Ge . CS ¥Gs [7732-°7¢ A es stp ‘ 4 . Se me 755 ee. be GRAY HERBARIUM OF HARVARD UNIVERSITY CONTRIBUTIONS 203 - 20 1973-74 Missouri: Boranicat \. GARDEN LisRansy Contributions from the GRAY HERBARIUM Reed C. Rollins a Otto T. Solbrig Arturo Gémez-Pompa d an Lorin I. Nevling, Jr. Tod F. Stuessy Gerald J. Gastony EDITED BY _ Reed C. Rollins 2 ee Kathryn Roby } | yee) 185 PUBLISHED BY ae . THE GRAY HERBARIUM OF HARVARD UNIVERSITY 1973 NO. 203 j INTERSPECIFIC HYBRI LESQUERELLA Z THE USE OF ELECTRO ! METHODS IN THE FLORA O PROGRAM owe 4 A SYSTEMATIC. REVIEW / OFC MELAMPODIINAE {COMER Contributions from the GRAY HERBARIUM 1973 NO. 203 Reeee Rollins INTERSPECIFIC HYBRIDIZATION IN Otto T. Solbrig LESQUERELLA METHODS IN THE FLORA OF VERA CRUZ Arturo Gémez-Pompa THE USE OF ELECTRONIC DATA PROCESSING and Lorin I. Nevling, Jr. PROGRAM Pod ol A SYSTEMATIC REVIEW OF THE SUBTRIBE eg Gen: MELAMPODIINAE (COMPOSITE, HELIANTHEAE) Gerald J. Gastony A REVISION OF THE FERN GENUS NEPHELEA EDITED BY Reed C. Rollins Kathryn Roby PUBLISHED BY THE GRAY HERBARIUM OF HARVARD UNIVERSITY Issue January 4, 1973 INTERSPECIFIC HYBRIDIZATION IN LESQUERELLA? REED C. ROLLINS AND Otto T. SOLBRIG The full ramification of interspecific hybridization as a mechanism for recombining the genetic materials of existing species, and the role this phenomenon may have played in the evolution of the present world biota is still without a complete and satisfactory formulation. This is true, in spite of the considerable research and the numerous publications on the subject. The research reported below contributes to an understanding of hybridization and its effects on populations. Frequently, motivations for the study of wild hybrids have been associ- ated with taxonomic problems that arise because of the complexity intro- duced by interspecific hybridization itself. However, in most of the auriculate species of Lesquerella such a problem does not exist. The wild hybrids participate in an existing situation that does not, at present, complicate the taxonomic picture of the species involved. The fact that populations of wild hybrids exist and are readily available for study, and that interspecific crosses can easily be made, makes certain of these species of Lesquerella desirable research material for examining the nature of species differences. We have been interested in various aspects of the genetic bases for the differences between species, and the way in which these differences are correlated with each other, both within and between species. The utilization of data from experimentally produced hybrids is an obvious way to approach these problems. During the course of our study, we have again been impressed with the striking, dynamic changes that occur in populations of hybrids, espe- cially in the Harpeth River valley. Some attention has been given to these aspects of the problem because of the general bearing they have on ways in which hybrids may participate in the recombination and dispersion of the genetic components of different species. We are aware of the possibility that entirely different taxa might arise and become stabilized at remote distances from the hybridizing parental species, a possibility suggested in a dramatic way by the Harpeth River valley populations. However, it should be emphasized that this has not occurred among the Lesquerellas of the area up to the present. ACKNOWLEDGEMENTS We have been assisted in the laboratory, greenhouse and experimental garden by Sharon Horn, Margaret Kunz, Mary Maly, Esperanza Vega and Roger Yocum, each o whom we thank very much for their committed efforts. We especially acknowledge the 1The first paper with this title, Interspecific Hybridization in Lesquerella (Cruciferae) by Reed C. Rollins was published in Contrib. Gray Herb. no. 181:1—40. 1957. 3 4 ROLLINS AND SOLBRIG help given by Dr. William Bossert with computation and pioerenmine problems. We ha crosses were grown. Dr. James Farris permitted us to use his Prim program deck for the — of linear similarity coefficients, which we very much appreciate. earch supported by National Science Foundation grant GB-2019 to Reed C. Rollins, Principal Investigator. OBJECTIVES Rollins (1957) examined the extent and nature of interspecific hybrid- ization among wild plants of the auriculate-leaved species of Lesquerella in Tennessee. It was established that there are virtually no genetical barriers between the four species of Lesquerella examined that were endemic to the Central Basin: L. stonensis, L. perforata, L. lescurii and L. densipila. Presumably these species arose in isolation from each other and have persisted allopatrically until recent times when, with man’s help, three species have come together to produce interspecific hybrids. Lesquerella densipila and L. lescurii hybridize near the junction of Arring- ton Creek and the Harpeth River (Fig. 1). From that point down river for a distance of over 40 miles, there are numerous populations of hybrids on the flood plain of the Harpeth. Lesquerella densipila hybridizes with L. stonensis at the junction of the East Fork and the West Fork of Stones River (Fig. 1). From that point down river, only hybrid populations are to be found. In both these instances of interspecific hybridization, the hybrid populations occur downstream from the sites of the parental populations. This is because local migrations of the species are strongly affected by seasonal flooding of the streams of the Central Basin, the seeds being carried downstream. This unidirectional movement, particularly of the interspecific hybrids, is an important factor in maintaining the relative purity of the parental species populations. The tendency is for the species to move downstream until they come together, and for the resulting hybrids to continue to move downstream away from the areas where the parent species exist on the upper tributaries. Such a situation is ideal to test several features of the hybrid popula- tions and other aspects of the hybrids and parental species that are of more general interest. These include: (1) Is the variation of the hybrid popula- tions greatest near the area where they are formed, or is it more or less uniform throughout? In other words, is natural selection favoring certain genotypes among the array present in the hybrid swarms in a way that the favored genotypes are more numerous farther downstream where presum- ably the influx of genes from the parental species is less intense? (2) What are the fluctuations exhibited by the hybrid swarms from year to year? That is, is the gene flow from both parental species uniform over time? (3) What is the inheritance mechanism of the characters by which these species are distinguished? (4) In these natural populations, do we en- counter the phenomenon of “genetic coherence” (Clausen and Hiesey, HYBRIDIZATION IN LESQUERELLA WILSON COUNTY ' ' ' N ' t) ‘ oe Sg ESN ice, ager oe ‘ ‘i Hy be, H ' 3 : %, t = Nashville»! ' vee Py Teena ; S \ ' Py Sort % 5 ' "st ‘ ' 7) ‘, - \ H ‘ %, 4 hN aire ; : W DAVIDSON COUNTY vas Pe Fe ' . - ' ! a a8 12 ' 1 ee Bi Ni ee ae ; Bi ' ‘ Py ee te a . i , ‘ - i ' ' * ~ , ‘ ” * ' ‘ “ 10 é : ‘ Pe “te / . , RY ts ‘< rs - ' be ’ 1 wa ‘ . ’ ' my 7 7 ' oT : i 4 ” Fong . : % MH ~ H ; 4 1 * r % ‘ 2 (8 . ' % U , Pegs 6 ot : ' oh ‘ ' WILLIAMSON COUNTY % 4 wt) i, RUTHERFORD COUNTY ! ‘ 2 : "big \ Baers ORG ! U ae Fic. 1. Map of central Tennessee showing the collecting sites of Lesquerella densipila *« L. lescurii hybrids along the Harpeth River and L. densipila x L. stonensis along the Stones River. ipil 1960), a mechanism that favors the persistence of the parental genotypes in a greater frequency than would be expected if free recombination were operable? Hybrids of the third combination, Lesquerella lescurii x L. stonensis, have only recently been discovered in LaVergne, Tennessee. These hybrids have not been studied in detail. The population is on a vacant lot in the town of LaVergne and is not associated with a stream system. PROCEDURE AND METHODS Progenies of wild populations (see Table 12 for localities of seed sources of five species of auriculate-leaved Lesquerella—L. stonensis, L. perforata, L. lescurii, L. densipila, L. lyrata) were grown in the green- houses in Cambridge, Mass. during the winter and spring of 1961, and crossed in all 10 possible combinations. Nine of these were successful, the exception being the cross between L. densipila L. lescurii. The following year F, progenies were grown in the greenhouse and sibling crosses per- formed. Since these species of Lesquerella are self-incompatible, no selfing was possible and sibling pollinations were required for viable seed set. There were some instances where cross incompatibility between individual 6 ROLLINS AND SOLBRIG plants was present. When this was discovered in any particular pairing, a different set of individuals was used. Occasionally cross incompatibility was present between two individuals in one direction only. In these cases, reciprocal pollinations were ineffective, but the progenies arising from unidirectional pollination were utilized. In the spring of 1963, some 5,000 plants of the F,, progenies of these crosses were grown in the experimental garden in Cambridge. Because of the previous failure, progenies of L. densipila and L. lescurii from newly acquired seeds were again grown in Cambridge in 1965 and hybridized; the F, seeds were produced the following year and some 1,500 progeny plants grown in the spring of 1967 at the Mathei Botanical Garden at the University of Michigan in Ann Arbor. The individual F,, plants were scored for flower color in the field, and when in fruit, samples were collected from each plant and taken to the laboratory where they were measured for additional characters (Table 1). These data were recorded on tally sheets and then punched on computer cards for further manipulations. Twenty-eight accessions (of approximately 200 plants each) of natural hybrids of Lesquerella densipila L. lescurii from 19 different localities and collected in one of five different years (1953, 1955, 1964, 1965 and 1966) spanning a 13 year period were also measured for the same 23 characters as the controlled hybrids and processed in the same way. Two sites are represented by four collections, three by two, and 14 by one. In addition, 12 samples of natural hybrids of L. densipila « L. stonensis were gathered and measured. These were from eight sites, and four collection years (1955, 1961, 1965 and 1966). Two sites were sampled in three different years, and six in only one. Finally, the following number of wild parental populations were sampled and measured in the same way as the artificial and wild hybrids: L. densipila, five populations over 13 years; L. lescurii, four populations over 13 years; L. stonensis, four populations over nine years; L. lyrata, two populations, both collected in 1955; and L. perforata, five populations over 11 years. For every artificially produced F,, progeny, as well as for every field sample, the mean, the standard deviation and the coefficient of variation of 23 characters was calculated. These are discussed in a subsequent section. In addition, simple correlation coefficients were calculated for every pair of characters for each F, progeny and each field sample. The results of these analyses can be found under the section Character Corre- lation. The calculations were performed on the IBM 7090 computer at Harvard University using programs written by one of us (OTS). Finally, a linear similarity coefficient between populations was calcu- lated according to the Prim diagram, using a program written by Dr. James Farris, and run on the IBM 7090 computer at the University of Michigan. The results of this analysis are discussed under the section on geograph- ical and temporal variation of natural hybrids. HYBRIDIZATION IN LESQUERELLA t TABLE |. CHARACTERS AND CHARACTER STATES Character Number — Character Character States 1 flower color l=white; 2, 3, 4=graded intermediates; 5=yellow 4 style length measured to nearest } mm. 3 replum length measured to nearest | mm. 4 replum width measured to nearest } mm. 5 silique width measured to nearest 0.1 mm. 6 septum perforation 0=none; 1=<3}; 2=3; 3=>3 3 replum apex 0=rounded; 1=angular 8 replum base 0=rounded; 1=angular 9 trichomes at style base 0=absent; 1=present 10 trichomes on valve outer surface 0=absent; 1=present ll trichomes on valve inner surface O0=absent; 1=present 13 trichomes on replum =absent; 1=present 13 bulbous simple trichomes on valve exterior 0=absent; 1=present 14 non-bulbous simple trichomes on valve exterior 0=absent; 1=present 15 branched non-stalked trichomes on valve exterior 0=absent; 1=present 16 branched stalked trichomes on valve exterior 0=absent; 1=present 17 type of trichomes on valve 0=none; 1=simple; 2=branched; inner surface = simple and branch 18 silique apex 0=angular; l=rounded; 2=depressed 19 length of pedicel measured to nearest mm. 20 number of pedicels in 5 cm. f raceme actual count PA trichomes >0.48 mm. long on valve exterior 0=absent; 1=present 22 trichomes 0.16—0.48 mm long on valve exterior 0=absent; 1=present 23 trichomes <0.16 mm. long on valve exterior 0=absent; 1=present FIELD SETTING The Central Basin of Tennessee is the residuum of a former giant lime- stone dome. Many of the resulting Ordovician limestone outcroppings form cedar glades throughout the basin. These glades are the favored sites of four sympatric species of Leavenworthia and they have been described in connection with studies of that genus (Rollins, 1963; Lloyd, 1965). The principal streams traversing this basin are the Stones and Harpeth rivers which arise in the eastern and south-central portions and flow northward into the Cumberland River. Cutting across the southern portion of the basin is the Duck River which flows almost directly west to join the Tennessee River. The Cumberland River transects the basin along its northern edge and flows westward for some distance past the limits of the basin before turning north. : The present distribution of the four auriculate-leaved species of Les- 8 ROLLINS AND SOLBRIG querella in Tennessee is related to the streams, on the one hand, and to the cedar glades on the other. The distribution of each species is given in turn. The most restricted species is L. perforata, so far known only from the vicinity of Lebanon and a few miles to the north near Spring Creek, both localities in Wilson County. However, the presently known distribu- tion may reflect, in part at least, the fact that no concerted effort has been made to extend the range by carefully searching for other occurrences of this species, especially downstream from the Spring Creek locality. Lesquerella perforata is on disturbed land in a glade area near Lebanon, but it is possible this is a recent invasion from elsewhere. It is not certain that this taxon is naturally in a glade situation but the population on Spring Creek confirms that it is at least associated with a stream in one area of its occurrence. Lesquerella stonensis is mostly associated with the East Fork of Stones River, occurring on the flood plain and adjacent areas, but it is also known from one location about one and one-half miles east of the river in an uncultivated cedar glade-like area. The species proper, unadulterated from hybridization, is at present known only from within Rutherford County. The most widespread species of the group is Lesquerella densipila but the present distribution appears to be somewhat more extensive than its pristine range. In the Central Basin it is abundant in the area of the upper Harpeth River, along the Duck River and on the upper portion of the West Fork of Stones River. There are many sites that are essentially cedar glades even though they are somewhat modified from the original by pasturage or other land use. Lesquerella densipila also occurs in south- ern Tennessee and in northern Alabama. Since our first observations of it in 1952, this species appears to have increased its geographic area, perhaps being carried about in hay or by farm machinery. Lesquerella lescurii is common in the northern portion of the Central Basin, especially in the area of Nashville, where it is found most com- monly on cedar glade sites. However, it has spread along the Cumberland River bottom up to the “rim” of the basin at some distance northwest of Nashville and is found on several tributaries of the Harpeth River. There is evidence that L. lescurii is actively expanding its range at the present time both by natural means and through man’s activities. It is clear that, basically the above four species of Lesquerella are allopatric in their distributions. Each appears to have evolved in relation- ship to one of the streams of the Central Basin, perhaps not in the stream valley itself but more likely in the glade-like areas exposed by the erosion of the land surface by small tributaries arising on high ground. Lesquerella perforata is associated with Spring Creek, L. stonensis with the Stones River, L. densipila with the Duck River and L. lescurii with the Cumber- land River. The fact that three of the four species have now come together, and produced interspecific hybrids does not, in our judgment, negate the HYBRIDIZATION IN LESQUERELLA 9 basic pattern of allopatry. Rather, this is explainable on the basis that the ranges of the species have become altered in recent years, most probably through intensive agricultural activities in the area. Lesquerella lyrata is restricted to a few known sites in the glade region of northern Alabama. The only other Lesquerella in Alabama is L. densi- pila, somewhat to the east of the range of L. lyrata. We believe L. densipila is probably a recent introduction into Alabama from Tennessee. In any case, the ranges of L. lyrata and L. densipila do not overlap. DESCRIPTION OF THE CHARACTERS Because of the convenience and ease of sampling and because the vegetative parts of all five species are quite similar, a large proportion of the characters measured or scored were those concerned with the siliques and infructescences. But there are other reasons for concentrating on fruit characters, as here many of the specific differences can be seen. Develop- mentally, the siliques reach a state of maturity quite rapidly and then remain the same for a relatively long period. Thus, problems of size changes associated with development were avoided by always sampling the oldest fully mature siliques. In this way, a comparability of samples was achieved that would be difficult to accomplish otherwise. The char- acters and the coding indicators are given in Table 1. However, some further comments on several of the characters are required for clarity. Flower color. The two basic patterns are all yellow with darker guide- lines in the center of the flower and all white except for the yellow center in which the guidelines are the deepest yellow. The coloring of the yellow- flowered species, Lesquerella lescurii, L. lyrata and L. densipila, is similar with only minor variations in the intensity of the yellow. The most intense color is in the center of the flower and this tends toward orange. White flowers are characteristic of L. perforata and L. stonensis. Crosses between species of the same flower color result in progeny with the same color in both F, and F,. However, crosses between yellow- and white-flowered species produce an intermediate flower color in the first generation and then segregated forms ranging from nearly white to nearly pure yellow among the F, individuals. In scoring these hybrids and those from field populations, we used 1 and 5 for the white and yellow respectively, 3 for the median intermediate, 2 for individuals tending toward white from the median and 4 for those tending toward yellow from the median. Style length. In Lesquerella as a whole, the length of the style differs considerably from species to species. However, among the auriculate- leaved species this is not the case. Measurements were taken mainly to test whether this easily taken statistic could be utilized in any definite way. Replum length. The replum, which can be readily exposed by taking off one of the valves, is easily measured. The length is also a measure of the silique length and important in showing the overall shape of the silique. 10 ROLLINS AND SOLBRIG Replum width. This second dimension of the replum provides the basis for knowing the thickness of the silique as well as the shape of the replum itself. Silique width. Of the species under consideration, Lesquerella lescurii with its flattened siliques is at one extreme and L. stonensis, with some- what distended valves is at the other. This dimension discriminates L. lescurii from the other species which have fruits nearly alike. eptum perforation. In most species of Lesquerella the septum is with- out any perforation. When a perforation is present (Fig. 2B) and of small size, it is usually at or near the center of the septum. The most extreme case of septum perforation is that of L. perforata, where nearly the entire septum is missing (Fig. 2C). This feature is less pronounced in L. ston- ensis but there is usually a hole present in the septum of this species. The other species of the group usually have imperforate septa. Replum apex and base. The angle at which the replum apex meets the base of the style varies from rounded, as in L. densipila, to angular, as in L. lescurii. This is true also at the base of the replum where it joins the receptacle. These two features (Fig. 2A, B) were scored separately for the apex and base of the replum. Trichomes: presence, absence and type. A number of variations in the types of trichomes and in their presence or absence occur among the auriculate species of Lesquerella in such a way as to offer discriminating a silique opex depressed silique apex rounded ee replum base t ‘ ' ' ! ' ' ' rounded Le Ssilique Ci width replum apex angula silique apex = angular replum base angular se F ptum perforation > 1/2 Fic, 2. Sketches of Lesquerella siliques. A. Replum without a septum perforation and with an replum length measurement limitations. B. Replum showin: an i replu i ™m gZ se . D. Silique showing depressed apex. E. Silique showing rounded apex and measurement limits for silique width. F. Silique showing angul. apex. HYBRIDIZATION IN LESQUERELLA 11 combinations. The extreme example of the lack of trichomes on the silique is L. lyrata where all parts are completely glabrous. The siliques of L. densipila are densely covered with simple trichomes while those of L. lescurii have two types of trichomes on the exterior as well as branched trichomes on the interior. In each of the character states separately scored, at least one species of the group features the alternative state. For example, trichomes are not at the style base in L. lescurii, whereas they are nearly always present in L. densipila and L. stonensis. Similarly, trichomes are always present on the interior of the valves in L. perforata but they are absent in L. stonensis. Of the simple trichomes found among the five species, the most extreme are the large, almost acicular bulbous- based type found in L. lescurii and the minute unthickened ones found in L. densipila. Branched trichomes are present on the valve surfaces of L. lescurii and on the interior of the valves of L. perforata. Silique apex. Whether the silique apex is angular (that is, more or less acute), rounded, or depressed (Fig. 2D, E, F) when viewed from the replum margin was recorded. Lesquerella lescurii has angular siliques while those of L. densipila and L. lyrata are rounded. Some siliques of L. perforata and nearly all of L. stonensis are depressed. Other characters. The last five characters listed in the tables are de- signed to show up differences in length of pedicels and trichomes and to give a measure of the density of fruits in the infructescences. The tri- chomes measured were those on the silique surface. ARTIFICIAL Hyprips First we will briefly describe the salient features of the artificial crosses. Since the analysis comprises a total of 5,400 plants, including 1,530 parental species individuals, 3,170 F., and 600 backcross progeny in 71 different genetical families (23 parental, 60 F, and 8 backcross) only a summary emphasizing the relevant aspects can be presented here.? Tables 2 to 11 list the range and mean for the 23 characters measured for each of the crosses. The comparison is between one particular artificial F, hybrid population grown in the garden and the sum of all populations from the wild of the parental species. Because the F, and the parents grew in different environments, a definite environmental effect is to be expected. An inspection of Tables 2 to 11 reveals that not all species differ in all 23 characters. When they do differ in a character, the hybrid may show a value intermediate between that of the parents, equal to that of one parent but not the other, or it may transcend the values exhibited by either parent. Out of a total of 230 individual pair-by-pair comparisons (23 characters in 10 crosses), 40 pairs show similar values for parents and 2The score sheets and detailed analysis have been filed in the —— of the Gray Herbarium together with this report and are available for scrutiny by interested indivi 12 ROLLINS AND SOLBRIG hybrids. Of the remaining 190, 96 (about 50%) show values that are intermediate between those of the parents, 25% have values that closely approximate one parent, and 25% have values transcending both parents. ertain characters, such as shape of the replum apex and base or length of pedicel, have a greater tendency to deviate from the expected inter- mediacy. Most of these are size characters. Since these are characters showing the greatest response to the cultural conditions, most, if not all, of the apparent dominance has to be attributed to luxuriance and not to genetic overdominance. Further inspection also reveals that the greater the magnitude of the difference in the two parental values, the greater the probability that the hybrid will be intermediate. If the environmental effect is fixed in its magnitude rather than proportional to the difference, the above observation is expected. Consequently, it can be assumed that the hybrids are intermediate in most, if not all, their characters. The lack of data on the actual parents and F, hybrids precludes drawing any conclusions regarding the genetic mechanism that controls the char- acters measured, other than the obvious one that none is controlled by a simple, 1- or 2-locus system. However, it is probable they are all inherited in a multigenic fashion, including the pubescence characters. LESQUERELLA DENSIPILA L. LESCURII—FIELD POPULATIONS The series of hybrid swarms produced by these species are by far the most interesting because of their extension over a distance of more than 70 TABLE 2. DATA FROM L. PERFORATA, L. STONENSIS AND THEIR F, PROGENIES perforata perforata X stonensis stonensis N=181 N=—212 N=216 Min Max x Min Max x Min Max .48 mm 0.11 043 0.18 0.00 0.31 0.14 0.00 0.02 0.01 nik: (how 048 0:71 055 -- 053 090 O77 098-100: 100 trich.<.16 0.17 0.68 0.34 0.77 1.00 0.84 098 1.00 1.00 HYBRIDIZATION IN LESQUERELLA 13 miles, from the junction of the upper Harpeth River and Arrington Creek in Williamson County to the junction of the Harpeth with the Cumber- land River in Cheatham County. We have sampled a total of 18 localities over this area and over a period of 13 years (Fig. 1 and Table 13). The hybrid swarms are usually found in alluvial flood plains on the upper side of long bends of the river. These areas are sometimes covered with the natural vegetation of such areas; in other cases they are pastures or are regularly cultivated. Our observations indicate that the populations are capable of maintaining themselves at least over a period of years. However, they show marked changes in numbers and occasionally also in resemblance to one or the other parent species, indicating that new seed is arriving at the sites from places upstream, and that the populations can be wiped out by unfavorable conditions. A good example is the population at the junction of Arrington Creek and the Harpeth River (Site 1). In 1955 this population was estimated by one of us (Rollins, 1957) to occupy an area of approximately 600 acres, and the plants to be intermediate in their characteristics to the two parental species. By 1964, the population area had been reduced considerably and the plants now resembled Lesquerella densipila more than L. lescurii. In 1966, an extensive search resulted in a find of less than 10 plants. The area by then had been con- verted into pasture land. Whether this observation is representative on how transient these populations really are is hard to ascertain. What we TABLE 3. DATA FROM L, PERFORATA, L. LYRATA AND THEIR F, PROGENIES perforata perforata X lyrata lyrata =) N=608 N=130 Min Max x Min Max x Min Max x flower color 1.00 1.00 1.00 1.95 2.54 2.19 5.00 5.00 5.00 style len 2.00 2 2.21 1.79 229 2.00 Lil: tt 14 replum length 3.75 4.79 4.19 440 5.02 4.69 3.22 3.38 3.30 replum width 3.85 5.00 4.32 3.57 440 3.94 3.29 3.46 3.37 silique width 4,34 5.43 4.78 449 498 4.80 3.62 3.65 3.63 septum perfor. 3.00 3.07 3.01 142 194 1.74 0.00 0.03 0.02 replum apex 0.07 0.75 0.51 0.63 0.75 0.71 0.00 0.02 0.01 eplum base 0.00 0.55 0.23 0.31 0.71 0.52 0.00 0.00 0.00 trich.: style-base 0.00 0.02 0.01 0.00 0 00 0.00 0. 00 trich.: out. valve 0.64 0.93 0.73 0.02 0.13 0.04 0.00 0.07 0.03 trich.: inn. valve 1.00 1.00 1.00 0.59 0.73 0.67 0.00 0 trich.: replum 0.00 0.59 0.31 0.00 0.01 0.00 0.00 0.07 0.03 trich.: bulbous 0.00 0.21 0.06 0.00 0.01 0.00 0.00 0.00 0.00 trich.: nonbulb G12: 0 0.33 0, 0.01 0.00 0.00 0.07 0.03 trich.: nonstalk 0.29 0.68 0.36 0.00 0.08 0.03 0.00 0.00 0.00 trich.: stalke 0.30 0.71 0.48 0. 0.11 0.02 0.00 0.00 0.00 trich. type: inn. 2.00 2.00 2.00 1.20 1.50 1.36 0.00 0.00 0.00 silique apex 1.89 2.00 1.93 121 162 1.33 1.82 1.83 1.82 icel length 0.74 1.03 0.93 112 126 148 0.88 1.00 0.94 pedicel no tii 167 1.38 146 1.71 1.60 Lil 150 13% ch.>.48 m 1 043 0,18 0.00 0.01 0.00 0.00 0.00 0 ROLLINS AND SOLBRIG TABLE 4, DATA FROM L. PERFORATA, L. LESCURIL AND THEIR F, PROGENIES flower color trich.<.16 flower color trich.> 48 mm. trich. .16—.48 mm. perforata perforata X lescurii lescurii N=181 N=84 N=324 Min Max x Min Max x Min Max x 1.00 1.00 1.00 1.78. 2.27 2.03 5.00 5.00 5.00 2.00 2.50 2.21 1.88 2.28 2.08 143 1.74 1.57 3.75 4.79 4.19 4.98 5.63 5.31 4.50 5.04 4.68 3.85 5.00 4.32 4.06 4.30 4.18 3.36 3.88 3.75 4.34 543 4.78 3.30 3.72 3.50 151 2.01 1.68 3.00 3.07 3.01 1.38 1.62 1.50 0.01 0.08 0.03 0.07 0.75 0.51 0.71 0.67 0.69 0.61 0.93 0.67 00 0.55 0.23 0.12 0.32 0.21 0.00 0.77 0.29 0.00 0.02 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.64 0.93 0.73 0.88 0.98 0.93 1.00 1.00 1.00 1.00 1.00 1.00 0.98 1.00 0.99 0.86 1.00 0.95 0.00 0.59 0.31 0.00 0.00 0.00 0.00 0.05 0.01 0.00 0.21 0.06 0.59 0.72 0.65 1.00 1.00 1.00 0.12 0.50 0.33 0.51 0.19 0.35 0.00 0.11 0.03 0.29 0.68 0.36 0.85 0.88 0.87 0.41 1.00 0.90 0.30 0.71 0.48 0.81 0.85 0.83 0.12 0.84 0.71 2.00 2.00 2.00 1.93 1.98 1.95 1.76 2.00 1.92 189 2.00 1.93 0.05 0.22 0.13 0.00 0.00 0.00 0.74 1.03 0.93 10Z 1.15 LOG 0.87 1.05 0.94 Lar 40 2:00 1.33 1.52 1.43 111 168 1.33 0.11 043 0.18 0.63 0.77 0.70 1.00 1.00 1.00 0.42 0.71 0.55 0.72 0.83 0.77 0.07 1.00 0.81 17 0.68 0.34 0.37 0.51 0.44 0.41 1.00 0.90 TABLE 5, DATA FROM L. PERFORATA, L, DENSIPILA AND THEIR F, PROGENIES perforata perforata x densipila densipila N=181 N=43 N=678 Min Max x Min Max x Min Max x 1.00 1.00 1.00 2.28 5. 5.00 5.00 2.00 2.50 2.21 2.23 150 192 173 3.75 4.79 4.19 4.76 3. 3.61 3.37 3.85 5.00 4.32 3.59 3.11 3.53 3.19 4.34 543 4.78 4.25 3.03 3.90 3.51 3.00 3.07 3.01 1.21 0.00 0.07 0.02 0.07 0.75 0.51 0.77 0.00 0.63 0.27 0.00 0.55 0.23 0.47 0.00 0.62 0.24 0.00 0.02 0.01 0.72 0.12 1.00 0.73 0.64 0.93 0.73 0.84 1.00 1.00 1.00 1.00 1.00 1.00 0.81 0.00 0.04 0.01 0.00 0.59 0.31 0.74 0.97 1.00 0.99 0.00 0.21 0.06 0.00 0.00 0.00 0.00 0.12 0.50 0.33 0.35 1.00 1.00 1.00 0.29 0.68 0.36 0.81 0.00 0.95 0.31 0.30 0.71 0.48 0.58 0.00 0.54 0.16 2. 2.00 2.00 1.79 0.00 0.11 0.03 189 2.00 1.93 1.30 0.98 1.72 1.19 0.74 1.03 0.93 1.06 0.79 1.18 0.95 Lil. Le? 1.38 1.62 131 1.58 149 0.11 0.43 0.18 0.02 0.00 0.01 0.00 0.42 0.71 0.55 0.56 0.00 0.30 0.08 0.17 068 0.34 0.84 1.00 1.00 1.00 trich. <.16 HYBRIDIZATION IN LESQUERELLA TABLE 6. DATA FROM L. STONENSIS, L. LYRATA AND THEIR F, PROGENIES stonensis stonensis X lyrata lyrata N=216 N=36 N=130 Min Max x Min Max x Min Max .48 mm. 0.00 0.02 0.01 0.09 0.11 0.08 0.00 0.00 0.00 trich. .16—-.48 mm. 0.98 1.00 1.00 0.36 0.50 0.42 0.00 0.00 .00 trich.<.16 0.98 1.00 1.00 0.36 0.67 0.50 0.00 0.07 0.03 TABLE 7. DATA FROM L. STONENSIS, L. LESCURI AND THEIR F, PROGENIES stonensis stonensis < lescurii lescurii N21 N=443 N=324 Min Max x Min Max x Min Max .48 mm. 0.00 0.02 0.01 ce pe ba : - ” trich. .16—.48 mm. 0.98 1.00 pegeaarpeynes trich.<.16 0.98 1.00 1.00 0.73 1.00 0.94 16 ROLLINS AND SOLBRIG TABLE 8. DATA FROM L. STONENSIS, L, DENSIPILA AND THEIR F, PROGENIES flower color trich.<.16 stonensis stonensis x densipila densipila N=216 N=312 =678 Min Max x Min Max x Min Max x 1.00 1.00 1.00 2.30 3.61 3.11 5.00 5.00 5.00 216. 2. 2.26 187. 220. 207 159 192. 1.73 3.60 4.33 3.95 3.53 3.89 3.72 3.06 3.61 3.37 3.54 3.99 3.71 2.97 3.31 3.21 3.11 3.53 3.19 4.09 4.91 4.44 3.63 4.25 3.91 3.03 3.90 3.51 148 1.73 1.59 0.01 0.20 0.08 0.00 0.07 0.02 0.40 0.72 0.57 0.10 0.42 0.23 0.00 0.63 0.27 0.02 0.15 0.07 0.17 0.71 0.42 0.00 0.62 0.24 0.90 1.00 0.97 0.84 1.00 0.94 0.12 1.00 0.73 100) 1, 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.01 1.00 1.00 1.00 0.90 1.00 1.00 0.97 1.00 0.99 0.00 0.00 0.00 0.00 0. 0.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 1. 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.47 0.77 0.66 0.00 0.95 0.31 0.00 0.03 0.01 0.02 0.70 0.18 0.00 0,54 0.16 0.00 0. 0.00 0.00 0 0.00 0.00 0.11 0.03 144 1.72 1.59 1.09 1.13 1.04 0.98 1.72 1.19 0.82 1.35 1.06 0.79 1.01 0.91 0.79 1.18 0.95 0.06 1.21 0.76 1.37 1.85 1.61 1.31 1.58 1.49 0.00 0.02 0.01 0. 00 0.00 0.00 0.01 0.00 0.98 1.00 1.00 0.23 0.70 0.38 0.00 0.30 0.08 0.98 1.00 1.00 1.00 1.00 1.00 1,00 1.00 1.00 TABLE 9. DATA FROM L. LYRATA, L, LESCURIT AND THEIR F, PROGENIES flower color lyrata lyrata x lescurii lescurii N=130 N=273 N=324 Min Max .48 mm. trich. .16—.48 mm. trich.<.16 TABLE 11. flower color . type: silique apex pedicel length trich.<.16 lyrata lyrata X densipila densipila N=130 N=546 N=678 Min Max x Min Max x Min Max x 5.00 5.00 5.00 5.00 5.00 5.00 5. 00 5.00 Lita AG 163 2.26 1.87 1.59 1.92 1.73 3.22 3.38 3.30 3.51 4.36 3.90 3.06 3.61 3.37 3.29 3.46 3.37 3.07 3.43 3.19 3.11 3.53 3.19 3.62 3.65 3.63 3.38 4.20 3.65 3.03 3.90 3.51 0.00 0.03 0.02 0.00 0.28 0.03 0.00 0.07 0.02 0.00 0.02 0.01 0.29 0.42 0.35 0.00 0.63 0.27 0.00 0.00 0.00 0.10 0.34 0.25 0.00 0.62 0.24 0.00 0.00 0.00 0.19 0.34 0,27 0.12 1.00 0.73 0.00 0.07 0.03 0.66 0.81 0.74 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.01 0.00 0. 0.04 0.01 0.00 0.07 0.03 0.61 0.75 0.70 0.97 1.00 0.99 0.00 0. 0.00 0.00 0. -00 0.00 0.00 0.00 0.00 0.07 0.03 0.03 0.20 0.12 1.00 1.00 1.00 0.00 0.00 0.00 0.66 0.81 0.73 0.00 0.95 0.31 0.00 0.00 0.00 0.01 0.47 0.05 0.00 0.54 0.16 0.00 0.00 06.00 0.00 0.02 0.00 0.00 0.11 0.03 1.82 1.83 1.82 1.00 1.33 1.10 0.98 1.72 1.19 0.88 1.00 0.94 0.92 1.14 0.98 0.79 1.18 0.95 Lit io 1e 1.64 1.85 1.74 131 1.58 1.49 0.00 0.00 0.00 0.00 0.02 0.00 0.00 0.01 0.00 0.00 0.00 0.00 0.00 0.67 0.05 0.00 0.30 0.08 0.00 0.07 0.03 0.66 0.81 0.73 1.00 1.00 1.00 324 1.00 0.81 0.90 lescurii X densipila densipila =608 N=678 Min Max x Min Max x 5.00 5.00 5.00 5.00 5.00 5.00 148 2.31 2.00 1.59 1.92 173 3.83 5.01 4.59 3.06 3.61 3.37 3.38 4.30 3.99 3.11 3.53 3.19 2.75 3.95 3.49 3.03 3.90 3.51 0.00 0.10 0.16? 0.00 0.07 0.02 0.89 1.00 0.99 0.00 0.63 0.27 0.50 1.00 0.93 0.00 0.62 0.24 0.00 1. 0.37 0.12 1.00 0.73 1.00 1.00 1.00 1.00 1.00 1.00 0.00 0.89 0.60 0.00 0.04 0.01 0.50 1.00 0.89 0.97 1.00 0.99 0.00 1.00 0.46 0.00 0.00 0.00 0.50 1.00 0.73 1.00 1.00 1.00 0.15 1.00 0.77 0.00 0.95 0.31 0.70 1.00 0.90 0.00 0.54 0.16 0.00 2.00 1.21 0.00 O.11 0.03 0.00 0.50 0.26 0.98 1.72 1.19 0.63 0.95 0.86 0.79 1.18 0.95 E21. © 1.65 iS). Foe: 2a 0.00 0.40 0.10 0.00 0.01 0.00 0.67 1.00 0.79 0.00 0.30 0.08 0.98 1.00 1.00 1.00 1.00 1.00 18 ROLLINS AND SOLBRIG can say is that 3 of every 4 populations sampled in 1953 were still in the same place ten or more years later. There are long stretches of the river where no hybrids are to be found. The larger populations appear to be on low-lying ground on the upper side of long river bends. This suggests that the seeds tend to settle out in areas of quiet water at a time when the river is at flood stage. The river follows a tortuous course, providing many places which seem ideal for the development of hybrid populations, but all such places are not occu- pied. This may be due to poor dispersal, or it may be that the flood plain is a marginal habitat for these plants. The seeds have no special dispersal mechanism. They are flat, round and very light in weight. When the capsules mature, the seeds fall to the ground, usually within inches of the mother plant. In this respect, some observations made in Ann Arbor are of interest. In the spring of 1967, F, hybrid plants of the Lesquerella lescurii < L. densipila cross were planted in rows 4 feet apart, with the plants spaced a foot from each other. One inflorescence was harvested from each plant for analysis, but several others remained. No cultivation was performed in the field which was soon overrun by agricultural weeds, largely plants of Amaranthus. In the fall, after the annual weeds had died, seedlings of Lesquerella were discovered. In the spring of 1968, the number of seedlings surrounding a number of previously selected and marked plants was counted, their distance to the site of the original mother plant measured and the amount of dry matter weighed. The results are depicted in Fig. 3 and given in Table 14. It can be seen that seeds were dispersed only 10-15 inches from the mother plant, most seedlings lying within 5 inches. Those seedlings that lay further away produced the largest plants and the largest number of seeds. In the summer of 1968 the field was plowed and used as an experimental field for other projects. Nevertheless, Lesquerella plants germinated in the fall, and in the spring of 1969 a uniform and dense crop of plants was found in the area. No plant was found, however, beyond five feet of the original plot. The cultivation of the field had resulted in a more uniform distribution of the seeds, but even so, few seeds had been dispersed out of the original area. Statistical problems. As stated previously, 28 populations were sampled over a period of 13 years from a total of eighteen sites. Ideally, a statistic is wanted that would provide an index of the amount and direction of change over all characters for the parameter time, and the parameter distance downstream. Unfortunately, we have not found a statistical pro- cedure that would properly handle our data. Furthermore, statistics that could be applied to a character at a time, such as a two-way analysis of variance or a discriminant function, cannot be applied because only some localities have been sampled more than one year, and no two localities were sampled over the same number of years. This is largely due to the fact that the early sampling was done when we were still trying to HYBRIDIZATION IN LESQUERELLA 19 demonstrate the existence of hybridization and no thought had been given to the statistical implications of the sampling procedure. Further- more, as stated before, the populations themselves are somewhat transient. We have therefore resorted to the following four way analysis in order to discover whether any trends exist: (1) a qualitative description of the peslnaiet of change for each character in the populations; (2) a Prim diagram, which shows the phenetic relations of all populations; (3) an analysis of the character by character simple correlation coefficients of all populations; (4) a hybrid index. Qualitative character by character description (see Table 1) Flower color (Character No. 1). Both parents have yellow flowers and so do all the hybrid populations. TABLE 12. GEOGRAPHICAL LOCATION OF WILD POPULATIONS SCORED Species Collection Number* —_ Location of Population L. densipila ama Rollins 57140 Lawrence Co.: 8 mi. east of Wre ollins 5987 Morgan Co.: 5 mi. ws of Falkville llins 6422 urge Co.: 5 mi. west of Falkville Rollins 55124 sichickes rd 0.: a 3 mi. southwest of Bethel near the west fork of the Stones River Williamson Co.: Kirkland Rollins & Solbrig 6656 Williamson Co.: ed south of Kirkland Rollins & Solbrig 6660 Williamson Co.: i. from the Harpeth River at the a ey Trinity Rd. and Co. 6183 L. lescurii Rollins & hg 6671 Chacashans Ue i mi. west of Ashland City Davidson Co.: flood plain south of the Harpeth Rollins 531 River at Linton Rollins 6475 Williamson Co.: 1 mi. east of Arrington Rollins & Solbrig 6657 Williamson Co.: 4 mi. south of Arrington L. lyrata Alabama Rollins 5599 Franklin Co.: 7 mi. east of Russellville Rollins 5981 Franklin Co.: 7 mi. east of Russellville and i mi. north of Cherry Hill Crossing L. perforata Tennessee Rollins 53139 Wilson Co.: 5 mi. north of Lebanon near Spring Creek Rollins 5514] Wilson Co.: 3 mi. west of Lebanon on US 70N Rollins 55145 Wilson Co.: 3 “ west o Rollins & Solbrig 6673 Wilson Co.: north of Lebanon near Spring Creek Rollins & Solbrig 6674 Wilson Co.: gercnale edge of Lebanon L. stonensis Tennesse Rollins 55138 Rutherford Co.: hae Tenn. 102 crosses the east fork of the Stones River Rollins 59120 Rutherford Co.: 1 mi. southwest of Lascassas Rutherford Co.: 1.5 mi. north of Walterhill Rutherford Co.: Stones River crossing at Lascassas Rollins 6147 Rollins 6462 ® Voucher specimens have been placed in the Gray Herbarium. 20 ROLLINS AND SOLBRIG Style length (Character No. 2). Lesquerella densipila has a slightly longer style (1.59-1.92, x=1.73, (see Table 11 and Figure 4) than L lescurit (1.43-1.74, x=1.57). The artificial hybrids showed slight over- dominance for this character, possibly due to the better culturing condi- tions of the garden. The natural hybrids varied from a low mean of 1.55 for population 1966-2’, to a high of 1.95 for population 1964-8’. In general, there are more populations (21) with a mean higher than 1.65 (halfway between both parents) than below (5) while 2 have exactly 1.65. Also, the populations downstream are consistently more likely to have means over 1.65, while upstream there is a great deal of fluctuation. The standard TABLE 13. LOCALITIES IN TEN NESSEE, SITE NUMBERS AND COLLECTION NUMBERS FOR FI ILD POPULATIONS OF HYBRIDS. SEE MAP > 1, Lesquerella densipila x L. lescurii Site No. Collection No. Location of Population 1 Rollins 55115 Williamson Co.: just below junction of Arrington ” 6474 Creek and the Harpeth River, SW side of river. 2 Rollins & Solbrig 6658 Williamson Co.: opposite side of river from site 1. ” & ” 6 9 3 Rollins & Solbrig 6660 Williamson Co.: near junction of Trinity Road and Arno Road, ca. 7 mile south of the Harpeth River. 4 Rollins & Solbrig 6661 Williamson Co.: oe side of the Harpeth River, just below Trinity R 5 Rollins & Solbrig 6662 Williamson Co.: "SW side of the Harpeth River near the crossing of Arno Road. 6 Rollins & Solbrig 6664 bidonogy i Co.: below bridge where interstate ute 65 crosses aes Harpe iver. 7 Rollins 6507 W illiamson Co.: N side of Harpeth River, about 3 miles east of Franklin 8 Rollins 53135 Ww Ses on Co.: S side of Harpeth River, 2 miles E ae, fe | of Franklin ”" 6472 9 Rollins & Solbrig 6666 Williamson Co.: near Harpeth — stad bridge where Hillsboro Road crosses the riv 10 Rollins 6514 Williamson C ore Mies where Snead Road crosses the river 11 Rollins 53133 Davidson Co.: near peth River just a ”" $5126 bridge where state route 100 crosses the riv: . 508 Rollins & Solbrig 6663 12 Rollins 55127 vidson Co.: r Harpeth River, 4 mile south- west of Reliohee | Post Office 13 Rollins & Solbrig 6667 Davidson Co.: just below bridge where interstate route 40 crosses the H 14 Rollins 55129* Davidson Co.: flood plain of aso sity ial miles W of junction 7 routes US 70N and US 15 Rollins 53130 Cheatham Co.: 9 miles E of White Blut flood plain of Harpeth River. 16 Rollins 53128 Cheatham Co.: near Harpeth River about 2 miles ” 55130 above bridge where US 70 crosses the river Rollins & Solbrig 6668 17 Rolli 3 Cheatham Co.; above bridge where highend Sane Rollins & Solbrig 6669 Ashland City road crosses the H Harpe 18 Rollins & Solbrig 6670 | Cheatham Co.: below bridge where state a 49 between Ashland City and Charlotte crosses the th River. HYBRIDIZATION IN LESQUERELLA 21 2. Lesquerella densipila x L. stonensis Site No. Collection No. Location of Population 1 Rollins 55175 Rutherford Co.: near Old Jefferson and just above junction of East and West forks of Stones River. 2 Rollins & Solbrig 6652 Davidson Co.: large bend of Stones River at end of Walker Road. 3 Rollins 55166 Davidson Co.; near Stones River, 4 miles NE of a Bee Lavergne. vee Rollins & Solbrig 6649 4 Rollins 55171 Davidson Co.: below and above bridge where = Couchville Pike crosses Stones River. Rollins & Solbrig 6651 5 Rollins & Solbrig 6653 Davidson Co.: flood plain of Stones River, } mile SW of Suggs Creek. 6 Rollins & Solbrig 6654 Davidson Co.: E side of Stones River off Cooke oad. 5 Rollins & Solbrig 6655 Davidson Co.: old corn field, 1 mile below site 6, off Cooke Road. 8 Rollins & Solbrig 6650 Davidson Co.: i mile below bridge of interstate 40 rossing over Stones River. *Studied but not scored or included in the data tables. deviation varies between .25 and .36 and the coefficient of variation between 15% and 22%. The comparable values of the F,, garden populations were s=.42 and cv=21%, indicating no loss of variation in the field popula- tions. Replum length (Character No. 3). This is one of the characters that best discriminates the two species. Lesquerella lescurii has a long replum (4.50-5.04 mm., %=4.68) while L. densipila has a short one (3.06-3.61, %=3.37). The artificial F, hybrid was intermediate but close to the larger parent (3.83-5.01, x=4.89) again probably due to cultural conditions. The natural hybrids varied from a low of 3.61 to a high of 4.93. About half the populations (21) had values below 4.00, the rest (17) having values above 4.00. The populations downstream again showed, in general, larger replums than the ones upstream which were likely to show greater differences from site to site. No appreciable difference existed in the standard deviation in most populations (values of .40 to .60). However, two populations had low values (s=.32, .38) and two very high (s=.75 and 76, see Table 15). The coefficient of variation varied from 11% to 16% with two populations showing low values of 8% to 10%. These values are equivalent to the parental populations (s=.57-.65, cv=14%-17%) and the artificial F, hybrids (s=.60, cv=13%) indicating no appreciable loss of variation in the natural hybrids. Replum width (Character No. 4). This is not a good discriminating character between the parents (Fig. 5). Lesquerella lescurii has a slightly wider replum (3.36-3.88, x=3.75) than L. densipila (3.11-3.53, x=3.19). The artificial hybrids showed marked overdominance (3.38-4.30, x=3.99). See Fig. 5. In the natural hybrid populations, a great deal of variation for this character was observed, from a low mean of 2.84 to a high of 4.01, that is, bo bo ROLLINS AND SOLBRIG 40+ Total plant wt. ight iD | it we. Total fru No. plants 0-7 14-21 28-35 Inches radius Fic. 3. Progeny plants, fruit weight and plant weight in grams plotted against distance from the three parent plants. HYBRIDIZATION IN LESQUERELLA 23 from considerably smaller than Lesquerella densipila to considerably larger than L. lescurii, as well as slightly larger than the artificial F,, plants. This variation is also shown by the standard deviation (.31-.53), and the coefficient of variation (9%-17%). The corresponding values of L. lescurii are s=.49, cv=13%; for L. densipila, s=.49, cv=15% and for the artificial F,, s=.40, cv=10%. Again, as with previous characters, the populations down- stream show a tendency to have larger values while those upstream are quite variable. Why this character shows such a range in variability we cannot say. Silique width (Character No. 5). This is the most discriminating character between the two species. Lesquerella lescurii has a flat silique while L. densipila has a globose one. Since the capsule of L. lescurii is larger, the actual thickness measurement differences are not as great as the “gestalt” differences. The mean thickness for true populations of L. lescurii is 1.68 mm. (1.51-2.01) while that of L. densipila is 3.51 (3.03- 3.90); the artificial F., has a value of 3.49 (2.75-3.95) similar to that of the largest parent (Fig. 6). In the natural hybrid populations, again a great deal of variation is encountered, from a low mean of 1.62 to a high of 3.82, compared to values observed in the parental populations. These values were all ob- tained from data taken on upstream populations. With variance from a low of s=.38 to a high of s=.63, the coefficient of variation varies from 13% to 27%, that is, twice as much. Thus, a great many rearrangements are indicated for this character. Therefore, it is of interest that the populations downstream from population 9 have more similar means (2.49-3.31) and coefficients of variation (15% to 23%) than those above it. Silique length/ width ratio. The silique length/width ratio is probably a better indicator than width alone as to similarity to either parent because it eliminates the effects of luxuriance. The length/width ratio of Les- querella lescurii is 2.80; for L. densipila it is 0.96; and for the artificial F,, hybrid, 1.65. In the natural hybrid populations, the lowest value observed was 1.06, the highest 2.40. That is, no population transcended its parents, confirming that they are all intermediates to a lesser or greater degree. TABLE 14. NUMBER OF SEEDLINGS, FRESH WEIGHT, AND FRUIT WEIGHT OF NATURALLY DISPERSED LESQUERELLA DENSIPILA X L. LESCURII SEEDLINGS Distance in cm. from mother plant A. Number of plants 0-7 7-14 14-21 21-28 28-35 Plant 1 288 139 28 14 7 Plant 2 Boe 125 8 51 10 Plant 3 155 46 50 is 7 B. Total fresh weight (g.) Plant 1 ses 57.8 136.2 90.4 48.4 46.5 Plant 2 36.2 65 6.3 90.7 Plant 3 53.1 56.5 29.4 69.7 95.9 C. Fruit weight ) « ~s 11 25.6 14.5 - . Plant 2 9.7 14.7 2.0 : : Plant 3 9.3 9.5 ida 0.8 12.7 24 ROLLINS AND SOLBRIG * °o 3 : A 5 e8 2 a a ag : ; eth es ° a we 8 ° = “¢ e — J s q ° >. 40+ ge ee S 3 & °@ ~ Cage Q J e 4 | s aa 35- a 2 i a a A " a e a 30- l ] l I ee Lee 14 16 18 20 2.2 24 Style Length ~ 4, an replum length (mm.) vs. mean style length (mm.) in populations of Lesquerella. Solid triangles: L. densipila, Open triangles: L. lescurii. Solid circles: L. densipila x eld populations. Open circles; L. densipila « L. lescurii, garden populations. However, the more interesting point is that the populations upriver (sites 2-13) show a great deal of variation (1.06-2.40) while those downstream are rather uniform (1.40-1.70). At the junction of the upper Harpeth River and Arrington Creek, the following values were obtained at the three sites sampled (1,2,2’): 1955 (1) 1.50; 1964 (1) 1.10; 1966 (2)2.40; 1966 (2’) 1.93 L. lescurii, Septum perforation (Character No. 6). Both parents have basically unperforated septa as do the hybrid populations. Occasional perforated septa in some plants may be due to stresses during the drying of the material. Replum apex (Character No. 7). This is another distinctive character that separates these two species but it is difficult to score because some subjectivity is involved. In Lesquerella densipila the replum apex is rounded (Fig. 2) scored as 0, and in L. lescurii it is pointed (scored as 1). Scoring each plant as 1 or 0, with mean values varying from 0 to 0.63 were found to obtain in wild populations of L. densipila (with a grand mean for the species of 0.27), while in populations of L. lescurii the corresponding values were 0.61 and 0.93 (species mean 0.67). Artificial HYBRIDIZATION IN LESQUERELLA 25 garden hybrids always had a pointed replum apex (mean for all plants 0.99). The natural hybrids, however, showed a great deal of variation, from populations with all rounded apices (value of .04) to all pointed (value of 1.00). Of the 27 populations analyzed, only 8 had values that can be considered intermediate (range of 0.30 to 0.65). It is our feeling, in view of the results of both the garden studies and the field studies, that this character has a big environmental component in its expression. Con- sequently, although it is very adequate to separate the parent species, it is not very useful as an indicator in hybrid populations. This is further reinforced when the variance values are analyzed. The standard deviation values vary from 0.00 to 0.50; the coefficient of variation from 0% to 466.832! Replum base (Character No. 8). What has been said of the replum apex applies as well to the base. Furthermore, there is no substantial difference between the parent populations in this character. Trichomes at the base of the style (Character No. 9). Lesquerella lescurii has no trichomes at the base of the style while they are nearly always present in L. densipila. Segregation of this character in the F, population is controlled by more than one gene. The overall mean for the F, population was 0.37. Field hybrids varied from populations entirely lacking plants with trichomes at the base of the style (pop. 53-11) to those where every plant had them present (pop. 65-10). However, the majority of the populations were intermediate for this character. Popula- tions markedly densipila-like, that is, with trichomes at the base of the style in the majority of the plants, were found only upstream and were in the minority; most populations either had a more or less even mixture or a majority of the plants were without trichomes at the base of the style. The latter were found mostly downstream. With respect to this character, there is a tendency for the populations to become more lescurii-like farther downstream. Values for standard deviation and coefficient of variation are not too meaningful because of the bimodal distribution of the character in the population. Trichomes on the outer surface of the valve (Character No. 10, 13, 16, 21-23). Both species have trichomes on the outer surface of the fruit valve. However, the trichomes of Lesquerella densipila are mostly simple, while those of L. lescurii are a mixture of unbranched trichomes with bulbous bases, and others that are dendritically branched. Artificial hybrids are intermediate in their indumentum, having all three types of trichomes. Natural hybrid populations show the same situation. Twenty-five per cent (pop. 64-17) to 100 per cent of the plants have simple trichomes and 13-90 per cent have branched ones, except population 64-1, where 2 per cent of the plants had branched trichomes. A slight tendency for the plants to become more lescurii-like downstream can be seen in the relative increase in the percentage of plants with branched trichomes in the popu- lations farther downstream. 26 ROLLINS AND SOLBRIG TaBLe 15, MEAN VALUES FOR CHARACTERS OF FIELD HYBRIDS OF L. LESCURIL & L. DENSIPIL Character xX s x s xX s x s x s i ee tee Ce, ON eRe Re ae 1 5.00 -00 .00 : : ; H : ‘ 2 1.60 33 1.69 a7 Loy. 29 b55 ot 1.92 34 3 3.42 4.02 52 3.00. 53 ots Bo 3.61 BE 4 3.33 40 3:65 BI 2.95 Ad 2.84 49 He es kl 38 5 2.49 63 3.82 51 1.63 43 1.93 3.43 45 6 0.02 14 0.00 00 0.07 26 0.08 28 0.02 14 ve 0.28 45 0.12 32 0.74 44 0.69 48 0.63 48 8 0.00 00 0.00 00 0.82 39 0.82 39 0.62 49 9 0.18 39 0.67 AT 0.07 26 0.07 25 0.76 43 10 1.00 1.00: 200 1.00 00 1.00 1.00 00 AY 0.48 50 0.00 -O0O 0.90 30 0.74 44 0.04 20 12 0.62 49 1.00 00 0.44 50 0.58 50 0.97 16 13 0.26 44 0.00 00 0.87 34 O37 00 0.00 00 14 0.83 38 1.00 00 0.56 50 0.68 47 0.99 07 15 0.74 44 0.02 14 0.96 19 0.86 36 0.30 46 16 0.66 48 0.00 00 0.80 40 0 6 0.54 50 fare 1.09:.: 1,20 0.00 00 2.02 79 1.94 1.20 0.11 54 18 0.44 1.03 oO 0.00 00 0 0.98 18 19 0.84 Ze 1.10 vale 0.83 21 OTF 18 0.97 20 20 1.50 15 2s valk 1.16 26 P22 22, 1.58 36 val O37 49 0.00 00 0.97 16 0.84 or 0.01 10 es 0.73 45 0.03 17 0.82 39 0.87 34 0.30 46 va! 1.00 .00 1.00 .00 0.99 1.00 00 1.00 00 Site 66-4 66-5 66-6 65-7 53-8 Character xX s x s X S x s xX s 1 5.00 .00 5.00 00 5.00 .00 5.00 .00 5.00 .00 2 139° 32 Yes) Guess: LG3 232 Le al Los. 4 S 3.61 61 3.67 58 3.67 <.52 3715 50 SOF BT 4 3.02 47 3.12 42 3.07 4] 3.46 33 3.40 44 5 3.37 56 3.40 55 2.50 a2 2.90 49 Aer Ard 50 6 0.02 13 0.09 28 0.08 27 0.00 0.08 27 3 0.81 40 0.72 45 0.91 29 0.04 21 0.90 Sl 8 0.53 50 0.62 49 0.76 43 0.01 0.95 22 9 0.81 40 0.50 50 0.25 O25 42 0.14 35 10 1.00 00 1.00 1.00 00 1,00 00 1.00 00 ll 0.11 a2 0.08 or 0.46 50 Oro ad 0.44 50 12 1.00 00 0.95 ee 0.74 44 0.82 38 0.57 50 13 0.00 00 0.00 00 0.12 33 0.01 09 0.33 AT 14 1.00 00 1.00 00 0.94 24 1.00 00 0.82 39 15 0.13 34 0.21 Al 0.39 49 0.25 43 0.29 46 16 0.26 44 0.40 49 0.83 38 0.29 46 25 17 0.26 79 0.19 -70 PA?) 437 0.40 79 L.09: 1,31 18 0.94 wag 0.97 we 0.18 39 42 0.39 49 19 Lis 27 HAS We 0.79 16 Lid 18 0.9 20 20 1.40 36 LoS «34 1.35 23 £32 23 1.36 21 0.00 00 .0O0 0.23 42, 0.04 19 0.13 pe 0.07 25 0.02 pa eS 0.80 40 0.43 50 0.70 46 Ag 1.00 00 1.00 .00 1 00 1.00 00 1.00 HYBRIDIZATION IN LESQUERELLA TaBLeE 15. (Continued ) 27 Site 64-8 64-8’ 66-9 65-10 53-11 Character x 5 x 5 x s xX s xX s hE 5.00 00 5.00: 00 5.00 00 5.00 .00 5.00 .00 2 1:70 io LObe 36 1.65 30 EL os Venere 9 f 1.68 25 3 4.67 .61 4.27 .76 3.74 48 3.87 38 AAS: = 32 4 3.92 48 3:19. 46 3.18 39 3.14 38 3.04 a 5 267)5-- 5S 360°. 343 270.49 2T1t. 40 9.49. 38 6 0.05 22 0.00 ~=.00 0.41 20 0.00 = .00 0.00 = .00 7 0.33 AT 0.3 AT 0.93 26 0.04 19 0.14 36 8 0.01 .07 0.00 ~=.00 0.78 42 0.00 = .00 0.00 .00 9 0.06 24 0.63 49 0.51 50 1.00 .00 0.00 00 10 0.99 07 1.00 .00 1.00 00 LO: C00 1.00 00 1l 0.59 AQ 0.20 Al 0.49 50 0.00 = .00 0.50: 52 iv 0.63 48 0.97 17 0.89 OL L066; .00 0.14: 36 13 0.41 A9 0.00 .00 0.07 26 0.00 = .00 O55 =) 00 14 0.63 48 1.00 .00 0.99 09 100: 06 Ot ear 15 0.78 A2 0.23.5 3 40 0.26 44 0:30... AT 0.71 AT 16 0.44 50 0.00 .00 0.93 26 0.04 19 O21 143 17 1.20.2 1:15 0.20: Al 1.28 1.40 0.00 = .00 L6 L398 18 0.19 40 0.91 45 0.64 £4.48 1.00 .00 0.36 350 19 1.08 ee V3 22, 0.81 16 135 17 0.94: — 22 20 1.24 24 1.40 28 Lol 28 U.92" 33 P10 ST A 0.38 AQ 0.00 .00 0.17 38 0.00. .00 00 321 22, 0.86 8D 0.23 A3 0.71 46 0.19 40 0.71 AT 23 1.00.00 1.00 00 1.00 .00 1.00 .00 1:00:00 Site 55-11 65-11 66-11 55-12 66-13 Character x s xX s x s x s x s 1 5.00 .00 5.00 .00 5.00 00 5.00: 00 5.00.00 2 1.84 31 1.62 28 1.91 ao 17% 28 Poa a8 3 3.78 A2 are AT o.03 59 4.01 .64 4.21 .64 4 326.38 3.98 2 ae cL Ng 8 35D... oe 3.39; 43 5 2.86 2 52 950. 38 3.31 48 DAG Uf ES se 949 SO 6 0.00 .00 O03: 15 0.06 oS 0.08 28 O04 = 22 1 1.00 .00 007 26 083: 238 023. AB 0.63 ~=~«.48 8 0.92 98 0.00 00 0.79 4l 0.00 .00 0.64 £4.48 9 0.45 50 O12. 33 0.54 50 O10 231 O22: at 10 0.98 15 100. -.60 1.00 .00 TOO. 00 1.00 .00 11 0.38 49 0.53 00 0.11 Si Ot =O 0:53.50 12, 0.72: 36 0.76 43 0.91 29 O72 146 0.66 48 i3 O01 SL 0.09 =-.28 0.00 0.19: 39 0.53 -.o0 14 0.98 = 15 O08 313 0.99 .09 090" 331 0.82 «39 15 0.21" AE G08 O80 O35 48 0.63.50 0.64 48 16 0.79 Al 0.53 50 0.60 49 0.67 48 O86 35 Me 083 Lili 1.09 1.16 0.23 fe 1:33 '20 127 325 18 055°" 50 0:45: 50 0.91 29 0.56 30 013° 35 19 0.90 14 1.07 14 Rit 25 0.89 17 £06 22 20 L356 26 Pi Bol 35 130. (| 139 ..30 21 0.15): 36 0.26 «44 0.01 09 0.21 Al 0.72 45 22, 0.77 43 0.90 38 0.17 38 000 at 0.98 15 23 098° 215 0.97 18 1.00 00 0.98 14 0.99 08 28 ROLLINS AND SOLBRIG TaBLeE 15. (Continued ) Site 53-15 53-16 55-16 Character x s x us 5.00 .00 5.00 00 5.00 00 2 Le oe 1.65:53,:.86 1.69 33 S. 4.56 'S 306.56 3.92 47 4 4.01 3 Jt 45 3.36 46 D oto > (6) 2.00 > 54 2.49 8 6 O02 15 0.05 D2 O05) .25 7 O28 45 025") 43 0.90 30 8 DOS Ss 0.00 .00 0.92 pag 9 0.038" 48 O12 oo 0.12 32 10 1.00 .00 100) 00 1.00 00 11 0.62 49 0.62 AQ 0.64 48 12 0.31 “| ° ° o = i ee e:55 a : & : eer 35-4 a, : a | A : A 3.04 . | T ] T | T | 1 | ! | qT | ] T ] 28 3.0 SZ 34 36 38 40 42 ae Replum Width Fic. 5. Mean replum length ( vs. mean replum width (mm. olid triangles: L. densipila. pers Peat ws L. lescurii. Solid circles: field populations. Open circles: L. densipila x L. lescurii, garden populat L. ) in populations of —— Fogg nsipila * L. lescuri 5O- ind ° 4 a ° A - z a ° a 9 Vv ° fe} s fe] °o — -* a . ° big 4 40- * é e ec “Fa ° 3 - se ° Ss . : & 7 . -+* . ° J os 3.5- 1 a at A A A Ke T T T | T T t T | T T T T | T T , T | tT T T | t ] 15 20 25 3.0 35 40 43 Silique Width © Fic. 6. Mean replum length (mm.) vs. mean silique width (m m triangles: L. —. ‘Solid triangles: L. densipila. — circles: field populations. Open circles: L. densipila L. lescurii, g | of Lesquerella. . de nsipila x L. lescurii, 32 ROLLINS AND SOLBRIG (3) Some populations of natural hybrids (4) are intermediate between the bulk of the hybrids and Lesquerella lescurii and are presumed to be introgressants, while others (5) are intermediate between the bulk of the hybrids and L. densipila and are also presumed to be introgressants. Finally, four populations of hybrids clump with L. densipila and away from the hybrid populations. These are presumed to be true L. densipila or introgressants with L. densipila strongly predominating and with very few L. lescurii characters. (4) No chronological trend was observed, the distribution among intro- gressants and intermediates is about the same for populations sampled in 1953, 1955, 1964, 1965 and 1966. (5) Of the 15 hybrid populations classed as intermediates, eleven were from down river (out of 16 sampled in that area) at least 25 miles away from the place where Arrington Creek meets the Harpeth River, and only four intermediates out of 12 sampled there came from near the junction of these two streams. Among the introgressed populations, 8 came from near this junction and only five were from downriver. Of these, only one was from beyond locality 11, about 25 miles from the junction. Correlation coefficients. In another section (see Genetic coherence) the reason for this part of the analysis is explained in more detail. In brief, character by character simple correlations were obtained, and their values averaged to obtain a value for the population. The character by character simple correlation of independent characters in natural populations of true breeding species should be zero or close to it, by definition. However, it was observed by Anderson (1949) and more recently by Clausen (1967) that in hybrids, characters that segregate independently in the parent plants do not necessarily segregate inde- pendently in the hybrids. The numerical value of the population obtained by averaging all the correlations can therefore be used as a kind of index of hybridity. Table 16 shows the values obtained. With the exception of one popula- tion (64-1) with a value of .065, all populations had values significantly higher than the parental species, confirming their hybrid nature. The values, however, were lower in general than those of the artificial F, hybrids. Furthermore, it can be seen that the populations downstream show much less variability from site to site (.133 to .176) than populations upstream (.065-.221). Hybrid index. A simple hybrid index was constructed and the popula- tions were scored accordingly (Table 16). It confirmed what the analysis of characters, the Prim diagram, and the correlation coefficient had shown: that populations upstream tended to be more similar to one of the parents (10 of 15) rather than intermediate (5 of 15). The reverse is true of the populations downstream where 6 of 8 populations had an intermediate value of 3 in a scale from 0 to 6, the other 2 having a value of 2, that is, toward Lesquerella lescurii. HYBRIDIZATION IN LESQUERELLA 33 LESQUERELLA DENSIPILA X L. LESCURID-SUMMARY At the beginning of this investigation, we addressed ourselves to the following question: is the variation of the hybrid populations greatest near the area where they are formed, or is the variation more or less uniform throughout the area of their occurrence? That is, we were interested in knowing whether natural selection was favoring certain genotypes among the array present in the hybrid swarms upstream. If such were the case, the favored genotypes should be more numerous farther downstream where presumably the influx of genes from the parental species is least intense. s P ~ ao ella lescurii. mae triangles: . densipila, Solid Prim diagram triangles: Lesquer - = oe som Fic. 7. . Open circles: LL haga Pd z.. aa field populations. Open circle: populati 34 ROLLINS AND SOLBRIG TABLE 16. AVERAGE OF CORRELATIONS OF ALL CHARACTERS IN WILD a gaia a OF LESCURIL X L. DENSIPILA AND HYBRID INDEX NUMBE stream populations Correlations .065—.221; index 0-6 Year & locality Correlations = Index Year & locality Correlations —_ Index 55-1 221 3 65-7 131 5 64-1 065 6 53-8 172 2 66-2 151 0 64-8 181 3 66—2’ 161 0 66-9 147 3 66-3 6 65-11 176 4 66-4 28 5 66-11 133 4 66-5 128 5 66-13 162 3 66-6 153 3 Downstream populations Correlations .156—.177; index 2-3 Year & locality Correlations = Index Year & locality _ Correlations _— Index 53-15 171 53-16 .156 3 66-16 157 3 66-17 .168 64-16 170 2 17 are 3 55-16 177 3 66-18 .163 3 *Index sense 0-6. 0=lescurii-like : lensiolle tks: After carefully studying these populations for a considerable number of years and over their total range of distribution, we find that the popu- lations downstream, taken as a group, are more uniform than those up- stream and show a more restricted range of morphological types. However, the variability of any single population as measured by their coefficient of variability is about the same, whether upstream or downstream. Our original model, based on three years of observation, was too simple. We assumed that at the junction of Arrington Creek (where true Les- querella lescurii populations are found only upstream) and the Harpeth River (where true L. densipila populations are found only upstream) a large permanent and more or less self-renewing hybrid swarm existed. It was presumed that seeds from the plants of this population were carried downstream, establishing new populations of hybrids. Seeds from these new populations, in turn, were presumed to be carried further downstream, and so on, until they had established a population at the very place where the Harpeth River flows into the much larger Cumberland River. What we had not appreciated fully at the time is the degree of unevenness of the contribution of propagules of the two parental species from year to year. In effect, seeds are dispersed largely as a result of occasional flooding of the rivers and creeks of the area, at which time they overrun their banks picking up seeds at one locality and depositing them at another. As a result of the pattern of highly localized spring storms in the area, the upper Harpeth River and Arrington Creek are likely to carry very uneven HYBRIDIZATION IN LESQUERELLA 35 amounts of water. When one of them floods and the other does not, only seeds of L. lescurii (from Arrington Creek) or L. densipila (from the upper Harpeth ) are deposited at the floodplain where these streams meet. When both flood, the hybrid swarm is more intermediate. In 1953 and 1955 the populations close to the junction of the two rivers were fairly intermediate. In 1964 and 1965, on the other hand, they had taken a decidedly densipila-like character. However, in 1966, those close to the junction of the upper Harpeth and Arrington Creek were very lescurii-like, while those a mile or two downstream were still densipila-like. Thus, it appeared that a very recent influx of lescurii seeds into the junction hybrid population had taken place. Populations downstream are somewhat buffered from these effects. Most of the seeds arriving there are probably from hybrid populations upstream rather than from populations of the parental species. Conse- quently there is less chance for either species to predominate. This is, of course, true the further downstream one goes. The situation is effectively shown in F ig. 8 where the populations sampled are given by year of collection ( 1953-1966) and arranged vertically according to a hybrid index of zero to six (see Table 16). Laterally, they are arranged by collection site along the Harpeth River beginning at the point where Arrington Creek enters and extending downstream to the point of its entrance into the Cumberland River. UPSTREAM DOWNSTREAM S— (64) 66) wae (6a66) 65) (65) 2 6a Pa a a i (66) (oa (64) (66) en) e4) (66) 68 (53) 6S = ao > = mee (53) (64) (66) 5 INSET | med miles ! Ee Ill ] | | | | | | aha — ‘5 ore 9 it 16 7 18 13 15 LOCALITY NUMBER Fic. 8. art showing samplings of Lesquerella densipila «x L. lescurii hybrid populations by year of collection (1953-1966) arranged vertically according to their position in a hybrid index from zero (L. lescurii) to six (L. densipila). Laterally, they are plotted by site position along the Harpeth River from the junction of Arrington Creek at left downstream to its entrance into the Cumberland River on the right at site 18. 36 ROLLINS AND SOLBRIG TaBLE 17. MEAN VALUES FOR CHARACTERS OF FIELD HYBRIDS OF L. DENSIPILA X L. STONENSIS Site 55-4 61-4 66-4 66-5 66-6 Character 4 s xX s x s 4 s x s 1 _ — Loko 58 2.0521, 10 1:62... 96 — oe 2. 2:06 2:36 1.80. 46 2.30. SAY CAG Mae PE (a 3 Wlaeaaeae: 8 3 es die) 65 ATG: )05 AIO 262 SAL. 49 4 Site Nae, 50 AT Oto ae 3.78 44 3.30 42, 5 392° 361 3.59 68 AGE 53 464 .58 S048 6 0.62 3.65 O63 5.75 O72 5d O77: 52 1,022 34 7 O22.— -<41 0.29 46 0.97'°= = 19 0.94 .24 0.98 = 85 8 0.06 24 0.07 .26 0.89 wl 0.92 BV 6 0.80 40 9 091-28 0.93 - .26 0.98 3.<;,13 097%27.18 00867 °.15 10 1.00 .00 1.00 .00 1:00.00 1.00 .00 100; 00 ll 0.15 36 O18: 38 0.06 24 0.11 wl. 0.11 ol 12 1:00 00 1:00 00 1.00 .00 0.99 .07 1.00 .00 13 0.00 .00 0.00 .00 0.46 50 0.36 48 0.24::.48 14 1.00 .00 1.00" 00 0.65 .48 0.78 Al 0.80 .40 15 6.08 2% 0.00 .00 0.00 .00 DOL OT 0.00 .00 16 O24 43 0.02. 513 0.05 -..23 O21 sat Ot); 31 ale 0.28 2.79 0.30 .76 0.08 36 OT 2281 O15 «1:52 18 103 Bt iG ..37 Lab. 66 ig Sy las |. ..53 19 104° 20 O54" "17 138. 26 ESL. 3.26 0.90.19 20 0.83 13 1.01 08 PO0l: 26 E21 2S Lol 5 21 0.00 .00 0.06.2 00 0.01 .09 0.0 907 0.00 .00 22, 069: AT 125 44 0.78 42 O73. 3.45 O71... 42 23 1.00 .00 1.00 .00 O22 22 42 O21. Ao O25... AZ Site 55-1 66-2 55-3 65-3 66-3 Character x s 4 s xX s x s 4 s 1 _ — 2:73: 126 _ = 1.00 .00 162°: 3.65 2 220): 42 220 38 Cl. .36 104. (45 2.10: 138 3 Da gale ss BL. Oo ST Weiaee Tf 3.44 .50 3.46 55 4 3.86 42 6.26.38 aI 336 S41 pao 42 5 ASS BL 4.05... ,52 24 44 3.59 47 4.00 .57 6 ae ie 0.50; bl igi. ..76 0.90 .69 0.87 + .40 fg 0.42 .5O 026. 44 O37. 48 0.01 ll Ue 2h 8 O0f 2 O05) 21 O11" 31 O02 << 16 053°" .38 9 1.00 .00 0.99 09 0:94 25 1.00 = .00 1.00 .00 10 100. 00 LOG: 0) 1.00 .00 1.00 .00 4,00... 06 Na 0.00 = .00 O06: 23 0.01 .09 G15 36 C05. = 22 12, 1.00 .00 1,00: 00 1.00 .00 £00 2.00 1.00 .00 13 0.00 .00 0.00.00 0.00 = .00 0.00 .00 0.00 .00 14 1.00 .00 100. 00 1.00 .00 100.2 00 Cor. lT 15 0.00 .00 O05 21 0.01 .09 0.01 ll COs 17 16 0.00 .00 0.35 48 0.02: 15 0.04 19 0.20 .40 17 0.00 .00 O12... 54 0.01 .09 0.16 .46 0.08 ~=Al 18 1.63 -°. 49 i: 2D 1.10 45 LOZ... 96 Lis 265 19 1.00. 20 1.05. 20 O97 . 20 1.08 19 0.88 .20 20 $05 18 bon 3 101 06 Ol 16 14° 18 21 0.0L... 09 0. .00 0.00 = .00 0.00 .00 0.00 .00 22, 0.90 .30 O5L....50 0.94 .24 0.51. ...B0 OAT... 5O 23 100. 16 LOO: 600 1.00 .00 1.00 .00 0.74 44 HYBRIDIZATION IN LESQUERELLA 37 TABLE 17. (Continued ) Site 66-7 66-8 Character x s x s 1 POS 117 2.14 83 2 40 2.19 40 a 4.28 70 3.89 62 4 300° ag K RS 44 5 4.32 57 4.20 Sy 6 0.75. 48 0.85 52 7 0.59 50 0.95 22 8 0.28 45 0.86 35 9 0:96 20 0.98 15 10 1,00: ..00 1.00 00 ll 0.04 .20 0.07 25 12 1.00 .00 0.99 09 13 0.00 .00 0.43 14 1.00 .00 0.67 47 15 0.08" 27 0.00 00 16 0:49. “2b 0.21 41 17 0.08 «44 O10: 42 18 0.98 . 24 iiss: 46 19 125. 33 1:14" 20 20 Lae Simeay | Lee 99 val 0.00 .00 0.00 .00 22, 0.82339 0.65 48 23 1.00 .00 0.36 .48 At this point there is no evidence to indicate that natural selection is favoring any particular genotype. The observed wide fluctuations appear to be related to catastrophic events (such as the storm pattern and sub- sequent flooding) rather than to evolutionary forces. However, we are much impressed with the dynamic ebb and flow of the genes of each species into the hybrid populations, and with the wave-like movement of the genetic impact of one species, then the other, on successive hybrid swarms down river from the point where the two species come together. LESQUERELLA DENSIPILA L. STONENSIS When first noticed, this hybrid, on the basis of the species then known, was classified as Lesquerella densipila var. maxima (Rollins, 1952). How- ever, the particular characters and variability of these populations sug- gested that they had arisen due to hybridization between L. densipila and an unknown species. Later, the putative unknown species was found and given the name L. stonensis ( Rollins, 1955). Lesquerella stonensis exists as a pure species at various localities along the East Fork of the Stones River (see Table 12 for localities of seed sources) while L. densipila is found as a pure species along the West Fork of the Stones River, as well as to the south and west of this area in the Central Basin (Fig. 1). Both species are found in the flood plains of 38 ROLLINS AND SOLBRIG the rivers and presumably some of the seed is washed downstream during the annual spring floods. Consequently, where the two forks of the Stones River come together, the two species meet and hybridize and from there to the point where the Stones River flows into the Cumberland, a distance of some 30 miles, only hybrid populations are found (Rollins, 1957). This situation was modified with construction of the J. Percy Priest Dam and Reservoir on the lower part of the Stones River. This reservoir flooded a great many of the hybrid populations and isolated the ones downstream below the dam. The Lesquerella densipila < L. stonensis hybrid swarm presents the same basic geographic situation as that of L. densipila x L. lescurii. The lower Stones River is shorter than the Harpeth, offering less opportunity for differentiation with distance, and the construction of the dam has LIS —2Z7 A> e/a", \ “ Neal. 7\ fp ; 5 19 7 y= eB Wes, KY Dae NN tye Lae, “a en WA KAZI ASX 7 ( TV Pac: ~*~] I7€&y 5 ALES Py, ) OO Z Nie Vv RK e ae Fic. 9, Correlation polygons showing levels of significance in simple character Xx character correlation for all 223 possible combinations: full 1 er; broken line. i i ine, correlation significant at the 1% level or bett line, co: io! ignificant at the 5% leve’ better, but less t 1%; no line, correlation significant at a level less than 5%. Upper left, Lesquerella densi (population 66056); upper right, Lesquerella densipila ions; ila, sum of all 66057); lower right, Lesquerella lescurii < L. densipila, segregating ¥. family from a single parental cross (population 67113). HYBRIDIZATION IN LESQUERELLA 39 destroyed any possibility of revisiting many of these populations in the future. Our account of this hybrid swarm is restricted to a character by char- acter description, mean values for the characters (Table 17), an analysis of the character correlation polygons and a hybrid index (Fig. 9 & 10). Qualitative Character by Character Description Flower color (Character No. 1). Lesquerella densipila has yellow flowers while L. stonensis has white flowers. On a scale of 1 to 5 (with 1=white ) the hybrids tend to be intermediate or toward the white parent and no significant difference is found between the various populations (Fig. 11). The natural hybrids are more frequently white in color than the artificial hybrids ——— 21 MONS Ay 7 B, See 19, MN Nos - A . U 18 aa ig t ® Og e6 PTS Ni Fic. 10. Same as figure 9. Upper left, Lesquerella densipila x L. lescurii, natural hybrid popula- tion from site 18 (66070); upper right, Lesquerella densipila x L. lescurii, natural hybrid population from site 1 (64074); lower left, Lesquerella densipila x L. lescurii, natural hybrid population from site 17 (64073); lower right, Lesquerella densipila L. lescurii, natural hybrid population from site 1 (55115). 40 ROLLINS AND SOLBRIG 5.0- 45- il Vv a i ee “I 40- S + v ®o6 ° rm ‘tl * D "I ” ° ° - . ‘ 35- : ‘ ° a A 4 a 7 A 30 . T T T T 0 lO 15 20 25.20. 36 40. 45. 50 Flower Color Fic. 11. Mean replum length (mm.) vs. flower color (l=white, 5=yellow) in populations of Lesquerella, Open triangles: L. stonensis. Solid triangles: L. densipila. Solid circles: L. densipila « L. stonensis, field populations. Open circles: L. densipila x L. stonensis, garden populations. Style length (Character No. 2). Populations of Lesquerella stonensis have longer styles (x=2.1 to 2.4 mm.) than L. densipila (x=1.6 to 1.9 mm.). The artificial hybrid population as a whole had an overall mean for this character of 2.06 with some families with a value as low as 1.87 and as high as 2.14. The natural field hybrids showed higher values. The lower value was in population 61-4 (x=1.80), the highest in population 66-4 (x=2.39), with most populations having values between 2.1 and 2.3, well within the range of the L. stonensis parent (Fig. 12). Replum length (Character No. 3). Lesquerella stonensis has a slightly longer replum than L. densipila (3.6 to 4.3 mm. to 3.0 to 3.5 mm, for L. densipila). The artificial hybrids are intermediate with a grand mean value of 3.7 mm. Natural field hybrids are varied, but approximate the L. stonensis parent more than that of L. densipila (%=3.44 to 4.28 mm.). No geographical or temporal trend could be observed (Fig. 13). The standard deviation and coefficient of variation for all populations was similar. Replum width (Character No. 4). The replum is wider in Lesquerella stonensis (x=3.7) than in L. densipila (x=3.3). In the artificial hybrids the HYBRIDIZATION IN LESQUERELLA 4] ©: 45- = 1 ee S ] v E i, ssl : we IS : Pyne a 2 4 ° > > . °° oe etl i a ° = . e ° é . a a a a 3.0 | T | q | T | T | T T 1 iS 7 RS) 21 paa 25 Style Length Fic. 12. Mean replum length (mm.) vs. mean style length ( mm.) in populations of Lesquerella. Solid triangles: L. densipila. Open triangles: L. stonensis. Solid circles: L. densipila x L. stonensis, field populations. Open circles: L. densipila x L. stonensis, garden populations. replum is slightly narrower than in either parent (x=3.2), while in the natural hybrids it is very variable, in some populations being as narrow as the artificial hybrid (x=3.26) and in others as wide or wider than L. stonensis (x=3.86). The coefficient of variation in each population is on the order of 10% which is lower than for some of the other characters. Why this character is so variable from population to population, but so relatively uniform within each population, is not known to us. Silique width (Character No. 5). This is a good character to separate the two species, the siliques of Lesquerella stonensis being much wider (x=4.4 mm. ) than those of L. densipila (x=3.51 mm. ). Artificial hybrids are intermediate (x=3.9 mm.) although there is some variation from one genetic family to another (range from 3.6 to 4.2 mm., Fig. 14). Natural hybrid populations tended to be intermediate or closer to the L. stonensis parent (x=3.92 to 4.62) although two populations had values reminiscent of L. densipila (x=3.89) and three had values well within the range of L, stonensis (x=4.64, 4.64 and 4.83). The coefficient of variation ranged from 11% to 19%. Septum perforation (Character No. 6). Lesquerella stonensis has a small perforation in the center of the septum that occupies about a third of the total septum area, while L. densipila has unperforated septa. Artificial hybrids tend to have very small perforations in their septa, as do 42 ROLLINS AND SOLBRIG Pdi Replum Length 5.0- 45-4 a os $ a Vv me a ee 4.0- “ &§ i ° ° e v ~ ; e > i" ° ° He a ° e 3.5- . - eae ad e a a a a 25 27 29 30 3.3 35 3.7 39 41 43 Replum Width 13. Mean replum ~~ (m vs. mean sesso width (mm.) in populations of — angles: L. densipila, Open eine L. stonensis. Solid circles: densipila « L. stone field populations. Open —. L. densipila « L. sears garden populations. 5.0 45- a e oy 2 Vv - 40+ - 4 e Vv ° ° * si e J ®o ° al ° A 3.5 - e Ae 4 e a _ Aa fae ] a 3.0- “5 42 “a6 48 Silique Width idee es oe het length (mm.) vs. mean silique width in populations of segreagssees Solid a, Open tri Honteoenen a stonensis. Solid circl L. densipila x stonensis, field are ae ‘ow sea L. densipila - stonensis, garden populations. HYBRIDIZATION IN LESQUERELLA 43 natural hybrids. A great deal of variation in this character exists in the population, however, and some hybrid plants have no perforation, while others have a perforation as large as that of L. stonensis. The size of the perforation is less of an indicator of hybridity in the population than the fact that the population segregates for this character. This points to a rather simple genetic base for the character, whose adaptive significance is hard to conceive. As for variation from population to population, it is not significant and no geographical or temporal trend was observed. Replum apex (Character No. 7). In Lesquerella densipila the replum apex is rounded (Fig. 2) and is scored as 0 while in L. stonensis the apex tends to be anguar and is scored 1. Scoring each plant as 1 or 0, approxi- mately 60% of all plants of L. stonensis are found to have pointed apices (%-0.56). The artificial hybrid populations are intermediate with an average index value of 0.23. Natural hybrid populations are tremendously variable for this character. One population has essentially all plants with rounded apices (index value .01) while 5 of the twelve populations have practically all plants with pointed apices (index values 0.93-0.98 ). Two populations have an index value similar to the L. stonensis parent (0.42 and 0.59) while the other four populations have values similar to the arti- ficial hybrids (0.29-0.22). Most populations either resemble the L. ston- ensis parent or transcend it in the pointedness of the replum. It is hard for us to see any selective reason for this pattern of variation, particularly in view of the fact that no geographical trend is apparent. A similar lack of pattern was observed for this character in the hybrids between L. lescurii and L. densipila, where we concluded that there probably was a large environmental component influencing the expression of this character, and that it was a poor indicator of hybridity. The same appears to be true here. Replum base (Character No. 8). What has been said of the replum apex applies as well to the base. Furthermore there is no substantial difference between the parent species populations in this character. Trichomes at the base of the style (Character No. 9). Both species, and consequently their artificial and natural hybrids as well, have trichomes at the base of the style. Trichomes on the outer surface of the valve (Character No. 10). Both species have trichomes on the outer surface of the valve and so do the hybrids. Trichomes on inner surface of the valve (Character No. 11 ). Neither of the species ordinarily have trichomes on the inner surface of the valve nor do the artificial hybrids. However, between one and 18 per cent of the natural hybrid populations have plants with at least some trichomes on the inner surface. Two of the populations of Lesquerella densipila studied had up to 4 per cent of their plants with some trichomes on the inner surface. That the genetic potential for trichomes on the valve interior is present in Lesquerella is shown by the fact that it occurs consistently in 44 ROLLINS AND SOLBRIG L. lescurii. However, we do not know the exact source in the case of these hybrid populations. Trichomes on replum (Character No. 12). Both species have trichomes on the replum and so do all the hybrids. Silique apex (Character No. 18). Both species have rounded or slightly depressed apices. Those of Lesquerella stonensis are somewhat more depressed than those of L. densipila. Both the artificial and the natural hybrids tend to have apices slightly more pointed than their parents. In general, the populations are quite uniform for this character. Infructescence (Character No. 19, 20). The infructescence of Les- querella densipila is more compact with shorter pedicels that are closer together than those of L. stonensis. The artificial hybrids resemble L. densipila both in pedicel length and in the number of pedicels per linear centimeter of infructescence. The artificial hybrids are slightly more inter- mediate, particularly in the length of the pedicels. It is possible that the more compact infructescences of the hybrids than expected is due to slower growth of the hybrids. ummary of character by character variation. Lesquerella densipila resembles L. stonensis more than it does L. lescurii and consequently there are only a few characters that discriminate L. densipila and L. stonensis. Other than flower color, they are mostly quantitative characters. The only pattern observed in the populations of natural hybrids is a tendency of many of them to resemble L. stonensis more than L. densipila. This is presumed to be due to an uneven overall contribution of the two species to the hybrid swarms. Circumstantial evidence suggests that L. stonensis was present along the Stones River in pure form before L. densipila entered. The natural hybrid populations studied appeared not to have become fully intermediate between the two species because of a predom- inance of L. stonensis in the valleys of the river system. No differences between upstream and downstream populations were observed. Correlation coefficients. Analysis of the correlation polygons of the Lesquerella densipila < L. stonensis hybrid populations reveals the fol- lowing: 1) no population shows high levels of correlation; 2) in several populations the character by character correlation is al- most negligible; 3) no geographical trend is visible. In our opinion these results reflect (1) the greater genetic similarity of Lesquerella densipila with L. stonensis when compared with the genetic similarity of L. densipila and L. lescurii, and (2) the lack of the geograph- ical mixing effect observed in the L. densipila L. lescurii hybrids. In this respect it reinforces the results obtained from the morphological analysis. Hybrid index. A simple hybrid index was constructed and populations scored accordingly (Table 18). It confirmed what the analysis of char- acters had already shown, namely, that the hybrid populations are either HYBRIDIZATION IN LESQUERELLA 45 TABLE 18. HYBRID INDEX NUMBERS FOR CHARACTERS IN WILD POPULATIONS OF L, DENSIPILA L. STONENSIS Character Number® Pop. site 1 o 3 4 5 6 if 8 a4 Total 66-6 0.50 025 1.00 0.50 0.50 0.00 1.00 025 66-7 0.25 0.00 0.00 0.50 025 0 0.00 1.00 025 275 6 050 025 0.00 050 025 050 0 1.00 0.50 L. densipila=1 L. stonensis= *For character number code see Table 1. intermediate or they tend to resemble Lesquerella stonensis more than L. densipila. The only exception was population 55-1 from near the junction of the East Fork and West Fork of Stones River which was quite similar to pure L. stonensis. The index also confirmed that there is no detectable geographical pattern, or for that matter, any pattern from year to year. LESQUERELLA DENSIPILA L, STONENSIS—SUMMARY The pattern of variation of the populations of this hybrid is more in keeping with the classical model of hybridization. The populations are either intermediate or they show a greater resemblance to one parent than the other, due presumably to backcrossing and introgression. From our experience with these particular populations, we cannot add anything of significance to the well-known model of interspecific hybridization. Genetic Coherence Segregations within hybrids between contrasting ecotypes of the diploid species Potentilla glandulosa (Clausen and Hiesey, 1958; 1960; Clausen, 1967) indicate that the traits distinguishing ecotypes are regulated by small systems of genes that have additive, inhibitory, complementary and epistatic effects. The individual genes that govern each character are not assembled on one or two chromosomes, but are distributed among several, if not all the chromosomes of the genome. Consequently, each chromo- some carries genes that contribute to the phenotypic expression of several characters (Clausen, 1967). This phenomenon, observed by Anderson (1949) before and called the “recombination spindle” by him, provides a system of hereditary components interlocking into a kind of coherence. Genetic coherence is a mechanism that favors the resegregation of the parental genotypes, with a greater frequency than would be expected if free recombination were operable. According to Clausen, it represents a cohesive force in evolution. The method employed for measuring genetic 46 ROLLINS AND SOLBRIG coherence is to determine the nature of the correlation of pairs of segre- gating parental characters as compared with free Mendelian recombina- tion. However, this measure requires an extended analysis. Statistically, measures of correlation and a test of their significance demonstrate only that two variates of a given sample vary or not in such a way as to main- tain a definite relation to each other. They tell nothing about the cause of the relationship. Assuming a correlation between two varieties (even when statistically significant) to indicate a relationship between the variates is a most common fallacy. Correlation by itself, then, is not a demonstration of genetic coherence. Another point worth emphasizing is that the coefficient of correlation r is dependent upon the magnitude of the difference between the means x and y, as well as the deviations of the x, and y; from the means. Conse- quently, if the mean of two characters differs by, let us say, only 0.1 mm., even if the two variables measured are completely independent variables, measure of r will be smaller and possibly less significant statistically than a case where the means x and ¥ differ by let us say 100 mm., and the two variables x and ¥ are only partially correlated. This presents an added problem when using the correlation coefficient to measure genetic cohe- sion. On the other hand, if genetic coherence exists, certain predictions can be made: (1) since coherence is a phenomenon that tends to keep the genotype intact in a segregating hybrid population, it should not be ex- pressed in a stable, non-hybrid population where presumably the genome is working harmoniously. Therefore, no correlation should be detected in such a situation, and any correlation found in the parental population should be attributed to other causes. Furthermore, cohesion should be strongest in an F, hybrid population and weakest in a stabilized hybrid population. Analysis of the correlation polygons indicates the following: 1) The parental populations show few correlations, 2) The artificial F, hybrid populations when taken as a group show a very high degree of correlation. 3) The individual F, families do not show as strong a correlation as the lumped F,. This is perhaps due to the smaller size of the sample, but possibly it represents a sample error in that only the offspring of one particular F, cross is being analyzed. 4) The field situation is more involved, but in general it indicates the following: (a) populations that resemble strongly one of the parents _ show a small degree of correlation; (b) populations that are hybrid inter- " mediates show more correlations; (c) populations suspected of being “primary” hybrids (64-1, 55-1) show the strongest correlation of all. that coherence acts as a conservative force in evolution that tends to slow down recombination. HYBRIDIZATION IN LESQUERELLA 47 On the other hand, an interesting phenomenon has been detected. Certain of the hybrid populations of Lesquerella densipila < L. lescurii morphologically intermediate or slightly similar to L. lescurii no longer show genetic coherence, and characters recombine independently. Pre- sumably, in these populations recombination has proceeded over a sufficient number of generations so that the original parental chromo- somes have been broken up by crossing-over, and most linkage combina- tions between L. lescurii and L. densipila genes are present in the popula- tion. Such a population presumably has now acquired unique linkage combinations and if crossed to either parent would show genetic coher- ence in favor of the hybrid combinations. Such may be the beginning of the process of hybrid stabilization, which would afford a certain amount of protection against occasional gene flow, particularly if lethal or semi- lethal genes were to accumulate in the population. SUMMARY AND CONCLUSIONS The detailed analyses of the two hybrid swarms presented above reveal an involved pattern of variation. The original model postulated that at the junction of the streams (Arrington and upper Harpeth in the case of the Lesquerella densipila < L. lescurii hybrids, and West and East forks of the Stones River in the case of the L. stonensis x L. densipila hybrid) a hybrid swarm was formed. It was further postulated that seeds of these F, and F, hybrids were carried downstream by the river during spring floods and deposited at near localities below, where hybrid populations arose. Seeds of these hybrid swarms were deposited at new localities below in subse- quent years, and so on. If this were the case, the hybrid material formed at the upper reaches of the river would move down river stepping-stone fashion and presumably natural selection would be acting on the hybrid populations all the time. One reason the present study was undertaken was to determine whether the original model was adequate, and to ascertain the course of change of the hybrid swarm if some effects of natural selection could be shown to be taking place. Our study has shown that the original model was incomplete and to some degree misleading. What we did not fully appreciate earlier was the fortuitousness of the input of propagules from each of the parental species. Rather than a true mixing of the genotypes of the two species where the rivers meet, with selection working on the hybrid swarm to narrow the variation, the two genotypes hybridize gradually down the river so that it is only fairly far downstream that intermediate hybrid swarms are found. The uneven seed contribution of each species from year to year is related to local conditions. In most years, one or the other of the upper tributaries of the main stream floods, bringing with it seeds of the pure species that grows near the river. These seeds are not only deposited at the junction of the two rivers but in diminishing amounts at localities down river from the junction. The following year seedlings of that species may germinate at these sites, together with seedlings of the 48 ROLLINS AND SOLBRIG other species and of hybrids that grew there before. Hybridization takes place, but since the number of seedlings of one of the species may be more abundant than that of the other, the resulting hybrid swarm is not truly intermediate. In the following year, the same pattern of flooding might take place so that more seeds of the same species are deposited, or it might be reversed, and the locality is flooded with seeds of the second species. The result is either a more or less intermediate hybrid swarm, or more likely one that is strongly influenced by one or the other species. Seeds from these upstream hybrid populations are carried by the river during spring flooding to localities further downstream, Here they germ- inate and cross with the hybrids that are present there. The arriving seed will be more like one parent some years, more like the other in other years. The resulting hybrids will be more intermediate than the arriving seed. Consequently the further downstream the more intermediate the hybrid populations become until true intermediacy is achieved. We cannot emphasize too strongly the fact that a very dynamic situation obtains along the Harpeth River. Not only are the hybrid populations subjected to the effects of new inputs of seeds of a diverse nature from upstream, but the sites where they grow are being changed from one kind of agronomic practice to another. Different crops are planted in different years or the land is shifted from cultivation to grazing, etc. Such changes can be a large factor in expanding or contracting or even obliterating a population on a given site. Some of the significance of the hybrid picture of the Lesquerellas of the Central Basin relates to the overall role hybridization has played in the origin and differentiation of plant species. It is easy to visualize similar patterns having occurred repeatedly over geological time. The movement of the hybrids away from the parental species provides the mechanism which would enable a new taxon different from either of the progenitor species to originate. LITERATURE CITED ANDERSON, Epcar. 1949. Introgressive Hybridization. John Wiley & Sons, N. Y. 190 p: Ciausen, J. 1967. Biosystematic consequences of ecotypic and chromosomal differ- entiation. Taxon 16:271—279. —— and W. M. Hresey. 1958. Experimental studies on the nature of species IV. Carneg. Inst. Wash. Publ. 615, 312 pp. —. | e balance between coherence and variation in evolution. Proc. 94-506 ‘ 1955. The auriculate-leaved species of Lesquerella (Cruciferae). Rhodora 57:241-264. - 1957. Interspecific hybridization in Lesquerella (Cruciferae). Contrib. Gray Herb. no. 181:1-40, - 1963. The evolution and systematics of Leavenworthia ( Cruciferae ). Contrib. Gray Herb. no. 192:3-98. THE USE OF ELECTRONIC DATA PROCESSING METHODS IN THE FLORA OF VERACRUZ PROGRAM! ARTURO GOMEz-POMPA AND Lorin I. NEvLING, JR. We are living and working in a period when floristic studies far exceed any similar emphasis of the past. They comprise a very significant portion of the systematic studies now underway and appear to be steadily increas- ing. Even a decade ago floristic studies were not fashionable and, in fact, received considerable criticism from many quarters. It is not intended to imply here that floristics are fashionable today. However, the realization that man is destroying the world before he has taken the opportunity to know and understand the diversity of plants occurring on the earth’s sur- face has had a profound influence on the thinking of many systematic botanists. Many systematists feel a strong moral obligation to document this diversity as best they can before it is too late, and find that floristics offers the most expeditious way to do so. The production of a flora for a given geographical area is a much more difficult task today than it was even a few decades ago. The pressure of time has been mentioned but several additional factors contribute their burdens. Among these is our steadily increasing knowledge of the processes of organic evolution that have to be dealt with and the desire, if indeed not necessity, of understanding the “biology” of the plant groups under study. Also, there are new and urgent demands for increased accuracy of all the data included in the ultimate treatment. Scientific knowledge grows unevenly and at some point it is most helpful to have this brought together in a synthesized? form. A flora can be the vehicle to accomplish this. It should not only be a work of synthesis, which is in itself an accomplishment, but it should also help spawn new and original research. In this way, a floristic study can help to build a new plateau of knowledge upon which other researchers can enlarge. Floras should never be considered as the end point of concern with the plants of a given region but as the beginning. STEPS IN THE PREPARATION OF A FLORA During the-initial consideration of the Flora of Veracruz, we made a 1Flora of Veracruz. Contribution No. 10. A joint project of the Instituto de Biologia, Universidad Nacional Auténoma de México and the Arnold Arboretum and Gray Herbarium of Harvard iver- sity. (See i v iol. Univ. Nal. Autén. México, a): 4, 9. 1071) partially supported by grant GB—20267X from the National Science Foundation to the junior author e authors wish to their appreciation to m lea: for helpful suggestions, corrections, and criticisms of this m 2Synthesized is used here in the sense of including original data gathering in addition to summarizing previously known information. 49 50 GOMEZ-POMPA AND NEVLING general analysis of the usual steps in the preparation of a flora. The basic data and procedures employed seem to vary somewhat from one project to another, just as content and quality vary. In spite of this, most floristic studies do share some common data sources, procedures and problems, each of which is discussed briefly as background information. Ideally, the ultimate flora would contain all the information known for each taxon regardless of the subject. At the present time, there are serious obstacles to this, one of which is gaining easy access to diverse biblio- graphic data sources. There are, of course, other problems inherent in gathering data to be used in a flora. Otherwise one might expect that most of the flora of the world would have been covered by this time. The bibliographic problem manifests itself in several ways. All floras begin with some kind of basic bibliography which probably will be most useful at the beginning of the study of a group and again toward the end. These two “peaks” of usage serve somewhat different functions. The first involves an initial search for papers dealing with taxonomy, nomenclature, distribu- tion, etc. These papers, along with specimens, form the basic building block for the development of plant descriptions and keys. The second bibliographic “peak” is usually reached as the rough framework of taxo- nomic structure has been completed. The usage here tends to be more specialized as the topics tend to be more specialized; for example, the search may be for detailed environmental data, ecology, chemical com- pounds, etc. One problem inherent in the bibliographic process is relative- ly simple in theory and difficult in practice: to generate a more or less complete bibliography. The second part of this problem, ie., the special- ized reference, is more complex because considerable individual judgment must come into play as to whether to include a paper or not. The reliability of published information always must be assessed. It is often difficult to determine which taxon the data should be assigned to; for example, many ecological and some phytochemical studies are not backed by voucher specimens. This is also the case with the earlier works in cytology and anatomy. A degree of uncertainty should be anticipated in all descriptive material whether vouchered or not. The ultimate responsi- bility for proper assessment lies with the individual worker. Another bibliographic problem concerns the extent to which references are cited in the published flora. The citation of bibliographic references might be considered a part of the description—those dealing with nomen- clature certainly are—but many floras have chosen to treat the remainder in a separate category. It is our impression that bibliographic references add significantly to a flora because they furnish the user with a baseline of information to interpret the floristic treatment. Furthermore, we feel that often there is a definite correlation between the number of references cited and the final quality of the work. One should be selective in this matter, however, and not include a reference solely because a particular species is mentioned. It is not especially helpful to have a bibliography ELECTRONIC DATA PROCESSING METHODS 51 more extensive than the text. In the final analysis, the most difficult prob- lem involving bibliographic resources is to maintain them in such a way as to keep them accessible. This task is compounded as the amount and diversity of information increases and it can be the main reason for not including information that is important. We doubt that it is necessary to direct much attention to the methods of obtaining original data resources, that is, living plants and herbarium specimens, as most of us are familiar with this aspect of systematic re- search. Suffice it to say that it is necessary to have a collection of specimens which is representative of the morphological variation and geographic distribution of the taxa. The combination of the two basic resources, bibliography and speci- mens, leads to the common procedure of preparing keys and descriptions of taxa (admittedly the content varies considerably from flora to flora). In most, however, the description constitutes the main mode of data presentation and accordingly forms the body of the flora. Descriptions of each species consist of some or all of the following categories: the scien- tific name and some or all of its synonyms, including typification; a morph- ological description based on data from literature and specimens; diverse data such as chromosome number, phenology, chemical compounds, uses, local name; ecology; illustrations; geographic distribution in the area and in a broader perspective, sometimes with maps. When specimen citations are included they most commonly consist of the collector's name, number, and the locality. In more comprehensive floras, herbarium abbreviations may be included as well. In our opinion, specimen citations add greatly to the usefulness of any flora but most particularly to those of tropical areas. It is the usual practice to develop generic descriptions from the species treated, unless the generic descriptions are meant to be worldwide in circumscription. Family descriptions are generated in a like manner from generic descriptions. Uniformity of descriptions, even within a single flora where a certain amount of format change may occur, is difficult to obtain. Often this means that descriptions are difficult to compare. The preparation of keys to all taxa is of prime importance. The keys are necessary to provide access to the proper data (description). Often the keys are the most used part of a flora and they should be constructed with the greatest care and with their frequency of use firmly in mind. Keys are one of the most variable portions of a flora because different criteria may be used for different families; for example, keys based on fruit, flower, leaves, or pollen may have a different diagnostic usefulness in different genera and families, and consequently some are stressed (weighted ) while others are not. The most reliable and easily used characteristics should be chosen for the keys. A related problem is that of compatibility with other floras, i.e., whether they can be used in conjunction with one another without serious data conflict. Frequently a flora of a particular region becomes the base for a SY GOMEZ-POMPA AND NEVLING flora either of a smaller area that is a portion of it, or a larger region that includes it. Even though there is a general awareness of this, it is a rare flora that has been planned for such contingencies. The greater the amount of data included in a given flora, the greater the probability that a useful data base will be established. Finally, the most frustrating procedures in flora preparation are those belonging to all the categories mentioned above, which might be thought of as being routine. Some examples of routine operations are: preparing duplicate labels for recently collected specimens; bibliographic search, particularly in peripheral areas such as the environmental aspects of the area; determination of localities for mapping purposes; and the elabora- tion of specialized catalogues, such as local names. In recent years several attempts have been made to solve some of the problems mentioned, with varying degrees of success, by using automatic data processing techniques. After an examination of these efforts, we came to the conclusion that they offered great possibilities, not fully explored, to free the professional botanist from many routine operations. The possibil- ity of increased efficiency, without greatly increased costs, appeared to be a realistic and attainable goal. Several methods were explored before a practical system for the Flora of Veracruz was developed. The decision to use electronic data processing methods for our flora was made in 1966, when the first pilot projects were completed. These included information retrieval of herbarium specimen data and automatic map production for range distribution. The results of these pilot projects were presented at our Mexico City Symposium on Information Problems in Natural Sciences in 1967 (Gémez-Pompa & Olvera, 1967; Scheinvar, Gomez-Pompa & Alonso, 1967). It is worth mentioning that during the initial developmental stages of our work other floristic projects were stim- ulated by our approach, results, and personnel (Ahumada, 1967). We are encouraged that in these few years the quality and number of other floristic programs using electronic data processing methods have increased considerably. The approaches for different applications have also in- creased and diversified (Morse, 1971; Bestchell & Soper, 1970; Scheinvar & Gémez-Pompa, 1969; Crovello & MacDonald, 1970). It is important to realize that many of these techniques were inspired by the pioneer work of Perring (1963) and Soper (1964), which are, to our knowledge the first large scale attempts to use EDP methods for specific floristic projects. Rogers’ work on Information Retrieval for Taxonomy (1960) should be mentioned here, even though not floristic in its approach, as it presents many of the basic ideas used for EDP-IR application to floristic studies. It is also worth mentioning the effort of the Smithsonian Institution in this field (Squires, 1966). Data PROCESSING FOR THE FLORA OF VERACRUZ The possibilities of the kinds and amount of data to be provided in a floristic work are endless and may range from the collection number of a ELECTRONIC DATA PROCESSING METHODS FLOW CHART OF VERACRUZ FLORA | HERBARIUM SPECIMENS | # See rs 5 4 CITATIONS OF SPECIMENS 3 SPECIALISTS » INPUT ee FROM Data Processing ! HERBARIUM LABELS (EXCHANGE OF SPECIMENS) OTHER | RESEARCH PROJECTS MAPS (DISTRIBUTION OF TAXA CORRELATION OF DATA) | ra (LOCAL, GENERAL) (LOCALITIES, COLLECTORS, ETC) CARD FILES (BIBLIOGRAPHY, CITATIONS, FAMILIES, ETC.) ANSWERS TO QUERIES Monographers Publications Fic. 1. Flow chart of the Flora of Veracruz. 