and presumably was destroyed during the earth- quake and fire of 1906. It seems desirable, therefore, to designate an isotype in the Gray Herbarium of Harvard University as lectotype. REPRESENTATIVE SPECIMENS: 24 mi. S. San Felipe, bap iis eae (CAS, DS); 1.5 mi. N Puertecitos, Porter 613 (CAS, DS); 34. Molino de Lacy, Porter 605 (CAS, DS); Bahia de San Luis ern Johnston 3338 (CAs); Calamajué, Porter 594 (CAS, DS); upper Arroyo de Calamajué, Wiggins 16029 (Cas, ad 29 mi. N Punta Prieta toward Calamajué, Porter 587 (CAS, DS) ; . E El Rosario, Wiggins 4332 (DS, GH, UC); a San Fer as Oreste 1344 (GH, uc); 6 mi. Wiggins 7580 sits DS, UC); between San Agustin & Catavina, Wig- gins 5320 (CAS, DS, GH, UC); Cataviia, Wiggins 4413 (CAS, DS, GH, UC) ; 24 mi. SE Laguna Seca Chapala, Ferris 9021 (DS); 14 mi. NE Punta Prieta toward Desengaho, Wiggins 7642 (DS, GH, UC); Agua Amarga, Wiggins 14821 (CAs, DS); 6.5 mi. NW Bahia de los Angeles, Porter 578 (CAS, DS) ; Bahia de los Angeles, Wiggins 7695 (AAH, DS, UC) ; 12.5 mi. N San Borja, Porter 562 (CAs, DS); San Borja, Moran 1988 (DS, UC) ; 2.5 mi. S Punta Prieta, Wiggins 15082 (CAS, DS); 15 mi. SW Punta 106 DUNCAN M. PORTER Prieta, Epling & Robinson, 10 Feb. 1935 (ps); 1 mi. from beach 35 mi. S Punta Prieta, Wiggins 11309 (Cas, DS, GH, UC) ; 4.5 mi. SE Rosarito, Porter 554 (CAS, DS); 8 mi. S Rosarito toward Miller’s Landing, Thomas 7980 (DS); Miller’s Landing, Porter 551 (CAS, DS); 15 mi. W Mezquital, Gentry 7893 (Ds, UC); 2 mi. S. Calmalli, Epling & Robin- son, 6 Feb. 1935 (AAH, Ds, UC); 10.2 mi. NW El Arco, Porter 208 (CAS, DS) ; Bahia de San Francisquito, Johnston 3582 (CAS); 17 mi. NW San Ignacio, Reed 6262 (Ds); 8 mi. S El Alamo, Wiggins & Ernst 604, 605 (CAS, DS); 35.5 mi. S San Ignacio toward La Purisima, Porter 525 (CAS, DS); 5 mi. N Santa Rosalia, Gentry 3683 (AAH, UC); coast SE Santa Rosalia, Moran 7478 (Ds, GH) ; Mulegé, Palmer 27 (ps, GH, UC) ; 49.6 km. SE Mulegé, Carter & Kellogg 2945 (DS, GH); 50.5 km. § El Médano, Carter, Alexander, & Kellogg 2465 (Ds); 19 mi. W Km. 38.5 on new highway W La Paz, Porter 434 (CAS, DS); La Paz, Johnston 3052 (CAS, GH), Jones 24074 (AAH, Uc) ; Bahia Pichilingue, Porter 390 (CAS, DS) ; 2 mi. N Bahia Pichilingue, Porter 419 (CAS, DS) ; San Pedro, Moran 7007 (Ds) ; 15.5 mi. 8 Todos Santos, Porter 366, 368 (CAS, DS); Isla Ventana, Bahia de los Angeles, Wiggins 14872 (cas, DS); Isla Angel de la Guarda, Moran 8640 (GH); Isla San Lorenzo Norte, John- ston 4194 (CAS); Isla Tortuga, Lindsay, 12 April 1947 (ps); Isla Ildefonso, Moran 9062 (GH); Isla San J osé, Moran 3756 (ps, UC); Isla Viscainoa geniculata var. geniculata is found on the sandy, usually granitic, soils of arroyo banks, outwash plains, and rocky hillsides from San Matias Pass in the northeast (Gold- man, 1916) and near El Rosario in the northwest, through- out Baja California and the islands in the Gulf of California, to 15.5 miles south of Todos Santos in the Cape Region. This variety also occurs on the Sonoran islands of Tiburon, San Esteban, and San Pedro Martir and on the Sonoran mainland near Guaymas. T. S. Brandegee (1891) reports Viscainoa geniculata from San José del Cabo, but herbarium specimens are unknown from this far south. This evergreen shrub is found throughout the area indi- cated above, with the exception of the Magdalena Plain, the Central Gulf Coast between Isla Ildefonso and Isla San José, and the Sierra Giganta. It is undoubtedly present in the latter two areas, but has not as yet been collected in these botanically little-known sections of Baja California. It is most common in the northern two-thirds of the Vizcaino Re- gion between San Ignacio and E] Marmol and in the south- ern part of the lower Colorado Valley between Bahia de los Angeles and San Felipe. It is in this latter area, on the ZYGOPHYLLACEAE OF BAJA CALIFORNIA 107 broad, sandy outwash plains between the Arroyo de Cala- majué and the Arroyo Miramar, where Viscainoa, together with Larrea tridentata (Ses. & Moc. ex DC.) Cov. and Fou- quieria splendens Engelm. become dominant members of the vegetation. It becomes a dominant nowhere else in Baja California, with the exception of a few small islands in the Gulf of California on which it may be the only shrub. Viscainoa geniculata var. geniculata is heavily browsed by cattle (and probably also by goats) when found near in- habited areas, as is the pinnate variety. Michael E. Soule, a graduate student in herpetology at Stanford University, has observed the chuckwalla Sawromalus hispidus climbing up into the shrubs and eating the leaves on several of the small islands near Bahia de los Angeles. 1b. V. geniculata var. pinnata Johnst., Univ. Calif. Publ. Bot. 7:439. 1922 Viscainoa pinnata Gentry, Madrono 5:161. 1940. Stipules tomentose to almost glabrate, 2-13 mm. long; leaves petio- late, odd-pinnate; petioles tomentose to glabrate, 2-17 mm. long; leaf- lets 38-5 (laterals rarely aborting to give 2 or 1), pétiolulate erg occasionally sessile on the rachis), entire, coriaceous, tomentose to glabrate, green to yellow-green, elliptic to ovate or obovate (occasion- ally inequilateral), abies leaflet the largest, 5-35 mm. wide, 15-65 mm. long, laterals 5-25 mm. wide, 1-5 cm. long, base cuneate (rarely oblique), apex mucronate to obtuse or retuse (rarely obcordate) ; flowers ca. 2 em. in diameter; sepals 5-6, 3-5 mm. wide, 5-8 mm. long; petals 4-5, to 8 mm. wide and 11 mm. long; stamens 7-8 mm. long; capsule to 1.5 em. wide and 3.5 cm. long, beak 5-10 mm. long; fruiting pedicel 5-12 mm. long Variety pinnata sis a gametic chromosome number of n = 13 ( Port- 08). Flowering and fruiting following the winter rains. TY: “San Raimonda Creek [Arroyo de San Raymundo], Baja California, April, 1889, Brandegee. Sheet No. 109442 in Herb. Univ. Calif.” a 1922, p. 439). SPECIMENS EXAMINED: El Carrizal, Porter 505, 508, 509 (CAS, Ds) ; 8 mi. N San Juanico, Gentry 4311 (DS, GH, UC); Cadajé, Porter 510 (cas, ps); Arroyo de San Raymundo, Brandegee, April, 1889 (uc!), Coristinee 3140 (uc), Porter 511 (CAS, DS), Thomas 8366 (DS, GH); Arroyo de San Juan, Porter 513 (CAS, DS); 4.5 mi. NW Arroyo de San Juan, Porter 517 (CAS, DS). This pinnate variety of Viscainoa geniculata is common in and along a series of sandy arroyos on the west coast of Baja California, near Lat. 26° 10’ N, Long. 112° 30’ W., between El Carrizal on the south and 4.5 miles northwest of 108 DUNCAN M. PORTER Arroyo de San Juan on the north, a distance along the road of about 24 miles. All of these arroyos appear to arise ata common point to the east in the Sierra de las Palmas. Pinnate leaves are found occasionally on Viscainoa geni- culata var. geniculata, e. g.: Moran 3756; Orcutt 1344; Por- ter 419, 434, 578, and 587; and Wiggins 15082. Leafy branches of the typical variety are also commonly found that have several leaves arising from the same axis on a stem; and V. geniculata var. pinnata will have an occasional entire leaf, e. g.: Constance 31 40; Gentry 4311; and Porter 908, 513, and 517. However, none of these collections is an example of the presence of one variety within the distribu- tional area of the other; the nearest V. geniculata var. geni- culata known to the territory of V. geniculata var. pinnata is 35.5 miles south of San Ignacio, still 35.5 miles north of the nearest collection of the pinnate variety. Further field work is needed to determine the nature and extent of the geographic and reproductive isolation of these two taxa. 2. Guaiacum L., Sp. Pl. 1:381. 1753 Trees or shrubs 1-10 m. high with hard resinous wood and stout branches with swollen nodes. Stipules minute, usually caducous. Leaves opposite, petiolate, even-pinnate, to 10 cm. long. Leaflets 1-6 pairs, subsessile on the rachis, coriaceous, glabrous to pubescent, linear to broadly obovate, apex rounded to mucronate, to 16 cm. long. Flowers 1-many, pedunculate from the axils of minute deciduous bracts, to ca. 3 em. in diameter. Sepals 4-5, unequal, free to somewhat united at the base, oblong to orbicular, deciduous. Petals 4-5, blue or purple (rarely white), obovate, apex rounded to mucronate. Stamens 8-10; filaments subulate or the lower hal Sagittate. Ovary 2-5-lobed, 2-5-loculed, obovate or clavate, glabrous to ovary. Fruit a smooth, coriac soid to obovoid, septicidal ¢ winged. Seeds 1 per locule. TYPE SPECIES: Guaiacum officinale L. cous, green to orange or reddish, ellip- apsule 1-2 em. broad, 1-2 em. long, angles A New World genus consisting of some six species found from Sonora, Mexico, to the Guianas and throughout the West Indies to southern Florida. A single endemic species is known from Baja California. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 109 1. Guaiacum unijugum T. S. Brandg., Univ. Calif. Publ. Bot. 6:183. 1915 A divaricate woody shrub growing to ca. 2 m. high; bark brownish- gray; youngest branches pubescent, older glabrate; stipules caducous; leaves yellow-green, crowded on short lateral branchlets 3-5 mm. long; leaflets 1 (rarely 2) pair, tomentose to almost glabrate, ovoid, mucron- ate, base oblique, 3-10 mm, wide, 6-13 mm. long; rachis 1-8 mm. long, tipped by a small caducous tomentose spine ca. 1 mm. long; petioles and rachises densely pubescent; flowers peduneulate from the axils of minute bracts on the lateral branchlets, 1-4 per branchlet, to 1.5 em. in diameter; pedicels 6-13 mm. long, pubescent; sepals 5, green, free, puberulent, oblong to obovate, rounded at the apex, 2-4 mm. wide, 3-6 mm. long; petals 5, clear blue drying to yellow, to 6 mm. wide and 12 mm. long; stamens 10, 7-8 mm. long; filaments subulate; anthers sub- basifixed, curved, sagittate, ca. 2 mm. long; ovary obovate, 5-lobed, 5- loculed, tomentose, on a short gynophore ca. 1 mm. long; ovules ca. 8 per locule, in 2 rows on the placenta, placentation axile; style slender, subulate, less than 1 mm. long; stigma entire; fruit a yellow-brown mucronate tomentose capsule, to ca. 2 cm. oni; are pedicels 7-16 mm. long; rarely more than 1 seed maturing per fru Flowering in August and fruiting in August and opal TYPE LOCALITY: “Collected by T. S. Brandegee near San José del Cabo, Baja California. Type, Herb. Univ. Calif. No. 109558” (T. S. Brandegee, (1915, p. 183). SPECIMENS EXAMINED: Bahia de los Frailes, Chambers 865 (DS, UC) ; San José del Cabo, Brandegee, 5 Sept. 1890 (uc!), Grabendorfer, 1899 (uc). This rare and little-known endemic from the southeastern Cape Region differs from Guaiacum coulteri Gray var. coul- teri, found from Sonora to Oaxaca, and from G. coulteri var. palmeri (Vail) Johnst., known from the vicinity of Carbé, Sonora, to San Blas in northwestern Sinaloa, in its smaller flowers and single pair of ovoid tomentose leaflets. The flowers of G. coulteri are up to twice the size of those of G. unijugum, and its leaves are 2-6 cm. long with 6-10 linear-oblong to elliptic-oblong glabrous leaflets that are 3-10 mm. wide and 10-25 mm. long. G. unijugum is probably closely related to G. coulteri, and further collections and morphological and cytological investigation may prove it to be a subspecies or variety of the latter. On the basis of present information, however, it is best to retain Brande- gee’s name, rather than to make a new combination. Chambers collected Guaiacum unijugum on “Sand dunes, Los Frailes Bay,” and it is probably restricted to the coastal sand dunes and sandy washes in this area and to those near 110 DUNCAN M. PORTER San José del Cabo, about 30 miles to the south. This distri- bution falls into an arid and botanically little-known south- ern extension of the Central Gulf Coast subdivision of the Sonoran Desert, which extends from the vicinity of Punta Arena del Sur to San José del Cabo, a distance of roughly 40 miles. G. unijugum is thus separated from the Mexican mainland and its nearest congeners by the mouth of the Gulf of California, a distance of about 200 kilometers. 3. Larrea Cav., An. Hist. Nat. Madrid 2 7119. 1800 short, 1-2 em. long. Stipules fleshy, reddish, triangular, pubescent, obtuse to short-acuminate, 1-4 mm. long, persistent; glands on the inner surface secreting a resin making the plants highly glutinous at brown, pubescent to glabrate. Flowers solitary, to ca, 2.5 cm. in diameter; pedicels pubescent, 3-15 mm. long. Sepals 5, free, unequal, ovoid, acute to obtuse, pubescent, persistent to deciduous. Petals 5, bright yellow, ovoid, acute or rounded to variously incised at the apex. irregularly laciniate membranaceous scale nearly as long as the fila- ment; anthers oblong, sub-basifixed, sagittate, 1-3 mm. long. Ovary deeply 5-lobed, 5-loculed, globose, on a short gynophore ca. 1 mm. long densely pubescent, 2-5 mm. long; style subulate to cylindric, 2-6 mm. long; stigma simple to slightly 5-lobed. Fruit a 5-lobed, 5-loculed, densely pubescent, globose, septicidal capsule, ca. 7 mm, long. Carpels 1-seeded, dehiscent or indehiscent at maturity. TYPE SPECIES: Larrea nitida Cav. A New World genus of five species, four in the arid re- gions of Argentina (two of these also ranging into Peru, Bolivia, or Chile), and one in the warm deserts of North America. i 1. Larrea tridentata (Ses. & Moc. ex DC.) Cov., Contr. U. S. Nat. Herb. 4:75. 1893 Zygophyllum tridentatum Ses. & Moe. ex DC., Prodr. 1:706. 1824. Larrea mexicana Moric., Pl. Nouv. Amér. 71. 1839, Guaiacum mexicanum (Moric.) Baill., Bot. Med. 2:886. 1844. Zygophyllum californicum Torr. & Frém. in Frém., Second Rep. 257. 1845. nom. nud. Larrea glutinosa Engelm. in Wisliz., Mem. Tour No. Mex. 93. 1848. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 111 Ae resinosum Berl. in Berl. & Chovel, Diario de Viaje 52. 1850. nom. Covillea seeanodiea (Cav.) Vail, Bull. Torr. Bot. Club 22:229. 1895. part Covillea tridentata (Ses. & Moc. ex DC.) Vail, Bull. Torr. Bot. Club 26:302. 1899. Covillea oan Sapam in Wisliz.) Rydb. in Vail & Rydb., N. Amer, F]. 25:1 Larrea plano var. glutinosa Jeps., Manual 604. 1925. Schroeterella glutinosa (Engelm. in Wisliz.) Briq., Ver. Geobot. Inst. Ziirich 3: 664. , Schroeterella tridentata (Ses. & Moc. ex DC.) Briq., loc. Neoschroetera glutinosa (Engelm. in Wisliz.) Briq,., pbc ti 2:514, 1926. Neoschroetera tridentata (Ses. & Moc. ex DC.) Briq., loc. cit. Ultimate branchlets appressed-pubescent with short white hairs; young branches with reddish bark, bark gray or blackish on older branches; black-banded and crooked at the nodes, banding caused by the resinous excresence of the persistent stipules; stipules obovoid, acute to short-acuminate, spreading or at least not clasping the stem, 1-4 mm. long; leaves subsessile to short-petiolate, petiole to ca. 2 mm long; leaflets 2, entire, divaricate, connate to each other and adnate to the rachis for 2-4 mm, at the base, obliquely lanceolate to falcate, inequilateral, mucronate, coriaceous, glutinous, ss appressed- pubescent to glabrate, 1-8.5 mm wide, 4-18 mm. long; a pubescent, green, deciduous mucro to 2 mm. long on the end of on rachis between the leaflets; flowers to ca. 2.5 em. in diameter, pedicels 3-12 mm. long; sepals appressed-pubescent, 3-4.5 mm. wide, 5-8 mm. long; petals oblong to oblanceolate, the short claw brownish, 2.5-5.5 mm. wide, 7-11 mm. long, twisted at the claw and appearing propeller-like; stamens 5-9 mm. long; filaments 4-8 mm. long, the scale % to as long as the filament, to 3 mm. wide; anthers 2-3 mm. long; hairs on the ovary 2-4 mm. long, carpels attached only on inner angle; ovules ca. 8 per locule, placentation axile; style cylindric, 4-6 mm. long, persisting on the young fruit; stigma minutely and obscurely lobed; fruit pilose-wooly, hairs silvery turning reddish-brown with age; fruiting pedicel 4-13 mm. long; seeds brown, boat-shaped, 4 mm. long. A somatic chromosome number of 2n = 52 has been reported for Larrea tridentata (Covas, 1949). Numbers have also been reported for the South American members of the genus, with the exception of L. ameghinoi Speg.: L. cuneifolia Cav., 2n = 52 (Covas & Schnack, 1946); L. divaricata Cav., 2n = 26 (Covas & Schnack, 1946); and L. nitida Cav., 2n = 26 ( Cea 1949; Rahn, 1960). Somatic numbers of 2n = 52 for L. divaricata, 2n = 52 for L. nitida, and 2n = 104 for L. tridentata listed by Darlington and Wylie (1955) are actually tetra- ploid counts (Porter, 1961). a In Baja California flowering from September to May and fruiting from September to June. TYPE ALITY: “In regno Mexicano” (DeCandolle, 1824, p. 706). The holotype is presumably in the Herbier DeCandolle, Geneva. the DUNCAN M. PORTER REPRESENTATIVE SPECIMENS: La Rumorosa, Cota, March, 1932 (UC); 12 mi. W Mexicali, Porter 618 (CAS, DS); Seven Wells, Mearns & Sehoonfetit 2872 (ps); Laguna Gardner, Mearns & Schoenfeldt 2910 (ps) ; San Matias Valley, Robertson 4 (UC); between La Ventana Pass & San Felipe, Wiggins 13011 (DS, GH) ; between El] Cajon & Algadones, Wiggins 9844 (DS); 13 mi. S San Felipe, Porter 616 (CAS, DS) ; 13.5 mi. N Puertecitos, Porter 614 (CAS, DS); 15 mi. S Puertecitos, Porter 611 (CAS, DS); 3 mi. NE Las Arrastras, Porter 602 (CaS, DS) ; 1.5 mi. NW Calamajué, Porter 596 (CAS, DS); 15 mi. NE El Rosario toward El Sauzalito, Porter 156 (cas, DS); 32 mi. E El Rosario toward San Agustin, Ferris 8551 (ps); Missi6n San Fernando, Harvey, 18 April 1954 (uc); El Marmol, Wiggins 4368 et DS, GH, UC); Catavina, Wiggins 4412 (AAH, CAS, DS, GH, UC); 15.5 mi. S Rancho Laguna Chapala, Porter 179 (CAS, DS); Sabha Porter 584 (CAS, DS); 6.5 mi. NW Bahia de los Angeles, Porter 577 (CAS, DS); Bahia de los Angeles, Wiggins 7693 (AAH, DS, UC); 12.5 mi. N Sam Borja, Porter 561 (CAS, DS); 3.5 mi. W San Ignacito toward Rosarito, Porter 200 (cas, DS); 30 mi. N Mezquital, Hammerly 66 (CAs, DS); 17.6 km. 5 Mezquital, Carter & Kellogg 2979 (pS, GH); 5 mi. W Calmalli, Thomas 7974 (DS, GH); 5 mi. W El Barril, Wiggins 7832 (AAH, DS, uc); 12 mi. SW El Arco toward La Banderita, Thomas 8273 (DS); 2.5 mi. NW La Cantina, — 537 (CAS, DS); 5 mi. W Los Martires, Porter 530 (CAs, DS); 40 mi. NW San Ignacio toward El Arco, Porter 209 (CAS, DS); 23.5 mi. g en Ignacio toward Cuarenta, Porter 527 (cas, DS); 6 mi. N Cuarenta, Thomas 8357 (ps, GH); 13.5 mi. NW Arroyo de San Juan, Porter 519 (cas, DS); El Carrizal, Porter 507 Santa Rosalia, Porter 217 (Cas, ps); 35 mi. S Mulegé, Porter 230 (CAS, DS) ; Comondé, Brandegee, Feb. 1889 (UC); 15.5 mi. SW San José Comondt, Porter 233 (cas, Ds); Loreto, Jones 27085 (AAH, DS, UC); 30 mi. NW El Crucero, Porter 485 (CAs, DS); 29 mi. S El Crucero, Porter 450 (cas, DS); 9.5 m. NE Estero Salinas toward Llanos de Hiray, Porter 468 (CAs, DS); 8.5 mi. NW Santa Rita toward Llanos de Hiray, Porter 472 (cas, Ds) ; 81 mi. NW La Paz, Porter 447 (CAS, DS); 42.5 mi. W La Paz, Porter 440 (CAs, DS); 19 mi. W Km. 38.5 on main highway W La Paz, Porter 433 (cas, DS); 3.5 mi. W Km. 38.5, Porter 436 (CAS, DS); 11 mi. W main highway toward Arroyo Seco, Porter 427 (CAS, DS); 15 mi. W La Paz, Hammerly 201 (cas, DS, GH); Los Aripes, Povior: 422 (CAS, DS); 10.5 mi. SE La Paz airport toward Todos Santos, Porter 453 (CAs, DS) ; Isla San Luis, Johnston 3323 (CAS, GH) Isla Angel de la Guarda, Johnston 3403 (cas) ; Isla San Marcos, Ferris 8662 (DS); Isla Carmen, Rose 16649 (GH); Isla San Francisco, Moran 3741 (DS, UC). T.S. Brandegee (1891) reported Larrea tridentata from Todos Santos, but herbarium specimens are known from no further south than 10.5 miles southeast of La Paz airport, about 43 miles north of Todos Santos. L. tridentata is rare in the region south of La Paz, but it is quite possible that the ZYGOPHYLLACEAE OF BAJA CALIFORNIA 113 plant ranges along the arid western coast of the Cape Region nearly to Todos Santos. The plant has also been reported from San Luis Gonzaga Bay, San Francisquito Bay, and Guadalupe Point (Johnston, 1924). The total distribution of this species is discussed by Shreve (1940), Rzedowski and Medellin Leal (1958), and Garcia, Soto, and Miranda (1960). There has been much controversy as to whether the North American Larrea tridentata is specifically distinct from the South American L. divaricata Cav. Vail (1895) was the first to argue that the two are conspecific, but she soon reversed herself (1899) upon examining a larger series of South American collections. Since Vail’s paper in 1895, there have been a number of authors arguing for and against the dis- tinctness of L. tridentata. Johnston (1924, 1940), Monticelli (1939), Axelrod (1950), and Morello (1958) have consid- ered the two as identical, while Robinson (in Gray, 1897), Garcia, Soto, and Miranda (1960), and Ragonese (1960) have regarded them as distinct. A careful comparison of North and South American ma- terial reveals that the two are readily separable on the basis of leaflet and stipule characteristics, as has been pointed out by Robinson (in Gray, 1897) and Vail (1899).2 Not only does the herbage of Larrea divaricata tend to be more copiously pubescent than that of L. tridentata, but the edges and veins of the leaflets of L. divaricata are lined with many white hairs 1-2 mm. long that are clearly discernible with the unaided eye. Leaflet veins in L. tridentata may at times be conspicuous, but they are usually dark, not lined with hairs. Occasional specimens of L. tridentata do, however, have some leaflets with the white-hairy veins and hairiness on the edges of the leaflets characteristic of L. divaricata, so this is not as infallible a criterion in separating the two spe- cies as are the following. There is a distinct difference be- tween the two in the shape of the leaflets, those of L. trident- ata being obliquely lanceolate to falcate, while those of L. divaricata are obovate to ellipsoid, the tips rarely curving to- ward one another. The stipules of L. tridentata are obovoid, acute to short-acuminate at the apex, 1-4 mm. long, and dis- *Differences in the internal petiolar and foliar anatomy are discussed by Ragonese (1960) 114 DUNCAN M. PORTER tinctly free to spreading from the stem or petioles (as they are also in the South American L. cuneifolia and L. nitida). The stipules of L. divaricata, on the other hand, are broadly ovate, obtuse or rounded at the apex, 1-2 mm. long, and dis- tinetly clasping the stem. he two taxa overlap in many of their morphological vari- ations. The differences discussed above, the geographic dis- tributions of the taxa, and the differences in chromosome number, however, provide sufficient evidence for considering them to be separate species. Larrea tridentata is found throughout the desert regions of Baja California, with the exception of the southeastern tip of the Cape Region. The species occurs on alluvial and voleanic soils, sandy plains, and the edges of alkali flats, being rarely found on granitic soils. It is found as a domi- nant and in many associations throughout arid Baja Cali- fornia wherever edaphic factors and the lack of competition are favorable. L. tridentata never occurs, however, in suc extensive stands as those in the Lower Colorado Valley. A possible exception is that in the valley surrounding Laguna Seca Chapala. 4. Fagonia L., Sp. Pl. 1:386. 1753 Prostrate-spreading to upright diffusely branched suffrutescent annual or perennial herbs or subshrubs, to 8 dm. high and 1.5 m. in diameter; branches angled at the occasionally minutely spinulose nodes, striate, green, yellowing with age; bark on oldest branches gray; herbage glandular, pubescent, or glabrous, occasionally tinged with purple. Stipules subulate to acerose, reflexed to spreading or ascend- ing, spinescent with a white tip, to 2 cm. long; occasionally making the plant quite prickly to the touch. Leaves opposite, compound, digi- tately 3-7-foliolate; leaflets entire, coriaceous, petiolulate, linear to broadly ovate, spinescent with a white tip; apical leaflet the largest, to 2 cm. wide and 3 em. long; lateral leaflets usually inequilateral, often caducous. Flowers solitary, axillary, few to many, 1-1.5 cm. in diameter; pedicels thin, to 1.5 em. long. Sepals 5, free, lanceolate to oblong, green to purple, spinescent with a white tip, deciduous. Petals 5, pink to dark red-purple (rarely white), spatulate, mucronate. Sta- mens 10; filaments filiform; anthers sub-basifixed, sagittate, 0.5-1 mm. long. Ovary 5-lobed, 5-loculed, ovoid or obovoid, stipitate glandular and pubescent to pubescent or glabrous, sessile or essentially so, 1-2 mm. in diameter; ovules 2 per locule, placentation basal; style subu- late, to 4 mm. long, persisting to form a beak on the fruit; stigma minutely and obscurely lobed. Fruit a moderately inflated, deeply 5- lobed, 5-loculed, cbovoid, loculicidal capsule, glandular and pubescent ZYGOPHYLLACEAE OF BAJA CALIFORNIA 115 to pubescent or glabrous, coriaceous, reticulate, to 7 mm. wide and 5 mm. high, green, yellowing with age; beak thickened at the base, commonly twisted clockwise, to 4.5 mm. long; fruiting pedicel abruptly reflexed; carpels separating at the inner angle, but remaining attached below the beak to the upper part of the axis, dehiscing ventrally, the shiny glutinous endocarp separating from the coriaceous exocarp. Seeds 1 or 2 per locule, dark, flattened, ovate, shiny, minutely reticu- late, to 2.5 mm. wide and 4mm long. TYPE SPECIES: Fagonia cretica L. Fagonia is a genus known from the warm arid regions of all continents except Australia. There has been some controversy concerning the status of Fagonia in North America. Bentham (1844), Standley (1911), and Johnston (1924, 1924a) have shown that the Fagonia of the New World are separable from the Old World members of the genus, but Johnston, differing from Bentham and Standley, assigned both the North and South American taxa (with the exception of F’. densa) to the South American Fagonia chilensis H. & A. An examination of the available South American material, however, shows the North Ameri- can taxa to be specifically distinct from the South American species of the genus. Standley (1911) emphasized glandularity and pubescence in his treatment of the genus, while Johnston (1924) empha- sized stipular development. A combination of these char- acters is used in the following treatment. Absolute sizes, however, are not reliable, as there is quite a difference in vegetative growth from year to year, depending upon pre- cipitation ; those plants growing with little water apparently also tend to be less open and more bushy in habit. Amount of pubescence or glandularity probably is dependent partial- ly upon precipitation, but the type of pubescence and stipule type are constant. The difference in leaflet size on the same plant is also striking. The young leaflets near the ends of the branches may be linear and scarcely 2 or 3 millimeters long, while the older basal leaflets may be 2 centimeters long and a centimeter wide, or larger. The flowering and fruiting dates given below have been compiled from herbarium collections. There is a good pos- sibility, however, that Fagonia produces flowers. whenever there has been sufficient rain. 116 DUNCAN M. PORTER KEY TO THE SPECIES Leaflets 5-7-foliolate 1. F. palmeri. Leaflets 3-foliolate Stipules acerose, ascending, as long as or longer than the petioles .... 2. F. densa Stipules subulate to linear-subulate, reflexed to spreading, not as ong as the petioles Stipules subulate, rarely spreading, usually 1-3 (never over 6) mm. lo plants upright to prostrate; flowers to ca. 1 em. in diameter Branches with small stipitate glands to glabrous Leaflets glandular; branches rarely scabrous; fruits and Po ee varies puberulent and glandular .......... alifornica. Leaflets glabrous; branches soon eet Fruits ae ovaries puberulent, but never glandular ...............006 4, F. laevis. Branches villous, never glandular 5, F. villosa. Stipules linear-subulate, rarely reflexed, to 8 mm. long or longer; plants prostrate-spreading; flowers to ca. 1.5 cm. in diameter Branches with large (1 mm. in diameter) Pa aee t to subsessile glands, glabrous toward the base ........:csssessereeere achyacantha. Branches with small (42 mm. in diameter sid a stipitate glands and many straight glandless hairs, more glands toward apices, more hairs toward bases .... 7. F. barclayana. 1. Fagonia palmeri Vasey & Rose, Contr. U. 8. Nat. Herb. 1:82. 1890 Fagonia chilensis var. palmeri (Vasey & Rose) Johnst., Proc. Calif. Acad., ser. 4, 12:1052. 1924. Flat-topped to rounded, prostrate to upright, to 0.5 m. high and 1 m. in diameter; herbage densely covered with small stipitate glands, older branches becoming glabrate; stipules acerose, ascending, 2-9 mm. long; leaflets 5-7 (occasionally 1 or more aborting), subfleshy, flattened, linear to linear-lanceolate, to 1 mm. wide, 2-14 mm. long; petioles 4-12 mm. long; flowers to ea. 1.5 ¢ n diameter; pedicels 2-7 mm. long, glandular-stipitate; sepals Sisto lakebiade agian hag 1.5-2.5 mm. wide, mm. long; petals rose, 2-4 mm. wide, 5-8 mm long; filavients 4 mm. long; ovary pubescent, style 2-3.5 mm. long; ’ fruiting pedicels 3-8 mm. long; fruit puberulent and glandular, 4- 6 mm. in diameter, 4-5 mm. high; beak 3-4 mm. long; seeds 1 per locule. ma species has a gametic chromosome number of n = 10 (Moran Known to flower and fruit from March to May. TYPE LOCALITY: “Santa Rosalia. Lower California.’ McVaugh (1956, p. 314) states that, “According to Palmer’s field notes, Nos. 197-210, inclusive, were collected in a range of Ravwarcnnsie (‘contiguous to Santa Rosalia’) made up entirely of pure gypsum.” SPECIMENS —— Santa Rosalia, Ferris 8703 (ps, NY), Palmer 209 (Us!; isotypes CAS, GH, NY, UC);Isla San Marcos. Ferris 863° (ps, NY), Johnston pea (CAS, DS, GH, NY, UC), Moran 3964 (Ds, UC). ZYGOPHYLLACEAE OF BAJA CALIFORNIA 117 This species is locally abundant in washes and on gypsum hills and cliffs on Isla San Marcos and on the peninsula near Santa Rosalia. It has also been collected on Isla Tiburon, Sonora [Moran 8795 (GH) ]. 2. Fagonia densa Johnst., Proc. Calif. Acad., ser. 4, 12:1052. 1924 Dense, compact, upright, globose subshrub, to 80 em, high and 1 m. in diameter; herbage densely covered with small short-stipitate and subsessile glands, becoming glabrate with age; younger herbage highly glutinous; stipules acerose to linear-lanceolate, ascending, 3-21 mm long; the numerous crowded leaves and stipules appearing together as whorled acerose leaves; leaflets 3 (rarely 4) acerose to linear-lanceo- late, to 1 mm, wide, 1-11 mm. long, deciduous; petioles 3-18 mm. long; flowers to ca. 1.5 mm. in diatater: pedicels 2-3 mm. long; sepals oblong-lanceolate to oblong, green to yellow, 1.5 mm. wide, 4-5 mm. long; petals pink to purple, 2-4 mm. wide, 6-9 mm. long; filaments 4-5 mm, long; ovary pubescent and with small subsessile glands, style 3-4 mm. long; fruiting pedicels 2-7 mm. long; fruit puberulent and gland- ular, ca. 6 mm. in diameter, 4-5 mm. high; beak 3-4.5 mm. long; seeds This species has a gametic chromosome number of n = 10 (Moran 8590, 8904, and 10368). Known to flower and fruit from March to Jun TYPE: “No. 1285, Herb. Calif. Acad. Sci., pas Bee May 9, 1921, by I. M. Johnston (no. 3532) from gypsum soil in a cafion on South San Lorenzo Island, Gulf of California” (Johnston, 1924, p. 1052). SPECIMENS EXAMINED: San Francisquito, Brandegee, 13 May 1889 (uc); Bahia de los Angeles, Moran 10368 (GH), Porter 572 (CAS, DS), Wiggins 14888 (CAS, DS); Isla Angel de la Guarda, Moran 7234, 8184 (DS), 8590 (GH); Isla San Lorenzo Sur, Johnston 3532 (cAs!, isotypes GH, NY, UC), Moran 8904 (GH). Fagonia densa is found in the southern part of the Lower Colorado Valley and on several islands in the Gulf of Cali- fornia opposite this botanically poorly known area. Recent botanical exploration has revealed it to be much more com- mon than previous records indicated. 3. Fagonia californica Benth., Bot. mre 10. 1844 Fagonia californica var. hindsiana Benth., loc. ci Fagonia rosei Standl., Proc. Biol. Soc. Wash. os 247, Fagonia cretica var. oe (Benth.) Engl. in ect, & Drude, Veg. der Erde 9(3) :731. 1915. Fagonia chilensis var. rosei (Standl.) Johnst., Proc. Calif. Acad., ser. 4, 12:1051. 1924. Fagonia californica ssp. rosei (Standl.) Wiggins, Contr. Dudley Herb, 4:19. 1950. 118 DUNCAN M. PORTER Prostrate-spreading; herbage densely covered with small stipitate glands and occasional glandless hairs, becoming glabrate and older branches becoming scabrous; stipules subulate, reflexed to spreading, 1-5 mm. long; leaflets 3, elliptic to oblong-lanceolate or oblong, glandu- lar, soon becoming glabrate, apical leaflet to 12 mm. wide and 18 mm. long, lateral leaflets to 9 mm. wide and 15mm. long, laterals commonly caducous; petioles 1-17 mm. long; flowers to ca. 1 em. in diameter; pedicels 1-6 mm. long; sepals oblong-lanceolate, glandular, becoming glabrate, green to purple, 1-1.5 mm. wide, 2-3.5 mm. long; petals pink to dark red-purple, 2-4 mm. wide, 5-7 mm, long; filaments ca. 4 mm. long; ovary pubescent and glandular, style 2 mm. long; fruiting pedi- cels 2-10 mm. long; fruit puberulent and glandular, 3-7 mm. wide, 3-5 mm. high, beak 1.5-2.5 mm. long; seeds 1 per locule. ¢ ahi from September to April and fruiting from November to pril. TYPE LOCALITY: “Bay of Magdalena, Lower California” (Bentham, 1844, p. 10). SPECIMENS EXAMINED: Arroyo de San Juan, Porter 514 (ps); La Purisima, Wiggins 15010A (DS); Bahia Magdalena, Brandegee s. n. (DS), Hinds, 1841 (K!); Isla Magdalena, Brandegee, 12 Jan. 1889 (GH, UC), Lung 24 (uc), Orcutt 7 (GH, NY); Isla Margarita, Johansen 611 (CAS, DS); 1.5 mi. SE Santa Rita, Porter 480 (cas, DS); 92 mi. NW La Paz, Porter 479 (cas, DS); 89 mi. NW La Paz, Porter 448 (CAS, DS) ; 76.5 mi. NW La Paz, Porter 444 (CAS, DS). Fagonia californica occurs in sandy and rocky soils from Arroyo de San Juan south across the Magdalena Plain to its southern boundary. Standley (1911), in his monograph of Fagonia in North and South America, considered typical Fagonia californica to extend from southern Baja California to southwestern Utah. An examination of a few of the speci- mens cited therein as F. californica (Abrams 3500; Hall 2794; Jones, 19 April 1906; Jones 3641; and Palmer 818) shows them to be Standley’s own Fagonia laevis. Standley considered F. laevis to be completely glabrous, and F. cali- fornica to have small stipitate glands. However, a thorough study of North American Fagonia shows that the former varies from glabrous to slightly glandular, with small stipi- tate glands usually on pedicels and sepals, occasionally on stipules and petioles, and rarely on the ultimate branches. The branches soon become scabrous, and the fruits are never glandular. F. californica, on the other hand, is densely glandular and has small stipitate glands on the fruits; oc- casionally the entire plant is covered with glands. Specimens of F. rosei from Isla Tiburén, Sonora, (Johns- ton 3528 (CAS, GH) and Rose 16779A (us!)) are small- ZYGOPHYLLACEAE OF BAJA CALIFORNIA 119 leaved, scabrous examples of F. californica. This is the only place outside of Baja California that F. californica is known to occur. 4. Fagonia laevis Stand]., Proc. Biol. Soc. Wash. 24:249. 1911 Fagonia chilensis var. aspera (Gay) Johnst., Proc, Calif. Acad., ser, 4, 12:1051. 1924. In part. Fagonia chilensis var. laevis (Stand].) Johnst., loc. cit. Fagonia i aiand ssp. laevis (Standl.) Wiseinis: Contr. Dudley Herb. 4:19. Upright ‘to pian ios to 2 ft. high and 3 ft. in diameter; essentially glabrous, but pedicels and sepals and occasionally stipules and petioles and rarely ultimate branches with small stipitate glands, some speci- mens entirely glabrous; older branches scabrous (younger parts oc- casionally scabrous as well); older branches becoming stoloniferous, bearing many erect smaller branches; stipules subulate, reflexed tc spreading, 1-6 mm. long; leaflets 3, linear-elliptic, apical leaflet to 5 mm. wide and 18 mm. long, lateral leaflets to 3 mm, wide and 15 mm. long, one or both laterals commonly caducous; petioles 2-15 mm. long; owers to ca. 1 cm, in diameter; pedicels 1.5-11 mm. long; sepals elliptic to lanceolate, green to purple, ca. 1 mm. wide, 2. 3 mm. long; petals pink to dark red-purple, 1.5-3 mm. wide, 4-7 mm. long; filaments 3-4.5 mm. long; ovary glabrous to pubescent, style 1-2 mm. long; fruit- ing pedicels glabrous to glandular, 1.5-11 mm. long; fruit glabrous to puberulent, 3-6 mm. wide, 4-5 mm. high; beak 1.5-2 mm. long; seeds 1 per locule. Flowering and fruiting from October to July. TYPE LOCALITy: “Type in the U.S. National Herbarium, No. 855, 582, collected near Yuma, Arizona, April 25, 1906, by Marcus E. Jones” (Standley, 1911, p. 249). REPRESENTATIVE SPECIMENS: Arroyo La Agua Amarga, Wiggins 9928 (ps); 35 mi. S Puertecitos, Porter 606A (CAS, Ds); Bahia San Luis Gonzaga, Johnston 3346 (CAS); 2 mi. N Las Arrastras, Wiggins 15929 (CAS, DS); 8 mi. SE El Rosario, Wiggins 5255 (CAS, DS, GH, NY, UC); San Fernando, Cronemiller 3065 (ps); Arroyo de Catavifnacito, Porter 169, 176 (CAs, DS); 34 mi. N Laguna Chapala toward San Felipe, Wiggins & Ernst 678 (cas, DS); 6 mi. S Laguna Chapala, Wiggins 15054 (CAS, DS); 29 mi. N Punta Prieta, Wiggins 15059 (CAs, DS); 19.5 mi. N Punta Prieta, Porter 585 (CAS, DS); 8 mi. SE Desengano, Porter 580 (CAs, DS) ; 9 mi. NW Bahia de los Angeles, Porter 579 (CAS, DS); Bahia de los Angeles, Palmer 546 (NY); 6.5 mi, N San Borja, Porter 560 (cas, DS); 12 mi. NE Rosarito toward San Borja, Porter 557 (CAS, DS); main road 4.7 mi. E Miller’s Landing, Porter 552 (CAS, Ds); 5 mi. N Mezquital Grande, Haines & Stewart, 9 Feb. 1935 (Ds, GH, NY, UC); 2 mi. E Mezquital, Shreve 6960 (GH) ; 30 mi, inland from Lagoon Head, Palmer 818 (CAS, GH, UC); Calmalli, Epling & Robinson, 5 Feb. 1985 (Ds, NY, UC); 4 km. NW El Arco, Carter, Alexander, & Kellogg 1916 (ps, Uc); Santa Gertrudis, Purpus 112 (DS); 4-5.5 mi. SW El Arco toward La Banderita, Thomas 8289 (DS); 5.5 mi. S El 120 DUNCAN M. PORTER Arco, Wiggins & Ernst 643 (CAS, DS); 15 mi. S El Arco, Wiggins 11348 (DS, GH, UC); Bahia San Bartolomé, Rose 16235 (NY): Las Tinajas, Gentry 7612 (Ds, UC); 1.5 mi. SW Los Martires, Thomas 8319 (DS); 17.5 mi. S San Ignacio, Porter 528 (cas, DS); 47.5 mi. S San Ignacio, Porter 522 (cas, DS); 59 mi. S San Ignacio, Porter 520 ( CAS, DS); Arroyo de San Juan, Porter 515 (CAS, DS); 3 mi. SW’ Rio de la Purisima toward road te Comondt, Porter 500 (CAS, DS); 2 mi. E San Ignacio, Porter 212 (CAs, Ds); 13.5 mi. NW Santa Rosalia, Porter 213 (CAS, DS); 4.5 mi. NW. Santa Rosalia, Porter 216 (cas, Ds); Santa Rosalia, Palmer 196 (GH); 4.8 mi, W Santa Rosalia toward Santa Agueda, Wiggins 7946 (DS); 7 mi. S Santa Rosalia, Porter 220, 222 (CAS, DS); 11 mi. S Mulegé, Porter 227 (CAS, DS); Isla San Marcos, Ferris 8640 (Ds); Isla Coronados, Moran 3904 (DS); Isla Cedros, Haines & Hale, 17 Feb. 1939 (CAS, DS, UC), Stewart 41 (cas); Isla Natividad, Stewart 78-79 (cas). The following specimens were probably collected somewhere in north- eastern Baja California within the confines of the Lower Colorado Valley, the data on the labels notwithstanding: Tijuana, Oreutt & Or- cutt, 30 June 1884 (GH, Ny, UC) ; Valle de las Palmas, Jones 3691 (CAS, DS, NY, UC); Ensenada, Jones, 1882 (GH). These localities are well within the more mesic Californian Region, and other collections of Fagonia are unknown from this area, Fagonia laevis, the most common Fagonia in Baja Cali- fornia, is found from the International Boundary south to an approximate latitude of 26° N , Where it extends slightly into the range of F. californica. It also occurs in the Mojave Desert and is found northward into the Great Basin Desert. It is known from southwestern Arizona south to the vicinity of Guaymas, Sonora. This species is commonly encountered in abundance in rocky and hilly surroundings, being more rarely found on sandy plains. It occurs in both granitic and volcanic soils, Fagonia longipes Standl., from the Lower Colorado Valley subdivision of the Sonoran Desert in southwestern Arizona and southeastern California, has long been considered a Synonym of F. laevis. However, F. longipes is a distinct taxon, separable from F. laevis by its long pedicels, glandu- lar-stipitate branches, pedicels, petioles, and leaflets, and its glandular fruits. Fagonia longipes differs from F. cali- fornica in its slender growth, longer pedicels and petioles, and linear leaflets. This species is not known from Baja California. 5. Fagonia villosa D. M. Porter, sp. nov. Prostrate-spreading; herbage densely to sparingly villous, hairs with thickened bases and ap Ree pressed toward the branch apices, older parts ZYGOPHYLLACEAE OF BAJA CALIFORNIA 121 becoming glabrate and occasionally scabrous;~stipules reflexed to spreading, subulate, 1-6 mm. long; leaflets 3, ovate or obovate to elliptic (rarely linear), apical leaflet 1-8.5 mm. wide and 2-23 mm. it a leaflets to 4 mm. wide and 16 mm. long; petioles 1-10 (rarely to 15) mm. rade flowers to ca. 1 cm. in diameter; pedicels 1-7 ‘ehnesy to 9) mm. long; sepals linear-oblong, villous, green to purple, 1-1.5 mm. wide, 2-4 mm. long; petals pink to red-purple (rarely white), 1-3 mm. wide, 4-6 mm. long; filaments 3-4 mm, long; ovary pubescent, style 1-2 mm. long; fruiting pedicels 1-6 (rarely to 10) mm. long; fruit puberulent, 3.5-6 mm. wide, 3.5-4.5 mm. high; beak 1-2 mm, long; seeds 1-2 per locule. Planta patuli-prostrata, dense vel sparse villosa, pili ad basem incrassati, apices ramulorum versus appressi, partes vetustiores gla- bratae vel aliquando scabrae; stipulae reflexae vel patulae, subulatae, 1-6 mm. longae; foliola 3, ovata, obovata, vel elliptica (raro linearia), foliolum apicale 1-8.5 mm. latum et 2-23 mm. longum, foliola lateralia ad 4 mm. lata et 16 mm. longa; petioli 1-10 mm. longi (raro 15 mm.) ; flores ad 1 cm. diam.; pedicelli 1-7 mm. longi (raro 9 mm.); sepala lineari-oblonga, villosa, viridia vel purpurea, 1-1.5 mm. lata, 2-4 mm. longa; petala rosea vel rubropurpurea (raro alba), 1-3 mm. lata, 4-6 mm, longa; filamenta 3-4 mm. longa; ovarium pubescens, stylus 1-2 mm. longus; pedicelli fructiferi 1-6 mm. longi (raro 10 mm.) ; fructus puberulus, 3.5-6 mm. latus, 3.5-4.5 mm, altus; rostrum 1-2 mm. long- um; semina 1-2 in loculo. TYPE: Porter 426, collected 8 March 1960 at ‘Roadside on gently sloping hill on old highway W of La Paz, He miles W of junction with new highway. Common. Flowers rose-purple, few in flower. Some prostrate-spreading; others rather Gaiieht: bushy.” The holo- type is deposited in the Dudley Herbarium of Stanford University (no. 445,587). Isotypes are deposited in the Herbarium of the Cali- fornia Academy of Sciences; Herbario Nacional del Instituto de Biologia de la Universidad Nacional de México; Conservatoire et Jar- din Botaniques, Geneva; Herbarium of the University of California, Berkeley; Herbarium of the University of Arizona, Tucson; Herbari- f the University of Texas, Austin; and the United States National Museum, Smithsonian Institution. SPECIMENS EXAMINED: 40 mi. E El Rosario, Wiggins 4476 (DS, GH) ; Lagoon Head, Palmer 827 (CAS, GH, NY); 12 mi. S El Solito, Wiggins 15139 (CAs, DS); 12 mi. NW Guadalupe, Porter 535 (CAS, DS) ; 0.5 mi. NW Guadalupe, Porter 531, 532 (CAS, DS); 7 mi. S Los Angeles, Wig- gins & Ernst 628 (CAs, pS); El Carrizal, Porter 506 (CAS, DS); 8 mi. N San Juanico, Gentry 4307 (DS, GH, UC); 1 mi. N San Juanico, Porter 504 (CAs, DS); Arroyo de Mezquital, Porter 502 (CAS, DS); 18 mi. NW road to Comondt toward La Purisima, Porter 494 (CAS, DS); 0.75 mi. NW road to Comondt toward La Purisima, Porter 491 (CAS, DS); 25.5 mi. SW San José Comondt, Porter 234 (CAS, DS) ; 51 mi. N El Crucero, Porter 487 (CAs, ps); 8 mi. N Santo Domingo, Thomas 8386 (DS); 30 mi. NW El Crucero, Porter 486 (CAS, DS); 24 mi. NW El Crucero, Porter 484 (cas, ps); S El Refugio, Wiggins 5525 (Ds, UC); 9.5 mi. S El Refugio, Porter 460 (cas, DS); 11 mi. NW Santa Rita toward 122 DUNCAN M. PORTER Llanos de Hiray, Porter 470 (CAs, DS); 8.5 mi. NW Santa Rita toward Llanos de Hiray, Porter 471 (CAs, DS); 4 mi. inland from Puerto Chale, Chambers 813 (DS, UC); 58.5 mi. NW La Paz, Porter 475 (CAS, DS); 52.5 mi. NW La Paz, Porter 443 (CAS, Ds) ; 46.5 mi. NW La Paz, Porter 441 (CAS, DS); 42.5 mi. NW La Paz, Porter 439 (cas, DS); 19 mi. W km. 38.5 on main highway W La Paz, Porter 435 (CAS, DS); 15.5 mi. W km. 38.5, Porter 431 (cas, D8); 2 mi. N Arroyo Seco, Porter 430 (CAS, DS); 22.5 mi. W new highway toward Arroyo Seco, Porter 429 (CAS, DS); 14.5 mi. W new highway toward Arroyo Seco, Porter 428 (CAS, DS) ; Todos Santos, Jones 24092 (cas, DS, NY, UC); 3 mi. S Todos Santos, Porter 371, 371A (cas, ps); San José del Cabo, Brandegee 81 (Ds, UC). This plant has long gone under the name Fagonia cali- fornica var. barclayana Benth. but a careful examination of the holotype of that taxon reveals that is has short straight hairs, stipitate glands, and glandular fruits, not the appres- sed villous hairs and glandless fruits of F. villosa. F. cali- fornica var. barclayana is, therefore, now applied to what has been hitherto known as Fagonia insularis Standl. 1 have chosen to name this species F agonia villosa because of the prominent pubescence, which differs from that of any other North American Fagonia. This is the most common species of Fagonia in the south- ern half of Baja California. It is known to occur along the western side of the peninsula from near El Rosario to San José del Cabo. It is most common from the southern end of the Magdalena Plain and northward across the Vizcaino Desert; it has been collected sparingly north and south of this range. Fagonia villosa is a plant of flat sandy areas or of sandy arroyo Margins, but it is occasionally collected in rocky soils. Flowers of Fagonia villosa are usually pink to red-purple, but white-flowered individuals are rarely found (Porter 931 and Wiggins 5525). These may or may not occur within populations containing the normal flower color. 6. Fagonia pachyacantha Rydb. in Vail & Rydb., N. Amer. Fl. 25:105. 1910 Fagonia californica var. glutinosa Pringle ex Vail, Bull. Torr. Bot. Club. 22 2229. 1895. Not F. glutinosa Del., 1813. Fagonia viscosa Rydb. in Vail & Rydb., N. Amer, Fl. 25:104. 1910. Not F’.. viscosa Hochst. ex Boiss., 1867. Fagonia chilensis var, glutinosa (Pringle ex Vail) Johnst. Proc. Calif. Acad., ser. 4, 12:1051. 1924, ZYGOPHYLLACEAE OF BAJA CALIFORNIA 123 Fagonia chilensis var. pachyacantha (Rydb. in Vail & Rydb.) Johnst., loc. cit. Fagonia californica ssp. pachyacantha (Rydb. in Vail & Rydb.) Wiggins, Contr. Dudley Herb. 4:19. 1950. Prostrate-spreading, to 4 ft. in diameter; ultimate branches thickly beset with large short-stipitate to subsessile globular (drying to cup- shaped) golden glands, older parts glabrate; ultimate branches usually appearing yellowish from a distance because of the many glands; stipules stout, linear-subulate, spreading to slightly reflexed, glandu- lar to glabrate, 3-16 mm. long; leaflets 3, ovate to elliptic, slightly obovate, or linear and terete, glandular, becoming glabrate, apical leaflet to 10 mm. wide and 1-26 mm. long, lateral leaflets to 7 mm. wide and 1-20 mm. long, one or both laterals commonly caducous; petioles glandular to glabrate, 2-16 mm. long; flowers to ca. 1.5 em. in di- ameter; pedicels glandular, 1-7 mm. long; sepals oval, glandular to glabrate, green to purple, 1-1.5 mm. wide, 2-2.5 mm. long; petals light to dark red-purple, 2.5-5.5 mm. wide, 5-8 mm. long; filaments 3.5-5 mm. long; ovary pubescent and glandular, style 2-2.5 mm. long; fruit- ing pedicels glandular, 1-6 mm. long; fruit puberulent and glandular, 4-5 mm. wide, 3.5-4 mm. high; beak 1.5-3.5 mm. long; seeds 1 per locule. Flowering and fruiting from February to May. TYPE LOCALITY: “Type collected in Baja California (date and exact locality not given), Leon Diguet (herb, N. Y. Bot. Gard.).” (Vail & Rydberg, 1910, p. 105). SPECIMENS EXAMINED: Baja California, Diguet s. n. (NY!; type fragment at uc); 24 mi. W Mexicali, Porter 619 (CAs, DS); 5 mi. La Ventana, Wiggins 15765 (CAS, DS); 9 mi. N San Felipe, Porter 617 (CAS, DS); 2.7 mi. N San Felipe, Raven 14768 (ps); 18 mi. S San Felipe, Wiggins 15817 (CAS, DS); 13.5 mi. N Puertecitos, Porter 615 (CAS, DS); 15 mi. S Puertecitos, Porter 612 (CAS, ps); 72 mi. S San Felipe, Wiggins & Ernst 683 (CAS, DS); 26.5 mi. S Puertecitos, Porter 607 (CAS, DS); 35 mi. S Puertecitos, Porter 606 (CAS, ps); 15 mi. N Bahia de San Luis Gonzaga, Wiggins 16040 (CAS, DS); 4 mi. N Bahia de San Luis Gonzaga, Wiggins 16035 (CAS. DS) ; 19.5 mi. N Las Arras- tras, Porter 604 (CAS, DS); 11.5 mi. NW Calamajué, Porter 600 (CAs, DS); 3.5 mi. NW Calamajué, Porter 597 (CAS, DS); 7.5 mi. N turnoff to Calamajué 24 mi. N Punta Prieta, Porter 588 (CAS, DS); 4.1 mi. NW Bahia de los Angeles, Porter 181 (CAS, DS), 181A (Ds); 3.5 mi. NW Bahia de los Angeles, Porter 575 (CAS, DS) ; Bahia de los Angeles, Palmer 546 (GH, NY); 2.5 mi. S village at Bahia de los Angeles, Porter 566 (CAs, DS); SE part Bahia de los Angeles, Porter 571 (CAS, DS) ; Bahia de San Francisquito, Johnston 3555 (CAS, GH)y Santa Rosalia, Palmer 180 (GH, NY); 1 mi. S Santa Rosalia, Porter 218 (CAS, DS) ; 7 mi. S Santa Rosalia, Porter 219, 221 (CAS, Ds); Isla Angel de la Guarda, Johnston 3385 (CAS, GH). Fagonia pachyacantha is easily recognized by its pros- trate-spreading, open habit, and its large golden short-stipi- tate to subsessile glands. This species occurs in sandy washes, on rocky hillsides, and on flat plains. It is found 124 DUNCAN M. PORTER in both granitic and volcanic soils along the Gulf Coast of Baja California from the International Boundary to a few miles south of Santa Rosalia. It occurs northward into southeastern California, southwestern Arizona, and north- western Sonora, all within the Lower Colorado Valley sub- division of the Sonoran Desert. Fagonia pachyacantha is found almost entirely within the range of F’, laevis, but there appears to be partial ecological isolation between them. F. pachyacantha tends to be a plant of flat, sandy areas, while F. laevis is more apt to be found in rocky, hilly surroundings. 7. Fagonia barclayana (Benth.) Rydb. in Vail & Rydb., N. Amer. Fl. 25:104. 1910 Fagonia californica var. barclayana Benth., Bot. Sulph. 10. 1844. Fagonia insularis Standl., Proc. Biol. Soc. Wash. 24:247. 1911. Fagonia chilensis var. barelayana (Benth.) Johnst. Proc. Calif. Acad., ser. 4. 12:1051. 1924, Fagonia chilensis var. insularis (Standl.) Johnst., loc. cit. Fagonia californica ssp. insularis (Standl.) Wiggins, Contr. Dudley Herb. 4:19. 1950. Prostrate-spreading, to 4 ft. in diameter; herbage on younger parts densely covered with small stipitate glands and with occasional small straight hairs, older parts densely pubescent with small straight hairs (becoming less glandular and more pubescent with age), oldest parts becoming glabrous and occasionally scabrous; stipules stout, linear- subulate, spreading, 3-9 mm. long; leaflets 3, linear to broadly ovate, glandular and pubescent, apical leaflet to 19 mm. wide and 28 mm. long, lateral leaflets to 10 mm. wide and 20 mm. long; petioles 3-15 mm, long; flowers to ca. 1.5 cm. in diameter; pedicels glandular and ; Sepals ovate, glandular, green to purple, 1.5-2 mm. wide, 2.5-4 mm. long; petals rose to purple, 2.5-5.5 mm. wide, 5-9 mm. long; filaments 4-6 mm. long; ovary pubescent and glandular, style 2.5-4 mm. long; fruiting pedicels glandular and pubescent, 3-12 mm. long; fruits densely glandular and sparingly puberulent, 4-6 mm. wide, 3-5 mm. high; beak 2-4 mm. long; seeds 1 per locule. Flowering and fruiting from November to J une, LOCALITY: “Magdalena Bay” (Bentham, 1844, p. 1 SPECIMENS EXAMINED: Bahia Concepcién, Shreve 7093 (Ds, GH); Punta Guadalupe, Johnston 4155 (CAS, GH, NY, UC); 19 mi. S Mulegé, Porter 229 (Cas, DS); 20 mi, S Mulegé, Wiggins 11411A (ps); 54.5 mi. S Mulegé, Porter 231 (CAS, DS); 6 mi. W Canipolé, Wiggins 11447 (DS, GH, UC) ; 2 km. NE San Isidro, Moran 7463 (DS) ; La Purisima, Gentry 4216 (DS, GH, UC); 10 mi. W Comonda, Gentry 4091 (ps) ; 15.5 mi. SW San José Comonda, Porter 2392 (CAS, DS); Bahia Magdalena, Barclay 8s. nm. (K!); 1 mi. S Missién Los Dolores landing, Wiggins, Carter, & Ernst 238 (CAS, Ds, UC); 13 mi. N El Pilar, Wiggins 15462 (CAS, DS); ZYGOPHYLLACEAE OF BAJA CALIFORNIA 125 Bahia de los Muertos, Wiggins 14444A (Ds); Isla Coronados, Johnston 3767 (CAS); Isla Carmen, Johnston 3806 (cas), Palmer 830 (us), Wosnessensky, 2 Feb. 1842 (GH); Isla Danzante, Moran 9239 (GH); Isla Monserrate, Moran 3892 (Ds, UC), Rose 16604 (NY); Isla San José, Moran 3786 (DS, UC); Isla San Francisco, Johnston 3958 (CAS), Moran 3709 (ps), Wiggins, Carter, & Ernst 395 (CAS, DS, UC). This taxon previously has been known under the name Fagonia insularis Standl.; however, an examination of the holotypes of North American Fagonia reveals that F. in- sularis is in reality a synonym of F. barclayana. The plants hitherto known under the name F. californica var. barclay- ana are now referred to F. villosa, q. v. This prostrate species is found in rocky canyons and on rocky hillsides from the vicinity of Bahia de la Concepcién south to Bahia de los Muertos, being found mainly on the eastern side of the peninsula. Fagonia barclayana occurs on the islands in the Gulf of California from Isla Carmen south to Isla San Francisquito and is also known from near Guaymas, Sonora. The collection of F. barclayana from Bahia de los Muertos approaches F’.. californica in its overall appearance, but it has the typical pubescence of Fagonia barclayana. 5. Tribulus L., Sp. Pl. 1:386. 1753 Prostrate to suberect herbs, annual or from a perennial rootstock; diffusely branching, branches striate, green, becoming yellow, to several meters long; herbage more or less sericeous or appressed- villous and somewhat hispid, youngest parts and nodes more densely so, oldest parts becoming glabrate. Stipules subulate to lanceolate, pinnate, one alternate leaf smaller than the other or aborting, dark green; leaflets 3-7 pairs, petiolulate, oblong to elliptic or slightly ovate, mucronate, inequilateral, base oblique, lower surface densely sericeous, upper surface (especially along midvein) less so to glabrate; rachis densely appressed-villous and hispid, a slender green mucro 1-2 mm long at the apex between the ultimate leaflets. Flowers solitary, pedi- cels appressed villous and hispid, in the axils of the alternately smaller leaves. Sepals 5, ovate to lanceolate, pubescent, edges membranaceous, deciduous. Petals 5, yellow (rarely white), obovate, apex rounded to slightly lobed. Stamens 10, inserted on the 10-lobed disc, the 5 opposite the petals exterior and usually slightly longer, adnate to the bases of the petals, the 5 opposite the sepals subtended by a small exterior gland, occasionally sterile; filaments linear-subulate; anthers sagit- tate, sub-basifixed. Ovary sessile or essentially so, 5-lobed, 5-loculed, ovoid, densely hirsute-pilose, the appressed white hairs totally obscur- ing it; ovules 3-5 per locule, placentation axile; style 1-4 mm. long, 126 DUNCAN M. PORTER stout to cylindrical; stigma globose, 5-lobed, ca. 1 mm. in diameter. Fruiting pedicels appressed-villous, hispid, terminally reflexed down- ward; the horizontally flattened fruit consisting of 5 indehiscent nut- lets (rarely 1 or more aborting), each with 2-4 stout spines, occasional smaller spines, and aristate tubercles on the dorsal surface; nutlets hispid at the base, inner faces reticulate, leaving no central axis when falling, divided internally by oblique transverse septa into 3-5 1-seeded locules. Seeds oblong, covered with a white aril, ca. 3 mm. long and 1 mm. in diameter, completely filling the locules. TYPE SPECIES: Tribulus terrestris L. This genus has its center of distribution in the Mediter- ranean region, where members are mainly plants of the open desert. Many species have been described from this area, but most of these are undoubtedly synonyms of the ubi- quitous Tribulus terrestris. Two species are known from Baja California, one introduced, the other probably a native. KEY TO THE SPECIES Plants annual; piri leaves to 45 mm. long; leaflets 4-6 pairs; flowers ca. 5 mm. in diameter; pedicels usually shorter than the leaves 1. T. terrestris. Plants perennial; prostrate to suberect; leaves to 75 mm. long; leaflets pairs; flowers to ca. 4 cm. in diameter; pedicels usually longer ‘ikon the leaves 2. T. cistoides. 1. Tribulus terrestris L., Sp. Pl. 1:387. 1753 Prostrate annual, branches radiating from the root to a length of 3 m.; stipules 1-5 mm. long, to 1 mm. wide; leaves 10-45 mm. long, usually longer than ie pedicels; adic 4-6 (rarely 3) pairs, ee to slightly ovate, 1-4 mm. wide, 4-11 mm. long; flowers to ca. 5 mm. diameter; pedicels 2-7 mm. long: sepals ovate, 1.5-2 mm. wide, 2-3 mm. long; petals bright yellow, occasionally lighter at the bases, 2-3 mm. wide, 3-5 mm. long; filaments ca, 2 mm. long, anthers ca. 0.5 mm. in diameter ; wpe stout, ca. 1 mm. po ng; fruiting pedicels 5-15 sine long; fruits to ca. 1 em. in diameter, excluding spines; nutlets ca. 5 mm. high, the 2 ee spines 4-7 mm. long, spines puberulent es almost glabrate (more so than the rest of the nutlet), occasionally with a few smaller spines, these 1-2 mm. long and recurved. The gers chromosome aan have been reported for this species: n = 12 (Sugiura, 1940); 2n = 24 (Negodi, 1939; Heiser and Whitaker, on 2n = 48 (Schnack sn Covas, 1947). TYPE LOCALITY: “Habitat in Europa australi ad semitas” (Linnaeus, 1753, p. 387). The determination of the full synonomy of this species is at present most an impossibility. The genus Tribulus is in need of revision, preferably with the aid of cytotaxonomic methods. SPECIMENS EXAMINED: El Crucero, Porter 483 (CAS, DS); 46.5 mi. W La Paz, Porter 442 (cas, DS); 36 mi. W La Paz, Porter 445 (CAS, DS)} 22 mi. W La Paz, Porter 474 (cas, ps); Los Aripes, Porter 420 (DS); ZYGOPHYLLACEAE OF BAJA CALIFORNIA 127 1.5 mi. W La Paz, Wiggins 15714 (CAS, DS, GH) ; 1 mi. W La Paz, Port- er 355 (CAS, DS); La Paz, Thomas 7803 (DS); 6.5 mi. N La Paz, Porter 405 (CAS, ee Rancho El Ciprés, 8 mi. NE La Paz, Porter 373, 373A (CAS, DS); 1.5 mi. S La Paz airport, Porter 372 (CAs, DS); Todos Santos, Porter 343 (CAS, DS). Tribulus terrestris has been collected in the Cape Region of Baja California, but it is probably more common through- out the peninsula than present collections would indicate. Johnson (1958, p. 222) reports finding the plant in the northwest “on the dunes at the deserted town of San Quin- tin.” The species also surely occurs in the northeast near Mexicali, but I have seen no specimens from that area, which is immediately south of the heavily infested Imperial Valley of California. Wherever I have seen Tribulus terrestris in Baja California, it has been extremely common. This was the situation both at E] Pescadero and San José del Cabo. This noxious weed has undoubtedly invaded Baja Cali- fornia both from the south through the Cape Region, where it may have been introduced with ships’ ballast or livestock, and from the north through California, where it was first collected at Santa Monica in 1902 (Davidson and Moxley, 1923). Tribulus terrestris has long been a serious agricul- tural pest in the Central Valley of California, but there is no evidence that it has yet become so in Baja California. 2. Tribulus cistoides L., Sp. Pl. 1:387. 1753 Kallstroemia cistoides (L.) Endl., Ann. Naturg. Mus. Wien, 1:184. 1836. Tribulus terrestris var. cistoides (L.) Oliver, Fl. Trop. Africa, 1:284. 1868. Prostrate to suberect perennial; stipules 3-9 mm. long, 1-5 mm. wide; leaves 15 to 75 mm. long, usually shorter than the fruiting pedicels; leaflets 7 pairs, oblong to elliptic, 4-8 mm. wide, 10-19 mm. long; flowers to ca. 4 cm. in diameter; pedicels 8-40 mm. long; sepals lanceolate, 2-3 mm. ia 6-9 mm. long; petals bright yellow, 5-16 mm. wide, 8-22 mm. long; filaments 4-6 mm. long, anthers ca. 2 mm. long; style cylindrical, 3-4 mm. long; fruiting pedicels 20-45 mm. long; fruits 5 cm. in diameter, excluding spines; nutlets ca. 10 mm. high, the 4 large spines 5-8 mm. long, more densely hispid than in the former species. a LOCALITY: “Habitat in America calidiore” (Linnaeus, 1753, p. 387). SPECIMEN EXAMINED: San José del Cabo, Porter 326 (CAS, DS). This large-flowered Tribulus is primarily a plant of the 128 DUNCAN M. PORTER coastal strand, especially common in the New World tropics and the Southwest Pacific, where it is apparently indigenous. T. cistoides is known to occur from Florida to Texas and Georgia in the southeastern United States (Small, 1933), and there is a collection of the plant in the Herbarium of the California Academy of Sciences (Skoss, 7 May 1948) labeled, “Kern River Valley 7 Miles below Hobo Hot Springs, Kern County [California].” This specimen is the only one of which I am aware from the western United States. Tribulus cistoides has been collected in Baja California only at San José del Cabo, where it is quite common, some- times growing side by side with 7. terrestris. According to Ira L. Wiggins (personal communication), the species is also abundant at the La Paz airport. Vail (1895) lists T. cistoides as occurring in Baja California, but I have seen no specimens from the peninsula other than my own. 6. Kallstroemia Scop., Introd. 212. 1777 Prostrate to suberect annual herbs; diffusely branching, branches decumbent, striate, green, becoming yellow, to a meter or more in length; herbage pubescent, youngest parts and nodes more densely so, oldest parts occasionally glabrate. Stipules subulate to lanceolate, green to somewhat membranaceous, pubescent. Leaves opposite, even- pinnate, one alternate leaf smaller than the other or aborting, light to dark green. Leaflets 2-6 pairs, petiolulate, oblong to elliptic, usually mucronate, inequilateral, base oblique, the terminal pair usually broad- er and shorter than the others, more or less pubescent; rachis densely appressed villous and hispid, a small green mucro at the apex between the ultimate leaflets. Flowers solitary; pedicels in the axils of the alternately smaller leaves, pubescent, thickened upward, longer or shorter than the subtending leaves. Sepals 5-6, lanceolate to subulate, densely pubescent, marcescent, persistent to deciduous. Petals 4-6, more or less reticulate on the inner faces, 1-seeded, falling away from ite Leste styliferous axis. Seeds obovate, completely filling the ocules. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 129 TYPE SPECIES: Tribulus maximus L. Kallstroemia is a genus of a dozen or so species found in the more arid tropics and subtropics of the New World. It occurs from the southern borders of the United States to Brazil and Argentina, but members are most common in Mexico and the West Indies. The genus was originally in- cluded in Tribulus and has from time to time been reas- signed there by a few botanists. However, the fruits and their development are of sufficient distinctness to warrant the separation of the two genera. KEY TO THE SPECIES Plants prostrate to suberect; leaves 15-65 mm. long; pedicels longer than the subtending leaves; petals 12-20 mm. long; beak of the fruit 5-10 mm. lon 1. K. grandiflora. Plants prostrate; leaves 5-40 mm. long; pedicels shorter than the sub- te ws leaves; petals 3-12 mm. long; beak of the fruit 1-7 mm. lon Petals 6-12 mm. long; sepals 5-8 mm. —— beak of the fruit cylin- ot drical from a conical base, 4-7 mm. long ............ parviflora, Petals 3-6 mm. long; sepals 2-4 mm. long; beak of the fruit conical, 1-3 mm. long 3. K. californica. 1. Kallstroemia grandiflora Torr. ex Gray, Pl. Wright. 1:28. 1852 Kalistroemia grandiflora var. detonsa Gray, loc. cit. Tribulus grandiflorus (Torr. ex Gray) Benth. & Hook. ex Brew. & Wats., Bot. Calif. 1:91. 1876. Tributes fisheri Kell. Proc. Calif. Acad. 7:162, 1877. Kallstroemia grandiflora var. arizonica Cockerell, Bull. Torr. Bot. Club 27:87. 1900. Prostrate to suberect; herbage hispid and appressed-villous or sericeous; stipules subulate to lanceolate, 1-2 mm. wide, 2-7 mm. long; leaves 15-65 mm. long; leaflets 2-5 pairs, elliptic to oblong, both sur- faces sericeous and appressed-villous, underside more so, 5-12 mm wide and 10-22 mm. long; pedicels usually longer than the subtending leaves, 10-55 mm. long; sepals lanceolate, 1-2 mm. wide, 6-10 mm. long, surpassing the fruit, but shorter than the style, persistent; petals orange, occasionally turning white with age, 7-14 mm. wide, 12-20 mm. long, apex broadly rounded; filaments subulate, 4-9 mm. long; anthers -3 mm. long; ovary 2-3 mm. in diameter; style 6-8 mm. long; stig- Matic areas extending along the upper half of the style; fruiting pedi- cels 15-65 mm. long; fruits to 5 mm. in diameter; nutlets with rounded tuberculae, 3 mm. high; beak of the fruit cylindrical, hardly thickened at the base, 5-10 mm. long. In Baja California flowering from August to March and fruiting from September to March. 130 DUNCAN M. PORTER TYPE LOCALITY: “Borders of the Gila [Arizona], Col Emory” (Gray, 1852, p. 28). SPECIMENS EXAMINED: 95 km. NW La Paz, Wiggins 15413 (CAS, DS) ; 24 mi. W La Paz, Porter 438 (Ds); Arroyo Seco, Hammerly 189 (CAS, DS); 15 mi. W La Paz, Hammerly 222 (cAs, Ds); 10 mi. SE La Paz toward Los Planes, Wiggins 15686 (CAs, DS); 1 mi. S San Antonio toward Santiago, Chambers 860 (Ds, UC) ; 12.8 km. N Santiago, Carter, Alexander, & Kellogg 2181 (Ds); Los Frailes, Porter 318 (CAS, DS); Arroyo Salado, Purpus 409 (uc);San José del Cabo, Brandegee 78 (uc), Brandegee, 5 Sept. 1891 (ps); 3 mi. NE San Lucas, Porter 331 (CAS, DS) ; 14 mi. N San Lucas, Porter 341 (CAS, DS). Kallstroemia grandiflora occurs from southern California east to Texas and south to Michoacan. Vail (1895) states that it is found south to Guatemala, but I have seen no speci- mens from that far south, nor have I seen any other ref- erences to substantiate her statement. In Baja California this plant is found in the Cape Region and the southern part of the Magdalena Region. It is mainly a plant of low sandy areas and beaches, but Johnson (1958) states that she saw it at a height of between 3500 and 4500 feet in the Sierra Laguna. Brewer and Watson (in Brewer, Watson, & Gray, 1876) erroneously cite the combination “T. grandiflorus Benth. & Hook.”, giving the source as “Gen. Pl. i. 264.” Bentham and Hooker, however, never made this combination ; they simply listed Kallstroemia as a synonym for Tribulus. 2. Kallstroemia parviflora Nort., Ann. Rep. Mo. Bot. Gard. 9:153. 1898 Kallstroemia laetevirens Thornb. in Woot. & Standl., Contr. U.S. Nat. Herb. 16:143. 1913. Prostrate; herbage densely to sparingly hispid and villous or serice- ous; stipules lanceolate, 1-1.5 mm. wide, 5-7 mm. long; leaves to 40 mm. long; leaflets 3-5 pairs, oblong, sericeous, underside more so than upper; 3-6 mm. wide, 6-13 mm. long; pedicels shorter than the subtending leaves, 10-20 mm. long; sepals subulate to lanceolate, to 1.5 mm. wide, 5-8 mm. long, more or less persistent; petals yellow to orange, 3.5-9 mm, wide, 6-12 mm. long; filaments linear-subulate, 3-6 mm. long; ovary 1-2 mm. in diameter; style 3-4 mm. long; fruiting pedicels 14-45 mm. long; fruits 4-5 mm. in diameter; nutlets with rounded tuberculae, 3-4 mm. high; beak of the fruit cylindrical, thickened at the base, 4-7 mm. long Known to flower and fruit in Baja California from October to Janu- ary TYPE LOCALITY: “Collected at Agricultural College, Miss., by Pollard, Aug., 1896, no. 1295” (Norton, 1898, p. 153). ZYGOPHYLLACEAE OF BAJA CALIFORNIA 131 SPECIMENS EXAMINED: Near km. 40 on main highway NW La Paz, Thomas 8439 (DS, GH); 15.5 mi. SE La Paz toward Las Cruces, Wig- gins 15669A (CAS, DS); near fork of road toward Punta Arena, Wig- gins 14454 (CAS, DS); 3 km. N Cabo San Lucas, Moran 7040 (CAs, DS). Kallstroemia parviflora is found from southern California east to Mississippi and south to Michoacan, in Mexico. It occurs sparingly in the Cape Region of Baja California. With the exception of Thomas 8439, all the specimens ex- amined approach Kallstroemia grandiflora in petal size, but their other characters indicate that they are in all prob- ability large-flowered examples of K. parviflora. 3. Kallstroemia californica (S. Wats.) Vail, Bull. Torr. Bot. Club 22:230. 1895 Tribulus californicus S. Wats., Proc. Amer. Acad. 11:125. 1876. KEY TO THE VARIETIES Leaflets 3-6 pairs, to 3 mm. wide and 7 mm, long; petals 1-2 mm. wide, 3-5 mm. long; beak of the fruit 1-2 mm. long; nutlets ssa tuberculate 3a, K. californica var. californ Leaflets 2-4 pairs, 3-10 mm. wide, 6-21 mm. long; petals 28 mm. athe 4-6 mm. long; beak of the fruit 2-3 mm. long; nutlets rounded tuberculate 3b. K. californica var. brachystylis. 3a. K. ealifornica var. californica Prostrate; herbage hispid and appressed-villous or sericeous; stipules subulate, to 1.5 mm. wide, 2-3 mm. long; leaves 5-35 mm. long; leaflets 3-6 pairs, elliptic, sericeous, especially on underside, becoming almost glabrate, to 3 mm. wide and 7 mm. long; pedicels shorter than the subtending leaves, 2.5-10 mm. long; sepals lanceolate, to 1 mm. wide, 2-3 mm. long, deciduous; petals cream to yellow, 1-2 mm. wide, 3-5 mm. long; filaments filiform, 2 mm. long; ovary ca. 1 mm. in diameter; style to 1 mm. long; fruiting pedicels 4-26 mm. long; fruits 5 mm, in diameter; nutlets sharply tuberculate, 3 mm. high, the 4 or 5 tuber- culae ‘eh ets mm. long; beak of the fruit thick, conical, glabrous, 1-1.5 mm. long. Flowering and fruiting in September and October in Baja Cali- fienin: TYPE LOC : “Collected by Dr. E. Palmer, in Lower California (1870), on the eaehiri lhc the peninsula” (Watson, 1876, p. 125). According to McVaugh (1956), in 1870 Edward Palmer collected in Baja California at and near La Paz (5 January) and Cabo San Lucas ie January and 7 February) and on Isla Carmen (1 January). He may also have been at the eng of the Colorado River in late August or early September of that y IMENS EXAMINED: ec California, Palmer, 1870 (GH!); 1.5 132 DUNCAN M. PORTER mi. SW Los Martires, Thomas 8317 (pS); San José del Cabo, Brande- gee 80 (UC), Sept. 1891 (Ds, GH), 28 Sept. 1893 (uc). Kallstroemia californica var. californica is known to occur from southern California and Arizona to Sinaloa. This variety is probably more common in Baja California than the above collections would indicate. It is not nearly so common, however, as K. californica var. brachystylis, either in Mexico or in the United States. None of the Baja California collections of this variety has such sharply tuberculate nutlets as specimens from Cali- fornia and Arizona. They do, however, have the many small leaflets that characterize Kallstroemia californica var. cali- fornica. 3b. K. californica var. brachystylis (Vail) Kearn. & Peeb., Journ. Wash. Acad. 29:485. 1939 Kallstroemia brachystylis Vail, Bull. Torr. Bot. Club 24:206. 1897. Tribulus brachystylis (Vail), Rob. in Gray, Syn. Fl. N. Amer. 1 354. 1897. te; herbage hispid and appressed-villous or sericeous; stipules subulate, to 1 mm. wide, 2-4 mm. long; leaves 5-30 mm. long; leaflets 2-4 pairs, oblong to elliptic, sericeous, more so on underside, to almost glabrate, 3-10 mm. wide, 6-21 mm. long; pedicels shorter than the subtending leaves, 2-20 mm. long; sepals subulate, to 1.5 mm. wide, 2.5-4 mm. long, deciduous; petals cream to orange, 2-3 mm. wide, 4-6 mm. long; filaments filiform, 2-3 mm. long; ovary ca. 2 mm. in di- ameter; style 1.5-2 mm. long; fruiting pedicels 8-25 mm. long; fruits 4-5 mm. in diameter; nutlets with rounded (rarely a few sharp) tuber- culae, 3 mm. high; beak conical, slightly thickened at the base, 2- mm. long. Flowering and fruiting from August to March. TYPE LOCALITY: “Mesa near Las Cruces, N. M., alt. 3900 ft. Col- lected by E. 0. Wooton, Aug. 12, 1895” (Vail, 1897, p. 206). SPECIMENS EXAMINED: 15 mi. N Bahia de San Luis Gonzaga, Wig- —_— 16037 (Cas, DS); 40 mi. N Laguna Chapala toward San Felipe, Wiggins & Ernst 679 (cas, Ds) ; Laguna Chapala, Thomas 8201 (DS) ; Bahia de los Angeles, Wiggins 15002 (cas, ps); 12 mi. E El Solito, Wiggins 15144 (CAS, DS); 6 mi. S Rancho Los Angeles, Wiggins 15166 (eats DS) ; 57 km. NW San Ignacio, Carter, Alexander, & Kellogg 1951 (DS); between San Ignacio and Los MArtires, Gentry 7873 (DS, UC); 7 mi. E. San Ignacio, Wiggins 11355 (cas, ps, GH, UC); 24 mi. NW Santa Rosalia, Chambers 979 (ps); 6 mi. NW Cuarenta, Thomas 8362 (Ds, GH) ; El Coyote, Hammerly 128 (cas, ps, GH); 14.4 km. SW Com- ae Carter, Alexander, & Kellogg 2122 (ps); Rancho del Cayuco, arter & Kellogg 3127 (ps), Jones 27113 (uc); 10 mi. S Missién los Dolores, Wiggins, Carter & Ernst 322 (CAs, DS, UC); 4.5 km. N Pozo Grande, Carter, Alexander & Kellogg 2129 (ps); 8.5 mi. N Santo ZYGOPHYLLACEAE OF BAJA CALIFORNIA 133 Domingo toward Pozo Grande, Thomas 8390 (Ds) ; 18 mi. S El Crucero, Porter 452 (CAS, eh 16 mi. S El Refugio, Wiggins 11495 (Ds, UC), 11497 (ps); 3.6 . N Punta del Cerro, Wiggins 15468 (CAS, DS); Rancho Colorado, greeting 17 Jan. 1890 (ps, uc), 19 Jan. 1898 (cAS); near km. 40 on main highway NW La Paz, Thomas 8439A (ps); 15 mi. W La Paz, Hammerly 194 (CAS, DS); 8.5 mi. N La Paz, Porter 384, 384A (CAS, DS); 8.9 mi. E Los Planes, Wiggins 14435 (CAS, ps); La Palmilla, Wiggins 15019 (CAS, DS); San José del Cabo, Bran- degee 79 (UC), 21 Sept. 1891 (GH), 28 Sept. 1893 (UC). Kallstroemia californica var. brachystylis occurs from southern California to Texas and south to Central America. It is found in Baja California in all regions except the Cali- fornian. It is by far the most common Kallstroemia on the peninsula and occurs mainly on flat sandy areas. Hammerly 128 and 194 have been elsewhere cited (John- son, 1958) as Kallstroemia glabrata Rydb. This name may prove to be a synonym of K. californica var. brachystylis. This variety is often difficult to separate from Kallistro- emia californica var. californica, as there is a distinct over- lap between the two in the size of the tubercles. According to Kearney and Peebles (1939, p. 485), “As compared with typical K. californica, this form has usually fewer leaflets and shorter, blunter tubercles on the carpels, but there is too much intergradation to warrant specific distinction.” Here is another instance in which cytotaxonomic methods would undoubtedly help delimit the differences between entities in this family. LITERATURE CITED AXELROD, D. I. 1950. Evolution of desert vegetation in western North America. Carnegie Inst. Wash. Publ. 590 pp. 215-306. BENTHAM, G. 1844. The Botany of the Voyage of H. M. S. Sul- phur. R. B. Hinds ed. London BRANDEGEE, K. 1888. Hotanical notes. I. Baja California. Proc. Calif. Acad., ser. 2. 1:227-237. BRANDEGEE, T. S. 1891. Flora of the oo Region of Baja California. Proc. Calif. Acad., ser 2. 3:108-1 . Plantae a Purpusianae. VII. Univ. Calif. Publ. Bot. 6:177-197. . H., S. Watson, & A. Gray. 1876. Botany of California. Vol. jhe Cambridge, Mass Covas, G. 1949. Estudios Envibibeions en antofitas. III. Darwiniana. 9:158-162. , & B. ScHNackK. 1946. Numero de chromosomas en antéfitas de la regién de Cuyo (Republica Argentina). Rev. Argent. Agron. 13:153-166. 134 DUNCAN M. PORTER DaruineTon, C. D., & A. P. WYLIE. 1955. Chromosome Atlas of Flowering Plants. ed. 2. London. Davipson, A., & G. L. Moxtey. 1923. Flora of Southern California. Los Angele DECANDOLLE, A. Pp. 1824. Prcertnias Systematis Universalis Regni Vegetabilis. Vol. 1. Pari GarclA, E., C. Soto, & F. iaisie 1960. Larrea y clima. An. Inst. Biol. Tinks, Méx, 31:133-190. GoLpMAN, E. A. 1916. Plant records of an expedition to Lower Cali- fornia. Contr. U. S. Nat. Herb. 16:309-371. Gray, A. 1852. Plantae Wrightianae Texano-Neo-Mexicanae, Part 1. Washington, D.C. . 1897. Synoptical Flora of North America. Vol. 1., No. 1. B. L, Robinson ed. Cambridge, Mass. HeIser, C. B., & T. W. WHITAKER. 1948. Chromosome number, poly- ploidy, and growth habit in California weeds. Amer. Journ. Bot. 35 :179-186. JoHNSON, B. H. 1958. The botany of the California Academy of Sciences expedition to Baja California in 1941. Wasmann Journ. Biol. 16:217-315. JOHNSTON, I. M. 1922. Undescribed plants mostly from Baja Cali- fornia. Univ. Calif. Publ. Bot. 7:437-446. . 1924. Expedition of the California Academy of Sciences to the Gulf of California in 1921. The botany (vascular plants). Proc. Calif. Acad., ser. 4. 12:951-1218. Taxonomic records concerning American spermatophytes. II. New or otherwise noteworthy plants. Contr. Gray Herb. 70:69-87. . 1940. The floristic significance of shrubs common to North and “South American deserts. Journ. Arnold Arb. 21:356- 363. KEARNEY, T. H., & R. H. PEEBLES. 1939. Arizona plants: new species, varieties, and combinations. Journ. Wash. Acad, 29:474-492. KELLOGG, A. 1863. Plants collected by Dr. John A. Veatch on Cerros Island. Proc. Calif. Acad. 2:21-27. LanJouw, J., & F. A. STAFLEAU. 1959. Index Herbariorum. Part 1. The Herbaria of the World. Regnum Vegetabile. Vol. 15. LinnaEus, C. 1753. Species Plantarum. Vol. 1. Stockholm. : McVaueu, R. 1956. Edward Palmer, Plant Explorer of the Amer!- est. Norman, Okla. MONTICELLI, J. V. 1939. El genero Larrea Cavanilles. Su historia ¥ revision. Physis 15:331-356. MOoRELLO, J. 1958. La Provincia Fitogeografica del Monte. Opera Lilloa. Vol. 2. Tucuman, Argentina NeEcopI, G. 1939. Cariologia delle Rutaceae e delle Zygophyllaceae- Sci. Genet. 1:168. Cited by Darlington & Wylie (1955). Norton, J. B. J. 1898. Notes on some plants, chiefly from the south- western United States. Ann. Rep. Mo. Bot. Gard. 9:151-157. PorTER, D. M. 1961. A note on the reported chromosome numbers for the genus Larrea. Contr. Dudley Herb. 5:169. ZYGOPHYLLACEAE OF BAJA CALIFORNIA 135 RaGonEsE, A. M. 1960. Estudio anatomico de las especies Argentinas de Larrea. Rev. Invest. Agricolas 14:355-370. RAHN, K. 1960. Danish scientific investigations in the Argentine under the auspices of the Fundacién Williams, Buenos Aires. Chromosome numbers in some South American angiosperms. Bot Tidsskrift 56:117-127. RzeEpowskI, J., & F. MEpELLiN LEAL. El limite sur de distribucién geografica de Larrea tridentata. Acta Sci. Potosina 2:133-147. ScHNACK, B., & G. Covas. 1947. Estudios cariolégicos en antofitas. Haumania 1:32-41. SHREVE, F. 1926. The desert of northern Baja California. Bull. Torr. Bot. Club 53:129-136. —___. 1936. The transition from desert to chaparral in Baja California. Madrono 3:257-264 ————. 1937. The vegetation of the Cape Region of Baja Cali- fornia. Madronio 4:105-113. 1 The edge of the desert. Yearbook Assoc. Pac. Coast. ; 951. Vegetation of the Sonoran Desert. Carnegie Inst. SMALL, J. K. 1933. Manual of the Southeastern Flora. New York. STANDLEY, P. C. 1911. The American species of Fagonia. Proc. Biol. Soe. Wash. 24:243-250. Sucrura, T. 1940. Studies in the chromosome numbers in higher plants. V. Cytologia 10:363-370. Vait, A. M. 1895. A preliminary list of the North American species of Malpighiaceae and Zygophyllaceae. Bull. Torr. Bot. Club 22: 228-231. 1897. An undescribed species of Kallstroemia from New Mexico. Bull. Torr. Bot. Club 24:206-207. —__________ 1899. Notes on Covillea and Fagonia. Bull. Torr. Bot. Club 26:301-302. —_—____—_., & P. A. Rypperc. 1910. Zygophyllaceae. N. Amer. FI. 25:103-11 6. Watson, S. 1876. Botanical contributions. III. Descriptions of new species of plants, chiefly Californian, with revisions of certain genera. Proc. Amer. Acad. 11:121-148. Wicarns, I. L. 1960. Biogeography of Baja California and adjacent seas. The origins and relationships of the land flora. Syst. Zool. 9:148-165. CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by : Reed C. Rollins and Robert C, Foster NO. CXCIIl A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA BY Siri VON Reis ALTSCHUL INFRASPECIFIC VARIATION IN THE GUTIERREZIA SAROTHRAE COMPLEX (COMPOSITAE-ASTEREAE) By Otto T. SoLBRic ed by ale THE GRAY HEREARIUM OF HARVARD UNIVERSITY ‘CAMBRIDGE, MASS,, U.S. A. i CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C. Rollins and Robert C, Foster NO. CXCIII A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA By SIRI VON REIS ALTSCHUL INFRASPECIFIC VARIATION IN THE GUTIERREZIA SAROTHRAE COMPLEX (COMPOSITAE-ASTEREAE) By OTTo T. SOLBRIG Published by THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U.S. A. Issued June 25, 1964 A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA SIRI VON REIS ALTSCHUL The small, tropical to subtropical, and strictly New World genus Anadenanthera formerly was considered as section Niopa of the genus Piptadenia and is the common legumi- nous source of the so-called narcotics known as Cohoba, Vilca and Yopo. Because of their unusual effects upon the human nervous system, the chemical constitutents of these materials are of current interest in the field of experimental psychia- try. Among the chemical compounds which have been iso- lated from the species of Anadenanthera may be included some hallucinogenic drugs designated as psychotomimetics (Hofmann, 1959). In contrast to other psychotropic drugs, which merely evoke modifications in the mood of the recipi- ent, the psychotomimetics appear to produce profound and acute changes in perception. An interest both in plants of possible medical significance and in species in need of taxonomic study provided a stimulus for the research whose results are published in part in this paper. An ethnobotanical treatment which also has been prepared will be published elsewhere. A detailed systematic treatment of the species of Anadenanthera has been needed for some time. While additional work still remains to be carried out on this group, as well as on related species in the genus Piptadenia, it is hoped that the data presented here from several branches of botany will help better to define Anadenanthera and to stimulate further research in the large and relatively neglected subfamily Mimosoideae. ACKNOWLEDGEMENTS I would like to express my deep appreciation to Professor Reed C. Rollins for guiding the course of my research and for his helpful criti- cism in the preparation of this paper. Professor John H. Welsh aided in the physiological aspects of the problem, with regard to current research on the effects of Anadenan- thera derivatives and related compounds upon the nervous system. Professors I. W. Bailey and Taylor A. Steeves were kind enough to suggest sources of information on leaf venation patterns and foliar anatomy. Professor Gordon R. Willey provided assistance with regard to archaeological data. In particular, I am grateful to Professor 4 SIRI VON REIS ALTSCHUL Richard A. Howard, to Dr. Richard E. Schultes and to Dr. Rolla M. Tryon for their careful scrutiny of the manuscript in the last phases. I also wish to thank the late Dr. C. E. Kobuski, and the other curators who made possible the loans of materials from the following herbaria: Arnold Arboretum (A); British Museum of Natural History (BM); Chicago Natural History Museum (F); Copenhagen Botanical Museum and Herbarium (c); Gray Herbarium (GH); Kew Herbarium (K); Rijksherbarium, Leiden (L); Missouri Botanical Garden (MO); New York Botanical Garden (Ny); Muséum National d’Histoire Naturelle, Paris (Pp); Naturhistoriska Riksmuseet, Stockholm (8); United States National Herbarium (us); Utrecht Botanical Museum and Herbarium (v). : Dr. Robert C. Foster’s scholarly interpretations helped to clarify questions about stylistic forms of early taxonomic literature. Dr. Carroll E. Wood, Jr., offered helpful suggestions as to presentation of materials. Dr. Lily M. Perry contributed from her understanding of species from Pacific regions. ; I would like to thank Dr. Joab L. Thomas for instruction in certain laboratory techniques and Dr, Alexander Grobman for exchanging data from his own current research, appreciation to Mr. Fred H. Chamberlin and to Dr. Alice F. Tryon for their help in preparing this paper. Finally, I am deeply grateful to my parents, Gustav and Tyyne von Reis, for their financial support and to my dear friend, Arthur GB Metcalf, for facilitating in countless ways the completion of this wee THE GENUS ANADENANTHERA: HISTORY AND THE CHARACTERS OF DISTINCTIVENESS The species comprising the leguminous genus Anadenan- thera as here conceived were treated originally as belonging to section Niopa of the genus Piptadenia Bentham (1840, 1841-42). The species of Piptadenia were not well known during Bentham’s lifetime, but that author was quick to recognize that a natural distinction could be made between sections Ewpiptadenia and Pityrocarpa with spicate inflores- cences, and section Niopa with capitate inflorescences. Ben- tham distinguished five species in section Niopa, though he wondered whether they might not better be considered aS A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA = 5 varieties of a single species. The fruits differed considerably in form and proportion, but the flowers and foliage were very much alike from one specimen to another. Bentham later included four of the original species in his “Revision of the Suborder Mimoseae” of 1874-75: Piptadenia peregrina (L.) Bentham, P. macrocarpa Bentham, P. falcata Bentham and P. colubrina (Vell.) Bentham. He placed a fifth species, P. microphylla Bentham, in the synonymy of P. macrocarpa. Since Bentham’s time, three additional species have been described for section Niopa of Piptadenia. Piptadenia novo- guineensis Warburg was published first as the monotypic genus Schleinitzia microphylla Warburg (1891) and, for reasons discussed in this paper, probably should be returned to its former status. Piptadenia Hassleriana Chodat (1904) is rcferred to synonymy under Anadenanthera colubrina var. Cebil in the present treatment. Piptadenia amazonica Ducke (1915) was recognized by Ducke in 1922 as belonging to Pithecellobium Martius (1837). Piptadenia leptoclada Baker (1887) was based on a speci- men with a globose-spicate inflorescence which could be mis- taken at first glance for a species intermediate between sections I and II and section III of Piptadenia sensu Ben- tham. The specimen was identified by Perrier de la Bathie in 1938 as Desmanthus Commersonianus Baillon (1883). In 1923, Spegazzini founded the genus Anadenanthera to accommodate Piptadenia peregrina and P. falcata. He — the genus on the flowers’ being “without anther glands”, character of doubtful generic value by itself. Saunt stated in his description that he did not include in Anadenan- thera five southern species he had not seen, but which were mentioned in Engler & Prantl (1892). An inspection of those authors’ work suggests that Spegazzini must have meant three, not five, other species. For the species named in Engler & Prantl are: Piptadenia peregrina, P. falcata, P. macrocarpa, P. colubrina and P. novo-guineensis. The last three have anther glands. In 1927, Britton & Rose proposed to elevate the whole of section Niopa to a genus. They offered, however, no reason for the change, added no new information and did not recog- nize the priority of the name Anadenanthera. In 1955, Brenan proposed a revision of the genus Pipta- 6 SIRI VON REIS ALTSCHUL denia according to a new set of characters involving, in particular, the mode of dehiscence of the pod and the struc- ture of the seed. The differences he prescribes are compar- able to those used elsewhere in the Mimosoideae to distinguish genera. Furthermore, they are not infrequently correlated with vegetative differences. Brenan presents fruit and flower keys to eight genera which have species formerly included in Piptadenia. These keys delimit groupings which seem to be more satisfactory than have been achieved here- tofcre. How well some of the genera will be upheld when their species become better known is uncertain, but by con- servative judgment, the most natural of them is Anadenan- thera, which Brenan extends to include all species of former section Niopa of Piptadenia. He distinguishes Anadenan- thera by the following combination of characters: the glob- ose inflorescence; the dehiscence of the pod along one suture only ; the suborbicular, narrowly or not winged seeds lacking endosperm; and the strictly American distribution. The evidence available on wood anatomy supports the reasonableness of upholding the genus Anadenanthera. Within the Leguminosae as a whole, genera of the Caesal- pinioideae and especially the Mimosoideae are noted for a relatively homogeneous wood structure (Co0zz0, 1951). Yet Tortorelli indicated preliminarily in 1948 that the element here called Anadenanthera colubrina var. Cebil is charac- terized by having incompletely paratracheal parenchyma, with rays three or more cells wide and up to 550 » high. By comparison, at least two other species of the Piptadenias have wood with definite paratracheal vasicentric parenchy- ma, rays uniseriate or biseriate and up to 350 » high : Pipta- denia excelsa (Grisebach) Lillo (1910) and P. rigida Ben- tham (1842). Recently, Brazier (1958) discovered that differences in wood anatomy within the whole of what now may be referred to as the ‘Piptadenia complex’ (Brenan 10 a letter, October 8, 1959) coincide almost perfectly with the eight generic distinctions proposed by Brenan. Brazier § key to these genera indicates that Anadenanthera may be characterized, and distinguished from other genera of the Piptadenia complex, by the presence of non-septate fibers and wide, typically (4-)5-seriate rays. An interest in the Piptadenias on the part of the medical A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 7 world has led to investigations into the chemical constituents of the reputedly hallucinogenic seeds. This research has shown that the species of Anadenanthera are in some ways physiolegically different from other species in the Piptadenia complex. In 1955, Fish, Johnson & Horning indicated that the seeds of what now are referred to as Anadenanthera peregrina var. peregrina and A. colubrina var. Cebil yielded a much larger amount of alkaloid material than was found in Piptadenia paniculata Bentham (1842). Fish reported (in a letter, January 7, 1958) a much lower concentration of bufotenine and related substances in three additional species from the Piptadenia complex than was found in the seeds of Anadenanthera peregrina var. peregrina and A. colubrina var. Cebil. The three species tested were Piptadenia com- munis Bentham (1842), P. contorta Bentham (1875) and P. leptostachya Bentham (1842). The information reviewed above, plus the data presented in the pages which follow, require that Anadenanthera be recognized as a well substantiated genus of the Mimosoideae. THE SPECIES OF ANADENANTHERA MATERIALS AND METHODS: This revision is based on her- barium specimens, field work and the literature. Because the fruits still are not known for many species of the Pipta- denia complex, flowering specimens usually are necessary for identification of material. Over 200 specimens were examined for this study. The leaflets reproduced in Plates VII-IX and Plate XI were cleared by a technique which is essentially that given by A. S. Foster (1950) for studying foliar venation in angio- sperms. The leaflets first were softened in water brought almost to the boiling point. Afterwards, they were soaked in 6% sodium hydroxide for 24-36 hours in an oven at 60 degrees Centigrade. The time required for decolorization varied with the samples. A small, capped vial was used for about a dozen leaflets from each field number. Some of the vials darkened quickly, and the solutions had to be replen- ished. Leaflets of Anadenanthera peregrina var. falcata, in particular, released what appeared to be large amounts of pigments or tannins. Differences in decolorizing suggest that a chemical study of the leaflets might uncover facts pertinent 8 SIRI VON REIS ALTSCHUL to the taxonomy of the genus. When the leaflets had become almost transparent, they were cleared in 85% lactic acid for 10 minutes over a boiling bath. They then were stored in capped bottles of fresh lactic acid for a day or two. The staining was accomplished by bathing the leaflets in distilled water and subsequently in a solution of 14,% safranin and 50% alcohol, for less than 20 minutes. Before becoming too heavily stained, the leaflets were removed, 1-3 to a glass slide. Before they could dry, the leaflets were flooded with successive solutions of 70%, 95% and 100% alcohol and, finally, xylol. A drop of damar was placed under the cover slip. In this study the petals and calyces were not used. They are extremely delicate, nearly clear and finely 1-veined. DISCUSSION: The genus Anadenanthera consists of two species. Each is distinguished by a constellation of a few constant morphological characters and a geographical dis- tribution which is partly overlapping with that of the other species. Specimens can be identified most easily as belonging to either of these two species on the basis of the pod texture. This division correlates with the presence or absence of a minute gland on each anther and with the position of a small involucre surrounding the peduncle. One species, Anadenan- thera peregrina, has dull, seurfy to verrucose pods, eglandu- lar anthers and the involucre is about three-quarters of the way up the peduncle. It is the more northern-ranging species occurring from southeastern Brazil to the Greater Antilles. The other species, Anadenanthera colubrina, has nitid, smooth to reticulated pods, glandular anthers and the in- volucre just under the receptacle. It is limited to the south- ern hemisphere and occurs from central Peru to northern Argentina to northeastern Brazil. Each of the above species may be divided further into two groups of individuals. These groups each can be distin- guished by a set of morphological characters correlated with a particular geographical distribution. The differences !- volved are relative, in that they may appear to be trends when a series of specimens within a species is observed. They also are outstanding enough for representatives 1 which the characters are fully expressed to have been con sidered as belonging to separate species. In this treatment, they are regarded as varieties. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 9 PLaTe I. Fig. 1. Anadenanthera peregrina var. peregrina in flower, from Puerto Rico (P. Sintenis ‘ig. 2. Anadenanthera peregrina var. peregrina in fruit, rom Venezuela (Humboldt 1159?, Willdenow Herbarium 19057). Fig. 3. Anadenan- thera peregrina var. falcata in fruit, from Brazil (E. Hemmendorff 285). Note the dull pod typical of this species; the faleate shape of the pod, the heavier twigs, thicker, fewer leaflets are characteristic of this variety. Fig. 4. Anadenanthe var. Cebil ra colubrina in flower, from Bolivia (J. Steinbach 6657). 10 SIRI VON REIS ALTSCHUL Anadenanthera peregrina var. peregrina (Plate I, figs. 1 & 2) is a tall tree with trunks and branches which are not very corky. The pinna pairs tend to be more numerous but the leaflets fewer, shorter, and straighter than in the other variety, A, peregrina var. falcata. The leaflets and other vegetative parts of var. peregrina generally are thinner. The heads are greener and may be smaller. The legume tends to be longer with a broader range of widths and usually is straight. This variety is found in northern Brazil, British Guiana, Colombia, Venezuela and probably is naturalized where found in the West Indies. Specimens show this to be the more widespread and more northern-ranging variety of A. peregrina. Anadenanthera peregrina var. falcata (Plate I, fig. 3) 18 a much shorter tree with suberose trunks and branches. The leaflets usually are falcate, coriaceous, contrasting dorsivent- rally and darker in dried specimens than those of var. pere- grina. The heads of var. falcata are yellowish, and the legume typically is faleate. This variety is found in southern Brazil and Paraguay. Specimens examined show it to be geographically adjacent to var. peregrina, representing the southernmost stronghold of the species as a whole. Anadenanthera colubrina var. Cebil (Plate I, fig. 4; Plate II, figs. 1 & 2) is a tall tree, with longer but flatter petiole glands than the other variety, A. colubrina var. colubrina. The pinna pairs of var. Cebil may be more numerous but the pinnze shorter than in the other variety. The leaflets of var. Cebil tend to be fewer, shorter, dilated in the middle, with prominent secondary veins. The heads may be larger and usually are borne axillary to the leaves and in a subterminal position, not frequently becoming arranged in racemose pat- terns in the branch apices. The legume is shorter, wider and thicker margined. The seeds are fewer and larger. This variety is the more variable of the two and is found in Arse tina, Bolivia, Brazil, Paraguay and Peru. Specimens exam- ined show it to be also the more widespread variety. It 18 broadly overlapping geographically with the distribution of var. colubrina. Anadenanthera colubrina var. colubrina (Plate II, figs. 3 & 4) is a shorter tree with leaflets which tend to be linear and nitidulous above, with only the mid-vein prominent. e — A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 1 ecw: sean Mtl ap Fe ae ee Pannier ane as PuLaTe Il. Fi oe colubrina var. Cebil, from empape (E. Hassler ries? s.n.). Fig. 3. Anade ie colubrina var. colub a i Scie. panne ike Brazil (Clas iou 5831). Here the tendency for inflorescences to be borne in terminal racemose-paniculate patterns is expressed fully. Fig. 4. hoe nanthera colubrina var colubrina in fruit, from Brazil (Widgren 2844), showing the long and uniformly con- tracted pods. “— 12 SIRI VON REIS ALTSCHUL heads are distinctive in being always whitish in the bud, with the involucre, unlike that of the other variety, clearly visible below each head at that time. The heads are borne in panicu- late patterns in the branch apices and have fewer flowers. The legume characteristically is elongated, and regularly contracted, often darker than in the other variety. Anaden- anthera colubrina var. colubrina is the most easily recog- nizable element in the genus Anadenanthera, because of the distinctive arrangement of inflorescence. It is found in Ar- gentina and Brazil. Specimens examined show this variety to have a more or less east coastal distribution and its repre- sentatives to be separated, at least locally, from those of var. Cebil. Although previous authors have recognized infraspecific entities among the species here included in Anadenanther4, the names used in the present treatment appear nowhere to have been validated. In 1900, Malme suggested that Pipta- denia falcata should not be distinguished from P. peregr™. Four years later, Chodat & Hassler referred P. falcata to varietal rank under P. peregrina but included neither a des- cription nor reasons for the change. Similarly, Macbride suggested in 1943 that P. macrocarpa was very closely re lated to P. colubrina and that perhaps the former should be broadened to include the latter as a more southern-ranging variety. MORPHOLOGY : Habit. — The representatives of Anadenan- thera are feathery-foliaged, elegant-looking trees and shrubs. Anadenanthera peregrina var. peregrina ranges from a shrub to a tall tree, 3-27 m. high, with a trunk that usually is leaning and twisted. The contorted, irregular branches spread out above the middle of the trunk into an umbrella-like crown. Sometimes the trunk is divided at the base into several shafts. Anadenanthera peregrina var- fal- cata is reported to be small, a shrub or tree 4-8 Mm. high, shorter and thicker in aspect than the other individuals of the species. Anadenanthera colubrina var. Cebil ranges from a shrub to a tall tree, 3-30 m. high. It commonly is erect and without a divided trunk. Anadenanthera colubrina var. coll- brina is a tree 3-24 m. high. : Plate III, fig. 1, shows a stand of Anadenanthera peregrme var. peregrina in northern Puerto Rico, in February, 1960. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 13 Estatal de Camba- 4 ~~ « 2 es —S = Cebil cultivated in Orlando, Anadenanthera colubrina var. Anadenanthera peregrina var. [wo trees of ? A stand of Fig. CI — — _ lache, Puerto Rico. SIRI VON REIS ALT brina var. Cebil in Orlando, Florida. adenanther Fig 1 i] 5 Fi Co o ) uo) ~ g » 4 % 7 n = o 5 oS D ° faa} — = [ea] =] ont 8 3 bg g > i=} os 5 © a 38 < res 2 3 = Sy 3 3 ~ rn ° oo Pa t& = a on 7) 3 a bs = ret ms . 3 = 3 3 = ) < ps ev lache, Puerto Rico. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 15 The slender trees were said to have been planted and were about 9-13 m. high. Plate III, fig. 2, shows two somewhat storm- and cold- damaged trees of Anadenanthera colubrina var. Cebil in Orlando, Florida. They had been introduced as seedlings from Sao Paulo, Brazil, by Mr. and Mrs. Mulford B. Foster and had attained, by February, 1960, a height of about 13-15 m. Trunk and branches. — The trunks may be unarmed or armed, especially toward the base, with mammillose projec- tions. The bark is gray to black and becomes suberose in dry climates. In Anadenanthera peregrina var. peregrina the trunk has a diameter at breast height of 20-40 cm. When armed, the lower trunk produces conical to wedge-shaped projections, sometimes intensely when young. The bark is gray to nearly black with many small lenticels. Anadenan- thera peregrina var. falcata reportedly produces bark often more than 5 em. thick, and heavily suberose branches. Ana- denanthera colubrina usually has a larger trunk than the above species, 30-50 cm. in diameter at breast height. The bark is grayish and smoother than in A. peregrina, even though it sometimes is striated with longitudinal fissures. The young twigs vary from very dark to light gray, with whitish or reddish warts. Anadenanthera colubrina var. colubrina has light gray bark, which is more or less smooth. When this tree is armed, the fortifications are short, fat and thorny. Plate IV, fig. 1, shows the bark of Anadenanthera pere- grina var. peregrina. In contrast, the relative smoothness of the bark of A. colubrina var. Cebil in Florida is shown in Plate IV, fig. 2. Orlando, Florida, lies at 28 degrees N. lati- tude, which is about the equivalent of Catamarca, Argentina, the source of the southernmost specimens of A. colubrina var. Cebil examined. This latitude may roughly represent the climatological limit in both hemispheres for A. colubrina. Twigs and foliage. — These parts are more or less pubes- cent, or glaucous, rarely glabrous when young. They usually become glabrous with maturity. Individual specimens vary greatly. Anadenanthera peregrina var. falcata appears to be more densely puberulent when young, but more glabrous when mature, than var. peregrina. The twigs and foliage, 16 SIRI VON REIS ALTSCHUL de as in El Bosque Estatal PLATE VY. ig. 1. Anadenanthera peregrina var. peregrina in El] Bosque Cambalache, Puerto Rieo. era colubrina var. Cebil in Orlando, Florida. ; . lenan- q 2, Anadena Note the Bromeliad on the left trunk. Fig. 2 y Storm damage caused defoliation. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 17 at least when dried, are also somewhat darker in var. falcata. Anadenanthera colubrina exhibits a higher degree of varia- tion than does A. peregrina. While some specimens of A. colubrina var. Cebil are golden-red tomentulose (from Per- nambuco) or partly puberulent, others are almost entirely glabrous. Specimens of var. Cebil appear, on the whole, to be less glabrous than specimens of var. colubrina. Plate V, fig. 1, is a photograph looking up into the crowns of Anadenanthera peregrina var. peregrina. A view of the branches and leaves of A. colubrina var. Cebil is shown in Plate V, fig. 2. Stipules. — These organs, which are only a few milli- meters long, bristly and fugacious, are seen almost never on dried specimens. However, broad basal bracts in the leaf axils and enclosing the young inflorescence buds may persist. af. — The main and secondary axes of the bipinnately compound leaves (including the petioles) are often reddish when young. These parts are furrowed by a single longi- tudinal channel on the ventral surface. Dorsally, they are perfectly rounded and somewhat lighter in color and less puberulent than on the ventral side. The petioles are enlarged basally and are frequently a slightly different and darker color in the lower region. In Anadenanthera peregrina the leaf is 12-30 cm. long, and the petiole is darkened at the base. In A. colubrina the leaves tend to be shorter than they are in A. peregrina. They are 4-20 cm. long (petioles included in measurements of both species). The petiole base in this species is wrinkled transversely and may be greenish. Ana- denanthera colubrina var. colubrina has a very deeply channeled rachis. The pinna pairs vary in Anadenanthera from 7-35 or more per leaf, and each pinna may be 1.2-7 cm. long. The pinna pairs are opposite to subopposite, and the pinnae may extend, in A. colubrina, in a fine point 1.5-2 mm. beyond the ultimate pair of leaflets. In A. peregrina var. peregrina the pinna pairs are 10-30 or more per leaf, and each pinna is 2-5 cm. long. Anadenanthera peregrina var. falcata may be recog- nized by its fewer, 10-18, pinna pairs per leaf. Anadenan- thera colubrina var. Cebil has 7-35 or more pinna pairs per leaf, each pinna being 1.2-7 cm. long. Anadenanthera colu- brina var. colubrina bears 10-25 pinna pairs per leaf, and 18 SIRI VON REIS ALTSCHUL , Uy Oe ‘ I. Terminal foliage of a representative of Anadenanthera peregrina var- peregrina in El Bosque Estatal de Cambalache, Puerto Rico. each pinna is 3.5-7 cm. long, often with those toward the apex becoming shorter. These differences are relative and variable. For instance, a single specimen of A. colubrina var. Cebil was seen to exhibit a range of from 7 pinna pairs per leaf to 35 pinna pairs per leaf! Plate VI shows terminal clusters of the leaves of Anade- nanthera peregrina var. peregrina. The foliage is extended palmately at the tips of otherwise bare branches. Glands. — These sessile protuberances appear to be ele- vations of portions of the margins bordering the channel that furrows the petiole and the rest of the main leaf rachis. In dried specimens they may be dark brown or black, rufous or white. One is found commonly on the petiole and 1-7 similar, smaller ones between or just below each of the ulti- mate pinna pairs. They may be absent. In Anadenanthera peregrina the main gland is dark, flattish, oval or oblong; -5-5 mm. long and located 5-15 mm. above the petiole base. in A. colubrina var. Cebil it more often is lacking than 12 A. peregrina or A. colubrina var. colubrina. One specimen of var. Cebil examined had two glands on one petiole. The main gland in var. Cebil is 1-5 mm. long and is found A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 19 anywhere between the base of the petiole and the first pinna pair; it is flattened and, in living material, reddish. The smaller glands are 1-3 (rarely 7). In var. colubrina the main gland is 1-3 mm. long and found just above the petiole base; it is erect, 1-1.7 mm. high, with a deep pore, and is blackish in dried specimens. The smaller glands are 1-2 (rarely 6). The foliage glands are most uniform as to color, number, position, presence and size in A. peregrina var. peregrina and A. colubrina var. colubrina. In A. colubrina var. Cebil they may be inconsistent upon single specimens. Leaflets. — The leaflets of Anadenanthera are sessile, opposite to subopposite, entire, at the base oblique or trun- cate, with margins usually ciliate or ciliolate. A pinna may bear 20-80 pairs of leaflets, each .9-8 mm. long and .5-1.5 mm. wide. In A. peregrina the leaflets usually are imbricate and at the apex more or less acuminate. In A. colubrina they are less often imbricate and at the apex relatively obtuse. In A. peregrina var. peregrina the leaflets are 25-80 pairs, 2-8 mm. long, usually straight and membranaceous. Those of A. peregrina var. falcata characteristically are more numerous, starting at 40 pairs, and longer, starting at 4.5 mm.; they tend to be falcate, coriaceous and nitid, with greater dorsiventral contrast than in var. peregrina. Ana- denanthera colubrina var. Cebil bears 20-80 pairs of leaflets, each .9-6 mm. long and often dilated in the middle. The leaf- lets of var. colubrina are fewer and more standardized in number, 40-60 pairs per pinna; they also are longer and of more uniform lengths, 3-6 mm. These leaflets usually are glabrous except for the lightly ciliolate margins ending in an often acute apex. They are nitidulous but less so than in A. peregrina var. falcata. The leaflets of A. colubrina var. colubrina tend to be dorsiventrally distinct, opaque, and coriacecus in texture. In general, the leaflets of A. colubrina are shorter than those of A. peregrina and are more oblong in appearance. Anadenanthera peregrina var. peregrina and A. colubrina var. Cebil commonly have thinner, more membranaceous leaflets than do the other varieties of the two species. Dried specimens of Anadenanthera colubrina var. Cebil often are grayish; some are green. In A. peregrina var. peregrina the foliage of herbarium material is brown or 20 SIRI VON REIS ALTSCHUL Sian am eas tian Se miele Daaps gs 0, GIG a ‘ 4 PS a 2 PLATE VII. Cleared and stained leaflets of Anadenanthera spp. Fig. 1. Anadenanthera peregrina var. peregrina, showing slightly excentric mid-vein (R. Schomburgk # é : ig. 2. Anadenanthera peregrina var. faleata, showing excentri vein and finely reticulated end-veins (Sellow 196, Brasilia). Fig. 3. Anadenanthera colubrina var. Cebil, showing prominent secondary veins (Gardner 1584, Brazil). b] ° ee] ia] w e = = ~) green. Dried specimens of A. peregrina var. falcata and A. colubrina var. colubrina are darker, shinier, thicker and more glabrous in the leaflets than are those of the other varieties of the species. Living material of A. peregrina vat: peregrina and of A. colubrina var. Cebil is bright green, and the leaflets wilt in minutes after a twig is cut. The species of Anadenanthera are partly deciduous. Trees A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 21 of A. peregrina var. peregrina growing in Puerto Rico were in full leaf during the relatively dry spring of 1960. Anade- nanthera colubrina var. Cebil has almost persistent foliage which may be absent only at the end of the winter season. The extremities dry out in winter if frost occurs (Clos, 1929; Hieronymus, 1882). According to Hoehne, Kuhlmann & Handro (1941), the leaves of var. colubrina are lost in winter. A fungus, Ravenelia Cebil Spegazzini (1909), has been described as parasitic on the leaves of the taxon here called Anadenanthera colubrina var. Cebil. Heringer (1954b) published a brief but concise notice on the difference between the elements here called Anadenan- th-ra peregrina var. peregrina and A. colubrina var. Cebil. Ho rezognized the critical morphological differences between them and produced sketches of the peduncular involucre, the cnthers and the leaflets. He did not, however, comment on the differences in venation suggested by the drawings. Plates VII-IX confirm these indications and illustrate, in addition, the patterns manifested in those entities which Heringer recognized but did not designate and which are treated here as A. peregrina var. falcata and A. colubrina var. colubrina. The venation pattern basic to the species of Anadenan- thera is a single, distinct vein which runs in nearly a straight line from the base to the apex of the leaflet. In Plate VII, fig. 1, the mid-vein of A. peregrina var. peregrina appears to be surrounded by a bundle sheath and is slightly eccentric. Plate VII, fig. 2, shows a leaflet of A. peregrina var. falcata, in which the end-veins seem to be more finely reticulated than in the other variety. Plate VII, fig. 3, shows a leaflet of A. colubrina var. Cebil, in which the secondary veins are much more prominent than in the other species. The leaflets of var. Cebil are also more variable, not only in venation but in size and shape, as well. In Plate VIII, fig. 1, are seen the marginal hairs which ordinarily are present on the leaflets but which usually do not survive processing. Some previously recognized varieties and forms lack distinctiveness and do not retain nomenclatural status in the present paper: Plate VIII, fig. 2, shows a leaflet of the type of Piptadenia macro- carpa var. genuina f. puberula; Plate VIII, fig. 3, shows a 22 SIRI VON REIS ALTSCHUL Se SSS Des : Sere Gs *LATE VIII. Cleared and ysis leaflets of Anadenanthera colubrina var. ive Fig. 1. Sellow 1412, Brasilia. marginal hairs which usually do not surv processing. Fig. 2 5 Silaster 7840, ei Fig. 3. Hassler 7466, Paraguay A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 23 ‘ Fa i des | Anadenan- thera colubrina Cebil (Hassler 8348, Paraguay). niheves colubrina var. Cebil (Hassler 6688, Paraguay). Fig. 3. Anaden thors colubrina var. colubrina (W Piate IX. Cleared and stained leaflets of a prenene h spp. Fig. 1 il eiro), a ak less prominent secondary venation than in A. colubrina var. Ce 24 SIRI VON REIS ALTSCHUL leaflet of the type of P. macrocarpa var. Cebil f. microcarpa; Plate IX, fig. 1, shows a leaflet of the type of P. macrocarpa var. vestita. The leaflet of the type of P. Hassleriana var. fruticosa in Plate IX, fig. 2, demonstrates one extreme form of vegetative variation common to the element here desig- nated as Anadenanthera colubrina var. Cebil. In A. colubrina var. colubrina, Plate IX, fig. 3, the secondary veins are seen to be less prominent than in A. colubrina var. Cebil. The value that can be placed on venation patterns for diagnostic purposes may not be consistent within the Pip- tadenia complex. Leaflets of the other genera included in Brenan’s scheme in 1955 were not cleared for the present study, but observations with the dissecting microscope re- veal the venation patterns to be somewhat variable even within the genera he cites. However, an adequate sampling was not available, and further studies on the problem are desirable. Arrangement of inflorescences. — The inflorescences of Anadenanthera are heads. They are produced upon slender peduncles which are fasciculate, commonly in the axils of the compound leaves and subterminal. In A. peregrina the heads are almost always disposed as described and are 1-5 per fascicle. In A. colubrina var. Cebil the heads are 1-4 per fascicle and usually are disposed as above but also may be arranged in racemose patterns in the branch apices, with the subtending leaves more or less reduced. Such patterns are found in A. peregrina var. peregrina in rare instances. In A. colubrina var. colubrina the inflorescences are 2-4 (rarely 7) per fascicle and are arranged in paniculate pat- terns in the branch apices, with the subtending leaves usually absent. Occasionally, incompletely developed foliar organs are seen below the fascicles of the paniculately borne heads. Often, a few simple axillary and subterminal fascicles of heads are present also along the branches. Peduncle. — Anadenanthera peregrina produces peduncles which are 1.75-4 em. long and puberulous. In A. peregrine var. peregrina they are filiform. In A. peregrina var. faleata they are slightly thicker and may appear shorter than in the other variety because of the larger mature heads of var. falcata. Generally, in Anadenanthera the peduncles are 1-3 times as long as the heads. Anadenanthera colubrina pro- A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 25 duces peduncles 2-4 cm. long, either puberulous or glabrous, and a little thicker than filiform. In A. colubrina var. colu- brina the peduncles become increasingly shorter toward the apex of the paniculate arrangement of the inflorescences. In both species of Anadenanthera, the peduncles thicken after fertilization. Involucre. — Each peduncle bears around its upper portion two membranous bracts which are more or less united in an annular involucre about 1 mm. long. In Anadenanthera pere- grina the involucre is puberulous, bidentate and campanu- late. It is borne about three-quarters of the distance to the head and, in dried specimens, often is detached, encircling loosely the base of the peduncle and leaving a girdle-mark above. In A. colubrina the involucre may be glabrous or puberulous. It is borne just below the receptacle and does not become detached. The involucre of A. colubrina var. Cebil appears as a thin band or annulus which is not readily noticeable below the immature head. The growth of the flowers hides the involucre and may obstruct its further expansion. The involucre of A. colubrina var. colubrina is bidentate, like that of A. peregrina, but with the tips extend- ed beyond a campanulate condition to nearly perpendicular to the main axis; it is borne just below the receptacle, as is characteristic for A. colubrina. Inflorescence. — The small, sessile flowers of Anadenan- thera are borne in globose-capitate heads, each with about 35-50 flowers crowded upon a globose-oblong receptacle com- monly 5 mm. long. The mature heads usually are whitish or pale. The receptacle often is seen below the maturing 1-3 fruits as a pockmarked knob to which a few flowers may adhere. In A. peregrina var. peregrina the heads are green- ish white to white, 10 or more mm. in diameter, including stamens, and several times shorter than the peduncles. In A. peregrina var. falcata the heads are white to creamy yellow and tend to be somewhat larger, up to 18 mm. in diameter. Upon one specimen of var. falcata auxiliary flower buds were found within the involucre of the peduncle. The heads of A. colubrina range in color from white to yellowish or, rarely, orangey. In var. Cebil they may be 15-20 mm. in diameter, including stamens. The heads of var. colubrina are only about 15 mm. in diameter; they reportedly are 26 SIRI VON REIS ALTSCHUL fragrant or aromatic and tend to have fewer flowers than do the heads in var. Cebil or A. peregrina. The heads of var. colubrina are minutely whitish-tomentose in the bud. Bracteoles. — The individual flowers each are subtended by a linear-spathulate or deltoid bracteole commonly about half the length of the mature corolla and often equal in length to the variable calyx. The bracteole attains a maxi- mum length of 2 mm. It is membranous-hyaline, delicately 1-veined, translucent and colorless to pale rufous in dried specimens, where the tip usually is darkened and lightly puberulent dorsally. The bracteole resembles in color, form, texture, venation and vesture the parts which make up the calyx and corolla. In the head it is much compressed and, in dried specimens, easily broken. Calyx. — The small, sessile flowers of Anadenanthera have campanulate calyces which are 5-dentate and relatively vari- able in length for their size, being 0.5-3 mm. long. Differ- ences in calyx length, as well as in color, have been seen within single heads and do not seem to be correlated with any other characters of the flowers. The calyx may be one- quarter to three-quarters the length of the corolla. It is, like the bracteole, membranous-hyaline, delicately 1-veined, translucent, and nearly colorless to pale rufous at the tips, which usually are at least lightly puberulent dorsally. Within a given element, vesture varies less in the flower parts than it does in the foliage. In some form, it is present nearly always in the inflorescence, even though limited to a few, fine epidermal hairs along the central vein at the tip of the bracteole, calyx lobe or petal. Specimens of A. colubrind which are otherwise glabrous still may retain this floral vesture. Corolla. — The corolla consists of 5 free or lightly coher ent petals. Like the calyx, its members are membranous-hyaline, delicately 1-veined, more or less translucent, and at least puberulent at the tips. The petals are off-white to flesh tone to pale rufous and sometimes, in dried specimens, dark at the tips. In Anadenanthera peregrina the corolla is 2-3.5 mm. long. In A. colubrina the corolla is 2.4-4 mm. long and often ruddier than in the other species. In A. peregrina the heads are more crowded and the flowers narrower than 1D A. colubrina. Usually the petals of both species cohere at A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 27 level more or less equal to the calyx tips. If the calyx is short, the petals may cohere to a point above the calyx tips; if the calyx is longer, the petals may cchere to a point below the calyx tips. Androecium. — The stamens of Anadenanthera are 10 in number, 5-8 mm. long, glabrous and exserted, usually 2-3 times the corolla length. The filaments are free and filiform. The anthers are bilocular, elliptical and longitudinally dehis- cent. In some specimens examined, the pollen appeared to have been released while the flowers were still in the bud. The pollen grains have nondescript, non-porous surfaces and are arranged in polyads, as Erdtman (1952) -has indicated for many of the Mimosoideae, including some species of Pip- tadenia. According to Spegazzini (1923), in both Piptadenia (no species named) and Anadenanthera, the polyads are 12- grained, rarely 8-16, moderate-sized and semi-lenticular ; they are paired in globose masses which are found 2-4 in each locule of the anthers. In Piptadenia (no species named), the individual grains are arranged in no particular order within the polyads; but in Anadenanthera four are central and eight peripheral (Spegazzini, 1923). In Anadenanthera peregrina the anthers are eglandular in the bud; in A. colubrina they are glandular. The gland usually is stipitate-substipitate but may be sessile at one extreme or long-stipitate at the other. Glands may vary from sessile to stipitate upon a single individual. Each globose gland is attached at the extremity of the anther to the con- nective between the two chambers. Dried specimens suggest that the gland is soon caducous and frequently a brighter or darker shade of the anther color. Sometimes the gland is white. The anther commonly is paler than the filament. The everall color of the mature heads reflects that of the stamens, which darken with preservation and usually are off-white to golden red. Gynoecium. — The simple ovary is unilocular, sessile- subsessile, many-ovuled and glabrous. It usually ranges from flesh color to dark in dried specimens and narrows into an elongated style, which enlarges apically into a tubular stigma at the level of or slightly above the anthers. (Because the filaments curl, the stamens appear shorter than they actually are). In Anadenanthera peregrina the fertilized 28 SIRI VON REIS ALTSCHUL ovary occasionally is reddish. In A. colubrina it may be similar or brighter red. The very young fruits are soon gray, at least in herbarium specimens; and the base begins to lengthen into a stipe. Fruit. — The legume of Anadenanthera is more or less flattened, with a glabrous surface; it is unilocular, though often falsely septate. The legume is light to dark brown or reddish within and not pulpy. It dehisces along one suture only. Usually just one fruit ripens per head, rarely 2-3. In Anadenanthera peregrina the fruit is oblongish to elongated, regularly, irregularly, vaguely, or not at all con- tracted between the seeds. The margins are slightly thick- ened, and the base is attenuate to obtuse. The surface is scurfy_t6 verrucose, and dull; it is dark brown with rufous scales in dried specimens. In A. peregrina var. peregrina thé legume is 5-35 cm. long (including the stipe but not the peduncle) and 1-3 em, wide. It is more or less straight, with the apex mucronate to acuminate or, if the tip has broken off, rounded. In A. peregrina var. falcata the legume tends to be shorter than in the variety above, up to 22 cm. long, and with a narrower range of widths, 1.4-2.2 cm. It usually is faleate, with the apex mucronate or, if the tip has broken off, rounded. The margins of the pod of A. peregr ima var. faleata may be also somewhat thinner than in the other variety. Faleate pods have been seen in dried material of A. peregrina var. peregrina from Venezuela, and specimens of var. falcata sometimes have straight pods. In Anadenanthera colubrina the fruit is attenuate to obtuse or truncate at the base, and the apex mucronate to acuminate to cuspidate or, if the tip has broken off, rounded. The surface is smooth to reticulated, and nitid. In dried specimens the fruit is light, dark or reddish brown, OF dark gray. In A. colubrina var. Cebil the legume is 10-32 cm. long (including the stipe but not the peduncle) and 1-3 cm. wide, straight to sometimes falcate, more or less oblongish t0 elongated. It is sinuate, or irregularly contracted, sometimes regularly contracted. The margins tend to be strongly thick- ened but occasionally are not at all so. Immature living material is bright yellow-green. The fruit of A. colubrina var. Cebil often is more oblong in dimensions, as well as thicker, than that of A. peregrina. In A. colubrina var. colu- A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 29 » 40 peregrina in El I regrina var. lnadenanthera pe root. how ing € nia I red », Seedling, 30 SIRI VON REIS ALTSCHUL brina the legume is more distinctive, less variable, than in var. Cebil. It tends to be longer, starting at 15 cm., and narrower, up to 1.9 cm. Also the fruit in var. colubrina is striaght, very elongated, very regularly contracted, very flattened and thin, with margins thickened but narrower than in var. Cebil. At the base it is truncate, at the apex mucronate. The surface is finely reticulated, nitid and dark brown to dark gray in dried specimens. Plate X, fig. 1, shows the nearly mature fruit of Anaden- anthera peregrina var. peregrina in Puerto Rico, in early February. The Jegume is borne somewhat below the termi- nal leaves. Seeds. -— The seeds of Anadenanthera are 8-16, flat, and orbicviar in outline. These smooth seeds are best described as wingless, but each has a rim or sharp margin. The seed is attached to a non-persistent, filiform funicle and is exal- buminous. The embryo is erect, the plumule conspicuously developed and divided. The thin cotyledons overlap and encase the stalk. In Anadenanthera peregrina the seeds are very thin, orbicular to suborbicular in outline and a dark chestnut brown to black. They are shining and 10-20 mm. in diameter. The seeds of var. peregrina will germinate about 6 days after planting (Heringer, 1947). In Anadenanthera colubrina the seeds are thin and dark chestnut brown. They are very shiny. The seeds of var. Cebil may attain oblong dimensions and are about 12-20 mm. in diameter, slightly larger than in A. peregrina. Mul- ford B. Foster has indicated in conversation that the seeds of var. Cebil germinate within only 2-3 days of planting. The seeds of A. colubrina var. colubrina are 10-16, more numer- ous than in the other taxa of Anadenanthera. They are orbicular to suborbicular, in contrast to the sometimes oblong seeds of var. Cebil. The seeds of var. colubrina are 12-15 mm. in diameter, uniformly smaller in size than 18 common in the seeds of var. Cebil. WOOD ANATOMY: In 1954 (a, b) Heringer indicated that in A. peregrina var. peregrina the wood becomes reddish upon exposure to sunlight, the sapwood is little differentiated from the heartwood, and the annual growth rings are only slightly apparent. In A. colubrina var. Cebil the wood was A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 31 said to be reddish, the sapwood much differentiated from the heartwood, the annual growth rings clearly apparent. Herin- ger described the element here considered as A. colubrina var. colubrina, unlike A. colubrina var. Cebil, as having the heartwood and sapwood not distinct one from another and the annual growth rings not very apparent. In 1958, Brazier described the wood of Anadenanthera, in general, as being hard, heavy timber of fine texture, usually with an interlocked grain. The heartwood was said to weigh about 60 lbs. per cubic foot, and up to 70 lbs. in A. colubrina var. Cebil. Anadenanthera cobubrina var. colu- brina is distinguishable, in addition, by its fewer and larger vessels. (Neither Brazier nor Heringer mentionedthe taxon here called A. peregrina var. falcata). a The differences between Anadenanthera colubrina var. Cebil and var. colubrina seem less extreme when the amoutit of anatomical variation reported for var. Cebil is considered. The literature on the wood of var. Cebil is relatively abun- dant, because that variety is important economically as a source of timber, particularly in Argentina. It not only has received many common names but has been divided, in col- loquial usage, into several ‘kinds’ of trees. The descriptions of these trees are confusing. Discrepancies in reporting are found as to: presence or absence and color of the dark stripes (perhaps annual growth rings observed in tangential or radial section) in the wood; the thickness of the secondary walls of the fibers; the predominance of pore multiples or of solitary pores; the correlation of any of these tendencies one with another or with grosser characters such as extent of fortification at the base of the trunk, the color of the bark, the amount of splitting or warping likely to occur when the wood is cut and used commercially (Brazier, 1958; Cozzo, 1951; Herrera, 1921; Koehler, 1928; Lillo, 1917; Tortorelli, 1948). On the basis of field work and anatomical studies, Cozzo (1951) has concluded that, unusual as such variation may be below the generic level, it must be regarded as an expres- sion of the natural variability of A. colubrina var. Cebil. In the light of available data, this may be the only reasonable conclusion. Since no patterns of variation correlatable with By SIRI VON REIS ALTSCHUL particular geographical areas or localities have been dis- cerned, the characters involved in the variation exhibited in this taxon cannot be used to distinguish any further varieties as that category is employed here. In summary, Anadenanthera peregrina has been studied but little anatomically. It is presumed to be fairly uniform in its wood characters. Anadenanthera colubrina, especially A. colubrina var. Cebil, appears to be variable in this respect as it is in others. GEOGRAPHICAL DISTRIBUTION AND ECOLOGICAL PREFER- ENCES: Representatives of the genus Anadenanthera are distributed widely throughout the tropical and subtropical New World. Maps 1 & 2 indicate partial distributions of the species examined in the course of this study. Notes accompanying herbarium materials suggest ecologi- eal preferences of the species of Anadenanthera. In general, they may be described as occupants of savannas, as defined by Beard (1953). Although the species appear to be rather specific in requirements for drainage and light, they seem to be relatively adaptable with regard to altitude. Anadenan- thera peregrina grows from sea level up to at least 1200 m. in altitude. Anadenanthera colubrina will survive up to 2100 m., possibly higher. Anadenanthera peregrina has the greater north-south distributional range but may be slightly more selective of habitat than is the other species. On the island of Hispaniola, var. peregrina (Map 1) inhabits riversides, waste ground, plains, slopes, open ridges, chalky limestone, sandstone hills, roadsides and eruptive foothills, up to an altitude of 400 ™- In Puerto Rico, it occurs in mountainous woods, along rivers, in riverside savannas, along roadsides, on hillsides and on red clay slopes. Venezuelan trees (Map 2) are reported varl- ously as being forest dominants, belonging to secondary forests, inhabiting savannas, light forests and riversides, up to 750 m. alt. In British Guiana, A. peregrina var. pere grina is found in savannas and riverside forests. Brazilian representatives of A. peregrina are mostly from the campos (the Brazilian equivalent for the term savanna, according to Ducke & Black, 1953) ; they also are found along rivers in Paré. The South American collections examined of “- peregrina came from Colombia, Venezuela, British Guian@ A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 33 — 70 65 60 Y @ ANADENANTHERA PEREGRINA ~ £ vaR. PEREGRINA is ° i“) f, fi , Sh) i o, s 7 = en 10 o;” = Fs An { ‘f Pe 23 ? = oe Nas fe Bey ¢ ] eh j ine ’; oe P ~~ Y ——— ? p ss guage A ae a, j + - weed eee 3 , ‘ tie, nts 8 : > i ‘ is “ 2M 2 Fa ae aS a ' ? ¢ 3 fh = y \ Pe eared ? at , 6 Maree . ae N. a Map 1. Distribution of Anadenanthera peregrina var. peregrina in the West Indies, on flowering specimens. Each symbol represents at least one collection. Only on Hispaniola are individual localities indicated. 34 _ SIRI VON REIS ALTSCHUL @ ANADENANTHERA PEREGRINA var. PEREGRINA © ANADENANTHERA PEREGRINA VAR. FALCATA © ANADENANTHERA COLUBRINA var. CEBIL \ O ANADENANTHERA COLUBRINA VAR. COLUBRINA P 2. Distributions of the species of Anadenanthera in South America, pased 07 flowering specimens. Each symbol represents at least one collection. Where an element was represented only once for a state or country and without a specific locality, - symbol was placed in the center of that state or country. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 35 and Brazil (Amazonas, Goyaz and Para). The regions which these countries encompass are interspersed with savannas. North of the Orinoco River, savannas are found which occur on soils of Quaternary alluvia, a type of land which also is found in lowland Guiana. A broad belt of savanna country runs through Venezuelan Guiana; but, in most of British Guiana and the Amazon, savannas exist only as islands in the rain-forest (Beard, 1953). Anadenanthera peregrina var. falcata (Map 2) is found in Minas Gerais, Brazil, in cut-over woods and on slopes. Specimens of this variety also have been collected in Mato Grosso, Parana and Sao Paulo, Brazil, and in Paraguay. Areas of the same kind of land as is found supporting savan- nas in lowland Guiana and north of the Orinoco also are found in Paraguay (Beard, 1953). Anadenanthera colubrina (Map 2) is limited in distribu- tion to the southern hemisphere but appears to be adapted to more varied environments than is the other species. In Peru, A. colubrina var. Cebil occupies riversides, dry slopes and terraces, up to 2000 m. altitude. According to the litera- ture, A. colubrina var. Cebil appears as a dominant tree in the xerophytic savanna-like formations of riparian valleys and on the western slopes, as far north as the Marafion Val- ley, at altitudes of 800-2700 m. (Weberbauer, 1945; Wil- liams, 1945). Specimens were examined from Bolivia, where, according to Beard (1953), soils are to be found similar to those supporting savannas in Paraguay and lowland Guiana. Herbarium specimens from Paraguay indicate that A. colu- brina var. Cebil grows along lakes and rivers, in mountains and on high plateaus, in rock declivities, valleys, fields, on woody riversides, in border woods and in forests. Common habitats for A. colubrina var. Cebil in northern Argentina include riverbanks with rich soil, forests, skirthills, campos, and the interiors of riparian mountains, at altitudes up to 1250 m. Specimens were examined from Catamarca, Cor- rientes, Jujuy, Misiones, Salta and Tucuman, Argentina. Sources from the literature report this variety as occurring in Argentina at altitudes of 340-900 m. in savanna country on laterite or sandy clays, in areas of winter frost. Anaden- anthera colubrina var. Cebil is characteristic for the Region del Cebil, or Zona Chaquefia Occidental, in northwestern Ar- 36 SIRI VON REIS ALTSCHUL gentina in the foothills of the Andes. The vegetation in that area is determined by high humidity and a long period of summer rains (Correa, 1925; Parodi, 1945; Tortorelli, 1948). Brazilian specimens from Minas Gerais have been collected on fertile ground in pedreiras (quarries, stone pits) and in the matta (heavy thickets, woods) ; in Sao Paulo, A. colubrina var. Cebil has been found in river areas. Other Brazilian specimens are from Bahia, Ceara, Maranhao, Ma- to Grosso, Pernambuco, Piauhy and Rio de Janeiro. Accord- ing to Léfgren (1896), A. colubrina var. Cebil occupies, in Sao Paulo, transition zones between campo, or savanna, and more arid areas. The distribution of this variety appears to follow, in general, the patterns of geographic elevation asso- ciated with the Brazilian and Peruvian highlands. Anadenanthera colubrina var. colubrina (Map 2) is re stricted in its distribution, it seems, to Bahia, Parana, Rio de Janeiro and Sao Paulo, Brazil, and to Misiones, Argentina. It generally is reported to occur along river edges; but one collection (Glaziou 10592), made in Rio de Janeiro, is from the restinga. The last term, as it appears on the label, must refer to the old maritime beaches of white sand and black humus which are covered with shrubs and trees, in south- eastern Brazil. (The term restinga, as used here, is not to be confused with the restinga of the Brazilian Amazon; OF should it be confused with the Amazonian caatinga, which it resembles superficially but from which it differs importantly with regard to rainfall and flora, according to Ducke & Black, 1953). Specimens of var. colubrina have been © lected only up to 700 m. Information accompanying herbarl um specimens is brief, and the literature is sparse. Hence, our present understanding of var. colubrina from a distribu- tional and ecological viewpoint is limited. Luetzelberg (1922- 23) reported this taxon as a constituent of what he called the tall caatinga. (This is not to be confused with the Amazonr an caatinga above of Ducke & Black). Luetzelberg further indicated var. colubrina as occurring: in the xerophytie mattas uniting with this caatinga; in riverside areas from the northern part of Piauhy to northern Bahia; and in the mountains of extreme northeast Brazil. Souza (1945), 100% mentions its presence in the arboreal caatingas of Braz. However, the specimens examined from the areas indicated A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 37 by Luetzelberg and Souza nearly all are better included in A. colubrina var. Cebil. PHYLOGENY AND HISTORY OF DISTRIBUTION : Fossil evidence is lacking for the Piptadenia complex. However, it may be possible, on the basis of available information, to construct a picture of the probable origin and development of the genus Anadenanthera. This small, strictly New World genus may be a relatively young offshoot of the Piptadenia complex. Such a probability is supported primarily by the arrangement of the flowers in a head, a character which could be explained as a reduc- tion in the length of the axis of the spike found in all other members of the complex. A second argument may be pre- sented from Spegazzini’s report (1923) that the pollen grains of Anadenanthera, unlike those of the other members of the Piptadenia complex, are arranged in a particular order within the polyads. If Anadenanthera did not arise as an offshoot of one of the existing genera of the Piptadenia complex, it may have originated from a more primitive stock ancestral, perhaps, to other genera as well, in that complex. Such an hypotheti- cal stock could be thought of as having had the following characters: a spicate inflorescence, pollen grains un-ordered in the polyads, plus incompletely paratracheal parenchyma with high, multiseriate rays. The wood characters just men- tioned are typical for Anadenanthera but are relatively primitive for the Piptadenia complex as a whole. The evolutionary implications of the few distinguishing features of the two species of Anadenanthera are not readily understood. In morphological terms alone, the lack of anther glands in A. peregrina may be seen as a reduction, since those appendages are typical for members of the Piptadenia complex. On the other hand, the form and position of the involucre surrounding the peduncle appear to be more primi- tive in Anadenanthera peregrina than in A. colubrina. In var. Cebil of the last species the involucre apparently has been reduced from two partly fused, pointed bracts situated somewhat below the head to a thin band located just below the receptacle, as though the distance between the involucre and head had been telescoped, as well. The possible evolu- tionary significance of the different textures of the fruit of the two species is not easily recognized, as there is no obvious 38 SIRI VON REIS ALTSCHUL adaptation associated with either the scurfy-verrucose and dull condition in A. peregrina or the smooth-reticulated and shiny condition in A. colubrina. In some specimens of A. peregrina the pollen appears to have been released before the flowers opened, as stated earlier. If cleistogamy is sug- gested thereby as a mode of reproduction in this species, the likelihood for establishment of characters through random fixation also might be enhanced. The texture of the pod of A. peregrina may have been established by chance, rather than by direct selection; or it may be linked with another character or characters of survival value to the species. Since the pod textures are constant distinguishing characters for the two species of Anadenanthera, they must be as old as the species themselves and may be associated in their incep- tion with much earlier times. The geographical distributions of the two species of Ana- denanthera suggest that the center of origin for the genus may be Brazil, where both species appear to be well repre- sented today. This probability is enhanced by the fact that the preponderance of the American species of the Piptadena complex grows in Brazil. Brazil may be a center for the evolution of New World Piptadenias in general. If Brazil is the center of origin for the genus Anadenan- tera, A. peregrina is the species which has migrated the greater distance from that center. Within A. peregrina, var. peregrina is the more widespread and, by far, the more northern-ranging of the two component varieties. Anaden- anthera peregrina var. faleata appears to be limited to the southern half of Brazil and perhaps exclusively represents A. peregrina in that region today. The broad longitudinal distribution of Anadenanthera colubrina is predominantly that of var. Cebil. This variety radiates geographically outward in nearly all directions from the relatively limited area occupied by var. colubrina. The areas occupied by Anadenanthera peregrina var. fal- cata and A. colubrina var. colubrina are sympatric and are presumed to lie within the postulated center of origin for the species of Anadenanthera. This center would fall in the Brazilian highlands, the southern part of which has bee? available for plant occupation since the Mesozoic and seems to have been the center of origin for other plant groups (Tryon, 1944). The altitudinal requirements and tolerances A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 39 of the species of Anadenanthera today could reflect well such a history. The present distributions of the species of Anadenanthera can be explained in terms which concur with the presumed geological history of the Amazon Valley as discussed by Seibert (1947) in a study of the genus Hevea. Anadenan- thera peregrina var. peregrina may have achieved most of its present distribution at an early time, in association with the southern part of an ancient eastern mountain range whose flora extended into the Amazon Valley. This range appears to have extended from the Venezuela-Guiana land mass southward to connect with the ranges which today skirt the Amazon Valley in Goids and Mato Grosso. The Amazon River is believed to have drained to the Pacific at that time. The relatively recent uplift of the Andes probably blocked that drainage, causing a huge inland lake to form in the central Amazon basin. Higher areas probably were isolated. Anadenanthera peregrina var. peregrina is represented to- day in the Amazon Valley within stands of savanna amid the rain-forest. These savannas are scattered from the edge of the Amazon flood plain both north and south toward the respective highlands of the Venezuela-Guiana land mass, and southeast Brazil. These savannas also are linked floristically with those two areas (Beard, 1953), and their flora is muc older than that of the surrounding rain-forest (Ducke & Black, 1953). The representation of closely related taxa in both the Venezuela-Guiana land mass and in the highlands of southeast Brazil today is not limited to Anadenanthera. The primitive members of the euphorbiaceous genus Micran- dra are found in the same two areas (Richard E. Schultes in conversation). The characters by which Anadenanthera peregrina var. falcata is distinguished from var. peregrina may reflect climatic or ecological changes which could have occurred in southern Brazil subsequent to the northward distribution postulated for the original var. peregrina. The smaller habit, thickened vegetative parts and heavily suberose bark of representatives of var. falcata could be interpreted as drought adaptations. Whether this variety occurs, typically, in drier habitats than does var. peregrina is not known to me at this time. Anadenanthera colubrina probably also arose in the Bra- 40 SIRI VON REIS ALTSCHUL zilian highlands. The movement of Anadenanthera colubrina var. Cebil westward into Paraguay, the mountains of Argen- tina, Bolivia and Peru most likely would have taken place subsequent to the northward migration of A. peregrina var. peregrina and after the establishment of the Orinoco or Amazon drainage of the Amazon basin lake. Prior to this drainage, it is believed, the lake drained southward through the Parana-Paraguay Rivers system. Such a southern outlet could have prevented for some time the westward dispersion of A. colubrina var. Cebil. Neither species of Anadenanthera is represented, to my knowledge, in western Amazonas nor in the Andes of north- ern Peru, Ecuador and Colombia. This circumstance is prob- ably due to an inability of the representatives of A. peregrina var. peregrina to tolerate the true rain-forest conditions through which they would have to pass in a westward dis- persal. There is equally little reason to believe that repre- sentatives of A. colubrina var. Cebil would descend from the Andes into lowland rain-forest. The absence of this taxon farther north in the Andes than north-central Peru may be due to the relative recentness of its dispersal. The greater variation exhibited by A. colubrina than by A. peregria could be related to a more recent distribution for the first species, as well as, perhaps, to its being the younger species. The absence of specimens of either species from centra Brazil may be due to lack of collections. Anadenanthera colubrina var. colubrina appears to be limited in its distribution to an east coastal pattern. Its representatives are found within a relatively small ge graphic range, seem to have a maritime preference and apparently constitute the single entity from the resting Anadenanthera colubrina var. colubrina is characterized by the most highly evolved arrangement of inflorescences in the genus. The tendency for clusters of axillary, subterminal heads to become disposed in racemose-paniculate patterns in the branch apices is expressed fully only in this variety. A reported floral fragrance also is peculiar, within Anaden- anthera, to var. colubrina. This fragrance and the relativé showiness of the inflorescences, because of the manner 10 which they are borne, may be related to the unusually larg® size of the petiole glands. Mulford B. Foster indicated in conversation (1960) that the petiole glands, at least 1m A. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 41 colubrina var. Cebil, nocturnally exude a substance attrac- tive to insects and possibly related to pollination. Repre- sentatives of var. colubrina are specialized further in possessing fewer and larger vessels than are found in A. colu- brina var. Cebil. Anadenanthera colubrina var. colubrina is also the most regular element in the genus with regard to measurable parts. These data suggest that var. colubrina probably arose from var. Cebil or an ancestral form of that variety. It may be that this variety has undergone relatively rapid evolution in some features coincident with certain eco- logical adaptations not wholly decipherable from either the specimens examined or the literature. Further study of A. colubrina var. colubrina should prove rewarding. The species of Anadenanthera seem to have been adapted to savanna conditions and to various altitudes for an extend- ed period. These trees, or shrubs, are semi-deciduous and apparently tolerant of long dry periods interspersed with only sporadic rains, which initiate the production of new foliage. The flat, rounded, light-weight seeds must be dis- persed easily by floods, and even by winds. Their quick ger- mination may be an adaptation to sudden inundations. Plate X, fig. 2, shows a subsoil enlargement of the basal portion of a young tree of A. peregrina var. peregrina, found in Puerto Rico. Such an enlargement could function in the storage of materials necessary for the tree during the period critical to establishment. The species of Anadenanthera do not appear to hybridize in nature. Their flowering periods overlap, and their dis- tributions are sympatric. Very little is known about the genetics of Anadenanthera. The chromosomes of A. colub- rina var. Cebil are small and similar one to another. In this taxon the haploid number is 13, the commonest base number for the Mimosoideae (Atchison, 1951; Darlington & Wylie, 1956). The chromosome number of Anadenanthera pere- grina var. peregrina still has not been determined (Berger, Witkus & McMahon, 1958). The seeds of Anadenanthera peregrina are known as the sources of certain Indian preparations, particularly snuffs and beverages, for magical, medical, religious and stimula- tive purposes. The seeds of A. colubrina also seem to have been used for those purposes. However, it appears unlikely that man may have played an important role in effecting the 42, SIRI VON REIS ALTSCHUL present distribution of these species. The two species of Anadenanthera are similar in appearance, and chemical analyses of their fruits have shown them to be nearly identi- cal with regard to the psychotomimetic compounds, bufote- nine and its derivatives (M. S. Fish in a letter, January % 1958; Fish, Johnsen & Horning, 1955; Pachter, Zacharias & Ribeiro, 1959; Stromberg, 1954). Therefore, A. peregrina and A. colubrina could have been used interchangeably where used at all. Furthermore, these species are distributed wide- ly enough to be readily available to man throughout many parts of South America without cultivation. The fact that the genus consists of trees would discourage planting. The peoples who use Anadenanthera seeds in northern South America today are nomadic and seek the trees in savannas for seasonal harvesting. The West Indies is the one region where man more reason ably accounts for the distribution of Anadenanthera than does distribution by natural means. There, A. peregrina var. peregrina is found today as a somewhat weedy tree, inhabit- ing roadsides and wastelands. Its representatives are T& markably uniform throughout the islands which suggests a relatively recent introduction. To the Island Arawak, who inhabited most of the islands where this taxon now is found, var. peregrina seems to have been of ritual importance. These Indians may have found it easier to plant the trees than to maintain communication with the mainland for their source of supply. The presence of war-like Carib on the smaller islands between them and South America would have added to the inconvenience. It is interesting to note that Safford (1916) believed that Linnaeus’ original description of A. peregrina was based on a seedling grown in Europe from the New World. The seed most likely came from the West Indies or northern South America and may have been obtained through trade. Anadenanthera peregrina var. peregrina is represented, among the specimens examined, from the islands of His- paniola, Puerto Rico, Grenada, Tobago and Trinidad. Trin dad and Tobago are equitable floristically with the mainland (Beard, 1946). However, A. peregrina var. pereg’ ina sh f may have been introduced into these islands. On @ sheet 0 Broadway 9258 from Trinidad it states that the species sy planted and only “perhaps” a native tree. Beard (1946) di A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 43 not include A. peregrina in a list of trees native to Trinidad. The only specimen of A. peregrina examined from the true Lesser Antilles, which have a distinctive flora, was from Grenada: Beard 164, labelled “In scrub on wasteland ; appar- ently not native but naturalized.” Collections simply could be lacking from the Lesser Antilles, but it seems more rea- sonable to suppose that A. peregrina is not found through most of these islands because the larger settlements of the Island Arawak existed predominantly in the Greater Antil- les, with the exception of a sizeable community on Trinidad. For the most part, the Lesser Antilles were occupied by the Carib, for whom there are no indications of the use of Ana- denanthera materials. In 1951, Cruxent presented surpris- ing archaeological evidence suggesting that the Carib were relatively late-arriving, aggressive, colonizing groups from Central America. If this interpretation is correct, the ab- sence of culture traits associated with Anadenanthera uses among the Carib can be explained by the fact that the genus is not represented and evidently was not known in Central America, or even in western Colombia. Finally, there would be no obvious reason why A. peregrina should not occur in Cuba if it were distributed naturally in the Greater Antilles, for the Cuban flora is predominantly South American in character (Seifriz, 1943). Rather, it is significant that the culture of the Island Arawak is not known to have extended westward beyond Hispaniola. SYSTEMATIC TREATMENT Anadenanthera Spegazzini, Physis 6: 313. qu LECTOTYPE: Anadenanthera peregrina (L.) Spe Piptadenia Benth. Sect. Niopa a Hook. Jour. /. 340, 1841. Lectotype: P. peregrina (L.) Ben Niopa (Benth.) Britton & Rose, stone 12: 37. 1927. Feathery-foliaged, often elegant-looking trees and shrubs of South America and the West Indies. Unarmed or the trunks armed toward the base with mammillose projections, the bark gray to black, becom- ing suberose in drier climates. Young twigs and foliage more or less pubescent, or glaucous, rarely glabrous, but becoming so w rith maturity. Stipules small, bristly and fugacious; but the broad basal bracts enclos- ing new shoots often persistent. Leaves alternate, bipinnately com- pound; main rachis and secondary rachises of young leaves often reddish and more or less channeled ventrally. Petioles usually enlarged basally and each commonly bearing, somewhere below the first pinna pair, a sessile, dark brown or biack, rufous or white gland; 1-7 similar, 44 SIRI VON REIS ALTSCHUL smaller glands often borne along the main rachis, one between or just below each of the ultimate pinna pairs. Pinna pairs often many, 7-35 or more, each pinna 1.2-7 cm. long, opposite or subopposite. Leaflets sessile, opposite or subopposite, imbricate or expanded, not always borne to the very tip of the pinna, which may end in a fine point up to 2 mm. beyond the ultimate pair of leaflets. Leaflets many, 20-80 pairs per pinna, .9-8 mm. long, .5-1.5 . wide, entire, linear to dilated in the middle, oblong to lanceolate, straight to faleate, at the base oblique or truncate, at the apex acute cuminate, apiculate or obtuse; margins usually ciliate or ciliolate; venation obscure, with one mid- vein and more or less prominent secondary reticulated veins; mem- branaceous to coriaceous and nitid or nitidulous, sometimes .differing in texture and color dorsiventrally. Inflorescence globose-capitate, the heads 10-20 mm, in diameter, including stamens, greenish white to yellow or orangey, in fascicles of 1-7, minutely whitish-tomentose to glabrous in the bud; the heads axillary to the leaves and subterminal, or becoming arranged in racemose-paniculate patterns and terminal in the branch apices with the leaves reduced. Peduncles 1.75-4 em. long, puberulous to glabrous, filiform or thicker, 1-3 times as long as the heads and each bearing a membranous, puberulous to glabrous annular involucre about 1 mm. long more than halfway up the axis or else directly under the receptacle and sometimes hidden by the mature head. Flowers about 35-50 per head, small, sessile, reportedly sometimes aromatic, crowded upon the globose-oblong receptacle about 5 mm. long and later visible as a pock-marked knob below 1-3 maturing legumes. Flowers each subtended by a linear-spathulate or deltoid bracteole half the length of the mature corolla. Calyx campanulate, .5-3 mm, long; 5-dentate, about one-quarter to three-quarters the corolla length. Corolla tubular-campanulate, 5-parted, 2-4 mm. long; petals lightly coherent below the top of the calyx. Bracteoles, calyces and corollas membranous-hyaline, delicately 1-veined, translucent, and colorless to pale rufous at the tips in dried specimens, where often at least lightly puberulent dorsally. Stamens 10, 5-8 mm. long, glabrous, exserted, 2-3 times the corolla length, whitish to reddish in dried specimens; filaments free, filiform ; anthers bilocular, elliptical and longitudinally dehiscent, each with or without a sessile to stipitate gland. Ovary sessile-sub- sessile, many-ovuled, glabrous, flesh color to darkish in dried specimens, Pipemermoas and nitid, or seurfy to verrucose, and dull; light brown, ark brown (sometimes with rufous scales), reddish brown or dar gray outside, light to dark brown or reddish within; dehiscing along A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 45 one suture only, often falsely septate between the 8-16 thin, flat, orbieu- lar to oblong, dark chestnut brown to black, shiny, exalbuminous seeds which are 10-20 mm. in diameter, wingless but with a rim or sharp margin and attached to a non-persistent, filiform funicle. KEY A. Anthers eglandular in the bud; involucre or scar - of the way up the peduncle; legume scurfy to verrucose, and (4. ve peregrina) B, B. Leaflets membranaceous, dull, straight; legume straight (north- ern Brazil, British Guiana, Colombia, Venezuela, West Indies). la. A. peregrina var. peregrina. B. Leaflets SR SPON nitid, faleate; legume falcate (southern Bra ay 1 . peregrina var. falcata A. Anthers elandular in 1 the bud; involucre just below the sa ar legume smooth to reticulated, and nitid. .......... (2. A. colubrina) C C. Leaflets aed with the mid-vein more prominent than secondary venation; inflorescences borne in paniculate patterns in the branch apices; heads whitish in the bud; involucre with tips extended and noticeable below the immature head; legume very elongated, regularly contracted (Argentina, Braz il). Sibdiilcidsbiovees 2a. A. colubrina var. colubrina. C. Leaflets dilated in the middle, with prominent secondary venation; inflorescences fasciculate in the leaf axils and subterminal, or borne in racemose patterns in the branch apices; heads not whit- ish in the bud; involucre a narrow band not noticeable; legume relatively short and wide, often irregularly contracted (Argen- tina, Bolivia, Brazil, Paraguay, Peru). 2b. A. colubrina var. Cebil. 1. Anadenanthera peregrina (L.) Speg. Shrub to tall tree, 3-27 m. high. Trunk 20-40 cm. in diameter at breast height, usually leaning, twisted, sometimes divided at the base into several shafts; more frequently the contorted, irregular branches spread out above the middle of a solitary trunk into an umbrella-like crown. Bark gray to nearly black with many small lenticels; unarmed or lower trunk producing conical thorns or wedge-shaped projections, sometimes intensely so when young, becoming tubercular-verrucose, corky, rugose and, in drier climates, very thick. Young twigs and foliage puberulent, occasionally glaucescent; mature foliage glabrous or nearly so. Leaves, including petioles, 12-30 cm. long, the main rachis more or less channeled ventrally. Petioles somewhat darkened at their bases, 5-15 mm. above which each bears a flattish, oval or oblong gland 5-5 mm. long; 1-4 similar, smaller glands borne one between or just below nes of the ultimate pinna pairs. Pinna pairs 10-30 or more, each p 5 em. or more long, opposite or subopposite. Leaflets sliuatlet ‘aborene 25-80 pairs, 2-8 mm. long, .5-1.5 mm. wide, linear, oblong or lanceolate, straight to faleate, at the base oblique or truncate, at the apex acute to acuminate to apiculate; venation obscure except for a single, nearly straight, slightly excentric mid-vein; membrana- ceous to coriaceous and nitid, sometimes differing in color and texture dorsiventrally. Heads 10-18 mm. in diameter, including stamens, green- 46 SIRI VON REIS ALTSCHUL ish white to creamy yellow, in fascicles of 1-5, puberulous to glabrous in the bud; the heads axillary to the leaves rot subterminal, rarely becoming arranged in racemose patterns in the branch apices. Pe- duncles 1,75-4 em. long, puberulous, filiform or thicker, each bearing about three-quarters of the way up the axis a puberulous bi-dentate, campanulate involucre which becomes detached and slides down to encircle loosely the base of the peduncle. Calyx .5-2.6 mm. long. Corolla 2-3.5 mm, long. Stamens 5-8 mm. long; anthers eglandular in the bud. Legume 5-35 cm. long (including the stipe but not the pe- duncle), 1-3 cm. wide, straight to falcate, oblongish to elongated, regu- larly to irregularly, vaguely or not at all contracted between the seeds, more or less flattened, with margins slightly thickened; at the base attenuate to obtuse, at the apex mucronate to acuminate to cuspidate or, if the tip has broken off, rounded; surface scurfy to verrucose, and dull; dark brown with rufous scales in dried specimens. Seeds 8-16, very thin, flat, orbicular to suborbicular, dark chestnut brown to black, shiny, 10-20 mm. in diameter This species is found in the West Indies and South America, from 20 degrees N. latitude to 26 degrees S. latitude. la. Anadenanthera peregrina (L.) Speg. var. peregrina Mimosa peregrina L., Sp. Pl. 520. 1753. Acacia peregrina (L.) Willd., Sp. Pl. 4: 1073. 1806. Piptadenia peregrina (L.) Benth: Hook. Jour. Bot. 4: 340. 1841. Niopa peregrina (L.) Britton & Rose, Addisonia 12: 37. 1927. Anadenanthera peregrina (L.) Speg., Physis 6: 313. 1923. Acacia microphylla Willd., Sp. Pl. 4: 1073. 1806. Inga Niopo Humb. & Bonpl. ex Willd., Sp. Pl. 4: 1027. 1806. Mimosa Niopo (Humb. & Bonpl. ex Willd.) Poir., Dict. Encyc. Suppl. 1: 48. 1810. pov tee psy ire & Bonpl. ex. Willd.) Humb., Rélat. Hist. 2: 620- 3. 1814- Piptadenia naib (Humb. & Bonpl. ex Willd.) Spruce, Notes Bot. Amaz. 2: 426. 1908. Mimosa parvifolia Poir., Dict. Encye. Suppl. 1: 74. 1810. Acacia angustiloba DC. esd 2: 470. 1825. Mimosa? acacioides Benth., Hook. Jour. Bot. 2: 132. 1840. : Mimosa peregrina L., a specimen from the Clifford Herbarium which is si locatable at present (W. T. Stearn in a letter, December 6, 1957). The type specimen was obtained from a seedling growing in the poeta garden in Holland (Safford, 1916). Mimosa peregrina, the earliest valid name by which Ana- denanthera peregrina has been known, appeared in Species Plantarum. It was based on a fuller diagnosis in Hortus Cliffortianus. Although not complete, the diagnosis fits A. peregrina var. peregrina in all particulars except one, an incidental comparison to what now is known as Mimosa cormgera L. A photograph of the last species from the Lin- naean Herbarium shows it to be represented by only a few A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 47 bipinnately compound leaves subtended by large, thorny stipules. The description in Hortus Cliffortianus and Species Plantarum suggests that Linnaeus’ familiarity with M. cor- nigera was founded exclusively upon that specimen. Consid- ering this and the gross similarity of foliage characters among the Mimosoideae, Linnaeus’ comment is more easily understood. The type specimen of Mimosa peregrina, or the seed from which the plant supplying the type specimen was grown, most likely came from the West Indies or, possibly, northern South America. Anadenanthera peregrina var. peregrina is thought to have been introduced into the Antilles from the mainland. Linnaeus’ description does not indicate why he applied the epithet peregrina. He may have been implying that the species was found from island to island or that it was not native to the West Indies. It is apparent from photographs of Acacia peregrina (L.) Willd. obtained from Willdenow’s herbarium (No. 19156) in Berlin that Willdenow interpreted Linnaeus’ description as referring to the element here designated as the type-includ- ing variety of Anadenanthera peregrina. The flowers are not visible in detail, but the characteristic annular involucre can be seen loosely encircling the peduncle after having be- come detached. Willdenow’s work is the earliest interpreta- tion of Linnaeus’ Mimosa peregrina as the element described here. Unless cogent evidence can be presented to the con- trary, the traditional application of this name ought, I feel, to continue to be accepted. SYNONYMY. Acacia microphylla Willdenow (1806) is included in the synonymy of Anadenanthera peregrina var. peregrina on the basis of the original description and a photograph of the type from the Will- denow Herbarium (No. 19155). Willdenow’s only citation was “Habi- tat ad Caracas .. . Bredemeyer.” Again, the flowers cannot be seen in detail, but the characteristic involucres are present. The spec is Bredemeyer s. n., from Caracas. Bredemeyer’s South American and West Indian collections are from Venezuela, Curacao and Puerto Rico, all areas where only the more northern-ranging Anadenanthera pere- wr The 1808 as Staal eng ree Plantarum contained the first scientific description of Inga Niopo. Willdenow attributed the name to Humboldt & Bonpland, and it evidently was based on material from the collection of those explorers, who had not yet published their famous account. Willdenow gave no collection number but mentioned the locality as being near Atures, which is along the Orinoco River, A photograph of the type from the Willdenow Herbarium (No. 19057) 48 SIRI VON REIS ALTSCHUL shows this plant, also, to belong to Anadenanthera peregrina var. pere- grina. The specimen is fruiting and without flowers, so that it is not unreasonable for Willdenow to have thought it to be distinctive. Never- theless, the foliage and fruits appear to be typical of var. peregrina, and the specimen comes from Atures. The epithet Niopo suggests that Humboldt & Bonpland were familiar with one of the native names attributed to var. peregrina in northern South America. It is referred to, in Willdenow, as having been used in place of tobacco. In 1810, Poiret placed Inga Niopo in the genus Mimosa but added nothing new in the way of interpretation. Four years later, in the first publication of Personal Narrative . . ., Humboldt mentioned, p. 405 (Vol. 2, 1900 ed.) “. . . the long pods of a mimosacea, which we have made known by the name of Acacia niopo, . . .” This was footnoted by the follow- ing: “It is an Acacia with very delicate leaves, and not an Inga”. The author cited no material and must have been suggesting a correction of Willdenow’s original publication. In 1824, the name appeared as Acacia Niopo (Humb.) H. B. & K., with an enlarged description but the flowers still unknown. Bentham (1841-42) subsequently placed this species in the synonymy of Piptadenia peregrina, doing so on the con- viction that, although he had not seen Humboldt’s material, its employ- ment by the natives, as reported by Humboldt and by Schomburgk (in Bentham, 1840, p. 132), was adequate to identify it with P. peregrina. Piptadenia Niopo (Humboldt) Spruce was published in 1908. The author said that his specimens agreed so well in leaves, flowers and fruits with Kunth’s description of Humboldt’s Acacia Niopo that they undoubtedly kelonged to the same species. But, as already stated, the flowers were unknown to Kunth and Humboldt and therefore not included in their diagnoses. Also, Spruce’s description gives the anthers as glandular. An examination of one of the two numbers cited by Spruce, 1786 (BM), showed it to be a typical specimen of Anadenan- thera peregrina var. peregrina in both flowers and fruits. This collec- tion was made near the mouth of the Amazon River, at Santarém. Possibly, Spruce assumed that the anthers were glandular on the basis of Bentham’s earliest formal description of Piptadenia (1841-42), which leads one to believe that all members of the genus have glandu- lar anthers. Even Bentham’s description of P. peregrina at that time did not indicate the eglandular condition of this species. Mimosa parvifolia Poiret (1810) was based on Acacia microphylla Willdenow and represents a change of name plus a translation into French of the earlier author’s Latin diagnosis. The original description of Acacia angustiloba DeCandolle (1825) was not very extensive, but, nonetheless, it was stated that the anthers were eglandular and that the tree was from the Caribbean islands. DeCandolle suggested, with a question-mark, that Acacia angustifoléa Lamarck might be a synonym of A. angustiloba. Acacia angustifolia does not belong to Anadenanthera peregrina, as is shown later in this paper. The type specimen of Acacia angustiloba therefore must be the uncited material upon which DeCandolle’s description was based. A photograph of the type specimen, labelled Acacia angustiloba DC, on loan from the Delessert Herbarium (F), provides the means for identi- A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 49 fying this species as belonging to Anadenanthera peregrina var. pere- eculiar, loose annular involucres surround the peduncles. Although the losality is not given, the common name on the label, Bois de tendre a Caillou, suggests that the collection may have been made on nde ns mosa? acacioides Bentham (1840) was published before Bentham had onelie the genus Piptadenia but after he had begun to doubt the correctness of the then extant classification of some of the Mimosas and Piptadenia peregrina. The specimens he cited in the original descrip- tion of Mimosa? acacioides agree with Anadenanthera peregrina var. peregrina. The original material was Robert Schomburgk 852, 866. I have examined the last number (F, BM, L and photographs from A, 8, NY), from woods, skirting savannas in British Guiana. Robert Schom- burgk 852 (K) belongs to A. peregrina var. peregrina and was collect- ed along the Rio Branco. A sheet examined from Paris had two labels, one of which was R. Schomburgk 852, 856; the other label did not indicate the collector and was merely numbered ‘ 33”. The material did not belong to Anadenanthera. Apparently, the collections had gotten mixed. m Shab to tree, up to 27 m. high. Trunk with bark less than 5 cm. thick; branches not heavily suberose. Pinna pairs up to 30 or more. Leaflets 25 or more pairs, 2 or more mm. long, usually straight, mem- branaceous. Heads 10 or more mm. in diameter, including stamens; greenish white to white. Peduncles filiform. Legumes up to 35 cm. long, up to 3 em. wide, usually straight. DISTRIBUTION. Herbarium specimens indicate that Anadenanthera peregrina var. peregrina occurs in Brazil (Amazonas, Goyaz, Para), British Guiana, Colombia and Venezuela. It probably is naturalized as found in the West Indies (Grenada, Hispaniola, Puerto Rico, Tobago and Trinidad). It has been cultivated on the islands and is reported not to be rare. In Puerto Rico conditions brought about by man no doubt have reduced its frequency. A small stand of cultivated trees exists in the Bosque Estatal de Cambalache in sunny, light mesophytic forest on flattish terrain in an area of well drained sandy, calcareous soil along the central northern coast at nearly sea level. According to ed to 70-1200 m. altitude as an occasional to frequent tree. It is abundant in savanna margins, characteristic of the edge of the selva de galeria, and at times is given as a forest dominant. It is also a well known element in the campos, on open ground or in cut-over woods or second- ary, light forest, in dry savannas and along rivers. In general, A. peregrina var. peregrina prefers semi-arid habitats. REP! NTATIVE SPECIMENS. Brazil. Amazonas: R. L. Frées 23075 (U); Krukof 6046 (A, BM, F, MO, NY, S, U, US). Goyaz: Glaziou 21034 50 SIRI VON REIS ALTSCHUL (c, 8). Para: Capucho 413 (F); Dahlgren & Sella 21 (F); Ducke 9884, 16519 (BM), 20186 (U); de Oliveira 1949 (NY). British Guiana: A. C. Smith 3217 (A, F, MO, NY, S, U, US). Venezuela: Aristeguieta 1612 (us); Pittier 8380 (F, GH, US), 11755 (A, NY, US), 13954 (F, NY, US), 14280 (Us); L. Williams 9973 (F, US), 12651 (F,S, US). WEST INDIES. Grenada: Broadway 1750 (F, GH, NY). Dominican Republic: Ekman 12624 (Ss, US), 14278 (NY, S, US); Fuertes 1561 (A, F, GH, NY, U, US); Valeur 423 (NY). Haiti: Ekman 3494 (A, 8, US); Holdridge 1069 (MO, NY, US). Tobago: Broadway 3809 (GH, L, US). Anadenanthera peregrina var. peregrina may be distinguished from A. peregrina var. faleata by the following characters: it is a taller tree with trunks and branches less corky. The pinna pairs tend to be more numerous but the leaflets fewer, shorter and straighter than in the other variety. The leaflets and olla vegetative parts are gen- erally thinner than in A. peregrina var. falcata. The heads are greener and may be smaller. The legume tends to be longer with a wider range of widths than in the other variety; it is usually straight; the apex is variable. Specimens examined show this more widespread and more northern-ranging variety to be found today from 20 degrees N. latitude to about 15 degrees S. latitude. 1b. Anadenanthera peregrina (L.) Speg. var. faleata (Benth.) comb. nov. Based on Piptadenia falcata Benth., Hook. Jour. Bot. 4: 341. 1841. Piptadenia peregrina (L.) var. faleata (Benth.) Chod. & Hass., Bull. Herb. Boiss. II. 4: 561. 1904. Anadenanthera falcata (Benth.) Speg., Physis 6: 313. 1928. HOLOTYPE: Sellow 196, Brasilia (K, with photographs of the type at A, NY). The material for the original description was cited by Ben- tham as “‘ — Brazil, Sello.” Of four sheets examined from Kew belong- ing to this variety and collected by Sellow, the one cited is the only one numbered. Shrub to tree, up to 8 m. high. Trunk with bark more than 5 cm thick; branches heavily suberose. Vegetative organs often densely pubernlent when young. Pinna pairs up to 18. Leaflets 40 or more pairs, 4.5 or more mm. long, usually faleate, coriaceous, nitid on the dorsal surface. Heads up to 18 mm. in diame ter, including shad fi white to creamy yellow. Peduncles thicker than filiform. Legume U to 22 em. long, 1.4-2.2 cm. wide, usually faleate; at the apex elie or, if the tip has broken off, rounded. DISTRIBUTION. Specimens come from Brazil (Matto Grosso, Minas Geraes, Parana, SAio Paulo) and Paraguay. This variety is common in the campo cerrado; in rocky fields; on high plains, in valleys; and along rivers. REPRESENTATIVE SPECIMENS. Brazil. Minas Geraes: Regnell III 517 (C, 8, US). Sao Paulo: de Andrade 26646 (A, NY, US); Regnell III 517 ra (s), Il] 717x (s). Paraguay: Hassler 4506, 5292 (A, BM, NY), 10516 (A, C, MO, NY, S, US), 10516a (A, BM). Also representative, though lacking a number, is a sheet labelled “Herb. Reg. Berolinense. Pipta- denia falcata Benth. Brasilia. Sellow legit.” (K). A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 51 Anadenanthera peregrina var. falcata may be distin- guished from A. peregrina var. peregrina by the following characters: it is a much shorter tree with suberose trunks and branches, thicker leaves and peduncles, resembling Stryphnodendron on herbarium sheets. Although often densely puberulent when young, the vegetative organs of var. falcata become more glabrous with maturity than do those of the other variety. The pinna pairs tend to be fewer but the leaflets more numerous, longer and more often fal- cate than in var. peregrina. The leaflets are generally cori- aceous, contrast dorsiventrally and are darker in dried specimens than those of the other variety. The heads are yellowish and may be larger. The legume tends to be shorter with a narrower range of widths than in var. peregrina; it usually is faleate; the apex is mucronate or, if the tip has broken off, rounded. Specimens examined show this less widespread and more southern-ranging variety to be found today from about 15 degrees S. latitude to 25 degrees S. latitude. 