JOURNAL 57655 OF THE tA ARNOLD ARBORETUM HARVARD UNIVERSITY B. G. SCHUBERT EDITOR T. G. HARTLEY C. E. WOOD, JR. D. A. POWELL CIRCULATION VOLUME 50 PUBLISHED BY THE ARNOLD ARBORETUM OF HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS 1969 MissouRi BOTANICAL DATES OF ISSUE No. 1. (pp. 1-158) issued 15 January, 1969. No. 2 (pp. 159-314) issued 15 April, 1969. No. 3 (pp. 315-480) issued 15 July, 1969. No. 4 (pp. 481-663) issued 17 October, 1969. ie . TER, INDIANA amy alee be Li TABLE OF CONTENTS CRETACEOUS ANGIOSPERM POLLEN OF THE ATLANTIC COASTAL PLAIN AND ITS EVOLUTIONARY SIGNIFICANCE. James A. Doyle COMPARATIVE ANATOMY AND RELATIONSHIPS OF COLUMELLIACEAE. William L. Stern, George K. Brizicky, and Richard H. Eyde “Notes on DIstTRIBUTION AND HasiTat oF COLUMELLIA. George K. Brizicky and William L. Stern THE Ecotocy oF AN EvFIN Forest IN Puerto Rico. HILutTor AND Forest INFLUENCES ON THE MICROCLIMATE oF Pico pet Orestr. Harold W. Baynton 4. TRANSPIRATION RATES AND TEMPERATURES OF LEAVES IN Coot Humip ENVIRONMENT. David M. Gates ................ 5. CHROMOSOME NUMBERS OF SOME FLOWERING PLANTS. Lorin I. Nevling, Jr. THE GENERA OF SENECIONEAE IN THE SOUTHEASTERN UNITED States. Beryl Simpson Vuillewmier ASPECTS OF THE CoMPLEX NopaL ANATOMY OF THE DIOSCOREA- CEAE. Hdward S. Ayensu ANATOMY OF THE PALM Ruapis EXCELSA, VII. Fiowers. N. W. Uhl, L. O. Morrow, and H. E. Moore, Jr. GLYCOSMIS PENTAPHYLLA (RUTACEAE) AND RELATED INDIAN AxA. R. L. Mitra and K. Subramanyam VASCULAR ANATOMY OF MONOCOTYLEDONS WITH SECONDARY GrowTH — an IntTRopucTION. P. B. Tomlinson and M. H. Zimmermann se ASPECTS OF REPRODUCTION IN SauRAvUIA. Djaja D. Soejarto .... THe Eco.ocy or AN ELFIN Forest IN Puerto Rico. Attnraz,-Roors: A. My Galle .o.0....c.cccc cece 7. Som, Root, anp EarrHworm Rexationsuips. Walter H. Lyford Srupies or Stem GrowTH AND ForM AND oF LEAF StRUC- TURE. Richard A. Howard LEctToryPiricaTion or Cacauia L. (ComMPosITAE-SENECIONEAE). Beryl S. Vuilleumier and C. E. Wood, Jr. ...........cc00cc A REVISION oF THE MALESIAN AND Paciric RAINFOREST CONIFERS, I. Popocarpacrar, IN PART. David J. de Laubenfels ............ A REVISION OF THE MALESIAN AND PaciFic RAINFOREST CONIFERS, I. Popocarpacran, IN part (Concluded). David J. de Lau- benfels .. i THE Vascuar System In THE Axis oF DRACAENA FRAGRAN (AGAVACEAE), 1. DistRIBUTION AND DEVELOPMENT OF PRI- MARY Stranps. M. H. Zimmermann and P. B. Tomlinson .... 9 370 COMPARATIVE MorPHOLOGICAL STUDIES IN DILLENIACEAE, IV. ANATOMY OF THE NODE AND VASCULARIZATION OF THE LEAF. William C. Dickison ANATOMY AND ONTOGENY OF THE CINCINNI AND FLOWERS IN NANNORRHOPS RITCHIANA (PALMAE). Natalie W. UAL ........ Aspects OF MorpHoLoGy or A US FORMOSANA WITH A OTE ON THE TAXONOMIC POSITION OF THE GENUS. Hsuan Keng A KaryoLocicaL Survey or Lonicera, II. Lily Riidenberg and Peter S. Green Nores on West Inp1an Orcuins, I. Leslie A. Garay ...........00000...: POLLEN CHARACTERISTICS OF AFRICAN SPECIES OF VERNONIA. C. Earle Smith, Jr. A New Species or Ficus rrom Suriname. Gordon P. DeWolf, Jr. A REVISION OF THE GENUS FLINDERSIA (RUTACEAE). Thomas G. Hartley A Stupy oF THE GENUS JOINVILLEA (FLAGELLARIACEAE). Thomas K. Newell THE Ecouocy or AN ELFIn Forest 1x Purrro Rico. 9. CHEMICAL STUDIES OF CoLorED Leaves. Richard J. Wagner, Anstiss B. Wagner, and Richard A. Howard .... THE GENERA OF PORTULACACEAE AND BASELLACEAE IN THE SOUTH- EASTERN Unitrep States. A. Linn Bogle STUDIES IN THE NortTH AMERICAN GENUS FOTHERGILLA (HAMAME- LIDACEAE). Richard E. Weaver, Jr. Tue Tripe MuvtTIsIEAE (COMPOSITAE) IN THE SOUTHEASTERN Unirep States. Beryl Simpson Vuilleumier A New Species oF ARENARIA FROM THE BHuTAN HIMALAYA. N.C. Majumdar and C. R. Babu THE Drrector’s REPORT INDEX TO VOLUME 50 VoLuME 50 NuMBER 1 JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY B. G. SCHUBERT EDITOR T. G. HARTLEY C. E. WOOD, JR. D. A. POWELL CIRCULATION Be ER ISS : PUBLISHED BY _~CO THE Sees ARBORETUM OF HARVARD UNIVERSITY — : _ CAMBRIDGE, MASSACHUSETTS THE JOURNAL OF THE ARNOLD ARBORETUM Published quarterly by the Arnold Arboretum of Harvard University. Subscription price $10.00 per year. Volumes I-XLV, reprinted, are available from the Kraus Reprint Corpo- 10017 RATION, 16 East 46TH Street, New York, N.Y. 1 PE pect ei and remittances should be addressed to Miss Duucie A. PowELt, ARNOLD ArBorETUM, 22 Drviniry AVENUE, CAMBRIDGE, Massa- CONTENTS OF NUMBER 1 CRETACEOUS ANGIOSPERM POLLEN OF THE ATLANTIC COASTAL PLAIN AND ITS EVOLUTIONARY SIGNIFICANCE. James A. Doyle CoMPARATIVE A AND RELATIONSHIPS OF COL COLUMELLIA- CEAE. William, ‘® 1 tom George K. Brizicky, and — H. Eyde Nores on DistrisuTion aND HABITAT OF COLUMELLIA. Georg K. Brizicky and William L. Stern seu pre ese ELFIN gence: In Puerto Rico. . Hixtor anp Forsst CES ON THE MICROCLIMATE or Pico pEL OxsTE. Harold W. Baynton 4. TRANSPORTATION RATES AND TEMPERATURES OF IN _— oS p ENVIRONMENT. David M. Gates ............ : CEAE. "Edward 8 Ayers “Tock “Fe AE a ’ (Roracean) AND Retarep Brows 76 JOURNAL OF THE ARNOLD ARBORETUM Vor. 50 JANUARY 1969 NUMBER 1 CRETACEOUS ANGIOSPERM POLLEN OF THE ATLANTIC COASTAL PLAIN AND ITS EVOLUTIONARY SIGNIFICANCE James A. DOYLE ONE OF THE MAJor problems in the study of the evolution of higher plants is the paucity of evidence from the fossil record on the origin and evolution of the angiosperms. Because of the relatively sudden appearance of angiosperms in the fossil record, the lack of recognized angiosperm pre- cursors, and the lack of any striking peculiarities of the macroscopic remains of Lower Cretaceous angiosperms (mostly leaves), almost all conclusions on the origin of the group and the nature of its primitive members have been based on comparative studies of its living representatives. Angiosperm paleobotany has been primarily concerned with the geographic vegetational and floristic changes in the Tertiary, which were due more to migration and extinction than to evolution. The methods of Tertiary angiosperm paleobotany, such as the procedure of identifying modern taxa for paleo- ecological information, have been far less productive in the Cretaceous, and the problematical nature of the results is undoubtedly largely respon- sible for the present low level of activity in Cretaceous megafossil paleo- botany. In the past decade a new method has been applied in Cretaceous paleo- botany which promises to shed light on the problems of angiosperm origin and evolution. This is the study of fossil pollen and spores. In contrast to the megafossil record, the palynological record shows clear-cut changes in the morphology and diversity of the angiosperms from the time of their first appearance. Most of the information on early angiosperm pollen has been obtained for stratigraphic purposes (cf. Couper, 1964), and is, from a botanical point of view, largely descriptive; the evolutionary implications have been frequently mentioned but not discussed in detail. The purpose of this paper is to review the characteristics of Lower and early Upper Cretaceous angiosperm pollen floras and to discuss the evo- lutionary and phylogenetic implications of the record. Much of this dis- cussion is based on my own work on Cretaceous angiosperm pollen of the Atlantic Coastal Plain, which appears to give a fairly representative picture of the world flora. Comparisons will be made with correlative sequences, which are now known in many other parts of the world. Much of the 2 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 detailed documentation of both the stratigraphic and systematic aspects is in progress and will be presented later in more complete form, but the general results seem clear enough to be summarized in this preliminary paper. In general, morphological terminology follows Erdtman (1952), though some terms of Pflug (1953) are used in discussing triporate pollen. GEOLOGICAL BACKGROUND The Cretaceous is one of the longer geologic periods, covering some 72 million years between about 136 and 64 million years before the present (Casey, 1964). The Cretaceous System is customarily divided into Lower and Upper Cretaceous series, which are subdivided into six stages each (TABLE 1). These stages were first recognized in western Europe and subsequently extended around the world; they are now operationally defined by ammonite zones of the Tethyan province. Continental or near-shore marine sediments favorable for palynological TABLE 1. Subdivisions of the Cretaceous SERIES STAGES Maestrichtian =) Campanian Santonian Senonian (most common usage ) Upper CRETACEOUS < Coniacian Turonian Cenomanian Albian Aptian i Barremian LOWER CRETACEOUS < Re Hauterivian : Neocomian (most common usage) Valanginian | Berriasian studies are fairly extensively developed in the Cretaceous, though few areas have large parts of the system represented by continuous continental deposition. For example, the uppermost Jurassic and much of the Lower Cretaceous are well represented in the Purbeck and Wealden of southern England, but late Lower Cretaceous rocks there are marine and only marginally suitable for palynological study. The Upper Cretaceous consists of the wholly unsuitable marine Chalk, and to extend the European Cretaceous pollen record we must go to Central Europe. TABLE 2, Presumed stratigraphic relations of ae Coastal Plain nmarine Cretaceous formation TIME-STRATIGRAPHIC UNITS ROCK-STRATIGRAPHIC UNITS SERIES STAGES SOUTHERN AND CENTRAL MARYLAND AREA RARITAN Bay AREA, NEW JERSEY ?—?P—? [6961 Cliffwood on Magothy Formation Magothy Formation Morgan Santonian Amboy nee Clay —?—?—?— Hiatus Coniacian —?—?—?— Upper Hiatus Old Bridge Sand Turonian South Amboy Fire Clay Raritan Formation Sayreville Sand es Bh Mek SS Woodbridge Clay Farrington Sand Raritan Fire Clay Cenomanian Subsurface Only “Raritan” Formation —?— —?—?—?— Subzone B Patapsco Fm. —Zone II— Albian Potomac Group Subzone A Lower —?—?—?— Arundel Clay Aptian a Zone I Barremian? Patuxent Fm. NATIOd WUAdSOIDNV SAOADVLAMO “ATAOG : JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 An excellent section of the late Lower and early Upper Cretaceous for pollen studies is found in the Atlantic Coastal Plain between Virginia and New Jersey. Pre-Campanian deposition in this area took place mostly in river flood plains and deltas. The result is a seaward-dipping wedge of unconsolidated clays, sands, and gravels, often very rich in organic matter, which is exposed as a wide northeast-trending band several hundred feet thick at the landward margin of the Coastal Plain. The presumed strati- graphic relations are shown in TABLE 2. In Maryland and adjacent states the basal unit is the Potomac Group, which until recently was defined as consisting of the Lower Cretaceous Patuxent, Arundel, and Patapsco formations (cf. Clark e¢ al., 1911). The Patuxent tends to be feldspar-rich and sandy or gravelly; the Arundel is a huge lens of dark, organic-rich clay with siderite nodules, while the Patapsco is rather heterogeneous, though its red and variegated clays are most characteristic. As is often the case with continental sediments, Potomac Group lithologies are highly variable, and there is much doubt that the formations can be consistently separated in the field. The Arundel, which was apparently deposited in a swamp belt, is the greatest exception to this, and it is an important marker in dividing the Potomac Group; however, it is definitely present only in the area between Washington, D.C., and Baltimore Co., Maryland. To these three formations have recently been added higher beds tradi- tionally designated Raritan Formation (Weaver et al., 1968), which appear to be earliest Upper Cretaceous. These sediments are generally sandy and lacking in fossils; the few samples that have been examined palynologically (discussed below) indicate an age between the typical Patapsco and the type New Jersey Raritan. In the absence of any distinctive lithological similarity to the type Raritan, this Maryland “Raritan” should probably be considered either part of the Patapsco or a new formation of the Potomac Group. In the Raritan Bay area of New Jersey, Coastal Plain deposits begin with the Upper Cretaceous Raritan Formation, which appears to be largely of deltaic origin (Owens et al., 1968). The Raritan has been divided into six locally recognizable members, excluding the Amboy Stoneware Clay, which on palynological and other grounds is better associated with the overlying Magothy Formation (Wolfe & Pakiser, ms.). The palynologically important Woodbridge Clay member is dark, massive, highly organic, and siderite-bearing, like the much older Arundel Clay of Maryland, but the South Amboy Fire Clay member is more varied lithologically, with lignitic beds and sands as well as dark, often laminated clays. Most of the Raritan Formation consists of deltaic sands. The last nonmarine Cretaceous unit, the Magothy Formation, occurs in both Maryland and New Jersey. It unconformably overlies both the Potomac Group and the Raritan Formation; this unconformity represents a considerable hiatus in deposition even in New Jersey, where older Upper Cretaceous is present. The Magothy is lithologically distinctive and unlike the lower units: it consists mostly of alternating sands and dark clays with “eo 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 5 considerable lignitic material, and it shows much more continuity of individual beds. It appears to be a deltaic deposit, with evidence of tidal influence (Glaser, 1967); it is overlain by the often glauconitic offshore shelf sediments of the Matawan and Monmouth groups. Unfortunately, controls on the age of the Atlantic Coastal Plain con- tinental units from marine fossils are poor. Except for an aberrant brackish water fauna of uncertain age from a deep well on the Eastern Shore of Maryland (Anderson, 1948), marine fossils are unknown in the Potomac Group. Marine mollusks from the Woodbridge Clay in New Jersey, recently restudied by Sohl (pers. comm.), date that unit as middle or late Cenomanian, while a late Santonian ammonite was recently found in the Magothy Formation of New Jersey (Owens & Sohl, pers. comm.). Bio- stratigraphic correlations must therefore be based almost entirely on the plant fossils, of which the pollen and spores are by far the most useful and readily obtained. Palynological study reveals a sequence of biostratigraphic zones which are consistent with the regional lithostratigraphy, and which compare closely with better dated sequences in other parts of the world. It is the angiosperms, which were apparently undergoing rapid evolutionary diversification in the mid-Cretaceous, that are most useful in defining these zones. Though questions might be raised on the exact correlation of the angiosperm pollen assemblages in the absence of independent age control, the relative times of appearance of major types are the same as elsewhere, and correlations based on the angiosperms agree well with those made with the spores and gymnosperm pollen alone. PATUXENT AND ARUNDEL FORMATIONS The pollen and spore flora of the lower two formations of the Potomac Group, first described in detail by Brenner (1963), is representative of mid-Lower Cretaceous floras of most of the world, just before the appear- ance of typical angiosperm pollen. It is dominated by pteridophytes and gymnosperms, notably: the fern families Cyatheaceae (or Dicksoniaceae), Schizaeaceae (Cicatricosisporites, Appendicisporites, and possibly Trilobo- sporites, Concavissimis porites, etc.), and Gleicheniaceae, as well as groups of less certain affinities; conifers representing the living families Pinaceae, Podocarpaceae, Cupressaceae (or Taxodiaceae), and Araucariaceae, and extinct forms such as Classopollis (which was apparently produced by plants known as the megafossil genera Cheirolepidium, Brachyphyllum, and Pagiophyllum, Pocock & Jansonius, 1961), the possibly related Exest- pollenites tumulus Balme, and the last of the seed ferns, the Caytoniales (Vitreisporites). Smooth monosulcate grains probably represent the gym- nospermous orders Cycadales, Bennettitales, and Ginkgoales, while the Gnetales are represented by grains of the Ephedra type. The picture of the flora obtained from the pollen and spores is in general agreement with that provided by the megafossils, which are also predominantly ferns, conifers, and cycadophytes (Fontaine, 1889; Berry, 1911). The genus Eucommiidites, common throughout the Potomac Group, 1s 6 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 of special interest since it was first described by Erdtman (1948) as a dicot from Jurassic rocks. Eucommiidites pollen is smooth and medium-sized, with three furrows which initially suggest the tricolpate condition typical of and restricted to dicots (Fics. 1a,b). However, one of the furrows is wider and more cycad-like than the others, and the general shape of the grains was shown by Couper (1958) to be more like that of monosulcate gymnosperm than tricolpate angiosperm grains. Subsequently, Eucom- miidites has been found in the micropyles of gymnospermous seeds in both England (Hughes, 1961a) and Virginia (Brenner, aria It was pre- sumably produced by an extinct group of gymnosper The Patuxent-Arundel flora does include one oe Clavatipollenites Couper, which has distinctive angiosperm characters. Clavatipollenites is generally monosulcate, with the exine finely pilate (clavate), retipilate (with free pila arranged in a reticulum), or reticulate (with the heads of the pila fused to form a true reticulum). Couper (1958), in describing the type species C. hughesii from the Barremian of England, pointed out that while the monosulcate aperture condition is prevalent in gymnosperms, pilate or retipilate sculpture is not known outside the angiosperms, and he noted the similarity of the grains to those of Ascarina in the dicot family Chloranthaceae. Clavatipollenites has been widely reported from the middle and late Lower Cretaceous: the Barremian through Albian of England (Hughes, 1958; Kemp, 1968), the Aptian and Albian of Portugal (as Apiculatisporis vulgaris Groot & Groot, 1962), the Barremian through Albian of West Africa and the Aptian and Albian of Central America (Couper, 1964), the Albian of Australia (Kemp, 1966), presumed pre- Albian rocks of southern Argentina (Archangelsky & Gamerro, 1967), and the late (Norris, 1967) and middle (pers. obs.) Albian of the Canadian Plains. A supposed latest Jurassic or earliest Cretaceous species, C. couperi Pocock, from Canada (Pocock, 1962) and Egypt (Helal, 1966) is dissimilar in its exine structure and is probably a cycadophyte (Pocock, pers. comm.; cf. Kemp, 1968) Clavatipollenites is so variable that it undoubtedly represents several natural species. The coarseness of the sculpture varies greatly, and there is every degree of fusion of the heads of the pila, up to a good reticulum with large lumina. The grains usually have a simple sulcus, consisting of a granulate or irregularly sculptured area in the pilate forms (Fics. Ic-e), or a well-delimited unsculptured membrane in the reticulate ones (Fics. 1f,g). But especially in the overlying Patapsco Formation, the pilate grains often have a more irregular, sometimes trichotomosulcate aperture (Fic. 1h), as figured by Groot and Groot (1962) as Apiculatisporis vulgaris from Portugal, or they may be inaperturate or have several weak colpoid areas (Fic. 1i).? * Recently Hedlund and Norris (1968. Spores and pollen grains from Fredericks- burgian (Albian) strata, Marshall County, Oklahoma. Pollen et Spores 10(1): 129- hese grains o be essentially identical to the irregular-aperturate specimens of Fg eae ney Hae the Potomac Group, but they show much more complete 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 7 The distinction between retipilate grains with an irregular sulcus and reticulate grains with a clearly defined sulcus and a te ndency for the reticulum to detach is important in Kemp’s (1968) separation of Clavati- Fic. 1. Potomac Group gymnosperm and probable angiosperm pollen. Num- bers in parentheses refer to slides. a and b, Eucommiuidites troedssonii, grain sa main sulcus on upper side, two od levels (Aq 45-Ic: Patuxent d, and e, Clavatipollenites sp., flattened pilate grain with sulcus on : lower si (Ag 27-1g: Patuxent Fm., Zone I); f and g, C/ P aseaierigsa or se culate grain, two focal levels (Aq eis. _Patu t Fm., ) a Clixaipllentes sp., trichotomosulcate grain (B 27-Ic: pores a Subzone B ); 1, Clavatipollenites sp., tetrac olpoidate grain 651-2 : Patapsco Fm., ae B of Zone II); j and k, Peromonolites sp. (sensu pie, grain with sulcus (?) on nie side, two focal levels (Aq 18-lc: Patuxent Fm., Zone I). All figures * 1000. intergradation from sulcoidate to colpoidate. This gives gaged support to the a pothesis that zonaperturate (including tricolpate) pollen is derived from monosul- cate through tri-, tetra-, or pentachotomosulcate Lamar 8 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 pollenites hughesii and her species C. rotundus. She also found size and shape were reliable characters, but the retipilate forms in the Potomac Group are much more variable in size and shape than C. hughesii in England, overlapping considerably with C. rotundus, while the reticulate forms often lack the characteristic infolding of the sulcus margin of C. rotundus and sometimes have coarser sculpture. Brenner (1963) referred the reticulate forms, which he reported only from the Patapsco, to Liliaci- dites dividuus (Pierce) Brenner. However, the Liliacidites type inter- grades with Clavatipollenites and does occur occasionally in the Patuxent- Arundel, favoring Kemp’s treatment of both types as one genus. The status of C. minutus Brenner, defined on the basis of smaller size, is doubtful, since Kemp found it falls within the size variation of C. hughesii. Another form which should be re-evaluated is the small, coarsely reticulate “Pero- monolites” reticulatus Brenner, which, as Worrie (1967) suggested, may be an angiosperm related to Clavatipollenites rather than a perinate spore (Fics. 1j,k). edit (1963) was skeptical about the angiospermous nature of Clava- tipollenites, and he suggested that it represents an extinct group of gymno- sperms. This possibility cannot be denied, but there is no concrete evidence for it, and it loses its force because definite (tricolpate) angiosperm pollen appears in the next formation, and all the morphological characters are quite at home among the angiosperms. Much of the range of variation (though not all the intermediates) may be found in the Chloranthaceae: Ascarina pollen resembles the fine clavate-retipilate forms, Hedyosmum the coarser clavate irregular-aperturate ones, while Sarcandra pollen is reticulate and nearly inaperturate. Similar retipilate sculpture is seen in the Myristicaceae and many dicots with tricolpate pollen, and variation in the aperture from monosulcate to trichotomosulcate is common in several monocot and “ranalean” families, e.g. Canellaceae (Wilson, 1964). In general, Clavatipollenites has more in common with the “‘ranalean’’ dicots than the monocots, which tend to have reticulate or tegillate rather than pilate exines If Clavatipollenites is tentatively regarded as of angiospermous origin, it is the oldest definite pollen record of angiosperms (cf. Couper, 1964). Older reports have gradually been rejected as more has been learned of Mesozoic gymnosperms. Eucommiidites has been discussed; the alleged nymphaeaceous pollen from the Scottish Middle Jurassic (Simpson, 1937) appears to have been grains of the coniferous genus Zonalapollenites and folded araucariaceous grains (Hughes & Couper, 1958), and Rouse’s (1959) Upper Jurassic Pterocarya was a corroded Classopollis grain (Pocock & Jansonius, 1961). Classopollis itself was originally misinterpreted by Pflug (1953) as a tricolpate grain. A possible older occurrence of tricolporates, in the Berriasian-Valanginian of the Netherlands (Burger, 1966), has not yet been restudied. The presence of primitive angiosperm pollen in the Patuxent-Arundel is consistent with the megafossil record. Fontaine (1889) described dicot leaves, Ficophyllum, Rogersia, and Proteaephyllum (in part), from the (catia sitieneteeeniaienaneeeaen 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 9 Patuxent near Fredericksburg, Virginia. Palynological study of the matrix (Harvard University Paleobotanical Collections: cf. Fontaine, 1889, p. 5) shows that this locality is indeed of lower Potomac age. Berry (1911) questioned that these leaves were dicots and suggested that they could be Gnetum, but reinvestigation by Wolfe (pers. comm.) shows none of the distinctive fine venation or fiber network characters of Gnetum, and instead a series of presumed primitive angiosperm characters found in the living Winteraceae. It should be noted that the distinctive permanent tetrads of the Winteraceae are absent in the Potomac Group pollen flora, so a direct affinity is questionable. Isolated entire margined dicot leaves are also reported from the Aptian of the USSR (Vakhrameev, 1952). No consistent way has been found to subdivide the Patuxent-Arundel palynologically, and Brenner (1963) included both in one biostratigraphic unit, Zone I. The age has not been defined more precisely than Bar- remian, Aptian, or early Albian. Clavatipollenites and ephedraceous pollen (Couper, 1964) and the schizaeaceous spore assemblage (Hughes, pers. comm.) indicate a post-Hauterivian age. The general assemblage suggests middle more than early Lower Cretaceous (cf. Pocock, 1962), and it is very much like the flora described from undifferentiated Aptian-Albian rocks of Portugal (Groot & Groot, 1962). Determinations of an early Neocomian age based on the megafossils (Berry, 1911; Dorf, 1952) were made when younger pre-Albian floras were practically unknown. The upper limit on the age is defined by the absence of tricolpate angiosperm pollen in the Patuxent-Arundel and its appearance at the base of the overlying Patapsco Formation. The appearance of tricolpates, discussed in the following section, is often taken to mark the Aptian-Albian boun- dary, but the sporadic record of Lower Albian tricolpates leaves open the possibility that the Arundel-Patapsco boundary lies within the Albian. PATAPSCO FORMATION Changes in the pteridophyte spore and gymnosperm pollen flora between the Arundel and the Patapsco are rather minor: the entrance of a handful of new species which Brenner (1963) used as index fossils for his Zone IT, and the decline of some groups such as Classopollis and the Schizaeaceae within the Patapsco. The most important event is the appearance of small reticulate tricolpate grains. This pollen type is unlike the “pseudotricol- pate” Eucommiidites type in having radial symmetry and a reticulate exine, and it is at present restricted to the dicots. In the lower part of the Patapsco (Brenner’s Subzone A) the tricolpates are very uniform and present only in low percentages; in the upper part (Brenner’s Subzone B, recognized by the entrance of several distinctive spores and gymnosperm pollen) they become more diverse and abundant; on rare occasions they constitute a majority of the pollen and spore flora. Their percentage variation from sample to sample (Brenner in fact encountered Subzone B assemblages with no tricolpates) suggests that they were restricted ecologi- cally to certain habitats. Clavatipollenites also increases in abundance in 10 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 the iy ere forms with irregular aperture morphology are not uncommon (Fics. 1h,i) . 2. Patapsco — LanfOir All specimens from Patapsco Fm., Sub- zone B of Zone II. a and b, colpopollenites cf. micromunus, two focal levels (65—1-2a); c pee - Tricolpo pollenit es cf. minutus, two focal levels (65-S—3h) ; e and f, Tricolpate type 1, pilate grain, two focal levels (65—-1-2a); g, h, and i, Tricolpate type | 2. nearly Hsu srg ~ focal levels (65-0-2g); j and k, V m, Tricolpopollenites aff. crassimurus, pte ocal levels (65-2a-1b); n and 0, i i t erc i , Ip (65-S—3i) ; s and t, Tricolporoidate type 2, oblate grain with subtriangular amb, two focal levels (65-2a—1j). All figures 1000. 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 11 Patapsco tricolpate pollen shows some differentiation; at least in the upper part of the formation perhaps a dozen form species might be recog- nized. They are typically small (10-20), prolate or spheroidal in shape, with a fairly thin retipilate or reticulate exine, and colpi without any specialized margins or wide membranes. Many of the dominant forms (Fics. 2a,b. Cf. Tricolpopollenites micromunus Groot & Penny, compared by Brenner to pollen of Tetracentron, or Tricolpites albiensis Kemp) have fine but well-defined retipilate or reticulate sculpture, without a continuous tegillum, and sexine somewhat thicker than nexine (1.0-1.5, total exine thickness). Tricolpopollenites minutus Brenner is very small (11, average axial dimension), with a reticulum which may be missed without oil im- mersion (Fics. 2c,d). Besides these and similar microreticulate species, there are tricolpates with a Clavatipollenites-like exine, with free pila not arranged into a reticulum (Fics. 2e,f), and at another extreme small forms with a nearly continuous smooth tegillum which were not reported by Brenner and are possibly restricted to the upper Patapsco (Fics. 2g-i). Less common species are “Retitricolpites” vermimurus Brenner with a loose vermiculate reticulum (Fics. 2j,k), and in the upper Patapsco large prolate tegillate grains close to Tricolpopollenites crassimurus Groot & Penny (Fics. 21,m), and a rather thick-walled spheroidal type with a coarse reticulum in the mesocolpia and internal thickenings (costae) at the margins of the operculate colpi (Fics. 2n,o). A previously unmentioned but possibly significant morphological feature of many of the Patapsco tricolpates, especially the Tricolpopollenites micro- munus and T. minutus complexes, is a frequent buckling-out of the center of the colpi, giving them a geniculate appearance and suggesting a rudi- mentary os. This “‘tricolporoidate” tendency is prevalent in the upper Patapsco (Fics. 2p-r). Another tendency, so far seen only in the upper Patapsco and later, results in oblate grains sub-triangular in equatorial outline, with the apertures at the protruding apices instead of sunken, as is the rule in Patapsco forms. These grains show differentiation of the nexine at the aperture and should probably be considered truly tricolporate (Fics. 2s,t). The appearance of tricolpate pollen seems to have been a major world- wide event, and in all areas which have been carefully studied there is a zone with small reticulate tricolpates but without triporates or typical tri- colporates (cf. Krutzsch, 1963; Muller, 1968). This appearance generally may be dated as early or middle Albian, but refinement is needed in most areas. In England, the Patapsco-type Tricolpites albiensis Kemp appears at the base of the Middle Albian, but Kemp (1968) found rare grains of another tricolpate species in one Lower Albian sample. In western Canada tricolpates are reported by Norris (1967) from the base of the Colorado Group, which is considered basal Upper Albian (Norris) or late Middle Albian (Jeletzky, 1968) on the basis of ammonites, while they are reported to be absent from the underlying Mannville Group (Singh, 1964), which is presumably Middle Albian at the top. However, mid-Albian tricolpates cannot be ruled out here since the contact between the two groups is an 12 JOURNAL OF THE ARNOLD ARBORETUM [VoL. 50 unconformity, and in fact rare tricolpates occur locally in the upper Mann- ville (Pocock, pers. comm.; pers. obs.). The same relation holds in the U.S. Western Interior: tricolpates are present in the Thermopolis and Mowry shales of Colorado (lower Colorado Group equivalents; Tschudy & Veach, 1965), and in the Fall River Formation (basal Colorado Group equivalent) of the Black Hills, but they are absent from the underlying Lakota Formation (Cahoon, 1968). In Portugal, they occur in undifferen- tiated Albian but not in lower Aptian-Albian rocks (Groot & Groot, 1962); there are other reports from Albian rocks in France (Taugourdeau-Lantz & Jekhowsky, 1959), Germany (Krutzsch, 1963), and USSR (Zaklinskaya, 1961). In the central USSR Bolkhovitina (1953) reported tricolpates from the Lower Albian on; Yedemskaya (1960) found them in the Albian of the Caucasus, plus two isolated grains in the Aptian. In the Southern Hemisphere, tricolpates occur in the Upper Albian or Cenomanian of New Zealand (Couper, 1960), and in the Albian of northwestern Australia (Kemp, 1966). In view of theories of a tropical origin of angiosperms, it would not be surprising to find earlier occurrences of angiosperm pollen or more pollen types in the present tropical areas. However, it appears that here, too, there is a zone with reticulate tricolpates and no triporates, and that the tricolpates do not appear demonstrably earlier than in present temperate areas. In North Borneo the Upper Albian or Cenomanian pollen flora contains angiosperms only of the same tricolpate and tricolporoidate types as in the Patapsco, associated with a very similar pteridophyte and gymnosperm flora (but lacking Pinaceae) (Muller, 1968). In northeast Brazil (Miller, 1966), the first angiosperm pollen is again reticulate tricolpates; the age is early Albian or possibly late Aptian. In higher zones (mid-Albian through Cenomanian, subdivision uncertain) these are joined by Didymeles-type tricolpodiorates and polyporates, and later triporates. In Upper Albian samples from Peru, Brenner (pers. comm.) found tri- colpates and polyporates but no other angiosperm pollen. A very similar sequence occurs in Africa: in Senegal and the Ivory Coast reticulate tricolpates and tricolporates (tricolporoidates?) are the only angiosperms in the Lower (?) through much of the upper Albian; polyporates and later triporates enter higher in the Upper Albian-Lower Cenomanian interval (Vachey & Jardiné, 1962; Jardiné & Magloire, 1965). The Albian and Cenomanian of Gabon yield similar tricolpates, tricolpodiorates, triporates, and polyporates (Boltenhagen, 1965). The most unusual elements are the tricolpodiorates and polyporates. The latter are compared with the Amaranthaceae, but it should be noted that similar pollen occurs in some monocots (e.g. Alisma spp.). These pollen types do appear earlier in South America and Africa: the tricolpodiorates in fact are unreported elsewhere, but rare polyporates are known from the Cenomanian of Okla- homa (Hedlund, 1966) and Bohemia (Pacltova, 1966). Still, the record is consistent with a pre-Upper Albian interval with only tricolpate and tricolporoidate angiosperms. Brenner (1963) considered the Patapsco to be Albian, and the record 1969} DOYLE, CRETACEOUS ANGIOSPERM POLLEN 13 of the first tricolpates as reviewed here indicates that it is almost certainly no older than Lower Albian and may in fact be younger. Considering the record in England and the North American Western Interior, it is quite possible, as suggested by Wolfe and Pakiser (ms.), that the underlying Patuxent and Arundel are largely Lower Albian in age. More work on the pollen flora near the Aptian-Albian boundary in well-dated sequences (e.g., in Texas) is clearly in order. Aside from the Fredericksburg material mentioned above, the first angiosperm leaves in the Atlantic Coastal Plain are found in the Patapsco (Fontaine, 1889; Berry, 1911), where they are still a subordinate element. Similar fossils occur in the (Middle?) Albian lower Blairmore flora of western Canada (Bell, 1956), the Cheyenne Sandstone of Kansas (Berry, 1922), deposits in the Kolyma basin in eastern Siberia (Samylina, 1960), Lower and Middle Albian rocks of Kazakhstan (Vakhrameev, 1952), and the Albian of Portugal (Teixeira, 1948); Vakhrameev (1952) and Takh- tajan (1960) have noted the characteristic small size of these Albian leaves and suggested a relation to a still unperfected conductive system. Brenner (1963) proposed an Upper Albian age for the upper Patapsco (Subzone B) on the basis of a close specific similarity to the Upper Albian- Lower Cenomanian angiosperm pollen flora of Portugal (Groot & Groot, 1962). It is generally in the Upper Albian that tricolpate pollen becomes a characteristic though still subordinate element in the flora and attains a certain low degree of diversity. Similar Upper Albian (-Lower Ceno- manian?) floras are seen in the lower Colorado Group of western Canada (Norris, 1967), the Thermopolis and Mowry shales of Colorado (Tschudy & Veach, 1965), and Upper Albian-Lower Cenomanian strata in the USSR (Bolkhovitina, 1953; Bolkhovitina et al., 1963). At the present time it is impossible to rule out a Lower Cenomanian age for part of Subzone B, considering the uncertain dating of the correlative deposits, the general lack of well-dated Lower Cenomanian pollen for comparison, and the only slightly more modernized floras of the Middle Cenomanian (cf. Hedlund, 1966, and below). It is in the Upper Albian and Lower Cenomanian that we see the first megafossil floras dominated by dicot leaves. Such floras are the Dakota flora of Kansas (Lesquereux, 1892), long considered Upper Cretaceous but now known to be in part correlative with the Upper Albian Mowry Shale, the upper Blairmore flora of western Canada (Bell, 1956), and a series of Upper Albian floras from Kazakhstan (Vakhrameev, 1952). Characteristic for these floras are a variety of entire leaves and a large number of lobate leaves with platanaceous venation which have been referred to several unrelated modern genera (Aralia, Sassafras, Sterculia, Liquidambar). A “‘Dakota” flora has not been described from the Patapsco, but this may be because the early collections were made mostly near the Potomac River, where the Patapsco appears to be pinching out; I have seen upper Patapsco localities (65—2a, 65—S) rich in simple entire marginal leaves of a type not described by Berry (1911). 14 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 RARITAN FORMATION The basal Coastal Plain unit in New Jersey, the Raritan Formation, has not been studied as comprehensively as the Potomac Group. The Wood- bridge Clay member, near the base of the formation in the Raritan Bay area, is best known palynologically (Groot, Penny, & Groot, 1961; Kimyai, 1966; Wolfe & Pakiser, ms.). This unit, dated as Middle or Upper Cenomanian by marine fossils, is significantly younger than the typical Patapsco of Maryland: the angiosperm pollen is much more diverse, with several definite tricolporates (the dominant pollen type in modern dicots) and low percentages of the first triporates, the genera Complexiopollis Krutzsch and Atlantopollis Krutzsch of the Normapolles group of Pflug (1953). Older beds which promise to close the gap between the Patapsco and Raritan are becoming known to the south of Raritan Bay and in the subsurface, as are younger beds of presumed Turonian age (the South Amboy Fire Clay member) in the Raritan Bay area. The Raritan is the first Coastal Plain unit in which angiosperms clearly dominate the pollen flora, but gymnosperms and pteridophytes are still important elements. These are mostly Pinaceae, Podocarpaceae (including Phyllocladus-like forms and perhaps the bizarre genus Rugubivesiculites Pierce, with a ruffled central body, which appears in the upper Patapsco but is most typical of the North American Upper Cretaceous) , Taxodiaceae, Cupressaceae, Araucariaceae, Cyatheaceae, and Gleicheniaceae; the Schi- zaeaceae are in decline, and most of the extinct groups represented by Classopollis, Eucommiidites, Vitreisporites, etc. are very rarely seen. The angiosperm pollen of the Woodbridge includes reticulate tricolpates of the Albian type, though many appear to be new species, and occasional mono- sulcates (Clavatipollenites, Liliacidites). A larger portion is assumed by small psilate tricolpates and tricolporates (Fics. 3a—d). Some of the most characteristic of these continue a trend seen in the upper Patapsco: they are oblate and triangular in equatorial outline, with apical apertures (Fics. 3e,f. Cf. Tricolporopollenites triangulus Groot, Penny, & Groot). Besides these small, simple tricolpates and tricolporates, there are larger forms with more complex exine structure (reticulate to completely tegillate) and apertures (e.g., Fics. 3g-i). An unusual new pollen type is represented by two forms with permanent tetrads: one is larger, with a smooth tegillum supported by large pila, and with somewhat obscure colpi arranged accord- ing to Fischer’s law (Fics. 3j,k); the other is smaller, psilate to retipilate, with very irregular colpoid areas (Fics. 3l,m). Neither is like the familiar tetrads of the Ericaceae; the smaller type is strikingly similar to pollen of the monogeneric family Myrothamnaceae of South Africa and Madagascar. The most striking new element is the first of the bizarre triporate Nor- mapolles, which are dominant in the Upper Cretaceous and earliest Tertiary of Europe. They are an extinct group, not directly comparable to any living angiosperms, though if the complex protruding apertures of some of the later form genera were reduced somewhat and the grains became less triangular, they might approach pollen of some modern amentiferous 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 15 Fic. 3. Lower Raritan and Patapsco-Raritan transition zone ea oe rm a All specimens except p-s from Woodbridge Clay member, Rari ‘m ind b, Tricolpate type 4, geen hg ae two focal levels (68-10-1b); c — a Tricolporate type 1, pro psilate grain, two focal levels (68 8-la); e and f, Tricolporopollenites cf. pt 1 focal levels (68—-10~1b); g and h, Tricol- porate type 2, reticulate grain with flat mesocolpia, two focal levels (NJ 2-1a); i, Tricolporate type 3, reticulate grain with marginate colpi (68~8-1a); j es k, Tetrad type 1, large, aie: grain, two focal levels (NJ 2-1a); land m trad type 2, small, scar te grain, two focal levels (68—-10—-Ic); n, Compieaosélie sp. (NJ 2-1b); Atlantopollis sp. (68—10—-1b); p and q, Tricolporate type 4 possible eens of mers aha group, oblique view, two focal levels (TR(1551-3)-I1c: up atapsco-Raritan transition zone); T and s, same, polar view, two focal pate (TR(1551-3)~Ic). All figures * 1000. 16 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 groups (Betulaceae, Casuarinaceae, Myricaceae, Rhoipteleaceae, Juglan- daceae, Urticales). In fact, as was pointed out by Goéczan et al. (1967) in their revision of the group, they cannot be rigidly separated from the Tertiary “Postnormapolles” of Pflug (1953), which include many of the modern “Amentiferae.”’ Complexiopollis and Atlantopollis are among the oldest Normapolles in Europe. Atlantopollis differs from the psilate or scabrate Complexiopollis in its coarsely reticulate or (in New Jersey) verrucate sculpture. Apertures in both genera are very short colpi or elongate pores, with the nexine differentiated into an endannular collar just inside the pore (Fics. 3n,o). The multiple endannular rings seen in European Turonian species and in the upper Raritan (Fic. 4a) are poorly developed in lower Raritan forms. The Normapolles and other pollen and spores provide an age determina- tion consistent with that from the marine fauna of the Woodbridge. A Normapolles assemblage with only Complexiopollis and its relatives was first described from the Lower Turonian of Germany (Krutzsch, 1959), but it has been extended an uncertain distance down into the Cenomanian. The Cenomanian Peruc Formation of Bohemia (Pacltova, 1966; Pacltova & Mazancova, 1966) is probably closest to the Woodbridge: very similar Normapolles are present in very low proportions, while the rest of the angiosperm flora contains reticulate and psilate tricolpates and tricolpo- rates, including a triangular form of the Tricolporopollenites triangulus type (but also polyporates unknown in the Raritan). Cenomanian deposits of Portugal (Groot & Groot, 1962) also contain Complexiopollis and Atlantopollis (as Latipollis). In North America, the Tuscaloosa Group of Alabama, believed to be of late Cenomanian age, yields a flora with Complexiopollis and Atlantopollis almost identical to that of the Wood- bridge (Leopold & Pakiser, 1964). Complexiopollis (as Punctatricolpo- rites) appears near the putative Cenomanian-Turonian boundary in the Eagle Ford Shale of Texas (Brown & Pierce, 1962). Wolfe and Pakiser (ms.) believe the Woodbridge is Upper Cenomanian, and considering the low percentages of Normapolles this is probably correct, though the range data permit a Lower Turonian age as well. It is probably younger than most of the Middle Cenomanian, since Normapolles are not reported from the Middle Cenomanian Woodbine Formation of Oklahoma (Hedlund, 1966), nor the “Dakota Group” of Minnesota (Pierce, 1959), though both these floras have post-Patapsco elements such as psilate tricolporates and similar conifers (diverse Phyllocladus-type and Rugubivesiculites) and spores (common large Sphagnumsporites, Camarozonosporites, and Glet- cheniidites ) . Triporates other than Normapolles appear in other parts of the world probably in the Cenomanian, though the age control is lamentably poor. Thus in North Borneo grains with simple round pores appear in a zone loosely dated as Cenomanian to Senonian (Muller, 1968); triporates com- pared with Sapindaceae or Proteaceae appear near the end of an Upper Albian to Lower Cenomanian interval in Senegal and the Ivory Coast 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 17 (Jardiné & Magloire, 1965) and late in the Upper Albian through Ceno- manian interval in northeast Brazil (Miiller, 1966). It is becoming clear that there was an interval in the Atlantic Coastal Plain after the typical Patapsco and before the Woodbridge with angio- sperm floras including tricolporates, many psilate and some triangular, but without Normapolles. Extinct gymnosperms such as Classopollis are often common in these floras. This Patapsco-Raritan transition zone is seen in surface samples from Elk Neck in northern Maryland and near Trenton, New Jersey (Wolfe & Pakiser, ms.), from the uppermost “Raritan” of Bodkin Point, Maryland (pers. obs.), in the subsurface “Raritan” near Delaware City, Delaware (Brenner, 1967), at the top of the “Raritan(?)- Patapsco” in a well near Waldorf, Maryland, and in a well some thirty miles downdip from the Raritan Bay outcrop area on the Toms River, New Jersey (pers. obs.). The data of Wolfe and Pakiser and from the Toms River well suggest that the first “Raritan”? elements to appear are the more prolate psilate tricolporates and, soon after, the triangular forms; other “Raritan” elements enter as Lower Cretaceous gymnosperms and ferns decline, until the flora is very close to the Woodbridge except for the absence of Normapolles. In one of the uppermost pre-Woodbridge samples from the Toms River well, an unusual tricolporate occurs with a shape and sculpture very much like Complexiopollis but with longer vestigial colpi and no typical annulus or endannulus. The apertures approach those of some of the more complex Raritan triangular tricolporates, suggesting a link between less bizarre tricolporates and the Normapolles (Fics. 3 , Wolfe and Pakiser (ms.) characterize the pollen flora of the South Amboy Clay member as essentially the same as the Woodbridge flora except for some new non-Normapolles triaperturates. However, samples from four localities in the upper Raritan, including the classic Kreischerville collections of Hollick (New York Botanical Garden Paleobotanical Collec- tions: cf. Hollick & Jeffrey, 1909), yield floras which appear to be significantly younger than the Woodbridge. Although many of the Nor- mapolles might be considered advanced members of the Complexiopollis group (Fic. 4a), most show characters of the mid-Turonian and younger Plicapollis and Vacuopollis groups. Intergradations render generic separa- tion difficult, but the genus Pseudoplicapollis Krutzsch, with rudimentary endoplicae and a characteristic pore structure, is definitely present (Fic. 4b). In most of the grains the apertures tend to protrude less than in Complexiopollis and consist of nearly round pores, with the nexinous collar retracted or reduced to produce a true vestibulum (as in Plicapollis Pflug) or atrium (as in Vacuopollis Pflug). Many show apparently structural folds (Fics. 4c,d), though these are generally less regular than the “endo- plicae” of typical (younger?) Plicapollis. Some of the subspheroical forms approach the myricoid genus Triatriopollenites Thomson & Pflug (Fic. 4g). Also present are small psilate triporates perhaps unrelated to the Normapolles (Fics. 4e,f), and large, oblate brevitricolporates (Porocol- popollenites, sensu Leopold & Pakiser 1964: Fic. 4h). Other new tricol- eT 18 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 porates appear to be forerunners of typical Magothy species (cf. Fics. Sh-k). Most of these forms are reported from the McShan and Eutaw formations of Alabama (Leopold & Pakiser, 1964), which are referred to the later Turonian by Wolfe and Pakiser (ms.). Although strictly com- parable floras have not been described from Europe, the range data of Goéczan et al. (1967) suggest a Middle or Upper Turonian age. MAGOTHY FORMATION The Magothy Formation, which extends from Maryland through Long Island, has a highly diversified angiosperm flora which has been described by Stover (1964), Groot, Penny, and Groot (1961), and more completely by Wolfe and Pakiser (ms.). As noted by Wolfe and Pakiser, the rich and advanced Normapolles element indicates a sizable break in deposition before Magothy time, though the presence of Turonian in the Raritan may close some of the gap. The Raritan marks the end of the nearly continuous mid-Cretaceous record; remarks on the Magothy flora will hence be only of a general nature. Normapolles are a dominant element, represented by at least a dozen genera. The Plicapollis-Pseudoplicapollis group (with Y-shaped thicken- ings and vestibula), the Vacuopollis group, now represented by typical 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 19 Vacuopollis (with large atria and thick annuli made up of minute inward Si apo rods), and Trudopollis Pflug (with thick annuli and endannuli ind a space or interloculum between inner and outer exine) are especially common (Fics, 5a—d). There are many small Normapolles (Minor pollis, etc.) and other simple triporates of the type seen in the upper Raritan (Fic. 5e; cf. Fics. 4e,f). Such an assemblage must be at least as young Fic. 5. Magothy angiosperm pollen. a, Plicapollis ‘7 (68-14-1a: Amboy Stoneware Clay member); b, — sp. (68-14-la); ¢, Vacuo pollis sp. (68-14 4-la); d, Trudopollis sp. (68-14-1a); e, eee type 3 (68-14-la); a Tricolporate type 6, two focal levels (68-16-1a); 1 and m, Monosulcate type ", grain with sulcus on lower side, two focal levels (Ch-Bf 127 (441-2)-1b: Parte Fm. undifferentiated). All figures < 1000. 20 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 as mid-Coniacian (cf. Géczan et al., 1967) and is probably younger, since most of these genera become abundant only in the Santonian (Krutzsch, 1957). More specific evidence is provided by close relatives or new species of the mid-Santonian and younger genera Praebasopollis Groot & Krutzsch (with two endannular lips extending into the large vestibula: Fic. 5f) and Pecakipollis Krutzsch & Pacltova (Plicapollis-like grains without clear endoplicae and with some Trudopollis traits: Fic. 5g). A Santonian age is verified by a late Santonian ammonite found in the upper Magothy of New Jersey (Sohl, pers. comm.). Tricolpates and tricolporates, many continuing from the upper Rari- tan, are highly diverse in the Magothy. Upper Raritan and Magothy pollen types commonly suggest modern families, but most species have anomalous features or characters now found only in related families. Thus the grain in Figures 5h and i has some nyssaceous and cornaceous characters but would not be at home in either family, while the grain in FIGURES 5j and k has rhamnaceous apertures but hippocrateaceous OF celastraceous sculpture, and the common myricoid grains (cf. Fic. 4g) have more of an endannulus than modern Myricaceae. Reticulate, psilate, and spiny monosulcates very suggestive of modern palms are abundant in some samples (Fics. 51m). Palm megafossils, among the oldest known, are also found in the Magothy (Berry, 1916). From the Turonian on, the world pollen flora is marked by provincial- ism which contrasts strongly with the cosmopolitanism of the early Cre- taceous. Zaklinskaya (1962) first pointed out the major provinces of the Northern Hemisphere in the Senonian and Maestrichtian: the Aquilapol- lenites province in Siberia and western North America and the Norma- polles province in Europe and eastern North America (cf. Gdczan et al, 1967). Aquilapollenites Rouse is a peculiar extinct form with a prolate, often heteropolar, central body and protruding arms bearing the apertures. It is often associated with pollen of possible proteaceous affinities which is common also in the Senonian of New Zealand, Australia, and Africa. Aquilapollenites has been found in equatorial Africa (Jardiné & Magloire, 1965) and Borneo (Muller, 1968), but not in the Normapolles province until the breakdown of provincialism in the Paleocene, when it occurs briefly in the Gulf Coastal Plain (Tschudy, pers. comm.). Normapolles are unknown in Africa and Borneo and very rare in Siberia. There are also strong similarities between the pollen floras of Africa and Brazil in the Upper Cretaceous, but these have not been studied as well (cf. Miiller, 1966). The peculiar distribution of the Northern Hemisphere provinces is clearly a reflection of the epicontinental seas which extended from the Gulf of Mexico to the Arctic Ocean and along the east side of the Urals (cf. Tschudy, 1966). The Magothy flora is a representative of the Normapolles province, but it illustrates that the province should be divided into American and European areas. Though many of the stratigraphically important genera and groups of genera are common to Europe, Wolfe and Pakiser (ms.) pa a at a 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 21 point out that many bizarre forms are restricted to Europe, and many of the Magothy genera are new, being most nearly represented only by relatives in Europe. They note that the Trudopollis group is absent from the American Turonian, and that most of the Magothy dicots other than Normapolles are still unreported from Europe. The Atlantic Ocean cer- tainly acted as a barrier to migration in the Senonian, but it is still sur- prising that it may have been less effective than the epicontinental sea of the American interior. GENERAL EVOLUTIONARY IMPLICATIONS In the late Lower Cretaceous the angiosperms are a very subordinate, undiversified element in the pollen flora; by the mid-Upper Cretaceous they are dominant and highly differentiated, though far from modern in total variation. The increase in diversity is regular, with new morpho- logical types appearing not at random but in what can be read as series that permit derivation of each type from an earlier one. Small etiilate monosulcates are joined by small retipilate tricolpates; these pass into tricolporoidates and then tricolporates of more diverse exine eichns and these into the first triporates, which in turn diversify. The pollen record by itself leads unambiguously to the conclusion that we are wit- nessing a major adaptive radiation of a new group. Since we have the time dimension, we can tell which way to read our series and hence de- termine which character states are primitive (i.e. ancestral) and which advanced (i.e. derived). The resulting trends of course apply directly only to the plants of the time observed, and many of them have doubtless been reversed in later evolution, but they are relevant to modern groups insofar as the present major alliances are the result of this radiation and much of the ancient range in morphology is retained today. Likewise, we observe directly only evolution in pollen morphology, but this tells us something about general phylogeny insofar as pollen morphology is useful in recognizing taxa today, and as primitive or advanced characters in different organs are loosely correlated as a result of lesser or greater evolutionary rates in a given line (cf. Sporne, 1954). The pollen record sets some limits on the time and place of origin of the angiosperms. The group must have originated at the beginning of the observed radiation (in the Barremian-Aptian) or earlier, though the possibility that the earliest pollen with angiosperm characters (Clavati- pollenites and the early Albian tricolpates ) was produced by plants which had not reached the angiosperm level in other organs should not be ig- nored, But at least in the Aptian, the typical dicotyledonous leaf morph- ology had been attained as well. So far, the pollen record provides no conclusive evidence on the hypothesis that the angiosperms appeared and diversified first in the tropics. The earliest tricolpates there are similar to those in the tem- perate zones, though after the Middle Albian some pollen types seem to have appeared earlier in the tropics. There is suggestion of a lag of 22 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 a third of a stage in invasion of the middle latitudes in reports of Lower Albian and even Aptian tricolpates in Africa and South America on the one hand and their poor record before the Middle Albian in England and North America on the other, but the stratigraphy requires much refine- ment before this can be considered established. In any case, the tropical belt in the early Cretaceous undoubtedly covered areas which are now temperate. The tree ferns and cycads in the Potomac Group suggest at least an equable (warm temperate?) climate, though the lack of bisaccate conifers in the tropics indicates some latitudinal differentiation. One area that is ruled out as a center of angiosperm origin and evolu- tion is the Arctic. Megafossil and microfossil floras from northern Siberia and Alaska north of the Brooks Range rarely contain angiosperms until well into the Cenomanian (Teslenko, 1958; Vasilievskaya, 1956; Smiley, 1966; Stanley, 1967; cf. Hughes, 1961b). Most of Seward’s Lower Cre- taceous angiosperm leaves from the Kome flora of Greenland appear to have come from the Upper Cretaceous, and the status of the one remain- ing leaf is uncertain (Koch, 1964). It is certainly possible that primitive, undiversified angiosperms existed as a subordinate part of the flora long before the Barremian. We need only compare the mammals, which originated in the late Triassic but did not undergo major radiation until the Tertiary. Rare angiosperms with cycad-like pollen, as in several “ranalean” families, might easily go un- noticed in the Jurassic or Triassic. However, theories that postulate that the angiosperms not only existed but diversified long before the Cretaceous in isolated areas such as the tropical uplands (e.g. Axelrod, 1952) and simply migrated into other areas in the Cretaceous do become implausible in the light of the progressive appearance of morphological types. While we might expect a gradual increase in the number of types as a result of such migration, we would expect a sequence of unrelated derived types rather than convincing evolutionary series. The record of early angiosperms is doubtless biased toward prolific pollen shedders and wind-pollinated offshoots. Even so, if much more highly evolved pollen was being produced by strictly insect-pollinated plants, we would expect to see it occasionally. Isolated large, more highly sculptured grains often found in the Potomac Group (e.g. ‘“Retitricol- pites” geranioides (Couper) Brenner or the form in Fics. 2l],m) may in fact represent such plants, but they are similar morphologically to their smaller and more common associates. The simplest assumption, that the pollen we see preserved is fairly representative of the morphological types that existed at the time, is followed here The concept of an evolutionary radiation of angiosperms beginning in the early Cretaceous may appear to be in conflict with the megafossil record. Lower Cretaceous leaves have been placed in such unrelated modern genera as Populus and Sassafras and form genera (e.g. Ficophyl- lum and Celastrophyllum) intended to suggest families as distant as Moraceae and Celastraceae. To reconcile such diversity with the uni- 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 23 formity of the pollen flora by postulating that pollen evolution lagged behind while other organs differentiated to nearly a modern level would require an incredible amount of mosaic evolution in many lines. The pattern of variation in modern angiosperms suggests rather that, in gen- eral, pollen morphology has not behaved markedly unlike other character complexes: families may be either very uniform or diverse palynologically. What is needed is a re-evaluation of the leaf determinations, which were important fossils. Pacltova (1961) found that cuticles of “Eucalyptus” from the Cenomanian of Bohemia bore no specific relation to that genus; the platanaceous venation of Dakota leaves placed in several unrelated genera and Wolfe’s case for the winteraceous affinity of Ficophyllum have been mentioned. A detailed study of the morphology of Cretaceous leaves might prove of more evolutionary interest than attempts at identification of taxa. SPECIFIC TRENDS The most striking evolutionary trends seen in early angiosperm pollen are in the apertures and shape of the grains. Other more questionable trends involve the exine sculpture and size. These trends are summarized in TABLE 3 (p. 28). It would appear that the monosulcate condition of Clavatipollenites, the first convincing angiosperm pollen, is very primitive. Within this group we see as later variants trichotomosulcates, inaperturates, and grains with several ill-defined colpoid areas. There is a definite trend to fusion of structural elements into a true reticulum, as in Liliacidites. The record is consistent with derivation of the tricolpates, the next most ancient major pollen type, from monosulcates of the Clavatipollenites type. The similar retipilate exine structure in Clavatipollenites and the earliest tricolpates favors this hypothesis over such alternatives as a completely independent origin or derivation from the Eucommiidites type. It is a general principle that in seeking ancestors for a group we should consider its most ng (here earliest) members rather than advanced (later) forms (Thorne, 1963). There are, unfortunately, no obvious in- termediates between monosulcates and tricolpates in the Cretaceous record, but the presence of trichotomosulcate apertures and irregular colpoids in Clavatipollenites may be significant. A theory of the origin of the tri- colpate condition by loss of the polar connection of the three arms In a trichotomosulcate grain has been advanced by Wilson (1964). However, intermediate forms with three colpi displaced toward the pole are lacking. Another possibility is that the tricolpate condition represents the stabili- zation of an irregular situation with several colpoids. In the Chloran- thaceae a similar process may have produced the longitudinal colpi (usu- ally six) in Chloranthus 2 * See footnote 1 on page 6. 24 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Within the triaperturate group trends are more readily documented, and the wealth of intermediates leaves no need to invoke independent origin of the more complex forms. Tricolporates may have originated from tricolpates via tricolporoidates with only a slight weakening at the center of the colpus membrane. In some lines this trend was associated with a change in shape from prolate or spheroidal to oblate with a triangular amb and apical apertures. A pervasive trend in all the triaperturate classes (as well as the monosulcates), but most common in the tricol- porates, was the fusion of exine structural elements into a complete tegillum, often resulting in psilate grains. Permanent tetrads also appear as a later offshoot in the triaperturate groups. The small size of the early tricolpates suggests that this may be a primitive character in the triaperturates. However, the occasional pres- ence of large grains suggests size was an unstable trait from the beginning, being subject to changes in pollination ecology. Small grains are often associated with wind pollination, but the Albian forms are even smaller than most amentiferous pollen. A comprehensive comparative study of size-pollination relations in modern angiosperm pollen would be desirable. The culmination of the trend toward apical apertures is evidently seen in the triporate Normapolles, which may have been derived from con- ventional dicots through triangular tricolporates in pre-Woodbridge time. The Normapolles show various peculiar trends, such as the evolution of atria, vestibula, and other elaborations of the pores, and the development of endoplicae suggesting the arci of the Betulaceae, Rhoipteleaceae, and Ulmaceae. Soon after the origin of the group, the shape trend was appar- ently reversed to produce subspheroidal grains, as in most of the modern “Amentiferae.” Other triporates, seen in the upper Raritan and parts of the world where Normapolles are lacking, may be of independent origin; they might originate by reduction of the colpus in a tricolporate grain or perhaps by contraction of the colpus in a tricolpate. Pollen with numerous scattered pores occurs in the Cretaceous of some areas, apparently always after the entrance of tricolpates. Polyporates are found today in both monocots and dicots: the fossil record is con- sistent with derivation from either monosulcates or tricolpates and does not indicate which alternative is correct. Occasional dicolpate and poly- colpate variants of tricolpate Raritan species and tetraporate Normapolles grains are within the normal variation of modern species; apparently such variation did not lead to major trends in the Cretaceous. The proposed evolutionary relationships among the major pollen types are shown in their stratigraphic framework in Ficure 6. This scheme is almost identical to the one proposed by Takhtajan (1959, 1966), based on the comparison of the pollen of angiosperms which are believed to be primitive or advanced in other characters. Many of the same trends are implicit in the writings of Bailey and coworkers (e.g. Money, Bailey, & Swamy, 1950) and of Wodehouse (1936). One of Takhtajan’s impor- tant trends, from monosulcate to monoporate, is not shown since mono- 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 25 Turonian (U. Raritan) other triporates higher Normapolles —_ oS ! (L. Raritan) (@ ? ? a tetrads Complexiopollis Cenomanian | (Patapsco- Raritan ; , Sranettien tricolporates tricolporates: | triangular amb ne nee : ae Altiien polyporates* tricolporoidates (Patapsco) ? 2 aa = ee tricolpates Barremian- Aptian * Africa, South (Patuxent - monosulcates America only Arundel) (Clavatipollenites type) Fic. 6. Suggested evolutionary interrelationships of the major angiosperm pollen types of the Potomac-Raritan interval (Barremian-Turonian). porates are not reported from Turonian or older rocks, but it is suggested by the record. Monoporates of a graminioid type are known from the Maestrichtian of Africa (Jardiné & Magloire, 1965). The trends proposed here are very different from those of Kuprianova (1966), who lists as primitive a large number of characters found in such groups as the Santalales, many amentiferous plants, and Upper Cretaceous fossils including the Normapolles which suggest an ancestry with trilete spores. This approach overlooks earlier Cretaceous fossils which point toward simple tricolpate or monosulcate ancestral forms; the fossil sequence clearly shows sporelike Upper Cretaceous forms and their modern analogs are secondarily derived. As a rule, I would suggest that no exclusively post-Middle Albian pollen types can be used directly to reconstruct primi- tive conditions in angiosperms. PHYLOGENETIC INFERENCES Although the Cretaceous pollen record does not show us the origin of 26 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 the angiosperms, it does allow us to make more secure inferences about the ancestors of the group. Thus the earliest monosulcate angiosperm pollen points to a group of gymnosperms with monosulcate pollen. This suggests an ancestor among the cycadopsids (i.e. the seed ferns and their presumed derivatives) rather than the coniferopsids or pteridophytes. An ultimate seed fern ancestor, making the angiosperms a parallel group to the cycads, Bennettitales, and Caytoniales, is suggested by the compara- tive morphology of the other plant organs (cf. Takhtajan, 1960; Cron- quist, 1968). Clearly more must be known of Triassic, Jurassic, and, as Hughes (1961b) emphasizes, Lower Cretaceous gymnosperms before a more definite hypothesis may be presented. Discussion of an ancestor of the angiosperms assumes the group is monophyletic, at least in the loose sense of Simpson (cf. Cronquist, 1968). This assumption is consistent with the record, which appears to show one major radiation, with the more ancient representatives of putative lines more instead of less similar to each other. Even the Normapolles may be derived from earlier tricolporates. This argument holds only for the basi- cally tricolpate groups and their immediate monosulcate ancestors (i.€. the bulk of the dicots), and it does not mean all the characters we asso- ciate with the angiosperm grade had evolved when the taxon originated. In any case, it is quite likely that still more primitive groups with mono- sulcate pollen reached the angiosperm level in several lines, resulting in much of the heterogeneity of the living “Ranales Speculation on the affinities of early sngioapern pollen might easily lead to unwarranted conclusions on the age of modern taxa. It is not difficult to find modern analogs of Albian pollen: we have seen that much of the morphological variation in the Clavatipollenites type may be found in the Chloranthaceae, while similar generalized tricolpates occur in the Lardizabalaceae and Menispermaceae, or the Tetracentraceae, Hama- melidaceae, Platanaceae, and related families of the Trochodendrales and Hamamelidales of Cronquist (1968). However, the monosulcate and tricolpate complexes were young and evolving rapidly in the Albian, and their total diversity could probably be accommodated in two or three closely related orders and perhaps five to ten families. In contrast, the modern families mentioned are relictual and isolated from each other by specializa- tion and extinction. It would probably be a mistake to believe the simi- larities indicate any more than that such taxa have retained a primitive pollen type, and hence perhaps other primitive characters. Higher dicot groups (e.g. Salicaceae) may have reticulate tricolpate and tricolporoidate pollen, but unlike the Ranunculales, Trochodendrales, and Hamamelidales they are usually dominated by tricolporate pollen (cf. Fagaceae, Elaeocarpaceae, Flacourtiaceae). In terms of the subclasses of Takhtajan (1966) and Cronquist (1968), it is possible that Albian angiosperms had not evolved beyond the level of the Magnoliidae and lower Hamamelididae, and that the higher Hamamelididae (most of the “Amentiferae”), Dilleniidae, Caryophyllidae, Rosidae, and Asteridae were 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 27 represented only by ancestors more primitive in pollen morphology and many other characters than their present members. These taxa may have differentiated in the radiation of basically tricolporate groups beginning in the Cenomanian. Some Upper Cenomanian tricolporates already sug- gest orders of the Rosidae such as the Cornales. The fossil record indicates that extinct dicot alliances, represented by the Normapolles and Aquilapollenites, flourished in the late Cretaceous (cf. Krutzsch, 1963). The possibility of extinct major groups is largely ignored in angiosperm phylogeny, but it is quite relevant, for example, in the ““Amentiferae,” which may be in large part relics of the group repre- sented by Normapolles pollen. Since the basic monosulcate pollen type of monocots is common among “ranalean” dicots, the fossil pollen record is ambiguous on the origin of monocots. Though typical Clavatipollenites is most like the pollen of some modern dicots, the usually younger reticulate Liliacidites type could be either “ranalean” or monocotyledonous. Some Cenomanian pollen is more convincingly monocotyledonous in origin. The discussion of the last paragraphs shows the consistency of the rec- ord with the systems of Takhtajan and Cronquist. In general, “ranalean” theories of angiosperm phylogeny are favored, since the earliest angio- sperm pollen is of types characteristic of or restricted to groups considered primitive in such theories. On the other hand, systems which make the wind-pollinated “Amentiferae” primitive become implausible. The Betu- laceae, Casuarinaceae, Myricaceae, Rhoipteleaceae, Juglandaceae, and Urticales all have basically triporate pollen (from the Normapolles?), a definitely derived, though ancient, type. The Fagaceae, with generally prolate tricolporate pollen, have a questionable status, but the unusual complex protruding apertures in Trigonobalanus doichangensis (Camus) Forman (Erdtman, 1967) suggest a relation to the Normapolles. CONCLUSIONS Angiosperm pollen types in the Cretaceous of the Atlantic Coastal Plain appear in essentially the same sequence as in other areas, including the tropics. In the Patuxent and Arundel formations (Barremian?-Lower Albian?) the retipilate monosulcate genus Clavatipollenites, apparently the oldest pollen with characters restricted to angiosperms, occurs In a ora dominated by pteridophytes and gymnosperms. Clearly dicotyle- donous reticulate tricolpate pollen appears at the base of the Patapsco Formation (Lower-Middle Albian?); tricolpates increase in abundance and diversity in the upper Patapsco (Upper Albian-Lower Cenomanian?), where many show tricolporate tendencies. Definite tricolporates, often psilate and with triangular amb, occur in beds transitional to the Raritan Formation; in the lower Raritan (Upper Cenomanian?) they are joined by the first triporates, Complexiopollis and Atlantopollis of the extinct (pre-amentiferous?) Normapolles group. In the upper Raritan (Middle- Upper Turonian?), these pass into more advanced Normapolles genera. 28 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The rich flora of the Magothy Formation (Santonian), which includes some forms suggesting modern families, is representative of the Senonian Normapolles province of Europe and eastern North America. The expansion and diversification of angiosperm pollen in the Cretaceous is believed to reflect the basic adaptive radiation of the group, within which morphological series documenting evolutionary trends and the origin of major types may be recognized. Though the angiosperms may have originated well before the observed radiation, the idea that they were highly differentiated at their first appearance in the fossil record conflicts with the low diversity of Albian angiosperm pollen and the regular sequen- tial appearance of morphological types. Trends such as monosulcate to tricolpate, prolate tricolpate to tricolporoidate to oblate tricolporate to triporate, and retipilate or reticulate to completely tegillate are in good agreement with trends postulated on the basis of comparative morphology and with systems in which the Magnoliidae and lower Hamamelididae are considered primitive and the ‘““Amentiferae” advanced. Considering the evidence for important evolution in pollen characters, it is hoped that the megafossil record of early Cretaceous angiosperms will be re-examined with modern techniques and a more evolutionary-morphological point of view. TABLE 3. Evolutionary trends in pollen morphology based on the Cretaceous pollen record GENERAL APERTURE TRENDS: monosulcate, bilateral symmetry — tricolpate, radial symmetry monosulcate or tricolpate — polyporate simple colpi > complex apertures MONOSULCATE GROUP monosulcate > eiietishicisatitiade, inaperturate, or with several colpoids pilate or retipilate — reticulate or completely tegillate gpeenteris GROUP colpate > iecokliee — tricolporate > triporate aa — triporate: prolate or subspheroidal — oblate, triangular amb, angulaperturate retipilate or reticulate — psilate, completely tegillate single grains — permanent tetrads small size — large size? NORMAPOLLES GROUP pores nearly siete — pores with atria or vestibula no endoplicae — endoplica triangular amb —> circular amb ACKNOWLEDGMENTS I am deeply indebted to Professor Elso S. Barghoorn, whose sponsor- ship made this study possible, for his continuing advice, encouragement, 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 29 and criticism. Dr, Alexandra Bartlett provided invaluable advice on palynological techniques and interpretation of pollen morphology. I also wish to thank the many persons acknowledged in the text for discussions and unpublished information. E. T. Cleaves, J. D. Glaser, and H. J. Han- sen, III, of the Maryland Geological Survey, J. P. Owens and C. F. With- ington of the U. S. Geological Survey, and H. F. Becker of the New York Botanical Garden supplied me with samples. I have received financial support from the Committee on Evolutionary Biology (NSF grants GB3167, GB7346; principal investigator, R. C. Rollins) and the Graduate School of Arts and Sciences of Harvard University. BIBLIOGRAPHY ANDERSON, J. L. 1948. Cretaceous and Tertiary subsurface geology. Maryland Dept. Geol., Mines, Water Res., Bull. 2: 1-113. ARCHANGELSKY, S., & J. C. GAMERRO. 1967. Spore and pollen types of the Lower Cretaceous * Patagonia (Argentina). Rev. Palaeobot. Palynol. 1: 211-217. Axetrop, D. I. 1952. A theory of angiosperm evolution. Evolution 6: 29-60. BELL, W. A. 1956. Lower ere floras of western Canada. Geol. Surv. 1916. " Paleobotanical sections in Upper Cretaceous. Maryland Geol. Surv. 1-1023 1922. The flora of the Cheyenne Sandstone of Kansas. U. S. Geol. 5 Surv., P BorkHovitina, N. A. 1953. Sporovo-pyl’tsevaya kharakteristika melovykh otlozheniy tsentral’ se) oblastey SSSR. Trudy Instituta Geol. Nauk, vyp. 145, geol. seriya, no. 61: 1-184. SSeS Korova, Vv. I, Samopurov, & YANG CHI-TUAN. 1963. Stratigrafiya kontinental’ nykh melovykh otlozheniy Nizhnesyrdar’inskogo podnyatiya oe Priaral’ye). Doklady Akad. Nauk SSSR 152(2): 392- ie ce. E. 1965. Introduction a la palynologie stratigraphique du bassin sédimentaire de l'Afrique équatoriale. Mém. Bur. Rech. Géol. Min. 3 305-326. BRENNER, G. J. 1963. The poe and pollen of the Potomac Group of Maryland. Maryland Dept. Geol., s, Water Res., Bull. 27: 1-215 . 1967. Early nic: pa ‘pollen differentiation in the Albian to Ceno- manian deposits of Delaware (U. S. A.). Rev. Palaeobot. Palynol. 1: 219- 227 Brown, C. W., & R. L. Prerce. 1962. Palynologic correlations in Cretaceous Eagle = Group, northeast Texas. Am. Assoc. Petrol. Geol., Bull. 46(12): 2133-214 Burcer, D. i. Palynology of uppermost Jurassic and lowermost Cretaceous Strata in the eastern Netherlands. Leidse Geol. Meded. 35: 209-276. Cason, E. J. 1968. Palynology of the Inyan Kara Group of res Dakota. Am. Jour. Bot. 55(6, pt. 2): 724, 725. Casey, R. 1964. The Cretaceous Period. Jn The Phanerozoic Time-scale. Quart. Jou ur. Geol. Soc. London 120s: 193-202. 30 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Criark, W. B., A. B. Brppins, & E. W. Berry. 1911. The Lower Cretaceous deposits of Maryland. Ju Lower Cretaceous. Maryland Geol. Surv. 23-98. CLEAVES, E. T. 1968. Piedmont and Coastal Plain geology along the Susquehanna Aqueduct, oe to Aberdeen, Maryland. Maryland Geol. Surv., Report Invest. 8: Couper, R. A. wae British Mesozoic microspores and pollen grains. Palaeon- pags resy 103B: 75-179. 960. New Zealand Mesozoic and rare plant microfossils. New Fella Geol. Surv., Paleont. Bull. 32: . 1964. Spore- pollen correlation of fa Cae rocks of the northern and southern hemispheres. Palynology in Oil Exploration, Soc. Econ. Pale- ontologists Mineralogists, Spec. Publ. 11: 131-142. Cronguist, A. 1968. The evolution and classification of flowering plants. Houghton Mifflin Co., Boston. 1-396. Dorr, E. 1952. Critical analysis of Cretaceous stratigraphy and paleobotany of the aoe Coastal Plain. Am. Assoc. Petrol. Geol., Bull. 36(11): 2161- pierces G. 1948. Did alee ica dg plants exist in early Jurassic time? Geol.. Foren, Forhandl. 70(2): 265-271. ; 52. Pollen eee and plant taxonomy. Part I. Angiosperms. Chronica Botanica Co., Waltham, Mass. 1-539. 1967. On the pollen morphology of Trigonobalanus (Fagaceae). Bot. Not. 120(3): 324— FONTAINE, W. M. non an oo or Younger Mesozoic flora. U. S. Geol. urv., Monograph 15: 1-3 GLASER, ; D. 1967. bad Cretaceous sedimentation in the Middle At- lantic Coastal Plain. Ph. D. dissertation, The Johns Hopkins Univ. [Un- published |. GoczAn, F., J. J. Groot, W. Krurzscu, & B. PacttovdA. 1967. Die Gattungen des “Stemma Normapolles Pflug 1953b” (Angiospermae). Palaont. Abhandl. 2B(3): 429-539. Groot, J. J.. & J. S. Penny. 1960. Plant microfossils and age of nonmarine Cretaceous sediments of Maryland and Delaware. Micropaleontology 6(2): 225-236 —.,& C.R. Groot. 1961. Plant microfossils and age of the Raritan, Tuscaloosa, and Magothy formations of the eastern United States. Palaeon- tographica 108B: 121-140. ——, & C. R. Groor. 