54 GOMEZ-POMPA AND NEVLING specimen from a certain locality to the number of ovules in the ovary of a certain species collection. The main problem is to decide what is wanted and needed from the system so that it may be developed, and to concen- trate on the data required. Another important aspect to be considered is the time involved in gathering the necessary data and their preparation for computer use. In our flora program, we have made the decision to use only the data for our most immediate needs; other data, for future use, has been included in free text so that they may be used if objectives change. The three main sources of data for any floristic work are: biblio- graphical material; field data along with specimens recently collected; and, specimens in herbaria (Fig. 1). We will describe the kinds of data we are processing and storing, how we are using them, and the tech- niques.* During the development of the computer programs we have had the generous and enthusiastic support of the personnel at the Computer Center of the UNAM, Centro de Investigacién de Matematicas Aplicadas, Sistemas y Servicios (CIMASS ). We are most grateful to the Director, Dr. Renato Iturriaga, for placing facilities at our disposal. Data from the literature. As the monumental annotated bibliography of Ida Langman (1964) for the Flora of Mexico was available, including references complete to 1950, the period for us to cover initially was self- evident; additional citations from 1950 related to the flora or those older but of great utility in the preparation of a flora. As the project includes the study of the plant resources and the environments of Veracruz, with the flora as one facet of the study, a great variety of citations are being included. The system that we are using includes the following steps: 1. Selection of the citation. Several approaches are used, including the examination of new periodicals, indices, personal card bibliographies, and retrospective examination of certain journals. 2. Selection of key words which describe the contents of a reference to be stored and by which the citation can be recalled from the computer. These include such words as: revision, keys, illustrations, chromosomes, etc. After the key word or words, a short free text can be added if desired; for example, illustrations: includes maps, photographs, and line drawings. 3. The entire citation, including journal citation presented in abbrevi- ated form in accordance with the B-P-H system (Lawrence et al., 1968) together with its analysis and free text, is given to the bibliographer who annotates the particular family, genus, or species which is the main subject of the citation. 4. The complete citation with key words is given to a key punch operator who punches a variable set of IBM cards (Fig. 2) for each citation in a fixed format with free text. This makes it possible to include additional information for each entry and at the same time the fixed format facilitates the rapid retrieval of information. 5. The punched cards are edited and the information is transferred to *Details of the format and programs are not included here. ELECTRONIC DATA PROCESSING METHODS get ae z Lees See ee ee ee = Saale“ tan ee cjvt 62 22 02 BI GL LPM GS mE dh i Ot 6 GGG BHG ECE SE GES SHEE HG 6E66 6666 6666 GSS i OE GAVE Le OC Gove Cz Cz tZpOe GI Bt L9G wt CE ut On 6 gia aoa ete cect eee tedeecderedeerdeecdecemeecdeecdercdeecderedccedecedeeedeeedecedeeedeeed ‘oe a US bl Ui (5M Gi FE ALG St9S Oh Ph BS , ik 4 it we a ok i iyt i] 9999599 999991999 9199991999 91999 91999 91999 919.99 9199991999 91999 919.99 91999 19. 9.9.91999 919 9 9 819 SGSSSTTS GES GIS SSSSSSSsesaieGggsSSSIESSSISSGGigGSGiggSoIESSGisS SisggG Gogicg Glcgs ” § O96! auf Qf wi f/ D> cb 99 Solve CF EF ib ieeseecest codecs $6 é9 79 i508 & GEIB MGEEES aid 4 = 2 < sf = i 4 > : > q berry rbhrerrrratiber PRU OPP PHP e rere bbe rbiped phd db heb bbb bb verrirerryesr ie aa vse wwe ||| CECEECECERLCC EEE CECE CECE EEC ERIC CE CCE E EERE RIECE Clee eCttt 7 f ce"oseee ]” CEC e eee elec eee eee e elec eee elec elec eee elle eee eee CULE TYPE E ALLA CLUE C CALCU EEE EAE UA UL UH EAU EEL AE Cubed ofl e0d00 aA MALO DRNOO OOO CHESS SARULOL Io 00 3 ve BSED pes aE Ree og cay tk Senet “Hen i : ae i 1903106 ; kee Spanos ATA ipa nto ay scien sate yet oh B51 So: aa TWD L6H pelea ec dipolpolivesCroqeeesreiestemeaieatey ao sign Serer so oeancreme peyote ia yates 2 es fone acter na wie ic ada eer, eel %, Bitte habe ld ae aera emer ) MANTA eee ath TALIA BA ig ie SiR rskte Ae LRN lk hsb neo t 3 ten is Th Ti 7 MOTE i {fi qi iy hit “is Fic. 2. Complete set of punched cards for one bibliographic entry. 56 GOMEZ-POMPA AND NEVLING BIBLIGGRAFIA PARA LA FLORA DE VERACRUZ TOMLINSON» Pe Be1969, ON THE MORPHOLOGY AND ANATOMY OF TURTLE GRASS,» THALASSIA TESTUDINUM CHYDROCHARITACEAE). IIe ANATOMY AND CEVELOPMENT OF THE ROOT IN RELATION TO FUNCTION.= BULL» MARe SCIe GULF CARIBBEAN 19:57"71, Neola HYDRC B001001 ANATOMIA ECOLOGICA RAICES BIBLIGCTECA INST. GEOL+ UNAM, Fic. 3. Printed bibliographic card. magnetic tapes and discs for processing. The first products the botanists see are bibliographic cards that are printed in any quantity either by the computer or by an electronic printer from the magnetic tape or from the punched cards (Fig. 3). 6. Periodically a selected bibliography for any of the key words, taxa, or any combination of them are printed as desired ( Fig. 4). 7. A basic bibliographic book is printed by the computer. It consists of all the citations listed in numerical order (each citation is assigned a sequential number ). This serves as a reference for queries to the computer: the answer will refer to a number in the basic book. The main reason for retrieving in this way is economic, but we are ready to develop a more sophisticated retrieval system that can be derived from files that have been constructed for this purpose. 8. Using this system we can print specialized bibliographies for our collaborators as required. Or, if we wish, we can print a cross index for the Flora of Veracruz that can be published more widely, in the same way as other abstracting publications. In our first computer generated bibliog- raphy (1971) we included more than one thousand citations. This has proven useful, not only for studies of the Veracruz flora, but for other projects as well. There are several other possibilities in bibliographic search that we are exploring. Among these is the citation of revisions in which specimens from Veracruz and the herbaria where they may be found are cited. Because of updating possibilities in the herbarium, we included a key phrase “specimen citation” to identify these references. Data from recent collections. We are giving emphasis to new field collections, even though many collections have been made previously in Veracruz. Unfortunately, the previous coverage is uneven as collecting was limited to a few heavily collected areas. Because of this and the need for new documentation, especially for the distribution of the taxa, a con- siderable number of specimens are being collected. ELECTRONIC DATA PROCESSING METHODS 57 As most of these collections are made under our auspices, we can control the label format and data presented. A field label was designed that meets our requirements but does not place an additional burden on the collectors (Fig. 4). The field labels consist of an original and two carbons; the original is kept in a master file and one carbon copy is allotted to each of the two cooperating institutions. The system for specimen data flow requires the following steps: 1. The collector completes the field labels for each collection number. In general, the data included are the same as those normally recorded. The only additional item necessary is to code the locality according to a grid coordinate system (Gémez-Pompa & Nevling, 1967). This particular coded entry permits the automatic printing of distribution maps. 2. The plants, with accompanying field labels, are identified. 3. The identification is recorded in the master file and the label data are prepared for card punching. This involves the assignment of a number for each collection and a numerical code for the genus and species. The genus number is derived from Dalla Torre & Harms (1900-07). If the genus is not included in that publication, a number is assigned arbitrarily, following the last generic number of the family. The species number is completely arbitrary and we keep a check list of species names with this information (Fig. 5). All important data fields must be filled, e.g., col- lector’s name, number, and date, etc. 4, The completed label then goes to the key punch operator and the information on each label is transferred to IBM cards. Each card is numbered in sequence, and for each entry on the label there is a card (Fig. 6). All information is punched in free text within each field which makes it simple for the operator and for revision. The final item is a replication number containing the number of duplicate labels to be printed. Menge Lean PAIS | ESTADO omeewedes LOCALIDAD : MAPA LATITUD LONGITUD P Cen mechs EGETACION ; rs te, palm — INF. AMBIENTAL \ es, Jr f SUELO ASOCIADA ABUNDANCIA | TAMARO AN. sa bata | OTROS DATOS os FRUTO eee > NOMBRE LOC. oe coL g USOS | ET. ° COL. ' No Fic. 4. Field label. 58 GOMEZ-POMPA AND NEVLING RANUNCULACEA ANEMONE MEXICANA HBKe RANUN 21 1 CLEMATIS CARACASANA DC. RANUN 22 6 CLEMATIS DIOICA L, RANUN 22 1 CLEMATIS GROSSA BENTH, RANUN 22 2 CLEMATIS SERICEA HBK, RANUN 22 3 DELPHINIUM AJACIS L, RANUN 19 1 RANUNCULUS CYMBALARIA PURSH RANUN 26 7 RANUNCULUS DICHOTOMUS MOC. g SESSE RANUN 26 RANUNCULUS GEOIDES HBK, RANUNCULUS HOOKERI SCHLECHT, RANUNCULUS MACRANTHUS SCHEELE RANUNCULUS PERUVIANUS PERS, 1 RANUN 26 2 RANUN 26 6 RANUN 26 3 8 a 5 RANUN 26 RANUNCULUS PILOSUS HBKe RANUN 26 RANUNCULUS SIBBALDIOIDES HBK, RANUN 26 THALICTRUM GIBBOSUM LEC, RANUN 28 1 THALICTRUM STRIGILLOSUM HEMSL, RANUN 28 2 Fic. 5. Family listing from checklist prepared by computer. 5. After editing and any necessary revision of the cards, a file is gen- erated according to a prepared program, and the labels are produced for the duplicate specimens ( Fig. 7). 6. With the information stored on the files it is possible to do many things with the aid of the computer: update the checklist; construct listings of collectors, localities, collector's numbers, etc. With appropriate software, a closed information system can be prepared if needed. We intend to print a basic collection book periodically with the collection data arranged numerically with a cross index at the end for the most important items. 7. The basic collection book serves as the base for queries to the com- puter. The answer is obtained faster and more cheaply by referring to the numbers in the book. This method of information retrieval is economical: the coding is done by the computer, the search for specific information is faster, it eliminates the inclusion of information that is not reliable; all e information is kept for future use in case a need for it develops. More sophisticated systems can be developed from this one if desired. 8. The use of the locality files (gazetteer) permits the automatic print- ing of distribution maps by a plotter. 59 ELECTRONIC DATA PROCESSING METHODS e238 o 5 a, 5 ” a Ss ro} rok nl a, There are numerous possible , such as vouch mens ific research 1 studies that are being done tten especially for that purpose ways of presenting data in relation to spec i If the data are to be used for a spec 9. can be wr mens for chemical, genetic, and ecologica the flora program or that are ment 1lves ioned in the literature. Our system is . Errors can be corrected in the arch ic citations open and we can add as many data cards to any collection as necessary, luding bibliograph at any time and will assist in keeping the data current. mc 2 eee We CI vndizee cw shame ove le UT aa RE ud 189 19 99 59/99 7 16j0a nie : Zl iF Golpy ov F iso oC OE Tea ie OF 62a IE Tis aide ot Of 16 BG 666.66 5/666 Galams soe GEVatad rae tiaeea eas Ry Eun citihivaraiiaacceets B08 Gt Bi Li]9L GL OL CLjce tf OF 69189 19 99 9 £9 79 19109 66 8S (E195 G¢ PS ESIZS 19 OG Lo OF Shiph Ce ty ib]Ob GE WE LE Or Se bE El ze Ue OF 2 GE GE ws 02 CE AZ]? Gi Ob C1; 94 Gi PL ELPA Mt OL Bt His ic 8 8/889 81888 8/888 81888 8/8 8 8 818. B G08 8 8888 88 88 G18 88 81988 88 5 88/8 Bk 8 8B 88k 8 R88 8 RB eS LEAL CCLRC CALC ee UC ek CELEL CELL EL }O8 GL BL LE}9C SL PL CLIZL AL OL G9}89 19 99 S9I¥9 C9 co GS 8S (9195 9G PS CI Is OS LY Ob Gb/d Eb Zp inlOP GE GE Lel uf GE we Cece Ie OF lac «2 90 Szjve tz ze GEM CO St mL Cet OL G HISHIIITIGISITGIIIGIINGIIIIIT GIG SI99Gg salsssslegssieresssdsssgeag leas sles sales: SSSgis"SGigSscigggciggssiss¢s SS959¢5 SSSsiss"isscsiss ™ SSSgSS” GiSScaigssciggcs gos” OG Gt BL LE9C Se be ELTA Ue DL BBD (9 99 Mudaene nuttin ua Ob Ly Sp Stee ee Zo ivlod BE FE LAGE SE Fe ree Le oF az (2 98 Sve er 22 | oa alo st Miele is © vont 3 rer oyy rors paaeere ey ave narra prvvp yyy he & be & BS he * 1] ’ i & PUVVIVV PV PPP Po Vis ve ERCEEC LEE LE ECE CEE Cece clece ¢ OL SL ML ELfZL We OL 689 19 99 SO}v9 C9 29 I9]0965 BS Lol 4G CS HG CAles 19 06 LE ezczzizz22 Sheree diet tgs teh ierpe or beeelez ze tit sb RE AD oor Gai. ole --* ee Re eR a Le ‘ ri Q00nnnnn Aan” "nA eo aad f qaoanondletg. 0" 0 010 0 Ark es ee ae as "MET: M2 ZIT LSM, aa 8 ano mega SINHA aera Us0 1344 ong ects Baal Seen eeeareccoee ertaeds eoahea teste aS eae 3H Hi oan iets pa ese mn poatae ae S0SNTMEAD — JgSTOOW Fic. 6. Punched cards for one specimen label. 60 GOMEZ-POMPA AND NEVLING The most important source of data for any floristic project is the speci- mens deposited in herbaria. Representation of a particular floristic area varies from herbarium to herbarium and this poses a serious information restriction. The organization of the herbaria of the world is such that it makes it impossible to use all the information (specimens) available from an area. We have begun a retrospective search for specimen data in our respective institutions (UNAM and Harvard) and in herbaria with personnel interested in cooperating. This procedure represents a compro- mise but it is hoped that the coverage can be increased gradually. After a few pilot trials (Scheinvar et al., 1967; Gomez-Pompa & Olvera, 1967) we decided that the simplest procedure was to follow the same format that was developed for the field labels of current collections. As done previ- ously, the data are presented in free text. This method overcomes the difficulties normally encountered in dealing with multiple languages. Two main problems are posed in organizing these older materials. The first is the locality names, as some of the older specimens are attributed to place names no longer in use, or which are known under several alternate spellings. To solve this problem, we compiled a gazetteer (Gémez-Pompa & Nevling, 1970) including localities of botanical interest, coded according to our grid system (Fig. 8). The second problem is the accuracy of the identifications: at the moment we are not attempting a broad-scale revi- sion of these. Instead, we are capturing the data on the labels, and changes in identification can be done at a later time from the archives via teletype. By using this approach, we will have a large quantity of data which will be useful in family treatments, the future Flora of Mexico, and other related research projects. In the future, we hope to uncover and correct other data from old voucher specimens, such as misidentifications. The possibilities of using the herbarium data for different purposes have been mentioned and dem- onstrated in other publications (Scheinvar & Gémez-Pompa, 1969; von Reis, 1962). A pilot project is being done at the Royal Botanic Garden, Kew, with some selected families to test the feasibility of expanding our coverage of Veracruz specimens. This test has been made possible through the gen- erous cooperation of Ms. Laura Vit under the sympathetic and technical supervision of Dr. David Hunt. OTHER SPECIALIZED PROJECTS Other research projects in the Flora of Veracruz series also are contrib- uting to broaden the coverage of the data bank. One example is the study of the climatic changes in Veracruz done by Soto ( 1969) with the aid of the computer. All the meteorological data used for this study is in machine readable form and will be helpful in looking for correlations of plant distribution with certain climatic factors. On a broader scale it should be helpful in more precisely correlating vegetation and species with climate. szaynduioo Aq paredaid sjeqe] eqwordnd *Z ‘org INSTITUTO DE BIOLOGIA HERBARIO NACIONAL DE LA UNIVERSIDAD DE MEXICO (MEXU) NO+RECe ING £001826 COMpOSITAE EUPATORIUM COLLINUM D,C, CATEMACO CARRETERA COYAME MAPA 22.0 57.0 ALT. 300M ACAHUAL VEGeSECUNDARIA SUELO ARCILLA Y GRAVA oN BEAULAR HIERBA REPADORA TAM,2.50M NUA LAS FLUORES SON AROMATICAS BLANCA 03/02/1969 COL.GUADALUPE MARTINEZ CALDERON NO. 1648 INSTITUTO DE BIOLOGIA HERBARIO NACIONAL DE LA UNIVERSIDAD DE MEXICO (MEXU) NO«RECeING. bod1d80 COMPSOITAE EUPATORTUM COLLINUM Oc. CATEMACO CARRETERA A COYAME MAPA 21.5 55.5 300M ALT. ACAHUAL VEGsSECUNDARIA SUELO ARCILLA Y GRAVA ABUND. wire HIERBA THEPADQRA TAM.2.50™M AS FLURES SON AROMATICAS BLANCA 03/02/1969 COL.GUADALUPE MARTINEZ CALUERON NO. 1848 INSTITUTO DE BIOLOGIA HERBARIO NACIONAL DE LA UNIVERSIDAD DE MEXICO (MEXU) NOeRECeINFe 0001826 COMPOSITAE EUPATORIUM CULLINUM 0,C, CATEMACO CARRETERA COYAME MAPA gee 57.0 ALT. ACAHUAL VEG SSECUNDARTA SUELO ARCILLA Y GRAVA ABUND+ REGULAR HIERBA REPADORA TAM,2.50M ANUAL LAS FLORES SON AROMATICAS BLANCA 03/02/1969 COL.«GUADALUPE MARTINEZ CALDERON NO. 1648 NSTITUTO DE BIOLOG HERBARIO NACIONAL DE LA UNIVERSIDAD DE MEXICO (MEXU) NOsRECeINEs D001826 COMPOSITAE EUPATURIUM COLLINUM O.C, CATEMACO CARRETERA COYAME MAPA 22.0 57.0 ALT. v ACAHUAL VEG+SECUNDARIA SUELU ARCILLA Y GRAVA sone REGULAR HIERBA REPADORA TAM.2.50M ANUA LAS FLORES SON AROMATICAS BLANCA 03/02/1969 COL,GUADALUPE MARTINEZ CALDERON NO. 1646 SGOHLAW ONISSHDOUd VLVd DINOWLOATA 62 GOMEZ-POMPA AND NEVLING IvOLo MAPA 67,0/21.0 INOLO ISLA NEL MAPA 66,5/20.5 IGNACIO ALTAMIRANO MAPA 51.0/25.5 IGNACIO DEL ROSARIO MAPA 38,0/26.5 IGNACIO DE LA LLAvVE IGNACIO MUNOS MAPA 26.0/42.0 MaPa 52.0/25.0 IGUANARA MAPA 14,0/65.0 INDEPENDENCIA MAPA 46,0/27.0 INDEPENNENCIA MAPA 44,0/25,0 INDEPENDENCIA MAPA 39.,5/24,5 INDIO MAPA 15,0/53.0 INFIERNILLO MAPA 32,0740,5 INFIERNTLLO MAPA 26,0/40,5 ISLA BLANQUILLA MAPA 33,5741,0 ISLA DE gurRrRos ISLA DE ENMEDIO MAPA 71,0/18,5 MAPA 32,0/43,0 ISLA DE JUAN ROSAS ISLA DE LOS PAYAROS ISLA DE Losos MAPA 50,5/23.5 MAPA 68.0/21,0 MAPA 67,.5/24,0 ISLA DEL FRONTON ISLA DEL TORO ISLA FRIVOLES MAPA 72,5/17.5 MAPA 69.0/19,5 MAPA 68.5/21.0 ISLA MARTINICA ISLA MATACARALTOS ISLA PAYAROS MAPA 70.5/18.5 MAPA 68,5/20.0 MAPA 33,0/41.0 ISLA SACRIFICIOS ISLA SALMENDRA MAPA 33,0/41.0 MAPA 31.5/43.0 Fic. 8. Partial page of gazetteer with grid system coding. ELECTRONIC DATA PROCESSING METHODS 63 CONCLUSION Our experience with using electronic data processing methods for a flora project generally is very favorable. We have proof that it is very helpful and the evidence suggests that it will be even more helpful in the near future. We have been able to produce duplicate herbarium labels, bibli- graphic cards, bibliographic indices, checklists, maps, gazetteers, special- ized data for specific research projects (Gémez-Pompa, in press) and have created a data bank that will be of great usefulness in the future. One of the things we have accomplished is to have a flexible and simple system, which is useful from the beginning. Our system is not expensive (very little computer time is required) and it will be of maximum help to the researchers of our flora who prepare family treatments. Our final objective is to write the Flora and a side product of it will be the creation of a data bank for future programs. Machines used. During the development of our work, we have had access to a variety of machines that have been used to perform different portions of our data processing program. From the beginning we used the computers available at the UNAM computer center (CIMASS): a CDCG20 and a CDCGI15 but in the last three years we have been using a Burroughs 5500 and a 6500. For primary data input, we have used IBM 80 column punched cards. The maintenance of the files has been by means of a teletype unit. We have used an IBM 447 tabulator machine and a UNIVAC printer for inexpensive printing of certain files and for printing labels and biblio- graphic cards when the high speed printer of the computer was not available. For making distribution maps, we have used a CALCOMP machine linked to an IBM computer. This machinery, as well as help and advice, was supplied by the Mexican Oil Company (PEMEX) to whom we are most grateful. The outline of our map was prepared by an OSKAR machine of the UNAM computer center. Programming. Specific programs have been prepared for the Flora by our different programmers in the last few years. During this time we have had the advice of the staff of CIMASS who have shown a continuing interest in our project. We are indebted especially to Juan Antonio Toledo for his advice and dedication to this program. We would be very happy to give advice on this subject to any floristic project interested in our results. REFERENCES CITED umapA, S. 1967. A bibliographic data retrieval system for botany. (Pilot FNA pew pes og oc FNS Report 5. 69 pp. Bescue, R. E. & J. H. Soper. 1970. The automation and standardization of certain herbarium ‘procedures. Canad. J. Bot. CROVELLO, S J. & R. D. Macponatp. 1970. ites of EDP-IR projects in systematics. Taxon 19:63-76. 64 GOMEZ-POMPA AND NEVLING Davia Torre, C. G. pe & H. Harms. 1900-07. Genera Siphonogamarum. Lipsiae. PP: Gomez-Pompa, A. (in press). Ecology of the vegetation of Veracruz. Elsevier Publ., Holland. ——----— & S. OtveRa. 1967. Proceso de datos para la Flora de Veracruz. Restimenes del Simposium sobre Problemas de Informacién en Ciencias Naturales, México, D. F. Dic. 18—20. 21. a & L. I. Nevuinc, Jr. 1967. Mapas del estado de Veracruz. Inst. Biol., UNAM, svete No. 1. Enero 15. (Reproducién en xerox ——————— 1970. Localidades del estado sd Veracruz. CIMASS, UNAM, México. No. 1, Enero 15. (Impresién de computa ——————— & A. Buranpa C. 1971. Bibliografia para la Flora de Veracruz, CIMASS, NAM, México. No. 1, Septiembre 14. (Impresién de computadora LANGMAN, L K. 1964. A Selected Guide to the Literature on the F lowering Plants of Mexico. Univ. Pennsylvania Press, agg ne 1015 pp. LAWRENCE, G. H. M., A.F. G. BUCHEM™M, G. S. Dantes & H. DoLEezA. 1968. Botanico- Periodicom- Ticstianusi. Hunt Botanical gay Pittsburgh, Pa. 1063 pp. Morse, L. E. 1971. Specimen identification and key construction with time-sharing computers. Taxon 20:269-282 Perrinc, F. H. 1963. Data processing for the atlas of the British Flora. Taxon 112:183-190 Roc D. J. & T. T. Tanimoto. 1960. A computer program for classifying plants. ties 132:1115-1118. Scuemnvar, L. & A. Gomez-Pompa. 1969. Algunos métodos automaticos para la elaboracion de etiquetas de herbario. Bol. Soc. Bot. México 30:73-93. . ALONSO. 1967. Sistema automatico de recuperacién de Informacién para el Herbario Nacional del Instituto de Biologia de la UNAM. Anales Inst. Biol. Univ. Nac. México, Ser. Bot. 38:203— SOPER, = J H. 1964. Mapping the distribution of plants by machine. Canad. J. Bot. 087-1100. Prtes M. 1969. Consideraciones rome del estado de Veracruz. Tesis Fac. - Ciencias, Dept. Biologia, UNAM, México. 4 Sgumes, D. F. 1966. Data processing and museum collections. A problem for the present. Curator 9:216—227. Von Reis, S. 1962. Herbaria: sources of medicinal folklore. Econ. Bot. 16:273—287. PR aialitp a or Sew A SYSTEMATIC REVIEW OF THE SUBTRIBE MELAMPODIINAE (COMPOSITAE, HELIANTHEAE)! Top F. Sturessy? In the course of recent revisionary studies on selected genera of the subtribe Melampodiinae (Stuessy 1969, 1970a, 197la, 1972, in press ), the difficulties encountered in determining generic relationships were realized to be, at least in part, symptomatic of the changing generic composition of the subtribe as well as of its morphological and chromosomal hetero- geneity. As a consequence of this realization, further studies have been initiated on the generic relationships within the subtribe. In the present paper the systematics of the Melampodiinae are reviewed from three aspects: (1) a sketch of the taxonomic history of the subtribe, including changes in generic composition that have occurred since the most recent survey of Hoffmann (1890); (2) a suggestion of new changes in the generic composition of the subtribe; and (3) an indication of generic affinities within the remaining assemblage. TAXONOMIC History Lessing (1830) named and described the subtribe Melampodiinae (as subtribe “Melampodieae” ) in his report of the Compositae in the collec- tion of Schiede and Deppe from Mexico. Due to the restricted scope of the study, the subtribe included only two genera, Polymnia.and Melam- podium, and it was described as follows (p. 149): “Rhachis [=receptacle] bracteolata. Antherae ecaudatae. Capitula aut dioica aut radiata, disco masculo.” Two years later in the Synopsis Generum Compositarum (1832), Lessing expanded and clarifted his initial description of the Melampodiinae (p. 213): “Capitula aut dioica vel subdioica aut radiata, disco masculo et radio uniseriali, foemineo. Rhachis saepissime bracteolata, aut ubi ebrac- teolata, ibi achaenium obcompressum et calvum. Pappus nullus aut obsoletus et bicornis.” In addition, the subtribe was subdivided into four described and named units ( without assigned rank) with included genera* as shown below ( pp. 213-217): 1. Silphieae: “Achaeniis ecorticatis aut bialatis, et apice bicornibus, aut compressis et calvis; capitulis neque dioicis, neque subdioicis; rhachide bracteolata.” Hidalgoa Llav., Polymnia L., Silphium L. 2. Millerieae: “Achaeniis ecorticatis, exalatis, calvis, lenticularibus com- pressis, vel pl. obcompressis, aut angulatis, sectione transversa figuram regularem sistente; capitulis neque dioicis, neque subdioicis; rhachide 1Publication No. 805 from the Department of Botany, The Ohio State University, Columbus. 2Cabot Fellow in the Gray Herbarium. Present address: Department of Botany, The Ohio State University, Columbus 43210. 3In this section on taxonomic history, the genera are listed as they were originally cited by the authors. No changes have been made to more recently adopted names. 65 66 TOD F. STUESSY saepe ebracteolata.” Bailleria Aubl., Delilia Spreng., Milleria L., Oswaldia Cass., Pronacron Cass., Riencourtia Cass.., Trigonospermum Less. 3. Euxenieae: “Achaeniis ecorticatis; capitulis homogamis, dioicis; rhachide bracteolata.” Astemma Less., Euxenia Cham., Phaethusa Gaertn. 4. Melampodieae: “Achaeniis calvis, corticatis; capitulis radiatis, radio uniseriali, foemineo, disco masculo; rhachide bracteolata.” C entrospermum Kunth, Melampodium L. Several years before Lessing published his scheme of classification for the Compositae, H. Cassini published his own reorganization of the family in the many volumes of the Dictionnaire des Sciences Naturelles edited by Cuvier (1816-1830). These numerous and detailed observations were then summarized in volume three of his Opuscules Phytologiques (1834). Although in this latter work Cassini did not recognize a subtribe Melam- podiinae by name, in his tribe Hélianthées, Section Millériés, his un- ranked “Millériées vraies” characterized by “Disque masculiflore” corre- sponded in large measure to the subtribe Melampodieae of Lessing (exclud- ing the dioecious Euxenieae ). Within the Millériées vraies were contained two units. The first, “Millériées vraies, réguliéres,” with the genera Alcina Cav., Centrospermum Kunth, Melampodium L., Polymnia L., Polymnia- strum Lam., and Zarabellia Cass., corresponded roughly to the subdivi- sions Silphieae and Melampodieae of Lessing. The second unit, “Millériées vraies, irréguliéres,” containing the genera Elvira Cass., Meratia Cass., Milleria L., Pronacron Cass., Riencourtia Cass., and Unxia L£., was approximately equivalent with the Millerieae of Lessing (Table 1). Candolle in the Prodromus (1836) modified the circumscription of the subtribe Melampodiinae established by Lessing by broadening the limits to include the genus Ambrosia L. and its relatives. Seven “Divisions” of the “Melampodineae” were recognized. The first four corresponded reasonably well to those subdivisions of Lessing; the last three were new additions to the subtribe (Table 1). Candolle’s classification with included genera is as follows: 1. Euxenieae: Astemma Less., Euxenia Cham., Petrobium R. Br. 2. Millerieae: Aiolotheca DC., Apalus DC., Chrysogonum L., Clibadium L., Elvira Cass., Fougerouxia Cass., Ichthyothere Mart. in Buchn., Latreil- lea DC., Milleria L., Pronacron Cass., Riencourtia Cass., Scolospermum Less., Trigonospermum Less., Unxia L.f., Xenismia DC. 3. Silphieae: Berlandiera DC., Espeletia Mutis ex H. & B., Guardiola Cerv. ex H. & B., Hidalgoa Llav., Polymnia L., Silphium L. 4. Melampodieae: Acanthospermum Schrank, Melampodium L. 5. Ambrosieae: Ambrosia L., Franseria Cav., Xanthium L. 6. Iveae: Euphrosyne DC., Iva L., Pinillosia Ossa in DC., Tetranthus S . w. 7. Parthenieae: Coniothele DC., Leptosyne DC., Mendezia DC., Par- thenium L., Tragoceras H.B.K. TABLE 1, HISTORY OF CLASSIFICATION OF THE SUBTRIBE MELAMPODIINAE Bentham and Lessing (1832) Cassini (1834) Candolle (1836) Hooker (1873) Hoffmann (1890) Tribe Tribe Tribe Tribe Asteroideae Hélianthées Senecionideae Heliantheae Heliantheae Subtribe Section Subtribe Melampodieae Millériées Melampodineae : . Division® } mrmampodione 5, Melampodieae Millériées vraies. » Subtribe 5 Subtri réguliéres Melampodieae cmcadines : ~ Division Silphieae o> sickle Millériées vraies, irréguliéres Division , Subtri ; Subtribe Millerieae —— >» Millerieae Millerieae Millerinae F Division Subtribe Subtribe > 2 _—_————— > ; a a A Euxenieae Euxenieae Petrobieae Petrobinae Division (To various Parthenieae ’ subtribes < bmiaer to Melampodieae ) Division Ambrosieae Subtribe . Sub Ambrosieae - ” Ambrosinae Division Iveae *Arrows indicate continuity of concepts of subdivisions from one author to another. AVNITGOdWVTAW AaMLans 68 TOD F. STUESSY Bentham and Hooker in the Genera Plantarum (1873) established the generic composition of the subtribe Melampodiinae (as well as that of the entire family) that is still in use today with but minor variations. Their subtribe “Melampodieae” was basically the Silphieae and Melampodieae of Lessing and of Candolle, and the Millériées vraies, réguli¢res of Cassini with several other genera added. The included genera are as follows: Acanthospermum Schrank, Aiolotheca DC., Baltimora L., Berlandiera DC., Chrysogonum L., Dicranocarpus A. Gray, Engelmannia T. & G., Espeletia Mutis ex H. & B., Guardiola Cerv. ex H. & B., Ichthyothere Mart. in Buchn., Lecocarpus Dene., Lindheimera A. Gray & Engelm., Melampodium L., Parthenice A. Gray, Parthenium L., Philoglossa DC., Polymnia L., Schizop- tera Turcz., Silphium L., Trigonospermum Less. This revised group of genera was characterized by heads with sterile disc florets, large numbers of florets per head, and paleaceous receptacles. As for the other Divisions of Candolle’s Melampodineae, the Millerieae and Euxenieae were made separate subtribes as the Millerieae and Petrobieae respectively, and the Ambrosieae and Iveae for the most part become the subtribe Ambrosieae (Table 1). Hoffmann (1890) followed the classification of the Compositae of Bentham and Hooker almost exactly. The subtribe Melampodiinae was characterized as before with two additional new genera: Dugesia A. Gray, and Moonia Arn. Since Hoffmann (1890) and Dalla Torre and Harms (1907) listed 22 genera, several generic shifts have been suggested within the Melam- podiinae by various authors. Small and Carter (1913) revived the genus Polymniastrum Lam. as a close relative of Polymnia. Blake (1917a) switched Clibadium to the Melampodiinae from its previous inclusion in the subtribe Milleriinae. Several years later Blake (1923) added a new genus, Rensonia, followed by Humbert’s addition of Tisserantia (1927), Mackenzie's addition of Smallanthus (in Small, 1933), and Sherff’s addi- tion of Oparanthus (1937), which brought the total number of genera to 28. Cronquist (1955) cancelled Humbert’s addition by indicating that Tisserantia should be included within Sphaeranthus L. of the tribe Inuleae. Based on morphological and detailed anatomical data, Carlquist (1957) transferred Oparanthus from the Melampodiinae to the subtribe Petro- biinae. Blake (in Sandwith, 1956) indicated that Philoglossa probably should be moved to the subtribe Liabinae of the Senecioneae. Still later, Polymniastrum and Smallanthus were resubmerged into Polymnia by Wells (1965). Rzedowski (1968) reported that Aiolotheca DC. was con- generic with the earlier described Zaluzania Pers. in the Helianthinae (=Verbesininae; Solbrig, 1963), and recently Unxia L.f. was resurrected from out of Melampodium (Stuessy, 1969). Presently, then, there are 23 genera in the Melampodiinae, and these are listed in Table 2 with avail- able studies that are at least partially comprehensive. SUBTRIBE MELAMPODIINAE 69 SUGGESTED NEW CHANGES IN GENERIC COMPOSITION The most recent criteria of Bentham and Hooker (1873) and Hoffmann (1890) for delimiting the subtribe Melampodiinae emphasize the char- acters of many and sterile disc florets and a paleaceous receptacle. These features contrast with those of the neighboring and similar subtribe, the Milleriinae, which also has sterile but fewer disc florets and naked recep- tacles. Based on an understanding of the variation in these features that is known to occur in the Compositae (Bentham, 1873; Shull, 1902; Nakano, 1910; Small, 1919; Gleason, 1919), the reliance on these few characters alone for subtribal differentiation might be viewed with suspicion. In particular, grouping genera by the sterility of the disc florets might seem artificial, because it is likely that such a feature has evolved independently in at least several instances in the Compositae as evidenced by its occur- rence in several different tribes of the family (Bentham, 1873). It is interesting to note that Linnaeus in his sexual system of classification (1735) recognized in the “Class Syngenesia” an “Order Polygamia Necessaria” that was characterized by fertile female ray florets and hermaphroditic sterile disc florets. This same grouping used by Linnaeus in the Species Plantarum (1753) contained 13 genera including Chrysogonum, Melam- podium, Polymnia, and Silphium, now of the Melampodiinae. The artificiality of the Melampodiinae, however, has not gone unnoticed, and several workers already have commented on its heterogeneous nature TABLE 2. LIST OF REVISIONARY STUDIES OF GENERA CURRENTLY INCLUDED N THE SUBTRIBE MELAMPODIINAE Genus Revisionary Studies anthospermum Schrank Blake (1921); euce (1970a) alteas L Stuessy (in pre: Berlandiera DC Turner and reall (1956); Pinkava (1967) hrysogonum L Clibadium L Schulz (1912); Blake (1917a) Espeletia Mutis ex H. & B. Guardiola Cerv. ex H. & B. ppp a re Mart. in Buchn. Melam Moonia Arn. Parthenice A. Gray Parthenium L. Polymnia L. Rensonia S. F. Blake Schizoptera Turcz. Silphium L. Trigonospermum Less. Unxia L.f us Dene. pyrene A. Gray & Engelm. podium L. urner and Johnston (1956) Sraniey ape é heron and Koch (1935); ecasas (1949, 1954) uatr ern (18 Cronquist (1971); Eliasson (1971) Turner and Johnston (1956) Robinson (1901); Stuessy (1972) Rollins ag Mears (1970) Wells (1965) Blake (1923) — (1916, 1917b) i] 937); Fisher (1959, 1966); rid and Fisher (1970); Sweeney (1970) McVaugh and Laskowski (1972) Stuessy (1969, 1971a) 70 TOD F. STUESSY (Gray, 1855; Turner and Johnston, 1956; Turner and King, 1962; Skvarla and Turner, 1966). Despite these comments, little has been done to modify the composition of the subtribe in order to increase its homo- geneity. One difficulty encountered is the necessity of having a familiarity not only with genera of the Melampodiinae, but also with those in the other subtribes of the Heliantheae. Although a survey of all genera of the tribe is beyond the scope of the present study (such a survey currently is in progress; Stuessy, in prep. ), the affinities of several of the genera of the Melampodiinae to those of other more clearly defined subtribes have become apparent. In particular, Dicranocarpus, Guardiola, and Moonia appear to belong more properly in the subtribe Coreopsidinae, and Parthenice and Parthenium seem placed better within the subtribe Ambrosiinae. Of all the subtribes of the Heliantheae, the Coreopsidinae as recognized by Hoffmann (1890) is one of the most cohesive. Although some genera probably are misplaced (e.g., Calyptocarpus Less., Synedrella Gaertn. ), the majority of the approximately 18 genera share the following character- istics: brown-orange longitudinal traces in the phyllaries and corollas and often also in the stamens and styles of disc and ray florets;* scarious- margined phyllaries; deeply divided or dissected leaves; pappus of two long awns (absent in Dahlia Cav. and Hidalgoa Llav.); and chromosome numbers on an apparent base of x=12 (Dahlia and Hidalgoa are based on x=16) (Solbrig, Kyhos, Powell, and Raven, 1972). Dicranocarpus with one species and Guardiola with about ten species are very similar to each other and also distinct within the Melampodiinae in possessing narrowly campanulate involucres. Both these genera have all the features of the Coreopsidinae mentioned above ( except Guardiola is without a pappus and has only dentate leaves) including compatible chromosome numbers; Dicranocarpus has been counted as n=10 (Turner and Johnston, 1961), and Guardiola is known with n=12 (Solbrig et al., 1972). Even A. Gray, in describing Dicranocarpus as new (1855), pointed out the relationship to Heterosperma Cav. of the Coreopsidinae. The former genus, in fact, is so similar to Wootonia Greene of the same subtribe (earlier mentioned by Turner and King, 1962), that any serious study of either genus should include examination of the other taxon as a possible congener. Moonia is a small, apparently monotypic, genus from India and Ceylon. Bentham and Hooker (1873) treated Moonia as a synonym of Chrysogo- num, but Hoffmann (1890) reinstated it as a distinct genus. The similarity of Moonia to Chrysogonum or any other genus of the Melampodiinae is not obvious, and in fact, among the genera of the Heliantheae, it is most similar to Dahlia and Hidalgoa of the Coreopsidinae. The details of this ‘These traces probably represent concentrations of anthochlor pigments (=chalcones and aurones) which in the Compositae are known to be distributed mainly in the Coreopsidinae (Gertz, 1938; Cronquist, 1955; Shimokoriyama, 1962). SUBTRIBE MELAMPODIINAE 71 affinity will be reported elsewhere (Stuessy, in prep.). It is sufficient to note here that Moonia possesses all of the characters of the Coreopsidinae mentioned above with the exception that chromosome numbers for the genus are not yet known, and no pappus is present (a feature also shared by Dahlia and Hidalgoa). The Ambrosiinae as traditionally recognized (Hoffmann, 1890) is prob- ably the most distinct and unified subtribe of the Heliantheae. This unity is so striking that several workers have preferred to recognize the group as a distinct tribe (Payne, 1964; Payne, Raven, and Kyhos, 1964; Skvarla and Turner, 1966). Several unique features are shared by all genera of the Ambrosiinae (in part from Payne, Raven, and Kyhos, 1964): alternate leaves; nodding flowering heads; sterile hermaphroditic florets with rudi- mentary, unbranched styles; anemophilous mode of pollination; free, short and translucent anthers with hooked appendages; bladder-like air cham- bers of the pollen walls; and chromosome numbers on a base of x=18. Parthenium with 13 species (Rollins, 1950) and the monotypic Parth- enice have been recognized by many workers as having affinities with the Ambrosiinae. Parthenice, in fact, was placed by A. Gray at the time of original description (1853) between Parthenium and Euphrosyne (of the Ambrosiinae) with the comment (p. 85) that Parthenice mollis has “_.. much the aspect of Euphrosyne xanthifolia. . . .” Even Bentham in the discussion of his new classification of the Conspositne (1873; p. 435) wrote that the Ambrosiinae “. . . are, without a doubt, connected with Artemisia as well as with Melampodineae, having much of the habit of the former and passing into the latter through Parthenice; but geograph- ically, as well as structurally, the relationship to Melampodineae appears to me to be the closest.” Since that time, several workers have continued the commentary on the relationships of Parthenium and Parthenice with the Ambrosiinae (e.g., Rollins, 1950; Alston and Turner, 1963; Skvarla and Larson, 1965; Skvarla and Turner, 1966), but no one has suggested yet that both genera should be moved from their present position in the Melampodiinae. In my opinion, little question exists that Parthenium embraces Parthenice as its closest relative. Both genera have little in common with other genera of the Melampodiinae, and both possess characteristics of the Ambrosiinae particularly in features of the leaves, anthers, and styles. In addition, the chromosome numbers of Parthenium (Parthenice has not yet been counted) coincide with those of genera in the Ambrosiinae. Three of the four sections of Parthenium are known chromosomally to have base numbers of x=18 [section Argyrochaeta (Cav.) DC. seems based on x=17]* (Rollins, 1950; Mears, 1970), and therefore the base number for the whole genus seems likely to be x=18. Recent anatomical studies of seedling anatomy (Anderson, 1970, pers. 5Three ee ce section have been counted chromosomally, and two of them, P. confertu ray and P. hysterophorus L., are clearly based on x=17 (n=36 and 34, gu n=17 ona. bipinnatifidum (Ortega) Rollins, however, has been reported from five populations as n=12 (Rollins, 1950). fe TOD F. STUESSY comm.) also support the relationship of Parthenium to the Ambrosiinae. Of the 13 genera® of the subtribe Melampodiinae that have been sampled, Parthenium is the only genus that has two traces associated with the midveins of the cotyledons—the same situation found in Xanthium L. of the Ambrosiinae. Electron microscopic studies of pollen grains also reveal the similarity of large cavea and internal foramina in the ektexine of Parthenium and genera of the Ambrosiinae (Skvarla and Larson, 1965; Payne and Skvarla, 1970), but Parthenice in palynological features is somewhat different from both these taxa. Recent chemical studies, al- though of a limited nature, also help substantiate the relationship of Parthenium to the Ambrosiinae. The sesquiterpenoid compound hymenin, previously known only from Hymenoclea T. & G. in the Ambrosiinae, now has been discovered in Parthenium (Rodriguez, Yoshioka, and Mabry, 1971). In addition, three other sesquiterpenoids (ambrosin, coronopilin, and parthenin ) have been found in Parthenium as well as in genera of the Ambrosiinae (Mitchell and Dupuis, 1971): Ambrosia (all three com- pounds ); Hymenoclea (ambrosin); and Iva (coronopilin and parthenin ). The most conflicting datum in moving Parthenium and Parthenice to the Ambrosiinae is the presence of an “achene-complex” (Rollins, 1950) in both genera as well as in the melampodioid genera, Berlandiera, Chrysogonum, Engelmannia, and Lindheimera. This unusual morpho- logical feature of the capitulum is a basally fused complex of one phyllary, one fertile female ray floret, two sterile hermaphrodite disc florets, and two to four receptacular paleae.? Such a structure is indeed unusual enough to suggest strongly that Parthenium and Parthenice must be related in some way to these other genera of the Melampodiinae, especially because Parthenium does show some overall morphological resemblance to Engelmannia. Because of the relationship of Parthenium and Parthenice to both the Melampodiinae and Ambrosiinae, and because of the morphologically specialized and polyploid condition of the latter subtribe, one might speculate that the Ambrosiinae s. str. was derived from a Parthenium-like ancestor which in turn came from an Engelmannia-like progenitor (Fig. 1). The ancestor of the entire generic assemblage may have been based on x-9, with one main line then leading directly to Engelmannia, another line leading by aneuploid reduction (x=8) to other genera of the Melam- podiinae, and a polyploid third line (x=18) to the Ambrosiinae (s. l.). These hypothetical relationships are compatible with the morphological and cytological data obtained by Payne, Raven, and Kyhos (1964) and with the ultrastructural features of pollen grains reported by Skvarla and Larson (1965). *Acanthospermum, Baltimora, Berlandiera, Chrysogonum, Clibadium, Engelmannia, Espeletia, Guardi- ola, Lindheimera, Melampodium, Parthenium, Polymnia, and Silphium. 7In the achene-complex of Parthenice the outer phyllaries are closely associated but not fused basally with the ray achenes. SUBTRIBE MELAMPODIINAE 73 ae ae Chrysogonum Le < sit Lindheimera sol Berlandiera io i= Engelmannia ee ae ee eee ere ena opcre nos ] | Ve Parthenice | | a X=18__ parthenium | | Pre es x=18 Ambrosiinae_ | | lat S. Str. | ae ee a ee re ee a Fic. 1. Hypothetical evolutionary relationships among Chrysogonum, Lindheimera, Berlandiera, and Engelmannia of the Melampodiinae, and Parthenice, Parthenium, and the Ambrosiinae s. str. The clarification of the evolutionary affinities of these taxa, although providing a better understanding of the biological relationships, poses a problem in classification. One might believe that because the entire generic complex shown in Fig. 1 is monophyletic, it should be treated as a separate subtribe of the Heliantheae. However, this alternative seems undesirable because the genera of the Ambrosiinae s. str. have many distinct features that are not found in the other four genera of the Melampodiinae. To leave the situation unchanged also seems unsatis- factory in view of the very strong similarity of Parthenium and Parthenice to the Ambrosiinae. In my opinion, the best alternative is to move Parthenium and Parthenice into the Ambrosiinae s. |.8 The remaining genera seem best left for the present as a natural group of the Melam- podiinae that may or may not have close relatives in other subtribes. If with further investigations no additional relatives are discovered, then the recognition of these genera as a small subtribe of the Heliantheae may be in order. Obviously, the solution to this problem must await the comple- tion of studies on the subtribal limits of the entire tribe (Stuessy, in prep. ). GENERIC AFFINITIES The removal of Dicranocarpus, Guardiola, Moonia, Parthenice, and Parthenium from the Melampodiinae still leaves a very heterogeneous assemblage of 18 genera. The diversity of morphological and chromosomal features that exist among these taxa is so great that it emphasizes again the ‘Although it is not intended here to evaluate in detail the desirability of tribal status for the - brosiinae (either as s. str. or s. I.), the evolutionary relationships of the Ambrosiinae to Engelmannia, etc., suggest that such a disposition is unwarranted (for agreement see Bentham, 1873; Small, 1919; Cronquist, 1955). TOD F. STUESSY 74 8 § ‘ueduy y Ain ‘y vioweypur'y 6 6 ‘S) YL, PruueUTfesUq 9T 9I ry wmnuososAryD 6 SI ST Od visrpurpiog 9 L 8 ‘L "T wnrydyrg - _ ‘ZOINT, B19\dozIYyog = - OAPI ‘A'S vuosuay 6 ‘8 ‘L ST SI Avi *y visasnq ra €8 “OE “L3 ‘IF SS ‘ES OL ‘08 ‘8T ‘BT ‘IT ‘OI ‘6 "] umnrpodurepayy II IT ‘guoq sndivs0oeT IT ‘OT II Il‘Ol yueriyos umutadsompjuroy Pp cI cI ‘ssarT wmnuttedsouosi. 7, ST ST ST “"] Blownleg € 9T 9T FT Brean, IFEE ‘OE ‘63 9I ‘ST ‘0G "89 ‘LT ‘OT ‘ST rT eruurAjog 91 61 61 "a 2°H %9 san ensjedsy G ae gg eo ‘uyong ur ‘yey ar0y,0AIQYOT 8 8 ¥E ‘OT ‘rT wanipeqyoD T dnoiZ opraues jo (x) snuos Jo (x) (u) staquinu autos snuey dnoiy jaquinu eseq eqeqoig Jaquinu aseq s[qeqoig -owlo1yo UMOUY AVNIIGOdNVTAW AGIULANS FHL AO VUANAD JO SANOUD AO SHAAIWAN AINOSOWOUHOD ‘§ ATIVE pe a o 53 Qs 5 Ow 8s ° o-5 5 22.48 wo 8 2 ow 355 5 Sia? 3 5 Bs subtribes for suitable generic relatives. In view of t ution that can be offered at this time to point out groupings of what appear to be closely related the subtribe. These units eventually may prove to have s polyphyletic nature of the group and the necessi present study, the best sol other subtribes. Six such informal groups have with generic clusters in gether numbers th available chromosomal data in Table 3. The chromosome wi SUBTRIBE MELAMPODIINAE 75 have been obtained from the following indexes and references: Darling- ton and Wylie (1955); Cave (1958-65); Ornduff (1967-69); Coleman (1968, 1970); Fedorov (1969); Powell and King (1969); Powell and Cuatrecasas (1970); Moore (1970-71); Stuessy (1970b, 1971b); Solbrig et al. (1972). The two genera of group one, Clibadium and Ichthyothere, are closely related South American taxa that have very short, tubular ray corollas, and small, narrow anthers, uniform in color with acute appendages. As Blake emphasized earlier (1917a, p. 2), the two genera are so similar that . it seems beyond dispute that Clibadium should be removed from the Millerinae and referred to the Melampodinae next to Ichthyo- there,....” Although chromosome numbers for both genera are available only from several out of a large number of species (ca. 30), the counts at hand of n=16, 24, and ca. 33 suggest a common base of x=8.° Because only one species of Ichthyothere has been counted, it is impossible to know whether x=32 is the base for the genus (as suggested in Table 3) or, if upon accumulation of additional reports, x=16 or x=8 might prove to be basic. As for close generic relatives outside the Melampodiinae, Des- manthodium Benth. of the Milleriinae seems very similar, but affinities of these three genera with other taxa are uncertain. Chromosome counts for Desmanthodium have not yet been obtained. Group two includes three genera of close affinity: Espeletia, Polymnia, and Unxia. Espeletia is a common member of the Andes of Colombia and Venezuela, Unxia is restricted to the northern portion of South America, and Polymnia ranges from South America to the eastern United States. All three genera have black, glabrous to subglabrous, subglobose fruits with no pappus. In addition, the ray achenes are partially enclosed by (but not fused with) the inner row of phyllaries. Unxia for many years was considered a synonym of Melampodium (Bentham and Hooker, 1873; Hoffmann, 1890), but recently the former genus has been reestablished as a close relative of Polymnia (Stuessy, 1969). The available chromosome numbers for these genera point clearly to a base of x=16 for both Unxia and Polymnia. The n=19 counts obtained from about 30 species of Espe- letia (Powell and King, 1969; Powell and Cuatrecasas, 1970) may have originated by aneuploid gain from an x=16 base for the entire complex of three genera. Many additional counts obviously are needed, especially because Espeletia is a large genus with perhaps as many as 74 species ( Cuatrecasas, 1954). The two genera of group three, Baltimora and Trigonospermum, seem *Coleman (1968) was the first to suggest this base number for the genus founded mainly on his new count of n=24. The camera lucida drawing of the meiotic configuration of the species counted, C. armanii Sch. Bip., shows 16 large and eight very small chromosomes, which gives strong re a to the x=8 eames The eight chromosomes are so small, however, that Co — mentio: (p. 232) the “.. ssibility that the 8 smaller chromosomes are B chromosomes * Tf th chromosomes do SE it prove to be supernumeraries, which are often vari — Ske eae & a taxon, then this reported count would be less substantial support for an x=8 b; 76 TOD F. STUESSY to be closely related in their annual habit, triquetrous fruits, and chromo- some numbers of n=15 (Stuessy, in press). However, Baltimora also shows some relationship to Wedelia Jacq. of the Helianthinae (=Verbesininae ), and Trigonospermum is similar to Sigesbeckia L. of the same subtribe (corroborated by McVaugh and Laskowski, 1972). It would not be sur- prising, therefore, that after a more careful appraisal of generic relation- ships in the entire tribe, these two genera justifiably might be separated. Group four includes three genera that have inner involucral bracts each enclosing and fused with individual ray achenes. Melampodium is the largest (37 species currently recognized; Stuessy, 1972), most morpholog- ically diverse of the generic trio, and with a distribution centering in Mexico and Central America. Acanthospermum contains six species (Stuessy, 1970a) and ranges mainly throughout Latin America with scattered introductions to parts of the Old World (Blake, 1921). Leco- carpus is the smallest and most restricted genus of the group with three species that are endemic to the Galapagos Islands (Cronquist, 1971; Eliasson, 1971). Chromosomal data supplement the morphological unity of these genera. Acanthospermum and Lecocarpus are clearly on a base of x=11, but Melampodium, for reasons explained elsewhere (Stuessy, 1971b), seems more likely to be based on x=10. The ancestral base for the entire complex, therefore, could have been either x=10 or x=11, although the former number has been suggested previously (Stuessy, 1971b). The relationship of these taxa to other genera of the Heliantheae is obscure at present. Other taxa such as Rumfordia L., Sigesbeckia, and genera of the subtribe Madiinae also have phyllaries enclosing (but not fused with) ray florets. Whether these similarities reflect parallelisms or homologies re- mains to be determined, but because of the otherwise divergent morphol- ogies of these taxa, it seems likely that the former condition has prevailed. The genera of group five, Dugesia, Rensonia, Schizoptera, and Silphium, form an interesting but probably heterogeneous group of taxa that have dorsally flattened and laterally winged achenes. Within the Melampodiinae these genera do comprise a comfortable morphological unit, but upon examining genera in other subtribes such as the Helianthinae (=Verbesi- ninae), the very close relationship of some of these taxa makes the cohesiveness of group five less striking. For example, the monotypic Rensonia is very similar to Wedelia. This similarity is particularly inter- esting because one species of the latter genus, W. parviceps S. F. Blake, has sterile disc florets and is found in the same geographic region (i.e., Central America) as Rensonia salvadorica S. F. Blake. As a result of a lack of clear understanding of these relationships at this time, the proposed base numbers for the group (x=7, 8, or 9) should be viewed with consid- erable scepticism. Berlandiera, Chrysogonum, Engelmannia and Lindheimera comprise generic group number six. All four genera are distributed mainly in the SUBTRIBE MELAMPODIINAE 77 southern United States and adjacent Mexico,'® and all have conspicuous, elliptic ligules, dissected or at least lobed leaves, and “achene-complexes” (=a basally fused ray achene, inner phyllary, and 2-4 paleae with included sterile disc florets). Based on this latter feature, Parthenium and Parth- enice might be included here, but for reasons explained earlier in this paper, these genera seem best placed in the Ambrosiinae. The chromo- some numbers for all four genera of this group are diverse with n=8, 9, 15, and 16. A base of x=9 is suggested for the entire generic complex with Engelmannia (n=9) retaining the ancestral number (Fig. 1). The other three genera are believed to have evolved from an aneuploid base of x=8. Lindheimera may have developed without change in chromosome number from the immediate ancestor, but Chrysogonum seems an obvious poly- ploid with n=16 and Berlandiera seems interpreted best as an aneuploid at the tetraploid level with n=15. CONCLUSION The observations presented here reveal again that the subtribe Melam- podiinae is morphologically heterogeneous and clearly polyphyletic (already mentioned by several writers, e.g., Gray, 1855 [“an incongruous group,” p. 321]; Turner and Johnston, 1956; Turner and King, 1962; Skvarla and Tumer, 1966). The characters of large numbers of disc florets and the presence of disc paleae have been emphasized to the exclusion of many other features. Even though most of the genera of the Melampo- diinae are now reasonably well known, the needed critical realignment of the taxa cannot be accomplished without first obtaining a better under- standing of the limits and affinities of genera in the other subtribes, particularly in the Helianthinae (=Verbesininae ). This latter assemblage of 58 genera (as recognized by Hoffmann, 1890) is very much in need of treatment. The subtribe Milleriinae also needs to be understood better from a generic standpoint (Stuessy, 1971a), for taxa such as Desmanthod- ium show obvious affinities with Clibadium and Ichthyothere in the Melampodiinae. ACKNOWLEDGMENTS r was written; Drs. R. H. Bohning and J. A. Schmitt, Dean of the College of Biological Sciences and Chairman of the Department of Botany, respectively, of the ” teachi i manuscript. 10Several species of Chrysogonum have been described from Madagascar and others from Ceylon and Australia. The proper disposition of these taxa is not clear, but they seem only distantly related to the type species, Chrysogonum virginianum L. of the southeastern United States. 78 TOD F. STUESSY LITERATURE CITED Aston, R. E., and B. L. i ae ete Biochemical Systematics. Prentice-Hall, Inc., En ngle wood Clifs, N. Jct ANDERSON, L, C. 1 sere norphlogy in Melampodiinae (Compositae) and its implications. Amer. J. Bo BENTHAM . Notes on et casietion, ees and geographical distribution of Compositae. J Linn. Soc. Lond. Bot. 13:335-577. ——--———— and J. D. Hooker. 1873. speech pp. 190-192 & 346-352. In, Gen era Plantiram, vol. 2. Reeve & Co., London BLAKE, S. F. 1916. Species novae vel minus cognitae. Hook. Icon. Pl. 31:t. 3058 (p. 2 —---—-—— ee otes on the systematic postion of Clibadium, with descriptions of some new ner Contrib. Gray Herb. no -_—— ew and noteworthy Compositae, chiefly Mexican. Contrib. Gray Hac. no. 52: 16-59. (cf. 4) —_—————. 1921. Revision of the genus Acanthospermum. Contrib. U.S. Nat. Herb. 20: 383-392. ———————. 1923. New composites from Salvador. J. Wash. Acad. Sci. 13:143-146. CaNDOLLE, A. P. DE. 1836. Melampodineae, pp. i & 500-533. In, Prodromus ystematis Naturalis Regni Vegetabilis, vol. 5. P Carviouist, S. 1957. The genus Fitchia ioucsses: “Daly, Calif. Publ. Bot. 29: Cassin H. 1834. Tableau systématique des Synanthérées, Millériées vraies, p. 60. spake eae rhe ge logiques, vol. 3. Paris. Fe M. (ed.). 8-65. Wites to deed Chromosome Numbers, 1956-64 & Suppl. Uni sin Carolin Press, Chapel H Chicasean, ik 8. Chromosome acinar in some Brazilian Compositae. Rhodora 70:228— ———— 1970. Additional chromosome numbers in Brazilian Compositae. Rhodora 72: : CRONQUIST, i‘ 1955. Phylogeny and taxonomy of the Compositae. Amer. Mid]. Nat. 53:478—5 TATA TE NS ES Lecocarpus, pp. 342-344. In, I. L. Wiggins oa st M. Porter, of the er Islands. Stanford Univ. ‘Press, Stanford, Caine ncaa, J 9. Les espécies del génere Espeletia, Butl. Inst. es Hist. Nat. 37:1-1 —__—, 1954, Distribution of the genus Espeletia. Proc. VIII Intern. Bot. Congr., Sect. 4:131—132. Cuvier, G. 1816-30. Wenrenerey des Sciences Naturelles, ed. 2, vols. 1-60. Paris. Datta Torre, C. G. and H. Harms. 1907. Melampodinae, pp- 544-545. In, Genera nhyshe aget 9 Lei ipzig. DaruinctTon, C. D., and A. P. Wy.iz. ee Chromosome Atlas of Flowering Plants, d. - George Allen and Unwin, Ltd., London Extasson, U. 1971. Studies - | Galpagos ai X. The genus Lecocarpus Decaisne. tig Bot. Tidskr. bal Feporov, A. A. (ed.). 8. Khromosomnye Chisla Tsvetkovykh Rasteny (Chromo- some Numbers of F. sacha Plants ). Leningra F — T. R. 1959. Natural hybridization between Silphium laciniatum and Silphium erebinthinaceum. Brittonia 50— —--—--—— . 1966. The genus Silphium in Ohio. Ohio J. Sci. 66: 259-263. Gertz, O. 1938. Uber die oe des Anthochlors bei den Compositen. Kungl. Fysi iog. Sallskap. Lund, F :62—80. GLeEason, H. A. 1919. sot in flower-number in Vernonia missurica Raf. Amer. Nat. 53:5 Gray, A. 1853. Plan stan Mes curved Texano-Neo-Mexicanae: Part II. Smithson. Contrib. Knowledge 5(6):1-119. ——__—. 1855. Plantae Novae Thurberianae. Mem. Amer. Acad., ser. 2, 2(5): 297-328. HorrMann, O. 1890. Melampodinae, pp. 214-219. In, A. Engler and K. Prantl, Die natiirlichen Pflanzenfamilien, vol. 4(5). Leipzig. SUBTRIBE MELAMPODIINAE 79 Humbert, H. 1927. Une Composée sae Pe au point de vue phylogéni- ue, Tisserantia africana, gen. sp. ompt. Rend. Acad. Sci. Paris 185:395-397. Lessinc, C. F. 1830. ee p- 149. In, D. Schlechtendal and A. de Chamisso, Plant tarum Mexicanarum : cel. viris Schiede et Deppe collectarum. Linnaea 5:72-174, 206-236, 5 ———————. 1832. weemiedioes, pp. 213-217, In, Synopsis Generum Compositarum. Linnars, s, C. 1735. Systema Naturae. Leiden 53. Polygamia Necessaria, pp. 919-927. In, Species Plantarum, vol. 2. olm. McVavex , R., and C. W. Lasxowsk1. 1972. The genus Trigonospermum Less. (Com- positae, Helianthese). C Contrib. Univ. Mich. Herb. 9:495-506. , Systematics of Parthenium Section Bolophytum: Biochemical, Morphological, ‘fia Ecological. Ph.D. Thesis. Univ. of Texas, Austin Mitcue.t, J. C., and G. Duputs. 1971. Allergic contact dermatitis from sesquiter- penoids g, the Compositae family of plants. Br. J. Derm. 84: 139-150. say R. J. (ed.). 1970-71. cape to Plant Chromosome Numbers for 1968-69. Regnum Veiga 68: 1-115; -112 Nakano, H. 1910. Variation and odie in rays and disk florets of Aster fastigiatus. Bot. Gaz. 49:371-378. gt R. (ed.). 1967-69. Index to Plant Chromosome Numbers for 1965-67. Regnum Veget. 50: 1-128; 55: 1-126; 59: 1-129. Payne, W. W. 1964. A re-evaluation of the genus Ambrosia (Compositae). J. Arn. Arbor, 45:401-438. aaa , P. H. Raven, and D. W. eae OS a Chromosome numbers in Com- positae. IV. Ambrosieae. Amer. J. Bot. ——————-, and J. J. Sxvarva. 1970. Shak microscope study of Ambrosia pollen 89-100. 281-297. Pinxava, D. J. 1967. Biosystematic study of GD Hacthove (Compositae). Brittonia 19:285-29 Powe ti, A. M., and sae pees 1970. Chromosome numbers in Compositae: Colombian a Venezuelan species. Ann. Mo. Bot. Gard. 57:374-379. and R. M. Kine. 1969. Chromosome numbers in the Compositae: oo plaealalacs species. Amer. J. Bot. 56:11 Rosinson, B. L. 1899. Revision of the genus Guardiola. Bull. Torr. Bot. Club 26:232— es AOE. Synopsis of the genus Melampodium. Proc. Amer. Acad. Arts & Sci. 36:455-466. Ropricuez, E., H. Yosuioxa, and T. J. Masry. 1971. The sesquiterpene lactone chemistry of the genus Parthenium (Compositae ). Phytochemistry 10:1145—-1154. Roxuivs, R. C. 1950. The guayule rubber plant and its relatives. Contrib. Gray Herb. no. 172:3-72. Rzepowskt, J. 1968. Nota sobre la identidad de los i Aiolotheca DC, y Tonalan- “thus T T. S. Brandegee (Compositae ). Brittonia 20: 166-168. SanpwirH, N. Y. 1956. Contributions to the flora oe tropical America: LXI. Notes on Philoglossa. Kew Bull. 1956 Scuuiz, O. E. 1912. Beitrage zur Kenntnis der Gattung Clibadium, Engl. Bot. Jahrb. 46:613-628. SertLE, W. J., and T. R. Fisner. 1970. The varieties of Silphium integrifolium. Rhodora 72:536—543. SuerFF, E. E. 1937. Some Compositae of southeastern Polynesia (Bidens, Coreopsis, Cosmos, and Oparanthus ). B. P. Bishop Mus. Occas. — 12(19) :1— - SHmmoxorivAMA, M. 1962. Flavanones, chalcones and aurones, pp. 286-316. In, T. A Geissman (ed.), The Chemistry of F cient ptt The role Co., SHULL, G. H. 1902. A quantitative study of variation in the bracts, rays and disk 80 TOD F. STUESSY florets of Aster shortii Hook., A. Novae-Angliae L., A. puniceus L., and A. prenanthoides Muhl., from Yellow Springs, Ohio. Amer. Nat. 36: 111-152. SKVARLA, J. J., and D. A. Larson. 1965. An el ctron microscopic study of pollen morphology in the 4 aad ‘with special reference to the he decvaiincs. Grana Palynologica 6: 210-2 -—--—-——— and B. L. Turner. 1966. Systematic implications from electron micro- scopic studies of Seocadne pollen—a review. Ann. Mo. Bot. Gard. 53:220-256. sees a 1919. The Origin and Development of caine New Phytologist rint, No. 11. William Wesley & Son, Lon Sneaes, - K. 1933. Manual of the Southeastern es .K.Sm intone, Od J.T, Caren: 1013, Flora of Lancaster ae : x Small and J. J. ter, 2 SMITH, bins C., and M. F. Kocu, 1935. The genus Espeletia: a study in phylogentic taxonomy. Brittonia 1:479-530. ini 0. T; iceg Subfamilial eat aah 2 rapes: Taxon 12:229— ee D. W. Kyuos, M. Powe Lt, an Raven. 1972. Chromosome num- ber: n Compositate VIII: a heres, Ape : Bot, 59: :869-878. Seieons P. C, 1915. The genus Espeletia. Amer. J. Bot. 2:468-485. Sturssy, T. F. 1969. Re-establishment of the genus Unxia (Compositae-Heliantheae ). Brittonia 21:314-321, ——___ - 1970a. The genus Acanthospermum ( Compositae- pi lain Melam- podinae ) : taxonomic changes and generic affinities. Rhodora 7 1970b —_—————— - Chromosome studies in Melampodium sloaietligs Helian- theae). Pst 20: 365-372. ———————. 197la. The genus Greenmania: a taxonomic synonym of Unxia (Com- positae: Pedicures). Rhodora 73:291-292, ———. Chromosome numbers and phylogeny in Melampodium (Com- prota Amer. J. Bot. 58:732-736 . Revision of the genus Melampodium (Compositae: Heliantheae ). Bhiadoen 74:1-70, 161-219. press. Revision of the genus Baltimora (Compositae, Heliantheae ). - In . Fieldia ana Bot rae C. R. 1970. Monograph of the genus a pets I. Silphium compositum ow (Compositae ). Ohio J. Sci. 70:226-233 "Poneeeie 1, C. Jounston. 1956. Chromosome numbers and geographic Tks of Lindheimera, Engelmannia, and Berlandier, a (Compositae-Helian- theae-Melampodinae ). Southwest. Natur. 1:125-132 —-—-——-— and —~——————. 1961. Chromosome numbers in the Compositae-III. Certain Mexican species. Brittonia 13:64-69. ———————, and R. M. Kine. 1962. A cytotaxonomic survey of Melampodium (Com- posta Helianthene ). Amer t. J. Bot. 49:263-269, WELLS, J. R. 1965. A taxonomic study of Polymnia (Compositae). Brittonia 17:144— 159. A REVISION OF THE FERN GENUS NEPHELEA GERALD J. Gastony* The genus Nephelea comprises a major evolutionary group of tropical American tree ferns. Because it was only recently recognized as a taxon, Nephelea lacks the long and often complex taxonomic history associated with many fern genera. It was described for the first time by Tryon (1970) as a result of a complete and long-overdue generic revision of the Cyathea- ceae in which eight genera, representing natural evolutionary lines, were recognized. These eight genera are Metaxya and Lophosoria with tri- chomes but without scales on the stem and petioles, and Sphaeropteris, Alsophila, Nephelea, Trichipteris, Cyathea, and Cnemidaria with scales and with or without trichomes on the stem and petioles. Usually the large bulk of the squamate members of the Cyatheaceae had previously been treated in the three classical and unnatural genera, Cyathea, Hemitelia, and Alsophila, based solely on characters of the indusium. Recognizing the unnaturalness of these classical genera, Copeland (1947) proposed a nearly inclusive genus Cyathea (ca. 800 species) and four minor genera (Trichopteris, Cnemidaria, Gymnosphaera, and Schizocaena—totaling about 50-60 species ). These were purportedly natural lines based largely on stem development, degree of leaf dissection, color of axes, articulation of pinnae, venation, and presence or absence of an indusium. Holttum (1963) circumscribed the Cyatheaceae in an inclusive sense with a more rigorously defined genus Cyathea. He recognized Cnemidaria as the only other genus among the squamate genera of the Cyatheaceae. Before the recognition of Nephelea as a natural alliance of generic rank, most of its species had been described in the genus Cyathea, two had been placed in Hemitelia, six erroneously in Alsophila, and one in the non-squamate genus Lophosoria. The name of the genus is derived from the greek vededn (nephéle) in reference to the usual cloud forest habitat of its species. Species occur in the Greater and Lesser Antilles and from Mexico to Argentina and east- ward in South America to Venezuela and French Guiana in the north and to Paraguay and Brazil in the south (Map 1). Nephelea exhibits a broad altitudinal distribution from sea level (N. cuspidata) to 2800 m. (N. erin- aced var. purpurascens). GENERIC AFFINITIES A key to the genera of the Cyatheaceae has been provided by Tyron (1970, pp. 14-16). This key and its accompanying discussion should be consulted for general generic determinations and for a thorough treat- ment of the generic characters employed. For use in identification and as an orientation to Nephelea itself, a synoptical arrangement of the key is 1Present address: Department of Plant Sciences, Indiana University, Bloomington, Indiana 47401 81 82. GERALD J. GASTONY given below together with reference to illustrations that emphasize certain points. A. Stem and petioles with trichomes, scales absent. _. . Lophosoria, Metaxya. A. Stem and petioles with scales, trichomes present or absent on the petioles ___._-B. B. Petiole scales structurally conform, the cells of the body similar in orientation, S B. Petiole scales structurally marginate, with a narrow to broad margin of ce different in orientation, size, and (usually) in shape and color from those of . 3, 4) C the central gp dyoatac el TUGNE, 8 MBC pian esis Maratea ait acs Aa es eG amon ele ‘ C. Petiole scales (or some of them) with a dark (very rarely lighter colored ) apical seta, similar setae sometimes also borne on the body of the scale or on its edge (Figs. 66, 79) scales (Figs. 7-9), the unexpanded croziers with well developed squaminate c helea. Evidence that Nephelea is most closely related to Alsophila is found not only in the unique common development of structurally marginate petiole scales with dark apical setae, as noted in the key, but also in characters of their sporangia and spores. Although a discussion of the sporangia and spores of the entire family will be presented elsewhere, the bearing of these characters on the generic affinities of Nephelea merits brief consideration here. A survey begun by the author during this study of Nephelea indicates that, whereas sporangia containing sixty-four spores characterize the other genera of the family, Nephelea and Alsophila are characterized by the presence of sixteen-spored sporangia (Fig. 12). This diminished spore output appears to result from an abbreviated sporo- genetic process whereby two of the four mitotic cell divisions typically preceding meiosis in sixty-four spored sporangia are absent. A survey of cyatheaceous spore morphology indicates that Nephelea and Alsophila are further characterized by a distinctive form of perine (Figs. 13-15). Although Nephelea cannot be separated from Alsophila on the basis of this perine morphology, both genera may be distinguished from the other genera in the family by this character, Further evidence of the phyletic affinity of Nephelea and Alsophila is found in the appendages of their petiole bases and in their common poten- tial for developing squaminate spines. Tryon has noted (1970, p. 28) that within the Cyatheaceae only Alsophila features fully developed aphlebiae at the base of the petiole. Nephelea exhibits a modified form of this char- acter in N. pubescens, N. setosa, and N. Grevilleana, species characterized by leaves with subaphlebioid basal pinnae (Fig. 61). These are structur- ally intermediate between the highly dissected or filiform aphlebiae of many species of Alsophila and the normal basal pinnae of other species of THE FERN GENUS NEPHELEA 83 Map 1. The distribution of Nephelea. Nephelea. As indicated in the key above, Nephelea is characterized by the presence of conspicuous squaminate spines on the petioles and unex- panded croziers. The tendency of Alsophila to develop indurated and even squaminate spine-like petiole scales has been recognized by Tryon (1970, pp. 27-28), particularly in certain Madagascan and Antillean species. The full development of squaminate spines in Nephelea may be considered the ultimate expression of the potential or tendency seen in Alsophila. The sporogenetic pattern, spore morphology, basal-pinnae structure, and potential for squaminate spine development, in addition to the char- acters of the petiole scales cited by Tryon, may therefore be taken to demonstrate the close phyletic relationship between Nephelea and Also- phila and to substantiate the expression of that relationship in his phyletic chart (Fig. 18, from Tryon, 1970, p. 12). The close relationship between these genera and certain morphological patterns within Nephelea raise the issue of the degree of monophylesis of Nephelea. The uncommon pinnate-pinnatifid to pinnate-pinnatisect lamina 84 GERALD J. GASTONY complexity and essentially cyathiform indusial morphology of N. pubes- cens and N. balanocarpa and the sub-aphlebioid basal pinnae of the former suggest a close affinity between these species and the pinnate- pinnatifid Antillean species of Alsophila. From an ancestral plexus similar to these two species, divergent geographic speciation may have given rise to the other species of Nephelea, which are characterized by a more dissected lamina. Such a derivation of N. setosa, however, may be particu- larly questioned. In its bipinnate-pinnatifid to tripinnate lamina, N. setosa resembles other species of Nephelea, but it is strongly characterized by sub-aphlebioid basal pinnae and is unique in the genus with a hemitelioid indusium. Its sub-aphlebioid basal pinnae may have been retained more or less unchanged from the condition of the ancestral complex while its lamina complexity increased and its indusial structure changed greatly, or these characters may reflect an independent close affinity of N. setosa to species of Alsophila with aphlebiae and hemitelioid indusia (for example A. capensis). If several independent origins of Nephelea from Alsophila did occur, Nephelea would still be considered monophyletic inasmuch as it was derived from a taxon of equal or lower rank (Simpson, 1961, p. 124). The issue cannot be discussed further with certainty until the infrageneric taxonomy of Alsophila is clarified by a revision of that genus. SPECIES AND SPECIATION I recognize only twenty-three of the fifty-nine taxa that have been formally proposed within the circumscription of Nephelea. The previous proposal of an excessive number of taxa was occasionally due to inade- quate study but probably more often was the result of fragmentary early collections, few of which consisted of more than a single pinna or pinna- portion. Due to this, early authors could not know the range of variation in a taxon and could easily recognize minor variants as species. Because Nephelea was only recently recognized as a natural assemblage, compari- sons of putative new species were often made with unrelated taxa, which served to enhance their distinctness. F inally, the biology of populations was not always appreciated and the species concepts of earlier authors, such as Hermann Christ, were strongly typological. Several early col- lectors, for example Wercklé in Costa Rica, were nurserymen with an eye for the unusual or even the aberrant, and the morphological extremes represented in some of their collections were quickly recognized as distinct new species. The biological species concept and evolutionary principles have guided the assessment of taxa in this study, and have conferred an evolutionary dimension on what might otherwise have been a simple cataloging of morphological types. The determination of taxonomic rank has been inferential, based largely on evidence from morphology, geography, and ecology. Specific rank expresses the probability that the members belong THE FERN GENUS NEPHELEA 85 to the same actually or potentially interbreeding population or group of populations. Infraspecific rank indicates a lack of that degree of morpho- logical divergence which is achieved in the species in this genus. Geographic speciation (involving the geographic separation of elements of a species) has probably been largely responsible for the taxonomic diversity within Nephelea. This conclusion is based on the geographically correlated morphological variation observed in some taxa and the apparent lack of amphiploidy in the scaly genera (Tryon, 1970). As outlying popu- lations migrate into areas at progressively greater environmental variance from that of the central population, the adaptive shift in gene frequencies may be reflected phenotypically. Given later isolation, differential selection pressure, and sufficient time, such populational divergence may proceed to speciation or to any of the taxonomically recognized infraspecific ranks. Taxa which are morphologically distinctive and which have substantial allopatric ranges have been inferred to represent taxa at the rank of species, as two species pairs Nephelea cuspidata and N. Sternbergii (Maps 19, 20-21) and N. Grevilleana and N. crassa (Maps 4, 5). The approxima- tion of other taxa to the rank of species has been inferred from their evolutionary performance in terms of morphological distinctiveness and geographic range relative to these species pairs. Although the species of Nephelea have not been extensively investigated cytologically, there is reason to believe that ploidal differences have not been a significant factor in speciation. Walker (1966) reported cytological analyses of N. Grevilleana, N. pubescens, and N. Tussacii (as species of Cyathea). They have chromosome numbers of n=69 or 2n=138, which are the numbers uniformly reported for every investigated species of the six squamate genera in the family (Tryon, 1970). MorPHOLOGY AND ANATOMY Stem. The stem in all species is erect. The maximum heights attained by the different species are unknown, but the maximum recorded height in the genus is 15 m. in Nephelea cuspidata and N. Sternbergii. Under appro- priate conditions, all species may develop a sheath of adventitious roots toward the base of the stem. Lucansky (1971, p. 53) has determined in N. polystichoides and N. erinacea var. erinacea (as N. aureonitens) that these roots arise both from the leaf bases and from the stem. The func- tional efficiency of these adventitious roots in a water-absorbing capacity is unknown, but they certainly add rigidity to taller stems and are their only means of radial growth. The greatest recorded stem diameter, includ- ing the adventitious root sheath, is 45 cm. in N. Imrayana var. basilaris from Estado Bolivar, Venezuela. The stem is dictyostelic with varying numbers of amphicribral meristeles, each surrounded by a dark scleren- chyma sheath. These, together with a hypodermal sclerenchyma sheath and the frequently developed adventitious root sheath, provide the major 86 GERALD J. GASTONY mechanical support of the stem. Medullary and cortical bundles are present (Fig. 19), but on the basis of Central American material examined in cross section, these accessory bundles appear to be less numerous in Nephelea than in the other squamate genera. A thorough discussion of the arrangement and fate of the medullary and cortical bundles is provided by Lucansky (1971, pp. 54-66). The leaves are disposed in vertical orthostichies and several vertical parastichies, or the spirals may be compressed and the internodes elon- gated so that the leaves are borne in whorls or apparent whorls. In Nephelea erinacea, N. mexicana, and N. polystichoides, I have frequently observed whorls of four leaves which give the stem a squarish cross sec- tion (Figs. 19, 20). Four to fifteen leaves per crown have been recorded in collections and their development may be sequential or apparently simultaneous in flushes. All species in which stem specimens have been seen bear persistent blackish spines on the stem (Fig. 23). Scales also invest the stem and may be best observed toward the apex, since they are eventually eroded below. The persistence and gradual erosion of dead leaf bases or their rapid and clean abscission is not consistent within a species, and I have been unable to determine environmental correlations. Adventitious buds have been observed in several species as well as lateral branching. Most specimens in this condition which I have personally observed had either been injured (see discussion below under Nephelea woodwardioides var. Hieronymi) or were growing in a disturbed habitat. In N. polystichoides, Lucansky (1971, pp. 53, 64) determined that adventitious buds arise on the latero- adaxial surface of the leaf bases and that the vascular supply to the bud is continuous with an adaxial leaf trace. I have observed positively geotropic buds on the stem of a plant of N. mexicana (Gastony & Gastony 993 [cH] ) which was growing in an apparently undisturbed habitat. Vegeta- tive reproduction by stolons has been reported by Brade (1971) in Cyathea Sampaioana. This is discussed below under N. Sternbergii var. acantho- me Leaf. Definitions of terms for the parts of the leaf are those of Tryon (1960). Petiole. The spines of the petiole provide one of the most useful generic characters, and the petiole indument is often diagnostic at the level of the species or species group. The petioles of all species are clearly character- ized by the presence of blackish spines (Figs. 6, 16, 17). The phyletic derivation of these spines from typical Nephelea scales by multiplication and induration of the cells of the scale body (Figs. 7-9) has been dis- cussed by Tryon (1970, p. 38). I am in complete agreement with his interpretation. Although the scales on the adaxial surface at the base of the petiole are often more persistent than those on the abaxial surface, their margins THE FERN GENUS NEPHELEA 87 . \ WY) i‘ — => \ =o Patsibbinen dace te stapes: Be figs —— ————> a. scales to squaminate spines, Nephelea erinacea var. erinacea, Gastony ; ; 8; 7, Petiole scale thickened basally; 8, Spine-like petiole scale; 9, Small squaminate spine. Fics 10, 11. Petiole scale apices representing the rounded filamentous condition, x 90; 10, Cyathea arvula, Proctor 5513 (Gu); 11, Trichipteris mexicana, Yuncker et al. 6015 (GH). 88 GERALD J. GASTONY are usually poorly developed. Those on the abaxial surface are more varied in structural detail and provide substantial diagnostic utility at the species level. All of these scales are structurally marginate and feature a dark seta at the apex, the maximum length of which is, to some extent, correlated with species groups. The petiole scales of most Antillean species tend to be uniformly brown with the apical setae measuring a maximum of 0.1- 0.25 mm. long, whereas the scales of continental species tend to be more delicate, with lighter colored margins and apical setae often reaching a length of 0.5-2.0 mm. The petiole scales of Antillean species and of several of the continental species lack dark setae other than at the apex ( Fig. 66), whereas some continental species often, or in some cases consistently, bear additional smaller dark apical and lateral setae (Fig. 79). Besides these scales, which are often 20 mm. Jong and 1 mm. wide or larger, and occa- sional minute trichomidia, the petiole bears minute squamules which frequently are also dark-setate and grade in size into the larger scales. In their spines and indument, there is no clear demarcation between petiole and rachis. In all species where I have seen adequate material, the petiole spines gradually diminish in size along the rachis. Lamina. The complexity of the lamina division in Nephelea ranges from predominantly pinnate-pinnatifid (as in N. pubescens) to predominantly tripinnate-pinnatifid (as in some plants of N. polystichoides). In most species, lamina complexity varies between the base and apex of the pinnae. The most basal ultimate segments of basal pinnules are often one degree of complexity greater than more distal ultimate segments or those on more distal pinnules. For this reason, the lamina complexity in the species descriptions is given both as the maximum degree commonly found in the species and as the predominant degree exclusive of the most proximal and distal portions of the primary and secondary segments. A strong dimorphism between sterile and fertile leaves does not char- acterize Nephelea. Most fertile and sterile leaves are practically mono- morphic, and where a slight dimorphism has been observed, it is discussed under the species involved. The texture of the lamina varies from papyra- ceous to strongly coriaceous, usually with a narrow range of variation. The form of the apex of the lamina, often poorly represented in her- barium collections, is generally correlated with the length of the apical seta of the petiole scale and serves to reinforce the distinction between the Antillean and continental species groups. The Antillan species gener- ally have a gradually acute lamina apex (Fig. 21), whereas in the conti- nental species the lamina is abruptly reduced to a distinct pinna-like apex (Fig. 22). Although the continental Nephelea Tryoniana is more or less intermediate between the two groups in the length of the apical setae of its petiole scales, the feature of a tapering lamina apex clearly places it with the Antillean species. The more or less abruptly reduced pinna-like lamina apex of the Lesser Antillean elements of N. Imrayana clearly associates them with the several continental species and supports the interpretation THE FERN GENUS NEPHELEA 89 of their derivation from South America. Further geographic implications are discussed under these species. In describing the indument of the non-vascular lamina surface, the extremely minute and usually two-celled trichomidia which are commonly encountered in fern leaves have been disregarded. These are present, although often obscurely so, in all material examined. Except when their color takes on a more white or claret cast ( usually randomly, but see the discussion under N. Sternbergii var. acanthomelas), they are inconspicu- ous even at magnifications of sixty diameters. Venation. The venation in all species is free, never reticulate. In most species the veins are predominantly once-forked, but are mostly simple in Nephelea concinna and in the most complex forms of N. polystichoides where they are borne on quintary axes. The venation pattern and the extent of lamina surface are correlated, although this was not recognized by some early taxonomists. In small or narrow ultimate segments, the veins are commonly simple as they also are toward the distal ends of larger segments. As the surface area of ultimate segments increases, the simple veins do not increase in number; they fork or branch instead. As ultimate segments approach a lobed condition, the venation becomes increasingly branched, or finally a midvein and simple laterals develop in large lobes. The veins of an ultimate segment are commonly branched proximally in the segment, once-forked medially, and simple distally. The primary forking is usually at or near the costule, but as the zone of simple veins is approached distally in the segment, the forking tends to occur farther from the costule. The descriptions of venation in the species treatments refer to the predominant condition more or less medially in the ultimate segments. The indument on the abaxial surface of the costa, costule, and veins is of particular diagnostic value; a diminutive form of the indument on the axis of one order of branching is often present on the axis of the next higher order. As squamules on these axes become progressively smaller, their morphology occasionally grades into that of trichomes, through trichomes which are uniseriate at the base and biseriate or multiseriate above. Such transitions suggest the phyletic derivation of the squamules from trichomes. In other instances the smallest squamules appear more or less stellate with various kinds of processes. As these approach the condi- tion of trichomes, it is diagnostically very useful to distinguish the various forms which are found to be species specific. For this reason, the stellate trichomes of Nephelea cuspidata (Fig. 80), the stellate squamules of N. erinacea (Fig. 67), and the highly trichomoid stellate squamules of N. Sternbergii var. Sternbergii ( Fig. 84) have been illustrated. Sorus and Associated Structures. Structures associated with the sorus provide very useful diagnostic characters, but they show as much variation as other aspects of the species morphology. Variability in these structures is to be expected in any outbreeding diploid population. 90 GERALD J. GASTONY Nephelea Tryoniana is the only species in the genus which is truly exindusiate. In N. setosa (Fig. 60), the indusia are hemitelioid, well- developed and attached to only one side of the receptacle. In all other species in the genus the indusia are attached completely around the base of the receptacle. In the latter case, the indusia may enclose the sori to varying degrees. Tryon (1970, p. 8) recognized such indusia as cyatheoid if they are open at the apex and as sphaeropteroid if they are closed at the apex. For purposes of greater refinement, I have recognized three subtypes of the cyatheoid indusium—meniscoid, cyathiform, and urceolate—and I have recognized a variant of sphaeropteroid as sub-sphaeropteroid. Menis- coid indusia are shaped like a concave meniscus or watch-glass; cyathi- form indusia are cup-shaped and a little wider at the top than at the bottom (Fig. 52); urceolate indusia are pitcher-like with the mouth more or less contracted; sphaeropteroid indusia completely enclose the sorus persist as the spines of the petiole in the mature leaf: 16, Young crozier beginning expansion, on g ns Gastony & Gastony 1025 (cu); 17, Expanding crozier showing spines of part of the petiole as well as of the still unexpanded crozier portion, Gastony & Gastony 1026 (cx). THE FERN GENUS NEPHELEA 91 and are closed at the apex where they often feature an umbo (Fig. 74). In sub-sphaeropteroid indusia, the apex is barely or not quite closed. These are degrees of development convenient for descriptive purposes and are often quite discrete in a given species although intergrading con- ditions can be found. The texture of the indusium may be quite firm, in which case it remains entire and persistent if meniscoid to cyathiform or else ruptures, but is usually persistent, if urceolate to sphaeropteroid. When its texture is delicate, the indusium is easily ruptured by the expansion of the sorus, as when the sporangia open at maturity, or upon drying. In proportion to their delicacy, such indusia are usually partially or almost completely fugacious, as in Nephelea mexicana, N. polystichoides, and N. Sternbergii var. acanthomelas. Careful examination reveals the presence of indusial remnants at the base of the receptacle in material of this nature, but these remnants were occasionally overlooked by some early authors who placed such specimens in the classical genus Alsophila. The indusia may be glabrous or variously pubescent, and the pubescence is usually a diminu- tive form of that found abaxially on the veins. The species of the Greater Antilles are characterized by meniscoid to cyathiform indusia of firm texture. With the exception of Nephelea incana and N. Sternbergii var. Sternbergii, the indusia of the continental species are more delicate or more enclosing than those of the Greater Antilles. In all species, the sporangia are borne on a receptacle which is situated NEPHELEA . CNEMIDARTIA CYATHEA scale body structurally marginate ALSOPHILA ; TRICHIPTERIS Ce ae Nee ie eee fee See Saree eet “Meal Satoshi Saitte moe Vie Neti OMe aie Sea scales ‘ scales not setate a setate SPHAEROPTERIS scale body structurally conform scales present eee ee wm wwe eee ee ee eee eee eee eee eee scales Lan LOPHOSORIA METAXYA Fic. 18. Phyletic chart of the Cyatheaceae (from Tryon, 1970, p. 12). 92 GERALD J. GASTONY at the primary forking of the veins or at or near the costule on simple veins. The shape of the receptacle is columnar or more or less globose on a very short stalk (Figs. 52, 60). In most species the form varies between these conditions but in some, the columnar (as in Nephelea crassa) or the globose form (as in N. cuspidata) may be constant or nearly so. For pews purposes, the receptacle is termed exserted when it extends . Stem, apex, and leaf apices of Nephelea. Fic. 19. Cross section vol a stem of N. wohl is es as in Fig. 20. Note the squarish shape of the section, the dark sclerenchyma sheath around each meristele, and the relatively few medullary and cortical bundles, Tryon ¢& Tryon 6990 (GH), X . Fre, 2 % i in i chelek ow siaake aati thei Page tioles and croziers each in an apparent whorl, Tryon ¢& yron 702 H), reduced. Fics. 21, 22. Apical portions of leaves of Nephelea, reduced: 21 Gradually trichomes present or absent, or these axes glabrescent. (Plants of species 3-5 with alate a-rachis—see headings G above. N. Costa and (or) costule abaxially with flattish scales, trichomes present or absent . Veins, or some of them, abaxially with stellate trichomes or strongly tri- chomoid stellate squamules, occasionally small dark-setate squamules as on the costule ma also be present . i ument le spicuous with less prominent arms and the indusia — to light brown, delicate and more or less a with indumen 5 aa ines petiole base abaxially wi or without more than o dark apical seta and dark lateral setae. Brazil, Pasi - N. Sternbergii. O. Veins ar lacking indument or the indument not stellt Poot ca Q. Indusia meniscoid to shallowly or deeply cyathiform or sub-sphaeroptero the more sphaeropteroid ones ruptured but persistent at m “0 often Suge es trichomes or trichomoid processes Colombia Hevador, ee a s. Q. Tnlviin sakes os al to sub-sphaeropteroid, often partially to almost complete i fu ees at maturity or when dried, not bearing dire or trichomoid processes R. ~» 98 GERALD J. GASTONY te 98); complexity of the lamina bipinnate-pinnatifid to tripinnate-pinnatifid. Costa Rica, Panama 18 1. Nephelea pubescens (Kuhn ) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 27, 28, Map 2 “92 aL Cyathea pubescens Mett. ex Kuhn, Linnaea 36: 164. 1869.rvpE: Jamaica, collector d r not stated but probably 1843, Purdie. A portion of a leaf (“comm. Sir W. Hooke 185: eS . apex and blackish-brown spines to 11 mm. long, adventitious buds Unexpanded croziers with blackish spines to 11 mm. long and brown scales with a q pinnae sessile, narrowly elliptic-la : long and 4.5 cm. wide; basal pinnae ighly reduced, to ca. 7 cm. long and sub- narrow tan scales, abaxially furfuraceous to a distally, lacking conspicuous e-segments to ca. 3.2 cm. long and 0.4 . n evident on both surfaces, adaxially usually glabrous, rarely sparsely pubescent abaxially glabrous or frequently with stiff tri i axially glabrous between the veins. Sori at the orking of the veins; indusium meniscoid to cyathiform (rarely urceolate), brown, of firm texture, persistent and entire at maturity, glabrous; receptacle columnar to globose, usually exserted, with inconspicuous, short (ca. 0.2-0.3 mm. long) paraphyses. In 1969, Rolla Tryon noted at BM that a collection in the herbarium of Rawson “Jamaica, ded. Hooker 1859” corresponds in detail with collections of Purdie, Jamaica, 1843 in the herbarium of John Smith. He reasoned that the Purdie 1843 collection was probably the one distributed by Hooker and received by Mettenius, among others. I have studied a BM sheet of Purdie 1843 from the herbarium of John Smith together with the type and agree that they appear to be parts of the same collection. These sheets in the John Smith herbarium and in the Rawson herbarium at BM may therefore be taken as isotypes. THE FERN GENUS NEPHELEA 99 Nephelea pubescens is notable because of its sub-aphlebioid pinnae to- ward the base of the lamina. Such sub-skeletonized pinnae are uncommon in Nephelea, being found elsewhere only in N. Grevilleana and N. setosa (Fig. 61). Although the degree of skeletonization is somewhat variable in N. pubescens, the pinnules of these lower pinnae are nearly always more dissected than their counterparts on other pinnae. Nephelea pubescens is endemic to the Blue Mountains of Jamaica at altitudes of 1250-2200 m., where it is common in wet, mossy forest, in ravines, and on moist, cloud forest slopes. Specimens commonly calle y this name from Cuba, Hispaniola, and Puerto Rico lack spines on the petioles and croziers and belong to species of the genus Alsophila. ADDITIONAL SPECIMENS EXAMINED: Jamaica. Eggers 3638 (r,M); Gilbert 80 (Ny); Jenman 44 (uc). Portland: Gastony 23, 120 (cu); Moore & Read 9557 (cH); Orcutt 5142 (uc); Proctor 4333 (Mo); Riba 233 (cH). Portland-St. Thomas: Maxon 9669 (GH.M,Ny,us); Maxon & Killip 1712 (¥,cH,Ny,us). St. Thomas: Hunnewell 14257 (cH); Maxon 1432, 1446 (us), 9987 (cu,Ny,us), 10138 (ny,us); Proctor 4182 (us); Underwood 2468, 2551 (nx). St. Andrew: Crosby et al. 312 (¥,Mo,Ny,uc,us); Hespenheide et al. 1467 (cH); 1932, Papenfuss (uc); Wilson 525 (A,us). St. Andrew- Portland: Chrysler 4547 (Mo); Clute 90 (mo,Ny,us); Eggers 3635 (us); Fisher 96 (Ny); Harris 7219 (Ny), 7719 (¥,Ny,us); Hatch 6 (us); Johnson 4 (F); Killip 287 (us); Maxon & Killip 661 (¥,cH,Ny,us); Underwood 1531 (Ny,us), 1533 (Ny). 2. Nephelea balanocarpa (D. C. Eaton) Tryon, Contrib. Gray Herb. 200: 38. 1970. Fics. 29-31, Map 3 a132e Cyathea balanocarpa D. C. Eaton, Mem. Amer. Acad. n. s., 8: 215. 1860.TyPE: Cuba orientali (Oriente), 1859, 1860, C. Wright 1063, yu! (two sheets); isotypes: cu!,Mo!,Nny!,us!. Stem to 7 m. tall and 14 cm. in diameter, bearing brown scales especially toward the apex and blackish spines to 5 mm. long, adventitious buds not known. Unexpanded croziers with blackish spines to 5 mm. long and brown scales with a dark apical seta 0.25-0.4 mm. long and without lateral setae. Petiole to 0.3 m. long, brown to grayish- o 0.5 mm.) long and without lateral setae. Lamina rigidly papyraceous, to ca. 2.5 m. long and 0.7 m. wide, bipinnate-pinnatilobate to bipinnate-pinnatifi (predominately pinnate-pinnatisect to bipinnate), with a gradually acute apex. Rachis brown, iall - ala long and 0.6 cm. wide, straight to sub-falcate with lateral margins entire-undulate to dentate-lobate (the teeth mostly simple) and usually auriculate especially on the basiscopic side, le £5 entire to te, margins slightly or not revolute, the more or less central innules partially adnate, basal pinnules often pinnatilobate to pinnatifid and to mm. tiolulate; costa or costule adaxially sparsely pubescent to SS abaxially stifly pubescent and with occasional flat to bullate usually dark-setate squamules to ca. 0.5-1.0 mm. long, not significantly clustered in the axil; veins once or more forked, 100 GERALD J. GASTONY near to ca. 0.5 mm. from the axis, in strongly lobate pinnule-segments and in the basal auricles the borne on a secondary axis, evident on both surfaces, adaxially glabrous, abaxially glabrous or rarely with occasional stiff trichomes on the veins of a nd . . or urceolate, brown, of more or less firm texture and mostly entire at maturity, persistent, glabrous; receptacle columnar to globose, inserted to exserted, with sparse paraphy: 3 mm. long. e pinnules and pinnule-segments of this species are usually persistent but may infrequently drop off in herbarium material. Although the fertile pinnules in Nephelea are often somewhat more contracted than their sterile counterparts, this condition is occasionally more pronounced in N. balanocarpa. Such characters are considered to be random, trivial, and are not of taxonomic importance. Nephelea balanocarpa is endemic to the mountains of Cuba, where it is known only from the provinces of Oriente and Las Villas (formerly Santa Clara). It is found at altitudes of 750-1350 m. (with one collection from about 400 m.) in sink holes and ravines and in wet or dwarf forest on mountain slopes. ADDITIONAL SPECIMENS EXAMINED: Cuba. Oriente: Clément 751 (F,us), 826, 995, 1272, 1659, 7571 (us), 1652 (vc), 1653 (cH), 1667, 1676 (m); Ekman 1639, 5326, 8132 (s,us), 3872 (us), 5163, 14645 (s); Hioram 7002, 7292, 7294, 7296, 7300, 7305, 7309 (us); Hioram & Clément 1653 (cH), 6372, 6408 (us); Leén 11174 (ny,us), 11182 (ny), 14021(us); Leén, Clément & Roca 10116, 10531 (us), 10215 (Ny,us); Maxon 4031 (GH,M,Ny,uc,us); Morton ¢ Acuna 3677 (us); Shafer 8033, 8224 (us); Taylor 508 (Ny). Las Villas (formerly Santa Clara): Bailey 12453 (us); Jack 7115, 7132, 7879 (¥,GH,NY,Us), 7248 (¥,CH,MO,Ny,us), 8053 (cH,us); Jack & Rowe 10421 (Ny); Morton 4238 (cu,uc,us). 3. Nephelea Grevilleana’ ( Mart.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 32-35, Map 4 THE FERN GENUS NEPHELEA 101 abaxially furfuraceous-glabrescent to pubescent distally, with usually Lary pat lateral pneumatodes at costa axils, often narrowly greenish-alate (to c mm wide on each side) between the most distal sessile pinnules. Pinnules to sty cm. long : ; ng to vein endings ), apex serrate, margins slightly or not revolute; costule a — usually with one to few stiff trichomes ‘istally abaxially pubescent and with usually dark- setate bullate scales; veins once-forked at or near the costule to eeteceecinin twice or mn rp 0.2-0.3 mm. long ) bullate sqiamules pace weet to globose, mostly — with inconspicuous, short (ca. 0.2-0.3 mm. long), rarely squamuloid, para- pnyses Three sheets in the Jenman herbarium at ny! bear an apparently unpub- lished varietal epithet referring to the coriaceous texture of the specimens; these specimens are within the normal range of variation of the species. A noteworthy character of this species is the presence of sub-aphlebioid pinnae toward the base of the lamina. Such sub-skeletonized pinnae are uncommon in Nephelea, being found elsewhere only in N. pubescens and N. setosa (Fig. 61). However, these sub-aphlebioid basal pinnae are not always present in N. Grevilleana. They are clearly present in fourteen of the eighteen collections I have studied which include the petiole base. Recently in Jamaica I examined and collected three individuals clearly of this species, two of which (Gastony & Gastony 967, 976) lacked any indication of these structures. Another interesting feature of Nephelea Grevilleana is the occasional squamuloid nature of the paraphyses at the apex of the receptacle. These are uncommon in other species of Nephelea, but appear to be more fre- quent here. Morphologically, they seem to be derived from abortive sporangia and their apical position on a receptacle which develops its sporangia somewhat acropetally suggests they may be related to nutri- tional or other metabolic inadequacies. These structures are best observed in immature sori, where they comprise a sort of cushion within the open- ing of the more or less urceolate indusium. Their possible role in pro- tecting the immature sporangia from desiccation or other environmental hazards at the mouth of the indusium possibly confers a selective advan- tage to plants possessing them. This species is endemic to Jamaica, where it occurs widely at altitudes of 200-950 m. in moist ravines and river valleys or in dense, wet forests, especially on cloud bathed mountain slopes. It is reported to be found on limestone and metamorphic rock. ‘AL SPECIMEN: Jamaica. Portland: Clute (Ny); Gastony & astoay, 967, 971 (cH); pitas ieee gg toe us); Maxon & Killip 149, 820 (¥,cu,Ny,us). 102 GERALD J. GASTONY ca aseR AN al elat a by, ; : YE y AVY 4 aA ERE BY SU NAD Fics. 32-35. Nephelea Grevilleana: 32, Central pinnules of a central pinna, Proctor 4815 (mo), x 0.5; 33, Central portion of a pinnule in 32, x 2; 34, 35, Two views of dark-setate, pouch-like bullate scales from the adaxial costule surface, both Gastony 101 (om), < 85. Fics. 36. 37. Nephelea crassa: 36, Central pinnules of a central pinna, portion of a pinnule as shown in 36, Ekman H15723 (F), X 1.5. Fies. 38, 39. Nephelea fulgens: 38, Central pinnules of a pinna, Gastony et al. 603 (cH), X 0.5; 39, Central portion of a pinnule in 38, x 2. Fics, 40, 41. Nephelea portoricensis: 40, Central pinnules of a central pinna, Sintenis 2594 (cH), x 0.5; 41, Central portion of a pinnule as shown in 40, Britton et al. 3909 C2), 4 3. : THE FERN GENUS NEPHELEA 103 St. Thomas: Day 2, 3, 4 lary Fredholm 3236 ( wih arrows 101 (cu); Hatch 14 (Ny,uc,us ); Maxon 2403, 443 (ny,us) 8896 ) (GH,NY,Us); Proctor 3 (mo,us); Wilson ¢ M urray 561 (a,us). St. Reta! Andrew: Maxon 10321 (GH,Ny,us). St. Andrew: Fisher 65 (xy); Gastony 27 (cH); Maxon 1020 (us); Maxon & Killip 1358 (¥,cu,ny,us); Proctor 4815 (Mo); Underwoood 1584, 1585, 1624 (ny). St. Catherine-St. Ann: Da ay 130 (GH,NY); Maxon 1945, 10453 (us); Underwood 1820 (ny). St. eee R. C. Alexander Side Campbell or Harris 7725 (F,Ny,us); Hatch 10 tj (us); How ; Ppa 15515 Manchester: thal (Ny). Trelawny: Gasto ony & Gastony 976 cH). St. Eliz ior Maxon & Killip 1473 (¥,cH,Ny,us). Westmoreland: Proctor 4633 (us); Wilson & Webster 505 (a,us). 4. Nephelea crassa (Maxon) Tryon, Contrib. Gray Herb. 200: 38. 1970. Fics. 36, 37, Mar 5 vues a cr ih Maxon, oe U.S. Nat. Herb. ms 40. 1909. type: Santo Domingo (Dominican Republic) in sylva montis “Isabel de la Torre,” 550 m., 8 Mgrs Pia Bueers 2735c, US 529878 (incorrectly cited by an as 523876); isotype: v Cyathea domingensis Brause, in Urban, Symb. Ant. 7: 153. 