2. Anadenanthera colubrina (Vell.) Brenan horizontally wrinkled and each commonly bearing, somewhere below e first pinna pair, a flattened or erect, oval to disciform, more or less centrally depressed gland 1-5 mm. long; 1-3 (rarely to 7) similar, smaller glands borne one between or just below each of the ultimate pinna pairs. Pinna pairs 7-35 or more, each pinna 1.2-7 cm. long, usual- ly opposite. Leaflets not always borne to the tip of the pinna, which may end in a fine point 1.5-2 mm. beyond the ultimate pair of leaflets. Leaf- lets imbricate or expanded, 20-80 pairs, .9-6 mm. long, .5-1.5 mm. wide, linear or slightly dilated in the middle, oblong to lanceolate, usually straight (not faleate), and the base oblique or truncate, at the apex acute to apiculate or obtuse; venation obscure, or with a straight or nearly straight, slightly excentrie mid-vein, and more or less promi- nent secondary reticulated veins; membranaceous to coriaceous and yellow to orangey, in fascicles of 1-7, minutely whitish-tomentose ‘to nearly glabrous in the bud; the heads axillary to the leaves and sub- terminal, or becoming arranged in racemose-panicu branch apices with the leaves reduced or absent. Peduncles 2-4 cm. 52 SIRI VON REIS ALTSCHUL long, puberulous to glabrous, thicker than filiform, each bearing just below the receptacle and often hidden by the mature head a narrow, more or less glabrous annular involucre. Calyx .6-3 mm. long. Corolla 2.5-4 mm. long. Stamens 5-8 mm. long; anthers each with a caducous land. Legume 10-32 em. long (including the stipe but not the thickened peduncle), 1-3 cm. wide, straight to sometimes faleate, oblongish to very elongated, regularly contracted to sinuate, or irregularly con- tracted where seeds have aborted, sometimes very flattened and thin, with margins often strongly thickened, even though sometimes narrow; at the base attenuate to obtuse or truncate, at the apex mucronate to acuminate to cuspidate or, if the tip has broken off, rounded; surface smooth to reticulated, and nitid; light, dark or reddish brown or dark gray in dried specimens. Seeds 8-16, thin, flat, orbicular to oblong, dark chestnut brown, very shiny, 12-20 mm. in diameter. The species is found in South America, with an approximate distri- bution of 0-30 degrees S. latitude. a. Anadenanthera colubrina (Vell.) Brenan var, colubrina Mimosa colubrina Vell., Fl. Flum. 11: tab. XVI. 1790. Acacia colubrina (Vell.) Mart., Herb. Fl. Bras. 107. 1837. Piptadenia colubrina (Vell.) Benth., Hook. Jour. Bot. 4: 341. 1841. Anadenanthera colubrina (Vell.) Brenan, Kew Bull. 2: 182. 1955. TYPE: the plate of Mimosa colubrina, tab. XVI, in Velloso’s Flora Fluminensis of 1790. This drawing delineates the fruits and foliage. It is not greatly detailed but conforms to the characters of Anadenan- thera colubrina var. colubrina. The interpretation of the type is aided by Velloso’s description, which was not published until 1881, The unarmed Mimosa with bipinnately compound leaves, the leaflets many- paired and lanceolate-linear; the legume linear, torulose or undulate. The author further designated this element as a few-branched tree, with the outermost pinnae of decreasing length, the legume curved on or more cm. long, up to 1.9 em. wide, very elongated, regu- larly contracted, very flattened and thin, with margins thickened but narrow; at the base truncate, at the apex mucronate; surface finely A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 53 reticulated, nitid; dark brown to dark gray in dried specimens. Seeds 10 or more, orbicular to suborbicular, up to 15 mm. in diameter. DISTRIBUTION. Herbarium specimens were examined from Argentina (Misiones) and Brazil (Bahia, Parana, Rio de Janeiro, Sao Paulo). The variety has been collected in woods, along rivers and in restingas. The material comes from localities of no greater altitude than 700 m. REPRESENTATIVE SPECIMENS. Brazil. Bahia: Ule 6955 (L). Parana: Dusén 11253 (Ss); Tessmann 6021 (A, U). Rio de Janeiro: Campos Porto 10415 (U), 10416 (8, U, US) ; Glaziou 10592 (A, C), 10596 (C, NY, 8); Widgren s. n. (S). Sao Paulo: Hoehne 1030 (GH), 28582 (A, F, GH, Ny, S). Other numbers from Brazil, but lacking locality designations, are Pohl 1445 (F, NY), s.n. (US); Widgren 419 (s). Anadenanthera colubrina var. colubrina may be distin- guished from A. colubrina var. Cebil by the following charac- ters: it is a shorter tree. The main leaf rachis is more deeply channeled ventrally. The petiole gland tends to be shorter but erect. The pinna pairs may be fewer but the pinnae longer than in A. colubrina var. Cebil, with the pinnae be- coming noticeably shorter toward the leaf apex. The leaflets tend to be fewer, longer and more often linear and nitidulous above, with only the mid-vein prominent. The heads may be smaller than in A. colubrina var. Cebil and always are whitish in the bud, with the two tips of the annular involucre extended and visible below each head at that time. The heads are borne in paniculate patterns in the branch apices, with the subtending leaves reduced to absent. The flowers are fewer per head than in A. colubrina var. Cebil, and report- edly are aromatic. The legume is longer and narrower than that of A. colubrina var. Cebil, very elongated, regularly con- tracted, flatter and thinner, with margins thickened but nar- rower than in the other variety ; the base is usually truncate, the apex mucronate, with the surface finely reticulated, nitid, dark brown to dark gray in dried specimens. The seeds of A. colubrina var. colubrina are more humerous and smaller. This variety tends to be generally more uniform in appear- ance than A. colubrina var. Cebil. Specimens examined show this less widespread variety to be found today from about 12 degrees §. latitude to 27 degrees S. latitude, overlapping geographically with the distribution of A. colubrina var. Cebil. ; 2b. Anadenanthera colubrina (Vell.) Brenan var. Cebil (Griseb.) 0 mb, nov. Based on Acacia Cebil Griseb., Goett. Abh. 19: 136. 1874. Piptadenia macrocarpa Benth., Hook. Jour. Bot. 4: 341. 1841. 54 SIRI VON REIS ALTSCHUL P. macrocarpa Benth. var. genuina Chod. & Hass., Bull. Herb. Boiss. IT, 4: 559. 1904. . macrocarpa Benth. var. genuina Chod. & Hass. f. puberula Chod. & Hass., ibid. macrocarpa Benth. var. Cebil (Griseb.) Chod. & Hass., Bull. Herb. Boiss. IT. 4: 560. 1904. macrocarpa Benth. var. Cebil (Griseb.) f. ruwpestris Chod. & Hass., ibid. Mey. ON macrocarpa Benth. var. Cebil (Griseb.) f. microcarpa Chod. & Hass., ibid. a") . macrocarpa Benth. var. vestita Chod. & Hass., ibid. P. macrocarpa Benth. var. AP cual Hoehne, Comm. Linh. Tel. Annex. 5 (Bot. Pt. 8): 28. Anadenanthera macrocarpa Pua litty Brenan, Kew Bull. 2: 182. 1955. Piptadenia microphylla Benth., Hook. Jour. Bot. 4: 341. 1841. P. Cebil (Griseb.) Griseb., Goett. Abh. 24: 121. 1879. P. Hassleriana Chod., Bull. Herb. Boiss. II. 4: 560. 1904. P. Hassleriana Chod var. fruticosa Chod. & Hass., ibid HOLOTYPE: Lorentz 194. Although no material was cited in the original publication of Acacia Cebil Grisebach, the description evident- ly was based on a collection of Lorentz. The description appeared not only in the journal referred to above, cited by Index Kewensis, but also earlier in the same year in Plantae Lorentzianae. I have examined Lorentz 194 from Grisebach’s herbarium. It is labelled Acacia Cebil. The original description of Acacia Cebil lacks references to flowers and Lorentz 194 has no flowers. However, both are consistent with the fruit and foliage characters of the present interpretation of Anaden- anthera colubrina var. Cebil. In 1879, Grisebach transferred Acacia Cebil to the genus Piptadenia. He added a description of the flowers and stated that he doubted whether the species was satisfactorily dis- tinguishable from P. macrocarpa. Grisebach treated Piptadenia Cebil as separate from P. macrocarpa because the petiole gland was absent and the legume contracted between the seeds of P. Cebil. In 1904, common name probably most often Paacciated with this variety in northwestern Argentina, where abundant collections have been made. SYNONYMY. Piptadenia macrocarpa Bentham (1841) is included in the synonymy of Anadenanthera colubrina var. Cebil on the basis of the original description and material examined. Bentham did not psi a holotype as such from the original material, which he cited “— Brazil, Pohl, Sello; Gardner 1584. No. 1107 of Martius, Herb. Pl. Bras. in éithex this or the preceding species.” The word “in” in the last sentence is a misprint for “is.” The “preceding species” is P. peregrina. However, Martius 1107 agrees with Bentham’s Piptadenta macrocarpa. I have examined two sheets cf that collection bearing A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 55 flowers (BM, S). Nevertheless, Bentham’s doubt about the identity of Martius 1107 at the time of publication and his citing Gardner 1584 ahead of it suggest that Gardner’s number should have precedence in the interpretation of the type of Piptadenia macrocarpa. Gardner 1584 also fits the description of P. macrocarpa. I have examined three sheets of that collection (BM, GH, NY). This leaves as the only other possibili- ties the unnumbered collections of Pohl and Sellow. Although they are the first named by Bentham, the exact sheets examined by Bentham may not be ascertained easily. Gardner 1584 seems to be the reasonable choice in interpreting the type, as it is the first number cited in P. macrocarpa all agree well with Anadenanthera colubrina var. Cebil, as here conceived. The original description of Piptadenia macrocarpa gives the pods as very smooth. The flowers in the two collections examined have glandular anthers and heads which are axillary and subterminal. Gardner 1584 was collected in Ceara, Brazil, which is outside of the known range of Anadenanthera colubrina var. colubrina, In 1904, Chodat & Hassler published three varieties of Piptadenia macrocarpa: genuina, Cebil and vestita, Specimens bearing the label P. macrocarpa var. genuina, if correctly identified now should be called Anadenanthera colubrina var. Cebil. Under Piptadenia macrocarpa var. genuina, Chodat & Hassler published f. puberula, with puberulent leaflets and up to 23 pinna pairs per leaf. I have examined material of the type, Hassler 7840, from near Bellavista, Paraguay (BM, NY) and identified it as Anadenanthera colubrina var. Cebil. Under Piptadenia macrocarpa var. Cebil, Chodat & Hassler distin- guished two forms, of which I have examined the types. Piptadenia macrocarpa var. Cebil f. rupestris has 10-12 pinna pairs per leaf, the pairs up to 1 cm. apart, and the leaflets perfectly glabrous. The only unusual character is the relatively wide spacing between the pinna pairs. The type is Hassler 6375, from central Paraguay (BM, NY). ducing a legume scarcely 10 em. long and inflorescences arranged in subpaniculate patterns. The specimen did not seem to represent a special taxonomic entity. Despite the sub- paniculate arrangement of the heads, it probably is better placed in the more variable Anadenanthera colubrina var. Cebil than in A. colubrina var. colubrina. It has a very short legume and was collected outside the range of var. colubrina. The material examined was Hass- ler 7459 (A, BM, MO, NY, S) and 7466 (A, BM, C, MO, NY, S), both from near Concepcién, Paraguay. Piptadenia macrocarpa var. vestita of Chodat & Hassler has conspic- uously puberulent foliage and heads which are arranged in racemose patterns. It falls within the wide range of variation found in Anaden- anthera colubrina var. Cebil; and the type, which I have examined, was collected in a region where A. colubrina is represented only by A. colubrina var. Cebil. The original material is Hassler 8348, from salty sand near Bellavista, Paraguay (A, BM, C, MO, NY, us). It may be significant that the arrangement of inflorescences in racemose patterns, 56 SIRI VON REIS ALTSCHUL which is relatively uncommon for A. colubrina var. Cebil, should crop up in a specimen from a saline habitat. Such a locality is unusual for this variety but may be characteristic for the coastal A. colubrina var. colubrina. Piptadenia macrocarpa var. plurifoliata (1919) Hoehne was to have been distinguished from P. macroca ar. macrocarpa by being a taller tree, over 20 m. high, by having more pinna pairs per leaf, 20-35, and by the different dimension of the heads, which were said to be 3 mm. in diameter in anthesis. Only the incredibly small size of the var. Cebil. The measurement published must be in error. The heads were said to be fasciculate in the leaf axils. Hoehne neither cited the original material nor indicated where it was collected, apparently in Brazil. I have been unable to locate the type but believe that the tree referred to cannot differ significantly from Anadenanthera colubrina var. Cebil. pairs per pinna. Finally, the original material was described as canescent-tomentellose, a condition more likely to be associated with A. colubrina var. Cebil. Piptadenia Hassleriana was described by Chodat in 1904. The type material, which I have examined, was cited as Hassler 6641, flowering (A, NY,S and a photo from F) and 6641a, fruiting (A, NY, S), both from woods in the Cordillera de Piribebuy, Paraguay. The description and the specimens agree with Anadenanthera colubrina. The tree in ques- tion was relatively small, 6-8 m. high. The pods were said to be not at all nitid. However, they were not scurfy or verrucose. Their small size and unusvally fine reticulation may have made them less shiny than is common. The heads were described as glabrous but proved to be puber- ulous, at least in the bud. Chodat added that Piptadenia Hassleriana had affinities with the species here called Anadenanthera peregrina, differing from it by having glandular anthers. It is strange that he A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 57 did not notice its closer relationship to A. colubrina. Although the tree described is small and has puberulous flower buds, the pinna pairs are 25-35, the leaflets 60-68 pairs, the petiole gland midway up the petiole, and the other glands 4-7, one between each of the ultimate pinna pairs. Hence, it is better placed in Anadenanthera colubrina var. Cebil than in the type-including variety. The locality of the collection is outside the known range of A. colubrina var. colubrina. Piptadenia Hassleriana var. fruticosa was based on Hassler 6688 from among barren cliffs in the valley of the Y-aca River, Paraguay NY). I have examined this material, which is both flowering and fruiting. Having glandular anthers, it belongs to Anadenanthera colubrina. The plant is shrubby, 1-2 m. high, with 10-24 pinna pairs, 40-50 very small leaflets, .9-1.1 mm. long, and small heads. The absence of the distinctive characters associated with A. colubrina var. colu- brina, and the fact that the material was collected in a region where, to my knowledge, only the other variety is found necessitates placing this unusual specimen in the all-accommodating A. colubrina var. Cebil. It may be that representatives of the characteristically variable A. colubrina var. Cebil respond radically to certain environments. Shrub to tree, up to 30 m. high. Petiole gland more or less flattened, up to 5 mm. long; 1-3 (rarely to 7) smaller glands borne between the ultimate pinna pairs. Pinna pairs up to 35 or more, each pinna 1.2 or more cm. long. Leaflets 20-80 pairs, .9 or more mm. long, often dilated in the middle. Venation with the secondary reticulated veins promi- nent. Heads up to 20 mm. in diameter, including stamens, in fascicles of 1-4, more or less glabrous in the bud; the heads axillary to the leaves and subterminal, or sometimes borne in racemose patterns in the branch apices. Legume 10 or more cm. long, up to 3 em. wide, margins strongly thickened; yellow-green in fresh, immature material. Seeds 8 or more, up to 20 mm. in diameter. DISTRIBUTION. Herbarium specimens indicate that Anadenanthera colubrina var. Cebil occurs in Argentina (Catamarca, Corrientes, Maranhao, Matto Grosso, Minas Geraes, Pernambuco, Piauhy, Rio de Janeiro, Sao Paulo), Paraguay and Peru. Sometimes cultivated, this variety is common to campos or fields, where it is solitary. It is also found in wooded fields, along rivers, by lakes, in riverside forests; in salty sand; on dry slopes, rocky hills, high woods on foothills : and in mountain regions in riparian localities. It is found up to altitudes of 2100 m Several seedlings, introduced into Orlando, Florida, from Sao ai Brazil, in 1940, have matured into large trees. They are esa e only living representatives of Anadenanthera colubrina in the northern hemisphere. They were to have been cut down in the spring of 1961, ding was to have changed t REPRESE (GH). Jujuy: Lillo 11034 (F. U): Venturi 5362 (F, US), 9730 (A, GH, MO, US). Salta: Rodriguez 1101, 1107 (NY); man: Venturi 1006, 1027 (A, US), 5455 (A, BM, 8). Bolivia: Steinbach 58 SIRI VON REIS ALTSCHUL 6657 (A, BM, F, S), 7226 (A, BM, F, MO, NY, S, U). Brazil. Minas Geraes: Glaziou 12653 (A, C). Pernambuco: Pickel 3133 (F, GH, S, US), 4A.20M- 12-16-29 (F, GH, MO, US). Paraguay: Anisits 1958 (s); Balansa 1419 (P, 8S); Fiebrig 119 (A, F, L); Hassler 3304 (A, BM), 7460 (A, C, MO, NY, S), 7600 (A, C, NY, S), 10657 (BM, K, MO, NY, S), 12230 (A, BM, C, F, GH, L, MO, NY, S, US), 12281 (A, BM, C, GH, L, MO, NY, US), 123809, 12320 (A, BM, C, GH, L, MO, NY, 8, US); Jorgensen 4405 (A, C, F, MO, NY, 8, US); Lindman A2057 (GH, NY, S, US); Malme 1096 (Ss). Peru: Mac- bride & Featherstone 1353 (F), 2421 (F, US); Weberbauer 6686 (F, GH); J. West 3679, 3845 (GH, MO). Anadenanthera colubrina var. Cebil may be distinguished from A. colubrina var. colubrina by the following characters : it is a taller tree. The petiole gland tends to be longer but flatter. The pinna pairs may be more numerous but the pinnae shorter than in the other variety. The leaflets tend to be more numerous, shorter and more often dilated in the middle, with the secondary reticulated veins prominent. The heads may be larger than in var. colubrina; they are not whitish in the bud, and the annular involucre is not easily visible below them at that time. In addition, the heads of var. Cebil are usually axillary to the leaves and subterminal, and less frequently borne in racemose patterns in the branch apices with the subtending leaves reduced. The legume 1s shorter and wider than that of var. colubrina, not as regu- larly contracted and with thicker margins. The seeds are fewer and larger. Anadenanthera colubrina var. Cebil is much more variable than is var. colubrina. Specimens exam- ined show var. Cebil to be found today from near the equator to about 30 degrees S. latitude, overlapping geographically with the distribution of var. colubrina. EXCLUDED NAMES Bentham suggested (1874-75) that Acacia grata Willdenow, “ex diagnosi nimis brevi”, might be the same as what he called Piptadenia macrocarpa. In 1919, Macbride stated that P. macrocarpa seemingly had not been given its first specific designation and that the epithet grata, published in 1809, might be available for it. The type material of Acacia grata, Hoffmannsegg s. n., “Habitat in Brasilia”, is no longer extant in Willdenow’s herbarium at Berlin, and I have been unable to locate a photograph of it. Furthermore, Macbride has informed me (in a letter, April 20, 1959), regarding his earlier proposal, that en probably the new combination was made as was the custom at one time on a statement of some previous investigator, or maybe simply on the statement of Index Kewensis . . . It is possible that Harms verified the status of grata as he was curator for many years at the Dahlem Herbarium”. Harms is no longer living, and Brenan suggested in 1955 A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 59 that it would be unwise to replace macrocarpa with an epithet of such uncertain application. Acacia grata is the name-bringing synonym for Mimosa grata (Willd.) Poiret, published in 1817 without adding any new information. In the absence of any further evidence as to the identity of Acacia grata, it seems best not to include that name in the synonymy of Anadenanthera colubrina var. Cebil. Mimosa angustifolia Lamarck (1783) was treated with a question- mark by DeCandolle, as possibly synonymous with Acacia angustiloba DeCandolle, which is a synonym of Anadenanthera peregrina var. peregrina. Lamarck’s original description appears to have been based on two sources of information. One of these was Nicolson’s description of Tendre a caillou france (1776), which was not accompanied by a Latin binomial and cited no material. The other source was non- flowering living material that Lamarck had seen, probably that which he mentioned as in cultivation in the “Jardin du Roi”. The tree was the most 12 per leaf. Without flowering speci would not want to speculate upon the i hera, as here interpreted pitied eee meiner considered here is Mimosa filicifolia (1783), which that author described from vegetative rate only and, in his own words, “par présomption” suggested that it might the same as Mimosa peregrina L. Lamarck cited no material. sae description alone is inadequate to identify it from among probably of the species of the: Mimosoideae. In 1844, Bentham transferred M:- 60 SIRI VON REIS ALTSCHUL mosa filicifolia to the genus Pithecellobium Mart. He had by then described the genus Piptadenia. Lamarck’s description of Mimosa filici- folia and Bentham’s description of Pithecellobium filicifolium differ somewhat as to the numbers of pinna pairs per leaf and of leaflets per pinna. Bentham did not indicate, either, whether he had seen Lamarck’s material. Specimens which I have seen labelled Pithecellobiwm. filici- folium in the Gray Herbarium come from Mexico and Central America, where none of the Anadenantheras is known to occur (E. Langlassé 103, E. W. Nelson 2671, C. A. Purpus 8251, all from Mexico; Tonduz 12503, from Costa Rica). The specimens examined appear to be some- what variable in numbers of foliage parts, approaching more closely the measurements given by Bentham. Both the descriptions, and the specimens examined, show fewer pinna pairs and fewer leaflets than is common for Anadenanthera peregrina. The leaflets also are larger. The heads are large by comparison, and the flowers are of a different structure. On the basis of the specimens seen and evidence from the literature, Mimosa filicifolia cannot properly be included in the f either species of Anadenanthera. The name Acacia trichophylloides Macfadyen was published in 1837, with a rather detailed description but without citing specimens. Later, Grisebach (1859-64) treated it as a synonym of Acacia Julibrissin Willdenow. He had seen Macfadyen’s material and apparently distin- guished it from Piptadenia isl na which also appeared in the same publication. Some time afterwards, Bentham (1874-75) stated that Macfadyen’s description of yeep trichophylloides was at variance with what he referred to as Albizzia Julibrissin but that it agreed well with Piptadenia peregrina. I have not seen Macfadyen’s material, but his description differs from Anadenanthera peregrina by referring to a smaller tree with angulose and purple branch extremities. The leaves are apparently much shorter and the pinna pairs many fewer than in A. peregrina. The ovary is reported to be white-villous with appressed hairs, a condition never found in Anadenanthera. Bentham may have been misled by the fact that Macfadyen gave Acacia angustiloba De- Candolle (1825), as a synonym of A. trichophylloides, for Bentham new A. angustiloba to be synonymous with Piptadenia peregrina. The herbarium specimens that I have seen of Albizzia Julibrissin from the West Indies are without flowers, but Macfadyen’s description of Acacia trichophylloides more appre Be eauerreay these specimens than it does those of Anadenanthera peregrina. Schefaiteis microphyll Outside of the New World, only one species esi raat to the Pipta- denia complex has been described with a truly capitate inflorescence. Now usually retevred to as Piptadenia novo-guineensis, it was original- ly described in the monotypic genus, Schleinitzia, as S. microphylla bringin (1891). The collection upon which the description was base s Hollrung 598; the type was destroyed at Berlin in 1943, and a Sepak at Breslau presently is unavailable for loan. The collection was made near Finschhafen, New Guinea. In describing Schleinitzia, Warburg stated that, dlthonet the tree was closely related to Pipta- denia, as Bentham interpreted it, the presence of glands on only 5 of A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 61 the 10 —— provided the basis for treating the species in a separate genus. Not long afterward, Warburg (1891) decided that Schleinitzia might abe be united with Pip snare on grounds that the number of anther glands was a variable soueaceit not well known in Piptadenia. The epithet microphylla was dropped because it already had been applied to a species of Piptadenia. Warburg added that, unless a new section were created to accommodate P. novo-guineensis, the species would have to be placed in the exclusively American section, Niopa. new section has not been published for that species, to my knowl- edge, and Piptadenia novo-guineensis has been mentioned rarely in the literature. The fact that the species is not American is a relatively eco tropical rain-forest. The leaflet of this species, as seen in Plate XI, ean be distinguished from that of Anadenanthera by the presence of three veins, instead of one, arising from the base of the leaflet. Further- ou complex. Finally, there is no ee from any soure satan this species is included in native pharmacopoeias, unlike the! species of nanthera. Collections of Piptadenia novo-guineensis from New Guinea, Rossel Island and two of the Solomon Islands, Bougainville and San Cristoval, show the species to be represented as an erect, thin, tall tree, 5-20 m. high, with open and spreading branches, and sensitive, pale green leaf- Cleared and stained leaflet of Schl dag microphylla. Hegelirry three PLATE veins arising from the base of the leaflet (L. J. Brass 2736, San Gyisiod 62 SIRI VON REIS ALTSCHUL lets. The bark is 1-4 mm. thick, whitish gray to purplish brown, with many broad, shallow longitudinal fissures in which are numerous trans- verse, light brown lenticels exuding an amber gum. The sapwood is 2 cm. deep and straw-colored. The heartwood is gray-brown (compared to reddish in Anadenanthera). The diameter of the trunk may reach 27 cm. in large individuals. The profuse flowers appear to be pale ink or cream-colored or white from a distance, due to the pink or purple filaments, which become whitish in older flowers. The anthers are yellow. The young fruits are red-brown, the mature ones dark brown or black. The tree is found in secondary growth of coastal rain-forests at altitudes up to 250 m. It has been reported also as fairly common in savanna or coral-limestone in the coastal plains up to 5 m. altitude. Specimens examined from Guadalcanal in the Solomon Islands represent the first in a series from the New Hebrides, New Caledonia, the Fiji Islands, Cook Islands and the Society Islands, all of which have been identified by previous botanists as Leucaena Forsteri Ben- tham (1842) or one of its synonyms. Bentham did not designate a type for this species. Of its two synonyms, Mimosa glandulosa Forster (1786) is a nomen nudum, and Acacia insularum Guillemin (1837) is not typified. These specimens from the eastern islands exhibit variation in: the the relative thickness and length of the peduncle; the position of the annular involucre upon the peduncle; the structure of the bract which immediately subtends each flower in the head and which may be filiform to broad at the base and sometimes extends into a point above the un- opened bud; the relative length of the calyx; the degree of coherence and the basal width of the petals; the size and shape of the anthers, whose chambers appear to be winged in some specimens (perhaps due to shrinkage in drying); the presence or absence of anther glands; the fruits, which range in color from reddish brown to black and may be pointed or rounded at the apex. When Bentham defined Leucaena in 1842, he recognized the similari- ties in habit and appearance between it and section Niopa of Pipta- denia. On the basis of characters enumerated by Bentham, plus my own observations of herbarium specimens of the genus, Leucaena appears to differ from Anadenanthera by having heavier peduncles, larger heads and flowers, basally narrowed and distinctly free petals, guineensis is more like that of Leucaena, being less membranaceous than that of Pityrocarpa. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 63 In view of the relative distinctness of the taxon now being called Piptadenia novo-guineensis, it would not be unrealistic to recognize it a genus separate from Piptadenia. In that case, it should be known as Schleinitzia microphylla Warburg. LITERATURE CITED ATCHISON, E. 1951. Studies in the Leguminosae VI: chromosome numbers among tropical woody species. Am, Jour. Bot. 38: 538-546. BEARD, J. S. 1946. The natural vegetation of Trinidad. Oxford Forest. Mem. 20: 1-152. ——_——.. 1958. The savanna vegetation of northern tropical Amer- ica. Ecol. Monogr. 23: 249-315. BENTHAM, G. 1840. Contributions towards a flora of South America. Hook. ae Bot. 2: 38-108, 127-146, 210-223, 286-324, 1841-42. Notes on Mimoseae. . . . Hook. Jour. Bot. 4: 323-418. 1844. Notes on Mimoseae.... Lond. Jour. Bot. 3: 82-112, 195-226. . 1874-75. Revision of the suborder Mimoseae.... Trans, Linn. Soc. 30: 335-664 BERGER, C. A., WITKUS, E. R., & MCMAHON, R. M. 1958. Cytotaxonomic studies in the Leguminosae. Bull. Torr. Bot. Club 85: 405-415. BRAZIER, J. D. 1958. The anatomy of some timbers formerly included in Piptadenia. Trop. Woods 108: 46-64 BRENAN, J. P. M. 1955. Notes on Mimosoideae. I. Kew Bull. 2: 170- 183. BRITTON, N. L., & ROSE, J. N. 1927. Addisonia 12: 37. CLOS, E. - 1929. Primera contribucién al conocimiento de los Arboles eae en la Argentina. Bol. Minist. Agr. Nac. 28: 29-63. CORREA, A. M. 1925. Geografia General de la Provincia de Tucuman. Buenos Aires, pp. 68, 70. c0zz0, D. 1951. Anatomia del lefio secundario de las Leguminosas Argentinas Mimosoideas. ... Revista Inst. Invest. Ci. Nat. CRUXENT, J. M. 1951 Venezuela: a strategic center for Caribbean archaeology. In Wilgus, A. C. The Caribbean at Mid-Century. Univ. Fla. Press, pp. 149-156. DARLINGTON, D. C., & WYLIE, A. P. 1956. ing Plants. Lond p. 151. DUCKE, A. 1922. Plantes nouvelles ou peu connues de la région Ama- zonienne. (Ile. Partie). Arch. Jardim Bot. Rio de Janeiro 3: 3- 269. Chromosome Atlas of Flower- DUCKE, Ay & ACK, G. A, 1953. Phytogeographical notes on the ENGLER, A., & PRANTL, K. VON. 1892. Natiirl. Pflanz. 3: 122. ERDTMAN, C. 1952. Pollen Morphology and Plant hiioanely. Stock- holt, pp. 225-227. ; FISH, M. 8., JOHNSON, N. M., & HORNING, E. C. 1955. Piptadenia alka- disiitin Indole bases of P. peregrina (L.) Benth. and related species. Jour, Am. Chem. Soc. 77: 5892-5895. 64 SIRI VON REIS ALTSCHUL FOSTER, A. S. 1950. Techniques for the study of venation patterns in leaves of angiosperms. Proc. Seventh Internat]. Bot. Congr. Stock- holm, pp. 586-587. ; GRISEBACH, A. H.R. 1859-44. Flora of the British West Indian Islands. London, 789 pp HERINGER, E. P. 1947. Contribuicaéo ao conhecimento da flora da Zona da Mata de Minas Gerais. Bol. Serv. Nac. Pesq. Agron. 2: 1-186. —————. 1954a. Alguns angicos e sua posigao sistematica. Chacaras e Quintais 39: 347-348. : . Piptadenia peregrina (L.) Benth. e P. macro- carpa Benth. Anais 5a. Reunido Anual S. B. B. 51-54. HERRERA, F. L. 1921. Contribucién a la Flora del Departamento del uzco. Primera Parte, Univ. Cuzco, Segunda Ed. Cuzco, p. 119. HIERONYMUS, J. 1882. Plantae Diaphoricae florae Argentinae. Buenos Aires, p. 88. HOEHNE, F. C., KUHLMANN, M., & HANDRO, 0. 1941. O Jardim Botanico de Sao Paul, Sao Paulo, p. 429. HOFMANN, A. 1959. Psychotomimetic drugs: chemical and pharma- ecological artis Acta Physiol. Phar. Neer]. 8: 240-258. KOEHLER, A. 1928. Tests on six Argentine woods. Trop. Woods 14: 1 5-20. LILLO, M. 1917. Segunda Contribucién al Conocimiento de los Arboles de la Argentina. Tucuman, pp. 30-31. LUETZELBURG, P. VON. 1922-23. Estudo Botdnico de Nordeste. Rio de Janeiro. Vol. II: 79; Vol. III: 55, 78, 269. LOFGREN, A. 1896. Ensaio para uma distribuigao dos vegetaes nos diversos grupos floristicos. . . . Bol. Comm. Geogr. E Geol. Sao Paulo 11: 3-47. MACBRIDE, J. F. 1919. Notes on certain Leguminosae. Contr, Gray Herb. 59: 17. - 1943. Flora of Peru. Bot. Ser. Field Mus. Nat. Hist. 13: 100- 103. MALME, G. 0. 1900. Ex Herbario Regnelliano. . . . Bihang Till Kung. Vet.-Akad. Hand. 25, Afd, III, No. 11: 36. NICOLSON, P. 1776. Essai Sur L’Histoire de St. Domingue. Paris, 374 pp. PACHTER, I. J., ZACHARIAS, D. E., ‘ RIBEIRO, 0. 1959. Indole alkaloids of ‘Acor eoccharias oe Dic yoloma incanescens, Piptadenia colu- brina and Mimosa hostilis. (ai Organ. Chem, 24: 1285-1287. PARODI, L. R. 1945. Las regiones fitogeograficas Argentinas y sus relaciones con la industria forestal. In Verdoorn, F. Plants and Plant Science in Latin America. Chron. Bot. 381 pp. SAFFORD, W. E. 1916, Ethnobotany. — Identity of Cohoba. . . . Jour. Wash. Acad. Sci. 6: 547-562. SEIBERT, R. J. 1947. A study of Hevea... in the Republic of Peru. Ann. Missouri Bot. Gard, 34: 261- 352. SEIFRIZ, W. 1943. The plant life of Cuba. Ecol. Monogr, 13: 375-426. SENN, H. A. 1938. Chromosome number relationships in the Legumi- nosae. Bibliogr. Genet. 12: 175- 836. A TAXONOMIC STUDY OF THE GENUS ANADENANTHERA 65 SOUZA, P. 1945. The Brazilian Forests. In Verdoorn, F. Plants and Plant Science in Latin America. Chron. Bot. 381 pp. SPEGAZZINI, C. 1909. Mycetes Argentinenses. Anal. Mus. Nac. Buenos Aires 19: 295. . 1923. Algunos observaciones relativas al suborden de las Mimosoideas. Physis 7: 308-315 SPRUCE, R. 1908. Notes of a Botanist on the Amazon and Andes. London, ae II, pp. 426-438. STROMBERG, V. 1954. The isolation = ana from Piptadenia peregrina. eg Am. Chem. Soc. 76: TORTORELLI, L. A. 1948. The Pr tnshee ciao timbers. Trop. Woods 94: 1-27. TRYON, R. M. 1944. Dynamic phytogeography of Doryopteris. Am. Jour. Bot. 31: 470-473. VELLOSO, J. M. DE C. 1881. Florae Fluminensis seu descriptionum plantarum. ... Arch. Mus. Nac. Rio de Janeiro 5: 1-467 WARBURG, 0. 1891. Beitriige zur Kenntnis der papuanischen Flora. Engl. Bot. Jahrb. 13: 336, 453 WEBERBAUER, A. 1945. El Mundo Vegetal de los Andes Peruanos. Lima, pp. 169, 410, 423, 430-433, 496, 508, 579, 601, 606. WILLIAMS, L. 1945. The phytogeography of Peru. In Verdoorn, F. Plants and Plant Science in Latin America. Chron. Bot. 381 pp. ate my Te INFRASPECIFIC VARIATION IN THE GUTIERREZIA SAROTHRAE COMPLEX (COMPOSITAE-ASTEREAE) OTTO T. SOLBRIG! The understanding of morphological variation in organ- isms is of fundamental importance for the student of sys- tematics and has received increased attention over the last fifty years. The presence of an array of diverse genotypes in a population allows evolutionary adjustments to environ- mental changes, and phenotypic differences to a certain extent reflect the amount of genetic variation in a species population. Yet phenotypic differences are not only governed y the theoretical or actual presence of a large number of different genotypes, but depend also on the buffering or “homeostatic” properties of the individual and the popula- tion, genetic homeostasis being apparently governed by the amount of heterozygosity in the gene pool of a population. The breeding system also plays a very important role in the amount and type of phenotypic variability present. Another important factor is the spatial relationship of the individual in the population, which will govern the amount of inbreed- ing vs. outbreeding in self-compatible plants, or influence the rate of recombination in self-incompatible species. Need- less to say, breeding systems and especially the actual amount of outbreeding are important determining factors of the type and amount of variation in a population of plants. Differences between populations can be due to a series of factors. The best-known is geographical variation correlated with environmental change. In such cases the differences can have a genetical basis as a result of different selection forces in response to changed conditions, or they may be the result «This work is an outgrowth of studies made in connection with my revision of the Nortt merican eae of Gutierrezia (Solbrig, 1960) and of studies of the South American species of the genus pisconpeidt n progress. many pers Shai helped i Particularly I want to thank Miss Julia Barth, Mr. "Joh n Chittum, De ae "Mrs. Via Ern, Dr. Peter Raven, Mrs. Lily Riidenberg, and Ing. Agr, Benno Schna: r. Reed C, Rollins read the manuscript and made oat suggestions. "Field Date in the United States and Mexico over the ve onne ‘ 67 68 OTTO T. SOLBRIG of different canalizations of development resulting from the altered environment. Other factors accounting for variation between populations are accidents of colonization in the expansion stages of a species, and in small populations chance elimination of certain genotypes from the gene pool. Unfortunately it is not always possible to tell in natural populations even after a detailed investigation, what the true genetic component of variability is, and a reasonable approx- imation is usually the best that can be hoped for. The experi- mental methods demanded to obtain even an approximate knowledge of the hereditable component of variation are not successfully applicable to all types of material and require, at best, several generations of breeding as well as the cultur- ing of large progenies before satisfactory results can be obtained. Therefore, experimental procedures are not always to be recommended, for example with perennial plants where a long generation period is involved. Even where a detailed analysis is not possible, a certain understanding of the pattern of variation is a necessary requirement for the student of evolution and the taxonomist alike. Different genotypes may give rise to similar pheno- types under a particular environmental condition, due to the phenomenon of “genetic homeostasis” as indicated above; furthermore, under different environmental situations, simi- lar genotypes may give rise to dissimilar phenotypes due to genotypic plasticity. Therefore, it is obvious that meas- urements of phenotypic characters cannot be taken as necessarily reflecting directly the nature of the genotypes. Nevertheless, natural selection affects phenotypes and only indirectly genotypes and, furthermore, the systematist inter- ested in the affinities and relationships of plants has by necessity to concern himself with phenotypic relationships. From this point of view, attempts to study variability of the phenotype without a knowledge of the genetic and de- velopmental bases are justified. Such studies are usually an integral part of any taxonomical work, although their accur- acy and the amount of detail possible vary appreciably. The present study is of the latter type. Gutierrezia sarothrae is a perennial globose subshrub relatively common in western North America west of the Great Plains. It extends from northern Mexico to southern British Columbia and Alberta in Canada. The plants are INFRASPECIFIC VARIATION IN GUTIERREZIA 69 ya oe SB) I my x « Fic. 1. Distribution map of G. sarothrae in the United States. Goode Base Map Labbe se by the University of Chicago. found growing in populations which have usually between 50 and 500 densely aggregated plants, although occasionally populations of up to several thousand plants are found. species is most abundant in the Rocky Mountain area, and adjacent plains, below 2000 m., in the valleys of the major drainage systems and neighborihg regions, such as the entirely absent from the Sonoran, Mohave and Great Basin desert areas, but is found on the western edge of this zone in southern and Baja California, around Reno, N evada, and in eastern Oregon and Washington (fig. 1). Over this area populations of Gutierrezia sarothrae show a great deal of variation, as is to be expected. This fact is even more strik- ing when the analysis is restricted to non-random samples, 70 OTTO T. SOLBRIG as herbarium specimens often are. Such a procedure has led to the description of 23 separate entities within this complex. Field studies have shown that the characters used to separate specimens in the herbarium show a continuous or almost continuous variation in the field, and “specific” characters are often found in different members of the same population, or even in different stages of the ontogeny of a single plant (Solbrig, 1960). One object of the present investigation has been to attempt to quantify and assess the morphological variation within and between populations of G. sarothrae and to attempt to establish and evaluate the relative impor- tance of the factors that account for this variability. It is hoped that the information presented will help in under- standing the evolutionary history of the species, and an attempt is made to relate it to the broader context of evolu- tionary theory. A second very special aim of this study has been to see if any quantitative differences can be discovered between the diploid and tetraploid races of Gutierrezia sarothrae. As was shown elsewhere, (Solbrig, 1960), there are no qualita- tive differences between the two levels of ploidy. In view of the fact that qualitative and/or quantitative differences were observed in the majority of the studies of polyploidy done to date, particularly those dealing with artificial tetraploids, the situation merits special consideration. The methods of investigation chosen, detailed measure- ments of random samples of populations, observations of greenhouse cultures, together with cytological studies, are believed to be the ones that are the most efficacious in pro- ducing useful data in the present situation. Techniques often used in the study of variation in annual plants, particularly crossing experiments, are unprofitable in the case of relative- ly large, self-fertile, perennial plants that require extensive greenhouse facilities and a minimum of a year between generations. These techniques have been applied randomly or not at all. CHROMOSOME STUDIES? Chromosome number has been determined from anther squashes, and root tip squashes in 59 populations (table 1). *I am particularly indebted to Mrs. Lily Riidenberg who prepared the slides used for the studies of chromosome morphology, and who also obtained some of the m eiotic counts. INFRASPECIFIC VARIATION IN GUTIERREZIA 71 Forty-nine populations proved to be diploid with n= 4, while the remaining 10 were tetraploids with n = 8. Meiosis was regular, and such irregularities as could be detected (occasional “sticky” bridges, and rare univalents) were present in low frequency and appear to be of no significance. One population from Texas had two supernumerary chromo- somes, which are apparently eliminated in the microgame- tophyte by formation of microsporocytes during meiosis. In order to determine whether there were any differences between populations in respect to the gross arrangement of the chromosomal arms by which populations could be cyto- logically identified, a study of root tip chromosome morphol- ogy was made. Plants of seven populations were investigated, grown from seed gathered at Dickinson, North Dakota (3244), Sheridan, Wyoming (3252), Evanston, Wyoming (38257), Dubois, Idaho (3254), Thompson, Utah (3260), Fountain, Colorado (3264), and Altus, Oklahoma (3289). The results of the study were largely negative, since no dif- ferences in the chromosome morphology of these populations could be detected. However, the data for a description of the karyotype of Gutierrezia sarothrae were obtained (Riiden- berg & Solbrig, 1963). The chromosomes of Gutierrezia are small, and therefore no attempt was made to measure arm length in the various populations, since the probable error would doubtless be too great to give meaningful data. Nevertheless the four chrom- osomes of G. sarothrae are qualitatively different and can be identified with relative ease. The centromere of one pair is median, while the position of the centromere in the other three pairs is subterminal. Since the median chromosome 1s y é IV Fic. 2. Metaphase plate of root tip mitosis and chromosome idiogram of G. sarothrae te ih pean UU s &” ae. IV —~_ 12 OTTO T. SOLBRIG TABLE 1. CHROMOSOME NUMBERS DETERMINED IN POPULATIONS OF G, SAROTHRAE NORTH DAKOTA. 12 mi. N. of Belfield on ity 85. Solbrig 3245. MONTANA. 3 mi. E. of Sanders. Solbrig SOUTH DAKOTA. Hwy. 385, on the N. & i ees border. Solbrig 3240. Hwy. 16, 10 mi. N. of Rapid City. Solbrig 3236 WYOMING. 17 mi. S.E. of Mountainview on road to Ticmeteee Solbrig 8134. Hwy. 30, 91 mi. W. of ae Bani S257. Outskirts of Sheridan. Solbri IDAHO. 12 mi. E. of Dubois on road . cae Solbrig 3254. OREGON. Field Creek. R. Ornduff s. NEBRASKA. 1 mi. N. of Lisco on oe a Orlando. ane $228. 2 mi. E. of Roscoe on road to Ogallala. Solbrig 3224. KANSAS. 10 mi. S. of Medicine Lodge. Solbrig 3292. COLORADO. 1 mi. W. of Hwy. 50, 22 mi. S. of Grand Junction. Solbrig $158. Colorado Nat. Mon., 1 mi. N. of Campground. Solbrig 3262. Fort Collins. A. Weber s.n. UTAH. Hwy. 24, 15 mi. S. of Jet. with Hwy. 6. Solbrig 3151. Hwy. 54, 7 mi. S. of Boulder. Solbrig 3150 Red Canyon, just before Bryce Canyon Nat. Park. Solbrig 3145. Diamond Valley, Diamond Mt. Rd., 16 mi. S.E. of Jct. Hwy. 44. Solbrig 3136. Hwy. 44, 16 mi. N. of Vernal. Solbrig 3258. OKLAHOMA. 4 mi. E. of Clinton, Hwy. 66. Sobriy 3290. TEXAS. Hwy. 62, 11 mi. W. of Altus, Okla. Solbrig 3289. Hwy. 67, 7 mi. E. of Alpine. Solbrig 3280. Hwy. 254, 5 mi. W. of Craford. Solbrig 3287. Hwy. 277, 5 mi. W. of Abilene State Park. pet 3286. Hwy. 67, 12 mi. W. of San Angelo. Solbrig 328: Hwy. 62-180, 23 mi. S. of White City, N. M. Sots 3271. NEW MEXICO. Hwy. 62-180, 3 mi. S. of White City. Solbrig 3270. Hwy. 285, 20 mi. N. of Roswell. Solbrig 3269. Hwy. 20, 12 mi. 8. of Ft. Sumner. Solbrig 3267. ARIZONA. 12.6 mi. E. of Ash Fork. Solbrig 2801. 0.8 mi. E. of Hyde Park. Solbrig 2805. CALIFORNIA. 12.3 mi. W. of Aguanga. Solbrig 2760. 9.9 mi. S. of Santa Ysabel. Solbrig 2763. 2 mi. W. of Temecula. Solbrig 2758, 9.8 mi. S. of Santa Ysabel. Solbrig 2765. 1.6 mi. W. of Rancho Santa Fe. shetitheg 2769. 8.5 mi. E. of Chula Vista. Solbrig 6.2 mi. E. of Chula Vista. Solbrig a 11.1 mi. E. of Idyllwild. Solbrig 2773. NEVADA. Reine City. R. H. Miller sn. se. R. H. Miller s.n. | 8 INFRASPECIFIC VARIATION IN GUTIERREZIA 73 B. n=8 SOUTH DAKOTA. Hwy. 385, 8 mi. S. of Oelvicks. Solbrig 3 COLORADO. ‘i mi. before Ww. entrance to Colorado Nat. oa) ‘batitle 326 TEXAS. ae 80, 11 mi. W. of Van Horn. Solbrig 3212. Hwy. 67, 6 mi. W. of McCamey. Solbrig 3282. Hwy. 54, 30 mi. N. of Van Horn. Solbrig 3273. NEW MEXICO. 3 mi. S. of Jet. Hwy. 285 & 286. Solbrig 3169. ARIZONA, 5 mi. N. of Payson. Solbrig 2794. 9.8 mi. W. of Seligman. Solbrig 2802. 19.8 mi. N. of Roosevelt Dam. Solbrig 2792. Jct. Payson-Phoenix & 488 Hwys. Solbrig 2793. also the shortest, it can easily be identified in root tips (No. IV in the diagram, fig. 2). The short arm of one of the other pairs is slightly longer than that of the other two. This characteristic, combined with the slightly shorter size of this chromosome as compared with the other subterminal chrom- osomes, makes this pair identifiable (No. II, fig. 2). Finally, the two remaining pairs can be separated by the presence of a secondary constriction (probably the nucleolus organizer) in one of them (No. I, fig. 2). The secondary constriction is located very close to the centromere, and is detected only in stages such as anaphase, when the chromosomes are not too contracted. The primary and secondary constrictions cannot always be told apart at metaphase, but due to a tendency for the satellite to break off easily when squashed hard (fig. 2), it is usually possible to tell the SAT chromo- some, even in stages of maximum contraction. Unfortunately no tetraploid plants were available for the study of chromosome morphology. Since the tetraploid plants are morphologically very similar to the diploids, an auto- polyploid origin is suspected. In such a case a certain amount of multivalent pairing at meiosis might be taken as a con- firmation of this view, but although a careful search was made, none was found. However, the finding of perfect dip- loid pairing in tetraploid plants does not necessarily rule out autopolyploidy, since pairing is apparently controlled by mendelian factors in at least some species (Lévquist 1956, Riley 1960, Johnson 1963). Another possibility is that these are segmental allopolyploids (Stebbins, 1950) or even true allopolyploids, although this last assumption is unlikely. (For more details see discussion under sibling and incipient spe- cies), 74 OTTO T. SOLBRIG tek Fic. 3. Distribution of diploid (dark dots) ee tetraploids (light dots) populations. Goode Base Map copyrighted by the University of Chicago. x STATISTICAL METHODS This work is primarily based on an intensive study of 65 natural populations more or less randomly distributed throughout the range of the species (for exact localities see appendix). The principal data were obtained by measuring up to 11 characters in each of 50 plants chosen at random from each population. In a study such as this one, involving a series of measure- ments, there are two major procedural problems: sample randomization and the choice of the statistical tests that will be most meaningful. An absolute prerequisite of any statis- tical work is making certain that no bias of any sort 1s INFRASPECIFIC VARIATION IN GUTIERREZIA 75 committed when the basic information is being gathered. This is accomplished by randomization. We now have several techniques available, borrowed from statistics and ecology, which can be used to randomize a sample. Before adopting any of these, it is essential to find out their relative degree of accuracy, that is, to find out to what degree bias will be eliminated, and then to determine whether the added lack of bias in the data obtained by using more laborious methods will justify the extra work needed to obtain the more precise result. Whenever the added accuracy is less than the expect- ed error from other sources, the limit beyond which no extra work is justified has been reached. Unfortunately, in the present investigation there was no way by which the total error due to sources other than sampling could be assessed (such as degree of phenotypic plasticity), and therefore an empirical test was used to assess the relative value of differ- ent sampling techniques. At first, samples were taken com- pletely at random. This was accomplished by the use of a grid and a pair of dice (a table of random numbers would have been equally useful). Next, samples were obtained by walking diagonally through the population and choosing each tenth plant. Since there was no appreciable difference in the results, the second, less time consuming method was adopted for the remainder of the investigation. The choice of statistical tests was a much harder problem. The major difficulty, of course, is that the nature of the data is being influenced by two major components: genotype and environment. If the usual statistical tests of significance are applied the level (usually 5% and 1%) at which the results are statistically significant or not can be learned. But such a statement is in my opinion not very meaningful biological- ly. F tests resulting from an analysis of variance indicate whether data can properly be considered to be from samples of the same statistical population. This information can be of use, but it is doubtful that this type of statistic has validity in the present instance, since the analysis of variance was developed for use with experimental data and not observa- tional data. The main difference is that the present samples Were produced under a variety of environmental conditions, and not under uniform or relatively uniform conditions as 1s usually the case. Significance tests (t tests) of the difference between diploids and tetraploids are valid, I feel, and they 76 OTTO T. SOLBRIG have been used. This is so because all the data gathered are lumped into two classes and the variable tested, chromosome number, is constant within each of the two categories. The standard deviation and standard error of the mean have been calculated for each character in each population. This is also a fairly straight-forward test of the amount of dispersion in each character measured in a population. Whenever pos- sible an effort was made to represent the statistics in a graphic manner, and only the more obvious conclusions as to correlations were made. STUDIES OF VARIATION The following study of the morphological variation of Gutierrezia sarothrae is based upon some 30,000 measure- ments made on over 3,000 plants. The total number of popu- lations for which there are herbarium samples is in the neighborhood of 5,000, and assuming an average of 100 plants per population (a rather low estimate), there are a minimum of 500,000 plants of G. sarothrae in nature. Prob- ably a figure twice as large somewhat approaches the num- ber of plants in nature, but it could easily be 10 times as large. Therefore our sample represents somewhere between 0.05 and 0.5% of all the plants. Fifty-three natural populations (fig. 4) were sampled and measured. The samples were obtained at random (see statis- tical methods). The characteristics measured or counted were: plant height, height and width of the involucre, num- ber of tubular and ligulate flowers, length of the pappus of the ligulate and tubular flowers, length and width of the achene of the tubular and ligulate flowers, pollen diameter and stomatal guard cell length. In addition, observations on blooming period were recorded in the greenhouse and in the field, and from herbarium specimens. PLANT HEIGHT. Measurements were taken with a steel measuring tape, the height recorded is that from the base of the plant to the tip of the longest branch. All measure- ments of plant height are expressed in inches, since no met- rically marked tape was available, and converting inches to centimeters would have introduced a source of inaccuracy. Height was measured in all 53 populations, the sample con- sisted of 50 plants in all but two populations; the sample in these two consisting of 35 plants, the total number of plants INFRASPECIFIC VARIATION IN GUTIERREZIA 77 Fic. 4. Mean height of populations of G. sarothrae. Goode Base Map copyrighted by the University of Chicago. . present. Plant height varied considerably both within and between populations. The largest single plant (38 inches) was over nine times as big as the smallest single plant meas- ured (4 inches), while the average height of the tallest pop- ulation (22.4 inches) was about 4.5 times higher than the shortest population. In spite of these differences the size distribution is continuous, both within and between popula- tions (fig. 4, table 2). When grown in the greenhouse and/or the garden, plants from different populations attained differ- ent heights (some were up to 2.5 times as tall as others), while the plants from the same locality were relatively uni- 78 OTTO T. SOLBRIG form. This suggests that the differences in plant height between populations are genetically controlled. On the other hand, the plants grown in the greenhouse were as a rule larger than the plants in the field, which in turn shows that there is a certain degree of unrealized potential under natu- TABLE 2. MEAN, RANGE, STANDARD DEVIATION AND COEFFICIENT OF VARIATION FOR PLANT HEIGHT. Coll.No. x + St. error Range 3 C(%) n NORTH DAKOTA 3 5.48 + 0.15 3.0- 7.5 1.09 24.8 50 6 6.81 + 0.18 3.5 - 11.0 1.29 18.9 50 MONTANA 3248 5.00 + 0.13 3.0- 7.5 0.94 18.8 50 3250 5.92 + 0.23 4.0 - 12.0 1.68 27.5 50 SOUTH DAKOTA 5.86 + 0.19 4.0 - 10.0 1.37 19.9 50 WYOMING 79 + 0.19 4.0- 9.0 1.13 19.5 35 3252 8.10 + 0.17 5.0 - 12.0 1.28 15.2 50 IDAHO 2911 9.12 + 0.70 4.0 - 16.0 4.92 54.1 50 2912 12.62 + 0.80 8.0 - 19.0 5.68 21.8 50 3254 6.06 + 0.09 4.0- 9.5 0.66 10.9 50 8255 7.65 + 0.07 5.5 - 12.0 0.49 6.4 50 NEBRASKA 2224 9.78 + 0.27 6.5 - 15.0 1.89 19.4 50 231 5.65 + 0.25 8.5 - 10.0 1.76 31.1 50 KANSAS 292 10.33 + 0.32 6.5 - 15.5 2.23 22.5 50 COLORADO 3153 9.61 + 0.30 6.0 - 16.0 2.14 22.3 50 3261 9.73 + 0.26 6.5 - 14.5 1.81 18.6 50 3262 03 + 0.32 3.5 - 18.0 2.23 27.7 50 3263 9.59 + 0.17 5.0 - 13.0 1.22 12.7 50 3264 7.36 + 0.17 4.5 - 13.0 1.17 15.9 50 3265 8.62 + 0.18 5.0 - 18.5 1.29 14.9 50 8161 9.02 + 0.30 5.5 - 13.0 2.10 23.3 50 UTAH 8150 10.56 + 0.37 6.0 - 16.0 2.64 25.0 50 3151 11.20 + 0.42 6.0 - 18.5 2.95 26.4 50 3256 10.38 + 0.32 5.0 - 16.5 2.29 22.3 50 3258 5.39 + 0.23 2.6- 8.5 1.60 29.6 50 3259 5.61 + 0.21 3.5- 9.0 1.48 26.3 50 3136 4.98 + 0.11 4.0- 6.5 0.76 15.2 50 3141 9.59 + 0.27 6.0 - 15.0 1.90 19.8 50 3142 9.85 + 0.38 5.0 - 16.5 2.66 27.0 50 3143 10.75 + 0.46 5.0 - 21.0 3.25 30.2 50 3145 7.77 + 0.23 5.0 - 11.5 1.66 21.4 50 OREGON 2905 9.26 + 0.68 4.0 - 14.0 4.84 52.0 50 OKLAHOMA 12.12 + 0.30 7.5 - 19.0 2.14 17.1 5 3290 8.91 + 0.22 6.0 - 14.5 1.59 17.9 50 3291 15.31 + 0.32 11.0 - 25.0 2.28 14.9 50 3212 10.96 + 0.43 6.0 - 15.5 3.04 27.8 50 3271 7.95 + 0.29 5.5 - 17.0 2.03 25.5 50 INFRASPECIFIC VARIATION IN GUTIERREZIA 79 3273 15.00 + 0.53 8.5 - 22.0 3.78 25.2 50 3280 11.50 + 0.32 8.0 - 16.0 2.24 19.5 50 3284 13.31 + 0.21 7.5 - 19.0 1,50 11.3 50 3286 11.02 + 0.51 6.5 - 21.0 3.58 32.4 50 3287 15.13 + 0.54 9.0 - 23.0 3.82 25.3 50 3288 13.78 + 0.39 8.0 - 19.0 2.73 19.8 50 NEW MEXICO 3169 8.97 + 0.31 4.5 - 14.5 2.22 24.8 50 3215 16.63 + 0.49 10.0 - 28.5 3.45 20.7 50 3266 12.46 + 0.25 5.5 - 17.5 L717 14.2 50 3269 9.70 + 0.30 5.0 - 15.0 2.15 22.2 50 ARIZONA 5 <0 2 @. 4.0 - 11.0 3.18 40.7 50 2794 17582 <= 1.16 10.0 - 28.0 8.14 33.2 50 13.10 + 0.81 6.0 - 19.0 5.74 62.7 50 CALIFORNIA 17.27 + 1.37 10.0 - 27.0 9.70 56.1 50 2769 20.3 + 1.26 13.0 - 29.0 7.48 36.8 35 2758 22.4 = 1.97 10.0 - 38.0 13.90 62.1 50 ral conditions. This is according to expectations based on theoretical formulations and observations in other species. There is a definite geographical correlation between plant height and latitude, the northern plants being generally smaller than the southern ones (fig. 4). This trend is slightly obscured by an east to west trend (western plants tend to be larger) of less magnitude and by the distributional rela- tionship with the Rocky Mountains. Gutierrezia sarothrae is absent from higher elevations, but it grows in some of the valleys at lower elevations in the Rocky Mts. of Colorado, Wyoming, and Montana. The influence of altitude is not quite clear, but in any event plants north and east of the Rocky Mts. are smaller than those growing south and west. In the beginning I thought the differences found in plants of the two areas might merit subspecific recognition, but the present data do not warrant such a conclusion. Tetraploid plants were found to be slightly larger than the average, but all populations fell well within the variation pattern of the diploids. Polyploid plants or populations can- not be distinguished from the diploids on the basis of plant height. INVOLUCRE. The length and width of the involucre are related to the number and size of flowers in a head, and are presumably under the same selective control that affects the floral characteristics. The character has been measured in 50 plants in each of 41 different populations (table 3; fig. 5) the measurements having been made in the field with the aid of a caliper. Of all the characters studied, involucre dimen- 80 OTTO T. SOLBRIG Fic. 5. Height and width of involuere; triangles drawn proportional to the mean values of the populations. Goode Base Map copyrighted by the University of Chicago. sions, proved to be the least variable, notwithstanding the fact that the mean involucre length of the population with the longest involucre was twice that of the one with the shortest involucre (2.9-5.7 mm.) and the mean width of the widest involucres was 2.5 times the mean of the narrowest involucres (1.1-2.5 mm.). The shortest measured involucre was 2.2 mm., the longest 6.5 mm.; the narrowest was 90.9 mm., the widest 2.9 mm.; most involucres were between 3.0 mm. and 3.5 mm. long and 1.5-1.7 mm. wide. The coefficient of variation fluctuated around 10% with minimum and maxi- mum values of 4.2 and 19.2%. There is a weak linear corre- INFRASPECIFIC VARIATION IN GUTIERREZIA 81 TABLE 3. MEAN, RANGE, oe DEVIATION AND COEFFICIENT OF V RE ARIATION FOR COLL. NO. LEN ae CTERS OF THE INVOLUCR xtst.er. range s C(%) NORTH DAKOTA 3243 3.95 + .05 3.1- 4.6 3246 3.89 + .05 3.2-4.9 .36 A 3248 4.08 + .06 3.2-5.0 .43 3250 8.97 + .06 3.2-5.0 .42 SOUTH DAKOTA 3238 3.57 + .06 2.4-4.9 .46 WYOMING 3252 415+ .06 3.5-5.4 .41 IDAHO 3254 8.80 + .05 32-4.5 .87 3255 415+ 03 3.4-46 .20 NEBRAS 3224 4.14+ .08 3.1-5.8 .55 3231 4.25 + .07 3.2-5.6 .48 KANS. 3292 4.01 + .05 38.0-4.5 .87 3261* 6.10 = 2 4.3-6.2 .44 3262 3.98 + .06 3.2-5.0 .41 3263 4.36 + .06 3.9-5.3 .42 3264 3.94 + .04 3.3-4.5 .32 OKLAHO 3290 $8.99 + .05 3.2-4.7 .84 3291 8.98 + .06 2.7-4.5 .43 3271 3.16 + .03 2.8-3.6 .22 3273* 8.12 + .05 2.7-3.8 .35 3280 8.51 + .04 2.9-4.2 .82 3284 3.82 + .05 2.2-4.3 38 6 3.90 + 04 38.3-4.5 .29 328 + NEW 3169* 357+ .02 3.1-4.0 16 3215 8.72 + 03 88-42 28 266 4.07 + .05 3.3-4.9 .36 3269 2.93+.08 25-33 .24 ZO. 2794" 5.65 +. 5 85 2801 3.78 + 02 27-43 .16 2805 8.92 + .04 8.5-49 .26 2766 3.51 + .08 2.5-4.5 .24 2769 3.48 + .08 2.7-3.9 .60 2758 4.36.08 35-49 .56 *Polyploid populations (n = 8). ~ ou rm H+ — oa ~ I+ mn =a + 1 Be cal in co Ht Ht a ie] Ht Ht 2 H+ I+ 2 HH 3 He + i+ Ht i+ a + H+ + 02 02 1,2 - 2.0 1.8 - 2.1 1.3 - 1.9 1.2 - 1,8 1.1 - 2.0 1.3 - 2.0 1.3-1.9 1.3 -1.8 1,2 - 2.9 1.3-2.1 -14 15 13 17 WIDTH xt st.er. range s C(%) 8.8 9.6 8.1 10.5 12.0 11.5 82 ; OTTO T. SOLBRIG 3243 N. DAKOTA 3248 ase" AN MONTANA i\ / x H4 \ \ / \ vs * 3250 , 3142 * \ MONTANA \ pee 3261 3252 3144 WYOMING WYOMING ve seaablet A ‘ ‘ \ \ \ \ \ \ 3238 3136 * 3262 PS ad COLORADO S.DAKOTA 3263 3231 COLORADO NEBRASKA 3224 pee COLORADO NEBRASKA es 0 Fic. 6. Frequency distribution of tubular (dotted lines) and ligulate (full lines) flowers per head in populations of G, sarothrae. Aste smn indicate samples of less than 50 plants. For exact localities see app INFRASPECIFIC VARIATION IN GUTIERREZIA 2794 i ARIZONA 83 3292 KANSAS 2801 329) ARIZONA OKLAHOMA 2805 3290 ARIZONA OKLAHOMA 3265 NEW MEXICO 3266 NEW MEXICO 2758 CALIFORNIA 3269 NEW MEXICO TEXAS 2766 CALIFORNIA 3215 3169 NEW MEXICO NEW MEXICO \ ry | 1 \ 2769 CALIFORNIA - Frequency distribution of tubular (dotted lines) and gia) (full lines) fionrers Fic per end in popuations of G. sarothrae. For exact localities see aj 84 OTTO T. SOLBRIG lation between length and width of the involucre, obscured in part by a large amount of scattering. No correlation could be observed between dimensions of the involucre and geo- graphical distribution, aside from the slightly larger sized involucre of the California populations (table 3; fig. 6). Degree of variation was relatively uniform, the differences observed being randomly distributed. Two polyploid populations (3261 and 2794) have signifi- cantly larger involucres than all the other populations meas- ured; the other three polyploid populations have involucres slightly smaller than the average. Although it is possible that many of the plants with very large involucres that have been collected repeatedly are polyploids, no generalization that polyploids always have larger involucres can be made as far as Gutierrezia sarothrae is concerned. The shape of the involucre in all cases, was the character- istic turbinate one of the species. No significant differences in this character were detected between any of the popula- tions studied. NUMBER OF FLOWERS. Of all the characteristics studied, number of flowers per head is the most interesting for sever- al reasons: it was the main “key” character used to distin- guish the many species erected within the complex; it shows the greatest amount of variability, at least among the char- acteristics investigated; it is potentially the feature most directly subjected to selection by pollinators; and it illus- trates best some of the puzzling problems encountered in the morphological analysis. There are two main components to consider: number of ligulate and number of tubular flowers in a head. These are not entirely independent of each other as we will see further on, but for the purpose of analysis they are best treated inde- pendently (table 4). The number of ligulate flowers in a head ranges between 2 and 8, but over 50% of the heads counted had 5 ligulate flowers. The mean of 30 of the 41 populations analyzed was between 4 and 5; of the remaining 12 populations, 8 had values between 5.01 and 6.00, one had a mean of 6.32 and two had values below 4. The coefficient of variability was usually between 11 and 15%, with a low of 9.5% and a high of 26.7%. Since number of flowers is a discrete and not a continuous variable it is obvious that we only have seven INFRASPECIFIC VARIATION IN GUTIERREZIA 85 TABLE 4. MEAN, RANGE, MODE, STANDARD DEVIATION AND COEFFICIENT OF VARIATION FOR NUMBER OF FLOWERS IN A HEAD POP. NO. __ LIGULATE FLOWERS __ TUBULAR FLOWERS et range c et range C st.er. &mode s (%) st.er. &mode s (%) NORTH DAKOTA 8243 4:60 4:09: 8=5-(5):) 617 182 8.08 + .09 2-6 (3) .64 20.8 3246 4.24-+ 09 8-5 (4) ..65 15.4 8.12 +.09 2-5 (3) .66 21.0 MONTANA 3248 4.99 12, 3255 (6) (588 7 195 2.96 +.10 2-4 (38) .67 22.7 3250 4.58.08 $=5,(5)...64 11.7 2.76 + .08 2-4 (3) .59 21.4 SOUTH DAKOTA 3238 ANTS 08 =i 8-6) (5) 346e. ie 3.02 + .09 2-5 (3) .66 21.7 WYOMING 8252 4.80+ .07 3-5 (5) .46 9.5 3.30 + .11 2-5 (3) .81 24.4 ID. 3254 ATO 3104 8-6 (6). 6%. 14ae 8.744 11 2-6 (4). 77) 20.6 3255 4.60 + 09 9-6 (5) 64 ‘189 3.52 +.08 8-5 (3) .57 16.8 NEBRASKA 3224 4.68+.09 8-6 (5) .65 14.0 $46-+.10 .2-4-(8) 70 205 3231 4.36+.10 2-5 (4) .70 16.1 2.64+.08 2-4 (3) .54 20.4 KAN 3292 4.56+.07 4-5 (5) .50 10.9 2.54+.08 1-4 (8) .57 22.5 co 3261 4.88+.08 4-6 (5) .56 11.4 4.54.11 8-7 (4) .76 16.8 3262 4.54 +08... 8. = 6.(6) ..457.-, 126 2.98 + .08 2-4 (3) .59 19.7 3263 6:82 + 12" 4-8 (6) "80> 184 4.76 + .18 2-7 (5) 1.24 26.0 3264 4.90 + .09 4-6 (5) .64 13.0 3.40 + .08 2-4 (8) .57 16.9 3265 5.20 10> 18256 1). eda bee 3.36.10 1-4 (3) .67 20.0 ITAH 3150 5.12+.10 4-7 (5) .69 13.5 4.96.11 8-7 (5) .81 16.3 8151 5.00 + .10 38-6 (5) 10.2 140 4.14+.09 38-6 (4) 64 5.5 3256 4.10+ .10 8-6 (5) 67 14.8 3.48 +.13 2-6 (3) .88 25.4 3258 5.58 ED a0" (6) 176 18.6 4.524.138 2-7 (4) .938 20.5 3259 4.80+.10 3-6 (5) .78 15.2 3.30 + .18 2-6 (4) .88 26.8 OKLAHOMA 3290 484+ .07 4-6 (5) .51 10.5 8.12 + .07 2-4 (3) .50 16.0 3291 4.82 + .09 8-6 (5) .62 12.9 2.84.09 2-4 (3) .64 22.5 TEX 3212 4.92 + .09 8-6°(6) 60° 12.2 5.06 + .19 2-7 (6) 1.36 26.9 8271 4.50 + .09 3-5 (5) 61: «18.5 2.98 4+-..08. 2-4.(8). 56... 18.7 3273 4.58 +.09 3-6 (5) .64 14.0 319 + 09... 2-5 (8) ...66 22 3280 5.40 18! OS OP GY OE” 16S 3.68 +.10 8-5 (4) .71 19.4 8284. 4.62 10. 46 (4). 6 146 2.68.10 2-4 (8) 68 25.5 3286 5.14+.09 4-7 (5) .62 12.1 3.04+.12 0-6 (3) -86 28.1 3287 4.72 + .11 3-6 (5) 8 16.5 ge EA 28S) NTS 2a NEW MEXICO 69 8.58 0-65.07. 82 ae (A), a Bk a 3.22 +.08 2-4 (3) ve cried 3215 4.92 + .08 4-6 (5) .57 11.6 4.52+.08 4-6 (4) a ries 3266 A94 1015) $5 6<(6) SLs 3.26+.09 2-4 (3) = ° 3269 8.94+.09 2-5 (4) .66 16.6 969 +- .07...2=8 (8). 8 ARIZO 2794 460+ .07 4-5 (5 50 10.9 3.60+.18 0-4 (4) - pS : 2801 4.40+.08 3-5 (4) .60 18.7 3.80.13 2-6 (4) ty tee 05 4.60 + .07 4-5 (5) .50 10.9 3.50 + 07 3-4 (4) : 0-6 2766 5.50 + .21 4-7 : =: 9 2769 5.80 + .12 4-8 (6) .84 14.5 590 2.10 4-716) te 2758 5.50 + .09 4-7 (6) .65 11.8 5.380-+.11 4-7 86 OTTO T. SOLBRIG classes altogether (2, 3, 4, 5, 6, 7, 8) and at first glance it appears that the differences observed between populations might be significant. Nevertheless three of the seven classes (2, 7, 8) are present only occasionally. In 32 populations the modal class was five, in six it was four and in three it was six. Only the populations with six and four could be considered as differing sufficiently to be of taxonomic impor- tance. However, if we look at the frequency distributions of ligulate flowers in a head (figs. 6 and 7), we see that there is a considerable overlap between most populations. The possible explanation for the variation in mean and modal number between populations is discussed below (see under discussion of eco-geographic differentiation). The picture revealed by the data on number of tubular flowers in a head is similar to that shown by the ligulate flowers, with minor variations. The modal class is three tubular flowers per head. This class characterizes only 24 out of 41 populations; four is the next most common, being the modal class in 11 populations ; five is the most abundant class in four populations ; and finally two populations have heads mostly with six tubular flowers. Individual heads with between none and seven tubular flowers have been found. It is clear that the number of tubular flowers per head is a more variable character than the number of ligulate flowers per head, and this is reflected by the coefficient of variation which fluctuates between 11.9% and 36.1%, with most populations having values around 20%. If we examine the relation between ligulate and tubular flowers we can see that they are not independent of each other but are directly correlated. On the basis of the ob- served frequencies of the moda] classes, the expected number of populations with a certain number of ligulate and tubular flowers was calculated and compared with the observed numbers (table 5). It will be noted that the classes with a similar number of ligulate and tubular flowers such as 4/3, 6/5, 5/3, 5/4, 6/5 and 6/6 are found in the expected numbers or in higher frequencies than expected, while in combinations such as 4/5, 4/6, 6/3 and 6/4 where the number of ligulate and tubular flowers is more unequal, the observed numbers (zero) are less than the percentage (10%) of the total ex- pected (4). The correlation between ligulate and tubular flowers is shown graphically in fig. 8, where in addition to INFRASPECIFIC VARIATION IN GUTIERREZIA 87 FLOWERS RAY f @oiPvoiw, v4 SAROTHRAE \ =) TETRAPLOID, N*8 OCD BRACTEATA ~-— MICROCEPHALA € — —— er or * DISK FLOWERS Fic. 8. Regression curve showing correlation between number of ligulate and tubular flowers in a head. Gutierrezia sarothrae populations, data from other species have been plotted. The regression can be clearly seen, and it also can be observed that it is parabolic rather than linear. The reason for this is clear. Since in a capitulum of Gutier- rezia there is only one row of ligulate flowers, as the total number of flowers in a head increases the ligulate flowers in the periphery increase in relation to the diameter of the head (a linear function), while the tubular flowers Increase in relation to the surface of the head (a geometric function). Some latitude is of course present as in all biological materi- al, and changes in relative size of the flowers or in the con- vexity of the receptacle, can affect this relationship to a certain extent. Finally, it should be pointed out that flowers at various stages of abortion were occasionally found. These were counted when discovered, but it is almost certain that some have been overlooked, introducing another source of error. It would be very interesting to investigate the mor- phogenetic processes involved in the formation of ligulate and tubular flowers where a better explanation of the corre- 88 OTTO T. SOLBRIG TABLE 5. EXPECTED AND OBSERVED MopAL NUMBERS OF TUBULAR AND LIGULATE FLOWERS IN A HEAD CLASSES EXPECTED OBSERVED Ligulate flowers| Tubular flowers % No. pop. % No. pop. 4 3 9 4 12.5 5 4 4 4 1 3 i 4 5 2 I, — — 4 6 0 af — a 5 3 47 19 47.5 19 5 4 20 8 25 10 5 5 8 3 5 2 5 6 4 1 3 1 6 3 4 1 _- — 6 4 z 1 — — 6 5 1 1 5 2 6 6 A — 3 1 102.4 AL 104, 41 lation between these two types of flowers might be expected to be found. There is some correlation between numbers of flowers and geographical distribution. All the populations north and east of Colorado are fairly uniform with mostly modal classes of 5 ligulate and 3 tubular flowers and mean values of approximately 4.50-4.80 for ligulate flowers and 2.80-3.10 for tubular flowers. The populations in Colorado and Utah are more variable, with a tendency to higher values for both ligulate and tubular flowers, and a similar situation is found in some Texas populations. Two populations in New Mexico had values of less than four for ligulate flowers ; and, finally, the three California populations studied had consistently higher values than the average. The significance of these variations is hard to assess, but it is possibly related to the more optimal conditions of the populations nearer to the center of distribution than those towards the periphery (for further information see under discussion). No correlation could be detected between chromosome number and number of flowers in a head. PAPPUS. The pappus in the flowers of Gutierrezia saro- thrae is formed by 6 to 9 short paleaceous bracts. In this respect Gutierrezia is an exceptional genus in the tribe Astereae where the genera generally have a pappus formed by well developed bristles, or wanting by reduction. The pappus of Gutierrezia is of a type that is also characteristic INFRASPECIFIC VARIATION IN GUTIERREZIA TABLE 6. COLL. NO. ae, Ot. CPs NORTH DAKOTA 3248 1.39 + .03 3246 1.48 + .03 MONTANA 3248 1.36 + .03 3250 1.44 + .03 SOUTH DAKOT 3238 vy se age WYOMING 3252 1.26 + .03 IDAH 3254 1.28. =.402 3255 1.34 + .03 N KA 3224 1.34 + .03 3231 1.38 + .04 KA 329 1.27 + .02 COLORADO 3261 1.40 + .03 3262 1.24 + .03 3263 1.22 + .03 3264 1.36 + .04 3265 1.22 + .03 UTAH 3150 1.05 + .03 3151 1.54 + .04 3256 1.31 + .02 3258 1.28 + .03 3259 1.21 + .03 OKLAHOMA 3291 1.22 + .02 TEX 3212 1.19 + .02 3271 1.13 + 02 3278 1.21 + .02 3280 1.19 + .02 3284 1,22 + .03 3286 + .08 NEW EX! 3169 1.11 + .03 3215 1.20 He 3266 1.30 + .04 3269 1.25 + .08 ARIZONA 2794 1.66 + .0 2801 1.06 + .02 2805 1.18 + .01 CALI NI 2766 =: 1.14 + .02 2769 -93 + .01 2758 1.46 + .02 range 1.0 - 1.9 1.0 - 2.0 1,0 - 2.0 1.0 - 2.0 8 - 1.6 TUBULAR FLOWERS s C(%) 24 17.1 18 17.3 25 18.2 24 16.7 23 19.8 21 16.4 17 13.6 14.5 23 17.1 27 19.3 15 12.1 20 14.6 22 17.6 21 16. 26 19.4 21 17.2 19 17.7 30 19. 17 13.0 20 15.5 «19 15.4 16 13.0 14 11.4 -16 14.4 14 11.2 14 12.1 22 17.8 +24 20 21 18.8 12 55 OS | 25 19.3 18 14.6 16 9.5 14 13.0 Al 8.9 -16 13.7 10 11.2 16 11.0 89 MEAN, RANGE, STANDARD DEVIATION AND COEFFICIENT OF VARIATION FOR Pappus LENGTHS LIGULATE FLOWERS 2 Sher: -03 02 2 Hit it = is + > pat ca ot) I+ —) ror) a3 i) + I+ & a on + Ht 28 on o I+ = xc) o o It Ht Ht HE . ° = LJ . r= + J+ 1+ 1+ H+ on ot H+ ra) i) ts He He Ht [+ i+ range s C(%) 21 82.2 21 29,2 18 © 21.6 22 © 30.5 10 28.4 a es a5 (aes 16 46-286 2229.0 1s. 2s 1119.0 te: S| 16 26.8 14... 22.9 lt “28. 13 25.6 09 «16.7 21 (02k 18 25.5 a 32 16 3 12 20.9 13 20.9 09 = «18.9 1827.8 12 28.4 1 oe it. 248 12 20.2 1428.4 14. 360 1529.5 12 10.9 07 18.6 20: 38.8 1T | 28.8 08 «22.4 1017.4 of a group of related genera including Grindelia, Amphia- chyris, Xanthocephalum and Olivaea. This pappus type is 90 OTTO T. SOLBRIG more like that of many genera of Heliantheae than of the Astereae generally, and it can be considered to be a primitive character. Measurements were made of the length of the pappus of the tubular and ligulate flowers. The pappus of the tubular flowers is roughly twice the length of the pappus of the ligu- late flowers. The results are summarized in table 6. The main function of the pappus is that of aiding in the dispersal of the fruit. The agent dispersing many of the Compositae fruits is the wind, and the bristly pappus of such genera of Astereae as Erigeron, Conyza, Haplopappus and Aster is particularly well adapted to dispersal by wind. The paleaceous pappus of the Heliantheae and Gutierrezia may be interpreted either as a transitional stage between the ancestral calyx and an advanced bristly pappus, or it may be considered as a particular adaptation to dispersal by agents other than wind, especially animals. More experimental evi- dence is needed to clarify this point and phylogenetic consid- erations should be brought to bear on the problem. In any event, in Gutierrezia the structure is somewhat reduced and no obvious function has been discovered. In view of this, it is of interest to see the amount of variation present in this structure. The length of the pappus is remarkably uniform from population to population, the mean varying from 0.93 mm. to 1.48 mm. for the tubular flowers and from 0.34 mm. to 1.07 mm. for the ligulate flowers. The coefficient of variation varies from 9.5% to 20% for the pappus of the tubular flowers and from 10.9% to 32.2% for the pappus of the ligu- late flowers. The greater variability in the pappus of the ligulate flowers is of interest. Part of the variation is due to experimental error, which in such a small structure becomes an important element, but not all of the higher variability of this structure compared to that of the tubular flowers can be ascribed to experimental error alone. The pappus of the ligulate flowers is very reduced, and in all likelihood no long- er functional. The variability could be due to a relaxation of selection pressure for whatever function it had, but I believe that this is not the likely explanation. In this character we can observe a slightly larger inter- deme variability towards the center of the range, than that observed towards the periphery. INFRASPECIFIC VARIATION IN GUTIERREZIA 91 a uJ = uJ = 18sded402.. - 92 15 27 24 19.02 ye 1% Oh: 4351208 5.7 WYOMING S14” IDG A0T 10-14 ~ 1.06 8.5 W422 1b 220° 138 7.4 NEBRASKA 3224 12.64.05 11-14 AT 6.1 209+ 2 17-24 1.50 7.2 3228 12.9 + .03 12-15 .39 3.0 18.8 + .2 15 - 23 1.78 9.2 KANS. 3292 12.4+ 08 1.14 1.20 9.7 194+ .2 15-22 1.25 6.4 COLORADO 3261* 14.7+.06 13-16 .83 5.6 20.9.8 18-24 214/012 $262) 42.0 05. 14 = 18 .69 5.7 19,6 .2 — 1622 — 1:26 6.6 UTAH 3136 =: 10.8 + .05 o=18 a2 6.7 172 ed) 1G 20° 26108 6.1 3145 180+ .05 12-15 64 4.9 173-258 — 1620 tae 8.2 SEO 41,7 08! 10 15 .37 2.2 18.0 2 © 96 2:20 “iG 6.4 3151 170 1.8 168i 20". «11:28 7.2 3258 142+ .06 18-16 18 5.5 19.2+.2 17-22 1.45 7.6 8260 I eT aia 0k 8.1 17.8+.2 16-20 1.15 6.5 OKLAHOMA $289 12.94.05 12-15 .68 5.3 17.84. 16-205. 14 6.4 2200 5455-1 04.5, 18 2.15. 52 3.9 136 34 16 22) i 4 o0 6.5 TEX. S212" 18)4 4072 11. 16 .93 6.9 18:9. 63°, [je<22. 5/208 eee StL T1251 05, 3 A 18 .67 5.5 6 + 2. Abe 20 tat 6.3 8273 13.7-+ .05 18-16 5.3 184 2° 16 2. 2 BS 6.4 8280 11.8+.05 11-13 74 6.3 17.2 dee Rome 6.0 3282* 12.2 + .04 it 14 55 4.5 18.2 + .1 17 - 20 1,02 5.6 3283 12.2 + .05 Iho tg 71 5.8 16.6 + .3 4-19 2.25 18.6 3286 18.24.05 12-15 64 4.8 WG es 2 16 20; 24.08 6.1 S287. 120 +05 = 1 4d 72 6.0 18:6 oe 16 oT ee 7.3 NE B1GOF 18.9 08 1 17 18 8.8 219+ .2 19-25 1.62 6.9 8269... 12.5 + .06- . 11-14 .86 6.9 16.0. 8 FIR ASL i Les 7.4 S200) 16 A ee ae eG ARIZONA 2792* 14.0 + .07 12-15 1.01 1.0 20.4 + .2 18 - 23 1.67 8.2 2IDSY 12.9 Sb 1s Ak 807 oe 19.2 + .2 16 - 21 1.08 5.6 ZIRE 18S S06 1a ae 36 6.4 AA B20 1s ea AS 6.9 2802* 19.8 +.2 17-28 ., 1.24 6.3 *Plants polyploid (n = 8) The low values for the coefficient of variation are due to the fact that the samples are from a single plant. No geogr aphi- cal pattern could be detected in the size variation of these two characters. INFRASPECIFIC VARIATION IN GUTIERREZIA 93 FLOWERING PERIOD When grown under uniform conditions in Berkeley, plants from different localities were observed to flower at different times (Solbrig, 1960). Similar observations were made again with plants grown at Cambridge. There is a gradient with plants from northern localities flowering earlier and usually being through the flowering period earlier than plants from farther south. For example, plants from Sheri- dan, Wyoming (3252) started blooming in Cambridge May 15; those of Dubois, Idaho (3254), a locality some 300 miles west but from about the same latitude, started blooming May 22, a week later; on the other hand the first date for the population from south of Pocatello, Idaho (3255), some 120 miles south from the previous populations, was June 10, about two weeks later than the second, and almost a month later than the first. A record was made of the dates on which flowering ma- terial of Gutierrezia sarothrae has been collected, on the basis of herbarium specimens. The results are shown graphi- cally in fig. 10. The investigation shows that there is no appreciable difference in first flowering dates but a definite trend for plants from northern localities to finish earlier was found. It is not too surprising to find this cline, which can almost be expected on the basis of the climatological JUNE | JULY | AUGUST ISEPTEMBER OCTOBER _|NOV. N.DAKOTA MONTANA . IDAHO a . : iu e S.DAKOTA WYOMING oes OREGON oe NEBRASKA bc KANSAS 363 OEE PO COLORADO ast de eo gh re : ; Le i 000 +00 ov coven esos | NEVADA pe Fe ) * OKLAHOMA = | ve | XAS o—_______+__»___err00-0- 0-9-0090 0- 9-0 —@ ARIZONA | gee Fic. 10. Dates of blooming of populations of G. sarothrae. For further details see text. 94 OTTO T. SOLBRIG factors of the environment (see under discussion). The earlier frosts in the northern area, and the shorter frost free period, diminish the growing season considerably, while the usually earlier summer rains in the more boreal areas apparently speed up the growth cycle. The earlier flowering dates of these plants under uniform conditions in culture as compared to the field are probably due to a faster rate of growth under the good conditions of the greenhouse or ex- perimental field. The earliest date of blooming of wild popu- lations throughout the range is most likely dependent on the date of the summer rains, a factor which fluctuates some- what every year. Comparisons would be more meaningful if they were restricted to plants collected in the same year, which is not the case with the data of fig. 10. Field observa- tions during the summer of 1961, made it very clear that there was this cline. Plants in North Dakota and Montana were in full bloom and forming seed at the end of August, while populations from Texas were just beginning to bloom in middle September. The variation in plant size observed, and also the increase in stature of the plants from north to south, is most likely another expression of the shorter growing season, and an adaptation to these conditions. POLLINATION? In a previous paper concerning the North American species of Gutierrezia (Solbrig, 1960) I put forward the idea that new populations of Gutierrezia are probably initiated by a few seeds blown in from surrounding populations and that under favorable conditions the population might expand from these few pioneers to a few hundred plants. In such a situation the genotypes of the first members of the colony will be determined strictly by chance from the genotypes present in the neighboring populations. The probability of a certain genotype being selected is in direct proportion of its frequency in the area and the distance to the new avail- able site. This is part of the process known as random "aE want i eapeaas was appreciation for the help received from Drs, P. D. Hurd, Jt-» : Linsley, J. W. McSwain and R. . Painter, in aecquainting me with the literature and some of the problems of pollinators as viewed by entomologists. I also want to thank especially W. H Anderson, K. V. Krombein, E. G. Linsley, C. F. W. Muesbeck, R. H. Painter, C. Sabrosky, T; J. Spilman and P. H. Timberlake for identifying my collections of insects, INFRASPECIFIC VARIATION IN GUTIERREZIA 95 genetic drift (Wright 1949; Dobzhansky, 1953), also known by the better designation of “founder principle’ (Mayr, 1942). Another postulate was that a combination of the founder principle and selection by oligolectic bees could explain the uniformity within populations of Gutierrezia, and the differences between populations, assuming that no drastic differences existed in the selection pressures between different populations. In order to test the role of pollinators in this scheme collections and observations of insects were made in a few different localities scattered throughout the range of G. sarothrae. The list of the particular pollinators collected is TABLE 8. POLLINATING INSECTS COLLECTED ON G. SAROTHRAE POPULA- TIONS, 1961 Pop. 3244. 7 mi. south of Dickinson, North Dakota. Aug. 27, 10 AM. Anastoechus barbatus O.S.; Peleteria aff. clara Curran; undet. Melo- ideae. Pop. 3245. 12 mi. north of Belfield, North Dakota. Aug. 28, 10:30 AM. Peleteria clara Curran; Poecilanthrax alpha (O.S.); P. willis- tonu (Coq.); Anastoechus barbatus O.S.; Cylindromyia californica (Big.) ; Villa eke Painter; Toscaute pennsylvanica (De Geer). Pop. 3246. Bridge over Little Missouri river and Hwy. 22, North Dakota. Aug. 28,2 PM. Anastoechus barbatus O.S.; Colletes sp.; Villa flavocostalis Painter. Pop. 3248. 2 mi. S. of Glendive on Hwy. 10, Montana. Aug. 29, 1 PM. Poecilanthrax alpha (0.S.) ; Villa flavocostalis Painter Pop. 3251. 5 mi. S. of Custer, Montana. Aug. 30,10 AM. Melissodes sp.; Colletes sp.; Poecilanthrax willistonii (Coq.); Peleteria malleola (Big.); Paradidyma singularis (Tns.); Chelonus sericeus (Say); Sronndyhoris apache Boh. Pop. 3265. 5 mi. south of Trinidad, Colorado. Sept. 8, 12 noon. Epicauta sp.; Chauliognatus lewisii Croth. Pop. 3269. 20 mi. north of Roswell on Hwy. 285, New Mexico. Sept. 10, 12 noon. Calliopsis coloratipes Ckll.; Bucercia-sp.; ; Ex lopsis solanii Ckll.; Melissodes sp.; Chauloghaihus lewisti Croth. shown in table 8. A series of Diptera, Hymenoptera and Coleoptera are involved. In addition table 9 shows a few additional records taken from the literature. Of the major groups of pollinators only the Lepidoptera are absent. Ob- servations were made at different times of day, and only insects observed visiting the heads of Gutierrezia sarothrae in medium to large numbers and feeding on pollen or nectar were collected, it was hoped that in this way occasional random visitors would not be picked up. 96 OTTO T. SOLBRIG At the time of flowering of Gutierrezia sarothrae, late summer and fall, that is from about August to late October or early November, depending upon the latitude blooming ordinarily begins first and finishes earlier in the northern areas, there are not many other plants in bloom in the scrub- land, juniper-pifon pine association, or semi-desert areas where G. sarothrae is found. The other plants in bloom at the time are for the most part yellow-flowered shrubby Com- positae very similar in their inflorescence characters to G. sarothrae. Common genera in bloom at the time are Chryso- soled bpicunctsait Grindelia, Tetradymia, Solidago, and ot The bene insects fall into two broad categories. The first group is the more abundant and is the group to which the bees (Hymnoptera) and bee-flies (Diptera-bombilids) belong. The behavior of these insects while they feed is to alight on a head inserting their tongues into one or several flowers and then to fly to another head, usually of another plant, where they repeat the process. They seem to be effec- tive pollinators and are mostly involved in cross-pollination. They inflict no visible damage to the flowers. The second group is that of the Coleoptera, particularly such genera as Epicauta, Chauliognatus and Crossidius. These insects feed on pollen, but instead of flying from flower to flower, they tend to craw] all over the plants, concentrating sometimes in large numbers (particularly Epicauta). They TABLE 9. INSECTS REPORTED FEEDING ON FLOWERS OF G. SAROTHRAE® Colletes laticinctus Timberlake; C. intermixtus Swenk; C. phaceliae Cockerell; Heriades variolosa Carivkeds (Cresson); Crossidius wick- hami Casey; C. jocosus (Horn) ; C. allgewahri Le Conte; Poecilanthrax sackenti sackenti (Coq.); Melissodes sabinensis nubila LaBerge; M. tepida Cresson; M. tepida timberlakei Cockerell; M. tesellata LaBerge; taken from the following works: Hurd, P. D, Jr. and C. D. Michener, “The ade Bees of California (Hymenoptera: Megachilidae)”, Bull. Calif. Insect Survey 3: 1-247, 1955; rge, W. E. “A Revision of the the Genus moulin in North and Central America (H menoptera, Apidae)’’ Parts I-10], niv sas Relenee Bull, 38-42: 1956-61; Linsley, E. G. ‘Host Relationships in the genus Crossidius’. Jour, Kansas Ent. Soc. 30: 83-89. 1957. Painter, R. H. and J. Halt, “A m ingraph of the Genus Poecilanthrax (Diptera: Bombe)” B PF. t] e Colletidae)”” Univ. Kansas Science Bull. 36: et 1954; Timber! ake, P. of the genus Perdita F. Smith, with special reference to the fauna of the Pacific Coast (Sihietoxbean: Y ealaore Parts I-V. Univ. Calif. Publ. Ent., 1954-62 INFRASPECIFIC VARIATION IN GUTIERREZIA 97 M. plumosa LaBerge; M. lustra LaBerge; M. stearnsi Cockerell; M. agilis Cresson; M. perlusa Cockerell; M. menuachus Cresson; M. semi- lupina Cockerell; M. bimatris LaBerge; M. fasciatella LaBerge; M. coreopsis Robertson; M. montana Cresson; M. tristis Cockerell; M. robustior Cockerell; M. hurdii LaBerge; M. pallidisignata Cockerell; M. subagilis Cockerell; M. limbus LaBerge; M. verbesinarum Cocke- rell; M. lutulenta LaBerge; M. utahensis LaBerge; M. Apne Laber ge; M. velutina (Cockerell); M. appressa LaBer sticta Cockerell; M. paulula LaBerge; M. melanura (Cockerell) ; M. personatella Cockérell : M. sabinensis nubila LaBerge; M. tessellata LaBerge; Perdita semicrocea Cockerell; P. austini Cockerell; P. stott- leri stottleri Cockerell; P. s. flavida Swenk & Cockerell; P. ericameriae Timberlake; P. oregonensis Timberlake; P. parilis Timberlake; P. luteola Cockerell; P. rhodwra Cockerell; P. gutierreziae Gockerell: P; melastoma Swenk & Cockerell; P. rectangulata Cockerell; P. phymatae Cockerell; P. apacheorum Timberlake. are most effective in performing self-pollination (since they tend to carry pollen from one head to another in the same p'ant). In addition, they also do some damage to the flowers since on occasion they seem to feed on petals, styles, and anthers as well as pollen. In addition to the insects mentioned, flowers of Gutierrezia sarothrae are also visited by parasitic wasps, apparently in the hope of ambushing their victims; diptera which para- sitize the ovaries of G. sarothrae; and many other kinds of insects use the capitula as mating places. Not much if any pollination is effected by these visitors. Although some species (such as members of the genus Crossidius) are larval parasites of roots of Gutierrezia and therefore spend their entire life cycle in association with this Species, such a close relationship appears to be rare. Also, it has been reported that some of the bees (Perdita) are oligolectic to Gutierrezia in their feeding habits. This seems to be an exception, also. As a rule, a spectrum of various diptera, hymenoptera, and coleoptera feed as adults on Guti- errezia and other plants in bloom at the time. They rely on what plants are available, and, according to the literature, Gutierrezia is not considered to be a delicacy by insects, other plants apparently being preferred over it. None of the pollinators collected had exactly the same geographical dis- tribution as G. sarothrae, although some, such as Poecilan- thrax willistonii, come quite close (Painter & Hall, 1960). Given the particular dispersed pattern of Gutierrezia where dense populations are often separated by considerable 98 OTTO T. SOLBRIG distances from each other, it is fairly safe to say that the amount of pollen carried from one population to the next is small, and probably is at zero or nearly so when the distance between populations is more than a mile. This situation with respect to the pollinating insects to a certain degree rein- forces the isolation between populations, while the non- specificity both of the plant to the pollinators and vice-versa, to a certain extent assures pollination for the plant when it invades a new area. In summary, Gutierrezia does not require a specific polli- nator, pollination being effected by many different insects. The lack of dependence on a particular insect is due probably to the advantage of the species, particularly when it invades new areas. Furthermore, pollinators tend to reinforce the physical isolation of populations, but effect a fairly active transport of pollen within the population. This is facilitated, no doubt, by the almost universal dense aggregation of indi- vidual plants in populations of G. sarothrae. Exchange of genes between populations is, therefore, mostly restricted to seed dispersal. Finally, the thesis that insect pollinators might reinforce differences between populations by selecting a particular phenotype is not tenable, because of the large and diverse number of pollinators. ENVIRONMENTAL FACTORS A very important element in the diversity of morphologi- cal forms encountered in G. sarothrae is the wide range of environmental factors found over the area in which this species grows. The nature of this study did not permit a detailed study of the microenvironment, but the broad fea- tures of the environment are illuminating by themselves. Temperature varies over the range of G. sarothrae be- tween an annual mean of 40° F to about 70° F. The mean temperatures in the months of May through September fluc- tuate from 40° to 70° F in North Dakota, and from 60° to 80° F in Texas. July average minimum temperatures are between 40° and 70° in the whole zone from North Dakota to Texas, and July average maximum ranges between 70° and 100° F. Normal summer temperatures above 68° are present in North Dakota only between July 1 and August 15; in Utah from the middle of June to the end of August; in California from the middle of May to the middle of Novem- INFRASPECIFIC VARIATION IN GUTIERREZIA 99 DICKINSON, POCATELLO, N. DAKOTA IDAHO AOS 43 Ax Z UTAH coLo, 43 Vivo 43 BRYCE CANYON, | GRAND JUNCTION ty LLL AAS J ff 7 arpa SID f Ai LPI LIS. Af, SAN DIEGO, CALIF, F Fic. 11. Main climatological data of six sites sustaining populations of G. ssrothrae in the immediate vicinity. Curve = mean temperature; hatched area = mean rainfall; ta arrow = frost free period. Area above 43° line is period of rapid growth. Da taken from the U. S. Weather Bureau. ber. The rapid growing season (mean temperature above 43° F) in North Dakota starts in May and lasts through September ; in Utah it starts in March and lasts through the middle of November; in California, Arizona and Texas tem- peratures above 43° F last throughout the year. The num- ber of frost free days varies also quite drastically ranging from only about 100 days in Montana to about 300 in Cali- fornia. 100 OTTO T. SOLBRIG The region where G. sarothrae grows has on the average an annual rainfall of not more than 20 inches but not less than 8 inches. Average July air humidity is less than 40% (noon and 8 PM data) in all the area but California where it is 60-70%. Monthly average figures for evaporation dur- ing the growing months of May through September vary from 5 to 12 inches; total normal average evaporation from May through October is 35-70 inches (in pans). The ratio of precipitation to evaporation for the normal frost free season varies from 0 to 20% ; 90% or more of the years for which records are available had less precipitation than evap- oration. Another variable is day length. The longest day in June where the North Dakota populations grow is 16 hours; the longest day in June for the Texas populations is 14 hours; the shortest winter day in Texas is 1014 hours, in North Dakota 8 hours, 20 minutes. It is evident that a marked change occurs from north to south. To summarize, G. sarothrae grows largely in an area with short, warm but not hot summers and cold winters (Mon- tana, Idaho, North and South Dakota, Wyoming, Utah, W. Colorado, N. E. Arizona and N. W. New Mexico) ; in this area it is widespread but not too abundant, particularly in the more northerly parts. In also grows in areas with hot summers and cold winters (Arizona, New Mexico, Nebraska, Kansas and northern Oklahoma) and in areas with hot sum- mers and mild winters (California, Texas and southern Oklahoma). The latter areas are less extensive than the first, but in them the species is in general more abundant. Where G. sarothrae grows it is very dry, as measured by air humidi- ty, precipitation, and the precipitation/evaporation ratio. There is also a very marked gradient from north to south in et of the growing season, frost free days and length of ay. The terrain where the plant is found varies, but in general it does not grow at elevations above 2,000 m. The soil is loose, rocky, gravelly or sandy, with a high pH (7-9), and a low concentration of nitrates (Solbrig, 1960). No precise data on microenvironmental factors are avail- able, but observations show that the plants tend to favor areas of slightly higher moisture (protected slopes, flood beds, depressions, etc.) than adjacent sites, and also where INFRASPECIFIC VARIATION IN GUTIERREZIA 101 soils are disturbed (soil slides, erosion gullies, etc.). No precise pattern emerges from these data aside from the generalities already enunciated. But one is impressed by the great amount of variability and by the extremes of environment the species is capable of withstanding. Al- though there is not too much obvious direct correlation between the morphological variation and the environmental conditions indicated, there is a parallelism in the enormous amount of variation of both aspects that is worth noting. DISCUSSION VARIATION AND ECO-GEOGRAPHIC DIFFERENTIATION From the foregoing presentation it is clear that we are dealing with a very variable group of plants. The variation can best be considered as that within and that between popu- lations. An understanding of interpopulation variability is important for a better taxonomy. Speculation on the past history of the group, the general model of evolution of the complex, and how general or special a case is represented by G. sarothrae are all items of considerable interest. Gutierrezia is not unique, neither in its wide tolerances nor in its high degree of plasticity. Many, if not most, plants have these characteristics to a greater or lesser degree. Transplant studies have shown that populations of such species vary in their specific tolerances to environmental factors and also in their morphology, and that these different responses are genetically fixed (Clausen, Keck & Hiesey, 1940; Bocher, 1963, etc.). The transplant studies at Berkeley and Cambridge showed that similar features are character- istic of G. sarothrae. Morphological variations between pop- ulations of a species correlated with environmental factors have been referred to as ecotypes. Turesson created the term ecotype, as an “ecological unit to cover the product arising as a result of the genotypical response of an ecospecies to a particular habitat” (Turesson, 1922). The fundamental idea embodied in the concept of ecotype is that the genotype best adapted to each particular environmental and edaphic situ- ation is selected in each instance, and hence the species — “ecospecies” in the sense of Turesson — is formed by a series of such types. Although initially the term was defined strict- ly on an ecological basis, the definition was later modified by Turesson (1929) to include genetical criteria, and was simi- o407 3260 INFRASPECIFIC VARIATION IN GUTIERREZIA 103 larly interpreted by Clausen, Keck & Hiesey (1940; Clausen, 1952). The ecotype concept presupposes a mosaic-like eco- logical situation formed by patches of more or less discrete and uniform conditions both in time and space to which each ecotype is adapted, and envisages a situation, where a certain phenotype and/or genotype or group of similar phenotypes and/or genotypes, which presumably are those best fit to a given place, are consistently being selected over any others. Gregor (1939, 1944) working with Plantago maritima along the coastal areas of Great Britain found that when the habitats in which this species grows are subjectively arranged in an ecological sequence, the populations within these habitats show a continuous variation between extremes in growth habit. A situation such as this has been designated by Huxley (1938) as a “cline.” Clausen (1952) feels that clines are accidents of sampling and/or the result of field rather than garden measurements, and that the term cline should be used only when referring to individual characters. However, it is true that environmental situations form a highly complicated picture in nearly every instance. The more divergent two habitats occupied by a particular species, the easier it will be to detect ecotypes, and the more gradual the change from one into another, the easier it will be to see clines. To a certain extent both concepts can be considered accidents of sampling. But even if an environment com- posed of discrete patches of uniform conditions in space and time did exist, no selection for a “type” adapted to these conditions would be possible in the presence of much gene flow from adjacent populations. In the case of G. sarothrae, if plants from populations from opposite extremes of ecological conditions are con- sidered, all the morphological and physiological characteris- tics necessary to place them into bona fide ecotypes can be detected ; if a series of intermediate situations is added, and particularly if these are arranged in a subjective order, arguments for recognizing a cline are present. But if the totality of the variation is studied, it will be seen that the pattern is a highly complicated one, probably related to a g. 12. Four week old seedlings grown under uniform conditions. Largest, smallest a from California. Note discrepancy in growth rate among the various populations 104 OTTO T. SOLBRIG highly complicated environment, as we have seen. The con- cepts of ecotype and cline are therefore not applicable in this particular case. VARIATION IN THE CENTER AND THE PERIPHERY OF THE RANGE In most species of animals, populations toward the geo- graphical center of distribution of the species are denser, that is, the individuals are spaced closer; they are more variable; and populations are more contiguous to each other (Mayr, 1963). The usual explanation is that populations in the center of distribution are better adapted ecologically and therefore more subniches are occupied, hence the greater variability (Mayr, 1945, 1963). Also two other forces sup- posedly play an important role in such situations: density dependent factors and gene flow. The first of these will favor a diversification into subniches; the second will contribute greatly to increase the amount of variability. The data for G. sarothrae do not show such a well marked pattern. If we take the coefficient of variability as a measure of the amount of variability present, populations in the geographical center, Colorado, Utah and New Mexico, are not significantly more variable than those around them; on the contrary, if any- thing they are less variable. The explanation for this is no doubt complex, but an important factor is the type of distri- bution. In animals, increased density of individuals in a population will result in losses of some individuals to neigh- boring populations with a resulting increase of gene flow. In plants, no direct loss of individuals to neighboring popu- lations is possible, and increase in the production of seeds decreases per individual rather than increases after a certain density is reached, due to the smaller size of the plants, as experiments with crop plants have shown. Consequently there is never such a population pressure in plants as there is in animals. The other important component usually pres- ent towards the center of the geographical range of species, that is more contiguous populations, is not present in the case of G. sarothrae. As indicated above, the pattern of groups of small populations in isolated pockets is fairly uniform throughout the range of the species. Towards the center of the range in Utah, Colorado and New Mexico, a rugged ter- rain is combined with very dry atmospheric conditions, S° that rather than being more hospitable, the center is perhaps INFRASPECIFIC VARIATION IN GUTIERREZIA 105 less so. As a result of the even and wide spacing of popula- tions throughout the range of the species gene flow is low and more or less uniform throughout the area of distribution. The similar degree of variability of the populations in the center and towards the periphery might be accounted for this way. VARIATION AND ADAPTATIVE RADIATION The relation of topography to patterns of evolution and speciation is an interesting one, but also a highly complicated one. A puzzling aspect is the lack of radiation into new habi- tats of the North American species compared to the South American representatives of the genus. Two of the North American species are annual, and they have diverged mor- phologically to a certain extent, particularly in the number of florets in a head which has increased, and also in the loss of the already reduced pappus of the ligulate florets. Beyond that there has not been much change between the annual and perennial species. All of the six perennial species of North America are very close and, aside from chromosome number, the differences between them are of degree rather than of kind. The South American species on the other hand have radiated into a series of new and different niches. All South American species are perennial, but they vary from the cushion forming G. baccharoides which grows at altitudes of 3,000 m. in the Andes, and is only 5 cm. tall; to the low, creeping G. repens of the Aconquija Mts. also at altitudes of 3,000 m.; to the shrubby G. paniculata of central Chile, one m. or more in height with woody stems sometimes 2 to 3 cm. in diameter ; to G. gilliesii of central Argentina which is very similar to G. sarothrae but has white rather than yellow ligu- late flowers. The ecological variation is matched by the morphological variation, with changes in leaf shape and size, habit, woodiness, color of flower, etc. having taken place. Evidently either the South American species have had a longer evolutionary history, or their rate of evolution has been greater. Since the tribe Astereae is most probably of North American origin and furthermore since the close generic relatives of Gutierrezia are found only in North America, it does not seem likely that the genus has had a longer history in the southern lands. An explanation for the higher rate of evolution in South America can be found in 106 OTTO T. SOLBRIG the combination of available niches, higher recombination index due to polyploidy, and isolation. These elements have not been present in North America to the same degree as they have existed in South America. It is also possible that the extensive Pleistocene glaciation of North America played an important role, by pushing the species south and off the higher places in the mountains, but there is not enough infor- mation to elaborate on this idea. INCIPIENT AND SIBLING SPECIES Sibling species, morphologically identical but reproduc- tively isolated, are well known among animals (for a detailed discussion see Mayr, 1963). Among plants they are difficult to define, due to the more complex systems of self-incompati- bility present which assure outbreeding. This makes testing for reproductive isolation much more difficult. Furthermore, in plants there often are modes of reproductive isolation between species other than genetic sterility, such as ecologi- cal isolation. As an example, in a population a single indi- vidual may be unable to breed with as much as one-half of all the plants present, while on the other hand, the individu- als of two good sympatric species may be completely inter- fertile (Raven, 1962). A mechanism which as a rule produces total reproductive isolation and which is independent of the self-incompatibility mechanism is polyploidy. Furthermore polyploidy is rela- tively easy to detect. Allopolyploids combine the characters of the two parental species while autopolyploids are morphol- ogically and genetically similar qualitatively but not quan- titatively. Segmental allopolyploidy stands in between allo- and autopolyploidy. All three types of polyploids to a certain extent, but particularly autopolyploids, may give rise to sib- ling species. Autopolyploid plants can be readily obtained artificially, and have been observed to occur naturally. As & rule they are highly sterile, due to the formation of multi- valent associations at meiosis. For this reason fertile natural polyploids are usually assumed to be allopolyploids. Such an assumption has frequently been shown to be correct when the plants were investigated in detail. The taxonomic treatment of the diploids and tetraploids will depend to a certain extent on the species concept that is adopted. According to a definition based primarily oD INFRASPECIFIC VARIATION IN GUTIERREZIA 107 morphology, diploids and tetraploids when morphologically alike have to be treated as one species; on the other hand, if a definition based primarily on the degree of crossing in nature is adopted, the two different levels of ploidy are better treated as separate species, when cytogenetical considera- tions indicate that there is no effective exchange of genes between them as is the rule. The interpretation of particular situations is, nevertheless, more complex than this simple analysis would suggest, due to the high degree of variability of biological entities and the small amount of effective interbreeding between allopatric populations. On account of their lack of mobility and their method of dispersal, plants have to be able to adapt to a variety of microenvironmental conditions. This is particularly true for perennial plants which are exposed to a variety of seasonal changes. Under these conditions selection favors systems which are flexible enough to compensate phenotypically for the differences from the paternal environment which the progeny will encounter. In addition selection will also favor the fixation of characteristics which are of general adaptive advantage to the population. The tremendous variation present in species of plants, and the lack of a definite pattern, is often perplexing. In Gutierrezia sarothrae a pattern of variation and distribution is present which represents an intermediate situation between total isolation and free gene exchange between populations. Both distance and pollinators effectively prevent cross-breeding between populations but there still can be some gene exchange due to blowing in of fruits from neighboring areas. Certain groups of popula- tions, particularly those in California are completely isolated, and it is here that some of the extremes of variation are found, and it is also in California that the polyploid G. brac- teata, a closely related species morphologically distinct from G. sarothrae has evolved (Solbrig, 1960, and unpublished). The diploid populations can best be regarded as forming a large, polytypic species. The treatment of the polyploid pop- ulations is more problematic. The range and mean of the measurements of 850 diploid and 250 tetraploid plants of Gutierrezia sarothrae taken as a group are shown in table 10. There is a considerable amount of overlap in the range of diploid and tetraploids in each 108 OTTO T. SOLBRIG CHARACTER RANGE MEAN be EXAMINED KP 0.05 1.96) DIPLOID |TETRAPL} DIPLOID |TETRAPL WZ PLANT HEIGHT 4- 38 |}45- 28 | 10.93 | 10.13 | 0.942 HEIGHT |25-5.0/26-62| 3.77 | 3.56 | 0.075 INVOLUCRE | WIDTH |0.9-29| 11-27] 1.77 | 181 | 0.031 ba os ; FLOWER DISK |0-7]/0-7] 3.67] 390] 1.118 i =NUMBE i RAY |2-7|3-6]| 487| 449 /0.85I DISK |05-20/\06-20) 1.2% 0.076 PAPPUS 05-2.01 06-20 1.24 LSi LENGTH ) Ray |o2-12 |0413| 0.58] 073 | 0.333 fo POLLEN DIAMETER |9-16 | Il-17 |12.57 | 13.59 | 2.930 amp STOMA LENGTH /|14-24 /15- 25 | 18.13 | 19.97 | |.870 TABLE 10. Comparison between diploid and tetraploid plants. Data from all popula- tions have been added. (n = 250 tetraploids and 1350 diploids). . instance, and although the mean of the polyploid group tends to be greater, only pollen and stoma measurements show any appreciable mean difference. Significance tests show the difference to be significant only in the case of pollen diameter, at the 5% but not at the 1% level. On the basis of the char- acters analyzed, no plant taken at random can be placed with certainty as a diploid or as a tetraploid. The pattern does not differ appreciably if we compare the same characters on a population basis, since there is still quite an overlap between diploid and tetraploid populations. The amount of overlap and the pattern differs with the character under consideration (see under morphological analysis and eco-geographic differentiation). From the mor- phological analysis, it is clear that in the G. sarothrae com- plex no separation into morphological categories is possible. Since the diploid and tetraploid populations are most prob- ably reproductively isolated, they can be regarded as sibling species. It is impossible to know at present whether the tetraploids in this group are descendants from a common ancestor oF whether they had independent origins. Nevertheless, thelr INFRASPECIFIC VARIATION IN GUTIERREZIA 109 scattered distribution and morphological differences seem to favor an independent origin for at least some of the popu- lations. It is very possible that the Arizona populations, for example, are all of the same stock, but it is unlikely that the tetraploids from Grand Junction, Colorado, had the same origin. An independent origin could explain the differences between different tetraploids and between them and the diploids, particularly if we assume that the diploids of both small and large stature have independently given rise to tetraploid plants. We could easily have a tetraploid plant showing an increase in size over its small diploid progenitor, but still being smaller than some of the large diploid plants. The large tetraploid plants would of course have originated from large diploid plants. Most of the tetraploid populations are separated by considerable distances and consequently they are reproductively isolated from each other. The ques- tion arises as to whether these reproductively isolated popu- lations which probably had an independent origin from different diploid parents could reasonably be considered to be one species. It is possible that some or all of the polyploid populations may evolve further and become morphologically distinct, as has been the case with G. bracteata. In such an eventuality there is no doubt that they will form separate species. In the meantime, I think it is legitimate to consider all the polyploids and diploids as part of one species with the polyploids polyphyletically tied into it. The present situation is similar to the problem presented when in the course of time several populations cross a threshold of difference into what is considered as a new, more advanced species. The principal difference is that the tetraploid populations of G. sarothrae have crossed the threshold suddenly and not grad- ually. AUTOPOLYPLOIDY AND ALLOPOLYPLOIDY A more difficult problem is to determine whether the poly- ploids are allopolyploids, segmental allopolyploids or auto- polyploids. The absence of closely related diploid species and the morphological similarities between diploids and tetraploids almost certainly precludes interspecific allopoly- ploidy. On the other hand only the study of certain crosses can tell us if we are dealing with autopolyploidy or segmental allopolyploidy. Polyploidy is a rather common phenomenon in plants, and 110 OTTO T. SOLBRIG in Gutierrezia it is very widespread. Out of the 12 species of the approximately 20 in the genus for which we have chromosome counts, 9 (75% of those with counts), are poly- ploid. Morphological analysis of the chromosomes indicates no noticeable change in the structure of the basic karyotype in the polyploid species analyzed (Riidenberg & Solbrig, 1963). Unfortunately, no tetraploid population of G. saro- thrae has been available for a similar study, but the closely related G. bracteata (n=8 & 12), as well as the more remotely related species G. californica (n = 12) and the South American G. gayana (n = 28), essentially repeat the basic karyotype of diploid G. sarothrae. The “diploidization” — that is the acquisition of diploid pairing behavior — of these species has apparently not been dependent on any gross structural change in the chromosomes, and it is probable that a similar situation exists in tetraploid G. sarothrae. Autopolyploidy is a rare phenomenon and some authors have completely ruled it out as a factor in speciation. But autopolyploidy when not accompanied by gross morphologi- cal variation is detectable only when the chromosome num- bers of a whole group of plants are investigated systematic- ally. In such a survey of the tribe Astereae of the Com- positae (Raven et al. 1960; Solbrig et al. 1964) a number of species were found to have polyploid populations. Al- though no detailed study has been made, it is possible that some of these polyploids are autopolyploids. In this connec- tion, the importance of counting large numbers of popula- tions of a species should be emphasized. SUMMARY The cytology, the morphological variation of 11 characters, and the pollinators, of 53 populations of Gutierrezia saro- thrae have been studied. Diploid and tetraploid populations have been discovered and their morphological variability compared. The geographical distribution and ecology of the species have also been investigated. The pattern which emerges is that of a very variable species, covering a large territory and growing under a series of very different con- ditions. No clear ecotypes or clines could be discovered, and this is assumed to result from the mosaic environment in which the species grows. No absolute qualitative or quanti- tative morphological deviation exists between the diploid and INFRASPECIFIC VARIATION IN GUTIERREZIA 111 tetraploid populations, which may be considered as sibling populations. No definite conclusion can be made as to the exact origin of the tetraploids. LITERATURE CITED AVERS, CHARLOTTE J. 1954. Chromosome Behavior in Fertile Triploid Aster Hybrids. Genetics 39: 117-126. 1957. Fertile Hybrids Derived from a Wide Species Cross in Aster. Evolution 11: 482-486. Bécuer, T. W. 1963. The Study of Ecotypical Variation in Relation to ing “afenar arian Morphology. Reg. Veg. 27: 10-16. CELARIER, R. P. AND J. L. MEHRA. 1958. Determination of Polyploidy rom Macturiuie Specimens. Rhodora 60: 89-97. CLAUSEN, J. 1952. Stages in the Evolution of Plant Species. 205 pp. thaca. , D. D. Keck anp W. M. Hiesey, 1940. Experimental Studies on the Nature of Species. I. Effect of Varied Environ- ments on Western North American Plants. Carnegie Inst. Publ. No. 520. DoszHANSKY, T. 1953. Genetics and the Origin of Species. 3rd. Ed., 364 pp. New York. Grecor, J. W. 1939. Experimental Taxonomy. IV. Population Dif- ferentiation in North American and European Sea Plantains Allied to Plantago maritima L. New Phytol. 38: 293-322. The Ecotype. Biol. Rev. 19: 20-30. UXLEY, J. S. 1938. Clines: an Auxiliary Method in Taxonomy. Nature 142: 219. JOHNSON, B. L. 1945. Natural Hybrids between Oryzopsis hymenoides and Several Species of Stipa. Am. Jour. Bot. 32: 599-608. 1960. Natural Hybrids between Oryzopsis and Stipa. ,, Oryzopsis hymenoides X Stipa speciosa. Am. Jour. Bot. 47: 736-742. 1962. Natural Hybrids between Oryzopsis and Stipa. Il. Oryzopsis hymenoides X Stipa nevadensis. Am. Jour. Bot. 49: 540-546. 963. Natural Hybrids between Oryzopsis and Stipa. IIT. bso hymenoides X Stipa pinetorum. Am. Jour. Bot. 50: 228-234. LERNER, M. 1954. ergy Homeostasis. 134 pp. Edinburgh. LEWONTHIN, R. C. ano L. C. DuNN. 1960. The Evolutionary Dy- namics of a a lard in the House Mouse. Genetics 45: 705- 722. LivKvist, B. 1956. The Cardamine pratensis complex. Outlines of Its Cytogenetics and Taxonomy. Symb. Bot. Upsal. 14: 1-131. Mayr, E. 1942. Systematics and the Origin of Species. 334 pp. ae 1945. Symposium on Age of the Distribution Pattern of Gene Arrangements in Drosophila pseudoobscura. Introduction and Some Evidence in Favor of a Recent Date. Lloydia 8: 69-83. 112 OTTO T. SOLBRIG ————.. 1968. Animal Species and Evolution. 797 pp. Cambridge. PAINTER, R. AND J. HALL, 1960. A Monograph of the Genus Poecil- anthrax (Diptera: Bombyliidae). Kansas State Univ., Agric, Exp. St., Tech. Bull. 106: 1-132. RAVEN, P. H. 1962. The Systematics of Oenothera subgenus Chylis- mia. Univ. Calif. Publ. Bot. 34: 1-122. Raven, P. H., O. T. Soxpric, D. W. KyHos AND R. SNow, 1960. Chromosome Numbers in Compositae. I, Astereae. Am. Jour. Bot. 47: 124-132. Ritty, R. 1960. The Meiotic Behaviour, Fertility and Stability of Wheat-rye Chromosome Addition Lines. Heredity 14: 89-100. AND V. CHAPMAN, 1958. Genetic Control of the Cytologically Diploid Behavior of Hexaploid Wheat. Nature 182: 713-715. Rotuins, R. C. 1963. The Evolution and Systematics of Leaven- worthia (Cruciferae). Contr. Gray Herb. 192: 1-98. RUDENBERG, L. AND O. T. SoLBRIG. 1963. Chromosome Number and Morphology in the Genus Gutierrezia. Phyton 11: 199-204. SOLBRIG, O. T. 1960. Cytotaxonomic and Evolutionary Studies in the oO American Species of Gutierrezia (Compositae). Contr. Gray Herb. 188: 1-63. RS te RAVEN, L. ANDERSON, L. RipENBERG AND D. KyHos. 1964. Chromosome Numbers in Compositae. IV. Astereae I. Am. Jour. Bot. (in press). STEBBINS, G. L. 1950. Variation and Evolution in plants. 645 pp., New York. TURESSON, G. 1922. The Species and the Variety as Ecological Units. Hereditas 3: 100-113. 1929. Zur Natur und Begrenzung der Arteneinheiten. Hereditas 12: 323-334, WADDINGTON, C. H. 1957. The Strategy of the Genes. 262 pp. London. WRIGHT, S. 1949. Population Structure in Evolution. Proc. Am. Phil. Soc. 93: 471-478, APPENDIX LOCALITIES OF G. sarothrae POPULATIONS STUDIED 2758. California. 2 mi. W. of Temecula on Hwy. 78. In a slight depression formed by a dry creek bed. Plants rather sparse. 2766. California. 8.5 mi. E. of Chula Vista on road to Dulzura, just below Otay Reservoir Dam. Near roadside. 2769. California. 1.6 mi. E. of Rancho Santa Fe on road to Escondido. In old wheat field, growing half way up a hill near a pond where the land had not been cultivated. 2792. Arizona. 19.6 mi. N. of Roosevelt Dam. Common at roadside and fields in slight depression in the valley, 2793. Arizona. % mi. N. of Jct. Payson-Phoenix road with Hwy. 488. ommon. 2794. Arizona. 5 mi. N. of Payson. Very common under a good stand of Juniper. 2801. Arizona. 12.6 mi. W. of Ash Fork. Transition zone between 2802. 2805. 2905. 2911. 2912. 3134. 3136. 3141. 3142. 3143. 3145. 3150. 3151. 3153. 3161. 3169. 3212. 3215. 3224. 3228. 3231. INFRASPECIFIC VARIATION IN GUTIERREZIA 113 open Pinus ponderosa forest and Pinus cembroides-Juniperus communis complex. Locally abundant. Arizona. 9.8 mi. W. of Seligman. In a depression muddy due to the recent rains. The plants restricted to the depression and surrounding area, although they are present in the rocky mts. surrounding the area. Arizona. 0.8 mi. W. of Hyde Park. Under Juniperus communis association. Growing sympatrically with Gutierrezia micro- cephala. No sign of hybridization. Oregon. 2.5 mi. E. of Grant re bere in John Day, on an old dirt road. Abundant in dry Idaho. 8.7 mi. S. of Hansen. Foot a rocky hills with sandy soil covered with sagebrush. Idaho. 8.7 mi. S. of Hansen. At roadside. Plants 500 yards away from pop. 2911, but in wetter spot Wyoming. 17 mi. SE of Mountainview on dirt road to Lonetree. Small population near roadside; soil whitish and sandy. Utah. 16 mi. SE of Jct. Hwy. 44 and Diamond Mt. Rd., at junction with Diamond Valley Creek. On a hillside, under sage- brush. Utah. Western Outskirts of Salt Lake City in abandoned field on Hwy. 50. Aiteona, Hwy. 91, 2 mi. S. of Utah-Arizona border. Common in spots near roadside. Utah. Hwy. 15, 9 mi. W. of Mt. Carmel Jct. Rocky hillside under Juniperus association. Utah. Red Canyon, just before Bryce Canyon Nat. Park. Mix- ture of Pinus ponderosa and Juniperus. Gutierrezia common along dry wash. Utah. Hwy. 54, 7 mi. S. of Boulder. Small pop. under Juniperus, abundant in places. Soil white-sandy, loose. Utah. Hwy. 24, 15 mi. S. of Jct. with Hwy. 6. Sagebrush com- munity, sandy soil: Gutierrezia abundant in this spot, but absent in the area. ipsidcty Ea mi. W. of Hwy. 50, 22 mi. S. of Grand Jct. Common n lava Colorado "tS mi. E. of Durango on Hwy. 160. Roadside popula- na Mexico. 3 mi. S. of Jet. Hwy. 285 with 286. Juniper-Pine association. “pinay ies common in the area. Texas. 1 mi. W. of Van Horn on Hwy. 80. Sandy-clay soil. Plants very common in fields and at roadsides. New Mexico. 12 mi. S. of Glenwood, on Hwy. 260. Extraordi- narily large plants with glaucous stems. Nebraska. 2 mi. E. of Roscoe, 9 mi. E. of Ogallala. Growing on loose, sandy soil on bluffs, and in gullies; large population. Nebraska. 1 mi. N. of Lisco on road to Orlando. Sandy soil, abundant at roadside and in fields. Nebraska. 10 mi. S. of Chadron. On roadcut and little hill. Very dry sandstone; overdispersed here, rare elsewhere. 114 3233. 3236. 3238. 3243. 3245. 3246. 3248. 3249. 3250. 3252, 3254, 3255. 3256, 3258. 3259. 3260. 3261. 3262. 3263. 3264. 3265. 3266. 3269. 3270. 3271. OTTO T. SOLBRIG South Dakota. Hwy. 385, 7 sic S. of Oelvichs. In open field, growing in reg hag occasio South Dakota. Hwy. 16, 10 mi. on of Rapid City, 1 mi. E. of Piedmont. In ditch, South Dakota. Hwy. 16, 10 mi. N. of Rapid City, and about 1-2 mi. E. of Hwy. and pop. 3236. Overdispersed population in open field depression. North Dakota. 8 mi. W. of Dickinson. Uncommon in this area; overdispersed population in roadside dite North Dakota. 12 mi. N. of Bellfield on Given: 85. Growing in roadside ditch. North Dakota. Bridge over Little Missouri river and Hwy. 85. Badlands; small population. Montana. Hwy. 10, 2 mi. S. of Glendive. Roadside population. © Montana. Hwy. 10, 20 mi. S. of Glendive. Roadside population; fairly common along roads and ditches. Montana. 3 miles E. of Sanders. Growing on Hwy. embank- ment. Wyoming. West outskirts of Sheridan. Growing on rocky hill- side. Idaho. 12 mi. E. of Dubois, on dirt road to Kilgore. Growing on very weathered sii field with sagebrush; overdispersed locally, but generally ra Idaho. Hwy. 91, 10 mi. Ss. of Pocatello. Very si in large rocky field at both sides of road; plants very unifor Utah. Hwy. 91 between Logan ‘and Ogden, 29 mi. s. “of Ogden. Rocky summit; overdispersed population. Utah. 16 mi. N . of Vernal on Hwy. 44. On rocky overhang; very common in overdispersed population. Utah. Hwy. 134, 1% mi. S. of Talmadge. Common in wash. Utah. Hwy. 50, 4 mi. W. of Thompson. dances population along Hwy., growing on loose, sandy so Colorado. % mi. before W. entrance se Colorado Nat. Mon. In roadside field; sandy soil; Juniper association. Colorado. Colorado Nat. Mon., 1 mi. N. campground, Grow- ing on eroded red sandstone; common in area. Cslornsos Hwy. 50, 15 mi. W. of Salida, At roadside; locally very r. Gallas Dry hills W. of Fountain. Locally abundant. Colorado. 5 mi. S. of Trinidad on road to Raton. Juniper-pifion pine open association. Very common in fields and roadside. New Mexico. Gallinas canyon, 10 mi. W. of Las Vegas on Hwy. 65. On rocky cliffs and road embankments. New Mexico. Hwy. 285, 20 mi. N. of Roswell. In small wash, both at roadside and in fields; red, sandy soil; population very dense. New Mexico. 3 mi. S. of White City on Hwy. 62-180. Common in roadside and fields, loose sandy-clay soil. Texas. 23 mi. S. of White City, N. M. , on Hwy, 67-180. Growing on rocky, limestone soil; very common iy SFr oe 3278. 3280. 3282. 3283. 3284, 3286. 3287. 3288. 3289. 3290. 3291. 3292. INFRASPECIFIC VARIATION IN GUTIERREZIA 115 Las Hwy. 54, 30 mi. N. of Van Horn. In dry wash; soil andy. Pusan 7 mi. E. of Alpine on Hwy. 67, Overdispersed popula- tion; locally common. Texas. Hwy. 67, 6 mi. W. of McCamey. Roadside population. Texas. Hwy. 67, 12 mi. W. of San Angelo. Common at roadside and in fields. Texas. Hwy. 67, 9 mi. W. of San Angelo. On roadbank and fields; abundant. Texas. 5 mi. W. of Abilene State Park, on road leading to Hwy. 277. Around and under oaks and Junipers; overdispersed but not too abundant. Texas. Hwy. 254, 5 mi. W. of Craford, Terry Co. In sandy- rocky, eroded wash. Texas. 2 mi. N. of Childress. Growing in abandoned field; over- dispersed. Oklahoma. Hwy. 62, 11 mi. W. of Altus. On mezquite-cactus land; soil red, sandy; overdispersed and common in patches. Oklahoma. 4 mi. E. of Clinton, on Hwy. 66. On field next to road. Oklahoma. Hwy. 281, 9 mi. S. of Bouse Jct. On eroded red sandstone; in roadside cliffs and fields. Kansas. Hwy. 281, 10 mi. S. of Medicine Lodge. Rare locally; growing on eroded red sandstone in creek bed CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C. Rollins and Robert C. Foster NO. CXCIV THE FERNS OF PERU POLYPODIACEAE (DENNSTAEDTIEAE TO OLEANDREAE) vi By RoLLa TRYON Published by Ess THE GRAY HERBARIUM OF HARVARD UNIVERSITY 5 a CAMBRIDGE, MASS., U.S.A. 1964 CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C, Rollins and Robert C. Foster NO. CXCIV THE FERNS OF PERU POLYPODIACEAE (DENNSTAEDTIEAE TO OLEANDREAE) BY, RoutuA TRYON i S # SOC ~2o0 Published by THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U.S. A. Issued November 19, 1964 Introduction Polypodiaceae Tribe Tribe 5. 6. ih 8. 9. Tribe 10. Tribe 11. Index to names THE FERNS OF PERU CONTENTS Dennstaedtieae Cheilantheae PrTiiti oe Pterideae Vittarieae Davallieae eeeeeee Lindsaeeae eeeeenee Oleandreae seeeeenee THE FERNS OF PERU INTRODUCTION Nearly ten years ago I undertook the preparation of a manual of the Fern Flora of Peru with the encouragement and active support of the late Theodore K. Just of the Chi- cago Natural History Museum. This work has occupied a considerable portion of my time since then and although it is still far from complete I have decided to bring out the portion that has been finished. The principle reason for offering this partial account is the general need for modern literature on tropical American ferns and especially on the species-rich Andean region from Bolivia to Colombia. The present portion treats the species of seven tribes of the Polypodiaceae. These tribes are represented by 33 genera and 176 native species. This is probably about a third of the genera of Pteridophyta to be found in Peru and about a quarter of the species. Three introduced species are also treated and eight that are to be expected in Peru, one of these in an additional genus. The scope of the treatment and the method of study of each genus has been to some extent dictated by convenience. In the smaller genera I have considered all of the American material and the Peruvian species have been treated in relation to this study of broad scope. In such genera I have sometimes also included species related to the Peruvian ones, or those that may be expected to occur in Peru. In the larger genera, and especially where difficult problems were encountered, I have limited my study to the Peruvian species, although not to Peruvian material only. The generic de- scriptions have sometimes been limited, in some characters, to the Peruvian species. Each genus has been treated in a monographic manner but I have felt it necessary to accept tentative conclusions when a problem went too far beyond the scope of the immedi- ate study. I have tried to determine the type of each name and to see the holotype, when possible, or at least other authentic material. In most cases this effort has been suc- cessful; I have examined the holotype or an isotype of about 4 ROLLA TRYON 80 per cent of the basionyms and photographs or other authentic materials for an additional 10 per cent. I have made no attempt to include a full synonomy under each name, but have usually limited it to those names based on Peru material and the commonly used synonyms. In a few cases, where I have examined a type of an obscure name, I have included that also. The classification adopted is essentially that of Carl Chris- tensen as set forth in Verdoorn, Manual of Pteridology (1938). This has been modified in some places by more recent studies. I have not made a special study of generic problems but have followed a conservative course unless the need for a change and a proper solution have both been clear. I can not accept the present tendency to divide the Polypodi- aceae into many families. The category of family loses its utility when restricted to groups of more or less closely related genera; such groups are better treated as subfamilies or tribes. The principle collections of Peruvian ferns are in the Herbarium of the Chicago Natural History Museum, the United States National Herbarium (Smithsonian Institution) and the Gray Herbarium and it is on these collections that my work is largely based. Other impor- tant collections are in the Herbarium of the New York Botanical Garden, the Herbarium of the University of California (Berkeley), the Herbarium of the Missouri Botanical Garden and the Herbario San Marcos, Museo de Historia Natural, Universidad Nacional Mayor de San Marcos, and I have also consulted these. Type specimens, especial- ly, have been studied at the Botanical Institute of the Academy of Sciences, U. S. S. R., Leningrad, the Paleobotanical Department, Natur- historiska Riksmuseum, Stockholm, the Botanisches Museum, Berlin- Dahlem, the Muséum National d’Histoire Naturelle, Paris, Royal Botanic Gardens, Kew and the British Museum (Natural History). 1 am indebted to the officers of all of these institutions for the loan of materials or for the privileges extended during periods of stud I have been aided in this work by many persons. Especial thanks are given to Alice F. Tryon who has assisted in many ways, including the preparation of the treatments of the genera Jamesonia, Eriosorls and Pellaea and to my several friends in Peru who were most helpful during my field work there and who have continued to send material to me: Dr. Ramén Ferreyra, Dr. César Vargas, Dr. Oscar Tovar, Dr. Emma Cerrate and the late Dr. Pedro Coronado. Dr. Karl Kramer has done much of the work on the treatment of Lindsaea. A. Sagastes¥ and A. Lépez (Trujillo), S. G. E. Saunders (Lima) and P. C. Hutehr son (Berkeley) continue to send valuable collections. The illustrations have been prepared over a period of time and by several artists. Mrs. . FERNS OF PERU 5 Bernedette Velick, Mrs. Joyce Todd, and Mrs, Alice Tryon have pre- pared the illustrations for about half of the species and Mrs. Ruth Chen the remainder. Mrs. Vivian Clement drew most of the habit sketches. The work has been supported by three grants from the National Science Foundation. It was initiated under Grant 1064 and it would have been impossible to undertake it without this first support. Grants 15949 and N-1565 have supported the preparation of the illustrations and the publication. GEOGRAPHY Some brief comments on the distribution of Peruvian ferns are presented here, accompanied by a map of the Departments, so that the information on habitat and range of each species may have some meaning to those unfamiliar with the country. The distribution maps do not always clearly show the general relation of the species to the type of vegetation in which it grows. This is due to the intricate system of deep valleys and high mountains, that is charac- teristic of the eastern Andes, and to the penetration of some of the larger rivers in deep valleys well into the central Andean region. It is commonplace to have a difference in elevation of some 2000 to 3000 m. within a relative few miles. Dots on a small scale map can not differentiate the major environments and vegetation types that may occur in one rather small area. The maps, then, must be inter- preted in relation to the habitat and altitude given for the species and this information will distinguish between the species of the low valleys and those of the adjacent high ridges and mountains. A somewhat fuller account of the distribution and ecology of the Peru ferns is to be found in Tryon, R. The Ecology of Peruvian Ferns, (Amer. Fern Jour. 50: 46-55. 1960) where I also include the more impor- tant references on the vegetation of Peru. There are four principal types of distribution of the ferns of Peru. These correspond to the areas of the principal types of vegetation : the Lomas, the Sierra Steppe and Scrub, the Ceja and the Montana. The lomas are locally green areas on the coastal foothills of the Andes (from La Libertad southward) where fog, mist and occasionally rain are concentrated sufficiently ROLLA TRYON The Departments of Peru. FERNS OF PERU 7 during the winter season so that a vegetation is developed in most years. During the summer the lomas are continu- ously dry. These are the only places in this part of the coast of Peru, below about 2500 m., in which ferns grow naturally. A few species have become naturalized along ditches or in other areas that are locally moist due to irrigation. Although none of the species treated here are confined to the lomas, some of them do occur there. The lomas are an important part of the range of Adiantum subvolubile, Map 32, and of A. digitatum, Map 36. The Sierra Steppe and Scrub forms the vegetation of central Peru; it occurs in a broad area in Puno and dimin- ishes northward through Junin and Huanuco into Caja- marca. The region, the Altiplano, is mostly above 3400 m. and is generally rather dry although many local mesic habi- tats occur. These latter are especially common in Cajamarca and northward where the typical Sierra vegetation is re- placed by the wetter and somewhat lower Jalca. The most characteristic ferns of the Sierra Steppe and Scrub are the xeric species of Cheilanthes, Notholaena and Pellaea: Cheu- anthes scariosa, Map 14; C. pruinata, Map 15; Notholaena sinuata, Map 19; N. aurea, Map 20; N. nivea, Map 22 and Pellaea ternifolia, Map 23. Species characteristic of locally mesic sites in this area are Pityrogramma calomelanos var. ochracea Map 9, Adiantum Raddianum, Map 33, and A. Poiretii var. Poiretii, Map 34. An extension of the Sierra fern flora on the Pacific side at a lower elevation is represented by some of the more common of the xeric species mentioned above and by two endemics: Saffordia induta (Map 26) and Cheilanthes fractifera. These latter species grow between 2500 and 3000 m. in a zone that is considerably drier than the higher elevations. The Ceja vegetation occurs along the higher eastern slopes and high ridges of the Andes. In some areas clouds, fog and frequent precipitation afford cool and rather constantly moist conditions and here the vegetation forms a dense shrubby thicket. Two species of Eriosorus, E. flexuosus, Map 4, and E. elongatus, Map 5, are predominantly in this area. In most places in the eastern Andes, however, there is a gradual transition from the Sierra Steppe and Scrub vegetation to the lower Montafia (the forest). Such species 8 ROLLA TRYON as Cheilanthes Poeppigiana, Map 17, and C. marginata, Map 18, are characteristic of this region. Below the Ceja vegetation the rainfall is greater, and with lower elevation the temperature increases. All of this eastern part of Peru is covered by the forested Montafa which includes all of the Departments of Amazonas, Loreto and Madre de Dios and portions of adjacent Departments. This is the richest area for species of ferns and although it has not been adequately collected, it seems clear that most species do not grow throughout the region. It is difficult to make clear distinctions within the fern flora but at least two rather typical types of range can be recognized. One is that of the upper elevations, above 1800 m. where there is gen- erally greater relief, higher rainfall and cooler temperatures. Species that center in this region are Adiantum peruvianum, Map 27, and A. macrophyllum, Map 29. Those more char- acteristic of the lower elevations are Saccoloma inaequale, Map 2, Pityrogramma calomelanos var. calomelanos, Map 7, Adiantum obliquum, Map 30 and A. latifolium, Map 31. A small area in the northern Department of Tumbes is forested but the fern flora is not sufficently known to relate it with the Montafia region, although several Montafia species grow there. Some species, such as Adiantum Alarconianum, are known in Peru only from this part of Tumbes. Prior to 1944 the Department of Pasco was a part of Junin. I have not attempted to disentangle the many J unin collections that were made in what is now Pasco but have cited collections under the younger Department only when it was given on the label. Some of the maps do not show the location of a few cited collections, since these were received after the blocks had been prepared. SYNOPSIS OF THE POLYPODIACEAE The following synopsis and key include the 57 genera of Polypodiaceae that I know grow in Peru and some additional ones that may be expected to occur there. Certain rare endemics of adjacent countries have not been included. The genera included in the present portion are numbered as = the text, and the number of species in Peru is enclosed parentheses ; the genera that are not treated have not been numbered. FERNS OF PERU 9 Tribe 1. Woodsieae Woodsia, Hypoderris, Cystopteris. Tribe 2. Dryopterideae Dryopteris, Thelypteris, Ctenitis, Polystichum, Cyclopeltis, Didymoch- laena, Tectaria, Polybotrya, Bolbitis. Tribe 3. Asplenieae Athyrium, Diplazium, Asplenium. Tribe 4. Blechneae Blechnum, Stenochlaena. Tribe 5. Dennstaedtieae 1. Dennstaedtia (9), 2. Microlepia (0), 3. Saccoloma (2), 4. Hypolepis (5), 5. Blotiella (1). Tribe 6. Cheilantheae 6. Eriosorus (9), 7. Jamesonia (9), 8. Pterozonium (1), 9. Pityrogram- ma (6), 10. Anogramma (1), 11. Gymnopteris (2), 12. Hemionitis (1), 13. Trachypteris (1), 14. Ceratopteris (1), 15. Cheilanthes (15), 16. Notholaena (13), 17. Pellaea (3), 18. Doryopteris (5), 19. Saffordia (1), 20. Adiantum (39). Tribe 7. Pterideae 21. Pteridium (1), 22. Paesia (1), 23. Lonchitis (1), 24. Histiopteris (1), 25. Pteris (20), 26. Acrostichum (1). Tribe 8. Vittarieae 27. Hecistopteris (1), 28. Vittaria (8), 29. Ananthacorus (1), 30. Polytaenium (4), 31. Anetium (1). Tribe 9. Davallieae 32. Nephrolepis (6). Tribe 10. Lindsaeeae 33. Lindsaea (10). Tribe 11. Oleandreae 34. Oleandra (3). Tribe 12. Polypodieae Platycerium, Polypodium, Dicranoglossum (Eschatogramma). Tribe 13. Elaphoglosseae Peltapteris (Rhipidopteris), Microstaphyla, Elaphoglossum. 10 ROLLA TRYON KEY TO THE GENERA OF POLYPODIACEAE In the following key I have tried to use the generic char- acters whenever possible and when they are adequately known. However, I have also been guided by convenience of identification, and characters that may be readily seen or that may be expected on most specimens have been given some preference. In certain poorly defined genera, such as Pellaea, Cheilanthes and Notholaena, I have keyed out only the species that grow in Peru. a. Sporangia not covered by a true indusium nor by a more or less modified portion of the margin (although the latter is sometimes present in nos. 7,15 and 16). b. b. Sporangia borne continuously all over the lamina or segment, or a large area of it. GROUP I ...... p. 10 b. Sporangia borne only on ‘the veins in definite clusters, lines or narrow bands (in no. 31 also sparsely scattered on the surface), sometimes becoming contiguous at Maturity. ......seese GROUP IL «sesso eee | a. Sporangia covered or enclosed, at least when young, by a true indusium or by a modified portion of the margin (or by both). ...... c. ec. Sporangia borne on a usually continuous marginal commissure connecting the vein-ends (in no. 14 sometimes also solitary on er veins), or on a long commissure between the margin and the p. 14 c. ena ta borne at the ends of free veins which are not or only slightly extended laterally, or in sori on the free or areolate i d. Indusium formed wholly by the modified margin, or by a modi- fied marginal lobe and a true indusium, these separate to ae joined. GROUP IV ...-+0+ p- d. True indusium back from the margin which is not — if relation to the sorus GROUP V. esseeee GROUP I a. Venation — PUP N OE ops ei aks vo ccncs teat seoed ivabisasstsv masa neon ee b. Lamina its segments with several principle veins, stellate hia: yee nen . PL b. Lamina and its rae die with a single principle vein, or vasa rial. covered with imbricate scales. .... 13. TRACHYPTERI TS secesees p- c. Fertile and sterile lamina with few or no scales on the lower WATE MUOOS “sscioscconvencibnconsssssvervivhssebineenvoesssinvexeotisesovessniveviasennssoess emsantn d. d. Rhizome large, erect; sterile pinnae finely areolate without dominant lateral veins (or groups of them), fertile age abundantly paraphysate., ........ 26. ACROSTICHUM ..-.+++ p. 2 FERNS OF PERU 11 d. Rhizome creeping; sterile pinnae (or the sterile lamina) rather coarsely areolate, with definite lateral veins (or groups of them), fertile pinnae without paraphyses or with a few abortive sporangia. BOLBITI a. Venation open, or only casually areolate, or the veins connected nie by a marginal strand. ,e. Sterile lamina simple, entire, pinnately vein ned. .... srpyeorsnaayre e. Sterile lamina deeply lobed or more complex, or flabellately veined. f. Fertile lamina entire or shallowly lobed, the sterile deeply pin- natifid or flabellately lobed or veined. , g. Sterile lamina flabellately lobed or veined. .......... PELTAPTERIS g. Sterile lamina pinnately lobed sdaanly pinnatifid) .......sssseee MICROSTAPHYLA f. Fertile and sterile lamina pinnate, 1-pinnate or more complex. h h. Pinnae not articulate, the fertile pinnae pinnatifid or more complex. POLYBOTRYA h. Pinnae articulate, the fertile pinnae entire. .... STENOCHLAENA GROUP II a. Rhizome with trichomes (sometimes bristle-like with several inflated cells near the base). " b. Lamina 1-pinnate or more complex. c. Lamina bipinnate or more complex, narrowly lanceolate to deltoid, or sometimes pinnate-pinnatifid and linear and the margins of the pinnae scarcely modified. 6. ERIOSORUS .......- p. 40 c. Lamina pinnatisect to 1-pinnate, linear, pinnae adi to shal- lowly lobed, the margins ciliate or pineal definite aE der 7 senereee border. b. Lamina ‘entire, reniform to ee . 8. PTEROZONIUM ........ p. a. Rhizome with scales (in no. 10 erichenttis sometimes also present). d. Lamina simple, entire; or furcate (dichotomously lobed) and the venation open. Lamina furcate, dichotomously or subdichotomously lobed, sa epiphytes (leaves ca. 1-2 cm. apie es open venation. ..........+++++ 27. HECISTOPTERIS ........ p 211 e. Lamina simple, entire, .......... f. Sporangia borne Lciausaaty (or nearly so) along all the bactoncoat veins (and sometimes also between them), or in two to several long lines. - g. Areolae in one series on ‘each side po CHG COREA... .ncceiccsecccneeceesen cis Anoka Be Bere ON LES . VITTARIA ........ p. 211 ae in two or more series on pe: side of the costa. .. h. line of sporangia on each side of the costa. ........ 29, ANANTHACORUS ......... p. 219 g. Areo h. A single poacdecctdcdctdecsaneveseotesssoesesesysseerse a he ROLLA TRYON h. Two or more lines of sporangia on each side of the ~ or sporangia along most of the veins. i. Rhizome long-creeping; leaves herbaceous (fresh), papyraceous (dry), sporangia borne superficially and sparingly along the veins and also between them. .......... 31, ANETIUM ........ p. 224 i. Rhizome short-creeping; leaves coriaceous; sporangia rne in more or less depressed grooves, in sovardl long lines or continuously along the veins. 30. POLYTAENIUM ........ p. 220 f. Sporangia borne in roundish to long-oblong clusters on ie papa portions of the veins. . Petiole not articulate, costa grooved and oe on is | apne side. ELYPTERI . Petiole articulate near its base or at the ee or not Taine algen and the costa more or less rounded on obec: upper side. amina variously lobed to 1-pinnate or more complex, cat dichot- “eenounlly lobed then the venation areolate. k. k. Venation areolate and the sporangia in lines or bands. ........+ L. ina pubescent. 12. HEMIONITIS ......... p. 1. Lamina scaly, at least beneath (sometimes the scales minute). m. Lamina with dense imbricate scales beneath, .......0« n. Sterile lamina spathulate (rarely lobed); sporangia borne on all veins. ............ 13. TRACHYPTERIS ........ p. 86 n. Sterile lamina more or less pedate; sporangia borne only on some veins in a narrow marginal band. ........ssssseseeser 19. SAFFORDIA ....... p. 133 m. ps with minute, appressed and scattered scales be- neath. DICRANOGLOSSUM ‘Bp k. Venation open, or if areolate, then the sporangia in a“ WOUNG, Cp Clomentes clerster gs, ics osuieseseseccscssecsecessucesssnceiceioossiunso nen o. Sori marginal, usually confined to seo vein-tips but rel extending about half-way down the Veins. ........+:::ss:ssen p. Lamina glabrous, glandular, ceraceous or pubescent be neath. .. 16. NOTHOLAENA «..+++ p- Be Dp. toinn ts i 2 , Cn ame ae q. Ultimate segments minute and subspherical; or the scales on the under surface of the pinnae strongly y and irregularly sy aa those on the upper side of the pinna-rachis linear. ............ 15. CHEILANTHES ........ p. 88 q. Ultimate weciunes moderately large and plane, the seales on the under surface of the pinnae finely pectinate- serrulate, or those on the upper side of the pinnae filiform or substellately dissected. avin sintvaders 16. NOTHOLAENA Perera i 105 0. Sori, or sporangia, dorsal, eg eter marginal vein-tips vo extending all BION The VEIN. ...000cccccovoccesecocsevsconcasssonsssensessessnenes® FERNS OF PERU 13 r. Sporangia wh definite, round to elongated clusters, borne on a receptac ‘ s. Petiole eae near its base or at the rhizome. .............. POLYPODIUM s. Petiole not articulate t. Rachis more or less rounded on the upper side, lamina usually pinnately lobed to pinnatisect (pinnae adnate), if 1-pinnate (or rarely more complex) then epiphytes and sce pendent leaves with long patent, eee hee chom: UM t. Rachis grooved and ridged on the upper a, 0 or re r less rounded, then terrestrial wi spreading leaves with articulate, catenate trichomes on e upper side of the axes (other types of — e sometimes also present). u. Upper side of rachis and pinna-rachises ek Vv. v. Only short to long-acicular trichomes present on the upper side of the axes, these usually ares and sometimes branched, ........:ssss9+ THELYPTERI vy. Articulate, catenate, usually short and rufous a chomes present on the upper side of the axes, sometimes other types of trichomes also present. .... CTENITIS u. Upper side of rachis and pinna-rachises glabrous, See scaly or with short to elongate, papilloid w. tt "Dita anes segments (pinnae or pinnules) equi- lateral or nearly so, not or hardly auriculate and not aristate, not toothed or with blunt teeth ; lamina with few scales. DRYOPTERIS w. Ultimate segments ( pinnae, pinnules or second- HUM pene sealy. . r. Sporangia borne in long or short lines te a arid ners veins. x & hirsute on the upper surface sand on the 6 11. GYMNO 82 x. Segments sidrons on the upper phoosne or rarely deciduously pubescent and then ae lower surface an rachis densely lanate. . y: y. Perennial with well developed stem; rhizome scales rigid or firm, atropurpureous to light brown. ...........++++++ si cebtavud Casebbecauecbdenswevsasedesctesas MA | aan SOB y. Annual with the stem surly developed; rhizome scales apove whitish, sometimes mixed with Gellar trichomes. 10. ANOGRAMMA ........ p- 81 Gade cdedesescoduses bipeesbesesedsessoesyere nr’ 14 ROLLA TRYON GROUP III a. Leaves pedate, the primary segments adnate or joined at the base. 18. DORYOPTERIS ........ p. 126 a. Leaves usually pinnate, rarely pedate and the basal pinnae sessile or stalked. b. b. Venation wholly or partially areolate. tel c. Aquatic; leaves thin, dimorphic, venation fully areolate-veined; stem poorly developed and short-lived. . 14, CERATOPTERIS ........ p. 87 ce. Terrestrial, stem well developed ‘and long-lived. ........:sss+seeesse d. d. Rhizome with trichomes (no scales) ; sori borne in the sinuses of the segments and lobes, ............00 5. BLOTIELLA ....000 p. 39 d. Rhizome with scales (trichomes sometimes also present in no. 24); sori borne along the sides of the segments, adjacent ones interrupted at the sinus (when present) unless it is pe: shallow. e. Rachis flat to rounded, or only shallowly depressed, on “the adaxial side, especially where the pinnae join; segments usually glaucous beneath; lower pinnae of larger leaves Ni stipule-like basal pinnules; rhizome slender and creep- 4, HISTIOPTERIS ......+- p. 186 e. "Rachie with a prominent and continuous groove on the adaxial side; segments not glaucous; the basal pinnules of the pinnae not stipule-like; rhizome cba or stout an creeping. 25. PTERIS. ...... p. 188 b. Venation wholly D8 except for the veins acliowe ve the fertile MRATGinal COMMISBUTE. .....00..cccccccercoesarsecrecserrsserrvevencssncscesossonsnsseneosswenes f. f. Segments in fertile segments with a nearly unmodified, usually flat margin extending beyond the prominent indusium which opens outward; rhizome with Scales. ......-ssssssrereeerensertee 33. LINDSAEA ....++ p. 285 f. Segments more or less equilateral. .........sscssssceseresssesesenesseseseeets g. g. Rhizome with trichomes (no scales); lamina glandular- pubescent or pubescent, at least ‘ae (rarely gabe h. Primary and other rachises (toward their apex) and ~ larger costae with convex, often separate, lobes between the segments; inner indusium a more or less continuous mem- brane or reduced to a row of trichomes. .......::s-ceeeeerreeeere 21. PTERIDIUM ....++ p. 181 h. Primary and other rachises and costae without a ae —s ars aes nts. wowancesensceseoseseeserr nl” . Inner indusium a well developed membrane; major oe of the leaf and usually the segments hard or firm, lamina usually glandular-pubescent, especially beneath, rhizome slender, hard, with dark firm trichomes. «....+-+-:+" 22. PAESIA «sees FERNS OF PERU 15 i. Inner indusium absent, whole leaf succulent, sparingly covered with whitish crispate to flattened trichomes simi- lar to those on the thick, fleshy rhizome. 23. LONCHITIS ........ p. 186 g. Rhizome with scales (trichomes sometimes also present in no. 24) ; lamina glabrous to scaly beneath, not pubescent. .... % A wisebnapes commissure marginal, the marginal indusium and th rangia usually distant from the midrib, sori planner .... 24, HISTIOPTERIS and 25, PTERIS, see heads e, . Fertile commissure near the midrib (in narrow segments marginal or intramarginal), the indusium and the mature sporangia nearly or quite extending to it, sori not para- physate BLECHNUM n~ GROUP IV a. True indusium present, separate from the opposed margin of the segment to fully joined with it and then forming a cup- or purse- shaped, or globular to cylindrical indusium. b. b. Marginal lobe joined to the true indusium to form a symmetrical cup- or purse-shaped, or globular to cylindrical indusium; rhizome long-creeping, pubescent. ........::2++ 1, DENNST. A cuntetee p. 18 b. Marginal lobe separate, or nearly so, from the true indusium and exceeding it, sometimes slightly eee) rhizome erect, with seales. 3. SACCOLOMA ......- p. 30 a. True indusium absent, sporangia covered by (or in no. 20, borne on) a modified and reflexed marginal lobe or band. C. c. Sporangia borne on veins that extend into the modified margin. .... 20. ADIANTUM ........ p. 135 c. Sporangia borne on veins that end in the leaf tissue back of the modified margin. ... d. d. Rachis and pinna-rachises grooved on the adaxial ja the grooves interrupted (or nearly so) where the axes join. ............. . 4. HYPOLEP d. Rachis and pinna-rachises pen on the adaxial side, the grooves fully continuous where the axes join; or the rachis flat to rounded on the adaxial side, or the pinna-rachis absent. ...... e e. ining more or less straw-colored, or atropurpureous and the + eee EAS sccisece p. 121 pinnae ternate. e. Paka more or less reddish- a to atropurpureous, the pinnae pinnate. . = f. Lamina sialon. a elandular or ceraceous beneath, or loose- ly pubescent and the lower surface oo 15. CHEILANTHES ........ 88 f. Lamina scaly beneath, or densely sina bia and the atti surface concealed. .. g. 16 ROLLA TRYON g. Lamina scaly beneath, the ultimate segments minute and subspherical, or the petiole scales ovate-lanceolate and the scales on the lower surface of the pinnae strongly and irregularly dentate-ciliate. . 15, CHEILANTHES ......... p. 88 g. Lamina pubescent beneath, or scaly and the ultimate Segments moderately large and plane, the petiole scales linear-lanceolate, or the scales on the lower surface of the pinnae finely pectinate-ciliate. Sin lahsnieicies . 16. NOTHOLAENA ........ p. 105 GROUP V a. Leaf, pinnae or pinnules articulate, deciduous (or easily detached in dried material). .. b. b. Leaf or pinnae articulate; indusium orbicular, reniform or lunate. Cc. c, Lamina 1-pinnate, pinnae simple, articulate; rhizome erect, . 32. NEPHROLEPIS p. 225 scdesoolibtirtttabescectevetiortbisisivbenesutbessestsesriaceciess G4, OLEANDRA «2.0 b. Pinnules articulate, dimidiate; indusium elongate along the vein; lamina bipinnate, the upper side of the pinna-rachis with digitate processes at the base of the pinnules, coecccoceeeccocec.-... DIDYMOCHLAENA a. Leaf and its segments not articulate, persistent. .d. d. Indusium elongate, attached along the vein. 4s e. Rhizome scales clathrate (the lateral cell walls dark and much thickened, the top and bottom walls very thin and nearly color- ; pinnae (or pinnules) commonly strongly inequilateral at WN eo ASPLENIUM e. Rhizome scales with the cell walls more or less equally and moderately thickened, the top and bottom walls brownish; iy (or pinnules) equilateral or slightly inequilateral at Bee BOS ee re f. Indusium about 3 ‘to 4 times as long as broad; ultimate seg- ments or lobes strongly toothed or laciniate. ce... ATHYRIUM lowly toothed. .... DIPLAZIUM d. Indusium usually about as long as broad, attached at essentially one point on the vein or receptacle (sometimes its two sides also pitimaapeape ili ecy haces vag PN 8: g- Indusium attached to the leaf-tissue on each side of the pas tacle, ooweeee h. Rhizome with trichomes, long-creeping; indusium and lamina pubescent. ............... . 2. MICROLEPIA .......- p. 29 FERNS OF PERU 17 h. Rhizome with scales, erect; indusium and lamina coma 3. SACCOLOMA tee eeeee g. Indusium attached only at the receptacle OF VEIN. ..eecececcccseeeseseee i. i. Indusium scale-like or saucer-shaped to glObOSE. ....e.cecs.cceese0+. 4, j. Indusium scale-like, attached at its base and arching over the sporangia; lamina bipinnate to tripinnate, thin, glab- rous. CYSTOPTERIS j. Indusium saucer-shaped to foaming of few to several seg- ments attached around the recep k, k. Venation open, lamina with many So ias to 1-pinnate pinnae; rhizome short-creeping. WOODSIA k. Venation areolate, lamina with a large, entire, central segment, deeply lobed at the base to form two smaller Be EIERS rhizome long-creeping, (some indusia may be elongate). HYPODERRIS ‘i . Indusium peltate or reniform. 1. Upper side of the rachis (and pinna-rachises when present) pubescent; if the lamina is tripinnate or more complex the basiscopic side) throug . Pubescence on the upper as of the axes wholly of sor 65 long, acicular trichomes, these usually one-celled a sometimes branched. ‘canis m. Pubescence on the upper side of the axes wholly or part- ly of articulate, catenate, usually short and rufous = omes, n. Venation open. Siieiees n. Venation areolate. TECTARIA l. Upper side of the rachis (and pinna-rachises when present) glabrous, eon scaly or with short to long papilloid . pubescent and the lamina tripinnate or more complex aa anadromous (the basal segment or branch on the acroscopic side) in the distal portions, or throughout. .... oO. o. Lamina 1-pinnate; pinnae simple, entire to crenate, the base truncate on the acroscopic side, prominently auricu- late on the basiscopic side, the auricle curved and over- laying the rachis. CYCLOPELTIS o. Lamina more than 1-pinnate; or 1-pinnate and the of the pinnae nearly equilateral, or better developed on the acroscopic side. p. p. Lamina less than tripinnate and the ultimate segments more complex and the pinnae broadest at the base and the pinnules acute; lamina usually scaly but not densely so. DRYOPTERIS 18 ROLLA TRYON p. Lamina less than tripinnate and the ultimate segments (pinnae or pinnules) inequilateral, auriculate or sub- auriculate, the apex, and usually the auricle, aristate; or tripinnate and the pinnae not or hardly broader at the base and the pinnules obtuse; lamina often densely scaly. POLYSTICHUM POLYPODIACEAE TRIBE 5. DENNSTAEDTIEAE 1. DENNSTAEDTIA Bernh. Jour. Bot. Schrad. 1800(2): 124. 1801. Type: Trichomanes flaccidum Forst. = Dennstaedtia flaccida (Forst.) Bernh. Terrestrial, the rhizome slender and long-creeping, pubescent, bear- ing the leaves at usually wide intervals; leaves large to very large, inner (true) indusium and the outer, opposed lobe of leaf-tissue, these fully joined basally to form a reflexed saucer- to purse-shaped or glob- ular to cylindrical whole which is bilabiate or not. — 11 American spe- cies. Tryon, R. A review of the genus Dennstaedtia in America. Contrib. Gray Herb, 187: 23-52. 1960. Most of the species grow in forests or along forest borders and have often long leaves that depend on the surrounding vegetation for support. All but two of the American species (D. distenta of Mexico, Central America and the Greater Antilles, and D. punctilobula of eastern North America) grow in Peru. KEY TO SPECIES a. Axis of the penultimate segments lacking perpendicular herbaceous wings on the upper surface, or perpendicular wings present put the one on the basiscopic side not decurrent onto the axis of the next order; trichomes on the under surface of the pinnules, when ir ent, whitish to brownish and usually subopaque. ....s.ss-sssssssersseneeeneett ending: well back HOE OF MOE march TEdUCE, ....0cccccceorcorseresssosssessessoseccecsrcenssssscessnseeseenert ¢. Many or most of the sori borne in a sinus; pinnules more OF lene pubescent beneath; lamina deltoid. wo 1, D. cioutar FERNS OF PERU 19 c. All or most of the sori terminal on lobes; pinnules glabrous to glabrate; lamina ovate-lanceolate to deltoid-lanceolate, ................ 2. D. glauca b. Sterile vein tips, on the upper surface, enlarged, ciate to punc- tate (rarely slender in D. obtusifolia); tertiary axes glabrate to usually subappressed, ascending pubescent beneath with the tri- chomes more or less curled. d. Lamina pinnate-pinnatifid or more complex; the lower Ean usually subsessile with the basal pinnules reduced. ..............+++ e. Sterile veins ending well back of the glabrous margin. ........ f. f. Pinnules 1-pinnate 'to pinnate-pinnatifid, apical segments of the pinna separate or nearly so to the prolonged apex, the separate ones closest to the tip obtuse; obtuse pinnules on oe apical pinnae deeply pinnatifid; lamina deltoid. .......... g. . Mature sori, at least the basal acroscopic ones on an Siilibausahe segment, ed 1.0-1.5 mm. broad and about half as thick. 5. D. dissecta g. Mature sori mostly 0.5-1.0 mm. broad and about as thick. 6. obtusifolia f. Pinnules entire to deeply pinnatifid, apical seguients of the pinna confluent back of the prolonged tip, the separate ones closest to the tip acute; obtuse pinnules on the apical pinnae entire to lobed; lamina evidently ovate. 7. D. arborescens e. Sterile veins nearly reaching the persistently pubescent mar- gin. Si DB. i d. Lamina 1-pinnate, bes PINNAE ENTITE. ...........0s00000 9. D. Wercklei a. Axis of the penultimate segments bordered on each side, on the upper surface, by a pronounced herbaceous wing perpendicular to the plane of the segment, the wing on the basiscopic side decurrent onto the axis of the next order either as an herbaceous wing or as a pronounced ridge; trichomes on the under surface of the pinnules wholly clear brown or tan (rarely subopaque or whitish). .............. h. h. Basal segments of the pinnules of the central pinnae usually subopposite to nearly opposite, rather or quite equal in size, the inferior not or slightly ascending; sori globular to less often s globular or rarely cylindrical; lamina deltoid. ...... 3. D. globulifera h. Basal segments of the pinnules of the central pinnae definitely alternate, quite unequal in size, the inferior ascending to strongly ascending; sori cylindrical to subcylindrical, rarely globular; lami- na ovate to lanceolate-ovate. 4. D. bipinnata 1. Dennstaedtia cicutaria (Sw.) Moore, Ind. Fil. xevii. 1857. Fie. 1, AP 1 Dishéonsi cicutaria Sw. Jour. Bot. Schrad. 1800 (2) :91. 1801. biel Jamaica, Swartz, s-pa! fragment and photographs Us!; isotype: Willd. 20156, B! photo GH icksonia rubiginosa Kaulf, Enum. Fil. 226. 1824. Type: Rio de Janeiro, Brazil, Herb. Mertens; isotype: LE! photo GH. Dennstaedtia rubiginosa (Kaulf.) Moore, Ind. Fil. xevii. 1857. 20 ROLLA TRYON Leaf ca. 1-4 m. long, lamina deltoid, tripinnate to quadripinnate- pinnatifid; lower pinnae stalked and with the basal pinnules not, or not much, reduced, axes of the penultimate segments without herbaceous ridges or wings on the upper surface, sterile vein-tips (on the upper surface) slender, ending well back of the glabrous to sparingly pubes- cent margin; pinnules more or less pubescent beneath (as well as above) with whitish to brownish usually dimorphic trichomes, some one-celled, short, acicular, rigid, others (rarely the only kind) multi- cellular, longer, sometimes lax and subtortuous; most sori borne in @ sinus, mature indusium saucer-shaped to purse-shaped. In addition to the characters mentioned in the key, this species and the following one, D. glauca, may often be sepa- rated by the color of the major axes. In D. cicutaria they are usually brownish and in D. glauca they are usually straw- colored. Central Mexico to Panama; Greater Antilles; Venezuela and Colombia to Bolivia and southern Brazil. In dense or open forests or at the borders of clearings, Loreto and San Martin south to Puno, 200-1700 m. Selected specimens: SAN MARTIN: near Tarapoto, Spruce 4338 (GH, K), 4346 (K); Zepelacio, near Moyobamba, Klug 3553 (F, GH, K, US). LORETO: mouth of Rio Santiago, Mexia 6128 (BM, F, GH, K, UC, us); La Victoria, Amazon River (“near Brazil”), L. Williams 2620 (F). HUANUCO: Cotirarda (Dist. Churubamba), Mexia 8218 (F, GH, UC, us); Quillasé, Soukup 3513 (Us). JUNIN: Colonia Perené, Killip & 25234 (F, US); La Merced, Killip & Smith 23699 (F, GH, US) ; Chancha mayo valley, Schunke 170 (F), 756 (F), 1351 (F, US); Satipo, AUB: 1940, Ridoutt (GH, USM). AYACUCHO: Estrella, between Huanta and Rio Apurimac, Killip & Smith 23072 (GH, Us); Ccarrapa, between Huanta and Rio Apurimac, Killip & Smith 22440 (GH, US). aes Potrero, 8 km. w. of Quillabamba, Tryon & Tryon 5372 (BM, Fy GBs ie FF USM); Yanayacu, Biies 2007 (us). PUNO: Churumayo, Soukup 871 F). 2. Dennstaedtia glauca (Cav.) Looser, Rev. Hist. Geog. Chile 69: 184. 19382. Fic. 2. Davallia glauca Cav. Descr. 278. 1802. Type: Cordillera de Planchon, Chile, Neé, ma, fragment BM! (Looser, loc. cit., and C. Chr. in Dansk Bot. Ark. 9(8) :28. 1937, discuss the identity of the type). . Dicksonia Lambertiana Remy, Gay, Fl. Chil. 6: 523. 1853. TyP® Chile, Herb. Bonpland, P ? (not seen) ; isotype: K! Dennstaedtia Lambertiana (Remy) Christ, Farnkr. 312. 1897. FERNS OF PERU 21 Leaf ca. 0.5-2 m. long, lamina ovate-lanceolate to deltoid-lanceolate, tripinnate-pinnatifid to quadripinnate-pinnatifid; lower pinnae stalked, with the basal pinnules not or hardly reduced, axes of the penultimate segments with an herbaceous ridge on the upper surface, sterile vein tips, on the upper surface, slender, ending well back of the glabrous margin; pinnules glabrous or nearly so; sori predominantly terminal on lobes, mature indusia usually purse-shaped. Specimens of the previous species, D. cicutaria, that are only slightly pubescent may be confused with D. glauca. The character of the sorus and especially that of the shape of the lamina, mentioned in the key, must then be employed for certain identification. Chile, northeast to adjacent Argentina, north to Bolivia and southern Peru. In moist, shrubby or rapeng ravines and on banks, Cuzco and Puno, 2900-3700 m Specimens seen: cuzco: Hacienda Chiraura, prov. Quispicanchis, saree 2623 (UC, US); Quebrada de Quispicanchi, Herrera 2598 Vargas 12011 (GH). PUNO: Cuyo-Cuyo, Weberbauer 934 (B). 3. mS Ea globulifera (Poir.) Hieron. Bot. Jahrb. 34: 455. 1904. Fic. pe globuliferum _ Semi Encycl. 5: 554. 1804. Type: Santo Domingo, Plumier, Fil. t. icksonia tenera Presl, Del. al rag. 1: 189. 1822. Type: Rio de Janeiro, Brazil, Pohl 4062, w! det. Presl, may be an isotype, an unan- gies duplicate i is at GH. onia exaltata Kze. Bot. Zeit. 8: 59. 1850. Type: Santo Domingo, car Fil. t. 30. (nom. superfl., illegit. Dennstaedtia tenera (Presl) Mett. Ann. Sci. Nat. V, 2: 261. 1864. Dennstaedtia exaltata (Kze.) Hieron. Bot. Jahrb. 34: 454. 1904. (illegit.) Leaf ca. 1-3 m. long, lamina deltoid, tripinnate to nearly quadripin- nate; lower pinnae stalked, with the basal pinnules not or scarcely reduced, basal segments of the pinnules of the central pinnae suboppo- site, nearly or quite equal in size, the inferior one not or slightly ascending, axes of the penultimate segments bordered on each side, on the upper surface, by a pronounced herbaceous wing perpendicular to the plane of the segment, the wing on the basiscopic side decurrent onto the axis of the next order, sterile vein-tips usually slender, ending well back of the glabrous margin; pinnules more or less pubescent beneath, the trichomes wholly clear brown or tan (rarely subopaque or whitish), rather straight and rigid; indusia globular to rarely cylindrical. 22 ROLLA TRYON nt, pyri H. H. Smith 1118, au. Fig. 4. D. bipinnata: A, pinnules, X % - 5. Williams 1259, cH; B, base of tertiary segments, X 5, Cuba, Hioram 6299, ee FERNS OF PERU 23 This species is rather closely related to the next, D. bipin- nata, and the differences are discussed under its treatment. The distinctive characters of the basal segments of the pin- nules are best developed in the central pinnae. In the apical pinnae, where the pinnules are small, the basal pinnules of the pinnae have the same characters. In the larger, lower pinnae both the basal and apical ‘gata may be similar to those characteristic of D. bipinna Texas; Mexico to Panama; pra a Antilles; Venezuela and Colombia south to Bolivia, Argentina and southern Brazil. In forests, Cajamarca, Junin and Cuzco, 1000-1800 m. Specimens seen: CAJAMARCA: Taulis, prov. Hualgayoc, Koepeke 676 (GH); Prov. Hualgayoc, Soukup 3809 (F, US). JUNIN: above San Ramon, Killip & Smith 24648 (F, US); Pichis Trail, Yapas, Killip & Smith 25473 (F, GH, US); La Merced, Soukup 1076 (F); Chanchamayo valley, Schunke 12 (F, US) Quebrada of Pariahuanca, Mathews 974 (K). CUZCO: Valle de Occobamba, Biies 879 (US). 4, a eee bipinnata (Cav.) Maxon, Proc. Biol. Soc. Wash. 61: 39. 1938. bide | een Cav. Descr. 174. 1802. Type: Porto Rico, Ven- tenat, MA; isotype: Herb. Willd. 20165-1, B! photo GH, fragment Us! (Maxon, loc. cit., discusses the Herb. Willdenow specimen accepted as an isotype). ksonia adiantoides Willd. Sp. Pl. 5: 488. 1810. Type: Caripe, Venezuela, Humboldt & Bonpland (466), Herb. Willd. 20165-2, B! photo GH, fragment Us!; isotype: Herb. Humboldt & Bonpland (466), P! Dennstaedtia adiantoides (Willd.) Moore, Ind. Fil. xevii. 1857. Leaf ca. 1-3 m. long, lamina ovate to lanceolate-ovate, tripinnate- Pinnatifid to quadripinnate-pinnatifid; lower pinnae stalked, with the basal pinnules not or scarcely reduced, basal segments of the pinnules of the central pinnae definitely alternate, quite unequal in size, the inferior ascending, axes of the penultimate segments bordered on each side, on the upper surface, by a pronounced herbaceous wing perpendic- ular to the plane of the segment, the wing on the basiscopie side decur- rent onto the axis of the next order, sterile vein-tips usually slender, ending well back of the glabrous margin, pinnules usually slightly pubescent beneath, the trichomes often nearly confined to the base of the tertiary segments, trichomes wholly clear brown to tan (rarely subopaque or whitish), rather straight and rigid; indusium cylindrical to rarely globular. D. bipinnata and D. globulifera often differ in characters other than those mentioned in the key. The lamina in D. 24 ROLLA TRYON bipinnata is usually coriaceous or firmly herbaceous and is usually shining beneath. In D. globulifera the lamina is usually softly herbaceous and dull beneath. The ultimate segments of D. bipinnata are acutely to subacutely toothed while those of D. globulifera are entire to usually bluntly toothed. Southern Florida; Mexico to Panama; Greater Antilles; Trinidad to Colombia, south to Peru and Bolivia. In forests, Loreto, San Martin to Ayacucho, 135-1800 m. ecimens seen: SAN MARTIN: Monte Campana, Tarapoto, Spruce 4690, 4890 (K); San Roque, L. Williams 7179 (F). LORETO: mouth of Rio Santiago, Mexia 6127 (BM, F, GH, K, UC, US) ; Puerto Arturo, below Yurimaguas, Killip & Smith 27796 (Us); Maynas, 1831, Poeppig (K). JUNIN: Satipo, Aug. 1940, Ridowtt (GH, USM); Chanchamayo valley, Schunke 145, 945 (F), 947 (F, US); La Merced, Soukuwp 1027 (F); Cahuapanas, Rio Pichis, Killip & Smith 26799 (Us); Colonia Perené, Killip & Smith 25188 (us). AYACUCHO: near Kimpitiriki, Apurimac valley, Killip & Smith 22875 (¥, Us), 22987 (BM, US); Estrella, be tween Huanta and Rio Apurimac, Killip & Smith 22651 (GH, US). 5. Dennstaedtia dissecta (Sw.) Moore, Ind. Fil. 305. 1861. Fic. 5. Polypodium dissectum Sw. Prod. 134. 1788, not Forst. 1786. Dicksonia dissecta Sw. Jour. Bot. Schrad. 1800 (2): 91. 1801. Type: Jamaica, Swartz, s-pA! photo and fragment vs! Dicksonia cornuta Kaulf. Enum. Fil, 227, 1824. Type: Brazil, “ex Spreng.” (Brazil, Sello, 8! may be the collection described). Dennstaedtia cornuta (Kaulf.) Mett. Ann. Sci. Nat. V, 2: 260. 1864. e indusia half The Swartz material at S-PA consists of two sheets of sterile specimens and one sheet of two fertile specimens: The latter has been appropriately marked “type” by Maxon, only a few of the sori on these specimens are as broad as 1? typical D. dissecta. FERNS OF PERU 25 This species and the next, D. obtusifolia, are very closely related. The only differences that I have been able to find to distinguish them are those of the size and shape of the indusia mentioned in the key. It is quite possible that the two represent a single species with variable sori, but the evidence for this is not conclusive. Mexico to Panama; Jamaica and Hispaniola; Trinidad to Colombia, south to Bolivia and southern Brazil. In forests, Hudnuco and Junin, 1300-1700 m, Specimens seen: HUANUCO: Cushi, Macbride 4844 (F, US); 12 km. w. of Puente Durand, Stork & Horton 9862 (F, UC); Puente Durand to Exito, Mexia 8248 (F, GH, UC, US). JUNIN: Schunke Hacienda, above San Ramén, Killip & Smith 24546 (Us); Pichis Trail, San Nicholas, Killip & Smith 26017 (us). 