1962. Plant microfossils from Aptian, Albian, and Cenomanian deposits of Portugal. Comunicacées dos Servicos Geoldgicos de Portugal 46: 133-176. Hansen, H. J., III. 1968. Geophysical log cross-section network of the Cre- taceous ee of southern Maryland. Maryland Geol. Surv., Report Invest. 7: 1-4 HEDLUND, R. W. a Palynology of the Red Branch Member of the Woodbine Formation Neo gern Bryan County, Oklahoma. Okla. Geol. Surv., Bull. 112: 1-69. He.at, A. H. a Jurassic plant microfossils from the subsurface of Kharga Oasis, ecole Desert, Egypt. Palaeontographica 117B: 83-— Hottick, A., & E. C. Jerrrey. 1909. Studies of Cretaceous coniferous — from sales at New York. Mem. New York Bot. Gard. 3: 1-13 Hucues, N. F. 1958. Palaeontological evidence for the age of the oe Wealden. Geol. Mag. 95(1): 41-49. 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 31 ——. 196la. ogee oe of Eucommiidites Erdtman 1948. Palae- ger 4 4(2): 2 ———. 1961b. dee die and angiosperm ancestry. Science Progress 49: rere ,&R Couper. 1958. Palynology of the Brora Coal of the Scottish Middle Jurassic. Nature 181(4621): 1482, 1483. JarpineE, S., & L. Maciorre. 1965. Palynologie et stratigraphie du Crétacé des bassins du Sénégal et de Céte d'Ivoire. Mém. Bur. Rech. Géol. Min. 32: 187~245. JeLetzky, J. A. 1968. Macrofossil zones of marine Cretaceous of the Western Interior of Canada and their correlation with the zones and stages of Europe and the Western Interior of the United States. Geol. Surv. Canada Paper 67-72: 1-66. Kemp, E. M. 1966. Aptian and Albian miospores from Southern England. Ph. D. thesis, Univ. Cambridge | Unpublished]. . 1968. Probable eee saben Loria from British Barremian to Albian rata. Palaeontology 11(3): 421 reese A. 1966. New plant ae from the Raritan Formation (Creta- ceous) in New Jersey. Micropaleontology 12(4): 461-476. Kocu, B. E. 1964. Review of fossil floras nonmarine deposits of west Greenland. Geol. Soc. Am., Bull. 75: 535-54 Krurzscu, W. 1957. Sporen- und ennai aus der Oberkreide und dem Tertiair Mitteleuropas und ihre stratigraphische Verteilung. Zeitsch. Angew. Geol. 3(11/12): 509-548. ———. 1959. Einige neue Formgattungen und -arten von Sporen und Pollen aus ee de ser aaiaea Oberkreide und dem Tertiadr. Palaeontographica 105B: . 1963. Beitrag zur Kenntnis der Microflora der Niederschénaer Schicht- ed cg kleine Microflora aus der Bohrung KOnigstein 1. Ber. Geol. Ges. 8(2): 224-236. a ole L. 1966. A tentative code of the ancient and derivative characters of pollen grains of angiosperms. Palaeobotanist 15(1, 2): 35-37. LeEopotp, E. B., & H. M. Pakiser. 1964. A preliminary report on the pollen and spores of the pre-Selma Upper Cretaceous strata of western Alabama. U. S. Geol. Surv., Bull. 1160E: 71-95. LesQueREUX, L. 1892. The flora of the Dakota Group. U. S. Geol. Surv., Monograph 17: 1-400. Money, L. L., I. W. Bartey, & B. G. L. Swamy. 1950. The morphology and relationships of the Monimiaceae. Jour. Arnold Arb. 31: 372-404 MU ter, H. 1966. Palynological investigations of Cretaceous sediments in north- eastern Brazil. Proc. 2nd West African Micropaleontological Eataneu Ibadan. Leiden. 123-136. Mutter, J. 1968. Palynology of the Pedawan and Plateau Sandstone Forma- pod (Cretaceous-Eocene) in Sarawak, Malaysia. Micropaleontology 14: 37 Norris, G. 1967. Spores and pollen from the lower Colorado Nene : Albian- *Cenomanian of Central Alberta. Palaeontographica 120B: Owens, J. P., J. P. Minarp, & N. F. Sout. 1968. Cretaceous ee in the northern: New Jersey Coastal Plain. New York State Geol. Assoc., Guide- book, 40th Ann. Meeting. 33-48. PacttovA, i 1961. Zur Frage der Gattung Eucalyptus in der bohmischen Kreideformation. Preslia 33(2): 113-129. 32 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 . 1966. Pollen grains of angiosperms in the Cenomanian Peruc Formation in Bohemia. Palaeobotanist 15(1, 2): 52-54 . Mazancova. 1966. Nachweis von Pollen der “Normapolles”- Gruppe in den Peruc-Schichten (Perutzer Schichten) des bohmischen Cenomans. Véstnik Ustredniho Ustavu geologického 41(1): 51-54. Priuc, H. D. 1953. Zur Entstehung und aecpeen ah a angiospermiden Pollens i in der Erdgeschichte. Palaeontographica 95B: Prerce, R. L. 1959. Lower or plant sae from Minnesota. Minn. Geol. Surv., Bull. 42: Pocock, S. A. J. 1962. naa hae analysis and age determination of strata at the Jurassic- Saal boundary in the Western Canada Plains. Palaeon- tographica 111B: 1-95. ,& J. pte, og icing pollen genus Classopollis Pflug, 1953. Micro- paleontology 7(4): 4 Rouse, G. E. 1959. a Gsatete from Kootenay coal-measures strata of British Columbia. Micropaleontology 5: 303-324 SAMYLINA, V. A. 1960. Po pier egg rasteniya iz nizhnemelovykh otlozhenly Kolymy. Bot. Zhur. 45(3): 3 2. Srmpson, J. B. 1937. Fossil ney in nee Jurassic coal. Nature 139(3520): 673. SincH, C. 1964. Microflora of the Lower enti! ‘pai Group, east- central Alberta. Res. Council Alberta, Bull. 15: SMILeEy, C. J. 1966. Cretaceous floras from Kuk Pai area, Alaska: strati- graphic and climatic interpretations. Geol. Soc. Am., Bull. 77: 1-14. Sporne, K. R. 1954. Statistics and the evolution of dicotyledons. Evolution 8: 55-64. STANLEY, E, A. 1967. Cretaceous pollen and = assemblages from northern Alaska. Rev. Paleobot. Palynol. 1: 229-23 Stover, L. E. 1964. Comparison of three pnb spore-pollen assemblages from Maryland and England. Palynology in Oil Exploration, Soc. Econ Paleontologists Mineralogists, Spec. Publ. 11: 143-152. TakHTAJAN, A. L. 1959. Die Evolution der Angiospermen. VEB Gustav Fischer Verlag, Jena. 1-344. ———. 1960. Proiskhozhdenie pokrytosemennykh rasteniy. “Vysshaya Shkola,” Moskva. 1-133. 1966. Sistema i filogeniya tsvetkovykh rasteniy. ‘Nauka,’ Moskva, Leningrad. 1-611. TAUGOURDEAU-LANTz, J., & B. DE JeEKHOwsKy. 1959. Spores et pollen du Ke euper, Jurassique et Crétacé inférieur d’Aquitaine. Comptes Rend. Soc. éol. France, sér. 7. 1(7): 167, 168. nahin : 1948. Flora Mesozoica Portuguesa. Servicos Geoldgicos de Por- tugal. 1-119. TESLENKO, ea 1958. Nakhodki ostatkov rasteniy apt-al’ba v rs roa -Sibirskoy nizmennosti. Doklady Akad. Nauk SSSR 121(5): 905-9 THorNE, R. F. 1963. Some problems and guiding SP of angiosperm phylogeny. Am. Naturalist 97: 287-305. TscHupy, R. H. 1966. Palynology of the Cretaceous-Tertiary boundary in the Mississippi Embayment and northern Rocky Mountain regions. Geol. Soc Am. Spec. Paper 101: 223. S. D. Veacu. 1965. Plant and miscellaneous microfossils from the Thermopolis and Mowry Shales. U. S. Geol. Surv. Open File Report, Denver 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 33 VacHEY, G., & S. JARDINE. 1962. Apercu sur la microflore des séries “albiennes”’ de Céte d'Ivoire. Comptes Rend. Soc. Géol. France 1962(4): 102-104. VAKHRAMEEV, V. A. 1952. Stratigrafiya i iskopayemaya flora melovykh otlo- zheniy zapadnogo Kazakhstana. Stratigrafiya SSSR I: 1-340. VASILIEVSKAYA, N. D. 1956. Nizhnemelovaya flora severnoy chasti lenskogo basseyna. es! Akad. Nauk SSSR voir 4 913-915. Weaver, K. N., E. T. CLeaves, J. Epw & J. D. re 1968. Geologic map of ae Maryland Geol. rome Baltimor Witson, T. K. 1964. Comparative morphology of the Paiste III. Pollen. Bot. Gaz. 125(3): 192~197. WobvenHouse, R. P. 1936. Evolution of pollen grains. Bot. Rev. 2(2): 67-84. Wotre, J. he, & H. M. Pakiser. Stratigraphic and biologic interpretations of some Cretaceous microfossil floras of the Mid-Atlantic States. [Unpubl. manuscript. | Voge. 7 L. 1960. Sporovo-pyl’tsevyye kompleksy iz aptskikh i al’bskikh otlozheniy vostochnoy chasti severnogo Kavkaza. Byul. Mosk. O-va Isp. Prirody, otd. geol. 35(6): 110-115. ZAKLINSKAYA, YE. D. 1960. O znachenii pyl’tsy pokrytosemyannykh rasteniy dlya teenie verkhnego mela i paleogena. Doklady Akad. Nauk SSSR 133(2): 431-434. 6 pce nes of angiosperm pollen for the stratigraphy of Upper Cretaceous and Lower Paleogene deposits and botanical-geographical prov- inces at the boundary between the Cretaceous and Tertiary systems. Pollen et Spores 4(2): 389 ——— 66. Pollen morphology of angiosperms and paleofloristic areas and provinces at the boundary of the Cretaceous and Paleogene. Palaeobotanist 15(1, 2): 110-116. APPENDIX Loca.LiTies CITED IN TEXT AND FIGURES PATUXENT FORMATION: All specimens figured are from the lower Potomac Group exposed in construc- tion of the Susquehanna Aqueduct between Baltimore and Aberdeen, Md., collected by E. T. Cleaves. Locality data are given in Cleaves (1968): Aq 18 = Cleaves sample no. 18 q ? = ”? 9 be 27 Aq 44 = ” ” ” 44 Aq 45 = 7 7 7 45 Patapsco FORMATION: 65-1: exposure on E side of parking lot behind eget Center on 52nd St., ca. 0.2 mi. N of junction of Kenilworth Ave. (Md. 201) and Baltimore- Washington Parkway, S of Bladensburg, Md. cee s priced 17). Gray clay with cupressaceous twig compressions, 40-50’ above surface of parking lot and Brenner’s sample, overlain and underlain by red and gray clay. Subzone B-1 of Zone II, vs. Subzone A of Zone II for Brenner’s sample. 65-2a: NW side of West Bros. Brick Co. pit on N side of Sheriff Rd., 1.0 mi. E of Washington, D. C. city limit, ca. 0.7 mi. NW of Highland Park, Md. (Brenner’s Seuttin 29). Thin gray clay lens with dicot leaf compressions near 34 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 top of predominantly red sit roughly same level as Brenner’s collection but toward N side of pit. Subzone B of Zone II. 65-O: exposure on NE corner of Branch Ave. and O St. SE, Washington, D.C. Gray clay lens with lignite bed, grading downward and laterally into red and white clay, and overlain by cross-bedded sands with ironstone concretions (“Raritan” Formation?). Subzone B of Zone II. 65-S: N side of Severn Clay Co. pit, 0.1 mi. N of road connecting Ritchie Highway and Md. Rt. 648, ca. 0.5 mi. SE of Harundale, Md. (Brenner’s Station 11). Gray clay lens with dicot leaf compressions just above base of pit, overlain by red clay. Subzone B of Zone II. B-27: James D. Bethards No. 1 well (Socony-Vacuum Oil Co.), ca. 5 mi. SW of Berlin, Md. (cf. Anderson, 1948). Gray clay core sample from 2735-2751’, provided by Maryland Geological Survey. Subzone B of Zone II. PATAPSCO-RARITAN TRANSITION ZONE: 68-65: exposure overlooking Chesapeake Bay, ca. 0.6 mi. S of Bodkin Point, Anne Arundel Co., Md. Gray clay exposed just above beach level, passing laterally into red clay, overlain by yellow-white sands with ferruginous ledges. “Raritan” Fm.: Patapsco-Raritan transition zone. Ch-Bf 127(536-7) and Ch-Bf 127(546-7): well ca. 1.5 mi. NE of Waldorf, Md. (Ch-Bf 127: cf. Hansen, 1968). Medium gray clay core samples from 536-537’ and 546-547’, obtained from Maryland Geological Survey. Near top of “Raritan(?) -Patapsco” Fm.: Patapsco-Raritan transition zone. TR(1551-3): Toms River Chemical Co. Test Well No. 84, 39° 59’ 3” N latitude, 74° 14’ 20” W longitude, Ocean Co., N.J. Gray clay core sample from 1551-1553’, obtained from H. E. Gill through J. P. Owens, U.S. Geological Survey. Near top of Patapsco-Raritan transition zone: ecigey from 1369-1371’ and 1298-1300’ yield typical Woodbridge pollen and spore RARITAN FORMATION: NJ 2: “Woodbridge, N.J.” Light gray clay matrix from specimen of Magnolia glaucoides Newberry, N.Y. Botanical Garden Paleobotanical Collections. Wood- bridge Cla 68-8: S side of Sayre & Fisher Brick Co. pit, on S side of Main St., just NE of Sayreville, N.J. Near top of massive dark gray clay exposure. Woodbridge Cla 68-10: E side of same pit. Medium gray clayey sand at top of massive dark aed Woodbridge 68-12: NE end of shies sand pit ca. 0.5 mi. NNE of Phoenix, N.J. Gray clay capping thick cross-bedded sands. Old Bridge Sand? 68-23: W side of clay pit on N side of Washington Rd., ca. 0.5 mi. E of Parlin, N.J. Near base of thin-bedded gray clay unit, underlain by light gray sand, at low elevations in pit. South Amboy Fire Clay. 68-25: same pit. Near top of same clay unit, exposed just to W and 10-20’ higher. South Amboy Fire Clay. 8-26: same le Thin bed of laminated gray clay exposed near top of small hill near NW corner of pit, underlain by white sand, and overlain by thin bed of thinly eaaOG lignitic sand (68-27). South Amboy Fire Clay or Old Bridge Sand. 68-27: see under 68-26. 68-28: S side of abandoned sand pit off E side of Hillside Ave., just N of high voltage wires, Sayreville, N.J., above and to E of Sayre & Fisher pit (68-8, 1969 | DOYLE, CRETACEOUS ANGIOSPERM POLLEN 35 68-10). Gray clay lens in predominantly coarse-medium grained sand. Sayreville Sand or South Amboy Fire Clay. MacotHuy ForMATION: 68-14: SW corner of Madison Township dump, 0.3 mi. E of U.S. Rt. 9, ca. 0.9 mi. S of junction with Ernston Rd., and 1.5 mi. SSE of Ernston, N.J. Dark gray clay overlain by thin-bedded alternating sands and clays of the Morgan beds of the Magothy Fm., near lowest elevations in dump. Amboy Stoneware Clay (J. P. Owens, pers. comm 68-16: bluff overlooking Raritan Bay NE of town of Cliffwood Beach, N. Gray silty clay just below contact with wag Merchantville Fm. exposed at top of bluff. Cliffwood beds of Magothy Ch-Bf 127(441-2): same well as Ch-Bf 127 samples under Patapsco-Raritan transition zone. Fine gray clayey sand core sample from 441-442’. Near top of Magothy Fm. Note: Brenner localities are those described in Brenner (1963). Unless other- wise indicated, samples were collected by J. A. Doyle. All slides are located at the Harvard University Paleobotanical Collections. DEPARTMENT OF BIOLOGY Harvarp UNIVERSITY 36 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 COMPARATIVE ANATOMY AND RELATIONSHIPS OF COLUMELLIACEAE WILLIAM L, STERN, GeorcE K. Brizicxy,' Fs AND RICHARD H. EypE Género dedicado 4 Junio Moderato Columela, antiguo espaiol, colocado por Linneo entre los padres de la Botanica, y que escribo elegantemente en prosa y verso de Labranza y cultivo de Jardines — Ruiz and Pavon 1794. In 1961, Brizicky summarized information on the Andean genus Col- umellia and presented a taxonomic synopsis of this puzzling group of plants. The genus was described in 1794 by Ruiz and Pavén and David Don estab- lished Columelliaceae in 1828. Eleven species have at one time or another been ascribed to the genus and through his critical examination of all available herbarium specimens, Brizicky reduced this number to four more or less well-defined species. Evaluations of the taxonomic position of Columellia and Columelliaceae have been set forth from the time of A. L. de Jussieu and Ruiz and Pavén, but even the latest authors have been unable to fix the relationships of these plants conclusively. ‘With its peculiar combination of opposite, exstipulate leaves; bisexual, epigy- nous flowers; somewhat irregular, sympetalous corollas; two stamens with plicate and contorted anthers resembling those of some Cucurbitaceae; two-carpellate, imperfectly two-locular ovaries; and imperfectly four- locular capsular fruits, Columellia is indeed a unique genus’ (Brizicky 1961). Although several positions have been proposed for Columellia and for Columelliaceae, taxonomists agree that a plausible understanding of the relationships of these plants requires comprehensive studies to clarify dis- puted points and to complete our knowledge of their anatomy. It was with this in mind that the present authors have examined the anatomy of the flower and fruit, node, leaf, and secondary xylem. Taxonomic position of Columellia A. L. de Jussieu (1801) considered Columellia as a genus of Oleaceae “hoc Genus ad Jasminearum ordinem pertinere.”’ Kunth (1818) placed the genus in Scrophularinae, but noted, “‘An Gesnereis affinior?” At first Reichenbach (1828) included the genus in Gesneriaceae (‘‘Gesnereae’’ as a tribe of Bignoniaceae) but later (1837) he transferred it to Oleaceae e K. Brizicky died in 15, 1968 in Cambridge, Massachusetts, during the ie a of the preparation f this — It is to his memory that the sur- yiving authors respectfully cook this pape 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 37 (“Jasmineae”). Bartling (1830) retained Columellia in Scrophulariaceae among “Genera incertae sedis.” Sprengel (1830) supposed the affinity of the genus to be with Gesneriaceae. In 1839, Endlicher placed Columellia near Ebenaceae among “Genera Dubiae Affinitatis”; later (1841), he in- cluded it in his classis (order) Petalanthae (Primulaceae, Myrsinaceae, Sapotaceae, Ebenaceae, and Styracaceae) as a genus “‘Petalanthis affinis.” Schnizlein (1843-1870) recommended an affinity with Saxifragaceae- Escallonioideae (‘‘Escallonieen’”’), and particularly with the genera Argo- phyllum J. R. & G. Forst., Brexia Nor. ex Thou., and Roussea Smith. J. D. Hooker (1873, 1875) suggested referring the genus to Loganiaceae. Baillon (1888) included Columellia in Gesneriaceae as a representative of the monogeneric series Columellieae (between series Gesnereae and series Cyrtandreae). Hallier at first (1901) placed Columellia in Rubiaceae as an anomalous genus and later (1903) included it in Scrophulariaceae as ques- tionably related to Veronica sect. HEBE Benth. of the tribe Leucophylleae. Finally (1908, 1910) he transferred it to Saxifragaceae-Philadelpheae. Herzog (1915) also regarded Columellia as a genus of Saxifragaceae. Taxonomic position of Columelliaceae David Don (1828), who founded the family Columelliaceae, considered it allied to Oleaceae (“Oleinae” and “Jasmineae”) as well as to Styra- caceae and Ebenaceae. Apparently following the suggestions of his brother, George Don (1838) showed Columelliaceae (“Columellieae”) to contain three genera: Columellia, Menodora Humb. & Bonpl., and Bolivaria Cham. & Schlechtd. (= Menodora Humb. & Bonpl.). He placed the family between Oleinae and Jasmineaceae. Grisebach (1839) presumed a close relationship with Gentianaceae. Meisner (1836-1843) favored the affinity of Columelliaceae with Oleaceae. De Candolle (1839) assumed a close relationship with Gesneriaceae. Adrien de Jussieu (1848) placed Columelliaceae in Rubiales between Caprifoliaceae and Valerianaceae. Lindley (1835) put Columelliaceae in his alliance (order) Cinchonales (Rubiales) between Vacciniaceae and Cinchonaceae (Rubiaceae) with which families and Onagraceae he thought it related. He also presumed an affinity of Columelliaceae with Caprifoliaceae. Agardh (1858) sug- gested a close affinity of the family with Lythraceae (“Lawsoniae”). Basing his conclusions on the contorted anthers in both Columelliaceae and Cucurbitaceae, Clarke (1858) asserted that, “... if the nearest affinity of this family [Columelliaceae] is not with Cucurbitaceae, yet there is no other to which it more closely approaches. . . .” Following de Candolle, Bentham and Hooker (1876), and several of the more recent taxonomists — Fritsch 1894, Engler 1892 (unchanged in Melchior’s 1964 edition of Engler’s “Syllabus der Pflanzenfamilien”), Schlechter 1920, Wettstein 1935, and Pulle 1952 —placed Columelliaceae near Gesneri- aceae. Fritsch emphasized the similarity with Bellonia L. (Gesneriaceae). Nevertheless, Wettstein stressed the continuing uncertainty of the sys- tematic position of Columelliaceae. Warburg (1922) placed Columelliaceae near Gesneriaceae also; however, he noted: “Am natiirlichsten diirfte die 38 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Stellung bei den Rubiaceen sein.” In 1959, Takhtajan allied Columelliaceae closely to Gesneriaceae, particularly with the genus Ramonda Rich. Here, and in his 1966 work, he stated that Columelliaceae is a derivative of Gesneriaceae. Hutchinson (1959) placed Columelliaceae in Personales with the families Scrophulariaceae, Acanthaceae, Gesneriaceae, Orobanch- aceae, and Lentibulariaceae. Airy Shaw (in Willis 1966) stated: “Despite the sympetaly, slight zygomorphy and curious anthers [in Columelliaceae], probably related to Escalloniac. and Hydrangeac.; perhaps also to Loga- niac.” In his recent conservative treatment of Saxifragaceae, Thorne (1968) treated Columelliaceae as a subfamily adjacent to Escallonioideae and Montinioideae. Columelliaceae is placed in Rosales by Cronquist (1968) near the Pittosporaceae and Grossulariaceae. Anatomists have examined the microscopic structure of Columelliaceae in an attempt to establish its affinities with more certainty. Solereder (1899) was able to study the structure of Columellia oblonga Ruiz & Pavon ssp. serrata (Rusby) Brizicky (= C. serrata Rusby) and concluded that the occurrence of scalariform perforation plates and fibrous elements with conspicuous bordered pits in the secondary xylem precluded any close affinity with Gesneriaceae. Rather, he thought, Columelliaceae showed anatomical similarities to Saxifragaceae. Van Tieghem (1903), having several species of Columellia at his disposal, confirmed Solereder’s anatomical observations, thus establishing the homogeneity of secondary xylem structure throughout the genus. However, van Tieghem believed Columelliaceae to be best placed in his alliance Rubiales near Rubiaceae. Metcalfe and Chalk (1950), having no further material at their disposal, repeated Solereder’s findings. Erdtman (1952) stated that pollen mor- phology of Columelliaceae does not give any positive indications of the affinity of the family. He does remark, however, that “The following families have been mentioned as possibly related [to Columelliaceae]: Ebenaceae, Ericaceae, Gesneriaceae (the grains oF Bellonia {Gesneri- aceae| are not similar to those of Columellia!). .. . Columellia, or Columelliaceae, has been considered related to families of both Sympetalae and Choripetalae, to families with superior ovaries and to others with inferior ovaries. Some proposed relatives have stipules and others are exstipulate; some proposed relatives have opposite leaves and others have alternate leaves; some proposed related families are largely herbaceous and others are mostly woody. Among the taxa sug- gested as relatives, the following seem to predominate: The first proposals indicated Oleaceae; later the Ericaceae-Vaccinioideae and Rubiaceae were recommended; Scrophulariaceae appeared a few times in the literature during the early 19th century; but Gesneriaceae seemed most strongly defended in the late 19th and early 20th centuries. Although alliance with Saxifragaceae was suggested in the mid-19th century, it was not until the early 20th century and later that the proposal seemed to gain strength. Several other families have been proposed, though not as often as the foregoing: Ebenaceae, Styracaceae, Gentianaceae, Loganiaceae, Capri- 1969 } STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 39 foliaceae, and Onagraceae. Today, both the gesneriaceous and saxifra- gaceous hypotheses of relationship seem to have equal standing among plant taxonomists, although the most recent treatments favor alignment with saxifragaceous taxa. It is clear, though, that the variety of families proposed as relatives of Columellia (Columelliaceae) could not be much more diverse. MATERIALS AND METHODS In drawing comparisons between Columelliaceae and other families, it has been necessary for convenience and clarity to accept certain taxo- nomic delineations and judgements. This is especially important in refer- ring to the Saxifragaceae which has been treated in different ways by different authors. Engler’s (1928) treatment is the most detailed to date and his concept of the family is very broad. He divides Saxifragaceae into several subfamilies, namely, Penthoroideae, Saxifragoideae, Lepuro- petaloideae, Parnassioideae, Tetracarpaeoideae, Pterostemonoideae, Iteoi- deae, Brexioideae, Kirengeshomoideae, Kanioideae, Baueroideae, Hydran- geoideae, Escallonioideae, Montinioideae, and Phyllonomoideae. Thorne’s (1968) outline is very reminiscent of Engler’s treatment. In our paper, when ‘“‘Saxifragaceae, sensu lato,” is employed, it is used in this broad Englerian sense. Other taxonomists have chosen to disassemble the Englerian conglom- erate into several smaller families; hence, Hutchinson (1967) treated Engler’s subfamily Escallonioideae as the family Escalloniaceae and his subfamily Hydrangeoideae as the family Hydrangeaceae. Engler’s tribe Philadelpheae of Hydrangeoideae is considered as Philadelphaceae by Hutchinson. The genus Rides L. is part of the subfamily Saxifragoideae in Engler but Hutchinson treated it as the basis of the monogeneric fam- ily, Grossulariaceae. Cronquist (1968), similarly, has dissected Engler’s Saxifragaceae. Because our comparisons among the vegetative parts of plants depend heavily on the information in Metcalfe and Chalk (1950), we have used their taxonomic designations for the Englerian subfamilies. The concept of Saxifragaceae employed by these two plant anatomists is wholly herbaceous, and the woody taxa in Engler’s Saxifragaceae are relegated to other families, e.g., Escalloniaceae, Grossulariaceae, and Hydrangeaceae (including Hutchinson’s Philadelphaceae). “Saxifrag- aceae, sensu stricto,’ as we have used it, refers to a strictly herbaceous family conforming to the sense of Metcalfe and Chalk. Terminology used in the descriptions of xylem anatomy follows that prescribed by the Committee on Nomenclature of the International Asso- ciation of Wood Anatomists (1957). Other terminology used in descrip- tions of anatomical structures is that in current use and deviations from common usage are explained where they occur. TABLE | contains a detailed listing of specimens employed in the study of the vegetative anatomy of Columellia; materials used for comparative 40 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 floral anatomy are cited in the text. Fluid-preserved material of about 30 flowers of C. oblonga ssp. oblonga was available from one of Tovar’s collections (4033, USM). All study specimens of wood, stems, leaves, and flowers (except for comparative floral material of Escallonia and Car- podetus), are supported by herbarium vouchers and their place of deposit is noted in TABLE 1 or in the text. Methods of preparing specimens for study followed standard laboratory techniques. Woods were boiled in water to hydrate and stored in 70 per- cent ethanol prior to microtoming. Transverse, radial, and tangential sections of wood were stained with Heidenhain’s iron-alum haematoxylin and counter-stained with safranin. Macerations of wood were prepared using Jeffrey’s fluid. Clearing of leaves was carried out using Arnott’s (1959) method involving 5 percent NaOH followed by a saturated aqueous solution of chloral hydrate. After washing in water, leaves were stained in aqueous safranin to accentuate vascular detail, dehydrated, and mounted on glass slides in Canada balsam. Transverse and paradermal sections of leaves were also prepared after embedding in paraffin, These were stained in Heidenhain’s iron-alum haematoxylin and safranin. Nodal and petiolar anatomy were studied from hand-cut sections treated with phloroglucinol and concentrated HCl to differentiate the lignified tissues. Observations of floral anatomy were performed from serial microtome sections (trans- verse and longitudinal), cleared thick sections, and cleared whole flowers of Columellia oblonga ssp. oblonga. These preparations were made using familiar microtechnical methods from flowers fixed in formalin-acetic acid- alcohol. ANATOMY The flower Transverse sections through the base of the Columellia gynoecium show two locules separated by a thick septum (Fic. 1, d, d'). In successively more distal sections the placentas appear first as single lobes on each side of the septum (Fic. 1, e; Fic. 3), then as deeply two-lobed structures bearing many unitegmic ovules (Fic. 1, f). In still more distal sections there is an opening between the locules (Fic. 1, g, h), but the uppermost level of the ovary may again be divided by a complete septum (Fic. 2) through which the stylar canal enters the ovarian cavity. If the stylar canal is followed distally its appearance in transverse sec- tion changes from that of a single cavity to that of a pair of tracts filled with pollen-transmitting tissue (Fic. 1, j, k). The pollen-transmitting tracts expand greatly below the two-lobed stigmatic surface, producing the unusual transectional effect shown in Fic. 4. The outer layers of gynoecial tissue, from the stylar base to the corolla, constitute a nectary of small cells with densely staining cytoplasm (Fic. 2). Flowers of Columellia are devoid of unusual histologic features that can be used as taxonomic markers. The hypanthium, like the foliage, is 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 41 Fic. 1. Columellia oblonga, flower. Camera lucida drawings of selected trans- verse sections, arranged sequentially from pedicel (a) to upper part of flower (k). D, dorsal carpel bundles; S, stamen supply. Fics. 2 and 3. Columelia aban: “ey in transverse section. Fic. 2. “Upper (free) part of gynoecium, showing nectary and upper ovarian septum, X 30. Fic. 3. Lower part taf ower showing basal septum, arrangement of vascular bundles (cf. Fic. le); arrows indicate bundles supplying the 2 stamens, 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 43 covered with simple, appressed trichomes. Floral tissues contain no con- spicuous tannin cells or sclereids and no crystal inclusions except for a few scattered druses. The anthers dehisce with the aid of the familiar subepidermal banded layer (Fics. 6, 7); moreover, the sporogenous por- tions, in spite of their peculiar external form, resemble in section the corresponding parts of ordinary four-locular anthers. The anther sacs, at least the young ones, are minutely glandular-hairy at the margins, the glandular trichomes being more or less club-shaped. The gynoecium con- tains a well-marked endocarp tissue, four to six cells deep on the dorsal side of the locule, gradually decreasing in thickness in the vicinity of the septum and the placentas. Cell walls of the endocarp are neither lignified nor greatly thickened in newly opened flowers, and there is no anatomical indication of a dehiscence line at this stage. Floral vascular bundles, many of them amphicribral, diverge from a continuous cylinder in the pedicel (Fic. 1, a, b). Well below the base of the locules, the cylinder expands into the pattern shown in Fic. 1, c, with an inner portion of the vascular tissue directed to the septum and the placentas and an outer portion directed to other parts of the flower. A few sections above this level, and still below the locules, the outer portion separates into two series of traces, a gynoecial series and a series supply- ing perianth and stamens. With additional branching at even higher levels (Fic. 1, d, d?, e), the gynoecial series contains as many as 20 bundles per carpel, and the other series (now outermost) contains about a dozen perianth traces plus two stamen traces. A stamen trace can be united for part of its length with the basal extension of a sepal midvein or it can be completely free of other bundles to the base of the flower. In either case, the position of the stamen traces is the same; they occupy roughly the same radius as the septum. The perianth traces, if followed distally, become the major veins of sepals and corolla lobes. As in many other kinds of flowers, there are lateral connections between these strands at the level where the calyx and corolla become free of the ovary wall, and minor strands diverge from the major ones within the perianth mem- bers. The vascular tissue of the stamen broadens within the filament (Fic. 1, k) and terminates in the connective with a great many short branches. The vascular supply to the placentas rises through the septum in a massive and irregular column or plexus (Fic. 1, c-f). Branches to the ovules diverge from the plexus all through the placental region, but this Portion of the vascular system does not continue above the placentas. The many outer gynoecial bundles, however, extend all the way to the base of the style (Fic. 1, g-i). Although the dorsal bundle is not easily distinguishable in sections through the lower half of the ovary, it is con- spicuous in higher sections because of its proximity to the locule (Fic. 1, g,h). The dorsal bundle can be followed into the style, which it enters as a single well-defined strand. About a third of the way up the style, it divides into two or more strands, which subdivide further into many [voL. 50 JOURNAL OF THE ARNOLD ARBORETUM I ‘bZ872 MA Sn ‘A ‘A ‘Vv lopeno gy Zz1 youquiry 96b6T MA A‘HD ‘A Jopensy Of Hequiry I Sn ‘AN ‘HD Jopenoy 9F802 Y2094ITH Ayoz gq (M'S'H) vadtuas ‘dss UOARY FY zZiny vsuojqo I sa nod 6PLT WaqiIN Y Yoo) I sn niad OS8 Weq[iH ®R yoo) I sa nia 8ObL SesIeA I SN ‘A ‘AN ‘H9 Jopenoy 19pp—q due} 14 SN ‘OW “A ‘V nad 16L2 Janeqiaqa 19 SN ‘HD nad P8ss Janeqiaqe 7 d nied ZS/T ugaeg ® ziNy I sa nag S8LE IPAOL, J} ‘T ‘p969¢ “MSO wsn nia E£LOh APAOL I ‘srsze MSA sn nia ZELT AOUPANM piuojqo ‘dss ugARg ® ziny vIuojqo I sn 001 auss “erquiooD) p8Lz SN I sa nia Ore S19qpaty 3 AN ‘HD ‘a Jopenoy OOsr 94puy 7 Hd ‘a Jopenoy brrI-M 2apuy uozauLiay) ®W Ansueq vpiany » GAIGALS AIHONOA NIOIUQ AOLOATION sa1oadS SLaVd pourwexg eijoumyog jo suaurtseds ‘T aTav STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 45 1969] “WIEYD PUL WING PUE (1961) WANG MOYO} sUOTAPO? poom [RUON NTU! JO EUOTIEAIIQAE ‘saMoy yA = 1 poom = AM yea = Le (0961) S19q — eee ee Sf ‘AN ‘HD ‘V g6Lres SA p6Lbes SA sa sopenoy £69, Teme, 9971 JYAOL GOLE F9YNN Opsyye Iste tauayy ZSbPs saNEqsaqa\\ ZLI1 Burg SLIIZ B1aqso4 Z7S1L punjdsy L9OT UAE] B Uospod 8197 Aaqung ‘u's uosowef fli-q Mad YOALY BY ZINY Pypacgo AU (Aqsny) ogpisas ‘dss uoaud R ZNyY PFuo/qo 4-7. Fic. 4. Columellia oblonga, of sole, xX 55. Fic. 5. Kohleria elegans (Gesneri GS. 6 and 7 transverse section through upper part aceae), transverse section I i pee ig transverse sections of mt at before and after dehiscence. Fic. S325. Pa, 7. 95. 1969] STERN, BRIZICKY, & EYDE, COLUMELLIACEAE ower. Fic. 8. Placental region, 25. Fic. x ee below placenta to show basal septum and ventral bundles (arrows), 9. Another flower, 48 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 strands just below the stigma. The remaining bundles of the gynoecium wall converge upon the dorsal at the base of the style (Fic. 1, i); how- ever, they do not appear to merge with the dorsal, because it enters the style with its cross-sectional shape and dimensions unchanged. The leaf and the node Hairs on leaves of Columellia are thick walled and simple, tapering to the obtuse tip and slightly swollen or bulbous at the base (Fics. 10, 11). Trichomes emanate from the center of saucer-shaped depressions in the lower epidermis. These are formed from several radially oriented cells each of which is thicker toward the periphery of the depression and thinner toward the center where the hair arises (Fics. 10, 11). The cuticle is thick and covers upper and lower epidermis. It is espe cially pronounced toward leaf margins and in the trichome-base depres- sions of the lower epidermis. It also covers all portions of hairs. The cuticle is strongly modified in the stomatal region; it covers the exposed surfaces of guard cells and it over-arches both the outer portion of the aperture producing a front cavity and the inner portion producing a back cavity (Fics. 11, 12). Stomata are restricted to the lower epidermis. The stomatal apparatus 3 is anomocytic (sensu Metcalfe and Chalk, 1950), ie., the guard cells are surrounded by cells of varying number which are indistinguishable 1 form or position from the remainder of the epidermal cells (Fic. 10). Guard cell walls are thickened along the inner surface facing the spongy mesophyll and on the outer surface (Fics. 11, 12). In paradermal view, guard cells are elongate-reniform (Fic. 10). ; The lower epidermis is uniseriate; the upper epidermis is biseriate (Fics. 11, 13, 15). Since developmental studies could not be conducted, it is not possible to determine if the inner layer is protodermal in origin or if it arose from the ground tissue. Inner cells of the biseriate uppeT epidermis are larger and conspicuously more rotund than those of the outer layer (Fics. 13, 15). Leaves are dorsiventral and the mesophyll is divided into a biseriate, upper palisade layer and a lower spongy lay = In the thickish leaves of Columellia lucida and C. obovata, the transition between palisade and spongy mesophyll is not sharp. Furthermore, 19 these two species, there is a tendency for a lower palisade layer to formed and an isobilateral condition (Fic. 15). : Leaves of all species of Columellia are glandular; in those species having serrate leaves, the tips of the teeth and the apical point are glandular, in species with entire leaves, the apex of the leaf may be glandular. * Although Metcalfe and Chalk (1950), Fahn (1967), and Esau (1965) do not agree, the first author would prefer to use the term stoma (Gr. a mouth) in its re- stricted sense to mean the actual aperture or pore in the epidermis which is sur- stoma, guard cells, and subsidiary (accessory) cells, if present. The maintenance of separate terms for the aperture and guard cells seems meritorious in that it provides for independent reference to each of these units and alleviates the possible redundant implications of referring to the “aperture of a stoma.” showing biseriate upper epidermis, biseriate palisade layer, and uniseriate lower €Pidermis. The cuticle overarches the unevenly thickened guard cells externally and internally to form front and back cavities. Bases of hairs are situated in Saucer-like depressions of the lower epidermis. 12-15. Fic. 12. Columellia lucida, Friedberg 240, transverse section of lower epidermis of leaf showing thickened cuticle and ae cuticular modification in association with stomatal apparatus, X 600. Fic. 13. C. oblonga ssp. oblonga, Tovdr 4033, transverse section through mid-vein of leaf showing biseriate upper epidermis, uniseriate lower epidermis, bun sheath, and bundle sheath extensions, * 210. Fic. 14. C. oblonga ne be Tovér 3785, cleared whole mount of leaf showing a single glandular ration; ark bodies in glan are fruiting structures of an aspergillous rea < 40. Fic. 15. C. lucida, Friedberg 240, transverse section through mid-vein of leaf showing biseriate u Ppe epi is, uniseriate lower epidermis, bundle sheath and bundle sheath r epider tensions, tendency to development of a —e palisade layer, and abundance of thick. walled fibers in the vascular bundle, 180. tte 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 51 Fics. 16 and 17. Sectional series through petioles of Columellia, (a) being = (c) proximal, showing increasing distal development of sclerenchyma. . C. oblonga ssp. sericea, Drew E-113. Fic. 17. C. oblonga ssp. oblonga, leer & Gilbert 1749. (x) xylem, (p) phloem, (s) sclerenchy ma. Glands are highly vascularized and massive (Fic. 14); proximally adjacent to the secretory epithelium is a cupulate reticulum of vascular elements. That the central portion of the gland contains a cavity is borne out by the occurrence there of aspergillous fruiting bodies in some specimens. Apices of glands are aperturate probably through schizogeny. Vasculation of the petiole is characterized by a single collateral strand of conducting tissue varying from crescentiform to cupulate to almost semiterete in transverse section (Fics. 16-20). Xylem is adaxial and phloem is abaxial. In all species examined, an abaxial sclerenchymatous region develops progressively from the proximal to the distal portion of the petiole (Fics. 16-18). In specimens of Columellia oblonga ssp. ob- longa (Fic. 17) and C. lucida (Fic. 19), a well-developed lunate layer completely subtends the phloem at the extreme distal end of the petiole; in specimens of other species (Fics. 16, 18, 20) the sclerenchyma seems not to develop into more than a series of widely-spaced rods at this point. However, sections through the mid-vein of the lamina in C. oblonga ssp. 52 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Fics. 18-20. Fic. 18. Columellia oblonga ssp. serrata, Bang 1172, sectional series through petiole, ~ bey * Lege (c) proximal, showing increasing dista development of scleren C. lucida, Friedberg 240, distal section of collie. showing one “cerenchynatos arc. Fic. 20, C. obovata, Vargas 7693, distal section of petiole wing sclerenchyma as an arc of rods at this point. (x) xylem, (p) phloem, () sclerenchyma. sericea (Drew E-113), which shows a series of sclerenchymatous rods at the distal end of the petiole (Fic. 16, a), show a complete sclerenchyma- tous layer subtending the phloem. It is likely, therefore, that in the laminae of all species of Columellia, the mid-vein is supported by an abaxial layer of sclerenchyma. The central vascular strand of the petiole branches into a series of minor strands toward the base of the lamina (Fics. 16-20). In Columellia oblonga the mid-vein of the lamina is characterized by secondary growth and several layers of secondary xylem and phloem are produced (Fic. 13). In C. lucida and C. obovata, secondary growth is not pronounced; furthermore, in these species most of the xylem in the mid- rib and secondary veins consists of thick-walled fibers (Fic. 15). Bundle sheaths surround secondary veins in all species. Bundle sheath extensions (Wylie, 1952) reach upper and lower epidermises in C. oblonga (Fic. 13); in C. lucida and C. obovata, there are no bundle sheath extensions asso- ciated with the bundle sheaths of secondary veins. The node in Columellia is unilacunar and a single trace emerges through each of the two opposite gaps in the vascular cylinder (Fic. 21). 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 53 F Transverse section of stem illustrating the unilacunar node in Colu- mellia: “a) xylem, (p) phloem, (It) leaf trace. The secondary xylem The wood of Columellia is generally without growth rings, although in the immature specimens of C. obovata, represented by Weberbauer 5482 and Nunez 3309, more or less sharply defined rings occur. However, both of these specimens show strong evidence of decay or disease and it is sus- pected that the growth rings are related to these conditions. All woods examined are diffuse-porous, the strictly solitary, uniformly-sized pores being distributed evenly across the transverse surface (Fic. 23). Vessel walls are thin and there are no tyloses. Pores are angular. Data for measurements of vessel diameter, vessel element length, bars per scalariform perforation plate, tracheid length, and heights of vascular rays are presented in TABLE 2. Because both mature and immature wood were examined, measurements for each are separated in the table to provide a more meaningful basis for comparisons with xylem in other taxa. Vessel elements are generally long and narrow although ligules as such are short and sometimes lacking. End wall angle ranges from 10° to 45°. Perforation plates are entirely scalariform (Fic. 27) and in some cases bars are so profusely branched they give the appearance of pits. Openings in scalariform perforation plates are completely bordered. Spiral thicken- ings occur in the cell walls of ligules throughout all species, being more prominent in some than in others. In specimens of Columellia oblonga ssp. oblonga, vaguely outlined spirals are seen in the body segment of vessel elements and they are strongly marked in the ligules; in C. oblonga 54 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ( a RPS 2 ee RK Soars se 8 aN | ae Owe oe Be ot Fics. 22-25. Fic. 22. Escallonia myrtilloides, Rimbach 13, Yw 16920, trans- verse section of xylem showing solitary distribution of angular pores, 100. Fic. 23. Columellia oblonga ssp. sericea, Rimbach 122, transverse section of xylem showing solitary distribution of angular pores, and scanty vasicentric and diffuse axial parenchyma, * 100. Fic. 24. E. myrtilloides, pranayese section of xylemswith biseriate vascular rays and spiral thickenings in tracheids and vessels, X 100. Fic. 25. C. oblonga ssp. sericea, oo section of xylem showing uniseriate vascular rays and tracheids, « 1 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 55 Fics. 26-28. Fic. 26. poet myrtilloides, Rimbach 13, Yw 16920, radial section of xylem showing scalariform perforation plates, x 150. Fic. 27. Colu- mellia oblonga ssp. sericea, Rimback 122, hi dial section of xylem showing scalariform perforation plates, x 100. Fic. c. degegens W eberbauer 5482 longitudinal section of xylem showing spiral Sry toni in \ 500 56 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ssp. sericea, spirals are tenuous at best and appear only in ligular portions; in C, cida spirals occur only in ligules; and in C. obovata ee are peach throughout the lengths of vessel elements (Fic. Intervascular pitting is generally absent owing to the nee nature of vessels; however, a suggestion of intervascular pitting is sometimes present in the overlapping ends of superposed vessel elements. In these areas, the circular to elongate pits are sparse and irregularly scattered but there is a tendency toward the alternate arrangement. Imperforate tracheary elements are tracheids, the pits in these cells being of the same order of magnitude as those which occur in the over- lapping ligulate portions of vessel elements (Fic. 25). Pitting in tracheids is ordinarily uniseriate; less commonly two rows of pits are present, stag- gered alternately. Inner apertures of pits are elliptical, crossed in face view, and included within the pit border. Tracheid walls vary from very thin to thick. Vascular rays are entirely uniseriate and comprise axially elongated or upright cells only (Fic. 25). These rays are homocellular and the ray tis- sue corresponds with Kribs’ (1935) Heterogeneous Type III. TABLE 2. Summary of Xylem Anatomical Measurements in Columelliaceae Marovre * IMMATURE ” VESSEL DIAMETER IN p Average: 45 25 Range: 22-105 12-45 MFR *°: 30-70 15-36 VESSEL ELEMENT LENGTH IN pb verage: Range: 308-1100 MFR: 375-828 BARS PER SCALARIFORM PERFORATION PLATE Average: 14 10 Range: 7-20 3-24 : 11-16 6-17 TRACHEID LENGTH IN p Average: 866 Range: 378-1260 MFR: 625-1110 HEIGHT OF VASCULAR RAYs IN CELLS Range: 1-6 1-47+ * Columellia oblonga : oblonga, Tovdr 4033, Wurdack 1732. C. oblonga ssp. sericea, Rimbach 122 and 3 olumellia oblonga ii ski Psi & Pavén 1/52; Weberbauer 5584 and 7791. C. lucida, André K— 1444 and 4500. C. obovata, Weberbauer 5482, Herrera 3451. Data from Nufez 3309, a d saat are not included here. ° MFR = Most frequent range. 1969} STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 57 Axial xylem parenchyma is largely scanty vasicentric, a few isolated strands occurring about the vessels (Fic. 23). In addition a few strands were seen embedded within the groundmass of tracheids (diffuse paren- chyma). DISCUSSION In view of the sympetalous corolla of Columellia, it is not surprising that taxonomists looked for its relationships among the sympetalous families, and especially those with inferior ovaries and opposite leaves. Among other features, the androecial peculiarities of Columellia, un- matched in any other known taxon, persuaded David Don to establish a separate family for this unusual group of plants. Time has shown him to have been correct in his assessment of the individuality of Columellia. Evidence from gross morphology There are such sharp differences in floral structure between Columel- ated on an axile placenta in each locule, can hardly be regarded as closely related to Columelliaceae. The mostly herbaceous Gentianaceae-Gentian- oideae show some similarities with Columelliaceae in their cymose in- florescences, in the structure of ovaries and fruits (2-carpellate ovaries with numerous unitegmic, tenuinucellate ovules on parietal intrusive to axile placentas, septicidal capsules, small seeds, etc.), as well as in the possession of opposite, exstipulate leaves. They are markedly different, however, in their regular flowers; in the usually contorted aestivation of corolla lobes; in their usually dorsifixed, introrse anthers; and in their superior ovaries. Loganiaceae (excluding Desfontainea Ruiz & Pavon) differ from Columelliaceae in their usually stipulate leaves, regular flow- ers, and superior ovaries; in addition, in the subfamily Buddleioideae, the presence of glandular and stellate hairs is widespread. Some relation- ship with Scrophulariaceae and especially Gesneriaceae appears possible, but both families have highly specialized, mostly hypogynous flowers (only Gesnereae of Gesneriaceae, sensu Fritsch 1893, 1894, have semi- inferior or inferior ovaries). Some genera of Rubiaceae agree with Col- umelliaceae in floral structure (except for the non-reduced number of stamens) and opposite leaves, but they differ in the presence of stipules. A close relationship with Caprifoliaceae seems equally doubtful. The only genera of this family which are perhaps comparable with Columel- liaceae in possessing multi-ovulate, 2-carpellate ovaries, are Diervilla Mill. and Weigela Thunb., genera apparently restricted to the temperate zones of North America and eastern Asia. The gross-morphological similarities between Lythraceae and Onagraceae and Columelliaceae are too scarce even to suggest a relationship. Within the saxifragaceous families — 58 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Hydrangeaceae, Grossulariaceae, and Escalloniaceae— almost all the gross-morphological characters in Columellia may be found: frutescent and/or arborescent habit; opposite, ba os often glandular-dentate leaves; 5-merous haplostemonous flowers (Escalloniaceae); sympetalous corolla (Roussea of Escalloniaceae) ; a semi-inferior or inferior, 2-car- pellate ovaries with parietal intruding placentas bearing numerous, ana- tropous and apotropous, unitegmic, tenuinucellate ovules (Escalloni- aceae and some genera of Hydrangeaceae). Septicidal capsules usually have numerous small endosperm-containing seeds (Hydrangeaceae and Escalloniaceae) with small embryos. Most of the features of Columellia are represented in the family Escalloniaceae. Although alternate leaves predominate in this family, opposite leaves are found in the genera Grevea Baill., Roussea, and Polyosma Blume. Other genera, as Valdivia Remy, have subopposite leaves. Evidence from floral anatomy It would not be practical, nor is it necessary, to attempt a detailed anatomical comparison of the Columellia flower to flowers of all the plant families with which Columellia has been allied. A brief commentary on the Cucurbitaceae seems to be in order, however, because androecial structure in that family has significance for the interpretation of the an- droecium in Columellia. Clarke (1858) considered the stamens of Columellia, because of their contorted anthers, to be almost identical to those of many Cucurbitaceae. He interpreted the androecia of certain cucurbits — those with three ap- . pendages, one two-locular and two four-locular — as comprising two and a half stamens, an opinion shared by some other 19th century botanists. If this view were correct, the two-staminate androecium of Columellia would not seem greatly different. In more recent times, however, an alternative interpretation of such cucurbitaceous androecia has been con- firmed again and again; that is, the two-sporangiate stamen is an entire one, and the four- -sporangiate stamens are duplex appendages. Evidence for the more modern view is now overwhelming. It is derived from on- togeny; from vascular anatomy (the duplex stamens sometimes contain two well-defined bundles that are derived from two different petal traces) ; and from comparative studies of male, female, and bisexual flowers 0 many genera, some of them exhibiting transitional stages between the five-staminate condition and the “two and a half’’-staminate condition. Reviews of the evidence are given by Miller (1929) and McLean (1947) and additional confirmation by Bhattacharjya (1954), Chakravarty (1958), and Quang (1963). Although Columellia stamens are superficially similar to the duplex stamens of Cucurbitaceae, the vascular supply in Columellia is a solitary bundle. In transverse sections through the filament or the connective, the bundle is often very broad and may occasionally seem to have two xylem patches, but its appearance within the inferior part of the flower gives no 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 59 hint of compound structure. Observing this, van Tieghem (1903) con- cluded that the two members of the Columellia androecium are solitary stamens, and most floral morphologists would accept his evidence. Thus, it can be argued rather convincingly that the evolutionary modification leading to the two-staminate condition in Columellia was a loss or “abor- tion” of stamens rather than any sort of phylogenetic union of stamens. The occasional occurrence of a third stamen in flowers of Columellia, re- ported by Brizicky (1961), supports this argument. This reasoning might be thought to favor the relationship of C olumellia other gesneriads there are only two stamens. In the latter case, as in Columellia, there are no staminodes. But the resemblance of Columellia to the gesneriads is not so close as this information would suggest, for in Gesneriaceae only genera with superior ovaries have the two-staminate androecium (Fritsch 1893, 1894). A satisfactory anatomical comparison of the Columellia flower with gesneriaceous flowers is not yet possible because floral anatomy of the Gesneriaceae has never been investigated to any great extent. Compara- tive information is presently available only for flowers of a Kohleria hybrid, K. amabilis & K. scladotydea (Teeri, 1968), and for those of Kohleria elegans (Dcne.) Loes. (H. E. Moore 8190; US, BH). Serial sections of the latter were prepared from fluid-preserved material especially for this paper. Anatomically, flowers of the two gesneriads are much alike, and they have several characters in common with Columellia. For instance, the floral tissues are devoid of tannins, and general features of placentation and vasculation do not differ greatly from those of Columellia. In addi- tion, both gesneriads have two-lobed placentas and many gynoecial strands (Fic. 8). On the other hand, there are differences in detail that may be important. The two gesneriads have no well-developed endocarp tissue, except for a single layer of transversely elongate cells adjoining the locule. The style is hollow for all of its length, with a single canal (Fic. 5). Floral trichomes are multicellular (but uniseriate). Vascular traces to the stamens are united with sepal traces for part of their passage through the inferior part of the flower, and the supply to the placentas is derived from two large septal bundles (duplex bundles representing paired hetero- carpous ventrals; Fic. 9) in the septum. Of course, a major floral dif- ference is that the anthers of the gesneriads are not contorted. Perhaps the most important difference aside from that is in the nectary: nectaries of Gesneriaceae are usually very well developed and deeply lobed or even divided into distinct appendages (Feldhofen 1933). It is somewhat easier to compare flowers of Columellia with those of Rubiaceae because a detailed survey of floral anatomy in Rubiaceae is available (Rao, Ramarethinam, & Iyer 1964). Rubiaceae is a large fam- ily, rather diverse in floral structure; therefore, it is almost to be expected that some of the members would have characters in common with Col- 60 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 umellia. For instance, some Rubiaceae have separate vascular traces to calyx, corolla, androecium, and gynoecium. And in some genera (e & Guettarda L. ) there are a great many gynoecial bundles. Piece is often similar to that of Columellia, and many genera have an epigynous nectary resembling that of C olumellia. A difference that strikes one im- mediately, when sectioned flowers of Columellia are compared with sec- tions of rubiaceous flowers, is the absence of conspicuous tannins in the former. Floral tannins are rarely lacking in Rubiaceae. Another differ- ence is that a single stylar canal seems to be of universal occurrence in the Rubiaceae. Furthermore, the peculiar androecial modification in Columellia has no counterpart among the rubiads. Floral anatomy of the more easily obtained members of Saxifragaceae, sensu lato, is fairly well known through the investigations of many work- ers, including Palmatier (1943), Morf Shee Dravitski (see Philipson, 1967), Gelius (1967), and Komar (1967). None of these studies has produced evidence to support Hallier’s (1908, 1910) opinion that Col- umellia belongs with the Philadelpheae. In fact, ontogenetic observations on Philadelphus (Gelius, 1967) suggest that evolution has favored an increase in stamen number in this group. In some other genera of Phila- delpheae, a reduction in the number of ovules has led to forms that bear little resemblance to Columellia (e.g., Jamesia Torr. & Gray, Whipplea Torr.). Schnizlein (1843-1870) proposed Brexia and Roussea as close allies of Columellia; however both Brexia and Roussea have superior ova- ries with distinctly two-ranked ovules. Argophyllum, another genus men- tioned by Schnizlein, is also very dissimilar to Columellia, for it has peculiar corolline ligules and T-shaped trichomes like its ally Corokia A. Cunn. (Eyde, 1966). If the relationships of Columellia are to be sought among the escallonioids, attention should be given to genera other than the aberrant Argophyllum and Corokia. Berenice Tul. can also be elim- inated from consideration, because it has recently been transferred to Campanulaceae on anatomical and palynological grounds (Erdtman & Metcalfe, 1963). From the standpoint of floral anatomy, Escallonia Mutis ex L. f. is not as close to Columellia as might be indicated by other evi- dence. Tannins are abundant in floral tissues of Escallonia species and the floral trichomes frequently are multicellular with globular terminal portions; also, the gynoecial bundles are few and the ventral bundles commonly accompany the dorsals into the style. Choristylis Harv. has stamens united with corolla tube, but in other respects the flowers are unlike those of Columellia. One difference is that the gynoecial bundles are few; another is that the nectary is located on the lower part of the corolla tube. The latter character may be sufficiently important to remove Choristylis from its position adjoining Forgesia (Engler, 1928) and to place it elsewhere in the Saxifragaceae, sensu lato. (Agababyan 1964, links Choristylis with Itea on palynological evidence.) Flowers of F orgesia have rather massive multicellular trichomes; otherwise they are anatomi- cally similar to Columellia flowers. To judge from our one sectioned her- 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 61 barium flower, the gynoecial vasculature and the placentation approximate those of Columellia. The nectary, if there is one (it is not easy to tell from dried material), is part of the free portion of the gynoecium, and the androecium shows indications of reduction (abortive locules in some anthers). Forgesia, like many other Saxifragaceae, sensu lato, has free styles that could be viewed as a precursor to the two-canal structure of the Columellia style.* In summarizing this section on floral anatomy, it must be conceded that the cited points of similarity and dissimilarity do not tell us much about the affinities of Columellia. The foregoing commentary includes no strong evidence against the proposed relationship with Gesneriaceae, nor does it include really firm evidence for such a relationship. The same can be said of the possible alliance with Rubiaceae or with the escallonioid group of Saxifragaceae, sensu lato. The reason for this is clear. observed characters in the flowers of Columellia are widely distributed in many plant families, except for the contorted anthers. Ironically, the latter character does not help in placing Columellia because it has not been found in any other group of plants, the resemblance to anthers of certain cucurbits being demonstrably superficial. Evidence from leaf anatomy It does not appear possible to compare all features of the foliar anatomy of Columelliaceae with those of families reputed to be allied to it, since complete foliar surveys of these families are lacking from the literature. An original study of leaves in all these families is surely outside the scope of this investigation. Nevertheless, certain comparisons can be made.* Leaves are dorsiventral in Gesneriaceae. Hairs are always multicellular and they are often situated on a pedestal. They may be glandular or non-glandular. A multiseriate hypodermis occurs in certain species. The stomatal apparatus is often very large and anisocytic. Vascular bundles in veins are not usually accompanied by sclerenchyma. Vasculation of the petiole is various and many genera show a single leaf trace; Alloplec- tus Mart., Besleria L., Episcia Mart., and others have three leaf traces and Klugia notoniana A. DC. shows a large number of separate strands. There is no “‘pericyclic’” sclerenchyma associated with the petiolar vascular strand in Gesneriaceae. Gesneriaceous leaves differ markedly from those in Columelliaceae in their multicellular and glandular hairs, anisocytic sto- matal apparatus, and lack of sclerenchyma associated with vascular tissue. * Observations on the floral anatomy of Escalloniaceae are based on serial sections prepared especially for this paper. Material was obtained from the following sources: Escallonia, fluid-preserved flowers from several cultivars growing in the Los Angeles State and County Arboretum, not vouchered; Carpodetus serratus, fluid-preserved owers from plants cultivated at the University of Auckland, New vouchered ; Quintinia fawkneri, pressed flowers, Brass 4719, US; Choristylis shirensis, pressed flowers, Swynnerton 607, US; Forgesia borbonica de V'Isle 216, US. ? > g ca, pressed flowers, amily circumscriptions follow those used by Metcalfe and Chalk (1950) for convenience in making comparisons. 62 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Rubiaceous leaves are generally dorsiventral; centric and homogeneous - leaf organization occur in a few species. Hairs may be unicellular, multi- cellular and uniseriate, tufted, and rarely peltate. A hypodermis occurs in many species. The stomatal apparatus is _paracytic (rubiaceous ) in most species, as might be expected. The petiolar vascular strand is usually shield shaped with more or less coe Se wings. There are also variously shaped median vascular strands, nearly always associated with smaller accessory bundles toward the wings. In such a large and anatomi- cally diverse family as Rubiaceae, it is not surprising to find foliar re- semblances to Columelliaceae. The only clear and consistent difference is the more or less ubiquitous occurrence of the paracytic stomatal appa- ratus in Rubiaceae. Caprifoliaceae usually have dorsiventral leaves, but the palisade tissue is poorly developed in species of Triosteum L. and Viburnum L. Hairs may be glandular or non-glandular and unicellular, simple and multi- seriate, tufted or stellate, and peltate. Glandular leaf teeth are present in some species. Stomatal organization is frequently anomocytic, but paracytic types occur in the same genera as anomocytic types. Except for Diervilla, a single layer of palisade mesophyll occurs; in Sambucus L. and Viburnum, cells of the palisade layer may have arms. The petiolar vasculation shows a considerable range of structure from a solitary, slightly crescentic bundle to an arc of 3—5 or more separate bundles to a closed vascular cylinder. The anomocytic stomatal apparatus and solitary petl- olar strand in Caprifoliaceae are similar to Columelliaceae, but the para- cytic stomatal apparatus, single-layered palisade mesophyll, and multi- strand and cylindrical vasculation of the petiole, which also occur in some species of Caprifoliaceae, are very different from the situation in Columel- liaceae. Leaves in Saxifragaceae, sensu stricto, are dorsiventral and isobilateral. Hairs are glandular and non-glandular and these may be simple, uniseriate and multicellular; shaggy; and multiseriate. Stomatal organization 1S anomocytic and sometimes subsidiary cells, smaller than neighboring epidermal cells, are evident. The mesophyll in some species of Saxifraga L. is undifferentiated, and in species where it is differentiated, the palisade segment may range from 1 to 7 cells deep. Hydathodes are of common occurrence. Petiolar vasculation is distinctive, especially in Saxifraga where one concentric bundle or one hemi-concentric bundle may occur. In other species of Saxifraga, there are three such bundles, each with its own endodermis. Some Saxifragaceae have the usual collateral bundles. but these may be scattered. The herbaceous Saxifragaceae resemble Col- umelliaceae in the presence of an anomocytic stomatal apparatus, appar- ently modified in some taxa; but the undifferentiated mesophyll in some species of Saxifraga and multilayered palisade tissues in others, are very different from the condition in Columellia. Petiolar vasculation in Saxi- fragaceae bears little resemblance to that in Columelliaceae. Leaves in Grossulariaceae are dorsiventral and bear unicellular and 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 63 also glandular hairs. Pairs of small, circular guard cells are characteristic. The petiole is characterized by three separate vascular strands at the base which fuse distally to produce a single crescentiform strand all supported by sclerenchyma in the “pericyclic” region. The specialized, small circu- lar guard cells vary from those in Columelliaceae but the abaxial scleren- chyma associated with the petiolar bundle also occurs in Columelliaceae. The proximally triple vascular strand differs from the condition in Col- umelliaceae. All leaves in Escalloniaceae are dorsiventral. In Escallonia, foliar hairs are thick-walled and unicellular; in Adbrophyllum Hook.f., hairs are glandular with unicellular heads; in some species of Escallonia hairs are glandular-shaggy with multiseriate stalks; in Quintinia A. DC. peltate hairs occur; and T-shaped hairs occur in Argophyllum. Stomatal organi- zation is variable and pairs of nearly circular, small guard cells, resem- bling those in Grossulariaceae, occur in Escallonia, Itea L., and other genera; the stomatal apparatus in Quintinia is paracytic; and the stomatal apparatus in Brexia is characterized by a double front cavity. A 1—3- layered upper hypodermis occurs in species of Argophyllum, Carpodetus, Escallonia, and other genera. A single-layered palisade mesophyll is pres- ent in two genera, Three vascular bundles enter the base of the petiole in Escallonia, but in E. macrantha Wedd. (= E. polifolia Hook.) and E. rubra (Ruiz & Pavon) Pers., a single crescentiform petiolar bundle with accessory strands is present. Apparently there is no abaxial sclerenchyma present in Escallonia. Brexia appears unique, for besides the abaxial, crescentiform vascular strand in the petiole, there is also a small cylinder of xylem in the medullary region and two abaxial xylem cylinders. Certain similarities between Columelliaceae and Escalloniaceae occur: unicellular, thick-walled hairs; presence of a hypodermis; and a single petiolar vascu- lar strand in at least two species of Escallonia. However, there are also marked differences and Escalloniaceae show glandular and multicellular hairs, grossulariaceous stomatal organization, and a triple vascular condi- tion in petioles of most species of Escallonia. Hydrangeaceous leaves are dorsiventra]l. Hairs are various with long, unicellular trichomes in Jamesia; tufted trichomes in Broussaisia Gaudich. and Pileostegia Hook. f. & Thoms.; and stellate, calcified, and unicellular Deutzia, and Philadelphus. A hypodermis occurs in Broussaisia and in species of Hydrangea, and the epidermis contains some _ horizontally divided cells in Carpenteria Torr. The stomatal organization is paracytic in species of Dichroa Lour. and Hydrangea L. and anomocytic in Phila- delphus. Palisade mesophyll is uniseriate in Deutzia and Philadelphus. The petiolar vascular strand differs throughout the family: It is single and crescent-shaped in species of Deutzia, Jamesia, Philadelphus, Hy- drangea, and Pileostegia; petioles of Decumaria sinensis Oliv., Dichroa febrifuga Lour., and Hydrangea petiolaris Sieb. & Zucc. are characterized by a main abaxial arc with several flat adaxial bundles between the ends. 64 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Additional strands are present in other species, including medullary bundles. Although some foliar similarities exist between some taxa of Hydrangeaceae and Columelliaceae — unicellular hairs, glandular leaf teeth, hypodermis, anomocytic stomatal organization, and arcuate petiolar vascular supply —the differences are equally clear. Multicellular and tufted hairs, paracytic stomatal organization, and multistranded petiolar vascular supply occur in Hydrangeaceae. The remaining plant families which have at one time or another been suggested as near relatives of Columelliaceae or Columellia — Scrophu- lariaceae, Ebenaceae, Loganiaceae, Oleaceae, Lythraceae, Vacciniaceae, Ericaceae, Gentianaceae, and Onagraceae — present a wide array of foliar anatomical features, some similar and others different from Columelliaceae. As should be apparent from the brief comparative summary above, no family presents a consistent foliar pattern which is similar enough in most respects to that in Columelliaceae to convince the critical botanist that leaf anatomy is a key to understanding the relationships of the fam- ily. To be sure, this is probably related to the lack of thorough anatomi- cal investigation in those taxa reputedly related to Columelliaceae, but as the situation stands now, foliar anatomy is at its best equivocal in pointing the way to the relationships of Columelliaceae. Evidence from nodal anatomy According to Sinnott’s (1914) survey of the nodal condition among seed plants, all members of the Tubiflorae, which include Scrophulari- aceae and Gesneriaceae, are unilacunar. However, three or five gaps are typical for Cyrtandra J. R. & G. Forst. (Gesneriaceae). Onagraceae, Ericaceae, Ebenaceae, Oleaceae, Gentianaceae, Loganiaceae, and Rubi- aceae, are also characterized by unilacunar nodes. In addition, some mem- bers of Gentianaceae are multilacunar and some Rubiaceae are trilacunar. Caprifoliaceae are generally tri- and sometimes pentalacunar. Cucurbit- aceae are all trilacunar. Rosales, which include Saxifragaceae (treated in the broad Englerian sense by Sinnott), are said to be mostly trilacunar although five gaps occur in Brunelliaceae and in a few Saxifragaceae, Rosaceae, and Leguminosae. Platanaceae exhibit seven Plant orders are remarkably constant with respect to their nodal con- ditions but Sinnott recognized that nodal anatomy is only one character, that nodal structure is not always invariable, and that further study will necessitate changes in his outline. In 1955, Marsden and Bailey presented their penetrating analysis of the node and interpretation of the primitive nodal condition. In contrast to Sinnott’s hypothesis that the trilacunar condition is basic and primitive, Marsden and Bailey provided evidence to indicate that the unilacunar, two-trace condition is ancestral and they indicated possible means for deriving both the unilacunar, single-trace condition and the trilacunar, triple-trace condition directly from it. Furthermore, they hypothesized that the unilacunar node could give rise to the trilacunar node through amplification, much as Sinnott derived the multilacunar form from the trilacunar. 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 65 Takhtajan’s (1964) scheme of nodal evolution is similar to that of Sinnott in that he accepted the primitiveness of the trilacunar node. However, the median lacuna has a double trace: ‘Thus, from all of these data one can conclude, it seems to me, that the node with three or more lacunae (Fic. 9) is the primary type of node in angiosperms. At present, it is impossible to determine more accurately the initial nodal type in angiosperms.’ ecause of the studies of Marsden and Bailey, it is apparent that a reassessment of the taxonomic value of nodal anatomy, as exemplified by Sinnott’s treatment, is very much in order. The derivation of the uni- lacunar, one-trace condition in Columelliaceae, rather than the condition itself, is the key to taxonomic understanding. This is also true of the largely characterized by trilacunar nodes, nor can we assign the relation- ship of Columelliaceae to those families with unilacunar nodes, if we agree with Marsden and Bailey that, ‘Structures which appear to be similar at the nodal level may not be truly homologous, and conversely differences which seem outstanding at the nodal level may acquire a dif- ferent significance where comprehensive developmental studies at succes- sive levels of the stem and leaf are made.’ Evidence from xylem anatomy * A brief recapitulation of the salient features in the xylem anatomy of Columelliaceae is in order here: perforation plates scalariform; pore dis- tribution exclusively solitary; intervascular pitting usually absent, except tending to alternate in regions of ligular overlap between superposed ves- sel elements; axial parenchyma vasicentric scanty; vascular rays exclu- sively uniseriate consisting solely of upright cells; spiral thickenings present in walls of vessel elements; and imperforate tracheary elements are tra- cheids. Gesneriaceae all have simple perforations in vessel elements. However, vestigial bars were noted in perforation plates of Solenophora calycosa Donn. Smith. In all woods examined, pores are solitary, in radial multiples, and in clusters except in Solenophora sp. (Yw 22822) where no clusters were observed. Intervascular pitting is exclusively alternate, i in Solenophora calycosa where transitional pitting was also seen. Axial par- enchyma distribution is various; however, it is paratracheal nase for um DC., and Solenophora calycosa, in which diffuse parenchyma occurs. In most species the vasicentric parenchyma is scanty; vasicentric parenchyma is abundant, however, in Columnea purpurata Hanst., Cyrtandra oenobar- ° Anatomical eat presented in this section are based on original observations in Gesneriaceae, Gro riaceae, Hydrangeaceae, and Escalloniaceae. Microscope slides examined were ae the Yale (Yw) and Say seapeon Sw) woed collections. Data pe g comparisons, onv families are considered as circumscribed in Metcalfe oi 1 Chalk (19 66 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 bata H. Mann, C. spathacea A. C. Smith, and Gesneria sp. (Yw 16832). In Drymonia sp. (Yw 17724), aliform and aliform-confluent parenchyma occurs. Vascular rays are absent in Besleria spp. (Yw 12217, 12225). In Columnea purpurata rays are 1 to 3 cells wide and in Solenophora calycosa, rays are mostly uni- and biseriate. In all other species investi- gated, rays are multiseriate. Rays are homocellular consisting solely of upright cells in Drymonia spectabilis, Columnea purpurata, Rhytidophyl- lum crenulatum, R. tomentosum (L.) Mart., and Rhytidophyllum sp. (Yw 20017). Heterocellular rays occur in Cyrtandra oenobarbata, Gs spathacea, Gesneria sp. (Yw 16832), Drymonia sp. (Yw 17724), Soleno- phora calycosa, and Solenophora sp. There are no spiral thickenings 1n vessels of Gesneriaceae. Imperforate tracheary elements are various: septate elements occur in Besleria spp., Gesneria sp., Drymonia spectabilis, Columnea purpurata, Rhytidophyllum crenulatum, R. tomentosum, Soleno- phora calycosa, and Solenophora sp. Only Cyrtandra did not show septate imperforate tracheary elements. Drymonia spectabilis exhibits only fiber- tracheids and Gesneria sp., Drymonia sp., Rhytidophyllum crenulatum, and Solenophora sp. show only libriform wood fibers. All other species investigated show both fiber-tracheids and libriform wood fibers. Except for the common occurrence of vasicentric scanty axial paren- chyma in Gesneriaceae and Columelliaceae, the wood anatomy of these two families is very different. Perforation plates in Columelliaceae are scalariform; in Gesneriaceae they are simple. Pore distribution is strictly solitary in Columelliaceae; in Gesneriaceae it is solitary and in radial multiples and clusters in most of the species studied. Intervascular pitting is virtually absent in Columelliaceae because of the independent distribu- tion of vessels; in Gesneriaceae it is alternate. All species of Columelli- aceae have vascular rays; in Gesneriaceae, Besleria lacks vascular rays. Vascular rays are uniseriate in Columelliaceae; in Gesneriaceae all species have vascular rays more than one cell wide. Vascular rays contain only upright cells in Columelliaceae; in Gesneriaceae species may show both heterocellular rays and homocellular rays with upright cells. Spiral thick- enings are present in the vessels of Columelliaceae; in Gesneriaceae, veS- sels lack spiral thickenings. In Columelliaceae all imperforate tracheary elements are tracheids; in Gesneriaceae both fiber-tracheids and libriform wood fibers occur, but no tracheids. Grossulariaceae have scalariform perforations in vessel elements, but some simple perforations were also observed. Pores are solitary, in radial multiples, and in clusters. Growth rings are pronounced and the wood is ring porous. Intervascular pitting is transitional and scalariform. Axial parenchyma is absent. Vascular rays are multiseriate, broad, and hetero- cellular. Sheath cells are of common occurrence in the rays. Spiral thick- enings are absent from vessel walls. Imperforate elements are septate tracheids and in Ribes viscosissimum Pursh, fiber-tracheids were also recorded. The presence of scalariform perforations and tracheids seems to pro- 1969] STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 67 vide the only common anatomical features between Grossulariaceae and Columelliaceae. Solitary, radial multiple, and clustered pores; ring poros- ity; broad, heterocellular vascular rays; septate imperforate tracheary elements; and the absence of axial parenchyma in the wood of Grossu- lariaceae are rather distinct anatomical characteristics which differ from Columelliaceae Perforation plates in vessel elements of Hydrangeaceae are scalari- form.® Pores are exclusively solitary in Broussaisia arguta Gaudich., B. pellucida Gaudich., Fendlera rupicola A. Gray, and Philadelphus sp. (Yw 11845). In Deutzia vilmorinae Lemoine & D. Bois, Hydrangea pana- mensis Standley, and Philadelphus coronarius L., pores are solitary and in radial multiples. Hydrangea bretschneideri Dipp. and Dichroa febri- fuga show pores in solitary, radial multiple, and clustered dispositions. Intervascular pitting is generally absent in Broussaisia arguta, B. pel- lucida, and Fendlera rupicola. However, in the overlapping vessel ligules of Broussaisia arguta, scalariform pitting was seen, whereas in this posi- tion Fendlera rupicola shows a tendency to alternate intervascular pitting. In Philadelphus coronarius, intervascular pitting is transitional; in Phila- delphus sp., pitting is alternate with some opposite. Pitting in vessel walls of Deutzia vilmorinae, Hydrangea bretschneideri, and H. panamensis is scalariform. Vessel walls in Dichroa febrifuga show both transitional These occur in conjunction with other heterocellular rays, two or more cells wide. Rays up to 8-cells wide occur in Broussaisia pellucida. Deutzia vilmorinae, Fendlera rupicola, and Hydrangea bretschneideri have only uni- and biseriate rays. Sheath cells are common in species with wide rays. In Deutzia vilmorinae, scalariformly perforated ray cells occur. Tenuous spiral thickenings occur in the cell walls of vessels and tracheids of Fendlera rupicola; in Philadelphus coronarius and Philadelphus sp., spirals occur in tracheids. Imperforate tracheary elements in Fendlera rupicola, Hydrangea bretschneideri, Philadelphus coronarius, and Phila- delphus sp., are exclusively tracheids. In Broussaisia arguta, B. pellucida, and Dichroa febrifuga, both tracheids and fiber-tracheids appear. Deutzia vilmorinae and Hydrangea panamensis show only fiber-tracheids. Imper- forate tracheary elements are septate in Hydrangea panamensis and Dichroa febrifuga. There are several similarities between the woods of some species of Hydrangeaceae and Columelliaceae: scalariform perforation plates, ex- clusively solitary pores and concomitant absence of intervascular pitting, a tendency to alternate intervascular pitting, and tracheids. Axial xylem parenchyma is vasicentric scanty in Columelliaceae with some diffuse; in ® Metcalfe and Chalk (1950) report simple perforation plates in Deutzia glabrata Kom. and in some species of Philadelphus. 