1911. tTypE: cae Domingo (Danis Republic) ad Maniel prope Barahona 600 m. alt. in sylvis montanis, Dec. 1908, H. von Tiirckheim 2715, presumably ay B (not seen); isotypes: F!,cu!,Mo!,Ny!,p!us!. Stem to 7.5 m. tall and 7 cm. in diameter, bearing brown scales especially toward the apex and blackish spines to 5 mm. long, adventitious buds not known. Unexpanded croziers with blackish spines to 5 mm. long and brown scales at a Ha apical seta .25 mm. or less long. Petiole to ca. 0.13-0.38 m. long, brown to purple-brown with blackish spines to 9 mm. (usually less) long, the scales on the abaxial surface sparingly persistent, brown, with a dark apical seta to 0.5 mm. (usually less) long and without ateral setae. Lamina rigidly papyraceous, to ca. 1.7 m. long and 1.2 m. wide, long and 21 cm. wide; basal pinnae highly reduced, to ca. 8 cm. lon ng; pin na-rachis the scales not significantly clustered in the axil. Segments straight to sub- floats with lateral margins serrate-crenate (the teeth often shallowly to moderately bifid, cor- responding to the vein endings), or rarely pin innatilobate to pinn innatifid, apex al y an tween the : forking of the veins; indusium cyathiform to urceolat cent of firm texture, per- sistent and usually entire at maturity, pypaus ‘recepta acle a inserted to slightly exserted, with sparse short (ca. 0.2-0.4 long) paraphyses Although the more strongly squamuloid paraphyses occasionally found in Nephelea Grevilleana have not been observed in this species, the morphology of some of the larger paraphyses here is also suggestive of 104 GERALD J. GASTONY sporangial derivation. This is especially true of paraphyses found toward the apex of the receptacle. Nephelea crassa appears to be closely related to N. Grevilleana and there has been some confusion between them in the past. Specimens from the Barahona province of the Dominican Republic in particular caused Maxon (1924, 1937 in Christensen) to vacillate in his recognition of these two taxa. He finally concluded that some Barahona specimens were Cyathea elegans. These specimens have also proved difficult for me, but I now feel confident that they are N. crassa and that N. Grevilleana does not occur in Hispaniola. These specimens are similar to N. Grevilleana in their cutting and the squamose indument on the abaxial costule surface, but I have seen no material of N. crassa with sub-aphlebioid basal pinnae, even though these pinnae may be quite reduced in size. Furthermore, the specimens most like N. Gravilleana in their cutting lack the stiff trichomes on the adaxial costule surface which characterize that species. Most specimens of Nephelea crassa have a glabrous or glabrescent abaxial costule surface. Maxon took this to represent the total variation of the species, but the more extensive collections now available for study show that this represents only one extreme of the total species variation. There is a clinal variation indicated by a higher incidence of scales and pubescence on the abaxial costule surface in the western and southwestern (Barahona) part of the species range to glabrescence in the northeast and east. This supports the view that the species originated in the western and southwestern portions of Hispaniola from an ancestor common to it and N. Grevilleana, that it lost the more primitive sub-aphlebioid condition of the basal pinnae, and that as it migrated to the east and northeast most of the costule indument was lost, while the pinnules became more triangular with a more prolonged apex. This species is endemic to Hispaniola where its range extends nearly throughout the length of the island mostly at altitudes of 100-600 m., although it also has been collected at 1065 m. and 1250 m. It is found in moist ravines and lowland rain forest and occasionally on mountain slopes. ADDITIONAL SPECIMENS EXAMINED: Dominican Republic. El Seibo: Abbott 2662 H15119 (cu,us), H15294 (wny,s,us). San Cristobal (formerly Santo Domingo): ( H na: Fuertes 741 (Ny,us), 742 (F,cH,Mo); Howard 8483 (GH,Ny,us); Tiirck- heim 2714 (¥,cGH,Mo,Ny), 2716 (us). Haiti. Quest: Ekman H5522 (vs). 5. Nephelea fulgens (C. Chr. ) Gastony, comb. nov. Fics. 38, 39, Mar 6 Cyathea fulgens C. Chr. Kungl. Svensk. Vetens.-akad. Hand. ser. 3, 16; 14, t. 1 (Figs. 9-12). 1937. type: Haiti, Nord, St. Louis du Nord, Morne Chavary, 900 m., on laterite, 25 Aug. 1925, Ekman H4721 s!; isotype: us! (two sheets). THE FERN GENUS NEPHELEA 105 tem ~ 4 m. tall and 6.5 cm. in diameter, bearing brown scales age toward the apex and blackish spines to 9 mm. long, adventitious buds not known. nded croziers with blackish spines to 7 mm. long and brown scales oh a ae. oie seta 5 mm. long and without lateral — Petiole to 0.36 m. long, dark brown, with blackish spines to ca. 10 mm. long, the scales on the abaxial surface usually deciduous above their blackish spine- -like bases, packed a scabrous surface, brown with a blackish central — -_— faa — seta to 0.1—-0.2 mm. long and without lateral setae. ong and 1.2 m, wide, tripinnate to tripinnate- pinnatifid (predominately Tipnemaieee tees with a gradu ally acute at dh Rachis dark brown, adaxially minutely pubescent to glabrescent, abaxially glabrescent to glabrous. Central pinnae sr to short (to ca. 1.5 cm.) petiolu — day hea tg to ca. 55 cm. long an m. wide; basal pinnae reduced, oft ca. . long; r c pinna-rachis adaxially pubescent, abaxially furfuraceous to mo oat glabrescent, with eenis crassa; 6, N. fulgens; 7, portoricensis; 8, N. woodwardioides var. woodwardioides; 9, N. Saks ddan var. cxibesiaie; Pag N. woodwardioides var. Hieronymi; 11, N. Tussacii; 12, N. concinna, Maps oi 2, Nephelea pubescens; 3, N. bal nance 4, N. Grevilleana 106 GERALD J. GASTONY dark apical setae; veins mostly once-forked at or near the costule, more forked or orne on secondary costules in more strongly crenate to lobed segments, evident on the veins; indusium sphaeropteroid to sub-sphaeropteroid, brown, of more or less firm texture, persistent but rupturing into usually 2-3 lobes at maturity, glabrous; receptacle globose to columnar, inserted, paraphyses to 0.1-0.4 mm. long. As in the case of Nephelea crassa, the morphology of the larger para- physes of N. fulgens suggests their sporangial derivation. Unlike all other Greater Antillean species of Nephelea, which have meniscoid to cyathi- form or urceolate indusia, the indusia of N. fulgens are sphaeropteroid to sub-sphaeropteroid. It is further distinguished from N. crassa, which may be its closest relative, by its coriaceous to strongly coriaceous lamina texture, the more blunted teeth of its ultimate segments, the usually conspicuous (at a magnification of 15 or more) stomatal apparatus abaxially and reticulate pattern of the dried lamina surface adaxially, the greater scabrosity of the petiole base due to the persistent bases of the more indurated scales, and by its occurrence at generally higher altitudes. Nephelea fulgens is endemic to Hispaniola, where it occurs at altitudes of ca. 800-2000 m. on shrubby to forested moist mountain slopes. It is reported from lateritic and calcareous soils. ADDITIONAL SPECIMENS EXAMINED: Dominican Republic. Duarte (formerly part of Pacificador): Abbott 2068 (cH,NY,Us); Ekman H12279 (s,us). Santiago: Jiménez 3863 (us). Santiago Rodriguez (formerly part of Monte Cristi): Ekman H12862 (s,us ). San Rafael: Gastony et al. 506, 603 ( GH,NY,Us ), 6. Nephelea portoricensis (Kuhn) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 40, 41, Mar 7 Cyathea portoricensis Spreng. ex Kuhn, Linnaea 36: 163. 1869. TyPE: Puerto Rico, Balbis, perhaps 8 (not seen); isotypes: s! in herb. Kunth, ny! fragment of isotype or seein of holotype ex herb. s, possible isotype portion P! herb. Luerssen 10486 ex erb, Lz. and 6.5 cm. in diameter, bearing brown scales especially toward ot known. Unexpanded croziers with blackish spines to 11 mm. long and brown scales with a dark apical seta 0.25 mm. or less long and without lateral setae. Petiole to ca. 0.5 m. long, dark brown and without lateral setae. Lamina rigidly papyraceous length not known, to ca. 1.2 m. wide, with a gradually acute apex. Rachis dark brown to purple-brown, adaxially pubescent and occasionally squamose, abaxially glabrescent. Central pinnae sessile to THE FERN GENUS NEPHELEA 107 in the axil. Segments sub- ‘falcate with lateral margins subentire to 0 crenate-serrate, a one to few stiff trichomes distally, abaxially pubescent with flat to sub-bullate squamules (to ca m. long) with several dark apical and lateral setae; veins once forked or occasionally more branched at, or about 0. m, the costule, evident on both surfaces, adaxially glabrous, abaxially usu ally with stiff trichomes; lamina surface adaxially and abaxially wv (rarely abaxially pubescent) between the veins. Sori at the ica ary forkin e pis indusium cyathiform to urceolate, brown (at maturity), of firm aie aoy persistent, and usually entire at maturity, pubescent with more or less stiff white fi cna (rarely these trichomes bi-seriate to multi-seriate and squamuloid, or with actual squamules ); receptacle glo obose to oc- casionally columnar, inserted, with inconspicuous paraphyses 0.1-0.2 m ae. — Nephelea portoricensis is the only species of Nephelea in the West Indies with pubescent indusia. Its affinities are certainly with the other Greater Antillean Nephelea species, although unlike them the scales on its abaxial costae and costules are never more than sub-bullate. A collection from Indiera Fria, near Maricao (Britton, Cowell & Brown 4520), is peculiar in its very narrow pinnae (largest pinnules ca. 4 cm. long and 1 cm. wide). However, the critical characters of the material are certainly those of Nephelea portoricensis. Evidently a juvenile form is represented. Nephelea portoricensis is endemic to Puerto Rico and is the only species of Nephelea on the island. It occurs at altitudes of ca. 500-1100 m. in ravines and on forested mountain slopes in a more or less central band nearly throughout the length of the island. ONAL SPECIMENS EXAMINED: Puerto Rico. Blomquist egy ere 11965 (F), 13134 yimty seco S| Britton 7207 (Ny,us); Britton & Cowell 913 (Ny,us); Britton et al. 3909. i (GH,M abies ou F,Mo,NY), 4102 (a,cH. 20s), 58 (GH,M,us); Stevens 1411 (nx); Tryon & Tryon 7082, 7085 (cu); Underwood My Griggs 272 (ny 2145 7, Nephelea woodwardioides (Kaulf.) Gastony, comb. nov. 124 Cyathea woodwardioides Kaulf. Enum. Fil. 255. 1824. type: without collector or ocality, weaeraree destroyed at Lz; fragment “Ex Herb. Kaulf. fragm. origin., ohne standortsangabe” P Stem to 9 m. tall and 30 cm. in diameter Lineloding the investing sheath of adventitious roots), bearing brown scales especially toward the apex and blackish spines to 15 mm. long, occasionally with adventitious buds. Unexpanded croziers with _ blackish spines to 8-15 mm. long and brown scales with a dark apical seta to 0.25-0.5 mm. long and without lateral setae. Petiole to ca. 0.5 m. long, stramineous to dark 108 GERALD J. GASTONY brown with blackish spines to 8-14 mm. long, the scales on the abaxial surface ene persistent, brown with a dark apical seta to 0.25 mm. long and without lateral se setae. Lamina rigidly papyraceous to coriaceous, to ca, 2.5 m. long and 1.3 m. wil ; : : d c blron Bvt, narrowly ovate to triangular or elliptic, the apex acute to acuminate; costa adaxiall pubescent, some w with occasional long, narrow, brown scales cone pubescent distally, the to with brown to black ete and sash margins a ne es severa pac we setae, not significantly clustered in axil. Segments ‘ruth to sub-falcate or aes with lateral margins entire to poet (rarely especially the basal segments lobate), apex subentire to serrate- crenate, margins slightly revolute; costule adaxially gla brous or rarely with one to scales to 0.5-1.0 mm. long and pubescent to glabrescent distally; veins once- orked at or near the costule, ge more forked, evident on both surfaces or somewhat obscure adaxially, adaxially wee _abaxially glabrous; lami adaxially and abaxially glabrous between the veins. Sori i the veins; indusium cyathiform to urceolate Sais. of more or less firm texture, persistent and entire at maturity, glabrous; receptacle globose to columnar, inserted to exserted, with mostly inconspicuous short (to 0.2 mm. long) paraphyses oa 5 > ~ Sa len! a. 5 ga ° Leal KEY TO VARIETIES Axes usually dark; axil of costa (and sieges of costules) with a conspicuously prominent dark pneumatode; lamina coriace costa surface brown to blackish-brown). J Axes usually stramineous to light brow aie axil of c ote without a dark pneumatode or the pneumatode not conspicuously prominent; ati rigidly papyraceous to sub- coriaceous. Segment margins subentire to (especially at the a et serrate- es ‘oad abaxi- ally sparsely pubescent to glabrescent distally and with bullate to sub-bullate scales; centers of largest scales on abaxial costa surface mostly brown {occasionally black- perrmown). La a Segment margins entire to serrate-crenate; rn gana costule ‘iuxtally ‘pubescent distally and with bullate to sub-bullate scales; centers of largest scales on —— costa surface blackish-brown to black. Hispanio OR eines 7c. var. Hierony wee! Jy. Nephelea woodwardioides var. woodwardioides Fics. 42-44, Mar 8 Cyathea nigrescens Auct. not (Hook.) J. Sm., e.g. Maxon North Amer. Fl. 16: 74. 1909. Cyathea araneosa Maxon, North Amer. FI. 16: 74. 1909. rvpE: Cuba, Oriente, en slopes and summit of Gran Piedra, 900-1200 m., April 14, 1907, Maxon 4035, 522680!; isotypes: F!,cu!,M!,ny!,uc! ws!. THE FERN GENUS NEPHELEA 109 ~ 970. Cyathea hystricosa Mort. Amer. Fern Jour. 5B: 30. 1965. Ty : Jamaica, ca. 1805 Wiles, nm, Morton photo 8097 (not seen; from oa eos iwiox gh cited enh ince on the na gradually acute apex. Rachis brown to purple gyal Central pinnae capes to short (to ca. 1 cm.) petiolulate, ove mueeaier i to 63 cm. long and 18 cm. wide; basal j e reduced to 19 long; ae eHerk abaxia ‘ep finfuracéons- glabrescent, pubescen t distally, with “swollen, dark, conspicuous lateral pneumatodes at costa axils. Pin nnules sessile to short (to ca. 3 mm.) petiolulate, narrowly ovate to triangular; costa ad pubescent with re anal long, narrow, brown scales oO: = 3 This variety is best distinguished by the usually quite conspicuous pneumatodes at the costa axils and occasionally at the costule axils. Similar but much less conspicuously developed pneumatodes are fre- quently found in other species of Nephelea at the costa axils, but only in this variety have they been observed (occasionally) at the costule axils. The portion of the holotype of Cyathea woodwardioides Kaulf. that I have seen is only a fragment; however, the application of the name is certain from its characters. The collection was probably made in Jamaica, where early collectors were active, and where var. woodwardioides is common. In describing Cyathea araneosa as new, Maxon was undoubtedly strongly influenced by what appears to be a delicately whitish araneose-ciliate margin of the indusium in the type collection. This condition has also been found in a collection from Jamaica, but in both cases close examination reveals that it is simply a fungus—the mycelium of a species of Nectria. Variety woodwardioides occurs in Jamaica at ca. 725-1430 m., predom- inantly in the Blue and John Crow mountains, with a disjunct station in the mountains of Clarendon. It is found in cloud forest (often associated with Podocarpus) and rain forest in exposed and wooded situations on ridges, in ravines, and on mountain slopes. In Cuba it is known in similar habitats only from the most southerly mountains of Oriente (the Sierra Maestra and Gran Piedra) at 900-1200 m. altitude. ADDITIONAL SPECIMENS EXAMINED: Jamaica. Jenman 41 (uc). Portland: Gastony 93 (cH); Raion & Killip 827 (¥,cu,ny,us); Tryon & Tryon 6979 (cH). St. Thom Maxon 9219, 9460, 9512, 9557 (Ny,us), 9426 — us), 9492 (us). Portland-St. Maxon & Killip 984 (¥,cu,ny,us). St. Andr bien 1804 (us); Proctor ), Pei (ny). Cuba. Oriente: Clément 460 (ny,us), 1665 (us); Hioram 7312 (us); oram & Clément 6374 (us), 6375 (cu,us); Leén et al. 10532, 10534 (Ny,us). 110 GERALD J. GASTONY 2132 Tb. Nephelea woodwardioides var. cubensis (Maxon) Gastony, comb. et stat. nov. Fics. 45-47, Mar 9 B a? Cyathea cubensis Underw. ex Maxon, North Amer. FI, 16: 73. 1909.“rveE: Cuba, Baracoa, base of El Yunque Mt., March 1903, Underwood ¢& Earle 1313, ny! (three s 41 %2a4 Nephelea cubensis (Maxon) Tryon, Contrib. Gray Herb. 200: 40. 1970. Adventitious buds on the stem not known. Unexpanded croziers with blackish spines to 8 mm. long and brown scales with a dark apical seta to 0.25-0.5 mm. lon a axils. Pinnules sessile to short t s : etiolulate, narrowly ovate to elliptic; costa adaxially pubescent, e basal segments lobate), apex more strongly serrate-crenate; costule adaxially glabrous, abaxially sparsely pubescent to glabrescent distally and with apically dark- setate bullate to sub-bullate scales, these often also with light-colored trichomoid In addition to the characters mentioned in the key, the often abruptly reduced pinna apex (sometimes nearly to the form of a pinnule) helps to distingush this variety. In comparison to var. Hieronymi, the ultimate segments in var. cubensis are more strongly and more frequently serrate- crenate at the apex. Of the three varieties, var. cubensis shows the strong- est tendency to possess sub-bullate squamules abaxially on the costule. Five collections from the Sierra Maestra (Ekman 5388, Ledén 11089, Morton 9292 and 9435, Taylor 449) seem at first rather distinct from var. cubensis. Less extreme states of each of the several characters which lend a distinctive appearance to these specimens can be found in other collec- tions of var. cubensis. It therefore seems best to regard the material avail- able as representing more or less extreme forms of var. cubensis. Perhaps most troublesome are the presence of squamules on the abaxial costule surface which are at most sub-bullate, making these specimens difficult to accommodate in the key to species. Variety cubensis is endemic to Cuba in the three major mountainous areas: southern Oriente, the Sierra de Trinidad in Las Villas, and the _ Sierra de los Organos in Pinar del Rio. It is found in moist situations in -Tavines and on ridges and mountain slopes in or at the edges of forest at ca. 400-1100 m. altitude, some specimens have been recorded from lime- stone areas. ADDITIONAL SPECIMENS EXAMINED: Cub, (us), 1654 (F), 1661 (uc), 1672 (M,us); Eggers 5252 (us); Ekman 3218, 3681 (s), 5388 (Ny,s,us); Hioram 6989, 6999, 7291, 7295, 7306 (us); Hioram & Clément 2 6381, 6384, 6387 (us); Hioram & Maurel 2420, 4094, 6033 (us); Leén 11089, 12638 THE FERN GENUS NEPHELEA 111 (ny,us); Ledn & Chateauvieux 3897 (ny); Leén & Maurel 3804 (nx); Leén et al. 10216 feel Maxon 4105 (cGuH,Ny,uc,us), 4206 (GH,M,Ny,us), 4226 (Ny,us), 4264 (F,GH,NY,Us ); Morton 9292, 9435 (us); Morton & Acuna 3620 (cH,Ny,uc,us); Pollard 1 7020 (GH,NY,us), 7117 (¥,GH,UC), 7231 (¥,GH,MO,Ny,US) & Clément 3989 (ny = fot 8108 (us); Leén & Roca 8104, 8108 (ny); er 4239 (GH,MO,NY,UC, Roig & Acuna 6125 (ny). Pinar del Rio: Ekman 10673 (us); Leén 1 12638 (wy, us), 12929 (s40,nx), 14140 (wx); Morton 4387 (at0,0s), 4873 (14 us); Roig Leén 6398 (x). YAO iy Calle ANAS =f Ch A hia Mbit Mh as LP Axa” NBG X MMMM ANNI SH Sica : eee UMM es 48 ANU Fics, 42-44, Nephelea woodwardioides var. woodwardioides: 42, Central ag of a | central i b of p his showing conspicuous dark p: s 45-47. Nephelea woodwardioides var. cu 4 entral pinnules of a central pinna, Morton & Acuna 3620 (cH), X 0.5; . Fao portion of a pinnule in 45, x 2; 47, cat pinnules of a > vari coria rm with no closely spaced segments, a longer tapering pinnule apex, and at seek sh mallets eager on the abaxial costule surface, Taylor "9 (Nx), x ‘a. 5. Fies. ay 49. Nephelea woodwardioides var. Hieronymi ar Central pinnules of a central pinna, Gastony et 203 (GH), 0.5; 49, Central portion o: bi dikes xX 2. 112 GERALD J. GASTONY 21225 Tc, Nephelea woodwardioides var. Hieronymi (Brause ) Gastony, comb. et stat. nov. Fics. 48, 49, Mar 10 brown centers. Segments sub-falcate to falcate with lateral margins entire to serrate- enate, apex subentire to serrate-crenate; costule adaxially occasionally pubescent distally and with apically dark-setate bullate to sub-bullate scales. Among the varieties of Nephelea woodwardioides, the spines on the croziers and petioles reach their greatest development in var. Hieronymi. This is the only variety where evidence of adventitious buds on the stem has been observed. In Gastony et al. 156, these buds gave rise to six branches at 0.3-0.5 m. above the ground and probably arose when apical ominance was broken by the previous decapitation of the main stem at 4m. The maximum diameter of this stem was 30 cm. in the region of the adventitious root sheath; above the adventitious roots and below the persistent petiole bases the stem diameter was only 10 cm. Variety Hieronymi is endemic to Hispaniola where it is fairly common and its range is quite extensive at ca. 300-1850 m. (especially above 800 m.). It occurs on calcareous and lateritic soils of mossy cloud forest and rain forest in partial to well-shaded situations, on mountain slopes and in ravines and sink holes. ADDITIONAL SPECIMENS EXAMINED: Dominican Republic. Duarte (formerly Pacifi- cador): Abbot 2031 (cx pro parte, Ny,us). San Cristobal (formerly Santo Domingo) : Allard 14304 (us); Ekman H11501 (F,NY,S,us). La Vega: Allard 13488 (Ny,us); Puerto Plata: Ekman 4 NY,Us ) zua: Ekman H6344 sj. Ss R (formerly Monte Cristi): Ekman H12699 (s). San Juan: Howard 944] (GH,Ny,us). B : Ab 666 (F,cH,Ny,us); Howard 12323 (GH,NY). San - : Gast ,Us). Haiti. No: Nash & Taylor ony et al. 435 (cH, 1743 (Ny,us). Ouest: Buch 1135, 1551 (us); Ekman H1291 (¥,GH,s,us), H1764 (us), H3227 (s,us); Holdridge 2051 (¥,nx); Leonard 3814 (cH,NY,us), 5356 (us); Proctor 10799 (us). THE FERN GENUS NEPHELEA 113 8. Nephelea Tussacii (Desv. ) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 50-52, Map 11 Cyathea Tussacii Desv. Mém. Soc. Linn. Paris 6: 323. 1827. (= Prod. 323. 1827). TYPE: Jamaica, De Tussac, P (not seen), photos cH!, uc!, us!. Another sheet at p! with this name and “In Jamaica,” Herb. Desvaux is taken as an i otype. Cyathea Tussacii var. magnifolia Jenm. Bull. Misc. Inform. Roy. Bot. Gard. Trinidad 3 (15): 53. 1898. (= West Ind. and Guiana Ferns, 53). TYPE: Jamaica, very abundant in forests 4,000-6,000 ft. alt., chiefly in ee —— ravines, probably Jenman, probably Trin or k (not seen); ‘possible isotypes: N Stem to 8 m. tall and 10 cm. in diameter, bearing brown to tan scales especially misc 0 the apex and blackish spines to 12 mm. long, adventitious buds not known. Unexpanded croziers with blackish spines to 11 mm. ong and with b to tan n the abaxial surface sparingly persistent, brown, some arker central stripe, with a single dark apical seta to 0.25 mm. (usually les) long and without lateral setae. Lamina rigidly papyraceous to sub-coriaceous, to m. long and 1.6 m. wide tripinnate ace wees bipisineteeiteabiia). with a gradually acute apex. Rachis light Hgts cep urfuraceous-pubescent and often squamose, some scales with abaxia m. wide; basal a. ig an uced, to ca. 36 cm. long; piss whe adaxially a : axi between the more distal sessile pinnules. Pinnules to ca. 14 cm. long and 3 cm. wide, deeply pee atifid (pinnate at the base), er i to Pips gael the = bes long, flat to bullate and usually with a sing dark nse seta ecenale more), occasionally with scales clustered in Rerccgpate sub-falcate to falcate with lateral margins and apex entire to ss gai (bas i a ~ ee: margins slightly revolute; costule st? ially gla brous or occasio ly w to stiff oo dista St abaxially pubescent and with diskactats to non-state bullate to e n immature ); gies or By rat to columnar, inserted to slightly exserted, with short (to ca. 0.2 mm. long) paraphyses, occasionally tufted at the apex of the receptacle. There are six sheets of Cyathea Tussacii var. magnifolia at Ny in the Jenman herbarium which bear his label “Type specimen—collected 1874- 1879.” The status of these specimens as isotypes cannot be determined. They are considered to be possible isotypes because they are in the Jenman herbarium and bear the varietal epithet in Jenman’s hand, not because of Jenman’s “Type specimen” labels. These labels are irrelevant to and should not be considered in typification, since there are several instances in which collections clearly without relevance to the nomen- clatural type bear such labels (e.g., five sheets at Ny and one at us anno- tated by Jenman as C. Tussacii Desv.). 114 GERALD J. GASTONY The combination of indusial type and strongly squamose axes abaxially makes Nephelea Tussacii a very distinctive species. Jenman (1898) noted that it drops all its leaves in the dry season. Its affinities with N. concinna are discussed under that species. Nephelea Tussacii is endemic to the Blue Mountains of Jamaica where it occurs in mossy cloud forest or on shaded wet slopes and in moist ravines at altitudes of ca. 750-1650 m. ADDITIONAL SPECIMENS EXAMINED: Jamaica. Portland: Gastony 21 (cH); Hatch 7, 9 (uc,us); Maxon & Killip 1288 (¥,cH,Ny,us); Moore & Read 9559 (cH); 1932, Papenfuss (uc,us); Proctor 5210 (mo,us), 5712 (mo). Portland-St. Thomas: Maxon 9805 (cH,Ny,us); Maxon & Killip 1191 (¥,CH,NY,Us). St. Thomas: Hatch 13 (vs). Portland-St. Andrew: Chrysler 1687 (mo). St. Andrew: Campbell 7724 (¥F,Nny,us); Clute 164 (ny,us); Hart 33 (us); Maxon 1043 (us); Underwood 1528 (Ny,us), 2166 (Ny). 9. Nephelea concinna (Jenm.) Tryon, Contrib. Gray Herb. 200: 38. 1970. Fics. 53-57, Map 12 Cyathea arborea var. concinna Bak. ex Jenm. Jour. Bot. 19: 52. 1881. type: Jamaica, 1879, Jenman 2, x! Cyathea concinna (Jenm.) Jenm. Bull. Dept. Jam. 26: 4. 1891. Stem to 6 m. tall and 6.5 cm. in diameter (20 cm. in diameter including the investing sheath of adventitious roots), bearing brown scales especially toward the apex and to 6 ds not kno g blackish spines to 6 m ng, adventitious bu Unexpanded croziers with blackish spines to 10 d brown scales with a dark apical seta without lateral setae. Lami aceous to rigi apyraceous, to ca. 2.5 m. long and m. wide, bipinnate-pinnatilobate to bipinnate-pinnatifid, wi adually acute apex. is b , adaxia pubescent-squamose to glabrescent, abaxia furfuraceous to glabrescent. Central pinnae sessile to short (to ca. 0.5 cm. iolulate, narrowly ovate-lanceolate, to ca. 45 cm. | nd 10 cm. wide; b innaé more or acute to blunt; costa adaxially sparsely pubescent to glabrescent, abaxially pubescent or distally so an uamose, the scales tan to brown, mostly bullate to sub-bullate, with one to few dark apical setae, occasionally clustered in the axil Segments vario' p fertile region where e may be merely crenate to lobate) straight to sub-falcate with lateral argins entire to crenate crenulate, margin htly or not revolute; costule adaxially glabrous or with one to cyathiform to urceolate, brown, of firm texture, persistent and usually entire at maturity, glabrous; receptacle globose to occasionally columnar, inserted to slightly exserted, with short (to ca. 0.2 mm. long) paraphyses, occasionally tufted at the apex receptacle. THE FERN GENUS NEPHELEA 115 A sheet of Cyathea arborea var. concinna at ny! in the Jenman herbari- um and bearing his “Type specimen” label is probably not part of the same collection as the holotype at xk, although it is an authentic representative of the name. A specimen at P! in the Hart herbarium annotated as C. con- cinna, presumably by Jenman and referred to by him in a note on the type sheet at x, is probably not part of the type series either. The status of Nephelea concinna as a species appears to be somewhat questionable. It is certainly very closely related to N. Tussacii, from which it is usually best distinguished by the size and degree of complexity of the leaf. Given that the leaf of N. concinna is smaller and less complex than that of N. Tussacii (maximally bipinnate-pinnatifid versus tripinnate), it is not unexpected that its veins are also less branched and that the indu- ment characters associated with axes of higher orders in the more complex leaves tend to be shifted toward axes of lower order in the more simple leaves (cf., major heading separating these species in the key ). The irregu- lar appearance of the partially contracted fertile pinnules found through- out the leaves of N. concinna suggests a teratological condition. This is quite similar to the condition found elsewhere in highly fertile leaves where the fertility extends to the very distal pinnae which are transitional to a lower degree of complexity than that predominant in the leaf. At least one collection I have seen (Gastony 24) would certainly be placed in N. Tussacii on the basis of its stem, croziers, and indument characters, were it not for the size and complexity of its leaves. Another point of congruity between these two taxa is their uniqueness among the Jamaican Cyatheaceae in reportedly dropping all their leaves in the dry season (Jenman, 1898). It might be suggested that what is called N. concinna is simply a form of N. Tussacii produced by ecologic catastrophes such as exposure resulting from landslides or exceptionally stressing dry periods. I have observed in the field and collected portions (Gastony & Gastony 951) of a large perfectly healthy fertile specimen, however, which grew among other tree ferns in a quite undisturbed habitat. The spores and sporangia of Nephelea concinna appear quite normally developed, but it is possible that N. concinna is a hybrid between N. pubescens and N. Tussacii. Nephelea pubescens has the least degree of laminal complexity in the genus and might contribute to the lesser laminal complexity and diminutive basal pinnae of N. concinna when compared to N. Tussacii. Hybridity could also account for the teratological-like aspect of the fertile pinnules of N. concinna. All three taxa are sympatric and endemic to the Blue Mountains of Jamaica. Until field population studies and related laboratory investigations of these taxa can be made, however, it seems best to maintain them as distinct species recognizable by the distinguishing characters indicated here and in the key to species. Nephelea concinna is endemic to the Blue Mountains of Jamaica where it occurs in cloud forest and rain forest on ridges and slopes and in ravines at altitudes of ca. 1065-1675 m. 116 GERALD J. GASTONY ADDITIONAL SPECIMENS EXAMINED: Jamaica. Hart 166 (us). Portland: Fisher 95 NY); Gastony 24 (cH); mare! ‘a Gastony 951 (cH); 1919, Pessin (us); Riba 198 (cH). Portland-St. Thomas: Maxon 10039 (cu,Ny,us). Portland-St. Andrew: Chrysler 1659 (us); Maxon 1576 (us); Bal ced 2635 (ny). St. Andrew: Harris 7721 (us); Proctor 6691 (us); Underwood 1532 (Ny,us); Watt 12 (us). 10. Nephelea setosa (Kaulf.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 58-61, Map 13 Alsophila — oper ews Fil. 249. 1824. type: Brazil, Chamisso, presumably destroyed at Lz; isotype: Hemitelia setosa PE att ) Mett. Fil. Lechl. 2: 30. 1859, Cyathea setosa (Kaulf.) Domin, Pterid. 264. 1929. Cyathea ed oainnia iis ex Hook. Icon. Pl. 7: t. 623. 1844 and Sp. Fil. 1: 21. 1844. (Presl, Tent. Pterid. 55. 1836. nom. nud.). Isosyntypes: Brazil: Sellow p!, upc photos uel, us!; sere , L photos ny!, us!; Rio de Capek Gardner 135 m!. 1 am ooker Hemitelia Beryichiana (Hook.) Presl, Peet rte Stipes der Farrn, 45. 1947 ( a sate Abhandl. béhm. Ges. V, 5:353. Amphic . Civ. re nevis dorsalis Fée, Crypt. Vasc. Brésil. 1:173. 1869. type: In Bras fluminensi, in loco wh Novo Friburgo, ad ripas rivulorum, Glaziou 2287, p! in hash. Cosson; isoty us!. Alsophila ps (Fée) Christ, Bull. Herb. Boiss. II, 2:648. 1902. Cyathea dorsalis (Fée) Domin, Pterid. 262. 1929. Cyathea leucosticta Fée, Crypt. Vasc. Brésil. 1. 182. 1869. type: In Brasilia fluminensi, Tijuca, Glaziou 1700. v! in herb. Casson; isotypes: fragment in herb. anpert F!, f, : Hemitelia setosa var. crenata Rosenst. Hedwigia 46:64. 1906. TyPE: Brazil, Santa Catarina, Lages, ne Annita Garibaldi, Spannagel 134, presumably s-pa (not ; isotype: us! o 10 m. tall and 10 cm. in diameter, scales not seen, bearing blackish spines (length penta, occasionally with adventitious buds. Unex expanded croziers not Petiole to ca. 0.05-0.15 m. long to the sub-aphlebioid basal pinnae, light brown, with blackish spines to 4 mm. long, the scales on axial surface usually numerous, occasionall, duous, tan to brown, usually with a dark brown central s , with a dark apical seta to 0 mm. long and ati with smaller additional apical “= later: mina ri apyraceous to riaceous, length not know : .2 m. wide, na cee (predominately bipinnate-pinna ng usually abruptly phic o a mo inct pinna-like a ong). Rachis light brown, adaxially pucgaee. pre ally fhe aden bent pinnae sessile to ca. 3.5 cm. petiolulate, reir go to ca. 64 cm. lon and 20 cm. wide; basal reduc rachis adaxially aa eet abaxially uraceous to mostly glabrescent, pu scent 2 ] . . . . ] margins crenate to sisladinies (occ: soar vee sears apex sible ae margins usually slightly revolute; costule adax y rous or with one to several stiff trichomes distally, abaxially pubescent aa with occasional squamules to ca. 0.5-0.75 mm. Jong, often with numerous trichomoid processes and wi or without dark i THE FERN GENUS NEPHELEA 117 FAM L anal "ey on ll oe MN TUG - Mle 58 ani rn Wh WM: ‘. Sh gy a ‘ wae . 50-52. Nephelea Tussacii: 50, Central pinnules of a central pinna, Maxon ¢& Killip 1191 jaye x 0.5; 51, Central portion of a pinnule in 50, x 1.5; 52, Gs secesedititive cyathiform indusiu as in 50. 5. Fics. 53-57. Nephelea concinna 53, Central sporangi : 5 a pinnules of a sterile central pinna, with adjacent pinnules of unequal length, Maxon 10039 (vs), x 0.5; 54, Partially fertile pinnule of a central pinna, Underwood 2365 (ny), 0.5; 55, Fe i segments. tr e pinnule in 58-61. Nephelea setosa: 58, Central pinnules of a cen’ pinna, Smith & Brade 2201 (cH), x 0. 5: 59, Central portion of a pinnule in 58, 2; 60, A peak indusium the receptacle toward the i pubescens and N. Grevilleana. pinna, Steyermark 30009 (us), x Os, 63, Central portion of a pinnule in 62, 118 GERALD J. GASTONY adaxially glabrous, abaxially glabrous or with an occasional stiff trichome; lamina surface adaxially and abaxially glabrous between the veins. Sori at the primary forking ics glabrous; sere columnar to ee with more or less conspicuous to Tryon (1971, p. 18) has discussed the evidence suggesting that the Fée tree fern specimens in the Cosson Herbarium at P are types. A sheet at P! has been annotated by Christ as a new species of Cyathea with an epithet honoring ee collector (Brazil, Parana, Ponta Grossa in sivula, 12 XII 1908, P. Dusén 3219), but this name apparently was never published. Because of its sub-aphlebioid basal pinnae and hemitelioid indusia, Nephelea setosa is one of the most distinctive species in the genus. It is unique in its indusial type and shares the basal pinnae condition only with N. pubescens and N. Grevilleana. In these two characters, it is reminiscent of Alsophila capensis (L£.) J. Sm. and is considered to be primitive. Its relations to other species of Nephelea are not clear. As in several other species of Nephelea, it occasionally shows a tendency toward dimorphism in the lamina contraction of its fertile pinnae and rarely has pinnules which disarticulate from the pinna-rachis, at least with drying. Nephelea setosa occurs in southeastern Brazil in the states of Rio de Janeiro, Sao Paulo, Parana, Santa Catarina, and Rio Grande do Sul, in the Misiones province of Argentina, and in the Sierra de Amabay in Paraguay. It is frequently found at low altitudes of ca. 20-1300 m. in ravines, campos and shady forests. SPECIMENS ieee pias Brazil. Rio de Janeiro: Brade 10018, 10610 (Rr), Claioa 980 (BM,us); LG. 99 (x); Lima 21 (Rr); Lutz 2051 (Rx); Sampaio 2055 (R,uc,us); Sick 8 (HB); baie & Brade 2201 (¥,cu,uc,us). Sao Paulo: 1913, Brade (Rn), 5824 (sp); Eiten et al. 2149 (ny,us); Gerdes 97 (vc); Leite 3565 (a); Luederwaldt 1197 {ve); 6619, 6621 bell 1925, Lutz (rR); Wacket 79 (uc). Parana: Dusén 3219 (P), 6764 (F,cH, m0 8), 14839 (s,us); Schwacke 82 ae Santa Catarina: Luederwaldt 6620 (sp); Miiller 163 (rn); Reitz 1 ee (R,s,us). Rio sige do Sul: Brauner 120 (¥,HB); Glaziou 319 (us); Jiirgens 64 (A,F,MO,s,UC,US), ens 68 (uc); Leite 696 (Ny); pal 1123 (s,us); Sehnem 1314 (us). Argentin 2 ieibenes de la Sota & Cuezzo 1432 (¥,HB); Meyer 11834 (cu); Rodriguez 414 (cu). Paraguay. Shering 6622 (sp). Amambay: Rojas 10428 (Mo,Nny,uc). 24(@ 11. Nephelea Tryoniana Gastony, sp. nov. Fics. 62, 63, Map 14 tiarum Fax eshy spinis ae ad 11 mm. longis esertim apicem versus uamis brunneis fulvisve praeditus, gemmis adventitis incogni ieres non- evolutae non visae ages ad 0.4 m. longus, pallide brunneus, spinis nigellis ad 7 mm. longis instructus, squamae pagina abaxiali sparsae, fulvae brunneaeve margine pallidae apice setiferae om fuscata, ad 0.5 mm. longa) et saepe alias — need apice margineque ferentes. Lamina a papyracea, longitudine ignota, ad 1 . lata, raro tripinnata (plerumque bipinnata-pinnatifida) apicem versus aes wee Rachis _pallide brunnea, adaxi iter pu bescens-squamosa, abaxialit Fijgusard THE FERN GENUS NEPHELEA 119 ovatae, ad 63 cm. longae et 16 cm. latae; pinnae basales non valde deminutae, ad 30 cm. longae ate pokien). rachis pinnae adaxialiter pubescens (squamis — fulvis pote intermixtis ), abaxialiter squamosa furfuraceaque (saepe cde post lapsum arum majorum), distaliter pubescens Squamosaque squam bullatis, snatinae8 nee lateralibus ad axillam costae inconspicuis, sine alis viridulis inter pinnulas ue distallissimas. Pinnulae ad 8 cm. longae et 1.5 cm. latae, valde pinnatifidae (interdum basi vix st sessiles Nad bake g (ad 3 mm ay petiolulatae, anguste i ee versus pleru mque acum vel acutae; costa adaxialiter pubescens, raro squamis paucis pine: pe He ie si et squamis planis vel sub-bullatis ifbie Sulvesceatibus apice margineque setiferis (setis fuscatis) in- indusium deficiens; Se vege um columnare vel globosum, oe Shee ios inconspicuis brevibus (ad ca. 0.25 mm. longis) praeditum TYPE: Guatemala, ape ec capa: pine-covered cahions bordering Rio Lima, Sierr. de las gat Picacd Finca Alejandria, alt. 2000 m., Oct. 14, 1939, J. A. Steyerm at 9 ¥! consisting of six sheets (F1054354, sctoagoie 1054405, 1054409, 1054412, 1054418), one fa eet ( F 1054409 ) now at cu; isotype: u Nephelea Tryoniana is unique in the genus in its lack of an indusium and is also distinctive in the copious and very characteristic flat to bullate scales found abaxially on its axes. Although I have not seen the croziers, I am confident that they bear conspicuous black spines because of the highly spinose nature of other parts (e.g., stem portions of Steyermark 29925 and 30009). Although Nephelea mexicana and N. polystichoides, also from Central America, may at times appear to lack an indusium, critical examination will show remnants of their delicate, sphaeropteroid to sub-sphaeropteroid indusium at the base of the receptacle. Sterile material can readily be distinguished from N. mexicana and N. polystichoides by the projections bearing the larger scales on the abaxial pinna-rachis which leave a scab- rous surface when the scales fall or are eroded, and by the bullate scales present abaxially on the costule and distal costa. Such bullate scales are not found in other species of Nephelea outside of the Greater Antilles. Their presence in this Central American species and the form of the lamina apex in this species (discussed above under Morphology and Anatomy) suggest that it may be a relict member of a group once common to the Greater Antilles and the region of Guatemala, Honduras, and Nicaragua. The former land connection between these areas from the Mesozoic through the upper Miocene-lower Pliocene (varying in extent and with a brief break in the upper Oligocene—Schuchert, 1935), may have influ- enced this distribution. I am pleased to name this species for my mentor, Dr. Rolla M. Tryon, in recognition of his outstanding contribution to the understanding of the systematics and evolution of the family Cyatheaceae. 120 GERALD J. GASTONY Nephelea Tryoniana is found in pine-covered canyons, cloud forest ravines and dense, wet, mixed mountain forest in Guatemala, Nicaragua and Honduras at altitudes of 1500-2500 m. ADDITIONAL SPECIMENS EXAMINED: Guatemala. Alta Verapaz: Standley 90730 (vs). Zacapa: Steyermark 29925 (F). Honduras. Francisco Marazan: Carlson 2661, 2662 (F). Nicaragua. Matagalpa: Molina 20369 (cH). 12. Nephelea erinacea (Karst.) Tryon, Contrib. Gray Herb. 200: 40. 1970. blackish spines to 10 mm. long, occasionally with adventitious buds. Unexpanded i i 13 mm. lon apical seta to 1.0 mm. long (many to only ca. 0.25 mm.) and without lateral setae. } lon setae. Lamina coriaceous, to 3 m. long and ca. 1.6-2.0 m. wide, (rarel artially d, usually abruptly reduced to a distinct pinna-like apex : chis brown to purple-brown, adaxially minutely pubescent (occasionally sli htly furfuraceous) to glabrescent, abaxially fur- uraceous to glabrescent. Central pinnae sessile to short (to ca. 1-2 cm.) petiolulate, elliptic-lanceolate or ovate-lanceolate to narrowly so, to 104 cm. Jon d 30 cm wide; basal pinnae not highly reduced, to ca. 20-40 cm. long; pinna-rachis adaxially pubescent, abaxially furfuraceous-glabrescent to more or less pubescent with crispate a ments often free on the ba ic si the acroscopic side adnate basal segments only rarely truly sessile or free), sessile to subsessile, occasionally short (to ca. 1 mm.) petiolulate, narro e to n triangular, apex < rs} = Ba I rom a = 4 S ® :F o § QO oO poe) wn 5 = < oO 3 5 ie) o. =) 3 = wn crenulate, apex entire to subentire, margins not revolute to nee so; costule axi i istally, abaxially bo ce obscure, ad y glabrous or with occasional minute whitish stellate squamules (ca ter ) ar the segment margins, abaxially wi rm texture, persistent, frequently ruptured at maturity into two to four lobes, glabrous to densely lepidote; receptacle globose to columnar, inserted to slightly exserted, with mostly inconsp: ( % THE FERN GENUS NEPHELEA 121 This is a widespread species but the variability of its critical characters is slight making it easy to recognize. It is readily separated from Nephelea cuspidata, with which it has been confused most frequently, by its cori- aceous lamina, by its occurance at higher altitudes, by the stellate squam- ules found abaxially on the veins (Fig. 67), by the pinna-rachis being non-alate between the more distal sessile pinnules, and especially by the presence of scales with only one dark apical seta and no lateral setae (Fig. 66) on the abaxial surface of the petiole base. KEY TO VARIETIES Indusia a oe to sparsely and obscurely lepidote. Costa Rica, Panama, Venezuela, ia, Ecuador, Peru, Bolivia. inacea Luts ciel and persistently lepidote, the scales often with dark setae. RouadT: 2 Cee a ye ae ee 12b. var. purpurascens. 12a. Nephelea erinacea var. erinacea Fics. 64-67, Map 15 Cyathea erinacea Karst. Linnaea 28: 453. 1857. type: Venezuela, Merida: crescit ad pedem grainy a eS. Meridensi, alt. 2000 m., Karsten, probably LE or w as ot seen); i ! in 16425. re at eco Pome niet Christ, Bu il. Herb Rs eage II, ir 948. 1904.“LecrotypeE: Pittier 2413 or Cartago, Rio Navarrit nl in herb. ore Sapper? tae ny! oe ex B, P! S S: ee a nk ro), Wercklé s.n. (specim unce jertain); eaves Tonduz 13332 cu!, M!, Ny!, ny! fragment ex B, us!; san 16496 us! fragment ex herb. Rosenstoc Nephelea aureonitens ( Christ ) Tryon, gree Gray Herb. 200:40. 1970. Cyathea cuspidata shek rigida Rosenst. Meded. Rijks Herb. Leiden 19: 6. 1913. TYPE: Bolivia, Espritu Santo, in silvis 8 montanis, 1600 m., Jun. 1911, Herzog 2237, L (not seen), fragment us!; isotypes: M!, s!, uc!, us!. Stem to 12 m. tall and 15 cm. in diameter, bearing blackish spines fo 10 mm. long, occasionally with adventitious buds. Unexpanded croziers with blackish spines to 13 mm. long and tan t nes scales with a dark apical seta to 1.0 mm hone m a dark brown. Central pinnae sy ae to 104 cm. long and 30 cm. wide; pinna-rachis with often swollen and conspicuous lateral pneumatodes at costa axils. Pinnules sessile to sub-sessile, narrowly ovate to narrowly triangular, the pte ; an j ie ; their several dark setae. Segments with lateral margins entire to crenulate and slightly or not revolute. Indusium glabrous to sparsely and Paige lepidote, the squamules more or less stellate and similar to those on the veins In the protologue of Cyathea aureonitens, Christ cited at least five collections: Wercklé 229 and s.n., Alfaro 16496 at Capelladas, Tonduz 13332 at Las Vueltas Tucurrique, and Pittier 2413 at Rio Navarrito 1400 m. I have seen Christ herbarium material of Pittier 2413(Br) and of two sheets of Wercklé without number in 1903 and 1904 respectively (Pp). The latter two are possible syntypes, but it cannot be determined if these are 122 GERALD J. GASTONY Tonduz 13332 and 1332 (see discussion below under Cyathea basilaris). I have chosen Pittier 2413 as lectotype because I have seen the sheet in the Christ herbarium and because of the confusion associated with the other collections. s in several other species of Nephelea, the pinnules often disarticulate from the pinna-rachis. The leaves are often borne in apparent whorls of four or five (the former giving the stem a squarish cross section). Variety erinacea occurs in the mountains of Costa Rica, Panama, Venezuela, Colombia, Ecuador, Peru, and Bolivia in wet ravines and on sparsely to heavily wooded slopes in cloud forest or rain forest at altitudes of ca. 700-2500 m. (predominantly above 1400 m.). ADDITIONAL SPECIMENS EXAMINED: Costa Rica. asec 1099 (us); Kupper 1471 (m). Alajuela: Scamman 7581 (cu,us); White & Lucansky 196825 (puKE,cH). Guanacaste: Brener 15546 (¥). Heredia: Chrysler & Roever 5613 (mo,uc); Nisman 48, 109 (cH). San José: Brade 94 (uc), Brade under Rosenstock 108 (A,M,MO,UC); Gastony & Gastony 753, ig pages 767 (cu); Nisman 23 (cu), 42 (cu Sheet 1); Scamman sod sha GH,NY,UC,US); Sca n & Holdridge 7856, 7858 (cH); nts 1384 (uc); Wercklé, b. Jim. no. 562, 583 sae SY. Herb. Nac. Costa Rica no. 16416 (Ny,us); White & epee 1968121, 1968127 (puKE,cH). Cartago: Alfaro 8072 (cH,Ny,us); Nisman & Tryon 7039 (cu); Wercklé, Herb. Nac. Costa Rica no. 16750 (us); White Lucansky 196833, 196834, 196835, 196836, 196858, 196861, 196892, oye 196895, 196898, 1968111, 1968148, lian 1968171, 1968192 ee GH); O. Williams 19787 (us). Puntar arenas: Nisman 147, 158 (cH). Panama. Chiriqui: jacaastal iReye (Mmo,uc,us), 1228 (u 5), 1229 ae uc); Killip 5072 (cu, salty Maxon 5575 (GH,Ny,us). Venezuela. Merida: White ¢& Lucansky 1969244a, 1969245 (puKE,cH). Colombia. Lindig. 283 (cH). Magdalena: H. H. Smith 2222 GHNY). Santander: Killip ¢ Smith : Daniel 1214 Amértegui 285 (us); Garcia-Barriga 12560, 12567 (us); Little dr Little 8574 (GH, us). Cauca: Dryander 2903 (¥ pro exe 2907 (¥F). Ecuador. Napo-Pastaza: Sydow 895 (BM,us). Pichincha: Solis 5822 (F). Tungurahua: Prescott ¢ Wiggins 20 (us). Santiago-Zamora: Dodson & Thien 1443 (Mo,us). El Oro: i gg A 53766 (F, us). Peru. Amazonas: Hutchison & Wright 6803 (cu,uc). Huanico: Macbride 4843 (F, sa Bolivia. Yungas: Rusby 121 (us). Cochabamba: Seembash. 9478 (¥,GH,MO, UC,US 12b. Nephelea erinacea var. purpurascens (Sod. ) Gastony, comb. et stat. nov. Fics. 68, 69, Map 16 Cy athea purpurascens Sod. Crypt. Vasc. Quit. 503. 1893. type: Ecuador, Imbabura: crece en los — occidentales del volcan Cotacachi cerca de hacienda ee ., Sodiro s. n. A sheet at p! leg. Sodiro 2/1874 with duplicates ex herb. P a (two ses sae mol, and a sheet at ote leg. 4/1874 are all authentic from “Cot Sate prope Quisaya” and may be type material. Before a lectotype sheet is chosen, the Sodiro collection mela at Q chould be consulte Nephelea purpurascens és od.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Cyathea oxyacantha Sod. Sert. FI. Ecuad. II, 6. 1908. type: Ecuador, Pichin crescit in silvis suband. vulc. Atacatzo, Sodiro s.n., possible holotype p! (leg. Sodiro 3/1906 ); isotype: us! ex Rosenstock. THE FERN GENUS NEPHELEA 123 Stem (ex descr.) to 2-3 m, tall and 12-15 cm. in diameter, bearing sod ig of unknown len ngth, adventitious buds not known. Unexpanded croziers Hides purpurascens is known only from the adjacent provinces of Imbabura and Pichincha in Ecuador where it occurs in wet forests on the slopes of a few volcanos at altitudes of (1900-) 2600-2800 m. It has apparently been derived from variety erinacea or their common ancestral form in the high altitude isolation of these Ecuadorean volcanos. ADDITIONAL SPECIME Pichincha: Mille 170 (us), s. n. (Ny,uc,us); 1900, Saltire re gs 1907, Fs rag ee P,Us pro parte). 13. Nephelea Imrayana (Hook. ) Tryon, Contrib. Gray Herb. 200: 40. 1970. o 10 m. tall and 45 cm. in diameter (including the investing sheath of adiveutisiecs roots), bearing light tan to brown scales especially toward the apex and blackish spines to 26 mm. ak ga y ui | long, adventitious buds not known Petiole to ca. 0.51-1.2 m. long, brown with blacki ish s spine to ca. 14 mm. (us usually less ) long, the scales on the abaxial surface Ret af pers inns dense and felt-like, ally with a an ; pinnatifid, abruptly reduced to a distinct Sicantile apex (to ca. 40 c cm. . te ng). Rachis brown to dark brown, adaxially squamose-pubescent to glabrescent, abaxiall furturaceo often leaving a scabrous surface. Centr innae sessile to short (to ca. 1.5 cm.) d a petiolulate, ovate-lanceolate, to 90 cm. long and 27 cm. wide; basal pinnae not highly uced, to ca. em. long; pinna-rachis adaxially more or less pu nt and axiall glabrescent distally, with mostly inconspicuous as esac at costa axils, not sessile, narrowly ovate to narro the apex acute to acuminate; costa a y pubescent, ab more or am s densely furfura aa game to pubescent istally, the scales to 2-3 er ones thout a brown central hae to brown and th i g dea single) dark apical sta, oun —— in Segments sub- _— to falcate, with later: _ rarely crenate, (bas ~ often auriculate on the ‘aniarciele side with ic margin or less crenate), apex subentire to crenulate, mar slightly or not revolute; candi adaxially glabrous or rarely with one to several stiff 124 GERALD J. GASTONY trichomes distally, abaxially distally glabrescent or ye stiff trichomes, proximally more or less squamose uamules to ca. 0.5 mm. long bearing severa prominent i for i a the eking of the veins; indusium preter, with or ak an umbo, oe to light brown, of mor less delicate texture, mostly persistent but irregularly Te at maturity, glabrous; receptacle globose to short-columnar, inserted, wit short (to ca. 0.2 mm, lon ng) paraphyses occasionally tufted at the apex of the receptacle. Maps a 13, Nephelea setosa; 14, N. T. ryoniana; 15, erinacea var. cea; 16, N. erimacea var. purpurascens; 17, N. Imrayana var. Imrayana; 18, x Imrayana var. eden 19, N. cuspidata. THE FERN GENUS NEPHELEA 125 In addition to the characters mentioned in the key, the dullness of the abaxial lamina surface and the lack of whitish stellate trichomoid squam- ules at or near the segment margins (such as chracterize Nephelea erinacea ), serve to distinguish this species. In plants from Guadeloupe to St. Lucia, the rachis, pinna-rachis, and costa are abaxially more or less densely and persistently furfuraceous- squamose, whereas in plants from St. Vincent, Trinidad, and the continent the costule is less densely squamose abaxially and the larger scales on the abaxial surface of the rachis and pinna-rachis are not as persistent. In the specimens from St. Vincent southward, however, there usually are clear scars and epidermal emergences where the larger deciduous scales were once attached. In some specimens from Trinidad (e.g., Broadway 9969, Britton et al. 2287), Venezuela (Wurdack 34117), Colombia (Grant 10879), and Panama (Maxon 5670), the pinna-rachis is rather persistently furfur- aceous-squamose abaxially, although perhaps not always as densely so as in material from Guadeloupe to St. Lucia. The number of veins in an ultimate segment is variable. However, the variation does not have a significant correlation with geography. Through- out the Lesser Antilles and into Venezuela (Wurdack 34117; Aristeguieta & Pannier 1927) and Colombia (Grant 10879) the specimens show tri- chomes abaxially on the costule, although these are absent or poorly developed in other material from Venezuela (Steyermark 58989) and in that from Panama (Maxon 5670) and Costa Rica. From Guadeloupe to the Paria peninsula of Venezuela the squamules borne more or less distally on the abaxial costule surface have a relatively small body and several prominent dark setae at the apex and sides ( Fig. 73), whereas elsewhere in Venezuela and westward into Central America these more distal squamules (or many of them) feature a conspicuously greater body area (Fig. 76) with the setae relatively less conspicuous and less numerous (many have a single seta at the apex and some lack dark setae altogther ). The lack of coordination among the several clinal morphological tendencies observed in this species does not facilitate the recognition of infra-specific taxa. The morphological divergence of the species toward each end of its range, however, warrants the recognition of two varieties which may best be distinguished as below. Their geographic separation occurs in Venezuela. KEY TO VARIETIES Abaxially on the costule the more distal squamules bearing several dark apical and lateral setae, most of the scale bearing setae (Fig. 73); conspicuous trichomes present abaxially on the costule. Guadeloupe, Dominica, Martinique, St. Lucia, St. Abaxially on the costule the more distal squamules bearing one to few dark apical setae (rarely lacking dark setae), most of the scale body lacking the setae (Fig. 76); Wh 126 GERALD J. GASTONY — trichomes absent abaxially on the costule in Central American and some uth American material. Venezuela, Colombia, Ecuador, Panama, Costa Rica. .... '13b. 13a. Nephelea Imrayana var. Imrayana Fics. 70-74, Map 17 7°& Cyathea Imrayana Hook. Sp. Fil. 1: 18, t. 9, B. 1844. tecroryre: Dominica, Couliaban Mountain, Dr. Imray, k (not seen), fragments ny! in herb. Jenman and in herb. Underwood ex x. Alsophila nigra Jenm. Bull. Misc. Inform. Roy. Bot. Gard. Trinidad 3 (15): 38. 1898, (=West Ind. and Guiana Ferns, 38), not Mart. Icon. Pl. Cr rypt. ik a 30 f. 5, t. 47. 1834. Type: locality and collector unrecorded, Trin!, phot Cyathea caribea Jenm. Bull. Misc. seg ctn Roy. Bot. Gard. Trinidad 3 ash 1898, (=West Ind. and Guiana Ferns, 57). LECTOTYPE: St. Vincent, collector aa spines to 8 mm. long and th te tan to brown ale with a dark eal seta to 0. 95-1. 0 mm. long and without Posie setae. Petiole to 0.51 m. long, with blackish spines to mm. (usually less) long. Lamina length not known, to ca. 1.6 m. wide. Rachis abaxially furfuraceous-squamose to sistiescete the larger scales oe or decidu- ie distally. Segments with latera margins ‘eutive to pie ntire; costule abaxially with stiff trichomes distally, proximally with squam diss s bearing several prominent dark pestle and lateral setae. Indusium sphaeropteroid, usually with an umbo. Hooker cited as Cyathea Imrayana collections of Dr. Imray in Dominica and Dr. Bancroft in Jamaica. Jenman explicitly limited the species to Dominica (excluding Jamaican materials) and so effectively chose the Imray Dominica collection as the lectotype (Jenman, 1898, p. 59). Cyathea Imrayana var. subnudata Hook., of Jamaica, is treated under dubious and excluded names. In the herbarium of John Smith at Bm! is an authentic early collection of Cyathea caribaea Jenm. collected on St. Vincent by Caley and bearing John Smith’s apparently unpublished epithet in the genus Cyathea honor- ing the collector. On a collection from St. Lucia (Proctor 17765), it is noted that the leaves are borne “in whorls of five, expanding simultaneously.” Similar apparently whorled leaf arrangements have been observed in Nephelea erinacea and other species in Central America. Variety Imrayana occurs in Guadeloupe, Dominica, Martinique, St. Lucia, St. Vincent, Trinidad, and the Peninsula de Paria of Venezuela at altitudes of ca. 350-980 m. on mountain slopes in mossy forest, cloud forest, and rain forest. THE FERN GENUS NEPHELEA 127 ADDITIONAL SPECIMENS EXAMINED: Guadeloupe: Bailey & Bailey 78 (us); Duss 4131 (Ny pro pate); 1895, Duss 4323 (Ny); 1897, Duss 4323 (4156) (us); Duss 435 (¥,cH,Us); Husnot 269 (F), 400 (¥,ny); L’Herminier 178 (Ny), s-.n. (¥,MO,Us ); ri 10132 Gm is Baery’ (x) Proctor 20040 (a,us); Questel 3292, 3250 ); A I 6234 (ny, uC us); Smith & Smith 962 (us), s.n. (GH). Trinidad: Britton et al. 1360 fence 2287 ( tea Gaye Broadway 6216 (¥,GH,Mo,s,us), 9969 (F,CGH,TRIN re Purdie 10 (TRIN); Seifriz 3 (us). Venezuela. Sucre: Steyermark 94812 (GH,NY,VEN 13b. Nephelea Imrayana var. basilaris (Christ) Gastony, comb. et stat. nov. Fics. 75, 76, Mar 18 ¢ Cyathea basilaris Christ, Bull. Herb. Boiss. II, 4:949. 1904. LecroryrEe: Costa Rica, Wercklé s.n. in herb. Christ presumably at Pp (not seen); possible isolectotype: ny! ex herb. Christ iter Wercklé); authentic material: 1903, Wercklé & Brune, P!, 1901-1905, Werckleé, Nephelea basilaris ( Cheri) Tryon, Contrib. Gray Herb. 200:40. 1970. Cyathea reticulata Wercklé ex Christ, Bull. Herb. Boiss. II, 5: 251. 1905. rype: Costa Rica, ect Irazi, 1800 m., Wercklé 6, Pp! in herb. Christ; isotypes: ny! fragment ex oe , ny!, Pl, usl. em to 45 cm. in diameter (including the investing _—_ of adventitious roots), Sie = blackish spines to 26 mm. (usually less) long. Unexpanded croziers with blackish cn to 17 mm. long and light tan to brown scales with a dark a In the protologue of iucahes basilaris, Christ cited at ick two collec- tions: Wercklé s.n. and Tonduz 1332, foréts de las Vueltas Tucurrique, 1000 m. In his search for types in 1969, Rolla Tryon found in Christ’s herbarium at P only two sheets under C. basilaris: Tonduz 1332 and Wercklé and Brune in 1903. The Tonduz number is probably an error for 13332, as noted by Maxon on vs sheet 74301. Because Tonduz 13332 is C. aureonitens Christ, I follow Maxon and take the cited Wercklé collec- tion as the lectotype of C. basilaris Christ. Unfortunately this sheet was not found in Christ's herbarium at p. The sheet in Christ’s herbarium at p (Wercklé and Brune in 1903), however, is C. basilaris Christ and may be taken as authentic material. A specimen with a stem diameter of 45 cm. (Wurdack 34117) from Estado Bolivar, Venezuela, represents the largest stem diameter recorded in the genus. 128 GERALD J. GASTONY , 64 j hj Lf / hifi, WN /) ij YY Pt AUG, Tan, ty \\ a Tare , \ VAN \ Wy PAN SUNS AAANRAA AN MOYO y UU KANN SIH MM ae AW: LUCAS Is) oe Gl iN M = Ny. AN THE FERN GENUS NEPHELEA 129 Variety basilaris is known from Venezuela, Colombia, Ecuador, Panama, and Costa Rica where it occurs at altitudes of ca. 1000-2 m. in wet forests on ridges and mountain slopes. eeahavees SPECIMENS EXAMINED: Nene: Bolivar: Steyermark 58989 (cu, MO.NY,US); Wurdack 34117 eo ny). Yaracuy: Aristeguieta & Pannier (us,vEN Colombia. Magdalena: Grant 10879 went thesis Napo-Pastaza: Grubb et al. 1334 (us). Panama. Chiriqui: Maxon 5670 (us 14. Nephelea cuspidata (Kze.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 77-80, Map 19 Cyathea cuspidata Kze. Linnaea 9: 1834. TypPE: Peru, prov. Maynas, stein in be paludosis, Febr., 1831, Poeppig sa 2286, igen destroyed at LZ; isotype: B!, Ny! fragments ex B and K, pl, us! ey a eB Mi "Cyathea oyapoka Jenm. Bull. Misc. Inform. Gard. Trinidad 3 (15): 58. 1898, (=West and Guiana s, 58). sng ayenne, French Guiana, in sylvis humidis ad torrentes Guyana Cera ole the 8 fe arigol Junio, 1833, Leprieur, Ny in herb. Jenman (not seen), photo Gu!; isotypes: F !, ny!, Pp! in herb. Cosson, vus!, us! — ex herb. Br. Guiana. Cyat. a punctifera Christ, in Pittier, Primit. Fl. Costaric. 3: 40. . TYPE: Costa ies fordts de Tuis, 650 m., Nov. 1897, Tonduz 11307, Br (2 iets det. Christ )!; es: cul, Ny! fragments ex BR ed us, P! (two gg annotated as n. sp. Christ ve not herb. Christ), us! six sheets, photo us! of BR shee Cyathea Hassleriana Christ, Bull. Herb. Boiss. I, 7: 936. 1907. Type: Paraguay, La Cordillera: in silvis humidis pr. Caacupe, mens Sept., Hassler 120, presumably at P (not seen); gl s-pa! gee in herb. Selim ain ex Rosenstock. em to cm. oe pte cota te an to brown scales uk ard the apex ee blackish spines to 1 cm. long, occasio nally with A seatiliods als: Udapended and t erous smaller apical and lateral setae. Petiole to 0.5 m. long, brown, with blackish spines to 9 mm. long, the scales on the a surface copious, tan to brown = with or without a darkened central stripe, with a dark apical seta to 0.5-1.0 mm. long ore with usually numerous sm maller apical — lateral setae. Lamina apyraceous, wry pinnatiid, abruptly ited to a distinct Sane tike apex (to 20-60 cm. long). Rachis occasionally blackish-brown), adaxially minutely furfuraceous- pu ab nt to pi leering abaxially — furfuraceous to never Central pinnae sessile Fics. 64-67. Ni , Nisman 152 (cH), X 0.5; 65, Central portion io a annane as je ala in 64, Macbride 4843 (Fr), a 2: 66, Distal portion of a scale from the abaxial petiole surface, structurally marginate with a single dark apical seta and without lateral setae, Gaston Gastony 753 (cH), x 43; 67, resentative stellate ar tomgesey ne = 7 vein _ Surface, with irregularly disposed processes, Maxon 5575 (GH), sg cee ephelea erinacea var. - purpurascens : 68, Central pinnules of a central in 68 from n gia), goo 70-74. Nephelea Dans var. Imrayana: 70, Ce pinnules of a central pinna, Chambers 2692 (us), X 0.5; 71, Adaxial view of basal portions of pinnules, with basal segments and auriculate on the basiscopic side, as sale ie Central portion of a pinnule as shown in 70, print 1669 (cH), X 2; 73, A squamule from axial costule surface, with a relatively small body and several prominent dark setae, Morton oie nay x 85; 74, A piasecuenbative unruptured sphaeropteroid indusium with an sips umbo, Wilbur et al. 8209 (Nx), X 15. Fics. 75, 76. Nephelea Imrayana var. basilaris: 75, Central pinnules of a central pinna, Steyermark 58989 (cH), x 0.5; 76, A squamule from the abaxial apg surface, with a conspicuously large body area relatively inconspicuous dark setae, as in 75, 130 GERALD J. GASTONY to short (to ca. 2 cm.) petiolulate, ovate-lanceolate, to 95 cm. long and 30 cm. wide; asal pinnae not highl reduced, to ca. 1 side) greenish-alate between the more distal sessile pinnules, Pinnules to 15 cm, long and 1.5-3.5 cm. wide, deeply pinnatifid (the basal segments often free on the basiscopic side with the acroscopic side adnate; basal segments only rarely truly frequently rupturing at maturity into 2-4 lobes, bearing stellate indument similar to that on the veins or glabrescent; receptacle globose, inserted, with short (to ca. 0.1-0.2 mm. long) mostly quite inconspicuous paraphyses. Although Maxon (1930) correctly associated Leprieur’s number 267 with the type collection of Cyathea oyapoka, “in sylvis humidis ad tor- rentes Guayanae centralis Oyapok superior, Junio, 1833,” there is at least one other collection of this taxon which bears the Leprieur number 267, viz., “in sylvis paludosis ad basin montium via Kaw, Guyane frangaise, Leprieur 1847.” The same number was evidently used for other collections of Cyathea oyapoka. Consequently the several sheets at Fl, Ny!, and vs! which bear the number 267 without locality and date cannot unequivocal- ly be designated isotypes. The sheets I have cited as isotypes are from the Leprieur herbarium and bear the locality and date of the type collection. Several of these and other sheets of this taxon collected by Leprieur are annotated with an apparently unpublished epithet in the genus Cyathea honoring the collector and attributed to Baker. nata,” for a collection (Brade 628) from Costa Rica are both apparently unpublished. Although it exhibits the greatest species range in the genus, Nephelea cuspidata is quite uniform in its critical characters and is easily recognized. Its most diagnostic features are the petiole and crozier scales which have several dark apical and lateral setae in addition to the main apical seta which is universal in the genus, the stellate trichomes found abaxially on the veins (Fig. 80), and the persistent, brown, firm, usually lobed and THE FERN GENUS NEPHELEA 131 often stellate-pubescent indusium. Its separation from N. erinacea, with which it has most frequently been confused, is facilitated by the first two of these characters. These features have been discussed under that species. It is adequately separated in the key from its closest relative, N. stern- bergii. The apparent disjunction in the range of Nephelea cuspidata between Colombia and French Guiana may be genuine inasmuch as the interven- ing areas of Venezuela, British Guiana, and Surinam have provided a number of other tree fern collections without yielding this species. Most of these collections, however, are from the more coastal regions of these countries. Because it is a lowland and gallery forest species and does occur on the eastern side of the Andes, its presence in Amazonas, Brazil, near the Colombian and Peruvian borders suggests that it may yet be found in the little collected southern reaches of Venezuela, British Guiana, and Surinam and in northern Brazil along the eastern and northern edges of the Amazon basin. If it does inhabit such an arc, its range may connect to Leprieur’s collection sites via the Oyapok River. Nephelea cuspidata is known from Nicaragua, Costa Rica, Panama, Colombia, French Guiana, Ecuador, Amazonas Brazil, Peru, Bolivia, and Paraguay. It commonly occurs along river banks, in ravines, and in more or less dense wet rain forest or cloud forest from sea level to 1200 m. It is replaced at higher altitudes by other species, such as N. erinacea. NAL SPECIMENS EXAMINED: Nicaragua. Rio San Juan: Bunting & Licht 890 isman 7 teas. Heredia: Burger e Stolze sev op rose Nisman 51, 1 cH); Scam oldrid H,US imon: Nisman 134 (c Scamman 7003 (cu,us), 7010 (cuH,uc,us); J. D. Smith 5074 (cu,us). San José: Brade 94 (ny). Panama. Canal Zone: Killip 2921 (us); Shattuck 1145 (Fr); Wetmore é& Woodworth 116 (cH). Darien: Schott 34 (¥,Mo,us); Stern et al. 434 (GH,Mo,uc, us). Colombia. Norte de Santander: Cuatrecasas 12978B (F,us). oni Haught 2001 (cH,NyY,uc,us), 4617 (Ny,us). Choco: Taylor 1242 (us). Cundinam ig 195 (cH). Meta: oa aerocurdl "4628 (¥,us), Philipson & Mieke 1801 : (es) Schudltet 922 (a,cu,us). Vall as 13754, 15110 (F,cH,Us), 14388, 16157 (F,GH,NyY,Us), 14843, 15222, 16004, - poten 1 "17662 (F,us). Narino: Schultes & Cabrera 19105 ( us). Ca & Juzepezuk 6419 (us). Amazonas: Schultes 6966 (cH). gir Duarte (up). French Guiana. Leprieur 137 (ny), 190 (¥,cGH,us), 1830, Leprieur 197 (cH), 1847, Leprieur 197 (F,us), 1847, Leprieur 197/267 (F), pecans 224 (us), 267 (F,Ny), 1847, Leprieur 267 (Ny). Ecuador. eraldas: Sparre pie (cH). Imbabura: Solis 12784 (F). Napo-Pastaza: Grubb et al. 1551 (Ny,us). Cotopaxi: Sparre 17196 (cH). Los Rios: Sparre 17900 (cH). Peru. 1829, Poeppig (st). Loreto: Mexia 6197 (¥,GH,MO,NY,UC,Us); Ll. Williams 3288 lard 2 : Killi ( F,GH,NY,US in: Al 0618 (cH); Ferreyra 437! i illip & Smi 86 (F,NY,Us). Cuzco Med age 11703, 14591, 15572, 16117 (cH). Ayacucho hg ¢+ Smith 22654 (cH,us : Vargas 16132, 18862, 18919 (cH). Boli (F,CGH,MO,NY,uS), 10841 (F,cH,MO,NY,uUC,US); R. §. Williams 1335 (cu,us). Cochabamba: Buchtien 2195 (¥,cH,us ); Cardenas 2031 (cu, us). Santa Cruz: Steinbach 16416 (cH). Paraguay. La Cordillera: Balansa 2861 (P pro parte). 132 GERALD J. GASTONY 15. Nephelea Sternbergii (Sternb.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Ste titious ‘ei bearing tan to brown scales cue toward the a wad: blackish spines to 11 mm. long, adventitious buds reported. Unexpanded croziers Pie blackish spines to 10 mm. long and mostly light tan scales with a dark apical s o 1.0 mm. ong, brown with blackish spines to 11 mm. long, the scales on ea axi a oe copious, tan, with or without a darker central stripe, with a dark Sapte to se mm. long and with or without additional smaller apical and lateral pire apyraceous to ca. 3.0-3.5 m. long and 1.4 m. wide, cet porally tpinate) as Seaicas parva ae cane ails to aprons (to ca. 1.5 cm.) petiolu late, ovate-lanceolate, to ca. 75 cm. long and 23 cm. wide; basal pinnae not highly reduced, 2 ca. 23-32 cm. long; inna ichis adaxially pubescent, abaxially furfuraceous- squamose or, when n glabrescent, was cabrous to spinose projections where the larger ca. 2 mm. ong, eg usually without a darkened centra with several dark apical and lateral setae, often clustered in the axil. eae abies to aloes with lateral margins entire to subentire, apex entire to crenulate, margins s ightly or not ff tri ; revolute; costule adaxiall frequently with one to f chomes distally, abaxially th dark-setate squamules similar to those of the costa but smaller, often with whitish omoid sq es, occasionally with sparse stiff trichomes distally; veins once-forked ( more) at or near the Aang or simple, evident on both surfaces, adaxially glabrous or ira ccasional, minute, highly aye fone ae peci: argins, sastalls | frequently wi a- similar receptacle globose, inserted to vey exserted, Bi short (to ca. 0.1-0.2 mm. long) mostly inconspicuous paraphyses The close relationship of this species to Nephelea cuspidata is espe- cially evident because of its indument of stellate trichomes or trichomoid squamules found abaxially on the veins. It is distinct from N. cuspidata in detailed aspects of this same indument in the nature of its indusia, and in its more variable petiole scales. Specimens which are most like N. cuspi- data in the color, texture, and persistence of the indusium are least like that species in their highly conspicuous whitish vein and indusial indu- one (Fig. 84) and in their petiole scales which generally feature a single dark seta. Those most like N. cuspidata in their stellate trichomoid indu- THE FERN GENUS NEPHELEA 133 ment and the condition of the setae on the petiole scales feature delicate and mostly fugacious indusia. Both species reach eastern Paraguay, but this is not the center of a cline between the two species. While those specimens of N. Sternbergii with indusial color and texture most like that of N. cuspidata occur more to the west in Brazil, they are found in var. Sternbergii which has the arms of the vein indument tending to be more numerous and irregularly disposed than in N. cuspidata and the petiole scales have a single apical seta and none laterally. The presence of several apical and lateral setae characterizes the petiole scales of N. cuspidata even in Paraguay. KEY TO VARIETIES Petiole eid Bevaix bearing light tan scales which usually lack a darkened central stri lateral setae, and more than one dark apical seta; costule abaxially with eee hoi whitish highly ER stellate eanteedies in addition to the dar setate squamules; abaxially the v ae stellate trichomes or highly trichomoid stellate sqhainiles € o ca. 0.5-0. n diameter with long whitish irregular arm (Fig. 84), often especially evident near the segment margins; indusium owe, more or less firm, cyathiform to urceolate, occasionally ob spleen, intact hoget gate costule and with spars less inconspicuous stellate trichomes or highly trichomoid squamules to ca. 0.2-0.3 mm. diameter; indusium diaphanous to tan or hs aad own, delicate, sphaeropteroid to sub-sphaeropteroid rupturing irregularly a cinturity and persistent to usuall more or less fugacious, glabrescent to ay stellate pubescent. Brazil. = eer ey eee See ee es 15 b. var. acanthomelas. 15a. Nephelea Sternbergii var. Sternbergii Fics. 81-84. Map 20 Cyathea oe Pohl ex Sternb. Fl. der Vorwelt 1:47, t.C. 1820. (Essai sur la Fl. du Monde Prim. 4:52, t. C. 1826.) type: habitat in Brasiliae Capitania Goyaz a Limoero non feline St. Lido, _ prc! or perhaps w; isotypes: BR (two sheets, not seen), BM! fragment and photos imens at BR, pRc!. Not Pathan: Sternbergii Domin, Pterid, 263. 1929, nom. nov. si Alsophila elegans Mart., not Cyathea elegans w. Cyathea Caesariana Christ, in Wettstein, Denkschr. Akad. Wien 79:18, t. 2, f. 2, t. 8, f. 5-6. (1908); as a te p- 12. (1906) 1907, fide Hieron. in Beiblatt zur Hedwigia 46(3):116-117. 1907. LecroryPe: Brazil, Séo Paulo: in tu urbis iro Cesar, ca. 500 m.s.m., VII, 1901, Wettstein & Schiffner, r! (nok heat Christ, Sco fous rhaps only an isotype); syntypes (not seen): In sylvaticis inter Apiahy et Yporan 2. 900-400 m. , VII 1901, Wettstein et Schiffner; Ad ripas fluminis Rio Branco — Conceicao de Ttahaen, 20-100 m ., VII 1909, Wettstein et Schi emitelia Caesariana (Christ ) Samp. Bol. “age Nac. Rio —_ 1: 19. sr Coahes Rojasii Christ, Fedde Repert. 6:348. 1901. Lecroryre: Paraguay bay: Sierra de Amambay, ad ripas rivulorum pr. Punta Porda, Date 10414 ayy h and thougl should also be included in ape holotype because it includes the petiole as bailiee GERALD J. GASTONY 134 79 MZ Mygare Ze axante ~ YL AG Ht ZS “ scale from the abaxial , Asplund 16330 (s), pinnules of a central pinna. x 2; 79, Distal portion of a ero > » Central 78, Central portion of a pinnule in 77 petiole surface, structurally marginate with Fics. 77-80. Nephelea cuspidata: 77 «OS; of a pinnule in 85, with unruptured sp shown in 85, but more fertile, Brade 42967 iv), THE FERN GENUS NEPHELEA 135 Cyathea schizolepis Copel. Univ. Calif. Publ. Bot. 17:27, t. 3. 1932. TypE: Brazil, inas Gerais, Vigosa, Fazenda de Aguada, in dense forest, alt. 725 m., Mexia 5059, UC 466087—-466090!; isotypes: F!, cH! (two sheets), Mo! ny!, us! (three sheets ). Stem diameter not known, bearing mostly light tan scales especially toward the apex and blackish spines to 11 mm. long Unexpan central stripe, with a dark apical seta to 1.0-1.5 mm long and usually without additional apical and lateral setae. ina bipinnate-pinnatifid. Basal pinnae not highly reduced, oc . long. Pinnules with the segments usually fr the basiscopic Ob aie or less firm texture, usually cyathiform to urceolate, occasionally acephanr teroid, intact to more or less rupturing at maturity, persistent, bearing conspicuous whitish stellate indument similar to that on the veins. Dr. Riedl at the Naturhistorisches Museum, Vienna, has informed me that the type of Cyathea Sternbergii cannot now be found there among Pohl'’s specimens. Likewise, Dr. Sojak at the Botanical Department of the National Museum in Prague, which holds the Sternberg Herbarium, indi- cated that the type material is not there. Dr. Sojak referred me to the herbarium of Charles University in Prague where a Pohl collection was ound. This variety generally occurs more inland than variety acanthomelas. In their most characteristic condition, the petiole scales of variety Stern- bergii have a single dark seta and are more or less golden tan (due to lack of a darkened central stripe) and the indusia are brown, firm, per- sistent, and conspicuously stellate pubescent. Variety Sternbergii occurs in Brazil in the states of Minas Gerais, Sao Paulo, and Goias and in Paraguay in the Departments of Amambay (Sierra de Amambay) and La Cordillera at altitudes of ca. 500-690 m. in forest and along rivers in gallery forest. ADDITIONAL SPECIMENS EXAMINED: Brazil. Clausen 257 (us). Goias: Dawson 14968 (us); Ule 526 (Rn). Minas Gerais: Mexia 4650 (¥F,GH,MO,NyY,R,UC,Us); Mosén 2037 (s), 2038 (ns); Santos & Castellanos 24124 (HB). Sao Paulo: Burchell 5339 (cx); 1913, Edwalle (sp); Regnell 1479 (¥,Mo,us). Paraguay. La Cordillera: Balansa 2861 (¥,P pro parte). 15b. Nephelea Sternbergii var. acanthomelas (Fée) Gastony, comb. et stat. nov. Fics. 85-87, Map 21 Cyathea acanthomelas Fée, Crypt. Vasc. Brésil 1:177, t. 64, f. 1. 1869. TyPE: in ry Brasilia fluminensi, Glaziou 379, p!; isotype P! in herb. Glaziou; isotype portion: us! ex herb. Rosenst. 136 GERALD J. GASTONY Nephelea Sternbergii — acanthomelas (Fée) Brade, Bradea 1 (10): 74. 1971. Cyathea Sampaioana Brade et Rosenst. Bol. Mus. Nac. Rio 7:137, t. 1, 4. 1931 TYPE: habitat in cea ae Rio de Janeiro, prope Teresopolim (Corrego Bei ija- Flor), Pi m., 7 X 1929, A. C. Brade 9601 (erroneously cited in the protolo ogre as 9631), . Mus. Nac. no. 20616, r! consisting of six sheets; isotypes s!, Nv, ucl. Nephelea Sternbergii forma Sampaioana (Brade et Rosenst.) Brade, Bradea 1 (10): 74. m. diameter, bearing tan to brown scales especially toward the apex heey blackish spines (length not known), adventitious buds reported on one specimen. Unexpanded croziers not known. Petiole with scales on the abaxial surface tan, often a additional smaller apical and ayer setae. Lamina (rarely sarily tepneoiie) pre- dominantly ee . Basal pinnae not highly reduced, to ca. 23 cm. lon Pinnules with the basa segments sata free on the basiscopic side, the acroscopic e wichomnotd sori (to ca. '0.2-0.3 mm. diameter). Indusium oa ea to tan or light brown, of delicate texture, en as to sub-sphaeropteroid, rupturing irregularly at maturity and Lge sistent to usually more or less fugacious, glabrescent to obscurely stellate pubesce This variety is ally more coastal than var. Sternbergii. In their most characteristic condition the scales borne abaxially on the petiole base have several dark apical and lateral setae and the indusia are diaphanous to tan, delicate, fugacious or partially so, and glabrescent or only obscurely stellate pubescent. Minute dark-setate squamules are generally more common abaxially on 1 the a dee and veins in this variety than in var. Sternbergii and the trich 1amules on the veins are less conspicuous. The minute one- or rected ‘icone which are commonly (but usually obscurely) present on the abaxial surface of fern leaves are cer- tainly present in both varieties of this species but are frequently more conspicuous (with a tan to claret cast) abaxially on the veins of var. acanthomelas. The scabrous abaxial surface of the pinna-rachis is occasionally more pronounced in var. acanthomelas than in the preceding one. This is espe- cially shown in some specimens from Parana (e.g., Dusén 10346) where these scabrous projections may be seen as small squaminate spines. Brade (1971) has recently concluded that Cyathea acanthomelas and C. Sampaioana are distinct only at the rank of forma. I am unable to assign even that degree of taxonomic recognition to them. In the same paper, Brade pointed out what appeared to be vegetative reproduction by stolons in a specimen of C. Sampaioana and discussed the presence of adventitious buds at the base of the stem in one individual. Although adventitious buds are known in other species of Nephelea, reproduction by stolons does not appear to have been reported elsewhere in the genus. The phenomenon reported by Brade may be similar to the stolon reproduction described by Halle (1966) in Alsophila Manniana (Hook.) Tryon (as Cyathea Man- niana Hook. ) in Africa. THE FERN GENUS NEPHELEA 137 Variety acanthomelas occurs in Brazil in the states of Bahia, Rio de Janeiro, Sao Paulo, Parand, and Santa Catarina at altitudes of ca. 50-1200 m. POS rivers and in rain ree especially along the Serra do Mar. DDITIONAL SPECIMENS EXAMINED: Brazil. Glaziou 2379 (ny); Luederwaldt SP 22027 res Bahia: renee herb. Luerssen 6145 (uc), Blanchet 2489 (mo ,NY,us), 2506 (¥,K,NY); Pohl 1 p). Rio de Janeiro: Brade 16582 (cu, MO,NY,S,US), 16642 “ss 3096 (P,R,s). Parana: Dusén 7 724a ( GH,MO), 6734 io GH,MO), 6 H,MO NY ,8,US), 10112 (cu,s), 10346 (Gu,Mo,Ny,s,uc,Us), 12037 (cH); Lye aes 11283 (HB). Santa Catarina: Schmaltz 47 (r, MO,NY,P,S,UC 16. Nephelea incana ( Karst.) Gastony, comb. nov. Fics. 88-90, Map 22 Cyathea incana Karst. Fl. Columbiae 1:75, t. 37. 1860. TypE: Columbia, Cundina- marca: crescit in silvis frondosis andium Bogotensium, altitud. 2500 m., Lindig; authentic material (Villeta, 1900 m.) Le! and p!, a sheet at s! and one at w (3M! photo and fragment ex w) are probably the same as these, but are labeled “1 leg. Karsten.” Cyathea sprucei Hook. Syn. Fil. 20. 1865. Lectotype: Ecuador, Montafia de Canelos, alt. 4500 ft., Tungurahua, alt. 6500 ft., Maio 1858, Spruce 5428 x! a collec- tion consisting of two erp isotypes: Ny! two sheets and fragment ex x, P!. Spruce 5744 pl is ie a type; it is Nephelea mexicana (Schlect. & Cham.) Tryon, (K sheet not see Cyathea canescens Sod. . Fl. Ecuad. II, 4. 1908. “Ecuador, Pichincha: cresit in silvis suband. et er Te ichincha et Atacatzo.” LECTOTYPE: 7/907 Sodiro in silv. suband. vulc. Atacatzo (ut C. incanae Karst. affinis, tamen diversa videtur) P!; pan etbness mo!, uc!, us!. The protologue af tiers usly based on more complete material, ae this fits the description and reflects Sodiro’s remarks in the rot es as Loe its affinity with C. incana Karst. Seana vule. Pinchincha, Sodiro Pp! (tw shee Neplieded canescens (Sod.) Tryon, Contrib. = Herb. 200: 40. 1970. camer O’D agree iana Alston, Lilloa 30:108, t. 1-2. 1960. ee Argentina, ta: . és, 1000 m., Willink 285° ‘pol; grees aia SYNTYPE; 24 May 1949, “O'Donell from Oran, cultivated at Instituto M. Lillo, ey wn i ed croziers with blackish st to 8 mm. lon tan ae with a usually solitary dark apical seta to 1-1.5 mm, long fps en) aa without lateral setae. Petiole to 0.15-0.35 m. oe (to the highly reduced basal pinnae), light brown to brown with blackish spines to 4 mm. long, the scales on the abaxial surface copious, my light tan, some with a darker central stripe, with a dark apical seta 1 m. long and without lateral setae (or an additional apical or Daphne seta only rarely). Lamina rigidly papyraceous, len not known, to ca. 1 m. wide, bipinnate-pinnatifid, abruptly reduced to a distinct — apex oe ca. 20 cm. long). Rachis light brown, roneny furfuraceous-pubescent, abaxially furfuraceous- glabrescent. Central pinnae short (to ca. 1 cm. ~ peti egy ‘narrowly e elliptic-lan late to narrowly ovate-lanceolate, to ca. 60 cm. long and 16 (usually less) cm. a pneumatodes at costa axils not evident, narrowly (ca. 0.25-0.5 mm. wide on each side) greenish-alate between the more distal sessile ghee Pinnules to ca. 9 cm. long and 1.8 cm. wide, more or less deeply pinnatifid (the basal segments usually free on the basiscopic side, the acroscopic side adnate), sete to short (to ca. 1 mm.) petiolulate, 138 GERALD J. GASTONY narrowly ovate to narrowly triangular, the apex acute to acuminate; costa adaxially pubescent, abaxially pubescent and with occasional small, tan, dark-setate scales, the margins and apex subentire to serrate-dentate, margins more or less revolute; costule re 0 adaxially glabrous, occasionally with a stiff trichome distally, abaxially pubescent to squamulose with highly trichomoid (rarely dark-setate) squamules; veins commonl simple but often once-forked at or near the costule, evident on both surfaces, adaxially = 2S dy = c =f owe £ : ‘aan oe Ae ae = Ey : f Pf, sea in Sout : bilge, Bells Siar an y NIALL don s+ - — & amet Maps 20-24. 20. Nephelea Sternbergii var. Sternbergii; 21, N. St ii : ? . > > . ernbe: ‘4 . N. incana; 23, N. mexicana; 24, N. polystichoides. Sot Neve pe veteran en, 73" THE FERN GENUS NEPHELEA 139 or deeply cyathiform or sub- fi arrige ile diaphanous to light brown, of more or less delicate texture, persistent with those more sp ptr deg often lobed or ruptured at maturity, pubescent with more or less por and often crispate trichomes or tricho- In addition to the characters in the key, the long, narrow, often almost filiform, light tan scales on the petiole base serve to characterize this species. The scales borne abaxially on the petiole base have a dark apical seta often up to 2 mm. long, the longest in the genus. I have seen only one petiole base from Argentina and that is from the syntype of Cyathea O’Donelliana cultivated at the Instituto M. Lillo. In view of the congru- ence of all other critical characters in the Argentine material, the lack of the characteristic long tan scales from this petiole is taken to be an aber- rancy of the individual, perhaps related to the conditions of its cultivation. As in Nephelea Grevilleana and N. crassa and occasionally elsewhere, the morphology of many of the paraphyses is suggestive of sporangial derivation, especially when the paraphyses are more or less tufted at the apex of the receptacle. The pinnules occasionally disarticulate from the pinna-rachis in this species as in several others in the genus. The distribution of Nephelea incana based on the few presently avail- able specimens is quite discontinuous, reflecting the need for more collec- tions from the Andes. Nephelea incana is known from Colombia, Ecuador, Peru, Bolivia, and northern Argentina in wet forest on mountain slopes, in ravines, and along rivers and road banks at altitudes of ca. 800-2500 m. ITIONAL SPECIMENS rena sheer eed Pichincha: 1907, Mille (us); 1907, Sodiro Sona? US 2427083 (a,us). Tungurahua: Bell 809 (cu,s). Guayas: André 3656 (Ny). Peru. Junin: Esposto 670 Bi ses tis & Iltis 249 (cu). Bolivia. La Paz: Herzog 1605 (uc). serrata Salta: Capurro 245, 279, 292 (iu); Nervoorst 440 (LIL). 17. Nephelea mexicana (Schlect. & Cham.) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics, 91-94, Map 23 Cyathea mexicana Schlet. & Cham. Linnaea 5:616. 1830. TYPE: — Jelapa, Schiede, B (seen and confirmed by Rolla Tryon in 1969); isotype: BM!, a et of Schiede 802 a ex B is doubtless authentic; photo uc! of isotype in ik . J. Sm. atk a articulata Fée, Mém. Fam. Foug. 8:1 1857. Type: Mexico, Villa-Al et Taleo, pales 6531, probably BR; istotypes: B ee seen), Ny! fragment ex B, o in pe emite - es ker, Jour. Bot. 15:161. 1877. sighs Ecuador, collected in the poe es of ne re 1875, Sodiro x!; isotypes: P! (two sheets), us! fragment ex Rose: ae “firma ( (Baker) Domin, Meese 264. 1929, not Mett. ex Kuhn, 1869. Cyathea firmula Domin, Acta Bot. Bohm. 9:115. 1930, nom. nov. for Hemitelia Bak t Cyathea eg ndiode ex — dt ‘catalan ri ‘ rv. Jard. Bot. Geneve 4:207. 1900. Costa Rica, bords du Rio ae Las ites araeitiaue. alt. 631 m., Nov. 1898, Tonduz 140 GERALD J. GASTONY Fics. 88-90. Nephelea incana: 88, Central piunules of 89, Central pinnules of a central pinna, July 1907. pinnule as shown in 89, with shallow » Cen al pinna 3904 (us), x 0.5; 92, Central pinnules o! tral pinna, Pringle 7892 (cH), < 0.5; 93, Central portion of a pinnule in 92, with unruptured sphaeropteroid indusia x2; 94, al view of distal pinna-rachis showing the alate n een more € pinm well as between the adnate pinnules Gastony & Gastony 784 (cH), 2. Fics 95-98: Nephelea polystichoides, all alate betw te segments a Dp Nisman 111 (cu), « 0.5: 96 pinnule of a central pimna, more complex than those in 95, 1! Wercklé (P), x 0.5; 97 Central pinnule of a ge representative of the most complex leaf form in the specie S in 95-97 but of intermediate complexity, ia, 1904, Wercklé, Herb. Nac. Costa Rica no. 16787 (2), x 2. THE FERN GENUS NEPHELEA 141 12783, no specimen found by Rolla Tryon in 1969 in Christ's herbarium at P; iso F! fragment in herb. Jeanpert, cu!, m!, ny! (two sheets, one a fragmen oe Christ), p! (two sheets, one in herb. Bon napart), us! (two sheets, one a en Nephelea patellaris (Christ) Tryon, Contrib. Gray sit 200:40. 1970. Alsophila costalis Christ, Bull. Herb. Boiss. I, 4:9 1904. type: Costa Rica, colectée s.i 2 et Brune ahha ty ent au si oat de Turrialba, leur domicile, p! (probable, annotated erb. Christ, but does not include Brune’s name as collector ); isoty y! fragment ex Christ. pe costalis (Christ) Domin, Pterid. 262. 19 ophila furcata Christ, Bull. Herb. Boiss. Il, 4:957. 1904. type: Costa Rica, ons a -Quelle, Werc klé et Brune, P! S pinky annotated as n. sp. herb. pre but does not include Brune’ e's name as ae a phila tenerifrons Christ, Bull. Herb. Baise IL 4:959, 1904. rypE: Costa Rica, isotypes: innule ex P isotype, ny!, p! in herb. Tonduz, us! fragment from P isotype or ag missing holotype Cyathea tenerifrons (Christ ) Domin, Pterid. 263. 1929. Nephelea tenerifrons (Christ) Tryon, Contrib. Gray Herb. 200:40. 1970. Cyathea arida Christ, em Herb, Boiss. II, 6:180. 1906. synrypEs: Costa Rica, Navarro Luna, 1400 m., Wercklé 16757, 16780; alto de Mano de Tigre, oe 700 itti name in ; 00 include “Luna” as part of the locality) m y be taken to fix the application of the name until the collections cited by Christ are ie facade: Five other sheets from Christ’s a gemmifera Christ, in Jiménez, Bol. Fomento Org. Min. Fomento San José, Costa Rica 3:661, c. fig. 1913. TYPE: Costa Rica, Cerros del Tremedal, San n. de imé i herb. Jiménez 914 (=Herb. Nac. Costa Rica no. 17565) F! on snipe uc!. Note that Figs. 1, 5, and 6 in this publication are of Jiménez, herb. Jim 909, Hacienda Violia, San Ramén, Julio 1913 and this specimen is Nephelea soe 8 ( Christ ) Tryon, F!, us!. Stem to 10.5 m. tall and 30 cm. in diameter (including the investing sheath of ate roots) bearing be scales especially toward the apex and blackish spines to 12 mm. long, occasionally with adventitious buds. Unexpanded croziers with the abaxial surface sparingly persistent, brown to tan with a darkened central stripe, ith a dark apical seta to 1.5 mm. long and — with smaller — apical and m. pinnatifid, ab : a . P brown, adaxial minutely pubescent Ds - rescent, a ly Central pinnae ' sessile le to sh a hort (to 2.0 cm.) barren ovate-lanceolate, to 87 em. long and 20 cm — basal ae reduced, to ca. 17-30 cm. long; pinna-rachis ovate to narrowly obovate, the a abaxially furfuraceous-squamose to glabrescent, sparsely pubescent lly, the scales 142 GERALD J. GASTONY . 2 mm. long, tan, with several dark apical and lateral setae, infrequently clustered in the axil. Segments falcate to subfalcate with lateral margins and apex abaxially glabrous or occasionally with an obscure more or less araneose squamule; mina surface adaxially and abaxially glabrous between the veins. Sori at the forking i t on si i oO veins or near the costule ple veins; indusium sphaeropteroid to sub- sphaeropteroid, with or without an apical umbo, diaphanous to whi tan, of delicate to very delicate texture, us ally fugacious at maturi f the ule except for an occasional dark seta; receptacle globose, inserted before loss of the indusium, with short (to 0.1-0.3 mm. long) mostly inconspicuous paraphyses. Four sheets were found under the name Cyathea arida Christ at p! by Rolla Tryon in 1969. Two are in Christ’s herbarium and were collected in Costa Rica by Wercklé in 1904 without specific locality, a third is without collection data or any indication that it is part of the Christ herbarium, and the fourth is in Christ’s herbarium and was collected by Wercklé in 1905 in Navarro, Costa Rica at 1400 m. All four are annotated C. arida n. sp. The data associated with the fourth sheet brings it closer than any others I have seen to the syntypes cited by Christ. Although it was collected at Navarro, it is not annotated Navarro Luna. I do not wish to choose a lectotype until specimens with the full locality and numbers cited by Christ are seen. Two factors which have contributed to the confused taxonomy and substantial synonymy associated with this taxon are the varying degrees of persistence of the indusium and the variation in details of size and shape of the leaf segments. Because the indusia are sphaeropteroid to sub-sphaeropteroid and quite delicate, they are often nearly completely fugacious both in fully mature and nearly mature sori which have been dried in the course of makin herbarium specimens. Although critical examination reveals the presence of an indusial remnant at the base of the receptacle in these specimens, such remnants were often overlooked and these specimens were described or identified in the genus Alsophila. Also indusial fragments were mistaken for the complete form of the indusium, for example, in the case of Hem- itelia firma. Those specimens which retained the indusium were attributed to the genus Cyathea. The size and shape of the pinnules and their associated venation pat- smaller to larger and more complex pinnules. This has not been previously recognized, especially in determining material from Costa Rica. Speci- mens with the smallest pinnules, simplest venation, and greatest uniform- THE FERN GENUS NEPHELEA 143 ity of ultimate segment cutting generally have been referred to Cyathea mexicana; those with somewhat larger pinnules, more forked venation, and some separation of sub-basal and basal segments (often with the basiscopic side of the sub-basal segments decurrent to the basal segments ) to C. patellaris; and those with the largest pinnules, the greatest incidence of forked veins, and more coriaceous texture to Alsophila tenerifrons. It is now clear that the venation pattern and the extent of lamina surface are correlated. Small pinnules or those with small ultimate segments have simple veins, whereas the larger pinnules with larger ultimate segments have forked veins (rather than a greater number of simple veins). All intermediate conditions may be observed. Also associated with the larger pinnules and larger ultimate segments is a tendency for the most basal segments to be less adnate than in the smaller pinnules. In addition to specimens of Nephelea mexicana from Panama and Ecuador which fall within the central variation of the species, a few collec- tions from these countries represent a more extreme variation character- ized by larger size and greater coriaceousness of their pinnules. Because no reliable critical characters are associated with the latter conditions, these plants are not considered to be of taxonomic significance. As in Nephelea erinacea and N. polystichoides, the leaves of N. mexi- cana have often been observed in apparent whorls of four or five (the former giving the trunk a squarish cross section). Furthermore, as in these and other species of Nephelea, the pinnules occasionally disarticu- late from the pinna-rachis. Nephelea mexicana occurs in Mexico, Guatamala, El Salvador, Hon- duras, Nicaragua, Costa Rica, Panama, and Ecuador at altitudes of ca. 50- m. in cloud forest or on rain forest slopes, in ravines, along streams, road banks and in more or less exposed to shaded sites of dense woods. In Guatemala it is often noted as occurring on south-facing slopes and in Mexico and Nicaragua it is frequently found in association with Quercus and Liquidambar. ADDITIONAL SPECIMENS EXAMINED: Mexico. Jurgensen 911 (Ny); cng 95a (NY); Liebold 50 (ny). San Luis Potosi: Graber Ge Say meen Hi : Moore 5051 (uc), 4255 (m,uc); Smith 2009 (cH,Ny), 2184 (F,mMo); Spence 2 (cH). pene Mickel 1089 ate 1324 (us). Chiapas: Purpus 6766 (F — NY,UC). Guatemala. Huehuetenango: Steyermark 48880 (F,us), 51736 (F,GH,us ). San Marcos: Sieger 36866 (¥F,cH). Quezaltenango: Skutch 1410 (cH); Standley 86662 (us); Steyermark 33309, "33494, 33540 (F,us), 33311 (F,cH). Suchitepequez: Brenckle 47— 192 (F,Ny); Steyermark 46778 (F,cH,vs), 46837 (us). Peten: Contreras 2272 (us). Alta : nson ; 144 GERALD J. GASTONY L. O, Williams et al. 23874 (¥,Ny,us). Zelaya: Proctor et al. 27359 (ny). Costa Rica. Guanacaste: Standley & Valerio 46175 (cH); White & Lucansky 196869, 196873 ; “ ; 1715 Nisman cH); White & Lucansky 196876 (puKE,cH). Here Burger & Stolze 5811 (cu); Nisman 52, 91, 120, 178 (cH); Scamman 7002 ( us), ; Scamman & Holdridge 7863 (cH). Limon: G ny & Gastony 777, 7 (cH); Ku 1 (m); Nisman 132, 137 (cH); Tonduz 14571 (ny). San José: Brade 21 suc); G Gastony 751, 752 (cH); Kupper 1401 (m); Molina et al 18098 (F); Nisman 39 (cH), 42 (cH sheet 2); Scamman 5878 (¥,GH,Ny,uc); Stork Burger & Stolze 5518 (cu); de la Sota 5230 (vs 8) Lent i (cx): Nisman 86, 142 3904 ey anal Zone: Aviles 4 (F,us); Kenoyer 6 (cH, us). Panama: Cornman 655 (Mmo,uc); Killip 2692 (cu,uc Us), 2741. (us). aia ag 1905, Sodiro (Mo). Pichincha: Bell 258 (cH); Sparre 13831 (cu). Chimborazo: Spruce 5744 (¥,P). El Oro: Little 6651 (¥F }. 18. Nephelea polystichoides (Christ) Tryon, Contrib. Gray Herb. 200: 40. 1970. Fics. 95-98, Map 24 Alsophila polystichoides ae Bull. ope Bot. Belg. 35 (Mém.): 177. 1896. Type: Costa Rica, “arborescent” Pentes boisées au dessus d’Ara Aragon, 600 m., 20 Oct. 1894, Pittier 9017, sr! (annot ae as n. sp. Chis but not as herb. Christ, two sheets), photo s!, probable portion ny! ex Chris sotypes: Babs us! (two sheets). Cyathea polystichoides (Christ ) ean Presid 26 3. 1929. Cyathea hastulata Christ, Bull. Herb. Boiss. I, 4:945. 1904. typPE: Costa Rica, Christ. : ; : : Costa Rica, La Palma, 1459 m. ( erroneously published as 1495 m.), Tonduz 12677, P!; isolectotypes: BI, us! (two sheets). Isosyntypes: Lagi 12528, s!, xl, ami ny! (two fra Navarro La Luna, Werklé, “arbre élevé, épineaux,” ead. unknown, perhaps sR. Authentic material which may serve to fix the application of the name until the holotype is located is at BM! (two Sinets in herb. Claistscann ex herb. Christ), ny! (one sheet ex herb. ist, one ex Rosenstock, Werklé pr! (two sheets in herb. Bona- parte), us! (two sheets ex herb. Christ t). Stem to 6 m. tall and 10 cm. in diameter, bearing brown scales especially toward = apex = blackish spines to 5 mm. long, occasionally with adventitious buds. croziers wi ] i aringl central stripe and with a dark apical seta to 0.5-1.0 mm. long and occasionally with several smaller apical and lateral setae. Lami yraceous a viel perecsoms, to wi predominantly tripinnate-pinnatifid, abruptly reduced to a tinct tna ike apex (to 48 cm. long). Rachis brown, adaxia ally pubescent to glabrescent, ie furfur- aceous-glabrescent. Central pinnae sessile short pe ca. 2 cm.) nar ovate- lanceolate, to 82 cm. long and 28 cm. ‘asal pinnae more or less reduced, to 30-45 cm. long; pinna-rachis adaxially onkaiies abaxially furfuraceous-glabrescent, THE FERN GENUS NEPHELEA 145 pubescent distally, pach inconspicuous to arene conspicuous lateral pneumatodes at the axils of the tertiary axes, narrowly (to ca. 0.4-0.7 mm. wide on each side) greenish-alate site the more distal sessile Aree ules. Pinnules to 15.5 cm. long and 4.5 cm. wide, predominantly pinnatifid (pinnate at the bas e) to predominantly pinnate-pinnatifid, sessile to short (to ca. 2 mm.) petiolulate, ovate to elliptic-oblong xiall gr i or pinnati then — to short (to ca. 0.5 mm.) petiolulate, ra subentire to serrate- Ske ay margins slightly or not revolute; quaternary axis adaxially usually wih one to few stiff has ati distally, occasionally glabrous, abaxially glabrescent arsely pubescent distally and aces sely and obscurely squamulose proximally, the dicate light tan, to ca. 0.5 ea ng, dark-setate or with non-dark trichomoid processes, in pinnatilobate to pinnatifid tertiary segments the quintary axes occasion- ally with a stiff trichome distally on the adaxial surface; veins in serrate-crenate margined tertiary segments once or more forked (at or near the quaternary axis), in pinnatilobate to pinnatifid tertiary segments simple on the quintary axes, evident on oth surfaces, adaxially glabrous or with an occasional trichome, abaxially — r (especially in tertiary segments) with a diminishing continuation of the agape of the costule; lamina surface adaxially and abaxially glabrous between Sori at or near the quaternary axis; indusium sphaeropteroid to subpart th or without an apical umbo, diaphanous to whitish tan, of delicat texture, usually fugacious at maturity or upon drying of the sorus, often only a rem- nant remaining around the base of the receptacle, glabrous or nearly so; receptacle globose, inserted before loss of the indusium, with short (to ca. 0.1-0.2 mm. long) mostly inconspicuous paraphyses. In the protologue of Cyathea Werckleana, Christ cited at least three collections: La Palma, 1495 m. Tonduz 12528, 12677, and Navarro, Wercklé. In his search for types at p in 1969, Rolla Tryon could find no material of the Wercklé collection in Christ’s herbarium, nor have I seen such label data on any sheet with this name from Christ’s herbarium. The isolecto- type of Tonduz 12677 at B was unidentified and bears no mark to indicate that Christ ever saw it. The two isolectotype sheets at us are annotated as determined by Christ, but bear the apparently unpublished epithet “pin- nula” in the genus Alsophila. Two sheets of Tonduz 12528 were found at p, but neither is stamped as belonging to Christ's herbarium. One is anno- tated “Cyathea Werckleana n. sp.,” the other “Cyathea (Hemitelia)” with the apparently unpublished epithet “pinnatilobula.” The B sheet of Tonduz 12528 was unidentified; the other isosyntypes are annotated “pinnati- lobula,” as is an Ny sheet from Christ’s herbarium Wercklé in 1903 without locality. All collections I have seen of Tonduz 12528 and 12677 which indicate the altitude give 1459 m. rather than 1495 m. as published by Christ, presumably in error. All of this material represents one taxon and may be taken as authentically representing the name. Because the only sheet I have seen from Christ’s herbarium bearing the annotation “Cyathea Werckleana n. sp.” is the Tonduz 12677 sheet at p, this collection is chosen as the lectotype. In his search for types at P in 1969, Rolla Tryon was unable to find the 146 GERALD J. GASTONY type collection of Cyathea hemiotis Christ. The sheets at BM, NY, P, and us cited above as authentic material do not bear the locality data of the type. I have seen no collection from Christ’s herbarium which would have provided him with the petiole characters cited in the protologue. It is obvious that the holotype consists of other material at P, BR or elsewhere bearing Wercklé’s locality data and forming a more complete specimen. In the complexity of its lamina, Nephelea polystichoides is the most variable species in the genus. Nevertheless, the former recognition of at least three species on the basis of degree of lamina complexity cannot be supported. Whereas no critical characters have been found to justify the indusial characters, and in all there is a tendency for the dried segments of the second and third orders to be deciduous. Although the full range of variation of lamina complexity from bipinnate-pinnatifid to tripinnate- deeply pinnatifid has not been observed in any one individual, many speci- mens do feature much of this variation and their respective ranges of variation overlap from one extreme of the species variation to the other. tecture. Adequate materials of these two taxa are clearly separable on the basis of the key characters, Altho gh N. mexicana is interpreted in the key decurrent to the basal segments. These are distinct from, but suggestive of, the condition in N. polystichoides noted in the key as “tertiary axes Nephelea polystichoides is known from the provinces of Alajuela, Heredia, San José, Cartago, and Puntarenas in Costa Rica and from the province of Chiriqui in Panama. It occurs on mountain slopes and along stream banks in undisturbed and cutover cloud forest at altitudes of ca. 700-1970 m. (with one collection recorded as 90 m.). ADDITIONAL SPECIMENS EXAMINED: Costa Rica. 1901-1905, Wercklé (ny), 1903, klé (wx), 1907, Wercklé (P,us), Wercklé 49,55 (us). Alajuela: Burger ¢ Stolze a); Gas Gaston cS : (GH,Ny,uC,Us); A, Smith H322 (¥F); Tonduz, Herb. Jim. no. 909 (¥F,us); White & Lucansky 196810, pe (DUKE,cH). Heredia: Burger & Stolze 5811 (cH); Scamman 8 THE FERN GENUS NEPHELEA 147 (cH); Gastony & Gastony 770 (cH); Nisman 94, 96 (cH); Scamman & Holdridge 7866 (cH,us); Tonduz 2244 (us); Wercklé, Herb. Jim. no. 553, 561 (us). Cartago: Kupper 778 (mM); Nisman 26, 28, 126, 151, 154 (cH); Scamman 7586 (cH); Tryon & Tryon 6990, 7021, 7037, 7038, 7040, 7041, 7042 (cH); White & Lucansky 196860, 196862, 196897, 196899, 1968109, 1968132, 1968133, 1968134, 1968140, 1968143, 1968144 (pUKE,cH). Puntarenas: Burger re Stolze 5518 (cH); Nisman 141 (cH). Panama. Chiriqui: Cornman 1359 (Mo,us ); Killip 5393 (us); Maxon 5706 (GH,NyY,US). DUBIOUS AND EXCLUDED NAMES Alsophila ne bate & Gal. Mém. mare Brux. 15:78, t. 23. 1842. =Cyathea fulva (Mart. & Gal.) Fée, Mém. Fam. Foug. 9:34. 1857. Contrary to Maxon’s suggestion (1909), the type of this name (Galeotti 6346 p!) is not a form of Nephelea mexicana oo & Cham.) Pliny but rather a aoe s of Cyathea. Cyathea glauca Fourn. Mex. Pl. 1:135. 1872, not Willd. Sp. Pl. 5:493. 1810. The Cyathea hexagona Fée et Schaffn. ex Fée, Mém. Fam. Foug. 8:111. 1857. Type: Mexico, pres Huatusco, Schaffner 237. Hooker (1865, in Hooker & Baker) and Maxon (1909) cited this as a syno nym © of C. mexicana Schlect. & Cham. The sterile element of the mixed collection comprising the sheet at x! (in Moore’s herbarium ex Fée) is C. mexicana. The description of the so oral position, the unar rmed nature of the iol Wi e and should ie the lectotype. I therefore exclude from this name the sterile element which is Nephelea mexicana (Schlect. & Cham.) Tryon. Cyathea Imrayana var. subnudata Hook. Sp. Fil. 1:18. 1844. This variety is —— from C. Imrayana, It is from Jamaica and was collected by Wiles and by Bancroft. It may be a synonym of C. Tussacii Desv., but its precise disposition must await a study of the types, presumably at x yathea mexicana var. boliviensis Rosenst. Fedde Repert. 25:56. spain oe Simaco supra Tipuani, 1400 m. alt., 3, 1920, ce 5140, F!, cul, ny!, us!. This is certain! not a variety of Nephelea mexicana thea oF N. woodwardioides var. cubensis, should not chosen as the lectotype collection because N. woodwardioides var. cubensis does not occur in the British West Indian Islands, from which Grisebach described A. nitens and ecause its indusium is not consonant with Grisebach’s description. — papyracea Christ, Bull. Herb. Boiss. Il, 4: 946. 1904. I have seen two ist’s herbarium at P bearing the annotation Cyathea —— n. sp. es labeled “1eF© feuille” and “2°™* feuille.” These are clearly two distinct taxa, the first sheet is Nephelea mexicana (Schlect. & Cham.) Tryon and we seco’ oa sheet is a species of Cyathea. The second sheet is chosen to represent the name because of the greater correspondence of its indusia and rachis with the description. In addition, Maxon (1909) said the fragment he ne seen is close to C. onusta Christ, w lies better to a second n to the first sheet, and finally only the second sheet a the emilying Wercklé 52. The first sheet is therefore excluded from the type var. squamosa Bosco, Nu. Giorn. Bot. Ital. n. ser. 45:142. 1938. Syntypes: seer in Plan de Milagro, 1500 m. et de Sapote, 2000 m., Crespi herbarium del Liceo Pareggiato “Don Bosco” a Torino (Turin), not seen. If this 148 GERALD J. GASTONY ariety is correctly — to C. Sprucei Hook., it is probably a synonym of N ophei 2a incana ( Karst.) Gastony. LITERATURE CITED Brabe, A. C. 1971. Cyathea Sampaioana Brade et Ros. somente uma “forma” de Cyathea Sternbergii Pohl. Bradea 1(10):73-76. CHRISTENSEN, C. 1937. The collection of Pteridophyta made in Hispaniola by E. L. Ekman 1917 and 1924-1930. Kungl. Svensk. Vince. -akad. Handl. ser. 3, 16:1- CopELAND, E.B 1947. Genera Filicum. Chronica Botanica. Waltham, Massachusetts. Hate, F. 196 66. Etude de la ramification du tronc chez quelques fougéres arbor- escentes. ee 6(3) :405-424. Ho.itrum, R. ait ae Flora Malesiana i, ee sors Lanjouw, J.« & F, A. STaFLEv. 1964. Index Herbariorum. Reg. Veget. 31:1-251. Lucansxy, T. W. 1971. Comparative studies of the nodal and vascular anatomy in the ae es Cyatheaceae. Unpubl. Ph.D. dissertation, Dept. of Botany, Duke Uni Maxon, W. R. 1909, Cyatheaceae. North Amer. Fl. 16:65-88. maa. Ww a noteworthy ferns from the Dominican Republic. Proc. Biol. Soc. _ Wash, 37: 4, Fern caaeines ¥ oc. Biol. Soc. Wash, 43: Purest, C. B. 1836. ‘Tentamen Preritopraliios seu Gener Five ScuucHERT, C. 1935. Historical geology of the Astilleds Caribbean ess Le Wiley & Sons, Inc. New York. pgm G. G. 1961. Principles of Animal Taxonomy. Columbia Univ. Press, New Y¥ Tryon, R. M. 1960. A glossary of some terms relating to the fern leaf. Taxon 9: ——. 1970. The classification of the Cyatheaceae. Contrib. Gray Herb. 200: ————-——~—. 1971. The American tree ferns allied to Sphaeropteris horrida. Rhodora ALKER, ‘T. G. 1966. A cytotaxonomic survey of the Pteridophyta of Jamaica. Trans. Roy. Soc, Edinb. 