6. Dennstaedtia obtusifolia (Willd.) Moore, Ind. Fil. 306. 1861. G. 6 Dicksonia obtusifolia Willd. Sp. Pl. 5: 483. 1810. Type: Caracas, Venezuela, Bredemeyer, Herb. Willd. 20163, B! photo GH, Us, fragment LE! Dennstaedtia ordinata (Kaulf.) Moore, Ind. Fil. 306. 1861. Dennstaedtia erosa (Kze.) Moore, Ind. Fil. 306. 1861. Dennstaedtia Orbignyana Kuhn, Linnaea 36: 146. 1869; Chaetopt. 384. 1882. Type: Bolivia, D’Orbigny 278 B!, photo GH; isotype: P! basiscopic side not decurrent onto the axis of the next order, sterile vein-tips, on the upper surface, clavate to punctate (rarely slender), ending well back of the glabrous margin; pinnules glabrate to usually pubescent beneath, especially on the axes, the trichomes brownish, subappressed, ascending, more or less curled; mature indusia cup- shaped, mostly 0.5-1.0 mm. broad and about as thick. This species is closely related to the previous one, D. dis- 26 ROLLA TRYON secta, and the few differences between them are discussed under that. species. Guatemala to Panama; Greater and Lesser Antilles; Trin- idad to Colombia, south to Bolivia, Paraguay and southern Brazil. In forests, San Martin to Ayacucho, 500-1700 m. Specimens seen: SAN MARTIN: near Tarapoto, Spruce 4246 (GH, K), 4346 (BM, K). HUANUCO: Pampayacu, Kanehira 160 (Us); Pampa- yacu, 1829, Poeppig (K). JUNIN: Pichis Trail, Killip & Smith 25550 (US); above San Ramon, Killip & Smith 24554 (F, Us), Schunke A230 (Us) ; near La Merced, Killip & Smith 23915 (BM, F, US), 24083 (US); Colonia Perené, Killip & Smith 24916 (BM, F, US); La Merced, So 1024 (F); Chanchamayo valley, Schunke 154 (F, US), 506, 971 (F). AYACUCHO: Estrella, between Huanta and Rio Apurimac, Killip & Smith 22647 (F, GH, US). 7. Dennstaedtia arborescens (Willd.) Maxon, \Proc. Biol. Soc. Wash. 43: 88.1930. Fic. 7. Davallia arborescens Willd. Sp. Pl. 5: 470. 1810. Type: Santo Do- mingo, Plumier, Fil. t. 6. Davallia concinna Pres], Rel. Haenk. 1: 66. 1825, not Schrad. 1818. Syntypes: “Chile, 1790,” Haenke, 2 sheets PR, photos GH, Us. Th species is not known from Chile; Presl was undoubtedly correct in later (Epim. Bot. reprint, 102) citing the Haenke collection from Peru. Dicksonia Pavonii Hook. Sp. Fil. 1: 74. 1844. Based on Davalha arborescens Willd., not Dicksonia arborescens L’Hérit. (The specimen cited is Dennstaedtia Sprucei). eparia Mathewsii Hook. Sp. Fil. 1: 85, t. 30B. 1844. Type: Peru, Mathews 1782 k! fragment P! us!, photo US; isotypes: BM! Us! Dennstaedtia concinna (Presl) Moore, Ind. Fil. xevii. 1857. Dennstaedtia Pavonii (Hook.) Moore, Ind. Fil. 307. 1861. Dennstaedtia Mathewsii (Hook.) C. Chr. Ind. Fil. 218. 1905. Leaf ca. 1.5-3 m. long, lamina evidently ovate, bipinnate to tripin- nate; lower pinnae usually subsessile, with the basal pinnules reduced, apical segments of the pinna confluent back of the prolonged tip, the separate ones closest to the tip acute, axes of the penultimate segments without perpendicular herbaceous wings, on the upper surface, or with them but the wing on the basiscopic side not decurrent onto the axis of the glabrous margin; pinnules glabrate to usually pubescent be neath, the trichomes brownish, subappressed, ascending, more or less curled; mature indusia purse-shaped to usually cup-shaped. The shape and size of the ultimate segments are quite variable in this species and, in its aspect, the lamina may resemble that of D. obtusifolia, or D. Sprucei. Dennstaedtia FERNS OF PERU 27 arborescens and the two previous species, D. dissecta and D. obtusifolia, sometimes have proliferous buds in the axils of the pinnae; they are evidently more common in this species and rather rare in the other two. These buds are easily detached and presumably can produce young plants when they fall to the ground. I have not, however, seen any material that would substantiate this conclusion. araee! “ang area ~ ~apgnaense EU SRA EESSS eS RNC . ar : i Ekman H7633, a Fig. 8. D. Sprucei: margin of ance aye x ae, ae Lape & Smith 25848, GH. Fig. 9. D. Wercklei: pinnae, X %%4, Colo ia, Dryander 28 ROLLA TRYON Mexico to Panama; Greater Antilles ; Venezuela to Colom- bia, south to Bolivia. In forests, San Martin to Puno, 1400-1700 m. Specimens seen: Mathews 1782 (BM, K, P, US). SAN MARTIN: near Tarapoto, Spruce 4346 (BM, GH), 4692 (K) ; Japalacio, Stiibel 1110 (B). HUANUCO: Huacachi, near dre Macbride 4176 (F, US); Cushi, Mac- bride 4842 (F, US). JU : above San Ramon, Killip & Smith "24642 (F, GH, US); Scehuaace valley, Schunke 34, 104 (F). cuzco: Amai- bamba, prov. Convencién, Vargas 9805 (UC); Valle de Occobamba, Biies 868 (US); Rio Arasoa, ne. on Cuzco, Sandeman 3689 (K); Cha- quelloccauca, prov. Convencién, Vargas 12939 (GH). PUNO: San Gaban (Rio), Lechler 2157 (L, B). 8. Dennstaedtia Sprucei Moore, Ind. Fil. 308. 1861. Type: Ecuador, Spruce 5350 (Herb. oe K!; isotype: GH! P!, photo and frag- ment ex C, us! a 1-2 m. long, lamina (shape uncertain) pinnate-pinnatifid to bipinnate-pinnatifid; lower pinnae sessile or subsessile, with the basal pinnules reduced, axes of the penultimate segments without herbaceous wings, on the upper surface, sterile vein-tips clavate 'to punctate, near- ly reaching the persistently pubescent margin; pinnae and pinnules pubescent beneath, especially on the axes, the trichomes brownish, curled; mature indusia purse-shaped to cup-shaped. Dennstaedtia Sprucei is evidently a rare species and ade- quate material has not been available to determine the shape of the lamina. The shape and size of the ultimate segments are quite variable and they parallel the similar variation in D. arborescens. The persistently pubescent margin, with the vein ends nearly extending to it, are distinctive characters of D. Sprucei. Ecuador and Peru. Dense forest, Junin, 1700-1900 m. Specimens seen: “Perou” (F' no. 809472). JUNIN: Oxapampa, Sou- kup 1826 (GH, US); Dos de Mayo, Killip & Smith 25848 (BM, GH, Us). 9. oe Wercklei (Christ) Tryon, Contrib. Gray Herb. 187: 50.1960. Fic. co Werekiei Christ, Bull. Boiss. II, 4: 1100. 1904. Type: a Rica, Wercklé 320, P!, photo and figs. in Amer. Fern Jour. 48: pl. 7 14, 1958 Dennstaedtia arcuata Maxon, Amer. Fern Jour. 35: 22. 1945. Type: Colombia, Killip 5565, us!; isotype: GH! Pog olepia Wercklei (Christ) Kramer, Amer. Fern Jour. 48: 116. FERNS OF PERU 29 Leaf ca. 0.5-1.5 m. long, lamina lanceolate, 1-pinnate, pinnae entire, sometimes subauriculate, subsessile, sterile vein-tips clavate to punc- tate, ending near the glabrous margin; pinnae slightly pubescent beneath, especially on the costa, the trichomes brownish, subappressed, hacendine. more or less curled; mature indusia purse-shaped, contigu- ous and often joined. The once-pinnate lamina makes this a very distinctive species. Costa Rica, Colombia and Peru. In forest, Hudnuco, 2000 m Specimen seen: HUANUCO: Pozuzo, Bryan 674 (US). 2. MICROLEPIA Presl, Tent. Pterid. 124. cat Type. Polypodium speluncae L. = Microlepia speluncae (L.) Moo Terrestrial, the rhizome long-creeping, pubescent, bearing leaves at intervals; leaves large, tripinnate to quadripinnate, softly pubescent, veins free; sorus roundish, not paraphysate, borne on a vein-tip back from the margin, indusium fully attached to the leaf-tissue along its sides (or in M. jamaicensis, the sides partially free), half cup-shaped, the leaf-tissue beyond the sorus not modified. — 2 species in America. Microlepia speluncae (also in the Old World) is evidently rare and sporadic in its distribution through South America and the West Indies. It has not been collected in Peru but is known from Bolivia. It will probably eventually be found in Peru in forest or ravines of the eastern Andes or adjacent Amazon region. The other American species, M. jamaicen- sts, is confined to the Greater Antilles. crolepia speluncae (L.) Moore, Ind. Fil. xciii. 1857. Rhizome trichomes rather soft, several cells long, these mostly alter- nately flattened (catenate), the side walls colorless, transparent, the end walls brownish, rather opaque; leaves ca. 0.5-1.5 m. long, long- petioled, lamina lanceolate to broadly lanceolate, pinnae broadest at or near the base, acuminate, softly pubescent beneath and less so above; indusium pubescent. The indusium and sorus of Microlepia speluncae are sim- ilar to those of Saccoloma inaequale. The two species may be readily separated by the creeping, pubescent rhizome and the pubescent lamina and indusium of M. speluncae and the erect, scaly rhizome and the glabrous lamina and indusium of S. inaequale. The general aspect of the lamina and the 30 ROLLA TRYON indument of Microlepia speluncae are similar to those of Dennstaedtia cicutaria. These species may be distinguishe by the sorus, which is borne back of the unmodified margin in M. speluncae and is marginal and united with the modified margin in D. cicutaria. 3. SACCOLOMA Kaulf. Berl. Jahrb. Pharm. 1820: 51. Type: Saccoloma elegans Kaulf. Orthiopteris Copel. Bishop Mus. Bull. 59: 14. 1929. Type: Davallia ferulacea Moore = Orthiopteris ferulacea (Moore) Copel. (the species has not been placed in Saccoloma). Ithycaulon of auths., including Copel. Univ. Cal. Publ. Bot. 16: 79. 1929. Type: Davallia moluccana Bl. = Tapeinidiwm moluccanum. (Bl.) C. Chr. This generic name has been erroneously applied to species of Saccoloma. Terrestrial, the rhizome erect, or decumbent with age, scaly, especial- ly at the apex, bearing the leaves in a crown or cluster; leaves of medium size to large, 1-pinnate to tripinnate-pinnatifid, glabrous or nearly so, veins free; sorus borne on a vein tip, not paraphysate, mar- Maps 1-2. Map 1, Dennstaedtia cicutaria. Map 2, Saccoloma inaequale. FERNS OF PERU 31 ginal (in S. elegans) and the indusium narrowly lunate, the pinna margin usually modified, or borne back of the margin (in S. inaequale) and the indusium half-conical, attached along its sides, the leaf-tissue beyond not or very slightly modified. — 3 American species. Tryon, R. The genus Saccoloma Kaulf, in Tax. Fern Notes, III. Cintab. ‘Gea Herb. 191: 100-106, 1962. KEY TO SPECIES a. Lamina bipinnate or more complex. b. b. Lamina rather finely cut into mostly small, strongly toothed or lobed ultimate segments, its apex cut almost to the rachis; sori distant, indusium narrowly cuneate, shorter than the unmodified, opposed mar; 1. S. inaequal b. Lamina ices cut into mostly large, entire . shallowly toothed ultimate segments, its apex pinnatifid; sori mostly on adjacent vein-ends, indusium broadly cuneate, nearly mre to the short, somewhat modified, indusiform marginal lobule; West Indies, Venezuela, Colombia, Brazil. ...... S. domingense ee C. Chr. a. Lamina 1-pinnate, the pinnae simple. eleg 1. Saccoloma inaequale (Kze.) Mett. Ann. Sci. Nat. IV, 15: 80. 1861. Fig. 10, MAP 2. Davallia inaequalis Kze. Linnaea 9: 87. 1834. Type: Yurimaguas, ayn ow Alto Amazonas, Loreto), Peru, Dec. 1830, Poeppig, Diar. 2113, probably destroyed at LZ; isotype: B! photo GH, GH! Saccoloma inaequale var. nadeia: Hieron. Hedwigia, 47: 207. 1908. bie near Yquitos and Cocha (in the Marafion valley), Peru, Stiibel 1129 see ie inaequalis (Kze.) Copel. Gen. Fil. 50. 1947. Rhizome quite erect; leaf ca. 1 m. (-1.5 m.) long, long-petioled, lamina bipinnate to quadripinnate, deltoid to long-triangular, rather finely cut into mostly small, strongly toothed or lobed ultimate seg- ments, the apex gradually reduced, cut almost to the rachis, long- triangular, veins more or less evident beneath; sori distant, indusium glabrous. There is considerable variation in the size of the ultimate segments, the degree to which they are lobed and the form of the apex of the penultimate segments. None of the varia- tions appears to merit taxonomic recognition. Mexico and Central America; West Indies; Guianas to Colombia, south to Bolivia; Brazil. In woods and dense forest, Loreto and San Martin to Puno, 100-1700 m. 32 ROLLA TRYON Selected specimens: SAN MARTIN: Tarapoto, Spruce 4691 (kK), L. Williams 6007 (F, GH); Cumbassauma Mts. (near Tarapoto), Steere (GH); Rioja, Stiibel 1056 (B). LORETO: Iquitos, Ule 6884 (K), Killip & Smith 30687, 37319 (¥F, US); Tierra Doble, alto Rio Nanay, L. Willams 1053 (¥F); Mishuyacu, near Iquitos, Klug 239 (F, US); Yurimaguas, Maynas, Poeppig (K), 2113 (B, GH); Yurimaguas, L. Williams 4421 (F, GH), Killip & Smith 28033 (us). HUANUCO: Tingo Maria, Asplund 12242 (K, US), Ferreyra 10320 (GH, USM), Tryon & Tryon 5223 (BM, Yapas, Pichis Trail, Killip & Smith 25548 (US); Schunke Hacienda, above San Ramén, Schunke A232 (GH, US); near La Merced, Killip & Smith 23988 (F, GH, US); Puerto Bermudez, Killip & Smith 26468 (F, US); Puerto Veins Killip & Smith 26368 (F, US). PUNO: San Gaban (Rio), Lechler 2293 (B, photo GH, K); Chunchusmayo, Sandia, Weber- bauer 1263 (B, photo GH). 2. Saccoloma elegans Kaulf. Berl. Jahrb, Pharm. 1820: 51. Type: None cited, but in Enum. Fil. 224, t. 1, 1824, Kaulfuss cites Sello, Brazil, which may be accepted as the type; a specimen at B! photo GH, is undoubtedly an isotype. FG. 11. Rhizome erect ‘to decumbent with age; leaf ca. 1-1.5 m. long, long- petioled, lamina 1-pinnate, ovate-oblong, apex with a conform terminal pinna, pinnae simple, long-elliptical to oblong-elliptical, crenulate to ae at the acuminate apex, veins prominent beneath; sori very numerous along the margin, contiguous, indusium glabrous. Central America; Greater Antilles; Trinidad and the Guianas to Colombia, south to Bolivia; Brazil. In woods and dense forest, Loreto, Junin and Puno, 100- 1700 m Specimens seen: LORETO: Iquitos, Killip & Smith 27121 (F, NY, us). JUNIN: San Nicolas, Pichis Trail, Killip & Smith 26072 (F, GH, NY; Us) ; Schunke Hacienda, above San Ramon, Killip & Smith 24574 (GM NY, US), Schunke A231 (GH, US); La Merced, Sowkwp 1090 (F). PUNO: San Gaban (Rio), Lechler 2416 (B). 4, Hypoteris Bernh. Neues Jour. Bot. Schrad. 1 (2) :34. 1805. (1806). Type: Lonchitis tenuifolia Forst. = Hypolepis tenuifola (Forst.) Bernh. Terrestrial, the rhizome slender, long-creeping, pubescent, bearing the leaves at intervals: leaves of medium size to very large, bipinnate pinnatifid to quadripinnate-pinnatifid, glabrate to usually pubescent, veins free; sorus borne on a vein-tip ending at a sinus, not paraphy- sate, the eporkegis more or less covered by the indusium which is formed from a reflexed, more or less modified, marginal lobe. — About 20 species in America. FERNS OF PERU 33 Some of the species, especially Hypolepis hostilis, are sim- ilar to Pteridium in their ecology and manner of growth; Fics. 10-13. Fig. 10. Saccoloma inaequale: A, pinnule, X 14, Peru, L, Williams 4421, GH; B, fertile tertiary segment, X 1, idem. Fig. 11. S. elegans: A, fertile pinna, K 44, enlarged, Ecuador, Mexia 8429, GH; Panama, Killip 2531, GH; B, basal view of spore, . ig. %4, Colombia, Killip & Smith 19368, GH; B, fertile tertiary segmen 7, idem. Fig. 18. H. parallelogramma: A, pinnule, X 14, Peru, Mexia 8149, GH; B, fertile tertiary 34 ROLLA TRYON the two may grow together in recently cleared areas. How- ever, the species of Hypolepis never become pernicious weeds as Pteridium sometimes does. The usually large leaves are not well represented as herbarium specimens, nor are the rhizomes or juvenile plants. The latter may have (see H. hostilis) a different type of indument than the adult leaf. It is anticipated that better and more representative collec- tions will lead to an improved classification of the American species. Hypolepis viscosa Karst. of Colombia may possibly be discovered in Peru. It is similar to H. Stuebeli in its ciliate indusium. However, the lamina is glandular-pubes- cent beneath (lacking the long stiff trichomes of H. Stu- ebelit) and it has numerous short, glandular trichomes on the upper surface. KEY TO SPECIES a. Indusium long-ciliate with few-celled ‘trichomes; lower surface of the segments with moderately long, straight or curved, few-celled, terete, stiff, pointed trichomes (shorter, gland-tipped ones may also be present); stipe, rachis and pinna-rachises aculeolate or very sparingly so. 1. H. Stuebelit a. Indusium glabrous, smooth on the edge ‘to erose-fimbriate; lower surface of the segments glabrous or with multicellular, often moni- liform, or gland-tipped trichomes. - b. Ultimate segments, or many of them, squarish or very y bluntly obtuse; lower surface of the segments glabrous or very slightly pubescent; rachis and pinna-rachises aculeolate, especially be- neath. 2. H. poral b. Ultimate segments rounded at the apex. sc.cccsecssssssecsecesneeseesenesnee c. Lower surface of the segments with nulticellalar, often ‘a tortuous and moniliform trichomes with a pointed apex, these sometimes forming a tomentum nif d. Rachis flexuous, subdichotonoin toward the bases "lower pinnae much larger than those above, strongly inequilateral. 3.H ote sis d. Rachis quite or nearly straight, lower pinnae not much, i at all, larger than ‘those above, nearly or quite equilateral. «+ 4. H. obtusata ce. Lower surface of the segments very slightly to rather densely pubescent, the trichomes various but some gland- -tipped oF Silents rachis and usually the pinna- -rachises ne 5. H. 1. Hypolepis Stuebelii Hieron. Hedwigia 48: 230, t. 10, fig. 8. 1909. Type: near San Florencio, prov. Manabi, Ecuador, Stiibel 796, B! frag- ment US!; isotype: GH! Fic. 12. FERNS OF PERU 35 o 1m. (or more) tall, more or less erect; lamina tripinnate- aaatend to quadripinnate, rachis straight, it and the pinna-rachises equilateral; segments pubescent beneath with moderately long, straight or curved, stiff, few-celled, terete, pointed trichomes, a few shorter gland-tipped ones may also be present, ultimate segments broadly rounded to subacute; indusium long-ciliate. The long-ciliate indusium as well as the distinctive pubes- cence on the lower surface of the segments amply distinguish this species. The related Hypolepis rigescens (Kze.) Moore has similar indument but a non-ciliate indusium. Greater Antilles; Venezuela to Colombia, south to Peru. In pastures, rocky open places and rocky woods, Junin and Cuzco, 1450-2300 m. Specimens seen: JUNIN: Chanchamayo valley, Schunke 689 (F), 942 (F, US). cUzco: Cerro de Cusilluyoc, Pennell 14025 (F, GH); Choquel- louanca, prov. Convencién, Vargas 12936 (GH) 2. Hypolepis parallelogramma (Kze.) Presl, Tent. Pterid. 162. 1836. Fie. 13. Cheilanthes parallelogramma Kze. Linnaea 9: 85. 1834. Type: Pam- payacu, Peru, July, 1829, Poeppig, Hb. Kze. presumably destroyed at LZ; fragment ex Kze., LE! photo GH. Leaf to 3-3.5 m. (to 7 m.) long, ont im lamina tripinnate to tri- pinnate-pinnatifid, rachis more or less straight, it and the pinna- rachises aculeolate, lower pinnae not ciek if any larger than those above, nearly equilateral; segments glabrous beneath or sometimes very slightly pubescent with multicellular, sometimes long, tortuous and moniliform trichomes with a pointed or rarely gland-tipped apex, ultimate segments squarish or very bluntly obtuse, or many of them so; indusium glabrous, smooth on the edge. The exceptionally long leaf is reported to be scandent over shrubs, or the apical portion pendant from the lower branches of trees. The shape of the ultimate segments and the usually glabrous segments serve to distinguish specimens of this species. In some portions of the lamina the ultimate segments are three times as long as broad with parallel and entire margins, and they present a distinctive type of leaf- cutting. However, in other portions, they may be strongly crenate or lobed and are often broader at the base. Venezuela to Colombia, south to Bolivia; Brazil. In forests, Huanuco, Junin and Cuzco, 1500-1800 m. 36 ROLLA TRYON Specimens seen: HUANUCO: Pampayacu, Poeppig (LE); Puente Durand to Exito, Mexia 8149 (BM, F, GH, MO, UC, US). JUNIN: Chan- chamayo valley, Schunke 83 (F, US), 508 (F); San Ramon, Schunke A182 (GH,US) ; Enefias, Pichis Trial, Killip & Smith 25760 (¥F, GH, US). cuzco: Valle San Miguel, prov. Convencién, Biies 2086, 2092 (US); Quince Mil, Vargas 10096 (UC). 8. Hypolepis bogotensis Karst. Fl. Columb. 2: 91, t. 147. 1865. Type: a hia ting Bogotensis, 2900-3000 m.”, Karsten; LE! photo GH; Arb Fig. 14. Pune Necuised Sod. Crypt. Vase. Quit. 634. iene Type: “Volcan el Pact 2900 m.”, Ecuador, Sodiro, P! photo G Leaf to 2m. (or more 2) tall, lamina to quadripinnate-pinnatifid or quinquepinnate-pinnatifid, rachis flexuous, subdichotomous toward the d the pi larger than those above, strongly i NG segments densely tomentose to slightly tie ki beneath with multicellular, usually long and tortuous, moniliform, pointed dara these shorter and straighter when the segment is slightly pubescent, ultimate segments rounded to narrowly rounded at the apex; indusium slightly to moder- ately and irregularly erose-fimbriate. The flexuous rachis which is subdichotomous toward the base and the large inequilateral lower pinnae are the most distinctive characters of this species. The penultimate seg- ments are often rather long stalked and triangular. Costa Rica; Venezuela to Colombia, south to Bolivia. At edge of woods and in dense jungle, San Martin and Huanuco, 3500-3700 m. Specimens seen: SAN MARTIN: Bagazan, Stiibel 1071 (B). HUANUCO: Tambo de Vaca, Macbride 4381 (F, US) ; 15 miles ne. of Hudnuco, Mac- bride & Featherstone 2199 (F, NY, US). 4. Hypolepis obtusata (Presl) Hieron. Hedwigia 48: me 1909. (Kuhn, Chaetopt. 347. 1882, not a proper transfer). Fic. 1 Cheilanthes obtusata Presl, Rel. Haenk. 1: 64: t. seh fig. 1. 1826. Type: Mountains of Peru, Haenke, pr Polypodium. fulvescens Hook. & Grev. Hook. Bot. Misc. 2: 239. Wi Type: Huaylluay (Hucayllay), near 2 hee Peru, A. Cruckshanks, ® photo and fragment us!; isotype: G Plecosorus perwvianus Fée. Gen. ni 151, 1852. Type: Peru, Pavom isotype: K! photo GH, US. Hypolepis pteroides Mett. Fil. Lechl. 1: 17, t. 3, figs. 7-13. 1856. Type: St. Gavan, (Rio San Gaban) Peru, Lechler 2152, B! photo GH, fragment us!; isotype: K! photos GH, US, fragment US! FERNS OF PERU 37 Leaf ca. 0.08-1.0 m. tall, erect, lamina bipinnate-pinnatifid to tripin- pubescent beneath with multicellular, long, tortuous, moniliform, pointed trichomes, ultimate segments rounded at the apex; indusium nearly smooth on the edge to moderately erose-fimbriate. Hypolepis obtusata is represented by rather few speci- mens and it is not yet sufficiently known. It is evidently Vee a> TAY we wet > N ye \ “ange. ize, AIRF oy) Lk EX Nr Ay can oe ANY. wy SS FED aa\ BS PAIN eH. 4 oa ae, Bae ve oH jzz A S) i768 Yor, My, Op Wh: LEX eo) Ms SSS SS ny ) Nn ~N \S s ore, aS) uy | Fics: 14-17. Fig. 14 Hypolepis bogotensis: A. small fertile pinnule, X 214, Colom- bia, Pennell & Hazen 10086, GH; B, petiole section, X 2. Colombia, Tryon & Tryon 6117, GH. Fig. 15. H. obtusata: pinna, X 1, Peru, Cruckshanks, GH. Fig. 16. H. hostilis: A, fertile tertiary segment, X 5. Bolivia, Steinbach 9302, GH; B, types of laminar trichomes, enlarged, idem. Fig. 17. tiella Lindeniana: fertile pinnules, X %, Colombia, Kalbreyer 995, GH; B, petiole section, X 4, Jamaica, Watt, Ny; C, petiole section, X 2, Brazil, Brade 9390, NY. 38 ROLLA TRYON allied to H. bogotense by the similar pubescence and non- aculeate axes. It differs from that species in the characters of the rachis and lower pinnae mentioned in the key. The leaves vary greatly in size, the smaller ones having a propor- tionately shorter and narrower lamina. Vargas 6976, per- haps collected when the leaves were rather young, has the upper surface of the segments shortly glandular-pubescent as in H. viscosa. Colombia to Peru. Among rocks, Junin, Cuzco and Puno, 2800-3400 m. Specimens seen: JUNIN: Hucayllay, Cruckshanks (GH, US). CUZCO: Ceaocco, Biies 1399 (US). PUNO: Achopampa, prov. Carabaya, Vargas 6976 (UC); “Tabina”, Lechler 2094 (B); San Gaban (Rio), Lechler 2152 (B). 5. Hypolepis hostilis (Kze.) Presl, Tent. Pterid. 162. 1836. Fic. 16, Map 3 Cheilanthes hostilis Kze. Linnaea 9: 86. 1834. Type: Mission To- cache, upper Huallaga, July, August, 1830, Peru, Poeppig (Diar. 1957), b. Kze. presumably destroyed at Lz; Poeppig, Peru, 1829, K! photo and fragment us! is probably authentic. Hypolepis parviloba Fée, Crypt. Vase. Brésil. 1: 53, t. 20, fig. 1, 1869. Type: near San Gabriel da Cachoeira, Rio Negro, Brazil, Spruce 2119, P! (Herb. Glaziou ex Fée) ; isotypes: K! GH! Us! Hypolepis Buchtienii Rosenst. Fedde Rep. Spec. Nov. 25: 58. 1928. Type: Hacienda Casana, valle Tipuani, Bolivia, Buchtien 7014, s-PA!; isotype: us! Leaf to ca. 2 m. tall, erect or nearly so, lamina quadripinnate to quadripinnate-pinnatifid, rachis straight, it and usually the pinna- rachises aculeolate, lower pinnae not much, if at all, larger than those above, equilateral or nearly so; segments very slightly to densely pubescent beneath with short, more or less moniliform, straightish oF subtortuous, multicellular or few-celled, gland- or clavate-tipped tr- chomes, or sometimes a few trichomes with pointed tips, ultimate s¢8- ments rounded to subacute at the apex; indusium nearly smooth on the edge to definitely erose-fimbriate. The aculeolate rachis and gland- or clavate-tipped tri- chomes serve to distinguish Hypolepis hostilis. The ultimate segments are usually small and somewhat distant so that the leaf-cutting has a more lacy appearance than in the other species. The least pubescent leaves have only a short, glan- dular pubescence while the most pubescent ones have also a mixture of longer, pointed trichomes. A collection from FERNS OF PERU 39 Tingo Maria (Tryon & Tryon 5240) shows a different type of leaf indument in juvenile and adult plants. The smallest leaves have only large, broad whitish, pointed, moniliform trichomes. This perhaps is parallel to the situation in Pteri- dium aquilinum in which the pubescence of the juvenile leaves, in some varieties, is different from that in the adult leaves. Colombia to Bolivia; Brazil. In dense forest or along forest borders, or in cut-over areas, Loreto and San Martin to Puno, 100-1300 m. Specimens seen: SAN MARTIN: Divisoria, Ferreyra 1048 (BM). : San Lor , between mouths of Rio Pastaza and Rio Hual- laga, Killip & Smith 29207 (us); Mishuyacu, near Iquitos, Klug 469, 1389 (F, US). HUANUCO: Tingo Maria, Allard 21678 (us), Tryon & Tryon 5240 (BM, F, GH, U, US, USM). JUNIN: San Ramon, Schunke A181 (us); Chanchamayo valley, Schunke 48 (F, US); Carpapata, G. Kunkel 550 (GH). PUNO: San Gaban (Rio), Lechler 2334 (B). 5. BLOTIELLA Tryon, Contrib. Gray Herb. 191: Wed 1962. Type: Lon- chitis glabra Bory = Blotiella glabra (Bory) T Lonchitis of auths., not L. Terrestrial, the rhizome stout, decumbent (in our species), pubescent, bearing the leaves in a cluster; leaves large, pinnate-pinnatifid to bipinnate-pinnatifid, more or less pubescent, veins anastomosing; sorus marginal, paraphysate, borne in the sinus of segments or small lobes, the sporangia borne on a short vascular commissure, covered by the indusium which is formed from the reflexed, modified margin. —i American species. Kiimmerle, J. B. Monographiae generis Lonchitidis prodromus. Bot. Kézlem. ates 166-188. on, he genera Lonchitis and Blotiella, in Tax. Fern Notes, ani, Contrib: Gray Herb. 191: 93-100, 1962. Blotiella erugee (Hook.) Tryon, Contrib. Gray Herb. 191: 99. 1962. Fie.1 Lonchitis ibiradiaies Hook. Sp. Fil. 2: 56. t. 89A. 1851. Type: Caracas, Venezuela, Linden 543, K! photo GH; isotype: BR photo GH, us! Rhizome trichomes long, of many long cells, these with whitish to brown side walls and brownish cross walls; leaf to 2 m. (or more ?) long, lamina lanceolate in small leaves, ? in large ones, pinnae (with rare exceptions) sessile or short-stalked, the basal pinnules usually reduced, especially on the lower pinnae, 4 orem entire to pinnatifid, 40 ROLLA TRYON obtuse to acute, all parts of the lamina more or less pubescent with long, acicular trichomes, which may be gland-tipped, or not; indusium more or less pubescent. There is a single species of this predominantly African genus in the American tropics. It is evidently rare and local and more adequate collections are needed in order to deter- mine the lamina shape and to understand apparently minor variations in characters of the indument. Costa Rica; Jamaica and Hispaniola; Venezuela to Co- lombia, south to Bolivia. Huanuco. Ss mens seen: HUANUCO: between Hudnuco and Pampayacu, Kanehira 140 (Gu, Us). TRIBE 6, CHEILANTHEAE. 6. Ertosorus Fée, Gen. Fil. 152, t. 13B, f. 1. 1852. Type: E’riosorus scandens Fée = Eriosorus aureonitens (Hook.) Copel. Psilogramme Kuhn, Fest. 50 Jub. Reals. Berl. (Chaetopt.) 332. 1882. Type: Gymnogramma elongata Hook. & Grev. = Eriosorus elonga (Hook. & Grev.) Copel. Terrestrial, the rhizome slender, creeping, pubescent or with bristle- like trichomes, bearing the leaves closely or widely spaced; leaves small to large, pinnate-pinnatifid to quadripinnate-pinnatifid, pubescent o glandular-pubescent, veins free; sporangia borne along a portion of the veins, back of the nearly unmodified margin, indusium and para physes absent. — About 30 American species. The treatment of this genus has been prepared by Alice F. Tryon on the basis of studies preliminary to a revision of the whole genus. Eriosorus is principally a genus of the South American Andes from Bolivia to Venezuela. In Pert it typically grows in the moist and cool zone of the eastern slopes of the Andes between 2500 and 3500 m. Some species have a determinate lamina that becomes fully expanded, while others have an indeterminate lamina that retains 4 coiled bud at the apex. KEY TO SPECIES a. Lamina long-deltoid or sublanceolate, some (or all) of the ne pinnae twice as long as those toward the apex. ....--s-rercsreeseerereeren” 3 b. Leaves erect or subscandent, bi- to tripinnate. ....-..+-++++ eeneseee . “a ce. Pinnae sessile or with stalks less than 1 cm. long, rachis — or nearly so, leaves determinate or the apical bud minute. ----~ FERNS OF PERU . 41 d. Pinnae rigid-herbaceous or chartaceous, vein-ends acute or clavate, both surfaces of the pinnae with abundant, short, capitate, glandular trichomes. 7. a arte d. Pinnae subcoriaceous, vein-ends flabellate, both surfaces of the pinnae with long, acuminate (rarely capitate) richomes e. Pinnae with rigid, bicolorous trichomes on both ee ss and along the margin. 2. E. rufescens e. Pinnae with sericeous, tan trichomes on both surfaces. __... 3. BE. Stuebelii c. Basal pinnae petiolulate, the stalks 1-25 cm. long, pas usually flexuous, leaves with indeterminate growth, the apical buds f. large. f. Lower surface of 'the pinna visible beneath the tan pubescence, vein-ends terminating at the margin. .............. . E. acerescens f. Lower surface of the pinna obscured by the rust-colored to- mentum, vein-ends protruding as a tooth from the margin. .... " E b. Leaves scandent, quadripinnate or more complex. ........cceeeeeeee g. g. Ultimate segments dichotomously divided or rium bifid, usu- ally several times longer than broad, with 1 or 2 veins, pinna- rachises flexuous, the pinnules ascending, or at a ce a to the pinna-rachis. 6. FE g. Ultimate segments ovate to flabellate, about as trond as long, with 4-15 veins, pinna-rachises arcuate, the pinnules mostly descending-curved. 1.E. a. Lamina linear-elongate, the pinnae of nearly the same nied throughout. h. Leaves bipinnate-pinnatifid, the pinnules 1-2 mm. distant, spores usually light amber-colored to light brown, rhizome trichomes with an acuminate apex. 8. E. flabellatus h. Leaves pinnate-pinnatifid, rarely bipinnate, the eee or pinnules imbricate or adjacent, spores dark brown, rhizome trichomes with a bulbous apical cell. 9, E. elongatus 1. Eriosorus Lechleri (Kuhn) A. F. Tryon, Rhodora 65: 56, 1963. Fic. 18. Gymnogramma Lechleri Kuhn, Linnaea 36: 71. 1869. ut San rien, ( Rio ‘Sag Gaban) Puno, Peru, Lechler 2262, B! photo G Peilogr amme Lechleri (Kuhn) Kuhn, Fests. 50 Jub. Reale. ‘Berol. (Chaetopt.) 339. 1882. Rhizome not seen; leaves erect, with indeterminate growth, lamina elongate-deltoid or sublanceolate, acuminate, bipinnate-pinnatifid to tripinnate, rachis castaneous, straight or nearly so, pinnae deltoid, or the apical ones ovate and decurrent on the rachis, the basal petiolulate with stalks less than 0.5 mm. long, slightly shorter than the pair above, pinnules deltoid or ovate, rigid-herbaceous, both upper and lower sur- faces with some long clear trichomes having acuminate apical cells and 42 ROLLA TRYON abundant short trichomes with the apex bulbous, ultimate lobes bifid or crenulate. Vein branches departing at moderately broad angles, the ends extending to, or nearly to, the margin, acute to clavate; spores dark brown. Peru. Cuzco and Puno, 2700 m. Specimens seen: cUzcO: Los Palmitos, Biies 1915 (US). PUNO: San Gaban (Rio), Lechler 2262 (B). 2. Eriosorus rufescens (Fée) A. F. Tryon, Rhodora 65: 56. 1963. Fig. 19. Gymnogramma rufescens Fée, Gen. Fil. 181, t. 19C, f. 3. 1852. Type: “Andibus”, Peru, Mathews. Gymnogramma mohriaeformis Mett. Fil. Lechl. 1: 9. 1856. Type: San Gavan (Rio San Gaban), Puno, Peru, Lechler 2255, B! photo GH, fragment Ny! Gymnogramma Mathewsii Hook. Sp. Fil. 5: 128, t. 290. 1864. Type: Peru, Mathews 1814, K! photo GH. Psilogramme rufescens (Fée) Kuhn, Fests. 50 Jub. Reals. Berol. (Chaetopt.) 336. 1882. Psilogramme Mathewsii (Hook.) Kuhn, Fest. 50 Reals. Berol. (Chaetopt.) 336. 1882. Rhizome long-creeping, dichotomously branching, 2-4 mm. in diam, with rigid, appressed, bristle-like trichomes with 1-5, usually 2, cells at the base, darker than the rhizome surface, the apex acuminate; leaves erect, determinate, lamina elongate-deltoid, acuminate, bip!- nate, rachis castaneous, straight, pinnae deltoid to ovate, subsessile oF with very short stalks, the apical ones decurrent on the rachis, pinnules subdeltoid or ovate, subcoriaceous, upper and lower surfaces with rigid, clear or bicolorous trichomes, acuminate or rarely with a bulbous apical cell, ultimate segments broadly lobed to crenulate, vein branches moderately acute the vein-ends extending to the margin, flabellate; spores usually light amber colored, rarely dark brown. This species is distinguished from the others in Peru by the rigid, bicolorous trichomes. There is considerable varie tion in the size of the leaves and the Lechler collection, which is the type of Gymnogramma mohriaeformis, is a0 unusually small-leaved specimen. Andean, Venezuela to Bolivia. _ On mossy banks, in sphagnum, montaiia, Libertad 10 Cuzco, 2440-3600 m. FERNS OF PERU 43 Fics. 18-20. Fig. 18. Eriosorus Lechleri: A, architecture of lamina, X %. Peru, Lechler 2262, B; B, pinna, X 1, Peru, Biles 1951, US. Fig. 19. EB. rufeecens: A, archi- tecture of lamina, X 14, Peru, Biies 2142, US; B, pinna, X 1, idem, Fig. 20. E. accres- cens: A, architecture of lamina, X 14, Peru. Biies £145, US; B, pinna, X 1, idem. 44 ROLLA TRYON Specimens seen: LIBERTAD: Pumatambo, Puerto del Monte, Lopez & Sagdstegui 3436 (GH). HUANUCO: Cushi, Bryan 683 (US); Pampaya- Convencion, Biies 2142 (US); Prov. Urubamba, Vargas 2897 (Us); Paucartambo, Vargas 12192 (GH). 3. Eriosorus Stuebelii (Hieron.) A. F. Tryon, Rhodora 65: 57. 1963. Gymnogramma Stuebelii Hieron. Hedwigia 48: 219, t. 9, f. 5. 1909. Type: Mojon-Cruz, inter Pacasmayo et Moyobamba, Peru, Stiibel 1058, B! photo and fragment GH! Rhizome not seen; leaves erect, determinate, lamina elongate-deltoid, acuminate, bipinnate, rachis straight, pinnae deltoid to ovate, the apical ones decurrent, the basal ones with very short stalks, nearly subsessile, pinnules ovate, ns aonb a upper and lower surfaces with tan, sericeous tomentum, the trichomes with acuminate apex, ultimate segments crenulate, vein ate moderately broad, the vein-ends flabellate, terminating in a sinus at the margin; spores dark brown (irregular or shriveled). This species is known only from the original collection from northern Peru. The region between Pacasmayo and Moyobamba has been little collected since Stiibel made his trip across the Andes. Eriosorus Stuebelii most closely resembles E. rufescens in the determinate leaves and the deltoid, nearly sessile, pinnae. The tan, sericeous tomentum, particularly on the upper pinna surface, of E. Stuebelii, however, is quite dis- tinct from the rigid, bicolorous trichomes on the pinnae of E. rufescens. Peru. Specimen seen: Mojon-Cruz inter Pacasmayo et Moyobamba, 3300 m., Stiibel 1058 (B). 4. Eriosorus accrescens A. F. Tryon, Rhodora 65: 57, 1963. Fie. 20. bd Puyupata to “Yuncapata”, Prov. Urubamba, Peru, Vargas 2921 Rhizome not seen; leaves subscandent or pendent, with indetermi- nate growth, lamine elongate-lanceolate or elongate-ovate, bipinnate- pinnatifid or tripinnate, rachis castaneous becoming lighter colored toward the apex, somewhat flexuous, pinnae deltoid, petiolulate, the stalks ca. 1.0 em. long, pinnules deltoid or ovate, subcoriaceous, uPPet and lower surfaces with tan, sericeous pubescence, somewhat denser on the lower surface along the veins, the trichomes with acuminate apex, ultimate segments bluntly lobed to crenulate, vein branches de- FERNS OF PERU 45 parting at moderately broad angles, the pining flabellate, terminat- ing in a sinus at the margin; spores dar In the habit and form of the leaves, in the stalked pinnae and the sericeous indument on the pinnae this species most closely resembles FE. aureonitens. The indument is tan and less dense in FH. accrescens and the vein-ends terminate at the margin, while in EF. aureonitens the lamina is covered by a rust-colored tomentum and the veins protrude in a tooth from the margin. Peru. Wooded ravine, Amazonas and Cuzco, 2950-3350 m. Specimens seen: AMAZONAS: entre Leimebamba y Balsas, Lépez et al. 4444 (GH). cuzco: Huadquifia, Biies 992 (US); Valle de Lares, Montafia de Colca, Biies 1925 (Us); Altura de Chaco, Biies 2135 (Us). 5. Eriosorus aureonitens (Hook.) Copel. Gen. Fil. 58.1947. Fie. 21. Gymnogramma aureonitens Hook. Icon. Pl. 9: t. 820. 1852 (prior to May, cf. Gard. Chron. 1852: 278). Type: Peru, W. Lobb, K! photo Gu, fragment Ny! Eriosorus scandens Fée, Gen. Fil. 152, t. 13B, f. 1. 1852 (“probably December,” cf. W. T. Stearn, Webbia 17: 207-222. 1962). Type: Peru, Ruiz. Rhizome not seen; leaves subscandent or pendent with indeterminate growth, lamina jastnniee dor etre bipinnate-pinnatifid, rachis atro- 25 cm. long, pinnules ovate or deltoid, the longer ones with few, blunt lobes, subcoriaceous, upper and lower surfaces with bright rust-colored sericeous tomentum, the trichomes with acuminate apex, ultimate seg- ments bluntly lobed to crenulate, vein branches departing at moderately broad angles, the vein-ends flabellate, protruding in a tooth beyond the margin; spores dark brown The dense, reddish tomentum covering the leaf surfaces clearly distinguishes the species from others in the genus. Heavy tomentum of this type characterizes some species of the allied genus Jamesonia and illustrates a parallel devel- opment of the character in these two genera. Peru and Colombia. Open areas, on clay banks and in woods, Amazonas and Huanuco, 2750-3000 m. 46 ROLLA TRYON Specimens seen: Lobb (K). AMAZONAS: Leimebamba to Balsas road, Wurdack 1738 (GH). HUANUCO: Hudnuco to Pampayacu, Kanehira 152 (GH, US); Playapampa, Macbride 4529 (F, US). 6. Eriosorus flexuosus (HBK.) Copel. Gen. Fil. 58. 1947. Fi. 22, Map 4. Gymnogramma flexuosa Desv. Ges. Naturf. Freunde Berlin Mag. 5: 306. 1811, ex char. Type: none cited. (Desvaux’s epithet, although the earliest, cannot now be transferred to Eriosorus). Grammitis flexuosa HBK. Nov. Gen. Sp. 1: 5. 1816, not HB. Pl. Aequin. 2: 167. 1809, as sometimes cited in error. Type: Venezuela, Humboldt & Bonpland, P! photo GH. \ 22c YY > 22A RAC Fic . - 21-22, Fig. 21, Eriosorus aureonitens: A, ge igeata of lamina, X %4 Pers Mocbride 4529, US; B, pinnule, X 1, idem. Fig. 22. E. flexuosus: A, architecture of central portion of lamina, X x 1/3, accom, Siepermenk isort, us: B, fertile segment, X 214, idem; C, pinnule, * 1 FERNS OF PERU 47 Gymnogramma Ruiziana Kl. Linnaea 20: 410. 1847. Type: Prov. Panatahuarum, Peru, Ruiz 74 B! photo GH. Psilogramme flecuosa (HBK.) Kuhn, Fests. 50 Jub. Reals. Berol. (Chaetopt.) 339. 1882. Gymnogramma flexuosa var. peruviana Hieron. Hedwigia 48: 220. 1909. Type: Cuesta de Lejia, prop. Molinobamba, Peru, Stiibel 1055, B! photo GH. Rhizome slender, elongate, 2-4 mm. in diam. with rigid, appressed, bristle or scale-like trichomes, with 1-10 cells at the base, dark brown to blackish, darker than the rhizome surface, with acuminate, rarely bulbous apical cell; leaves scandent, scrambling on other vegetation, indeterminate, sometimes exceeding 3 m. in length, lamina elongate, branching in several planes, 4-or 5-pinnate, rachis castaneous to straw colored, darker near the rhizome, flexuous, pinnae elongate, broadest near the rachis, petiolulate, pinnules ovate to deltoid, ascending or nearly at right-angles to the pinna-rachis, herbaceous, the upper and lower surfaces glabrous or sparsely pubescent, the trichomes clear or tan with acuminate apex, ultimate segments slender, usually several times longer than broad, bifid to deeply lobed, with 1 or 2 veins, vein branches departing at moderately broad angles, vein-ends not or slight- ly enlarged, extending to, or short of, the margin; spores usually dark brown, sometimes light brown. There is considerable variation in the size of the ultimate segments in this species, although the ratio of length to breadth is relatively constant. The amount of indument on the lamina varies with age and with the location on the leaf. The size of segments also varies in different parts of the lamina. Such variation is difficult to interpret from herbarium specimens for most of them represent only pinnae or portions of the leaf apex. Mexico, Central America; Hispaniola; British Guiana to Colombia, south to Bolivia. Under trees and shrubs in moist areas, along river banks or mossy slopes, scrambling on other vegetation, Cajamarca to Cuzco, 2200-3600 m. 10011 (cH), Tryon & Tryon 5323 (GH, us); Huanuco to Pampayacu, Kanehira 137 (GH), 139, 157 (US); Playapampa, Macbride 4518 (F, Us). HUANCA : Prov. Tayacaja, e. of Surcubamba, Stork & Horton 10393 (F, GH). CUZCO: Rio Urubamba valley, Biies A30, A35 ee Map 4, Eriosorus flexuosus. Map 5, ps Mars 3-6. Map 3, Hypolepis hostilis. elongatus. Map 6, Jamesonia. FERNS OF PERU 49 (US); Valle de Lares, Biies 1906 (Us); Convencién, Biies 2107, 21381 (Us) ; Lucumayo valley, Cook & Gilbert 1359 (US); Pillahwata, Cerro de Cusilluyoc, Pennell 13941 (F, GH, NY); Paucartambo, Vargas 1906 (GH, US). PUNO: entre Ayapata y Kahualluyoc, Vargas 10750 (GH). go Eriosorus Orbignyanus (Kuhn) A. F. Tryon, Rhodora 65: 56. 1963. G. 238. Gymnogramma Orbignyana Kuhn, Linnaea 36: 70. 1869. Syntypes: Yungas, Bolivia, D’Orbigny 174, B! photo GH (dupl. F!); Yorocares, Bolivia, D’Orbigny 299, B! photo GH Gymnogram ehensibilis Relea Syn. Fil. oo = 517. 1874. Type: Sandillano, 8-9000 ft., Ecuador, Pearce, K! photo not seen; leaves scandent, evidently indeterminate, lamina Pr cekenag Sead rachis castaneous, lighter toward the apex, flexuous, pinnae elongate, broadest near ie rachis, petiolulate, pin- nules elongate-triangular, descending in an arc, herbaceous, the upper and lower surfaces sparsely pubescent, the trichomes clear with acumi- nate apex, more dense amo the sporangia, ultimate segments as broad or broader than long broadly lobed to crenulate, with mostly 4-15 veins, vein branches departing at moderately broad angles, the vein- ends acute or clavate, extending to or nearly to the margin; spores dark to light brown. This and the previous species are scandent with large, dissected leaves and in these characters they are superficially allied. They differ, oe in the form of the pinnules and ultimate segments, the position of the pinnules and other characters, an are probably not closely related. Colombia to Bolivi Junin and Cuzco, 1500-27 45 m. Specimens seen: JUNIN: Porvenir, Killip & Smith 25947 (NY, US). cuzco: Chaupimayo, Biies 1949, 1950 (US). 8. Eriosorus flabellatus (Hook. & Grev.) Copel. Gen. Fil. 58. 1947. Fig, 24. Gymnogramma flabellata Hook. & Grev. Jour. Bot. 1: 61, t. 120. 1834. Type: Surucucho, near Cuenca, 9000 ft., Ecuador, W. Jameson, E; isotype: K! photo GH, fragment Us! Psilogramme flabellata (Hook. & Grev.) Kuhn, Fests. 50 Jub. Reals. Berol. (Chaetopt.) 336. 1882. Rhizome slender, elongate or compact, 2-4 mm. in diam., with ri appressed, bristle-like trichomes with 1-4, usually 1 or 2, cells at the base, darker than the rhizome surface, with an acuminate apex; leaves erect, indeterminate, lamina linear-elongate, 2-4 cm. wide, the base sometimes slightly narrowed, bipinnate-pinnatifid, rachis castaneous or 50 ROLLA TRYON Fias. 23-25. Fig. 28. Eriosorua Orbignyanus: A, architecture of pinna, * %, — Biles 1950, v8; B, pinnule, X 1, idem. Fig. 24. E. : lamina (central omitted), X %. Peru, J. B. Steere, us; B, pinnae, X 2, idem. Fig. 26, B. A, lamina, X %4, Peru, Tron & Tryon 5319, aw; B, pinnae, X idem. FERNS OF PERU 51 dark brown, lighter near the apex, straight, pinnae elongate-ovate or deltoid, petiolulate, usually with 4 or more pairs of pinnules 2-4 mm. distant, herbaceous with clear, multiseriate trichomes on both surfaces, longer and curled on the lower surface and usually appressed on the upper, the ultimate lobes bifid, vein branches departing at moderately broad angles, the vein-ends clavate or acute, ending short of the mar- gin; spores light amber-colored or light brown. This species is very close to E. elongatus and perhaps may better be regarded as a variety of it. The final treatment of these taxa awaits a more complete survey of variation throughout their ranges. In the northern portion of the range, EF. flabellatus is more abundant than EF. elongatus, while in Peru the latter is more abundant. The collections cited below are included in EF. flabellatus mainly on the basis of the complexity of the pinnae and the spore color. Colombia to Bolivia. Specimens seen: Contumarca, J. B. Steere (GH, US); Ruiz (G, US). 9, msg elongatus (Hook. & Grev.) Copel. Gen. Fil. 58. 1947. Fig. 25, Map sche ar elongata Hook. & Grev. Jour. Bot. 1: 61, t. 119. 1834. Type: Surucucho, near Cuenca, 9000 ft., W. Jameson, E; isotype: K! photo GH Gymnogramma angustifrons Bak. Syn. Fil. 380. 1868, based on Gymnogramma elongata Hook. & Grev., not Gymnogramma elongata (Sw.) Hook. which is Polypodium astrolepis; an illegitimate new name for the earlier! homonym. Psilogramme elongata (Hook. & Grev.) Kuhn, Fests. 50 Jub. Reals. Berol. (Chaetopt.) 335. 1882. Rhizome slender or compact, 2-4 mm. in diam., with rigid, appressed bristle-like trichomes with 1-4, usually 1 or 2, cells at the base, darker than the rhizome surface, with a bulbous apex; leaves erect, indetermi- nate, lamina linear-elongate, 0.5-2.0 cm. wide, the base and apex some- times narrowed, pinnate-pinnatifid, rarely bipinnate, rachis castan- pinnules which are imbricate, herbaceous, with clear, multiseriate tri- chomes on both surfaces, rarely glandular, often longer and eo on the lower surface and appressed on the upper surface, the lobes bifid, vein branches departing at moderately broad ee the vein-ends acute or clavate, extending to, or nearly to, the segment margin; spores dark brown. 52 ROLLA TRYON The collections from Hudnuco of Kanehira, Macbride, and Stork & Horton, are exceptional in having glandular indument on the lamina. On rocky, foggy or dry slopes and road cuts or in open, grassy puna, Piura to Cuzco, 2500-3900 m. Selected specimens: PIURA: above Huancabamba, Hutchison 1616 (GH); e. of Huancabamba, Weberbauer 6096 (F, GH, US). CAJAMARCA: pass s. of Conchan, Stork & Horton 10072 (US). AMAZONAS: mts. e. 0 Balsas, Osgood & Anderson 84 (F, US); Prov. Chachapoyas, summit of Puma-ureu, Wurdack 1161 (GH). LIBERTAD: Prov. Bolivar, entre Unamen y Bolivar, Lopez & Sagdstegui 3332 (GH); Las Quinuas, Longotea to Bolivar, Lépez & Sagdstegui 3348 (GH). HUANUCO: Pam- payacu, Kanehira 143 (GH, US), 175 (US); Playapampa, Maebride 4523 (F, US); Mito, Macbride & Featherstone 1792 (¥F, US); Carpish, Stork & Horton 9910 (¥, GH), Ferreyra 8172, 10033 (GH), Hodge 6272 (GH); 35 km. ne. of Hudnuco, Tryon & Tryon 5319 (GH). AYACUCHO: Pampalea, Killip & Smith 23249 (F, GH, NY, US); above Zanamonté, Weberbauer 5658 (F, GH, US). Cuzco: Huadquifia, Biies 978, 981 (US) ; Valle Lares, Biies 1777 (US); La Convencién, Biies 2099, 2160 (US); Lucumayo Valley, Cook & Gilbert 1360 (us) ; Cerro de Cusilluyoe, Paso de Tres Cruces, Pennell 13850 (F, GH, NY, US); Paucartambo, Vargas 12239 (GH). Fic. 26. Jamesonia peruviana, Cordillera Vileabamba (Cuzco). FERNS OF PERU 53 7. JAMESONIA Hook, & Grev. Icon. Fil. 2: : 178. 1830. Type: Jame- sonia pulchra Hook. & Grev. FIG. 26, MAP Terrestrial, the rhizome slender, creeping, pubescent or with bristle- like trichomes, bearing the leaves closely or widely spaced; leaves small to medium sized, pinnatisect to usually 1-pinnate, glabrous, glandular, pubescent or glandular-pubescent, veins free; sporangia borne along a portion (or most) of the veins, hardly or not covered by the reflexed, usually modified mere in Hogar and paraphyses absent. — An American genus of 19 spec Tryon, A. F., A monograph of the fern genus Jamesonia. Contrib. Gray Heth. 191: 109-197. 1962. The treatment of this genus has been adapted by Alice F. Tryon from the monograph cited above. Jamesonia is a genus mainly of the paramos of Ecuador to Venezuela, where it is a characteristic element of the flora, and of the puna of Peru and Bolivia. The genus is characterized by the linear, elongate, often densely pubescent, leaves with indeterminate gro In most species the margin is usually strongly enrolled with a modified membranaceous border which is entire or variously dentate or ciliate. This genus has clearly evolved from the previous one, Eriosorus, or from a common ancestor, and the two genera are not sharply differentiated. The unusual morphology and ecology of Jamesonia, however, justify their separation. KEY TO SPECIES a. Pinnae asymmetrical, inequilateral and usually cordate at the base; spores tan or light brown . b. Lower pinna surface with dense, matted tomentum. ................0.0. c. Apex of the lamina vermiform, with appressed ssichitied: glandular and usually vernicose. ..........:..ss:s0000+- 2. J. Scammanae ce. Apex of the lamina about the same width or broader than the lower portions, a fh ese to slightly appressed trichomes, pubescent or tom d. d. Pinnae, ethers the older ones, convex on the upper surface (patelliform), glabrous or somewhat pubescent, the stalk at an oblique angle to the rachis. ..............0.... . 1. J. rotundifolia the plane of the pinna, meinehce glandular on the upper surface, the stalk perpendicular to the rachis. mpi J. imbricata var. imbricata 54 ROLLA TRYON b. Lower pinna surface glabrous, glandular or with sparse trichomes. e e. Pinnae, especially the older ones, convex on the upper surface (patelliform), the stalk often bent, at an oblique angle to the rachis. 1. J. rotundifolia e. Pinnae plane or slightly concave on the upper surface, the stalk straight, nearly perpendicular to the rachis, «1... f. f. Pinnae usually of a rigid-herbaceous texture, the upper sur- ce conspicuously pubescent, with the epidermal cells un- thickened; the lamina 0.3-0.7 em. wide. 2 g. Upper pinna surface with ea appressed trichomes, espe- cially on the distal portions, these usually enveea several pinnae; the pinna margin with a wide, more 0. less papillate, border several cells broad. 8. J. perio g. Upper pinna’surface with short, erect, rigid, capitate tri- chomes; the pinna margin with a narrow border usually 1 cell broad, with cilia similar to the trichomes. ......--..-:-+-+ OJ; bokvianall f. Pinnae coriaceous, the upper pinna surface glabrous, rarely with appressed glands or a few trichomes, with the Baer cell walls thickened; the lamina usually about 1 em. wide. . h. Lamina about the same breadth throughout or mee narrower near the base; tomentum of the bud and rachis concolorous; the pinna margin with an contiguous border usually several cells broad. 4, J, Alstonit h. Lamina clavate, broadest near the apex; tomentum of the apex and rachis bicolorous with a prominent dark streak; the pinna margin usually with an irregularly dentate border extended at the vein ends, not contiguous or scarcely so. 5. 3. G oudotit sag icles symmetrical, the base equilateral; spores dark brown. . . Pinnae with poecaed stalks about 1 mm. long (or longer) and ye "egies Or Sire PAPI i Oe A ee j. Pinna base hie lower surface of the pinna glandular oF with a few short trichomes, the border narrow, usually of a single row of or ciliate, veins at wide angles, the branches often short. 7.3.8 calaris . Pinna base a lower surface of the pinna usually — tose (to rarely glabrous), the border usually broad, dentate an or ciliate, veins usually at acute angles, the branches long and somewhat parallel, _ 8b. J. imbricata var. glutinos? i. Pinnae adnate, subsessile, or with short broad stalks less than l mm. long and at an oblique angle to the rachis, veins at acute angles, the branches long and nearly parallel. ........ 9, J. blepharum el FERNS OF PERU 55 . Jamesonia rotundifolia Fée, Mém. Fam. Foug. 7: 41, t. 10. 1857. eee Colombia, Schlim 363; isotypes: G! K! P! photoGH. Fic. 27. Rhizome trichomes castaneous or light brown, with 1 (-4) cells near the base; lamina ca. 14-65 em, long, 0.7-1.5 cm. wide, about the same width throughout, the bud with trichomes patent or slightly appressed; pinnae rotundate-cordate, patelliform, especially the older pinnae, the margins enrolled, inequilateral at the base, herbaceous; upper surface more or less pubescent, especially the distal portion, the epidermal cell walls thickened, lower surface pubescent with tan or clear trichomes or glabrous, stalk bent at an oblique angle to the rachis, veins at acute to moderately wide angles, the branches long, border moderately broad, extended at the vein ends, irregularly dentate, sometimes little modi- fied and similar to the pinna in texture; spores light ‘tan. This species represents one of the least specialized forms in the genus and can be readily distinguished by the large, straw-colored tomentum of the apical buds and the rotun- date-cordate pinnae which have a patelliform shape, par- ticularly in the older pinnae. It is known in Peru from two widely disjunct collections; the Hudnuco station is the most southern extension of the species. It is expected that other records will be found since the species is one of the com- moner ones in Colombia. Costa Rica; Colombia to Peru. Grassy steppe, Piura and Huadnuco, 3200-3500 m. Specimens seen: PIURA: e. of Huancabamba, Weberbauer 6067 (B, F, GH, US). HUANUCO: Cushi, Bryan 612 (F). 2. Jamesonia Scammanae A. F. Tryon, Contr. Gray Herb. 191: 164. 1962. Type: Cerro de la Muerte, Costa Rica, Scamman & Holdridge 7929, GH! Fic. 28. Rhizome — lustrous, light to dark brown, usually 1 (-5) cells near the base; lamina ca. 11-45 em. long, 0.2-0.6 em. wide, about the same size Pinched. the bud pcsnly sae glutinous, with appressed trichomes; pinnae ovate, sometimes lobed, usually with a central fur- row on the upper surface, the margin enrolled, inequilateral or nearly so at the base, upper surface glandular and vernicose or crustose, the epidermal cell walls unthickened, lower surface tomentose with whitish or tan trichomes, stalk bent, short to nearly absent, veins with broad to moderately broad angles, the branches moderately long, border moderately broad, entire or slightly undulate, firm-membranous or similar to the pinna texture; spores pale amber-colored or tan. This species is readily recognized by the vermiform apex of the leaves and the nearly subsessile pinnae which are vernicose or crustose on the upper surface. 56 ROLLA TRYON Costa Rica ; Colombia to Bolivia. In grassy uplands and among rocks, Lima and Hudanuco to Puno, 2750-4270 m. Specimens seen: HUANUCO: Hudnuco, Macbride & Featherstone 2182 (F, GH, US); Mito, Bryan 183 (F, US). LIMA: Huarochiri, Sawnders 399 (BM). JUNIN: Acopalca, prov. Huadncayo, G. Kunkel 485 (GH). APURIMAC: Abancay region, Santander et al. in 1935 (uc); Bosques de Ampay, Vargas 1060 (GH). CUZCO: Salleantay, Biies 1008 (Us) ; Pinas- niocj, Cook & Gilbert 1242, 1833 (US); Cerro de Cusilluyoc, Pennell 13870 (F, GH, US); Paucartambo, Soukup 386 (F); Achirani, Paucar- tambo, Vargas 11165 (F, K, UC) ; entre Lares y Hierbabuinayoc, Vargas 11999 (GH). PUNO: “Tabina,” Lechler 2032 (B, E, K, LE, P). 3. Jamesonia peruviana A. F. Tryon, Contr. Gray Herb. 191: 167. 1962. Type: Tambo de Vaca, Peru, J. F. Macbride 4404, F! Figs. 26, 29. Rhizome trichomes castaneous or atropurpureous, with 1 or 2 cells near the base; lamina ca. 17-40 em. long, 0.3-0.5 cm. wide, broadest at the apex, the bud with tan or light brown patent or somewhat ap- pressed trichomes, pinnae reniform or orbicular-cordate, concave on the upper surface, the margins enrolled, inequilateral at the base, rigid- herbaceous or slightly coriaceous, upper surface with long, discrete trichomes, strongly bent near the base of the trichome, appressed and usually enveloping several pinnae, with epidermal cell walls sometimes slightly thickened, lower surface with sparse, short, capitate, clear or tan trichomes and a tuft of longer trichomes on the stalk and adjacent veins, sometimes pubescent along all veins, stalk straight, perpendicu- lar to the rachis, veins with wide angles, the branches short or moder- ately long, border broad, with more or less uniform papillae firm- membranous; spores tan or straw colored. Jamesonia peruviana is a unique species having long, appressed trichomes on the upper surface of the pinnae and a rather uniform fringe of papillae on the border. The northernmost collection from Libertad differs from the others in having rather dense pubescense along the veins on the lower pinna surface and bulbous tipped rachis trichomes. Although the species is known from several collections, the early ones are incomplete and without locality data and most recent collections lack the rhizome. Ample and complete specimens of this species are much needed. eru and Bolivia. In open grassy slopes, jalea, and in wet places, Amazonas to Puno, 3400-4300 m. FERNS OF PERU 57 AN HN TANG z . : aA. 13873, GH; B, pinna margin (two vein ends), X 35, gaia Pies Sh obs asaeiigge pinna, lower surface, X 10, Colombia, Cuatre ee Naas anwe 16886. vein ends) (heavy lines indicate fold in the tissue), X 35, oe : GH, 58 ROLLA TRYON Specimens seen: Dombey 18 (G, L, P), Pavon 137 (G). AMAZONAS: Prov. Chachapoyas, summit of Cerros de Calla-Calla, Wurdack 1216 (GH). LIBERTAD: Las Quinuas, Lopez & Sagdstegui 3346 (GH). AN- CASH: Huari, above Ponto, Weberbauer 3302 (B, G, US). HUANUCO: Tambo de Vaca, Bryan 648 (F, US). JUNIN: Tarma, Née (F); Muna, Pearce, in May 1863 (kK). cuzco: Biies 1530, 1531 (US); Vileabamba, Biies 1598 (US); Convencién, Biies 2161, in part (US); Cerro Cusil- luyoc, Pennell 13873 (F, GH, US); Tres Cruces, prov. Paucartambo, Vargas 12240 (GH). PUNO: Lechler 2153, in part (B, E, G, K, LE, P, S-PA). 4. Jamesonia Alstonii A. F. Tryon, Contr. Gray Herb. 191: 168. 1962. Type: Los Farallones, del Valle, Colombia, Cuatrecasas 21884, GH! Fig. 30. Rhizome trichomes usually dark brown or blackish, sometimes light brown, 1 (-4) cells near the base; lamina ca. 15-40 em. long, 0.4-1.0 em. wide, about the same width throughout or the base slightly narrowed, the bud with straw colored, matted or slightly appressed trichomes; pinnae ovate-cordate or orbicular-cordate, plane, the margins enrolled, inequilateral at the base, coriaceous, the upper surface glabrous, rarely glandular with epidermal cells thickened, lower surface with sparse short clear or tan trichomes and usually with a tuft of longer trichomes on the stalk and adjacent veins, stalk straight, perpendicular to the rachis, veins with wide angles, the branches short or moderately long, border moderately broad, extended at the vein ends, dentate, with sparse, short, bulbous cilia or entire, rigid-herbaceous; spores light amber-colored. Jamesonia Alstonii most closely resembles J. Goudoti par- ticularly in the coriaceous texture of the pinnae. There are several collections of it from Peru which are mixed with J. peruviana and these two must grow together. The collection of Stork and Horton from Libertad is exceptional in having dense glands on the upper pinna surface and very strongly enrolled margins. Southernmost Mexico, Guatemala and Costa Rica; Colom- bia to Bolivia. Under boulders in dry soil and among bunch grasses 00 wet slope, Libertad to Puno, 3355 and 4000 m. Cerrate 2547 (GH). HUANUCO: Mito, Macbride & Featherstone 1888 (F, G, US). JUNIN: Huayllay, Mathews 979 (, GL, K). cuzco: Bues 1378 (US); Prov. Urubamba, camino a Savacmarca, Aug. 1941, © Dreyfus (GH). PUNO: Lechler 2158, in part (B, E, G, K, LE, P, S-PA). FERNS OF PERU 59 5. Jamesonia Goudotii (Hieron.) C. Chr. Ind. Fil. 373. 1905. Fue. 31. Gymnogramme Goudotii Hieron. Engl. Bot. Jahrb. 34: 476. 1904, Type: Colombia, Goudot, B! photo GH; isotype: G! photo GH. Rhizome trichomes light brown to lustrous black, with usually 1 (-3) cells near the base; lamina ca. 5-40 em. long, 0.5-1.2 em. wide clavate, broadest at the apex, narrowed toward the base, the bud bicolorous, with erect or slightly appressed trichomes, these light brown and with a patch or streak of brown trichomes: pinnae reniform or orbicular- cordate, plane, or the margins incurved, sometimes strongly so, forming a pouch-like structure, inequilateral at the base, coriaceous, the upper surface usually glabrous, sometimes glandular, with epidermal cell walls thickened, lower surface with sparse, short, clear or tan, bulbous trichomes and a tuft of longer trichomes on the stalk and adjacent veins, stalk straight, perpendicular to the rachis, veins at wide angles, the branches moderately long, border irregularly dentate, usually ex- tended at the vein-ends and ciliate, rarely entire or nearly so, rigid- herbaceous; spores tan or light amber-colored. This species most closely resembles J. Alstonii and there are only slight differences between them in the form of the pinnae, the elaboration or the pinna margins and the color of the tomentum. At the Market in Huancayo a few leaves of J. Goudotii were obtained by Kunkel (486, GH). In boggy places between rocks on a glacier, Junin and Cuzco, 3660-4500 m. Specimens seen: JUNIN: Hacienda Runatullu, near Comas, Weber- bauer 6629 (F, GH, US). CUZCO: Biies 1406 (US 6. Jamesonia boliviensis A. F. Tryon, Contr. Gray Herb. 191: 174. 1962. Type: Bosques de Ampay, Apurimac, Peru, Vargas 1060a, GH! Fig. 32. Rhizome trichomes lustrous gs aa or blackish, 2 or usually m. long, 0.3-0.7 cm. wide, the same plane, the margins incurved, inequilateral at the base, rigid-herbaceous, ith short, clear, capitate trichomes dermis slightly thickened, stalk straight, perpendicular to the rachis, veins with wide angles, the branches short, or moderately long, border arrow, a row of cells mostly broader than long with capitate cilia; spores light amber-colored. 60 ROLLA TRYON This species is unique in having the same form of indu- ment on both surfaces of the pinnae and on the margin. The collections are mainly from Bolivia although the most com- plete and ample material, from César Vargas, has been designated as the type. Peru and Bolivia. “Woods,” Apurimac, 3200 m. Specimens examined: APURIMAC: Abancay region, Oct. 1935, V. San- tander et al. (UC); Bosques de Ampay, Vargas 1060a (GH). — oy fe KA ie iS ote Fi aye at epee ‘ 1 wey aN Pee A Ari hays = 4 ENG: Lp AR MN} Hy “ive WY ¥ 2, m, pinna, lower surface, x 10. P ae vein , » Peru, Var ; i margin (two ends), X 35, idem. gas 1060a, GH; B, pinna FERNS OF PERU 61 7. Jamesonia scalaris Kze. Bot. Zeit. 2: 738. 1844. Type: Pillao, Peru, Ruiz 49,B! photo GH. Fie. 33. Rhizome trichomes light to dark brown or atropurpureous, usually with 2 (-5) cells near the base; lamina ca. 10-33 cm. long, 0.2-0.5 em. wide, about the same size throughout or the base slightly narrowed, the bud with tan or rust colored, patent or slightly appressed trichomes; pinnae ovate-cordate, usually lobed, patelliform, with the upper surface often depressed near the stalk, the margins enrolled, equilateral or nearly so at the base, herbaceous, the upper surface with papillate glands, the epidermal cell walls unthickened, lower surface with erect, clear or tan, capitate trichomes sometimes tomentose and usually with a tuft of trichomes on the stalk and adjacent veins, stalk straight or slightly bent perpendicular to the rachis, veins with wide angles, the branches short or moderately long, border narrow, a row of cells mostly broader than long and with capitate cilia; spores dark brown. This species is readily distinguished from the others in Peru with small pinnae, by the patelliform and lobed form of the pinnae, the glands on the upper surface and the dark spores. It appears to grow with J. peruviana and J. Scam- manae for these species are sometimes mixed in collections. Peru and Bolivia. In sphagnum, montafia on moist slopes, Libertad to Puno, 2750-4270 m. pecimens seen: Pacechac, Hill 553 (K). LIBERTAD: Pumatambo, Puerta del Monte, prov. Pataz, Lépez & Sagdstegui 3435 (GH). HUANU- co: Punta de Panao, Asplund 13718 (GH, S); Playapampa, sient 4524 (F, US). PASCO: Goyllarisquisca, Asplund 11922 (GH, S). JUNIN: arma, Pavon 137 (G). CUZCO: Biies 1533, 1788, 1784, 1931, 2163 (US); La Convencién Biies 2161, in part (US) ; Cochapatu, Biies 2190 (US); Marcapata, Stafford 989 (K); Paucartambo, June, 1937, Vargas (GH). PUNO: Ayapata, Lechler 2036a (B, E, G, K, LE, P). 8. Jamesonia imbricata (Sw.) Hook. & Grev. Icon. Fil. 1: 2. 1831. Rhizome trichomes tan or ruddy brown to blackish, with 1 (-3) cells near the base; lamina ca. 7-80 cm. long, 0.2-1.2 cm. wide, about ¢ : Same width throughout, the bud with tan, brown or bicolorous, 2908 or appressed trichomes; pinnae orbicular, ovate, patelliform, or t : upper surface plane or sunken, the margins enrolled, equilateral an truncate and incurved at the base or slightly inequilateral and some- what cordate, rigid-herbaceous, upper surface glandular sometimes al with a few trichomes, rarely glabrous, the cell walls unthickened, lower surface tomentose with clear, white or rust colored trichomes, stalk usually bent or twisted sometimes nearly straight, perpendicular to the 62 ROLLA TRYON rachis; veins with acute angles and the branches long, or with wide angles and moderately long; border moderately broad, entire or den- tate and sparsely ciliate; spores tan or light to dark brown. The species is known with certainty in Peru only from recent collections of var. glutinosa in Amazonas. Variety imbricata is recorded from Peru by the type collection of J. Jussieu but the locality for this material is not certain. A third variety, var. meridensis, occurs in the Andes of Vene- zuela. Peru, Figs. 33-35. Fig. 38. Jamesonia scalaris: A, pinna, lower surface, * 10, Asplund 13718, GH; B, pinna margin (two vein ends), X 35, Peru, Biies 2163, us. Fig. 84. J. imbricata var, imbricata: A, pinna, lower surface, X 10, Peru, ussieu, Pi ge margin (two vein ends), X 35, idem. Fig. 35. J. blepharum: A — X 10, Peru, Kanehira 148, GH; B, pinna margin (three vein ends), FERNS OF PERU 63 8a. Jamesonia imbricata var. imbricata Fic. 34. Pteris orbiculata Poir. Lam. Encyel. 5: 710. 1804, not Houtt. Nat. Hist. 14: 108. 1783. Type: “Pérou”, J. Jussieu (Herb. Jussieu no. 1332) P! photo GH. Pteris imbricata Sw. Syn. Fil. 102. 1806, based on Pteris orbiculata Poir. Rhizome trichomes amber colored or light tan, lighter than the rhizome surface; lamina indeterminate, the bud of the same size or larger than the mature pinnae with tan or light brown, patent tri- chomes; pinnae orbicular or ovate, the upper surface plane or sunken, slightly inequilateral at the base and somewhat cordate, the upper slightly bent, veins with wide angles, the branches widely diverging, moderately long, border rust colored or tan; spores tan or light brown, sometimes shriveled This variety is known from Peru only by the type collec- tion. There is a rather unusual combination of characters in this material and a few other collections from Colombia and Kcuador are identified as this variety. Peru, Ecuador and Colombia, 3385-3400 m. Specimen seen: Herb. Jussieu 1332 (P). 8b. Jamesonia imbricata var. glutinosa (Karst.) A. F. Tryon, Contrib. Gray Herb. 191: 182. 1962. Jamesonia glutinosa Karst. Fl. Columb. 2: 85, t. 143. 1865. Type: monte Guadalupé, Bogota, Colombia, H. Karsten, LE! photo GH. Rhizome trichomes ruddy to dark brown, the same color or more ruddy than the rhizome surface; lamina usually indeterminate, the bud usually smaller than the mature pinnae with tan, brown or usu- ally bicolorous, appressed or patent trichomes; pinnae orbicular or ovate, patelliform, with the base equilateral, truncate and incurved adjacent to the stalk, upper surface usually glandular and vernicose, rarely glabrous, lower surface with clear, opaque, white or tan tri- chomes, sometimes nearly glabrous, stalk strongly bent, veins with acute angles, the branches long and somewhat parallel, border whitish or tan; spores dark brown. This variety is relatively widespread in the Andes from Venezuela to Colombia but has just recently been discovered in northern Peru. It has undoubtedly been overlooked earlier for it is noted to be locally abundant. Plants in Venezuela and Colombia are quite variable, particularly where they 64 ROLLA TRYON occur with other species. Characters such as rhizome indu- ment, the size, shape, border and indument of the pinnae, which are fairly constant in other species, are variable in these plants. The bent pinnae stalks, the patelliform shape of the pinnae and their equilateral base with truncate mar- gins at the stalk and the dark brown spores best characterize this variety. Venezuela to Colombia, south to Peru. In open scrub forest and at the base of rocks, Amazonas, 3100-3300 m. Specimens seen: AMAZONAS: Prov. Chachapoyas, Cordillera Calla- calla, Leimebamba to Balsas, Lépez, et al. 4447, 4448 (GH); Summit of Puma-urcu, se. of Chachapoyas, Wurdack 1160 (GH). 9. Jamesonia blepharum A. F. Tryon, Contrib. Gray Herb. 191: se 1962. Type: Pampayacu, Huanaco, Peru, Ryozo Kanehira 148, GH: Fig. 35. Rhizome trichomes lustrous, light to dark brown or atropurpureous, with 1 or 2 cells near the base; lamina ca. 10-55 cm. long, 0.5-1.0 cm. wide, the lamina nearly the same width throughout or the base and apex more slender, sometimes determinate, the bud with dark brown, appressed, glutinous trichomes; pinnae orbicular, entire, usually patel- This species is readily distinguished from others in Peru by the nearly adnate pinnae with long nearly parallel veins and ciliate margins, Colombia to Bolivia. : Huanuco and Cuzco, 2900 m. and 4575 m. respectively : . : ia Specimens seen: HUANUCO: Carpish, entre Hudnuco y Tingo pene Ferreyra 6702 (GH). cUzcO: Biles 1824, 1914 (Us); Valle de Lares Biies 1908 (us). - 202. 8. PTEROZONIUM Fée, Mém. Soc. Mus. Hist. Nat. Strasbourg ‘ He 1850. Type: Gymnogramma reniformis Mart. = Pterozonium reni (Mart.) Fée. FERNS OF PERU 65 Terrestrial, the rhizome small to moderately stout, short-creeping, densely pubescent, bearing the leaves in a rosette or loose cluster; leaves small, the lamina entire, reniform to orbicular-cuneate, glabrate, veins free; sporangia borne for a short distance along the veins, form- ing a band well back of the hardly modified margin, intermixed with short trichomes, indusium absent.— A South American genus of 5-7 species Pterozonium reniforme (Mart.) Fée, Gen. Fil. 178, 385. 1852. Fic. 87 Gymnogramma reniformis Mart. Icon. Crypt. Bras. 88, t. 26. 1834. Type: Mt. Cupati, Rio Japura, Brazil, Martius. Rhizome trichomes yellowish brown to reddish brown, many cells long, one cell wide, gland-tipped; leaf ca. 5-20 em. long, the long petiole dark reddish brown to blackish, with a pale ridge on each side extend- ing down from the base of the lamina; lamina coriaceous, the outer margin crenulate, with a cartilaginous border. Brazil and Peru. Very rare on acidic rocks in San Martin and western Loreto, 1000 m Specimens seen: SAN MARTIN: Monte Guayrapurima, Tarapoto, Spruce 4641 (B, GH, K, NY). LORETO: Cerro de Isco (evidently between Yurimaguas and Tarapoto), Ule 6887 (B, P); Cuesta de Yento, near Balsa Puerto, Raimondi 25 (B). 9. PITYROGRAMMA Link, Handb. Gewiichse, 3: 19. 1833. Type: Acros- grad chrysophyllum Sw. = Pityrogramma chrysophylla (Sw.) Link. 1G eh Ae Link, Fil. Sp. Cult. 141. 1841. (nom. superfl., illegit.) ria Fée, Mém. Fam, Foug. 5 (Gen. Fil.): 164. 1852. Type: Trismeria aurea Fée = Acrostichum trifoliatum L. = Pityrogramma trifoliata (L.) Tryon. Terrestrial, the rhizome small to often stout, erect or decumbent, Sealy, especially at the apex, bearing the leaves in a crown or cluster; leaves of medium size to large, 1-pinnate to tripinnate, usually with white or yellow ceraceous indument beneath, less often pubescent, rarely glabrous, veins free; sporangia borne along the veins, somewhat back of the nearly cannodified margin, at maturity often confluent over most of the surface, indusium and paraphyses absent. — 12 American species. Domin, K. The hybrids and garden forms of the genus Pityrogram- 929. ma (Link), Rozpr. II. Tr. Ceské. Akad. 38(4). 1 Tryon, R. Pi a (including Trismeria) and Anogramma, Tax. Fern Notes II, Contrib. Gray Herb. 189: 52-76. 1962. 66 ROLLA TRYON Some of the species grow in open habitats and on road banks and in other disturbed places. Many are adaptable to horticulture and were widely cultivated in the 19th century as ornamentals. The species with white or bright yellow indument were especially popular and were known as silver ferns and gold ferns. A somewhat confused account of the numerous horticultural forms is found in Domin’s publica- tion. hy 5, ris sie HS 3 ot Fic. 36. Pityrogramma calomelanos, Iquitos (Loreto). FERNS OF PERU 67 KEY TO SPECIES a. Secondary and tertiary segments (if present) moderately, if at all, toothed, or with broad lobes with several veins. b. Vascular bundles of the petiole (at about 1/3rd the distance to the lamina) roundish, oval to C-shaped; pinnae with numerous, pinnately arranged lobes or segments. c. c. Petiole, primary rachis and secondary gaits glabrous, cer- aceous or thinly and irregularly pubescen d. Rachis and petiole deep reddish OCI te blackish; scales of the apex of the rhizome and the base of the petiole with a long portion one cell wide below the terminal cell. .........:.-.+.- e. e. Pinnae equilateral, the pinnules ascending, those on the basiscopic side more strongly so than those on the acro- scopic side; scales of the apex of the rhizome and the base of the petinls with a long-conical terminal cell. EERE om ealamalinide e. Pinnae inequilateral, the pinnules, on both sides of the pinna, at nearly right-angles to the pinna-rachis (or in large leaves the tertiary segments so disposed on the pinnule-rachis) ; scales of the apex of the rhizome and the base of the petiole with a spherical terminal cell. ..........0+0+ . P. tartarea d. Sasniy and the upner portion of the petiole. or all of it except the base, straw colored to light brown; scales of the apex of the rhizome and the base of the petiole with only a short portion one cell wide below the spherical terminal cell. ............ 3. P. chrysoconia e. Primary rachis and secondary rachises densely and persistently lanate, the petiole similar but sometimes deciduously lanate. ...... ne 4g ‘pine ea b. Vascular bundles of the petiole (at about 1/3rd the distance to the lamina) C-shaped with the back of the C curved forward. ........ tees P. trifoliata and hybrids of it. a. Secondary and wets segments ety laciniate into linear lobes, each with one vein 6. P. Pearcei 1. Pityrogramma calomelanos (L.) Link, Handb. Gewichse 3: 20. 833. Rhizome scales (and those at the base of the petiole) with a long portion, one cell wide, below the long-conical terminal cell; leaf ca. 0.5-1 roseate ceraceous, or pubescent. 68 ROLLA TRYON Two of the varieties, awreoflava and ochracea usually grow in the Andes above 1500 m., while var. calomelanos typically grows below 1000 m. It is widely distributed through the American tropics and seems to grow more rapidly and is more often an occupant of disturbed habitats than the other varieties. Some specimens of P. calomelanos and P. tartarea are not easily distinguished and some characters additional to those presented in the key to species are mentioned under the latter species. la. Pityrogramma calomelanos var. calomelanos. Fics. 36, 38, Map 7. Acrostichum calomelanos L. Sp. Pl. 2: 1072. 1753. LINN sheet 1245.19, photo A, GH, is this species. Gymnogramma calomelanos var. denudata Harr. Jour. Linn. Soc. Bot. 16: 37. 1877. Type: Pébas, Peru, Steere, K! photo GH. Lamina ceraceous beneath, the wax white to pale (lemon) yellow or pale roseate; or rarely the lamina glabrous. Southern Florida; Greater and Lesser Antilles; Mexico to Panama; French Guiana west to Colombia, Ecuador and Galapagos Islands south to Bolivia and Argentina; Paraguay and Brazil. Cleared hillsides, road banks, river banks, thickets and sometimes in forests, Tumbes to Cuzco, 100-1500 m., most commonly below 1000 m. Selected specimens: TUMBES: between Tumbes and Caucho, ps mado 224 (GH, UC). AMAZONAS: Aramango, prov. Bagua, Lopez et a 4161 (GH). SAN MARTIN: 4 miles e. of Tarapoto, Woytkowski 352. : (uc); Lamas, near Tarapoto, L. Williams 6360 (F, GH). , Caballo-cocha, L. Williams 2201 (F, GH); Santa Ana, upper Rio phage L, Williams 1257 (F); above Pongo de Manseriche, Mexia 6101 (F; " ’ UC) ; Iquitos, Tryon & Tryon. 5167 (BM, F, GH, U, US, USM) ; peg roe” Padre Abad, Woytkowski 34379 (uc); Yurimaguas, Killip & Sm me 27961 (US); Gamitanicocha, Rio Mazan, Schunke 224 (F, GH, UC, an Hacienda San Jorge, 55 km. w. of Pucallpa, Ferreyra 13008 (GB . F, US); Pampayacu, Kanehira 145 (GH, US); Pozuzo, Maebride 4616 (F; Us); Villeabamba, Macbride 4988 (Fr, Us). JUNIN: San Ramon, FERNS OF PERU 69 nado 267 (UC); Puerto Yessup, Killip & Smith 26254 (F, us); La Merced, Killip & Smith 23389 (US). AYACUCHO: Rfo Apurimac valley, near Kimpitiriki, Killip & Smith 22958 ( Us). Cuzco: Hacienda Cadena, valle de Marcapata, Scolnik 945 (US); Valle de Marcapata, Herrera 1199 (US); Machu-Piechu to Quillabamba, Mexia 8089a (uC). 1b, Pityrogramma calomelanos var. aureoflava (Hook.) Bailey, Man. Cult. Pl. 64.1926. Maps. Gymnogramma. calomelanos var. aureoflava Hook. Gard. Ferns, t. 50 text. 1862. Lectotype: Seeman 948, K. Pityrogramma austroamericana in, Publ. Fac. Sci. Univ. Charles 88: 7, 1928. (Also Kew Bull. 1929: 221.) Lectotype: Bolivia, Mandon 1549 bis, K! photo GH; isotype: GH! Pityrogramma calomelanos var. austroamericana (Domin) Farw. Am. Midl. Nat. 12: 280. 1931. Lamina ceraceous beneath, the wax bright yellow to orange-yellow. Costa Rica; Venezuela, Colombia, Ecuador and Galapagos Islands south to Bolivia and northwestern Argentina; Bra- zil. Maps 7-8. Map 7, Pityrogramma calomelanos var. calomelanos. Map 8, P. calo- nos var. aureofilava. 70 ROLLA TRYON Open, rocky places, grassy slopes, clay banks, thickets and rarely along irrigation ditches (Libertad), Lambayeque and Cajamarca to Cuzco, 680-2200 m. Specimens seen: LAMBAYEQUE: 32 km. from Olmos, on road to Jaén, Correll & Smith P831 (GH); 25 km. from Olmos, on road to Jaén, Correll & Smith P799 (GH). CAJAMARCA: entre Abra de Porculla y Jaén, Ferreyra 13638 (GH, USM). LIBERTAD: Hacienda Mochal, prov. Trujillo, Sagdstegui 425 (GH); Campifia de Moche, Lépez 1311 (GH); Huaranchal, Sagdéstegui 94 (GH). HUANUCO: between Chinchao and Puente Durand, Coronado 89 (GH, UC); Yanano, Macbride 3667 (F, US). JUNIN: La Merced, Soukup 2368 (F, GH), 3410 (GH, US), Killip & Smith 23801 (F, US); Colonia Perené, Killip & Smith 24984 (US). AYACUCHO: Aina, between Huanta and Rio Apurimac, Killip & Smith 22697 (F, GH, US). CUZCO: Pumachaca, valle de Santa Ana, Herrera 3288 (GH, US); Torontoy, Herrera 1333 (US); Valle de Lacco, Herrera 2072 (US); Santa Rosa, Urubamba valley, Cook & Gilbert 1728 (US); Machu-Picchu, Vargas 2154 (us), Coronado 106 (UC); valley of the Sambray, Mexia 803 (Us); near Quillabamba, Mexia 8045 (F, GH, US); Potrero, 8 km. w. of Quillabamba, Tryon & Tryon 5365 (BM, F, GH, U, US, USM), Vargas 1739 (GH). lc, Pityrogramma calomelanos var. ochracea (Presl) Tryon. Contrib. Gray Herb. 189: 61.1962. Fic. 39, MAP 9. Gymnogramma ochracea Presl, Rel. Haenk. 1: 17. 1825. Type: Peru, Haenke, PR (Herb. no. 24360), photo GH, Us; probable isotypes: K fragment NY!, B fragment NY! Gymnogramma Ballivianii Rosenst. Fedde Repert. 6: 314. 1909. Type: Bolivia, Buchtien 1038, s-pA! fragment ex Rosenst. Us! Ceropteris adiantoides var. peruviana Hieron. Hedwigia 48: 221. 1909. Type: Peru, Stiibel 1096, B, not P. calomelanos var. peruviand (Desv.) Farw. Pityrogramma ochracea (Presl) Domin, Publ. Fac. Sci. Univ. Charles 88: 8, 1928, Pityrogramma Ballivianii (Rosenst.) Domin, op. cit.: 10. Pityrogramma perelegans Domin, op. cit.: 8. Type: Tarapoto, Peru, Spruce in 1855-56, K! photo GH. Lamina pubescent beneath, not ceraceous. The spores of some specimens of var. ochracea have well defined dark ridges typical of the genus. Other specimens have spores that are more spherical than the usual type although somewhat flattened in three planes on the commis- sural face, and the tan exospore is smooth or only slightly roughened. It is exceptional to find such different spores in the same species or variety. FERNS OF PERU 71 A hybrid of var. ochracea and P. trifoliata is treated under that species. Honduras ; Venezuela to Colombia, south to Bolivia. Open rocky places, river banks and road banks, cleared hillsides, cliffs, less often in thickets, forest or along forest borders, rarely along irrigation ditches, San Martin to Puno, 400-2600 m. or at 50 m. along irrigation ditches in Libertad. Selected specimens: SAN MARTIN: near Moyobamba, Woytkowski 35314 (UC); Juan Jui, alto Rio Huallaga, Klug 4177 (F, GH, UC); Divisoria, Aug. 3, 1942, Ridoutt (GH, USM). LORETO: Boquerén Padre Abad, Allard 22106, 22070 (GH, US). LIBERTAD: Huaca del Sol, Sagds- tegui 393 (GH); Hacienda de Mochal, Sagdstegui 425a (GH) ; between Poroto and Samne, Angulo & Lépez 464 (GH) ; Mampuesto, Lépez 1318 (GH). HUANUCO: entre Puente Durand y Chinchavito, Ferreyra 6787 (GH, USM) ; Puente Durand, Stork & Horton 9446 (F, UC); Villeabam- ba, Macbride 4994 (F, GH, US) ; Tingo Maria, Tryon & Tryon 5243 (BM, F, GH, U, US, USM) ; Hacienda Exito, Mexia 8166 (F, GH, US). PASCO: Oxapampa, Soukup 2360 (GH). JUNIN: La Merced, Macbride 5276 (F, US), rm 3411 (GH), Cerrate 2833 (GH, USM) ; Colonia Perené, Killip & Smith 25421 (F, us); San Ramén, Coronado 267 (GH); Carpapata, G. Kunkel 591 (GH). AYACUCHO: Estrella, between Huanta and Rio Apurimac, Killip & Smith 22667 (Us). cUzcO: Machu-Picchu, Herrera 3301 (GH, US); Pumachaca, valle de Santa Ana, Herrera 3288a (US) ; Potrero, 8 km. w. of Quillabamba, Tryon & Tryon 5366 (BM, F, GH, U, US, USM); Machu-Picchu to Quillabamba, Mexia 8089 (F, GH, US); Santa Rosa, Cook & Gilbert 1715 (US) ; Coshipata, Vargas 11293 (GH). PUNO: near Puno, Soukwp 449 2. Pityrogramma tartarea (Cav.) Maxon, Contrib. U.S. Nat. Herb. 173: 178...1918, Rhizome scales (and those at the base of the petiole) with a long portion, one cell wide, below the spherical terminal cell; leaf usually ca. 0.4-1 m. long, its axes glabrous or thinly ceraceous or thinly pubes- cha petiole and the rachis deep reddish brown to blackish; lamina long-triangular to deltoid, pinnate-pinnatifid to tripinnate, pinnae in- equilateral, those on the basiscopic side better developed, pinnules at nearly right angles to the pinna-rachis, the lower surface usually white, pale yellow or bright yellow ceraceous, or pubescent. The following key is supplementary to the headings in the key to species. It may be useful for the identification of certain material that is not typical of either P. calomelanos nor P. tartarea. Vz ROLLA TRYON Fics. 37-42. Fig. 37. Pterozonium reniforme: fertile —_ * tag Peru, ae se GH, Fig . 88. Pityrogramma calomelanos var calomelanos: pin x %, Page 6101, cH, Fig. 89. P. — var. oohrucea: A, pinna, x 1, hess. Mexia 8106, afi B, ion of petiole, Seg Peru, Tryon & Tryon 5366, cH. Fig. 40. P. pa? , central pinnae of small leaf, X 14, Peru, Tryon & Tryon 5429, GH? ‘“ H. ig. 41. Col chrysoconia: central pinnae, X 1%, C olombia, H. H. Smith 1061, ou. Fig. 42. P. fe ir ginea: A, central pinnae, X 14, Peru, Tryon & Tryon 5451, GH; B, section of petiole, enlarged, idem; C, spores, three necied enlarged, idem FERNS OF PERU 73 Lamina lanceolate to ovate-lanceolate, or in large leaves long-tri- angular, its apex often more or less acute or abruptly acuminate; apical pinnae strongly ascending; pinnules acute, serrate, or acutely pinnatifid; basal inferior pinnules more strongly ascending on the upper than on the lower pinnae P. calomelanos a long-triangular, to deltoid in large leaves, its apex evenly long-acuminate; apical pinnae nearly or quite at right angles to the rachis; pinnules obtusely lobed or pinnatifid; basal inferior bi dend at about the same angle to the pinna-rachis on all pinnae (or in large leaves the basal inferior ‘tertiary segments so in relation to the pinnule- rachis). P. tartarea 2a. Pityrogramma tartarea var. tartarea. Fic. 40, Map 10. Acrostichum tartareum Cav. Descr. 242. 1802. Type: near Guaman- tanga, Peru, Née, seen by C. Chr. at MA (Dansk Bot. Ark. 9(3): 10. 1937). G ramma peruviana Desv. Ges. Naturf. Freunde Berl. Mag. 5: 329. 1811. Type: Peru, Jos. de Jussieu (Herb. Jussieu no, 1009) P! photo GH Pityrogramma peruviana (Desv.) Maxon, Contrib. U. S. Nat. Herb. 17: 173. 1913. Pityrogramma calomelanos var. peruviana (Desv.) Farw. Am. Midl. Nat. 12: 280. 1931. Lamina ceraceous beneath, the wax white to pale (cream) yellow, rarely very sparse and almost colorless; rarely partly ceraceous- pubescent. Greater Antilles; Mexico to Panama; Venezuela, Colom- bia, Ecuador and Galapagos Islands south to Bolivia; Brazil. ocky and shrubby hillsides, in crevices of rocks and Inca walls, on shrubby slopes and on cliffs, Piura to Puno, 800- 4000 m Selected specimens: PrURA: arriba de Canchaque, Ferreyra 3106 rs USM) ; Prov. Piura, N. Angulo 2168 (GH). LAMBAYEQUE: 29 km. f Olmos, on road to Jaén, Correll & Smith P803 (GH). ‘atu Macash, prov. Celendin, Ridowtt 507 (GH). AMAZONAS: between Doni- las and Cohechan, Soukup 4132 (F); Pomacocha, prov. Chachapoyas, Lépez et al. 4394 (GH). LIBERTAD: Retamas, prov. Pataz, Lépez & Sagdstegui 3604 (GH). HUANUCO: Mito, Macbride & Featherstone 1429 (F, Us), Bryan 380 (F); Carpish, Coronado 60 (GH, UC), 87 (UC); Hacienda Paty, prov. Hudnuco, Ferreyra 9415 (GH, USM). JUNIN: Carpapata, prov. Tarma, Soukup 2476 (F, GH), Cerrate 2776 (GH, UsM) ; Huacapistana, Ferreyra 11245 (GH, USM), Coronado 273 (UC), Tryon & Tryon 5429 (BM, F, GH, U, UC, USM); Palea, Correll & Smith P763 (GH). cuzco: Machu-Picchu, Ferreyra 9913 (GH, USM), Coronado 108 (GH, UC), Mexia 8086 (F, GH, US); Torontoy, Herrera 1598 (US); ROLLA TRYON Map 12, P. Mars 9-12. Map 9, Pityrogramma calomelanos var. ochracea. Map 10, P. . FERNS OF PERU 75 Valle de Cositipata, Scolnik 883 (Us); Toccorochayoc, prov. Paucar- tambo, Woytkowski 95 (GH, USM). PUNO: near Puno, Soukup 450 (F). 2b. Pityrogramma tartarea var. aurata (Moore) Tryon, Contrib. Gray Herb. 189: 65. 1962. Gymnogramma tartarea var. aurata Moore, Gard. Chron. 1870: 493. Type: Cultivated by Veitch from a collection by Pearce in Peru. (The specimen of Pearce 218 at K! photo GH, and Domin, Rospr. II. Tr. Ceské Akad. 38(4): t. 1. 1929, may be the type as Domin says; it does clearly represent Moore’s name). Pityrogramma Presliana Domin, Publ. Fac. Sci. Univ. Charles 88:6. 1928. Type: Peru, Matthews 1823, K! photo GH. Lamina ceraceous beneath, the wax bright yellow. Colombia, Ecuador and Galapagos Islands to Argentina. Open places, clay banks and among rocks, southwest Lor- eto to Cuzco, 800-1800 m. Specimens seen: LORETO: Divisoria, Ferreyra 1668 (GH, USM). HUAN- uco: near Pozuzo, Macbride 4793 (F, US); Villeabamba, Rio Chinchao, Macbride 4995 (F, GH, US). JUNIN: Chanchamayo valley, Schunke 38 (F, US); near San Ramon, Schunke A177 (Us); between Huacapistana and San Ramon, Coronado 259 (GH, UC). cUzcCO: Tanamayo, prov. Paucartambo, Vargas 6479 (UC). 2c. Pityrogramma tartarea var. Jamesonii (Baker) Tryon, Contrib. Gray Herb. 189: 66. 1962. Lamina pubescent beneath, not ceraceous. Colombia and Ecuador; this variety may be expected to occur in Peru in the northern Andes adjacent to Ecuador. 3. Pityrogramma chrysoconia (Desv.) Domin, Publ. Fac. Sci. Univ. Charles 88:10. 1928. Fic. 41. Acrostichum chrysoconium Desv. Mém. Soe. Linn. Paris 6: 212. 1827. Type: Peru, evidently Dombey, P, photo GH; isotype: B! photo GH, frag- ment us! Gymnogramma Ornithopteris Kl. Linnaea 20: 413. 1847. Type: Venezuela, Moritz 288, B! photo GH, fragment US!; isotype: K! photo Pityrogramma Ornithopteris (K1.) Knuth, Fedde Rep. Beih. 43: 95. 1926. Rhizome scales (and those at the base of the petiole) with only a short portion, one cell wide, below the spherical terminal cell; leaf ca. 25-80 em. long, its axes glabrous or thinly ceraceous, petiole (at least 16 ROLLA TRYON the upper portion) and the rachis straw-colored to light brown; lamina long-triangular, bipinnate to bipinnate-pinnatifid, pinnae inequilateral, those on the basiscopic side somewhat better developed, pinnules nearly at right-angles to the pinna-rachis, the lower surface white or yellow ceraceous, Plants with white wax on the leaves and those with yellow wax both occur nearly throughout the range of the species and there seems to be no reason to recognize these variants. Costa Rica to Panama; Jamaica; British Guiana to Co- lombia, south to Bolivia. Rocky and shrubby slopes and hillsides and Inca walls, Huanuco to Cuzco, 2000-2800 m. Specimens seen: HUANUCO: Mito, Macbride & Featherstone 1393 (F, Us), Bryan 182 (F); Mitotambo, arriba de Mito, Ferreyra 10348 (GH, USM). HUANCAVELICA: Chuspi, Tocas, Tovar 2052 (GH, USM). CUZCO: Huadquifia, Biies 1319 (US); Valle de Occobamba, Biies 851 (US); Santa Rita, prov. Urubamba, Vargas 2683 (US); Valle de Amparaes, prov. Paucartambo, Biies 1836 (US). 4. Pityrogramma ferruginea (Kze.) Maxon, Contrib. U.S, Nat. Herb. 17: 173.1918. Fic. 42, Map 11 gramma, ferruginea Kze. Linnaea 9: 34. 1834. Type: Que brada de Chinchao, Peru, Jul. 1829, Poeppig, Diar. 1160; isotype: P! photo GH. : Eriosorus Ruizianus Fée, Gen. Fil. 152, t. 13, f. 2. 1852. Type: Peru, uiz. Rhizome scales (and those at the base of the petiole) with a rather long portion, one cell wide, below the long-conical terminal cell; leaf ca. 0.25-1 m. long, its axes densely and persistently lanate (the petiole Sometimes deciduously so), petiole and rachis atropurpureous a blackish ; lamina narrowly elliptic, pinnate-pinnatifid to pinnate-PI~ natisect, pinnae equilateral, the pinnule-lobes usually somew ascending, the lower surface densely, to rarely thinly, lanate. The dense lanate covering of the whole leaf makes this a most distinctive species. It is the only species in the genus with a markedly disjunct range. A hybrid of P. ferrugin@ and P. trifoliata is treated under the latter species. Guatemala to Panama; Peru. , Moist and dry cliffs, rocky banks, crevices of rocks, Hust: uco to Ayacucho, 750-1800 m. FERNS OF PERU 77 Specimens seen: HUANUCO: Rio Huallaga cafion, below Santo Domin- go, Macbride 4261 (F, US); 25 km. toward Lima from Tingo Maria, Allard 21530 (GH, US), 21531 (US). PASCO: Oxapampa, Soukup 1822 (F, US), 3351 (F, GH). JUNIN: east of La Merced, Hutchison 1205 (Gu, San Ramén, Tryon & Tryon 5451 (BM, F, GH, U, UC, US, USM) ; vicinity of San Ramon, Constance & Tovar 2233 (UC). AYACUCHO: Cearrapa, between Huanta and Rio Apurimac, Killip & Smith 22482 (us). 5. Pityrogramma trifoliata (L.) Tryon, Contrib. Gray Herb. 189:68. 1962. Fic. 43, MaP 12. Acrostichum trifoliatum L. Sp. Pl. 2: 1070, 1753. LINN a 1245.9, photo A, and Sloane, Hist. Jam. t. 45, f. 2, are both this sp rismeria aurea Fée Mém, ae Foug. 5 (Gen. Fil): "168. 1852. (nom. superfl. illegit. = Acrostichum trifoliatum L.) Trismeria m pee at Fée, Mém. Fam. Foug. 5 (Gen. Fil.) : 165. 1852, ex char. “Peruvi Trismeria trifoliata “(L) Diels, Nat. Pflanz. 1(4) :265. 1899. Rhizome scales (and those at the base of the petiole) with a moder- ately long portion, one cell wide, below the long-conical to enlarged terminal cell; leaf ca. 0.5-1.25 m. long, its axes glabrous to thinly Ceraceous, petiole and rachis light reddish brown to blackish; lamina Narrowly lanceolate to usually elongate, 1-pinnate to bipinnate, pinnae equilateral or inequilateral, pinnules (when present) ascending, the lower surface of the pinnae glabrous to usually white or yellow cerace- ous Pityrogramma trifoliata usually has somewhat dimorphic leaves, the fertile ones being taller and more erect than the Sterile. In this species and its hybrids the two vascular bundles in the petiole (about 1/3 the distance to the lamina) are C-shaped with the back of the C curved forward. In the Other species the vascular bundles are roundish, oval or C-shaped. The pinnae of P. trifoliata are usually simple, bifoliolate or trifoliolate, rarely to 7-foliolate. The hybrids with P. calomelanos and P. ferruginea differ in having the Pinnae with many segments. Southern Florida; Greater Antilles; Mexico to Costa Rica (notably absent from Panama); Venezuela to Colombia South to Bolivia, Argentina and Chile (Arica) ; Paraguay, Uruguay and Brazil. Open rocky ground, or in gravel, along road borders, river banks, irrigation ditches, etc., Piura to Loreto, south to Cuzco and Arequipa, 150-2200 m. 78 ROLLA TRYON 43-47. Fig. 43. Pityrogramma trifoliata: A, B, sterile pinnae, X % Fics, -— Rica, Shut ch 2295, GH; C, pangs pinna, X %, idem; D, petiole section, oe idem. Fig. 44, P. ca nos var. ochracea X trifol iata: sterile pinna, X % Fel® Prem & Tryon 5440, cH. — is, ‘Pp. ferruginea X trifoliata: A, sterile pinna, > Peru, Tryon & Tryon 5449, G ; B, petiole section, gia rged, idem. Fig. 46. 7 ma leptophylla: A, sterile pinne, OL » Mexia 8083, GH; B, sterile pinna, * : Peru Peru, Tryon & Tryon 5416, ; ©, sterile stidaisiies segments, X 214, idem am pate ultimate segments, X 214, Pace , Mexia si GH. Fig. 47. A. chaeropholla Pinnule, 214, Brazil, L. B. Smith 1202, FERNS OF PERU 79 Selected specimens: PIURA: Serran to Canchaque, Ferreyra 10782 (USM). LAMBAYEQUE: 20 km. from Olmos, on road to Jaén, Correll & Smith P796 (GH). CAJAMARCA: South of Ichocan, on road to Cajabam- ba, Correll & Smith P913 (GH); alrededores de San Benito, prov. Con- tumaza, Sagdstegui 3740 (GH). SAN MARTIN: Tarapoto, Spruce 4153 (BM, GH, US). LORETO: lower Rio Huallaga, L. Williams 4640 (F); Boquerén de Abad, Ferreyra 1109 (USM). LIBERTAD: Trujillo, Osgood & Anderson 28 ha 29, 30 (F, US), Worth et al. 8887 (GH, UC), HUAN- uco: Huanuco to Muna, Mexia 4100 (GH, MO, UC, US); Huanuco, Mac- bride & ene targa 2066 (F, US); Muna, Macbride 4051 (F, Us); Tingo Maria, Soukwp 2191 (F), Allard 22051 (US). LIMA: Santa Eulalia, Coronado 1 (GH, MO, UC, US); Huaral, Saunders 154 (BM), Chosica, Macbride 2856 (F, GH, MO, US), Tryon & Tryon 5342 (BM, F, GH, MO, NY, U, UC, US, USM) ; Sayan, Ferreyra 3505 (BM, USM). JUNIN: La Merced, Killip & Smith 23496 (F, US), Soukup 1095, 1107 (F); Chanchamayo valley, Schunke 1377 (F). AYACUCHO: Ayna, Killip & Smith 23112 (US). APURIMAC: 45 km. from Abancay, on road to Chall- huanca, Saunders 764 (GH) ; Kairanka, prov. Grau, Vargas 5868 (UC). cuzco: Sisal to Cunyace, prov. Anta, Vargas 7411 (MO, UC); San Miguel, Cook & Gilbert 1087 (US). AREQUIPA: Huario, below Chuqui- bamba, D. Stafford 1149 (BM); Majes valley, north of Arequipa, D. Stafford 1191 (BM). 5a. Pityrogramma calomelanos var. ochracea X trifoliata. Fic. 44. Gymnogramma Herzogii Rosenst. mone Rijks Herb. 19: 21. 1913. Type: Bolivia, Herzog 2000; isotype: US Petiole and rachis thinly pubescent to glabrous, segments pubescent beneath. Peru and Bolivia. Dry, rocky stream bed, Junin, 750 m. and Cuzco, 1200 m. Specimens seen: JUNIN: La Merced, Tryon & Tryon 5440 (GH). cuzco: Puente Chaupemayo, sobre el Rio Sambray, Biies 1936 (US). 5b. Pityrogramma ferruginea X trifoliata. Fic. 45. Petiole and rachis closely and densely lanate, although deciduously The only collection has the pinnules markedly irregular in their length. Peru. Exposed crevices of rocks, Junin, 1000 m. Ss ns seen: JUNIN: 10 km. sw. of San Ramén, Tryon & Tryon 5449 (BM, GH, U, US). 80 ROLLA TRYON 6. a en Pearcei (Moore) Domin, Publ. Fac. Sci. Univ. Charles 88:9. 192 Gymnogramma ile Moore, Gard. Chron. 1864: 340. Type: “Cult. Veitch ex Pearce”; two specimens, Pearce 274 and “Gymnogramma Pearcei, n, sp.” K! photos GH, are authentic. Lamina long-triangular, quadripinnate with slender ultimate lobes, pinnae equilateral. The original material of this species was probably collec- ted in Peru. It has not been gathered again and its status is uncertain. It may represent a valid species or it may be a highly dissected leaf-variant of another species. Specimens seen: “Chile,” Pearce (K); “Eastern Andes,” Pearce 274 (K); Hort. Veitch. in 1889 (K). Fic. 48, Anogramma leptophylla, Loma Lachay (Lima). FERNS OF PERU 81 10. ANOGRAMMA Link, Fil. Sp. Cult. 137. 1841. Type: eee leptophyllum L. = Anogramma leptophylla (L.) Link. FIG. Terrestrial, the rhizome erect, small to minute, with a few thin scales (or also similar trichomes), bearing the leaves in a cluster; leaves very small to medium sized, bipinnate to quadripinnate, glabrous, veins free; sporangia borne along the veins back of the unmodified margin, often confluent over the segment at maturity, indusium and paraphyses absent. — 5 species in America. , K. Generis Pityrogramma (Link) species ac sectiones in avai cualgticaie — (4. Anogramma), Publ. Fac. Sci. Univ. Charles no. 88: 9. 1928. Tryon, R. neg rasinin, in Tax. Fern Notes, II, Contrib. Gray Herb. 189: 74-76, 1962. The sporophyte is annual; it grows rather rapidly and lives for only a single season. The gametophyte is said to be perennial, producing buds which survive an adverse season and produce new sporophytes the following year. The lon- gevity of both the sporophyte and gametophyte of the differ- ent species needs to be confirmed in nature. Anogramma represents a reduced and specialized evolutionary develop- ment closely related to Pityrogramma. There is a single species in Peru but another, Anogramma chaerophylla, grows in Bolivia and may eventually be found in Peru. KEY TO SPECIES Ultimate lobes obtuse, 2-veined, bifid lobes frequent, pinnules obovate-cuneate. A. leptophylla Wittens lobes acute, poy. 1-veined, pinnules ovate to ovate- lanceolate. . chaerophylla (Desv.) Link, Fig. 47. Anogramma leptophylla (L.) Link, Fil. Sp. 137. 1841. Fies. 46, 48. Polypodium leptophyllum L. Sp. Pl. 2: 1092. 1753. Syntypes: Magn. Monsp. 5, t. 5; Barr. Rar. 1270, t. 431; LINN 1251.56, photo A. Leaves ca. 5-15 em. long, the petiole long, light brown to straw- colored near the apex, to dark reddish-brown at the base; lamina herbaceous, lanceolate, ovate or deltoid, pinnules aan the ultimate lobes obtuse, often bifid and with two vein This species is known only from some of the lomas of the Department of Lima and from Machu-Picchu. At Loma de 82 ROLLA TRYON Lachay it grows under the edges of large rocks and is well hidden from the casual collector. Mexico and Central America; Ecuador and Peru; Old orld. Among rocks, under boulders, on Inca walls, Lima and Cuzco, 500-2400 m. Specimens seen: LIMA: Loma de Lachay, Coronado 14 (GH, UC, US), Tryon & Tryon 5416 (BM, F, GH, U, US, USM); Loma de Quilmana, Coronado 26 (GH, UC, US); Loma de Granados, 12 km. n. of Huaral, Stork & Vargas 9332 (GH, UC). cUzcO: Machu-Picchu, Mexia 8083 (F, GH, UC), Coronado 94 (GH, UC, US); Machu-Picchu Station, Vargas 2591 (US). 11. GyYMNopTERIS Bernh. Jour. Bot. Schrad. 1(2) :297. 1799. Type: Aecrostichum rufum L. (not Pteris ruffa L. as often stated) = Gymnop- teris rufa (L.) Underw. Terrestrial, the rhizome rather small, erect or nearly so, scaly, bear- ing the leaves in a cluster; leaves small or medium sized, 1-pinnate or bipinnate, pubescent, veins free; sporangia borne along the veins for most of their length, slightly back of the unmodified margin, indusium and paraphyses absent.— 4 American species. A few specimens seem somewhat intermediate between the two well marked extremes of the Peruvian species. A critical study of all of the American material of these species is needed in order to understand their relation to each other. KEY TO SPECIES All pinnae short-stalked, the dark color of the stalk ending abruptly at the usually cuneate (to subcordate) base, basal pinnae slightly smaller than those above, pinnae entire. | L, Gee Lower pinnae with (often long) stalks, longer than the upper oes» the dark color of the stalk entering the usually cordate (to broadly cuneate) base, the basal pinnae larger than those above, pinnae entire, lobed or 1-pinnate. ... 2. G. tomentosa eo Cravens rufa (L.) Underw. Bull. Torrey Cl. 29: 627. 1902. Acrostichum rufum L. Syst. Nat. ed. 10, 1320, 1759. Type: Sloane, t. 45, £. 1. (Not Pteris ruffa L. Sp. Pl. 2: 1074. 1753 which is mal tained in Syst. Nat. 1321). Bernhardi did not transfer this species to Gymnopteris, as often cited; Underwood did, although he may not have been the first one to do go. Leaves ca. 20-60 em. long, petiole usually long, it and the rachis dark reddish-brown to atropurpureous, pubescent with short, acicular = FERNS OF PERU &3 xX 14, Venezuela, Fendler sa: A, sterile lamina, X 14, Peru, Tryon & Tryon ake GH; B, fertile ‘alas x %, pk Vargas 1733, GH. Fics. 49-50. Fig. 49. Gymnopteris rufa: A, a" pier 302, GH; B, fertile lamina, X 14, idem. Fig. G. tomen 84 ROLLA TRYON chomes and fewer, long, multicellular ones; lamina narrowly elliptic to lanceolate, 1-pinnate, the pinnae entire, usually cuneate (to subcor- date), short-stalked, the dark color of the stalk not continued into the pinna, both surfaces more or less pubescent with acicular, multicellu- lar, more or less appressed, trichomes. Southern Mexico to Panama; Greater: Antilles; Surinam to Colombia, south to Peru. In forest, San Martin, 400-800 m. Specimens seen: SAN MARTIN: Juan J ui, Klug 4174 (F, GH, UC). 2. Gymnopteris tomentosa (Lam.) Underw. Bull. Torrey Cl. 29: 627. 1902. Fig. 50. Asplenium tomentosum Lam. Encycl. 2: 308. 1786. Syntypes: Brazil, Commerson, Dombey. A sheet in Herb. Lamarck, P! photo GH, “As- plenium tomentosum e Brasilia” (without collector) is this species. Peru to Argentina, Paraguay and Brazil. Forest slopes, stony clearings and in woods, San Martin to Cuzco, 700-1600 m. Selected specimens: SAN MARTIN: Tarapoto, Spruce 3992 (B, GH); Woytkowski 35228 (Uc), L. Williams 5484 (F). HUANUCO: near Pozu- 20, Macbride 4788 (¥, US). JUNIN: La Merced, Macbride 5303 (F); Rio Paucartambo, near Perené Bridge, Killip & Smith 25345 (US); Rio Pinedo, n. of La Merced, Killip & Smith 23592 (GH, US). CUZCO: Quel- launo, prov. Convencion, Vargas 13557 (GH); Quillabamba, Coronado 122 (GH, UC, US); Santa Ana, Cook & Gilbert 1537 (GH, US), 1690 (US) ; Santa Rosa, Urubamba valley, Cook & Gilbert 1724 (US) ; Toron- Y, Herrera 1331 (us); Potrero, 8 km. w. of Quillabamba, Tryon & Tryon 5374 (GH, U, US, USM). 12. HEMIONITIS L. Sp. Pl. 2: 1077, 1753; Gen. Pl. 485. 1754. Type: Hemionitis palmata L. Terrestrial, the rhizome small, erect or nearly so, scaly, bearing the leaves in a cluster; leaves small, pedately lobed, pubescent, veils anastomosing; sporangia borne all along the veins, nearly to the WU modified margin, indusium and paraphyses absent.—5 American species, FERNS OF PERU 85 Hemionitis palmata L. Sp. Pl. 2: 1077. 1753. Type: LINN 1248.3, photo A, is this species. Fic. 51 Sterile leaves ca, 3-10 cm. long, 3-lobed or pedately 5-lobed, clustered at the base of the few, erect, pedately 5-lobed fertile leaves which are ca. 10-25 cm. long; principal lobes shallowly lobed or crenate on the tawny to rufous pubescent with scolltaatiain, slender-tipped trichomes, these especially evident along the margins, the brown to dark brown petiole with similar trichomes, Fics. 51-52. Fig. 51. Hemionitis palmata: A, fertile leaf, X %4, Guatemala, Deam ie GH; B, portion of sterile segment, X 1, nag - C, sterile leaf with young plants, X 14, Mexico, Dressler & Jones 76, GH. Fig. 52. rachypteris pin “op = aor X 44, Argentina, Schreiter 8876, GH; B, sterile nt x 4, idem; C, reir surface o sterile leaf, X 2 86 ROLLA TRYON This is a most attractive species, the sporangia exactly tracing the pattern of the veins. It has been cultivated at Iquitos: Williams 3559 (F). West Indies; Mexico and Central America; Guianas to Colombia, south to Peru and Bolivia; Brazil. In forests, San Martin and Cuzco, 500-900 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce 3993 (GH, US) ; Juan Jui, alto Rio Huallaga, Klug 4288 (F, GH, MO, UC, US, USM); near Tarapoto, L. Williams 6793 (F, US). CUZCO: Santa Ana, Cook & Gilbert L477 CUS), 13. TRACHYPTERIS Christ, Denkechr. Schwe'z. Naturfors. Gesells. 36: (Monogr. Elaph.) 150. 1899. Type: Acrostichum aureonitens Hook.= Trachypteris aureonitens (Hook.) Christ = Trachypteris pinnata (Hook. f.) C. Chr. Terrestrial, the rhizome rather small, erect or decumbent, scaly, bearing the sterile leaves in a rosette, the few fertile ones in a cluster; leaves small, the sterile with the lamina entire, the fertile with the lamina deeply pinnatifid to 1-pinnate, densely scaly beneath, glabrate above, veins anastomosing; sporangia borne all along the veins nearly to the unmodified margin, indusium and paraphyses absent. — 1 Amer- 1can species. Trachypteris pinnata (Hook. f.) C. Chr. Ind. Fil. 634. 1906. Fie. 52. Hemionitis pinnata Hook. f. Trans. Linn, Soc. 20: 167. 1847. Type: Charles Isl., Galapagos Isls., Darwin. Acrostichum aureonitens Hook. Ic. Pl. t. 933 (Cent. Ferns t. 33). 1854. Type: Galapagos Isls., Cuming 109 (the plate was drawn from 3 itt aon og: Gillianum Bak. Jour. Bot. 1882: 310. Type: Arassnahy, nL Gerais, Brazil, Gille (Glaziou no. 13341) K! photo GH; isotype: ie Several, ca. 5-8 cm. long, more or less prostrate, spathu- the bynes se rhs sessile or nearly so, rarely with a few lobes, scales on fertile fives Pe brownish, appressed, imbricate, more or less dentate} with a lon, titi ca. 12-25 em, long, erect, subdeltoid to short-oblong, pater 7 rown to dark reddish-brown, deciduously scaly petiole, n the lower surface similar to those of the sterile leaf. faa ae study of the American material may show that ree species, or geo . ae ies, are present. geographic varieties of one spec lain Islands; Peru, Bolivia, Argentina and Brazil. n rocky woods and in forest, Cajamarca, San Martin and Cuzco, 850-2750 m. , FERNS OF PERU 87 Specimens seen: CAJAMARCA: Jaén, Rauh P2160 (B); Rio Chamaya, Olmos to Rio Marajion, prov. Jaén Hutchison 1424 (F, GH, UC). SAN MARTIN: Tambo de Carrizal, Stiibel 1016 (B). CUzCO: Santa Ana, Cook & Gilbert 1480 (US); Huadquifia, Biies 1339 (US); Sahuayacu, Biies 837 (US). 14. CERATOPTERIS Brongn. Bull. Soc. Philomath. Paris 1821: 186. Type: Acrostichum thalictroides L. = Ceratopteris thalictroides (L.) Brongn, Aquatic (floating on water or rooting in mud), the stem small, erect, with a few scales, bearing the sterile leaves in a rosette, the fertile ones in a cluster; leaves small to medium sized, the sterile lobed to bipinnate- lobed, the fertile 1-pinnate to quadripinnate, glabrous, veins anastomos- ing; sporangia borne sparingly along the marginal commissure, some- times also on the lateral veins, more or less covered by an indusium formed from the reflexed, somewhat modified margin, paraphyses absent. — 4 species in America. Benedict, R. C. The genus Ceratopteris: a preliminary revision, Bull. Torrey Cl. 36: 463-476. 1909. Benedict, R. C. Ceratopteridaceae, N. Amer. Fl. 16(1) :29-30. 1909. DeVol, C. E. The geographic distribution of Ceratopteris pteridoides, Am. Fern Jour. 67-72. 1957. (The photograph and the map are errone- ously labeled C. thalictroides). Ceratopteris pteridoides (Hook.) Hieron. Bot. Jahrb. 34: 561. 1905. Fig. 53. Parkeria pteridoides Hook. Exot. Fl. 2: t. 147. 1825. Type: District of Essequibo, British Guiana, Parker; isotype: GH! Ceratopteris pteridoides: A, sterile leaf, X 44, Florida, Curtiss, GH; B, Fic. 58. fertile pinna, X 14, Cuba, Killip 44595, GH; C, portion of fertile segment, enlarged, idem. 88 ROLLA TRYON Sterile leaves more or less prostrate, ca. 3-20 cm. long, with a fleshy petiole which is broadest at its apex and narrowed toward the base, or is inflated, lamina deltoid, thin-herbaceous, with broad segments, old leaves often with adventitious buds; fertile leaves more or less erect, ca, 5-35 cm. long, with linear segments; leaves intermediate between the typical sterile and fertile ones are sometimes present. Southern Florida; Central America; northern South America to Peru, Argentina and Brazil; Old World. In slow streams and along rivers, Loreto, 100 m. Specimens seen: LORETO: Rio Itaya, Iquitos, Asplund 14626 (US); Iquitos, Killip & Smith 27427 (us). 15. CHEILANTHES Sw. Syn. Fil. 5, 126. 1806, conserved name. Type: Cheilanthes micropteris Sw. Adiantopsis Fée, Gen. Fil. 145. 1852. Type: Adiantum pauperculum Kze. = Adiantopsis paupercula (Kze.) Fée = Cheilanthes paupercula (Kze.) Mett. Terrestrial, the rhizome small to moderately stout, compact to creep- ing, scaly, bearing the leaves in a cluster or closely spaced; leaves usually small to medium sized, pinnate (pinnate-pinnatifid to quadri- pinnate), ternate or radiate, glabrous, glandular, ceraceous, pubescent or scaly, veins free; sorus borne on the vein-tips, not paraphysate, marginal, covered by the indusium which is formed from the reflexed, modified margin or (in species 1-3) ‘the strongly revolute margin scarcely modified. — About 35 species in South America. A genus of 15 species in Peru, these inhabiting sheltered rocky places in the Altiplano and adjacent valleys of the Andes, or some of them growing in exposed situations. Cheilanthes incarum and C. scariosa, for example, are extreme alpine xerophytes. KEY TO SPECIES a. Lamina pubescent or scaly; petiole and rachis terete or rarely the latter grooved on the upper side toward the apex (or in no. 6 moet idl Sige, Fn Se SRO SERCO eC A MPT a b. Lamina ee a a eo ee c. Seales of the rachis predominantly narrowly linear to acicular; ultimate segments suborbicular, minute; rhizome scales with * dark sclerotic central portion and pale thinner MargiNs. «-+-+-+-+*""" 1. C. myriophylla ce. Scales of the rachis ovate-acuminate; rhizome scales browns” oi aaa d. Pee eee ew et esesoes FERNS OF PERU 89 d. Ultimate segments small, suborbicular, their upper surface partially concealed by a persistent pubescence produced from the dissected tips of the scales beneath. ...........0. 2. C. scariosa d. Ultimate segments mostly of moderate size, ovate-deltoid, their upper oe hardly concealed by the deciduous tips of the scales eath. 3. C. incarum b. Lamina pubescent. e. e. Rachis and pinna-rachises pubescent, often becoming glabrous; indusium deeply crenate or lobed; rhizome scales b to atropurpureous, rigid, shining, with or without very narrow, pale, thinner margin; blade narrowly linear, pinnae mostly alternate, adjacent, very NUMETOUS. .....-+ss00-0-s0000000 . C. pruinata e. Rachis nal pinna-rachises pubescent, often becoming glabrous on the under side, the trichomes moderately long (rarely only those of the pinnae so ¢ f. Scales at the apex of the rhizome dark brown, a shining, with or without very narrow, pale, thinner margins. ...............« 5. C. pilosa » i Sealed at the apex of the rhizome whitish to light Naat, te dull, concolorous. g. Petioles deciduous, breaking sharply and evenly ara their base; rachis grooved on the upper side or the apical Pp te. 6. C. fractifera g. Petioles neces nerd breaking ethers rachis terete or nearly so h. h. Pinna-rachis fea pean the pinna-stalk greenish to blackish; older rhizome scales often with a rigid, dark, sclerotic benkrad MOPRMOM. wicieciselciinisons 7. C. notholaenoides h. Pinna-rachis atropurpureous to blackish above (as is the pinna-stalk) well beyond the basal gia or nearly to the tip; older rhizome scales sometimes with a Saas semi-sclerotic base. . C. Moritzia a. Lamina glabrous or glandular, or ceraceous-glandular; rachis a” _— the petiole grooved on the upper side. i. i, Segments whitish ceraceous-glandular beneath. steee 11. C. farinosa i. Bb tere glabrous or with scattered non-ceraceous glands eath. 10. C. vaibale k. i Lamina ternate to usually radiate. j. Lamina pinnate. k. Indusium confined to the segmen l, Rhizome scales eee ap a \ father thick, dark center and lighter, thin m m. Apical part of the sabia on the upper side, with a li sek colored, submembranous ridge on each Side. ..........20+-se0++00 . chlorophylla m. Apical part of the petiole, on the upper side, with a rounded ridge on each side, these concolorous with the petiole (or rarely the petiole terete). Ss 12. C. Poeppigiana 90 ROLLA TRYON 1. Rhizome scales concolorous, wholly brown to atropurpure- ous, rather thick. n. n. Apical part of petiole, on the upper side, flat to convex between the small lateral ridges; rhizome erect, ..........:00 13. C. Orbignyana n. Apical part of petiole, on the upper side, deeply sulcate between the relatively large lateral ridges; rhizome short- creeping, prostrate. 14. C. rufopunctata k. Indusium extending onto the segment stalks. 15. C. marginata 1. Cheilanthes myriophylla Desv. Ges. Naturf. Freunde Berl. Mag. 5: 328. 1811. Type: Peru, P! photoGH. Fic. 54, MAP 13. Cheilanthes elegans Desv. Ges. Naturf. Freunde Berl. Mag. 5: 328. 1811. Type: “Chile,” P! photo GH, a portion of a pinna, probably from a specimen collected in Peru. Dombey, Peru, P! photo GH, det. Desv., may supplement the holotype. Rhizome moderately stout, short-creeping, multicipital, scales lance- subulate, with a dar sclerotic central portion and paler margins; densely scaly beneath, nearly glabrous above, the ultimate segments small, suborbicular; indusium more or less modified, rather continuous. This species is characterized by its small bead-like ulti- mate Segments that are easily seen on the upper side of the lamina and usually not completely concealed by the scales on the under side. The upper surface of the segments is often slightly pubescent; the narrow scales of the petiole and rachis superficially resemble a mat of hairs. The related ©. lendigera (Cav.) Sw. differs in having long brownish hairs on the under surface of the segments, a definitely intra- Marginal indusium and an elongate, slender rhizome wi tawny, concolorous scales. C. A. Weatherby (Contrib. Gray Herb. 114: 22. 1936) has discussed the type specimens of Desvaux’s two names. Hispaniola; Mexico to Colombia and Venezuela, south to Bolivia and Argentina; Brazil. Rocky soil, shrubby hillsides and cliffs, Piura to Puno and Arequipa, 1550-3250 m. S 8); Cerro Chologday, prov. Otuzco, Sagdstegui 77 (GH). ANCASH* uasta, Cerrate 2164 (USM) ; Chiquidn, Cerrate 546 (USM). HUANUCO: FERNS OF PERU 91 Ambo, Macbride 3197 (F, US); Acomayo, Tryon & Tryon 5828 (BM, F, MO, US, USM). LIMA: Chancay, e. of Sayan, Goodspeed 33082 (GH, MO, UC, US); Surco, Ferreyra 3479 (GH, USM). JUNIN: Palca to Car- papata, Stork 10958 (F, GH, MO, UC) ; Nar caper Tryon & Tryon 5420 (BM, F, GH, MO, U, US, USM). HUANCAVELICA: of Pampas, Stork & Hovtii 10242 (F, UC). AYACUCHO: Ayacucho, West 8641 (MO, UC); 45 km. from Nasca on road to Puquio, Correll & Smith Pi70 (GH). APURI- MAC: Chirhuai, Vargas 2301 (cuz); Challhuanca, Saunders 772 (GH). CUZCO: Cuzco, Vargas 368 (GH, MO); Anta, Tryon & Tryon 5863 (BM, F, MO, U, US, USM); PUNO: Sandia, Weberbauer 724 (B). AREQUIPA: 14 equipa, Eyerdam & Beetle 22164 (GH, MO, UC); Arequipa, Pennell 13184 (F, GH, US). 2. Cheilanthes scariosa (Sw.) Presl, Rel. Haenk. 1: 65, 1825. Fie. 55, MAP 14, Acrostichum scariosum Sw. Syn. Fil. 16. 1806, based on Acrostichum lanuginosum Willd. Schrift. Acad. Erfurt, 1802: 31, t. 3, f. 4, not Desf. ge lip e: Peru, Malaspina Exped. (Herb. Willd. 19554-1), B! photo GH, Cheliidittes ornatissima Maxon, Smiths. Misc. Coll. 65(8) = Earhia Type: gage 8 back of Lima, Peru, W. E. Safford 996 Us!; iso- types: GH! mo! Cheilanthes Weis Copel. Univ. Cal. Publ. Bot. 19: 301, t. 58. 1941. Type: Chincheros to Andahuaylas, Dept. Apurimac, Peru, W 3724, uc! Rhizome rather stout, very short-creeping, rie agegy scales long and filiform, concolorous, light reddish brown; leaves petiole densely scaly, terete, as is the rachis; lamina oe to rather narrowly elliptic, tripinnate, pinnae densely scaly beneath, the scales and especially their tips curving over the glabrous upper surface and concealing it, ultimate segments small, subdeltoid; the margin strongly recurved, only very slightly if at all modified into an indusium. The ultimate segments are small and bead-like as in C. myriophylla but they are nct easily seen for the under surface is completely concealed, except in age, by the scales, and the upper surface is somewhat concealed by the dissected tips of those scales that extend over the margin. Some of the scales on the rachis have such strongly curved auricles that they give the appearance of a small scale attached to the base. This is one of the most scaly of all ferns and it is perhaps unfortunate that Maxon’s appro- priate name can not be taken up. C. A. Weatherby (Contrib. Gray Herb. 124: 19. 1939.) has discussed and illustrated the type specimen, and the confusion in the application of the 92 ROLLA TRYON C. Mars 13-16. nay 13, Cheilanthes myriophylla. Map 14, C. scariosa. Map 16, pruinata. Map 16, C. pilosa FERNS OF PERU 93 name C. scariosa that caused Maxon to redescribe the species. Peru and Bolivia. Exposed calcareous rocky places and cliffs, Libertad to Puno, 2700-4300 m. Selected specimens: LIBERTAD: Yanazara to Huaquil, Lépez & ee tegui 3395 (GH). ANCASH: Mahuay, Cerrate 2200 (USM). : Matucana, Macbride & Featherstone 424 (F, US), Goodspeed eae io UC) ; Rio Blanco, Killip & Smith 21681 (Us). SUNIN: Tarma, Macbride & Featherstone 1049 (F, US); near Tarma, Tryon & Tryon 5457 (BM, 8. Cheilanthes incarum Maxon, Smiths. Misc. Coll. wed ng 75. 1915. Type: Cuzco, Peru, Mr. & Mrs. J. N. Rose 19061, us! Fi. 5 Rhizome rather stout, short-creeping, multicipital, scales light brown, subulate, attenuate to a capillary tip; leaves ca. 10-25 cm, tall, the petiole densely scaly, terete, as is the rachis; lamina bipinnate to tripinnate, linear-lanceolate, pinnae densely covered with scales be- neath, with a few readily deciduous ones above, the ultimate segments mostly ovate-deltoid, their upper surface only partially concealed by the tips of the scales from beneath; indusium rather narrow, more or less continuous. In general appearance this species is like the previous, C. scariosa, and may easily be confused with it. In addition to the characters mentioned in the key, the leaves of C. incarum have a definite petiole of 5 cm. or more while those of C. scariosa are nearly sessile or have a very short petiole. The lowest pinnae are reduced but not as strongly so as in C. scariosa and the rhizome scales are broader, elongate-attenu- ate and dentate-spinescent. The leaf-cutting is similar : C. Moritziana and it may be that this species is a xerl alpine derivative from such a progenitor and not panes related, by its densely scaly leaves, to C. scariosa. Peru and nw. Argentina. Exposed rocky aa Huancavelica to Cuzco, 2600-3600 m. Specimens seen: HUANCAVELICA: Mejorada, Tovar 1014 (GH, USM), 94 ROLLA TRYON Hutchison 1960 (GH). APURIMAC: 5 km. n. of Huancarama, West 3903 (uc); Posocoi, prov. Andahuaylas, Vargas 8792 (UC); 80 km. from Abancay on road to Challhuanca, Saunders 768 (GH). CUZCO: Cerro Sape, Cuczo, Ferreyra 2668 (BM, GH, USM); Saxihuaman, Herrera 212, 2194 (F), Tryon & Tryon 5343 (BM, F, GH, MO, U, US, USM) ; Urubamba, Vargas 7627 (CUZ); near Cuzco, Herrera 1 (BM, F, UC, US), Mr. & Mrs. J. N. Rose 19061 (US); near Pisac, Hunnewell 15860 (GH). 4, arn pruinata Kaulf. Enum. Fil. 210. 1824. Type: “Peru.” Fic. 57, Map pc ce Pe peli Presl, Tent. Pterid. 160. 1836, nomen nudum, is perhaps this species; Dombey, Peru and Meyen, Peru, both at B! are labeled with Presl’s name. Cheilanthes Mathewsii Kze. Farnkr. 1: 50, t. 25. 1840. Type: Peru, Mathews 605, BM! photo GH. This specimen, identified by Kunze, may be the holotype, or it may serve to replace the one destroyed at LZ Cheilanthes fasciculata Goldm. Nov. Act. Acad. Caes. Leopold.-Carol. Nat. Cur. 19, suppl. 1 (Meyen, eit zur Bot.) : 456. 1848. Type: Peru, 9000-12000 ft., Meyen, B! photo G Rhizome moderately stout, cr rt multicipital, scales dark reddish brown, concolorous or with v very narrow pale borders; leaves ca. 20-50 em. tall, petiole terete, short- bles: Son usually densely so, rachis terete or rarely somewhat grooved, otherwise like the petiole; lamina linear, bipinnate-pinnatifid to tripinnate-pinnatifid, pinnae deltoid, deciduously pubescent above and beneath except along the axes and midveins be- neath; indusium moderately broad, seeply lobed, crispate, or as separ” ate lobes on small ultimate segment; The dense but very short pubescence of the petiole and rachis are especially distinctive characters. The pinnae are pinnate to bipinnate-pinnatifid; they are said to be glutin- ous, in drying they appear dull or shiny varnished. The related C. micropteris Sw. has a very narrow linear lamina, small pinnatifid pinnae about 3-6 mm. long, moderately long hairs on the rachis and narrow, tawny, concolorous rhizome scales. Peru to Argentina. : In crevices or on ledges of cliffs or in rocky soil, Cajam- arca to Puno and Arequipa, 2800-4400 m. Selected specimens: CAJAMARCA: Cajamarca to Cajabamba, Ferrey™@ USM). 3224 (BM, GH, USM) ; Cajamarca to Chilete, Ferreyra 3324 (oe Man- caine g above Cachicadan, Stork & Horton 9975 (F, UC); La BM) nas, Lépez 1119 (uC, US). ANCASH: Chiquidn, Ferreyra sie est as (USM), Cerrate 500 (GH, USM), 3350 (USM). HUANUC as Maebride & Featherstone 1490 (F, GH, US); Panao, Ferreyra@ u, UC)} (USM). LIMA: between Parac and Toncuyo, Coronado 314 APes FERNS OF PERU 95 Matucana, Macbride & Featherstone 420 (F, US). JUNIN: Tarma, (F, GH, US). HUANCAVELICA: se. of Pampas, Stork & Horton 10251 (F, UC) ; between Conaica and Laria, Tovar 920 (GH, USM). AYACUCHO: near Puquio, Ferreyra 7214 (GH, USM). APURIMAC: Andahuaylas, Stork & Horton 10717 (F and UC, in part; GH). CUZCO: Ollantaytambo, Cook & Gilbert 360 (GH, US); near Anta, Tryon & Tryon 5862 (BM, F, MO, U, US, USM). PUNO: Puno, Mexia 7780 (F, GH, MO, UC, US); Araran- ca, Pennell 13465 (F, GH, US). AREQUIPA: Arequipa, Pennell 13199 (FP, GH, US); 12 km. s. of Arequipa, Eyerdam & Beetle 22126 (GH, MO, UC). 5. Cheilanthes pilosa Goldm. Nov. Act. Acad. Caes, Leopold.-Carol. Nat. Cur. 19, suppl. 1(Meyen, Beitr. zur Bot.) : 455. atte: Type: Peru, Meyen, B! photo GH; isotype: s-PA! Fic. 58, Map J Fics. 54-62. Fig. 54. ene. myriophylla: ap upper en lo surface, igre Safford inn, 0. C tholae Pringle 449, GH. Fig. 61. C. Moritziana: central pinna, X 1, Ven GH. Fig. 62. C. chlorophylla: central pinna, X 4, Argentina, Schwarz 6191, 96 ROLLA TRYON Cheilanthes Macleanii Hook. Sp. Fil. 2: 93, t. 110B. 1852. Type: Andes of Peru, John MacLean, K! photo GH, US; isotype: GH! Cheilanthes andina Hook. Sp. Fil. 2: 115. 1852. Type: Andes of Peru, John MacLean, K! photo GH, US. Rhizome rather slender, short-creeping, scales lance-subulate, dark brown, rather sclerotic, with or without paler borders; leaves ca. 10-40 em. tall, petiole terete, somewhat pubescent, the rachis similar but sometimes grooved on the upper surface; lamina lanceolate to narrow- ly so, bipinnate-pinnatifid to tripinnate, pinnae pubescent above and more so beneath; indusium broad, deeply crispate or lobed. The soft spreading trichomes of the leaf and the rigid, shining rhizome scales are characteristic of this species. The rhizome is strongly multicipital although the individual branches are rather slender. Cheilanthes andina Hook. is a variant of C. pilosa, perhaps a response to drier or more exposed conditions. The ultimate segments are not as broad as usual and are thicker in tex- ture. The trichomes are frequently shorter than in the common form. I have seen one collection that is evidently a hybrid be tween C. pilosa and C. pruinata: Visachani, Dept. Cuzco, Biies 1782 (GH). Peru to Argentina. On ledges and in crevices of cliffs, Ancash to Puno, 2300- 4200 m. Selected specimens: ANCASH: Chiquidn, Cerrate 1548 (USM); be tween Llamac and Jahuacocha, Cerrate 2337 (USM). JUNIN: near Huancayo, Killip & Smith 23365 (us), Saunders 647 (GH); 15 km. & of Huancayo, Tryon & Tryon 5467 (F). HUANCAVELICA: Cerro sie Barbara, near Huancavelica, Tovar 3117 ( GH, USM). cuzco: Cerro 4 Cusilluyoe, Pennell 13994 (F, GH, US); Velille, Vargas 6541 ie Valle de Lares, Biies 1825 (US). PUNO: Puno, Mexia 7783 (F, see: cH UC, US); Araranca, Pennell 13454 (F, GH, US); Juliaca, Williams * (US), D. Stafford 404 (BM); Sandia, Weberbauer 717 (B). 6. Cheilanthes fractifera Tryon, Rhodora 62: 7. January, 1960. Type Dept. Ayacucho, Peru, Correll & Smith P1 69,GH! Fie. 59. os Cheilanthes Saundersii Alston, Lilloa 30: 110, t. 6. August, 19 afi Type: Dept. Lima, prov. Huarochiri, Peru, S. G. E. Saunders 350, BM the upper side, the ridges rounded and concolorous with t deciduous, breaking sharply and evenly toward the base, thinly w FERNS OF PERU 97 pubescent, as is the rachis, with large whitish scales at the base; lamina deltoid to broadly ovate, bipinnate-pinnatifid, pinnae moder- ately whitish pubescent with 2-5 celled trichomes beneath, thinly pubes- cent above; indusium more or less crenate or of small lobes, well odified. This species was first discovered in 1954 by S. G. E. Saunders and since then it has been collected several times. The petiole that fractures evenly toward the base, the large whitish scales at the base of the petiole and the aspect of the lamina combine to make it a distinctive species among the Peruvian ones. It is closely related, however, to C. Brandegei of Baja California. eru. Rocky hillsides, Lima to Arequipa, 1600-2200 m. Specimens seen: LIMA: Rimac valley, dist. Surco, Saunders 219 (BM, GH); Rimac valley, Rauh & Hirsch P143 (B). AYACUCHO: 45 km. from Nasca on road to Puquio, Correll & Smith P169 (GH). AREQUIPA: Cerros de Caldera, Rauh & Hirsch P570 (B). 7. Cheilanthes notholaenoides (Desv.) Weath. Contrib. Gray Herb. 114: 34.1936. Fic. 60. Pteris notholaenoides Desv. Mém. Soc. Linn. Paris 6: 298. 1827. Type: “Hispaniola,” P! photo GH. Cheilanthes micromera Link, Hort. Berol. 2: 36. 1833. Type: “Mexi- co,” B! photo GH. Rhizome rather slender, creeping, more or less knotted, scales lance-subulate, brown, with or without a darker center; leaves ca, 10-20 cm. long, petiole terete, pubescent, as is the rachis; lamina narrowly lanceolate, mostly bipinnate to bipinnate-pinnatifid, pinnae slightly pubescent to nearly glabrate, except on the axes; indusium rather broad, more or less crispate, continuous. This species is closely related to the next, C. Moritztana; the characters of the pinnae and rhizome scales given in the key afford a sufficient separation. In addition, the lamina of C. notholaenoides is rather regularly bipinnate and the pinnae are pinnate nearly to the tip, while that of C. M orit- ziana is bipinnate-pinnatifid and there is a definite pinnatifid apex of the pinnae. The rhizome is creeping but rather shortly so and it has a tendency to be multicipital. . West Indies; Mexico to Venezuela, Peru and Argentina. Cliffs and rocky places, Piura to Junin, 2800-3300 m. 98 ROLLA TRYON Specimens seen: PIURA: Chira valley, Rawh P1935 (B). CAJAMARCA: 52 km. w. of Cajamarca on road to Chilete, Correll & Smith P838 (GH). LIBERTAD: Retamas, Lépez & Sagdstegui 3603 (GH). HUANUCO: Yana- huanca, Macbride & Featherstone 1242 (F, GH, US). JUNIN: Ingahuasi, between Huancayo and Izcuchaca, Tovar 3870 (GH, USM) ; Uspachaca, Macbride & Featherstone 1307 (F, US). 8. Cheilanthes Moritziana Kze. Linnaea 23: 307. 1850. Lectotype: La Guayra (Caracas), Venezuela, Moritz 263, B! photo GH; isotype GH! Fie. 61. Rhizome rather slender, creeping, sometimes knotted, scales lanceo- late to lance-ovate, light brown to brown, concolorous; leaves ca. 15-40 cm. long, petiole terete, somewhat fibrillose especially on the upper side, rachis similar; lamina narrowly lanceolate, bipinnate-pinnatifid to tripinnate, pinnae only slightly pubescent above and below; indusium s more or less crispate, continuous. The differences from the closely related C. notholaenoides are discussed under that species. A critical study will be necessary to define C. Moritziana properly in relation to the Central American and Antillean C. microphylla Sw. It is traditionally maintained here as a segregate. Venezuela to Colombia and Bolivia. heltered, rocky places, Cajamarca and Amazonas to Cuz- co, 2000-3000 m Specimens seen: CAJAMARCA: alrededores de San Benito, Sagdsteg@ 3739 (GH); Celendin, Stiibel 1045 (B); Cantanoe, Celendin to Rio Marajion, Lépez & Sagdstegui 3365 (GH); Leimabamba, valle de Utcu- bamba, Stiibel 1021 (B). SAN MARTIN: Salinas de Pilluana, Ule 699 = JUNIN: Mufia, Macbride 3921 (F, US). cUzco: Yucay, Soukup 918 F). 9. Cheilanthes chlorophylla Sw. Kongl. Vet. Acad. Handl. 1817: 76. Type: Villa Rica, Brazil, Freyreis, (Herb. Sw.) s-PA! photo US: The material is fragmentary: a lower pinna and the central portion of a lamina. Fic. 62. Adiantopsis chlorophylla (Sw.) Fée, Gen. Fil. 145. 1852. _Rhizome compact to creeping, scales narrowly lanceolate to subulate, bicolorous, with a dark, sclerotic central portion and lighter, dis margins; leaves ca, 15-80 em. long, petiole shallowly grooved on bmembra- FERNS OF PERU 99 Ecuador s. to Bolivia and Argentina, to Brazil. In shady woods, Cuzco, 2500-2800 m. Specimens seen: CUZCO: Rio Chaupimayo, prov. Convencién, Soukup 806 (F); Hacienda Sahuayaco, prov. Convencién, Vargas 1659 (GH); Torontoy, Herrera 1332 (us). 10. on re radiata (L.) J. Sm. Jour. Bot. Hooker 4: 159. 1841. Fic. pposeete radiatum L. Sp. Pl. 2: 1094. 1753. LINN 1252.1, photo a, is this species ARintovnte radiata (L.) Fée, Gen. Fil. 145. 1852. ? Adiantopsis ternata Prantl, Gartenfl, 32: 101. 1893. Type: Orinoco- gebiet, Humboldt, B! photo GH. Rhizome erect to decumbent, often rather stout, scales subulate, bicolorous, with a dark, sclerotic central portion and lighter, thin margins; leaves ca. 10-50 cm. long, petiole glabrous, usually terete, to shallowly grooved at the apex, rachis glabrous or nearly so, grooved on asymmetrically circular, bipinnate, pinnae glabrous or nearly s0; indusium suborbicular to oblong, well modified. The radiate arrangement of the pinnae make this a very distinctive and attractive species. I am uncertain of the status of Adiantopsis ternata Prantl, Fic. 63. In addition to the type, I have seen Pennell 3685, Haught 2402 and Lehmann 2649 from Colombia and Buchtien 7029, Herzog 211 and R. S. Williams 1328 from Bolivia. This material may represent juvenile plants of C. radiata or, perhaps, plants growing under unusual conditions. Tropical America. In rocky ravines, on wooded slopes and in dense forest, Cajamarca and Amazonas to Cuzco, 200-2800 m. Selected specimens: CAJAMARCA: Jaén, Feb. 17, 1954, Cerrén (GH, UC); entre Jaén y Chamaya, Lépez et al. 4136 (GH). AMAZONAS: Rio Uteubamba, 40 km. s. of Bagua Grande, Hutchison 1469 (GH). SAN MARTIN: Juan Jui, Alto Rio Huallaga, Klug 4170 (F, GH, MO, UC, US, USM); San José de Sisa, Prov. Ferreyra 7889 (GH, USM). JUNIN: La Merced, Killip & Smith 23797 (F, US), Macbride 5375 (F, Us), Soukup 1114, 1162 (F). cUzco: Torontoy, Herrera 1299 (Us); Sonar aie: prov. Convencién, Vargas 1135 (GH). 100 ROLLA TRYON 11. Cheilanthes farinosa (Forsk.) Kaulf. Enum. Fil. 212.1824. Fic. Prors farinosa Forsk. Fl. Aegypt.-Arab. 187. 1775. Isotype: Forskal, BM! Rhizome rather stout, very short-creeping, decumbent to nearly erect, scales narrowly long-triangular, brown, semi-sclerotic, con- colorous; leaves ca. 10-70 em. tall, petiole terete, to grooved on the upper side, the ridges rounded, concolorous with the petiole, slightly scaly or not, glabrous to farinaceous, rachis grooved on the upper side, glabrous to farinaceous; lamina long-triangular to ovate-triangular or narrowly so, pinnate-pinnatifid to pinnate-bipinnatifid, pinnae glabrous to somewhat ceraceous on the upper surface, conspicuously to densely white-ceraceous beneath; indusium crispate, lobed, more or less con- tinuous, well modified. This is a most distinctive species because of the white waxy covering of the segments which is especially evident on their under surface. West Indies ; Mexico to Peru ; Old World. Rocky places, Hudnuco to Cuzco, ca. 2000 m. Specimens seen: HUANUCO: Muna, Bryan 555 (F, US). JUNIN: mie Huacapistana, Tryon & Tryon 5435 (F). cuzco: Nevada Salleantay, Biies 986 (us). 12. Cheilanthes Poeppigiana Kuhn, Linnaea 36: 84. 1869. Type: Peru, Poeppig. Fig. 66, MAP 17. a the fid, pinnae glabrous, pinnules adnate to somewhat narrowed at base; indusium continuous, intramarginal, well modified. Small laminae of this species may be lanceolate; i ones are ovate or broadly triangular. The segments se relatively broad and herbaceous and their broadly attac ase is characteristic. Ecuador to Argentina. , ines Crevices of cliffs or Inca walls, among rocks in *1700- and on shrubby slopes, Tumbes to Puno, 150- (usually 2200) 3000 m. FERNS OF PERU 101 Selected specimens: TUMBES: between Tumbes and Cancho, Coronado 226 (GH, UC). LAMBAYEQUE: entre Beatita de Humay y km, 38 (Carre- tera Olmos-Marajion), et al. 4042 (GH); 27 km. e. of Olmos on road to Jaén, Correll & Smith Ps02 (GH). CAJAMARCA: entre C Contumazaé, Lépez et al. 3725 (GH). LIBERTAD: Huaranchal, Sagdstegui 194 (GH). HUANUCO: Piedra Grande, Macbride 3672 (F). JUNIN: Huacapistana, Tryon & Tryon 5436 (BM, F, MO, U, US, USM); Carpa- Figs. 63-69. Fig. 63. Cheilanthes ternata: lamina, X %4, Colombia, aged — ge Fig. 64. C. radiata: lamina, X 4, Venezuela, Fendler 67, GH. Fig. 65. C. 9 inosa: lamina ico, Pringle 10816, GH. Fig. 66. C. Poeppigiana: central pinnae, base of fe fertile pinna, X 4, Peru, Macbride 3326, F. 102 ROLLA TRYON pata, Cerrate 2808 (GH). CUzCO: Machu-Picchu, Tryon & Tryon 5396 (BM, F, MO, U, US, USM), Mexia 8085 (F, GH, MO, US). PUNO: Sandia, Weberbauer 566 (B). 13. Cheilanthes Orbignyana Kuhn, Linnaea 36: 82. 1869. Type: Prov. La Laguna, Bolivia, D’Orbigny 388; isotype: P, photo GH. FI. Rhizome erect, moderately stout, scales lanceolate or narrowly so, brown to dark brown, concolorous, semi-sclerotic; leaves ca. 25-50 cm. long, petiole shallowly grooved on the upper side, at least toward the apex, flat to convex between the small, rounded ridges which are con- colorous with the petiole, glabrous or nearly so, rachis similar but the ridges somewhat lighter in color and rather sharp; lamina narrowly ovate, tripinnate-pinnatifid to quadripinnate, pinnae glabrous to slight- ly glandular; indusium suborbicular to lunate, well modified. Peru and Bolivia. Among rocks, Cajamarca, 2750 m. Specimen seen: CAJAMARCA: La Pampa, Guzmango, Sagdstegui 2937 (GH). Mars 17-18. Map 17, Cheilanthes Poeppigiana. Map 18, C. marginata. i | i f ; FERNS OF PERU 103 14. Cheilanthes rufopunctata Rosenst. Meded. Rijks Herb. Leiden 19: 9. 1913. Type: Araca, Bolivia, Herzog 2366; fragment Us!; L, photo GH. Fig. 68. Rhizome creeping, often knotted, scales lanceolate long-triangular, reddish oe wholly sclerotic; leaves ca. 15-35 cm. tall, iole n the upper side, at least toward 'the apex, deeply sulcate between the relatively large, rounded ridges which are concolorous with the petiole, glabrous, rachis grooved and glabrous, with green wings toward the apex; lamina long-deltoid to broadly lanceolate, tripinnate to tripinnate-pinnatifid, pinnae glabrate above and beneath, or with numerous reddish, sessile glands, especially beneath, indusium broad, crispate, well modified. Cheilanthes glauca (Cav.) Mett. of Chile, which is related to this species, has a pentagonal lamina, the segments of which are thickly beset beneath with short, dark trichomes which also extend in lines along the sides of the pinna- rachises and the rachis. Peru and Bolivia. Rock crevices, Lima to Puno, 2600-4000 m. Selected specimens: LIMA: Matucana, Macbride & Featherstone 285 (F, GH, Us), Rauh P181 (B); Viso, Macbride & Featherstone 759 LICA: near Conaica, Tovar 316 (GH, US, USM). CUZCO: Valle de Lares, Soukup 29 (F, US) ; Calea, Vargas 4004 (CUZ, UC, US) ; Canchis, Vargas 5007 (MO, US). PUNO: near Puno, Vargas 29 (MO), Soukup 29 (Uc) ; Vileanota, Rauh P706 (B). 15. Cheilanthes marginata HBK. Nov. Gen. Sp. 1: 22. 1815. Type: Penipe, Quito, Ecuador, Humboldt & Bonpland, P! photo GH. Fc. 69, AP 18, Rhizome short-creeping, somewhat multicipital, scales ovate- lanceo- late, attenuate, dark brown, sclerotic; leaves ca. 8-25 cm. tall, the petiole suleate, the ridges rounded, concolorous with the petiole, some- what scaly, rachis suleate, with green wings on the upper side above the base; lamina broadly deltoid to lanceolate, mostly tripinnate, pin- nae glabroas. ultimate segments stalked or narrowed at the base; indusium glandular-fimbriate, continuous enti the margins and de- current on the axes of the pinnules and pinnae The unusual condition of the indusium which extends from the segments along their stalks and along the pinna- rachises mark this species as one of the most distinctive in Peru. Venezuela to Colombia, south to Argentina. 104 ROLLA TRYON Cliffs, among rocks and in rocky soil, Piura to Puno, 2200-4000 m. pay specimens: PIURA: near Ayabaca, 1927, Berry (US). LIBER- : Huaranchal, Sagdstegui 207, 2664 (GH). CAJAMARCA: Cajamarca i 5 ae Ferreyra 3294 (USM); El Puquio, Guzmango, Sagdstegu $918 ( GH). ANCASH: e. of Huasta, Cerrate 2264 (USM). HUANUCO: cayo, G. Kunkel 481 (GH). HUANCAVELICA: Colcabamba to Paurcar- amba, Tovar 1969 (USM). AYACUCHO: Pampalea, Killip & Smith 22234 (US). CUZCO: Machu-Picchu, Tryon & Tryon 5401 (BM, F, MO, U, US, USM), Ferreyra 2684 (GH); Valle de Lares, Biies 1798 (Us). PUNO: Sandia, Soukup 157 (F), Weberbauer 718 (B). Fic. 70. Notholaena nivea, Cuzco area. FERNS OF PERU 105 16. NOTHOLAENA R. Br. Prod. Fl. Nov. Holl. 145. 1810: Type: Acros- tichum Marantae L. = Notholaena Marantae (L.) Desv. Fic. 70. Terrestrial, the rhizome small to moderately stout, compact to creep- ing, scaly, bearing the leaves in a cluster or loosely spaced; leaves small to medium sized, pinnate-pinnatifid to quadripinnate, glabrous, ceraceous, pubescent or scaly, veins free; sporangia borne on the apical % to % of the vein or at its tip, partially covered by an indusium formed from the reflexed, slightly modified margin, or the margin reflexed to plane and unmodified, paraphyses absent. — 62 American species, Tryon, R. A revision of the American species of Notholaena. Contrib. Gray Herb. 179. 1956. Weatherby, C. A. The Argentine species of Notholaena. Lilloa 6: 251-275. 1941. These are small ferns with tufts of leaves, characteristic of the xeric and semi-xeric rocky places of the Altiplano and the higher elevations of the eastern and western Andean valleys. The leaves resist desiccation to a considerable de- gree and the roots evidently draw on local sources of water (perhaps night condensation in part) for the plants are usually fresh when other vegetation is dry and dormant. The rhizome contains a persistent (up to 50 years in her- barium specimens) oily substance which may well be related to its water-holding capacity. Notholaena tomentosa Desv., with pubescent leaves and a strongly sulcate petiole and rachis, is reported from Peru by Pichi-Sermolli (Webbia 8: 181, 186, 1951) but the records must be considered as dubious until they have been confirmed by modern collections. KEY TO SPECIES a. Indument of the lamina of scales or trichomes, or both, not cerace- b ous. b. Tidument of the lamina of scales (in N. sinuata some of these may be dissected into capillary lobes), sometimes also of trichomes. c. c. Indument of scales only. seen d. Lamina lanceolate or broader, bipinnate to tripinnate, pinnae few; petiole commonly at least half as long as the lamina, with one vascular bundle. e. 106 ROLLA TRYON e. Scales of the rhizome entire or minutely serrulate with very short ascending teeth; scales of the upper surface of the lamina soon deciduous, oo flat, or if narrow and subpiliform then 1 mm. or less long, scales of the lower surface erose-serrulate with mostly deltoid teeth. ...........0 a f. Lamina lanceolate, mostly 12-18 em. tall; rhizome scales entire or nearly so; scales of nie blade dark chestnut brown; dilated soriferous vein ends projecting into small, hyaline, marginal lobes. 1. N. peruviana f. Lamina deltoid-ovate, mostly 2-4.5 cm. tall; rhizome scales minutely serrulate; scales of the lower surface of the lamina bright or pale brown; soriferous vein-ends in slight, unmodified crenations of the Margin. ..........+0 . arequipensis e. Reales of the rhizome remotely Girone eerie scales of the upper surface of the lamina piliform, more or less persistent, more than 1 mm. long, scales of the lower surface cepa -serrulate, with often piliform, although broad- based, teeth. . g g. cae ovate or elliptic, about equaling, or shorter than the petiole; piliform scales of the upper surface of the lamina brown, those of the lower surface lance-ovate or ovate, 3-4 mm. long. 3. N. squamosa g. Lamina Sain much longer than the petiole; pili- orm scales of the upper surface of the lamina whitish, those of the lower surface lance- or linear-subulate, uP to 2 mm. long. 4. N. lonchophylla d. Lamina linear with numerous merely lobed pinnae; petiole much shorter than the lamina, with two vascular bundles 5. eae © Tndugnent of both scales and trichomes. ........ 6. N. cantangensis b. Indument of the lamina of trichomes OnLy. ....s..sss+sssssssseeseseeneeeetes h. ' hh. Indument of simple trichomes. .....c..scscsssscssscsvssssvsessssesssssessseneseeeets i. Lamina bipinnate, much longer than the el is segments appressed and ascending pubescent beneath. 7. N. obducta i. Lamina not much longer than the petiole, or it so, then uni- formly PUM ALS-PANNAEIG, onic coscnnicecesstcesssccacicncssecetossavessenrenaters j. j. Lamina gradually reduced at base, usually much longer? than the petiole, uniformly pinnate-pinnatifid, the pinnae equilateral, long-deltoid to oblong; rhizome short, — knotted, the tips of the scales dark brown. ......-.-- 8. N. @ 4 Latitina not conspicuously reduced at ‘the base, nor te longer than the petiole, pinnate-pinnatifid, the pinnae inequilateral, deltoid; rhizome creeping, horizontal, wie tips of the scales light brown 9. N. Fir h. Indument of stellate trichomes. a. Indument of lamina wholly ceraceous, or lacking. «1... k. Rhizome scales bicolorous, with a dark esa to plackis portion and narrow, lighter mar ns; pinnules adnate or dominantly so : e : 11. N. sulphures FERNS OF PERU 107 k. Rhizome scales concolorous, brown; pinnules stalked, or some of them sessile. 1, 1. Ultimate segments persistent, mostly orbicular or nearly so; me merle scales oily. 12. N. Stuebeliana 1. Ultimate segments articulate, oblong to ovate, or some of them nearly orbicular; rhizome scales Ary. ......sscsseeseeees 13. N. nivea 1. Notholaena peruviana Desv. Mém. Soc. Linn. Paris 6: 220, 1827. Type: Peru, Dombey, P! photo GH, UC, fragment B! Fie, 71. Notholaena Brackenridgei Baker, Syn. Fil. 371. 1868. Type: lean (evidently Bafios, Peru, Wilkes Exped.), K! te a sotyp Rhizome erect, stout, scales linear-ligulate, long-attenuate, margins entire or nearly so; leaves to 30 em. tall, the petiole somewhat shorter to much shorter than the lamina, with one vascular bundle; lamina narrowly lanceolate, not narrowed below, bipinnate to tripinnate, with some pinnules lobed, the lower surface densely covered with dark brown, lanceolate to ovate, somewhat pectinate-serrulate scales; spor- angia borne on the vein-ends, the margin modified, narrow. This and the next species are closely related and may be separated by the characters mentioned in the key. From N. lonchophylla, they are distinguished by the deciduous rather than persistent scales on the upper surface of the lamina, of different shape, and the erose-serrulate, rather than pectinate-serrulate scales, on the lower surface. The lamina of N. lonchophylla is proportionately narrower than in N. peruviana or N. arequipensis and the pinnae are more regularly divided. Peru. In soil and on rocks of lomas, tee on exposed rocky slopes, Lambayeque to Moquegua, 300-3900 m. Selected specimens: LAMBAYEQUE: 44 km. from Olmos on road to Jaén, Correll & Smith P822 (GH). CAJAMARCA: entre Cascas y Contu- maza, Lépez et al. 3688 (GH); El Puquio, Guzmango, Sagdstegui 3913 (GH). LIBERTAD: Huaranchal, prov. Otuzco, Lépez et al. 2665 (GH) LIMA: High Mts. above Tama, Safford 999 (p, us), 1000 (Us); Matu- cana, Macbride & F 284, 421 (F), Bryan 43 (F); above San Bartolomé, Ferreyra 9742 (GH, usM); above Surco, Ferreyra 3458 (GH,USM). AREQUIPA: Mollendo, Dora Stafford 824 (BM); Mollendo (loma), Weberbauer 1528 (B); Lomas de Capac, near Chala, Ferreyra 1426 (UsM), Coronado 39 (GH, UC, US); Lomas de Atiquipa, Coronado 35 (GH, UC, US) ; Lomas de Chaparra, cerca Chala, Ferreyra 11968 (GH, USM). MOQUEGUA: Ilo, Dora Stafford 933 (BM). 108 ROLLA TRYON 2. Notholaena arequipensis Maxon, Smiths. Misc. Coll. 65(8) :9. 1915. Type: near Arequipa, Peru, Rose & Rose 18797, US! photo GH. FI. Rhizome erect or ascending, scales linear, long-attenuate, distantly denticulate-with ascending teeth; leaves to 8 cm. tall, petiole somewh longer than the lamina, with one vascular bundle; lamina deltoid- oblong, bipinnate, the pinnules hardly if at all lobed, the lower surface densely covered with large, ovate-oblong, long-acuminate, light reddish brown, erose-denticulate scales; sporangia borne on the vein-ends, the margin somewhat revolute, unmodified. A local species and one closely related to the previous; the characters by which the two may be distinguished are presented in the key. Southern Peru to n. Chile and nw. Argentina. Rocky places and crevices of rocks, Arequipa, 2200-2750 m. 3. Notholaena squamosa (Hook, & Grev.) Lowe, Ferns Brit. & Exot. 1: 49.1856. Fig, 73. Cheilanthes squamosa Hook. & Grev. Icon. Fil. t. 151. 1829. Type: Cerro del Morro, San Luis, Argentina, Gillies, E; isotype: K! Pellaea Lilloi Hicken, Anal. Soc. Cient. Argent. 62: 210. 1906. Type: Tucumaén, Argentina, Lillo 5021, SI! photo GH. Notholaena Lilloi (Hicken) Hicken, Apuntes Hist. Nat. Buenos Aires 1: 117. 1909. Rhizome short, erect or oblique, scales thick, subsclerotic, linear, tapering to a long capillary point, pectinate-serrulate, the teeth often antrorse; leaves to 30 cm. tall, petiole about half as long as, to some- what longer than, the lamina, with one vascular bundle; lamina oblong- lanceolate to broadly ovate, bipinnate, the lower surface densely ith soft, imbricate, brown, lance-ovate to ovate, pectinate- fimbriate scales; sporangia at the apex of the veins, margin unmodified. Southern Peru to Argentina. Calcareous hills, Puno, 3900 m, Specimen seen: puno: Camjata Hacienda, Capachica peninsula (Lake Titicaca), Tutin 1039 (BM), FERNS OF PERU 109 Fics. 71-74. Fig. Notholaena peruviana: A, leaf, X %4, Peru, J. B. Steere, Mo; B, segments, lower Singite X 414, idem; C, scale from lower surface of segment, X 9, i i i is: A, leaf. . ide eaf, 610, K; B, seale from lower hese of segment, X 9, idem; C, ibiaei scale, 110 ROLLA TRYON 4, Notholaena lonchophylla Tryon, Contrib. Gray Herb. 179: 19. 1956. Type: Peru, Mathews 610, K! photo GH. Fa. 74. Rhizome short, horizontal or ascending, scales linear-ligulate, spar- ingly antrorsely denticulate; leaves to 15 em. tall, petiole shorter than the lamina, with one vascular bundle; lamina lanceolate, pinnate-pin- natifid to bipinnate, the lower surface densely covered with pale brown to whitish, linear-subulate, long-acuminate, pectinate-dentate scales; sporangia at the apex of the veins, margin unmodified. Although one of the most recently described species from Peru this is also the rarest; quite possibly it is extinct. The single plant collected by Mathews in 1835 is our sole record. For the benefit of anyone keen enough and fortunate enough to find it, the differences from the related N. peruviana and N. arequipensis are discussed under the former species. Endemic to Peru. Specimen examined: PERU: 1835, Mathews 610 (XK). 5. Notholaena sinuata (Sw.) Kaulf. Enum. Fil. 135. 1824, var. sinuata. Fic. 75, MAP 19. Acrostichum sinuatum Sw. Syn. Fil. 14. 1806. Type: Peru, “Squa- maria sinuata, Lagasca Herb.” (Herb. Sw.), S-PA! Notholaena Tectaria Desv. Mém. Soc. Linn. Paris 6: 219. 1827. Type: Peru, P! photo GH. Tarma, Dombey, “N. Tectaria Desv. eX Desv.” B! is probably an isotype, as is Tarma, Dombey, BM! ; Rhizome short, horizontal, scales linear to linear-subulate, pectinate- ciliate to entire; leaves up to 45 em. tall, petiole usually less than 4 the length of the lamina, with two vascular bundles; lamina linear, pinnate-pinnatifid, the lower surface thickly covered with castaneous or pale brown, deltoid to lanceolate, acuminate, short-fimbriate scales, these overlying a tomentum of smaller scales dissected into long capil lary segments; sporangia borne on the terminal portion of the veins, margin unmodified. The two vascular bundles in the petiole serve to distin- guish this species from the others of the genus. In addition, the scaly, linear lamina, much longer than the petiole, and the lobed pinnae are characteristic. Many of the scales on the segments are dissected into capillary processes, giving the indument as a whole a tomentose appearance. : The species is distributed from Argentina to Colombia and Venezuela, and northward to the southwestern United FERNS OF PERU 111 States and in Hispaniola. In Mexico and the southwestern United States there also occur two other varieties. Crevices of rocks, rocky banks and hillsides, Lambayeque, Cajamarca and Amazonas to Cuzco and Arequipa, 1000- 3800 m. elected specimens: LAMBAYEQUE: 20 km. from Olmos on road to Jaén, Correll & Smith P795 (GH). CAJAMARCA: entre Cascas y Contu- maza, prov. Contumaza, Lépez et al. 8687 (GH). AMAZONAS: Tambo de Carrizal, Stiibel 1041, in part (B). LIBERTAD: Huaranchal, Sagdstegui 210 (GH). ANCASH: s. of Chiquidn, Cerrate 7589 (USM), HUANUCO: Ambo, Macbride 3190 (F, US); Mito, Macbride & Featherstone 2310 (¥, GH, US). LIMA: Santa Eulalia, Goodspeed 11307 (GH, MO, UC); between Surco and Matucana, Ferreyra 5434 (USM). JUNIN: Tarma, Killip & Smith 21808 (us); Cerro San Sebastian, Tarma, Ferreyra 580 (GH, USM). AYACUCHO: Nasca to Puquio, Correll & Smith P1i71B (LL). CUZCO: Quillabamba, Stork et al. 10449 (MO). AREQUIPA: near Are- quipa, D. Stafford 1289 (BM); Tiabaya, Pennell 13071 (F, GH); Yura, prov. Arequipa, Vargas 8018 (UC). 6. Notholaena cantangensis Tryon, Rhodora 63: 81. 1961. Type: Can- tange, prov. bane ge Dept. Cajamarca, Peru, Lépez & Sagdstegui 3366, GH! G. 76. Rhizome short, scales lanceolate with a dark, sclerotic central portion and brown, pectinate-serrulate margins; leaves 10-15 cm. tall, petiole shorter than the lamina, with one vascular bundle; lamina lanceolate- elliptic, bipinnate-pinnatifid, the lower surface covered with narrowly ovate-lanceolate scales, the upper surface moderately pubescent with rather short and thick trichomes; sporangia borne at the apex of the veins, margin slightly modified. Peru. Among rocks, Cajamarca, 1450 m. Specimen seen: CAJAMARCA: ages ruta Celendin-Rio Marajion, Lopez & Sagdstegui 3366 (GH 7. Notholaena obducta (Kuhn) Baker, Syn. Fil. ed. 2, 515. 1874. Fic. 77. Cheilanthes obducta Kuhn, Linnaea 36: 83. 1869. Type: Bolivia, D’Orbigny 386, B! photo GH Rhizome short-repent, scales narrowly linear, pale brown, or with a castaneous, sclerotic central portion and lighter margins; leaves to 55 em. long, petiole much shorter than the lamina, with one vascular bundle; lamina lanceolate, bipinnate, the lower surface densely sub- 112 ROLLA TRYON 75A 3 PUG ER ath hee Dihees rn) G3 4 f\ tts eo f vs rhicss y OG 8 at WB mt Figs, 75-78. Fig, 75, Notholaena sinuata var. sinuata: A, leaf, X 1%, New —_ U.S., Wooton i 89 : ed, id ; ; lamina, enlarged, idem. Fig. 76. N. antangensis: one ie GH; B, central pinnae, lower surface, 1, ide 24 rs “> me So 13656, MO; B, segment, lower s Peldnctge dy Paes 44, Texas, U.S., E. J. Palmer 30608, mo; B, D , z m, Fig. 77. N. obducta: urface, X 7, idem. Fig. inna, lower FERNS OF PERU 113 appressed pubescent, the trichomes rather straight, sporangia borne at the apex of the veins, margin unmodified. The trichomes on the lower surface of the pinnae are unusual, having from 3 to 6 of the lower cells branching from the trichome proper. Colombia, Peru, Bolivia to Paraguay. Under shrubs and trees, Cajamarca, 680 m. Specimen seen: CAJAMARCA: Valley of the Rio Chamaya, Mesones to Muro, highway between Olmos and Rio Marafion, Hutchison 1425 (UC). 8. typtaameen aurea (Poir.) Desv. Mém. Soc. Linn. Paris 6: 219. 1827. Fig. 78, Map Pteris aurea on oir. Lam. Encycl. Méth. 5: 710. 1804. Type: Peru, Joseph de Jussieu ( Sigh, Jussieu no. 1033), P! photo GH. Notholaena ferruginea (Link) Hook. Second Cent. Ferns, sub t. 52. Notholaena bonariensis (Willd.) C. Chr. Ind. Fil. 6. 1905; 459. 1906. Maps 19-20. Map 19, Notholaena sinuata var. sinuata. Map 20, N. aurea. 114 ROLLA TRYON Rhizome short-repent, knotted, scales lance-linear, entire; leaves to 60 em. tall, petiole 1/3 as long as the lamina or less, with one vascular bundle; lamina linear-elliptic, long-attenuate at the base, pinnate- pinnatifid, the lower surface covered with a dense tawny (white when young) tomentum of fine matted trichomes; sporangia borne on the vein-ends, margin modified with a very narrow subhyaline band. Notholaena aurea is one of the characteristic ferns of the Altiplano. It is usually quite distinctive in the aspect of its pubescent lamina but some variations must carefully be distinguished from N. Fraseri. Southwestern United States to Chile and Argentina; West Indies. On soil banks, rocky slopes, shrubby hillsides or cliffs, Lambayeque and Amazonas to Arequipa and Puno, 1200- 3200 m. Selected specimens: LAMBAYEQUE: 44 km. from Olmos on road to Jaén, Correll & Smith P821 (GH). AMAZONAS: Prov. Chachapoyas, Matthews 205 (BM, K), 3289 (B); Tambo de Carrizal, Stiibel 1041, in part (B). LIBERTAD: Huaranchal, Sagdstegui 211 (GH); Munmalca, prov. Huamachuco, Lépez & Sagdstegui 2826 (GH), ANCASH: Chiquian, Ferreyra 5786 (UsM), Cerrate 788 (USM). HUANUCO: Mito, Macbride & Featherstone 1921 (¥, US); between Ambo and Huanuco, Ferreyra 9234, 10394 (GH, USM). LIMA: Oroya RR., Safford 991 (F, GH, P, US); Cajatambo, Ferreyra 3545 (USM) ; Matucana, Macbride & Featherstone 370, 423 (F). JUNIN: Carpapata to Huacapistana, Ferreyra 11020 (GH, USM) ; Vileabamba, Tryon & Tryon 5421 (F, BM, MO, U). HUAN- AREQUIPA: Arequipa, D. Stafford 589 (BM); near Laspinas, Prov: Arequipa, Eyerdam & Beetle 22144 (Gu). 9. Notholaena Fraseri (Kuhn) Baker, Syn. Fil. ed. 2, 83. 1874. F1¢. 79, Map 21 Cheilanthes Fraseri Kuhn, Linnaea 36: 83. 1869. Syntypes: Ecuador, Fraser Wagner and Peru, Ruiz & Pavon, B! (a sheet containing both collections) photo GH. Rhizome short- to moderately long- ly linear- subulate, repent, scales narrowly entire or remotely serrulate; leaves to 40 cm. tall, petiole about halt as long as the lamina, with one vascular bundle; lamina narrowly linear, pinnate-pinnatifid or bipinnatifid at the base, the “a ie y tomentose with rather coarse, m tted pale anes ne sporangia borne on the vein-tips, margin slightly FERNS OF PERU 115 The differences between this species and the sometimes similar N. aurea are presented in the key. Notholaena Buch- tienii Rosenst., a species of Bolivia, may well also occur in Peru. It is quite closely related to N. Fraseri and the two may be separated as follows: ‘3 ue Coe, anfext> Bo” ets Zi tr f 3 GP ey "e D, ae cain 79A Fics. 79-80, Fig. 79. Notholaena Fraseri: A, leaf, X 44, Peru, aisles eg Aa B, pinna, upper surface, X 414, Peru, Mexia 8055, MO; C, rhizome scale, xX 18, idem. Fig. 80. N. mollis: A, leaf, X 14, Chile, Wagenknecht 18571, Mo; B, pinnule, X 4%, idem; C. trichome from lamina, enlarged, idem. 116 ROLLA TRYON Rhizome with numerous pale scales with conspicuously tortuous capillary tips, giving a tomentose appearance; no scales on rachis or upper part of petiole; median pinnae mostly deltoid, veins not visible on the upper surface. N. Frasert Rhizome usually with a few and inconspicuous pale, capillary-tipped scales, not appearing tomentose; a few deciduous scales among the tomentum on the upper part of the petiole and the lower part of the rachis; median pinnae mostly lanceolate or deltoid-lanceolate, the veins usually subimpressed and visible on the upper surface. .... N. Buchtientt Ecuador to Bolivia. Rocky places, often shaded, Piura to Cuzco, 1000-3500 m. Selected specimens: PIURA: Huancabamba, Scolnik 1435 (UC). CASA- MARCA: Santa Cruz, Weberbauer 4139 (B); between San Marcos and Cajabamba, Correll & Smith P909 (GH). HUANUCO: Hudnuco, Mac- bride & Featherstone 2328 (F, GH, US); Ambo, Ferreyra 6912 (GH, USM). LIMA: km. 68, Carretera Central, Saunders 220, in part (BM). HUANCAVELICA: Rio Mantaro, 70 km. s. of Mejorada, Hutchison 1689 (GH, UC); entre Pampas y Salecabamba, prov. Tayacaja, Tovar 3848 (GH, USM). APURIMAC: Quisapata, prov. Abancay, Vargas 8910 (MO, uc) 3 Ca. 10 km, above Pajonal, West 3676 (GH). cUzco: Rio Sambray, . — 8055 (F, GH, MO, UC, US) ; Quillabamba, Stork et al. 10449 (F, GH, 10. Notholaena mollis Kze. Linnaea 9: 54. 1834. Type: Playa Ancha, near Valparaiso, Chile, Poeppig, Lz, destroyed. Fic. 80. Rhizome short-creeping, knotted, scales narrow-linear, entire; leaves to 30 cm. tall, petiole usually shorter than the lamina, with one vascu- lar bundle; lamina lanceolate, to tripinnate, ultimate segments small, suborbicular, their lower surface densely covered with whitish to ferrugineous, stellate trichomes; sporangia borne on the vein-ends, margin unmodified, strongly revolute. N otholaena mollis is amply distinguished by its indument of stellate trichomes, and its tripinnate lamina with small, suborbicular ultimate segments. Southern Peru and Chile. On lomas, Arequipa, 100 m. Specimen examined: AREQUIPA: Mollendo, Weberbauer 1545 (B). 11. Notholaena sulphu : tthe phurea (Cav.) J. Sm. Bot. Voy. Herald 1: 233. Pteris sulphiaen Cav. De i ‘ ji . Deser, 269. 1802. Type: Chimapan (Zimapan), Hidalgo, Mexico, Née, MA; isotype: F! Gg! - ao Cheilanthes Borsigniana Koch, Wochen. Gartn. Pflanzenkr. 1: 2 FERNS OF PERU 117 1858. Type: Peru, Warszewicz, B! photo GH. Notholaena candida var. lutea Hook. Sp. Fil. 5: 111. 1864, not Pteris lutea Cav. Type: Hudnuco, Peru, Mathews 981 >< Notholaena sulphurea var. flava Kuhn, Abhandl. Ges. Halle 11: 32. 1869. Type: “Peru.” Rhizome short-horizontal, scales lance-ovate, attenuate, with decidu- ously glandular-ciliate margins; leaves to 20 em. tall, petiole usually much longer than the lamina, with one vascular bundle; lamina penta- gonal, bipinnate-pinnatifid at the base, pinnate-pinnatifid above the basal pinnae, ultimate segments adnate, the lower surface densely yellow to whitish-yellow ceraceous; sporangia borne on the vein-ends, margin thick, otherwise unmodified, somewhat revolute. Among the three Peruvian species with waxy indument, N. sulphurea is easily distinguished by its adnate segments, N. Stubeliana and N. nivea having stalked ones. The wax is usually yellow but it varies to a very pale yellow. Mexico to Chile. Open rocky places, Amazonas to Arequipa, 1500-2500 m. 19 Maps 21-22. Map 21, Notholaena Fraseri. Map 22, N. nivea. 118 ROLLA TRYON Selected specimens: CAJAMARCA: Cantange, Celedin to Rio Marafion, Lopez & Sagdstegui 3367 (GH). AMAZONAS: Sunibamba, Utcubamba valley, Stibel 1048, in part (B). ANCASH: Caraz, Weberbauer 3007 (B). ANU Grande, Macbride & Featherstone 2051 (GH, US), Stork & Horton 9395 (F, UC). LIMA: Mts. back of Lima, Safford 993 (US); Matucana, Macbride & Featherstone 426 (US). HUANCAVELICA: Valle de Mantaro, entre Pampas y Saleabamba, Tovar 3840 (GH, USM). APURIMAC: 80 km. from Abancay, on road to Challhuanca, Saunders 769 (GH). AREQUIPA: Arequipa, Rauh P567 (B). 12. Notholaena Stuebeliana (Hieron.) Tryon, Rhodora 63: 83. 1961. Fig. 82 Pellaea dealbata var. Stuebeliana Hieron. Hedwigia 48: 225, t. 12, f. 15, 1909. Type: Sunibamba, Dept. Amazonas, Peru, Stiibel 1048, in part, B! Rhizome short, erect or decumbent, scales linear, margin entire, brown, oily, concolorous ; leaves 10-25 cm. long, petiole as long as, or longer than, the lamina, with one vascular bundle; lamina deltoid to long-triangular, tripinnate to quadripinnate, the lower surface densely white to very pale yellow ceraceous, ultimate segments persistent, most- ly orbicular or nearly so; sporangia borne on the terminal %4 of the vein, margin unmodified. Peru. Among boulders and on ledges on shrubby slopes, moist ravines and rocky soil, Cajamarca and Amazonas to Huan- uco, 1800-3000 m. Specimens seen: CAJAMARCA: 40 km. from Cajamarca on road to Chilete, Correll & Smith P841 (GH, LL, Us) ; Celendin, Lépez & Sagas- tegui 3104 (GH). AMAZONAS: Sunibamba, valle Rio Uteubamba, Stiibel 1048, In part (B). LIBERTAD: Camino de las Quishuas, Bolivar, Lépez & Sagdstegui 3288 (GH); Retamas, Lépez & Sagdstegui 3606 (GH). HUANUCO: Mito, Bryan 192 (F). FE etc nivea (Poir.) Desv. Jour. Bot. Appl. 1: 98. 1813. Pellaea nivea (Poir.) Prantl, Bot. Jahrb. 3: 417. 1882. Rhizome short, erect or somewhat oblique, scales linear-subulate, margin entire; leaves to 30 cm. tall, petiole shorter than the lamina or about equalling it, with one vascular bundle; lamina lanceolate, deltoid-lanceolate to ovate, to tripinnate, the ultimate segments stalked, the lower surface densely yellow or white ceraceous or glabrous; — borne on the terminal] % to % of the veins, margin unmodi- FERNS OF PERU 119 This species is a characteristic fern of the Altiplano me the rocky, semi-arid valleys to the east and west. The fou varieties all occur in Peru and often two may be Rind growing in the same habitat. = em AD > Le q oret. . eo {] ‘i. ae a Fics, 81-85. Fig. 81. Notholaena sulphurea: A, leaf, X 14, Me ee i a, xX 414, idem; G, ‘ 882, MO; B, segments, lower surface, 1% : i a Wie EON. o; B, segm ta: segment, lower surface, X 444, Argen- tina, “Lossen. 2 242, mo. Fig. 85. N. nivea var. tenera: segment, lower surface, X 4%, Argentina, Venturi 10639, Mo. 120 ROLLA TRYON Colombia to Argentina and Brazil. Rocky hillsides and crevices of rocks, Cajamarca to Puno and Arequipa, 1200-3800 m. KEY TO VARIETIES a. Indument white. - be b. Ultimate segments suborbicular to broadly oblong, the terminal one often lobed, petiole castaneous, rhizome scales often strongly i v 13a. var. nwvea b. Ultimate segments oblong and entire, or subdeltoid and trilobate, the terminal one commonly simple, petiole bright castaneous, rhi- zome scales not strongly crisped. ceecccccccccsscoee.s... 3b. var. oblongata a. Indument yellow or none be ce. Indument none. ................... 13. var. tenera snsaidloncadehunasiens 13d. var. flava 13a. Notholaena nivea var. nivea. Fics. 70, 83. Pteris nivea Poir. Lam. Eneyel. 5: 718. 1804. Type: Peru, Joseph de Jussieu (Herb. Jussieu no. 1047), P! photo Gu. Ecuador to w. Argentina, Libertad to Puno and Arequipa. Selected specimens: LIBERTAD: Otuzco to Huamchirco, Ferreyra 2987 Stork & Horton 1 0823 (F, UC) ; between Coleabamba and Paucarbamba, Tovar 2110 (GH, USM). AYACUCHO: Ayacucho, West 3642 (UC). APURI- wac: Andahuaylas, Stork & Horton 10718 (F, uc), cUzCO: Saxi- huaman, Tryon, & Tryon 5344 (BM, F, GH, MO, U, US, USM), Coronado 158 (GH, UC). PUNO: Baja Isla, Lake Titicaca, Mexia 7788 (F, GH, MO, UC); Tequefia, Aguilar 133 (USM). AREQUIPA: Arequipa, Pennell 13245 (F); El Misti, D. Stafford 572 (BM). 13b, Notholaena nivea var. oblongata Griseb. Abh. Kénig. Ges. Wiss. Gott. 24(Symb. Fl. Argent.): 342. 1979, Type: Salta, Argentina, Hieronymus & Lorentz 142 and 161, B! photo aH. Fic. 84. Peru, Argentina and Brazil. Huanuco to Cuzco, Specimens seen: HUANUCO: Chavenillo, Woytkowski 1028 (UC). FERNS OF PERU 121 APURIMAC: Andahuaylas, Herrera 1498, in part (GH). CUZCO: Cuzco, Soukup (USM), 76 (GH). PUNO: Macusani, Lechler 1830 (B). 13c. Notholaena nivea var, tenera (Hook.) Griseb. — Konig. Ges. Wiss. Gott. 24(Symb. Fl. Argent.) : 342.1879. Fic. 8 Notholaena tenera Hook. Curtis, Bot. Mag. t. 3055 a text. 1831. Type: Spec. cult, ex Mendoza, Argentina, Gillies, K! photo GH. Pellaea tenera (Hook.) Prantl, Bot. Jahrb. 3: 417. 1882. Pellaea peruviana Copel. Univ. Cal. Publ. Bot. 19: 302. 1941. Type: Abancay region, Apurimac, Peru, pupils of V. Santander C. vc! A collection by Soukup (1488, USM) is said to be from Cerro Jeronimo, a loma near Lima, but this represents the only occurrence of Notholaena nivea on a loma and the record needs to be confirmed. Peru to Argentina. Libertad to Puno Selected specimens: LIBERTAD: arriba de Casmiche, prov. Otuzco, Lépez et al. 3962 (GH). ANCASH: Cerro de Huasta, Cerrate 2456 (GH, USM). HUANUCO: between Huanucay and Ambo, Ferreyra 6566 (USM). LIMA: Matucana, Macbride & Featherstone 82 (F, GH); entre Matu- cana y San Mateo, Ferreyra 5306 (GH, USM). JUNIN: Palea, Correll & unin P758 (GH). HUANCAVELICA: posrnipeia Tovar 1000 (GH, USM). UzCO: Cuzco, Tryon & Tryon 5353 (BM, F, MO, U, US, USM); Yucay, Cini 142 (GH, UC). PUNO: Tequefia, a 19, 1948, Aguilar (USM). 13d. Notholaena nivea var. flava Hook. Sp. Fil. 5: 112. 1855. Type: ? Acrostichum tereticaulon Desv. Ges. Naturf. Freunde Berl. Mag. 5: 310. 1811. Type: perhaps Peru, Dombey, P! photo GH. Notholaena chrysophylla Kl. Allg. Gartenzeit. 23: 265, 1855. Type: spec. cult. ex Peru, Warszewicz, B Pellaea flavens (Sw.) C. Chr. ea, Fil. 480. 1906. Colombia to Argentina and Brazil. Cajamarca to Cuzco. Specimens seen: CAJAMARCA: Between San Marcos and Cajabamba, Correll & Smith P908, P912 (GH). HUANUCO: Huanuco, Pearce 54 (BM) ; Piedra Grande, Macbride & Featherstone 3706 (F, US). HUAN- CAVELICA: Valle de Mantaro, entre Pampas y Saleabamba, Tovar 3847 (GH, USM). CUZCO: Potrero, Tryon & Tryon 5378 (BM, F, MO, U, US, USM), Coronado 111 (GH, UC, US), Vargas 1807 (GH). 17. PeLtara Link, Fil. Sp. Hort. Bot. Berol. 59. 1841, conserved name. Type: Pteris atropurpurea L. = Pellaea atropurpurea (L.) Link, Terrestrial, the rhizome moderately stout and compact to slender 122 ROLLA TRYON and creeping, scaly, bearing the leaves in a cluster or spaced; leaves small to large, 1- pinnate to quadripinnate, glabrous or nearly so, veins free; sporangia borne on the apical %4 of the vein or at its tip, covered by an indusium which is formed from the reflexed, modified margin, paraphyses absent. — About 25 species in America. Tryon, A. F. A revision of the fern genus Pellaea section Pellaea. Ann. Mo. Bot. Gard. 44: 129-193. 1957. The treatment of this genus has been adapted by Alice F. Tryon from the monograph cited above. The American species are well represented in the southwestern United States and Mexico. The three Peruvian species have a Cor- dilleran distribution, ranging from Texas and Mexico to Bolivia and Argentina. They are especially common in the Altiplano and adjacent valleys, in dry, rocky hillsides or among rocks of old walls. KEY TO SPECIES the petiole and those of the buds bicolorous, tan to dark brown, the central portion often lustrous, sclerotic. .......s:000e00++++ 3. P. ovata 1. Fellaea ternifolia ( Cav.) Link, Fil. Sp. Hort. Bot. Berol. 59. 1841, var. ternifolia. Fic. 86, Map 23. Pteris ternifolia Cav. Descr. 266, 1802. Type: Née, Peru, MA; iso- type: F! Pteris peruviana Poir. in Lam. Encycl. Méth. Bot. 5: 718. 1804. Type: Joseph de Jussieu, Peru, (Herb. Jussieu no. 1334A), P! photo GH. Rhizome moderately stout, elongate, decumbent, multicipital, scales of the rhizome (and those at the base of the petiole) straight or fal- cate, bicolorous with a slender sclerotic stripe usually narrower than nnate, pinnae ternate or entire, » Without pinna-rachises; indusium formed by the FERNS OF PERU 123 This species is characterized by the blackish petioles and ternate form of the pinnae. It has the widest distribution of the New World species of Pellaea, occurring along the Cordillera from Arizona to Argentina and in Hispaniola and the Hawaiian Islands. Some specimens from high altitudes in Bolivia and Argentina have been segregated as Pellaea Weddelliana on the basis of the nearly simple pinnae and almost concolorous scales but they are not sufficiently distinct or consistent in these characters to warrant recog- nition. Southwestern United States to Nicaragua; Hispaniola; Venezuela to Colombia to Argentina ; Hawaiian Islands. In crevices of igneous rocks or in Inca walls, in sun and semi-shade, Cajamarca and Amazonas to Puno and Are- quipa, 1800-4000 m. Selected specimens: CAJAMARCA: cumbre El Gavilan, Cajamarca, Namora, Correll & Smith P888 (GH). AMAZONAS: Carrizal, Stiibel 1025 (B). LIBERTAD: n. of Cachicadan, prov. Santiago de Chuco, Stork & Horton 9997 (F, Uc); Huaranchal, Lépez et al. 2691 (GH). ANCASH: Chiquian, Cerrate 622, 786 (USM), Ferreyra 6273 (USM). HUANUCO: near Panao, Asplund 13679 (US); Mito, Bryan 198 (F). LIMA: cerca Tupe, Cerrate 1089 (USM) ; Matucana, Macbride & Featherstone 286 (F). JUNIN: Tarma, Killip & Smith 21817 (Ny, US); Huacapistana, Tryon & Tryon 5424 (BM, F, MO, U, US, USM). HUANCAVELICA: Conaica, Tovar 976 (GH, : km. 45, Nasca to Puquio, Correll & Smith P171A 2. Pellaea sagittata (Cav.) Link, Fil. Sp. Hort. Bot. Berol. 60. 1841, var. sagittata. Fic. 87. Pteris sagittata Cav. Descr. Pl. 267. 1802. Type: Née, Cerro de Guadeloupé, Mexico, MA. Rhizome moderately stout, compact, decumbent, multicipital, straight or nearly so, concolorous, tan to rust-colored, elongate, late-triangular, usually cordate, the margins irregularly dentate, the apex filiform, more or less tortuous; leaves 16-78 cm. long, erect, stiff, Seales of the base of the petiole and the buds straight, concolorous, scales anceo- 124 ROLLA TRYON Fics. 86-88. Fig. 36. Pellaea ternifolia var, ternifolia: 4%, Colombia, Killip & Smith 18751, GH; als. 87. P. sagittata var. sagittata: rhizome and attached leaf, X 1h, weap » lamina, X 14, Peru, Vargas 1051, GH; B, A, rhizome and attached leaf, B, fertile ultimate ent, enlarged, idem. Fig. Skutch 805, cu, Fig. 88. P. ovata: A \4, idem. FERNS OF PERU 125 tawny, broadly ovate-lanceolate, cordate or pseudopeltate, the margin erose, the apex filiform, tortuous, petiole and rachis convex or plane pinnae entire or divided into 3-18 segments, long-stalked; indusium formed by the enrolled, partly membranaceous segment margin, spores with sparse, prominent rugae, 32 per sporangium. This species is distinguished by the light colored petioles, the sagittate form of the segments and straight rachises. All of the South American specimens have 32 spores per sporangium, in which they differ from the cordate segmented var. cordata (Mexico to Texas) which has 64 smaller spores per sporangium. The South American plants seem to be apogamous derivatives of var. cordata. Central Mexico to Guatemala ; Colombia to Bolivia. On dry banks, in open sun or shade, among limestone rocks or stone fences, south-central Peru, Pasco to Cuzco, 1700-3000 m. Selected specimens: PASCO: entre Saleahupan y Cerro de Pasco, Ferreyra 6621 (GH, USM). LIMA: Matucana, Macbride & Featherstone 422 (F, GH). JUNIN: Huacapistana, Tryon & Tryon 5431 (BM, F, GH, MO, U, US, USM) ; Carpapata, Cerrate 2798 (GH, USM). HUANCAVELICA: Conaica, Tovar 978 (USM); entre Coleabamba y Paucarbamba, Tovar 2072 (USM). APURIMAC: Upper Rio Marino, prov. Abancay, Stork et 3. Pellaea ovata (Desv.) Weath. Contrib. Gray Herb. 114: 34. 1936. Fic. 88, Map 24. Pteris ovata Desv. Mém. Soc. Linn. Paris 6: 301. 1827. Type: Peru, Herb. Desvaux, P! photo GH, UC. strongly flexuous, straw to ruddy brown becoming gray with age, lamina elongate triangular, bi- to quadripinnate, usually tripinnate, the pinnae descending from the rachis, pinnae divided into 5-60 seg- ments, long stalked; indusium formed by the enrolled, somewhat mem- 126 ROLLA TRYON branaceous segment margin, spores with a prominent light brown, rugose exospore, 32 per sporangium. Pellaea ovata is unique among most rock ferns in its slender long-creeping rhizome and usually strongly flexuous rachises with descending pinnae. Specimens from southern Texas and northern Mexico have 64 spores in the sporan- gium while those from central and southern Mexico, Hispan- iola and South America have 32. Southern Texas to Nicaragua; Hispaniola, Jamaica; Venezuela to Colombia to northern Argentina. Subscandent, in open sun, brushy savanna, among lime- stone rocks, on shaded banks or stone fences, Cajamarca to Puno, 1200-2850 m. Selected specimens: CAJAMARCA: 52 km. w. of Cajamarca on road to Chilete, Correll & Smith P839 (GH). LIBERTAD: Retamas, Lépez & Sagdstegui 3608 (GH); Sinsicap, prov. Otuzco, Lépez et al. 2270 (GH). HUANUCO: entre Ambo y Huanuco, Ferreyra 9232 (GH, USM); Muna, Bryan 438 (¥F); Acomayo, Feb. 1940, Ridoutt (GH, USM). JUNIN: 2 km. arriba de Huacapistana, Ferreyra 425 (GH, USM); Huacapistana, Try- on & Tryon 5423 (BM, F, GH, MO, U, US, USM). HUANCAVELICA: La Mejorada, G. Kunkel 452 (GH). APURIMAC: Andahuaylas, Herrera 1497 (F, GH) ; w. of Pincos, prov. Andahuaylas, Stork & Horton 10712 (F, UC). cUzco: San Miguel, Urubamba valley, Cook & Gilbert 1765 (US) ; Yucay, Soukup 754 (F, GH), Vargas 1051 (GH). PUNO: Sandia, Weberbauer 841 (B). 18. Doryopreris J. Sm. Jour. Bot. Hooker, 4: 162. 1841. Type: Pteris palmata Willd. = Doryopteris palmata ( Willd.) J. Sm. = Doryopteris pedata var. palmata (Willd.) Hicken. Terrestrial, the rhizome small to moderately stout, erect to decum- bent or creeping, scaly, bearing the leaves in a crown or loose cluster; leaves small, the lamina pedate, deeply lobed to tripinnatifid, glabrate, veins free or anastomosing; sporangia borne on a continuous or inter- rupted marginal commissure, covered by an indusium which is formed from the reflexed and modified margin, paraphyses absent. — 23 species in America. Tryon, R. A revision of the genus Doryopteris. Contrib. Gray Herb. 143, 1942. Doryopteris is distinguished by its black or atropurpur- eous petioles, pedate laminae and usually long-stalked spor- angia borne in a continuous marginal line covered by an FERNS OF PERU 127 24, P. ovata. Map 25, ea iad dot), D. Lorentzii Maps 23-26. Map 23, Pellaea ternifolia var. ternifolia. Map Doryopteris crenulans (northern dot Si lomariac (southern dots). Map 26, Saffordia Suduh 128 ROLLA TRYON indusium formed from the greatly modified margin. None of the five species in Peru is common; the genus is primarily one of eastern Brazil and diminishes in numbers through Argentina and Bolivia. The region along the eastern side of the Andes in the Department of Cuzco and southeastward is the center in Peru and the area in which species new to the country will most probably be found. KEY TO SPECIES a. Venation free, single areolae present by rare exception. ........00+ b. tion. b. Petiole suleate to wing-angled on the upper side, at least toward the base of the lamina, with one vascular bundle. . 5. D. concolor b. Petiole terete or subterete, with two vascular bundles at least toward the base. c. ce. Fertile lamina usually with numerous ultimate segments, the pinnae and primary segments surcurrent and, except ‘the basal pair, also decurrent, the bases forming wings along the rachis that are broad at their top and bottom and narrowed in their middle. ..... 1. D. erenulans at the base of the lamina. ..... 3. D. lorentzt d. Fertile and sterile laminae with completely areolate venation, pro- liferous buds present at the base of the lamina. .......... 