68 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 all Hydrangeaceae studied, where axial xylem parenchyma was present, it it vasicentric scanty and some strands were diffusely arranged. On the other hand, there are also pronounced anatomical differences between these families: Pores are exclusively solitary in Columelliaceae; in several species of Hydrangeaceae pores are in both solitary and other arrange- ments. Intervascular pitting tends toward alternate in Columelliaceae; in Hydrangeaceae scalariform and transitional intervascular pitting occur in several species. All species of Columelliaceae show axial parenchyma; several species of Hydrangeaceae lack this tissue. In Columelliaceae, vas- cular rays are all uniseriate and homocellular; all Hydrangeaceae have uniseriate rays plus rays which are two or more cells wide and hetero- cellular. Imperforate tracheary elements in Columelliaceae are tracheids; some species of Hydrangeaceae show both tracheids and fiber-tracheids, while other species have only fiber-tracheids. Perforation plates in Escalloniaceae are exclusively scalariform except in Brexia, where plates are mostly simple, and in Kania Schlechter,“ where they are exclusively simple. All Escalloniaceae have solitary pores, 10 Escallonia floribunda H.B.K., E. fonkii Phil., and E. myrtilloides Lif, pores are exclusively solitary. In E. pulverulenta (Ruiz & Pavon) Pers., E. revoluta (Ruiz & Pavoén) Pers., E. rubra (Ruiz & Pavon) Pers., and E. tortuosa H.B.K., pores are also in radial multiples. Pores are solitary and in radial multiples in Brexia madagascariensis Thou. ex Ker-Gawl., Itea sp. (Yw 20142), Ouintinia acutifolia T. Kirk, Q. serrata A. Cunn., and Q. sieberi A. DC. In Quintinia, however, multiples are rare but tangentially oriented groups of pores are conspicuous. Solitary, radial multiple, and clustered dispositions are seen in Anopterus glandulosus Labill., Argo- phyllum ellipticum Labill., and in all Polyosma species studied. Inter- vascular pitting is sparse in Quintinia acutifolia and Q. serrata; pitting 1n Q. sieberi is alternate with a tendency to opposite. In those species of Escallonia with exclusively solitary pore distribution, the widely overlap- ping vessel ligules provide areas of intervascular communication showing alternate intervascular pitting. Species of Escallonia with radial pore multiples show alternate intervascular pitting. Alternate intervascular pitting also occurs in Anopterus macleayanus F. Muell., Argophyllum el- lipticum, Brexia madagascariensis, Itea sp., and in all Polyosma species studied except P. integrifolia Blume and P. serrulata Blume which have exclusively opposite pitting. In addition to alternate intervascular pit- ting, Anopterus macleayanus and Escallonia floribunda show transitional pitting. Anopterus glandulosus only has transitional intervascular pitting. In addition to alternate pitting, /tea sp. shows transitional and scalariform pitting. All species of Polyosma with alternate intervascular pitting also show opposite pitting. Escalloniaceae are characterized by apotracheal axial parenchyma and all species studied show either a diffuse and/or *Erdtman and Metcalfe (1963) have assigned this genus to Myrtaceae on anatomi- cal and palynological grounds. Their evidence is so strong, that Kania will not be considered further in this discussion. 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 69 diffuse-in-aggregates pattern. In Escallonia revoluta, E. rubra, all species of Polyosma, and Quintinia sieberi, both diffuse and diffuse-in-aggregates patterns occur. In Anopterus glandulosus, Escallonia myrtilloides, E. tor- tuosa, Itea sp., Quintinia acutifolia, and Q. serrata, only diffuse axial parenchyma was observed. Parenchyma in Brexia madagascariensis con- sists of multiseriate bands. Short uniseriate bands occur in Escallonia floribunda, in addition to the diffuse-in-aggregates pattern. Axial paren- chyma is absent in Argophyllum ellipticum. All species of Escalloniaceae have some uniseriate rays, although none was observed in Anopterus glandulosus where rays are exclusively multiseriate. All species have some heterocellular rays except for Brexia madagascariensis. The following species has uni- and biseriate rays only: Anopterus macleayanus, Brexia madagascariensis, Escallonia myrtilloides, and E. tortuosa. All other spe- cies studied have both uniseriate rays and rays which are two or more cells wide. Vascular rays are exclusively heterocellular in Anopterus glan- dulosus, A. macleayanus, Argophyllum ellipticum, and Escallonia flori- bunda. Rays in Brexia madagascariensis are homocellular and cells are upright. In the following species, multiseriate and biseriate rays are heterocellular and uniseriate rays are homocellular containing only upright cells: Escallonia fonkii, E. myrtilloides, E. pulverulenta, E. revoluta, E. rubra, E. tortuosa, Itea sp., and all species of Polyosma and Quintinia. Species with wide multiseriate rays commonly exhibit sheath cells. Spiral thickenings occur in walls of vessels in Escallonia floribunda, E. myrtil- loides, E. rubra, and E. tortuosa. In E. myrtilloides and E. tortuosa, spiral thickenings also occur in tracheid walls. Only tracheids occur in Anopterus glandulosus, Escallonia floribunda, E. myrtilloides, E. revoluta, E. rubra, E. tortuosa, Polyosma cunninghamii Benn., and Quintinia. Both tracheids and fiber- tracheids occur in Anopterus macleayanus, Escallonia pulver- ulenta, and Itea sp. Argophyllum ellipticum, Brexia madagascariensis, Escallonia fonkii, Polyosma cambodiana Gagn. (?), P. ilicifolia Blume, P. integrifolia, P. mutabilis Blume, and P. serrulata, exhibit only fiber-tra- cheids. Septate fiber-tracheids appear in Argophyllum ellipticum. The xylem anatomical similarities between species of Escalloniaceae and Columelliaceae are striking: exclusively scalariform perforation plates (except in Brexia), exclusively solitary pore distribution (in some species of Escallonia and in Polyosma cunninghamii), spiral thickenings in vessels (in some species of Escallonia), and exclusively tracheids (in Anopterus glandulosus, in some species of Escallonia, Polyosma cunninghamii, and Quintinia). The only major anatomical differences between these two families are the presence of vascular rays which are two or more cells wide and exclusively apotracheal axial parenchyma in all Escalloniaceae. Among the species studied, the xylem anatomy of Escallonia myrtilloides can hardly be distinguished from that of Columelliaceae, except for the biseriate condition of some of the rays and exclusively diffuse axial paren- chyma in the former (cf. Fics. 22 and 23, 24 and 25, 26 and 27). Among the remaining families which have been suggested as close rela- tives of Columelliaceae — Ebenaceae, Styracaceae, Gentianaceae, Logan- 70 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 iaceae, Caprifoliaceae, Rubiaceae, Onagraceae, Oleaceae, Vacciniaceae, and Scrophulariaceae — xylem anatomy provides serious bases for compar!- son only with Styracaceae, Caprifoliaceae, and Vacciniaceae. Styracaceae typically show scalariform perforation plates and uniseriate, homocellular vascular rays in some species; some Caprifoliaceae have scalariform and simple perforation plates, spiral thickenings in vessels, and imperforate tracheary elements with distinctly bordered pits; and most Vacciniaceae have scalariform or scalariform and simple perforations and imperforate tracheary elements with distinctly bordered pits. Ebenaceae, Loganiaceae, Rubiaceae, Onagraceae, Oleaceae, and Scrophulariaceae, are characterized by simple perforations. In addition, Gentianaceae-Gentianoideae univer- sally possess internal phloem and medullary vascular bundles; Logania- ceae-Loganioideae are characterized by included phloem; and Onagraceae have internal phloem in the axis and a few genera show included phloem. These dispositions of phloem are very specialized and are ordinarily in- dicative of close relationship within specific taxa, sometimes on an ordinal basis (e.g., internal phloem in families of Myrtales). CONCLUSION In reviewing the foregoing presentations of evidence and discussions, it is clearly impossible to assemble an array of data from each form of evl- dence presented — gross morphology, floral anatomy, foliar anatomy, nO- dal anatomy, and xylem anatomy — which would affirm unequivocally the relationships of Columelliaceae with any one of the several families to which it has been allied. The similarities in gross morphology of flow- ers and fruits among many families of various alliances probably indi- cates parallel evolution rather than close genetic relationship. The evo- lutionary development which has culminated in Columellia has proceeded in such a manner that the complex of its characteristics is different from any known taxon today. What baffles us now baffled our predecessors and it is time to admit once and for all that Columelliaceae is a unique plant family, probably with no really close living relatives. The clearest line of evidence for the possible relationships of Columelliaceae is pro- vided by xylem anatomy and it appears not too far from reality to assert that this family belongs in the great saxifragaceous assemblage with the Escalloniaceae, Hydrangeaceae, and Grossulariaceae. Data from gross morphology, floral anatomy, palynology, etc., at least do not contradict this probability. Perhaps its nearest relatives are in the Escalloniaceae. If there was a common saxifragaceous ancestor, phylogenetic departure must have occurred long ago, for transitional forms seem to have been lost in the development of the modern plants of which this taxon is com- posed. Evidence from xylem anatomy seems equally persuasive in negat- ing an alliance with any other family or group of families. Unfortunately, data from cytotaxonomy, embryology, and biochemistry, which might be helpful in resolving our somewhat equivocal stand, are not available for Columellia. 1969] STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 71 ACKNOWLEDGMENTS This study has been carried out under the sponsorship of the Yale School of Forestry, the Smithsonian Institution, the Arnold Arboretum of Harvard University, and the University of Maryland. We are grateful to administrators of these institutions for the privilege of using their fa- cilities. For their encouragement, suggestions, and critical advice, we wish to acknowledge warmly Dr. Sherwin Carlquist, : rt F. Thorne, Dr. Arthur Cronquist, and Dr. Hugh IItis. The peingiinarb of various herbaria have been cooperative in allowing us to see specim in situ, to borrow specimens for study, and to use bits of stems, ives and flowers for microscopic observations: Royal Botanic Gardens, Kew; Yale School of Forestry, New Haven; Herbario San Marcos, Museo de Historia Natural, Lima; Arnold Arboretum and Gray Herbarium, Har- vard University, Cambridge; Field Museum of Natural History, Chicago; U.S. National Herbarium, Smithsonian Institution, Washington; and New York Botanical Garden, Bronx. Microscope slides and woods for sectioning of Columellia and other taxa were made available from the Record Memorial Collection of the Yale School of Forestry and from the collections of the Division of Plant Anatomy, Department of Botany, Smithsonian Institution. We appreciate the generous cooperation of botanists in these institutions. Dr. Oscar Tovar of the Herbario San Mar- cos, Lima, provided the only fluid-preserved anatomical specimens of Columellia available to us and we are especially thankful for his efforts in our behalf. Dr. John J. Wurdack, Smithsonian Institution, kept our needs in mind during a collecting trip to Peru and provided us with a fine wood sample of C. oblonga ssp. oblonga. Mr. James Teeri, of the Univer- sity of New Hampshire, and Miss Carolyn Bensel, who is working on the floral anatomy of the Saxifragaceae, sensu lato, in Dr. Barbara Palser’s laboratory at Rutgers University, were most kind to share their observa- tions with us in advance of publication. We also extend our thanks to Mr. Austin Griffiths, Jr., of the Los Angeles State and County Arboretum for preserved flowers of Escallonia; to Miss Brenda Gee, of the University of Auckland, for preserved flowers of Carpodetus; and Dr. Judy Morgan for help with the sectioning and examination of floral material. LITERATURE CITED AGABABYAN, V. SH. 1964. Evolutsiya pyl’tsy v poryadkakh Cunoniales i Saxi- fragales v svyazi s nekotorymi voprosami ikh sistematiki: filogenii. Izv. d. Nauk Armyanskoi SSR, Biol. Nauki 17(1): 59-72; tab. I-11. AGARDH, J G. 1858. emer systematis plantarum. xcvi + 404 pp. pls. 28. C. W. K. Gleerup. Lun ARNOTT, H. J. 1959. Leaf poate Turtox News 37: 192-1 BAILLON, H. 1888. Gesnériacées. Histoire des plantes. 10: ere Librairie Hachette. Paris. BARTLING, F. T. 1830. Ordines naturales plantarum. Dieterichianus. Gottingae. 72 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 BENTHAM, G., & J. D age 1876. Columelliaceae. Genera plantarum. 2: 989. L. Reeve & Co. Lon BHATTACHARJYA, S. S. 1954. - Beitrag zur Morphologie des Androceums von Benincasa hispida (Thunb.) Cogn. Ber. Deutsch. Bot. Ges. 67: 22-25. Brizicky, G. K. 1961. A synopsis of the genus Columellia (Columelliaceae). Jour. Arnold Arb. 42: 363-372. CANDOLLE, A. P. pe. 1839. Columelliaceae. Prodromus systematis naturalis regni vegetabilis. 7: 549. Truettel & Wiirtz. Paris. Cuaxkravarty, H. L. 1958. Morphology of the staminate flowers in the Cucur- bitaceae with special reference to the evolution of the stamen. Lloydia 21: 49-87. CxarkE, B. 1858. On the anthers of Columelliaceae and Cucurbitaceae. Ann. Mag. Nat. Hist. London. III. 1: 109-113; pl. VJ. CoMMITTEE ON NOMENCLATURE, INTERNATIONAL ASSOCIATION OF Woop ANATO- MISTS. 1957. oe glossary of terms used in wood anatomy. Trop. Woods 107: 1-3 Cronguist, A. See The evolution and classification of flowering plants. x + 396 pp. Houghton Mifflin. Boston. Don, D. 1828. Descriptions of Columellia, Tovaria, and Francoa; with remarks on their affinities. Edinburgh New Philos. Jour. 1828-1829: 46-53. Don, G. 1838. Columellieae. A general nage of the dichlamydeous plants. 4: 57, 58. J. G. & F. Rivington. Lon ENDLICHER, S. 1839. Columelliaceae. aie plantarum. 1839: 745. Fr. Beck. Vindo bonae. . 1841. Columelliaceae. Enchiridion botanicum. 366 pp. Guil. Engelmann. Lipsiae-Viennae ENGLER, A. 1892. ‘Syllabus der Vorlesungen iiber specielle und medicinisch- pharmaceutische Botanik. xxiii + 184 pp. Gebriider Homtiaeeee: Berlin. 1928. Saxifragaceae. Nat. Pflanzenfam. ed. 2 ErpTMAN, G. 1952. Pollen morphology and plant taeonomy. Angiosperms. xii + 539 pp.; frontis, Almquist & Wicksell. Stockholm. C. R. Metcalfe. 1963. Affinities of certain genera incertae sedis sug- gested by pollen morphology and vegetative anatomy. Kew Bull. 17: 249- 256; pl. 2. Esau, K. 1965. Plant anatomy. ed. 2. xx + 767 pp. John Wiley & Sons. New York. Eype, R.H. 1966. Systematic anatomy of the flower and fruit of Corokia. Am. Jour. Bot. 53: 833-847. Faun, A. 1967. Plant anatomy — by SysiL Bromo-ALTMAN). vii + 534 pp. Pergamon Press. ork. FELDHOFEN, E. 1933. Hee zur physiologischen Anatomie der nuptialen Nektarien aus den Reihen der Dikotylen. Beih. Bot. Centralbl. 50: 459-634; Taf. HI-XX XI. FritscH, K. 1893, 1894. Gesneriaceae. Nat. Pflanzenfam. IV. 3b: 133-185 (133-144, 1893; 145-185, 1894). ———. 1894. Columelliaceae. Nat. Pflanzenfam. IV. 3b: 186-188. Getius, L. 1967. Studien zur Entwicklungsgeschichte an Bliiten der Saxifragales ssa lato mit besonderer Beriicksichtigung des Androeceums. Bot. Jahrb. 8 —303. GriseBacH, A. H. R. 1839. Genera et ~— gentianearum. viii + 364 pp. J. G. Cottae. Saran et Tubinga 1969 | STERN, BRIZICKY, & EYDE, COLUMELLIACEAE 73 Hauer, H. 1901. Uber die Ne der Tubifloren und nalen. Abh. Naturw. Ver. Hamburg 16(2): 1 9 "1908. Ueber die Abgrenzung und Vervandtscat pe einzelnen Sippen bei den Scrophularineen. Bull. Herb. Boiss. II. 3: 1908. Uber Juliania, eine Terebinthaceen- ee, ae ‘Cupula, und die wahren Stammeltern der Katzchenbliitler. 210 pp. C. Heinrich. Dresden. . 1910. Ueber Phanerogamen ie unsicherer oder unrichtiger Stellung. Meded. Rijks Herb. Leiden 1: 1-4 Herzoc, T. 1915. Die von Dr. Th. piace auf seiner zweiten Reise durch Bolivien in den Jahren 1910 und 1911 gesammelten Pflanzen. II Teil. Meded. Rijks Herb. Leiden 27: 1-90; Hooker, J. D. 1873. “Editor’s note.” ” In: E. LE Maout & J. apse A general system of botany. (Transl. by Mrs. Hooker; edited . Hooker.) xii + 1066 pp. Longmans, Green, and Co. London. Editor's note, 594 ——.. “

m - r=3 = oa m oO ~< a = o ” @ | 4 a = «<¢ 7 Z. ! oe | 9 (2 q 4 } | Ss i Zl y | T vl T SAYND ALISNALNI JWIL GYVONVLS 1V907 ny | SAYND ADNINOSYS Van ns 2 bf —* = Nn MM/HOUR Fic A representation of the diurnal variation of intensity and frequency of Bont occurrences. night as in the day, whereas daytime rains were almost twice as heavy as nighttime rains. Presumably daytime rains are predominantly convec- tive, while nighttime rains are orographic. 84 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The effect of the forest on rainfall rate is illustrated in Ficure 4 where the 1-minute accumulations of rainfall above and below the forest are compared for a typical shower. During an 11-minute interval 0.54 inches of rain fell above the forest. Within a minute the rain began below the forest, continued for 12 minutes and totalled 0.38 inches. The forest acts like a filter, delaying the onset of rain at ground level, lessening the peaks and smoothing the variations of intensity, and prolonging the rain slightly as it drips from the leaves. T T T T T T T > lOr ~ x————x ABOVE se = o——-—-o BELOW E BE 4 p= 4= = rs) 6b aa = wn tec i ee <3 / 2 | a ao) 4r / S | < j Uy a = / = *e 7 S x ‘O--O. / N ea i i i N 2 4 6 8 10 2 14 TIME IN MINUTES . A representation of the rainfall rate above and below the forest versws ‘eo near noon, 20 July 1966. The rain above that is not caught by the lower rain gage has either been intercepted by the trees or has reached the ground by means of the trunks. Hydrologists define interception as that portion of precipitation that never reaches the ground either as rain or trunk flow. This relation- ship is expressed in the simple continuity equation: Rain Below = Rain Above — Trunk Flow — Interception (1) Wisler and Brater (1959) point out that after the initial wetting of the leaves, branches, and trunks of trees, the interception rate becomes equal to the evaporation rate from those surfaces. Since Pico del Oeste is usu- ally shrouded in fog it follows that the interception rate is usually zero and that trunk flow must account for almost all of the difference between rain above and below the forest. This result contrasts with the rather large 1969 | BAYNTON, ELFIN FOREST, 3 85 interception and small trunk flow reported by Clegg (1963) in much taller stands at lower elevations in the Luquillo Mountains. Rainfall below the forest as a function of rainfall above was analyzed by standard regression techniques. One hundred and twenty-four rains were selected for the analysis. Each had the property that it was preceded by a 6-hour drying period as shown by the hygrograph trace. Thus part of each rain was used in wetting the foliage. The balance either ran down the trunks or dripped through. Using the notation X = rain above the forest in inches, Y = rain below the forest in inches, the analysis yielded: Y = 0.768X — 0.034 with a correlation coefficient of 0.99 between X and Y. A correction was dictated by slight differences in the volume of water required to tip the buckets in the two gages. The design value is 18.5 ml. Calibration of the two gages in place gave 18.5 for the upper gage and 18.9 for the lower gage. Appropriate adjustment leads to the final result: Y = 0.786X — 0.035 On rewriting the equation in the form: Y = X — 0.214X — 0.035 and comparing it to equation (1) we can identify trunk flow as 21.4 per- cent of the rain above and interception as 0.035 inches. Clegg cites other investigators as setting trunk flow no higher than 10 percent of the total rainfall. The explanation for large trunk flow on Pico del Oeste is un- doubtedly found in the unusually high number of stems, a feature of this forest that is well illustrated in Ficure 5. The interception figure of 0.035 inches implies that the vegetation over each square meter of ground is able to store 886 ml/m?. Studies of the U.S. Forest Service reported by Wisler and Brater indicate storage capacities of about 0.14 inch for hardwoods and 0.23 inch for pines in North Carolina. Storage in the cloud forest of Pico del Oeste would be expected to be very much less since it is only 10 to 12 feet high. It should be noted that, for rains occurring when the forest is already thoroughly wet, the relationship between rain below and above simplifies to Y = 0.786X Cloud Water. The difficulty with all cloud water studies is to relate the observations of a collecting device to the amount of water that the foliage itself extracts from the cloud. Although the same difficulty besets the interpretation of the data collected on Pico del Oeste, different lines of argument support the conclusion that cloud-water is of secondary im- portance in this region of abundant rain. In the first place, cloud water is not a means of sustaining the forest during drouth since cloud-free periods coincide with rainless periods. Secondly, four distinct analyses, each of which by itself is imprecise, give very similar results. First ANALYsIs. The cloud-water sampler was in service for 258 days during the year from June 1966 to May 1967. By extrapolating the data 86 JOURNAL OF THE ARNOLD ARBORETUM a me wh Pot oy ex * RY ‘ Pt % : . - [voL. 50 at number of stems in its composition. A view of the elfin forest illustrating the gre 1969 | BAYNTON, ELFIN FOREST, 3 87 to a full year the annual total of cloud water is estimated at 325 liters/ square meters (1/m*). Since 1 mm. of rain is the same as 1 1/m?, the an- nual rainfall total of 453 cm. may be expressed as 4530 1/m?. Although the unit cross section is in a vertical plane for cloud water and in a horizontal plane for rain, no adjustment is needed for trees such as those and vertical planes. Moreover the wind speed through the thermometer shelter housing the cloud-water collector was found to be nearly the same as the wind in the forest halfway to the top, namely 17.6 and 16 percent, respectively, of the 20-foot wind. Deferring for the moment any discus- sion of differences in the collection efficiency of the aluminum shadescreen and the foliage, and differences in the sampling period, the data imply that cloud water is only 7.2 percent of rain water. SECOND ANALYsIs. Another approach was based on a 1,000-hour period from 18 July to 29 August 1966. Since cloud water intercepted by the trees contributes to rain measured below the forest, the record of the 20-pen event-recorder was examined for occurrences of rain below the forest without rain above. The event of interest, “rain below without rain above,”” was defined as no rain above during the three hours including the hours preceding and following an occurrence of rain below. Any run of three hours without rain above the forest is a possible occurrence of the event of interest. There were 324 possible occurrences and only 15 observed occurrences. During the 1,000 hours the total rainfall below the forest was 15.83 inches. Of that total 0.15 inch occurred without rain above and must therefore be attributed to cloud water. Additional cloud water is also collected during rain but the exact amount cannot be deter- mined. Again the implication is that cloud water is only a small fraction of rain water. Tuirp ANALysIs. The 15 cases were then examined in detail in an attempt to relate the observed cloud water collection to the observed rainfall below the forest. The approach was to count the number of tips of the cloud-water collector during the time for 0.01 inch of rain to accumulate in the below-canopy rain gage. The data are summarized in TABLE 1. The entry for August 19 is suspect because of the long accumu- lation time indicating that the foliage must have dried out and had to be rewetted before the process of foliage drip could resume. The same might be true for August 16. The analysis is imprecise because in many cases some rain fell during the accumulation time. Omitting August 19 the mean is 5.4 units of cloud water per 0.01 inch of rain below the forest. Earlier it was shown that 78.6 percent of rainfall drips through when the foliage is wet and the same will be true for cloud water. Thus the collection of; < 0.01 inch! or 259 ml/m? of water below the canopy 18.9 i. : is the correction factor that accounts for 18.9 ml. being the actual volume (18. of water to tip the bucket rather than the design volume of 18.5 ml. 88 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 259 0.786 of cloud water by the foliage. During the same period this analysis shows that 5.4 (50) = 270 ml/m? of water were collected by the cloud-water detector. Therefore, the foliage is 1.2 (i.e. 330/270) times as efficient as the cloud-water collector. during a rainless period results from the collection of = 330 ml/m? TABLE 1. For fifteen occurrences of 0.01 inch rain below the forest without rain above, the amount of cloud water collected during the accumulation time, the accumulation time. Amount cloud water Accumulation time for Date during accumulation time 0.01 inch rain below July 29 8.1 units * 4 hours 11 min 30 9.6 3 37 Aug. 2 6.6 4 50 2 5.4 4 30 5 5.0 3 12 5 13 1 40 7 1.4 3 12 9 ye: 3 43 11 2.5 ! 9 13 42 5 14 13 4.0 4 54 14 55 4 7 16 10.0 11 8 19 16.5 24 39 22 5.0 3 10 * One unit, or bucket tip, equals 50 ml/m”. Fourtu ANAtysis. The main shortcoming of the third analysis was the truncation error associated with the collection of rain and cloud water in discrete steps of a bucket. It appeared possible to sharpen the analysis by replacing the tipping buckets with bottles and measuring exact vol- umes of water collected by the two rain gages and the cloud-water col- lector under personally observed weather conditions. The fourth analysis summarizes the results of this approach carried out in December of 1967. Five separate attempts were made to collect rain and cloud water under known boundary conditions. When the data were analysed, errors in eXx- perimental technique became evident. Generally there was doubt about the boundary conditions. No interpretation of the first attempt was pos- sible because it became apparent that the foliage was neither fully wet nor fully dry. Another error in technique may be illustrated by the anal- ysis of the data collected on 9 December. The collecting of water began at 9:55 a.m., immediately after a mod- 1969] BAYNTON, ELFIN FOREST, 3 89 erate shower, and continued until 4 p.m. The error in technique was that the sampling began so soon after the shower that its effect was still being felt as drip from the foliage. The collected water was equivalent to 3227 ml/m? of rain above the canopy, 2746 ml/m? below the canopy, and only 70 ml/m? of cloud water. Because the foliage was always fully wet the relationship Y = 0.786X should apply where Y = 2746 ml/m*. But because of the faulty tech- nique, we have = Rain Above + (Cloud Water) E + A\R, with E being the ratio between the collecting efficiency of the vegetation and the cloud-water collector, and /\R being the unknown amount of rain above the canopy immediately before 9:55 that is in the process of getting to the ground via trunks and drip-through. Substituting both Y and X in the equation gives: 2746 = 0.786 (3227 + 70E + AR) whence: E = 3.82 — 0.014 AR We conclude therefore that E is less than 3.82 since /A\R is not zero, but that is all we can sa Two of the attempts were for periods that began with dry foliage, i.e. no liquid water attached to plant surfaces. The collections in the three gages should therefore be related by: Y= 0.786X — 886 where the units of X and Y are ml/m? so that 886 replaces 0.035 in the original equation because each hundredth of an inch of rain = 254 ml/m*. Between 4:35 p.m. December 7 and 1 p.m. December 8 the amounts collected were 3570 ml/m? of rain above, 2004 ml/m? below, and 16 ml/m* of cloud water. With so little cloud water the equation is too sensitive to slight errors in the collected amounts above and below to per- mit estimates of E, the relative collecting efficiency of the vegetation. We can, however, get independent estimates of the Y-intercept, 886, by sub- stituting E = 1.2, the value obtained earlier or E < 3.82,? the upper limit. These choices of E yield a Y-intercept = 817 and <850. Conditions were also dry on the peak at 11:45 a.m. December 3 when collections were begun. By 3:45 p.m. December 4 the amounts were 1936 ml/m? of rain above, 906 ml/m? below, and only 22 ml/m? of cloud water. Setting E = 1.2 and <3.82 gave values of 637 and <682 for the Y-intercept. The average of these two trials was 727 for E = 1.2 and <766 for E <3.82, providing fair confirmation of the value, 886, obtained from the regression equation. One attempt combined enough cloud water with light rain to permit an estimate of E, the relative collecting efficiency of the forest. Between ? < is the symbol for “is less than.” 90 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 4:30 p.m. on December 2 and 11:20 a.m. on December 3 the observed collections were 2128 ml/m? of rain above, 310 ml/m? of cloud water, and 1151 ml/m? of rain below. On December 2 the peak had been clear for four hours during the day but had fogged in shortly before 4:30 p.m. FREQUENCY IN PERCENT nm w i.e) oe) D WIND SPEED IN MI/ HOUR oO Ol T 3JWiL GYVGNVLS 1V901 b2 ! Lee | NIVY JO AONINOSYS J 8 Q33dS QNIM Fic. frequency of rain. 6. A graphic comparison of the diurnal variations of wind speed and 1969 | BAYNTON, ELFIN FOREST, 3 91 Presumably the forest was substantially, but not fully, dry. Substitution in the regression equation gives: 1151 = 0.786 (2128 + 310 E) — 886, from which E = 1.5 or slightly less since the constant, 886, is for fully dry foliage. While not providing the hoped for “acid test,”’ the fourth analysis con- firms that the foliage is only slightly more efficient than the cloud-water collector and that its storage capacity is substantially less than that re- ported for other forests. Correcting the first analysis for the greater collecting efficiency of the foliage, we have annual cloud water of 1.2 (325) = 390 1/m?, which is 8.6 percent of the annual rainfall. Although this amount is relatively unimportant to the water budget of Pico del Oeste it is equal to the nor- mal annual precipitation for Denver, Colorado. Wind Speed. The diurnal variation of wind speed above the forest is shown in the upper half of Ficure 6. The data were for a month with little daytime clearing, August, and a month with considerable daytime clearing, October. Both months showed the same pattern and were therefore com- bined. For comparison the diurnal variation of rainfall frequency is in- cluded in Figure 6. The night maximum and day minimum show the influ- ence of convection. During the daytime there is a downward flux of mo- VERTICAL PROFILE OF WIND SPEED or T T 1 T T q T Ly T iB u pb: 16E : Z al | os : 2 ak TREE TOPS ve é . ERQOL BOOAA é Br Pe 7 . sar sy, tal / i / 2+ ; ral I l l i l | 1 i l l a 10 20 30 00 WIND SPEED AS A PERCENT OF 20-FT. WIND Fic. 7. A vertical profile of wind speed. 92 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 mentum below the mountain top and a consequent decrease in wind speed. The same anomalous cycle has been reported on towers several hundred feet above flat ground. The almost identical cycle of rainfall frequency supports the interpre- tation that the frequent nighttime rains are mainly orographic, and that daytime rains are mainly convective. The vertical profile of wind speed above and below the tree tops was investigated by installing a sensitive Casella anemometer at various heights on the tower. Wind speed averaged over an hour was expressed as a percent of the 20-foot wind. The results are presented in FIGURE 7. neem points below the tree tops might modify that portion of the SUMMARY Rainfall on Pico del Oeste, although twice as frequent by night, is only half as intense as during the day. Rain is mainly orographic by night and convective by da Trunk flow accounts for 21 percent of the rainfall. The canopy has a storage capacity equal to a depth of 0.035 inches or 886 ml/m?. : On the average, water extracted from the clouds by the foliage 1s slightly less than 10 per cent of rainfall. Winds are strongest at night and weakest during the afternoon. REFERENCES ALAKA, M. A. Problems of tropical meteorology (A survey). Tech. Note no. 62, p. 20. World Meteorological Organization, 1964. Baynton, H. W. e ecology of an elfin forest in Puerto Rico, 2. The micro- climate of See del Oeste. Jour. Arnold Arb. 49: 419-430. 1968. Ctiece, A. G. infall interception in a tropical forest. Carib. Forest. 24(2): 75- 79. 1963. é EKERN, P. C. Direct ger capes ee ane water on Lanaihale, Hawaii. Soil Sci. Soc. Am. Proc. 28(3): 4 1964. Howarp, R. A. The ecology of an ger forest in Puerto Rico, 1. Introduction and composition studies. Jour. Arnold Arb. 49: 381-418. ; Kraus, E. B. The diurnal precipitation change over the sea. Jour. Atmos. Sci. 20: 551-556. 1963. Wister, C. O., & E. F. Brater. Hydrology, 2nd ed. p. 195. John Wiley & Sons, New York, 1959. NATIONAL CENTER FOR ATMOSPHERIC RESEARCH BouLpEer, CoLorApo 80302 1969} GATES, ELFIN FOREST, 4 93 THE ECOLOGY OF AN ELFIN FOREST IN PUERTO RICO, 4. TRANSPIRATION RATES AND TEMPERATURES OF LEAVES IN COOL HUMID ENVIRONMENT ? Davip M. GATES THE PURPOSE OF THE STUDIES reported here is to contribute some under- standing of the adaptation, growth, and behavior of plants in the mist forest at the top of Pico del Oeste, Luquillo Mountains, Puerto Rico. The primary influence of climate on a plant is through the transfer of energy. All physiological processes consume energy. Biochemical re- actions are temperature dependent and some are light dependent. The vitality of a plant depends on its temperature and its energy content. If a plant is too warm, its vital processes slow down; and above certain temperatures many physiological processes stop and denaturation of pro- teins occurs. If a plant is too cool, its vital processes slow down. The plant will not survive below certain temperatures. Most plants grow best at an optimum temperature. The energy content of a plant determines its temperature. Several factors affect the energy exchanged between a plant and its surroundings. The significant environmental factors are radiation, air temperature, wind and humidity. In order for these factors to be translated into their effect on the plant, they must be expressed as energy flow. The incident radia- tion is a specific amount of energy. The air temperature and wind speed are translated into energy flow by the concept of convection. The humid- ity of the air affects the energy exchange for a leaf by means of the transpirational cooling. The leaf temperature and transpiration rate are dependent variables which are functions of the four independent variables: radiation, air temperature, wind, and humidity. Therefore, it is seen that one must deal with a six-dimensional problem. This is complicated, but there is no choice. It is not valid to ask for the influence of air tempera- ture on transpiration rate without specifying the values of all other vari- ables simultaneously. It is this simultaneity of factors which makes eco- logical problems complex. ENERGY EXCHANGE A leaf absorbs an amount of radiation which is designated Q,,, in cal cm~-? min-!. The absorbed radiation is the sum of absorbed direct sun- *Supported by grant No. AT(11-1)-1711 from the U.S. Atomic Energy Commis- sion and the Center for the Biology of Natural Systems under PHS grant No. 1 P10 ES 00139-03 ERT. Field facilities and transportation were provided under the NSF grant GB: 3975 to R. A. Howard. 94 JOURNAL OF THE ARNOLD ARBORETUM [vor. 50 light, scattered skylight, reflected light, and emitted thermal radiation from ground, vegetation, and atmosphere. The leaf absorbs each incident stream of radiation according to the absorptivity of its surface and the leaf orientation. This is discussed by Gates (1968a) in detail. The leaf consumes a very small fraction, maybe one or two percent, of the ab- sorbed radiation in photosynthesis. The major portion of the absorbed radiation is lost by radiation emitted from the leaf surface, by convection and by transpiration. The energy budget for the leaf is given as follows: 1/2 od, (T1) — rh. sda (Te) Que = co TY +k( ~) (T, — Ta) +L — (1) where ¢ is the emissivity of the leaf surface, o is the Stefan-Boltzmann radiation constant, k is a constant, V is the wind speed in cm sec~’, the width of the leaf in cm, T,; and T, the leaf and air temperatures respec- tively, L the latent heat of vaporization of water (580 cal gm~' at 30°C), the tribes. The hairs he calls ag : hairs (Zwillingshaare) consist of two more or less isometric basal ce a ne pulvi Numerous elaborations and/or reductions of this basic type of double hair have occurred in different species and genera of the family. (Cf. Senecio.) 110 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 hemispherical, composed of a single row [rarely more] of erect, usually free, flat [keeled], green bracts; supernumerary bracts sometimes present; receptacle slightly convex [flat], naked, foveolate. Florets dimorphic (perfect and carpellate) or all tubular and perfect; pappus setose-capillary, soft, white; ray florets, when present, carpellate, in a single outer series,® the corolla with an irregularly toothed ligule, yellow; disc florets perfect, the corolla tubular, shortly 5-fid, yellow to orange (rarely white or light purple), the anthers with terminal appendages and truncate bases; pollen spherical, more or less spiny, prominently tricolporate (cf. Greenman) ; style branches of perfect florets truncate (to penicillate) or with a short, pointed apex. Achenes subterete, 5—10-nerved, variously pubescent. LECc- TOTYPE SPECIES: S. vulgaris L.; see Cassini, Dict. Sci. Nat. 48: 454. 1827. (Name Latin, applied to groundsel, S. vulgaris; derived from senex, old man, referring to the soft, white pappus which suggests the beard of an old man). — RAGWoRT, GROUNDSEL. An ubiquitous genus, possibly the largest of the flowering plants, esti- mates varying from 900 (Bentham & Hooker) to 3000 (Cabrera) species. There has been no treatment of the entire genus since that of De Candolle (1838) in which he divided the genus into several “series” based on geo- graphical distribution. Most authors have arbitrarily accepted this treat- ment as a basis and have worked within one geographical area (e.., Muschler, Africa; Cufodontis, northern Eurasia; Cabrera, Chile; Green- man, North and Central America). Greenman placed the North and Central American species in 21 sections; five sections (some of which are not very distinct) with eleven species occur in the Southeast and four introduced and 15 native species are known from the eastern United States as a whole The European section Senecio ($ Annui Hoffm.), composed of weedy annual herbs, is represented in our area only by the now almost cosmopoli- tan S. vulgaris L., 2n = 40, which differs from our other species in its annual habit and uniformly discoid heads of yellow flowers. Two species of sect. SANGUISORBOIDEI Greenman, characterized by the perennial habit of its species and the more or less glabrous, once or more pinnately parted leaves, occur in the Southeast. Senecio glabellus Poir., 2n = 46, is wide ranging in wet habitats from Mexico eastward to Florida and north to Oklahoma, Kansas, Missouri, Illinois, Indiana, Kentucky, Tennessee, and North Carolina, whereas S. Millefolium Torr. & Gray (including S. Mem- mingeri Britton), with basal leaves two or three times pinnate, is restricted to rocks and cliffs in a few counties in the mountains of southwestern Vir- ginia, western North and South Carolina, and northernmost Georgia. The remainder of the species of this pence are found predominantly in the uplands of Mexico and Central Amer Most of our species fall into sect. ja Rydb., which is composed of * The presence of ray florets in normally radiate species is not an absolute character. The number of ray florets is also subject to great variation, mainly according to the number of involucral bracts 1969 | VUILLEUMIER, GENERA OF SENECIONEAE 111 perennial, usually glabrous herbs with petiolate simple or lyrately parted basal leaves and cauline leaves reduced upward. Six of the 22 species of this group reach our area: S. aureus L. (including S. gracilis Pursh), 2n = 44; S. Robbinsii Oakes ex Rusby, 2m = 46; S. obovatus Muhl. (including S. rotundus (Britton) Small), 2n = 40; S. Smallii Britton, 2n = 44; S. pauperculus Michx. var. Crawfordii (Britton) T. M. Barkley; and S. plattensis Nutt. Senecio aureus, S. Robbinsii, and S. pauperculus var. Crawfordii all frequent moist to wet meadows and bogs, S. obovatus and S. plattensis prefer drier areas, and S. Smallii grows primarily in fields, roadsides, and open woods. Senecio Robbinsii occurs in the Southeast only as a remarkably disjunct population on Roan Mountain (Tennessee- North Carolina border) with the principal populations located far to the north in the mountains of New York and New England and in adjacent Canada. Many of these species commonly hybridize where their ranges and habitats overlap, which often makes identification of intermediate plants difficult. However, hybrids are usually restricted to “hybrid’”’ habitats and do not seem to swamp out the parental species. The species of sect. AurEI, their ecology, natural history, and evolution, have been thoroughly discussed by Barkle Members of sect. TomeNtos1 Rydb. differ from those of sect. AUREI primarily in a tendency toward being permanently tomentose. The major- ity of the species are centered in the Rocky Mountains, but the range of the type species, Senecio tomentosus Michx. (including the glabrous-leaved f. alabamensis (Britton) Fern.; S. alabamensis Britton), 2n = stretches across the country in weedy areas from Arkansas and Texas to Florida and north to southern New Jersey. Hybrids between S. tomen- tosus and S. aureus are fairly common (Barkley), and the hybrid of S. tomentosus and S. Smallii also occurs. The latter has been studied cyto- logically (in the first meiotic division 2m = 21-22 bivalents and 1-3 univalents). This evidence and other studies of Barkley have shown that some of the sections used by Greenman are artificial and should possibly be abandoned. In the southwestern United States (Colorado, New Mex- ico, Texas) hybrids are formed between S. mutabilis Greenm. (sect. TOMENTOsI) and S. neomexicanus Gray (sect. ToMENTOsI), S. mutabilis and S. multilobatus Torr. & Gray (sect. Lopatt Rydb.), and between S. mutabilis and S. tridenticulus Rydb. (sect. AUREI). An earlier study had already indicated that the species of the S. multilobatus group be- longed in an integrated complex with species formerly considered to belong to sects. BoLANDERANI Greenm., LosaTI, and AUREI. (See also the cytological data of Ornduff et al., 1967. ) The last section, RUGELIA (Shuttlew. ex Chapm.) Greenm., contains only the unique Senecio Rugelia Gray (Rugelia nudicaulis Shuttlew. ex Chapm.), winterwell, 2n = 56, a perennial herb with alternate, undivided leaves and large, nodding discoid heads of white or light purple florets in a simple corymbose raceme. This species grows in partial shade in 112 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 cool woods (usually of Picea rubens Sarg. and/or Abies Fraseri (Pursh) Lindl.) at high elevations (ca. 1300 m.) in the Smoky Mountains of western North Carolina and eastern Tennessee. On the basis of both the morphology and the chromosome number, Ornduff e¢ al. have reiterated that this species should be removed from Senecio. Considering the size of the genus, relatively few studies have been made on its embryology, cytology, and anatomy. Palmblad (1965) and Ornduff et al. (1967) have recently added much new cytological infor- mation and discussed some of the possible significance of chromosome numbers within the genus, but too many species have still not been counted to allow decisions concerning the genus as a whole. Gustafsson reported no apomixis in the species of Senecio he examined, a finding corroborated on other species by Afzelius and Haskell. The breeding system in the few cases studied appears to be one of facultative out- breeding with occasional inbreeding (Knuth, Haskell). Hauman postu- lated that the arborescent senecios (sect. ARBOREI Hoffm.) of Africa are all obligate inbreeders. Anatomical studies have been made most extensively on the African arborescent species of Senecio (cf. Hare). Recent c comparative work by apparently because in the Senecioneae stem anatomy is easily modified under different environmental conditions. Yet within Senecio itself, Carl- quist found that clustering of species on the basis of wood anatomy was, in some cases, consistent with groupings based on other morphological criteria. Hare and Carlquist concur that the woody members of the Senecioneae are derived from herbaceous ancestors and that the stem structure of Senecio is advanced in comparison with other genera. A recent study by Drury & Watson on some of the Eurasian sections of Senecio has revealed that the leaf and achenial hairs, pappus types, and the kinds of ovarian crystals — when carefully and critically examined —provide useful taxonomic characters. They call for a reassessment of many characters usually superficially examined in species of the Com- positae and imply that the use of these characters might help in pro- ducing a more natural classification of such troublesome genera as Senecio. The specialized anatomy of the achenial double hairs of Senecio vul- garis (see footnote 5 under Arnica) has been described by Macloskie and J. Small and that of several other species of Senecio by Hess. The basal cells, as in the double hairs of most Compositae, act as pulvini sensitive to moisture. In several species of this genus, the two hair cells are further specialized and are filled with a spiral tongue of a mucilaginous substance which is extruded when pressure due to water absorption forces the hair cells to separate. Apparently, the seer sticks the achenes to soil particles and thus helps to insure germina Alkaloids reported in at least 75 species (cf. ‘Willaman & Schubert) un- doubtedly account for the medicinal use of various species of Senecio. 1969} VUILLEUMIER, GENERA OF SENECIONEAE 113 In the United States, only S. aureus was extensively used, the leaves first being dried, then steeped in water, and the liquid used as a stimulant, diuretic, and uterine tonic. The last use of this brew by North American Indian women led to the common name of squaw-weed. In other parts of the world, shoots and leaves of several species are eaten raw or cooked. Some species (especially those of Cineraria L., if this genus is merged with Senecio) are cultivated as ornamentals. Much attention has been directed toward Senecio and the Senecioneae because of the writings of James Small, who attempted to prove in an elaborate series of papers (1917-1919) that Senecio was the ancestral genus of the Compositae. His theory has, however, been dismissed by most workers with only a partial explanation. It thus seems worth noting here that four general concepts, now considered to be erroneous, lay at the base of his argument: (1) derivation of the Compositae from the Campanulaceae subfam. Lobelioideae; (2) acceptance of the now refuted Age and Area hypothesis of Willis; (3) the uplift of the Andes in the early Cretaceous; (4) belief in the doctrine of evolution by saltation. As a consequence of these tenets, Small proposed that the ancestral pre- Composite had a woody habit, a zygomorphic bilabiate corolla, and united anthers (free anthers are now considered primitive); * that Senecio, the largest and most widespread genus of the family was naturally the oldest; that the uplift of the Andes in the Cretaceous (rather than in the Pliocene- Pleistocene as is now accepted) gave the genus ample time to spread around the world; and finally, that evolution by saltation, combined with this (presumed) early Andean uplift created a situation in which the lobelioid pre-Composite evolved and radiated as the Andes rose and thereby created a plexus of species able to migrate throughout the world. Small also had a number of ideas concerning morphology which re- inforced his conviction that Senecio was the ancestral Composite: 1. The pappus was developed from a structure that was morphologically a hair. Consequently, a fine capillary pappus (as is found in Senecio) should be primitive. The pappus now considered by most taxonomists to be the ancestral type is composed of broad, flat bristles resembling the a of the calyx, from which it is thought to be derived. . The inflorescence of the pre-Composite was an umbel with all of the pedicellar bracts except the outermost series already suppressed. Fur- ther reduction would have resulted in a head with a flat or convex naked receptacle and, correspondingly, a uniseriate involucre. Additional series of receptacular or involucral bracts would be produced by the abortion of florets in the head. Although it is still debated whether the primitive Composite possessed an umbel or a panicle, most authors now accept entham’s view that receptacular bracts and a multiseriate involucre are unspecialized. A uniseriate involucre and a naked receptacle, as in Senecio, are now considered to be advanced reductions. “See Cronquist (1955) for a discussion of the evolution of the Compositae and an enumeration of characters considered to be unspecialized in the family. 114 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 3. Through a series of drawings, Small showed how all the types of style branches now found in the Compositae could be derived from the flat, truncated style arms of Senecio. Similarly, he derived all the anther types from the senecionid type with its terminal appendage and tailless base. Yet, since this kind of hypothetical derivation from a selected prototype can be made using almost any form (except for the obviously highly modified ones) as a starting point, it really has little biological meaning 4. Several corolla characters in Senecio were also suggested by Small as primitive. Yellow, for example, was considered the ‘‘unspecialized” flower color. Bilabiate corollas found in the ray florets of some species of Senecio were deemed unspecialized because they were like those of Lobelia. Al- though yellow may be a basic flower color in the Compositae, the tubular corolla is now believed to be primitive and to have given rise to both the bilabiate and the ligulate corolla (cf. Koch). 5. The chromosomal evidence available to Small suggested that five was the base number for Senecio. More recent evidence, however, indicates that ten is actually the base number for the genus (Ornduff et al.). In spite of these mistaken ideas about Senecio, Small’s studies provide one of the most complete comparative morphological surveys ever made on the Compositae. Even without the bibliographies and summaries of previous work, his research is an indispensable reference on Senecio and the Compositae i in general. REFERENCES: All references listed under subtribe Senecioninae are pertinent. AFzELIUs, K. Embryologische und zytologische Studien in Senecio und verwand- ten Gattungen. Acta Horti Berg. 8: 123-219. 1924. ALEXANDER, E. J. Senecio Rugelia. a 20: 29, 30. pl. 655. 1937. Sene- cio Millefolium. Ibid. 31, 32, pl. 6 Bark_ey, T. M. A revision of — aureus Linn. and allied species. Trans. Kansas Acad. Sci. 65: 318-408. 1962. icooeay treatment of Senecio aureus fa ap group with comments on evolut intergradation of Senecio plattensis ey Sélacie pauperculus in Wis- consin. _Rhodors 65: 65-67. 1963. . Taxonomy of Senecio see arrecg and its allies. Brittonia 20: 267— 284. 1968, "Tncludes S. Millefolium, 2 . Intergradation of Senecio sections se Tomentosi and Lobati through Senecio mutabilis Greenm. (Compositae). Southwest. Nat. 13: 102-115. 1968. re saci A. L. El género Senecio en Chile. Lilloa 15: 27-501. 1949. [In- cludes 208 spp., numerous illustrations, little of generic or specific rela- tionships. | CLutTe, W. N. The meaning of plant names. XLVII. Senecios and others. Am. Bot. 37: 105-109. 1931. Cotron, A. D. The megaphytic habit in the tree Senecios and other genera. Proc. Linn. Soc. Bot. London 156: 158-168. 1944. 1969] VUILLEUMIER, GENERA OF SENECIONEAE 115 Cronguist, A. Phylogeny and taxonomy of the Compositae. Am. Midl, Nat, 53: 478-511. 1955. [A general work on evolution of family with list of charac- ters considered primitive and key to tribes. ] Curopontis, G. Kritische Revision von Senecio sectio Tephroseris. Repert. Sp. Nov. Beih. 70: 1-266. pls. 1-5. 1933. Drury, D. G., & L. Watson. Anatomy and taxonomic comes of gross vegetative morphology in Senecio. New Phytol. 64: 307- 65. & . A bizarre pappus form in Senecio. Taxon - enn 311. 1966. Cie pappus. ] GREENMAN, J. M. Monographie der nord- und centralamerikanischen Arten der Gattis Senecio. Bot. Jahrb. 32: 1-33. 1902. [A general treatment for this area, with discussion of anatomy, morphology, relationships, evolution, and key to sections. | - Monograph of the North and Central American species of the genus Senecio — Part II. Ann. Missouri Bot. Gard. 2: 573-626. pls. 17-20. 1915; 3: 85-194. pls. 3-5. 1916; 4: 15-36. pl. 4. 1917; 5: 37-107. pls. 4-6. 1918. [The basic treatment. ] Gustarsson, A. Apomixis in higher plants. Lunds Univ. Arsskr. II. Sect. 2. “2: 1-68. 1946; 43: 69-372. Vs Hare, C. L. The arborescent senecios of reece a study in ecological anatomy, Trans. Roy. Soc. Edinb. 60: 355-371. 1940. HASKELL, G. Adaptation and the breeding coating in groundsel. Genetica 26: 468-484. 1953. [S. vulgaris. ] Hauman, L. Les “Senecio” arborescents du Congo. Etude morphologique, phytogéographique et systématique. Revue Zool. Bot. Afr. 28: 1-76. pls. 1-11. 1935. [Also includes photographs and comparisons with the paramos of S. Am. Kocu, M. F. — in the anatomy and morphology of the Composite flower II. The corollas of the Heliantheae and Mutisieae. Am. Jour. Bot. 17: 995-1010. pls. 51, 52. 1930. Mactoskie, G. Achenial hairs and fibers of Compositae. Am. Nat. 17: 31-36. 883. [A short preliminary attempt, but interesting. For more detail, see Hess under subtribal references. | d MuscHter, R. Systematische und pflanzengeographische Gliederung der afri- kanischen Senecio-Arten. Bot. Jahrb. 43: 1-74. 1909 Ornourr, R. Evolutionary pathways of the Senecio lautus alliance in New Zealand and Australia. Evolution 18: 349-360. 1964. PALMBLAD, I. G. Chromosome numbers in Senecio (Compositae). I. ——_ Jour. Bot. 43: 715-721. 1965. [65 collections representing 30 spp.; in- cludes 5S. aaron Britton, 2n = 46, endemic to shale barrens of Virginia, W. Virginia, and Penn. to the north of our area. | Smatt, J. The a aa: ‘mechanism in the Compositae. Ann. Bot. 29: 457-470. 1 oe Wittaman, J. J., _ G. Scuusert. Alkaloid-bearing plants and their con- tained alkalcids, S. Dep. Agr. Tech. Bull. 1234: 1-287. 1961. [Senecio, 71-75.] 3. Cacalia Linnaeus, Sp. Pl. 2: 834. 1753; Gen. Pl. ed. 5. 362. 1754. Tall caulescent herbs arising from a rosette of alternate, petiolate, Spathulate, ovate, reniform, or hastate, entire, undulate, crenate, or 116 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 toothed [or lobed] basal leaves; stem leaves petiolate or sessile, decreas- ing in size toward the inflorescence. Inflorescence a compound cyme of numerous campanulate discoid heads with cylindrical or campanulate involucres composed of a single series of herbaceous, lanceolate, winged or flat bracts (sometimes with an outer series of supernumerary brac- teoles); receptacle flat, naked, or with a fleshy projection in the center. Florets monomorphic, perfect; pappus capillary, white; corolla deeply 5-cleft, white, cream, or pinkish; anthers with terminal appendages and obtuse bases; style branches truncate or with short conical [or elongate in some Mexican species] appendages. Achenes fusiform to cylindrical with a variable number of ribs, smooth. (Synosma Raf. ex Britton & Brown, 1898 [Hasteola Raf. ex Pojark., 1960, nom. superfluum]; includ- ing Arnoglossum Raf., 1817 [Mesadenia Raf., 1838, nom. superfluum].) Lectotype SPEctEs: C. hastata L., largely typified by the removal of the original species to other genera; see Miller, Gard. Dict. Abr. ed. 4. 1754; De Candolle, Prodr. 6: 327. 1838; Kitamura, Mem. Coll. Sci. Kyoto Univ. B. 16: 170. 1942; Shinners, Field Lab. 18: 79. 1950; Pojarkova, Fl. URSS 26: 684. 1961.8 (Name from Greek, kakalia, a name given by Dioscorides to a plant believed to be a Tussilago.) — INDIAN PLAN- TAIN. A genus of perhaps 40 species distributed from eastern Europe to eastern Asia, in eastern North America, Mexico, and Central America, and in South America along the Andes southward to Bolivia. The Mexican species were referred to Psacalium Cass., Odontotrichium Zucc., and Pericalia Rydb. by Rydberg (1924) and recently to these plus Digitacalia Pippen by Pippen (1968). Some Asiatic species have been removed to Syneilesis Maxim. and Méiricacalia Kitamura. Nine species distinctly divisible into two sections occur in the southeastern United States. Sec- tion Cacaria (treated as the genus Synosma by Small and by Britton & Brown, ed. 2) is represented only by Cacalia suaveolens L., 2n = 40, which occurs in moist woods from Massachusetts to Minnesota, south- ward to Missouri, Tennessee, and western North Carolina. Morphologi- cally, it is distinct from all the other eastern North American species in having hastate leaves with pinnate venation, large heads with 12-15 involucral bracts (plus a ring of bracteoles), a naked receptacle, and numerous florets.® Its affinities lie with the Eurasian C. hastata L., 2n = “The typification of Cacalia will be discussed in a subsequent paper. Rydbe (1924), Cuatrecasas (1960), and Pippen (1968), contrary to the course followed spin have maintained that Cacalia should be typified by C. alpina L., which was remove from Cacalia as the type species of Adenostyles Cass. This choice restricts the name to a genus of four or five species of Central Europe *T have seen one atavistic specimen of Cacalia suaveolens (Moore, Rosendahl & 1969 | VUILLEUMIER, GENERA OF SENECIONEAE 117 60, and its Asiatic relatives rather than with any other North American ecies The seven other species in the Southeast form a closely knit distinctive group (cf. Pippen) which constitutes sect. ConopHoRA DC. (Arnoglossum Raf., Mesadenia Raf.). All are morphologically similar in having palmately nerved leaves, five involucral bracts, a fleshy projection in the center of the receptacle, and five florets.’ Cacalia Muhlenbergii (Sch.-Bip.) Fern. (C. reniformis Muhl.), 2n = 50, occurs in woodlands from New Jersey and Pennsylvania, west to Minnesota and south to Missouri, Alabama, and Georgia. Cacalia lanceolata Nutt. var. lanceolata, 2n = 56, occurs in moist to wet habitats from eastern Texas and Louisiana to Florida, northward into southeastern North Carolina, and C. lanceolata var. Eliottii (Harper) Kral & Godfrey (M. Eliottii Harper, C. Elliottii (Har- per) Shinners) '° occurs from peninsular Florida northward into South Carolina. Cacalia diversifolia Torr. & Gray also occurs in swampy areas of southern Georgia and northern oe westward to Louisiana, and C. floridana Gray is endemic to the dry, sandy oak and pine woods of central and northern Florida. Cacalia striplicifolia LL, 20 = 50, 52, 54, 56, and C, apa rape (probably including C. plantaginea (Raf.) Shin- ners), 2n = 54, wide ranging: the former in dry woodlands from New York to helio and Nebraska, south to Oklahoma, and east to Mississippi, Alabama, and Georgia, and the latter in damp prairies from Ontario to Minnesota, south to Oklahoma and Texas. The last species of this section, C. sulcata Fern., is restricted to sandy bogs in southern Georgia and western Florida Within the North American species, evidence supports the distinctive- ness of the two sections, but the inclusion of both into one genus. The morphological differences between the two groups of species are reinforced y differences in germination. Three species of sect. CoNopHORA which have been investigated (Cacalia tuberosa, C. atriplicifolia, and C. Muhlen- bergii) have achenes which need four days to two weeks for germination and cotyledons which are strongly curved upon emergence. In contrast, C. suaveolens (sect. CacALIA) needs only 48 hours for germination and the cotyledons are only slightly curved when they emerge. Afzelius reported the infrequent occurrence of two embryo sacs in ovules of C. suaveolens, each formed from a separate megaspore mother-cell. In one sac, the €gg apparatus was invariably crushed but the antipodals were normal, * The placement of Cacalia ovata Walt. is at present uncertain. This taxon, pre- treatment is followed, C. cocoa var. Elliottii becomes C. ovata var. ovata, and - lanceolata var. lanceolata will req under C. ovata a new combination based on the oldest legitimate varietal cpithet “Gf available). 118 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 while in the other, the egg apparatus appeared normal, but the antipodals were crushed. Greene later reported that good seeds of C. suaveolens were difficult to find, and Wadmond stated that it was impossible to locate viable seed of C. Muhlenbergii. These two observations suggest a similar sort of meiotic irregularity in C. Muhlenbergii. Another feature which links the Southeastern American species (sect. ConopHora) and the Asiatic species (sect. CAcALIA) is the presence of the same type of asexual reproduction. Kral & Godfrey reported, as a general phenomenon in the Florida species, the production of lateral rosettes which become disconnected from the parent plant by disintegra- tion of the connecting stolons. Liubarsky described the same phenomenon in greater detail for several Russian species. Two Japanese species, Cacalia auriculata DC. var. bulbifera Koidz. and C. farfarifolia Sieb. & Zucc., produce bulbils in the leaf axils (cf. Ohwi). Within sect. ConopHora, the species are very similar morphologically, seemingly closely related, and apparently rather removed from other species in the genus. Hybrids within this section appear to be rare, how- ever. The only natural hybrid reported, that between Cacalia atriplicifolia and C. Muhlenbergii (also produced artificially by Coleman), was ex- ceedingly sterile (2—9 per cent pollen staining and no seed set). Generically, Cacalia is ill defined from Senecio L. and its satellite genera. Originally, Linnaeus included in Cacalia the herbaceous perennials treated here (and other species) and a group of shrubby African plants now con- sidered to constitute either the genus Kleinia Mill. or Senecio subg. Kleinia (Mill.) Hoffm. Bentham placed both the herbaceous and the shrubby groups in Senecio, while Hoffmann separated the two, retaining the herbaceous species as Cacalia and referring the species of Kleinia to Senecio. In opposition to Hoffmann’s treatment, however, large and soli- tary crystals of calcium oxalate (rather rare in the Compositae according to Metcalfe & Chalk) have been found in both Cacalia and Senecio subg. Kleinia. Chromosome numbers of the 27 species reported as Cacalia are 2n = 40, 50, 52, 54, 56, 58, 60, 70, and 120. The 16 counts recently reported by Pippen as species of the segregate genera Digiticalia, Odontotrichum, Pericalia, and Psacalium are all 2n = 60, with the exception of that for O. Palmeri which is 2n = ca. 50. Ornduff et al. (1967) suggest that the basic chromosome numbers in Cacalia are 20 and 30 and that other num- bers have been derived by aneuploid reduction from m = 30. More chromosome counts and further study of the generic limits of Cacalia on a worldwide basis undoubtedly are needed. It seems possible, especially in view of the morphological continuity with Senecio in Africa, that the genus is of Old World origin, but Liu- barsky’s postulation of an origin in the region of the upper Amur River is highly questionable. , PS rate 1969] VUILLEUMIER, GENERA OF SENECIONEAE 119 REFERENCES: Under scepin references see BENTHAM, on & Hooker, De CANn- DOLLE, HOFFMAN, ORNDUFF et al., and RICKE AFZELIUS, K. aeames und aay Studien in Senecio und verwand- ten Gattungen. Acta Horti Berg. 8: 123-219. 1924. [C. suaveolens, 162.] ARANO, H. Cytological studies in subfamily Carduoideae (Compositae) of Japan XVIL. The karyotype analysis in Cacalia and open is. Bot. Mag. To- kyo 77: 86-97. 1964. [Gives counts for 9 spp. Cacalia CoLeMaN, J. R. Natural and artificial cag of Gade atriplicifolia and C, M uhlenber ii. Rhodora 67: 55-58. Cuatrecasas, J. Studies on Andean acne ikea Brittonia 12: 182-195. 1960. [Includes typification of Cacalia GREENE, E. L. Studies in a Compositae. eT Part 3. The genus Mesadenia. Pittonia 3: 180-183. GREENE, H. C. Differences in Senge characters and germination in some species of Cacalia L. Am. Midl. Nat. 39: 758-760. 1948. [C. atriplicifolia, C. Muhlenbergii, C. suaveolens, C. tuberosa. | Harper, R. M. Mesadenia lanceolata and its allies. Torreya 5: 182-185. 1905. 1905. Kitamura, S. Recognition of the genus Syneilesis Maxim. (In Japanese.) Jour. Jap. Bot. 10: 699-703. 1934. : — mia du Japon. (In Japanese.) Acta Phytotax. Geobot. 7: 236- 251: AL, R. ‘ = K. Goprrey. Synopsis of the Florida species of Cacalia (Com- positae). Quart. Jour. Florida Acad. Sci. 21: 193-206. 1958 Lruparsky, E. L. Notes on the genus Cacalia in the southern par of the Mari- . 1961 Metcatre, C. R., & L. CuHatk. Compositae. Anat. Dicot. 2: 782-804. 1950. [Cacalia and Kleinia, 786.]| Ouw1, J. Flora of Japan (in English). F. G. Meyer & E. H. WALKER, eds. ix + 1067 pp. Wash. D.C. 1965. [Miricacalia, 882; 13 spp. of Cacalia, 882- 884; Syneilesis, 887, 888. Pippen, R. W. Mevicna * ‘cacalioid” ge allied to Senecio (Compositae). Contr. U.S. Natl. Herb. 34: 363-448. Poyarkova, A. Notae criticae de genere aes de fb Lee Russian.) Not. Syst. Leningrad 20: 370-391. 1960. [Lectotype sp. = C. atriplicifolia; adopts Hasteola Raf. (ex Pojarkova) for C. hastata, C. suaveolens, an relatives. | . Cacalia. Fl. URSS 26: 683-697. 1961. [Lectotype sp. = C. hastata L.] Rypgerc, P. A. Some senecioid genera—I. Bull. Torrey Bot. Club 51: 369- 378. 1924. [Includes incorrect typification of Cacalia. | Suinners, L. H. The Texas species of Cacalia (Compositae). Field Lab. 18: 79-83. 1950. [C. plantaginea and C. lanceolata, with comments on the application of the name Cacalia TaKEsHiITA, M. Cytological studies on Cacalia and its related genera. I. The chromosome number of three species and one variety of acalia and one variety of Miricacalia. (In Japanese.) Jap. ie Genet. 36: 217-220. 1961.* Wapmonp, S. C. The Indian plantain. Asa Gray Bull. 6: 52. 1898. [C. reni- formis = C. Muhlenbergii.] 120 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 4. Erechtites Rafinesque, Fl. Ludov. 65. 1817. Robust annual [perennial] caulescent herbs with fibrous roots and alter- nate, toothed or parted, glabrous or pubescent leaves. Inflorescence a “panicle” or cyme of numerous heads, each with a basally swollen in- volucre composed of a single series of narrow lanceolate scabrous bracts often surrounded by a series of supernumerary bracteoles; receptacle flat, alveolate or fimbrillate. Florets monomorphic and perfect [or in some cases florets dimorphic, the ray florets then carpellate with filiform 4-5-parted corollas]; pappus thin, white, soft, copious; corolla tubular, regular, 5-toothed, whitish or yellowish; anthers with obtuse bases; style branches of perfect florets with a terminal appendage of fused papillose hairs surrounded at the base by a semicircular crown of collecting hairs (cf. Belcher). Achenes oblong to linear in outline, more or less 10[—20]- ribbed, glabrous. (Erechtites sensu Bentham & Hooker and Hoffmann, in part.) Type species: EF. praealta Raf. = E. hieracifolia (L.) Raf. ex DC. (Name from Greek, Erechthites, a name given by Dioscorides to a species of Senecio.) — FIREWEED, PILEWORT. Two sections with five species native to the Americas (sensu Belcher) and adventive in the Pacific region, Asia, and Europe. One wide-ranging species, Erechtites hieracifolia (L.) Raf. ex DC., of sect. ERECHTITES (§ Hieracifoliae Belcher), occurs in weedy habitats in the southeastern United States. Erechtites usually has been distinguished from Senecio and its allies by its two to several series of outer carpellate florets with filiform, eligu- late corollas. Belcher, however, narrowed the genus to include only New World species which possess what he considers the diagnostic feature of the genus: style branches with an appendage of fused papillose hairs with a semicircular crown of collecting hairs at its base. He returned several Australian and Indonesian species traditionally included in Erech- tites to Senecio and placed five New Guinean species (one population in New South Wales) in Arrhenechthites Mattfeld. The characters used to separate Arrhenechthites from Erechtites include the presence of func- tionally staminate disc florets and reduced, astigmatic (and thereby with- out the critical character of Belcher’s Evechtites) style branches. In view of the obvious correlation between sterile (abortive) ovaries and reduced stigmas, the question arises as to whether these species are not simply inbreeders derived from the outbreeding American species. Also, the chromosome number of Senecio (Erechtites) minimus Poir., the only Old World species counted, has a diploid count of 2m = 60, less like the most common number in Senecio, 2n = 40, than E. hieracifolia with 2n = 40. Regardless of which circumscription is used, the taxonomy of our species is unchanged. Our species, Erechtites hieracifolia, has three varieties, two of which are now widely distributed weeds. Varietas hieracifolia (including vars. intermedia Fern. and praealta (Raf.) Fern.) occurs naturally from Can- 1969 | VUILLEUMIER, GENERA OF SENECIONEAE 121 ada southward through the Greater Antilles. It is distinguishable from the other varieties by its short bracteoles less than 1/4 the length of the involucre and by its 10-ribbed achene. It has been introduced into Hawaii and Europe. Varietas megalocarpa (Fern.) Crong. (E. megalo- carpa Fern.), separable from var. Aieracifolia in its much larger receptacle (twice as wide) and its 16—20-ribbed achene, is endemic to sandy coastal habitats from southeastern Massachusetts to New Jersey. Belcher’s sug- gestion that the plants may be tetraploids derived from var. hieracifolia seems not to have been tested. The third variety, var. cacalioides (Fisch. ex Spreng.) Griseb., is found in Central America, the Lesser Antilles, and South America to Argentina, and is now established as a weed in Asia. It differs from the other two varieties in its longer bracteoles with multi- cellular hairs (instead of being glabrous or bearing unicellular hairs). Intermediates between this and var. hieracifolia occur in the West Indies. The success with which Erechtites hieracifolia has managed to colon- ize new areas is apparently due to its adaptability to new environmental conditions and to its easily dispersed achenes. Ridley listed it as one of the first plants to recolonize Krakatau after the volcanic eruption of 1883. Although this species is usually an annual herb, Carlquist has exam- ined a Hawaiian specimen over six feet tall which had secondary xylem. The change from the annual habit in the Hawaiian population has occurred, Carlquist postulated, because plants on oceanic islands are “released” from the selection pressures of a cyclical climate. However, there must be some positive selection for woodiness on islands (see also Carlquist, 1965). In Brazil, the leaves of Erechtites hieracifolia (and of E. valerianifolia (Wolf) DC.) are cooked with palm oil (Corréa), and Ochse reports that in the East Indies the upper leaves, called “lalab,” are eaten raw or steamed with rice and are rumored to be beneficial for nursing mothers. REFERENCES: Under subtribal peau see BENTHAM, BENTHAM & Hooker, HOFFMANN, OrnvurF et al., and RICKET BELCHER, R. O. A revision of the genus Erechtites paacaany” wan inquiries into Senecio and Arrhenechthites. Ann. Missouri Bot. Gard. 43: 1-85. 1956. Carxouist, S. Wood anatomy of Senecioneae (Compositae). - 5: 123-146. 1962. [Comparative anatomy with conclusions as to relationships within tribe and correlations with ecology. ] Island life. 451 pp. New York. 1965. [Chapter 8. Some remodeled plants, ] Cooper, G. O. Cyt tological investigations of Erechtites hieracifolia. Bot. Gaz. 98: 348-355. 1936. [Describes both micro- and mega sporagenesis. Corréa, “~ P. Diccionario das plantas uteis do Brasil e das exoticas cultiva- vols. Ministerio da Agricultura, Rio de Janeiro. 1926, 1931. [Erech- tites, * 96. 1931.] 122 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 FERNALD, M. L. The genus Erechtites in temperate North America. Rhodora 19: 24-27. 1917. [E. hieracifolia, its varieties, and E. megalocarpa. | OcuseE, J. J. Vegetables of the Dutch East Indies. 1006 pp. Buitenzorg, Java. 1931. [Erechtites, 132-134. ] Riptey, H. N. The dispersal of plants throughout the world. xx + 744 pp. pls. 1-22. oy Kent, England. 1930. [Erechtites, 133, 160, 656. SANForD, S. N. Erechtites megalocarpa in Rhode Island. Rhodora 28: 111. 1926. [See on H. K. Svenson, Rhodora 41: 256. 1939. | 5. Emilia Cassini, Bull. Sci. Soc. Philom. Paris III. 1817: 68. 1817. Annual or perennial caulescent herbs arising from a rosette of lyrate- pinnatifid or spathulate, dentate [entire], glabrous or glaucous leaves. Stem leaves alternate, dentate or lyrately lobed, decreasing in size toward the lax corymbose inflorescence [plants sometimes monocephalous] of discoid heads. Involucre tubular, often swollen at the base, composed of a single row of lanceolate slightly scarious bracts; receptacle flat, naked. Florets perfect; pappus setose, soft, white or purplish; corollas tubular, shortly 5-fid, lavendar [white] or red; anthers truncate at the base; style branches terete with penicillate appendages surrounded by a ring of hairs. Achenes 5-angled, truncate at both ends. Typr species: Cacalia sagit- tata Vahl, nom. illegit. = Emilia javanica (Burm.) C. B. Robinson. (Derivation of name not explained but apparently from the French proper name Emilie.) — Cupip’s PAINTBRUSH. A genus of about five species native to the Tropics of Africa and the Far East. Two weedy species, Emilia sonchifolia (L.) DC. ex Wight, 2n = 10, and E. javanica (Burm.) C. B. Robinson (E. coccinea (Sims) G. Don in Sweet, E. sagittata (Vahl) DC., E. flammea Cass.), 2n = 20, are naturalized in disturbed and weedy habitats in the warmer parts of peninsular Florida, where the latter is more frequently encountered. Both are also rather widely naturalized in the West Indies, Central America, northern South America, and Brazil. The application of the names of these two species has been the source of much confusion. There has never been any doubt that there are two entities: one a species with lyrate-pinnatifid lower leaves, small heads, lavender (rarely white) corollas which barely exceed the involucre, laven- der anthers and styles, and white pollen; the other with ovate-spathulate dentate leaves, carmine corollas which extend conspicuously beyond the involucre, orange anthers and styles, and bright yellow pollen. The first species is Emilia sonchifolia, and the other long has been known as E. sagittata in the Old World and FE. coccinea in the New. When Cassini described Emilia he listed Cacalia sagittata Willd. as type species. Will- denow, in turn, referred to Cacalia sagittata Vahl, excluding the synonym Hieracium javanicum Burman. Because Willdenow excluded the synonym (Vahl’s inclusion of this earlier legitimate name as a synonym made the name Cacalia sagittata superfluous) and because the type specimens of 1969] VUILLEUMIER, GENERA OF SENECIONEAE 123 Vahl’s and Burman’s names had not been examined, it has been assumed that two different taxa were involved. However, Mattfeld (1929) estab- lished that Vahl’s type belongs to the large-headed, red-flowered species, and Fosberg (1966) finally located the Burman type and reported that it, also, was this species. Since Burman’s is the oldest legitimate name available, it must replace the other names now used. That Cassini had the showy red-flowered species in mind when he described the genus Emilia is evidenced by his changing the name of the type species to EF. flammea (nom. illegit.). The two species apparently do not interbreed in nature, and attempts to produce artificial hybrids (cf. Lee) have failed, indicating that a sterility barrier (as well as the difference in chromosome number) is involved. The two species are not only incompatible, but also have dif- ferent mechanisms of reproduction (at least in Jamaica): Emilia sonchi- folia is an obligate inbreeder, while E. javanica is outbreeding. e genus seems to be a natural group of species closely related to, but distinct from, Senecio L. Three of the five species counted have a chromo- some number of 2” = 10, and the other two have 2m = 20. The species with five pairs of chromosomes are annuals apparently derived from less specialized species with 2n = 20 (cf. Ornduff et al.). The genus has no true commercial value, although Emilia javanica is sometimes used in tropical areas as an ornamental. Baldwin mentioned that the plants were eaten in the Far East, but not in the New World. REFERENCES: Under subtribal references see BENTHAM, BENTHAM & Hooker, DE CANDOLLE, HOFFMANN, and RICKETT BaLpwin, J. T., Jr. Cytogeosraphy of Emilia Cass. in the Americas. Bull. Torrey Bot. ‘Club 73: 18-23 6. ———.. Cytogeography of mais in Wes Africa. Ibid. 76: 346-351. 1949. FosBerc, F. R. Miscellaneous notes on Hawaiian plants. Occas. Pap. Bishop Mus. 23: 129-138. 1966. [Typification of Emilia sagittata Vahl and Hieracium javanicum Burm. GaRABEDIAN, S. A revision of Emilia. Bull. Misc. Inf. Kew 1924: 137-144. 1924. Goopine, E. G. B., A. R. Lovetess, & G. R. Proctor. Flora of Barbados. 486 pp. London. 1965. [E. sonchifolia, 437, fig. 27.] Koster, J. Notes on Malay Compositae III. Blumea 7: 288-291. 1952. [Dis- cusses application of names E. sagittata and E. javanica; but see Fosberg. ] Ler, B. Jamaican species of Emilia. Sci. Notes sie Jamaica 2: 14, 15. 1966. [Breeding systems of E. sagittata and E. javanica. | MarttFELp, J. Die Compositen von Papuasien. a Jahrb. 62: 386-451. 1929. [Emilia, 443-447, nomenclature. | Sims, J. Cacalia coccinea. Scarlet-flowered Cacalia. Bot. Mag. 16: pl. 564. 1802. (E. javanica. | THE ARNOLD ARBORETUM Present address: THE GRAY HERBARIUM OF OF HARVARD UNIVERSITY HARVARD UNIVERSITY 124 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ASPECTS OF THE COMPLEX NODAL ANATOMY OF THE DIOSCOREACEAE ? EDWARD S. AYENSU THIS PAPER IS AN ATTEMPT to explain how the vascular tissue of two successive internodes maintains continuity in the complex nodal structure between them in stems of the Dioscoreaceae, especially in the genera Dioscorea and Tamus. Because of the economic importance of this family early emphasis (Mason, 1926) was placed on the relation between struc- ture and function, This led physiologists to take a look at the anatomy before they had a full knowledge of how food substances are translocated in the plant. The Dioscoreaceae is a monocotyledonous family which is distributed throughout the tropics and subtropics of the world. It is, by all standards, one of the most economically important foodstuffs in the diet of most tropical peoples (cf. Coursey, 1967). Attention has recently been focused on this family, especially the genus Dioscorea, because a precursor of cortisone and other related steroidal drugs is derived from the tubers of some species. The unique anatomy of the nodes of the stems of the Dioscoreaceae was brought to attention by Mason (1926) when he studied the rate of sugar transport in Dioscorea alata L. Earlier, Falkenberg (1876) had called the glomerulus of the node an imperfect knot in his study of D. villosa L. Mason noted that the phloem was of a markedly abnormal type. He further observed that the sieve tubes of the successive internodes did not join with each other directly but through a glomerulus which was composed of a great number of oblong thin-walled parenchymatous cells, each with a distinct nucleus, running fairly parallel with each other. Behnke (1965a) questioned the presence of nuclei in the glomerulus cells. Present studies show that nuclei occur at certain stages in the ontogeny of these cells (Fic. 2). In his study of the ontogeny of the stem of Tamus communis L., Bur- kill (1949) disproved Mason’s claim that glomeruli were absent from the nodes of T. communis. Present studies reveal that glomeruli are cer- tainly present in the nodes of Tamus (Fic. 12) and, although they cannot be easily overlooked, it should be emphasized that the glomeruli in this genus are not so pronounced as those of most species of Dioscorea (Fics. 4-11). A full account of the vegetative anatomy of the Dioscoreaceae will be includ in Bs Anatomy of the Monocotyledons. Dioscoreales, ed. C. R. Metcalfe, Oxford University Press. 1969 | AYENSU, ANATOMY OF DIOSCOREACEAE 125 Happ (1950) wrote his thesis on the nodes of the Dioscoreaceae but a copy is not available to me. However, a comment on it appeared in Braun’s (1957) work. Essentially, Happ investigated by means of serial sections the interlacing of the xylem-phloem glomeruli in the vascular system of the node. Brouwer (1953) published his account of the arrangement of the vas- cular bundles in the nodes of Dioscoreaceae and presented a diagram of the elements of the node. Brouwer concluded that the sieve tubes of two successive internodes were connected in the following manner: sieve tubes, funiculus cells, bast tubulus cells, glomerulus cells, bast tubulus cells, funiculus cells, and sieve tubes. Brouwer, following Mason (1926), con- cluded that the phloem-glomerulus cells were (a) densely filled with cyto- plasm; (b) with a persistent nucleus with nucleolus; and (c) without sieve areas. A comprehensive study of the nodal anatomy of Dioscorea batatas Decne. and Tamus communis was conducted by Braun (1957). He con- cluded that (a) the xylem-glomerulus consists of very numerous short tracheids of various sizes, the orientation of which is difficult to trace; (b) the phloem-glomerulus, which is divided into several partial glomeruli, is composed of a new type of translocatory cell, called phloem-glomerulus cells; and (c) the phloem-glomerulus cells possess thin walls without sieve pores and without visible pitting; they are distinguished from parenchyma cells by their lack of starch. Behnke’s (1965c) electron microscopic studies show that sieve areas are, in fact, present in the phloem-glomerulus cells. The present study, involving more species than were available to earlier investigators, essentially supports and extends their conclusions. MATERIALS AND METHODS My observations are based on 180 specimens of 112 species. A com- plete list and citations are given elsewhere (Ayensu, 1966). Most of the specimens examined were fluid-preserved in formalin acetic alcohol. Microscopic details were studied in serial sections at 10u, and those produced on a sliding microtome usually at 162. Depending upon the nature of some specimens, sections were cut up to 90u. The sections were stained in safranin and counterstained with Delafield’s haematoxylin followed by conventional differentiation, dehydration, clearing in xylene, and mounting in Canada balsam. NODAL ANATOMY As pointed out in an earlier paper (Ayensu, 1965), the vascular strands between the petiole and the stem at the nodes of many species of Dios- coreaceae are highly distinctive and are believed to be unique in the family. Longitudinal serial sections of the node reveal two groups of interlacing vascular elements, each forming a plexus close to the petiole insertion. 