66: 169-237. Contributions from the GRAY HERBARIUM 1973 NO. 204 THE BIOSYSTEMATICS/ OF T : we Thsan A. Al-Shehbaz aa e es PURPLE-FLOWERED ARABI Reed C. Rollins COAST OF NORTH AMERI ONSIDERATION Sas Reed C. Rollins A RECON SN - JAEGERI es | Missour: Borawcar = es Sesh Contributions from the GRAY HERBARIUM 1973 NO. 204 THE BIOSYSTEMATICS OF THE GENUS Ihsan A. Al-Shehbaz THELYPODIUM (CRUCIFERAE) : as PURPLE-FLOWERED ARABIS OF THE PACIFIC Reed C. Roltins COAST OF NORTH AMERICA ‘ A RECONSIDERATION OF THELYPODIUM Reed C. Rollins JAEGERI EDITED BY Reed C. Rollins Kathryn Roby PUBLISHED BY THE GRAY HERBARIUM OF HARVARD UNIVERSITY IssuED OcrosER 1, 1973 THE BIOSYSTEMATICS OF THE GENUS THELYPODIUM (CRUCIFERAE) THsAN A. AL-SHEHBAZ! INTRODUCTION Those familiar with the Cruciferae readily agree that among the most acute and frequently faced problems in the systematics of this family are those having to do with generic delimitation. It is often claimed that the genera of the Cruciferae are not natural, but perhaps the problem may be better expressed by stating that the generic boundaries are sometimes artificially drawn. Undoubtedly one of the most controversial aspects of the systematics of the family involves its classification into tribes and sub- tribes. There is little agreement between the 15 or more classifications consulted as to the number of tribes recognized or in the number of genera assigned to a given tribe. This is particularly true of the tribal disposition of the various primitive genera of the family. The two most recent treatments of the Cruciferae (Schulz, 1936; Janchen, 1942) place the ten related primitive genera and their segregates in four to seven tribes. Neither of these is natural nor can they be accepted. The species dealt with in the present detailed study belong to the genus Thelypodium and other closely related genera. They were investigated earlier by Payson (1922, 1923). His treatment helped significantly to es- tablish the proper generic boundaries between Thelypodium and Cau- lanthus, but he did not solve the generic placement of certain species that have been referred to Thelypodium at one time or another. Further- more, he (1922) did not satisfactorily treat four species of the alliance by placing them in Sisymbrium. Shortly after his two treatments, the species he excluded from Thelypodium reappeared in the genus again in some of the major floristic books on western United States. The treat- ments of Jepson (1936) and Schulz (1936) strongly diverge from that of Payson (1923). Schulz recognized Thelypodium as a monotypic genus and maintained all of its previously proposed generic segregates, while Jepson preferred a much broader generic concept. He included in Thely- podium species that are generally placed in Streptanthella and Caulanthus. In addition to the difficulties encountered in establishing the generic boundaries of Thelypodium, the tribal classification of the genus has been even more problematic. The genus was placed in the tribes Thelypo- dieae (Prantl, 1891), Hesperideae (Schulz, 1936) and Sisymbrieae (Jan- chen, 1942). The present research consists of a number of approaches, including chemical and palynological, which have not been used before in the study *Present address: Dept. of Biology, College of Sciences, University of Baghdad, Baghdad, Iraq. 3 4 IHSAN A. AL-SHEHBAZ of the primitive Cruciferae. This investigation consists of the following: (1) a systematic treatment of Thelypodium; (2) establishment of the generic boundaries between the primitive genera of the family as well as their tribal classification; (3) a presentation of some of the evolutionary trends among these primitive genera; (4) palynological analysis of the various species of Thelypodium as well as selected genera of the sub- family Cleomoideae-Capparaceae using the scanning electron micro- scope; and (5) the presentation, for the first time, of a major chemosys- tematic study of Thelypodium involving the distribution of mustard oil glucosides on inter- and intraspecific levels. In this, both paper and gas- liquid ch tographic app Ss were utilized. ACKNOWLEDGEMENTS Professors Rolla M. T: on, Carroll 00 , Otto T. Solbrig, Elso $ Barghoorn, G. Torrey; Drs. Charles Cleland, Lorin I. Nevling, Jr., Eli A. Shaw and Alice Tryon; and James E. Rodman, Kathryn Roby, Umesh Banerjee, Richard Leo and Lily Riiden Seed samples have been received from Drs. Arthur S. Barclay, Martin G. Ettlinger, Professor Arthur R. Kruckeberg, Miss Barbara Lund. Drs. uane N. Atwood, Janice C. Beatley, Larry C. Higgins, Noel Holmgren, Patricia K. Holmgren, and James L. Reveal have kindly sent me some of their collections of Thelypodium for (Professor Reed C. ollins, Harvard University, Principal Investigator) ) to the Evolu- i i Fernald Fund of Harvard Last, but certainly not the least, I would like to express my deepest appreciation and gratitude to the Iraqi Government for financial support throughout my entire graduate ies, History OF THE GENUS During the earlier stages of botanical exploration of the New World, particularly the last part of the eighteenth and the early decades of the nineteenth centuries, plant specimens were mostly sent to European tax- onomists for study. The new species discovered were often matched with known species that belong to Old World genera, and in many instances they were placed in these genera. This was the case with the first species of Thelypodium discovered. In 1830 Sir William Jackson Hooker described a new species of Cruci- BIOSYSTEMATICS OF THELYPODIUM 5 ferae collected by David Douglas near Wallawallah and Priest’s Rapids on the Columbia River in what is now the state of Washington. The species was given the name Macropodium laciniatum. Macropodium was then known to consist of a single eastern asiatic species, M. nivale. Following his discussion on M. nivale, Hooker (1830a) wrote the following on M. laciniatum: “In the long spiked raceme and the general structure of the flowers, this has a very close affinity with the preceding species; but there the affinity ceases: for, even in the inflorescence, when it comes to be minutely examined, a very marked difference will be discovered.” Hooker then discussed a number of differences in the petals, anthers, pedicels, and gynophores of the two species. The first attempt to place this American species in a distinct genus was made by Nuttall in his manuscript published by Torrey and Gray (1838). He proposed the name Pachypodium to include, in addition to P. la- ciniatum, two new species, P. integrifolium and P. sagittatum. These three species have been consistently placed in the same genus by most taxonomists even though the generic position of the last species was questioned by Torrey and Gray (1838) who suggested that P. sagittatum may belong to Sisymbrium section Cardaminopsis. The name Pachy- podium of Nuttall was illegitimate because it was a later homonym to that of Lindley (1830) in the Apocynaceae as well as to Pachypodium of Webb and Berthelot (1836) in the Cruciferae. It was replaced by Thelypodium, a name proposed by Endlicher (1839). These three species were trans- ferred from Pachypodium to Thelypodium three years after the publication of the latter genus. In 1871 Sereno Watson transferred certain species originally described in Streptanthus to Thelypodium. The generic limits of Thelypodium were further expanded by Robinson (1895) to include 23 species. Robinson recognized the heterogeneity of such an assemblage and attempted to reorganize the genus by dividing it into the three subgenera: Hesperi- danthus (one species), Heterothrix (two species), and Euthelypodium (20 species ). In 1907 Per Axel Rydberg raised the first two subgenera of Robinson to the rank of genus and proposed three additional genera, Pleuro- phragma (three species), Stanleyella (one species), and Thelypodiopsis (four species) which he segregated from Euthelypodium of Robinson. Rydberg added two species to Thelypodium (T. palmeri and T. leptose- palum) and one to Pleurophragma (P. platypodum), but it is now clear that they represent a renaming of species previously named by others. A critical evaluation of these segregate genera is presented in the section on the generic limits of Thelypodium. In his second edition of the “Flora of the Rocky Mountains and Adjacent Plains,” Rydberg (1923) assigned ten species to Thelypodium, the first five of which belong to one species, T. sagittatum. 6 IHSAN A. AL-SHEHBAZ Payson (1923) presented a good taxonomic treatment of Thelypodium and its immediate allies that helped to establish the generic lines be- tween Thelypodium, Caulanthus, Streptanthella, Chlorocrambe, and Warea. He gave a critical evaluation of some of Rydberg’s segregate genera, but his treatment has a few inadequacies. In my opinion, Payson mistakenly recognized Stanleyella as a distinct genus from Thelypodium using the orientation of the sepals and the cellular pattern of the septum as a basis for distinguishing the two. Furthermore, he ignored a number of species originally described in Thelypodium and did not attempt to discuss their generic status. Payson recognized three species related to T. integrifolium (T. lilacinum, T. affine, and T. rhomboideum), but taxonomists that subsequently dealt with this complex were in fair agree- ment that a single polymorphic species is involved (Jepson, 1936; Abrams, 1944). One of the strong points in Payson’s treatment is his transfer of four species commonly placed in Thelypodium to Caulanthus, but it is unfortunate that he did not present more convincing reasons for trans- ferring these species (C. cooperi, C. lasiophyllus, C. flavescens, and C. anceps) because this undoubtedly is the proper generic position for them. In any case, it is surprising to see some or all these four species placed back in Thelypodium (Jepson, 1936; Abrams, 1944; Munz, 1959; Hitch- cock, 1964). In 1922 Payson transferred four species known for a long time in Thely- podium to Sisymbrium (T. ambiguum, T. elegans, T. aureum, and T. linearifolium) even though he admitted that such a transfer would ex- pand the limits of the latter genus beyond what was then generally rec- ognized. In 1933 O. E. Schulz transferred a few species from Thelypodium to Thelypodiopsis and in 1936 he recognized the former genus as monotypic. This was probably because he accepted all the genera segregated from Thelypodium by Rydberg and Greene. However, his treatment received no recognition from the botanists who dealt with Thelypodium in their floristic treatments. Furthermore, Schulz left two species that unques- tionably belong to Thelypodium (T. eucosmum and T. flexuosum) unas- signed to any genus. Finally, Rollins (1946, 1957a) added four species to Thelypodium (T. texanum, T. repandum, T. paysonii, and T. tenue) and presented a criti- cal evaluation of the segregate genus Stanleyella which helped signifi- cantly in establishing the proper generic limits of Thelypodium. THELYPODIUM AND RELATED GENERA The family Cruciferae is known for two facts to most students of sys- tematic botany. First, it is one of the most natural families of flowering plants, probably because the flowers are in general very conservative structures. It has been suggested that the Cruciferae represents an adap- BIOSYSTEMATICS OF THELYPODIUM fd tive peak in the evolution of its flowers (Stebbins 1950, 1951). Second, the Cruciferae is one of the most difficult families in terms of the de- limitation of genera (Davis & Heywood, 1963; Rollins, 1938b, 1959). It is altogether possible that the difficulties faced resulted from the lack of high extinction rates that families with more easily delimited genera have suffered (Stebbins, 1950). The species in this family are easy to recog- nize for the most part (Rollins, 1959) and clearly more “natural” than the genera (Stebbins, 1950), but this does not necessarily mean that there are no problems at the species level, nor does it imply that the genera of the family are artificial. In fact, there are a number of well-defined and readily recognized genera in the family. However, problems on the species level are quite pronounced in genera like Cakile (Dr. James E. Rodman, personal communication) and Isatis (Davis, 1964; Davis & Heywood, 1963). The generic limits of Thelypodium have been the subject of a continu- ous controversy. There has been a lack of agreement among the taxono- mists who have dealt with this genus in terms of the number of species assigned to it. The number has varied from one species according to Schulz (1936) to 14 (Payson, 1923) and 23 (Robinson, 1895). Thely- podium is said to contain 45 species according to Willis’ Dictionary of Flowering Plants, while over 70 binomials are listed under the genus in Index Kewensis and the Gray Index. The various species that have been placed in Thelypodium by one author or another have fluctuated between a total of at least 20 generic names. Some of these “genera” are nothing but segregates based on minor morphological characters. In the present treatment 18 species, two of which are new, and seven subspecies are recognized. Table 1 summarizes the generic disposition of 31 species as recognized in six major treatments of Thelypodium and its allies. It clearly shows that the largest number of segregate genera in this complex were recognized by Schulz (1936) and Rydberg (1907, 1923), while both Rob- inson (1895) and Jepson (1936) preferred a much broader generic con- cept. Thelypodium. The generic limits of Thelypodium, as recognized in the present treatment, are essentially the same as those of Payson (1923) with a single exception. Payson recognized the segregate genus Stanley- ella which here is merged with Thelypodium. Rydberg (1907) segregated T. wrightii as a genus (Stanleyella) mainly on the basis of spreading to reflexed sepals which are erect in Thelypodium. He claimed that the glands in T. wrightii are inconspicuous as compared to the well-develop- ed glands of Thelypodium but admitted that this species is “close to the typical Thelypodia” in its siliques, flowers, and habit. The difference in the glandular tissue, as presented by Rydberg, is unrealistic. Further- more, the orientation of the sepals within Thelypodium is variable. The sepals in T. laxiflorum, a very close relative of T. wrightii, vary from erect 8 IHSAN A. AL-SHEHBAZ TABLE 1. THE gees 2 OF SPECIES IN THELYPODIUM AND SOME OF ITS SEGREGATES OR RELATED GENERA AS RECOGNIZED BY SIX AUTHORS Rydberg Payson Robinson 1907, 1922, Jepson Schulz Present Genus 1895 1923 1923 1936 1936 Treatment Thelypodium Endl. 1, 411, Bop: S210: 146-5 (1) 1-12, 26— 13-16, is "13 aay a) 17- 31 (18) 23 (23) 1 (11) Stanleyella Rydberg 11:43 TEL) LECT) Pleurophragma Rydb 6 (3) 6 (4) Streptanthella Rydb. 141) 17 (1) 17 (1) It) Hesperidanthus Rydb 16 (1) 16. (1) Pennellia Nieuwland 92.23 23 (6) 22° 33 (P) ( Heterothrix Rydberg) (2) Thelypodiopsis Rydb. 14.45. 3-4, 7— 3-16, 24— DA D5. §.:10, 12, . (6 ) (5) 1 ‘ 26 (14) Sisymbrium L. 13-16, 94, 25 Caulanthus Watson 18-21 18-21 (15) (18) Microsisymbrium Sch 18 (6) Lamprophragma Schulz 92 12)) Guillenia Greene 19, 21 (4) umbers in parenthesis are the total number of species recognized by the author in Lae genus. 1. Thelypodium laciniatum, 2. T. milleflorw a m T. brachycarpum, 5. T. m. 6. T. inte; mm TT, petalum, 8. T. howellii, 9. T. flexcuosum, 10. T. aa ee aL. T. wrightii, 12. T. paniculatum, 13. Thelypodiopsis ambigua, 14. T. elegans, 15. T. aurea, 16. T. lineari- folia, 17. Streptanthella aginiaiits, 18. Caulanthus lasiophyllus, 19. C. flavescens, 20. C. — 21. C. cooperi, 22. Pennellia longifolia, 23. P. micrantha, 24. Thelypodiopsis jw deere 25. wyomingensis, 26. Thelypodium texanum, 27, T. repandum, 28. T. tenue, 29. T. paysonii, = T. rollinsii, 31. T. laxiflorum. to spreading. The former species was treated as a variety of the latter by Payson. Spreading and erect sepals are known in a number of genera of the Cruciferae. This is particularly true in Caulanthus, Brassica, Sisym- brium, Leavenworthia, and some others. It is the writer's strong feeling that one character difference is not enough to provide the basis for segre- gating a genus, particularly if there is some doubt of its value or con- sistency. It is very clear, as pointed out by Rollins (1957a), that the segre- gate genus Stanleyella does not deserve independent recognition. Pleurophragma of Rydberg (1907) was based on the single character of the septum which was said to possess a “strong and broad midrib,” while the species of Thelypodium were not supposed to have a distinct midvein. Not only was this character seriously questioned by Payson (1923), but it was found to be unreliable and inconsistent. I have studied the various aspects of the septum throughout the genus and failed to recognize any features of taxonomic value. Therefore, Pleurophragma does not merit recognition. BIOSYSTEMATICS OF THELYPODIUM 9 Thelypodiopsis. Controversy still exists as to whether Schoenocrambe Greene, Thelypodiopsis Rydberg, and Hesperidanthus Rydberg should be maintained as separate genera or merged with Thelypodium or Sisym- brium. Hesperidanthus is based on a single species, H. linearifolius. The main difference between this species and the true members of Thely- podium is found in the stigma, which is entire and small in species of the latter genus (Plate 1) and conical in Hesperidanthus. Both Robinson (1895) and Rydberg (1907) claimed that the stigma in H. linearifolius is not lobed, while Payson (1922) and Schulz (1936) stated that it is distinctly bilobed. I have carefully looked at this feature of the stigma on more than 200 herbarium sheets and in no case have I been able to find an entire stigma in this species. On the contrary, it is strongly bilobed with the lobes opposite the replum, firmly touching each other, and al- most completely connivent (Plate 1). Because of its stigmas, H. lineari- folius is very distinct from all the species placed in Thelypodium in the present treatment. In its stigma and other morphological aspects, it sig- nificantly resembles the species often placed in Thelypodiopsis. Therefore, Hesperidanthus does not need to be recognized and its single species should be placed in Thelypodiopsis. Thelypodiopsis was segregated from Thelypodium by Rydberg (1907) on the basis of having strongly bilobed stigmas (Plate 1) with the lobes opposite the replum. He recognized four species in Thelypodiopsis (T. aurea, T. elegans, T. wyomingensis, and T. bakeri). In 1923 he added a fifth species by transferring Sisymbrium juniperorum. In 1933 Schulz transferred a number of species from Thelypodium to Thelypodiopsis without giving any reasons supporting his transfers. Three years later (1936) he distinguished the two genera mainly on the basis of having obtuse anthers in the former genus and acute or apiculate anthers in Thelypodiopsis. This alleged difference is far from being realistic, since it is very hard to decide in many cases whether the anther is apiculate or not. Furthermore, this feature does not seem to be significant enough to call for such a massive transfer of species. It is quite clear that none of the transferred species fit the concept of Thelypodiopsis as defined by Ryd- berg (1907). Therefore, in the following discussion we use the generic limits of Thelypodiopsis as presented by Rydberg rather than those of Schulz. Needless to say, all the species that Schulz transferred from Thelypodium to Thelypodiopsis are perfectly at home in the former genus. Payson (1922) did not recognize Thelypodiopsis and he transferred its four species to Sisymbrium. He recognized 11 species of Sisymbrium native to North America north of Mexico, six of which seem to be at home in this genus, but Payson mishandled the nomenclature of two of these species (cf. Rollins, 1960a). The remaining five species (S. ambi- guum, S. aureum, S. elegans, S. juniperorum, and S. linearifolium) make One group of interrelated species, which appear not to be phylogenetically 10 IHSAN A. AL-SHEHBAZ ee Puate 1. Stigmas and paces of piesa: 1 (1) sage Thelypodion (2). Fic. 1. Left to right: Thelypodium rollinsii, ightii, and T. texanum. Fic. 2. Left to ri hel sypdionis wyomingen- juniperorum, T. elegans, and T. Piseetag Scale: one millemet BIOSYSTEMATICS OF THELYPODIUM 11 related to any of the species that he maintained in Sisymbrium. The present writer strongly feels that these species must not be placed in Sisymbrium for the following reasons. First, the presence of a distinct gynophore that can reach a maximum length of 1-1.5 cm. in three of the five species is not found in any species of Sisymbrium. The siliques in this genus are almost always sessile or subsessile. Second, two of these five species, ambiguum and aureum, have other morphological features in the flowers and inflorescences that tie them more with the primitive members of the family. Admitting these species to Sisymbrium carries the implication that this genus has to be associated with the primitive genera of the Cruciferae. Such an action is not likely to receive support from botanists familiar with these plants. Third, the chromosome num- bers known from two of these five species (ambiguum and linearifolium ) are based on n=11 (Rollins, 1966; Rollins and Riidenberg, 1971), while 14 out of the 24 species of Sisymbrium listed by Bolkovskikh et al. (1969) have a haploid number of n=7 and six of the 10 remaining species have higher chromosome numbers, based on x=7. Although our knowledge of the chromosome numbers in this complex is far from complete, the pat- tern seems to be clear enough to support the exclusion of these five species from Sisymbrium. Finally, the close relationship of these five spe- cies to each other and not to any known species of Sisymbrium or any other genus supports their placement in a genus of their own. Therefore, Thelypodiopsis seems to be the best alternative for treating the five spe- cies under consideration. The genus Schoenocrambe was segregated from Sisymbrium by Greene (1896) on the basis of a single vegetative character (root stock), which is present in S. linifolium and lacking in the other species of Sisymbrium. He listed two other “species” and Rydberg (1904) added a fourth one, but it is clear, as pointed out by Payson (1922), that a single polymor- phic species is involved. Sisymbrium linifolium shows some superficial resemblance to certain species of Thelypodiopsis in its flowers and siliques. Payson held the view that this species is closely related to what we are calling Thelypodiopsis linearifolia, but the two species are very different in their flower morphology. Because S. linifolium is not closely related to any species of Sisymbrium native in North America, and that it super- ficially resembles species of Thelypodiopsis, creates a problem in terms of its generic disposition. Placing S. linifolium with species of Thely- podiopsis would mean that Schoenocrambe has to be recognized because it was published before Thelypodiopsis. However, we strongly believe that S. linifolium is more at home in Sisymbrium than in Thelypodiopsis, and that Schoenocrambe is not worth recognition for the following rea- sons. First, S. linifolium appears to be very closely related to a perfectly good Sisymbrium, S. polymorphum (Murray) Roth. (=S. junceum), which is native and widely distributed in Europe (Ball, 1964) and the U.S.S.R. 12 IHSAN A. AL-SHEHBAZ (Vasil’chenko, 1939). Such a close relationship was pointed out by Greene (1896), Payson (1922), and Schulz (1924). In fact, the last author treated S. linifolium directly after S. polymorphum, while Payson pointed out that that “the differences are very slight between the American plant and the Asiatic and European S. junceum Bieb.” Furthermore, plants of S. lini- folium were misidentified as S. junceum by Hooker (1830b) and Nuttall (in Torrey & Gray, 1838). Second, the chromosome number of S. lini- folium fits very well in Sisymbrium as it appears to be based on n=7 (Rollins, 1966; Rollins & Riidenberg, 1971), while Thelypodiopsis seems to be based on n=11. Third, the lack of a distinct gynophore in S. linifolium supports the disposition of this species in Sisymbrium and not Thely- podiopsis. Finally, the genus Schoenocrambe is not recognized because it was based on a vegetative character. Vegetative characters are hardly of significance in establishing generic relationships or limits, nor can the be of major value in classification when used alone (cf. Rollins, 1952, 1957b). Caulanthus. Four species of Caulanthus (C. lasiophyllus, C. flavescens, C. cooperi, and C. anceps) have fluctuated back and forth between Caulanthus and Thelypodium. Caulanthus lasiophyllus was transferred to Thelypodium by Greene (1886). The four species were recognized in Thelypodium for more than 35 years. In 1923 Payson transferred them to Caulanthus, where they should be retained. Caulanthus cooperi presents the fewest problems of the four species. The various aspects of the calyx, petals, and siliques are fairly similar to the typical species of Caulanthus and it should always be maintained in that genus. However, Jepson (1925, 1936) treated it as a Thelypodium even though he admitted (1936) that it “would rest as well in Streptan- thus” and that “it might as well be considered a primitive Streptanthus.” Jepson united Caulanthus with Streptanthus, an action that I do not sup- port for reasons to be discussed below. Caulanthus flavescens is no more difficult than C. cooperi if one takes the characteristic features of the petals into consideration. In both of these species as well as those central to Caulanthus, the petal blades are nar- rower than the claws and they are crisped or coiled. The claws are differentiated from the blades and they are mostly obovate to oblanceo- late. These features of the petals are very important in distinguishing between Caulanthus and Thelypodium since they are not found in any species of the latter genus. The only two species that deviate from this general pattern are C. lasiophyllus and C. anceps, but these are properly placed in Caulanthus if other characters are taken into consideration (Table 2). Jepson (1936) did note the streptanthoid flowers of C. flaves- cens but he preferred to maintain the species in Thelypodium even though he stated that “In Streptanthus it seems related to S. pilosus Jep- son and S. hallii Jepson.” We believe this species is a Caulanthus for the reasons summarized in Table 2. Character Stem Sepals Pedicel Pedicel orientation Gynophore Stigma Chromosome mbers C. cooperi solid glabrous glabrous or pub. curved, reflexed 1 mm. long, very broa bilobed, lobes opposite valves n=14 TABLE 2, COMPARISON BETWEEN SPECIES OF THELYPODIUM AND CAULANTHUS C. flavescens hollow glabrous or pubescent mostly pubescent strongly curved, erect or reflexed 1 mm. or less, very broad bi lobed or C. anceps mostly hollow glabrous or pub, glabrous or pub, strongly curved, erect or reflexed 1 mm. or less, very broad bilobed or subentire, lobes opp. valves n=14 C, lasiophyllus hollow or solid glabrous or pubescent glabrous or pubescent strongly curved, erect or reflexed 1 mm. or less, very broad entire or rarely slightly bilobed n= 14 other spp. of Caulanthus hollow in 6 spp., solid in 3, hollow or solid in 2 glabrous in one sp., pubescent in 3 spp., glab. or pub. in 7 spp glabrous in one sp., pubescent in 5 spp., glab. or pub. in 5 spp slightly to strongly curved, erect, reflexed, or spreading 1 mm. or less, very broad bilobed in all remaining species, obes — valves in 9 spp., opp. replum in 2 species n = 14 in 4 species, n = 10 in 2 species spp. of Thelypodium solid in 17 spp., hollow in one glabrous in all 18 species glabrous in all 18 species pes 7 slightly i trongly ne sp., eilteaby pa. 1-8.5 mm. long in 14 spp., 1 mm. or less in 4 spp., not broad entire in all 18 species n=13in9 species WOIGOdATAHL JO SOLLVWSLLSASOIA &T 14 IHSAN A. AL-SHEHBAZ Caulanthus anceps is more of a problem because its flowers do not suggest its generic placement. The spreading nature of the sepals, petals, and stamens of C. anceps is also present in five species of Thelypodium that also have petiolate cauline leaves. Considering such characters, C. anceps could easily be placed in Thelypodium. However, if this is done, the species would have no relatives, close or remote, within Thelypodium. Rather, its nearest relative would be a species of Caulanthus, C. flavescens. The various aspects of the siliques, pedicels, sepals, glandular tissue, in- florescence, and, to a certain extent, the leaves of these two species show remarkable similarities. The petals in C. anceps are neither crisped nor chanelled or coiled and the blades are attenuate into a narrow and slender claw-like base. Caulanthus anceps and C. lasiophyllus have recently been treated as Streptanthus by Hoover (1966, 1970). Of the four species of Caulanthus discussed above, C. lasiophyllus per- haps presents the most problems. One reason is that it does not have the characteristic caulanthoid flowers that are present in most species of the genus. The calyx is not urceolate and the petals are neither crisped nor chanelled. Because of this, the species was, for the most part, ex- cluded from Caulanthus. However, there is scarcely any significant mor- phological evidence that supports its placement in Thelypodium. In fact, C. lasiophyllus resembles certain species of Sisymbrium in some features more than it does those of either Caulanthus or Thelypodium. Therefore, it is not surprising to find that some of the collections of this species were described as new under such names as Sisymbrium deflexum, S. re- flexum, and S. acuticarpum. However, we are retaining C. lasiophyllus and its relatives in Caulanthus for reasons summarized in Table 2. Plants of this species grown from seeds of three population samples in the greenhouse showed a remarkably high percentage of autogamy. This was roughly estimated to range between 80 and 100 per cent. A number of floral adaptations to such a mode of reproduction were observed: small flower size, reduced glandular tissue or its absence, enlarged stig- mas, reduced anther size. Also, anthers that are found to be dehisced and touching the stigmas before full anthesis seem to be clear adaptations to autogamy. The lack of true caulanthoid characters in the flowers of C. lasiophyllus may be related to its breeding system. In conclusion, I am convinced that the four species of Caulanthus discussed in the preceding paragraphs and compared with the remaining species of the genus and Thelypodium in Table 2 clearly should be placed in Caulanthus and not Thelypodium. The content of this table needs no elaboration as the items are self-evident. However, it is perhaps significant that the pubescence on the sepals in Caulanthus (including C. flavescens, C. anceps, and C. lasiophyllus) is often made up of very few trichomes restricted to their tips. The generic limits of Caulanthus are somewhat indefinite. Watson (1871) BIOSYSTEMATICS OF THELYPODIUM 15 segregated the genus from Streptanthus on the basis of its having terete siliques and incumbent cotyledons instead of the flattened siliques and accumbent cotyledons present in the species of the latter genus. Jepson (1925, 1936) united Caulanthus under Streptanthus because of the over- all similarities in the flower morphology of the two genera. He neglected the differences in the siliques, seeds, and cotyledonary position in rela- tion to the radicle, which are used as the basis for distinguishing the two genera. It is my strong feeling that the two genera should be main- tained, and I fully agree with Rollins (1971) that “it is in the interest of a reasonable and workable classification to accept both Caulanthus and Streptanthus.” Caulanthus seems to be made up of at least 15 species including the recently described C. divaricatus (Rollins, 1971). Four of the 18 species recognized by Payson (1923) present some problems. Three species, C. amplexicaulis, C. heterophyllus, and C. simulans, appear to be more prop- erly placed in Streptanthus than in Caulanthus because they have some- what compressed siliques, accumbent cotyledons, and a narrow or rudi- mentary wing at the distal end of the seed. Caulanthus sulfureus of Pay- son is unquestionably Brassica. I have seen the holotype of C. sulfureus and agree with both Kearney and Rollins in annotating it as B. campes- tris. The exact number of species in Streptanthus is not known, but an estimate of 30 to 35 species may not be far off the mark. Streptanthella. This monotypic genus is based on a species originally described as Arabis longirostris by Watson (1871). It was known as a Streptanthus for nearly 30 years after its transfer there by Watson (1889). Streptanthella was proposed by Rydberg in 1917 and since then the genus has been recognized by the majority of taxonomists who have dealt with its only species. Streptanthella longirostris resembles certain species of Caulanthus in its flowers, incumbent cotyledons, and the orientation of its siliques, but differs significantly in having flattened siliques that are beaked and by its winged seeds. Streptanthella resembles Streptan- thus in its flattened siliques, winged seeds, and flowers but it differs in having incumbent instead of accumbent cotyledons and its style-like beak with the distal part of the silique often remains indehiscent and contains 2-4 seeds. The treatment of Streptanthella longirostris as a species of Thelypo- dium by Jepson (1925, 1936) is not supported by present morphological evidence. The streptanthoid flowers, winged seeds, beaked siliques, and stamens that occur in three pairs of different lengths are characters not found in any species of Thelypodium. The distinctness of Streptanthella longirostris from the closely related genera Caulanthus and Streptanthus as well as from Thelypodium argues for its recognition as a monotypic genus. Pennellia. Two species of Pennellia, P. micrantha (Gray) Nieuw. and 16 IHSAN A. AL-SHEHBAZ P. longifolia (Bentham) Rollins, were originally described as species of Streptanthus and later transferred to Thelypodium by Watson (1882). Watson's concept of the generic limits of Thelypodium was so broad that some of the species he placed in this genus (1871) are currently placed in the genera Caulanthus, Iodanthus, Pennellia, and Thelypodiopsis. Robinson (1895) was the first to recognize that the two species mentioned above are anomalous in Thelypodium and placed them in a subgenus of their own, Heterothrix, which was raised to the rank of genus by Rydberg (1907). The name Heterothrix Rydberg is illegitimate because, as shown by Nieuwland (1918), it is a later homonym of Heterothrix Muell. (=Ech- ites L.) in the Apocynaceae. Nieuwland proposed the name Pennellia to replace Heterothrix of Rydberg. Rollins (1960b) showed that Thely- podium longifolium is congeneric with Pennellia micrantha. Pennellia is clearly different and readily distinguishable from Thely- podium by a number of morphological features. First, the dendritic tri- chomes that are mostly present on the lower half of the plants of Pennellia are not found in any species of Thelypodium. Most species of Thelypo- dium are glabrous and those that are pubescent have only simple tri- chomes. In addition to the prevalent dendritic trichomes, Pennellia has simple trichomes as well. Second, the inflorescence in Pennellia is defi- nitely a lax raceme and both the floral buds and flowers are small ( gen- erally less than 6 mm. long) and mostly spherical. In Thelypodium the inflorescence is racemose or corymbose and usually densely flowered but neither the buds nor the flowers are spherical (mostly oblong) and the flowers may be up to 2 cm. long. Finally, the majority of Thelypodium species have exserted stamens, coiled anthers, distinct gynophores, and the petals are differentiated into claw and blade, while the stamens in Pennellia are included, the anthers are hardly coiled, the gynophores are mostly obsolete, and the petals are not differentiated into a distinct claw and blade. In fact, the differences between the two genera are so great that Pennellia is not particularly related to Thelypodium. CLASSIFICATION OF THE PRIMITIVE CRUCIFERAE There is general agreement among those of us involved in systematic studies of the various genera of the Cruciferae that none of the proposed systems of classification above the genus level are fully adequate. For reasons to be discussed below, some of the earlier systems, such as those of De Candolle (1821b, 1824), Bentham and Hooker (1862), and Prantl (1891), and some of the more recent ones, e.g., Schulz (1936) and Jan- chen (1942), fail to provide a natural tribal classification for the primitive genera of the family. The lack of agreement between the main systems in the tribal placement of these primitive genera (Table 3) may have resulted from the use of one or a very few characters as the basis for the tribal divisions rather than grouping the genera on the basis of their relationships to each other. BIOSYSTEMATICS OF THELYPODIUM 17 Our discussion is limited to the genera placed in the tribe Thelypo- dieae by the present writer and starts from the system of De Candolle (1821b). The tribal classification of the primitive Cruciferae in the treat- ments of Endlicher (1839), Walpers (1842), Bentham and Hooker (1862), and Baillon (1874) was essentially the same as that of De Can- dolle (1821a, 1821b, 1824), the main difference being the use of different names above the tribal level by some of these authors and the addition of the genus Streptanthus to the tribe Arabideae and Warea and Thely- podium to the Sisymbriae. The cotyledonary position in relation to the radicle, accumbent versus incumbent, and the nature of the silique are the basic features on which these two tribes are defined in the system of De Candolle. A system that places Streptanthus and Macropodium in the tribe Arabideae and their relatives Stanleya, Warea, and Thelypo- dium in the tribe Sisymbrieae is obviously artificial. Stanleya and Thely- podium are improperly classified because the former genus has both accumbent and incumbent cotyledons (Rollins, 1939) while the latter has obliquely accumbent to obliquely incumbent cotyledons. Prantl (1891) differed strongly from De Candolle’s system by using the characteristics of the stigmas (capitate versus bilobed) and the tri- chomes (presence versus absence, simple versus branched) as the basis for his major divisions. He recognized 20 subtribes distributed in the four tribes Thelypodieae, Sinapeae, Schizopetaleae, and Hesperideae. The tribe Thelypodieae was defined to include glabrous plants or those with simple trichomes as well as entire or capitate stigmas. It was divided into four subtribes: Stanleyinae, Cremolobinae, Heliophilinae and Chami- rinae. Under the subtribe Stanleyinae, Prantl included the genera Nototh- laspi, Pringlea, Warea, Stanleya, Thelypodium, Caulanthus, and Streptan- thus. The system of Prantl was intended to be natural but in fact, it is no less artificial than that of De Candolle. The nature of the pubescence is hardly of value when used in classifications above the genus level. There are genera with various types of pubescence ranging from simple to branched and even peltate. The best examples are the genera Lesquer- ella (Rollins & Shaw, 1973) and Arabis (Rollins, 1941c). The use of tri- chomes as the basis for tribal classification led to the separation of the genus Macropodium to a different tribe (Hesperideae) from its relatives found in the subtribe Stanleyinae-Thelypodieae. The tribe Thelypodieae, as defined by Prantl, contained an extremely heterogeneous assemblage of genera that show a wide range of diversity in fruit morphology. Only five of the 17 genera placed in Thelypodieae are distributed in the Northern Hemisphere. They are the last five of the subtribe Stanleyinae (Table 3). Prantl believed that the most primitive forms of the Cruciferae are found in the Southern Hemisphere, but we believe this is not the case. All ten genera of the tribe Thelypodieae, as recognized by the present writer, have a Northern Hemisphere distribu- Endlicher, 1839 Subordo I Pleurorhizeae Tribu Streptanthus Macropodium Subordo IT TABLE 3, TRIBAL CLASSIFICATION OF THE PRIMITIVE CRUCIFERAE Prantl, 1891 Tribe I. Thelypodieae 1. Subtribe tanleya Thelypodium Caulanthus Streptanthus 2. Subtribe Cremolobinae 3. Subtribe Heliophilinae 4. Subtribe Chamirinae Tribe IV. Hesperideae ubtribe Turritinae Macropodium Hayek, 1911 Schulz, 1936 Tribe I. Tribe II. Stanleyeae Thelypodieae Stanleya 1. Stanleya Warea © Wares Chlorocrambe 3M di Macropodi So Wigderebdis seeinceiess Tribe III. Romanschulzieae 4. Streptanthus Romanschulzia 5. Euclisia Tribe IV. Streptantheae 6. Microsemia Subtribe hae reptanthus alpina ge Streptanthella ‘ Disccanthus 9. Thelypodium Cartiera Agianthus Tribe VI. Mitophyllum Schizopetaleae Microsemia ribe lisi Schizopetalinae Stanfordia Stanfordia Tribe XVI. Arabideae (29 genera ) Pleurophragma Tribe XVII. Tribe XVIII. Pleiocardia Subtribe Caulanthinae Matthioleae (17 gen.) Hesperideae (20 gen.) helypodium Thelypodiopsis Tribe XIX. mia (74 gen.) Subtribe Sisymbrinae M ceealtueniciem Stanleyella Present Treatment Tribe Thelypodieae . Stanl Romanschulzia Thelypodiopsis Streptanthus Caulanthus Streptanthella CHINA Owe _ — — (ee) ZVdGHAHS-TV “V NVSHI BIOSYSTEMATICS OF THELYPODIUM 19 tion. The most primitive known Cruciferae are among these ten genera (Table 3). Pringlea and Notothlaspi are neither related to each other nor to other members of the subtribe Stanleyinae. Although Pringlea possesses certain features that are present in the truly primitive Cruci- ferae (such as the exserted stamens and long racemes), these features are very likely to have originated independently. The reduction in petal size, the exserted stamens, and the long and densely flowered racemes are perhaps adaptations to wind pollination which is not known else- where in the Cruciferae. The lack of the septum might have been an- other reason for overestimating the primitiveness of Pringlea. However, this feature evidently has evolved several times, independently, within the Cruciferae. All the truly primitive genera have a complete septum. In my opinion, the features of the septum, inflorescence, and flowers in Pringlea are derived and the genus is not related at all to those that I place in the tribe Thelypodieae. Prantl listed Stanfordia Watson with 17 other genera of uncertain posi- tion at the end of his treatment. He indicated that it possibly belongs to the tribe Schizopetaleae as he apparently followed Watson (1880) in associating the genus with Tropidocarpum, which is often placed in this tribe. However, we believe that Stanfordia californica is no more than a good species of Caulanthus. Stanfordia was based primarily on the pres- ence of unusual trifid cotyledons, a feature that Watson overlooked in a species that he described as Caulanthus coulteri. Robinson (1895) presented the best tribal classification of the primi- tive Cruciferae of North America. He placed the genera in the tribe Stanleyeae, a name that was used on the tribal level for the first time, instead of the earlier published Thelypodieae of Prantl (1891). Stanley- eae was proposed first by Nuttall (1834) as a family including the genera Stanleya and Warea. The tribe Stanleyeae, as treated by Robinson, was a truly natural assemblage of related genera. However, he maintained Stanfordia and listed it between Streptanthus and Caulanthus. In 1911 Hayek presented a system of classification for the whole fam- ily. He adopted the tribal name Thelypodieae to include, with the ex- ception of Schoenocrambe, all the known primitive genera of the family. However, he probably overlooked the work of Rydberg (1907) since he did not mention any of the segregates of Thelypodium, or Chlorocrambe. The placement of Schoenocrambe with the primitive crucifers is not justified for reasons mentioned above. There are three other points of weakness in Hayek’s classification. First, he recognized Euclisia and Mi- crosemia that were segregated from Streptanthus by Greene (1904). We believe that these are not worth recognition. Second, he derived Caulan- thus and Streptanthus from totally different ancestors. Caulanthus and Thelypodium were said to be derived from Schoenocrambe while Strep- tanthus was derived from Warea and both Warea and Schoenocrambe 20 IHSAN A. AL-SHEHBAZ were derived directly from Stanleya. Caulanthus and Streptanthus are so similar to each other in their floral morphology that their generic limits are often confused and the boundaries between them are hard to define. It is difficult to visualize the independent origin of such remarkable flowers from ancestors that are extremely different from each other and from both Streptanthus and Caulanthus. Third, he placed Stanfordia, a genus that we believe to be at home in Caulanthus, in the tribe Schizo- petaleae. It is probable that Hayek overlooked Robinson’s (1895) treat- ment, since the latter author was the first to associate Stanfordia with Caulanthus. One significant improvement of Hayek’s system over that of Prantl (1891) is the removal of Notothlaspi and Pringlea to the tribes Schizopetaleae and Pringleae respectively. It is beyond the scope of the present study to evaluate the tribal position of these two genera, but we certainly agree that they are not related to the primitive crucifers. In 1936 Otto E. Schulz published a comprehensive monograph of the family Cruciferae that includes a number of strong as well as weak points. The detailed descriptions of the tribes and genera supported by illustra- tions and keys are undoubtedly very useful. However, Schulz has often been criticized for the large number of genera and tribes he recognized and for the placement of certain closely related genera in widely sep- arated tribes. There is no doubt his treatment of the primitive Cruciferae, both on the tribal and generic levels, created more problems and con- troversy than any other system of classification extant. Our evaluation of his system with regard to these crucifers is presented in the following paragraphs. as T. milleflorum, T. eucosmum, and T. laciniatum, have distinctly ex- serted stamens and usually long gynophores that can be longer than those of Chlorocrambe. Had Schulz looked at these features carefully in these three species of Thelypodium, he would have found that the genus is at home with the other four genera he placed in the Stanleyeae. How- ever, he placed Thelypodium in the tribe Hesperideae and this led him to choose a tribal name other than Thelypodieae for the four genera mentioned above. The fact that both Thelypodium and Stanleya should be placed in the same tribe has been elegantly pointed out by Rollins (1939) in the following words: “O. E. Schulz (1936) has placed Stan- leya and Thelypodium in widely separated tribes of the Cruciferae, but on the basis of observed facts such a disposition is decidedly unwar- BIOSYSTEMATICS OF THELYPODIUM 21 ranted. Insofar as these genera are concerned, the retention of the tribe Thelypodieae of Prantl (1891) as modified by von Hayek (1911) more nearly presents the actual phylogenetic relationships in this group than does the system of O. E. Schulz.” Schulz maintained Thelypodiopsis and placed it with Thelypodium in the tribe Hesperideae. However, neither the generic limits of Thelypodiopsis nor its tribal disposition were cor- rectly treated. Thelypodiopsis is to be associated with its nearest rela- tives which we place in the tribe Thelypodieae. The genus Romanschulzia was given a tribe of its own, Romanschulz- ieae, by Schulz, which was maintained by Janchen (1942). The char- acters used in defining this tribe are primarily the spreading nature of the floral parts and the insertion of the filament bases in distinct depres- sions on the receptable. Romanschulzia is not the only primitive genus with spreading floral parts. Five species of Thelypodium, Caulanthus anceps, and all the species of Stanleya and Warea have spreading parts. Therefore, this character has little value in establishing a tribe for Ro- manschulzia. The filaments in this genus are distinctly dilated at the base and mostly are surrounded by glandular tissue. Upon the removal of the filaments, the glandular tissue appears with six depressions corresponding to the insertion points of the filaments. This feature alone is not strong enough to be used as the sole basis for establishing an independent tribe. Rather, Romanschulzia fits very well in the tribe Thelypodieae because it shares a number of features of the flowers, fruits, and inflores- cences with other members of the tribe, particularly Thelypodium. The evidence supports the conclusion that the tribe Romanschulzieae does not deserve recognition. Such a conclusion was reached earlier by Rollins (1942b) in his treatment of Romanschulzia. He stated “I cannot agree with O. E. Schulz (1936) that this genus represents a separate tribe in the family. Rather it should be placed in the same tribe with Thely- podium.” The characters Schulz used to define the tribe Streptantheae are the diagonal or oblique receptacle, the campanulate or + bilabiate calyx, and the undulate or twisted petals. The claim that the receptacle is horizontal in all the Cruciferae but the Streptantheae does not seem to be realistic. It is true that Streptanthus (sensu lato) and Caulanthus (sensu lato) possess certain features in the petals and the calyx that are uncommon or lacking in the other primitive Cruciferae, but these differ- ences are not strong enough to warrant putting these genera into a tribe of their own. The relationship of Caulanthus and Streptanthus to Thely- podium, Chlorocrambe, and Thelypodiopsis is fairly clear even though certain species of the former two genera may seem remote when com- pared with some species of Thelypodium and Thelypodiopsis. There are species in Caulanthus and Streptanthus that lack some features, such as the twisted or crisped petals and the campanulate or urceolate calyx, 22, IHSAN A. AL-SHEHBAZ that were considered significant in defining the Streptantheae by Schulz. He divided the Streptantheae into two subtribes the first of which, Eu- klisiinae, was characterized by its entire stigmas and siliques that are flattened parallel to the septum, while the subtribe Caulanthinae was said to have bilobed stigmas and siliques flattened against the replum. We believe that these subtribes are neither useful nor are they based on accurately determined characters. In addition to Streptanthus and Strep- tanthella, Schulz recognized eight other genera in his subtribe Euklisi- inae (Table 3). These were segregated by Greene (1904, 1906a, 1906b ) from Streptanthus on the basis of minor differences in the flowers and siliques. It is clear that these segregates of Streptanthus are not worth recognition since the genus otherwise is made up of a natural assembly of species. The genus Guillenia was erected by Greene (1906a) to include spe- cies with reflexed siliques that were previously placed in Thelypodium. The two species of Guillenia recognized by both Greene and Schulz are G. cooperi and G. flavescens, both of which are better accommodated in Caulanthus, as pointed out above. However, not only did Schulz over- look the very characteristic streptanthoid flowers of these two species, but he placed Guillenia in the tribe Arabideae. The first three species listed by Greene (1906b) under Guillenia seem to represent one poly- morphic species that we refer to as Caulanthus lasiophyllus. This species is certainly the type of the genus because it was the first one listed by Greene who stated (1904): “In arranging the sequence of species my custom is to place those first which seem to have the clearest claim to represent a genus; and therefore these stand as its type.” The unfor- tunate thing is that Schulz maintained Guillenia even after he removed its type species to a different genus (Microsisymbrium) which he put in a different tribe, Sisymbrieae. If Guillenia were to be recognized, it ought to be placed in the tribe containing Caulanthus. However, the evidence from the morphology and cytology of the group strongly sug- gests that the streptanthoid genera belong to the tribe Thelypodieae. Perhaps the best example showing the artificiality of Schulz’s system is found in Thelypodium. The generic limits of this genus, as recognized by the present writer, include the segregate genera Pleurophragma and Stanleyella. The two segregates cannot be maintained for reasons dis- cussed earlier. However, Schulz placed Pleurophragma in the tribe Ara- bideae, Stanleyella in the Sisymbrieae, and Thelypodium in the Hesperi- deae. The characters that Schulz used to distinguish these three tribes are the accumbent versus incumbent cotyledons, open versus closed ca- lyx, and acute versus obtuse anthers. I find it very difficult to rely on these characters alone in tribal delimitations. It is unfortunate to find that the most comprehensive monograph of the Cruciferae is the least satisfactory with respect to the tribal classifi- BIOSYSTEMATICS OF THELYPODIUM 93 cation. The use of a single character in tribal delimitation within the Cruciferae inevitably leads to an artificial classification no matter what character is used. In my opinion, the tribes should be defined on the basis of a number of characters that do not always have to be of a discontinuous nature. The tribal disposition of a given genus should be based on asso- ciation with its nearest relatives and how they all fit in that particular tribe. It is true that the genera of the Cruciferae are generally difficult to determine and that their generic boundaries can sometimes be artificial, but the tribes are by no means easier to define and their boundaries may be even more artificial. In 1942 Erwin Janchen presented a modification of Schulz’s classifi- cation on the tribal and the subtribal levels only. He recognized the tribes Stanleyeae, Romanschulzieae, and Streptantheae which we believe to constitute the single tribe Thelypodieae. He placed the genera Thely- podium and Thelypodiopsis in a subtribe of their own, Thelypodiinae, which he put under the Sisymbrieae. However, such a disposition cannot be accepted since these two genera are more related to those of the three tribes mentioned above than to the members of the Sisymbrieae. Dvorak (1971b) maintained Thelypodium in the subtribe Thelypodi- inae as presented by Janchen (1942). He derived it from a questionable group with incumbent cotyledons and a basic chromosome number of x=6 and indicated that the basic chromosome number in the genus is x=24. However, this number is not found in any species of the genus. It is almost certainly based on a single count made by Snow (1959) from Caulanthus lasiophyllus (2n=48) which was listed as a species of Thely- podium. Our chromosome counts in the genus, as well as those made by Rollins (1966), clearly indicate that a fundamental haploid number of n=13 characterizes the genus. Present classification. The tribe Thelypodieae, as defined in the present treatment, is made up of a natural group of ten genera, most of which occur in North America. Some of these genera are presumed to be the most primitive known members of the Cruciferae. It is generally sup- posed that the tribe Thelypodieae might represent a link between the rest of the family and its presumed ancestors which are thought to be something like members of the subfamily Cleomoideae-Capparaceae. However, it is beyond the scope of the present paper to speculate on how the other crucifers and their tribes are connected to the Thelypo- dieae. It is not deemed even advisable to draw direct “phylogenetic” lines between the various genera of the tribe. The genera Stanelya and Warea are perhaps more related to each other than to the remainder of the tribe. However, the latter genus is morphologically somewhat isolated and does not seem to tie in with the other genera as does Stanleya. A clear relationship appears to be present between Thelypodium and Stanleya (Rollins, 1939). The relationship be- 24 IHSAN A. AL-SHEHBAZ tween the latter genus and Thelypodiopsis is not entirely clear. The flowers of T. aurea resemble those of Stanleya in their color and the shape and the orientation of their petals and sepals, but there the similarities cease. Therefore, the relationships between these two genera appear to be loose and indirect. Certain species of Thelypodium resemble in their flowers and inflorescences some species of Thelypodiopsis. This is par- ticularly evident when Thelypodium sagittatum and Thelypodiopsis ele- gans are compared. In fact, the two species are sufficiently similar to each other so that the latter is often misidentified as T. sagittatum. However, this poses a problem with respect to the relationship between the two genera, since both species appear to be advanced in their respective genera. The genus Thelypodium seems to be somewhat related to Roman- schulzia (Rollins, 1956). Such a relationship becomes evident when spe- cies with spreading floral parts of the former genus are used in the com- parison. Romanschulzia seems to be somewhat advanced because of its small flowers and calyptra-like calyx, but we cannot say much about the genus without having a more detailed knowledge of its individual species. Romanschulzia apetala is somewhat similar to species of Macro- podium in its siliques, racemose inflorescence, and long gynophores, but the two genera are distinct in their floral morphology. Macropodium resembles Thelypodium in its flowers and inflorescences more than it does other genera of the tribe. This is particularly true if T. laciniatum, T. milleflorum, and T. brachycarpum are used for the comparison. Although the first species of Thelypodium was originally described as Macropo- dium by Hooker (1830a), it should be understood that the two genera are very different in their siliques, seeds, and pubescence and I only associate them loosely. The relationship between Chlorocrambe and Thelypodium was clearly indicated by Payson (1923). The similarities between C. hastata and T. laciniatum and T. milleflorum in the inflorescence, petiolate leaves, some aspects of the flowers (excluding the petals) and the pedicels are fairly clear. The former genus also resembles the three streptanthoid genera Streptanthus, Caulanthus, and Streptanthella in the shape of its petals, as they have broad and obovate claws, and in the broad siliques that re- semble some species of Caulanthus. However, Chlorocrambe is different from any of these three genera in its toothed petals, long and distinctly exserted stamens that are somewhat equal in length, and by its distinct gynophore. The three streptanthoid genera are very clearly related to each other. The characteristic aspect of the flowers is not shared by the other genera of the tribe. The tribe Thelypodieae is not defined by a single character. Rather a combination of characters are utilized in its definition. The following is a brief description of the tribe: flower parts erect to spreading; petals BIOSYSTEMATICS OF THELYPODIUM 25 differentiated into claw and blade or not, crisped, chanelled, or not so; stamens exserted to included, equal in length to somewhat tetradyna- mous or in three pairs of different lengths; anthers sagittate at base, usually coiled after dehiscence; glandular tissue mostly low and flat, usually subtending the base of filaments; siliques usually gynophorate, terete or flattened parallel to the septum; stigmas sessile or on a distinct style, entire or bilobed; seeds winged to wingless; cotyledons accumbent to incumbent. The number of species known in the ten genera of the Thelypodieae are distributed as follows: Stanleya (6 species), Warea (4 species), Thely- podium (18 species), Macropodium (2 species ), Chlorocrambe (1 spe- cies), Romanschulzia (11 species), Thelypodiopsis (6 species), Strep- tanthus (approximately 30-35 species), Caulanthus (15 species) and Streptanthella (1 species ). KEY TO THE GENERA OF THE TRIBE THELYPODIEAE A. Stamens distinctly exserted, equal in length or nearly so, filaments at least 2-4 mm. above the sepals and/or petals; siliques usually long stipitate; gynophores rarely 1-2 mm. long; stigmas usually sessile. B. Siliques 3-6 e; seeds winged; plants of Siberia and temperate eastern MR St ep ues aie eye beatae Poe ea, es acropodium. B. Siliques 1-2(3) mm. wide; seeds wingless (Chlorocrambe sometimes with a be pe distal edge ); North America Sepals widely spreading to reflexed; claws and/or filaments usually a or tuberculate near the base; gynophores 1-3 cm. long (less than 1 cm. *PP- of Warea). Flowers yellow to creamy white; inflorescence racemose; biennials or pace. gas claws broader at base than above; western United States ...... D. Flowers white to purple; carey corymbose; annuals; claws ee throughout; southeastern United States ...................--..--. Warea. C. Sepals erect; claws and filaments sibel gynophores mostly less than 1 cm. lon 7.) = E. cess yellow-green; petals dentate between claw and blade, claws obovate, blades narrower than claws; stigmas nearly sessile; siliques 2-3 mm. wide; eh te Oren: Tes Sits a Chlorocrambe. E. Flowers white to purple; petals entire, claws slender or somewhat broader at the base, blades as wide or wider than claws; . — sag ‘ate 5(2) distal end of it; phd fectenad parallel to sep G. Cotyledons accumbent or nearly so; rn present; valves separating fro the a along their entire length; southwestern and western — States and n i WRK i Streptanthus. G. Cations incumbent; siliques attenuating into ve beak, valves separat- ing most of their length, indehiscent above, usually with 2-4 seeds remaining in beak portion; western United States ...............----...-- Streptanthella. F. gi wingless; —— mostly terete, or if —— _ floral parts spreading. H. Calyx urceolate to campanulate; petals crisped or chanelled, claws as broad or 26 IHSAN A. AL-SHEHBAZ broader than the limb, often obovate to spatulate; western United States ...... dee ee ie at alae eee wie oe cue se ie ee eae lee autantnus. H. Calyx not urceolate, sepals spreading to erect; petals entire, rarely crisped (a few spp. of Thelypodium ), claws narrower I. Sepals usually shed as flowers open; median glandular tissue surrounding the base of paired filaments; Central America to southern Mexico .... Romanschulzia. . Sepals remaining attac and usually shed after anthesis; median glandular tissue subtending paired stamens or absent; northern Mexico and western United States. J. Stigmas bilobed, lobes opposite the replum J. Stigmas entire. K. Siliques subsessile on a broad gynophore; pedicels stout, strongly curved, often reflexed but sometimes erect; siliques and/or pedicels usually pubescent; siaiaiis al csrtsnery? Hh csi ng TRACE ee oe eH ee Pee au K. Siliques gynophorate, or if subsessile, then gynophore narrower than the fruit width, pedicels slender or stout, straight, or very slightly curved, rarely reflexed; siliques and pedicels glabrous; stems solid Thelypodiu IRE Es acea ed Thelypodiopsis. There are a few genera needing some mention because they have been either associated with or compared to Thelypodium. Glaucocarpum is clearly distinct from Thelypodium (Rollins, 1938a) and does not seem to resemble other genera of the tribe Thelypodieae as it is presently con- stituted. Ornithocarpa is similar to both Thelypodium and Romanschulzia in its spreading parts, coiled anthers, gynophorate siliques, and filaments that are nearly equal in length, but as suggested by Rollins (1969) the genus does not seem to be phylogenetically closely related to these primi- tive genera. Ornithocarpa resembles Schizopetalum in its divided petals, nectar glands, divided leaves, and compressed fruits, but they are quite different in their stigmas, fruit shape, pubescence, and orientation of floral parts. The two genera were placed by Schulz (1936) in the tribe Schizopetaleae. It is not the purpose of the present paper to find a tribal classification for Ornithocarpa, but we feel that it does not be- long to the Thelypodieae. Pennellia presents difficulties. We have excluded it from an association with Thelypodium and the other primitive genera for reasons discussed previously. Schulz (1936) placed it immediately after Halimolobos in the subtribe Arabidopsidinae of the tribe Sisymbrieae, and the super- ficial resemblance of P. micrantha to certain species of Halimolobos have been pointed out by Rollins (1943). The relationship of P. micrantha and P. longifolia to the genus Arabis was recognized by Asa Gray (1864). The two species were then known under Streptanthus, but were ex- cluded from that genus by Gray (p. 187) who suggested that they “may probably be referred to Arabis.” Furthermore, Rollins (1941c) pointed out the remarkable resemblance between P. micrantha and A. tricor- nuta in their flowers, inflorescences, and general habit. All these resem- blances are not accidental. It is very likely that Pennellia is more readily accommodated in the Arabideae than in any other tribe. BIOSYSTEMATICS OF THELYPODIUM 27 Iodanthus has been the center of a continuous controversy in terms of its tribal disposition. Prantl (1891) placed it in the subtribe Cardamini- nae of the tribe Sinapeae. Robinson (1895) placed it in the Arabideae while Hayek moved it to the subtribe Hesperidinae of the tribe Alysseae. Schulz (1936) placed Iodanthus pinnatifidus in the tribe Matthioleae and treated another species that unquestionably belongs to Iodanthus (cf. Rollins, 1942a) in a separate genus, Chaunanthus, and placed it in the tribe Sisymbrieae. Finally, Moggi (1965) maintained Iodanthus in the Hesperideae, while Dvorak (1970b) maintained it in the subtribe Cardamininae of Prantl. However, the last two authors treated Iodanthus as a monotypic genus since they both overlooked the latest treatment of the genus by Rollins (1942a). It is clear that there is no general agree- ment on the tribal disposition of the genus. The present writer has no suggestion as to where the genus should go, but it cannot be placed in the Thelypodieae since it does not appear to be related to the genera we have placed in this tribe. MorPHOLOGY Habit. All species of Thelypodium are typically herbaceous. One spe- cies, T. flexuosum, possess a distinct caudex. The majority of them are biennial, but T. texanum, T. tenue, and T. paysonii are winter annuals, while T. flexuosum and T. eucosmum are perennial. Under favorable conditions the potential of being short-lived perennials is present in T. crispum, T. paniculatum, T. repandum, and T. sagittatum where lateral crowns of basal leaves may be produced a few years in a row. Each crown is like a basal rosette of a biennial plant. Stems. The stems are simple or branched from the base in the’ ma- jority of species. They are mostly branched from the base in Thelypo- dium stenopetalum and, to a lesser extent, in T. sagittatum subsp. ovali- folium, while in T. milleflorum and T. rollinsii the stems are mostly simple to the inflorescence. In all species basal branching can occur as a result of grazing or the failure of the main stem to grow normally. Thelypo- dium flexuosum almost always has basal branching from the caudex, and in T. stenopetalum the basal branches are decumbent or subdecumbent. In one species, T. milleflorum, the stems can be hollow throughout or only to the inflorescence and the main stem is slightly to moderately in- flated. The majority of species have glabrous and somewhat glaucous stems, but in T. brachycarpum, T. crispum, T. howellii, T. sagittatum, and T. laxiflorum the stems can be glabrous or pubescent near the base within the same population or in different populations. In T. paysonii the stems appear to be consistently hirsute throughout most of their length. The trichomes in the genus are simple and usually spreading or retrorse. 28 IHSAN A. AL-SHEHBAZ T. Jaxiflorum $ sm Tmilleflorum I. ee T. laciniatum TJ. paysonii Puarte 2. Basal lea of Thelypodium with petiolate cauline leaves BIOSYSTEMATICS OF THELYPODIUM 29 none T-rollinsi! T-paniculatum Lge T. integrifolium $ 9 T- crispum deae ae MERAY T. stenopetalum T-eucosmum 9 T. flexuosum PLATE 3. Basal leaves of species of Thelypodium with sessile cauline leaves 30 IHSAN A. AL-SHEHBAZ The range of stem length between the various species of Thelypodium can be remarkable. In T. texanum, T. paysonii, T. repandum, and T. flex- uosum the stems rarely reach 7 dm. in length and in some of these and in T. crispum they can be as short as one decimeter. The maximum stem lengths are reached in T. wrightii, T. integrifolium subsp. affine and subsp. longicarpum, T. milleflorum, and T. laxiflorum, where they may exceed two meters in length. In the first two species I have seen plants close to three meters high. These are among the tallest known herbaceous cru- cifers. Basal leaves. The basal leaves of Thelypodium are usually thickish ( dis- tinctly fleshy in T. repandum) and always petiolate. They are glabrous in all species but the six having pubescent stems. Basal leaves can be quite variable in shape, ranging from lanceolate to oblanceolate, spatu- late, oblong, ovate, and obovate (Plates 2 & 3). They are typically pin- natifid or lyrate in eight species, entire in six, and variously dentate, repand, and shallowly lobed in three. In T. howellii they are nearly entire in subsp. spectabilis and usually lyrate in subsp. howellii. The petioles are glabrous in eight species, glabrous or pubescent in three, apparently always pubescent in T. paysonii, and ciliate near the base in at least six species. The basal leaves are poorly known in most species of Thelypodium be- cause they are mostly biennials where the rosette leaves wither when the flowering stems are produced. The rosettes are seldom collected in- dependently. They are taxonomically very useful, particularly in distin- guishing between certain related species, such as T. laciniatum and T. milleflorum (Plate 2) and T. integrifolium and T. rollinsii (Plate 3). Leaf margins can be quite variable within the same population in species having pinnately lobed leaves (Plate 4). Cauline leaves. As recognized by Payson (1923), species of Thelypo- dium can be divided into three groups according to their cauline leaves. In eight species, the cauline leaves are distinctly petioled; they include T. laciniatum, T. wrightii, and their relatives. Here the cauline leaves are morphologically the same as the basal leaves but they gradually be- come smaller, narrower, shorter petioled, and usually less lobed toward the upper parts of-the plant. In T. integrifolium and its five subspecies, the cauline leaves are sessile, entire, and not auriculate. Minute auricles are very rare in this species, being found in only eight plants out of more than one thousand studied. In the remaining nine species, the leaves are auriculate, sagittate, or amplexicaul. In T. rollinsii the leaves are ap- pressed to the stem (Plate 17). This feature is occasionally present in T. crispum and T. brachycarpum but is not found elsewhere in the genus. The auricles can be quite variable in size within a given species, but they are characteristically small in T. rollinsii while in T. sagittatum and T. howellii they may exceed two centimeters in length. BIOSYSTEMATICS OF THELYPODIUM 31 Puate 4. Variation in basal leaves of Thelypodium laciniatum (68145 and 68152) and T. crispum (68116 and 68130). ie IHSAN A. AL-SHEHBAZ Inflorescences. As in the majority of the Cruciferae, the inflorescence in Thelypodium is an ebracteate raceme or corymb terminating the main stem and the lateral branches. In six species (T. paniculatum, T. flexuo- sum, T. laxiflorum, T. paysonii, T. rollinsii, and T. integrifolium) the inflorescence is typically corymbose, elongating in fruit in the first four species but usually not so in T. rollinsii. The inflorescence of T. integri- folium may or may not elongate in fruit and it is a short raceme or it may be nearly corymbose in subsp. gracilipes. Eight species have a typical raceme and in five of these. (T. laciniatum, T. milleflorum, T. crispum, T. brachycarpum, and T. eucosmum) it is densely flowered. However, lax inflorescences are occasionally found in the last three spe- cies mentioned. The inflorescences of T. laciniatum and T. milleflorum may reach a length of slightly more than one meter, but in the majority of species it is only a few centimeters long. In T. howellii, T. steno- petalum, and T. repandum the inflorescence is characteristically a lax ra- ceme, while in the remaining four species it is usually corymbose to short racemose. The lowest flowers of the inflorescence may or may not arise from the axils of the uppermost cauline leaves, and this feature does not have any taxonomic value in Thelypodium. Floral buds. The floral buds are oblong or ovate in the majority of species, but they are characteristically oblong-linear in Thelypodium steno- petalum and T. eucosmum. This serves as an excellent character to distinguish these two species from the rest of the genus. Flowers. The flowers of Thelypodium show a substantial amount of variation in their size, color, and shape and in the orientation of the various floral parts. The largest flowers are found in T. laciniatum, T. milleflorum, and T. howellii, where they may reach 1.5 to 2.0 centimeters in length, while T. repandum is characterized by the smallest flowers, shorter than half a centimeter. Flower color ranges from white to dark purple but not yellow. It can be uniform in a given species or variable within the same population. The color in T. eucosmum is dark or red purple, while in T. milleflorum and T. brachycarpum it is always white. Various shades from white to purple are known in T. laciniatum, T. sagittatum, T. integrifolium, and T. rollinsii and these can be found within the same population or between populations. The remaining spe- cies of the genus are predominantly white or lavender-flowered. On the basis of orientation of the floral parts to each other, two types of flowers can be recognized within Thelypodium. The first type, present in the majority of species, is characterized by having erect sepals, petals, and stamens. These parts can be erect throughout their length or along their lower half. The stamens are usually spreading or ascending above when they are exserted and clearly erect when included. This type char- acterizes species such as T. laciniatum, T. eucosmum, T. rollinsii, T. steno- petalum, T. sagittatum and their relatives (Plate 5). The second type, where the sepals, petals, and stamens are spreading from their bases, is BIOSYSTEMATICS OF THELYPODIUM 33 present in T. wrightii, T. texanum, T. repandum, T. tenue, and less clearly so in T. paysonii (Plate 5). The sepals of T. wrightii are some- times reflexed, a feature not found elsewhere in the genus. The stamens, on the other hand, are usually spreading to an angle of 45 degrees and ATE 5. Flowers of selected species of a ena ~~ 1. Flower of T. sagittatum, side view (4x). Fic. 2. Flower of T. texanum, side view ( . 3. Flowers of T. wrightii, side view (3x). Fic. 4. Flower of T. eucosmum, side view (Tx), son 5. Flower of T. texanum, top view (5x ). Fic. 6. Flower of T. stenopetalum, side view (12 ). 34 IHSAN A. AL-SHEHBAZ equally spaced between each other so that the single and the paired stamens do not stand out since they are all equal in length or nearly so. However, the single stamens are always opposite to the outer pair of sepals, and this serves to distinguish between the two types of stamens. The paired stamens of T. wrightii are sometimes closer to each other than to the single stamens. In T. laxiflorum, a close relative of T. wrightii, the sepals are ascending to spreading or erect, and the petals and sta- mens are erect at the base (Plate 22). Sepals. The sepals are glabrous in all species of Thelypodium. The variation in their color and orientation is discussed above. One feature worth mentioning is that the sepals are somewhat saccate and unequal at the base in T. sagittatum, T. paniculatum, T. howellii, T. flexuosum, and less clearly so in some other species. In T. laciniatum. T. mille- florum, T. eucosmum, and less clearly in T. crispum and T. brachycarpum they are saccate and nearly equal at the base, while in those species where they are spreading or reflexed the sepals are nonsaccate and equal at the base. Petals. The petals of Thelypodium show a wide range of variation in size, shape, orientation, and color (Plate 6) and therefore are often use- ful taxonomically. They may be differentiated into a distinct claw and blade in species such as T. laciniatum, T. milleflorum, T. crispum, T. brachycarpum, T. eucosmum, and T. paysonii, or the blades are attenu- ated to a narrow claw-like base, as in most of the remaining species. The blades are narrowly linear in T. stenopetalum and linear in T. cris- pum, T. brachycarpum, and T. laciniatum. The broadest, often obovate, petals are found in T. paniculatum and, to a lesser extent, in T. flexu- osum. The petals may be crisped throughout most of their blade, as in T. crispum, and T. brachycarpum, or between the claw and blade, as in T. stenopetalum and T. howellii. The petals in most of the remaining species are usually not crisped. The claws in T. laciniatum and T. mille- florum are most often thicker than the thin blades, while in T. paysonii they are covered with short gland-like trichomes near their bases. These features of the claws are very useful in distinguishing these particular species of Thelypodium from the rest. Filaments. The filaments of Thelypodium show two interesting fea- tures. First, they are spreading in species such as T. texanum, T. tenue, T. wrightii, T. paysonii, and T. repandum, while they are erect below in the majority of the remaining species. The orientation of the sta- mens (erect versus spreading) served as an excellent character to dis- tinguish between the closely related species T. wrightii and T. laxiflorum (Plate 22). Second, in the majority of species with exserted anthers the filaments are equal in length or nearly so. This is particularly true of T. lacini , T. milleflorum, T. eucosmum, T. texanum, and T. wrightii. On the other hand, in T. laxiflorum, T. stenopetalum, T. sagittatum, T. i m, and T. flexuosum the filaments are tetradynamous. The re- BIOSYSTEMATICS OF THELYPODIUM 35 ak Nf nA | E 6. Petal ppd of the various species of Thelypodium. Fic. 1. T. laciniatum; Fic. 2. T. rehis Fic, 3. T. um; Fic. 4. T. brachycarpum; Fic. 5. T. eucosmum; Fic. 6. T. flexuosum Pe. 7. T. illineli, Fic, < r. pie Fic. 9. T. at oe Fic. 10. T. Asan Fic rum; Rie os Hae! § i i + paysonii; Fic. 12. u 14, sag subsp. ovalifolium; Fic. 15. anic eth Fic. 16. T. wrightii; ie sf fie pa sah si tabilis; Fic. 18. T. howellii pen howellii; Fie. 19. T. repandum IHSAN A. AL-SHEHBAZ 36 ‘ i ; Fics. 0 ssp. complanatum; Fics. 1-24 may be found within ssp. integrifolium. BIOSYSTEMATICS OF THELYPODIUM 37 maining species of the genus are intermediate between the two species groups. One feature of T. paysonii, not found anywhere else among the primitive Cruciferae except in the genera Stanleya and Warea, is the presence of short trichomes on the lower parts of the filaments and claws. Anthers. The anthers in all species of Thelypodium are introrse, basi- fixed, and sagittate at the base. In a few species they are apiculate while in the rest they are not. No characters of taxonomic value are of- fered by the anthers. Where they are long and exserted, the anthers become strongly coiled after dehiscence, a feature common in the most primitive genera of the tribe Thelypodieae. On the other hand, the an- thers, in the species where they are included, remain straight or slightly curved after dehiscence. Glandular tissue. The nectar glands of the family Cruciferae received considerable attention from a number of investigators around the turn of the 20th century, or earlier, starting with the work of Hildebrand (1879). Bayer (1905) was the first to elaborate a classification system for the family based primarily on the characteristics of the nectar glands. In con- nection with Thelypodium, Payson (1923) relied heavily on them to dis- tinguish three out of four “species” related to T. integrifolium. As in other aspects of the plant, nectar glands are fairly constant in certain species and extremely variable in others. Similar observations were made by Snogerup (1967) in the genus Erysimum. Two basic types of glan- dular tissues are present in Thelypodium. The first is characterized by the presence of lateral glands only, which are developed more above the base of single stamens than beneath. They are usually flat and low, but can be tooth-like in most populations of T. integrifolium. Other species that have lateral glands only are: T. crispum, T. brachycarpum, T. flexuosum, and most populations of T. howellii subsp. howellii and T. sagittatum subsp. sagittatum. The second type is a continuous disc sub- tending the bases of paired filaments and surrounding or subtending the bases of single filaments. This type is characteristic of the remaining taxa of the genus. Thelypodium integrifolium is the most variable species of the genus in its nectar glands. As shown in Plate 7, the glands vary considerably in terms of being flat or tooth-like and whether the “teeth” are united or separated. However, it is not possible to distinguish between the various subspecies of T. integrifolium on the basis of their glands in the way that Payson did (1923). Fruiting pedicels. Characters of the fruiting pedicels, such as length, orientation, stoutness, and the nature of their bases, are very useful in distinguishing between the various species of Thelypodium. The shortest pedicels in the genus are found in T. brachycarpum, where they hardly exceed 1-2 mm. in length, while in T. tenue they can be as long as 2-3.5 cm. Pedicel orientation is usually a conservative feature in most species, but it may vary from divaricately ascending to divaricate or variously 38 IHSAN A. AL-SHEHBAZ reflexed in species such as T. wrightii, T. laxiflorum, and, to a lesser ex- tent, T. integrifolium subsp. longicarpum. Pedicels are horizontally at- tached to the rachis in the majority of species, but they are erect or nearly so and usually appressed to the rachis in T. crispum, while in T. howellii they are mostly ascending. In most species they are straight or slightly curved, but in T. milleflorum they are strongly curved up, usually with an erect tip. Pedicel length may show considerable variation within a given species. This is particularly true in T. sagittatum, T. lacin- iatum, and T. integrifolium. In most species, the lowest pedicels of the infructescence are longer than the uppermost ones. Siliques. As stated by Rollins (1941c), the siliques are extremely valu- able in the identification of genera and species in the Cruciferae. They are very useful in distinguishing species or species groups within Thely- podium and, more important, in separating this genus from related gen- era. Payson (1923) depended heavily on the cellular pattern of the sep- tum in defining the generic limits of Thelypodium. He excluded T. wrightii from the genus and accepted its disposition in a genus of its own, Stanleyella Rydberg, because it was said to lack the central band of elongated cells and the lateral tortuose cells that characterize the septum of Thelypodium. The other characters on which Stanleyella was based were evaluated in the section on generic relationships. These al- leged differences in the septum are not consistent. In my opinion, the cellular patterns of the septum have a very limited value, if any, at the various taxonomic levels within the tribe Thelypodieae. Dvorak (1970a) used them in distinguishing species groups with Malcolmia and as a criterion of evolutionary relationships within the Cruciferae (Dvorak, 1971a), but Stork (1972) questioned their value in Malcolmia, since she found significant variation within a given taxon. With the exception of Thelypodium paysonii, the siliques are torulose throughout the genus. In T. laciniatum, T. milleflorum, T. wrightii, T. repandum, T. texanum, and T. paysonii they are somewhat flattened parallel to the septum, while in the remaining species they are clearly terete. Mature fruits of T. tenue are not known. Silique length has a very limited taxonomic value in this genus. A maximum length (10-12 cm.) is reached in T. laciniatum and T. milleflorum and minimum lengths (0.6-1.5 cm.) are found in T. crispum, T. brachycarpum, and T. flexuo- sum. Silique width is only useful in distinguishing between the two closely related species T. paniculatum and T. sagittatum. Styles and stigmas. The style in most species of Thelypodium is slen- der and usually narrows from the base to the stigma. In some species, such as T. wrightii, T. laxiflorum, T. repandum, and occasionally in T. laciniatum and T. milleflorum the style is subclavate. The stigmas are entire and almost always equal to or smaller than the diameter of the style. Although these features of the style and stigma have little or no BIOSYSTEMATICS OF THELYPODIUM 39 significance on the species level, they are most useful in distinguishing between Thelypodium and the related genus Thelypodiopsis (Plate 1). Gynophore. All species of Thelypodium have a distinct gynophore that is always narrower than the fruit itself. It can be as short as 4 mm. or as long as 8.5 mm. In T. paniculatum, T. flexuosum, T. laxiflorum, T. texanum, and T. repandum the gynophore hardly excedes one millimeter in length, while in T. milleflorum, T. laciniatum, and T. eucosmum it is usually longer than 2 mm. Gynophore length is usually quite variable within the same population. This is particularly true in species where it is most frequently longer than 1 mm. (Plate 8). Pate 8. Variation in gynophore length of Thelypodium laciniatum, Scale in millimeters. 40 IHSAN A. AL-SHEHBAZ PLaTEe 9. SEM photomicrographs of seeds of Thelypodium and Stanleya. Fics. 1-4. T. texanum, Barclay 713; Fies. 5—6. T. laciniatum, Al-Shehbaz 6922: Fics. 7-8. Stanleya pinnata, Al-Shehbaz 6986. Magnifications: 1. 25x; 2. 49x; 3. 98x; 4. 245 «: 5. 37™&: 6. 98x; 7. 100 x; 8. 250x. > BIOSYSTEMATICS OF THELYPODIUM 41 Seeds. The seeds in all species of Thelypodium are wingless, uniser- iate, faintly reticulate, non-mucilaginous, oblong to ovate, and brownis in color. The cotyledons are obliquely accumbent to obliquely incumbent. However, one occasionally finds accumbent or incumbent cotyledons in species that have flattened or terete siliques respectively. The percentages of these two cotyledonary positions in a given seed sample range be- tween 1 to 10% in the majority of the species. As stated by Payson (1923), seed morphology does not contribute much to the taxonomy of Thelypodium. The differences in seed dimen- sions, if present, are so slight as to be insignificant. Seeds of T. panicula- tum and T. brachycarpum are decidedly plump, while their respective relatives, T. sagittatum and T. crispum, have somewhat flattened seeds. Seed sculpture is uniform throughout Thelypodium. According to the terminology used by Murley (1951), the seed surface is minutely reticu- late. Seeds of 16 of the 18 species of the genus were examined under the scanning electron microscope. The sculpture (Plate 9) is mostly made up of hexagonal or pentagonal units that are not arranged in any particular fashion. However, those units above the radicle are tetragonal and arranged in lines parallel to its long axis. Seeds of Thelypodium are fairly similar to those of the related genus Stanleya, the main differences being the presence of pustulae (one in the center of each unit) and the angulate nature of the units in the former genus, and their absence from the single species studied of the latter (Plate 9). DistTriBuTION, EcoLocy, AND ECONOMIC IMPORTANCE Distribution. The genus Thelypodium occurs primarily in the western part of the United States. Two of its species, T. laciniatum and T. mille- florum, are also present in Canada, but these are known from old col- lections made from southern British Columbia (Henry, 1915). One spe- cies, T. paysonii, is endemic to northern Mexico and appears to be re- stricted to southern Coahuila and adjacent Durango. However, very few collections of this species were available for study and more are needed before a clear picture of its distribution can be presented. The only spe- cies found in both Mexico and the United States is T. wrightii. Within the United States, Thelypodium is distributed in an area de- fined on the east by an imaginary line extending from central North Dakota southward through western South Dakota, western Nebraska, eastern Colorado, northwestern Oklahoma, eastern New Mexico, and southwestern Texas. On the west, the area is defined by the eastern slopes of the Cascade Mountains in Washington and Oregon and the eastern slopes of the Sierra Nevada Mountains and San Bernardino Mountains in California. Seventeen of the twenty-five taxa recognized in the present treatment occur in the Great Basin, with the majority of them (twelve) found in Nevada and Utah. The most widely distributed species are T. 42 IHSAN A. AL-SHEHBAZ integrifolium, T. sagittatum, T. milleflorum, and T. laciniatum (Maps 1, 3, and 4). The most restricted species are known from one or very few counties. Thelypodium repandum is known only from Custer County, Idaho, while T. stenopetalum is endemic to Bear Valley in San Bernar- dino County, California. One remarkable species, T. tenue, is known only from the type collection made in Presidio County, Texas. Finally, three species, T. eucosmum, T. texanum, and T. rollinsii, are known from very few counties in Oregon, Texas, and Utah respectively. Thelypodium texa- num occurs in a number of localities in Brewster County, but has not been collected, to my knowledge, from the neighboring areas of northern Mexico. Altitudinal ranges. There is a wide range of variation in the altitudinal limits reached by the various species of Thelypodium. Species that are narrow endemics show very little variation in altitude, but widely dis- tributed species, such as T. laciniatum, T. integrifolium, T. wrightii, T. milleflorum, and T. crispum, have maximum and minimum limits that differ by 6,000 feet or more. The lowest elevation (100 to 500 feet) is reached by T. laciniatum and T. milleflorum, while T. laxiflorum, T. sagit- tatum, and T. paniculatum are known to grow at elevations of ca. 9,000 feet. Certain populations of T. crispum have reached the highest limits for the genus. A few plants have been collected from altitudes as high as 10,500 feet in the Virginia Lakes region of California (M. & L. Smith 335, Cas). Habitats. Species of Thelypodium are found in various types of habi- tats with alkaline to non-alkaline soils that are wet or dry, sandy, loamy, or clay-like; limestones, sandstones, or serpentine rock and talus; in open meadows, rocky slopes, sunny flats or slopes, shaded canyons, stream- sides, and sand dunes (Table 4). Some of the widely distributed species, such as T. integrifolium and T. laciniatum, occupy a variety of habitats, while the endemic species have a rather narrow habitat preference. For example, T. repandum prefers the loose and decomposing shale on the steep canyons of the Salmon River in Custer County, Idaho. The soil in such a habitat is unstable and dries out quickly. Plants of this species are probably adapted to such conditions by having a rather deep root system, short stems, and dis- tinctly fleshy leaves. At least eight species are known to grow in alkaline meadows or flats and some of them, T. crispum, T. brachycarpum, and T. flexuosum, can tolerate strongly alkaline soils in which very few other plants are found. Plants of T. flexuosum have slender and weak stems that receive their support by being twisted between the branches of desert shrubs under which they grow. Some of these shrubs are the greasewood, Sarcobatus vermiculatus, sagebrush, Artemisia tridentata, Atriplex spp., and Chrysothamnus spp. Being long-lived perennials, plants of T. flexuosum are well-established in such habitats. Unlike other species of Thelypodium, the stems are very few-leaved in this species and the BIOSYSTEMATICS OF THELYPODIUM 43 TABLE 4, SPECIES OF THELYPODIUM, THEIR HABITATS AND THE ASSOCIATIONS IN WHICH THEY GROW n on oO - 3 B fae ne 5 = § g a “— o ‘2 a ~ « n nw oe w Qy o oS n =] = | meee Seat a a gs po = so o S wn is) +4 ws — = ao o a ° =} = e hy a b ee ee tu ae a S a By ee sf See ee a fos € BY 8 sa ee eee er ae ee | | 2 3 8 oS eS 8 4 8 fa 3 Se oS poe ee ee ee ee 8 ae . 8 £:.23° 6 ace 2 § 2 se =] 2) 8 8a oe fe ee a ay Oo 8 Bae eee 2.8 4.2% ae ie eee Coie eS. Be we ee af 8 Bog 2 ee a. ¢¢ 8 ¢ > ov a =) an eS. S. Se a 5 o 8 3 & @:28 8 24 2 § 8 8 2 8 fh T. laciniatum x x x T. milleflorum x T. crispum x x x T. brachycarpum x x T. eucosm x x © T. rollinsii x T. integrifolium x x x oe oe Ay T. stenopetalum x T. howellii x . T. sagittatum x * T. paniculatum x * T. flexuos agent T. wrightii x x x x T. laxiflorum x x x x T. repandum x T. texanum x x major photosynthetic part is the basal leaves that are shed at the end of the growing season. In the remaining species of the genus, basal leaves are shed as the stems grow and the cauline leaves are formed. Thelypo- dium sagittatum was said to grow in association with Sarcobatus (Holm- gren, 1972), but it is likely that the species meant was T. flexuosum. nas The only species of Thelypodium clearly adapted to wet conditions is T. paniculatum. It grows in wet sedge meadows where the water level may cover the basal portion of the plant. Its closest relative, T. sagitta- tum, is also found in meadows, but it usually prefers much drier sites and alkaline soil. With the exception of a few species, Thelypodium is adapted to some- what arid or semiarid conditions. Seven species are listed in Table 4 as growing near streamsides, river banks, and creek beds, but otherwise most of them occur in dry places. Thelypodium milleflorum and T. flexu- osum are the only two species that appear to tolerate the high degrees of aridity of open deserts, while T. crispum, and, to a lesser extent, T. in- 44 IHSAN A. AL-SHEHBAZ tegrifolium can withstand the highly mineralized soils adjacent to hot springs. The majority of species grow in open areas that are directly ex- posed to sunlight, but T. wrightii, T. laxiflorum, T. eucosmum, and T. laciniatum grow in shady as well as open habitats. Economic importance. Very few reports are available on the economic value(s) of the various species of Thelypodium. Vasey and Rose (1890) published the following field note made by Edward Palmer on the plants of T. integrifolium subsp. affine: “The leaves are cooked and eaten by the Indians. Cattle do not seem to feed on this.” A similar note was made by Elzada Clover on herbarium sheets (Clover 5113, ps & MicH) that the plants of T. integrifolium subsp. longicarpum are “used as spinach by Supais.” Hermann (1966) mentioned that T. laciniatum “appears to be valuable for sheep early in the season where it grows in sufficient quantity” and that T. integrifolium “appears to be of some forage value under certain conditions.” From my field experience with most species of Thelypodium growing in open ranches where cattle were present, the plants were mostly healthy looking and relatively few of them were eaten. It is likely that the plants are avoided by cattle and that occasional eat- ing might be accidental. No other usages or values are known and it is clear that the genus has no economic significance. REPRODUCTIVE BIOLOGY Flowering. In the annual species of Thelypodium, such as T. texanum, T. tenue, and T. paysonii, the life cycle of the plant appears to be di- rectly related to the availability of rain, and flowering probably takes place a few weeks after seed germination. Photoperiodism does not ap- pear to be a crucial factor in the flowering of T. texanum, since the plants flowered in growth chambers with 8, 12, and 16 hours of daily light periods. As for the biennial and perennial plants, photoperiodism alone has no effect. The main requirement to induce flowering probably is subjection to cold. The fifteen biennial and perennial species of Thelypodium seem to have an obligate cold requirement, since the untreated rosettes failed to flower and some of them remained in such a stage from one to three years. Biennials with an obligate cold requirement grown under non- inductive conditions, in which they remain growing vegetatively, can be induced to flower by treating them with gibberellic acid (GA,) (Lang, 1957, 1965; Michniewicz & Lang, 1962). This might be a useful approach to follow in order to study the flowering and the breeding systems of Thelypodium. The GA, solution used was prepared as follows: 25 mg. of gibberellic acid (Eastman Organic Chemicals) dissolved in 500 ml. of 0.05% tween 20 (Fisher Scientific) as a wetting agent. The resulting solution (concentration 50 »g/ml.) was applied dropwise to the growing center of the rosette, which received 2 drops (approximately 5 pg of GA, ) every other day. BIOSYSTEMATICS OF THELYPODIUM 45 The response of the various species of Thelypodium to the gibberellin treatment was different. Three species, T. brachycarpum, T. panicula- tum, and T. howellii, showed no signs of bolting and/or flowering and remained in the rosette stage after a period of gibberellin treatment extending nearly five months. The best, and by far the most consistent, response to this treatment was shown by T. stenopetalum. Five to four- teen applications were sufficient for bolting and three to six additional Ones were necessary to induce flowering. Plants of T. paysonii and T. tenue were not available for such a treatment. The other eleven species that flowered showed no signs of consistency in terms of the number of applications needed for bolting and flowering. In certain species the response was fast in some rosettes and much slower in others. It is beyond the scope of this study to go into the details of this treatment, but it is clear that it is a valuable approach to obtain flowering for the study of breeding systems in some species of Thelypodium. Pollination. Almost all species of Thelypodium have showy flowers that range in color from white to dark purple. They often produce nectar and are usually grouped in dense inflorescences, undoubtedly contributing to the attraction of insects for pollination. By far the most common visi- tors are two types of bees belonging to the genera Bombus (Bumble- bee) and Apis (Honey-bee, A. millefera) of the family Bombidae. Car- penter bees of the genus Xylocopa (family Xylocopidae) were fairly common on plants of T. integrifolium in Arizona, Utah, and Nevada. A few unidentified species of bees also were found. One species of butter- fly of the family Lycinidae was seen, but only a few times, on plants of T. rollinsii. Some members of the butterfly genus Pieris (probably P. sis- ymbrii) of the family Pieridae were fairly common visitors to the flowers of T. rollinsii, T. integrifolium, T. laciniatum, and T. wrightii. Breeding systems. The evidence available from all 13 species that flow- ered in the greenhouse was that autogamy did not occur in most species. Representatives of eight species are clearly self-compatible. They are: Thelypodium integrifolium (subsp. affine, subsp. integrifolium, and subsp. longicarpum were tested), T. laxiflorum, T. milleflorum, T. repandum, T. rollinsii, T. stenopetalum, T. texanum, and T. wrightii subsp. wrightii. Plants of two of the remaining five species, T. sagittatum and T. flexuosum, showed that they are probably self-incompatible. However, the evidence was not conclusive. Whether these findings will hold in other populations of the species so far investigated in a preliminary way remains to be de- termined. It should be kept in mind that both self-compatibility and self- incompatibility may be present in the same taxon (Rollins, 1963a; Lloyd, 1965). The flowers of Thelypodium possess certain features that appear to reduce the chances of selfing. In the species with distinctly exserted an- thers, the stigmatic surface is always below the anthers in the recently opened flower. As the anthers dehisce, they become distinctly coiled and 46 IHSAN A. AL-SHEHBAZ the filaments become ascending or somewhat recurved. At the same time, the pistil elongates and the stigmatic level is raised, but almost never touches the anthers. This type of flower is characteristic of T. laciniatum, T. milleflorum, T. eucosmum, T. crispum, T. brachycarpum, and, to a lesser extent, T. integrifolium and T. rollinsii. In three of the five species with spreading floral parts, T. wrightii, T. texanum, and T. repandum, the filaments are well spread by the time the anthers start to dehisce and the pistils are at some distance from the anthers. In some plants of Thelypodium texanum grown under greenhouse con- ditions, the styles were observed to emerge from the floral buds two to three days before the flowers opened. This is a clear case of protogyny. Unfortunately, we do not know how common protogyny is in natural populations of this species. Protogyny, protandry, and dioecy were said to be lacking in the Cruciferae (Bateman, 1955a). Dioecism occurs in three species of Lepidium from New Zealand (Kirk, 1899; Bateman, 1955b; Allan, 1961). Protandry is known in three species of Streptanthus ( Krucke- berg, 1957; Rollins, 1963b). The earliest report on protogyny in the Cruciferae probably dates back to Kerner (1895), who stated that “. . . and Cruciferae exclusively protogynous.” However, neither he nor Faegri and van der Pijl (1966) gave any examples. Protogyny is now well-known in Arabis (Rollins, 1971; Johnson, 1970) and Thlaspi (Riley, 1956; Holm- gren, 1971). Isolating mechanisms. As a general rule, species of Thelypodium are either wholly or for the most part allopatric. Various types of isolation, such as seasonal, ecological, reproductive, and geographical operate to maintain well-defined species in the genus. Natural hybridization has not been detected. One of the most effective types of isolation is the geo- graphical separation of species. As shown in Maps I, 2, 4, and 5, the dis- tributional ranges of the various taxa hardly overlap, and when this happens, other types of isolation seem to be effective. For example, T. sagittatum and T. flexuosum have different distributions (Map 4), but in northern Nevada, where they are both present, the latter species is found primarily in strongly alkaline desert flats and in association with shrubs, while the former species grows in meadows that are somewhat alkaline. Seasonal isolation is generally considered to be “leaky,” because it is not an absolute barrier and is likely to break down (Solbrig, 1970). The present writer has observed a few cases of contact between pairs of unre- lated species of Thelypodium where seasonal isolation was evident and effective. In the examples mentioned below, plants of the first species of each pair were completely dry and some of them had shed most of their seeds, while those of the second species were at the beginning of their flowering season. These species pairs are as follows (including the collec- tion numbers): Thelypodium sagittatum (6919) and T. integrifolium subsp. integrifolium (6920); T. sagittatum (6925) and T. integrifol- ium subsp. complanatum (6926); T. flexuosum (6936) and T. integrifolium BIOSYSTEMATICS OF THELYPODIUM 47 subsp. complanatum (6935); T. howellii subsp. spectabilis (6957) and T. integrifolium subsp. complanatum (6956); T. laciniatum (6968) and T. integrifolium subsp. integrifolium (6967). Crossing experiments were done to a very limited extent, because it was not possible to bring plants of closely related species to flower at overlapping periods. The only results worth mentioning are those from the reciprocal crosses between Thelypodium rollinsii and two subspecies of T. integrifolium (subsp. integrifolium and affine). Nearly fifty flowers were crossed in each direction and no fruits were formed. This suggests that the two species are genetically isolated. Seed dispersal. Seeds of Thelypodium are quite small and always very light in weight. The lightest seeds in the genus belong to T. rollinsii and T. wrightii, where the average weight of a single seed ranges between 0.15 and 0.18 mg. The heaviest seeds are found in T. paniculatum and T. brachycarpum, where the average weight ranges from 0.58 to 0.62 mg. Seed weight for the other species of the genus falls between the two sets of species given above but it is much closer to the former pair of species than to the latter. Being very light in weight, the seeds of Thelypodium are probably very easily distributed. Wind may be an important factor in their dis- persal in open habitats, such as deserts, flats, and open slopes, but the action of rain wash is perhaps equally, if not more, important in such open habitats. Flooding is likely to be significant in seed dispersal of spe- cies growing along streamsides, creek beds, and river banks and may also be important for those species found in meadows (Table 4). PALYNOLOGY Based on light microscope studies alone, pollen grains were said to be similar throughout the Cruciferae. Erdtman (1952, 1969, 1971) suggested that the Cruciferae is “. . . a very stenopalynous family.” The advantages of the scanning electron microscope (SEM) in palynological studies have been pointed out by Echlin (1968), who published one of the earliest known SEM photomicrographs of cruciferous pollen belonging to the genus Cheiranthus. Pollen SEM photomicrographs of other crucifers have been published by Martin and Drew (1969, 1970) and Stork (1972). These few reports suggest that some interesting differences in pollen grains of the genera studied do exist and the findings raised some ques- tions on the validity of the conclusions reached by Erdtman from light microscopy alone. With this in mind, a comprehensive pollen survey of the Cruciferae, using the SEM is being conducted by R. C. Rollins, E. A. Shaw, J. E. Rodman and myself. The results will be published inde- pendently. The pollen grains of the Cruciferae are said to be more or less similar to those of the Capparaceae and different from the Papaveraceae-Fuma- roideae (Erdtman, 1951, 1971). It is generally agreed that the Papa- 48 IHSAN A. AL-SHEHBAZ -_ M J” ae mor yy . PxraTe 10. SEM photomicrographs of pollen grains of Thelypodium milleforum, Al-Shehbaz 6898 (Fics. 1-6) and T. stenopetalum, Al-Shehbaz 6947 (Fics. 7-8). Magnifications: 1. 490; 2. 980; 3. 2450x ; 4. 4900 ; 5. 9800; 6. 5000 (acetolysed); 7. 700; 8. 3500. BIOSYSTEMATICS OF THELYPODIUM 5 =, PLaTE 11. SEM PRS aN of ai ollen grains ee Thelypodium oe Al- sick (Fies. 1-3), T. kaeitoram Shehbs 2 (Fic. 4), T. flexuosum, Hitchcock & Muhlick 222 (Fics. 5-7), and T. pay : ryon 58296 ( Si . 8). EON 1. 760 1900 » 3. 3800 ; 4. 770%: ; 5. 750%; ; 6. 1875 x; 7. 