4, D. pedata 1. Doryopteris crenulans (Fée) Christ, Schwacke, Pl. Nov. Mineiras 2: 26.1900. Fic. 89, 25. Pellaea crenulans Fée, Crypt. Vasc. Brésil 2: 27, 't. 87, f. 3, 1872-73. Type: Tijuea, Brazil, Glaziou 5345, fragment “ex Fée” Ny!; isotype: P, photo GH, cu! and Sterile leaves quite dimorphic, coriaceous, ‘the sterile to 35 cm., the fertile to 50 cm. tall, ultimate seements numerous, with free venation, the pinnae and primary segments surcurrent, and, except the basal broad at the top and bottom and narrowed in the middle; soral lines broken by all or nearly all of the sinuses. FERNS OF PERU 129 Doryopteris crenulans is distinguished by its free vena- tion, essentially terete petiole with two vascular bundles, the rather strongly dimorphic fertile and sterile laminae, the pinna or primary segment bases that form wings along the rachis that are narrowed in the middle and the soral lines that are broken at most of the sinuses. 90c 89-90. pis nip detsidden crenulans: A, fertile lamina, x 14, Bolivia, Buch- x 36, a: Fics Fig 3 tien 7088, US: B: sterile lam 4, Brazil, Reiss ee es i ry ph dons Jamina, ae orm kk B. Smith 1856, GH; B, sterile lamin na, x poe Ripe Kunert 26, us; C, cei. and recepta ‘ ent, wy, eRe: bundles near base of baitale: enlarged, t 130 ROLLA TRYON Peru and Bolivia; southern Brazil. On hillside, in humus and clay soil, Cajamarca, 3200 m. Specimens seen: CAJAMARCA: Socota, Stork & Horton 10127 (F, UC, 5 2. Doryopteris lomariacea K]. Linnaea 20: 343. 1847. Type: British Guiana, Schomburgk 1197; isotypes: BM! K! Fic. 90, MAP 25. Rhizome creeping, its scales and those of the leaf-buds long and narrow, ‘the cells of the hyaline borders (or of the wholly hyaline scales) mostly three or more times as long as broad; petiole usually smooth, with two vascular bundles; fertile and sterile leaves coriaceous, strongly dimorphic, the sterile to 50 em., the fertile to 120 cm. tall, ultimate segments 5 to several, less often numerous, with free venation, the pinnae and primary segments surcurrent and, except the basal pair, also decurrent, forming wings along ‘the rachis that have essen- tially parallel sides; soral lines continuous around the sinuses. ; This species is similar to the preceding in its free vena- tion, the terete petiole with two vascular bundles and the strongly dimorphic fertile and sterile laminae, but differs in the wings along the rachis formed by the bases of the pinnae and primary segments having essentially parallel sides and the soral lines continuous around the sinuses. Southern Brazil and Paraguay; Peru; British Guiana. Locally wet places, Junin. Specimens seen: JUNIN: Enefias, Killip & Smith 25709 (Ny, US). 3. Doryopteris Lorentzii (Hieron.) Diels, Engl. & Prantl, Nat. Pf. 1(4): 270.1899. Fie. 91, Map 25. Pellaea Lorentzii Hieron. Bot. Jahrb. 22: 392. 1896. Lectotype: Cordoba, Argentina, Lorentz 1 9, B! photo GH. Sterile leaves moderately dimorphic, moderately coriaceous, the sterile to 18 em., the fertile ‘to 35 cm. tall, ultimate segments numerous, with areolae along the midnerves, the veins free toward the margin, remotely pas » the sinuses black and semi-sclerotic; soral lines broken by the Doryopteris Lorentzii is distinguished by its partially neces venation, the areolae being commonly confined to € area along the midrib. The petiole is terete or rarely FERNS OF PERU 131 suleate toward the apex and has one vascular bundle. The sinuses are usually black and semi-sclerotic. Peru to Argentina and southern Brazil. Rocky places, Cuzco, 1200-2500 m. Specimens seen: CUZCO: Potrero, Coronado 115 (GH, UC, US), Tryon & Tryon 5377 (BM, F, MO, U, US, USM); Yucay, Soukup 921 (F); Dist. Santa Ana, Herrera 871a (US); Santa Ana, Cook & Gilbert 1479 (Us, in part); Valle de Lares, Herrera 1635 (US); Machu-Picchu to Quil- labamba, Mexia 8053a (F, UC, in part). 4. Doryopteris pedata (L.) Fée, var. palmata (Willd.) Hicken, Rev. Mus. Plata 15: 253.1908. Fic. 92. Pteris palmata Willd. Sp. Pl. 5: 357. 1810. Type: Caracas, Venezuela, Bredemeyer, (Herb. Willd. 19957), B! photo GH! Ny! Us! Rhizome scales and those of the leaf buds usually ovate-lanceolate, the cells of the hyaline borders about as long as broad; petiole usually pubescent, with one vascular bundle; fertile and sterile leaves moder- ately dimorphic, moderately coriaceous, the sterile to 37 cm., the fertile to 40 em. tall, both with proliferous buds at the base of the lamina, ultimate segments several to many, with areolate venation; soral lines usually broken by the sinuses. This variety is easily identified by the proliferous buds that are borne at the base of the lamina, one on each side of the petiole apex. The buds develop as the leaf ages and old leaves that have come to rest on the ground have young plants developed from them. These continue to grow and are an effective means of reproducing the species locally. Rarely some leaves of a plant lack buds, and in young leaves they are rather inconspicuous. The petiole is usually plane on the upper side, at least in part, and the sinuses are only occasionally blackish and are cartilaginous, not sclerotic. The closely related var. multipartita (Fée) Tryon is known from Bolivia; it may be distinguished by the lack of proliferous buds, the more divided lamina and the spores which lack perispore; in var. palmata a perispore is present. The completely areolate venation separates var. multipartita from D. Lorentzii. Doryopteris nobilis also is a marginal species distinguished by large laminae, usually with buds at the base of the lamina as in D. pedata var. palmata but with the sterile tips of the segments crenate or crenulate with 132 ROLLA TRYON QIA / / gic 93Cc 93A 93B ‘ Figs. 91-93. Fig. 91. Doryopteris Lorentzii: A, fertile lamina, X tien 3397, US; B, sterile lamina, « 4, Argentina, Maldonado 402, sterile segment, « 5 rgentina, Vattuone & Bianchi 165, us. Fig. 92. D pa : oa 1%, Bolivia, Buch- GH; C, venation of : 84, US. Fig, 93. D. concolor var. concolor: Xx ¥, Brazil, Wacket, us; B, sterile lamina, X 1%, Peru, Vargas 1709, GH; C, v and receptacles of fertile segment, x 10, El Salvador, Calderén 1714, US. FERNS OF PERU 133 ascending teeth; in D. Lorentzii and D. pedata var. palmata the sterile tips are entire or have spreading teeth. Venezuela to Colombia; south to Bolivia and north to Mexico. Rocky woods, Cajamarca and Cuzco, 1200-2200 m. Specimens seen: CAJAMARCA: prov. Hualgayoc, Soukup 3814 (us). CUZCO: Potrero, Coronado 114 (GH, UC, Us), Tryon & Tryon 5369 (BM, F, GH, MO, U, US, USM), Vargas 1705 (GH); Torontoy, Herrera 1304 (Us); Yucay, Soukup 920 (F); Colpani, Cook & Gilbert 1062 (GH). 5. Doryopteris ne (Langsd. & Fisch.) Kuhn, v. d. Decken, Reisen Ost-Afrika 3(3): 1879. var. concolor. FIG. 93. Pteris concolor nue & Fisch. Ic. Fil. 19, t. 21. 1810. Type: “Archi- pelagi ape aeons insula Nucahiva”; isotype: (Herb. Willd. 19961-1), B! photo Rhizome scales and those of the leaf buds usually ovate-lanceolate, vascular bundle; fertile and sterile leaves similar, herbaceous, the sterile to 15 cm. the fertile to 35 cm. tall, ultimate segments very numerous, with free venation; soral lines usually broken by the sinuses he American Doryopteris concolor all belongs to var. concolor, the other variety, var. Kirkii, being in Africa, India and Ceylon. The venation is free which distinguishes it from the two preceding species and the single vascular bundle in the petiole will distinguish it from the first two, as does the sulcate or wing-angled petiole. The texture of the lamina is herbaceous while in all of the other species it is coriaceous. Tropical America. Rocky places and clay banks, Cajamarca to Cuzco, 750- 2000 m. Selected specimens: CAJAMARCA: Jaén, Feb. 17, 1954, R. mackie UC, US). SAN MARTIN: Tarapoto, L. Williams 5451 (F, GH), Si 4045 (K). cuzco: Valley of the Vilcanota, Mexia 8053 Saget US); Santa Ana, Cook & Gilbert 1512, 1553 (US); Fuirers, ay. 113 (GH, UC, US), Tryon & Tryon 5375 (BM, F, GH, MO, U, UC, US 9. SAFFORDIA aig Smiths. Misc. Coll. 61(4): 1. 1913. Type: Saf- f ee induta Max 134 ROLLA TRYON Terrestrial, the rhizome rather small, erect, scaly, bearing the leaves in a rosette or cluster; leaves small, the lamina pedate, bipinnatifid, densely scaly beneath, glabrous above, veins anastomosing; sporangia borne along and between the veins in a narrow band close 'to the un- modified margin, indusium and paraphyses absent.— A Peruvian genus of 1 species. Ballard, F. Saffordia induta, Hook. Ic. Pl. t. 3599 & text. 1962. The genus was long known only from the original collec- tion made in Lima by Safford in 1892; it was recollected at about the same place in 1954 and by 1959 its range had been extended to La Libertad and Cajamarca. Saffordia induta Maxon, Smiths. Misc, Coll. 61(4): 2, t. 1-2. 1918. Type: Along the Arroya (La Oroya) Railway, in the mountains back of Lima (probably near Matucana), Peru, Safford 989, Us!; isotype: B!GH! K! uc! Fic. 95, MAP 26. Rhizome scales subulate to lance-subulate, light brown to brown, con- colorous, the margins toothed; leaves ca. 15-30 cm. long, petiole decidu- ously scaly, reddish-brown to nearly atropurpureous, longer than the blade, with one vascular bundle, the scales at its base lanceolate to ovate-lanceolate, reddish-brown or pinkish, the margins toothed to sub- fimbriate; blade more or less suborbicular, the scales (beneath) densely Bele od rat deltoid-ovate to long-deltoid, prominently fimbri- ate. The pink color present in the scales at the base of the petiole is a unique one among ferns. This color was espe cially evident among the many living plants at Kew (1960) which had been grown from spores sent by Mr. S. G. E. Saunders, Young plants among these cultures had the small- est leaves nearly orbicular, larger ones were oblong-cordate to long deltoid-sagittate, and the largest were 3-lobed. Specimens seen: CAJAMARCA: between Santiago and Guzmango, prov. Contumaza, Sagdstegui 2936 (GH); between Cascas and Contumaza, Lépez et al. 3683 (GH). LIBERTAD: Huaranchal, prov. Otuzco, Sagds- FERNS OF PERU 135 tegui 205 (GH). LIMA: Mountains back of Lima, Safford 989 (B, GH, K, Uc, US); km. 68, Carretera Central, dist. Surco, Saunders 218 (BM); Matucana, Rauh P186 (B). 20. ADIANTUM L. Sp. Pl. 2: 1094. 1753; Gen. Pl. 485. 1754. Type: Adiantum Capillus-Veneris L. FIG. 94. Terrestrial, the rhizome small and suberect, stout and short-creeping, or slender and long-creeping, scaly, bearing the leaves in a cluster or well spaced; leaves small to very large, pinnate (1-pinnate to 6-pin- nate) or helicoid (in A. patens), usually glabrous, less often sparingly scaly, pubescent or ceraceous beneath, veins free; sorus marginal, not paraphysate, the sporangia borne along (or also between) the vein-tips which extend into the indusium which is formed from a reflexed, modi- fied lobe. — About 75 species in South America. Kramer, K. U. A contribution to the fern flora of French Guiana, (Adiantum). Acta Bot. Neerland. 3: 481-486. 1954. Fic. 94. Adiantum Capillus-Veneris, sea cliffs at Miraflores (Lima). 136 ROLLA TRYON Adiantum is one of the best defined genera of ferns, the sporangia-bearing veins that enter the marginal indusium being a distinctive feature. Also, the aspect of the leaves is characteristic, although variable and difficult to define, but once a few species of the genus are known others can readily be placed there. The genus is widely distributed in Peru and is one of the common ones both in number of species and individuals. Most of the species fall into a few rather well marked geographic groups. One group, species 10-24, is character- istic of the Amazon Basin and the lower elevations of the eastern slopes of the Andes. Another, species 1-8 and 25, is characteristic of the eastern slopes of the Andes and their valleys. A third, species 33-35, of the valleys and sheltered places of the Altiplano and the higher elevations of the eastern and western Andean valleys. A fourth, species 31 and 38, occurs primarily on the costal lomas, although also present in the Altiplano. Finally, a single species, 30. A. Capillus-Veneris, is probably an escape from cultivation; with a single exception it grows near Lima. KEY TO SPECIES eva ee ee ae ‘ 5. A. macrophyllum d. Rachis pubescent-scaly or scaly, pinnae alternate, rarely sub- eg tna Rage "PRR AS So A ls : e. Rhizome scales dull, rachis pubescent-scaly or scaly on . Sides, rarely glabrate in A. Poeppigianum. sasecenecsnrerssensanecens Be Rachis densely scaly, the scales freely ciliate, pinnae with long brown trichomes beneath, lamina 1-pinnate. .....--+++- Spica gaia 6. A. scalare f. Rachis moderately pubescent-sealy, the scales mostly ciliate only at the expanded base, pinnae glabrous . BS Sch UTC INE 7. A. Poeppigianum FERNS OF PERU 137 g. Dark color of the segment-stalk not or hardly entering the base of the segment, midvein indistinct or lacking 8. ALh in the apical half. ucidum . Rhizome scales iridescent, rachis pubescent-scaly on the upper side, glabrous below, lamina 1- to bipinnate. .............. 9, A. alarconianum ce. Several oblong or arcuate sori on each edge of the pinna or pinnules, or only on the upper edge, lamina 1- to bipinnate. .... h. h. Segments dull, patron glaucous, beneath, the sterile ones or decent evenly se i . Rhizome shortereering usually nodose, with the scales appressed, the petiole es approximate; lamina bipinnate, segments with saudi numerous simple trichomes beneath. 23. A. humile i. Rhizome short- to long-creeping, with the scales spreading, the petiole bases usually well spaced; lamina usually 1- pinnate, segments glabrous or nearly so beneath. .........-.00 10. A. petiolatum h. Segments ier usually shining, beneath, the sterile ones or portions unevenly biserrate; rhizome short-creeping, with appressed eee the petiole bases approximate. .......cceeeerees 11. A. obliquum b. Lateral ultimate segments asymmetrical in their apical half, or in A. latifolium and A. villosissimum, if symmetrical, then the lamina laterals and usually narrowed at the base; lamina bipinnate, at least above the basal pinnae, or 3- to 4-pinnate in A. pectinatum. j. j. Rachis and ee glabrous or puberulent or pubescent with simple trichom k. k. Pinna-rachises berate or pubescent only on the upper side, glabrous below 20. A. tomentosum k. Pinna-rachises ade or puberulent or pubescent on all sides. L l. Pinnules nearest the rachis definitely asymmetrical. ........ m. m. Pinnules spaced, acute, pinna-rachises puberulent. ........ 15. A. ae m. Pinnules closely imbricate, broadly obtuse, pinna- rachises glabrous. ...... . 21. A. macrocladum l. Pinnules nearest the rachis, or at the base of the penulti- n overlaying the rachis. -.-.. 0. o. Pinna-rachises whitish pubescent. .. sip A. sessilifolium o. Pinna-rachises puberulent. .......... 27. A. Henslovianum 138 ROLLA TRYON n. Lamina with a divided rachis and no terminal pinna, or with a single rachis and a distinct terminal pinna that is somewhat reduced at the base and with the lateral pinnae with their inner pinnules hardly overlaying tthe rachis. .... 29. A. patens j. Rachis and pinna-rachises scaly or rufous pubescent-scaly. .... D. p. Segments bearing usually numerous (sometimes few) simple trichomes beneath, in no. 23, these sometimes mixed with scales. q- q. Pinna-rachises scaly with long, freely ciliate scales, one to three (rarely four) long sori on the upper edge of the fertile segments, segments acute. 16. A. villosissimum q. Pinna-rachises pubescent-scaly, the scales sparsely if at all ciliate, five to usually 8-10 sori on ‘the upper edge of the fertile segments, segments glaucous, obtuse. sie r. Pinnules dull green to faintly glaucous beneath, those near- est the terminal segment of the pinna reduced, less than half as long as the longest pinnule (rarely more), usually 8-10 sori on the upper edge of the pinnule. ee wvee sl hesasiiiie 22. A. terminatum r. Pinnules glaucous beneath, those nearest the terminal seg- ment of the pinna not much reduced, more than half as long as the longest pinnule, usually 5-6 sori on the upper edge of the pinnule. .................. 23. A. humile p. Segments glabrous beneath or bearing usually few, sometimes stellate, scales. . s. Lamina commonly 4- to 5-pinnate at the base, segments deep- ly incised-lobed. Gcansibhoswhheiace 25. A. pectinatum 8. Lamina bipinnate, segments entire to coarsely serrate. ...... t Pe eee eres eeesesareeees usually ve greatly reduced, about half as long as the long- POCO eee rreneeseeeseseessce a hd u. Pinnules mostly herbaceous, glaucous, ‘the indusia-bear- ing Margins flat, terminal pinna reduced at the base, sori usually on the lower as well as the upper edge of the pinnules. susie snsneeda cued. . 12. A. latifolium u. Pinnules coriaceous, not glaucous, the indusia-bearing margins revolute, terminal pinna not or hardly reduced at the base, lower edge of pinnules without sori. .......- cies “A ctanrerretbemerrpentovenisiebitdesades eu veischdboneli 12. A. serratodentatum t. . - Segments or portions coarsely and often unevenly trate, terminal segments of pinnae usually sterile, acute SO PEGMMNOES cocci 8 +> v. Fertile pinnules with about 10-15 small sori on shallow lobes, none on the lower edge, some pinnules with 1-2 sori nl Ee Maher elie ee 24, A. cayennense FERNS OF PERU 139 v. Fertile pinnules usually with fewer than 10 large, arcu- sori on the hardly incised margin, sometimes some on = lower edge, none on the inner edge. w. . Pinnules nearest the terminal segment of the pinna ier greatly reduced, about half as long as the longest pinnule, segments green beneath, indusium-bearing —— flat. 14. A. villosum . Pinnules nearest the terminal segment of the pinna oir ays reduced, less than half as long as the soli pinnules. x. Several sori on the upper and outer edges of is fertile pinnules. y. Rhizome long-creeping, the petiole bases distant, pinnules with sterile tips mostly turned toward the apex of the pinna and acute. 17. A. tetraphyllum y. Rhizome usually short-creeping, often knotted, the petiole bases adjacent, pinnules with sterile tips mostly straight and obtuse. ........ 18. A. fructuosum x. One, rarely two, sori on the upper edge of the fertile pinnules, rarely another on the peuged GEG, icicssiotinsacen A. pulverulentum a. Ultimate ee definitely, although ceomeet shortly, ey or most of Z. Rachis and ae axes of the lamina glabrous, or in A. Pere rarely with some ceraceous indumen a. Segments not articulated, the stalk passing = into ae ed veins, deciduous by irregular fracture of the stalk. ...... bb. bb. Sori borne on both sides of the a cues 0 ins ute to acuminate segments. 4. yphyllum. bb. Sori borne on the convex so edge of ‘the Erah en to flabel- late-orbicular, rounded, segmen ce. ec. Sterile margins of the apart with each vein ending in a tooth, segments cuneate, with a pronounced cartilaginous border. 30. A. Capillus-Veneris cc. Sterile margins of the segments with each vein ending in a sinus between the definite or indefinite teeth. ................ dd. dd. Pinnae subsessile, the lower as well as the upper agin the inner upper pinnules overlaying the rachis. ............ ee. Inner upper pinnule of lower pinnae simple. .......-........ 31. A. subvolubile ee. Inner upper pinnule of lower pinnae divided into two ultimate segments. 32. A. concinnum dd. Pinnae, or at least the basal or lower ones, definitely stalked, the inner upper pinnule on them not or hardly overlaying the rachis ff. ff. Fertile lamina “ rarely bi-) 3- to 4-pinnate, triangular to ovate-lanceola gg. 140 ROLLA TRYON gg. Many ultimate segments symmetrically or asym- metrically cuneate, sori orbicular to suborbicular. .... A. Raddianum gg. Many ultimate segments orbicular to suborbicular, sori oblong or lunate, or many so. ........ 34, A. Poiretu ff. Fertile lamina 1- to bipinnate, oblong to oblong-taper- ing, sori orbicular, usually closely spaced. ......s.ssss0 36. A. Ruizianum aa. Segments articulated at the junction with the stalk, deciduous by a sharp, clean break, the apical seis of the stalk enlarged (in the mature lamina or portions o hh. hh. Ultimate segments oo nae to cuneate, sori borne on the convex outer e ii. i, Lamina bi- to isi sterile margins of the segments with each vein ending in a sinus between the usually definite teeth, rachis with segments to the apex, oe TOOTING. «...+000000 35. . Orbignyanum i. Lamina 1-pinnate, sterile margins of the nat sharply serrate with each vein ending in a tooth, some leaves with a long, barren, usually rooting, rachis t 39. A. deflectens hh. Ultimate segments trapeziform to ovate-acuminate, sori borne on the upper and outer edges ode jj. Segments mostly trapeziform, bluntly acute or eben kk. SNe short-stalked, terminal segments of the pinnae acute to acuminate. ........66.0... 1. A. Mathewsianum kk. pees Aelia terminal segments of the pinnae bluntly ac 2. A. peruvianum jj. Segments Ss ovate-cuneate, occasionally with an acuminate auricle, concavely acuminate. 3. A. anceps z. Rachis and other axes of the lamina minutely puberulent (on the upper side) or short-pubescent (rar ely glabrate), segments often TOCA Mi NaN go ses sicccinn cs tbo esccuinncs be eelacs aca I. ll. Rachis and other axes of the lamina minutely puberulent only on the upper side, segments glabrous beneath. ...... 28. A. lobatum ll. Rachis and other axes of the lamina short-pubescent all over, “es rarely glabrate, segments pubescent beneath, to ip ce ra eee eeereee mm. Ultimate segments not or shallowly and ae “cet lamina bipinnate at the base, 1-pinnate above, segments & brous above, sori orbicular to suborbicular. .. 37. A. sae mm. Ultimate segments deeply cleft into ca. 4-7 spreading lobes, lamina bi- to tripinnate, sezments pubescent above as well as beneath, rarely they and the axes of the lamina glabrate, sterile margins of the segments with each vein ending in 4 tooth, sori straight or nearly so 38. A. digitatum 9 eee enwesseseenaerese & FERNS OF PERU 141 1. Adiantum Mathewsianum Hook. Sp. Fil. 2: 35. 1851. Type: Chacha- poyas, Peru, Mathews 3296, K! photo Us, fragment s.m. ex K, NY! FIG. 96. Rhizome not seen; leaves evidently large, perhaps to 1 m. or more, petiole glabrous, ebeneous or slightly glaucous, rachis similar; lamina probably broadly ovate-triangular, to quadripinnate, pinnae widely alternate, the basal very large, ultimate segments trapeziform, bluntly acute (the terminal one acute ‘to acuminate) glabrous, short-stalked, articulate, the dark color of the stalk not entering the segment, of moderate size, ca. 2-2.5 em. long, the inner edge often overlying the axis; sori commonly borne on all but the basal edge, many, roundish to short-oblong or lunate. A glabrous species with trapeziform and blunt segments. The segment stalks are definite although rather short. In the shape of the segments and length of their stalks it differs from the related A. trapeziforme which has longer and slender segment-stalks and the segments themselves are concavely acute or acuminate. This latter species was erro- neously reported from Peru by Hieronymus (Hedwigia 48: 237. 1909) on the basis of Stiibel 1084. Peru to Paraguay. Amazonas to Junin, 1200 m. Specimens seen: AMAZONAS: Chachapoyas, Mathews (NY), 3296 (K). SAN MARTIN: Tabalosos, Stiibel 1084 (B). JUNIN: La Merced, Chancha- mayo, Soukup 1116 (F); Chanchamayo, 1918, Esposto (USM). 2, Adiantum peruvianum KI. Linnaea 18: 555. 1845. Syntypes: Vitoc, Peru, Ruiz 25, B! Chachavani, Ruiz 27, B! fragments ex B, us! Fie. 97, MaP 27. Rhizome large and massive, very short-creeping; leaves to 1 m. tall, petiole ebeneous, glabrous, shining or slightly glaucous, rachis the same; lamina broadly ovate-triangular, to tripinnate ( small ones elongate-triangular or oblong, once to bipinnate), pinnae widely alter- nate, the basal largest, ultimate segments ovate-trapeziform, bluntly acute or rounded (the terminal bluntly acute), glabrous, long-stalked, articulate, the dark color of the stalk not entering the segment, large, commonly ca. 5 em. long; sori borne on the upper and outer edges, few if any on the inner edge, many, roundish to short-oblong or lunate. The bluntly acute segments distinguish this species from the related A. anceps and A. platyphyllum. The long seg- ment stalks and the larger segments distinguish it from the previous species A. Mathewsianum. The fertile leaves may be 1-pinnate, or in larger specimens bi- or tripinnate. 142 ROLLA TRYON ent PN eS ies Mis ai fa Pane Wy 3 i ae el ae ‘ = > eA PIV a |, GH; B, Fics. 95-97. Fig. 95. Saffordia induta: A, leaf, X 14, Peru, Sagdstegui 29: Hehe ee mnent.. lower Eps Mi, iden, Fig. 96. Adiantum Mathewelanim: Z ‘iy ; ., a , » Soukup 1116, ry, Fig. 97. A. peruvianum: fertile pinna %, Peru, Soukup rise pi FERNS OF PERU 143 Ecuador to Bolivia. In rich humus in deep woods or on canyon sides, Loreto to Cuzco, 750-2800 m Selected specimens: LORETO: Boquerén del Padre Abad, Aug. 10, 1943, Ridoutt (USM). HUANUCO: Cayumba, Ferreyra 1873 (GH, USM); Puente Durand, north of Huanuco, valley of Rio Chinchao, Stork & Horton 9448 (F, GH, UC); gorge of Rio Chinchao, Tryon & Tryon 5813 (BM, F, MO, U, US, USM) ; Tingo Maria, Tryon & Tryon 5236 (BM, F, MO, U, USM). JUNIN: La Merced, Soukup 1115 (F), 2568 (US); Puente San Felix, Cerrate 2869 pv —— 10 km. sw. of San Ramon, Tryon & Tryon 5450 (BM, F, MO, USM). CUZCO: Santa Rosa, Urubamba val- ley, Cook & Gilbert aa (us); Quillabamba, Biies 1236 (Us). 3. Adiantum anceps Maxon & Morton, Amer. Fern Jour. 24: 15, 1934. Type: La Merced, Junin, Peru, Killip & Smith 24069, Us!; isotype: NY! Fic. 98, Map 28. Adiantum Crespianum Bosco, Nuovo Giorn. Bot. Ital. ns. big 148, t.x, fig. 2. 1988. Type: Indanza, Ecuador, C. Crespi; isotype: U Rhizome large and massive, short- or very short-creeping; leaves to 1 m. tall, petiole ebeneous, glabrous, shining or slightly glaucous, rachis the same; lamina broadly ovate-triangular, to tripinnate (small ones elongate-triangular or oblong, 1- or bipinnate), pinnae widely alter- nate, the basal largest, ultimate segments ovate-cuneate, concavely acuminate, glabrous, long-stalked, articulate, the dark color of the stalk not entering the segment, large, commonly ca. 6 cm. long; sori borne on the upper and outer edges (the inner and lower not well defined), numerous, roundish to mostly short-oblong to oblong or lunate. This species is rather closely related to A. peruvianum from which it was segregated. It differs especially in the concavely acuminate rather than bluntly acute segments. A number of specimens have some segments with an acumi- nate auricle. The related A. subcordatum of Brazil and British Guiana has the rhizome creeping and stout with the petiole bases spaced, the scales are secund, falcate, rigid and entire, the petiole and rachis are reddish-brown and the segments are glaucous beneath, and acute to acuminate with straight sides. Adiantum anceps, on the other hand, has a massive rhizome with the petiole bases adjacent; the scales are toothed or ciliate, not rigid and more or less straight; the petiole and rachis are atropurpureous to nearly black; and the segments are dull to green beneath and concavely acuminate. Ecuador and Peru. 144 ROLLA TRYON In forests, San Martin to Cuzco, 230-1500 m. Selected specimens: SAN MARTIN: Lamas, L. Williams 6445 (F, GH, NY, US); Juan Jui, Alto Rio Huallaga, Klug 4255 (BM, F, GH, MO, NY, UC, US, USM); Pongo de Cainarachi, Rio Cainarachi, Klug 2688 (BM, F, GH, MO, NY, US). HUANUCO: Tingo Maria, Tryon & Tryon 5288 (BM, F, GH, MO, NY, U, UC, US, USM). JUNIN: Monte Rico, La Merced, Soukup 3507 (GH, MO, UC); La Merced, Killip & Smith 23478 (F, GH, NY, US), Cerrate 284 (GH, USM). CUZCO: Quellouno, prov. Convencién, Vargas 13556 (GH). 4. Adiantum platyphyllum Sw. Svenska Vet. Akad. Handl. 1817: 74, t. 3, fig. 6. 1817. Type: Villa Rica, Minas Geraes, Brazil, Freyreis (Herb. Sw.) s-pa! fragment us! Fic. 99. Rhizome moderately stout, short-creeping, the petiole bases crowded; leaves up to ca. 0.7 m. tall, petiole ebeneous or nearly so, glabrous, Shining or somewhat glaucous, rachis the same; lamina broadly tri- The long-stalked, glaucous and long-acuminate segments combine to make this a distinctive species. It is easily Separated from any of the preceding three species by the character of the Segment-stalk. In A. platyphyllum the dark color enters the base of the segment; in the other species the stalk is articulated at its apex and the dark color abruptly ends there. Ecuador to Bolivia and Brazil. In forests, San Martin to Junin, 700, to usually ca. 1000, to 1700 m. Selected specimens: san MARTIN: San Roque, L. Williams 7446 (F; Us); Zepelacio, near Moyobamba, Klug 3490 (F, GH, MO, NY, US) HUANUCO: Pozuzo, Macbride 4570 (F, US). JUNIN: Chanchamayo Pye iagsee 4 (F, US), 100 (US), 152 (F), 445, 1366, 1368 (¥F, US), : 5. Adiantum macrophyllum Sw. Nov. Gen. Sp. Prod. 135. 1788. Type: Smaies, Swartz, (Herb. Sw.), 9-Pal Fic. 100, Mar 29. gen moderately stout, short to moderately creeping; leaves sag y 80-60 cm. tall, petiole ebeneous or atropurpureous, glabrous, ining or slightly glaucous, rachis the same; lamina oblong-acute or FERNS OF PERU 145 Maps 27-30. Map 27, rennin peruvianum. Map 28, A. anceps. Map 29, A. macro- phyllum. Map 30, A. obliq 146 ROLLA TRYON elongate-triangular, 1-pinnate, pinnae essentially opposite (rarely some alternate), entire, long-triangular, to ovate-triangular, broadly cune- ate, occasionally auriculate or biauriculate, acute to acuminate, gla- brous, glaucous or dull beneath, subsessile, not articulate, the dark color of the short stalk entering the base of the pinna, large, commonly 6-9 cm. long; one long sorus on each side of the pinna, rarely two or three on some pinnae. Several characters combine to make this one of the most distinctive species in the genus. The lamina is 1-pinnate with opposite (rarely sub-opposite), glabrous, glaucous and broad pinnae. Typically there is a single long sorus on each side of the pinna. Guianas to Colombia to Bolivia and Brazil; Mexico; West Indies. It grows in woods, in dense forests, on shady banks and in rocky places with soil, San Martin and Loreto to Madre de Dios and Puno, 200-2500 m. Selected specimens: SAN MARTIN: Tarapoto, Woytowski 35247 (MO, Tingo Maria, Soukwp (us), 3090 (F, MO); Tingo Maria, Ferreyra 10308 (GH, USM). JUNIN: Pichis trail, between San Nicolés and Azu- pizi, Killip & Smith 26112 (Fr, NY, Us); La Merced, Soukup 1111 (F). AYACUCHO: Estrella, between Huanta and Rio Apurimac, Killip & Smith 23056 (F, GH, NY, US). CUZCO: Potrero, Vargas 1704 (GH, M0), 8225 (MO, UC), Tryon & Tryon 5388 (BM, F, MO, U, US, USM). MAD! DE DIOS: Alto Madre de Dios, Rauh P1667 (B). PUNO: Chunchosmayo, prov. Sandia, Weberbauer 1267 (B). 6. Adiantum scalare Tryon, Amer. Fern Jour. 47: 141, t. xv. 1957. Type: Rio Santiago, Loreto, Peru, Mexia 6162, UC!; isotypes: BM! Gu! K! MO! PH! Fic. 101, Rhizome moderately stout, short-creeping, scales dull; leaves ca. 60- ° 5 cm. tall, petiole ebeneous, deciduously scaly, rachis similar but per- sistently scaly; lamina ovate or oblong-ovate, 1-pinnate, pinnae closely alternate, mostly 8-13 em. long, more or less long-triangular, entire, often auriculate, with long brown trichomes beneath; one very long sorus on each side of the pinna, This distinctive species is 1-pinnate and has the rachis scaly and the under surface of the pinnae pubescent. Other 1-pinnate species, or those occasionally 1-pinnate, have the rachis glabrous or pubescent-scaly and the segments gla- FERNS OF PERU 147 brous. The long, narrowly triangular pinnae usually have a small auricle on the upper edge at the base and sometimes a smaller one on the lower edge near the base. Endemic to Peru. Fr i : fertile ultimate segment, X 1%, ss age ch Al lo se e 1%, Peru, Killip & Smith 23056, ver 6, Peru, Killip & Smith 28995, us. Fig. 103. 4. lucidum: fertile pinna, X %, Trinidad, Broadway 4310, MO. Fig. 104. A. alarconianum: fertile pinnae, X 1, Ecuador, Barclay, us. Fig. 105. A. petiolatum: fertile pinna, x i. eru, Schunke 377, cH. Fig. 106. A. obliquum: fertile pinna, X 1, British Guiana, A. C. Smith 2432, cH. 148 ROLLA TRYON In dense forest, Amazonas and Junin, at 200 to 340 m. Specimens seen: AMAZONAS: Mouth of Rio Santiago, Tessmann 4276 (B, NY); Rio Santiago, Mexia 6162 (GH, MO, PH, UC). JUNIN: Cahua- panas on Rio Pichis, Killip & Smith 26741 (F, GH, NY). 7. Adiantum Poeppigianum (Kuhn) Hieron. Hedwigia 48: 231. 1909. Fic. 102. Adiantum lucidum var. Poeppigianum Kuhn, Jahrb. Bot. Gart. Ber- lin, 1: 340. 1881. Type: Prov. Maynas, (Loreto), Peru, Poeppig 2268, B!; isotypes: BM! GH! LE! Us! Rhizome rather slender, long-creeping, the petiole bases spaced, scales dull; leaves ca. 30-50 em. tall, petiole ebeneous, deciduously pubescent-scaly, rachis similar; lamina roundish-ovate to ovate-oblong, usually 1-pinnate, sometimes sparingly bipinnate, entire pinnae lanceo- late-cuneate to ovate-lanceolate and subcordate, mostly 6-11 cm. long, subopposite to alternate, glabrous or slightly scaly beneath, not articu- late, the dark color of the stalk entering the base of the pinna, midvein distinct to the apex; a single very long sorus on each side of the pinna or pinnule. This species is somewhat variable in the shape of the pinnae. Usually the sterile ones are about twice as broa as the fertile, although occasionally the latter may also be broad. The base of the pinna varies from cuneate to nearly cordate. It is separated from the allied A. lucidum by the distinct midvein of the pinna, which is dark colored at the base. The related A. phyllitidis has a long dark colored portion of the midvein and the sterile margins of the pinnae are slightly and remotely crenate-serrate while A. Poeppig'- anum has a short dark colored portion of the midvein and sterile margins are minutely and densely serrate. Endemic to Peru. In dense forest, Amazonas to Junin, 135-500 m. Selected specimens: AMAZONAS: entre Aramango y Montenegt0, prov. Bagua, Lépez et al. 4227 (GH). SAN MARTIN: Tarapoto, ™ Williams 6141 (F, US); upper Rio Huallaga, L. Williams 6190 (F, US) 5 Juan Jui, Ferreyra 4525 (GH, USM). LORETO: Puerte Arturo, lower Rio Huallaga below Yurimaguas, Killip & Smith 27724 (F, NY, us); Santa Rosa, below Yurimaguas, Killip & Smith 28995 (NY, US); Yuri- maguas, Killip & Smith 28013 (¥, Ny, US); lower Rio Huallaga, L. Williams 5097 (F, US). JUNIN: Satipo, Aug. 1940, Ridoutt (GH, USM): 8. Adiantum lucidum (Cav.) Sw. Syn. Fil. 121. 1806. Fie. 10 Pteris lucida Cav. Deser. 266. 1802. Type: Guaranda, Ecuador isotype: s-PA! a: ‘ Née; j | . FERNS OF PERU 149 Rhizome moderately stout, short- -creeping, scales dull; leaves mostly 30-60 cm, tall, petiole ebeneous to dark brown, deciduoualy pubescent- scaly, rachis similar; lamina bhi: 1-pinnate, or ovate-triangular, bipinnate, entire pinnae and the pinnules asymmetrically ovate-lanceo- late to long-triangular, unequally cuneate, ca. 4-10 cm. long, alternate, glabrous or slightly scaly beneath, not articulate, the dark color of pinna or pinnule, or the latter with only one on the upper edge. Guianas to Colombia and Panama, south to Peru ; Trinidad and Tobago Forests, Junin. ecimens seen: JUNIN: Rio Satipo, 1940, Ridoutt (Us, USM); Pichita Caluga, Chanchamayo, Walden 64 (BM). 9. Adiantum alarconianum Gaud. Voy. Bonit. Bot. t. 99. 1846. Type: egg Ecuador, April, 1836, Gaudichaud, P! fragment Us! isotype: Fig. 104. yp iin ti incisum Pr. Rel. Haenk. 1: 61, t. 10. f. 3. 1825, not Forsk, 1775. eae cnr aere Ecuador, Haenke, 9 (var. a), 10 (var. B); isotypes: BM! Rhizome moderately stout, short-creeping, scales iridescent; leaves ca. 30-50 cm. tall, petiole reddish-brown to dark brown, glabrous or slightly pubescent-scaly on the upper side, rachis similar but definitely pubescent-scaly on the upper side, glabrous elsewhere; lamina linear to broadly linear and 1-pinnate or broadly ovate-oblong and bipinnate, entire pinnae or pinnules oblong-faleate to long-triangular, strongly asymmetrical at the base, ca. 1-4 em. long, alternate, glabrate beneath, Sessile or subsessile; a single long sorus borne on the upper edge. The iridescent rhizome scales and the indument confined to the upper surface of the rachis characterize this species. The sterile pinnae are deeply incised and the fertile usually have a single long sorus on the upper edge. In bi-pinnate blades only the basal or lower pinnae are 1-pinnate. Ecuador and adjacent Peru. Woods, Tumbes, 650 m. Specimens seen: TUMBES: between Cancho and Cotrina, Coronado 218 (GH, UC); Tumbes to Cancho, Coronado 228 (UC). 10. Adiantum petiolatum Desv. Ges. Naturfr. oe Berl. Mag. 5: 326. 1811. Type: uncertain, not seen at P. ‘FIG. 10: ‘ Adiantum Kaulfussii se Linnaea 21: 221. 1848. ‘Gas: Sieber, Fl. Martin. 371; isotype: 150 ROLLA TRYON izome slender, long-creeping, with spreading scales, the petiole bases usually well spaced; leaves ca. 20-40 em. long, petiole ebeneous or atropurpureous, glabrate, rachis often somewhat lighter; lamina more or less oblong and 1-pinnate or deltoid and bipinnate, with the pinnate pinnae few, entire pinnae ovate-lanceolate to oblong, strongly asymmetrical at the base, ca. 3-5 cm. long, pinnae and pinnules dull, usually glaucous beneath, sterile margins evenly serrate; sori several, oblong-arcuate to long-arcuate or nearly straight, more numerous on the upper than on the outer edge. This is a rather critical species and somewhat variable; the rachis, for example, is rarely glabrate. The characters given in the key to separate it from A. obliquum and A. humile are quite satisfactory with adequate material at hand but with partial specimens identification is more dif- ficult. It is apparently also closely related to A. latifolium. In that species the terminal pinna is conform with the several lateral ones and it is usually reduced at the base; its segments are sessile or subsessile. In A. petiolatum the terminal pinna in bipinnate laminae is much larger than the few lateral ones and broadest at the base, its basal segments have definite stalks. Guianas to Colombia, to Bolivia and Brazil; Mexico; West Indies, In forests, Loreto, Junin and Cuzco, to 700 m. Selected specimens: LORETO: near Iquitos, Tryon & Tryon 5191 (BM, F, GH, MO, U, UC, US, USM) ; Salinas, Rio Mazdn, Schunke 377 (F, GH, NY, UC, US, USM); Rio Santiago above Pongo de Manseriche, Mexia 6235 (BM, F, GH, MO, PH, US). JUNIN: Rio Satipo, Aug. 1940, Ridoutt (USM). cuzco: Atalaya, prov. Paucartambo, Vargas 13301 (GH). . 11. Adiantum obliquum Willd. Sp. 'Pl. 5: 429. 1810. Syntypes: Caracas, PS iaiagonce Bredemeyer (Herb. Willd. 20067-1) B! photo BM, GH; Po reat at (Herb. Willd. 20067-2) B! photo BM, GH. FIG. 106, ona slender, long-creeping but with usually closely spaced petiole bases, often branched, with appressed scales; leaves ca. 20-60 pants ly biserrate; sori several, short to long-oblong, straight oF €, more numerous on the upper than on tthe outer edge. FERNS OF PERU 151 This species is closely allied to the previous, A. petiolatum. Adequate material may be satisfactorily determined by the characters in the key. Bipinnate leaves may be confused with a number of the strictly bipinnate species. However, in A. obliquum as in A. petiolatum, the terminal pinna of a bipinnate lamina is larger than the few laterials, broadest at the base and its basal segments are usually stalked. Guianas to Colombia to Bolivia and Brazil; Mexico; West Indies. It grows in dense forests or on wooded slopes, San Martin and Loreto to Junin, 350-1500 m. Selected specimens: SAN MARTIN: Across the Huallaga on trail to Monson (Tingo Maria), Allard 21853 (US). LORETO: Balsapuerto, lower Rio Huallaga basin, Killip & Smith 28508 (F, NY, US); near mouth of Rio Santiago, above Pongo de Manseriche, Mexia 6206a (GH, UC, US); Rio Itaya, near Iquitos, Tryon & Tryon 5201 (BM, F, MO, U, US, USM). HUANUCO: Tingo Maria, Tryon & Tryon 5253 (BM, F, MO, U, US, USM), 5293 (BM, F, USM), Ferreyra 10259 (USM). JUNIN: Chan- chamayo valley, Schunke 95 (F), 763 (US), 831 (F). 12. Adiantum latifolium Lam. Encycl. 1: 43. 1783. Type: uncertain, there is no specimen in Herb. Lamarck; Brazil, Commerson (Herb. Jussieu no. 1408) P! photo GH, Us may serve to fix the application of the name. Fic. 107, Map 31. Rhizome slender, very long-creeping; leaves ca. 30-70 cm. tall, petiole brown to atropurpureous, deciduously pubescent-scaly, the rachis per- sistently so; lamina deltoid to ovate-deltoid or ovate-oblong, bipinnate, pinnae ca, 10-15 cm. long, pinna-rachises pubescent-scaly, the terminal pinna reduced at the base, pinnules herbaceous, glaucous beneath, gla- brate, with the sterile ones evenly serrate, those nearest the usually c several, oblong, usually on the lower as well as the upper edge. This species is distinguished from the other bipinnate species by its long-creeping rhizome with the petioles borne at intervals, its glaucous segments, the sterile ones with the margins finely and evenly serrate, and the usually distinct midvein of the pinnules. ropical America. In woods and dense forest, Loreto to Cuzco and Madre de Dios, 100-1000 m. Selected specimens: LORETO: Rio Putumayo, Peru-Colombia bound- ary, Klug 1631 (F, GH, MO, NY, US); Iquitos, Killip & Smith 27236 (F, GH, NY, US); Fundo Indiana, dist. Iquitos, Mexia 6389 (BM, F, GH, MO, 152 ROLLA TRYON NY, PH, UC, US); Rio Itaya, near Iquitos, Tryon & Tryon 5170 (BM, F, MO, U, US, USM). JUNIN: east of Quimiri Bridge, near La Merced, Killip & Smith 24003 (NY, US). AYACUCHO: Rio Apurimac valley, near Kimpitiriki, Killip & Smith 22925 (GH, Ny, US). cUzco: Hacienda Kosnipata, Socorro, prov. Paucartambo, Vargas 10202 (MO, UC). MADRE DE DIOS: Maldonado, Rauh P722 (B). 13. Adiantum serratodentatum Willd. Sp. Pl. 5: 445. 1810. Type: Caripe, Venezuela, Humboldt & Bonpland (Herb. Willd. 20088, the two sterile leaves) B! photo BM, GH. FG. 108. Rhizome slender, very long-creeping; leaves ca. 35-65 cm. tall, petiole brownish to atropurpureous, deciduously pubescent-sealy, the rachis persistently so; lamina deltoid-ovate to oblong-ovate, bipinnate, pinnae ca. 6-11 cm. long, pinna-rachises pubescent-secaly, the terminal pinna not or hardly reduced at the base, pinnules coriaceous, not glaucous, glabrate or slightly scaly beneath, sterile ones evenly serrate, those nearest the usually obtuse terminal segment of the pinna not much reduced; sori few to several, roundish to oblong or sublunate, borne on the upper and outer, rarely the inner, edges. This species is characterized by its coriaceous, usually small, oblong pinnules and the terminal pinna which is little, if at all, reduced at the base. The pinnules are commonly fertile except for the entire inner and lower edges; the other margins, when sterile, are usually finely and evenly serrate. The rhizome is cord-like and long-creeping as in the allied A. latifolium. Guianas to Colombia, to Bolivia and Brazil; to Mexico; West Indies. It grows in forests and in open habitats among grasses, San Martin and Piura, 750-1500 m. Selected specimens: PIURA: Cerro de Chiriris, Pajonal, Biies 1711 (US). SAN MARTIN: Zepelacio, near Moyobamba, Klug 3462 (F, GH, MO, NY, US); Tarapoto, Woytkowski 35103 (Mo, Uc), L. Williams 5896 (F)- 14. Adiantum villosum L. Syst. Nat. ed. 10, 2: 1328. 1759. Type: LINN 1252.10, photo A, is evidently this species. G. 109. Rhizome moderately slender, rather short-creeping; leaves ca. 40-70 cm. tall, petiole dark reddish-brown to atropurpureous, deciduously pubescent-scaly, much reduced; sori few anc outer edges, FERNS OF PERU 153 ium: base of fertile pinna, X 1, Peru, f fertile pinna, X 1, Peru, Klug Fics, 107-111. Fig. 107. Adiantum latifol Mexia 6389, Mo. Fig. 108. A. serratodentatum: base 0 3462, Mo. Fig. 109. A. villosum: A, base of fertile pinna, X 1, G 947, GH; B, apex of sterile pinna, X 14, idem. Fig. 110. 2 pinna, X 1, Peru, Meria 8175, GH. Fig. 111. A. villosis X 1, Peru, Macbride 5648, F. 154 ROLLA TRYON The pinnules near the apex of a pinna are about 7 as long as the longest on the pinna, rather than much smaller, and this difference distinguishes A. villosum from related Species that also have coarse and irregular serrations on the Sterile margins. In general aspect the leaves are otherwise | imilar to those of A. tetraphyllum. ike not been able to account satisfactorily for A tum Cecileae Alston (Lilloa 30: 109, t. 4, 5. 1960, Type: moist sandstone on edge of forest, South bank of Rio Perené, within 15 miles of confluence of Rio Chanchamayo and Rio Paucartambo, Junin, Peru, Gascoyne-Cecil 50 BM . para- type: 47, BM!). The species seems to be close to A. villosum in the characters of the dimidiate, subtrapeziform pinnules, the short-creeping rhizome, the shape of the sori and the type of indument (or lack of it) on various parts of the leaf. All of these characters are within the range of variation of A. villosum, although they are not correlated as in A. Ceeil- eae. The pinnules have longer stalks than any I have seen in A. villosum. Venezuela to Colombia, to Peru and Brazil; Mexico; West Indies Forests, Loreto, Junin and Cuzco, 700-1500 m. Specimens seen: LORETO: Rio Itaya, near Iquitos, Tryon & gee 5197 (BM, F, GH, Mo, USM). JUNIN: Chanchamayo valley, Schunke ) (F); La Merced, Killip & Smith 23445 (NY, US), Soukup 1110 (F). CUZCO: Quelloune, prov. Convencién, Vargas 13359 (GH). 15. Adiantum Kalbreyeri C. Chr. Ind. Fil. 28. 1905, based on pilosum Bak., Ann. Bot. 5: 207. 1891, not Fée, 1852. Type: Colombia, Kalbreyer 956, K! photo cu. Fie. 110. terminal segment greatly ro ; 8, roundish, borne on the upper and outer edges: Costa Rica to Peru. Forested slope, Huanuco, 1200 m. Specimens see n: HUANUCO: Hacienda Mercedes, Balsa-Playa, dist. Churubamba, pr 8 ov. Hudnuco, Mexia 8175 (BM, F, GH, MO, US) FERNS OF PERU 155 16. Adiantum villosissimum Kuhn, Linnaea 36: 73. 1869. Type: Turbo, Panama, Schott 65, Mo! fragments ex Engelm. B! =‘ Fi. 111. Rhizome moderately slender, long-creeping; leaves ca. 50-80 cm, tall, petiole dark reddish-brown to atropurpureous, glaucous, persistently or deciduously scaly, the rachis scaly; lamina ovate to deltoid-ovate, bipinnate (sometimes the basal pinnae with a single branch on the lower side and then tripinnate), pinnae ca, 15-20 em. long, pinna- rachises scaly, the terminal pinna reduced at the base, pinnules with n few, oblong to linear, sometimes arcuate, borne on the upper and outer edges, The bipinnate lamina with the rachis and pinna-rachises scaly with freely ciliate scales and the segments with tri- chomes on the under surface combine to make this a very distinctive species. Related species with scales on the rachis have narrow scales and trichomes as well. Some leaves are tripinnate, the inner lower pinnule of the basal pinnae being well developed and fully pinnate. anama and Peru. Dense forest and forested slopes, San Martin to Junin, 1100-1500 m. Selected specimens: SAN MARTIN: Tarapoto, Spruce 4665 (K). HUANUCO: Balsa Playa, prov. Hudnuco, Vargas 5317 (UC; USM, s.n.). Tingo Maria, Aguilar 314 (USM). JUNIN: La Merced, Chanchamayo, Soukup 1078 (F); Chanchamayo valley, Schunke 105 (F), 872 (F); Schunke Hacienda, La Merced, Macbride 5648 (F, US), Killip & Smith 24691 (F, NY, US). 7. Adiantum tetraphyllum Willd. Sp. Pl. 5: 441. 1810. Syntypes: “Sad Willd. 20082-1, ex Vahl, and -2, Caripe, Humboldt & Bonpland, B! photos GH. Fic. 112. Rhizome rather slender, long-creeping, the petiole bases distant; leaves ca. 50-90 cm. tall, petiole dark reddish-brown to atropurpureous, more or less deciduously pubescent-scaly as is the rachis; lamina broad- ly ovate to deltoid-ovate, bipinnate, pinnae ca. 10-18 cm. long, pinna- rachises pubescent-scaly, the terminal pinna reduced at the base, pin- nules glabrate or slightly scaly beneath, the usually acute sterile tips sori few to several, short-oblong to lunate, borne on the upper and also often on the outer . 156 ROLLA TRYON The two species here recognized in the alliance of A. tetra- phyllum, the true A. tetraphyllum and A. fructuosum, are maintained with some hesitation. The characters seem ade- quate but until the whole complex is worked out in the American tropics, their status will remain doubtful. The rhizome is long-creeping and the sterile apex of the pinnules is acute and mostly turned upward in A. tetraphyllum, while in A. fructuosum the rhizome is short-creeping, often mas- Sive with crowded petiole bases, and the sterile apex of the pinnules is obtuse and rather straight. In either species identification may be difficult in cases where the rhizome 1s lacking on a specimen or in some heavily fertile leaves that have few pinnules with a sterile apex. Tropical America. Dense forest and wooded slopes, Loreto, Hudnuco south to Ayacucho and Madre de Dios, 100-1000 m. Selected specimens: LorETO: Soledad on Rio Itaya, Killip & Smith 29573 (F); lower Rio Huallaga, L. Williams 5053 (NY). HUANUCO: Pampayacu, mouth of Rio Chinchao, Macbride 5029 (F). JUNIN: Chan- chamayo valley, Schunke 9 (F, US). La Merced, Tryon & Tryon 7 (BM, F, GH, MO, U, UC, Us, USM). AYACUCHO: between Huanta and Rio Apurimac, Killip & Smith 22623 (r, NY, US), 22714 (NY, US). MADRE DE DIOS: Maldonado, Rauh P749 (B). 18. Adiantum fructuosum Spreng. Syst. Veget. 4: 113. 1827. Type: Cuba, 1822, Poeppig; isotype: B! us! L, photo GH, Us. 1G. 118. Rhizome moderately stout to slender, rather short-creeping, the petiole bases commonly closely approximate; leaves ca. 60-100 em. tall, petiole dark reddish-brown to atropurpureous, more or less persistently pubescent-scaly, Sometimes densely so, the rachis similar; lamina > Sori few to several, roundish to usually short-oblong the upper and usually also on the outer edges. The differences from A, tetraphyllum are discussed under that species. Venezuela to Peru and Brazil; Mexico; West Indies. Dense forest, Tumbes to Loreto, south to Cuzco, 135- 1200 m. FERNS OF PERU 157 Selected specimens: TUMBES: between Cancho and Cotrina, Coronado 223 (GH, UC). AMAZONAS: ca. 40 km. s. of Bagua Grande, Hutchison 1463 (GH). LORETO: Rio Nanay, near Iquitos, Tryon & Tryon 5177 (BM, F, GH, MO, U, US, USM) ; Yurimaguas, Ferreyra 4990 (USM); La Victoria, on the Amazon, L. Williams 2549 (F). HUANUCO: Pampayacu, Kanehira 153 (GH, US); Tingo Maria, Tryon & Tryon 5294 (BM, F, GH, MO, U, US, USM). CUZCO: Quellu-uno, Biies 1702 (us). 19, Adiantum pulverulentum L. Sp. Pl. 2: 1096. 1753. Type: uncertain. Fig. 114. Rhizome moderately stout, rather short-creeping; leaves ca. 30-90 em. tall, petiole dark reddish-brown to e neous, more or less persist- serrate, those nearest the acute to long-acuminate terminal segment greatly reduced; one long sorus (rarely two) borne on the upper edge, rarely a shorter one also on the outer edge. The single long sorus on the upper edge of the pinnules is sufficient to characterize this species. Occasionally there may be two sori, or rarely an additional one on the outer edge. Guianas to Colombia, to Bolivia and Brazil; Mexico; West Indies. In dense or partially open forest, Amazonas and Loreto south to Cuzco, 100-1800 m. Selected specimens: AMAZONAS: mouth of the Rio Santiago, Mexia 6133a (GH, NY, UC, US), Tessman 4277 (B). LORETO: Rio Itaya, Rip & Smith 29301 (NY, US), 29339 (F, NY, US), 29431 (F, GH, NY, US); near Iquitos, Klug 311 (F, NY, US), 1399 (F, NY, US), Tryon & Tryon 5194 (BM, F, GH, MO, U, USM); Rio Mazdn, Schunke 108 (F, cada UC, Us, USM); lower Rio Huallaga, L. Williams 5299 (F). JUNIN: Polonia, Satipo, 1940, Ridoutt (us). cuzco: Rio Sambray, prov. Con- vencién, Mexia 8066 (F, GH, MO, UC, US). 20. Adiantum tomentosum KI. Linnaea 18: 553. 1845. Type: British Guiana, Schomburgk 1202, 8! fragment, GH! Us! Fie. 115. Rhizome moderately stout, rather long-creeping; leaves to ca. pe eae tall, petiole dark reddish-brown to atropurpureous, more nt hei or ously puberulent, the rachis persistently so; lamina ova nf . gw: broadly so to broadly ovate-oblong, bipinnate, pinnae ca. er 5 eh & pinna-rachises puberulent to pubescent on the upper si e apes od brous beneath, the terminal pinna reduced at the base, day es Ee brous, usually imbricate, sterile ones rather coarsely but evenly serrate, 158 ROLLA TRYON Figs. 112-117, Fig. 112. Ad TYyYONn 38, i see iantum tetraphyllum: A, apex of fertile pinna Peru, Tryon & T MO; B, portion A - fructuosum: fertile pinna Y%, wie ; Fig. 118. of pinna-rachis, enlarged, idem. PH ig. 117. A- A. macrocladum: base of fertile pinna, X 1, Peru, Mexia 8292, GH. Fig. terminatum: base of fertile pinna, X 1, Peru, Klug 1270, US. FERNS OF PERU 159 those nearest the usually acute terminal segment moderately reduced; sori numerous, roundish to short-oblong, borne on the upper and outer edges, sometimes a few also on the inner edge. This is one of the most distinctive species, the pinna- rachises being puberulent or short-pubescent on the upper side only. Except for the next species, it is also distinctive in its closely imbricate and very obtuse pinnules that are often curved. Guianas to Colombia, to Peru and Brazil. Dense forest, Loreto, Hudnuco and Junin, 100-400 m. Selected specimens: LORETO: Iquitos, Killip & Smith 27415 (NY, US), 27490 (NY, US); near Iquitos, Killip & Smith 29946 (NY, US); Rio Nanay, L. Williams 1065 (F), 1070 (F). Serb Tingo Maria, Agui- lar 301 (uc), Tryon & Tryon 5295 (BM, F, GH, NY, U, UC, US, USM). JUNIN: Puerto Bermudez, Killip & Smith peri get NY, US); Puerto Yessup, Killip & Smith 26376 (F, NY, US). 21. Adiantum macrocladum KI. Linnaea 18: 554. 1845. Red Peru, Poeppig 1147, B! photo BM! isotypes: GH! LE! Us! Fic. Adiantum Meviae Copel. Univ. Cal. Publ. Bot. 19: 303. ore Type: Huanuco, Peru, Mexia 8292, UC; isotype: GH! Us! upper side into two to four major divisions and tripinnate, pinnae ca. the tnde. pinnules glabrous, imbricate, acts ones mostly evenly ser- rate, those nearest the long-acuminate terminal segment greatly reduced; sori numerous, very short-oblong, borne on the upper edge. In the general aspect of the pinnae this species is similar to the previous, A. tomentosum, but the wholly glabrous pinna-rachises, in addition to the petiole and rachis, and the acroscopically branched basal pinnae easily distinguish it. e pinnules are quite long and obtuse and the sori small and numerous. Endemic to Peru. Dense forests, Hudnuco and Junin, 340-860 m. Specimens seen: Poeppig 1147 (B, GH, LE, us), 1829, Poeppig (Mo). HUANUCO: Tingo Maria, Stork & Horton 9531 (F, UC); near aan of Rio Cayumba and Rio Huallaga, Mexia 8292 (GH, Us). JUN Cahuapanas on Rio Pichis, Killip & Smith 26790 (NY, US). 160 ROLLA TRYON 22. Adiantum terminatum Mig. Het. Instit. Versl. Meded. Ned. Instit. Wet. 1842: 3. 1843 (cited by C. Chr. Ind. Fil. as Diar. Inst. Reg. Bat.). Type: Bergendall, Surinam, Focke, vu! Fig, 117. Rhizome rather slender, short-creeping, the petiole bases approxi- mate; leaves ca. 20-45 cm. tall, petiole dark reddish-brown to atropur- pureous, deciduously pubescent-scaly, the rachis persistently so; lamina broadly deltoid to oblong or oblong-ovate, bipinnate, pinnae ca, 8-15 em. long, pinna-rachises pubescent-scaly, the terminal pinna reduced serrate, those nearest the acuminate terminal segment reduced, sori several to numerous, very short-oblong, borne on the upper and often also on 'the outer and inner edges. Adiantum terminatum is evidently a critical species for none of the identified material I have seen from Peru has been properly named. The numerous small sori, the pubes- cent-scaly rachis and pinna-rachises, the reduced apical pin- nules and the pinnules with trichomes on the under surface serve to distinguish it. Maps 31-32, Map 31, Adiantum latifolium. Map 32, A. subvolubile. FERNS OF PERU 161 Guianas to Colombia, to Bolivia and Brazil; Guatemala; Trinidad In forests, Loreto and San Martin to Puno, 100-800 m. Specimens seen: SAN MARTIN: near Tingo Maria, Allard 20451 (us), 22542 (US). LORETO: near Iquitos, Klug 1270 (NY, US); lower Rio Huallaga, L. Williams 5058 (F); Nauta, Ferreyra 5134 (USM), HUANU- co: Tingo Maria, Stork & Horton 9520 (F, UC), Tryon & Tryon 5827 (BM, F, GH, MO, NY, U, UC, US, USM), Allard 21505 (GH, US). JUNIN: Satipo, Aug. 1940, Ridoutt (GH, USM); Puente Perené, Coronado 262 (GH, UC). PUNO: San Gaban (Rio), Lechler 2319, 2319a (B). 23. Adiantum humile Kze. Linnaea 9: 80. 1834. Type: Upper Hual- laga, Mission Tocache, Peru, June, 1830, Poeppig, royed. An authentic specimen: Tocache, Aug. 1830, Poeppig, W w! may cone the holotype. Wn 3 Adiantum Killipii Maxon & Weath. Amer. Jour. Bot. 19: 166. 1932. Type: Ancén Hill, Canal Zone, Panama, Killip 2752, us! Rhizome moderately stout, very short-creeping, the petiole bases approximate; leaves ca. 30-75 em. tall, petiole dark reddish-brown to atropurpureous, deciduously scaly, the rachis persistently so; lamina deltoid to oblong-ovate, bipin innate, pinnae ca. 8-15 cm. long, pinna- outer British Honduras to Panama; French Guiana and Trini- dad to Colombia and south to Peru ; Brazil. In forest, San Martin and Junin, 500-700 m. Specimens seen: SAN MARTIN: Tocache, Poeppig (W). JUNIN: Qui- miri Bridge, La Merced, Killip & Smith 24003 (NY, Us). 24. Adiantum cayennense KI. Linnaea 18: 552. 1 1845. Type: Herb. Willd. 20084, B! photo GH; paratype: British Guiana, Schomburgk 1201,8! Fic. 119. izome moderately stout, rather short-creeping; leaves ca. 80-100 em. tall ea dark reddish-brown to atropurpureo 15-20 cm. long, pinna-rachises densely pubescent-scaly, pinnules gla- brate or scaly beneath, sterile ones coarsely and often unevenly serra those nearest the broadly acute to acuminate terminal segment greatly redu n blong, borne on the upper and outer edges, often a few on the inner 162 ROLLA TRYON This species is closely related to the two preceding ones; however, it has the pinnules glabrous or only with a few scales beneath, while A. terminatum and A. humile have the pinnules with simple trichomes beneath. Guianas to Colombia, Peru and Brazil. In forests, Loreto, San Martin and Hudnuco, 135-800 m. Specimens seen: SAN MARTIN: east of Tingo Maria Allard 21623 (US). LORETO: Balsa Puerto, lower Rio Huallaga basin, Killip & Smith 28510 (NY, US); below Yurimaguas, Killip & Smith 28785 (NY, US); lower Rio Huallaga, L. Williams 4133 (F, US). HUANUCO: Tingo Maria, Tryon & Tryon 5276 (BM, F, MO, USM). 25. Adiantum pectinatum Ettingsh. Farnkr. 85, t. 45. figs. 14-16. 1865. Type: Goyaz, Brazil, Pohl 1481, w!; isotype: B! (s.n.), GH! Fi. 120. Rhizome very stout, moderately creeping ; leaves very large, up to 2 m. tall or more, petiole very stout (to nearly 1 cm. in diameter at the base), dark reddish-brown to atropurpureous, deciduously pubes- cent-scaly, the rachis rather persistently so; lamina broadly deltoid- ovate to deltoid-circular, 4- to 5-pinnate, rarely 6-pinnate, the basal pinnae largest, pinnae widely alternate, 30-90 cm. long, pinna-rachises pubescent-scaly, ultimate segments glabrate or slightly scaly beneath, sterile ones deeply incised-lobed, those nearest the rather obtuse to acute terminal segment of ‘the penultimate segments moderately to greatly reduced; sori several, roundish to lunate, borne on the upper and sometimes also a few on the outer edge. This species is characterized by the deeply incised pin- nules, or ultimate segments, and the highly divided lamina that is commonly 4- to 5- pinnate at the base. It is the largest of the Adiantums, some leaves reaching a length of at least 2 m. It differs from the related A. polyphyllum Willd., with Which it has been confused, in having the several sori borne only on the upper edge of the pinnules and in having the stipe, rachis and pinna-rachises pubescent-scaly rather than glabrous. Costa Rica to Argentina and Brazil. In woods and dense forests, San Martin to Cuzco, 700- 2000 m. a oe + sical SAN MARTIN: Moyabamba to Huallaga, Stiibel hira 130 re yo an ee 4781 (K). HUANUCO: Pampayacu, Kane- 1082 (F), 3 PM Giga Pearce 552 (K). JUNIN: La Merced, Soukup r » 3405 (GH, Us), Killip & Smith 23448 (GH, NY, Us), Tryon & te eee (BM, F, MO, U, Us, USM). cU cién, Pots iene pid 1292 (Us) ; Baling de Rosalina, prov. Conven- FERNS OF PERU 163 12068, A. eayennense: basal portion of fertile pinna, X 1, Peru, & Smith f fertile pinna, X 1, Peru, Soukup 3405, a EF ig. Figs. 118-124, Fig. 118. Adiantum humile: fertile pinna, X 14, Panama, Killip Killip GH. Fig. ey = . 164 ROLLA TRYON 26. Adiantum sessilifolium Hook. Sp. Fil. 2: 44, t. 85B. 1851. Type: Chachapoyas, Peru, Mathews 1855, K! photoGH. FIG. A Adiantum Henslovianum var. macrosora Hieron. Hedwigia 48: 288. 1909. Syntypes: Peru, Stiibel 1031, 1046 B! Rhizome rather small, short-creeping; leaves ca. 15-30 cm. tall, petiole dark straw-colored to dark reddish-brown, more or less persist- ently short-pubescent, the rachis definitely so; lamina elongate-deltoid, bipinnate, pinnae ca. 2-6 cm. long, the terminal one broadest at ‘the base, pinna-rachises whitish short-pubescent, pinnules (and undivided pinnae) whitish-pubescent beneath, sterile ones moderately lobed, those nearest the obtuse to broadly rounded terminal segment not much reduced, the basal pinnule of the upper pinnae overlying the rachis; sori few, oblong-lunate to reniform, borne on the upper and outer edges. Endemic to Peru. Rocky slopes and among rocks of Inca walls, Cajamarca and Amazonas to Cuzco, 2000-2700 m. Specimens seen: CAJAMARCA: west of Balsas, Osgood & Anderson 67 (F), Osgood 67 (US); Celedin, Stitbel 1046, 1046a (B); 40 km. from Cajamarca on road to Chilete, Correll & Smith P842 (GH). AMAZONAS: Chachapoyas, Mathews 1855 (kK); Leimebamba, Stiibel 1031 a LIBERTAD: between Alpamarca and Retamas, prov. Pataz, Lopez " Sagdstegui 3634 (GH). HUANUCO: Mufia, Bryan 428 (F, GH). CUZCO: about Cuzco, 1854, Wm. Lobb (BM). 27. Adiantum Henslovianum Hook. f. Trans. Linn. Soc. Lond. 20: 160. 1847. Syntypes: Galapagos Isls., Darwin: James Is. CGE, tracings nee Charles Is. K! photo GH. : : Adiantum laetum Kuhn, Linnaea 36: 76. 1869. Type: Chachapoyas, Peru, Mathews 3295; isotype: K! photo GH, US. Rhizome rather small, very short-creeping; leaves ca. 40-90 cm. pre petiole light to rather dark reddish-brown, glabrate to puberulent, a rachis puberulent; lamina more or less ovate, bipinnate, pinnae ca. 6- cm. long, the terminal pinna broadest at the base, pinna-racnst puberulent, pinnules whitish-pubescent beneath, or slightly 50, nee ones moderately lobed, those nearest the acute to broadly es terminal segment greatly reduced or not, the basal pinnule of at lea the upper pinnae overlying the rachis; sori few, roundish, junate or reniform, borne on the upper and outer edges. The flabellate-cuneate upper basal pinnules that are borne very close to the rachis so that they overlap it, the nearly sessile pinnae and the puberulent rachis and pinna-rachises serve to distinguish this species. The pinnules are delicate in texture. Venezuela to Ecuador and Peru. FERNS OF PERU 165 Rocky slopes, Lambayeque and Amazonas, ca. 1500 m. Specimens seen: LAMBAYEQUE: 31 km. from Olmos on road to Jaén, Correll & Smith P807 (GH). AMAZONAS: Chachapoyas, Mathews 3295 (K). 28. Adiantum lobatum Presl, Rel. Haenk. 1: 62, t. 10, f. 4. 1825. Type: Guayaquil, Ecuador, 1790, Haenke, pr, photoGH. FIG. 123. Rhizome rather slender, short-creeping, the petioles spaced but not distant; leaves 'to 60 cm. tall, petiole atropurpureous, glabrous; lamina deltoid to long-triangular, bipinnate to tripinnate, rachis and pinna- rachises minutely puberulent on the upper side, terminal pinna some- what reduced at the base, ultimate segments glabrous, mostly oblong to trapeziform, the basal ones more or less flabellate, entire to some- times rather strongly incised, stalks very short to usually short, the dark color entering the base of the segment, sterile vein-tips end be- tween indistinct marginal crenulations; sori orbicular to suborbicular, borne on the upper and outer edges. Ecuador and Peru. Brushy and rocky hillsides, Lambayeque, 1250-1400 m. pecimens seen: LAMBAYEQUE: 27 km. from Olmos on road to Jaén, Correll & Smith P801 (GH, LL, US); 31 km. from Olmos on road to Jaén, Correll & Smith P808 (GH, LL). 29. Adiantum patens Willd. Sp. Pl. 5: 439. 1810. Type: Caracas, Vene- zuela, Bredemeyer (Herb. Willd. 20078), B! photo GH; isotype: w! Fic. 124. Rhizome rather small, short-creeping; leaves ca. 20-45 cm. tall, petiole light to dark reddish-brown, glabrous or slightly puberulent, rachis (or rachises) similar; lamina deltoid-ovate to nearly circular, bi- to tripinnate, penultimate segments on the upper side of each part of the divided rachis, or the lamina more or less pinnate in division and the terminal pinna somewhat reduced at the base, penultimate seg- short-pubescent, sterile ones moderately lobed, or evenly serrate, those nearest the broadly rounded terminal segment reduced or not, the basal ultimate segments hardly or not overlying the adjacent axis; sori few, roundish, lunate or reniform, borne on the upper and outer edges. The manner of division of the lamina is similar to that in A. pedatum of North America, that is the rachis is divided and the pinnae arise only from the acroscopi¢ side of each of the two branches. This character alone will identify the species; however, in small or poorly developed leaves the 166 ROLLA TRYON lamina may be essentially pinnate and then the glabrous rachis and pinna-rachises (rarely the latter are partially puberulent) must serve to distinguish it from the preceding species. Venezuela and Colombia to Bolivia; Mexico. In woods or on shaded rocks, Tumbes to Lambayeque and Cuzco, 150-2050 m. Specimens seen: TUMBES: between Cancho and Cotrina, Coronado 233 (GH). PIURA: Talara, Haught 92 (Us); Huancabamba, Ferreyra 10877 (GH, USM). LAMBAYEQUE: prov. Lambayeque, Lépez et al. 4045 (GH); Olmos to Jaén, Correll & Smith P806 (GH). cuzco: Machu- Picchu, Vargas 3352 (MO, US). 30. Adiantum Capillus-Veneris L. Sp. Pl. 2: 1096. 1753. Type: LINN 1252. 9 chosen by Pichi-Sermolli in Webbia 12(2): 678. 1957. FIGS. 94, 125. Rhizome rather slender, short- to long-creeping; leaves ca. 15-45 om. tall, petiole reddish-brown to atropurpureous, glabrous, or slightly scaly at the base, rachis similar, glabrous; lamina elongate-deltoid to ovate to elongate-ovate, bi- ‘to tripinnate, pinnae stalked, ultimate seg- ments cuneate-flabellate to suborbicular, rather symmetrical or not, glabrous beneath, not articulate, the color of the apex of the stalk passing into the base, sterile margins with a cartilaginous border, each vein ending in a tooth; sori few, rather squarish ‘to oblong. The Peruvian material is evidently identical to the Euro- pean. It is distinct among other Peruvian species by the pronounced cartilaginous border of the segments, the light grayish-brown rhizome scales and the veins that each end im a tooth on the sterile margins. Adiantum Capillus-Veneris is unique among Peruvian ferns in having its distribution restricted to the coastal portion of the Department of Lima and a valley in Ica. Its close correlation with city, suburban and resort areas strong- ly suggests that it has been introduced as a garden plant and become locally naturalized. Tropical and temperate America; Old World. Damp, gravelly sea cliffs, banks of irrigation ditches and local seepage areas, Lima and Ica, sea level to 2000 m. Selected specimens: LIMA: Lomas de Amancaes, Ferrey?Ta 6257 - SM); Laguna de Villa, Coronado 4 (GH, Mo, UC, US); Choisica atucana, Vargas 4774 (uc) ; Miraflores, Tryon & Tryon 5216 (BM, FERNS OF PERU 167 F, GH, MO, U, UC, US, USM) ; Chorrillos, Tryon & Tryon 5460 (BM, F, MO, U, USM). ICA: Huamani, collector unknown (USM). 31. Adiantum subvolubile Kuhn, Linnaea 36: 77. 1869. Type: Puento de Bajios, Ecuador, Spruce 5318, B!; isotypes: BM! GH! K! ‘Fig. 126, Map 32. Rhizome rather slender, moderately long-creeping, with tufts of petiole bases borne at intervals; leaves ca. 20-60 cm. long, petiole light to dark reddish-brown, glabrous or slightly scaly at the base, rachis similar, glabrous; lamina more or less elongate-ovate, bipinnate, or tripinnate in the center, pinnae subsessile, the basal pinnule undivided, overlying the rachis, basal pinnae often reduced or withered, ultimate segments cuneate-flabellate or broadly so to suborbicular, rather sym- metrical or not, glabrous beneath, not articulate, the color of the apex of the stalk passing into tthe base, sterile margins with each vein ending in a usually well developed sinus; sori few, roundish to lunate to nearly reniform. This species is characterized by the subsessile pinnae with the inner pinnule borne very close to the rachis and overlapping it. It differs from the related A. excisum of Chile in having a glabrous rather than a deciduously scaly rachis. In addition A. excisum is a smaller and more delicate species and has large, thin, pale brown scales at the base of the petiole. Adiantum subvolubile has been more frequently confused with A. concinnum, a common species northward, but differs in the character of the inner upper pinnules of the lower pinnae mentioned in the key. Ecuador and Peru. Adiantum subvolubile is one of the characteristic ferns of the coastal lomas. In crevices of rocks, or at the base of rocks, less often on the ground, Piura to Cuzco and Moque- gua, 50-3000 m. Selected specimens: PrURA: Canchaque, Ferreyra 3095, 10813 (USM). CAJAMARCA: Summit of Cerro Prieto, Haught 280 (US). LIBERTAD: Lomas de Viru, Coronado 283 (GH, UC); Lomas Campana, Coronado 282 (GH, UC). ANCASH: Lomas de Mongon, Coronado 303 (GH, UC). LIMA: Loma de Atacongo, Pennell 14762 (F, GH, NY, pH), Ferreyra 2440 (BM, GH, USM); Loma de Lachay, Coronado 24 (UC, Us); Loma de Amancaes, Coronado 13 (GH, MO, UC, US), Tryon & Tryon 5214 (BM, F, MO, U, US, USM) ; Rio Chillén, above Obrajillo, Pennell 14368 (F, GH, NY, PH); Lomas de Quilman4, Coronado 28 (UC, US); Lomas de Mon- gomarca, Velarde (Coronado 19) (GH, UC, US); Lomas de Pativilea, Coronado 308 (GH). HUANUCO: Gorge of Rio Chinchao, Tryon & Tryon 5315 (BM, F). cUzCO: Sicuani, D. Stafford 419 (BM). AREQUIPA: Mol- lendo, I. M. Johnston 3573 (GH, US); Lomas de Capac, Coronado 38 168 ROLLA TRYON (UC, US); slope of Pichu-Pichu, Sandeman 3833 (K). MOQUEGUA: Ilo, D. Stafford 931 (BM). 32, Adiantum concinnum Willd. Sp. Pl. 5: 451. 1810. Type: Caracas, Venezuela, Humboldt & Bonpland (Herb. Willd. 20099), B! photo GH. Fig. 127. Rhizome moderately slender, rather short-creeping, often multici- pital; leaves ca. 20-80 cm. tall, petiole reddish-brown to atropurpureous, glabrous or slightly scaly at the base, rachis similar, glabrous; lamina ovate-oblong to linear-lanceolate, bipinnate (except at the very base) to generally tripinnate, pinnae subsessile, the basal pinnule divided into two ultimate segments, at least at the base of the blade, overlying the rachis, ultimate segments cuneate-flabellate to broadly so to suborbicu- 128A woe seco — 125. Adiantum Capillus-Veneris: A, fertile pinna, X Mego ryon MO; B, sterile co mee se ¥ ~~ om B; gment, X 114, idem. ; subvolubile: A, he basal pinna, X %, Peru, Grant 7444, GH; B, sterile — acca a 1 , Coronado 289, cu Fig. 1 127, A concin al pinna, X n . Eotgoaglan 229, GH. Fig. 128. A. Raddia wanum: A, fertile basal aR, xX % P mado 157, ct; B, sterile ultimate segment, X 114, Peru, J. B. Steere, 6 FERNS OF PERU 169 lar, rather symmetrical or not, glabrous beneath, not articulate, the color of the apex of the stalk passing into the base, sterile margins with each vein ending in a sinus; sori few to several, roundish to lunate to reniform. The differences from the closely related A. subvolubile are discussed under that species Venezuela to Ecuador and adjacent Peru; to Mexico; West Indies In woods and along irrigation ditches, Tumbes to Ca- jamarca, 200-2200 m Specimens seen: TUMBES: between Tumbes and Cancho, Coronado 229 (GH, UC) ; between Cancho and Cotrina, Coronado 222 (GH, UC). PTURA: cerca a Chanchaque, Ferreyra 10813 (GH, USM). ae ae aes apt km. from Olmos on road to Jaén, Correll & Smith P791 (GH). CAT alrededores de San Benito, prov. Contumaza, Sagdstegui 3742 is 7m. 33. Adiantum Raddianum Presl, Tent. Pterid. 158. 1836, based on Raddi, Pl. Bras. 1: t. 78, f. 2.1825. Fic. 128, Map 33. Adiantum cuneatum Langsd. & Fisch. Ic. Fil. 23, t. 26. 1810, not Forst. 1786. Type: Ins. St. Catharina, Brazil, Langsdorff, LE! photo GH! isotype: BM! Adiantum colpodes Moore, Gard. Chron. 1865: 530. Type: Ecuador, Pearce; Cult. Chelsea, Herb. Moore, K! photo GH Adiantum a Moore, Gard. Chron. 1862: 932. Type: Peru, Hort. Veitch, K! photo Adiantum cae, Moore, Gard. Chron, 1868: 866. Type: Bolivia, Hort. Veitch, K! photo GH ‘A diab decorum tours, Gard. Chron. 1869: 582. Type: Peru, Pearce, K! photo GH. Adiantum Moorei ere Gard. Chron. 1873: 811, based on A. ama- bile Moore, Gard. Chron. 1868: 1090, not Liebm. 1849. Type: Peru, Pearce, K! photo GH. Adiantum rufopunctatum Kuhn, Jahrb. Bot. Gart. Berlin 1: 350. 1881. Type: evidently Yungas, Bolivia, D’Orbigny 165, B!; isotype: P, photo GH, Us. Adiantum boliviense Christ & Rosenst. Fedde Rep. Spec. Nov. 5: 230. 1908. Type: Bolivia, Buchtien 459 s-PA!; isotype: P! us! yates Remyanum Espinosa, Bol. Mus. Nac. (Santiago, Chile) 15: 96, t. 3. 1936. Rhizome rather stout, very short-creeping, multicipital; leaves ca. 15-55 em. tall, petiole dark reddish-brown to blackish, glabrous or slightly scaly at the base, rachis similar, glabrous; lamina elongate- triangular to lance-ovate, broadly ovate or deltoid, commonly tripin- nate, less often bipinnate or quadripinnate, pinnae stalked, ultimate 170 ROLLA TRYON segments cuneate and symmetrical to broadly cuneate-flabellate and asymmetrical, glabrous or with sessile glands beneath, not articulate, the color of the apex of the stalk passing into the base, sterile margins with each vein ending in a more or less well developed sinus, sori few to several, roundish to suborbicular-reniform. An entirely satisfactory treatment of this species or species complex must await monographic study. There are four variants within the species in Peru that, on some grounds, might warrant specific recognition but, on others, seem to represent minor variations. The characters con- cerned are not of the same value as in related species and are more variable and not as well correlated as one would like. Two of the variants have small segments that are cuneate and the lamina is usually tripinnate. One of these elements, to which the name A. Raddianum evidently should be strictly applied, has glabrous segments, while the other, which is A. rufopunctatum, is glandular beneath. The other variants have the segments larger and asymmetrically or broadly cuneate and the lamina often bipinnate. One, prob- ably A. colpodes, is glandular beneath, and the other, A. tinectum, is glabrous. The glandularity is variable and may be present or absent in a single collection, for example, Tryon & Tryon 5446. Leaves with large asymmetrical segments and small, cune- ate, symmetrical ones both occur in Macbride & Feather- stone 1435 and Tryon & Tryon 5368 (the small-segment kind at F). Tryon & Tryon. 5876 is also variable in the Adiantum Raddianum may be separated from the related A. Poiretii by its orbicular to suborbicular sori and its mpi rhizome with the petioles always closely clus- ered. Tropical America, In forests, open woods, thickets, rocky banks and cliffs, border of irrigation ditches, Tumbes and Amazonas 10 Cuzco, 600-4000 m. FERNS OF PERU 171 Selected specimens: TUMBES: El] Cancho to Cotrina, Coronado 233 (uc). CAJAMARCA: Hacienda Limén, w. of Balsas, Osgood & Anderson 58 (F); El Puquio, Guzmango, Sagdstegui 3917 (GH). AMAZONAS Poma Cocha, Steere (GH). SAN MARTIN: Tarapoto, Spruce 4664 (K) near Tingo Maria, Allard 21133 (US). LIBERTAD: Samne, prov Angulo 1426 (GH). ANCASH: Llamac, Cerrate 2367 (USM); Huasta, Cerrate 2442 (GH, USM). HUANUCO: Mito, Macbride & Featherstone 1435 (F, GH, US); Carpish, Coronado 80 (GH, UC, US). LIMA: San Buenaventura, Pennell 14559 (F, GH, NY, PH, US). JUNIN: La Merced, Soukup 2570 (F, US, USM), 2571 (Us, USM) ; 12 km, sw. of San Ramén, Tryon & Tryon 5446 (BM, F, MO, U, US, USM). HUANCAVALICA: a 4 km. de Conaica, Tovar 979 (GH, USM). APURIMAC: Quisapata, Vargas 8898 (uc); Chincheros, West 3694 (MO, UC). CUZCO: Potrero, near Quilla- bamba, Tryon & Tryon 5368, 5376 (BM, F, GH, MO, U, US, USM) ; Saxi- huaman, Coronado 157 (GH, UC). PUNO: Juli, near Lake Titicaca, Shepard 14 (NY). * . * ’ 34. Adiantum Poiretii Wikstr. Vet. Akad. Handl. 1825: 443. 1826, based on Adiantum crenatum Poir. in Lam. Encycl. Suppl. 1: 187. Sept. 1810, not Willd. March, 1810. Type: Tristan d’Acunha, Petit-Thouars, (Herb. Jussieu no. 1427) P! Maps 33-34. Map 33, Adiantum Raddianum. Map 34, A. Poiretii var. Poiretii. 172 ROLLA TRYON Rhizome slender, rather long-creeping, with tufts of petioles borne at intervals; leaves ca. 20-50 cm. tall, petiole light or dark reddish- brown to atropurpureous, glabrous or slightly scaly at the base, rachis similar, glabrous or rarely slightly ceraceous; lamina elongate-deltoid, to broadly ovate-deltoid, tripinnate (rarely bi- or quadripinnate), pin- nae stalked, ultimate segments cuneate- to subcordate-flabellate, mostly suborbicular, rather symmetrical, glabrous or occasionally ceraceous or glandular-pubescent beneath, not articulate, the color of the apex of the stalk passing into the base, sterile margins with each vein ending in a more or less well developed sinus; sori few, oblong to long-lunate, or the smallest ones roundish. The many orbicular to suborbicular segments with oblong to lunate sori and the slender, long-creeping rhizome with loose clusters of petioles borne at intervals distinguish this Species from the previous one; it differs from the next, Adiantum Orbignyanum, by the articulate segments and orbicular to suborbicular sori of that species. Adiantum chilense differs principally in having the rhizome scales and those of the petiole bases entire while in A. Poiretii they are freely to rather slightly ciliate. Pichi-Sermolli (Webbia 12: 693-695. 1957) considers the Tristan d’ Acunha plant to be a distinct species and restricts the name A. Poiretii to it. The widespread tropical species, then, bears the name A. thalictroides Schlecht. However; until a thorough study is made of the variations of such 4 widespread species as A. Poiretii, it does not seem advisable to accept the validity of a local segregate. Adiantum sulpuwreum does not seem to be specifically separable because it agrees with A. Poiretii except for the characters of indument. Adiantum Poiretii commonly has yellow wax among the sporangia and this condition grades into the one where the wax is distributed over the segment surface. Mexico and West Indies to Chile and Argentina; Old World. ; Wooded hillsides, open woods, thickets, lomas and a varl- ety of rocky habitats, Amazonas to Arequipa and Puno, 400-4000 m. 34a, Adiantum Poiretii var. Poiretii. Fic. 129, Map 34. Segments glabrous or often with yellow wax among the sporangi® FERNS OF PERU 173 Mexico and West Indies to Bolivia and Uruguay. Amazonas to Puno, 1500-4000 m. Selected specimens: AMAZONAS: Conila, Soukup 4166 (US); Chacha- poyas, 1838, Mathews (K). LIBERTAD: between Huamachuco and Caja- 2371 (GH, USM); Huasta, Cerrate 2465 (GH, USM). HUANUCO: Muja, Macbride 3932 (F, US); Mitotambo, Ferreyra 10382 (GH, USM). LIMA: hurin, Ferreyra 5356 (GH, USM); Tupe, Cerrate 1073 (GH, USM). JUNIN: 10 km. east of Huancayo, Tryon & Tryon 5468 (BM, F, MO, U, US, USM); Huacapistana, Ferreyra 11303 (GH, USM). HUANCAVELICA: Andaimarca, Tovar 1816 (GH, USM); Saleabamba, Tovar 8624 (GH, USM). AYACUCHO: between Huanta and Rio Apurimac, Killip & Smith 22320 (F, NY, US). CUZCO: near town of Machu-Picchu, Tryon & Tryon 5403 (BM, F, MO, U, US, USM); Paucartambo, Vargas 4359 (MO, UC), Biies 4339 (US). PUNO: Salcedo, Soukup 1 (F, GH, UC, US). AREQUIPA: slopes of Chachani, Sandeman 3828 (K). 34b. Adiantum Poiretii var. sulphureum (Kaulf.) Tryon, Amer. Fern Jour. 47: 139. 1957. Adiantum sulphureum Kaulf. Enum. ge 207. 1824. Type: Chile, C isso, LE! photo GH; isotype: P! pho Adiantum Williamsii momo Gard. hc 10: 45, f. 4. 1878. Type: Mts. of Peru, 12,000 ft., Willia The under surface of the segments is sparingly to abundantly covered with yellow wax and also with a few to many, short to moderately long, gland-tipped trichomes. Peru to Argentina and Chile. Lomas and rocky hillsides, Ancash, Lima and Arequipa, 600-4000 m. Specimens seen: ANCASH: Chiquidn, Cerrate 789 (USM). LIM Chicla, 1882, Ball (GH) ; km. 75, Carretera Central, Saunders 251, 258 (B UIPA: Loma of Atiquipa, Coronado 31 (GH, UC, US); Are- quipa, Pennell 13189 (F); 14 km. s. of Arequipa, Eyerdam & Beetle 22123 (GH, UC). 34c. Adiantum Poiretii var. hirsutum (Hook. & Grev.) Tryon, Amer. Fern Jour. 47: 141.1957. Fic. 130. Adiantum chilense var. hirsutum Hook. & Grev. Ic. Fil. 2: t. 173. 1830. Type: Chile, Gillies, K! photo GH Adiantum. glanduliferum ear Hort. Berol. 2: 18. 1833. Type: Chile, Poeppig; (Hort. Berol. 1846) B Adiantum Poiretii f. “penal (Hook. & Grev.) Hicken, Rev. Mus. La Plata 15: 261. 1908. Adiantum Weatherbyanum Espinosa, Bol. Mus. Nac. ( ir ea Chile) 15: 93. 1936. Type: I. M. Johnston 5307, SGO; isotype: G 174 ROLLA TRYON Fics. 129-133. Fig. 129. Adiantum Poiretii var. Poiretii: A, fertile lower pinna, X x, Peru, Hunnewell 15834, GH; B, sterile ultima t, X 144, idem. Fig. ee A. Poiretii var. hirsutum: sterile ultimate segment, lower surface, X 1%, Peru anum: A, fertile pinna, X %, Peru Killip & Smith 25846, cu; B, sterile pinna, : - Fig. 183. A. imbricatum: A, fertile upper pinna, X %, Peru, Biles 1305, US; B, fertile Pinnule, x 1%, idem. FERNS OF PERU 175 The under surface of the segments has long, gland-tipped trichomes and no wax. Peru to Chile Lomas, Arequipa, 400-600 m. Specimens seen: AREQUIPA: Loma de Capac, Coronado 42 (GH, UC, Us) ; Loma de Atiquipa, Coronado 33 (UC, US). 35. Adiantum Orbignyanum Kuhn, Linnaea 36: 78. 1869. Lectotype: Bolivia, Mandon 52, B! photo GH. FIG. 131, Map 35. Rhizome slender, long-creeping, often branched, with the petioles spaced or borne in tufts at intervals; leaves ca. 15-40 cm. tall, petiole light or dark reddish-brown to atropurpureous, glabrous or slightly scaly at the base, rachis similar, glabrous; | lanceolate to nar- rowly ovate-deltoid, bipinnate to rarely tripinnate, pinnae stalked, ultimate segments cuneate-flabellate to suborbicular, rather symmetri- cal or not, glabrous beneath, articulated at the junction with the stalk, the apical portion of the stalk enlarged, its color abruptly distinct from that of the segment base, the segment deciduous with a sharp clean break, sterile margins with each vein ending in a usually well de- veloped sinus; sori few to several, roundish to orbicular-reniform. This species is clearly marked by its articulate segments; the apical portion of the segment-stalk is enlarged and the segments are decidous at that point. In addition it may be distinguished from species of similar appearance by the flabellate-orbicular to flabellate-cuneate segments, the orbic- ular to suborbicular sori and the inner upper pinnules especially of the upper pinnae that overlie the rachis. The rhizome is similar to A. Poiretii in being slender and long- creeping with loose clusters of petioles at intervals, and with scales that are slightly ciliate. Peru and Bolivia. Most common in open rocky places, Cajamarca to Cuzco, 2700-4000 m. Selected specimens: CAJAMARCA: San Miguel, July 30, 1952, A. Dias (USM). HUANUCO: Pachachupam, Feb. 1940, Ridowtt (USM). JUNIN: below Palea, Correll & Smith P766 (GH). APURIMAC: Cachora to Huill- cayoce, Vargas 9104 (UC). CUZCO: Yucay, Herrera 714, 718, 1198 (Us), Coronado 146 (GH, UC); San Sebastian, Pennell 13621 (¥, GH, NY, PH). 36. Adiantum Ruizianum KI. Linnaea 18: 551. 1845. Type: Peru, Ruiz 26,8! Fie. 132. Adiantum Veitchianum soil Gard. Chron, 1868: 1090. Type: Mujiia, Peru, Pearce, K! photo GH 176 ROLLA TRYON Adiantum Steerei Harr. Jour. Linn. Soc. Lond. 16: 34. 1877. Lecto- type: Poma Cocha, Peru, Steere, K!; isotype: GH! Mo! Us! Adiantum microsorium C. Chr. Ind. Fil. 30. 1905, based on A. Veit- chianum but an unnecessary new name as Ballard in Kew Bull. 1954 (4): 560 has pointed out. spaced. Most of the material is 1-pinnate with rather large reni- form to flabellate pinnae on long slender stalks ; sometimes the lamina is bipinnate at the base, or even above, and then the segments may be smaller, elongate-flabellate and with MPs 35-36. Map 35, Adiantum Orbignyanum. Map 36, A. digitatum. FERNS OF PERU 177 short stalks. The rhizome is slender and long-creeping; occasionally the petioles are loosely clustered. The next species, A. imbricatum, is closely related and the differences are discussed under it. Adiantum Ruizianum has frequently been confused with A. grossum, from which it differs in having orbicular sori rather than long-oblong or lunate ones. Also, the sterile margins have the veins ending in a sinus of (or slightly prolonged beyond) the slightly, if at all, toothed margin, while in A. grossum the veins run to the prominent teeth of the sharply serrate margin. Endemic to Peru. amp or shaded rocky places, Amazonas, Hudnuco and Junin, 1600-2900 m. (GH) ; Piedra Grande, near Rio Santo Domingo, Macbride 3709 (F, Us). JUNIN: Dos de Mayo, Killip & Smith 25846 (BM, F, GH, NY, US); Huca- pistana, Tryon & Tryon. 5434 (F); Pichita Caluga, Walden 68 (BM), Gascuyne-Cecil 109 (BM). 37. Adiantum imbricatum Tryon, Amer. Fern Jour. 47: 142, t. 15. 1957. Type: La Tranca, Cuzco, Peru, Biies 1377, Us!; isotype: Cuz! F! GH! Fie. 133. Rhizome not seen; leaves ca, 12-30 cm. tall, petiole atropurpureous, brownish short-pubescent or glabrate, rachis similar, brownish short- pubescent; lamina long-oblong to elongate-triangular, bipinnate, basal pinnae stalked, those above short-stalked, the innermost pinnule over- lying the rachis, pinna axes brownish short-pubescent, ultimate Segments broadly cuneate-flabellate to suborbicular, rather to quite Symmetrical, mostly imbricate, pubescent beneath, subarticulated at the junction with the stalk, ‘the apical portion of the stalk slightly enlarged, its color abruptly distinct from that of the segment base, sterile margins not seen; sori several to numerous, roundish to orbicu- lar-reniform, often closely spaced. This species is related to A. Ruizianum but is amply dis- tinct in a number of characters. The dark color of the Segment stalk stops abruptly at the base of the segment, while in A. Ruizianum it enters the base of the segments; the under surface of the segments, the segment-stalks and the rachis are pubescent while in A. Ruizianum they are glabrous. In addition the segments have a strong tendency 178 ROLLA TRYON to be imbricate and their stalks are shorter than in A. Ruizianum and the lamina is often more fully bipinnate. The somewhat similar A. sessilifolium has lunate sori and many of the segments are nearly sessile and the indu- sium is pubescent while A. imbricatum has orbicular sori, short-stalked segments and a glabrous indusium. Endemic to Peru. Rather dry, rocky places, Cuzco, 1600-2000 m. Specimens seen: cuzco: Puente de Collpafii, Biies 1303 (us); La Tranca, Rio Mapillo, Biies 1305 (us), 1377 (CUZ, GH, F, US). 38. Adiantum digitatum Hook. Sp. Fil. 2: 38. 1851 (Presl, Tent. Pterid. 159. 1836, nomen nudum). Type: Brazil, Sello, K! photo GH. Fic. 184, Map 36. Adiantum speciosum Hook. Sp. Fil. 2: 45, 't. 85C. 1851. Lectotype: Sasaranga, Ecuador, Seemann 952, K! photo GH. Adiantum palmatum Moore, Gard. Chron. 1877: 40, f. 5. Type: Peru, Roezl, ex hort. B. S. Williams, K! photo GH. Rhizome moderately stout, long-creeping; leaves ca. 45-150 em. tall and subscandent, petiole light to dark reddish-brown, tawny short- pubescent or glabrate, sometimes slightly scaly toward the base, rachis similar but not scaly; lamina deltoid, tripinnate or less often bipinnate, pinnae stalked, the axes tawny short-pubescent or rarely glabrate, ultimate segments more or less suborbicular, deeply cleft into about 4-7 spreading lobes, quite symmetrical or not, pubescent beneath, rarely glabrate, articulated at the junction with the stalk, the apical portion of the stalk definitely or hardly enlarged, its color abruptly distinct from that of the segment base, sterile margins with each vein ending in a tooth; sori few to several, oblong, straight or nearly so. Adiantum digitatum is a distinctive species with its flab- ellate, deeply cleft segments and frequently much elongated leaf that is up to 1.5 m. in length. Most of the axes, espe cially those of the pinnules, are at right angles to the axis that bears them. Ecuador to Argentina, Uruguay and Brazil. In rocky places or at the base of rocks, in woods or brushy hillsides in soil, Tumbes to Puno, 400-4000 m. In the north- ern half of Peru this species is frequent on the lomas and characteristic of them. Selected specimens: TuMBES: between Cancho and Cotrina, Coronado 217 (GH, UC). PruRA: w. of Canchaque, Stork 11396 (GH, Us); Cam chaque, Ferreyra 10894 (GH, USM). LAMBAYEQUE: Olmos to Jaén, Correll & Smith Ps29 (GH). CAJAMARCA: El Puquio, Guzmang% FERNS OF PERU 179 Sagdstegui 3915 (GH); entre Cascas y Contumaza, Lopez et al. 3689 (GH). LIBERTAD: Lomas Campana, Coronado 280 (GH, UC); Lomas de Viru, Coronado 287 (GH). ANCASH: Huasta, Cerrate 2478 (USM). LIMA: Lomas de Pativilca, Coronado 306 (GH, UC) ; Lomas de Chancay, Ferreyra 8700 (GH, USM); above San Bartolomé, Ferreyra 9730 (GH, USM). JUNIN: 10 km. below Palea, ‘Walden 3, 6 (BM). CUZCO: Yucay, Herrera 1361 (Us); Vileabamba, Vargas 4018 (UC, US). PUNO: Olla- chea, Vargas 6917 (MO, UC, US). Fics. 134-135. Fig. 134. Adiantum digitatum: A, base of fertile pinna, X %, Peru, Vargas 4018, us; B, sterile ultimate segment, X 144, Peru, Vargas 6917, ‘MO. Fig. 135. A. deflectens: A, fertile lamina, X %, Peru, Vargas 1658, GH; B, sterile pinna, idem. . ’ 180 ROLLA TRYON 39. Adiantum deflectens Mart. Ic. Crypt. Brasil. 94. 1834, Type: San- tarém, prov. Para, Brazil, Martius. Fic. 135. nearly or quite symmetrical, widely pinnae stalked, ultimate segments (pinnae) more or less cleft into 2-7 close lobes, glabrous, articulated at the abruptly distinct in color from the segment base, sterile margins finely and sharply serrate, with each vein ending in a tooth; sori few, oblong to linear, nearly straight to long-arcuate. This is one of the most distinctive of the Peruvian species. The lamina is 1-pinnate; the veins end in a tooth on the sterile margins; there are projecting lobes on one or both sides of the sori and the rhizome is small and short and bears slender petioles. A number of names have been applied to this, or closely related, Species, and the proper one will remain in doubt until a study is made of the whole group. Adiantum delica- tulum Mart. (Ic. Crypt. Brasil. 93, t. 56, fig. 2. 1834) and Adiantum dolabriforme Hook. (Ic. Plant. t. 191. 1837) are Mart., Adiantum fitiforme Hook., Adiantum phillipense L. Adiantum lunulatum Burm., Adiantum flagellum Fée and Adiantum subaristatum Fée. Guianas to Colombia, to Peru, Brazil and Paraguay; to Guatemala. Damp, shady soil, Cuzco, 820-1700 m. Specimens seen: cuzco: Hacienda Sahuayaco, (prov. Convencién), Biies 833 (us), Vargas 1658, 1660 (GH) ; Puente de Collpafi, Biies 129 af (US) ; Machu-Piechu to Quillabamba, Mexia 8088a (GH, MO, UC, US)} Yanayaco Grande, Biies 1035 (Us) ; Quellouno, prov. Convencién, Var- 94s 13553 (GH). -.cieeeaa FERNS OF PERU 181 TRIBE 7. PTERIDEAE 21. PTERIDIUM Scop. Fl. Carn. 169. 1760. Type: Pteris aquilina L. = Pteridium aquilinum (L.) Kuhn. Fie. 136. Terrestrial, the rhizome rather slender, extensively creeping, pubes- cent, bearing the leaves at intervals, leaves large to very large, tripin- nate-pinnatifid to quadripinnate, glabrous to usually pubescent beneath, veins free; sporangia borne on a marginal commissure connecting the vein-tips, covered by a fertile indusium which is formed from the re- flexed, modified margin, (a similar sterile indusium, a covering indusium variously developed, paraphyses absent. —1 ‘iieldseride species with 6 varieties in America, 2 of them in South America. Tryon, R. Revision of the genus Pteridium. Rhodora 43: 1-81, 37-67. 1941, asi Contrib. Gray Herb. 134. 1 The genus Pteridium, as presently interpreted, is repre- sented by a single widely distributed species which may be OSS Za ES S SSUKAW (Cuzeo Fic. 136. Pteridium aquilinum var. arachnoideum, Inca terraces at Macchu-Picchu ie 182 ROLLA TRYON divided into twelve geographic varieties. Only two of the varieties occur in South America; the one represented in Peru is typically South American ; the other, var. caudatum, is primarily Antillean but also occurs in Venezuela and Colombia and north to Mexico. Pteridium aquilinum (L.) Kuhn, var. arachnoideum (Kaulf.) Brade, Zeitsch. Deut. Ver. Wissen. Kunst. SAo Paulo 1: 56.1920. Fics. 136, 137, MaP 37. Pteris arachnoidea Kaulf. Enum, Fil. 190. 1824. Type: Brazil, Cha- misso, LE! photo GH. Pteridium arachnoidewm (Kaulf.) Maxon, Jour. Wash. Acad. Sci. 14: 89, 1924, Leaves to 3 m. ‘tall, petiole usually shorter than the lamina; lamina ovate-triangular to long-triangular, to quadripinnate, free lobes pres- ent along the axes between the segments, segments arachnoid-pubescent beneath, or rarely short-pubescent or glabrous, usually having a fari- naceous covering beneath the pubescence, midnerve usually with dark or bicolorous ‘trichomes and with membranous wings along it and the beers ; fertile indusium no broader than the sterile on the same seg- The small lobes distributed along the axes of the lamina, some of which may be attached to the base of the segment immediately above, but some of which are free, characterize var. arachnoideum. The related var. caudatum has the seg- ments caudate but lacks the free lobes. Some specimens of var. arachnoidewm may lack these free lobes but they have its characters of indusium and indument. The leaves of Juvenile plants have a sparse, setose pubescence quite dif- ferent from that of the leaves of mature plants. Due to the nature of the rhizome and the natural vigor, this species tends to be a bad weed in agricultural areas in many parts of the world. In Peru it is locally an important weed, for example in the Cerro Azul region, in Cajamarca and elsewhere. The deep main rhizome is widely creeping and sends branches upward that in turn produce the leaves. Latent buds will renew the growth if the area is burned, cut or plowed. South America; less common in the West Indies, Central America and Mexico. Open slopes, thickets, pastures and cleared land, Lambaye- que to Amazonas, south to Puno, 400-3000 m. FERNS OF PERU 183 Selected specimens: LAMBAYEQUE: 20 km. e. of Olmos, on road to Jaén, Correll & Smith P797 (GH). CAJAMARCA: nw. of Hualgayoc, Stork & Horton 10027 (F, UC); between Cascas and Contumaza, prov. Contumaza, Lépez et al. 3686 (GH). AMAZONAS: alrededores de Cha- chapoyas, Lépez et al. 4342 (GH); entre Ingenio y Pomacocha, prov. tegui 129 (GH). HUANUCO: 5 km. ne. of Acomayo, Tryon & Tryon 5224 eA 137 —althana Yaa we Town 138c Fics, 137-138. Fig. 137. Pteridiwm aquilinum var. arachnoideum: apex o* Pinna, X %, Ecuador, Camp E4411, cH. Fig. 138. Paesia viscosa: A, fertile pinnules, X 1, Colombia, Killip & Smith 15923, GH; B, fertile pinnules, X 1, Colombia, Daniel 660, GH; C, sterile pinnules, X 1, Colombia, Kalbreyer 1298, GH. 184 ROLLA TRYON fous F, MO, U, US, USM) ; Tingo Maria, Tryon & Tryon 5241 (BM, F, MO, US, USM). LIMA: Loma Lachay, Ferreyra 9767 (GH, USM). JUNIN ae armies aE 307 (GH, USM) ; Carpapata, Cerrate 2796 (ae UsM). AYACUCHO: Estrella, Killip & Smith 23095 (F, NY, US). CUZCO: Machu-Picchu, Soukup 189 (F, GH), Ferreyra 2705 (GH, USM); Cerr de Cusilluyoc, Pennell 13936 (F, GH, NY, US). PUNO: Tabina, Lede 2031 (B). 22. PaEsta St.-Hil. Voy. Distr. Diamans 1: 381. 1833. Type: Paesia viscosa St.-Hil. Terrestrial, the rhizome slender, long-creeping, pubescent, bearing the leaves at intervals; leaves small to large, bipinnate to tripinnate- pinnatifid, glabrate to glandular-pubescent, veins free; auate peer on a marginal commissure between the outer indusium which is formed from the reflexed, modified margin and the membranous inner (true) indusium, paraphyses absent. — 2 American species. Paesia viscosa St.-Hil. Voy. Distr. Diamans 1: 381. 1833. Type: Serra da Piedade, Minas Geraes, Brazil, St. Hilaire 2260, P!; isotype: GH! Fic. 138, Map 38. llosorus acclivis Kze. Farnkr. 2: 6. 1848. Lectotype: Prov. Merida, , Pteridium aquilinum var. arachnoideum. Map 38, P' aes Maps 37-38. Map 37 viscosa, FERNS OF PERU 185 Venezuela, Funck & Schlim 1222; isotype: GH! K! photo GH, P! Pteris scalaris Mett. Abhandl. Senckenberg. Naturf, Ges. Frankfort 2: 282 (Uber ein. Farngatt. III: 9). 1858. Type: Spec. cult. Hort. Bot. Lips. ex Colonia Tovar, Venezuela, Moritz 399, Herb. Mett. B! photo GH (Moritz 399, B! GH! K!). Pteris scalaris Moritz, Bot. Zeit. 1854: 856 and Pteris resistens Mett. Fil. Hort. Bot. Lips. 59. 1856, nomina nuda, were both also “based” on Moritz 399. Pa acclivis (Kze.) Kuhn, Festsch. 50 Jub. Reals. Berlin, (Chaetopt.) 347. 1882. Paesia sealaris (Mett.) Kuhn, op. cit. 347. 1882. teris amazonica Christ, Hedwigia 44: 364. 1905. Type: Cerro Po- nasa (“Vonasa”), Loreto, Peru, renee 1908, Ule 6899 Herb. Christ, P!; isotype: B! photo Gu, K! photo G Paesia amazonica (Christ) C. Chr. Ind. Fil. 476. 1906. Rhizome trichomes dark brown, rigid, terete or flattened in age; leaves 0.5 to ca. 2 m. long, long-petioled, lamina haga oA to Nisin eding the rachis flexuous, pinnules with the basal segment on the acroscopic side, coriaceous to herbaceous, all parts of the lamina with usually rather abundant (sometimes sparse), short, gland-tipped trichomes, sometimes also with sessile glands or with long trichomes that may be gland-tipped, or not; outer indusium thin, somewhat fim- briate or glandular-pubescent. There is considerable variation in the shape of the seg- ments, their texture, the density of the pubescence and the presence or absence of sessile glands, short gland-tipped trichomes or long gland-tipped trichomes. None of these characters affords a basis for the recognition of other taxa. The other American species, P. anfractuosa of Central America, has the basal segment on the basiscopic side of the pinnules, while P. viscosa has it on the acroscopic side. It has been suggested by C. Christensen (Ark. f. Bot. 9(11) : 18-20. 1910, that Cheilanthes glandulosa Sw. (1817) may be a Paesia. If this is true, then it would be an earlier name for P. viscosa. A study of the material at S-PA, and the original description, leads me to the conclusion that it is more likely to be a species of Hypolepis. Costa Rica; Greater Antilles; Venezuela to Colombia, south to Bolivia; Brazil. Shrubby slopes, edge of forests and rocky places in woods, Amazonas to Puno, 1400-3600 m. AMAZONAS: Molinobamba, Spee 1069 hy SAN LORETO: Ce Specimens seen: rro UAN- MARTIN: Tarapoto, Spruce 4666 (GH, K, NY, Ponasa, (between Yurimaguas and Tarapoto), Ule 6899 (B, ang H 186 ROLLA TRYON uco: Playapampa, Macbride 4501 (F, GH, US), 4511 (F, US). CUZCO: Cerro de Cusilluyoe, Pennell 14024 (F, GH); Calea, Biies 1902, 1915 (us); Valle San Miquel, Biies 2055 (US). PUNO: Tatanara, Lechler 2536 (B). 23. Loncuitis L. Sp. Pl. 2: 1078. 1753; Gen. Pl. 485. 1754. Type: Lonchitis hirsuta L. Anisosorus Maxon. Sci Surv. Porto Rico & V. I. 6: 429. 1926, (nom. superfl., illegit.). Type: the same as that of Lonchitis. Terrestrial, the rhizome thick and fleshy, creeping, pubescent, bear- ing the leaves at close intervals; leaves large to very large, bipinnate- pinnatifid to tripinnate-pinnatifid, pubescent, veins free; spo: rangia borne on a marginal commissure connecting the vein-tips, covered by an indusium which is formed from the reflexed, more or less modified margin, paraphyses absent. — 1 American species. mn, R. The genera Lonchitis and Blotiella, in Tax. Fern Notes, II. Contrib. Gray Herb. 191: 98-96. 1962. Lonchitis pisces Sp. Pl. 2: 1078. 1753. Type: Martinique, Plumier, ie ‘ 20. Fig.1 teris laciniata we Sp. Pl. 5: 897. 1810. Type: Ind. Occ. Fligge oe (Herb. Willd. no. 20013) B! photo Gu. Pteris lonchitoides Desv. Mém. Soc. Linn. Paris 6: 301. 1827. Type: Pteris hirsuta (L.) J. "hi Jour. Bot. Hooker, 4: 165. 1841, not Poir. Lam. Encyel. 5: 719. 180 Anisosorus hirsutus ers Maxon, Sci. Surv. Porto Rico & V. I. 6: 429. 1926. Mexico and Central America: Venezuela to Colombia and south to Bolivia. In forests, San Martin and J unin, 1300-1600 m. Specimens seen: SAN MARTIN: Tar Spruce 4667 (GH, K). JUNIN: Yapas, Pichis Trail, Killip & Smith 25488 (oH, Us); 1a Merced, Macbride 5652 2 (us). 24.HIstTiopTeris (Ag.) J. Sm. Hist. Fil. 294, 1875. “4 Pteris sect. Histiopteris Ag. Rec. Gen. Pterid. 76. 1839. Type: Ptert vespertilionis Labill. = = Histiopteris vespertilionis (Labill.) J. 5m. Terrestrial, the rhizome slender, long-creeping, scaly and sometimes also pubescent, bearing the leaves at intervals; leaves small to very FERNS OF PERU 187 large, bipinnate-pinnatifid to tripinnate-pinnatifid, glabrate, veins free to anastomosing; sporangia borne on a marginal commissure connect- ing the vein-tips, covered by an indusium which is formed from the reflexed, modified margin, paraphyses present. — 1 species in America. Histiopteris incisa (Thunb.) J. Sm. Hist. Fil. 295.1875. Fic. 140. Pteris incisa Thunb. Prod. Fl. Cap. 171. 1800. Type: Cap. Bon. Spei, 1774, Thunberg; isotype: S-PA! Rhizome scales brown, mostly long-triangular, more or less clathrate, trichomes, when present, brownish, long and rather soft; leaves ca. 0.5-3 m. (or more) long, petiole short to long, scaly toward or at the base with scales similar to those on the rhizome, lamina ovate-triangu- lar to long-triangular, pinnae opposite, the basal pinnules reduced, especially toward the base of the lamina of large leaves, where they become stipule-like, pinnules often opposite, glaucous beneath and labrous or with scattered, brownish, large trichomes; indusium mem- branous, glabrous, entire to crenulate. Fics, 139-140. Fig. 139. Lonchitis hirsuta: A, fertile pinnules, indument and vena- tion, X 34, Venezuela, Fendler 100, GH; B, petiole section, X 2, Honduras, Steeves & Ray 385 GH. Fig. 140. Histiopteris incisa: A, sterile, pinna, X %, Colombia, Killip & segment, X 14, Bolivia, Steinbach 8961, GH; C, , idem. 188 ROLLA TRYON The venation is unusually variable in this species. Most commonly it is partially areolate; less often it is almost wholly areolate and rather rarely it is wholly open, all of the veins being free. Tropical America; Old World. Shrubby slopes and dense forest, Hudnuco to Puno, 1000- 3000 m. Specimens seen: HUANUCO: Carpish, Sandeman 5193 (K); Panao, Macbride 3603 (F, US). JUNIN: Huacapistana, Weberbauer 2194 (B); Villa Amoreti, G. Kunkel 637 (GH); San Nicolas, Pichis Trail, Killip & Smith 25961 (us). cuzco: Valle de Pillahuata, Paucartambo, Her- rera 3337, 3341 (US). PUNO: Sandia, Weberbauer 714 (B); between Sandia and Rio Chunchusmayo, Weberbauer 1341 (B). 25. Preris L. Sp. Pl. 2: 1073. 1753; Gen. Pl. 484. 1754. Type: Pteris longifolia L. The typification of the genus has been discussed by Maxon (Jour. Bot. 61: 7. 1923). Terrestrial, the rhizome usually stout, scaly, erect and bearing the leaves in a crown or cluster, to long-creeping and bearing the leaves at intervals; leaves of medium size to large or very large, 1-pinnate to 5-pinnate, glabrous to pubescent or sparingly scaly, veins free or anas- tomosing; sporangia borne on a marginal commissure connecting the vein-tips, covered by an indusium which is formed from the reflexed, modified margin, paraphyses present.— About 50 species in South America. Maxon, W. R. Pteridophyta (Pteris), in Sci. Surv. Porto Rico & V- I. 6: 432-436. 1926. The genus Pteris is notable for the manner in which the leaves, especially their basal pinnae, are branched. The eX- tent of the branching may change on the same plant from the small leaves to the largest ones. The type of branching 1s difficult to use as much as its importance probably warrants, because it is often insufficiently known. Most herbarium Specimens consist either of small leaves or of fragments 0 large leaves. Another unusual feature in Pteris is that many species have awns on the upper side of the costa of the pel- ultimate segments. In these species there is a ridge on each side of the costa which is interrupted at each costule and 1s prolonged into a free awn. The species predominantly grow in forests or in more open, shrubby or rocky habitats on the eastern slopes and valleys of the Andes. One species, P. coriacea, grows in the FERNS OF PERU 189 Andes proper, and a few others also grow in the Amazon basin. Pteris vittata and P. critica are local escapes from cultivation. KEY TO THE SPECIES a. Veins free (rare areolae may be present). b. b. All pinnae, or those above the base, entire. c. S ina 1-pinnate, pinnae 10 to many pairs, cordate to sub- cordate, the basal ones reduced. 1. Pteris vittata c. Lamina with the basal pinnae usually with a single (rarely more) large pinnule, rarely entire, pinnae one to five pairs, cuneate or decurrent, the basal largest or nearly SO, ......ee 2 ris cretica b. All pinnae pinnatifid or more complex (rarely reduced apical ones entire). d. d. Basal pinnae 1-pinnate (or more) beyond the basal pinnules, the pinnules stalked or sessile (constricted at least on the acro- scopic side); penultimate segments mostly with the basal in- ferior segment reduced, or at least shorter than the superior. .. e. e. Penultimate segments 1-pinnate, except sometimes toward the apex; ultimate segments narrowed at the base, at least on the acroscopic side. . f. Basal pinnae the largest but each much smaller than the remaining portion of the lamina, ultimate segments entire. -'& Pte obed. e. Penultimate segments deeply pinnatifid, or 1-pinnate only at the base; ultimate segments broadest at the base. ....... seeeeees g g. Costa of the penultimate segments (and other axes) smooth beneath, with deciduous trichomes (rarely scales); seg- ments herbaceous to herbaceous-coriaceous. FSi ime ore ae 6. Pteris deflexra to subpinnatisect beyond the one (rarely ‘two or none) basal, enlarged, inferior, pinnatifid pin- nule, the ultimate segments fully adnate or laterally joined; penultimate segments mostly with the basal superior segment reduced, or at least shorter than the inferior, or both basal ~ te segments, or 190 ROLLA TRYON h. One or two veins, especially in the basal portion of the penul- timate segment, arising from the costa of the penultimate seg- ment about half way between adjacent costules; pinnae, except the basal pair, with both basal segments reduced to about half or less the length of the longest. ............:00.40+ 8. Pteris pungens a. Venation areolate, or predominantly so, or at least a row of ior along the costa of the penultimate segments. i. Ultimate segments (usually pinnae or pinnules) stalked, a usually large (mostly 15-40 cm. long’). .......ssssssssserrrseerssesnnensenensenenens j. j. Lamina 1-pinnate (rarely the basal pinnae each with a large pinnule) ; costal areolae long, with the long axis divergent from the costa, those toward the margin progressively shorter; pinnae minutely tortuous-pubescent beneath, especially on the veins, to glabrate, the sterile margins entire. ............ 9. Pteris grandifolia . Lamina bipinnate toward or at the base, or rarely tripinnate; costal areolae short and broad, those toward the margin longer, those at the margin shorter; pinnae and pinnules minutely straight appressed-pubescent heat to glabrate the sterile margins serrate, 10, Pteris Haenkeana i. Ultimate segments joined or sessile, or lobed, often small (5-10 ORNs 09) OE ME ieee as es ks a k. ry k. Veins free except for a single row of areolae along the costa of the penultimate segments (rarely a few suet a caicbccolesoucebiveanrteee Sinitsgaeusaecs . Pteris biaurita k. Venation predominantly areolate. ..sssssssesssssseessssecssnessessseseeeeeet 1, 1, Upper side of tthe costa of the penultimate segments awned at the base of the costules; basal pinnae pinnate-pinnatifid to bipinnate beyond the hexat pinnules (except in no. 16 where they are only pinnatifid or are pinnatifid beyond the _ enlarged, basal, inferior pinnatifid pinnule). ..........-eese m. Two or more areolae, with their long axis parallel tii a costa of ‘the penultimate segments, between adjacent cos tules. 12. Pteris altissima m. A snipe areola, with its long axis parallel to the costa of the penultimate segment, een adjacent costules. ....-.- n. n. Base of the het, segments ae to cuneate, at decurrent (apical ones may be sessile). .......ss-ss o. Separate portion of the longer ultimate ane about 11% to 2%, rarely 3, times as long as broad; eooonececsesee® FERNS OF PERU 191 p. Penultimate segments with an abrupt base, not at all cuneate or decurrent and, all but the apical ones, subpinnatisect to 1-pinnate at the base with at least the basal segments slightly narrowed toward their base; ultimate segments serrate-crenulate to bluntly serrate apically. 15. Pteris livida p. Penultimate segments cuneate at the base, pinnati- partite, with the basal segments broadest at their base; ultimate segments entire to sharply serrate apically. q. q. Areolae converging on the sinus of the ultimate segments few and broad; the vein that forms the long, costal areola arising from the costule or near to it. 16. P. reticulatovenosa q. Areolae converging on the sinus of the ultimate segments numerous and very narrow; the vein that fe) well beyond the costule that is basal to it. ......0-+000 17. P. speciosa 1. Upper side of the costa of the penultimate segments not awned; basal pinnae entire to pinnatifid beyond the single enlarged, basal, inferior pinnule. r. r. Pinnae entire or irregularly lobed. 19. Pteris petiolulata s. r. Pinnae regularly lobed or pinnatifid. - 8. s. Penultimate segments hirsute on both surfaces and on he margin and especially so on the COSTAE. ....-.---+--ssseseer teris Lechleri wg s. Penultimate segments glabrous to minutely pubescent. .... i t. Lobes or ultimate segments sharply serrate at the apex; pinnae lobed to shallowly or moderately pinnati- fid. 17. Pteris speciosa Ultimate segments bluntly serrate at the apex; pinnae (except the basal pair) deeply pinnatifid. ........--s-eseee+ - ... 20. Pteris horizontalis st 1. Pteris vittata L. Sp. Pl. 2: 1074. 1753. Type: China, Osbeck, LINN 1246.3, photo a.‘ Fi. 141. Rhizome stout, short-creeping, the leaves forming a rosette in small plants, a cluster in large ones, leaves 5-150 cm. (or more) long, the petiole much shorter than the lamina; lamina 1-pinnate, narrowed to the base, pinnae at least 10 pairs in small leaves to usually numerous, the sterile margins serrate, short-petiolulate, the base cordate to sub- cordate, glabrous or nearly so beneath except for the moderately to sparingly pubescent costa, veins free. This species is very distinct from the other Peruvian ones. In the past it has been confused with Pteris longifolia 192 ROLLA TRYON (Mexico, West Indies to Brazil) but that species has articu- late pinnae, the petiolule being abruptly attached to the rachis and somewhat enlarged, while in P. vittata the petio- lule is evenly and somewhat decurrently joined to the rachis. A species of the Old World, often cultivated and sometimes becoming naturalized in tropical and subtropical America. Sea cliffs near Lima and on banks of the Rio Rimac, Lima, sea level to ca 1500 m. Specimens seen: LIMA: north of Chorillos, Stork 9367 (GH, UC) ; Santa Eulalia, Coronado 2 (GH, UC, US); Chosica, Soukup 2047 (F, US) ; Miraflores, Tryon & Tryon 5217 (BM, F, GH, U, UC, US, USM) ; km. 75, Carretera Central, Saunders 371 (BM). 2. Pteris cretica L. Mant. 130. 1767. Type: Crete, LINN 1246.7, photo A. Bre, 142, A native of the Old World and perhaps of Mexico; culti- vated in Peru and probably sometimes becoming naturalized. pecimens seen: AMAZONAS: Poma Cocha, Steere (GH). LIMA: Botanical Garden of Lima, 1954, Coronado 189 (UC). 3. Pteris coriacea Desy. Mém. Soc. Linn. Paris 6: 300. 1827. TyP?: Peru, Dombey, P! photo GH. The sheet in Herb. Desvaux is marked as the type, it contains a sterile leaf of this species and a portion of a fertile leaf of Pteris muricata; the sheet in Herb. Gen. P! contains . fertile and sterile leaf of this species. Fic. 1438. : Pteris Jamesonii Hook. Sp. Fil. 2: 198, t. 183A. 1858. Lectotype: Quito, Ecuador, Jameson, K! Rhizome moderately stout, short-creeping, leaves 30-70 cm. (or more) long, the petiole about as long as the lamina, or longer; lamina ae nate to tripinnate at the base, the basal pinnae largest, pinnae short- . definitely petiolulate, with the basal inferior pinnule usually shorter than the superior (except on the basal ones) , the costa unawned above or very bluntly awned, smooth to muricate-spiculate beneath; ultimate FERNS OF PERU 193 Fis. 141-145. Fig. 141. Pteris vittata: portion of gig & Tryon 5217, GH. Fig. Fig. 145. P. muricata: A, fertile penultimate segment, X : J0s0h/ crx Bi, cuciasls aacments anil «xis, lower warface, X 2, Eeundor, Bimbork 206, GH, 194 ROLLA TRYON segments narrowed at the base on both sides or at least on the acro- scopic side, the sterile margins entire to serrate, glabrous beneath ex- cept for the deciduously scaly costa, veins free. This species differs from P. muricata principally in its 1-pinnate rather than pinnatisect penultimate segments, and it is perhaps not wholly distinct. It is a rare species and has been frequently confused with the more common P. muricata. Ecuador and Peru. Crevices of shaded rocks and canyon walls, Ancash to Apurimac, 2800-3500 m. Specimens seen: ANCASH: cerca a Llamac, prov. Bolognesi, Cerrate 2372 (GH, USM). HUANUCO: Chasqui, Macbride & Featherstone 1755 (F, US). LIMA: between San Mateo and Parac, Coronado 312 (GH, UC)- JUNIN: Tarma, G. Kunkel 406 (GH). APURIMAC: Ampay, Vargas 1062 (GH, Us). 4. Pteris Bakeri C. Chr. Ind. Fil. 593. 1906, based on P. decomposiia Baker, Syn. Fil. ed. 2, 479. 1874, not Gaud. 1827. Type: Mujfia, Peru, Pearce, K! fragment Ny! Fic. 144. Rhizome evidently rather slender and long-creeping but perhaps stout and short-creeping in large plants, leaves 0.5 to perhaps 2 ™. long, the petiole about as long as the lamina; lamina broadly trianguU- lar, tripartite, each of the basal pinnae almost as large as the upper portion, quadripinnate at the base, bipinnate to tripinnate above the basal pinnae, pinnae petiolulate, penultimate segments with the basal inferior pinnule usually shorter than the superior, their axes bluntly to definitely awned above, deciduously scaly and long muricate-spiculate beneath; ultimate segments usually lobed, often deeply so, cuneate at bis sessile base, glabrous beneath or slightly scaly on the costa, veils ree. ; The tripartite lamina and the sessile, usually lobed, ulti- mate segments are evidently sufficiently distinctive char- acters of this species. However, the scanty material available does not allow a proper assessment of it. Peru. Forest, Hudnuco and Junin, ca. 2800 m. Specimens seen: HUANUCO: Mufia, Pearce, (K); Pozuzo, Pearce 534 a JUNIN: Carpapata, above Huacapistana, Killip & Smith 24488 Us). 5. Pteris muricata Hook. Sp. Fil. 2: 198, t. 123B. 1858. Type: Antio- quia, Colombia, Jervise, kK! Fic. 145, Map 39. FERNS OF PERU 195 Rhizome short- to long-creeping, ascending to decumbent, leaves 0.5 to 2 m. (or more) long, the petiole about as long as the lamina, or longer; lamina bipinnate to 'tripinnate at the base, tripartite or nearly so, usually 1-pinnate above the basal pinnae which are the largest and regularly 1-pinnate or with a single, enlarged, basal, inferior, 1-pinnate pinnule, penultimate segments pinnatisect or subpinnatisect (rarely 1-pinnate at the base), short-petiolulate to subsessile, the basal inferior segment usually shorter ‘than the superior, the costa bluntly to definite- ly awned above, deciduously scaly (rarely also pubescent) and shortly muricate to muricate-spiculate beneath; ultimate segments mostly as broad at the base as above, or broader, the sterile margins entire to strongly serrate, glabrous beneath except for the slightly scaly or pubescent costule, veins free. I am not certain of the validity of maintaining this species distinct from P. coriacea, or from the next one, P. deflexa. It differs from P. coriacea in its pinnatisect penultimate seg- ments, although in some cases they are 1-pinnate at the base. It differs from P. deflexa in the firmer texture of the leaf and in its muricate, rather than smooth, costae which bear scales. However, sometimes the costae are only slightly muricate and rarely they may bear trichomes, as in P. deflexa. Specimens of this species have often been identified as P. coriacea, perhaps because Hooker (Sp. Fil. 2: t. 124) con- fused the two; his plate of P. coriacea is P. muricata. Mexico to Colombia and Bolivia. Forests and among moist rocks, Hudnuco to Puno, 1700- 2400 m. Selected specimens: HUANUCO: Mito, Macbride & Featherstone 1619 (F, US) ; Huacachi, near Mufia, Macbride 4167 (F, US). PASCO: Quillesd, Soukup 3283 (GH). JUNIN: Huacapistana, Killip & Smith 24191 (US); Pariahuanca, Mathews 1295 (K); Carpapata, G. Kunkel 649 (GH). APURIMAC: Chirhuai, Vargas 2302 (US); arriba de Abancay, Ferreyra 9808 (GH, USM). CUZCO: Machu-Picchu, Tryon & Tryon 5398 (BM, F, between Ayapata and Kahualluyoc, prov. Carabaya, Vargas 10751 (GH); Tabina, Lechler 2030 (B, K). 6. Pteris deflexa Link, Hort. Berol. 2: 30. 1833. Type: Brazil, Hort. Bot. Berol. 8! photoGH. Fic. 146. Pteris polita Link, Hort. Berol. 2: 30. 1833. Type: Brazil, Hort. Bot. Berol. B! photo GH. ry stout (6-8 cm. in diameter), short- the petiole about as long tripinnate at the base, 1- Rhizome moderately to ve creeping or ascending, leaves 0.5-3 m. long, as the lamina, or longer; lamina bipinnate to 196 ROLLA TRYON pinnate to bipinnate above the basal pinnae which are the largest and may have one or two enlarged, basal, inferior, 1-pinnate pinnules, pen- ultimate segments pinnatipartite to usually pinnatisect, petiolulate to subsessile, the basal inferior segment usually shorter than the superior, the costa awned above, smooth beneath and deciduously pubescent (rarely with scales) ; ultimate segments as broad at the base as above, or broader, the sterile margins sharply to coarsely serrate, glabrous beneath or the costule slightly pubescent or scaly, veins free. This species is close to the previous ene, P. muwricata, and the differences are discussed under that species. Greater Antilles; South America. In dense forest, San Martin to Ayacucho, 800-1800 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce 4326 (BM, GH, XK, US). LORETO: Sinchono, prov. Coronel Portillo, Aguilar 898 (GH, USM). JUNIN: La Merced, Soukup 1028 (F), Killip & Smith 23902 (us) ; Hua- capistana, Ferreyra 503 (GH, USM); Chanchamayo valley, Schunke 2 (F), 1386 (F, US). AYACUCHO: Cearrapa, between Huanta and Rio Apurimac, Killip & Smith, 22408 (F, GH, US). 7. Pteris quadriaurita Retz. Observ. Bot. 6: 38. 1791. Type: Ceylon, Konig. Fic. 147. Pteris edentula Kze. Linnaea 9: 75. 1834, as Pteris biaurita var. ? P. edentula. Type: Pampayacu, Peru, Poeppig 223 (Diar. 1106); iso- type: B! photo cu. Rhizome moderately stout, erect or decumbent, leaves about 0.5-2 oe long, the petiole about as long as ‘the lamina, or longer; lamina bipin- nate at the base, 1-pinnate above the basal or lower pinnae, the basal pinnae about the same length as those above, each with an enlarged, basal, inferior, pinnatifid pinnule, (rarely with a second such pinnule, ments as broad at the base as above, or broader, entire, glabrous oF slightly and usually minutely pubescent beneath, mostly on the costule, veins free, The suggestion of T. G. Walker (Evolution 12: 91. 1958) “... that the name P. quadriaurita should not in future be used in any sense other than that of Retzius sens. strict. ..- (for the plant of southern India and Ceylon) is based on 4 cytological study but can hardly be accepted as a taxonomi¢ judgment, Tropical America ; Old World. FERNS OF PERU 197 Dense forests, woods and thickets, in soil or in rocky places, Lambayeque to Cuzco, 700-2200 m. Selected specimens: LAMBAYEQUE: km. 32 on road from Olmos to Jaén, Correll & Smith P830 (GH). HUANUCO: Rio Huallaga, below Rio Santo Domingo, Maebride 4233 (F, US); Pampayacu, Macbride 5043 (F, US). JUNIN: La Merced, Killip & Smith 23473, 23686 (F, US), Cer- rate 2832 (GH, USM), Soukup 3409 (US); Huacapistana, Aguilar 508 (GH, USM). cuzco: Potrero, Tryon & Tryon 5367 (BM, F, GH, U, US, usM), Vargas 8236 (uc); Rio Lucumayo, Urubamba valley, Sandeman 8546 (K). 8. Pteris pungens Willd. Sp. Pl. 5: 387. 1810. Type: Hispaniola. Fig. 148. long, the petiole about as long as ‘the lamina, or longer; lamina bipin- beneath or nearly so, veins free. Pteris pungens is rather similar to the previous species, P. quadriaurita, from which it differs especially in the veins that arise from the costa about half way between adjacent costules (that is, more or less beneath the sinus). These occur particularly toward the base of the penultimate seg- ment. In P. quadriaurita all of the veins arise from the costules or very near to them. Also, both basal segments of the penultimate segments are reduced in P. pungens while in P. quadriaurita only the superior one is reduced. Mexico to Panama; West Indies; British Guiana to Colom- bia, south to Bolivia. Specimens seen: SAN MARTIN: Tarapoto, Spruce 4750 (K), Woy est ski 35215 (UC). LORETO: Yurimaguas, L. Williams 3865 (F, US), “ wp & Smith 28012 (us); Balsapuerto, Killip & Smith 28517 (GH, US). HUANUCO: Ganzo Azul, Rio Pachitea, Sandeman 3361 (K). 9. Pteris grandifolia L. Sp. Pl. 2: 1073. 1753. Type: Dominica or Martinique, LINN 1246.1, photo A. Fic. 149, Map 40. “ ney Pteris grandifolia var. campanae Rosenst. Fedde Rep. aigyre toa 291. 1909. Type: Tarapoto, (monte Campana) Peru, Spruce ; . . Rhizome about 1 cm. in diameter, rather extensively creeping, leaves 198 ROLLA TRYON AAAs ray TL) y KY y ae Q he VM YU MEE, KS G | ark SS SS i, 48 Al i oo pinna, X 1, idem; ©, portion of sterile pinna, X 1, Peru, Tryon & T ith ”, GH. Fig. 150. P. Haenkeana: A, fertile pinna, X 14, Colombia, Killip & xi 16510, GH; B, portion of sterile pinna, X 1, idem. FERNS OF PERU 199 1-5 m. long, the petiole about as long as the lamina, or longer; lamina 1-pinnate (rarely bipinnate at the base with each basal pinna with an enlarged, basal, inferior, entire pinnule), the lower pinnae somewhat shorter than those above, pinnae simple entire, (except rarely the basal), long- to short-petiolulate, minutely tortuous-pubescent beneath, especially on the veins, to glabrate; venation areolate, the costa] areolae long, with their long axis divergent from the costa, those toward the margin progressively shorter. Small portions of a leaf may be distinguished from the next species, P. Haenkeana, by the pattern of the venation mentioned in the key and by the entire, rather than serrate, sterile margins. West Indies ; Mexico to Peru. In forest or more commonly, thickets, forest borders, and along stream banks, Amazonas to Junin, 135-1100 m. Selected specimens: AMAZONAS: entre Aramango y Montenegro, prov. Bagua, Lépez et al. 4214 (GH). SAN MARTIN: Chazuta, Rio Huallaga, Klug 4013 (F, GH, K, UC); Tarapoto, Spruce 4668 (K), 4751 (BM, K). LORETO: Rio Marafion, above Pongo de Manseriche, Mexia 6182 (F, GH, K, UC, US); Puerto Arturo, Yurimaguas, L. Williams 5109, 5348 (F), Killip & Smith 27852 (us). HUANUCO: Tingo Maria, Tryon & Tryon 5287 (BM, F, GH, U, US, USM) ; Puente Durand, 1945, Vargas (GH, USM). PASCO: Oxapampa, Soukup 2349 (F, GH). JUNIN: Puente Perené, Coro- nado 277 (uc); La Merced, Killip & Smith 23693 (F, GH, US), Cerrate 2825 (GH, USM). 10. Pteris Haenkeana Pres], Rel. Haenk. 1: 55. 1825. Type: Haenke, PR; authentic specimen: Colombia, Linden 1028, w! det. Presl. Fic. 150, Map 41. Pteris ampla Kze. Linnaea 9: 74, 1834. Type: Pampayacu, Peru, 1829, Poeppig (Diar. 1154); isotype: K! Rhizome about 1 cm. in diameter, extensively creeping, leaves 1-2 m. long, the petiole about as long as the lamina, or longer; lamina fully The differences between P. Haenkeana and the previous species, P. grandifolia, are discussed under that species. It is closely related to, if not conspecific with, Pteris mexicana of Mexico and Central America. Colombia to Bolivia and Brazil. ROLLA TRYON 200 . Map 40, P. grandifolia. Map 41, P. = rica Maps 39-42. Map 39, Pteris mu keana, Map 42, P. altissima. FERNS OF PERU 201 In forests, San Martin, to Cuzco, 600-1400 m. Specimens seen: SAN MARTIN: Juan Jui, Klug 4256 (F, GH, K, UC, us); Zepelacio, near Moyobamba, Klug 3492 (F, GH, K, us); Tambo Maria, Stiibel 1108 (B). HUANUCO: Pampayacu, Kanehira 147 (GH, us), Poeppig 1154 (K); Cushi, Bryan 724 (F, GH, US). JUNIN: La d . Schunke 27,70 (Ff). AYACUCHO: Aina, between Huanta and Rio Apuri- mac, Killip & Smith 22717 (F, GH, US). CUZCO: Ceochayoc, Biles 1726 (US). 1. Pteris biaurita L. Sp. Pl. 2: 1076, 1753. not Martinique, Jamaica or Hispaniola, LINN 1246.19, photo A. Fia. 1 izome moderately stout, erect to decumbent, leaves about 0.6-1 m. long, the petiole about as long as the lamina, or longer; lamina bipin- nate at the base, 1-pinnate above the basal pinnae which are about the same length as those above and have a single enlarged, basal, inferior, pinnatifid pinnule, pinnae petiolulate to sessile, the costa awned above and glabrous or minutely appressed-pubescent beneath, deeply pinnati- fid (except for the pinnule on ‘the basal pinnae), with the basal inferior segment usually longer than the superior; ultimate segments broadest at the base, glabrous or minutely appressed-pubescent beneath, the sterile margins entire to subcrenulate, veins free except for a single costal areola between adjacent costules, or occasionally a few other basal veins joined. The venation is characteristic of this species and easily separates it from the others in Peru. Small sterile leaves (the lamina less than 10 cm. long) have the basal pinnae with the enlarged pinnule typical of adult leaves but the re- mainder of the lamina is only pinnatifid. Mexico to Panama; West Indies; Guianas to Colombia to Peru and Brazil; Old World. In forest, in clearings and along forest borders, Loreto to Cuzco, 400-1500 m. Specimens seen: : Boquerén Padre Abad, Allard 22115 (Us). HUANUCO: Pozuzo, pre 4596 (F); Tingo Maria, Tryon & Try 5246 (GH, U, US). JUNIN: Satipo, Ridoutt (US) ; San Ramon, Schunke A180 (us); Rio Paucartambo, near Perené Bridge, Killip & Smith 25277 (F, US); Chanchamayo valley, Schunke 24, 1001 (F). CUZCO: Chacanares, prov. Convencién, Vargas 13680 (GH). 12. Pteris altissima Poir. Lam. Encycl. 5: 722. 1804. Type: Porto Rico, Ledru (Herb. Lamarck) P! photoGH. Fic. 152, Map 42. Pteris elata Ag. Rec. Gen. Pterid. 63. 1839. Type: Panama, Cuming 1267, K!; isotype: GH! 202 ROLLA TRYON Pteris Kunzeana Ag. Rec. Gen. Pterid. 62. 1839. Syntypes: Peru, Mathews 1802, Kk! fragment Ny!; Peru, Poeppig 225, B! photo GH. (Two other specimens cited, not seen). Rhizome erect in young plants, becoming short- to rather long-creep- ing, leaves 0.8-2.5 m. long, the petiole about as long as the lamina, or longer; lamina bipinnate, the basal pinnae much the largest, pinnae petiolulate to subsessile, penultimate segments with the costa awned than the superior, or not; ultimate segments broadest at the base, minutely appressed-pubescent beneath, the sterile margin entire to sharply serrate, venation areolate, two or more costal areolae with their long axis parallel to the costa, between adjacent costules. Pteris altissima is a variable species, especially in charac- ters of the texture, shape and disposition of the ultimate segments. I believe that Pteris sclerophylla Sod. and P. esmeraldense Sod. are synonyms but I have not seen authen- tic material of them. The leaves of young plants have the lamina more complex in relation to their size, than leaves of old plants. Small leaves are bipinnate with narrow segments, older leaves on the same plant are progressively less complex and have broader segments. ; Mexico to Panama; West Indies; Guianas to Colombia, south to Bolivia; Brazil. In woods and forests, Amazonas to Cuzco, 100-1500 m. Selected specimens: AMAZONAS: Rio Utcubamba, 40 km. s. of mp Grande, Hutchison 1486 (GH). SAN MARTIN: Tarapoto, Spruce 406 (BM, K); Zepelacio, near Moyobamba, Klug 3704 (F, GH, K, us). RETO: mouth of Rio Santiago, above Pongo de Manseriche, M ‘ey 6125 (F, GH, K, UC, US), 6125a (GH, K, UC, US) ; Iquitos, Killip & Smit 27440 (US). HUANUCO: Tingo Maria, Tryon & Tryon 5229 (BM, F, GH, Soukup 1098 (F), Killip & Smith 23684 (us). ayacucno: near Kim 13. Pteris propinqua Ag. Rec. Gen. Pterid. 65. 1839. Lectotype: Jamaica, Bancroft, Kk! Fic. 153. Rhizome moderately stout to stout, erect, leaves 0.5-1.5 m. long, the petiole about as long as the lamina, or longer; lamina bipinnate, ~ basal pinnae the largest, pinnae petiolulate to subsessile, penultimate Segments with the costa awned above and glabrous or minutely ee pressed-pubescent beneath, deeply pinnatifid to pinnatipartite, W? FERNS OF PERU 203 the basal inferior segment shorter than the superior, or not, the base narrowly decurrent on its stalk or onto the next axis; ultimate seg- iv with its long axis parallel to the costa, between adjacent costules. Among the species with areolate veins, P. propinqua is distinctive in having the base of the penultimate segments decurrent on the stalk, or onto the next axis. ‘a daw Wns | LAI VD. If NN ee N y Fics. 151-154. Fig. 151. Pteris biaurita: A, fertile pinna, X %4, Mexico, Langlassé 707, GH; B, portion of fertile pinna, X 114, Peru, Tryon & Tryon 5246, GH. Fig. 152. P. alti. fertile pinna, X 14, Costa Rica, Dodge & Thomas 6421, GH; B, portion pinna-rachis, X 234, Peru, Tryon & Tryon 5229, GH. Fig. 153. P. propinqua: fertile pinna, X %, Colombia, Haught 3772, GH. Fig. 154. P. podophylla: fertile pinna, X %4, Colombia, Pennell et al. 8610, CH. 204 ROLLA TRYON Mexico to Panama; Jamaica; Colombia to Bolivia and Brazil. Forests, Loreto, Hudnuco and J unin, 100-700 m. Specimens seen: LORETO: Rio Itaya, Killip & Smith 29620 (us); Gamitanicocha, Rio Mazan, Schunke 282 (F, GH, UC, US); Paraiso, Smith 27895 (us), L. Williams 5253 (F). HUANUCO: Tingo Maria, Allard 21920 (us). JUNIN: Puerto Bermudez, Killip & Smith 26639 (Us). 14. Pteris podophylla Sw. Jour. Bot, Schrad. 1800(2) :67. 1801, based on Lonchitis pedata L. Sp. Pl. ed. 2, 2: 1536. 1763, not Pteris pedata » Type: Jamaica, P. Browne, LINN 1249.1; Browne, Nat. Hist. Jam. & Wy hd, Do eg ee Pteris trialata Sod. Crypt. Vase. Quit. 107. 1893. Type: Niebli, Ecua- dor, Sodiro; isotype: P! photo Gu. Rhizome very stout, evidently creeping; leaves 0.75-2 m. (or more) long, the petiole about as long as the lamina; lamina 3- to 5-pinnate and subpalmately branched at the base, the basal pinnae much the largest, pinnae long-petiolulate to sessile, penultimate segments sessile or short-stalked with the costa awned above and pubescent ‘to glabrate beneath, shallowly pinnatifid to pinnatipartite, the base not decurrent, the basal inferior Segment usually shorter than the superior, or not; ultimate segments ‘Minutely puberulent to hirsute beneath, venation areolate, one long areola with its long axis parallel to the costa, be- tween adjacent costules. Pteris podophylla is a rather variable species, especially n the size of the ultimate segments and the degree to which they are pubescent. The nearly palmate type of branching of the lamina is distinctive, although rarely represented in speciments. Mexico to Panama; Greater Antilles ; Colombia to Bolivia. Forests, San Martin to Cuzco, 1000-1800 m. Selected specimens: SAN MARTIN: Zepelacio, near Moyobamba, Klug 3495 (F, GH, K, US). LORETO: Sinchona, cerca a Divisoria, prov. Coronel Portillo, Aguilar 844 (GH, USM). HUANUCO: Carpish, Sandeman 5208 (K), Ferreyra 1831 (GH, USM), 2123 (BM); Tingo Maria, Ridoutt (GH, USM). PASco: Oxapampa, Soukup 2344 (F, GH). JUNIN: San Ramon, Killip & Smith 24686 (F, US); Chanchamayo, Schunke 149 (us), 743; 966 (F). CUZCO: San Miguel, Cook & Gilbert 1114 (Us); Valle de Pilla- huata, Herrera 1609 (us). 15. Pteris livida Mett, Ann. Sci. Nat. V, 2: 222. 1864. Lectotype: Choachi, Bogota, Colombia, Lindig 179, B! photo GH; isotype: BM! GH: ! us! G. 155. FERNS OF PERU 205 Pteris Andreana Sod. Crypt. Vasc. Quit. 102. 1893. Type: Ecuador, Sodiro; isotype: P!; authentic specimens: GH! Ny! Us Rhizome evidently stout, creeping, leaves 2-6 m. long, the petiole about as long as the lamina; lamina bi- to tripinnate at the base, most- ly 1- to bipinnate above the basal pinnae which are much the largest, pinnae short- to long-petiolulate, penultimate segments with the costa awned above and minutely pubescent to glabrate beneath, pinnatipar- tite to pinnatisect, or 1-pinnate at the base, the base abrupt (not cuneate or decurrent), the basal inferior segments about the same serrate-crenulate, venation areolate, one long costal areola with its long axis parallel to the costa, between adjacent costules. Most of the Peruvian material of P. livida has been iden- tified as Pteris gigantea Willd. That species, however, has two areolae, with their long axis parallel to the costa, be- tween adjacent costules and it also differs in characters of the arrangement and shape of the segments. Costa Rica to Panama, Colombia to Bolivia. Forests, Loreto and Hudnuco, 1500-2400 m. Specimens seen: Poeppig 186 (P). LORETO: Cerro de Escaler (be- tween Yurimaguas and Moyabamba), Ule 6885 (B). HUANUCO: Muna, Macbride 4322 (F, US); Hacienda Mercedes, Prov. Hudanuco, dist. Churubamba, Mexia 8219 (F, GH, K, UC, US); Pampayacu, Kanehira 150bis (GH, US), Poeppig (B); Mirador, prov. Huanuco, 1940, Ridoutt (GH, USM) ; Cochero, Dombey 53 (P). 16. Pteris reticulatovenosa Hieron. Hedwigia 48: 243. 1909, based on P. reticulata Kuhn, Linnaea 36: 91. 1869, not Desv. 1811. Lectotype: Peru, Pavon, B! photoGH. Fic. 156. Pteris nuda Copel. Univ. Cal. Publ. Bot. 19: 302. 1941. Type: Huanu- co, Peru, Mexia 8299, UC!; isotype: GH! Rhizome stout, evidently decumbent, leaves about 1-2 m. long, the petiole about as long as the lamina; lamina 1-pinnate, or bipinnate at the base and each basal pinna with an enlarged, basal, inferior, pinnati- partite pinnule, the basal pinnae about as long as those above, pinnae petiolulate to subsessile, the costa awned above and glabrous or minute- ly appressed-pubescent beneath, pinnatipartite (except the basal when a pinnule is present), with the basal inferior segment usually longer than the superior, the base cuneate; ultimate segments broadest at the base, the sterile margin entire to serrulate, minutely appressed-pubes- cent beneath, venation areolate, with a single costal areola, with its long axis parallel to the costa, between adjacent costules. 206 ROLLA TRYON This species and the next one, P. speciosa, are rather close- ly related; the characters of their venation which may serve to identify them are mentioned in the key. Colombia to Peru. In forests and on wooded, rocky slopes, San Martin to Cuzco, 600-1700 m. Fics. 155-157. Fig. 155. Pteris livida: apex and base of fertile pinna, X Vas Colomb™ Ariste-Joseph, GH. Fig. 156. i > Ha i rl p dere x vay Peru, Klug 3680, GH; B, portion of sterile pinna, venation, X 114, Kind lat’ & Smith 23962, cH; C, portion of fertile pinna, indument, X 114, Peru, , GH. FERNS OF PERU 207 Specimens seen: Pavon (B); Ruiz 40 (B). SAN MARTIN: Tarapoto, Spruce 4752 (B, K). HUANUCO: Pozuzo, Macbride 4718 (F, US); Ria- chuela Chontalagua, Mexia 8299 (GH, K, UC). JUNIN: La Merced Soukup 1031 (F); Dos de Mayo, Pichis Trail, Killip & Smith 25812 (us); Chanchamayo, 1909, Schunke (BM). CUZCO: entre Quincemil y San Lorenzo, Vargas 11724 (GH). 17. Pteris speciosa Kuhn, Linnaea 36: 91. 1869. Lectotype: St. Gavan, (Rio San Gaban), Peru, Lechler 2326, B! photo GH, fragment Ny! Us! Fic. 159. Rhizome not seen, leaves about 0.5-1.0 m. long; lamina 1-pinnate, the basal pinnae about as long as those above, pinnae petiolulate, lobed to moderately pinnatifid, with the basal inferior segment (or lobe) shorter than the superior, the costa unawned or awned above; lobes or ultimate segments broadest at the base, the sterile margins sharply serrate, glabrous or minutely pubescent beneath, venation areolate, a single costal areola, with its long axis parallel to the costa, between adjacent costules. Pteris speciosa appears to be a distinct species but the material available for study has been so limited that its status remains in some doubt. It is unusual in having the costa either awned or unawned. The characters of venation na separate it from P. reticulatovenosa are given in the ante to Peru In forests, San Martin and Puno, ca, 1000-1500 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce 4757 (BM), 4759, a marked 4757 (K). PUNO: San Gaban (Rio), Lechler 2326 (B, NY, Us). . ris Lechleri Mett. Fil. Lechler. 2: 13. 1859. Type: Tatanara A hae Garbaya, Dept. Puno), Peru, Lechler 2533, 8B! photo GH. FIG. Sink vestita Baker, ihe Eb 169. 1867. Type: Tarapoto, Spruce 4063, K! photo us, fragmen Pteris Killipii Maxon, pay Rem Jour. 23: 107. 1933, (nom. superf.., illegit.). Type: the same as that of Pteris vestita. Rhizome small to moderately stout, erect, leaves about 0.5-1.25 m. long, the petiole about as long as the lamina, or longer; lamina 1- to bipinnate at the base, 1-pinnate above the basal pinnae which are about as long as those above and have a single enlarged, basal, inferior, mod- erately to deeply pinnatifid segment or pinnule, pinnae short- petiolulate to sessile, with the costa unawned above and hirsute beneath, moderate- ly pinnatifid to pinnatipartite (except for the basal pair), the basal segments reduced; ultimate segments hirsute on both surfaces and on 208 ROLLA TRYON the margin, sterile margins subentire to crenulate, venation areolate, one costal areola, with its long axis parallel to the costa, between adja- cent costules. The unawned costa and the markedly hirsute segments are distinctive characters of this species. Leaves of young plants (the lamina about 5-9 cm. long) have the basal pinnae pinnatifid and the remainder of the lamina is pinnatifid. Wea XK 1588 Fids. 158-160. Fig. 158. Pteris petiolulata: A, fertile pinna, X %, Peru, Maebride 5714, US portion of fertile pinna, 14, idem. Fig. 159. P. speciosa: apex and base of sterile pinna, X 14, Colombia, Daniel 1211 F 60. P. horizontalis pe i 0 FERNS OF PERU 209 Peru and Bolivia. Forests, San Martin to Puno, 840-1500 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce 4063 (K); Zepelacio, near Moyobamba, Klug 3680 (F, GH, K, US); Lamas, L. Williams 6400 (F, US). HUANUCO: Tingo Maria, Allard 21195, 21201, 21205a, 21594 (US). JUNIN: near La Merced, Killip & Smith 2391% (US), 23962 (F, GH, ; K Smith 24697 (us). Cuzco: Tocate, Biies 1740 (US). PUNO: Tatanara, prov. Carabaya, Lechler 2533 (B). 19. Pteris petiolulata Tryon, Rhodora 62: 9. 1960. Type: La Merced, Peru, Macbride 5714, us!; isotype: F! Fig. 158. Rhizome rather small, erect, leaves about 0.75-1.5 m. long, the acme about as long as the lamina; lamina bipinnate at the base, 1-pin above the basal or lower pinnae; which are about as long as those orn and have a single enlarged, basal, inferior, simple pinnule, pinnae short-petiolulate to sessile, pubescent above and beneath, at least on the costa and costules and sometimes on the margin, entire, or with a basal pinnule, or irregularly and incompletely pinnatifid or lobed most- ly on the basiscopic side, the pinnatifid or lobed pinnae with a long entire apex, the costa unawned above, sterile margins entire to serru- late at the tip; venation areolate, several costal areolae between adja- cent costules (when present). This species has been confused with Pteris Fraseri of Ecuador from which it differs, most obviously, in the several stalked or sessile pinnae. Pteris Fraseri has only the basal pinnae stalked; the lamina above them is pinnatipartite. Pteris denticulata of the West Indies to Brazil, Argentina and Bolivia is also related and differs in its serrate-spinescent sterile margins. Venezuela to Per Ferests, Junin aie Cuzco, 500- 1500 m. Specimens seen: JUNIN: San Ramén, Killip & Smith 24696 (F, GH, Us); La Merced, Macbride 5714 (F, US); Chanchamayo, Schunke 965 (F). cuzco: Rio Pachiri, Biies 1767 (US). 20. Pteris horizontalis (Fée) Rosenst. Meded. Rijks Herb. 19: 10. 1913. Fic. 160. Litobrochia horizontalis Fée, Crypt. Vasc. Brésil, 1: oo t2:12t, 3. 1869. Type: Brazil, Glaziou 2314; fragment ex Fée, NY Rhizome moderately stout, short-creeping, pean ue leaves about 1 m. leng, the petiole about as long as the lamina; lamina bipinnate at the base, 1-pinnate above the basal pinnae which are about as long as 210 ROLLA TRYON those above and have a single enlarged, basal, inferior, pinnatipartite pinnule, pinnae short-petiolulate to sessile, the upper ones decurrent onto the rachis, the costa unawned above, and glabrous beneath, pin- natipartite, the basal inferior segment longer than the superior, or not; ultimate segments glabrous or minutely appressed-pubescent, the sterile margin subentire to crenulate-serrate; venation areolate, one (or less often two) costal areolae, with their long axis parallel to the costa, between adjacent costules. The single collection of this species that I have seen from Peru bears no indication of the habitat or altitude. Peru, Bolivia and Brazil. Cuzco. Specimen seen: CUZCO: Valle de Pillhuata, prov. Paucartambo, Her- rera 1621 (US). 26. ACROSTICHUM L. Sp. Pl. 2: 1067. 1753; Gen. Pl. 484. 1754. Type: Acrostichum aureum L. Terrestrial, or palustral, the rhizome very stout, erect, scaly, bearing the leaves in a crown or cluster; leaves very large, 1-pinnate, glabrous to pubescent, veins anastomosing; sporangia borne all over the surface of the fertile pinnae, nearly to the plane, slightly modified margin, indusium absent, paraphyses present. — 2 species in America. Maxon, W. R. Pteridophyta (Acrostichum), in Sci. Surv. Porto Rico & V.I. 6: 401-402. 1926. This genus has only recently been discovered in Peru (1962) and I did not examine the material until it was too late to have illustrations prepared. However, the genus is 4 distinctive one and the palustral habitat, the large 1-pinnate leaves with sporangia completely covering the paraphysate fertile pinnae are characters that will readily identify it. The other American species, A. awrewm, may also be found in Peru and the following key, adapted from Maxon (0P. cit.), will serve to distinguish the two: Fertile lamina with only tthe upper pinnae fertile, pinnae well spaced, coriaceous, the long axis of the areolae oblique to the costa of the A. aureum L. Fertile lamina with all, or nearly all, of the pinnae fertile, pinnae crowded, chartaceous, the long axis of lthe areolae nearly at Ti angles to the costa of ‘the pinna. A. daneaefolum Acrostichum daneaefolium Langsd. & Fisch. Ic. Fil. 5, t. 1. 7 Type: Ins. St. Catharina, Brazil, Langsdorff, LE. FERNS OF PERU 211 Rhizome scales brown, linear, thick and firm; leaves ca. 1.5-3 m. tall, the fertile with few or no sterile pinnae, petiole shorter than the lamina, stout, with large, ligulate, rather soft, fimbriate scales at the base, lamina linear, pinnae usually close, sometimes imbricate, the fertile somewhat smaller than the sterile, entire, chartaceous, with a cartilaginous edge, glabrous to finely pubescent beneath, venation finely areolate, the long axis of the areolae nearly at right angles to the costa. Tropical Ameri Swamp, with Typha and Eichhornia, Tumbes, 20 m. Specimen seen: TUMBES: Corrales, prov. Tumbes, Lépez & Sagdstegui 4034 (GH). TRIBE 8, VITTARIEAE. 27. HECISTOPTERIS J. Sm. Lond. Jour. Bot. 1: 193. 1842. Type: Gym- nogramma pumila Spreng. = Hecistopteris pumila (Spreng.) J. Sm. Epiphytic, the rhizome very small, creeping, scaly, bearing the leaves in a cluster; leaves very small, furcate, glabrous, veins free; sporangia superficial, tories along a portion of the foe indusium absent, para- physes present. — An American genus of 1 Hecistopteris pumila (Spreng.) J. Sm. Lond. Jour. Bot. 1: 193. 1842. Fig, 161. Gymnogramma pumila Spreng. meet Suppl. Syst. Veg. 31. 1828. Type: priedieconeye Weigelt; isotype: B Rhizome scales brown, linear or sublinear, clathrate, slightly irides- cent, leaves ca. 1-2 em. long, lamina thin, narrowly to rather broadly flabellate, variously dichotomously or subdichotomously branched or ’ ually tapering at the base to the petiole, sporangia borne on the distal (but not the apical) portions of the veins. Tropical America; widespread but not commonly collect- Epiphyte or on fallen trunks or branches of trees, Huanu- co, 62 Specimen seen: HUANUCO: Tingo Maria, Allard 20490 (Us). 28. Virraria J. E. Sm. Mém. Acad. Turin 5: 413. 1793. Type: Pteris lineata L. = Vittaria lineata (L.) J. E. Sm. Epiphytic, the rhizome small, erect or short-creeping, scaly, bearing the leaves in a cluster; leaves small to medium sized, the lamina simple, entire, glabrous, veins anastomosing, forming one series of areolae on each bigs of the costa; sporangia borne in a continuous, more or less n, line ial each of the intramarginal commissures, indusium absent, paraphyses present. — 12 species in America 212 ROLLA TRYON Benedict, R. C. A revision of the genus Vittaria J. E. Smith. Bull. ip ai Cl. 41: 391-410. 1914. Try - Taxonomic fern notes, IV.— Some American vittarioid ates sate 66: 110-117. 1964. The rhizome scales are brown or darker (light brown in V. latifolia) and clathrate; usually they are iridescent, especially when the lateral cell walls are heavily sclerotic. he term costate is used in reference to the rhizome scales of Vittaria to describe the dark cell walls that form more or less distinct lines the length of the scale. Scales that are predominantly two cells broad are 1-costate (Fig. 164D), those that are three, four or more cells broad are 2-costate, 3-costate, to many costate, respectively. The spores are sticky and this character is doubtless related to the epiphytic habit of the genus. The typification and correct application of the names V. graminifolia, V. Moritziana and V. Ruiziana are discussed in my paper cited above. KEY TO SPECIES a. Petiole pale, lighter than the green lamina (or if darkened in age rin drying, then concolorous with the brownish lamina); stem — b. ea rcuanainaes w V, graminif ola a. Petiole dark, agen reddish-brown to oe a darker ap ‘he lamina; stem fadial, .o00050. 0 c. Petiole terete or secur (it may be flattened only at the very = Me Re, BON a es pe ae . Rhizome icing l-costate (sometimes 2-costate at the 2 those on ‘the petiole neuer but longer; sporangia in deep grooves near the mar 3. V. stipitata d. Rhizome scales naar ra to 7-costate, those on the petiole longer and narrower; sporangia in shallow grooves back from the wiarem, ooo 4. V. Moritziana c. Petiole flattened and two-angled throughout, wholly or mostly narrowly alate, firm or rather soft, usually irregularly wrinkl or grooved in drying. ... - & e. Rhizome scales mostly. dark brown to atropte puedes “with sclerotic lateral cell walls and stout, short teeth; lamina usually 15 or more (rarely less than 10) times longer than broad. .....+- f. FERNS OF PERU 213 f. Lamina narrowly elliptic to linear-elliptic, often faleate, with a costal nn on the upper surface from the base to the center or beyo g. g. She nici of the areolae oblique to the costa (a lateral vein reaches the margin above the point where the next one on the same side arises from the costa). voce. 5. V. remota g. Long axis of the areolae parallel to the costa (a lateral vein reaches the margin below the point where the next one on 'the same side arises from the costa V riana f. Lamina narrowly linear (or narrower), or straight, lack- ing a costal ridge on the upper surface at t 7" (one may be somewhat developed beyond the ) eee V. Ruiziana - Rhizome scales light brown, with slightly siete fog cell pels: with long, slender cilia; lamina ca. 5 8 times longer than broad (above.the alate petiole), elliptic; sporangia born well back from the ma argin. 8. V. latifolia 1. Vittaria lineata (L.) J. E. Sm. Mém. Acad. Turin 5: 421. 1793. Fig. 162. Pteris lineata L. Sp. Pl. 2: 1073. 1753. Type: Santo Domingo, Plumier, Fil. t. 143, Vittaria filiformis Cav. Deser. 270. 1802. Type: Peru, Née, MA, photo F, NY, US; seen by C. Chr. (Dansk Bot. Ark. 9(3) : 24. 1987) and placed here with some doubt, the type has no rhizome or petiole base. hizome dorsiventral; leaves ca. 10-100 cm. long, pendent, the petiole pale, lighter than the lamina (or in drying concolorous with it) soft somewhat flattened; lamina very narrow; sporangia in deep grooves back from the inairgin. The characters of the spores and of the paraphyses that afford the best characters to distinguish this species from V. graminifolia are presented in the key. The following charac- ters, in addition, may be mentioned. The rhizome scales of V. lineata usually have a 1-costate, filiform tip and cell walls that are all of the same thickness. In V. graminifolia, the rhizome scales usually have a short, 1-costate tip, or none, and the cell walls at the margin of the scale are thinner than those in the center. I have seen a few specimens (not from Peru) that have the paraphyses of V. lineata and the type of spore of V. graminifolia, or vice versa. Tropical America. Epiphyte in forest, San Martin to Junin, 150-1000 m. 214 ROLLA TRYON Specimens seen: SAN MARTIN: Tarapoto, L. Williams 6321 (F, US). LORETO: Santa Maria, ca. km. below Yurimaguas on Rio Huallaga, Allard 22471 (US). HUANUCO: Tingo Maria, Asplund 12356 (Us). JUNIN: La Merced, Weberbauer 1914 ie), 162c 162D I62E ‘ I6IB i e Ht ‘yes : LEY ye ; Fics. 161-163, Fig. 161. Hecistopteris pumila: A, ei on twig, X 1, Colombia, Haught 2080, GH; B, portion of fertile lamina, X 2, Fig. 162. Vittaria lineata: A, fertile leaf, X 14, Paraguay, Fiebri 5950, GH; B, pase of fertile lamina, X 1%, C, petiole nae x m; D, rhizome aie e paraph me es, X n; x 4, 7 duende; Pittier 9971, GH paraphysis, enlarged, idem » oo es aiinubiadabtiés A, leaf. B, visieadhes wale x y ee Cc, FERNS OF PERU 215 2. Vittaria graminifolia Kaulf. Enum. Fil. 192. 1824. Isotype: BE! (Herb. Greville, Kaulfuss misit 1827, seen in 1964; this must replace my choice of Sello, B, as a lectotype). FG. 163. Vittaria filifolia Fée, Mém. Fam. Foug. 3: 20, t. 3, f. 6. 1851-1852. Lectotype: Guadeloupe, Herminies (Herb. Cosson), Rhizome dorsiventral; leaves ca. 10-50 cm. long, pendent to suberect, the petiole pale, lighter than the lamina (or in drying concolorous with it), soft and somewhat flattened; lamina very narrow; sporangia in deep grooves close to the margin. The characters of the paraphyses and spores that distin- guish this species from the preceding one, V. lineata, are presented in the key. Tropical America. Epiphyte in forests, Amazonas and Junin, ca. 1200 m. Specimens seen: AMAZONAS: Tazan, prov. Chachapoyas, Lépez et al. 4261 (GH). JUNIN: La Merced, Macbride 56387 (F, US). 3. Vittaria stipitata Kze. Linnaea 9: 77. 1834, Type: Pampayacu, Peru, Jul. 1829, Poeppig (Diar. 1121), evidently illustrated in Kze. Anal. Pterid. t. 18, f. 1; isotype: P!; authentic specimen: Poeppig, B! photoGH,K! Fic. 164. Rhizome radial; leaves ca. 10-75 cm. long, rygenge the petiole dark reddish-brown to atropurpureous, hard, terete or nearly so; lamina narrow; sporangia in deep grooves near the margin. This species and the next, V. Moritziana, are most readily identified by the characters of the sporangial grooves men- tioned above and the characters of the rhizome and petiole scales. In V. stipitata the rhizome scales are 1-costate or at most 2-costate at their base; in V. Moritiziana they are mostly 4- to 7-costate. The petiole scales are similar to those of the rhizome but longer in V. stipitata, while in V. Moritzi- ana they are also narrower. Guatemala to Panama; Greater Antilles; Venezuela to Colombia and south to Bolivia; Brazil. Epiphyte in forests, San Martin to Cuzco, 600-2000 m. seen: SAN MARTIN: Cumbassauma Mts., Steere (GH); Specimen. Monte Guayrapurina, prope Tarapoto, Spruce 4773 (K, P). HUANUCO: Pampayacu, Poeppig (B, K, P); Tingo Maria, Allard 21932 (Us). Idma, prov. Convencién, Weberbauer 5019 (B). 216 ROLLA TRYON 1648 166A i64c 164A } = 164 165A 167C 167A RWC Fics. 164-167, Fig. 164. Vittaria stipitata: A, bain: leaf, X Ke Basie Steinbach idem 8759, GH; B, section of fertile lamina, & 4, idem; C, petiole section, X 10, ? i: rhizome seale, X 12, idem. Fig. 165. V. Mittal: A, section * fertile green x 3, Colombia, Pennell 9250, GH; B, rhizome scale, X 12, idem. Fig. 166. V. remota By leaf, X %, Colombia, Schultes & Villarreal 5249, G , portion of ‘ort lamina, lower surface, , idem; C, portion of lami upper surface, ' r. Fig. 16h V. Gardneriana: A, fertile leaf, X uador, teri ot E3446, GH; B, aes of fertile lamina, X 114, idem; C, petiole atic: X 10, idem; D, rhizome seale, X 1 FERNS OF PERU 217 4. Vittaria Moritziana Mett. Ann. ea mh V, 2: 207. 1864. Type: Canoas, Colombia, Lindig 319, B! Rhizome radial; leaves ca. 10-75 cm. long, pendent, the petiole reddish-brown to atropurpureous, hard, terete or flattened basally, lamina narrow; sporangia in shallow grooves back from the margin. The characters that best serve to distinguish this species from the preceding, V. stipitata, are discussed under that species. Benedict interpreted this species and name correctly but erroneously applied the earlier name V. Ruiziana to it. Costa Rica and Hispaniola; Venezuela to Colombia, Ecua- dor and Bolivia; not known from Peru. 5. Vittaria remota Fée, Mém. Fam. Foug. 7: 26, t. 20, f. 1, 1857. Type: prov. Ocafia, Colombia, Schlim 611. Fic. 166. Rhizome radial; leaves ca. 5-30 em. long, erect or suberect, petiole brown to atropurpureous, narrowly alate, flattened, rather soft, wrinkled in drying; lamina narrowly elliptic to elliptic-linear, some- times subfalcate, with a costal ridge on the upper surface from the base to the center or beyond; sporangia in shallow grooves back from the margin This species is rather similar to the next, V. Gardneriana, and two can best be distinguished by the characters of the venation mentioned in the key. reater Antilles; Surinam to Colombia, Ecuador and Bolivia; not known in Peru. 6. Vittaria Gardneriana Fée, Mém. Fam. Foug. 3: 15, t. 3, f. 1, 1851- 1852. Lectotype: Brazil, Gardner 147; isotype: B! photo GH, BM! K! P! us! Fie. 167. Rhizome radial; leaves ca. 5-20 em. long, suberect or pendent, petiole brownish to atropurpureous, narrowly alate, flattened, rather soft, wrinkled in drying; lamina linear-elliptic, often subfalecate, with a costal ridge on the upper surface from the base to the center or beyond; sporangia in shallow grooves back from the margin. This species is close to the previous one, V. remota, and can best be distinguished from it by the characters of vena- tion mentioned in the key. Costa Rica and Panama; Sa British Guiana to Colombia, south to Peru; Braz Epiphyte in dense forest, juni, ca. 1500 m. 218 ROLLA TRYON Specimens seen: JUNIN: Yapas, Pichis Trail, Killip & Smith 25552 (F, GH, US). 7. Vittaria Ruiziana Fée, Mém. Fam. Foug. 3: 16, t. 3, f. 3. 1851-1852. Type: Peru, Ruiz; authentic specimen: Peru, Dombey (det. Fée), B! photoGH. fFiG1 Pteropsis wittinsoiles Desv. Mém. Soc. Linn. Paris, 6: 219. 1827. Type: Peru, Herb. Desv., P!, the single leaf evidently taken from Huasi-Huasi, 1779, Dombey, P! (a duplicate of this is in Herb. Kunth, B! photo GH Vittaria Oniiighatand Kuhn, Linnaea 36: 66. 1869. Type: Yungas, Bolivia, D’Orbigny 229, B! photo GH; isotype: GH! P! (an excellent specimen), w! 170A RHC UR Fics. 16 8-170. Fig. 168. lap oe leaf, X 1%, , Vargas 1588, © 169. V. latifolia: A, leaf, X 14, Bolivia, R. S. Williams eons cH; B, rhizome > aeale x = Ani heey oukup 2851, GH. Fig. 170. Ananthacorue angustifolius: Cc me, X 4, Bolivia, R. S. Williams 1845, GH: B, portion po apg gig Ser Haught 1649, GH; C, section of fertile etry x 3, FERNS OF PERU 219 Vittaria vittarioides (Desv.) Weath. Contrib. Gray Herb. 114: 34. 1936, not (Thouars) C. Chr. Ind. Fil. 655. 1907. Rhizome radial; leaves ca. 15-70 cm. long, pendent, petiole dark brown to usually atropurpureous, narrowly alate, flattened, firm ally grooved in drying; lamina narrow to very narrow, rather straight, lacking a costal ridge on the upper surface at the base, although one may be developed beyond 'the base; sporangia in rather deep grooves back from the margin. Venezuela to Colombia, south to Bolivia. Epiphyte in forest, Hudnuco and Cuzco, 1800-3100 m. Selected specimens: HUANUCO: near eee Macbride 4136 (B, F, GH, Us). cuzco: near Achirani, prov. Paucartambo, Vargas 1588, 11157 (GH); Cerro Chuyapi, Biies A44 (GH, cp Valle de San Miguel, Bies 2187 (US); Huadquifia, Biies 1344 (US). 8. bition latifolia Benedict, Bull. Torrey Cl. 41: 403, t. 17. 1914. Type: Arbara, Bolivia, R. S. Williams 1337, NY!; isotype: GH! us! Pam 169 Rhizome radial; leaves ca. 5-18 cm. long, erect ?, petiole dark brown, narrowly alate, flattened, firm, grooved in drying; lamina elliptic, grad- ually narrowed at the base to the petiole; sporangia in shallow grooves well back from the margin. The characters of the rhizome scales mentioned in the key are distinctive for this species. Their light brown color cor- responds to the only slightly sclerotic cell walls. Peru and Bolivia. Pasco. Specimen seen: PASCO: Oxapampa, Soukup 2351 (GH). 29, ANANTHACORUS Underw. & Maxon, Contrib. U. S. Nat. Herb. 10: 487. 1908. Type: Pteris eet Sw. = Ananthacorus angustifolius (Sw.) Underw. & Maxo Epiphytic, the rhizome small, short-creeping, scaly, bearing the leaves in a loose cluster or spaced; leaves small, the lamina simple, entire, pri tinanty veins leaded, forming a few series of areolae on each side of the costa; sporangia borne in a long sunken line along each of the intramarginal commissures, indusium absent, paraphyses present. — An American genus of 1 species. Ananthacorus angustifolius (Sw.) Underw. & Maxon, Contrib. U. 8S. Nat. oe 10: 487. 1908. G. angustifolia Sw. Prod. 129. 1788. Type: “Jamaica and His- rt jiatitoda, > (Fi. Ind. Occ. 3: 1599). 220 ROLLA TRYON Vittaria costata Kze. Linnaea 9: 77. 1834. Type: Tocache, Huallaga, Peru, Jun. 1830, Poeppig; isotype: w! (At B, Kze. Anal. Pterid. t. 18, f. 2, is mounted in the herbarium and has the same data as the holotype written on it; it is probable that the figure was prepared from the holotype at Lz, now destroyed). Rhizome scales brown to atropurpureous, linear to narrowly lanceo- late-attenuate, clathrate, iridescent; leaf ca, 8-35 cm. long, the lamina coriaceous, narrow and tapering gradually to base and apex (sterile ones much smaller and relatively broader, often elliptical), nearly or quite sessile, straight to subfaleate, the costa extending to the apex; the fertile commissure continuous or irregularly discontinuous, some- times the sporangia in short, interrupted lines. Tropical America. Epiphyte in forest, San Martin and Hudnuco, ca. 650 m. Specimens seen: SAN MARTIN: Tocache, Poeppig (w); Cerro Cam- pana, Tarapoto, Spruce 4670 (P). HUANUCO: Supte river, n. of Tingo Maria, Stork & Horton, 9598 (F, US); junction of Rio Monzén and Rio Huallaga, Asplund 12678 (Us). 30. POLYTAENIUM Desv. Mém. Soc. Linn. Paris 6; 218. 1827, not Polytaenia DC. 1829 (heteronym). Type: Hemionitis lineata Sw. = Polytaenium lineatum (Sw.) J. Sm. Fic. 171. Epiphytic, the rhizome small, short-creeping, scaly, bearing the leaves in a loose cluster; leaves small, the lamina simple, entire, gla- brous, veins anastomosing, forming a few to many series of areolae on each side of 'the costa; sporangia more or less sunken, borne in several long lines, or along most of the veins, indusium and paraphyses absent. — An American genus of 10 species. Benedict, R. C. The genus Antrophyum-I. Bull. Torrey Cl. 34: 446 458. 1907. Benedict, R. C. The genera of the fern tribe Vittarieae. Bull. Torrey Cl. 38: 153-190. 1911. Tryon, R. Taxonomic fern notes, IV.— Some American vittarioid ferns. Rhodora 66: 110-117, 1964. This genus is frequently treated as part of the larger, Pre dominantly Old World, Antrophyum. The reasons for main- taining it, largely on the bases of geography and the absenc® of paraphyses in Polytaenium, are discussed at length a gi paper cited above, as are the changes in the application ° names that are necessary. eat The three species with a broad lamina require additional study before their classification can be considered to be ade- FERNS OF PERU 221 quate. The petiole color is not always easy to determine in dried specimens and if this character does not actually cor- relate as well as it seems to with the others some change in the classification may be necessary. Although material from Central America, of this group, and that from the Guianas is relatively uniform, there is greater variation and less cor- relation of characters in the central Andes. The rhizome scales of Polytaenium are brown, clathrate and sometimes iridescent ; they are sometimes pectinate-toothed. The scales may offer some useful characters; however, the differences I have observed in South America do not correlate with other characters. 6a J 7} a { Pe we ‘ MSW y bh) aAR Fic. 171. Polytaenium guayanense, Rio Monzon, Tingo Maria (Huanuco). 222 ROLLA TRYON KEY TO SPECIES a. Soral lines few and long, parallel to eat costa and deeply sunken in the tissue; lamina narrowly elliptic-linear. ...........+++++++ 1. P. lineatum a. Soral lines many, long and short, ‘Gotee from the costa and superficial to partially sunken in the tissue b. b. Petiole greenish beneath (or drying to light or dark brown), con- colorous, or darker than the adjacent leaf-tissue, usually strongly alate; lamina usually coriaceous, broadly to ag oblanceolate. eg b. Petiole pale (straw-colored) beneath (or eciheesBs paris than . adjacent leaf-tissue) ; lamina usually papyYraceousS. .......+--+sersere c. Lamina rather narrowly elliptical or with ri parallel RL petiole narrowly alate. 3. P. guayanense ce. Lamina oblanceolate, petiole strongly alate. .......::sssssesssssreeneneee 4, P. brasilianum z 1. Polytaenium lineatum (Sw.) J. Sm. Jour. Bot. Hooker, 4: 68. 1841. 1G. 172 Hemionitie lineata Sw. Prod. 129. 1788. Type: Jamaica, Swartz; iso- type: Herb. Willd. 20033, B! photo GH. Vittaria lanceolata Sw. Ges. Naturf. Freunde Berl. Mag. 2: 188. hag based on Hemionitis lineata Sw., not Vittaria lineata (L.) J. E. Antrophyum lineatum (Sw.) Kaulf, Enum. Fil. 199. 1824. Pe aaa lanceolatum (Sw.) Desv. Mém. Soc. Linn. Paris 6: 218. Leaves ca. 8-30 em. long, petiole strongly alate, straw-colored to light green beneath, concolorous with the adjacent leaf-tissue; lamina nat rowly elliptic linear, more or less coriaceous; soral lines few, long and parallel to the costa. Tropical America. Epiphyte in woods and forests, Hudnuco, Junin and Ayacucho, 400-1800 m. Specimens seen: HUANUCO: Mufia, Bryan 529 (F, US). JUNIN Pe pampa, Soukup 2352 (GH); Huacapistana, Killip & Smith 24810 (03); f near La Merced, Killip & Smith 23859 (us). AYACUCHO: Rio Ap — valley, near Kimpitiriki, Killip & Smith 22869 (us). 2. Polytaenium cajenense (Desv.) Benedict, Bull. Torrey Cl. 38: 169. ik... Fe178. Hemionitis cajenensis Desy. Ges. aonb \ bbecnges Berl. Mag. 5: 311. 1811. Type: French Guiana, Herb. Des Antrophyum cajenense (Desv.) Spreng. - Veg. 4: 67. 18 - Antrophyum discoideum Kze i 6: 702. 1848. Lectotype* Colombia, Karsten 30; isotype: B! igh 1 ee lytaenium discoideum (Kze.) Soar Bull. Torrey Cl. 38: 169. FERNS OF PERU 223 Leaves ca. 10-35 cm. long, petiole usually strongly alate, greenish beneath (drying to light or dark brown), concolorous with, or darker than, the adjacent leaf-tissue; lamina narrowly to broadly oblanceolate, papyraceous to usually coriaceous; soral lines many, following nearly all of the veins. This species has usually been called Polytaenium (or Antrophyum) brasilianum. Guatemala to Panama; Greater Antilles; Guianas to Co- lombia, south to Bolivia; Brazil. Epiphyte in dense forest, Loreto, Huanuco and Cuzco, 100-1200 m. I73A x 14, Brazil, Dusén 14886, B, portion of fertile lamina, X 114, Colombia, Lindig 170, GH; C, section of fertile lamina, X 3, idem. Fig. 173. P. cajenense: A, leaf, X 14, Peru, Schunke 283, portion of fertile lamina, X 1, idem. Fig. 174. P. guayanense: leaf, X 4, Guiana, A. C. Smith 2872, GH. Fics. 172-174. Fig. 172. Polytaenium lineatum: A, plant, H; i cH; 5, British 224 ROLLA TRYON Selected specimens: LORETO: Gamitanicocha, Rio Mazan, Schunke 283 (F, GH, US); San Antonio, Rio Itaya, Killip & Smith 29367 (US). HU- ANUCO: Hacienda Mercedes, prov. Hudnuco, Mexia 8177a (GH); Tingo Maria, Asplund 12222 (Us), Ferreyra 10230 (GH, USM), Woytkowski 1097 (GH, USM). cUzcOo: Hacienda Potrero, near Quillabamba, Coro- nado 120 (GH, Us). 3. Polytaenium guayanense (Hieron.) Alston, Kew Bull. 1932: 314. Fics. 171, 174. Antrophyum guayanense Hieron. Hedwigia 57: 212. 1915. Lectotype: Trinidad, Fendler 151, B!; isotype: GH! Leaves ca. 10-30 em. long, petiole narrowly alate, straw-colored be- neath (or darker in drying but lighter than the adjacent leaf-tissue) ; lamina rather narrowly elliptical or with nearly parallel sides, sub- coriaceous to usually papyraceous; soral lines many, following nearly all of the veins. This species has usually been called Polytaenium (or Antrophyum) cajenense. Guianas, Trinidad to Colombia and Peru, adjacent Brazil. ' Epiphyte in forest, San Martin, Loreto and Hudnuco, 135- 00 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce 3993 (P). LORETO: Santa Rosa, Rio Huallaga below Yurimaguas, Killip & Smith 28951 (GH, US). HUANUCO: Rio Monzon, near Tingo Maria, Tryon & Tryon 5300 (GH, U, US, USM). 4. Polytaenium brasilianum (Desv.) Benedict, Bull. Torrey Cl. 38: 169. 1911. Hemionitis brasiliana Desv. Mém. Soc. Linn. Paris 6: 216. 1827. Type: Brazil, Herb. Desv. P! photo GH, Us. Antrophyum subsessile Kze. Anal. Pterid. 29. 1837. (nom. superf. illegit.). Type: the same as that of Hemionitis brasiliana. Antrophyum brasilianum (Desv.) C. Chr. Ind. Fil. 59. 1905. Leaves ca. 10-30 em. long, petiole strongly alate, straw-colored ai neath (or darker in drying but lighter ‘than the adjacent leaf-tissue) 5 lamina oblanceolate, papyraceous; soral lines many, following nearly all of the veins. This species is not known from Peru but it has been col- lected in Bolivia and it may be expected to occur I the Department of Puno or Cuzco. It has previously usually been called Polytaenium (or Antrophyum) discoideum. 31. ANETIUM Splitg. Tijdsch. Nat. Gesch. 7: 395. 1840. Type: Ae7® tichum citrifolium L. = Anetium citrifolium (L.) Splitg. FERNS OF PERU 225 Pteridanetium Copel. Gen. Fil. 224. 1947, based on Anetium Splitg. not Anetia Endl. (heteronym). Epiphytic, the rhizome slender, long-creeping, scaly, bearing the leaves at intervals; leaves small to rarely large, the lamina simple, entire, glabrous, veins anastomosing, forming many series of areolae on each side of the costa; sporangia borne superficially and sparingly along the veins and also between them, indusium and paraphyses ab- sent. — An American genus of 1 species. C. V. Morton (Amer. Fern Jour. 43: 71. 1953) has point- ed out that Anetiwm is a different name than Anetia, not a homonym of it as Copeland considered it to be. Anetium citrifolium (L.) Splitg. Tijdsch. Nat. Gesch 7: 395. 1840. Fic. 175. Acrostichum citrifolium L. Sp. Pl. 2: 1067. 1753. Type: Martinique, Plumier, Fil. t. 116. Pteridanetium citrifolium (L.) Copel. Gen. Fil. 224. 1947. Rhizome scales brownish, mostly broadly ovate-attenuate, clathrate, iridescent; leaves herbaceous-fleshy (papyraceous in drying), pendent, 10 'to usually 15-30 to 100 em. long, the lamina elliptical, to oblanceolate to ligulate, narrow to broad, nearly or quite sessile to short-petioled (especially in large leaves) the margin sometimes ruffled, costa extend- ing ca. %ths the length of the lamina, apex obtuse to acute to abruptly acuminate. Tropical America. Epiphyte in dense forests, Loreto and Hudnuco, 100-800 m. imens seen: LORETO: Santa Rosa, below Yurimaguas, Killip & Smith 28881 (us); Puerto Arturo, Rio Huallaga, below Yurimaguas, Killip & Smith 27900 (F, GH, US) ; Rio Mazan, Schunke 301 (F, GH, US). HUANUCO: Rio Monzon, near Tingo Maria, Tryon & Tryon 5300 1/2 (GH). TRIBE 9. DAVALLIEAE. 32. NEPHROLEPIS Schott, Gen. Fil. no. 3. 1834. Type: Polypodium exaltatum L. = Nephrolepis exaltata (L.) Schott. Terrestrial or epiphytic, the rhizomes small to rather stout, erect, usually stoloniferous, scaly, bearing the leaves in a Crown OF loose cluster; leaves of medium size to very long, 1-pinnate, glabrous, slightly sealy or pubescent, the pinnae articulate, veins free; sori orbicular to lunate, borne on the anterior branch of a vein, toward the margin, not paraphysate, indusium orbicular, attached at the base of the narrow sinus, to lunate and attached along the basal side.—6 species in erica. 226 ROLLA TRYON ‘Maxon, W. R. Pteridophyta (Nephrolepis), in Sci. Surv. Porto Rico & V. 1. 6: 484-486. 1926. Morton, C. V. Observations on Cultivated ferns, V— The species and forms of Nephrolepis. Amer. Fern Jour. 48: 18-27. 1958. The species are terrestrial or casually epiphytic. In the latter case or when they grow on the edge of a bank, some, such as N. biserrata and N. cordifolia, may develop very long leaves (up to 3 or 4? m.). In many specimens of these species, regardless of the length of the leaf, the apex is a loose bud and the leaf is thus apparently indeterminate. In other specimens the apex is fully expanded but usually not well developed. Many species of Nephrolepis are cultivated and several of them are represented by variants that depart rather widely from the wild form. The recent paper by C. V. Morton, cited above, treats the kinds that are commonly cultivated and includes references to other more detailed treatments. Here I have included only the two exotic cultivars that have evi- dently escaped from cultivation in Peru. KEY TO SPECIES the hase of the petiole, 8-7.) csciie3socascbiccesssecdooséscsctiosSscvcsciencsjmoitie eam b Re PWS MAREE 565i ss Se sd istincasscesansoudonesiincar ee C. c. Pinnae, the basal ones excepted, with the base very unequal, cuneate to convexly so on the inferior side, acutely to subacutely auriculate on the superior side; pinnae minutely appressed- 1. N. 10 vularis d. Longer pinnae rather abruptly reduced to a prolonged i to acuminate; indusia all orbicular, with a very narrow sinus; pinnae commonly hirtellous beneath, very rarely only squamur lose; vascular bundles 5-7, 3. N.biserralt d. Longer pinnae obtuse to shortly acute; indusia variable, ia least some toward the base of the pinna orbicular to orbicular reniform, at least some toward the apex reniform to foe (rarely all reniform) ; commonly squamulose or glabrate Ps neath; vascular bundles 3 (-5). .. 2, N, enol b. Pinnae, or at least some of them, pinnatifid or more complex. +" Hesserescecoreessedoncccebccccese 2a. N. exaltata cv. Boston? FERNS OF PERU 227 a. Indusia reniform to lunate, mostly facing the apex of the pinna, the sinus broad or absent, the mature sporangia projecting on the ope side; vascular bundles, above the base of ‘the petiole, 1-3. ...........+. e. Pinnae simple and orbicular, or with two | 4a. N. cordifolia cv. e. Pinnae simple and elongate. f. Upper pinnae, but not necessarily the apical ones, rounded a’ the base on 'the inferior ~~ uf sometimes shortly so) to all pinnae cordate; vascular bundles g. Pinnae coriaceous, ed veins obscure, rarely thin and the veins evident; stolons wiry and relatively stout, usually numerous, often tuber-beahing (but not native plants in Peru); plant developing a substantial rhizome. .......sssseee 4. N. cordifolia g. Pinnae thin-herbaceous, the veins evident; stolons reshar soft, slender, usually few, usually or always tuber-bearin plant apparently annual from a tuber, only a small thizome developed. 5. N. occidentalis f. Upper pinnae cuneate at the base on the inferior side, or all pinnae except the basal ones so, auriculate on the superior side; vascular bundle 1; stolons not tuber-bearing. ...... 6. N. pectinata 1. Nephrolepis rivularis (Vahl) C. Chr. Ind. Fil. 455. 1906. Fic. 176. Polypodium rivulare Vahl, Eclog. Amer. 3: 51. 1807. Type: Montser- rat, West Indies, Ryan. Aspidium eminens Wickstr. Vet. Akad. Handl. 1825: 436. 1826. Type: Guadeloupe, Forsstrém, s-PA! Nephrodium crenatum Desv. Mém. Soc. Linn. Paris 6: 252. 1827. Type: “Trop. Amer.” P, photo GH! Suchen intermedia Sod. Rec. Crypt. Vase. Quit. 57. 1883. T San Miguel de los Colorados, Ecuador, Sodiro; isotype: K!; authentic specimens: uc! us! Vascular bundles in the petiole, above the base, 3-5; pinnae sessile, very unequal at the base, the inferior side cuneate, the superior acutely auriculate, apex obtuse to acuminate, minutely appressed-squamulose beneath; indusium orbicular, mostly facing the margin of the pinna, sinus very narrow, the sporangia projecting on all sides when mature The combination Nephrolepis rivularis was not made by Krug (Engl. Bot. Jahrb. 24: 122. 1897) as commonly cited. It was definitely made, although unintentionally, by C. Christensen; it may also have been properly made earlier. Tropica] America. In forests, terrestrial and on tree bases, San Martin to Puno, 100-1000 m Specimens seen: SAN MARTIN: Soritor, Woytkowski 6246 (GH). LORETO: Pinto-Cocha, Rio Nanay, L. Williams 812 (F, US); between Rio Nanay and Rio Napo, L. Williams 703 (F); San Juan, Iquitos, L. 228 ROLLA TRYON Williams 3733 (F); Mishuyacu, near Iquitos, Klug 176 (F, US) ; Gami- nicocha, Rio Mazan, Schunke 219 (GH, UC). HUANUCO: Tingo Maria, Allard 21492 (US). MADRE DE DIOS: Maldonado, Rauh P751 (B). PUNO: San Gaban (Rio), Lechler 2515 (Gu, K). 2. Nephrolepis exaltata (L.) Schott, Gen. Fil. no. 3. 1834. Fic. 177. Polypodium exaltatum L. Syst. Nat. ed. 10, 2: 1326. 1759. Lectotype: (by Alston, Phil. Jour. Sci. 50: 182. 1983) Sloane, Jam. t. 31; Sloane Herb. 1: 52, BM! Vascular bundles in the petiole, above the base, 3 to rarely 5; pinnae subsessile to petiolulate, usually rather unequal at the base, the inferior 9Sa0ggs se SSoSssnassesas > ene Fics. 175-178. Fig. 175. Anetium citrifolium: A, plant, X 14, Jamaica, Mazon rivu- Killip 790, GH; B, portion of fertile lamina, X 1, idem. Fig. 176. age? Tis H 1. H GH; ; fertile ‘ St. Lucia, Howard 11385, cH. Fig. 177. N. exaltata: ns ps pinnae, X 114, Porto Rico, Holm 157, GH; B, portion of fertile pinna, X 21 Fig. 178. N. biserrata: fertile pinna, X 1%, Guatemala, Deam 478, GH. FERNS OF PERU 229 side rounded to auriculate, the superior side auriculate, the apex obtuse to acute, fibrillose-squamulose to glabrate beneath; indusium orbicular to orbicular-reniform toward the base of the pinna, to reniform or lunate toward the apex, the orbicular ones mostly facing the margin, the reniform and lunate ones facing the apex, mature sporangia mostly projecting on all sides. The wild species is native to Florida, Central America, the West Indies and northern South America. 2a. Nephrolepis exaltata cv. Bostoniensis. Nephrolepis exaltata var. bostoniensis Davenp. New England Florist 2: 137. 1896. Type: from F. W. Fletcher, Davenp. Herb., Gu! The Boston Fern and the many cultivars derived from it are commonly and widely cultivated. Discarded plants may persist and spread by means of the stolons in tropical and subtropical areas. The collections cited below are perhaps from such a source. I have not tried to identify the particu- lar kind of Boston Fern that these specimens represent. LORETO: Caballo-Cocha, L. Williams 2381 (F): Yurimaguas, L. Wil- liams 4051, 4053, 4330 (F); Bersalles, prov. Iquitos, Vargas 11471 (GH). 3. Nephrolepis biserrata (Sw.) Schott, Gen. Fil. no. 3. 1834, Fi. 178. Aspidium. biserratum Sw. Jour. Bot. Schrad. 1800(2): 32. 1801. Type: Mauritius, Gréndat, Herb. Sw. s-PA! Tectaria fraxinea Cav. Descr. 250. 