126 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 VT ) | ) f 00 ) \ i Fic. la (LEFT). Schematic diagram illustrating the arrangement of the ele- ments of xylem-glomerulus in the nodal region of stems of Dioscorea and amus. Fic. 1b (RIGHT). Vessel-like tracheid showing a reticulate perforation plate (lower) and bordered pits (upper). 1969 | AYENSU, ANATOMY OF DIOSCOREACEAE 127 Xylem-glomerulus. Serial sections and macerations reveal that the mature xylem glomerulus is mainly composed of short tracheids of vari- able shape closely fitted together, thus resembling the distinct parts of a composite jig-saw puzzle. These peculiar tracheids are confined to the node and have large bordered pits. Presumably in the internodes water moves freely from vessel element to vessel element through the scalariform perforation plates. Exactly how materials are translocated through the nodal region is not clearly understood. The phyllotaxy determines the width of the glomerulus in the nodes. In the species having simple, alternate leaves, a single glomerulus oc- cupies about one-third of the area of the node. In an opposite (or decus- sate) arrangement, the glomerulus occupies about two-thirds of the nodal area. In species that exhibit a whorled arrangement, the glomerulus oc- cupies almost all the nodal area. The tracheids vary in width and length within species. The widths varying from 40» to 110u, and lengths from 80, to 260 have been re- corded for different species. These tracheids are closely fitted together, and have numerous pit-pairs on their common walls. The exact pathway of the contiguous tracheids between successive internodes is very com- plicated and variable within a species. (See Fics. 4-12.) Longitudinal | ©, Fic. 2 (LEFT). Schematic diagram of the phloem glomerulus. : : Fic. 3 (RIGHT). Schematic representation of the stem showing the relationship of and the position of the xylem and phloem glomeruli in the region of the leaf Insertion. 128 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 sections and macerations of the node give a partial elucidation of the complicated sequence of the tissue structures. As Braun (1957) inter- preted D. batatas and Tamus communis, a vessel just about to enter a node is attached to 1, 2, or 3 cells which Braun referred to as “‘vessel-like tracheids.” The end wall of the vessel-like tracheid (VT) facing the vessel (v) has a reticulate perforation plate, while the other end wall has bordered pits (Fics. 1a, b). The elements that constitute the bulk of the xylem-glomerulus lie between the vessel-like tracheids. The tracheids of the first group (Tj) are closely fitted to those of the second group (T2) and to other successive tracheid groups, thus establishing the normal com- munication between them. The lengths of the tracheids vary from one node to the other within a species. In this respect variation in tracheid length does not have any taxonomic value. Those of the first few groups (T; —Ts3) are shorter than those of T, and T;. It is also observed that the tracheid groups increase in number from T; to T;, presumably for enlarging the water conducting tissues in the node. The surface area of the water conducting tissues is further increased by the complex arrange- ment of many xylem-glomeruli at a node. Each glomerulus is S-shaped and longitudinally orientated. A xylem-glomerulus diagram (Fic. 1) is presented for the sake of simplicity, but the full complexity of it is demon- strated by Fics. 4-12. Phloem-glomerulus. The construction of the phloem-glomerulus (Fic. 2) follows essentially the scheme presented for the xylem-glomerulus (Fic. 1). The phloem-glomerulus is made up of what Braun (1957) named ‘‘glomerulus sieve-tubes” (GS). Earlier, the same tissues had been called “funiculus cells” by Brouwer (1953) and “funnel-cells” by Mason (1926). Recently, Behnke (1965a) has called the same tissues “connect- ing sieve-tubes.”’ Essentially, these tissues are composed of somewhat funnel-shaped, thin-walled cells having numerous small simple pits at the end walls adjoining the PH. They differ from ordinary sieve tubes in the presence of sieve plates only at the end adjoining the sieve tubes. The glomerulus sieve-tubes adjoin the cells that make up the bulk of the phloem glomerulus. These cells were designated “phloem-glomerulus cells” of the first (PH), second (PH), and third (PH;) orders by Braun (1957). Similar cells had earlier been called “bast tubulus” and ‘“‘glomer- ulus cells” for PH, and PHg orders, respectively, by Brouwer (1953). PH, and PH: had also been called “Nodal sieve-tubes” and “Nodal sieve-elements” respectively by Behnke (1965a). The PH; of Braun may actually be the over-lapping ends of PH» and PHoel. The phloem-glomerulus cells vary in length, and as was observed in the case of the xylem-glomerulus, some of the cells of the phloem groups are shorter than others. In this case, the cells of the PH, order vary from 20p to 60u in length, while those of PH» vary from 60, to 140,. The cells of PH; and PHg have thin walls (about 1p thick) with simple pits that can hardly be seen with a light microscope. Whether the walls are interconnected by cytoplasmic threads (plasmodesmata) or by any 1969 | AYENSU, ANATOMY OF DIOSCOREACEAE 129 other mechanism has not been demonstrated with the light microscope. Dr. Behnke of Bonn University informed me that his electron microscope studies show that plasmodesmata are indeed present in the cells of PH; and PH». His recent publications (Behnke, 1965a, b, c) support his find- ings. It is, however, certain that these phloem cells are specialized and differ from sieve tubes and sieve cells of ordinary phloem tissue. Micro- chemical tests reveal the absence of starch-grains from the phloem-glo- merulus cells; the surrounding parenchyma cells possess starch. The his- tochemistry of the phloem will have to await critical studies. Cleared and stained portions of young and old stems reveal that at the node (Fic. 3) three major vascular bundles (LT) enter the petiole from the stem through the node without joining other vascular bundles, coming through the underlying internode as peripheral vascular bundles. These leaf-trace bundles are V-shape The vascular bundles of the stem axis lying in front of the point of entry of the leaf traces, and those of the inner and outer circles become enlarged and join to form the xylem and phloem glomeruli (X, PH). These glomeruli lie obliquely above each leaf insertion at the same height as the axillary bud (AB). Opposite the outer circle of the vascular bundles in the internode they appear somewhat towards the outside and project into the base of the axillary bud or the lateral shoot. Five cauline vascular bundles leave a glomerulus into the internode above (CB), but only two enter it from below (GB). The latter are the characteristic large vascular bundles which are arranged in the gaps between the three leaf- trace bundles, which lie on the inside of the stem furrows. Hence the five cauline vascular bundles forming the circle are made up of the two vascu- lar bundles from the glomerulus (GB) and the three vascular bundles of the leaf-trace (LT). The vascular bundles of the axillary buds come from the glomerulus directly. Just after they leave the glomerulus, each divides into two (an upper xylem branch and a lower phloem branch), which come from the upper and lower regions of the glomerulus respectively. Occasionally the lower phloem branch subdivides into two with one establishing itself above the xylem branch. DISCUSSION The structure of the xylem and phloem glomeruli in the nodes of the Dioscoreaceae seems to be unique amongst the monocotyledons. Futher- more, the presence of tracheids and the distinct type of sieve elements in the node has considerable implications regarding the evolutionary history of these tissues in the angiosperms. The anatomical studies of the xylem by Bailey and Tupper (1918) showed that the most logical phylogenetic sequence is the derivation of vessels from tracheids in the angiosperms. Cheadle (1943) working with the xylem of monocotyledons confirmed Bailey’s work. In the light of the above theory it is interesting to examine the developmental aspects 130 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 of the tracheal elements in the node of the Dioscoreaceae. The bulk of the xylem glomerulus is made up of tracheids which are considered primi- tive in the phylogenetic sense. Similarly, the cells of the phloem glo- merulus are considered to be of a relatively primitive type (cf. Braun, 1957, and the papers he quotes). It is significant that such a difference can occur in a stem with primitive structures in the nodes and more advanced structures in the internodes. Bailey (1956) stated that “It is now clearly demonstrated that evolu- tionary modification of the xylem of stems and roots is not necessarily closely synchronized with phylogenetic trends in the specialization of the angiospermic flower. Either trend of evolution may be accelerated or retarded in relation to the other.” The above can be extended with a statement that vessel development in an individual part of an organ can be delayed or advanced within that particular part as demonstrated in the node and internode of the Dioscoreaceae respectively. This study demonstrates that in the midst of the complex nodal vascu- lar system lies an orderly and systematic mechanism that permits the transport of assimilatory materials through the stems of the Dioscoreaceae. However, any attempt to gain full understanding of the exact pathway, and therefore, the movement of material through the phloem glomerulus must first confirm the present observations which are based on a recon- struction from serial microtome sections. A more reliable understanding of the pathway will hopefully be gained when the writer is able to study the vascular system of the Dioscoreaceae using the motion-picture analy- sis technique employed by Zimmermann and Tomlinson (1965, 1967). The complexity of the phloem glomerulus in the Dioscoreaceae raises some fundamental questions about the current hypotheses on transport mechanisms in plants. Esau, Currier and Cheadle (1957) summarized the hypotheses as (a) mass or pressure flow; (b) mass flow together with activities of parenchyma cells associated with the phloem that account for the turgor gradients necessary for mass flow; (c) transport of solutes in the sieve tube along protoplasmic interfaces; (d) accelerated solute move- ments in sieve tubes resulting apparently from some special kind of cytoplasmic movement or flow; (e) independent solute movement result- ing from one or more as yet unknown active transfer processes that occur in the sieve element cytoplasm. The unique anatomical characteristics of the phloem glomerulus in this family seem to suggest that perhaps more than one of the above methods is responsible for the movement of assimilatory substances in the Dio- scoreaceae. Arisz’s (1952) suggestion that every substance moves its own way, and that different mechanisms may be involved in translocation should be considered in the light of the anatomical variation in the phloem of this family. Although I have no proof as to the exact function of the phloem glomerulus, it seems likely that rapid translocation is achieved by the numerous cells that form the bulk of the nodal region. 1969] AYENSU, ANATOMY OF DIOSCOREACEAE 131 SUMMARY The complex nodal anatomy which is unique and basically uniform in the Dioscoreaceae, especially in Dioscorea and Tamus, is described. The width of the two masses of tissues referred to as glomeruli is correlated with the phyllotaxy in each species. The xylem-glomerulus is composed of numerous short tracheids of various sizes and shapes which are closely fitted together. The phloem glomerulus, whose construction is essentially that of the xylem-glomerulus, consists of thin-walled cells without visible pitting and sieve areas. Because of the presence of primitive xylem and phloem structures in the nodes in contrast to more advanced structures in the internodes, it is postulated that vessel development in an individual part of an organ can be delayed or advanced within that particular part as shown in the node and internode of the Dioscoreaceae respectively. The peculiar nature of the vascular bundle glomeruli is presumed to have some effect on the rate of fluid transport in the stem. It is suggested that another technique, such as the motion-picture analysis method, should be employed to study further the nodal structure and its relation to trans- location. ACKNOWLEDGMENTS I am very grateful to Drs. P. B. Tomlinson, R. H. Eyde, and H. Robin- son for reading the manuscript. LITERATURE CITED Arisz, W. H. 1952. Transport of organic compounds. Ann. Rev. Pl. Physiol. 3: 109-130. AYENSU, E. S. 1965. Notes on the anatomy of the Dioscoreaceae. Ghana Jour. Sci. 5(1): 19-23. 1966. Vegetative anatomy and taxonomy of the Dioscoreaceae. Ph.D. Thes esis. University of London, pp. 383 [unpublished]. gee I. W. 1956. Nodal anatomy in retrospect. Jour. Arnold Arb. 37: 269- . Tupper. 1918. Size variation in tracheary cells. I. A com- Parison between the secondary xylems of vascular cryptogams, gymno- BEHNKE, H.-D. 1965a. Uber das phloem der Dioscoreaceen unter besonderer i. “thick Phloembecken. I. Zeitschr. Pflanzenphys. 53(2): 97-12 3 ed Uber den Feinbau “gitterartig” aufgebauter Plasmaeinschlusse in den Siebelementen von Dioscorea reticulata. Planta [ Berl.| 66: 106-112. mit besonderer Beriicksichtigung eines neuartigen Typs assimilateleitender Zellen. Ber. Deutsch. Bot. Ges 322. Brouwer, R. 1953. The arrangement of the vascular bundles in the nodes of the Diowcsireaceec: Acta Bot. Neerl. 2: 66-73. 132 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 BurkitL, I. H. 1949. The ontogeny of the stem of the Common Bryony, Tamus communis Linn. Jour. Linn. Soc. Bot. 53: 313-382. CuHeapie, V. I. 1943. The origin and certain trends of ir apse of the vessel in the si ea a Am. Jour. Bot. 30: Coursey, D. G. 1967. Yams. Longmans, London, pp. 230. Esau, K., H. B. canes. & V. I, CHEADLE. 1957. Physiology of phloem. Ann. Rev. Pl. Physiol. 8: 349-374. FALKENBERG, P. 1876. Vergleichende Untersuchungen iiber den Bau der Veg- etationsorgane der Monocotyledonen. Stuttgart. p. 202. Harr, H. 1950. Die Dioscoreaceen-Knoten. Staatsexamensarbeit, Miinchen [unpublished]. Mason, T. G. 1926. Preliminary note on the physiological aspects of certain undescribed structures in the phloem of the great yam Dioscorea alata Linn. Proc. Roy. Dublin Soc. 18: 195-198. ZIMMERMANN, M. H., & P. B. ToMLINson. 1965. Anatomy of the see Rhapis excelsa, I. Mature vegetative axis. Jour. Arnold Arb. 46: 160-178. & . Anatomy of the palm Rhapis excelsa, W. Vascular development in apex of vegetative aérial axis and rhizome. /bid. 48: 122- Hela oF BoTANY SMITHSONIAN INSTITUTION WasHINGTON, D.C. 20560 EXPLANATION OF PLATES 4-12. Longitudinal sections of the stem ee region illustrating the copaetec of the xylem and phloem glomeruli, « 8 PLATE I Fic. 4. Dioscorea hirtifora Benth., showing an example of the meeting se between the phloe Map tale a cells (PH.) of the second order, and a trans- verse section a a gece tu scorea discolor Kunt h, interlacing of xylem glomerulus cells. Arrow points es a transverse section of a vessel (V) just entering the node. PLA sen ne Fie: '6. Rrmcaage? cies gh ana Hochst., exhi oh faa orientation of xylem and phloem glomeruli. Vessel Gstieal (vi glomerulus of the first (PH,) and ee (PH.) orders. Xylem Aare galls (XG), phloem glomerulus cells (PHG). FI Dioscorea multiflora Mart., showing a vessel element (V) and xylem glomerulus cells (XG). PLATE III Fic. 8. Dioscorea —— Schauer, showing transverse sections of phloem glomerulus cells (PH Fic. 9. De composi ita Hemsl. (D. tepinapensis Uline ex Knuth). Arrows pointing to vessel (V), vessel-tracheid (VT) and xylem anmeraiun in transverse section (XG). Fro. 1 a ome elem Dioscorea dregeana (Kunth) Th. Dur. & Schinz, end plate of vessel- fdea (VT) and phloem glomerulus cells (PHG). PLAT so pentaphylla L., she hbo of a vessel-tracheid (VT) and PLATE V 12. Tamus communis L., exhibiting the presence of xylem (XG) and pret (PHG) glomeruli. Jour. ARNOLD Ars. VoL, 50 AYENSU, DIOSCOREACEAE Jour. ARNOLD ARB. VOL. 50 PriateE II AYENSU, DIOSCOREACEAE Jour. ARNOLD ARB. VOL. 50 PLaTE II AYENSU, DIOSCOREACEAE Jour. ARNOLD Ars. VoL. 50 PuaTe IV opt bgt ba tom - ie CY ff he y s& ~ et Fg {* OY Like F ie AYENSU, DIOSCOREACEAE PLATE V Jour. ARNOLD Ars. VoL. 50 AYENSU, DIOSCOREACEAE 138 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ANATOMY OF THE PALM RHAPIS EXCELSA, VII. FLOWERS * N. W. UHL, L. O. Morrow, anv H. E. Moore, Jr. Tur GENus Rhapis is one of a group of six genera in the subfamily Coryphoideae centered in the southeastern United States (Rhapidophyl- lum) and southeastern Asia (Liberbaileya, Maxburretia, Trachycarpus, Rhapis) with a Mediterranean outlier (Chamaerops). These genera are notable for complete apocarpy coupled with an apparently specialized in- florescence (relative to the subfamily as a whole), polygamy or dioecism, and slight to marked morphological distinction between staminate and pistillate flowers. Among them, Rhapis appears to be most highly special- ized in having greater dissimilarity between staminate and _pistillate flowers and a gamophyllous corolla. That perfect flowers may sometimes occur is suggested by the forma- tion of apparently normal seed on an isolated pistillate plant at Cornell University and further by the comments of Tomlinson and Zimmermann (1968). In young stages of pistillate flowers, anthers appear normal but in the mature flowers they are small and do not normally contain pollen. There is a basic similarity between staminate and pistillate flowers in young stages. Differences — functional versus abortive carpels and anthers and elongate staminate corolla tube—are obvious only in later and mature stages. The nature of most palms makes morphogenetic experi- a pot plant in greenhouses. When the present anatomical series is com- plete, it may prove an excellent subject for studies of development and morphogenetic experiments on different aspects of flowering. The purpose of this paper is to describe the anatomy of the two morphological types of flowers in order to continue the anatomical series on Rhapis, to add to a survey of floral anatomy in palms, and to provide the basis for further WOrkK. MATERIAL AND METHODS A partial description of the floral anatomy of Rhapis was previously prepared by one of us (Morrow, 1965) from collections vouchered by * This study was undertaken at the request of and in collaboration with Drs. P. B. Tomlinson and M. H. Zimmermann. We would like to thank them for this appeared in Jour. Arnold Arb. volumes 46 (1965), 47 (1966), 48 (1967), and 49 (1968). 1969} UHL, MORROW, & MOORE, RHAPIS EXCELSA 139 Read 701 and 774. Further study of this material and of new collections (Moore & Uhl 9561, 9562) has resulted in the more complete description presented in this paper. Staminate and pistillate flowers (Moore & Uhl 9561, 9562) at anthesis were cleared and sectioned as described previously (Uhl, 1966). Serial sections were studied in polarized light and by cine- matography (Zimmermann & Tomlinson, 1965), a technique we are find- ing most useful for analyzing flowers where many bundles are present. RACHILLAE As described in a previous paper of this series (Tomlinson & Zimmer- mann, 1968), both staminate and pistillate inflorescences of Rhapis are small panicles with up to three orders of branching. Bract to branch relationship, although somewhat obscured by adnation, reveals a simple monopodial system similar to that described for Nannorrhops ritchiana (Tomlinson & Moore, 1968). Flowers are inserted in irregular spirals (Fics. 1, 7) on branches of the first, second or third orders and on the terminal part of the main axis, these axes being rachillae as defined by Tomlinson and Moore (1968). 2 Fics. 1-17. Fic. 1, portion of staminate rachilla, x 2; Fic. 2, staminate flower in vertical section, x 8; Fic. 3, stamin ah flower, x 4; as as — swt Calyx, X 4; Fic. 5, staminate flower, calyx ved, X 4 6, stam flower expanded, X 4; FiG. 7, portion of pistillate ictal, x - om 8, Distillate flower, x 4; Fic. 9, pistillate calyx, exterior view, X 4; Fic ‘10, pstillate — foteeiae view, x 4: Fic. 11, pistillate flower, calyx remov FI Pistillate flower, vertical section, < 8; Fic. 13, gynoecium, x 8: Synoecium, vertical section, X 8; Fic. 15, carpels in transection, X 8; nh fas ie bead. dorsal view, X 8; Fic. 17, one ports ventral view, X 8. DETAILS: r, bractl 140 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Pistillate inflorescences (Fic. 7) have fewer, commonly shorter branches than staminate, with flowers more widely spaced (1-2 mm. apart). In staminate inflorescences (Fic. 1), third order branches are more common, often longer, and flowers are more crowded (0.5—1 mm. apart), sometimes opposite, or in pairs. Many more flowers are produced in a staminate than in a pistillate inflorescence (TABLE 1). Anatomy. Anatomically all rachillae are similar. Epidermal cells are small with rounded to slightly papillose outer walls. A thin cuticle is present. The cortex is moderately wide and of unspecialized parenchyma cells which increase in diameter centripetally. Some of the cells contain tannins. The vascular complement consists of both large and small bundles. Each larger bundle has one or two large vessels, a single phloem strand, and a fibrous sheath four to five cells wide next to the phloem and two to three cells wide next to the xylem. Smaller bundles have fewer vascular elements; a few may contain only a phloem strand or be com- pletely fibrous. In general these axes differ from the main axis in having less lignified ground tissue and fewer cortical fibrous strands. There is a definite arrangement and orientation of axial bundles in rachillae of many palms. In Rhapis, a transection of a rachilla at any level shows some large central bundles, one or two peripheral groups of smaller bundles, and some scattered fibrous or very small vascular bundles in the inner cortex. This configuration is easily explained in terms of origin of bundles to the flowers. Slightly below and opposite a floral insertion, six to ten axis bundles branch (Fic. 18, fls) to form the bundles supplying the flower. A single axial strand may produce one to four small branches in close vertical succession or in a horizontal plane. Commonly the vertical derivative continues as an axial bundle; however all branches of a bundle may be- come floral traces. The peripheral clusters of small bundles are traces to higher flowers; consequently, the number of small bundles varies de- pending on the proximity to a floral insertion. One or two axis bundles as well as branches from many others extend directly into each flower. The total number of bundles in a rachilla is thus progressively reduced distally (TABLE 1). Bundles in the axis branch fre- quently, providing the numerous traces to flowers. Absolute numbers of bundles are difficult to determine because bundles branch frequently, the levels at which bundles are counted cannot be considered perfectly com- parable, and fibrous sheaths of main strands and branches are often con- fluent. Mere vigor or order of the branch may also affect the number of bundles in a rachilla. However, the number of bundles in floral stalks and organs seems to vary within definite limits. Approximately 6 to 9 bundles are present in staminate floral stalks below the abscission zone, and a larger number (20-25) in pistillate floral stalks. Rachillae are not terminated by flowers. In pistillate branches 4 rounded or pointed projection of the axis extends beyond the flower; some 14 to 16 vascular bundles are present in this reduced tip. Staminate 1969] triangular group of bun stalk; UHL, MORROW, & MOORE, RHAPIS EXCELSA abs, abscission zone; oh wn pe, petal traces; se, gee ra, rachilla. fis, 142 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 rachillae usually end less abruptly, one to seven abortive flowers being present. Bracts subtending these abortive flowers are more prominent than bracts of normal flowers, which are often obscured as the axis and flower increase in size. A difference in growth patterns is suggested in the two types of inflorescences. More branches and more flowers per branch are formed in staminate inflorescences, suggesting that factors affecting branch and floral initiation are more active and that cessation of growth is less abrupt. TABLE I. Flowers and Bundles per cm. of Length in Rachillae , Pistillate rachilla Staminate rachilla 1 cm. intervals, base to apex bundles/cm. flowers/cm. bundles/cm. flowers/cm. base 50 42 1 43 4 36 3 2 40 3 3D 7 Bi 40 + 35 10 - 39 4 Sl 9 5 ot 3 28 15 6 16 1 28 12 7 16 26 13 9 19 14 10 12 13 11 5 7 abortive PISTILLATE FLOWER Bracts. Each pistillate flower is subtended by a bractlet (Fic. 7, br). Bractlets subtending basal flowers on rachillae may be larger than bract- lets of distal flowers which are usually small, crowded between the flower and the axis, and apparent only when flowers are detached (Fic. 7). A small trace, originating as a branch of an axis bundle, is usually present in the bractlet. One or more floral traces may originate from the same stelar bundle from which the trace to the bractlet diverged at a lower level. Morphology (Fics. 7-17). Although considerable connation and ad- nation are present in floral organs, a 3-3—6-3 floral plan is obvious both morphologically and anatomically. Sepals of pistillate flowers (Fics. 8, 9, 10) are connate forming a shallow parenchymatous cup about 1 mm. high with three pointed lobes 1-1.5 mm. long. The three petals (Fic. 11) are also connate for approximately 3 mm., above which the free lobes are briefly imbricate and then valvate reaching an additional length of 1-2 mm. The staminodes (Fic. 12) resemble the stamens in staminate flowers but are smaller. The filaments are linear, adnate to the petal tube for 1 mm., and free above that for about 0.5 mm. In the material studied the reduced anthers did not produce pollen. 1969 | UHL, MORROW, & MOORE, RHAPIS EXCELSA 143 The three separate carpels (Fic. 13) are wedge-shaped with flat ventral sides and rounded and grooved dorsal sides (Fics. 13, 16). Each carpel has a distinct stalk which is fused with the petal-staminode tube for a very short distance basally (Fic. 24). A locule with a single basal ovule occupies the lower half of the carpel (Fics. 12, 14). Distally the style is wide; the upper part is distended abaxially and converges abruptly toward the ventrally situated, conduplicate, tube-shaped stigma (Fic. 28). Thus the styles of these carpels are enlarged and are also histologically specialized, as described below. A single, hemianatropous ovule with a large funicular aril is attached basally in the ventral angle of the locule (Fic. 27). There are two integu- ments which are free for about 1/3 the length of the ovule. The outer integument is six to seven cells wide and increases to about nine cell layers around the micropyle. The inner integument consists of two cell layers and is widened to three to four cells around the micropyle to form a short beak. The inner layer of the inner integument is specialized as an integumentary tapetum. Anatomy (Fics. 18-32). Pistillate flowers appear to be sessile (Fic. 7). Anatomically, however, a very short stalk with a distinct group of floral traces, can be recognized (Fic. 19). As explained above, the major- ity of the bundles of the floral axis originate as branches from strands in the rachilla, one or two of which also extend directly into the flower without branching. The number of bundles supplying the pistillate flower (Fic. 19) is about 23 An abscission zone forms a characteristic feature of floral stalks of both staminate and pistillate flowers (Fic. 18, abs). This zone is distinguished by the absence of fibrous bundle sheaths and by smaller ground paren- chyma cells (Fic. 20) through which bundles can be followed. Generally, in palm flowers, even when organs are connate, the origin of their traces indicates a spiral insertion. This is not apparent in the Sepals or petals of Rhapis. Directly above the abscission zone, most bundles of the floral axis branch at about the same level (Fic. 21) to form about 30 sepal traces. The origin and horizontal divergence of so many bundles at one level results in a collar-like complex in which inner bundles extend radially between outer strands and some lateral fusion of bundles occurs (Fic. 21). Individual bundles may be followed through this complex. Ficure 38 is a radial plot of a single major bundle of the floral axis. The sepal trace (se 1) originating from this bundle branches to form three other sepal traces (Fic. 39, se 2, se 3, se 4) and these bundles in turn branch forming the continuing vertical bundles VB 2, VB 3, and VB 4. Smaller (minor) bundles of the floral axis may produce only a single sepal trace or extend directly into the sepal. Above the sepal complex about 30 vertical bundles form a central group (Fic. 22). Some 30 to 40 petal traces diverge at an acute upward angle (30° to 40°) from these as opposed to a near 90° angle of divergence for sepal traces (Fic. 18). Smaller vertical bundles (Fic. 39, VB 2 and 144 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 VB 3) may extend directly into a petal without branching. Most petal traces, at the level of their origin, contain phloem only and fibrous sheaths of main bundles and branches are often confluent (Fic. 23, pe). At higher levels where traces are separate, a few scalariform xylem elements are present. Sclerenchymatic sheaths of petal traces are thinner walled than those of sepal traces. As in the sepals, a few lateral bundles may branch and a median and two lateral veins extend into each petal tip. 38. 24-27. Pistillate flower, continued. Fic. 24, transection through stalks of the three carpels, outer ring of bundles are petal traces, inner six large bundles supply staminodes, all bundles in carpel bases are provascular, X 18; Fic. 25, transection through petal-staminode tube and three carpels at level of funicular attachments, X 18; Fic. 26, transection showing petal-staminode tube and expanded styles of carpels, x 18; Fic. 27, longitudinal section of one carpel, 35. DETAILS: ar, aril; cs, carpel stipe: pe, petal trace; stm, staminode trace. 1969 | UHL, MORROW, & MOORE, RHAPIS EXCELSA 145 Fic. 28. Three dimensional drawing of one carpel to show vascular supply. Dorsal and ventral bundles labeled, remaining are lateral bundles. Seven lateral bundles are not completed for clarity, ca. 50. For details see Fics. 29-32. About 20 relatively large receptacular bundles, each with a complete fibrous sheath (Fic. 23, central bundles) are present above the origin of the petal traces. Just above this level, considerable reorientation and 146 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 branching of strands takes place. Traces to staminodes (Fic. 24, stm) are formed, often as a central branch of a trifurcating receptacular bundle. Traces to antisepalous staminodes diverge at a slightly lower level than those to the antipetalous ones. The remaining bundles become oriented into three groups, one group representing the supply to each carpel. Fibrous bundle sheaths extend as far as the stalk of each carpel but are absent in the carpel base where all bundles are procambial. Three or four of the bundles in each carpel base are Jarger than the remainder and possess xylem elements which are birefringent. The central of these larger strands extends across the carpel base and distally around the locule to the base of the stigma (Fic. 28, db). Two of the other larger strands remain in ventral positions (Fics. 28, 29-32, vb). Thus a dorsal Fics. 29-32. Series of transections through the base of one carpel drawn with Wild M20 research microscope and drawing tube, to show origin of ovular supply. Ovule traces shaded, bundles with birefringent xylem shown divided, > 60. Derarts: db, dorsal bundle; Ib, lateral bundle; Ic, locular canal; vb, ventral bundle. 1969 | UHL, MORROW, & MOORE, RHAPIS EXCELSA 147 and two ventral bundles can be recognized by size, position, and maturity (Fics, 29-32). Remaining strands form the 20 to 24 lateral bundles present in an irregular ring in the carpel wall (Fics. 31, 32). Four of these are larger and more mature (Fics. 31, 32, lb). Some lateral bundles fuse with others near their upper limits, the major ones extending toward the locular canal (Fic. 28). The ventrals extend slightly higher and the dorsal ends just below the stigma (Fic. 28). The origin of the vascular supply to ovules varies in palms (Uhl, un- published). In Rhapis, a branch from the dorsal bundle, a branch from each of four or five lateral bundles, and a branch from one ventral bundle form a group of strands (Fics. 29-32) which extends into the funiculus (Fic. 28). These strands fuse near the chalaza and the resulting large with traces derived from dorsal and ventral carpellary bundles are not frequent. Such taxa occur in groups usually considered to be primitive, as Magnoliaceae (Canright, 1960) and Nymphaeaceae (Moseley, 1961). STAMINATE FLOWER Morphology (Fics. 1-6). A comparison of staminate and pistillate flowers shows both differences and similarities. Bractlets are alike in both types of inflorescence. Sepals in the two flowers (Fics. 3, 4) are also Similar in shape and size; those of staminate flowers are perhaps slightly less fleshy. Petals are about the same length (4-5 mm.) and are 2/3 to 3/4 connate (Fics. 3, 5). In staminate flowers, however, the petal tube is definitely obovoid or clavate and much less fleshy than that in pistillate flowers (cf. Fics. 8 and 11 with 3 and 5). The diameter of the petal tube immediately below the free lobes is approximately 2 mm. in staminate flowers and 3 mm, in pistillate flowers. Staminate petals are valvate and often incompletely connate, a groove of varying depths showing the limits of each petal. In pistillate flowers, however, the petal-tube is smooth and free lobes are briefly imbricate. Filaments (Fic. 6) are wider in Stamens than in staminodes and bear well-developed, latrorse anthers with dark, tannin-containing connectives (Fics. 33, 37). Three very tiny vestigial carpels are present (Fics. 33, 35, vc). 148 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Fics. 33-37. Staminate flower. Fic. 33, longitudinal section, X 18; Fic. 34 cleared staminate flower, levels of riGuRES 35~37 indicated by one Op sbe beo 5* Fic. 36, numbers, * 10; Fic. 35. transection through base of flower, 3 transection through petal-stamen tube, * 18; Fic. 37, tr cota ES of ae part of flowe <18. DeETAILs: pe, petal traces; se, sepal; st, stamen trace; vestigial t carpel. 1969 | UHL, MORROW, & MOORE, RHAPIS EXCELSA 149 14404 1080- 720+ 360- in microns Length 60 40 240 360 720 Distance from center of axis in microns Fic. 38. Diagram of the radial path of a major betyrs ¥ hime floral axis. woe c, carpel traces; pe, petal trace; se, sepal trace; stm, staminode trace VB 1, ontinuing vertical bundle. Dotted lines indicate any Ghrou bundle sheaths ; are conflue bundles. About three strands remain in the floral receptacle above the origin of the stamen traces. These disappear just below the vestigial carpels. Histology. Histological features in floral organs are sometimes diag- nostic in palms (Uhl, unpublished). In R/apis, tannins are present ran- 150 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 domly in sepals and filaments, near the adaxial surfaces of petals and in all cells of connectives. Fibrous bundle caps are lacking in receptacular, lower petal, and stamen bundles of staminate flowers; in carpels; and in abscission zones of both flowers. It is perhaps significant that there are few, if any, crystals in fleshy sepal bases and petal tubes. The abaxially distended styles are histologically the most specialized parts of the flowers, containing raphides, tannin cells in radial rows, and distal, cap-like layers of sclereids (Fic. 26). DISCUSSION Comparison with vegetative organs. Emphasis in this series on Rhapis has been on vascular pathways throughout the plant. Careful analysis of the flowers shows continuity of bundles from those of the se | Fic. 39. Part of a transection, drawn with the Wild M20 research microscope and drawing tube, to show the continuation of the sepal trace (se 1) diagrammed in Fic. 38; se 2, se 3, and se 4 are sepal traces derived as branches of se 1. Each of these branches to form a continuing vertical bundle (VB 2, VB 3, and VB 4) as indicated, 125. 1969] UHL, MORROW, & MOORE, RHAPIS EXCELSA 151 rachilla to the ovule or stamen. The pattern of origin is a simple one. Bundles of the floral axis, derived as branches of rachilla bundles, branch in turn at appropriate levels to provide traces to sepals, petals, staminodes or stamens, and carpels. This vascular continuity throughout RAapis, which is now completed in the description of floral vasculature, shows a similar pattern throughout every kind of axis on the plant (e.g. seedling, aérial axis, rhizome, inflores- cence axis, rachilla, and pedicel). The same principle of vascular organ- ization is expressed in the flower, but it is somewhat more difficult to recognize here than in the vegetative organs because the floral axis is condensed and the lateral organs are small. Nevertheless we may say that the divergence of traces to sepals, petals, and staminodes or stamens, involving axial continuity, is comparable to the departure of leaf traces in rhizome and aérial stem (Tomlinson & Zimmermann, 1966; Zimmermann & Tomlinson, 1965). This is most obvious when an individual bundle is followed through the floral axis. The radial path resulting from such an analysis is presented diagrammatically in Ficure 38. In addition very short bundles, which may be interpreted as bridges (Zimmermann & Tomlinson, 1965), often link diverging traces with bundles of the re- ceptacular system. More detailed comparison of floral and vegetative vascular pathways must await a more complete understanding of mono- cotyledonous vascular development. Comparison with other palms. Among coryphoid palms, Rhapis may be considered intermediate in specialization. The connation in sepals and petals and corresponding derivation of sepal and petal traces in whorls are evidences of specialization, as is also the adnation of stamens and staminodes to the petal tube. Several features of the carpel are note- worthy. Completely free, stipitate, spirally inserted carpels are con- sidered primitive in palms and angiosperms. However, the large dorsally extended styles and completely closed ventral sutures of Rhapis indicate specialization. The orientation of the ovule is intermediate between the primitive anatropous and the most advanced orthotropous position. The multiple derivation of traces to the ovule from the dorsal, several laterals, and a ventral carpellary bundle suggests laminar placentation (Eames, 1961) and may be a basic pattern in palms. In a preceding paper of this series it was stated that Rhapis has a relatively specialized inflorescence (Tomlinson & Zimmermann, 1968). Similarly it may be said that among the Coryphoideae the flowers are relatively specialized. LITERATURE CITED Canricut, J. E. 1960. The comparative ey a relationships of the Magnoliaceae. III. Carpels. Am. Jour. Bot. 47: —155. Eames, A. J. 1961. Morphology of the ey aa -Hill. N.Y. Morrow, L. O. 1965. Floral morphology and anatomy of certain Coryphoideae (Palmae). Ph.D. Thesis. Cornell Univ. [Unpublished]. 152 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Mose ey, M. F. 1961. Morphological studies of the Nymphaeaceae. II. The flower of eae Bot. Gaz. 122: 233-259. Tomitnson, P. B. & H. E. Moore, Jr. 1968. Inflorescence in Nannorrhops ritchiana fies Jour. Arnold Arb. 49: 16-34 Tom.inson, P. B. & M. H. ZIMMERMANN. 1966. Anatomy of the palm Rhapis excelsa, II. Rhizome. Jour. Arnold Arb. 47: 248-261. & . 1968. Anatomy of the palm Rhapis excelsa, V. Inflorescence. Ibid. 49: 291-306. Unt, N. W. 1966. Morphology and anatomy of the inflorescence axis and flowers of a new palm, Aristeyera spicata. Jour. Arnold Arb. 47: 9-22. ZIMMERMANN, M. H. & P. B. TomLinson. 1965. Anatomy of the palm Rhapis excelsa, I. Mature vegetative axis. Jour. Arnold Arb. 46: 160-180. L. H. Battey Hortortum CORNELL UNIVERSITY IrHaca, NEw York 14850 (Uhl and Moore) A ND RANDOLPH MACON COLLEGE ASHLAND, VIRGINIA (Morrow 7 1969 | MITRA & SUBRAMANYAM, GLYCOSMIS 153 GLYCOSMIS PENTAPHYLLA (RUTACEAE) AND : RELATED INDIAN TAXA | R. L. Mirra AND K. SUBRAMANYAM THE PUBLICATION of a new series, Limonia arborea, by Roxburgh (PI. Coromandel. 1: 60. ¢. 85. 1798) and his providing the plant which he believed to be “Limonia pentaphylla Retz.” (Roxb. loc. cit. t. 84) with a detailed description and illustration, as well as the subsequent discovery of the authentic type specimen of Limonia pentaphylla Retz. by Tanaka (Bot. Not. 1928: 156-160. 1928), has led to some controversy in the , nomenclature of these two species now included in the genus Glycosmis. In the interest of clarity, relevant parts of the earlier works are reviewed in brief. Tanaka (loc. cit.) pointed out that Limonia pentaphylla Retz. is con- specific with Limonia arborea Roxb. and is entirely different from the plant treated by Roxburgh as “Limonia pentaphylla Retz.” He therefore treated Glycosmis arborea (Roxb.) Correa (= Limonia arborea Roxb.) as a synonym of Glycosmis pentaphylla (Retz.) Correa (= Limonia pentaphylla Retz.), and in Botaniska Notiser (1928: 159. 1928) proposed Glycosmis mauritiana (Lam.) Tanaka (= Limonia mauritiana Lam.) for the plant erroneously treated by Roxburgh as “Limonia pentaphylla Retz.” , Narayanswami (Rec. Bot. Surv. India 14(2): 26. 1941) did not agree ; with Tanaka’s view and maintained Limonia pentaphylla Retz. and Limonia arborea Roxb. as distinct from each other; accordingly the cor- rect names in the genus Glycosmis should be G. pentaphylla (Retz.) Correa and G. arborea (Roxb.) Correa, respectively. Brizicky (Jour. Arnold Arb. 43: 88. 1962) upheld Tanaka’s view on the conspecificity of Limonia pentaphylla Retz. and Limonia arborea Roxb. and remarked, “Narayanswami (1941), apparently having overlooked Tanaka’s article on the type of Retzius’ species, came to the conclusion . . . that Tanaka’s interpretation of L. pentaphylla was entirely incorrect. . ..” Brizicky also pointed out that De Candolle (Prodr. 1: 538. 1824), instead of Correa (Ann. Mus. Hist. Nat. Paris 6: 386. 1805), should be assigned the author- ship of these two binomials, G. pentaphylla and G. arborea, since De Can- : dolle made these combinations for the first time in the sense of the Code. However, Brizicky’ s conclusion on their nomenclature is untenable, not being in accordance with the existing Code. Brizicky (Joc. cit., P. 87) is of the opinion that “...... Glycosmis pentaphylla DC. was based on the plant identified and illustrated by Roxburgh as ‘Limonia pentaphylla Retzius’ and only questionably on Retzius’ species 154 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ea eee, . . . Limonia pentaphylla Retz. obs. 5. p. 24? Roxb. 84.’).” Brizicky (loc. cit., p. 89) further argues, Then Gly- cosmis pass: DC., based on Roxburgh’ s plant, not on that of Retzi- us, must be regarded not as a new combination, but as a new name in combination G. mauritiana (Lam.) Tanaka... Since G. pentaphylla DC. cannot be applied to Retzius’ Limonia pentaphylla, the next available name for the latter species is Glycosmis arborea (Roxb.) DC.” In treating Glycosmis pentaphylla as a new name and not as a new combination Brizicky was probably applying the provisions of Art. 72. However, this article is not operative in this case; it is clear from Roxburgh’s treatment of “Limonia pentaphylla Retz.” that he was not describing a new species under a homonymous name, but was only misidentifying Retzius’ plant. hus, the question of De Candolle’s basing the binomial G. pentaphylla on “Roxburgh’s plant — Limonia pentaphylla’ does not arise. Moreover, De Candolle, in making the transfer (Prodr. 1: 538. 1824), gave a direct reference to Retzius’ plant, though with a question mark, “.... - - Limonia pentaphylla Retz. obs. 5. p. 24? Roxb. cor. |. t. 84.” It is evident from above that De Candolle was not certain about the identity of the two plants involved in the confusion. Hence, Brizicky’s argument for bag menclature (ed. 1966) clearly states, “When, on transference to another genus, the specific epithet has been applied erroneously in its new position to a different species, the new combination must be retained for the species *Dr. Brizicky, who died on June 15, 1968, saw an earlier but hardly different version of this paper and, on May 4, 1968, set down the comments which follow. These comments were duly communicated to the authors of this paper, who are pee not agreeable to the arguments placed by Dr. Brizicky. e authors of this paper believe that my view of Glycosmis pentaphylla DC. as a new name, rather than combination, is untenable from the standpo int of the Code, making transfer, gave a direct reference to Retzius’ plant, “Limonia pentaphylla Retz. . 5. p. 24?.. .”’ Curiously, though, applying the Code mechanically [Art. 557], the authors disregard the question mark which follows the complete citation of a ri sumed to be possessed by a taxonomist who publishes a new combination, new status, 1969] MITRA & SUBRAMANYAM, GLYCOSMIS 155 to which the epithet was originally applied, and must be attributed to the author who first published it.” Glycosmis pentaphylla (Retz.) DC., there- fore, must be retained as a new combination based on L. pentaphylla Retz. The nomenclature of the relevant taxa follows: Glycosmis pentaphylla (Retz.) DC. Prodr. 1: 538. 1824, quoad basionym; Tanaka, Jour. Indian Bot. Soc. 16: 229. 1937. ie pentaphylla Retz. Obs. Bot. 5: 24. 1789. Limonia arborea Roxb. Pl. Coromandel. 1: 60. t. 85. 1798. G. pe ee (Roxb.) DC. Prodr. 1: 538. 1824; Narayanswami, Rec. Bot. Surv. India 14(2): 20. 1941; Brizicky, Jour. Arnold Arb. 43: 90, 1962 oe pieces es var. linearifoliola Tanaka, Jour, Indian Bot. : 230. 1937, “linearifoliolis.” G. arborea var. linearifoliola (Tanaka) Narayanswami, Rec. Bot. Surv. India 14(2): 26. 1941, “linearifoliolata.’ Glycosmis mauritiana (Lam.) Tanaka, Bot. Not. 1928: 159. [4 Apr. ] 1928; Bull, , A 2a excl. syn. G. june DC Limonia mauritiana Lam. Encycl. Méthod. Bot. 3: 517. 1792. Limonia pentaphylla auct. non Retz.: Roxb. Pl. Coromandel. 1: 60. ¢. 84. 1798. G. Rnlaresicaty sensu Narayanswami, Rec. Bot. Surv. India 14(2): 12. 1941, xcl. syn. L. pentaphylla Retz. Tanaka proposed this combination in a manuscript sent to the Bulletin de la Société Botanique de France on January 13, 1928. Ina subsequent paper sent to Botaniska Notiser on February 18, 1928, he referred to his etc. There certainly have been cases when because of unavailability of the ig pes and/or ned o of misinterpretation of some taxa the new combinations tur ave bee made xa different from those for whi ey were intende os ceaare, in all these cases the authors of the combinations sincerely believed that their taxa and those the epithets of which were taken as io were identical (conspecific, con- varietal, etc.) parently, there are so ene if any, combinations which : on the epithets of doubtfully conspecific a that there has been no necessity to mention them in the Code, and the ee ee those cases has been left to the good t ‘ “Thus, since De Ca ndolle himself indicated that Limonia pentaphylla Retz. cries t as basionym, there is no reason to r d Glycosmis pentaphyl new combina- tion based on Limonia pentaphylla Retz, On the contrary, pestis pentaphylla, gesting a combination was not that, but a new name. Thus, A. Gray took Malvastrum, the name of De Candolle’s section of Malva L., as the name of his genus without having based the genus on De Candolle’s section. yeaa & Ree 156 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 proposed combination “Glycosmis mauritiana (Lamk.) Tanaka in Bull. Soc. Bot. Fra with basionym and a few more synonyms; but this latter paper appeared in print first. Tanaka’s awareness of this changed situation is evident from his subsequent citation in the Journal of Botany (68: 226. 1930). Glycosmis mauritiana var. andamanensis (Narayanswami) Mitra & Subr., comb. nov. G. pentaphylla var. andamanensis Narayanswami, Rec. Bot. Surv. India 14(2): 16. 1941. Publication of Glycosmis mauritiana var. andamanensis Tanaka (Jour. Indian Bot. Soc. 16: 229. 1937) was not valid, for a Latin description was not given. Glycosmis mauritiana var. insularis (Kurz) Tanaka, Jour. Indian Bot. Soc. 16: 229. 1937 G. arborea var. insularis Kurz, Jour. Bot. 14: 38. 1876, pro parte. ‘ G. pentaphylla var. insularis (Kurz) Narayanswami, Rec. Bot. Surv. India 14(1): 20. 1941. Glycosmis mauritiana var. fuscescens (Kurz) Mitra & Subr., comb. nov. G. trifoliata var. fuscescens Kurz, Jour. Bot. II. 5; 37. 1876. G. pentaphylla var. fuscescens (Kurz) Narayanswami, Rec. Bot. Surv. India 14(2): 20. 1941. We could examine only one sheet present at CAL (King’s Collector 1884, acc. no. 74984) from which the diagram given by Narayanswaml (Joc. cit. page 21, fig. 5) was drawn. We agree with Narayanswami in treating this as a distinct variety. Noses cymosa var. linearifoliola (Tanaka) Mitra & Subr., comb. G. cyanocarpa var. aft ea Tanaka, Jour. Indian Bot. Soc. 16: 229. 1937, “linearifoliol G. cymosa var. ecestbiia Narayanswami, Rec. Bot. Surv. India 14(2): 32. 1941. DOUBTFUL TAXON Glycosmis pentaphylla var. latifolia (Kurz) Narayanswami, Rec. Bot. Surv. India 14(2): 20. 1941 We are doubtful whether this plant needs a distinct rank. The lone sheet examined at CAL (Helfer 525, acc. no. 74981) has only two twigs mounted on it, the inflorescence being altogether lost. The leaf 1969 | MITRA & SUBRAMANYAM, GLYCOSMIS 157 character of the right-hand specimen appears to be that of the Glycosmis mauritiana var. andamanensis while the left-hand one is closer to G. mauritiana var. insularis. BOTANICAL SURVEY OF INDIA 76 ACHARYA JAGADISH Bose Roap Catcutta-14, INDIA VoLuME 50 NuMBER 2 JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY B. G. SCHUBERT EDITOR T. G. HARTLEY Cc. E. WOOD, JR. APRIL, re a ae ay & ACO scenes! MAY 9 - 09 THE JOURNAL OF THE ARNOLD ARBORETUM Published quarterly by the Arnold Arboretum of Harvard University. Subscription price $10.00 per year. Volumes I-XLYV, reprinted, are available from the Kraus Reprint CorPo- RATION, 16 Fast 46rH Street, New Yorx, N.Y. 10017. Subscriptions and remittances should be addressed to Miss Dune A. PowWELL, ARNOLD ARBORETUM, 22 Drv1 nity AVENUE, CAMBRIDGE, Massa- CHUSETTs 02138. CONTENTS OF NUMBER 2 Vascutark ANATOMY OF MOoNOCOTYLEDONS WITH SECONDARY GrowTH — An Intropuction. P. B. Tomlinson and M. H. Zimmermann ... 159 teeeeer ASPECTs OF REPRODUCTION IN SaurautA. Djaja D. Soejarto ........ ipsa dled AN Erin Forest 1x Purrto Rico. 6. Aéptat Roots. A. M. Gill . 197 ae is: Root, AND Ranerwonm Reuationsuies. Walter H. - Lyford ee 8, lg Gives as as ts a Se TURE. Richard A. Howard 225° JOURNAL OF THE ARNOLD ARBORETUM VoL. 50 Aprit 1969 NUMBER 2 VASCULAR ANATOMY OF MONOCOTYLEDONS WITH SECONDARY GROWTH — AN INTRODUCTION P. B. TomMLINSON AND M. H. ZIMMERMANN ARBORESCENT PLANTS with secondary growth from a vascular cambium represent a small minority of monocotyledons. Nevertheless their sig- nificance is out of proportion to their abundance because they exhibit a growth habit comparable to that of familiar dicotyledonous and gymno- spermous trees. Monocotyledonous secondary vascular tissue, however, unlike that of other trees, includes discrete vascular bundles. Earlier botanists in their study of palms and other arborescent monocot- yledons devoted considerable attention to those few forms with secondary vascular tissues and it is largely on the efforts of these nineteenth-century anatomists that our present knowledge is based. It was therefore inevita- ble that in our own studies of palms we should follow this earlier tradition and turn our attention to monocotyledons with secondary growth, espe- cially as we had found that early work on the vascular system of monocot- yledons had been incomplete and furthermore had become misrepresented in modern textbooks (Tomlinson & Zimmermann, 1966). In the present article we review the vascular anatomy of monocotyledons with secondary growth in order to provide the background for our observa- tions which will be published separately. Such a lengthy and independent introduction is justifiable because few botanists have first-hand familiarity with these plants and the literature about them is old and not readily available. Nevertheless our introductory review is necessarily very selec- tive, because the early literature is extensive and some of it is no longer very informative. We have largely cited those articles which contributed significantly to knowledge about these plants. Reviews of the subject do already exist. Those by Cordemoy (1894) and Cheadle (1937) are lengthy. Cheadle’s paper is relatively recent and accessible, but it is re- stricted to a study of small tissue samples. General organization, growth habits, and overall distribution of vascular tissues are not discussed in it. This is somewhat in contrast to the approach adopted by earlier genera- tions, as indicated by Cordemoy’s review, and reflects the way in which anatomists have lost sight of the plant as a functioning whole in their 160 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 preoccupation with histological detail. It is with the object of re-instating the approaches of botanists concerned with overall aspects of growth and construction that our own studies have been undertaken. TAXONOMIC DISTRIBUTION The occurrence of aborescent genera in certain families of monocoty- ledons which possess secondary vascular tissues is indicated in TABLE 1. We treat these genera in a very broad sense but appreciate that some, notably Yucca (e.g. Trelease, 1902) have been subdivided. Two systems of classi- fication are compared, that of various authors in Engler and Prantl (1930) and that of Hutchinson (1959) in order to emphasize that the taxonomic distribution of these plants is far from certain. We do not intend to dis- cuss the evolutionary relationship between arborescent and herbaceous monocotyledons, but we suggest that anatomical investigation is likely to contribute significantly to a resolution of these problems. Of the listed genera, Beaucarnea, Cordyline, Dasylirion, Dracaena, and Yucca consist of species ranging in size from trees to shrubs and rhizomatous herbs. The genus Aloé is predominantly herbaceous, but includes a few aborescent TABLE 1. Taxonomic distribution of larger monocotyledons with secondary vascular tissues. ENGLER and PRANTL REPRESENTATIVE GENERA HUTCHINSON AMARYLLIDACEAE AGAVACEAE subfamily Agavoideae Agave, Furcraea tribe Agaveae LILIACE subfamily reais tribe ceae Yucca tribe Yucceae tribe Nolineae ron alte (Nolina) tribe Nolineae Dasylirion tribe Dracaeneae c oii, Dracaena tribe Dracaeneae (Pleomele subfamily Asphodeloideae LILIACEAE tribe Aloineae 0é tribe Aloineae Pen Ya RS Sec esi eNO tribe Lomandreae Lomandra, Xanthorrhoea XANTHORRHOEACEAE tribe Calectasieae Kingia PSUR SSH Date INN Notte me eee ne No TRIDACEAE IRIDACEAE subtribe Aristeinae Aristea (Nivenia), Kiattia, tribe Aristeae Witsenia 1969] TOMLINSON & ZIMMERMANN, MONOCOTYLEDONS 161 species, notably A. bainesti and A. dichotoma, Agave and Furcraea do not really fit a strict definition of a tree although some species achieve mas- sive proportions. The same seems true of the Xanthorrhoeaceae although there is little information about their size, growth habit, and the extent of secondary tissue except in the work of Floresta (1902). The iridaceous genera are listed, although they are little more than shrubs, because sec- ondary tissue is extensive and has been well described (e.g. Adamson, 1926; Scott & Brebner, 1893). On the other hand, we have omitted many monocotyledons which possess a limited amount of secondary growth but are otherwise essentially herbaceous. These include a number of genera in the Liliaceae, like Aphyllanthes, Veratrum and others in Hutchinson’s Agavaceae. Fleshy rhizomes with secondary tissues, as in the Dioscorea- ceae, are also disregarded. Vascular tissue which is by definition secondary may be quite common in other, unrelated, herbaceous monocotyledonous families [e.g. Bromeliaceae, Krauss (1948); Musaceae, Skutch (1932); Zingiberaceae, Chakraverti (1939) |] where it seems to be associated with root insertion. However, before any major evolutionary significance can be attached to secondary cambial activity, we must attempt to under- stand it from a developmental point of view. MORPHOLOGY Growth habits (Fics. 1-11). Growth form is quite diverse although it can be seen to depend on a common pattern of development. Leaves are linear, usually rigid, often thick and fleshy. They are rarely distinctly petiolate as in some smaller species of Cordyline and Dracaena. Axes are made up of short, often very congested internodes. In slow-growing plants this results in the characteristic terminal tufts of leaves or, if the main axis is very much shortened, in the basal rosette which characterizes the Agave-habit (Fic. 7). Branching is usually sparse; the reason for this is discussed below. In Agave and Furcraea the vegetative axis may be unbranched so that the plant is monocarpic. Otherwise the rosette in these plants is propagated by basal and usually stoloniferous suckers. Stoloniferous shoots are not usually present in other genera but they are common in herbaceous relatives (e.g. Sansevieria). The habit of most arborescent monocotyledons is quite tree-like, and some may even be mistaken for a dicotyledonous tree by a superficial observer, as noted by Wright (1901). However, some species of Dracaena, especially those in its segregate genus Pleomele, look more like shrubs with their much- branched crown and fine twigs (Fic. 8). It is evident that shoot diameter on a single plant is not entirely dependent on the amount of secondary growth. In smaller and much branched species variation in crown diameter is considerable and seems related to the vigor of the shoot. Basal, erect shoots are thickest and most vigorous; distal horizontal shoots are narrow and least vigorous. We have noted a range in primary shoot diameter of 6 to 30 mm. in Pleomele. Some of the simpler growth forms can be looked upon as juvenile stages 162 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 fff ( Ml i fi My | SIO P| WA Da aa i i} d AN a rs — (N ) Yj ie fh t : Nb 1 | . / WY ww, NY Y/ WA Wy i) VWNAN Ww si a Ree eee ee RNG Ry 7 NA Aly j y avid WW . Lay / a Sy NN NS’ Q\\\y) Ne) i). ANA, NY WWM y ee ) LAK \i 4 Ziw Growth habits in arborescent er eee with secondary 1-4. Beaucarnea recurvata. Fics. 1-11. thickening (all to approximately same scale). Fic 1969] TOMLINSON & ZIMMERMANN, MONOCOTYLEDONS 163 in the development of the larger forms, which are fixed permanently. De- velopment of a large Beaucarnea (Fics. 1-4), for example, begins with a rapidly growing main axis which remains unbranched for several years. Leaves may be long persistent so that they clothe the axis of quite tall specimens. Many species of Yucca do not develop much beyond this stage (Fic. 10). A link between the specialized rosette of Agave and this juve- nile habit is provided by a number of species of both Agave and Furcraea with relatively tall stems (e.g. F. longaeva illustrated in Engler & Prantl (1930) p. 419). Otherwise, normal development of the tree form con- tinues with branching, the loss of leaves from the older stem parts, thick- ening of the base of the stem, and development of a fissured bark. The evolutionary relation between ontogeny and phylogeny is suggested by Cordyline in New Zealand. Cordyline indivisa can be equated with the unbranched juvenile stage of C. australis (Fic. 9) and in turn the low rosette of C. pumilio with a younger stage still. A disproportionate thickening of the base of the stem characterizes mature plants (Fics. 4, 5, 13) and has probably led to some exaggerated statements about their longevity. Speculations about possible great age have particularly centered around Dracaena draco. The early literature about this is summarized in the paper of Wossidlo (1868). Perhaps the most famous individual tree in this respect was the specimen of Dracaena draco of Orotava on Teneriffe, described by Alexander von Humboldt (1850). Its historical record goes back to the fifteenth century. But estimates that it dated back to the period of the building of the pyramids (4,500 years) are probably exaggerations, especially in view of the known rate of growth of Dracaena reflexa (Wright, 1901). In 1799 the famous tree of Orotava had reached a height of about 70 feet and a circumference of 48 feet at the base of the trunk. A hurricane destroyed it in 1821. There is no certain method of telling the age of a specimen in the absence of planting data. A more meaningful time scale is given by a specimen of Beaucarnea recurvata (Fic, 13) in Fairchild Tropical Garden which is 25 feet high, 19 feet in circumference at a height of 2 feet and yet is known to be not more than 50 years old. Rates of growth otherwise appear not to have been determined for any of these plants. Inflorescences are always terminal. On unbranched axes they are large and very conspicuous as in Yucca and Xanthorrhoeaceae (Fic. 11) and €ven on young specimens of Beaucarnea. They reach massive proportions in Agave and Furcraea. In the much-branched forms flowering is usually simultaneous on all or most distal shoots and renders the tree very conspicuous. In temperate species flowering is seasonal as in Cordyline Plants of successive ages to show development of massive trunk. 1, Unbranched sapling 5-6 years old. 2-3, Early wahiarkonaue of branch system in older stages. 4, Mature specimen in flower. Flow begins in saplings of the size a in Fig. 1 hte it is ei branching 5, Aloé dichotoma. 6, Cordy australis. sp. 8, mele (Drac aena) reflexa. 9, Cordyline indivisa 7, Aga 10, Yucca aloifolia. tt. ssshieohned quadrangulata. 164 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 IG. 12. Specimens of _Cordyline australis, growing in their natural habitat near ana New Z Beaucarnea recurvata, ay Dies ma Nee he ‘ety aise “| a pie < : ? ope, ae 9 te \ * A etl oe 4" a ® ® - * é = J ~ os tone S a'78e @e cep o2® @@s ve Otic ©. 8's ae8 8 Ge. ess s ay . 8s atta “en at Ys G23 ¥ ucca aloifolia, big aies eae of stem eaeety low ae pe Nap 38 show 7“ primary y ( sue. Growth rings ar evident in the secondary tissue particularly if nae illu ustri oe is viewed pee a distance C. Leaf siLT)t sige Do radially. Fics. 24, 25. Dracaena fragrans, pete se otis of ro t 33. A small amount of secondary tiss ssue is present. [Note the position of endodermis show- ing that cambial activity began in places inside it, in other places outside of it.] 7 ss i=) Q a8 er always thickest in the region of insertion of lateral roots, and suggested that secondary thickening is initiated in this region. It is quite obvious that the somewhat conflicting observations of differ- ent workers have a rational explanation in terms of growth and the factors which influence cambial development and activity. The problem has to be studied by following the origin and subsequent growth of adventitious roots in seedlings of different age, as Wright (1901) suggested. Wright 174 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 also made the observation that the cambium originates in the pericycle of the very short hypocotyl, thence spreading upward into the stem and downward into the root. Further observations of this kind are needed to establish a clear understanding of secondary growth in roots of Dracaena. THE RELATION OF PRIMARY TO SECONDARY GROWTH The earlier studies on growth and development of these monocotyledons were carried out at a time when the understanding of plant growth in general was at a very primitive stage. It was also inevitable that theories of plant growth were dominated by concepts derived from studies of dicotyledons, and some of the earliest interpretations of monocotyledonous growth made unfortunate comparisons between monocotyledons and dicot- yledons. To this early period belong a series of studies concerned with the relation between the secondary “thickening ring” and the meristematic tissues of the shoot apex proper as in the investigations of Karsten (1847), Schacht (1852), Nageli (1858) and Sanio (1863). It seems that these studies were based on examination of single sections cut in transverse and longitudinal planes and that no attempt was made to trace the distribution of developing vascular bundles and the 3-dimensional relation between primary and secondary growth. We will show in a later article that this kind of investigation is crucial to the understanding of this relation. One of the features of arborescent monocotyledons which captured the interest of earlier workers was the apparent continuity between the secondary meristem and the meristematic tissues of the crown. Some authors considered these two meristems to be discontinuous (e.g. Scott & Brebner, 1893). This discontinuity is also implied by Millardet (1865) who gave measurements of the distance below the apex at which the secondary meristem could be first recognized. This varied from as little as 3 mm. in Yucca aloifolia to as much as 22 cm. in Dracaena marginata. On the other hand many authors regarded the two meristems as continuous (e.g. Wossidlo, 1868; Lindinger, 1908). Hausmann (1908) reviewed the extensive literature on this topic and himself supported the latter point of view, concluding in fact that the distinction between the two meristems was rather artificial. In a developmental sense this is true, because establish- ment and activity of secondary tissue is dependent upon growth of the primary meristem. Nevertheless, earlier authors have often adopted a very dogmatic point of view, largely in an effort to establish whether the secondary meristem originated in tissue which had completed its matura- tion or not, and was therefore, by definition, truly “secondary.” A similar dogmatic preoccupation which is also largely a semantic one, was with the level, in a radial direction, at which divisions which initiated the secondary meristem occurred. The problem was to decide whether there was a region in the monocotyledonous stem, to which the term “pericycle” could be given. This is entirely an artificial concept, since in most monocotyledonous stems, cortex and central cylinder each ends where the other begins. A true understanding of the development of that region 1969] TOMLINSON & ZIMMERMANN, MONOCOTYLEDONS 175 of the stem in dicotyledons for which the older term “pericycle’’ was devised has been forthcoming only in recent years (Blyth, 1958). The term pericycle has no application in monocotyledonous stems (Carano, 1910). In terms of the overall distribution of the monocotyledonous cambium, one factual error does deserve comment. Réseler (1889) and apparently some earlier authors stated that the cambium does not extend into the leafy zone of the shoot. This is manifestly so untrue a generalization, whatever may have been the situation in the material on which it was based, that it is not surprising that it was soon corrected (e.g. by Corde- moy, 1894). The presence of functioning leaves, the traces of which must cross the cambium and secondary tissues, does raise interesting physiologi- cal and developmental questions to which we will return in a later article. One reason for the conflicting reports on these topics which appears in the literature was that many authors failed to appreciate the variability in the time of appearance of the cambium and its vigor, which in turn seems largely to depend on the vigor of the shoot. We have already com- mented upon the variation in vigor expressed in the different diameters of shoots in one plant. This variation extends to the secondary cambium and may depend largely on the type of shoot. Seedling axes, for example, initially produce secondary tissue very actively. This activity declines on distal branches. Newly released buds, either below inflorescences or decapitated shoots, are dependent on an active production of secondary tissue in the early stages of growth in order to establish vascular continuity with the parent axis. In view of this variation it is not surprising that reports by early authors conflict, since they are probably based on com- parison of shoots in different positions and of differing vigor. COMPARATIVE INVESTIGATIONS A few authors have been concerned with the relation between those monocotyledons with secondary growth and those without. Notable are Mangin (1882) and Petersen (1893). Chouard (1936) was concerned with the same topic, but his interpretations of monocotyledonous growth are not easy to comprehend. Petersen studied a number of monocotyledons which together represented a wide variety of families and growth forms. He came to the conclusion that in the group as a whole there was a con- tinuous series with all intermediate steps, from those, like the orchids with no trace of a secondary meristem, via those in which one is briefly active, as in the Bromeliaceae, to the continually active cambium of Dracaena which permits unlimited growth. Mangin (1882), on the other hand, was concerned with the way in which adventitious roots develop and establish vascular continuity with the conducting tissues of the parent axis. Adventitious roots arise in a meristematic region (couche dictyogéne) between cortex and central cylinder. This meristem also gives rise to a plexus of vascular tissue (réseau radicifére) which connects conducting tissues of root and stem. 176 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The extent of this plexus varies in different kinds of monocotyledons. Mangin considered that in some arborescent monocotyledons, like Agave, this meristematic region remains active throughout the life of the plant. In others, like Dracaena and Yucca, the root meristem is replaced by the secondary meristem. When more is known about the factors which stimulate and maintain an active cambium in monocotyledons it will be possible to approach the topic on a comparative base. Nevertheless Man- gin’s contribution to anatomical literature remains a notable one. CONCLUSIONS It is obvious from the previous pages that a reappraisal of this subject from first principles is needed. We hope to present in future articles the results of studies which to a large part resolve much of the conflicting literature. In particular we will describe the course and developmental pattern of the primary vascular bundles, the constructional relation be- tween primary and secondary vascular bundles and demonstrate how the initiation and activity of the secondary meristem is dependent upon shoot growth. These will be related to growth of the shoot system as a whole. LITERATURE CITED ApaMson, R. S. 1926. On a anatomy of some shrubby Iridaceae. Trans. Roy. Soc. S. Afr. 13: 175-1 BiytH, A. 1958. Po segs sf primary extraxylary stem fibers in dicotyle- dons. Univ. Calif. Publ. Bot. 30(2): 145-231. 23 pls. BraNNER, J. C. 1884. The course and cities of the fibro-vascular bundles in palms. Proc. Am. Philos. Soc. 21: CANDOLLE, A. P. pe. 1813. Théorie raSaUd ib de la botanique. viii + 527 pp. Déterville, Paris. Carano, E. anges Su le or secondarie nel caule delle Monocotiledoni. . Roma 8: 2 Coats aie e oats Growth rings in a monocotyl. Bot. Gaz. 72: 293-304. Cuakraverti, D. N. 1939. The occurrence of fugacious cambium in the rhi- zome va Curcuma longa Linnaeus. Philipp. Jour. Sci. 69: 191-195. CHEADLE, V. I. 1937. Secondary growth by means of a thickening ring in cer- tain monocotyledons. Bot. Gaz. 98: 535-555 CHOoUARD, P. 1936. La nature et le réle des formations dites ‘‘secondaires” dans Védification de la tige des Monocotylédones. Bull. Soc. Bot. Fr. 83: 819- bee cae H. J. pe. 1893. Sur le réle des tissus secondaires 4 réserves des Meneses arborescentes. Compt. Rend. Acad. Paris 117: 132-134. . 1894. Recherches sur les Monocotylédones 4 accroissement secondaire. Dp. 108. 3 pls. Thesis. Paris. DEsFONTAINES, R. L. 1798, Sur l’organisation des Monocotylédons, ou plantes a une feuille seminale. Mém. Acad. Sci. Paris. 1: 478-502. Dupetit-THovars, L. M. A. 1809. Essais sur la végétation. 1. Sur l’accrois- sement en diamétre du tronc des Dracaenas, quoique Monocotylédones. pp. 10. Arthur Bertrand. Paris. 1969| TOMLINSON & ZIMMERMANN, MONOCOTYLEDONS 177 ENGLER, A., & K. PRANTL. 1930. Die natiirlichen Pflanzenfamilien. Ed. 2. 15a. Wilhelm Engelmann, Leipzig. FLoresta, P. La. 1902. Struttura ed accrescimento seg ig sia fusto di “Xanthorrhoea”. Contr. Biol. Veg. Ist. Bot. Palermo 3(1): 208. HAUSMANN, E. 1908. Anatomische ee an Nolina Sort Hems- ley. Beih. Bot. Centralbl. 23(2): Humpotpt, F. H. A. von. 1850. Pot of nature, or contemplations on oy sublime ee of creation (English ed.). xiv + 452 pp. Henry Bohn. Bienes, 1 1959. The families of flowering plants. Vol. II. Monocotyle- dons. ed. 2. Clarendon Press, Oxfor Karsten, H. 1847. Die Vegetationsorgane der Palmen. Abh. Akad. Wiss. Ber- lin 1847: 73-236. Kny, L. 1886. Ein ste ts zur Entwickelungsgeschichte der ‘Tracheiden’. Ber. Deutsch. Bot. Ges. 4: 267-276. Krappe, H. G. 1886. Das gleitende Wachsthum bei der Gewebebildung der GefinenBansen. vii + 100 pp. 7 pls. Gebriider Borntraeger. Berlin. (See . H. Scorr in Ann. Bot. (Lond.) 2: 127-136. 1888.) Krauss, B. H. 1948. Anatomy of the vegetative organs of the pineapple, Ananas comosus (L.) Merr. I. Introduction, organography, the stem and the lateral branch or axillary buds. Bot. Gaz. 110: 159-217 Linpincer, L. 1906. Zur Anatomie und Biologie der Monokotylenwurzel. Beih. Bot. Centralbl. 19: 321-358. . 1908. Die Struktur von Aloé dichotoma L., mit er allge- sie Betrachtungen. Beih. Bot. Centralbl. 24(1): 211-2 ————. 1909. Jahresringe bei den Monokotylen der Sk ee Na- a Wochenschr. Jena. N.F. 8: 491-494 Manan, L. 1882. Origine et insertion des racines adventives et modifications corrélatives de la tige chez les monocotylédones. Ann. Sci. Nat. Bot. 14: 216-363 Merusg, A. D. J. 1961. The Pentoxylales and the origin of the Monocotyledons. Proc. Nederl. Akad. Wet. C. 64: 543-559. 1965. Angiosperms — past and present. /mu: Advancing Frontiers of Plant Sciences 2: pp. 228. MENEGHINI, G. 1836. Ricerche sulla struttura del caule nelle piante Monoco- tiledoni. 110 pp. 10 pls. Minerva, Padua MILiarpet, A. 1865. Sur l’anatomie et le développement du corps ligneux dans les genres Yucca et Dracaena. Mém. Soc. Sci. Nat. Cherbourg 11: 1-24. Mrrect, C. F. B. pe. 1809. Nouvelles recherches sur les caractéres anatomiques et pliysiclogicuste qui distinguent les plantes monocotylédones des plantes dicotylédones. Ann. Mus. Hist. Nat. Paris 4-86. ———. 1843. Recherches anatomique et physiologiques sur quelques végétaux monocotylés. Ann . Nat. Bot. II. 20: 5-31. . 1845. Suites ei mre anatomiques et phy coakien sur quelques végtiaus Sagas Ann. Sci. Nat. Bot. III. 3: 321-3 Mont, H. von. 1824. De palmarum structura. In: K. F. P. von Martius, His- toria Naturalis Palmarum 1: pp. I-LII. 16 pls. ———.. 1849. On the structure of the palm stem. Rep. Roy. Soc. 1849: 1-92. ————. 1858. Ueber die Cambiumschicht des Stammes der Phanerogamen und ihr Verhiltniss zum Dickenwachsthum desselben. Bot. Zeit. 16: 185-190, 193-198 178 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Morot, L. 1885, Recherches sur le péricycle. Ann. Sci. Nat. Bot. VI. 20: 217- 309. Miwcn, E. 1938. Untersuchungen iiber die Harmonie der Baumgestalt. Jahrb. Wiss. Bot. 86: 581-673. NAGELI, C. 1858. Ueber das Wachsthum des Stammes und sie ota bei den Gefasspflanzen. Beitr. Wiss. Bot. Heft 1. pp. 1-156. Pls. PARTHASARATHY, M. V., & P. B. Tomitnson. 1967. Anatomical one of meta- phloem in stems of Sabal, Cocos and two other palms. Am. Jour. Bot. 54: 1143-1151 PETERSEN, O. G. 1893. Bemaerkninger om den gr ONS staengels Tyk- kelsevaext anatomiske Regioner. Bot. Tidsskr. 18: ~124. ROsELER, P. 1889. Das Dickenwachsthum und die Dedede der secundaren Gefassbiindel bei den baumartigen Lilien. Jahrb. Wiss. Bot. 20: 9 8. Russow, E. 1882. Ueber den Bau und die Entwicklung der Siebroéhren und Bau und Entwicklung der secundiren Rinde der Dicotylen und Gymnospermen. Sitzungsber. Naturforsch. Ges. Univ. Dorpat. 6: 257-327 SANIO, C. 1863. Vergleichende Untersuchungen iiber die Zusammensetzung des Holzkérpers. Bot. Zeit. 21: 357-363; 367-375; 377-385; 389-399; 401- 12 412. ScHACHT, H. 1852. Die Pflanzenzelle. Berlin (original not seen). Scnoute, J. C. 1902. Uber Zellteilungsvorginge im Cambium. Verh. Akad. Wet. Amsterdam. Afd. Natuurk. sec. 2. 9: 1-59. . 1903. Die Stammesbildung der Monokotylen. Flora [Jena] 92: 32-48. . 1918. Uber die Verastelung bei monokotylen Baumen. III. Die Vera- stelung nce baumartigen Liliaceen. Rec. Trav. Bot. Néerl. 15: 264-335. Scott, D. H. 1889. On some recent progress in our knowledge of the anatomy of plants. ne Bot. (Lond.) 4: 147~ G. BREBNER. 1893. On ae secoudany tissues in certain monocotyle- Ann. Bot. (Lond.) 7: 21— citeaee G. 1964. The nature 7 eee wood. IX. Anomalous cases of reaction mension Austral. Jour. Bot. 12: 173-184 SkuTcu, A. F. 1932. Anatomy of the axis of banana. Bot. Gaz. 93: 233-258. STRASBURGER, E. 1884. Das botanische Practicum. ed. 1. xxxvi + 664 pp. Gus- tav Fischer. Jena Tumann, K. V. 1964. (See the discussion following the paper by A. B. War- drop, p. 451.) 7m: The formation of gg in forest trees, M. H. ZmmMER- MANN, ed., ee Press, New Yi ToMLINSoN, P. B. 1964. Stem Sova in arborescent monocotyledons. Im: M. H. ZIMMERMANN, ed. The formation of wood in forest trees. pp. 65- 86. Academic Press, New York M. H. ZIMMERMANN. 1966. Vascular bundles in palm stems — their bibliographic evolution. Proc. Am. Philos. Soc. 110: 174-181. : 7. The “wood” of monocotyledons, Bull. Int. Assoc. Wood Anatomists. 1967(2): 4~24. TRELEASE, W. 1902. The Yucceae. Ann. Rep. Missouri Bot. Gard. 13: pp. 133. P Ch WILSON, B. F, 1964. Structure and — of woody roots of Acer rubrum L. arvard Forest Papers 11: pp. 1 WossipLo, P. 1868. Ueber Wachsthum _ Structur der Drachenbaume. Jahrb. Re alsch, Zwinger, Breslau 1868: 1969] TOMLINSON & ZIMMERMANN, MONOCOTYLEDONS 179 WricuT, H. 1901. Observations on Dracaena reflexa Lam. Ann. Roy. Bot. Gard. Peradeniya 1: ' ZIMMERMANN, M. H., & P. B. Tomuinson. 1968. Vascular construction and de- velopment i in the aerial stem of Prionium (Juncaceae). Am. Jour. Bot. 55: 1100-1109. 1969. The vascular system in the axis of Dracaena fragrans (Agavace sa), I. Distribution and development of primary strands. Jour. Arnold Arb. 50: in the press. [PBT] [M.H.Z. | FAIRCHILD TROPICAL GARDEN HARVARD UNIVERSITY 10901 OL_p CUTLER Roap CaBoT FOUNDATION MriamI, Fiorwa 33156 PETERSHAM MASSACHUSETTS 01366 180 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ASPECTS OF REPRODUCTION IN SAURAUIA Dyaja D. Soryarto ! THE GENUS Saurauia is a widespread tropical member of the Actinidia- ceae with representatives in both the Old and the New World. The American range of distribution extends from Central Mexico to southern Bolivia, through Andean South America. According to a recent study by Hunter (1966), 22 species occur in Mexico and Central America, and m present study indicates that 49 species are represented in South America. The genus is not represented in the West Indies, and there are no records of its occurrence in the Guianas or Brazil. During the course of field work in southern Colombia in 1965, I ob- served that some individuals of Saurauia tomentosa (H.B.K.) Sprengel have flowers with sessile stigmas, in contrast to the flowers with long styles (5-7 mm.) of individuals commonly held to be characteristic of the spe- cies. Later herbarium studies indicated that several other South Ameri- can species are similar to S. tomentosa in this respect. To be certain that such a phenomenon had not previously been described in Saurauia, I have searched the literature and found that nothing con- clusive has ever been published. There are several references, however, to the reproductive system of Saurauia. Gilg (1895) and Gilg and Werder- mann (1925) described the flowers of Saurauia as hermaphroditic to polygamo-dioecious. Brown (1935), who observed the flowering pattern of S. subspinosa Anthony, an Asiatic species, noted that the ovary de- velopment in this species lags behind the development of the anthers by about five days, suggesting that cross-pollination may be dominant. Hunter (1966) mentioned that some species in Mexico and Central America have flowers with “aborted” pistils. Killip (Jour. Wash. Acad. Sci. 16: 570. 