3750 700 23992 50 THSAN A. AL-SHEHBAZ veraceae is not closely related to the other two families for morphological (Cronquist, 1968; Merxmiiller & Leins, 1967) and chemical reasons ( Heg- nauer, 1964, 1969; Ettlinger and Kjaer, 1968). It has been suggested that the tribe Thelypodieae (Stanleyeae) is the link between the Cruci- ferae and the ancestral Cleomoideae-Capparaceae (Takhtajan, 1969). With the hope that pollen grains of the Thelypodieae and the Cleomoi- deae might give further insight into the relationship between the two families, a survey of all the genera of the Thelypodieae and selected genera of the Cleomoideae was conducted. Our findings on all the genera of the Thelypodieae will be presented in the comprehensive survey of the family mentioned above. Thelypodium is used as a representative of the tribe in the present discussion, since its poilen grains are fairly simi- lar in many respects to most of the other members of the tribe. The ter- minology used follows that illustrated by Kremp (1965). Pollen grains of all samples were taken from herbarium specimens and mounted, without treatment, on aluminum holders using double-stick tape. A very few samples of Thelypodium and Cleome were acetolized by Mr. Umesh Banerjee following the standardized method of Erdtman (1960). All samples were coated in a vacuum chamber with a thin film of carbon (50-100 A thick) followed by a gold-palladium film (200-300 A thick) for conductivity and to prevent charging. Polaroid photomicro- graphs were taken on an AMR Model 900 Scanning Electron Microscope. Pollen of Thelypodium. Like most genera of the Thelypodieae, Thely- podium species are fairly uniform in their pollen morphology (Plates 10 and 11). The following description is based on pollen of almost all species of the genus: pollen prolate, reticulate, tricolpate, 19-29 y» long, 9-15 » wide, rounded at the poles; colpi rather long and nearly reaching the poles but remaining distinct (Plate 10, Fig. 6); reticulum smooth, fine in size; lumina rounded or seldom slightly angled, variable in size and shape, 0.1-1.5 (2.7) » in diameter, smaller ones nearly isodiametric, Jarger irregular and sometimes elongated, the smaller lumina are nearer the polar region than the middle; reticulum distinctly elevated by sup- porting columellae that are sometimes visible, free columellae rare; muri 0.3-0.7 uw wide. There is some degree of variability in the shape of the lumina. Those of Thelypodium stenopetalum are rounded, while those of T. milleflorum are slightly subangular (Plate 10, Figs. 4, 5, and 8). The description given, excluding measurements, presents a fairly good picture of the pollen in the Thelypodieae except the genera Warea and perhaps Strep- tanthella which have a decidedly coarser reticulum. Pollen of certain Cleomoideae. Pollen of the family Capparaceae was said to show some variation in surface configurations by Erdtman (1952, 1971), but it was concluded by the same author (Erdtman, 1969) that the family is “markedly stenopalynous.” Our preliminary studies of a few species of the family, most of which belong to the Cleomoideae, do not BIOSYSTEMATICS OF THELYPODIUM 51 ted genera of the Capparaceae. Fics t PLaTE 12. SEM JE spaageernctaars of pollen grains of selec 1-2. Polanisia dodecandra, Nelson pony Fics, 3 +4. Oxystylis lutea, Wiegand 803: Fic. 5. Cra acva ia, H is _ 6. Wislizenia retrofracta, Peebles 14558; Fx. 7. Isomeris arborea, Munz 11941; Fie. 8. Cleo nag iongipes, ‘Maller 1 2. 4850 3 15 ; 6. 990 x; 7. 208: 1100. Magnifications: 1. 97 9X; 4. 3798; 5. 2500 2450 IHSAN A. AL-SHEHBAZ ain species of Cleome. Fics. 1—2 a. C. a. Potter As A7T5 x 4950 PLA . SEM ee of pollen of ¢ chuiae: gph 642; . 3-4 serratula, a 5660. Fre 6-8. C. ha nage eriana gd gorse Magnifications: 1. 980x 2450: 3. 5. 2450 ; 6. 970 . 1698; 8. 4850 x 53 BIOSYSTEMATICS OF THELYPODIUM of certain species of Cleome, Fic. 6G 31: Fics. 5—8. C. platycarpa Cronquist 4900 x; 5. 1000x en as 572 EM — of pollen Fics. 2-4. C. lutea, Rollir 50x; 3. 2500X ( 8. 5000 (ac ileeud) lt 8. 14. § Kennedy 1991; Fic 300; 2. 2500 PLar TE sparcifolia, 7185. PA agen 2500 <: 7.2 (acetolysed ) _ ~_ GO o 3 ( acetoly. sed ); 54 IHSAN A. AL-SHEHBAZ support the last conclusion. All the genera discussed below except Cra- taeva belong to the subfamily Cleomoideae. The pollen of this genus is finely pitted (Plate 12, Fig. 5) and the unacetolized grains look nearly smooth, Both Polanisia dodecandra, a native to the new world (Iltis, 1958), and Cleome gynandra, an introduced species from Africa (Iltis, 1960), have striate-reticulate pollen grains (Plates 12 and 13). The colpi in the two species extend nearly to the poles. The former species has prolate and the latter subspherical grains. In C. hassleriana the grains are dis- tinctly echinate (Plate 13). Pollen grains of C. platycarpa, C. serratula, C. sparsifolia, C. lutea, Isomeris arborea, Cleomella longipes, and Wis- lizenia refracta are prolate, shallowly reticulate, and with shallow colpi extending nearly to the poles (Plates 12, 13, and 14). The muri, if the term is applicable here, are very low and the surface between them is distinctly pitted (Plate 14). The pollen is distinctly tricolporate in C. platycarpa and C. lutea (Plate 14), but C. serratula and W. refracta appear in Plates 12 and 13 as though they are tricolpate. However, the last two species are clearly tricolporate as judged from the study of their acetolysed pollen by light microscopy. The pollen grains of C. muticaulis and Oxystylis lutea resemble each other in being elliptical-subspherical, tricolporate, and faintly reticulate. The two species are not related. In fact, O. lutea is considered to be at the end of a reduction series in the order Cleome, Cleomella, Wislizenia, and Oxystylis (Itis, 1957). Although reticulate pollen grains are found in the Cleomoideae, they are very clearly different from those of the Cruciferae. The reticulum in the latter family is very sharply defined and clearly elevated by support- ing columellae. The reticulum in the Cleomoideae, when present, is much shallower and neither the columellae nor the lumina are recognizable. Furthermore, the pitted surface between the “muri” in the Cleomoideae has not been detected in the Cruciferae. In addition to these significant differences, pollen of the Cruciferae are colpate, while the Cleomoideae appear to be exclusively colporate. Two conclusions can be drawn from the comparison of pollen grains of the tribe Thelypodieae and the subfamily Cleomoideae and both are contrary to those reached by Erdtman (1969, 1971). First, the family Capparaceae is not a stenopalynous family. Second, pollen of the Cap- paraceae is very different from that of the Cruciferae. The similarities between them appear to be less significant than previously supposed. Pollen grains of the Thelypodieae are typically cruciferous and pollen data alone do not seem to support the suggestion that the tribe is inter- mediate between the rest of the Cruciferae and the Cleomoideae. CYTOLOGY Until mone cytological information concerning the genus Thelypo- dium has been lacking. The only available report containing chromo- BIOSYSTEMATICS OF THELYPODIUM 55 some counts of species that are true members of the genus is that of Rollins (1966). An earlier report by Snow (1959) included a chromosome count of T. lasiophyllum, but the species is here excluded from Thelypo- dium and placed with its relatives in Caulanthus. Rollins presented for the first time counts on T. flavescens (n=14), T. lemmonii (n=14), fi flexuosum (n=13), T. texanum (n=13), and tentative counts on T. lacinia- tum (n=ca. 12) and T. milleflorum (n=ca. 14). Both T. flavescens and T. lemmonii are considered members of Caulanthus in the present treat- ment. We have been able to confirm the earlier chromosome count of T. texanum from greenhouse plants grown from seeds collected by A. S. Barclay (see below). The approximate numbers given by Rollins (1966) for T. laciniatum and T. milleflorum are very close to our finding of n=13 in both species. All the chromosome counts presented here were made from bud ma- terials fixed in three parts of 100% ethyl alcohol and one part glacial acetic acid. After one day of fixation, the buds were washed twice in 70% alcohol before they were stored or studied. The results are presented below. Thelypodium crispum Greene ex Payson. n=13: Ormsby Co., Nevada. I. & M. Al -Shehbaz 68110, GH (Plate 15, Fig. 4). Thelypodium howellii Watson subsp. spectabilis (Peck) Al-Shehbaz. n= 13: Union Co., Oregon. I. & M. Al-Shehbaz 68151, GH (Plate 15, Fig. 2). Thelypodium integrifolium (Nutt.) Endl. subsp. integrifolium. n=13: Gunnison Co., Colorado. I. & M. Al-Shehbaz 6903, GH. Thelypodium integrifolium (Nutt.) Endl. subsp. affine (Greene) Al- Shehbaz. n=13: Clark Co., Nevada. I. & M. Al-Shehbaz 6988, GH ( Plate 15, Fig. 3). Thelypodium laciniatum (Hook.) Endl. n=13: Mono Co., California. I. & M. Al-Shehbaz 68112, GH. Thelypodium milleflorum Nelson. n=13: Elko Co., Nevada. I. & M. Al-Shehbaz 6898, GH. Thelypodium rollinsii Al-Shehbaz. n=13: Sevier Co., Utah. I. d M. Al-Shehbaz 6913, GH. Thelypodium sagittatum (Nutt.) Endl. subsp. ovalifolium (Rydberg) Al-Shehbaz. n=13: Garfield Co., Utah. I. & M. Al-Shehbaz 6911, GH (Plate 15, Fig. 1). Thelypodium texanum (Cory) Rollins. n=13: Brewster Co., Texas. From seeds collected by A. S. Barclay 713, GH. It is very likely that the fundamental chromosome number for Thely- podium is n=13. More counts from the species listed above and the re- maining members of the genus are needed before any rigid conclusions can be made. Although a haploid number of n=13 has not yet been counted from other genera of the tribe Thelypodieae, it is very close to n-14, which is the most common number in the tribe. The latter number is very common in Streptanthus, Caulanthus, Stanleya, and the monotypic 56 THSAN A. AL-SHEHBAZ : Rae » sora of selected species of Thelypodium. Fic. 1. PMC, half of anaphase I, : p. ovalifolium, Al-Shehbaz 6911; ~ be at late anaphase I, T. howellii subsp. spectabilis, — 8151; Fic. 3. PMC, half o . integrifoliu ‘ae 51; Fic. m subsp. affine, 6988; Fic. 4. PMC, diakinesis, T. crispum press sh All figures x ca. 2800. BIOSYSTEMATICS OF THELYPODIUM ar Streptanthella. The least common numbers in the tribe appear to be n=10, present in two species of Caulanthus, and n=11 which is found in two species of Thelypodiopsis (Rollins, 1966; Rollins and Riidenberg, 1971). Both Thelypodiopsis linearifolia and T. ambigua are listed as species of Sisymbrium in the two treatments, but, as suggested by these authors, the two species are out of place in Sisymbrium. Twenty-nine of the 50 species known cytologically (including the counts of Thelypodium pre- sented here) that belong to the tribe Thelypodieae have chromosome numbers based on n=14 (Rollins, 1966; Rollins and Riidenberg, 1971; Bol- khovskikh et al., 1969). The presence of such a number in the three spe- cies of Stanleya, which is probably one of the most primitive genera of the tribe, and the occurrence of n=13 in nine species of Thelypodium is of interest. It is probable that a number of x=7 is basic for the tribe Thelypodieae. Furthermore, chromosome counts from other genera, such as Romanschulzia, Warea, and Chlorocrambe, as well as additional species of Stanleya, Thelypodium, Streptanthus, Caulanthus, and Thely- podiopsis are very necessary before one can present a clear picture of the cytology of the tribe. BIOCHEMICAL SYSTEMATICS Thelypodium, like many other primitive genera of the family Cruciferae, has received only minimal attention in general surveys of natural pro- ducts. As far as the literature shows, only one species of Thelypodium (T. texanum) has been investigated for mustard oil glucosides (Daxen- bichler et al. 1964) and seed proteins and fatty acids (Mikolajczak et al. 1961; Jones and Earle, 1966). Using paper chromatography, Dr. Martin G. Ettlinger and Charlyne P. Thompson (personal communica- tion) analyzed seed samples of two species of Thelypodium (T. texanum and T. milleflorum) for their glucosinolate contents. Their results agree completely with those I have obtained by paper chromatography on these two species, The mustard oil glucosides (glucosinolates) represent a unique class of natural products, restricted in their distribution to a limited number of families of the dicotyledons. The distinctive pungent flavor or odor present in some of the common mustards (species of Brassica, Sinapis, and other related genera) has been recognized for several centuries, and a few species have been utilized in the preparation of condiments (Vaughan and Hemingway, 1959). The flavoring principles (mustard oils or isothiocyanates) are only found in minute quantities, if at all, in the intact plant tissues, but are readily recognized when the tissues are crushed or injured (Ettlinger and Kjaer, 1968; Tang, 1971). The glucosinolates are present in practically all the species of the fam- ily Cruciferae (Kjaer, 1960). Other families known to contain them are the Capparaceae, Tovariaceae, Resedaceae, Moringaceae (these five fam- 58 IHSAN A. AL-SHEHBAZ ilies constitute the well-defined order Capparales sensu Cronquist (1968), Limnanthaceae, Caricaceae, Tropaeolaceae, Gyrostemonaceae, Salvadora- ceae, and the genus Drypetes (including Putranjiva) of the family Eu- phorbiaceae (Ettlinger and Kjaer, 1968). Fifty glucosinolates were structurally known in 1968 (Ettlinger and Kjaer, 1968). Since then an additional 19 have been discovered from natural sources (cf. Elliott and Stowe, 1970; Gmelin et al. 1970; Kjaer and Schuster, 1970, 1971, 1972a, 1972b, 1973; Kjaer and Wagnieres, 1971; Underhill and Kirkland, 1972). The general formula of the glucosinolates was established by Ettlinger and Lundeen in 1956. The side chain (R) can be any organic group from methyl up (Ettlinger and Kjaer, 1968). As shown below, the glucosino- lates are hydrolysed by myrosinase to isothiocyanates (mustard oils), sulfate, and D-glucose. R-C-S-C,H,,0, + H,O enzyme R-N=-C-S + HSO, + C,Hy.0¢ ——————> N-OSO,”° Glucosinolate Isothiocyanate D-glucose Many of the isothiocyanates volatilize readily. However, those contain- ing sulfoxide (SO) or sulfone (SO,) groups volatilize only with difficulty. Moreover, isothiocyanates that have a hydroxyl group on a carbon atom next to the one bearing the isothiocyanate group are neither volatile nor stable, and spontaneously cyclize giving rise to oxazolidine-2-thiones. The oxazolidine-2-thiones, among other glucosinolate derivatives, are known to have antithyroid or goitrogenic effects in man and a few other animals (Astwood et al. 1949; Langer, 1966; Van Etten et al. 1969; Gillie, 1971). Nothing is known about the exact functions of the glucosinolates in the plants that contain them. Whittaker (1970) and Wareing (1965) sug- gested that they are probably seed germination inhibitors, while Kuta- éek (1964) and Elliott and Stowe (1971) proposed a function in the nature of allelopathy. Finally, the glucosinolates are said to have a de- fensive function against certain animal species (Fraenkel, 1959; Whit- taker, 1970). The idea may have merit, but we know of certain species of insects that seem to have evolved to withstand any toxic effects ex- erted by the glucosinolates or their derivatives (Whittaker and Feeny, 1971; Ehrlich and Raven, 1964). In fact, it has been shown that these compounds stimulate feeding and oviposition in certain species of the insect genera Pieris and Plutella (Thorsteinson, 1953; Gupta and Thor- steinson, 1960; Nayar and Thorsteinson, 1963; Dethier, 1970; Schoon- hoven, 1968, 1969). The skeletal similarities between the glucosinolates and some of the BIOSYSTEMATICS OF THELYPODIUM 59 protein amino acids were clarified by Ettlinger and Lundeen (1956). We now know that these amino acids are direct precursors of the gluco- sinolates, and that glucosinolates may also be derived from higher homo- logues of protein amino acids, produced by a chain-lengthening mecha- nism utilizing acetate (Underhill et al., 1962; Underhill and Wetter, 1966; Ettlinger and Kjaer, 1968; Josefsson, 1971). Very little is known about the chemosystematic value of the gluco- sinolates. The first attempt to determine this value was made by Kjaer and Hansen (1958) with four species of the cruciferous genus Arabis. Using paper chromatography, Ettlinger and Thompson (1962) inves- tigated a large number of cultivars belonging to a relatively few species of the genus Brassica and related genera. Their results showed certain distributional patterns of the glucosinolates in these taxa, but the most significant differences were between Brassica and Sinapis. The present investigation was conducted with the hope that the distribution of the glucosinolates within Thelypodium might provide some information of chemotaxonomic value, particularly in connection with certain species complexes. Experimental. The paper-chromatographic analysis involves the en- zymatic hydrolysis of the glucosinolates by myrosinase (in the presence of sodium ascorbate as a co-factor, 5 ml. of distilled water, and 100 ml. of diethyl ether) to isothiocyanates. Small portions of the ether extract of the isothiocyanates are checked for the presence of p-hydroxybenzyl isothiocyanate and oxazolidine-2-thiones. The remainder is converted into thioureas by the addition of % of its value of ethanolic ammonium hydroxide. The thioureas and oxazolidine-2-thiones are then chromato- graphed with the reference standard phenylthiourea in a few solvent systems (one to six), depending on the nature of these compounds. The extraction of the glucosinolate derivatives from the seeds, the paper chromatography techniques, and the solvent systems are essentially the same as those used by Ettlinger and Thomson (1962) and Ettlinger et al. (1966) and therefore will not be described again. The thioureas and oxazolidine-2-thiones were identified by comparing their R,,, values (the ratios between the distances travelled by the thioureas and oxazolidine- 2-thiones and by the phenylthiourea standard (Kjaer and Rubinstein, 1953) ) with those of known compounds from tables kindly provided by Dr. Martin G. Ettlinger (Ettlinger and Thompson, unpublished results ). Several standards of the thioureas, oxazoeidine-2-thiones, and isothiocya- nates were generously given by Dr. Ettlinger. The extraction of the glucosinolates from the leaves was as follows. Fresh rosette leaves (20-30 grams) were put into 500 ml. of boiling 70% methanol for 15-20 minutes. The leaves were blended for one to two minutes in a high-speed blender. They were then filtered through a few layers of cheesecloth, and the filtrate was boiled on a hot plate until all 60 IHSAN A. AL-SHEHBAZ of the methanol and most of the water had evaporated. After cooling, 10 ml. of distilled water was added. The diluted extract was then washed four to six times with generous portions of petroleum ether until most of the chlorophyll and other pigments were removed. The nearly colorless water extract of the glucosinolates was hydrolized and converted to the thiourea derivatives as outlined by Ettlinger and Thompson (1962) and Ettlinger et al. (1966). The thioureas and oxazolidine-2-thiones were then identified by paper chromatography as mentioned earlier. The extraction of the isothiocyanates for GLC analysis was as follows. Seeds (0.5 gm.) were finely ground and defatted with 75 ml. of diethyl ether for two to three hours. The ether was decanted carefully, and the ground seed was washed with 20-25 ml. of ether, filtered off, and washed three or four more times with ether on the filter. The defatted seed pow- der was air-dried and put in a 50 ml. Erlenmeyer flask, and 30 ml. of boiling 70% methanol was added. The flask was placed on a hot plate for approximately 15 minutes until the seed extract was freed of the metha- nol and some of the water. The concentrated extract was cooled and sub- jected to enzymatic hydrolysis by the addition of three drops of myro- sinase preparation, 0.1 ml. of sodium ascorbate solution, 5 ml. of distilled water, and 10 ml. of diethyl ether (free of alcohol and peroxide). The mixture was let stand 10-12 hours at room temperature, in the dark, with occasional shaking. The ether extract, containing the isothiocyanates (and oxazolidine-2-thiones), was carefully removed from the aqueous mixture by pipette and used in the GLC analyses. Boiling the samples in methanol was necessary to denature the various enzymes present in the seeds that might convert the glucosinolates or other substrates to volatile products other than isothiocyanates. Such products, which appeared as peaks of various sizes in a number of sam- ples, were completely eliminated when the samples were boiled. An F. & M. high efficiency gas chromatograph Model 402 linked with a Hewlett-Packard recorder Model 7127A was used in all analyses. The chro- matograph was operated isothermally at a column temperature of 80° C. for the characterization of the highly volatile isothiocyanates (flame detector temperature was 145° C. and flash heater 130° C.), and at 160° C. for the characterization of those less volatile (temperatures of the flame detector and flash heater were 170° C. and 200° C. respectively ). These isothiocyanates are shown in Table 8. The gas chromatograph was equipped with a hydrogen flame ionization detector. A U-shaped glass column (6 feet < % inch) packed with 6% DC-560 and 2% EGSP-Z on 100/200 Gas-Chrom Q (Applied Science Laboratories) was used. The helium (carrier gas) flow rate was approximately 60 ml. per minute. A range of 10, an attenuation of 1, and a recorder speed of 4 inch/minute were used. The sample size was 2ul. for the most part. The quantitative determination of the volatile isothiocyanates was accomplished by calculat- BIOSYSTEMATICS OF THELYPODIUM 61 ing their peak areas (done by the multiplication of their peak heights by the peak widths at half the height) as a measure of their amounts, which are expressed as percentages of the total volatile contents area. No attempts were made to quantify the oxazolidine-2-thiones, although it is possible to do so by UV absorption as described by Youngs and Wetter (1967). Results and discussion. A relatively large number of glucosinolates have been identified from the various species of Thelypodium (Table 5). None of the p-hydroxybenzyl glucosinolate has been detected in any of the samples prepared for paper-chromatographic analysis. Three oxazoli- dine-2-thiones encountered in Thelypodium are 4-methyloxazolidine-2- thione, 4-ethyloxazolidine-2-thione, and 5-vinyloxazolidine-2-thione, as shown below. These are derived from spontaneous cyclization of the iso- thiocyanates resulting from hydrolysis of the glucosinolates 2, 5, and 12 respectively. Each of these three glucosinolates is a very close relative of the one preceding it in Table 5. CH,-CH NH CH.CH NH CH,———NH H.C C-S H.C C=S a C=S = 8 ae ve H,C-CH rae vs 0 4-Methyloxazolidine- 4-Ethyloxazolidine- 5-Vinyloxazolidine- : 2-thione 2-thione 2-thione Two stereochemical configurations are known for the 2-hydroxy-3- butenyl glucosinolate and the derived 5-vinyloxazolidine-2-thiones. Ast- wood et al. (1949) discovered the first 5-vinyloxazolidine-2-thione in the genus Brassica. It was shown later to be the (S) compound by Kjaer et al. (1959). The (R) isomer was obtained from the seeds of Crambe abyssinica by Daxenbichler et al. (1965). No attempts have been made in the present treatment to determine which isomer is produced in Thely- podium. The results of the paper-chromatographic analyses are shown in Table 6. The comparison between these and the GLC results in Tables 9, 10, 11, and 12 shows that some of the minor constituents found in a number of species were not detected by paper chromatography. For example, the benzyl and sec-butyl glucosinolates in Thelypodium rollinsii and the 4-methylthiobutyl glucosinolate and 5-methylthiopentyl glucosinolate in T. laciniatum were not detected by paper chromatography. Moreover, the isobutyl isothiocyanate was not detected by the paper-chromatograph- ic analysis of its derived thiourea in all samples of T. laciniatum and 62 THSAN A. AL-SHEHBAZ TABLE 5. GLUCOSINOLATES PRESENT IN THE GENUS THELYPODIUM* Compound No. Systematic Name Side Chain, R 1 Isopropyl glucosinolate (CH, ) .CH- 2 2-Hydroxy-1-methylethy] glucosinolate HOCH,CH(CH,)- 3 Isobutyl glucosinolate (CH, ),CHCH,-— 4 sec-Butyl glucosinolate C,H,CH( CH, )- 5 1-(Hydroxymethyl) propyl glucosinolate © C,H;CH(CH,OH )- 6 3-Methylthiopropy! glucosinolate CH,S( CH ).- a Allyl glucosinolate H,C=CHCH,- 8 4-Methylthiobuty] glucosinolate CH,S(CH,).- 9 4-Methylsulfinylbutyl glucosinolate CH,SO( CH.) .- 10 4-Methylsulfonylbutyl glucosinolate CH,SO,(CH,).- ll 3-Butenyl glucosinolate H,C=CH (CH,)- 12 2-Hydroxy-3-butenyl glucosinolate H,C=CHCHOHCH,- 13 5-Methylthiopentyl glucosinolate CH,S(CH,);- 14 5-Methylsulfinylpentyl glucosinolate CH,SO(CH,)- 15 4-Pentenyl glucosinolate H,C—CH(CH,),- 16 nzyl glucosi C,H,CH,- 17 2-Phenylethyl glucosinolate C,H,( CH, ).- *Arranged according to Ettlinger and Kjaer (1968) with the addition of the recently discovered isobutyl! glucosinolate ( Underhill and Kirkland, 1972). T. milleflorum. This failure is due to the fact that isobutylthiourea and sec- butylthiourea give similar R,, values, making it impossible to resolve them by paper chromatography (Kjaer and Rubinstein, 1953). Further comparison between the GLC and paper chromatography results show the superiority of the former approach for both qualitative and quanti- tative determinations. However, the latter approach is still very valuable, particularly in connection with the nonvolatile glucosinolate derivatives. There is some variation in the glucosinolate composition between the basal leaves and the seeds of the same species (Table 6). The basal leaves of Thelypodium wrightii (Correll 34106), T. repandum (6921), and T. laciniatum (68133) contained small amounts of 5-methylsulfinyl- pentyl glucosinolate, which was not detected in the seeds of the same samples. Moreover, the leaves of T. wrightii showed a trace of the benzyl glucosinolate, which was not detected in the seeds, while the seeds con- tained 2-hydroxy-3-buteny] glucosinolate, which appeared to be lacking from the leaves. Qualitative variation in the glucosinolate contents of different parts of a given plant has been demonstrated by Gmelin and Kjaer (1970), Josefsson (1967), and Ettlinger and Kjaer (1968). The Ry, values of the thioureas and oxazolidine-2-thiones, as encoun- tered in Thelypodium, hold good to + 0.03 or + 0.04 and rarely vary as much as + 0.05. These values, as shown in Table 7, are only slightly different from those of Ettlinger and Thompson (1962 and unpublished). Most of the variability in the R,,, values is probably caused by fluctuation in temperature. BIOSYSTEMATICS OF THELYPODIUM 63 The identification of the various isothiocyanates by GLC was always aided by injection of the isothiocyanate standards either separately or mixed with small proportions of the sample. This precaution was very necessary, because the retention time (the time from injection of the sample to the appearance of the top of the peak) of a given isothiocya- nate was not always the same. The variation in the retention time was probably caused by slight differences in column temperature and flow rate of the carrier gas. The variability in retention time increases as the volatility of the isothiocyanate decreases (Table 8). TABLE 6. PAPER CHROMATOGRAPHY RESULTS Plant Species Collection No. art Glucosinolate® Thelypodium brachycarpum 6949 seed (11), 14? T. crispum 6939 aa (7), 8, 9, 10, (11), 13, 14?, 15 T. eucosmum 6972 ° (11) T. flexuosum 6953, - (12) T. howellii ssp. spectabilis 6958 eg 1, (11) T. integrifolium ssp. affine 68128 leaf (12) Lund-B seed (12) ssp. complanatum 6926, 6934, 6954, 6956 = (12) ssp. gracilipes 699. a (12) 3, 69 ssp. integrifolium 6967, 6973, 6974, 6977, 6995 ” (12) ssp. longicarpum 6991 - fi laciniatum 68152 is 4, (11 ), 15 6970, 6971, 6968, 6965, 6963 “ef 1, 4, (11), 15 6959, 6960 id 14, (it) 15, 17 68133 - 4, : ” leaf 4, (11), 14, 16, 17 T. milleflorum 68154, 68160, 6966 seed (1), (4),17 6961, 6962, 6964 (1), (4). 16, 17 T. repandum 6921, Hitchcock 14120 # {12} 6921 leaf 32), 34 T. rollinsii 6979, 6980, 6981, 6982, 6983 seed (7),11 T. sagittatum ssp. sagittatum 6919, 6925, 6929 . (1), (2), 4,5 ssp. ovalifolium 6984 zi (1), (2),4,5 ai stenopetalum 6947 ag (1),4 T. texanum Barclay 713, Cory 43996, Rollins 6176, Rollins 61106 se (12) T. wrightii 6992 a = : Correll 34106 ” (a1), (12) ” ” leaf (11), 14, 16 *Numbers correspond to those in Table 5. Compounds in parentheses are primary, others secondary constituents. 64 IHSAN A. AL-SHEHBAZ TABLE 7. Rp, VALUES OF THIOUREAS AND OXAZOLIDINE-2-THIONES* Solvent Systems Compound But. 3-Tol. 10-Tol. Chlor. Benz. Tol-Ac. Isopropylthiourea - 0.85 0.60 0.41 0.32 0.26 4-Methyloxazolidine-2-thione - 0.84 0.82 0.91 0.95 0.69 sec-Butylthiourea ~ - 0.96 0.74 0.72 0.62 4-Ethyloxazolidine-2-thione - - 1.13 1,18 1.44 1.34 iourea - 0.78 0.45 0.25 0.25 0.18 4-Methylthiobutylthiourea = ~ 112 0.96 1.02 0.82 4-Methylsulfinylbutylthiourea 0.42 0.08 0.0 0.0 0.0 0.0 4-Methylsulfonylbutylthiourea 0.54 0.15 0.03 0.03 0.0 0.0 3-Butenylthiourea - - 0.83 0.60 0.58 0.43 5-Vinyloxazoldine-2-thione - - 1.07 1.09 1.33 1.10 5-Methylthiopentylthiourea - 1.28 1.12 1.41 1.23 5-Methylsulfinylpentylthiourea 0.56 017 0.03 0.02 0.0 0.0 4-Pentenylthiourea ~ - 115 0.90 1.02 0.88 Ithiourea ~ - 1.18 0.92 1.11 0.82 2-Phenylethylthiourea ~ - 1.29 1.07 1.41 1.24 cs pound nged aie pened to Table 5. Solvent Systems: But. = 1:3:1 toluene-butyl alcohol-water; 3-Tol. = 3:1:2 os alcohol- water; 10-Tol. = 10:1:2 toluene-butyl alcohol-water; Chlor. = 5:3 Shan wether = Ble benzene-ethanol-water; Tol-Ac. = 5:2:4 toluene-acetic acid-water TABLE 8. RETENTION TIMES OF SOME ISOTHIOCYANATES Approximate Retention time range of variation Isothiocyanate Minutes Seconds (+ seconds ) Group I Isopropyl isothiocyanate 2 06 6 Ally] isothiocyanate 2 57 12 sec-Butyl isothiocyanate 3 44 12 Isobuty] isothiocyanate 4 19 14 3-Buteny] isothiocyanate 5 39 25 4-Penteny] isothiocyanate 10 45 45 oup if 3-Methylthiopropy] isothiocyanate 2 57 8 nzyl isothiocyanate 3 45 15 4-Methylthiobuty] isothiocyanate 4 55 17 2-Phenylethyl isothiocyanate 5 57 22 5-Methylthiopentyl isothiocyanate 8 02 32 Group I was characterized at 80°C, group II at 160°C. BIOSYSTEMATICS OF THELYPODIUM 65 Two unknown volatile compounds were detected in the GLC analyses of seed samples of certain species of Thelypodium. One of them was present in very small quantities in some of the samples of T. wrightii, T. laxiflorum, and T. integrifolium. This compound is probably a cyanide derivative, but no attempt has been made to characterize it. The second unknown was sometimes present in concentrations close to 11% of the total volatile content, but was only found in the species that had high concentrations of 3-butenyl isothiocyanate (Table 15, compound 11). This compound is probably 3-buteny] cyanide (Dr. Martin G. Ettlinger, personal communication). Neither of the unknowns were included in the quantifications of the volatile isothiocyanate contents and will not be mentioned in the rest of the discussion. In the following, the terms isothiocyanate and glucosinolate are occa- sionally used interchangeably. However, it should be remembered that the isothiocyanates are the hydrolysis products of the glucosinolates. Thelypodium laciniatum is the most varied chemically of all the spe- cies of the genus. Five to nine glucosinolates have been identified in the seeds of each of the various population samples of this species. The major isothiocyanate-producing glucoside in this species is 3-buteny] glucosino- late (Table 9). It tends to be present in slightly higher concentrations in the populations of Oregon and Washington than in those of Idaho or California. The second significant component, which has also been found in all the population samples of this species, is sec-butyl glucosinolate (Table 9). The 4-pentenyl glucosinolate was a minor constituent in all the population samples but the three from Idaho. These three samples were collected from the central portion of the state, which is the eastern- most limit of the species range. The isopropyl glucosinolate was found in small amounts in all the samples but those from Idaho. Instead, the three from Idaho had traces of the allyl glucosinolate, which was not detected in the other samples of the species. The amounts of 4-pentenyl glucosinolate in the Idaho samples were higher than anywhere else in the genus. The isobutyl glucosinolate was present in very small quan- tities in T. laciniatum as compared to its concentrations in the closely related T. milleflorum. The benzyl, 4-methylthiobutyl, and 5-methylthio- pentyl glucosinolates were present either alone or in various combina- tions, but always in small quantities. They were not detected in some of the samples (Table 9). The distribution of these three glucosinolates does not seem to offer any recognizable pattern. Substantial amounts of 2-phenylethy] glucosinolate occurred in the three population samples from California. In Thelypodium milleflorum isopropyl and sec-butyl glucosinolates were the major constituents in all the samples analyzed. The first gluco- sinolate was the dominant compound in most of the samples, particularly those from the state of Washington (Table 10). However, in samples 66 THSAN A. AL-SHEHBAZ TABLE 9. ISOTHIOCYANATES OF THELYPODIUM LACINIATUM (%) 8 8 3 © S) 4 n a ieee ee ee ore eas Se me ae ee Boo 2S ee eG ake See oe Se 6 eS ee gS ae ee \ ee ae OE Boe ee Oe ee Collection No. State Somer ef oe ioe oo oo se 68132 California - 6 179 T 611° T° O08 30° 157 25 68133 ” 13 0 142 T 647. 22 06 45 99 26 6951 ” oe wer ns 8 TO a TS 6922 Idaho 5 fo S6°T 616 mA TO o8 6923 , O99 8s Fe 684 999.0 <8 er, 6924 ” OF 185 °T S842 901.0. 0 Of 68145 Oregon oe, 8 08 oT 8. GA. AO 8 6955 ” oe 0 307 2 780...1.9.00 0 G20 6959 ” toe ae ee S68 21 Bw 6960 ” 13.6 O66. Y 661 24 06 32 16.7 6970 . woe i665 7 765 SA 0 14° 1 8 6971 ” 7 8 ise fT wea 62. 88 G8 Howell 1880 ” aye hae ses TT 8 OF 8 Washington T 0 100 T 790 2.0 12 29 35 14 6965 iv orm 169 F704 Se TO aT Te 6968 - eo 76 Tens 1d Tt 26 Te T = trace, 0 = undetected 68154 and 68160 from Idaho and sample 2540 that was collected by Rollins and Chambers in Nevada, the concentration of the sec-butyl glu- cosinolate was slightly to moderately higher. Benzyl and 2-phenylethy! glucosinolates were both present in minor amounts, with more of the latter than the former. Isobutyl isothiocyanate has been recently reported by Underhill and Kirkland (1972) from the fresh parts of the crucifer Conringia orientalis (L.) Andrz. As found earlier by Kjaer et al. (1956), this compound is absent from the seeds. We now know that it is present in the seeds of two species of Thelypodium (T. laciniatum and T. milleflorum) and Cochlearia officinalis (Ettlinger and Al-Shehbaz, unpublished). Isobuty] isothiocyanate was a minor compound in all the samples of Thelypodium milleflorum but the concentrations were clearly higher than those seen in T. laciniatum (Tables 9 and 10). In his treatment of the genus Thelypodium, Payson (1923) recog- nized three varieties in T. laciniatum. These are vars. laciniatum, mille- florum, and streptanthoides. In the present treatment, however, var. milleflorum is considered a distinct species, while var. streptanthoides is not given any taxonomic recognition. Chemically, the two species are BIOSYSTEMATICS OF THELYPODIUM 67 TABLE 10. IsOTHIOCYANATES OF THELYPODIUM MILLEFLORUM (%) iS ” ered a oy O O ar) +> Baa * es 8 a Begs Ao 6 & s & - RB Com 8G oe lS Collection number State Oo a Oo m Cy st 68154 Idaho SED Od soe 4 3.0 68160 o 405 47.9 075 515: 7 Bs hy f Palmer 417 si 00.3 06 BOO d 6933 Nevada ma oe 5.0 50 sR: 78 Rollins & Chambers 2540 és oot bi: 61 jy 7. 68140 Oregon GOLG. S08 428-0 bs 6961 Washington S24 120 47 ° 0 dé & 0.9 6962 1 $4.0: 120° 24-36: 08 1.0 6964 sh 18.1;°-158 > 45°. 0 T 1.0 6966 - Tal 201 3A: 6 i 3.4 St. John 7650 ” 668 48h 15° 0 0° 6 T = trace, 0 = undetected very distinct in a number of ways. Thelypodium laciniatum is charac- terized by having high concentrations of 3-butenyl glucosinolate, low to moderate amounts of the sec-butyl glucosinolate, traces or none of iso- propyl glucosinolate, low to moderate amounts of 4-pentenyl glucosino- late, and bare traces of isobutyl glucosinolate (Table 9 and Plate 16). Thelypodium milleflorum, on the other hand, has high concentrations of isopropyl glucosinolate, and high to moderate amounts of sec-butyl glu- cosinolate. Measurable amounts of isobutyl glucosinolates were always present, but no detectable quantities of 4-pentenyl and 3-butenyl glu- cosinolates were found in any of the samples (except 68160 which needs further checking). Such significant differences in the chemistry clearly favor the maintenance of these two taxa as distinct species (Table 10 and Plate 16). In one locality (near Beverly, Washington) where they were growing near each other, Thelypodium laciniatum and T. milleflorum were as distinct chemically as they were elsewhere in their ranges (sample 6963 in Table 9 and sample 6962 in Table 10). Neither chemical nor morphological intermediates were found. It is clear that genetic isolation is present between the two species in nature. Variety streptanthoides was based mainly on the differences in flower color. However, this character does not seem to have any significance in the genus. Five samples (6959, 6960, 6965, 6968, and 6970 in Table 9) of Thelypodium laciniatum that clearly fall in var. streptanthoides sensu 68 IHSAN A. AL-SHEHBAZ 3 0) or UW) o ‘oh 68140 7p) oO “ 6979 c- © G 5 U 4 = Pc foe Oo Y) : O F od i) a 7 - 68160 D . a 6982 O U © , a t+. ms ee ee j . 4 8 min. A Seay wales ae Piare 16. Chromatographic tracings of of isothiocyanates of Thelypodium laciniatum (6959, 6923), 1 Peet saps pear agis and T. rollinsii (6979, 6982). 2 wl injected at 80° C. Numbers ao ee isothiocyanates derived from their respective glucosinolates in Table 5. BIOSYSTEMATICS OF THELYPODIUM 69 Payson are chemically indistinguishable from other samples of this species as far as the pattern of their glucosinolate distribution is concerned. It is clearly shown that the glucosinolates are very stable compounds by the comparison of the recently collected samples with much older ones obtained from herbarium specimens, such as those collected by Howell and Palmer in the years 1880 and 1895 respectively (Tables 9 and 10). One of the taxonomically most confused groups of taxa in Thelypo- dium is the T. integrifolium complex. As many as six species were de- scribed in this group. Payson (1923) recognized six taxa, including four species and two varieties. In the present treatment one species and five subspecies are recognized. Thelypodium rollinsii, a new species added to this complex, differs in many significant morphological features from T. integrifolium. The glucosinolate distribution agrees well with the taxo- nomic decisions. All of the subspecies of T. integrifolium have one major glucoside, 2-hydroxy-3-butenylglucosinolate. No volatile isothiocyanates have been detected in this species (Table 6). Thelypodium rollinsii has a completely different pattern of glucosino- lates (Plate 16) from that of the closely related T. integrifolium. The major isothiocyanate-producing glucoside in this species is allyl gluco- sinolate, forming over 70% of the total. Small amounts of 3-butenyl gluco- sinolate were found in the three samples from Juab County, but signifi- cantly higher concentrations were present in samples 6982 and 6983, collected from two other counties of Utah, Sanpete, and Sevier respec- tively (Table 11). These two samples contained slightly smaller amounts of allyl glucosinolate than the other three. Other differences are present (Tables 6 and 11) but it is clear that we are dealing with two signifi- cantly different patterns of glucosinolates that justify, from the chemical point of view, the recognition of two distinct species. Thelypodium laxiflorum has been considered a variety of T. wrightii under the name tenellum since 1895. However, the two taxa are so ex- tremely different in their lower and fruit morphology (see the taxonomic treatment and Plate 22) that they are best treated as distinct species. The major compound in T. wrightii is 2-hydroxy-3-butenyl glucosinolate. This was not detected in T. laxiflorum (Table 12). The 3-buteny] isothio- cyanate was found in measurable amounts in sample 34106 but not in 6992. This compound was the major mustard oil in T. laxiflorum, constitut- ing more than 80% of the total isothiocynate content in the seeds (Table 12). The remainder included small amounts of isopropyl, allyl, sec-butyl, 4- pentenyl, benzyl, 4-methylthiobutyl, and 5-methylthiopentyl isothiocya- nates, none of which were detected in T. wrightii. Unfortunately, we are dealing with a very few samples which may not reflect the main variability in the two species. Nevertheless, the two species appear to have two distinct patterns in their glucosinolates which is solid evidence against uniting them in a single species. 70 IHSAN A. AL-SHEHBAZ TABLE 11. IsOTHIOCYANATES OF THELYPODIUM ROLLINSII (%) 6 x” S A, ES S O O pase 5 ro 5 O ll ne ll Oo, xq 1e} Collection number County in Utah = cs ot 6979 Juab 93.9 T 4.4 6980 “s 95.2 i 3.8 6981 . 94.5 é 4.) 6982 Sanpete 74.1 ~ 24.7 6983 Sevier 73.6 T 25.3 TABLE 12. ISOTHIOCYANATES OF THELYPODIUM LAXIFLORUM (%) Se oe S| CCH Nes os Ree oe eS SS O Os pee ee A ee ee eae Ae fc Te, eee gaa 2 2 ad Poa et a Collection number State a ae} " co (7) mm. long; filaments (4.5)6.5- 14(15.5) mm. long; —_— pedicels sto D. Fruiting pedicels straight, horizontal; a solid, not inflated; basal and lower cauline leaves sidaails obed to laciniate; ep es of basal leaves glabrous; 2 spreading; petal blades linear, 0.3-0.7(1 5) mm. wide; widely distrib- pn Ho ee es A eer ye 1. T. laciniatum. D. Fruiting pedicels strongly curved upwards, with the tips usu ; stems laciniate or pinnately lobed; _ of basal leaves ciliate at least near the base; siliques erect to divaricately asc ending, often — or stagnet gai appressed to the fade petal blades oblanceolate to spatulate, 1-2 mm. ee: —— ag WOME ee eh RE Oe a a C. Inflorescence dense to lax, corymbose to short racemose, 0. Rog 5 (0.7) ger long; sepals spreading; petals spreading to erect, usually not arma into claw and blade, or if so, then claws cylindrical, slender, and uni width claws 1-2.5(3) mm. long; filaments erect to spreading, 2.5-6.5(8. 3) mm. long; fruiting pedicels slender. E. Sepals ascending, sometimes spreading or erect; — nts and — bases erect; siliques submoniliform, terete, replum mostly tricted een the seeds; stems pubescent near the base to glabrous; Utah: Novia, and western — ee re Be a 14. T. laxifloru E. widely spreading; filaments and ei spreading; siliques torulose, mst flattened parallel to the septum, replum straight; stems glabro ting pedicels filiform or at a very slender, divaricately “ascending, 235 cm. long, southwestern WN a ee: 17. T. tenue. F. Fruiting pedicels rigid, slender to somewhat stoutish, widely divaricate to Mss ge ee ed, 04-1. 4(1.7) cm. long. G. Flowers levensies to purple; atk (2.5)3-4.5 mm. long, 0.5-1 mm. wide; filaments 2.5-3.75 mm . long; basal and lower cauline leaves greyish, fleshy, ovate or less often obovate to elliptic, repand or sometimes ea or lyrate; ee wee 15. T. repandum. 82 IHSAN A. AL-SHEHBAZ G. Flowers . white, rarely lavender See a petals (3.5)4-7.5 (9) mm. long, 1— wide; filaments (3)4-6 5) mm. long; basal and lower cauline leaves sarally green, thin to mcnranog thickish, lanceolate or oblan- ceolate, mostly pinnately lobed, rarely lyrate. H. Plants biennial; styles clavate to subclavate; uppermost cauline leaves northedn Mexico, New Mexico, ssontiwestenn Oklahoma, and southwestern is. oa nie 6 GER Se UB SOO GE: AR ER a na NN ranSi a cate Rue en ert T. wrightii. H. Plants annual; styles conical to subconical; uppermost cauline leaves mostly pinnately lobed; stems 1.3-4.8(6.1) dm. high; petals spatulate to oblanceo- late, attenuate to short cap ene base; southwestern Texas . ie num. aul, or auriculate at the base, lower and middle leaves mostly cai aE entate; — nig styles c lindrical or more d I. Plants queens caudex woody, thick, covered with papery remains of petiole bases of previous ee seasons; stems rather weak, usually flexuous, few lea ved eae alte ete Lo Ne MR Ree AR Ree a ha ae RE el een ee ee me eae ee bo het Owl we et be ae a ee eee uosum. I. Plants biennial, or if perennial, then rootstock slender, nearly naked; stems erect, slender or stout, few to several leaved; basal leaves usua withering early in the on, or if somewhat . then petioles ciliate near the base. ending, not ina hive to s kom, owe se, a anil ing, auricles a raed ae to reduced; median glandular tissue present or abse: K. Terminal eck of the inflorescence racemose, elongated to somewhat short- ened, lax to congested. ovate to lanceolate- ovate, acuminate: sone plum UL 3)1.5-2 mm. long, : (0.75) 1( 1.25) mm. wide; Oregon and northern California pe es Secs T. brachycarpum. L. Inflorescence mostly a lax raceme; flowers lavender to te rarely white, fruiting pedicels widely — or ascending; petal blades not crisped or only neg at their Lvactees with the ilaments paired nace pets to completel sitar Oregon and assisting tomcat Gti ee ee .. a ig ne T. howelli = Filaments of paired stamens completely O. Petals narrowly linear, 0.3-0.5 i cons en mm. isk floral buds linear to nar- rowly apa aoiitern Colfomis 2 5s 8. T. stenopetalum. oO. = narrowly spatulate to seakahas oblanceolate, (0.5)1-3(4) mm. wi fecha buds m pared ovate to lanceolate, if linear or narrowly oblong, rescence s congested and P. Floral buds narr prea ie peta at k p ly oblong to linear. oO hore ()25-6(75) mm. ; perennial; anthers completely ware ye : BIOSYSTEMATICS OF THELYPODIUM 83 P. Floral buds ovate to lanceolate; gynophore (0.2)0.5-1(3.5) mm. long; almost always biennials; anthers often partly to completely included. siliques ascending, straight, pedicels and siliques —— in a straight line; weet sea oblong to Byars: hors si eastern Orego . 9. T, howellii. wabh upwards, often slightly flattened at the ae siliques straight to curved, usually with erect tips, siliques and pedicels usually forming an arc; “ap Be leaves er vate to oblanceolate; —. oo but absent from Orego sagittatum les inflorescence n wie, pe elongated in fruit; siliques slender, strongly incurved, young siliques extending above the top of inflorescence; stems _(4)6-16(20) m. high; floral buds oblong; central Utah... . rollinsii. R. Cauline leaves ascending, oblong to lanceolate; basal leaves vied (2)6— 20(29) cm. long, (0.6)1-4(5) cm. wide, petioles ciliate near the base; anthers ly e what curved, young siliques not extending above the top of inflorescence; stems (1.4)2-8(12) dm. ae floral buds ovate. T. Siliques (0.5)0.75-1.( 1.2) mm. wide; petals (0.5)1-3(4) mm. wide; median glandular tissue mostly absent; seeds somewhat flattened, (0.75) 1-1.3(1.5) mm. dong; widely: distributed (200. 10. T. sagittatum. dy sree (1.3)1.5-2.3 mm. wide; petals 2.5-5(6) mm. wide: median andular tissue mostly present; seeds plump (1.3)1.5-2 mm. long; western snore northeastern ‘lads and adjacent Montana __.. 11. T. paniculatum. s eas meee laciniatum (Hooker) Endlicher acropodi Walkestai and at Priest’s Rapid on ‘the genset iy Douglas $.n. te. not seen Pachypodium laciniatum ( Hook.) Nutt. in Torr. & Gray, Fl. N. Am. 1:96. 1838. type: U Thelypodium laciniatum (Hook.) Endl. var. saestholiie (Leib.) Pays., Ann. oe 1923. Bot. Club 343433. 1907. Holotype: lypodium leptosepalum Rydb., Bull. Torr. Bot. Clu olo Lewiston, Frees ace , Idaho, May 6, 1896, A. A. & E. G. Heller 3022 (nv); iso : MO,ND-G,U Plants ee int erect, often virgately branched above the base, sometimes simple or branched from the base, glabrous and somewhat glaucous, solid, (1.3)2.6— 4.5) dm tus tween lobes usually extending to midrib below and becoming gradually i toward leaf apex; petioles somewhat expanded at base, aaa (1)1. — ) cm. long; cauline leaves reduced upwards, lanceolate to linear-lanceolate, often acuminate, entire to sinuately toothed or laciniate; eka a densely flowered raceme, rarely lax, slightly elongated in fruit, usually branched, (0. eee ee “es ag dm. long; calyx somewhat urceolate to rarely tubular; sepals creamy white above 84 IHSAN A. AL-SHEHBAZ d green below or lavender to purple throughout, glabrous, oblong to linear- oblong, rarely oblong-lanceolate, obtuse, scarious at margin, becomin ha and spreading above, filaments of single stamens strongly curved at the base; anthers well exserted above the calyx, linear, rarely oblong, usually purple, apiculate, i t i 5.5 d ? lar tissue continuous and disc-like, slightly raised above the receptacle, subtending i filame of the paired filaments and the ophore; iting pedicels horizontal to rarely divaricate, straight or sometimes slightly curved, usually stout, often flattened at the base, (2)3-6.5( 15) mm. long; siliques somewhat flattened parallel to septum to terete, rarely 4-sided, torulose, straight to curved, sometimes irregularly flexuous, i preading to divaricately ascending, rarely recurved, (2.5)3.4—10(12.1) cm. -5(2) mm. wide; op ong; styles stout, cylindrical to occasionally subclavate, (0.3)0.75-2.5(4) mm. long; stigmas entire; seeds (0.75) 1-1.5( 1.75) mm. long, 0.5-0.75( 1 ) mm. wide; cotyledons oblique, very rarely accumbent. istributed in northern and western Nevada, northern and eastern Cali- fornia, Oregon and western Idaho, northward through southern and central Washington to southern British Columbia (Map 1). REPRESENTATIVE SPECIMENS. Canada: British Columbia. Lake Osoyoos, between lat. 40 : and lat. 49°05’, long. 119°20’ and long. 119°35’, Macoun 70851 (F,CH,NY ). Alpine Co.: 4.2 miles E. of Woodfords, Breedlove 3587 (cH); Inyo Co.: s Fish Springs in t Canyon near mouth of Sheep Creek, Meyer 263 ke River, Packard 410 (cGH,Mo,UC, na. Us,ws); Lemhi Co.: ca. 2 miles E. of Indiano Ranger Station, north bank of Salmon ga Hitchcock & Muhlick 14320 Ss -: Victory & Howell 238 (cas,cu); Owyhee, Jones 25314 (CAs,Ds,Mo,ND-c,PpoM ); Eureka Co.: P. alisade, Jones 3767 (CAS,Ds,MO,NY,POM,RM,UC,US,UTC): Humboldt Co.: S. & E. of ey Summit of Santa Rosa Range of Humboldt National Forest, Constance, Molseed BIOSYSTEMATICS OF THELYPODIUM CANADA California/ e ~ © © $ F @ © 2, : ~ CO ae ae e © Nevada Utah e 4 ae 3 e Thelypodium laciniatum oT. milleflorum O 100 200Miles Map 1. Distribution of Thelypodium laciniatum and T. milleflorum. 86 IHSAN A, AL-SHEHBAZ & Ornduff 3728 (NY,TEX,UC,uTC); Ormsby Co.: Carson City, Jones s.n., 1897 (cas, MO,POM,RM,US,UTC); Storey Co.: 3% miles N. NW. of 5-mile House, Adams 158 (Rsa,uc); Washoe Co.: Pyramid Lake, Kennedy 1003 (ny,nM,us); Badger Mountain, Sheldon Antelope Range, Murie 2868 (NA). Oregon. Baker Co.: 22 miles N.E. of : wy. 26 and Oregon Hwy. 7 intersection, Davidse ¢> Collotzi 514 (NY,UTC,wTv); Crook Co.: along Rimrock above Crooked River Valley, % mile W. of Prineville, : uglas C Yoncalla, Gale 187 (ps); Gilliam Co.: 5 miles W. of Blalock on U.S. Hwy. 30 along Columbia River, I. ¢& M. Al-Shehbaz 68145 (cH); Pine Creek, Leiberg 172 (cH.Mo, NY,ORE,POM,us); Grant Co.: walls of John Day Canyon, Henderson 5295 (cas.ps, avatar 7 miles from Vantage Bridge on road to Quincy, Howell 28470 (cas,cu, > S' 7 miles N. of Wenatchee, Hitchcock 17363 ( COLO,Ds,GH,NY,RM,RSA,UC,US,UTC,WS,WTU); Douglas Co.: between Coulee City and Waterville, Spillman s.n., 1896 (ws); Franklin Co.: Palouse F. alls, Daubenmire 37236 (ws,wru); Co.: Wawawai arfield Ferry, St. John 6076 (uc,ws); Grant Co.: 8 miles N. of Dry Falls, Rogers 599 (cas, )NY,POM, »WIS,WS,WTU ); near L CH,MO,NY,PH,POM,wTu ); Kittitas Co.: near Ellensberg, Jones 10161 (wis); Klickitat 0.: Lyle, Suksdorf 841 (DS,F,MO,NY,UC,US,Ws ); Okanogan Co.: W. of Hess Lake, Fiker 895 (Ds,Us,ws); Walla Walla Co.: Wallula, Booth s.n., 1942 (ws); Whitman ank of Snake River opposite Clarkston, Constance, Clarke, Dillon, Machlis & ,OSC,PH,RM,UC,US,UTC,WIs,ws,wTU ); Almota, Constance, Clarke, Machlis & Rollins 1077 (CaS, F,CH, JEPS,MO,NY,OSC,PH,RM,UC,UTC,WIS,WS,WTU ) ; Yakima Co.: mouth of Selah Creek, E. of US. Hwy. 97, Witt 1929 (ny,ws,wtv). e calyx is said to be saccate in T. streptanthoides and not so in T. matum ( Hitchcock, 1964). However, this character is also found to be BIOSYSTEMATICS OF THELYPODIUM 87 various color intensities have been found. It is deemed wise, therefore, not to give any taxonomic recognition to the purple-flowered form. Thelypodium laciniatum appears to grow primarily on silty (rarely sandy) soil in crevices of basalt cliffs or outcrops along roadsides and canyon walls. Occasionally it grows in association with sagebrush on gravelly or rocky hillsides. In addition to the common occurrence of this species on basaltic rocks, there are scattered collections made from “limestone ledges” (St. John 3477), “crevices of sandstone rocks” (Rollins 57349), “on serpentine rock and talus” (Hitchcock & Muhlick 8482), and “sandy bank” (Hitchcock & Muhlick 14320). Thelypodium laciniatum shows great variation in basal leaf morphology. , Such variation can usually be seen within any sizeable population. Plate 4 shows the variation in basal leaves of two population samples grown under uniform greenhouse conditions. The siliques show a considerable amount of variation in length. Silique width is much less variable. Plants of northern California tend to have broader siliques than those in other parts of the species range, but such a tendency does not provide the basis for division of the species into subspecific taxa. Thelypodium laciniatum is most closely related to T. milleflorum which it resembles in many important morphological features. The latter has been considered to be a variety of T. laciniatum since Payson’s treatment of the genus in 1923. The evidence presented by Payson to support such a place- ment is misleading. He claimed that the two species intergrade, and suggested that, since they have similar geographic distributions (Map 1), they are best treated as varieties of the same species. In fact, the two species are very distinct ecologically, and rarely grow in the same habitat. Thely- podium laciniatum tends to grow mostly on silty soils in crevices of basaltic cliffs or rocky outcrops, while T. milleflorum almost always grows on sandy soil in open desert or along river banks. The writer has collected the two species growing a few feet apart in one locality (near Beverly, Washington) where T. milleflorum was growing on sand accumulated at the base of a cliff along the Columbia River, and T. laciniatum was very abundant in the crevices of the same cliff. No hybrids were detected. Moreover, the two species found in this locality are chemically very istinct as they are elsewhere in their ranges (see the section on bio- chemical systematics). Of the two species, T. milleflorum appears to be more advanced morphologically than T. laciniatum from which it may have been derived. The overall resemblances of Thelypodium laciniatum to certain species of the closely related genus Stanleya have been clearly pointed out by Rollins (1939). Several morphological features of the leaves, inflorescences, petals, stamens, nectar glands, and siliques are strikingly similar in these taxa. On the basis of such similarities, Payson (1923, p. 244) had sug- gested that Thelypodium may have been derived from Stanleya. 88 IHSAN A. AL-SHEHBAZ Thelypodium laciniatum is easily distinguished from aT. milleflorum by its solid stems, basal and lower cauline leaves pinnately lobed to laciniate, glabrous petioles, fruiting pedicels straight and horizontal, siliques spreading to divaricate, and linear petal blades. In typical T. milleflorum the stems are almost always hollow, often slightly to moderately inflated below, the basal leaves and lower cauline leaves are often coarsely toothed to sinuate and rarely laciniate or pinnately lobed. The petioles of the basal leaves, and occasionally the lower cauline leaves, are ciliate at least near the base. Fruiting pedicels are strongly curved upwards and the siliques are erect to divaricately ascending, often partly or completely appressed to the rachis. Finally, the petal blades are mostly oblanceolate and rarely spatulate or narrowly oblanceolate. Most of the above morphological differences have been, unfortunately, overlooked. The morphological, ecological, and chemical differences are clearly sufficient to maintain the two taxa as distinct species. This interpre- tation is further strengthened by the fact that no hybrids were detected in an area where the two species were growing together. 2. Thelypodium milleflorum Nelson Thelypodium milleflorum Nels. Bot. Gaz. 52:263. 1911. Holotype: New Plymouth, Payette Co., Idaho, June 10, 1910, J. F. Macbride 234 (nm); isotypes: Ds,¥,GH,MO, NY,UC,US,WS. ium laciniatum (Hook.) Endl. var. milleflorum (Nels.) Pays., Ann. Mo. Bot. Gard. 9:274. 1923. Plants biennial; stems erect, single from the base, often branched above, glabrous and glaucous, almost always hollow at least to the inflorescence, often slightly to moderately inflated above the base, usually purple below and creamy white above, (1.8)4.5-13(21) dm. high; basal and lower cauline leaves petiolate, glabrous and margin often variously dentate to sometimes pan i pinnately lobed, (3.8)6-23(28) cm. long, (0.9)2-6.5(10) cm. wide; petioles of ; 2 mm ; . sh ,e low, somewhat g above, nearly equal in length, (6)7-14.5(15.5) mm. long; anthers linear to narrowly oblong, apiculate, Sagittate at base, circ y co when fully , we © often purple, 2.5-4.5(6) mm. long; glandular tissue continuous, slightly rai above the rece , subt of paired stamens, surrounding base of single stamens, more developed opposite petal insertions; fruiting pedicels slightly flatten — erence curved upwards, with the tips erect, usually stout, (1.5)2.5-5(7) mm. long; siliq hat flattened p Hel to septum, rarely terete, narrowly linear, BIOSYSTEMATICS OF THELYPODIUM 89 torulose, erect to ascending or sometimes divaricately ascending, often partly or completely appressed to rachis, straight to somewhat irregularly aioe (1.8)3.3- 8.5(10) cm. long, (0.75)1-1.5(1.75) mm. wide; gynophores wea (0.5) 1-4(6) mm. long; styles stout, phir to siete subclavate, (0.5)0. 5(3) mm. long; stigmas entire; seeds (0. 75)1-1.5(2) mm. long, 0.5-0.7 5(1) mm, on cotyledons oblique, very rarely accumbent. Flowering from middle April through early August but mainly early May to early July at elevations from 500 to 7,100 feet in western and south- ern Idaho, northwestern Utah, northern and central Nevada, northeastern California, Oregon, southern and central Washington (Map 1). ‘ATIVE SPECIMENS. Canada. British Columbia: Hope, Anderson 732 (ws); Seema Aid 733(ws). California. Lassen Co.: 2 miles N. of sie on N. side of Honey Lake Valley, Babcock & Stebbins 1782 ee: Mono Co.: along Deep Wells road S.E. of Mono Lake, gd & 4, T.2.N., BR. 26-E., J. & A. Reveal 412 (utc). Idaho. Ada Co.: along hwy. E. of Boise, Cronquist 149.56 (utc); Bannock Co.: Pocatello, Blankinship s.n., 1902 (cu,RM); Bingham Co.: Blackfoot, Jones s.n., 1909 (vom); Bonneville Co.: Idaho Falls, E. & L. Payson 1728 (CAs,GH,MO,NY,RM, uc); Butte Co.: 2 miles N. of Butte City, Hitchcock & Muhlick 22753 (ps,Ny,osc, Ys us); Canyon Co.: Caldwell, Tucker 609 (ny); Cassia Co.: Malta, Cottam "3008 (uT); Clark Co.: 9 miles $.W. of Dubois, Hitchcock & Muhlick 22785 (coLo,Ny 1Ws ,wTv); j g Co.: jet. of U.S. Hwy. 26 and Idaho Hwy. 23 below Carey, L ri M. Al-Shehbaz 68154 (cH); Minidoka Co.: Heyburn, Christ 9499 (Ny); Payette Co.: 5 miles S.W. of Fruitland, NY (Ny,wru); Washington Co.: Olds Ferry, Christ 9378 (Ny). Nevada. Chure 1 mile W. of sia Mills ¢+ Beach 664 (Na); Elko Co.: 10 miles S. of Montello on Nevada Hwy. 30, I. ¢ M. Al-Shehbaz 6898 (cu); 29 miles N.E. of Elko on Inter- state Hwy. 80, I fog M. Al-Shehbaz 68104 (cH); — pio = 1891 (cu, ,TEX,UC,US,ws,wtTu ); Eureka Co.: 14.3 miles ka on Nev: ada Hwy. 51 from its jct. with U.S. Hwy. 50, I. & M. ‘AL Shehbaz 68106 5 (cit Palisade, Jones 3772 (cAs,Ds,NY,POM,UC,US,UTC); 5 miles W. of Eureka, Rollins & Chambers (Ds,GH,RM,UC,UTC ); Humboldt Co.: 11 miles W. o mliesaonente on Nevada R. 36 E., Sec. 28, N. & P. Holmgren 3964 (cH); 13 miles N. of Sulphur on road to Quinn River Crossing, T. olm, 3 (cH); Lander Co.: 3 miles N. of Nye and Lander Co. line, mouth of gg Canyon, Train 2765 (ps,Na,Ny); Lyon Co., igs miles N.E. of Dayton on U.S. Hwy. 50, W. Weber & Rose 8404 (coo,ps,RSA,TEX,UC,wtu); Nye Co.: near Ione, Beach 62 (Na); Ormsby Co.: Eagle Valley, ee 1020 ( F,GH,MO,NY,POM,RM,UC us); Pershing Co.: 4.5 miles N. of Lovelock, Alexander ¢> Kellogg 4606 (ps,uc,utc); Washoe Co.: 3.9 miles $.S.W. of Pyramid, "Simontacchi 508 (ms; uc). Oregon. Gilliam Co.: 5 miles W. of Blalock, Raven 18388 (cH); Harney Co.: 26 miles S. of Folly Farm in T. 32 S., R. 35 E., Hitchcock & Muhlick 21129 (ps,NY,RM,ws,wTu ); Lake Co.: = miles N. of Paisley, Leach s.n., 1927 (ore); Malheur Co.: -. miles S.E. of Vale in T. 19 S., 43 E., S. 23, Cronquist 8362 (Ny,uTC,ws,wTu); Morrow Co.: near Lexington, Leiberg 12 (onx ,Us); Umatilla Co.: 10 miles E. of Aramtaad Hills 18 (osc); Wasco Co.: along Columbia River Hwy. E. of The Dalles, A A. & R. Nelson 612 (ps,Mo,RM). Utah. Box Elder Co.: 12 miles W. of Snowville, Jensen 137 (utc). Washington. Adams Co.: oe Sandberg & Leiberg 190 (GH,NY,ORE,UC,US,ws); Benton Co.: 2 | les N.W. of Richland, Rose 50092 ( CAS,COLO,NY,RSA,UC,US); Douglas sar ( : S. of McAdam, Daubenmire 3839 be? Grant : 2.9 miles S. of R.R. Bridge on Columbia River near Beverly, I. & M. Al- Shekbac 6962 (cH); near Euphrata, —* IHSAN A. AL-SHEHBAZ Thompson 6120 (ps,cH,Mo,osc,wtv): Kittitas Co.: near Wenatchee, Whited 1246 (us,ws); Klickitat Co.: Bingen, Suksdorf 8208 (ws); Walla Walla Co.: Wallula, Leckenby s.n.. 1898 (cu,us,ws); Yakima Co.: near Yakima, Thompson 11315 (cas, GH,MO,NY,PH,POM,UC,US,WTU ). The most unique feature of Thelypodium milleflorum, that is not present in any other species of the genus, is the predominant occurrence of hollow stems that are slightly to moderately inflated above the base. The stems are neither inflated nor hollow in all the other species of Thely- podium. There the pith tissue is either solid and continuous throughout or loose to irregularily vacuolated. Older stems are occasionally hollow due to the breaking up of the pith tissue. Over 500 plants of T. milleflorum were checked for this feature. All but two were consistently hollow, but there is some variability as to how much of the stem is hollow. In some plants the stem is hollow only to the inflorescence, while in others the rachis and the branches of the inflorescence are hollow for most of their lengths. The most conspicuous variation in Thelypodium milleflorum is in the margin of the basal and lower cauline leaves, and the length and orienta- tion of the siliques. Although the leaf margin tends to be mostly dentate or sinuate, there are varying degrees of transition to pinnately lobed leaves. However, the latter type is rather uncommon, and seems to be restricted to the northern parts of the species range. Basal leaf morphology appears to be less variable in this species than in T. laciniatum. The siliques are quite variable in terms of their orientation and length. Such variation can be seen within any sizeable population. The siliques in T. milleflorum tend to be erect and somewhat appressed to the rachis, but ascending siliques, that are only appressed or parallel to the rachis near their bases, are by no means uncommon. The proximal parts of the petioles of the basal leaves are almost always Thelypodium milleflorum grows predominantly on sandy soils, and it rush. It is rather common on sandy dunes in banks. No other species of Thelypodium occurs BIOSYSTEMATICS OF THELYPODIUM 91 3. Thelypodium crispum Greene ex Payson Thelypodium crispum Greene ex Pays., Ann. Mo. Bot. Gard. 9:264. 1923. Holotype: Eagle Valley, 1446 m., Ormsby Co., Nevada, June 28, 1902, C. F. Baker 1191 (RM); isotypes: CAS,GH,MICH, MO,NY ,POM,RM,UC,US. Thelypodiopsis crispa (Greene ex Pays.) O. E. Schulz, Bot. Jahrb. 66: 98. 1933. agi Sage brachycarpum Torr. var. crispum (Greene ex Pays.) Jeps., Fl. Calif. 2:40. SNe io leptostachyum Greene, West Am. PI. 1:19. 1902. This name was a nomen nudum, therefore fe validly published. Thelypodiopsis leptostachya (Greene) O. E. Schulz, Bot. Jahrb. 66:98. 1933. This was based on the invalidly ‘editsehed: Thelypodium leptostachyum, thus the transfer ss Schulz also was in error. Herbaceaus biennial or short lived perennial; stems solid, usually slender, often glauco ous, glabrou: us throughout or sparsely to sometimes densely pilose toward the b : vecossainonig to spatulate, rarely obovate or lanceolate, often pinnately lobed to lyrate, metimes sinuate or dentate, rarely entire, (1.5)2. 2-15(25) cm. arth (0.6) 1-3.5(5) cm. wide; lobes entire, linear to oblong or rarely deltoid, acute at apex, remote and narrower on lower part of the leaf, with the divisions between the lobes becoming shallower toward leaf apex; petioles slender, mostly ciliate, (0.7)2-8.5 cm. long; cauline leaves sagittate to auriculate at base, linear to linear-lanceolate or lanceolate. rarely oblong or ovate, usually appressed most of their lengths to the stem and with € upper parts ascending, acute to acuminate at apex, mostly entire, nes dentate or _Tepand, somewhat eleven, wy labrous, or sometimes sparsel fe pil ose —— (11) m long rescence a densely ered raceme, sometimes a little lax, a slightly in fruit; sepals oblong to linear-oblong, obtuse, somewhat scarious at n, greenish above and white below or lavender throughout, often slightly keeled, saceate at the base, (3)3.5-6(8) mm. long, (0.75) 1-1.75(2) mm. wide; petals w r lavender with purplish veins, mostly linear to narrowly oblanceolate ponies pata = strongly crisped, (6)6.5-11(14.5) mm long, 0.5-0.75(1 - long; filaments year or ~~ lavender, abaxially; glandular tissue only surrounding base of single stamens, sometimes failing to develop | beneath them, usually with a lateral extension; fruiting pedicels mostly spe not flattened at base, erect or ascending-erect, partly or ora wae 4 appressed to rachis, (1.5)2-5(10) mm. long; sliques terete, torulos e, often incurved, sometimes s traight. divaricately ascending or occasionally as siadide —- coe oper pion poy e apex oo ened mm. we wide; styles slender, eoninet owar per sigan flattened, 1-1.5( 1.75) mm pass .5-0.75(1) mm. wide; sayledons mostly obliquely incumbent, rarely incumbent Flowering from early June to early August in eastern California and adjacent western Nevada (Map 2). The usual altitudinal range of Thely- podium crispum is from 4,000 to 8,500 feet, but it has been collected from elevations as high as 10,500 feet near Moat Lake in California (Smith 335, cas ). This is ies highest known altitudinal limit reached by any species of Thelypodiu 92 IHSAN A. AL-SHEHBAZ M,uc); Hot Creek in Long Valley, Train s.n., 1937 (coLo,Ds,NA,POM,uC,Us ); Lassen Co.: 5 miles S. of Ravendale, Harris 24965 (cu); 3 miles N. of Omira, Howell 11859 (CAs,F,POM); Milford, Ripley & Barneby 5684 (cas,ny); Mono Co.: Bodie Creek, Alexander & Kellogg 4373 (ps, oo Howell 14389 (cas,ps,F,GH,MICH,NY,US,wtu); 1 mile E. of Bridgeport, Munz 11900 ( COLO,Ds,NY,RSA,UT,wTU); Gem Lake, Peirson 6112 (rsa); June Lake, Rose 40729 (mo,us); below Moat Lake, Smith 335 (cas); Placer Co.: hot springs at McFarland Mill, Sonne s.n., 1886 (¥,uc); Plumas Co.: Vinton, Rose 55149 (CAs,COLO,NY,RSA,UC); Sierra Co.: along road to Beckwourth from Calpine, Bacigalupi, Mason & Sweeney y RSA,UC,US,UTC,Wis ); Ormsby Co 8 mile on hot springs road from U.S. Hwy. 395 at Carson City, I. & M. Al-Shehbaz 68110 (GH); near Carson City, Archer 6063 (N i Thelypodium crispum has been confused with T. brachycarpum for a long time, because the two species are very similar in many morphological features. In their treatment of T. brachycarpum, Gray (1865), Greene (1891), and Robinson (1895) stated that this species occurs in western Nevada and eastern California. However, we know that it is restricted to northern California and Oregon, and that the plants from western Nevada and eastern California that were misidentified as T. brachycarpum are, in fact, T. crispum. Thelypodium crispum actually was described by Payson from Baker’s collection (no. 1191) that was labelled “Thelypodium crispum Greene.” The name, therefore, must be written T. crispum Greene ex Payson instead of the customarily used T. crispum Greene. Although there is considerable variation in a number of morphological features of Thelypodium crispum, the most noticeable is that found in the margin of the basal leaves. The majority of plants have lyrate to pinnately divided leaves, but a number of individuals have the leaves nearly entire to variously dentate. However, such variation is of a continuous nature and usually may be found within the same population. Plate 4 shows the variability in the basal leaves of two population samples. Each leaf was taken from one of the rosettes grown under uniform green- house conditions. Cauline leaves are more variable in shape than is their margin. Typically, they are li margin, but there are populations with sporadic occurrence elsewhere in the m, the sepals and petals of T. crispum BIOSYSTEMATICS OF THELYPODIUM 93 can be white or less often colored, but the flower color does not correlate with any other character nor does it have any geographical basis. Finally, the fruiting pedicels are quite variable in their length, but such variation can easily be seen in any sizeable population. Thelypodium crispum grows predominantly on loamy to sandy clay soils in meadows or flats of various degrees of alkalinity. It is fairly common on mineralized soils (often disintegrated travertine) in the vicinity of hot springs at a number of localities in eastern California. a Washington Oregon © wo evada .. @ N © a ry J a aThelypodium crispum eT. brachycarpum % oT stenopetalum @ \ *I- eucosmum . AT. howellii SSP. howellii eT. howellii ssp. spectabilis California 0 100 200miles Map 2. Distribution of Thelypodium crispum, T. brachycarpum, T. eucosmum, T. howellii, and T. stenopetalum. 94 THSAN A. AL-SHEHBAZ The species most closely related to Thelypodium crispum is T. brachy- carpum. They are so similar to each other in the morphology of their basal and cauline leaves, inflorescences, siliques, and a few features of the flowers that they might be considered conspecific. In fact, Jepson (1936 ) treated the former species as a variety of the latter. However, the two have been maintained as distinct species by a number of authors since Payson's treatment of Thelypodium in 1923. Thelypodium crispum is also somewhat related to T. eucosmum, which it resembles in some aspects of the flowers, inflorescences, and siliques, but the two are sufficiently different in a number of features that they should only be indirectly associated. Thelypodium crispum can be easily distinguished from T. brachycarpum on the basis of the fruiting pedicels that are erect to ascending, usually slender, not flattened at the base, and with a length of (1.5)2-5(10) mm. In typical T. brachycarpum the fruiting pedicels are often horizontal to occasionally divaricate, always stout, flattened at the base, and only 12 (2.5) mm. long. From T. eucosmum, T. crispum is easily distinguished by its erect or ascending instead of horizontal fruiting pedicels, linear, white instead of oblanceolate, dark purple petals, appressed instead of divaricately ascending leaves, shorter siliques, and narrower divided basal leaves. 4, Thelypodium brachycarpum Torrey Thelypodium brachycarpum Torr., Bot. U.S. Expl. Exped., Plate 1, 1862; Wilkes’ Expl. Exped. 17:231. 1874. Holotype: Wilkes’ Expedition, between Clamet and San Francisco, no. 1596 (us). Two specimens of the Wilkes’ Expedition were seen branched from base, usually branched above, somewhat glaucous, glabro throughout or § ly to densely pubescent near base, rarely pubescent to the inflorescence, (1.3)3.4-8.3( 12) high, trichomes simple, often long and spreading; basal leaves thickish, glaucous, glabrous or sparsely to densely hirsute, wi , remote narrower on | rt of leaf; petioles slender, glabrous or pubescent, (1.5) sibel x . pe : 0 lanceolate-linear, or stem, gradually reduced late, (1)2-6.4(8.7) om. teh : wide; inflorescence a aensely red raceme, slightly elongated in fruit; sepals lanceolate to linear-Janceolate Ot BIOSYSTEMATICS OF THELYPODIUM 95 ovate, somewhat scarious at margin, often acuminate, erect, slightly saccate at base, white, often with a prominant midvein, slightly keeled, median (outer) sepals slightly inserted above the lateral ones, (3) 3.5-5(5.5) mm. long, 1-1.5(2) mm. wide; petals linear to narrowly oblanceolate, often crisped, white, 8-12.5( 16) mm. long, 0.3-0.5(1) mm. wide, claws erect, narrow, thickish, 2.5-4 mm. long; filaments white, slender, erect, slightly dilated at base, nearly equal in length or slightly tetradynamous, (2)2.5-6.5(10) mm. long; anthers sagittate at base, often strongly apiculate, linear to narrowly oblong, exserted, somewhat circinately coiled when fully dehisced, upwards, 1-2(2.5) mm. long; siliques linear, terete, torulose, straight to curved, (2) n long, (0.75)1(1.25) mm. wide; cotyledons obliquely incumbent or sometimes in- Flowering mid April to late August, but mainly late May through July, at elevations from 2,000 to 7,500 feet in northern California and southern Oregon (Map 2). REPRESENTATIVE SPECIMENS. California. Napa Co.: just W. of Knoxville on Lower Lake-Knoxville Road, Rattenbury s.n., 1949 (uc); Shasta Co.: Pitt River, Smith 305 (cas,us); Siskiyou Co.: 8 miles W. of jct. California roads 139 and 161, near Lower Klamath Lake, I. & M. Al-Shehbaz 68139 (cH); 3.8 miles S.E. of Gazelle on U.S. Hwy. 99, near railroad tracks, I. & M. Al-Shehbaz 6949 (cH); N. side of Mt. Shasta, Brown 469 (Mo,Nny,uc,us); Montague, Heller 8011 (cAs,Ds,F,GH,MO,NY,PH,UC ); N. end of Shasta Valley near Klamath River, Heller 14441 (CAs,cOLO,DS,F,GH,LA,MICH,MO, NY,ORE,PH,RM,RSA,UC,Ws,wtu); Yreka, Parker s.n., 1948 (DsS,GH,RM,RSA,UC,UTC,WS, wtu); Weed, Smith 285 (cas); 3 miles S. of Grenada, Wheeler 3637 (GH,MO,NA,ND, NY,POM). Oregon. Klamath Co.: Harriman Lodge Station, Abrams 9724 (Ds,Mo, POM,RM); near Dairy, Applegate 42 (ps,cH,us); near mouth of Williamson River, Coville 1241 (us); Modoc Point, Evans 340 (orr,pH); 1 miles S.E. of Keno, Peck 9401 (GH,MO,NY,ws ). In his original description of Thelypodiopsis brachypoda, Schulz (1933) gave the following description for the new species: Pedicelli 0.5 mm. longi. Flores parvi.” Later (1936, p. 583), he distinguished this species from Thelypodiopsis brachycarpa by giving pedicels 0.5 instead of 1.5 mm. long. The four isotypes of the former “species” that I have studied are indistinguishable from plants of Thelypodium brachycarpum in every morphological aspect. Moreover, the pedicels in these four isotypes are 1-2 mm. long. The placement of this species and its close relatives in Thelypodiopsis by Schulz (1933, 1936) has been discussed previously under generic relationships. = The single collection from Napa County cited here is somewhat disjunct from the main distribution center of the species. Whether this disjunction is the result of a recent introduction by man or possibly an old relict population is not known. There is relatively little variation in most of the morphological features in Thelypodium brachycarpum. Fruiting pedicels seem to vary 96 IHSAN A. AL-SHEHBAZ only a little in length. The most conspicuous variation is in the margin and the degree of pubescence of basal leaves. This is similar to the variation found in T. crispum. Thelypodium brachycarpum grows primarily in meadows and open flats with varying degrees of alkalinity. It can tolerate highly alkaline soils where it was found either growing by itself or at least more abundantly than any other plant species in the most alkaline parts of a given meadow. In this respect, it resembles T. crispum, which occupies somewhat similar habitats in eastern California and western Nevada. 5. Thelypodium eucosmum Robinson Thelypodium eucosmum Robins. in Gray, Synop. Fl. N. Am. 1:175. 1895. Holotype: 5 (cH) or sometimes attenuate at epa and somewhat glaucous, (2.8)3.5-8.8(11.1) cm. long; (0.7) 1-2.5(3.5) mm. wide; petioles slender, usually purple, ciliate near the base, wi omes simple an straight, (0.9)1.4—3.1(4.5) cm. long; cauline leaves lanceolate to oblong-lanceolate arious at margin, 5-7(8) mm. long, (0.75)1-1.5(1.75) mm. wide; a purple, mostly spatulate to oblanceolate, obtuse, entire to somewhat repand, attenuate at base, (6.5)7.5-10(11.5) mm. long, 1-1.75(2) mm. wide; claws slender, nearly rect, 3.5-5(5.5) mm, long; filaments slender, purplish, erect below and some- Flowering May to July in central and eastern Oregon. T: helypodium eucosmum was said to be present in Idaho by Rydberg (1917, 1923), Peck (1941, 1961), and Davis (1952). However, I have seen no specimens of this species from Idaho, and it seems unlikely that it grows outside of Oregon. TIVE SPECIMENS. Oregon. Baker Co.: Blue Mountains, Griffiths & Hunter 142 (xyx,uc); Blue Mountains, Howell 345 (cH); Grant Co. Canyon City, Howell s.n., 1885 (¥,NA,NY,ORE,PH,UC,US); moist ground along John Day River, 15 miles above Dayville, Peck 16035 (ps,wru); 17 miles E. of Prairie City, Fox Valley, Co.: BIOSYSTEMATICS OF THELYPODIUM 97 — — (ny); Wheeler Co.: Sutton Mountain, 10 miles N. of Mitchell in T. 10S., 9, Cronquist 7249 (cAs,Ds,GH,MO,NY,OSC,RM,RSA,UC,UTC,WS,WTU ); same ee oe, Al-Shehbaz & Cronquist 6972 (cH Thelypodium eucosmum is the prettiest species in the genus because of its bright purple flowers. However, it appears to be very rare, as judged from the very few collections that exist in the major herbaria of this country. Two different observations have been made with regard to the habitat where T. eucosmum grows. The present writer and Dr. Cronquist collected it from shady places under the junipers on one of the dry slopes of Sutton Mountain, while Peck collected it from moist ground along John Day River. Thelypodium eucosmum does not appear to be very closely related to any other species. It stands as morphologically intermediate between the previous four species and most of the remaining auriculate-leaved species. Payson (1923) considered T. eucosmum the most primitive species of the genus, but the evidence at hand indicates that this is not the case. The relationships of T. eucosmum to the rest of the species were discussed above under the section on evolution within the genus. The combination of the following characters easily distinguishes Thely- podium eucosmum from the other species of the genus: long gynophores, dark purple flowers arranged in a raceme, exserted anthers, and oblong- linear floral buds 6. Thelypodium rollinsii Al-Shehbaz, sp. nov. a biennis; caulibus erectis, tenuibus farctis, simplicibus ad basin et supern spathulatis vel oblanceolatis, dentatis vel repandis, ngis 1.3(1.8) cm. latis; foliis caulinis capree apap linearibus vel lineari- Seiceobatia: vel acuminatis, ad caulem appressis, (0.8) 1.3-4.5(6.1) cm. long (1517 “71(8) x mm. latis; auriculis 0.5-2.5(5) mm. longis, 0.5-1.5(2.5) mm. latis; sepalis linearibus vel re scariosis, 4-6(7) mm. longis, 0.75-1(1.5) mm. latis; petalis erectis, spathulatis vel obovatis, non crispis, lavandulis vel purpureis, (6)7-9. 5(12) mm. longis, (1.2)1.5— longis; inibus erectic, filamentis gestis (1)2-10.5( a) longis; pedicellis fructiferis oe satis, divaricatis vel horizontalibus, non complanatis, (6)7—-14(18) mm. longis; sliquis 7 pa teretibus, tenuibus, torulosis, valde incurvatis, (1.8)2.8-5 75(1) mm. latis; stipite 0.5-2.5(6) mm. longo; stylis (0.5)0.75-1.5(2) mm ee sti igmati tibus parvis, integris; seminibus uniseriatis, oblongis, exalatis; cotyledoni- bus oblique incumbentibus ‘ set in the Gray Herbarium, collected on alkalaine soil in the vicinity o Sevier ope about 12 miles north of Scipio on U. S. Hwy. 91, Juab County, Utah, July 29, 1969, Ihsan A. é Mona M. Al-Shehbaz 6913; isotypes to be distributed. lant herhackous biennial, glabrous and somewhat glaucous throughout; stems erect, often slender, solid, single from the base, branc hed at the inflorescence, (4) 6-16( 20) dm. high; basal leaves thickish, ish, spatulate or less often oblanceolate to obovate, obtuse to sometimes acute at apex, dentate or repand, very rarely entire, mostly attenuate at 98 IHSAN A. AL-SHEHBAZ base, 1.3-4(7) cm. long, (0.4)0.6-1.3(1.8) cm. wide; petioles slightly expanded at base, glabrous, 0.5-1.3(1.5) c¢ linear or sometimes oblong, obtuse, scarious at margin, purple to lavender throughout, or lavender at the margin and green in the center, equal at base, sometimes slightly (7 terete, slightly tetradynamous, often lavender, erect, filaments of paired stamens (5)5.5-7.5(8.5) mm. long, filaments of single stamens (4)5-6.5(7.5) mm. long; 10.5(18) cm. long; fruiting pedicels straight, slender, striate, mostly divaricate to horizontal, rarely divaricately ascending, mostly not flattened at the base, (6)7—14( 18) : 0.25-0.5 . wi narrower and subtending base of paired stamens; infructescence congested, (1)2— rarely incum Flowering late June through mid August at elevations from 5,000 to 6,500 feet in central Utah. REPRESENTATIVE SPECIMENS, Utah. Carbon Co.: 15 miles E. of Wellington, Malburtt s.n., 1959 (ut,uTc); Juab Co.: 14.5 miles $.W. of Levan on US. Hwv. 91. L Al-Shehbaz 6979 (cH); 9 miles S.W. of jet. U.S. Hwy. 91 and Utah State Hwy. 28, I. & M. Al-Shehbaz 6981 (cu); ca. 10 miles N. U.S. Hwy. 91, Higgins 1794 (cu); Millard Co.: 32 miles N. of Filmore, Holmgren & Tillett 9654 ( CAS,GH,MO,NY,UC,US,UTC,ws,wTU ); Piute Co.: ca. % mile S. of Piute-Sevier County line on USS. Hwy. 89, I. & M. Al-Shehbaz 6910 (cH); Marysvale, Jones 5965B (NY, POM,us ); along Sevier River above Marysvale, Rydberg & Carlton 6926 (GH,NY,RM,US ); Sanpete Co.: i ing on US. wy. 89, Nine-Mile Reservoir, I. M. Al-Shehbaz 6907 (cx); same locality I. & M. Al-Shehbaz 6982 (cH); Manti, Jones 5521 (Nx); Sevier Co.: 0.1-0.2 mile S. of Redmond, I. ¢ M. Al-Shehbaz 6908 (cH); same locality, I. & M. Al-Shehbaz 6983 (cH); 1.6 miles S.W. of Salina on U.S. Hwy. 89 at Sevier River, I. & M. Al-Shehbaz 6909 (cH). Thelypodium rollinsii is an abundant species on the alkaline flats along the Sevier River and adjacent areas in central Utah. However, it is surpris- ing to find very few collections of it in the major herbaria of this country. More surprising is the fact that some specimens of this species were available to taxonomists such as Rydberg, Jones, and Payson, who, for one reason or another, failed to describe it. In fact, Payson (1923) cited the only collection that he saw (Rydberg & Carlton 6926, nm) under T. lilacinum, while others have misidentiified it as T. saggittatum. There is some variability among the specimens cited above. This applies BIOSYSTEMATICS OF THELYPODIUM 99 especially to certain features such as the size of the auricles, length of the fruiting pedicels and infructescences, and the color of the flowers The closest relative to Thelypodium rollinsii is T. integrifolium, which it resembles in habit and several features of the flowers. The new species is morphologically intermediate between T. integrifolium and T. sagit- tatum in a number of features, but it is quite different from either one in the nature of its fruiting pedicels, cauline and basal leaves, and siliques. It is also very different in its mustard oils as we have shown above. Thelypodium rollinsii can be easily distingushed from its nearest rela- tive, T. integrifolium, by the following characteristics: the basal leaves are very small, 1.3-4(7) cm. long, (0.4)0.6-1.3(1.8) cm. wide, often dentate or repand (Plates 3 and 19). The cauline leaves are almost always auriculate and partly to completely appressed to the stem (Plate 17). The calyx is somewhat 4-sided, and median glandular tissue is present (Plate 18). The fruiting pedicels are slender, (6)7-14(18) mm. long, and often not flattened at base. Finally, the siliques are slender, often strongly curved, and 0.75(1) mm. wide (Plate 20). In T. integrifolium, as represented by its five subspecies, the basal leaves are (3.7)5-31(54) cm. long, (1.2)1.6-7.8(14) cm. wide, often entire (Plates 3 and 19). The cauline leaves are not appressed to the stem (Plate 17), but they are usually ascending, sessile and without auricles (reduced auricles were found in 8 plants out of more than one thousand studied). The median glandular tissue is absent (Plate 18). Fruiting pedicels are slender to stout, mostly flattened at the base, (2)3-9(13) mm. long. The siliques are slender to stout, straight or curved, 1-1.3(2) mm. wide. Thelypodium rollinsii is distinguished from T. sagittatum by its glabrous instead of ciliate basal leaf petioles, smaller and dentate instead of the larger and entire basal leaves, its appressed instead of ascending cauline leaves, its branching habit, exserted stamens, longer stems, and inflorescences that are hardly elongated in fruit. 7. Thelypodium integrifolium (Nuttall) Endlicher Herbaceous biennial; stems often simple or sometimes branched at base, yay or somewhat paniculately branched at the inflorescence, slender to stout, purple whitish green, glabrous and somewhat ya throughout, (2)4.5-17(28) dm. ck basal leaves thickish, glabrous aa somewhat glaucous, mostly oblong or elliptic to oblan i tly entire occasionally denticulate or rep to less often acute at apex, eate to sometim at base, earlier leaves usually broader and with longer —— than later ones, (3.7 1( em. long, (1.2)1.6-7. 8(14) cm. wide; petio TO at base, (0.5)1.2-10(15) cm. long; ca aves ichish, glabrous, slightly to moderately glaucous, ob to more often acute or acuminate, en or ges rae acini seldom rasa, uriculate, lower cauline leaves lanceolate, upper leav eo to mos: anceo) an e linear, sessile, Feces Rompe cm. Tong oe) mm. ; corymbose or very short racemose t de to linear, erect, obtuse. a oe. bove 100 IHSAN A. AL-SHEHBAZ white to lavender below, or lavender to dark purple throughout, equal at base, the lateral sepals sometimes slightly saccate, glabrous, (3)3.5-5.5(7.5) mm. long, (0.5)0.75-1(1.5) mm. wide; petals oblanceolate to spatulate, not crisped, obtuse, white to lavender or dark purple, erect, (4.5)6-9(13) mm. long, (0.5)0.75-1.5( 2.3) mm. above base of lateral stamens, open or emarginate, absent or reduced below, glands flat, bilobed or with two strongly developed tooth-like processes that rarely exceed 1 mm. in length; infructescence racemose or subumbellate, (0.6)1.5-14.5(28) cm. long, lax to congested; fruiting pedicels slender to stout, straight or rarely curved, we o di re i 7. to divaricate, rarely ascending or reflex mewhat stri at least near the tip, glabrous and sometimes glaucous, strongly to moderately flattened at bas so’ es not flattened, (2)3-9(13) mm. long; siliques terete to somewhat flattened, torulose or rarely submoniliform, straight re often strongly incurved, divaricate to divaricately ascending or horizontal, rarely reflexed (0.7) 4(8 long, e; styles slender, taperi ard tip, rarely subclavate, 0.5- 1.3(3.2) mm. long; ntire; gynophores slender, (0.3)0 ) mm. long; stigmas e ; seeds (1)1.2-2(2.3) mm. long, 0.75-1(1.25) mm. wide; cotyledons obliquely incum- t a, q = & | a & 5 lagi 17. Cauline leaves of Thelypodium integrifolium (Fic. 1) and T. rollinsii (Frc. 2). Natural BIOSYSTEMATICS OF THELYPODIUM 101 PuaTe 18. Nectar glands of Thelypodium rollinsii and T. integrifolium. Figs. 1, 2. T. rollinsii; Fics. 3-6. T. integrifolium. Figures 1, 3, 5 lateral view; 2, 4, 6 median view. Scale 4% mm. KEY TO THE SUBSPECIES A. Flowers white; mature fruiting pedicels — whitish, 6-13 mm. long. B. Siliques (2.2)3.5-6.3(8.0) cm. long; fruiting pete (6)7-11(13) mm. long, horizontal to divaricate or reflexed; pyaainaes (1)1.5-4(5.5) mm. long; petals (7. (75)8-11(18) mm. long; plants of the Grand Canyon of the Colorado River in Te. subsp. longic a g raceme. cm. long; gynophores slender (0. 5)0.75-3 western Colorado, northwestern New Mexico, and aie astern “Azo oe eg Un ee One Ee ean eG ar a subsp. . grailipes. nce short raceme, often strongly congested, fete kh (0.6) 1.2— 79) cm. pay gynophores stout, 0.5-1(3) mm . long; plants of northwestern Utah, Nevada, eastern California, and southern Oregon oe : C. Mature fruiting irepee a0 slender - cprationy flattened at the base, or mm. shorter and not flattened at base, infructescence ane pe aa widely distributed .............. Ta. subsp. integrifolium. 102 IHSAN A. AL-SHEHBAZ 7a. Thelypodium integrifolium (Nuttall) Endlicher subsp. integrifolium Pachypodium integrifolium Nutt, in Torr. & Gray, Fl. N. Am. 1:96. 1838. Holotype: “Elevated plains o ’ the peel ” elec towards the Oregon, as far as Wallawa Nah,” Nuttall s.n. (not seen); iso Thelypodium Sstepritolin ho "Endl Walp. Repert. 1:172. 1842. Pleurophragma integrifolium (Nutt.) Rydb., Bull. Torr. Bot. Club 34: 433. 1907. elypodium lilacinum Greene, Pl. Baker. 3:9. 1901. Holotype: Doyle’s, alt. 8,100 ei July 29, 1901, C. F. Baker 635 (ND~c); isotypes: GH,MO,NY,POM,US. rophragma i (Greene) Rydb., Brittonia 1:89. 1931. te el lilacinum Greene var. subumbellatum Pays., Ann. Mo. Bot. Gard. 9:281. 1923. Holotype: near Mammoth Hot Springs, Yellowstone National Park, Wyoming, alt. 6,200 feet, August 1893, F. H. Burglehaus s.n. (Mo); isotypes: MICH, US,TU. Flowers lavender to purple or rarely white; petals (4.5)5.5-8(10. 5) mm. long; i mce racemose to su umbella te, _ congested to somewhat lax , central rachis straight to slightly Pr upwards, mostly divaricate to divaricately ascending, rarely ee slightly fla ienod or oe at base, (4)5-10(13) mm. long; siliques rved, divaricately ascending to ascending, (1.0)1.5-3.0(4.1) cm. long; gynophores 00 0.5-1.2(2.5) mm. long; glandular tissue flat and low, — or with two strongly developed tooth-like processes that are free to variously unit Flowering from early June to late September, but mainly late June through August, at elevations from 2,000 to 8,100 feet. Distributed in central Washington, Oregon, Idaho, southwestern Montana, Wyoming, central and northern Utah, northern and central Colorado, western Nebraska, western South Dakota, and central North Dakota (Map 3). ENTATIVE SPECIMENS. Colorado. Eagle Co.: about 20 miles N.E. of Glenwood , Rollins 1887 (ps,cH,MO,ND,NY,US. sel: Fremont Co.: Wet Mt. li ar NY ringfield, Christ 8944 oe Butte Co.: 112 miles N.W. ~ Howe, I. & M. peor 6920 (cu); Canyon Co.: 4.4 miles N.W. of Notus on Hi I. & M. Al-Shehbaz 6973 (cu); Caribou Co.: 1 mile E. of Soda Springs, pat 16167. (xy); Canin Co.: 23 miles E. of ee J. & C. Christ 18541 (NY); Custer Co.: Clayton, Macbride & Payson 3360 (CAs,Ds,cH,MO,NY,POM,RM,UC,US ) ; Elmore Co.: near King Hill, Nelson ¢¢ Macbride s.n., 1911 (nm); Lemhi Co.: a Henderson 3881 (us); Power Co.: American Falls, N elson & Macbride 1380 (vs O,NY,POM,RM,UC,US,UTC,WS owru): Twin wie ies Twin Falls & Shoshone Falls, — & — 1349 _ ,MO,RM ). Montana. op Lodge ae Jones s.n., 190 her jows POM,us); Gallatin ree Forks, posal er 23 (MO,NMC,POM,RM,UC); Lewis lark Co.: Helena, “Kebey $.n., — (ps,F,POM ); and Co.: Musselshell River Bottom, 4 miles of S Shawm 0c. ka, i o— 711 (cGu,uc,us); Kidder Co.: near Lake Ta appen, Stevens s.n., e Berthold Reservation, Heidenreich B36 (wa); Ward Co.: Ryder, oe s.n., 1923 (Mo). am Co.: near Prineville, Leiburg 817 (DS,F,GH, JEPS.MO,NY,ORE,POM,UC,US ); Harney : Stein’s Mountain, Troin s.n., 1935 :} Creek, Summer rg Constance 9542 (pH); Malheur Co.: ked Creek, Train s.n. (us). South Dakota. Custer Co.: Pringle, Over 15957 (RM, BIOSYSTEMATICS OF THELYPODIUM 103 us); Pennington Co.: Rapid Creek, Rapid City, McIntosh 209 (nm). Utah. Cache Co.: 1 mile W. of Logan, Smith 2 (CAS,DS,RM,UTC); Piute Co.: ca. 7 miles W. of Co.: near Ellensburg, Brandegee 636 (uc); Okanogan Co.: near Tonasket, Thompson 7106 (Ds,cH,Mo,osc,uC us,wtu); Yakima Co.: Squaw Creek, Cotton 874 (cH, Mo, us,ws ). Wyoming. Albany Co.: S. of Laramie, Beetle 4894 Uc,wis); Fremont Co.: Red Canyon of the Wind River, 13 miles S. of Dubois, C. & e Ae) Q m 9 yD ie 9 m ry io) Ss = ry a : Dayto tw. , Nelson 4140 (nm); Uinta Co.: Blacks Fork River, 4 miles N.W. of Lyman, Rollins Mufioz 2922 (ps,cH,RM,uc,uTC); Yellowstone National Park: Mammoth Hot Springs, I. & M. Al-Shehbaz 68161 (cx). Thelypodium integrifolium subsp. integrifolium as defined here shows a rather wide range of variation in a number of characters. Despite the fact that some of the variation, as is shown below, falls in certain geographic patterns, the present writer believes that the taxonomy of the group is better served by maintaining subsp. integrifolium as a poly- morphic taxon rather than splitting it into five or six varieties. Nectar glands are more variable here than in the other four subspecies of Thelypodium integrifolium (Plate 7). They tend to be low and flat in most plants of central and northern Colorado, Wyoming, and central and northern Utah. On the other hand, plants of North Dakota and South Dakota have somewhat bilobed glands, while those of Oregon, Washing- ton, and some parts of Idaho have well-developed glands that are tooth- like and up to 1 mm. high. However, the variation is of a continuous nature and does not seem to characterize all the populations of any given part of the geographic range. Moreover, plants of the same popula- tion may vary considerably in the size, shape, and height of their nectar ands. The fruiting pedicels are equally variable. Although most of the populations of subsp. integrifolium are characterized by having slender pedicels that are somewhat flattened at the base, populations of centr and northern Colorado, southwestern Wyoming, and northern Utah tend to have stouter pedicels that are hardly flattened at the base. The length of the pedicel is also subject to considerable variation. The shortest pedicels are found in the populations of southwestem Wyoming and northern Utah, while the longest pedicels are characteristic of the plants from Washington, North Dakota, and South Dakota. Flower color ranges from lavender to various shades of purple, or 104 IHSAN A. AL-SHEHBAZ e Thelypodium rollinsii = 9 T. integrifolium SOF © E 40} 3) U = © © ee () ee (3) = 6 © 3O0F ad © . (0) Oo Ww) © © > % D 6 © e @° eur sete Hae ® ane ° 00 7 ©2 4 8 10} F 4 B30. 40. 44 Basal Leaves Width (cm.) bao roe showing relationship of length to width of basal leaves of Thelypodium sometimes white. In central Colorado the populations of this subspecies tend to have dark purple flowers, while those of North Dakota and South Dakota are made up of white-flowered forms. Flower color may be variable in the same population. The present writer has collected plants BIOSYSTEMATICS OF THELYPODIUM 105 Finally, the infructescences tend to be shorter in the populations of western Wyoming, northern Utah, Montana, and adjacent Idaho than in other parts of the subspecies range. However, forms with short infructes- cences are by no means restricted to the populations of the states mentioned. In fact, they are distributed throughout the subspecies range ut with much less frequency in areas outside of that indicated. Plants of this nature were the basis of Payson’s Thelypodium lilacinum var. subumbellatum. The type collection of this variety has been compared carefully with an isotype of T. integrifolium at the Gray Herbarium. There are no significant differences between the two and nothing to justify the recognition of the former taxon. The three features claimed by Greene (1901) to distinguish Thely- podium lilacinum from T. integrifolium (corymbose inflorescence that lengthens in fruit, rich lilac-purple flowers, and branching from the base) are clearly variable within the type collection of T. lilacinum. Moreover, the characters used by Payson (1923) to distinguish the two have no claim to consistency. Thelypodium lilacinum may deserve a rank no higher than variety at best, but if this course is followed, then subsp. integrifolium should be further split into a number of varieties separated from each other by only a few inconsistent characters. This would certainly confuse rather than clarify the taxonomy of the complex. Subspecies integrifolium is the most widely distributed of the five sub- species of T. integrifolium. It has been reported from Owyhee County in Idaho (Baker 1963), and Dawes Co., Deuel Co., Box Butte Co., Sheridan Co., (Peterson 1912, 1923) and Garden Co. (Winter 1936) in Nebraska. I have not seen specimens from any of these counties, but there is no reason to doubt the occurrence of it there. 7b. Thelypodium integrifolium (Nuttall) Endlicher subsp. complanatum Al-Shehbaz subsp. nov. talis lilacinis vel albis, (5)6-8.5(9.5) mm. longis; infructescentibus racemos' fae (0.6) 1.2-7(9) cm. longis; panna oo — vel fama Striatis, rectis, crassis vel tenuibus, ad basin mp. mm. longis; siliquis stipitatis, divaricatis vel reales aa a. OR “(08)1. 4-2.7(3.4) cm. longis; stipite crasso, 0.5-1(3) mm. longo. Holotype in the Gray Herbarium, collected on low alkaline sane 10 miles east of Tronsid, US. . 26, about % mile south of the way, Malheur County, ao Anges 16, Ot, Toe & 56; i Hae wr ie deena: os horizontal, or sometimes divaricate, strongly flattened at base, 3-5 sy mm. long; siliques straight or curved, divaricate to divaricately ascending or a (0 _ ae w, ies 2.7(3, 4) cm. long; aes stout, 0.5-1(3) ap Jong; nectar glands more often bilobed, with the lo bes partly united to Flowering mainly from late June through August at elevations from 2,900 to 8,000 feet in eastern California, Nevada, southeastern Oregon, and northwestern Utah (Map 3). IHSAN A. AL-SHEHBAZ PLaTE 20. Infructescences (main branches) of — integrifolium subsp. complanatum (Fic. 1) and T. rollinsii (Fic. ; Scale: one centim ENTATIVE SPECIMENS. California. Inyo Co.: Black Canyon, White Mountains, Duran 532 cog tar RG spite ,UC,US,UTC,wis ); Lassen Co.: Rus reek, ] miles E E. - Hwy. 395 and 1.3 miles W. of Nevada state line, ste Pel & Robbins ey ( bee, Mono Co.: Sherwin Grade summit, Kerr s.n BIOSYSTEMATICS OF THELYPODIUM 107 (cas,cH,us). Nevada. Elko Co.: 2.7 miles S. of Jackpot on U.S. Hwy. 93, I. & M. Al-Shehbaz 6926 (cH); 13.8 miles N.E. of Elko on Interstate Hwy. 80, I. & M. Al- Shehbaz 6934 (cH); vicinity of Nevada Industrial School, U.S. Hwy. 40, Maguire Holmgren 22169 (Ds,GH,NA,NY,UC,US,UTC,ws,wTu); Hum Co.: Virgin Creek gorge, Sheldon Range, Murie 2752 (Na); Lander Co.: 3-18 miles N. of Austin along Hwy. 8A, Goodner ¢ Henning 874 (na,Ny); Lyon Co.: 6% miles S.E. of Rockland, Hendrix 655 (uc); Mineral Co.: 1 mile S. of Fletcher, Alexander & ‘Kellogg 4487 (Ds,GH,UC,Us,UTC); 1.3 miles from mouth and u Cory Creek, Wassuk Range, Archer 6874 ( NA,NY,RSA,TEX,UC,UTC,wru); Nye Co.: 23 miles S. of Dieringer, Goodner & Or Harney Co.: 6.3 miles E. of Burns on Oregon Hwy. 78, then 0.9 miles S. of the highway on a dirt road, I. & M. Al-Shehbaz 68141 (cH). Utah. Box Elder Co.: north end of Salt Lake, Anderson 855 (utc); 18 miles W. of Park Valley, Maguire & Holmgren 22241 (ps,cH,Ny,UTC,wTv). Thelypodium integrifolium subsp. complanatum corresponds fairly well with T. rhomboideum sensu Payson (1923). However, the type specimen of T. rhomboideum Greene, which Payson did not see (Payson, 1923, p. 276), differs in many significant morphological features from the speci- mens cited by him under this “species.” In fact, the type specimen of T. rhomboideum is hardly different from that of T. affine Greene. The type of T. rhomboideum was collected from the West Humboldt Moun- tains by Greene. These mountains are located, according to McVaugh and F Oosberg (1941), in Pershing County (Nevada), and are now known as the Humboldt Range. The mature fruiting pedicels of the type of T. rhomboideum are 7-8 mm. long. This clearly falls within the range of subsp. affine and not subsp. complanatum. The fruiting pedicel length of plants of T. integrifolium from central Nevada, placed here in subsp. complanatum, hardly exceed 5 mm. in length. The nectar glands on the type of T. rhomboideum are very well-developed and approximately 1 mm. in length. This is characteristic of most plants of subsp. affine and not subsp. complanatum. The only clear difference between the type collec- tion of T. affine and T. rhomboideum is the shape of the basal leaves, which is spatulate-oblanceolate in the former and rhombic-lanceolate (broadly elliptic-lanceolate) in the latter. The type collection of the latter “species” is made up of a rosette and a flowering specimen each mounted on a separate sheet. Our experience with basal leaf shape in T. integrifolium and other species leads us to the conclusion that this is One of the most variable and least satisfactory taxonomic characters to use in distinguishing taxa at all levels in Thelypodium. All the evidence at hand supports the following conclusions: (1) the type of Thelypodium rhomboideum is nothing but a minor variant of T. affine; (2) subspecies complanatum is the sole representative of T. integrifolium in central and northern Nevada; and (3) it is probable that the type specimen of T. rhomboideum was collected from an area very near to the known range of subsp. affine. - 108 IHSAN A. AL-SHEHBAZ 7c. Thelypodium integrifolium (Nuttall) Endlicher subsp. gracilipes (Robinson ) Al-Shehbaz, comb. nov. Thelypodium coy, eet (Nutt.) Endl. var. gracilipes Robins. in Gray, Syn. N. Am, 1:176. 1895. Holotype: southwestern Colorado, T. S. Brandegee 1233 pia Meccan gracilipes (Robins. ) 7. 1906. Pleur eum Greene var. gracilipes (Robins. ) ek. Ann. Mo. Bot. Gard. 9:277. 1993. Pleurophragma platypodum Rydb., Bull. Torr. Bot. Club 34:434. 1907. Holotype: pis Grand County, Utah, August 30, ag Marcus E. Jones s.n. (NY); isotypes: ‘AS,DS,GH,MICH,MO,NY,POM,RM,TEX, UC, US,W more curved, horizontal to aa, crepe ee ate) cm. long; ich slender to mm. long; nectar glands ion and flat, or more often fused. the lobes — to completely bilobed, with th North ° Dakota es ana South Ji ies Dakota e e e ig Wyoming ° 2 . Nebraska : eee ane re Pei elase™ Nevada .» ssh ¥ — oe Z : ° ae 4% J . ke e A iJ a Colorado ) < . A z oo ~~ * . . en * 4 Thelypodium rollinsil aa wr Tintegrifolium no a * . ssp. integrifoliu bal ve. ae © ssp. complana a Ep Pe, @ . ssp. longicarpN * Mea » ssp gracilipes Arizona XICO pi ssp. affi 16) 100 200Miles ~ aes Seaman nr | Mar 3. Distribution of Thelypodium rollinsii and T. integrifolium BIOSYSTEMATICS OF THELYPODIUM 109 Flowering July through August at elevations between 3,500 to 8,300 feet in northwestern New Mexico, northeastern Arizona, western Colorado, and eastern Utah. (Map 3). REPRESENTATIVE SPECIMENS. Arizona. Apache Co.: Navajo Indian Reservation, N. end of Carrizo Mts., Standley 7412 (us); Coconino Co.: Painted Desert, near Tuba, Clute 82 (cH,Mo,Ny,RM,us). Colorado. Grand Co.: Middle Park, near Hot Sulphur Springs, Ramaley & Robbins 3627 (co.o,nM); Jackson Co.: Lake John, Johnson 907 (coLo); Moffat Co.: S. bank of Yampa River, 1 mile W. of Maybell, Hermann 5352 (cH,PH); Montrose Co.: Bedrock, Walker 369 (GH,RM,Us,ws); Montezuma Co.: 5 miles N.W. of Cortez on U.S. Hwy. 160, I. & M. Al-Shehbaz 6994 (cu); Rio Blanco Co.: Rio Blanco, Robbins 7060 (coLo); Routt Co.: Hayden, Goodding 1789 (coLo,cu, MO,NY,RM,UC); San Miguel Co.: along Dolores River at the bridge on Colorado Hwy. 80, Ownbey 1494 (cOLO,GH,MO,NY,RM,UC,WSs,wTu). New Mexico. San Juan Co.: Farmington, Wooton 2783 (NMc,us); vicinity of Cedar Hill, Standley 7992 (us). Utah. Garfield Co.: 15.1 miles E. of Escalante, Utah , oA, reek Recreation Area, N. & P. Holmgren 4721 (cH); Grand Co.: Hanging Garden, N. of trail to Delicate Arch, Arches National Monument, Welsh, Harrison & Moore 2337 (Ny); San Juan Co.: Natural Bridges National Monument, I. & M. Al-Shehbaz 6993 (cx); Lime Creek, 5 miles N. of Mexican Hat, Cutler 2781 (ps,cH,NA,NY); Uintah Co.: S. of Split Mt. Gorge, Dinosaur National Monument, Bradley 5338 (coLo). Thelypodium integrifolium subsp. gracilipes is probably the best defined of all the five subspecies. Occasionally it has been confused with the plants cited here under subsp. complanatum. Payson (1923), for instance, treated it as a variety of T. rhomboideum, but confused the two by citing some of the collections that unquestionably belong to subsp. gracilipes under T. rhomboideum proper. This was probably the result of his use of the gynophore length as the basis of distinguishing the two taxa. Rydberg (1923), on the other hand, recognized two species, namely Pleurophragma gracilipes and P. platypodum. He claimed that the two are distinguishable on the basis of having pedicels that are not flattened and about 1 cm. long in the former “species,” while they are flattened at the base and about 3 mm. long in the latter. These alleged differences have no claim to reality. The type collections of the two taxa are very similar, and the pedicels in both are flattened and less than 5 mm. long. Moreover, Rydberg’s own annotations show that he had designated the isotypes of his P. platypodum differently. For example, two isotypes at Ny were an- notated as Thelypodium platypodum, one at NY and one at US as T. integri- folium, and the cu isotype as T. integrifolium var. gracilipes. Subspecies gracilipes is rather variable as far as flower color is con- cerned. The populations of northern Colorado tend to have mostly lavender to dark purple flowers, while those of eastern Utah and south- western Colorado often have lavender as well as within the same population. The populations of Arizona and New Mexico appear to be ae white flowered. The infructescence length also is somewhat variable. Populations of northern Colorado and adjacent northeastern Utah tend to have shorter infructescences than those else- 110 THSAN A. AL-SHEHBAZ where in the subspecies range, but the variation is of a continuous nature. Subspecies gracilipes is most closely related to subsp. complanatum from which it is distinguished by its longer infructescence and longer and more slender gynophores. The two subspecies can be distinguished from the other three of T. integrifolium by their shorter pedicels that are often strongly flattened at the base. 7d. Thelypodium integrifolium (Nuttall) Endlicher subsp. affine (Greene) Al-Shehbaz, comb. nov. helypodium affine Greene, Pittonia 4: se 1901. Harsrve’ Tehachapi, Kern Frain California, June 22, 1889, E. L. Greene s.n. (ND-G Thelypodium rhomboideum Greene, Pitonia, 4: ait "1901. Holotype: West Humboldt Mountains, July 1894, E. L. Greene s Pleurophragma rhomboideum (Greene) O.£.. Schulz, Bot Bot. Jahrb. 66:98. 1933. Sepals whitish-green; petals white, (5)6-9(10) mm. long; infructescence racemo often congested, central rachis (3.7)4.5-12(16.5) cm. long; fruiting pedicels straight ightly curved Vv arica em. long; gynophores stout, (0.5)1-3 mm. long; nectar glands large, up to 1 mm. high, well-developed above the base of single stamens, lobed, with the lobes partly to completely fused, or sometimes free, often strongly flattened and erect in the middle. Flowering late June to mid October, but mainly late July through September, at elevations from 2,000 to 5,800 feet in southeastern California, southern and southeastern Nevada, and southwestern Utah (Map 3). REPRESENTATIVE SPECIMENS, California. Kern Co.: Tehachapi, Eastwood s.n., 189 (cH); Los = eles Co.: Rosamond Lake, Eastwood ¢+ Howell 3979 (cas); see RES Jones s.n., POM,uC ); San Bernardino Co.: El Vaquero Rancho, ca. 29 miles of Victorville as California Hwy. 18, I. & M. Al-Shehbaz 68128 (cH); Rabbit nie nea cere bh oe Ball 14391 (cas,cOLO,NA,RMRSA); 1 mile N.W . cegcepsic 66 (DS,F,MO,POM,UC). Nevada. Clark Co.: ca. 8 m of 5S. Hwy. 95, I. & M. Al-Shehbaz 6988 (cu); Cactus jaca Indian Valley, eee een oe Lincoln Co.: 4 miles S. of Caliente, below road to road to oring roa Farm, Sec. 5, T. 18 S., R. 50 E., Reveal 1537 (cu). Utah. Co.: miles S.W. of Mollie’s Nipple and 40-45 miles N.E. of Kanab, Hester Ss (rsa); age ounig - Pumice, near Black Rock, Jones s.n., 1913 (rom); Wash- 0.2 miles E. of Rockville on Utah Hwy. 15, 1. => " Al-Shehbaz 6987 (cH); ce of St. George, Maguire & Richards 20460 (ps,uT Subspecies affine and longicarpum are the least variable of the five subspecies of T. integrifolium. The tallest plants in Thelypodium are found in these two subspecies, but short plants are found here as well. The only two characters that are variable to a significant extent are the pedicel and fruit length, but these two characters are equally or more variable in the other three subspecies. Subspecies affine was known only from southeastern California for a BIOSYSTEMATICS OF THELYPODIUM lll long time. It was first reported from Utah by Maguire (1942) as Thely- podium affine, and the range as presently given extends to cover southern and southeastern Nevada. The closest relative to subsp. affine is subsp. longicarpum from which it is distinguished by its shorter siliques and petals. The two are easily dis- tinguished from the remaining three subspecies of Thelypodium integri- folium on the basis of their white flowers, whitish- -green sepals, and fruiting pedicels that are stout, whitish-green, and longer than in the others. 