1802. Type: Obragillo, Peru, Née, MA, seen by C. Chr. (Dansk Bot. Ark. 9(3):15, 1937), and referred here. nate, fibrillose-squamulose beneath and also hirtellous, very rarely sed scaly; indusium orbicular, mostly facing the margin of the pinna, the sinus narrow, the mature sporangia projecting on all sides. The lamina of this species is usually considerably broader than that of any other one; even in rather small leaves it is rarely less than 15-20 cm. broad, and in larger leaves, some of which may be up to about 3 m. long, it is commonly 30-40 cm. broad. The taxonomy of this species is not well understood. Al- though it is evidently distinct among other American spe- 230 ROLLA TRYON cies, it intergrades with the Asiatic N. hirsutula (Forst.) Presl, at least in the characters presently employed to distin- guish the two. Also, the American material is very common- ly hirtellous, while the Asiatic N. biserrata is very rarely hirtellous. The proper status and name of the American N. biserrata can be clarified only by a monographic study. Tropical America; Old World. In clearings, forests, or along forest borders, terrestrial or epiphytic, especially on palm trunks, San Martin, Loreto and Ayacucho, 100-900 m. Selected specimens: SAN MARTIN: Rioja, Woytkowski 6155 (GH); Tarapoto, Spruce 4747 (K), L. Williams 5449, 5615, 6581 (F), Woyt- kowski 35230 (UC). LORETO: Iquitos, Mexia 6489 (F, GH, UC, US), Tryon & Tryon 5163, 5176 (BM, F, GH, U, US, USM). AYACUCHO: Kim- pitiriki, Rio Apurimac, Killip & Smith 22856 (F, US). 4. Nephrolepis cordifolia (L.) \Presl, Tent. Pterid. 79. 1836. FI. 179, Map 44, Polypodium cordifolium L. Sp. Pl. 2: 1089. 1753. Type: Petiver, Pterigraph. Amer. t. 1, fig. 11 (not seen in Petiver, Hort. Sice. in Hb. Sloane, BM). Aspidium tuberosum. Willd. Sp. Pl. 5: 234. 1810. Type: Bourbon, Herb. Willd. 19759, B! photo Gu. Aspidium pendulum Raddi, Opuse. Sci. Bolog. 3: 289. 1819. Type: Brazil, Raddi, F1; isotype: “Aspidium pendulum Raddi” ex Raddi, X! Nephrolepis pendula (Raddi) J. Sm. Jour. Bot. Hooker, 4: 197. 1841. Vascular bundles in the petiole, above the base, 1-3; pinnae sub- Sessile to petiolulate, rather unequal at the base, rounded, sometimes shortly so to semicordate on the inferior side, acutely to usually obtuse- ly auriculate on the superior side, apex usually obtuse to acute, glabrous to rarely minutely fibrillose-squamulose beneath; indusium reniform to lunate, mostly facing the apex of the pinna, the matu sporangia projecting on the open side. The Petiver illustration is not adequate for the purposes of accurate identification but I am applying the name in its usual sense. Some epiphytic plants have very long, pendent leaves, UP -to 3 m. and perhaps longer. However, in comparison wit the long leaves of N. biserrata, the lamina of those of '- cordifolia are narrow, about 8-10 cm. broad. I have seen N° evidence of tubers on the stolons of Peru material; they ar¢ frequent on plants from the West Indies. ; I have not been able to maintain the usually recognized FERNS OF PERU 231 a f ‘ = pot Hl t ° ; $ “ee i al ml i f oe ; shy) 44 ( ; } o e U F \ Pa ‘a rr - a @ aoa Br ee | a — ae Ns Lae Xe la tbe ’ ae miei tj { : | L | Jf \, § f Bre J ee 43 N e) VS \ fy ) ‘ } AYs 3 Maps 43-46. Map 43, Vittaria. Map 44, Nephrolepis ‘cordifolia. Map 45, N. pecti- nata. Map 46, Lindsaea 232 ROLLA TRYON N. pendula, being unable to find any character by which it might be clearly distinguished. The tuber-bearing habit of N. cordifolia (in the West Indies, for example) is evidently not constant and the long pendent leaves of N. pendula (in South America) gradually grade into shorter and erect ones. Tropical America; Old World. In dense forest, on shrubby slopes, rocky places and on road banks, Amazonas to Cuzco, 200-3500 m. Selected specimens: AMAZONAS: oo Mathews 3280 (K); entre Chachapoyas y Caclic, Lopez et al. 4350 (GH). SAN MARTIN: Tarapoto, L. Williams 6129 (F); Zepelacio, near Moyobamba, Klug 3736 (F, GH, US). LORETO: Rio Santiago, above Pongo de Manseriche, Mexia 6163 (F, GH, UC, US); Boquerén del Padre Abad, Ridoutt (GH, USM). HUANUCO: Mufia, Macbride 4053 (F, US); Carpish Pass, Coro- nado 71 (GH, UC); Tingo pions Allard 20425, 22362 (US); pair Paty, Ferreyra 9353 (GH, USM). JUNIN: Huacapistana, Ferreyr 11308 (GH, USM), Cerrate Bie (GH, USM), Coronado 258 (GH, se Tryon & Tryon 5452 (BM, F, GH, U, US, USM) ; near Perené Bridge, Rio ‘West 7125 (Uc); Machu-Picchu, Vargas 3167 (F); Valle de Santa Ana, Herrera 2627 (us). 4a. Nephrolepis cordifolia cv. Duffii. Nephrolepis Duffii Moore, Gard. Chr ns. 9: 622, f. 118. 1878. Type: Hort. Veitch, ex Duff, Duke of York's Island, Territory of New Guinea, Herb. Moore, K! In this unusual form, the pinnae are either orbicular or consist of two orbicular lobes, and the rachis is often dichot- omously branched, sometimes several times. Widely culti- ey) Pa 1798 180 18] Fics. 179-181. Fig. 179. se cordifolia: A, — pinna, X 1%, ~~ Tryon 5452, GH; B, portion of fertile pinna, X 5 Fig. 180. N. o¢¢ talis: fertile pinnae, X 1, prin Hinton 9341, GH. Fig. anes N. pectinata: fertile pinnae, X 114, Colombia, Haught 1934, cu. FERNS OF PERU 233 vated and evidently sometimes becoming established from discarded plants. LORETO: Iquitos, L. Williams 3578 (F); Pébas, Rio Amazonas, L. Wu- liams 1825 (F). 5. Nephrolepis occidentalis Kze. Linnaea 18: 343. 1844. Type: Leibold 27, Mexico, Lz destroyed; Leibold 127, B! is this species. Fic. 180. Vascular bundles in the petiole, above the base, 1-3; pinnae sessile or subsessile, unequal at the base, rounded to subauriculate on the inferior side, auriculate on the superior side, apex acute, glabrous to minutely glandular-pubescent beneath; indusium reniform to lunate, mostly facing the apex of the pinna, mature sporangia projecting on the open side. The pinnae are usually long-triangular and thin in texture with the veins evident. The rhizome is poorly developed and evidently most plants are annual from a tuber. Some of the stolons of a mature plant always (7?) bear tubers. Mexico to Panama: Greater Antilles; Colombia, Venezue- la, Peru and Brazil. Rocky mountain side, Huanuco, 2100 m. Specimens seen: HUANUCO: Mufia, Bryan 415 (F, GH). 6. Nephrolepis pectinata (Willd.) Schott, Gen. Fil. no. 3. 1834. Fic. 181, Map 45, Aspidium. pectinatum Willd. Sp. Pl. 5: 223. 1810. Type: Herb. Willd. 19753, B! photo GH. Vascular bundle in the petiole, above the base, 1; pinnae sessile or subsessile, very unequal at the base, cuneate on the inferior side, auric- ulate on the superior side, apex obtuse to shortly acute, glabrous be- neath; indusium reniform to lunate, mostly facing the apex of the pinna, mature sporangia projecting on the open side. The type of Willdenow’s name has been uncertain due to the fact that his citation of “Aspidiwm trapezoides Schkuhr” has been taken to refer to a latter homonym of Aspidium trapezoides Sw. If this were true, the type of Aspidium pectinatum would be the type of Aspidium trapezoides Schkuhr. However, a study of Schkuhr’s treatment clearly shows that he was using Swartz’s name and hence Willde- now’s reference to it, rather than to Swartz, must be taken as a reference to the taxon and not to the name. Aspidium 234 ROLLA TRYON pectinatum Willd. is a new species and typified by Willde- now’s material. Mexico to Panama; Greater Antilles ; Colombia to Bolivia. In forests, ravines, on stream banks, often on rocks or a low epiphyte on trees, San Martin to Puno, 200-2900 m. Selected specimens: SAN MARTIN: Tarapoto, Spruce 4082 (BM); pe Tarapoto, Woytkowski 35245 (uc); San Roque, L. Williams 7517 (F, US). LORETO: above Pongo de Manseriche, Mexia 6219 (F, GH, UC, US); meee, between pate and Moyobamba, Klug 3246 os Ferreyra 10228 (GH, USM). : Satipo, ape (GH, USM) ; Pichis Trail, Killip & Smith 25506, Giese (F, Us); above San Ramén, Killip & Smith 24613 (F, GH, US). CUZCO: Coshipata, Vargas 10217 (uc); San Pedro, prov. Quispicanchi, Vargas 9748 (GH). PUNO: La Pampa, prov. en Watkins (us); San Gaban (Rio), Lechler 2155 (B, K), 3312 Fic. 182. Lindsaea lancea var. falcata, Tingo Maria (Hydnuco). FERNS OF PERU 235 TRIBE 10. LINDSAEEAE. 38. LINDSABA J. E. Sm. Mém. Acad. Turin 5: 401. 1793. Type: Lindsaea trapeziformis Dryand. = Lindsaea lancea (L.) Bedd. FIG. 182, Map 46. Terrestrial, the rhizome slender to rather stout, short-creeping or de- cumbent, scaly, bearing ‘the leaves in a usually loose cluster; leaves small to large, 1-pinnate to bipinnatte, glabrous, veins free; sorus borne on a long marginal commissure connecting the vein-tips, not paraphy- sate (minute trichomes may be present), indusium attached along the commissure, opening toward the nearly plane and unmodified margin. — 46 American species. Kramer, K. U. A revision of the genus Lindsaea in the New World. Acta Bot. Neerland. 6:97-290. 1957 The treatment of this genus has been adapted with the aid of K. U. Kramer from his monograph cited above. Most of the materials have been cited from his revision; to these I have added a few of my own identifications of recently ac- quired material. The center of species of Lindsaea, both in r and abundance, is in northern South America. A little over 1/5th of the American species grow in Peru but they have not been commonly collected. They are often quite local in areas where suitable habitats seem to be frequent. KEY TO SPECIES a. Apical pinnules (apical pinnae of 1-pinnate laminae) strongly re- duced, the terminal segment small and more or less confluent with fe) De b. Lamina bipinnate and the pinna-rachises with lateral wings or angles on Bs pe side that are lighter in color than the pinna- UIE SON ssc ncakcmsncecerendpdasoomenreinenovrennencenasseesneenatevenee nae Cc; c Pailin’ waciides with the wings continuous at their base, a spreadin . 8. L. divaricata ng. c. Pinna-rachises with the wings or angles Grcgalaty interrapted at their base, pinnae peabinastat ascending. ........ 5. L. portoricensis b. Lamina 1-pinnate, or bip e and the pinna-rachises lacking wings or stele on the Hesial i or they are present and con- colorous with the pinna-rachis PTOper. ..-.-.-.s-sssecereessseenenreneeenenentees d. d. Ultimate segments (pinnae fe pinnules) mostly 3 or more o BS Long as BrOAM. c.ccccccecocscccssenseccesecvvesconsesssvscssssscsnneccesnscnnvescemoscsces e. Largest ultimate segments 15 mm. or More ONG. -.-++---+--+++++ ag L. arcuata e. Latoas ultimate segments 5-9 mm. long. ......-- L. Spruceana d. Ultimate segments (pinnules or pinnae) mis 2% times or less as long as broad. bide nee lini eeebectbesentiet £. 236 ROLLA TRYON f. Indusium 0.15 mm. wide, entire to minutely erose, segments herbaceous, veins evident. .... 4. L. guianensis ssp. guianensis f. Indusium ca. 0.3-0.5 mm. wide, pepaxid-erces ‘to strongly erose or lacerate. g. Segments firmly herbaceous to chartaceous, veins ome evident; indusium ca. 0.3 mm. wide, mostly repand-erose. .... 5. L. portoricensis g. Segments usually coriaceous, veins obscure; indusium 0.3- 0.5 mm. wide, mostly strongly erose to lacerate. .......sssss00 . L. stricta a. Apical pinnules (apical pinnae of 1-pinnate laminae) but little re- duced, tthe termi a segment large, free or nearly so from the adja- h. cent lateral o h. Petiole tain on the abaxial side; castaneous to bipesnee pinna- rachises with pale lateral wings. . L. divaricata h. Petiole angular or winged on the abaxial side, at ee toward - apex. i, Terminal segment of the ete (or of the pinnae in Renova laminae) acute or subacu! 7. L. lancea i. Terminal segment of the patra (or of the pinnae in bipinnate laminae) very obtuse to concave. j. j. Terminal segment flabellate, petiole reddish to dark brown. .... 8. L. Schomburgkit j. Terminal segment triangular-cuneate or crescent-shaped, petiole stramineous to pale brown. kK k, Ultimate segments 3 to 4 times as long as ry oauaiy dark olivaceous to brown when dry, .....:s00000+ L. latifrons k. Ultimate segments 4 to 6 times as i one as heirs usually bright green when dry. 10. L. hemiglossa . Lindsaea arcuata Kze. Linnaea 9: 86. 1834. Type: Pampayacu, sie Poeppig 1133, Lz, destroyed; probable isotype:B. Fic. 183. Petiole ca. 10-40 em. long, stramineous to pale brown, the abaxial side rounded to usually angular or sulcate; lamina 1-pinnate or bipinnate, cea. 10-45 em. long, apical pinnae (or apical pinnules of bipinnate lami- nae) reduced to the small, lanceolate terminal segment which is more or less confluent with the lateral segments adjacent to it, rachis of 1-pinnate laminae abaxially angular, sulcate, pinna-rachises abaxially rounded at the base, becoming concolorously angled beyond, ultimate segments 3 to 3% times as long as broad, 12-35 mm. long, herbaceous, the veins evident to rather obscure; indusium ca. 0.2 mm. wide, sub- entire. Mexico to Costa Rica; Greater Antilles; Venezuela to Bolivia; Brazil. Dense forest, Loreto to Cuzco, 100-1500 m. Specimens seen: LORETO: Tierra Doble, alto Rio Nanay, L. Williams 1071 (¥, US). HUANUCO: Pampayacu, Kanehira 177 (GH, US), Poeprg FERNS OF PERU 237 (B). JUNIN: Yapas, Pichis Trail, Killip & Smith 25522 (F, NY, US); Villa Amoretti, G. Kunkel 578 (GH). CUZCO: Bajada, Rio Tocate, Biies 1739 (US). 2. Lindsaea Spruceana Kuhn, Linnaea 36: 79. 1869. Type: Mt. Guay- rapurima, Tarapoto, Peru, Spruce 4023, B; isotypes: BM, BR, F, G, GH, K, FI Lindsaea tarapotensis C. Chr. Ind. Fil. 398. 1906, based on Lindsaea Spruceana Kuhn. Petiole 2-9 em. long, stramineous to pale reddish-brown, the abaxial side angled to rounded; lamina 1-pinnate or bipinnate, 8-16 cm. long, apical pinnae (or apical pinnules of bipinnate laminae) reduced to the small, lanceolate-linear terminal segment which is more or less con- fluent with the lateral segments adjacent to it, rachis of 1-pinnate laminae and pinna-rachises abaxially concolorously angular or sulcate, largest ultimate segments 3 to 3% ‘times as long as broad, 5-9 mm. long, herbaceous, the veins obscure; indusium 0.15-0.25 mm. wide, sub-entire. This species may be only a dwarf form of the former one, L. arcuata. Additional material is needed in order to assess its status properly. Peru. San Martin. Specimens seen: SAN MARTIN: Mt. Guayrapurima, near Tarapoto, Spruce 4023 (B, BM, F, G, GH, K, LE, W). 8. Lindsaea divaricata Kl. Linnaea 18: 547. 1845. Type: British Guiana, Schomburgk 368,8. Fic. 185. Petiole ca. 10-60 cm. long, castaneous to nearly black, the abaxial side rounded; lamina bipinnate, ca. 20-90 cm. long, apical pinnules reduced, or not, to the small to moderately large, triangular-lanceolate terminal segment which is more or less confluent with, or free trom, the lateral segments adjacent to it, pinna-rachises abaxially with con- tinuous pale wings beyond the base, ultimate segments ca. 2% times as long as broad, ca. 10-20 mm. long, herbaceous, the veins usually obscure; indusium 0.10-0.15 mm. wide, entire or subentire. The pinnae which are laxly spreading at an angle of about 45-60 degrees afford a useful character for the distinction of some specimens of this species from some of those of L. bortoricensis. In that species the pinnae are strongly ascend- ing. Mexico to Panama; Lesser Antilles; Guianas and Brazil to Colombia, Venezuela, Bolivia and Paraguay. reo forest and hillside forest, Loreto and Hudnuco, 100- m. 238 ROLLA TRYON Specimens seen: LORETO: mouth of Rio Santiago, Mexia 6133b (uC, US) ; near Iquitos, Klug 69, 1337 (F, NY, US). HUANUCO: Tingo Maria, Tryon & Tryon 5339 (F, U, USM), Aguilar 301, in part (UC). 4. Lindsaea guianensis (Aublet) Dryand. Trans. Linn. Soc. 3: 42. 1797, ssp. guianensis. Fic. ; Adiantum guianense Aublet, Hist. Pl. Guian. 2: 963. 1775. Type: French Guiana, Aublet, p? Petiole ca. 10-60 cm. long, stramineous to rarely castaneous beyond the dark brown to blackish base, the abaxial side rounded; lamina rounded (always so at the base) to concolorously angled or sulcate, largest ultimate segments ca. 2 times as long as broad, ca. 10-12 mm. Fics, 183-188. Fig. 183. Lindsaea arcuata: apical half of fertile pinna, X 1, Brazil, L. B. Smith 1965, GH. Fig. 184. L Spruceana: ba Spruce 4023, GH Fig. 185. L. div ta: A, apex of fertile pinna, X 1, Belin Steinbach 5809, cH; B, base of fertile pinna, X 1, idem. Fig. 186. L. guianensis 89D- guianensis: apical half of pinna, X 1, Colombia, Killip & Smith 148738, GH. Fig. i le stiri * base of fertile pina, X 1, British Guiana, Gleason 603, cH. Fig. 188. @ var. stricta: fertile pinna, X 1, Brazil, Mexia 5486, GH. FERNS OF PERU 239 long, herbaceous, the veins evident; indusium ca. 0.15 mm. wide, entire to usually minutely erose-denticulate. This subspecies has long-acuminate pinnae with the apical pinnules minute (1-2 mm. long); subspecies lanceastrum Kramer, of Brazil and Paraguay, has acute to short-acumi- nate pinnae with the apical pinnules longer (ca. 5 mm. long). Nicaragua; Lesser Antilles; Guianas and northern Brazil to Venezuela, Colombia and Peru. Hillside forest, Hudnuco, 750 m. pecimens seen: HUANUCO: Tingo Maria, Tryon & Tryon 5275 (F, U, USM), 5289 (BM, F, GH, MO, U, US, USM), Allard 21502 (Us). Lindsaea portoricensis Desv. Ges. Naturf. Freunde Berl. Mag. 5: 326, 1811. Type: Herb. Desvaux, P. FG. 187. Petiole 5-50 cm. long, brownish-red to dark brown, the abaxial side rounded; lamina 1-pinnate or bipinnate, ca. 15-50 cm. long, apical pinnules (or apical pinnae of 1-pinnate laminae) reduced to the small, lanceolate terminal segment which is more or less confluent with the lateral segments adjacent to it, rachis of 1-pinnate laminae and pinna- rachises abaxially with pale angles or wings which are irregularly interrupted toward its base, ultimate segments ca. 2 times as long as broad, 5-14 mm. long, firmly herbaceous to chartaceous, wil veins usually evident; indusium ca. 0.3 mm. wide, usually repand-eros The characters of the pinnae which aid in distinguishing some specimens of this species from some of those of L. divaricata are mentioned under that species. Mexico to Guatemala; Greater Antilles; Guianas to Co- lombia, south to Bolivia, Brazil. Phi woods, in sandy soil, Loreto and San Martin, 100- 0 m. ee seen: LORETO: near Iquitos, Tryon & Tryon 5180 ere da MO, U, US, USM); Bersalles, prov. Iquitos, Vargas 11455 (GH). MARTIN: Soritor, Woytkowski 6247 (us). 6. Lindsaea stricta (Sw.) Dryand. Trans. Linn. Soc. 3: 42. 1797. Petiole ca. 5-80 cm. long, stramineous ‘to pale brown or reddish- brown, the abaxial side rounded; lamina 1-pinnate or bipinnate (rarely to tripinnate), ca. 5-70 em. long, apical pinnules (or apical pinnae of 1-pinnate laminae) reduced to the small, elliptic-lanceolate to roundish terminal segment which is more or less confluent with the lateral seg- ments adjacent to it, rachis of 1-pinnate laminae and pinna-rachises abaxially rounded to concolorously angled or sulcate, ultimate segmen 240 ROLLA TRYON roundish or to 114 times as long as broad, herbaceous to usually cori- aceous, the veins mostly obscure; indusium 0.3-0.5 mm. wide, usually strongly erose to lacerate. Mexico to Panama; Greater Antilles; Guianas to Co- lombia, south to Bolivia; Brazil. In forests, San Martin to Puno, 840-2400 m. KEY TO VARIETIES Axes with thick, laterally projecting ridges on the adaxial side, these especially on the upper part of the petiole, on the rachis near the base of the pinnae and at the base of the pinna-rachises. .... 6b. var. parvula Axes lacking such ridges (some intermediates OXISE)... -sccressrenteinae PEERED A eit a var. stricta 6a. Lindsaea stricta var. stricta. Fic. 188. Adiantum strictum Sw. Prod. 135. 1788. Type: Jamaica, Swartz, S-PA, Specimens seen: SAN MARTIN: San Roque, L. Williams 7760 (F, GH); Lamas, L. Williams 6382 (r, US); near Moyobamba, Klug 3418 (B, F, G, GH, K, MO, NY, S, US). CUZCO: Beatriz, Maranura, Biies 894 (US). PUNO: Sandia, Vargas 11850 (GH). 6b. Lindsaea stricta var. parvula (Fée) Kramer, Acta Bot. Neerland. 6: 230. 1957. Lindsaea parvula Fée, Mém. Fam. Foug. 11: 17. 1866. Type: Trini- dad, Germain; isotype: P, Specimen seen: san MARTIN: Pascomayo to Moyobamba, Stiibel 1061 (B). 7, Lindsaea lancea (L.) Bedd. Ferns Brit. India Suppl. 6. 1876. : Petiole ca. 10-50 em. long, stramineous to nearly black, the abaxial side with sharp angles or wings, or obtusely angled to rounded in the basal half; lamina 1-pinnate or bipinnate, ca. 5-50 em. long, apical pinnules SG apical pinnae of 1-pinnate laminae) not or not much t of 1-pinnate laminae and Pinna-rachises abaxially with often pale ridges or wings (the pinna-rachises abaxially rounded at the base), ultimate segments ca. 2 to 3 times as long as broad, 11-45 mm. long, bhai the veins rather evident; indusium 0.2-0.3 mm. wide, entire. KEY TO VARIETIES Lamina bipinnate, or 1-pinnate and the terminal segment usually longer than broad, not very asymmetrical, the pinnae ca. 2 to 24% times FERNS OF PERU 241 as long as broad and the upper ones about half as long as the lower. ; lancea Lamina 1-pinnate, the terminal segment about as long as broad, very asymmetrical, pinnae up to 3 times as long as broad, upper ones little or not reduced (some intermediates exist). ........:::ceeee0 Tb. var. falcata 7a. Lindsaea lancea var. lancea. G. 189. Adiantum lancea L. Sp. Pl. ed. 2, * sgh 1768. Type: Surinam, Seba, Thes. 2: t. 64, f. 7, 8 (see Kramer, op Lindsaea ee Pamala Dryand. Trans. pi ‘Soe. 3: 43. 1797. Type: Grenada, Smeathma Tropical America. Dense forest and rocky woods, Amazonas and Loreto to Junin, 100-1600 m. Specimens seen: AMAZONAS: Campamento, prov. Bagua, Lépez et al. 4237 (GH). LORETO: near mouth of Rio Santiago, Mexia 6207a (GH, UC, US); Iquitos, Killip & Smith 27021 (Ny, US). HUANUCO: Tingo Maria, Tryon & Tryon 5291, 5296 (BM, F, MO, U, US, USM), 5297 (F, U), Allard 21209, 21369, 21491 (US). JUNIN: La Merced, Macbride 5715 (F, US), Schunke 96 (F), 458 (F, US), Soukup 1079 (F) ; Puente Perené, Coronado 263 (GH, UC); above San Ramén, Schunke A233 (us), Killip & Smith 24698 (NY, Us); Rio Pinedo, n. of La Merced, Killip & Smith 30676 (Us), Santa Rosa, Pichis Trail, Killip & Smith 26162 (US); Quimiri, Tarma, Esposto (USM). 7b. Lindsaea lancea var faleata (Dryand.) Rosenst. Hedwigia 46: 79. 1906. bila 182, 190. a Jaleate Dryand. Trans. Linn. Soc. 3: 41. 1797. Type: French pe a Aublet, P? Panama; Guianas to Colombia, south to Bolivia; Brazil. Forests, Loreto and Junin, 100-1300 m. Specimens seen: LORETO: Mishuyacu, near Iquitos, Klug 379, 1510 (F, NY, US); Timbuchi, alto Rio Nanay, L. Williams 954 (F); Pongo dd. Marner riche, Tessmann 4850 (B). JUNIN: La Merced, Killip & Smith 23966, in part (F, US). 8. Lindsaea Schomburgkii KI. Linnaea 18: 545. 1845. Type: British = Schomburgk 278, 8B. Fic. 191 ca. 20-85 mm. lon g, firmly herbaceous to coriaceous, veins evident; indusium 0.2 mm. a entire to minutely esi 242 ROLLA TRYON British Guiana, Venezuela, Colombia, Peru and Brazil. In wet, open sand, Loreto and San Martin, 200-1400 m. Specimens seen: SAN MARTIN: Pajonal, between Rio Negro and Rioja, Stiibel 1059 (B); Rio Negro, Woytkowski 6206 (GH). LORETO: Chamicuros (prov. Loreto, dist. Parinari), Bartlett (w). 194B I94A 195 Fics. 189-196. Fig. 180 _Tindsaes lances var. lancea: fertile pinna, X %, Bess Krukoff 7302, cu. Fig. . L. lancea var. viduals fertile lamina, X 14, Brazil, Krew i koff 7298, cu. Fig. 191. ‘. shears ctl central _— of lamina (14), X 4 wi Spruce 2648, cH. Fig. 192. L. latifrons: apex of fertile lamina, X 1%, Penge Klug 2899 GH. Fig. 193. L. he portion ‘of fertile —. Oh), X %, Peru Schunke . istoem- . A234, GH. Fig. 194. Oleandra articulata: A, fertile leaf attached to rhizome, x 600, ion of sterile la %, idem. wba H O. Lehmannii: leaf attached to rhizome, X 14, Peru, Killip & Smith 25870, G . pilosa: leaf attached to rhizome, X 44, Colombia, Schultes & Cabrera 1486h ot FERNS OF PERU 243 9. Lindsaea latifrons Kramer, Acta Bot. Neerland. 6: 256. 1957. Type: Balsapuerto, Peru, Klug 2890, US; isotypes: B, BM, F, G, GH, K, MO, NY, s. Fie. 192. Petiole 15-45 cm. long, stramineous to pale brown, the abaxial side angled to palely ridged; lamina usually 1-pinnate, rarely bipinnate, ca. 15-40 cm. long, apical pinnae (or apical pinnules in bipinnate laminae) not or slightly reduced, the large, triangular-cuneate to crescent-shaped terminal segment free, rachis and pinna-rachises abaxially angled or palely ridged, ultimate segments 3 to 4 times as long as broad, 6-9 cm. long, firmly herbaceous, the veins evident ‘to obscure; indusium 0.1-0.2 mm. wide, entire. The next species, L. hemiglossa, is closely related to L. latifrons and they are both discussed under that species. Peru. Dense forest, Loreto, 150-350 m. Specimens seen: LORETO: Balsapuerto, Killip & Smith 28596 (US), 28614 (NY, US), Klug 2890 (B, BM, F, G, GH, MO, NY, S); Tierra Doble, alto Rio Nanay, L. Williams 1068 (F). 10. Lindsaea hemiglossa Kramer, Acta Bot. Neerland. 6: 257. 1957. Type: Schunke Hacienda, above San Ramé6n, Peru, Schunke A234, UC; isotypes: GH, Us. Fic. 193. Petiole ca. 10-40 em. long, stramineous to pale brown, the abaxial side angled to palely ridged; lamina 1-pinnate, ca. 20-40 em. long, apical pinnae little reduced, the large, triangular-cuneate to crescent- shaped terminal segment free, rachis abaxially ridged, often palely so, ultimate segments 4 to 6 times as long as broad, ca. 4-9 cm. long firmly herbaceous, the veins mostly obscure; indusium ca. 0.1 mm. wide, entire. When dry, the pinnae of L. hemiglossa are usually bright green, while those of the previous species, L. latifrons, are dark olivaceous to brown. These two species are represented by few collections and additional materials may indicate that they are variants of a single species. Ecuador and Peru. In forests, San Martin and Junin, 1200-1800 m. Specimens seen: SAN MARTIN: Tarapoto, Spruce (K). JUNIN: Chan- chamayo, Schunke 102 (F, US), 511 (F), 806 (F, US); San Ramon, Killip & Smith 24571 (Fr, NY, US) ; La Merced, Macbride 5624 (F, US); Schunke Hacienda, above San Ramén, Schunke A234 (GH, UC, Us). 244 ROLLA TRYON TRIBE 11. OLEANDREAE. 34. OLEANDRA Cav. Ann. Hist. Nat. 1: 115. 1799. Type: Oleandra nertiformis Cav. ia dgresiens the rhizome slender, long-creeping or climbing, scaly, ring the leaves in loose clusters or at wide intervals; leaves small to sabeatdchanel: the lamina simple and entire, glabrous to pubescent, the costa slightly scaly, the petiole articulate at a prominent joint, the portion below the joint (the phyllopodium) short or long, similar to the portion above the joint or scaly like the rhizome, veins free; sori Posastieh, borne rather irregularly on the veins, not genie: covered by an orbicular to reniform indusium which is attached a the sinus. — 9 species in America. Maxon, W. R. The American species of Oleandra. Contrib. U. S. Nat. Herb. 17: 392-398. 1914. The treatment of this genus has been adapted from that of Maxon cited above. Some of the species of Oleandra have an erect, short-climbing stem which, with its branches, often produces a shrubby growth. Among the Peruvian species, O. Lehmannii and O. pilosa have this habit which is unique in this genus. The petiole is articulate, the portion of it below the joint being called the phyllopodium. KEY TO SPECIES a. Rhizome widely creeping, its scales widely spreading; phyllopodia mostly 5-30 mm. long, slender and naked like the petiole above the joint. b. b. Lamina and indusia glabrous 1. O. articulata b. Lamina pubescent, indusia wahewcent on the surface, not ciliate on the margin. 2. O. he a. Rhizome ascending or erect-climbing, its scales closely appreee™ ce. Lamina glabrous, phyllopodia mostly 10-20 mm. long, slender Br naked like the petiole above the joint, indusium glabrous. +..." 3. O. Lehmannit ~ ‘Lamina pubescent, at least in part, phyllopodia 1-3 (-5) mm. loné; stout and at first scaly like the oe indusium pubescent on the surface and with ciliate margins. .......csssesseeseeeeee 4, O. pilosa asvececceesoeee® a Oleandra articulata (Sw.) Presl, Tent. Pterid. 78. 1836. FIG Aspidium articulatum Sw. Jour. Bot. Schrad. 1800 (2): 30. 1801. e: Martinique, Plumier, Fil. t. 186 (See Maxon, Contrib. U. sie Herb. 17: 394, for a discussion of the choice of the type set the corre application of the name). FERNS OF PERU 245 Aspidium nodosum Willd. Sp. Pl. 5: 211. 1810, seg Tak ille- git.). Type: the same as that of A ium a Oleandra nodosa (Willd.) Presl, Tent. Pterid. 7 tan ‘(illegit.) Rhizome widely creeping, scales spreading; phyllopodia ca. 5-30 mm. long, slender, without scales; lamina glabrous, ca. 10-35 cm. long, about 1/4th to 1/10th as broad, caudate to rather abruptly acute at apex, gradually narrowed to usually broadly cuneate, rarely sub- cordate at the base; indusium glabrous. The costal scales of this widely distributed species are most commonly ovate-deltoid to long-deltoid and acuminate. The Peru and Bolivian material I have seen has them subu- late to narrowly lanceolate and acuminate. Scales of the same shape, however, occur in other parts of the range of O. articulata, either by themselves, or mixed with the more common kind; there is no evidence that the variation has taxonomic importance. Tropical America. Dense forest, Junin, 1400-1900 m. Specimens seen: JUNIN: Enefias, Pichis Trail, Killip & Smith 25756 (F, NY, Us); Schunke Hacienda, above San Ramon, Killip & Smith 24878 (NY, US). 2. Oleandra hirta Brack. in Wilkes, U. S. Expl. Exped. 16: ar Atlas t. 29. 1854. Type: Organ Mountains, Brazil, Brackenridge, U Rhizome widely creeping, scales spreading; phyllopodia ca. 5-30 mm long, slender, without scales; lamina pubescent, ca. 10-20 cm. long, spr 1/5th as broad, caudate to abruptly acute at the apex, narrowly to broadly cuneate at the base; indusium pubescent on the surface, glabrous on the margin. This species is not known from Peru but I have seen a specimen from Bolivia (Tate 1152, NY) and it may well be discovered in Puno or Cuzco. Brazil, Venezuela and Bolivia. 3. Oleandra Lehmannii Maxon, Contrib. U. S. Nat. Herb. 17: 395. 1914. Type: Colombia, Lehmann XLII, us! Fic. 195. Rhizome ascending or erect and climbing, scales closely appressed; Phyllopodia ca. 10-20 mm. long, slender, without scales; lamina gla- brous, ca. 15-25 em. long, about ee as broad, peor tapered at apex and base; indusium glabrou 246 ROLLA TRYON The earlier name, Oleandra micans Kze., discussed under O. pilosa, may apply to this species. Venezuela and Colombia to Peru. Dense forest, Junin, 1800 m. Specimens seen: JUNIN: Dos de Mayo, Pichis Trail, Killip & Smith 25870 (F, GH, NY, US). 4. Oleandra pilosa Hook. Gen. Fil. t. 45 : & text. 1840 Type: Berbice, British Guiana, Schomburgk 416, K! 196. spidium pendulum Splitg. Tijds. Nat. Gesch. 7: 412. 1840, not Raddi, 1819. ao: Surinam, Splitgerber; British Guiana, Schom- burgk 416; isotype: Rhizome ascending or erect and climbing, scales closely appressed; phyllopodia 1-3 (to rarely 5) mm. long, stout, at first scaly like the rhizome; lamina pubescent, sometimes sparingly so, ca. 10-35 em. long, about 1/5th to 1/10th as broad, caudate, acuminate or acute at the apex, gradually narrowed to rather rounded at the base; indusium pubescent on the surface and with a ciliate margin. The pubescence on the lamina is variable in its distribu- tion. Most specimens are somewhat pubescent on all parts but others may be partially glabrous, especially on the leaf tissue beneath. Oleandra micans Kze. Bot. Zeit. 1851 : 346. (Type: Hual- laga superior, Peru, 1830, Poeppig, (Diar. 1958), LZ, des- troyed) may be a synonym of this species, or it might be an earlier name for O. Lehmannii. The description mentions only one character of value in Seen the aes of the name and that is “caudice . . adpresse palea but this serves only to place it in the general allianes of O. pilosa. Guianas to Colombia, south to Bolivia. In dense forest, Hudnuco to Puno, 600-1300 m. Pigerecioag seen: HUANUCO: Tingo Maria, Allard 21215, 21999 (GH, Us). : Rio Marcapata, de Lantreppe (NY); San Pedro, prov: Pavcartato, Vargas 6801 (UC, US). PUNO: Tatanara, Lechler 2539 B, K INDEX INDEX TO NAMES Accepted names are in roman type, synonyms are in italic Acrostichum 9, 210 glanduliferum 173 aureum 210 aureonitens 86 guianense 238 calomelanos 68 Henslovianum 164 chrysoconium 75 var. macrosora 164 chrysophyllum 65 humile 161 citrifolium 224, 225 imbricatum 177 daneaefolium 210 i Gillianum 86 Kalbreyeri 154 lanuginosum 91 Kaulfussii 149 Marantae 105 Killipit 161 rufum 82 laetum 164 scariosum 91 sinuatum 110 latifolium 151 tartareum 73 lobatum 165 tereticaulon 121 lucidum 148 thalictroides 87 var. Poeppigianum 148 trifoliatum 65, 77 lunulatum 180 Adiantopsis macrocladum 159 chlorophylla 98 macrophyllum 144 paupercula 88 Mathewsianum 141 radiata 99 Mexiae 159 ternata 99 microsorium 176 Adiantum 9, 135 Moorei 169 alarconianum 149 obliquum 150 amabile 169 Orbignyanum 175 anceps 143 palmatum 178 boliviense 169 patens 165 Capillus-Veneris 135, 166 pauperculum 88 cayennense 161 pectinatum 162 ileae 154 pedatum 165 chilense 172 peruvianum 141 var. hirsutum 173 petiolatum 149 colpodes 169 phillipense 180 concinnum 168 phyllitidis 148 erenatum 171 platyphyllum 144 Crespianum 143 Poeppigianum 148 cuneatum 169 Poiretii 171 decorum 169 var. hirsutum 173 defiectens 180 f. hirsutum 173 delicatulum 180 var. Poiretii 172 digitatum 178 var. sulphureum 173 dolabriforme 180 polyphyllum 162 exci pulverulentum 157 filiforme 180 Raddianum 169 flagellum 180 rad: iatum 99 fructuosum 156 Remyanum 169 248 INDEX rhizophyllum 180 Asplenieae 9 rubellum 16 Asplenium 9 rufopunctatum 169 tomentosum 84 Ruizianum 175 Athyrium 9 sealare 146 Blechneae 9 serratodentatum 152 Blechnum 9 sessilifolium 164 Blotiella 9, 39 speciosum. 178 glabra 39 Steerei 176 Lindeniana 39 strictum. 240 Bolbitis 9 subaristatum 180 Ceratopteris 9, 87 subcordatum 143 pteridoides 87 subvolubile 167 thalictroides 87 sulphureum 172, 173 Ceropteris 65 terminatum 160 adiantoides var. peruviana 70 tetraphyllum 155 Cheilantheae 9, 40 thalictroides 172 Cheilanthes 9, 88 tinctum 169 andina 96 tomentosum 157 Borsigniana 116 Veitchianum 175 Brandegei 97 villosissimum 155 chlorophylla 98 villosum 152 digitata 94 Weatherbyanum. 173 elegans 90 Williamsii 173 farinosa 100 Allosorus acclivis 184 fasciculata 94 Ananthacorus 9, 219 fractifera 96 angustifolius 219 Fraseri 114 Anetia 225 glandulosa 185 Anetium 9, 224 glauca 103 citrifolium 224, 225 hostilis 38 Anisosorus 186 incarum 93 hirsutus 186 lendigera 90 ogr OL Macleanii 96 chaerophylla 81 marginata 103 leptophylla 81 Mathewsii 94 Antrophyum 220 micromera 97 brasilianum 224 micropteris 88, 94 cajenense 222 Moritziana 98 discoideum 222 myriophylla 90 guayanense 224 notholaenoides 97 lineatum 222 obducta 111 subsessile 224 obtusata 36 spidium articulatum 244 Orbignyana 102 biserratum 229 ornatisst eminens 227 parallelogramma 35 nodosum 245 paupercula pectinatum 233 pilosa 95 pendulum 230, 246 xX pruinata 96 trapezoides 233 Poeppigiana 100 tuberosum 230 pruinata 94 tripinnata 91 Ctenitis 9 Cyclopeltis 9 Cystopteris 9 Davallia arborescens 26 na 26 Davallieae 9, 225 Dennstaedtia 9, 18 cicutaria 19 glauca 20 globulifera 21 Lambertiana 20 Mathewsii 26 obtusifolia 25 Dicranoglossum 9 Didymochlaena 9 nobilis 131 mata 126 pedata var. multipartita 131 var. palmata 126, 131 a 46 Gymnogramma amie 51 aureonitens 45 Ballivianii 70 calome lanos var. aureoflava 69 var. denudata 68 elongata 40, 51 Mathewsii 42 250 INDEX mohriaeformis 42 var. imbricata 63 ochracea 70 peruviana 56 Orbignyana 49 pulchra 53 Orthiopteris 75 rotundifolia 55 Pearcei 80 sealaris 61 peruviana 73 Scammanae 55 prehensibilis 49 Lindsaea 9, 235 pumila 211 arcuata 236 reniformis 64, 65 divaricata 237 rufescens 42 falcata 241 Ruiziana 47 guianensis ssp. guianensis 238 Stuebelii 44 ssp. lanceastrum 239 tartarea var. aurata 75 hemiglossa 243 Gymnopteris 9, 82 lancea 235, 240 rufa 82 var. falcata 241 tomentosa 84 var. lancea 241 Hecistopteris 9, 211 latifrons 243 umila 211 parvula 240 Hemionitis 9, 84 portoricensis 239 brasiliana 224 Schomburgkii 241 cajenensis 222 Spruceana 237 lineata 220, 222 stricta 239 palmata 84, 85 var. parvula 240 pinnata 86 var. stricta 240 Histiopteris 9, 186 tarapotensis 237 incisa 187 trapeziformis 235, 241 vespertilionis 186 Lindsaeeae 9, oderris 9 Litobrochia horizontalis 209 Hypolepis 9, 32 Lonchitis 9, 186 bogotensis 36 hirsuta 186 Buchtienii Lonchitis 39 flexuosa 36 glabra 39 hostilis 38 Lindeniana 39 obtusata 36 pedata 204 parallelogramma 35 tenuifolia 32 parviloba 38 Microlepia 9, 29 pteroides 36 lun 9 rigescens 35 Wercklet 28 a9 N idan oulaiin 227 Ithycaulon 30 biserrata 229 Jamesonia 9, 53 cordifolia 230 Alstonii 58 ev. Duffii 232 blepharum 64 Ffit 232 boliviensis 59 exaltata 225, 228 ev. Bostoniensis 229 var. bostoniensis 229 hirsutula 230 var. glutinosa 63 intermedia 227 occidentalis 233 pectinata 233 pendula 230 rivularis 227 Notholaena 9, 105 arequipensis 108 riensis 113 Brackenridgei 107 Buchtienii 115 candida var. lutea 117 cantangensis 111 chrysophylla 121 ferruginea 113 Fraseri 114 Lillot 108 lonchophylla 110 Marantae 105 mollis 116 nivea 118 var. flava 121 var. nivea 120 var. oblongata 120 var. tenera 121 obducta 111 peruviana 107 sinuata var. sinuata 110 amosa Stuebeliana 118 articulata 244 hirta 245 Lehmannii 245 micans 246 neriiformis 244 modosa 245 Oleandreae 9, 244 Orthi INDEX scalaris 185 viscosa 184 Parkeria pteridoides 87 Pellaea 9, 121 atropurpurea 121 ns 1 251 dealbata var. Stuebeliana 118 1 sagittata var. cordata 125 var. sagittata 1 tenera 121 ternifolia var. ternifolia 122 var. calomelanos 68 var. ochracea 70 X trifoliata 79 var. peruviana 70, 73 chrysoconia 75 chrysophylla 65 ferruginea 76 x trifoliata 79 ochracea 70 Polypodiaceae 9, 18 Polypodieae 9 252 astrolepis 51 globuliferum 21 leptophyllum 81 rivulare 227 var. arachnoideum 182 var. caudatum 182 arachnoideum 182 Pteris 9, 188 sect. Histiopteris 186 altissima 201 angustifolia 219 aquilina 181 arachnoidea 182 atr chelate 121 3 ca 192 decomposita 194 INDEX deflexa 195 denticulata 209 edentula 196 elata 201 esmeraldense 202 horizontalis 209 imbricata 63 incisa 187 Jamesonii 192 lineata 211, 2138 livida 204 lonchitoides 186 longifolia 188, 191 lucida 148 lutea 117 ovata 125 palmata 126, 131 pedata 204 peruviana 122 petiolulata 209 podophylla 204 polita 195 a 205 a eon 205 ruffa 82 sagittata 123 sealaris 185 selerophylla 202 speciosa 207 sulphurea 116 ternifolia 122 trialata 204 vespertilionis 186 vittata 191 abet roe apa age 218 Pterozo sate eT meen a prey moluccanum 30 Tecta aerate 229 Thelypteris 9 Trachypteris 9, 86 aureonitens 86 INDEX pinnata 86 Trichomanes flaccidum 18 65 graminifolia 215 lanceolata 222 latifolia 219 lineata 211, 213, 222 Mortiziana 217 Orbignyana 218 remota 217 Ruiziana 218 Vittarieae 9, 211 Woodsia 9 Woodsieae 9 CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C. Rollins and Robert C. Foster NO. CXCV EVOLUTION OF SELF-COMPATIBILITY AND RACIAL — | DIFFERENTIATION IN LEAVENWORTHIA (CRUCIFERAE) By Davin G. LLoyD A TAXONOMIC REVISION OF CREMOLOBUS (CRUCIFERAE) By Kutpip R. KHANNA AND REED C. ROLLINS Published by 3 ae THE GRAY HERBARIUM OF HARVARD UNIVERSITY oe CAMBRIDGE, MASS., U.S.A. 1965 CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY dited by Reed C. Rollins and Robert C. Foster NO. CXCV EVOLUTION OF SELF-COMPATIBILITY AND RACIAL DIFFERENTIATION IN LEAVENWORTHIA (CRUCIFERAE) BY Davin G. LLOYD A TAXONOMIC REVISION OF CREMOLOBUS (CRUCIFERAE) By Kuuprie R. KHANNA AND REED C. ROLLINS Published by THE GRAY agginircnneD OF HARVARD UNIVERSITY MBRIDGE, MASS., U.S.A. Issued April 30, 1965 - EVOLUTION OF SELF-COMPATIBILITY AND RACIAL DIFFERENTIATION IN LEAVENWORTHIA (CRUCIFERAE) DAVID G. LLOYD* Charles Darwin (e.g., 1876) showed that, in the majority of plant species he studied, progeny from crossing two plants are usually superior to those resulting from self- fertilization, and that many plant species have devices which ensure that the flowers are not regularly self-fertilized. In the century since Darwin’s work, it has become apparent that one of the commonest means whereby angiosperm species restrict self-fertilization is through the operation of self-incompatibility systems. Self-incompatible species have been found in at least 78 families and occur in every major phylogenetic line (Pandey, 1960). Brewbaker and Majumder (1961) estimate, from a compilation by Fryxell (1957), that approximately 250 of 600 genera studied con- tain one or more self-incompatible species. Many genera contain both self-incompatible and self- compatible species (Fryxell, 1957; Stebbins, 1957). Within the Cruciferae, of 182 species studied, 80 are self-incom- patible and 102 are self-compatible, and many genera con- tain both self-incompatible and self-compatible species (Bateman, 1955). A change from self-incompatibility to self-compatibility, or vice versa, has apparently occurred many times. The majority of recent authors (e.g., Mather, 1943a; Crowe, 1964) believe that the change has been from self-incompatibility to self-compatibility in most, if not all, cases where multi-allelic systems are involved. In many genera, the self-incompatible species have adap- tations to insect pollinations such as nectaries and con- spicuous flowers, whereas the self-compatible species have partially or completely lost these adaptations and have evolved devices to ensure the success of self-pollination. In addition, the acquisition of self-compatibility may have secondary effects on the distribution, variation, and evo- lutionary potential of such species. It is often said that self- 1Present address: Botany Department, University of Canterbury, Christchurch 1, New Zealand. 4 LLOYD compatible species are able to adapt more closely to the environment in which they live, but have a reduced genetic variability and a more limited evolutionary potential. In some groups too, self-compatible species are more widely distributed than their self-incompatible relatives (Baker, 1953). This has led to considerable speculation on the selective advantage of the change in the breeding system. The nine- teenth century naturalists observed that self-fertile species are more common at high altitudes and high latitudes in Europe, and that many of these species have devices to en- sure self-pollination if cross-pollination fails; they believed that self-fertilization enabled a plant to set sufficient seed when unfavorable weather conditions or a lack of insect visitors restricted cross-pollination. Since the development of the theory of genetic systems (Darlington, 1939; Mather, 1943a) this explanation has been used less, and many authors have placed more emphasis on the apparent ability of self-compatible races to adapt more closely to their m- mediate environment. Rollins (1963) has recently given an account of the systematics and evolution of Leavenworthia, which includes a description of the evolution of self-compatibility and asso- ciated morphological characters in certain populations of L. alabamica and L. crassa and at least two other phyletic lines in the genus. The existence in both Leavenworthia crassa Rollins and L. alabamica Rollins of self-incompatible and self-compatible geographical races in a very small aree of northwest Alabama (Rollins, 1963) suggests that the change in the breeding system has occurred recently in these species. Both species therefore offered an exceptional isc tunity to study the factors influencing the evolution of self- compatibility and the effects of the change in the breeding system on the morphology and distribution of the self- compatible populations. F The present study on L. crassa and L. alabamica oe revealed a much more complicated pattern of geograpnit variation, largely associated with the evolution of sell- compatibility, than was previously known. Fifteen ee graphic races of L. crassa and four geographic races a - alabamica were distinguished amongst the population EVOLUTION IN LEAVENWORTHIA 5 grown in a greenhouse, and these are informally described below. The self-compatible races differ in the degree to which they have lost adaptations for insect pollination and have gained structural devices which increase the efficiency of self-pollination. The morphology, ecology and distribution of the self-incompatible and self-compatible races of L. crassa and L. alabamica have been studied in detail, and factors affecting the number of insect visitors in the natural habitat of Leavenworthia have been examined. GENERAL METHODS The projects reported in this study were done on plants growing in nature and on plants grown in a greenhouse in Cambridge, Massachusetts, from seeds collected in nature. Wherever possible measurements were made on both ma- terial collected in nature and on material grown in the greenhouse under relatively uniform conditions. L. crassa and L. alabamica are winter annuals endemic to northwest Alabama. The seeds germinate in the fall, and the plants grow slowly throughout the winter and flower from approximately March 10 to April 20 or later, de- pending upon how rapidly the spring rains diminish and the temperature increases. The fruits mature and the seeds are shed at the end of April and beginning of May. Field studies were conducted in the flowering seasons of 1961, 1962 and 1964. Thirteen of the nineteen races of L. crassa and L. alabamica described below are restricted to a small area approximately 414 miles (from west to east) by 4 miles (from south to north) near Massey’, Morgan county, Alabama (map 1). This area, referred to henceforth as the Massey-Lebanon district, was searched and studied more intensively than the areas to the west, where the remaining races of L. crassa and L. alabamica are scattered over a much larger area. The greenhouse material was grown in two lots, from November 1961 to March 1962, and from January to June 1963. These two plantings are referred to as the 1962 and 1963 greenhouse material, respectively. The 1962 material 2The small village of Massey is the same as that referred to as McKendry by Rollins (1963); the name has recently been changed. 6 LLOYD included plants from 13 populations belonging to seven races of L. crassa and three races of L. alabamica. The 1963 material included plants from 30 populations from all of the described races of both species. The seeds were germinated in a layer of peat moss over soil in three inch clay pots, and the seedlings were trans- planted to three inch pots, and later to four inch pots as their growth required it. Despite extreme care considerable difficulty was experienced in growing the plants, and many plants, particularly of the 1963 material, became unhealthy in the later stages of flowering. Only plants on which less than five per cent of the flowers were abnormally arrested in their development were measured ; these were considered to be reasonably representative of growth under conditions which were relatively uniform and close to optimal for the greater part of the lives of the plants. The races were recognized on the basis of differences expressed in the greenhouse. A number of measurements were made on the plants of each population grown in the greenhouse, and features distinguishing the races were noted. Many of the races are not distinguishable under natural conditions, and several of the races were not even suspected of being distinct until they were grown in 1963. The identification of populations which were not grown It the greenhouse is based on observations and measurements made in the field in 1962. In some cases the racial identity of these populations is tentative: it is quite possible that additional races might be distinguished if more populations were grown in a greenhouse. The two westernmost races of L. alabamica have not been grown in the greenhouse — Se€ undescribed races of L. alabamica below. Voucher specimens of plants collected in nature from one or more populations of each race of L. crassa and L. ala bamica, and of one plant of every population grow? in the greenhouse in 1963, have been deposited in the Gray Her- barium of Harvard University. Insects collected on Leaven- worthia flowers have been placed in the Museum ° Comparative Zoology, Harvard. Negatives of photographs of young plants, flowering plants, leaves, flowers and fruits of all the described races of L. crassa and L, alabamice have been deposited in the negative collection of the Gray Her- barium EVOLUTION IN LEAVENWORTHIA 7 THE RELATIONSHIPS BETWEEN L. ALABAMICA, L. CRASSA AND OTHER SPECIES OF LEAVENWORTH Leavenworthia is a small, well-defined and isolated genus of the Cruciferae confined to the southeastern United States. The genus most closely related to Leavenworthia is possibly Selenia Nutt., but the relationship is not close (Rollins, 1963). All of the species are highly specialized in their ecological requirements; they are winter annuals whose natural habi- tat is confined to limestone, dolomitic limestone or dolomite outcrops with shallow soils unable to support forest but with sufficient moisture in the fall and spring to enable the Leavenworthia plants to complete their reproductive cycle (Quaterman, 1950; Rollins, 1963). These areas are known as cedar glades, since the dominant tree surrounding them is the ‘red cedar’, Juniperus virginiana L. Amongst the seven species of Leavenworthia there are three basic chromosome numbers, n=11, 15 and 24 (Rollins, 1963). Each chromosome level has probably been attained only once by the living species, so that species with the same chromosome number represent natural groups within the genus. Three species, L. stylosa Gray, L. torulosa Gray and L. uniflora (Michx.) Britton, have n=15 chromosomes. The center of distribution of these three species (and of the genus as a whole) is the Central Basin of Tennessee. Within the Central Basin these three species are sympatric and often grow intermingled amongst each other. Leavenworthia stylosa, the most primitive species of the three, is self- incompatible throughout its range and is wholly confined to the Central Basin. Leavenworthia torulosa is self-compati- ble and has lost many of the adaptations to insect polli- nation: it has a wider range than L. stylosa in Tennessee and also occurs in southern Kentucky. Leavenworthia uni- flora, the most highly autogamous species in the genus, also has the widest distribution of all Leavenworthia species. The species at the n=15 chromosome level have thus become secondarily sympatric after their prior speciation in geographic isolation. The reproductive isolating mecha- nisms between them are complete — no hybrids have ever been found in nature, and the species cannot be artificially crossed (Rollins, 1963). 8 LLOYD The three species with 11 chromosome pairs, L. alabamica, L. crassa and L. exigua Rollins, offer a strong contrast to the species with 15 chromosome pairs in the pattern of distribution and in the hybridization of the species. The glade populations of these species are completely allopatric, although the distribution of the races of L. alabamica and L. crassa is very complicated. L. exigua is the most ad- vanced species in this group in regard to its breeding system: all of the populations tested are self-compatible, and have to a large extent lost the ancestral adaptations to cross-pollination by insects, such as large flowers and a noticeable odor (Rollins, 1963). Leavenworthia exigua is also the most widespread of the species with 11 chromosome pairs, occurring in the Central Basin, northwest Georgia, central Alabama and north Kentucky. Leavenworthia exigua is less closely related to L. crassa and L. alabamica than the latter species are to each other, and it cannot be crossed with either L. crassa or L. alabamica (Rollins, 1963). Leavenworthia crassa and L. alabamica are clearly the two most closely related species in the genus. They differ only in several silique characters. The siliques of L. crassa never exceed 13 mm in length and are fleshy, whereas the siliques of all races of L. alabamica exceed 17 mm in length under favorable conditions and are not fleshy. The number of seeds in each silique depends upon the environmental conditions and varies between the races in both species, but there are often more than twelve seeds per silique in L. alabamica, and usually less than ten seeds per silique 10 L. crassa. In addition adjacent seeds in each locule usually overlap in L. crassa, but in L. alabamica the seeds in each locule are well spaced, or barely touching (fig. 5). Apart from these fruit characters there are no constant differences between the two species, and the self-incompatl- ble races of L. crassa and L. alabamica are otherwise almost identical. The close relationship between L. crassa an L. alabamica is also shown by the fact that they hybridize easily on cultivated fields where they have met, and unde ‘ experimental conditions, to produce vigorous and fertile offspring (Rollins, 1963). The siliques of L. crassa are the most distinct of any species in the genus. All other species have much longer EVOLUTION IN LEAVENWORTHIA 9 siliques in which the seeds are well spaced, as in L. ala- bamica. Indeed, L. alabamica was at first confused with both L. stylosa and L. aurea and was not separated from L. stylosa until very recently (Rollins, 1963), It seems proba- ble that the unique silique characters of L. crassa are de- rived and have arisen in the relatively recent divergence of L. crassa from L. alabamica. The geographical distribution of L. crassa and L. ala- bamica provides some additional evidence that L. crassa has diverged only recently from a self-incompatible ancestor of L. alabamica, perhaps within a few miles of its present very restricted distribution (the most distantly separated glade sites of L. crassa are only 11 miles apart). Leavenworthia crassa is confined to southeast Lawrence Co. and southwest Morgan Co., in the Moulton Valley of northwest Alabama (map 1). Leavenworthia alabamica has a somewhat wider distri- bution than L. crassa, although the most distantly separated glade sites of L. alabamica are only about 62 miles apart. Populations of L. alabamica occur at the western end of the Moulton Valley in Franklin County and the adjacent western half of Lawrence County, and at the eastern end of the Moulton Valley in southwest Morgan Co. In addition there are several populations in the Tennessee Valley (Colbert Co.). The two species do not occur anywhere in the same glade although the distribution of L. crassa lies entirely within that of L. alabamica, and within the Massey-Lebanon district of Morgan County the glade populations of L. crassa are almost entirely surrounded by those of L. alabamica, Ap- parently both species have migrated slowly over the Moulton Valley as suitable cedar glades have shifted in position. It is remarkable that the two species have not met in any of the present glades, and it seems unlikely that both L. crassa and L. alabamica could have migrated into the Moulton Valley region and have maintained their identity in close proximity during a long period of slow wanderings around the valley. I believe instead that L. crassa diverged from L. alabamica after the ancestors of the two species reached the Moulton Valley, and perhaps after L. alabamica had spread throughout the length of the valley that it presently occupies. LLOYD ee . . * *. « : se . Ulsid 1818B0D = ‘d'D “(TT “SID ‘SPG6T) Aedaey Wioay UMBAIp a1B SSdOUTAOAd OTydBIFZOISAYd oy, °*(g 03 % SdwU 9a8) ba ge jo SooBI ZG PUB DssDL9 *T JO SIVVI TT SULBJUOD OLI}SIP UOUBGa]-ABSSBI, BY} ‘BaIB payod}eY-sso19 ayy, “HW apisaq payeoiput aovt ay} Aq perdnoso st vote payoyey Youy "vureqeyy 4seAyWOU UI DEsDLa “JT puB voLUDQnID “rT JO wornqiysiq ‘T AVI pssoi9*y FY poi1wpqnjo "7 Fy) Ps /) en. ARF : teas ell Ayune> uljyunsy OO XK KKK SKK eveteces Wotetetetererere RRR KR eveteterecenere SKS Woteteterererere atetetorereee Weestatetetetatet Weteteterecerens SST foretere! ae. = < Aas w = - < -_ Q COT) Ayuno> ges Re oe a2Uua1MD4 40) N SO 41> 1- eS SASSO reat Ayun0> ~—s upBuow t e DIquinosny Ayuno> j4aaqjo> EVOLUTION IN LEAVENWORTHIA 11 One species of L thia, L. aurea Torrey, has 24 pairs of chromosomes; it is a self-compatible polyploid most closely related to L. exigua and L. alabamica (Rollins, 1963). L. aurea has a disjunct distribution in Oklahoma and Texas, completely outside the range of the other species of the genus. The species of Leavenworthia are thus spread over a considerable number of discontinuous areas in ten states. Leavenworthia has extremely limited powers of dispersal (see below) and it is highly probable that members of the genus must once have occurred in the intervening areas between those presently occupied. Moreover, the widely different diploid chromosome numbers (n=11 and n=15) and the isolated polyploid, L. aurea (n=24), which cannot be derived directly from the existing diploid chromosome numbers, indicate that Leavenworthia has previously evolved forms which are now extinct. It is thus impossible to trace the full history of speciation and migration in the genus. Nevertheless, from the distribution of the species of Leavenworthia and of other cedar glade plants one can make certain suggestions concerning the geographical and evo- lutionary history of the ancestors of L. crassa and L. alabamica. The most important question in this regard is whether the common ancestor of L. crassa and L. alabamica diverged from the other species in Alabama or whether it reached the area secondarily and became extinct elsewhere. I believe that several lines of evidence indicate that the ancestor of L. alabamica and L. crassa, (and, independently, L. exigua and L. uniflora) migrated into Alabama from the north. The area occupied by Leavenworthia can be broadly divided into a western and an eastern section. The western section in Missouri, Arkansas, Oklahoma and Texas occurs within the Interior Highlands, except for the populations of L. aurea in Texas, which are in the Coastal Plain Province (Fenneman, 1938). The western section is occupied only by L. uniflora (n=15) and L. aurea (n=24). The species in the eastern section from Ohio and Indiana to Alabama and Georgia occur mostly in the Interior Low Province, with peripheral populations of L. exigua, L. torulosa and L. uniflora in the Appalachian Plateau Province. All of the 12 LLOYD Species except L. aurea occur in the eastern section. The eastern and western sections are at present separated from each other by the Mississippi Alluvial Plain which is generally quite unsuitable for occupation by Leavenworthia. Steyermark (1934) has pointed out that a number of other plants have disjunct distributions similar to that of the genus Leavenworthia, with populations in the Interior Highlands and others in the southern Appalachian Plateau and adjacent areas. Braun (1950) states that the Tertiary connections between the Appalachian and Interior High- lands, northward around the Mississippi Embayment have been of fundamental importance in vegetational develop- ment. It seems likely that this was also the route taken by Leavenworthia in migrations between the Appalachian and Interior Highlands. There is no conclusive evidence to suggest whether Leavenworthia originated in the western or eastern sections of the area it occupies at present. The resemblance between the flora of the cedar glades and that of the prairies of the Great Plains (Harper, 1926; Erickson et al., 1942) perhaps suggests a western origin for Leaven- worthia, If Leavenworthia has indeed come into the Interior Highlands from the west by a northern route, the species must have come into Alabama from the north. Leaven- worthia exigua at present occurs from northern Kentucky to central Alabama, and it is striking that the most priml- tive (least adapted to autogamy) variety, var. laciniata, 1s in Kentucky, and the most advanced variety, var lutea, occurs in Alabama. This perhaps suggests a southward migration by L. exigua. A second and more important line of evidence lies in the nature of the cedar glades in Tennessee and Alabama. Many authors have stressed that the cedar glades are most eX- tensive and characteristic in the Central Basin, where they are typically developed on thin-bedded Lebanon limestone of Ordovician age. This formation frequently weathers to form large areas of denuded outcrops with low relief — the Sites of the cedar glades. Quaterman (1950) estimated that there are approximately 300 square miles (5 to 6 per cent of the Central Basin) of rocky barren flats supporting cedar glade growth in central Tennessee. The cedar glades of northern Alabama are much smaller EVOLUTION IN LEAVENWORTHIA 13 than those of the Central Basin. The Bangor and Tuscumbia limestones of Mississippian age on which they occur are predominantly massive to thick-bedded. These limestones form much smaller areas of denuded outcrops, mostly on mountain slopes (Harper, 1926; Johnston, 1930). Harper did not even recognize these areas as cedar glades, though it is apparent from their species composition that they resemble the cedar glades of the Central Basin more closely than they do any other habitat (Harper, 1926; Braun, 1950). The suitable limestone formations in north- ern Alabama are sandwiched between younger and older non-calcareous rocks, whereas the entire Central Basin is caleareous. The cedar glades of Alabama have probably always been small and widely scattered, although in past times they have undoubtedly occurred in places different from those they occupy now. The cedar glades of Alabama are not only geologically and ecologically marginal; they are geographically periph- eral too. They occur mostly on the southern margin of the Highland Rim section of the Interior Low Plateau. To the west, south, and east of northwestern Alabama the Paleo- zoic limestones are largely covered by younger non-calcare- ous sediments of the coastal plain and the Appalachian Highlands. North of Alabama, however, calcareous for- mations have been extensively developed for a long time and could well have provided a source for the population of the cedar glades of Alabama. The geological nature of the Interior Low Plateau sug- gests that the Paleozoic limestones of the area were exposed in the Central Basin of Tennessee earlier than in Alabama. The Nashville Dome is the major structural feature of the present topography and geology of the Interior Low Plateau. The crest of this dome is in central Tennessee. As this dome rose, the sediments above the Paleozoic formations were eroded away and the Paleozoic sediments were exposed. “That the Ordovician rocks in Limestone County [Alabama] and in Tennessee to the north were exposed prior to the exposure of Mississippian strata south of the Tennessee River is entirely probable owing to the greater amount of uplift in the proximity of the crest of the dome” (C. W. Copeland, pers. comm.). If the limestone areas occupied by Leavenworthia were 14 LLOYD ants of IG. 1. (above) Photograph of a medium-sized cedar glade occupied by pl race al (L. alabamica). F rrowing in 4 plants of race al (L. alabamica) & ‘IG. 2. (below) Photograph of cedar glade, EVOLUTION IN LEAVENWORTHIA 15 exposed in Tennessee before being exposed in Alabama it is more likely that the species of Leavenworthia have migrated into Alabama from Tennessee, and not vice versa. Ac- cording to this hypothesis, the common ancestor of L. alabamica and L. crassa formerly existed in Tennessee, then migrated to Alabama, and subsequently became extinct except in Alabama, A similar migration was proposed by Baldwin (1943) to account for the distribution of Sedum pulchellum Michx. Baldwin believes that the Central Basin has been the center of variability and dispersal of the poly- ploid series within Sedum pulchellum. From the Central Basin the diploid form has succeeded in extending itself to the south and west (including Lauderdale and Madison Counties, Alabama) and the tetraploids have extended beyond the Central Basin to the north. THE DESCRIPTION OF L. CRASSA, L. ALABAMICA ND THEIR RACES As described previously, L. crassa and L. alabamica are closely related species with the same chromosome number (n=11), differing in only a few silique characters. Within L. crassa, fifteen distinguishable races have been recognized from field and greenhouse studies. Within L. alabamica four races have been recognized amongst populations grown in the greenhouse; two additional races seen in the field have not been studied in detail. Since the species have been previously described (Rollins, 1963) and the races are not being formally recognized, both the species and races are given only informal descriptions below. The descriptions of the species and races are all based on observations on plants growing in nature, and on plants grown in a greenhouse in the winters of 1962 and 1963. All of the races which have been recognized were grown during 1963, and their separation is based primarily on data col- lected then. The measurements given in the descriptions below will all be discussed later and are averages obtained on the 1963 greenhouse material with the following ex- ceptions: 1. The flower colors given are those growing in the wild. 2. The fraction of posit present in populations ive results in self-pollination trials 16 LLOYD is based on the combined averages of all tests made on the populations of each race, as described below. 3. The lengths of the siliques (excluding styles) are those of the largest fruit of any population of each race, from collections of plants growing in the wild in 1962 and in the greenhouse in 1962 and 1963. The 1963 greenhouse material alone could not be used, since some of the plants grown then became unhealthy later in the flowering period, and many of the fruits did not mature properly. Planar shapes are described on the system published by the Systematics Association Committee for Descriptive Biological Terminology (1962). To simplify the descriptions of L. crassa, L. alabamica and their races, a combined description of L. crassa and L. alabamica (including the differences between the species) is given first. This is followed by descriptions of the races of L. crassa and L, alabamica, which mention only the dis- tinguishing features of each race, including measurements. In the following combined description of L. crassa and L. alabamica, characters which differ consistently between the species and characters which are shared by all races of both species are given in Roman type. Characters which vary between races of either species are given in italics; the expression given for each of these characters is that prevail- ing in most races of both species. Measurements given in italics are those of the lowest and highest racial averages amongst the races of L. crassa and L. alabamica, except for the silique lengths which show the range in the longest measurement obtained in each race. A COMBINED DESCRIPTION OF L. CRASSA AND L. ALABAMICA mucronate; the apices of the terminal lobes are rounded; the bases of the terminal lobes are variable (cordate to obtuse). The lateral lobes EVOLUTION IN LEAVENWORTHIA 17 are always variable in shape, with irregularly scattered anthocyanin spots on the upper surface. The branches average 4.1 to 29.8 cm long, are initially suberect, and later extend beyond the scapes of the central flowers, becoming procumbent. The petals average 7.81 to 12.07 mm long and consist of a narro proximal claw parallel to the axis of the flower and a broader distal limb. The petal limbs are broadly obovate, emarginate (emarginations petal limb are three nectar-guides radiating from the middle of the petal towards the distal end of the limb, The nectar guides are composed of superimposed brown and orange pigments; the orange pigment predominates; the lateral nectar-guides are only slightly shorter than the central one. Apart from the nectar-guides which are always present, the following petal color patterns have been recog- nized (the first two are by far the most common): a. ‘Yellow’: a yellow pigment is uniformly present over the petal limbs. b. ‘Yellow- centered’: the proximal portion of the petal limbs is yellow, and the distal portion is white; the yellow portion occupies half, or less, of the limbs. The yellow portion ends abruptly, is conspicuous and contrasts strongly with the white portion, and the flowers thus have a yellow ‘eye’ surrounded by the white portions of the petals. c. ‘orange-centered’: as for yellow-centered, except that the proximal portion of the petal limbs is orange. d. ‘Imperfect-centered’: the proximal portion of the petal limbs is yellow and does not end abruptly, but blends into the distal white portion, and is longer than in yellow-centered flowers, particularly towards the petal margins. Other patterns occur rarely but are of little importance. In nature conditions the petals are without visible anthocyanin. The flower color patterns present in nature are described separately under each race, The stamens are strongly tetradynamous. The anthers of single stamens ere always introrse, and are approximately level with the petal claws. The anthers of the paired stamens are approximately at the level of the stigma, beyond the plane of the petal limbs, arne than the style, which averages 2.2 to 4. alabamica the ovary is —— bee longer (ca. 3 mm) than the style, which averages 2.0 to mm The mature siliques pasa styles) of L. crassa are fleshy and 18 LLOYD terete (2.5 to 5.0 mm thick); the septa are oblong to broadly elliptic and average 8.8 to 14.8 mm long, with a rectilinear to round apex. The mature siliques of L. alabamica are not fleshy and flattened (1.5 to 2.5 mm thick); the septa are narrow-oblong to oblong, and average 17.6 to 27.9 mm long with an acute to obtuse apex. In both species the septa of mature siliques sometimes have one (or occasional- ly more) small holes near the base. The seeds are flattened, winged, circular to broadly elliptic and ca. 3 mm wide. The seeds in each locule partly overlap in L. crassa, but are usually entirely separated in L. alabamica (fig. 5). Some races are self-incompatible (with varying frequencies of pseudo-compatibility); other races are self-compatible; the fraction of positive results on self-pollination averages between 0.17 and 1.00. The fraction of flowers which set seed spontaneously in the green- house (are auto-fertilized) averages between 0.01 and 0.30. The advancement indices of the races range from 0 to 100. It is apparent from maps 1 to 5 and the following dis- cussions that the distinguishable racial entities are a allopatric geographical races. But since the distribution pattern is very complex and concentrated in a small area, and the differences between races are minor and not always apparent in nature, I believe that it is not advisable to give these races any formal taxonomic recognition. A limited number of crosses between the races has been made. Within L. crassa, the ten attempted inter-racial combinations were all successful, and within L. alabama the two attempted inter-racial combinations were successful (Lloyd, unpublished). Crosses between self-incompatible and self-compatible races, however, give a lower seed set when the self-compatible race is used as the male parent, indicating the development of incipient reproductive 180 lating mechanisms. There is also a possibility that one OF more of the races of L. crassa and L. alabamica may have chromosome numbers other than the n=11 found in the populations studied by Rollins (1963). However, no chromo- some counts were made in the present study. The counts made by Rolling were on the races designated as cl, c2, ©, c7 and c15 (L. crassa) and al, a2, and a4 (L. alabamict) below, and included the most primitive and most advanced race in both species. There is therefore no evidence at present that any of the racial entities described below de- serve specific recognition. In the following sections it will be shown that t self-incompatibility in some of the races of L. cras he loss of sa and EVOLUTION IN LEAVENWORTHIA 19 L. alabamica has been accompanied by a whole complex of associated changes in other characters. These changes, and other differences which are not related to the change in the breeding system, provide the distinguishing features of the races of the two species. The races of L. crassa and L. alabamica are designated by the first letter of the species to which they belong, followed by a number indicating their rank within the species in the order of increasing advancement indices. The advancement indices, described on page 70, indicate how far each race has evolved in the trends in ten characters associated with the evolution of self-compatibility. The evolution of these characters, including self-compatibility itself, is discussed in detail after the race descriptions. The description of a race as primitive or advanced refers only to its position in regard to the characters associated with the evolution of self-compatibility. The two intraspecific categories described by Rollins (1963), L. crassa var. elongata and L. alabamica var. brachystyla are the most advanced of the recognized races in their respective species. Leavenworthia crassa var. elongata is therefore designated as race c15 and L. ala- bamica var. brachystyla is designated as race a4. To make the following descriptions of the races of L. crassa and L. alabamica as brief as possible a character is mentioned in a description only if its expression in a par- ticular race differs from that prevailing in most races of both species. If a character is not mentioned in a racial description, its expression is the same as that given in italics in the species description. In addition the racial averages for all measurements mentioned in the species description in italics are given for each race. Where two petal color forms (morphs) of a race were common in nature and were grown in the greenhouse in 1963, the floral measurements for each morph are given separately. More than one population of some of the races was grown, and for these races the figures given are the average of the figures for each population; in the case of silique lengths the figure given is the length of the longest fruit amongst all the field and greenhouse collections of populations cf a race. Many of the characters studied are quite variable, and the differences between races may be small. Consequently, ac3 ecs i) Jmile MAP 2 Massey ocs Lebanon aad x9 Ac10 ei Imile @CIl wCl2 ec 4 MAP 3 wCl4 M TO Distribution of the races of L. crassa and L, alabam Massey-Lebanon district. The narrow co indicate peters above = level. “eee areas indicate forest mining around the hills. Patches of forest on the valley floor are _- shown. Str are shown in MAP 4. Cedar glades Seoul by L. crassa . alabamica are hile by crosses in MAP 2. Glade populati ecis indicated by solid symbols, with the population number beside the symbol. In MAP ; — en are. from left to right, numbers 748, 89, 791, = 1375, 86 an 118. Populations on secondary sites are shown as hatched more are : z and 742); the number beside each mixed population corresponds to that in the description of these areas (page 89). 22 LLOYD some of the racial differences were not detected under the more variable conditions in nature, and appeared only in the greenhouse material, grown under relatively uniform con- ditions. Small differences between races have been ignored, and it is believed that all of the differences discussed re- present true racial characteristics. Although many of the characters discussed appear trivial, the number of these small differences distinguishing the species and races is often surprisingly large, and provides a distinct visual impression of each race. The description of each race is accompanied by an account of its known distribution. Pre-agricultural glade sites, and sites on cultivated land are considered separately. The ybrid swarms occurring on cultivated fields have been noted under the appropriate races. THE RACES OF L. CRASSA E Cl. LAWRENCE AND MORGAN COUNT Population ie was grown in the greenhouse in oat and 1963. Distinguishing features. Terminal lobes of later leaves typically crenate, varying from shallowly crenate to deeply crenate and almost lobed. Most plants of every population have yellow-centered flowers, narrowly hong to Hes with an to as Tire ae Branches long, 23.5 em. Number of flowers per plant large 465; Petals longer than in car 0 other L. crassa race, 12.07 mm; emargi- nations deep, 0.97 mm. Styles long, 3.70 mm; pistils long, 5.88 mm. Pollen:ovule index higher than in any other race of either species, 9520. Silique longer than in any other race of L. crassa, up 4. mm. Plants self-incompatible with some pseudo-compatibility; fraction of positive results on self-pollination 0.17, lowest of any race of either species. Percentage of flowers auto-fertilized low, 2. Advancement index 0. ace cl is the most primitive race of L. crassa, and the most similar to L. alabamica, especially in the length and shape of the siliques. Its distinguishing sete i not indicate a particularly close relationship with any other r Glade sites: None is known, but ‘the probable areas have not been thoroughly searched. Cultivated sites: There are numerous populations in corn fields and s in the region one and a quarter to two and a quarter miles southwest of Speake, in Lawrence and Morgan counties. Vicinity of the ta gaa populations, but more than one glade aie may exist in the EVOLUTION IN LEAVENWORTHIA 23 C2, LAWRENCE COUNTY Population 354 was grown in 1962 and 19638. Distinguishing features. Terminal lobes of later leaves crenate Anthocyanin on the ventral surface of the lateral lobes of later leaves usually present, in diffuse, dense Spots on the proximal lobes, concen- horizontally from the beginning, and are stiffer and elongate more idly than in other races (except c4). Most plants of every popu- lation have yellow-centered flowers, but other flower colors are usually present in very low frequencies. Siliques narrowly oblong to elliptic. Branches long, 26.5 em. Number of flowers per plant large, 514. Petals long, 10.94 mm, emarginations deep, 0.80 mm. Styles long, 64 mm, pistils long, 5.95 mm. Pollen:ovule index intermediate, 4400. Siliques longer than in any other L. crassa race except cl, up to 13.1 mm, Plants self-incompatible with some grins fraction of positive results on self-pollination met flowers auto-fertilized low, 2. Advancement in 5. In a large number of nome me race c2 is nit quite as primitive as race cl, but the differe are generally small, and only the difference in the pollen beats ina is reflected in the advancement rapidly elongating branches, and the relatively short styles and pistils for a self-incompatible race, suggest that race c2 is closely related to race c4. The two races are clearly separated only by the mucronate crenations of the terminal leaf lobes of race c4, and the weaker self-incompatibility of race c4. Glade sites: An extensive glade oun one and a half miles south- surrounding forest. Other glade sites may exist to the northwest of Danville, in Lawrence county. ; Cultivated sites: a large series of populations, principally in pastures, surround Danville to the west, north and east, in Lawrence and Morgan counties. RACE C3. MORGAN CO’ Population 37 was grown in 1962 and ses: dil population 89 was 2: grown in 196 Distinguis ae features. Terminal lobes of later leaves sinuate, n a r Most plants of every population have yellow or yellow- centered flowers, but other flower colors are someti low frequencies. Yellow center of yellow-centered flowers extends halfway or more along the petal limbs. Branckis long, 24.5 thy Flowers per plant many, 404. Petals long, 11.63 mm for yellow-centered flowers, 10.47 mm for yellow flowers; emarginations deep, 0.93 mm for yellow-centered flowers, shallower, 0.50 mm for yellow flowers (but the 1962 figures show less contrast 24 LLOYD between flower colors, see table 4). Styles long, 3.85 mm for yellow- centered flowers, 4.17 mm for yellow flowers; pistils long, 6.03 mm for yellow-centered flowers, 6.35 mm for yellow flowers. Pollen: ovule index high, 6910 for yellow-centered flowers, intermediate, 5050 for yellow flowers. Siliques long, up to 12.7 mm. Plants self-incom- patible with considerable pseudo-compatibility; fraction of positive results on self-pollination 0.54. Percentage of flowers auto-fertilized low, 8. 5 Nac index 10 (5 for plants with yellow-centered flowers, 15 for plants with yellow flowers). Race c3 is most ce related to race c5 in its petal color poly- morphism and in the long yellow portion on the petal limb of yellow- centered flowers. The latter character is also shared with race c6, but this race is quite distinct otherwise and not very similar to race ¢3. Only the self-incompatibility of race c3 reliably separates it from race c5. The unique elongation of the primary axis during flowering appears erratically in less than a half of the plants, but it was present in some plants every time race c3 was grown, and has seen in nature. Glade sites: Populations 89 and 61, one and a half miles northeast of Massey, Morgan county, are two large but considerably disturbed pulations in table 10). Bathe, 743 probably belongs to race c3, although the frequency of p heey vn after selfing is higher race c3 and c5 to this complex is ‘not known. Race c3 has appare not spread south from its glade populations, since all field aio of L. crassa south of the race c3 glade populations are self-compatible. RACE C4. LAWRENCE COUNTY Population 38 was grown in 1963. istinguishing features. Terminal lobes of later leaves crenate; the ventral -cente wers, but other flower colors may be present. Oran ge-centered wers are more common than in any other race of L. crass EVOLUTION IN LEAVENWORTHIA 25 ovule index high, 5733, Silique length unknown (see below). Plants self-incompatible with considerable seudo-compatibility; fraction of positive results on self-pollination 0.53. Percentage of flowers auto- fertilized low, 4. Advancement index 15. Race c4 was one of the two least healthy in the 1963 greenhouse material, and ‘the fruits were too poor for their shape and maximum length to be measured. All measurements were made on plants with yellow-centered flowers. This primitive race is most similar to race e2, but is clearly Separated from it by the mucronate crenations on the terminal lobes of the later leaves, and its weaker self-incompatibility. The relatively high frequency of orange-centered flowers ( up to 25 per cent) is unique in L. crassa. Glade sites: There are two known glade populations about sixty yards apart, two and a quarter miles east-northeast of Oakville, Cultivated sites: Race c4 has spread, probably by downstream dispersal of seeds, onto two adjacent fields one and a half miles east of Oakville. RACE C5, MORGAN COUNTY Populations 58 and 41 were grown in 1962 and 1963. Population 100 grown in 1962. Populations 86 and 118 were grown in 1963. Distinguishing features. Most plants of every population have yellow or yellow-centered flowers, but other flower colors are some- times present in very low frequencies. Yellow center * yellow- mb. ran Petals long, 10.41 mm for yellow-centered flowers, shorter, 9.96 mm, for yellow flowers; emarginations deep, 1.06 mm for yellow-centered flowers, shorter, 0.89 mm, for yellow flowers. Styles long, 3.73 mm for yellow-centered flowers, slightly longer, 3.89 mm, for yellow flowers. Pistils long, 5.70 mm for yellow-centered flowers, slightly longer, 6.07 mm, for yellow flowers. Pollen:ovule index intermediate, of positive results on self-pollination, 0.96. Percentage of flowers auto-fertilized intermediate, 12. Advancement index 22.5 (20 for plants with yellow-centered flowers, 25 for plants with yellow flowers). Race c5 is closely related to race c3 (see under race c3), but is somewhat more advanced in several characters. It is more distantly related to race c8. ; Glade sites: Population 86, one and three-quarter miles east-north- east of Massey, is separated from the corn fields to the north only by a fence and an incomplete line of trees, and is almost continuous with the corn field populations. Population 118, one and three-quarter miles east-northeast of Massey, is about sixty yards southeast of Population 86 and is also separated from the corn field populations 26 LLOYD to the north by only a narrow line of trees. Population 791, one and a half miles northeast of Massey, occupies a very marginal site about twenty yards within the forest from population 37 of race c3, and is surrounded on three sides by race ¢3, but it is probably a pre-agri- cultural site. Population 1375, one and three-quarter miles east- northeast of Massey, occurs as several scattered sub-populations in a small patch of forest in which many trees have been cut. Cultivated sites: Race c5, like races cl5 and a4, has moved long distances over cultivated fields in several directions and it has met and hybridized with several other races in these fields to produce a complicated distribution and variation pattern. The field populations of race ¢c5 closest to glade populations 86 and 1375 are a complex series of populations one and a half miles north- east to one and three-quarter miles east-northeast of Massey, where races c3 and c5 have come into contact as described under race ¢3 (see also map 2). From population 86 and perhaps also from population 1375, race cd has also spread south over a distance of almost two miles, There is a series of populations from a half mile east of Massey to two and one-quarter miles east-northeast of Massey. Included in this series are several populations surrounding glade population 60 of race a4, which are mixtures of race c5, race a4 and their hybrids. Also in this The movement of race c5 continued south to a scattered series of : RACE MORGAN COUNTY Population 268 was grown in 1961 and 1962. tinguishing features. Base of the terminal lobe of later a 0 of positive results on self-pollination, 0.99. Percentage of flowers auto-fertilized intermediate, 0.16. Advancement index 30. Race c6 is geographically the most isolated of all L. crassa races; EVOLUTION IN LEAVENWORTHIA 27 and it is not particularly closely related to y other race, The leaves are quite distinct, even though I could ey only one constant tangible difference between them and those of other races. e yellow center of the flowers may indicate a relationship to races c3 and c5. Glade sites: None is known. One or more must exist, almost certainly to the west of the cultivated site, but all attempts to find a glade site failed. Cultivated sites: This race is known only from two sotaty oh ce constituting population 268, three miles southwest of Massey, Co. CE C7, MORGAN COUNTY Population 361 was grown in 1963. Distinguishing features. Terminal lobes of later leaves shallowly crenate to crenate. Terminal lobes, and especially the lateral lobes of later leaves, slightly incurved exposing the paler edges of the leaves. Most plants have yellow-centered or yellow flowers but other color forms are present. Stigma and anthers of paired stamens at approxi- mately the level of the petal limbs. Filaments of paired anthers turn nches n i mi ys 361. Petals long, 10.33 mm; emarginations deep, 0 mm. Styles, 2.21 mm, and pistils, 4.28 mm, shorter than in any other race of L. crassa. Pollen:ovule index intermediate, 4120. Siliques short, up to 9.3 mm. Plants self-compatible, fraction of positive results on self-pollination 0.95. be a of flowers auto-fertilized intermediate, 15. Advancement index 4 measurements were made on ie with yellow-centered flowers. Race c7 is not particularly similar to any other, but the incurved leaf lobes may saree a closer relationship to the adjacent race c14 than to any Glade sites: ‘Popiesiaie on 361, one mile southeast of Massey, is probably a pre-agricultural sibs, although the forest around it has been completely destroyed, and the population has probably expanded beyond its pre-agricultural size Cultivated sites: The area immediately to the south and west of population 361 contains a complex group of populations bi contri- t have contains at least one population (298) of race c7 which mus ae pulations 298 and 361. The margin rom a small scattered obviously hybrid plants near the margins of the two populations. Other nearby populations may contain elements of race c7 (see under race c5). RACE C8. MORGAN C Population 72 was grown in 1962 and 1968. 28 LLOYD Distinguishing features. Terminal lobes of later leaves cordate, crenate. Most plants have yellow or yellow-centered flowers, but other flower colors are present in very low frequencies, Anthers of paired stamens often widely separated from each other and the stigma. Branches of intermediate length, 14.0 cm. Number of flowers per plant intermediate, 225. Petals of yellow-centered flowers long, 10.78 mm, those of yellow flowers considerably shorter, 8.73 mm; emargi- mediate length, 3.21 mm for yellow-centered flowers, slightly longer, 3.57 mm, for yellow flowers; pistils of intermediate length, 5.39 mm mm flowers, low, 3360 for yellow flowers. Siliques short, up 8.8 mm. Plants self-compatible; fraction of positive results on self-pollination 0.98. Percentage of flowers auto-fertilized intermediate, 10. Advance- ment index 47.5 (40 for plants with yellow-centered flowers, 55 for plants with yellow flowers). } Race c8 is probably most closely related to races c3 and c5, but 1s considerably more advanced in many features than either of them. It is outstanding in the marked difference between yellow-centered and yellow flowers in petal length, petal emargination depth, and style length. Glade site: None is known, though the land surrounding the only known cultivated population was carefully searched. Cultivated site: This race is known only from population 72, on grazed land half a mile west of Lebanon, Morgan count RACE C9. MORGAN COUNTY Population 769 was grown in 1963. istinguishing features. Terminal lobes of later leaves ga Branches of intermediate length, 15.5 em, Number of flowers pet plant intermediate, 217. Petals moderately long, 9.77 mm; emargl- nations deep, 1.06 mm. Styles short, 2.53 mm; pistils short, 4.66 mm. Pollen:ovule index low, 3560. Siliques short, up to 0.97 mm. ee self-compatible, fraction of positive results on self-pollination pp Percentage of flowers auto-fertilized intermediate, 10. Advanceme? index 50. Race 9 is most closely related to race cll (see under race ¢11) but is more primitive. f Glade site: Population 769, one and a half miles east-southeast © Massey, Morgan county, is the only glade site. EVOLUTION IN LEAVENWORTHIA 29 Cultivated site: Population 771 is a small population about fifty yards downhill from population 769, just outside the forest margin. RACE C10. MORGAN COUNTY Population 698 was grown in 1963. Distinguishing features. Mature later leaves strongly curved, always in a counter-clockwise direction when ooking down on the entire, polygonal (heptagonal to octagonal); angles apiculate. Plants with yellow or yellow-centered flowers. Petal limbs usually bend at less than 90° to the claws in open flowers. Yellow center of yellow- centered flowers not as bright as in other races (cf. race cl4). Orange pigment of nectar guides poorly developed, sometimes completely masked by the brown pigment. Lateral nectar guides much shorter than the central one. Filaments of paired stamens turn through 135°, sometimes less, anthers extrorse or slightly introrse. Branches of intermediate length, 11.0 cm. Number of flowers per plant small, 126. Petals long, 10.02 mm; emarginations of inter- mediate dest. 0.61 mm. Style length intermediate, 3.29 mm; pistil length intermediate, 5.10 mm. Pollen:ovule index low, 3690. Siliques short, up to 9.4 mm. Plants self-compatible, fraction of positive results on self-pollination 0.95. ‘Percentage of flowers auto-fertilized high, 30. Advancement index 70. No yellow-flowered plants were sufficiently healthy in the 1962 material for measurements to taken from them. flower measurements above refer only to yellow-centered flowers. Race c10 characters but differs in having a petal color polymorphism and in its very distinctive leaf shape and curvature Glade site: Population 698, one and | a half miles southeast of Cultivated site: None is known, and it seems unlikely that said ane ever been produced since population 698 is surrounded by fore CE Cll. MORGAN COUNTY Population 742 was grown in 1963. istinguishing features. Mature leaves short, usually less than 7 em long. Terminal lobes of later leaves wider than long, cordate, entire or slightly sinuate, oc! appressed to the ground. Lateral ially the distal ones, usually with a con- yellow-centered flowers (but see ng oer! os pax te Petal li ‘a usually bend at less than 90° to the claws i flowers and anthers of paired stamens at sspetimately a level of the sat lim bs. Branches shortest of any race of either species, 4.1 cm. Number of 30 LLOYD flowers per plant small, 163. Petals of intermediate length, 9.56 mm; emarginations of intermediate depth, 0.71 mm. Styles short, 2.68 mm; pistils short, 4.45 mm. Pollen:ovule index low. 3910. Siliques short, up to 8.8 mm. Plants self-compatible, fraction of positive results on self-pollination 1.00. Percentage of flowers auto-fertilized high, 21. Advancement index 75. _ Race cll is very distinct, but the cordate and entire or sinuate terminal leaf lobes and prominent anthocyanin on the lateral leaf lobes shared with the geographically adjacent race ¢8 indicate that these two races are more closely related to each other than to any other race. Glade site: Population 742, one and a half-miles east-southeast of Massey, and about 80 yards southeast of population 769 (race c8), is the only known glade population; it is in the process of fusing with population 741 (race cl12?)—see page 94. ultivated site: None is known, and the position of population 742 within the forest makes it unlikely that any have ever been produced. RACE C12. MORGAN COUNTY Population 739 was grown in 1963. Distinguishing features. All plants have yellow flowers. P etal limbs oblong to obovate, often bending at more than 90° to the claw in open flowers, and rotated so that the limbs of the anterior petals are inclined towards each other and the limbs of the posterior petals are inclined towards each other. In flower closing, petal limbs often bend back past the vertical plane until they become horizontal over the pistil, and are stacked like the four sections of the top of a cardboard box. All siliques (greenhouse grown material only) with one (oF occasionally more) holes in the septum. Branches short, 7.1 cm. Advancement index 80. The yellow flowers and rather short and narrow leaves of rare oF indicate that it is probably most closely related to the geographically adjacent race c13, but it is unique and very distinctive in its short petals, peculiar and complicated petal movements, and the invariably present silique septum hole (the latter is not invariably present 17 siliques collected in nature), f Glade sites: Population 739, one and a half miles southeast : assey, consists of two separate subpopulations in the same glade (see page 94). The fruits of population 741 collected in 74 resemble those of race c12 rather than those of race ¢13, but identity of populations 739 and 741 needs checking. ore Cultivated site: None is known, and as both populations are " EVOLUTION IN LEAVENWORTHIA 31 than forty yards within the forest margin it is unlikely that any have ever been produced RACE C13, MORGAN COUNTY Population 745 was grown in 1963. Distinguishing features. Mature later leaves short, usually less than 7 em long. Terminal lobes of later leaves cuneate to obtuse, margin unusually variable, sinuate to deeply crenate or even lobed. Branches short, not extending beyond the scapes of the central flowers, almost vertical, scapose fi racemose flowers densely crowded together. All plants yellow-flowered. Sepals often open to a horizontal position before the petals unfold. Petal limbs usually bend at less on 90° to the claws in open flowers. Filaments of paired stamens turn through 90° (occasionally more or less), anthers introrse. * Siliques se Branches short, 8.8 cm. Number of flowers per plant small, 196. Petal length intermediate, 9.52 mm, emarginations deep, 0.82 . Style length intermediate, 3.19 mm; pistil length intermediate, 5.29 mm. Pollen:ovule index low, 3240. Siliques short, up to 9.2 mm Plants self-compatible; fraction of positive results on self-pollination 1.01. Percentage of flowers auto-fertilized high, 20. Advancement index 80. Race c13 is probably most closely related to the geographically adjacent race c12, but it is very distinctive in the (sometimes) deeply incised terminal lobes of the leaves, the crowded inflorescences, and the (sometimes) delayed opening of the petals. Glade sites: Population 745, one and a half miles southeast of Massey, is about fifty yards south of population 739 (race c12) and separated from it by continuous forest. Population 746 is west of population 745, and the edges of the two populations are separated by bout twenty yards of forest. These two ae nega are sufficiently isolated to be distinct, but population 746 was say in the ouse, so it is provisionally included in this rac Cultivated site: None is known, and as both Seinialieal 745 and 746 are well within the forest margin it is unlikely that any have ever been produced. CE C14. MORGAN COUNTY Population 699 was Pie in Disti Terminal lobes and especially the er f not as bright as in other races (cf. race cl0). Orange pigment of nectar guides poorly developed, sometimes completely masked by the brown pigment. Lateral nectar guides much shorter than ae central one. Filaments of paired a turn through 90 to 135 the anthers are introrse to extrors Branches rather short, 10.2 cm. Num ber of flowers per plant small, 115. Petals short, 8.13 mm, emarginations shallow, 0.47 mm. Style 32 | LLOYD length intermediate, 3.20 mm; pistil length intermediate, 5.0 mm. Pollen:ovule index low, 3480. Silique length unknown (see below). Plants self-compatible; fraction of positive results on self-pollination 1.00. Percentage of flowers auto-fertilized high, 22. Advancement index 85. This population was one of the two least healthy in the 1963 greenhouse material, and the fruits were considered too poorly devel- oped to describe and measure. Race c14 closely resembles race c10 in its floral characters, but it is monomorphic and the distinctive leaf curvature of race c10 is barely evident in race c14, and the shape of the terminal lobes of the later leaves is very different in races c10 and cl4, Race cl4 appears to be more advanced than race e10 in a number of characters. Glade site: Population 699, one and a half miles southeast of incomplete line of trees. Even so these two populations have remained distinct and must be effectively geographically isolated. Cultivated site: Population 700 is about 100 yards north of popu- lation 699 in a field which had not been ploughed for some years prior to 1962. It consisted in 1962 of a few scattered plants with introrse anthers, apparently of race cl4. An unused forest road exists between population 699 and this field and the plants could easily have moved along it RACE C15. MORGAN COUNTY (L. crassa var. elongata Rollins) Population 370 was grown in 1962 and 1963. Populations 53 and 750 were grown in 1963. Distinguishing features. Terminal lobes of later leaves sinuate, occasionally shallowly crenate. All plants yellow-flowered. Petal limbs almost always bend at less than 90° to the claws in open flowers. Filaments of paired stamens turn through 90° (occasionally more or less), anthers introrse. Siliques oblong to elliptic, apiees Plants self-compatible; fraction of positive results on self-pollinatiom, 0.97. Percentage of flowers auto-fertilized high, 27. Advancemen index 100. Race c15 is the most consistently advanced of any race of wig species, but even this race is only moderately advanced in silique length. It does not appear to be particularly closely related to any other race, ¥ Glade sites: Population 53 is one and three-quarter miles sout southwest of Lebanon. Population 27, two and a half miles southwest : which 18 EVOLUTION IN LEAVENWORTHIA 33 quarter miles southwest of Lebanon, occupies a high (700 + 5 feet above sea level) site on the brink of a quarry, and on a forest road; this site is doubtfully pre-agricultural. Cultivated sites: In 1961 there were two separate populations * race cl5 in the area two and a half miles south-roathwes of (map 4). One of these was a mixture of plants of races ¢15 and and their hybrids. The other was a small but vigorously growing population in a grazed field; but in 1962 and 1964 there were no plants there at all. A series of Ai eren een extends from population 27 to the northeast for a distance of almost three miles. This distribution pattern is admirably explained “ee “tiie movement of race cl5 seeds downstream from population 27 (Rollins, 1963). The populations farthest from population 27 are a mixture of races cl15 and a4 (map 4 ). There are several ghee of race cl5 in the area one mile southeast of Massey. These res of race cl5 and race c7 There are several ooudidaens containing race c15 in the area one and a half to two miles south-southeast of Massey. In addition to the —— with elements from races cb and ¢c15 (and c7?— map 4) there are a number of populations of race ¢15 alone. The Pred ei population containing race cl5, population 275, two and three-quarter miles south-southeast of Massey, in 1961 had a few yellow-flowered plants which were hybrids between race c15 and race a3 and/or race a4 (see under race a3). These plants in population 275 are apparently a relict of populations which must have spread overland from the race ¢15 populations to the north. THE RACES OF L. ALABAMICA RACE Al. FRANKLIN AND LAWRENCE COUNTIES Population 81 was grown in 1962. Populations 81, 82 and 923 were grown in 1963. Distinguishing features. All plants with yellow-centered flowers in most sangre ag several populations also have other flower colors in low frequenc Anthocyanin always present on the dorsal surface of petal eetage of yellow-centered flowers Branches long, 29.3 em. Number of flowers per plant la arge, 503. ovule index high, 7030. Siliques very long. self-incompatible with some oot gj Paeinge ~ age (lacing ) f > si om each fi shown. ca. X ower. é 8 EVOLUTION IN LEAVENWORTHIA L. Fig. 5. Dehiseed fruits of all the described races of L. igi f£ L. alaba- The seeds in one locule are shown for race c9 of L. crassa and race a! ob % 40 LLOYD UNDESCRIBED RACES OF L. ALABAMICA Two additional races of L. alabamica have been dis- tinguished in the field but not grown in the greenhouse. They therefore can not be described fully, or be given an advancement index. Both races have been designated by the town nearest which they occur, and their distinguishing features and known distributions are briefly indicated. A. THE ‘RUSSELLVILLE RACE’ OF L. ALABAMICA In 1964 a population of L. alabamica which was clearly different from all other known populations was discovered in a ploughed field three and a half miles northeast of Russellville, Franklin County. This field is in a narrow valley with steep hills on either side. The glade source is unknown but is probably at or near the base of the hills, and is clearly separated from the glades of race al to the east and south by several miles of hilly ground unsuitable for L thia. The Russellville race appears to be ap- proximately as advanced in characters associated with the breeding system as is race a4, but is easily distinguished from race a4 by the presence of anthocyanin on the petal limbs. The terminal lobes of later leaves appear to be relatively broader than those of race a4. The flowers are approximately the same size as those of race a4, and are mostly yellow-centered; but plants with orange-centered flowers (unknown in other L. alabamica races) are quite common. The petal limbs usually bend at less than 90° to the claws in open flowers, and are often incurved (cf. race a4). The filaments of paired stamens turn through 90° so the anthers are introrse. The styles are approximately as long as those of race a4. Forty-four self-pollinated flowers all gave positive results, so the plants are self-compatible. B. THE ‘TUSCUMBIA RACE’ OF L. ALABAMICA A number of populations southwest and west of Tuscum- bia, Colbert County, are probably distinct from other races of L. alabamica, but have not been grown in the greenhouse. These occur in or near scattered glades from 13 miles west of Tuscumbia to four miles southwest of Tuscumbia (Rol- lins, 1963). It is possible that the separate populations are not all identical, but I have examined only the population four miles southwest of Tuscumbia. This population appeats EVOLUTION IN LEAVENWORTHIA 41 to be somewhat more advanced than either race a4 or the Russellville race in characters associated with the breeding system. It can be distinguished from race a4 by the presence of anthocyanin on lower surface of the petal limbs, and from the Russellville race by the much shorter and less conspicuous nectar guides and yellow portions of the petal limbs. All of the flowers examined were yellow-centered, and approximately as large as those of race a4. The petal limbs of open flowers are often incurved, and bend at less than 90° to the petal claws. The filaments of paired stamens turn through 90° so the anthers are introrse. The styles are approximately as long as those of race a4 and the Russell- ville race. Thirty-four of thirty-five self-pollinated flowers gave positive results, so the plants are self-compatible. Race a4 and the Russellville and Tuscumbia races of L. alabamica share several common features which distinguish them from the more primitive races (al, a2 and a3) of L. alabamica, viz. short petals and styles, incurved, incomplete- ly opening petal limbs, and introrse anthers ( paired stamens). It is possible that these three races form a natural group within L. alabamica, but these common features could equally well result from parallel evolution towards features associated with increasing autogamy. The presence of race a3 near race a4, and of race al near the Russellville race, perhaps suggests that race a4, the Russellville and Tuscumbia races have become advanced in characters asso- ciated with self-compatibility independently of each other, but the relationships between the three races can not be settled until all have been studied in the greenhouse. THE EVOLUTION OF SELF-COMPATIBILITY IN L. CRASSA AND L. ALABAMICA Rollins (1963) discovered the presence of self-incom- patible and self-compatible races in both L. crassa and L. alabamica. The direction of evolution in these species has almost certainly been from self-incompatibility to self- compatibility, since race cl of L. crassa possesses a Sporo- phytic, one locus, multiallelic incompatibility system ( Lloyd, unpublished). This type of incompatibility system 1s un- common, and is known only in the Cruciferae and Com- positae, so it is unlikely to have been acquired independently in Leavenworthia. 42 LLOYD In the present study controlled cross- and self-pollinations were made in the field and greenhouse on all the races of L. crassa and L. alabamica described above. In three series of tests — in the field in 1962 and in the greenhouse in 1962 and 1963 —the resulting pollen tube growth in each style was examined. In two other series of tests, in the green- house in 1962 and 1963, the subsequent silique growth and seed set was examined. For the pollen tube growth tests the pedicel of one flower bud from each of a number of plants of a population was cut off the afternoon before the flower was due to open and put into a vial containing water. The next morning, when the anthers had dehisced, pollen was transferred to the stigmas with an arrowhead needle until they were abundant- ly covered with pollen. About 2/3 of the flowers of each sample were self-pollinated, and the remainder were cross- pollinated. The needle was cleaned after each pollination, and the flowers put back in the vials. In the greenhouse tests additional flowers from the same plants were pollinated each day until about 40 self-pollinations had been made on each population. The styles were cut off approximately 24 hours after pollination and put into a solution of lactophenol containing traces of acid fuchsin and fast green. After another 24 hours or longer the styles were examined under a microscope to see if there were any pollen tubes in the style. The self- incompatibility barrier in Leavenworthia, as in other Cruciferae, is at the stigma. If there were any tubes at all present in the style, the pollination was considered as suc- cessful and counted as a positive result. With few exceptions the styles contained either no pollen tubes or a considerable number, usually more than thirty. e fraction of positive tests from self-pollinations of the flowers of a population was calculated. If more than one population of a race was tested in one series, the ge0- metric mean of the fraction of positive results on self- pollination in each population was calculated. The cross-pollinations showed that there were almost no failures due to poor pollen or unsuitable pollinating conditions. In the 1962 field tests 333 out of 336 (99 per cent) of the cross-pollinations on all populations gave positive results. In the 1962 greenhouse tests 403 of 407 (99 percent) cross- EVOLUTION IN LEAVENWORTHIA “UO!FBUITJOd-ss019 UO UBYy} UOIBUTT]Od-jJjas uo (Av IGQIVBduI00-opansd yo sarouenbaay SNOMBA YA) s@jGQI}VduUOoUI-J[as == (10°>d) AeMOT AMUBOYUs!sS seM syINSat AATZISOd Jo UOTOVIy AYT, oe» ayqiyeduros-jjas <= +- ae tT + 16°0 00°L 160 — _ 00°T yu + 66°0 00°T 00°L - _ L6'0 gu + 18°0 TL‘0 £6°0 — 00°T eu - 810 «#620 90°0 «e610 aa P8'O eeZ1'0 ye DIUYQDID “7 + 160 16°0 £60 00'T 96'0 66°0 ho) + 00°T — 00°T _ 00°1 pro + ToT S0'T 00°T _ 00°T g12 + 00°T LO'T 96°0 — 00°T ra) + 00°T 00°T 00°T _ 00'T 180) + 960 £6°0 160 _ 00°T 019 + ¥6°0 00°T £8°0 — _— 00°T 6? + 86°0 00°T 66°0 T0°T 00°T 16°0 go + 66'0 00°T 98°0 — — 00'T i) + 66°0 00'T 00°T 10'T 66°0 00°L 99 + 96°0 96°0 r6'0 66'0 86°0 L6°0 go = £9°0 ##69'0 080 a €8°0 po = PS'0 26°0 g8"0 xab8'0 #«%69°0 69'0 go = 12'0 #830 12'0 anL3'0 L¥v'0 ##90'0 (a) cag LT'0 «e910 12'0 #610 «#900 ax88'0 ja D88DL9 “77 1sn484g Sjsay, [[B jo yywMois yours aqn} yinorgs YyIMo1s aqny yyaors aqny avy pur Buyles Uva, JLIJeUIOay anbys Ag -uayjod Ag anbiis Ag -uaTjod Ag -uayjod Ag soroedg 841s9., ASNOYyUdeIyH EOGT 8189], eSNOyUDeIy Z9GT NOLTWAITT'T 7 8189. PIPlA c96T NO SLIONSAY AAMLISOd JO NOILOVUE AHL NO SLSaL ‘T a1av,L 44 LLOYD pollinations on all populations gave positive results. These figures are so close to 100 per cent that it was considered unnecessary to do any cross-pollinations in the 1963 green- house tests by pollen tube growth. In the greenhouse tests on the growth of siliques after cross- and self-pollination, the flowers to be pollinated were emasculated the evening of the day before they would open. The next day the stigmas were examined with a hand lens to check that there were no pollen grains on them, and pollen from another flower of the same plant or of a different plant of the same population was put on each stigma with an arrowhead needle. The flowers were not bagged, either before or after pollination, since the greenhouse was insect- free. When the pollinated flowers were examined about 5 weeks later there was a clear distinction between those that had been fertilized and contained one or more seeds in an en- larged silique, and those that had not been fertilized, in which neither the ovary nor the ovules had enlarged. For each plant, the fraction of positive results on self-pollination (z.e., the fraction of flowers fertilized) divided by the fraction of positive results on cross-pollination, was calcu- lated. This figure, which may be called the relative fraction of positive results on self-pollination, corrects the self- pollination results for occasional failures due to poor polli- nating conditions. As an extra precaution against the possibility of incompatible cross-pollinations, the few plants on which less than 3/4 of the cross-pollinations succeeded were eliminated from the results. In most populations, controlled pollinations were done on between four and ten plants in either 1962 or 1963. The geometric means of the relative fractions of positive results on self-pollination, for all the plants of a population tested in One year, were calculated. If more than one population of a race was tested, the population averages were in turn averaged to obtain a race average. The race averages for the fraction of positive results on self-pollination in the pollen tube growth tests and the relative fraction of positive results on self-pollination in the silique growth tests are shown in table 1. The fractions vary from 0.04 to 1.07, but the races fall clearly into two groups. In races cl, c2, c3, c4 and al the fractions vary EVOLUTION IN LEAVENWORTHIA 45 between 0.05 and 0.92, and in at least one of the tests on each of these races there were significantly lower (P<.01) fractions of positive results on self-pollination than on cross-pollination. These races may therefore be considered self-incompatible, although the frequency of pseudo-com- patibility is sometimes high. In races c5 to c15, a3 and a4, the fractions of positive results on self-pollination are usually between 0.95 and 1.00, although they range from 0.71 to 1.07. There were no significant differences between cross- and self-pollinations in the fractions of positive results in any of the tests on races c5 to c15, a3 and a4, so these populations may be considered self-compatible. There were no consistent differences between the pollen tube and silique growth tests in the fraction of positive results on self-pollination. This provides additional evidence that the incompatibility barrier in Leavenworthia is re- stricted to the stigma. The self-incompatible and _ self-compatible races are clearly separated when the figures for the separate tests on each race are combined as the geometric mean of all tests on the fraction of positive results on self-pollination (table 1). The means of all tests are between 0.17 and 0.54 for the self-incompatible races, and between 0.87 and 1.01 for the Self-compatible races. The figure of 0.87 for race a2 is principally due to the failure of self-pollinations (in silique growth tests only) on one of the plants grown in 1963. In addition, it is apparent that the self-incompatible races differ in the strength of their incompatibility barriers. The means of all tests on races cl, c2 and al are less than 0.25, while the means of all tests on races c3 and c4 are slightly greater than 0.50. There is no overlap between the two groups of races in any of the five series of tests, so it Is evident that races cl, c2 and al have lower frequencies of pseudo-compatibility, i.e., are more strongly self-incompati- ble, than races c3 and ¢4. The fraction of positive results varies considerably in the different series of tests on the same self-compatible race. For example, the five series averages on population 354 (race ¢2) vary from 0.06 to 0.47. Significant differences between the series of tests were found by x° tests for heterogeneity in populations 171 (race cl), 354 (race c2) and 38 (race c4). These differences are probably due to 46 LLOYD variations in the environmental conditions in the separate tests. It is well known that the frequency of pseudo-compati- bility depends greatly on such factors as the temperature, humidity and the age of the plants (e.g., Lewis, 1943). These factors undoubtedly varied between the separate series and even from day to day in each series. In most of the self-incompatible populations there were also significant differences between individual] plants in the Silique growth tests on the fractions of positive results on self-pollination. For example, the results on the six plants of population 171 (race cl) tested in 1962 varied from 0.11 to 0.37. Since the plants were pollinated on the same days, and only healthy plants in mid-flowering were tested, the variation must be due to differences in the strength of the incompatibility reactions in the individual plants. All of the self-pollinations produced siliques on several plants of self-incompatible populations tested in 1962 and 1963. In each of these cases, however, only twelve or fewer flowers were self-pollinated, and the data are insufficient to distinguish between a high level of pseudo-compatibility and full self-compatibility. Many of the self-compatible races have gained adaptations to increase the efficiency of self-pollination, such as introrse anthers, have a high spontaneous seed set in the absence of insects and must be largely self-pollinated in nature (see below). But in all races, including the self-compatible ones, fewer seeds are produced in each silique after self-pollina- tion than after cross-pollination (Lloyd, 1964, and un- published). This is probably due to a low level of inhibition through a self-incompatible reaction which is too weak to prevent the success of self-pollinations, but still operates to restrict the number of seeds after self-pollination. So it may not be entirely accurate to describe races c5 to c15, a3 and a4 as fully self-compatible. By the usual criteria of equally frequent pollen tube or fruit development after self- and cross-pollination, however, these races would be considered self-compatible, and they will be considered so ere. The comparative behavior of cross- and self-pollina- tions in the races of L, crassa and L. alabamica will be considered in detail in a later paper, In summary, four races of L. crassa (cl to c4) and one race of L. alabamica (al) are self-incompatible, with vary- EVOLUTION IN LEAVENWORTHIA 47 ing degrees of pseudo-compatibility dependent on the par- ticular plant and race tested and the environmental conditions. Self-compatibility has evolved in L. crassa (races c5 to c15) and in L. alabamica (races a3 and a4). Even in the self-compatible races fewer seeds result, on the average, from self-pollinations than from cross pollinations. CHARACTERS ASSOCIATED WITH THE EVOLUTION OF SELF-COMPATIBILITY A considerable number of changes in other characters has accompanied and followed the evolution of self-compati- bility in L. crassa and L. alabamica. Some of these changes have been noted in the genus as a whole (Rollins, 1963). The present discussion is restricted to L. crassa and L. alabamica, and only racial differences which are associated with the loss of self-incompatibility will be discussed here. The relatively few characters which appear to have evolved independently of the breeding system have been considered in the descriptions of the races, In the following discussions, an expression of a character is described as primitive if it is usually associated with self-incompatibility, and as advanced if it is found only in self-compatible races. The description of a race as primitive or advanced refers to its overall primitiveness or advance- ment in characters associated with the breeding system, and is based on the calculation of the advancement indices, discussed below. The percentage of auto-fertilization: The significance of the presence or absence of self-incompatibility lies in its effect on the frequency of self-fertilization, and thus on the level of heterozygosity and genetic variability in natural populations. The only way in which the relative frequency of self-fertilization in natural populations can be measured directly is by growing and classifying the natural progeny of rare, naturally occurring mutants. This has not been done for the races of L. crassa and L. alabamica, but a rough indication of the comparative frequencies of self- pollination in the different races can be got from counts of the percentage of fruits in each race which set seed spon- taneously under uniform, insect-free conditions in the greenhouse. Accordingly, the number of flowers produced on the plants grown in the greenhouse in 1962 and 1963, 4 2. THE NUMBER OF FLOWERS PER PLANT AND THE PERCENTAGE OF FLOWERS WHICH SET SEED SPONTANEOUSLY (WERE ‘AUTO-FERTILIZED’) N A GREENHOUSE IN 1962 AND 1963 vy. no. Av. no. Av. % Species Popula- No. of flowers flowers of flowers and Race ion plants per plant auto-fert. auto-fert. 1962 L. crassa Race: 171 7 363 4 1 c 354 10 189 9 5 3 37 10 208 44 24 5 58 6 161 77 44 4 6 150 63 42 Race average 156 70 43 c6 268 8 101 63 61 c8 72 9 133 88 68 c 370 3 40 36 91 L. alabamica Race: al 81 5 262 9 4 19638 L. crassa ace: cl 171 5 465 7 2 c2 354 4 514 11 2 3 37 6 404 34 8 c4 38 3 447 18 4 c5 86 4 460 40 9 58 6 295 28 10 100 8 342 54 18 41 8 297 25 9 Race average 349 37 12 c6 268 9 241 38 16 c7 361 7 361 55 15 8 72 6 225 21 10 9 769 4 217 22 10 c10 698 10 126 34 30 ell 742 7 163 31 21 c12 739 11 144 34 24 13 745 9 196 38 20 c14 699 3 115 25 22 c15 370 3 98 31 31 53 5 122 29 23 Race average 110 30 27 L. alabamica Rises: al 81 2 540 0.5 0.1 82 4 576 2 0.4 923 4 394 11 4 Race average 469 5 2 a2 91 7 297 53 19 a3 275 9 295 40 14 a4 445 2 295 52 18 EVOLUTION IN LEAVENWORTHIA 49 and the number of flowers which set seed without being manipulated, were counted. The results (table 2) show that in both 1962 and 1963 there is a clear difference in the percentages of flowers which set seeds spontaneously in self-incompatible and self- compatible populations. In 1962 the average percentage of flowers which set seed spontaneously (‘auto-fertilized’ — Drayner, 1959) was 24 per cent or less in the four self-incompatible populations grown, and 42 per cent or more in the five self-compatible populations. The failure of some of the flowers to set seed spontaneously in the self-compatible populations is largely or wholly due to a failure of the pollen grains to be trans- ferred from the anthers to the stigma of the same flower. In addition, there are conspicuous differences between the self-incompatible populations, and also between the self- compatible populations. In the three populations with relatively strong incompatibility barriers, populations 171 (race c1), 354 (c2) and 81 (a1), 5 per cent or less of the flowers were auto-fertilized, but in population 37 (race ¢3) which has a weaker self-incompatibility system, 23 per cent of the flowers were auto-fertilized. In the self-compatible populations of races ¢5, c6 and c8, which have extrorse anthers and petal limbs which bend at right angles to the petal claws in open flowers, between 42 and 68 per cent of the flowers were auto-fertilized. But in population 370 of race c15, 91 per cent of the flowers were auto-fertilized. Race c15 has introrse anthers and semi-erect petal limbs ; these characters may increase the possibility that pollen is deposited on the stigma of the same flower. Plants of at least one population in each of the described races of L. crassa and L. alabamica were grown in 1963. The results (table 2 & fig. 8) parallel those obtained in 1962, ut the percentages of flowers auto-fertilized are consistent- ly lower, particularly for the more advanced races. This is Probably due to the fact that the 1963 plants were in general less vigorous, particularly in the later stages of flowering, than the 1962 plants. The amount of auto-fertilization apparently varies considerably within one race and is Probably dependent on the vigor of the plants. In 1963 the Percentage of auto-fertilization was again consistently higher in the self-compatible populations than in the self- 50 LLOYD incompatible populations, and almost all of the populations in which more than 20 per cent of the flowers were auto- fertilized have both introrse anthers and semi-erect petal limbs. The appreciable auto-fertilization in all self-compatible populations under insect-free conditions and in the absence of strong wind currents indicates that these races are well adapted to autogamous reproduction. The percentage of self-fertilization in nature in the races with additional adaptations to facilitate the transfer of pollen from the anthers to the stigma of the same flower might well be as high as 95 or even 99 per cent, since the flowers of these races receive few insect visits (see below). Inflorescence measurements: The size of the plants of each population grown in the greenhouse in 1962 and 1963 was measured by the number of flowers produced on healthy plants (those with less than 5 per cent of abnormally arrested flowers). The results (table 2 & fig. 9) are given as population and race averages. In those populations for which figures are available from 1962 and 1963, the 1963 figures are considerably higher and in some cases double the 1962 figures. This is perhaps surprising since the 1962 plants were more vigorous in other respects, e.g., average seed weight. The 1963 plants benefited from several im- provements in technique, however, and their early growth was superior to that of the 1962 plants. The 1963 plants did not begin to show signs of poor health until after flower- ing had begun. In addition, the heavier seed set in 1962 by auto-fertilization may have reduced the number of fiowers produced that year, particularly in self-compatible popu- lations. The population averages of the total number of flowers produced on each plant show a reasonably close agreement between the populations of one race, in either 1962 or 1963. The differences between populations of a race are generally small compared with the differences between races and are probably due to differences in the health of the plants of the separate populations. The race averages show a clear trend of decreasing number of flowers per plant from the most primitive to the most advanced race in both species. Apparently the loss of self-incompatibility has been accompanied by an evolution- EVOLUTION IN LEAVENWORTHIA 51 ary trend toward fewer flowers per plant (under near- optimal conditions). This trend is apparent in the 1962 and 1963 data for L. crassa races, and in the 1963 data for L. alabamica (only one race of L. alabamica was grown in 1962). In both years the most primitive race of L. crassa (cl) had more than four times as many flowers per plant on the average as the most advanced race (c15). The difference between the races is not as great in L. alabamica, where in 1963 the most primitive race (al) had an average of almost twice the number of flowers per plant as the most advanced race (a4). In general, races which are intermediate in the overall advancement in regard to the characters associated with the loss of self-incompatibility have intermediate numbers of flowers, but it is apparent that the order of the plant sizes is not precisely the order of advancement of the races. In 1963 particularly, several races such as ¢c7 and cl0 had considerably more or fewer flowers per plant than would be the case if the evolution towards fewer flowers per plant was completely in step with the other characters which have evolved in association with the loss of self-incompatibility. The production of flowers was analyzed further in the 1963 material by separate counts of the numbers of scapose and branch flowers, and the number of branches produced on each plant (table 3). The results show that the evolution towards fewer flowers per plant has been achieved by a reduction in the number of branch flowers and not by a reduction in the number of scapose flowers. There has been no consistent change at all in the number of scapose flowers in L. crassa; there is actually a slight increase in the number of scapose flowers in the more advanced races of L. ala- bamica, but the differences between the races are small compared with the decrease in the number of branch flowers. The average number of branch flowers can itself be divided into two components —the average number of flowers per branch, and the average number of branches at plant. Table 3 shows that the difference between primitive and advanced races of L. crassa and L. alabamica in the average number of branch flowers is in turn due to the number of flowers per branch. There is no clear difference between primitive and advanced races in the average number of branches, except that the two most advanced 52 LLOYD TABLE 3. BRANCH AND SCAPE MEASUREMENTS AND FLOWER COUNTS ON PLANTS GROWN IN GREENHOUSE IN 1963 Z m ov = z $8 35 z : g a ‘ ae es En 8a Ge res] ree 33 a. vs #8 & 4, & $82 gb de 3= feat es. 6COelCUOU ee URE EE ne cate eae d= sk gf 2S Se “ie as Pa oa EY a) 20 < S$ 00"s> 006 006> 0s°0> 00°6 > 00¢> o°orI> Or paouvapy 61-0T 99°0-88°0 6668-000P of'S-00'S of8T-.06 ss, GL°0-09°0 66°6-00'°6 662-002 6'61-0'0T $ 9} BIPouLezUT ol> £8°0-00°0 0009< ogs< ef8I<= 0 06 gL'0< 00°0T< 00s= o07<= OO OATIUAL or 6 8 L 9 9 v g j I ul} pezyiytey — ~BUT[od-J[98 xepul quant “un “uid -oyne uo 64[Nset eynAao *UtUL squsUIe[y pue me Yy}dep uol “UU s1aMmoy yisus] s1aMoy JO eArT}Isod Jo 2 ua]jod yisuel uauieys Aq [Bjod useM ~-vUld.eUIe yysue] jo younaq WM sy woul “AV [sid ‘ay peye}or s[suy -Joq e[suy = [Bjed ‘AV [eyed “ay ‘ou “AV *XBUL ‘AY a109g SUIT] ssBlO PUv StoPBIBYO seep SsBIO SAQIGNI LNAWSONVAGVY AHL OL DNILAGIMLNOO SYSLOVYVHO TIV 40d SAUOOS SSVIO UNV SLINIT SSVIO °8 aTav dL, 72 73 EVOLUTION IN LEAVENWORTHIA “39 78 UMOYs Jepso oy} Ul ‘*sydiour J0[OO JeMOy YIOq J0OJ g put YL ‘Pp ‘g BUUINOO UT AjazwAUdes UsAIB 91" BO1008 OY] *g2 puw go ‘go saovt ofydiourAjod ay} 10g “g 2148} Ul UAAIS are saroos sBBlO PUL BzTUNI] SsETO e4UL “8 8144} Ul SB pasequINU oIv S1d}OBABYD oJ, OL OL g or or g OT or oT 9 g 0 paieyued-moT a4 pe 0g 0g g or g 0 0 0 0 0 g gs pe.iaqued-Mo][aA gu SZ SZ g OT Ss 0 0 0 0 0 g 0 pozezted-Moppod qu 0 0 0 0 0 0 0 0 0 0 0 0 P949}190-MO][OA ye teony Dovwpgnyy "Ty Oo oor Or or or OT or Or Or OT or or MOTTA 910 98 Sg or or or or g or 9 Or or g pate] U99-MoT[a4 at) 08 08 Or or Or g or or 0 g Or or MOTI[AA g1e 08 08 or or or OL 0 0 Or or or or MOTTO a SL SL Or Or or g 0 or g g Or or pate} Uad-Mo][ad Tr OL OL OL Or Or g ¢ or g 0 OL g pe.10}Ua0-Mo] [ak ie) og og g Or or Or 0 0 0 S s g po.iequad-Mo] a4 6° MOT[AA Q°Ly e¢'0P g or Ors o*g 0 0 ors oro g g *p2.1a]U90-Mo] [ah §9 or OF S Or g OL g 0 0 0 0 g pa.i9e}u90-Mo] [04 Le og 0g S ol g 0 0 0 0 0 g g pose} uaa-Mol[a4 90 MO]Ja4 9°%z S202 ¢ or o‘g 00 0 0 00 $0 0 0 ‘pet9}ua0-Mol[a& go ST ST 0 g g g 0 0 0 0 0 0 pesequao-mo]jads po 74 LLOYD ces. Races cl12 and ¢13 both have advancement indices of 80; race c12 has extrorse anthers, and has therefore been ranked ahead of race c13, which has introrse anthers and may be more frequently self-pollinated. The advancement indices indicate roughly how far each race has evolved in the characters associated with the evolution of self-compatibility. It is apparent from table 9 that the races of L. crassa and L. alabamica show a wide variety in the extent to which they have evolved in these characters. The advancement indices probably also broadly parallel the percentage of self-pollination in each race; the correspondence will not be perfect, however, since some of the characters contributing to the advancement indices do not directly affect the ease of self-pollination, while others have a strong effect. THE PATTERN OF EVOLUTION OF CHARACTERS ASSOCIATED WITH THE EVOLUTION OF SELF-COMPATIBILITY The discussion of the individual characters associated with the loss of self-incompatibility emphasized that the order of increasing advancement in the various characters often did not correspond exactly. Apparently the evolution of these characters has not followed a completely constant pattern. For example, race c15 is the most advanced race of either species when all characters are considered, but it is the most advanced race in only a few of the single charac- ters. In race c7 the pistil is shorter than in any other race of either L. crassa or L. alabamica, yet overall this race is only moderately advanced. Similarly race c11, and not race cl, has the shortest (most advanced) branches of any race of either species, race c12 has the shortest petals, etc. There is nevertheless a strong parallel in the evolution of these characters in each race, and the pattern of gain of advanced characters seems to be fixed to some extent. The most advanced races in both L. crassa and L. alabamica are moderately to highly advanced in all characters. And the most primitive self-compatible races (c5, c6 and a2) are similar to the self-incompatible races in all characters. The degree of constancy in the pattern of evolution of advanced characters may be seen best when the measure- ments obtained on each race for a given character are plotted against the advancement index of the race concerned. EVOLUTION IN LEAVENWORTHIA 75 The resulting graphs are shown in figs. 7 to 14 for the ten characters which contribute to the advancement indices. The two semi-quantitative characters, the degree of opening of the petal limbs and the angle rotated by the anthers of the paired stamens, are shown in fig. 8, where the possession by a race of the advanced expression of either character is shown by the addition of a short line to the dot marking the position of that race. For each of the remaining eight characters the largest measurement among the racial aver- ages of all races of both species was noted, and the racial averages were converted to percentages of this maximum measurement. The percentages of the maximum measure- ment were then graphed against the advancement indices of the races. The resulting graph for each character shows how that character has evolved in relation to the ten charac- ters contributing to the advancement indices, considered as a whole. A straight line for all eight characters would indicate that they evolved simultaneously and at the same rate in relation to each other in all races. The slope of a line would indicate the magnitude of the change in relation to the absolute size of a character. A consideration of figs. 7 to 14 shows that for six of the characters contributing to the advancement indices — name- ly the percentage of auto-fertilization, the maximum branch length, the number of flowers per plant, the average petal length, the average petal emargination depth and the aver- age pistil length — the points show a reasonable agreement with a straight line. These characters have evolved gradually and at a relatively uniform rate in relation to each other in all races. But it is obvious from the varying amounts of scatter in the graphs that some characters, e.g., the average number of flowers per plant, have evolved more uniformly than others such as the average petal emargination depth. Moreover, there are occasional points widely separated from all others on a graph, e¢.g., that for race a4 (advancement index 70) in fig. 10. Such points indicate exceptions to the relatively uniform evolution of a character. The amount of change in relation to the magnitude of the character in- volved in the six graphs approximating straight lines varies considerably. For example, in the trend towards a decrease im the length of branches, the lowest average measurement is only 14 per cent of the highest average measurement, 76 LLOYD whereas the lowest average petal length measurement is still 62 per cent of the highest average measurement. The graphs for the percentage of positive results on self- pollination versus the advancement index and for the pollen: ovule index versus the advancement index obviously do not follow a straight line. This reflects the fact that self- «100 “ ie *3e 100 Z ‘ a . " Pa S| S ° Z 80 80 . 6 6 ‘ 2 z 4 5 bes wy i 60 E 60 i ud a w ” -« 7 ° 2 fe) wos ie 5 O 40 < 40 e Z >< oe O < 4 . ¥ = Zab x 29 nu >< FIG,7 < = ~ rel lL i 0 20 60 100 ADVANCEMENT INDEX oor @ 1 ° . . He be . e ee x = . . 5 ales z r4 * re} we . = 60) mo Gee Lae . c= 5 . fe) Fd : ya . ys a “ < = 2 le 5 » . * ° ood . < z ae = ~ e 7 e . 20) = a 20 FIG.9 i 1 1 1 atid _ 40 60 80 100 ADVANCEMENT INDEX 101 100 . vd , ° e ee. ° x . mf 80 . « w * oe sf e = ” 7 * ° Aa ied z - = 3 ae” 3 . z 53 Ei Ww 60 Z 60 g 9 . < < " z ° an . =e 4° ° = to ° a a ee 20 < = 20 FIG.11 sd EVOLUTION IN LEAVENWORTHIA 77 oor bed ey e . Be ee so® < gob . _ eae a ¢ mo ° = o ° e . w e & 60F ° - > 60 be ra t i 5 B ie ° a a = = bed > € >< 40F oO 40 . . . * < a ° . = sé . ~ < 2 20- pe 20F FIG, 13 FIG, 14 ° 0 4. 1 1 0 20 40 60 80 100 0 20 40 60 80 100 ADVANCEMENT INDEX ADVANCEMENT — INDEX . 7 To 14. Graphs of the relationships between race averages of measurements i bed races of wee characters and the advancement indie f the nineteen descri ra of L. crassa. The measurements on the ciate are shown as percentages of the largest racial average for a charac 8, a horizontal line attached to ag er. In FIG, a point indicates a race with regularly introrse anthers, and a vertical line indicates a race in which the petal limbs bend at less than 90° es the petal claws in open ers. incompatibility was lost before the other characters had evolved very far, and most of the decrease in the pollen: ovule indices occurred during the loss of self-incompatibili- ty, and there has been only a small amount of subsequent evolution of this character in the self-compatible populations. On the other hand, the evolution of introrse anthers and of semi-erect petal limbs has occurred only in races with advancement indices of 70 or more (fig. 8), that is, only after the self-compatible populations have evolved to a con- siderable extent in other characters associated with the loss of self-incompatibility. And as mentioned previously, all of the races of L. crassa and L. alabamica have retained odoriferous flowers. The loss of odor seems to ne one of the final stages in the evolution of Lea thia towards Increasing autogamy and loss of adaptations to insect pollination, for it has occurred only in the uniformly self- — species of Leavenworthia, which are all more advanced in their reproductive systems than any of the races of L. crassa and L. alabamica. In summary, it may be said that the evolution of self- “De Ec and of all the other characters associated with as been gradual and with few exceptions has followed a sited constant pattern. Most characters have evolved at a uniform rate in relation to each other, but the evolution 78 LLOYD of self-compatibility and the reduction in the pollen :ovule index have largely preceded other changes, and the evolution of introrse anthers, semi-erect petal limbs and particularly of non-odoriferous flowers has occurred only after the other characters have become quite advanced. THE EVOLUTION AND RELATIONSHIPS OF THE RACES OF L. CRASSA AND L. ALABAMICA The loss of self-incompatibility and the concomitant changes in thirteen other characters have occurred in some or all of the races of both L. crassa and L. alabamica, And at least some of these changes have occurred twice in the ancestors of other species of Leavenworthia (Rollins, 1963). This parallel evolution in at least four separate phyletic lines in the genus suggests the possibility that the changes have occurred more than once in either or both L. crassa and L. alabamica; so characters associated with the evo- lution of self-compatibility have a very limited use as indicators of relationships between the races of L. crassa and L. alabamica. Since these characters are believed to have evolved only in one direction (towards increasing autogamy and a decreasing attractiveness of the flowers to insect pollinators) they may be used instead to indicate how primitive the common ancestor of two or more races must have been. For example, as there are self-incompatible races in both L. crassa and L. alabamica, and self-incom- patibility has been lost but not regained, the common ancestor of L. crassa and L. alabamica must have been self-incompatible. By similar arguments the common ancestor of L. crassa and L. alabamica was also primitive in other characters associated with the change in the breeding system. If a phylogeny of the races of the two species could be established, it would be possible to work out from this whether characters such as self-compatibility and introrse anthers have evolved more than once in each species. Fortu- nately there are characters distinguishing between the races of each species which are not associated with the change in the breeding system. Some of these are shared by two or more races and may be used to indicate relation- ships between the races. EVOLUTION IN LEAVENWORTHIA 79 100 100 L.crassa INDEX L.alabamica ADVANCEMENT Fic. 15. Diagram of the suggested phylogeny of the races of L i the common a L. crassa. The point O represents neestor € advancement index ertical axis) indicat race has evolved in characters associated with the evolution of self-compatibility. The distance ce from O on the h ontal is represents approxi h ar ace has evolved in Related ra e sed oriz axi m characters not associated with the breeding system. ces are enclo within a curve; less inclusive curves indicate a closer relationship between enclosed races, Moreover, the large number of races in the narrow Moulton Valley, and particularly within an area of ap- proximately ten square miles in the Massey-Lebanon district, Suggests that the races have recently diverged in this area and have migrated only short distances, so that the most closely related races may be geographically adjacent. This is apparently the case for several pairs of races, and offers valuable supporting evidence for the proposed phylogeny of the races of L. crassa. The distinguishing characters and the relationships of each race have been given in the racial descriptions and 80 LLOYD only the general features of the movements and evolution of the races will be considered here. The relationships described under each race are shown in fig. 15, based on a modification of the hypothetical scheme used by Sporne (1956, fig. 7) to indicate the relationships in a group when the component taxa can be placed on an axis of increasing advancement. The position of each race on the vertical scale in fig. 15 is determined by its advancement index. The most primitive races (advancement indices=0) are at the base of the figure and the most advanced race (c15, with an advancement index of 100) is at the top of the figure. The point O repre- sents the latest common ancestor of L. crassa and L. ala- bamica. The horizontal distance of a race from O shows approximately how far it has diverged from the common ancestor in characters which are not associated with the evolution of self-compatibility. The position of a race on the horizontal axis is approximate, and based on a subjective assessment of characters which vary independently of the breeding system. The most closely related pairs of races are indicated by dotted lines around each pair. Less closely related races are shown by light solid lines around them. The races of L. crassa and those of L. alabamica are enclosed by heavier solid lines. The relationships are based on only a few characters and only those affinities which are reasonably certain have been included in fig. 15. The relationships of the more distantly related races in each species are uncertain and are not included in the diagram. If the affinities of the races of L. crassa and L. alabamica given in fig. 15 are correct, it is apparent that many of the characters associated with the evolution of self-compatibility have evolved in parallel directions in several phyletic lines. There is no doubt that the changes have occurred inde- pendently in L. crassa and L. alabamica, and many of the quantitative and qualitative changes appear to have occu more than once in L. crassa at least. : The quantitative characters vary continuously and it 1s highly probable that these characters have evolved gradually and repeatedly by gene changes at many loci. There }§ little point, therefore, in describing the extent of the quantitative changes in each phyletic line. The loss of self-incompatibility, and the evolution of sub- EVOLUTION IN LEAVENWORTHIA 81 erect petals and introrse anthers (paired stamens) can be considered qualitatively, and it is possible to decide the minimum number of times these changes have occurred in L. crassa and L. alabamica. The loss of self-incompatibility has occurred at least once in L. alabamica and perhaps as many as four times, since the four self-compatible races are all geographically distant from each other. Self-compatibility has evolved three (or more?) times in L. crassa — in the ancestors of race c6, of races c5 and c8, and once or more in the ancestors of the remaining self-compatible races. The evolution of introrse anthers has occurred one or more times in L. alabamica (race a4 and the ‘Russellville’ and ‘Tuscumbia’ races) and at least twice in L. crassa —in races c10 and c14, in race c13, and perhaps independently also in race c15. The evolution of semi-erect petal limbs has occurred in race a4 and perhaps independently in the ‘Russellville’ and ‘Tuscumbia’ races of L. alabamica, and in L. crassa three or more times — in races c10 and cl14, in race cll, in race c13 (where it is often combined with a delayed opening of the petals) and possibly independently in race c15. Thus the evolution of self-compatibility, introrse anthers, semi-erect petal limbs and associated quantitative characters has occurred independently in several phyletic lines in L. crassa and L. alabamica (and at least twice in the ancestors of other Leavenworthia species — Rollins, 1963). These changes have probably been facilitated under the appropri- ate ecological conditions (discussed below) by the ease of self-pollination in Leavenworthia flowers. The anthers of the paired stamens closely surround the stigma, and the unaided transfer of pollen from the anthers to the stigma of e same flower is readily accomplished. The close proximity of the anthers and the stigma allows autogamous pollination to be effective as soon as self-compatibility has evolved; this is confirmed by the considerable auto-fertilization found under greenhouse conditions in the extrorse self-compatible races, The self-incompatible races of L. crassa and L. alabamica may be compared with the self-compatible races with regard to the magnitude of the differences between the races, 1.€. the amount of evolution that has occurred in self-incom- patible and self-compatible phylads. In general, the self- 82 LLOYD compatible races are more distinct than the self-incompatible races. This applies both to characters which are associated with the change in the breeding system and to those which are not, and is reflected in the generally greater distances between self-compatible races along both the vertical and horizontal axes in fig. The self-incompatible races are more similar to each other in the characters associated with the breeding system, since the majority of these characters are adaptations to bee pollination, and apparently have been held constant in the self-incompatible races of L. crassa and L. alabamica (and in L. stylosa) by stabilizing selection caused by the need for cross-pollination. The self-compatible races, on the other hand, are variously adapted to a mixture of cross- and self- pollination, and there are considerable differences between the races in the characters associated with the evolution of self-compatibility. In other words, there is only one adaptive peak in Leavenworthia for those species and races which rely on cross-pollination, but numerous adaptive combi- nations of the same characters have been adopted by those races which are frequently or predominantly self-pollinated. The greater variety of the self-compatible races in charac- ters which are not associated with the evolution of self- compatibility is more surprising. Many of these characters are leaf characters, and it is very striking that most of the aberrant leaf characters occur in advanced races of L. erassa and L. alabamica. It may be that the more marginal habitats occupied by self-compatible races (see below) vary more than the relatively favorable sites occupied by the self-incompatible races. Morley (1959) believes that “in general the rate of evolution has been independent of popu- lation characteristics . . . and is principally a function of rates of changes in the environment.” Whether or not this is the explanation for the generally greater differences between self-compatible races than between self-incom- patible races, there is no evidence in L. crassa and L. aa- bamica that a lower genetic variability in self-compatible races has restricted their evolutionary potential. THE ‘eae DISTRIBUTIONS OF SELF-INCOMPATIBLE ND SELF-COMPATIBLE RACES att donnie species are often more widespread than EVOLUTION IN LEAVENWORTHIA 83 their self-incompatible relatives (Baker, 1953) ; this is very striking in those genera with one or more widespread self- compatible weed species. In the Leavenworthia species with 15 chromosome pairs, L. stylosa, the only self-incompatible species, has the narrowest distribution; L. uniflora, the species with the most advanced breeding system, is also the most widespread species of the group (and of the genus as a whole — Rollins, 1963). This suggests that the self- compatible species are better able to spread into new areas. The distribution of the races of L. crassa offers a striking contrast to that of the species with 15 chromosome pairs. Not all of the glade sites are known, particularly for the self-incompatible races, but present knowledge suggests that the self-incompatible races are in general more widespread than the self-compatible races. Except for race ¢3, the glade sites of the self-incompatible races of L. crassa are incompletely known, but the distri- bution of the cultivated populations of races cl, c2 and c4 Suggests that there may be a number of well separated glade sites for each of these races. In addition the self-incompati- ble races are spread over a distance of approximately ten miles. On the other hand all of the self-compatible races of L. crassa are extremely local and all except race c6 occur in the Massey-Lebanon district of Morgan County where the most widely separated glades of L. crassa are only one and a half miles apart. Race c6 occurs approximately five miles southwest of the Massey-Lebanon district, and has probably ome self-compatible independently of the Massey- Lebanon races of L. crassa. The more restricted distribution of the self-compatible races of L. crassa is probably related to their very recent differentiation, They simply have not had sufficient time to Spread very far. It is noteworthy that the most advanced races in both L. crassa and L. alabamica — races ¢15 and a4, respectively — are the most widespread races in the Massey- anon district, although they have spread less than two miles. In addition, L. uniflora is more advanced than any Trace of L. crassa or L. alabamica, and is more widespread in the Massey-Lebanon district and surrounding areas than either race c15 or a4, Soona very small scale in the Massey- Lebanon district the races of L. alabamica, L. crassa and L. uniflora show a distribution pattern similar to that of the 84 LLOYD n=15 species; but within L. crassa as a whole, there is a reverse pattern, with the more primitive races being more widespread. In L. alabamica, glade sites of the self-incompatible race al are spread over a distance of approximately 15 miles in Franklin and Lawrence counties (map 1). The glade popu- lations of the self-compatible races a2 and the ‘Tuscumbia’ race occur over an almost equally large area, but the self- compatible races a3 and a4 are restricted to the smaller Massey-Lebanon district. No glade site is known for the self-compatible ‘Russellville’ race but the position of the only known population at the head of a narrow valley suggests that the Russellville race has the narrowest distribution of all L. alabamica races. The four self-compatible races of L. alabamica are widely separated in the Moulton and Tennessee valleys. This does not mean, however, that a common self-compatible ancestor of these four races has migrated a long distance through these valleys, as there is a strong possibility that self-compatibility has evolved several times in L. alabamica. There is, therefore, no clear difference between the distribution of self-incompatible and self-compatible races of L. alabamica. It will be shown below that the advanced races of L. crassa and L. alabamica flower earlier than the primitive races. Within the n=15 group of species too, L. stylosa, the most primitive species, continues flowering the longest, and L. uniflora, the most advanced species, has the earliest flowering period. I believe that the earlier flowering of advanced races and species of L thia explains their wider distribution in the Massey-Lebanon district, and in the n=15 species. These races and species are capable of existing in more marginal habitats, and can consequently spread over areas that provide an impassable barrier to the primitive races and species. According to this explanation, the distribution of the races of L. crassa is controlled primarily by historical factors, viz. the recent origin of self-compatible races, and the ability of advanced races to occupy marginal habitats has had little chance to influence their distribution. But 1 the older species with 15 chromosome pairs the relative distribution of the species is primarily controlled by their ecological tolerance, so there is an inverse relationshiP EVOLUTION IN LEAVENWORTHIA 85 between primitiveness and area occupied. The situation in L. alabamica is intermediate between that in L. crassa and that in the species with 15 chromosome pairs. THE SPREAD OF L. CRASSA AND L. ALABAMICA ONTO CULTIVATED LAND The clearing of forest in the Moulton Valley since about 1840 has made large areas of ground near cedar glades available for occupation by L. crassa and L. alabamica populations. All of the races of L. alabamica and all but four of the races of L. crassa (races cll, c12, c13 and ¢c14) have spread to some extent onto land from which the origi- nal forest has been cleared. Some of these races, such as c9 and cl0, have spread only a very small distance from the parental glade site, but others, particularly races c2, ¢5, cl15 and a4, have moved up to three miles over cultivated land and have produced a large number of populations growing on suitable cultivated land. On regularly cultivated land the secondary populations may contain ten or a hundred times as many plants as the glade sites from which they came, and many of the individual plants may be as large as, or larger than, the plants grown in the greenhouse. The cedar glades are unsuitable for cultivation and have remained relatively undisturbed, except where the forest has been cleared from their borders. Consequently most if not all of those glades that originally contained Leaven- worthia populations still do. It is therefore possible to trace the origins and movements of the populations on cultivated land with considerable certainty in most cases. This does not apply of course to those races (cl, c6, c8 and a3) for which no glade site is known. The distribution of the field populations of the races of L. crassa and L. alabamica is shown in maps 1 to 5. The movements of the populations of each race are discussed under the description of the appropriate race above, and only the general aspects of the spread of L. crassa and L. alabamica onto cultivated ground will be dealt with here. There is no relation between the breeding system of a race and the extent of its spread onto cultivated land. Instead the extent of the spread of each race seems to be principally determined by the accessibility of suitable land adjacent to the cedar glade populations. The most important 86 LLOYD factor in this regard is whether any of the forest surround- ing a glade has been removed. Only two of the known olades still completely surrounded by forest have produced any populations on cultivated land. One of these glade popu- lations, number 769 of race c9, has produced a single small population immediately outside the forest, directly downhill from population 769 in a drainage depression. The other glade population, number 698 of race c10, is connected to cleared land by a narrow farm road and a few seeds have apparently moved along this, perhaps on farm machinery, to form a small population in the adjacent field. The Massey-Lebanon district is the only area where both the glade and derived field populations have been exhaustive- ly traced. It is obvious here that almost all of the movement of populations across cultivated land has been a slow spreading of the populations over suitable land. Only two races, cl5 and a4, appear to have spread through down- stream dispersal of their seeds. The reason for this is that the Massey-Lebanon district is on the southern edge of the Moulton Valley, and only a few small streams run through the area occupied by L. crassa and L. alabamica. None of these originates in the low hills containing cedar glades on their slopes and at their bases. Consequently the races of L. crassa and L. alabamica in the Massey-Lebanon district have reached the streams at some distance from their source, where the streams flow between well-developed banks, below the surface of the land. At this stage in their course the streams seem to offer a barrier to the movement of Leavenworthia rather than an avenue for dispersal. The other areas occupied by Leavenworthia are less wel known; many of the known cedar gilades in the Moulton Valley outside of the Lebanon-Massey district occur beside streams flowing over exposed bedrock, and it is possible that downstream transport of seeds has played a much more important role in the movement of L. crassa and L. ala- bamica over cultivated ground elsewhere than in the Massey- Lebanon district. Once a population has spread onto land cleared of forest its size and subsequent migrations are controlled by the nature of the agricultural practices in the immediate vicin!- ty. Rollins (1963) demonstrated experimentally that L. crassa, L. alabamica and their hybrids are unable to with- EVOLUTION IN LEAVENWORTHIA 87 stand competition from more aggressive weeds. Corn fields are ploughed in the spring, usually after Leavenworthia has fruited but before most other weeds set seed; consequently Leavenworthia is virtually free of competition from other plants and grows most luxuriantly and abundantly in such fields. Fields which have been left uncultivated for some years have a tall but incomplete cover of weeds and Leaven- worthia is less successful in these. Fields sown in pasture are marginal sites for Leavenworthia, and populations cannot survive indefinitely unless the fields are reploughed. The precarious existence of L thia on grazed land was dramatically demonstrated by a population of race c15 two and a half miles south-southwest of Lebanon. In 1961 this was a small population of perhaps 50 plants in a wet depression within a pasture. The plants were mostly large and vigorous and they set abundant seed, but in 1962 and 1964 there was not a single plant, although there had been no apparent change in the farming of this field. The Leavenworthia plants on regularly cultivated land (corn fields, or occasionally cotton or oat fields) are usually in the last stages of flowering when the fields dry sufficiently in the spring to be cultivated. In most years they apparently manage to produce large numbers of viable seeds, for Leav- enworthia may cover up to several acres with one or more plants within each square foot. Even on cultivated fields, however, the number of Leavenworthia plants may be drastically reduced in one year if the field is ploughed earlier (or more deeply?) than usual or is left uncultivated. For example, in 1961 population 37 (race 3) covered the whole area of the corn field it occupied, and there were many large plants. In the spring of 1961 only about two- thirds of the field was ploughed (on April 23); it was one of the first fields in the area to be cultivated. The following year there were many smaller plants in the uncultivated portion, but on the cultivated portion the number of plants was perhaps only one hundredth of the number present in the previous year. Within many cultivated fields, and in all pastures and neglected fields, Leavenworthia occupies only a portion of the total area, and often occurs as several separate and Scattered populations. The limited distribution of the races in many suitable fields, and the smal] distances travelled by 88 LLOYD all of the races (never more than 3 miles in the Massey- Lebanon district at least) confirm the extremely limited powers of dispersal of L. crassa and L. alabamica noted below. The seeds have no special adaptations for dispersal, and much of the movement over the fields is probably through the movement of soil on farm equipment. INTERSPECIFIC AND INTER-RACIAL HYBRID POPULATIONS IN THE MASSEY-LEBANON DISTRICT The distribution pattern of the races of L. crassa and L. alabamica is even more complicated in fields which are occupied by more than one race. This occurs only in the Massey-Lebanon district, where the glade sites of thirteen races are within an area of approximately ten square miles. In each case where separate races of L. crassa and/or L. alabamica have come into contact there has been some hybridization, although none of the mixed populations is uniformly intermediate between the component races. For example, populations 276 (race a4) and 298 (race c7) are contiguous, but the bulk of the plants appear to be either pure race a4 or pure race c7, and there are only a few hybrids at the point where the races are in contact. The absence of large numbers of hybrids is probably due to the predominant inbreeding in both races and to the limited dispersal of seeds, rather than to the presence of isolating mechanisms, I have, however, done no work on the compo- sition of the hybrid populations other than to establish their existence and identity and to plot their distribution and extent. The maps of the distributions of the races of L. crassa and L. alabamica in the Massey-Lebanon district (maps 2 to 5) show that there are many interspecific and inter-racial hybrid populations, and sometimes several combinations are present in one field. The complexity of the racial distri butions is greatest in the series of fields one and a half to one and three-quarter miles southeast of Massey where four races (c5, c7, cl5 and a4) have spread onto the same fields. The following is a brief summary of the hybrid popu lations occurring in the Massey-Lebanon district. The individual interspecific and inter-racial mixtures and hybrid swarms are discussed in more detail under the descriptions EVOLUTION IN LEAVENWORTHIA 89 of the races. With one exception the mixed populations occur on secondary sites from which the forest has been removed. The single exception, described on page 94, is the small amount of admixture between the contiguous glade populations 742 (cll) and 741 (race ¢c12?). Adjacent populations in a series of fields with the same racial combination may be considered as due to a single contact between the races in that area. Using this simplifi- cation, there are ten distinct areas where races of L. crassa and/or L. alabamica have met and hybridized; nine of these are on cultivated sites. The numbers given here for each contact between two or more races correspond to those on the maps of the Massey-Lebanon district (maps 2 to 5). 1. Races c3 and c5 have become mixed in a complex series of corn field populations one and a half crac northeast to one and three quarter miles north-northeast of Mas 2. Races eh and cl5 (and c7 gets are mixed in four adjacent populations one and a half miles southeast of Massey. 3. Races c5 and a4 are mixed in the area one and a quarter to one and a half miles east of Massey. 4. A population of race c5 one and three-quarter miles southeast of of Massey. There are probably minor contributions from race ¢e5 also in the easternmost populations of the series. 6. Races c7 and a4 are mixed at the common border of contiguous populations of the two races one and a half miles south-southeast of Massey. 7. Races c11 and c12 have mixed to a small extent in the contiguous glade populations 741 (race c12?) and 742 (race cll) one and a half miles east-southeast of Massey. 8. Races cl5 and a4 are mixed in a corn field two and a half miles South-southwest of Lebanon. 9. Races c15 and a4 are also mixed in scattered populations one and a quarter miles south-southwest to three-quarters of a mile east of on. 10. Races a3, a4 and cl5 are mixed in portions of the only known a seston of race a3, two and three-quarter miles south-southeast of Massey, Ignoring the possibility that race c7 may occur in area 2 above and that race c5 may occur in area 5, the ten areas represent eight different racial combinations within an area of approximately ten square miles. Two combinations (races 90 LLOYD e5 and a4, races c15 and a4) have occurred twice in widely separated areas. Four races of L. crassa (c3, c5, c7 and c15) and two (a3 and a4) of L. alabamica contribute to the mixed populations on secondary sites; two other races of L. crassa (c11 and c12?) have mixed slightly at one glade site. With regard to the species involved, four of the contacts (including the one on the glade site) involve two races of L. crassa, five involve one race of L. crassa and one race of L. alabamica, and one involves two races of L. alabamica and one race of L. crassa. Races ¢5, c15 and a4 have come into contact most often with other races. These three races are also the ones which have spread most widely over the fields; their larger number of contacts is due to their greater spread. There are several areas where separate races of L. crassa and/or L. alabamica occur on different sections of the same field, or in adjacent fields, but have not yet met and hybrid- ized. If future agricultural practices permit the spread of these populations, additional racial mixtures will un- doubtedly occur. THE GLADE POPULATIONS OF L. CRASSA AND L. ALABAMICA Leavenworthia alabamica and L. crassa are endemic to northwest Alabama. They occur naturally on cedar glades — small isolated areas where the horizontally-bedded limestone develops shallow, denuded soils unable to support forest. The limestone outcrops in northwest Alabama occur in two parallel valley systems — the Tennessee Valley proper and the Moulton Valley (map 1). The Tennessee Valley contains only a few populations of L. alabamica, in Colbert County, on Bangor or Tuscumbia limestones of Mississippian age The Moulton Valley is a narrow valley approximately fifty miles long in an east-west direction, and three to tem miles wide (Johnston, 1930). The floor of the valley consists entirely of Bangor limestone of Mississippian age, except at the western end, where the limestone is overlain in parts by a thin veneer of coastal plain deposits of the Tuscaloosa (Cretaceous) formation. To the north, the Moulton Valley }s separated from the Tennessee Valley by Little Mountain, 4 low lying sandstone plateau 100 to 250 feet high. The Moulton Valley is bounded on the south by the Pottsville EVOLUTION IN LEAVENWORTHIA 91 escarpment, which rises 200 to 500 feet above the Moulton Valley to the Sand Mountain Plateau, an extension of the Cumberland Plateau. Most of the Bangor limestone of the Moulton Valley is covered by a thin layer, up to 25 feet deep, of regolith (material overlying solid rock). The cedar glades are scattered throughout the Moulton Valley in places where the topography and the character of the underlying rock have caused denuded limestone outcrops to be exposed. The Moulton Valley was searched for cedar glades con- taining Leavenworthia as extensively as was possible in three field seasons. The number of known glade populations of each race varies from more than ten (race al) to zero (races cl, c6, c8, and a3). The number of undiscovered glade sites is difficult to estimate, but there are probably few in the Massey-Lebanon district, which has been intensively searched. Four races of L. crassa (races cl, ¢2, c4 and c6) and two races of L. alabamica (a1 and a2) occur outside of the Massey-Lebanon district, and all of these probably have undiscovered glade populations. Race c6 is known only from one cultivated site, and may well have only a single glade population. The other races outside the Massey-Lebanon district are each known from many glade or cultivated sites, and they may all have several glade populations. In the Massey-Lebanon district there are 25 known glade sites (counting the adjacent clearings of population 743 as one glade, and the contiguous areas occupied by populations 741 and 742 as two glades) shared by eleven races of L. crassa and L. alabamica (maps 2 to 5). I would have considered that all the glades in the Massey-Lebanon district are known but for the fact that two races (c8 and a3) in the area are each known from a single cultivated population. These two cultivated populations must have come from glade Sites originally, although intensive searches of the adjacent areas have failed to reveal any glade populations of races 8 and a3. It is possible that these two races have migrated a considerable distance from their glade source, but it is still surprising that an existing glade could have been overlooked. Another possibility is that the original glade sites have had the surrounding forest removed, and have become too dry to Support Leavenworthia populations. Assuming that there is (or was?) one glade site for race 92 LLOYD c8 and one for race ¢3, there are 27 glades occupied by thirteen recognized races in the Massey-Lebanon district; that is, an average of 2.1 glade populations per race. Some of the glade populations that have not been grown in the greenhouse may represent distinct races, so the average number of glades per race may be less than two. On the other hand, several populations occupy rocky outcrops that I do not consider to be pre-agricultural glade sites, but since these sites have had the surrounding forest removed, it is impossible to be certain whether or not they are second- ary sites. The Massey-Lebanon district is only 4.5 miles long and 4 miles wide, so there are a large number of allopatric races in an extremely small area. Actually, all of the glade sites for the ten races of L. crassa in the Massey-Lebanon district for which glade populations are known occur in an area approximately 3 miles long by half a mile wide. THE LIMITED DISPERSAL CAPACITY OF LEAVENWORTHIA The existence in the Massey-Lebanon district of distinct races separated by such small distances, or even in the case of populations 741 and 742 of distinguishable populations in contact at their margins, demands that both pollen and seed dispersal must be extremely restricted. Nothing is known of the distances that non-social bees, the natural pollinators of Leavenworthia, fly, but there must be very few flights from one glade to another. Some species at least prefer to forage close to their nests (Linsley, 1958). Both honeybee and bumblebee (Bombus species) individuals characteristically forage in very small areas, sometimes for days at a time, even when suitable plants are available over a wide area (Butler, 1954; Free and Butler, 1959). The fact that the ratio of native bees to honeybees is much lower in field populations than in glade populations (see below), even where field populations are adjacent to glade popu- lations, suggests that the native bees do not habitually travel far. The self-compatibility of many populations probably also restricts the amount of outcrossing and therefore of hybridization between races (Rollins, 1963). Leavenworthia seeds lack adaptations for either wind or animal dispersal. Most seed movement probably occurs EVOLUTION IN LEAVENWORTHIA 93 after heavy rains, when some of the glade soil is carried away by the flow of water over the shallow glade soils. The glades in the Massey-Lebanon district are small, however, and have no streams running through them. so the seeds cannot be carried far by water movements. The movements of the Leavenworthia populations in the cedar glades must have been primarily controlled by the rate at which erosion exposed suitable ecological territory contiguous with that already occupied. Most of the known glades of L. crassa and L. alabamica outside of the Massey-Lebanon district have streams running through them, and it is probable that there seeds are carried downstream and establish populations at considerable distances from the parent source. It may be Significant that outside of the Massey-Lebanon district the races of L. crassa and L. alabamica are much farther apart than they are within the Massey-Lebanon district. That small distances can be effective geographical barri- ers is shown dramatically by several instances in the Massey-Lebanon district where two glades are less than one hundred meters apart, but contain different races. For example, populations 698 (race c10) and 699 (race c14) are quite distinct when grown in the greenhouse, although the margins of these two populations are only about twenty-five meters apart. If there has been any gene flow between these two populations, it has not occurred with sufficient frequen- cy to cause them to become alike. The coexistence of populations 742 (race cll), with yellow-centered flowers, and 741 (race c12?), with yellow flowers, in one glade is even more remarkable. The glade is a narrow area approximately 66 meters long and nowhere more than 4 meters wide. The southeastern section of the glade (occupied by population 742) is approximately 37 meters long, and in 1962 contained 221 plants with yellow- centered flowers, eight yellow-flowered plants, and six Plants with imperfect-centered flowers (probably inter- racial hybrids). In this section, all of the plants with yellow flowers and imperfect-centered flowers were within one meter of the northwestern section of the glade occupied by Population 741. The area occupied by population 741 in 1962 was about twenty-nine meters long and contained 218 yellow-flowered plants, 20 plants with yellow-centered flowers and 11 plants with imperfect-centered flowers. Most 94 LLOYD of the 31 non-yellow-flowered plants were close to the common margin of the two populations. Thus these two sections of the glade are occupied by contiguous, but morphologically distinct, populations. The plants of the two populations are quite continuous, and there is no break or even a decrease in the number of plants at the junction of the two populations. At both ends of the area occupied by these two races the stratum on which they occur continues around the hillside, but elsewhere the outcrop is at present less well developed and not occupied by Leavenworthia. Presumably these two races were formerly growing separately on different sections of the outcrop. As the outcrop has eroded, the flat areas suitable for Leavenworthia have moved together until recently they have become continuous. These movements must have been slow, and yet the amount of mixture of the races is still very small. There are several glades in the Massey-Lebanon district which are entirely unoccupied by L wworthia, although . they appear to be quite suitable and are occupied by other glade plants, and are only short distances from glades containing Leavenworthia. These glades are marked with a cross in map 2. In addition, there are only three glades which are occupied by both L. crassa and L. uniflora, and no glades which are occupied by L. alabamica and L. uni- flora. L. uniflora is reproductively isolated from both L. crassa and L. alabamica, and coexists with both species in many cultivated sites. Moreover, in Tennessee L. uniflora may occur in cedar glades together with up to three other species of Leavenworthia. The rarity of glades with both L. uniflora and L. crassa or L. alabamica must be attributed to the poor ability of these species to migrate over anything but continuously suitable ground. Apparently, in the ml- grations of the Leavenworthia species in the Massey- Lebanon district very few glades have become accessible to both L. uniflora and L. crassa or L. alabamica. It might be argued that differences between narrowly Separated or contiguous glade populations are maintained in spite of considerable gene flow between the populations, by ecotypic adaptation to the particular soil conditions in ea glade. Several factors suggest, however, that this is not the case. First, few of the characters separating races are of a EVOLUTION IN LEAVENWORTHIA 95 sort which can be readily thought of as reflecting edaphic adaptations. Second, in the most extreme cases, where the distances between populations are least, the distinguishable populations occur on the same stratum of the same geologi- cal formation where appreciable soil differences are least likely. Third, exactly comparable situations are equally common in the recent, radically different environment of cultivated fields, where adaptive differences parallelling those in the glades are unlikely; these situations can be much more readily explained in terms of a limited dispersal capacity and population movements in response to changing farm practices. The distribution of the species and races on cultivated ground in the Massey-Lebanon district indicates that the species have a very restricted dispersal capacity. Even if a field is totally suitable for Leave thia at present, it may be occupied by separate populations which have not been able to spread over the whole field and come into contact. This demonstrates that Leave thia pollen and seeds disperse very short distances even when there is no ecological barrier to their movement. A COMPARISON OF SELF-INCOMPATIBLE AND SELF-COMPATIBLE GLADE POPULATIONS The extremely limited dispersal powers of L thia have caused each glade population to be effectively isolated from all others. The lack of gene flow between populations has allowed many of the populations to diverge until they are now distinguishable under uniform conditions. Within the Massey-Lebanon area particularly, there are many distinct but closely related races in a very small area. These races must have evolved very close to the sites they presently occupy, and probably under conditions very similar to those existing at present. Most of the differences between the races are associated with the evolution of self-compatibility, and therefore the races of L. crassa and L. alabamica offer an exceptional opportunity to study the selective factors which may have been responsible for the evolution of self-compatibility and the concomitant changes in other characters. A detailed Study was made of the plants and the pollinators in natural 96 LLOYD glade populations of L. crassa and L. alabamica in the hope of revealing differences between the populations of self- incompatible and self-compatible races. The cedar glade populations are all relatively small (usually with less than 1000 plants). In those glades con- taining less than 1500 plants, the number of plants in each glade known in 1962 was counted at the height of the flowering season with a mechanical hand-counter. The numbers obtained (table 10) are probably slight under- estimates, since plants that had produced no flowers by this time could be overlooked, and occasionally two plants grow- ing closely together might be counted as one. For the few glade populations with more than 1500 plants, the counts are estimates made by counting over the population ten plants at a time. These counts are probably correct to within less than five per cent. There were more than 500 plants in ten of the twelve glade populations of self-incompatible races of L. crassa and L. alabamica (races c1 to c4 and al). The two self- incompatible glade populations with less than 500 plants in 1962 (populations 743d and 66) are very close to other glade populations of the same race. The number of plants 1n the glade populations of self-compatible races is strikingly different; only two of the twenty glades had more than 500 plants, and the greatest number was 710 in population 745. A number of the glades have had some or all of the forest removed from their margin and are considerably disturbed as a consequence (see table 10). The counts oe these populations may not reflect the pre-agricultural size very accurately. The remaining glades (eight of self- incompatible races, fifteen of self-compatible races) have been disturbed very little or not at all by agr icultural practices, and the figures given should be fair estimates of the pre-agricultural size of the populations. : The geometric mean of the number of plants in the eight self-incompatible populations in undisturbed glades was 1884 in 1962; the geometric mean of the number of plants in the fifteen self-compatible populations in undisturbed glades was 278. That is, on the average there were more than six times as many plants in the self-incompatible populations than in the self-compatible populations. If only the 23 undisturbed glade populations are COP EVOLUTION IN LEAVENWORTHIA 97 sidered, the populations of self-incompatible and _ self- compatible races may be statistically compared. For this purpose the distribution of the number of plants per population is obviously highly skewed. The distribution was made approximately normal by taking the logarithm to the base ten of the number of plants. The logarithms of the number of plants per population in the self-incompatible and self-compatible populations were significantly different (P<.001) by a two-tailed ‘Student’s ?’ test, i.e., the known glades of self-incompatible races contained significantly more plants in 1962 than the glades of self-compatible races. To obtain measurements of the number of flowers pro- duced by the plants in glade populations, a sample of plants was collected from each population at the end of the 1962 flowering season, and the number of buds and flowers (very few) and fruits was counted. I was unable to devise any practicable random or systematic sampling technique that would satisfactorily measure the average number of flowers produced by the plants in each glade. Any such technique would have to take into account the fact that the plants are very irregularly distributed in each glade. Leav thia plants usually occupy only a portion of each glade, and are concentrated in the most favorable parts of the glade, where they may be several times larger than elsewhere. Instead, a block sample from the portion of each population where the plants were densest and largest was taken. Fifty plants, comprising all the plants in a small area, were dug up and e number of buds, flowers and fruits was counted. (In the smallest populations only twenty-five plants were sampled). Since the counts were made at the end of the flowering season, the average number of flowers per plant in each sample may be taken as the average number pro- duced by the plants of the sample during the flowering Season. The average number of flowers per plant in a Population is taken as half the average number of flowers per plant in the sample. This method is quite approximate and probably underestimates the average number of flowers per plant in most populations. ; The estimated average number of flowers per plant in each of the undisturbed glades in 1962 is shown in table 10. The average of the estimates for the eight undisturbed self-incompatible populations is 4.92 flowers per plant, and LLOYD 98 60'S 083'T 06°F 08"6 or'z 19% arb + Lac) 8°LT rar LET 88% SIZ 89°T ess ore Let 869 i or 00 S19MOY ON PA's 6ZL‘T 98°S GL’ Tt Lvs 963 69L 2 §? UAOUY SI oS epRls ON + go - = — — _ = = — Osh = “Ba sl98 + Lo uUMOUY SI 9PIS apRls ON + go — - = — = = = ~ LIT 21811 8st 61 ePl ge"e 8827'S se°s 99°0T 69°S 6IP BSIT L°0 IT ost LP's 6F6'S 09°F 02°6 T8°% Tro 98 3°6 aT 8ST 86% 6S8 I6°T T8°s $9°S TSP T6L + go — “= an — a — — noe Sos 299 3°98 oor 9IT OL’s T66'F L9°S PE'h ers 098‘T 88 - po iaovy S’rT &@ 6ST 9's TeP'T 62°S ss°9 $9% cep PEPL 8'FT GZ 691 ¥2's 6IL‘T 66°% L6°S 9L°% 9Lg SShL 6°02 ve 89T g8°6 SOT‘L 88°L $9°ST 96°S 806 Berl = ay ae oe a, Sil ee ee 08$'‘T OSPL 9°89 c6 LST Sa'P T26‘LT 98°L LST 98°s 083% qshL eee a 3 aad a = = er oos*T ‘Bo el9 aa ae ie go UMOUY SI o}IS epBls oN - go UAMOUY ST ays epvys on ~ Jo :eoRy MSsDio “T BunemMoy sultamoy payunoo peonpoad uoljujndod yuyjd sad = zquejd aed sjuvid sjuyd uol} msnqwig sony pus sjuvid % squvyd sjuyjd siamoy Jo AQ paonpozd = s1aMmoy S1a MO] yo ‘ou jo "ON -eyndog Suypeg se1vedg “ON ‘ON ‘ou ‘4s0 s1aMOy JO ‘ou "AB jo "ON yo “lzory oz Wady *IZOT ‘ou "8 payBULLyST uo pazunoo s}uxjd Jo ajdwueg 696l NI VOINVEVIV “I GNV YSSVuO “I dO SNOLLVINdOd ACVID 40 SENONOD YAMOTA GNV INV Id ‘OT Gav, 99 EVOLUTION IN LEAVENWORTHIA yy JO SZIS [vaNqNoIAse-aid 94} eQUleser Alasop you Avur wey} ur e4L ‘A[sno108ta 4sour Sulmois ateM squRld ey} W1ayM s1aMOY ON SIOMOH ON slamoy ON OT OLT 8L 861 00T oET 00T TIT 8 e9T ST SFT. 6 6IT v vIT 8 611 SIaMOY ON *suol}ei[ndod SyuB[d Jo Joquinu ey} PUB ‘paqanyzsIp A[qusop[su0d e1B suOTZY[Ndod eBay J, ‘e[dures oY} Ul JequINU eSvI0AB oY} JIVY ST JuB[d z0ed saamoy jo Joquinu aeSsvaeae peyeuryyse worE[ndod yous Jo vere ey} uy epeul e1eM JuY[d 10d saomoy Jo soquinu oY} JO SjuNoD oY T, TL’s *9[qIyVdui00-J[98 UMOUY SI ozs apels on UMOUY ST ozIs apRls ont 000% “Bo = + ‘e[qryeduioour-jjos = — , 2 100 LLOYD the average of the estimates for the fifteen undisturbed self-compatible populations is 3.03 flowers per plant. That is, on the average, plants in the self-compatible populations produced approximately three-fifths as many flowers in 1962 as the plants in the self-incompatible populations. The difference is significant (P<.001) by a two-tailed ‘Student’s t test. So far it has been shown that self-incompatible popu- lations of L. crassa and L. alabamica contained more plants with more flowers per plant than the self-compatible popu- lations. The two parameters are significantly correlated in the 23 undisturbed populations (r=.54, P<.01). This is shown graphically in fig. 16, where the estimates of the number flowers per plant are plotted against the logarithm of the number of plants in each glade population. The self- incompatible populations are concentrated in the upper right section of the graph, and the self-compatible popu- lations in the lower left section. The number of plants in a population multiplied by the estimated number of flowers per plant in the same popu- lation gives an estimate of the number of flowers produced in 1962 in each undisturbed glade (table 10). As expected there is a considerable difference between self-incompatible and self-compatible populations in the number of flowers produced in each glade. All but two of the eight self-incom- patible populations produced more than 2000 flowers im 1962, and all but two of the fifteen self-compatible popu lations produced less than 2000 flowers. The geometric mean of the number of flowers produced in the 1962 flower- ing season in the eight self-incompatible populations was 8,650; the geometric mean of the number of flowers produced in the fifteen self-compatible populations was 736. The difference between self-incompatible and self-compatible populations was again significant (P<.001) by a two-tailed ‘Student’s # test; this is to be expected, since the tw0 contributing parameters, the number of plants per POPU lation and the estimated average number of flowers bed plant, were both significantly different in self-incompatible and self-compatible populations. It is apparent that the self-compatible populations occupy poorer glade sites than the self-incompatible populations: In the poorest glades, such as those of population 698 (race EVOLUTION IN LEAVENWORTHIA 101 8-0 ° ° 7:0. — a < = a 60F & ° a. & * ° = @ 3 5:0 6 ° e uw 1) . #O) O ie) z e ° a ° ° < as oe ° - e no a e ° r= +54 oe 20 e ® @®? A Ee eee | pn | ! 1S 20 25 3-0 aS 40 45 LOG... NO. OF PLANTS Fic. 16. The relationship between the number of plants and the estimated average number of flowers per plant in undisturbed glade populations of L. crassa and L,. “abamica in 1962. Open circles represent self-incompatible populations; closed circles represent self-compatible populations. ¢10) and 699 (race c14), it is estimated that there are less than two flowers per plant on the average (table 10). Under Poor conditions there are also fewer seeds per silique — Perhaps an average of two seeds per silique. These popu- lations clearly live under very marginal conditions. eavenworthia species are amongst the most hydrophilic of cedar glade plants (Freeman, 1933). The principal factor limiting the distribution and growth of Leavenworthia is the availability of soil water in the spring (Quarterman, 1950; Rollins, 1963); the production of flowers and the 8rowth of seeds are severely curtailed by the drying of the 102 LLOYD glades in the spring as the temperature increases and the rainfall decreases. If the small number of plants and of flowers per plant in the self-compatible populations is due to the populations being on drier sites, it would be expected that these popu- lations would stop flowering earlier than those of self- incompatible races. This was investigated by counting the percentage of plants that were still producing flowers late in the flowering season (April 25, 1962) in a sample from the most favorable portion of each glade. The results (table 10) show that, in general, there is a much higher percentage of plants still flowering in the self-incompatible populations than in the self-compatible populations. In fact, the dis- tinction between self-incompatible and_ self-compatible populations is clearer in these figures than in the figures for the number of plants per population and the estimated average number of flowers per plant. Only one self-compa- tible population (698, race c10) had a higher percentage of plants flowering on April 25, 1962, than any of the self- incompatible populations. The average percentage of plants flowering in the eight undisturbed self-incompatible popu- lations was 50 per cent; the average in the fifteen undis- turbed self-compatible populations was five per cent. The difference is significant (P<.001) by a two-tailed ‘Student’s t’ test. Furthermore, the percentage of plants flowering in each population is significantly correlated (r=.77, p<.001) eh the logarithms of the number of flowers produced in 962. Thus the average number of plants, the estimated average number of flowers per plant, and the percentage of plants flowering in a population late in the flowering season are all correlated, and presumably determined by the same et vironmental factor, the drying of the soils in the spring. It is clear that the self-compatible populations in gene occupy much poorer (drier) glade sites than those of the self-incompatible populations. As a number of the know? and unknown glade sites have been excluded from the statistical comparisons and the comparison was made 1 only one year, there is a possibility that the association between self-compatibility and poor glade sites is fortuitous. Several additional considerations, however, strongly suggest EVOLUTION IN LEAVENWORTHIA 103 100 Zz Oo ° ° Oo Z 80 ry $ > 12) — i N 7 2 60 — 6 - =- i. za — % & 40h ‘ Som Ww g = r=-77 w 20— ° U e ao re) a e °o r J 8 ee 0 PP ge » + a l j 2-0 2:5 3-0 3.5 4:0 45 5-0 LOG. ESTIMATED NO. OF FLOWERS PRODUCED e relationship between the estimated number of flowers produced in centage of plants flowering on April 25, 1962, in the undisturbed : f L. crassa d labamica. Open circles represent self- incompatible populations; closed circles represent self-compatible populations. that the association is a real phenomenon that bears on the evolution of self-compatibility. First, the difference between self-incompatible and self- compatible populations is also apparent in the cultivated fields, where Leavenworthia continues to flower one to two weeks later than in the glades. Cultivated fields differ little in their soil moisture contents, and the later flowering of self-incompatible populations in cultivated fields may be Partly genetically determined. This suggests that the more advanced races and species are adapted to earlier flowering, and presumably that they have lived under poor conditions, Where earlier flowering is forced upon them, for a consider- able time. only self-incompatible species in Tennessee, continues flowering later than the other species in Tennessee. Leaven- 104 LLOYD worthia uniflora, the species most adapted to autogamy, is the first to stop flowering in Tennessee. (This was also noted by Rollins, 1963). The third line of evidence that the association between self-incompatibility and better glade sites has existed for a considerable time lies in a difference in the physiographic positions of the glades of self-incompatible and self-com- patible populations. The glade sites of races al, c2, c3 and c4 are the only known sites for self-incompatible races. The glades containing races al, c2 and c4 in Franklin and Lawrence Counties occur in the Moulton Valley at some distance from the Pottsville escarpment, which marks the boundary between the Moulton Valley and Sand Mt. Plateau. These sites are on undulating to gently rolling ground, and many have permanent streams running through them, The glade sites of race c3 (self-incompatible) in the Massey- Lebanon district do not have streams running through them, but these glades are larger, and perhaps wetter, than those of other races in the Massey-Lebanon district. The glade sites in the Massey-Lebanon district — most of the known glade sites of self-compatible races, as well as those of the self-incompatible race c3 — occur at the base or on the lower slopes of a line of low hills projecting from the Pottsville escarpment (maps 2 to 4). These hills are only 80 to 220 feet high and have few streams running from them. The glade sites are consequently much drier than those occupied by self-incompatible populations. It is probably significant that the two most primitive races of the Massey-Lebanon district, races c3 and c5, are the farthest north in the area (farthest from the Pottsville escarpment) and at the lowest elevation (approximately 610 feet above sea level), while the most advanced race 10 both L. crassa and L. alabamica, races c15 and a4, occurs farthest south (closest to the Pottsville escarpment), and includes the populations at the highest elevation (approx! mately 680 feet above sea level). Three self-compatible races —c6 and the ‘Russellville’ and ‘Tuscumbia’ races of L. alabamica — occur outside the Massey-Lebanon district. All of the known glade sites of the ‘Tuscumbia’ race are on the slopes of limestone hills in situations very similar to those occupied by the self-compat ible races in the Massey-Lebanon district. Both race EVOLUTION IN LEAVENWORTHIA 105 and the Russellville race are known only from single field populations at the base of hills on the edge of the Moulton Valley. Thus the nature of the glades of self-compatible popu- lations is determined by physiographic features. At least for the races in the Massey-Lebanon district, there can be little doubt that these races have existed for some time under conditions similar to those existing today. The large number of L. crassa races in the Massey-Lebanon district can best be interpreted as meaning that the races of the Massey-Lebanon district have actually evolved in the area. There is a primitive, self-incompatible race (c3) of L. crassa in the Massey-Lebanon district, so the loss of self-incom- patibility and concomitant changes in the Massey-Lebanon races of L. crassa have occurred in the Massey-Lebanon district, that is, while the populations were occupying poor glade sites similar to those occupied today. It is possible that L. alabamica was already self-com- patible before it reached the Massey-Lebanon district, since the two races of L. alabamica there are both self-compatible. But race a3 has an advancement index of only 30, so L. alabamica must also have been relatively primitive when it reached the Massey-Lebanon district. So most of the evo- lution of the features associated with self-compatibility in race a4 may also have occurred in the Massey-Lebanon district. Race c6, and the ‘Russellville’ and ‘Tuscumbia’ races of L. alabamica are each geographically isolated from other races of L. crassa and L. alabamica, and it is not known whether they evolved self-compatibility and other advanced characters before or after reaching the area they occupy at present. The arguments presented above strongly suggest that the evolution of self-compatibility and concomitant changes in other characters are associated with the nature of the glades, It appears that on several separate occasions popu- lations of L. crassa and L. alabamica came into hilly areas, where the glades are poorer. As a consequence the popu- lations in these areas have fewer plants, with fewer flowers per plant, and flowering is completed earlier. In the follow- ing sections evidence is presented that the changes in the breeding system have actually taken place in response to a 106 LLOYD relative paucity of insect pollinators in the earlier flowering populations of the poorer glade sites. THE POLLINATION OF LEAVENWORTHIA FLOWERS It has been frequently suggested that self-compatibility has arisen in various plants under conditions where there are relatively few suitable insect pollinators, or when the pollinators are unable to effect sufficient pollination, for example when there is frequent bad weather. Unfortunately there are very few circumstances in nature where this hypothesis can be tested. Most self-compatible taxa have arisen long ago, or have spread from the area in which they originated, so that their present ecological conditions may not closely resemble those under which they arose. And in many self-compatible plants other changes have followed the acquisition of self-compatibility, and have resulted in a lesser attractiveness of the flowers to their natural polli- nators, so that even if there are close self-incompatible relatives still existing, insect visits to the two groups are not strictly comparable. Leavenworthia alabamica, and particularly L. crassa, offer an unusually favorable opportunity for an investigation of the selective factors producing self-compatibility. Self- incompatible and self-compatible races exist in both species, and the pattern of distribution of the races, particularly those of L. crassa in the Massey-Lebanon district, suggests that the self-compatible races have arisen recently under conditions closely approximating those existing in the cedar glades today. Moreover, in both species the flowers of the more primitive self-compatible races differ appreciably from their self-incompatible relatives only in the ease of self-fertilization, and may therefore be assumed to be equally attractive to insects. For these reasons detailed studies of the numbers of the various insect visitors to Lea thia flowers, and their manner of operation, were made during the 1961 and 1962 flowering seasons, To reduce the possibility that any differ- ences in the insects on each population were caused by racial differences in the attractiveness of the flowers, or to ge graphical differences in the pollinators available, the de- tailed comparisons of the populations have been restricted EVOLUTION IN LEAVENWORTHIA 107 to glade populations which belong to large-flowered races of L. crassa in the Massey-Lebanon district, and occupy sites which have not been appreciably disturbed by agri- cultural practices. In addition, supplementary observations were made on glade populations of L. crassa and L. ala- bamica outside the Massey-Lebanon district, on glade popu- lations of L. stylosa in Tennessee (1961 only), and on cultivated populations of L. crassa and L. alabamica. It quickly became apparent that there is a considerable variety of insects visiting Leavenworthia flowers. Collections of insects on both L. crassa and L. stylosa were made on a number of occasions (in 1961 and 1962 on L. crassa, in 1961 only on L. stylosa). A list of the species collected and later identified by specialists in the appropriate groups is given in table 11. Many of the species were collected only once, and undoubtedly additional species which visit Leaven- worthia flowers less frequently would have been added if larger collections had been made. All the bees collected, except the honeybee, are nonsocial and native to the eastern United States. Apart from special attempts to catch the most easily distinguished species, the collections of the native bees were made more or less randomly by my attempting to net the closest bee on each occasion. The numbers of the native bee species caught are therefore a rough indication of their relative abundance. This is not so for the other insects, each of which was caught after a special search. A considerable number of counts of the various insect visitors on glade and field populations of Leavenworthia was made in 1961 and 1962 (table 12). To count the insects, I moved slowly and cautiously over a population, noting only those insects actually on the Leavenworthia flowers, and apparently feeding or collecting pollen. Oc- casional insects merely resting on the flowers (mostly Diptera) were omitted from the counts. Honeybees, native s, and non-Hymenoptera were counted separately. The small black beetle, Meligethes nigrescens Stephens was very common on Leavenworthia flowers, but it never moves between flowers during the day and probably achieves very little pollination, so it was excluded from the counts. In counting the insects on the glade populations (except Population 89) I walked once over the whole population, LLOYD 108 DSSDLI "qoy YPDUMI]VI YUIYDLAD DSSDLa ze : ‘ds ppnwon evpidy DSSDLO 'T Z ‘ssolgQ, vjounluoo visa aVpl[lyow.sa wsophiis “osspsa "'T Cr (Wg) DyDUNnD DI}ALOYI0bNy Dsojhizs “Mssp4a "'T + Keg vund piopyoobny DSSDMI “'T T (‘uIS) snufiydaz "qT DSSDLO "'T T ("qOy) SsnznSuaa *G DSSDLI “'T T (qoy) stunpnbaz “7 DSSDLA "'T G (wg) snnaidsuocour "(7 DSSDLO 'T +t YJ wuhoodyn *q DSSDM2 “'T Z ‘qoy Ssnypmoun “gq msojhiys 7 G (‘ypuRs) snpupdswmDy snjoyniug DSSDLI “'T g Aeg snobby "HH DSSD4a “'T T "ul snsnfuoad snzoyv eepiyoley psojphizs “T 4 amiziz (1B0U) “PY DSSDLI "'T T ‘qoy anjoi “PP DSSDLI “'T 1 (2) ‘qoy vwuvdhiznid ‘py DSSDAI "'T Z "qoy vIDUOSLad “YP DSSDla “'T OL "QOy wpwoasoyjzou 4 VP wsojhys 'T ii nuowyjhiaa (ABaU) “YP DSSDLI "'T e *qoy Uuo0ssawa “VY msojhys *T + “OTA 8200819 °Y DSSDLI *'T E, "QO" Signup DUaupuyp aepluatpuy msojhizs “'T T Aeg snzsapow snanjhyq eeprqza[[OD vtoqydousautrs FT soroeds pe4zoe[[09 soredg pue snuds) ATTUre sy IapsIoO BIYPLOMUBA BAT tequin N SYUAaMOTT VSOTALS "I GNV VSSVUO “I NO G4LOaTIOON SLOUSNI dO LSIT “TT aiavyL 109 EVOLUTION IN LEAVENWORTHIA Dsojhys “J UO peayda[[od sjosuT DSSDAO DSSDAI Dsozfiys “nsspso DSSDAO DSSDAIO psojhijs ‘vsspsa DSSDAI af FHA SANNA ion bo . 98 Onn MAAN OO * | Nim wake wei wo Or GT ‘aessouuay, ‘SeltJUNOD UOS|IA, PUB PAOFABYYNY, Wry 9.18 (‘yde1g) suassaubrw sayzabyapy ‘slog swvqnbn shy y ‘S1Oq srun{ frp sume ‘ds sopuny J ‘qony, snauhub vanazyipy "] snaozyd nuanofv’] Amid sisuauibue vssaun A cauBig addiiw nwmeing ‘slog awmayjhana svyog ‘qe vynuab siwmo0yjup (Avg) snznurmab snsamoxn y (Avg) vinuibsnu nidvsbosa py CPM) Ssnunaiwaumn snydsisnjya yy MOOT SU0L{14VD) "H 2S M snzpwsn{ snpydojayH ‘ds nubysoqu'T (psonporjul) "Ty vuafiyjam sidy ‘US vMoypDJaWMW *D oeprNpiN ‘wuleqe[y ‘SeTJUN0D UBSIO| PUB 9DUDAIMET WOT 218 YSSDLO “TT UO poydoel[od syoosuT Blaydoa[og aepIpmAayyL oupitaodsoyy aeplusevoA'T eepryeydusAN aepliarg aeprydaAg aeprAuioryeayg ‘eaoqydopidery e1aqydiq 110 LLOYD and attempted to glance briefly at every flower. Ideally this method enables every insect working on the Le thia flowers of one glade population at one time to be observed and counted. The numbers observed are probably somewhat below the true figures however, as even a slow movement over the population must distract the insects to some extent, particularly those approached while flying between two flowers. And although all the insects are easily seen on the light-colored background provided by the Leavenworthia flowers, a small percentage of the insects on the flowers was probably overlooked. Glade population 89 (race ¢c3) and populations on cultivated fields were too large to examine completely for insect visitors, and the counts represent the samples of insects seen while I walked over a portion of each population. Most populations on which the insects were counted were observed on several occasions in either 1961 or 1962; only a few populations had insect counts made on them in both 1961 and 1962. Table 12 shows the number and percentages of honeybees, native bees and non-Hymenoptera in the counts on each population of L. crassa and L. stylosa in 1961 and 1962. Considering all of the insect visitor counts on L. crassa and L. stylosa in 1961 and 1962, a total of 2432 insects was observed during 127 separate counts on 23 populations of L. crassa and 9 populations of L. stylosa. Of the insects observed, 1393 (57 per cent) were honeybees, 686 (28 per cent) were native bees, and 353 (15 per cent) were non- Hymenoptera. The relative proportions of honeybees and native bees were similar in the 1961 and 1962 counts, in- cluding those on populations 37, 88, 89, 100, and 171 on which insect counts were made in both years. Non-Hymenoptera constituted less than a quarter of the insects counted on most populations in both 1961 and 1962, but made up as much as 37 per cent of the insect visitors (population 37 in 1961). Table 12 shows that the percentage of non-Hymenoptera wag generally higher in the 1961 counts on L. crassa populations than in the 1962 counts. This is principally due to the fact that there were many more individuals of the Helophilus species observed in 1961 than in 1962 (Lloyd, 1964). There were also differences between geographical areas in the species of non-Hymenop- EVOLUTION IN LEAVENWORTHIA 111 tera observed. For example, Bombylius major and the Helophilus species were common on L. crassa in ‘Morgan and Lawrence counties, Alabama, in 1961 and 1962, but were not found on L. stylosa in Tennessee. Although there are appreciable numbers of honeybees, native bees, and non-Hymenoptera on both glade and culti- vated populations, it is apparent that their relative pro- portions are very different in glade and field populations. In all 12 field populations of L. crassa which were counted in 1961 and/or 1962 there were more than three times as many honeybees as native bees. But in the 12 glade populations of L. crassa (excluding population 89) and L. stylosa the situation is completely reversed; there were more than twice as many native bees as honeybees in every case, and the honeybees were outnumbered by as much as 23:1 (popu- lation 38). There is no doubt that the difference in pro- portions of honeybees and native bees between undisturbed glade and field populations is quite general, and is not based on one or two anomalous populations. The pollinators on population 89, which has been greatly disturbed by the removal of forest around its margin, are more similar to those on the field populations than those on other glades. There are, I believe, two possible explanations for the higher proportion of honeybees en field population. Glade and field populations differ in their size, and in the nature of their soils and surrounding vegetation. It is well known that honeybees are attracted to large masses of flowers, and that foragers returning to the hive may transmit information about good foraging sites to other bees, thus stimulating an increased activity on a favored population. The field populations, and population 89, are much larger than the glade populations, and may attract relatively more honeybees for this reason. This alone cannot explain the higher percentage of honey- bees on field populations, however; if the population size is the controlling factor it is necessary to postulate that the Native bees are less influenced by the size of a population than honeybees are. There is no evidence that there is any relationship between the size of a population and the relative numbers of honeybees and native bees, either in the glade or. in the field populations. Amongst the field populations there ° are most native bees (relative to the number of honeybees) LLOYD 112 syoesur jo ‘ou [BIO], viaydousuls -uou jo aSe4u9010T -ejndod 6 sé O°8L 902 L°8 beard Tr PC-8Z peqinjstp eutos ‘sepyys 10} [BIO} ‘aassouua, ‘UISBG a[[[AYSBN ‘SeljUNOD [BA1aA95 psojplizs a | g ST g L’éh 8% T 8T PIey Uso. 893 go g £°82 L 0°09 8T I L Pley us09 oF ya 6g o'st ST 23s £9 v4 TT-L spUBUUAT JsatOF YPM Ploy TLT ic 8 8's § 0°88 18 a It-h PIey Uso. OLT pP “Burege|y ‘sarjunoo aUaIM BT pues uBs10W ‘spLIjsIp Tayo 0 8°S T 2°16 cs T 8T Pley peysnojdun 00T go 8 Be a 9 3°38 Lg z 8I-¢ Pley uso. 69 go ai L’9T 8 r°09 6G & 8I-s P[Py U0. sg go 68 PST &% L°99 Pel L 8T-§ Pley Ust0o 88 go TOL DLE 0s 0°Ss GPT or 8I-s Ppley us00 1g go 68 g'9T 63 19 8oT L 8T-s ape]s peqinjsip 68 g9 :e0Ry ‘euEqLTy “OO UBsIOPY *7011}sSIP UOUUqoe']-AossBy DSsDLo “T SINQOO 1961 js0) 4 ~ y i ce Soni B 4 & So ms mo o p ° at oO r=] ~ pag fe GE aE fb ar Bag 2 of ORS 5B & 3 ag >a a 5 2 =o 36 ee pip gh G2 Feo Pees Bee Fog oft 3 ce Fh = ° S i=] 296 GNV T96I NI VSOTALS “I GNV VSSVUO “140 SNOILLVINdOd NO SINN00 LOaSNI “él Gav 113 aA oO FF DO HO ri EVOLUTION IN LEAVENWORTHIA oot MOH OO - Tha L 6°SL 6% T ra SJUBUUS 4SaIOJZ YA Poy TLI Tj? 2°06 9P 6°e 3 T VG opes 88 po “BUILGBLY ‘Se1zUNOD sdUeIMETT pu UBsIOW ‘s}L148Ip 19410 L’8 ca 0°L8 yas v 13-9 PIPy ust0d 63 go L’9 a 8°88 96 z 6 “9 PlP¥ peysnojdun QOT go PLT v £°8L 8T T VI PIPY pezeis 8hL §9 L’LT LT 66h 9L & 6I-FT PIPy uUt0o 88 go 8°61 6P 8°ch 88T 5 Te-9 PIey u.to0o LE go = 0 = 0 Pp 13-6 apes B8IT go O°SL & 0°0 0 9 13-6 apes PEPL £9 8°88 2 0°0 0 Pb Te-FT opes T6L gy 0°88 xa Li § g 9 12-6 apes 98 go EL a ie) T § 13-6 apes orh TT Po 6 9°86 v S 13-6 opels OSPL go 0°8L c& 9FT 9 8 13-6 opyls VEPL go 6°89 TS aed Or 9 16-6 apels peqinqsip 98PL go S°09 6P B'8s £3 9 16-6 epels q8Phl £9 L’st rae QoL L0G 6 Té-9 apes paqanqsip 68 g9 taony ‘Bureqe[y ‘Ajunoo uBsIOW 4011ISIp uourqgoy-Aasseyy DESDLI "7 SINNOO 72961 114 LLOYD on population 46, yet this is one of the largest populations on which the insects were counted. Also, populations 59, 69, and 100 had relatively fewer native bees than other field populations, but populations 69 and 100 were rather small field populations, while population 59 was a large population. It seems much more likely that the difference in the visitors to the glade and field populations is due to the position or nature, rather than the size, of the sites the populations occupy. The fact that the disturbed glade population 89 has proportions of honey and native bees closely resembling those on the field populations suggests that it is the surroundings of a population that determine the relative numbers of honeybees and native bees. The glade site of population 89 is quite similar to that of the other glade populations except for having most of the forest removed from its margins, and is totally different from the fields occupied by Leavenworthia. Most of the native bees collected are ground-nesting species, which cannot complete their life cycle in nests in ploughed land (Malyshev, 1936). On the other hand the glades themselves and the forest which surrounds them provide suitable habitats for the nests of both ground- nesting and wood-nesting bees. The cedar glade popu- lations of Leavenworthia are closer to the nesting sites of the native bees than are the field populations. It is known that bumblebees and at least some solitary bees prefer to forage close to their nests (Free and Butler, 1959; Linsley, 1958). Braun et al. (1956) have shown that bumblebees are more or less evenly dispersed over small clover fields, but in large fields the number of bees decreases from the edges of the field. If the native bees visiting Leavenworth also prefer to forage close to their nests, this would explain their lower percentages on cultivated populations. The honeybees would not be similarly affected since they have to fly some distance to both the glade and field populations. Population 89 (on a disturbed glade) and populations 37 and 88 (in corn fields) are less than 50 yards from the margin of the forest containing glade populations 743 and 791. All of these populations are probably visited by honey- bees from the same hives and yet there are many fewer native bees (relative to the number of honeybees) in the disturbed glade and corn field populations. Apparently the EVOLUTION IN LEAVENWORTHIA 115 native bees do not often fly even this short distance onto the cultivated fields. The differences between the various field populations in the proportions of honeybees and native bees appear to agree with the hypothesis that the distance from the nesting sites is largely controlling the number of native bee visitors to a Leavenworthia population. Amongst the field populations the number of native bees most nearly approaches the number of honeybees in populations 46 and 171: population 46 is adjacent to a stream with waste land along the margins, and population 171 is in a field which had not been completely cleared of forest. And populations 59, 171, and 100, which have the lowest percentages of native bees, are the farthest from any forest. It is obvious from the above discussions that the intro- duction of the honeybee and the spread of Leavenworthia crassa onto cultivated land have had considerable effects on the abundance of the insect visitors to L. crassa populations. With regard to the introduction of the honeybee into North America, Pearson (1933) states that “this introduced Species appears to exert an influence upon native bees which is visible to any student of the native species. So efficient is it as a collector of pollen and honey and so ubiquitous has it become, there can be no question but what its inroads cause a serious diminution in the food supply of native bees, Particularly in bad seasons.” Fortunately, this effect has en much greater in the field populations of Le thia than in the glade populations. It may be that the develop- ment of extensive field populations of L. crassa close to the glade populations has minimized the effect of the honeybee in the glades by offering alternative, much larger, popu- lations for the honeybees to use. All of the glade populations of L. crassa (except population 89) on which insects were counted have been little disturbed by human activities, and it seems likely that apart from the introduction of the honeybee, the plant-pollinator relationships in these ‘undis- turbed’ glades of L. crassa are similar to those under which the races differentiated. THE RELATIVE EFFECTIVENESS OF DIFFERENT INSECTS AS POLLINATORS 3 The figures given above suggest that although native bees 116 LLOYD are the predominant natural insect visitors, the non-Hy- menoptera play an appreciable role in the pollination of thia flowers. Considerations of the structure of L thia flowers, the mode of action of the Hymenop- tera and non-Hymenoptera visiting the flowers, and a comparison of the rapidity of the movements of Hymenop- tera and non-Hymenoptera indicate, however, that the non- Hymenoptera are much less important than their frequency would suggest. The flowers of L worthia are erect; the petal claws are also erect and appressed to the circle of filaments which closely surround the pistil (fig. 3). The petal limbs bend at right angles to the claws in fine weather and thus form a horizontal platform. The stigma projects approximately 1 mm above the outspread petal limbs, and is closely sur- rounded by, and sometimes completely concealed within, the anthers of the four paired stamens. In the more primitive races, including all of those on which the pollinators were studied, the anthers of the paired stamens face obliquely or directly away from the stigma and the pollen is freely exposed to insect visitors. The single stamens are more Or less strongly curved in the open flower and their anthers lie slightly below the level of the petal limbs. Consequently there is a narrow channel on either side of the flower, enclosed by a single anther, the adjacent petal claw margins and the filaments of the paired stamens. These spaces lead to the nectaries at the base of the anthers, and may there- fore be called nectar channels. Their length and width vary somewhat between the races, and with the vigor of the plant, but they are approximately 5 mm deep, and less than 1 mm in diameter. Leavenworthia flowers thus have completely concealed nectar and exposed anthers (those of the paired stamens surrounding the stigma in a cone in the center of, and slightly above, the outspread petal limbs. According t? Knuth (1906, page 20), H. Miiller claimed that “it may be ... correct that in all... flowers in which the anthers are orne upon short, stiff filaments, and enclose the conical style, bees are the necessary agents of cross-pollination. This arrangement is rather uncommon in the Crucif erde, most of which have only partly concealed nectar, but it 1S common within the Compositae. Miiller (1883, page 581) EVOLUTION IN LEAVENWORTHIA 117 summarizes the insects he noted on 25 Compositae species; the insect groups and their relative frequencies on these species are very similar to those observed on Leavenworthia. The majority of the native bees land on or climb onto the central cone containing the stigma and the anthers of the paired stamens and crawl rapidly over or “wallow” (Maly- shev, 1936) on the cone. Their actions are usually very vigorous and there is little doubt that they often succeed in separating the anthers and touching the stigmas, and at the same time in picking up abundant pollen on their legs and undersides. Most of the native bees appeared to collect both pollen and nectar, although some of the smaller bees made no attempt to insert their proboscises into a nectar channel, possibly because the nectar was too deeply concealed. The action of the honeybees in visiting the flowers of Leavenworthia is very different from that of the native bees. A honeybee lands on one of the petal limbs, with its head towards the center of the flower. Its weight causes the pedicel of the flower to bend over and the honeybee grasps the petals with its legs while it quickly inserts its head into the top of one or both of the nectar channels. In doing so the petals and the single anther surrounding the nectar channel are forced apart, and the bee’s head is brought against the paired stamens and the pistil. The forceful actions of a honeybee presumably result in a considerable amount of cross-pollination. All of the honeybees visiting Leavenworthia flowers appeared to be seeking nectar, for they invariably pushed their heads into one or both of the nectar channels. Many of the honeybees seemed to be collecting pollen, too, for the pollen baskets of most honeybees foraging on Leavenworthia contain pollen. Samples of the pollen masses from two honeybees collected on L. crassa were examined and proved to contain cruciferous pollen, presumably that of Leaven- worthia. The actions of the non-Hymenopteran species are such that they rarely or never touch the stigmas. Bombylius major and the Lepidoptera species stand on the petal limbs with their heads above the nectar-channels but they are able to reach the nectar without putting their heads close to the entrance of the nectar-channels, and thus they do not usually come into contact with the anthers of the paired stamens. 118 LLOYD If they do touch the paired stamens they probably rarely do so with sufficient force to separate them and touch the stigma. Miiller (1883, page 580) notes that butterflies are often quite useless as pollinating agents in their Visits to bee flowers, and this seems to be the case in L I e remaining species of non-Hymenoptera (chiefly Diptera and Coleoptera) feed on the exposed pollen of Leavenworthia flowers. The species of Syrphidae are the principal pollen feeding insects. The syrphids stand on the petal limbs and take pollen from the anthers of the paired stamens. They may cause some pollen to be scattered onto the stigmas of the same flower, but as Robertson (1924) notes, they are light and rest so superficially on the flowers that they can eat the pollen without carrying it on their bodies, and are therefore useless as pollinators of flowers with exposed pollen. The importance of the Hymenoptera in the pollination of Leavenworthia is even clearer when the time the various insects spend on a single flower is considered. Individuals of all the major groups of insect visitors were followed for some time, while the number of flowers and the time they took to visit them were noted (table 13). In both 1961 and 1962 several individuals of Apis melli- fera spent an average of almost exactly 4 seconds on each TABLE 13, AVERAGE TIME INSECTS SPENT ON L. CRASSA FLOWERS Av. time No. of No. - — on si insects flow Order Species Year tinea visited (eesman pest Hymenoptera Apis mellifera 1961 5 260 1060 4,08 1962 7 383 1510 3.94 Halictus ligatus* 1962 if 65 585 8.23 Andrena cressonii.. 1962 1 89 630 7.08 Andrena?nothoscordi! 1962 1 40 430 10.75 Ceratina metallica 1962 1 12 135 11.25 Species unknown 1962 1 18 105 5.83 Species unknown 1962 1 26 150 5.77 Species unknown 1962 1 4 430 10.75 Species unknown 1962 1 129 1060 8.22 Diptera Bombylius major 1961 2 52 10 _ 1962 1 58 310 5.85 Lepidoptera Anthocaris genutia 1962 3 214 1615 7.55 Thyris lugubris 1962 1 13 1 14.62 Caenurgia ?erechtea 1962 4 16 915 57.19 "Individual collected after it was timed. The = ies of Syrphidae (Di iptera) im Nitidulidae ee eoptera) were not timed. The syrphids spent up to se veral pee on each flower, and the nitidulids were never seen moving from one honey to ano during the asi EVOLUTION IN LEAVENWORTHIA 119 flower. Individuals of several species of native bees spent between 5.77 and 11.25 seconds on each flower. It appears that there are differences between the native bee species in their speed of operation, but that all spend somewhat more time on each flower than do honeybees. No accurate timings of the syrphid species were taken; they nearly all spend a long time on each flower (often several minutes) and they rest for considerable periods either on the ground or on a flower after relatively few flower visits. Bombylius major, on the other hand, moves almost as quickly from flower to flower as honeybees do. The Lepidoptera species vary tremendously in the rapidity of their movements from flower to flower, but on the whole they are comparatively slow workers. When the structure of Lea thia flowers is con- Sidered together with the mode and speed of action of the various insect visitors there is no doubt that the bees are the only significant pollinators of L. crassa and L, stylosa flowers. They probably achieve 99 per cent or more of the cross-pollinations and a considerable proportion of the self- pollinations of L. crassa and L. stylosa flowers. THE EFFECT OF POPULATION SIZE ON THE NUMBER OF POLLINATORS It was shown above that glade populations of L. crassa and L. alabamica differ considerably in their size, and that in general the self-compatible races have relatively small glade populations which cease flowering earlier than the larger glade populations of self-incompatible races. Further- more, self-compatibility has apparently arisen under glade conditions similar to those of the self-compatible races today. It is possible therefore that self-compatibility may have arisen in response to a paucity of insect visitors to small, early-flowering glade populations. 0 examine the effect of population size on the number of Pollinators it is necessary to have estimates of both the number of pollinators foraging on a series of populations and the number of flowers in those populations. The number of pollinators divided by the number of flowers (i.e. the fraction of flowers with pollinators on them at any one time) gives an estimate of the relative abundance of polli- nators, 120 LLOYD The counts on the glade populations of L. crassa (except population 89) were made by observing the insects on all the flowers of a population. Since the populations were small and could be examined sufficiently quickly that there was little possibility of counting an insect more than once in one count, the number of insects counted may therefore be taken as the number on a population at any one time. The insect counts on the undisturbed glade populations of L. crassa were made between the 9th and the 21st of April 1962. On April 16, 1962, the number of flowers in each glade was counted (table 14). These numbers may be taken as estimates of the average number of flowers in the glades throughout the period when the insects were counted. The non-Hymenoptera are not important pollinators of Leavenworthia flowers, so these have been excluded from the calculations of the relative numbers of pollinators in each population. The average number of honeybees seen on one count of each glade population was then divided by the number of flowers in the same population on April 16, 1962, to give an estimate of the fraction of flowers with honeybees on them at any one time when bees were active throughout the period of the counts (table 14). Similarly the average number of native bees per count divided by the number of flowers gives an estimate of the fraction of flowers in each population with native bees on them at any one time. It should be emphasized that the counts were made only on fine days between 10 AM and 3 PM, when the bees were fully active. The results show that there were between 0 and 0.22 per cent of the flowers with honeybees on them at one time In the various glade populations, and between 0 and 1.62 per cent of the flowers with native bees on them. That is, only a very small fraction of the flowers have bees on them at any one time. The calculations of the speed of movement of honey and native bees from flower to flower may be combined with the fraction of flowers with bees on them at one time to give am estimate of the number of bees which, on the average, visit each flower in a given time. Then if the time each flower 1S open while the bees are foraging is known one can obtain finally an estimate of the average number of bees which visit a flower in the various glade populations. EVOLUTION IN LEAVENWORTHIA 121 For this purpose, it was considered that a honeybee takes 4 seconds to go from flower to flower. The native bees vary considerably in the rapidity with which they move from flower to flower (table 12). For the purposes of the subse- quent calculations, it was considered that a figure of 8 seconds per flower would be a reasonable estimate for the native bees as a whole. The majority of the flowers remain open for only one fine day. If a flower is not pollinated on the first day it may, however, remain open for a second, or even a third day. Moreover, if a flower becomes ready to open on a cold, dull or rainy day it may remain closed for one or two days until the weather improves sufficiently for it to open. The weather conditions necessary for bee flight and for Leavenworthia flowers to open are quite similar; in general if the flowers open, the bees are more or less active, and vice versa. On a fine, warm day in the middle of April the bees become active at approximately 9 AM, and forage until about 3 PM. Although there is considerable variation from day to day in flower and insect periodicity it seems reasona- ble to say that on the average a single Le thia flower is open for approximately 6 hours when the bees are active. If a honeybee were to forage continuously for 6 hours it would visit x 60 X 6 = 5400 flowers; a native bee would visit xX 60 X 6 = 2700 flowers. Of course one bee does not forage continuously for 6 hours, but the average number of visits to a single flower is equal to the fraction of flowers with bees on them at any one time multiplied by the number of flowers that a single bee would visit if it were active throughout the time that a flower is open. The estimated number of honeybees and native bees that on the average visited a flower in each of the glade popu- lations studied is given in table 14, together with the total (honeybee plus native bee) visits per flower. There is Considerable variation between populations in the estimates for both the number of honeybees and the number of native bees which visit each flower. The estimated number of honeybees visiting each flower ranges from 0 for popu- lations 118a, 791 and 743, to 11.9 for population 86. The 122 LLOYD TABLE 14, ESTIMATED NUMBER OF BEES VISITING EACH FLOWER IN GLADE POPULATIONS OF L. CRASSA IN 1962 Av. no. of bees Fraction of flowers Estimated no. of bees in each count? with bees on visiting each flower® n hy a a a | ane n 8 cm = 8 2 tee 2 2. 20 woes EE 2 3s $e at 2 3S Ss _ a c irs =| b> - © em me s 3 Sy ie oS Bees 5 3 32 63 65 & . Pe Meee Se ba 48 BB SY a cee Lies ae: Es = a A E, Fy c3 743b 3500S 3.83 8.17 1,09 10-3 2.88 10-3 5.9 6.3 12.2 3 748c 8=0«16 1.67 5.20 1.0410-3 2 033 5.6 8 14.3 ell 742 251 0.20 2.20 0.80x10-3 8.77102 -— 4.3'- O87 eae 5 86 226 0.50 3.67 2.21K10- 16.28X10-8 11.9 44.8 56.2 5 791 215 0.00 1.20 00 0.56 X 10-3 0.0 1.5 1.4 3 7434 185 0.00 0.50 0.00 0.27 10-3 0.0 Onf Met C5 118 =184. 0.00 =—0.00~—s*0«.00 0.00 0.0 0.0 = 8©=— 0.0 4 gures are taken the average number of flowers present during the These fi a period when the insects were counted (April 9 to April 21, 1962). *The raw data for these figures is given in table 12. These figures are taken as the average number of bees on each population at any one moment. ’ Each flower is assumel to be open for one day, during which the bees are foraging for 6 hours, estimated number of native bees visiting each flower ranges from 0 for population 118a to 44.3 for population 86, The estimated total number of bees visiting each flower ranges from 0 for population 118a to 56.2 for population 86. At this point I must emphasize that these figures are based on four parameters — the number of flowers in 4 population, the number of bees foraging on a population at One time, the speed of movement of the bees from flower to flower, and the time that each flower is available for insect Visits. All of the four component figures are approximate only, so there may be considerable error in any one of the final figures. In general, however, the errors are likely to be in the same direction for all populations. Even if these figures are incorrect by a factor of two or ree there can be no doubt that each flower is visited by only a small number of bees. And these calculations refer only to those flowers which are open on fine warm days. the early part of the flowering season particularly there may be several successive days during which there are no suitable foraging conditions, or only short periods when the bees are active. Flowers which are ready to open at such times may never open fully and receive any insect visits. That this 18 EVOLUTION IN LEAVENWORTHIA 123 so is demonstrated by the fact that a small proportion (up to 16 per cent) of the fruits collected from samples of the glade plants are aborted, i.e. were never pollinated. And this is in spite of the fact that a considerable number of flowers must be self-pollinated without the aid of insects. The glade populations on which the numbers of bees visiting each flower were estimated are arranged in Table 14 in order of the number of flowers present on April 16, 1962. The number of flowers ranged from approximately 3500 in population 743b to 134 in population 118a. It is apparent that there is no consistent trend in the estimated number of honeybees and native bees visiting each flower from the population with the largest number of flowers to the population with the smallest number of flowers, although the three smallest populations received many fewer bee visits to each flower than the other populations. The high figure for population 86 is probably due to the fact that this population experienced a greater decline in the number of flowers during the period of the insect counts than did the other populations. Since the number of flowers was counted on April 16, in the latter half of the period of the insect counts, the average number of flowers throughout this period was probably greater than the number observed on April 16. Consequently the average number of bees Visiting each flower of population 86 has been overestimated. There is a similar source of error in the estimations for the other populations, but the number of flowers in them de- clined much more gradually than in population 86, and the error is accordingly much smaller. The large differences between the three populations with the fewest flowers (791, 743d, and 118a) and the remaining glade populations in the number of honey and native bees visiting each flower probably represent a true difference between the larger and smaller glades. I noted on several Occasions other than those of insect counts that there were no bees on the smaller glade populations. The three smallest Populations, however, differ little in size from populations 742 and 86, which received many more bee visits. Perhaps smaller populations do not necessarily attract relatively ewer bees, but for some reason are less likely to be visited frequently; their smaller size or shallower soils in the 124 LLOYD surrounding forest may decrease the chance that there are suitable nesting sites in their immediate vicinity. POLLINATOR CHANGES DURING A FLOWERING SEASON The counts of the number of bees visiting the Leaven- worthia populations were also examined to see if there were any significant changes in the course of a flowering season. The number of bees visiting L thia flowers in the early and late parts of the flowering season would ideally be compared by studying the number of bees visiting each flower in the early and late flowering periods of the glade populations. Unfortunately my data are not sufficiently extensive for a direct comparison to be made in this way — the number of flowers in each undisturbed glade population was counted only once, and the insect counts were not begun until the peak of the flowering period. It was noticed, however, that the ratio of native bees to honeybees appeared to increase throughout the flowering season, on both glade and field populations of L. crassa in 1961 and 1962. This observation was examined more closely by dividing the 1961 and 1962 insect counts on L. crassa populations into those made in the first half of the period of the insect counts in each year, and those made in the second half of this period. In 1961 the insect counts on L. crassa populations were made between April 3, when flowering was at its peak, and April 18 when flowering had almost ceased. Counts made from April 3 to April 8 are considered to have been made in the mid-flowering period, and those made between April 11 and April 18 are considered to have been made in the es period. No counts were made on April 9 and 0 The 1962 counts were made between April 6 and April 21. Those made between April 6 and April 10 are com sidered to have been made in the mid-flowering period, though the number of flowers was beginning to decline * this time; those made between April 14 and April 21 are considered to have been made in the late-flowering perlo™: The number of native bees divided by the number of honeybees in each population, in the mid- and late-flowering periods, is shown in table 15 for those populations on whic EVOLUTION IN LEAVENWORTHIA 125 TABLE 15. COMPARISON OF BEES VISITING L. CRASSA POPULATIONS DURING MID-FLOWERING AND LATE-FLOWERING PERIODS IN 1961 AND 1962 Total Total a cv : native bees Popula- Flowering : f no. of Ri awe ones : 3 honey- native No. of eriod counts bees Race tion bees honeybees 1961, glade population c3 89a middle 3 "7 7 0.09 late 3 31 22 0.71°* 1961, field populations c3 37 middle 8 87 12 0.14 late 4 55 18 0.33° c3 88 middle 3 71 5 0 late 4 53 18 0.34°* ce 58 middle 1 14 2 0.14 late 2 15 6 0.40 5 69 middle 1 15 1 0.07 late 1 22 4 0.18 1962, glade populations e3 89 middle 5 157 18 0.12 4 50 34 0.68** 3 743b middle 1 6 12 2.00 5 17 37 2.12 3 743¢ middle 2 8 7 0.8 ate 4 2 24 12.00°* 3 748a middle 2 8 4 1.33 late 6 3 28 9.33 ¢3, ¢5, 6 glades' middle 12 7 28 4,00 cg late 18 1 22 22.00 1962, field populations c3 37 middle 3 166 39 0.24 late 2 22 10 0.56 ce 58 middle 2 88 0 0.00 late I nd 12 0.38** ‘The data from populations 743e, 742, 86, 791, 743d and 118a was grouped to obtain a sufficient number of observations for comparison of the two periods. *The ratio of native bees to honeybees was significantly higher (P = .05 —.01) in the late-flowering period than in the mid-flowering period. “The ratio of native bees to honeybees was significantly higher (P<.01) in the late-flowering period than in the mid-flowering Period. a sufficient number of insects were counted in 1961 or 1962 for a comparison of the two periods to be made. In addition the 1962 counts on six smaller glade populations on which 126 LLOYD only a few bees were observed have been pooled to obtain sufficient numbers for the comparison. In all twelve possible comparisons the ratio of the number of native bees to the number of honeybees is greater in the late-flowering period than in the mid-flowering period. The difference is signifi- cant (P<.05) for six populations; the lack of significant differences in the other populations can be attributed principally to the small number of observations. Considering the totals from the six populations observed in both periods in 1961, in the mid-flowering period the ratio of native bees to honeybees was 27 :264 or 0.10:1, and in the late-flowering period the ratio was 68:176 or 0.39:1. Thus compared with the number of honeybees, there were 0.39/0.10 = 3.78 times as many native bees in the late- flowering period as in the mid-flowering period. The mid- points of the mid- and late-flowering periods were only 9 days apart, so a large change in the proportions of native bees and honeybees occurred in a very short time. Considering the totals from the eleven populations ob- served in both periods in 1962, the ratio of native bees to honeybees was 108 :435 or 0.25:1 in the mid-flowering period and 167 :127 or 1.31:1 in the late-flowering period. Compared with the number of honeybees there were 1.31/0.25 = 5.30 times as many native bees in the late-flowering period as In the mid-flowering period. The mid-points of the mid- and late-flowering periods were 814 days apart, so in 1962 the . ratio of native bees to honeybees showed more than a five- fold increase in little more than a week. : There are two possible explanations for the relative increase in the number of native bees in the late-flowerms period; it might be due to the honeybees leaving Leaven- worthia in favor of other plants as the number of Leaven- worthia flowers decreases, or it might be due to an increase in the number of native bees. The change in the ratio of native bees to honeybees ous be explained as being due to the transfer of honeybees t0 other plants only if the percentage of honeybees working on Leavenworthia decreases four or five times more rapidly than the percentage of native bees working on Leaven- worthia. This is possible, but unlikely. The populations studied in both periods include glade populations and culti- vated populations. The largest field populations contained EVOLUTION IN LEAVENWORTHIA 127 more than a thousand times the number of flowers in the smaller glade populations, and yet the relative increase in the number of native bees was similar in all populations. If the increase was due to the honeybees alone transferring to another crop one might expect the effect to be more pro- nounced on the smaller populations. It seems much more likely that the change in the ratio of native bees to honeybees is due to an increase in the number of native bees foraging in the area. The available infor- mation on the seasonal activities of honeybees and native bees appears to support this conclusion. Honeybees, unlike all other temperate Apoidea, do not hibernate during the winter, but remain intermittently active in the hives. Honeybees are foraging on Leaven- worthia in suitable weather from the beginning of the flowering period (approximately March 15 in 1961 and March 20 in 1962 for L. crassa in the Massey-Lebanon district), but it is probable that the number of honeybees foraging increases somewhat as the Le thia flower- ing season advances. The native bees, on the other hand, hibernate during the winter, and emerge at some time during the warmer months of the year —the species differ widely in their dates of emergence and flight periods. There has been no detailed study of the flight periods of native bees in Alabama, but available information from studies elsewhere in the eastern United States (Michener and Rettenmeyer, 1956; Mitchell, 1960, 1962; Robertson, 1929) indicates that the number of bee species foraging increases greatly throughout the flower- ing period of Leavenworthia. All of the native bees collected have a wider distribution than the genus Leavenworthia, and must forage on other plants outside its range. Un- doubtedly, these bee species forage on a variety of plants through a longer season than the relatively short flowering period of Leavenworthia. The relative times of the flight periods of the native bees and the flowering of L thia suggest that the in- crease in the ratio of native bees to honeybees in the latter part of the flowering period of Leavenworthia is due to an increase in the number of native bee species foraging as the spring progresses, although a proportionately greater trans- fer of honeybees to other plant species cannot be ruled out as 128 LLOYD a contributing factor. It is unfortunate that I have no polli- nator counts taken during the first half of the flowering season when the number of flowers is increasing. A similar increase in the ratio of native bees to honeybees during that time could only be interpreted as due to an increase in the number of native bees. In addition the temperatures are lower earlier in the spring, and the daily flight periods are probably shorter at this time. Michener and Rettenmeyer (1956) observed that the daily period of flight of Andrena erythronii (possibly one of the species found on Le thia —table 11) increases as the season becomes progressively warmer. Also, the average rainfall in the Tennessee Valley decreases as the spring progresses, so it seems likely that the percentage of days on which any flights are possible increases as the spring progresses. The combination of fewer native bees, fewer days suitable for flight, and shorter flight periods on suitable days earlier in the spring leaves no doubt that the earlier flowers are visited by fewer native bees than the later flowers. Consequently the flowers of those populations which are forced by the nature of the glade sites to cease flowering earlier in the season are on the average visited by fewer native bees than the flowers of populations on more favorable sites. Because of the association between self- compatibility and poor glade sites noted previously, this means that in general the flowers of self-compatible popu- lations receive fewer native bees than the flowers of self- incompatible populations. FACTORS AFFECTING THE EVOLUTION OF SELF-COMPATIBILITY IN L. CRASSA AND L. ALABAMICA The nature of the selective forces responsible for the evolution of self-compatibility has been widely discussed 1n the century since the demonstration by Charles Darwin tha cross-fertilization is generally advantageous to plants, and that many plant species have structural features whi prevent or restrict self-fertilization. The earlier workers, including Darwin and Hermann Miiller, generally attribut the evolution of self-compatibility to the need for an ade- quate seed set under conditions where cross-pollination W4* insufficient for this purpose. Since the rise of ™ exe EVOLUTION IN LEAVENWORTHIA 129 evolutionary theory many workers have replaced or supple- mented this explanation with arguments based on differ- ences in the genetical structures of cross-fertilized and self-fertilized species. Genetical explanations for the evolution of self-compati- bility are based on the theory of breeding systems developed principally by Darlington (e.g. 1939) and Mather (e.g. 1943a). These authors pointed out that the genetical structure of a population represents a compromise between the extent to which a population is adapted to its current environment, expressed as its immediate fitness, and its capacity to adapt to changes in the environment by pro- ducing new gene combinations, expressed as the flexibility of a population. Self-fertilization increases the degree of homozygosity in a population and therefore increases the number of genes which are expressed in a population; Mather consequently argued that a self-fertilizing plant can become more ‘closely adapted’ to its immediate environment. Self-fertilization also leads to a decrease in the genetic variability and therefore to a loss of flexibility. Consequently these authors believe that self-fertilizing plants achieve an increase in immediate fitness at the expense of a decreased flexibility. Stebbins (1950, 1957) and others have extended this line of reasoning to suggest that self-compatibility may arise in species living in temporary, unstable habitats in response to the need for many well-adapted seeds under these conditions. In plants living under such conditions “a high premium Is placed on a genetic system which favors opportunism and enables a favorable gene combination, once it has been achieved, to spread over a large number of genetically Similar individuals” (Stebbins, 1950). We may now examine the studies on the glade populations of L. crassa and L. alabamica and their pollinators to see which of the alternative hypotheses best explains the results. Because of their physiographic position — at the base or on the lower slopes of limestone hills — the self-compatible glade populations of L. crassa and L. alabamica occupy drier glade sites than those occupied by most of the self- incompatible populations on the floor of the Moulton Valley. Asa result, the self-compatible glade populations studied in 1962 were smaller and ceased flowering earlier in the spring 130 LLOYD than the self-incompatible glade populations. Moreover, the loss of self-incompatibility in several phyletic lines in L. crassa and L. alabamica probably occurred while the ancestors of the present self-compatible populations occu- pied glades similar to those of the self-compatible races today. Estimates of the average number of bees visiting each tlower in glade populations of various sizes in 1962 showed that the three smallest glade populations had many fewer bee visits than the six larger glade populations. The three smallest glade populations differed little in size, however, from two others which received many more bee visits to each flower. It may be that there is less chance of suitable nesting sites for bees existing near the smaller glades, rather than that the bees are less attracted to the smaller populations. But in either case the smaller size of the populations ancestral to the modern self-compatible races may have led to a diminished seed set through a lack of sufficient insect visitors. The effect of the earlier flowering of glade populations in poor sites is clearer. Our knowledge of the seasonal foraging activity of the natural bee pollinators of Leaven- worthia, and the relative increase in their numbers (com pared with honey bees) in the second half of the flowering period of Leavenworthia, offers strong evidence that the earlier flowering populations receive relatively few bees to each flower; and even in the second half of the flowering period, when the bees are most numerous, each flower receives only a small number of visits (between 10 and 20 honeybee and native bee visits combined in most popu lations). An examination of the samples of plants from each of the undisturbed glade populations showed that up to 16 per cent of the flowers were not fertilized; and some of the flowers were undoubtedly self-pollinated without the aid of insects, even in the self-incompatible populations. This shows that there are indeed too few bees on those POP lations, or insufficiently frequent periods suitable for bees to forage, for all flowers to be pollinated by their activities. It seems likely therefore that the number of bee visits 1S inadequate to secure a full seed-set on the flowers produced earlier in the flowering season. According to the above arguments, the loss of self-in- EVOLUTION IN LEAVENWORTHIA 181 compatibility in L. crassa and L. alabamica took place in small, earlier-flowering populations, the flowers of which received relatively few bee visits. The evolution of self- compatibility and of other characters associated with it increases the ‘spontaneous’ seed-set in the absence of insects, and may well have occurred in response to selection pressures favoring a fuller seed set. The results obtained in Leavenworthia thus agree well with the hypothesis that the evolution of self-compatibility and concomitant changes in L. crassa and L. alabamica have occurred as a response to the need for an adequate seed-set under conditions where cross-pollination was inadequate for this purpose, On the other hand, there is some evidence that the supposed ability of self-fertilized plants to produce more closely adapted offspring has not been a factor in the evo- lution of self-compatibility in L. crassa and L. alabamica. I have no information on the relative growth and vigor of Leavenworthia plants resulting from self- and cross-polli- nations, but in the only population tested (population 171 of the self-incompatible race cl) 653 out of 672 (97 per cent) seeds from cross-pollinations germinated, and only 39 out of 60 (65 per cent) seeds from self-pollinations germinated. The difference in percentages of germination was significant (P<.01). Outcrossing is presumably advantageous in all self- incompatible races; otherwise, as Mather (1943b) pointed out, any suitable mutant conferring self-compatibility would increase in a population, since it would succeed in self- Pollinations as well as in cross-pollinations. An increase in inbreeding in an outbreeding species will thus inevitably lead to inbreeding depression. It can hardly be argued, therefore, that the increase in inbreeding accompanying the actual loss of self-incompatibility produces an immediate gain in the average fitness of a population. The evolution of self-compatibility in Leavenworthia must have had another adverse effect on seed production. It was mentioned above that in both self-incompatible and self- compatible populations, on the average, fewer seeds are produced in each silique after self-pollination than after cross-pollination. The increase in inbreeding accompanying the evolution of self-compatibility would therefore be ex- pected to decrease the average number of seeds produced in Sper sesnteene ieee reser ah aaa pe 132 LLOYD each silique, and in this respect would decrease the fitness of a population The lower suber of seeds per silique after self-polli- nation and the presence of heterosis in the self-incompatible races suggest that the evolution of self-compatibility in eavenworthia can only be due to a selective advantage acting prior to fertilization that is, a greater percentage of flowers being pollinated. This advantage appears to have arisen through the paucity of bees on smaller, earlier flowering populations occupying drier glades, and must have been sufficient to compensate for decreases in the number and quality of seeds resulting from each self- pollination. ACKNOWLEDGEMENTS I wish to express my gratitude to Professor Reed C. Rollins, who suggested this study for a Ph.D. thesis topic. He has given me the benefits of his knowledge of Leavenworthia and has provided in- valuable assistance and criticism throughout this study. Professor T. B. Mitchell, Dr. Ellis G. MacLeod, Professor Reginald H. Painter, Dr. Luella M. Walkley and Dr. Howard V. Weems kindly identified the insects I collected. Dr. Howard E. Evans ses and criticized the sections on the pollination of Leavenworthia flowers. Field work in Alabama and Tennessee was seabed by National Science Foundation grants to the Committee on Evolutionary Biology, Harvard University. LITERATURE CITED BAKER, H. G. 1953. Race formation and reproductive method 1 flowering plants. Symp. Soc. Exp. Biol. 7: 114-145. BALDWIN, J. T., JR. 1943. Forpinny - Sedum pulchellum. I, Cyto- geography. Bull. Torrey Club 70 BATEMAN, A. J. 1955. ania systems in Angiosperms: III. Cruciferae. Heredity 9:53-68. Oth BRAUN, E., et al. 1956. Pollination studies on red clover. Proc. 1 Tak. Cote Ent. 4:927. BRAUN, E. L. 1950. Deciduous os of eastern North America. Philadelphia and Toronto, Blakiston. Pe d BREWBAKER, J. L., AND S. K. MAJUMDER. 1961. Incompatibility ” the pollen grain. Recent advances in botany 2:1503- 1508. BUTLER, C. G. 1954. The world of the honeybee. London, Paar Crowe, L. K. 1964. The evolution of outbreeding in plants. I Angiosperms. Heredity 19:435-457. ‘dge DARLINGTON, C. D. 1939. Evolution of genetic systems. m—_ Cambridge University Press. EVOLUTION IN LEAVENWORTHIA 133 DaRWIN, C. 1876. The effects of cross- and self-fertilisation in the vegetable kingdom. London, John Murray. DRAYNER, J. M. 1959. Self-and cross-fertility in field beans (Vicia faba L.). Jour. Agr. Sei. 53:387. ERICKSON, R. O., et al. 1942. Dolomitic glades of east-central Missouri. Ann. Missouri Bot. Gard. 29:89-101 FENNEMAN, N. M. 1938. Physiography of eastern United States. New York and London, McGraw-Hill. FREE, J. B., and C. G. Butler. 1959. Bumblebees. London, Collins. FREEMAN, C. P. 1933. Ecology of the cedar glade vegetation near Nashville, Tennessee. Jour. Tenn. Acad. Sci. 8:141-228. FRIscH, K. von. 1950. Bees. Their vision, a Draipcren senses and language. Ithaca, Cornell University FRYXELL, P. A. 1957. Mode of reprodactiia’ sl of higher plants. Bot. Rev. 23:135-233. Harper, R. M. 1926. The cedar glades of middle Tennessee. Ecology 7: 4 1942. Natural resources of the Tennessee Valley region in | Alabama. Geol. Survey Ala. Special Report 17:1- ra JOHNSTON, W. D., JR. 1930. Physical divisions of northern Geol. Survey. Ala. Bull. 38:1-48. KnuTuH, P. 1906-1909. Handbook of flower pollination. 3 vols. Oxford, Clarendon Press. Lewis, D. 1943. The physiology of incompatibility in plants. I. The effect of temperature. Proc. Roy. Soc. London B 131:13-26. LInsLey, E. G. 1958. The ecology of solitary bees. Hilgardia 27 2543-599. Lioyp, D. G. 1964. Breeding systems and — in Leavenworthia (Crncifatee). Ph.D. thesis, Harvard Univers: MALYSHEY, S. I. 1936. The nesting habits of ae bees. A com- parative study. Eos 11:201-309. MANNING, A. 1956. Some aspects of foraging behavior of bumble- s. Behaviour 9:164-201. : MATHER, 7 1943a. Polygenic inheritance and natural selection. Biol. Rev. 18:32-64 ~———en——y, -1943b.. Specific ie aE in Petunia. I. Incompati- sk Jour. Genet. 45:215-2 MICHENER, C. D., AND C. W. pirate 1956. The ethology of ype erythronii with comparative data on er species aati Andrenidae). Univ. Kansas Sci. Bull. 37:654- eat T. B. 1960, 1962. Bees of the eastern United States. North and racial differentiation in plants. Cold Spring Harbor Symp. Quant. Biol. 24:47-56. : ULLER, H. 1883. The fertilisation of flowers. London, Macmillan. PANDEY, K. K. 1960. Evolution of gametophytic and sporophytic systems of self-incompatibility in Angiosperms. Evolution 14: 98-115. 134 LLOYD Pearson, J. F. W. 1933. Studies on the ecological relations of bees in 'the a region. Ecol. Monogr. 3:373-442. QUARTERMAN, E. 1950. Major plant communities of Tennessee cedar glades. Acie 31:234-254. sir re C. R. 1953. The behavior and social life of honeybees. , Bee Research Assoc. Ltd. ak C. 1924. Flowers and insects. XXIII. Bot. Gaz. 78:68-84. . 1929. Phenology of oligolectic bees and favorite flowers. Psyche 36:112-118. Rotuins, R. C. 1963. The evolution and systematics of Leaven- worthia (Cruciferae). Contr. Gray Herb. 192:1-98. SPORNE, K. R. 1956. The phylogenetic classification of Angiosperms. STEBBINS, G. L., JR. 1950. ageuniby and evolution in plants. New York, Columbia University Pre 1957. Self fertilization and population varia- bility in the ighvar plants, Am. Nat. 91:337-354. STEPHENS, S. G., AND M. D. FINKNER. a Natural crossing in cotton. Econ. Bot. 7:257-269 STEYERMARK, J. A. 1934. Some features of the flora of the Ozark region of Missouri. Rhodora 36:214-233. : Systematics Association Committee for Descriptive Biological Terml nology. 1962. IT. pe eer of simple symmetrical plane shapes (Chart I). Taxon 11:145-1 A TAXONOMIC REVISION OF CREMOLOBUS (CRUCIFERAE) KuLpIPp R. KHANNA AND REED C. ROLLINS Cremolobus, which literally means “the pendant lobes” and refers to the hanging valves of the silique, is a very distinctive genus of the Cruciferae, It was separated from Biscutella by De Candolle (1821) because of the thick pyramidal style, pendulous valves that are not adnate along their entire length and stipitate siliques. Three species were originally placed in Cremolobus by De Candolle. Since then a total of 17 species has been ascribed to the genus at one time or another. More recently, Cremolobus has received treatments by Schulz (1936) and by Macbride (1938). The latter author discussed the Peruvian species in some detail and dealt with nearly all of the species included in the present treatment. The exceptions were those whose oc- currence in Peru was doubtful. The criteria used by these and earlier authors have been critically examined by us. Our study has been based on a larger number of collections than was previously available, making possible the careful comparison of various transitional types over a wide geo- graphic range. As a result, we have found it prudent to place a number of previously recognized species together, particularly under C. chilensis. Each of two species of Cremolobus, C. rhomboideus and C. stenophyllus, has previously been raised to generic rank. Muschler (1908) erected the monotypic genus Urba- nodoxa, basing it on C. rhomboideus. Later Schulz (1933) described Loxoptera, ascribing the single species C. steno- phyllus to it. The characters presumed to separate these Species as genera distinct from Cremolobus, though definite in occurrence, are either variable intraspecifically or show Sporadic occurrence in other species of Cremolobus. Hence, it is unrealistic to maintain Urbanodoxa and Lozoptera as monotypic genera apart from Cremolobus. However, a discussion of each case is given individually under the Species concerned. GENERIC RELATIONSHIPS Despite the early association of species of Cremolobus with Biscutella, it is clear that the more significant relation- 135 136 KHANNA AND ROLLINS ship is in the direction of Menonvillea. An extreme re- duction of the replum and consequently of the septum, the frequent occurrence of a distinct gynophore, a strong tendency for the individual valves of the silique to become almost completely closed around the single seed and the usual presence of wings or their remnants on the valves are all features common to Cremolobus and Menonvillea. How- ever, the orientation of the valves with respect to the replum is quite the opposite in the two genera. The placing of Cremolobus and Menonvillea together accords with the treatment of Schulz (1936) where they are admitted to the tribe Cremolobeae. That author subdivided the Cremolobeae into subtribe Cremolobinae and subtribe Menonvilleinae, each with three genera. However, in our studies including previous work (Rollins, 1955), only two genera are recognized in place of six. Unlike the Schulz presentation, we would neither subdivide the Cremolobede nor place genera such as Biscutella and Dithyrea completely outside of the tribe including Cremolobus and Menonvillea. GENERAL CHARACTERISTICS OF THE GENUS A majority of the species of Cremolobus is centered in Peru, which is also the area of greatest diversity. However, the genus as a whole has a much wider distribution in the Andes of South America than Peru alone. It extends a few degrees above the Equator and is reported as far south as Coquimbo in Chile, although this record is in doubt. Creme lobus is mostly a genus of high altitudes, the majority, of the members growing above 6,000 feet and some reaching 13,000 - 14,000 feet. However, the plants of some species occur at lower elevations as well. Some collections of C. chilensis have been made from as low as 1,000 - 1,500 feet and C. stenophyllus and a variety of C. bolivianus are rr to occur at an intermediate range of 4,000 - 5,000 eet. The genus is represented by tiny herbs barely 3 cm. tall at one extreme and on the other by shrubby types like C. peruvianus that are reported to grow as tall as 5 meters: With the exceptions of C. peruvianus and the imperfectly known C. suffruticosus and C. subscandens, the rest of the species are strongly herbaceous, only occasionally having 4 tendency towards suffrutescence at the very base. REVISION OF CREMOLOBUS 137 Siliques of Cremolobus. Siliques of C. rhomboideus. FIG , 4 Sige i 4, 1G. sw on . 2 Siliques of C. peruvianus. FIG. 3. Siliques of C. stenophyllus. Fic. 4. Siliques of C. bolivianus. The stems are ordinarily profusely branched. In some specimens of C. chilensis, the branching is noticeably pro- fuse right from the base. The stems may be solid or fistular and range from less than 1 mm. to 8 mm. in diameter as 138 KHANNA AND ROLLINS found in some branches of C. peruvianus. The thickness of the stem must be much greater in this species than in any other, although no complete specimens have been observed to fully authenticate such a statement. The stems may be rounded or somewhat angular, with longitudinal furrows running along the entire length, as may be seen in some herbaceous types. The stem pubescence, like the pubescence of the other parts of the plant, is a very variable character in this genus, although the pubescence of each part is inde- pendent of every other. The differences in stem pubescence are a very noticeable feature of C. peruvianus, C. chilensis and C. bolivianus. The trichomes are dendritic, consisting of 6 - 8 rays or branches in C. peruvianus and C. bolivianus, while in the other species, when present, they are simple, unbranched and with flattened bases. The leaves are opposite in two species, C. bolivianus and C. subscandens, and alternate in the rest. Sometimes in the alternate leaved species, the leaves are opposite or sub- opposite. They vary in shape from linear-lanceolate to ovate-oblong or ovate-rhomboidal. Their size ranges from those present on the smallest plants of C. chilensis, where they are not more than 0.5 em. long and 0.4 cm. wide, to those of the large shrubby plants of C. peruvianus, where they are up to 12 cm. long and 4.5 cm. in width. In the latter, the leaves are quite hard and leathery in texture, this being the only species of the genus with this characteristic strong- ly developed. The leaves are usually markedly petiolate, the petiole being well developed in most species except in ©- chilensis, where the leaves are sometimes subsessile. The margin is somewhat serrate in all species, becoming dentate to variously pinnatifid in some plants of C. chilensis. +#€ leaves may be shallowly lobed, with the lobes close together, or deeply lobed with the lobes remote from each other. fH¢ lobes themselves may be entire or serrate, and are usually acute. Sometimes there is a tendency toward a bipinnatifid condition in C. chilensis. The leaves of C. peruvianus alt C. bolivianus are the most strongly pubescent of the genus, although here, as in the other species, some specimens were observed to lack pubescence completely. The trichomes, like those of the stem in these species, are stellate and dendritic and are situated mostly on the lower surfaces of the leaves especially along the veins. In the other species of the genU® REVISION OF CREMOLOBUS 139 the leaves are only sparsely pubescent with minute tri- chomes or they are glabrous. In these cases, the trichomes are simple, with bases flattened to a varying degree, and are scattered on the leaf surface. Here again, the pubescence is almost wholly confined to the lower surfaces. The leaves of C. stenophyllus and C. suffruticosus are completely glabrous. The flowers are borne in elongated or semi-corymbose racemes. The size of the flower, as reflected by the petal length, ranges from 2 mm. to 5 mm. and in several cases considerable within-species variation was observed in the flower size, particularly in C. chilensis. The flower color in the genus as a whole ranges from white to creamy or sulphur yellow becoming somewhat darker on drying in C. peruvianus. The structure of the flower is that of a typical crucifer and is quite uniform for the entire genus. The Sepals are green or yellowish green in color, boat-shaped and hyaline-margined. The venation is sparse and is pres- ent only in the central non-hyaline region. The sepals are usually glabrous but may sometimes possess a few trichomes on the outer basal portion in C. peruvianus. They are in two pairs which are differentiated from each other. The outer pair consists of relatively narrow sepals arising from narrower bases than those of the inner pair. The members of the outer pair are also less deeply keeled than those of the inner. The petals are about one and half times longer than the sepals but rarely the sepals and petals are subequal. The petals are entire to wavy margined and are spatulate to distinctly clawed. The nectaries are inconspicuous to well developed. The stamens are tetradynamous with filaments gradually and not abruptly flattened at the base. The anthers are sagittate and dehisce by longitudinal slits on the inner side. The ovary in most species is supported on a distinct gynophore. The gynophore is rudimentary in C. bolivianus and C. subscandens. The ovary is initially bilobed and flattened at right angles to the septum. It continues to develop in the same pattern, ultimately becoming the characteristic fruit of Cremolobus. The styles are long, somewhat flattened to triangular, except in C. peruvianus (fig. 2) where they are peg shaped. The stigma is capitate, usually exceeding the upper portion of the style in diameter. In C. bolivianus (fig. 4) and C. subscandens, like the gyno- 140 KHANNA AND ROLLINS phore, the style is rudimentary and is completely covered over by the capitate stigma at maturity. The fruits are borne on pedicels that may be pubescent or glabrous and vary considerably in length. They may be horizontal, sigmoid or ascending. In C. subscandens, the pedicels are nearly erect, paralleling the axis of the infruc- tescence. In C. suffruticosus, they are very distinctive, being the longest in the genus, i.e., up to 2.5 em. long. The pedicels in this species are also more flattened in contrast to the cylindrical ones of the other species. The shape of the fruit is distinctive because the valves of the silique not only hang from the replum but are also flattened in a plane that is perpendicular to the face of the replum. Exceptions to the pendant position are found in C. subscandens, C. bolivianus and C. peruvianus, where the valves are horizontally ori- ented to the main axis of the replum. The presence of a gynophore is also a very characteristic feature of the genus, even though it is very much abbreviated in C. bolivianus and C. subscandens. The development of the gynophore in most species is very striking, being almost as long as the style (cf. fig. 1, 3 & 8). However, it is sometimes con- siderably shorter than the style in C. suffruticosus. The valves of the fruits are laterally winged or wingless, but the extent of winging varies to some extent. C. bolivianus, C. subscandens and C. rhomboideus have completely wing- less valves. The most variable species with respect to this character is C. chilensis (fig. 5-10). Here, the development of the wings takes place to a varying degree. The valves of the other species have well developed wings, the largest of which are found in C. peruvianus. The wings are veine in this species (fig. 2), a character not present elsewhere in the genus. The wings are translucent in C. stenophyllus (fig. 3) where, in addition to the principal wings, the valves also have minor cross wings rolled inward, both above and below. The surfaces of the valves are smooth in C. peru- vianus and in some individuals of C. chilensis. They are radially wrinkled and thick walled in C. rhomboideus (fig- 1) and reticulate in C. bolivianus (fig. 4) and C. subsean- dens. The valve surfaces in some plants of C. chilensts have various sizes of simple trichomes (fig. 6, 9 & 10). The margin of the valves, when not winged, ranges from simple to dentate. REVISION OF CREMOLOBUS - 10. Siliques of ae. chilensis showing range of variation slice re: re d brev | Fic. ater Fig. 5, - Pubescent silique with well develo reticulate silique. crenate wings. Fic. 10. rai nt and elongated valves, ely pubes tionk dtticdan with nearly wingless sve gute 142 KHANNA AND ROLLINS The seeds are orbicular to suborbicular, flattened and rown in color. They are attached by either a long or a short funicle. The cotyledons are always accumbent. ACKNOWLEDGEMENTS We are indebted to the Directors and Curators of the following herbaria for making materials in their care available for our study: British Museum (Natural History), London; Chicago Natural History Museum; Conservatoire de Botanique, Geneva; Herbarium of the University of California, Berkeley; Instituto de Botanica Darwinion, Buenos Aires; Instituto Miguel Lillo, Tucuman; The New York Botanical Garden; and the United States National Herbarium, Smithsonian Institution, Washington. These materials and the speci- mens in the Gray Herbarium are cited according to the abbreviations given by Lanjouw and Stafleu (1964). TAXONOMIC TREATMENT CREMOLOBUS DC., Systema 2: 418. 1821 Annual or perennial; small herbs to undershrubs or shrubs; stems unbranched to profusely branched, often from the base; leaves region, inner and outer pair differentiated from each other; petals white to cream, mostly clawed, entire or with a wavy margin; nectaries inconspicuous to well developed; stamens tetradynamous; style absent or short and peg shaped to elongated and triangular; attached valves; valves triangular, suborbicular to orbicular, cana to reticulate, glabrous to pubescent, winged to wingless, rarely Ww! h cross wings, each valve enclosing a single seed; seeds orbicular; cotyledons accumbent. KEY TO THE SPECIES A. Leaves opposite; silique reticulately ridged, sessile, wingless; stigmas sessile. B. Plants densely pubescent with conspicuous trichomes; leaves ovate-lanceolate to linear-lanceolate, lower about 9 cm. long and em. wide; pedicels mostly horizontal ............ 1. C. bolivianus. PROPOR POC OS HOD Seb OR SEs eebenoeees sees. REVISION OF CREMOLOBUS 143 C. Plants perennial, suffrutescent, undershrubs or vinelike in habit. D. Plants pubescent at least in ‘the upper part; pedicels about 1 cm. or less long; valves 10 mm. across, horizontal; styles short and thick . C. peruvianus. D. Plants glabrous; pedicels 1-2.5 em. long; valves 2-3 mm. across, pendant; styles thin and elongated 4. C. suffruticosus. C. Plants annual, herbaceous throughout. E. Valves with secondary wings at right angles to the principal wings; margins of the principal wings rolled, thin an translucent 6. C. stenophyllus. E. Valves without secondary wings; margins of the wings not u que, F. Plants up to 1.25 m. tall: pedicels up to 2 cm, long; valves triangular, turgid, hard and woody in texture, completely wingless, stigma not markedly different from the style in diameter 7. C. rhomboideus. . Plants 3-35 em. tall; pedicels not more than 1 em. long; valves triangular to orbicular, mostly flat or lens shaped in the center, thin and papery, winged to wingless; stigma distinct 5. C. chilensis. rj 1. Cremolobus bolivianus Britt. Bull. Torrey Club 16:17. 1889 Perennial; stems branched, densely pubescent with simple or dendritically branched trichomes; leaves opposite, shortly but dis- tinctly petiolate, ovate-lanceolate to linear-lanceolate, lower 9 cm long and 3 cm. wide, minutely and distantly toothed, acute to ac- mostly horizontal, 5-7 mm. long, pubescent; sepals green, oblong, other at the point of attachment, 4 mm. across, horizontal, reticulate, Wingless, margin strongly dentate, sometimes pubescent; seeds about 1 mm. across, funicle very short; cotyledons accumbent. Fig. 4. BOLIVIA: Unduavi, 8,000 ft., October, 1885, Rusby 1816 (NY, Type; GH, Us, isotypes); in the vicinity of Ananea, Parjchani (Prov. Larecaja), June, 1860 Mandon 905 bis (G). Opposite leaves and opposite branching are strongly fixed features of C. bolivianus and C. subscandens and the two 144 KHANNA AND ROLLINS species are obviously very closely related if not variants of the same species. The pubescence is definite and per- vasive on the specimens we have seen of C. bolivianus, while it consists of minute flattish trichomes scattered along the veins of the under surfaces of the leaves or is completely absent in C. subscandens. Other minor differences, as indi- cated in the key and in the descriptions, tend to correlate to give some substance to the evidence for the recognition of two taxa. However, more adequate specimens and fiel work are needed to provide the data for a more reliable interpretation than we can give at present. If both taxa had not been described at the specific level, we would be inclined to consider one of them as a subspecies of the other. 2. Cremolobus subscandens Kuntze, Rev. Gen. 3 pt. Perennial; stems branched, suffrutescent, sparsely pubescent with short simple trichomes, branching opposite, at least above; leaves opposite, ovate to ovate-rhomboidal, 4-5 em, long and 3-5 cm. wide, serrate, acute, veins sparsely and minutely pubescent on the lower side, texture stiff; pedicels not exceeding 0.5 cm. in length, ascending to nearly erect, glabrous; sepals green, outer and inner pairs different in width and point of attachment, ca. 2 mm. long; petals white, spatulate, 2.5-3 mm. long; nectaries poorly developed; stamens distinctly short and long, anthers elongate, sagittate; gynophore 4,000 ft., 1885, Rusby 1820 (GH, NY, US). ARGENTINA: Ledesma, (Prov. Jujuy), October 2, 1906, Dinelli 5108 (GH, LIL). From limited herbarium material, it is difficult to gain any real conception of what the plants of this species are like. The name Kuntze applied suggests that growing plants are weak stemmed and at least dependent on other plants for support, if not actually climbing in habit. The specimens are in accord with this idea. Not a single one has the basa portion of a plant present and the material of each collection appears to have been taken as ultimate flowering or fruiting REVISION OF CREMOLOBUS 145 branches. We have no evidence as to height of the plants or distance of the remotest branches from the ground. Not one of the seven sheets we have for study shows more than one or two developed fruits per inflorescence branch even when the number of pedicels with an abortive ovary is 20 or more. This suggests that pollination is relatively inefficient, at least in areas where specimens have been collected. A general remoteness of one plant from another, either spatially or by interspersion of other plants, may also be inferred as a possibility. Both C. subscandens and C. bolivianus, if they are in fact distinct species, would be rewarding for further study and collecting. True climbers are certainly rare if not absent in the Cruciferae, but these are the best candidates we know of to possess the climbing feature. 3. Cremolobus peruvianus (Lam.) DC., Systema 2: 419. 1821 Biscutella peruviana Lam., Encycl. Méth. 3: 620. 1791. Cremolobus pubescens Hook., Icon. Pl. t 81. 1837. Perennial; stems woody, branched, glabrescent and smooth below, pubescent in the upper parts with 6-10-rayed stellate trichomes; leaves alternate, petiolate, ovate-lanceolate, lower large, up to 12 em. long and 4.5 em. wide, upper much shorter, 2.5 em. long and 1-3 cm. wide, serrate, acute, rarely acuminate or ct aretiaic, Racor surface deep green, sparsely pubescent, lower surface pale n ore profusely pubescent, leaves often completely a veins Spay: on the lower surface, petioles 5-9 mm. long; pedicels pubescent, up to 1 em. long when mature, horizontal to slightly curved hig ag A sepals greenish, glabrous, or ae ae a few trichomes on the lower portion, obovate, 3-4 mm. lon argin hyaline, ie veins confined to the central region, not ant the hyaline portion, keeled, outer pair linear and less deeply keeled as compared to the inner, which is broader and more pouch-like; petals cream to sulphur yellow, some- what orange colored when dry. linear obovate, distinctly clawed, 5-7 mm. long, blade margin wavy or entire; stamens distinctly long and short, the longer ones 4 mm. long and the shorter ones 2 mm. long, anthers sagittate, introrse; nectaries 4, prominent; gynoecium raised on a smooth gynophore, style short, thick, about the same diameter as the replum ze below it, smooth when young but ridged in the mature fruit; stigma capitate, covering the style like a triangular cap, depressed in the center, reddish in color, nearly of the same diameter as the style when young; valves attached horizontally on the replum, orbicular, glabrous, wings well developed, reticulately veined, entire; seeds attached by a thin funicle about 1 mm. long, brownish reticulations, one seed in each locule; eotsiodans accumbent, Fig. 2. 146 KHANNA AND ROLLINS Cremolobus pubescens, as described and illustrated by Hooker, is similar to C. peruvianus in all the characters given, such as the suffrutescent habit, the pubescence, the panicled inflorescence, and particularly the shape of the fruit and style, which are very characteristic features of . peruvianus. The only noticeable difference from C. peruvianus is in the illustration of the flower where the sepals and petals look nearly equal. Probably the material examined by Hooker was young both as to flowers and fruits. In both figures where fruits are represented, younger material appears to have been the basis for the illustration and the same is true of the flower, which does not seem to have opened fully. In such a stage of development, the relative size of the sepals and petals is likely to be mis- judged, since the full size of the petals is manifested only when the flower is fully open. DISTRIBUTION: Colombia, Ecuador and Peru. De Can- dolle cited a Dombey specimen from Chile in his Systema (1821) but in an earlier treatment of C. peruvianus (1811), it was given as Peru. Since we have no subsequent evidence that the species extends to Chile, we assume the citation in the Systema is in error. COLOMBIA: Cauca. ‘San Jose’, San Antonio, altitude 2100-2500 m., July 1, 1922, Pennell 7630 (GH); Hoya del rio Tambito, Mt. Munchi- que, Occidental Cordillera, 2000-2500 m., July 16, 1939, Arbelaez and Cuatrecasas 6209 (F, US); near Puenta de Tierra, 2200 m., September 13, 1944, Core 1249 (WvA). ECUADOR. Corazon, Mission Scientifique de Ed. Andre K160 (¥, GH, NY, US); Mojanda, 1906, Herb. Lehmannia- num 5400 (F, GH, NY, US); Malchingui to Pomasqui, Pichincha, 3000- 3600 m., August 13, 1923, Hitchcock 20870 (Us); Mindo, June 26, 1876, Mission Scientifique de Ed. Andre K161 (NY); San Miguel om margin of paramo between Atuntiaqui and Hacienda Pinon (Prov. Imbabura), 11,300 ft., June 20, 1944, Wiggins 10,334 (GH); western side of Pichincha, Jameson (G, GH, NY, US); Cinto Santuario, S.E. de Lloa (Prov. Pichincha), 3200 m., July 28, 1927 Firmin 153 (US)3 Quito, 1848, Jameson (G); Andes of Quito, 1859, Jameson (6, GH, NY); Quito-Popayan-Bogota, 1843, Hertweg 886 (G); Supra ™° Cristal, Occidental Cordillera, July 10, 1876, Mission Scientifique 4 Ed. Andre 4005 (¥, GH, NY, US); Nudo de Portete, pass between head waters of the rios Tarqui (Atlantic) and Giron (Pacific) (Er Azuay), about 9000 ft., March 10, 1945, Camp E-2176 (NY); 1-8 kM» north of the village of Servilla de Oro, Eastern Cordillera, (Prov. Azuay), 8000-9000 ft., July 28-August 12, 1945, Camp E-4351 (GH); Andes, 1857-1859, Spruce 5019, 5506, 6049 (F, G, GH, NY); 1860, Fraser (G); Tropical South America, 1862, Fraser (BM). PERU: Jussieu (frag. of type, F); 1827, Pavon (F, G); Ruiz and Pavon REVISION OF CREMOLOBUS 147 1778-1788 (F); Rio Masamerich (Dept. Junin), 2200-2300 m., 1909- 1914, Weberbauer 6673 (F, GH, NY); upper slopes and summit of Cerros de Calla-Calle, near kms. 403-407 of Balsas-Leimebamba road, Prov. Chachapoyas (Dept. Amazonas), Aug. 18, 1962, Wurdack 1695 Ss) Cremolobus peruvianus is of considerable interest because of the vine-like growth habit and the divided paniculate raceme with the subdivisions most frequently in the axils of bracts. In turn, the branches bearing the inflorescences are in the axils of leaves. The stem is definitely woody. The habit of growth, the inflorescence form and the woody nature of the stem are all unusual characteristics for the family Cruciferae. From specimens alone, it is difficult to reconstruct an exact description of the whole plant because the basal portions are not included. Notes on a few of the labels indicate the Species may be either vine-like or shrub- by, with stems up to 5 meters in length. On a number of specimens, the leaves are all oriented in one direction, indicating that the stem on which they were borne was reclining, Aside from the habit of growth, there are other deviating characteristics evident in the specimens the taxonomic significance of which cannot be completely assessed from the specimens alone. For example, among the collections here referred to C. peruvianus, some specimens have com- pletely glabrous leaves, in others the leaves are sparsely pubescent with either simple trichomes or with dendritic trichomes and in one collection (Camp E-4351) the lower leaf-surfaces are densely pubescent with short simple tri- chomes. It is possible that these differences have some taxonomic significance but we have not been able to find any correlation with other distinctions. Until special studies can be made to carefully elaborate on the nature of the srowing plants and the significance of such deviations as we have pointed out, we have felt impelled to provide a conservative treatment that emphasizes the similarities we can assess rather than the differences. It should be pointed out that the type of C. peruvianus has glabrous leaves. 4. Cremolobus suffruticosus DC., Systema 2: 419. 1837 Biscutella suffruticosa DC., Ann. Mus. Hist. Nat. Paris 18: 296, tab. V. 1811. Perennial shrubs or under shrubs, height unknown; stems alternate, 148 KHANNA AND ROLLINS woody, glabrous; lower leaves unknown, upper ones alternate, petiolate, ovate-oblong, 2.5 em. long, 0.6 cm. wide, entire or minutely to strongly serrate in the distal portion, glabrous both on the upper and lower surfaces, petiole up to 5 mm. long; pedicels comparatively long, slender, up to 2.5 em. when mature, sigmoid to irregularly directed upwards, glabrous; sepals green, broadly ovate, obtuse, 3 mm. long, hyaline margined, boat shaped, differentiated into an outer and an inner pair, upper part of outer pair with large flat trichomes; petals whitish, somewhat darker when dry, 5-7 mm. long, margin in the region of the blade wavy; stamens tet Sudysininece filaments narrowed towards the point of anther attachment, anthers elongate, sagittate, introrse; gynoecium raised on a short thin gynophore; style short and thick when young, thin and elongate when mature, stigma capitate; valves pendant with respect to replum, more or less orbicular, wings present, glabrous; seeds not observed As compared to C. peruvianus, the flowers of C. suffrutt- cosus are considerably larger, and both pedicels and styles are more than twice as long. The two species should not be easily confused because of these striking differences but it is clear that Macbride (1938) did not properly distinguish between them. C. suffruticosus is evidently quite rare. To our knowledge, it has not been collected outside of Peru, even though it has been attributed in the literature to both Chile and Ecuador. The above description is based on rather poor specimens. We have seen only a fragment of the original material collected by Dombey and not the Lagasca specimen. How- ever, the plate accompanying De Candolle’s proposal of the species is very helpful in interpreting the few available specime ns. PERU: Dombey (frag. of type, F); Chechin, Dombey (F); 1827, Pavon (F, G); aia of Bafios, Wilkes Expl. Expedition 1838-1842 (Us); Mahuay, Prov. Bolognesi, Dept. Ancash, Peru, on May 22, 1954. Cerrate 2216 (GH). 5. Cremolobus chilensis DC., Systema 2:419. 1821 Seite Chilensis DC., Ann. Mus. Hist. Nat. Paris 18: 297, tab. I, 1811. B. cuneata Lagasea, Gen. Spec. Pl. Nov. 20. 1816. Cremolobus sinuatus Hook., Icon, Pl. t 99. 1837. C. pinnatifidus Hook., Icon. Pl. t 100. 1837. C. aphanopteris Gray. Bot. U.S. Expl. Exp. 55. 1854 C. ta var. integrifolius Gray, Bot. U.S. Expl. Exp. 56. C. parviflorus Wedd., Ann. Sci. Nat. Ser. 5. 1: 283. 1864. C. humilis Musch., Engl. Bot. Jahrb. 40: 269. 1908. REVISION OF CREMOLOBUS 149 C. Weberbaueri Musch., Engl. Bot. Jahrb. 40: 170. 1908. C. Paysonii Schulz in Macb, Field Mus. Pub. Bot. 8: 80. 1930. C. Benoistii Macbr. Field Mus. Pub. Bot. 13 (2): 941. 1938 sometimes pubescent in the younger parts; leaves alternate, the lower ones occasionally opposite, subsessile 'to petiolate, petiole up to 1 cm. long, blade linear-ovate or rhomboidal, serrate to deeply pinna- tifid, leaves including petiole 0.5 cm. to 5.0 em. long, 0.4 em. to 1.5 em. wide, glabrous to minutely pubescent on both the upper and lower side; pedicels up to 1 em. long when mature, horizontal, sigmoid or divaricately ascending, glabrous or minutely pubescent on the upper side with flat pointed trichomes; sepals greenish, hyaline margined, veined only in the central region, 2 mm. long, outer pair narrow and with a narrow base, inner pair broader with a broader base and deeper keel; petals white, 2.5-3 mm. long, spatulate to distinctly clawed, entire or Wavy in the upper region; nectaries well developed; stamens tetradynamous, distinctly long and short, filaments flattened at the base, anthers heart shaped or slightly more elongate, to variously pubescent, creamy to brown in color; seeds single in each valve, attached by a short white funicle, yellowish to dark brown in color, suborbicular; cotyledons accumbent. Fig. 5-10. Cremolobus chilensis is a highly polymorphic species and has the widest distributional range in the genus. The largest number of collections are from Peru, while only a few collections have been made from adjoining regions of Bolivia, Argentina and Chile. This species ranges in altitude from 1,300-10,000 feet. As might be expected in a plant of such diverse altitudinal and climatic zones, there exists a considerable difference in time of flowering and fruiting (February to September). This species is very variable with respect to size, branching of the stem and the charac- teristics of the leaves and fruits. These characters seem to independently determined genetically. In most cases, they are uncorrelated, making it hard to study any one of them consistently with the rest. Any one of these characters might have been of specific importance had its development en associated with an isolating mechanism of some sort. To be effective, such an isolating mechanism controlling the 150 KHANNA AND ROLLINS flow of genes that determine a particular character would have to coincide with ecological or geographical niches of importance to the specific populations. But the occurrence of a whole array of transitional types over the entire range of the species, points in another direction. In the following, the characters which formed the basis for a number of proposed species given in the synonmy list, are discussed in some detail. The size of the plant varies from 4-35 cm. in height. There does not appear to be any definite correlation between the size and other characteristics of specific rank or even with the geographic distribution of the species. However, of the few collections available from southern Peru, Bolivia, and Argentina, most of them had much smaller sized plants than are usually present in the more abundant collections farther north. These small plants may represent only the chance factor involved in making the collections or possibly they grew under the less favorable circumstances than the larger specimens. Size in annual plants is known to be extremely variable and sensitively reflects the quality of the environment. We suspect that this is the case in C. chilensis. Since size as a character may be both environmentally and polvgenically determined, it cannot, by itself, be assigned a high degree of taxonomic importance. The leaf characters are similarly variable. The leaf size, which is to some extent correlated with the height and general habit of the plant, ranges from 0.5-5.0 cm. and is of little specific value. The authors describing the species given in the synonymy list described the leaf shape various- ly as ovate, oblong, ovate-oblong, spatulate, subrhomboidal, lanceolate, oblanceolate, elliptic, etc. By examining the available collections, we cannot escape the conclusion that these leaf shapes are closely interconnected and any one of them is by no means a specific criterion. Similarly the nature of the leaf margin is a very flexible character, some times observably so in the same plant. The lower leaves may be entire when the upper are pinnatifid, More often the variability of this character is observed in the members of the same population where some plants have entire OF serrate leaves and the others have deeply pinnatifid ones- This may be neatly observed in the collection, Stork, 11441 (GH). REVISION OF CREMOLOBUS 151 The fruit characters have been greatly relied upon as the key features for distinguishing the purportedly separable species whose names are listed above in the synonymy. These include size of fruit, shape of the valves (triangular to suborbicular to orbicular), development or the lack of development of the wings, margin of the wings (entire or variously lobed), and pubescence (its presence or absence or unequal development). These characters (cf. fig. 5-10) may vary from one plant to another in the same population and their expression is sometimes dependent upon other features as well. The size of the fruit, for instance, is often related to the vigor of the plant or to the development of the wings. The ultimate shape of the fruit is also considerably influenced both by the shape of the valves and development of the wings. The dentation and lobing of the margin of the wings seem to have both a genetic as well as a develop- mental basis for their expression. This is evident because of their variation from plant to plant as well as in the same individual. The presence or absence of wings and of pubes- cence are also probably under genetic control. The wingless form was named C. aphanopterus by Gray from material of the Wilkes Expedition collected at Obrajillo, Peru. It is probably significant that certain specimens collected at the same time have winged fruits and perhaps were part of the same population. A great amount of variation in these characteristics has been observed among the individuals of the same collection which obviously represent the same population. Examples illustrating variation in the wing development, ranging from presence to absence, are found in the following col- lections: Ferreyra 7279 (GH), Ferreyra 8945 (GH), and Ferreyra 7520 (us). Examples illustrating variation in pubescence are found in the following collections: Ferreyra 8945 (GH), Ferreyra 6088 (US), Ferreyra 7520 (US), peta and Featherstone 467 (F) and Pennell 14398 NY). The great variability within this species is clearly shown by a study of the more recent collections. All present infor- mation indicates that the names given in the synonymy list apply to variants of C. chilensis. Placing them in synonymy is the only reasonable course of action if the close inter- gradations of the morphological variants are taken fully 152 KHANNA AND ROLLINS into account. More information, particularly of a cyto- genetic nature, is highly desirable and perhaps would permit a more precise taxonomic treatment of what appears to be a rather complex assemblage of minor variants. The origin of the type material of C. chilensis collected by Lagasca is given as “chili” by De Candolle. We have not seen the type, but the illustration accompanying the original description reflects a similarity to specimens from Peru rather than any we have seen from Chile. Lagasca, in describing Biscutella cuneata, which presumably was based on the same original material, merely gives, “Habitat in America meridionali.” It is suspected that the type of C. chilensis came from what is now Peru rather than from Chile. The scarcity of C. chilensis in Chile is indicated by the fact that other than an early Dombey collection which may actually have been gathered in Peru, we have seen only one other collection, that of Ricardi, Marticorena and Matthei from over 10,000 ft. in the Department of Arica. DISTRIBUTION: Northern Argentina, Bolivia, northern Chile, Peru. PERU: Andes of Peru, Wilkes Exped. (GH, type; isotype of C. pinnatifidus var. integrifolius Gray); without locality, 1837, Dombey 619 (G). La Libertad: Canduay, Sinsicap (Prov. Otuzco), 2750 m., May 1, 1954, A. Lopez M. 1059 (us); Chiputur Mountains (Trujillo), 400 m., August 1949, A. Lopez M. 368 (GH). Ancash: near San Rafael between Casma and Huarmei (Prov. Santa), sandy soil (Prov. Bolognesi), foot hills, May 22, 1954, Cerrate 2228 (GH). Lima: Ambar (Prov. Cajatambo), annual rainy season herb on gravelly hill side, 2010 m., April 16, 1939, Stork 11441 (GH, UC); Obrajillo, Wilkes Expedition (GH, type; US, isotype, C. aphanopters Gray); above Obrajillo along Rio Chillén, rocky slopes, 2800-3200 m., June 13-23, 1925, Pennell 14398 (F, GH, NY, US); San Buenaventura, wet shaded banks beside streams, 2700-2900 m., Pennell 14588 (F); Huaros, bushy slopes, 3200-3400 m., June 23, 1925, Pennell 14702 (F, NY, US); Matucana, moist soil among rocks on steep western , , stone 84 (F, US); Matucana, rocky slope, 8,000 ft., April 12-May % REVISION OF CREMOLOBUS 153 1922, Macbride and Featherstone 112 (F, NY); Matucana, moist swales on northern mountain side, about 8,000 ft. April 12-May 3, May 3, 1922, Macbride and Featherstone 428 (F, GH); Matucana, steep moist eastern slope, 8,000 ft., April 12-May 3, 1922, Macbride and Featherstone 467 (F); Matucana, rocky, grassy shady slope, 8,000 ft., April 12-May 3, 1922, Macbride and Featherstone 557 (F, GH); Matucana (Lima-Oroya rail road), 2400 m., 1910, Weberbauer 5256 (F, GH, US); between Matucana and Tambo de Viso (Prov. Huarochiri) April 9, 1910, Weberbauer 5209 (F); Viso, shallow soil on rocks about 9,000 ft., May 5-14, 1922, Macbride and Featherstone 601 (F, NY); Viso, sandy sunny soil, 2800 m., April 23, 1929, Good- speed et al. 11518, (GH, UC); Rio Blanco, decomposing rock outcrops, 12,000 ft., May 8-19, 1922, Macbride and Featherstone 696 (F, NY) and 697 (F, GH); Rio Blanco, between San Mateo and Casapalea (Prov. Huarochiri), rocky mountain hea, 3,500-3,550 m., May 25, 1950, Ferreyra 6975 (GH); Rio Blanco, between San Mateo and Casapalea (Prov. Huarochiri), evergreen mountain forests, 3300- 3400 m., March 9, 1953, Ferreyra 8945 (GH ); Rio Blanco, open hill Side, 3,000-3,500 m., April 15-17, 1929, Killip and Smith 21685 (us) and 21700 (Ny, Us): Cerros de Surco, clay iy Be slopes, 2100- of mountains 3200-3300 m., May 22, 1949, Sakuno 4088 (GH, US); near Surco, 2100-2200 m., April 3, 1955, Ferreyra 10529 sg near Chicla, 12 000-13,000 ft., ‘Aor 1-23, 1882, Ball (ox, NY). Arequipa: 1925, Pennell 13223 (F, GH, NY, US). Maquega: Gganinee 3200 m., F, m., March 17-18, 1925, Weberbauer 7416 (F, US). BOLIVIA: Prov.y¥ Larebiia near Torata, rocky cliffs, sub alpine, 3700 m., March-April 1960, Mandon 905 (F. GH, NY, isotypes of C. parviflorus Wedd.) ARGENTINA: Tucuman. Estancia Las Pavas-Cerro Boyo. (Dept. Burhart 5281 (st). CHILE: Dombey (photo at F); Subida al Porte- zuela de Chapiquifia, Dept. Arica, Prov. Tarapaca, 3360 m., March 26, 1961, Ricardi, Marticorena and Matthei 157 (CONC, GH). 6. Sernelpny stenophyllus rues Engl. Bot. hrb. 50. Beibl. 111:7. 1913 Loxoptera eepscrwe : mal Schulz, Bot. Jahrb. Annual herbs, 25-30 em. in stems ee ed profusely, angulate, glabrous throughout; leaves alternate, linear, 4-5 cm. long, less than 0.7 em. broad, dentate, acute, glabrous on both sides, short petiolate, petioles very slender; pedicels about 1 cm. long when fully mature, oo sigmoid, minutely pubescent; sepals greenish, oblong, ca. 2 mm. lo ong, hyaline margined, inner pair deeply keeled with a broad 154 KHANNA AND ROLLINS base; petals white, clawed, 3-3.5 mm. long, entire; nectaries 4, poorly developed; stamens distinctly tetradynamous, filaments uniform in breadth throughout, up to 3 mm. long in paired stamens, anthers sagittate; gynoecium supported on a thin gynophore which may be of the same length as the style in mature fruit; style rigid, conical elongated, stigma slightly broader in diameter than that of the style, separated from the style by a distinct constriction; valves slightly directed downwards with respect to replum, orbicular, 2.5-3.0 mm, across, glabrous, wings very well developed, thin, translucent, some- times rolled over at the edges, margin entire to slightly wavy, minor cross wings present in the center on both sides of the valves, extent of development variable, minor wings curled inward forming a pouch like structure; seeds elongate orbicular, attached by a thin funicle, i in the position of the cross wings, dark brown; cotyledons accumbent. Fig. 3. P Mountains near Chosica (Lima-Oroya railroad), 1400-1500 m., April, 1910. Weberbauer 5335 (Isotypes: F, GH, US); near Lima, July 5, 1914, Rose 19479 (us). Schulz (1933, 1936) treated Cremolobus stenophyllus as a separate genus, Loxoptera, on the basis of a number of presumed distinctive characters such as the orientation of the nectaries, the heart-shaped anthers, the gynoecium being longer than the paired stamens, the long and thin style, the stigma being smaller than in other species of Cremolobus and the development of the minor wings on the valves. The heart-shaped anthers are not characteristic of this species alone. The relative length of the stamens and gynoecium 1s a variable feature, the latter being influenced considerably by the age of the flower. The long and thin style and the size of the stigma is similar to that of other species of Cremolobus. The only major character which distinguishes this species from others of Cremolobus is the presence of secondary wings on the valves. The extent of secondary wing development is variable as may be seen by observing different specimens of the same population. Unfortunately, the available material of the species is too meagre to perm! an assessment of the full range of variability of this character. However, it is pertinent to mention here the parallel situation in the related genus Menonvillea. In 4 previous study of Menonvillea (Rollins, 1955), specimens with 6 and 10 winged fruits (formerly known in the two genera, Hexaptera and Decaptera, respectively) were shown to be present in the same collection. The partial develop- ment of secondary wings in Menonvillea scapigera was aiS° stressed. On this basis it was concluded, “that the presence REVISION OF CREMOLOBUS 155 or absence of additional wings may not be of marked signifi- cance.” Because the extent of wing development did not coincide with other significant natural generic boundaries, Hezxaptera and Decaptera were merged into the single genus Menonvillea. A similar treatment is indicated by the evi- dence for the present case as well. Thus, the generic name Loxoptera is not maintained. In the four sheets of specimens we have for study, the leaves are shallowly dentate to entire and the lower ones in particular have a very slender petiole. The branching is all well above the base and it appears that no truly basal leaves are produced. Plants of this species are undoubtedly rapid growers, presumably responding to moisture for their growth period more than to season. 7. Cremolobus rhomboideus Hook., Icon. Pl. t 32. 1837 Urbanodoxa rhomboidea (Hook.) Muschler, Bot. Jahrb. Annual herbs, up to 1.25 m. tall; stems rather sparingly branched in the upper region, glabrous throughout; leaves alternate, tending be opposite on the lowermost portion of the stem, short petiolate, rhomboidal to ovate-lanceolate, 2.5-3.0 cm. long, 1-1.5 cm. broad, serrate acute, glabrous; pedicels very short when young, up to 2 cm. when mature, divaricate, minutely pubescent on the upper side; sepals ca. 1.5 mm. long, outer pair less hollow than inner and attached by a narrow base; petals spatulate, not sharply clawed, 2.5-3.0 mm. long; stamens long and short, longer ones 2 mm. in length, anthers heart-shaped, sagittate; gynoecium supported on a thin gynophore; Style thick, conical elongated when mature, stigma about the same as the style in diameter; valves narrowed to a limited point of attach- ment on the lower side of the replum swelling, valves ca. 4 mm. long, 3 mm. across, wingless, rugose, dentate, pubescent; seeds attached by a distinct funicle, flat, 2 mm. across; cotyledons accumbent. Fig. 1. PERU: Carumas (Prov. Moquegua), 3,200 m., February 21-March 6, 1925, Weberbauer 7296 (Us); Viso, about 9,000 ft., May 5-14, 1922, acbride and Featherstone 572 (NY). Muschler (1908) separated C. rhomboideus from Cremo- lobus and erected the monotypic genus Urbanodoxa. The stated basis for this course of action was the absence of alternate leaves, the presence of poorly developed nectaries, and the lack of development of the wings on the valves. These features, he thought, were distinctive as compared to the usual characteristics of Cremolobus. Essentially the Same treatment was further elaborated by Schulz (1936), who ascribed a number of additional characters to Urbano- 156 KHANNA AND ROLLINS doxa that were purported to be different from Cremolobus. These included the presence of a gynophore nearly of the same length as the elongated conical style, the triangular valves not keeled or winged and the presence of radial nerves or wrinkles on the valve surfaces. An examination of all these characters in relation to the other species of Cremolobus shows that they all occur to a varying degree throughout the genus. None of them, either singly or in combination, provides the distinctiveness re- quired to support this taxon at the rank of a genus by itself. The leaves of C. rhomboideus are not always opposite as suggested by Muschler, they are opposite only on the lower portions of the stem or in the younger plants. Other species of Cremolobus are not characterized entirely by alternate leaves. Two very distinct species, C. bolivianus and C. subscandens, invariably have opposite leaves. The nectaries, though very small and inconspicuous in most species of Cremolobus, may be quite large in others, par- ticularly those having large flower size, as in C. peruvianus and C. suffruticosus. Most of the species of Cremolobus have the gynoecium raised on a thin gynophore which is nearly as long as the style. Furthermore, the style is coni- cal in C. suffruticosus, C. chilensis and C. stenophyllus. In some variants of C. chilensis, the distinction between the diameter of the mature style and that of the stigma is only slight. The triangular shape of the valves with radial veins and ridges and the complete lack of the wings is comparable to certain forms of C. chilensis with suborbicular triangular valves with almost no wings. The size of the fruit in C. rhomboideus is larger than in the wingless forms of ©. chilensis, and the texture is much harder, but these charac- ters alone hardly justify generic rank for C. rhomboideus. The weak segregation of Urbanodoxa from Cremolobus was pointed out by Macbride (1938). Actually, the dis- tinctive features of C. rhomboideus make it a clearly marked species, but the combination of features it possesses place it unmistakably with the other species here included in Cremolobus. SPECIES EXCLUDED ; A special comment on the name Cremolobus linearifolius H. & A. (Hook. Bot. Misc, 3: 138. 1833) is required. This name is based on Cuming 905 collected at Coquimbo, Chile; REVISION OF CREMOLOBUS 157 and a fragmentary specimen which apparently is the type is now in the herbarium of the University of Glasgow. A photograph of the type has been examined. The plants are somewhat comparable in habit to the smaller forms of C. chilensis with linear, ribbon shaped and deeply pinnatifid leaves. However, no complete fruits are present on the specimens. The valves are missing but the gynophore, the replum with two lateral points of valve attachment and the conical style could refer to either Menonvillea or Cremo- lobus. We cannot be absolutely sure of the generic identity of this taxon at the present time. However, we feel confi- dent, from the evidence discernable on the photograph of the type, that it is Menonvillea and not Cremolobus. The association of C. linearifolius with C. rhomboideus [Urbanodoxa rhomboidea (Hook.) Muschler] as suggested by Schulz (1936) is certainly not the proper disposition of the name. In Menonvillea, Cremolobus linearifolius could apply to the species we have referred to by the name M. Gayi Ph., or it might possibly be the same as M. chilensis (Turcz.) Jackson. The type is so fragmentary that a very careful examination will need to be made, and even then it may not be possible to fix the application of the name C. linearifolius with certainty. LITERATURE CITED Britton, N. L., 1889. An Enumeration of the Plants Collected by Pe Bo usher in South America. Bull. Torr. Bot. Club 16: 13-20. DE CANDOLLE, A. P., 1821. Reg. Veg. Syst. Nat. 2: 418-419. Lansouw, J. AND F. A. STAFLEU, 1964. Index Herbariorum, Fifth Edition. Rec. Veg. 31: 1-251. MACBRIDE, J. FRANCIS, 1938. Flora of Peru. Chicago Mus. Nat. Hist., Bot. 13(3): 665-1162. MuscHter, R., 1908. Cruciferae andinae in Urban, Plantae novae andinae imprimis Weberbauerianae III. Bot. Jahrb. 40: 267-277. Rotuins, REED C., 1955. A revisionary study of the genus Menon- villea (Cruciferae). Contr. Gray Herb. 177: 1-57. , 1958. The genetic evaluation of a taxonomic character in Dithyrea (Cruciferae). Rhodora 60: 145-152. soo O. E., 1933. Kurze Notizen iiber neue Gattungen, Sektionen Avion der Cruciferen. Bot. Jahrb. 66: 91-102. ———__________, 1936. Cruciferae in Engler and Prantl, Natiirlichen Pflanzenfamilien (ed. 2) 17b: 227-658. INDEX CREMOLOBUS acknowledgements 142 Biscutella chilensis 148 Biscutella cuneata 148 Biscutella peruviana 145 Biscutella suffruticosa 147 characters o u 136 Cremolobus aphanopteris 148 C. Benois’ 149 C. bolivianus 1438 C. chilensis 148 C. humilis 148 C. linearifolius 156 parviflorus 148 C. Paysonii 149 C. peruvianus 145 C. pinnatifidus 148 C. pinnatifidus var integrifolius 148 C. pubescens 145 C. rhomboidens 155 C. sinuatus 148 C. stenophyllus 153 C. subscandens 144 C. suffruticosus 147 C. Weberbaueri 149 flowers 139 generic relationships 135 inflorescences 139 138 mae stenophylla 153 pubes 138 ce pee 1-4, fig. 5-10) 137, 140, 141 taxonomic treatment Urbanodoxa rhomboidei 155 LEAVENWORTHIA acknowledgements 13 advancement index 69, 76 nthers, rot. 62 _ area of origin 12 auto-fertilization, percentage AT cedar glades, origin 13 chromosome numbers 11 description, L. alabamica & crassa dispersal capacity 92 distribution of races 82 distribution, L. alabamica & L. crassa 9, 10 uitedtivetiads of insect pollinators 115 evolution of characters 74 factors affecting evolution of breeding system field studies a flower color pattern 62 flower measurements 53 hybrid populations 88 inflorescence measurements 50 insects of pollination 108 nectar guides 62 ovules, no. per pone 58 phylogeny of race 79 ollen: ovule ee 58 pollination 106 pollinator change 124 no. of pollinator populations on cultivated land 85 populations of glades 90 race relationships bs races of L. alabamica races of L. crassa seed weight if self-compatibility, evolution self-incompatible and self- voucher specimens, deposited 6 Oe Ate Se oe CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reep C. Ro.tLiIns AND Ropert C, FosrTer NO. CXCVI THE TAXONOMY OF THE SALIX GLAUCA : COMPLEX IN NORTH AMERICA By GEORGE W. Arcus Published by ee THE GRAY HERBARIUM OF HARVARD UNEVEESIFY . CAMBRIDGE, MASS., U.S. A. 1965 CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reep C. ROLLINS AND ROBERT C. FOSTER NO. CXCVI THE TAXONOMY OF THE SALIX GLAUCA COMPLEX IN NORTH AMERICA By GEORGE W. ARGUS Published by THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U.S. A. Issued September 10, 1965 TABLE OF CONTENTS INTRODUCTION ACKNOWLEDGEMENTS THE SALIX GLAUCA COMPLEX: Characteristics, Composition, and Intrageneric Position TECHNIQUES Population Sampling Morphological Measurements Stomatal Measurements Cytology Systematic Materials and Methods VEGETATIVE MORPHOLOGY AND BIOLOGY ms Distribution of Stemata Preliminary Survey of Stomatal Distribution Intraindividual Variation Environmental Variation Stomatal Length sapien of Stomatal Length Environmental Influence on Stomatal Length Seed Bias atiGn and Seedling Development REPRODUCTIVE MORPHOLOGY AND BIOLOGY Inflorescence s Peaks and Seeds Pollen Pollination Correlation of Stomatal Length and Ploidy Level Geographic Variation in Stomatal Length SOME GEOGRAPHIC ASPECTS OF VARIATION IN SALIX GLAUCA Results Discussion KEY TO SPECIES KEY TO PHASES - Salix glauca KEY TO TAXA - Salix brachycarpa SALIX GLAUCA Synonymy of Salix glauca (Western Hemisphere) Discussion of Synonymy General Description of Salix glauca Beringia Phase Principal Synon Amplified Description Distribution Discussion Ecology Selected Specimens Western Phase Principal Synon Amplified Description Distribution Selected Specimens Rocky Mountain Phase Principal Synon Amplified Description a iscussion colo Selected Specim Selected Specimens ae “ Ferruginous Form, Rocky Mt. Phase Eastern Phase Principal Synonymy Selected Specimens Selected Specimens from Transitional Area Between the Western, Eastern, and Rocky Mountain Phases Salix glauca in Greenland Selected Specimens SALIX BRACHYCARPA Salix brachycarpa ssp. brachycarpa eetereeS of Synonymy Dese ponies Discussion Ecolo ted Specimens Salix brachycarpa ssp. niphoclada cee of Synonymy Deser fist ctea Discussion ected Specimen Salix Seckeures ssp. ahaa var. fullertonensis Discussion of Synon Description poeta Discussion Ecolo: Selected Specimens HYBRIDS Salix arctica ss oe ssp. brachycarpa g Salix brachycarpa ssp. : ee < candida Salix brachycarpa ssp. brachycarpa X chlorolepis alix brachycarpa ssp. brachycarpa X glauca Salix brachycarpa ssp. brachycarpa X turnorii Salix brachycarpa ssp. niphoclada XK glauca Salix glauca * padophylla Salix glauca X pedicellaris LITERATURE CITED THE TAXONOMY OF THE SALIX GLAUCA COMPLEX IN NORTH AMERICA THE TAXONOMY OF THE SALIX GLAUCA COMPLEX IN NORTH AMERICA? GEORGE W. ARGUS? The species in the Salix glauca L. complex have long been recognized as a source of taxonomic difficulties in northern botany, and more than one student has commented on the variation within the group and urged a detailed study. In i Flora of Alaska and Yukon, Hultén (1943) observed at “The many closely related forms of the S. glauca group - arctic and subarctic America, S. brachycarpa, S. glauca, S. glauca desertorum, S. cordifolia, S. anamesa, S. pseudo- lapponum and others are very much in need of thorough study.”’ The confusing variation occurring in these species was commented on by Polunin (1940) who noted that in arctic Canada intermediates occur linking S. glauca and S. cordifolia var. callicarpaea and “. . . a continuous series could probably be constructed connecting almost all the Salices of our region, ...” In 1959, Raup wrote “I do not think that the taxonomic realities in S. glauca or in the group of entities related to it will be understood until the full range of their variability is better known, and until we have some light on its causes.” One of the purposes of this study was to contribute to a knowledge of the variation in the Salix glauca complex, suggest some possible causes for the unusual variation pres- ent, and to introduce a measure of order into the classifica- tion of the American taxa, thereby starting a much nee study of this important group of circumpolar willows. 1Financial support for this study was provided by the following sources: The the Fernald Fund for Aretie Institu orth America e-nainet AINA-39 Field Study in Systematic Botany at vard Tiadearaiter established by the late unnewell; ational a of Health Training Grant, provided Universit ational Research f h et Ea g° 8 > ct =: @ o & 3 bt) + BS oO 5 3 ° ee Biology arvar Council of Canada Postdoctorate aggre at th ; rsity of Saskatchew This paper is ed on an unpublished thesis presented to Harvard University in pereene Reve ana of the requirements of the degree of Doctor of Philosophy. *Present address: W. P. Fraser Herbarium, University of Saskatchewan, Saskatoon, Skates, Canada 4 GEORGE W. ARGUS The question of whether “true” Salix glauca L. occurs in the Western Hemisphere has been considered by Raup (1931, 1943, and 1959), Polunin (1940), and Hultén (1943) and will not be considered in detail in this paper. These authors concluded that at least some of their American material was identical to S. glauca of Europe. The variation in the European material of S. glauca that I have studied is very similar to that in American material, and I am of the opinion that S. glauca is widespread in America. However, until comparative population samples are available, and further cytological study is done a discussion of this ques- tion is mainly speculative. The present study of Salix glauca, a circumpolar species ranging in the Western Hemisphere from Alaska to New- foundland and Greenland, and south in the Rocky Moun- tains to New Mexico, was restricted to North America for several reasons. First, this area was in particular need of study; second, it was relatively accessible; and third, the amount of herbarium material available from this area alone was overwhelming. I have been aware of the pitfalls of a restricted study of a widespread taxon and the inade- quacy of my familiarity with the Eurasian S. glauca and its relatives. For this reason, as much as any other, my taxo- nomic treatment of the circumpolar S. glauca differs from that of the American S. brachycarpa. In S. glauca the varia- tion has been described in detail and related to geography, but no formal taxonomic categories have been proposed, whereas, in S. brachycarpa infraspecific status has been assigned to several geographic variants. Throughout the study every effort was made to consider the taxa in terms of populations exhibiting genetical, onto- genetical, ecotypic, and ecophenic variation. The sampling of local populations in the field and the herbarium has con- tributed to an understanding of population variation, and the degree of ontogenetic variation to be expected in indi- viduals was studied in successive collections of tagged plants. The cultivation of cuttings has contributed to an under- standing of phenotypical plasticity, and several artificial hybridizations have pointed to the value of further such studies. More information for almost every aspect of the problem is needed and further studies in cytology, hybridi- SALIX GLAUCA COMPLEX . zation, and morphology are certain to modify and supple- ment the conclusions presented here. ACKNOWLEDGEMENTS Many people have provided assistance during the course of this study and it is a pleasure to formally acknowledge their numerous %, I am particularly indebted to Professor Reed C. Rollins, Director of the Gray Herbarium, under whose guidance this study was con- ducted, for his contributions of an abstract as well as a concrete nature, including a general philosophy of plant systematics, an atmos- phere conducive to independent study, a critical reading of the manuscript accompanied by provocative comments and Saal a rt. To Professor Hugh M. Raup, Director of the Harvard Forest, I express my sincere appreciation for his personal interest, and for an abundance of ideas and information. I also wish to thank my associates at the Harvard Herbaria for their suggestions ond stimulating peat Dr... We ¥. A. Barclay, Dr. R. C. Foster, especially for answering my endless nomenclatural Fahy Dr. O. Solbrig, Dr. and Mrs. R. Tryon, Dr. K. Wilson, and Dr. J. L. Thomas. My appreciation to the collectors of living material, Dr. C. L. Porter, the cultivation of this material by the Arnold Arboretum, and Mr. A Fordham, propagator, gave invaluable assistance and advice. I am indebted to the curators of the herbaria from which specimens were obtained, especially Dr. C. E. Kobuski, then sagen of the Harvard University Herbaria, now deceased, and Dr. W. Archer, National Arboretum Herbarium, for making yon "the C. R. Ball Herbarium. The other herbaria are listed in the section Tech- es, Finally, I wish to acknowledge my wife, Mary, for without her infinite patience and assistance this study would not have been completed. THE SALIX GLAUCA COMPLEX: ITS CHARACTERISTICS, COMPOSITION, AND INTRAGENERIC POSITION Characteristics. The species belonging to the Salix glauca complex can be recognized by the following combination of characteristics: low shrubs, 3-6 feet tall, or sometimes prostrate, with pubescent branchlets. Leaves glaucous and pubescent beneath and pubescent to glabrate above; margin 6 GEORGE W. ARGUS entire. Flowers arranged in coétaneous aments borne on short leafy reproductive shoots (peduncle of authors). Bracts of the flowers light brown to yellowish, pubescent. Staminate flowers with two stamens, filaments pubescent, subtended by an adaxial and an abaxial gland. Pistillate flowers with a pubescent ovary subtended by an adaxial gland. Composition. Fundamentally, the Salix glauca complex in the Western Hemisphere consists of two highly variable species, S. glauca and S. brachycarpa. A multitude of names has been applied to these two taxa and their various forms, including S. desertorum, S. glaucops, S. pseudolapponum, S. cordifolia, S. callicarpaea, S. villosa, S. anamesa, S. niph- oclada, S. muriei, and S. fullertonensis, and numerous infra- specific names. These names and others are discussed in the sections on synonymy. Other species including S. wol/it, its var. idahoensis, and S. eastwoodiae (incl. orestera) seem to be closely related to this complex and further study may include them. Intrageneric Position. To place the Salix glauca complex in its proper intrageneric position is difficult and hazardous because of the fragmentary nature of our knowledge of the affinities within the genus. The subdivision of Salix into sections is generally unsatisfactory, and many of the com- monly recognized sections are clearly artificial. However the following review of the historical development of the intrageneric classification of Salix will provide a background for further study. Throughout this review emphasis 1S placed on the Salix glauca complex, although related taxa will be discussed. Attempts to subdivide the genus Salix into intrageneri¢ units have great antiquity. Even prior to Linnaeus, who in 1737 and 1753 presented a classification based on stamen number and leaf characteristics, Theophrastus and Plinius proposed subgeneric groupings. Muhlenberg (1805) was the first author to treat the North American willows. In this treatment, which did not include species related to Salix glauca, he subdivided the genus into two groups, those with leafy peduncles, and those with naked peduncles. SALIX GLAUCA COMPLEX 7 Between 1825 and 1828 “natural” classifications were proposed by Dumortier, Fries, and Koch. Dumortier (1825) described five sections and grouped them into two series based on, 1) stamens coming from the side of the nectary, and 2) stamens coming from the center of the nectary. Fries presented his first classification of Salix in 1825. It was followed in 1828 by a second treatment, assumed to be the same as his 1825 treatment (Toepffer, 1915), in which he based his primary classification on whether the capsules were sessile or pedicellate, with a further sub- division relating to the height of the plants. In this treat- ment tribe Glaucae included S. lanata L., S. glauca, S. limosa Wahl., and S. versifolia Wahl. The same year, but after Fries, Koch (1828) classified the European willows using as primary subdivisions, 1) sessile aments produced at the tip of a branchlet, with leaves from buds below the aments; 2) “pedunculate” aments produced from the tip of the previous years branchlets, peduncle increases in size and bears leaves with buds in the axils; and 3) terminal bud, and generally others beneath it, pro- ducing vegetative shoots, the aments originating beneath these. Further subdivision was based on bract-color, an- ther-color, and pedicel-length. His cohors Frigidae included S. limosa, S. glauca, S. pyrenaica Gouan, S. waldsteiniana Willd., S. prunifolia Sm., S. caesia Vill., S. myrsinites L., and S. jacquinii Hort. y Hooker (1830) in his British Flora used a classification suggested by Borrer. In it Salix was divided into equal units previously used by Dumortier, Fries, or Koch ; how- ever, some were originated by Borrer. Three species, S. glauca, S. arenaria L. and S. stuartiana Sm., were included in Glaucae. No rank was assigned to these groups ; however, later authors assumed they were sections (Toepffer, 1915), and in some treatments Borrer’s name appears as the author of these “sections”. = Fries in 1832 and again in 1840 revised his original classi- fication, drawing on the work of Dumortier and Koch. In 1832 he erected tribe Amerina (sect. Amerina Dumortier) and included in it species with pedunculate aments borne on lateral twigs. This included: S. pentandra L., S. amygda- 8 GEORGE W. ARGUS lina L., S. fragilis L., S. viridis Fries, S. alba L., S. glauca, and its varieties appendiculata Wahl., nivalis Wahl., and pallida Fries. The artificiality of this tribe was recognized and in 1840 he again modified his classification. Salix glauca was placed in Chamelix-Frigidae (rank undesignated) with S. glauca var. pullata Laest., S. glauca var. pallida, S. arbus- cula L., S. myrsinites L., and S. pyrenaica var. norvegica Fries. Trautvetter in 1833 expanded Koch’s cohors Frigidae by combining it with cohors Chrysantheae. The following names were included in this emended taxon: S. arenaria, S. lanata, S. glauca and vars. microcarpa Ledeb., and macrocarpa Ledeb., S. arctica Pall., S. myrsinites, S. prunifolia Sm., S. stbirica Pall., and S. brayi Ledeb. The willows of North America were arranged in groups by Barratt in 1834, but this treatment was unpublished until 1838 when part of it was used by Hooker in Flora Boreali- Americana. Finally, in 1840 a revised and shortened edition of this classification was published by Barratt. His classi- fication closely followed that of Borrer, but several new groups were erected to contain American species. In 1836 Trautvetter radically revised his infrageneri¢ classification of Salix. He divided the genus into two taxa Chrysolepideae and Allolepideae based on bract-color. With- in the former he recognized two subdivisions based on bract-persistence and within the latter he recognized four subdivisions based on leaf-texture and shape. The grouP Platyphyllae included S. glauca, S. glauca var. melanolepis (Ledeb.) Trautv. S. phylicifolia L., S. hastata L., S. arbus- cula, S. arenaria, S. aurita L., and S. caprea L. Before Barratt’s manuscript on the North American wil- lows was published it was used extensively by Hooker (1838). Names which are now regarded as part of Salix glauca, S. desertorum Richards. and S. cordifolia Pursh, were placed in different sections. Salix desertorwm Was placed in sect. Arbusculae, small shrubs of the arctic and alpine, with coétaneous aments; and S. cordifolia in sect. Prostratae, low shrubs of arctic and alpine areas, aments coétaneous and pedunculate. Both groups contained um related species. SALIX GLAUCA COMPLEX 9 In his Flora Rossica, Ledebour (1850) gave the cohors of Trautvetter the rank of section for the first time. In sect. Frigidae, he included S. lanata, S. lapponum L., S. glauca, S. reptans Rupr., S. arctica, S. myrsinites, S. ovalifolia Trautv., S. arbuscula, et al. In 1858 N. J. Andersson treated the North American wil- lows following the 1840 classification of Fries. In Chamae- lix-Frigidae he included S. glauca, its var, villosa Hook., S. desertorum, S. arctica R. Br., S. subcordata Anderss., S. alpestris Anderss., S. myrsinites, S. pseudomyrsinites Anderss., S. curtifolia Anderss., S. arbuscula, S. rhamnifolia Hook., S. ovalifolia, and S. glacialis Anderss. In 1867 An- dersson began a revision of the classification of the willows of the world which was completed in De Candolle’s Prodro- mus in 1868. In this treatment the classification was based primarily on stamen-number and secondarily on style-length, and geographic distribution. Salix glauca and its relatives were placed in Niveae s. Glaucae c. Sericeae Anderss. In- cluded in this group were S. glauca and its vars. sericea Anderss., pullata, virescens Anderss., alpina Anderss., S. xX glaucops Anderss., and its var. villosa (Hook.) Anderss., and var. glabrescens Anderss., S. desertorum and its vars. elata Anderss., var. stricta Anderss., and fruticulosa An- derss., S. lingulata Anderss., S. pyrenaica, S. reptans, S. subcordata, and various hybrids. Be Wichura in 1865, after a lengthy dissertation on hybridi- zation in the genus Salix, proposed a classification which “ . . brought the true species into natural groups.” His general classification did not find much support but his use of the nectary-number as a diagnostic characteristic did. Nectary number was used by Otto von Seemen (1903) in his treatment of Salix in Salices Japonicae. His first division was based on the number of nectaries in the stam- inate and pistillate flowers; further subdivision was based on the number of stamens, whether they were united or not, and on style-length. His infrageneric classification was used by C. Schneider (1904) in his Handbuch der Laubholz- kunde, and by A. Toepffer (1915) in Salices Bavariae. The use of nectary-number by Schneider resulted in the un- natural separation of S. desertorum from S. glauca. Salix 10 GEORGE W. ARGUS glauca, S. pyrenaica, and S. arctica were placed in sect. Sericeae Koehne (Deutsche Dendrol., 1893) and S. desert- orum was placed in sect. Lanatae Koehne subsect. Villosae Anderss. along with S. lapponum and S. helvetica Vill. In the late 19th and early 20th century the taxonomy of North American Salix received an unprecedented amount of attention. Prominent among the students of the genus were M.S. Bebb, P. A. Rydberg, C. R. Ball, and C. Schneider. In Coulter’s Manual of the Botany of the Rocky Mountain Region, M. Bebb (1885) presented a synoptic key to Salix. Under the headings, “Aments lateral or terminal with or without bracts: scales persistent, usually darker at the tip: stamens 2; filaments glabrous - capsules tomentose - pedicels short or none - styles distinct’? he included the following names: S. chlorophylla Anderss., S. candida Willd., S. glauca var. villosa, S. desertorum, and its var. wolfii Bebb, S. are- tica R. Br., and its var. petraea Anderss. Rydberg (1899) was disturbed by Bebb’s conservative approach to the species of western America and felt that, “S. desertorum and S. reticulata, as understood by him (i.¢. Bebb), consist . . . of at least three species each.” Rydberg was of the opinion that characteristics of significance to Salix were not being properly regarded. He stated “.. - if characters generally accepted ag good by European salicolo- gists should be applied to American species, . . . the number of our native species would be doubled.” In applying this reasoning to the arctic and Rocky Mountain species Ryd- berg multiplied the American species of Salix at a rapid rate. Most of his taxa have since been reduced to synonymy or to infraspecific rank. In his 1899 treatment he described a hew group, Arcticae, to include “Caespitose willows, - - - with entire-margined leaves, catkins at the ends of short leafy branches, appearing with the leaves, more or less densely white-tomentose or villous capsules and an evident style.” This group was very large primarily due to the many new species described by Rydberg. Undoubtedly several of the taxa included in the Arcticae do not belong there, but he was probably correct in including the S. glauca complex and the S. arctica complex in the same group. : In his Flora of Colorado Rydberg (1906) added Salut chlorophylla, and S. pseudolapponum v. Seem. to the Are SALIX GLAUCA COMPLEX 11 ticae, and changed the name of S. stricta to S. brachycarpa Nutt. Three years later C. Ball (1909), treating Salix for Coul- ter and Nelson’s Flora of the Central Rocky Mountains, used Rydberg’s sub-generic units, adding S. candida to the Are- ticae and transferring S. chlorophylla to the Phylicifoliae. Rydberg (1917) expanded his concept of the Arcticae in the Flora of the Rocky Mountains and the Adjacent Plains to include S. idahoensis (Ball) Rydb., S. subcordata, S. austinae Rydb., S. maccalliana Rowlee, S. saskatchewana, S. cascadensis Cock. and S. nelsonii Ball. During 1912 and 1914 treatments of Salix were published which provided rank for many intrageneric units that had been used in the past without designated rank. The first was Frye and Rigg (1912) who assigned the rank of sub- genus to Rydberg’s sub-generic units, and the second was C. Moss (1914) who, in the Cambridge British Flora, named four sections, Amerina, Chamaetia, Vetrix, and Vimen, and described series within each of them. Series Glaucae was used to include the species of the British flora related to S. glauca. This treatment was followed by Rehder in 1949. C. Schneider, a German citizen interned in the United States during World War I, was permitted to continue his botanical studies at the Arnold Arboretum under the super- vision of C. Sargent. From 1918 to 1921 he published his outstanding monograph of the American willows, a treat- ment which was to provide the basis for much of the later work on American Salix. In 1921 he summarized his classi- fication, which had appeared previously in twelve separate papers, and noted that stamen-number provided a basis for recognizing two large natural groups within Salix. One of these groups he named Pleiandrae and the other was left unnamed. No rank was assigned to these groups. The section was his primary sub-generic taxon. Schneider’s work has played a prominent role in the formulation of our present day concepts of the genus. However, it should be noted that his idea of a section was a narrow one and often corres- ponds to the species complex. He wrote “. . . I thought it best not to unite species of apparently no close affinity in the same group but to propose new sections for those species which show good characters.” In his treatment of S. glauca 12 GEORGE W. ARGUS and related species (Schneider, 1918b) he used sect. Glaucae to include the species of the S. glauca complex and sect. Ovalifoliae to include the S. arctica complex. In the discus- sion of differences between Glaucae and Ovalifoliae he noted the difficulties involved but felt that the bracts were of diagnostic significance. He characterized sect. Glauca as having bracts uniformly yellowish, light-brown or straw- colored compared with the bicolored bracts of sect. Ovali- foliae which tend to be pale at the base and dark-brown, fuscous, or even black at the apex. Furthermore, the pubes- cence of the bracts in Glaucae was shorter, less straight and rarely distinctly silky in contrast to the long straight, silky hairs on the bracts of the Ovalifoliae. Flederus (1931) in Holmberg’s Skandinaviens Flora proposed a classification utilizing sections and series of Du- mortier to divide the genus into three groups, Chamaetia, Caprisalix, and Amerina. Under Caprisalix he listed Glaucae Fries which included S. glauca, S. stipulifera Flod. and a host of hybrids. This classification was followed closely by Nasarov (1936) in the Flora USSR. Nasarov assigned subgeneric status to the three main groups used by Floderus and sectional rank to their subdivisions. In section Glaucae he included the following species: S. glauca, S. stipulifera, S. seemannii Rydb., S. altavica Kar. & Kir., and S. reptans. He also used sect. Arcticae of subgenus Chamaetia to in- clude S. arctica, S. torulosa Trautv., S. pallasii Anderss., S. pulehra Cham. and S. divaricata Pall. Hultén (1943) followed essentially the same sectional classification as Kom- arov and placed the species related to S. glauca in sect. Glaucae and those related to S. arctica in sect. Arecticde- He correctly transferred S. pulchra from sect. Areticde, where it was assigned by Nasarov, to sect. Phylicifoliae. In his studies of Saliz, Raup (1943, 1959) has regarded the formal sectional category with some skepticism. In 1943 he wrote “Neither in the arrangement of the list nor in the keys have I indicated sectional subdivisions of the genus: With our present understanding of the relationships, past and present, among the species, it is somewhat hazardous to do so.” However, he did give a list of the tentative sec tional disposition of his species. In 1959 he made a furtl er break with the use of formal infrageneric taxa by placin’ SALIX GLAUCA COMPLEX 13 the taxa into groups named after the major species of the group. Many of these were species complexes. However, despite the reluctance to use sectional categories Raup did list informal categories in the general order of the sections as they were used by C. Schneider. This skepticism seems justified in view of the diversity of opinion concerning the alignment of species within Salix and the lack of reliable diagnostic characters on the subgen- eric level. The characters used as diagnostic for subgeneric taxa have included habit (Hooker), branching pattern (Koch), nectary-number (Wichura), bract-color and -persistence (Trautvetter), stamen-number (Andersson) ef al. These criteria generally have been inadequate for the purpose either because of their variability or because of in- adequate study. A complete subgeneric revision of Salix based on thoroughly studied meaningful characters is much needed. The infrageneric rank of the species related to Salix glauca may be treated in two general ways. They may be placed in a narrowly conceived sectional group which should be named Salix sect. Glaucae (Fries) Schneider, or in a broadly con- ceived section named Salix sect. Frigidae (Koch emend. Trautv.) Ledebour. The treatment depends on whether or not the author believes Salix glauca and S. arctica should be placed in the same section. There is precedent for each view, with Trautvetter, Ledebour, Bebb, and Rydberg plac- ing these species in the same group and Andersson, Schnei- der, Floderus, and Hultén placing them in different groups. In my opinion the broadly conceived section, Salix sect. Frigidae, is the more natural taxon. However, my earlier statement on the difficulty of making infrageneric assign- ments and our lack of information remains pertinent. TECHNIQUES Population Sampling. Two types of population-samples were used to analyze variation, 1) the field local population, and 2) the “herbarium local population.” In both types of populations the unit sampled was defined in terms of a taxon, a geographic area, and the seasonal development of reproductive structures. In the field local population samples habitat was an additional criterion. 14 GEORGE W. ARGUS The field local population of a taxon was sampled in a small geographic area of relatively homogeneous vegetation and habitat. A single shoot bearing reproductive structures was selected from each of about 100 plants. Randomness was achieved by the use of a grid in which a sample was taken every 25 paces (used at Churchill, Manitoba and Lake Athabasca, Sask.) or by “unselectively” collecting a sample as the taxon is encountered in walking through an area. The latter method is less objective than the former, but it is applicable to areas in which the taxon is sparsely distributed or where time is limited. Such local population samples are of the “mass collection” type described by Anderson (1941). The “herbarium local population” was sampled by assem- bling herbarium specimens of a taxon from a particular eographic area and selecting out all specimens in a com- parable stage of ontogenetic development. In this study each specimen was in anthesis or early postanthesis and bore mature leaves. No attempt was made to segregate speci- mens on the basis of habitat and the range of habitats repre- sented in each set of samples approximated the range of local habitats occupied by the taxon. The size of the geo- graphic area and the number of specimens studied varied with the intensity of local collecting. However, the geog: raph- ic area was kept as small as possible and the sample size as large as possible. The value of the “herbarium local population” in deter- mining the mean and range of variation in a taxon can be illustrated by comparing the variation based on a herbarium- sample of 27 specimens from Churchill, Manitoba (popula- tion 15., figs. 21, 23, and 25) with field local population samples of 100 specimens from the same area (figs. 1 an 9). The degree of variation and the estimated mean values are similar in each case and for the purposes of this study are not significantly different. Morphological Measurements. In order to obtain com- parable measurements an effort was made to measure organs from the same position on the plant and in the same stage of development. Measurements of leaf-blade, petiole, and stipule dimensions were based on the largest mature leaf on the shoot. Measurements were made using a milli- meter scale without magnification, except in the case of stipule measurements and the short petiole in Salix brachy- SALIX GLAUCA COMPLEX 15 carpa in which 9X magnification was used. Ament and flower dimensions were measured using a millimeter scale and 9X magnification. Ament length was measured from the base of the lowermost flower to the apex, and width was measured near the middle of the ament. Ovary, capsule, pedicel, and bract-length were measured on several] flowers selected from near the middle of the ament. Stomatal Measurements. In the stomatal studies leaves were examined from dry herbarium specimens. In each case one or two mature leaves were selected from the distal end of the shoot. In two specimens all the leaves from several annual shoots were removed and studied. The leaves were cleared and stained using a modification of the tannic acid — ferric chloride stain (Foster, 1934) as described by Wilson (1958). The clearing was accomplished by treating the leaves in 5% aqueous sodium hydroxide until discolored and then bleaching in 50% aqueous “Clorox.” The leaves were washed thoroughly and then treated in 5% tannic acid in 30% ethyl alcohol for about 5 minutes, rinsed and then treated in 5% iron chloride in 30% ethyl alcohol for about 5 minutes, The leaves were then dehydrated and mounted in diaphane. Trichomes were removed by gentle scraping during the 95% dehydration stage. The length of stomata was determined by measuring 10 to 20 abaxial stomata located near the center of one half of the blade. Measurements were made using a 40% Wild Fluotar objective and a 15x ocular with an ocular microm- eter. Drawings of representative epidermal sections were made using a camera lucida and a 44 objective and a 15x ocular. ‘ The Dice-Leraas method of graphically comparing the mean, standard deviation, standard error, and the range of several samples is used in fig. 2. For a modification and a discussion of this method refer to Hubbs and Perlmutter (1942) and Critchfield (1957). The symbols used are: the range, a thin horizontal line; the mean, a vertical line; one standard deviation on either side of the mean, a wide stip- pled bar; and 2 standard errors on either side of the mean, a solid black bar. ; An indication of the “significance” of the differences be- tween any two means can be obtained by Peers Fe overlap or separation of the two standard errors on either 16 GEORGE W. ARGUS side of the mean. The difference between the means may be considered to be “significant” if the standard error bars do not overlap. The variable sample-size undoubtedly influences the accuracy of this method, and the interpretation of “sig- nificance” must be considered with some reservations. Cytology. Chromosome counts were based on root-tip squash preparations which were fixed and stained essentially as described by Live and Sarkar (1956). Root-tips were excised and pretreated in saturated aqueous paradichloro- benzene for 3 hours. They were washed and then fixed in 6:3:1 (6 parts absolute alcohol: 3 parts chloroform: 1 part propionic acid) for 12-24 hours and then transferred to 20% ethyl aleohol, The squash and staining process involved the treatment of the material in 45% acetic acid for 1 hour followed by 5-10 min. in equal parts of concentrated HCl and 95% ethyl alcohol. One to two millimeters of the root- tip was excised onto a microslide and stained with ferric acetocarmine. Following gentle heating, the material was firmly squashed under a cover glass. Slides containing good mitotic figures were made permanent and finally studied under a microscope with a 90 X oil-immersion objective and 15 X oculars. Photomicrographs aided in making the counts and the preparation of drawings. The sectioning of feulgen stained root-tips was tried, but good preparations were not obtained. Systematic Materials and Methods. The systematic treat- ment is based on field and herbarium study. Field work was carried out in central Alaska and the Alaska Range; Church- ill, Manitoba; Mt. Albert, Quebec; and northern Saskatche- wan. Valuable recent collections from the Bering Sea coast of Alaska (Johnson, Viereck, and Melchior) and the Great Slave Lake region (Thieret and Reich) were placed at my disposal by the collectors, and undistributed collections from the Mackenzie Mountains and the Alaska Highway were made available by Prof. H. M. Raup. Special collections of living material from Wyoming (C. L. Porter), Colorado (Levi), Manitoba, and Quebec (Argus) were made for this study. Herbarium specimens were studied from the following herbaria (the abbreviations are according to Lanjouw and Stafleu (1959) except for GWA, MTSM, and RMNP): Arnold Arboretum (A) ; University of Alaska (ALA) ; National Museum of Canada (CAN); Science Service, Dept. of SALIX GLAUCA COMPLEX 17 Agriculture, Ottawa (DAO); Dudley Herbarium, Stanford (ps) ; Chica- go Natural History Museum (Ff), Thieret and Reich collections only; Gray Herbarium (GH); University of Minnesota (MIN); Missouri Botanical Garden (MO), type only; Michigan State University (Msc), Petrides collections only; Musée de 1|’Institution des Sourds-Muets, Montreal (MTSM), one specimen; United States National Arboretum (NA); New York Botanical Garden (NY); Rocky Mountain Herbar- ium, Laramie (RM); Rocky Mountain National Park Herbarium, Banff, Alberta (RMNP); University of Saskatchewan (SASK); Uni- versity of British Columbia (UBC); United States National Herbarium (Us) ; University of Wisconsin (Wis) and the author’s personal collec- tion (GWA). Approximately 7000 specimens were studied, ange-maps were plotted on Map 202 of Goode’s Series of Base Maps, published by the University of Chicago Press. The maps of the phases of Salix glauca and the infraspecific ‘taxa of S. brachycarpa are for general graphic purposes only, and the limits of the units are only approximate. The specimens cited here have been selected for a variety of reasons. They may be historical specimens, unusual specimens worthy of note, specimens at the edge of the range of the taxon, specimens referred to in the text, or measured in the geographic variation study of S. tioned above. The specimens from Alaska and Yukon are arran in the districts proposed by Hultén (1941). In other areas political sub- divisions have been used. VEGETATIVE MORPHOLOGY AND BIOLOGY Stems. The stems of the Salix glauca complex are highly branched to form shrubs of variable stature. The tallest shrubs occur in S. glauca, which may attain a height of 10 feet. However, the same species may be prostrate under cer- tain environmental conditions, illustrating the great plastic- ity of this characteristic. At Churchill, Manitoba, prostrate individuals of S. glauca were observed on a ridge-top exposed to strong northwest winds and a few yards away, in the lee of a boulder, an erect individual was growing as high as the boulder that was protecting it (figs. 44 & 45). Although such modification is common, one taxon in this complex, S. brachy- carpa ssp. niphoclada var. fullertonensis, seems to be habitu- ally prostrate, and S. brachycarpa ssp. niphoclada is a decumbent. In general, the species of the S. glauca complex are erect shrubs about 2 to 4 feet tall. 18 GEORGE W. ARGUS Shoot development in Salix is commonly sympodial ; how- ever, in the seedling stage and immediately following the first dormancy monopodial development has been observed (Sugaya, 1956a and 1956b). Monopodial growth was observed in Salix glauca seedlings from Churchill (see dis- cussion of seed germination and seedling growth). The in- ternode-lengih is usually greater in S. glauca than in S. brachycarpa, giving the shoots of the latter species a charac- teristic “fan-like’ appearance. However, this is a variable feature that requires further study. The annual shoots (branchlets) are usually clothed with a pubescence varying from a dense woolly mat in Salix brachycarpa ssp. brachycarpa to thinly pubescent or glabres- cent in some S. glauca. The trichomes are lost in time and after 2 and 3 years the branches are usually glabrescent. The branchlets vary in color from light green to dark red- dish-brown. The branches are usually reddish-brown or grayish, although in some they are yellowish or black. The older branches usually have a grey exfoliating epidermis. Branchlets may be pruinose, but observations of the pruinose condition are incomplete due to the dense pubescence on some stems and due to the loss of pruinosity during the drying of specimens. Roots. The roots of the Salix glauca complex are woody structures often considerably branched, the main secondary branches running parallel to the surface of the ground a few inches beneath it. The depth of the roots is probably often controlled in arctic and subarctic regions by the depth at which permafrost occurs. The possibility of vegetative shoots originating from the roots was investigated in the field. Both Salix glauca and S. brachycarpa were excavated at Churchill, Manitoba. An individual of S. glauca growing on the outcrop ridge was excavated until about 9 feet of roots were exposed, and an individual of S. brachycarpa was excavated, exposing three secondary roots of 5, 11, and 15 feet each. In neither case were root-shoots found to occur. On the Gaspé Peninsula 4 plant of S. argyrocarpa Anderss. growing in stream-8t avels was also excavated with similar results. Some species of Salix, including S. interior Rowlee and S. exigua Nutt., do have vegetative shoots sprouting from the roots and this question must be investigated separately for SALIX GLAUCA COMPLEX 19 each species. In the species included in the present study, sprouts originate only from stems, most frequently from the base of the main shoot. Leaves. The leaves of the Salix glauca complex are dorsi- ventral structures ranging in shape from oblanceolate to narrowly lanceolate through elliptical and obovate to oval or suborbicular in some forms of S. glauca. Leaf-shape is of general taxonomic importance in spite of its wide varia- tion within populations. The leaf-base generally varies from round to obtuse to acute or attenuate, with one extreme form being cordate or subcordate. The apex similarly runs the gamut of variation from obtuse or rounded through acute or even apiculate in some instances. The general shape of the leaf-extremities depends somewhat on the shape of the leaf, with the broader leaves having rounded or obtuse extremities and the narrower leaves the more tapering ex- tremities. The leaves also vary in size and shape according to their position on the shoot. The lowermost leaves (proximal) are more or less modified in shape and pubescence and mature very rapidly, almost as soon as the shoot begins to elongate and the leaves unfold. In Salix brachycarpa ssp. brachycarpa (at Churchill) there are usually 3 to 4 such modified leaves which have parallel sides and a rounded apex and an obtuse or acute base. They lack, or have very scattered, pubescence on the upper leaf-surface even when enclosed in the bud. This is in marked contrast to the uppermost (distal) leaves which are completely clothed with a dense pubescence in the bud. In S. glauca (at Churchill) there are only 2 or some- times 3 modified proximal leaves which are similar in shape to those of S. brachycarpa, but with an acute or apiculate apex. They have long straight trichomes on the lower leaf surface and lack vesture above; the distal leaves are general- ly sparsely pubescent on both surfaces. ; The leaf-blade in the Salix glauca complex is always glaucous beneath and variously pubescent on both surfaces. The trichomes vary in length and degree of wayiness. They originate as an outgrowth of a single modified epidermal cell and remain a single cell. The base of the trichome cell . surrounded by a radiating series of 4 to 6 modified epiderma cells (see figs. 3-3 and 4-8). There are no apparent taxo- 20 GEORGE W. ARGUS nomic differences in this characteristic within the S. glauca plex. In Salia the stomata are generally of the rubiaceous type, in which two subsidiary cells flank the stomatal pore parallel to the long axis (Metcalf and Chalk, 1959). However, varia- tion in this pattern is common (see figs. 4-2, 4-3, and 4-9). Stomata are abundant in the abaxial (lower) epidermis and may or may not be present in the adaxial epidermis. The size of stomata varies with the species. The length of sto- mata in S. brachycarpa ranges from 17-22.5 micra and in S. glauca from 19.8-33.4 micra (see fig. 2). The import- ance of the presence or absence of stomata in the adaxial epidermis and stomatal size will be discussed below. The leaf-margin is generally entire, although often the proximal leaves have small glandular teeth. In some cases rot the distal leaves have small teeth near the base of the ade. I+ og ° 4 2-0 Geng +o e fc 4-0 ° ° e + ee ry ee ° e * ait’ 2 e 2 ee a 2 e ee 4 eels e @ee0 e880 @ ® ne rg °* @@@® eee. @ IT © @20@ e@208@ e8 8 © zy bd * @® @6e @ 54 e e208 e208 ° @ ° ee a ° @otT * e® 0@ 00 0 0 0 pee O09 oo ° ° o + ° ° 0000 Oo: nc Oo a 200 000 Oo00000 o I+ ° ©9000000 COoo ' ' , ' PCa a 15 2.5 3.5 45 5.5 Leaf Length (cm.) correlation of petiole-length and leaf-length in Salix glauca and S The e and S. oe (open circles) in the adjacent willow-shrub community. js these n peti are readily pseesineea Be on the basis of a correlation betwee length and leaf-length. SALIX GLAUCA COMPLEX 21 The leaves are always petiolate; however, in Salix brachy- carpa the petiole may be very short. The length of the petiole is a relatively constant characteristic and is of taxonomic importance in the S. glauca complex. The correlation be- tween petiole-length and leaf-length is shown in fig. 1. These characters are of diagnostic value in the species studied. A pair of glandular-margined stipules flanks the base of the petiole. The stipules are usually prominent although variable in shape, and may be reduced to small glandular processes, Their shape and size vary in different species and in relation to the vigor of the shoot. The time of abscis- sion of the stipule has been regarded by Raup (1959) to be of diagnostic importance, and Salix glauca var. perstipula was characterized by stipules persistent for several years. I have not yet studied this characteristic exhaustively, but I suspect that stipule-persistence is often related to the en- vironment and is of limited taxonomic value. Distribution of Stomata. In Salix stomata always occur in the abaxial leaf-epidermis. However, in some species or groups of species, stomata also occur in the adaxial epider- mis, and a study of the distribution of stomata is primarily concerned with this variation. When stomata do occur in the adaxial epidermis they may be either distributed gener- ally over the surface or localized at the extreme apex. Ad- axial stomata are characteristically larger and more widely spaced than abaxial stomata. ne ae Sectional importance was attributed to the distribution of stomata by A. and E. Camus (1904), and they described it as a fixed characteristic. It was used by these authors as the primary dichotomy in their anatomical key to the sec- tions of Salix of Europe. Their section Frigidae (which in- cluded S. glauca) was described as lacking adaxial stomata. During the years in which Schneider studied the Salix of North America (ca. 1918-1921) he paid particular attention to this character, as his annotations on herbarium-specimens and his comments in the literature testify (Schneider, 1918a, 1918b, and 1919). However, he found extensive intrasec- tional variation in the distribution of stomata and reduced the use of it from sectional to specific application. Schneider (1918b) utilized the distribution of stomata to distinguish some closely related species, e.g. S. brachycarpa from S. 22 GEORGE W. ARGUS pseudolapponum (= Rocky Mt. phase of S. glauca). He did note the occurrence of intraspecific variation but he did not elaborate upon it. Since Schneider’s work, no taxonomic use has been made of stomatal distribution in Salix, although both M. L. Fernald and H. M. Raup considered it and found it to be undependable (Raup, pers. comm.). In this examination of the taxonomic value of stomatal distribution a preliminary survey of 8 taxa has been made, and two possible sources of error are considered, Three of the eight taxa, Salix chlorolepis Fern., S. arctica, and Si arctophila, are not members of the S. glauca complex but they are closely related and may at times be involved in hybridization with members of the complex. The results of this preliminary survey are presented in Table I. In each specimen the largest distal leaf on a shoot was examined and the distribution of adaxial stomata noted. The following classes of stomatal distribution in the adaxial epidermis were recognized: 1) adaxial stomata absent, 2) adaxial stomata present at the extreme apex, and 3) adaxial stomata present (i.e. generally distributed over the surface). TABLE I, DISTRIBUTION OF STOMATA Taxon Number of Adaxial stomata’ specimens Absent Apex Present only S. brachycarpa ssp. brachycarpa 10 7 2° ssp. niphoclada var. niphoclada 3 4 b 0 var. fullertonensis 0 1 3 S. brachycarpa ssp. brachycarpa S. chlorolepis 7 1 3 3 S. chlorolepis 4 0 3 1 S. glauca (Europe and Asia) % 3 4 0 (North America)* 38 26 12 0 (Colorado and Wyoming) 5 0 1 4 S. arctica 11 0 0 11 S. arctophila 2 1 1 0 1All specimens have stomata in the abaxial epidermis. 