1926) referred to the flowers of S. micayensis Killip as uni- sexual, while Benoist ( Bull. Soc. Bot. France 80: 334. 1933) described the flowers of his S. Aypomaila as staminate. A few field workers have noted the existence of “male” and “female” plants in some species of Saurauia. Lorenzo Uribe Uribe, for example, noted the peculiarity in S. isoxanthotricha Busc. (L. Uribe U.’s collection number 4802): “Este pie, que crecia cerca a mi No. 4801, no tenia sino flores femeninas.” (This tree, which grew close to my No. 4801, had only female flowers. ) The flowers of Saurauia are borne in a thyrsiform inflorescence, con- sisting of a peduncle, rachis, and axillary scorpioid cymes arising in 4 spiral pattern. Each cyme is borne in the axil of a bract. The flowers are *The author is currently engaged in the revision of the South American species of Saurauia. 1969 | SOEJARTO, REPRODUCTION IN SAURAUIA 181 actinomorphic, pedicellate, each subtended by a bract and two lateral bracteoles; basically, the flowers are pentamerous and are usually de- scribed as bisexual or hermaphroditic. To the best of my knowledge, there is no true “male” or “female” plant; in other words, there is no true sexual dioecism in Saurauia. The observations discussed in this paper were made to obtain more conclusive evidence about the reproductive system and its operations, and to suggest the implications for evolution in the South American species of Saurauia. This paper is the basis for more detailed studies on the breed- ing systems of the group which are in progress. MATERIALS AND METHODS The present investigation has been based primarily upon data obtained from herbarium specimens. Initially, the work consisted simply of sort- ing specimens with reproductive parts into long- and short-(obsolete-) Styled groups. The next step was examination of the pollen grains (their morphology, size, and fertility) of individuals in each of the two groups. Pollen fertility count was obtained either from open flowers or from ma- ture flower buds. The best results were obtained by boiling the flowers or flower buds (sufficiently mature) to obtain the anthers for maceration. Boiling restores the dried material to a natural texture, which makes dis- section and measurement of the floral parts more accurate. The pollen grains were mounted in glycerine jelly and stained with cotton blue dis- solved in water. All pollen fertility counts reported here were obtained by using a Wild M20 phase contrast microscope, with bright-field il- lumination with or without a green filter. Percentage numbers were based upon a count of between 100 and 500 pollen grains on a single preparation. From two to five samples were prepared from one individual. Stamen counts were made for taxonomic purposes. More important to this study, however, was to ascertain whether or not stamen number has any significant relation to floral dimorphism. All counts were made by boiling the (mature) flower buds, since counts based upon open flowers may be inaccurate, as some stamens may have aborted or others may have been broken and fallen during the process of drying and handling. Measurements of style length were made mainly from open flowers and/or fruits, since the styles are persistent in Saurauia, Style length is not reduced much by drying, so boiling was only occasionally necessary. When neither open flowers nor fruits were available, measurement was made from the mature flower buds. This is a valid and reliable substitute, as will be obvious from the following discussions. OBSERVATIONS Analysis of data. I have examined all species from South America for my taxonomic revision, but; due to lack of data, only species with sufficient representation are included here for discussion. These are Sau- 182 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Individuals Scored Individuals Scored Distribution of bistribution of ? N 2 Stamen Number i Stamen Number jai ; RW °%50 100 160 200 : Q 0 To 20 #30 40 3 15 Ou S Y BE > pat rt 2 Y Yy . Yy 2B! = ie vs i £E “ts & 0 i 104 sgt a a* Be 3 re aS Bink z rat a3 ADO 4 of on 1 0 we > 9 : Yi; & © f Style par 5 ao Li rs Length a ao (ces ) oe 2% . 2 » 5} 7 -) 4 or 2 oO mo fp Se Bis a3 va Z | a & & y/ 4 Individuals Scored tyle Ld Ld canat : ' 2 3 5 6 71 ees 2 Distribution of oo Stamen Number pod abage bao Fag Stamen Num 300 40 <% wo P od Lan A +e ial Uj, ws. i= ® g nl a o De Fruit not recorded N e Cf - al et A » é S o a 3 a, Pollen Fertility 0% Fruit recorded Fruit not recorded . : Ficures 1-4, Graphic representation of the distribution of long- and short- s aed son fepiciented by herbarium collections, together with histogram dis- tribution o e stamen numb . Sample examined for each species consists ‘of short- or long-styled forms in more or less equal numbers. rauia bullosa Wawra, S. brachybotrys Turczaninow, S. excelsa Willdenow, S. Humboldtiana Buscalioni, S. tomentosa (H.B.K.) Sprengel, S. omich- lophila R. E. Schultes, S. putumayonis R. E. Schultes, and S. ursina Triana & Planchon. Data for each species, such as presented in TABLE 1 for S. bullosa, and in Taste 2 for S. omichlophila, have been converted into graphs, Fics. 1-8. Of the eight species, seven show a definite correla- tion between low pollen fertility (or absolute pollen sterility and a long- Table 1. Saurauia bullosa Wawra POLLE STYLE COLLECTOR i 2, je . FERTILITY LENGTH he veesond FRUIT oS ALTITUDE (%) (mm.) m. Soejarto 496 _ — §.5 _ + Aug. 2900 Soejarto 1504 — 6 — + Aug. 3000 Soejarto 1533 - — 6.5 _~ + Aug. 2700 Soejarto 1472 + 0 5 160 — Aug. 3100 Soejarto 1336 - - 7 — + Aug. 2900 Soejarto 495 + 0 6 115 + Aug. 2900 Soejarto 1015 + 0 5 127 + July 3200 Soejarto 1478 + 0 5.5 220 ~ Aug. 2900 Soejarto 1595 + 0 7 160 Sept. 3000 Soejarto 500 + 0 5 125 ~ Aug. 2900 Cuatrecasas 20805 + 0 6 100 + Apr. 3100 Cuatrecasas 20414 + 1 4.5 85 - March 3200 Cuatrecasas 23316 + 2 5 70 Nov. 3000 Jorge Castro 78 + 0 5 140 - Apr. 3400 . Uribe Uribe 5278 + 0 55 100 + July 3000 nes Mexia 7599 + 10 5 140 - Aug. 3000 Soejarto 1496 on 75 0 175 Aug. 3100 Soejarto 1435 + 80 0 225 _ Aug. 3100 Soejarto 1484 + 85 0 150 _ Aug. 3000 Soejarto 1491 a 93 0 164 - Aug. 3000 Soejarto 1045 + 80 0 150 - July 3600 Soejarto 1508 + 92 0 127 —- Aug. 3000 Soejarto 1474 + 88 0 149 - Aug. 3000 Soejarto 1335 + 40 0 240 _ Aug. 3200 Soejarto 1473 + 96 0 145 a Aug. 3000 Fajardo G. 81 + 96 0 125 -— Apr. 3400 I. F. Holton 23 + 97 0 a _ Jan. 3000 Cuatrecasas 20997 + 98 1 145 — Apr. 3000 Hitchcock 20888 + -- 1.5 150 - Aug. 3400 L. Uribe Uribe 5328 + 90 0 160 — July 3100 VINVUAVS NI NOILONGOUdAY ‘OLUV[AOS [696T est 184 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Individuals Scored Distribution of 10: Stamen Number 20) 0'$0 100 150 200 250 Individuals Scored 20. Distribution of 10 Stamen Number i] o i 050 100 156 200 250 5 g 3 of 3 nd - Sng 3 > 2% a5 4% Yt © 5 oO iy Se & > or a Pear ners aq i ef: tS Styl Yj, 6 Wa ran te} 2 3 4 6 6 a (mn. ) Distribution of Stamen Number Distribution of Stamen Number Pollen Fertility (25-)65-97 % Fruit not recorded Pollen Fertility 0-10(-70) % Fruit recorded Style Style 3 . Lengt 8 ps Length . (mm. ) oe 2 3 4 5 6 7 (mm. ) Ficures 5~8. Graphic representation of the distribution of long- and short- styled forms represented by herbarium collections, together with histogram distribution of the stamen numbers. Fic. 5, Saurauia brachybotrys Turcz.; Fic. 7, S. Humboldtiana Busc.; Fic. 8, S. Sample examined for each species consists of short- and long-styled forms in more or less equal numbers. styled condition, and between high pollen fertility and a short-(obsolete-) styled condition. This correlation breaks down in S. omichlophila, where both long- and short-styled plants have high pollen fertility. Plants of each type occur in approximately equal numbers within a sampling col- lection of each species, which may reflect the distribution in the natural - populations. Another significant correlation is that specimens bearing fruits have been recorded only from plants with long styles. This is cer- 1969 | SOEJARTO, REPRODUCTION IN SAURAUIA 185 tainly not mere coincidence, since all eight species discussed here (and many others for which statistics are not included) show this condition throughout. Morphological examinations from free hand sections of ad- vanced ovaries in short-styled flowers show that these are aborted and they simply “do not grow” (PLaTeE II, Fic. 19). In long-styled flowers fruiting is accompanied by good seed set, except in several individuals where seed set is poor, notably in Soejarto 1043 (S. tomentosa). Short or long condition of the styles is not in any way correlated with high or low number of the stamens. As is obvious from Fics. 1-8, and from TaBLes 1 and 2, the distribution of the stamen number is continuous throughout the population, regardless of the style length. From the mea- surements of flower parts (data not included here), it also appears that a short- or long-styled condition is not correlated with the size of the flowers. From field observations and from herbarium records, there is no indi- cation of any particular flowering and fruiting season among the South American species of Saurauia. Flowering is usually associated with the wet months of the year. In most species, however, flowering and fruiting are continuous throughout the year, although fruiting is recorded in the herbarium collections (at least, in the eight species under discussion) only from February through October. Pollen grains (PraTe I). All eight species have 3-colporate pollen, which is oblate spheroidal (cf. also Erdtman, 1952). Several individuals of S. excelsa have 3-colpate, prolate pollen grains. Most of the South American species that I have examined have oblate spheroidal pollen grains, although occasional prolate pollen is also present. No single pollen shape is restricted to a particular species. This condition applies only to fertile pollen grains, where the cell content stains uniformly with cotton blue and appears light to dark blue with bright-field illumination. The cell wall is smooth, with no observable wall sculpturing (at least with the present processing technique). The (fertile) pollen grains are binucleate at the time of anthesis (PLATE I, Fic. 14; cf. also Brewbaker, 1967); the generative cell is ellipsoidal or spindle-shaped, and the vege- tative cell is roundish. The vegetative cell usually does not take acetocar- mine stain so well as the generative cell. The binucleate condition of the grains may be seen (with cotton blue stain) in a sufficiently mature flower bud, prior to anthesis, and it is assumed that this condition indt- cates that the pollen is fertile. The sterile pollen grains, on the other hand, have no fixed shape or any orientation. They may be lenticular, roundish or simply irregular in shape, but, lacking contents, do not stain. The cell wall usually is shriv- elled. Some roundish pollen grains have minute dark granules within. The size of the pollen varies, and no serious attempt has been made to measure size variation species by species. I am convinced, however, that pollen size is not taxonomically significant. Pollen size variation is always present in any preparation from a single plant, and size variation 186 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 between species is very slight. According to Erdtman (loc. cit.) the pollen size of S. Prainiana Busc. from Pert is 18.5 20 microns (oblate spheroidal), and that of a species from Bolivia identified as S. brachy- botrys Turez. (probably a misidentification, since S. brachybotrys occurs only in southwestern Colombia) is 19 % 15 microns (subprolate). Ac- cording to my rather crude measurements, fertile pollen grains vary in pollen grains from a short-styled plant and those from a long-styled plant; nor between sterile pollen grains from a long-styled and from a a eis plant. Androecium and gynoecium (Prater II). There is no pollen dimor- phism in Saurauia, in the sense of two types of pollen grains differing morphologically and correlated with floral dimorphism. The correlation in most cases is straightforward: long styles and low pollen fertility (ab- solute sterility) vs. short styles and high pollen fertility. The term long used here is relative, depending on the individual species involved. Spe- cies with large flowers (3-5 cm. in diameter), such as S. bullosa and S. tomentosa, have long styles 5-7 mm. long, and short styles 1-2.5 mm. long, whereas species with smaller flowers (0.5—1 cm. in diameter), such as S. pseudoleucocarpa Busc. and S. micayensis Killip, have long styles 3-5 mm. long, and short styles 0.5-1 mm. long. Styles less than 0.5 mm. long are considered to be obsolete. The ovary of the American species of Saurauia is mostly five-carpellate, but in some species, e.g. S. yasicae Loes., S. peruviana Busc., and S. leuco- carpa Schlecht. may be three- to five-carpellate or, in Saurauia sp. (a new species from Bolivia to be described by me), five- to seven-carpellate. Each style of the long-styled flower is surmounted by a capitate stigma. The stigmatic surface is either roundish or cordate (1-2 mm. broad in S. bullosa), covered by minute papillae. The size of the stigma — and for that matter of all other floral parts — varies with the size of the flower. At the time of anthesis, the stigmas turn dark brown and become sticky. This condition lasts, in S. bullosa, for four to seven days. On the other hand, the styles of a short-styled flower are tipped by simple stigmas which are non-papillate, and according to my field observations, there is no change in color or stickiness during anthesis. Pollination. It appears from field observations that pollination in Saurauia is promiscuous. Most flowers have persistent green sepals an white petals (free for most of their length, but coherent at the base, falling as a unit with the stamens’). Occasionally, some species (e.g. S. #50- xanthotricha Busc.) have both white and pink flowers, but I have never seen species with only pink flowers.* The stamens, with ‘white filament and * As a result, flowers examined after anthesis are often described as “unisexu al *S. Conzattii Busc. from Mexico has red, beautiful flowers (Schultes, personal comm.). 1969 | SOEJARTO, REPRODUCTION IN SAURAUIA 187 yellow anthers, characteristically form a yellow clump at the center of the corolla. The anthers consist of two thecae, versatile and extrorse at the time of anthesis; the point of attachment of the filament is at the junction of the two thecae, which fork in most cases about two-thirds the _ distance from the (embryonic) base. The versatile anthers and the pale, morphologically unspecialized flowers represent, in a way, an adaptation for wind pollination. There is no definite nectary present in the flower, but nectar-secreting tissue is found inside at the base of the corolla, partly hidden by the stamens (cf. Brown, 1935); also, most flowers have a faint, Sweet scent, which in some species, especially S. omichlophila, is moder- of S. peduncularis, S. omichlophila, S. brachybotrys, and S. chiliantha R. E. Schultes.® Fruit and seed dispersal. The fruit of Sawrauia is a berry filled with numerous small seeds embedded in a mucilaginous pulp. The color of the fruit is green, even when mature, although sometimes there is a purple to purple-red tinge on the green, glabrous pericarp. The sepals are per- sistent, as are the styles. Maturity of the fruit is indicated by an abundance of mucilage, which is rather sticky, clear, sweet and edible. Dehiscence of the fruit is septicidal along the longitudinal sutures, the septa often being membranaceous; the central column and the septa remain intact after dehiscence. In S. budlosa, dehiscence may occur from one to three days after a ripe fruit is detached from the tree (faster when conditions are wet) with little mechanical stimulation. The dehiscence lines start at the apex of the fruit and run gradually towards the base, at the same time discharging, or more precisely, exuding the mucilage which includes the seeds. This is, I believe, the way that the seeds are dispersed naturally, aided by the rain wash. Dispersal by birds certainly is not uncommon. Birds have been seen frequently feeding on Saurauia fruits (common name: moquillo or dulumoco, referring to the mucilage of the fruit). How- ever, the effectiveness of bird dispersal must be further investigated. In all probability, diaspores may not be transported great distances in Sau- rauia; survival is insured by an abundant production of the seeds.’ “Pollen of wind- pollinated plants is sioner characterized by simplicity of structure, and by the small size of the grains (betwee a microns, cf. P. Echlin, Sci. Amer., Apr., old such is the case in Saurauia ane matica R. E. Schultes and S. eeceaenes R. E. Schultes were given their enithets tai of the strong and heavy scent of their flowers . Uribe Uribe (no, 2888) has observed numerous bees visiting the flowers of S. Py mee urauia are usually minute, areolate, dark brown; the testa is fragile (cf Pee I, FIG. 21). That Saurauia seeds are viable for relatively long periods is evident from the following notes, S. kegeliana Schlecht. (1836) was “described from living plants at Halle, Germany, that grew from seeds in soil found about the Toots of plants imported from Guatemala” (note by Standley & Steyermark, Field- iana 24(6): 431. 1949). S. spectabilis Hook. (Bot. Mag. 69: pl. 3982. 1842) was de- Scribed from a “plant raised by Mr. Knight, of the Exotic Nursery, King’s Road, 188 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Geographical distribution. The center of distribution of Saurauia excelsa lies in the Venezuelan Cordillera de Mérida, while that of S. Hum- boldtiana is found in the Cundinamarca region, Cordillera Oriental of the Colombian Andes. There is an overlapping of range between these two species in the Santander region. S. ursina is centered in Antioquia, along the Cordillera Central, and its range overlaps that of S. Humboldtiana and, perhaps, that of S. excelsa as well. S. brachybotrys centers in the Cauca- Valle region, between the Cordillera Central and Occidental, but its range extends north to Antioquia, and south to the Narifio-Putumayo region. The Narifio-Putumayo area is located near the Colombian-Ecuadorian frontier, where the species concentration of the genus is highest. S. budlosa, S. tomentosa and S. omichlophila have their centers of distribution in this region also. S. bullosa and S. tomentosa have the broadest ranges of the South American species. S. putumayonis occurs in the Putumayo re- gion, along the Cordillera of Portachuelo. The four species, Saurauia excelsa, S. Humboldtiana, S. ursina, and S. brachybotrys are not effectively isolated from one another geographically or altitudinally. Although S. bullosa and S. tomentosa overlap geo- graphically with other species, they are effectively isolated from the others altitudinally and are themselves frequently sympatric in their distribution, geographically, ecologically, and altitudinally. S. omichlophila and S. putu- mayonis are effectively isolated from the others, particularly ecologically, and they are spatially allopatric. Cytology. I have examined the meiotic chromosomes of seven of the eight species under discussion: Saurauia bullosa, S. brachybotrys, S. Humboldtiana, S. tomentosa, S. omichlophila, S. putumayonis, and S. ursina (Soejarto, 1969). Chromosome behavior at meiosis in these species appears to be normal, and the chromosome size and morphology are re- markably stable. The haploid chromosome number of all seven species is n = 30. Cytokinesis is of a simultaneous type, and the tetrad arrange- ment is tetrahedral. Chromosome counts were all made from pollen mother cells. DISCUSSION In Saurauia, at least among the species from South America, two kinds of flowers can be distinguished. The differences lie in the size and morphology of the styles, and in the degree of pollen fertility. Anther height and pollen size and morphology appear to be fixed. It is for this reason, perhaps, that the existence of floral dimorphism in Saurauia has passed unnoticed for so long. Most workers on this group considered a short-styled condition to be peculiar to a particular individual or species, Chelsea, England, from seeds imported from the Republic of Bolivia, in 1838.” How- ever, I have attempted several times to germinate Saurauia seeds for cytological studies without success. 1969 | SOEJARTO, REPRODUCTION IN SAURAUIA 189 or a sign of immaturity, and apparently did not appreciate the biological significance of their observations. The present study shows that floral dimorphism does exist in Saurauia, but that this type of dimorphism is not distyly or heterostyly in the true physiological sense of the word, since it appears (at least now) that no incompatibility system is involved. Low pollen fertility (to complete sterility) in a plant with long-styled flowers, and high pollen fertility in a plant with short-styled flowers is a mechanism that promotes outcrossing. In this respect, the flower of Saurauia must be described as functionally dioecious. The short-styled form with high pollen fertility may be considered a functionally staminate plant (the pistil being nonfunctional), while the long-styled form with low pollen fertility (to complete sterility) is a functionally carpellate plant (the stamens being nonfunctional). For those individuals, partic- ularly populations of S. omichlophila, which are truly hermaphroditic (Taste 2; hermaphrodites are characterized by a long-styled flower with pollen fertility 80% or more) within the dimorphic populations, further investigation is needed to demonstrate whether any self-compatibility be- tween the pollen and the stigma of the same flower exists. From the pollen size and morphology, there seems to be no reason why it should not occur. If this is the case, species like S. omichlophila must be described as an- drodioecious. The widespread occurrence of functional dioecy, an outbreeding sys- 6 6 °O Fic 23A-F. Anther orientation in the flower of Saurauia Humboldtiana Buse. "Se pals and petals removed to show details. A, short rt-styled form, bud stage; B, cee styled form, at anthesis; C, mcd form, bud stage; D, long- styled fo orm, at anthesis; E, anther, bud stage; F, anther, at anthesis. As a re- 190 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 tem, is further enforced by the peculiar anther orientation during an- thesis in the American species of Saurauia (Fic. 23, A-F). The end of the anther (the embryonic base) is directed away from the center of the flower as it opens, and the anther rotates about 130° on the filament so that the pollen discharge is directed away from the stigmas. Pore openings and dehiscence of the anther start at the embryonic (morpho- logical) base and “zip” ventrally about two-thirds the length of the thecae. My field observations of this dehiscence and reorientation of the anther during anthesis are confirmed by Hunter’s (1966) interpretation, from histological observations of the vascular trace of the stamen (Hunter interprets the reorientation of the anther at anthesis as 180°, with which I cannot agree). Therefore, in individuals which are truly hermaphroditic, like those in S. omichlophila, self-pollination is averted as much as possible. Prevention of self-pollination among the hermaphrodites is indicated by the position of the stigmas which is well above the surface of the androe- cium. Nevertheless, the chances of self-pollination are rather good. From an evolutionary point of view, the immediate consequence of outbreeding is its capacity for genetic recombination to produce variability or the action of selection and other external forces which direct the evolution of populations (Stebbins, 1950). The greater part of the geno- typic variation within a cross-breeding population is due to segregation and recombination of genic differences which have existed in it for many generations. As a result, in a comparable environment, the outbreeders may show great, more or less continuous, morphological variation, which is an expression of genetic variability from plant to plant. Most species populations of Saurauia, those which are functionally dioecious, are characterized by this type of morphological and environmental continuity. Because of a low selective pressure, variability within a population tends to obscure any clear-cut distinction between populations. The situation is further confounded by a more or less free gene flow between species populations, due to an incomplete isolating mechanism: spatial, ecological, ethological or, perhaps, genetical; this last mechanism must be further in- vestigated. The lack of a complete genetic barrier is demonstrated by the frequent occurrence of natural hybridizations where two or more species populations are in contact or where they are sympatrically distributed. I have collected several natural hybrids of Sauwrauia from southwestern Colombia, where the greatest concentration of species is located, and the hybrid status of at least four of these plants has been confirmed by meiotic irregularities of the chromosomes (Soejarto, unpubl.). Although altitudinal isolation is usually effective, nevertheless some population contact is al- ways present. There seems to be no effective barrier against interspecific pollination in most cases, which is reflected in the relatively uniform floral morphology. Only size variation of the flower, which is conspicuous, exists within as well as between species. Correspondingly, selective pres- sures are relatively weak at the stage of flowering and also at the fruiting, or dispersal stage. On the chromosome level, the differences between species populations appear to be even less significant; that is, as far as Table 2. Saurauia omichlophila R. E. Schultes FL. BUDS POLLEN STYLE STAMEN COLLECTING COLLECTOR OR FLs. FERTILITY LENGTH NUMBER FRUIT DATE ALTITUDE (%) (mm. ) (m.) Soejarto 1493 + 90 5 18 + Aug. 2900 Core 1019 + 95 5 30 + July 2700 Soejarto 1176 + 90 4.5 28 + July 2500 Soejarto 1046 ao 96 4 14 + July 3200 Soejarto 977 + — 4.5 _ + July 3000 Soejarto 1511 - — 4 — + Aug. 2800 Garcia-Barriga 13023 + 85 4 21 — July 2800 Uribe Uribe 3876 + 98 4 25 Sept. 3000 Soejarto 1501 + 90 0 23 Aug. 3000 Soejarto 1509 + 88 1,5 20 Aug. 3100 Schultes 3236 + 95 0 26 - Feb. 3200 Schultes 7560 + 90 0.5 26 March 2900 101 4 96 0.5 37 s July 2700 Schultes 7560A + 88 0.5 ay — May 2900 Soejarto 1502 + 85 0.5 29 - Aug. 3000 Hernandez 79 + 95 0 22 - -- 3000 Soejarto 1598 + 95 0 20 — Sept. 3000 Schultes 20098 + 92 0 21 — June 2800 Schultes 7550 +- 85 0 21 - June 3000 Schultes 7771 + 90 0 26 _ June 3000 Soejarto 914 + 0-80 0 21 _ July 3000 [6961 TI NOLLONGOWdAY ‘OLUV[AOS VIAVANVS T6T 192 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 my present investigations on the cytology of the South American Saurauia show. The inevitable consequence of all this is the difficulty of drawing clear-cut boundaries between species populations, and, consequently, in the delimitation of the species within the genus. It is unfortunate that species of Saurauia are unfavorable subjects for garden experiments be- cause of climatic intolerance, poor seed germination, the large size of the plants and the length of time before they reach the flowering stage. These drawbacks, however, should not discourage workers on the group from continuing their efforts. There are several other things that can and must be done; one of these is more vigorous field work and collecting of her- barium material. The more herbarium collections accumulate, the better we can evaluate and analyze the limits of variation within and between species. Considering the relatively young age of the group, Tertiary (Eocene?; cf. Langeron, 1900, Hollick, 1936), it appears to me that evolutionary differentiation is proceeding in the genus Saurauia. Finally, the realization that functional dioecy is prevalent in Saurauia may further confirm our opinion with regard to the phylogenetic relation- ship of the genus with the closely allied, predominantly dioecious Actinidia. SUMMARY The reproductive system(s) of the following eight South American species have been described and discussed: Saurauia bullosa, S. brachy- botrys, S. excelsa, S. Humboldtiana, S. tomentosa, S. omichlophila, S. putu- mayonis, and S. ursina. As far as the present data show, seven of the eight species appear to be functionally dioecious, and one, S. omichlophila, is androdioecious. The flowers of these plants are dimorphic: a long-styled form with high pollen sterility (functionally carpellate) vs. a short- styled form with high pollen fertility (functionally staminate). Anther height is fixed, and pollen dimorphism related to style dimorphism has not been seen. Although data have been compiled exclusively from herbarium examina- tions (presented here in graphic form, Fics. 1-8), the following observa- tions, based upon field and laboratory studies, have also been briefly de- scribed: pollen grain, androecium vs. gynoecium, pollination, fruit and seed dispersal, geographical distribution, and cytology. The widespread occurrence of functional dioecy may be a useful guide in confirming the phylogenetic relationship between Saurauia and the closely allied, predominantly dioecious genus Actinidia. It is further suggested in the discussion, that the extensive morphological variability is the result of the outbreeding nature of the group, because the immediate consequence of outbreeding is its capacity for genetic recombination to produce variability in the action of selection and other external forces which direct the evolution of populations. 1969 | SOEJARTO, REPRODUCTION IN SAURAUIA 193 ACKNOWLEDGMENTS I am deeply indebted to Professor Reed C. Rollins for his generous suggestions and painstaking criticism of the manuscript. Dr. Richard E. Schultes has very kindly helped me edit the English of the text, for which I wish to express my thanks. I also wish to thank Dr. Carroll E. Wood, Jr. and Dr. Beryl Vuilleumier for several discussions on heterostyly. I have received financial support from the Committee on Evolutionary Biology (NSF grants GB3167, GB7346; principal investigator, R. C. Rollins), Harvard University. Finally, I want to thank my wife, Mariela, for her constant and cheerful encouragement, and for typing the first draft of the manuscript. Any errors or misinterpretations found in this paper, however, are my sole responsibility. LITERATURE CITED BREWBAKER, J. L. 1967. The distribution and phylogenetic significance of binucleate and trinucleate pollen grains in the angiosperms. Am. Jour. Bot. 54: 1069-1083. Brown, E. G. S. 1935. The floral mechanism of Saurauia subspinosa Anth. re Crowe, L. K. 1964. The evolution of outbreeding in plants, I. The angio- sperms. Heredity 19: 435-457. Erptman, G. 1952. Pollen morphology and plant taxonomy. 539 pp. Alm- : : jae Gite, E. 1895. Dilleniaceae — Actintoiinan: Actinidieae and ariel Sau- rauieae. Jn: Engler & Prantl, Nat. Pflanzenfam. III. 6: & E. WERDERMANN. 1925. Actinidiaceae. Nat. era ed. 2. 21: 7. Hoxiicx, A. 1936. The tertiary floras of Alaska. U.S. Geol. Surv. Prof. Paper 1 Hunter, G. E. 1966. Revision of the Mexican and Central American Saurauia (Dilleniaceae), Ann. Missouri Bot. Gard. 53: 47-89. Bie ie M. 1900. Contribution a Vetude de la flore fossile de Sézanne. Bull. c. Hist. Nat. Autun 13: 333-3 Sorsatte, D. D. 1969. Saurauia species and their chromosomes. Rhodora [in pre: . . Sreteties G. L., Jk. 1950, Variation and evolution in plants. 643 pp. Columbia Univ. Press, New York & London. DEPARTMENT OF BIOLOGY Present address: DEPARTAMENTO DE BIOLOGIA Harvarp UNIVERSITY UNIVERSIDAD DE ANTIOQUIA A CAMBRIDGE, MASSACHUSETTS MEDELLiIN, Cotompta, S.A. 194 JOURNAL OF THE ARNOLD ARBORETUM {voL. 50 EXPLANATION OF PLATES PLATE I Ficures 9-16. Pollen grains in Saurauia. Fics. 9, 11, 14, 15, 16, fertile pollen; 9, prolate pores As ins. except FIG. 14 prepared from herbarium samples, stained in cotton blue; Fic. 14 prepared na anthers fixed in Carnoy’s solution and stained in ig cio only a generative cell clearly visible). Fics. 9 and 10 approx. < 400, the others aA xX 1600. PLATE II big view of the stigmas at anthesis. Fic. 21, seeds. All to the same scale . FIG. 19, and all from S. bullosa. Fic. 19 photographed from dried flowers (boiled he eee all others from material fixed in Carnoy’s solution. Scale in FIG. 19 is Jour. ARNOLD Ars. VoL. 50 PLaTE I SoEJARTO, REPRODUCTION IN SAURAUIA Jour. ARNOLD ArB. VOL. 50 Prate II SOEJARTO, REPRODUCTION IN SAURAUIA 1969 | GILL, ELFIN FOREST, 6 197 THE ECOLOGY OF AN ELFIN FOREST IN PUERTO RICO, 6 AERIAL ROOTS ! A. M. GILi IN TEMPERATE REGIONS aérial roots are rare and although they may be found on a few vines they are absent from the trees and shrubs. In the moist elfin forest of Puerto Rico, however, many of the trees, shrubs, vines, and herbs form aérial roots. The tree fern Cyathea and the lowly Selagi- nella also form aérial roots in this environment. Many of the aérial roots hanging freely from the plants are very characteristic of the species while some other species are difficult to dis- tinguish by the characters of their aérial roots alone. In this study some of the distinctive characters of the roots are described and the frequency of aérial root formation on Pico del Oeste is documented. OBSERVATIONS ON THE DISTRIBUTION OF ROOTS IN TOTO The roots in the study area are found in four general habitats: in the soil; immediately above the soil beneath a layer of cryptogams and/or leaf litter; appressed to the trunks and branches of the trees and shrubs; and hanging freely in the air. All the roots in the last three habitats named may be considered “aérial.” Those in the second category occur in a gaseous environment immediately below the forest floor and above the soil. A mat of roots up to five centimeters thick may be formed (Fic. 1) which appears to have arisen not merely by erosion of soil but by the growth of roots out of the soil and over its surface. On steep slopes roots of sufficient rigidity may even grow through the forest floor into the atmosphere. On gentle slopes this achievement has been attained by growth along tree trunks and fallen branches beneath a layer of cryptogams and thence out to the atmos- phere. The roots of many of the vines and of the bromeliad Vriesea are found closely attached to rigid organic surfaces. They are often found beneath a mantle of cryptogams but are also found where such a covering is lacking. This latter type of root may also be considered “aérial’” but the affinity of the roots to their supports distinguishes them from the final group which is the main subject of this paper. The aérial roots to be considered here are those found hanging freely in the atmosphere. They arise above ground and are not closely appressed ‘The first two papers in this series were published in Jour. Arnold Arb. 49: 1968. See: R. A. Howarp, The Ecology of an elfin forest in Puerto Rico, 1. Introduction and composition studies, 381-418; and H. W. BaynTon, 2. The Microclimate of Pico del Oeste, 419-430. 198 cryptogam layer but Fic. 1. Mat of roots immediately below the litter and immediate} ly above the soil. to any surface. They may become anchored in the substrate and undergo considerabie secondary thickening and in such cases have been termed rop” or “stilt” roots by other authors. The aérial portions of such anchored roots may exhibit phenomena different from roots of the same species in the freely-hanging stage — those to be considered here. AERIAL ROOTS OF THE TREES AND SHRUBS Many of the data Sew. to the aérial roots of the trees and shrubs of the area are shown in TAB rigin. Aérial roots ee arise from the undersides of branches and from the main axis of the plant. They are often associated with the for- mation of sprouts (probably arising from dormant buds) and in such cases are found at the base of the sprout where it joins the main stem. This condition was observed in Ocotea, Ilex, Miconia pachyphylla, Calyp 1969 | GILL, ELFIN FOREST, 6 199 TABLE 1. The aérial roots of the trees and shrubs Tip PROPERTIES AT ROOT ORIGIN Max incre- Min. ment Max. Lateral Min. oe —— Max. before replace- Rigidity roots stem i diam. none ment and without diam. bane hee SPECIES (mm.) (cm.) tips Color alignment injury* (mm.) (cm. ) ne Prestoea 17 19 4 pale orange stiff & + 58 45 — montana to pale pink brittle, simple curves Hedyosmum 3.5 80 3 white apex, unbent, — 4 5 + arborescens then lemon, flexible then green nd cotea 36 4 pink to unbent, ~ 12 30 _ spathulata brown flexible Trichilia (1.2) 9 1 creamy brown’ unbent, _ 15. HS ? pallida to pink flexible ex 0.7 5 3 white to unbent, -- 13.5 110 ? sintenisii flexible Torralbasia 0:6: 12 5 orange unbent, a 3 30 + neifolia flexible za 89 2 white apex, unbent, - 13.5 SO + grisebachiana yellow and flexible brown behind Calyptranthes 1.2 14 1 white to unbent, - 2S) Ws + krugii red-brown flexible Eugenia 24 1 white to unbent, — “HO 2m. + borinquensis red-brown flexible Calycogonium 1.55 14 3 bright unbent, — = 2 squamulosum pink flexible Mecranium 9 2. white to unbent, = 2 2 Ss amygdalinum pink flexible iconia pink — 2 3 ? foveolata Miconia 10 2 white to unbent, ae 3s ob + bachyphylla pink flexible Grammadenia 2.0 9 2 white to unbent, = .) oe Sintenisii light brown flexib]l Wallenia 10 10 4 white to unbent, 5 5 1 + yunquensis pink xible Micropholis 27 kT ? white unbent, _ 47 ~ &50 ? garciniaefolia flexible Symplocos 0.8 4 6 white unbent, = — 55 6 + micrantha flex. t 200 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 TABLE 1 — continued Tie PROPERTIES AT ROOT ORIGIN incre- Min. ment Max. Lateral Min. distance Second. Max. before replace- Rigidity roots stem to __ thick. diam. laterals ment and without diam. _ leaves before | SPECIES (mm m.) tips Color alignment injury* (mm.) (cm.) ground Haenianthus 5° 23 6 cream ochre unbent to — 8 15 + salicifolius to brown hang in cluster Tabebuia 2 10 5 creamy lime unbent, = 1 ae rigida to weak flexible yellow Gesneria 0.5 4 7 white to unbent, — 3 3 ? sintentsti tan flexible Psychotria 0.7 7 2 beetroot unbent = 4 11 ? berteriana to pale flexible white Lobelia white to unbent, ~ vf 0 ? portoricensis pale green flexible * + represents presence, — represents absence. tranthes, Grammadenia and Torralbasia. The same thing may occur below the ground with some species of trees in temperate areas. In Massa- chusetts it has been seen in Fraxinus americana: when a tree is cut down and new shoots arise at the base from beneath the soil-surface, new roots may be formed at the junction of the new shoot and the parent stem. No root was found within the leafy zone (distal to the most proximal leaf and with or without a few leafless nodes included) of the trees and shrubs except in Lobelia. In Tasie 1 the minimum distance of an aéria root from the leaf zone has been noted and also the minimum stem diam- eter on which an aérial root has been found. Most of the species have aérial roots very close to leaf zones but not within them. Miconia pachy- phylla was recorded with roots at the junction of the leaf zone and for most of the species aérial roots have been found within 50 centimeters of the leaf zone. The aérial roots are usually found too far away from the leafy zone to determine if there is any association between the root origin and the nodes of the stem. With Miconia pachyphylla (Fic. 2) one root was found at a node and no roots were observed where a definite lack of such an association could be seen. No anatomical observation of the origin of the aérial roots was made but association with a node would suggest develop- ment of a preformed primordium giving rise to the aérial root. Aérial roots may also arise laterally from other aérial roots, a condition discussed in a later section. Tip properties. The maximum diameters of the root tips vary widely between species. Maximum values were taken since they are more dis- 1969 | GILL, ELFIN FOREST, 6 201 2. Roots of Miconia oat ta with droplets of water at the apices. fae eesti of Dr. R.A tinctive than average values. Tip size may decrease with the order of branching, with distance from the point of origin, and in some species served with smaller tips than those on large trees). Prestoea has tips up to 17 mm. in diameter (Fic. 3) while the maximum recorded for Torral- basia was 0.6 mm. The root tips of the other species were of diameters intermediate between these values. The significance of tip size in between- species comparisons is not known, but in the roots of the trees in central Massachusetts at least, tip size within a given root system is an important parameter, and other properties of the root are associated with it, e.g., lateral frecmency: number of protoxylem poles, and other anatomical features, as well as the probability of secondary thickening. Color variations in the tips may be distinctive. Torralbasia roots are often orange in color, those of Miconia spp. usually a bright pink, and those 202 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 rial roots of the palm, Prestoea aioe 8 inhibited mage Fic Aé root te At in the absence of aérial roots of Hedyosmum arborescens. Fic. ve Brox 90m - like cluster of nérial — of Tabebuia rigida, developed as a response to deg oeey iauey. ” Secondaty thickening is bag evident near the point of attachment. Fic. 6. Distally anchored agvial roots of Clusza a showing devel mae of root tips and con- siderable oe thicken of Hedyosmum most often lemon. However, the color of the roots may be considerably muted under some conditions and for many of the species darker environments may cause all color to be lost from the root tips. 1969 | GILL, ELFIN FOREST, 6 203 Most of the newly emerged tips are straight and flexible. The palm root tips, however, may be stiff and rather brittle and they may curve down towards the soil. When the roots are longer their alignment and rigidity may change. The roots of Clusia become long and rubbery and the roots of Tabebuia and Torralbasia tend to hang in clusters. The roots of Ocotea, however, usually maintain the initial direction of growth with some bias downward. The living roots of Hedyosmum arborescens are usually bathed in a gelatinous fluid.* This may hang down from the apex of the tip for two millimeters as part of a drop around and below the apex (Fic. 4). Shortly above the apex the material becomes much thinner and is very thin 4 or 5 centimeters from the apex. Two roots with a considerable drop on their ig ebe apices were stripped of their coating, using two fingers to remove . The volume from both was approximately 2 milliliters. aye experiment was initiated in an attempt to discover how quickly the material covering the root was replaced. After a week there was par- tial replacement of the material. After this period approximately 1 milli- liter of fluid was removed from the three stripped roots. It should be noted however, that similar iy ia (with care not to exert pressure) eventually resulted in root death (R. A. Howard, personal communication). In addition it may be noted that the material does not occur on dead roots nor those with brown, apparently inactive, tips. The gelatinous substance is common on the healthy roots of Hedyosmum but has been found only rarely on the aérial roots of other species in this area. It has been found only twice on the aérial roots of Calycogonium and once on a root of Miconia pachyphyila. Many aérial roots of the latter two species were observed, but in only these cited cases was the gelatinous material seen. On these specimens the material was less gelat- inous and more readily removed than that on roots of Hedyosmum. How- ever, it was not dislodged as easily as a drop of water might be and it was jelly-like in texture. A cut in the aérial roots of three species caused drops of milky exudate to be formed. Such material is common in the leaves and stems of the three families represented and this observation indicates that the lactif- erous system does extend into the aérial roots. The plants concerned are Clusia grisebachiana, Lobelia portoricensis and Micropholis garciniaefolia, members of the families Guttiferae, Campanulaceae, and Sapotaceae, re- spectively. Lateral root formation. Patterns of lateral root development con- tribute to the specific character of many of the aérial roots. The rope-like Clusia roots and the broom-like Tabebuia roots (Fic. 5) are very distinc- tive for this reason. In many cases the lateral root formation from the freely hanging aérial roots appears to be entirely dependent upon injury. With such a stimulus, one to seven replacement tips may arise behind the *This fluid has a high content of algae including diatoms and desmids. Some six Carbon sugars were found in the material but no higher sugars (R.A.H.). 204 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 injured portion (TABLE 1). These tips may arise not only close to the injury but also several centimeters behind it (e.g. Tabebuia and Torral- basia). With the aérial roots of the palm Prestoea lateral roots develop regu- larly without injury (Fic. 3) but have very limited growth. The laterals are short (up to 5 mm.) and pear-shaped. Their bases are narrow but their diameter increases markedly (to 2 mm.) within a short distance and then tapers to the apex. They arise in three to eight regular rows depend- ing on parent root diameter. Some roots may achieve great lengths before any lateral is formed. The length attained is a reflection of the growth rate and the interval between injuries, which in Clusia and Hedyosmum may amount to 2.5 to 3 meters. However, the maximum length without laterals for most of the aérial roots is usually between 4 and 40 centimeters in this area. When a freely hanging aérial root becomes anchored in the substrate, prolific lateral root formation may occur in the subterranean portion. In addition, however, new laterals may arise on the aérial portion with no apparent stimulus from injury (Fic. 6). This contrasts with the develop- ment of lateral roots before anchorage and is evident in the aérial por- tions of anchored Clusia and Ocotea roots. Growth rate. The growth rate appears to be very variable through time as single uninjured roots are found to have variations in diameter and color suggesting growth pulses. Injury, of course, prevents growth and a decrease in length may result. Injury may be environmentally in- duced and desiccation is a probable agent. The growth in length of the aérial roots of ten species was measured as it occurred during a 7 to 11 day period in December, 1967. The growth rates varied from 0 to 2 millimeters per day. This may be contrasted with the rate of growth of the aérial roots of Rhizophora mangle (the red mangrove) in Miami, Florida, which grew up to 7 millimeters in length per day during the months of April and May, 1968 (P. B. Tomlinson and author). In the soils of New England, roots may achieve growth rates of 12 millimeters per day during summer (W. H. Lyford and author). Thus the growth rates of aérial roots in the equable climate of the elfin forest may be regarded as low. Secondary thickening. Many of the aérial roots of trees and shrubs may commence secondary thickening well before they reach the substrate. Examples of this may be found among the roots of Ocotea, Clusia, Miconia pachyphylla, Tabebuia, Haenianthus and Eugenia. In FiIGURE 5 thick- ening of the root of Tabebuia near the point of attachment to the branch may be observed readily. The aérial roots of Ocotea near the base of the plant may become an- chored and considerably thickened. In cross section these roots are oval with the larger axis vertical and the morphological center in the lower half. In Clusia however, similar roots are more nearly circular in cross section and arise from up to 3 meters above ground. 1969 | GILL, ELFIN FOREST, 6 205 AERIAL Roots OF THE VINES Origin. The aérial roots of vines are often found within the leafy zone of the plant (in contrast to those of trees and shrubs). Not all roots are found within the leafy zone but this is a common occurrence. In the species under study here all the aérial roots were associated with a node in one way or another. This is not always the case with vines: in some vine species the roots are apparently formed at random along the stem as in the ornamental Hydrangea anomala ssp. petiolaris and in the native Rhus radicans in Massachusetts. The associations with the node were varied. Marcgravia sintenisii has up to four aérial roots produced in a row parallel to the axis of the stem and running proximally from the leaf base. Gonocalyx, an ericaceous vine, has a similar arrangement of aérial roots but on the distal side of the leaf base. The genus Mikania of the family Compositae has roots formed between the leaf bases or in an axillary position. Psychotria guadalupensis (Rubiaceae) has aérial roots formed just distal to and be- tween the nodes. Such specific positions of origin suggest a regular for- mation of root primordia in these positions as the shoot grows. In Marcgravia at least, it appears that new roots may be formed on TABLE 2. The aérial roots of the vines Trp PROPERTIES Roots Max. oots assoc. number Max. Rigidity Laterals in leaf with per diameter _and without SPECIES zone? nodes ? node (mm.) Color alignment injury? Rajania yes yes 1 0.3. + white flexible, fat cordata wrinkled r yes yes 1 0.1 white nt, os emarginella flexible delicate Marcgravia yes yes 4 0.6 cream unbent, = Sintenisii (2.0) tend to be rigid Gonocalyx yes yes 5 0.3 white flexible, + bortoricensis crinkled Hornemannia yes yes 1 0.3 white to crinkled, + racemosa pale pink flexible to brown Ipomoea yes yes 2 0.5 white weak ian repanda flexibility, curves Psychotria yes yes 4 0.5 cream to flexible, + guadalupensis light crinkly green Mikania yes yes 3 0.5 white to flexible, pachyphylla pale crinkled — 206 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 old parts of the vines where they may be 2 or 3 centimeters in diameter. In this case the tips produced may have different properties from those formed within the leaf zone. In Taste 2 (summarizing the data collected for the vines) the value for tip diameter recorded on older wood is noted in parentheses. Whether these new roots develop from latent primordia formed in association with the leaves is not known. Tip properties and lateral root formation. The tips of the aérial roots formed within or close to the leaf zone are usually very fine. Ap- proximately 0.1 mm. to 0.6 mm. diameter is the range encountered. The larger value in this range was recorded for Marcgravia, which has rapidly tapering aérial roots — from 0.8 mm. to 0.4 mm. over one centimeter of length. In this species aérial roots with a length greater than a few centi- meters have not been found free-hanging in or near the leafy zone. Some of the species were observed to have lateral roots formed appar- ently without injury to the parent. The species in which this was ob- served are recorded in TABLE 2. The aérial roots of vines appear rather fragile in comparison with those of trees and shrubs, both because of their small diameter and the fact that they are often irregularly bent. AERIAL Roots oF THE HERBS Origin. The aérial roots of herbs may be found within the leafy zone in most species. In most cases also the roots are formed at well defined morphological positions. TABLE 3 presents a summary of the data. Selagi- nella and Dilomilis both form roots at the branch junctions. Most of the other species root only at the nodes but internodal roots have been ob- served in Pilea yunquensis. Tip properties and lateral root formation. The roots of the Sela- ginella are green as are the apices of the aérial roots of Dilomilis. In the latter species the region of the tip behind the apex was creamy in color and green only at the apex. A few of the species were observed to have lateral roots formed in the absence of injury and these are recorded in TaBLe 3. DISCUSSION Frequency of formation of aérial roots in species. Some species such as Miconia pachyphylla are usually found with aérial roots but other species have been found to have no aérial roots. The reasons may be that too few specimens have been examined or that they do not in fact ever form them under the conditions experienced in this area. Woody plants such as Cleyera and Ardisia were not seen with aérial roots, but these species are not common on the site. Some of the epiphytes, such as the bromeliad Vriesea, had readily visible roots but these were not observed hanging free of the host. Simi- larly no root of the vine Peperomia hernandiifolia was seen hanging free. 1969 | GILL, ELFIN FOREST, 6 207 TABLE 3. The aérial roots of the herbs Min. Tip PROPERTIES distance Roots leaf Max. aterals . zone at diameter Rigidity and without SPECIES (cm.) nodes? (mm.) Color alignment injury Selaginella QO branch 04 ~~ green flexible - li junctions but wiry, unbent Tsachne yes 0.6 white straight & — angustifolia to pale flexible green Dilomilis 2 branch 1.8 green corrugated, — montana junctions apex, flexible cream behind ilea QO yes 0.3. white to curled, obtusata pink flexible Pilea 0 no 0.2 reddish curled, + yunquensis brown flexible Sauvagesia QO yes 0.2 pale straight & — erecta cream flexible Begonia 1 yes 0.4 white to straight & _ decandra flexible One large Cecropia peltata was examined and found to have secondarily thickened “prop” roots, but no tip was seen above ground and in this Case it appeared that the roots had been exposed by erosion. Some of the grasses and carices had a preponderance of leafy tissue above ground, and very little stem tissue and no aérial root was observed. The scrambling grass [sachne has many stems above ground and in the most humid situa- tions aérial roots are found. The tree fern Cyathea forms aérial roots in some cases but these were not studied. Other ferns were also omitted tom the investigation. The environment and aérial root formation. Two stages may be distinguished in aérial root formation. The first is the production of a Primordium either in association with normal growth and development of the shoot or formed de novo under certain conditions and in older In Populus nigra for example, root primordia are present in the aérial stem but no aérial root is formed. Under the proper conditions of mois- ture and darkness however, aérial roots may be forced into active growth (Shapiro, 1962). However, the “proper conditions” for the appearance 208 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 of aérial roots in various species differ. The aérial roots of Rhizophora mangle in Florida, for example, often show great development in environ- ments where there is always a high light intensity. In the elfin forest light intensities are low and roots often arise beneath a mass of crypto- gams, but the importance of the light factor can only be surmised at present. Mechanical tissue and/or lack of injury may be important in some species as roots are often associated with new shoots, which may cause wounding of the parent shoot as they grow and which are composed of relatively soft tissue. The humid environment may be essential to out- growth of roots as desiccation seems to be an important cause of injury to apices. Lateral root formation. The freely hanging aérial roots of the plants in the study area rarely produced laterals in the absence of injury. How- ever, when these roots enter the soil they branch immediately. Thus there are two types of control to the lateral root formation, an external (environ- ment) and an internal (through injury). In the external environment of the aérial root the high humidity appears to be incapable of inducing lateral root development in many of the species, and some other environ- mental factors such as light intensity and nutrient-environment may be involved. Some plants with aérial roots fail to develop laterals in the absence of injury, although many herbs and vines do not. The major member of the latter group is the palm Prestoea, but its lateral roots are inhibited. One cause of injury seems to be desiccation. The dead apices of the roots sometimes show no signs of physical injury but the rare periods of desiccation seem a likely cause of death. One case of physical injury was observed on a root of Clusia, which appeared to have been chewed. Function of the aérial roots. Aérial roots may enable a plant to spread vegetatively to the surface of another plant or to the soil away from the base of the parent plant. If the roots from the shoot system reach the ground the path that nutrients have to travel will be shortened and this may be an advantage. Vegetative spread from detached portions of a tree is possible as broken branches with no connection to a parent tree or the soil have been seen with new roots and shoots. In an area where trees and shrubs may be pulled over by vines and/or the weight of epi- phytes and water, or toppled on the steep slopes after a little soil erosion, the ability to form aérial roots may constitute a valuable property for survival. The presence of copious quantities of a gelatinous material on the apices of the aérial roots of the Hedyosmum is remarkable. Samtsevitch (1965) has noticed relatively small gel-like caps on the roots of some plants such as Zea mays in soil and artificial media, and considers that they have several important functions including protection of the root apex from mechanical injury, improvement of the root penetration of soil, and pro- motion of root hair growth. In the area of study it seems that protection from desiccation is the most likely function of the material, as death of tips follows its removal (R. A. Howard, personal communication). 1969 | GILL, ELFIN FOREST, 6 209 The anchored roots of Clusia and Ocotea certainly provide support to their parent trunks. In their absence however, subterranean roots may provide the plants with the same stability. Thus the adaptive value of these roots for support is questionable. SUMMARY In the humid conditions of the Puerto Rican elfin forest many freely hanging aérial roots are found on the trees, shrubs, vines, and herbs. Those of the trees and shrubs are not found in the leafy zone of the shoot system and lateral root development in the absence of injury is rare. In contrast the aérial roots of the vines and herbs arise in definite morphological posi- tions within the leafy zone of the shoot system, and more commonly de- velop laterals in the absence of injury. Patterns of lateral root develop- ment may be distinctive, but other properties of the root tips such as color, rigidity, alignment, diameter, and the presence of secretions, may also contribute to the character of the aérial roots of the various species. ACKNOWLEDGMENTS This study was carried out over a period of 15 days at Pico del Oeste in the Luquillo mountains of Puerto Rico. The trip was made possible by a grant to Dr. R. A. Howard, director of the Arnold Arboretum of Har- vard University, by the National Science Foundation (Grant # GB-3975). Excellent housing facilities close to the site were kindly provided by Mr. J. B. Martinson. Mrs. R. J. Wagner helped with the identification and checking of plant specimens. REFERENCES CITED SAMTSEvITCH, S. A. 1965. Active excretions from plant roots and their signifi- cance. Fiziol. Rast. 12: 837-846. [ Biol. Abstr. 48: 1968 SHAPIRO, S. 1962. The role of light in the growth of root primordia i in 7“ stem of the lombardy poplar. In: The Physiology of Forest Trees, ed. K. V. THIMANN, Ronald Press. pp. 445-465. Wesster, T. R., & T. A. SrEEvES. 1967. Developmental morphology of the roots of Selaginella martensii Spring. Canad. Jour. Bot. 45: 395-4 Casot FounpATION, HARVARD UNIVERSITY VARD ForEST, PETERSHAM MASSACHUSETTS 01366 210 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 THE ECOLOGY OF AN ELFIN FOREST IN PUERTO RICO, 7 SOIL, ROOT, AND EARTHWORM RELATIONSHIPS WALTER H. Lyrorp ! ECOLOGICAL STUDIES Of a small area of elfin woodland on the narrow ridge top of Pico del Oeste, a 1000 meter-high mountain on the extreme eastern tip of Puerto Rico where rainfall is about 453 centimeters per year (Baynton, 1968), were made over a period of two years by several investigators. The background for the overall study has been given by Howard (1968). This report points out some soil and vegetation relationships and the possible importance that earthworms and other fauna have on soil genesis. PROCEDURE Soils were examined and described in three trenches, 6 to 15 meters long and about 1 meter deep, dug across the 10 to 15 meter-wide ridge of Pico del Oeste (Fic. 1) at representative sites. Soil horizons were mapped at a scale of 1:12 and a few soil samples collected for approximate deter- Fic. 1. The broad western side of Pico del Oeste. The soil study was along the narrow spine and extended from a location near the right hand side of the photo to the summit. * Field work was carried out during March 1-13, 1967, as a part of National Sci- ence ieee Grant GB:3975 to R. A. Howard, whom I thank for making the study possible. L. Theobald, W. E. Gensel and R. W. and Mrs. Wagner gon laboratory and ns assistance, and A. R. Gill, a photograph, for all of which I a very grateful. 1969 | LYFORD, ELFIN FOREST, 7 211 minations of ignition loss, pH, and moisture content. Weights of forest floor and the epiphyte-soil blanket on tree stems were obtained. Earth- worms were collected and their influence on the soil studied. RESULTS Soil characteristics. On the narrow ridge of the mountain the soil, developed on residuum from fine grained volcanic rock, is wet and has a muck-like surface 25—30 centimeters thick. This mucky surface thins out and disappears as the soil becomes steep. Under the mucky surface there is a gray gleyed horizon mottled distinctly with browns, yellows, and reds. Under this, in turn, lies reddish yellow, massive, plastic, nonsticky clay, in places with a noticeable content of soft, weathered rock fragments. With greater depth the soil becomes redder and more fragments of weathered rock are present. The relationship between the horizons is shown in the scale diagram of the three trenches (Fic. 2). Following is a detailed description of the soil. The terms are those in common use in the United States (Soil Survey Staff, 1951). O1 Forest floor: a continuous cover of recently oo leaves and twigs horizon with up to 25 percent of the surface covered with living green bry- 3—0 cm. ophytes (mostly liverworts) and algae; most ‘sae are fragmented 0 : 1 i _»__§ werens Fic. 2. Scale ees of the soil horizons in three trenches dug across the Narrow spine of the mountain. 02 horizon 0-30 cm. Al horizon 0-30 cm A2g horizon 30-60 cm. B21, B22 horizons 60-90 cm. JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 and lie directly on the surface of the underlying soil or on roots; in ee places under the leaf and twig cover there is a 2—2.5 cm. thick oot floor” consisting of clean coarse and fine roots. Dark brown or very dark brown (7.5YR 3/2 or 10 YR 3/2 wet) mucklike material so well decomposed the original plants cannot be identified, about 40-60 percent organic matter as judged from igni- tion loss; massive in place and well permeated by fine grass-like roots and some woody roots; nonplastic, nonsticky; very large earth- on. be identified from the remains in the soil takes the material out of the peat class. Reddish brown (5YR 4/3 wet) or dark reddish brown (SYR 4/2 with many medium, cre mottles of dark brown (7.5YR 3/2) and olive gray (SY 5 , in some places with mottles of strong brown (7.5YR 5/6) ae yellowish red (5YR 5/6); “silty clay i friable, plastic, nonsticky; many roots; distinct 1 mm. red borders around many of the dead roots; large earth- worms present and many earthworm tunnels; no or few coarse fragments of strongly weathered rock. This horizon is adjacent to the wetter A2g horizon and is at the top of the very steep slopes. As a whole this horizon has a brown color in contrast to the overall gray color of the A2g. The texture feels like a silty clay loam of the northeastern United States but the soil probably is mostly clay Dominantly olive gray (5Y 5/2 wet) with 10-20 percent dark brown (7.5YR 3/2) or reddish brown (5YR 4/3) distinct moderate size mottles ; ; massive; firm in place, plastic, non- sticky; many grass-like roots, many large earthworm tunnels filled with dark brown organic matter. In places this horizon can be divided into a somewhat browner upper portion in which the colors are dark brown (7.5YR 3/2) and light olive gray (SY 6/2) in about a 60-35 proportion with the re- mainder made up of red and black fine mottles. The gray mottling in this upper portion is distinct and as a whole this portion appears to be gray, but somewhat less gray than the lower part. Reddish yellow (7.5YR 6/8, 7/8 wet) with reddish yellow (SYR 6/8), yellowish red (SYR 5/8) and red (2.5YR 5/8) distinct mot- tles; “silty clay loam”; massive; firm to friable and digs out readily with a shovel, plastic, nonsticky, dense and nonporous; in places with up to 5 percent very pale brown (10YR 7/3) or nearly white fine scattered mottles; few 5-15 cm. pieces of saprolite agi with a noticeable yellow or red rind, hard but can be broken with the edge of the trowel; organic matter-filled earthworm tacks are common but fewer than in the A2g horizon. The B21 and B22 horizons are generally yellowish in the upper part, becoming redder with depth. Mottles are distinct and not 1969] B23 horizon and deeper LYFORD, ELFIN FOREST, 7 213 noticeably in a reticulate (network) pattern. The B21 horizon shown in Fic. 2 has more conspicuous mottling than the B22 and was designated as a gleyed horizon (B21g) in the field. It does not, Weak red (10YR 4/4) and strong brown (7.5YR 5/6) “silty clay loam”; massive; firm, dense, plastic, nonsticky; contains up to 60 percent angular, hard but thoroughly weathered rock fragments that break readily by a blow from the edge of a trowel. The overall color of this horizon is reddish and this contrasts strongly with the yellowish colors of the B21 and B22 horizons. Below the B23 horizon at depths ranging from one to several me- ters is red, yellowish red, and reddish yellow thoroughly weathered rock (saprolite) that retains its original stratification but is par- tially fragmented and the fragments can be broken readily. The bedrock probably is a fine-grained volcanic. Analyses. The ignition loss, moisture and pH analyses of soil samples collected from the trenches are listed below. These analyses are approxi- mate and are presented as preliminary data to provide a rough idea of the range of some of the soil characteristics. Moisture to Ignition loss p nearest 5 to nearest 5 Glass percent percent electrode Horizon (Oven dry basis) = (Oven dry basis) 1:1 water 02 (mucky) 270 50 4.4 320 40 4.3 sie 65 mies Al (edge of slope) 90 20 4.7 den 20 es A2g (grey, gleyed) 100 25 4.5 80 20 4.9 — 10 a = 15 a _ 10 ae B22 (yellowish) — 15 = —_— > — B23 (reddish) 60 5 SS 10 4.7 — 5 = 5 oe Although the mucky horizon (02), by feel, seems to be mostly organic matter, the ignition loss analyses indicate that about half is mineral matter. The gray, gleyed A2g horizon, on the other hand, seems by feel to be highly mineral but actually has 10 to 20 percent organic matter content. 214 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The soil is very strongly acid throughout as would be expected for any soil developed from strongly weathered rock under extremely high rainfall. By feel, the soil in the mineral horizons seems to be sandy, but it is probably quite high in clay-sized particles. Clay, in many soils of tropical regions, is aggregated into silt or sand-sized particles and these are not readily dispersed between the fingers by rubbing, or even by the use of dispersing agents in the laboratory. Determination of moisture charac- teristics, however, reveals that a high proportion of the soil is of clay size. This property is indicated in many of the soils sampled in connection with soil surveys currently being made in Puerto Rico. (Soil Survey In- vestigations Report No. 12, August, 1967. U.S. Dept. of Agri. Soil Con- servation Service in cooperation with Puerto Rico Agri. Exp. Station.) Analysis of a soil (S58PR—11~-1) collected about two miles away from the Pico del Oeste site shows that on the basis of moisture characteristics some of the “sand” and “‘silt” particles have clay-like properties. The soil on Pico del Oeste seems to have the requisite properties for an oxic horizon and it qualifies for an Oxisol in the 1965 USDA Soil Classification (Soil Survey Staff, 1960, 1967). It qualifies as a Latosol in the older classification scheme. Water in the soil. The peaty soil of the narrow ridge top is continu- ously wet judging from observations made during the course of the study. The path that bisects the area is always muddy and water-proof foot- wear is required if the feet are to be kept dry. There is a more or less continuous drip of water from the epiphyte-blanketed trees even when the sun is shining. When the trenches were being dug water moved into the excavations pri- marily by seepage rather than by overland flow. This seepage water came into the trenches from all horizons suggesting that the whole soil is waterlogged and not just the upper highly organic portions. For ex- ample, when the trenches were being dug, earthworm tunnels in the B2 horizons were full of water and under a hydrostatic head, and they drained suddenly and conspicuously when exposed by the shovel. Moisture content of the mucky horizon is high and so is that of the mineral horizons. This high content of moisture is evident if a clod of freshly collected soil is left in the sun; almost a whole day of good dry- ing conditions is necessary before the surface of the clod dries. A few observation wells 20 to 30 centimeters deep were made to ob- tain some idea about the fluctuation of the water tables in the mucky surface horizon. In these simple unlined wells, depth to the water sur- face varied considerably from day to day during the period March 1 to April 17, 1967, and there were only a few days when water did not stand within these shallow wells. During the period April 17 to June 15, some water generally was in the wells but the fluctuation from day to day was less. Forest floor characteristics. A forest floor horizon, (01 horizon) about 2 to 4 centimeters thick lies on the mucky horizon (02 horizon). 1969 | LYFORD, ELFIN FOREST, 7 zA3 i 9 MARET-6 West Peak PR 3. Components of the forest floor from an area 30 X 30 cm. square. ABO : Tree leaves in various stages of disintegration are on the right, living plants (mostly liverworts) in the middle, roots on the left. BELow: Entire and fragmented tree leaves 216 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 This forest floor is readily separated into three components; namely, liv- ing plants growing on the fallen leaves (mostly liverworts and algae), fallen tree leaves and twigs, and roots (Fic. 3). Samples were collected from six areas 30 30 centimeters square and the average oven dry weight was determined for each component. Living plants 314 g/m? (2800 pounds /acre) Fallen tree leaves 78 g/m? ( 700 pounds/acre) ts 577 g/m? (5200 pounds/acre) Bryophytes (mostly liverworts) growing on the fallen leaves occupy per- haps 10-25 percent of the area. Some are growing in place; others ap- parently were growing on the leaves while the leaves were still attached to the tree and continued growth after the leaves fell. In addition to the bryophytes many of the fallen leaves are about half covered by a thin layer of green algae. Fallen leaves and twigs completely cover the surface of the soil and no bare areas can be seen. Apparently leaves of these evergreen trees fall one by one throughout the year. Most are entire when they fall but after being on the surface of the soil for a short while show some signs of disintegration: parenchyma is removed and the leaves are broken into fragments (Fic. 3). The amount of organic material (other than roots) above the mucky 02 horizon at any one time is about equivalent to the amount that falls yearly in most deciduous forests in the northeastern United States. Clean woody roots lie in a layer between the fallen leaves and the mucky 02 horizon. This “root floor” is about 2 to 2.5 centimeters in thickness and there are many places where the roots arch away from the surface leaving open spaces beneath, Roots in the soil. Both aérial roots and roots in the soil are common. The characteristics of aérial roots on the study area have been described by Gill (1969), Many of the roots in the soil grow between the litter and the mucky surface and constitute the previously described “root floor.” The roots in this layer are essentially free of adhering soil and are as clean as though washed with water (Fic. 4). Possibly some of these roots are clean because they have never grown in the soil beneath. Gill (1969) observed some roots growing above the soil surface; in fact, some were even exposed to the atmosphere. Woody roots on or just under the soil surface, extend for at least 7 to 8 meters laterally and are well exposed in the path (Fic. 4). On the basis of observations made while digging the trenches it is estimated that 80—90 percent of the woody roots are either just under the fallen leaves in the “root floor” or are within the upper 2 to 10 centimeters of the surface of the soil. These woody tree roots seem to be much the same in overall growth habits as those of forest trees at the Harvard Forest in central Massachusetts (Lyford & Wilson, 1966). ok clue, = < ia me “geet Ore gait i a P = A ‘. 1 NIATA GYOAAT » “LSAUYO / Fic. 4. Woody roots in the soil. Lerr: Layer of clean roots that commonly exists just under the leaf litter. RIGHT: Woody roots in the upper portion of the soil that have become exposed in the path, 218 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Vertically descending grass-like roots are numerous in the gray gleyed A2g horizon and give the soil material a sod-like character. Some of the dead roots in this horizon are bordered by a thin layer of red soil and these vertical “pipes” are noticeable when the soil is examined. These pipes are well known features in many wet soils. Amount of epiphytes and soil on tree stems. A rather large mass of epiphytes and roots grows on the stems and branches of the trees. In addition there is brown soil-like material adhering to the bark and inter- mingled with the roots and green plants that blanket the stem. This brown mucky material is essentially identical in appearance to the brown mucky material that makes up the surface layer of the soil. To obtain some idea of the amount of this soil-like material on the trees all material around a portion (20 centimeters long) of the stem of each of six trees was removed. This was subdivided into portions comparable to those used for the forest floor, namely, living plants (mostly liverworts), roots, and soil-like material. Subdivision was made under water to enable com- plete separation of the soil-like material. Following are the results ex- pressed as grams of oven dry material per square meter of stem area. Living plants 265 g/m? Roots 125 g/m? Soil-like material 112 g/m? Weight of green material on the stems is 265 grams per square meter ~ compared with 314 grams for the leaves on the forest floor. Together green material and roots on the stem total 390 grams per square meter as com- pared with 392 grams for the combined leaves and bryophytes on the for- est floor. In other words, the amount of organic matter (other than that which makes up the soil-like material) on tree stems is not far different from that on the surface of the soil if compared area for area. A few determinations of ignition loss were made on the soil-like ma- terial on tree stems and branches. Most of the material has a 90 to 95 per- cent loss on ignition so its origin probably is from the decomposition of plant material in place. The one sample with an ignition loss of 64 percent may have had a different origin. Probably the soil-like material on stems and branches is largely the result of the decay of plant tissue in place. Some of the material however, is granular and definitely coprogenic. Fauna of various kinds are common under, and in, the plant material which clothes the stem. Millipedes, centipedes, enchytriads, beetle larvae, tiny red ants and sow bug-like in- sects are numerous and a black earthworm 10 to 15 centimeters long, like those in the soil, was found on one tree stem at a height of about 1 meter. This amount of faunal activity raises a question about the origin of the soil-like material. Conceivably a large part of it could originate in the soil and be carried into the trees within the bodies of fauna. In such a case, however, the casts would probably have a rather large content of mineral matter; larger than that indicated by the analyses of the 1969 | LYFORD, ELFIN FOREST, 7 219 Fic. 5. Soil occurs in some trees. Lert: Termite nest in a tree by a roadside within sight of the study area. RicHT: Termite tunnels on a tree stem. W. L. Theobald is used for scale. soil-like material observed on the study area. At lower elevations termites have large nests high in the trees (Fic. 5). They carry a good deal of material into the trees so the process leading to the presence of soil-like material in trees is not unusual. Earthworms and their significance. Earthworms are common in the soil of the study area and were collected from each of the three trenches (Fic. 6). Two species were collected but have not yet been identified. One species is black, about 15 to 20 centimeters long and weighs about 5 grams. It seems to be most common in the upper part of the soil and, indeed, a single specimen was observed on a tree stem about 1 meter above the soil surface where it was well protected by the wet epiphyte blanket. The second species is very large and has a dark gray or dark olive color. It is up to 60 centimeters in length and 10 millimeters in diameter. Each of the earthworms weighs about 30 grams. These large earthworms are com- mon to depths of 50 centimeters and are in all the upper horizons. Their tunnels are conspicuous when filled with dark-colored soil material and especially so in the gray gleyed A2g and the yellowish B2 horizons (Fic. 7) Roughly 1 to 5 percent of the soil mass in the upper 50 centimeters is occupied by earthworm tunnels. Tunnels in the mineral portion of the soil are made by ingestion of soil material rather than by simple pushing aside, and some of the ingested mineral material later is expelled on the surface. Organic matter also is ingested in large amounts. Leaves on the surface of the soil serve as food and some of the disintegration of leaves 220 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ste tee BDRM: Fic. 6. Two species of earthworms from the study area. Apove: Close-up of one of the large olive specimens. BELow: The four large olive pg shown in the upper part of the photo. are up to 60 cm. long and on in diameter. The smaller black earthworms shown in the bottom part of ne atts are up to 20 cm. long shown in Fic. 3 is the direct result of earthworm action. Earthworms probably ingest some of the soil material near the root floor and, in fact, these roots may have their clean appearance because of the earthworms. Puddled masses of earthworm casts 5 to 10 centimeters in diameter are on roots at intervals of about 40 to 50 centimeters, showing that earthworms are active on the surface; and it is possible the open spaces beneath the root floor are the result of earthworm action. 