7e. Thelypodium integrifolium (Nuttall) Endlicher subsp. longicarpum Al-Shehbaz, subsp. noy. rectis vel arcuatis vel incurvatis, horizontalibus vel sellcnla. Ga 2.)3.5-6.3( 8.0 longis; stipite crasso, bee 5-4(5.5) mm. longo. Holotype in Herbarium, collected ca. 0.5 mile below Indian Garden on Bright Angel Trail, a Canyon n National Pa rk, Coconino County, Arizona, Septem- ber 13, 1969, Ihsan ¢ Mona Al-Shehbaz 6991; isotypes to be distributed. creamy white to white below and greenish above; petals white, very rarely lavender, (7.5)8-11(13) mm. long; infructescence racemose, moderately to stich ed, central rachis (4.3)8-20 cm. long; fruiting pedicels straight, stout, moderately to strongly striate, whitish, often be agl flattened at base, mostly hori- zontal to reflexed, rarely divaricate, (6)7—11(13) mm. long; “aap often strongly torulose, straight to incurved, sometimes arcuate, horizontal to reflexed, rarely divaricate or or divaricately os (2. rae 5-6.3( 8.0) cm. long; gynophore stout, “d )1.5-4(5.5) . long; nectar glands up to 1 mm. long, each with two well- developed tooth-like eee that are partly to completely free Flowering early August through October at elevations from 1,900 to 4,200 feet in the Grand Canyon National Park and the adjacent areas in northern Arizona. Havasu Falls, Havasupai TATIVE SPECIMENS. Arizona. Coconino Co.: : ‘ ogy lig in ai nate 4408 (ps,MICH); clone Mooney 'F Falls, Clover 5263 (ps —— upai _— Clover 7016 (micu,us); Shuiamo Creek, Pilshy s.n., ring, Bright Angel Trail, Seler 4744 ee GH 8 — Agel Tra at Cohiratis River, Wolf 3180 (cas,Ds,GH,RSA ); aster Canyon and Upper Travertine Falls, Lake ice) Choose. 4284 te), Saddle Horse 13526 (ut : ov Subspecies Shceinien has been confused with other taxa of the Thelypodium integrifolium complex. It was cited as T. integrifolium by Clover and Jotter (1945) and Kearney and Peebles (1960), and as T. thomboideum by the first two authors and by Kearney and Peebles (1942). The latter authors failed to distinguish it from the very distinct subsp. gracilipes, as judged from their citation of collections belonging to the two subspecies. 112 IHSAN A, AL-SHEHBAZ The long siliques and pedicels of subsp. longicarpum serve to distinguish it easily from the other subspecies of Thelypodium integrifolium. One might argue that these features are different enough from the rest of T. integrifolium for it to be recognized as a distinct species. However, these features are connected with the rest of T. integrifolium through subsp. affine. Furthermore, the similarities in the flower and leaf morphol- ogy as well as the mustard oils (Table 6) in the whole complex are evi- dence against the recognition of subsp. longicarpum as a distinct species. Thelypodium integrifolium is clearly the most variable and most widely distributed of any species of the genus. It occupies an extremely wide range of habitats. It grows on clay, loamy or sandy soils that are slightly to strongly alkaline, on open flats or meadows, shady areas under shrubs, dry deserts, and along wet streamsides. It is also common on mineralized soils near hot springs. None of the subspecies is restricted to one par- ticular type of habitat, although subsp. gracilipes and longicarpum appear to favor sandy soils. Plants of the southern and western portions of the species range tend to be taller than those of the northern and eastern parts. However, there are a few exceptions. The tallest plants are found in subsp. longicarpum and affine, while the shortest are found in subsp. integrifolium particularly in the populations of northern Utah, Colorado, and Wyoming. In these populations some of the plants do not exceed one foot, while those of subsp. affine may exceed nine feet in height. Thelypodium integrifolium is a complex species. It is made up of a large series of populations that fall into morphologically distinct groups, which are geographically defined, and seem to represent natural infra- specific taxa. Some botanists, namely Payson (1923), recognized four species and two varieties in this complex. The characters he used in distinguishing these “species” (nectar glands and gynophore length) are probably the most variable and inconsistent in this complex. Abrams (1944), on the other hand, took the position that a single polymorphic species is represented, and he did not recognize any infraspecific taxa. The present writer believes that the taxonomy of this complex can be best served by recognizing these natur There seems to be a certain more can be profitably said about this. Each of the five subspecies can be determined rather easily, if the characteristic of the pedicels and the siliques are carefully studied and the geographic fruiting pedicels and siliques (Plate 21). BIOSYSTEMATICS OF THELYPODIUM 113 Thelypodium integrifolium can be easily distinguished from other species of the genus on the basis of its cauline leaves that are nearly always without auricles and by the nectar glands that are usually elevated and tooth-like, but present only around the lateral stamens. ) * SO} e ssp. integrifolium © ssp.complanatum e@ ssp. longicarpum 70] * ssp. gracilipes * * — a SSp.affine E e E 6O . ¢ Ps = s ae e® ® ) e 2 ee © ae Sige = =: @, 8 LJ - e* e ~ 40 4 . A af é a sec a : Q @ ® A e Aa ® . -* oo 30 © © § Pog tig a Py : ee a ae s ea a aa e 2 = * *e coe a sts a Bi Hx yO, © 0 cece e . 20 Pee hie tt fe 20 on al EY * se? x © Oe 6°? . von NaN . . ‘ mm ox"* eo®@ 10] m © 8 e° r) i 26.4 5 6 78 9°10 1 2 73 PEDICE & LENGTH (mm) Pate 21. Scatter diagram showing relationship of fruiting pedicel length to silique length in the subspecies of Thelypodium integrifolium. 114 IHSAN A. AL-SHEHBAZ 8. Thelypodium stenopetalum Watson Thelypodium stenopetalum Wats., Proc. Amer. Acad. 22:468. 1887. Holotype: stony hillside, Upper Lake (now known as Baldwin Lake), Bear Valley, San Bernardino Mountains, San Bernardino County, California, June 1886, Samuel B. Parish 1794 Plants biennial, glabrous and glaucous throughout; stems mostly branched from the base, decumbent to subdecumbent, sometimes simple or branc u ul at the ase, rarely auriculate, mostly lanceolate to oblong-lanceolate or oblong, rarely linear to lanceolate-linear, entire, obtuse. or t apex, (1.3)1.64. ) cm. long, floral buds oblong-linear; sepals erect, non-saccate linear to oblong-linear, obtuse, scarious at margin, purple above and green below or purple-green throughout, (6)6.5-9(10) mm. long, 1-1.5(1.75) mm. wide; petals linear to narrowly linear, i : .3-0.5(0. crisped mainly n claws and blades, (8)9 5-15(16.5) mm. long 75) mm. wide, blades horizontal, mos y lavender, or rarely laws 5) long; filaments erect, white to so t lavender, slightly dilated at e what tetradynamous, single filaments (5.5)7- long, paire filaments ( 7)8-12.5(14) mm. g; anthers exserted, Sagittate at base, linear to narrowly oblong, coiled when full dehisced, often purplish, mm. long; narrower and subtending base of paired filaments; fruiting pedicels stout, not flattened at base, ascending to divaricately ascending, rarely divaricate to horizontal, terete to rarely more or less quadrangular, slightly torulose, usually stout, ascending to divaricately ascending, rarely divaricate (2.2)2.8-4.9(6.3) cm. long, 1-1.5( 1.75) mm. wide; styles slender, terete, 1-2(2.5) mm. long; stigmas entire; gynophores stout, 0.5-3.5(5) mm. long; seeds 1-1.3( 1.5) mm. long, 0.75 mm. wide; cotyledons obliquely incumbent, or rarely incumbent. F lowering late May to early August at elevations from 6,500 to 6,800 feet in the Bear Valley of California. Lake, I. ¢& M. Al-Shehbaz 6947 (cH); E. end of Bear Lake, Munz 10463 (rom); N. shore of Baldwin Lake, Peirson 674] (RsA); Bear Valley, Parish 3787 (¥,cH,JEPS,ND-C, Us); near Big Bear Post Office, Yates 6634 (uc). Thelypodium stenopetalum is one of the most retricted species of the genus. It appears to be endemic to alkaline soils of the shores of Baldwin e and the nearby areas of Bear Valley. It was said to grow on stony hillsides by Robinson (1895), Payson (1923), Jepson (1936), and Munz (1959). However, these authors apparently depended on the original description of the Species, which cites such a habitat, as there are no other collections, to my knowledge, carrying that information. The pres- a gg! failed to find this species on the stony hillsides of Bear Valley in 2 BIOSYSTEMATICS OF THELYPODIUM 115 Thelypodium stenopetalum is rather unique in the genus in having narrowly linear petals, which distinguish it easily from the other species of Thelypodium. It can be distinguished also from the other auriculate- leaved species by a combination of the following characters: stout and usually ascending pedicels, oblong-linear to linear floral buds, and race- mose inflorescences that are long and lax. 9. Thelypodium howellii Watson Herbaceous biennial; stems mostly single from the base, usually branched above, sparsely to densely hirsute or hirsute-hispid near the base, glabrous and somewhat glaucous above or throughout, 1-8(9.3) dm. high; ‘ae! leaves Sabet to spatulate, less often oblong to lanceolate, lyrately lobed to dentate, sometimes entire or repand, obtuse or sometimes acute at apex, sparsely pubescent to more often erie usually glaucous, 2-10( 13.5) cm. long, 1-2.3(4) cm. wide; petioles slender, 0,.5-2 c long, ciliate near or at the base; —— — lanceolate to ———— some- times linear to oblong, acute to acuminate at apex, entire, often reat sagittate to amplexicaul at base, fu 8 auriculate, Dake ‘ous ead is ucous, (08) 1. .7(14.5) cm aoe 0.2-1(3) cm. wide; basal lobes lanceolate to oblong or ovate, so to obtuse, 1.5(2.3) cm. long, (0.5)1-6(11) mm. wide; ‘sling raceme, lax; sep ere mes scarious at margin, green in the center and lavender or purple at margin, or lavender throughout, (3. 5)5-8. 5(11.5) mm. long, median sepals linear to lanceolate- linear apa respon ee d near the apex, with the subapical seg e small] or up to 1 ong, sometimes slightly saccate at base, Spee )1-1.5(2) mm. wide, “yr pre often seadeclaie usually saccate at base, 1.5-2(2. linear or oblanceolate to spatulate, partly pe completely cris risped, lavender to purple, (6)9-16. 5(22) mm. long, (0.5)0.75-2.7(3) mm. wide; claws poet (2.5)4—7(11) mm. long; filaments erect, white to lavender or hater somewhat tetradynamous, single filaments curved at the base, (2. (10) mm. long ed filaments erect, my to completely fused or free, (3)5. 5-9( 12) mm. long; anthers purple, somewhat serted to nearly included, sagittate at base, linear-lanceolate to linear, (1.5)2-4(5 ong; glandular tissue flat, well-develo and surrounding base of single stam i ; fruiting pedicels stout, mostly mending. strai ght to sometimes slightly ¢ curved, 1-1.25(2) mm, wide; gynophores reer stout, — to nearly subclavate, (0.5) 1-2. 7(4. 3) mm. long; stigmas entire; (1)1. 7(2) mm. long, 0.5-0.7 mm. wide; cotyledons obliquely incumbent, rarely ote KEY TO THE SUBSPECIES Petals linear to narrowly oblanceolate, ents partly to se ene united; California, and southern Washington .............----+---- = — mostly spatulate, rarely oblanceolate, (1.2)1.5-2.7(3) mm. wi pai e pei mts free; eastern Orepon ... 9b. subsp. Ce Ya. Thelypodium howellii Watson subsp. howellii Thelypodium howellii Wats., Proc. Amer. Acad. 21:445. 1886. Holotype: Camp Polk, Crook County, Oregon, June 1885, Thomas Howell 343 (cx). ae Streptanthus howellii ( Wats.) Jones, Proc. c gals Sci., = ser., Thelypodiopsis howellii ( Wats.) O. E. Schulz, Bot. Jahrb. 66:98 116 IHSAN A, AL-SHEHBAZ Thelypodium simplex Greene, Pittonia 4:200. 1900. Holotype: adobe meadows, Dixey Valley, Lassen County, California, July 6, 1894, M. S. Baker & Frank Nutting i e: UC $n. (ND-G); iso 0G. Streptanthus coombsae Eastw., Proc. Calif. Acad. Sci, ser. 4, 20: 145. 1931 Holotype: Williamson River, Klamath County, Oregon, July 1913, A. L. Coombs s.n. (cas); isotype: GH. Basal leaves oblanceolate, mostly lyrately lobed, rarely dentate to entire; petals mostly linear, rarely oblanceolate to spatulate, with the blades usua y crisped most of their length, 0.5-1.2 mm. wide; paired filaments usually fused most of their length, seldom only fused along their lower third; siliques 1.5-4.6(7) cm. long; plants usually pubescent near the base. Flowering late May through July at elevations from 3,600 to 5,200 feet in central and southern Oregon, northeastern California, and southern Washington (Map. 2). REPRESENTATIVE SPECIMENS. California. Lassen Co.: 7.3 miles E. of Bieber, Cantelow 3764 (cass); Modoc Co.: Big Valley, Baker & Nutting, s.n., 1894 (cu, JEPS,POM,UcC). Oregon. Crook Co.: Big Summit Prairie. Ochoco Mts., between Prineville and Mitchell, T. 14 Sat E., S. 15, Cronquist 7461 (cu C : (F,CH,Mo, »NY,ORE,POM,RM,UC,US,ws ); Harney Valley, Howell s.n., 1885 (ps,F,GH, NA,NY,PH,UC,US,WTU ); e Paulina Marsh, Const 40 (PH,ORE,ws); Hunters Springs, near Lakeview, Constance 9541 (oRE,PH); Klama' o.: Harriman 9b. Thelypodium howellii Watson subsp. spectabilis (Peck) Al-Shehbaz, comb. nov. Thelypodium howellii Wats. var. spectabilis Peck, Torreya 32:150. 1932. Holotype: low alkaline meadow, 10 miles E. of Ironsi e, Malheur County, Oregon, June 19, 1927, Morton E. Peck 16066 (not seen); isotype: GH. Basal leaves oblong-lanceolate to oblong, mostly entire to repand; petals mostly spatulate, rarely oblanceolate, (1.2)1.5-2.7(3) mm. wide, i tween ; paired fil us blade and claw; aments free siliques (1.8 )3-6.5(8.2) cm. long; plants usually glabrous near the base. Flowering June through July in eastern Oregon. REPRESENTATIVE SPECIMENS, Oregon. Baker Co.: 3 miles N. of Baker City, eck 10416 (ps,ny); Harney Co.: Whitehorse Ranch, Peck 25641 (uc); Malheur Co.: 10 miles E. of Ironside on U.S, Hwy. 26, I. & M. Al-Shehbaz 6957 (cH); Union Co.: 18 miles N. of jet. US. Hwy. 30 and Anthony Lakes road at North Powder, I & M. Al-Shehbaz 68151 (cH); between Wolf Creek and North Powder, Howell 28611 (CAS,CH,osc,wtv ), The one unique feature of Thelypodium howellii subsp. howellii, that t present in any other taxon of the genus, is the union of the paired filaments. This character is well-known in the two related genera. Strep- BIOSYSTEMATICS OF THELYPODIUM 117 tanthus and Caulanthus, but it is far more common in the former. The union of paired filaments in Thelypodium was overlooked by Payson (1923), and was first reported by Jepson (1925). In her description of Streptanthus coombsae, Eastwood (1931) mentioned this feature of the filaments, but probably was unaware of its occurrence in T. howellii. This might have been the main reason for not recognizing the collection she had as a Thelypodium, but instead describing it as a Streptanthus. The lack of siliques in the type collection lead Morrison (1941) to recognize S. coombsae as a good species. He placed it in the subsection Pulchelli, section Euclisia of Strepthanthus. Kruckeberg (1958) questioned its inclusion in this section because of the different habitat in which it grows. We are certain that S. coombsae is properly referred to T. howellii subsp. howellii. Thelypodium howellii subsp. spectabilis is sufficiently distinct from subsp. howelli in the general aspects of the petals and filaments that it would be possible to consider it to be a distinct species. However, I have refrained from taking such a position because of the fact that the two subspecies are very similar in their fruiting pedicels, siliques, habit, and cauline leaves. Of the two subspecies, subsp. howellii has a wider distribution, but it appears to be more common in southern Oregon than elsewhere in its range. I have seen four specimens, unquestionably belonging to this taxon, from the state of Washington (one collected by Leckenby in 1898, and three by Vasey in 1889), but these are very old collections and it is possible that the species no longer grows there. Thelypodium howellii is easily distinguished from the other auriculate- leaved species of Thelypodium by its fruiting pedicels that are stout, straight, ascending, often forming a straight line with the siliques; by its lax racemes, ovate to lanceolate-ovate floral buds, and paired filaments that are usually united. 10. Thelypodium sagittatum (Nuttall ) Endlicher tu: younger ones, glabrous or sometimes sparsel : leaves smaller and broader than later ones, (2)6.5-20(29) cm. long, 1-4.2(5) scam wid petioles slender, ciliate at least near the base, (1)2-8( 14) cm. ; leav sometimes the lower ones sparsely pub somewhat thickish, sagittate to nearly clasping at base, uppermost leaves ate-linear and usu acuminat 1.3(2.8) cm. wide; basal (1)2-11(23) mm. long, 1-4.5(8) mm. wide; inflorescence ensely flowe ra corymbose, strongly elongated in fruit; sepals or linear-oblong, obtuse, scarious at margin, slightly to moderately sacca igh _ erect, green in the middle and lavender at margin, or purple throughout, (2.5)3-6(10) 118 IHSAN A. AL-SHEHBAZ mm. long, (1)1.5-2(2.5) mm. wide; petals pene to broadly oblanceolate or spatulate, obtuse to subtruncate at apex, white to lavender or purple, some slightly crisped between claw and blade, (5)7-14( 49) mm. one (0.5)1-3(4) mm. ide; claws erect, tapering toward base, (2)3-6.5(10) mm. long; filaments erect, somewhat dilated . “es tetradynamous, white to lavender or purple, filaments of 54.5 varic: siliques terete, rarely quadrangular, torulose, ey or somewhat incurved, erect to = goer — to stoutish, (1.0)1.3-5.3(6.9) cm. long, (0. 5)0. 75-1(1.2) rag ies psa a Doe “(0.25 )0.5-1(2) mm. long; see as (0.75) 1- 1.3(1.5) mm. long, 0.5-0.75(1) mm. wide; cotyledons obliquely incumbent, rarely incumbent. KEY TO THE SUBSPECIES Petals riers oblanceolate to ONG seldom narrowly oblanceolate, 1-3(4) mm. wide; fruiting pedicels 5-11(20) mm. long; inflorescence corymbose or sometimes racemose; median glandular tissues usually absent; widely distributed ............ ee rg he as Be oe da es + ney me © 10a. subsp. sagittatum. etals linear to narrowly oblanceolate, 0.5-1(1.5) mm. wide; fruitin pedicels He 5-7(9) mm. long; inflorescence ex raceme; median eae tissue esi southwestern Utah and adjacent Nevada ........... b. subsp. ovali lifolium. 10a. Thelypodium sagittatum (Nuttall) Endlicher subsp. sagittatum Pachypodium sagittatum Nutt. in Torr. & Gray, Fl. N. Amer. 1:97. 1838. Holotype: — the west side of the Rocky Mountains, T. Nuttall s.n., 1834 (photo cx of type at BM Thelypodium sagittatum ( Nutt.) Endl. in Walp. Repert. 1:172. 1842. Streptanthus sagittatus Nutt., Jour. Acad. Nat. Sci. Phila. 7:12. 1834. Holotype: Little Goddin River (now known as the Little Lost River), ppt towards the sources of the Columbia, N. B. Wyeth s s.n., 1833 (photo cu of type at Thelypodium sagittatum Die ) Heller, Bull. Torr. Bot. Club 25: 365. 1898 (based on Streptanthus s tt.). helypodiopsis sagittata (Nutt, ) O. E. Schulz, Bot. Jahrb. 66:99. 1 oes Laem nuttallii Wats., Bot. King’s Exped. 26. 1871 joerg on n Streptanthus ag Thelypodiopsis nuttallii (Wats.) O. E. Schulz, Bot. Jahrb. 66:99. 1933 Thelypodium amplifolium Greene, +H Sots :173. 1896. seed oe he Ranch, Pine Valley, Nevada, July 25, 1896, E. L. Greene s.n. ); isolectot Thelypodium + one Heller, Bull. Torr. Bot. Club 25:265. 1308. eae" on Pachypodium sagittatum vata trier (elle) O. E. Schulz, Bot. Jahrb. 66:99. 1933. a macropetalum Rydb., Bull. Torr. Bot. Club 29:233. 1 Holotype: 902. armington, "Devs County, “ng altitude 4,300 feet, May 14, 1881, Marcus E. Jones Ba (Ny); isotypes: Ds,NY,ORE,POM,US,UT Stems simple or branched from the base, with the branches cee © rai ci rape A mepiee — _. near the base; inflorescence corymbose, strongly elon: in fruit; sepals 3-6.5(10) mm. Jon tals ae to apeealane (5)7-14( . mm oe 1-3(4) mm. wide; claws 365 10) - ong; ge penne oe ng n stamens usually absent; fruiting pedicels horizontal varicate, , rarely divaricately saw to ascending, 5-11(20) mm. long, slender to stout; siliques (1.3)1.8-5.3(6.9) em, ae EPI) & BIOSYSTEMATICS OF THELYPODIUM 119 ° Thelypodium flexuosum . eo noapic - Sagittatum ssp. sagittatum * T- sagittatum ssp. ovalifolium O 100 200Miles Montana Wyoming e020? + (} 2? Oa Nevada California Colorado Map 4. Distribution of Thelypodium sagittatum, T, paniculatum and T. flexuosum. Flowering early May through July at elevations from 4,300 to 9,000 feet in central and northern Utah, northern Nevada, central Washington, central and southern Idaho, northern Colorado, southern Montana, and southeastern Wyoming (Map 4). REPRESENTATIVE SPECIMENS. Colorado. Jackson Co.: Nelson, Ripley & Barneby 8996 (cas); Routt Co.: Hayden, Osterhout 5222 (romM,nm). Idaho. Bannock Co.: Pocatello, Allen s.n., 1955 (TEx); Bear Lake Co.: 1 mi (cH); Caribou Co.: 3 miles W. of Soda Springs, Christ 16110 (Nx); Fr. 0.: mhi Co.: 10 miles E. of Gilmore, averhead Co.: Grasshopper Valley, Watson 39 miles N. of Contact on U.S. Hwy. 93, I. & M. bride 2085 (GH,MO,NY,RM,US,UTC, UC,US,UTC ); Humboldt Co.: é Barneby 9385 (CAS,NY,UTC); Lander Co.: 44 miles 120 IHSAN A. AL-SHEHBAZ W. of Austin, Goodner & Henning 645 (NA); Nye Co.: vicinity of Currant, Bently s.n., 1916 (Ds,POM,RM); White Pine Co.: 30 miles W. of Ely, Delameter s.n., 1947 (ps,RsA). Utah. Box Elder Co.: Brigham, Smith 2121 (cas,NA & . . 0.; near Loa, Jones 5649C (Pom,us). Washington. Douglas Co.: W. of Leahy, Daubenmire 59217 (ws); Grant Co.: Coulee City, Piper 3883 (GH,Ny,Us,ws ); Linco Co.: near Plum, Thompson 9146 (vus,wtv). Wyoming. Albany Co.: Tie Siding, Stevenson 108 (us); Carbon Co.: N. of Baggs, Osterhout 3196 (RM). In no other species of Thelypodium has the nomenclature been more confused than in T. sagittatum. As stated by Payson (1923), this confu- sion has resulted from the attempts of various taxonomists to maintain both Streptanthus sagittatus and Pachypodium sagittatum of Nuttall as distinct species. It is very likely that Nuttall himself considered the two to be distinct species, because he did not mention anything about S. sagittatus in his original description of P. sagittatum. Moreover, he cited a different collection for P. sagittatum which was published later. The statement that Nuttall may have considered P. sagittatum as a change of the generic status of S. sagittatus, as indicated by Payson (1923), does not have any support. Furthermore, the recognition of the two as distinct species by Watson (1871), Robinson (1895), Heller (1898), and a number of other taxonomists argue against Payson’s position. The types of the two “species,” as judged from their photos at the Gray Herbarium, show slight differences in the size of their flowers, cauline pedicels. However, such differences fall easily within the general range tunately been overlooked. Thelypodium nuttallii, a name proposed to accommodate the transfer of Streptanthus sagittatus to Thelypodium, originally represented a very heterogeneous assemblage as judged from the collections cited by Watson. This was clearly pointed out by Robinson (1895) in the following words: A species of considerable variability as interpreted by Dr. Watson, but dif- ficult to render more definite, owing to Nuttall’s brief description and fragmentary type.” Thelypodium sagittatum, as interpreted by Watson (1871), was even more heterogeneous, since he included collections that not only belong to T. sagittatum but also to T. flexuosum and Thelypodi- opsis elegans, Heller (1898) mishandled the nomenclature by arouj that the specific epithet sagittatus, first used in Streptanthus, should : es anes BIOSYSTEMATICS OF THELYPODIUM 121 transfer to Thelypodium, and that a new name (T. torulosum) should replace T. sagittatum (Nutt.) Endl., an earlier transfer to Thelypodium, that was based on Pachypodium sagittatum. Such an action by Heller was not in accordance with the present code of nomenclature. The type specimen of Thelypodium sagittatum must be the one col- lected by Nuttall. This is contrary to Payson’s (1923, p. 268) typification. He considered Pachypodium sagittatum to be a mere transfer of Strep- tanthus sagittatus to Pachypodium and not a newly published species. Therefore, he incorrectly assumed that the type specimen of S. sagittatus, collected by Wyeth, was also the type of T. sagittatum. Subspecies sagittatum as recognized in the present treatment is some- what heterogeneous. However, any attempt to give taxonomic recogni- tion to even some of the leading variants would not be of real help because most of these variants are neither well-defined morphologically nor do they have a geographical basis. The most variable characters in this subspecies are petal shape, flower size, inflorescence length, silique length, and pedicel length and orientation. 10b. Thelypodium sagittatum (Nuttall ) Endlicher subsp. ovalifolium (Rydberg) Al-Shehbaz, comb. nov. Jones 6015E (us); fragments from the type at Ny,POM. Thelypodium palmeri Rydb., Bull. Torr. Bot. Club 34:432. 1907. Holotype: Southern , 1877, E. Palmer 25 (Ny); isotypes: GH,MO,NA,NY,Us. According to an un- published “Miscellaneous Plant List” in the Gray Herbarium Library, this collection was made from Beaver Valley near Rio Creek. Stems simple or more often branched from the base, with the branches often subdecumbent; basal leaves glabrous or pubescent, petioles moderately to densely ciliate; inflorescence a lax raceme, slightly elongated in fruit; sepals 3-4.5(5) mm. long; tals oblanceolate-linear or narrowly oblanceolate, 5-7.5(8.5) mm. long, 0.5-1(1.5) mm. wide, claws 2-3.5(4) mm. long; median glandular tissue present; fruiting pedicels divaricately ascending or sometimes ascending to divaricate, stout to slender, (2.5)3.5-7(8) mm. long; siliques 1-3(3.8) cm. long. Flowering late May through mid August at elevations from 6,000 to 8,400 feet in southwestern Utah and adjacent Nevada. REPRESENTATIVE SPECIMENS, Nevada. White Pine Co.: Duck Creek, Jones $.n., 1893 (pom); 2 miles E. of junction U.S. Hwys. 50 & 93, Becky Spring, Train sorrd (Na). Utah. Garfield Co.: Panguitch Lake, between cabins and road, I. & M. z - Shehbaz 6911 ¢& 6984 (cH); Bryce Canyon National Park, Eastwood & Howell 766 (cas); Iron Co.: Sec. 13, T. 33 S., R. 7 W., Gierisch s.n., 1936 (utc). Thelypodium sagittatum subsp. ovalifolium shows some variation in the pubescence of the basal leaves and in silique length. These charac- ters were used by Rydberg (1923) to distinguish T. ovalifolium from T. palmeri. However, the present writer has collected both glabrous and 122 IHSAN A, AL-SHEHBAZ pubescent forms that also varied considerably in silique length within the same population. The basal leaves in subsp. ovalifolium are only ovate at early stages of rosette growth, becoming oblanceolate to spatu- late at later stages. This is usually the case in subsp. sagittatum, T. panicu- latum, and to a lesser extent in T. flexuosum. In his treatment of Thelypodium, Payson (1923) recognized T. ovali- folium as a species distinct from T. sagittatum. In fact, he considered it more closely related to and derived from T. howellii (p. 243). The only evident resemblance between these two is the presence of lax racemose inflorescences. However, subsp. ovalifolium is perfectly at home in T. sagittatum, if other characters are taken into consideration. Furthermore, corymbose as well as lax racemose inflorescences are found in subsp. sagittatum. Finally, these two subspecies are essentially indistinguishable in their mustard oil contents, and both are very distinct from T. howellii (see section on chemistry ). Thelypodium sagittatum often has been confused with certain auricu- late-leaved species of Arabis, particularly A. drummondii, but these two species have nothing in common other than superficial resemblances in their cauline leaves and similar inflorescences. Thelypodium sagittatum is sometimes confused with the species originally described as T. elegans. The latter species belongs, however, to the genus Thelypodiopsis which can be easily distinguished from Thelypodium by its bilobed stigma that has the lobes opposite the replum. The indications that Thelypodium sagittatum is present in Oregon (Davis, 1952: Peck, 1961) and California (Hitchcock, 1964) apparently are in error. The present writer has not seen a single specimen of this species from these two states, 11. Thelypodium paniculatum Nelson Thelypodium Paniculatum Nels., Bull. Torr. Bot. Club 26:126. 1899. Holotype: Fossil, Lincoln County, Wyoming, June 12, 1898, Aven Nelson 4673 (nm); isotypes: GH,MO.NY,vs, Thelypodiopsis paniculata ( Nels.) O. E. Schulz, Bot, Jahrb. 66:99. 1933. Thelypodium Sagittatum ( Nutt.) Endl, var. crassicarpum Pays., Ann. Mo. Bot. Gard. BIOSYSTEMATICS OF THELYPODIUM 3 9:269. 1923. Holotype: Yellowstone National Park, Yellowstone Canyon, Wyoming, August 24, 1899, Aven Nelson & Elias Nelson 6663 (Mo); isotypes: GH,NY,RM,US. Plants biennial or short-lived perennial; stems erect, simple or branched at base, mostly branched above, glabrous or seldom sparsely hirsute-hispid near the base, glaucous throughout, solid, (1.4)2-6.5(7.5) dm. high; basal leaves glaucous and often glabrous, mostly oblanceolate, sometimes oblong or spatulate, obtuse to rarely acute, entire, attenuate or occasionally cuneate at base, (3 )6—15( 22) cm. long, (0.6) 1-2.5(4) cm. wide; petioles ciliate at or near the base, slender, (0.8)2-4(6.5) cm. long; base, (1.7)2.2-6(9) cm. long, (0.3)0.5-1.5(3) cm. wide; basal lobes ovate to oblong or occasionally lanceolate-oblong, obtuse to slightly acute, (1)3-8(11) mm. long, 1-3(5) mm. wide; inflorescences corymbose, usually densely flowered, strongly elongated in fruit; sepals oblong to ovate-oblong, obtuse, scarious at margin, lavender to purple, slightly to moderately saccate and unequal at base, erect, (3)3.5-5(6) mm. long, (1)1.5-2 mm. wide; petals broadly spatulate to obovate or broadly obovate, slightly torulose, glabrous, somewhat glaucous before drying, erect to somewhat ascending, often slightly incurved, sometim i ( 2 (1.3)1.5-2.3 mm. wide; styles slender, (0.5)1-1.7(3) mm. long; stigmas entire; gynophores stout, 0.5-0.75(1.2) mm. long; seeds plump, (1.3)1.5-2 mm. long, 0.75-1 mm. wide; cotyledons obliquely incumbent, rarely incumbent. Flowering mainly from early June through July at elevations from 6,000 to 9,000 feet in western Wyoming, but mainly in northwestern Wyoming, Idaho and Montana; disjunct in northern Colorado (Map 4). REPRESENTATIVE SPECIMENS. Colorado. Larimer Co.: Camp Creek, Gooding 1466 ( COLO,DS,GH,MO,NY,PH,RM,UC,Us ). Idaho. Fremont Co.: Henry's e, Davis s.n., 1935 (xM,uTC). Montana. Madison Co.: Alaska Basin, A. & E. Nelson 5474 (GH,MO,ND,NY, POM,RM,Us ). Wyoming. Fremont Co.: ca. 24 miles S.W. of Dubois near upper end of Warm Springs Creek, Wind River Range, C. & M. Porter 7530 (DS,NY,RM,RSA,UC,UTC ) ; K N ); Platte Co.: Wheatland, Nelson s.n., Co.: 20 miles W. of Big Piney, E. & L. Payson 2647 (nM); Merna, E. & L. Payson 2741 (nM); Teton Co.: Jackson Hole, Hayden s.n., 1860 (mo); Bacon Creek, Nelson 9 (MoO,ND-G,NY,RM,us ); Uinta Co.: i ; Hayden Valley, n River, Hawkins 410 (us); Base of Bunsen Mountain, about 6 miles S.E. 0: Hot Springs, Rollins & Davis 57193 (GH,MICH,MO,NY,uCc ); Pelica (GH,NY). Thelypodium paniculatum is a relatively invariable species. The char- acters that are slightly variable, aside from the vegetative parts, are the silique and petal width. The populations of southwestern Wyoming have somewhat narrower siliques and petals than those of Yellowstone National Park and adjacent areas. 124 IHSAN A, AL-SHEHBAZ In contrast with the majority of species of Thelypodium, T. paniculatum is less sharply defined. It resembles its closest relative, T. sagittatum, in a number of morphological features. The two species have often been confused, and certain taxonomists who have dealt with them, including Payson, maintained a single species. However, Payson (1923) recognized the material from northwestern Wyoming and adjacent areas as a variety (crassicarpum) distinct from the rest of T. sagittatum on the basis of its having broader siliques. Although the type collection of T. paniculatum, collected from southwestern Wyoming has no mature siliques, there is no doubt that a single auriculate-leaved taxon, distinct from the typical T. sagittatum, is present in western Wyoming. Whether this distinct taxon is considered to be a variety of T. sagittatum or as a distinct species is a matter of judgment. In the present treatment, the rank of species is favored, based on a number of morphological, ecological, and chemical characteristics. The single collection cited from Colorado ( Goodding 1466) is disjunct from the main distribution center of the species. There is no doubt that this collection belongs here, since it is indistinguishable from the typical plants of Thelypodium paniculatum from northwestern Wyoming. It is not certain whether its presence in Colorado is the result of a recent introduction by man or whether it is a relict of a previous wider distri- ution. Thelypodium paniculatum can be easily distinguished from its nearest relative T. sagittatum on the basis of its wider petals and siliques, larger and plump seeds, and median glandular tissue that is almost always present. In T. sagittatum the petals and siliques are usually narrower, the seeds are somewhat flattened and smaller, and the median glandular tissue, with the exception of subsp. ovalifolium, is mostly absent. Meadows and stream bottoms that remain wet for most of the season appear to be the most favorable sites for Thelypodium paniculatum. The present writer has collected it from two localities in Yellowstone National Park, where it was growing in very wet sedge meadows. Dr. Rollins (personal communication) has made the same observation in a different locality. Thelypodium sagittatum, on the other hand, favors alkaline meadows that are often dry, but may be wet in the early part of the season. Such differences in the habitat of the two species were first pointed out by Nelson (1899), but were neglected by subsequent taxon- omists who dealt with the two species. The two species are extremely different in their mustard oil contents. As discussed above, the presence of two unique oxazolidine-2-thione producing glucosinolates in Thelypodium sagittatum and their absence in - paniculatum, and the presence of the 3-butenylglucosionlate as the dominant constituent in the latter species and its total absence from the former species, besides other differences, clearly stand against the idea of merging the two species. BIOSYSTEMATICS OF THELYPODIUM 125 Thelypodium paniculatum does not appear to occur in Washington. Payson (1923) cited one collection from that state (Vasey 194) under T. sagittatum var. crassicarpum, but the collection is, without question, T. howellii subsp. howellii. 12. Thelypodium flexuosum Robinson Thelypodium flexuosum Robins. in Gray, Synop. Fl. N. Amer. 1:175. 1895. Holotype; near Carson City, Nevada, 1865, C. L. Anderson 140 (cx) Plants perennial; tap roots s woody, rather long; caudex woody, simple or branched, covered with papery petiole remains of previous years; stems few to several sa th caudex, slender, weak, somewhat flexuous, subdecumbent or less often 0 ascending, solid, few-leaved to nearly naked above, glabrous and srg i pei cous, mostly branc ed at the orescence, 1.5-5.6(8) dm. high; ] leaves several, glabrous and somewhat seaoes, petioled, mostly lanceolate, ee oblong to oblanceolate, rarely ovate, acute to obtuse, entire, cuneate to attennate at base, what clasping at base, entire, ere rous and somewhat glaucous, much reduced above, ascending, 1-7(11) cm. long, 2-7(14) mm. wide; basal lobes linear to lanceolate-linear or lanceolate, acute to obtuse, 1-9(14) mm. long, 0. .5) mm. wide; inflorescence corymbose, few-flowered, si in fruit; sepals oblong to narrowly oblong, obtuse, somewhat scarious at margin, erect, green in the middle and white at margin, some- what saccate, 4.5) m loge 1-1.5(¢1.75) mm. e; petals mostly spatulate r sometimes oblanceolate to obovate, lavender to white, obtuse, not crisped, 6-9( 10) mm. long, (1.5)2-3(3.5) wide; claws erect, 2-3.5(4) mm. long; filaments erect, (2.5)4-9(16) mm. ee siliques terete, torulose, straight to slightly curved, erect to ascending, 1.1-2.5(4.2) cm. long, 0.75-1(1. 5) mm. wide; styles. slender, se 3)1-2(3) mm. long; stigmas entire, smaller than the diameter of style; hores 0.2— 0.5(1) mm. long; seeds uniseriate, wingless, non-mucilaginous, faintl ciara oblong, (1)1.3-1.5 mm. long, 0.5-0. 75(1) mm. wide; iia ie cbiicjucly nt to incumbent. Flowering late April through June at elevations from 3,000 to 8,000 feet in northeastern California, Oregon, Nevada, and western Idaho (Map 4) TIVE SPECIMENS. California. Lassen Co.: Dokin Unit, Honey Lake 1963 (cas); Modoc Co.: Surprise Valley, Lemmon s.n., 1879 — Battle Creek Ranch, between Riddle one ere Piemeisel 31-83 NA); P. ee Co.: 6 miles S.W. of Fa lend. Christ 942 Cc hill 28 Payete Buffalo Canyon and Ione Valley, Mills dx Beach C-87 (uc); : i i in 3547 NA,NY); Elko Co.: 20 miles S. suey 5 miles §. of Carson City, Train 3547 (cx, a a eee ts > Pi Valley, Kennedy se (nm); Lander Co me ee Eastwood rig 1 165A (cas,poM); Nye Co.: vicinity of Reese River Ranger Station, Maguire é> Holmgren 25485 (Ds,GH,NY,UC,US, 126 IHSAN A, AL-SHEHBAZ utc,wtu); Pershing Co.: 5 miles N. of Oreana, Train 28 (GH,NA); Washoe Co.: Empire City, Jones 3771 (CAS,DS,NY,POM,UC,Us,UTC); White Pine Co.: near Connor’s Station, Eastwood & Howell 9363 (CAS,DS,F,GH,POM,ws ). Oregon. Harney Co.: 21 miles N. of Frenchglen, Hitchcock ¢+ Muhlick 21198 (Ds,NY,RM,Ws,wTu); between gui Klamath Co.: White Lake, W. of Merrill, Applegate 3481 (ps,F,GH,UC,ws); Lake Co.: 5 miles S.E. of Paisley, Hitchcock 6763 (DS,MO,NY,POM,UC,UTC,ws,wTu ); 2 miles S. of jet. U.S. Hwy. 395 and Silver Lake Road, Steward & Smith 7120 ( CAS,DS,NY,0SC,UTC, ws,wtu); Malheur Co.: about Vale, Henderson 8445 (cAs,ORE); 40 miles from Fields on road to Crooked Creek Springs, Hitchcock & Muhlick 22292 (ps,Ny,osc,RM,UC,UT, ws ); Union Co.: near Medical Springs, Gale 263 (ps). Thelypodium flexuosum grows primarily on moderately to strongly alkaline sandy-loam or clayish soils of open deserts. It is unique in the genus in being a weak-stemmed perennial that survives by its thick and woody base. The weak, slender, nearly leafless stems derive their support by being tangled among desert shrubs. The charcters that readily distinguish Thelypodium flexuosum from other auriculate-leaved species of the genus are: woody caudex covered with petiole bases of previous years growth, persistent basal leaves that wither only after the fruits are dry, petioles that are glabrous, stems that are weak, slender, and usually somewhat flexuous. Although Thelypodium flexuosum is fairly widely distributed, it is probably one of the least variable species in the genus. The basal leaves, cauline leaves, and auricles vary in size as in other species, but the varia- tion in the length of the siliques and the size of the flowers is much less than in the related T. sagittatum. The degree of branching of the caudex and its size are sometimes variable within a given population, but this is probably due to the differences in the age of the different plants. 13. Thelypodium wrightii Gray Plants biennial, glabrous and somewhat glaucous throughout; stems erect, simple or rarely branched from the base, paniculately branched above, (1.5)3.2-16.3(28) dm. high; basal leaves lanceolate to sometimes oblanceolate in i i entate near the tip, rarely entire, (6.5)9.5-22.5(28) cm long, 2 = wide; lobes linear to oblong to deltoid, acute, entire or denticulate, (0.5)0.9-2.5(3.2) long, 0 . wide; petioles glabrous, (2) 7) cm. long; cauline leaves short petioled, ascending, lanceolate to linear-lan ¢ more I or short racemose, strongly elongated in fruit; sepals mostly linear-oblong, Sometimes oblong or linear, green in the middle and white at margin, or white to lavender throughout, obtuse, scarious at margin, entire, equal and not saccate at base, spreading to mostly horizontal at anthes (3)4-6(7) mm. long, (0.75) 1-1.3(1.5 — eG, 4-7.5(9) mm. long, 1—1.75( 2) mm. wide, blades mostly oblong, sometimes i me or Sometimes slightly tetradynamous, somewhat dilated at base, spreading at anthesis, (3)3.5-6.5(8.5) mm. long; anthers oblong to linear-oblong, exserted, sagittate at base, BIOSYSTEMATICS OF THELYPODIUM 127 (1)1.5-2.5(3) mm. long; glandular tissue continuous, low or sometimes elevated, subtending base of paired stamens, noe ing base of =. stamens, usually flat, or sometimes 6-toothed, with the teeth extending betw n bases of filaments; in- attened parallel to septum or more or less terete, torulose, straight to somewha curved, often horizontal to reflexed, sien an! divaricate, rarely divaricately ascend- ing, (2. 5)3.8-7.4(9) cm. long, 1-1.2(1.5) mm. wide; styles stout, davis to sub- clavate, rarely cylindrical, (0.5)0.75-2(3) mm. long; nee entire; gynophores stout, 0.2-2(5) mm. long; seeds 0.75-1.3(1.5) mm. long, 0.5-0.75 mm. wide; cotyledons obliquely accumbent, rarely accumbent. KEY TO THE SUBSPECIES 5 a am Aas — 75( bas mm. long; siliques (3.8)4-7.4(9) ~ long; plants of as, Ariz New Mexico, and northern Mexico .......... subsp, wrightii. ete (i! 5)9-4(5) m mm. eg ret siliques (2.5)3-4.8(5) — long, oan of north- western Oklahoma and southeastern Colorado .......... subsp. oklahomensis 13a. Thelypodium wrightii Gray subsp. wrightii sir ge rie Gray, Smiths. Contrib. (Pl. Wright.) 3:7. 1852. Holotype: Pass of the Limpia, Jeff Davis County, Texas, August 1849, Charles Wright 7 (cx); iso otypes 2GH,uUS. Stanleyella i (Gray) Rydb., Bull. Torr. Bot. Club 34:435. 1907. Glandular tissue low, more or less evenly developed, seldom slightly pointed between filament bases; siliques divaricate to horizontal or reflexed, rarely divaricately ascend- ing, straight or occasionally incurved, (3.8)4-7.4(9) cm. long; gynop phores obsolete, 0.25—0.75( 1) mm. lon Flowering mid-June through October, known to flower at earlier dates in Baja California (late March, Harbison 14846). Distributed in south- western Texas, New Mexico, Arizona, and central to northern Mexico (Map 5) at elevations from 2,000 to 8,500 feet. RESENTATIVE SPECIMENS. Arizona. Apache Co.: Salina coal mine, Deaver tei}, Cochise Co.: a ~ hig? Park, ea Mts., Goodman & Hitchcock 1208 (cas,F,GH,MICH,MO,P ; Coconino Co.: Bright Angel Trail, Grand Canyon National Park, l.& M. AL Shehbex 6992 (ct); Cosnino, Jones 4051 (GH,NY,POM,UTC); ila Co.: 10 miles N.W. of Pine, Wolf 2448 (cas,cH,POM M,RSA ); Maricopa Co.: near Saddle Mt., T. 7 N., R 8 E., Tonto National Forest, "Tuttle 4151 (NA); ‘Mojave Co.: anyon National Monument, Cottam 14069 (cu,vuT); 2 miles H,uc). New Mexico. Catron rae Aa gre Mts., ‘ax i 7 miles oO (om); Ri Arriba Co.: E El Rito 0 picnic groun , Carson Nations? Forest, El “deg; Neboa 470 (uc); Sandoval Co.: Jemez Springs age tise Nelson peopl (GH, uc); San Miguel C of Las Vegas, pial ge prasd-ts Bea sees Greene s.n., 1880 (¥,ND-G,NY,PH,POM 128 IHSAN A, AL-SHEHBAZ Torrance Co.: Manzano Mts., N.W. of Manzano, Barneby 12851 (cas). Texas. Brewster Co.: Paisano campground, about 10 miles W. of Alpine on Hwy. 90, Correll 34106 (cH); Chisos Mts., Big Bend National Park, Mueller 8003 (¥,Ny,TEX); Culberson Co.: 30 miles N. of Van Horn, Sierra Diablo Mts., Correll 31617 (cx); im ; Vv (uc); Davis Mt., Palmer 31839 (NY,PH,POM); Presidio Co.: San Esteban Lake, 12 miles - of Marfa, Hinkley s.n., 1940 (cH). Utah. Garfield Co.: Marvine colite, Jones s.n., 1894 (pom). Mexico. Baja California: Santa Catarina, 64 miles S.E. of Ensenada, Broder 330 (us); San Matias Pass, near Diablito Spring, Valle de la Trinidad, Harbison 14846 (ps); road N. of Sam’s Corral, 31°04’ N, 115°34’ W, Sierra San Pedro Martir, ; Sierra del Carmen, Muller 621 (GH,NY); ) 13b. Thelypodium wrightii Gray subsp. oklahomensis Al-Shehbaz, subsp. nov. Siliquis erectis vel adscendentibus, rectis vel incurvatis, (2.5)3-4.8(5) cm. longis; stipite tenui vel crasso, (1.5)2-4(5) mm. longo. Holotype in the Gray Herbarium, collected among rock crevices, 4 miles E. of Kenton, Cimarron County, Oklahoma, August 18, 1950, G. J. Goodman & R. W. Kelting 5358; isotypes: COLO,NY,RSA,TEX,WTU 0.75 mm. high, single teeth inserted between and outside base of paired stamens; siliques erect to ascending, sometimes divaricately ascending, rarely divaricate, often ongly incurved, rarely straight, (2.5)3-4.8(5) cm. long; gynophores slender or rarely stout, (1.5)2-4(5) mm. long. Flowering late July through early October in northwestern (panhandle) Oklahoma and southeastern Colorado, This subspecies was first reported from Oklahoma by Goodman (1936) as Stanleyella wrightii. ENTATIVE SPECIMENS. Colorado. Baca Co.: Sand and Gallinas canyons, 27 miles S. of Pritchett, T. 35 S., R. 48 W., Sec. 5 & 8, Weber 4322 (coto); 4 miles N. of Oklahoma line, N. of Kenton, Rogers 6432 (coto); Chaffee Co.: Salida, Harper s.n., ; Fremont Co.: Canyon of the Arkansas, 18 miles $.W. of Canyon City, Waterfall 10888 (rsa); Las Animas Co.: Trinidad. Osterhout 5755 (nam). Oklahoma. Cimarron Co.: John Regnier Ranch, Kenton, Demaree 13386 (cH,NY); 3 miles E. of Kenton, Goodman 2287 (CAS,GH,NY). Thelypodium wrightii grows primarily on clay-loam to sandy-loam soils that are usually moist. It is often found in sheltered areas of oak-wood- lands or pinyon-juniper associations along rocky slopes, cliffs, and ledges of canyons and river banks. The characters that are most variable in Thelypodium wrightii are leaf margin, length and orientation of fruiting pedicels and siliques, length of BIOSYSTEMATICS OF THELYPODIUM 129 the inflorescence, and density of siliques in the infructescence. These characters are, by and large, variable within a given population, and do not seem to fall in any particular geographical patterns. Thelypodium wrightii, as discussed before, was the basis for establish- ing the genus Stanleyella by Rydberg mainly on the grounds of its having spreading instead of erect sepals as in the majority of Thelypodium species. However, systematists working with the Cruciferae realize how insignificant such a character can be. The only other character, found in T. wrightii, that is not common in Thelypodium is the shape of the style, which is clavate to subclavate. However, this type of style characterizes two other species, T. laxiflorum and T. repandum, that also have spread- ing sepals and it occurs sporadically in other species, including T. lacini- atum. In his treatment of Thelypodium and its immediate allies, Payson (1923) stated that the filaments are folded in the floral buds of T. wrightii (as Stanleyella). However, the present writer failed to detect such a character in any of the materials studied, including herbarium specimens as well as living material in the field and greenhouse. Thelypodium wrightii was said to be present in Colorado and Utah by Payson (1923), Rydberg (1923), and Tidestrom and Kittell (1941), but the plants these authors were referring to are placed in T. laxiflorum in the present treat- ment. Kearney and Peebles (1960) reported T. wrightii from Pinal, Pima, and Greenlee Counties of Arizona, but I have not seen specimens from these counties. I have found no reason to question the correctness of their identifications, but it is not known where the specimens they used are presently located. Until recently, the distribution of Thelypodium wrightii in Mexico was thought to be restricted to the northern portions of the country. Dr. Rol- lins has kindly brought to my attention a specimen sent to him by G. C. Rzedowski, who collected it from the state of Hidalgo. This collec- tion certainly expands the distributional range of the species much farther south than what was known before 1973. Whether the species is disjunct in Hidalgo or has a continuous distribution with its range in northern Mexico is not known. The single collection from Utah (Jones s.n., 1894, POM) is somewhat disjunct, but there is no doubt that it belongs here. It is not unlikely that T. wrightii will be found in other localities of southern Utah. 14. Thelypodium laxiflorum Al-Shehbaz, sp. nov. i i ~ &) Thelypodium wrightii Gray var. tenellum Jones, Proc. Calif. Acad. Sci. (ser 5:622. 1895. ee abel Slate Canyon, Provo, Utah County, Utah, alt. 6,000 feet, July 2, 1894, M. E. Jones 5559 (Pom); isotypes: NY,US. Stanleyella wrightii (Gray) Rydb. var. tenella (Jones) Pays., Ann. Mo. Bot. Gard. 9:317. 1923. 130 IHSAN A. AL-SHEHBAZ | a ® Thelypodium laxiflorum & e T- wrightii ssp. wrightii @ 7: wrightii ssp. oklahomensis Idaho : i repandum are e ° ° 97 es s MEXICO ye © %% © 4 Map 5. Distribution of species of Thelypodium with spreading sepals. Herba biennis; cau ramosis, farctis, inferne hi ispidis-hirsutis vel glabrati: Superne orm i 5)3. Senh din, altis; foliis radicalibus petiolatis eS ectieen yratis, glabris vel sparse pubescentibus, (4 )7.2-20.5(30) cm. longis, bn din. 10) < cm. latis; lobis i ed vel oblongis, integris vel dentatis; foliis superne ibus vel sracoreen integris ve] dentatis, acutis vel acuminatis, 15-7(10) cm. longis, 0.3-1.2(2) em. lati i bus erectis, tetradynamis, 2.5-4.5(5.5) mm. longis; antheris sagittatis, inclusis, (1) 1.5-2.2(2.5) mm. longis i iferi i ; pedice ctiferis tenuibus, divaricatis vel reflexis, (4)5- 13(19) mm, = = stipitatis, teretibus, ragenags submoniliformibus, divaricatis vel reflexis, curvatis, (2)3-6.6(7.4) cm. longis, 0.75-1(1.5) mm. latis, stylis lavatis, (0 (0. 5)0. 75-2(5) mm. mm. longis; stigma tibus parvis, integris; seminibus , oblo: , Uniseriatis: in the Gray =netamicoge on sandy soil along roadside, ca. 2 miles east of Glenwood spring, Garfield Prooniags Chivas, une 3, 1968, Larry C. Higgins 1487; isotypes to be distributed. i 3 BIOSYSTEMATICS OF THELYPODIUM 131 Plants biennial; stems simple or sometimes branched from the base, branched at the ascending, ee linear to pate Pee or lanceolate, sometimes oblanceolate t oblong, sinuate to dentate, occasionally entire or repand, rarely lobed near the base sae to acuminate, 1.5-7( 10) cm. long, 0.3-1.2(2) cm. wide; inflorescence few bose, somewhat lax, greatly elongating in fruit; sepals white or some- times lavender, oblong, obtuse, slightly scarious at the margin, spreading or, more often, ascending, non-saccate, equal at base, (2.5)3-5(5.5) mm. long, 1-1.7(3) mm. wide; petals white or rarely lavender, mostly spatulate, sometimes oblanceolate or obovate, obtuse to subtruncate e ae no se erect below and horizontal above, C15 in s clavate to subclavate, slender to stout, (0.5)0.75-2(5) mm. long; stigmas a phores stout, 0.5-0.75( 1.5) mm. long; seeds ( ape 3-1.75(2) mm. long, 0.5-1 mm gyno s wide; cotyledons obliquely incumbent, rarely incum Flowering late May to early September at elevations from 4,100 to 9,000 feet in southern Nevada, Utah, and western Colorado (Map 5). SPECIMENS, Colorado. Archuleta Co.: along trail to Chimney Rock REPRESENTATIVE Mesa, Piedra, Schmoll 1496 (coLo); Eagle Co.: road from Dotsero towards Coffee i White River Plateau, Weber 12049 (coLo); Minturn, Osterhout 2553 (ny, Pot Springs, ); Wolcott, Osterhout 4237 (Ny,PoM.RM); Garfield Co.: near Glenwood Springs, Rollins 57151 (cu); G on Co.: road from Kebler Pass towards 277 (coLo); Lead King Basin, Crystal, Langenheim 2120 ( ); La Plata Co Durango, Baker, Earle ¢> O (¥,GH,ND-G, zum. 3 Spruce anyon near jct. with Navajo Canyon, Weber 3628 Pree ek ; Pitkin Co Coal Creek Basin, Langenheim 1432 (coo) —— 0.: Panaca, Jones s. 1912 (Ny,pom); Kershaw Canyon, Ripley & Barneby 4395 (cas,cH,NY); Ryan State ye Co.: southwestern rim of Imes Road, S.W. end of Belted Range, Beatley 9426 (cx); tributary of South Silent Canyon, E. of Well 9, N. of Pahute Mesa Ri Troy pe western base of Troy Mountain, Sharsmith 4819A (cu,uc). Utah. Carbon Co.: E. of S apne W. of Tavaputs ye Welsh ¢> Murdock 9145 (cu); Iron Co Co.: shots Allen Creek, Mt. Linnaeus, Abajo | Cronquist & N. Webiees 9487 (y,ws); Utah Co.: Bridal Veil Falls, Provo easy I. & M. Al-Shehbaz 6879 (cx); Washington Co.: Zion National Park, Thackery 528 (Na,uc). The ecological counterpart of Thelypodium wrightii in Nevada, Utah, and western Colorado is T. laxifloru rum. The two species inhabit slightly 132 IHSAN A. AL-SHEHBAZ moist areas that are partly shaded, and sometimes in association with Populus tremuloides, scrub oak, Artemisia-Pinyon-Juniper, and others. Thelypodium laxiflorum exhibits an interesting pattern of stem pubes- cence. Plants of the populations from Nevada seem always to be glabrous- stemmed, while those from Utah and Colorado are primarily pubescent near the base. However, no taxonomic recognition is given to the popu- lations from Nevada for two reasons. First, the presence or absence of pubescence in the Cruciferae can be under simple Mendelian genetic control, as discovered by Rollins (1958) in the genus Dithyrea, and second, both glabrous and pubescent forms were found within the same population in a number of localities. Plants of Thelypodium laxiflorum were placed in T. wrightii by earlier workers, who overlooked the significant differences in the flowers and siliques. Payson (1923) maintained T. wrightii var. tenellum (under the segregate genus Stanleyella), but cited only the type collection of this variety. He cited a number of collections from Colorado, Utah, and Nevada, that unquestionably belong to T. laxiflorum (var. tenellum sensu Payson), under T. wrightii. As did Jones, Payson recognized that the plants collected from Slate Canyon (the type locality of var. tenellum) were different from typical T. wrightii in their “delicate” cauline leaves, pubescent stems, and lax infructescences, but he too failed to detect the significant differences in the flowers and siliques (Plate 22). There is no question but that two distinct species exist. The present writer favored the presentation of this taxon as new at the species level rather than making a new combination based on var. tenellum whose fragmentary type collection has no flowers. The series of populations that belong in Thelypodium laxiflorum re- semble those of the closely related T. wrightii in certain morphological features such as leaf shape and margin, fruiting pedicels, glandular tissue, and style shape. However, this is where the significant similarities stop. The flowers and siliques of the two species are very different (Plate 22). In T. laxiflorum the sepals are ascending to spreading. The petals are spatulate to oblanceolate, 1.5-2.5(3.5) mm. wide, not sharply differenti- ated into claw and blade, and with their claw-like lower half almost always erect. The stamens are tetradynamous and erect. The siliques are submoniliform, with the replum variously constricted between the seeds. In T. wrightii the sepals are spreading to variously reflexed, the petals spreading and differentiated into claw and blade, with the blades oblong to linear-oblong and 1-1.7(2) mm. wide. The stamens are always spreading and more or less equal in length. The siliques are torulose, with the replum straight and not constricted between the seeds. 15. Thelypodium repandum Rollins Thelypodium repandum Rollins, Contrib. Dudley Herb. 3:371. 1946. Holotype: 20 miles south of Challis, on soft shale cliffs west rp River, Custer County, Idaho, BIOSYSTEMATICS OF THELYPODIUM 133 June 15, 1944, C. L. Hitchcock & C. V. Muhlick 8985 (ps); isotypes: Ds,MO,NY,RSA,UC, WS,WTU Plants biennial or short-lived perennial; stems simple or branched at base, usuall branched above, fairly robust, glabrous and somewhat glaucous, 1-4.4(6.0) dm. high; basal leaves and lower cauline leaves glabrous and distinctly glaucous, usually gray to bluish-gray, petiolate, rather fleshy, mostly ovate, less often obovate or elliptic, rarely orbicular or spatulate, obtuse to acute, cuneate at base, usually repand and lyrate, sometimes sinuate-repand or dentate, rarely entire, (2.2)4-10.5(14) cm. long, (1) 1.5-4.3(5.5) cm. wide; petioles glabrous, 14(6) cm. long; uppermost cauline leaves short-petioled, elliptic or often lanceolate, entire or repand; inflorescence rather densely- ered, raceme short; sepals purple to lavender, spreading to horizontal at anthesis, oblong, obtu: hat scarious at the margin, not saccate, (2.5)3-4 mm. long, paired stamens; fruiting a divaricately ascending, slightly flattened at the base, 4-12(15) mm. long; siliques n linear, torulose, somewhat flattened parallel to septum, erect to ascending, 1-1.5(1.75) mm. wide; styles cylindrical to subclavate, 0.5-1.2( 1.5) mm. long; stigmas entire; gynophores 0.5-0.75(1) mm. long; seeds (0.75)1-1.5 mm. long, 0.5- 0.75 mm. wide; cotyledons obliquely accumbent, rarely accumbent. Flowering in June at elevations around 5,700 feet in Custer County, REPRESENTATIVE SPECIMENS. Idaho. Custer Co.: 8.1 miles S. of Challis on U.S. Hwy. 93, along west facing steep canyon bank of Salmon River, I. & M. Al-Shehbaz 6921 (cH); 5 miles up South Fork of Salmon River, Stanley-Challis Road, Hitchcock & Muhlick 10805 (ps,wru); 8 miles S. of Challis, Hitchcock & Muhlick 14120 (cas,ps, F,MO,NA,NY,PH,RM,RSA,UC,UTC,WS,WTU ); miles E. of Clayton, Hitchcock 15674 (wu); 8 miles E. of Clayton, Salmon River Canyon, Ripley & Barneby 8837 (GH,NY). glaucous. 16. Thelypodium texanum (Cory) Rollins Sisymbrium texanum Cory, Rhodora 39:418. 1937. Holotype: Terlingua Creek, 18 miles north of Terlingua, cared County, Texas, April 13, 1936, V. L. Cory 18564 ( GH). Stanleyella texana (Cory) Rollins, Madrofio 5:134. 1939. Pialgpidinues texanum (Cory) Rollins, Contrib. Dudley Herb. 3:371. 1946. 134 IHSAN A. AL-SHEHBAZ e, ss TE 22. Flowers rage a of Siacing mage laxiflorum (A) and T. = (B). Fic. 1. ower, top view: Fic. 2. er, side ; Fic. 3. Androecium, median view ; Fie. 4. ‘Adeenitun: lateral view; Fic. 5. otto cS one mm. BIOSYSTEMATICS OF THELYPODIUM 135 Plants annual; stems usually simple or rarely branched from the base, mostly branched above, glabrous and somewhat glaucous, 1.3-4.8(6.1) dm. high; basal leaves smaller and remote on lower portion of the leaf, divisions between lobes usually extending to midrib below and becoming gradually shallower toward leaf apex; upper- most leaves short-petioled, nearly pectinate, with the lobes usually linear; inflorescence a densely flowered raceme, short, greatly elongated in fruit; sepals oblong to linear oblong, obtuse, entire, scarious at margin, green above and white below, or lavender throughout, not saccate, equal at base, spreading or sometimes reflexed at anthesis, usually coiled later at anthesis, (2.5)3-5 mm. long, 1-1.5 mm. wide; petals mostly spatulate or sometimes oblanceolate, not crisped, obtuse, attenuate to claw-like base, straight, spreading at anthesis, white, undifferentiated into claw and blade, (3.5)4— 6.5(7) mm. long, (0.5)1-2 mm. wide; filaments equal in len or rarely tetra- air slightly reflexed, more or less flattened at the base, striate, strai nore slightly incurved, (4)6-14(17) mm. long; siliques narrowly linear, often divaricately styles conical to subconical, rarely cylindrical, 0.75-1.5(3) gynophores 0.5—1(2) mm. long; seeds (1)1.25-1.5 mm. long, 0.75-1(1.25) mm. wide; cotyledons mostly obliquely accumbent, rarely accumbent. Flowering early February through April in southwestern Texas on clay or sandy soil on barren hillsides and gravelly creek beds. TATIVE SPECIMENS. Texas. Brewster Co.: Terlingua Beds, 7 mil REPRESEN les E. o Lajitas, Correll 30685 (cu,t); Tornillo Creek,, near Rio Grande Hot Springs, Big an ll ; 65 miles S. of Co.: abo iles SE. of Redford, Correll 30681 (x11); 0.: about 24.5 miles S.E. o ord, Corre owes . above Fresno mine, between Fresno Canyon and Rio Grande River, Rollins & Correll 61106 (cH). Thelypodium texanum, like T. tenue, is endemic to southwestern Texas, but it is certainly far more common than the latter species. It is some- what variable in the various aspects of its leaves, but aside from the slight variability in fruit orientation and length of pedicels, the species as a whole is only slightly variable in other morphological — Like Thelypodium tenue and T. paysonii, T. texanum is a winter ann é The greenhouse plants of the last species took a minimum of 8 to 10 wee to grow from the seedling to the flowering stage. No dormancy require- ments were evident, since mature seeds, removed from plants that were still growing, germinated readily with a high percentage (80-1002). 136 IHSAN A. AL-SHEHBAZ One interesting feature found in some plants of Thelypodium texanum that was not observed in other species of the genus, is the emergence of pistils from the mature floral buds before they open. This is a clear-cut case of protogyny. The significance of this phenomenon has already been discussed under the section on breeding systems. 17. Thelypodium tenue Rollins Thelypodium tenue Rollins, Rhodora 59:64. 1957. Holotype: bed of Fresno Creek, about 1 mile below Smith Mine, Presidio County, Texas, January 25, 1942, L. C. Hinckley 2336 (cu). Herbaceous annual; stem single from the base branched above, glabrous and glaucous, over 3 dm. high; basal leaves and lowermost line leaves somewha thickish, oblanceolate in outline, petiolate, pinnately lobed, glabrous and glaucous, long, 2-4 cm. wi g wide; lobes acute, irregularly dentate or somewhat sinuate, smaller and remote on lower portion of lea cu on rachis, divisions between lobes extending to midrib; petioles expande base; uppermost anthesis, oblong, 3-5 mm. long, 1-1.5 mm. wide; petals straight, spreading at anthesis, spatulate, not crisped, attenuate at base to short claw, 5-8 mm. long, 1-1.5 mm. wide; laments spreading at anthesis, white, equal in length or slightly tetradynamous, 5-8 mm. long; anthers purplish, oblong to linear-oblong, sagittate at base, exserted, 2-2.5 mm. long; glandular tissue continuous, subtending of filaments; pedicels very slender, striate, divaricately ascending, slightly flattened at base, 2-3.5 cm, long; young siliques stipitate, narrowly linear, slightly flattened par- allel to septum; styles sh seeds not seen. The fact that Thelypodium tenue is similar to T. texanum in many morphological features of the leaves and flowers is readily shown by comparing specimens of the two species. However, the present writer has seen only the holotype of the former species. Further collections should provide valuable information particularly on fruit morphology, clearly pointed out by Rollins (1957a). It is easily distinguished from the other species of Thelypodium by its pedicels that are, by far, the longest in the genus (2-3.5 cm. long). Long and slender pedicels are also found in the remotely related T. wrightii and T. laxiflorum, but they are shorter than those of T. tenue. The shape of the Rollins (1957a), T. conical styles are more related to each other than to T. wrightii and T. laxiflorum which have clavate to subclavate styles. This is supported by 5 ical P . BIOSYSTEMATICS OF THELYPODIUM 137 to T. texanum than to T. wrightii or T. laxiflorum on other morphological grounds, but resembles the last two species in having clavate to subclavate styles as well as being a biennial. 18, Thelypodium paysonii Rollins Thelypodium paysonii Rollins, Rhodora 59:61. 1957. Holotype: Cajion de Jara, east of Socorro, lower part of canyon near its mouth, about 30 km. west of Cuatro Cienegas, Coahuila, Mexico, February 1-15, 1941, Albert H. Schroeder 12 (cn). cm. long, 1.8-6(8) cm. wide; lobes narrowly to broadly oblong, rarely ovate or oblong- lanceolate, acute to obtuse, glabrous or rarely ciliate, dentate or sometimes sinuate- dentate, rarely repand or entire, smaller and remote on lower portion of the leaf; petioles slender, mostly pubescent, 2-6(8.5) cm. long; upper cauline leaves reduced, with the lobes linear to lanceolate-linear; inflorescences densely flowered, corymbose, greatly elongated in fruit, rachis usually pubescent; sepals oblong, obtuse, somewhat scari the margin, non-saccate, equal at the base, green above and white below, ascending or less often spreading at anthesis, 3-4(4.5) mm. long, (1)1.5-2 mm. wide; petals differentiated into claw and blade, with the blades spatulate to obovate, white, ot crisped, obtuse or rarely subtruncate, spreading at anthesis, 4-6(6.5) mm. long, ightly flattened parallel to septum, rizontal or usually pendulous, seldom divaricately ascending, (0.6)2—4.8(5.3) cm. long (1)1.3-2 mm. wide; styles slender, subconical or less often cylindrical, (0.75) 1.5-2.5 : 1-1.5 mm. Known to flower in November and February in northern Mexico. This variation in flowering dates and growth cycle may well be related to the availability of moisture. The species appears to grow primarily on rocky slopes or canyon walls. REPRESENTATIVE SPECIMENS. Mexico. Coahuila: casei sean rb : va Tpus ; Si de Parras, Pu 0. cH); miles E. Torreon, Pu 130 (cu,uc); Sierra de rpus er ee of Torreon, Rollins & Tryon 58296 (GH,MICH,MO,NY,UC,US ); Duran Torreon, Rollins & Tryon 58291 (GH,MICH,MO,NY,UC). The specimen collected by Palmer (2146), and identified by Watson as Sisymbrium, is different from the other specimens cited above in hav- ing much shorter siliques (0.6-2 cm. long) borne on much shorter ped- icels (2-5 mm. long) that are straight and divaricate. It may represent an undescribed infraspecific taxon of Thelypodium paysonii but I have refrained from describing it because the specimen is too fragmentary. 138 IHSAN A. AL-SHEHBAZ Thelypodium paysonii is somewhat anomalous in the genus in having non-torulose siliques and pubescent filaments and claws. The character- istic short gland-like trichomes on the lower parts of the filaments and claws serve as an excellent character to distinguish this species from others of the genus. The adaptive significance of these trichomes is not known, but they may help in the accumulation of large drops of nectar in the center of the flower. It is doubtful that they have any secretive function. Regardless of the “anomalous” features mentioned above, Thelypodium paysonii fits within the genus, and appears to be related to the other annual species, T. texanum and T. tenue. It resembles these species in having pinnately lobed leaves, spreading floral parts, subconical styles, and an annual habit. Very little is known about the ecology and the overall variability of this interesting species which is unique in a number of features as men- tioned above. This is probably due to the fact that the species is a narrow endemic, and that it has been collected only a very few times in the past century as judged from the collections available. SPECIES EXCLUDED FROM THELYPODIUM OR NAMES OF DOUBTFUL APPLICATION unless it is otherwise indicated 1. Thelypodium ambiguum Wats., Proc. Amer. Acad. 14:290. 1879. =Thelypodi- opsis ambigua ( Wats.) Al-Shehbaz, comb. nov. (see the discussion under the generic relationships in connection with this species as well as the others placed in Thelyodi- oO 2. (Benth.) Rollins, Rhodora 62:16. t 3. Thely ays Rydb., Bull. Torr. Bot. Club 34:432. 1907, 4. Thelypodium auriculatum (Gray) Wats., Proc. Amer. Acad. 17:321. 1882. =Sisymbrium auriculatum Gray, Smithson. Contrib. (PI. Wright.) 3:8. 1852. j 1904. —Romanschulzia australe The types and/or isotypes of the following taxa were seen by the present writer, ed. . Thelypodium bakeri Greene, Plantae Bakerianae 3:8. 1901. =Thelypodiopsis elegans (Jones) Rydb., Bull. Torr. Bot. Club 34:432. 1907, 7. Thelypodium cooperi Wats., Proc. Amer. Acad, 12:246. 1877. =Caulanthus cooperi ( Wats. ) Pays., Ann. Mo. Bot. Gard. 9:293. 1923. 8. Thelypodium crenatum Cr ne, Pittonia 4:20. 1899. —Lepidium montanum — ex Torr. & Gray subsp. spathulatum (Robins.) C. L. Hitche. Madrofio 10:158. 10. Thelypodium diehlii Jones, Contrib, Western Bot. 12:2. 1908. Type collection not seen. This species probably belongs to Sisymbrium, since it was su a = mak ansfi Thelypodiopsis, but it is not advisable to UL. Thelypodis rs Present time because mature fruits are not known. ‘ tum elegans Jones, 4:265. 1893, —T/ Ssieat Rydb., Bull. Torr. Bot. Club 34.492. 1907. helypodiopsis elegans (Jones) BIOSYSTEMATICS OF THELYPODIUM 139 2. Thelypodium speak Muschl. ex O. E. Schulz in Notizbl. Bot. Gart. Berlin 11:390. 1932. Listed as a synonym of Arabis ae Schulz (=Sibara vier- eckii (Schulz) Rollins var. Endlichii (Schulz) Rollins) but it was not validly published. 13. Thelypodium flavescens (Hook.) Jeps., Fl. West. sonoge Calif. 212. 1901. ot. 9:301. 19 There is no doubt that the species is at home in Caulanthus and not the other genera for the reasons mentioned = the discussion of Caulanthus. I have seen the holo- type of Streptanthus dudleyi Eastw. (Proc. Calif. Acad. Sci., ser. 4, 20:145. 1931.) and Streptanthus lilacinus chia (Leaflets West. Bot. 1:226. 1936.), and it is clear that they are additional synonyms of Caulanthus flavescens. 15. Thelypodium flexicaule (Dusén) Gilg & Muschl., Engl. Bot. Jahrb. 42:438. eens i) specks was transferred from Sisymbrium (S. flexicaule Dusén, oe for eek 07). The species is a native of Patagonia. I have not seen th patagonicum (Speg.) Schulz (based on Sisymbrium patagonicum Speg.). Whether the s oe belongs to Sisymbrium or Chilocardamum is beyond the scope of the present 16. eile eae cron Jeps., Fl. West. Middle poe 212. 1901. =Caulanthus flavescens ( Hook.) Pays., Ann. Mo. Bot. Gard. 9:301 3. 17. Thelypodium harmsianum Muschl., Engl. Bot. soe 40:267. 1908 —Po Schulz, psecadium harmsianum (Muschl.) O. E. S a 86, (IV, 105): 177. = he species was based on a collection from em via. I have not seen collection of this species, but I have seen an iso to) rmsianum. var dentata Muschl. at US that was also collected from southern Bo This species is = Pree aaa Thelypodium h divecslaaien var. dentata Muschl., Engl. Bot. Jahrb. 40:268. 1908. Polypsecadium harmsianum var, dentatum (Muschl. ) Schulz, Pflanzenreich 86, av, 105): 177. 1924. —. hookeri Greene, FI. aegere 263. 1891. =Caulanthus flav- oieas ( Hook.) Pays., Ann. Mo. Bot. Gard. 9 hag oe a. 20. Thel ium jaegeri Rollins, Contrib. Dudley Her In his deserptio on of Tr jaegeri, Rollins (1941a) pointed out the differences betwee: this species and thos e belonging to Thelypodium proper. He adopted the Soadlie defined concept of Thelypodium, because of the fact that the generic lines were not satisfactorily established between the various rate : - una League ty seg : a en : b ene ngage ae by Dr Rollins, the lack of basal leaves form- non-torainee ae 1: ‘ hi the lateral sepals in all species of Thelypodium. It is beyond the scope sf - present study to place this species ies in some other genus, g in 140 IHSAN A. AL-SHEHBAZ 21. + agate lasiophyllum (Hook. & Arn.) Greene, Bull. Torr. Bot. Club 13: 142. 188 aulan thus lasiophyllus (Hook. & Arn.) Pays., Ann. Mo. Bot. Gard. "22, Thelypodium lost phattas (Hook. & Arn.) Greene var. inalienum Robins. in Gray, Synop. Fl. N. Amer. 1:177. 1895. =Caulanthus lasiophyllus (Hook. & Arn.) Pays. var. inalienus ( Robins.) Pays., Ann. ie Bot. Gard. 9:306. 1923 23. Thelypodium lasiophyllum (Hook. & Arn.) Greene var. rigidum (Greene ) Robins. in Gray, Synop. FI. Amer. 1:177. 1895. =Caulanthus ee al ( Hook. -) Pays. var. rigidus (Greene) Pays., Ann. Mo. Bot. Gard. 9:307. 24. Thelypodium lasiophylum (Hook. & Arn.) Greene var. utahense (Ryde. ) ep a Man. FI. Pl. Calif. 413. =Caulanthus phe head (Hook. & Arn.) Pays. Pa 25. Thelypodium lasiophyllum ( Hook. & Arn.) G reene f. xerophilum (Fourn.) Thell. Mitt. Bot. Mus. Ziirich 83:735. 1919. —Caulanthus lasiophyllus (Hook. & Arn.) “36. Thelypodium lemmoni Greene, West. Amer. Sci. 3:156. 1887. =Caulanthus anceps Pays., Ann. Mo. Bot. Gard. 9:303. 1923. (See the discussion under Caulan- us. 27. Thelypodium on (Gray) Wats., Aes King’s Exped. 25. 1871. =Thely- podiopsis linearifolia (Gray ) Al-Shehbaz, comb. n 28. Thelypodium bbaioin Bi Brandeg., Univ. of Calif, Publ. ns 4:178. 1911. =Sisym- ~~ auriculatum Gray, Smithson. Contrib. ( Pl. Wright.) 3:8. 1852. - Schulz (1924) recognized lobatum as a variety er E dskihiicins auricu- nellia longifolia (Benth. ) Rollins, Rhodora es 16. 1960 30. ee tanthella ong stris (Wats.) Rydb., Fl. Rocky Mts., 364. 1917. 32. Thelypodium macrorrhizum Muschl. , Engl. Bot. Jahrb. 40:268. 1908. This is another of those species from South Ameri ca that were originally described in “cya a Muschler and later transferred to monotypic genera by Schulz (1924). In this ase, T. macrorrhizum was placed in a genus of its own called P Mceokes | Schulz, which was said to be different from Sisymbriu aving a septum 2-4 nerves instead of being nerveless, 1-nerv , or 2-nerved. It is very doubtful that such a ch acter is si orr, i ium by Macbride (S. macrorrhizum (Muschl.) re Canis 5. 355. 1934). 33. Thelypodium mexicanum Brande eg., Zoe 5:179. =Romanschulzia australe ( Brandeg. ) Rollins, Contrib. Dudley ee 3:225. 1 stg - Thelypodium micranthum (G ay) Wats., Proc. Amer. Acad. 17:321. 1882. on ernie (Gray) Nieuw. iy a Midl. Nat. 5: 294. 1918. lypodium pea sone Jones, Amer. Nat. 17: soil _ —Caulanthus lasio- ph (Ton h Ara yP ays., Ann, Mo. Bot. Gard. 9:303. Debates ssllidsars Rose, Contrib. U.S. Nat. aoa . 1905. —Roman- = sy (DC.) Rollins var. lasiocarpa (Schulz) Rolling, Contrib. Dudley an tum petiolatum Hemsl., Diagn. Pl. Nov. Mex. 2. 1878. —Iodanthus Ds ide (Hemsl.) Rollins, Contrib. Dud Dudley Herb. 3:211. 1949. This species was as placed by Schulz (1934) in «genus ofits own named Chounan thus. ie pst it belongs in us as Iodanthus pinnatifidus was clearly oO (1942a). Schu (1 1936) t aintained Iodanthus and Chaunanthus, but he placed cach of them in » difereo eS me * BIOSYSTEMATICS OF THELYPODIUM 141 38. Thelypodium pinnatifidum Wats., Bot. King’s Bren 25. 1871. =Iodanthus pinnatifidus (Michx.) Steud., Nom. Bot. ed, 2, pt. 1:812. 1 39. Thelypodium procerum (Brewer) Greene, FI. “crate 263. 1891. =Caul- a Hook. ) 23. nthus flavescens ( Pays., Ann. Mo. Bot. Gard 9:301. 40. Thelypodium purpusii Brandeg., Zoe 5:232. 1906. =Sisymbrium purpusii ( Brandeg.) O. E. Schulz, Pflanzenreich 86, (IV, 105): 58. 19. 41. Thelypodium rigidum Greene, Pittonia 1: er wi —Caulanthus lasiophyllus (Hook. & Arn.) Pays. var. rigidus aig tear Pays., Ann. Mo. Bot. Gard. 9:307. 1923. 42. Theiypodun (?) salsugineum ( Pall.) R Robins. i in Gray, Synop. Fl. N. Amer. 1:175. 1895. =Sisymbrium salsugineum Pall., Reise Prov. Russ. Reich. II. App. 740. sca pe t5, 1 Thelypodium stamineum poi cna West. Bot. 2:7. 1937. =Caulanthus panes Wats., Bot. King’s Exped. 27. 44. Thelypodium suffrutescens a is Ann. Carnegie Mus. 26:224. 1937. =Glauco- carpum suffrutescens (Rollins) Rollins, Madrono 4:233. 45. Thelypodium tularense Jones, Contrib. West. Bot. 18:21. 1935. Type not seen, but judging from the description, this “ a may well prove to be a synonym o Caudontns lasiophyllus (Hook. & Arn. ‘. Pays. Thelypodium utahense Rydb., Bull. Torr. Bot. Club 29:233. 1902. =Caul- sates ied gic sang (Hook. & Arn.) Pays. var. utahensis (Rydb.) Pays., Ann. Mo. Bot. Gard. 9:307. 1 47. reeled vaseyi Coult., Contrib. U.S. Nat. Herb. 1:30, 1890. =Sisymbrium shinnersii Johnst., Southwestern Nat. 2:129. 1957. 48. Thelypodium vernale Woot. & Standl., Contrib. U.S. Nat. Herb. 16: Har 1913. Several attempts to locate the type specimen of this species in the U.S Serres Herbarium have failed, but judging from the description and the absence of apiiee aag species of Thelypodium in New Mexi patie! species appears to rad Sisymbrium. In fact, it was transferred to Sisymbrium iy 8 hulz (Pflanzenreich 86, [IV. 105]: 57. she This may prove to be the right treatment if the species is indeed a good on ‘ ore Thebirpodii versicolor Brandeg., Univ. Calif. Publ. Bot. 4:178. 1911. =Sis- mbrium versicolor (Brandeg.) Schulz, Pianaenreich 86, (IV, 105): 57. 1924. LITERATURE CITED Asrams, L, 1944. Illustrated Flora of the Pacific States. vol. 2. Stanford Univ. Press, alifornia. — se si Sic, Flora of New Zealand. vol. 1. Government Printer, Wellington, Ne io R. E. 1967. Biochemical Systematics. Evol. Biol. 1:197—305. he ETTLIN 1949. 1-5-Vin sg rte AsTwoop ae Greer, and M. G. don: Antithyroi id Compound from Yellow Turnip and from Brassica S a Biol —— 181: 121-130. Baton, H. 1874. The Natural History of Plants. 3:179-291. English Translation by . M. Saute L. Reeve, London Baxer, W. H. 1963. Notes on the Flora of Idaho—III. Leafl. West. Bot. 1 Ba.t, P. W. 1964. — m. In Tutin, T. G. et al., eds., Flora ee 1:264-266. Cambridge University Pre: - BaTEeMAN, A. J. 1955a. Seif. incompatibility Systems in Angiosperms. III. Cruciferae. Heredity 9:53-68. ———————,' 1955b. Note on Diocey in the Cruciferae. Heredity 9: Bayer, A. 1905. Beitrage zur systematischen Gliederung der caress, Beih. Bot. _— 18:119-180. : tag D. Hooker. 1862. Genera Plantarum. vol. 1. L. Reeve, London. adit + Grir, T. Matveyeva, O. Za yEvA, 1969. Chromosome a of Flowering Plants. V. L. Komarov Botanical Institute, Acad. Sci. USSR. Gieaaa A. P. De. 182la. — sur Ja Famille des Cruciféres. Mem. Soc. Phys. Hist. Natur. Genéve. Genév 1821b. Systema Neturale II. Paris. 142 JHSAN A. AL-SHEHBAZ ——— 1824. eo een In Prodr. System. Nat. 1:131-236. Ciena. 3 E. U., and L. Jorrer. 1945. Floristic eatey in the cee “of ‘the Colorado and Tributaries. Amer. Mia Nat. 32:591-64 pe seaiee. A. 1968. The Evolution and iis of Flowering Plants. Houghton Mifflin, Boston Davis, P. H. 1964. “Materials ng os dae of Turkey: VIII. Cruciferae I: Isatis. Notes Roy. Bot. Gard. Edinb. 26: aaa . and V. H. Heywoop. ee Principles of Angiosperm Taxonomy. D. Van Nostrand, Princeton, New Jersey. 558 pp. Davis, R. J. 1952. Flora of Idaho. Wm. . Brown, Dubuque, Iowa AXENBICHLER, M. E., C. H. VAN Erren, F. S. Brown, and Q. J ONES. 1964, Oxazoli- dinethiones and Volatile adiocyainice in Sa an treated pone Meals from 65 Species of Cruciferae. Agric. Food Chem. 12: 127-130. AXENBICHLER, _ HL Van Erren, and I. Woxrr. 1965. A New Thio- glucoside, (R)-2-Hydroxy-3-butenylglucosinolate from Crambe abyssinica Seed. iochemistry DetHier, V. G. 1970. 4 Interactions between Plants and Insects. In Son heimer, E., a J. B. Simeone, eds. Chemical Ecology. pp. 83-102. AvgAetio DvorAx _F, 1970a. Study of the Characters of the Genus Malcolmia R. Br. Feddes cna 81:387-416 ———— : Ob. s the Genus Iodanthus Torr. & Gr. Subsume under the Tribe sn ab ay (Austria) 14:113-117. e Evolutionary Relationship in the Family Brassicaceae. 72. Sore : a. Feddes — “" 357. ome ctt ir A Study on the Species Arabis glandulosa Kar. & Kir. Feddes Repert. 82: Heaps ASTWoop, A. 1931. New — of Plants from Western North America. Proc. Calif. Acad. Sci. (ser. 4) 20: Ecuuin, P. 1968. Pollen. sie ee 218(4) :80-90. Exruicu, P. R., and P. H. Raven. 1965. Butterflies and Plants: A Study in Co- evolution. Evolution 18:586-608. Exuiort, M. C., and B. B. Stowe. 1970. A Novel Sulphonated Natural Indole. Phyto- chemistry & 9: 1629-1639. —————~—~. 1971. Distribution and Variation of Indole Glucosinolates in Woad ( Isatis ha L. ) Plant ——— 48:4 — EnouicHer, S. 1839. Genera Plantarum. Vindobon AN, G, 1952. Pollen Shacchalogy and Plant Taxonomy. Angiosperms. Chronica Botanica, Waltham, Mass. 539 pp. EN e douse Method, a Revised Description. Svensk. Bot. Tidskr. 7061 ~~; 1968, Handbook -of Palynology. —— Copenhagen. 486 pp. ——-———~. 1971. Pollen Morphology and Plant Taxonomy. Hafner, New York. and A. KJAEn. a Sulfur Compounds in Plants. In Mabry, T oe E. Alston, and V, C. Runeckles, eds., Recent Advances in Phytochemistry. a 1, pp. 59-144. Appeton-Century Crofts, New York. TTLIN . Lunn: M. G., and A 1956. The Structure of Sinigrin and Sinalbin; an Enzymatic Rearrangement. ee Amer. Chem. Soc. 78:4172-4 2. AD no. sab gic Office of Technical Services, U.S. Department of Commerce, Washington, D.C Ernunces, M. G., A. Ksaen, C. P, Tuo HOMPSON M. Wacnieres. 1966. and M a New Mustard Oil, from Heliophila longifolia DC. pacino s Acta Chem. maak 20: 1778-1782. Farcnt, K., and L. van per Piyx. 1966. Principles of Pollination Ecology. Pergamon FRAENKEL, G. $. 1959. The Raison d’Etre of Secondary Plant Substances. Science 129: 1466-1470. BIOSYSTEMATICS OF THELYPODIUM 143 Guu, R. B. 1971. an Goiter. Sci. Amer. 224(6):93-101. GME rn, R., and A. Kjaer. 1970. ae eee in Matthiola fruticulosa and Related Species: a Ralanieeicatien, Phytochemistry 9:569-573. GMELIN, R., A. KjJArEr, and A. Soneuue 1970. Glucosinolates in Seeds of Sibara virginica (L.) Rollins: Two New Glucosinolates. Acta Chem. Scand, 24: 3031- Goopan, G. J. 1936. Two Western Plants New to Oklahoma. Leaflets West. Bot. 1:208. Gray, A. 1864. On Streptanthus Nutt., and the Plants which have been referred to that Coviie: Proc. Amer. Acad. 6: 182-1 88. —---———— 1865. Charsctess of Some New Plants of California and Nevada. Proc. Amer. Add. Sci. 6 GREENE, E. L. 1886. “tie Californian Polypetalae. toa Torr. Bot. Club 13:142. . 1891. Flora Franciscana. San Francise ————. 1896. Studies in the Cruciferae.—I. 2. re Proposed New Genus, Schoeno- crambe. Pittonia 3:124—128. —-————— 1901. Plantae Bakerianae 3:9. ———————, - 1904. Certain West American Cruciferae. Leaflets Bot. Observ. Crit. 1:81-90. —--———— . 1906a. Four Streptanthoid Genera. Leaflets Bot. Observ. Crit. 1:224-229. ——————— 1906b. Icianthus and Sprengeria. Leaflets Bot. Observ. Crit. 1:197—199. Gupta, P. Dp. and A. J. THorsTEINSON. 1960. Food Plant Relationships of the Diamond- hac k Moth (Plutella maculipennis ( Curt.) ). Entom. Exp. Appl. 3:241-250, 305- 314. AYEK, A. von. 1911. Entwurf —— Cruciferen-System auf Phylogenetischer Grund- lage. Beih. Bot. Cael 27: 127-335. 1 HEGNAUER, saa der Pflanzen. 3:358-365, 586-607. Basel, iia wcoeges eee -———_—_—_———. _ Chemotaxonomie der Pflanzen. 5:264-293. Basel, Birkhaiiser Verlag. Heuuer, A. A. 1898. New Plants from Western North America.—II. Bull. Torr. Bot. Club 25:265. HEMINcway, J. S., H. J. Scnorte.p, and J. G. VAUGHAN, 1961. Volatile Mustard Oils of Brassica juncea Seeds. Nature 192:993. Henry, J. K. 1915. Flora of Southern British Columbia and Vancouver Island. W. J. HERMANN, F. J. 1966. Notes on Western Range Forbs: —— through Compositae. Forest Service, Agricultural Handbook No. 293. U. Hitpesranp, F, 1879. Vergleichende ‘Untersuchungen ‘aes die Saftdriisen der Cruciferen. eo Wissensch. Bot. Hircucock, C. L. 1 864, Thelypodium In ponies C. L. et al. Vascular Plants of the Pacific pote nage venoms: ae N. 1972. Plant Geography of the sal aE Region. In Cronquist, A g Tastestorses Flora 1:77-161. Hafner, ‘ce wne P. K. 1971. A Biosystematic ers fe sree North American Thlaspi montanum and its allies. Mem. N.Y. Bot. Gard. Hooker, W. J. 1830a. Description of a New anes of Macropodium. Bot. Miscell. 1:339-342. _—_. 1830b. Flora Boreali Americana. vol. 1, p. 61. Henry G. Bohn, London. Hoover, R. G. 1966. Miscellaneous New Names for Pusan Plants. Leaflets West. Bot. 10: sea 0. The Vascular Plants of San Luis Obispo County, California. Ganaay “ pelted Press, Berkeley. : Intis, H. H. 1957. Studies in the Cecmucdacone Ill. Evolution and eg ed of the Western North American Cleomoideae. Ann. Mo. Bot. Gard. 44:77-1 —. 1958. Studies in the Simian tv. Polanisia Raf. ame 10: —————-. l tudies in the Capparidaceae—VII. Old World Cleomes Adventive in the New nis Brittonia 12:279-294. 144 IHSAN A. AL-SHEHBAZ JANCHEN, E. 1942. Das System der Cruciferen. Osterr. Bot. Zeitscher. 91:1 —28. Jepson, W. L. 1925. A Manual of the Flowering Plants of California. Associated Students Store, rgcroueals ——————. 1936. A Flora of California. 2:20-42. Associated Students Store, Berkeley. Jounson, T. F. 1970. Investigations in the Floral Biiay of Arabis holboellii Complex. aster’s Thesis, University of Washington, Sea Jones, Q., and F. R. Earre. 1966. Chemical ree el of Seeds II: Oil and Protein Content of 759 Species. Econ. Bot. 20:127-155. Josersson, E. 1967. Distribution of Thioglucosides in Different Parts of Brassica Plants. Phytochemistry 6: 1617-1627. ——. 1971. Studies of the Biochemical Background to Differences in Glucosin late Content in Brassica napus L. II. Administration of Some Sulphur-35 a Carbon-14 Compounds and Localization of Metabolic Blocks. Physiol. Plant. 24:161-175. Kearney, T. H., and R. H. PeEsies. 1942. Flowering Plants and Ferns of Arizona. Government Printing Office, Washington, D.C. . 1960. Arizona Flora, University of California Press, Berkeley Kennen, A. A. 1895. The Natural ae of Plants 2:310-312. English ie ate, by F. W. Oliver, Henry Holt, New York. Kirk, T. 1899. The Student’s Flora - New Zealand and the Outlying Islands. co aime Printer, Wellington, New Zealand. Kyarr, A. 1960. Naturally Derived thiocyanate Loiaonts Oils) and their Parent Glucosides. Fortschr. Chem. O Org. Naturst. 18:122— rin) in rae of Sisymbrium austriacum Jacq. Acta Chem. Scand. 13:1575-1584. —————_, B. W. CurisTENnsEN. 1962. -2i Atcelioa XL. Glucosisaustricin, a Novel Glucose Present in Seeds of Sisymbrium austriacum Jacq. Acta vote Scand. 16:71~77. —, and S. E. Hansen. 1958. Isothiocyanates XXXI. The Distribution of Mustard Oil over in Some Arabis Species. A Chemotaxonomic Approach. Bot. Tidsskrift 54:374-378. ——————~—, and K. “sii atin 1953. Paper Chromatography of Thioureas. Acta Chem. Roda 7: —— and A. Scuustrer, 1970. uaeainage in Erysimum hieracifolium L.; Three Sake scan Glucosinolates. Acta Chem. Scand. 24:1631-1638. and HUST Tit G marordbinasadh in Capparis flexuosa of Jamaican Origin. Phytochemistry 10:3155-3160. —————~, an ScuustER. 1972a. Glucosinolates in Seeds of Arabis hirsuta (L.) Scop.: — New Naturally Derived Is sothiocyanates. Acta Chem. Scand. 26:8—14. . ScuustER. 1972b. Glucosinolates in Seeds of Neslia paniculata. Piytochmisy 11:304 A — Se Ra ya Congeners in Sesds of Erysimum et ae ams 12: "929-933. ———————, and M. Wacnrires, 197]. 3,4,5-Trimetho: oxybenzylglucosinolate: A — of Lepidium sordidum. Phytochemistry 10:2195-2198. MELIN, and R. B. JENSEN. 1956. Iso iocyanates XVI. Glucocon- ringiin, cn Natural Precursor of 5,5 Dimethyl-2-crazolidinethione. Acta Chem Scand. 10:432--438. » B. W. Curistensen, and S. E. Hansen. 1959. Isothiocyanates XXXIV. The Absolute Configuration of (— Hipp ony Schermer pericyre a and its _ oo ( Progoitrin ). Acta em. Scand. 13:1 es 1965. Morphologic estar ners of nat te npn of ag ‘Tormon. Aires A. R. 1957. Variation in a of Fase se between Isolated Populations of the tine be Spsies, Streptanthus glandulosus Hook. Evolution 11:185-211. - 1958. The Taxonomy of ey “sat Complex, Streptanthus glandulosus Madrojio 14:217-297, Suviten M. 1964. Glucobrassicin a Potential Inhibitor of Unusual Type Affecting the BIOSYSTEMATICS OF THELYPODIUM 145 Germination and ipa of Plants; Mechanism of its Action. Biologia Plantarum (Praha) 6(2):88— Lane, A. 1957. The effect of Gibberellin upon Flower Formation. Proc. Nat. Acad. Sci. “— ———————. 1965. Physiology of Flower Initiation. Encyclopedia Plant Physiol. 15(1): tov 1536. Lancer, P. 1966. Antithyroid Action in Rats of apy, Doses of Some Naturally Oc- curring Compounds. Endocrinology ie 117-11 the World, ed. 5, Regnum Vegetable. vol. 3 Lioyp, D. G. 1965. Evolution of Self-com ban and Racial Differentiation in Leavenworthia (Cruciferae). Contrib. Gray Herb. 195:3-134. Macurre, B. 1942. Great Basin Plants—VII. Cruciferae. Amer. Mid]. Natur. 27:468. Martin, P. S., and C. M. Drew. 1969. Scanning Electron Photomicrographs of Southwestern Pollen Grains. J. Ariz. Acad. Sci. 5:147-176. —————~——. 1970. Additional Scanning Electron Photomicrographs of Southwestern Pollen Grains. J. Ariz. Acad. Sci. 6: 140-161. McVaucu, R. and F. R. Poiasel 1941. apse to the Geographical Names of Nevada. —— ntrib. N. 29, U.S.D.A., Washington, D . and P. Lens. 1967. Die Fe ceodks bis beens der Kreuz- bilities und Mohngewiichse. Bot. Jahrb. 86: 113-129. Micuniewicz, M., and A. Lane. 1962. Effects of Nine Different een eea on Stem Elongation and Flower Fo ormation in Cold- -Requiring and Photoperiodic Plants Grown under ans panasio Conditions Planta 58:549-563. MikoLaJczaxk, K. L., T. K. M . EARLE ae Wo rr, and Q. Jones. 1961. Search for New. Industrial O Oils. wv. Oils of Cruciferae. phe Amer. Oil Chemists’ Soc. 38:678-681. Mocc1, G. 1965. Osservazioni Tassonomiche e Corologiche sulle Hesperideae (Cruci- Soran), Webbia 20: 241-273. Mo es J. L. 1941. A Monograph of the Section Euclisia Nutt., of Streptanthus Ph.D. Thesis, University of California, Berkele ron °. A. — A California Flora. Universi ity of f California Press, Berkeley. Murtey, M. R. 1951. — of the Cruciferae of Northeastern North America. Amer. Midl. Natur. 46:1-8 Nayar, J. K., and A. J. i ORSTEINSON. 1963. Further Investigations into the Chemical Basis of Insect-Host Plant Relationships in an Oligophagous orig: ( Plutella IEUWLAND, J. A. 1918. Heterothrix oof Ls. Robins. ) Rydb. a Synonym, and Other Notes. Amer. Mid]. Natur. 5:224-22 Nutrau., T. 1834. A Description of Some of the Rarer or Little Known Plants Indigenous to the United States, from the Dried Specimens in the Herbarium of ie yeas of Natural Sciences in Philadelphia. Jour. Acad. Nat. Sci. Phila. Sietiaty E. i 1922. Species of Sisymbrium Native to America North of Mexico. Univ. Wyom. Publ. Sci. 1(1):1-27. Peer —emenrernee,: LAND: Menger Study Ann. Mo. Bot. Gard. 9 Peck, M. E. 1941. A Manca! adi ie Higher Plants of Oregon. Binfords & Mort, Port- land. 2nd edition, 1961. Pererson, N. F. 1912. Flora of Nebraska. Privately Published, Lincoln, Nebraska. 3rd edition, 1923. PRANTL, K. 1891. —— In Engler, A., and K. Prantl, Die Natiirlichen Pflanzen- familien 3(2):145— Ritey, R. 1956. The 00 uence of the Breeding System on the Genecology of T hlaspi alpestre L. New Phytologist 55:319—330. oo. B. L. 1895. gs In Gray, A. and S. Watson, Synoptical Flora of North America 1(1):98—180. of Thelypodium and its Immediate Allies. 146 IHSAN A. AL-SHEHBAZ Ro..ins, re C. 1938a. Glaucocarpum, A New Genus in the Cruciferae. Madrofo 4:23 —_—-——-. om Smelowskia and Polyctenium. Rhodora 40:294-305. ——-—-——-, = fo The Cruciferous Genus ator t Lloydia 2:109-127. - Some New or Noteworthy N American Crucifers. Contrib. idles at. 3:174-175. ———. 1941b. The Cruciferous Genus Dryopetalon. Contrib. Dudley Herb. 3: 199-207. ———————. 1941c. Monographic Study of Arabis in Western North America. Rhodora 43:289-325, 348-411, 425-485. 1 ———————. 1942a. A Systematic Study of Iodanthus. Contrib. Dudley Herb. 3:209- 215. ————. 1942b. A Tentative Revision of the Genus Romanschulzia. Contrib. ales Herb, 3:217-226. -——————— . 1943. Generic Revisions in the Cruciferae: Halimolobos. Contrib. Dudley Herb. 3: foe —265. 6. Some North American Cruciferae II. Contrib. Dudley Herb. 3: em ———————. 1952. Taxonomy i nd Tomorrow. Rhodor : : ——————. 1956. Some New Primitive Mexican Cenatioshe, Risdars 58: 148-157. ———————. 1957a. Miscellaneous Coiaieecae of Mexico and Western Texas. Rhodora 59:61-71. ———————. 1957b. Taxonomy of the Higher Plants. Amer. Jour. Bot. 44:188-196. ———-———. 1958. The Genetic Evaluation of a Taxonomic Character in Dithyrea stag Rhodora 60: 145-152. —_—-——-— 9. The Genus Synthlipsis (Cruciferae). Rhodora 61:253-264. Some Sisymbriums (Cruciferae) native to Texas and Northeastern 2 aie Rhodor 62:55-60. 1960b. The American Cruciferae of Sessé and Mocino. Rhodora 62: 11-20 1963a. The Evolution and Systematics of Leavenworthia ( Cruciferae ). Coral. fed Herb. 192:3-98. an, AOD, Pctaaey in Two Species of Streptanthus (Cruciferae). Rhodora 65: 45-49. - 1966. Chromosome Numbers of Cruciferae. Contrib. Gray Herb. 197: - 1969. A Remarkable New Crucifer from Mexico. Contrib. Gray Herb. age —— ory ef in the Cruciferae and Notes on Arabis and Caulanthus. Contrib. Gray Herb. 20 1:3-10. ag tanieeae stage uDENBERG, 1971. Chromosome Numbers of Cruciferae II. Contrib. crs Herb. 201:117-133. , and E. SHaw. 1973. The Genus Lesquerella (Cruciferae) in North ana Harvard Univ. Press, Cambridge ss. Rypserc, P. A. 1904. Studies on the Rocky Newaskaie Flora—XI. Bull. Torr. Bot. Club 31:409-410. 1907. Studies on the Rocky Mountain Flora XVIII. Bull. Torr. Bot. Club 34:428-496,_ 7. Flora of the cane Mountains and Adjacent Plains. Privately Published. New York. 2nd edition, 1923. OONHOVEN, L. M. 1968. Chaise acs Bases of Host Plant Selection. Ann. Rev. Entomol. fies 115-136. snare and Food Plant Selection in Some Lepidopterous Larvae. Entomol. eae Appl. 12:555-564. , O. E. 1924. ponciferae-Sisymbrieae. Pflanzenreich 86 (IV, 105) :1-388. ——_—__, 1933. urze Notizen iiber neue Gattungen, Sektionen und Arten der oe dues Bot. Jahrb. 66:91-102. "———————. 1936. Cruciferae. Die Natur. Pflanzenfam, 2nd ed. 17b:227-658. BIOSYSTEMATICS OF THELYPODIUM 147 SNocERuP, S. 1967, Studies in the Aegian Flora. VIII. Erysimum sect. Cheiranthus. . Taxo onomy. Opera Botanica 13:1—70. Snow, R. 1959. Chromosome Numbers of erw eee Plants, with Notes on Some Cases of eee He Interest. Madrofio 15:8 Sotsric, O. 1970. Principles and iit of Plant Biosystematics. Macmillan 226 pp. Stessins, G. L. JR. 1950. Variation and Evolution in Plants. Columbia University Press, rk. 643 pp. Sanna 1951. Neves Selection and the Differentiation of Angiosperm Families. Evolution = 299-3 scien Adaptive Radiation and Trends of Evolution in Higher Plants. Evol. Biol. sa 101-1 Stork, A. 1972. Studies in the Aegean igi XX Biosystematics of the Malcolmia maritima Complex. Opera Botanica 33:1-118. TAKHTAJAN, A. 1969. Flow wering Plants , Origin and Dispersal. English Translation by C. Jeffrey. Smithsonian Institution Press, Washington, D.C. Tanc, C.-S. 1971. Benzyl I sothicyanate of Papaya Fruit. Phytochemistry 10:117-121. THORSTEINSON, A. Chemotactic Responses that Determine Host Specificity in an Oligopingos isa (Plutella maculipennis ( Cutt.) ) pa raaeg Can. J. > T. Kirrect. 1941. A Flora of Arizona and New Mexico. The Catholic University of America Press, Washington, D.C. Torrey, J. G., and A. Gray, 1838. Flora of North America. 1:91-97. Wiley & Putnam, Y TURNER, B. L. 1969. sp ieagens Jonas Recent Developments. Taxon 18:134— UNpERHLLL, E. W., D. F. Kirxianp. 1972. A New Thioglucoside, pales ghicosinolate. ‘Phytochemistry 11: "2085-2088. NDERHILL, nd L. os —— 1966. Biosynthesis of Mustard Oil Glucosides. In eno ie of Aromatic Compounds, Proceedings of the 2nd Meeting of the Federation of European Biochemical rpg . 129-1 UNDERHILL, E. . D. CuisHoiM, and L. R. ER. 1962 . Biosynthesis of Mustard “i Glucosides. 1. Admi cseatint of C14-Labelled Compounds to Horseradish, asturtium, and Watercress. Canad. i VAN japan a Oe i. .E. DAXENBICHLER, and I. A. lates ( Thioglucosides) in Foods and Feeds. fos A Food Chem. 17:483-491. Vasey, G., and J. N. Rose. 1890. Plants from Southern California. Contrib. U.S. Nat. erb, 17 VASIL CHENKO, ‘L. T. 1939. Sisymbrium. In Komaroy, V. L., and N. A. Bush, eds. Flora of the USSR. (English Lcerony bes R. Lavoott, 1970) 8:36 VaucHAN, J. G., and E. I. 3. A Taxonomic Study of B a june Using the Techniques of conn aden Gas-Liquid ra elt ape cae . Ann. Bot. 37: 167-184. Vavenia J he Pad S. Hemincway. 1959. The Utilization of Mustards. Econ. Bot. J. ScHOFIELD. 1963. Contributions to a Haas 1 es . S. Heminecway, and H. tudy of Variation in Brassica juncea Coss. & Czern. Jour. Linn. Soc. ( Bot.) Waniina. G. G. 1642 — Botanices Systematicae 1:123-192. Lipsia AREING, P. F. Endogenous Inhibitors in Germination and Seca Encyclo _ po Physiol. oe 909-924. : 187 5. Geolog. Expl. Fortieth Parallel. V: 14-31. Warson, S. 1 Botany. aaametaas 1380. Botany o of California. 2:479. _ 1882. Contributions to American Botany X —_—_———_. 1889. Contributions to American Botany IX. Proc. Amer. Acad. 25: 127. Saag R. H. 1970. The Biochemical Ecology of Higher Plants. In Sondheimer, and J. B. Simeone, eds., pp. 43-70. Academic Press, New Yor P. P. Feeny. 1971. Allelochemics: Chemical Interaction Between —_————,, an Species. Science 171:757-770. University Press, Cambridge, Mass. XVIII. Proc. Amer. Acad. 17:321. 148 IHSAN A. AL-SHEHBAZ Winter, J. M. 1936. An Analysis of the Srey 3 Plants of Nebraska. Contrib. Bot. Survey Nebr., Cons. Dept., Nebr. Bull. 13 Younes, C, G., and L. R. Werrer. 1967. M ok eae te of the Major oe Isothiocyanates and Oxazolidinethione in Rapeseed. Jour. Amer. Oil Chem Soc. 44:551-554. PURPLE-FLOWERED ARABIS OF THE PACIFIC COAST OF NORTH AMERICA REED C. ROLLINS The genus Arabis (Cruciferae) is well-represented in the flora of western North America. Some species are abundant and widespread while others are very restricted in distribution, often being confined to specialized habitats and/or soil types. In my treatment of the genus in North America west of the hundredth meridian (Rollins, 1941), I recognized 53 species together with a number of infraspecific taxa. Since then, names for several additional taxa that occur in the same area have been published ( Boivin, 1951; Rollins, 1946, 1971). Our present concern is with one of the evolu- tionary “bursts” within Arabis which has resulted in a group of related species found in the hills, mountains and coastal areas of northern Cali- fornia and southern Oregon. The most spectacular feature shared by these species is a comparatively large and conspicuous lavender to purplish flower. This feature, coupled with that of a rosette (usually flattened ) of obovate to broadly oblanceolate leaves from which the fertile stems arise marks these species as being distinctive among the American members of the genus. The known species involved are A. aculeolata, A. blepharo- phylla, A. medonaldiana, A. modesta, and A. oregana. None of these are widespread in occurrence. Presumably the most restricted, A. mcdonaldi- ana, is known from but a single locality. In 1942, Lincoln Constance and I hiked to this site at the summit of “Big” Red Mountain southeast of Bell Springs in northern Mendocino County, California. There we found A. medonaldiana growing abundantly on red serpentinized soil in open areas between such shrubs as Arctostaphylos, Ceanothus prostratus, and Quercus vaccinifolia. A few days before, we had collected A. aculeolata at four different locations in Josephine County, Oregon and in at least two in- stances the site was a serpentine barren. More recent collections of these purple-flowered species show that serpentine soils figure prominantly in the habitat of at least three of the species. Adaptation to serpentinized soils or at least the reduced competition provided by such sites as in the genus Streptanthus (Kruckeberg, 1951, 1969) appears to be a significant factor involved in the speciation of the group. The need for a careful and detailed study of the situation in this respect is clearly indicated. Addi- tional field research in the area would be rewarding. Some further evidence suggesting a highly local occurrence of this type of Arabis which is at the same time associated with outcrops of serpentine rocks comes from added new material. In 1971, Chester A. Ground sent to me for determination specimens collected on a serpentine ridge of the south slope of Preston Peak in northern Siskiyou County, California. Although obviously related to A. aculeolata, these specimens appeared to 149 150 REED C. ROLLINS be a sample of an undescribed taxon but they were without mature fruit. During the field season of 1972, Mr. Ground and Gilbert J. Muth were able to collect a number of specimens of the same taxon in various stages of maturity. From these, it is evident that an undescribed species is indeed represented. In order to provide a quick guide for those wishing to dis- tinguish among the species of this purple-flowered group, the following key is presented. A. Seeds wingless; plants glabrous throughout ........ ....... sane PHgtS A. Seeds winged at least distally; plants pubescent or hirsute (A. per aa te only a few simple trichomes on teeth of basal leave B. Seeds oblong; winged primarily at the distal bee ‘siliques 4-6 cm. long, 1.5-2 wide, — at apex; Napa and Yolo Counties, California northward to ouahell Orego: C. Pants low, less than 1.5 dm. high, glabrous except for a few rae trichomes on basal leaves; petals 2.5-3 mm. wide; usually truncate . 3. A. medonaldia C. Plants taller, 2-5 ye high, hirsute or pubescent at least below; petals 4-8 mm. wide, rounded at a D. Trichomes isthe: forked or three-pronged at apex of long stalk, large with bulbous base; leaves ciliate; stems coarsely hirsute with spreading trichomes, rarely glabrous. . Leaf trichomes simple; basal leaves dentate, narrowly obovate, densely rosulate, 1-2.5 cm. long, 0.5-1 cm. wide; stems several from a closely bra p> — Ce ee ce OO Soe a HE Ds 2. A olata. stems ee a simple or loose web branched caste e poner talk, comparatively small with- at apex; aoa sera or in the outer ni ranges Sak Sonoma ee C sa ee esperar tal tere 1.5-2.5 cm : oe | 1-15 cm. lati, fli caulinis pn Sita non auric riots th an spathulatis —— 5 12-15 mm. longi gis, siliquis divarica i immargina dy mm. tis .5 mm. suey ca. 1.5 mm. latis, cotyledonibus accumbentibus. on a steep and rocky, west-facing, serpentine slope, upper end of Rattlesnake Meadows, Preston Peak, 41°49.8’ N. jataae 123°38’ W. longitude, Siskiyou County, California, June 26, 1972, ete & round and Gilbert J. Muth 1 op aed stems simple, one from each single rosette or a few from several neste on a br 5-3 dm. hi ; basal leaves in a iolate, broadly obovate, rounded distally, entire to remotel and shallowly dentate, thicki 2.5 cm. long, 1-1.5 cm ; cauline leaves 2- 5, sessile, not auriculate, remote, 3-6 m ong, e; inflorescences race- mose, elongating in fruit; sepals urple, erect, oblong, 4.5-5.5 mm. long, outer pair saccate, inner pair plain; petals spatulate, not sisigherngge into blade and claw, light purple to lavender, econ . long, mm. wide; stamens erect, strongly ; filaments atk ot chove: guthess ce. “ mm. long; fruiting pedicels divaricate, 4-8 mm. mm. long, ihe in length from base to apex of infructescence; PURPLE-FLOWERED ARABIS 151 siliques divaricate to divaricately ascending, strongly single-nerved from base to apex, . , 15+ ide, sessile or with a very short gynophore; styles 0.5 mm. to obsolete; seeds in a single row, wingless, oblong, brown, 2-2.5 mm. long, ca. 1.5 mm. wide; cotyledons accumbent. Additional specimens studied, all from the same general locality as the holotype. Dry serpentine ridge, 5,800 ft., July 31, 1971, Ground 888 (cu); same population as holotype, Aug. 2, 1972, Ground & Muth 1794 (cx); 6,000 ft. Aug. 2, 1972, Ground & Muth 1795 (cH). The significant differences between Arabis serpentinicola and A. acule- olata involve the seeds, the presence or absence of trichomes and the length or absence of a style on the fruit. Arabis serpentinicola is wholly glabrous while A. aculeolata is always hirsute. The portion of the seeds occupied by the embryo is somewhat similar in the two species but there is a pronounced distal wing and minor wing-margins on the sides of the seeds of A. aculeolata whereas the seeds of A. serpentinicola are wingless. In some siliques of the latter species, there is a very short rather stout style about 0.5 mm. or less long but in others the stigma is virtually sessile. On the other hand, the siliques of A. aculeolata have a slender style 1.5-2 mm. long. Another difference that appears to have some value for dis- tinguishing these species is the shape of the silique apex, being accuminate in A. aculeolata and blunt in A. serpentinicola. There is no doubt about the close relationship of these two species which are similar in habit, the nature of the basal and cauline leaves and the flowers. On the whole, Arabis aculeolata has longer siliques and pedicels than those of A. serpentinicola, and the siliques tend to be more erect. The flowering and fruiting period does not coincide in the two species. Arabis aculeolata begins flowering in March in some years but is usually at its flowering peak in April or May. In June and July, A. serpentinicola reaches its lowering maximum. Part of this difference may be associated with the differences in altitude where the two species grow. Arabis aculeolata is found mostly below 2,000 ft. while A. serpentinicola occurs above 5,000 ft. 9. Arabis aculeolata Greene Although we have seen in excess of twenty different collections of this species, it is not certainly known outside of Josephine County, Oregon. There is a single collection at the California Academy of Sciences made by Alice Eastwood in extreme northern Del Norte County, California, and referred by her to A. medonaldiana, which I have tentatively placed in A. aculeolata. The material is in flower only and is hard to deal with on that account. However, it clearly is not A. medonaldiana. The specimens are small for A. aculeolata and the trichomes are less ascicular than in typical specimens but otherwise they seem to fall within the range of varia- tion of that species. Most of the specimens available for stu plants are most conspicuous at that stage an dy are in flower because the d collectors are attracted to 152 REED C. ROLLINS them. Of the six collections with reasonably mature siliques, all but two have sessile siliques. In one collection, Constance and Rollins 2967 (ps,cH), some plants have shortly stipitate siliques while in others the siliques are sessile. In one collection, Eastwood and Howell 1424 (cas), the siliques are uniformly stipitate and the stipe is ca. 2 mm. long. Such a character, although sometimes variable within a species, usually does not range as widely as in this case. Flower size is variable as indicated by the specimens examined, also considerable variation in the height of the plants is evident. Looking at the species as a whole, there is an uncommon amount of varia- tion present even though the total geographic range is relatively limited. This may be explained by the probability that the serpentine areas where A. aculeolata grows are somewhat separated and the individual popula- tions have evolved in slightly different directions. From such evidence as we can see and measure from the specimens alone, this species would appear to be a good subject for the detailed study of population variation and population disparity within a single species. 3. Arabis mcdonaldiana Eastwood There are only three collections of this species known to me, two more than I had to work with in 1940. All three are from the same place, summit of (Big) Red Mountain, northern Mendocino County, California. Both the type series, Eastwood in 1902 (cas,cu), and Constance and Rollins , 3002 (ps,cH) are mostly in flower with, at most, immature siliques. Fortu- nately a newer collection, Gankin, Hildreth, N. and I. Knight 2722 (cas), has mature fruits. Thus, for the first time it has been possible to determine that the siliques are accuminate at apex and tipped with a style about 1 mm. long. The seeds are oblong, about 2 mm. long, 1-1.2 mm. wide, with a definite distal wing and narrow or nearly obsolete wing-margins along the sides. A distinctive feature of A. mcdonaldiana is the short few-fruited infructescence. Only 2 to 4 siliques mature near the apex of the slender stems. 4. Arabis oregana Rollins Although there are a number of newer collections available since my treatment of this species in 1941, they do not provide much additional information. However, it is clear that Arabis oregana is not a serpentine inhabiting species. At two stations about three miles apart where Dr. Constance and I found it growing with Holodiscus, Cercocarpus and Ceanothus, south of Ashland, Oregon, the plants were in a rich deep soil and there was no evidence of serpentine rocks in the vicinity. Most of the known collections are from Jackson County, Oregon, but we found the species near Scott River, 10 miles southwest of Scott Bar, Siskiyou County, California. The specimens of this collection are unusual in being nearly glabrous but are otherwise similar in most respects to the Oregon PURPLE-FLOWERED ARABIS 153 material. The California collection, Constance and Rollins 2915 (ps,cu), has the characteristic trichomes of A. oregana on the margins of the basal leaves which along with the general features of the plants point to their identity with that species. 