2Specimens from Mt. Albert, Quebec. *Excluding specimens from Colorado and Wyoming. The presence of a few adaxial stomata at the extreme apex of the leaf is part of individual variation occurring 1n species which otherwise lack adaxial stomata on mature leaves. In SALIX GLAUCA COMPLEX 23 this preliminary survey, specimens having adaxial stomata at the apex (class 2) are tabulated separately, but in the discussion they are grouped with either class 1 or class 3, whichever is the predominant type in that taxon. An examination of Table I reveals that, although there is considerable intraspecific variation in stomatal distribu- tion, two general groups of species can be recognized. One, those lacking adaxial stomata, and two, those having adaxial stomata. Both groups include some individuals with adaxial stomata localized at the extreme apex. In general the Salix glauca complex is characterized by the absence of adaxial stomata, and S. arctica is characterized by the presence of adaxial stomata. However, Salix brachycarpa ssp. nipho- clada var. fullertonensis and the Rocky Mt. phase of S. glauca possess adaxial stomata and S. arctophila, sometimes considered conspecific with S. arctica (Drury, 1962), lacks adaxial stomata. Although general patterns are suggested and exceptions to these patterns can be observed, general taxonomic use of the characteristic is not warranted until two important sources of error, individual variation and environmental variation, are considered. Individual variation. The correlation of the distribution of stomata and the developmental position of leaves on the annual shoot was studied in two species. The leaves from three shoots of Salix brachycarpa ssp. brachycarpa (Argus 499-58, Churchill, Manitoba) and S. glauca (Argus 918- 58, Churchill, Manitoba) were removed, numbered succes- sively from the base, cleared, and stained. Each leaf was examined for the distribution of adaxial stomata. The fol- lowing is a summary of the data obtained: Salix brachy- carpa — leaves 1, 2, and 3, adaxial stomata present; leaf 4, adaxial stomata at extreme apex; leaves 5 through 8, adaxial stomata absent. Salix glauca —leaves 1 and 2, adaxial stomata present; leaves 3 through 5, adaxial stomata at ex- treme apex; leaf 6, adaxial stomata absent. : Although both of these taxa are generally characterized by an absence of adaxial stomata they may sometimes have a few adaxial stomata at the leaf-apex. The 2 or 3 proximal leaves which have been observed to have adaxial stomata are modified in shape and size from the distal leaves and 24 GEORGE W. ARGUS would not ordinarily be selected for use in a taxonomic study. However, this individual variation does emphasize the importance of using comparable organs and points out that the generalization that these taxa lack adaxial stomata is true only in reference to the distal leaves on the shoot. This variation may also explain the tabulation of some speci- mens of these taxa as having adaxial stomata at the leaf- apex. The leaves studied may have been selected from the intermediate zone between the proximal and the terminal leaves in which adaxial stomata are usually found to occur at the leaf-apex. Environmental Variation. It was noted by A. and E. Camus (1904) that the distribution of stomata in the leaves of Salix varied in different habitats. However, the nature and extent of this variation was not discussed. Three cases can be cited in which different stomatal distribution patterns are apparently correlated wth environmental differences. In each case a comparison is made between leaves from the same individual under natural field-conditions and under cultivation at Jamaica Plain. Massachusetts. 1) Salix glauca (Argus 501-58, Churchill, Manitoba). Field: adaxial stomata at apex. Cultivated: adaxial stomata absent. 2) Salix glauca (Levi 3, Gothic, Colorado) Rocky Mt. phase. Field: adaxial stomata present. Cultivated: adaxial stomata absent. 3) Salix chlorolepis (Argus 84-59, Mt. Albert, Quebec). Field: adaxial stomata present. Cultivated: adaxial stomata at apex. In each case adaxial stomata are absent or of more Te stricted distribution under cultivation than under natural field-conditions. More examples can be cited in which no difference is apparent between specimens under cultivated or field conditions, but, nevertheless, the occurrence of this variation is provocative. It is possible that the variation 1 stomatal distribution is not related to differences in environ- ment, but to leaves originating from cuttings. However, 11 each case the shoots on which the leaves were borne were not adventitious, but produced from axillary buds. Although these data are not conclusive they do support the observa- tions of A. and E. Camus (1904) suggesting that environ- SALIX GLAUCA COMPLEX 25 mental influence on stomatal distribution may provide a source of error. It is doubtful if sections, as presently understood, can be characterized by stomatal distribution. However, the intra- generic scope of this study is limited and an expanded survey may modify this conclusion. A recent paper by Watson (1962) in which the sectional importance of stomatal dis- tribution in the Epacridaceae was established should prompt an extensive survey of Salix. The specific differences with respect to this character are inadequately understood and presently are of little or no taxonomic value. Stomatal Length. The length of stomata in Salix was reported by A. and E. Camus (1904) to vary with the spe- cies, and even a cursory comparison of representative epi- dermal sections will corroborate this (compare figs. 3 and 4). The possibility that variation in stomatal length may be correlated with chromosome-number (see section on Cytology) enhances the potential significance of this charac- teristic and justifies a detailed consideration of it (see sec- tion on Techniques). The length of stomata was determined for five species and the means and total ranges are presented in fig. 2. These taxa are the same as were considered in the section on stomatal distribution; only two, Salix brachycarpa and S. glauca, are members of the S. glauca complex. The length of stomata in the three infraspecific taxa of Salix brachycarpa is essentially the same, with a mean vary- ing from 19.6 to 20.6 micra. Salix brachycarpa ssp. nipho- clada var. fullertonensis is an exception in this characteris- tic, as in others, and there is a strong suggestion that two stomatal lengths are represented in this taxon, one within the range of S. brachycarpa ssp. brachycarpa and the other within the range of S. glauca. For this reason, two Means are given for this taxon. Stomatal length-data for Salix glauca (fig. 2) are pre- sented for 4 phases, plus an intermediate zone 1 North America, and the area Europe and Asia (see Saliz 9 muee) Stomatal length varies from a mean of 26,3 micra in the Eastern phase to 30.1 micra in Europe and Asia. These mean lengths are in marked contrast to the report of stomata 8-15 micra long in European S. glauca by A. and E, Camus 26 GEORGE W. ARGUS (1904). Although an error by these authors may be suspect- ed, it is possible that a form of this species with short stomata, similar to S. brachycarpa, occurs in Europe. TAXON O28 NOG & wn 20 L 25 l l 1 lL 30 STOMATAL LENGTH (MICRONS) 1G. 2. Stomatal length sf selected species of Salix. The mean, range, standard deviation, and two standard errors are presented as desered in section Techniques. carpa ssp. niphoclada (4); 4, S. ehidedleats (8); 5. Ss. glauca, Europe and Asia (8); 6 i i cP a, Rocky M ; 10. S. phase (15); . S. arctica (5): 12. S. aretophila (2). See con for discussion The range of stomatal length measurements in Salix glauca is very great, from 22.1 to 33.4 micra, and tends to obscure any intraspecific differences. Nevertheless, a signifi- cant difference in the means of material of the Eastern phase in North America and of Europe and Asia is observed. These components of S. glauca can be distinguished on the basis of stomatal length. The mean length of stomata in Salix chlorolepis is 21.0 micra (within the range of S. brachycarpa). The mean length of stomata for S. urctica is 27.4 micra and for S. arctophila is 26.2 micra. Stomatal length in these taxa falls within the range of S. glauca. SALIX GLAUCA COMPLEX 27 Camera lucida drawings of representative epidermal sec- tions for four species are illustrated in figs. 3 and 4. The possible correlation of stomatal length and chromosome- number; is discussed below (see section Cytology). Environmental Inflence on Stomatal Length. To examine the possibility that the environment, which often has a S. arctica 8 = Z Fic. 8. Representative sections of the abaxial epidermis of Salix brachycarpa, S. chlorolepis, and S. arctica. 1-3. S. brachyearpa ssP- brachycarpa. 2 ae yp North Saskatchewan R., Alberta. 2. Argus 409-58, Churchill, Manitoba. 3. ra é hoc and Melchior 673, Kukpuk R., a. 5-6 hycarpa ssp , tonensis. 5 17270, Victoria Isl., N 6. Malte 120565, Chesterfield — N 7. S. chlorolepis, Fernald 5159, Mt. Albert, Quebec. 8-9. S. arctica. 8. Argus Lake, British Columbia. Stippled lucida. 28 GEORGE W. ARGUS A 20 Salix glauca aes Sweden Alaska Alaska / Wyoming 8 Manitoba Baffin Island Quebec Fic. 4. Representative sections of the abaxial epidermis of Salix glauca. 1. As- plund, 27 July 1949, Lapland, Sweden. 2. Johnson, Viereck, and Melchior 303, Ogoto- Cr., Alaska. 8. Argus 1145, Phelan Cr., Alaska Range, Alaska. 4. Lewis 842, Great Slave Lake, N.W.T. 5. Raup and Correll 11061, Watson Lake, Yukon. 6. J. and M. Reed 1668-69, Medicine Bow Mts., Wyoming. 7. Argus 340-58, Churchill, Manitoba. Malte 118709, Lake Harbor, Baffin Island. 9. Vietorin and Cermain 18925, Mingan Archipelago, Quehee. The stippled cells are trichomes. The drawings were made using a camera lucida, marked influence on leaf size, may have an influence on stomatal size, material from seven specimens growing under natural field conditions was compared with the same indi- viduals growing under cultivation at the Arnold Arboretum, Jamaica Plain, Massachusetts (fig. 5 and table II). For each specimen the relative leaf area of a representative leaf from each specimen was determined by leaf-length X leaf- width, and the mean stomatal length was based on ten measurements. SALIX GLAUCA COMPLEX 29 2 Aboxial Epidermis Leat Area (mm* ) and mean length of stomata (u) wild cult. wild 20u cult. 1500 ED: 1000 500 - 26.0432 ft ALA x’) a eT 1 and cultivated Comparison of leaf-area and stomatal length u capa ae, The mean is - en, and the standard deviation are given beneath the e aes aaa sta sentative section of the abaxial epidermis. 1. Sa glauca Argus — Mani ar glauca, Argus 58 hill, Manitoba ; , 3 ie Gothie, Colorado. 4. S. brachycarpa ssp. brachycarpa, Argus 81-59, Mt. i ; lucida. Quebee. The drawings of epidermal sections were made using a camera 30 GEORGE W. ARGUS TABLE II. ADDITIONAL COMPARISONS OF LEAF-AREA AND STOMATAL LENGTH UNDER NATURAL AND CULTIVATED CONDITIONS Taxon Relative area of leaf Stomata length (mm’) icrons) Wild Cult. Wild Cult. Salix glauca, Rocky Mt. phase (Levi 8, Elk Mt., Colorado) 330 800 24.5+2.9 22.542.1 S. chlorolepis (Argus 84-59, Mt. Albert, Que.) 128° 612 21.5432 185424 S. arctica (Argus 82-59, Mt. Albert, 330 1560 25.52.9 27.5242.7 In two specimens of Salix glauca from Manitoba (figs. 5-1 and 5-2) the leaf-area was increased about three times in the cultivated specimens, but this was accompanied nei- ther by a significant change in stomatal length nor by change in the size of epidermal cells. Leaf enlargement, in these cases, is due primarily to increased cell-divisions and not to a marked increase in cell-size. In material of S. glauea from Colorado (fig, 5-3 and table II) leaf area is about two times greater in the cultivated specimens than in the field collections. In these cases the epidermal cells are greatly increased in size, indicating that the gr eater leaf-area is due to cell enlargement as well as cell-division. However, the mean stomatal length remains unchanged. 4 representative of S. brachycarpa ssp. brachycarpa (fig. 5-4) also showed a marked increase in leaf-area under cultivation which, although accompanied by a small increase in epidermal cell-size, showed no change in stomatal length. The leaf area of S. chlorolepis and S. arctica (table II) was increased almost five times under cultivation, but no significant change in stomatal length could be detected. It can be concluded that favorable environmental con- ditions which apparently elicit a marked increase in leaf- area, both through increased cell-divisions and cell enlarge- ment, do not result in a significant change in stomatal length. This evidence is very important for it permits a greater confidence in the use of stomatal length particularly ™ wide ranging taxa which occupy a variety of habitats. SALIX GLAUCA COMPLEX 31 Seed Germination and Seedling Development. If provided with a suitable environment, seeds will begin to germinate within 12 to 24 hours. Within 48 hours 90 to 100 per cent germination may take place. The length of seed-viability is generally regarded to be only a few days, but under the proper conditions Salix seeds may remain viable for 9 to 10 months. Several authors including Toepffer (1915) and Sugaya (1956a) note that seeds may persist through the winter and germinate the following spring. This has also been observed in S. glauca where most of the seeds germ- inate in the year they are produced, but some may over- winter and germinate the following spring (Sdyrinki, 1939). Seeds of S. glauca from Churchill, Manitoba, showed some ability to germinate after storage for 20 days under warm dry conditions. Fresh seeds of S. calcicola and S. planifolia at Churchill placed on wet filter paper in a petri dish showed 90-100 per cent germination in 3-4 days. The per cent germination decreased with storage under warm, dry conditions. During the first 24 hours of germination the seeds of Salix glauca assume a bowed position resting on the radicle and the tips of the cotyledons. Elongation of the hypocotyl continues, rupturing the seed coat. At this time a ring of hairs (“hypocotyledonary bush” of Reed (1955) ) appears from the zone of brownish tissue located between the colorless radicle-cone and the yellow-green hypocotyl. The hairs in this ring are about 0.5 mm. long and may serve in anchorage and the absorption of water. The elongation of the now erect hypocotyl continues and radicle elongation is initiated. After 2 to 3 days the radicle is about 1.5 mm. long and root hairs are present along its length. The hypo- cotyl continues its elongation for about 7 days, at which time the cotyledons, having spread apart, reveal the be- ginning of epicoty] development. After about 30 days, the seedling may be from 5-10 mm. long and bear 2-3 unfolded leaves. If the seedling is placed under “winter conditions” of short days and cool tempera- tures a terminal bud may be formed. This bud is formed by the clasping of the apical meristem by the two r eddish- green subterminal leaves. The occurrence of an apical bud in Salix has been reported by Sugaya (1956a and 1956b) 32 GEORGE W. ARGUS in seedlings of Salix pauciflora Koidzumi and in young shoots on cuttings of S. reticulata L. A terminal bud is ordinarily formed in Salix only during the seedling stage and at the time of its first dormancy. REPRODUCTIVE MORPHOLOGY AND BIOLOGY Inflorescence. The inflorescence of Salix is a highly reduced, simple, usually erect, unisexual structure called an ament or catkin. The species are dioecious, the individual plants bearing either stam inate or pistillate aments. The relationship between the opening and development of the inflorescence and foliage buds is generally described using three terms: 1) precocious, inflorescence appearing before the foliage; 2) coétaneous, inflorescence appearing at about the same time as the foliage; and 3) serotinous, inflorescence appearing after the foliage. The pistillate aments are composed of pistillate flowers each consisting of a single pistil and its associated gland (s) borne in the axil of a bract. The staminate aments are composed of staminate flowers consisting of a variable number of stamens, two in the Salix glauca complex, and their associated glands borne in the axil of a bract. In the S. glauca complex the pistillate flowers have only one gland, rarely two, located adaxially, between the ovary and the rachis, and the staminate flowers have two glands, one located adaxially and the other abaxially. The ament in the Salix glauca complex is always borne at the end of a reproductive shoot (Figs. 6 & 6a). This Fics.6-8. hree general types of reproductive shoots in Salix. FIG. Salix glauca pies 196-58, 12 July 1958, Churchill, Manitoba). ii Salix glauea the Po nT (A) is located between ae distal leaf and the proximal flower. In s duais the reproductive shoot may be shorter and bear small bracts Salix reticulata (Thieret and Reich 5818, 30 July —2 Aug. 1959, Horn Mts., In S. retiewl (right), 24 May 1962). In this, and other icecsiids pp Fiance ay be develops before the vegetative eae appear and fruits may 5 the vegetative shoots have matu (Fig. 8, arate branch). In precocious rie reproductive shoots are very 7 or open eaatlt absent and bear reduced “ae SALIX GLAUCA COMPLEX 33 34 GEORGE W. ARGUS reproductive shoot is referred to in the literature as a peduncle and aments are described as “leafy-pedunculate” or “sessile”. In some species such as S. reticulata the re productive shoot is as prominent as the vegetative shoot and in these cases only the upper portion of the shoot above the leaves is called the “peduncle” and it is referred to as “a long leafless peduncle” (Fig. 7). Those species in which the reproductive shoot is very short or absent are the precocious flowering species such as S. discolor Muhl. and S. planifolia (Fig. 8). The “peduncle” terminology common usage is unfortunate because it is inconsistent and inaccurate. Actually, the difference between the re- productive and the vegetative shoot is simply that the former is terminated by an ament. The internode between the last leaf on the reproductive shoot and the ament has 4 zone of abscission a variable distance above the last leaf. The region between this zone and the lowermost flower may be properly referred to as the peduncle and the region below it the reproductive shoot. In most cases, the ament is cut off by the abscission layer; however, at times the whole reproductive shoot may dry up and fall off. In S. glauca the staminate aments are borne on very short reproductive shoots and the aments are characteristically cut off above the last leaf. Since the shoot bearing the ament is not called a “peduncle” in this paper, there is od reason to refer to the leaves borne thereon as “bracts”. This makes it possible to reserve this term for the struc- tures subtending each of the flowers (see Bracts). The problem of the relationship between the ament, peduncle, and reproductive shoot requires detailed morphological and anatomica] study, In the Salix glauca complex the length of the pistillate aments has been considered to be of taxonomic value an taxonomists have characterized Salix brachycarpa as havings Shorter aments than S. glawea. Schneider (1919) noted that S. brachycarpa has pistillate aments varying from 1.5-2.5 em. long in fruit, in contrast with the longer aments of S. cordifolia (eastern Canadian populations of S. glauc@ ) which vary from 3.0-9.0 cm. long. When this generaliza- tion is analyzed in terms of local populations a wide range of variation is found. The length of fruiting pistillate SALIX GLAUCA COMPLEX 35 aments was measured in two field local population samples from Churchill, Manitoba. Salix glauca (Fig. 9) ranges from 1.4-3.8 cm. long, and S. brachycarpa (Fig. 10) ranges from 0.7-1.9 (-2.2) cm. long. These data do not fully corroborate the observations of Schneider and illustrate the wide variation and overlap which may be encountered on the local population level. The variation described here applies only to populations in the Churchill, Manitoba 3.0 CM %| 10 ce AMENT LENGTH 20 Pay 9-S. GLAUCA | : 4 10-S. BRACHYCARPA Aad f- STAMINATE | O- PISTILLATE 54 WZ 10 A 90°” GM lauca and n field local in a gravel thee habitat e habitat. See text for Fics. 9 AND 10. i length variation in local populations of g Population samples om lix glauca (Fic. 9) was sampled and Salix PEL iat se (FIG. 10) was sampled in a sand dun 36 GEORGE W. ARGUS region. If populations of S. glauca from Alaska were com- pared with S. brachycarpa, overlap in pistillate ament length would be virtuaily absent and the characteristic would be of diagnostic value. The staminate aments of both species in the Churchill, Manitoba region are short and their ranges almost coincide (Fig. 9 and 10). However, those of S. brachycarpa do show a tendency to be shorter. The general shape of aments, particularly the pistillate, provides a more diagnostic characteristic than does ament- length when contrasting Salix brachycarpa and S. glauca. The pistillate aments of S. brachycarpa are often referred to as subglobose to oblong or short-oblong to spherical, in AMENT LENGTH / WIDTH % i 11 -S. GLAUCA 10 12 -S. BRACHYCARPA [a -STAMINATE C)-PISTILLATE 5 Ci i ee Zs YG ACA CZACZS g, : e 30 ia 50 58 %} |2 i > 3 13 o a. ‘ ° : e o 2 ee Co Be he ie pares WepeaP dt shh. 20 Go 2388008 5) a 2 a | ad | 2 3 cM 10 AMENT LENGTH ° S, GLAUCA I3 -. S$ BRACHYCARPA & 7 pete on . 4 Ke) 20 brachycarpa at Chur shen 11-13, > Clana variation in local populations of Salix glauca and chill, Manitoba. FIG. 11. Salix glauca sampled in a 8T@¥ ridge habitat. Fic. 12. vel iB ¥iockgauila sampled in a sand du Si Correlation of pistillate passa h teenie and ament-length/width. The data are based on field local population samples. See text for discussion. one ee, SALIX GLAUCA COMPLEX $7 contrast to the cylindrical pistillate aments of S. glauca. An analysis of local population variation (Figs. 11 and 12) shows overlap almost as great as found in ament-length. However, a correlation of ament-length and ament-length/ width provides a sounder basis for distinguishing these two species. (Fig. 13) Bracts. The bracts of Salix are simple, entire structures subtending each flower, and have a vascular supply similar to that of normal leaves (Fisher, 1928). Fisher does not regard them as true bracts of the flower because their vascular supply is derived from the stele of the flower and not from the stele of the ament. She interprets them as pertaining to the flower and arising from the pedicel. Hjelmquist (1948) does not agree with this interpretation and cites examples of similar coalescence of bracts and flowers in other Amentiferae which in the case of Salix led to the vascular bundles becoming partly united. In the pres- ent paper, this structure is held to be a bract subtending the flower. Variation in the size, shape, color, and pubescence of bracts is great even within a species. They vary in shape from narrowly linear to broadly oval. In the Salix glauca complex, certain bract shapes and sizes are superficially distinctive. For example, S. brachycarpa ssp. brachycarpa often has longer, broader bracts than does S. glauca; and linear bracts are common in S. brachycarpa ssp. niphoclada. A local population analysis of pistillate bracts (Figs. 14-2 and 15-2) reveals that although extremes may be distinctive, bract-shape and size are not of diagnostic value. Further, the staminate bracts (Figs. 14-1 and 15-1) are virtually identical in both S. brachycarpa and S. glauca. Although certain species seem to be characterized by bracts of a cer- tain shape, this is a highly variable characteristic and of limited taxonomic use. Bract color in Salix varies from yellow, tawny to green, black or bicolored (black at apex, light color ed below). Bract-color has been used as an important characteristic at both the sectional and specific levels. Schneider (1918a) noted that color and pubescence of bracts afforded reliable criteria on which to distinguish section Ovalifoliae (the S. arctica complex) and section Glaucae (the S. glauca com- 38 GEORGE W. ARGUS sé plex). The former with bracts usually more or less bicolor, being pale at base and dark brown, fuscous, or even blackish toward the apex .. .” and the latter with “... uni- formly yellowish, light brown, or straw colored bracts.” In general this characterization is accurate, but dark-brown to SALIX compiea (CHURCHILL) ce @eoer 14 ostAamae BRACTS 2 ~PISTILLATE BRACTS i bea ee ae ee eh@aaeanaasaa SALIX BRACHYCARPA (CHURCHILL) SAND DUNE 2 g Se é € 15 I -staminate practs | 2 -PISTILLATE BRACTS MM. -eaaaaaeeaad& Figs. AND t ion a (1) stamin- } s of Sal ix and size-variation in two local popul glauca 2 Ss. brachiear at lant ll, fag initoba. FIG. 14. Salix gla acts ate bracts, (2) pistillate ‘acts. FIG Salix brachycarpa (1) staminate ee fe (2 : <) pistillate bracts. Ilustr: ation pre we: using a microprojector discussion SALIX GLAUCA COMPLEX 3Y black bracts are observed in S. glauca in the southern Rocky Mountains and bicolored bracts occur in some material of S. glauca from Arctic Alaska. Bract-color seems to vary with the ontogenetic stage of development of the flowers. Smith (1954) reported a color change in Salix glauca from Colorado (as 8S. pseudolap- ponum), from green to pinkish and finally to brown. A similar series of colors from green to yellow or light brown has been observed in S. brachycarpa at Churchill, Manitoba. In 1905 Fernald used the green bracts of Salix chlorolepis to distinguish this taxon from S. brachycarpa. An analysis of a local population sample of S. brachycarpa from Mt. Albert, Quebec, where the endemic S. chlorolepis also occurs, shows a striking correlation between floral development and bract-color (Fig.16). Aments in anthesis have predomi- nantly green bracts and those in postanthesis have yellow to tawny bracts. There is a continuous sequence of inter- mediate colors if ontogenetic processes and genetic differ- ences between individuals are taken into account. The high frequency of green bracts in flowering specimens of S. brachycarpa brings into question the validity of the use of bract-color to distinguish it from S. chlorolepis. The extensive occurrence of greenish bracts in Saliz brachycarpa on Mt. Albert also may be related to its hybridization with S. chlorolepis (see Salix brachycarpa x chlorolepis). However, the occurrence of green bracts in S. brachycarpa at Churchill, Manitoba and the correlation between green bracts and the flowering stage in this taxon on Mt. Albert suggest that, although the green color may have been re-enforced in the Mt. Albert population, the correlation of bract-color and the stage of floral development is real and should be carefully considered. Nectaries. The nectaries of Salix are usually small, flat- tened structures which secrete nectar from their apex. There may be one, two, or several associated with each flower. In the S. glauca complex one, or rarely two, 1S found in the pistillate flower and two in the staminate. If a single nectary is present it is located adaxially between the ament- axis or rachis and the sporophyll. If two are present, the second is located abaxially between the bract and the sporo- phyll. The abaxial gland is the smaller of the two. If more 40 GEORGE W. ARGUS BRACT COLOR S. BRACHYCARPA ; (MT. ALBERT) fe 20. i \ — ANTHESIS i \ --—- POSTANTHESIS 4 \ aevine Lid fae) = Seg ova, uf / / pe ee ae J oe |; a8 Se tk owes Biase np : S. brachycarpa ssp. niphoclada xX glauca. Vierec ile A The small nectaries to the right of the dash in 4, 5, a id 9 3 ney € Although second nectaries occupying this |} ion are uncommon in e flo three individuals having them are 1 The abaxial , and uncleft in contrast to e a g drawings were ‘ " The use of the number of glands per flower as an Impor- tant sectional characteristic, as was done by Schneider (1922), should be reexamined. He did not incl in the same section as S. glauca because its stamit have two glands. As we have seen, the staminate flov of the species in the S. glauca complex also have two glands, although the pistillate flowers usually have only one. The ude 5. wolh nate flowers vers 42 GEORGE W. ARGUS difference which was supposed to exist between these two species does not in fact exist. Gland number appears to be somewhat variable and should be used with caution. Stamens. The stamens vary in number in Salix with the diandrous group having one or two per flower and the pleiandrous group having a larger number, usually five. The filaments are commonly free although they may be more or less united toward the base. In the Salix glauca complex this is variable and stamens with free or united filaments may be found in a single specimen. A pubescence of long simple trichomes may be found near the base of the fila- ments. Each stamen is supplied with a single vascular trace (Fisher, 1928). The anthers dehisce by a longitudinal split between the two pollen-locules. Pistil. The pistil is a flask-shaped structure which may or may not be pedicellate. In the Salix glauca complex both pedicellate (S. glauca) and non-pedicellate or sessile (S. brachycarpa) pistils occur. The length of the pedicel pro- vides a reliable diagnostic feature to distinguish S. glauca from S. brachycarpa. In comparing field local population samples of these taxa from Churchill, Manitoba, S. brachy- carpa was found to have a pedicel-length of 0.0 to 0.25 mm. in 80 per cent of the population and S. glauca a length of 1.0-2.0 mm. in 90 per cent of the population. An overlap of less than 10 per cent was found in the 0.5-1.0 mm. range. The pistil may be glabrous or pubescent; in the taxa studied pubescent pistils are the rule with some becoming glabrate in age. The carpels are two in number and are united to form a unilocular ovary. They were considered by Rainio (1926) to be median with median placentae. How- ever, this view is not generally accepted and Velenovsky (1904), Fisher (1928), and Hjelmquist (1948) all agree that the carpels should be interpreted as being transverse with the placentae formed at their edges, Fisher reports that each pistil receives six bundles and that the ventral ones are fused into two large placental bundles. The pla- centae are parietal, each bearing several erect, anatropous ovules (Hjelmquist, 1948). A single style is present in all the taxa considered in the present study, although in some it is very short. In Salix glauca it varies from entire to bifurcate. The stigmas SALIX GLAUCA COMPLEX 43 the S. glauca complex are mostly four-lobed with each lobe more or less linear. In some cases the lobes may be short and only two in number. Fruits and Seeds. The fruit of Salix is a capsule with loculicidal dehiscence. At the time of dehiscence the enclosed seeds are exposed revealing a coma of white or sometimes yellowish trichomes produced as outgrowths of the placentae (Takeda, 1936). At fruit maturity, the stigmas and often the styles drop off; the distal end of the capsule splits and the two halves curve backward. This releases the seeds which are small, exalbuminous, and provided with a coma attached loosely to their bases. The seeds are generally ellipsoidal to cylindrical and vary in length, depending some- what on the species. In the S. glauca complex they vary from 1 to 2 mm, long. Toepffer (1915) reports the range of color within the genus to be from black to brownish or greenish. In the material I have examined the seed-coat is colorless and translucent, and any color that the seed pos- sesses depends on the enclosed embryo. Viable seeds, and even some that are not viable, are generally greenish. How- ever, when studied on herbarium specimens brownish or blackish seeds may be noted. The coma consists of trichomes often two or more times the length of the seed. In the species studied, trichomes four to seven times as long as the seed are not uncommon. The trichomes are fused at their proximal end into a ring that is loosely attached to the seed. The coma provides a “parachute” for each seed presumably greatly facilitating its dispersal. Pollen. The Salix pollen-grain is generally prolate, tri- colpate, intectate, and reticulate. The colpus is usually boat- shaped and it is usually regarded as lacking a germ pore. The reticulum is formed “. . . by isolated granules, columel- lae, standing on a continuous endexine.” (Faegri, 1953). The columellae are not fused into a continuous tectum and the grains are intectate. The upper parts of the columellae are fused into a “reticulum tectum” (Faegri, 1953), The grains range in size from 20-24 microns in some small grains, to 33-36 microns in larger ones (Erdtman, 1952). The size seems to be related to the chromosome-number but Erdtman cautions against making any assumptions based on this information. 44 GEORGE W. ARGUS The pollen-grains are extremely variable in size, shape, number of colpi, pattern of sculpturing, etc. (Straka, 1950; Faegri, 1953). The degree of variation which had been noted by Héeg (1929), and Erdtman’s (1952) cautioning against relating grain-size and ploidy-level suggest that the use of pollen-grains as a taxonomic character should be avoided or used with extreme caution. I have seen two attempts to distinguish certain species on the basis of pollen- grain morphology. These were by Straka (1950) and Faegri (1953). For a non-palynologist the keys presented by these authors are of limited value. Faegri points out that “. . . the morphological details of the Salix pollen grain are so difficult to observe and interpret, being in fact of the same order of magnitude as the maximum resolution power of the visible light microscope, that only the very best optic equipment is adequate for the task.” The species which Straka was able to distinguish on the basis of their pollen-grain morphology were S. glabra, S. silesiaca, S. herbacea, S. daphnoides, 8. pentandra, and S. alba. Faegri provided a key to distinguish S. herbacea, S. myrsinites, S. m. ssp. weigliana, S. polaris, S. glauca, S. glandulifera, S. lapponum, S. reticulata, and S. lanata. A study of six hybrid species by Straka (1950) showed their grain type to be dominated by one parental type, mal formations to be frequent, and new combinations rare. This suggests that hybrids could be distinguished on the basis of malformed pollen-grains. I have not been able to detect such grains in any of the hybrids I have examined. To my knowl- edge, this characteristic has not been used taxonomically, but it deserves some attention. Pollination. Salix ig generally regarded to be insect- pollinated, although considerable quantities of pollen may be found to be transported by wind. Hgeg (1929) in his report on the pollen-grains carried by two species of Bomous on Ellesmere Land found that Salix grains were most com monly carried. Insects are frequent visitors of Salix and the presence of nectaries suggests that the flowers at largely adapted to an entomophilous mode of pollination Knuth (1909) noted, “. . . as insect visitors go from ont species of willow to another indiscriminately, it is difficult to assign them insects to individual species.” Howeve!s the SALIX GLAUCA COMPLEX 45 importance of wind-pollination cannot be overlooked, espe- cially in the northern regions where strong winds often prevail in summer. The relative importance of these two means of pollen-transfer is of particular importance in esti- mating population size in this genus. At any rate the pollination is promiscuous, no matter which agent is at work, and hybridization resulting from this non-specific transfer of pollen is probably common. Phenology. The relationship between the opening and development of the inflorescence and the foliage (see In- florescence) is also reflected in the time of flowering of the species. Precocious species flower early in the season and the coétaneous and serotinous species late in the season. During 1958, seasonal notes were made at Churchill, Mani- toba, on the phenology of most species of Salix in the area; the flowering time data are presented in Fig. 18. The data lO 1 n 4 30 + seth cae L—cALCICOLA 4 + ARC TOPHILA———4 H—PLANIFOLIA-——4 L—ALAXENSIS—I KRETICULATAY [+ GLAUca 4 E-CANDIDAS K-VESTITA tL ™BEBBIANA I—BRACHYCARPAA JUNE JULY 18. Comparative flowering time of the species of Salix occurring at Churchill, rence The data are based on field-observations made during June and July 1958. See text for discussion. are ‘based in part on casual observations and in part on successive collections, and represent the dates of anthesis of the species studied. There will be variation in flowering time from year to year but the flowering sequence can be expected to remain relatively constant. Some of the species plotted, such as S. vestita, S. bebbiana, and S. alazensis, were observed only rarely and their flowering period should be regarded as only approximate. 46 GEORGE W. ARGUS A general pattern of variation in flowering time can be noted, with the precocious species Salix calcicola and S. planifolia flowering early, and those coétaneous species, such as S. glauca and S. brachycarpa, flowering later. This pat- tern is supplemented by characteristics related to the adaptation of some species to a short arctic growing season. S. arctophila and S. reticulata are examples of arctic-alpine species which flower early although they are coétaneous. S. bebbiana and S. brachycarpa are examples of boreal species which are coétaneous and flower late in the season. The spread in time of flowering of the Salix species at Churchill may represent an isolating mechanism which re- duces the degree of hybridization between these species. Salix glauca and S. brachycarpa are probably significantly isolated by this mechanism. Further detailed phenological information from various localities would be of particular interest in studies of hybridization and evolution in Salix. Cytolegy. Polyploidy is important in the evolution of Salix and a knowledge of the chromosome levels in the Salix glauca complex is essential to an eventual understanding of its evolution. Unfortunately, only a few counts are available for species in this complex and these are inadequately docu- mented by herbarium vouchers. During the course of this study some chromosome counts were made. However, due to a series of difficulties the counts are too few and the results not unequivocal. In spite of limitations, these data are of some value and contribute to our understanding of polyploidy in the S. glauca complex. The chromosome counts that have been reported for European Salix glauca indicate that this species is octoploid with two basic numbers, 22 and 19. Material having a basi¢ number of 19 (2n = 152) is reported by Marklund Floderus (1931), and Léve and Léve (1948), and material with a basic number of 22 (2n = 176) is reported by Wil- kinson (1944). The 22 number is believed by WilkinsoD to be due to the fragmentation of chromosomes in the base type. A second “species”, S. callicarpaea (eastern Canadian populations of S. glauca), has been reported to be decaploid (2n = 190) by Léve (1954). Counts based on Salix glauca and S. brachycarpa from Churchill, Manitoba, and on S. brachycarpa ssp. brachy- SALIX GLAUCA COMPLEX 47 carpa X glauca from the Medicine Bow Mts., Wyoming are reported for the first time in this study. Since completely accurate counts proved to be difficult to make, the numbers of chromosomes actually counted are presented here with no attempt to adjust them to their probable ploidy level. See section on Techniques for a discussion of methods. The results are presented in Table III and in Fig. 19. Several interesting features of these data should be noted. | ee *¢@ es to ee eo -.8 9358 yay ee” °¢ or - sept oe ee 80 rs 3’ rd od Fa oetees® S em vf a 48 otet 200 OF “9 Se ‘ e 8 3 . » 202 2 ee «6 *° . oe, ee ad wad 2 Pd $52%00 ose : s Bm + a :, a oes, es °ee. ee b ° poy es gs . ave @8 @e¢@ eee ms * 2°Fe ae SEEM Si Tat e oe @ ®@ “23 e Fic. 19. Root-tip mitoses. 1-4. Salix glauca. 1. Argus 15S, Courehill, Manione (2900K). 2. Argus 26S, Churchill, Manitoba (2900X). 3. Argus 517-58, Churchill, Manitoba (1960X). 4. S. brachycarpa ssp. brachycarpa, Argus 608, rerenon Man- itoba (2900X). 5. S. brachycarpa ssp. brachycarpa X glauca, C. L. Porter, August 1958, Medicine Bow Mt., Wyoming (1960X). First, Salix brachycarpa ssp. brachycarpa is diploid. Sec- ond, S. glauca from Manitoba may contain more than one level of ploidy, The material counted varies from 4 to > and although the counts are not exact, they are not as = as 10X reported by Léve (1954) for S. callicarpaea (= d glauca in eastern Canada). However, it should be hi d that the specimens on which these S. glauca counts ar i oe represent the ferruginous form of the species (see Kocky Mt. phase of S. glauca) and due to the uncertainty — rounding this form it cannot be regarded as typical = the eastern Canadian populations of S. glauca. Third, the ap- 48 GEORGE W. ARGUS TABLE III. CHROMOSOME COUNTS OF TAXA IN THE SALIX GLAUCA COMPLEX Species Count - ploidy Source of Material Reference Counts reported in the literature, no vouchers seen. S. glauca ES eee Seandinavia ? ...... Wilkinson (1944) Ss. uca Be OR: Sects Scandinavia ? ....... Floderus (1931) S. glauca ssp. stipulifera ......... BGS 3: BR sesices Scandiavia 12 > Pe fice: af & eo « x * ~~ s Ps 4 @ ° ee Ba ae Pre ay q i T 35 4 : 3 bd ° s & hd 3 ¥ Ee 30- 3 3 s % e °* « = mt tate ig Sos + e e " > 25 - by? ° 0 c 38 2, 3 © e eo + ° 2 a | - eke, A ~ & 20- ry ° ° 00 w 7 Oo re) . ° re) 15 L 1 , : @ Salix glauca O Salix brachycarpa ssp. brachycarpo A Salix brachycarpa ssp. niphociada O) Salix brachycarpa var. follertonensis Fic. 20. Geographic variation of mean stomata length in uca and : brachycarpa. Each symbol represents the mean Woon hay a one individua (see section Techniques.). See text for discussio On the basis of stomata length, Salix brachycarpa ssp. niphoclada can be assumed to be diploid. Although the sample size is small, this taxon generally seems to differ markedly in mean stomata length, and presumably also in ploidy level from S. glauca in Arctic Alaska. The alignment of these taxa, as proposed by Wiggins (Wiggins and Thomas, 1962) is not supported by these data. 52 GEORGE W. ARGUS There is an indication that Salix brachycarpa ssp. nipho- clada var. fullertonensis has a greater tendency toward longer stomata than is typical for the species, In Fig. 20 this variation is shown to be within the limits of variation for the species, and the “tendency”? may be due to chance. The occurrence of intraspecific polyploidy within Salix glauca, discussed above, is supported by the data presented in Fig. 20. The mean stomata length, and presumably ploidy level, is greater in Europe, Asia, and northwestern Alaska than in Manitoba and Quebec. In general, there is clinal variation in stomata length from west to east in North America, and there is a possibility of two or three levels of ploidy occurring in this region. Clearly there is a serious need for additional information on chromosome-number in the North American Salix glauca. SOME GEOGRAPHIC ASPECTS OF VARIATION IN SALIX GLAUCA An examination of a large series of specimens which had been named Salix glauca, S. cordifolia, and S. pseudolappo- num (glaucops) revealed the basic similarity of these taxa and suggested that, although populations in certain areas may represent a “typical” manifestation of one or another of these names, there is a large region in which intermediate populations predominate. Furthermore, attempts to dis- tinguish entities to which these names apply break down as soon as the problem is approached on a population basis over a wide geographic area. In order to compare the variation in these taxa the characteristics that have been used to distinguish them were studied over the North American range of the entities concerned. Unfortunately, comparable jocal populations of S. pseudolapponum from the southern Rocky Mountains were not available and the population-variation for this entity is considered only in the northern portion of its range. In Table V, a comparison is made between Salix glauca, S. cordifolia and S. pseudolapponum (glaucops). The data are derived from the literature (mainly Schneider, 1919 and Raup, 1943 and 1959) and illustrate the difficulty in distn- guishing between these presumed taxa. Morphological data based on my observations are recorded in Figs. 21 through 26, and in the systematic treatment; these do not agree @ every instance with the data in Table V. Even if the entiies 53 SALIX GLAUCA COMPLEX ‘spuv[s Bulpssoxe yony “UIE §-S"9 ‘peiamoy Ajasuaq ‘qINIy Ul "WO g-z *peonpat yonut 10 auoN *pepunor 0} eveuND ‘aqnoeqns 0} aynoy ‘SuO[qo-914dI][9 10 eyBposouR] -9JBAOGO 10 JVdIT[a A[MOLIBU ‘aye[OaDURT ~d14d][[9 ‘ayB[oa.URI GO ‘aye[osoUEr] ZZ jnoqe ‘uld p JaA0 you wunuoddvjopnasd *s ‘spux[s Surpesoxe jon “uu (0T-) 8-9 “pAleMoy Ajasuep ssa] 10 a10W “"HOAy UL “UD Q-Z “wut g-g 0} PU0U *‘a}Bpaoogns 0} e}yveuNdD A[pvorq 10 esnjIqO "aynov PAOYS 10 osNzqO ssa] 10 V10[W ‘OIdI[]9 97BAO 10 Suojqo ~9JVAOGO ‘IAAI I[[e-o7BAogo ‘[Bao ‘OAT $°3-3'0 “wo (9-) 9-8 (-2°T) mOfpsoo *g A @Tavib spux[s ey} sour Z 0} ‘uur (§-) 8-b ‘asuq PlBMO} patomoy AlesooryT “‘yInay UT “WA 1-G"g “uu gT OF 9°% ‘ayeound A[[VNpBAs 10 A[Jdnaqe 10 osnjqo ‘ayBuluNoe -j.1048 A[IBVaU IO 9}NIB AO OsNzIGO ‘O19A][[9-ayBAogGo 10 Buolqo-azyuaogo ‘Suol[qo-d14d[[9 ‘aye[OaouR]qo ‘ayelooouRry P-2'S “utd (OI-) S-9°P (-%) ponnjo “sg Yue[-porpeg yyuel-e[nsduy BupoMOLyT YW sue]-quewre oye] [1981 4 uel-epndys O8Bq-7 Ver] xedu-jeary odvys-juary M/[ year] yySue|-suer] 892981.L020DLDY9 54 GEORGE W. ARGUS in Table V are treated in a narrow sense they are difficult to distinguish. However, certain characteristics may be of diagnostic importance and their geographic variation is considered. These are leaf-length, leaf-shape (as length/ width), pistillate ament-length, stipule-length, pedicel- length, and petiole-length. Each of the aforementioned characteristics was studied in twenty-one herbarium samples of local populations. The herbarium samples of local populations were selected from over the North American range of the species, with the exception of the southern Rocky Mountains. The popula- tions were delimited and the characteristics measured in the manner described in the section on techniques. The sample size varies from 44 specimens in the Umiat population (No. 7) to 14 in the Wiseman population (No. 5), with an aver- age of 33 specimens for the 21 populations. The data were plotted as frequency histograms and compared visually. Leaf-length data were plotted on a logarithmic scale to eliminate distortion and more accurately represent com- parative intrapopulation variation. If the leaf-length data were plotted on an arithmetic scale the short-leaved eastern populations would appear to have less intrapopulation varia- tion than the long leaved Alaskan populations. Data from the following local populations are included in Figs. 21 through 26. The geographic area sampled in each population and the sample size are indicated. The popula- tions are: 1. Naknek, also including King Salmon, Katmai, Lake Iliamna, Island L., and Ugashik L., 18 specimens. 2. Mt. McKinley National Park, 43 specimens. 3. Fairbanks, also including Livengood, 25 speci mens, 4. Circle, also including Miller House and Ft. Yukon, 23 speci- mens. 5, Wiseman, also including Old John L., 14 specimens. 6. Whitehorse, also including Lake Atlin, Bennett, Skagway, McRae, and Vista, 26 specimens. 7. Umiat, also including Kurupa R., Sadlerochit R., and Saganavirkotk, 44 specimens. 8. Aklavik, also including Eskimo Lakes, Eskimo Basin, Williams Isl., Reindeer Sta., Warten Pt., Caribou Hills, and the east branch of the Mackenzie R., 21 speci mens. 9. Jasper National Park, also including Brule and Beauvert L. 35 specimens. 10. Glacier National Park, 16 specimens. 11. Great Bear Lake, including Leith Pen., McTavish Arm, Ft. Franklin, Gun- barrel Inlet, Pt. Radium, and Sawmill Bay, 20 specimens. 12. Great Slave Lake, including Yellowknife, Barabant Isl., Hardisty Isl., Lon& Isl., Ft. Rae, Gros Cap Isl., Caribou Isl., Moraine Pt., Indian Village, Ft. Providence, Trout R., and Horn Mts., 28 specimens. 13. Lake SALIX GLAUCA COMPLEX 55 5 16 17 1819 2 . : a 2 oat al a ie | 2 = = m a aye = ns - m= ee: = : * a : = : ASE -— mm 2 — a » we "I 3 me og . P PPLE REE GL i =k iF ae - Las " F LEAF LENGTH : | 2 » i ; SCALE 4 5 Fic. 21. Geographie variation of leaf- “eae in Saliz glauca. See the in the area represented by each of the 21 herbarium local populations ee peony The scale is in maven oe individuals. The nel pm seale is plotted on a logari- thmic scale - Oa = = oa a b (= IO ree Re he ee Viv ON Viva ON G 5 Ow — oF > i ai ore WIDTH RATIO is ee | 3.45678. 9.10 Il 12131415 1617 18 19 202 E Tee di = "hs pas at ni | Zo SCALE | 5 PISTILLATE : ; + AMENT LENGTH 3 a In ‘fli ee i MM 6 7 10 1 13 14 15 16 s sia 10 : tie LENGTH muy | 9 10 IL 12 13 14 IS i819 05 20 26 | PEDICEL, LENGTH as expressed by the length-width ratio. Fic. 23. is otatSom in pistillate ament- -length. - iation in stipule length, Fic. 25. Variation in — length. Fic. 26. riation in pedicel-length. See text for discussion. The scale is in number of 56 GEORGE W. ARGUS 9 Il 12,13, 14 GAT. pA oe ee 4 19 20 2 Fics. 22-26. Geographic variation in Salix glauca, FIG. 22. Variation in leaf-shape Stalhotdieata: SALIX GLAUCA COMPLEX 57 Athabaska, 23 specimens, 14. McKague, also including Tisdale, 18 specimens. 15. Churchill, 27 specimens. 16. Richmond Gulf, also in- cluding Great Whale R., 30 specimens. 17. Baffin Island, including Ponds Inlet, Lake Harbour, Pangnirtung, Frobisher Bay, and Griffin Bay, 29 specimens. 18. George River, also including Indain House L., Helen Falls, Hades Hills, Mt. Pyramid, Mt. Bridgeman, and Camp Misery, 22 specimens. 19. Cape Mugford, also including Windy Tickle, Hopedale, Main, Sumavik, Davis Inlet, Port Manvers, Kikkivitok Isl., Christie Isl., and Jack Lane Bay, 27 specimens. 20. Mingan Archi- pelago, 25 specimens. 21. Strait of Belle Isle, including Blane Sablon, and Forteau, 15 specimens. Results. Leaf-length (Fig. 21) is strongly clinal from Alaska to eastern Canada. Populations with long leaves occur in the Alaska-Yukon area, and populations with short leaves occur in the Rocky Mountains, eastern, and north- central Canada. The extremes correspond to Salix glauca in the northwest, S. cordifolia in the east, and S. pseudo- lapponum in the Rocky Mountains. However, connecting the extremes, which more or less overlap, is a series of intermediate populations. Several of the intermediate popu- lations (e.g. Great Slave Lake — 12, Great Bear Lake — 11, and Whitehorse — 6) have a greater intrapopulation varia- tion than do some of the extreme populations (e.g. Fair- banks — 3, Mingan Archipelago — 20, and Glacier National Park — 10). The clinal variation in leaf-length makes this characteristic of much less diagnostic value than had been previously recognized. : Leaf-shape, as represented by length/width (Fig. 22) is also clinal and provides no basis for separating eastern, western, and Rocky Mountain populations. Pistillate ament-length (Fig. 23) is weakly clinal. Al- though the eastern and Rocky Mountain populations have slightly shorter aments than the western populations the overlap is too great for this characteristic to be of diagnostic value. The wide intrapopulation variation in ament-length is partly due to the different stages of ontogenetic develop- ment represented in the samples. It is difficult to find, in herbaria, sufficiently large collections of individuals in the same stage of development to eliminate this source of error. Stipule-length (Fig. 24), although slightly clinal, comes closest to providing a characteristic to separate the north- western populations from the eastern and Rocky Mountain ones. In the Alaska and Mackenzie Delta populations sti- 58 GEORGE W. ARGUS pules are usually prominent, although sometimes short, and in the eastern and Rocky Mountain populations the stipules are usually short and rarely prominent. However, even in this case the distinction is not unequivocal and there are exceptions which lower its diagnostic reliability such as the Naknek population (No. 1) with many small-stipuled plants and the numerous intermediates in the Mackenzie Delta population (No. 8). The shape of the stipule (not plotted) varies as well as length, with the northwestern populations having narrower stipules than the eastern or Rocky Moun- tain populations. If the northwestern populations are to be treated as taxonomically distinct from the eastern and Rocky Mountain populations the stipule characteristics may provide some basis for it. In cases in which the stipule is recorded as absent it was either caducous or less than 0.5 mm. long. Petiole-length (Fig. 25) variation is presented as an abso- lute value rather than as a ratio for two reasons. One, it illustrates the uniformity of this characteristic throughout North America, independent of leaf-length, and two, if it were considered as a ratio with leaf-length its curve would be similar to that presented for leaf-length. Throughout the entire range of the entities included in Salix glauca, petiole- length provides a means to distinguish it from the short- petioled S. brachycarpa. Pedicel-length (Fig. 26) variation has been plotted to illustrate a second characteristic which remains relatively constant over the entire range of the Salix glauca complex, and one which is in sharp contrast to the very short pedicel of S. brachycarpa. Discussion. The clinal trends and the wide overlap in characteristics thought to be useful in distinguishing Salix glauca, S. cordifolia and S. pseudolapponum (glaucops) support a treatment of these names as comprising a single species — Salix glauca. However, each of these names can be applied to geographic variants of the species which inter- grade through a series of intermediate populations. If these extreme variants, the intermediate populations, and Pleisto- cene events are considered, an hypothesis concerning the evolution of Salix glauca in North America can be proposed. First of all, it is possible to relate the present distribution SALIX GLAUCA COMPLEX 59 of each of the variants to a Pleistocene refugium. The Alaska-Yukon variant (S. glauca and vars.) may be related to a refugium in central Alaska, western Yukon, and the Arctic Slope of Alaska (Hultén, 1937; Flint, 1957); the eastern Canadian variant (S. cordifolia and vars.) may be related to a postulated refugium on the now submerged coastal plain of eastern North America and elsewhere south of the eastern icefront (Hultén, 1937; Raup, 1946); and the Rocky Mountain variant (S. pseudolapponum), not clearly distinguishable from the eastern variant in the data presented here, may be related to complex refugia along the eastern edge of the northern Rocky Mountains (Calder and Savile, 1959) and throughout the southern cordillera which was not continuously glaciated (Flint, 1957). If the popu- lations now represented by these variants survived the Pleistocene glaciations in these refugia they may have evolved their distinctive morphological features here under the combined pressures of different climatic conditions, hybridization, and isolation. However, it is possible that these variants evolved elsewhere and the refugia may repre- sent centers of survival and migration. Just how and when the three variants evolved cannot be determined by the data assembled here. However, if it is assumed that they did evolve from some common ancestry without developing reproductive barriers the numerous intermediate popula- tions now present in the glaciated portion of north central Canada may be attributed to the postglacial migration and “hybridization” of the variants. There is morphological evidence to indicate that since the end of the Wisconsin Stage populations from central Alaska have migrated as far east as Great Slave Lake; that populations from eastern Canada have migrated into the Great Bear Lake region; and that Rocky Mountain populations migrated north to the Yukon and eastward into Saskatchewan and Manitoba. For a discussion of this evidence see the systematic section. The area of overlap of these variants is the area in which the intermediate populations occur (Map 2). An evolutionary scheme similar to this has been proposed to explain variation in other plant and animal taxa e.g. Armeria maritima, Hultén, 1948; Abies, Haliday and Brown, 1943; Rana sylvatica (wood frog), Martof and 60 GEORGE W. ARGUS Humphries, 1959; and Zonotrichia leucophrys (white- crowned sparrow), Rand, 1948. Although this hypothesis may supply a plausible explanation for the geographic varia- tion observed in the components of Salix glauca in North America it may be found to be inadequate after future studies of polyploidy, hybridization, and population-varia- tion. The most important feature of the geographic varia- tion is the intergradation and clinal variation demonstrated for characteristics which have been used to distinguish between the various components of Salix glauca. SYSTEMATIC TREATMENT KEY TO SPECIES A. Petiole 3-10(-15)mm. long, usually yellowish; pedicel 0.5-1 (-2) mm. long; anthers usually longer than 0.5 mm.; leaves on repro- ductive branchlets and proximal leaves on vegetative shoots usually obovate to oblong, apex acute to obtuse; pistillate aments cylindrical, densely flowered. alix glauca. A. Petiole 1-3 mm. long, reddish (except in some ssp. niphoclada) ; pedicel 0-0.25 (-0.5) mm. long; anthers usually shorter than 0.5 mm.; leaves on reproductive branchlets and proximal leaves on vegetative shoots strap-shaped, apex rounded to obtuse; pistillate aments subspherical and densely flowered or narrowly cylindrical and loosely flowered. Salix brachycarp4- KEY TO PHASES OF SALIX GLAUCA olate to narrowly 8% long; pistillate f northwest B A. Leaves generally large, 4-10 cm. long, oblance L/W 2.8-4; stipules prominent 4-10 (-17) mm. aments 3.5-7 em. long; bracts light-colored; plants © Canada and: Alamlea. clisiil..cctistiiscksincideesdecemscbzsvennencsnisabbtatin steht 1 ta B. Leaves pubescent on both surfaces, becoming glabrate above; never villous-sericeous beneath; stipules variable in prominence; pistillate aments sometimes slender, not as long as in Western phase; plants of Arctic coastal Alaska and Yukon. «7 7" | Beringia phase obovate, se eeeeseneee A. Leaves smaller than above, 2.4-4 (-5) cm. long, elliptical to ie L/W 1.6-8; stipules inconspicuous, 0-4 mm. long; pistillate ai short, 2-4 em. long; bracts light-colored or sometimes dark; P ants . kenzie River, ae” Rocky Mountains, northern Canada east of Mac a os ceusensbaenesbeereoopenvernnrterra tea ummm C. Shrubs 3-4 (-10) feet tall; branchlets pubesce brescent, often pruinose; leaves narrowly elliptica leaves occasionally with ferruginous trichomes; bracts SALIX GLAUCA COMPLEX 61 to blackish; plants of Rocky Mountains, northern British Colum- bia to New Mexico. D D. Ferruginous trichomes lacking. ............ Rocky Mountain phase D. Ferruginous trichomes on leaves and sometimes on ovaries. Ferruginous form of Rocky Mountain phase C. Shrubs prostrate to 2-3 (-5) feet tall; branchlets white-tomentose to finely pubescent, pruinose in northern Quebec and Baffin Isl.; leaves elliptical to oval or obovate; ferruginous trichomes lacking; bracts light-brown; plants of eastern Canada and Greenland. Bead Eastern phase KEY TO SUBSPECIFIC TAXA OF SALIX BRACHYCARPA A. Pistillate aments subspherical, if cylindrical then densely flowered; leaves coarsely pubescent on both surfaces; shoot internodes short, leaves appearing fan-like; branches mostly thick and stout; style 0.5-0.8 mm. long; numerous aments borne below vegetative shoots. ssp. brachycarpa A. Pistillate aments long, narrowly cylindrical, loosely flowered; leaves appressed-pubescent beneath, thinly pubescent to glabrescent above; leaf apex acute-attenuate; branches thin and flexible; style 0.2-0.5 mm. long; few aments borne below vegetative shoots. —. oeeee B. Shrubs usually erect; leaves (2-) 2.5-3.2 (-4.5) cm. long; pistil- late aments 2-3.5 (-5) em. long; bracts narrow, oblong, yellowish ; branchlets densely tomentose, yellowish. «.-..-+++++++ ssp. niphoclada B. Shrubs prostrate; leaves (1.2-) 1.7-2.2 (-2.8) cm. long; pistillate aments 1.5-2.5 em. long; bracts broad, reddish, drying blackish; branchlets thinly pubescent, reddish DOWN. «-.-+---++-+s+sssssssersmesrerre z ssp. niphoclada var. fullertonensts Salix glauca L. Sp. Pl. 1019. 1753 Photograph of 1158.52 and 1158.53 in the Linnaean Herbar- ium, London : Previous treatments have considered Salix glauca in North America as three species, an eastern species (S. cordifolia and vars.), a northwestern species (S. glauca and vars.), and a Rocky Mountain species (S. pseudolapponum or S. glaucops). On the basis of studies of geographic population variation and the examination of a large series of herbarium specimens I have concluded that these three are conspecific and are best treated as variants of a single wide ranging circumpolar species, Salix glauca. Salix glauca can be distinguished from the related A brachycarpa on the basis of its longer petiole and . 4 its ovate to oblong leaves borne on the reproductive -_ od lets and at the proximal end of vegetative shoots, and i stout, cylindrical, usually densely flowered, pistillate aments. 62 GEORGE W. ARGUS The polyploid nature of the species is reflected in its longer stomata and anthers in comparison to the diploid S. brachy- carpa. Only the continental North American range of the species is treated here, although some comments will be made con- cerning the species in Greenland. It is hoped that the Eu- ropean and Asian representatives of the complex may be studied at a later date when field work is possible. Map 1. Range map of Salix glauca in North America and Greenland. Within the North American Salix glauca a large number of variants have been observed and named by earlier authors. Some of these variants are more or less correlated with certain geographic areas and as geographic variants would commonly be assigned the rank of subspecies or geogYr aphi¢e variety. I am reluctant to assign formal nomenclatural rank to these variants for several reasons. First, many of the characteristics used to distinguish the variants vary clinally (see section on geographic variation in S. glauca) and the area containing intermediate populations is very extenslve (see Map 2.). Second, the present treatment is concerned only with a portion of the range of S. glauca, and until the species is studied in Europe and Asia a well-founded infra- SALIX GLAUCA COMPLEX 63 specific classification is impossible. Third, a complete under- standing of the variation in this taxon in North America depends on further studies of hybridization, environmental modification, and polyploidy. And fourth, the use of in- formal nomenclature to designate the major geographic variants seems adequate and will facilitate discussion with- out implying a final solution or cluttering the already overwhelming nomenclature. For these reasons, the geo- graphic variants of S. glawca are termed “phases.” we ESS WS a PS Qa 9" | et Dg vr Th! 4 . SS | / EOGGC Pam Lip ya Se s 7 bm eo ir , }3 / — We Ly, aD waa m4 > Ga { Map of the ap America and Greenland. The areas occupied by the phases, the overlap between them, and the disjunctions are based on herbarium specimens. All boundaries are approxi- Four phases are recognized in North American Saliz glauca. Each of the phases has its center in a different geographic area. However, each of the phases is highly variable and their ranges overlap considerably. The Beringia phase is centered in arctic coastal and western coastal Alas- ka, the Western phase is centered in central Alaska and the Yukon, the Rocky Mountain phase is centered in the Rocky Mountains ranging from British Columbia to New Mexico, and the Eastern phase is centered in eastern Canada and Greenland. The major area of overlap between the phases occurs in the western Canadian Shield region (see Map Be 64 GEORGE W. ARGUS The center of each of the phases may be related to Pleisto- cene refugia and postglacial migration (see section on geo- graphic variation). The variation within each of the phases is complex and hybridization as well as geographic isolation has undoubtedly played a major role in their evolution. In this treatment, a general description of Salix glauca in North America will be followed by an amplified description and discussion of each of the phases. SYNONYMY OF SALIX GLAUCA (Western Hemisphere) S. cordifolia Pursh, Fl. Am. Sept. 2:611. 1814. ex char. desertorum Richardson, Frankl. Jour. App. ed. 1:753. (p. 25 in rep.) 1823; ed. 2:765 (p. 87 in rep.). 1823. Richardson 397, Ft. Frank- lin on the Mackenzie River, N.W.T. (K., lectotype; A!, fragment and photograph). S. glauca (ssp.?) desertorum (Richards.) Anderss. Ofvers. Vet.-akad. Férh. 15:127. 1858. S. callicarpaea Trautvetter, Nouv. Mém. Soc. Imp. Nat. Mosc. 2:295- 296. 1832. ex char. and illustration. S. cordifolia var. callicarpaea (Trautv.) Fernald, Rhodora 28:184. 1926. S. cordifolia ssp. calliear- paea (Trautv.) A. Love, Bot. Not. 1950:38. 1950. S. glauca ssp. callicarpaea (Trautv.) Bécher, Meddel. Groenl. 147:19. 1952. S. villosa Hooker, Fl. Bor.-Am. 2:144, 1838. Drummond 7, Rocky Mountains, Alberta, Herb. H.B.&T. (K, lectotype; A!, fragment and photograph). S. glauca var. villosa (Hook.) Anderss. Ofvers. Vet- akad. Forh. 15:127. 1858; also Amer. Acad. Arts and Sci. 4:22. 1858. S. X glaucops a villosa (Hook.) Anderss. in DC. Prodr. 16:281. 1868. S. villosa 8 acutifolia Hooker, Fl. Bor.-Am. 2:144. 1838. Richardson 76, Ft. Franklin on the Mackenzie River, N.W.T., Herb. H.B. & T. (& lectotype; A!, fragment and photograph; NY}, isolectotype)- glauca var. acutifolia (Hook.) Schneider, Bot. Gaz. 66 :327-329. 1918. S. glaucops B glabrescens Anderss. in DC. Prodr. 16:281. 1868. Bourgeau, Rocky Mountains, 1858. (GH!, apparent syntype)- glauca var. glabrescens (Anderss.) Schneider, Bot. Gaz. 66 2329-330. 1918. S. nudescens Rydberg, Brittonia 1:86. 1931. S. desertorum « elata Anderss. in DC. Prodr. 16:281. 1868. Drum coun 18830, Ellis Bay, Anticosti Is]l., Quebec, 7 Sept. 1883. (CAN, holotype; A!, photograph). S. rydbergii Heller, Cat. North Am. Pl. ed. 2:4. 1900. S. vacciniformis Rydberg, in Britton, Man. Fl. N. St Can. 319. 1901. S. cordifolia var. macounii (Rydb.) Schneider, Bot Gaz. 66:347. 1918. S. atra Rydberg, Bull. N.Y. Bot, Gard. 1:272. 1899. Bell 18825, Ford’s Harbor, Labrador, 31 July, 1884, (CAN!, lectotype). S. cord folia f. atra (Rydb.) Schneider, Bot. Gaz. 66:346. 1918. SALIX GLAUCA COMPLEX 65 S. labradorica Rydberg, Bull. N.Y. Bot. Gard. 1:274. 1899. Wag- horne 36, Turner’s Head, Hamilton Inlet, Labrador, 6 Aug. 1892. S. pseudolapponum v. Seemen, Engl. Bot. Jahrb. 29 (65) :28, 1900. Baker, Earle, and Tracy 300%, Mt. Hesperus, Colorado, 2 July 1898. (NY!, holotype; A!, GH!, MIN}, isotype). S. wolfii var. pseudolap- ponum (v. Seem.) Jones, The willow family of the Great Plateau, p. 17. 1908. S. glauca var. pseudolapponum (v. Seem.) Kelso, Biol. Leafl. 34:10. 1946. S. seemannii Rydberg, Bull. N.Y. Bot. Gard. 2:164-165. 1901. Williams, Dawson, Yukon Terr., 12 June 1899. (NY!, syntype). S. glauca var. seemanii [sic] (Rydb.) Ostenfeld, Skr. Vid.-Akad. Oslo, 1909 (8): 34-35. 1910. S. cordifolia f. hypoprionota Schneider, Bot. Gaz. 66:346. 1918, Fernald & Wiegand 3226, Blane Sablon, Labrador, 1 Aug. 1910. (GH!, holotype). S. anamesa Schneider, Bot. Gaz. 66:348-350. 1918. Lundholm, Tua, Greenland, 15-31 May 1889. (MO!, holotype). S. glauca var. acutifolia f. poliophylla Schneider, Bot. Gaz. 67:61. 1919. E. & A. Preble 139, Ft. Rae, N.W.T., 28 July 1901. (US!, holo- type). S. glauca var. poliophylla [as poliophila] (Schneid.) Raup, Jour. Arnold Arb. 17:233. 1936. S. cordifolia var. intonsa Fernald, Rhodora 28:185. 1926. Fernald & Long 28030, Deer Pond Brook, Highlands of St. John, Newfound- land, 20 Aug. 1925. (GH!, holotype & isotype). S. cordifolia var. eucycla Fernald, Rhodora 28:187. 1926. Fernald & Wiegand 3204, Pointe Riche, Newfoundland, 4 Aug. 1910. (GH!, holotype). S. cordifolia var. tonsa Fernald, Rhodora 28:187. 1926. Fernald & Long 27971, Ha-Ha Mt., Newfoundland, 5 Aug. 1925. (GH!, holo- type). S. cordifolia f. tonsa (Fern.) Polunin, Nat. Mus. Canada Bull. 1940. S. glauca var. aliceae Ball, Univ. Cal. Publ. Bot. 17:416-417. 1984. Eastwood 614, Whitehorse, Yukon Terr., 12 July 1914. (USI, Al, isotype). sy Fo daititihe\iuslch var. subincurva Kelso, Rhodora 363195. 1934, Kelso 3503, Rocky Mt. National Park, 5 Aug. 1931. (GH!, isotype). S. glauca var. subincurva (Kelso) Kelso, Biol. Leafl. $4:10. 1046. . glauca var. stenolepis Polunin, Nat. Mus. Canada Bull. 92:163- 164. 1940. Malie 118812, Lake Harbor, Baffin Isl., 25-26 Aug. 1927. (CAN!, holotype). . glauca tides eer ine var. sericea Hultén, Fi. Alaska and hocig 3:527. 1943. Mendenhall, Dall R., 55 mi. above mouth, 23 June % (US!, holotype and isotype). 66 GEORGE W. ARGUS S. glauca X farrae walpolei Cov. & Ball, of Hultén, Fl. Alaska and Yukon, 3:528. 1948. (in part). S.g eudomonticola Ball, of Hultén, Fl. Alaska and Yukon $:528-529. 1943. (including the “forms” subglabra and intermedia). S. pseudolapponum var. kenosha Kelso, Biol. Leafl. 25:3-4, 1944. L. & E. Kelso 534, Kenosha Pass, Colorado, 28 July 1936. (personal herbarium of L. Kelso, holotype). S. glauca var. kenosha (Kelso) Kelso, Biol. Leafl. 34:10. 1946. : a var. perstipula Raup, Sargentia 6:154. 1947. Raup & Soper 9321, Brintnell L., N.W.T., 30 June 1939. (GH!, holotype). DISCUSSION OF SYNONYMY Salix desertorum Richards. No specific name in this complex has caused more discussion, speculation, and con- fusion than this one. I have seen a photograph and a frag- ment of the lectotype (Richardson 397) and a sheet consid- ered by Schneider to be identical with the type (Richardson 70). Both of these specimens are juvenile and are similar to other juvenile specimens from Churchill, Manitoba (Ar- gus 67-58, 68-58, and 158-58), Coppermine, N.W.T. (Findlay 65), and Saskatchewan (Breitung 75). I regard the type material as representing juvenile material of S. glauca as found in central Canada. I agree with Raup (1959) that it has “no viable taxonomic status”, but I do not agree that it is necessary to regard it as “.. . an aberrant form of S. glauca caused by local site factors ...’’. In 1943, Hultén applied the combination Salix glauca ssp- desertorum (Richards.) Anderss. to material from western Alaska. He was of the opinion that this combination should be used for the Arctic American S. glauca distributed from Newfoundland to Alaska, including material named S. cordi- folia var. callicarpaea by some authors. Again the juvenile condition of the type-material of desertorum proved to be a stumbling block and all the material he cited from Alaska (I have seen all but two specimens) was juvenile. Hulten further described a var. sericea to include those specimens with leaves densely villous on the lower surface. The tyP® material of this varietal name is also juvenile and, in this case, it is the common form of the Western phase of »- glauca. An understanding of the ontogenetic development of Salix would have prevented this unnecessary descriptio? of a new taxon. Successive collections from the same shru SALIX GLAUCA COMPLEX 67 made throughout the growing season are useful in acquir- ing such knowledge. For example, compare Argus 311 (14 June 1955) and 781 (14 Aug. 1955), both specimens were collected from the same shrub near College, Alaska. The former is similar to var. sericea and the latter, using Hul- ten’s classification, to the hybrid S. glauca pseudomonti- cola. Both are simply S. glauca, Western phase. N. J. Andersson, in De Candolle’s Prodromus, recognized S. desertorum and three varieties, elata, stricta, and fruticu- losa. The latter two are S. brachycarpa and will be discussed under that species. The former is a juvenile specimen bear- ing ferruginous pubescence on the leaves. This form of S. glauca will be discussed later in reference to S. athabascensis and S. fallax. Salix glauca var. acutifolia (Hook.) Schneid, This varie- tal name has been recognized by many taxonomists with the notable exceptions of N. Polunin and E. Hultén. In 1940, Polunin noted the “fickleness of the characters” which iden- tify it, and preferred not to recognize this or any other variety of S. glauca. Hultén (1943) referred much of the material called var. acutifolia to his hybrid S. glauca * pseudomonticola. It is supposedly characterized by leaves mostly acute at the apex and somewhat smaller and more narrowly elliptic or lanceolate than the typical form. Raup (1959) is of the opinion that var. acutifolia shows geo- graphic segregation which justifies its varietal status. The holotype (S. villosa 8 acutifolia Hook., Richardson 76), of which I have seen a fragment and a sheet of the isotype, does have leaves that are sharply acute at the apex and narrower than usual; but this material is young, and other, better developed, Richardson collections (72 and 73) have leaves noticeably broader and less acute or with nearly ob- tuse apices. I cannot agree that this name represents a taxon that is of any value to the interpretation of the species. Its characteristics are highly variable and are influenced both by the age of the plant and by the environment. Salix glauca var. stenolepis Polunin. This name was given to material from northeastern and central Canada, ee od termediate between S. glauca and S. cordifolia, oe approaching much more closely the former. It ariel terized by long narrow bracts and stipules. The holotype 68 GEORGE W. ARGUS has persistent narrow stipules about 4 mm. long, and in the paratypes the stipules may be narrow as in Soper 132196 or very small as in Malte 126895. The bracts of the pistillate aments averaged about 2.5 mm. long and I could not find any 3-4 mm. long as described. I feel certain that the charac- teristics used to distinguish this variety are insignificant in the species. It is remarkable that Polunin, noted for his re- fusal to recognize any of the other varieties of Salix glauca, should have described it at all. Salix glauca var. perstipula Raup. This name was de- scribed in 1947; however, Raup later equated it with var. stenolepis after he saw Polunin’s type and noted that its stipules also were persistent. The var. perstipula was characterized by “. . . conspicuous, lance-attenuate, persis- tent stipules.” Although some American S. glauca do have persistent narrow stipules I do not feel that this characteris- tic deserves any greater stress than any other single charac- teristic. Furthermore, it has not been shown that it has any significant geographical or genetical importance, and it may simply represent an ecological modification. Salix glauca var. acutifolia £. poliophylla Schneider and S. glauca var. aliceae Ball have been used to identify mater- ial with permanently pubescent leaves and shoots. The former (f. poliophylla) was based on material from Great Slave Lake, N.W.T. Only one specimen was considered typical and others were thought to be intermediate between it and var. acutifolia. When Raup (1936) found that several of his collections from Lake Athabasca (Raup 4508 and 4597) had a similarly dense leaf pubescence he felt that this characteristic had a “. . . geographic affinity in the central part of the Mackenzie basin . . .”, and therefore deserved varietal status. He later (1947 and 1959) equated it with var. aliceae. When Ball described var. aliceae he knew of Schneider’s form but it is difficult to tell if he regarded it as synonymous with his variety or not. He claimed that var. aliceae, which “inclines toward S. cordifolia”, differs from S. glauca in its broader leaves sometimes with subcordate bases, broader stipules, longer capsules 8-10 mm.), and in the usually up divided styles. From the isotype before me (Eastwood $14 SALIX GLAUCA COMPLEX 69 Whitehorse, Yukon Terr.) I cannot understand Ball’s char- acterization. The leaves and stipules are certainly no broader than the species, the bases of the leaves are rounded, not subcordate, and long capsules are more typical of the species than his variety. Ball’s concept of this name changed with time and whereas in the beginning he included those specimens which were permanently pilose, he later included many specimens with leaves sericeous-tomentose below and glabrescent above, typical of the species in central Alaska. Our knowledge of pubescence-variation does not support the use of this characteristic as a proper basis for recognizing infraspecific taxa. Salix glauca X pseudomonticola, f. subglauca and f. inter- media of Hultén. In his treatment of Salix for the Flora of Alaska and Yukon, Hultén (1943) was very much influenced by the possibility of hybridization in the genus. The bulk of the material formerly called S. glauca var. acutifolia, var. glabrescens, and var. aliceae was identified as hybrids. I fully agree with comments by Ball and Raup that evidence of hybridization, especially between these two species, 1S difficult to support. Although hybridization does occur In Salix there is no concrete evidence for the origin or presence of this hybrid and to base an unwieldy taxonomy on such an untested hypothesis is undesirable (see discussion of hybrids). : : Salix glauca X farrae walpolei of Hultén. Six specimens from the Bering Straits District of Alaska were cited under this hybrid name (Hultén, 1943). One of the specimens (Jones 9045) was also cited as the only representative “2 Alaska of S. fullertonensis. After examining all but one of the cited specimens I find that two species are included in the material, S. glauca (Kellogg, 29 Aug. 1909; Jones 9034; and Anderson 5002), and S. speci pend ssp. nipho- clada (Jones 9045; and Walpole 1752). siviud “ae athabascensis Raup and S. f allax Raup. at toh some hesitation that I include these names In Salix 9 sig Both may represent hybrids including S. glauca iwc ‘ laris or even S. pedicellaris x planifolia. The rh ms these forms is highly uncertain and they are treated cong as synonymous with S. glauca because of numerous in 70 GEORGE W. ARGUS mediates linking them to S. glauca in populations at Churchill, Manitoba, northeastern Saskatchewan, and in the northern Rocky Mountains. The material which represents these two names is very interesting and its relationship to S. glauca deserves careful study. Salix athabascensis was described in 1930 and based on material from northeastern Alberta. The type-material is in late fruiting condition. Its glabrous capsules, long pedicels, and coriaceous leaves sug- gested S. pedicellaris Pursh, and its putative relationship to that species was discussed. Salix fallax was described four years later from material from eastern British Colum- bia. This material ig also in late fruiting condition. It was tentatively related to S. pedicellaris but no mention was made of S. athabascensis. In neither of this discussions of these species was S. glauca mentioned, although that species is strongly suggested by several characteristics, including the pubescent ovaries, pedicels, stems, and leaves; the nar- row neck of the ovary and capsule, and the long style. Material from the Canadian Rocky Mountains shows all forms of intermediacy between S. athabascensis, S. fallax, and S. glauca. The first taxonomist to note the resemblance of Salix athabascensis and S. fallax was Breitung (1957) and Scog- gan (1957) apparently concurred in this judgment. A. J. Breitung made numerous collections throughout central Saskatchewan which have contributed importantly to the understanding, such as it is, of these names. A charac- teristic of these “species” which has not been noted pre- viously is the occurrence of ferruginous trichomes on the leaves. These rusty-colored trichomes are often rare but Sometimes very common and may render the structure red- dish in color. This characteristic has been subsequently found throughout central Canada from Hudson Bay west- ward and definitely seems to link these names to the glabrescent forms of S. glauca, The type material of S. desertorum « elata Andersson also bears ferruginous ttl- chomes and further relates S. athabascensis and S. fallax to S. glauca. Salix pseudolapponum v, Seem. This names has been ap- Plied to Salix glauca in the southern Rocky Mountains. ! am unable to distinguish this small leaved alpine form from SALIX GLAUCA COMPLEX 71 S. glauca and prefer to regard it as a portion of the Rocky Mountain phase of that species. The Eastern phase of Salix glauca is represented by what has been called S. cordifolia, with numerous varieties and forms. The basic outline of its infraspecific classifica- tion was presented by Fernald in 1926. The variation in the Eastern phase is similar to that described in the Western phase, and I do not propose to recognize any of the variants as more than forms or ecological modifications. Salix cordifolia Pursh. I have not seen the type-specimen of this name, but the description gives a clear picture of it, and specimens having similarly broad leaves with cordate bases are known from Newfoundland (Fernald & Long 27078, 27977; Fernald & Wiegand 3219). The material in the Hooker Herb. from G. Anderson’s garden, of which | have seen a tracing, does not have the broad cordate leaves of Pursh’s description, but rather looks like S. callicarpaea. Salix callicarpaea Trautv. and S. cordifolia var. calli- carpaea (Trautv.) Fernald. These names represent the most common form of S. glauca in eastern Canada. | have not seen the type of S. callicarpaea and my judgment is based on the description and figure, both of which adequately characterize it. I agree with Schneider (1918b) that Traut- vetter’s concept of S. cordifolia and S. callicarpaea merely represents two stages of the same species. The former name is based on fruiting material and the latter name on flower- ing material. The name callicarpaea is characterized by its more or less glabrate leaves which are oblanceolate to elliptic with acute bases. Although this form is abundant through- out eastern Canada, it intergrades with the Western phase west of Hudson Bay. n : Salix glauca ssp. callicarpaea (Trautv.) Bocher. In his discussion of S. glauca in west Greenland Bécher includes the “smooth Greenland willows” in this combination. He does not agree with other taxonomists that the Saar willows should be called Salix callicarpaea because it an S. glauca are not distinct enough to be regarded oe we species. In this area he recognizes two a. es morphologically different types”, one broad-leaved and ee on sunny exposures, presumably S. glauca, ee ner narrow-leaved and smooth and growing in moister places, te GEORGE W. ARGUS S. glauca ssp. callicarpaea. If the Eastern phase of Salix glauca is to be treated as a subspecific entity this is the name that should be used. Salix cordifolia var. macounii (Rydberg) Schneider. I have not seen the type of S. macounii, but the description and the specimens later referred to the variety by Fernald depict a low, spreading shrub with small leaves tending to be elliptical to obovate with acute apices. This form inter- grades in one direction with var. eucycla Fern., whose leaves are more or less orbicular and rounded at the apex, and in the other direction with var. callicarpaea. Even one speci- men can almost span this variation (see Fernald, Long, & Dunbar 26570). It may be that the material from New- foundland with its orbicular leaves may represent intro- gression with S. arctica. Many of our specimens show other characteristics of S. arctica as well, including pruinose buds and stems, long straight trichomes on the underside of the leaves producing a “beard” at the apex. Salix cordifolia var, tonsa Fernald and var. intonsa Fer- nald. These names describe extremes of pubescence-varia- tion. The first is the glabrous form and the latter the pubescent form. Numerous intermediates connect the two making their value as varieties doubtful. Polunin (1940) reduced var. tonsa to a form on the grounds that it lacked any geographical range. In commenting on this reduction Raup (1943) wrote, “To be consistent, one should perhaps reduce them all (i.e. the other varieties of S. cordifolia) to forms, since some of the others show scarcely any better geographic segregation than does var. tonsa.” Salix anamesa Schneider. After examining the type of this name I find it to be simply S. glauca. I cannot find any sign of S. arctica in it, although hybridization between S. glauca and S. arctica is apparently common in Greenland. The Salix of Greenland are in need of a careful field study before the complexities found there can be understood. GENERAL DESCRIPTION OF SALIX GLAUCA Erect shrubs 1 to 3 feet tall, sometimes up to 15 feet, or prostrate. Branches reddish-brown to grayish, the epidermis often flaky, glabrate or variously pubescent, often with persistent pubescence, sometimes pruinose. Branchlets pubescent to white-tomentose, sometimes prui- SALIX GLAUCA COMPLEX 73 nose. Buds reddish-brown, pubescent and pruinose similar to the branchlets. Leaves highly variable in shape and size, from oval or suborbicular to narrowly lanceolate; length/width (1.2-) 2.0-3.8 (-5.4); length 3-9 cm. in the west, 2.5-4.5 em. in the east. Apex acute to attenuate, obtuse or sometimes rounded. Base tapering, acute or sometimes obtuse or cordate. Margin generally entire, sometimes with glands along the lower third of blade which may be raised on small teeth. Blade green above and glaucous beneath, often glossy above. Blade- pubescence varying from villous-tomentose on both sides to more commonly glabrate above and pubescent below. Petiole always longer than the bud, (2-) 4-10 (-16) mm., pubescent or glabrate and mostly yellowish. Stipules present, minute to 0.5-8 (-17) mm, long, glandular- margined, generally glabrous, variously persistent. Staminate aments coétaneous, cylindrical, borne on reproductive shoots generally shorter than those of the pistillate aments. Bracts light-brown to stramineous, ovate to oblong, pubescent on both sides to glabrate on the outer surface, usually with short wavy trichomes, sometimes with long straight trichomes, Glands usually in pairs, one adaxial and the other abaxial. The adaxial gland usually larger and often cleft into two lobes. The abaxial gland usually narrow, short and uncleft. Filaments two, free or united at the base, glabrous or pubescent near the base. Anthers reddish in preanthesis, light brown Branchlets white-tomentose, some simply pubescent. oblanceolate, some narrowly so, apex and base acu ; Blade usually pubescent on both sides, becoming glabrate el in age, the pubescence beneath sometimes of short appressed trichomes, never of the villous-sericeous type. Blade-size and te to attenuate. ate aments generally shorter than in the Western phase and loosely flowered, approaching - “iad rowly cylindrical type found in S. brachycarpa ssp. "2p Styles predominately divided, rarely entire. 74 GEORGE W. ARGUS nN. specime Salix glauca. rigs, 27 & 28. Representative ste eral Fic, 28. - riew. 2 569, 18 June vet Lappland Sweden. Fic. 27. General v & 30 ‘ingia phase. 26 0. Representative of the Beringia ¥ 4. Johnson, et al 1 i g. 195 seup of Bis ament. ska. > 1959. Bering IG. 30. Clos A. Johnson, et al. 03, 5 July 1959, Straits District, Ala SALIX GLAUCA COMPLEX 75 Distribution (Map 2). The Beringia phase occurs from westernmost Alaska Peninsula, probably including Unimak Island, eastward to Kodiak Island and Lake Iliamna. North- ward along the coast of western Alaska to the Seward Peninsula thence eastward across the Arctic slope of Alaska into the Mackenzie River delta area, N.W.T. The morpho- logical similarity of the species throughout this crescent is the reason for its inclusion in one phase and it is not a suggestion that the phase represents a single Pleistocene isolate. There is some doubt whether Salix glauca occurs on the Aleutian Islands, although it is known from the nearby Shumagin Islands and the Alaska Peninsula. I have seen a specimen (Beale, 11 June 1941) from Unimak Island which may be in the Salix glauca complex; however, its juvenile condition and insect damage preclude positive identification. The westernmost specimen of positive identity is Schofield 2331 from Cold Bay, near the tip of the Alaska Peninsula. Discussion. The Beringia phase is, in some parts of its range, virtually inseparable from “typical” European Salix glauca (compare Figs. 27 & 28 with 29 & 30). The material I have seen from Kodiak and the Shumigan Islands is re- markable in this resemblance and undoubtedly Coville’s (1901) opinion concerning S. glauca was strongly influenced by his experience with the species in this area. However, there is also a tendency for it to approach S. brachycarpa ssp. niphoclada in certain characteristics. This is especially noted in juvenile specimens. Characteristics such as narrow leaves, short petioles, and narrowly cylindrical aments, which ordinarily identify ssp. niphoclada are also found in the Beringia phase and they must be appraised with caution. Salix brachycarpa ssp. niphoclada occurs over most of the range of the Beringia phase with the exception of Kodiak Island and the western Alaska Peninsula. In these latter areas S. glauca does not seem to vary in the direction of ssp. niphoclada, suggesting that the intergradation else- where is the result of hybridization and introgression. The effect the environment may have in the formation of ap- parent intergrades is unknown. This problem is an im- portant one on the Arctic slope and the Seward Peninsula. The populations on the Arctic slope could, for example, be 76 GEORGE W. ARGUS regarded as hybrid swarms and the “typical” species in- frequent. A case for this view could be supported by the herbarium material I have seen. However, in the absence of experimental information concerning the relative influence of hybridization and the environment, it seems to be an undesirable one. In my treatment I have used S. glauca in a somewhat broader sense than S. brachycarpa ssp. nipho- clada. Hybrids have been recognized only when they are fairly certain. Actually, S. glauca, as characterized here, probably includes many specimens of hybrid origin. Her- barium material, in spite of its abundance, is still inadequate for a clear understanding of this problem. Intergradation of the Beringia phase and the Western phase occurs in the Cook Inlet and the foothills of the Arctic Slope regions. The former area is possibly a region where migrants from the Alaska Range and the Alaska Peninsula overlap, and the latter where migrants from central Alaska entered the Brooks Range. Ecology. The habitat of the Beringia phase varies widely. On the Arctic Slope, Spetzman (1959) reports it from flood- plains and cutbanks especially on the sand shores of rivers and streams. On the Alaska Peninsula it also occupies tun- dra habitats, and on Kodiak Island it occurs on sea cliffs, Carex, Sphagnum Swamps, and grassy slopes. (US), 58°45’, 156°37’, Ugashik L., 239* Schofield 2057A* » 2831 (DAO); Kodiak Is.: Sturgeon R., Coville & Kearney 2248 Lake Iliamna, Gorman 90* (NA, NY, US), 139* (US). Bering Sea ict: Coville & Kearney 1875 (A, US); mee A); Ogotoruk Cr. drainage, Johnson, Viereck, - 1909 (US) ; Teller, Scamman. 5453, Kotzebue, - and Norton Sd., B. Seemann 1423, Chamisso SALIX GLAUCA COMPLEX 77 Is., 1793 (NY); Noatak and Kugururuk R., Spetzman 4209 (MIN, A); Imuruk Basin, Cobblestone R., Walpole 1687 (US). Arctic Coast district: Cape Lisburne, Anderson 4498 (NA); Pitmagea R., Cantlon & Gillis 57-254, Jago L., 57-110 (GH); Okpilak L., Cantlon & Malcom 59-237 (GH) ; 69°26’, 151°28’, Chambers 165* (DS) ; Umiat, Churchill 134*, 166b*, 192*, 216*, 221a*, 283*, 373*, 383*,385*, 386", 494*, 496*, 548*, 563a*, 618*, 763*, (NA); LePage 23594* (NA); Wiggins 12931* (DS, RM); Hodgdon 8939*(US), 8973*, 8974* (GH), Kurupa R., 8890* (US); Kurupa L., Hodgdon & Riedeman 8602* (GH), above jct. of Kurupa and Coville Rs., 8904* (DAO); Pitmegea R., Shetler & Stone 3298 (GH, MICH); Kanayut Cr., Spetzman 1991 (DAO, MIN, NA), Sadlerochit R., 1033* (MIN, NA), Canning R. and Ignek Cr., 38&4 (MIN); East Oumalik, Ward 1502* (DS, RM). Yukon: Firth R., McEwen 164 (CAN); Head Pt., Oldenburg 44352 (GH); between King and Kay Pts., A. Porsild 7181 (CAN); King Pt., Lindstrom, 4 July 1906. WESTERN PHASE. FIGS. 31-33 Principal Synonymy. Salix glauca var. acutifolia, S. glauea var. glabrescens, S. glauca var. aliceae, S. glauca var. stenolepis, S. glauca ssp. desertorum var. sericea, and S. glauca * pseudomonticola ( pado- phylla). Amplified Description. Erect shrubs 3-7 or up to 15 feet tall. Branchlets long and pubescent, not tomentose, but sometimes densely pubescent, especially in the northern and eastern parts of the range. Leaves generally oblanceolate, 4-10 cm. long, characteristically dark- green and glabrescent above, and white villous-sericeous (in 50% of the specimens) or sparsely appressed-sericeous to glabrescent beneath. Petiole long, but shorter in relation to leaf-length than the Eastern phase. Stipules long and prominent, rarely absent. Pistillate aments long and densely flowered, but often loosely flowered at the base. Styles predominantly divided, but entire styles occur in 30-40% of the specimens. Distribution (Map 2). The Western phase of Salix glauca occurs throughout the western and central parts of Alaska and eastward to the Yukon Territory and into the Great Bear Lake and Great Slave Lake region. In Alaska it ex- tends northward into the Brooks Range where it overlaps the Beringia phase and southward into the Cook Inlet and Coastal Mountain area where it also overlaps the Beringia phase. In the vicinity of Great Slave Lake there 1s a major transitional region where the Western, Eastern, and Rocky Mountain phases overlap and intergrade extensively. The southern edge of the Yukon Plateau, near the British Columbia border, seems to represent the southern limit of this phase and here a narrow transitional region between it and the Rocky Mountain phase can be detected. 73 GEORGE W. ARGUS Representative specimens o FIGS. 31 & 32, G. Argu view, note prominent st FIG. 32. Closeup of Distillate ament. 1008, 9 June 1957, College, Alaska. FE wcd. f the Western phase — parser 8 613, 21 July 1956, Alaska Range, Alaska. Fig. gate apex: ipules and variation in the attenuation of the carga Fig. 33. Close-up of staminate ament. G. : . t Discussion. The Western phase of Salix apse ne is developed in the central Alaska-Yukon Plateau region. SALIX GLAUCA COMPLEX 79 a refugium for Salix glauca during the Pleistocene glacia- tions and it is possible that sometime during this isolation the Western phase evolved its unique character. The charac- teristic of villous-sericeous leaf-pubescence is distinct enough so that it can be plotted on a map (see Map 3) and used as AP 3. The distribution of villous-sericeous leaf-pubescence in the Western phase of Salix glauca. This characteristic is very abundant in central Alaska and western Yukon and becomes less common to the eastward. an indication of extent of migration. The phase probably had its center in central Alaska and from there spread in all directions. Although the map does not show its occurrence in western Alaska this is probably due to insufficient collect- ing in the area. To the eastward it reaches its apparent limit near Fitzgerald on the Slave River, and to the north it can be traced along the Porcupine River into the Mac- kenzie River Delta region and thence eastward to Copper- mine, where it is relatively abundant. In the Alaska Range it is prominent and as far south as the Coast Ranges it retains its identity, although it becomes highly variable in habit, leaf-size and pubescence, and ament-length. The wide variation in this part of Alaska may be related to 1) the Wide range of habitat variation in this mountainous region, 80 GEORGE W. ARGUS 2) hybridization with S. brachycarpa ssp. niphoclada, or 3) to intergradation with the Beringia phase. Two major transitional areas between the Western phase and other phases deserve discussion. The first is the transi- tion between the Western phase and the Rocky Mountain phase located in the general area of the Yukon — British Columbia border along the southern edge of the Yukon Plateau. The herbarium local population at Whitehorse, Y. T. (population No. 6, Figs. 21-26), although within the boundaries of the Western phase, shows a definite tendency toward the Rocky Mountain phase in its shorter leaves, stipules and aments, a lower leaf-length/petiole-length ratio, as well as a general absence of villous-sericeous leaf-pube- scence. Even as far north as Dawson, Y. T. certain in- fluences possibly attributable to the Rocky Mountain phase can be detected. In this area there is an increase in glabre- scence in leaves, shoots and aments. Such glabrescent indi- viduals have been described by Schneider as var. glabrescens and by Hultén as S. glauca * pseudomonticola f. subpseu- domonticola, The leaves are glabrescent or finely sericeous beneath and bear glands on the margin. These glands are not unique, but are simply more noticeable on the glabre- Scent specimens. The ovaries are sericeous, becoming glabrate in age. In some respects these specimens suggest the ferruginous form of the Rocky Mountain phase (see below) except for the absence of rusty colored leaf pube- scence. The second major transition occurs between the Western phase and the Eastern phase in the region between Great Bear Lake and Great Slave Lake. The species in this region 1s well illustrated by the herbarium local population study (Figs. 21-26). The populations from Great Bear Lake and Great Slave Lake are intermediate in most of the charac- teristics studied and smooth out an otherwise stepped cline (see section on geographic variation in S. glauca). The many specimens I have studied from the Great Slave Lake area support the view that intergradation between the phas- bess » almost complete in this region. Individuals charac- teristic of the Western phase (Fig. 34) (Thieret & Reich 4963, 4637, 7748; Thieret 4316; Lewis 849), and those char- acteristic of the Rocky Mt. and Eastern phases (Fig. 35) SALIX GLAUCA COMPLEX 81 + 2% a al et et 5, our glauca from the transit <5 . IGS. 34-35. Representatives of S¢ wlix gla from the t 5b . 24 A specimen ‘ ne the Western, Eastern, and “aes Mountain phases. 34 i : suggest t the Eastern or Rocky Mt. phases. 5. A specimen resembling the Roc ocky } Great Slave Lake. This sp aves. Thieret & Reich 5927, 82 GEORGE W. ARGUS (Thieret & Reich 4728, 5027, 5859, 5927, 5923; Lewis 842) are represented, but the majority of the specimens are In- termediate. The influence of the Eastern phase becomes more evident in the northern Lake Athabasca region. As discussed below, the Rocky Mountain phase also probably plays a role in modifying the variation of Salix glauca in central Canada, but the differences between it and the East- ern phase are so subtle that the influence of the two cannot be separated. : Ecology. The Western phase of Salix glauca is principally a shrub of forests and muskegs. In central Alaska and the Yukon it is very common in Picea mariana muskegs and occurs in openings and along drainage channels in Picea glauca forests. In the mountainous parts of Alaska, Yukon, and the Northwest Territories it also occurs in the forests, but here it extends into the subalpine Salix, Betula scrub, slide rocks, and in some areas above timberline. It is one of the pioneers on glacial outwash plains and is frequent along the gravel floodplains of glacier-fed rivers. In the western Northwest Territories it is prominent in the scrubby tundra as well as in muskegs and forests. SELECTED SPECIMENS. Alaska: Central Yukon district: Harding L., Scamman 6486* (GH); Salcha Slough, O. Murie, 19 June 1922* (NY, US); Fairbanks, Palmer 154* (NA); College, Argus 287* (GWA, RM), 311 (GWA, RM), 436*, 781 (DAO, GWA, RM), 1008, 1009*, 1020*, 1106*, 1163*, 1164*, 1165*, 1166*, 1167*, 1168*, 1169*, 1171*, 1174* (GWA), Chena Bluffs, 363* (GWA, RM); Ester Dome Rad., Williams 819* (NA); Alatna R., 30 mi. above mouth, 1901 (US); Galena, Péwé F-125 (NA); Ruby, Rouse 46 (NA); Kaltag, W. & C. Setchell 459a, Nulato, 495 (NA); Pedro Dome, Argus 569* (GWA, RM), Livengood Rd., 38 mi. from ject. with Steese Hwy., 1193* (GWA); Livengood, Scamman 1717*, 4888* (GH); Steese Hwy., Mi. 140, Cody & Webster 5352* (DAO); Miller House, Scamman 121*, 733*, 2042*, 5165* (GH), Independence Cr., 3493B* (GH); Circle, Anderson 2505* (NA); Dutilly, LePage, & O'Neill 20920* (NA); W- C. Setchell 389* (MIN, NA), 290*, 542% (NA), 543* (GH, NA, US), 544" (NA); Williams 489* ( 125* (GH); Ft. Yukon . + Marshall, 18 Aug. 1929* (NA); Arctic Village, — 3626*, mountains NW of Old John L., 2788*, Old John 8 3825 (NA); Shetler 948*, 1146* (ALA, GWA); Smith 25546 ~ SALIX GLAUCA COMPLEX 83 (GWA). Central Pacific Coast district: Kenai Pen., Ptarmigan L., Klein 60 (NA); between Skilak and Tustumena L., Palmer 5, 51, 58 (NA); Anchorage, LePage 23461 (DAO, NA); O. Murie, 2 Aug. 1922 (US); Tazlina Gl. L. & T. Viereck 2192 (GWA); Chitina, W. & C. Setchell 52 (GH, NA); Kennicott Gl. L. & T. Viereck 2207 (GWA). aska Range district: Mt. McKinley Nat. Pk.: Teklanika R., Argus 612* (GWA, RM), 613* (DAO, GWA, RM), 615*, 617* (GWA, RM), Camp Eielson, 642, Camp Denali 649* (GWA, RM), 650* (GWA), 651* (DAO, GWA, RM), Toklat R., 683* (GWA, RM), Igloo Cr., 691* (DAO, GWA, RM); Mi. 6, H. & V. Bailey 4612*, 4613*, Mt. Healy 4652*, between Toklat R. and Polychrome Pass, 4853*, Savage R., 5012*, 5030* (NA); Toklat Camp, Bolinger 18* (NA); Savage R. camp, Henderson 15054a* (NA); Muldrow Gl., Mexia 2137* (A, MIN, NA, NY), 2138* (A, GH, MIN), Wonder L., 2245*, 2264* (A); Mile 1, A. Murie 2*, Mile 7, 56*, Mile 50, 117*, 124*, Mile 69, 19*, Igloo Cr., 16*, 28* (NA), 66a* (NA, NY); Park Headquarters, A. & R. Nelson 3627* (NA, RM, US), Mi. 35, 3688* (GH, NA, RM, US), 3691* (NA, RM, US), Igloc Cr., 3764* (GH, NA, RM, US), Polychrome Pass, 3779* (NA, RM, US), Toklat Cabin, 4094* (GH, NA, NY, RM, US); Savage R., W. & C. Setchell 192* (GH, NA); Thorofare R., and Glacier Cr., Viereck 1095* (GH, GWA), Wonder L., Viereck 1713* (DS, MIN), Mi. 70-80, 1775* (GH). Phelan Cr., Argus 1145 (GWA); Richardson Hwy., Mi. 242, Cody & Webster 5730 (DAO); Summit .. W. & C. Setchell 1046 (NA); Black Rapids Gl., L. & T. Viereck 2133 (GWA). Yukon: Bear Cr., Calder & Billard 3221 (DAO, MIN, US); Dawson, Eastwood 386, 504 (A, NA, US), 812 (A); Williams, 12 June 1899 (NY); Hunker Cr., Macown 54385 (GH, NY); Mayo, Gillett & Mitchell 4205 (DAO); Whitehorse, Anderson 9606* (NA); Gillett 3838*, 3845* (DAO); Gillett & Mitchell 3246*, 4421*, 4514* (DAO), 3247* (DAO, RM); Porsild & Breitung 9142* (GH, US); Teslin, Argus 274 (GWA); Canol Rd., Mi. 105, Porsild & Breitung 9244 (GH, US); Mi. 127, 10012, Mi. 132, 9511, Mi. 136-138, 9777 (GH, NY, US): Burwash, H. Raup, Drury, & K. Raup 13975 (A); Kluane L., H. & L. Raup 12217, 12430 (A). Northwest Territories: District of koa nzie: Mackenzie Mts., Brintnell L., H. Raup & Soper 9228, 9321, . delta, east channel, Cody 10353*, Eskimo L. Basin, 10103A* (DAO) ; New Aklavik, Cody & Ferguson 9846*, Reindeer Sta., 10024*, 10026*, 10032*, 10033*, Williams Is., 10391*, Eskimo L., 10455*, 104 74%, 10525*, 10598*, 10675*, 10782*, 10828* (DAO); Aklavik, Cody & Gutteridge 7896*, 7913* (DAO); Caribou Hills, Cowan 54* (DAO), A. Porsild 16879* (GH); Aklavik, Dutilly 18060* (GH, NA). ROCKY MOUNTAIN PHASE. FIGS. pone ee 3 Principal Synonymy. Salix * glaucops, S. pse pponum, glauca var. pseudolapponum, and in the Tee Rocky Mountains and Plains S. fallax and S. athabascensis, in Amplified Description. Shrubs generally ei Ae tall, but ranging from 10 inches to 18 feet. Branchlets pubescent, commonly becoming 84 GEORGE W. ARGUS Fics, _ FIGS. 36-39. Representative specimens of the Rocky Mountain phase of Salix gla —— al view of a pistillate specimen in fruit. Smith sade Co! ng red auc is. 38. CG came cake: view. Fig. 39. Closeup of ‘abitnnte ament, note the dark- SALIX GLAUCA COMPLEX 85 glabrescent, often pruinose. Leaves generally small and narrowly elliptical, sometimes oblanceolate, pubescent on both sides or glabres- cent. Stipules present, mostly less than 0.5 mm. long, up to 4 mm. lo in some. Pistillate aments usually short and densely flowered, sometimes long and loosely flowered. Ovaries tomentose, capsules tend- ing toward glabrescence. Pedicels ranging from 0.25-1-1.75 mm. long. Styles short, about 0.5-0.8 mm. long, predominantly entire. Distribution. The Rocky Mountain phase of Salix glauca occurs in the mountains of British Columbia, Alberta, Mon- tana, Wyoming, Colorado, New Mexico, and Utah. Its ap- parent influence extends northward into southern Yukon and the Northwest Territories, and eastward across Saskat- chewan and into Manitoba, as far east as Churchill and York Factory. It is primarily a mountain element which overlaps the Eastern phase in the boreal forest and subarctic region between the Rocky Mountains and Hudson Bay. Discussion. The Rocky Mountain phase differs from the other phases in degree only, and individual specimens are often difficult to assign to this phase unless their geographic location is known. This phase is generally characterized by a tendency toward glabrescence in leaves, stems, and cap- sules, its short leaves, short undivided styles, and short to absent stipules. These characteristics are all in marked contrast to the Western phase and it is possible to distin- guish between the two even in the zone of intergradation in southeastern Yukon. However, the contrast with the Eastern phase is less well marked and presumably the inter- graduation between these phases is more complete than be- tween the Rocky Mt. phase and the Western phase. For purposes of discussion the Rocky Mt. phase will be divided into two parts, the northern Rocky Mountain phase and the southern Rocky Mountain phase. The dividing line between these parts can be represented by the International Boundary between the United States and Canada. An interesting feature of the variation in the northern Rocky Mountain phase is the occurrence of ferruginous tri- chomes. This, so-called, ferruginous form occurs widely throughout central Canada from eastern Manitoba to west- ern Alberta, and northward (Map 4). It is of special inter- est because in some areas it comprises a significant portion of the populations and two names, Salix fallax and S. atha- bascensis, have been based on it. 86 GEORGE W. ARGUS A PT hae Se | lak aus A p Map 4, The distribution of ferruginous leaf- and capsule-pubescence in Salix glauca. This type of rusty-colored pubescence is relatively common in Saskatchewan vie Manitoba and is frequent in Alberta. Th nt specimens which r to the ferruginous form bu ae r e open circles rep x are simila t lack the characteristic rusty-colored trichomes. In addition to the occurrence of ferruginous trichomes this form has spa rsely pubescent to glabrescent leaves, stems, and capsules, long pedicels (about 1.5-2 mm. long), ee widely spreading capsule-valves. Although sometimes is fo s rorm can be segregated on the basis of the above charac- teristics they are more out Salix glauca in adian Rocky Mountains, thinly pubescent viduals lacking ferruginous trichomes are open circles on Map 4). In northeastern to glabrescent indi common (see the SALIX GLAUCA COMPLEX 87 Saskatchewan and Churchill, Manitoba complete intergrada- tion has been observed between glabrescent and pubescent leaves and stems, glabrescent and pubescent ovaries and capsules, long and short pedicels, and ferruginous and color- less trichomes; so that any attempt to distinguish two taxa on the basis of these criteria becomes hopeless. The ferruginous form raises the question of the im- portance of hybridization to the variation-pattern of Salix glauca. The limited distribution of the ferruginous form in central Canada and its frequent association with sterile, putative hybrids, suggest that hybridization may be in- strumental to its formation. This possibility is supported by the resemblance of this form to other species and by the frequent occurrence of sterile ovaries. Its leaves often resemble those of S. pedicellaris in shape, texture, and vena- tion (Raup 9122, 10998, 1101; Thieret & Reich 4630; and Jack 2749). The long pedicels resemble those of S. pedicel- laris in Breitung 106, 166, and 542. The ferruginous pubes- cence may be linked to S. planifolia, S. scouleriana, or perhaps S. maccalliana. The occurrence of patchy pubes- cence on the ovaries also has been noted in the hybrid S. brachycarpa X chlorolepis. In some areas these indications of hybridization are more common than in others. Along the southern edge of the boreal forest in Saskatchewan (McKague) and as far north as the Lake Athabasca region the ferruginous form of Salix glauca may occur in company with S. pedicellaris, S. plani- folia and apparent hybrids of these taxa. In this region we may have the hybridization of three taxa producing a com- plex hybrid swarm, or it is possible that some specimens referred to the ferruginous form of S. glauca are really the hybrid S. pedicellaris X planifolia. However, in other areas including Churchill, Manitoba, indications of hybridization are absent and the ferruginous form occurs in populations in all habitats. Specimens such as Argus 517-58, 490-58, 456- 58; Breitung 2154; and Scoggan 6062 and 6088A represent the ferruginous form and yet produce good seed and show no obvious signs of hybridization. In this treatment the ferruginous form is dealt with as a part of the Rocky Mountain phase of Salix glauca. At the 88 GEORGE W. ARGUS same time, I recognize that some of the individuals which exhibit this character are probably hybrids which in some cases may not even include S. glauca. Any attempts to recognize this variant as a specific taxon (either S. fallax or athabascensis) seem unsupportable at this time. Studies now underway in northern Saskatchewan may result in a clarification of this problem and perhaps a modification of this treatment. The southern part of the Rocky Mountain phase, particu- larly in southern Wyoming and Colorado, presents such a high degree of complexity that I have been unable to fully understand it on the basis of herbarium material alone. The phase in this area has long been a source of confusion (Ball, 1899; Schneider, 1918b; and Smith, 1942) and will require detailed field study to be fully resolved. Ag has been noted previously, in this area Salix glauca has gone under the name S. pseudolapponum or S. glaucops. The literature reveals perennial confusion and vacillation concerning the nature and diversity of the species in this area. Rydberg (1917) recognized S. pseudolapponum, S. glaucops, S. seemannti, and S. desertorum; the latter two names are synonyms of the Western phase of S. glauca. Smith (1954) recognized S, pseudolapponum, S. glauca glabrescens, and S. cordifolia, the latter names representing the Western and Eastern phases respectively. Kelso described several varie- ties of S. pseudolapponum which he later (1946) transferred to S. glauca as his concept of the species changed (see dis- cussion of synonymy). Each of these authors recognized that S. glauca occurs in the southern Rocky Mountains in ; An examination of a large series of specimens makes the reason for the confusion evident. BF oF; although the material has certain distinctive charac- teristics such as plane, glabrate, short-petiolate leaves ; glab- bracts; yellow branchlets; and stom- most accurate to regard them as a part of the Rocky Mountain phase of S. glauca. SALIX GLAUCA COMPLEX 89 Three factors which seem to have influenced the variation of Salix glauca in the southern part of the Rocky Mountains are hybridization, isolation, and phenotypical modification. Hybridization has probably played a major role. I suspect that hybridization hag been, and is, more widespread in this part of the species range than in any other part of North America. Salix brachycarpa ssp. brachycarpa seems to be commonly involved in hybridization with S. glauca and in some areas, such as the Medicine Bow Mountains of Wyoming, hybrid swarms may occur. Hybridization has gone on to such an extent that neither of these species is completely recognizable (Argus, 1957). However, an accu- rate assessment of the situation will require field-study and experimentation. Other species which may be involved in hybridization with S. glauca are S. arctica, S. wolfii, S. wolfit var. idahoensis, and S. eastwoodiae (including S. orestera). The second factor, isolation, raises the question of recent morphological divergence of populations. Repeated refer- ence has been made to the influence of the Pleistocene glacia- tions and the importance of isolation to the divergence of the major phases of Salix glauca. However, today in the southern Rocky Mountains there exists a reverse situation in which populations which were probably more continuous during the Wisconsin glaciation are now relatively isolated in what may be called “interglacial refugia”. Populations in such areas as the Big Horn Mts., and the Medicine Bow Mts. of Wyoming, in southwestern Utah, and throughout the mountains of Colorado, are probably more isolated than the range-maps would suggest. The degree to which such isolation permits genetical divergence is as yet unknown for S. glauca and its relatives, but it may prove to be considera- ble. The third factor, phenotypical modification, should be evaluated, especially in reference to elevation. Specimens from Colorado (H. Levi) and Wyoming (C. L. Porter) which were cultivated in the greenhouse and nursery at Boston, Massachusetts, showed such a remarkable pheno- typical plasticity that the importance of such modification must be given careful consideration. Phenotypical modifica- tion may assume greater importance than usual in the southern Rocky Mountains where great environmental ex- 90 GEORGE W. ARGUS tremes are so closely related to elevation differences. This problem will require field study coordinated with transplant experiments. ; Ecology. The Rocky Mountain phase occurs in a variety of habitats from openings in spruce woods, groves of Populus tremuloides, pine forests, and along river and lake shores below timberline. It is less abundant on alpine slopes in the Canadian Rocky Mountains, although in the Rocky Moun- tains of the United States it is very abundant in both alpine and subalpine situations, and occurs up to 12,000 feet in parts of Colorado. The ferruginous form seems to be most common in Picea mariana muskegs and bogs along the south- ern edge of the boreal forest extending northward; it also may occur in willow thickets along rivers in the boreal forest. SELECTED SPECIMENS. Alaska: Eastern Pacific Coast district: Glacier Bay, Cooper 166 (US), 200 (NA); Skagway, Eastwood 793* (A). Yukon: MacRae, Gillett 3529*, 3530* (DAO), Careross, 7788* (DAO); L. Bennett, Seale, June 1898*, (DS). British Columbia: Haines Rd., Mi. 81, Taylor, Szezawinski, & Bell 1383 (UBC), Mi. 98, 973, 975 (DAO, UBC); Atlin L., Aitken 18*, 19*, 20* (DAO) ; Eastwood 655*, 657" (A, US); Atlin, Setchell & Parks 10* (NA); Bennett, Malte 122171* (A); Mitchell 177* (DAO), 178* (DAO, UBC); Vista, Wal- bole 1273*, 1275* (US); Azouzetta L., Calder, Savile, & Ferguson 14092 (DAO); Caribou, Flying U. Ranch, Eastham, 11528 (NA, UBC); White Pas t Pass, Alaska Hwy., Taylor, Szezawinski, & Bell 101 (DAO, UBC); Elk R. valle (GH, NA, NY, UBC. , Fraser 9*, 13* (NA); asca R., Jack 2574* (A, NA); Signal Mt., 2610* (A), trail to Whistlers Mt., 2695* (A) 2704* (A, NA) ; Beauvert L., Macoun 95382 (A, GH, NA, NY), 196* 95797*, Brule, 99*, 109* (A), 111* (A, » 116", 117* (A), Old Man Mt., 133* (A, DAO), SALIX GLAUCA COMPLEX 91 below Mt. Cavell, A. & R. Nelson 4854* (NA); Angel Gl., Mt. Cavell, Scamman 2441*, Medicine L., 2520*, mountain above Maligne L., Pk.: Lake Louise, Eggelston 21756 (US); Jack, 10 Aug. 1904 (A); 2618*, 2619* (GWA), Columbia Ice Field, 2745* (GH); Banff Nat. Macoun 68894 (GH, NA, NY); Castlemount Ranger Sta., Malte & Watson 116769 (A, GH, NA); Bow summit, Weber 2467 (GH, NA, Y); Waterton Lakes Nat. Pk., Chief Mt. Inter. Hwy., Breitung 15806 (NA, NY); Wood Buffalo Pk., Pine L. district, H. Raup 2186 (US). Montana: Glacier Nat. Pk.: Jack 1539*, 1540* (A); Gunsight Pass, H. & V. Bailey 1533* (NA); Many Glacier, Jack 2056*, trail NA, US). Fergus Co.: Big Snowy Mts., Hitchcock & Muhlick 11986 (DS, GH, NA, NY). Wyoming: Albany Co.: Nash’s Fork, A. Nelson 7831 (RM); La Plata Mines, E. Nelson 5166 (RM). Fremont Co., Christina L., Porter 5231 (DS); Lincoln Co., Dead Mans Pk., Williams 1289 (NA, NY, RM); Park Co., Clay Butt, Porter & Rollins 5872 (RM); Sheridan Co., Big Goose Cr., Tweedy 216 (US); Sublette Co., near Green R. Lake, FE. & L. Payson 4570 (GH, NA); Teton Co.: Hoback Canyon, Wehmeyer, Martin, & Loveland 5467 (NY); Yellowstone Nat. Pk., Amethyst Mt., Knowlton, 18 Aug. 1887 (US). Colorado: headwaters of Clear Cr. and alpine ridges east of “Middle Park”, Parry 341 (GH, NY); Boulder Co.: Redrock L., Ramaley & Robbins 5071; Smith 2272 (NA). Chaffee Co.: Cotton- wood Canyon Rd., Beetle 2173 (GH). Clear Creek Co.: Pass, Clokey 3752 (GH). Custer Co.: Brush Cr., Cockerell, 1899 (NY). Delta Co.: Grand Mesa, Maguire 12756 (NY). Eagle Co.: Tennessee Pass, Eggleston 11783 (US). El Paso Co : Bacigalupi 771 (GH). Garfield Co.: Newcastle, Harrington 4317 (NA). Grand Co.: near Eldora, Demaree 38706 (NA). Gunnison Co.: Virginia Basin, Langenheim 298 (NA); Blain Basin, Levi 3 (GWA). Gunnison-Pitkin Co. line: E. Maroon Pass, Levi 6 (GWA). Hinsdale Co,: Slum Gullion Pass, Levi 1 (GWA). Jackson Co.: mts. west of Cameron Pass, Williams 2435 (US). Ridge, Rocky Mt. Pk., Smith 2239, 832 (NA), 1779 (GWA). Co.: Leadville, L. & E. Kelso 4886 (DAO). Mesa Co.: top of Grand al ; Routt Co.: Ethel Pk., Gooding 1893 (NA). ac. s guire 14144 (NA). Duchesne 92 GEORGE W. ARGUS Co.: Uinta Basin. Mt. Emmons, Grahm 8518 (NA, US). Iron Co.: Dixie Nat. Forest, Cedar Breaks Canyon, Maguire 17561 (NY). Kane Co.: northwest of Orderville, Maguire 18831 (NA). Rich Co.; Bear R. Canyon, Watson 1100 (GH, US). Salt Lake Co.: Big Cotton- wood Canyon, Garrett 1595 (US). Summit Co.: Uinta Mts., Henry Fork Cr., B. Maguire, Hobson, & R. Maguire 14361 (NY). Utah Co.: trail to Mt. Timpanogos, Maguire 17485 (NY). Wasatch Co.: near Midway, Carlton & Garrett 6707 (GH). New Mexico: Castilla alley, Wooton, 4 Sept. 1913 (NY, US). Colfax Co.: Baldy Pk., Standley 14364 (US). Taos Co.: Taos Mt., Bailey 884, 885 (US). SELECTED SPECIMENS OF THE FERRUGINOUS ORM, ROCKY MOUNTAIN PHASE Alaska: Clearwater R., Argus 827 (GWA, RM). British Columbia: Beaton R. Crossing, Taylor, Szczawinski, & Bell 64 (DAO, UBC); Alaska Hwy., Mi. 347 and Liard R., H. Raup & Correll 10998, 11001, 11005 (A); Liard Hot Springs, 10907 (A). Alberta: Calgary, Macoun 94426 (NA); Jasper Nat. Pk.: Jack 2749 (A, NA); Mt. Cavell, Ledingham 49-645* (DAO); Athabasca R., Macoun 95377 (A, GH, NA, NY); Moose L. dist., Raup 8122 (GH), 8125 (CAN); Edmonton, Turner 1299, 1782 (A, DAO, NA). Manitoba: Otterburne, Bernard 58/398 (MTSM); Riverton, Scoggan 9024 (CAN); York Factory, 6062 (CAN, MIN), 6088A (CAN); Churchill: Argus Saskatchewan: McKague, Breitung 73 (A, DAO, NA, NY), 162 DAO, NA, NY), 542 (A); Prince Albert, ae EASTERN PHASE. FIGS. 6, 6a, 40-45 a Synonymy. Salix cordifolia, S. callicarpaea, S. cordifolia r. ‘ca ‘carpaea, S. labradorica, S. glauca var. stenolepis, S. glauca ee callicarpaea, and in Greenland S. anamesa. tiga ete Description. Shrubs prostrate to 5 feet tall, averaging o 3 feet tall. Branchlets sometimes white-tomentose in the escent below denne’ lire usually glabrate above and thinly pub- the base in some. pubescence uncommon, margin entire, glands near Stipules present but not prominent and rarely SALIX GLAUCA COMPLEX 93 exceeding 1 mm. in length. Pistillate aments short, 2-5 cm, long in ruit. Ovaries rat tomentose ~ oming glabrescent to thinly pub- escent in age. Pedicel from 0.25 to more commonly 0.5-1 mm. long. Style often short and entire, some divided part way to the base, reddish in preanthesis and yellow-green in anthesis ix glauca. Fernald cimen of the Eastern phase of Salix 9 Folbe f aeme ap Mt. Jacques Cartier), Quebec n & poor at 212, 5 and 12 Aug. 1906, Tabletop Mt. h in up i Fic. 40. General view, note persistent ament of previous year branc pp center. FIG. 41. Closeup of pistillate ament. 94 GEORGE W. ARGUS from F Representative specimens of the Eastern phase of Saliz _ si Churchill, Manitoba. Fic 42-43 ig 8. Staminate specimen, Argus 4S, 15 42. General view, note pubescent nrmneies and shiny, glabrescent b Closeup of s nate ament. Fic. 44, An erect specimen of Salix glauca, 3.5 feet growing in ‘ic. fae of the outcrop idee at Churchill, Manitoba (Argus A prostrate individual of S. glauca growing on the outcrop ridge a ait from the individual in FIG. 44 (Argus 501-58). See Figs. 6 & 6a for ‘aptbtitnbe speci- men from Churchill, Manitoba. SALIX GLAUCA COMPLEX 95 Distribution (Map 2). The Eastern phase of Salix glauca occurs from southern Keewatin and northern Manitoba southward along the coast of Hudson and James Bays into northern Quebec, Labrador, and Island Newfoundland. It extends northward into southeast Baffin Island and Green- land, and southward to the southern coast of Quebec, the Mingan Archipelago, Anticosti Island, the Shickshock Mts. of the Gaspé Peninsula, and St. Paul Island. The influence of the Eastern phase extends westward into the Great Slave Lake region where it intergrades with the Western and Rocky Mountain phases. Discussion. The Eastern phase of Salix glauca has long been regarded as a distinct species, S. cordifolia (usually as the variety callicarpaea). All the evidence I have been able to accumulate supports the conclusion that it is best regarded as being conspecific with S. glauca. However, if its chromosome-number is shown to be consistently different from S. glauca, as suggested in the section on cytology, it may have to be reassigned specific rank. The Eastern phase is relatively homogeneous throughout its entire range, especially east of Hudson Bay. This is in contrast to the other phases which vary widely and contain scattered individuals morphologically resembling other phases. I have not seen a single specimen from east of Hudson Bay that could be assigned to one of the other phases. In northern Manitoba the Eastern phase, although the dominant form, shows an increasing influence of the Rocky Mountain phase. This influence is reflected in a general diminution of leaf, stem, and capsule-pubescence, slightly longer pedicels and petioles, and the occurrence of occasional ferruginous leaf-pubescence. The influence of the Western phase is noted in central Keewatin and to the west where oblanceolate leaves, occasional villous-sericeous leaf- pubescence, and longer stipules and aments begin to appear. At Churchill, Manitoba, Salix glauca is an ubiquitous spe- cies occurring in most habitats and is dominant in several. Although morphological variation in the species 1n different habitats is not evident, variation due to microenvironmental differences, such as protected niches, or south-facing expo- sures, is striking (compare Figs. 45 & 44, the former is @ prostrate individual growing on the outcrop ridge, and the 96 GEORGE W. ARGUS latter is an erect shrub growing in the lee of the same out- crop ridge). It is under such microenvironmental conditions that forms referred to the Western phase by some taxo- nomists have been produced (C. R. Ball, in Ritchie, 1956). East of Hudson Bay the Eastern phase occurs throughout the region roughly corresponding to the forest-tundra transition and the artic tundra of eastern Canada. In this area leaf-pubescence is highly variable, ranging from dense- ly pubescent on both surfaces to almost completely glabres- cent. These extremes are connected by a continuous array of intermediates. In some areas (especially James Bay) densely pubescent specimens suggest hybridization with Salix brachycarpa. Similarly, pruinose stems found in about 20-30 per cent of the specimens from northern Quebec and Baffin Island suggest hybridization with S. arctica. How- ever, in neither case is the evidence unequivocal. Pruinose stems may have been derived through hybridization and introgression with S. planifolia as well as S. arctica, and such characteristics may represent introgression from “ancient” hybridization which is no longer going on and the characteristics may now be fully integrated into the species. On Baffin Island Salix glauca and S. arctica appear to con- verge in their morphology, and although hybrids may occur they are virtually impossible to identify because of being s0 similar to their putative parents. This problem of hybridi- zation and the recognition of hybrids is acute in Greenland as well as on Baffin Island and critical field studies are very much needed. Cytological information may also be of value because S. arctica is tetraploid and S. glauca is either hexa- ploid or octoploid. Along the Labrador coast Salix glauca exhibits very wide phenotypical variation. Shrubs range from prostrate to 4-5 feet tall and in some specimens broad leaves with subcordate bases occur, However, in general the shrubs are 1-2 fee tall and the leaves elliptical-oval and obtuse at both ends. n Newfoundland, numerous specimens with small oval to suborbicular leaves of the “macounij” or “eucycla’”’ type have been collected. These leaf types occur as prostrate individuals and have been collected also from the Strait of Belle Isle coast and other localities on the western coast. There are also erect forms and various intermediates in this SALIX GLAUCA COMPLEX 97 area but there seems to be a definite tendency toward pros- trate, small-leaved forms. In this area the specimens show a closer morphological resemblance to those from the Labra- dor area than to those from the Gulf of St. Lawrence. However, in spite of this considerable variation in vegeta- tive structure there is very little variation in their repro- ductive structures. The Eastern phase does not seem to occur in the spruce-fir forest of southern Quebec but appears only along the coastal fringe which experiences a rigorous environment apparently unfavorable for tree growth. It is fairly common on the Mingan Archipelago but has been collected from only one locality on Anticosti Island. The specimens I have seen from this general area are thinner-leaved than usual and have very short pedicels. A southern outlier of the Eastern phase in the Shickshock Mts. probably represents an isolated remnant of what was once an extensive population in eastern North America. It has been collected on Mt. Logan, Mt. Jacques Cartier ( Table- top Mt.), and on Mt. Albert. In the latter locality it is rare and causes some confusion with Salix brachycarpa which tends to vary in the direction of S. glauca in favorable habitats and may even hybridize with it. Ecology. The Eastern phase of Salix glauca is the com- monest willow along the east coast of Hudson Bay (Hustich, 1950), and throughout its entire range in eastern Canada. It is principally a species of the forest-tundra transition, extending into the adjacent tundra and the boreal forest. In the forest-tundra of Quebec and Labrador it occurs In Picea glauca forests and Picea mariana, Larix laricina muskegs, in meadows, and on sandy and gravelly beaches, and river margins. In Newfoundland it is common along the sea cliffs both on quartzitic cliffs and the serpentine tableland (Bonne Bay). It also occurs in the lichenheath tundra, spruce thickets, and on dry limestone barrens. In the region of the Hudson Bay lowlands it occurs on river banks, coastal meadows and gravel ridges. It is most commonly collected from river flood plain thickets, and gravel or clay river banks. At Churchill, Manitoba, it occupies a variety of habitats from the tundra of the out- crop ridge to dense willow thickets along the Churchill 98 GEORGE W. ARGUS River, and in both Picea glauca and Picea mariana, Larix laricina forests. It is much less common in the forests than in the tundra or along water courses. In the Gulf of St. Lawrence area it occurs principally in calcareous situations such as limestone-shingle beaches, calcareous slopes and cliffs, but it also occurs on granitic cliffs. It is an alpine species on the Gaspé Peninsula where it occurs only on the mountains in alpine and subalpine meadows, brooksides, and on cliffs and talus slopes. On Mt. Albert it occurs in both serpentine and amphibolite situa- tions. SELECTED SPECIMENS. Northwest Territories: District of Keewatin: (GH) ; Christie Is., Dutilly, O’Neill, & Duman 87861 (NA, US); Cairn Is., E. & L. Abbe, & Marr 3481* (DS, MIN), 3540* DAO); E. & L. Abbe 3540*, 3541*, 3037* (MIN). Manitoba: Nueltin L., Baldwin 2249 (GH), 2250 (CAN, GH); Churchill, Argus 4-S, 34-S, 196-58, $36-58, 364-58, 414-58*, 431-58*, 434-58", 436-58*, 438-58*, 457-58", 462-58", 465-58", 470-58*, 471-58*, 482-58*, 485-58*-488-58*, 503-58", 507-58*-509-58, 523-58, 524-58* (DAO, GH, GWA, RM) ; Macoun 79159 (A, GH, NY); York Factory, Scoggan 5942 (CAN, MIN); Nelson R., 6341, 6360 (CAN, MIN). Ontario: Moosonee, Dutilly & LePage 12075 (GH); Cape Henrietta Marie, 21245 (NA): jet. of Fawn R., and Savile 2277, 229%, 230%, 275*, 276*, 388* (DAO), 417* (DAO, US), 418", 612*, 681*, 687*, 755* (DAO) ; Richmond Gulf, Dutilly & LePage 13120*, 13213* (GH); E. & L. Abbe 3303* (MIN); Wiachewan R., mi. from bay, Rousse L., 628 (DAO, US), 778 (DAO) ;P L, 892 (GH), 981 (DAO); Wakems, ayne L., Rousseau (GH) Bay, Gardner 719 (GH, NA); Malte 126917 (GH, SALIX GLAUCA COMPLEX 99 NY, US); Ft. Chimo, Calder 2276 (DAO, RM); George R., various localities, Rousseau 531*, 597*, 654*, 659*, 694*, 697%, 714*, 774%, 800*, 804*, 876*, 878B*, 902*, 903*, 921*, 935*, 998*,1089* (DAO), 598*, 878A (DAO, GH); Korak R., 70 mi. from bay, 860 (NA); Mollie T. Lake, Harper 3758 (MIN); Knob L., Hustich 450 (GH); St. Law- rence region: Mingan Is., Brunel 149* (GH); St. John 90836*, 90838*, 90846* (GH); Victorin & Germain 18915*, 18927*, 25187 (GH, US), 18916*, 18923*, 18932*, 25932* (GH), 18926*, 25188 (GH, NY), 25184*, 25922*, 25925* (A, GH), 25183* (A, GH, US), 18925, 18930*, 18931 (GH, NY, US), 18922* (DS, GH, NY, US), 18924* (DS, GH, NY); Anticosti Is., Bay St. Clair, Victorin 4351 (A, GH, NY, US); Strait of Belle Isle, Blanc Sablon, Fernald & Long 28031* (A, GH); Fernald & Wiegand 3223* (GH), 3225* (A, GH, NA), 3224* (A, GH, NA, NY, US); Griscom 1* (GH); Gaspé Pen., Mt. Auclair, Louis-Marie, et al. 34402 (GH, NA); Mt. Blane, Pierce & Hodge 9A (GH, NY); Mt. Logan, Lévesque 48004 (DAO, NA); alpine pass bet. Mts. Logan and Fortin, Fernald & Pease 25008 (A, GH, NA, NY, US); Pease Basin, bet. Mts. Logan and Pembroke, Fernald, Griscom, & Mackenzie 25673 (A, GH, NA, NY); Tabletop Mt., Fer- (A, GH, MIN, NY, US); LePage 3898 (NA, RM). Newfoundland: Labrador: Komaktorvik Fjord, Dutilly, O'Neill, & Duman 7865 horne 4* (A), 7* (MIN, US); Island Newfoundland: St. Anthony, Abbe 238 (GH, MIN); Trinity, Ayre, Aug. 1935 (GH); Highlands of St. John, Doctor Hill, Fernald & Long 27977, 27978 (A, GH); Flower Cove, Fernald, Long, & Dunbar 26569 (A, GH, NY), 26570 (A, GH); Bonne Bay, Fernald, Long & Fogg 1612 (A, GH, NA, NY); Fernald & Wiegand 3230 (A, GH, NA, NY); Ingornachoix Bay, $219 (GH); Cape St. George, Mackenzie & Griscom 11024 (DS, NY, US). Nova Scotia: St. Paul Is., Eskine 58810 (DAO); Perry & Roscoe 170 (A, GH, NY). 100 GEORGE W. ARGUS SELECTED SPECIMENS FROM TRANSITIONAL AREA BETWEEN THE WESTERN, EASTERN, AND ROCKY MT. PHASES Northwest Territories: District of Mackenzie: Coppermine, Findlay 65, 218 (DAO, RM, US); Norman Wells, Cody & Gutteridge 7441 (DAO); Ft. Good Hope, Preble 330, 332 (US); Great Bear L., Port Radium, Cody 2790* (DAO), Gunbarrel Inlet, 2827*, 2828* (DAO); Sawmill Bay, Corcoran 15* (DAO); Leith Pt., Preble 291*, 292*, 293*, 294 (US); Ft. Franklin, Richardson 58 (NY), 70 (GH), 397 (A), Mackenzie R., Ft. Franklin?, 72, 72 (NY), 76 (A, NY), 399 (GH); Port Radium, Shacklette 2752*, 2881*, Leith Pen. 3081a*, 3100*, 2838*, McTavish Arm, 2792*, 3200*, Hunter Bay, 3211, 3212 (NA, US). Indin L., Cody & McCanse 3357 (DAO); Great Slave L., Snare R. Power Sta., Cody 2627* (DAO, US), Yellowknife, 7 Aug. 1949* (DAO); Cody & McCanse 2403* (DAO), 2404* (DAO, MIN), 2493*, Harditsy Is., 2907*, 2920* (DAO); Moraine Pt., Lewis SS) Ss nN Oy * . Plateau, 2140 (GH, NY, US); L. Athabasca, Sand Pt. H. Raup & 4528 (GH, MIN, NA, NY). Saskatchewan: Lake Athabasca: Charlotte Pt., H. Raup 6178*, 6181*, 6182*, 6184* -6136* (GH, NA, ae nisi (GH, NA), 6223* (GH), 6297* (GH, NA, NY), 6354* ; 6355 7 6356*, 6357*, Cornwall Bay, 6458*, 6459* (GH, NA, Hd 6459a* (GH), 6460*-- 6468*,6569* (GH, NA, NY), 6513* (GH, a Hasbala Lake region: “Quillwort” L., Argus 849-62, 1031-62, 4-62 (DAO, SASK, GWA), 854-62 (DAO, GH, GWA, SASK). SALIX GLAUCA IN GREENLAND In the early stages of this study Salix glauca in Green- land was not to be included because of its known complexity and the need for special field study. However, as an under- standing of the North American S. glauca developed it be- came apparent that a general treatment of the species in this area would be of value. This change in plans has made it Impossible to examine all of the material available from this area, particularly that in European herbaria, nor has SALIX GLAUCA COMPLEX 101 it been possible to typify all of the infraspecific taxa applied to this species in Greenland. This work, although important, must be left to the future. Salix glauca in Greenland has been treated taxonomically in three ways: 1) the species is sometimes recognized as consisting of several infraspecific taxa based on its highly variable leaf-pubescence, leaf-shape, and habit (Lange, 1880, et al.) ; 2) the species may be regarded as rare and the bulk of the material identified as double or triple hybrids, in- volving S. glauca, S. arctica and/or S. arctophila (Floderus, 1923) ; and 3) the species may be recognized as widespread and highly variable but treated in a broad sense as S. glauca Bl. The recognition of hybrids was common during the 1930’s under the strong influence of Floderus. Seidenfaden ( 1933) deserves special mention because of his attempt to combine anatomical and morphological characteristics to understand S. glauca, S. arctica, S. arctophila and their putative hybrids. Although I have not yet examined all of the anatomical characteristics he used, those that I have studied are either too similar in the various taxa to be of diagnostic value (e.g. the number of cells in a tangential view of medullary rays) ; or apparently inaccurate (e.g. the size of epidermal cells). He reports that S. arctica and S. arctophila have large epidermal cells (400-900 per square mm.) contrasted with the small cells of S. glauca (2000-4000 per square mm.). My work shows that S. glauca and S. arctica are very similar in cell size (2000-3500 cells per square mm.) ; compare Figs. 3-8 & 3-9 with Fig. 4. The largest cells noted in S. arctica were only 1500 cells per square mm. (in Heusser 4T from Southeastern Alaska (GWA) ). However, through the use of combined anatomical and morphological characteristics Seidenfaden classified all his material into 5 levels of hybrids from almost pure Salix arctophila, through three degrees of hybrids to almost pure S. glauca (S. arctica was not present in the area he studied, south of 64°N.). The improbability of this treatment was even recognized by its author who noted that, “A division into the above-mentioned five types will, of course, always be greatly subjective, and a later more thorough investiga- tion will probably arrive at a different result.” 102 GEORGE W. ARGUS In 1938, T. W. Bécher expressed some concern about the treatment of Salix as hybrids and suggested that the hybrids be verified experimentally. However, he considered S. glauca to be rare and ordinarily was involved in hybridization. He did note that pure S. glauca occurred in scrub or form scrub and cited as authority for this Hartz, Kruuse, and Rosenvinge, all pre-Floderan botanists. In 1952, his view changed somewhat and he regarded the hybrid S. arctophila Xx glauca as “ill-founded”. He recognized S. glauca as “a variable and critical group”, and referred the glabrous form to S. glauca ssp. callicarpaea and the hairy form to S. glauca (see discussion of synonymy of S. glauca). This trend back to recognizing Salix glauca in a broad sense and down-grading hybridization was followed by Polunin (1943) who noted S. glauca as being common and “extremely and tiresomely variable”. He not only disagreed with the general recognition of Salix hybrids but stated that S. arctophila and S. glauca seem not to hybridize even when growing in the same marsh. Polunin also pointed out the resemblance of the Greenland S. glauca to the species in eastern Canada. This general interpretation was expanded by Lagerkranz (1950), who proposed to call the Greenland species S. callicarpaea. Later Bocher (1952) proposed the combination S. glauca ssp. callicarpaea. I regard Salix glauca as being widespread in western Greenland, less common but present in east Greenland, and involving S. probably oce and identifi In “pure” Salix glauca, leaf-indument varies from densely pubescent o n both sides to almost glabrate. I have not been SALIX GLAUCA COMPLEX 103 able to associate this with habitat, although Bécher (1952) relates the densely pubescent-leaved forms to sunny expo- sures and the glabrescent-leaved forms to moister places on northern slopes or near running water. The size and shape of leaves also vary widely, possibly due to different en- vironments. The branchlets are more or less pubescent, becoming glabrate in age. Many of them are pruinose, a characteristic reminiscent of S. arctica. The styles are entire or divided almost to the base; however, the majority show little or no division. The bracts of the flowers are often darker brown than in S. glauca and may bear the straight pubescence of S. arctica, further suggesting intro- gression with that species. The habitats occupied by Salix glauca in Greenland are very diverse, similar to the species in eastern Canada. An account of its ecology is given by M. P. Porsild (1920) who notes that it forms copses in places providing sun and moisture during the summer. “Thus often on the sunny side of rocky walls, these lying sheltered from the wind. Besides it is always present in the heath-vegetation, both in the dry part and in the moist. It is to be found in bogs right down to the lake shore and is sometimes growing in the most dry and barren places, dunes, gravel among boulders and stones, on rock-ledges and in crevices.” That its occur- rence in diverse habitats may account for its wide pheno- typical variation has been suggested by M. P. Porsild (1920), Seidenfaden (1933), et al. SELECTED SPECIMENS, West Greenland: Tasermiut Fjord, A. & M. Porsild, 27-28 July 1925 (GH, NY, US); Godthaab, Lagerkranz, 30 August 1936 (A, RM); Ivigtut, Dutilly 9483 (NA); Neria, Hugenn’s, 13 July 1925 (GH, NY, US); Ameralik, Lagerkranz, 26 August 1936 (A, RM); Séndre Stromfjord, Erlanson 2376 (GH, NA); Ikinea, M. Porsild 3622 (GH); Godhaven, Bartlett 63 (GH, US); Diske, Warming & Holm, 20 July 1884 (DAO); Ritenbank Flor, Lagerkranz, 18 August 1984 (RM); Nugsuak Pen., Patut. Hrlanson. 3329 (NA); Kingigtok, 3287 (CAN); Umanaq, Enander, Cape York, Wetherill 209, 213, 214 ( GH). East Greenland: Angmags- salik Fiord, Kungmuit, Bartlett 448 (US); Angmagssalik, Lager- kranz, 8 August 1946 (MIN.). Salix brachycarpa Nuttall, N. Amer. Sylva 1:69. 1842 Salix brachycarpa is restricted to North America and is treated here as consisting of three infraspecific taxa, SSp. 104 GEORGE W. ARGUS brachycarpa, ssp. niphoclada, and ssp. niphoclada var. ful- lertonensis. Each of these infraspecific taxa has been regarded, in the past, as a separate species. However, some taxonomists have favored the reduction of one or another of them to infraspecific rank. Rydberg (1899) regarded S. niphoclada (S. brachycarpa ssp. niphoclada) as most closely related to S. stricta (S. brachycarpa ssp. brachy- carpa). This view was supported by Ball (1934) when he described S. brachycarpa var. mexiae, which was recognized later to be the same as Rydberg’s S. niphoclada (Raup, 1959). This judgment of a close relationship between S. brachycarpa and S. niphoclada has not been concurred in by all taxonomists (see Hultén, 1943, and Raup, 1959), but no one has presented a clear case against it. Some of the confusion regarding the position of Salix brachycarpa ssp. niphoclada in the S. glauca complex has stemmed from the widespread hybridization between ssp. niphoclada and S. glauca in parts of northern Alaska. This hybridization is thought to have resulted in a modification of ssp. niphoclada, in that area, in the direction of S. glauca. Some authors were led to regard it as a subspecies of S. glauca (Wiggins and Thomas, 1962). This view is unsub- stantiated when the two taxa are studied over their entire North American range (see Beringia phase, S. glauca). Salix brachycarpa ssp. niphoclada var. fullertonensis has been generally incompletely understood because of insuffi- cient collections. However, it was aligned with S. brachy- carpa by Ball (1948) and with S. niphoclada by Raup (1959). Raup chose to reduce it to a variety of S. nipho- clada on the grounds that its prostrate habit was all that distinguished the two. Although there are other character- istics to distinguish them I have maintained fullertonensis as a variety of niphoclada and placed them both within S. brachycarpa. A comparison of the infraspecific taxa of Salix brachy- carpa (Table VI) shows that all are similar in numerous characteristics of taxonomic importance. They can be related by pairs in even more characteristics. As a group they may be distinguished from S. glauca by their short pedicels and petioles, and the distinctive oblong leaves with obtuse apices which are borne near the proximal end of vegetative shoots and on the reproductive branchlets. The SALIX GLAUCA COMPLEX 105 TABLE VI. COMPARISON OF THE INFRASPECIFIC TAXA OF SALIX BRACHY- CARPA, SSP. BRACHYCARPA (B), SSP. NIPHOCLADA (N), VAR. FULLER- TONENSIS (F). CHARACTERISTICS TAXA’ B N A. Common to all taxa. Petiole short, 1-3 mm. long + + + Ovaries small and short-beaked . Oe ee J Pedicel 0-0.25-0.5 mm. long + + + Proximal leaves on shoot oblong, apex rounded, obtuse ++ + Bracts sometimes dark-brown to blackish .....+:-++++++++ + + Anthers small, 0.4-0.5 mm. long .... + + t+ Stomata short, 19.6-20.6 micra long ine f/f Petiole reddish ...........:c.cscssssssssescssssseeresesssonsanensnesnenenesoone Sie Se. Co: n to ssp. brachycarpa and ssp. niphoclada. Branchlets densely white-Villous .......-s-sssssersersereneneeres fftf— Shrub 1-2 feet tall ............ssscssssseeeresssosensnsennsesseesesannanens f{+e— C. Common to ssp. niphoclada and var. jullertonensis. Aments long and narrowly cylindrical ......--s-++-s+ss r+ + Shrubs prostrate or decumbent «....-..ssssessererenererererees r f£f + Leaves generally appressed-pubescent -..-+---+-+sererseree oe Leaf apex attenuate-acute .......ssesereresersersssrseneeneenenens ey LZ Branchlets thin and flexible ...........:-ssssssesseseenerseterenenees roo Style short, 0.2-0.5 mm. lOMg «..-+-+-+-seseressesereserenessenerseees r+. + D. Peculiar to ssp. brachycarpa. Aments small and subspherical «.....+-+-ssssssesersnerenesress f—-— Leaves coarsely PpUbeSCeNt ......ssseseseeressersrensenrenseenenrnees +-— Bracts greenish in amthesis ....-.-.:-sssseresrensnerssseneneres® 1 ROR Style long, 0.5-0.8 mm. omg .....scsecsecsseeeresseneserserensenensenees £2 Bact Internodes short, leaves appear “fan-like” .....-------+ f — Aments crowded on the shoot below the leaves ...------ weagase E. Peculiar to ssp. niphoc Bracts narrow, ODlONg «....-sss-ssessrsresesnssssseensnenennes eee r _ Branchlets yellowish .........::-:sssssssssssssseeenssenennanenensnnsnsnens® eS SS oa F. Peculiar to ssp. niphoclada var. fullertonensis. Bracts Teddish ...ccccccsoscssosesssscsscoosenascosnscssonsnansosoonomsceonors r —f Branchlets fine appressed-pubesce ae en Tee come 4 symbols: +- = 1The peau of bets miter rea is indicated by the following always present, f = frequent, r = , and — = absent ahah clit mele igh (less ‘ei 0.5 mm. long) and the short stomata (19.6-20.6 micra long) are possibly related to the diploid condition of S. brachycarpa 1D contrast to the polyploid S. glauca. i 46-50 Salix brachycarpa Nuttall ssp. prachycarpa. Figs S. brachyearpa Nuttall, N. Amer. Sylva 1:69. 1842. o Nuttall, Beer ! Springs encampment, Bear R ., Rocky Mountains, Idaho, 1834. (GH!, type collection). S. desertorum B stricta Anderson, i DC. Prodr. gre’ og a Bourgeau, Dans les Marais, Saskatchewan, 1 16 June 1 ecto- 106 GEORGE W. ARGUS xs Salix brachycarpa \ \ e ssp. brachycarpa o ssp.niphoclada — s yar. ful/lertonensis Map 5. Range map ot Salix brachycarpa ssp. brachycarpa, ssp. niphoclada, and ssp. niphoclada var. fullertonensis. type, A!, fragment and photograph; GH!, syntype 7). S. stricta (Anderss.) Rydberg, Bull. N. Y. Bot. Gard. 1:273. 1899. S. desertorum rt fruticulosa Andersson, in DC Prodr. 16(2) :281. 1868. ex char. S. brachycarpa var. glabellicarpa Schneider, Bot. Gaz. 66:338. 1918. Macoun 95374, Athabasca R., west side of discharge of Beauvert Lake, Jasper Park, Alberta, 30 July 1917. (CAN, holotype; A!, GH!, NY!, isotypes). S. brachyearpa var. sansoni Ball, Univ. Calif. Pub. Bot. 17:414-416. 1934. Sanson 119, Starbird Glacier valley, Rocky Mountains Park, Alberta, July 1928. (RMNP!, isotype). - brachycarpa var. psammophila Raup, Jour. Arnold Arb. 17:280. 1936. Rawp 6888, Vicinity of William Pt. Lake Athabasca, Saskatche- ? in Idaho by Thomas Nuttall in 1834 while he was a member erence to Lake Timpangos (now original discussion. The collection SALIX GLAUCA COMPLEX 107 Li Khe x ‘e i ee ell H WAS. \ Liat \ ; la ———— aia [ ol -& A ; uy ee a ly : if 4 Reg P WF Bo io we f 4 ull uw Pri ie. “ aN 5 Ng a} ¢ i > ——¥ i { . ‘ ) ME AW yy G +2) Te = i? Tiyge a SS rn BRACHYCARPA / cee a FULLERTORENSIS Map 6. Generalized distribution map of the infraspecific taxa of niger Os ae The zones of apparent overlap between the taxa a and some probable disjun indicated. was actually made along the Bear River in the Soda Springs area of southwestern Caribou Co., Idaho. A number of specimens very similar to the type have been collected there. I have not seen the holotype which may be represented by a specimen in the British Museum (P. Raven, personal com- munication) ; however, a specimen from the type collection is in the Gray Herbarium. The scone wae desertorum f stricta and + fruticulosa represent S. brachycarpa. I have seen a fragment and . photograph of the lectotype of stricta which is at Kew, an a specimen in the Gray Herbarium which is probably a syntype. The latter specimen is incorrectly dated 1859; it should read 1858. Both of these specimens are clearly S. Brachicarpa as‘ was noted by Gehmelder in-1018 (19185). The identity of the name fruticulosa is less certain because I have not seen any of the type material nor any of the 108 GEORGE W. ARGUS original material studied by Andersson (1868) from Ft. Hall (probably in southeastern Idaho). However, on the basis of its description it can safely be regarded as S. brachycarpa. Salix brachycarpa var. glabellicarpa is an unusual form of the species which may be of hybrid origin. It will be discussed below. Salix brachycarpa var. sansoni was described by Ball on the basis of material from the Rocky Mountains of Alberta. I am unable to understand how Ball proposed to distinguish this varietal name from the species. All of the characteris- tics used to distinguish this name, including the smaller size of the plant, the shorter and relatively broader leaves, and the subsessile, nearly globular aments, are typically those of the species and could not be used to distinguish a variety. The type material (Sanson 119) which is supposed to be in the C. R. Ball herbarium (now included in the United States National Arboretum Herbarium) and at the Uni- versity of California has not been found in either collection. Apparently the only extant specimen of the type collection is an isotype in the Rocky Mountain National Park Her- barium, Banff, Alberta. The name Salix brachycarpa var. psammophila is re- garded as a pubescent, broad-leaved form of the species which does not require formal recognition. It will be dis- cussed below. scription. Shrubs erect, commonly 1-3 feet tall, some up to 6-9 feet. Branches reddish-brown, often covered with persistent grayish tomentum, in some cases glabrescent and with a flakey epidermis. Branchlets densely white- or gray-tomentose, some with coarse matted trichomes or only thinly pubescent. Internodes often very short giving the numerous leaves on the shoot a fan-like appearance. Buds reddish- brown, pubescence similar to the branchlets, sometimes pruinose. Leaves obovate to oval to broadly or narrowly elliptical or rarely oblanceolate, length/width (1.5-) 2.8-3 (-4); length (1.2-) 2.3-3 (-4) cm. long. Apex acute or obtuse depending on the leaf-shape; base ai taas Pehoneses eS Margin entire, sometimes with a few glands near ae firtaan: eset econ on mature leaves. Blade pubescent on : y with densely matted grayish-white trichomes, Sometimes only thinly pubescent; glaucous beneath. Petiole generally shorter than bud (.5-) 1-3 (-4) mm. long; often reddish, sometimes hip age the color extending into the midvein. The proximal three ienéa tds @iedadae pa from the distal ones, oblong or mie trichomes below: iulsehigcine ia ae ooo sea a ae t ; glandular-margined. Stipules presen, SALIX GLAUCA COMPLEX 109 Fics. 46-50. Representative specimens Sa nee 46 & 47. Staminate specimen, Argus ef 4. “U0 4, 24 J 1962. w view. FIG. 47. Closeup of sts note the aie ; “ R and the very small anthers Pistillate s Argus -62B as above Ic. 48. General view ote I Jean se ‘ short, leafy reproductive branchlet. Clos g , . note the subglobose shape. Fic. 50. Habit photogra i growls from Mt. Albert, Quebec, Argus 76- 2 July 19 s spec egiie tall, the hand lens is 5 em. long. Th 2 ter sc n th A ate is the scale for FIGs. 46 & 48 110 GEORGE W. ARGUS usually short (less than 0.5-1.5 mm. long) and broad, often obscured by dense branchlet-pubescence. Staminate aments coétaneous, short-cylindrical to often spherical, length X width 6X5, 1011 or 155 mm., borne on very short leafy reproductive shoots rarely over 10 mm. long. Leaves on the reproductive shoot are of the modified type, strap-shaped with rounded apex and base. Aments crowded on the branch, often 14-18 aments on a branch 7-8 cm. long. Bracts light-brown, rarely blackish, often greenish in preanthesis and anthesis; pubescent on both surfaces with straight or curly trichomes; oval to broadly so. Glands borne adaxi- ally and abaxially, usually narrow. Filaments two, free or sometimes united at base, glabrous or sometimes pubescent near base. Anthers small, less than 0.5 mm. long, ellipsoid to globose, yellowish when dry, reddish in preanthesis. Pistillate aments coétaneous, short-cylindrical to nearly spherical, length X width 15x10 to 18X10 or 209 mm. in postanthesis, borne at the end of short leafy reproductive shocts (peduncle of authors) varying from 3-20 mm. long. Aments commonly very numerous on branches, often 12-15 aments on a branch 13-15 cm. long. Bracts as in the staminate flowers. Ovaries densely white- absent or, if present, very short 0.25-0.5 mm. long, pubescent. Glands adaxial, broad or narrow flattened structures, often half as long as ovary. Style mostly entire, about 0.5-0.8 (-1.5) mm. long, sometimes partly divided into two lobes, yellow-green. Stigmas two, each divided into two lobes. Distribution’. (Maps 5 & 6). The distribution of Salix brachycarpa ssp. brachycarpa is extensive and includes several disjunctions which suggest an even wider distribu- tion in the recent past. The taxon is restricted to North America. Its center of abundance is in the Canadian Rocky Mountains of southern British Columbia and Alberta. From there its abundance diminishes in all directions. It extends southward into Montana, Idaho and northwestern Wyoming, and then into southeastern Wyoming and Colorado. The latter regions are now disjunct but were probably con- tiguous in the late Pleistocene. West of the center of abundance there are three small disjunct areas in Wash- ington and Oregon. To the north the subspecies ranges into northern British Columbia where it overlaps with ssp. AP ES ie disjunct population of S. brachycarpa ssp. brach his pool Nevada Mts., California, came to my attention too late to be included in ai + A See Major, J. & S. A. Bamberg. 1963. Some cordilleran plant species neW e Sierra Nevada of California, Madrono 17: 93-109, yearpa, on the eastern slope of SALIX GLAUCA COMPLEX 111 niphoclada in the region between Muncho Lake and Lake Atlin. The northernmost outlier of this subspecies is at Norman and Norman Wells on the Mackenzie River, N.W.T., where it again overlaps with ssp. niphoclada. To the east of the center of abundance it occurs sparsely in the forest-prairie transition of Alberta and Saskatchewan ex- tending northward to the southern shores of Lake Athabasca and to Great Slave Lake. It then occurs in the Hudson Bay lowland region of Manitoba, Ontario, and on the shores of James Bay in Quebec. The gap between Saskatchewan and the Hudson Bay coast is thought to be a real one; however, it may prove to be a collecting gap. In Quebec there are three small disjunct areas, one in northern Ungava at Wake- ham Bay, a second on Mt. Albert where the subspecies is isolated on a serpentine plateau, and a third on Anticosti Island in the Gulf of St. Lawrence. The general pattern of distribution seems to be a large central area in the northern Rocky Mountains with smaller, disjunct areas radiating in all directions. This center cannot be interpreted as a center of distribution because the scat- tered disjunctions indicate a wide band of “survival” all along the southern fringe of the ice sheet and, in the north, the probable effects of post-Hypsithermal climatic change. The center in the Canadian Rocky Mountains may be re- garded as a “center of contraction” rather than a center of distribution. Discussion. The wide range of Salix brachycarpa ssp. brachycarpa and its numerous disjunctions are correlated with some morphological diversity, but not enough to under- mine the uniqueness of this taxon, one of the most distinctive in the S. glawca complex. . In British Columbia and Alberta ssp. brachycarpa 1s relatively constant morphologically but there is some ten- dency for the species to vary in the direction of S. glauca. It must be emphasized that this is only a tendency and that most material is of the “typical” form. This tendency can be noted in certain characteristics including an increase in petiole-length up to 2-3 times the length of the bud, the occurrence of plane leaves correlated with fine appressed pubescence, the occurrence of lanceolate to oblanceolate leaves, and longer peaked capsules. These 112 GEORGE W. ARGUS characteristics occur in much of the material from northern: British Columbia and are much less evident in the south. Whether these characteristics are indicative of hybridiza- tion and introgression with S. glauca is unknown. An arti- ficial cross between S. glauca (Argus 82S, pistillate, Churchill, Man.) and S. brachycarpa (Argus 106-59, stami- nate, Mt. Albert, Que.) made on 11 April 1960 was success- ful to the extent that seeds were produced, germinated on 6 June 1960, and the seedlings grew vigorously until killed by a fungus infection in August 1960. It is certainly possible that hybridization may be taking place at present or may have been common at some past time, resulting in the intro- gression of S. glauca characteristics into S. brachycarpa. A modification from the Waterton Lakes area of the Canadian Rocky Mountains deserves comment. This is the glabrous form named var. glabellicarpa by Schneider. This variant resembles S. chlorolepis, the endemic occurring on Mt. Albert, Que., and the numerous hybrids in that area between S. brachycarpa ssp. brachycarpa and S. chlorolepis. Its most distinctive characteristics are the glabrous-pruinose ovaries and capsules, the generally sparse shoot-pubescence, pruinose stems and buds, longer petioles, and sparsely pubescent leaves. In all of these characteristics it resembles the hybrid S. brachycarpa X chlorolepis. It resembles S. brachycarpa in leaf-shape, pedicel-length, and ament-size and shape. If it were not for the wide disjunction between Mt. Albert and the Canadian Rocky Mountains the speci- mens of this form would be assigned to the above hybrid. Hybridization may still be the explanation for this form, with S. pedicellaris being one of the parents. This is sug- gested by the long petioles and pruinose caspules. One specimen from the Sunwapta River, Alberta, has patchy pubescence on the ovaries, a characteristic often found in hybrids between species with glabrous and pubescent ovaries. This form requires field study and experimental tests in order to arrive at a more definite conclusion. In- vestigations could profitably be carried out along the Banff- Jasper Highway in Alberta where this form and the “tyPi- cal” S. brachycarpa occur together. South of the “center of abundance” there are two large, but isolated, areas of ssp. brachycarpa. The northern area, SALIX GLAUCA. COMPLEX 1138 southwestern Montana, eastern Idaho, and northwestern Wyoming,; includes the type locality. at Soda Springs, Idaho. The subspecies in this region is relatively uniform and indi- viduals with tendencies toward S. glauca are virtually unknown. Most of the collections from Idaho are from the salt flats along the Bear River and are more or less identical with the type in having small, elliptic leaves, In this area, leaves on vigorous shoots are broad and haye a rounded to subcordate base and rounded apex. This leaf form is found on robust. shoots throughout the range of the subspecies. The southern area, southeastern Wyoming, Colorado, and Utah, is characterized by an increasing difficulty in distin- guishing ssp. brachycarpa from the closely related S. glauca and S. arctica. In the Medicine Bow Mts. of Wyoming individuals with-longer pedicels, blackish bracts, and prui- hose stems and buds become increasingly common. I have assumed, in dealing with herbarium specimens, that this form is due to hybridization and environmental modifica- tion; however, experimental evidence is much needed. _The influence of the environment in the southern Rocky Mts. may be greater than in the north due to the increasing importance of altitude and the sharp environmental changes associated with it. A collection that I have had under culti- vation (C.L. Porter, Aug. 1958) from the Medicine Bow _Mts., Wyoming, has shown such striking leaf and habit modifications that I am reluctant to pass judgment on her- barium material without more personal field experience in this area. The only field studies of environmental modifica- tion in S. brachycarpa that I know of have been made by Bliss (1956) in the Medicine Bow Mts. Comparisons were made between individuals on north-facing and south-facing slopes. The information obtained in this study is not imme- diately usable because the identity of the individuals measured is in doubt (voucher specimens Bliss 419 is identi- fied as S. brachycarpa X glauca) and the sample size was too small for an adequate comparison of variation. — Southward in Colorado, the problem of recognizing ssp. brachycarpa is acute and the distinction between ssp. brachycarpa and S. glauca is virtually impossible. Speci- mens with long petioles, yellowish stems and large aments are common, suggesting widespread introgression with S. 114 GEORGE W. ARGUS glauca. The present treatment of the S. glauca complex in Colorado leaves much to be desired, and it is obvious that field study is required. Although numerous herbarium specimens are available, they are inadequate for a clear understanding of the problems involved. Other taxonomists, including Ball (1899), Smith (1942), and Schneider (1918b), have come to a similar conclusion. Along the northern border of British Columbia, from about Muncho Lake to Lake Atlin, ssp. brachycarpa is in contact and overlaps ssp. niphoclada. Unfortunately, a sufficiently large number of collections of these two taxa has not been made and it is difficult to clearly describe the intergradation that occurs there. Some of the specimens of ssp. brachycarpa (McCabe 8718, Dease L.; Setchell & Park 11, Atlin) suggest ssp. niphoclada, the former in pedicel- length and the latter in ament-length; and a specimen of ssp. niphoclada (Taylor, et al. 250) shows variation in its vegetative structures in the direction of ssp. brachycarpt. The variation found in these few specimens may be found in either subspecies and is not in itself unequivocal evidence of intergradation. The northernmost extension of ssp. brachycarpa is at Norman and Norman Wells, N.W.T., where it appears to be disjunct from the main population. Two specimens from this area (Crickmay 162 and A. Porsild 16706) strongly resemble ssp. brachycarpa and two others (Cody & Gutter- tdge 7442 and 7656) are in many respects intermediate be- tween ssp. brachycarpa and ssp. niphoclada. The Crickmay specimen is a peculiar one with the dense pubescence of the “psammophila” type identical to that of Lainge 222 from Lake Athabasca. From the region between Norman and Lake Athabasca I have Seen several collections of ssp- brachycarpa from Great Slave Lake (Thieret & Reich 6838, 7658, 7659, 8268) and several specimens which seem to represent S. brachycarpa X< glauca (Raup & Abbe 4508 and 4597; Loan 312; and Cody & Matte 8613). A possible ex- Planation of the outliers at Norman and Great Slave Lake is that they represent a formerly greater range of the Species in the Mackenzie River valley, which has contracted as the region became more stabilized after the retreat of the ice sheet and the drainage of lakes. It is also possible that the gaps are due to inadequate collecting. SALIX GLAUCA COMPLEX 115 There occurs on the south shore of Lake Athabasca a distinctive form of S. brachycarpa named var. psammophila by Raup (1936). This form is restricted to the sand-dunes. Raup stressed the broadly ovate leaves and the strict up- right habit. However, neither of these characteristics is unknown or rare in S. brachycarpa. If leaf-length/width is compared in four populations of S. brachycarpa ssp. brachycarpa (Fig. 51) it is evident that considerable varia- tion and overlap can be observed in this characteristic. The 30; Percentage 3.0 ener Ye Fig. 51. brachycarpa ssp. brachycarpa. The triangles represent a popu on the IV, 11 July 1959). The open circles represent a sand-dune population from Churchill, Manitoba (Ar M III, 16 Aug. 1958). The closed circles represent a salt-marsh population from Churchill, Manitoba (Argus M Il, 11 Aug. 1958). All local population samples are in the author’s collection. broader leaves in the Lake Athabasca population are part of a continuous variation of this characteristic in the species. The strict upright habit in the sand-dune material seems to be produced by a marked shoot elongation which may be related to the sand dune environment. Low growing, almost prostrate, specimens of S. brachycarpa from Mt. Albert, Que. (Fig. 50) assumed an erect habit when grown in the 116 GEORGE W. ARGUS greenhouse and nursery in Massachusetts (Argus 104-59 and 106-59). The reason for the denser pubescence of both leaves and stems in the sand dune material is not completely understood, but it may also be related to environmental influences. It is interesting to note that S. silicicola Raup, the densely pubescent relative of S. alaxensis Cov., also occurs in the sand dunes on the south shore of Lake Atha- basca. ; cou In the Hudson Bay lowlands region ssp. brachycarpa is abundant. At Churchill, Manitoba, the subspecies is dis- tinctive, although it is similar to S. glauca in leaf-size and shape. A series of characteristics including the short pedi- cels and petioles, the coarse leaf-pubescence, short inter- nodes, subglobose to short cylindrical aments and the numerous aments per shoot in ssp. brachycarpa are sufficient for positive identification. In this area ssp. brachycarpa is a late flowering species which serves both as an identifying characteristic and as a possible means of isolation. On Mt. Albert, Quebec, ssp. brachycarpa is very similar to the type-material of S. brachycarpa from Idaho. This may be related to the soils which are ultrabasic in both areas, or it may indicate a similar genetic constitution due to past connections. These strong morphological affinities between the Rocky Mountain and eastern elements of S. brachycarpa are reminiscent of that which exists between the Rocky Mountain and the Eastern Phases of S. glauca. These east-west ties in both species support the hypothesis that the southern portions of both S. glauca and S. brachy- carpa Were relatively continuous and made up a gene-pool continuum during a time when the northern portions, the Beringia and Western Phases of S. glauca and ssp. nipho- clada and var. fullertonensis of S. brachycarpa, were rela- tively isolated. It is possible that the disjunct nature of the southern populations is of relatively recent origin. Ecology. Salix brachycarpa ssp. brachycarpa is one of the few Salix, including S. exigua Nutt., S. candida Fluegge: and S. chlorolepis Fern., which can occur in alkaline and/or Saline habitats. It occurs on the serpentine barrens of Mt. Albert, the salt-marshes of the Hudson Bay lowlands, the alkaline meadows of Saskatchewan and Colorado, and hydro- magnesite swamps and hot springs in British Columbia. In SALIX GLAUCA COMPLEX 117 Idaho it occurs on salt-flats and on the cones of mineral- springs. However, it is not restricted to these habitats and over its range it may be found to occur in a wide variety of habitats from open woodlands, to bogs, muskegs, meadows, stream- and river-banks, alpine slopes, unstable scree slopes, and moraines. It seems to be particularly adapted to an existence in habitats which experience repeated severe dis- turbances, such as active sand dunes, seasonally flooded banks of streams and the low-lying estuary-zones of rivers. SELECTED SPECIMENS. Northwest Territories: District of Mackenzie: Mackenzie R., between Blackwater R. and Norman, Crickmay 162 (CAN); Norman, A. Porsild 16706 (CAN); Great Slave Lake, Thieret & Reich 6838, 7619, 7657, 7658, 7659, 8268 (F, GWA); Ft. Reliance, Larsen, July 1962 (GWA, WIS). District of Keewatin: Seal R., 40 mi. northwest of Churchill, EZ. and A. Preble 41 (26 before correction) (NA, NY, US). British Columbia: Flying U Ranch, Cariboo, shore of Green L., Eastham 11523, 11525 (NA, UBC); Kin- basket, Big Bend Highway, Eastham 16008 (DAO, NA, UBC); Sum- mit Lake, Fyles (DAO); Anahim L., Hatter 1 (UBC); Yoho Nat. Park, Yoho R. near Takakhaw Falls, Hitchcock & Martin 7681 (A, DS); Mt. McLean, near Lillooet, Macoun 97770 (GH); 2-3 mi. S. of Kinbasket L., McCabe 6278 (NA); head of Dease L., McCabe 8718 (NA); Kicking Horse R., near Field, McCalla 6998 (NA); Emerald L., Pease 22361 (GH); head of Findlay R., near Thutade L., E. Preble & Mixter 671 (US); Muncho L., H. Raup 10855 (A, CAN); Alaska Hwy., Mi. 347, near Liard R., H. Raup 11011 (A); Peace, 4 mi. above Carbon R., H. Raup & Abbe 4308 (A, MIN, NY); Atlin, Setchell & Parks 11 (NA); Liard Hot Springs, Taylor, Szezawinski, & Bell 337 (DAO, UBC); Elk R., 34 mi. N. of Natal, Weber 2323 (GH, NA, Ice L., Peck 18442 (NA). Wallowa Co.: upp 22487 (NY). Idaho: Teton Co.: Driggs, Christ 5382 (NA, bebe Caribou Co.: Bear R., north of Soda Springs, Christ 8977 (NA, ); Soda Springs, Davis 825 (DS, NA): Bannock Co.: Soda Springs, 118 GEORGE W. ARGUS E. & L. Payson 1723 (GH, NY). Custer Co.: Pahsimeroi R., Wilson 2658, 2658a (GWA). Montana: Beaverhead Co.: Pioneer Mts., near Sheep Cr., Hitchcock & Muhlick 12886 (DS, GH, NA, NY). Jefferson Co.: Elk Park, Rose, 25 July 1982 (NA). Glacier Nat. Park: Cracker L., Jack 2158 (A), Piegan Pass, 2210 (A, US); trail to Altyn Mt., McKelvey, 28 Aug. 1921 (A); Appekunny Cr., Standley 16906 (US), trail to Piegan Pass, 17457 (US). White R. trail, Kirkwood 2284 (GH). Wyoming: Yellowstone Nat. Park, Lower Geyser Basin, Rydberg & Bessey 3913 (GH, NY). Teton Co.: Gros Ventre R., Tweedy 298 (NY). Sublette Co.: Vicinity of Green R. Lakes, E. & L. Payson 4528 (GH, NA, RM); Union Pass, A. Nelson 913 (GH, RM), Union Pk., 1072 (RM). Albany Co.: Telephone Mines. A. Nelson 7869 (NA), Centennial, 8682 (RM). Utah: Iron Co.: Cedar Breaks, Maguire 20169 (NA). Summit Co.: Wasatch Forest, Haas 86 (NA). Colorado: Boulder Co.: Niwot Ridge, 10 mi. NW of Nederland, Weber 5532 (DAO). Conejos Co.: along Coneuos R., Cumbres Pass and Platoro Rds., Weber 7850 (MIN). El Paso Co.: Pikes Pk., Johnston 2788 (NA); Manitou Reservoir, Christ 619 (NA). Gunnison Co.: Virginia Basin, Langenheim 376-48; Mt. Carbon, Tidestrom 3835 (NA). Huerfano Co.: Sulphur Springs, Wooton (A. US). Lake Co.: Leadville, L. & E. Kelso 4864 (DAO). Larimer Co.: Rocky Mountain Nat. Park, Trail Ridge, Smith 208 (NA); Estes Park, Smith 1031A (NA). Park Co.: N. of Antero Jet., Weber, Rollins, & Livingston 6603, 6604 (GH). Summit Co.: Breckenridge, Christ 3003 (NA). Latitude 39°-41° (near South Clear Cr.), Hall & Harbour 184, 523 (GH); South Park, Wolf & Rothrock 819 (GH, NY), 829 (GH, US). Saskatchewan: Meath Park, 32 mi. NE of Prince Albert, Boivin & Breitung 6202 (NA); Beverly, Breitung 5844 (DAO); Saskatoon, Fraser 1A (NA), 31 May 1939 (DAO); Argus & Ledingham 90-62A, 90-62B (GWA, SASK); Grand Trunk Railway: Bare Hills, Macoun & Herriot 70292 (NA, NY); Spy Hill, 70294 (GH, NA, NY); Lake Athabasca: McFarlane R., Laing 222 (NA, US); Argus 666-62, SALIX GLAUCA COMPLEX 119 106-59 (DAO, GH, GWA); Clausen & Trapido 2875 (MIN); Collins & Fernald 65* (A, GH, MIN, NY, US); Fernald & Collins 518%, 513a* (GH); Fernald, Griscom, Mackenzie, & Smith 25664* (GH, NA, NY, US); Victorin, Germain, Brunel, & Rousseau 17601 (A, GH, NY, US). Salix brachycarpa ssp. niphoclada (Rydb.) Argus comb. nov. Figs. 2-55 S. lingulata Andersson, in DC. Prodr. 16 (2) :281-282. 1868. Kastal- sky (as Kostalsky), Alaxa, 1827. (NY'!, type collection). S. niphoclada Rydberg, Bull. N.Y. Bot. Gard. 1:272-273. 1899. Tay- lor 60, 30 mi. north of Arctic Circle, Mackenzie R., N.W.T., 18 July 1892. (CAN!, holotype; US!, isotype). S. glauca ssp. niphoclada (Rydb.) Wiggins, im Wiggins and Thomas, Flora Alaskan Arctic Slope, AINA Sp. Publ. 4:144-145. 1962. S. brachycarpa var. mexiae Ball, Univ. Calif. Publ. Bot. 17:412-414. 1934. Mexia 2131, Muldrow Glacier, Mt. McKinley Nat. Pk., Alaska, 23 July 1928. (A!, GH!, MIN!, NY!, US!, lectotype collection). (in part). S. glauca X farrae walpolei Cov. & Ball, of Hultén, Fl. Alaska and Yukon 3:528-529. 