1969] LYFORD, ELFIN FOREST, 7 221 Fic. 7. Dark colored areas in the A2g clods are earthworm tunnels filled with soil of high organic matter content. Scale in the photo. above is in centimeters. Organic matter ingestion is also evident because of the dark color and the high ignition loss of the casts in the tunnels. In the A2g horizon the ignition loss for two samples of earthworm casts from the dark-colored 222 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 tunnels was 65 and 75 percent, whereas that of the soil immediately adja- cent was 10 and 15 percent. A single sample of earthworm casts from a tunnel in the B22 horizon had an ignition loss of 50 percent; that of the soil immediately adjacent was 5 percent. Thus there is considerable mixing of the organic and mineral matter of the soil within the bodies of the large earthworms, and a rather great amount of transportation from one place to another. This mixing and transportation is not rapid enough, however, to cause the soil horizons to lose their identity. In addition to the mixing and transportation there is some segregation of the larger particles because they are too large to be ingested. DISCUSSION The climate and vegetation of the study area very likely is much the same now as it has been for centuries because of the particular location of the Pico del Oeste in respect to the ocean and the constantly blowing easterly trade winds. The soil, therefore, probably represents a kind that has developed and been maintained under continuously wet tropical con- ditions. Soil development processes may have been modified from time to time by addition of volcanic ash, but no known additions of this kind have been made recentl Continuously wet conditions have permitted the build-up of a mucky mineralization of the organic matter. A considerable amount of organic matter is returned to the surface of the soil each year not only from the dead leaves of trees but also from epiphytes that grow on the leaves. The B horizon in this soil seems to qualify as an oxic horizon. This in- dicates that the minerals of the original bedrock from which the present soil materials have been derived, are thoroughly weathered. The remain- ing materials are probably high in kaolinitic clay and the amount of iron and aluminum compounds present are considerably higher than in the un- weathered bedrock. In general, this would mean a lack of available nu- trients, but where the soil has a highly organic surface this is probably not the case, because organic matter has a high cation exchange capacity and holds the nutrients so they are not readily leached out. Action of large earthworms in soil development seems to be significant. Their large tunnels allow rapid movement of water from place to place in the soil. Their movement and mixing of mineral material and organic matter by ingestion is estimated to occur actively in at least 5 percent of the soil volume. They do not seem to use the same tunnels all the time but make new ones, so much of the soil material in the upper foot or two has been passed through their bodies at one time or another. Earthworm casts are numerous at the surface and are readily observed when the forest floor is removed. The rate of this deposition on the surface suggests that there is a minimum of a centimeter or two of soil added to 1969 | LYFORD, ELFIN FOREST, 7 223 the surface every 100 i.e This estimation is based on similar studies with ants (Lyford, 1964). But if there is this much action there is a question how the various fone are able to preserve their identity. Possibly, the earthworms we see now are recent colonizers or perhaps they are now experiencing an unusual population surge. The large earthworms are known to occur elsewhere in the island (Luis Maldonado, pers. comm.), but probably are confined to certain restricted habitats. On steep slopes, the action of earthworms in returning soil to the surface is considerable. Their action was observed not only on the steep slopes of the study area but also in two other places in the Luquillo Mountain area, on steep slopes. This constant return of earthworm casts to the surface may have some rather important geomorphological implications because fresh soil material is always available on the surface for trans- port by running water or gravity. Earthworms not only return soil to the surface, they also consume the leaves that fall on the surface of the soil. They do not, however, consume all the leaves but allow enough to remain to form a complete cover of the soil, thus, their activities are not completely detrimental and, perhaps they have achieved a desirable ecological balance. Presence of a considerable amount of soil, or at least soil-like material, on the stems and branches of trees suggests that some epiphytes may have their roots in soil even though the entire plant is on a tree. “Soil” in trees raises a problem for the soil scientist because he is not in the habit of finding it here, and indeed, this makes some revision of the definition of soil in order. Yet termite mounds several meters in height are common in tropical regions. These insects not only return soil material to the surface of the soil, they raised it several meters above the original surface. It is not too much of a mental jump then to think of soils in trees, particularly when it is readily apparent that some kinds of termites, ants, and other soil-moving fauna make their nests in trees. SUMMARY Soil on the steep narrow ridge of Pico del Oeste is wet and the mucky surface, 25 to 30 centimeters thick, has about 50 percent organic matter. Fallen tree leaves, on which grow liverworts and algae, lie on a layer of clean roots and these in turn are on and in the soil. Tree stems and branches are blanketed with liverworts, algae, and other epiphytes, and the amount of organic matter on the stems per unit area approximates that on the surface of the soil. Soil-like material occurs on the stems and branches of trees. Some may be carried up from the soil by fauna. Large earthworms up to 60 centimeters long and 1 centimeter in diameter are common in the soil and ingest and move large amounts of it. LITERATURE CITED Baynton, H. W. 1968. The ecology of an elfin forest in Puerto Rico, 2. The microclimate 7 Pico del Oeste. Jour. Arnold Arb. 49: 419-430. 224 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Gitt, A. M. 1969. The ecology of an elfin forest in Puerto Rico. 5. Aérial roots. Jour. Arnold Arb. 50: 197-209 Howarp, R. A. 1968. The ecology of an elfin forest in Puerto Rico, 1. Intro- duction and Composition Studies. Jour. Arnold Arb. 49: 381-418. Lyrorp, W. H. 1963. Importance of ants to Brown Podzolic aoe genesis. Har- vard Forest Paper 7. Harvard Univ., Petersham, Mass. & B. F. WItson. 1966. Controlled growth of forest a roots: technique and a. Harvard Forest Paper 16. Harvard Univ., Petersham, Mass. 0 SOIL camel penn 1951. Soil Survey Manual. U.S. Dept. Agriculture Hand- book 18. . 1960. Soil Classification. A Comprehensive System. 7th apg 265 pp. (Supplement 1967. 207 pp.) Soil Conservation Service, U.S. Dep Agriculture. HARVARD FOREST HARVARD UNIVERSITY PETERSHAM, MASSACHUSETTS 01366 1969 | HOWARD, ELFIN FOREST, 8 225 THE ECOLOGY OF AN ELFIN FOREST IN PUERTO RICO, 8. STUDIES OF STEM GROWTH AND FORM AND OF LEAF STRUCTURE Ricuarp A. Howarp* IN THE FIRST PAPER of this study (Howard 1968), the species compris- ing the elfin forest on the summit of Pico del Oeste were listed in systematic order, classified by growth form, and indicated in their frequency through transect and plot studies. It was shown that 14 taxa of monocotyledons and 40 taxa of dicotyledons were the common components, and that these could be distinguished as 25 taxa of woody trees or shrubs, 19 taxa as herbs or herbaceous vines, 4 taxa as woody climbers, and 6 taxa as epiphytes. Various suggestions to explain the reduced stature of such a forest were also listed. These suggestions involved the poor aeration of saturated soils, the reduced light reaching the forest through a heavy persistent cloud cover, the reduced transpiration suggested by the high humidity of the atmosphere and frequent and high precipitation, and the restrictive effect on growth of the high wind velocities. The frequent occurrence of trees which had blown over or had toppled due to the subsurface erosion were Suggested as factors increasing the density of the woody vegetation The present study will consider the nature of the growth of the indi- vidual and component species as factors in the form of the forest and will document the presence or absence of characteristics of plant structure which have been observed in comparable forests elsewhere by other workers. GROWTH AND FORM The short stature of the elfin forest on Pico del Oeste is in part due to the nature of the growth and of the mature form of the individual species. The growth and form of a particular plant may be the expression of a normal genetic factor exhibited at the specific, generic, or familial level as the herbaceous, climbing, woody, rhizomatous, or rosette-forming habit. Unusual or sbnortasl variation from the basic pattern may be an indi- vidual characteristic due to biotic factors of the environment or to muta- This study was possible only with the financial support of a grant from the Na~- on ae Foundation (GB:3975) for which I am deeply indebted. The present rs in tabular form data which represent much detailed work of counting ig dn measuring. I repeat my gratitude to my wife Elizabeth Howard and our children Jean, Barbara, and Bruce for their efforts in collecting, counting, and measuring leaves. The work of drying foliage material, of weighing and obtaining water contents, and of a the pH values of leaves involved the cooperation of sad ger Wagner and his wife, Anstiss Wagner. Mrs. Helen Roca-Garcia prepared m f the slides, patiently d on the leaf section and stomatal measurements, and eaten: ‘the: illustrations. 226 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 tions. We did find on rare individuals a fasciation of branches resulting from insect damage to an apical meristem. Witches’ brooms or comparable fasciations due to fungi or mutations did not appear within the forest under study. A few shrubs, e.g., Cleyera and Symplocos, had scrambling branches when intermediate in height. Aberrant leaf forms were encoun- tered and these were the result of abnormal laminal development following insect damage of primordia or very young leaves. The genetic-based habit of the plant can often be a growth pattern associated with the production of flowers. The production of a terminal inflorescence may impede the normal vegetative extension of a stem and result in a branch dichotomy. Vegetative development may be slowed by the development of a resting or short shoot area. Die-back or self-pruning of a limited amount of shoot development occurs in some species to restrict ultimate growth. By contrast axillary or cauliflorous inflorescences can be produced below the apical meristem of the branch or in the axils of sub- terminal leaves and have no effect on the ultimate habit of the plant. MONOPODIAL GROWTH Growth can be described as continuous if a dormant terminal bud is not formed and if leaves are produced in a regular sequence throughout the year. Such a condition is often attributed to tropical forests. The terminal growth of a branch can be interrupted by a resting period, whether a bud is formed or not, and new growth can be by a flush of new leaves with the subsequent lengthening of the internodal zones. The last leaves formed in a flush of growth may be closely associated, indicating that internodal elongation was reduced. The subsequent new apical growth may also have reduced basal internodes, and in fact, the first leaves formed may be aberrant in size and shape and be termed scales or cataphylls. In- ternodal areas in or above the zone of scales or cataphylls may be longer than average and the flush may again terminate in an apparent rosette of leaves. Such a growth pattern is basically a monopodial production of a long shoot terminated by an area of a short shoot. Jlex sintenisit ex- hibits this growth pattern on Pico del Oeste. SYMPODIAL GROWTH Occasionally the long shoot is produced not from the terminus of the short shoot, but sympodially from an axillary bud of a lower leaf. When the growth of a single leader extends vertically, the lateral offset of the new shoot may be noticeable for only a short period of time. It is evident, however, that the leader was terminated by a short shoot development and although the sympodial offshoot continued growth in the same direc- tion, a restriction was imposed in the rate of apical elongation. Such a growth pattern was found in Torralbasia. Within the Pico del Oeste forest nine genera exhibited a restriction in vertical growth by the production of long shoots and short shoots and 1969 | HOWARD, ELFIN FOREST, 8 227 > = ‘i Va NN SS : if iy WW) Ficure 1. Wallenia yunquensis. Plant grown from seed in greenhouses of the Arnold Arboretum. Short shoots, or terminal rosettes of leaves, and the single unit sympodial branching are all shown. were further affected by the development of lateral branches. Wallenia yunquensis commonly exhibits a vertical shoot producing a single lateral flush of growth as a branch (Fic. 1). The erect main shoot of Wallenia 228 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 may be interrupted by short shoot areas separated by areas of long shoot development. From one or more leaf axils in the short shoot zone a single fast-growing lateral branch may develop which appears naked at the base but does in fact have widely separated cataphylls of very short duration. The naked shoots are terminated with a short shoot zone possessing an aggregate of leaves. The lateral branch may also originate from the area of cataphylls. In Wallenia the lateral flush shoots never branched or con- tinued growth beyond the initial flush. An example of repeated sympodial lateral branching is readily seen in Ocotea spathulata. The development of flush shoots appeared to be from a short shoot zone in all cases, but there developed additional and com- parable lateral shoots from the terminal short shoot zone of the lateral branch. This growth pattern results in a sympodial development of lateral branches in a flat plane. The principal branches are tiered in appearance, the tiers being separated by an unbranched, seemingly naked stem. This growth form has been described as candelabra-branching, Terminalia- branching, or as pagoda trees. Corner (1952, p. 32) described this growth pattern for Terminalia as follows: “The leader-shoot rather suddenly lengthens into a long vertical fea clothed with a lax spiral of leaves . . . its growth slackens. . . . another terminal rosette is produced. From the base of this ashe several twigs grow out to form the next tier of branches . . . The positions of the branches in successive tiers usually alternate so that only those of every other tier are superimposed.” Lateral branches from the terminal short shoot area may also grow vigorously, producing scales or cataphylls before developing normal leaves and, ultimately, each its own terminal shoot. Corner (1952, p. 31) de- scribed the lateral growth as follows: “Each twig which grows from the leader-shoot of the tree does so rapidly and at a wide angle from it; then, as its growth slackens, it turns up at the end and from its lower side, just at the bend, a branch arises to grow out as another twig which will follow the same course by turning up at the end and branching in its turn. - - In the first horizontal part of such a twig the internodes are lengthened; the leaves, or their scars, are widely spaced on the slender stem; and the growth has been rapid so that the new shoot has quickly been thrust be- yond the parent rosette of leaves. In the second, vertical or upturned, part of the twig the internodes are very short or absent and the leaves, or their scars, are very crowded on a stout stem so that, while many more leaves are being produced than in the previous stage of the twig’s develop- growing, withers and falls off. When such a limb .. . is growing out from the trunk of the tree, it diverges from its neighbours and begins to branch sideways: this it does by producing every now and again not one twig but a pair of twigs, or even three, which grow out from each other at a wide angle; and thus the limb develops into a fan-shaped leafy spray.” The Terminalia-type of branching was particularly conspicuous in Ocoted spathulata (Lauraceae), one of the dominant trees of the elfin forest. 1969 | HOWARD, ELFIN FOREST. 8 229 Ib W 3 Z 6 he 3b 3a Ss le 2b 3b 4a Tc 2b “ Ib 2a . 4b ? 3a IGURE 2. Diagram of growth pattern of Ocotea spathulata. A, side view showing tiers of repeated sympodial branching; B, view from above showing the relationship of the tiered branches. Drawing by Pamela Bruns. a 230 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Ficure 2A shows, diagrammatically, a plant of Ocotea of average stature in the forest with four tiers of sympodially developed lateral branches. It is evident that the terminal growth is impeded while the lateral sympodial growth proceeds. Lateral branches at 4a or 4b show four additional flushes of growth while only 3 periods of reactivation of the terminal shoot are evident. The lateral branches, la and Ic, have each had one reactiva- tion of growth while the terminal remains static. The relationship of the lateral branches is illustrated in Ficure 2B and shows a conflict with Corner’s observation that the branches of alternate vertical flushes are superimposed. Specimens of Ocotea were found within the elfin forest with 27 flushes of sympodial lateral growth on the oldest branch. Only the last 13 of these sympodial flushes retained any foliage demonstrating a die-back of the upturned short shoot after a considerable period of continued lateral sympodial growth. Terminalia-branching was also observed in Ardisia and Grammadenia (Myrsinaceae), in Torralbasia (Celastraceae), and in Calycogonium (Melastomataceae). Terminalia-branching is conspicuous in Hedyosmum arborescens and is due to the naked areas of elongation and the terminal production of leaf pairs or of an inflorescence. Continued growth of the sympodial branches in this species was restricted in many examples by the production of an inflorescence. The sympodially branched lateral shoots can develop vertical extensions as well as horizontal branches. Vertically developed shoots as elongate leaders of vigorous growth have been observed on the lateral branches of specimens of Ocotea and Calycogonium, but in all cases additional sympodial branching also occurred from the same upturned short shoot and beyond it. Attempts were made to induce either vertical elongation of the up- turned shoot or the production of new or additional sympodial branching by pruning the leader shoot. During the three year time interval of the study all of the branches which had been pruned of lateral growth failed to respond by any new development from the areas of the upturned shoot. Likewise, vertical leaders when partially or completely decapitated by pruning failed to develop any sympodial lateral branches. Although many branches were marked along the trail to record growth phenomena, we were unable to draw conclusions on the frequency with which sympodial branching occurred normally. In no case where sympodial branching was noted in early stages of development (and the branch tagged for observation) was there any further sympodial branching. We could only conclude that the sympodial branches were not produced an- nually on any branch we had marked for observation. Gill has reported the occurrence of adventitious or aérial roots on the species within this forest. Although adventitious roots were observed on the horizontal branches of the species exhibiting sympodial lateral branch- ing, the roots did not appear to be associated with the short shoot area or the curved portion of the lateral branch. Ocotea, which had the most “ _“-s----- 1969 | HOWARD, ELFIN FOREST, 8 231 conspicuous Terminalia-branching, rarely produced adventitious roots from the horizontal branches but did develop “prop” roots from the base of the stem. The development of sympodial branching or of vertically continuous long shoot-short shoot growth patterns was not associated with flowering in Ocotea, Grammadenia, Torralbasia, Calycogonium or Wallenia. Ardisia, however, did develop a terminal inflorescence, and following the maturity of the fruit and the fall of the inflorescence axis, a lateral but vertical continuation of the stem developed as a sympodial flush of growth. DICHOTOMOUS BRANCHING The dichotomous branching of upright shoots was observed in a number of the components of the elfin forest. Dichotomous growth and branching was most conspicuous in Calyptranthes where it occurred, on the average, every three internodes. The new shoots developed in pairs and normally two pairs of leaves developed in each flush before elongation stopped. At the apex of each shoot there was a terminal aborted primordium. Subse- quent branching occurred lateral to the terminal aborted primordia but remained consistently in one plane. Calyptranthes appeared to be a col- lection of upright fans of dichotomous branches. he three species of Miconia always developed upright dichotomous shoots when growth was terminated by the production of a terminal in- florescence. Two lateral buds continued the upright vegetative growth after the inflorescence had matured fruit and had fallen. Subsequent growth consisted of but one or two pairs of leaves per flush. In mature plants flowering followed the maturation of each flush of growth on an annual basis. In Eugenia borinquensis one or two pairs of leaves formed each flush of growth. Flowering occurred only on mature stems and terminated the branch or was formed in an axillary position on the old wood. Both species of Psychotria produced a terminal inflorescence and there was no further vegetative growth on that shoot while the inflorescence persisted. With the maturity and desiccation of the inflorescence, how- ever, two basal axillary buds developed in Psychotria berteriana, produc- ing a dichotomous growth pattern. In Psychotria guadalupensis, however, only a single bud developed at the base of the inflorescence and the result- ing growth was falsely monopodial. Among the herbaceous vines Jpomoea repanda, with alternate leaves, produced a terminal inflorescence of many flowers which matured over an extended period of time. Axillary vegetative shoots often developed while the inflorescence was only partially mature. In Mikania pachyphylla, with opposite leaves, a terminal inflorescence appeared to restrict apical growth while the terminal inflorescence matured, but then a dichotomous growth pattern developed through activity of two axillary buds. The heteroblastic growth of Marcgravia sintenisii also showed an asso- Ciation with the production of a_ terminal inflorescence. Subsequent oa2 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 growth was by the development of axillary buds below the inflorescence. In all cases observed, leaves of the initial production on this axillary shoot were of the juvenile form whether or not the branch was in contact with a trunk or branch Plants of Tabebuia rigida formed the second major component of the forest and these plants produced flowers throughout the year mostly from new growth. The terminal flush of growth consisted of an average of branches. Most of the plants of Tabebuia which were observed in the canopy of the forest also showed a significant die-back of the flush growth during the winter season. Subsequent apical growth, therefore, came from adventitious buds in the axils of lower leaves or from opposite buds of such a node DIE-BACK Regular die-back of terminal and, less frequently, of lateral or sym- podial shoots was observed in Jlex sintenisii, Cleyera albopunctata, Eugenia borinquensis, Hornemannia racemosa, Ardisia luquillensis, Micropholis garciniaefolia, Alloplectus ambiguus, Gesneria sintenisii and Lobelia porto- ricensis. Although regular die-back has been described for plants of temperate areas, its occurrence as a factor in the size of a plant in tropical areas has not been recognized previously (Garrison & Wetmore 1961). Relatively long flushes of shoot development were observed in Gono- calyx and Hornemannia consisting of 5~10 leaves or internodes per flush. The young leaves were brightly colored and soft until full extension of the shoot, or the full development of the leaf size, was completed. Both species were climbers and the soft shoot development was often injured mechanically and the entire flush of growth abscissed. The height of the forest may be affected by the environmental factors previously suggested, but clearly the low stature of the component woody species may also be due to genetic factors expressed as long shoot-short shoot development, dichotomous branching associated with a terminal inflorescence, the abortion of the shoot tip, and the die-back of seasonal flushes of growth. Continuous production of single leaves or leaf pairs occurred in all of the herbaceous species in the elfin forest. Continuous production of single leaves or leaf pairs appeared to occur in both juvenile and adult shoots of Marcgravia, in Symplocos, Cleyera, Ilex, Gesneria, Clusia, and Micro- pholis. The herbaceous vines Rajania, Ipomoea ani Mikania also ap- peared to produce leaves continuously unless affected by flowering The rosette and epiphytic habit of the two members of the ee found within the elfin forest can be regarded as a family genetic character. Following flowering, however, the two species continued growth in differ- ent patterns. Rosettes of both species died following flowering, but plants of Guzmania produced one or, rarely, two basal vegetative rhizomes which 1969 | HOWARD, ELFIN FOREST, 8 233 developed laterally before terminating into a rosette or crown of leaves. This growth pattern caused the plants to form a ring around the host tree and Guzmania was most commonly found on the large trunks of Prestoea montana. Vriesea sintenisii by contrast, produced-a single basal rhizome which tended to grow upward immediately and formed a new crown in close competition with the parent rosette. The new growth could be to the right or the left of the parent plant but always extended upward. Plants with 7 generations of rosette-rhizome vertical development were found. When Vriesea sintenisii occurred on a branch extending horizontal- ly, the plants persisted for only 2 or 3 growth generations before being extended slightly off center and, seemingly top heavy, falling over to break free and drop to the ground. BUD PROTECTION Richards (1952, p. 77) notes that “buds of rain-forest trees and shrubs, as might be expected, are less well protected than those of trees in other climates.” An examination of the terminal foliage buds or the shoot apex in resting condition revealed that the leaf primordia are better protected in the plants of the elfin forest than might be expected from Richards’ statement. (PLATE I). Protective stipules are present in Hillia and Psychotria of the Rubiaceae and in Calyptranthes of the Myrtaceae. In Hillia (PLATE Ib) the stipules form a sheath around the young leaves which is compressed at the apex. The developing young leaves force an opening in the apex of the sheath. Psychotria species have smaller stipules consisting of an ochrea-like base with four short free apices. The apices tend to be closely associated in very young buds but their protective function would be of short duration. Calyptranthes possesses a peculiar type of stipule protection for which we have not found a description elsewhere (PLATE Ig). In fact Berg, in a monograph, reports the family to be estipulate, as have subsequent the original description of Calyptranthes krugii, Kiaerskou notes, “Quaque innovatio e duobus internodiis constat, quarum alterum breve duo cata- phylla opposita cito decidua, alterum longum duo euphylla fert.” In a footnote he equates “cataphylla” with “Niederblatter Germanorum.” In our observations of Calyptranthes krugii on Pico del Oeste the vegetative shoot increase is by production of 1 or 2 pairs of leaves in a flush. The apical meristem aborts although an inflorescence of one, rarely two, flowers may be produced in one or both terminal leaf axils. Subsequently, after flowering or resting, two axillary shoots develop and in each the apex is covered with a pair of laterally folded bud scales. The young leaves in- crease in size uniformly and are appressed by their ventral or adaxial surfaces. As the leaves increase in size, the bud covering is forced apart or torn free at the base, and the two halves separate as conduplicate folded sheaths. The bud scales are a light yellow or cream color in contrast to 234 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 the green shoots. When separate they dry white, then brown and shrivel before falling from the shoot. Although these bud scales are rarely found on herbarium sheets, they were conspicuous in the living plants of Calyp- tranthes krugii in the study area and were also found on a population of Calyptranthes which may represent a different species in the Cerro de Punta area. The term cataphyll although broadly inclusive for the early leaves of a plant or shoot as cotyledons, bud scales, etc. (Jackson) is scarcely descriptive of the folded protective scales of the young leaves of Calyptranthes. Apical buds may be protected by leaf bases as in Clusia which has opposite leaves or by the cluster of leaves in the several genera which produce terminal or lateral long shoots where the terminal apical elonga- tion is reduced. In Clusia (PLATE If) the mature leaves conceal the young buds as the leaf bases of opposite leaves of a pair are tightly appressed. In Wallenia, Ocotea, Grammadenia, Torralbasia, Ilex (PLATE Ic), and others, the vegetative shoot in resting condition is terminated by a dense cluster of small leaves or primordia. In subsequent development of the shoot represented by these primordia, the outer ones enlarge only slightly, frequently failing to develop a leaf blade even though the petiole may elongate. Such scales or cataphylls are found at the base of the long shoot and the internodes between them may or may not have elongated. Clearly these cataphylls have served a function of protection for the inner leaves and the apical meristem. The buds or apical meristem of shoots of Hedyosmum are enclosed within the sheathing stipular base of the leaves (PLaTe Ie). Bud protec- tion here is evident in the enclosure of the primordia in the sheathing leaf base. The apex of the stem of Micropholis and Symplocos have the young leaf primordia tightly invested in a protective covering of brown trichomes. As the leaves expand these trichomes are separated and in many cases break off. In Tabebuia rigida the young leaves or primordia are tightly and completely encased in a shield of brown peltate scales. Again with leaf enlargement the scales are separated and often persist in isolated posi- tions on the mature leaves. The species Gesneria sintenisii appears to have a large naked meristem where the leaf primordia are separated and evident from an early age (PiateE Id). The leaf primordia and the apex of the stem have a dense resinous covering. As the leaf starts to expand the resinous covering is cracked and usually flakes off although sections of the covering may per- sist even on the mature leaf blade and the petioles. The young leaves of Marcgravia (Pate Ia) and Cleyera are convolute in bud and appear to unroll in development. The apical meristem is en- closed within this pointed bud and the youngest leaf primordia do receive some protection. It is clear from these examples that the young leaves are not without protection in the majority of the species that comprise the woody com- ponents of the elfin forest on Pico del Oeste. 1969 | HOWARD, ELFIN FOREST, 8 235 LEAF SIZE AND MORPHOLOGY Plants of tropical forests have been grouped on the basis of leaf size: dimensions and areas. Raunkiaer proposed a classification of life forms on leaf-size classes which has been used for comparison and description by many authors. Leaves have been termed leptophylls if their area does not exceed 0.25 cm.*; nanophylls if their area is between 0.25 and 2.25 cm.*; microphylls if the area is 2.25-20 cm.?; mesophylls if the leaf area is 20-182 cm.”; and macrophylls if the area is 182-1640 cm.” Cain et al., found, in a Brazilian rain forest, that the phanerophytes are strongly mesophyllous and reported a tendency for the small leaf size classes to have a higher percentage in taller strata than in lower ones Brown, in his study of the mossy forest on Mount Maquiling in the Philippines, found only the leaf-size classes of microphyll and mesophyll represented in approximately the same numbers. The elfin forest on Pico del Oeste had a single species (Peperomia emarginella) of a size class smaller than the nanophyll classification and the majority of plants were of the microphyll size class. The total classifi- cation in numbers of taxa and percentage of the totals is the following: leptophylls 1 1.9% nanophylls 6 11.5% microphylls 30 57.6% mesophylls 13 25.0% macrophylls 2 3.8% Compound leaf types were represented only by Trichilia pallida, a species clearly only surviving and not reproducing in the elfin forest zone. At lower elevations Trichilia pallida becomes a small tree while most of the plants found on Pico del Oeste were weak saplings, dependent for support on the surrounding vegetation and nearly scrambling through the elfin forest. A single plant of Weinmannia pinnata (Cunoniaceae) with compound leaves was found on the peak but was not encountered in the transects. Brown did not have a compound-leaved plant in the Philippine study area and Lebrun notes such plants are less than 15% in African elfin forests. The largest leaves, macrophylls, were those of Prestoea montana, a palm, restricted to the leeward erosion valleys and Anthurium dominicense, an epiphytic member of the Araceae. When grouped according to habit the following leaf-size classification was obtained: LEAF SIZE VINES-SCRAMBLERS HERBS EPIPHYTES TREES & SHRUBS leptophylls 0 1 0 0 nanophylls 1 2 2 l microphylls 3 10 2 a mesophylls 1 2 2 8 macrophylls 0 0 1 1 Brown added data on leaf dimensions and leaf margins to his study of the Mt. Maquiling forest in the Philippines. He found the leaves were 236 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 0-10 cm. long in 11 species or 70%, and 10-20 cm. long in 5 species or 30% of the plants. The leaves were 0-5 cm. wide in 12 species or 75%, and 5-10 cm. wide in 4 species or 25%. In the elfin forest of Pico del Oeste the leaves were 0-10 cm. long in 7 species of monocotyledons and 31 species of dicotyledons or 70% of the total flora; and 10-20 cm. long in 7 species of monocotyledons and 9 species of dicotyledons or 30% of the total. The leaves were 0-5 cm. wide in 10 species of monocots and 33 species of dicotyledons or 79%, and 5—10 cm. wide in 4 species of monocotyledons and 7 species of dicotyledons or 21%. The leaves selected for these measurements were taken from the mature growth and were averaged for the plant. Variation in leaf size within a given plant ranged from the cataphylls and reduced leaves of initial growth of long shoots to the larger leaves of vegetative shoots, when compared with those of flowering branches. Heterophylly was found in dimorphic pairs of leaves in Pilea krugii, Pilea yunquensis (Urticaceae) and in Alloplectus ambiguus (Gesneriaceae). Heteroblastic growth was found only in Marcgravia sintenisii with appressed smaller leaves on juvenile and climbing shoots and larger leaves on the free arching branches. Heterophylly with age was observed in Ocotea spathulata and Symplocos micrantha where the leaves of seedling plants appeared to be quite dif- ferent in size and shape from those of adult plants. Although Cleyera albopunctata appeared to have larger than average leaves on some vigor- ous growing branches this could not be documented with measurement of samples. However, the leaves of sterile or vegetative branches of Clusia grisebachiana did possess larger leaves than were found on shoots which were mature or produced inflorescences. Macrophylly on adventitious shoots was not encountered within this forest. Much attention has been given in existing studies of tropical forests to the shape of the leaf, the nature of the margin, apex and base of the blade, and to the presence of a cuticular layer in relation to the retention of water or the presence of epiphyllous organisms. Brown noted in his study of the mossy forest at 1000 meters in the Philippines that as the altitude increases there is a marked increase in the percentage of small leaves and a decrease in the percentage of leaves with entire margins. It has been suggested that the presence of marginal teeth aids the runoff of water from the leaf surface, and Brown found entire leaves in eleven species of Philippines plants in the study area and five species in which the margin was not entire. In Puerto Rico on Pico del Oeste 29 taxa had entire leaves while eleven taxa of dicotyledonous plants had leaf margins with coarse or blunt teeth or with marginal un- dulation. The extended leaf tip, often called a drip-tip, has a popular association with wet tropical forests. The conclusion of Junger has been cited re- peatedly that the function of the pointed leaf tip was to hasten the run- off of water from the leaf, and thus help prevent insects and lower plants from attacking them. Baker recorded 37 of 41 species of plants belonging to 20 families with leaves drawn out into a tip, in a forest in Ceylon, and 1969 | HOWARD, ELFIN FOREST, 8 237 Richards observed, “Pointed tips to leaves are characteristic of plants of wet regions and especially of tropical rain-forests, but I doubt whether any rain forest can show the phenomenon more markedly than the Sin- haraka.” Richards observed that drip tips are common and better devel- oped in the lower than in the upper strata of the forest and in juvenile than in mature leaves of tall trees. Cain e¢ al. reported that 70.6% of the leaves in the Brazilian forest they studied had acuminate tips and 28% of the 150 species studied had rather abruptly long tips of the drip point type. By contrast, Shreve found drip tips uncommon in the montane rain forest of Jamaica, and Vaughan and Wiehe reported a similar observation for upland climax forest of Mauritius. In the Pico del Oeste forest 13 of the 14 taxa of monocotyledons had the leaves acuminate at the apex and the other taxon had leaves acute. Among the dicotyledonous plants the apex could be classified as acuminate in 15 taxa of which 6 would qualify as drip tips; 14 taxa had the leaves acute at the apex and 11 had the leaves obtuse, blunt, or emarginate. Although previous authors have not considered the leaf base, it seems that if leaf shape is important for drainage in one direction, it is equally so in the other. Only 5 of the 14 taxa of monocotyledons have petioles and of those, Prestoea montana, the mountain palm, has lacerate or com- pound leaves; Rajania cordata and Anthurium dominicense have the basal lobes extended and Renealmia antillarum and Brachionidium parvum have the leaf base obtuse. Of the dicotyledonous plants 24 taxa had the leaf base blunt, acuminate, or decurrent on the petiole while 16 are best de- scribed as cordate to peltate at the base. All of the leaves which were cordate, hastate, or peltate at the base had an acuminate apex or a drip tip. All leaves which were blunt at the apex or rounded or emarginate had acute or decurrent leaf bases except for Micropholis garciniaefolia and Eugenia borinquensis. In these two taxa the attitude of the leaves to the stem tended to be either upright or droop- ing and in this manner adapted to the runoff of water. Excepting Eugenia, those leaves with short petioles or with petiole:blade ratios 1:10 or larger, all had tapering blade bases with the leaves mostly arranged up- ward in attitude. Cleyera albopunctata has short petioles but the leaves have a noticeable curvature. Marcgravia sintenisii, again with a short petiole, also has a slight curvature and a drip tip. The frequency of taxa having leaves of strongly curved form suggests a selective value can be attached to this growth form. The blades may be noticeably curved longitudinally as well as laterally or in but one plane. This curved form is particularly evident in taxa of Calycogonium, Cleyera, Gesneria, Gonocalyx, Hornemannia, Ilex, Miconia pycnoneura, Symplocos, Tabebuia, and Torralbasia, that is in 10 of the 40 taxa of dicotyledons or 25% of the flora. ; A heavy upper cuticle was found in 22 taxa or 55% of the dicotyledon- ous plants. Ardisia, Cleyera, Clusia, Calyptranthes, Calycogonium, Eu- genia, Gonocaylx, Haenianthus, Hillia, Hornemannia, Ilex, Marcgravia, 238 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Miconia foveolata, Miconia pachyphylla, Miconia pycnoneura, Micropholis, Ocotea, Psychotria guadalupensis, Symplocos, Trichilia, Torralbasia, Tabe- buia, and Wallenia, that is, all woody taxa except Mecranium amygdali- num, Grammadenia sintenisii, Hedyosmum arborescens and Psychotria berteriana possess a heavy upper cuticular layer. Junger found that leaves with drip tips were less frequently overgrown with algae, fungi, lichens and bryophytes than those without. He be- lieved that the presence of these epiphyllae interfered with assimilation to such an extent as to be a serious handicap to the plant. Micropholis garciniaefolia and Eugenia borinquensis, which stand out as the only taxa of the 40 dicotyledons or 29 woody plants in which the leaves were rounded or cordate at the base and rounded at the apex and appear to lack any special adaptation for getting rid of surface water, seemed to support the larger populations of epiphyllous plants. The abundance of epiphyl- lous leafy Hepaticae on different species will be considered later in this paper in relation to the metabolism of the forest. In a superficial classification of the texture of leaves within the forest components, the leaves would be considered as membranaceous in all of the monocotyledons except Anthurium, which had leathery leaves. Among the dicotyledonous plants 13 taxa would have the leaves classified as membranaceous, 8 taxa would be described as fleshy or succulent, and 21 taxa as having the leaves leathery or coriaceous. The high percentage of water in the tissues or the relatively small amount of material forming dried weight will be considered later and is indicated in TABLE 1, column 8. The heavy texture of the leaves, the thickness of the blade, and the amount of succulence all support previous suggestions that the flora of the mountain summit shows many xeromorphic characteristics. Bews re- gards the rain forest type of leaf as xeromorphic and associates its charac- ters with the low specific conductivity of the wood for water. Shreve re- marks that the prolonged occurrence of rain, fog, and high humidity at relatively low temperatures places the vegetation of a montane rain forest under conditions which are so unfavorable as to be comparable with the conditions of many extremely arid regions. Xeromorphy is usually in- terpreted from such anatomical characteristics as cuticle, hypodermis, thin palisade layers, pubescence or idioblasts. Wylie (1954) noted that a xerophytic flora may have a high proportion of representatives with leaves having a hypodermis. In his studies of plants of North Island in New Zealand, Wylie, even though avoiding “ex- treme xeromorphs and succulents,” concluded that the species studied re- vealed a high average thickness of leaves, extensive spongy mesophyll and palisade parenchyma areas, great cuticular depth, and “the proportion hav- ing a hypodermis were greater than for any group previously studied.” Wylie (1946) compared his studies of the New Zealand plants with previous ones of his own, based on plants of Florida and of other temper- ate areas. Wylie reported that the leaves of the New Zealand species studied, ranged in blade thickness from 731 » for Pseudopanax to 172 p for Olearia, 1969 | HOWARD, ELFIN FOREST, 8 239 and the 38 species averaged 406 ». This was much greater than the corres- ponding thickness of 216 » for 121 Florida dicotyledons and 80 » for 80 species of northern dicotyledonous trees. Philpott reported a mean blade thickness of 234 p for 24 species of Ficus growing in Florida and Cooper found a mean laminar thickness of 336 » for 19 species of woody dicotyle- donous plants in the climax chaparral in western California. Within the Pico del Oeste forest the woody plants by comparison had leaves ranging in thickness from 787 p» in Clusia grisebachiana to 146 p in Psychotria berteriana and averaged 379.6 » in thickness. The herbaceous flora had leaves ranging in thickness from 625 p» in Peperomia hernandii- folia to 141 » in Sauvagesia erecta, and all herbs had leaves averaging 281.6 w in thickness (TABLE 2, column 1). Wylie reported that a hypodermis was found in 24 or 63% of the 38 species examined in the New Zealand study area. Eighteen species had a hypodermis on both the upper and lower surface, 5 species had only an upper hypodermis, and one species is described as having only a hypoder- mis on the lower side. In the Pico del Oeste elfin woodland 19 of 40 taxa or 47% have a hypo- dermis. Two taxa, Begonia decandra and Hillia parasitica had both an upper and a lower hypodermis. No plant was observed with only a lower hypodermis. Seventeen taxa had an upper hypodermis alone. Twenty- one taxa did not possess a hypodermis (TABLE 2, columns 2, 3). The presence of a hypodermis is often regarded as a xeromorphic charac- ter, although a multiple hypodermis is also an anatomical characteristic of taxonomic value. Carlquist noted that “continued periclinal division of the epidermis is of taxonomic importance in certain families such as the Piperaceae.” Within the plants of Pico del Oeste, multiple hypodermal layers were found in taxa of Peperomia (Piperaceae), Hedyosmum, (Chloranthaceae), Ocotea (Lauraceae), Clusia (Guttiferae), Calycogo- nium (Melastomataceae), and Hornemannia (Ericaceae). A ratio was determined between the thickness of the upper epidermis and that of the upper hypodermis in this mossy forest. Ratios varied from 1:1 in most plants with a hypodermis, to 1:15 in Psychotria guadalupensis and averaged 1:4.1. Although Wylie did not use such a figure calculation, the figures in table 2 of his paper (1954) suggest a ratio range in the New Zealand plants from 11:1 to 1:21.3 but an average of 1:3.8, or less than that found in the Puerto Rican vegetation. Stalfelt considers the mechanical strengthening of leaves through the de- velopment of sclerenchyma as particularly common among xerophytes as a means of reducing the injurious effect of wilting. Branched idioblasts were found in but four taxa within the mossy elfin forest (TABLE 2, col- umn 8 and Prats IIa). Watson concluded that the formation of palisade tissues in leaves might be a morphological response to light. He suggested that the cigar-shaped palisade cells are formed in increasing number with increasing light in- tensity during leaf development. We examined the leaves on Pico del Oeste to see if palisade mesophyll development was reduced with the re- 240 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 duced light intensities we have reported there (Baynton 1968, 1969). Two taxa, Lobelia portoricensis and Mikania pachyphylla, seemed to be without a definite palisade layer in the leaves examined (PLATE Id). Further, 19 of the remaining 38 taxa possessed a palisade mesophyll of but a single cell in thickness. The remaining 19 taxa had a palisade mesophyll in part exceeding a single cell layer to 3 to 4 cells in thickness (TABLE 2, col- umn 9). In taxa which could be measured, the palisade layer exceeded the spongy layer in thickness in only 4 taxa, while the ratio of palisade to spongy, in 34 taxa, ranged from 1:1 to 1:7.4 and averaged 1:2.4 (TABLE 2, column 10). Referring to Wylie’s study of New Zealand plants, of the 38 taxa he examined 7 had a thicker palisade layer than spongy layer and the comparable ratios determined from the figures he gave show a range from 1:1 to 1:3.3 with an average of 1:1.5. On the basis of limited comparative data it appears that the leaves of the plants growing in the elfin forest on the summit of Pico del Oeste are thicker than usual and approach leaves of admittedly xeromorphic type. The frequency of a hypodermal layer or multiple hypodermal layers is high. The ratio of thickness of palisade and spongy mesophyll layers suggests that the plants surviving on Pico del Oeste have adjusted to the low light values through a reduction in the palisade mesophyll zone and an increase in the amount of spongy mesophyll. LEAF DEVELOPMENT Richards has reviewed the earlier literature which claimed that a few species in Buitenzorg were ever-growing and showed no foliar periodicity whatever. On further study it was shown that one plant at least was in continuous leaf production when young, but when older leaf produc- tion was distinctly periodic. Richards (1964, p. 193) concluded that “it is certainly true that most rain-forest trees produce new leaves, not con- tinuously, but in periodic flushes, so that a single shoot bears several ‘gen- erations’ of leaves at the same time.” Our observations on the development of stems were in relation to the production of leaves (Taste 3). Initially we observed that certain plants did grow in obvious flushes where the young leaves were brightly colored or soft in texture in comparison with the mature leaves. We recognized 20 taxa which grew in flushes, 8 of which had conspicuous terminal long shoot-short shoot growth patterns. Twelve taxa were considered to be in continuous production of leaves but this varied from branch to branch on a given plant. A large specimen of Clusia grisebachiana, for example, failed to produce a single new leaf during the period of this study. marked plant of Trichilia pallida did not add a single leaf, or lose any, for a period of three years after the plant was tagged for observation. //ex sintenisii which appeared to have young green leaves all of the time proved to have only some of the individual shoots on the plant in a stage of growth or expansion at any given time. A shoot of J/ex tagged for observa- tion was shown to produce a flush of leaves and then remain in a mature 1969 | HOWARD, ELFIN FOREST, 8 241 but quiescent stage before renewing its growth. Some of the shoots re- newed growth with no change in the size of the leaves while others had an initial renewal of growth in the production of leaves, or a single leaf of smaller size or reduced to cataphyll proportions. When the internodes along a stem were measured carefully there was evidence that growth of the internodes had been reduced in some areas giving further evidence to a periodicity of growth. Clearly, it is difficult to determine that a given shoot has not added a leaf, but the majority of plants observed in the elfin forest did exhibit some degree of periodicity of growth and leaf production during the period of study. The suggestion has been made that leaves develop quickly in tropical forests. Studies of leaf expansion within a temperate area at the Arnold within a 10-day to three-week period in most native and cultivated species. Although many species studied within the elfin forest did complete the expansion of leaves within that period, there were notable exceptions. Tagged shoots where fairly large leaves were counted and observed at regular intervals showed the following times for development from a no- ticeable leaf primordium to full expansion. Symplocos micrantha 5 weeks Clusia grisebachiana 7 weeks Miconia pycnoneura 14 weeks NUMBER AND PERSISTENCE OF LEAVES PER PLANT Regular observations of the elfin forest components impressed upon us the fact that some plants had many leaves and that others, equally characteristically, had few leaves. Although the leaves may have been produced in flushes of growth or seemingly continuously, there was a leaf fall that in most plants seemed to equal leaf production. We selected 24 plants of comparable size and age of Miconia pachyphylla, Wallenia yunquensis and Dilomilis montana, counted the leaves, and found less than 5% variation in the number of leaves on a given plant of the species. Branches of Miconia foveolata or Psychotria berteriana characteristically had but 3 pairs of leaves at the end of a shoot. When new growt curred the new shoot had a comparable number of leaves and the leaves of the former growth generation abscissed. The largest number of leaves on a mature plant was found on Micropholis garciniaefolia with 10,487, while Brachionidium parvum characteristically had but 4 leaves per plant. The following table indicates the plants with the greatest number of leaves in comparison with the total photosynthetic area represented on the plant, and the rank of the plant in frequency counts for transects reported in the first paper of this series. Leaf numbers and total photosynthetic area for all species is given in TABLE 1. The leaf count was obtained as a 242 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 by-product of gathering foliage material for a chemical survey of the plants within the elfin forest. TOTAL TOTAL NUMBER OF LEAVES PHOTOSYNTHETIC AREA FREQUENCY * Micropholis 10,487 Prestoea 289,460 cm.” Pilea krugi Tlex 8,684 Micropholis 96,480 Wallenia Calyptranthes 4,539 Tabebuia 60,040 Calycogonium Tabebuia 2,680 Hedyosmum 44,908 Vriesea Hedyosmum 2,339 Eugenia 23,000 Ocotea Calycogonium 1,345 Lobelia 20,300 Calyptranthes rdisia 1,857 Calyptranthes 18,609 Pilea obtusata Gonocalyx 1,345 Psychotria 18,093 Dilomilis montana berteriana Haenianthus 1,294 Haenianthus 17,339 Miconia pachyphylla Marcgravia (adult) 1,280 Ardisia 17,458 Tabebuia rigida Cleyera 678 Ilex 15,631 Eugenia borinquensis * Frequency = descending order of frequency in transects. Holttum noted that in the uniform climate of Singapore, trees of a number of deciduous species change leaves annually, many in February, others in August, apparently because of leaf senescence. Within the elfin forest of Pico del Oeste the greatest noticeable leaf fall in the dominant plants of the forest occurred in February for Eugenia, Ocotea, and Tabebuia and was conspicuous in March for Lobelia. In each of these plants the leaf fall preceded the development of new year’s growth, and the plants presented a barren appearance for a short period in con- trast to their normal condition. Other species developed new growth before the erratic abscission of the older leaves. Richards notes the many widely different types of behavior among trop- ical trees in regard to leaf fall and leaf persistence. According to Warming and Graebner the average length of leaf life of tropical species is about 13-14 months. We made an attempt to mark branches and to record the number of leaves, the nature of the new growth, and the length of time individual leaves persisted (TaBLe 3). In general, the results were unsatisfactory. Often the tagged branch failed to develop any new leaves during the period of observation, while an adjacent branch of the same plant for which data had not been recorded produced a flush of leaves, oF flowered, or died. It is not possible to report with accuracy that the growth flush in a long shoot-short shoot growth pattern represented an annual increment of growth as may be done in temperate areas with deciduous or bud-forming plants. We did observe that some leaves re- mained on the plant during two full years of observation. Branches of Ilex sintenisti which appeared to have two flushes of leaves per year re- tained some leaves through 20 internodes, which represented 7 flushes as determined by areas of short internodes and by cataphylls. Torralbasia cuneifolia, which also grows in flushes with the production of many cata- phylls, also retained leaves for 20 nodes representing 7 flushes in this plant. Only one or two of the larger leaves persisted while cataphylls and 1969 | HOWARD, ELFIN FOREST, 8 243 smaller leaves abscissed. Gonocalyx produced 3 to 4 leaves per flush and retained 20 leaves in 5 recognizable flushes with all leaves persisting. Tabe- buia tended to retain only 1 pair of leaves of each flush of 2 pairs and the oldest persisting leaves were 10 internodes below the apex, suggesting that some leaves have persisted for five years. In general, younger plants in the undergrowth tended to hold more leaves per shoot for a longer period of time than did the plants with shoots exposed in the canopy. FACTORS OF PRODUCTIVITY OF THE LEAVES Although it has been suggested that the persistent cloud cover, high humidity saturated soil, and the growth form of individual plants all in- fluence growth rate or development of the forest, we found additional factors worthy of mention. The very slow growth of some component trees within the elfin forest has been recorded by Wadsworth and Bonnet in their comparative study of the tabonuco (Dacryoides excelsa) rain forest and the colorado (Cyrilla) forest in Puerto Rico. Although Cyrilla racemiflora was not encountered in the elfin forest of Pico del Oeste, four other taxa of the colorado forest were. No distinctive growth rings have been seen in the woody trunks of the Pico del Oeste plants. Wadsworth and Bonnet grouped the trees in diameter-size classes and estimated the age of the trees by summing the period required for a plant to pass from one diameter class to another. They concluded that a 4” trunk of Ocotea spathulata was 200 years old; one of Micropholis garciniaefolia, 170 years old; and one of Calycogonium squamulosum, 80 years old. The annual growth rate for saw timber and polewood species in the Luquillo Mountains was 0.07 inches for Tabebuia rigida, 0.05 inches for Micropholis garciniaefolia, and 0.04 inches for Calycogonium squamulosum and Ocotea spathulata. They concluded that the soil is the common factor most important in the forests they studied. The saturated, poorly aerated organic soil inhibited root penetration and the absorption of water and resulted in the very slow growth. The cloud and fog cover, the high humidity and abundant rain docu- mented by the studies of Baynton suggest that photosynthetic activity in the elfln forest is low. We were unable to test the amount of photo- synthesis carried on by the component species. Tests of evaporation with potometers and of transpiration with cut branches within the forest were complete failures. Gates, however, demonstrated by infra-red temperature measurements that transpiration did occur during brief periods of sun- shine and clear sky. As there were longer periods, even days of full sunshine on the peak, the plants grew even though the growth rate was slow. A survey was made of the stomatal types, size, and distribution to determine any specializations that might occur within the elfin forest components (TaBLEs 3, 4). Sinnott suggested that xerophytes tend to ave a high stomatal frequency but cites no reference. Regrettably, we have failed to find any comparative data for other forest zones. 244 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Although stomatal apparatus types are commonly associated at the family level, there are variations and exceptions as reported throughout the work of Metcalfe and Chalk. We found the anomocytic type of stomatal apparatus (PLATE IIIa) to be most common as represented in 16 taxa of dicotyledons and 4 taxa of monocotyledons. The paracytic type (PLATE IVa) was present in 13 taxa of dicotyledons and 3 monocoty- ledons. Anisocytic type (PLATE IIIc) was present in 9 taxa of dicotyledons. The gramineous type (PLATE IVd) was present in all 6 taxa of Cyperaceae and Gramineae. A didymocytic stomatal apparatus (PLATE IIIb) was represented in 1 taxon of monocotyledons and 1 of dicotyledons (TABLE 3, column 9; TABLE 4, column 3). Stomatal openings of varying sizes were found in Justicia martin- soniana where large numbers of the stomatal apparatus appeared to abort before the final cell division which, we suspect, would have formed the guard cells. The openings, therefore, were of varying sizes. Pilea krugit (Pirate Va) also had stomatal apparatus of varying size with very small guard cells approximately 0.002 mm. long appearing over the veins, while mesophyll tissue was surmounted by guard cells averaging 0.023 mm. in length. Marcgravia sintenisii with heteroblastic growth showed the same number of stomata per square mm. for juvenile and adult foliage, but the guard cells were 0.037 mm. long on the juvenile leaves and only 0.028 mm. long on the adult leaves. The stomatal apparatus, however, appeared to be broader in the adult leaves. Stomata occurred in definite patterns in many of the monocotyledons, as expected, but they were found in groups of 2 to 3 or 3 to 7 in Gram- madenia (PLATE Vb) and in groups of 6 in Gesneria sintenisii and with 2 to 3 very closely associated, almost united, in Sauvagesia. On leaves of Pilea yunquensis stomata were found only on the upper surface of the leaf. Metcalfe and Chalk refer to work of Mohler, who found stomata on the lower surface in the species of Pilea he examined except for Pilea spruceana, where they were on the upper surface. Stomata tended to be oriented around the long hairs on Cleyera. Accessory or subsidiary cells to the guard cells were usually clearly defined and commonly contrasted with those of the epidermis. Unusually shaped subsidiary cells appeared in Alloplectus (Pitate IIIc) and in Justicia martinsoniana and in Renealmia antillarum (PLATE I1Id). he subsidiary cells had a characteristic homogeneous yellow-brown pigmentation in Clusia in contrast to the adjacent epidermal cells. In Tabebmuia rigida the subsidiary cells were generally clear in contrast to the mottled appearance of the epidermal cells. The walls of the subsidiary cells of Micropholis were straight in conspicuous contrast with the sin- uous walls of the other epidermal cells. No conspicuous elevation of stomatal apparatus was discerned in the components of the elfin forest. Torralbasia was the only taxon with the guard cells noticeably sunken and overlain by 6 epidermal cells (PLATE er. The length of the guard cells was measured and those found in 14 taxa “Y 1969 | HOWARD, ELFIN FOREST, 8 245 of monocotyledons averaged 0.035 mm. in length while in 39 taxa of dicotyledons the guard cells averaged 0.028 mm. in length. Within the monocotyledons the largest guard cells were in Eleocharis, measuring 0.048 mm. long, while /sachne had the smallest, 0.023 mm. long. Within the dicotyledons the longest guard cells were found in Hedyosmum arbores- cens and Peperomia emarginella, each 0.048 mm. while the smallest were those of Miconia pycnoneura, 0.010 mm. in len In considering the length of the guard cells in ise to the habit of the plant we found the following lengths: 8 taxa of herbaceous plants average 0.031 2 taxa of woody epiphytes average 0.030 24 taxa of trees or shrubs average 0.028 5 taxa of woody climbers average 0.027 The number of stomatal openings ranged from 18 per square mm. in Guzmania berteroniana to 230 per square mm. in /sachne angustifolia. Within the dicotyledons Peperomia emarginella had 22 stomata per square mm. while Miconia pycnoneura had 2230. While Guzmania had only 18 stomata per square mm., there were 96 stellate hair clusters in the same area. Vriesea sintenisii, another bromeliad, was examined in several sections of the leaf. The upper portion of a mature leaf showed 11.4 stomatal ap- paratus per square mm. with 41.4 stellate hair-glands in the same area; the middle portion of the leaf had 28 stomata per mm.” and 19 stellate glands, while a basal portion above the water level showed 49.4 stomata per mm.” and 19 glands in the same area. It has been suggested that metabolic activity of individual plants or leaves might be impaired by the presence of epiphyllous algae and leafy hepatics. The young leaves of most species within the forest are a bright green color when they first develop. In other species the young leaves were colored when young or expanding and developed a green color when near maturity. Young leaves of Rajania and Ipomoea, herbaceous vines, were bronze in color when young. Young leaves of the woody climbers Gonocalyx and Hornemannia were pink to red or orange-red in color. Cleye- ra and Symplocos also produced young leaves bronze in color, while Caly- cogonium had the young leaves reddish. The herbaceous Peperomia hernan- diifolia had reddish young leaves. Miconia pachyphylla was unique in losing the green pigments and having the leaves turn a bright red or Orange immediately before falling. Upon reaching mature size, the leaves of most species in the Pico del Oeste elfin woodland acquired a covering of epiphyllous non-vascular plants. Spores, gemmae, gemmalings and sporelings, or fragments of liver- worts are wind-borne and settle on the new leaves of most species. They appeared most quickly on species with depressed midrib or veins such as Marcgravia, Ilex, Symplocos, Gonocalyx or Tabebuia, and were rarely seen on the pubescent leaves of Miconia foveolata or the rugose leaves of M7- conia pycnoneura. Dr. Margaret Fulford, in work to be reported later, 246 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 examined a collection of leaves from plants within the forest and in 94 collection numbers found approximately 680 specimens belong to 40 genera and more than 75 species. There appeared to be an average of eight species per sample with a maximum of 18 species. Preliminary data showed the following epiphyllous species distribution on representative leaves: Anthurium dominicense Ardisia luquillensis Calyptranthes krugii Eugenia borinquensis Gonocalyx portoricensis Grammadenia sintenisii Ilex sintenisit Micropholis garciniaefolia Miconia pachyphylla Miconia pycnoneura Ocotea spathulata Peperomia emarginella Symplocos micrantha Trichilia pallida Wallenia yunquensis FOr RK DO UI —— COOnmarThy Kw OW + _ The number of epiphyllous taxa on any given leaf is not indicative of the leaf size or the percentage of the surface covered. The speed with which epiphyllae grew and covered the surface of the host was startling. Leaves of Eugenia borinquensis were completely and densely covered with liverworts in less than five months after leaf expansion. The amount of light reaching the photosynthetic area of the leaf is certainly reduced by the abundant epiphyllous growth. Epiphyllous growth occurred on leaves exposed at the summit of the canopy although the number of seem- ingly dead or desiccated plants was high. The leaves of the lower and inner branches of the forest components were more densely covered. Epiphyllae were less common on leaves of the truly herbaceous species. LEAF DAMAGE An additional factor in reducing the potential metabolic production of the plants in the elfin forest is evident in the amount of damage to the foliage of individual species. This has generally been attributed to wind. Damage done by animals as found in the Pico del Oeste forest has not been recorded. The sheared effect and directional growth of woody plants along sea coasts have been attributed to wind and to salt spray. Beard and Gleason and Cook have suggested the same factors are important in the shaping of the mountain-top forests in the West Indies. The effects of wind were observed in the canopy of Pico del Oeste. Within a few feet of the roof of our observation tower a slender stem of Eugenia borinquensis had worn a circular opening in the canopy of surrounding species. Branches of Tabebuia rigida were worn smooth through the cambium to the xylem by friction against each other due to movement in the wind. The leaves of Prestoea montana were broken and lacerated when they exceeded the shelter of the lee forests. The soft leaves of Psychotria berteriana were severely lacerated on a few plants growing in open areas. The soft flush growth of Hornemannia, Gonocalyx and Marcgravia was broken and leaves torn when the leading branches were whipped about in gusting winds. nice, _einenieenaeemmeneaninntin 1969 | HOWARD, ELFIN FOREST, 8 247 Microscope slides which were exposed to collect wind-blown particles also revealed crystals of salt. We failed to find any quantities of salt crystals on leaves or any indications of leaf damage due to salt spray from ocean storms. Apparently the large amounts of rain water or pre- cipitation from the clouds washed the leaves free of salt. The succulent young leaves were severely damaged by the populations of insects which existed on Pico del Oeste. In column 6 of TABLE 1 is recorded the percentage of leaves of each species that was affected by animal damage. The program of collecting foliage material for drying and future chem- ical tests permitted an assessment of damage to leaves on representative plants. Among the monocotyledons, animal damage was relatively light and only plants of Rajania cordata and Brachionidium parvum appeared to be severely affected. Many dicotyledonous species, however, were eaten with great regularity. In one plant of Clusia grisebachiana selected for study 98% of all leaves had been partially eaten and the reduction in leaf surface was 24%. These figures were computed by an actual count of the leaves which had been eaten by insects, and the degree of surface loss computed by reconstructing the outline where possible, and measur- ing the area lost by planimeter. Eighty percent of all leaves on repre- sentative plants of Haenianthus salicifolius, Wallenia yunquensis, Eu- genia borinquensis and Miconia foveolata were similarly eaten. Miconia showed a reduction in leaf surface of 35% and Eugenia borinquensis ex- hibited loss of 25%. The following table shows the plants most susceptible to insect damage. PERCENT PERCENT REDUCTION pH PERCENT DAMAGED IN SURFACE AREA WATE Clusia grisebachiana 98 24 3.9-4.3 66 Haenianthus salicifolia 86 19 5.1-5.2 62 Wallenia yunquensis 84 16 3.5-4.2 7S Eugenia borinquensis 81 25 4.8 30 Miconia foveolata 35 3.9-4.0 71 Miconia pachyphylla 79 21 3.7-4.3 60 Grammadenia sintenisii 77 _— 4.1-4.5 84 aan = racemosa 73 — 3.9-5.3 67 Rajania cordata 70 — 49 = Peberomin hernandiifolia 70 oa 4.6-5.1 —_ The nature of the leaf damage varied. In most cases the insect began on the margin and ate for varying distances towards the midrib. In other cases the apex of the leaf was chosen as the point of initial attack. Mi- conia pachyphylla and Miconia foveolata, with the characteristic reticu- late network of veins of the family, were characterized by holes in the leaves. No evidence was found of insect or animal attacks on the petioles or pulvini. Insect attacks seemed to be present at all months of the year. Damage to unexpanded primordia was rarely seen. Young leaves of a flush might be consumed completely as quickly as they began to expand in po Cn and Eugenia. Other leaves were attacked only as the 248 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 lamina developed. The majority of the damaged leaves persisted on the plant following the insect attack. The leaves subsequently developed cal- lous tissue or what appeared to be cork in many instances along the mar- gin of laminar tissues that had been eaten. Regrettably, we have been unable, up to this stage, to obtain scientific names or determinations for the insects seen or collected during this study. The following tabulation suggests that the food habits of the insects were often specific: Gray caterpillar which stings: Cleyera, Grammadenia, Miconia pycnoneura, Gray caterpillar with tufts of orange hairs and longer white hairs: JJex, Clusia. Slender green walking stick: Eugenia, Marcgravia. Tan-colored stouter walking stick: Miconia pycnoneura. Large spiny walking stick: Ardisia, Calycogonium, Micropholis, Wallenia. A weevil: Calycogonium. Spittle bugs: Eugenia. Green grasshopper with white lines: Tabebuia. Black grasshopper: Tabebuia. Leaf hopper: Cyathea. Leaf miners: Hornemannia. A slug (Gaeotis nigrolineata Shuttleworth): Lobelia. A snail (Luquillia luquillensis Shuttleworth): Lobelia. Gall-producing insects: Ocotea. No domatia were encountered in leaves of species within the elfin for- est, although domatia were found in other species of plants in forests of lower elevations. The nature of the attraction in the leaves of the component species to the insects cannot be determined. It was evident that the leaves varied in their texture, the amount and the color of the liquid within the tissues, their aromatic constituents and the pH of the cell contents. Clusia (Gut- tiferae), Micropholis (Sapotaceae), and Lobelia (Campanulaceae) pos- sessed a latex, as is characteristic for the families involved. Ninety-eight percent of the leaves of Clusia, 33% of the leaves of Micropholis and 11% of the leaves of Lobelia were damaged by insects or snails. [pomoea reé- panda also has cells containing a yellow material, although this did not flow when the leaf tissue was broken, and 61% of the leaves of this plant were damaged by insects. Hedyosmum arborescens, Calycogonium squamulosum, Miconia foveolata, and Symplocos micrantha could be clas- sified as “bleeders,” for the leaves, petioles or stems exuded a clear liquid when cut or broken. The percentage of leaves damaged by insects in these taxa were: Hedyosmum, 61%; Calycogonium, 21%; Miconia foveolata, 81%; and Symplocos, 14%. Aromatic principles were present in the leaves or bark of some species and can be described as follows: Calycogo- nium — cider odor; Llex—odor of hay; Mecranium — sweet; Miconta foveolata — rank; Miconia pachyphylla — sweet; Miconia pycnoneura — sweet; Symplocos — rank and acidic; Tabebuia — medicinal; and Wal- lenia — odor of spinach. The nature of insect damage in these taxa 1S reported in TABLE 1, column 6. 4 1969 | HOWARD, ELFIN FOREST, 8 249 In the preparation of several pounds of dried material of leaves or branches for shipping and subsequent chemical analysis, it was evident that the species of the elfin forest contained different amounts of liquid and solid materials. Large quantities of certain plants would produce only a few pounds of dry weight material while other species clearly had less liquid to evaporate. Standard weight samples of leaves were obtained and dried in an oven to obtain the percentage of water in the leaves of each species. For mature leaves the percentage of water ranged from 93% in Psychotria guadalupensis and Begonia decandra to 44% for Calyp- tranthes krugii. Only 1% of the leaves of Psychotria guadalupensis were damaged by insects, 26% of the leaves of Begonia and 41% of the leaves of Calyptranthes. In the list of 10 most severely damaged species pre- viously given, the percentage of water in leaf tissue varied from 60% to 84% of the taxa for which we have data. Clearly some factor other than the amount of liquid in the plant tissue was responsible for the in- sect damage. The abundance of liquid in some leaves led us to a simple measurement of the pH of the plant liquid which could be extracted (TABLE 1, column 7). Further details on these tests will be given in a later paper. For each species, leaves of a size normally eaten by insects were crushed between clean microscope slides and several drops of liquid were tested immediately with a Beckman pH meter. The acidity varied from 2.4 in Begonia de- candra to 6.5 in Justicia martinsoniana. The values of the plant sap in the 10 most commonly eaten species ranged from 3.5 in Wallenia yun- quensis to 5.3 in Hornemannia racemosa but averaged 4.4. LITERATURE CITED Baker, J. R. Rain-forest in Ceylon. Kew Bull. 1938: 9-16. 1938. Baynton, H. W. The ecology of an elfin forest in Puerto Rico, 2. The micro- climate of Pico del Oeste. Jour. Arnold Arb. 49: 419-430. 1968. The ecology of an elfin forest in Puerto Rico, 3. Hilltop and forest in- fluences on the microclimate of Pico del Oeste. /bid. 50: 80-92. Bearp, J. S. The Natural be spina of the Windward and Leeward Islands. Oxford Forestry Mem. 21. 192 pp. Clarendon Press, Oxford. ; Berc, O. Myrtaceae. In: Aart s, C. F. P. von, FI. Brasil. 14*. 1-527. 1858. Bews, M. A. Studies in the sper eae hee evolution of the angiosperms. New Phy- tol. 26: 1-21. 1927. Brown, W. H. Vegetation of the Philippine mountains. Bureau of Sci. Publ. . ; ‘ a Carn, S. A., G. M. pe Oxiverra Castro, J. M. Prres, & N. T. pa Sttva, Appli- cations of some phytosociological ideas > Brazilian rain forest. Am. Jour. Bot. 43: 911-941. 1956. Carxouisr, S. J. ere plant anatomy. 146 pp. Holt, Rinehart & Win- ston, N.Y. 1961 Cooper, W. C. The broad-sclerophyll vegetation of California. Carnegie Inst. Publ, 319. 1922 Corner, E. J. H. Wayside trees of Malaya. ed. 2. Vol. 1. Singapore. 195 Garrison, R., & R. H. Wetmore. Studies of shoot tip dec Syringa ok Am. Jour. Bot. 48: 789-795. 1961. 250 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Gates, D. M. The ecology of an elfin forest in Puerto Rico, 4. Transpiration rates and temperatures of leaves in cool humid environment. Jour. Arnold Arb. 50: 93-98. 1969 Grit, A. M. The ecology of an elfin forest in Puerto Rico, 6. Aérial roots. Jour. Arnold Arb. 50: 197-209. Gueason, H. A., & M. T. Coox. Plant ecology of Porto Pops Pts. 1 and 2. Sci. Surv. Porto Rico Virgin Is. 7: 1-173 + 50 pls. 19 Hottum, R. E. On periodic leaf-change and flowering ep pane in Singapore. Gard, Bull. Straits Settlements 5: 173-206. 1931. Howarp, R. A. The ecology of an elfin forest in Puerto Rico. 1. Introduction and composition studies. Jour. Arnold Arb. 49: 381-418. 1968. Jackson, B. D. A glossary of botanical terms. ed. 4. 1928. London. Juncer, J. R. Anpassungen der Pflanzen an das Klima in den Gegenden der regenreichen Kamerungebirge. Bot. Centralbl. 47: 351-360. 1891. Kraerskou, H. Myrtaceae ex India occidentali a dominis Eggers, eee Sintensis, Stahl aliisque collectae. Bot. Tidsskr. 17: 248-292. pls. 7-13. 1890. Leprun, J. Répartition - la forét équatoriale et des formations végétales limitrophes. Brussels. 1936. McVauecH, R. Tropical caw Myrtaceae. Fieldiana Bot. 29: 145-228. 1956. Mercatre, C. R., & L. Cuatk. Anatomy of the dicotyledons. 2 vols. Oxford, Clarendon Press. 1950. Mouter, P. Beitrage zur pharmakognosie der Urticales. Anatomie des Laub- blattes. Thesis. Basel. pp. 84. 1936 [Not seen Puitport, J. A blade tissue study of forty-seven species of Ficus leaves. Bot. Gaz. 115: 15-35. 1953. RAUNKIAER, C. The life-forms of plants and statistical plant geography. 632 Ricuarps, P. W. The tropical rain forest. 450 pp. Cambridge Univ. Press 1964. [Second photocopy of 1952 edition. SHREVE, F. A montane rain-forest. Carnegie Inst. Publ. 199. 110 pp. Washing- ton, D.C. 1914. Smnnott, E, W. Plant morphogenesis. x -+ 550 pp. McGraw Hill. New York. STAALFELT, M. G. Morphologie und anatomie des Blattes als Transpirationsorgan. Handb. der Pflanzenphysiol. 3: 324-341. 1956. VauGHAN, R. E., & P. O. WiEHE. Studies on the vegetation of Mauritius. il. The structure and development of the upland climax forest. Jour. Ec ol. 29: 127-160. 1941 WapswortH, F. Thirteenth Annual Report. Carib. Forester 14: 1-33. 1953. & aN Bonnet. Soil as a factor in the occurrence of two types of mon- tane forest in Puerto Rico. Carib. Forester 12: 67-70. WarMIno, E., & P. GRAEBNER. Lehrbuch der dkologischen Pflanzengeographie. 988 + (64) pp. ed. 4. Berlin. 1933 Watson, R. W. The mechanism of elongation in palisade cells. New Phytol. 41: 206-221. 1942. WIESNER, - hap Absteigende Wasserstrom und dessen physiologische Bedeutung. Bot . 47: 24-29. 1889 WYLIE, - ms Relations between tissue organization and vascularization in leaves of certain tropical and subtropical dicotyledons. Am. Jour. Bot. 33: 721-726. 1946. . Leaf organization of some woody dicotyledons from New Zealand. Am. Jour. Bot. 41: 186-191. 1954. Table 1 CoLtumNn 1— Total number of leaves per plant. Cotumn 2— Total photosynthetic area in cm.” CoLtuMN 3—average blade are cm.” Corumn 4— Ratio, blade length:width. Corumn 5— Ratio, blade length: Laden length. Corumn 6 — Percent of gre prover by animals. Cotumn 7 —pH of leaf sap. Cotumn 8 — Percent of water in leaf tiss COLUMN NUMBER 1 2 3 4 5 6 7s 8 GRAMINEAE Arthrostylidium sarmentosum 168 523 12 ipa sessile ms — — Ichnanthus pallens 12 97 8.1 re | sessile 0 — — Tsachne angustifolium 18 144 8.0 10:1 sessile 0 = — CYPERACE Carex nae chya 27 2240 11.9 38:1 sessile 0 — — Eleocharis yunquensis 54 609 — terete sessile 0 — — Scleria secans 14 3101 62.0 70:1 sessile 0 — — PALMAE Prestoea montana 10 289,460 28,946.0 — -— 0 5.1-6.2 — ARACEAE Anthurium dominicense 10 918 91.8 6.8:1 Ast 0 5.0-5.8 _ BROMELIACEAE Guzmania berteroniana 20 4,392 225.0 — sessile 0 — Vriesea sintenisii 19 1,581 80.0 Svial sessile 0 oct 83 DI0SCOREACEAE Rajania cordata 20 370 18.0 2 S01 Lt 70 4.9 — ZINGIBERACEAE Renealmia antillarum 8 736 92.0 S071 constr. 