5. Arabis modesta Rollins At the time Arabis modesta was described, mature fruiting material was not available. Now with ten new collections for study, we still have only two with mature fruits and seeds. In one of these, Ripley and Barneby 9622 (cas), from Rogue River Canyon, 5 miles above Galice, Josephine County, Oregon, both pedicels and siliques are divaricately ascending and the siliques are accuminate at the apex. The style is slender, 1.5-2 mm. long. The seeds are oblong, ca. 2 mm. long, slightly wider than 1 mm., and are very dark, the distal wing being nearly black. The other collections, Harris 21397 (cH), from near the mouth of Scott River, Siskiyou County, California, is similar in most respects but the styles are less than 0.5 mm. long and the siliques are straighter than in the Oregon specimens. In some of the new collections, the specimens are up to 6 dm. tall which is about 1.5 dm. more than we had seen previously. However, the most notable collections are those of Donald V. Hemphill (cas,cu) from near Monticello Dam, Napa-Yolo county line, California. These specimens not only represent a considerable southward extension of the species range but they also show minor differences when compared to those of the California Siskiyou region or of southern Oregon. Arabis modesta overlaps the range of A. oregana in northern California and southern Oregon and there is one collection, Applegate 4598 (ps), from about 10 miles southeast of Ashland, Oregon, that suggests the pos- sibility of hybridization between these species. The specimens are nearest to A. modesta but the trichomes are slightly larger than in unadulterated plants of the species, and the lower stems and basal portions of the petioles have simple trichomes present to a limited degree. The trends toward larger trichomes and toward simple trichomes are in the direction of A. oregana. 6. Arabis blepharophylla Hooker & Arnott There is considerable variation in the coarseness of the pubescence on different plants of Arabis blepharophylla both within populations and between populations. In the most extreme forms in one direction, repre- sented by the collections from Bodega Bay in Sonoma County, California, for example, the pubescence is quite uniform on leaves and stems. The trichomes are small, dendritically branched with mostly four prongs, and the leaves show no suggestion of being ciliate-margined. In another direc- tion, exemplified by the type of the species and other specimens from the Presidio area of San Francisco, the trichomes are much coarser and usually 154 REED C. ROLLINS three or two pronged but often with unbranched ones mixed with them. Furthermore, the leaves are most often ciliate and the lower portions of the stems usually possess spreading trichomes. Other types of variations include the length of the siliques and whether they are blunt or somewhat tapered at apex, the length and thickness of the styles and the height of the plants. However, in spite of the rather considerable evident variation, this species hangs together quite well and I cannot distinguish any infraspecific taxa. Characters that appear to be distinctive turn out to be uncorrelated with other characters that might otherwise be utilized to set off distinctive taxa. LITERATURE CITED Borvin, BERNARD. 1951. Centurie des plantes canadiennes—II. Canad. Field-Naturalist 65: 1-22 cogeorery A. R. 1951. Intraspecific variability in the response of certain native plant species to serpentine soil. Amer. Journ. Bot. 38:408—41 ———————. 1961. Soil diversity and the distribution of plants, with examples from western North America. Madrofio 20:129-154. Rous, R. C. 1941. A monographic study of Arabis in western North America. Bhodox a 43:289-325; 348-411; 425-48]. pom Contributions nea the Gray ean of Harvard University, no. 138. —. 1946. Some new or noteworthy North Ameri rican beeen II. Con- A RECONSIDERATION OF THELYPODIUM JAEGERI REED C. ROLLINS The need for a reevaluation of the generic lines in the Thelypodium complex of the Cruciferae has been recognized for some time (Rollins 1941, 1957). In the last several years one of my graduate students, Dr. Ihsan Al-Shehbaz, has been studying this problem as part of his thesis research (1973). From a very exhaustive assembling of evidence and the careful evaluation of it, he has concluded that Thelypodium should be interpreted as being less inclusive than it is given in some of the current floras. Although it differs in many respects, his treatment of Thelypodium comes fairly close to that of Payson (1923) as far as its general interpreta- tion of generic limits is concerned. My own assessment of the species to be included in Thelypodium coincides in a general way with the findings of Payson and of Al-Shehbaz. Under such an interpretation, one of the species I described in Thelypodium, T. jaegeri, has to be excluded. This ultimate exclusion of T. jaegeri from Thelypodium was actually anticipated at the time the species was describe As a general principle, the Glasstficatiod of a group of organisms is best portrayed if related species are properly aggregated into genera even if some species are more remote in their relationship than others to the biologically central core of species of the genus. Monotypic genera should be avoided whenever possible because they leave the included species without a formally indicated close relationship position within a taxonomic framework. However, there are cases where species simply do not fit into established genera and it is not possible to say what their closest rela- tives might be. The position of Thelypodium jaegeri is a case in point. This species does not fit with those of an established genus of the Cruciferae. For this reason, we have concluded that an independent genus must be established to receive it. Fortunately, there is now enough material to provide the basis for a reasonably adequate study of T. jaegeri. This was not the case when the species was first described in 1941. Caulostramina Gen. Nov. Herba glabra caespitosa multicaulis, caulibus gytosis simplicibus vel rare superne ramosis, foliis petiolatis, inflorescentiis racemosis, sepalis purpureis oblongis non Saccatis, petalis purpureis spathulatis, sefrmeteacee ies laxis, siliquis linearibus; semini Type species: Foe i a Gabel on The helypodium jaegeri a de generic name is suggested by the fact that the older persistent stems become colored. Caulostramina jaegeri ( Rollins ) comb. nov. ms udley Herb. 3:174. 1941. Holotype: shaded ee ee ace ee mao Inyo Mts., about 25 st (es north of 155 156 REED C. ROLLINS Canyon, Sept. 19, 1954, P. A. Munz and J. C. Roos 20175 (CAs,RSA); same location, June 29, 1954, J. & L. Roos 6185 (rsa,uc); Marble Canyon, July 11, 1941, Annie M. Alexander and Louise Kellogg 2528 (cas,ps,GH,POM,UC,US,UTC,WSC,WTU); Teufel Canyon May 20, 1938, E. C. Jaeger s.n. (cH,Pom); Teufel Canyon, about 17 miles north of Darwin, May 31, 1940, E. C. Jaeger s.n. (DS,GH,NY,RSA); just below summit of Cerro Gordo Peak, June 27, 1942, Alexander and Kellogg 3033 (CAS,COLO,DS,GH,NY,RSA,RM,UC, USs,UTC,wsc,wTu); north slope of Cerro Gordo Peak, east of divide, July % 1942, WTv). ? Alexander and Kellogg 3033a (cas,Ds,F,GH,MO,NA,NY,POM,PM,UC,UTC,WSC, There is some variation in leaf shape and leaf margin in Caulostramina jaegeri but all of the leaves are basically of the same pattern. Young seed- lings grown at the Rancho Santa Ana Botanic Garden show even the earliest leaves to be long petioled. The petiole is relatively slender, broad- ening gradually only slightly then flaring abruptly into the broad nearly orbicular blade. Internodes develop between the early formed leaves so that a basal cluster or rosette of leaves with shortened internodes is not formed as in many genera of the Cruciferae. In this respect, the plants are somewhat like the genus Stanleya or the genus Chlorocrambe. Mature leaves occurring at relatively long intervals on the active stems are slender petioled. The blades vary from broader than long, when they tend to be slightly cordate, to longer than broad when they are broadly cuneate toward the petiole junction. The blade margin may be entire to slightly undulate or with broad and blunt teeth. The leaves are bluish green which is characteristic of many arid area crucifers. An outstanding feature of Caulostramina jaegeri is the caespitose habit with the numerous stiffish, almost wiry stems intermeshed in a thick cluster on the upper portion of a heavy root-stalk. The stems are mostly un- branched and are gyrate at least in the lower portion. They are very persistent, the lower portions forming a dense covering over the elongated root-stalk. The older stems become straw-colored and are eventually broken off leaving the lower portion as a part of a thick thatch that makes up the basal part of the plant. Some collectors describe the petals of Caulostramina jaegeri as being white, drying lavender, while others say “flowers lavender.” Mr. E. C Jaeger, the discoverer of the species noted, “Flower color and veining as RECONSIDERATION OF THELYPODIUM JAEGERI 157 Marble Canyon, kindly provided by Mary DeDecker of Independence, California, shows the siliques diverging from the rachis of the infruc- tescence at a very wide angle. These fruits are approaching maturity but instead of being arcuate, they are irregularly curved or bent back and forth in an undulating fashion. Caulostramina jaegeri grows in limestone crevices, usually on north- facing slopes or cliffs. Its habitat is assumed to be an ancient one from the point of view of potential plant habitation because it is an area where Bristle Cone Pine, one of the oldest, if not the oldest, known living plant species occurs. LITERATURE CITED AL-SHEHBAZ, IHsAN. 1973. The Biosystematics of the Genus Thelypodium (Cruciferae). Contrib. Gray Herb. no. 204. pp. oe BE, B. 1023. A apes oe Study ol Thelypodium and its Immediate Allies. n. Mo. Bot. Gard. 9:2 Rows, R. C. 1941. Some tte or Noteworthy North American Crucifers. Contrib. Dudley Herb, 3:174— ———————. 1957. Miscellaneous Cruciferae of Mexico and Western Texas. Rhodora 59:61-71. “fe eo eo Aer 73 Contributions from the GRAY HERBARIUM 1974 NO. 205 yy jy Vine SYSTEMATICS AND/EWOLUTION OF THE eee Roden GENUS CAKILE (CRGCIFERAE) . \CE \OF CRUCIFERAE El REVISIONS OF SOME \GENERA |O eaten A ae NATIVE TO AUSTRALIA \) hi | REVISION OF THE G = Jorge V. Crisci (COMPOSITAE: MU # EDITED By — Reed C. Rollins yy a Kathryn Roby Missourr BoTANIERE APRO- 10m an og LipRARY : THE GR CRAY, HERBARIUM | OF HARVARD UNIVERSITY ee se Cae oh a Contributions from the GRAY HERBARIUM 1974 NO. 205 SYSTEMATICS AND EVOLUTION OF THE James E. Rodman GENUS CAKILE (CRUCIFERAE) Pic hah & Sh REVISIONS OF SOME GENERA OF CRUCIFERAE Pabhevied NATIVE TO AUSTRALIA REVISION OF THE GENUS MOSCHARIA Jorge V. Crisci (COMPOSITAE: MUTISIEAE) AND A REINTERPRETATION OF ITS INFLORESCENCE EDITED BY Reed C. Rollins Kathryn Roby PUBLISHED BY THE GRAY HERBARIUM OF HARVARD UNIVERSITY IssueD Apri 1, 1974 SYSTEMATICS AND EVOLUTION OF THE GENUS CAKILE (CRUCIFERAE) JAMEs E. RopMaAn! INTRODUCTION On November 14th, 1963, a submarine volcano erupted in the North Atlantic 20 miles south of Iceland and eventually created the new island of Surtsey. Access to the island was restricted by the Icelandic government in the interests of scientifically monitoring its developing biota. In June, 1965 the first vascular plants growing on Surtsey were discovered—seed- lings of the sea rocket, Cakile. In the summer of 1967 plants of Cakile were found in flower and fruit, and along with lyme grass (Elymus arenarius ) and angelica ( Angelica archangelica) these initiated the primary coloniza- tion of Surtsey by the higher plants (Thorarinsson 1967). The Surtsey event epitomizes two of the salient characteristics of the biology of sea rockets: dispersal and colonization. It was certainly not an isolated incident although it is rare to find such a well-documented one. In the western Caribbean, for example, Stoddart (1963, 1969) described the colonization by Cakile lanceolata of sand cays off Honduras following their devastation by Hurricane Hattie. Several authors (Davis 1942; Heslop-Harrison 1954; Hewett 1970) have noted the fluctuating appear- ance and disappearance of sea rockets locally along a coast. The dynamic nature of dispersal in the genus is even more evident in accounts of its accidental dispersals by man, resulting in tremendous extensions of the distributional range of some species. Cakile edentula and C. maritima, for example, have been introduced to, and become naturalized throughout, the Pacific Coast of North America (Barbour & Rodman 1970) and temperate coastal Australia (Sauer 1965; Sims 1968). The Old World C. maritima has been collected on ballast ground near ports on the Atlantic and Gulf coast of North America, in Argentina and Uruguay, and even in New Caledonia. The result of these migrations, historically recent and geologically old, has been the creation of a confusing array of sympatric forms throughout much of the range of the genus. Particularly in the New World, two or more taxa of Cakile are now known from every major area of its distribu- tion. Combined with a lack of systematic investigations based on field studies, this situation has led to confusion in the taxonomy of the sea rockets which recent work has not dispelled. In turn, this has limited the comprehension of evolutionary patterns and relationships in the genus and led occasionally to facile conclusions concerning general biological phenomena. A recent monograph of the genus (Pobedimova 1963, 1964) appeared 1 Present address: Yale University, Osborn Memorial Laboratories, New Haven, Conn. 06520. 3 4 JAMES E. RODMAN concurrently with a treatment of the European sea rockets resulting from work on the Flora Europaea project (Ball 1964a,b); they offer quite dif- ferent interpretations of the taxonomy of Old World Cakile. Moreover, a number of authors have expressed dissatisfaction with the treatment of the American taxa (Rollins 1966; Taylor & Mulligan 1968). Omission of validly published names, incorrect citation of types, failure to designate lectotypes, among other shortcomings, apply to the latest treatments. This study was undertaken to provide a more critical taxonomy of Cakile as a basis for the investigation of those aspects of its biology of more general interest such as dispersal, hybridization, and speciation. Data included here derive from more than 100 field collections and from approximately 3700 herbarium collections assembled from 41 institutions. In addition to morphological studies, particularly on the flowers, fruits and seeds, chemical analyses of the seeds for mustard oil glucosides ( gluco- sinolates) by paper and gas chromatography were conducted on ap- proximately 100 field or greenhouse collections as well as seed samples sent from Australia, Iceland, Norway, and Denmark. This chemosys- tematic investigation, and that of Dr. Ihsan Al-Shehbaz done on the genus Thelypodium, constitutes the first extensive survey of these second- ary plant products for monographic purposes. Forty-seven collections, representing most of the taxa in the genus, were grown in environmental chambers to assess genotypic differences under uniform conditions. In conjunction with this, observations were made on seed germination, seed- ling growth, and photoperiodism. Aspects of the breeding systems of Cakile were studied: self-compatibility, floral phenology, and spontaneous fruit set. A limited number of artificial hybridizations were made, and the resulting F, plants were grown to maturity and tested for pollen viability and fertility. An interest in the dispersal ecology of these plants led to experiments on the duration of time the fruits can float in sea water and on the effects on seed germination. Preliminary counts of chromo- somes and a review of the literature suggested that cytological analyses would not be of taxonomic value within the genus; consequently, no extensive chromosome study was made. The combined data from a study of the morphology, glucosinolate chemistry, breeding systems, and genecology of Cakile have been utilized to construct a classification of the taxa and a probable phylogeny. Then, an interpretation of the geographical distribution of the genus through geological time is offered. While this classification purports to be phenetic, acquaintance with modern evolutionary theory has influenced my inter- pretations of the observed patterns. A rigorous numerical analysis of the data would be valuable but was not undertaken. ACKNOWLEDGEMENTS I am grateful to Prof. Reed Rollins, who su i i i ; ggested this study, for his advice and encouragement through the years. Profs. Martin Ettlinger, Otto Solbrig, Elso Barghoorn THE GENUS CAKILE 5 and Carroll Wood, Jr. were especially helpful, materially and in many other ways, as were Dr. Ihsan Al-Shehbaz, Mr. Richard Leo, Mrs. Lily Riidenberg, and TAXONOMIC History The history of the taxonomy of Cakile may be divided into three over- lapping phases corresponding to three major intellectual concerns of the times. The first period, rooted in the pre-Linnean works of Tournefort and the early herbalists, spanned the years from Linnaeus’ “Species Plantarum” to the great “Prodromus” of De Candolle and witnessed the development of a generic concept or circumscription regarded as “natural,” that is, based on several characters of different parts of the plant. The principal features of this period of botanical ferment—the increasing pre- cision in terminology accompanied by a quest for new characters, the standardization of nomenclatural procedures, the slow progress of empiri- cal, inductive classification over essentialist systematizing (Stafleu 1971)— can be illustrated by the systematic treatment of Cakile at that time. From an array of names and descriptions published under the genera Cakile, Bunias, I satis, Rapistrum, and Raphanus there emerged a generic concept based principally on the indehiscent, two-jointed, corky fruit with ac- cumbent seeds, the fleshy, glabrous leaves and stems, and a vague associa- tion with a particular habitat, the maritime strand. This concept was in a sense codified in the “Systema Naturale” and “Prodromus” of A. P. De Candolle, and the subsequent taxonomic history of the sea rockets remains remarkably free of the generic vicissitudes which have plagued so many groups of Cruciferae. The second phase, from the time of De Candolle long into the present century, reflects considerable taxonomic activity of a typological nature. The methodology was the same as that of Linnaeus and many of his contemporaries, but it was applied to the recognition of Species and especially of infraspecific taxa, well over 50 names in infra- specific categories being published. Characteristically, taxa were described on the basis of differences in fruit or leaf morphology uncorrelated with other characters and with little or no attention directed to the nature and extent of variation in populations (cf. Schulz 1903 especially). Indeed, it is a concern with population variation and character correlations as well as with such larger questions as evolutionary origins and phylogeny which permeates, to widely varying degrees, the modern work of the third phase, found principally in the publications of Ball (1964a,b) and Pobedimova (1963, 1964). 6 JAMES E, RODMAN The nomenclatural history of Cakile begins in 1753 with the “Species Plantarum” in which Linnaeus classified Old and New World sea rockets (having seen specimens from both hemispheres) as the species Bunias Cakile. The plants bear only a superficial resemblance to Bunias, and Linnaeus’ treatment reveals his limited knowledge of the indehiscent- fruited, nucamentaceous Cruciferae. A distinct generic concept antedated Linnaeus’ rather novel idea; he acknowledged this by, using Tournefort's generic name, Cakile, as the epithet in his new binomial.” His concept of the plants was apparently not well-defined, however, since he also de- scribed a specimen of Cakile maritima under the name Isatis aegyptica in “Species Plantarum.” The diagnosis is inadequate for resolving the issue, and unfortunately no specimens exist in the Linnean Herbarium, a situa- tion early noted by Willdenow (1800; cf. Savage 1945). The association of this name with the genus Cakile was, however, made by Linnaeus’ con- temporaries and immediate followers and by a number of later taxon- omists. In the herbarium of Linnaeus’ student Pehr Forskal in Copenhagen are sheets of C. maritima (seen on microfiche) marked “Isatis aegyptiaca F.”; these presumably constitute the basis for the record of Isatis aegyptia (note the variant orthography) in Forskal’s “Flora Aegyptiaco-arabica.” Forskal at least was more consistent than his teacher in his classification of sea rockets; specimens with pinnatifid leaves were described by him, not in Bunias, but as Isatis pinnata. In accord with the rules of the International Code of Botanical Nomen- clature, the genus Cakile was first described by Philip Miller in the 1754 edition of “The Gardeners Dictionary . . . Abridged” in which he did not employ Linnean binomials (Dandy 1967). Scopoli, in the 1772 edition of his “Flora Carniolica,” where he adopted the Linnean system of bi- nomials, provided a new name, Cakile maritima, for Bunias Cakile L., to avoid coining a tautonym. Cakile maritima Scop. thus constitutes the type species of the genus. Prior to Scopoli’s name, Crantz had published the combination Rapistrum cakile (L.), but this generic association was accepted only by Bergeret who then effected the superfluous combination Rapistrum maritimum (Scop.). The lack of consensus on a generic cir- cumscription is further illustrated in Willdenow’s “Species Plantarum” where he published Cakile aegyptiaca and Raphanus lanceolatus, the latter described from a specimen collected in “Antillis.” The identity of this species as a Cakile was not established until 1903, and for nearly a century the sea rockets of the West Indies were known by the name C. aequalis. In 1814, in his “Florula Bostoniensis,” the Boston physician Jacob Bigelow described Bunias edentula from plants collected on the sandy * The name Cakile is a Latinization of an Arabic term [transliterated “qaqulleh” (Black 1924) or whose popular manuscript on ee a er? SS. cogen Shani begagteti 5 simp: ? simy translated in the 13th century and was first printed in 1473 (Stillwell 1970). Joseph Gaertner commemorated him the name Cakile serapionis. THE GENUS CAKILE 7 beaches of Cape Ann and South Boston. Four years later Thomas Nuttall, more aware of the progress in European taxonomy but apparently un- acquainted with that in his adopted country, described the same species as Cakile americana. Linnaeus’ ideas must still have been fresh, however, for in so doing, Nuttall (1818) wrote: “in fruit this species approaches Bunias, and seems to evince the property of again uniting these 2 genera.” The confusion of generic concepts ended with A. P. De Candolle. In his treatment of Cakile in the “Systema Naturale” and the “Prodromus” he synthesized the results of several developing lines of research on the fruits and seeds of Cruciferae and assembled a suite of reproductive and vegeta- tive features characterizing plants associated with a particular habitat, the coastal strand. This habitat was virtually a generic character until 1911 when C. arabica from the Nefud Desert was described by Velenovsky and Bornmiiller. The taxonomic work on sea rockets subsequent to De Candolle’s synthesis has accepted implicitly the “natural” character of his circumscription of the genus. For the remainder of the 19th century taxonomic work on Cakile con- sisted largely of the description of infraspecific taxa of C. maritima by both Old and New World systematists. One noteworthy study of intra- specific variation in this species, though ultimately of little taxonomic value, was made by Jordan (1864) in his pioneering genecological in- vestigations. In 1900 the American taxonomist C. F. Millspaugh, following his collecting trip through the West Indies, monographed the genus. Millspaugh’s recognition of eight species in the New World reversed a long tradition maintained by 19th century North American botanists who had treated the American sea rockets as varieties of C. maritima. Un- doubtedly, Millspaugh’s field experience strongly influenced his decision to emphasize at species rank the differences between the American plants. Three years later a second monograph of Cakile appeared in Urban’s “Symbolae Antillanae,” prepared by O. E. Schulz in Berlin. Schulz ac- commodated the eight species recognized by Millspaugh along with sev- eral newly described taxa within an extensive hierarchy of infraspecific categories of a single American species of sea rocket, C. lanceolata. His curious treatment provoked from N. L. Britton ( 1906) the comment: “.. . the attempt of Mr. Schultz [sic] to classify the plants of this genus into named forms and varieties of various ranks serves no useful purpose whatever, and does not express their real relationships at all; the only advance that he has made in their study is to point out an older name for the species long known as C. aequalis L’Hér.” In 1923 Schulz pub- lished a treatment of Cakile for “Das Pflanzenreich” in which he recog- nized two American species, C. edentula and C. lanceolata, within his “species collectiva,” C. lanceolata. Basically this treatment duplicated his earlier one with only changes in rank being made. The judgment of his peers remained much the same; Ball (1964a), for example, remarked 8 JAMES E. RODMAN that “Schulz’s account of the variation within C. maritima and C. eden- tula was not very satisfactory and has not been generally accepted.” Schulz’s work epitomizes a taxonomy almost completely divorced from considerations of the biological meaning of character variation. In method- ology it reduces itself virtually to the nomenclatural recognition of indi- vidual herbarium specimens. The third and latest monograph of the genus, in which 15 species are recognized, was prepared by E. G. Pobedimova (1963, 1964) in Lenin- grad. It appeared concurrently with the treatment of the European sea rockets by P. W. Ball (1964a,b) for the Flora Europaea project. The two offer rather different interpretations of the variation in European Cakile although both share a common philosophical approach to taxonomy, viz. that taxa are recognized on the basis of character correlations and possess a particular geographic range. Their differences stem in part from dif- ferences in the herbarium collections utilized by each, but largely from a difference of opinion concerning the rank at which taxa should be treated, with Pobedimova adopting a very narrow view of species. In her treatment of the New World sea rockets Pobedimova virtually dupli- cates that of Millspaugh in his 1900 monograph; this approach may have been conditioned by the fact that she studied specimens from only three American herbaria (Dao, usc, us). Consequently she contributes little to an understanding of the bewildering array of forms on the shores of the Caribbean and Gulf of Mexico. In addition to the classification, Pobedi- mova offered a phylogeny of the species, admittedly speculative, along with an extensive discussion of the geological setting for the evolution of the species of Cakile. Dissatisfaction with this latest treatment has been expressed by a number of workers (Rollins 1966; Taylor & Mulligan 1968) and seemed to invite yet another attempt to elucidate relationships in this genus. GENERIC AFFINITIES The generic affinities of Cakile, as suggested in the several systems of Cruciferae proposed since the time of De Candolle, are presented in Table 1. In general, Cakile has been associated with genera of the tribe Brassiceae sensu Schulz (1936) on the basis of fruit and nectary morphol- ogy. Only Prantl’s (1891) system deviates significantly from this pattern, and his curious juxtaposition of such disparate genera as Sisymbrium and Cakile has understandably attracted few supporters. The alliance of Cakile with genera of the Brassiceae is clear: its position within the tribe, whether a basal, primitive stock, intermediate, or an advanced end-point of evolu- tion, is much less clear. Pobedimova (1963) considered Cakile to be derived from Erucaria by reduction in the number of seeds per fruit segment, loss of pubescence (she was unaware of the pubescence in C. arabica), and change to ac- cumbent seeds. The inland “sea rocket,” THE GENUS CAKILE 9 C. arabica, appears, in habitat and pubescence, to be a link between the strand species of Cakile and Erucaria, a genus of desert annuals of the Mediterranean region. In what TABLE 1. GENERA ASSOCIATED WITH CAKILE IN THE SYSTEMS OF CRUCIFERAE Author Date ‘Tribe Genera De Candolle 1821 Cakilineae Chorispora Rapistrum Cordylocarpus De Candolle 1824 Cakilineae Chorispora Cordylocarpus Bentham & Hooker 1862 Cakilineae Crambe Hemicrambe Enarthrocarpus Morisia Erucaria Muricaria Fortuynia Physorhynchus Guiraoa Rapistrum Wettstein 1889 Cakilineae Chorispora see air Cordylocarpus __ Erucar Crambe apeiron Didesmus Prantl 1891 er subtribe Ammosperma Myagrum Sisymbriinae dreoskia Pachypterygium Boreava Sameraria epin chimpera Chartoloma Sisymbrium rucaria pirorhynchus Goldbachia Tauscheria Isatis exiera Bayer 1905 _— Brassiceae Brassica Erucastrum ramb: Moricandia Diplotaxis Raphanus ruca Rapistrum Erucaria Sinapis Hayek 1911 Brassiceae, subtribe Calepina Hemicrambe Raphaninae Ceratocnemum Kremeria Cordylocarpus Morisia Cossonia arpus ambe Physorhynchus Enarthrocarpus Raphanus Erucaria Rapistrum Fortuynia Reboudia Guiraoa illa Villani 1926 Sinapeae Cordylocarpus _Hirschfeldia rambe Morisia Didesmus rales Enarthrocarpus arpus Erucaria ction AI Fortuynia Rapis Guiraoa Reboudia Hemicrambe Sinapis Schulz 1923, Brassiceae, subtribe Erucaria 1936 Cakilinae Janchen 1942, Brassiceae, subtribe Erucaria Cakilinae 10 JAMES E. RODMAN is probably the most extensive discussion of evolution within the tribe Brassiceae, Rytz (1936) also argued that Cakile had evolved from Erucaria but, in addition, Cakile represents the basal stock for a group of genera constituting his subtribe Rapistrinae (Calepina, Ceratocnemum, Crambe, Didesmus, Kremeria, Muricaria, Rapistrum). Rytz’s views focused pri- marily on aspects of fruit morphology. In the Brassiceae he distinguished etween “valvar” and “stylar” or “beak” portions of the mature fruit and postulated an evolutionary trend toward reduction of the valvar portion (reduction in size and in number of ovules and seeds), primary increase in the stylar portion or its elaboration into an appendage, and secondary reduction of the stylar portion of the fruit. He considered Cakile an inter- mediate in the reduction series from Erucaria leading to such presumably recent products of evolution as Crambe, Calepina, and Kremeria. These trends in fruit morphology appear to be valid, but the neat linear series which Rytz proposed for the evolution of the genera undoubtedly over- simplify a more reticulate pattern of relationships. Specifically, the series from Erucaria through Cakile to Rapistrum would involve a change in cotyledon position from conduplicate to accumbent back to conduplicate (Zohary 1966) and a change in ovular position along the replum from alternating on both sides to restricted to one side back to alternate (Schulz 1936). It appears simplest to postulate a parallel series of reductions, with Cakile the end-point of one of these lines of evolution. An ultimate rather than intermediate status seems more consonant with the nearly unique strand habitat of Cakile (excepting C. arabica), otherwise rare in the Cruciferae. MoRPHOLOGY AND ANATOMY Habit. All the species of Cakile are succulent, branched, erect to prostrate, herbaceous plants. In favorable sites, large plants may develop a woody caudex, becoming suffrutescent. Plants rarely exceed a meter in height and vary from a few decimeters to 1.5 m in diameter. A collection of C. lanceolata ssp. lanceolata from Jamaica (Harris 11628, cH) bears the note: “plant up to 5 feet high,” though whether erect or, more likely, sprawling over other plants is not indicated. A collection of C. lanceolata ssp. fusiformis from Dade County, Florida (Gillis 7659, cu), bears the observation: “plants up to 2 m long.” In both cases, it is probably the length of fruiting racemes which was estimated. The erect habit is generally encountered in Cakile edentula in its native range and in C. geniculata while C. maritima, C. constricta, and C. lanceolata vary from erect to more commonly prostrate and sprawling. Numerous collections of C. edentula from Australia, where it is natural- ized, indicate that the plants are often prostrate, a habit occurring only rarely in the species’ native range and along the Pacific Coast of North America where it is also naturalized. I have not seen C. arabica or C. THE GENUS CAKILE ll arctica in the wild but to judge from herbarium collections, the former commonly has an erect habit and the latter varies from erect to often prostrate. In the species with an erect growth habit, the branches (which arise near the base in all species) curve outward and upward. The plants often become quite compact and if uprooted when dead and dry may be blown along the beach like tumbleweeds (Cowles 1899). Branches in the more prostrate species vary from procumbent to decumbent. Root. Sea rockets develop a stout taproot, rarely exceeding 4 dm in length and 2 cm in diameter, along with several long, thin horizontal roots. The latter may extend over 15 dm (a specimen of Cakile geniculata, Thieret 25753, cu) and usually grow only a few centimeters below the surface. These probably function quite efficiently in anchoring the plants in the sand. They may also be adapted to absorbing the water which results from “internal dew formation” in dune sands (Salisbury 1952; Chapman 1964). Stem. The stems of sea rockets are succulent to varying degrees, curved, sinuous, or flexuous, smooth or weakly striate, and terete in cross-section. They are commonly green, becoming tan with age, but occasionally developing a purple coloration especially at the nodes. The stem possesses a thick epidermis with outer cell walls 10 » thick in Cakile edentula (Starr 1912) and C. maritima (Wright 1927). The epidermal cells are hetero- geneous in size as in the leaf. In C. maritima, long tubular myrosin cells occur primarily in the outer cortex but are also found sporadically in the pith near vascular bundles (Wright 1927). Leaf. The cotyledons and true leaves of sea rockets are characteristically succulent, a feature widely interpreted as an adaptation for water storage. The degree of succulence varies between taxa but is of little diagnostic value, especially in pressed herbarium specimens. In the American taxa studied in the field, I found the leaves of Cakile lanceolata to be less suc- culent than those of C. edentula and C. geniculata, with C. constricta intermediate. A collection of C. geniculata (Thieret 29350, 4) bears the note: “leaves about 1.5 mm thick.” Chrysler (1904) reported the thickness of leaves of sea rockets collected at Woods Hole, Massachusetts (un- doubtedly C. edentula var. edentula), as 1.17 mm and plants from Chicago (probably C. edentula var. lacustris) as 0.76 mm. He stated that less succulent leaves were characteristic of lacustrine plants compared to their maritime relatives, which appears to be true in C akile. Leaf morphology varies from broadly ovate with = entire margins to deeply pinnatifid. Leaf shape distinguishes some taxa but varies in others. In Cakile arabica, the leaves are finely pinnatifid, with lobes and rachis rarely exceeding 2 mm in width and with the lobes widely separated and themselves occasionally lobed. Deeply pinnatifid to entire leaves occur in C. maritima and C. lanceolata ssp. fusiformis. In all the species, leaves formed late in the season are usually much smaller and less dissected than 12 JAMES E. RODMAN the early leaves. Occasionally one of these small leaves subtends the first flower and fruit of the raceme, which is otherwise ebracteate. Leaf anatomy has been investigated by a number of workers (Chrysler 1904; Harshberger 1908°& 1909; Starr 1912; Wright 1927). The summary presented here is supplemented by my own observations on leaf material of Cakile edentula var. edentula collected near Boston, paraffin-embedded, and stained with safranin-fast green. The leaf epidermis is thinly cutin- ized, and its cells are heterogeneous, a few being much larger than the rest. Stomates, more numerous abaxially, are of the typical cruciferous type and are flush with the surface in C. maritima (Wright 1927) and in C. edentula (Harshberger 1909, as “C. maritima”) but partly sunken in C, lanceolata (Harshberger 1908, as “C. aequalis”). The leaf paren- chyma is not clearly differentiated into distinct palisade and spongy mesophyll. Wright (1927) stated that the mesophyll is undifferentiated in C. maritima, with the outer layers simply possessing more intercellular spaces. Harshberger (1908) and Starr (1912), however, described a pali- sade layer of vertical cells on both sides of the leaf (a “diplophyll”) with a central spongy mesophyll of more compact cells in American taxa. Metcalfe and Chalk (1950) repeated this distinction, but the differences may well be artifacts of limited sampling. Myrosin cells, elongated and tubular, were observed throughout the leaf of C. maritima, following staining with Millon’s reagent or orcin and HCl (Wright 1927). I ob- served darkly green-staining cells, irregular in shape, scattered throughout the leaf of C. edentula; these are probably the myrosin cells although safranin-fast green is not a differential stain for these. Schweidler (1905) placed Cakile in his class “Exo-Idioblastae,” as possessing idioblasts (myrosin cells) in the leaf mesophyll but not in the vascular bundles. Wright (1927) found that myrosin cells occur throughout vegetative and floral structures of C. maritima, including the vascular system of the root, and concluded that the genus more appropriately belonged in Schweidler’s class, “Hetero-Idioblastae,” although Schweidler had based his classes on the presence of myrosin cells in leaf tissue alone. Hyda- thodes are usually present at the end of the leaf, terminating the three major vascular strands; in young leaves they are often purplish. The succulent cotyledons of sea rockets are commonly oblate-lanceolate, about 2-4 cm long, and + equal in length. Lubbock (1892) stated that the cotyledons of “Cakile americana” (probably C. edentula) are “very unequal,” but this condition was not observed in cultivated plants nor herbarium specimens. Lubbock’s observation that the first leaves of C. americana are “very hairy” is also in error, Flower. The flowers of Cakile are typically cruciferous: regular, actino- morphic, tetradynamous, and borne on elongating terminal racemes. The A Is vary from 1-10 mm long and 0.5-2.5 mm wide, are varyingly vergent, and provide characters, particularly in fruiting material, of taxonomic value. The sepals are borne in two indistinct series with the THE GENUS CAKILE 13 outer, lateral pair usually slightly gibbous at the base; they are erect at anthesis and soon deciduous. Sparse, simple trichomes were observed occasionally at the apex of the sepals of all the taxa of Cakile. The margins of the sepals are consistently hyaline. In length, they vary from 3-5 mm, and they are commonly dimorphic at the apex, the two laterals being + straight and acute or obtuse and the two medians tending to being cucul- late inward. The petals, when formed, vary from 5-15 mm long and from distinctly unguiculate in large-flowered taxa to spatulate or obovate in smaller- flowered taxa. Petal size is taxonomically useful although there is consider- able overlap and few discontinuities. Table 2 presents the results of measurements of petals from field collections of flowers preserved in FAA or from herbarium specimens (the latter soaked in water or FAA) made under a binocular dissecting microscope with an eyepiece grid. For field collections, measurements were usually made on 10-15 flowers from a population sample of 50 plants. It is my impression that herbarium flowers give consistently smaller values than material preserved in FAA; measure- ments made on herbarium specimens should, therefore, be scaled slightly upward when compared with FAA-preserved flowers. It is evident that the smallest flowers are found in Cakile geniculata, C. constricta, and C. edentula (and possibly C. lanceolata ssp. alacranensis) and the largest in C. maritima and C. arctica, with C. lanceolata and C. arabica being inter- mediate. Table 3 presents the results of measurements on petals from owth-chamber cultures; measurements were usually made on 10 plants in each collection. The cultivated plants consistently produced slightly larger flowers than wild plants, as a comparison of Table 2 with Table 3 demonstrates. However, the rankings of the taxa by size remained gen- TABLE 2, AVERAGE PETAL LENGTH AND WIDTH (MM) IN WILD COLLECTIONS OF CAKILE Collec- tions Taxon No. No. Length Range Width Range C. maritima ssp. maritima (Pacific Coast) 10 92 10.9 83-138 46 29-62 arctica* 87 75-94 42 35-53 ll lanceolata ssp. fusiformis 49 75 49-94 32 2442 lanceolata ssp. lanceolata 922 #72 +5490 3.1 21-43 25 5 6 lanceolata ssp. lanceolata® 74 5.4— 9.0 27. 44 arabica* Ot 78-116 26. Poo) lanceolata ssp. pseudoconstricta > GU 70 25. S225 edentula var. edentula (Atlantic Coast) : 1.4-3.3 edentula var. edentula (Pacific Coast) SO 72 55-97 21 14-32 edentula ssp. Harperi 39: 66 55-80. 21 15-32 constricta BA 65. 53-60 20 13-26 C. lanceolata ssp. alacranensis* S 64 62-T0 20 19-26 C. edentula var. lacustris 61. 50-70 20 AS26 C. geniculata 44 53 40-61 15 12-19 *Measurements made on herbarium specimens. All others made on FAA-preserved field collections. qgAANnANnAANANnAA bo ro oR OD GOOF uw UR O ee wo co io») So ~ se ~l © bo bo 14 JAMES E. RODMAN erally the same although C. constricta (based on two collections ) tended to approach C. lanceolata in petal size under the growth-chamber con- ditions. At anthesis the claws of the petals are erect, enclosed by the erect sepals, and the blades are reflexed about 90°. The blade varies from entire to emarginate within most populations. Elongated, translucent, apparently tubular idioblasts were observed in most petals of many of the collections; these are probably myrosin cells, which Wright (1927) observed in petals of C. maritima. During field work in North America I observed, in populations of Cakile edentula ssp. edentula on the Atlantic and Pacific coasts and of C. edentula ssp. Harperi, varying proportions of plants which formed flowers with less than four petals, often none at all. In other respects the flowers appeared normal as did the plants. In five populations of C. edentula var. edentula on the Pacific Coast the proportion of plants with incomplete flowers varied from 50-100%, in five populations on the Atlantic Coast, 40-90%, in five populations of C. edentula ssp. Harperi, 10-75%, and in four popula- tions of C. edentula var. lacustris, 0-15%. Allusions to this phenomenon are rare, both in the literature and in collectors’ notes. A collection of C. edentula var. edentula from Lane County, Oregon (Cronquist 6104, GH), bears the comment: “petals none.” Meehan (1892) observed that a population of this taxon at Atlantic City, New Jersey, included many plants with incomplete flowers; he also noted that later the same season plants of C. edentula in Maine possessed a full complement of petals. In 16 growth-chamber cultures of C. edentula I found that flowers were consistently formed with less than four petals, frequently none at all. In four of these cultures, where careful records were kept, I found that flowers borne on axillary racemes late in the season usually had a complete set of petals. Perhaps a temporal shift toward formation of full-petaled flowers occurs in populations of this species. The implications of these observations are discussed in the section on breeding systems. The flowers of Cakile possess four small green nectar glands, two lateral glands subtended by the unpaired stamens and two median glands be- tween and subtending the paired stamens, Hildebrand (1879) and Bayer (1905) provide diagrams of the nectaries in Cakile. The lateral nectaries are compressed horizontally and are usually bilobed. The medians are conical or pyramidal and vary from 0.1 to 0.7 mm high. The size of the median nectaries is of some taxonomic value. Small nectaries (0.1-0.3 mm) are characteristic of C. edentula, C. constricta, and C. geniculata; medium nectaries (0.3-0.5 mm) are found in C. arabica and three of the four subspecies of C. lanceolata; and large nectaries (0.5-0.7 mm) characterize C. maritima, C. arctica, and C. lanceolata ssp. fusiformis. Nectary size appears to correlate positively with petal size in general. Measurements of nectar glands can be made easily on fresh or FAA-preserved flowers, but measurements on herbarium flowers, soaked in water or FAA, probably THE GENUS CAKILE 15 underestimate the true dimensions since the delicate tissues do not fully expand, The stamens of Cakile are tetradynamous at anthesis but become less so as the flower ages and the unpaired stamens elongate proportionally more than the paired stamens. The filaments are stout, about 0.5 mm wide, and taper abruptly to the point of attachment to the anthers, which are dorsifixed near their base. The bilocular anthers are broadly ovate, slightly cordate basally, and with the connective visible as an apiculate tooth at the apex; they dehisce longitudinally and introrsely in all the taxa though somewhat “latrorsely” in the larger-flowered taxa (e.g., C. maritima, C. arctica, and C. lanceolata ssp. fusiformis). Anther size is of some taxonomic value. Table 4 presents the results of measurements of anthers from field collections of flowers preserved in FAA or from herbar- ium specimens (soaked in water or FAA) made under a binocular dis- secting microscope with an eyepiece grid. For each field collection, measurements were usually made on 10-15 flowers from a population sample of 50 plants. The anthers of the unpaired, lateral stamens con- sistently tended to be slightly larger than those of the paired, median stamens. The largest anthers occur in C. maritima, C. lanceolata, and C. arabica and the smallest in C. edentula ssp. edentula and C. geniculata, with C. edentula ssp. Harperi, C. constricta, and C. arctica being inter- mediate. In general, anther size correlates positively with petal size and nectary size. It was strongly suspected that smaller anthers also produced fewer pollen grains but this was not measured. At anthesis the paired anthers are exserted and overtop the stigma, and the unpaired anthers are lower, within the “cup” of the flower, level with the stigma or below it. The immature pistil is green, ensiform, and indistinctly articulated transversely. The short-papillose stigma, atop the stylar beak, is hemi- TABLE 3. AVERAGE PETAL LENGTH AND WIDTH (MM) OF ROWTH-CHAMBER COLLECTIONS OF CAKILE Collections Plants No. No. Taxon Length Width C. lanceolata ssp. fusiformis 4 33 10.9 5.4 C. lanceolata ( Bermuda ) 1 6 9.7 49 C. lanceolata ssp. lanceolata 6 51 9.3 4.9 C. arctica 2 16 9.1 4.7 C. maritima ssp. baltica 1 20 10.4 4.6 C. maritima ssp. maritima ( Pacific Coast ) 2 20 11.5 4.2 C. constricta 2 11 9.1 3.6 C. edentula ssp. Harperi 2 12 8.6 so C. edentula var. lacustris S 30 7.8 24 C. geniculata 2 26 6.8 2.4 C. edentula var. edentula ( Pacific Coast) 1 10 78 a2 C. edentula var. edentula ( Atlantic Coast ) 4 25 7.6 1.8 16 JAMES E. RODMAN TABLE 4, AVERAGE ANTHER LENGTH (MM) IN WILD COLLECTIONS OF CAKILE Collections Plants Lateral No. No. A Median Taxon nther Range Anther Range C. maritima ssp. maritima (Pacific Coast) 10 92 LS 2 15-2.2. 16 = 14-21 C. lanceolata ssp. pseudoconstricta 1 6 16 16-17 16 ——— C. lanceolata ssp. fusiformis 9 49 16 13-18 15 13-17 C. lanceolata ssp. lanceolata 4 22 160 74-17 2-18. 14-48 C. lanceolata (Bermuda)* 9 9 1S. 222 14 <2 S80 C, arabica* 5 5 1S le8 1s tee C. lanceolata ssp. lanceolata® ya 25 4 13-07: 13 tie C. lanceolata ssp. alacranensis*® 3 3 14 13-15 13 12-15 C. arctica* ll LB 13° 13-14 13 13-14 C. constricta 6 54 13 -T1-Le Le oe C. edentula ssp. Harperi 5 39 13 10-14 12 10414 C. geniculata 5 44 12 07-13 11 061.3 C. edentula var. edentula (Pacific Coast) 5 50 11 09-14 11 £0O8-1.3 C. edentula var. edentula (Atlantic Coast) 5 38 10 04-14 09 0.4-1.3 C, edentula var. lacustris 4 30 0.9 07-13 0.9 0.6-1.3 *Measurements made on herbarium specimens. All others made on FAA-preserved field collections. spherical to slightly bilobed. The articulation of the stigmatic lobes is perpendicular to the plane of the septum. The characteristic lateral horns on the lower segment of the mature fruit of Cakile maritima are absent at this early stage. The immature ovary is initially unilocular, with the linear placenta disposed vertically in the elongated chamber and the ovules, usually two, attached in the region of what is to become the articulation between the two fruit segments. A thin elongated septum develops with the ovules on one side of this membrane; a diagrammatic cross section of the young pistil is given by Hannig (1901) and Markgraf (1963). Following fertilization, the ovary wall grows inward between the two ovules, pinching off two segments, the upper one with an erect ovule and the lower with a pendulous one. The small suture which is evident on the articulating surface of the mature fruit segments results from the in- complete fusion of this ingrowth of the ovary wall. The articulating surface is, therefore, not homologous with the septum of other crucifer fruits, nor are the two fruit segments homologous with the two “carpels” of a crucifer fruit. Each mature fruit segment can be split lengthwise into two equal halves, usually revealing a single oblong seed appressed against a thin, papery membrane—that segment’s portion of the septum. In maritima, Wright (1927) observed in immature pistils the regular forma- tion of four ovular primordia, two of which consistently aborted. The chance development and fertilization of one or both of _ these ovules, in addition to the typically developed two, would explain the infrequent but regular occurrence of two-seeded (and rarely three-seeded) fruit segments in Cakile. Wright (1927) interpreted the gynoecium of Cakile as tetracarpellary, following Edith Saunder’s views on the general THE GENUS CAKILE 17 nature of crucifer carpels. More recent opinion inclines toward a bi- carpellate interpretation of the cruciferous gynoecium (Carlquist 1969). Fruit. The distinctive fruit of Cakile is an indehiscent, hard-corky (“nucamentaceous”), two-segmented silique. Zohary (1948) described the fruit as a special case of his “valvo-nucamentoid” type and con- sidered the upper and lower segments to be homologous to, respectively, the “stylar” and “valvar” portions of fruits typical of the tribe Brassi- ceae. The upper segment is deciduous; the lower remains attached to the raceme. The fruit is thus an example of what Zohary has termed “hetero- mericarpy” (van der Pijl 1969). Both segments are (falsely) one-loculed and usually one-seeded (although a septum develops, seeds are formed on only one side of it). Size and shape of the fruit provide important characters for distinguishing infrageneric taxa (Plate 1). Features pertain- ing to the articulating surface between the two fruit segments have some taxonomic significance although certain characters, such as number of small teeth on the surface of the lower segment, have been emphasized without regard to their variation within populations. The number of seeds per segment has, unfortunately, also been used as a taxonomic character, particularly by Small (1903, 1913, 1933), again without regard to the variation within populations. Characteristics which have been found to be taxonomically valuable include the shape of the beak, whether retuse, obtuse, or sharply acute at the apex; the length of the segments; the out- line of the upper segment in cross-section, whether distinctly four-angled, eight-ribbed, or + sulcate; the development of a membranous margin at the base of the upper segment; the presence of lateral horns on the lower segment; the shape of this segment, whether cylindric or distinctly broadened at the top; the consequent width of the articulating surface; and the development of various projections on this surface. While fruit characters are among the most reliable in the genus, nearly all of them intergrade to some degree; and there are few sharp discontinuities between the taxa. This is the case with nearly all the characters in Cakile and it is perhaps to be expected in an assemblage of (primarily) strand plants with contiguous geographic ranges and great dispersibility. One species which can be readily identified from its fruits alone is C. arabica, which is the only geographically isolated species in the genus. The averages resulting from measurements of pedicel length (of fruits ) and fruit length (and length of the constituent lower and upper segments, usually ), made on herbarium specimens, are presented in Table 5, with the taxa ranked by decreasing fruit length. For comparison, measurements to the nearest half-millimeter were made only on one of the four lower- most mature fruits of a raceme (usually the main raceme) for which the segments of the fruit were judged to be fertile. In this way, confounding errors resulting from variation along a raceme or from variation between fertile and sterile segments were avoided. In a few cases, when none of Z | 2 =) ie) mG fe n fx] fet a < a NYT ETT c T O1NL3W Oeeeeeeeneen | oly. THE GENUS CAKILE 19 the four lowermost fruits was fully fertile, measurements were made on one fertile segment (the other being recognizably sterile from its smaller size). Consequently, the combined values for the lower and upper seg- ments may not always equal the value for fruit length for a particular taxon in Table 5. For each value for fruit length, the coefficient of vari- ability, V (Simpson, Roe & Lewontin 1960), was calculated from the mean and standard deviation to provide an estimate of the amount of variation in the samples of collections measured. In addition, for many airs of values for fruit length, Student’s t-test was calculated to assess the statistical significance of the differences between taxa. For the most part, the longest fruits and the longest pedicels are found in Cakile arctica of Iceland, the Faeroes, and northern Scandinavia. These sea rockets have often been treated as conspecific with C. edentula var. edentula of the Atlantic Coast of North America, but the latter is clearly distinct and possesses shorter fruits, the differences between the two being highly significant statistically (P < 0.001). The fruits of C. edentula ssp. Harperi are nearly as long as those of C. arctica but are clearly distinguish- able morphologically. Cakile edentula var. lacustris of the Great Lakes bears longer fruits than does C. edentula var. edentula of the Atlantic Coast, and the difference is statistically very significant (P < 0.001), as is the difference in fruit length between C. edentula var. lacustris and the Great Lakes collections of C. edentula var. edentula. Patman and IItis (1961) also demonstrated a rather clear separation of the two varieties by fruit size in Wisconsin collections. Surprisingly, the difference in fruit length between Atlantic Coast C. edentula var. edentula and the Great Lakes collections of this variety is also very significant (P < 0.001). The difference could have resulted from a bias in measuring, longer fruits being considered auto- matically as C. edentula var. lacustris, thus leaving a residue of small- fruited typical C. edentula. Such a bias, however, would most likely have resulted in a much higher value of V for C. edentula var. lacustris than was found. The Great Lakes material of C. edentula var. edentula does not appear to differ in any other respect from typical Atlantic Coast material, and I would attribute the difference in fruit length to a founders’ effect. The difference in fruit length between Cakile edentula ssp. edentula var. edentula and C. edentula ssp. Harperi of the Atlantic Coast is also highly significant statistically (P < 0.001). Intermediates do occur, however, in the transition zone between the two along the Outer Banks of North Priate 1. Fruits = fered A. Left to right: C. arctica (Rodman 107); C. maritima ssp. baltica (Rodman 182): C. maritima ssp. maritima (Rodman 78); C. maritima ssp. euxina (Stojanoff s.n.); C. arabica (Dickson posts B. Left to right: C. edentula var. lacustris (Rodman 132); C. edentula var. edentula (Rodman 96—segments dite and sepa cag C. edentula ssp. Harperi (Rodman 45— i onstricta (Rodman 60). C. Left to right: C. geniculate (Rodman 64); C. lanceolata ssp. alacranensis (Fosberg 41910); C. lanceolata ssp. fusiformis (Rodman 53); C. lanceolata ssp. lanceolata (Rodman 114); C. lanceolata ssp. pseudoconstricta isa aan 57). TABLE 5, AVERAGE PEDICEL AND FRUIT LENGTH (MM) IN CAKILE Collections Lower Upper Taxon or Area No. Pedicel Range Segment Range Segment Range Fruit Range ve C. arctica 10 65 5 -10 85 6 -ll 163 14 -20 248 21 -31 12.4 C. lanceolata ssp. lanceolata 134 2.7 15-4 7.2 5 -10 17.6 13 -24 23.8 19 -31 9.6 C. edentula ssp, Harperi 49 36 2-5 8.7 7.5-10 in). = 4 19.5. S38. «26-20 9.3 C, geniculata 32 38 2-5 82 6 -ll 15.7 13 -18 23.6 20 -27 7.6 Mexico; Gulf collections, all taxa 15 30 15-45 75 5 -10 15.3 12 ~-18 21.8 15 -27 17.9 C. lanceolata (Bermuda only) 19 B82: °°0 2 4 1.335729 14.7 - 10-20 212 14 -29 18.7 Louisiana collections, all taxa 14 v0. 2 45 72 8-10 13.9 10.5-18 91.0 17 25 14.8 C. edentula var. lacustris (Great Lakes) 147 44 2-8 ey Ge Ee i606. - glucose rosina ae y Inase _ MeOsoL BNC Ss 68 aie + D-glucose Fic. 1. General reaction for the enzymatic hydrolysis of glucosinolates (I) to isothiocyanates (II). THE GENUS CAKILE 37 likewise reported finding only this one compound in their studies on seeds of C. maritima. By paper-chromatographic analysis of the unripe fruits, Delaveau (1957) detected seven compounds in C. maritima, four of which he tentatively identified as allyl, 3-butenyl, 4-pentenyl, and phenylethyl isothiocyanates. Daxenbichler et al. (1964), analyzing seeds of C. edentula, reported allyl and sec-butyl thioureas as well as a third, unidentified compound. Results from a paper-chromatographic analysis of two samples of Cakile, done previously by Dr. Martin Ett- linger, who was an Honorary Research Associate in the Gray Herbarium 1969-70, demonstrated two distinct patterns of glucosinolate composition. These results encouraged a belief in the systematic value of the com- pounds in this genus which my investigations have subsequently con- firmed. Material and Methods. Seeds from mature (i.e., dry and corky) fruits were used for all the analyses, both by paper and gas chromatography. The fruits were stored at room temperature in paper packets for up to three years or, in a few cases, in a refrigerator at 5° C. for one year and shelled just prior to analysis. The different storage conditions did not appear to affect the glucosinolate composition of the seeds. Appendix I gives the identity and place of collection of all the samples analyzed. Whenever possible, mass samples were analyzed, that is, samples of seeds collected from 25 to 60 (usually 50) plants in a local area. In such instances, the seeds originate from one of the four lowermost mature fruits on the main or central raceme of the plant. When seeds from a single plant or only a few plants were analyzed, mature fruits from the entire length of a raceme were used. For the analysis by paper chromatography, the glucosinolates are not chromatographed directly; rather, the seeds are treated with a preparation of the myrosinase enzyme to yield isothiocyanates which are then treated with ammonia to produce thioureas. It is the resulting thioureas which are chromatographed in up to six different solvent systems, and the spots, developed by spraying with Grote’s reagent, are then identified by their mobilities relative to that of a reference compound, phenylthiourea. Separate tests are run to detect the presence of p- hydroxybenzyl isothiocyanate and oxazolidine-2-thiones. The detailed procedure has been described previously (Kjaer & Rubinstein 1953; Ettlin- ger & Thompson 1962; Ettlinger et al. 1966). The six solvent systems employed were: n-butyl alcohol, toluene, and water, 3:1:1 ( “butanol solvent); toluene, n-butyl alcohol, and water, 3:1:2 (“3 toluene solvent); toluene, n-butyl alcohol, and water, 10:1:2 (“10 toluene” solvent); toluene, acetic acid, and water, 5:2:4 (“toluene-acetic” solvent); benzene, ethanol, and water, 5:1:2 (“benzene” solvent); and chloroform and water, 5:3 (“chloroform” solvent). The distance in centimeters of the center of a spot from the starting point on a chromatographic strip was measured, and this number, divided by the distance traveled by the 38 JAMES E. RODMAN reference compound, phenylthiourea, gave an “Ry,” value which charac- terizes the mobility of that particular compound in that solvent. Occa- sionally, a compound can be identified from its R,,-value in a single solvent system; allylthiourea, for example, has an R,,, of 0.25 + 0.05 in the chloroform solvent which appears to be nnique. More often, the pattern of mobilities in more than one solvent must be used to identify a com- pound. While this paper-chromatographic procedure does not discrim- inate all known glucosinolates, it serves to identify with reasonable accuracy a majority of the glucosinolates found thus far in higher plants. The technique of gas chromatography for the analysis of glucosino- lates was first reported by Kjzr and Jart (1957) and by Jart (1961), who chromatographed the enzymatically released isothiocyanates. Sub- sequent reports on gas chromotography of isothiocyanates confirmed its great utility for quantification of the compounds (Youngs & Wetter 1967) and for discrimination of complex mixtures (Binder 1969). Seeds, 0.1 to 1.2 g, were ground with a mortar and pestle, transferred to a 250 ml Erlenmeyer flask, and overlaid with 100 ml of anhydrous ethyl ether, free of alcohols and peroxides. The flasks were left in the dark at room temperature for three to five hours with occasional shaking. The defatted seed meal was then filtered off, washed with ether, air- dried a few minutes, and transferred to a 50 ml Erlenmeyer flask. To this were added 7-10 ml of ethyl ether, 5 ml of distilled water, 0.1 ml of ascorbate solution, and two drops of the myrosinase preparation ( Ettlin- ger & Thompson 1962). The flasks were placed on a mechanical shaker, operated slowly, and left overnight in the dark at room temperature. In the morning the ether solution was pipetted into 10 ml volumetric flasks, which could be stored in a refrigerator. A variant of this procedure was used after it was noticed that spurious compounds were appearing on chromatograms as a probable result of some endogenous enzymatic activity. To the ether-extracted seed meal, before addition of the myrosinase, boiling 70% methanol was added, and the mixture was left to boil for 15 minutes, by which time the methanol had evaporated leaving an aqueous mixture. After cooling, the ascorbate solution and myrosinase were added, and the procedure was continued as outlined above. The technique effectively eliminated these spurious compounds, probably by destroying all native enzymatic activity. The ether solutions of isothiocyanates were chromatographed on an F & M Model 402 gas chromatograph with a hydrogen flame ionization detector, linked to a Hewlett-Packard Model 7127A strip chart recorder and operated isothermally at 80° C. (flash heater 130° C. and flame detector 145° C.) and at 160° C. (flash heater 200° C. and flame detector 170°C.). A single glass U-column, 6 ft * 3 mm LD., packed with 6% plus 2% EGSP-Z on 100/120 Gas Chrom Q (from Applied Science Laboratories, State College, Pa. 16801), was employed, with helium as the carrier gas at an approximate flow rate of 60 cc/min. The THE GENUS CAKILE 39 size of the injected sample varied from one to five microliters. For quantitative comparison, peak areas on the chromatograms were com- puted by height x width at half-height. There was no difficulty with overlapping compounds in the two runs, at 80° C. and at 160° C., since all peaks appearing within the first 13 minutes at 80° C. came off in less than one minute at 160° and were usually obscured by the ether. No attempt was made to quantify absolutely the glucosinolate content of the seeds; only relative amounts of the isothiocyanates appearing on the chromatograms were determined. Identification of the peaks was made by judicious comparison with the results from the paper-chromatographic analyses of the same sample and by comparison with gas chromatograms of synthetic isothiocyanates or extracts from species known to possess particular glucosinolates. In questionable cases, the sample and the suspected isothiocyanate (synthetic or derived from a known plant source) were mixed together and chromatographed, and complete over- lap of the peaks was assumed to indicate identity of these compounds. Twenty-one enzymatic products were detected by paper and gas TABLE 14. GLUCOSINOLATE HYDROLYSIS PRODUCTS FROM SEEDS OF CAKILE. Compound Retention time in min. Number Name R (Fig. 1) @80°C. 1 unknown 1 id : 1627 2 isopropy!* (CH,),CH 2.0- 2.2 3 allyl" CH,—CHCH, 2.8- 3.1 4 sec-butyl* C.H.CH(CH,) 3.5- 3.9 5 3-butenyl* GH,—CH(CH,), 5.3— 6.1 -6 unknown 2 ? 7.8- 9.4 7 4-pentenyl* CH,—CH(CH,), 10.0-11.5 Retention time in min. @160°C. 8 unknown 3 ? 1.3- 1.4 9 unknown 4 ? 16— 1.7 10 unknown 5 ? 15- 22 11 3-methylthiopropy]* CH,S(CH,); 2.7— 3.1 12 own 6 is 25- 3.5 13 benzyl" C,H,CH, 3.5-— 3.9 14 4-methylthiobuty!" CH,S(CH,), 45-54 15 2-phenylethy!* C.H.(CH,). 5.6— 6.5 16 5-methylthiopenty]* CH,S(CH,), 13-86 17 6-methylthiohexy!* CH,S(CH,)., 11.5-12.4 R,, in “Butanol” solvent 18 3-methylsulfinylpropy!” CH,SO(CH.,); 0.37 19 4-methylsulfinylbutyl” CH,SO(CH.), 0.43 20 5-methylsulfinylpenty!” CH,SO( CH, ); 0.58 21 4-methylsulfonylbuty!” CH,SO,(CH:), 0.52 a = isothiocyanate; b = thiourea. 40 JAMES E. RODMAN chromatography following treatment of the seed meals of Cakile with myrosinase, and 15 of the presumed parent glucosinolates have been identified chemically (Table 14). The procedure for identification —involving comparison by paper chromatography of sample R,,, values with a table of the known mobilities of 42 thioureas and oxazolidine-2- thiones (Dr. Martin Ettlinger, unpublished results) and comparison by gas chromatography of sample retention times with the retention times of synthetic isothiocyanates or of extracts from plants known to yield a particular mustard oil—is reasonably accurate and discriminating though not conclusive. The paper-chromatographic procedure does not discrim- inate, for example, isobutyl and sec-butyl glucosinolates (Ettlinger & Kjer 1968) although there is evidence that gas chromatography does dis- tinguish the two (Binder 1969; Dr. Ihsan Al-Shehbaz, personal com- munication). Such a caveat applies to all chromatographic analyses. Absolute identification of the constituents would require spectroscopic studies and even chemical syntheses, which a taxonomist usually cannot undertake. The 21 glucosinolate hydrolysis products are listed in Table 14 with their gas-chromatographic retention times at 80° C., 160° C. or the R,,, values in butanol solvent. Neither p-hydoxybenzy] isothiocyanates nor oxazolidine-2-thiones were encountered in the tests for these, and the Sn glucosinolates are presumably absent from seeds of akile. No effort was made to analyze other plant parts for glucosinolates except for collection 107 from Iceland, for which fresh leaves of greenhouse-grown plants were analyzed by paper chromatography in chloroform solvent. One compound, allylthiourea, was detected, which corresponds to the major constituent of seeds of this collection. It should not be expected, however, that the same glucosinolates will be found in other parts of the same plant. Working with species of Brassica and with Sinapis alba, researchers have reported significant differences in glucosinolate composition in different plant parts (Josefsson 1967, 1970; Bergman 1970; Kondra & Downey 1970). Moreover, developmental differences in glucosinolate composition have been demonstrated in various plant organs (Kondra & Downey 1969; Bergman 1970; Tang 1971). The use of mature seeds in this study of Cakile eliminates these sources of variation. Table 15 presents the results of paper-chromatographic analyses of 27 seed samples from 21 populations of Cakile along with the unpub- lished results from two analyses (M. G. Ettlinger and C. P. Thompson, personal communication) on C. maritima ssp. maritima from California (no voucher) and C. geniculata from Tamaulipas, Mexico (voucher: Johnston 2547, cu). An estimate of the quantitative relationships in each sample was made on the basis of the size and intensity of color of the spots after spraying with Grote’s reagent, and each compound was classi- fied as either primary (P) or secondary (S$). Secondary compounds include THE GENUS CAKILE 41 TABLE 15, GLUCOSINOLATE COMPOSITION OF SEEDS OF CAKILE DETERMINED BY PAPER CHROMATOGRAPHY OF THE THIOUREA DERIVATIVES. Compound (see Table 14) D627 89 10 1) 42g a 1S 16 2. AS ie oo S S S S S Sample gee aS | 94a/L ae} —_ si (ea NNN Ny s ‘send ANNNN Pot Ne a eM A A oes Pe re re a S P P P S S S S S S S S S S S AANDNANANANNAMNNN ANnNnNN n ~~ i i" Ble ~ ee” thas lg « beri lg Fe = ils = ain Beg « Bes ba eke? en cg oa ola? plan! oe’ OME? ola») a ®) —~ NANNNNHNHHARAURHNU ? S S S P > S Ss 3 = 6 C. gen. | ee Ss Legend: P, primary compound; S$, secondary or trace compound. those appearing as only trace spots on a few chromatograms. The results of the gas-chromatographic analyses of 96 seed samples are presented in Table 16. Tracings of gas chromatograms of four samples are shown in Figures 2-5. The taxonomic identity of the collections, locality, date of collection, and voucher are given in Appendix 1. Of the 27 samples which I analyzed by paper chromatography, all but two (collection 94b from Oregon and R-s.n—2 from St. John, Virgin Islands) were also analyzed by gas chromatography. With some of these samples, however, lower seeds (that is, seeds from the lower fruit seg- ment) were used in the one type of analysis and upper seeds in the other, or lower or upper seeds were used in one and a mixture of both in the other type of analysis (a mixture being indicated by the absence of / L or “/U” after the collection number in Tables 15 and 16). One obvious 42 JAMES E. RODMAN TABLE 16, VOLATILE ISOTHIOCYANATE COMPOSITION OF SEEDS OF CAKILE AS DETERMINED BY GAS CHROMATOGRAPHY ( PEAK-AREA PERCENTAGE BY TRIANGULATION ) ee es ae aa} ed se tie 4 eG a 4 ae =i —s S ao S =| = Sc a eee ee oe EE CE ee OE Bee a eae ia eee 5 II, £68.48 eet Ge aoe ee ee ee Se ee oe ee ee Oe ee tS bl! 8. Sb a SG oe Australia *P_sn. S162 35005 44 6 0.65 0 T2150 .0 679: 0-28 0 Pacific Coast *98-9/U Se tree tO ee 8 ea OT OO OO US Ue *9] Sit esee 7 Oo 66 oT OT 6 oT 8! UO er eS 90 0 200 T 566.0. .0. 6.24.0 (5.203 6: 012.10 42 4 89 oe eee ee oT 8) 1A 8 OS 0.48828 94a/U oo ee oe bee Pe Oe UT OU. UTS Od6 03.06 68 eee 7 So 0 oon 0 60 6° Tt OCF OCT CIS OC U180 6 16S 34 *89 wo 7 Be OC 0: Oe eT 180 20°: 67:50. 2a 8 °79 O60 28.7. 6 6 Gg eT ITA O 0 108: 0: 42% °76 O30) tae to 6 Tt CF 8120 8: 70.0 388 73 6 > 6 8. OOP oP 6 P 0 Rae ast A | C7) 200 fr 8 se Too se 08:0 70 0.3825 69 oo | eee 0 oO 8S 8 18 182 6° 0-100 0. SAE Great Lakes Wee 6 t O44 4) ft 6 6 oe eo Tse Tt OT 6 Oe 132 0 08S £4 16 28 8 FF Ooo Teo Oo 808 133 1 Soe 48.68 © 0.0 oT 8: 40.0. To 6 Te *97.2/U . > 10 Oe 6 6 an ht 8 ak 8 8 8 8 Atlantic Coast of North America lee @ 16h F-66866 6 ts 8 0 0° 8 137/L Ste eeeee 48 6 6 6 6 ie 6 61: 8 O..0. 2 136/L Oe 1 ee 34.180, 6 6 FS te ea 48 oo 88 103 0 32455508 T oO © © 49°60 7 @ 0 0 8 °34-2/U O10 400 i 7 280 64.0: 6 FT) 6 8 6 8 0. 8 © 21419663 06.0:.0. 6 .7 6.08.0: 7 8.0 .9..8 *95/L 2 wee e008 0. 8 8 e RB) 6S Bb OO UF OF O16 14 Tt 8 8 OT 6 Se OO Tr 8 Oo Ce 26/L ee 8s Ot ee 6 6 8 er lu 6 lO lO 26-2/L 0.7 908 60.0804 6 6. 7 6.98 6 05. 0..0.0: #8 151/L oO PA AL ibe 6 6 ft. Un iA 6 to 68. 88 150/L M0 M4 62 1352 6 a TG UIs 6 Gt’ Oo OO 8 100/L oc Ol 98 08 1 6 6. Fy 6 To: 89 8. 0. 0 8 Legend: R = variable side-chain; cf. Fig. 1. T = trace amounts p . P = present but not penites /L= seeds from only lower fruit segments used./U = seeds from pat a upper fruit segments used. xtracted in boiling 70% methanol. Boldface = mass sam ple analyzed. THE GENUS CAKILE cs © oie Peete eee ee 3 Shee Fo ee ee eee Sf ie top oa eg ae a eS Sa 8 OB Se SB eee oe be eS Atlantic Coast of North America (cont. ) 98/L 6. T° 8 87 69 0626-082 6 8. oe eo ee 97/L O° T2018 TF 19 Toe eS Te ee SC ee ee 99/L OPA SIOALE 1208 00 (Oo Ee 8 OT Ot 8 eee 96/U 06.7 °910-59 09.0% 02.02 Ts 8 ie 2 te 0 8 aw 96/L 6. T BOOK LE 6. 06s Te se ee 96-2/L oT O88 35 O08 OS OOO Oe eT eae 39/U 6: 065802 48 19 0: 00 6 To ase OT 8 Oe 40 6...O0883 40198 0.26 © Doe ee oe 8 ee 41 46 6 88% © 680 6 On tN Te oe Le 43 69 o -S% o2770 6 43° FT 5 Os FO te es tO 44 Soo -390 09804 0 (100P 7 So 06 7 (0 Et FOS. 0 *45/U an 6. SO. To peace Ot Fr To 8 Tt 8 *45/L 64.6 86 64002 6 O08 FT © Oe Ce ert 45 tk OG Se ORGST- 0 O68 TOT YT Cer OU Geo se 8 24-3 194.0 44 OF780°0 18°07 7 0 08 7.90 ft roe 8 47/L Ji 6. 8h 69808 © 16 Te TE Te 8 a Ta 8 131/L 142 Oo MIMO OD 1k bt fF Oe 7 8 FP 8 *131-2 es ag 96 GA0en 0 0 fT © O88. et , 6 48 en G@ 88 boese 6 4) TY FT 8 1o TT 05 -0 49/L ns 6 9) S198 6. 0-7 0. 2 08 ©. 8 ee | ES 50 94.0 60 73668 0° 0 T 0 48 0608 9 af: O° On 0 52 bo 6 £0 fT 710 6 t TC bet. 8 41 0 06 0 51 Gah. 640.788 OT FT 8. OR OF te 22°86 2.0 G-7659 ac G6 ks 6:62.60 0 28 0.64 27 7 8 7 6. T 6 W-9138/U 34 0 116 0 7530 0 0 O 06 28 0 ® 56 6 06 0 Wess 86 9 62 0 723 6 © T 0 10 tit o sh 6° 7 98 55 aa 0° 68 7 Bh 08 tO UT 88 8 Oo 38 5.70 53 "A 8 BS 6 769 6 OT 8 te Oe Ft 8 ec a aE Gulf of Mexico 56 ian 6s 7 O08 6 6 Fe Ft O48 nh 660 Tt D 130/L 15s 6 io 6 oe be 8 eG 8 Yan 8. Fr 8 57 * § 036 0 e658 0 8 12 0.10 14 6 0 4.6 7-6 58 to 6 i560 858 6 0 68.0. ft Ss Oo 6. 65.0 0 <0 60 Bo p 19) 45600 0 @ T 0 24 O06 ": 5° 28 60 0° 0 62 as 5 66 ise © 0. TO OT OS so 24: 0 0 8 62 “P,’ 1” @ 164 P7060 6. TF 8 Oo Ff oe 8. F 6 -T 0 62, “H” ao 6 63 Oo 8 0 0 TF Oe 7 “at ye Oo te 62 “P,” 90 lanceolata ssp. lanceolata (R-s.n. ) . edentula ssp. Harperi (24-2) x C. 80 1l >90 lanceolata ssp. fusiformis (W-9138) C. lanceolata ssp. lanceolata (R-s.n.-2) x 68 17 53 C. edentula var. edentula (26) C. lanceolata ssp. lanceolata (R-s.n.-2) X 75 4 60 C. geniculata (25) C. lanceolata ssp. lanceolata (R-s.n.-2) x 56 17 >90 C. lanceolata ssp. fusiformis (W-9138-2 ) C. lanceolata ssp. fusiformis ( W-9138-2 ) 43 52 >90 x C. lanceolata ssp. lanceolata (R-s.n.-2) C. lanceolata ssp. fusiformis (W-9138-2) 24 9 >90 x C. geniculata (25) C. lanceolata ssp. fusiformis (W-9138-2) 44 7 89 x C. edentula var. edentula (26) Legend: *pistillate plant listed first; collection number is given in parentheses; **fruit from artificial selfing. environmental conditions, percentage of spontaneous fruit set (number of maturing fruits/number of flowers produced) in F, plants grown in growth chambers under uniform conditions, and percentage of pollen viability in these F, hybrids. To assess pollen viability, 100 grains from 4-11 F, plants were counted in methylene blue or acetocarmine and scored; aborted grains were much smaller than normal and did not take up the stain. It is evident from Tables 25 and 26 that viable seeds were produced from all the crosses and that the resulting F, hybrids were fertile, with a spontaneous fruit set ranging from 24 to 86% (in 69 « 96 F, plants, where spontaneous fruit set was very low, plants were artificially selfed to test their fertility, giving a value of 45% fruit set). No attempt was made to grow an F, generation. The F, hybrids were generally intermediate between their parents in vegetative and floral morphology. Moreover, as demonstrated in the section on biochemical systematics, individual F, plants from the crosses C. edentula ssp. Harperi x C. edentula var. edentula, C. lanceolata ssp. lanceolata x C. edentula var. edentula, and C. lanceolata ssp. fusiformis x C. edentula var. edentula exhibited a seed isothiocyanate composition intermediate between their parents. Table 17 presents data on average petal length and width for F, hybrids from all the crosses. With two exceptions, the F, plants were consistently intermediate between their THE GENUS CAKILE yes parents in petal size. In two cases involving the small-flowered C. eden- tula ssp. Harperi as pistillate parent in crosses with C. lanceolata, the F, hybrids had slightly larger petals on the average than the larger-flowered, staminate parent. The two cases may constitute an example of heterosis. These results from artificial crosses indicate that natural hybridization is a possibility for most of the taxa of Cakile if not for all. Crosses involv- ing C. constricta, C. arctica, and especially C. arabica would complete our understanding of the genetic barriers, or lack of them, in this genus. Herbarium specimens of plants with intermediate morphologies and the patterns of character variation in mixed populations of sea rockets sug- gest that hybridization does occur in nature. Eichler (1965) suggested that hybridization occurred in mixed popula- tions of Cakile maritima and C. edentula in South Australia. Specimens he labeled as hybrids (Eichler 12119, av, cu.) do show some intermediate characteristics, However, Eichler 14198 (av, Nsw), identified as a hybrid from its entire leaves, appears to be normal C. maritima. Plants of this taxon with entire leaves occur sporadically throughout its native range as well. Along the Pacific Coast of Nerth America where the two species have also become naturalized, hybrids appear to be quite rare (Barbour & Rodman 1970). Dr. Dennis Anderson of Humboldt State College stated in a personal communication (1969) that he was unable to find any evidence of hybridization between the two in northern California. In the nearly 400 collections from the Pacific Coast studied, only three appear to be hybrids: Schreiber 2185 (uc) from Marin County, Calif., Tracey 18062 (Ncu) from Humboldt County, Calif., and Stuckey 1930 (os) from Lincoln County, Oregon. No hybrids were detected during my own field work along the West Coast; in mixed poulations, plants could be readily identified as one or the other species. I attribute this lack of hybridization to the different breeding systems of the two taxa. The small-flowered C. edentula is probably highly self-fertilizing. How- ever, C. maritima is primarily allogamous and probably dependent on insect activity for good fruit set. In their foraging activity, insects can be expected to consistently favor the larger, showier blossoms of this species. Heslop-Harrison (1953, 1954) reported plants of Cakile maritima and C. edentula in the Outer Hebrides and a sterile hybrid between the two. The plants identified as C. edentula, however, are probably not this species as Clapham, Tutin, and Warburg (1962) pointed out. Rather, they may be C. arctica, native in Iceland and the Faeroes, or, as Ball (1964a) suggested, simply variants of C. maritima. The Icelandic plants have been treated as C. edentula by many botanists (Ball 1964a), and Heslop-Harrison (1953) speculated that the Hebridean and Icelandic plants were conspecific. Icelandic plants could have migrated by sea to the Outer Hebrides on the North Atlantic Current. Hybrids between C. arctica and C. maritima would be difficult to identify without fruiting material since the two are very similar in floral characters and even in TABLE 27, AVERAGE PETAL LENGTH AND WIDTH (MM) OF PARENTAL AND F, PLANTS OF CAKILE Pistillate Parent F Hybr d. Staminate Parent Hybrid (N)*® Length Width (N) Length Width (N) Length Width C. maritima ssp. maritima (69) x C. edentula var. edentula (96) ** 10 120 4.4 ll 16g 38 2. 60. 16 C, edentula ssp. Harperi (24-2) x C. edentula var. edentula (34) 10 89 3.6 6: 8.7 °-30 10°: (8% bi C. edentula ssp. Harperi (24-2) x C. lanceolata ssp. lanceolata (R-s.n.) i0 §6©89 3.6 0” 88° 68 10 9.0 48 C. edentula ssp. Harperi (24-2) x C. lanceolata ssp. fusiformis (W-9138 ) 16. 45° 35 10 10.4 5.0 10.101 ° 48 C. lanceolata ssp. lanceolata (R-s.n.-2) x C, edentula var. edentula (26) 10 90 46 10° 802 S37 4. 68 25 C, lanceolata ssp. lanceolata (R-s.n.-2) x C, geniculata (25) 10 9.0 46 4. @05 38 10° 64-22 C. lanceolata ssp. lanceolata (R-s.n,-2) x C, lanceolata ssp. fusiformis (W-9138-2 ) 10 90 46 190 11,1°°458 10: 11358 C. lanceolata ssp. fusiformis (W-9138-2) > C. lanceolata ssp. lanceolata (R-s.n.-2 ) 10. 11.3. 58 10 106 5.8 10 90 46 C. lanceolata ssp. fusiformis (W-9138-2) % C. geniculata (25) 10° 13.388 10 10.0 4.6 10--64- 32 1011.3 35 10° 1002 45 4 6822.0 C. lanceolata ssp. fusiformis (W-9138-2) % C. edentula var. edentula (26) *Number of plants measured; **Pistillate plant first and collection number in parentheses. NvWaouw ‘a sanwvt THE GENUS CAKILE 75 leaf morphology (considering the entire-leaved forms of C. maritima). A sterile F, hybrid would be unusual in view of my results from artificial crosses (albeit not between these two species ). On beaches of Lakes Michigan, Erie, and Ontario the native Cakile edentula var. lacustris occasionally grows intermixed with the rarer C. edentula var. edentula, which is most likely a recent introduction to the Great Lakes. Hybrids between the two appear to be rare. In Wisconsin, Patman and IItis (1961) demonstrated a clear distinction in their mate- rial between the two varieties, and Dr. Hugh Iltis of the University of Wisconsin stated in a personal communication (1971) that hybrids were absent from that state. A few collections possibly of hybrid origin were noted among the nearly 400 collections studied from the Great Lakes region: Bessey s.n. (Msc), Muskegon Co., Mich. (upper fruit segment 9 mm long X 3.5 mm wide), McVaugh 12633 (micH), Manistee Co., Mich. (14 X 5.5 mm), and Gross s.n. (pH), Lincoln Co., Ontario (12 4.5 mm). It has not been determined, by attempting artificial crosses between the two varieties, whether genetic barriers to hybridization exist. The rarity of hybrids is probably conditioned by the strength of the autogamy in both taxa; in addition, their sympatry is presumably a recent phenomenon. Along the Outer Banks of North Carolina Cakile edentula ssp. edentula gives way southward to C. edentula ssp. Harperi. A small sample (col- lection 40) and a mass sample (collection 41) from two populations in this region, when analyzed for seed glucosinolates, exhibited intermediate chemical profiles which may have resulted from hybridization (see sec- tion on biochemical systematics). Since synthesized F, hybrids were intermediate for seed glucosinolate composition, some support for hybrid- ization in natural populations is adduced. The herbarium record indi- cates that hybrids are not common, however. Despite the absence of genetic barriers between the two subspecies, hybridization is, I think, restricted because of the presence of autogamy in both taxa. Herbarium collections indicate that on the Atlantic Coast of Florida south to St. Lucie County Cakile edentula ssp. Harperi and C. constricta occasionally grow in mixed populations. Their sympatry is apparently not extensive. Five populations sampled in this region were found to be homo- geneous, showing no evidence of hybridization. A few herbarium speci- mens appear to be hybrids: Creager 41 7 (ras), Duval Co. (fruit 19 mm long) and Arnold s.n. (FLas), St. Johns Co. (23.5 mm). In Volusia County and southward C. lanceolata ssp. fusiformis occurs and becomes the predominant sea rocket in Dade, Monroe, and Collier Counties. It grows with C. constricta in Indian River County (Palmer 12 & 13, cH) and perhaps in adjacent counties and may hybridize sporadically with that species (Schallert 3997, smu, from Brevard Co.). Moreover, typical C. lanceolata occurs sporadically in the Florida Keys and north to Mar- tin County although apparently never forming extensive populations. 76 JAMES E. RODMAN Hybrids between it and C. lanceolata ssp. fusiformis were reported by Patman (1962) who stated that the hybrid “demonstrates many inter- mediate characters.” However, the two subspecies are not markedly dis- tinct, and hybrids between them would not be conspicuous. Sporadic collections in the Florida Keys and north to Pinellas County along the Gulf of Mexico approach C. lanceolata ssp. alacranensis. Millspaugh (1900) identified a specimen from Palm Beach County as a hybrid between “C. alacranensis” and “C. aequalis” (typical C. lanceolata) and remarked that such a hybrid was “a highly possible result, as the fruit of C. alacranensis could reach that locality on the current of the Gulf Stream which sweeps the shores of the Alacrans on its way to the Florida Keys.” The specimen (Webber 243, mo) is C. lanceolata ssp. fusiformis which is indeed intermediate in fruit morphology between C. lanceolata ssp. alacranensis and the typical subspecies (Plate 1). Cakile lanceolata ssp. fusiformis may have arisen as a hybrid between C. lanceolata ssp. alacranensis and the typical subspecies, most likely in the area of the Gulf of Honduras from where it could be transported to the Florida Keys by the Gulf Stream. Its presence in Florida prior to the last glaciation is problematic since most if not all of southern Florida was periodically inundated during the Pleistocene (Flint 1971). Plants of C. lanceolata ssp. fusiformis breed true for floral and fruit characters under cultivation but show considerable variation in leaf form, from nearly entire to deeply R-62: epee Xone O 6 3 ep ot ) — abaenipilaccil arene Pao E Ee ‘ “ ieee .: O if Sig : i 16 eo 2 min fruit length be ie ee i... THE GENUS CAKILE ae pinnatifid. On the Gulf Coast from Lee to Pinellas counties, Florida, C. lanceolata ssp. fusiformis occasionally grows with the more common C. lanceolata ssp. pseudoconstricta, and hybrids may be expected between them. Peninsular Florida is a great meeting ground for sea rockets which presumably differentiated to varying degrees in other areas of the Gulf of Mexico and Caribbean and which subsequently gained a foothold on the more recently emergent Florida Platform. Repeated dispersals and hybridizations have blurred morphological differences which may not have been sharply discontinuous initially and have created a confus- ing array of forms reflected by the conflicting taxonomies of Millspaugh (1900), Small (1933), and Pobedimova (1964). Hybridization also appears to contribute to the variability found in populations in Louisiana and eastern Texas where Cakile constricta and C. geniculata often grow sympatrically. Figures 7 and 8 present histo- grams of fruit length for 50-plant population samples of collections 62 from Grand Isle, Louisiana, and 64 from Brazoria County, Texas. These collections comprise a broad range of plants intermediate between the average values for fruit length for C. constricta (arrow 1) and C. genicu- lata (arrow 2). As noted in the section on morphology and anatomy, the difference in fruit length between these species is statistically signif- icant; in the absence of hybridization, therefore, one would expect a bimodal distribution for fruit length in mixed populations of the two. R-64: 0) @ 107 | | number of plants IN cee 2 O =, ' 1G. 21 26mm fruit length Fic. 8, Frequency histogram of fruit length in population sample R-64 (Brazoria Co., Texas). Arrow 1. Average fruit length of Cakile constricta; Arrow 2. Average fruit length of C. geniculata. 78 JAMES E. RODMAN The observed distribution strongly suggests natural hybridization. Fur- ther west along the Texas coast, C. geniculata frequently grows inter- mixed with C. lanceolata ssp. pseudoconstricta, and a few herbarium collections (e.g., Whitehouse 21169, smu, from Aransas Co.) suggest hybridization between the two. The absence of genetic barriers to hybridization combined with great dispersibility have created a complex array of forms in the genus Cakile, especially on the shores of the Caribbean and the Gulf of Mexico. The present taxonomy surely oversimplifies a dynamic situation, one which is not likely to stabilize until the appearance of strong genetic or geo- graphic barriers to intermingling. However, strong selective pressures may actually exist against genetic barriers in strand plants as Wilson (1965) suggested in a theoretical context. Colonizing plants which allow speciation to occur in effect create their own competitors; genetic differ- entiation may proceed to a point where hybridization results in sterile or stunted offspring, and consequently, effective population size and ranges are reduced. Despite the evident differentiation in sea rockets, therefore, there may be no selective advantage to genetic isolating mechanisms; since these plants are self-compatible, adaptive gene com- plexes can be preserved by inbreeding. Complex patterns of variability arising in part through frequent hybridizations may, therefore, remain an essential manifestation of the biology of the strand species of Cakile. HABITAT AND RANGE The species of Cakile, with one exception, are strand plants of the North Atlantic Ocean and adjoining bodies of water: the Baltic, North, and Barents seas north to Spitsbergen; the Mediterranean and Black seas; the Caribbean and Gulf of Mexico; and the Great Lakes of North America. The exception, C. arabica, is endemic to desert regions in a broad are around the Persian Gulf but does not occur on the shores of the Gulf (contra Hultén 1945). Two strand species, C. edentula and C. maritima, have become widely naturalized along the Pacific Coast of North America (Barbour & Rodman 1970) and along much of the west- ern, southern, and eastern shores of Australia. Sporadic collections outside the native ranges of these and other species undoubtedly represent cases of introduction by man in ships’ ballast. Plants of C. edentula, for exam- ple, have been collected in the Azores and in New Zealand, and C. mari- tima has been found near major ports of eastern North America, Uruguay, Argentina, and New Caledonia. The ephemeral occurrence of members of this genus as ballast weeds may be more widespread than is presently documented by herbarium collections. Where native and naturalized, the maritime species of Cakile typically grow on sandy beaches just beyond the reach of high tides in the pioneer zone of beach vegetation. The sea rockets commonly grow in thin, linear belts paralleling the shore, often in nearly pure stands at middle and ‘ THE GENUS CAKILE 79 northern latitudes. They are frequently the vascular plants growing closest to the sea. The plants of the fresh-water Great Lakes occupy an analogous habitat, often growing, as Cowles (1899) observed, just above the influence of summer waves in nearly pure linear stands along the shore. Where dunes are developed, occasional sea rockets grow among the grass- and shrub-dominated dune vegetation though never as extensively as along the foredune above high tides. Rare, isolated plants have been collected in coastal salt marshes and along roadsides near a_ coast. The strand habitat, called the middle beach by Cowles (1899), the storm beach by Johnson and York (1915), and the driftline by Chapman (1964), can be defined as the strip of shore above the influence of sum- mer or tidal waves but washed by winter storm waves and thus corre- sponds to the berm of geologists (Bascom 1960; Schuberth 1970). Physical factors of the strand environment have been summarized by a number of workers (Cowles 1899, in a classic paper on the Lake Michigan sand dunes, Oosting 1954, Chapman 1964). For Cakile maritima in California, Barbour (1970a) considered the important factors to be saline ground water, shifting sand, dry surface soil, salt spray, high winds, and low soil nitrogen. The poor water-retaining capacity of sand, high daytime tempera- tures, high winds, and high insolation combine to create an “edaphic desert,” in Salisbury’s words (1952), on the strand. The succulent nature of sea rockets has been interpreted as an adaptation for water storage in this xeric environment (Martin 1959) and alternatively as a response to salt spray (Boyce 1954) and to accumulation of salts from the sand (Uphof 1941). Controversy also exists as to whether the soil water in which sea rockets grow is saline (Johnson & York 1915; Barbour 1970a, who reported 450-1800 ppm. of soluble salts from strand soils near Cakile) or fresh (Kearney 1904; Boyce 1954; Martin 1959), becoming salty only when storm waves wash over the beach (Oosting 1954). The succulent character of the foliage appears to be genetically controlled since greenhouse and growth chamber plants, watered with 4-strength Hoagland’s solution, retain this feature. Because he found seed germina- tion and seedling growth to be inhibited by low salt concentrations, Barbour (1970c) termed Cakile an “intolerant halophyte,” which he distinguished from facultative and obligate halophytes. In laboratory experiments on C. maritima, Barbour (1970a) found the seedlings to be highly tolerant of salt spray, and from experiments on plants in situ, Martin (1959) stated that C. edentula was “completely tolerant to wind borne salt spray.” These results weaken Veldkamp’s (1971) assertion that sea rockets manifest no “typical special adaptations to the sandy beach.” He also contested Barbour’s (1970a) statement that low soil nitrogen characterizes the strand habitat and pointed out that sea rockets often grow over or near old driftlines where decaying organic debris 80 JAMES E. RODMAN would enrich the soil. Barbour (1972) later published measurements of soil nitrogen made on the strand near Bodega Bay, California, which indicated low levels in the areas he sampled. Johnson and York (1915) stated without supporting evidence that C. edentula requires a well- drained soil, and Salisbury (1952) cited unpublished research demon- strating the poor growth of C. maritima in clay soils. The sand in which sea rockets grow presents challenges as well, however, since young seed- lings may be buried in drifting sand and both young and old plants may be severely abraded by blowing sand particles. Older plants and clumps of seedlings frequently form mounds of sand, and this sand accumulation and the humus resulting from decay contribute to the initial formation of dune ridges and eventually of dune systems (Cowles 1899; Harshberger 1901; Salisbury 1952; Chapman 1964). Cakile is often among the first colonizers of blowouts in dunes (Erskine 1960) and of beaches recently swept by storms (Stoddart 1963, 1969). Thus, the sea rockets play a valuable though minor role in the reclamation of land at the edge of the sea. Seed germination and its adaptive significance have also been investi- gated by a number of researchers. Becker (1912) and Barbour (1970a) reported similarly high germinability for seeds from shelled upper and lower fruit segments which had been recently collected. I also observed high rates of germination in greenhouse and growth-chamber collections and found that germinability was enhanced by nicking the seeds lightly to break the seed coat. The longevity of Cakile seeds has not been studied in detail although Binet’s (1961) work on C. maritima suggests that the germinability of mature seeds increases over a period of several months and then declines. Ewart (1908) reported that a 67-year-old seed lot of C. maritima var. americana (= C. edentula var. edentula) and a 56 year-old seed lot of C. maritima var. maritima exhibited no germination. A very low rate of germination is observed when intact fruit segments are sown (Binet 1961; Barbour 1970a). Binet postulated seed dormancy in Cakile arising from a double inhibition: first, by the mature pericarp, an essentially mechanical inhibition resulting from its postulated imperme- ability to water, and second, by the seed coat, possibly a chemical inhibi- tion although the nature of the inhibitor was not determined. Barbour (1972) reported that upper fruit segments which had been soaked in sea water had higher rates of germination than fruit segments that were not. He hypothesized that the sea water had weakened the pericarp and/ or dissolved the chemical inhibitors in the seed coat. Furthermore, he suggested that such exposure to sea water imitated the natural dispersal conditions of upper fruit segments of Cakile. The botanical literature has many references to the dispersal of Cakile fruits by water but I have found no references to actual observations of the fruits of Cakile at sea. Heatwole and Levins (1972) reported whole plants of C. lanceolata floating at sea near Puerto Rico; these were THE GENUS CAKILE 81 green and, if in fruit, probably immature. Howard (1950) also men- tioned that the whole plant of C. lanceolata is carried by sea currents, and he stated that land crabs dispersed the fleshy (immature ) fruits of this species in the Bahamas. The fruit dimorphism of Cakile presumably reflects two different modes of dispersal ( Millspaugh 1900, 1916; Zohary, cited in van der Pijl 1969; Harper, Lovell & Moore 1970): the upper, deciduous segments may be blown about the beach until lodged in the sand or carried out to sea and transported while the lower segments remain attached to the dead plant, eventually to be buried locally in the sand. The latter mode would explain the frequent occur- rence of clumps of seedlings on small sand mounds observed during field studies and which Barbour (1970d, 1972) reported on the California strand. Dead plants may also be uprooted and blown about like tumble- weeds on the beach (Cowles 1899; Bowman 1918). The sea rockets thus appear adapted both to short-distance dispersal along the sandy beach and to long-distance dispersal at sea. Fruit dispersal in the desert species, C. arabica, has not been studied, but it is probable that the upper fruit segments of this species are wind-dispersed. The buoyancy of the upper fruit segments and the viability of the enclosed seeds after exposure to salt water are two striking examples of Cakile’s adaptation to the strand habitat. Experimental studies of this aspect of the ecology of sea rockets are not numerous, however. The most extensive study was reported by the naturalist H. B. Guppy (1906, 1917). My own studies indicate that considerable variation in the capacity for floating exists within a species and, therefore, that the results from testing only a few fruits may not provide an accurate assessment of the potential for water dispersal. Figure 9 presents the results of my studies on flotation of the upper fruit segments and viability of the seeds following exposure to salt water in five species of Cakile. For these tests, 100-110 upper segments for each collection were first washed for five minutes in a commercial bleach, “Clorox” (to inhibit fungal growth), rinsed well with water, then placed in gallon jars of sea water (collected from Boston Harbor). The jars were shaken once daily (to simulate some wave action) and counts made each day of the number of floating segments for a period of ten weeks. To test for seed viability, 10 (usually sunken ) fruit segments were removed at weekly intervals, rinsed well in water, and shelled; the seed coats were nicked to hasten imbibition and the seeds placed on wet filter paper in Petri dishes in the light at room tem- perature. The fruit segments used in these tests originated from popula- tion samples of 50 plants (in the case of collections 45, 58, 73, and 96) or from a few plants (collections 26 and B-103) or a single plant (65a). In collection R-s.n.-2, fruits from a few growth-chamber plants were used. In Figure 9 the percentage of upper segments still floating is indi- cated by broken lines and the percentage of seed germination by bars. It is evident that considerable variation exists in the buoyancy of the 82 JAMES E. RODMAN 2) hs 61 B40 10 wee edentula var. edentula constricta I ip ad ; O edentula var. 10 edentula ssp. edentula Har peri 100 - | r | + 100 F 7 \ 4 gables ai SU ia Sa 4 me T —— 1 aT eee F = Ue aap lanceolata ssp. 10 aritima lanceolata Ne ea r T T _ SE BE qT a geniculata 10 [ lanceolata ssp. 10 lanceolata Reietacieiesbiiuiitinaviniaineiiinsdedseiac.: Fic. 9. ae ee cae Percentage seed germination (bars) and p ntage broken lines). THE GENUS CAKILE 83 fruits. For most of the collections the fruit segments had all sunk in one or two weeks; fruits of C. edentula ssp. edentula (96) and C. lanceolata ssp. lanceolata (R-s.n.-2), however, were still floating at the end of 10 weeks when the tests were concluded. The maximum floating time which Guppy found in his tests was for a collection of C. lanceolata (ssp. lanceo- lata most likely) from Jamaica. It is interesting that the longest floating times were found in two of the most widely distributed taxa. The variation in buoyancy does not appear to be a simple function of the weight of the upper fruit segments or of the relative proportions by weight of the segment and the seed. This is evident from Table 28. Ridley (1930) stated that the variation in buoyancy was caused by vari- ation in the amount of “spongy” tissue in the fruit as well as variation in TABLE 28, AVERAGE WEIGHT OF UPPER FRUIT SEGMENTS (INCLUDING SEED ) AND RATIO OF WEIGHT OF FRUIT SEGMENT ( WITHOUT SEED ) TO WEIGHT OF SEED IN CAKILE. Collection Average weight (mg) Number Taxon of upper segment Ratio C. edentula ssp. edentula 96 31.4 2.0 C. constricta 58 11.5 22 C. edentula ssp. Harperi 45 46.0 27 C. geniculata 65a 30.6 3.9 C. lanceolata ssp. lanceolata B-103 20.5 4.2 C. lanceolata ssp. lanceolata R-s.n.