1943. (in part). : S. muriei Hultén, Fl. Alaska and Yukon 3:531. 1943. O. Murie, Salcha Slough, Alaska, 18 June 1922. (A!, holotype [pistillate shoot] ; US!, isotype). S. niphoclada var. muriei (Hultén) Raup, Contr. Gray Herb. 185:60. 1959. Discussion of Synonymy. Salix lingulata Anderss. I have studied the specimen in the New York Botanical Garden (ex Herb. Fischer) which Schneider regarded as a cotype of this name. It was collected by Kastalsky in Alaska (as Kostalsky in Alaxa). There are only a few fragments in this collection including two almost complete aments, one ‘pistillate and the other staminate. On the basis of this material I would regard the collection as representing S. brachycarpa ssp. niphoclada. The filaments are pubescent and more or less united at the base, as in S. glauca, but the anthers are small, ca. 0.4 mm. long, as typical of ssp. nipho- clada. The pistillate ament is in anthesis but its small ovaries and short pedicels strongly suggest ssp. niphoclada. Hultén (1943) discusses this name and notes a specimen he has seen at Riksmuseet, Stockholm, collected by Kastal- sky at Unalaska. From his account, this appears to be a possible portion of the type-collection and the equivalent of Ssp. niphoclada, as he suggests (as S. niphoclada). Since the two specimens, known to me, collected by Kastalsky and called S. lingulata are probably ssp. niphoclada, this name is placed as a synonym of that subspecies. However, 120 GEORGE W. ARGUS additional material which is apparently located at Leningrad should be studied before a final judgment is made. Saliz brachycarpa var. mexiae Ball. I agree with Raup (1959) that this taxon is indistinguishable from S. nipho- clada (ssp. niphoclada). However, when all the material cited by Ball is taken into account, a number of hybrids were noted, supporting Hultén’s (1943) view that the Alas- ka population included under var. mexiae is a hybrid popu- lation. The staminate type (Mexia 2132 ) as well as some of the other specimens (Mexia 2270, and Henderson 15045) are probably the hybrid S. brachycarpa ssp. niphoclada X glauca. Salix muriei Hultén. Again I agree with the judgment of Prof. Raup (1959) who notes that this name simply represents a glabrescent extreme of ssp. niphoclada. The type-material I have seen certainly cannot be distinguished from ssp. niphoclada on any basis except its tendency toward glabrescence. Salix glauca ssp. niphoclada (Rydb.) Wiggins. In this combination the taxon ssp. niphoclada has been incorrectly associated with S. glauca. Wiggins apparently based his judgment on a familiarity with the taxa involved in the Arctic Slope of Alaska region. In this area extensive hybrid- ization seems to be taking place between S. glauca and S. brachycarpa ssp. niphoclada which somewhat breaks down the morphological boundary between these taxa (see Hy- brids). If these taxa are studied over their entire ranges, the differences between them, which may include cytological differences, and the very close relationship between S. brachycarpa and niphoclada become apparent. Description. Low shrubs, erect to prostrate, often spreading, 1-3 (-6) feet tall. Branches thin and flexible, reddish-brown, grayish to rmis exfoliating; sometimes with persistent pubescence, Branchlets densely white-tomentose, often with inter- mixed long villous trichomes; sometimes only thinly appressed-pubes- cent. Buds reddish-brown, with pubescence similar to the branchlets, rarely pruinose, Leaves obovate, elliptic-oblong, oblong, oblanceolate to lanceolate; length/width (2A- ) 2.9-3.7 (-5.5). Blade (2.1-) 2.5-3.2 (-8.5-4.5) cm. ; base cuneate to rounded. Margin entire, n lcuous. Blade pubescent on both surfaces; appressed- spreading trichomes, glaucous; SALIX GLAUCA COMPLEX 121 FIG. 53. Closeup of fruiting pistillate ament. epee 2 ae Tene 1 from western Arctic Alaska $15, Ogoterak OF. eee ie s Z Ale , ial FIG. 54. General vie bit, broad leaves and narn wily mae os : se¢illate @ t. pistillate aments. FIG. 55. Closeup of flowe! istillate amen 122 GEORGE W. ARGUS thinly pubescent to glabrate above, sometimes ciliate on margin and glabrescent toward center of blade. The proximal leaves on the shoot characteristically oblong, glabrous above and with long straight tri- chomes below; apex and base rounded; veins originating near the leaf base running parallel to margin and terminating near apex. Juvenile leaves often sericeous-appressed-pubescent. Petiole relative- ly short, but longer than bud, 1-3-4 mm. long; reddish to yellowish. Stipules present, sometimes prominent, 2 mm. or less up to 4 mm. long; lanceolate; glandular-margined; pubescent. Staminate aments coétaneous, long and narrowly cylindrical, 16-37 mm. long; borne on leafy reproductive shoots (peduncle of authors). In one specimen (Spetzman 2701 ) the flowers are “whorled” on the on both sides with short trichomes, or glabrate abaxially and some- times pruinose; oval to narrowly oblong. Ovaries densely white- tomentose, short, compact, and short-beaked, occasionally long-beaked. Capsules thinly pubescent, light-brown or sometimes greenish, may be deciduous from rachis at maturity. Pedicels (0-) 0,25-0.5 mm. long, » pubescent. Glands adaxial, usually simple narrow Sometimes divided into two lobes; often longer than ovary and up to 2 mm. long; 0.5-1.5-2.0 mm. long. Style 0.2-0.5 mm. long, undivided or divided to the base. Stigmas two, each with two lobes 0.2-0.3 mm. long, rarely longer. Distribution (Maps 5 & 6). Salix brachycarpa ssp. nipho- a northwest American distribution, found most ,on the Arctic Slope of Alaska and in the Alaska region and southeast into the Great Bear Lake area. In its occurrence is scattered and it is poorly tern Alaska. It extends east and south into northwestern British Columbia where it PD. brachycarpa. In the Coppermine area it var. fullertonensis. Salix brachycarpa ssp. niphoclada is a dis- Which can be separated from closely related ort petiole, densely white-tomentose branch- cylindrical often lax aments, short-beaked overlaps with ss Overlaps with Discussi taxa by its sh lets, narrowly SALIX GLAUCA COMPLEX 123 Ovaries borne on short pedicels. It is distinguished with difficulty from var. fullertonensis, but a combination of characteristics including its erect habit, densely tomentose branchlets, larger leaves, and the common occurrence of narrow bracts will usually serve the purpose. For further discussion, see var. fullertonensis and Table VI. The leaves of ssp. niphoclada are of two general shapes, broad or narrow. The former are generally elliptic and the latter oblong, oblanceolate or lanceolate. The apex is usually acute-attenuate in both types, although the broad leaves are sometimes more or less obtuse. These two general shapes occur at random throughout the range of the subspecies. Individuals with broad leaves, such as Raup 12185, Lake Kluane, Yukon; and Johnson, et al. 215 (Figs. 54 & 55), Bering Straits area, Alaska, are not only similar in leaf- shape but also in their general prostrate habit, and narrow, elongate, flexible pistillate aments. Individuals with narrow leaves, the most common type, are distributed from the Yukon westward to western Arctic Alaska and include Porsild & Breitung 9716; A. Porsild 16705; Taylor, et al. 780; Buckley 110 (Figs. 52 & 53); and Anderson 4707. Numerous other examples of these types as well as numer- ous intermediates could be cited. In central Alaska and the Yukon Plateau, typical material of ssp. niphoclada is less common than in the Alaska Range or the Arctic Slope of Alaska. The form found in this region tends to be the glabrescent one named S. muriei by Hultén. In addition, many of the specimens seem to represent hy- brids. For example, a specimen from the Porcupine River (Buckley 111) which is in most respects like ssp. niphoclada (glabrescent form) proved to have large epidermal cells and stomata typical of S. glauca suggesting a possible influ- ence of the latter species. The incidence of hybridization of Salix brachycarpa ssp. niphoclada and S. glauca is apparently very high in arctic Alaska and in the Alaska Range. The intergradation noted in herbarium material from that area is thought to reflect a combination of influences of hybridization, introgression, and environmental modification. From what is known about the plasticity of both S. glauca and S. brachycarpa, it is con- ceivable that under the influence of a rigorous environ- 124 GEORGE W. ARGUS ment S. glauca could approach ssp. niphoclada, and on favor- able sites ssp. niphoclada could resemble S. glauca. Our understanding of the effect of the environment on the morphology of these taxa is wholly inadequate. Field obser- vations correlated with experimental transplants could profitably be made on the Arctic Slope region at the Arctic Research Laboratory at Umiat. Salix brachycarpa ssp. niphoclada is treated here with a relatively narrow range of variation. Specimens of pre- sumed hybrid origin are placed under the name S. brachy- carpa ssp. niphoclada X glauca, Characteristics given par- ticular attention in delimiting the subspecies are petiole- length, pedicel-length, ament-length’ and width, and the density of flowering. The pedicel-length is one of the most reliable, but it is best used in combination with other charac- teristics. In treating this taxon in a fairly narrow sense I have attempted to eliminate many of what appear to be hybrid specimens which have been included in it in the past and which led Hultén (1943) to consider this taxon to be composed largely of hybrids. The area of overlap between ssp. niphoclada and ssp. brachycarpa in northern British Columbia has been dis- cussed under the latter subspecies. Ecology. The ssp. niphoclada occurs in a variety of habi- tats, usually, but not always, in arctic or alpine areas. In the Arctic it occurs on limestone scree, talus slopes, tussock meadows, upland Dryas heaths, willow scrub along river and creek banks and on sandy and gravelly floodplains. In the alpine region of Mt. McKinley Nat. Park, it occurs in woods, on moraines, and most commonly on gravel glacial outwash plains. SELECTED SPECIMENS. Alaska: Aleutian Is., “Alaxa” (Unalaska), Kastalsky (NY). Western Pacific Coast district: Cold Bay, across from Kenai Pen., Beals 24 (NA); Koggiung, Muller 1046 (US); northeast of Naknek Airfield, K. Raup 11 (US). Bering Straits dis- trict: Nome, Jones 9045 (DS, NA); Kotzebue, Anderson 4707 (NA); Scamman 6458 (GH); Grantly Harbor, Walpole 1752b (US); O80 toruk Cr., Johnson, Viereck, & Melchior 148 (ALA), 215, 215A, Kuk- puk R., 673 (ALA, GWA). Central Yukon district: Gerstle R., Arg¥8 529, 580 (GWA, RM), 556 (DAO, GWA, RM); Porcupine R., Buckley 110, 111 (GWA); Funston 185 (NA. NY 1795 (NA, US); Coldfoot, Piper 270 (NA, US); John R., Schrader, SALIX GLAUCA COMPLEX 125 10 July 1901 (US); Anaktuvuk R., Sehrader, 5 Aug. 1901 (US); Ft. Yukon, W. & C. Setchell 397 (NA). Alaska Range district: Mt. McKinley Nat. Pk.: Teklanika R., Argus 608 (GWA, RM); trail to McKinley bar cabin, 658 (GWA, RM); Toklat R., 687 (RM); Savage R., Mexia 2083, 2084 (A, GH, MIN, NA, NY, US); Thoroughfare R., Viereck 1097 (GWA); jct. of Glacier Cr. and Thoroughfare R., 1320 (GWA), Jarvis Cr., Cody & Webster 4999 (DAO); Nabesna Rd., Mi. 91, Dutilly, Lepage, & O’Neill 21499 (NA); Black Rapids Glacier, L. & T. Viereck 2139 (GWA); Paxson, West S.4 (NA). Arctic Coast district: Anaktuvuk Pass, Spetzman 1766, 2701 (MIN, NA); Umiat, Churehill 477 (NA); Lepage 23688 (DAO); Spetzman 2709 (MIN, NA); Jago R., Cantlon & Gillis 57-622 (GH); Okpilak R., 57-1844 (GH); Okpilak L., Cantlon & Malcolm 58-59 (GH); Lake Schrader, Spetzman 666 (MIN); Canning R., and Ignek Cr., 3881 (MIN); Meade R., Ward 1193 (DS); Utukok R., 1388 (DS, RM); Icy Reef, Johansen 160 (NA). Yukon: bet. King and Kay Pts., A. Porsild 7173 (CAN); Dawson, Eastwood 723 (A); Kluane L., Anderson 9379 (CAN, NA); H. Raup 12185, 12404; Burwash Landing, 13287A 13940, 13961, 13968 (A); Alaska Highway, Mi. 1021, 11819, 11902 (A); Carcross, Eastwood 722 (A); Mt. Caribou, Gillett & Mitchell 4527; Teslin L., Dutilly 28439 (NA); Canol Rd., Lower Lapie R. cross- Ferguson 10923 (DAO); Schooner Landing, mouth of Anderson R., 10939 (DAO); Ft. Barbant, Dutilly 18406 (DAO, NA); Mae- kenzie R. delta, east branch, A. Porsild 7382 (CAN); Liverpool Bay, Nicholson Is., A. & R. Porsild 2862 (GH) ; Coppermine, Cemetary Is., Findlay 121 (DAO), 122 (DAO, MIN, NY); Great Bear L., east end of McTavish Arm, Schacklette 3245 (CAN, NA, US) ; Good Hope, 50 mi. kelow, A. Porsild 16739 (CAN, GH); Norman, 16705 (CAN); Norman Wells, Hicks 1 (DAO); Canol Rd., Mi. 32E, Cody & Gut- teridge 7691 (DAO, NA), 7692 (DAO). British Columbia: Lake Atlin, Eastwood 654 (A, US); Alaska Hwy., Mi. 410, McDonald Cr., Taylor, Szczawinski, & Bell 250 (staminate and pistillate sheets, DAO, UBC). Salix brachycarpa ssp. niphoclada var. fullertonensis (Schneider) Argus, comb. nov. Figs. 56 & 57 ; S. fullertonensis Schneider, Bot. Gaz. 66:340. 1918. Macoun Soe Fullerton, Hudson Bay, N.W.T., 4 Sept. 1910. ( CAN}, ee od isotype). S. niphoclada var. fullertonensis (Schneider) pes = * Gray Herb. 185:60. 1959. The holotype is supposed to be loca - the Gray Herbarium but it has not been located and I have designa’ the CAN specimen as the lectotype. ; Description. Shrubs prostrate, often mat-for Retices inches. Branches thin, trailing in some, an ; reddish-brown, sometimes yellowish, the ol grayish. Branchlets shiny, sometimes pruinose, 126 GEORGE W. ARGUS ———s ic 0 [METR II I HVT Representative specimen of Salix brachycarpa ssp. niphoclada var- 14 July 1950. FIG. ents. fllertonenis. ; Savile & Watts 960, Chesterfield Inlet, N.W.T - General view, note sle nder, spreading branches and small p leaves and am hae tad Close of staminate ament, note dark-colored bracts trelidioh) and small er appressed-pubescence, rarely densely tomentose. Buds reddish-brown to yellowish, rounded at apex, pubescent as the branchlets, some- times pruinose. Leaves narrowly elliptic to oblanceolate, obovate or ee length/ Width 2.7-3.1-4.2; blade (1.2-) 1.7-2.2 (-2.8) em. long, (0.4-) 0.7-08 (-1) cm. wide; apex acute-attenuate on narrow ae obtuse on broader leaves; base cuneate to rounded. Margin entire, veins promi- nent beneath. B ade pubescence of short Lg trichomes on both surfaces, somewhat less above, glaucous beneath; some pubescent on t m. long, narro w, acute, and glandular-margined. ———- ements eo&taneous, narrowly cylindrical, 1-1.5 cm. long; SALIX GLAUCA COMPLEX 127 borne on short reproductive shoots. Bracts light-brown to almost black, commonly reddish, puberulent on both surfaces. Glands adaxial and abaxial. Filaments two, free, glabrous, and sometimes pinkish. Anthers 0.2-0.5 mm. long, reddish in anthesis, drying dark or some- times yellowish Pistillate aments coétaneous, narrowly cylindrical, 1.5-2.5 cm, long, loosely flowered, borne on short reproductive shoot. Bracts as in the staminate flowers. Ovaries densely tomentose, short-beaked. Capsules thinly pubescent and often reddish or light-brown. Pedicels 0-0.3-0.5 mm. long, pubescent. Gland adaxial, narrow and often half as long as ovary; in some several narrow glands are arranged around the ovary. Style entire or minutely divided, 0.2-0.5 mm. long. Stigmas two, divided, short. Distribution (Maps 5 & 6). This variety is restricted to the western Canadian Arctic, its range not extending south of the tree line. It occurs from the western Canadian Arctic Archipelago, Victoria and Banks Islands, south to the Mac- kenzie Delta region, and northeastern Great Bear Lake; from there eastward across the Northwest Territories to the west coast of Hudson Bay as far south as Cape Eskimo. Discussion. Salix brachycarpa ssp. niphoclada var, fuller- tonensis is an arctic variety in the S. glauca complex (Ball, 1948) and clearly related to S. brachycarpa (Table VI). It can be distinguished from the other infraspecific taxa in S. brachycarpa by its prostrate habit, long trailing branches with appressed pubescence; small leaves with acute atten- uate apices; its small, narrowly cylindrical, few-flowered aments; and flowers with reddish bracts, drying blackish (see Figs. 56-57). The presence of stomata in the upper epidermis was noted by Schneider (1918b). This observa- tion was verified in the specimens I have examined (Table I); however, its taxonomic impcrtance is not certain. Stomata size also diverges somewhat from the “typical” pattern in S. brachycarpa so that two mean stomata lengths have been cited for the variety (Fig. 2) ; one brachycarpa- like with a mean of 20.2 microns and the other glauca-like with a mean of 27.8 microns. This evidence suggests that var. fullertonensis ig not a homogeneous unit, but may In- clude environmentally modified forms of other taxa (e.g. S. glauca) or perhaps hybrids. : The type-material of ssp. fullertonensis and ssp nipho- clada is strikingly similar and not easily separated. How- ever, var. fullertonensis can be distinguished from the rest 128 GEORGE W. ARGUS of ssp. niphoclada by its prostrate habit and characteristi- cally diminutive vegetative and reproductive structures. If it were not for the geographic integrity of var. fullerto- nensis it would be possible to treat it merely as an environ- mental variant. The common occurrence of reddish bracts in var. fuller- tonensis and the occasionally pink filaments of the stamens are of interest because a similar feature in S. brachycarpa ssp. brachycarpa was noted at Churchill, Manitoba. There, the “pinkish” color in the aments occurred at random and was relatively uncommon in the populations, in contrast to the common occurrence of reddish bracts in var. fuller- tonensis. Herbarium specimens from the Cape Eskimo area of Hudson Bay have a more robust aspect with leaves similar to typical ssp. niphoclada and aments resembling those of S. glauca (Eastern phase). The taxon in this area is further complicated by the possibility of hybridization between it and S. arctophila. Some of the specimens assigned by Schneider (1919) to S. hudsonensis Schneid. are very similar to the Cape Eskimo material of var. fullertonensis. It should be noted that S. hudsonensig was first recognized by Schneider (1918b) as the hybrid S. fullertonensis groenlandica (= S. arctophila X brachycarpa var. fuller- tonensis). On the basis of the material I have available, 1 cannot formulate an adequate understanding of either S. hudsonensis or var. fullertonensis in this area, and I prefer not to recognize the above hybrid. Collectors are urged to collect large series of specimens, in various stages of onto- genetic development in this general northwestern Hudson Bay region. For the present, Cape Eskimo material is in- cluded in var. fullertonensis with some reservation. E. Hultén (1943) reported S. fullertonensis from Nome, Alaska. If it is correct, this would be a considerable west- ward extension of the range of this western Canadian Arctic taxon. The specimen (Jones 9045) on which this report was based is pistillate and differs from var. fullertonensis in its large, broad leaves, long stipules, long aments, and stout short stems lacking the flexible trailing character of the variety. In some ways it suggests a depauperate individual of S. glauca (as noted also by C. R. Ball). In the present treatment, it is included in ssp. niphoclada. SALIX -GLAUCA -COMPLEX 129 Ecology. Salix brachycarpa ssp. niphoclada var. fuller- tonensis occurs in the dry rocky tundra where it forms compact mats of vegetation. It is common in highly exposed situations and accordingly modified by the wind. Polunin (1940) reports it chiefly as a plant of the “damper types of heath, and especially in the outermost zones of snowdrift areas.” He comments that it has no apparent ecological significance. SELECTED SPECIMENS. Northwest Territories: District of Mackenzie: Muskox L., Back R., Chilleott 177 (DAO); Toker Pt., Cody & Fergus- son 9851 (DAO); Great Bear L.: north shore of Dease Arm, A & E. Porsild 4685 (CAN, NA); Cape McDonnel, 51387 (CAN, GH). Dis- trict of Franklin: Banks Is., DeSalis Bay, Dutilly 18956 (GH, NA); Victoria Is., Holman Trading Post, A. Porsild 17270 (CAN, NA). District of Keewatin: Cape Eskimo, Macoun 79161 (GH, NA, NY); Mistake Bay, A. Porsild 5652 (GH); Ranken Inlet, Macoun 79163 (NA, NY), 79165 (NY); Fullerton, 79164 (CAN, NY); Chesterfield Inlet, Malte 120565 (NA, NY, US); Savile & Watts 960, 1206, 1207, and 1459 (DAO, NA). S.arctice X brachycarpa ssp. brochycorpe Srctica X ci Xx condida up peayhi SSP. creat orpe : ole Ss. - ” x eee i ae Sit s. . x furnorit sp. niphocioda X glovce x S. cheese x podophylte TO — — pubescent 7. Style — undivided —partly divided — more than % divided SALIX GLAUCA COMPLEX 133 The hybrid population as plotted (Fig. 58) seems to be a relatively accurate representation of the situation prevailing on Mt. Albert. A trimodal curve is produced in which Salix brachyearpa and the hybrids are most common and S. chlorolepis relatively uncommon. The intermediate taxa described by Schneider, var. antimima and X gaspeensis, fall into the hybrid zone (between values 2 and 7) and actually outline the hybrid swarm. 7 anf NUMBER Lal ik tnd 4 At -ocPchinded de B.4 AGieds VALUE - 58. Histogram of hybrid index values based on specimens of Salix brachycarpa ssp. brachycarpa, S. chlorolepis, and their hybrids from Mt. Albert, Quebec. Typical rial has va Yaives::0-3, and hybrids have values 4-8. The type specimens of S. chlorolepis var. wehbe have values of 2, 3, 4, 5, and 7.. The type of S. X gaspeensis has a value of 5. Although no one charactisic is sufficient to distinguish the hybrids, a combination of characteristics is more or less definitive. However, problems are often posed by certain individuals, making their identi- fication highly subjective. On Mt. Albert it is difficult to ascertain whether or not Salix chlorolepis is being swamped by hybridization. Since the species is a local end few seeds, although the ovary was somewhat abnormal. The ovaries 4 rapid swamping of the rarer species. 134 GEORGE W. ARGUS SELECTED SPECIMENS. Quebec: Mt. Albert, Allen, 26 July 1881* (A, GH, MIN, NY); Argus 63-59*, 65-59*, 70-59*, 78-59%, 83-59*, 96-59, 97-59*, 101-59*, 103-59*, 107-59* (DAO, GH, GWA) ; Cerlson, 31 July 1938* (NA); Collins & Fernald, 8-15 Aug. 1905* (GH, NY), 64* (A, GH, MIN, NY); Fernald and Collins 512a*, 512d*, 512e*, 512, 516, 517*, 519* (GH); Fernald, Griscom, Mackenzie, & Smith 25667* (GH, MIN, NY), 25668* (A, GH, NA, NY, US); Gosselin 3624*, 3681* (NA). Salix brachycarpa ssp. brachycarpa x glauca. S. wyomingensis Rydberg, Bull. Torr. Bot. Club 28:271. 1901. Tweedy 3434, head- waters of Clear Cr. and Crazy Woman R., Big Horn Mts., Wyoming, 20 July-15 Aug. 1900. (NY, holotype; RM!, US!, para- type). S. brachycarpa var. alticola Kelso, Rhodora 36:195-196. 1934. Kelso 3459, near Fairplay, Colorado, 10 Aug. 1982. (GH!, isotype). Hybrids between ssp. brachycarpa and S. glauca are probably more common than the specimens cited here would indicate. I have been conservative in the recognition of hybrids and the few specimens finally identified as hybrids probably represent only a portion of those present in herbaria. That hybrids can be formed by these species has been partly supported by an experimental cross made in Boston, Massachusetts, in 1960. A pistillate specimen of S. glauca from Churchill, Manitoba (Argus 82S) and a staminate specimen of S. brachycarpa ssp. brachycarpa from Mt. Albert, Quebec (Argus 106-59) were crossed. Seeds were produced in several capsules and several were germinated and grown into seedlings which ranged in height from 22 to 27 mm. The seedlings grew vigorously for three months until they were killed by a fungus infection. In spite of this evidence few individuals readily identifiable as of hybrid origin were noted in the field (Manitoba or Quebec) or in the herbarium. This may be due to different flowering times, or it may be that hybrid inviability or sterility, not revealed by the single cross made, are important isolating factors. Characteristics, or combinations, indicative of hybridization are not at present well enough known. In the artificia! hybridizations made by Nilsson (1980) examples of F.’s which most strongly resembled one parent, and others which looked like a third uninvolved species were noted. This hybrid has been recognized throughout the area of overlap of the two taxa. Few of the specimens referred to it are intermediates; more commonly they resemble one species or the other but have certain characteristics that suggest hybridity. For example, Dutilly, et al. 32312, from Old Factory, Quebec, most closely resembles S. SALIX GLAUCA COMPLEX 135 like pubescence on the leaves of some (Raup & Abbe 4508, 4597; and Loan 312), the small ovaries and short pedicels (Loan 312; Cody & Matte 8613), and the reddish petioles on (Loan 312) suggest S. brachy- carpa ssp. brachycarpa. Throughout the southern Rocky Mountains the hybrid seems to be very common. The intergradation which occurs in certain areas be- tween ssp. brachycarpa and S. glauca suggests that hybridization has been going on for a long time. Specimens from Colorado and Wyoming are often difficult to assign to one or the other species making the identification of putative hybrids highly subjective. Specimens recog- nized as hybrids have been distinguished by a series of characteristics including pedicels 0.25 mm. long or longer; long cylindrical aments; long beaked capsules, petioles more than three times the length of the buds; and leaves sparsely pubescent. Field study is needed in the southern Rocky Mountains both to understand the parental species and to assess the importance of hybridization between them. SELECTED SPECIMENS. Northwest Territories: Mackenzie R., Mills L. camp, Cody & Matte 8613 (DAO); Ft. Smith, Loan 312 (DAO). Alberta: Jasper Nat. Pk., Yellowhead Pass, Jack 2774 (NA); The Palisade, Jenkins 8002 (DAO); Lake Athabasca, east of Sand Pt., Raup & Abbe 4508 (GH); north of Sand Pt., 4597 (GH, NA). Wyoming: Albany Co.: Centennial, Jack 1070 (A); Snowy Range, Bliss 419 (RM); Brooklyn L., Goodwin & Vestal 801 (GH); Lost L., A. & R. Nelson 1140 (RM, US). Colorado: Clear Cr. Co.: Silver Plume, Rydberg, 24 Aug. 1895 (NY); Denver Co.: South Park, Hoosier Pass, Penland & Hartwell 1310 (NA); Lake Co.: Leadville, L. & E. Kelso 5128 (DAO); Larimer Co.: Rocky Mt. Nat. Park, Trail Ridge, Smith 1277 (DAO); Cameron Pass, 270 (NA); Park Co.: Alma, Ewan 12597 (NA); Antero Res., Buffalo Cr., Lane 2282 (NA); San Miguel Co.: Ophir, Maguire (NA, NY). Utah: Kane Co.: Orderville, B. Maguire 18830 (NA); Utah Co.: Mt. Timpanogos, Larsen 6798 (US); Weight 221 (NA); Wasatch Mts., Clayton Pk., Stokes, 12-26 Aug. 1903 (US). Manitoba: Churchill, Beckett M-17-9 (MIN). Ontario: Weenusk, Dutilly & Lepage 16901 (DAO). Quebec: Old Factory, Dutilly, Lepage & Duman 32312 (DAO, NA); Ft. George, Dutilly, O'Neill & Duman 97257 (NA); Anticosti Isl., Southwest Pt., Victorin & Germain 24730 (A, GH). Salix brachycarpa ssp. brachycarpa X turnorii, A specimen repre- senting this putative hybrid was collected in the sand dunes on the south shore of Lake Athabasca in 1962. It resembles S. turnorii in its red branchlets, prominent, half-cordate stipules, minute glandular teeth on the leaf margin, and pedicels about 1 mm. long; and S. brachycarpa ssp. brachycarpa in its pubescent capsules, densely pubescent bracts, and densely tomentose immature leaves. Its aborted ovules further suggest its hybrid origin. Salix turnorti Raup is is most closely related to S. lutea Nutt. This hybrid is rare in northern Saskatchewan and, although the specimen was growing vigorously, it was apparently sterile and of no evolutionary significance. Saskat- 136 GEORGE W. ARGUS chewan: sand dunes east of William R., Lake Athabaska, Argus 221-62 (DAO, GH, GWA, SASK). Salix brachycarpa ssp. niphoclada < glauca. The specimens iden- tified as this hybrid are not identical with each other in their external morphology, but exhibit various combinations of characteristics suggesting S. brachycarpa ssp. niphoclada and S. glauca. Charac- narrowly elliptic-oblanceolate leaves with acute-attenuate apices, small stipules, and narrow, loosely flowered aments: Characteristics sug- gesting S. glauca are the long petioles, large oblanceolate leaves, large stipules, and densely flowered broadly cylindrical aments. Specimens that I have studied from the Arctic Slope of Alaska suggest that this hybrid may be more common in that region than indicated by the specimens cited here. Intermediates between ssp. niphoclada and S. glauca are very common and populations often appear to be hybrid swarms. This situation has led some authors to regard ssp. niphoclada as a subspecies of S. glauca rather than S. brachycarpa as it is treated in the present paper. (Raup, 1959, and Wiggins and Thomas, 1962). This hybrid is also widespread in the Alaska Range area, especially in Mt. McKinley Nat. Park where extensive collections have been made. The staminate type of S. brachy- arpa var. mexiae Ball (Mexia 2132, staminate) referred here to this hybrid also has teratological anthers. My treatment of this hybrid has been conservative, and I have often placed specimens in one species or the other rather than label doubtful material as a hybrid. Clearly this is the safest course until much more information is available concerning environmental modi- fication, polyploidy, and interspecific hybrid relationships in both parental species. SELECTED SPECIMENS. Alaska: Western Pacific Coast district: Bristol Bay, Kvichak R., Jones 9257 ( DS, NA); Lake Kulik, K. Raup 415 (US). Central Pacific Coast district: Eklutna Flats, Dutilly, Lepage, & O’Neill 20081 (NA). Alaska Range: Mt. McKinley Nat. Pk.: Scamman 5083 (GH); Inspiration Pt., W & C. Setchell 572 (A, DS, NS, NY, US); Teklanika R., Argus 605; Highway Pass, 629 (GWA, RM) ; Muldrow Glacier, Mexia 2132, staminate (A, GH, MIN, NA, NY); Rapids Roadhouse, W. & C. Setchell 111, 113 (GH, NA); Falls Cr., Cody 6319 (DAO); Chitna, Anderson 2021 (NA). Central Yukon R. district: College, Argus 1089 (GWA); White Mts., Gjaervoll 859 (CAN); Ft. Yukon, Henderson 15046 (NA); Fortymile dist., Franklin, Anderson & Gasser 7212 (GH, NA); Wiseman, 5818 (NA). Arctic Coast district: Pitmagea R., Shetler & Stone 3256-A Rusia. 2 se Kurupa R. valley, Hodgdon 8486 (GH); Middle Chdiites . Ace an ae oe US) ; tributary west of Chandler Ba Siete bon ); agavanirktok R., Spetzman 97 (MIN); et » Cantlon & Gillis 57-1937; Okpilak L., 58-41 (GH). ritish Columbia: Haines Rd., Mi. 60, Taylor, Szezawinski, & Bell ; Mi. 98, 972 (DAO, UBC). Yukon: Dawson, Eastwood 161 (A, CAN, NA, US); Canol Rd., Mt. Sheldon, Porsild & Breitung 11610 SALIX GLAUCA COMPLEX 137 (CAN). Northwest Territories: Coppermine, Dutilly 184 (GH). Salix glauca < padophylla. S. glauca xX pseudomonticola f. swb- pseudomonticola of Hultén, Fl. Alaska and Yukon 3:529. 1943. Ss. inceps-ourayt Kelso, Biol. Leafl. 34:11. This hybrid was recognized by Hultén (2948) in the Flora of Alaska and Yukon under the name of S. glauca X pseudomonticola. Salix pseudomonticola is antedated by S. padophylla and the latter name is used in the present treatment (Little, 1953; Argus, 1957). Hultén was of the opinion that this hybrid was very widespread in Alaska and he described three “forms” of it (see-the synonymy of S. glauca). Only-his f. subpseudomonticola shows definite signs of hybridity, the others represent the “typical” form of the Western phase of S. glauca. Specimens identified as this hybrid in the present treatment have glabrate leaves with. glandular-toothed margins. Kelso also recognized this hybrid and named it S. princeps- ourayt (1946). The specimen I have seen (Kelso 5249) has aborted ovaries, and leaves with toothed margins and rounded bases. It is probably a hybrid but its ‘parentage is not without some doubt. SPECIMEN CITATIONS. Yukon: Dawson, Eastwood 211 (A, US); Hunker Hill, Macoun 54399 (GH). Colorado: San Juan Co.: Kendall Mt;:\L. & E. Kelso. 5249 (GH)... 64. > . a Salix glauca X< pedicellaris. Much of the material cited as the tung material from central Saskatchewan (106, 166, 448, 94, 368, 554, 549, et al.).. The parents maybe S. glauca and S. pedicellaris but it is possible that similar offspring could be produced by the hybrids S. glauca X planif olia, S. glauca X scouleriana, or even S. pedicellaris X planifolia. n some parts of the range of the ferruginous form ho Salix glauca there is no suggestion of hybri idity and many specimens produce seeds which have been germinated (Argus 517-58, 490-58, 456-58; Breitung 2154; and Scoggan 6026, 6088A). At this time it is impossi- e to ascertain just how much of the ferruginous form is of hybrid origin, or the exact nature of the hybrid. More study is needed and investigations now underway in Saskatchewan may help answer some perplexing questions. See also, discussion of S. athabascensis and S. in (synonyms of S. glauca), and the Rocky Mountain phase of SELECTED D SPECIMEN. Saskatchewan: McKague, Breitung 94 (DAO), 106, 166 (A, DAO), 363 (DAO), 448 (A, DAO, NA), 549, 554 (A). Northwest Territories: Enterprise-Mackenzie R. Hwy., Mi. 41, Thieret & Reich 5416; Mi. 44, 5402 (F, GWA). LITERATURE CITED ANDERSON, E. 1941. The technique and use of mass collections in plant taxonomy. Ann. Missouri Bot. Gard. 28: 287-292. | 138 GEORGE W. ARGUS Suigeomeates N. J. 1868. mage Boreali-Americanae. A synopsis of rth American willows. Proc. Am. Acad. 4: 3-32. het G. W. 1957. The Sallb ie of Wyoming. Univ. Wyo. Publ. 21: 1-63. . 1961. The taxonomy of the Salix glauca L. complex in North ‘Aitisri¢a. Unpubl. Ph.D. Thesis, Harvard University. BALL, C. R. 1899. Notes on some western willows. Trans. Acad. St. Louis 9: 69-90. —_———. 1909. Salix. In Coulter, J. and A. Nelson. New Manual of Botany of the Central Rocky Mountains. New York. 646 pp. ——————.. 1934. New or little known west American willows. Univ. Calif. ges Bot. 17: 399-434, ———_———.. 1948. Schneider’s three new Canadian willow species (S. anamesa, S. fullertonensis and S. hudsonensis). Canad. Field-Nat. 62: 150-152. BARRATT, J. 1840. Salices Americanae: North American Willows. Middletown, Conn. pp. 7, unnumbered. rc M. S. 1885. Salix. In Coulter, J. Manual of the Botany of the Rocky Mountain Region. New York. 452 Buss, L. C. 1956. Comparison of plant development in microenviron- ments of arctic and alpine tundras. Ecol. Monog. 26: 303-337. BoécHeEr, T. W. 1938. Biolcgical distributional types in the flora of Greenland. Meddel. Gronl. 106: 5-339. . 1952. Contributions to the flora and plant geography of west Greenland, III. Vascular plants collected or observed during the botanical expedition to west Greenland 1946. Meddel. Gronl. 147: 1-85. BrEITUNG. A. J. 1957. Annotated catalogue of the vascular flora of Saskatchewan. Am. Midl. Nat. 58: 1-72. CaLprr, J. A. AND D. B. O. SAvILE. 1959. Studies in the Saxifraga- ceae, I. The Heuchera cylindrica complex in and adjacent to Brit- ish Columbia. gic 11: 49-67. Camus, A. AND E. G. 1904. Classification des Saules d’Europe et meray ae des “ta de France, I. Paris 386 pp. CovILLE, F. V. 1901. The willows of Alaska. Proc. Wash. Acad. Sci. 3: 297-362. CRITCHFIELD, W. 1957. Geographic Myspsotie in Pinus contorta. Maria Moors Cabot Foundation, Publ. 3. Drury, W. 1962. Patterned ground oe oceslatie on southern Bylot ner Northwest Territories, Canada. Contr. Gray Herb. 190: Dumortier, B. C. 1825. Verhandeling over het geslacht der Wilgen (Salix) en de Natuurlijke Familie der Amentaceae. Bijdr. ot Nat. Wetensch. pp. 44-61. oe G. 1952. Pollen Morphology and Plant Taxonomy. Stock- 0 Farcri, K. Bite, Ob the peri-glacial flora of Jaeren, with notes on the Papa of the Salix pollen grain. Norsk Geogr. Tidskr. 14: SALIX GLAUCA COMPLEX 139 FERNALD, M. L. 1905. An anomalous alpine willow. Rhodora 7: 185- 186 _ 1926. Two summers of botanizing in Newfoundland. Salix siaepetaes Pursh. Rhodora 28: 181-188. FIsHER, M. J. 1928. The morphology and anatomy of the flowers of the Salicaceae, I and II. Am. Jour. Bot. 15: 307-326, 372-394. FLINT, R. 1957. Glacial and Pleistocene Geology. New York. 553 pp. FLODERUS, B. 1931. In Holmberg, Skandinaviens Flora, Ib. Stockholm. pp. 160. _ 1923. Om Grénlands Salices. Meddel. Grenl. 63: 61- 204. Foster, A. 1934. The use of tannic acid and iron means for staining cell walls in meristematic tissue. Stain Tech. -92. Fries, E. 1828. Conspectus descriptionis pala, "Sueciae. plantarum novarum, Ratisbonae. pp. . 1832. Novitiae Florae Sueciae. pp. 1-84. _ 1840. Svenska Pilarterne efter naturliga férandskaben Sintdneder: Bot. Not. 2: 145-153, 177-188, 193-206. FrYE, T. C. AND G. RicG. 1912. Northwest Flora. Seattle. pp 45 GREENLEAF, W. 1938. Induction of polyploidy in Nicotiana. Jour. Her. 29: 451-464. HALLIDAY, W. AND A. BRowNn. 1943. The distribution of some im- portant forest trees in Canada. Ecology 24: 353-373. HsELMaQuist, H. 1948. Studies on the floral morphology and phylogeny of the Amentiferae. Bot. Not. Suppl. 2: 1-171. Hgec, O. A. 1929. Pollen on the eauaie from Ellesmere Land. Det. Kong. Norske Forh. 2: 55- HoLMEN, K. 1952. Cytological te in North Greenland. Meddel. Gronl. 128: Hooker, W. J. 1838. Flora irealsaeeecaein 2: 144-153. London. Sylloge -38. Mantissa Prima. Lundae. the flora of Peary Land, Husss, C. AND A. PERLMUTTER. 1942. Biometric compariso f sev- eral samples, with particular reference to racial investigations. Am, Nat. 76: 582-592. of — and boreal biota Huitén, E. 1937. Outline of the history during app Period. Stockholm. 165 p 1941. Flora of Alaska and Yukon, Part Resse: qi: Sect. 2. 37: 3-127. = . 1948. Flora of me and Yukon, Part III. Lunds Univ. ies. II. Sect. 2. 39: 415-567. 1948. Flora of Alaska and Yukon, Part VII. Lunds Univ. Dekker: II. Sect. 2. 44: 1203-1341. ¥ : 960. Flora of the Aleutian Islands. Second edition. Wein- heim/Bergstr. 376 pp Husticu, I. 1950. pt bornde on the forests on the east coast of Hudson Bay and James Bay. Acta Geogr. 1: 3-83. ; JORGENSEN, C., T. Sorensen, and M. Westergaard. 1958. The aged plants of Ginen! land. A taxonomical and cytological survey. Skr. Dan. Vid. Selsk. 9: 1- 172; I. Lunds Univ. 140 GEORGE W. ARGUS Ketso, L. 1946. The Rocky Mt. Flora I. The willows. eer nine 34, KnutH, P. 1909. Handbook of Flower Pollination, Vol. HI. Trans- lated by J. Ainsworth Davis. Oxford. 644 pp. Kocu, G. D. 1828. De Salicibus Europaeis commentatio. Erlangae. pp. LAGERKRANZ, J. 1950. Observations on the flora. of west and east eR omOyn Acta Reg «. Soe; : Sci. Upsaliensis. Ser. IV. 14: im 9-14 LANGE, ee 1880, Conspectus Piarse Groenlandicae. Meddel. Gronl. 83: 1-281. Lasrouw, J. AND F. STAFLEU. 1959. Index Herbariorum. Pt. I. The ~: herbaria of the world. Ed. 4. Regnum. Vegetabile 15: 249 pp. LEDEBOUR, C. F. 1850. Flora Rossica. 3: 493-684: Stuttgartiae. LITTLE, E.\ 1953. Check List of Native and Naturalized Trees of the United States cela etic Taegan U.S. For. Serv., Agric. Hand- book 41, 472 pp. LonG, F. ann F. ieee 1934, The method of collodion films for stomata. Am. Jour. Bot. 21: 7-17. Love, A. 1954. Snap cosapppamii evaluation of corresponding taxa. tisniairooe 5-6: 212-224-0777 ————. AND D. Love. 1948. Chromosome numbers of northern 4 siigs tsi Teel. Univ. Inst. Appl. Sci., Dept. Agric., Rep. B. 3: 31. a ——_.. . Cytotaxonomical conspectus of the fesandi flora. Acta Horti Gothob. 20: 65-290. DP. SARKAR. isi het icoeaueaich Mimeographed. 41 pp. MarTor, B. AND R. Wome 1959. Geographic variation in the wood frog, Rana sylvatica. Am. Midl. Nat. 61: 350-389. METCALF, C. R. AND L. CHALK. 1950. Anatomy of the Dicotyledons. London. 2 vols. 1500 pp. Moss, C. E. 1914. The Cambridge British Flora, Vol. 2. Cambridge. MUHLENBERG, H. F. 1805. ttber die Nordamerikanischen Weiden. Neue Schr. Ges. Nat. Freunde 4: 233-242. Munz, P. AND J. LAUDERMILK. 1949. A neglected character in petals ashes (Fraxinus). Aliso 2: 49-62. Nasarov, M.H. 1936. Salix in Komarov, Flora USSR. 5: 91-242. Lenin- grad. Nitsson, N. H. 1930. Synthetische Bastardierungsversuche in der Gattung Salix. Lunds Univ. Arsskr. II. Sect. 2. 27: 3-97. POLUNIN, N. 1940. Botany of the Canadian Eastern Arctic, I, Nat. us, Canada Bull. 92: 1-408. . 1943. Contributions to the flora and phytogeography of south-weeters Greenland: an enumeration of the vascular plants, with critical notes. Jour. Linn. Soc. Bot. 52: 349-406. Porsitp, A. E. 1955. The vascular plants of the western Canadian “Archipelago. Nat. Mus. Canada Bull. 135: 1-226. ' Porsitp, ‘Mf. P. 1920. The flora of the Disko Island and 'the agjaceme coast of West Greenland. Meddel. Grenl. 58: 1-156. SALIX GLAUCA COMPLEX 141 Rarnio, A. J. 1926. ther die Intersexualitaét bei der Gattung Salix. Soe. Zool.-Bot. Vanamo Ann. Bot. 5: 2. RAnp, A. 1948. Glaciation, an isolating factor in speciation. Evol, 2: 314-321 RANDOLPH, a 1932. Some effects of high temperatures on polyploidy and other variations in maize. Proc. Nat. Acad. Sci. 18: 222-229. “eg H. M. 1930. A new species of Salix from the Mackenzie Basin. Rhodora 32: 111-112. 1931. Salix glauca L. and its allies in the Athabaska- Great slave Lake region. Rhodora 33: 241-244. 1935. Botanical investigations in Wood Buffalo Park. Nat. Mus. Canada Bull. 74: 1-174. 1936. Phytogeographic studies in the Athabaska-Great Slave L pike region, I. Catalogue of the vascular plants. Jour. Arnold Arb. 17: 180-315. 1943. Willows of the Hudson Bay region and the Labra- dor Peninsula Sargentia 4: 81-127. . Phytogeographic studies in the Athabaska-Great Slave inthe ene Il. Jour. Arnold Arb, 27: 1-85. 947. The botany of southwestern Mackenzie. Sargen- tia 6: 1- 275. 1959. The willows of boreal western America. Contr. Gray Hert. 185: 3-95. REED, J. F. 1955. Seeds bo seedlings of Salix brachycarpa Nutt. Colo. -Wyo. Acad. Sci. 4 REHDER, A. 1949, ioahesataaal of Cultivated Trees and Shrubs. The Arnold i Harvard Univ., Jamaica Plain, Mass. pp. 825 RITCHIE, J. C. 1956. The native plants of Churchill, Manitobs, Canada. Canad. Jour. Bot. 34: 269-320. RypBerc, P. A. 1899. Caespitose willows of Arctic America and the Rocky Mountains. Bull. N.Y. Bot. Gard. 1: 257-278 906. Flora of Colorado. Fort Collins, Colorado. . 1917. Flora ag the Rocky Mountains and Adjacent Plains. Nes York. 1110 p SAWYER , JR. 1932. Seis apparatus of the cultivated cran- berry anoint macrocarpon, Am. Jour. Bot. 19: 508-513. Sax, K. anp H. Sax. 1937. Stomata size and distribution in diploid and polyploid plants. Jour. Arnold Arb. 18: 164-172. eel C. 1904. Handbuch der Laubholzkunde. Vol. 1. Jena. pp. 918a. Notes on American willows. I. The species re- lated to Sal arctica Pall. Bot. Gaz. 66: 117-142. 1918b. Notes on Aiserican willows, II. The species related to Sees glauca L. Bot. Gaz. 66: 318-353. ae eee SOL ee oo on American willows, III. A conspectus of ee and varieties of sections Reticulatae, Her- baceae, Ovalifoliae, and Glaucae. Bot. Gaz. 67: 27-64. ee Notes on American willows, XII. Systematic enumeration of the sections, species, varieties, and forms of Ameri- can willows. Jour. Arnold Arb. 3: 61-125. 142 GEORGE W. ARGUS Scoacan, H. J. 1957. Flora of Manitoba. Nat. Mus. Canada Bull. 140: 1-619. SEEMEN, O. VON. 1903. Salices Japonicae. Leipzig. pp. 83. SEIDENFADEN, G. 1933. The vascular plants of south-east Greenland 60°04’ to 64°30’ N. lat. Meddel. Gronl. 106: 1-129. SmitH, E. C. 1942. The willows of Colorado. Am. Midl. Nat. 27: 217- 252. —_——. 1954. Salix. In aia H. D. Manual of the Plants of Colorado. Denver. 666 p S6YRINKI, N. 1939. Studien tiber nae Generative and Vegetative Vermehrung der Samenpflanzen in der Alpinen Vegetation Pet- amo-Lapplands, II. Soc. Zool.-Bot. Vanamo Ann. Bot, 14: 1-404. SPETZMAN, L. 1959. Vegetation of the Arctic Slope of Alaska. U.S. Geol. Surv. Prof. Paper 302-B. 58 Stone, D. 1961. Ploidal level and stomatal size in the American hickories. Brittonia 13: 293-302 STRAKA, H. 1950. Untersuchungen iiber Salix-Pollen. Proc. 7th Int. Bot. Cong. Stockholm. pp. 874-876. Sucaya, 8. 1956a. Terminal bud formation in Salix, Tohoku U. Sci. Rpt. Ser. 4 (Biol.) 22: 5-11. ————. 1956b. Further notes on the Meese bud formation in Salix. Tohoku U. Sci. Rpt. Ser. 4 (Biol.) 22: TAKEDA, H. 1936. On the coma or hairy tuft on ae seed of willows. Bot. Mag. Tokyo 50: 283. TOEPFFER, A. 1915. Salices Bavariae. Bayer. Bot. Gesellsch. 15: 17- ee E. R. 1833. Salix. In Ledebour, D. Flora Altaica. 4: ————_—————. 1836. Uber die Weiden des Hortus Hostianus und der Dendrotheca ‘askae Linnaea 10: 571-581. VELENOVSKY, J. 1904. Vergleichende studien iiber die Salix-Blute. Beih. Bot. Centr. 17: 123-128. Watson, L. 1962. The taxonomic significance of stomatal distribu- tion and morphology in Epacridaceae. New Ph ytol. 61: 36-40. WicuurA, M. 1865. Die Bastardfruchten im Pflanzenreich erliutert an den vty as der Weiden. Breslau. pp. 95. Wicerns, I. D J. THOMAS. 1962. A Flora of the Alaskan Arctic Slope. Worst 425 WILKINSON, J. 1944. The cytology of Salix in relation to its taxonomy. Am. Bot. IT. 8: 269-284. Witson, K. 1958, Ontogeny of the sporangium of Phlebodium (Poly- podium) aureum. Am. Jour. Bot. 45: 483-491. CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C, Rollins and Robert C. Foster NO. CXCVII THE SOUTH AMERICAN SPECIES OF GUT By Otto T. SoLBric CHROMOSOME NUMBERS OF CRUCIFERAE By Reep C. Roiuins PUBLISHED BY THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U. S. A. 1966 kee, SH gr) CLIT 7) ne, 1/97 196?S CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY No. *1. Roprnson, B. L. Descriptions of New Plants, Chiefly Gamope- talae, collected in Mexico by C. G. Pringle in 1889 and 1890. (Reprinted from Proc. Am. Acad. XXVI. 164-176. 1891.) $ .50 *2. pes wed ey B. L. Description of sh Plants collected in Mexico, Cc. G. Prin ngle, in 1890 and 1891, with Notes upon a few ate eer (Repri inted from ag Am. Acad. XXVII. —-185. 1892.) *3. Ropinson, B. L. anp SEaToN H. E. Additions to the Phaeno- gamic Flora of Mosicn, discovered by C. G. Pringle in 1891-92. (Reprinted from Proc Am. Acad. XXVIII. 103-115. 1893.) 50 *4. Seaton, Henry E. New and Little Known Plants collected on Mount Orizaba i in the Sanimel of 1891. (Reprinted from Am. Acad. XXVIII. 116-123. 1893.) 50 *5. Rosrnson, B. L. The North American Sileneae and Polycarpeae. (Rep rinted from Proc. Am. Acad. XXVIII. 124-155. 1893.) 50 . Rosrinson, B. I. The North American Alsine Il. es © of New nary “Hitherto Imperfectly Known nts contained in C. G. Pringle’s Mexican Collections a 1892 and 1893. * o soga. IV. Miscellaneous Notes and New Species. Dapp ae from Proc. Am. Acad. X XIX. 273-330. 1894.) 75 . Rosinson, B. si anp GreEeNMAN, J. M. Further New and Im- perfe ctly Known Plants collected i in ites by C. G. Pringle in the Summer of 1893. (Reprinted from Proc. Am. Acad. X XIX. 382-394. 1894.) 50 ‘ Ropixson, B. L. anp Fernatp, M. L. New Plants collected by ess ” Hartman and C. E. Lloyd upon _Archaeolo ogical Expe dition to Northwestern Mexico under the Direction of Dr. Carl Lumholtz. (Reprinted from Proc. Am. Ac ad. Vol. 30, 114-123. 1804.) 50 . Rosrnson, B. L p GREENMAN, J. M. I. On n the Flo Flora = the “Galapagos Islands, as shown by the II. New ar oteerthy Prahia: chiefly from Oaxaca, col- lec =e _C. G. Pringle, TN ie rg 4 * i?) 3 Th A “aber Revision of the Genus Lamouro IV. Miscellaneous New Species. ‘Reprinted fr from m Am. Journ. J 135-176. 1895.) 50 *10. RoBINson, B. L. a Ap roms od - M I. Revis wor the Genus Tri : : Il. Synncais of ae Mexican as ‘Central American Species of e Genus ania. IT. Sievision of f the ‘Genius Zin IV. Revision of the Mexican ar Central American Species of nus Calea. V. A Provisional Key to the Species of Porophyllum ranging Abeta! of the Isthmus of Panama. VI. Description of New and Little Known Phanerogams, chiefly trai Dares (Reprinted from Proc. Acad. a. @. EG 3-51. 1896.) *Out of print. _ _ . GREENMAN, J. M. be Revision of he Mexican and Central American Species of II. Key it the I Mexican Species of Liabum III. Descriptions of New or little known Plants from Mexico. ae ted from Proc. Am. Acad. XXXII. 283-311. —_ bo . FERNALD TT: Sone Rare a a ede eribed Plants collected by Dr. Edward hina: at Acapulco, Mexico. (Reprinted from Proc. Am. Acad. XXXL 57-94. 1897.) — i) 5 me nm ° ae cae . Revisi Vision ¢ of the North American and Mexican Species of Il. Revision « of the North American Species of Neptunia. (Reprinted from Proc. Am. Acad. XX XIII. 305-334. 14. GREENMAN, J. M. I. Revision of the Mexican and Central American Species of Galium and Relbunium II. Diagnoses of New arg Critical Mexican Phanerogams. isn fen Proc. Am. Acad. XXXIII. 455-489. . Fernazp, M. L. I. Eleocharis ovata and its American Allies II. Scirpus Eriophorum and some Related eatha. With one plate. oo from Proc. Am. Acad. XXXIV. 485- . RoBINson, B. L. anD GREENMAN, J. M. I. Revision ion of he emia: "Montanoa, Perymenium, and ~~ i — lor) Il. Symons of = the Saries Verbesina, with an analytical Key to ene ENMAN, IM III. Some new Species, extended Ranges, and — noted dentities among the Mexican Phaner printed from Proc. Am. Acad. XXXIV. 507-378. " 1899.) 17. Rosrnson, B. = AND GREENman, J. M. I. Revisi —* Pre nus Gymnolomia. Supplementary notes ine ea, idax ikania. i fons ee Bon t. Soc. Nat. ve tee E ge and Varieties of Mexican nts. RoBINson, B. ee I ore of the Genera Jaegeria and Russelia, Uuing, E. lL. Ne i ™m Mexico. Rosinson, B. L. Ay. Ne ew Phaenogams, chiefly ag ar from Mexico and oe America. (Reprin rinted Am. Acad. pgm 342. 1900. I ay Pee S wn Species of Sa Il. A Revision of the Mexican and 15 75 75 75 2.00 iw) ~— 8 3 8 . Roprson, B, L. I. Synopsis of me Genus Melampodium. II. Synopsis of the Genus Nocca. III. New Species and newly noted Synonymy among the Spermatophytes of Mexico and Central America. (Reprinted from Proc, Am. Acad. XXXVI. 1901). . Fernaxp, M. L. me new Spermatophytes from Mexico and Central ress apenas ted from Proc. Am. Acad. XXXVI. 491-506. 1901.) . Fernatp, M. L. I. The or possi of the Section Hyparrhenae. II. The Variation of some Boreal Carices. With te plates. (Reprinted ome Proc. aa Acad. XXXVITI. 447-5 1902.) aM . The World Birches. With two la tes. (Reprinted from Am. Jour. ; RoBINsoN, B. L. Flora of ‘he eu. Islands. With three ates. (Reprinted from Proc. Am. Acad. XX XVIII. 77-269. . GREENM AN, J. M. Newand Otherwise Noteworthy Angiosperms from Moras and Central America. (Reprinted from Proc. Am. Acad. XXXIX. 69-120. 1903. . Jonnston, J. R. A Revision of the Genus Flaveria. (Reprinted from Proc. Am. Acad. XX XIX. 279-292. 1903.) Gre nus Sabazia. II. Revision of the Mexican and American Species of Trixis. III. Revision of the Mexican sae Central American Species of ‘Hieracium. F ernald, _ ve Synopsis of the Mexican and Central American Species REENMAN, M;Y¥, oses and Synonymy of Fee and cane American S ecawieea ew Spe of Mexican and Nic: cies sane Dicotyledons. ‘ean nied geo Proc. Am. Acad. XL. 3-57. Jounston, J.R. New Plants from the Islands of Margarita and he, , Venezuela, (Reprin rom Proc. Am. Acad. XL. 5. Jones, W. W. A Revision of the Genus Zexmenia. (Reprinted from Proc. Am. Acad. XLI. 143-167. 1905. - GRE f Spermatophytes from See ENMAN, J. M. I. Descriptions of Spe: ia hacer Southwestern United States, Mexico, an Roginson, B. L. II. Diagnoses and Notes Relating to Anak: potas Eupatoriae. (Reprinted from Proc. Am. Acad. XLI. 235- 1905. Roginson, B. L. Studies in dra sage Fp em I. Revision of the Genus Piqueria. : the Genus Oe ape patio III. The Genus Helogyne sae ‘its Synonyms. IV. Diagnoses and Synonymy of Eupatorieae and of ce rtain other Scanian which = been erg with them. (Reprinted from Proc. Am. Acad. XLII. 3-48. 1906.) 3.00 75 1.00 75 1.00 F 2 oo I ww =) ae a &S ; oorE, ALBERT Hanrorp. Revision of the Genus Spilanthes. * iteprnted from Proc. Am. Acad. XLII. 521-569. 1907.) . GREENMAN, : New Species “ ag 80 and Schoenocaulon from Mexico. RoBINson, B. New or otherwise Note- worthy Spermatophyta chiefly from Mexico. Rosinson, Bartiert, H. H. III. New “Plants from Guatemala and Mexico paleciod chiefly by C. eam. Frrnazp, M. L. V. Diagnoses of New Spermatophytes from Mexico. (Re- 19 07. . Fernatp, M. L. The Soil Preferences of Certain Alpine and Sub- Alpine Plants. (Reprinted from Rhodora, [X. 149-193. 1907.) . Eastwoop, A. Synopsis of the Mexican and Central American Species of Castilleja. Rosinson, B. L. II. A Revision of the Genus Rumfordia. Barr.ert, H.H. UI. A Synopsis of the American Species of Litsea. Eastwoop, A. IV. Some Unde- scribed Species of Mexican Depene each Bartiett, H. H. V. Notes on Mexi ican on ral Amerie n Alders. Rosin- H. I. Description of Mexican Phanerogams. (Reprinted from Proc. Am . Acad, XLIV. 363-687. 1909.) . JOHNSTON, JoHn Roper. Flora of the Islands of Margarita oes oche, Venezuela. Wit plates. (Reprinted from ost at. His 1909. q WEATHERBY, C. A. I. A preliminary igen of the Genus Echeandia. Rosinson, B. L. II. Spermatophytes, new or Sarai chiefly pg pao and Gentianaceae. WEATHERBY, C. A. . erican Forms of Lycopodium sorepiaane FERNALD, M. L. IV. — au little known Mexican Plants, chiefly Labiatae. Wraru , C. A. V. Mexican Phan — gams—Notes and S Proc. Acad. XLV, 387. 428. Species. (Reprinted from Pr - Rozinson, B. L. On the Classification of net Bupsiods ane: II. Revision of G. I. Desmodium: Sacer ag Studies— i: modium procumbens and lated Species. B. Misbtilencous Notes and Records. Smith, L. B. II. Studies in the Bromeliacese— With 3 plates. (Published by the ray Herbarium, 1-35 p. 1 130. Rosspacu, Rutu P._ Spergularia in n North and “sie greg With 8 sie: (Reprinted from Rhodora, XL pp 1940.) 1,25 2.00 1.25 2.00 1.00 1.25 1.50 1.25 1.00 1.25 . Rouuns, Reep C. onographic Study 7 . Rouins, REED C. I. Studies in the Genus Hedysarum in North Fernald, M. L. sch Some Spermatophytes of East- ern Nort merica. Rollins, Ree n two weedy Crucifers. With 29 plates. Tinaieied from Rhotora, XLII. . 40. . ScamMAN, Eprru_ A list of Plants from Interior Alaska. (Re- 1940. printed from Rhodora, XLII, 309-349 pp. . Fernatp, M.I. A Century of Additions to the Flora of Virgin- 161 ‘a. With 24 plates. (Reprinted from Rhodora, XLII. . Trve ‘, JR., R. M._ Revision of the Genus eps Big h 4 pp. ple tes. ; . Foster, RopertC. I. Studies in Sn aerate Schubert, Pernice G. IJ. Desmodium: Preliminary Phe Type W.H. TIT. ‘the Genus Charianthus. IV. 2S sia af Heliconia, eet; plates. (Published by a cS erba- Pp ; WHEELER, Lovis C. Euphorbia Subgenus Chamaesyce in Can- ada and the United States Exclusive »f Southern Florida. With 15 plates. (Reprinted from Rhodora, XLIII. 160 pp. = SMrrH, L. B. Studies in the Bromeliaceae—XII. With 3 ates. (Reprinted from Lilloa, VI. sarge pp. ; roeh is in of A orth America. Wit illustrations. (Reprinted in p. 1941.) . FerNatp, M. L. Another Century cf Additions to the Flora of Virginia. Pr esha 26 plates. (Reprinted from Rhodora, XLIJI. With 13, pi ates. fe rinted from Rhodora, LIV. . WeaTuersy, C. A. A List of Type Specimens in Elliott’s Her- 2 ‘ — Reotated from Rhodora, XLIV. 249-264 pp With 6 Ricuarp A. Studies of the Icacinaceae IV and V. With 6 plates. "(Published by the Gray Herbarium, 1-92 pp. ; tne JR. ,R.M. A Revision of the Genus Doryopteris. wes : illustrations. (Published by the Gray Herbarium, 1-80 p . Fernatp, M.L. Critical Notes on Carex. wey plates. (Re- 194 oo from Rhodora, XLIV. 281-334. - Fernatp, M. L. The Seventh Centu ury of Adiitio ns to the Flora of Vir nia. With 28 plat ted f: Rhodora, XIV. = 1942) plates. (Reprin rom E North America North of M lates. (Reprinted Sam Hide or h ce ah With 3 plates. (Rep M . FerNatp, M. L. I. Five saat thei Rhizomatous Species of Muhlenbergia. II. Notes on Danthonia. III. Erianthus Angelica triquinata‘? VIII. } on Hieracium. With 21 plates. (Reprinted from Rhodora, LY. 3.) . FERNALD, . L. Virginian Botanizing under Restrictions. pp- ra plates. (Reprinted from Rhodora, XLV. 2.00 . FerNauLD, M. L. Overlooked Species, Transfers and ge in the Flora of ey North America. ¥ bor 10 plates. (Re- printed from Rhodora, X 47 p GaLs, Suiritey. Rhyncho nigh Section ‘Trive chospora, in Canada, the United States and the West Indies. ith 18 plates. (Reprinted rout 8 Rhodora, XLVI. 152 pp. 1944.) ; se te BERNARD. American Thalictra and their Id World Lier 5 plates. (Reprinted from Shadeea XLVI. . FuRNatp, M. L. Ruellia in the Eastern United States. With 37 plates. wi ee from Rhodora, XLVII. 77 pp. 1945.) . ScHUBERT, BERNI Fae be lication of Jacquin’s Icones Plantarum Ravlorais. Smith, Lyman B. & Schubert, Bernice -II. Studies in deg Begoniaceae, —I. A. Miscellaneous Nov- elties. B. Bego of Sessé and Mocifio. me» Lyman ar ew 6 apy ‘Published by the Gray Herbarium, 1-45 pp. wer . Foster, Rose C. I. Studies in the Iridaceae,—III. Three Ne i i rt. mer Genus Rigidella Lindl. 4. Notes on Mastigostyla Waker . A Revision of the North American Species of 2 a sche Nutt. iscellaneous Notes an agnoses. 7. Tenta- tive Keys to the Indigenous American ec Il. A mon r Zuce. I Rediscovery of Riesenbachia Pres] Note on the Localit — Ecklon Zeyher and Dre ége. V. Miscellaneous gnoses and "ehafere: (Published by the Gray Herbarium, 1-72 pp. 1945 ‘ FERNALD, M.L. Botanical papel or of eed Seward Forest and Adjacent Areas of Southeastern Virgi With 36 plates. (Reprinted from Rhodora, LXVIL OB p . Fernatp, M.L, I. Key to Antennaria of rhe “Manual oy 7 i, Transfers in oe f oo vere on I Senecio congestus. IV. Notes on eastern American Luzula. ery 51 plates. (Reprinted from Rhadors: Rui 41 pp. . Wear rr. A Apams, JouN. A List 2 the Vascular HERBY, C. A. Plants of Grand M anan, Charlotte Count ty, New Brunswick. 1945.) 2 plates. (Published by the Gray Herbarium, 1-96. . Fern Some North Ameriean Corylaceae (Betu- M. L. lncene). With 27 plates. (Reprinted from Rhodora, XLVII. -) pp. . FERNALD, M L. Technical Studies on North American Plants, e ian Transfers in Polygonum. 4. Novelties in our Flora. ith op plates. (Reprinted from Hhodors, XLVIII. 63 pp. 6. . Foster, Rosert C. I. Studies in the Flora < Bolivia,—I. Schu ubert, Bernice G. Il. The eerie of Sessé and mi ? lished by the Gray Herbarium, 1-38 p . Pate, M.L. Identifications and Reidentifications of North erican Plants. With 20 plates. (Reprinted from Rhodora, XLVIII. 50 pp. 1946.) 1.00 2.00 1.50 1.50 1.50 171, 173. . FERNALD, M.L. Additions to and Subtractions from the Flora of Virginia. With 30 plates. (Reprinted from Rhodora, XLIX., 88 pp. 1947.) _ Smira, Lyman B. & Scuusert, Bernice G. The Begoniaceae of ce. With 18 plates. (Reprinted from Caldasia IV. 98pp. 1946.) . A Misceviany In Honor oF Merritr LyNDON FERNALD. With 6 — — —encot e (Published by mi Gray Herbarium, 1-15 osteR, Rospert C. I. Studies he Iridaceae,—V._ II. Studies i in the Flora of Bolivia, 1 “Published by the Gray Herbarium, 1-43 pp. 1948.) . FERNALD, M. L. & ScuuBert, Bernice G. Studies of American ¥ Types in ge Herbaria. With 21 plates. (Reprinted from Rhodora, L. 73 pp. \ — STANCE, LINc A on of Phacelia Subgenus osmanthus (H aecihy isis With 1 car’ & 7 figures. (Published by the Gray Herbarium, 1-48 pp. 1949.) . FERNALD, M. L. & Scuusert, BERNICE z. Part I. Some Identities i in Breweria. Fernald, M. L. Part i Studies 7; astern American Plants. . Blackberries, old and new es e-characters in and minor Forms of Viola. 3. Som a! in Oenothera. 4. Emendations in the Order Tubi- 5. Some Northern Astereae. 6. Transfers and minor Variations eae 30 plates. (Reprinted from Rhodora, LI. 60 pp. . Woop, Carrotu E., Jr. The American Barbistyled_ Species of Tephrosia (Legu inane With 4 plates. (Reprinted from Rhodora, LI. 185 p 949.) Ameri i figure lished ip. the Gray Herbarium, 1-53 pp. . Rotus, Resp C. The Guayule Rubber Plant and Its Rela- tives. With 1 late and 17 fi : lished by the Gray Herbarium, 1- 2p D. 1950 950.) Ss : — QuenTIN. A taxonomic Study of the Genus Cyto Dis- in North America. es 7 figures. (Published by the Gray Hocharicm, 1-39 pp. 1951.) 4 174. — Ricuarp A. The Vegetation of the Grenadines, Wind- lands, British West Indies. With 12 plates and 5 maps. (Published by the Gray Herbarium, 1-129 oF 1952.) ‘haze. Sarrn, C. Eartt, Jn. The New World —e of Sloan ea (Elaeo- carpaceae). With, 3 B pintes (Published by the Gray Herba- rium, 1-114 pp. 1.50 2.00 3.50 1.25 2.00 1.50 2.00 2.00 1.25 3.00 No. 8 176. X 177. K 178. X 179. ¥ 180. a xX 182. Asse, Ernst C. Vascular plants of the Hamilton River area. Labrador, WensTER, Grapy L. Studies of the Euphorbiaceae, Phyllanthoideae. I. Taxonomic notes on the West Indies eeciet of Phyllanthus. Sige 5 figures. (Published by the Herbarium, 1-63 p 1955.) oa: REeep C. A revisionary study of the tg Menon- villea (Cruciferae). RLEY, WINIFRED erns of Liberia. With 5 plates. (Published by ‘the yi Her- barium, 1-103 pp. 55.) Drury, aoa M., Jr. aes: flats and phynoersente eee esses in the Upper Kuskokwim River region, Alas og (Published by he. Gray Herbarium, mn pp 956. Tryon, Rota. A — of the American hasty ies of Notho- laena. With 1 port., 58 i 67 maps. (Published by the Gray Herbatone 1-106 p 1956. ) CLEMENT, I. D. Studies in Sida Ses I. A review of the genus and monograph of the sections Malacroideae, Physalodes, Pseudomalvastrum, Incanifolia, Oligandrae, Pseu- er rec ark ge age T Stantata (Published by the Gray Herbarium, 1-91 pp. 1957. Rouuins, REED C. nterpei Sigmar in ate (Cruciferae ). Pee 3 R. A re avake of the enus 2 {Comastae). With ng maid and 5 sry Mes by the Gray otboru 1-132 pp. DreEssLER, ROBER The genus Pedilanthus (Euphorbi- aceae). With 21 slates, 28 par 6 maps. (Published by the en Herbarium. 1-188 p 1957.) Thm systematic and evolution of — —— ioe si ae “With 23 plates, 17 maps, 2 charts. lia (Comps the a4 Herbarium, 1-151 pp. 1957, ) . Foster, Rosert C. nies. g% of the Ferns and Flowering Plants ¥ oo (Publishe by the Gray Thain, 1-223 Pp. . Raup, Hucoh M. The Willows of ae — America. ith 28 maps. Wison, KeNNETH A. Sporangia 0 of the Fern Genera Allied with Polypodium an Vi ttaria. With 5 plates. Prrpur, Rosert E., Jr. The — Chromo- somes of Rudbeckia and Related Genera of the Compositae. Wi figures, 37 graphs. (Published by the tee Her- barium, 1-162 pp. 1960.) . THoMas, JoAB , ee Study of the Cyrillaceae ergs 45 5 figures (Published by the Gray Herbarium, 1-114 960. ) ‘ See Eprru, The Maidenhair Ferns (Adiantum) of Costa Rica. With 5 plates. Tryon, Rotra. A Review of = eng De a ee in America. With 8 _ plates. KrennetH A. The Leptosporangium of the New Jaey Fern Anarthropteris Dictyopterié: With 29 figures $ 1.50 2.00 3.50 3.00 TRYON, Observations on the Leaves of Pellaea Androl, "(published by the Gray Herbarium, 1-68 x 188. So.sric, Orro T. _Cytotaxonomic and Evolutionary Studies in the North American amg: s of Gutierrezia (Compositae). with. 16 figures. SoLpBri Oro T. The South American Sections of Erigeron and Their Relation to Celmisia. With 27 oi oF (Published by the Gray Herbarium, 1-86 pp Xiso. ADAMS, he ie a in the Guttiferae. I. A Synopsis of Hypericum Section Myriandra. With 3 figures. TRYON, Rotia. Taxonomic Fore Notes. II. Pityrogramma (includ- ne. Trismeria) and Anogramma. With 6 figures. (Pub- lished by the ie Herbarium, 1-76 pp. 1962.) \, 190. ies Wiis Jn. Patterned Ground and Vegetation n So heen ‘Bylot land, Motews st Territories, Canada. Vith 24 a 2 —. ( Published iby the Gray Herbarium, 1-111 pp }.191. Sorsric, Orro T. The South American Species of Erigeron. Wi th 69 figures. ScAMMAN, EDITH nus Eriosorus Costa Rica 5 figures. Tryon, Roya ‘axO- n ern Notes. III. With 13 figures on, ALIC mic : gures. T EF, A Monograph of the Fern Genus Jamesonia. With 4 plates, — 19 5 a (Published by A Gray Herbarium, 1-203 p 1962 192. — ne C. The Evolution and lg te of i a (Cruciferae). With 15 plates, 4 figures, 3 maps. woe Duncan M. The Taxono th "and Distribution of ed Zygophyllaceae of Baja California, Mexico, Wi Tes (Published by the Gray Herbarium, 1-135 pp. 193. Ax Srrt von Reis. A Taxonomic Study of the Genus Anadenanthera ‘With 11 plates, 2 maps. So. ees OTTo Infraspecific tion in the Gatienexia Sarothr Com- < ex itnote eee With 12 figures. (Published y the Gray Herbarium, 1-115 pp. 1964.) Kiss. Tryon, Rotta. The Ferns of Peru; Polypodiaceae (Denn- sta Heer to Oleandreae). With 196 Spee 46 maps. (Published by the Gray Herbarium, 1-253 pp. 1964.) 195. Lroyp, Davm G. Evolution of oF Sees and are: Differentiation in Leavenworthia (Cruciferae). SS th figures, 5 — Kuanna, Kutpire R. anp REED eas Lins. A Taxonomic Revision of o Siied sibs (evndiorss): wah 10 Sears (Published by the Gray Herbarium, 1-158 Yuse. pas Greorce W. = a of the Salix i Com- jlex in North America. With 58 figures, 7 maps. (Pub- ed by the Gray Fecheitech 1-142 pp. 196 3). 97. ai 48 Heures 4 — — American gy wae of — S rs of Crucif . “ Can ca Gareth oe With 3 3 rae (Published va e Gray ene a A ee ee oe Oe ee eres en 2.50 4.50 CONTRIBUTIONS FROM THE GRAY HERBARIUM OF HARVARD UNIVERSITY Edited by Reed C. Rollins and Robert C. Foster NO. CXCVII THE SOUTH AMERICAN SPECIES OF GUTIERREZIA By Orro T. SoLsric CHROMOSOME NUMBERS OF CRUCIFERAE By Reep C. RoLiins PUBLISHED BY THE GRAY HERBARIUM OF HARVARD UNIVERSITY CAMBRIDGE, MASS., U. S. A. IssueD OcrosBer 7, 1966 PRINTED BY THE SON PRINTING COMPANY CAMBRIDGE + MASSACHUSETTS * U.S.A. THE SOUTH AMERICAN SPECIES OF GUTIERREZIA Otro T. SoL_sric Gutierrezia is a small genus of 19 known species in the Com- positae-Astereae, Eight species are found in western North Ameri- ca from northern Mexico to southern Canada. The other eleven species grow in South America, from southern Bolivia to the Straits of Magellan. The genus is relatively uniform, particularly in its oral characteristics. Nevertheless, some variability is pres- ent in such characters as growth-habit, leaf-size and shape, di- mensions of capitula and number of flowers in a head. These fluctuations are most evident among the South American species, where they often can be correlated with the diverse habitats occupied by plants of the genus. This is the sixth paper in a series dealing with diverse aspects of the evolution and systematics of Gutierrezia. The principal objective has been to study the genus from various approaches, such as cytology, morphology, distribution and ecology, including detailed investigations of populations in nature. Previous articles have dealt with generic relationships, variability, cytotaxonomy, and evolution of the North American species (Solbrig 1960a, b, 1964, 1965) and with chromosome morphology (Riidenberg and Solbrig 1963). The present work covers these aspects for the South American species. A forthcoming work will summarize the evolutionary relationships of the species and the phytogeo- graphical implications of the disjunct distribution of Gutierrezia. ACKNOWLEDGEMENTS ago; Sr. Rodolfo Wagenknecht, Serena, and the late Dr. Celestino Castro, Antofagasta. The author is also very grateful to Dr. Teodoro Meyer of Tucuman, Ing. Agr. Armando Hunziker, of Cérdoba, and Sra. Rebeca Acevedo de Vargas, of Santiago de Chile, for allowing the use 4 OTTO T. SOLBRIG of the facilities and equipment of the herbaria in their custody. Last but not least, the author acknowledges his debt of gratitude to Dr. Ramon Ferreyra of Lima, Pert, and to Mr. Guido Pabst, of Rio de Janeiro, Brazil, for their help and generosity while the author visited their countries during these collecting tri This work would not have been possible without the financial as- sistance of a grant in aid of research from the National Science Foun- dation, which is gratefully acknowledged. am also indebted for able laboratory assistance in various phases of this work to Miss Susana Cabrera, Miss Marilyn Wasser, Miss Julia Barth, and Miss Karen Niecke. I have studied material from the following herbaria, and hereby express my appreciation to their curators for making this possible: British Museum of Natural History; Botanisches Museum, Berlin- Dahlem; Chicago Natural History Museum; Gray Herbarium and the Arnold Arboretum of Harvard University; Instituto Botanico Dar- winion, San Isidro; Instituto Miguel Lillo; Instituto de Botanica Agri- cola, Castelar; Missouri Botanical Garden; Museo Botanico, Cérdoba, Argentina; Museo de La Plata; Museo de Historia Natural, Santiago, Chile; New York Botanical Garden; Royal Botanical Gardens, Kew; Staatsinstitut fiir Allgemeine Botanik, Hamburg; United States Na- tional Museum; University of Utrecht; and University of California, Berkeley. In addition Dr. Adrian Ruiz Leal (RL), of Mendoza, Ar- gentina, sent to me on loan his personal herbarium of Gutierrezia for study, for which I am most grateful. History oF THE Genus The genus Gutierrezia with one species, G. linearifolia, was described in 1806 by Lagasca. The type material is unknown since no specimens named by Lagasca exist, and the original de- scription is so general that it fits several modern species. Lagasca indicated the original locality only as “Hab. in N. H.” (Habitat in Nova Hispania?). The name was subsequently used for at least three different species; G. resinosa and G. mandonii ssp. gilliesii from South America and G. bracteata from California. Hooker and Arnott (1830) described specimens collected in Chile during Beechey’s voyage as “Galinsoga? resinosa.” Mean- while Nuttall (1818), unaware of Lagasca’s paper, described the new genus Brachyris, for the North American species B. eutham- We 4a (, sarothrae). De Candolle (1836) was the first to study the totality of the material then known. He accepted Nuttall’s Brachyris, with seven species, five from North America, and two om South America: G. neaeana, which was based on material gathered by the Malaspina expedition, and G. paniculata (= G. THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 5 resinosa), which he described without reference to the work of Hooker and Arnott. In addition, De Candolle established the new genus Odontocarpha, where he once again redescribed G. resin- osa, this time as O. poeppigii. He placed Odontocarpha in the tribe Vernonieae, its clearly asteroid characters notwithstanding. Hooker and Arnott (1836) recognized that Gutierrezia and Brachyris were congeneric; Bentham and Hooker (1873) recog- nized the true identity of Odontocarpha. Meanwhile several species from South America had been added to all three genera by Remy, Schulz-Bipontinus, Philippi and Grisebach. No monographic study of Gutierrezia has been attempted since De Candolle. The genus has been treated twice for the flora of Chile: first by Remy (in Gay, 1846) and later by Reiche (1902). With the increase of botanical exploration in South America at the end of the last century, more taxa of Gutierrezia were added to the floras of Chile and Argentina by Philippi, Spegazzini, O. Kuntze, Reiche, Cabrera, and Rebeca Acevedo de Vargas. Never- theless, the areas where Gutierrezia grows are not fully explored, and it is likely that new species will be added to the genus in the future. MorRPHOLOGY AND ANATOMY The importance of a clear understanding of the morphological variation of taxa for taxonomic and evolutionary studies needs no emphasis. Likewise, comprehending the anatomical basis of mor- phological characters provides illuminating insights for the sys- tematist. Both these approaches have been used in the present investigation. In addition, chromosome number and pollen and stomata size have been investigated whenever feasible. HABIT: Six of the eight North American species are globose shrubs. This growth-habit also prevails among the austral mem- bers of the genus, where it is characteristic of nine of the eleven species. Nevertheless, there is far more variation in these than in the North American species. The other two South American Species are appressed cushion-plants or semi-creeping, habits of growth not found elsewhere in the genus. None of the South American species is annual. rae Gutierrezia mandonii ssp. gilliesii is in habit most similar to the North American species, particularly G. bracteata and to populations of large-sized plants of G. sarothrae. It is a shrub cm. tall, very much branched from the base, with narrow 6 OTTO T. SOLBRIG leaves and of globose or semi-globose form. Gutierrezia mandonii ssp. isernii is similar but less globose, and with more compact branching. Gutierrezia spathulata, G. ruiz-lealii, G. ameghinoi, and G. neaeana are shorter (10-25 cm. tall), more compact and globose, with wider leaves and with a loose basal rosette of leaves. Similar characters can be observed in G. mandonii ssp. mandonii of northern Argentina. Of the Chilean species, G. ga- yana is close in habit to G. mandonii ssp. gilliesii. On the other hand, G. resinosa is a large erect shrub with stems up to 5 cm. in diameter, and up to 1.50 m. tall. Gutierrezia taltalensis and G. espinosae ‘are close to G. resinosa in woodiness and other as- pects, but they are shorter and more compact in their habits of growth. Gutierrezia repens is a low, mat-forming or semi-creeping perennial herb, with woody, rhizomatous stems up to 1 mm. in diameter. Gutierrezia baccharoides is a true cushion plant that grows in the high Andes above 3,000 m. in Mendoza, and in pro- gressively lower altitudes farther south, It is found at sea-level at the Straits of Magellan. FOLIAR AND NODE CHARACTERS, Studies of the nodal anatomy of species of Gutierrezia always revealed a node with three lacunae and three traces. The traces depart from the stele at different levels, and are normal collateral bundles with exteriorly lying sclerenchymatous fibers (fig. 1). Foliar anatomy is also very uniform throughout the South American species. To be noted is a very thick cuticle covering an epidermis, formed by large cells, that envelops both foliar sur- faces. At regular intervals in the epidermis, there are narrow cavities, where one to several, multicellular, multiseriate resin- glands are found (fig. 10). The resin-glands can also be observed with the naked eye, and are very evident when the leaf is held against the light. They are found on both surfaces but are most abundant in the adaxial surface. Stomata are also to be found on both surfaces but in greater abundance on the abaxial side. The palisade parenchyma is formed of one to three layers depending upon the species, and the spongy parenchyma is normal. The collateral vascular bundles are surrounded by a sheath of col- lenchyma. The large bundles also include a large resin-canal, and bundle sheath extensions. Leaf-shape is fairly varied in the different species. Although striking to the eye, the changes involve only variations in length THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 7 Ry ane Say Ret EY = te RH re en! a5 ILO Aa Sees o, Sore CONNER U SN iettgay } KORY ro ay sa ses meaner: v.8 iiss A ae tip BA CPO eS. gOCK we! noel CO yy eo! fitecel 26 NG =, r iN tae! Fic. <7 Cameta 1 : rezia gayana (Solbrig 3070, Gu), showing some cellular detail, r.c.=resin canal; p tp ucida drawing of cross-section of stem and leaf of Gutter- 2 Pee photosynthetic parenchyma; p.f hloem fibers; c.b.=collateral bundles. and width. All species have entire leaves (or minutely toothed under the microscope), linear or lanceolate in shape, with no clearly marked petiole, and with a smooth glutinous surface ( figs. 2-18). The changes in shape parallel those in habit, with the more globose species having narrow leaves, that increase in width and decrease in length the more compact and appressed the habit of the species. The extreme condition is reached in G. baccharoides where the internodes are very short and the appressed leaves completely cover the stems. INFLORESCENCE. The capitula are borne on short pedicels in groups of 2-5 in most South American species. In G. resinosa, ea taltalensis, and G. espinosae, the heads are sessile, forming rather tight glomerules, while in G. repens, G. neaeana, G. ruiz-lealii, and G. baccharoides, the heads are solitary at the end of the branches. The capitula in general are small or medium-sized. The only exception is the capitulum of G. gayana which is 10-12 mm. in length and up to 15 mm. in width (fig. 22). The next largest OTTO T. SOLBRIG oo 1G. 2-18. Leaf and involucral bract due to sclerenchyma. Only major veins d - Gut 10-XI-1941, cu); 3. Leaf of G. resinosa (Solbrig 3044, cu GH); 8. Leaf of G S (Venturi 4117 5658, GH); 10. Resin gland in section ilar resin glands ar af cross-section of G baccharoides (S§ olbrig 3070 (Solbrig 30 GH); 13. Portion 1 veins drawn in; 14 af of G. ma Portion of same 1 mandont ssp. gilliesii with leaf o i d. majo aes im; 16. Cross-section of leaf of G. mandonii ssp. gilliesii; 17. Leaf of G. — mandonit (Venturi 4853, cu); 18. Leaf of G. mandonii ssp. isernii (Solbrig 3400, ‘ THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 9 capitulum is found in G. repens and G., ruiz-lealii which have heads up to 7 mm. in length and 8 mm, in width, that is, about one half the size of those of G. gayana (fig. 20). The smallest heads are those of G. mandonii, which are sometimes only 4.5 mm. high and 3 mm. wide. These are only half as large as those of G. repens (fig. 21). The other eight species have capitula that range in size between those of G. mandonii and those of G. re- pens. The shape of the head varies from species to species, rang- ing from turbinate to broadly campanulate. The shape is also correlated with the number of flowers: the larger the number of flowers, the more campanulate the capitulum. In general, the involucral bracts are fairly uniform. However, there are some slight changes from species to species in color, texture, and the degree to which they are appressed. These are of some secondary diagnostic value. FLOWERS. The flowers of Gutierrezia are quite uniform in all the species. There is an outer row of ligulate pistillate flowers, and a number of tubular, hermaphroditic disk flowers. The pap- pus is formed by 10 to 15 paleae, and the achene is pubescent in all species. The most significant variations are in the color of the flowers, in their absolute size, and in the relative size of their parts. Gutierrezia gayana is the only species with exclusively white flowers. Six other species have white-flowered populations and/or polymorphic populations. These are: G. mandonii ssp. gilliesii and G. ameghinoi, with a large number of entirely white-flowered populations, and G. mandonii ssp. isernii, G. resinosa, G. taltal- ensis and G. espinosae with some polymorphic populations con- taining both white- and yellow-flowered plants. There appears to be no obvious correlation between flower-color and habitat preference other than the fact that all high montane species are yellow-flowered. Length of the tubular corolla is about 3 to 5 mm., while the tube of the ligulate corolla is a little shorter. The ligule is from 3 to 7 mm. long, and from 1 to 3 mm. wide. Four species, G. gayana, G. repens, G. ruiz-lealii, and G. baccharoides, have slight- ly larger flowers and longer ligules than the rest. Gutierrezia resinosa has a short and narrow ligule. The length of the pappus is about 1.5 mm. in the tubular flow- ers and 1 mm. in the ligulate ones. Exceptions are G. gayana, 10 OTTO T. SOLBRIG , €erresia mandonit SSp. man lo» it . : a4 i (¢ ondoriac : . ‘ ‘ . itiey } 1doriaco, Coquimbo, ( hile): 23 Chile THE SOUTH AMERICAN SPECIES OF GUTIERREZIA ll where the pappus is usually less than 1 mm. long, and G. resin- osa, where it is 2.5-3.5 mm. long. The number of flowers in a head also varies from species to species. The largest number is found in G. gayana, where over 50 flowers are present: 15-25 ligulate and 30-40 tubular; G. repens and G. ruiz-lealii have 10 to 20 ligulate and 17-31 tubular flow- ers; G. baccharoides, G. ameghinoi and G. neaeana have more than 20 flowers per capitulum (7-10 ligulate and 10-15 tubular). All the other South American species, with the exception of G. resinosa, have 5-10 ligulate flowers and 8-12 tubular flowers; G. resinosa has only 4-7 ligulate and 5-12 tubular flowers per head (figs. 19-23). CHROMOSOME NUMBER AND POLLEN, Unfortunately very little is known about the chromosome number of the South American species of Gutierrezia. Altogether only eleven counts have been made (table 1), ten by the author and one by Schnack and TABLE 1. CHROMOSOME NUMBERS OF SOUTH AMERICAN SPECIES OF GUTIERREZIA Gutierrezia mandonii ssp. mandonii: n=12. Argentina, Jujuy, Mai- mara, pie del cerro Colorado, Solbrig 3325; Argentina, Jujuy, Tilcara, Quebrada del Chorro, Cabrera & Solbrig 16885; Argentina, Jujuy, 5 Km. N. of Volean, Cabrera & Solbrig 17023; Argentina, Jujuy, Volcan, Chilcayo, Cabrera & Solbrig 16838; Argentina, Catamarca, Tinogasta, Cuesta de Zapata, Cabrera & Solbrig 16749. Gutierrezia mandonii ssp. gilliesii: n=12. Argentina, Mendoza, Papa- gallos, Schnack & Covas (no voucher extant). Gutierrezia mandonii ssp. gilliesii n=20. Argentina, Cordoba,, San Pedro Norte, Solbrig 3402. Gutierrezia ruiz-lealii: n=12. Argentina, Jujuy, cerro Horqueta near Volcan, 3,500 m., Cabrera & Solbrig 16995. Gutierrezia gayana: n=16. Chile, Coquimbo, Cuesta Vineta near Condoriaco, Solbrig 3055. Gutierrezia resinosa: n=28, Chile, Coquimbo, Rivadavia, Solbrig 3384; Chile, Coquimbo, Estancia Huaynay, 45 K. S. of Coquimbo, Solbrig 3044, Covas. This last count was reported as Gutierrezia sp. and no voucher is extant. However, Ing. Agr. Schnack led the author to the exact place where he gathered his cytological material (Papagallos, Prov. Mendoza). The only species growing there and in a radius of 50 miles is G. mandonii ssp. gilliesii. Since the number of n—12 was found in other populations of 12 OTTO T. SOLBRIG G. mandonii (but not ssp. gilliesii), it appears very likely that Covas and Schnack studied G. mandonii ssp. gilliesii. From the available counts it appears that the South American species of Gutierrezia are based on x=4 (as are all the North American species studied) and that they are high polyploids. More counts are needed to confirm this hypothesis. In order to obtain a possible insight into the level of ploidy of the chromosomally unknown species, the diameter of the pol- len was measured in 25 samples belonging to nine of the 11 species. Few size-differences were found and in general the pollen-diameter did not correlate with known differences in chro- mosome ploidy levels. Consequently, pollen-diameter is not a reli- able indicator of ploidy level in this group of species. DIsTRIBUTION AND ECOLOGY Gutierrezia mandonii is the most widely distributed of all South American species of the genus. It is found throughout the region extending from southern Bolivia (Depts. Tarija and Potosi) along the eastern foot-hills of the Andes and the Pam- pean formation to Mendoza and Cérdoba. It also grows in the Sierra de la Ventana in the province of Buenos Aires, and sporad- ically in south-central La Pampa, and in the southernmost part of the province of Buenos Aires (fig. 24). Gutierrezia repens is endemic to the Aconquija Mts. in prov. Tucuman, Argentina, and adjacent areas of Catamarca and Salta, and is found between 2,800 m. and 4,000 m. (fig. 25); G. ruiz-lealii is also found at altitudes over 3,000 m. in the mts. of Jujuy and Salta (fig. 26). Gutierrezia ameghinoi grows in eastern and northern Patagonia, western Rio Negro, Neuquén, and southern Mendoza in Argen- tina, at elevations ranging from sea level to 2,000 m. (fig. 25); G. spathulata is found on the western slopes of the Andes of southern and central Mendoza between 1,500 and 3,500 m. (fig. 27 ). Gutierrezia baccharoides is a species of alpine to subantarc- tic environments, It grows from the high Cordillera of Mendoza, where it is found at altitudes ranging from 3,000 to 4,000 m., to the Straits of Magellan, where it grows at sea level (fig. 24). Finally, five species are found in Chile: G. resinosa, in the dry mediterranean belt between Santiago and Copiapé at low eleva- tions (fig. 25); G. gayana in the dry, pre-Cordilleran “pampas,” north and east of Serena (fig. 26); G. taltalensis and G. espinosae THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 13 that grow in the extreme desert area of northern Chile, the former in the vicinity of Taltal, the second extending from Taltal to Antofagasta (figs. 24, 27); and G. neaeana known only from the type collection (locality unknown) and from one other collection from the high Cordillera of central Chile (fig. 27). Considering the entire area, the various species of Gutierrezia grow in diverse environmental conditions. Unfortunately, prob- lems of transportation and distance made impossible detailed analysis of soils and other ecological factors, as were made for the North American species. Nevertheless, field-observations and data taken from the literature allow a preliminary evaluation of the ecological situation. sor. Analysis of 19 soil samples supporting growth of North American species (Solbrig, 1960a ) failed to reveal any differences in edaphic requirements between the species, but provided some indication of the over-all requirements of the genus as a whole. The North American species of Gutierrezia seem to grow on loose, sandy, alkaline or neutral soils, and apparently those with also a low content of organic matter and nitrates. Field-observa- tions tend to confirm this set of conditions for the South American species also, at least for seven of the eleven species, which are the ones studied in the field. In addition many of the South American species are found growing among rocks in mountainous areas, and in at least one instance plants were seen growing in scree on a rather steep slope. No South American species grows in a special edaphic situation, such as is the case with the North American species that grow on serpentine. Nevertheless, in cer- tain cases (Gutierrezia mandonii ssp. gilliesii in Cordoba ) Gutier- rezia appears to grow on limestone outcrops. BLOOMING PERIOD. Species of Gutierrezia bloom principally in late summer and fall. All North American species of the genus show this blooming regime. Rainfall affects blooming time to a certain extent, even within a species (Solbrig, 1964). Observa- tions in the greenhouse, where water was always plentiful, indi- cate that blooming is a photoperiodic response in the North American species, since under cultivation they also bloomed in late summer and fall. Of the eleven South American species, only one, G. ruiz-lealii, appears to bloom exclusively in late sum- mer and fall. However, this might be only an artifact resulting from the very few observations available on this taxon. Four species, G. repens, G. spathulata, G. baccharoides and G. man- 14 OTTO T. SOLBRIG re is wel = eo a = 4 Be Pig Vay 0 REPENS O RESINOSA * AMEGHINO! @ 2 SSP. MANDONII @ SSP. GILLESII © SSRISERNI + wees & 3B Vy WS... TALTALENSIS © 2D = es lew ie 1a aes —— ‘ a A ESPINOSAE © NEAEANA 1% SPATHULATA @ RUIZ-LEALIT © GAYANA * \ BACCHAROIDES @ \ x | aA) iw, Lie 7 yed pr sep | | = { Elle. w ee 3 Fic. 24-27, Distribution of the South American species of Gutierrezia. Each dot ss ag ong a locality; several collections might be wn from each locality. Goode ase Map, Copyrighted by University of Chicago, THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 15 donii, bloom throughout the summer and early fall, approximate- ly from November-December to May. These species grow in areas with summer rains. Populations of G. mandonii ssp. gilliesii from Cordoba, an area that has summer droughts, usually bloom later, from March to July. On the other hand, the four Chilean species and G. ameghinoi, from Argentine Patagonia, bloom in late spring and early summer, roughly from October to February. This is, no doubt, an adaption to the spring rains of the region where these species grow. Gutierrezia gayana is the earliest bloomer of the whole genus, flowering only from September to November. Field-observations further show that sufficient rainfall is needed before blooming will begin in any of the South Ameri- can species. In very dry years, when rainfall is insufficient, the plants will not bloom at all. Therefore, it appears that in some of the South American species, blooming is controlled by moisture and not photoperiod. OTHER OBSERVATIONS. North American species of Gutierrezia spread into range lands when these are overgrazed, This is ap- parently due to a disruption of the dynamic balance between grasses and Gutierrezia in the competition for water and nutrient matter in the soil. Normally the grasses are able to crowd Gutierrezia out from the better soils, the reverse being true in the poorer ones (Solbrig, 1960a). When land areas are overgrazed, Gutierrezia can move into the edaphically better places. A similar situation holds true for at least two South American species, G. resinosa and G. mandonii ssp. gilliesii. This is most notable in the region south of Serena, Chile, where extensive overgrazing in open ranges, particularly by goats, has led to a depauperization of the grassland vegetation and the spread of G. resinosa over hundreds of acres (figs. 28-30). Fenced, well managed pastures were devoid of Gutierrezia plants. The plants are not palatable to cattle, sheep, or goats on account of their high resin-content. There are unconfirmed reports that the plant when ingested in large amounts is poisonous. The North American species form dense populations of 50 to 1,000 plants. These populations are often separated from each other by considerable distances. This type of distribution is also found in most of the South American species, notably G. man- donii ssp. gilliesii and G. resinosa. On the other hand, the moun- tain species are less likely to form large populations, and this is particularly true for the two species from northern Chile, G. 16 OTTO T. SOLBRIG Natural populations of Gutierrezia. 28. Natural population of G. gayana IG. -30. b Condi, ae mbo, Chile. The species is distributed locally over several acres, u d Plants are fairly separated from each other; 29. adside population of G- mandonit ssp. gilliesii, near Los Cocos, Cordoba, Argentina. “a this area G. mandonit THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 17 espinosae and G. taltalensis. These two species are found in the “quebradas” along the extremely dry coast between Taltal and Antofagasta. Rainfall in this region is, on the average, only 1 mm. per year with many years when absolutely no rainfall occurs. Plants obtain needed moisture from fog condensation in these steep, sea-facing canyons. As a result, the vegetation is very sparse. Bushes of Gutierrezia are very rare, often only one in a locality. BREEDING SYSTEMS AND POLLINATORS. The North American per- ennial species are self-compatible, but normally outbreeding. This is due to the long exserted styles that make selfing mechan- ically improbable. The annual species on the other hand have shorter styles and are normally self-pollinated. Unfortunately, no systematic study of the South American species could be made because most species could not be brought into cultivation. The two species that were grown, G. resinosa and G. mandonii ssp. gilliesii, were self-compatible. However, as is the case with the North American species, they were not normally selfed on ac- count of the long exserted styles. Field-observations and herbari- um studies show a similar exsertion of the style in all the South American species. It is possible that they all might be self-com- patible, but in all probability they are normally outbreeding. Species of Gutierrezia are pollinated by a variety of different insects, particularly small hymenoptera, diptera, and coleoptera. This is true both for the North American species (Solbrig 1960a, 1964) and for those of South America. SYSTEMATIC TREATMENT Gutierrezia is both a polymorphic and polytypic genus. Conse- quently the precise delimitation of allopatric taxa is difficult, particularly when there is a lack of biosystematic information. The phenotypic plasticity of the plants complicates matters fur- ther. The criteria adopted in the delimitation of taxa are what is usually labelled as “conservative.” An effort has been made to delimit natural groups stressing similarities within the taxon rath- er than differences. Even so, in some cases, certain herbarium specimens—especially incomplete and scrappy ones—do not fall easily in the groups recognized. Future field and experimental work are needed to resolve these problems. Another difficulty is the imprecise floristic knowledge of many forms dense stands, that occupy at most a few hundred square feet at each — 30. Population of G. resinosa south of Serena, Chile. Due to overgrazing of — plants, G. resinosa has spread over hundred of acres forming rather dense stands. 18 OTTO T. SOLBRIG of the areas sustaining Gutierrezia, especially certain areas of Chile. From experience, we know that only in time will the need- ed collections be assembled. Gutierrezia Lag., Nov. Gen. et Spec. 30. 1816 Brachyris Nutt., Gen, 2: 163, 1818. Brachyachyris Spreng., Syst. 3: 574. 1825. Hemiachyris DC., Prodr. 5: 313. 1836. Odontocarpha DC., Prodr. 5: 71. 1836. Suffruticose, erect, creeping or cushion-forming, glabrous, per- ennial shrubs. Leaves petiolate or sessile, alternate, subcoria- ceous, linear-lanceolate to lanceolate, entire. Heads in corymbose or paniculate inflorescences, in clusters, or solitary at the ends of branchlets. Capitula turbinate or campanulate, the involucral bracts in two to many rows, imbricated. Flowers yellow or white, the ray flowers pistillate, ligulate, the disk flowers tubular, her- maphroditic; corolla with a very short tube and narrow throat, five-lobed; stamens 5; styles of tubular flowers with long branches very much exserted at maturity, long stigmatic hairs present on the upper part and stigmatic papillae on the lower part of the stigmatic branches; styles of the ligulate flowers only papillate; pappus of short to medium-long squamellae, shorter in the ligu- late flowers; achenes turbinate, slightly flattened, tomentose, with hairs arranged in loose rows. Type species, Gutierrezia linearifolia Lag. (No specimens iden- tified by Lagasca have been found ). KEY TO THE SECTIONS A. Plants perennial (shrubs). B. Involucre more than 10 mm. in height and width. Flowers white. Plants from South America. section Megaloce section Gutierrezid. A. Plants annual. Flowers always yellow. Plants from North America. section Hemiachyris. KEY TO THE SPECIES ange or mat-forming, or low creeping perennials, less than 10 THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 19 B. Small, compact, — or mat- éguaitiee plants. Leaves a iy 0 mm. long or .... 6. G. baccharoides. B. Low, creeping, ie ues s plants ‘Leaves 30 mm. or haine . repens. A. Erect to globose shrubs, more than 10 em. tall ( generally 20-80 cm, ta C. Involucres more than 10 mm. in height and width; Some white . gayana. C. nvelieces — than 81 mm. in n height oan wilt: git yellow or white. D. Involucres narrowly turbinate 6-8 mm. in height, 3-5 mm. in width; erect shrubs with woody bases (plants of Chile). E. Ligules 1. S mm. long or shorter; leaves narrow, 1.5 mm. wide or narrower; narrow, open shrubs, 50-150 cm. tall G. resinosa. E. Ligule 2. 5S: Sn mm. n. long; leaves broader, particularly the basal ones, 2 mm. wide or more; compact shrubs 25—75 cm. tall. F. Leaves linear, more than 50 mm. long; stems et oth Bre ALES. seeds ah ere eee aie eee taltalensis. F. Leaves lanceolate, less than 45 mm. long ee less than mm.); stems with pee elevated leaf scars ‘ G. espinosae. D. Involucres turbinates: “4-6 m high, o + campanulate and then 7 mm. high (Plants of Ataadhting a Bolivia. ) G. Leaves ei 2 mm. wide or less; involucres turbinate, 2-4.5 mm. wi 1. G. mandonii. H. Plants 50-100 | cm. tall: leaves | narrow, “hy mm. wide; flowers ite or occasionally yellow .. lb. G. mandonii ssp. gilliesii. H. plants less than 50 cm. tall; leaves 1-2 mm. wide; flowers ellow. I. Leaves short, mostly scale-like, 20 mm. long or less; invo- lucre less than 4 mm. in ater long; in nvolucre 4-6 mm. high .. 1c. G. mandonii ssp. isernii. G. Leaves lanceolate, shikibellnte or spathulate (at least the basal ones), 3 mm. wide or more; involucres campanulate, mm. wide. J. Leaves oblanceolate; capitula solitary at the end of naked or sparsely leafed scape-like branches, 3-10 cm. om omits et her 5 . neaeana. K. betes ag Raton as wide as high, 5 mm. or more in width; ligulate flowers 8-10; tubular flowers 10-15. L. Ligules 5 mm. phe or more, yellow; pares 7-10 mm. broad .... G. ruiz-lealii. L. acpi 3 mm. or less in n length, white (or yellow) involu- e 5-7 mm. wide . meghinoi. 20 OTTO T. SOLBRIG K. Involucre turbinate, higher than broad, 4 mm. or less in width; ligulate flowers 5-7; tubular flowers 8-12 5. G. spathulata. SECTION GUTIERREZIA 1. Gutierrezia mandonii (Sch. Bip.) Solbrig, comb. nov. Brachyris mandonii Sch. Bip., Linnaea 34: 354, 1865. Small to medium-sized shrub, 25-100 cm. tall, from a strong, usually fascicled, root. Caudex strong, woody, up to 2 cm. in diameter, with a thick, usually dark bark, profusely branching to the top. Branches erect, angular, leafy to the end, light gray or green in color, up to 50 cm. long. Leaves linear, with an acute apex, sessile, often somewhat attenuate towards the base, 5-30 mm. long, 1-2 mm. broad, glabrous, resinous-punctate. Heads borne at the ends of the branchlets on short pedicels 1-3 mm. long. Involucre turbinate, 4.5-6.0 mm. long, 3.0-4.5 mm. wide; involucral bracts 15 or less, 4-5 mm. long, 1-1.5 mm. wide, gla- brous, yellow to yellow-green, with an acute tip, often squarrose. Ligulate flowers 7-12, 2.5-3.5 mm, long, yellow or more frequent- ly white, ligule 2.5-3.5 mm. long, 1.2-1.8 mm. wide; tubular flowers yellow or white, 1.5-2.5 mm. long. Pappus of scales, 0.5- 1.5 mm. long in the tubular flowers, 0.5-1.2 mm. long in the ligu- late flowers. Achenes 1-2 mm. wide, densely pubescent. TYPE: Bolivia, prov. Larecaja, “vicinis Sorata,” 2,680 m., Man- don 228 (Isotype cu! us! F! Ny!). DISTRIBUTION: From southern Bolivia to northern Patagonia, and from the foot of the Andes to Tucuman, Cérdoba, Sierra de la Ventana (Buenos Aires) and the Atlantic Ocean. la. Gutierrezia mandonii ssp. mandonii Gutierrezia gilliesii Gris. var. scabriuscula Gris., Goett. Abh. 19:173. 1874; based on Catamarca, unter Nacimientos; Lorentz 434 (Isotype coro! ) This subspecies is characterized by short leaves, yellow flow- ers, small campanulate capitula with appressed involucral bracts, and the small globose habit, rarely exceeding 25 cm. tall. The four chromosome counts available indicate that this subspecies is a hexaploid (n = 12). DISTRIBUTION: Southern Bolivia to Tucuman, Argentina. It is found in the valleys (“quebradas”) and mountains of the foot of THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 21 the Bolivian plateau, the Aconquija Mts., and other mountains of the Pampean formation of northern Argentina, between 1,000 and 4,000 m. SELECTED SPECIMENS: ARGENTINA. PROV. JUJUY. Dept. Yavi: La Quiaca a Villazon, 3,442 m., Schreiter 10900 (Lu); La Quiaca, 3,450 m., Meyer 33021 (uu), Fidel Zelada (um), Parodi 9623 (cH). Dept. Cochinoca: Abra Pampa, 3,700 m., Venturi 9380 (us); Cerro Incahuasi, 4,500 m., Venturi 10127 (us). Dept. Hu- mahuaca: Tres Cruces, 3,700 m., Fabris & Marchioni 1801 (us); 20 Km. W. of Humahuaca, 3,600 m., West 6300 (cH, Mo); Mina Aguilar, 4,100 m., Sleumer 3375 (Lu), Petersen & Hjerting 101 (11); Laguna Colorada, 4,000 m., Budin I (iL). Dept. Tilcara: Quebrada del Chorro, 2,500 m., Solbrig 3329 (cH), Cabrera & Solbrig 16885 (cu, Lp), 17023 (cH, Le), Cabrera 7692 (cu, F, LP); Tilcara, 2,600 m., Venturi 4853 (GH, A, F, US), Balls 5951 (us), Venturi 6221 (us), 6222 (us), Pereyra 19 (us), Meyer 33027 (cu, tu), Cabrera, Fabris ¢> Marchioni 13288 (us, te), Schreiter 10040 (1u.). Dept. Purmamarca: Mai- mara, 2,500 m., Solbrig 3325 (cH). Dept. Tumbaya: Vol- can, 2,200 m., Venturi 3411 (us), Cabrera & Solbrig 16838 (cu, tp); 5 Km. N. of Volcan, 2,200 m., Solbrig 3335 (cH), Cabrera & Solbrig 17023 (cu, LP). PRov. saLta. Dept. Candelaria: Cerro del Chorrillo, 1700 m., Venturi 3846 (a, us). Dept. Santa Victoria: Lizoite, 3340 m., Meyer & Bianchi 33033 (GH, LL). Dept. Poma: Incahuasi, Cabrera 8218 (GH, F, LP). Dept. Rosario de Lerma, Puerta Tastil, 2,800 m., Venturi 8156 (GH, A, Us, MO, LiL). Dept. Cachi, Cachi Pampa, 3,400 m., Romero s.n. (ui). Prov. CATAMARCA. Dept. Belen: Cuesta de Zapata, 2,000 m., Cabrera & Solbrig 16749 (cu, Lp). Dept. Poman: Cuesta de Poman, Schickendantz 118 (us). Dept. Capayan: Santo Domingo, Peirano s.n. (cH, LiL). Dept. Andalgala, El Suncho, Jorgensen 1508 (us, GH, MO). PROV. LA RIOJA. Dept. Chilecito, camino a La Mejicana, 2250 m., Parodi 8061 (cH). Dept. Paganzo: Sierra de Paganzo, Hunziker, Cocucci d> Subils 15859 (corD). PROV. TU- cuMAN. Dept. Barruyacu: Agua Negra, 1400 m., Peirano (GH, A, um). Dept. Trancas: Colalao del Valle, 2800 m., Schreiter 5673 (A, tL). Dept. Tafi: Tafi del Valle, 2000 m., J. Hunziker 7160 (cH, BAB), Lillo 5050 (A, LIL). ee BoLiviA. Dept. Tarija: Puna Patanca, 3800 m., Fiebrig 2910 (GH, A). Dept. Potosi: Uyuni, Asplund 4952 (us); Miraflores, 3,600 m., Cardenas 4319 (us); west of Chocaya, 3700 m., West a2 OTTO T. SOLBRIG 6094 (cH, MO); Quechisla, Cardenas 53 (cH); near Villazon, 3400 m., West 8225 (cH, Mo): Tupiza, Fiebrig 3106 (¥, cH, vs, HBG ). lb. Gutierrezia mandonii ssp. gilliesii (Gris.) Solbrig, stat. nov. Gutierrezia gilliesii Gris., Goett. Abh. 19: 173. 1874; G. spathu- lata var. gilliesii (Gris.) O. Ktze., Rev. Gen. PI. 3(2): 156. 1898. Gutierrezia leucantha Cabrera, Rev. Museo de La Plata n.s., Bot. 4: 61. 1941. Based on “Argentina, Buenos Aires, Sierra de la Ventana, entre peas, A. L. Cabrera 4475, 23-V-1938" (Holo- type LP!) This subspecies is characterized by narrow (1 mm.) and long (4-6 cm.) leaves, white flowers (occasionally yellow), turbinate capitula, involucral bracts with green swollen tips, and the large shrubby habit. One population from Cérdoba known cytological- : is a decaploid (n=20), and one from Mendoza is a hexaploid nz12), Type: Argentina. Cordoba, “in collibus ab urbe occidentalibus, in montibus inter S. Pedro et Horcosuni,” Lorentz 183 VI-1871. (Isotype corp! xr!, photo cx!). Distribution: Central Argentina, from Tucuman to San Luis and Cordoba, in low mountainous areas (less than 1000 m.); Sierra de la Ventana (Prov. Bs. Aires); and La Pampa. SELECTED SPECIMENS, ARGENTINA. PROV. MENDOZA. Dept. Las Heras: Villavicencio, 1700 m., Ruiz Leal 8043 (ni), 1040 (RL), 4361 (RL), 8645 (RL), Burkart, Troncoso & Nicora 14385 (GH, us, st), O'Donell 1330 (a, um); Papagayos, Ruiz Leal 4493 (RL); Dept. Lujan: Cerro de Las Cabras, F. A. Roig (nL); Arroyo de la Manga, F. A. Roig (ru). Dept. Tupungato: Cacheuta, Sol- brig 3396 (cH); Potrerillos, road to Pampa de La Polcura, 2,000 m., Solbrig 3398 (cu). Dept. San Rafael: San Rafael a Malal-hue, Lourteig 1020 (us, LL); Monte Colman, Ruiz Leal 21724 (RL). PROV. SAN JUAN. Dept. Ullun: Bajada Cuesta Vieja, Hosseus 2600 (corp); cumbre Cuesta Nueva, Hosseus 2540 (corp); desde Ullun hasta La Angostura, 1800 m., Hosseus 2517 (corp). PROV. SAN Luts. Dept. San Antonio: Quebrada de los Bueyes, Galander Sn. (CoRD); Sierra de las Quijadas, Hunziker & Cocucci 16380 (corp); Bajo del Jume, Guinazt: 8 (US). PROV. LA PAMPA. Dept. Lihuel Calel: cerro Lihuel Calel, Burkart 15960 (st); Sierra de THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 23 31-32. Gutierrezia mandonit. 31. G. Argentina; note the dense, globose, branching system; 32. G. ™. SSp- i Ct ehe a Polcura, Mendoza, Argentina, app. 3,000 m. altitude; note the flatter growth, and t . much less developed branch system, probably an adaptation to he high altitude and winter snow. G. m. ssp. gilliesti near Los Cocos, Cordoba, 2. G. n ’ . } 1 ssp. isernit cerro de a+ 24 OTTO T. SOLBRIG Lihuel Calel, Schwabe & Fabris 2002 (Le); Anzoategui, O'Don- ell 1767 (1.); Rio Colorado, O’Donell 1628 (LiL). PROV. COR- posa. Dept. Santa Maria: Malagueno, Hunziker 6658 (corp, LIL, A, GH, us, MO), O'Donnell & Rodriguez 323 (LiL, F). Dept. Puni- lla: Sierra Chica, Lossen 223 (cu); Los Cocos, Cabrera 6390 (LP, GH, US, F). Dep. Ischillin: Ongamira, Solbrig 3403 (cH); San Pedro Norte, Solbrig 3402 (cH). Dept. Pocho: Sierra de Pocho, faldeo E., Hunziker 13643 (corp). Dept. Sobremonte: cerca de La Plaza, Hunziker 12137 (corp); Sierra de Copacabana, faldeos NE, Hunziker 14801 (corp). PROV. BUENOS AIRES. Partido Torn- quist: valley south of Pico Ventana, Bartlett 20036 (cH); Sierra Ventana, Lorentz s.n. (cH, us). Partido de Patagones: Carmen de Patagones, Krapovickas 2012 (11). lc. Gutierrezia mandonii ssp. isernii (Phil.) Solbrig, stat. nov. (fig. 32) Brachyris isernii Phil., Anal. Univ. Chile 27: 337. 1865. Gutierrezia isernii (Phil.) Phil. Anal. Univ. Chile 87: 427. 1894. Subspecies isernii is characterized by stiff branches and heads borne solitarily or in groups of 2 or 3 at the ends of the branches. This subspecies is closely related to ssp. gilliesii, ssp. isernii be- ing smaller (15-30 cm.) than ssp. gilliesii and with wider leaves (1-2 mm.). From ssp. mandonii it can be separated by the longer leaves and larger capitula (4.5-6.0 mm. long, 3-4.5 mm. wide). The flowers are usually yellow in ssp. isernii, but populations with yellow and white flowered plants are known. No cytological information is available on this subspecies. TYPE: Argentina, “inter St. Rosa et Mendoza,” Philippii 1868/69. (SANT!, Isotype tp! ). DISTRIBUTION: Argentina, Mendoza, in the precordillera be- tween 2,000 and 3,000 m. SELECTED SPECIMENS: ARGENTINA. PROV. MENDOZA. Dept. Tunu- yan, Corte Amarillo, Ruiz Leal 2991 (rt), 3020 (x); en la pre- cordillera a 2,500 m., Ruiz Leal 1125 (ri); cerro de la Polcura, Solbrig 3400 (cx); Los Arboles, Ruiz Leal 1699 (RL); Quebrada del Arroyo Grande, Ruiz Leal 13125 (ni); Cuchilla de los Ala- zanes, Ruiz Leal 2020 (nx). Dept. Las Heras, Cerro de la Cal, Carette s.n. (RL); Pampa Seca, 2,400 m., Ruiz Leal 5133 (RL); en lomadas, 2,600 m., Semper s.n. (RL) THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 25 2. Gutierrezia repens Griseb., Goett. Abh. 19: 173. 1874 Rhizomatous, low, slightly creeping, woody, perennial, up to 10 cm. tall, forming mats 10-30 cm. in diameter, branching from the base. Rhizome 5-10 cm. long, 3-8 mm. in diameter, woody. Branches short, 5-15 cm. long, 2-5 mm. in diameter, black or dark gray in color, woody, densely leafy to the top; in older plants only the dried-up petiole-bases remaining. Leaves lanceo- late to slightly oblanceolate, 15-35 mm. long, 2-4 mm. wide, acute, petioled, glabrous, resinous-punctate. Capitula borne singly at the ends of leafy flowering branches, that may be undivided or branched, and then with a head at the end of each branchlet. Involucre campanulate, 5-8 mm. long, 7-8 mm. wide; involucral bracts 20 or less, 4-8 mm. long, 2-3.5 mm. wide, oblong, with a broadly acute or obtuse apex, glabrous, resinous, coriaceous, usually yellow-brown in color. Ligulate flowers 10-20, yellow, tube 2-3.5 mm. long, ligule 3.5-7 mm. long, 1.5-3 mm. broad; tubular flowers 15-30, 3-6 mm. long; pappus of scales 1-1.8 mm. long; achenes 1-2 mm. long, 0.5 mm, wide, pubescent. TYPE: Argentina. Tucuman, “in jugo montis supra Cienaga,” 30-III-1872, P. G. Lorentz 150 (Isotype corn! ). DISTRIBUTION: This species is endemic to the mountains of Tucuman and adjacent areas of Salta and Catamarca in Argen- tina, at altitudes between 2,800 and 4,000 m. SELECTED SPECIMENS: ARGENTINA. PROV. CATAMARCA. Dept. An- dalgala: Cerro del Medio, 3,700 m., Jorgensen 1366 (cH, UC, LIL, SI, US, MO); valle del Rio Bolson, 4,000 m., Rohmeder (.1L); Capillitas, Schickendantz 96 (corp). prov. TUCUMAN. Dept. Tafi: Cumbre del Chorro, 3,300 m., Venturi 4117 (A, LIL, st, us); Pa- bellon, Castillon 36 (im); Tafi del Valle, Castillon 620a (LIL); Real del Derrumbado, 3,800 b., Venturi 4253 (xu); Infiernillo, arriba del rancho de Felipe Diaz, 3,500 m., Sleumer 1989 (LL); Cerros Calchaquies, Castillon 2776 (1); Cumbre del Malamala, 3,300 m., Lillo 3471 (1m), 2722 (im), 3453 (m.); La Cienaga, 2,500-3,300 m., Schreiter 900 (Lu), Descole 1614 (u1L), Sleumer 292 (111), Lillo 3605 (11.), 3721 (x1L), 3605 (Lu); Lara, 3,200 m., Lillo s.n. (uu); Cumbre del Lara, 4,000 m., Schreiter 1338 (x1); Cerro Mufioz, 3,000 m., Monetti 283 (LL), Schreiter 1330 (A, LL); Cuesta entre Tafi y Malamala, 2,800 m., Lillo 4348 (tu.); Chasquivil, Potrero Grande, Lillo 1924 (um); Loma del 26 OTTO T. SOLBRIG Diablo, 2,900 m., Rodriguez 457 (a, uit, st), Lillo 11312 (1m). Dept. Chicligasta: La Cascada, Meyer 14129 (iL); Est. Las Pavas, Puesto El Bayo, 3,300 m., Venturi 3113 (A, LmL, sI, Us), 4651 (im.); Est. Santa Rosa, 3,500 m., Meyer 15086 (LiL). Dept. Trancas, Abra de la Quenoa, 2,900 m., Bellanio 283 (1m); La Cascada a Las Cuevas, 3,000 m., Meyer 14888 (LIL). PROV. SALTA. Dept. Trancas: Camino Cuesta del Arca, Trancas, C. Spegazzini (Les 16824). 3. Gutierrezia ruiz-lealii Solbrig, spec. nov. Suffrutex compactus, erectus, 10-25 cm. altus; caulibus parce ramosis, glabris, 25 cm. longis. Folia lanceolata 15-30 mm. longa, 2-4 mm. lata, sessilia, margine integerrima, glandulosa. Capitula ad apicem ramulorum solitaria, vel 2-3, sessilia vel brevissime pedunculata. Involucrum hemisphaericum, 5-7 mm. altum, 6-8 mm. crassum; bracteis 15-20, pluriseriatis, subaequilongis, latis- sime triangulatis, acutis, 2-3 mm. latis, 5-7 mm. longis; flores mar- ginales 10-14, uniseriales, feminei, ligulati, lutei, tubulo 1.5-2.5 mm. longo, ligula 5-7 mm. longa, 1.5-3 mm. lata; flores disci 12-20, lutei, hermaphroditi, corolla tubulosa 2-4 mm. longa, quinque- dentata; squamae papposae aequilongae, 2-3 mm. longae; achae- nia pubescentia 1-3 mm. longa, 0.5-1 mm. lata. Compact, globose subshrub, 10-25 cm. high, from a strong, woody taproot, branching from the base. Root-crown 1-3 cm. in diameter, woody, with dark, grey, thick, fissured bark, usually dichotomously or trichotomously branched. Branches arising in great numbers from the root-crown, up to 25 cm. long, with a woody, thick, grey base, 3-5 cm. long, 5-10 mm. thick, leafless, abruptly changing into the green, semi-woody stems, 1-3 mm. thick, usually undivided, leafy throughout, internodes 5-15 mm. long. Leaves lanceolate, 15-30 mm. long, 2-4 mm. wide, acute, glabrous, glandular-punctate, sessile. Capitula borne at the ends of the branches or seldom in groups of two. Involucre campanu- late 5-7 mm. high, 6-8 mm. wide; involucral bracts 15-20, broad- ly triangular, 2-3 mm. wide, 5-7 mm. long, green, with a well- marked triangular tip of darker color. Ligulate flowers 10-14, yellow, tube 1.5-2.5 mm. long, ligule 5-7 mm. long, 1.5-3 mm. wide; tubular flowers 12-20, 2-4 mm. long; pappus of scales 2-4 oo achenes 1-3 mm. long, 0.5-1 mm. wide, pubescent. FIG. THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 27 : rear Volcan, Jujuy, Argentina, Fic. 33-3 Gutierrezia ruiz-leali, on cerro Horqueta, near nha —— E Note the Dr S, at app. a m. Aika 33. View of plant, 34. Close-up view WB chcdue habit, “oa head and buds 28 OTTO T. SOLBRIG tyre: Argentina. Prov. Jujuy, Dept. Tumbaya: cerro Horqueta, cerca de Volcan, 3,500 m., A. L. Cabrera and O. T. Solbrig 16695 (Holotype cx!, Isotype Lp! us! si! ). DISTRIBUTION: Known so far only from five localities, all above 3,000 m., in Jujuy and Salta, on relatively humid mountain tops. MATERIAL STUDIED: ARGENTINA. PROV. SALTA. Dept. Santa Victo- ria: Santa Victoria, 3,340 m., Meyer and Bianchi (iu.); Cuesta del Obispo, 2,500-3,000 m., Meyer 12068 (11), 12072 (1m), 12071 (LiL). Dept. Caldera, Cuesta del Carancho, arriba del Potrero del Castillo, ca. 3,200 m., Sleumer and Vervoost 2890 (LiL, Us); Dept. San Antonio de los Cobres, camino a Mina Concordia, Cabrera y Schwabe 56 (Bas). Gutierrezia ruiz-lealii is a very distinct species on account of the large heads, the long yellow ligules and the broad leaves (for a Gutierrezia). It is closest to G. repens, from which it differs principally in the globose, cushion-like habit of growth, and the smaller leaves. It gives me great pleasure to dedicate this species to Dr. Adrian Ruiz Leal, the enthusiastic, pioneer botanist of Mendoza, Argentina. 4. Gutierrezia ameghinoi Speg., Rev. Fac. Agr. La Plata 3: 527. 1897 Gutierrezia paniculata (DC) Phil. var. patagonica Speg. Rev. Fac. Agr. La Plata 3:608, 1897; Gutierrezia brachyris patagonica Macloskie Rep. Princeton Univ. Exp. Patag. 8:780. 1905 based on, Argentina, “Chubut, M oyano, 1897” (ips 11432!). Small globose shrub, 10-25 cm. tall, from a strong woody root, branching from the base. Root-crown up to 1 cm. in diameter, woody, with dark grey, thick, fissured bark. Branches up to 25 cm. long with a woody base, up to 5 mm. in diameter; branches leafy throughout, internodes 1-10 mm. long. Leaves linear to lanceolate, 10-20 mm. long, 1-3 mm. wide, acute, glabrous, gland- ular-punctate, the blade attenuating into a pseudo-petiole, or sessile. Capitula borne on slender pedicels 3-15 mm. long, in groups of 1-8 at the ends of the branches. Involucre campanulate, 5-7 mm. long, 5-7 mm. wide; involucral bracts 15 or less, 4-6 mm. long, 1.5-2.5 mm. wide, broadly oblong, with a broad triangular ip, usually resinous, membranous margins and a glabrous sur- face, usually yellow or yellow gray in color. Ligulate flowers 7-10, yellow (white occasionally ) tube 1.8 mm. long, ligule 3 mm. long, THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 29 1.5-2.5 mm. wide; tubular flowers 10-15, 2.5-3.5 mm. long. Pap- pus of scales 2.5 mm. long in the tubular flowers, 1.5 mm. long in the ligulate ones. Achenes 1-3 mm. long, 0.5-1 mm. wide, pu- bescent. TYPE: Argentina. Prov. Santa Cruz “rarissime in pratis glareo- sis secus Rio Deseado, anno 1894,” C. Ameghino 94 (xpPs!). DISTRIBUTION: Argentina. Eastern Patagonia from Santa Cruz to Rio Negro; central and western Rio Negro, Neuquén, southern Mendoza and San Luis. Found in dry, rocky “pampas” between 1,000 and 2,000 m., and in the coastal areas of the Patagonian plateau, down to sea level. MATERIAL STUDIED: ARGENTINA. PROV. SANTA CRUZ. Pto. San Julian, Carette s.n. (Lp); Pto. Deseado, Carette s.n. (LP) Eyer- dam, Beetle & Grondona 23871 (st, Mo); s.loc., C. A. Muello 9078 (st). PROV. CHUBUT. Peninsula Valdez, Biraben & Biraben 449 (Lp), 454 (Le); Cabo Raso, Valentin s.n. (Les); Rio Chico, C. Ameghino s.n. (LPs); alrededores de Golfo Nuevo, Guarrera s.n. (Lp); Puerto Madryn, Pennington 70 (corp), 68 (corp), Hicken 63 (st), 73 (st); Trelew, O’Donell 3292 (11L); Gaiman, Krapo- vickas 3880 (Lu); Puerto Piramides, Hicken & Hauman 58 (st), 235 (st); Comodoro Rivadavia, Hicken & Hauman 94 (st); 20 Km. al E. de Ulsen, Soriano 2801 (Bas); cerca de las Chapas, Soriano 2830 (BAB); travesia Kel-la, C. Burmeister s.n. (BAB 2119). prov. R10 NEGRO. Vicinity of Gral. Roca. Fisher 29 (cu, SI, MO, F, BM, K), J. Hirschhorn 163 (Lp), 683 (Le); Allen, O’Don- ell 1931 (a, um); Dique Valcheta, Meyer 7231 (a, LiL), 7207 (xu); Nahuel Niyen, Meyer 7273 (im), 7180 (t1L); Puerto San Antonio Oeste, A. T. Hunziker 4258 (tp), Gomez 65 (st), 661 (st); 85 Km. S. of San Antonio, Eyerdam, Beetle & Grondona 23533 (st, uc, MO, GH); Dept. San Antonio, ruta 3, 1160, Correa, Mendoza & Movia 2470 (Bas); El Bolson, Rasp 102 (LP); camino de Conesa a San Antonio, Biraben & Biraben 405 (Lp); barrancas del Gualicho, Maldonado 280 (tp); Rio Colorado, S. Renato s.n. (st). PROV. NEUQUEN. Antes de llegar a Cerro Negro, O’Donell 2235 (a, tL); meseta de los Chihuidos, Castellanos 20165 (11 ), 20210 (111); alrededores de la ciudad de Neuquén, Castellanos 20137 (11), O’Donell 1893 (A, Lu); Currunco Centro, Maldo- nado 713 (uP), 688 (LP); lago Moquehue, Lagayo 3283B(xP); 40 Km. S. de Zapala Lagayo 3322 (LP); Plaza Huincul, al E. de la Barda Negra, Cabrera 11005 (cu, LP, LIL); Sierra Auca Ma- huida, Fabris 813 (Lp, F, us). PROV. MENDOzA. Dept. San Rafael: 30 OTTO T. SOLBRIG los chanares, Ruiz Leal & Roig 18550 (ri); Santa Elena, 1,700 m., Ruiz Leal 7333 (xv); El Nihuil, Reales 2044 (xu); La Pin- tada, Ragonese & Macola 163 (LP); mina Santa Elena, Ruiz Leal 7333 (Le). Dept. San Carlos: arroyo Hondo, 1,820 m., Ruiz Leal 9627 (RL); arroyo Carrizalito, 1,750 m., Ruiz Leal 7521 (LIL, RL); divisadero del Aguila, Ruiz Leal 6905 (xP, RL); El Papagallo, 1,600 m., Araque 1506(Lu.), Balegno & Palacios 4370 (um), Ruiz Leal 11330 (tm.); Ea. Gaucha, Boelcke 4021 (iP); Ruta 40, 114 Km. al S. de Tunuyan, Boelcke, Bacigalupo y Correa 23 (LP). PROY. SAN Luis. Sierras de San Luis, Deletay-Guinazu 1625 (BAB). 5. Gutierrezia spathulata ( Phil.) Kurtz, Bol. Acad. Nac. Ciencias. Cordoba (Argentina), 13: 194. 1893 (fig. 35) Brachyris spathulata Phil., An. Univ. Chile 27; 336. 1865. Gutierrezia spathulata (Phil.) Kurtz, var. ochroleuca Kurtz, Bol. Acad. Nac. Ciencias, Cérdoba, 13: 186. 1893, based on “Mendoza, inter arroyo Papagayos et arroyo Hondo, 27-XII- 1892,” Kurtz 7431 ( Holotype corp!, Isotype Lp! ). Small globose shrub, 10-20 cm. tall, from a very strong woody root, branching from the base, occasionally with short stolonif- erous branches. Caudex and root crown up to 1 cm. in diameter, woody, with dark grey, thick, fissured bark. Branches up to. 15 em. long, and up to 0.5 cm. thick, leafy throughout; the inter- nodes short and the leaves appressed at the base forming a pseudo-rosette of leaves, more elongate towards the tip of the branches; the upper % of the branches dies back every year so that an old plant at blooming time is formed by a tight cushion of leaves and branches 3-5 cm. thick, from which several flow- ering shoots emerge. Leaves spathulate, 5-25 mm. long, 3-8 mm. wide, obtuse, glabrous, resinous-punctate, margins sometimes shortly ciliate, basal leaves in particular often coated with resin. Heads borne on peduncles 1-5 mm. long, 1-5 at the ends of-the branches. Involucre turbinate, 4.5-5.5 mm. long, 3.5-5 mm. wide; involucral bracts less than 15, 3.5-4.5 mm. long, 1.5-2 mm. wide, glabrous, acute, margins membranous, coriaceous. Ligulate flow- ers 5-8, yellow or orange, tube 1.5-2 mm. long, ligule 3-3.5.mm. long, 1.5-2 mm. wide; tubular flowers 9-12, 3 mm. long, yellow or orange; pappus of scales 2-2.5 mm. long in the tubular flowers, THE SOUTH AMERICAN SPECIES OF GUTIERR (photo Adrian Fic. 35-36. 35. Gutierrezia spathulata, in the cordillera of Mendoza m Ruiz Leal); 36. Gutierrezia baccharoides e La olcura, Mendoza, Argentina. Note the small size of the plants, and the PU eed mat-forming growth habit 32 OTTO T. SOLBRIG 1-1.5 mm. long in the ligulate flowers. Achenes 1-3 mm. long, 0.5-1 mm. wide, densely pubescent, slightly ridged. TYPE; ARGENTINA, “prope in La Guardia in prov. Mendoza reperit orn. Max Landbeck.” (sant.! Isotype Lp! ). DISTRIBUTION: ARGENTINA, on the eastern slopes of the Andes of southern and central Mendoza, Neuquén, and northern Chu- but, between 1,500 and 3,500 m. SELECTED SPECIMENS. ARGENTINA. PROV. MENDOZA. Dept. Tupun- gato: Confluencia A.° Novillo Muerto y Los Monteros, 1,740 m., Roig sn. (RL 15533). Dept. San Carlos: El Cepillo Nuevo, 1,500 m., Sanzini 1844 (RL); Quebrada Alavarado, Ruiz Leal 11330 (RL); entre A.° Cortaderas y A.° Hondo, Ruiz Leal 11005 (vp, LIL, RL), 7521 (LP); camino Pareditas-E] Sosneado, Boelcke 4150 (Lp); El Sosneado, 1700 m., Carette 354 (st). Dept. Mala- hue: Chos Malal, 2,000 m., Combs 180 (x); Los Colgados, Ruiz Leal & Roig 16015 (x1); Portezuelo del Viento, Ruiz Leal 9736 (LIL, RL), 9735 (Lm, RL); Portezuelo del Choique, 2,400 m., Ruiz Leal & Roig 16079 (x); La Valenciana, 2,100 m., Ruiz Leal 7815 (nt.); Bardas Blancas, Braun, 1889 (RL); Portillo de la Mareta, 3,500 m., Kurtz 11971 (corp); Los Molles, Kurtz 7569 (corp); Rodeo Viejo, Kurtz 7189 (corp), PROV. NEUQUEN: Arroyo Toil, 1100 m., Ruiz Leal & Roig 22416 (LP); alrededores del A.° Chaca-y-co, Chichi 47 (Le); Corusico, Dawson 1245 (LP); ruta 40, 25 Km. al S. de Zapala, Perez-Moreau 3146 (LP, BAB). PROV. CHUBUT: entre Trelew y pie de la cordillera, Mori- beau 7 (tp). 6. Gutierrezia baccharoides Sch. Bip., Flora 38: 115. 1855 (fig. 36) Gutierrezia hoffmanii O, Ktze. Rev. Gen. 3(2): 156, 1898; G. baccharoides Sch. Bip. var. hoffmanii (O. Ktze.) Hauman, Anal. Soc. Cient. Arg. 86: 321. 1918: based on Moreno and Tonini 523, “Patagonia” (Isotype ny! tp!, photo of type B!). Gutierrezia pulviniformis Cabrera, Darwiniana 4; 135. 1940. Based on Carette s.n., 1-1916, Mendoza, Dept. San Rafael, Cerro Nevado (xp!), Small, cushion-forming perennial, subshrub, woody, up to 10 cm. tall, but usually forming hemispheric cushions 3-5 cm. in diameter and 3-5 cm. tall, growing together in groups that form mats up to 50 cm. in diameter, branching from the base. Branches short, 3-6 cm. long, 1-3 mm. in diameter, brown in color, woody, THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 33 densely leafy to the top; in older specimens only the dried-up petiole-bases remaining. Leaves oblanceolate to spathulate, 5-15 mm. long, 2-5 mm. wide, obtuse or somewhat acute, petiolate, surface glabrous, slightly fleshy, resinous-punctate, margins slight- ly ciliate, often inrolled. Capitula borne singly at the ends of the leafy undivided branches. Involucre campanulate, 5-6 mm. long, 4-7 mm. wide; involucral bracts less than 15, in three loose series, 4-5 mm. long, 1-2 mm. wide, oblong, yellow, coriaceous, with a broadly acute triangular tip that is usually green, margin mem- branous and glabrous. Ligulate flowers 6-10, yellow, tube 1.5 mm. long, ligule 2.5 mm. long, 1-1.5 mm. wide; tubular flowers about 10, yellow, 3 mm. long; pappus of scales 0.8-1.5 mm. long; achenes about 1-2 mm. long, turbinate, densely pubescent. “type: Chile, Prov. Magallanes “ad sinum Oazy Harbour, ad terminum accesus maris” Febr., Lechler pl. mag. 1241 (x!, Iso- type F!, ny!). DISTRIBUTION: Argentina and Chile. From the high Cordillera of Mendoza to the Straits of Magellan, in alpine, subalpine, or subantarctic environments; in Mendoza it is found between 3,000 and 4,000 m., in Magallanes and Santa Cruz at sea level. SELECTED SPECIMENS. ARGENTINA. MENDOZA: Pampa de La Pol- cura, 3,000 m., Solbrig 3399 (cu, Lp, us), Roig 4204 (GH, RL), Ruiz Leal 21671 (x); Potrerillos, canchas de Esqui, 3,000 m., Cuezzo & Say 2549 (xiL, Lp); cerca del nacimiento del arroyo Las Cuevas, 3,200 m., Ruiz Leal 3128 (LP, LIL, RL); Vallecito, 2,700 m., Sanzini 1311 (LP, RL); Portezuelo Morado, 3,800 m., Semper, 1949 (LP, RL); Portezuelo de Las Osamentas, 3,000 m., Ruiz Leal 7192 (iP, LiL, RL); Cuicenita, Carette, 1907 (corD). PROV. NEUQUEN: Cerro Colohuincul, Comber 1091 (x, Lp); Rincon Grande, 1,000 m., Neumayer 384 (Lp); Lago Lagar, Roth, 1896 (Lp); fuentes del Rio Caleufu; 2,000 m., Roth, 1896 (Le), Mauri, 1897 (Les); Lago Villarino, Roth, 1896 (LP). PROV. RIO NEGRO: s. loc., Wolffliigel 20 (st). PROV. CHUBUT: Carrenleofu, Illin s.n. (ces 16813); entre Rio Pico y Lago Vintter, Soriano 3075 (LP, BAB); 20 Km. de Rio Pico en camino a Lago Winter, Krapovick- as 4112 (Bas); lat S. 44° 24’, long 0 71° 22’, Hogberg, 1902 (st, BAB). PROV. SANTA CRUZ: Puerto San Julian, Blake 107 (3M, K); San Julian-Rio Deseado, C. Ameghino, 1899 (LPs). CHILE. TIERRA DEL FUEGO: Punta Anegada, Spegazzini 104 (LP). This is the most distinct of all species of Gutierrezia because 34 OTTO T. SOLBRIG of its growth-habit. Nevertheless, it is a polytypic taxon also, and specimens from Mendoza differ in size of plant, size of capitulum, number of flowers, and dimensions of the leaves from specimens from Magallanes. Since the distance separating both localities is about 2,000 Km., and the difference in altitude is over 3,000 m., it is not surprising to find such variations. I have seen enough inter- mediate specimens from localities lying between the northern and southern extremes mentioned, to convince me that we are dealing with only one taxon. However, the Cordillera of Chubut and Santa Cruz is incompletely known botanically, and more field-work may show that Gutierrezia baccharoides is a complex of more than one taxon. 7. Gutierrezia neaeana (DC.) Blake, Contr. U. S. Nat. Herb. 26: 23. 1930 (fig. 37) Brachyris neaeana DC., Prodr. 5: 313. 1836. Small, fruticose, semiglobose, woody shrub, up to 60 cm. tall, from a strong, woody taproot. Root-crown woody, up to 10 mm. in diameter. Branches all from the root-crown, woody and yellow in color and covered with the bases of dried leaves in their first third; leafy, green, and less woody above; up to 5 mm. in diam- eter, striate. Internodes short below, longer above, up to 20 mm. in length. Leaves entire, oblanceolate, 15-20 mm. long, 2-3 mm, wide, glabrous, sessile, narrowing towards the base, broadly acute at the tip. Heads borne solitarily at the ends of the branches or in groups of 2-4 on long (10-30 mm.) pedicels; involucres broadly turbinate or campanulate, 5-7 mm. high, and 7-8 mm. wide, with membranaceous margins and tip; ray flowers 9-11, ligulate, yellow, ligule up to 5 mm. long and 2.5 mm. wide; disk flowers 11-13, yellow, 5-6 mm. long; pappus approximately 1.5 mm. long; achenes terete, pubescent, 1-2 mm. long at maturity. FIG.) 374. ‘TyPE: Locality unknown, “Chile or Mexico?”, fide De Can- dolle. Née leg: (Microfiche of specimen in De Candolle her- barium!) . DISTRIBUTION: Known only from the type (locality unknown ) and one collection from the high Cordillera of Central Chile. MATERIAL STUDIED: CHILE. PROV, ATACAMA: Dept. Vallenar, Quebrada Alfalfa, 3,500 m., Johnston 6000 (GH) THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 35 php ie =< SAS eo 5 SS ing ty nd NEL Fic. 37, Gutierrezia neaeana (Johnston 6000, cu). Drawing by Arnold Clapman. Gutierrezia neaeana was described by De Candolle from ma- terial collected by Née of the Malaspina expedition. The type does not have any annotation as to locality, and De Candolle in his description says only “Chile or Mexico?” The plant was re- described by Blake, who rightly points out that it is not any of the known Mexican species, but also that it has not been col- lected since in South America, The specimen studied is to my knowledge the only other known collection of the species. It coincides well with the description of De Candolle and the ampli- fied one of Blake, except as to size, the Née specimen being much smaller. Nevertheless, height within a species often varies con- siderably, particularly in high mountain plants. Gutierrezia 36 OTTO T. SOLBRIG neaeana is characterized by its oblanceolate leaves, broad invo- lucres, and the lax arrangement of the heads at the ends of the branches. It was collected in the high Cordillera east of Coquim- bo, a region poorly explored botanically. Johnston tentatively identified it as aff. copiapina Phil., but later changed it to repens var. nov.? Gutierrezia copiapina (= resinosa) has much nar- rower leaves, is bushy and woodier, and is a species of low altitudes. Gutierrezia repens, although similar in its foliar char- acteristics, is a creeping plant found in the Aconquija Mts. of Argentina. Its leaves are also lanceolate rather than oblanceolate as in G. neaeana. The lack of collections of G. neaeana (as well as those of G. espinosae and others ) clearly shows the need for more collecting in the high Cordillera and in northern Chile. 8. Gutierrezia resinosa (H. & A.) Blake, Contr. U. S. Nat. Herb. 26: 232. 1930 (fig. 43) Galinsoga resinosa H. & A., Bot. Beechey Voy. 32, 1830; Bahia resinosa DC. ex Hook. f. & Jacks., Index Kew. 1: 264. 1893. Brachyris paniculata DC., Prodr. 5: 313. 1836; Gutierrezia paniculata (DC.) Gray, PI. Wright. 2: 78. 1853 (in text); Gutier- rezia brachyris Macloskie, Rep. Princeton Univ. Exped. Patag. 8: 780. 1903-06; based on “in Chili circa Coquimbo” Gaudichaud (Microfiche G-DC! ) Odontocarpha poeppigii DC., Prodr. 5: 72. 1836; based on “Chili, in collibus arid. inter Caucon et Tavolingo,” leg. Poeppig (photo cu! ), Gutierrezia compacta Phil., Anal. Univ. Chile 87: 427. 1894; based on “prope La Serena, Jan. 1883,” leg. Phillippi (sant!, Isotype ip!) Gutierrezia laricifolia D. Don, ex Hook. & Arn., Comp. Bot. aie 2: 51. 1836; based on “Coquimbo,” leg. Mr. Calcleugh (not seen). Brachyris floribunda Phil., Linnaea 33: 137. 1864; G. paniculata var. floribunda ( Phil.) Reiche, Anal. Univ. Chile 109: 23. 1901; based on “prope S. Felipe de Aconcagua, Martio 1863,” leg. Phillippi_ (sanr!, Isotype LP!). Large, woody shrubs, 80-150 cm. tall, with one to several well- developed stems up to 15 mm. in diameter, branching in the THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 37 upper portion; stems striate, with brown-grayish bark; secondary branches lighter, or greenish, and resinous. Leaves linear to linear-lanceolate, 20-40 mm. long, 1-2 mm. wide, entire, acute, sessile, with entire margins, resinous on both surfaces, but amounts of resin varying with plants and environmental condi- tions. Heads borne in corymb-like groups at the ends of the branches, each branchlet with 1-5 heads. Involucres turbinate, 6-8 mm. long, 3.5-5 mm. wide, involucral bracts in three loose rows, triangular-lanceolate, coriaceous, yellow, with green tips; ray flowers ligulate, 5-7, yellow or white, ligule 2-3 mm. long, 1-2 mm. wide; disk flowers 8-11, yellow or white, 4-5 mm. long; pap- us of awns 2-3 mm. long; achene terete, 2-4 mm. long, pubes- cent. TYPE: Chile, Coquimbo (not seen) DISTRIBUTION: Coastal and interior, extra Andean, central Chile, from Vallenar and Coquimbo south to Santiago and Valparaiso. Locally abundant. SELECTED SPECIMENS: CHILE. PROV. SANTIAGO: Dept. Santiago: Cerro de La Cruz near Limache, Solbrig 3395 (cu, Us, LP), 3394 (cH, us, LP); Garaventa 2155 (cH); Limache, Looser 659 (cH); Dept. Valparaiso: Valparaiso, Hartwig 39 (cH); Vina del Mar, Hicken 200 (st). prov. coguimso: Dept. Ovalle: Fray Jorge, Munoz 229 (cu); Cerrillos de Tamaga, Worth & Morrison 16418 (mo, cH); 3 Km. N. Empalme ruta a Ovalle, on Santiago-Serena road, Solbrig 3044 (cH). Dept. Coquimbo, Est. E] Tangue, Sol- brig 3041 (cH); lomas de Peiuela, Solbrig 3031 (cH), 3387 (cH); La Serena, Barros 111 (xP), 1713 (LP), Philippi s.n. (LP). Dept. Rivadavia, Vicufia, Cabrera 104 (ip); Rivadavia, Werdermann 173 (cu), Solbrig 3384 (cH). Although G. resinosa is one of the most distinct species of the genus it has been described six times, and ascribed to five dif- ferent genera! This is due, no doubt, to the fact that it grows in the area of Chile that was most frequently visited by the early botanical expeditions. Gutierrezia resinosa is replaced north of its range by G. taltal- ensis and G. espinosae, two closely related, and most probably derived species. The area north of Serena is very incompletely known botanically, and it is likely that further exploration will reveal a whole complex of small, isolated populations with dis- tinctive characteristics. Which of these deserve specific status has to be decided from the evidence at hand. At present, I feel it is 38 OTTO T. SOLBRIG Fic. 38 . 38-41, Flowering branches of G. taltalensis and G. espinosae showing ex remes of variation in leaf size and shape, and arrangement of capitula. pinosae (Johnston 5722, Gu); 39. Id. taltalensis (Johnston 5665, GH); 38. Gutierrezia es- d (Espinosa, 10-XI-1941, Gu); 40. Gutierreza 41. Id. (Johnston 5533, Gu) best to recognize only the three named species, Even so, the separation of G. taltalensis from G. espinosae is not always easy (see figs. 38-41). 9. Gutierrezia taltalensis Phil., Anal. Univ. Chile 87: 426. 1894 ( fig. 40, 41) Gutierrezia copiapina Phil., Anal. Univ. Chile 87: 425. 1894, based on “prope Chajarcillo invenit orn, Guillermo Geisse 1885. (sant!, Isotype xr! ). Medium to large shrubs, 40-120 cm. tall, from a woody base up to 10 mm. in diameter, branching dichotomously; bark light gray, often with the bases of dried up leaves; secondary branches thinner, leafy, internodes short. Leaves linear to linear-lanceo- late, 20-40 mm. long, 1-3 mm, wide, entire, glabrous, resinous, acute, sessile. Flowering heads arranged in tight groups of 3-8, sessile or on short pedicels 1-3 mm. long, or occasionally solitary; involucre broadly turbinate, 5-6 mm. long, 3-5 mm. wide; in- volucral bracts in three obscure rows, lanceolate, acute, coria- ceous, with slightly membranaceous margins, and a darker tip; ray flowers ligulate, 7-10, white or yellow, ligule 3-4 mm. long, THE SOUTH AMERICAN SPECIES OF GUTIERREZIA 39 1-2 mm. wide; disk flowers tubular, 10-15, white or yellow, 4-5 mm. long; pappus of awns 1-2 mm. long; achenes terete, 1-3 mm. long, pubescent. Fic. 40, 41. type: Chile. “Proper Taltal in deserto Atacama, legit Dr. Lud. Darapsky” (sant! ). MATERIAL STUDIED: CHILE. PROV, ANTOFAGASTA: Dept. Taltal, Cerro Perales, ca. 1,000 m. alt., Johnston 5625 (cu, us); Quebrada de San Ramon, I. M. Johnston 5150 (cu, us); Posada Hidalgos, I. M. Johnston 5665 (cu, us); Hills S. E. of Taltal, I. M. Johnston 5112 (cu); Agua de los Mantos, I. M. Johnston 5658 (GH, US); Aguada Panulcito, I. M. Johnston 5472 (cu, us); El Rincon, I, M. Johnston 5533 (cu, us); Aguada Cachina, I. M. Johnston 5723 (cH); Aguada de Miguel Diaz, I. M. Johnston 5331 (cH); Aguada Grande, I. M. Johnston 5751 (cH). 10. Gutierrezia espinosae Acevedo, Bol. Museo Hist. Nat. Santiago 24: 82. 1949 fig. 38, 39) Small, woody shrub, 40-50 cm. tall. Primary branches up to 10 mm. in diameter, with a rough, gray bark. Secondary branches woody, rough, with prominent elevated leaf scars, leafy in their upper part. Internodes short, 1-3 mm., the branches covered with a gray to gray-brown bark, never green. Leaves mostly oblanceo- late, short and broad, 5-15 mm. long, 3-4 mm. broad, or towards the top in well-developed specimens larger and up to 25 mm. long and 5 mm. broad, acute, sessile, surface scrobiculate (at least when dry) due to the presence of large cavities which house the resin-glands, margin entire, with denticulate trichomes. Heads borne in tight glomerules of 3-5 sessile heads or on very short pedicels, 1-5 mm. long, or occasionally solitary. Involucre broadly turbinate, 5-6 mm. high, 4-6 wide; involucral bracts in about 3 unequal rows, lanceolate, acute, coriaceous, yellow with a green tip; ray flowers ligulate, 7-9, white or yellow, ligule 3-5 mm. long, 1-2 mm. wide; disk flowers tubular 10-15, white or yellow, 4-5 mm. long; pappus of short awns 1-1.5 mm. long; achenes terete, pubescent, 1-3 mm. long at maturity. FIG. 38, 39. TYPE: Chile. Prov. Antofagasta. Dept. Antofagasta, Quebrada de la Chimba, Marcial R. Espinosa s.n., 10-XI-1941 (sant! Isotype GH! ). asl Chile, from Taltal to Antofagasta, in “Quebra- 40 OTTO T. SOLBRIG das” and “Aguadas” near the coast, where some humidity is present. MATERIAL STUDIED: CHILE. PROV. ANTOFAGASTA: Dept. Taltal. Cerro de la Cachina, ca. 14 Km. S. of Aguada Cachina, I. M. Johnston 5722 (cx, us); Dept. Antofagasta. Cerro Gordo, 35 Km. S. of Antofagasta, alt. 1,000 m., West 6066 (cH); La Chimba, Barros 70 (ie), 1354 (Le); Cerro Moreno, Barros 1319 (LP). SECTION MEGALOCEPHALA SOLBRIG, sect. nov. Capitula magna, campanulata, ligula alba. Species typicus, Gutierrezia gayana (Remy) Reiche, Anal. Univ. Chile 109: 22. 1901. 11. Gutierrezia gayana (Remy) Reiche, Anal. Univ. Chile 109: 22. 1901 (fig. 22, 42) Odontocarpha gayana Remy, in Gay, Flora de Chile, Laminas, Tabl. 44. 1846; Brachyris gayana (Remy) Remy, in Gay, Flora de Chile 4: 36. 1849. Globose shrub, 50-100 cm. tall. Stems woody, gray or grayish- yellow, up to 5 mm. in diameter, branching profusely; secondary branches striate, green or green-gray. Leaves linear, 3-5 cm. long, 1-2 mm. wide, entire, glabrous, acute, sessile, very glutinous. Heads borne solitary at the end of the branches. Involucre broad- ly campanulate, 8-12 mm. high, 8-15 mm. wide; involucral bracts arranged in 2-3 very loose rows, broadly lanceolate, 5-10 mm. long, 3-4 mm. wide, membranous with a thin green tip, up to 3 mm. long. Ray flowers 10-15, ligulate, white, ligule 5-7 mm. long, 3-4 mm. wide; tubular flowers white, 4-6 mm. long; pappus about 1.5 mm. long; achene terete, pubescent, 1-3 mm. long at maturity. Fic. 22, 43. TYPE: Chile. Tabl. 44, Laminas, Gay, Flora de Chile, 1846. DISTRIBUTION: Endemic to the dry pre-Cordilleran “pampas” north and east of Serena, Chile. MATERIAL STUDIED: CHILE. PROV. COQUIMBO: Bajada Cuesta Vineta, 108 Km. from Serena, Solbrig 3073 (cu), 3072 (cH), 3070 (cH); Llano “El Potrero,” 84 Km. from Serena, Solbrig 3055 (cH); Incahuasi, Jiles P. 2226 (LP). PROV. ATACAMA: Dept. Huasco. Carretera Panamericana, 35 Km. al $. de Vallenar, Ca- brera 12681 (xr); Id., 30 Km. S. de Vallenar, Cabrera 12621 (LP); 2 ~ = ce) 72) £3) S) io) mu 2) Z < QO ~ % <3} fend a < co fa ~ © 7) be x 2a note Chile), several main, Serena, a, (near La (Cuesta Vinheta, Coquimbo, Chile); note the sia resinos uiterre. sia gayana p45. G Gutierre. 42 42-43, Fic. large heads of white flower the erect habit, with very little side branching, and the tendency to form trunk e branches. -lik 42 OTTO T. SOLBRIG camino a Domeyko, entre 36-40 Km. al S. de Vallenar en la quebrada, Mufioz & Johnson 2001 (xP). Chile, s. loc., Gaudi- chaud 104 (cH), Gay s.n. (US, NY). EXCLUDED SPECIES Brachyris sectifolia Cessati, Att. Acc. Sci. Nap. 5: 7. 1871. Accord- ing to the description this species from Villavicencio, Mendoza, Ar- gentina, has divided and opposite leaves. Although I have not seen type-material this species is probably not a Gutierrezia. Gutierrezia cupressiformis Sch. Bip., Flora 38: 114. 1855 = Lepido- phyllum cupressiformis (Lam.) Cassini, Dict. Sci. Nat. 26: 37. Gutierrezia ledifolia Griseb., Goett. Abh. 19: 173. 1874 = Chilio- trichiopsis ledifolia (Gris.) Cabrera, Notas Museo La Plata 9: 244. 1944. LITERATURE CITED BentuaM, G. & J. D. Hooker. 1873. Genera Plantarum, Vol. 2(1), 544 pp. London De CaNnpo_te, A. 1836. Prodromus Systematis Naturalis, Vol. 5, 706 . Paris. Hooker, W. J. anp G. A. Arnott. 1830-1841. The Botany of Captain Beechey’ S varies 485 Pp. London. 1836, oo Towards a Flora of South America. Comp. Bot. Mag. 214 Nutra.t, T. 1818. The genera of eon paeaias Plants and a Cata- logue of the Species to the Year 1817. Vol. 2, 254 pp. Philadelphia. Remy, J. 1848-49. Ex Gay, Historia Fisica y Politica de Chile, Botani- ca, Compositae, Vol. 3 & 4. Paris and Chile. Reicue, C. 1902. Flora Chile, Vol. 3, 425 pp. Santiago de Chile. RipenseRc, L. & O. T. Sousric. 1963. Chromosome Number and Morphology in the ts Gutierrezia (Compositae-Astereae). Phy- ton 20: 499-20 SoLsric, O. T., ce Cytotaxonomic and Evolutionary Studies in the North American Species of Gutierrezia (Compositae). Contr. Gray Herb. no. 188: ee 960b. The Status of the Genera Am- phipappus, eRe Greenella, Gutierrezia, Gymnosperma, and Xanthocephalum. serve 62: 43-54. Infraspecific Variation in the Gu- 1964, tierrezia sarothrae Complex oe -Astereae). Contr. Gray as no. 193: 67-115. . The California i of Gutier- rezia (Cimajiolites: as ess 18: 75-84 CHROMOSOME NUMBERS OF CRUCIFERAE Reep C, Rows There has been a steady increase in our knowledge of chro- mosome numbers of the family Cruciferae following the com- prehensive paper by Manton (1932). However, the point has scarcely been reached where these data can be put to their maximum usefulness. Many more counts, together with authentic determinations of the taxa involved, supported by voucher specimens, are needed before a new comprehensive evaluation of cytological data on the internal classification of the family is justified. The problem of generic delineation is an unusually difficult one in the Cruciferae (cf. Rollins, 1960, 1962). In some instances, a consistency of chromosome number within genera coupled with differences between related genera are helpful in determin- ing where generic lines of demarkation are rightfully placed. The reliability of any such patterns that emerge will be dependent upon the completeness and accuracy of the chromosome counts. Proper application of these data to solutions of particular prob- lems demands broad considerations of diverse kinds of informa- tion which can be convincing only if they are evaluated and presented in considerable detail. Obviously, this is not our pres- ent purpose. The following listing of chromosome numbers and the accompanying comments are intended to help in rounding out the needed chromosome data and to highlight some cytologically interesting problems that have turned up in the Cruciferae dur- ing the past three decades of research on this interesting family. ACKNOWLEDGEMENTS terial has been contributed by a number of collaborators, especially by Dr. Peter Raven. I wish to thank those who have contributed in any 44 REED C. ROLLINS way to this study, particularly Dr. Kuldip R. Khanna, who actively collaborated with me in of this research. Financial support for portions of this work has been given by the National Science Founda- Arabis A. crandallii Robinson 2n = 14: Gunnison Co., Colorado. Ripley and Bameby 10206, cu. 2n = 14: Hinsdale Co., Colorado. Rollins 51165, cu. A. crandallii x A. Holboellii Hornem., var. retrofracta (Grah.) Rydb. 2n = 21: Gunnison Co., Colorado. Rollins 5194, cH. A. demissa Greene, var. russeola Rollins 2n = 21: Sweetwater Co., Wyoming. Rollins & Porter 5134, GH. A. divaricarpa A. Nels 2n = 14: Park Co., Colorado. Rollins & Weber 51290. cu. 2n = 14: Conejos Co., Colorado. W. A. Weber 7845. cH. 2n = 14: Park Co., Montana. Rollins & Porter 51244, GH. 2n = 14: Park Co., Wyoming. Rollins & Porter 51252, cu. 2n = 22: Siskiyou Co., California. J. T. Howell 15193, cu. A. drummondii Gray 2n = 14: Park Co., Colorado. Rollins & Weber 51289, cu. 2n = 14: Park Co., Montana. Rollins & Porter 51246, cu. 2n = 14: Park Co., Wyoming. Rollins & Porter 51250, cH. A. fendleri (Wats.) Greene, var. spatifolia (Rydb.) Rollins 2n = 14; Douglas Co., Colorado. Rollins 5147, GH. A. holboellii Hornem., var. pendulocarpa (A. Nels.) Rollins 2n = 14: Yellowstone National Park, Wyoming. Rollins & Porter 51281, cu. A. holboellii var. pinetorum (Tidestr.) Rollins 2n = 21: Sweetwater Co., Wyoming. Rollins & Porter 5135, GH. A. holboellii var. retrofracta (Grah. ) Rydberg an =. |4- Siskiyou Co., California. L,. Constance s.n., GH. 2n = 14: Park Co., Montana. Rollins & Porter 51245, cH. A. lemmonii Watson . 2n = 14: Park Co., Wyoming. C. L. Porter 5888, GH. lyrata L.., var. kamchatica Fischer ree — gies River Valley, Arctic Slope of Alaska. Hodg- CHROMOSOME NUMBERS OF CRUCIFERAE 45 2n = 32: Nixon Mine, Kuskokwim Mts., Alaska. Drury 3380, GH. A. perstellata Braun, var. ampla Rollins n = 7: Davidson Co., Tennessee. Rollins, Solbrig, Hilferty & Lloyd 6012, cu. Polyploidy was firmly established in certain species of Arabis by evidence presented earlier (Rollins, 1941), Also, the data showed that x — 7 was a common fundamental number in the genus. Mulligan (1964) suggests that all species of Arabis ex- clusively North American have a chromosome number based on x = 7, whereas the European and Asiatic species are based on x = 8. We have no data contrary to this division but I suspect the correlation is with phylogenetic relationship rather than with geography. It just happens that we do not have counts on any of the exclusively North American species that are obviously related to those of Eurasia. To test the point, counts are needed in such species as Arabis blepharophylla H. & A., A. oregana Roll., A. modesta Roll., A. aculeolata Greene, A. furcata Wats., A. Nuttallii Robins., and A. crucisetosa Const. & Roll. The discovery of apomixis in Arabis holboellii Hornem. by Bécher (1951) opened the way to a more reasonable explanation of the inconsistent chromosome numbers in Arabis than was available earlier. Where apomixis is operative, triploids and various aneuploid numbers can persist in wild populations with- out difficulty and several different chromosome numbers within a given species are then not surprising. There is very good evidence (Rollins, 1946) that interspecific hybridization occurs in Arabis and I am convinced that hybridiza- tion between taxa at specific and infraspecific levels is wide- spread in the genus. Hybridization, polyploidy and og soameana undoubtedly operate together to provide several polymorphic groups within the genus. Taxonomically, the most difficult and puzzling complexes are the following, designated by the name of the species with ee other less prominent taxa are associated: (1) the Arabis holboe i complex, (2) the Arabis sparsiflora complex, (3) the Arabis divaricarpa complex, (4) the Arabis fendleri complex, (5) the Arabis lemmonii complex. Although we do not have direct evi- dence as yet, it is very probable that all three phenomena (i.€., hybridization, polyploidy and apomixis), occur separately o oF gether to provide the complex patterns of variation found in eac complex. 46 REED C. ROLLINS Barbarea B. orthoceras Ledeb. n = 8: San Luis Obispo Co., California. Breedlove 2030, GH. This count is in line with that of Mulligan (1964) for the species and of other authors for the genus as a whole. The funda- mental number x = 8 is well established. Cakile C. fusiformis Greene n = 9: Aransas Co., Texas. Rollins & Correll 5964, GH. C. geniculata (Robins. ) Millsp. n = 9: Galveston Co., Texas. Riidenberg. March, 1966, c.u. Count by L. Riidenberg. Previous counts on other species of Cakile are from more north- erly stations in Europe, Asia and North America, However, all counts agree with a fundamental number of x — 9. There is still a major need for studies of variation patterns in populations of Cakile. Pobedimova’s (1964) recognition of eight species in North America and the West Indies requires a careful evaluation. Cardamine C. breweri Watson n = 42-48: Fresno Co., California. Breedlove 5242, cu. C. parviflora L., var. arenicola (Britt.) Schulz hs 99 34. Morgan Co., Alabama. Rollins et al. 6103, cH. C. digitata Richardson 2n = 32: Umiat, near Colville River, Alaska. Thompson 1217, cu. settled upon. It is clear that x — 8 is a fundamental number in Cardamine and that polypoidy is widespread in the genus. Caulanthus C. coulteri Watson n = 14: Santa Barbara Co., California. Breedlove 1929, cH. C. flavescens ( Hook.) Payson n = 14: Alameda Co., California. Breedlove 4295, cu. CHROMOSOME NUMBERS OF CRUCIFERAE 47 PLate 1. Upper left, chromosomes of Lesquerella re” ; Rollins and Correll 5950; upper right, pager i ner lore ta ow: rre 'T lemmonii, n = reedlove. 1954; lower right, chromosomes of Tropi- docarpum gracile, n = &, eotae 1822. All figures x 3900. 48 REED C. ROLLINS C. heterophyllus (Nutt.) Payson n = 14: San Diego Co., California. Breedlove 1831, cH. C. inflatus Watson n = 10: Kern Co., California. Rollins 4160, ps. n = ca. 10: Fresno Co., California. Rollins 4159, ps. lasiophyllus (H. & A.) Payson n = 14: Kern Co., California. Breedlove 1951, cu. lasiophyllus var. inaliens (Robins.) Payson. n = 14 ?: Marin Co., California. Breedlove 4402, cx. lasiophyllus var. utahensis (Rydb.) Payson n = 14: San Diego Co., California. Breedlove 1859, GH. lemmonii Watson n = 14: Kern Co., California. Breedlove 1954, cu. Plate 1. n = 14: Monterey Co., California. Breedlove 4312, cx. The long-standing question as to whether Caulanthus should be maintained as a genus distinct from Streptanthus is not af- fected by the chromosome numbers now known. Species in both genera are quite consistently n = 14. The exceptions, in addition to Caulanthus inflatus given above, are n = 12 in C. crassicaulis and Streptanthus cordatus (Rollins, 1939) and 2n — 48 for Caulanthus lasiophyllus reported by Snow (1959) under the name Thelypodium lasiophyllum. Our findings are different for C, lasiophyllus, but this merely suggests a complex chromosome number pattern paralleling a known complex and puzzling tax- onomic situation. There is a great need for extensive and detailed studies of C. lasiophyllus because of the morphologically diver- gent plants at present accepted as belonging to this species. The nature and range of variation have not been established. A second known problem involving C. lasiophyllus involves its generic placement. Schulz (1924) associated it with a small group of Asiatic species comprising the genus Microsysimbrium but this does not seem to be a well founded solution. The chromosomes of GC. lemmonii are shown in Plate 1. Nese a ee 2 Cochlearia C. groenlandica L. 2n = 14: Prince Charles Island, Canada, W. K. W. Baldwin 1894, cu. The number 2n = 14 is in accordance with numerous counts from Greenland, Canada and Iceland (Saunte, 1955) for this CHROMOSOME NUMBERS OF CRUCIFERAE 49 species. The genus has two polyploid series based on x = 6 and Dentaria D. integrifolia Nuttall n = 16: Santa Barbara Co., California. Breedlove 1773, cu. D. integrifolia, var. californica ( Nutt.) Jepson n = 8: San Mateo Co., California. Rollins 2947, ps. n = 16: San Mateo Co., California. Rollins 4196, ps. The fundamental chromosome number x = 8 is the same for Dentaria and Cardamine and no evidence is contributed to the problem of whether both of these genera should be recognized or whether all species should be placed in Cardamine. D. in- tegrifolia var. californica occurs both in open valley swales and on wooded slopes. In a limited area in San Mateo County, Cali- fornia, we found the polyploid in open areas and the diploid on the slopes of the Santa Cruz Mountains. A worthwhile problem for investigation would be to see whether such a correlation is widespread and to determine the significance of such a correla- tion if it does exist species-wide. Dithyrea D. californica Harvey n = 10: Mohave Co., Arizona. Rollins 4164, ps. n = 10: San Diego Co., California. Breedlove 1855, GH. D. wislizenii Engelmann n = 9, 2n = 18: Pinal Co., Arizona. Rollins 4168, GH. n = 9: Presidio Co., Texas. Rollins & Correll 61100, cu. D. wislizenii, var. palmeri Payson — 9: Howard Co., Texas. Rollins 53116, cH. The number n = 10 for D. californica is the same as that of Lewis (1959) and of Raven et al. (1965), and n = 9 appears to be a common number in D. wislizenii. More counts need to be made on the latter species, especially the annual winter-bloom- ing populations of Arizona. In addition, data from other species of the genus are required before a clear pattern of chromosome numbers will emerge. Draba D. glabella Pursh 2n = ca. 75: Point Jay, Alaska. J. H. Thomas 2297, GH. 50 REED C. ROLLINS D. lanceolata Royle n = 16: Park-Summit Co. line, Colorado. Rollins, Weber & Livingston 5155, cu. D. oligosperma Nutt. 2n = ca. 60: Albany Co., Wyoming. Ripley & Barneby 10536, cu. The taxonomy of Draba is very confused. This is particularly true of the Arctic and subarctic species and those of high altitudes in the mountains. Chromosome counts on many of the species are high, ranging upward from n — 16. A polyploid pattern based on x = 8 for Draba appears to be emerging but the chromosomes are so small in many instances that it is extremely difficult to obtain a certain count. We have no solid evidence that apomixis occurs in the genus. However, on the basis of the frequent and probably variable chromosome numbers found, it is a fairly safe prediction that apomixis together with interspecific hybridization and polyploidy are responsible for the confused taxonomic pic- ture in the genus. Erysimum E. capitatum (Dougl.) Greene n = 18: Contra Costa Co., California. Breedlove 4282, cu. n = 18: Santa Clara Co., California. Breedlove 4673, GH. - concinnum Eastwood n = ca. 18: Marin Co., California. Breedlove 4449 GH. . pallasii (Pursh) Fernald 2n = 36: Lake Noluk, Brooks Range, Alaska. H. J. Thomp- son, Ds. Most of the definitive counts made on Erysimum indicate a fundamental number of x — 9. Our counts on E. capitatum are the same as that given by Raven et al. (1965) for E. capitatum and E, capitatum var. bealianum. The number 2n — 36 for E. pallasii is in line with other counts in the genus but is somewhat different from an estimated count of 2n — ca. 28 by Holmen (1952) for this species. Our count also differs from the counts of n = 12 and 2n — 24 given by Mulligan (1966) for E. pallasii. e >| ~ Eutrema E. edwardsii R. Brown = 18: North slope, Brooks Range, Alaska. Thompson 1342, cu. CHROMOSOME NUMBERS OF CRUCIFERAE 51 Eutrema edwardsii is a widespread species of arctic and sub- arctic regions occurring on all continents that extend into these high latitudes. It is morphologically variable and also appears to have several chromosome races. The counts of 2n — 28 and 2n = 42 by Mulligan (1964) substantiate the same counts by others. Our count of 2n — 18 introduces a complication that is not at present open to resolution. Halimolobos H. perplexa (Hend.) Rollins n = 14: Adams Co., Idaho. M. Ownbey 3293, cu. Previous counts (Jérgensen, Sgrensen and Westergaard, 1958, Mulligan, 1964) on H. mollis agree on 2n = 16, pointing to a ase number of x = 8. However, our finding of 2n = 14 for H. perplexa suggests x = 7 may be another fundamental number in the genus. Lepidium ™ . densiflorum Schrader n = ca. 16: Morgan Co., Alabama. Rollins et al. 6115, cu. . jaredii Brandegee n = 8, 2n = 16: San Benito Co., California. Wiggins & Rollins 18, cu. . perfoliatum L. n = 8: White Pine Co., Nevada. Breedlove 5814, cu. . strictum ( Wats.) Rattan n =ca. 16: Santa Cruz Co., California. Breedlove 4635, cH. Lepidium continues to check out as “a very uniform poly- ploid genus,” as suggested by Manton (1932 ) in her early gs gal on the cytology of the Cruciferae. L. jaredii is a very distinct localized species of California and it is interesting to find that its chromosome number conforms to the pattern otherwise known in the genus. IN ig — Leavenworthia Reference is made to table 1, pages 9 & 10, Contributions ag the Gray Herbarium No. 192, 1963, where a detailed ae 2 chromosome numbers is given. The numbers n = ia 1 omg n = 24 are found in the genus. No new counts have been made. iN by Ee i ee! ite, REED C, ROLLINS Lesquerella . alpina ( Nutt.) Watson, var. spathulata (Rydb.) Payson 2n = 12: Custer Co., South Dakota. Ripley & Barneby s.n., GH. . angustifolia Nuttall n = 5: Choctaw Co., Oklahoma. Rollins 597 ly:Gn; n = 5: Choctaw Co., Oklahoma. Rollins 6151, cu. - arenosa (Richards. ) Rydberg 2n = 18: Custer Co., South Dakota. Ripley & Barneby 10559, cu. . argyrea (Gray) Watson n = 6: Victoria Co., Texas. Rollins 5361, cH. n = 7: Victoria Co., Texas. Rollins 5566, GH. n = 7: South of Saltillo, Coahuila, Mexico. Rollins & Tryon 58121, cu. n = 8: Kennedy Co., Texas. Rollins & Correll 5961, cu. n = 8: Webb Co., Texas. Rollins & Correll 5944, cu. Pate 1. n = 9: Uvalde Co., Texas. Rollins & Correll 5942, cu. n = 12: Refugio Co., Texas. Rollins 5359, GH. n = ca. 15: Refugio Co., Texas. Rollins 5564, cH. oo We, eS TT SS Webb Co. Téxas’ Rollins & Correll 5946, cH. n = 18: Llano Co., Texas. Rollins 53104 and 53105, cu. . arizonica Watson nm = 95: Mohave Co., Arizona. Rollins 4167, cu. : duriculata (Engelm. & Gray) Watson n = 8: Comanche Co., Oklahoma. Rollins 53123, cu. n = 8: Grady Co., OKMahoma. Rollins 53126, cH. . densiflora (Gray) Watson n = 7: Llano Co., Texas. Rollins 53103, cu. n = 7: Llano Co., Texas. Rollins 5574, cx. 7 Gillespie Co., Texas. Rollins 53106, cu. n = 7: Dewitt Co., Texas. Rollins 5560, cu. : densipila Rollins = 8: Williamson Co., Tennessee. R. & D. Rollins 5215, GH. 8: Williamson Co., Tennessee. Rollins 5315, cH. 8: Williamson Co., Tennessee. Rollins 53137, cu. 8: Marshall Co., Tennessee, Rollins 5321, cu. 8: 8: S323 3 Hdl WT Al Rutherford Co., Tennessee, Rollins 55124, cu. CHROMOSOME NUMBERS OF CRUCIFERAE 53 : Maury Co., Tennessee. Rollins 55146, cu. : Morgan Co., Alabama. Rollins & Chambers 5710, cu. : Morgan Co., Alabama. Rollins 5924, cu. : Morgan Co., Alabama. Rollins et al. 6105, cu. Lawrence Co., Alabama. Rollins et al. 6127, cu. nena b ae losourti = 8 Cheatham Co., Tennessee. Rollins 5326, cu. = 8 Cheatham Co., Tennessee. Rollins 53130, cu. = 8 Williamson Ca. Tennessee. Rollins 5325, cu. = 8 Williamson Co., Tennessee. Rollins 53135, cu. : igen (Gray) Watson = 18: Guadalupe Co., Texas. Rollins 5366, cu. fender (Gray) Watson = 12: Brewster Co., Texas. B. H. Warnock s.n., GH. n = 6: Howard Co., Texas. Rollins 53117, cu. n = 6: Jeff Davis Co., Texas. Rollins 53114, cu. = 6: South of Saltillo, Coahuila, Mexico. Rollins & Tryon 58131, cu. n = 12: Andrews Co., Texas. Rollins & Correll 61149, cu. . filiformis Rollins n = 7: Dade Co., Missouri. Rollins 61158, cu. . globosa (Desv.) Watson n = 7: Davidson Co., Tennessee. Rollins 5312, cH. n = 7: Davidson Co., Tennessee. Rollins 53132, cx. n = 7: Davidson Co., Tennessee. R. & D. Rollins 5213, cx. n = 7: Maury Co., Tennessee. Rollins & Quarterman 55150, GH. oe dH | 90 20 00 0 O mn oe Oa i th i i L. gordonii (Gray) Watson n = 6: Baylor Co., Texas. Rollins 53120, cu. PLATE 3. 2n = 12: Brewster Go. Texas. B. H. Warnock s.n., GH. PLATE a L. gracilis (Hook. ) Watson — = pee Co., Mississippi. Rollins et al. 5644, cH. n = 6: Bryan Co., Oklahoma. Rollins 5970, cx. no GD: Ellis Co. Texas. Rollins 5347, cH. moe B: Kauluian Co., Texas. Rollins 5343 and 5344, GH. n = 6: Leon Co., Texas. Rollins & Correll 5968, GH. n = 6: McLennan Co., Texas. Rollins 5349, cH. n = 6: Williamson Co., Texas. Rollins 5554, GH. Z, delay var. repanda (Nutt.) Payson = 6: Cotton Co., Oklahoma. Rollins 53121, cH. 54 REED C. ROLLINS m i I~ im i~ o ™ Co . grandiflora (Hook.) Watson n 9: Austin Co., Texas. Rollins 5352, cu. n = 9: Austin Co., Texas. Rollins & Correll 5965, cu. n = 9: Dewitt Co., Texas. Rollins 5561, cu. n = 9: Gonzales Co., Texas. Rollins 5363, cu. n = 9: Wilson Co., Texas. Rollins 5364 and 5365, cu. . intermedia (Wats.) Heller 2n = 18: Garfield Co., Utah. Rollins 51200, cu. . lasiocarpa (Hook. ex Gray) Watson n = 7: Cameron Co., Texas. Rollins & Correll 5950, cH. PLATE 1. latifolia A. Nelson n = 5; 2n = 10: Clark Co., Nevada. Clokey 8358, ps. . lescurii (Gray) Watson parses 8: Cheatham Co., Tennessee. Rollins 53131, cu. n = 8: Davidson Co., Tennessee. Rollins 53127, GH. n = 8: Davidson Co., Tennessee. R. & D. Rollins 5209, cu. n = 8: Rutherford Co., Tennessee. Rollins 55174, GH. n = 8: Williamson Co., Tennessee. Rollins 53136, GH. n = 8: Williamson Co., Tennessee. Rollins 55111, cu. lindheimeri (Gray) Watson eS 6: Refugio Co., Texas. Rollins 5562 and 5563, GH. n = 6: Victoria Co., Texas. Rollins 5360, GH. n = 6: Victoria Co., Texas. Rollins 5565, GH. . ludoviciana ( Nutt.) Watson 2n = 10: Moffat Co., Colorado. Rollins & Porter 5115, cu. . lyrata Rollins n = 8: Franklin Co., Alabama. Rollins 5547 and 5548, GH. n = 8: Franklin Co., Alabama. Rollins et al. 5599, GH. = 8: Franklin Co., Alabama. Rollins 55188, GH. n - X maxima Rollins (L. densipila < stonensis) n = 8: Davidson Co., Tennessee. Rollins et al. 5222, cu. = 8: Davidson Co., Tennessee. Rollins 5313 and 53142, cu. = - mevaughiana Rollins 2n = 12: Brewster Co., Texas. Warnock & Turner 8646, GH. n = 12: Pecos Co., Texas. B. H. Warnock s.n. . ovalifolia Rydb., var. alba Goodman n = 6: Caddo Co., Oklahoma. Rollins 53125, cu. nm = 6: Comanche Co., Oklahoma. Rollins 53124, cu. . palmeri Watson n = 95: Pima Co., Arizona. J. Poindexter 1, ps. CHROMOSOME NUMBERS OF CRUCIFERAE 55 of Lesquerella gor- Streptanthus PLATE 2. Upper left, chromosomes in a tapetal cell donii, 2n = 12, s.n.; upper right, chromosomes of cutleri, n = 14, Rollins and Correll 61111; lower, chromosomes of Lesquer- ella purpurea, n —9, Rollins and Correll 61117. All figures x 3900. 56 REED C. ROLLINS L. perforata Rollins n = 8: Wilson Co., Tennessee. R. & D. Rollins 5207, cu. n = 8: Wilson Co., Tennessee. Rollins 5304, 5306 and 53145, GH. L. purpurea (Gray) Watson n = 9: Brewster Co., Texas. Rollins & Correll 6181, cH. n = 9: Hudspeth Co., Texas. Rollins & Correll 61117, GH. PLATE 2. n = 18: Val Verde Co., Texas. Rollins & Correll 6160, cH. L. recurvata (Engelm.) Watson n = 5: Comal Co., Texas. Rollins 5387, cu. n = 5: Gillespie Co., Texas. Rollins 53101, cu. n = 5: Sutton Co., Texas. Rollins 53110, cu. L. sessilis (Wats.) Sma n = 6: Gillespie Co., Texas. Rollins 53100; 53107 and 53109, GH. n = 6: Kimble Co., Texas. Rollins & Correll 5937, cu. L. stonensis Rollins n = 8: Rutherford Co., Tennessee. Rollins 55177, GH. L. subumbellata Rollins 2n = 10: Uintah Co., Utah. Rollins & Porter 5119, cH. L. wardii Watson 2n = 12: Piute Co., Utah. Rollins 51221, cu. Most species of Lesquerella have chromosomes large enough to work with beyond that of merely counting them. However, we have not had cytological study as a goal in itself, hence no at- tempts have been made to characterize individual chromosomes or genomes. An aneuploid series of chromosome numbers between species extends unbroken from n = 5 through n = 9. Fundamental numbers appear to include x = 5, x = 6, x = 7, x — 8 and x = 9 and there are polyploid species or populations based on x = 6 and x = 9. Polyploids based on x = 6 include L. arenosa, 2n = 18; probably L. argyrea, n = 18, from Llano Co., Texas; L. engelmannii, n = 18; L. fendleri, n = 12, one count from Andrews Co., Texas and L. intermedia, 2n — 18. The polyploid population of L. purpurea, n = 18, from Val Verde Co., Texas, appears to be based upon x = 8. The most complex chromosome number situation so far en- countered in Lesquerella occurs in L. argyrea. The taxonomy of what must at present be termed the “L. argyrea complex” is not CHROMOSOME NUMBERS OF CRUCIFERAE 57 at all clear. There may be several taxa present instead of one. Further, it is fairly certain that natural hybridization is a factor in producing the complex taxonomic pattern found. Excellent chromosome number integrity based on x = 8 is shown by the majority of the group of related annual species bearing auriculate cauline leaves (Rollins, 1955). The one certain exception is L. grandiflora with n = 9. L. lasiocarpa, rather doubtfully to be associated in the same subgeneric grouping, has High chromosome numbers have been reported for taxa at extremes of the distribution range of Lesquerella if the early report of Manton (1932) is taken at face value. Her report for L. mendocina was 2n = ca. 50. The one fact that raises a ques- tion in this case is that the species is attributed to Chile and the seed is supposed to have come from a wild plant. If one relies only on undisputed evidence, Lesquerella does not occur in Chile, thus making this particular count slightly open to ques- tion. On the other hand, counts of n = 30 and 2n = 60 (cf. J@rgensen et al. 1958) seem well established for L. arctica. The chromosomes of L. argyrea and L. lasiocarpa are shown in PLATE 1, those of L. gordonii and L. purpurea in PLATE 2, and of L. gordonii in PLATE 3. Lyrocarpa L. coulteri Hooker & Harvey 2n = 20: Pinal Co., Arizona. Nichol 23, vs. Raven (1959) presented a count of n = 20 for L. coulteri var. palmeri (as L. palmeri). As far as I am aware, only two counts have been made in the genus. Polyploidy is obviously present but it would be unsafe to make any assumptions as to the fun- damental number for the genus without further evidence. Nerisyrenia N. camporum (Gray) Greene n = 9: Brewster Co., Texas. Rollins & Correll 6180, GH. n = 11: Torreon-Saltillo, Coahuila, Mexico. Rollins & Tryon 58293, cH. N. linearifolia (Wats.) Greene n = 9: Culberson Co., Texas. Rollins & Correll 61144, cH. ; We now have three separate chromosome numbers for N. 58 REED C, ROLLINS camporum, n = 7 (Rollins, 19392) and the two given above. Plants identified as N. camporum form extensive populations at frequent intervals from trans-Pecos Texas and New Mexico far to the South and West in the Chihuahuan Desert of Mexico. The variation present is extensive and puzzling. The differing chro- mosome numbers is a clue that suggests sexual reproduction is not strictly adhered to throughout the species. The possibility that several taxa are being masked by the presently accepted taxonomy has to be considered also. Physaria P. acutifolia Rydberg 2n = 8: Gunnison Co., Colorado. Ripley & Barneby 10200, GH. P. australis (Pays.) Rollins 2 n = 8: Boulder Co., Colorado. Rollins 5145, cH. 2n = 10: Uintah Co., Utah. Rollins 3091, cH. 2n = 14: Albany Co., Wyoming. Ripley & Barneby 10543, GH . chambersii Rollins 2n = 10: Emery Co., Utah. Rollins 51183, cH. . chambersii, var. membranacea Rollins 2n = ca. 20. Garfield Co., Utah. Rollins 51207, cu. oregona Watson 2n = 8: Idaho Co., Idaho. Ripley & Barneby 10729, cu. Earlier (Rollins, 1939b) it appeared that the chromosome numbers in Physaria would be straightforward, based on x = 4 At least, this assumption could be made if the first three counts of three different species were indicative of the broader picture in the genus. The count of n — 8 for P. vitulifera (Weber and Brewbaker, 1950) did not disturb the assumption that x = 4 is the fundamental number even though polyploidy was then established in the genus. However, the presently reported counts show that a more complex situation exists in P. australis than was shown earlier, and it is clear from counts of P. chambersii that x = 5 must also be a fundamental number in the genus. This does not take into account the odd number of 2n — 14 in P. australis from Albany Co., Wyoming, which shows no relation- ship to the other counts, a) a) ~ CHROMOSOME NUMBERS OF CRUCIFERAE 59 Rorippa oy curvipes Greene n = 8: Gunnison Co., Colorado, Rollins 51172, cu. . curvisiliqua (Hook.) Bessey n = 8: Josephine Co., Oregon. Constance & Rollins 2943, cu. . sinuata (Nutt.) A. S. Hitchcock n = 8: Thomas Co., Kansas. Rollins 5101, cu. . subumbellata Rollins n = 5: Eldorado Co., California. Rollins 3027, cu. The fundamental number x = 8 has become well established in Rorippa with most of the recent counts merely confirming and extending earlier records. A polyploid series exists but deviations from a multiple series were not recorded prior to the present count of n = 5 for R. subumbellata. This disturbance of an other- wise consistent chromosome number pattern in the genus is an- omalous and the significance of it is not known. ww aw as) Selenia S. aurea Nuttall n = 23: Garland Co., Arkansas. Rollins & Chambers 5756, GH. S. grandis Martin n = 12: Dimmit Co., Texas. Barclay 706, cu. Selenia grandis grows very well under greenhouse conditions and we were able to sample the material repeatedly for reas- surance of an accurate count. Material of S. aurea was fixed in the field. These first known counts for the genus suggest poly- ploidy. The disrupted ranges of most of the species provide a basis for an evolutionary pattern that could prove to be exceed- ingly interesting. This is a genus that deserves careful re-study even though it was the relatively recent subject of a paper by Martin (1940). Sibara S. pectinata Greene n = 14: Desierto Viscaino Region, Baja California, Mexico. Gentry 7396, cu. S. virginica (L.) Rollins 2n = 16: Marshall Co., Tennessee. Sharp et al. 11188, cu. 60 REED C. ROLLINS Of the two previous counts in Sibara (Rollins, 1947) 2 n = 26 for S. desertii and 2n = 28 for S. viereckii, the latter fits with the count for S. pectinata. The other two counts show little relation- ship to each other or to the n = 14 number. All species of Sibara, except S. virginica, are limited in distribution and are in- frequently collected. It will probably take many years to resolve what is at present a puzzling series of chromosome numbers in the genus. Sisymbrium S. altissimum L. n = 7: Lake County, California. Breedlove 5134, cu. S. linearifolium (Gray) Payson n = ll: Las Animas Co., Colorado. Rollins 1818, cu. n = 11: Brewster Co., Texas. Rollins & Correll 6168 and 6139, cu. S. linifolium Nuttall n = 7: Albany Co., Wyoming. Rollins & Porter 5113, cu. n = 8: Uinta Co., Wyoming. Rollins 1773, cu. S. orientale L. n = 7: San Diego Co., California. Breedlove 1816, cu. The two counts for the introduced species, S. altissimum and S. orientale are the same as those of most other European species of Sisymbrium. S. linifolium appears to fit the same pattern but the count discrepancy of n = 7 and n — 8 may have some sig- nificance. This species is morphologically very variable and the infraspecific taxonomy requires intensive study for a better un- derstanding than is now available. The very different count of n = 11 for S. linearifolium lends support to taxonomic treatments that place this species outside of Sisymbrium. Stanleya S. pinnata (Pursh) Britton, var. integrifolia (James) Rollins n = 14: Brewster Co., Texas, Rollins 6191, cu. The count of n = 14 does not accord with my previous counts (Rollins, 1939¢) which indicated x — 12 as the fundamental number for Stanleya. On the other hand, n = 14 fits an emerging x = 14 that is widespread in Caulanthus, Thelypodium and Strepthanthus, genera somewhat related to Stanleya. CHROMOSOME NUMBERS OF CRUCIFERAE 61 Strepthanthella WN . longirostris (Wats.) Rydberg 2n — 28: Sweetwater Co., Wyoming. Rollins & Porter 5144, GH. longirostris, var. derelicta J. T. Howell n = 14: San Diego Co., California. Breedlove 1865, cu. Although Streptanthella is usually given the status of a mono- typic genus in current manuals and floras, it is by no means cer- tain that this is the correct taxonomic interpretation. It is prob- able that the one species, S. longirostris, should be associated with such species as Caulanthus cooperi but perhaps not in the genus Caulanthus. The chromosome number does not contribute any- thing toward solving the problem. a Streptanthus a barbiger Greene n — 14: Lake Co., California. Breedlove 5145, cH. . breweri Gray n — 14: Colusa Co., California. Breedlove 5181, cu. n = 14: Napa Co., California. Breedlove 5088, GH. . carinatus Wright n = ca. 14: Brewster Co., Texas. Rollins & Correll 6178, GH. n — 14: Presidio Co., Texas. Rollins & Correll 61105, cH. PLATE 3. . cordatus Nuttall n —12: Mohave Co., Arizona. Rollins 4166, cH. . cutleri Cory n — 14: Brewster Co., Texas. Rollins & Correll 61111, cH. PLATE 2. . diversifolia Watson n = 14: Fresno Co., California. Breedlove 5270, GH. glandulosa Hooker n = 14: Lake Co., California. Breedlove 5158, GH. n = ca. 14: Santa Clara Co., California. Breedlov GH. . insignis Jepson n ty te ee Co., California. Breedlove 2375, GH. n — 14: San Benito Co., California. Wiggins & Rollins 24 and 34, GH. i) nN N 2) iS) wa e 4986, 2) 62 REED C. ROLLINS eo A. * a 2. *e @ : es . bl y + Pirate 3. Left, chromosomes of Lesquerella gordonii, n = 6, Rollins 53120; upper right, chromosomes of Streptanthus carinatus, n = 14, Rol- lins and Correll 61105; lower right, chromosomes of Selenia grandis, n = 12, Barclay 706. -All figures x 3900. S. niger Greene n = 14: Marin Co., California. Breedlove 4962, GH. . secundus Greene n = 14: Sonoma Co., California. Constance & Rollins 2863, GH, . tortuosus Kellogg n = 14: Siskiyou Co., California. Constance & Rollins 2901, GH. n = 14: Tuolumne Co., California. Breedlove 4830, GH. There is near uniformity of the chromosome number n = 14 in Streptanthus. The known exception is S. cordata with n = 12. Further counts are needed in the group to which S. cordatus wn Ww CHROMOSOME NUMBERS OF CRUCIFERAE 63 belongs. The species involved were segregated from Streptan- thus and placed in the genus Cartiera by Greene (1906) and by Schulz (1936). The one count pointed to here suggests the pos- sibility of chromosome number support for such a separation. However, the uniformity elsewhere in Streptanthus does not support a wholesale breaking up of the genus as attempted by Greene (l.c.) and followed by Schulz (l.c.). Chromosomes of S. cutleri are shown in PLATE 2. Those of S. carinatus are shown in PLATE 3. Synthlipsis S. greggii Gray n = 10: Northeast of Durango, Durango, Mexico. Rollins & Tryon 58280, cH. This is the first count in Synthlipsis. The three known species (Rollins, 1959) are found mainly in Mexico. Thelypodium T. flavescens (Hook.) Watson n = 14: Monterey Co., California. Breedlove 2180, cu. T. flexuosum Robinson n = 13: Harney Co., Oregon. Raven 18452, GH. T. laciniatum (Hook.) Endlicher, var. milleflorum (A. Nels.) Payson n = ca. 14: Eureka Co., California. Raven 18533, cH. T. laciniatum, var. streptanthoides (Leiberg) Payson n = ca. 12: Grant Co., Washington. Raven 18487, GH. T. lemmonii Greene n = 14: San Benito Co., California. Wiggins & Rollins 36, GH. T. texanum (Cory) Rollins n = 13: Brewster Co., Texas. Rollins & Correll 6188, cH. n = 13: Brewster Co., Texas. Rollins & Correll 6176, CH. A relatively high fundamental number (or numbers ) 7 ia erging for Thelypodium. This finding is in general accor = the presence of similar numbers in related genera such as Stan- leya and Caulanthus. The counts given for T. oni es tentative because good figures for counting could not be foun in the material available for study. 64 REED C. ROLLINS Thlaspi T. fendleri Gray 2n = 14: Mt. Ord, Brewster Co., Texas. B. H. Warnock s.n., GH. 2n = 28: Hinsdale Co., Colorado. Rollins 51107, cu. . glaucum A. Nelson 2n = 14: Douglas Co., Colorado. Rollins & Livingston 5148, GH. — T. parviflorum A. Nelson n = 7: Sheridan Co., Wyoming. Williams 3092, cx. Thlaspi is widely distributed in the Northern Hemisphere but also occurs sparingly in temperate areas of the Southern Hem- isphere. The genus is relatively well-marked and is readily dis- tinguishable from others of the family. The chromosome number too is relatively uniform, based on x — 7. Our material shows polyploidy to be present in T. fendleri. Tropidocarpum T. gracile Hooker n = 8: San Diego Co., California. Breedlove 1822, GH. n = 8: Santa Barbara Co., California. Breedlove 1904, cH. Thysanocarpus T. curvipes Hooker n = 7: Humboldt Co., California. Constance & Rollins 2884, CH, T. elegans Fischer & Meyer 2n = 28: San Luis Obispo Co., California. M. P. & A. G. Vestal s.n., GH. Manton (1932) gave a count of 2n — 28 for T. curvipes. If all identifications are correct, this means polyploidy is present within T. curvipes. The presence of 2n — 28 in T. elegans proves that multiple chromosome numbers are present in the genus, at least. The taxonomy of Thysanocarpus is very much in need of a careful study and revision. LITERATURE CITED B6cHER T. W. 1951. Cytological and Embryological Studies in the Amphi- Apomictic Arabis holboellii Complex. Biol. Skr. Dan. Vid. Selsk. xxxx. 6: 1-59 Greene, E. L. 1906. Four Streptanthoid Genera. Leafl. Bot. Obs. and Crit. 1: 226-28. CHROMOSOME NUMBERS OF CRUCIFERAE 65 Hoimen, K. 1952. Pipa Studies in the Flora of Peary Land, North Greenland. Medd. SEs 8 JORGENSEN, CA. TH. proans AND M. WESTERGAARD. ee Png Flower- ing Plants of Greenland. Biol. Skr. Dan. Vid. Selsk. 9 Lewis, H. 1959. Documented Chromosome Numbers es mae "Madrofio 15: 49. Manton, I. 1932. Introduction to the Cytology of the Cruciferae. Ann. Bot 46: 509- Mutuican, G. A. 1964 Chromosome Numbers of the Family Cruciferae. 1. Can. ph Bot. 42: 1509-19. -—--—— . Chromosome Numbers of the Family Cruciferae. III. Can. Jou oe 44: 309-319. Pssios. E. 1964. The Genus Cakile. Novitates Systematicae Plantar- um Vascularium 90-128. Raven, P. H. 1959. Documented Chromosome Numbers of Plants. Madrono 15:49. Raven, P. H., D. W. KyHos anp Avsert J. Hiv. 1965. Chromosome Num- bers of Spermatophytes, Mostly Californian. Aliso 6: 105-113. Rowuiys, R. C. 1939a. Notes on Certain Cruciferae of Mexico and South- western United ene Madrono 5: 129 39b. The Crecchiace Gaus Physaria. Rhodora 41: 391-414. _ 1939c. The Cruciferous Genus Stanleya. Lloydia 2: 109-27. ie ee Study of Arabis in Western North America. Rhodor 43: 3034 Some Si or Noteworthy North American Cruciferae. II. ar Dudley Herb. 3: 366-73. 1947. Generic Revisions in the Cruciferae: Sibara. Contr. feay Herb. spi 133-43. 955. The Auriculate-leaved Species of Lesquerella ( Cruciferae ). is 57: 241-64. 1959. The Genus Synthlipsis (Cruciferae). Rhodora : 253-64. 1960. Some Sisymbriums (Cruciferae) Native to Texas and Northisastarn Mexico. Rhodora 62: 55-60. 1962. A New Crucifer from the Great Slave Lake Area of Canada. ‘Rhodora 64: 324-7. SaunTE, L. H. 1955. Cytogenetical Studies in the Complex. reg 41: 499-515. Scuu.tz, O. E. ee ae Pflanzenr. 86: 159-63. 6. Cruciferae. Die Natur. Pflanzenfam. 17b: 303- Cochlearia officinalis 4. Snow, R. 1959. Chromosome Numbers of California 8180, with Notes on Some — of Cytological Interest. Madrono 15: 81-89 Weber, W. A. anv J. L. Brewsaker. 1950. hysaria_vitulifera, a Tetra- uid | Species of Cruciferae. Univ. Colo. Studies, Biol. 1: 24-