0 4.8-5.3 — 8 ‘LSAYOA NIATA ‘GUVMOH [6961 Sz [voL. 50 JOURNAL OF THE ARNOLD ARBORETUM 9°S-6'F Oo 1:6 T:8°Z SLO 6'EST S0z DjIa49 Disadvanvs 252 AVAOVNHOO O'S-L'b i 1:81 1:Z Ll Z9L'T 622 DYyOofiaund vISvg]D440, 7 AVAOVALSV1a) 'S—T'S Or T¢°S el LT TE9‘ST $89'8 usmajuis Xaq] AVAOVITOAINOY 9°S-2's 91 1:22 1:9'2 Vor 819'T OT opyjod vytyot4 J AVAOVITA I'S-8"p SZ T:L2 TiZ'T I'bz zsr‘OI feb Dyvinyyjogs 094090 avaovanv’y] b'9-S'S bT he a Lot 0°6 OOI II CJT 19B1v[) stsuanbunk vattd O'9-1'S ZI 1:01 1:8°Z $°8 LET 91 CHL dosrey) minsy vapid AVAOVOILA) I's ) 1:9°T ET S'6b2'T LOb‘L 9 vyoyjad mdos9ag AVANVAOW 9'S-S*b 19 L352 tee 161 806 ‘bb 6FE'Z SUAISALOGAD MNUSOKpaH AVAOVHLNVAOTHS) 9°S—9'b OL IT TZ'T o'r Obl Or pyofupunusay vimodsadad Ss-05 0 IT IT 61'0 ZI z9 DauisAviMa Dimosadag aVAOVAdAdI _— 0 "Indap itt 08 Ott rat DUDIUOML SHMMONLGT = SZ IZ ca a4 s'T SL t unadsogd unipiuoiysDvag AVAVGIHOAGC) L 9 s b £ Zz I wadWAN NWAT09 HOWARD, ELFIN FOREST, 8 253 1969] 1:92 alIssas fOr EZ 609‘8T 601'T 9S6'T 199° p78'eT O82‘T Sisuanbunk viuajyp usiuaqus DiuappUwuDdty sisuayimbny vistpay AVAOV NISHA DSOMAIDA DIUUDUWANAO HT sisuant4oj40g xk]DI0U0 aVAOVOTay Danauourkd pimor pw AVAOVLVWOLSY Taf sisuanbuisog viuaing mindy sayjunadajoy AVAOVLUAT, Dépudrap DiNodsag AVAOVINODAgG Duplyspqgasi4s DIsN{D aVaadILLAs) pypjaungoqp vaakaty aVAOVAH L, yeoy ynpe jeay apruaant USIUMIBIUIS DIADATIAD AVAQVIAVADOAV I COLUMN NUMBER 1 2 4) 4 5 6 | 8 SAPOTACEAE Micropholis garciniaefolia 10,487 96,480 9.1 1,6:1 8.0:1 33 4.1-4.8 52 SYMPLOCACEAE Symplocos micrantha 32 268 8.3 233i 20.0: 1 14 4.0-4.8 60 OLEACEAE Haenianthus salicifolius 1,294 17,339 13.4 V foal a 8.0:1 86 5.1-5.2 62 CONVOLVULACEAE Ipomoea repanda 18 205 113 25 5.031 61 4.5-6.2 —_ BIGNONIACEAE Tabebuia rigida 2,680 60,040 22.4 19:1 8.3:1 60 5.0-5.5 ip GESNERIACEAE Alloplectus ambiguus 16 168 10.5 22°) 9.5:1 43 4.6-5.5 _ Gesneria sintenisii 256 9,022 35:2 2321 8.9:1 67 5.3-5.7 84 ACANTHACEAE Justicia martinsoniana 9 47 52 2521 11.4:1 0 4.1-6.5 — RUBIACEAE Hillia parasitica 48 564 SB vy ie | 5.6:1 29 4.7-4.9 83 Psychotria berteriana 305 18,093 59.3 20:1 4.5:1 62 5.0-5.7 85 Psychotria guadalupensis 213 410 1.9 2.0:1 1.4:1 1 4.9-5.0 93 CAMPANULACEAE Lobelia portoricensis 416 20,300 48.7 3.4:1 4.5:1 11 4.7-5.0 79 COMPOSITAE Mikania pachyphylla 46 274 5.9 1.4:1 4.5:1 10 5.1-6.0 —- SZ WOLAXOUAV GAIONYV AHL AO TYNYNOL 0s “Toa] Table 2 Cotumn 1—Leaf thickness in ». CoLumMn ead a oe . x, absent 0). Corumn 3 — Lower hypodermis. oo 4— Ratio, upper epidermis:upper hypodermis. sear 5 — Upp e. CoLuMN 6 —Lower cuticle. Cotumn 7—Crystals (ra raphides, rh rhombic, d = druses, f = vas CoLu bee ee? asts or sclereids present. CoLtumNn 9 — Multiple edi jayer. CoLUMN 10 — Ratio of palisade layer to spongy mesophyll. Isodiam. = isodiametric cells only COLUMN NUMBER 1 2 3 4 5 6 7 8 9 10 PIPERACEAE Peperomia emarginella 275 Ps 0 1:10 0 0 d,ra 0 0 4. Peperomia hernandiifolia 625 x 0 1:3.0 0 0 d 0 0 isodiam. CHLORANTHACEAE Hedyosmum arborescens 311 2% 0 i332 0 0 0 0 0 ey. MoRACEAE Cecropia peltata 91 0 0 —_ 0 0 d 0 0 1:0.39 URTICACEAE Pilea krugit 146 0 0) —_ 0 0 f re) 0 jal Pilea yunquensis 146 0 0 _ 0 0 f 0 O 131 LAURACEAE Ocotea spathulata 475 2x 0 Mia x x 0 0 x RE oe: MELIACEAE Trichilia pallida 209 0 0 _ x Ps rh 0 0 1322 AQUIFOLIACEAE lex sintenisi 421 0 0 —_ x x d 0 x 1:1.8 CELASTRACEAE Torralbasia cuneifolia 458 0 0 — X = d 0 x 1:4.5 OCHNACEAE Sauvagesia erecta 141 X 0 io .; 0 0 d 0 0 is [6961 8 ‘LSAYOI NIATA ‘GYVMOH JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 256 oo i oo oo oo HOO uvTU yi oo x Oo oo oo 6 xO xO Me MK om 0 ) 67€ = MsMmazMIS DILAapDMUDAsT) se 0 0 vLZ Sisuayinbny vis~pay AVAOVNISUAJL TT fe) XZ b8E DSOWmarDs DINUDMAUAO H Tt 0 x $29 sisuartsojiog xxjpI0U0y avaovorny IT 0 x 602 panauourkd mmonpy 7:1 0 x SoZ oyxydkyovd piuomnpy TT 0 x O8T Dynjoanof vmMonW ae 0 x O8l wmnuyopskup wniupisa yy ae 0 XZ 262 unsomnuonds UNIMOSOIKIDI AVAOVLVWOLSV Tay ome 0 0 zs sisuanbutsog miuaing a 0 0 6bS ninsy sayquvigdaqvy AVAOVLAA L’b:T x x IT¢ papuvzap piuosag AVAOVINODAG SOL 0 Xo—p L8L Duby IDgAastAS vIsN{D avaaalitar i 0 0 z1s pyojaundgoqpy vsakata aVAOVAH J, — 0 0 cis Mswmajuts DInDssI4D AVAOVIAVADOAV IAT b £ Zz I qad#ANON NWA109 HOWARD, ELFIN FOREST, 8 257 1969] “WIeIpOst “UIRIpOsT oo oo ooo 1 oo xO oo * x OO oo HO K oo¢ oycydakyovd viupyryw avLisodNog sisuantsoj4og pyaqoT AVAIVINNVANVS sisuadnippons viaqoyoksg DuIAaI4ag MLAJOYIKS DaysD4vg DIL aVAOVIAN Y DUDINOSUJADI «DIIYSHS aVAOVHLNVOY ustMaquIs DidaUusary sunsiqup snyradoynpy AVAOVINANSA‘) Dpisid DINgaQvy AVAOVINONODIG AVANVTINATOANOD) ppuvngas vaomod] SnyOflayDvs snyzuDiuaD aVAOVAIO DYIUDAIIUL SOI0IGUKS AVADVIOTAWAS pyofaniui9408 syoydosn py AVAIVLOAVS sisuanbunk pimayv 4 ear i i aa tre a JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 258 sAs0wouy =— «OFT Z£0" L-02 a ad Ss V Ss Dyoftaund vIspgq]D440. | AVAOVALSVTa,) mAsowouy — OFT €20° L-02 b-¢ ad V V S ustuazuts xa] ] AVAOVITOAINOY ayAsowouy = $02 ze0" -9 I — KR >V Ss ppyyog vyryrtd L AVaAOVITAY mnAoeed = 90F zeo0 SST (4) e-1 mes S V L pypjnyyogs 094090 avaovanv’y] sn ADOWOUY OIl ¢Z0° - “ _ Ww Vv H sisuanbunk Ddajtd onAsostuy = $9 £20" m= 2 _ W V H uindy Daft AVAOVOILAL) mAoowouy = 00 810° - 9) = V V L vyoyjad 0140199) aVAOVAO on ADoWOUYy cL 9b0° - id¢ — S Pi Ss suassasogav wunusokpa fT AVIOVHINVAOTH)D ondoostuy = $9 Z£0" L 6 a WwW L H vyofupurusay viwmosadad mA00wlouy = ZZ 9b0° - o oT WwW di H myjauisaniua Diuosagad avaovaadl 6 8 Z 9 S + e Zz I waaWaAN NWNTO9 “snqeiedde yeyeuroys jo adh. —6 NWATOD “wut “bs jod vyewo}s Jo Jaquiny—g NWaATOD ‘wut ur s]je9 prend jo yIBueT— ZL NWA -102 ‘pajuaseidas saysny jo Jaquinu pue seave, jo Joquinu jo sajduiexa svafo 0} s9jo1 sioquinu poyeusydAyy “juasaid saavat JO Joaquin — 9 NWATOD “Ysny ¥ ul paonposd saavay Jo sired 10 saava] OF Jojo stoquinu ‘juejd uo AJsnonunuos Zuuvadde = Q ‘uonsnposd yea jo adAy, —s§ NWOTIOD “xoeq-aIp MoyYs IO j10Ge sopoueig — NWOATOD ‘snowojJoyIg = d ‘away = V fyeipodwAg = g ‘[vipodouo-_w = W :Surppuviq 10 yIMoIZ Jo sdk —¢ NWATOD ‘shos0pyNey = O ‘aqyeusayy = VY f[euyuuay = | :auassaioyut jo uonlsog —Z NWAT0D ‘qhydidg = J {equi snoarqiey = OH ‘equi Apoom = OM !9L = L fanys = § ‘q9H = H ‘queyd yo W1qeH — 1 NWAT0D € 9198.1 HOWARD, ELFIN FOREST, 8 1969] on Adeleg on AIOstuy wn Ad0stuy I ANvIEg on Adevieg InAIOWOUY on Av0stuy on AdOstuy on AdvIeg oT ADvIeg op ADIVT oN AIOWIOUY or Ad0uwlouy a1 ADvIeg OL2 LLT £81 $¢0° 9£0° LEO" 8¢20" cro’ —— Or a 8 dd aoa Wun AaxzAAA How O00 MMMM WN Sisuanbunk piuayp 4 SMaIUIS DiNapoUMDsD Sisuayinbn visipay AVAOV NISUA DSOWIIDA DIUUuDUWALAO $18u99140}40g XKJDI0U0L aVaOvoOINy Danauourkd vinonp unsojnuonbs wniuo30rKo7 AVAIOVLVWOLSV 14a, Sisuanbutsog piuaing usnay Sayquoigdakyoy AVAIOVLYAT, Dépunrap viMosag AVAOVINODA Duviyavgasi4s visnyD aVaadILLAS) pypjoundoqp viakayy avaovadH |, USIMAIUIS DIADAIIAD Py AVAIVIAVADONV JL 019948 Disasvanvy AVAIVNHIOO COLUMN NUMBER 1 2 3 4 5 6 7 8 9 SAPOTACEAE : Micropholis garciniaefolia i A A DB 2-6 20-4 027 230 Anomocytic SYMPLOCACEAE Symplocos micrantha S A A — 2-6 - .030 230 Paracytic OLEACEAE Haenianthus salicifolius sg DS — Se - 025 327 Anomocytic CONVOLVULACEAE Ipomoea repanda wc A A,T DB 9- 032 110 Paracytic BIGNONIACEAE Tabebuia rigida T s D DB 1-2 pr 10-4 .023 230 Anomocytic GESNERIACEAE Alloplectus ambiguus H A M DB i 22- 032 56 Anisocytic Gesneria sintenisii S A M DB c 20- 035 168 Anisocytic ACANTHACEAE Justicia martinsoniana H si D — — - 025 170 Paracytic RUBIACEAE Hillia parasitica S,E T D — a - 037 100 Paracytic Psychotria berteriana S 9 D — 2 pr 3- 029 210 Paracytic Psychotria guadalupensis S,E 4 YD — 8- .025 110 Paracytic CAMPANULACEA Lobelia portoricensis SH T A DB — 37- .040 190 Anomocytic COMPOSITAE Mikania pachyphylla & A,T M — —_ - 035 196 Anomocytic ~S a “= 092 WOLHYOdUY GIONYV AHL AO TYNUNOL OS “I0A] 1969 | HOWARD, ELFIN FOREST, 8 261 Table 4 STOMATAL STOMATA TYPE TAXON SIZE IN MM. NUMBER OF STOMATAL MM * APPARATUS GRAMINEAE Arthrostylidium sarmentosum 0.027 205 Gramineous Ichnanthus pallens 0.043 140 Gramineous Isachne angustifolium 0.023 230 Gramineous CYPERACEAE i Carex polystachya 0.041 102 Gramineous Eleocharis yunquensis 0.048 120 Gramineous Scleria secans 0.030 110 Gramineous PALMAE ; Prestoea montana 0.025 140 Paracytic ARACEAE ; Anthurium dominicense 0.046 56 Paracytic BROMELIACEAE J ; Guzmania berteroniana 0.044 18 Didymocytic Vriesea sintenisii 0.039 28 Didymocytic DIOSCOREACEAE - Rajania cordata 0.035 120 Anomocytic ZINGIBERACEAE ‘ Renealmia antillarum 0.027 140 Paracytic ORCHIDACEAE , ; . Brachionidium parvum 0.039 47.5 Didymocyt “i Dilomilis montana 0.032 120 Anomocytic ARNOLD ARBORETUM HARVARD UNIVERSITY 262 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 EXPLANATION OF PLATES PLATE I a. Terminal bud of an adult branch of Marcgravia sintenisii. b. Stem of Hillia parasitica showing the flattened stipular sheath. c. Stem apex of /lex sintenisu i a eria and enclose and protect the terminal bud. g. Three views of the stem apex an terminal leaf pair of Calyptranthes krugii, left figure shows the mature leaves ; central figure shows the plicate stipule pair separated at the base; right figure shows the plicate stipule pair separated at the apex. PLATE II PLATE III Epidermal cells and stomatal apparatus of: a, Miconia pachyphylla; b, Brachionidium parvum; c, Alloplectus ambiguus; d, Renealmia antillarum. P PLATE IV Epidermal cells and stomatal apparatus of: a, Anthurium dominicense ; be Hillia asitica; c, Torralbasia cuneifolia; d, Scleria secans; e, Haenianthus salicifolius var. obovatus; £, Gonocalyx portoricensis. PLATE V a. Lower epidermal surface of Pilea krugii showing the cluster of small sto- mata over a vascular bundle, a multicellular gland, and two of the larger stomatal apparatus. b. The lower epidermal surface of a leaf of Grammadenia sintenisu showing the stomatal apparatus grouped in clusters of three. Jour. ARNOLD ARB. VoL. 50 PiaTeE I Howarp, ELFIN ForEST, 8 Jour. ARNOLD Ars. VOL. 50 Howarp, ELFIN Forest, 8 ty Or os €; ‘s Sbes Da Ny Ga QS eo: CRE oe 1% as par, e Howarb, ELFIN Forest, 8 268 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 LECTOTYPIFICATION OF CACALIA L. (COMPOSITAE-SENECIONEAE) + BERYL S. VUILLEUMIER AND C. E. Woop, Jr. Tue INTERNATIONAL Cope of Botanical Nomenclature (1966) provides, by means of nomenclatural types, stability in the application of names at, or below, the rank of family. The use of a name is determined by the no- menclatural type of that name, and changes may result when, or if, the choice of a type is shown to be incorrect. To avoid disadvantageous changes in the application of names provision is also made in the Code to preserve current usage and to avoid the confusion which could result from varying opinions concerning the choice of a lectotype. The sene- cionid genus Cacalia, as circumscribed by Linnaeus (1753, 1754) was quite heterogeneous, and the process of choosing a lectotype has been both complicated and subject to individual interpretation. Three dif- ferent species have been proposed as lectotype, and even now there are conflicting opinions (Cuatrecasas, 1960, and Pippen, 1968, vs. Pojarkova, 1961, and Vuilleumier, 1969). It seems expedient to review the typifica- tion of this genus once more. Cacalia as typified here (by C. hastata L., a choice made by Kitamura, 1942) provides an example in which the choice of a lectotype in accordance with the guides outlined in the Code also complies with the recommendation (7B) that a lectotype be so chosen as to preserve current usage. In the first edition of Species Plantarum (2: 834-836. 1753), Lin- naeus described and named ten species of Cacalia which he divided into two groups: ‘‘Frutescentes,”’ consisting of four shrubby species, and ‘“Herbaceae,” with six herbaceous species. In 1754, Miller (Gard. Dict. Abr. ed. 4. 2: ord. alph.) split Cacalia, placing the four shrubby species in Kleinia Mill. Cacalia was thus restricted to the herbaceous species, and it is from the six original species of this group that all choices of lectotype have been made. To our knowledge, no one has suggested that Cacalia be typified by one of the species removed to Kleinia. There has been, however, considerable disagreement as to which of the six herbaceous species should be designated as lectotype. Rydberg (1924) concluded that Cacalia alpina should be the type species; Cuatre- casas (1955, 1960) and Pippen (1968) have concurred in this choice. e of an informal series of peripheral papers arising from research toward a Generic Flora of the Southeastern United States being carried on through the gen- erous help of the National Science Foundation (Grant GB-6459X, C. E. Wood, Jr., principal investigator). We acknowledge with thanks the assistance of Dr. Bernice G. Schubert and Dr. Elizabeth Shaw and we are indebted to Andrey I. Baranov for his translation of a portion of A. Pojarkova’s treatment of Cacalia in Flora URSS. 1969 | VUILLEUMIER & WOOD, CACALIA 269 Hitchcock and Green (1927) chose another type, C. atriplicifolia, ene Pojarkova (1960) also adopted. Kitamura (1942), however, ma third choice, C. hastata, and Pojarkova (1961), changing her pact agreed with him. Quite independently, Shinners (1950) came to the conclusion that either C. suaveolens or C. hastata should be the lectotype species. After a careful review of the arguments and of the nomen- clatural and taxonomic history of Cacalia in conjunction with the appli- cation of the Rules and Recommendations of the Code, we are convinced that C. hastata should be the lectotype species. Since the nomenclature of species of vascular plants begins in 1753 and that of genera in 1754, in most instances the application of a name prior to 1753 should be given little weight relative to a post-1753 application, especially since Linnaeus not infrequently reapplied older names in a completely different sense. However, one of the principal arguments ad- vanced, first by Rydberg, and then by Cuatrecasas and Pippen for the selection of Cacalia alpina as the type species of Cacalia is an historic one. Rydberg wrote (loc. cit., p. 370), “Of the species of the second [herba- ceous| group only the last two, Cacalia atriplicifolia and C. alpina, had been known as Cacalia before Linnaeus’ time.2 The oie Cacalia, applied to the last one, dates back to Vaillant and L’Obel. C. alpina L. or Ade- nostylis alpinus is therefore the historical type of Cacalia.” Cuatrecasas (1960, p. 182) reiterated, “There is no doubt that the name Cacalia was first applied to C. alpina and that Linné had this species in mind when he established the genus in his Genera Plantarum. Therefore, Cacalia alpina is the type of Cacalia.’ Pippen added (1968, p. 377), “It seems clear that C. alpina L. is the most logical lectotype of Cacalia. This species, named C. alpina by Linnaeus (1753), embodied the Linnaean and pre-Linnaean concept of Cacalia in that essentially all of the species of Cacalia described by pre-Linnaean botanists (L’Obel, 1581; Clusius, 1601; Bauhin, 1623; —— 1699; Tournefort, 1700) actually repre- sented the same species The adoption of this ‘historical argument would restrict the name Ca- calia to a genus consisting of four or five species of Europe which has been known since 1816 as Adenostyles Cassini. Contrary to all three authors, however, the use of arguments concerning the application of names before the starting point for botanical nomenclature can result only in further confusion, as is shown below In the first edition of Genera Plantarum (1737, p. 252), Cacalia in the sense of Tournefort (that is, C. alpina L.) is found in the synonymy of Tussilago: “TUSSILAGO*. Tournef. 276. Vaill. A. G. 1720. f. 46. Cacalia *In the next paragraph Rydberg added, “. . . only Linnaeus himself had used Cacalia for C. suaveolens in his Hortus Upsaliensis.” Rydberg probably was misled by Linngeus’s mace viet to Hortus Upsaliensis immediately following the diagnostic phrase name cif ame) 0 itp suaveolens in Species omaha (1753, p. 835). In Hortus UgeaBions eaten p. - a byeevgy this a species of Kleinia (“Kleinia caule Shee ceo,” etc. ); the diagnos the same in sage ies Plan- tarum, except that Cacalia has been Sou wea for Kleinia, We have not found any mention of Cacalia in Hortus Upsaliensis. 270 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Tournef. 258. Petasites Tournef. 258. Vaill. A. G. 1719.” Following the description the observation is added: “‘Cacalia T. caule ramoso est, & corollulis hermaphroditis quadrifidis, sine radio ligulato.” On the same page is found Kleinia: “KLEINIA. Cacalianthemum Dill. elth. 54. 55. An Tithymaloides ? Klein. monagr.”” The corolla is described as with the limb “quinquefido, erecto,” and the stigmas as “duo, oblonga, revoluta.” In the second edition (1742, p. 401), the synonymy of Tussilago is am- plified by the addition of “Vaill. A. G. 1719.” to the reference to Cacalia, and to that of Kleinia (p. 394) is added “Porophyllum Vaill. A. G. 1719. t. 20. f. 39.” which Linnaeus had maintained as distinct in the Hortus Cliffortianus (1738, p. 494). Neither description was changed in any way in this edition. The synonymy and descriptions of Tussilago and Kleinia in the editions of 1743 and 1752 are identical with those of the second, but these editions were not prepared by Linnaeus. In the fifth and crucial edition (1754, p. 362), Linnaeus treated Cacalia of Vaillant and Tournefort as congeneric with Kleinia and combined the two under the name Cacalia: “CACALIA.* Vaill. A. G. 1719. Tournef. 258. Kleinia edit. prior. Cacalianthemum Dill. elth. 54. 55. An Tithyma- loides ? Klein. monagr. Porophyllum Vaill, A. G. 1719. #. 20. f. 39.” Most interestingly, the generic description is identical with that of Kleinia of the first four editions of Genera Plantarum! There is no mention of the tetramerous corolla of Cacalia alpina which had appeared in earlier editions under Tussilago. Rydberg argued (/oc. cit.) that “Linnaeus’ description of the genus [Cacalia] points to this species [C. alpina] especially the description of the style tips: ‘Stigmata duo, oblonga, revoluta.’ This is characteristic of Adenostylis alpinus which on account of its oblong style branches had been placed in the tribe Eupatorrear, but which Dr. B. L. Robinson rightly restored to the SENECcIONEAE. C. atriplicifolia as well as C. suaveolens has a true Senecioid style, with truncate style-branches.” Cuatrecasas further argued, “Among all the species of Cacalia in Linné’s Species Plantarum, Cacalia alpina is the only one with elongate, curled stigmas and 4-merous corollas.” Cuatrecasas is correct, but Linnaeus’s description (1754) of the corollas of Cacalia as five-fid and the stigmas as “‘duo, oblonga, revoluta” does not apply to C. alpina but to the species which he had formerly placed in Kleinia, a name which he abandoned in favor of Cacalia in 1753. Moreover, the description of the corolla and stigmas is precisely the same as in Senecio (Gen. Pl. ed. 5. 373). A note under Cacalia alpina in Species Plantarum (1753, p. 836) reads: “Hanc speciem genere cum antecedentibus convenire docuit autopsia, hinc genere conjugenda: Cacalia cum Kleiniis.” In the second edition (1763, p. 1171) this note is clearer and is amplified: “Hanc speciem genere cum antecedentibus convenire docuit autopsia, hinc genere conjugendae Ca- caliae cum Kleiniis. Calyx hujus speciei flosculis 3 s. 4 tantum.” We translate this to read: ““My observation has shown this species to agree generically with the preceding ones; hence the Cacalias are to be joined in a genus with the Kleinias. The involucre of this species with only 3 1969 | VUILLEUMIER & WOOD, CACALIA 271 or 4 florets.” From this comment and the placement of the species last in the genus it appears that Linnaeus regarded C. alpina as somewhat aber- rant in, but belonging to, the genus which he had formerly called Kleinia. The use of the plural of Cacalia also suggests that he had in mind more than one species. Linnaeus did not change the description of Cacalia in the sixth edition of the Genera, but the authors of the seventh and eighth editions noted the departures of C. alpina from the others of the genus. Reichard (Gen. Pl. ed. 7. 411. 1778) observed, “C. alpina foliolis calycis conglutinatis corollulisque quadrifidis differt.”’ Schreber (Gen. Pl. ed. 8. 545. 1791) in- cluded in the description of the corolla “limbo quadri-f. quinquefido, erecto,’ and Haenke in his edition (Gen. Pl. ed. 8. 709. 1791) had pre- cisely the same description and observation as Reichard. Cuatrecasas (1960, p. 182) attributes a comment to Schreber (1791, p. 545) which we have been unable to locate in the copy available to us: “ ‘Cacalia dif- fert a Senecione flosculis quadrifariam scissis.’ ”’ To return to 1753, Cacalia as set forth by Linnaeus in the works which are the starting points for botanical nomenclature has a protologue which is that of Kleinia of the first four editions of the Genera Plantarum, with the exception of the pre-Linnaean Vaillant and Tournefort references to Cacalia, both of which apply to C. alpina. Cacalia alpina does not agree with the generic description in either corolla or stigmas, but a number of the other species do, among them species currently assigned to Cacalia and those removed by Miller to Kleinia. It appears that this is but an- other example of Linnaeus’s changing the name of a genus (Kleinia) to one which he liked better (Cacalia), even though the species which had borne that name historically was somewhat aberrant within an already heterogeneous genus. In the interests of nomenclatural stability, it seems to us most unwise and unwarranted to do anything but to begin the no- menclatural and taxonomic history of Cacalia at 1753 and to proceed from that year in the choice of a type species. The species chosen should be in agreement with the protologue and must be one of the ten described in Species Plantarum in 1753, taking into consideration those which have been removed to other genera. In reaching the conclusion that Cacalia hastata must be the lectotype species, our reasoning follows essentially the same arguments as those succinctly presented by Shinners (1950). All of the ten original species have been transferred to one or more other genera at one time or another. The chronological sequence of the more important of these transfers follows: C. papillaris, C. anteuphorbium, C. kleinia, and C. ficoides: Segregated as the genus Kleinia by P. Miller (Gard. Dict. Abr. ed. 4. 2: ord. alph. 1754), although the combinations under that genus were not made until much later by Haworth (1812) and De Candolle (1838). f C. alpina: Transferred to Tussilago L. by Scopoli (Fl. Carniol. ed. 2. 2: 156. 1772) as T. Cacalia Scop. (not T. alpina L.). Placed in a new genus, Adenostyles, by Cassini (Dict. Sci. Nat. Paris 1(Suppl.): 59. 1816). C. Porophyllum: Removed by Cassini (Dict. Sci. Nat. Paris 43: 56. 1826) to Porophyllum Guett. as P. ellipticum Cass. 272 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 C. suaveolens: Placed in Senecio L. by Elliott (Sketch Bot. S. Carol. & Ga. 2: 328. 1823); later placed in Synosma Raf. ex Britton & Brown (Illus. FI. No. U.S. Canada 3: 474. 1898) C. sonchifolia: Tentatively removed to Crassocephalum Moench (= Gynura Cass., nom. cons.) by Lessing (Synop. Comp. 395. 1832). Later placed in Emilia Cass. by De Candolle as E. sonchifolia (L.) DC. ex Wight (Contr. Bot. India 24. 1834). C. hastata L.: Tentatively referred to Ligularia L. by Lessing (Synop. Comp. 390. 1832), but the combination under Ligularia not made. Later trans- ferred to Senecio as S. sagittatus by Schultz Bipontinus who united Cacalia with that genus (Flora 28: 498. 1845). C. atriplicifolia L.: Transferred to Mesadenia Raf. (nom. superfluum = Arnoglossum Raf.) by Rafinesque (New Fl. N. Am. 4: 79. 1838). Treated as a Senecio by Hooker, who combined Cacalia with Senecio (Fl. Bor.-Am. 1: 332. 1834). By the beginning of 1838 only Cacalia hastata and C. atriplicifolia of the original species had not been transferred formally to other genera. Early in 1838 appeared Volume six of A. P. de Candolle’s Prodromus, which included a revision crucial in the typification of Cacalia. De Can- dolle divided Cacalia into four sections, retaining only three of Linnaeus’s original species: C. hastata and C. suaveolens, which he placed in section Eucacalia DC., and C. atriplicifolia, which was a member of section Cono- phora DC. Thus, he effectively limited the choice of lectotype to either C. hastata L. or C. suaveolens L. Since C. suaveolens had been trans- ferred to Senecio by Elliott (1823), who left C. atriplicifolia in Cacalia, C. hastata becomes the lectotype species. This species is quite in accord with the original description of the genus (Linnaeus, Gen. Pl. ed. 5. 362. 1754), and we can see no reason for choosing another species. The selection of Cacalia hastata as lectotype for Cacalia L., primarily on the basis of the removal of the other species to different genera, is in accordance with both the Guide for the Determination of Types set forth in the International Code and the recommendation (7B) that the lecto- type should be so selected as to preserve current usage. As so typified, Cacalia is a genus of North America and Asia (including easternmost Europe). The question of whether a number of genera should be segrega- ted from Cacalia as now circumscribed taxonomically is not yet settled, but no matter what the outcome of future investigations, Cacalia as typl- fied by C. hastata, will be stable and a minimum of new combinations will have to be made. We regard the species of Cacalia in eastern North America as belonging to two sections: CAcaLtA, represented in North America by C. suaveolens L. and in Asia by C, hastata L. and a number of allied species; and CONO- PHORA DC., restricted to eastern North America. (Cf. Vuilleumier, 1969.) The pertinent synonymy of these sections is shown below: Cacalia L. Sp. Pl. 2: 834 1753; Gen. Pl. ed. 5. 362. 1754. Sect. Cacalia. 1969 | VUILLEUMIER & WOOD, CACALIA 273 Cacalia sect. Eucacalia DC. Prodr. 6: 327. 1838. LectoTYPE SPECIES: C. hastata Synosma Raf, ex Britton & Brown, Illus. Fl. No. U.S. Canada 3: 474. 1898. Type sPEcIES: S. suaveolens (L.) Raf. ex Britton & Brown (C. suaveolens Hasteola Raf. ex —— Not. Syst. Leningrad 20: 380. 1960, nom. super- uum. TYPE SPECIES: H. suaveolens < Pojark. Se suaveolens 1.) [ Not validly published by Rafinesque, New FI. N. Am. 4: 79. 1838; as validated by Pojarkova includes the type species 7 Cacalia.] Sect. Conophora DC. Prodr. 6: 329, 1838. LecrotyPE species: C. atriplicifolia L. a aplaag Raf. Fl. Ludovic. 64. 1817. Type species: A. plantagineum Raf. C. plantaginea (Raf.) Shinners (C. tuberosa Nutt.). Mesadena Raf. New Fl. N. Am. 4: 78. 1838, nom. superfluum. LEcTOTYPE ECIES: M. atriplicifolia (L.) Raf. (C. atriplicifolia L.). [Includes the “eto species of Arnoglossum Raf. | LITERATURE CITED CuatTrREcasas, i A new genus and other novelties. Brittonia 8: 151-163. 1955. [ Cacalia, : : = a on Andean Compositae —IV. bid. 12: 182-195. 1960. Hircucocx, A. S., & M. L. Green. Standard-species of Linnaean genera of Phanerogamae. Pp. 111-199 im International Botanical Congress, Cam- bridge, England, 1930. Nomenclature. Proposals by British Botanists. 1929. [Cacalia, 180.] (Reprinted in Brittonia 6: 114-118. 1947.) Kiramura, S. Compositae Japonicae. Pars tertia. Mem. Coll. — Kyoto Univ. B. 16: 155-292. pls. 1-7. 1942. [Lectotype of Cacalia, 170.] Pippen, R. W. Mexican “Cacalioid” genera allied to Senecio (Compositae). Contr. U.S. Natl. Herb. 34: 363-448. 1968. Poyarkova, A. Notae criticae de genere Cacalia L. s. l. (In Russian.) Not. Syst. Leningrad 20: 370-391. 1960. . Cacalia L. Fl. URSS 26: 683-697. 1961. RypBERG, 2 A. Some senecioid genera —I. Bull. Torrey Bot. Club 51: 369- 378. ibaa Tk H. The Texas species of Cacalia. Field Lab. 18: 79-83. 1950. VuILLEUMIER, B. The genera of Senecioneae in the southeastern United States. Jour. Arnold Arb. 50: 104-123. 1969. [ Cacalia, 115-119.] Gray HERBARIUM ARNOLD ARBORETUM OF IN HARVARD UNIVERSITY HarvARD UNIVERSITY 274 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 A REVISION OF THE MALESIAN AND PACIFIC RAINFOREST CONIFERS, I. PODOCARPACEAE, IN PART Davip J. DE LAUBENFELS THE RAINFOREST FLORA of a considerably submerged land area extend- ing from southeast Asia through Indonesia and New Guinea to the Tonga Islands includes an important conifer element for much of which systematic examination is essentially lacking. It will be the purpose of this study to present in three parts a critical account of the genera of Coniferales that occur primarily in this area, that is to say, all of the tropical rain- forest conifers beyond the mainland of Asia except two species of Pinus whose ranges extend into Indonesia and the Philippines, plus such species in southeast Asia as belong to the island rainforest groups. Conifers are, in general, strongly divided into northern hemisphere and southern hemisphere elements (Li, 1953). The characteristically southern hemisphere families are Podocarpaceae and Araucariaceae, both of which reach their greatest luxuriance in the area under consideration. In addition, several genera of Cupressaceae occur in the southern hemi- sphere as does one genus of Taxodiaceae, but of these only Libocedrus 1s truly a part of the rainforest and included here. Taxaceae, formerly con- sidered in the Coniferales, is not important, posing no taxonomic prob- lems here, and will be omitted also, The rainforest conifers to be studied involve twelve genera and well over one hundred species. New Caledonia alone, centrally located with respect to the floristic region but recently re- markably isolated, has preserved some forty species, all endemic, while the extensive forests of New Guinea have yielded nearly as many again. Indonesia in general has a conifer flora of equal richness to New Guinea, sharing many species, while the rainforests of Queensland, Fiji, and lesser areas have fewer elements many of which are endemic. : More than a third of the species being described, both new and prev!- ously recognized, were studied in their natural state during two extensive field trips to the Pacific in 1957 and in 1964—65. In addition, the directors and personnel of many herbaria contributed greatly to the completeness of the study by their sympathetic help, and I should like to extend my deep gratitude to them. The following key identifies the herbaria whose specimens were consulted: A Arnold Arboretum of Harvard University, Cambridge BM ___ British Museum (Natural History), London BRI Botanic Museum and Herbarium, Brisbane FI Herbarium Universitatis Florentinae GH Gray Herbarium of Harvard University, Cambridge iLL ~——sdU niversity of Illinois Herbarium, Urbana 1969 | DE LAUBENFELS, PODOCARPACEAE 275 ad Royal Botanic eee Kew L Rijksherbarium, Lei LAE Department of sei Division of Botany, Lae NA U.S. National Arboretum, Washington Nsw National Herbarium of New South far Sydney NY New York Botanical Garden, New Y P Muséum National d’Histoire unin oe RSA Rancho Santa Ana Botanic Garden, Claremont spt Hortus Botanicus Bergianus, Stockholm Us U.S. National age (Department of Botany), Smithsonian Institu- tion, Washin De Z Botanic Pearcy pie Institute of Systematic Botany of the University, Ziirich In addition, I should like to thank M. Corbasson, Director, Bureau address, M. Schmid, Centre O. R. S. T. O. M., Nouméa, Lucien Lavoix of Nouméa, and J. W. Parham, Department of Agriculture, Suva, Fiji, for oes help both in the field and i in obtaining additional important speci- men The following comments apply to the citation of specimens: 1. Each specimen is accompanied by a symbol denoting its development and sex. These include ‘“‘j” for juvenile, ‘‘s” adult but sterile, “2” female structures present, and 3” a structures present When more Ren one stage is in- cluded on the same sheet, more than one symbol will be used. Where appro- priate and available, elevation figures will ae included (“m. ig for meters or “ft.” for feet). 2. Where collections are numbered in series (sometimes without a collector’s name) the standard abbreviation will be used. These include: ANU Herbarium ee Seine agape BRUN Brunei, Forest Departm BSIP__ Bri itish Solomon inode Pade BW Boswezan, Forestry Division, Netherland New Guinea NGF N Guinea Forest Departm NIFS_ Netherlands Indies Forest ice bb: series bossen buitengewesten — islands outside Jav: SAN North Borneo Forest Department, Sandakan SFN Singapore Field Number Podocarpaceae is a well differentiated family that is distinguished by holz, 1941). The seeds in many genera are produced on structures so modified from the cone morphology that one can not easily refer to them as cones, although true seed cones and intermediate structures are found in the family. Pollen cones are always truly cone-like and for all but two 276 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 genera are developed strictly on separate plants from the seed structures. Nine of the twelve genera being recognized here occur in the tropics and four of these are endemic to the tropics. All but a few species of the family grow in very moist areas, some, particularly in Tasmania, being found beyond the tree line as small or prostrate shrubs. Most, however, are large forest trees, many with broad leaves quite unlike the usual conception of conifers. Within Podocarpaceae there has been great variation in the size and complexity of the recognized genera. The genus Podocarpus as generally treated involves up to eight sections and well over one hundred species, the differences between some of these sections being every bit as great as those which separate most other genera. It is being proposed here to divide Podocarpus into five separate genera in order to produce a more balanced treatment of the family. In addition, one new genus is separated from Dacrydium because of the different form of the fertile shoots and the strikingly different foliage morphology. The result is a total of twelve genera in the family, of which nine are to be considered in this study, six in part I and three in part IT. KEY TO THE GENERA OF PODOCARPACEAE 1. Seed cone cdi seeds not subterminal. 2. Ovules in ; a produce on ordinary foliage branches; adult leaves in the form of scale 4. Seed ned covered by an epimatium; leaves opposite, decussate. igi e teen en ns a icrocachrys, not tropical). 4. Seed completely enveloped in the fertile scale or epimatium; MOOVES GOUEMIEY GUIANBED. <5 cso ce nnumi sen vneswendenenes KE on ee BEA ea aes ee (some species of Dacrydium, not tropical). 3. ag shoot specialized; adult leaves linear, flat, constricted at the OCI eee Treen) renee (Saxegothaea, not tropical). a. pedis a Fertile as lacking; adult leaves developed. ............---- 002° ie hosts, Seta ie See ines PA eid al eit Re: Microstrobus, not tropical). 5. Fertile scale an epimatium; adult leaves suppressed in favor of phyl- i: a SION OF Pore ane Ee SPU bee Se EAE ee Phyllocladus. 1. Seeds one or a few, subterminal or dispersed near the end of a fertile branch. 6. Seed free, projecting above an epimatium (fertile scale). 7. Seed structures terminal on ordinary foliage branches; leaves crowded, awl-like, linear, or scale-like. ......... Decrydium (most species). 7. Seed structures cia on specialized shoots; leaves bilaterally flat- OE SE I a es is ins ae eee en aten's Falcatifolium. 6. Seed covered by or aed with the scale. 8. Fertile bract forming a terminal crest over seed complex; leaves awl- PRO RIES HT aint at Rt onl ge Sea arene Dacrycarpus. 8. Fertile bract separate from the seed complex; leaves 9. Seed complex becoming erect; leaves bilaterally flattened. - vig Pee Ae RtU ees eee es AE Acmopyle. 1969 | DE LAUBENFELS, PODOCARPACEAE 277 9. Seed complex remaining inverted, leaves bifacially flattened. 10. Fertile shoot terminal on ordinary foliage branches; leaves scale-like; parasitic shrub. [Podocarpus sect. Microcar pus}. r Specialized fertile shoot, usually A oi leaves broad and flat, usually distichous; not parasitic. 11. Fertile shoot scaly; leaves never with both hypoderm and accessory transfusion tis 12. Seed with a beak; paary with hypoderm, usually amphistomatic and decussate, oval or lanceolate. . BR ry ieee remanent ee Orne Decussocarpus, 12. Seed without a beak; leaves without hypoderm, spirally placed and hypostomatic, linear. .......... 2 seat nee tok leashed Ee ee a rumno pitys. . Fertile shoot divided into a naked peduncle and a spe- cialized fleshy receptacle; leaves with both hypoderm and accessory transfusion tissue. ............ Podocar pus. _ ° ~_ — Phyllocladus L. C. & A. Rich. ex Mirbel, Mém. Mus. Hist. Nat. Paris 13: 48. 1825, nom. cons. Type species: Phyllocladus billardieri Rich. ex Mirbel [Phyllocladus aspleniifolius (Labill.) Hooker]. Podocarpus Labill. Nov. Holl. Pl. Sp. 2: 71. t. 221. 1806. Type species: Podo- carpus aspleniifolius Labill. [ Phyllocladus aspleniifolius (Labill.) Hooker]. Brownetera L. C. Rich. Ann. Mus. Paris 16: 299. 1810. Nomen nudum based on Podocarpus aspleniifolius. Thalamia Sprengel, Anleitung zur Kenntniss der Gewiichse. ed. 2. 2: 218. 1817, based on Podocarpus aspleniifolius. Small to large trees; bark dark brown or blackish and smooth, reddish and fibrous within, shed in large thin flakes; abundantly branched, branches often in whorls; juvenile leaves linear or slightly broader near the apex, acute or rounded but with a small spine-like point, 1 mm. or more wide and about 1 cm. long, changing rapidly on small plants to flat- tened leaf-branch complexes or phylloclads with scale-leaves on non- foliage branches; leaves represented by small spurs on the margins of the phylloclads, ‘strongly keeled on the dorsal side, triangular in cross section and on older plants scarcely or not distinguishable; phylloclads extremely variable in shape, broad, dorsiventrally slightly differentiated in some cases, reaching several cm. in length or aggregated along branches in complexes to more than 20 cm. long or transitional as a large deeply lobed phylloclad; monoecious, but individual trees may be verge pollen cones in clusters but the central axis of the cluster in most cas continuing growth, nearly sessile or stalked; seed cone consisting of seat or numerous scales some of which are sterile, single ovules erect in the axil of a scale: seed cones terminal or marginal on fully grown or reduced phylloclads or clustered as are the pollen cones, becoming swollen, fleshy or leathery; erect seeds as many as 20 per cone but usually only - or 3, with a filmy aril (symmetrical but rough edged epimatium) growing as *To be taken up as a genus elsewhere. 278 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 a cup around the lower half, protruding beyond the scale when ripe, oval, wider than thick, with the micropyle as a crooked tip, about 3 mm. long. The genus consists of five closely related species in mild to cool and very moist climates, three in New Zealand, one in Tasmania, and one in mountain areas from the Philippines to New Guinea. PAyllocladus is sharply distinguished from related taxa by the distinctive phylloclads which give it the popular name of “celery-topped pine.” The bark con- tains abundant tannin and the wood is of good quality but, because all of the forest species grow as scattered individuals, its commercial value is limited. One species in New Zealand and part of the population in Tas- mania grow as bushy pioneer plants around mountain meadows. The one tropical example of this genus is the only podocarp species growing in the tropics whose seeds are produced in a recognizable cone. This cer- tainly suggests that the family Podocarpaceae, so abundantly developed within the tropics, had its origins in cooler climate areas. 1. Phyllocladus hypophyllus Hooker f. Icon. Pl. ¢. 889. 1852. Type: n., Mt. Kinabalu. ge arwred hypophyllus var. protracta Warb. Monsunia 1: 194. 1900. Synt ane: ey S. Mindanao, mountain forest of Dagatpan and 18272, Batjan (not seen). Phyllocladus pesthstoen ‘(Warb.) Pilger, Pflanzenreich IV. 5 (Heft 18): 99. Phyllocladus major Pilger, ig a 54: 211. 1916. Type: Ledermann 9872, Lordberg, NE. New Gui Common small tree on ridges or becoming quite large in the forest, 30 m. or more high; bark hard, rough with large lenticels, dark brown, breaking off in large scales; inner bark straw color; hence: more or less whorled around the main stem and densely ramified: foliar buds on young plants with long thin and somewhat spreading bracts, these becoming tighter and more globular on older plants; phylloclads sometimes glaucous, particularly underneath, variable in shape, deeply lobed on young specimens but be- coming ese lobed in maturity, margins nearly entire to wavy with indi- vidual lobes ca. 5 mm. wide and 2 mm. long, oval to triangular, 3 or 4 cm. long and 2 cm. wide, single or aggregated alternately along lateral branches of limited growth: pollen cones clustered around a shoot that continues growth, peduncle 5-25 mm. long; mature pollen cones to 15 mm. long, 3 mm. in diam.; seed cones clustered on stalks about 1 cm. long oF terminal on a slightly modified phylloclad or any possible intermediate con- dition, small, with 1-3 or more fertile scales, first red when mature, then brown and leathery. DistRIBUTION. Luzon and Borneo to New Guinea, scattered and often common in moist forests and on ridges generally, from 1,500 to 3,200 meters, and occasionally from 900 to 4,000 meters. Map 1. Sarawak. Mt. Poi, upper cave, Clemens 20026 j (Ny). Mt. Laiun, Richards ee eee 1969 | DE LAUBENFELS, PODOCARPACEAE 279 out loc. Beccari 2391 s (kK), 3220 2 (kK). Brunei. Mt. Ulak, Ashton BRUN 1033 s 4,300 ft. (K, L). North Borneo. Jesselton, Kumu Rengis, Wyatt-Smith [?] 71650 2 80 ft. [sic] (x, 1, Us). Penampang, Leano-Castro 5992 s 6,000 ft. (k, L), Clemente 6217 s 5,000 ft. (K). Ranau, Meijer SAN 21968 2 5- 6,000 ft. (kK), Mikil 56277 s 7,000 ft. (K), Burgess SAN 25167 s 4,500 ft. (xk). Mt. Kinabalu, Low s.n. 2 8,000 ft. (K-holotype), s.2. 6 10,000 ft. (Kk), Gibbs 4088 j 7,000 ft. (BM, K), 4152 2, j 6,000 ft. (Bm, K), 4238 s (BM), 4273 @ 9-12,000 ft. (Bm, kK), Clemens 10556 & (A), 10565 @ (A, GH, K), 10654 2 (A), 10957 s (BM), 27930 2 6,000-13,500 ft. (A, BM, K, L, NY), 29328 ? 10,000 ft. (A, BM, ILL, K, L, NY), 29743 2 8-9,000 ft. (BM, K, NY), 30029 2 7,000 ft. (A), 30030 2 10,500 ft. (K, Ny), 31838 & 7,000 ft. (A, L), 31927 8 89,000 ft. (Ny), 32459 s (BM, L), 50626 s (BM), 50784 2 7-9,000 ft. (A, BM, L), 50797 2 10,000 ft. (BM, L), 51220 s (Bm), Haviland 1092 2, 11,000 ft. (A, BM, K, L), Sinclair & Kadim 9053 s 6,950 ft. (L), Chew & Corner RSNB 710 2 7,500 ft. (k, Ny), RSNB 4172 2 5,000 ft. (x), RSNB 4824 9 6,000 ft. (K), Smythies S10622 2 9,000 ft. (K, L), Wyatt-Smith 80370 s (x), 80371 2 (kK, L), Anderson S27089 8 11,800 ft. (K), 527090 2 11,300 ft. (kK, L), Meijer SAN 22114 s 4,000 ft. (k), SAN 29271 & 9,700 ft. (K, L). Nicholson SAN 17823 2, 6 8,800 ft. (K, L), Fuchs & Colenette 21430 2 3,375 m. (Kk), Carr SFN 27617 s 11,500 ft. (sm). Trusmadi Kudat, Mikil SAN 31784 8 (1). Sobong Peak, Lobb (1857) s 4,000 ft. (BM, K). Borneo. W. Region, Bengka- jang, Banan, NIFS 669671 j 1,400 m. (Lt), bb24777 j 1,200 m. (a, L). B. Raja, Winkler 1036 s 1,600 m. (x). Ulu Kelan, Molengraaf B3477 s (t). Top of Semedum, Hallier 697 @ (A, K, L, Ny). Mt. Palimasan, W. Kutei (Belajan R.), Kostermans 12894 5 700 m. (pM, K). Mt. Niapa on Kelai R., Kostermans 21482 $ 1,000 m. (x, L). Philippines. Luzon: Mt. Panai (Benguet), Gillis 27257 s (A, K, L, us), Merrill 4753 j (K, L, NY, US), Quisumbing & Sulit 82404 Ss 7,700 ft. (Ny). Mt. Sifigakalsa (Benguet), Sulit 7669 8 2,500 m. (A, L). Benguet, Alvarez 18364 s (pm). Lepanto Dist., Curran 10957 & (kK, L, NY, US). Mt. Data, Steiner 2150 j 2,200 m. (L). Mt. Pukis (Bontoc), Ramos & Edano 37757 8 (a, us). Mt. Tabuan-Buan (Cagayan), Ramos 77401 2 5,800 ft. (xk, Ny). Center, Loher 4851 s (K), 5203 2 (aA, K, Ny, US). Mrnporo: Mt. Halcon, Merrill 5788 s (xk, Nx, us). Mt. Dulangan, Whitehead (1896) s 5,000 ft. (BM). Minpanao: Mt. Katanglad (Bukidnon), Sulit 10052 2 2,200 m. (a, L), 10124 $ 2,300 m. (A). Mt. Candoon (Bukidnon), Ramos & Edano 38738 2 (A, US). Kaatoan Chinchona (Bukidnon), Britton 439 2 1,380 m. (L). Mt. Apo (Davao), Elmer 11463 s (a, BM, FI, K, L, NY, US, Z), Clemens 15675 j (A, NY), Mearns Hutchinson 4679 s (L). Mt. KcKinley, Kanehira 2676 j (NY). Celebes. Mt. Tampai, Palu (Menado), NIFS bb15154 2 2,500 m. (L). Parigi Lombok (Menado), NIFS 6b15026 s 1,100 m. (tL). Sawito (Enrekang), NJFS 6b20782 S 1,750 m. (L). Mt. Tahole, Labu (Malili), Burki bb24089 2 1,500 m. (L). Porehu (Malili), NJFS 6619564 2 1,500 m. (A). Makale-Toloko (Manggala), NIFS bb20270 s 1,200 m. (a, L). Moluccas. Batjan, de Haan bb23236 s 2,199 m. (L). Obi, de Haan 6b23812 s 700 m. (x). Buru, N/JFS 6621509 s 800 m. (t), Binnendyk s.n. j (K, L). Middle Ceram, G. Sofia, Stresemann 133 s 2,200 m. (L). New Guinea. VocELKop: Mt. Nettoti, van Royen 3873 s 1,960 m. (L), van Royen & Sleumer 7403 s 1,750 m. (K, L), Versteegh BW 10407 s 1,700 m. 280 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 (L). Neentjapaki Mts., Kebar Valley, Kalkman BW 6373 s 1,090 m. (L). Adjar, Kebar Valley, Koster BW 6887 2 1,110 m. (L). Tobie Mts., Kebar Valley, Schram BW 7972 s 720 m. (x). Anggi Lakes, Gibbs 5992 2 7~9,000 ft. (BM, K), Versteegh BW 248 s 2,000 m. (A, K, L), BW 253 2 2,100 m. (A, K, L), BW 281 s (A, L), Kanehira & Hatusima 13704 s (A), 14096 s (A), Stefels BW 2008 j 1,875 m. (L), BW 2010 s 1,860 m. (t), BW 2031s 2,200 m. (L). Koebri Ridge, Gibbs 5657 s 8,500-9,000 ft. (pm, K). Ransiki, Sioriep, Mangold BW 2262 s 1,200 m. (K, L). Mt. Mundi (Ransiki), Mangold BW 2254 s 1,900 m. (tL). WeEsTERN Harr: Mt. Genofa (E. of Arguni Bay), Salverda bb22564 s 750 m. (L), Versteegh BW 7596 2 1,000 m. (L). Wissel Lakes, Eyma 4954 ¢ 1,750 m. (A, K, L), 5228 6 (a, K, L), 5371 2 (A, K, L), Versteegh BW 3009 @ 1,750 m. (A, L), Johannes BW 3262 s 1,750 m. (L), Vink & Schram BW 8764 2 1,500 m. (L), BW 8945 s 1,850 m. (1). Nassau Mts., Docters v. Leeuwen 10906 s 2,600 m. (A, K, L). Mt. Doorman (Mamberamo R. Region), Lam 1628 & 3,250 m. (1), 1647 2 3,500 m. (L), 1742 2 3,250 m. (x), 1984 s 2,560 m. (z). Lake Habbema, Brass 9058 & 3,225 m. (A, BM, K, L), 9090 2 (A, BM, K, L), 10528 % 2,800 m. (A, BM, K, L), Brass & Meyer-Drees 10432 @ 3,225 m. (a, 1), Brass & Versteegh 10446 9 2,840 m. (A, BM, L), 10446A 8 3,200 m. (A, BM, L). Barnhard Camp, Brass & Versteegh 11931 2 1,850 m. (A, BM, K, L), 12523 s 1,100 m. (A, BM, L), 12523A @ 1,150 m. (A, K, L), 13520 @ 900 m. (A, BM, L), 135204 2 (a, L), Brass 12191 2 2,100 m. (A, L). Cycloop Mts., van Royen isotypes of Phyllocladus major), Wabag-Maramuni Road, Saunders 1025 s 10,000 ft. a Rjcoenge (Mt. eae pee & Pullen 5871 s 7,600 ft. (A, BM, K, US). Wichmann, Pulle 982 s 2,500 m. (Kk, L), 1018 s (kK, L), 1042 é - 100 m. He i, 2). Upper Minj ae. Pullen 273A j 9,000 ft. (A, L). Al River Mts. (Nondugl), Womersley NGF 5351 2 7,000 ft. (A, BM, XK, L). Mt. ae Sirimbki, Walker ANU 859 2 9-9,500 ft. (A, K, L), 859A j 9,500 ft. (A, K, L). Chimbu, Cavanaugh NGF 3333 2 (kK). Wa imambuno (Chimbu u), Saunier 824 s 9,000 ft. (A, BM, K, L, us). Mt. Wilhelm, Page 5651 s 2,600 m. (K, L, z). Lake Inim, har haps "ANU 2177 5 8,300 ft. (kK, L). sissegeraigs Clemens 4942 s 6-7,000 ft. (A, z), 5117A 2 6,000 ft. (A). Samanzing, Clemens 9384 3 7-8,000 ft. (a), 9549 3 8~9,000 ft. (a). Mt. Enggom, Sarawaket Range, van Royen NGF 16182 & 11,000 ft. (k, L), Mannasat, Cromwell Mts., Hoo8- land 9482 2 7,600 ft. (kK). Bolan, Lauterbach 303 s 2,400-3,000 m. (pM). Mt. Kaindi (Bulolo), Brass 29692 2 2,150 m. (A, K, L, Ny, us), Millar & Womers- ley NGF 12255 s 7,000 ft. (a, K), McVeagh NGF 7580 2 3,000 ft. (A, BM, KE, L), de Laubenfels P4381 4 6,500 ft. (A, K, L, RSA, sBT), P431A j (a), Toropai NGF 17153 @ 6,900 ft. (k, L), Havel & Kairo NGF 17341 @ 7,000 ft. (K). Wau-Salamaua Road, Millar NGF 22785 92 6,400 ft. (xk). Mt. Amungwiwa, Wau, Womersley NGF 17946 s 11,400 ft. (tL). Wagau, Buang Region, Womersley NGF 17902 s 4,500 ft. (kK, x). Papua: Mt. Giluwe, Schodde 2014 ? 8,800 ft. (k, L). Mt. Tafa (Cent. Div.), Brass 4035 : ay ny). Murray Pass, Wharton Range, Brass 4578 s 2,840 m. (A, BM, NY), 4584 @ (A, K, L, NY, US). Mt. Obree, Owen Stanley Range, Lane-Poole (1923) s “mn 000 ft. (A, K). Mt. Dayman, Maneau Range, Brass 22453 2 2,230 m. (A). Mt. Maneao, — 519 s 7,500 ft. (K). Mt. Mon [Mau?], Crutwell 896 j 6,800 ft. (K). Vinevo, Goodenough, Crutwell 1423 s 7,000 ft. (kK). ILLUSTRATION. Hooker, f. Icon. Pl. t. 889. 1852. 1969 | DE LAUBENFELS, PODOCARPACEAE 281 ° —— ee ea eres te si —<—$$__ ~——+- 7 — e's : a | tte s a \ BALANSAE SS atum (Roxburgh) Wallich (dots west o : i tensi benfels, known only from the 3, D. pectinatum de Laubenfels (dots west of line), D. nidulum de lansae Brongniart & Gris, known through- Fiji Islands; ; Laubenfels (dots east of line , D. ba out New Caledonia, and D. cupressinum Solander ex Lambert, known from d New Zealand. 282 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 In addition to its completely disjunct distribution, Phyllocladus hypo- phyllus can be differentiated from all other species of the genus by its distinctly larger phylloclads. In other species the larger structures are deeply lobed and transitional to branch systems, those without deep lobes are less than 20 mm. wide or 25 mm. long. PAyllocladus hypophyllus is unique also in frequently having the seed cones terminal rather than lateral on the phylloclads, and in having peduncles on both pollen and seed cones up to twice the length observed in other species. The two species with intermediate sized phylloclads approaching the lower limit of those of P. hypophyllus are P. glaucus and P. aspleniifolius, both of which share the glaucous habit with P. hypophyllus. The former has seed cones with numerous fertile scales and the latter has particularly small and nearly sessile pollen cones. The species P. major and P. protractus have been differentiated from P. Aypophyllus on the basis of the shape of the cladodes, the position of the seed cone, and by their glaucous aspect. These dif- ferences, however, are found within local populations related to age of the tree or even on different parts of the same specimen. Dacrydium Solander ex Lambert, Descr. Genus Pinus 1: Appendix 93. 1807. Type species: Dacrydium cupressinum Solander ex Lambert. Lepidothamnus Phil. Linnaea 30: 730. 1860. Type species: Lepidothamnus fonkit Phil. [Dacrydium fonkii (Phil.) Benth.]. Shrubs and trees varying considerably in stature; juvenile leaves awl- shaped (falcate needles), longer than the adult, or in some species bi- facially flattened and linear; adult leaves quite variable among the species from scale leaves to leaves resembling the juvenile needles and with either gradual or abrupt transitions uniting the different forms dur- ing their ontogeny; dioecious (or rarely monoecious in some New Zealand species); pollen cones cylindrical, terminal, or lateral and sessile, or both; seed cones much reduced, with bracts hardly modified from foliage leaves, often becoming fleshy when ripe, terminal, often on a short lateral branch; ovules inverted on bracts in a nearly terminal position and partly covered by an epimatium; seeds usually becoming erect, projecting well beyond the apex of the modified cone, occasionally occurring in pairs or three together, sometimes surrounded by the leaf-like extremities of the cone bracts, oval with the micropyle forming a small tip, usually somewhat flat- tened, on some species remaining inverted and covered by the fertile scale. The genus Dacrydium occurs in a wide range of temperature and soil conditions but rarely in anything less than a very moist climate. It is readily divisible into two subgroups based on the internal morphology of the wood, leaves, and pollen (Tengnér, 1965). In one of the groups, called by Florin (1931) Group C, the adult leaves are more or less overlapping; broad, bluntly keeled scales (in one species, a prostrate alpine shrub, plants with juvenile type short flat leaves sometimes are fertile). The other group, called Group B, lacks scale leaves in all but two species where the scale is narrowly and sharply keeled and strongly appressed. 1969 | DE LAUBENFELS, PODOCARPACEAE 283 The seeds in this group always become more or less erect, while in Group C some species have inverted mature seeds covered by the fertile scale. Group C is entirely extra-tropical and will not be treated here. Group B is primarily tropical, the two groups overlapping in New Zealand where most of the Group C species are found. In Group B the juvenile leaves are scarcely distinguishable between the various species, being lanceolate, slender, and bifacially flattened on the seedling but soon be- coming strongly keeled and awl-shaped. For the most part, the seeds are also very similar throughout, so that the species are distinguished pri- marily by the form of the seed and pollen cones, and by the adult leaf form. Four common leaf types occur, one with fairly short needles (2- 5 mm.) changing gradually from the juvenile form (cupressinum, balansae, nidulum, pectinatum), a longer type with more flexible needles (beccarii) , a type with narrow, flat, and lanceolate leaves (xanthandrum), and one with scale leaves changing abruptly from the juvenile needles (elatum, novo-guineense). In addition, there are a number of local species, usually with distinctly bifacially flattened leaves and, in most cases, rather rare. Most of the species are too small in growth form or are too rare to be useful, but a few, as D. cupressinum, are valuable lumber trees. KEY TO THE SPECIES OF DACRYDIUM 1. Trees or prostrate shrubs with adult leaves broad, imbricate, bluntly keeled scales (Group 1. Trees or bushes with adult leaves narrow, appressed, sharply keeled scales or longer spreading leaves (Group 2. Abrupt change between juvenile and adult leaves, which are minute (not more than 1.5 mm. lon 3. Bracts in the fertile area similar to scale-like foliage leaves. ... 3. aay in the fertile area distinctly longer than the foliage aor or 1. en and pollen cones not small; foliage leaves scale-like. Henin deduces Mo hiats ath auc niet ee eee are 3. D. novo- guineense. 4. Seeds and pollen cones small; foliage leaves spreading. . . D. nausoriense. 2. Gradual change from juvenile to adult leaves, which are at least 2 mm. ong. _ Bracts in the fertile area not surpassing the epimatium and not longer that the foliage leav me 6. sroerreca aa og ate lanceolate; leaves pe (0.6 mm.) eee ele ee ee D. seen 6. tin ab long triangular; leaves less than 0.4 mm. thic curved upwards at the ti Pp. 7. Slender, linear oo. with - og turned upwards. Bet hades atc aes D. pectinatum var. ‘pectinatum. k, ly taperin, ny spreading leaves page Send : es : een 5b. D _ pectinatum v. var. - robustum. Bracts in the fertile area distinctly longer than the epimatium and, where the foliage leaves are not long, distinctly elongated by contrast. - JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 8. Bracts.in the fertile area distinctly longer than the foliage leaves of the subtending branch; gin ye a triangular 9. Bracts, in the fertile area, and foliage — strongly keeled, triangular or quadrangular in cross sect 10. Developing seed extending well oa the elongated bracts of the fertile area; foliage leaves not more than 0. wide. 11, Leaves staging Me as thick as wide, tip more or less Blunt and Hat incurved, ......2ciesccvcaends RCs aE era oe Bib ay a nidulum var. nidulum. _ — D. . Leaves noticeably wider than thick, tip distinctly pai with a slight prickle, generally incurved and rowded. ........ 6b. D. nidulum var. araucarioides. Hire a completely surrounded by elongated racts, the tip protruding slightly on maturity; foliage leaves ae ee than 1.0 mm. wide. 12. Pollen cones 2.0 mm. in diameter; leaves curved upwards but not phe wg markedly tapering, quad- rangular in cross section. .......... 7. D. balansae. . Pollen cones 2.5-3. aan mm. in diameter; leaves strong- ly incurved, linear, axial surface concave towards EAs 65< oF vv Acs aes ou8s 8. D. araucarioides. 9. Bracts in the fertile mar and foliage leaves distinctly flat, more than twice as wide a: 13; it and alien: cone small: leaves srg 3-4.5 mm. Re eat a td oat ers al oak Iycopodioides. 13. Sed and pollen cone not small; leaves Tina, 4-7 m SE ae ee tee eee Pere spore 8. Bracts in ee fertile area no longer than the ee eis of the subtending branch; microsporophylls elongated, lanceolate. 14, iy bract not surpassing the mature seed, leaves 5-10 mm. — = _ i) 1S. » Pollen cone 20-25 mm. long by 5-7 mm. in diameter; re less than 0.8 mm. wide Seed cone terminal on ordinary foliage — mature seed surrounded ne bracts. 11. D. magm . Seed cone terminal on shoots with reduced ee seed well exposed when mature. 17. Leaves quadrangular or triangular in cross sec- tion, imbricate. 18. Leaves uniform, more than 5 mm. long, at least ten times as long as wide. 19. Leaves spreading outward. ......----: 12a. D. beccarii var. beccarii. 19. Leaves ante and compact. ...--- Qn oe 12c. D. beccarii var. rudens. Leaves variable, sometimes less than 5 mm. long, less se eight times as long as wide. sees Seta . beccarii var. subelatum. 17. Leaves twice as eae as thick, spreading at nearly right angles to the stem. .....-----°° 13. D. xanthandrum. — ON —_ 2 1969} DE LAUBENFELS, PODOCARPACEAE 285 15. Pollen cone 20-25 mm. long by 5-7 ee in diameter; leaves at least 1.0 mm. wide. .......... gay Nene bract much longer than the seed, for 12- 20 m ~ - guillauminii. comosu 2. Dacrydium elatum (Roxburgh) Wallich, London Jour. Bot. 2: 144. 1843.” Juniperus i aia Fl. Indica 3: 838. 1832. Lectotype: Wallich 6045, Malay iendians sung Miquel, Pl. Junghuhn, 1: 4. 1851. Type: Junghuhn s.n., Sumatra. Decrydium ed Hickel, Bull. Soc. Dendr. France 76: 74. 1930. Lecto- type: Pierre 1396, Cochin China, Phu Quoc Island.* Tree to 40 m., much branched with masses of erect twigs forming a dome-like crown; bark furrowed and flaky, reddish-brown, inner bark pink; juvenile leaves acicular, to at least 12 mm. long, gradually becom- ing shorter and more robust before changing abruptly on young trees, about 6-8 mm. long, sharply keeled on four sides and nearly straight, spreading, acute; mature foliage branches cord-like, 1-2 mm, in diam covered with imbricate scales which are acute and sharply keeled, 1—1.5 mm. long by 0.4-0.6 mm. wide, occasionally passing through a semi-adult or transitional stage of short spreading leaves about 1.5 mm. long; branches with juvenile leaves occasionally fertile; pollen cones terminal, usually on short lateral branches, thus sometimes almost lateral, cylindrical, 4-5 mm long and 1.2 mm. wide: microsporophylls triangular, acute; seed cone terminal, generally on short lateral branches, bracts of the cone becoming slightly enlarged, red and fleshy when mature; the solitary naked seed becomes almost erect, tapering to a blunt apex, reaching 4-4.5 mm. in length DistrisuTion. In humid mountain forests from north central Thailand and Tonkin to central Sumatra and Sarawak, from 500 to 1,700 meters 1n elevation or even down to sea level where suitable conditions exist. Map 2. Thailand. NortuH CENTRAL: Loei, Phu Krading, Tham Nam, Royal Forest Dept. 3631 j, 1,045 m. (us), Kerr 8727 j (K), 8727A @ (kK), 8727B 3 yee a 2263 j 1,300 m. (A). Without loc., Smitinand 19058 j 1,200 m. (XK). TRAL: N. akhaun Nayok, Phengkhlai 691 8, j (K, L). Cambodia. Plateau Pega Gulf of Siam, Showe (1927) s, j 3,000 ft. (sm). North of Kampot, Poilane 14707 2 (ny). Near Komplon (Phnom Penh), Bejoud 717 ° (1LL). Without loc., Pierre 19074 s (k). Tonkin. Than Moi, Balansa 596 ¢, j (ILL, K). An- nam. Summit Mt. Bani, near Da Nang, Clemens 4280 j (K, NY, US). Bana, near Tourane, Poilane 1539 s, j 1,200 m. (N¥-syntype of D. pierrei), 7095 2 (A- Specie cutively through the whole paper. Thee I did ay siete a bbe: but list ed many specimens of which Balansa 576 is the first. The one specimen collected by Pierre is here chosen as the lectotype be- Cause of the specific epithet. 286 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 syntype of D. pierrei). Kontum Prov., Poilane 33351 s 1,200 m. (ILL). Nha- trang, Poilane 25 j (A-syntype of D. pierrei), 3455 2 (a-syntype of D. pierret), 3782 Q (A, K-syntypes of D. pierrei), 4411 2 (a-syntype of D. pierret). Cochin China. Phu Quoc Island, Gulf of Siam, Pierre 1396 8, j (A, K, Ny-iso- types of D. pierrei), Harmand (Godefroy) 901 é (a-syntype of D. pierrei). Without loc., Poilane 32825 8 (ILL), Godefroy-Lebeuf sn. 8 (k). Malaya. Thailand border (Botong), G. Ina, Kerr 7554 s, j (K). Penang, Wallich 6045 s (BM-lectotype of Juniperus elata; K-isotype), Sinclair 39094 s, j (K, L), Walker 70 j (kK), Maingay 2262 s (kK), 2753 s (kK), Curtis 2880 s, j (K). Perak, Ernst 1213 s, j (z). G. Butu, Wray 1028 j (x), 3899 s, j (kK). Pahang, G. Tahan, Hanif & Nur sn. 8 (x), SFN 7959 s, j 5,500 ft. (a, K), Wray & Robinson 5354 s 3,300 ft. (kK), 5380 j (kK). Pahang, G. Lesong, Wakau 4155 j (kK). Ja- hore, Mt. Ophir, Maingay 1503 2 (FI, GH, K, L), Moxon s.n, s (L). Sumatra. Between Tapanuli and Silindong, Junghuhn s.n. j 2,000 ft. (L-holotype of D. junghuhnii). Pajakumbuh, W. Taram, Meijer 6938 8, j 500-1,000 m. (Xk, £), 7040 2 (L). Poya Kombo, Teysmann 21647 8 (Kk), s.n.s (kK). Without loc., Praetorius sm. j (L). Sarawak. Merurong Plateau (Bintulu), Brunig S9991 s 750 m. (L). Mt. Dulit, Richards 1962 s 1,250 m. (BRI, K, L, US). Between Biak R. and Sut, Pickles 2991 8 2,360 ft. (L, us). Lawas, Brunig 510673 4, j 900 m. (L). Borneo. Without loc., De Vriese s.n. j (L). ILLustrations. Riptey, H. N. Fl. Malay Peninsula ¢. 227. 1925. Corner, E. J. H. Gard. Bull. Straits Settlements 10: ¢. 5. 1939 Dacrydium elatum differs from D. novo-guineense in the form of the female cone, in the form of the juvenile leaves, size of the pollen cone, size of the mature tree, and in its occurrence generally at lower elevation. Specimens of D. pectinatum have been much confused with D. elatum be- cause the pectinatum foliage is similar to the juvenile foliage of D. elatum but, the leaves of D. pectinatum are, in fact, shorter and distinctly curved. The known range of these two species overlaps only in Sarawak. The name elatum has further been applied to almost any uncertain Dacrydium specimen from Borneo to the Fiji Islands. Hickel described D. pierrei, contrasting it with D. beccarii, which he mistook for D. elatum. 3. Dacrydium novo-guineense Gibbs, Contrib. Phytogeography and Flora of the Arfak Mountains 78. 1917. Lectotype: Gibbs 5648, New Guinea, Arfak Mountains. Tree to about 10 m. with branches rigidly ascending into a rounded crown; juvenile leaves acicular, spreading and incurved, lanceolate, acute, keeled on the back, to 7 mm. long by 0.7 mm. wide but variable in size, changing abruptly to the adult form, occasionally passing through a semi- adult or transitional stage of short spreading leaves about 1.0-1.5 mm. long; mature foliage in imbricate scales, acute and sharply keeled, 1.0- 1.5 mm. long by 0.4-0.6 mm. wide; foliage branches 1.0-2.0 mm. in di- ameter, penultimate branches becoming larger; pollen cones terminal, usually on short erect lateral branches, cylindrical, 8 mm. long, micro- sporophylls triangular; seed cones terminal on short curved lateral branches, bracts long and spreading, reaching 3 mm. at the cone apex, 1969 | DE LAUBENFELS, PODOCARPACEAE 287 the whole cone becoming red and fleshy when mature, the single apical seed becoming almost erect and extending well beyond the cone bracts, 5 mm. long, edges slightly keeled, tapering to a small blunt apex. DistriBuTion. In open to mossy forests, often on ridge tops from 1,300 to 2,750 meters in elevation, occasionally lower. Locally common but apparently localized; from Obi and the mountains of western New Guinea at least as far as the Western Highlands of the Territory of New Guinea. The collections from the Celebes are tentatively included here until it can be determined whether these represent Dacrydium elatum or D. novo- guineense. Map 2. Celebes. Manado, Poso, Eyma 1623 s 1,700-1,800 m. (L), 3642 4 (tL). Ma- samba, Kuniapu, N/JFS bb24964 s 1,500 m. (L). Masamba, Omboan, N/FS 6626288 j 1,800 m. (L). Enrekang, NJFS bb20786 j 1,900 m. (L). Moluccas. W. Buru, Stresemann 395 s, j 1,800-2,000 m. (L). Buru, Martin s.n. j (1). Obi, de Haan bb23813 s 700 m. (L), bb23814 s, j (L). New Guinea. VocELKoP: Tamrau Mts., Van Royen & Sleumer 7219 s, j 2,000 m. (L). Kebar Valley, Van Royen 3857 s, j 1,980 m. (L). W. of Mt. Nettoti, Van Royen & Sleumer 7948 2 2,100 m. (K, L, LAE), 7948B j (L). Arfak Mts., Gibbs 5648 2 9,000 ft. (sa-lectotype; K-isotype), 5508 s, j 7,000 ft. (sm, K-syntypes), Kanehira & Hatusima 13518 s 2,000 m. (a), Gjellerup 1032 s, j 1,800 m. (L). Anggi Lakes, Versteegh 256 2 2,100 m. (1), 262 2 (1), 269 (1), Stefels BW 2015 j 1,860 m. (L), BW 2033 s, j 2,100 m. (L). WESTERN HALF: Wissel Lakes, Eyma 4422 S, J 1,750 m. (a, K, L), 4519 2 1,760 m. (A, K, L), Vink & Schram BW 8746 S 1,820 m. (L). Hellwig Mts., Pulle 663 2 1,300 m. (L), 966 s 2,600 m. (K, L). Barnhard Camp, Brass & Versteegh 11967 $ 1,520 m. (A, BRI, kK, L), 12507 j 2,100 m. (A, BRI, L), TERRITORY oF NEw GUINEA: Western Highlands, Mt. Hagen, Cavenaugh NGF 3337 s, j (A, BRI, L). Tagen R., Jimmi Valley, Womers- ley & Millar NGF 7680 2 4,300 ft. (A, BRI). Minj-Jimmi Divide, Robbins 598 3, j 6,500 ft. (A, BRI, K, L, us). Nondugl, Womersley NGF 4420 &, j (A, BRI, K, L). Papua: Sibium Range, Pullen 5930A j (L). ILLUSTRATION. Gipps, L. S. Contrib. Phytogeography and Flora of the Arfak Mountains, t. 3. 1917. Being one of the scale-leaved species of Dacrydium, D. novo-guineense cannot be distinguished in the sterile form from D. elatum from which it differs in the elongated bracts of the seed cone and to a lesser extent in the form of the juvenile leaves and the size of the pollen cone. Juvenile specimens can often be separated from D. beccarii, with whose range it overlaps, by their coarser and less dense growth. From D. nidulum the juvenile leaves differ in their variable size including, for the most part, greater length. The rapid change from juvenile to adult form is so strik- ing and comes when the tree is yet quite small so that collectors generally include mature leaf forms when dealing with D. novo-guineense. The rigid ascending branches are another distinctive character. 4. Dacrydium nausoriensis de Laubenfels, sp. nov. Arbor ad 25 m. alta, ramosissima. Folia plantarum iuvenilis acicularia, ad 9 mm. longa, ad formam adultam abrupte convertentes; folia plantarum 288 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 —v —_ — — — — ee ee Eccl 7 a, Dacrydium anion ig de Laubenfels, portion of the holotype, de Laubenfels P302 (A), enlarged; b, ena de i _ nfels var. pectima- tum, portion of the isotype, Nicholson SAN 292 (L), ged: c, D. pectina- tum, var. mee a de Laubenfels, portion a the boi: ce "Meijer SAN 3790 (L), enlarged; b and c are at same magnificati adultarum parva, patula, acuta, dorsaliter carinata, densa, 1 mm. longa, 0.4 mm. lata. Strobili masculi cylindracei, terminales vel laterales, saepe utroque, parvi (?), ad 2.5 mm. longi. Strobili feminei ad apicem ramulorum saepe brevi; folia ad basem seminis longiora, ad 2 mm. longa; semen pro- trudendum, 3.5—4 mm. longum. Holotypus: de Laubenfels P302 (A) ), Fiji, Nausori Highlands. Fic. la DistRiBuTION. In slightly open forest on the leeward sides of the large islands of Fiji and apparently of limited extent. Fiji. gs Levu: Nausori “ee de Laubenfels P302 2 1,900 ft. (A holotype RSA, SBT-lsotypes), P303 j (A, RSA, ait P304 & (A, RSA, SBT), ae NH19 2 (x), NH23 & (x), Bas 29 _ Kuruvoli 13326 & (K). A Levu: Lambassa, Sarava, Damanu L14 5 (e) E), pri FD832 2 400 it. o 1969 | DE LAUBENFELS, PODOCARPACEAE 289 The species of Dacrydium with sharp scale-leaves, changing abruptly from juvenile to adult form (D. elatum and D. novo-guineense) stand apart from the other species, with D. nausoriensis representing a somewhat transitional position. The abrupt change from fine juvenile needles to the more robust and very short adult leaves is in accord with the scale-leaved species, while the still spreading orientation is the common condition for other species. Occasional specimens of D. elatum and of D. novo-guineense have transitional leaves abruptly marked off from the juvenile leaves and closely resembling the adult leaves of D. nausoriensis. The bark of this new species is virtually the same as in all other species of the group, with large thick flakes, fibrous and brown within but with a tough smooth surface generally well supplied with lenticels and weathering gray. The seeds are also of the usual type showing a slight marginal keel and be- coming a rich brown color. The pollen cones seen may not be fully grown. 5. Dacrydium pectinatum de Laubenfels, sp. nov. Arbor ad 40 m. alta, ramosissima; cortex canus vel rufulus; folia brevia, oblique adscendentia, patentia pectinatum, apice paulo incurva, dorsus carinata, 2-5 mm. longa, 0.4-0.8 mm. lata (iuvenilis ad 20 mm. longa). Strobili masculi cylindracei, terminales, 9-12 mm. longi, 2 mm. lati. Stro- bili feminei ad apicem ramulorum, saepe ramulorum brevium; folia ad basis parviora; folia strobilorum sub semine maturo parva, crescentes carnosa rubra; 1-2 folia ultima fertilia. Semen 4.5 mm. longum, non tegens foliis strobilorum. Holotypus: Nicholson SAN 17292 (a), North Borneo, Sandakan. Fics. 1b and 2. The short bracts in the fertile area not even surpassing the epimatium and not longer than the foliage leaves, distinguish this new species from all but two others, one of which, Dacrydium elatum has distinctly smaller pollen cones and scale-like foliage leaves abruptly marked off from the juvenile leaves, while the second, D. cupressinum, has very elongated microsporophylls and thick straight stubby foliage leaves. The short spreading needles distinguish sterile specimens of D. pectinatum from other species with which its range overlaps. Two varieties have been recognized because of rather marked differences in leaf form. Sa. Var. pectinatum. Folia gracilia, linearia, acicularia, 2-5 mm. longa, ssidinedl saacianane DistrisutioN. From Hainan through the Philippines to Billiton Is- land, at low elevations up to 1,500 meters but mostly below 600 meters. Several specimens are reported from sandy soils. Map 3. Hainan. Yaichow, Liang 62041 s (NY), 62619 j (Ny), 62670 a (NY, US), 63214 s top of mt. (a, Ny, US). Hung Mo Mt., Tsang & Fung LU 18100 9 (A, NY), LU18152 2 (a, Ny), McClure 18303 j 1,000-1,500 m. (Ny). Po-ting, How 72869 2 (a). Five Finger Mt., Chun 1367 j (A), 2089 s (a). Dai Land, Dung Ka, Chun & Tso 4380 2,400 ft. (A, Ny). Chim Fung Mt., Lau 5283 $ (A). 290 JOURNAL OF THE ARNOLD ARBORETUM ® ARNOLD ARBORETUM | FLORA OF NEATH BORNEO Te GO THe PRBA® Attend: 2s Hedin ee HARWARD UNivesei* HER BA TS E 2. Dacrydium pectinatum oS eens rar. pectinatum, photogré aph of rt pera Nicholson SAN 172 Without loc., hae ce Chun 70144 2 2,000 ft. (A, ny, us), Liang 63693 3 (NY, US), 65094 3 (A , Wang 33651 s (A, NY), 36532 & (a, NY), Tang 457 @ (A); Hance 22162 j (pm). Billiton, NIFS bb32284 92 (a, L), Rossum 122 (L), 784 2 (L). Sarawak. Bako National Park ~ NE. of Kuching), Purseglove P5066 j 400 ft. (K, L), P5553 & 350 ft. (x, ), Brunig §12073 3 120 m. (L), S12074 s 13 , Sinclair & Kadim pita al s (A, K, L), Sing JC/59 s 300 ft. (kK), Rashid S9546 5 | hap (L), Nicholson 1319 s 200-300 ft. Brunig S1101 2 1,4 (K, 1). K), ‘ * Pa’ Ke. GUILLAUMINI} \Comosum { - » ° e : - «@ x : D oth ~ ~ pettect ‘ 7 ® os> ‘oo ~ = ee ? a . a a » a se LYCOPODIOIDES — SW ocs> . 9 + — —__+_4—— * NS ( a TAXOIDES Maps showing distribution of: 4, Dacrydium beccarii Parlatore (dots), D guillauminii Buchholz, known only from New Caledonia; 5, D. xanthandrum Xe 1969] DE LAUBENFELS, PODOCARPACEAE 303 Haviland 2070 s (x). Philippines. Leyte: Biliran, Sulit 21694 s 1,350 m. (L). Gros: Dumaguete, Or, Herre 1150 s 4-6,000 ft. (A, ny). Mt. Canlaon, Edano 21936 j 1,860 m. (L). Mt. Marapara, Curran & Foxworthy 13612 s (L, NY, US). Mt. Silay, Everett 4227 j (Ny, us). Without loc., Britton 343 s 1,700 m. (t). Mrinpanao: Mt. Malingdang, Mearns & Hutchinson 4547 s (K, L, ce US), 4731 s (NY, US). Moluccas. Taliabu, Hulstijn 126 2 (L). New Guinea. Vogel- kop, Upper Aifat Valley, Moll BW 12853 s 870 m. (L); Tamrau Ra., Van Royen & Schram 7791 s 920 m. (K, L, LAE). ohn Mts., Gjellerup 572 s 600-1,500 m. (A, K, L). Hellwig Mts., Lorents 1698 s 2,100 m. (K, L). Wissel L., Maiare, Eyma s.n.s (L). Norman “ I., Brass 25660 2 mt. crest (A, K, L, LAE, US). New Britain. Mt. Tangis, Frodin N GF 26902 s 3,500-5,000 ft. (L). Solomon Is. Santa Ysabel, Baea BSIP 2475 : well above 3,000 ft. (ridge top) (kK, L, LAE), Brass 3264 4, } 1,400 m. (A, Hex, 1). Guadalcanal, Mt. Popomansiu, Braithwaite 4810 2 (kK), Hill 9004 j *,000 ft. (K). ILLUSTRATION. CorNER, E. J. H. Gard. Bull. Straits Settlements 10: t. 6. 1939 The branches of this variety have a definite bushy aspect because of the fine dense growth of needles. Several specimens with leaves more robust than normal for the species have been included here, although their status is a little uncertain. These include Van Steenis 8357, Brass 3264, and Brass 25660. 12b. Var. subelatum Corner, Gard. Bull. Straits Settlements 10: 243. 1939. Type: Corner SFN 33224, Malaya, Pine Tree Hill. Adult leaves noticeably less bushy than in the typical variety, variable in length, 3-6 mm. long; up to three seeds in a fertile structure. DistrIBUTION. Mixed with var. beccarii in the mossy forests and ex- posed ridges of Malaya, from 1,200 to 2,300 meters. Malaya. G. Bubu (Perak), Wray 3875 2 5,000 ft. (a, K). G. Tahan (Pahang), Hanif. & Nur SFN 7994 & 5,500-7,000 ft. (K). G. Tapis (Pahang), Symington & Kiah s.n. 2 4,600 ft. (K). Fraser Hill (Pahang), Cubitt 6519 s (x). Pine Tree Hill (Pahang), Corner SFN 33224 s 4,200 ft. (K-isotype). G. Padang (Treng- ganu), Moysey SFN 31072 s 4,000 ft. (K), SFN 31841 s 3,800 ft. (x). ILLUSTRATIONS. CorNER, E. J. H. Gard. Bull. Straits Settlements 10: t7&8 Only the shorter needles distinguish this variety from variety beccari, and intermediates between them can be found 12c. Var. rudens de Laubenfels, var. nov. Folia patula incurvata conferta in forma rudenti. Holotypus: Brass 27821 (A), Sudest Island. Fic. 4b. Pilger (dots), D. comosum Corner, known only from the Malay agp D. lycopodioides Brongniart & Gris, known only from New Caledonia; 6, Falca tifolium falciforme (Parlatore) de Laubenfels (dots west of line), F po bieasiesme de Laubenfels (dots east of line), F. taxoides (Brongniart & Gris) de Lauben- fels, known only from New Caledonia 304 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 DIsTRIBUTION. New Guinea to Sudest I., from 300 to 3,000 meters in elevation. New Guinea. WESTERN Hatr: Mt. Goliath, de Kock 42 s 3,000 m. (L). With- out loc., Brandenhorst 132 s (L), 133 s (L), van Romer 1233 s (1). Sudest (Tagula) I., Brass 27821 2 500-600 m. (A-holotype; K, L, Us-isotypes), 28187 9 300 m. (A, K, L, US), 28188 j (A, L, US). This variety with incurved leaves forming a compact and smooth rope- like branch system contrasts strongly with the two varieties which have spreading leaves and a ragged appearance. Otherwise var. rudens does not differ significantly from the remaining varieties of this species. 13. Dacrydium xanthandrum Pilger, Bot. Jahrb. 69: 252. 1938. Lec- totype: Clemens 4504, New Guinea, Morobe District. Tree to 30 m. high, sometimes stunted on ridges; densely branched; bark chocolate brown or reddish, peeling in thick flakes, bearing lenticels; leaves spreading obliquely, slightly incurved, linear-lanceolate, generally wider than thick, keeled on the back, acute, 6-10 mm. long, or longer on vigorous branches and when juvenile, 0.6-0.8 mm. wide, not crowded; pollen cones lateral or terminal and subtended by several reduced leaves, oval to cylindrical, 5-13 mm. long; microsporophylls narrowly triangular to lanceolate, acute, 2—2.5 mm. long; seed cones terminal, often on very short branches, fertile bracts in the form of reduced leaves; seeds rich tan, 2-angled, 5 mm. long, more or less protruding when mature, Fic. 5. DistripuTion. The island of Borneo and the Philippines to the Solo- mons, in the mountains from 1,000 to 2,400 meters, rarely down to 500 meters above sea level. Map 5. Sumatra. Road from coast to Tapanuli (Toba L.), Bangham 1070 2 4,100- 4,500 ft. (A, K, NY). Sarawak. Mt. Luiga, Beccari 3948 6 (Fr). Baram, Ander- son 4545 2 4,800-7,000 ft. (k, L). G. Mulu, Hotta 14597 8 1,200-1,600 m. (L). orth Borneo. Kinabalu, Nicholson SAN 17827 2 8,800 ft. (BRI, K, L), Clemens 32502 s 6,000 ft. (A, K, L, NY), 34341 2 5-6,000 ft. (A, K, L, NY). Ranau, Nichol- son SAN 39768 2 8,000 ft. (k), Meijer SAN 29153 s 7,000 ft. (kK, L). Tambu- nan, Mikil SAN 32086 8 montane (x, L). Penampang, Clemente 5980 s 5,000 ft. (k, L), Leano-Castro 5985 s (kK, L). Mt. Alab, Keith 5965 j 6,000 ft. (K, L). . Borneo. B. Raja, Winkler 1037 & 1,600 m. (L). Philippines. Mt. Umingan (Nueva Ecija), Luzon, Ramos & Edafo 26510 2 (a, K, us). Mt. Halcon, Min- doro, Rabor 20485 &, j 1,600 m. (t), Edafio 3265 s 780 m. (A), Merrill 5714 s (us), 5789 j (NY, Us). Calapan, Mindoro, Vidal 3910 2 (a, K). New Guinea. Cycloop Mts., Karstel BW 5440 s 510 m. (L, LAE). Sepik region, Ledermann 9395 s (L). Chimbu, Cavenaugh NGF 3334 j (a). Morobe District, Ogeramnang, Clemens 4504 8 (a-lectotype; z-isotype), 5390 2 5,900 ft. (A-syntype), 6398 5,850 ft. (a-syntype), 6408 s 5,850 ft. (A), 6488 s 4,500 ft. (a). Bougainville Is. Kajewski 1694 & 950 m. (A, BRI), 1709 2 1,000 m. (A, BRI, L). Solomon Is. Guadalcanal, Walker BSIP 247 2 1,500 ft. (A, BRI, K, L), Kajewski 2652 s 1,200 m. (A, BRI, L). The range of this species overlaps that of Dacrydium beccarii with which it is often confused, both being found for example, at Ranau and 1969 | DE LAUBENFELS, PODOCARPACEAE 305 er ‘ if id “wniasoguy aAONUy | PLANTS OF NORTHEASTERN NEW GUINEA Me. Mga M.S. Comens wine aatee panThbs disc hee . & Mosebe Thstnet Ficure 5. Dacrydium xanthandrum Pilger, photograph of the lectotype, Clemens 4504 (a). on Mt. Kinabalu, D. xanthandrum differs in the noticeably flattened leaves which are widely spreading and distinctly less dense. It also grows with D. gibbsiae and its leaves strongly resemble the transitional leaves 306 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 of that species, but, not only are the adult leaves of D. gibbsiae much more robust, the pollen cone is much larger, and the fertile shoots have unre- duced leaves. The specimen from Sumatra cited here has much more robust leaves similar to nearly mature leaves of D. gibbsiae and may, with more material, be found to represent a distinct taxonomic unit. 14. Dacrydium gibbsiae Stapf, Jour. Linn. Soc. Bot. 42: 192. ¢. 4. 14. Type: Gibbs 4162, North Borneo, Mt. Kinabalu. Dacrydium beccarii var. kinabaluense Corner, Gard. Bull. Straits Settlements 10: 244. 1939. Type: Carr SFN 26437, North Borneo, Mt. Kinabalu (not seen; photo included in description). Much branched tree to at least 12 m. high; juvenile leaves acicular, 12-20 mm. long, spreading but slightly incurved; mature foliage leaves becoming wider and thicker but distinctly flattened, incurved and im- bricate (an angular keel on the dorsal side), acute, aggregated into rope- like shoots about 8 mm. in diameter, individual leaves 5—7 mm. long, 1.0- 1.3 mm. wide, rigid; pollen cones terminal or lateral, cylindrical, 20-25 mm. long by 5—7 mm. in diameter; microsporophyll lanceolate, 5 mm. long; seed cone terminal, often on a very short lateral branch, formed of largely unmodified leaf- ile structures and with one or two fertile apical leaves, becoming reddish when mature; seeds becoming almost erect, surrounded by but spreading apart the subtending leaves, oval and tapering slightly towards the apex, 4.5 mm. long. DistrIBuTION. On the slopes of Mt. Kinabalu, in serpentine soils where it is common from 1,500 to 3,300 meters. North Borneo. Mt. Kinabalu, Gibbs 4162 2 over 6,000 ft. (BmM-holotype; K- isotype), 4050 j (BM), Clemens 1 10685 @ (A, GH, K), 10879 j (a), 11091 6 (A), 28542 s 11,000 ft. (K), 30922 j 45,000 ft. (a, L, Ny), 33037 2, j 5,000 ft. (A, , a i: 40151 2 6,500 ft. (A, NY), Griswold 67 j (A), Haviland 1183 s 6,600 ft. (x), C hew & Corner 4303 j (K), 4361 j 7,000 ft. (kK), 8024 2 (xk), Nicholson SAN 17826 2 9,000 ft. (art, L), Meijer SAN 21097 s 5,500 ft. (x), SAN 21098 j, 5,000 ft. (k), SAN 23500 s 6,000 ft. (kK), Colenette 543 s 8,000 ft. (K). Pinosok Plateau, Colenette 542 2 5,100 ft. (x). ILLustRaTION. CorNeR, E. J. H. Gard. Bull. Straits Settlements 10: t. 9. 1939, as Dacrydium beccarii var. kinabaluense. This is one of the many distinctive endemics of Mt. Kinabalu and, like many, is characteristically robust in form. The pollen cone is unique. With the discovery of fertile Dacrydium xanthandrum specimens well up on Mt. Kinabalu, many of the “juvenile” specimens may actually be that species. 15. Dacrydium guillauminii Buchholz, Bull. Mus. Hist. Nat. Paris il. 21: 282. 1949. Type: Buchholz 1728,° New Caledonia, Riviere des Lacs. "In the — of this species the collection number given is 1278, clearly a typographical e 1969} DE LAUBENFELS, PODOCARPACEAE 307 Erect bush 1-2 m. high; bark with small dark rough flakes, fibrous brown within, surface more or less smooth at first and covered with numerous small lenticels, developing many small cracks with age; pro- fusely branched; leaves becoming denser and less spreading with age but not at all reduced in size, acute, needle-like or slightly compressed, bushy imbricate, 13-17 mm. long, 1.0 mm. wide; pollen cones terminal and lateral, the lateral ones at the base of a terminal cone and smaller, 8—14 mm. long, tapering from the base; microsporophylls with a long lanceolate tip from 5 mm. at the base of the pollen cone to not more than 2 mm. long near the apex; seed cones terminal, sometimes on very short lateral branches; bracts of the seed cone unmodified or slightly reduced leaves; seeds up to five in a cone, subterminal, oval, wider than thick, laterally keeled, the tip rounded with the micropyle projecting, 4.5 mm. long. DistRIBUTION. Probably the most restricted species of the genus, found only for a few kilometers along the Madelaine River (Riviére des Lacs) and on the margins of Lac en Huit, from which that river flows, and only at the very edge of the water. New Caledonia. Riviére des Lacs, Buchholz 1728 $ (1Lt-holotype; K, P- isotypes), de Laubenfels P341 2 (A, RSA), P341A & (A, RSA), Bernier 323 j (P), sn. 8 (p), Sarlin 242 s (p), Déniker 205 p.p. (z), Baumann-Bodenheim & Guil- laumin 11798 s (p, z), Hiirlimann 3471 s 146 m. (z), Bernardi 9360 s (P, 2), Blanchon 1162 s (p). Lac en Huit, de Laubenfels P116A é (spt), P116B 2 (x, sBT), McKee 3385 4 (A, K, P, US). ILLUSTRATION. SARLIN, P. Bois et Foréts de la Nouvelle-Calédonie, t. 21, 1954. This distinctive bush, a component of the serpentine maquis, bears strong resemblances to Dacrydium beccarii and probably represents an endemic pedomorphic variant of that species. 16. Dacrydium comosum Corner, Gard. Bull. Straits Settlements 10: 244. 1939. Type: Corner 33222, Malaya, Pine Tree Hill. Tree 4-12 m. high; profusely branched with an umbrella-shaped crown; bark in small flakes; foliage branches bushy, densely leafy; leaves spread- ing at an angle and then incurved near the base, lanceolate-pungent, dis- tinctly flattened, 12-20 m. long and 0.7-1.3 mm. wide; juvenile leaves up to 33 mm. long; pollen cones unknown; seed cone on a short lateral branch, often with two seeds; seeds 4-5 mm. long. DistriBuTION. Mossy forest on exposed ridges, from 1,200 to 2,000 meters elevation in parts of Malaya, common locally but of restricted range. Malaya. Pahang. Pine Tree Hill, Corner SFN 33222 s 1,500 m. (k-isotype), Burkill & Holttum 8536 s (a, K), Melville & Landon 4814 s (x). G. Tahan, Hanif & Nur SFN 8307 s 1,500-2,000 m. (A, K). 