-2 37.8 5.3 C. maritima ssp. maritima 73 44.2 ao the size of the suture on the articulating surface of the fruit segment. The importance of this latter feature had also been suggested by Guppy (1917). It is unknown whether these factors or others, such as the texture of the fruit or nature of the epidermis, account for the variation. Seed viability, indicated by the bars in Figure 9, was retained by nearly all the collections even after 10 weeks in salt water. A decline in viability is evident after the fourth week, but only in collection B-103 of Cakile lanceolata ssp. lanceolata were the seeds completely killed by the sea water. Over 100 years ago the French botanist Charles Martins (1857) tested the viability of seeds of C. maritima with an ingeniously constructed box enclosing the fruits and attached to a buoy in the harbor of Séte on the Mediterranean Coast of France. After 45 days in the water the fruits were removed (no comment was made on whether they were floating ) and planted; seeds from 13 of the 20 fruit segments of C. mari- tima tested (65%) germinated. The figure agrees well with the value I obtained for collection 73 of C. maritima ssp. maritima from California. Martins repeated his experiment on 20 upper fruit segments left for 93 days in the sea and found at the end of that time that all of the fruits had rotted, a result paralleling what I found in all the taxa tested. Diaspores must manifest three properties for dispersal by sea currents 84 JAMES E. RODMAN according to Lesko and Walker (1969): (1) the ability to float for several days; (2) the retention of seed viability after soaking in sea water; and (3) inhibition of germination by sea water. Regarding the third property (required to prevent germination at sea), I observed no germination from any of the fruits tested while they were in sea water. In addition, studies of the effects of sea water on germination have demonstrated that seeds of Cakile are inhibited by even low concentra- tions of salts (Barbour 1970a) and would probably not germinate under saline conditions. Thus, diaspores of Cakile possess the necessary charac- teristics for dispersal by sea currents and thereby manifest a sensitive adaptation to the strand environment. Biotic factors affecting Cakile in the strand habitat are poorly known. Wilson (1907, 1908) reported that plants of C. edentula serve as hosts for the fungus Albugo candida which infects many Cruciferae. Flowers preserved in the field in FAA frequently contained small (up to 1.5 mm), white (larval) to brown (adult) thrips. It appeared to be the same species of thrips from all 35 collections where it was found, from the Atlantic and Pacific coasts of North America, the Gulf of Mexico, Jamaica, and Bermuda. A conspicuous infestation by small black beetles was observed in a mixed population of C. edentula and C. maritima near Coos Bay, Oregon, where many plants of both species bore dam- aged leaves, buds, and young fruits. Barbour ( 1970d) hypothesized that herbivory (by unknown agents) may be a factor in the exclusion of Cakile from the dune communities back of the strand. Sea rockets have greatly extended their range within historical time as a result of their probable dispersal by man in ships’ ballast. Where introduced and naturalized, they occupy a habitat similar to their native one. This is strikingly evident for Cakile maritima ssp. maritima which is abundant along the California coast and also naturalized in western and southern Australia—areas with a Mediterranean climate (Raven 1971). The herbarium record (discussed below) indicates that the rapid migration of C. maritima in California and its replacement of C. edentula there (Barbour & Rodman 1970) is paralleled by a similar migration of C. maritima in Australia and its replacement in South Australia and Victoria of the previously introduced C. edentula. The aggressive occupation of the strand habitat in these regions by the Old World C. maritima suggests that a similar naturalization might well occur in two other areas with a Mediterranean environment: the coast of Chile and the Cape region of South Africa. The immigrant status of Cakile edentula in the Pacific region is not uncontested. Pobedimova (1963, 1964) accepted Heller’s (1907) species, C. californica, as native, and, furthermore, she stated that the Australian material treated by her as C, californica was somewhat distinctive and might prove to be an indigenous species. I consider the Australian and West Coast plants to fall within the range of variation of the Atlantic THE GENUS CAKILE 85 Coast C. edentula ssp. edentula var. edentula and find the herbarium and literature record a convincing proof of the migration and naturalization of the East Coast C. edentula. The case for the introduced status of Cakile edentula on the Pacific Coast of North America has been presented by Barbour and Rodman (1970) and was made earlier by Jepson (1936). Our argument was basically historical and took for granted the taxonomic identity of Pacific and Atlantic coast C. edentula, concurring with the opinion of Hitch- cock (1964). Here I hope to have presented sufficient evidence bearing on the morphology, glucosinolate chemistry, and reproductive biology of the two geographic entities to confirm Hitchcock’s judgment. A few additional comments will round out the story. The first accounts of Cakile by West Coast botanists are conflicting but suggest that the plants were initially limited in distribution although locally abundant in places. Behr (1888), for example, in the first pub- lished report of Cakile on the Pacific Coast, recorded it only from “Berkeley salt marshes.” Greene (1891), however, remarked that Cakile was “common along sandy beaches about the Bay of San Francisco at West Berkeley, Alameda, etc., also at Half Moon Bay; doubtless not rare on the coast and probably indigenous.” This last comment was disputed by Katharine Brandegee (1892) who treated C. edentula as an alien and observed that it occurred “about the shore sparingly from Black Point to the outlet of Lake Merced, and very abundant a short distance south of it.” LeRoy Abrams did not record Cakile in his early floras of Los Angeles and vicinity (1904, 1911, 1917) although it had been collected in nearby Orange County in 1907 (Davidson 1770, vs) and in Ventura County in 1916 (Eastwood 5024, cas). In their flora of the Northwest Coast, Piper and Beattie (1915) noted that C. edentula was “very rare along the seacoast.” Relevant comments on herbarium labels are scarce, but a revealing one on the collection of E. A. McGregor s.n. (ps) from Pacific County, Washington, in 1907 reads: “this specimen extends dist[ribution] from Golden Gate.” Eastern botanists of the period, for example Robinson (1895) and Millspaugh (1900) who monographed the genus, believed the sea rockets were introduced. But in 1922 Fernald published his new com- bination, Cakile edentula var. californica (Heller), and provided a plausible rationale for the existence of a native West Coast taxon by invoking the argument of a Pleistocene disjunction in a previously boreal distribution, No fossil record of Cakile exists apparently to test such a hypothesis, and the genus is absent from the Alaskan and Canadian Arctic and from Greenland [the report of C. edentula from Hudson Bay was stated by Scoggan (1957) to be a misidentification of a young plant of Chrysanthemum arcticum, the leaves of which superficially resemble those of the sea rocket]. Fernald’s idea undoubtedly influenced Hultén’s acceptance of the West Coast plants as native (1936, 1968) although he 86 JAMES E. RODMAN chose to treat them as C. edentula ssp. californica (1945). While accept- ing Fernald’s combination, Jepson (1925) maintained that the taxon was introduced and provided a brief historical argument for his opinion in 1936. That same year, George Neville Jones made the interesting obser- vation that “it is perhaps significant that the Makah Indians [of Wash- ington] have neither name nor use for the plant.” Heller (1907) and Fernald (1922) considered the West Coast plants to be somewhat more robust than their East Coast congeners, a characteristic which, if valid at all, is probably environmental (my growth-chamber collections showed no vegetative differences). Fernald (1922) remarked that the “fruits are in form essentially identical with those of many eastern specimens” but maintained that the two could be distinguished by the presence of small teeth on the articulating surface of the lower fruit segment of West Coast plants. This feature is encountered in East Coast material, however, and it varies within cultures from both coasts grown under uniform environmental conditions (Calder & T aylor 1968). Pobedimova (1964) conceded that it was difficult to distinguish East and West Coast plants and sometimes impossible without knowledge of their geographic provenance. While the plants of both coasts are virtually indistinguish- able, a perfect identity can not be expected; ecotypic differentiation or a founders’ effect could obscure their common origin. The tendency toward slightly larger petals in West Coast collections of C. edentula (see section on morphology and anatomy ) may reflect such phenomena. The relatively rapid replacement of Cakile edentula by C. maritima along the California coast, following the latter’s introduction around 1935, has been documented by Barbour and Rodman (1970). The phenomenon has also been adumbrated in a number of recent floristic works on Cali- fornia where the authors have noted the greater abundance of C. maritima —for example, for Marin County (Howell 1949, 1970) and for San Luis Obispo County (Hoover 1970). The argument for a reproductive advan- tage in C. maritima as an explanation for this replacement, proposed by Barbour and Rodman (1970) and Barbour (1970d), is developed here in the section on breeding systems. In Australia a remarkably parallel sequence of events occurred, with the introduction and naturalization of Cakile edentula from the Atlantic Coast of North America followed by the introduction and spread of the Old World C. maritima. Both species migrated widely and rapidly along the coast of temperate Australia, establishing extensive ranges which currently overlap on the shores of the states of South Australia and Victoria. In this zone, plants with intermediate characters suggest that some hybridization is occurring. The situation is a dynamic one, and a similar process of replacement of C. edentula by C. maritima, now nearing completion in California, appears to be underway on the Aus- tralian littoral. The first collectors on the continent found no Cakile on its shores; at THE GENUS CAKILE 87 least, Bentham did not record the genus in his F lora Australiensis ( Ben- tham & Mueller 1863) although he studied all the major collections of such early visitors as Banks and Solander, Robert Brown, James Drum- mond, Allan Cunningham, and J. D. Hooker. Localities where Cakile now grows had been visited (e.g., the famous Botany Bay area, Spencer Gulf, the mouth of the Swan River), and coastal plants with which Cakile is now associated were listed by Bentham: Spinifex hirsutus and S. longifolius, Scirpus nodosus, Sporobolus virginicus, Lepidosperma gladiatum, and Poa australis. The lack of an early record did not result, therefore, from a lack of botanical collecting in suitable habitats. The earliest collection of Cakile from Australia which I have seen is of C. edentula ssp. edentula var. edentula, made by Ferdinand von Mueller (s.n., GH,K,MEL,NY) on Phillip Island, Victoria, near the port of Mel- bourne, in 1863. It was this collection on which Hemsley (1879) based the first published report of Cakile in Australia (as C. maritima). And it was probably this collection or some other very early one of C. edentula to which Ridley (1930) referred when he wrote: “C. maritima [sic] was in Australia before 1867. It was no doubt transported there in sand bal- last.” The me sheet, #1004009, bears the comment (not apparently in Mueller’s hand): “known there wild since 20 years.” It is possible, there- fore, that C. edentula was introduced as early as the 1840’s near Mel- bourne, a major port, in ballast. American seal hunters from New Eng- land, enroute to Kangaroo Island, sailed the area repeatedly in the early 19th century (Dr. Elizabeth Shaw, personal communication; contra Sims 1968). A later specimen sent to Mueller by John Mason of Belfast, western Victoria (s.n., MEL #1004017), includes his letter of 18 Septem- ber 1876 which reads in part: “. . . its first appearance was about three years since and it continues to spread through the sand down to the margin of the sea and promises to cover in a short time the whole of the sand patches on the coast.” In 1870 W. Woolls sent Mueller a specimen of C. edentula (s.n., MeL #1004032) collected from Manly Beach near the port of Sydney, New South Wales. It has been collected repeatedly from the state since then and was reported (as C. maritima) by Moore and Betche (1893) and by Maiden and Betche (1916). In 1881 R. Tate collected C. edentula at D’Estrees Bay on Kangaroo Island (s.n., AD), which is apparently the first collection from the state of South Australia. Tate (1890) recorded the sea rocket in an early local flora of the state (again as C. maritima, in accord with prevailing custom), and Black (1909) also reported “C. maritima” (clearly C. edentula from his figure ) and commented that it was “. . . accepted by some botanists as native, but not recorded in Australia until 1869 [sic] when it was found on the shores of Port Phillip Bay and F rench Island.” In 1893 J. Bufton (s.n., MeL #1004025) and J. H. Maclaine (s.n.,MEL #1004022) extended its known range to Tasmania with collections from Port Davey and Clarke Island, respectively. In 1898 G. King (s-n., nsw) collected C. edentula 88 JAMES E. RODMAN on Lord Howe Island where it has apparently persisted. The earliest collection from Queensland I have seen is C. T. White 1680 (Brt,Nsw) from Stradbroke Island, near Brisbane, in 1922, and the earliest from New Zealand may be Philson, Doore, and Earle 340 (cH), collected on Stewart Island in 1935. [I have not seen material, however, from New Zealand herbaria; Pobedimova (1963) mapped C. maritima in New Zea- land, but I suspect this is in error for C. edentula.] Recently, C. edentula has penetrated into Western Australia; D. E. Symon 4685 (ap,Nsw) and Parsons 8 (ap), collected in 1967 from Eucla near the state border with South Australia, are the only specimens of this taxon I have seen from Western Australia. Sims (1968) reported C. edentula at Esperance in 1950, but I suspect that he mistook forms of C. maritima with poorly developed lateral horns on the lower fruit segments (common in Western Australia, e.g., Eichler 20300, av,cu, from Esperance ) for it. The herbari- um record thus reveals a progressive migration of C. edentula outward from the state of Victoria, east into New South Wales, to Lord Howe Island, and eventually north into Queensland, west to South Australia and recently into Western Australia, south to Tasmania and New Zea- land. Whether dispersed on sea currents or transported in ballast, C. edentula rapidly colonized southeastern Australia while it achieved a similar success, nearly contemporaneously, on the Pacific Coast of North America. In Western Australia where Cakile edentula was yet unknown, the second invader, C. maritima, appeared. The earliest collection I have seen is R. Helms s.n. (pertH) from Fremantle, a suburb of Perth, made in 1897, some 30 years later than the first collection of C. edentula in spreading horns . . .” in his flora of the state, and Black (1935) and Cleland and Black (1941) reported “C. maritima var. pinnatifida” from Kangaroo Island where only C. edentula had been found previously. Cakile maritima then migrated farther eastward into the state of Vic- THE GENUS CAKILE 89 toria. The collection Tagdell s.n. (MeL) from Beaumaris, near Melbourne, made in 1922, is the earliest I have seen of this species in Victoria. Again, the herbarium record (ca. 300 collections from Australia) suggests a pro- gressive migration: Western Australia, 1897; South Australia, 1918; Vic- toria, 1922. In his flora of Victoria, Ewart (1930) recorded a single species of sea rocket, referring to the then well-collected C. edentula as “|. C. maritima . . . the Victorian plant belong[ing] to the variety edentula (C. edentula Bigel.) . . . First recorded from Port Phillip and French Island in 1869.” He was apparently unaware that true C. maritima had invaded the state. I have seen no specimens of C. maritima from Tasmania; but Black (1948) reported it there, and Curtis (1956), after recording the presence of C. edentula, stated: “C. maritima Scop., native in Europe and introduced in temperate Australia, may also be found in Tasmania.” Migrating into South Australia and Victoria, Cakile maritima came into contact with the earlier introduction, C. edentula, and must have coexisted with it for some time and perhaps hybridized with it to a limited extent. The herbarium record indicates, however, that C. mari- tima has replaced C. edentula in these two states as it did in California. This phenomenon has not been recognized in the literature but it is clearly supported by the record of collections in Australian herbaria. Of the 71 collections made in the last three decades in South Australia which I have seen, only two are of C. edentula. Of the 19 collections I have seen from Victoria, made in the same period, only one is of C. edentula. As in Western Australia (Sauer 1965), C. maritima is common and appar- ently abundant along the South Australia littoral (Sims 1968). Cakile edentula, however, was probably never very common there. The two may still coexist in places; Eichler (1965) suggested that the two were hybridizing in South Australia, but the opportunities for this would appear to be decreasing. The replacement is most dramatic in Victoria where an early record of C. edentula over several decades has given way to an almost exclusive record of C. maritima. While I have not seen specimens of C. maritima from New South Wales, the species has recently migrated to that state. In a personal communication (1972), Dr. R. H. Groves of the Division of Plant Industry of the CSIRO wrote: “As far as we can understand, what is happening on the N.S.W. south coast at the moment is that the first plants of C. maritima have arrived (presumably from Victoria where it occurs commonly). Only C. edentula occurs on N.S.W. and Queensland beaches at present. Thus we may be witnessing the interesting ecological situation which occurred in California, on the basis of your herbarium evidence, some 35 years ago.” A competitive struggle analogous to that on the strand of the Pacific Northwest appears to be developing in New South Wales. It is surely an unstable situation, inviting more extensive field studies of the population dynamics of these two colonizing species. 90 JAMES E. RODMAN The same two species have been collected from other areas outside their native ranges, but nowhere else have they duplicated the success- ful colonization of such vast stretches of sandy shore. In eastern North America, an area of indigenous sea rockets, Cakile maritima has been collected from ballast ground near the ports of New York, Philadelphia, Norfolk, Mobile, and Wilmington. Dense populations may build up locally (Brown 1879), often with the native sea rocket intermixed (Ahles 1951), only to disappear after a few years. Cakile maritima has also been collected near major ports in Uruguay and Argentina (Hauman 1925; Rollins 1940), but whether appearing intermittently as a ballast weed or persisting in stable populations is unknown. The reverse invasion of the Old World by the North American C. edentula has also occurred. The species has been collected from the English Channel Islands ( Millspaugh 1900) and from the Azores (Millspaugh 1900), where, according to Guppy (1917), it was found as early as 1842. Reports of C. edentula in northern Scotland and the Outer Hebrides (Wilmott 1949; Allen 1952; Lousley 1953; Heslop-Harrison 1953a,b, 1954, 1955, 1956, 1958; Love & Love 1956) have resulted from mistaken identifications (Clapham, Tutin & Warburg 1962; Ball 1964a), understandable since Icelandic plants were usually treated as C. edentula at this time (Wilmott 1949; Heslop-Harrison 1953a). If not forms of C. maritima, as Ball (1964a) suggested, these plants in northern Scotland are likely to be of C. arctica, transported by sea currents from the Faeroes or Iceland. The occurrence of true C. edentula is sporadic in the Old World. Cakile edentula ssp. edentula var. edentula has also invaded the Great Lakes where the closely related C. edentula ssp. edentula var. lacustris is native. The two grow intermixed on the lake shores and, except for rare hybrids, maintain their identity. The more widespread native variety probably evolved on the shores of the ancestral Great Lakes following the Wisconsin glaciation while the rarer typical variety probably invaded in historical times, perhaps in ballast. The recognition of two taxa in the Great Lakes area dates back at least to Millspaugh (1900) who treated them as C. edentula and C. americana Nutt. Nuttall based his species concept on material from both the Atlantic Coast and the Great Lakes (from which material a lectotype should be designated, but I have not seen his specimens; and whether lectotypified by an Atlantic Coast plant or one from the Great Lakes, the name would remain in synonymy ys His name was widely used by those unaware of Bigelow’s earlier name for the East Coast species or by those who chose to distinguish, at specif- ic rank, between Northeastern plants with short- or long-beaked upper fruit segments. Beak length varies within Atlantic Coast C. edentula ssp. edentula but on the average is shorter than in Great Lakes material (see section on morphology and anatomy). Fernald (1922) also recognized two taxa in the Great Lakes region (the varieties accepted here), but his criterion for distinguishing the two—the presence of teeth on the articu- THE GENUS CAKILE 91 lating surface of the lower fruit segment—is unreliable. Of the 394 herbarium collections of Cakile studied from the Great Lakes, 30 (7.6%) were clearly C. edentula var. edentula, and four (1.0%) were probably hybrids. I have seen no collections of the typical variety from Lake Huron or Lake Superior, and the herbarium record suggests that it occurs only sporadically on the beaches of Lakes Michigan, Erie, and Ontario. Most of the collections (26) are from the southern end of Lake Michigan, usually in the vicinity of Chicago. Even here, however, typical C. edentula is less common than the native variety. Swink (1969), for example, stated: “Most of our material is referable to var. lacustris, although the type does occur . . .”; and in his “Flora of the Indiana Dunes,” Peattie (1930) remarked that “. . . the typical species is found elsewhere on the shores of Lake Michigan, but no specimens from the Indiana coast have been seen.” Deam (1940), likewise recorded only C. edentula var. lacustris; however, 1 have seen a few collections of typical C. edentula from the Indiana shore (Umbach 11744, wis; Pennell 6448, ny). Patman and IItis (1961) imply that it is more common along the Wisconsin shore of Lake Michigan, but relative to the native variety it appears to be much less abundant. The sporadic occurrence of Cakile edentula var. edentula in the Great Lakes region suggests that it is a recent introduction. It might have entered the area as a ballast plant following the opening of the Erie Canal in 1825. Unfortunately, I have encountered no specimens of either variety ante- dating that event. Nuttall’s Great Lakes material must have been var. lacustris; at least, his name, C. americana, was widely adopted for long- beaked specimens. The earliest collections from the Great Lakes I have seen, 1838, are of C. edentula var. lacustris (Wright s.n., MICH; Clinton S.n., F,MICH,MSC,Ny). However, the collection of I. A. Lapham (s.n.), made in 1841 at Milwaukee, includes both C. edentula var. lacustris (F, MEL,MO,NY,RSA,Wis) and C. edentula var. edentula (pu). It is by no means certain that these specimens represent the same gathering in the Same year. The early introduction of typical C. edentula at Milwaukee, however, might explain its more common occurrence along the Wisconsin shore. In 1843 Torrey cited a specimen of C. edentula var. edentula col- lected by P. D. Kneiskern (s.n., Ny) from Lake Ontario. The herbarium record, therefore, provides no clear indication for the recent migration of typical C. edentula into the Great Lakes region. Its status as an alien must remain an hypothesis, but one which is most consistent with the view that the var. lacustris evolved in the Great Lakes area as a geo- graphically isolated entity. EVOLUTION No fossils of Cakile are known (Pobedimova 1963). In the absence of a physical record of its evolutionary history, a phylogeny of the group can only be reconstructed by inferences drawn from structural and 92 JAMES E. RODMAN ancestral species C. fusiformis C. arabica C. lanceolata C. euxina C. californica C. geniculata C. maritima C. arctica C. alacranensis | ene eamgace aussi aohietnminetiiteeee et C. baltica C. edentula C. monosperma C. lapponica C. cubensis hci td C. lacustris FicurE 10. PHYLOGENETIC SCHEME OF CAKILE (PoBEDIMova 1963). THE GENUS CAKILE 93 functional similarities, interpreted within a geological framework rele- vant to the present distribution of the species and their presumed age. Such a reconstruction is attempted here by applying the evolutionary mechanism of speciation by geographic isolation to the biology of Cakile as now understood. In her monograph of the genus, Pobedimova (1963) discussed the probable evolution of the species and summarized her ideas in a phylo- genetic “scheme” reproduced here as Figure 10. The genus was con- sidered to have been derived monophyletically from Erucaria, with an ancestral species referred to as Tethyan in age and distribution. This presumably widespread Tethyan ancestor diverged into two archaic species, C. arabica in the Old World and C. fusiformis in the New, and these eventually gave rise to an additional 13 species, probably since the Pliocene. Speciation by geographic isolation was emphasized as the primary mode of differentiation in Cakile. This was manifested in the Old World by the evolution of C. euxina in the Black Sea, C. maritima in the Mediterranean, C. baltica in the Baltic Sea, and C. monosperma in Western Europe, in that chronological sequence. In the, Americas the tropical section Xiphocakile sensu Pobedimova (1964: C. fusiformis, C. lanceolata, C. geniculata, C. alacranensis, and C. cubensis) was stated to have evolved in response to fluctuating sea levels. From this primitive section, the northern section Integrifoliae sensu Pobedimova (C. cali- fornica, C. arctica, C. edentula, C. lapponica, and C. lacustris) evolved, with a secondary penetration of the Old World by the northern C. arctica sensu Pobedimova (a “pseudorelict” in the White Sea) and by C._Yap- ponica, the more recent migrant. In agreement with Fernald (1922), Pobedimova considered C. californica and C. edentula as disjunct de- scendants of a boreal-American taxon. Cakile lacustris was presumed to have evolved in the Great Lakes region after a marine invasion of that area in the Quaternary. In many respects my views on the probable phylogeny of the sea rockets coincide with those of Pobedimova; differ- ences in our taxonomies, however, necessitate a different evolutionary reconstruction for several taxa. The tribe Brassiceae, with which Cakile is associated and in which it most likely evolved (see section on generic affinities), is native exclu- sively to the Old World, primarily to the Mediterranean region, with a distinctive dispersal ecology. The inland species of Cakile, C. arabica, links the strand taxa with genera of arid habitats, Erucaria in particular, which represent the phyletic stock from which Cakile evolved. Morpho- logically, C. arabica is the only species in the genus possessing charac- teristically pubescent leaves, a feature often encountered in other genera of Brassiceae. This desert species presumably gave rise to a strand spe- cies in the Mediterranean area (the shores of the ancient Tethys Sea) 94 JAMES E. RODMAN which inherited several features preadapting it to a coastal environment: an annual growth cycle, fleshy leaves, and a corky fruit capable of floating. In this new habitat Cakile expanded in range and speciated in regions where a degree of genetic isolation was imposed by geography, becoming the only crucifer and one of the few vascular plants growing widely on sandy beaches in the northern temperate zone. Pobedimova (1963) proposed that Cakile arabica is secondarily a desert plant and was primarily a coastal species now relict following the retreat of Tethyan water from western Asia. The idea is plausible but an unnecessary complication. Cakile arabica does not grow on the strand of the Persian Gulf (contra Hultén 1945) nor on the shores of the Red Sea or Indian Ocean where it might be expected had it once been a strand plant of the eastern Tethys. Its very small fruits would appear to be poorly adapted to water dispersal. There is no compelling reason to assume that the present habitat of this species is anything other than its original one. An age can be postulated for the strand species of Cakile based on dis- tributional evidence of a negative sort. In the Old World, Cakile is absent from the shores of the Persian Gulf and Indian Ocean where the ancient Tethys extended into early Miocene time (Kummel 1970). If present on the strand of the Miocene Tethys Sea, sea rockets should occur today on the coasts of the Indian Ocean and Persian Gulf. In the New World, Cakile is absent from the Pacific Coast of Central and South America and, except where naturalized, absent also from the Pacific Coast of North America. A water passage between the Caribbean and Pacific existed, however, into the late Pliocene when the Isthmus of Panama arose (Dunbar & Waage 1969; Kummel 1970). If present in the Caribbean before late Pliocene times, Cakile should have been transported by the currents to the Pacific Coast, where in fact two species have become widely naturalized. From its present natural distribution, therefore, I would argue that Cakile is no older than early Miocene and possibly only late Pliocene in origin. The age of the Cruciferae as a family is not well established, Couper (1960) reported fossil pollen of the family from Cretaceous deposits in New Zealand, and Becker (1961) described a fruit impression of Thlaspi primaevum from the late Oligocene Ruby River shales of Montana. Thus the Cruciferae had apparently evolved by Miocene times and possibly much earlier. For the Old World Cakile euxina (treated here as C. maritima ssp. euxina) Pobedimova (1963) concluded that it must be of late Pliocene age or even younger since the species is absent from the Caspian Sea which was long united with the Black Sea during the Pliocene (Kummel 1970). At the beginning of the Pleistocene the Black Sea and the Mediter- fanean were joined (Kummel 1970), and at that time Cakile may have first entered the Black Sea. The connection with the Mediterranean was recurrently closed and opened during the Pleistocene in response to THE GENUS CAKILE 95 fluctuating sea levels associated with glaciation (Frenzel 1968; Flint 1971), and the genetic isolation imposed on the Black Sea Cakile during periods of separation must have been sufficient to allow the differentia- tion of a distinct subspecies. At the last, Weichselian glacial maximum, ca. 18,000 years B.P., the Black Sea was separated from the Mediter- ranean (Flint 1971) and again united by a narrow channel with the Caspian Sea (Frenzel 1968; Flint 1971). This connection was appar- ently not wide enough or long enough in duration to permit Cakile to migrate into the Caspian. Following the last glaciation, this channel was eliminated by crustal uplift, and the connection between the Black Sea and the Mediterranean was reestablished. Only recently (ca. 3000-7000 years B.P.) has the Black Sea changed from freshwater to saltwater con- ditions under the influence of this reunion (Degens, Watson & Remsen 1970). If pre-Weichselian in age, therefore, C. maritima ssp. euxina has experienced a change in habitat from freshwater to saltwater strand— the reverse of what must have occurred as sea rockets from the Atlantic Coast of North America colonized the shores of the freshwater Great Lakes following the Wisconsin glaciation. Pleistocene glaciation must also have influenced the evolution of Cakile maritima ssp. baltica, growing presently on the shores of the Baltic Sea and the Skaggerak and Kattegat channels. At the last glacial maximum the Baltic Sea region was completely covered by the Scan- dinavian Ice Sheet, which in conjunction with the British Isles Ice Sheet also covered great portions of the North Sea (F renzel 1968; Flint 1971). With the subsequent melting of the glacial ice, a cold lake (Baltic Ice Lake) formed ca. 10,000-12,000 years B.P., which eventually united (as the Yoldia Sea, ca. 9000 years B.P.) with the contemporaneously developing North Sea (Flint 1971). It is probable that Cakile entered the ancestral Baltic Sea area at this time. Crustal uplift, in the wake of the ice sheet's retreat, then isolated this body of water (the Ancylus Lake) for roughly 2000 years, ca. 7000-9000 years B.P., during which time a distinct sub- species could have diverged under genetic isolation from typical C. maritima. Union with the North Sea was re-established ca. 7000 years B.P. (the Littorina Sea), continuing to the present via the Skaggerak and Kattegat Channels where plants intermediate between C. maritima ssp. baltica and typical C. maritima now occur. Pleistocene glaciation must also have eliminated habitats in Iceland, the Faeroes, and northern Scandinavia where Cakile arctica now grows. Glacial refugia in this area (Steindorsson 1963) may have provided a suitable habitat for sea rockets, but their absence from Greenland argues against the adaptation of Cakile to such cold conditions. The possibility of the origin of progenitors of C. arctica from North America by long-distance dispersal via the Gulf Stream (Love 1963) can not be easily discounted. The Gulf Stream system may be geologically old; Schuchert (1935) stated that it developed in late Miocene-early Pliocene 96 JAMES E. RODMAN time. Cakile arctica, however, exhibits no preponderant morphological similarities to the American taxa. Pobedimova (1963) maintained that the subarctic sea rockets were derived from North American Cakile because they share similar fleshy, + entire leaves. However, plants of C. maritima ssp. maritima with fleshy, -+ entire leaves occur in Western Europe and the Mediterranean region, and in floral morphology C. arctica is more similar to C. maritima than to the American C. edentula. Cakile arctica could have evolved from Western European C. maritima after the last glacial maximum following the chance dispersal of propagules to Iceland and differentiation there in response to insular isolation; popula- tions in the Faeroes and northern Scandinavia could then have been established by disseminules carried by the major currents in this region. The origin of C. arctica must in any case have been from southern stocks, but whether American or European cannot be determined on the basis of present information. Since I consider the sea rockets to be monophyletic and derived from Old World ancestors, their presence in the Americas raises prob- lems of origin and dispersal. Unless one assumes an origin in the early Tertiary when North America and Europe were physically closer to one another than now (Pitman & Talwani 1972; Phillips & Forsyth 1972) long-distance dispersal must be invoked to account for the entry of Cakile into the New World. A chance dispersal might have occurred via the Canaries Current. Pobedimova (1963) argued that C. fusiformis (treated here as C. lanceolata ssp. fusiformis) was the ancestral stock of the American taxa, and she linked this taxon with C. arabica because of the similarly pinnatifid leaves. She also interpreted the occasional occur- rence of incumbent seeds in C. fusiformis as indicating a relationship with Erucaria (which, however, commonly has conduplicate cotyledons; cf. Schulz 1936). In addition, Pobedimova described the distribution of the two species as relictual from an ancient Tethyan pattern and, citing Guppy as authority, referred to the presence of C. fusiformis in the Azores as an intermediate station linking their present ranges. This last idea, however, involved a misreading of Guppy (1917) who wrote of the occurrence of C. edentula in the Azores as a probable ballast weed. Moreover, her morphological argument is insupportable. Pinnatifid leaves are encountered sporadically in C. lanceolata ssp. lanceolata, more com- monly in C. lanceolata ssp. fusiformis, and characteristically in C. lanceo- lata ssp. pseudoconstricta as well as in C. maritima. Incumbent cotyle- dons are found in all the taxa of Cakile in a small proportion of any large population sample. Since the present range of C. lanceolata ssp. fusiformis encompasses areas which were inundated during the Pleistocene, it seems plausible to regard this taxon as recent in origin; it may have arisen as a hybrid between the Antillean C. lanceolata ssp. lanceolata and the Yucatan endemic C. lanceolata ssp. alacranensis. A pivotal role for C. lanceolata ssp. fusiformis in the evolution of the American taxa THE GENUS CAKILE 97 is not indicated by the morphological and distributional data. No existing taxon, in fact, provides a clear intermediate between the New World and Old World sea rockets. From considerations of the patterns of seed glucosinolate composition, however, a guess can be made concern- ing a probable route of entry for Cakile into the New World. The seed glucosinolate compositions of Cakile constitute three general chemical patterns: Mediterranean, Amphi-Atlantic, and Caribbean (to use the descriptive terms suggested by Dr. Martin Ettlinger in a personal communication). The Mediterranean pattern is found in C. maritima ssp. maritima on the Pacific Coast of North America and in Australia (see section on biochemical systematics) and in the Mediterranean sea rockets (unpublished results). It is distinguished by high amounts of sec-butyl and isopropyl glucosinolates and a series of methylthioalky] glucosinolates. The Amphi-Atlantic pattern, distinguished by high amounts of sec-butyl and/or allyl glucosinolates, is encountered in C. edentula ssp. edentula and C. maritima ssp. baltica (as previously noted ) as well as in Western European sea rockets (unpublished results). The Caribbean pattern, characterized by the predominance of 3-butenylglu- cosinolate, is found in C. edentula ssp. Harperi and in the sea rockets of the West Indies and Gulf of Mexico. The presence of a similar glucosino- late composition in Western European and North American sea rockets is an affinity possibly arising from common descent. The progenitors of the American taxa may have originated from the coasts of Western Europe and have colonized first the Atlantic Coast of North America. The introduction, by chance long-distance dispersal, presumably involved only a single phyletic group. This founder group then colonized south- ward and differentiated into the several taxa now found in the West Indies and Gulf of Mexico. The change to the Caribbean pattern of seed glucosinolates apparently occurred early in the evolution of the American taxa since it characterizes C. edentula ssp. Harperi on the Atlantic Coast south of Cape Hatteras. My bias in this speculation for an original locus of introduction on the Atlantic Coast rather than in the West Indies also arises from a consideration of the annual growth cycle of Cakile. The European sea rockets, as with many other strand plants of that area (Chapman 1964), are annuals. Such a growth cycle is adaptive on the Atlantic Coast of North America where seasonality is pronounced. An annual growth cycle in the strand plants of the subtropical West Indies is anomalous, however, as Sauer (1967, 1972) has pointed out. The move- ment of Cakile into the West Indies may be a quite recent ( ?Pleistocene ) event. This would explain its absence from the Pacific Coast of Central America despite the presence of a sea passage enduring until late Plio- cene time. From an original locus of colonization on the Atlantic Coast, migra- tions occurred southward into the islands of the Antilles and perhaps, to some degree, northward. On the coast of the Yucatan Peninsula or on the 98 JAMES E. RODMAN islands formed in this area during periods of higher sea level in the Pleistocene (Wilhelm & Ewing 1972), Cakile lanceolata ssp. alacranensis became differentiated. Cakile geniculata evolved in the northwestern corner of the Gulf of Mexico, presumably then in relative isolation, al- though it is widely sympatric today with C. constricta in the east and C. lanceolata ssp. pseudoconstricta in the west. East of the Mississippi Delta, C. constricta evolved along the coast of the southeastern United States or on the northernmost islands of the Florida Platform during Pleistocene time. Since the Oligocene and particularly during the Pleisto- cene, the Florida Peninsula has been emergent not as a continuous landmass but as a series of low, scattered, calcareous islands, sensitive to fluctuating sea levels (MacNeil 1950; Kummel 1970). Cakile lanceo- lata ssp. pseudoconstricta may have differentiated on the southernmost of these islands, south of C. constricta. With the emergence of the Florida Peninsula after the last glaciation the range of C. constricta was inter- rupted and the gap in southern Florida was occupied by C. lanceolata ssp. pseudoconstricta and C. lanceolata ssp. fusiformis. The latter is probably a recent introduction via the Gulf Stream from the eastern shores of the Yucatan Peninsula or the Bay of Honduras. Fosberg (1963) considered it remarkable that minor but very real geographic variation occurred in widespread tropical strand species, presumably plants with great dispersibility thus promoting rapid gene exchange between popu- lations. Geographic variation in the Gulf and West Indies sea rockets is no less real, though often minor, and it would undoubtedly be more comprehensible were more known of the shifting mosaic of land and sea in this region. The confused historical geology of the Caribbean-Gulf of Mexico region (Schuchert 1935; Woodring 1954; Graham & Jarzen 1969; Wilhelm & Ewing 1972) dampens any speculation beyond the relatively crude reconstruction attempted here. Along the Atlantic Coast of North America, Cakile edentula evolved into a northern race, ssp. edentula, and a southern race, ssp. Harperi, both of which are presently sympatric on the Outer Banks of North Carolina. The transition between the two subspecies, occurring in the vicinity of Cape Hatteras, correlates with a major biogeographic and geomorphological boundary (Burk 1968; Milliman, Pilkey & Ross 1972). Ecological associates of the northern C. edentula ssp. edentula, such as the dune grass Ammophila breviligulata and Hudsonia tomentosa, also terminate their ranges on the Outer Banks. Seneca and Cooper ( 1971) and Seneca (1972) have shown that the transition from Ammophila breviligulata north of Cape Hatteras to Uniola paniculata southward involves a complex of interactions with temperature, water, and biotic factors in the environment. The evolution of two subspecies of Cakile north and south of Cape Hatteras may have similarly involved adjust- ments to some of these differing climatic and biotic conditions. While no obvious geographic barrier separates the two today, the environments of THE GENUS CAKILE 99 these subspecies are strongly influenced by two very different current systems of the North Atlantic Ocean, the warm Gulf Stream south of Cape Hatteras and the cold Labrador Current north of it (Cotter 1965). In the Hatteras offshore waters where they meet, they create a barrier to migration as effective as any landmass separating two bodies of water. Despite the clear potential for gene flow through hybridization, the integrity of each seems little affected by the other. Strong environmental selection or an absence of significant dispersal due to the opposing cur- rent systems, or both, may operate in maintaining the identity of these two subspecies. The range of Cakile edentula ssp. edentula must have been severely restricted by the Laurentide Ice Sheet about 20,000 years B.P. when it completely covered the Great Lakes Basin, the valley of the St. Lawrence River and the Gulf of the St. Lawrence, and the eastern coast of North America south to Long Island (Flint 1971). With the melting of the glaciers, coastal habitats were reopened to northward colonization by C. edentula. Moreover, the Great Lakes were formed after the last glacial maximum, about 15,000 years B.P. (Hough 1963). Cakile and other plants may have entered the Great Lakes region nearly this early by long-distance dispersal overland from the Atlantic Coast. More plausibly, however, Cakile could have colonized the region about 11,500- 9,500 years B.P. when a marine embayment known as the Champlain Sea extended from the Gulf of St. Lawrence to, or nearly to, old Lake Ontario (Hough 1958, 1963; Elson 1969; Dorr & Eschman 1970). This embayment would have provided a continuous route for the migration of strand species from the Atlantic Coast into the Great Lakes Basin. Isostatic uplift in the St. Lawrence River valley forced the retreat of these marine waters (Hough 1963), which could then have isolated the Great Lakes Cakile from its maritime congener. This geographic isolation may then have enforced the differentiation of a new taxon, C. edentula ssp. edentula var. lacustris. Seneca and Cooper (1971) demonstrated ecotypic differences in Ammophila breviligulata from the Great Lakes strand and the Atlantic Coast of North Carolina. Adapta- tion to the freshwater littoral has probably occurred in C. edentula var. lacustris as well although the differences in fruit and seed morphology distinguishing it from C. edentula var. edentula do not appear to be of adaptive significance. Based on a limited study, Chrysler (1904) reported that the Great Lakes Cakile has less succulent leaves than Atlantic Coast C. edentula, a phenomenon often encountered in plants which shift from a saltwater to freshwater strand habitat (Boyce 1954). The affinities be- tween several Atlantic Coast and Great Lakes plants were early pointed out by John Torrey (1843). Knowing of the occurrence of marine clay deposits near Lake Ontario, he hypothesized that the disjunctions arose following the migration of these plants after the melting of the ice sheet when the Great Lakes were a marine body of water united to the Atlantic 100 JAMES E. RODMAN Ocean via the St. Lawrence. Drummond (1867) and Hitchcock (1871) also supported this interpretation. Increased knowledge of the geology of the Great Lakes negated this view of an extensive marine stage, however. As the concept of the Champlain Sea embayment gained acceptance, Marie-Victorin (1916) and Svenson (1927) proposed that many of the disjuncts used this as a migratory route. This route is the most plausible one for those plants primarily dispersed over many miles by water. It may not have served as such for all the strand disjuncts between the Great Lakes and the Atlantic Coast (e.g., Ammophila breviligulata, Euphorbia polygonifolia, Hudsonia tomentosa), nor was it likely the only means of entry for the larger number of Atlantic Coastal Plain species disjunct in the Great Lakes basin (Peattie 1922), many of which could have been dispersed overland. Pleistocene conditions, particularly the general cooling induced by the glaciation, may also have influenced events in Bermuda. As noted in the section on biochemical systematics, my analyses of two Bermuda samples revealed a distinctive seed glucosinolate composition of the Amphi-Atlantic pattern. This pattern is characteristic of northern taxa of Cakile such as C. edentula ssp. edentula, C. arctica, and C. maritima ssp. baltica. Morphologically, these Bermuda plants had fleshier leaves and smaller fruits than is typical of C. lanceolata with which most of the Bermuda herbarium collections agree. It is possible that a distinct race evolved in Bermuda during cooler Pleistocene times and that this race possessed a characteristically northern glucosinolate composition, either because the populations were founded by migrants which possessed an Amphi-Atlantic pattern or because there was selection for this character under the cool conditions then prevailing. Recurrent introductions of C. lanceolata via the Gulf Stream, perhaps accelerating with post-Pleisto- cene warming, suppressed any further evolution and created the hetero- geneity in Bermuda material observed by Pobedimova (1963, 1964) and others. With some hesitation, F igure 11 is presented to summarize the phylo- geny of Cakile. Any scheme such as this grossly oversimplifies the argu- ments behind it which in the absence of a fossil record, must oversimplify a dynamic process. Throughout, an attempt has been made to find a plaus- ible background for geographic isolation to account for the observed differentiation in Cakile. Isolated bodies of water obviously provide a locus for speciation, but island settings, too, may lead to fragmented populations and reduced gene flow and thereby promote evolution. Insular evolution undoubtedly accounts for the origin of C. arctica and may well have generated most of the patterns of geographic variation in the Caribbean and Gulf of Mexico. Little more than speculation can be offered regarding the selective value, if any, of the characters distinguish- ing the taxa. Increased fruit length in C. arctica and C. lanceolata ssp. lanceolata may, for example, confer greater buoyancy and improve / C. geniculata c. constricta ce C, lanceolata ssp. alacranensis C, lanceolata ssp. fusiformis ancestral strand species ancestral American species ne THE GENUS CAKILE Erucaria (eee ? C. arabica ? C. edentula ssp. edentula C, edentu ssp. Harperi L. edentula var. lacustris, lanceolata LC, lanceolata a. it C. maritima C. maritima ssp. euxina C. maritima ssp. baltica c. arctica Figure 11. Proposed phylogeny of the genus Cakile. Read from top. 102 JAMES E. RODMAN dispersibility. Small flowers may be part of a syndrome for autogamy promoting high fruit set in seasonally restricted habitats. In some cases the differentiated features may be without adaptive value but result instead from random genetic drift. Ultimately, ecological work like that of Barbour (1972) and my own experiments on fruit flotation and seed viability may contribute to an understanding of natural selection in the evolution of Cakile. Economic UsE “There is no Beauty or Use in these Plants at present known, but they are preserved in Botanic Gardens for Variety. (Philip Miller, 1748 Notwithstanding the noted English gardeners comment, sea rockets have been utilized by man in a number of ways through the ages. Mark- graf (1963) reported that in the Old World Cakile was employed as an antiscorbutic, diuretic, and purgative in folk remedies. The leaves were eaten in salads, and plants were even cultivated on a small scale in France for this purpose (Pobedimova 1963). In the New World, Lin- naeus’ student Peter Kalm observed that in time of famine the Indians along the St. Lawrence River pounded the roots of “Bunias cakile” into powder which, mixed with other flours, was made into bread (Benson 1966 ). The colonial Europeans also learned to exploit the plants. As in the Old World, sea rockets were eaten for their antiscorbutic properties; and Small (1900) stated: “[T]he scurvy dispelling properties of this plant [Cakile lanceolata] have been tested and found effective.” From De la Sagra’s account (1853) one may surmise that sailors in the West Indies were not especially opposed to the remedy for scurvy since it usually consisted of “la raiz en infusion en vino blanco.” Leaves of sea rockets were likewise eaten in salads (Questel 1951; Marie-Victorin & Rolland- Germain 1969). On personal examination I found the young leaves to have a pleasant, mildly biting flavor, reminiscent of horseradish. The flavor is caused in part at least by the plant’s mustard oils. Fernald, Kinsey, and Rollins (1958) report an observation made by Harold St. John that Labrador fishermen often gather large quantities of the young plants of C. edentula to cook as a potherb. In the Bahamas, sea rockets are occasionally gathered as forage for the local burros, hogs, and goats (Howard 1950). A specimen of C. edentula collected in Grays Harbor County, Washington (Foster 1534, us), bears the charming comment on its label: “Cattle are very fond of it.” Undoubtedly the chief benefits which man derives from these plants are the indirect ones arising from their ecological role as pioneers in beach and dune successions. This aspect of the biology of Cakile is treated more fully in the section on habitat and range. THE GENUS CAKILE 103 TAXONOMIC TREATMENT Cakile Miller, The Gardeners Dictionary . . . Abridged, 4th Edition, G31 Plants annual or fortuitously perennial, herbaceous to occasionally suffrutescent, fleshy, glabrous to rarely pubescent. much-branched near the base, the branches deciduous, ovate, 3-5 mm long, 1.5-3 mm wide, margins hyaline, subcucullate at apex, slightly dimorphic, the laterals somewhat gibbous at the base and usually more acute at the apex; the outer upper surface often with sparse simple trichomes, etals 4 (rarely absent or present only as bristles), obovate or spatulate to distinctly un- the apiculate at apex, saggitate at base, the laterals dehiscing introrsely within the flower, th occasionally bilobed, capitate to retuse at apex of stylar beak. Immature fruit fleshy, green, becoming dry and corky, of 2 indehiscent segments, both falsely l-loculed and usually 1-seeded (occasionally aborted or 2- 3-seeded ), the septum a white papery- thin membrane appressed longitudinally to each seed, the upper beaked segment iculati conical to flattened in one plan base occasionally with a scarious margin, the articulating surface pits corresponding to the teeth on the lower segment. Seeds accumbent to occasionally obliquely incumbent, ellipsoid to narrowly oblong, flattened longitudinally, puncticulate to smo variously br or bronze, dimorphic, the distinctly geniculate. Cotyledons fleshy, + equal, lanceolate or oblanceolate. Pollen grains yellow, tricolpate, reticulate, spherical to oblate-spheroidal when turgid. CHROMOSOME NUMBER 2n=18. Lectotype species: Cakile maritima Scop. KEY TO THE SPECIES A. Leaves deeply and finely pinnatifid, usually pubescent; fruits < 4 mm wide, the articulating surface of the lower segment with 2 stout vertical, conical eo 1-2 mm hi : A. Leaves entire to deeply pinnatifid, never pubescent; fruits > 3 mm wide, articulating surface of the lower segment flat or with teeth 1 mm high or ies. 2 B B. Fruiting pedicels 5-10 mm long; fruits constricted at the articulation, > 20 mm long when both segments fertile; petals usually > 3 mm mide 5 2 S C. arctica. B. Fruiting pedicels usually < 5 mm long; fruits constricted or not at articulation, 13-31 mm long; petals 1.5-6 mm SHI i be ear C. C. Lower fruit segment with 2 opposite lateral horns or the sides prolonged upward into 2 sharp triangular wedges, concave between; petals 3-6 mm wide, lavender to a ce 1. C. 104 JAMES E. RODMAN C. Lower fruit segment without lateral horns and if prolonged upward, the wedges < 1.5 mm high; petals < 4.5 mm wide, often < 3 mm, white to lavender D. D. Infructescence usually > 2 dm long; leaves not gested fleshy; ‘plants usually sprawling; petals wally > 3 mm wide and w C. lanceolata. D. Infructescence usually 1-2 dm long; leaves a. ke Sia to compactly prostrate; petals usually < 3 mm wide, lavender to white. E. E. nee scence geniculate, the pedicel as broad as the rachis in fruit; petals at ee er ie a . geniculata. E. Hire aes linear or sinuous, the pedicels not as broad: as the rachis in fruit; — 15-3 mm _ wide. F. Fruit 8-ribbed, or 4-angled and short-beaked, 5-9 mm wide, or slender ee long-beaked the beak somewhat flattened in one plane and retuse or blunt at Wire, PRO Ot in ete ee ee 3. C. edentula. F. Fruit terete to % -angled, 3-4 mm wide, the beak conical and acute at the apex. 7. C. constricta. 1. Cakile maritima Scopoli Fleshy, herbaceous to sievy nee 2 suffrutescent annuals or fortuitous aera es erect to prostrate, occasionally form dense mounds up to 8 dm high and 12 dm across, much- branched, the ate pene or lax; first leaves large, up to 10 cm i lobe 0.4-0.7 mm high; stamens tetradynamous, the anthers 1.4-2.2 mm long, dehiscing introrsely to latrorsely; pistil lanceolate; stigma capitate; mature fruit 12-27 mm lon mm wide, + hastate and with two reflexed lateral horns on the lower segment or deltoid; the lower fruit s segment + conical, a peteen tow ard the apex, this surface broad and flat, concave, or convex; the upper hee segment mitriform or conical, terete or tetragonal, acute to blunt at the apex, with a scarious basal ma Lai seeds ellipsoid and dimorphic; cotyledons accumbent to occasionally incumbent o contorted. KEY TO THE SUBSPECIES A. Leaves entire to pinnatipartite, the lobes obovate or — rarely sinuately lobed themselves, and rarely less than 5 mm wide... . maritima ssp. maritima. A. Leaves deeply pinnatifid, often 2- -pinnatipartite, os ‘Sokis linear and 1-5 m ee re se Pe epee se I Ge ee eh eg ye oe CC. maritima ssp. baltica. B. Fruits deltoid in outline, not constricted at the point of articulation eee ee EEE GON see as erie MOLNAR et age ed ies Acer Nl FORE a c. C. maritima ssp. euxina. la. Cakile maritima Scop. ssp. maritima PiLate 1; Map 1. Cakile maritima Scop., Fl. Carn. ed. 2, 2: 35. 1772; based on Bunias Cakile L., Sp. Pl. 2:670. 1753. Lectotype: Fortngsl, “$. “Ybes [Setubal] in littore marini,” Loefling $.n, (LINN 847.6, not see seen; photo: Rapistrum cakile (L.) Crantz, Class. Crucif. 106. 1769. Bunias littoralis Salisb., Prodr. 273. 1796. An avowed new name for B. Cakile L. Cakile Cakile (L.) Karsten. Deutsche Fl. Pharm.-med. Bot. 663. 1882. Crucifera cakile (L.) Krause in Sturm, Fl. Deutsch. ed. 2, 138. 1902. THE GENUS CAKILE 105 Rapistrum maritimum (Scop.) Berger.. ie compen 173. 1784. Cakile maritima var. latifolia Desf., Fl. Atl. 2: 77. 1798. Holotype: without data een on microfiche: cs hg 38: i. 3). Cakile latifolia ( Desf.) Poir., Ency. Bot. Suppl. 2: 88. Cakile aegyptia var. latifolia (Desf. ) rere Fl. Afr. Nord. 1: 402. 1 Cakile maritima var. integrifolia Boiss. non Koch, Fl. Or. 1: 365. 1867. oe Egypt, Alexandria, Samar s.n., and T Turkey, Moeotidum, Stev s.n. (?c; not s Cakile maritima var. laciniata Hallier, Fl. Deutsch. 15: 44. 1883. Tene eaiay. Helgoland (?yr; not seen - Cakile maritima var. pand aria Terraciano, Ann. Acad. Aspir. Nat. Ser. 3, 1: 5. 1884. Type: Italy, Naples eens ot seen Cakile maritima f. pandataria (Terrac. ) 0. E. Schulz in Urb., Symb. Antill. 3: 504. Cakile maritima var. auriculata Post, F1. Syr. Pal. Sin. 103. 1896. Type: not known. Cakile maritima var. amblycarpa O. E. Schulz in Urb., Symb. Antill. 3: 503. 1903. Type: none specifie Cakile salateans var. oxycarpa O. E. Schulz in Urb., Symb. Antill. 3: 503. 1903. Type: none speci Cakile maritima var. sessiliflora O. E. Schulz in Urb., Symb. Antill. 3: 504. 1903. Type: none specifie Cakile maritima f. pygmaea O. E. Schulz in Urb., Symb. Antill. 3: 504, 1903. Type: none s specifie kile maritima var. susica Maire, Weiller & Wilczek, Bull. Soc. Hist — 26: 120. 1935. Type: Morocco, “in arenis ad ostium fluminis Sous,” ue ‘et a rd) " Gokile aegyptia var. susica (Maire, Weiller & Wilczek) Maire, Fl. Afr. Nord 12: 402. 1965. Isatis aegyptica L., Sp. Pl. 2: 671. 1753. Type: not known; not in LINN (cf. Savage Cakile aegyptiaca (L.) Willd., Sp. Pl. 3: 417. 1 Cakile maritima var. aegyptiaca a (L.) Delile, L. hesvek 19. 18 Cakile maritima var. integrifolia Koch non Boiss., Syn. Fl. ams 77. 1835. An avowed new name for the preceding Cakile maritima ssp. aegyptiaca (iL. ) Nyman, 17. 29.1 Isatis pinnata Forsk., Fl. Aegypt. 121. 1775. Lestcspe: “ex Aegypto inser.”, 8. Cakile pinnatifida Stokes, Bot. Mat. Med. 3: 484. 1812. Type: not se Cakile sinuatifolia Stokes, Bot. Mat. Med. 3: 485. 1812. Type: “the shore of the Adriatic,” Agerius s.n. (?). Cakile maritima var. sinuatifolia (Stokes) DC., Reg. Veg. Syst. Nat. 2: 429. 1821. Bunias i Ye Fl. Lib. 35. t. 16, fig. 3. 1824. Type: Libya, “in littore Penta- poleos” (?GE; not seen). Cakile jim een Syst. Veg. 2: 852. 1825. An avowed new name for the precedin Cakile crenata Jordan, Diagn. 1: 346. 1864. Type: none specified. Cakile edentula Jordan non le Hook., Diagn. 1: 344. 1864. Type: none specified. Cakile maritima var. edentula (Jord.) Loret in Loret & Barr., Fl. Montp. ed. 2. 50. 1888. vagal maritima “proles” edentula (Jord.) O. E. Schulz in Urb., Symb. Antill. 3: Cakile. aegyptia var. edentula Davee ) Biggs FI. Afr. Nord 12: . 1965. Cakile hispanica Jordan, Dia : none s Cakile SOT Ln ane (Jord. ) Paol. in Fiori & Paol., Fl. Anal. Ital. 1: 453. 1898. 106 JAMES E. RODMAN Cakile aegyptia var. —— Aa ) Maire, Fl. Afr. Nord 12: 403. 1965. Cakile littoralis Jordan, Diagn. 1 . 1864. Type: none specified. Cakile maritima var. Sate Cosson in Loret & Barr., Fl. Montp. 1: 64. 1876. An avowed new name for the prece Cakile aegyptia var. australis (Come. ) Maire, Fl. Afr. Nord 12: 402. 1965. Cakile maritima var ppearetianten Paol. in Fiori & Paol. Fl. Anal. el 1: 452. 1898. An avowed new name for C. litt is Jor Cakile maritima ‘ ‘proles” toralie (Jord. ) O. E. Schulz in Urb., Symb. Antill. 3: 503. 1903. Cakile monosperma Lange, Descr. Icon. Ill. 1: 5. t. 7. 1864. Type: Spain, Coruna, IX-1852 (?c; not seen). sige maritima var. monosperma (Lange) O. E. Schulz in Urb., Symb. Antill. 3: 503. 1 wide, the lo wit s obovate or eke. 5-15 mm long, 4-8 in number, Ba dat z apex, rarely themselves lobed, to broadly ovate or lanceolate and sinuately to dentately tees pedicels in fruit divergent to occasionally reflexed, 1.5-7 mm long, mm wide; mature fruit 14-27 mm long, usually hastate with two opposite lateral horns on the upper part of the lower segment; the lower fruit segment conical, expan ing toward apex, with two lateral horns projecting Sclow the articulating surface, often reflexed, <1-3 mm long; upper fruit segment mitriform or conical, usually 4- angled, apex acute to blunt and usually compressed, the base with a scarious margin; when aborted, the lower segment often becomes a small janet extension of the pedicel and the upper fruit segment is fusiform. CHROMOSOME NUMBER: 2n=1 DISTRIBUTION: shores of the Mediterranean Sea and of the Atlantic Ocean from the Pacific Coast of North America and the mperate coast of Australia; more widely introduced as a ballast weed. Flowering probably throughout the year in southern parts of its range, but more seasonally further n ENTATIVE SPECIMENS (native range). Algeria: Hussein Dey, 4-IX—1866, poem 220 (ps). Belgium: Oostduinkerke, 26-VI-1928, Robyns s.n. (pao). Egypt: exandria, 15-III-1926, Jepson 10912 (uc). England: Cornwall, Port Luney, 24- VUI-1929, Thurston 2920 (mo,us); Kent, Sandwich Bay, 11-X-1960, Raven 15852 sa). France: Aude, La Nouvelle, 1902, Sennen s.n. (ps,cH); — 1821 n. rsica, Bonifacio, 1880, Reverchon 286 (ps,ny); Landes, Mimizan Plages, 11-IX-1930,. Tidestr s.n, (us); Normandie, Fermanville, 24-VII-1936, Tidestrom ] 1883, aE earreneese 10 os). Mocece: "Tangier SRK. Moors iad s.n. (cv). Bohol ds: Terschelling Island, 15-VIII-1951, van — & Nijenbuis 128 (pao, MO,NCU,NSW,SMU,WIS ). Portugal: Porto, 1891, Bu Poti s.n, (us). Scotland: Luce Bay, Wigtownshire, Henderson 1538 (pao,ncv). | An s, Wielkomm 391 ( MO); Majorca, 13-IX-1907, Knoche 2285 iy. oe near View, VIII-1926, Exell 36 (sm). Tunisia: Gabes, III-1907, Pitard 37 (mEL,Mo.Nsw,Ny). Yugoslavia: Brioni, 25-VI-1905, Kanche sn. (Ds REPRESENTATIVE SPECIMENS (naturalized or introduced). Argentina: Buenos Aires, Villa Gesell, 3-III-1961, Burkart 22368 (uc); San Clemente del Tuyu, Partido de Lavalle, dunas, 29-I-1939, Cabrera 4919 (cu,ny); $. Clemente del Tuyu-Medanos, 10-II-1961, Fabris-Cullen 2532 (rsa). Australia. South Australia: near Beachport, ca. km N.W. of Mount Gambier, 12-XI-1959, Jackson 188 (ap); Kangaroo Island, Hog Bay, ee. Segre s.n, (AD); pce XI-1918, Andrew s.n. (ap or uth of THE GENUS CAKILE 107 Perry sn. (CANB). Victoria: Altona at the mouth of Skeleton Creek, 21—XI-1967, Cullimore 104 (MeEL,Nsw); Beaumaris, X-1922; Tagdell s.n. (MeL); Phillip Island, Cowes, III-1950, Dunk s.n. (MEL); Shelly Beach near Portland, X-1944, Beauglehole s.n. (MEL); Warrnambool, 5-I-1960, Howard s.n. (ap). Western Australia: Esper- ance, 17—-X-1968, Eichler 20300 (ap,cH); Fremantle, 19—-XII-1897, Helms s.n. (PERTH); Geographe Bay, E. of Busselton, 23-III-1898, Morrison s.n, (kK); Guilder- ton, mouth of Moore River, 21—II-1963, Sauer 3438 (uc); King George’s Sound, VIII- 1 20-IV-1963, Moran 10765 (ps,cH,RSA,sD,UC,Us); San Martin Island, Hassler Cove, 10-IV-1963, Moran 10530 (sp). New Caledonia. Plage de Anse Vata pres Noumea, 6—VIII-1950, Baumann 5161 (us). Uruguay. Dep. Maldonado: Punta del Este, 3-I- 1941, Descole 65 (cH); Dep. Rocha: La Paloma, 7—XI-1934, Herter 1704 (ny); Maravillas, 16-XI-1948, Herter 1704a (Mo). U.S.A. Alabama: Mobile Co., ballast ground of M & O R.R. wharf, 16-VIII-1891, Mohr 21 (us). California: Alameda Co., Bay Farm, 4-IX—1936, Covel 584 (cas); Contra Costa Co., Pt. Orient, 19—VII-1958, Robbins 3905 (uc); Del Norte Co., Pt. St. George, 20-VII-1947, Tracy 17872 (uc); Humboldt Co., near Samoa, 21-IX-1940, Tracy 16724 (DAO,Ds,GH,MO,NyY,UC); Los Plant, IV-1961, Bennett s.n. (rsa); Marin Co., Stinson Beach, 19-V—-1935, Rose 35178 (cAs,Ds,F, .MO,NY,POM,RSA,UC,US); Mendocino Co., beach at creek mouth 4 miles S. of Fort Bragg, 11—VII-1941, Heller s.n. (ps); Monterey Co., sea beach at Monterey VII-1940, Hesse 134 (uc); San Diego Co., Imperial Beach, 26-IV—1954, Harbison s.n, (uc); San Francisco Co., near Daly City, 29-V—1949, Weatherby 504 (rsa); San Luis Obispo Co., along beach between Pismo and Oceano, VIII-1941, Reed s.n. (cas); San Mateo Co., Half Moon Bay, 12-II-1941. Hoover 4753 (uc); Santa Barbara Co., beach southeast of Ellwood, 15—-X-1957, Pollard s.n. (cas); a a ie} o = 1-VIII-1958, Pollard s.n. (TEx). New Jersey: Ballast near Communipaw ; IX-1880, Brown s.n. (¥,N¥). New York: Bronx Co., at Longefellow and Viele Ave., 28-IX-1948, Ahles 897 (Ny). North Carolina: Wilmington ballast grounds, 14—VIII— Florence, 15—VII-1965, Clark s.n. (ps); Lin oln Co., Ne rt 2 1963, Stuckey 1929 (os); Tillamook Co., Netarts Bay, 13-VIII-1969, Ro 7 (cH) ennsylvania: Philadelphia, Greenwich Point ballast. 26-VIII-1877, Parker 10990 ( MICH,MO,NY ) ls Harbor, Reverchon s Mo). Virginia ews : : po ballast, 30-V-1921, Leonard & Killip 61 (cas,F,GA,Mo,us ). Washington: Clallam Co., Mukkaw Bay, 22-VII-1940, Ownbey 2260 (ps,cu,uc); Grays Harbor Co., south of Westport, 12—VII-1952, Bartlett & Grayson 643 (MicH). Scopoli (1772) published Cakile maritima as a new name for Lin- naeus’ Bunias Cakile to avoid coining a tautonym. Therefore the type of C. maritima ssp. maritima is the same as the type of B. Cakile. However, Linnaeus (1753) founded the species on several specimens as is evident from his comment: “Habitat in Europae, Africae, Americae maritimus. A lectotype must therefore be chosen, but Pobedimova (1964) and Ball (1964a) neglected to do this. I am here designating as lectotype a sheet in the Linnean Herbarium annotated: “Bunias Kakile . . . S. Ybes,” a collection from Steubal, Portugal, made by Loefling in 1751 (Ball 1964a ). The specimen has pinnatifid leaves and hastate fruits with both seg- ments apparently fertile, thus resembling most Mediterranean collections and, hence, also corresponding to Scopoli’s concept of the species. 108 JAMES E. RODMAN Pobedimova (1963, 1964) separated the Western European plants (as Cakile monosperma) from Mediterranean plants (as C. maritima) on the basis of the frequently aborted lower fruit segments exhibited by Western European sea rockets. This distinction was apparent in the material (ca. 300 collections) available to me for study (see section on morphology and anatomy) and must also have been evident to Schulz (1903) who made the combination C. maritima var. monosperma and commented that these plants occurred “ab ins. Helgoland ad Lusitaniam.” Ball (1964a,b), however, discounted this character as too variable but proceeded to recognize a Western European subspecies (C. maritima ssp. maritima) distinct from a Mediterranean subspecies (C. maritima ssp. degyptiaca) on the basis primarily of the flat or concave nature, respectively, of the articulating surface of the lower fruit segment. I have found this character to vary over a wide range of material although the trend is there. Ball (1964a) also had to concede that: “Plants which are almost identical with subsp. maritima occur sporadically throughout the range of the subspecies [ssp. aegyptiaca].” I prefer to make no taxonomic distinction between Western European and Mediterranean sea rockets on the basis of the characters currently investigated. While trends can be perceived (e.g., toward aborted lower fruit segments and lower segments with a flat articulating surface in Western European plants ), the discontinuities are not substantial enough to warrant formal nomencla- tural recognition, at least at the species or subspecies rank. The plethora of infraspecific names for Cakile maritima ssp. maritima attests to the variation in leaf morphology, largely in the degree of lob- ing, variation in fruit shape, and the degree of development of the lateral horns present in these plants. However, most of this variation is apparent within individual populations and even, occasionally, on individual plants, as a critical reading of Schulz (1903, 1923) reveals. Pobedimova (1964) remarked that she could find no correlations between this variation and geographical range and consequently chose to recognize no varieties or forms, a position which I think is fully justified. 1b. Cakile maritima Scop. ssp. baltica (Rouy & Foucaud ) P. W. Ball PLATE 1; Map 1, Cakile maritima ssp. baltica (Rouy & Foucaud ) Hylander ex P. W. Ball, Rep. Spec. Nov. 69: 37. 1964, Cakile maritima f. baltica Rouy & Fouc., Fl. Fr. 2: 69. 1895. Syntypes: “F. Schultz Herb. norm., nov. ser., n° 1318; Reichb. n° 361.—Non encore observe en France” (?P; not seen). ; Cakile maritima ssp. baltica ( Rouy & Fouc.) Hyl., Forteck. Nord. Vaxter 64. 1955. No citation of basionym, hence illegitimate. Cakile baltica Jordan, Diagn. 1:345. 1864. Nomen nudum. Egy maritima var. baltica (Jord.) Paol. in Fiori & Paol., Fl. Anal. Ital. 1: 452. THE GENUS CAKILE 109 a ile baltica Jord. ex Pobed., Not. Syst. Leningrad 15: 66. 1953. Holotype: ak; Bie TOV. calcd pela, distr. Maritimus, Kurskaja kossa (Kurisch-Nerung ) ee ope Schwarzerberg, ad sinum Kurisch,” 16-VII-1949, Pobedimova & Haesing 1216 (LE, not seen Caki € maritima var. bipinnat ta O. E. Schulz in Urb., Symb. Antill. 3: 504, 1903, pro porte Syntypes: “Hab. ad mare balticum prope Pillau. in Turcia, Graecia.” (?B; no en). Leaves deeply and finely pinnatifid, the lobes ea rachis rarely more than 4 mm wide, the lobes up to 3 cm long, often 1 cm apart and = alternate along the rachis, linear to oblanceolate, the lobes of large leaves often further lobed, sinuately to laciniately; leaves seemingly not as fleshy as in typical ng fruiting pedicels divergent to reflexed, 3-5 mm long, 1-2 mm wide; mature f + hastate, usually constricted at the point of arti culation; the lower fruit Sade c oaial expanding toward the apex, the lateral horns usually 1-2 mm long, the articulating surface + flat across to slightly convex; upper fruit segments mitriform to obconical, weakly 4-angled to terete, apex acute to blunt, the base with a scarious border. CHROMOSOME NUMBER: 2n=18. ISTRIBUTION: sandy shores of the Baltic Sea and the Kattegat and Skagerrak chan- ne Flowering from June to October. REPRESENTATIVE SPECIMENS. Denmark. ane Degho. 16-VII-1862, Leth s.n. (ps); Jutland, Alrg, 20-IX-1964, Larsen e ps). Estonia. Rakvere coast, 12-VIII-1934, Meinertzhagen s.n, (BM). Finlan ee Storrevet, 13—VII-1911, Johansson 695 (Mo,Ny,uc); Karelia, ee 14-VI I-1937, Kohonen s.n, (Ncu); Nylandia, Helsinki, 13-VIII-1961, ‘Oinonen s.n. (rsu); near Hanko, 2—VIII-1964, Alava 4469 (uc); Satakunta, Law. o-_1X_1965, Kause ¢ Seikkula s.n. (uc). rmany. Greifsvalde, IX-1933, Guyot s.n. (NY); Kiel, 1827, Meyer s.n. (Mo); Rugen, 31-V-1868, Englemann 3635 (mo); Warnemunde, Dethar ding 361 (MEL,NY,POM). 5 ® C. arctica 4 C. maritima ssp. baltica ® C. maritima ssp. maritima © C. maritima ssp. euxina * C. arabica (an rail , ~ OP CE et | , oie reed oes a ~y Bick oe X Wein an nO ~~ ei ye ay ( \ sel are a4 a teat iets 8 Nita | der * ee ag = e ve ra et i eos * is? Pe eet, iE yg ei, Cd Ro: ge : i ba fon , oy > / ee : aa ae . 2 bout Map 1. Distribution of Cakile arctica, C. maritima ssp. baltica, ssp. m maritima, and ssp. euxina, and C. arabica in the Old World ( ble f C. arctica in a and of C. maritima in ). occurrence the Canaries not indicated; range of C. maritima where naturalized not m 110 JAMES E. RODMAN Poland. Hel, Jastarnia, 22—VIII-1937, Trela s.n.(us); Wolin, 1949, Kobendza s.n. (pao). Sweden. Goteborg, Wargo, VIII-1904, Thedenius s.n. (micu,pH); Gotland, Ljugarn, 20-VII-1941, Fries s.n. (pAo); Halland, Vallda, 13-VII-1934, Skottsberg s.n. (uc); Skane, Falsterbo, 10-IX-1964, Hekking 3469 (Nsw); Sodermanland, Asko, 24— VIII-1926, Asplund 861 (pao,Ncu,NyY). Rouy and Foucaud’s (1895) name for the Baltic Sea plants was pub- lished in the rank of “forme,” by which term these authors did not mean the forma of current usage but a rank more or less intermediate between species and variety, corresponding to the “proles” of Schulz (1903). Two specimens were cited in the protologue, but Ball (1964a), in effecting the subspecies combination adopted here, did not choose a lectotype; indeed, the herbarium of deposit is not known for certain. Pobedimova (1953) legitimately published Jordan’s (1864) nomen nudum as Cakile baltica Jordan ex Pobed. and specified a type in Leningrad. Later, how- ever (1964), she interpreted this species as C. baltica (Jordan ex Rouy & Fouc.) Pobed. and cited her type as a neotype, apparently implying that Rouy and Foucaud’s specimens were lost or destroyed. The Baltic Sea plants are not sharply distinct from typical Cakile maritima, but they can usually be distinguished by an ensemble of leaf, pedicel, and fruit characters. The leaves are deeply pinnatifid, and there appears to be a trend toward very finely dissected leaves in the eastern end of the Baltic Sea. Pedicels are longer on the average than in typical C. maritima, and the fruits are usually smaller and with a flatter articu- lating surface. The seed glucosinolate composition, with a predominance of allylglucosinolate, is distinctive. 1c. Cakile maritima Scop. ssp. euxina (Pobedimova ) N yarady PLATE 1; Map 1. Cakile maritima ssp. euxina (Pobed.) Nyarady in Savul., Fl. Reip. Pop. Roman. 3: 480. 1955. Cakile euxina Pobed. in Grossheim, Key Pl. Cauc. 386. 1949. Nomen nudum. Cakile euxina Pobed., Not. Syst. Leningrad 15: 71. 1953. Holotype: U.S.S.R., “Prope oppidum Scadowsk, in litore marino arenoso, ponti Euxini,” 15-VIII-1947, Pobedimova 48 (Lx; not seen); isotypes: A,MO,NY,US Cakile maritima var. bipinnata O. E. Schulz in Urb., Symb. Antill. 3: 504. 1903, pro ai Syntypes: “Hab. ad mare balticum prope Pillau, in Turcia, Graecia.” (?B; not seen ). Leaves finely and deeply pinnatifid, the lobes and rachis 1-4 mm wide, the lobes widely and alternately or irreguarly spaced, rounded at the apex, often further lobed mm long, 1.5-3 mm wide; mature fruit angulate-ovate in outline, 17-21 mm long, without prominent lateral horns; the lower fruit segment conical, prolonged upward at the articulating surface into 2 sharp-pointed triangular wedges, the surface concave ] between; the upper fruit segment conical and with a scarious basal margin. CHROMO- SOME NUMBER 2n—18. DISTRIBUTION: coast of the Black Sea and sporadically on shores of the Aegaean. Flowering from May to September. THE GENUS CAKILE 111 REPRESENTATIVE SPECIMENS. Bulgaria. Sozopol, 29-VIII-1932, Stojanoff s.n. (cas). Gree Plaka, 31—-VIII-1891, Sintenis & Bornmuller 1161 (PH); 4 miles S. of cleanin: VIII-1918, Ramsbotious sn, (BM). Romania. Dobrogea ad Capul Sabla, : ) 25, Borza a Turkey. Samsun, 6-VII-1931, Krause 3893 (uc). U.S.S.R. Azoff, Gandoger s.n. (Mo); outh of the Kuban River, 11-VIII -1916, Woronow 897 (Ny); Odessa, Kelk s.n. (Ny); Strelka, 9—-VII-1959, Kuznetzova sn. (NY). Cakile maritima ssp. euxina is poorly represented at least in American herbaria; hence, it is difficult to assess the full range of variation in the Black Sea plants. Hedge (1965) questioned whether the characters delimiting this taxon were clearly related to geographical distribution. The material available to me for study, however, was homogeneous and distinct enough to merit taxonomic separation from typical C. maritima. The differences are of degree, however, not kind, and it therefore seems reasonable to adopt the view of Nyarady (1955) and Ball (1964a,b) and treat the Black Sea plants as a subspecies of C. maritima, from which they most likely evolved since the late Pliocene. 2. Cakile lanceolata ( Willdenow ) O. E. Schulz Plants herbaceous to suffrutescent, annual to fortuitously perennial, not especially succulent, erect to more often straggling or prostrate, much-branched, the branches often more than 5 dm long; ae bogie up to 4 x 15 cm, broadly ovate or ovate-deltoid and petiolate to ovate-lanceo e and more attenuate basally, + entire to variousl lobed; fruiting racemes ile S ihe n 2 dm long, linear or sinuous; pedice els ca. mm long in rut eas oa rt 3.5-5 mm long, dimorphic, ir nectaries conspicuous, ‘the madiace: 0.3-0.7 7m fra stamens tetradynamous, the anthers 1.1-2.2 mm ger dehiscing introrsely to latrorsely; pistil ensiform to KEY TO THE SUBSPECIES A. Fruit turbinate, the beak less than the length of the seminiferous part. .......... 2c. C. lanceolata ssp. “alacranensis. A. Fruit fusiform or lanceolate, the beak Siliee or exceeding the length of the SENG OVOLS AEE i ee ee B. Fruit hl ee lanceolate with a long beak, the upper es eager — twice e length of the | egment or greater; leaves usually sinuately or dentately — Pe i 2a. C. lanceolata ssp. lanceolata. B. Fruit fusiform or lanceolate with a short beak, the upper fruit segment usually less than twi e length of the lower segment; leaves pinnatifid to less ey sinuately to pha aa We C. Fruit fusiform, + oval in cross-section and 4(-8)-sulcate. .................. a cane Se 2b. C. cola ssp. fusiformis. C. Fruit rpg easton lanceolate, usually eels at the point of articulation, weakly 4-angled to C. lanceolata ssp. pseudoconstricta. 112 JAMES E. RODMAN 2a. Cakile lanceolata ( Willd. ) O. E. Schulz ssp. lanceolata PLATE 1; Map 2. Raphanus donceolatius Willd., Sp. Pl. 3: 562. 1800. Holotype: “Antillis,” anon. s.n. (8, not seen; pho Cakile lanceolata (Willd. ) 0. E. = in Urb., Symb. Antill. 3: 504. 1903. Cakile dom — Tussac, FI. ene : 119. Pl. 17. 1808. Type: Dominican Republic, Port Francois, (?) Tussac s.n. (?p;n en). Pine lanceolata ssp. slates "(Tussse) O. E. Schulz in Urb., Symb. Antill. 3: 505. 1 Cakile ‘aequalis L’Hér. ex DC., Reg. Veg. Syst. Nat. 2: 430. 1821. Holotype: 00. 89: I. 8. Cakile americana var. cubensis DC., Reg. Veg. Syst. Nat. 2: 429. (May) 1821. Lectotype: Cuba, est Bonito, Humboldt & Bonpland s.n. (P; seen on microfiche: IDC 6209. 119: I Cakile maritima var. cubensis ae é pongemy FI. S. U.S. ed. 2, Suppl. 606. 1883. Cakile cubensis Kunth in HBK, N . Sp. Pl. 5: 75. (Sept.) 1821. Holotype: Cuba, Cayo ee March 1801, Humboldt & ites s.n. (P; seen on microfiche: IDC 6209. 119: I. 3.). vatially more shen 3 dm long; fruiting pedicels 1.5-4 mm fin neni 4-5 mm long, i i Bane oe i i rarely pale lavender; mature fruit 19-31 mm long, the upper se t usually more than stricted at the point of articulation of the 2 segments; ‘io ruit segment 10 mm , su teeth projecting upward; upper fruit segment slenderl ereun weakly t etragonal or curved. CHROMOSOME NUM MBER: on DISTRIBUTION: sandy beaches of the Greater and Lesser Antilles, the Bahamas, Bermuda, and the Carribbean Coast of Venezuela, Colombia, and Central America; sporadic i in cakgroea ang Florida and the Mexican coast of the Gulf of Mexico. Flower- ing throughout the yea TATIVE SPECIMENS. Anguilla. Island Harbour, 3-I-1959, seers 18613 (ane see Bahamas. Acklin’s Island, Atwood Ca , 4-XII-1907, Wilson 3 (F,GH, NY); — Island, Little Golden Cay, 6-III-1966, Dawson 26700 (cH, ca Grand oi Island, Russell Town, since ae Gillis 7788 (cu); Great Abaco, Cook’s Cove, 6-VIIL_ 1953, Robertson 2 4 (ps); Inagua, 18—X-1904, Nash & Taylor 1155 (F, NY); Mayaguana, Abr aham Bay, 8-XII-1907, Wilson 7531 (¥F,cH,NY); New Provi- dence, West Bay, 26-II-1946, Degener 18924 (cH, NY,PH); Salt Key Bank, Water Key, V-1909, Wilson 8144 (¥,Mo,Ny); San Salvador near French Bay, 19-III-1963, Gillis 5257 (puxe,1J,Msc); South Bimini Island, 28-III-1965, Stimson 1061 Pes. Brown & Britton 314 (r, ag ,PH,us ). Colombia. Cartagena, XII-1964, Sarmiento oo (coL,us). Cuba. Cam maguey, Punta Arenas, 9-III-1909, Shafer 698 (¥,ny,us); Isle of Pines, Caleta, 8—III-1916, Bebton. Wilson & Leon 15284 (F,Ny,us). Dominican Re- public. Beata Island, 7-VIII_-1950, Howard 12444 (BM,cuH,1J,MicH). Guadeloupe. East of Moule, 9-XI-1959, Proctor 19888 (BM apd _ Tortue Island, 31—-XI- 1928, Leonard ¢ 11396 (cH,Ny,us). Honduras. La Ceiba, 6-VII-1938, Yuncker, Koepper & Wagner 8237 (BM,F,CHK, MICH,MO,NY ean amaica. Grand Cay- man, 6—VIII-1962, Sauer 3334 (wis); St. Anne’s Bay, 7111-1908, Harris 10353 (BM, F,K,NY,Us); St. James Parish, near Little River, 14-VIII-1963, Crosby dr Anderson THE GENUS CAKILE Lis @C. geniculata eC. lanceolata ssp. lanceolata © ; 2 ssp. fusiformis ( ; e ssp. alacranensis ssp. pseudoconstricta Nee Map 2. Distribution of Cakile geniculata and C. lanceolata ssp. lanceolata, ssp. fusiformis, ssp. alacranensis, and ssp. pseudoconstricta in the New World (occurrence of C. lanceolata in Bermuda not indicated ). 1195 (DUKE,F,GH,MICH,MSC,NY,RSA,UC ). Mexico. Campeche: Isla Aguada, 11—-VII-1962, Sauer & Gade 3182 (micu). Quintana Roo: Tancah, 5-VIII-1932, Steere 2516 (micu). Panama. Viento Frio, 8-VIII-1911, Pittier 4113 (cu,Ny.Us). Puerto Rico. Cabo Rojo, 9—II-1885, Sintenis 591 (BM,F,GH,MSC,NY,us); Culebra, 12-III-1906, Britton & Wheeler 206 (¥,Ny,us); Santurce, 23-I-1899, Heller ¢> Heller 256 (cu,¥,x, Ny,us). U.S.A. Florida: Dade Co., Key Biscayne, 7-IX-1959, Atwater 138 (FLAs); ‘ ) 1859, Cruger s.n. (kK). Virgin Islands. St. Croix, 11—XI-1895, Ricksecker 68 (ps.F,GH, Mo,Ny,uCc,Us); Tortola, 16—XII-1918, F ishlock 272 (GH,K,PH). The Caribbean and West Indian sea rockets were for a century known by the name Cakile aequalis L'Hér. ex DC. until Schulz (1903) resur- rected Willdenow’s earlier name. While Schulz specified that the type was from “Antillis,” the Berlin specimen (a photograph of which was very kindly provided me by Dr. Th. Eckardt) carries a label reading Habitat in Guinea (w).” If this label belongs with the specimen, the plant could have originated from the Guianas (which would then represent a range extension southeast from Barbados and Venezuela) or less likely from African Guinea. Cakile could have been transported in ship ballast during 114 JAMES E. RODMAN the slave trade, but this seems doubtful because it is otherwise com- pletely unknown from the coast of tropical Africa. Minor but real variation characterizes many island collections of Cakile lanceolata, apparently as a result of local isolation or small population effects. Two somewhat distinctive examples include the Cayman Islands, where the plants tend toward pinnatifid leaves and short-beaked fruits, and Bermuda, with fleshy plants with short-beaked, somewhat tetragonal fruits. The differentiation of what are here interpreted as subspecies of C. lanceolata appears to be an intensification of the pressures isolating these rather distinct island forms. 2b. Cakile lanceolata ( Willd.) O. E. Schulz ssp. fusiformis reene ) Rodman comb. nov. PLATE 1; Map 2. Cakile fusiformis Greene, Pittonia 3: 346. 1898. Holotype: Florida, “Mangrove Key,” 10-III-1898, Pollard 19 (p-c); isotypes: F ,NY,US. Cakile lanceolata var. fusiformis (Greene) Patman, Quart. Jour. Fla. Acad. 25: 198. B ; Cakile alacranensis x aequalis Millsp., Field Mus. Bot. 2: 130. 1900. Holotype: Florida, Palm Beach Co., Palm Beach, 28—XII-1895, Webber 243 (mo); isotype: F. Cakile Chapmanii Millsp., Field Mus. Bot. 2: 130. 1900. Lectotype: Florida, Dade Co., Miami, VII-1877, Garber 6472 (us); isotypes: F,PH. Leaves ovate-lanceolate, nearly entire to dentately lobed to most commonly pinnati- fid; fruiting pedicels 2-4 mm long; sepals 4-5 mm long, distinctly bigibbous; petals distinctly clawed, 4.9-9.4 mm long and 2.4-4.2 mm wide, white to very rarely pale lavender; median nectaries 0.4—0.7 mm high; anthers 1.3—1.8 mm long; mature fruit 16-25.5 mm long, 4-6 mm wide, broadly to slenderly fusiform, striate to suleate (4-8 striae or sulci), somewhat compressed longitudinally; lower fruit segment 5-10 mm long, narrowly terete basally and expanded into a broad oval articulating surface, usually with 2 small teeth projecting upward; upper fruit segment 10.5-18 mm long, + tumescent, tapering somewhat abruptly to an acute beak, the basal margin scarious. CHROMOSOME NUMBER; 2n=18. DISTRIBUTION: sandy coast of the Bay of Honduras from Honduras to the Yucatan Peninsula and southern Florida. Flowering throughout the year. REPRESENTATIVE SPECIMENS. British Honduras. Stann Creek, 7-I-1932, Schipp 284 (F,Mo,uc); Turneffe Islands, Deadman I Cay, 10-IV—1962, Stoddart 127 (1J jus). uras. Cortez, Travesia, 30-V-1970, Hernandez & Barkley 40488 (herb. of North- eastern University ); Tela, 8-V—1919, Blake 7279 (us). Mexico. Quintana Roo: Tancah, 5-VIII-1932, Steere 2519 (micu). Yucatan: Progreso, 15-VIII-1932, Steere 3114 (micH,smu ). U.S.A. Florida: Broward Co., Ft. Lauderdale, XI-1903, Small & Carter 951 (pH); Collier Co., Marco Island, 25~XII-1961, Duncan 21854 (GA,LAF,NCU ); a 2 & 6 3 & c¢ 5 ® iv] S $ Z QO SG q Dn sy “ <) z. oO ie) i?) F fo _ Ly \ ~ Ke) ioe) i) an ; > cH,Mo); Lee Co., Sanibel Island, 9-III-1954, Cooley 2560 (FLAS,GH,MICH,NY,USF ); e Co., Bradentown, 4-III-1935, Weber s.n. (rLas); Monroe Co., Knights Key, THE GENUS CAKILE 115 30-IV-1896, Curtis 5645 ( F,FLAS,GH,MO,MSC,NY,POM,UC ) ; Marquesas Keys, IV—1886, aay (BM,F,GH,NY,PH,US ); Palm Beach Co., Palm Beach, 3-V-1969, Cassen 470 LAS ). The southern and western Florida sea rockets present a nightmare of variation to the taxonomist. A number of epithets have been applied to these plants (aequalis, cubensis, fusiformis, Chapmanii), but Patman (1962) recently considered all southern and western Florida material to be Cakile lanceolata var. fusiformis. Both herbarium collection studies and field work indicate that there is enough geographic integrity to certain morphologically recognizable characters to justify the formal recognition of three taxa in this area: C. constricta to the north, C. lanceolata ssp. fusiformis primarily in the Florida Keys, and C. lanceolata ssp. pseudo- constricta between. This necessitates the unfortunate synonymization of Millspaugh’s (1900) C. Chapmanii, which has been widely but varyingly used for over 50 years. Millspaugh cited eight collections in the proto- logue without designating a specific type. I have studied all eight col- lections, and one, Garber s.n. (us #6472), bears the annotation by Millspaugh: “Cakile Chapmani [sic] sp. nov. Type C. F. Millspaugh.” This collection most closely, though not perfectly, conforms to the original description, and I feel compelled to designate it as the lectotype of the name. This specimen, from Miami, Florida, is clearly indistinguishable from C. fusiformis Greene (1898), and consequently, C. Chapmanii be- comes, as here lectotypified, a synonym of C. lanceolata ssp. fusiformis. Pobedimova (1964) treated C. Chapmanii (without designating a type) as a synonym of C. cubensis, a name which, after study of the type material in Paris (seen on microfiche), I consider synonymous with C. lanceolata ssp. lanceolata. Indeed, Pobedimova annotated two sheets of Brown ¢& Britton 314 (us) from Bermuda as C. lanceolata and C. cubensis although I can see no differences between them. In the original description of Cakile fusiformis, Greene (1898) charac- terized the lower fruit segments as sterile. Millspaugh (1900) and Small (1903) confounded things by describing the lower segments as 2-spermous and using this feature as a key character. As Pobedimova (1964) pointed out, however, the fruit segments are typically one-seeded. In any large population sample some 2(-3)-spermous or aborted lower segments will be encountered. Cakile lanceolata ssp. fusiformis can be distinguished from other C. lanceolata plants by its fusiform fruits which have a somewhat bloated appearance and which are commonly 8-striate or 8-sulcate longi- tudinally. The taxon occurs disjunctly in southern Florida and on the coast of the Bay of Honduras to the Yucatan Peninsula where it inter- grades with C. lanceolata ssp. lanceolata to the south and C. lanceolata ssp. alacranensis to the north and where it may have originated by hybrid- ization between these two taxa. 116 JAMES E. RODMAN 2c. Cakile lanceolata ( Willd.) O. E. Schulz ssp. alacranensis Millspaugh ) Rodman comb. nov. PLATE 1; Map 2. Cakile alacranensis Millsp., Field Mus. Bot. 2: 130. 1900. Lectotype: Mexico, Alacran Shoals, Pajaros Island, 8-IN-1899, Millspaugh 1764 (¥F). Cakile lanceolata var. alacranensis (Millsp.) O. E. Schulz in Urb., Symb. Antill. 3: 103 . 1903. Cakile edentula var. alacranensis (Millsp.) O. E. Schulz, Pflanzenreich IV, 105: 27. 1923. thin, 2-3.5 mm long; sepals 3.5-5 mm long; petals clawed, 6.2-7 mm long and 1.9-2.6 mm wide; median nectaries 0.3-0.5 mm high; anthers 1.2-1.5 mm long; mature fruit DISTRIBUTION: sandy coast and offshore islands of the Yucatan Peninsula. REPRESENTATIVE SPECIMENS. Mexico. Quintana Roo: Puerto Juarez, 19—VII-1962, Sauer & Gade 3270 (¥); Isla Mujeres, 1-I-1895, Millspaugh 13 (cu,Micu); Cozumel Island, 11-VI-1885, Gaumer 139 (¥,GH,K). Yucatan. Alacran Islands, 8—III-1899, Millspaugh 1767 (¥), 30-VI-1961, Fosberg 41864, 41868, 41900 (puKE,Us,USF); Progreso, IV-1887, Millspaugh 7 (¥), Cakile lanceolata ssp. alacranensis is distinguished by its plump, tur- binate fruits. It is endemic to the coast of the Yucatan Peninsula and nearby islands, including the Alacrans. Sauer (1967), who has done field work in this area, thought it likely that the Alacran plant was a “seaborne immigrant from the Yucatan Peninsula upstream.” Furthermore, Sauer (1967) described sandy beaches of Quintana Roo with “a nearly con- tinuous row of sea rocket, Cakile alacranensis.” While plants of C. lanceo- lata ssp. lanceolata and C. lanceolata ssp. fusiformis have been collected in this region, C. lanceolata ssp. alacranensis would appear to form a series of populations, especially in the northern part of the peninsula, with sufficient geographic integrity to justify the recognition of a discrete subspecies. In fruit morphology the plants intergrade with C. lanceolata ssp. fusiformis but appear to have somewhat smaller flowers and a dis- tinctive seed glucosinolate composition (based upon a single sample). 2d. Cakile lanceolata (Willd. ) O. E. Schulz ssp. pseudoconstricta Rodman ssp. nov. Piate 1; Map 2. ao pinnatifida profunde; folia prima ad 3 x 12 cm, folia serotiniara parviora ob- ta + integra; pedicelli divergentes 1.5-5 mm longi; sepala lanceolata 4-5 mm longa bigibbosa; petala unguiculata, 6-8.2 mm longa xX 2-2.8 mm lata, alba vel THE GENUS CAKILE it? lavandula raro; nectaria mediana 0.3-0.6 mm alta; antherae 1.2-1.7 mm _ longae; siliculae lanceolatae 15-24 mm longae constrictae plerumque ad locum articuli; articulus inferior 5.5-8 mm longus cylindricus vel conicus leviter + teris; articulus superior tenuis 9-17 mm longus, 3-4 mm latus, fusiformis vel conicus, teris vel tetra- gonus apice acuto et labio basali exiguo. Holotype: Florida, Hillsborough Co., “sea-shore,” 6-III-1904, Fredholm 6324 (cH); isotypes: MO,us. DISTRIBUTION: coastal strand of southwestern Florida, Texas, and Tamaulipas. Flowering throughout the year. Cuthbert 1533 (Fas); Pinellas Co., Redington Beach, 20-V-1962. Patman 1088 (BM, GH,UsF); Sarasota Co., Longboat Key, 20-XI-1964, Godfrey 65244 (rsu). Texas: Aransas Co., Goose Island State Park, 11—-X-1960, Traverse 1774 (GH,SMU,TEX); Brazoria Co., Freeport Beach, 20-IV-1952, Killip 42102 (us); Cameron Co., Padre Island, 16-III-1929, Tharp 5573 (Ny,TEX,us); Nueces Co., Corpus Christi, 11—-Il- 1917, Palmer 11246 (_0,POM,TEX,UC); San Patricio Co., 8 miles S. of Taft, 2-VII-1951, Jones 590 (smu); Willacy Co., Port Mansfield, 1-V-1959, Traverse 1185 (F,MO,SMU, TEX,Us ). The plants here treated as Cakile lanceolata ssp. pseudoconstricta have usually been identified with C. lanceolata ssp. fusiformis because of their pinnatifid leaves, showy flowers, and sprawling habit of growth. How- ever, the plants develop fruits similar to and often indistinguishable from those of C. constricta which typically possesses fleshy, + entire leaves and smaller flowers. Cakile lanceolata ssp. pseudoconstricta is the predominant sea rocket on the Gulf of Mexico strand of south-central Florida and also occurs sympatrically with C. geniculata in Texas and Tamaulipas. I interpret this sympatry as a secondary phenomenon, most likely arising from the recent introduction of these plants probably in ship ballast from the Tampa Bay region of Florida. The taxon may have evolved during the Pleistocene on islands of what was, during periods of higher sea level, southern Florida. 3. Cakile edentula ( Bigelow) Hooker Succulent, herbaceous to rarely suffrutescent annuals (or fortuitous perennials), o 8 dm, much-branched, the branches spreading and ascending, im: especially early in the season; nectaries very small, the medians 0.2-0.4 mm high; stamens a first tetrad becoming nearly equal as fl es, anthers 0.4-1.4 tradynamous, satel! eo ance mm long, dehiscing introrsely; pistil lanceolate to + turbinate; stigma capitate to s 118 JAMES E. RODMAN in retuse beak; mature fruit 12-29 mm long, 4-angled or 8-ribbed, beaked, the beak + flattened in one plane and acute, blunt, or retuse at the apex, the point of articulation of segments usually + —— lower fruit segment terete, + cylindric to — slightly toward apex rt fruit segment truncate-fusiform to turbinate, 4-angled or 8-ribbed; seeds ellipsoid to oblong, dimorphic; cotyledons accumbent to occasionally incumbent. KEY TO THE SUBSPECIES A. Upper fruit segment 4-angled; articulating surface of lower fruit segment + flat to pore and with 2(6) small teeth projecting upward. __ 3a. C. edentula ssp. edentula. Upper fruit segment 8-ribbed; phenleune surface of lower fruit segment flat and ee Oe ee as 3b. C. edentula ssp. Harperi. 3a. Cakile edentula (Bigel. ) Hook. ssp. edentula KEY TO THE VARIETIES A. Upper fruit segment plump, 5-9 mm wide, short-beaked, aad beat less than the length of the seminiferous part, the base without a scarious a ee ee een ape ee eee eee Way BS ern, Pee ae ee ns 3al yea ce var. edentula. A. Upper fruit segment slender, 3-5 mm wide, tie ecko, the beak equalling or exceeding the length of the seminiferous part, the base with a slight scarious border a2. C. edentula var. lacustris. 3al. Cakile edentula ( Bigel.) Hook. var. edentula PLATE 1; Maps 3, 4, 5. Bunias i heprita sng Bigel., Fl. Bost. 157. 1814. Lectotype: Massachusetts, Boston, Bigelow 43 (1L1NN; seen on microfiche: IDC 5073. ee Ase ae Sed edentula ( Bigel.) Hook., Fl. Bor.—Amer. 1: akile lanceolata ssp. edentula (Bigel.) O. E. Eee in ae Symb. Antill. 3: 504. = Cakile americana Nutt., Gen. N. Amer. Pl. 2: 62. 1818. Syntypes strand of the sea-coast, aod also on the shores of cot North Weanin Lakes “of oa St. Lawrence.” (?BM; not seen) pro parte. Cakile maritima var. americana ( Nutt.) Torrey, Comp. Fl. N. Middle States 246. Cakile e geniculata x edentula jog ae laa. ee Bot. 2: 127. 1900. Holotype: New ors iam Island, Kearney s.n. (Ny lanceolata var. he O. E. Schulz in Urb., Symb. Antill. 3: 506. 1903. A new name for the preceding, Cakile _ var. Millspaughii (O. E. Schulz) O. E. Schulz, Pflanzenreich IV, 105; 26 Cakile gaa olata var. apetala O. E. Schulz in Urb. Symb. Antill. 3: 506. 1903. Holotype: New York, Staten Island, Rabenau s.n. (?p; not seen pees edentula subvar. apetala (0. E. Schulz) O. E. Sokuke Pilanzenreich IV, 105: 27. 1923. Cakile californica Heller, Muhlenbergia 3: 10. 1907. Lectotype: California, beach at Monterey, 29-VI-1903, H Heller 6856 (GH); isotypes: ps,F,MO,NY,PH,POM,UC,US. Cakile ne var. californica (Heller) Fern., Rhodora 24: 23. 1922. akile p. californica (Heller) Hult., Lunds Univ. Arsskr. II. Sect. 2. 41: Cakile lanceolata var. australiensis Thell. in Hegi, Ill. Fl. Mittel-Eur. 4: 183. 1916. Type: none specified. , THE GENUS CAKILE = O, ger NO) “ (*) : Gs 0) en % ne ©> © sees . Wife oS. i ° Nee ea : (*) \ eo () , (*) =O © C. edentula var. edentula ee! mn AN ee \ (*) bs (+) Big Fe ve (>) C)( }O (c) (*) (>) pO . r. age / Poa i f ) ribution of Cakile edentula v Ao pate of C. edentula where coated sik mapped ). in eastern North America (excluding Great 119 120 JAMES E. RODMAN ae © C. edentula var. edentula Q © C. edentula var. lacustris Map 4. Distribution of Cakile edentula var. edentula and C. edentula var. lacustris in the Great Lakes region. Plants fleshy, usually erect, much-branched, the branches smooth to weakly striate or sulcate, sinuous to flexuous; fruiting racemes 1-2 dm long, rarely to 3 dm, linear to icel: . : petals 4.6-7.9 mm long and 1.4-3.3 mm wide (on the Pacific Coast: 5.5-9.7 mm long and 1.4-3.2 mm wide), often fewer than 4 or reduced to bristles; anthers 0.4-1.4 mm long, debi introrsely; mature fruit 12-24 mm long, 4-8 mm wide, somewhat r ges; upper fruit segment 7-15 mm ca tangled and seat. ats 4 pores “anak ridges between the ribs, beaked, the beak short, compressed oginsinedy. acute to retuse at the apex; seeds large, ellipsoid, dimorphic. cHROMOSOME NUMBER: 2n=18, DISTRIBUTION: coastal beaches from pectin i Labrador, the Gulf of St. Law- rence, _ the St. Lawrence River east of o the Outer Banks of North sporadic on the shores of Lakes Michigan, Erie, and Ontario; naturalized on the Pacific Coast of North America and the shores of temperate a aa ceanenely introduced as a ballast weed. Flowering from June to November in its nativ ge- ATIVE SPECIMENS (native range ). Canada. Labrador: Aone Bee & Wiegand 3476 (cu,cH). New Brunswick: Bathurst, Blake 5393 emmels © ampobello Island, Smith s.n. (uc,us). Newfoundland: Cow Head, Fernald & gand 3475 (GH, K,NY); Frenchmans Cove, Mackenzie & Griscom 10295 (ext); a Ce Bridge, Rouleau 6425 (pao,Fsu CH,NY ,us). Nova Scotia: Cape Breton Co., Scatari Island, THE GENUS CAKILE 121 Smith et al. 5249 (pao); rear Co., South Cove E. of Joggins, Barrington 388 (cu); Sable Island, St. John 1241 GHX.NY.PHUS). Prince Edward Island: Cape Aylesbury, Fernald, Long & St. John 7511 (cu,pH); Souris, Erskine 1444 (pao). Quebec: Blanc Sablon, Fernald, Wiegand & Long 28314 (cu,pH); Gaspé Co., Riviere a Claude, Dore & Jones 17308 (pao); Magdalen Islands, Hospital Pond, F ernald et al. 7510 (cx, pu); Rimouski, Basset & Crompton 4266 (pao). U.S.A. Connecticut: Fair- field Co., Greens Farms, ‘Pollard 128 (us); New Haven Co., New ics Young 198 (osu); New London Co., Norwich, Setchell ~~ (xy). Delaware: Kent Co., Woodland Beach, Commons s.n. (pH); New Castle Co., 4 mile S. of River-Bay pelea Larsen Maine: Cumberland Co., Cape Elizabeth th salt marsh, Seymour 20125 (smu); Hancock Co., Seal Harbor, Mt. Desert Island, Redfield 5108 (mo,PH); Knox Co., Port Clyde, Friesner 1121 (cas,DUKE,F,NY,SMU ); Lincoln Co., Monhegan Island, Jenney, Churchill & Hill sn. (Mo); Sagadahoc Co., Hermit Island, Salamun 191 (ws); Waldo Co., Islesboro, Roesbach 3520 (Ncvu); Washington Co., Jonesport, a n. (cH); York Co., York, B Blake s.n. (Ny,PH). Maryland: Anne Aran del Co., Sandy Point, Ewan 17374 (cas, cu); Calvert Co., Scientists Cliffs, Seymour 16789 (smu,w wis); St. Marys Co., Millstone, atone 5136 (micu); Talbot Co., Deep Water Point, Earle 3817 iat, Worcester Co., Ocean City, Fogg 11448 (cn). Massachusetts: Barnstable Co., Province- town sand dened, Goodale s.n. (DUKE); Dukes Co., Gay Head, Marthas Vineyard, © C. edentula ssp. edentula eC. edentula ssp. Harperi *«. constricta Wy ay \ ra ry H \ Q 2 2 ~€ : ode wn ge ote _ Distribution of Cakile edentula ssp. edentula and ssp. Harperi and C. constricta in south- eastern United States 122 JAMES E. RODMAN shire: Rockingham Co., beach at Rye, Miller 5621 (cu). New Jersey: Atlantic Co., Atlantic City, Letterman s.n. (Ny,pH); Cape May Co., Cape May, Gershoy 325 (cu Moldenke & Moldenke 10655 (MO,NY,POM,sMuU). North Carolina: Brunswick Co., Southport, Ahles & McCrary 58377 (Ncu,smu); Carteret Co., Beaufort, Blomquis 10387A (puKE,cH); Currituck Co., Kitty Hawk, Godfrey 5274 (cu,us); Dare Co., Co., Block Island, Fernald & Long 9544 (cu,pH); Washington Co., Narragansett, Williamson s.n. (pH). Virginia: Accomack Co., S. end of Assateague Island, Iltis 24110 U ancaster Co., Wind- mill Point, Edwin 389 (rsu); Middlesex Co., Urbanna, Hermann 10426 (cu); Norfolk Co., near Ocean View, Kearney 1225, 1448 (us); Northampton Co., west of Kiptopeke, Fernald, Long & Fogg 5309 (cu,pH); Princess Anne Co., Virginia Beach, Heller s.n. (pH); Warwick Co., Newport News, Bartram s.n. (pH); Westmoreland Co., Camp Potomac, Iltis 148 (smu). Between Botany Bay and Sydney Head, XII-1871, Daintree s.n. (MEL); Currarong, 7-X-1960, Constable s.n. (a,Nsw); Hat Head, E. of Kempsey, 18-I-1953, Constable McDonald 435 ( BRI); Deception Bay, 13-XI-1932, White 8684 (sri,uc); Heron Island, 5-X-1960, F osberg 41300 (Nnyx,us); Palm Beach between Currumbin and 1922, White 1680 (sri,nsw). South Australia: Beachport, 1—XII-1917, . (aD); Fowler’s Bay, 12-IX-1960, Wilson 1598 (ap); Head of Great Australian Bight, 6—-I-1971, Orchard 3195 (ap); Kangaroo Island, D’Estrees Bay, XII-1881, Tate s.n. (ap); Victor Harbour, II-1921, Cleland s.n. (ap). Tasmania: Kings Island, 1882, Spong s.n. I Bay, 5-VIII-1931, Eyerdam 525 (~M,cas,ps ); Middleton Island, 3-VIII-1956, Thomas 6360 (ps,us); Yakutat, 14-VII-1945, Stair s.n. (pH). California: Alameda Co., beach at West Berkeley, VII-1882, Greene s.n. (cH); Del Norte Co., Crescent City, 19-IX-1912, Eastwood 2281 (cas); Humboldt Co., Samoa, 7-VIII-1901, Tracy 1266 (uc); Los geles Co., Anaheim Landing, 7-X-1930, F osberg 53174 (ny); Marin Co., beach near Bear Valley, 9-IX-1897, Eastwood s.n. (F); Mendocino Co., Fort Bragg, 1914, Mathews 161 (uc); Monterey Co., Monterey Bay, 1887, Hickman s.n. (Np-c); Orange 0., Newport, 7-IX-1907, Davidson 1770 (ps); San Diego Co., San Diego, VI-1906, Brandegee s.n. (Ny,uc,u ); San Francisco Co., San Francisco, 11-VII-1893, Congdon s.n. (ps); San Luis Obispo Co., Morro sand dunes, 23-V—1908, Condit 22 (uc); San Mateo Co., Spanish Town, 1893, Dudley s.n. (ps); Santa Barbara Co., Surf, V-1902, Elmer 3628 (Ds,GH,MO,Ny,us); Santa Cruz Co., Capitola, 4—-VIII-1894, Burnham s.n. (cu); Sonoma Co., Bodega Bay, 21—-VII-1932, Purer 4040 (sp); Ventura Co., Ventura, 17-IV-1916, Eastwood 5024 (cas). Illinois: Cook Co., Lake Shore, 12—-VII-1891, Moffat 162 (wis). Indiana: Porter Co., Mineral Springs, 6-IX-1911, Wolcott 71 (¥). THE GENUS CAKILE 123 Michigan: Berrien Co., St. Joseph, 27-IX-1949, Jordal 2431 (micu); Mason Co., Ludington, 25—VII-1959, Miller & Milstead 169 (Ncu). New York: Erie Co., “Shore Lake Erie,” Kneiskern s.n. (Nyx). Oregon: Clatsop Co., Clatsop Spit, XI-1902, Stubenrauch s.n, (uc); Coos Co., Charleston Bay, 6-V-1911, Smith 3667 (F,Ny); urry Co., Rogue River, VIII-1917, Hawkins s.n. (wis); Lane Co., W. of Florence, 28-VIII-1949, Cronquist 6104 (CcAs,GH,MICH,NY,RSA,SMU,UC,Us ); Lincoln Co., New- port, 1917, Hawkins s.n. (wis); Tillamook Co., Nestucca, VIII-1901, Kirkwood 123 (Ny). Washington: Clallam Co., Port Angeles, 26-VI-1908, Flett 3381 (us); Grays Harbor Co., Copalis, XI-1911, Foster 1534 (Ny,us); Island Co., Deception Pass State Park, 30-VII-1936, Smith 1384 (uc); Jefferson Co., Pt. Hadlock, VII-1931, Jones 3279 (pH); Pacific Co., Long Beach, 13-VIII-1907, McGregor s.n. (ps); San Juan Co., Argyle, 1-VIII-1917, Zeller & Zeller 866 (cu,Mo,Ny,us); Whatcom Co., Pt. Roberts, 9-VII-1937, Muenscher 7896 (cu). Wisconsin: Kenosha Co., Kenosha, 11—IX-1932, Wadmond s.n. (wis); Manitowoc Co., Point Beach State Park, 2-IX-1962, Clarke 3-2 (wis); Milwaukee Co., Atwater Beach, 24—VII-1939, Shinners 629 (sp); Sheboygan Co., Cedar Grove, 21-IX—1958, Iltis 12732 (wis). Ball (1964a) expressed doubt about recognizing Cakile edentula as being distinct from C. maritima. His argument was based on the assump- tion that the Icelandic sea rockets represented a subspecies of C. eden- tula, which would encompass a range of variation for several characters overlapping that of C. maritima. The Icelandic plants, however, consti- tute a distinct species; when they are excluded, C. edentula remains easily distinguishable from C. maritima on the basis of vegetative, floral, and fruit morphology and breeding system. The southern limit of Cakile edentula var. edentula on the Outer Banks of North Carolina corresponds closely to the influence of the Labrador Current which reaches southward nearly to Cape Hatteras (Cotter 1965). A number of coastal plants with which typical C. edentula is associated, such as Ammophila breviligulata and Hudsonia tomentosa, also reach their southern limits on the Outer Banks (cf. Radford, Ahles & Bell 1968). The range of C. edentula var. edentula thus appears to corre- spond to a natural biogeographic area, at least for coastal organisms, which is undoubtedly strongly influenced by coastal currents. Recently, Patman (1962) reported Cakile edentula var. edentula from Indian River, St. Johns, Santa Rosa, Volusia, and Wakulla counties in Florida. The plants probably are C. constricta, which is vegetatively sim- ilar to C. edentula var. edentula but has larger flowers and much thinner fruits. Sporadic, ballast plants of C. edentula would not be wholly unexpected, however. Within populations of typical Cakile edentula, on both the Atlantic and Pacific coasts, one usually finds a varying proportion of plants with very purplish stems and young fruits and purple petals. Marie-Victorin (1935) early pointed this out and suggested that nomenclatural recogni- tion be accorded these plants. However, since they appear to represent only color morphs within populations, they deserve nothing more than forma status if any. I have the impression, gained from my field work, that the proportion of purplish forms increases northward on both coasts, but it would require a more thorough population study to document this. 124 JAMES E. RODMAN 3a2. Cakile edentula ( Bigel. ) Hook. var. lacustris Fernald Pirate 1; Map 4. Cakile edentula var. lacustris Fern., Rhodora ie 23. 1922. Holotype: Indiana, Lake Co., Millers, 4-IX-1911, Sherf s.n. ( (cu): isotype: Cakile edentula ssp. lacustris (Fern.) Hulkt., Lands Univ. Arsskr. II. Sect. 2. 41: 824. 1945. Cakile lacustris ( Fern.) Pobed., Nov. Syst. P]. Vasc. 1964: 108. 1964. Plants generally less robust and re st the typical Maye posi usually with 4 petals; petals 5-7 mm fog 8 and 1.5-2. wide; anther 1.3 mm long; sige. pedicels divergent, 2-8 mm long and nant wide; eae Hae 16-26 mm long, mm wide, slightly seuaitotea at the point of igihedlaon: lower fruit segment + nog : cylindric, parece: 8 lon ng, the articulating surface + flat and usually with 2 small teeth Ganoatcsonte NosEDER: on= DISTRIBUTION: sandy beaches of the Great Lakes of North America and bed sporadic on the St. Lawrence River east of Quebec. Flowering from May to October REPRESENTATIVE SPECIMENS. Canada. Ont ario: Ipperwash Beach, isbn ne llan ( GH,NSW,TEX); it Stanley: 26-VIII-1880, Burgess s.n. (DAO); Rondeau Provincial Park, sie msi Iltis 24104 (smu,wis); Sault Ste. Marie, Pitcher s.n. (Ny); Shore of Lake a Macoun 180 (cu). U.S.A. Illinois: Cook Co., Chicago, 1860, Scam 4a (¥,cH); Lake Co., Waukegan, 28-VITI-1908, oe ian on MICH). nalts ae Co., Whiting, 25-VII-1875, Hill s.n. (F); Michigan City, 7-IX-1903, Lansing 1901 (¥); Porter Co., Dune Park, Oe VIL-1806. Chase 472 (pH). Michigan: Alcona Co., 4 miles N. of Ha rrisville, 12—VIII-1951, McVaugh 12520 ety Allegan Co., . miles S. of Douglas, 10-X-1948, Cain, egadas—Vianna % Hagenah s.n. (pao); Arenac Co., Point Lookout, 10-VII-1900, hr 360 (msc); Benzie Co., Betsie Point, 16—-VIII-1951, McVaugh 12626 so ae . St. Jose . 10-VIII- 1838, ibe $.n. feeston Cheboygan ner lodge, 12-VIII-1961, Bourdo 3218 (msc); Mason Co., Ludington, 1910, abate 212 (¥,NY stanley onroe a Monroe, siete ae Chandler s.n, (us) fal 1838, Clinton s $.n. rapes Msc an. Monroe Co., tectum separated from each other by a thin layer not parallel to the nexine (zig-zag). ts e ( Moscharia is a member of the tribe Mutisieae subtribe Naussauviinae which is characterized by bilabiate corollas with the outer lip three- toothed and the inner lip bifid, anthers long-tailed at the base, and style bifid with branches truncate at the apex. Moscharia can be easily distin- ished from all the other genera of the subtribe by the presence of pseudocephalia composed of 8-10 capitula, attached at the same level and by outer capitula with the involucre reduced to two bracts, one of them keeled. The greatest affinity of Moscharia is with Polyachyrus, which also has pseudocephalia but with many more than 10 capitula; the capit- ula are disposed in the pseudocephalium at different levels. The name of the genus comes from Latin for “musky” because of the strong musky odor of the plants. The common name is “almizclillo” or “almizcle,” which in Spanish also means musky. The two species of Moscharia can be easily distinguished as shown below: Moscharia pinnatifida Moscharia solbrigii Pseudocephalia with pseudoinvolucre Pseudocephalia without pseudoinvolucre Flowers lavender eeled bract with apex acute, abaxial rface densely pubescent, oT fleshy, Flowers pale pin Keeled bract with apex incised, abaxial < surface scarcely pubescent, hump not su fleshy, bract conduplicate around the bract conduplicate aroun e outer two flowers and the i ct. ver and the inner bract, not the inner flower. Only the outer flower of the outer capit- ula with a pappus. All flowers with a pappus. JORGE VICTOR CRISCI % a me Ee 2 27 Zz Ae 4 | f ba ] zs | =a | t + LJ 4 ¢ + | + 28° i 4 + - a + * = s ‘ eS ( d ¥ did a aie sane ota ; cae xO Pad * + \ . \ , o > M on 9 “Nd COQUIMBO, ? 30° ceria ARR Fe x to nk 5 A es 2 pe! *. * (b t + ® + 3 a M. PINNATIFIDA ye | x a tx? j t } »* : i M.SOLBRIGI! *% | t < | “ | + | ieee — » 32° = Giea rg ~. ‘ ~~ 4 oe *, ee Ht { t t * * vw > : ae = > 17] ° 33 ele er hoc ‘ * + ¥ dice ang ag 2 i! as eee, 34 fy + po i ne \ ; yw! | + i a” | ¢ | of | Se + | a Yar 9 s TALCA can — 4, | \ ¢ i Meno ? NN, t ‘, nine ~~ ys ae 36 ‘ AG ic i ; ey . wn f + a < e na Pa rey 3 AVE ~. et ad A \ + Map I. Distribution map of the species of Moscharia Ruiz et Pavén. THE GENUS MOSCHARIA 169 Moscharia pinnatifida Ruiz et Pavon Figure 4 Moscharia pinnatifida Ruiz et Pavon, Systema Vegetabilium: 186. 1798. Moscaria pinnatifida (Ruiz et Pavén) Persoon, opsis Plantarum 2:379. 1807. Moschifera pinnatifida (Ruiz et Pavén) Molina, Saggio 2: 294. 1810. ieee : pr ngel, Plants annual with a strong odor, 30-70 cm. tall, stem erect, pubescent, much branched toward the apex, with leaves all along it. Basal leaves and the 5 or 6 lowest rmos in pseudocephalia. Pseudocephalia hemispherical, mm. in diameter and 10 mm igh. ag eoriens ie of 5-8 foliaceous bracts; bracts acute at the apex, densely Liga n the adaxial surface, ovate, membranaceous, 6-8 mm lon a the apex, 4-5 mm. igh; inne gee scarcely jae one nt on ore margin, ee lanceolate, 3.5-4 mm. long aad 0.5-0.8 mm. wide, 2 or 3 toothed at the a flowers two, hermaphroditic; the inner flower oie a pappus. Inner capi um Mr an aaah of 4-7 bracts, bracts lanceolate, scarious, 2 or 3 toothed at the apex, m. long and ide; flowers n mm. wide; > rmaphroditic, with a pappus Corclla, pale pink, pubescent, eaganaone . long; tube 4-6 mm. long, 0.5-0.7 mm in diameter at base and 1—1.1 mm. at apex; outer lip 4-veined, three toothed at apex, 3- . long and 2-2.5 mm aie inner lip bifid, 3-4 mm. long, each lobe 0.5-0.6 mm. wide. Anthers tailed, glabrous 5-6 mm. long; style bifid, the branches 1.3-1.5 mm. long. Achenes 1-. m. long, 0.3-0.4 mm. in diameter, scarcely pubescent. Pappus of hae scales 1 mm. high, uniseriate, white, present only in outer flowers of outer capi DISTRIBUTION: Central Chile from Coquimbo to Maule, from sea level to almost 2,000 meters altitude; flowering from August to December. SPECIMENS EXAMINED, Chile. Provincia de Coquimbo. Fray Jorge, Skottsberg 755 (¥,Ny), Mufioz 209 (cx); La Serena, Punta de Teatinos, Wedermann ( ¥,GH,M,NY, uc,us); La Serena-Vallenar route, Morrison 16281 (cH,uc); Coquimbo, Harvey s.n., a sea de ao neage 294 (uc); Los Chorrillos, Poeppig 237 (nx). Provincia de Val- . Val : Buchtien, 1895 ae Claude Joseph 3575 (us), Bertero s.n. te S oitis), Borchics s.n., 1883 (F), Jaffuel 552 (cH), Behn s.n., 1922 (F); Curauma: Behn Sn, 1929 (F,uc); Laguna Verde: stein 1642 (cH); Quillota: Cerro Caquis, Morrison 16898 (cuH,Uc). Provincia de Aco’ = Zapallar: Quebrada del Barraco, Kausel 2585 (¥). Provincia de Santiago. pony ae erro San Cristobal: Skottsberg 981 (¥,Nx), Hastings 80 (Ny,uc,us),Montero 00 (¥), Looser 678, 776 (cH), Kausel 2475 (¥); Santiago: Philippi s.n., 1888 (us), ay Joseph 513, 514 (us), Germain s.n. (F); (ny); Gillis s.n. bes fe ertero - (yx); 1834 (cH); Caldcleugh s.n. (F, ex G syn- type of Gastrocarpha ncinata?); Rio Claro, Guillemin s.n. (cH), Bertero s.n., 1828 (cH); Mufioa, Claude Joseph 2097 (a us). This species is very widespread and abundant. According to Reiche 170 JORGE VICTOR CRISCI Te <—.sS =) = tot Py f ay ze THE GENUS MOSCHARIA 171 RP 2 2 AZ \ ao _ Moscharia solbrigii Crisci sp. nov Teale ¢é Morrison = a. Habit,