308 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ILLUSTRATION. Corner, E. J. H. Gard. Bull. Straits Settlements 10: t. 10. 1939. Like Dacrydium guillauminii, D. comosum is apparently a_pedo- morphic variant of some other species, perhaps D. xanthandrum. The distinctly flattened and much longer leaves set it apart from D. beccarit which grows in the same area. The relationships between the flattened but falcate-leaved Dacrydium species (xanthandrum, comosum, gibbsiae, spathoides, and lycopodioides) are unclear. They may be a group with a common origin or each may have developed separately from other stock. It is worth noting that, where known, their juvenile leaves at inter- mediate stages have an unflattened form. Thus the flattening, for some at least, does not represent a continuation of the seedling flattened- leaf condition. This is in distinct contrast to the flat and not falcately incurved leaves in other genera of the family. Falcatifolium de Laubenfels, gen. nov. Type species: Falcatifolium falciforme (Parlatore) de Laubenfels. brevissimis. Strobili feminei in ramulis brevissimis, axillares; squamula ultima sola ovulifera; ovulum unicam inversum, epimatio involutum; se- tandem suberectum, epimatio cristato basi breviter involucrato, crista lateraliter prominens; strobili maturi carnosi. This new genus was previously included as a part of Dacrydium, identi- fied as group A by Florin (1931, pp. 256-259) because of its differences from other members of that genus. Tengnér (1965) also discussed the distinctions between Florin’s group A and the rest of Dacrydium. Sev- eral basic differences justify the separation of Falcatifolium as a new genus. The fertile structures in Falcatifolium are produced on specialized axillary shoots whereas in Dacrydium they grow terminally on ordinary foliage branches. The epimatium of the new genus has a pronounced hump which projects laterally from the mature cone, in contrast with the smaller epimatium of Dacrydium which becomes a cup-like fringe at the base of the mature seed, not projecting at all. Very striking In Falcatifolium are the bilaterally flattened leaves which spread out dis- tichously, contrasting not only with the fertile shoots and basal scales of new growth, but also with the bifacially flattened juvenile leaves which give way rapidly to the adult form at about the second year of growth. In Dacrydium bilaterally flattened leaves do not occur. The name Falcatt- folium reflects the basal falcate curvature of the leaves away from the branch. Tengnér (1965) further reports a lack of vascular fibers and pollen differences which separate this new genus from Dacrydium. Four spectes can be differentiated, primarily on the basis of leaf form, distributed from Malaya to New Caledonia in moist mountain forests, where they occur as undershrubs or small understory trees. we 1969 | DE LAUBENFELS, PODOCARPACEAE 309 KEY TO THE SPECIES OF FALCATIFOLIUM 1. Leaves broad and flat. 2. Leaves blunt to acute, normally more than 20 mm. long and 3 mm. wide. 3. Pollen cone 20-30 mm. long by 2-3 mm. in diam.; upper edge of the leaf normally curved upwards, leaf variable in size er a more Chvae: DS naee, De gods SS i os See F. falciforme. . Pollen cone 15-25 mm. long by 1.5~2.0 mm, in Ps m.; upper edge of leaf cle even slightly curved upwards, leaf rarely as much as 5 mm. 18. F. taxoides. 2. Leaves a ieais: 12-17 mm. long by 2-3.5 mm. wide. .. 19. F. papuanum, ke Awe pra Tere i 5 Svc a ek onc os 20. F. angustum. Ge 17. Falcatifolium falciforme (Parlatore) de Laubenfels, comb. nov. Podocarpus falciformis Parlatore in DC. Prodr. 16(2): 685. 1868. Lectotype: Beccari 2437, Sarawak, Mt. Poe. Nageia falciformis (Parl.) Kuntze, Rev. Gen Dacrydium falciforme (Parl.) Pilger, ce ; oe 18): 45. 1903. Tree 3-10 (rarely to 25) m. tall; bark more or less smooth, rich purple- brown, inner bark dark reddish; leaves variable in size, on mature fruit- ing trees from 20 to 65 mm. long and 5-7 mm. wide, smoothly curved out- ward from near the base to the widest part (about one third of the length from the base), then tapering and curving more or less gradually towards the acute tip, smaller leaves which may be almost straight and probably not fully developed, sporadically occurring along with normal leaves, nar- rowed at the base to a short, angled petiole and then decurrent; pollen cone axial or terminal on a short, 2-3 mm., scaly stalk, cylindrical, 20-30 mm. long and 2—3 mm. in diam.; microsporophyll small, triangular-acute; seed cone on a short scaly shoot up to 5 mm. long, the cone made up of about a dozen larger, acuminate scales, the apical one fertile, the whole cone becoming fleshy on maturity; seed with a humped epimatium at the base, oval, flattened and narrowed to a blunt apical ridge, 6 mm. long, 5 mm. wide, and 4 mm. thick. DistripuTion. Mostly an understory tree in open rainforests from 600 to 1,650 meters in elevation, from Malaya and Luzon to Obi in the Moluccas. Map 6 Malaya. Mengkuang, Wyatt-Smith 93115 8 5,000 ft. (kK, L, us). Batu Gajah, Perak, Ridley 5695 8 (x). G. Tahan, Hanif & Nur — 7851 & (xk), Ridley 16026 & (kK), 16178 s (k). Pine Tree Hill, Penang, Poore 6228 s 4,300 ft. (K). Fraser Hill, Nur 10507 s 4,000 ft. (A). Lingga Archipelago. Teysmann 169 9 (L), Hullett 5695 8 (a, BM). Sarawak. Santubong top, Beccari 2126 3 (F), Haviland (1890) $ 2, 800 ra (Kk). Mt. Dulit, ae 1834 $ 900 m. (A, BM, K, L), 1836 j (pM, kK). Mt. Poe, Beccari 2437 2 (rI-lectotype; A, k-isotypes), Clemens 20238 s 6,000 ft. (Ny), 20263 s 5,000 ft. (A, Ny). Mt. Mattang, Bec- Cari 1331 s (Ft), 1697 s (FI), 2941 2 (FI), Koley 11669 s (k). Trusan, so S8743 s (K, L). Meruong Plateau, Brunig 59994 s 800 m. (L). Without loc. An- derson 8365 & 2,000 ft. (K, L), Gibbs 4400 s 3,000 ft. (BM, K). Brunei. Ash- 310 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ton BRUN 1031 s 4,300 ft. (K, L), 1066 s 4,750 ft. (K, L). North Borneo. Kina- balu, Clemens 10962 s (A, K), 27851 j 7,000 ft. (BM, K, NY), 33078 & 5,000 ft. (A, K, L, NY), 5.%. s 4-5,000 ft. (A, L, NY), Gibbs 4067 s (kK), Chew & Corner 1863 & 5,500 ft. (K), 4847 s 5,000 ft. (x). Lahad Datu Dist. (Mt. Silam), Wood SAN A4179 s 2,500 ft. (A, BRI, L), Meijer & Anak SAN 37497 6 2,000 ft. (kK, L), SAN 22705 s (k). Penampang, Clemente 5995 s (kK), Leano-Castro 5986 s (x). Ranau, Meijer SAN 20953 s (K), Anon. SAN 20279 j 4,000-4,500 ft. (1). Borneo. Bengkajang, NJFS bb9664 s 1,400 m. (L), bb24778 s 1,200 m. (A, L), bb25157 s 1,100 m. (L). G. Damus, Hallier 506 s (L). Mt. Palimasan, Koster- mans 12779 s 500 m. (L). Lianggagang, Hallier 2688 s (L). Philippines. Mrn- DANAO: Davao, De La Cruz 27746 j (us). Minporo: Mt. Halcon, Merritt 4425 2 (F, us), Merrill 5744 s (kK, L, Ny, US), Rabor 20482 s 1,600 m. (L). Without loc. Whitehead (1896) s, j (BM). Luzon: Mt. Umingan, Nueva Ecija, Ramos & Edaiio 26394 & (a, ny, us). Mt. Camatis, Tayabas, Edaio 4508 2 (A). Celebes. Manado, Eyma 3671 j (Lt), NIFS bb17544 s 1,400 m. (A, L), 0b21294 2 1,200 m. (L), bb24778 s (a). Obi. de Haan bb23815 j 700 m. (1). ILLUSTRATIONS. PiicER, R. Pflanzenreich IV. 5 (Heft 18): fig. 4 D-G. 1903; Nat. Pflanzenfam. ed. 13: fig. 227 D-G. 1926; Gress, L. S. Jour. Linn. Soc. Bot. 42: ¢. 8. 1914, all as Dacrydium falciforme. Shape of pollen cone and mature leaf size and shape distinguish Falcati- folium falciforme from other species in the genus. In contrast, F. taxoides has a more slender pollen cone and mature foliage leaves with only sporadically the slightest upward curve of the upper leaf margin, while in F, falciforme such a curve is normally pronounced and only sporadical- ly absent. The mature leaf size of F. papuanum is completely below the great size range of F. falciforme, differing also in a straight profile and apiculate tip. The larger, probably deep-shade-grown leaves of F. falct- forme with the sweeping curve of their upper part are attractive and quite unique, paralleled only in F. angustum whose leaves are quite narrow. 18. Falcatifolium taxoides (Brongn. & Gris) de Laubenfels, comb. nov. Dacrydium taxoides Brongn. & Gris, Ann. Sci. Nat. Paris V. 6: 245. 1866. Lectotype: Vieillard 1259 p.p. New Caledonia, Balade. Podocarpus taxodioides Carriere, Traité Conif. 2: 657. 1867. Type: Vieil- lard 1259 p.p. New Caledonia, Wagap. Podocarpus taxodioides var. gracilis Carriére, ibid. 658. Type: Vieillard 1259 p.p. New Caledonia, Balade. Nageia taxoides (Brongn. & Gris) Kuntze, Rev. Gen. Pl. 800. 1891; as N. taxodes. Bush or small tree from 2 to perhaps 15 m. high, bark thin, more OF less smooth, scattered with lenticels, light reddish brown and fibrous within, occasionally breaking off a flake; loosely branched; juvenile leaves bifacially flattened, long ovate, almost linear, tapering to a sharp tip, keeled on the lower surface, 15-20 mm. long and 1.5 mm. wide; ma- ture foliage leaves somewhat variable, smoothly curved outward at the base and expanding to the greatest width at about one third their length, then tapering slightly toward the rounded or acute apex, sometimes al- aw 1969 | DE LAUBENFELS, PODOCARPACEAE 311 most linear, the tip usually straight and pointing directly outward or occasionally bent slightly towards the branch apex without a corres- ponding bend in the upper leaf edge (or rarely a slight curve), more or less narrowed at the base to a petiole and then decurrent; pollen cone axillary or terminal, often with several on a short axillary branch with minute scales, cylindrical, 15-25 mm. long and 1.5—-2.0 mm. in diam.; microsporophyll with a minute acuminate tip; seed cone on a slender scaly branch up to 6 mm. long, the cone with about a dozen larger elongated scales up to 2 mm. long, the apical one fertile, the whole cone becoming fleshy on maturity; seed with a humped epimatium at the base, oval, strongly keeled on the sides with an elongated blunt tip, 7 mm, long, 4 mm. wide, and 3 mm. thick. DIsTRIBUTION. In moist rainforests (but not mossy forests) as an understory shrub or small tree throughout New Caledonia wherever these conditions occur, which is most commonly in the 800 to 1,200 meter range but occasionally reaching almost to sea level and to at least 1,400 meters. New Caledonia. Balade, Vieillard 1259 p.p. s (p-lectotype of Dacrydium taxoides and holotype of Podocarpus taxodioides var. gracilis). Ignambi, Comp- ton 1571 9 3,500 ft. (Bm). Upper Diahot, Hiirlimann 1887 3 (p, z). Mt. Col- nett, Hiirlimann 1965 & (p, z). Tao, Baumann-Bodenheim 15881 s (P, Z). P), Thorne 28705 s (Pp), Baumann-Bodenheim 5654A s (P, Z), 15680 4 (Pp, 2), Baumann-Bodenheim & Guillaumin 11259 s (P, Z), 11262 s (P, z), 11286 s (P, Z), 11287 s (Pp, z), 11292 s (P,z), 11296 s (P,z), Blanchon 340 s (P). Mt. Dzumac, Bar- ets 7 s (Pp), Blanchon 1247 s 700-900 m. (Pp). Dumbea, Sunshine Mine, a 1587 s 650 m. (Pp, z), 1609 s (Pp, z). Mt. Koghis (Mone, Bebo), Pancher 379 (p), Blanchon 566 s 300 m. (P). Upper R. Bleue, Bernier 301 s (P). agrees Bodenheim 15021 2 (p, z), de Laubenfels P400 3 800 m. (RSA, SBT), / on ville & Heine 187 s, j (ve), Bernardi 9404 s (P, z). Upper Mois de Mai, a 2 1390 s (ILL, P). NE. of Lac Naoué, Hirlimann 3180 s 500 m. (Z). Bois : ud, Baumann-Bodenheim 12492 s (P, Z), 14996 @ (P, z). Upper Kuébini, Hur a 3542 2 265 m. (z), 3543 s (z). Without loc., Balansa 184 & (Pp), Deplanche 169 s (Pp), Mueller s.n. s (p), Sarlin 229 s (e), Baudouin 387 s (P). 312 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ILLUSTRATIONS. BroNcGNIART & Gris, Nouv. Arch. Mus. Hist. Nat. Paris 4: t. 3. 1868, as Dacrydium taxoides; Pircer, R. Pflanzenreich IV. 5 (Heft 18): fig. 4 H-L. 1903, as Dacrydium falciforme; Nat. Pflan- zenfam, ed. 2. 13: fig. 227 H-L. 1926, as Dacrydium taxoides; SARLIN, P. Bois et Foréts de la Nouvelle-Calédonie, t. 19. 1954, as Dacrydium taxotdes. From Falcatifolium falciforme this species differs in its smaller leaves and pollen cones and in the straight rather than upwardly curved leaf tips. From F. papuanum it differs in lacking a pungent leaf tip and hav- ing distinctly larger leaves. These two species and F. taxoides are clearly quite closely related, being geographic segregates. F. taxoides is sometimes the host to another conifer as a root parasite (de Laubenfels 1959). 19. Falcatifolium papuanum de Laubenfels, sp. nov. Arbusculus vel arbor ad 22 m. altus; folia patentia, ad apex apiculata, linearia vel ovato-linearia, 12-17 mm. longa, 2—3.5 mm. lata. Strobili masculi ignoti; strobili feminei cum ramulis brevissimis, squamis lanceo- latis, 1.0-1.5 mm. longis, bracteis strobilorum ca. 2 mm. longis; semen lateraliter et terminaliter carinatum, 6 mm. longum, 4.5 mm. latum, 3 mm. crassum. Holotypus: de Laubenfels P483 (a), New Guinea, Morobe District. Fic. 6a, b. DistRIBuTION. In moist rainforests as an understory plant in the east- ern part of the island of New Guinea (possibly in the Vogelkop), from 2,000 to 2,400 meters in elevation. Map 6. New Guinea. VocELKop: Mt. Nettoti, Van Royen & Sleumer 8203a j 1,920 m. (L), Terr. New Guinea: Al R. Mts., Womersley NGF 5354 s 7,000 ft. (4, BRI, K, L). Mt. Hagan Sta., Hoogland & Pullen 5891 2, j 7,600 ft. (A, BRI, i. t. Kum, Womersley NGF 9419 s 7,000 ft. (BRI, K, L). Nondugl, Womersley NGF 4483 s 7,000 ft. (a, K, L). Morobe Dist., Edie Creek (Mt. Kaindi), de Laubenfels P483 2 6,500 ft. (A-holotype; K, RSA, SBT-isotypes), Brass 29127 s, 7,200 ft. (L), Womersley NGF 11038 s 6,700 ft. (BRI, K, L), NGF 13922 g 7,200 ft. (Kk, L). Papua: Mt. Tafa, Cent. Div., Brass 5107 s 8,000 ft. (BRI, NY). Ridge betw. Adai and Turui Rivers, Lane-Poole 397 s (A, K). The apiculate and somewhat small leaves, whose mature size is com- pletely below the considerable range of both Falcatifolium falciforme and F. taxoides, distinguish this new species. The leaf profile is straight as in F, taxoides, but without the rounded tip of that species. The juvenile leaves reach 22 mm. in length and 4 mm. in width. The bark, gray to dark brown, and flaky with large lenticels, and a red-brown inner bark, is not unusual. A remarkable specimen from the Vogelkop, an entire small plant of Falcatifolium, has distinctly smaller leaves, 6-10 by 2 mm. (Fic. 6b). Inasmuch as juvenile leaves are usually distinctly larger than those of the adult, it may be that this isolated specimen represents a distinct entity. 20. Falcatifolium angustum de Laubenfels, sp. nov. 1969 | DE LAUBENFELS, PODOCARPACEAE 313 py Wy HW Seeman = ~— — eae : -_ = a oe _ a, Pee let papuanum de ecieen are portion of the holotype, de Lawbenfel P483 slightly enlarged; b [in he same, fragment of Van Royen & Sleumer $2030 from the Vogelkop, ew Guinea (L), enlarged. Arbor ad 20 m. alta; folia plantarum iuvenilis acicularia, crassiora quam lata, lanceolata, falcata, patentia, e basi curvata extrinsecus, ad apici Curvata sursum, ca. 7 cm. longa, basem versum 1.2 mm. crassa; folia plantarum adultarum minus curvata vel quasi recta, pungentia, carinata 314 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ¥¥ ep me FicurE 7. a, Falcatifolium angustum de Laubenfels, portion of the holotype, Brunig S8866 (1); b, Dacrycarpus expansus de Laubenfels, portion of the holo- type, Hoogland & Schodde 7463 (1): a and b. approximately natural size. a latere, 18-35 mm. longa, 1—2.5 mm. crassa: strobili masculi terminales vel laterales, immaturi ovati, 8 mm. longi, 2 mm. diametro; strobili feminei ignoti. Holotypus: Brunig S8866 (1), Sarawak, Bintulu. Fic. 7a. DIstRIBUTION. At low elevation along the coast of Sarawak. Sarawak, Bintulu, Brunig $8860 & 300 ft. (L), 58866 8 400 ft. (t-holotype), S963 j 500 ft. (kK, L). Kuching, Anderson 12448 s 800 ft. (K). This distinct new species with its narrow but nevertheless bilaterally flattened leaves is intermediate between the other species of F alcatifolium and Dacrydium, and seems to represent an early stage of the development of the genus. In the transition between seedling leaves and norma foliage leaves of F. taxoides are found leaves of identical morphology to the adult leaves here. The bark is purplish-brown, irregularly flaky to scaly, weathering gray. [To be continued | VotumE 50 NuMBER 3 JOURNAL OF THE ARNOLD ARBORETUM HARVARD UNIVERSITY B. G. SCHUBERT EDITOR T. G. HARTLEY C. E. WOOD, JR. D. A. POWELL CIRCULATION | fi 23 908 a 2 r THE JOURNAL OF THE ARNOLD ARBORETUM Published quarterly by the Arnold Arboretum of Harvard University. Subscription price $10.00 per year. ss RaTION, 16 East 467TH Srreet, New York, N.Y. Volumes I-XLV, Seegtos won = available from the Kraus Reprint Corpo- 10017. Se see riptions and remittances should be oo to Miss Dutcirs A. Divinity A RIDGE, Massa- Subscriptions PowELL, ARNOLD ARBORETUM, 22 VENUE, CAMB cHusETTs 02138. 52. CONTENTS OF NUMBER 3 A Revision oF THE MALESIAN AND Paciric RAINFOREST CONIFERS, -ACEAE, IN PART (Concluded). David J. de Lau- Sia Mastsvane AND Paiesenioke oF PrI- iM. H. Zimmermann and P. B. Tomlinson .... a ee AND FLOWERS IN NA (Pazatat. Natale W. 1) Cees 315 370 411 JOURNAL OF THE ARNOLD ARBORETUM ._—,,:° VoL. 50 Jury 1969 NUMBER 3 A REVISION OF THE MALESIAN AND PACIFIC RAINFOREST CONIFERS, I. PODOCARPACEAE, IN PART * Davin J. DE LAUBENFELS Dacrycarpus (Endlicher) de Laubenfels, stat. nov. Podocarpus sect. Dacrycarpus Endlicher, Syn. Conif. 221. 1847. Type species: Podocarpus imbricatus Blume [Dacrycarpus imbricatus (Blume) de Lau- benfels]. Podocarpus sect. Dacrydioideae Bennett ex Horsfield, PI. jav. rar. 41. 18 8. ype species: Podocarpus dacrydioides Rich. [Dacrycarpus dacrydioides (Rich.) de Laubenfels]. Folia parva vel squamata, Strobili feminei terminales; receptaculum verruculosum, tandem carnosum; unus vel duo bracteae terminalae fer- tiles, cum ovulo connatum in forma crista superans; ovulum inversum epimatium contingens. The distinguishing character of Dacrycarpus is the union of the bract with the seed and seed scale on one side, forming a projecting crest particularly noticeable on immature fruit. As in most of the family, the seed is inverse and Dacrycarpus resembles Podocarpus, of which it has long been treated as a section, because of the union of the fertile scale with the seed and its distinct receptacle. In addition to the fusion of the fertile bract with the corresponding scale and seed, however, is the fact that the cone is produced terminally on leafy branches and not on specialized shoots or peduncles. The leaves of Dacrycarpus also differ markedly from Podocarpus and resemble those of Dacrydium, being sometimes difficult to distinguish in the sterile form. There are, in addi- tion, New Zealand species of Dacrydium in which the seed is covered by - * Continued from volume 50, p. 314 316 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 the leaves are distinctly bilaterally flattened, often spread out distichously, yet this character is usually lost in the mature form. Elsewhere among the conifers only in Falcatifolium and Acmopyle are bilaterally flattened leaves found but without change in the adult form. Where the foliage leaves are bilaterally flattened or acicular, sterile specimens of Dacrycar- pus can be identified by the sharply dimorphic leaves because the penul- timate branches always have bifacially flattened leaves or scales. Dacrycarpus is composed of nine species ranging from Burma to New Zealand. Some of the species are geographically isolated and the one species in New Zealand (D. dacrydioides) is not tropical in habitat. The various species are distinguished primarily by the shape of the involucral leaves subtending the receptacle and the shape of the foliage leaves. Most discussion of specific differences in the morphology of the seed-complex simply involves degrees of maturity. Among the various species there are several wide ranging groups. A more or less scale-leaved type in- volves D. dacrydioides and D. imbricatus and is the most important for lumber and afforestation. Longer acicular leaves and a relation to moist habitats characterize D. steupii and D. vieillardii. Long involucral leaves and a mid-mountain distribution are characteristic of D. cumingii and D. cinctus. The remainder of the species are mostly localized and are found at high elevations. KEY TO THE SPECIES OF DACRYCARPUS 1. Involucral leaves spreading and not oo the seed and receptacle at all; mature leaves not distinctly flatten 2. Leaves short to scale-like (less nee 2m 3. Pollen cones terminal, linear; ni leaves very short, longer than the foliage leaves. ................ D. Hoong are 3. Pollen cones lateral, ovoid; involucral leaves about as long as the re- ceptacle and longer than the foliage leaves 4. Leaves slender (0.4-0.6 mm. wide). 5. Leaves imbricate......... 21a. D. imbricatus var. imbricatus. 5. Leaves spreading. .......... 21b. D. imbricatus var. patulus. 4. Leaves robust (0. 61 .O mm. as, eaves spreading. .......... 21c. D. imbricatus var. robustus. 6. Leaves Sib acate eRe are ee 21d. D. imbricatus var. curvulus. 2. Leaves elongate (at least 2 mm.). 7. Involucral leaves less than 2 mm. (shorter than the foliage leaves); foliage leaves strongly variable and more or less imbricate. ....---- a eh > on Cae ens oh SEs ely 72. D. vieillardii. 7. Involucral leaves more than 3 mm. (longer than the foliage leaves) ; foliage leaves more constant and strongly spreading. ......------*: a Cue Np an eee ea te gt Ss gel ER Cie Pee, CoM AAA Glen EN A 23. D. steupii. 1. Involucral leaves clasping the seed and receptacle; mature foliage leaves ttened, 8. Leaves bilaterally flattened 9. Involucral leaves long ‘(7-10 _ surpassing the mature seed; fo- liage leaves slender. 24. D. cumingit 1969] DE LAUBENFELS, PODOCARPACEAE 317 9. Involucral leaves short (5-7 mm.), not covering mature seed; fo- Hage leaves POuuUSt. . <5 isnds cwieoems sccee die, 25. D. kinabaluensis. 8. Leaves bifacially flattened. 10. Involucral leaves long (5-6 mm.); foliage leaves long and narrow (2-$ mm, by GAO6 mii Ve oboe cs ac no oon bce, 26. D. cinctus. d wide (2-4 mm. by 0.6—1.0 mm.). 11. Pollen cone lateral; seed not large (5-6 mm. long); foliage SOAVOR SURCACH. Ss o55 a de sk cnn wae cubes : nSUS. - Pollen cone terminal; seed large (7-8 mm. long); foliage leaves She eo adda osha eee eee 28. D. compactus. — ji — 21. Dacrycarpus imbricatus (Blume) de Laubenfels, comb. nov. Podocarpus imbricata Blume, Enum. Pl. Javae 1: 89. 1827. Lectotype: Blume 5.n.,. W. Java. Podocarpus cupressina R. Br. ex Mirb. Mém. Mus. Hist, Nat. Paris 13: 75. 1825 (nomen); R. Br. ex Horsfield, Pl. Jav. Rar. 1: 35. ¢. 10. 1838. Type: Horsfield 5.n., Java. Podocarpus horsfieldii Wallich, Cat. No. 6049. 1832. Nomen nudum, Nageia cupressina (R. Br.) Muell. Phyt. New Hebr. 20. 1874. Tree up to at least 30 m. tall; bark dark brown or blackish on the surface, weathering gray, inside a rich red-brown and granular (slightly fibrous), breaking off in small thick scales with a rough surface; juvenile leaves bilaterally flattened and distichous, nearly linear, curving outward from the base and upward at the tip, narrowing rapidly to a fine mucro, 10-17 mm. long and 1.2-2.2 mm. wide, shorter toward the branch tip and base, the first leaves at the branch base short and acicular, the whole foliage branch of limited growth; leaves on seedlings and on penultimate branches quite distinct, bifacially flattened, lanceolate, mucronate, im- bricate, decurrent, 2-4 mm. long and 0.7—1.0 mm. wide; terminal shoots On young plants sometimes very long, whip-like, up to 20 cm.; on older plants more compact, the foliage leaves becoming progressively smaller, fertile specimens sometimes having distichous and bilaterally flattened leaves 3-5 mm. long and 0.6—-0.8 mm. wide; foliage leaves in older trees eventually becoming short and needle-like or more or less scale-like, about 1-1.8 mm, long, strongly keeled and acute but neither flattened nor distichous; pollen cones lateral or rarely terminal, subtended by a few scale leaves on a branchlet 1-3 mm. long, oval but elongating with the shedding of pollen, to 6-12 mm. long and 2-2.5 mm. in diam. (about 5 mm. long before elongating); microsporophyll triangular, acute to apiculate; seed cone terminal, often on a short lateral branch bearing scales which become elongated just below the receptacle, forming an in- volucre, the involucral leaves spreading and generally less than 4 mm. long, acicular and sharply pointed; seed cone a short, warty, glaucous receptacle 3-4 mm. long, formed of enlarged bract bases, the tips of one Or two bracts (resembling the involucral leaves) projecting from the receptacle, one or two terminal bracts fertile, the whole receptacle be- 318 JOURNAL OF THE ARNOLD ARBORETUM [vor. 50 coming red upon maturity; mature seed globose, slightly ribbed on the back with a blunt crest, 4-6 mm. in diam., 5—6 mm, lon The short acicular or scale leaves and the lateral more or less oval pollen cone distinguish this widespread species from other members of the genus. Longer leaves occur on fertile specimens but, if present, are less than 5 mm. long and very robust (var. robustus) or are distichous. Dacrycarpus imbricatus can be subdivided into four varieties on the basis of mature foliage leaf form (Fic. 8). The reproductive structures and immature leaves of the varieties are indistinguishable. When identifying these varieties, care must be taken to compare the leaves of only the ultimate foliage branchlets and not the distinct penultimate scale-cov- ered branches (the penultimate branches of all varieties resemble the foliage branches of var. imbricatus). 21a. Var. imbricatus. Mature foliage leaves strongly appressed, slender, about 1.5 mm. long and 0.6 mm. wide, the whole foliage branch 0.75—-1.25 mm, in diam.; in- volucral leaves 2-4 mm. long. Fic. 8a. DisTRIBUTION. Scattered and common in rainforests from low elevation up to about 3,000 meters, but particularly from 700 to 2,400 meters (agriculture has commonly destroyed the forests at low elevation); in Java and the Lesser Sunda Islands, and occasionally in Celebes and Borneo. Map 7. s (NY). G. Pangranggo, Schiffner 1475 & 1,900 m. (L), Palmer & Bryant 988 j 2,900 m. (us), Winkler 1866 j 2,400 m. (x). Gegerbintang (Preanger), Den Berger 549 2 1,100 m. (x). Mt. Tankuban Prau, Anderson 67 s (K). Lembang, Junghuhn sn. 2 (1). Bandang, Junghuhn sm. s (L). Takokok (Preanger), 1,150 m. Koorders 15535 @ (A, L), 27704 s (L). G. Besser, Winckel sm. (L). G. Ungaran (Semarang), 1,000-1,350 m. Koorders 1283 s (L), 1284 j (L), 1285 j (1), 27705 @ (x, L). G. Kukusan (Lawu), Elbert 52 j 1,500-1,700 m. (L). Ngebel (Madiun), 1,450 m. Koorders 1278 s (1), 1279 j (L), 1280 ¢ (L), 1281 s, j (L), 1282 j (t), 29188 j (L), 29189 2 (a), 38626 j (L), 38652 j (). G. Ardjuno (Pasuruan), Koorders 38187 2 2,100-2,400 m. (L). Ngadasarl, Koorders 37922 s (t), 37923 j (K, L). Ngadiwono, La Rinere s.n. s 1,600 m- (t). Idjen Plateau (Besuki), 1,700 m. Koorders 1290 j (L), 1296 2 (2); 1297 j (t). G. Kendeng, Koorders 28507 9 (a). G. Tapandajan, Coert 637 j 1,750 m. (L). G. Tenga (Parverua), Dugeh 1382 j 1,600 m. (L). Parverua, Oillerings 175 s (ut). G. Guntar, Anderson 429 j 4-6,000 ft. (x). E. Java, Coert 1437 $ VIEILLARDII a Mars showing distribution of: 7, Dacrycarpus imbricatus (Blume) de Lau- nfels, var. imbricatus and var. patulus de Laubenfels; 8, D. nape ws tae de Laubenfels (dots north of line), and var. "curvulus (Miq ubenfels (dots south of line); 9, D. steupii (Wasscher) de pr a D. vieillardii known only from New Caledonia. 320 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 (L), Went s.n. j (L). Without loc. Blume 492 @ (tL), s.n. 2 (L-lectotype of Podocarpus imbricata), Junghuhn s.n. j (L), Horsfield sn. 2 (K-holotype of Podocarpus cupressina; GH-isotype), 108 s (K), 1166 2 (Kk), Korthals s.n, } (L), van Hasselt s.n. s, j (L), Simmoro s.n. s 3-4,000 ft. (L), Coert 1209 s (L), Zollinger 2262 2, j (A, 2), De Vriese sn. 2 (kK), Miquel sn. s (K). Lesser Sunda Is. Batt: Mt. Batukan, Kostermans, Kuswata, Sugeng & Supadmo KK & SS 138 2 1,300 m. (A, K, L), Sarip 371 j 1,930 m. (1). Buleleng, N/FS bb17269 8 1,300 m. (A, L). Lompox: Mt. Rindjani, Elbert 2266 j 700-1,700 m. (A, L). Lenek (mid.), NJFS bb15504 2 700 m. (K, L). Plambi (SW.), Elbert 2428 j 200-400 m. (1). SumBawa: Batu-Lanteh Mts. (N.), Elbert 4191 j 1,500- 1,700 m. (A, L, us). Frores: Rensch 1307 j (x). G. Kasterso, Posthumus 3235 j 1,800 m. (1). Sumpa: Lairondja (E.), [but 547 j (L), NIFS bb9003 j 1,000 m. (k, L). Trmor: Nenas (mid.), NIFS 6b11803 2 1,600 m. (1). Mt. Perdido (cent. Port.), Van Steenis 18267 8 1,600-1,750 m. (L). Without loc., Forbes 3855 8 (A, L). Sarawak. Kuching, Clemens s.n. j (Ny). Mt. Dulit, Richards 1768 j 1,300 m. (K, L). Kapit, Upper Rejang R., Clemens 21066 j (NY). Brunel. B. Ulak, Ashton BRUN 1032 j 4,300 ft. (kK, x). B. Pagon, Ashton BRUN 1065 8 4,750 ft. (x, L). North Borneo. Ranau, Sadau 42890 2 4,920 ft. (K, L), Sario SAN 32246 & (x), Lajangah SAN 33085 j (x). Tambunan, Mikil SAN 32070 j (K). Borneo. Sakumbang, Korthals s.n. s, j (1). B. Raja, Winkler 1035 j 1,700 m. (L). Celebes. G. Bantaeng, Biinnemeyer 11903 s 2,300 m. (K, L), 12019 2 2,060 m. (A, L), NIFS bb5460 2 2,000 m. (L), Everett 42 j 7- 10,000 ft. (x). Roto (Masamba), NJFS bb24957 j (L). G. Kambuno (Ma- samba), Eyma 1369 j (L). Enrekang (Rantelmo), NJFS bb29195 j 1,600 m. , L). Upper Binuang, NJFS 6b20202 j (A, K, L, NY). Mt. Mambuliling, De- Froidville 173 j (x). Betw. Angin-Angin and Pintealon (Enrekang), Eyma 570 j 1,550-2,600 m. (A, x, L). ILLusTRATIONS. BENNETT, J. J., Pl. Jav. Rar., ¢. 10. 1838, as Podo- carpus cupressina; BLuMer, K. L., Rumphia 3: t. 172 & t. 172B. 1849, as Podocarpus cupressina; Pricer, R., Pflanzenreich IV. 5 (Heft 18): fig. 7E. 1903; Nat. Pflanzenfam. ed. 2. 13: ¢. 124E. 1926, as Podocarpus im- bricatus; Koorprers, S. H., & Tu. VALeton, Atlas der Baumarten von Java 3: t. 585 & 586. 1915, as Podocarpus imbricata; WASSCHER, J., Blumea 4: ¢. 4, fig. 2, 1941, as Podocarpus imbricata. The variety imbricatus is well known in Java and the Lesser Sunda Islands and is widely cultivated. Because there have been only scat- tered collections elsewhere, the possibility of artificial introduction must be considered. Juvenile plants can not be identified to variety in this species so they have been assigned to whatever mature form is known in the vicinity. Some rather large juvenile leaves appear in the col- lections from the Lesser Sunda Islands. 21b. Var. patulus de Laubenfels, var. nov. Podocarpus kawaii Hayata, Bull. Econ. Indochine 20: 439. 1917. Type: Hayata in 1917, Tonkin. Folia patula, acicularia, falcata, basi carinata, acuta, 0.8-1.5 mm. longa, 0.4-0.6 mm. lata; ramuli foliis inclusis 1-2 mm. diametro; folia involu- 1969 | DE LAUBENFELS, PODOCARPACEAE 321 tae fii) i i obidudatedes dlitiliidwulia Hitliuai wily ijubudlinda 9 stout nln vata uth it IGURE 8. a, iia iis eg ml (Blume) de Laubenfels var. chigtonn fragmenta showing mature foliage form; b var. oe “yon de Lauben a of holotype de Laubenfels P- ma hg ete < 0.9 _ var. robustus de ” jc fels, fragments showing mature foliage d, var. Curt — ee ) de Laubenfels. fragments, ane ace nea form, c and x 0.8 = a. cralia 1-3 mm. longa. Holotypus: de Laubenfels P328 (A), Fiji Nandari- vatu. Fic. 8b DistriButTion. Scattered and common in rainforests from low eleva- tion up to 2,500 meters, particularly from 700 to 1,700 meters and lower where moist forests occur: from Upper Burma to Fiji, particularly from South China to Sumatra, otherwise apparently in a more or less discon- 322 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 tinuous distribution overlapping with other varieties of the species and in isolated populations east of New Guinea. Map Burma. Hukong Valley, Hole 21 s (kK). Serpentine Mines (S. of Hukong Valley), em 5007 j 1,600-2,600 ft. (cH, K). Tampyu (Kachin), Thompson (1896) ¢ (xk). Northern Triangle, Arahku, Kingdon-Ward 20626 j (A, BM), 21295 j (A, BM), 21393 s 4-5,000 ft. (A, BM), 21626 j (A, BM). Thailand. Nakhan Rachasima, Phengkhlai 568 j (K). Pulom Lo, Dan Sai, Kerr 5788 2 1,000 m. (BM, K). Kao Kuap, Kerr 17715 s 500 m. (B BM, K). Kao Soi Dao, Trang, Kerr 19435 j 500 m. (kK). Botong, Pattani, Kerr 7648 6 600 m. (K). Laos. Betw. Dasia and Cateng, Fania Prov., Poilane 16092 @ (a). Tram-la, Tranninh Prov., Poilane 2147 j (A, K, L). Boloven near Attopeu, Poilane 15922 2 (NY). Without loc., ace 1932) s (L). Cambodia. Kuang Repoe, Opong Prov., Pierre 5528 s (in mts.) (A, K). Sckral Mts., Samrongtong Prov., Pierre 5528 PP. j a x, NY). Phnom Penh Forest, Bejaud 718 3 (ILL). Elephant Podocarpus kawaii). Annam QUANG Trr PROV.: Dent de Tigre, Poilane 10293 Q@ (A, K, L). Bach Ma (N. of Da Nang) Poilane 29960 2 (ILL). Dong-tri Mas- sif, Poilane 10995 j (a, L). Dong-co-pah Massif, Poilane 11110 2 (A, L, NY). Without loc. Poilane 13644 2 (a, K). SOUTH: ‘Near Dakto, Pozlane 35595 j (1), Dalat (Lang Bian Massif), Evrard 1779 @ (a, NY), 238 j (A, NY), Che- valier 30027 j 1,400 m. (A), Poilane 4038 j (A). Nonh-hoa (near Nhatrang), Poilane 6509 2 (a, NY). Nhatrang, Poilane 3387 j (A, K), 9103 j 500-1,500 m. (ILL). Chapu, Petelot sm. 2 1,500 m. (NY). Without loc. Delacour & Low (1927) 2 (Bm), Kloss s.n. s 5,200 ft. (pM), Vim sn. @ 1,500 m. (us). China. Kwangtung-Tonkin border, Tsang 27332 j (A, K). Chen Pien Dist, Kwangsi, Ko 55900 j (a). Tsin Hung Shan, N. Him Yen, Ching 7034 j 4,000 ft. (A, NY, us). Kwangsi, Wang 39608 2 (a). Hainan. Fan Ya (5 Finger Mt.), Chun & Tso 44250 s 4,000 ft. (a, 1, Ny), McClure 8705 2 700-1,000 m. (A, BM). Seven Finger Mts., Liang 61783 j (a, Ny). Dung Ka, Chun & Tso 43955 &, j 2,400 ft. (A, Ny). Kan-en Dist., Lau 3556 j (A). Without lo oc. McClure 18304 j 1,000-1,500 m. (ny), 18279 9 (xy), How 72870 2 (Bm), Chun 1390 2 (A); Liang 65187 j (A, NY), 65257 j near summit (A, NY, us), Tang 438 j (a), Wang 35591 2 (NY, US). Malaya. Kedah Peak, Low 28 j (K), Kochumen 70988 j 3,200 ft. (Kk, L). Penang, Curtis s.n. j (us). Gov't Hill, Penang, Maingay 2239 s (Kk). G. Batu Pateh, Perak, Wray 1198 2 (K). G. Benom, Pahang, Whitmore 3268 2 (K); ahs Telom, Strugnell 23931 j 2,800 ft. (A); Kluang Terbang, Barnes 10907 2 (x). Selanger, Pahang Track, Ridley 8636 s 1,500 ft. (A). Fraser Hill, Deris 22563 2 (xk). Batang Padang, Selangor, Murdoch 11964 j K). B. Etam, Selangor, Kekall 19814 s (K). Malacca, Mt. Tapah, Werner 13509 @ 1,000-1,600 m. (x). Karoland, Sigurunggurung, NJFS 6b5443 g 1,500 m. (L). Karoland, Tongkoh, NJFS bb28147 s (A, K, L). Karoland, NIFS bb27 68 9 1,400 m. (L), bb7708 @ (x). Simelungun, Yates 2148 j (t, Ny), Esche bb35321 s 1,200 m. (1), Lérzing 11508 j 400 m. (L); Marehat Huta, N/FS 664866 j 700 m. (t). re Yates 1987 j (t). Mt. Singalan, Upper Padang, Beccari 49 j 2,000 m. (Ft, K, L), Schiffner 1473 j 1,700 m. (L), 1474 s 2,500 m. (L), Ernst 851 j (z). Solok, NIFS 6b4130 j 1,000 m, (x). Kerintji Indrapura, 1969 | DE LAUBENFELS, PODOCARPACEAE 323 NIFS bb18752 j 1,200 m. (A, L). Siolok Daras (G. Kerintji), Robinson & Kloss S.m. $ 3,000 ft. (kK). G. Tudjuh (G. Kerintji), Meijer 6584 é 1,500-1,700 m. (L), 7267 j 1,500-2,000 m. (L), Jacobs 4483 j 2,000-2,200 m. (k, L). Taram, R. Tjampo, Meijer 6780 j 500-1,000 m. (L). Bengkulen, Redjang, Paja Magelang, Renwarin bb2436 $ (1). Kriu, Waimengaku, N/FS 6b8737 j 950 m. (L). Beng- kulen, G. Pesagi, Rappard P19 j 1,700 m. (a, 1). G. Pesagi, Liwa, De Voogd 119 s 1,800 m. (L), 134 j 1,700 m. (L). Lae Pondom, Surbeck 532 j 1,600-1,800 m. (L). Leaukavear, Balten Pooll s.n. s 1,630 m. (L). Palembang, Seminung, Rappard S28 s 1,800 m. (a). Philippines. MrnpANAo: Mt. Katanglad, Bukidnon Prov., Sulit 9896 2 1,800 m. (A, L). Lanao Prov., Alvarez 25176 j (A, us). Mt. Malindang, Mizamis Prov., Mearns & Hutchinson 4666 & (K, L, NY, US). Mt. Batangan, Warburg 14721 j (NY). Celebes. Pamula dama, B. Koroué (Masam- ba), NIFS 6b24951 2 2,000 m. (a, L). Ululu (Masamba), N/FS 6b24956 s 1,700 m. (A, L). Palu, Wuka Tampai Mt. (Masamba), NJFS 6b15155 s 2,500 m. (L). Porehu (Malili), NJFS bb19559 8 1,200 m. (A, L). Moluccas, Buru, Fakal, Toxopeus 485 j 1,100 m. (L). Morotai, Kostermans 1215 j 1,000 m. (a). Middle Ceram, Stresemann 158 j 1,000 m. (tL), 354 8 1,450 m. (t), 363 j 1,100 m. (L). Guinea, Cycloop Mts., Karstel BW 5441 $ 510m. (L). Terr. New Guinea, E. Highlands, Osaka, Womersley NGF 24928 2 4,000 ft. (LAE). New Britain. Mt. Tangis, Frodin NGF 26889 j 3,000-4,500 ft. (L). New Hebrides. Erromanga, Corbasson 18123 j 200 m. (Pp). Aneityum, Kajewski 849 Q 500 ft. (A, NY, US, z). Fiji. Nandarivatu, Gibbs 775A & Bs (BM), Smith 4901 j 800- 900 m. (L, us), 6245 2 850-970 m. (a, L, US), Degener 14315 2 (ny, us). Gil- lespie 4263 } 900 m. (ny), Lam 6876 8 850 m. (t), Vaughn 3258 s (BM), de Laubenfels P328 2 2,000 ft. (A-holotype of Dacrycarpus imbricatus var. patu- lus; K, RSA, SBT-isotypes)/ P331 j (A, RSA). Nausori Highlands, de Laubenfels P306 j 1,900 ft. (A, RSA). Namboutini, de Laubenfels P310 j 1,000 ft. (a, RSA). From its distribution, one might guess that Dacrycarpus imbricatus variety patulus is the primitive representative of the species which has been largely displaced over much of its range by other varieties, but sur- vives alone both on the western and the eastern parts of the range. Specimens from Fiji, when compared with specimens from Sumatra and southeast Asia, can not be distinguished. In the Philippines, Celebes, and New Guinea where overlap with other varieties occurs, specimens are difficult to identify because juvenile and transitional stages are indis- tinguishable and all too often are all that is collected. In the Philippines forms transitional to var. robustus apparently occur, while in Borneo and Celebes the very few mature specimens seem transitional between var. imbricatus and var. patulus. Several specimens from lower elevations in New Guinea do not have a robust form and have been referred to the var. patulus, 21c. Var. robustus de Laubenfels, var. nov. Podocarpus papuanus Ridley, Trans. Linn. Soc. London. II. 9: 158. 1916. Syntypes: Kloss in 1913, New Guinea, Mt. Carstensz and Giulianetti & English in 1897,8 Wharton Range. Podocar pus leptophylla Wasscher, Blumea 4: 414. 1941. Type: De Kock 39, New Guinea, Mt. Goliath (not seen). * Ridley refers only to Giulianetti. 324 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Folia brevia, patula, acuta, ad apicem incurvata, fortiter carinata, ro- busta, 1.2-1.8 mm. longa, 0.6-0.8 mm. lata, ramuli foliis inclusis 1.5—2.5 mm. diametro; folia involucralia 2-3 mm. longa. Holotypus: Brass 30568 (A), New Guinea, Mt. Wilhelm. Fic. 8c. DIsTRIBUTION. Scattered and common in moist rainforests from near sea level to 3,300 meters, but mostly 1,000 to 2,700 meters from North Borneo and the Philippines to the eastern end of New Guinea. Map 8. Sarawak. Mt. Poe, Beccari 2431 j 3,000 m. (FI), Clemens 20134 j summit (A, Ny). Mt. Mah, Beccari 2812 2 (FI, K). North Borneo. Penampang, Clemente 5981 j 5,000 ft. (A, K, L), 6216 s (K), Leano-Castro 5988 j (K, L), 5991 j 3,500 ft. (kK, L). Tenompok Pass (Kinabalu), Smythies S10601 2 4,500 ft. (K, L), Clemens 28631 8 5,000 ft. (A, ILL, K, L, NY), 29779 j (A, K, L, NY), Melegrito A471 j 4,700 ft. (K, L). Pentaran Basin (Kinabalu), Clemens 33618 9 8,000 ft. (A, K, L, NY). Masilan R. (Kinabalu), Clemens 51635 2 8,000 ft. (A, K, L). Mt. Gedeh (Kinabalu), Clemens 30371 j 6—9,000 ft. (Ny). Kinabalu, Colenette 579 s 8,000 ft. (kK), Clemens 28954 j 8,000 ft. (Bm, K), Chew & Corner RSNB 4084 j (x). Tiong Pass, Keith 5930 6 5,500 ft. (K, L), 5967 j 5,300 ft. (K, L). Philippines. Luzon: Mt. Santo Thomas (Benguet), Elmer 6550 2 (K, NY, us), 6551 Q (kK, Ny, us), Williams 1298 2 (cH, K, NY, US), 1299 2 (Ny). Panai (Benguet), Mearns 4405 2 7,000 ft. (x, us), Santos 31817 2 (a, us), Gillis 27255 @ (a, us), Sulit 7586 & (srt). Mt. Osdung (Benguet), Quisumbing & Sulit 82481 j (Ny). Benguet Dist., Leafio 20673 s (us), 20674 j (us). Lepanto Dist., Curran 10960 s (us), Darling 14498 j (L), Vidal 1818 s (K). Mt. Data (Lepanto), Alcasid 1847 $ (1), 1897 j (L), Merrill 4503 j (K, NY, US), 4546 J (K, L), Stern 2242 j 7,050 ft. (tL), Stern & Rojo 2289 j 7-8,000 ft. (ILL), 2292 j (ILL). S. of Bontoc, Walker 7526 j 6,000 ft. (us). Mt. Banahao, Barthe (1857) s (A). Mrnporo: Merrit 8529 j (k, Ny, us). Mrnpanao: Mt. McKinley (Davao), Kanehira 2652 j (Ny), 2726 s (xy). Tupi, Mt. Matutum (Cotabato), Sumajit (1966) 2 293 ft. (L). New Guinea. VoceLKop: Nettoti Ra., Versteegh W 10411 4 1,700 m. (L), Van Royen & Sleumer 7948A j 2,100 m. (L). Kebar Valley, Van Royen 3895 j 1,750 m. (L). Anggi Lakes, Gibbs 5540 7-9,000 ft. (kK), Versteegh BW 250 2 2,000 m. (a, L), Kostermanns 2197 s 2,000 m. (t), Stefels BW 2014 j 1,860 m. (1), BW 2006 j 1,875 m. (L). Arfak Mts., Hatam, 2,100 m. (kK, L). Cycloop Mts., Versteegh & Koster BW 14 s 750 m. (A, K, L)- Terr. New GumneA: 12 miles N. of Wabag, Womersley NGF 11260 @ 7,000 ft. (K, L), 11067 j 7-8,000 ft. (L, Nsw). Wabag, Saunders 1048 j 7,100 ft. (Z, LAE). Tambul (Mt. Hagen), Womersley NGF 14253 j 8,000 it. (L). Mt. Kum (Mt. Hagen), Womersley NGF 9430 s 7,000 ft. (a, L, Nsw). Wankl (Mt. Ha- 1969] DE LAUBENFELS, PODOCARPACEAE 325 gen), Hoogland & Pullen 5868 s, ] 2,300 m., (A, K, L, Us). Mt. Hagen, Cava- naugh NGF 3322 j (a, kK). L. Inim, Flenley ANU 2176 s 8,300 ft. (K, L). Al R. Mts. (Nondugl), Womersley NGF 5338 j 7,000 ft. (A, K, L, Nsw), NGF 5353 j (A, K, L). Waimambuno (Chimbu), Saunders, 823 j 9,000 ft. (A, L). Mt. Wil- helm, Brass 30568 2 2,650 m. (A-holotype of Dacrycarpus imbricatus var. robustus; K, L, NY, USs-isotypes), 30570 j (x, L, NY, US). Chimbu, Cavanaugh NGF 3332 j (A, K, L), Stauffer 5652 j 2,600 m. (K, L, z). Fatima R., Marafunga- Chimbu Div. (Goroka), Womersley NGF 24563 s 7,700 ft. (K, L). Marafunga, Upper Asaro Valley (Goroka), Womersley & Sleumer NGF 14013 2 8,200 ft. (K, L), Anden JARA 7 s 8,300 ft. (K). Danlo (Goroka), Saunders 861 j 8,500 ft. (L), 865 (L). Above Goroka, Womersley & Floyd NGF 6138 & 8,300 ft. (A, K, L). Purosa (Okapa), Brass 31660 j 1,950 m. (a, L, NY, US), 31852 & (a, K, L, NY, US). Wagau, Sayers NGF 21613 j 4,500 ft. (L). Samanzing, Clemens 3323 2 4,600 ft. (a, z), 5473 j (a), 8848 j (A). Sarawaket, Clemens 5586 2 7,000 ft. (A). Mt. Rawlinson, Hoogland & Craven 9553 2 6,000 ft. (kK), 9354 } (K), 9355 j (kK). Wau, Womersley & Millar NGF 8324 s 5,500 ft. (A, L, NSW), Mt. Kaindi (Edie Creek), McVeagh NGF 7581 & 5,850 ft. (A, K, L, NSW), Womersley & de Laubenfels NGF 19460 (P485) 2 7,500 ft. (A, K, L, RSA, SBT), de Laubenfels P482 j 6,500 ft. (A, K, RSA, SBT), Brass 29577 s 2,060 m. (L, US), 29598 s 2,250 m. (A, L, NY, US). 29599 s (A, L, NY, US), Havel & Nauari NGF 17134 s 7,300 ft. (kK, L). Morobe Dist., Anon. NGF 3128 j (L). Papua: Anga Valley near Ebenda (S. Highlands), Schodde 1561 s 6,500 ft. (K, tL). Alola, Carr 14194 2 6,000 ft. (A, L, Ny). Boridi, Carr 13264 j 4,700 ft. (a, Ny). Mt. Mau, Crutwell 897 j (kK). Murray Pass, Brass 4768 j 2,840 m. (NY). Mt. Scratchley, Giulianetti (1896) s 12,200 ft. (K). Wharton Ra., Giulianetti & English (1897) @ 11,000 ft. (K-syntype of Podocarpus papuanus). Mt. Tafa, rass 4962 s 2,400 m. (A, NY), 5115 j (Ny). Owen Stanley Ra., Lane-Poole 264 j 5,000 ft. (A). Sibium Ra., Pullen 5914 j 2,650 ft. (a, L), 5930 8 3,520 ft. (a, K, L). Mt. Dayman (Milne Bay), Brass 22582 Q 2,000 m. (A, K, L), 23393 3 1,700 m. (A, K, L, us). ItLustrations. Grpss, L. S., Contrib. Phytogeography and Flora of the Arfak Mountains, ¢. 4. 1917, as Podocarpus papuanus; WASSCHER, J., Blumea 4: t. 4, fig. 3. 1941, as Podocarpus papuana. There has been a great deal of difficulty in separating this variety, when treated as a species (Podocarpus papuanus), from the type (Podocarpus imbricatus) (Gibbs, 1917; Wasscher, 1941). When speci- mens of fully mature forms are placed side by side they are definitely dis- tinct, but the various juvenile and transitional forms so often met with can not be distinguished and have been greatly confused in the her- baria. Inasmuch as the reproductive structures are essentially identical, it seems best to maintain it in varietal status. Certainly where var. ro- bustus occurs, other varieties are usually rare or absent. A few mature specimens in the Philippines are more or less intermediate between varieties robustus and patulus, including Leano 20673 and (see var. patulus) Mearns & Hutchinson 4666. Perhaps these two varieties tend to merge in the Philippines. Certainly typical var. robustus specimens have come from Borneo. Variety robustus differs most from the typical variety, imbricatus. From var. curvulus it differs in the same way that var. patulus differs from var. imbricatus. From var. patulus it differs in 326 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 the same way (except for habit) that var. curvulus differs from var. imbri- catus. The status of Podocarpus leptophylla is uncertain as I have not seen the type. From its description it appears to belong to this species but perhaps not to this variety. 21d. Var. curvulus (Miquel) de Laubenfels, comb. nov. Podocarpus cupressina var. curvula Miquel, Pl. Junghuhn. 1: 4. 1851. Lecto- type: Junghuhn s.n., Podocarpus imbricata var. curvula (Miquel) Wasscher, Blumea 4: 398. 1941. Mature foliage srs strongly adpressed, robust, ca. 1.2—-2.0 mm. long and 0.8-1.0 mm. wide, the whole foliage branch 1-1.25 mm. in diam. and drooping; eine leaves 2.5-4.5 mm. long, more or less clasping the receptacle. Fic. 8d. DistripuTION. Generally on mountain ridges, often in solid stands and sometimes dwarfed or procumbent, from 1,350 to 3,300 meters in eleva- tion but mostly above 2,000 meters, from Sumatra and Java. Map 8. Sumatra. Atjeh, Gajoland, Van Steenis 8423 2 2,100-2,250 m. (a, K, L) and 8 (K, L, Nsw). Java. Mt. Gedeh (Pengalengan), Junghuhn s.n. 8 4-~7,000 ft. (L-syntype). Dieng Mts., Mt. Prahu, Junghuhn s.n. 8 5-7,000 ft. Ae Kedec, Wonosobo, Zwart O5te. 3 (t). Without loc., Junghukn sn. @ (L), % (Ny), 4 j (L), Blume s.n. Q (1). ILLUSTRATION, WasscHER, J., Blumea 4: ¢. 4, fig. 28. 1941, as Podo- carpus imbricata var. curvula. The most striking character of var. curvulus is its weeping habit, but herbarium specimens can be readily distinguished by their robust branches with adpressed scale leaves. 22. Dacrycarpus vieillardii (Parlatore) de Laubenfels, comb. nov. Podocarpus taxodioides var. tenuifolia Carriére, Traité Conif. 2: 658. 1867. Type: Vieillard 1260, New Caledonia, Paita (juvenile form). Nevada elatum Wallich var. compactum Carriére, ibid. 693. Type: Vieil- lard 1262, New Caledonia, Paita Dacrydium elatum Wallich var. tenuifolium Carriére, ibid. Type: uncertain.” ger otiey vieillardii Parlatore in DC. Prodr. 16(2): 521. 1868. Type: Vieillar Podocarpas tenuifolia (Carriére) Parlatore, (based on Dacrydium ela- m var. tenuifolium) T: N pore vieillardi (Parlatore) Kuntze, Rev. a Pl. 800. 1891. Nageia tenuifolia (Carriére) Kuntze, ibid. Tree to ca. 25 m., often much less; bark hard, slightly rough with scat- tered low lenticels, breaking off in small thick flakes or short strips, dark but weathering gray, brown and slightly fibrous or granular within; juve- arently Carriére meant to replace this by his Podocarpus taxodioides var. Pa “Cae the same type specimen) but failed to delete it from the manus cript. 1969 | DE LAUBENFELS, PODOCARPACEAE 327 nile leaves bilaterally flattened and distichous, up to 10 mm. long and 1.0 mm. broad, spreading and acute with a minute spine turned upward more or less parallel to the branch, smaller towards the base and apex of a branch, gradually reduced in size, thickened and losing the dis- tichous habit: adult leaves acicular, sometimes not bilaterally flattened, Straight, spreading at an angle of about 30°, acute, with a minute spine turned upward, not distichous, from 2 to at least 4 mm. long in the mid- dle, but beginning as scales at the base of a branch growth unit, ca. 0.4— 0.6 mm. wide, 0.4-0.8 mm, thick, sometimes continuing growth into addi- tional growth units; non-foliage leaves of main shoots scale-like, appressed, bifacially flattened, at least 2 mm. long; pollen cones lateral and sub- tended by a short stalk with a few small scales or rarely terminal on a short branch, linear, 7-12 mm. long and 1 mm. in diam.; microsporophyll triangular and acute; seed cone on a lateral or terminal scaly shoot 6-8 mm. long, the scales 0.6—0.8 mm. long and appressed, the cone subtended by 6-10 spreading involucral leaves 1-2 mm. long, robust, keeled, acute, the cone itself formed of a small warty receptacle 2-3 mm. long with one projecting sterile bract and one or occasionally two apical ovules; seed oval or globular, generally with a blunt double crest and somewhat elongated at the base, 4 mm. in diam., 5.5-6 mm. long. DistRIBUTION, Throughout New Caledonia, particularly in areas of serpentine rock, along river banks and in moist draws generally where flooding is common, from sea level to 800 meters. w Caledonia. Mt. Paéoua, McKee 17029 s 600-900 m. (Pp). Mt. Boulinda, McKee 17199 j 750-850 m. (P), 17176 j (vp), Stauffer, Blanchon & Boulet 5778 2 (P, z), Veillon 142 j 750 m. (P). Baraua R., McMillan 5173 4 & K, P), g (s K, NY, z), 2419 & (A, K, NY, P, Z). errs y pean 181 2 (K, P), Pancher s.n. 2 (Bm, Pp), White 2112 o (a), 2285 s, j (A, K, P), Franc 35 j (K, 7, gig 69 s (Pp), 205 j (P), 253 j (P), 257 3 (®), ' Buchhols 1140 s (ILL, K, P), oe K, % Sere 1040 s (P, z), McKee 2353 8 (P), 2567 6 ), mann 1533 s (Pp, z), de hairs P389 Q 165 m. Pr k, re P389a j . RSA), P4448 160 m. (a, ee Baumann-Bodenheim 15040 s (P, Z), 15041 s (P, 2) Aubréville & Heine S R. Blanche (Upper Yaté), Bernier 206 s (P), non, 241 s (P), Deckiok 1349 2 (ILL, K, - oes Q (ILL, K, P), 1464 ‘ ut, K), 1465 s (ILL, K, Pp), 1553 @ (ILL, P), 1 Q (ILL, K, P), de Lau = P11 s (spr), Baumann-Bodenheim & pee 10843 9 (Pp, z). Mare 328 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Hiirlimann 3109 j (z), 3158 6 (z), 3159 j (z). Canyon of Yaté R., Bernier 254 j (Pp), 255 s, j (2), 256 j (P), 258 2, j (P). Plaine des Lacs, McKee 1142 2 (A). Prony, Le Rat 222 j (p), 1719 s (p). Southwest, Moore 4 é (xk). Without loc. Balansa s.n. 2 (BM, K), Pancher 4s (p), Mueller 68 s (Pp), Sarlin 237 s (P), 341 s (p), Baudouin 335 j (P). ILLUSTRATIONS. PincER, R., Pflanzenreich IV. 5 (Heft 18): fig. 7F. 1903; Nat. Pflanzenfam. ed. 2. 13: fig. 124F. 1926; Sar in, P., Bois et Foréts de la Nouvelle-Calédonie, ¢. 25. 1954, all as Podocarpus vieillardii. The elongated pollen cones distinguish Dacrycarpus vieillardit from other species except D. dacrydioides (the pollen cones of D. steupii are not known). The leaves of D. dacrydioides are far shorter than those of D. vieillardii and the pollen cone is normally terminal rather than lateral. The leaves of D. steupii are shorter and more spreading while the involu- cral leaves are longer than the foliage leaves, opposite to the condition in D. vieillardii. The species with leaves resembling those of D. vieillardii have much longer involucral leaves and sharply spreading and not im- bricate foliage leaves. The low elevation river-bank habitat is also a unique character. 23. Dacrycarpus steupii (Wasscher) de Laubenfels, comb. nov. Podocarpus steupii Wasscher, Blumea 4: 405. 1941. Type: NIFS 6622837, Celebes, Rantelmo (not seen). Tree to 36 m. but usually much less; bark brown or gray, inner bark pink, peeling in thin strips; juvenile leaves bilaterally flattened and distichous, up to 8 mm. long and 0.9 mm. thick, becoming shorter and not distichous, transitional leaves (sometimes fertile) variable in length, the longest in the middle of a shoot, 3-4 mm. long, acicular, tip pungent but turned upward parallel to the branch, becoming more constant in size at 2 or 2.5 mm. in length as a mature form, strongly keeled on the sides and back and spreading at an angle of 60° or more from the stem, leaves on non-foliage branches lanceolate, bifacially flattened, almost ap- pressed, 2-3 mm. long; pollen cones unknown; seed cone on a short leafy lateral shoot 3-5 mm. or more long, the involucral leaves at the base of the cone elongated and becoming widely spreading as the seed develops, 3-5 mm, long, the cone made up of a small warty receptacle 2-3 mm. long with a sterile bract protruding on one side, one or two terminal bracts fertile; seed globular with a small crest, 5—6 mm. long and 4.5-5 mm. in diameter. DistriBuTIoN. Locally common but widely dispersed on high wet peaks or in bogs from 1,000 to 3,420 meters in elevation, mostly 1,600 to 3,000 meters, from Borneo to eastern New Guinea. Map 9. Borneo. Peak of Balikpapan, Kostermans 7350 2 1,000 m. (A, K, L). Philip- pines. Luzon, Benguet, Curran 10829 s (L). Celebes. ENREKANG: near Pin- tealon, spur of Pokapindjang, Eyma 572 2 2,350 m. (A, BRI, K, L). Tinabang, . side of Rante Mario, Eyma 675 2 3,000 m. (A, BRI, K, L, LAE), 778 j (® tL), ~~ 1969] DE LAUBENFELS, PODOCARPACEAE 329 Manado: Palu, E. of Linden Sea, Blumbergen 3976 s 2,250 m. (A, L), 3977 j (L). New Guinea. VocELKop: Aifat Valley, Moll BW 12820 s 860 m. (L), 12840 Q 920 m. (L), 12876 s 1,050 m. (L). WesTERN HALF: Wissel L., Eyma 5101 s 1,750 m. (A, K, L). Kadaitadie, E. of Motito, Wissel L., Vink & Schram BW 8667 s 1,900 m. (L, LAE). Baliem R., Brass & Versteegh 11187 2 1,600 m. (a, K, L, LAE). Terr. NEw GUINEA: Wabag near L. Inim, Flenley ANU 2175 s (K, L), 2769 2 8,300 ft. (kK, L). Aiyura, Womersley NGF 4428 2 6,000 ft. (a, BRI, K, L, LAE). Sattleberg, Sambanga, Clemens 7258 s 5,000 ft. (A), 7562A s (a), 7902B 2 6,000 ft. (A). Mt. Amungwiwa, S. of Wau, Womersley NGF 17939 s 11,400 ft. (L). Papua: Ialibu, L. Buneh (S. Highlands), Pullen 2716 2 6,950 ft. (BM, L, LAE), 2716A j (L). Uriko, road from Woitape to Kosipi (Cent. Div.), Van Royen NGF 20289 ° 6,500 ft. (x; 4). ILLUSTRATION. WasscHER, J., Blumea 4: ¢. 4, fig. 4. 1941, as Podo- carpus steupii. The preference for wet conditions which appears to characterize this species probably explains why it is only occasionally found over its broad range. Many specimens have been filed with other species. Sterile speci- mens are distinguished by the short spreading acicular leaves becoming nearly uniform in size on mature specimens. The leaves are generally shorter and less (or not at all) bilaterally flattened than for comparable stages of Dacrycarpus cumingii. On the other hand the leaves of D. com- pacta are short and uniform but differ in being distinctly bifacially flat- tened, fairly broad, and nearly appressed. The seed cones in each case Sive positive identification. The one specimen from the Philippines is somewhat uncertain because it is sterile and more or less juvenile. 24. Dacrycarpus cumingii (Parlatore) de Laubenfels, comb. nov. Podocarpus cumingii Parlatore in DC. Prodr. 16(2): 521. 1868. Lectotype: Cuming 803, Luzon, Mt. Banahao. Nageia cumingii (Parlatore) Kuntze, Rev. Gen. Pl. 800. 1891. : Podocarpus imbricatus Blume var. cumingii (Parlatore) Pilger, Pflanzenreich IV. 5 (Heft 18): 56. 1903. Tree to at least 20 m.; juvenile leaves bilaterally flattened and dis- tichous, up to 12 mm. long and 1.2 mm. thick, the tip curved and parallel with the branch, soon losing the distichous habit and becoming coarser; Mature foliage leaves bilaterally flattened, spreading, somewhat falcate, acute with a fine spine curved upward, strongly variable in length, the longest in the middle of a branch unit, 6 mm. long and 0.6 mm. thick; leaves on non-foliage branches bifacially flattened, lanceolate, nearly ap- Pressed, 2-4 mm. long and 0.6 mm. wide; pollen cones lateral on short shoots 2-5 mm. long, oval, 8-10 mm. long and 2-3 mm. in diam., micro- sporophylls lanceolate; seed cone on a short, usually lateral shoot 6-10 mm. long or more, leaves elongated greatly at the base of the cone so that the curving involucral leaves surround even the mature seed, the longest at least 10 mm. long and 0.5 mm. thick, the cone formed ofa small warty receptacle 2-3 mm. long with one or rarely two apical fertile 330 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 bracts; the mature seed with a distinct asymmetrical crest, 4.5—-5 mm. in diam. and 5—6 mm. long. DistRIBUTION. In mountain forests up to 3,000 meters in the Philip- pines, Borneo, and (according to Wasscher, 1941) in Sumatra. Map 10. Sarawak. Mt. Penrissen, S. of Kuching, Jacobs 5017 2 1,400 m. (Kk, L, US). Philippines. Luzon: Mt. Polis (Mountain Prov.), Steiner 2207 s 2,040 ft. (L). Mt. Pulog (Benguet), Curran, Merritt & Zschokke 18049 2 (1), Ramos & Edano 45005 s (a), Steiner 2032 j 2,400 m. (L). Mt. Banahao (Tayabas), Cum- ing 803 @ (a-lectotype; F, K, L-isotypes), Foxworthy 2387 & (L). Loher 7137 Q (kK, us), 7138 8 2,250 m. (kK), Curran & Merritt 7886 2 (Ny, us), Ramos 19557 s (us), Klemme 66 6, j (a), 874 2 (ny, us), Whitford 951 2 (x, NY, us), Holman 4 @ (a), Vidal 623 9 (a, K, L), Barthe (1857) @ (a-syntype), Ocampo 27926 s (A), Robinson 5656 2 (srt), Sulit 30051 s (Bri), Lucban (Tayabas), Elmer 7465 2 (A, K, L, z). Mt. Mahaihai (Luconia), Wilkes s.n. s (cx), Central Luzon, Loher 4852 2 (A, K, us). Without loc. Loher 2138 s, J (us). Mrnxporo: Mt. Halcon, Merrill 5563 s (Ny). Panay: Mt. Midiaas (An- tique), Yoder (1905) 2 (L). Necros: Canlaon Volcano along lake (E. Negros), Edano 21935 j 1,860 m. (L), 21944 j (L). Mrnpanao: Mt. Apo (Davao), Elmer 11684 @ (A, K, L, NY, US, z). Mt. McKinley (Davao), Edano 993 s (A). ILLUSTRATION. WaASSCHER, J., Blumea 4: t. 4, fig. 5. 1941, as Podo- carpus cumingii. The long involucral leaves are the most distinguishing character of this species, being approached only by Dacrycarpus cinctus which has very different leaves. The bilaterally flattened mature foliage leaves are the same as D. kinabaluensis but not as robust. Juvenile leaves of D. steupit resemble mature leaves of D. cumingii. 25. Dacrycarpus kinabaluensis (Wasscher) de Laubenfels, stat. nov. Podocarpus imbricatus Blume var. kinabaluensis Wasscher, Blumea 4: 400. 41. Type: Clemens 27854, North Borneo, Mt. Kinabalu. Small tree or shrub down to 2 m. high; juvenile leaves bilaterally flattened and distichous at first, at least 10 mm. long and 1.2 mm. thick, falcate with a long upturned acuminate apex; mature foliage leaves bi- laterally flattened, robust (stiff), falcate, spreading at about a 30 angle, strongly curved upwards at the apex and pungent, the spine not projecting, markedly variable in length, becoming reduced on older plants so that the longer spreading leaves may be as short as 2 mm., 0.5 mm. thick, and nearly quadrangular in cross section; leaves on non- foliage branches bifacially flattened, lanceolate and pungent, curved up- wards, 1-2 mm. long and 0.5-1.0 mm. wide, the size varying with the robustness of the branch but not within a given branch; pollen cones lateral on a short branchlet about 3 mm. long, globular, 8 mm. long an 3 mm. in diam.; seed cone on short lateral or terminal shoots 5-15 mm. long and bearing, as is usual for the genus, the non-foliage type leaves; at the base of the cone the leaves elongated to the size of foliage leaves “ ~~ 6 ae 8 G ats Te 3 & c ‘ ra POA iS e: na : 4 pr ry) \ é ee / Ie Pi Z ~ . . a a Fhe yj e re ss set € bd ay, & ee, oe a BS See ° AL SS" . ' : rs n ~~ COMPTONII! ig KO Be Tey & i 5 Maps § a er distribution of: 10, Dacrycarpus cumingii (Parlatore) de ar (dots west of line), and D. cinctus (Pilger) de oe (east of line); 11, D. compactus (Wasscher) de Laubenfe ts), D. b , known 12, De see sede vitiensis (Seemann) de epithet (dots), D. comptonii (Buchholz) de Laubenfels, known ad from New Caledonia; 13 ichi i (Hickel) 4 ee (dots north of line), , D. nagi (Thunberg) de Laubenf S32 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 and nearly surrounding the young ovule but not reaching beyond the middle of the ripe seed, up to 7 mm. long and 1.0 mm. thick, the cone made up of a small warty receptacle 2-4 mm. long with one or two sterile protruding bracts and an apical fertile bract; the mature seed with a distinct asymmetrical crest, 5—5.5 mm. in diameter and 7 mm. long. DIsTRIBUTION. In mountain dwarf forests sometimes forming almost pure stands on Mt. Kinabalu from 2,750 to 4,000 meters at the timberline. North Borneo. Mt. Kinabalu, Paka Cave area, Clemens 10636 2 (A, GH, K), 10662 2 (A), 10686 s (A), 27092 s 13,000 ft. (A, K, L, NY), 27854 @ (A, K, L- isotypes), 28910 2 11,000 ft. (A, K, L, NY), Wood & Wyatt-Smith SAN A4493 s 10,500 ft. (a, BRI), Meijer SAN 21988 2 9,500-11,000 ft. (k), SAN 29265 6 10,000 ft. (x). Mt. Kinabalu, Marai Parai, Clemens 32316 2 10-11,000 ft. (4, K, L, NY), 32317 @ 11,000 ft. (a, L, NY), 32318 j 11,000 ft. (a, L, Ny). Mt. Kinabalu, Gurulau Spur, Clemens 51201 2 (L). Mt. Kinabalu, side of Granite Dome, Clemens 29914 2 12,500 ft. (K, L); S. slope, Jacobs 5755 2 3,600 m. (K, L, us). Mt. Kinabalu, Clemens s.n. 2 11,000 ft. (pm), Nicholson SAN 17825 2 10,000 ft. (BRI, k, L), SAN 39766 2 9,000 ft. (K), Sinclair & Kadim 9146 2 10,700 ft. (kK, L), Haviland 1094 j 11,000 ft. (K), 1095 2 11,000 ft. (a, K), Chew & Corner RSNB 868 2 10,500 ft. (kK), RSNB 5887 @ 9-11,000 ft. (K), Gibbs 4216 4 12,000 ft. (x), Anderson $27079 @ 11,300 ft. (K). ILLUSTRATION. WASSCHER, J., Blumea 4: ¢. 4, fig. 2y. 1941, as Podo- carpus imbricata var. kinabaluensis. The robust form of this high-elevation species is characteristic of conifers in such places, and in general habit Dacrycarpus kinabaluensis resembles D. compactus from high mountains in New Guinea, although in detail their leaf form is, of course, quite different. D. kinabaluensis is most closely related to D. cumingii, differing in the markedly robust foliage leaves and distinctly shorter involucral leaves. The pollen cones also differ somewhat in shape. Perhaps it could be treated as a variety of D. cumingii but not, certainly, of D. imbricatus. 26. Dacrycarpus cinctus (Pilger) de Laubenfels, comb. nov. Podocarpus cinctus Pilger, Bot. Jahrb. 69: 253. 1938. Type: Clemens 5261, New Guinea, Busu River. 5 Podocarpus dacrydiifolia Wasscher, Blumea 4: 410. 1941. Type: NIF bb13633, Celebes, Pawreang Mts. Shrub of less than 4 m. to a tree up to 30 m. high; bark brown to black, hard and uneven, inner bark reddish, breaking off in rough scales or plates; juvenile leaves slightly bilaterally flattened and distichous at first, the longest 12 mm. long and 0.8 mm. thick, falcate and bending upward to the pungent tip, gradually changing to resemble the mature leaves; mature foliage leaves of uniform size on a branch, slightly bifacially flattened, lanceolate, falcate, eventually reduced to 2-3 mm. in length and spreading 1969 | DE LAUBENFELS, PODOCARPACEAE 333 at an angle to give the branch system a diameter of 3—4 mm., about 0.4-0.6 mm. wide, mature specimens including leaf sizes ranging up to 5 mm. in the center of a branch unit, often glaucous; leaves of non-foliage branches the same or more distinctly bifacially flattened, pollen cones terminal or lateral on a very short branch 2—3 mm. long, globular or oval, 4-10 mm. long and 2-3 mm. in diam., microsporophylls acuminate; seed cone lateral on a short branch or terminal, 5—15 mm. long, involucral leaves much longer than foliage leaves and clasping the young seed but generally not reaching past the middle of the mature seed, the longest 6-7 mm. long, the cone formed by a small warty receptacle 3-4 mm. long, with one or two project- ing sterile bracts, seed and receptacle becoming red when ripe; mature seed with a small asymmetrical crest, 6-7 mm. in diam. and 6—7 mm. long. Fic. 9a. DistripuTiIon. Mountain forests to high mountain shrubbery from 900 to 3,600 meters but mostly 2,200 to 3,200 meters, from the Celebes to the high mountains of New Guinea. Map 10. Celebes. Masamba, N/FS bb24958 s 900 m. (L). Pawreang Mts., Ulu Salu (Upper Binuang), NJFS 6b13633 2 1,800 m. (L-holotype of Podocarpus dacry- diifolia). Pinapuang (Manado), Eyma 3873 s (£4). Ceram. G. Sofia, Central Mts., Stresemann 125 j 1,300 m. (Lt). G. Pinaia, Middle Ceram, Eyma 2276 s 3,030 m. (L), Stresemann 251 s 3,010 m. (L), 276a j 2,530-2,750 m. (L). New Guinea. WesTERN HALF: Mamberamo R. (Mt. Doorman), Lam 1773 ? 3,260 m. (L). Hellwig Mts., Pulle 964 2 2,600 m. (kK, L), van Romer 736 s (L). Lake Habbema, Brass 10513 2 2,800 m. (A, K, L), 10514 j (A, L), 10675 2 3,000 m. (A, K, L), Brass & Versteegh 10447 2 2,840 m, (A, L). L. Quarles, Versteegh BW 2537 s 3,600 m. (x, L). Terr. New Guinea: Wapu R. (Wabag), Hoogland & Schodde 7166 s 9,500 ft. (A, L). Tomba, Mt. Hagen—-Wabag Road, Flenley ANU 2819 2 8,900 ft. (K, L), Robbins 238 2 8,000 ft. (A, K, L, LAE, US). Lal Valley (Wabag), Robbins 3112 @ 7,500 ft. (A). Minj-Nona Divide, Pullen 5052 2 10,600 ft. (L), 5267 s 9,500 ft. (K, L). Keglsugl (Chimbu), Saunders 804 s 8,000 ft. (L, LAE). Toromambuno Mission (Upper Chimbu), Pullen 313 2 9,000 ft. (A, BRI, K, L, LAE, US), 313A j (K, L, LAE). Mt. Wilhelm Track, Chimbu Valley, Robbins 673 2 9,000 ft. (A, BM, L, LAE). Mt. Wilhelm, Brass 30412 s 2,770 m. (A, K, L, NY, US), 30707 2 3,180 m. (A, K, L, LAE, NY, us), Stauffer 5670 8 3,250 m. (z). Kerigomna Camp (Goroka), Hoogland & Pullen tw. Mt. Kerewa and Mt. Ne, Vink 17188 j 2,890 m. (L). 2 2,880 m. (x). Mt. Giluwe, above Klareg, Schodde 2021 ¢ 8,800 ft. (K, L, oe 2104 j 9,100 ft. (x, LAE). Woitapi-Kosipi Road (Cent. Div.), Van Royen 334 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 20309 2 6,300 ft. (K, L). Murray Pass, Wharton Range, Brass 4688 2 2,840 m. (A, K, L, NY, US). ILLUSTRATIONS. WasscHER, J., Blumea 4: ¢. 4, fig. 6, as Podocarpus cincta, and fig. 7 as Podocarpus dacryditfolia. 1941. Confusion has existed between this species and Dacrycarpus compactus with which it overlaps in range, although they are not at all the same. The involucral leaves of D. cinctus are long and narrow, clasping the smaller seed, while those of D. compactus are short and triangular, barely reach- ing the base of the distinctly larger seed. Foliage leaves contrast in the same way. The terminal position of the pollen cone which is usual in D. cinctus is a character shared only with D. compactus and D. dacrydioides. The essential non-dimorphic quality of the mature foliage is shared only with D. compactus, D. expansus, and D. imbricatus (in part). There are several specimens of D. cinctus which differ from the typical form in the direction of D. compactus and could, perhaps, be recognized as forming a variety (Fic. 9b). These are: Brass 4688, 10513, 10514, 10675, Brass & Versteegh 10447, Versteegh 2537, Pulle 964, Hoogland & Pullen 5574, Pullen 5052, 5267, Lam 1773, Schodde 2021, 2104, van Romer 736, Vink NGF 12430, and Womersley NGF 14018. They differ in that the foliage leaves are somewhat broader (up to 0.8 mm.), as are the involucral leaves (up to 1.0 mm wide). The possibility exists that these are hybrids, being found at and not far below the lower elevation limit of D. compactus. 27. Dacrycarpus expansus de Laubenfels, sp. nov. Arbor ad 25 m. alta; cortex squamosus. Folia plantarum iuvenilium dimorpha, ad ramulos breves compressa bilateraliter, patentia, falcata, pungentia, ad 12 mm. longa, 1.5 mm. lata, biseratim expansa; ad ramulos magis elongatos compressa bilateraliter, imbricata, lanceolata, pungentla, ad 4 mm. longa, in basi 0.8 mm. lata; folia plantarum adultarum com- pressa bifacialiter, expansa, falcata, acuta, dorso carinata, 2-4 mm. longa, 0.6-1.0 mm. lata. Strobili masculi laterales ad ramusculis 1-2 mm. longis, ovoidei, 6 mm. longi, 3 mm. crassi. Strobili feminei ad apicem ramulorum saepe brevi 4-5 mm. longi, foliis parvis; folia involucra longiora, 3-4 mm. longa, 0.6 mm. lata; receptaculum parvulum, verruculosum, 2-3 mm. longum; semen globosum, cristatum, 4 mm. diametro, 5—6 mm, longum. Holotypus: Hoogland & Schodde 7463 (L), New Guinea, Yobobos Grass- land. Fic. 7b. Distripution. Locally common in disturbed forests in the highlands es New Guinea at 2,600-2,670 meters. New Guinea. Terr. New Gurnea: Yobobos Grassland, Laiagam Subdistrict (Wabag), Hoogland & Schodde 7463 2 8,500 ft. (L-holotype; BRI, Lak-isotypes), 7440 s (L, LAE), 7682 j (L, LAE), Robbins 3214 8 8,700 ft. (BRI, L, LAE). PAPUA: E. foot of Mt. Ambua, Tari Subdist., Vink 17502 2, é 2,670 m. (L), 17499 J (L), 17500 j (L), 17501 j (x). we Cro nn 1969 | DE LAUBENFELS, PODOCARPACEAE = ~ 2 s ~ FIGURE 9, a. Dacrycarpus cinctus (Pilger) de Laubenfels, portion of Hartley 13263 (A), typical form of the species; b, the same, fragments showing variation toward D. compactus: c, D. compactus (Wasscher) de Laubenfels, fragments. — The short distinctly bifacially flattened involucral leaves clasping the receptacle, but not the somewhat small seed, distinguish this new species from all others in the genus. The only other species which have distinctly bifacially flattened involucral leaves are Dacrycarpus cinctus and compactus. Those of D. cinctus are twice as long, clasping the seed, 336 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 while those of D. compactus are up to twice as broad below a much larger seed. As in both of these species, D. expansus lacks dimorphic foliage when mature, contrasting in appearance because of the strongly spreading rather than imbricate leaves which are also distinctly broader than those of D. cinctus. The normally lateral pollen cones also distinguish D. ex- pansus from these two species. Except for the great contrast in shape of the involucral and foliage leaves D. expansus resembles D. steupii in gross morphology, differing in habitat preference. 28. Dacrycarpus compactus (Wasscher) de Laubenfels, comb. nov. Podocarpus compacta Wasscher, Blumea 4: 411. 1941. Type: Brass 4284, New Guinea, Mt. Albert Edward. Small tree 2-15 m. high; bark hard, rough, warty, dark gray, breaking off in scales, inner bark reddish straw color; juvenile leaves bilaterally flattened, lanceolate, falcate and curved upward at the tip, acute, strongly keeled laterally, not distichous, 2.5 mm. long and 0.6 mm. thick; mature foliage leaves not dimorphic, bifacially flattened, spreading slightly, fal- cate, lanceolate, pungent, keeled on the back, 2—3 mm. long, 0.6-1.0 mm. wide (the wider probably on older plants); pollen cones lateral on a short branch about 3 mm. long or more usually terminal, 7-8 mm. long and 3 mm. in diam., microsporophylls lanceolate, acute; seed cone terminal, generally on a short branch 6-17 mm. long, involucral leaves robust, 4-5 mm. long and 0.8-1.2 mm. wide, clasping the receptacle, the cone itself made up of a small warty receptacle 3-4 mm. long with a sterile bract protruding; seed globular, with a blunt crest, 7-8 mm. long and 7 mm. in diam. Fic. 9c. DistripuTion. In high mountain forests and shrubberies often as an emergent, and sometimes the dominant tree at the tree line in New Guinea, from 3,200 to 3,900 meters. Map 11. New Guinea. WESTERN Hair: L. Habbema, Brass 9291 2 3,225 m. (A, K, L), 21104 2 3,225 m. (A). Terr. New Gurnea: Mt. Kinkain, Cent. Kubor Range (Minj), Saunders 708 2 11,800 ft. (a, L), Pullen 5111 8 11,770 ft. (a, K, 2); 5138 @ 12,000 ft. (K, x). Mt. Wilhelm, Robbins 718 2 12,000 ft. (1), Brass 29861 2 3,650 m. (A, K, L, NY, US), 29935 2 13,100 ft. (A, K, L, NY, US), 5.” § 3,320 m. (us), Millar NGF 14671 2 12,000 ft. (x, L), Womersley NGF 8852 Q 11,870 ft. (A, K, L), 8861 8 (a, K, L), Pullen 338 12,500 ft. (a, L), Hoogland & Pullen 5650 & 11,700 ft. (A, BRI, K, L, US), 5703 2 12,500 ft. (A, BRI, K, L, us); Havel NGF 17421 2 11,500 ft. (x), Stauffer 5670 8 3,250 m. (x, L), Balgooy 287 2 3,650 m. (L). Mt. Otto (E. Highland), Brass & Collins 31021 ¢ 3,460 m. (A, K, L, Ny). Mt. Piora, Kaimantu Subdiv. (E. Highland), Henty & Carlquist NGF 16566 2 10,500 ft. (k, x). Papua: Mt. Dickson, Goilala Subdist., Hartley TGH 12958 2 11,500 ft. (L). Mt. Albert Edward, Brass 4284 @ 3,630 m. (A, ny-isotypes), 4284A j (A, Ny), 4347 j (Ny), 4348 s 3,680 m. (A, NY). ILLUSTRATION. WaAsSCHER, J., Blumea 4: ¢. 4, fig. 8. 1941, as Podo- carpus compacta, 1969] DE LAUBENFELS, PODOCARPACEAE 337 The particularly large seed completely free of the involucral leaves and the small bifacially flattened not widely spreading leaves distinguish Dacrycarpus compactus from other species. D. expansus, with rather similar, though spreading leaves, has much smaller seeds and more lan- ceolate involucral leaves. The wild tree of D. compactus is a very strik- ing plant, often rising above the other shrubs near the tree line and stand- ing out with a dark green color. The short juvenile leaves are the most primitive in the genus and apparently only in this species are the juvenile leaves never distichous. ADDITIONAL SPECIES: Dacrycarpus dacrydioides (Rich.) de Laubenfels, comb. nov. Podocarpus dacrydioides Rich. Essai d’une Flore de la Nouvelle Zéland, 358. t. 39, 1832. Type: D’Urville in 1827 (not seen). Podocarpus thujoides R. Br. ex Mirb. Mém. Mus. Hist. Nat. Paris 13: 75. 1825 (nomen). Dacrydium excelsum Cunn. Ann. Nat. Hist. 1: 213. 1838 (nomen illeg., based on Podocarpus dacrydioides). Nageia excelsa (Cunn.) Kuntze, Rev. Gen. Pl. 800. 1891. Acmopyle Pilger, Pflanzenreich IV. 5 (Heft 18): 117. 1903. Type species: Acmopyle pancheri (Brongn. & Gris) Pilger. Small trees; foliage leaves linear, bilaterally flattened, distichous, with scale-like, triangular and bifacially flattened; pollen cones terminal and lateral together; seed cones on short branches which are lateral or terminal Or grouped together, becoming enlarged and warty as a receptacle with a single subterminal seed; the ovule at first inverted and partially covered by the epimatium, eventually becoming nearly erect, fused with the epimatium, and fleshy. This genus is characterized by the unique combination of the seed fused with the epimatium (fertile scale), together with an inverted ovule which becomes gradually nearly erect as it matures. The seeds of other genera of Podocarpaceae which are fused with the fertile scale do not become erect. Acmopyle shares bilaterally flattened and distichous leaves with Falcatifolium and juvenile forms of Dacrycarpus. With the latter it also Shares a warty receptacle. Two species are known, differing in the charac- ter of seed and receptacle as well as in size of the pollen cone and details of leaf-form. Both are island endemics. 29. Acmopyle pancheri (Brongn. & Gris) Pilger, Pflanzenreich IV. 5 (Heft 18): 117. 1903. Dacrydium pancheri Brongn. & Gris, Bull. Soc. Bot. France 16: 330. 1869. Type: Pancher in 1869, New Caledonia, Mount Mou. 338 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Nageia pancheri (Brongn. & Gris) Kuntze, Rev. Gen. Pl. 800. 1891. Podocarpus pectinatus Masters, Gard. Chron. III. 9: 113. 1892. Type: Hort. Sander s.n., of New Caledonian origin. Acmopyle alba Buchholz, Bull. Mus. Hist. Nat. Paris II. 21: 281. 1949. Type: Buchholz 1704, New Caledonia, Bois de Mois de Mai. Tree from 5 to 25 m. high; bark hard and smooth, weathering to a gray color, brown to tan and fibrous within, with scales breaking off on older trees; foliage leaves bilaterally flattened and decurrent, distichous, linear and spreading 60° to 75° from the branch, tapering somewhat towards the apex which is turned slightly in the direction of the shoot apex, or slightly falcate, at first with two glaucous bands on each surface asso- ciated with the stomatiferous areas but with further development this condition suppressed on the upper surface but the white bands remaining prominent below, the midrib marked by a faint line on the upper surface and more pronounced below, leaves shorter at the beginning and end of a sequence of growth with the leafy shoots never producing a second cycle of leaves but commonly continued into fertile shoots; shade leaves spread out into a flat and almost solid plane, except for the smaller leaves at either end of the branch, 16-21 mm. long and 2.8-3.0 mm. wide, slightly revolute on the margins; leaves exposed to the sun less regularly placed and more noticeably keeled, often overlapping and weakly spread into a plane, 10-15 mm. long and 1.8—2.2 mm. wide, with intermediate forms sometimes found; non-foliage leaves scale-like, triangular, bifacially flattened, keeled on the back, less than 2 mm. long, on main branches bearing foliage shoots broadly decurrent and dispersed, on fertile branches (and occasionally at the base of a leafy shoot) more or less crowded; pol- len cones terminal or often a pair (one of which is lateral) produced at the apex of a leafy shoot or on a scaly shoot which may itself be terminal or lateral either at the apex of a leafy shoot or a main branch bearing leafy shoots (on vigorous trees all of these together), subtended by a few small scales, 10-20 mm. long by 2-3 mm. in diam. (fide Hooker, 1902, to 35 mm. by 4 mm.), the microsporophylls small and triangular; seed cones terminal or lateral at the apex of a leafy shoot, or on a main shoot, or terminal or lateral on a scaly shoot which may be either terminal on a leafy shoot, or lateral on a main shoot (on vigorous trees a combination of these) ; the seed cone subtended by a peduncle 9-22 mm. long, densely covered by small overlapping scales and slightly enlarged toward the cone to a diameter of about 2 mm.; the cone formed by a fleshy warty receptacle 8-18 mm. long involving about 4 to 8 bracts whose free tips each surmount a bulge; one ovule inverted and protruding from the enveloping epimatium in the axil of a sub-apical bract, becoming almost erect and fleshy, the epimatium completely fused to the mature ovule and apparently attached for about half its length (marked by a roughened area on the seed and a charac- teristic ridge on the dried fruit); seed globular, 10-11 mm. in diameter, thick and hard. DIsTRIBUTION. Scattered in moist rainforest over serpentine rocks 1n 1969 | DE LAUBENFELS, PODOCARPACEAE 339 most of New Caledonia from near sea level to at least 1,200 meters. Growing as a canopy tree in drier areas and sometimes found in the understory within the mossy forest where it is fully fertile. New Caledonia. Upper Diahot, Tendé Forest, McKee 17540 j 500 m. (P). Mt. Colnett, Hiirlimann 1964 2 1,200 m. (Pp, z). Mt. Paéoua, McKee 17057 s 900- 1,100 m. (Pp), Bernardi 10151 2 900-950 m. (P, z). Mt. Boulinda, Veillon 136 $ 1,200 m. (P), Schmid 137 s (orstom). Crest W. of Col de Rousettes, de Lau- benfels P429 s 700 m. (A, RSA). Me Arembo, Bernier 1007 s (K, Pp). Mt. Koun- gouhaou N., McKee 17954 2 1,000-1,100 m. (P). Mt. Mou, Pancher (1869) @ 1,200 m. (p-holotype of Dacrydium pancheri), Balansa 2862 2 (BM, K, NY, P), Compton 485 2 (BM), Franc 170 @ (A, BM, K, NY, P, Z), Virot 10s (a, Pp), Le Rat 697 s (P), 980 Q (kK, P), 2594 s (A, P), Bernier 278 2 (Pp), 1309 2 (P), Buchholz 1451 2 (ILL, K, P), 1587 9 (ILL, K, P), 1587S j (ILL, P), 1593S j (IL, P), 1790 é (ILL, K, P), McMillan 5013 2 (P), 5014 s (P), de Laubenfels P130 2, 8 1,140 m. (sBT), P355 @ (A, K, RSA), P356 3 (a, RSA), Brousmiche s.n. s (P), Thorne 28704 s (p), Baumann-Bodenheim & Guillaumin 11260 s (P, z), Baumann-Boden- heim 15632 s (Pp, 2), 15633 9 (Pp, z), McKee 3517 @ 1,100 m. (A, K, P). Mt. Ouin, McKee 9795 s (K, p). Col de Mt. Dzumac, McKee 9773 s (K, P), 9774 j (K, P), 12922 8 (p), de Laubenfels P447 2 900 m. (A, RSA), Baumann-Bodenheim & Guillaumin 12714 s (ve, z), Blanchon 930 s (P). Mt. Koghis, Pancher sn. & 800 m. (P), Alleizette 142 2 (P), Brousmiche 9 2 (with Prumnopitys ferrugi- noides), Hiirlimann 1657 s 1,050 m. (Pp, z). Mt. des Sources, Hiirlimann 911 s (P, z). Mois de Mai, Bernier 276 s (P), 277 s (P), 279 s (P), 280 8 (P), 281s (P), 321 j (P), Buchholz 1354 s (ILL, P), 1388 s 200-250 m. (ILL, P), 1388A j (ILL), 1388M 8 (Pp), 1698 s (tL, P), 1698L (shade) s (Pp), 1704 2 (11L-holotype of Acmopyle alba; x, p-isotypes), McKee 3454 s 200 m. (A, K, P), Baumann-Boden- heim 13964 s (p, z), 14258 & (P, Zz), 14263 2 (P, 2), 14988 s (P, Z), 14992 s (®, Z), 15096 & (p, z), 15097 9 (P, z), 15098 s (P), 15130 s (P, Z), 15208 s (P, Z), 15213 s (z). Slope N. of R. Bleue, de Laubenfels P136 s 700 m. (sBt), P382 250 m. (A, RSA), P383 6 (A, K, RSA, SBT), P383A (shade) s (A, RSA), P4465 @ 770 m. (A, RSA), Baumann-Bodenheim & Guillaumin 10929 s (P, 2), Baumann- Bodenheim 15043 j (z), 15055 s (Pp, z), McKee 12653 s 200 m. (Pp). Bois Elec- trique, Foster 206 9 (p), de Laubenfels P377 2 240 m. (A, RSA), P378 s (A, RSA), Hiirlimann 3411 s 220 m. (z). Without loc., Mueller 44 s (111, P). Cult., Hort. Sander s.n, s (K-holotype of Podocarpus pectinatus). ItLustraTIons. Hooker, J. D., Bot. Mag. t. 7854. 1902, as rine pus pectinata; PiiceR, R., Pflanzenreich IV. 5 (Heft 18): fig. 24. yes : Gray, Jour. Arnold Arb. 28: i IB. 1947; SarLIN, P., Bois et Foréts de la Nouvelle-Calédonie, t. 22 & 23, as Acmopyle alba. 1954. The difference between shade and sun growth forms suggests that two entities are included under Acmopyle pancheri but this difference occurs regularly on single plants. A. alba differs from A. pancheri ean! larger pollen cones (18-20 mm. long by 3 mm. in diam. in A. @ : - 10-13 mm. long by 2 mm. in diam. in A. pancheri). There are absolutely no other differences between these two taxa, while even larger pollen cones are described by Hooker (1902). Because only a few examples of pollen cones are available to show variations in size and because speci- 340 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 mens can not be distinguished in the absence of pollen cones, it is felt that A. alba should not be separated from A. pancheri at this time. In the future it might seem advisable to separate them at the varietal level. 30. Acmopyle sahniana Buchholz & Gray, Jour. Arnold Arb. 28: 142. 1947. Type: Gillespie 3273, Fiji, Mt. Vakarogasiu. Small and gnarled tree 3-5 m. high; leaves bilaterally flattened, dis- tichous, spreading at an angle of 60° to 80°, linear, slightly tapering and curved in the direction of the branch apex near the blunt tip or falcate, decurrent, 10-19 mm. long and 2.0-3.2 mm. wide (wider according to Buchholz and Gray but not according to their illustrations or to herbarium upper surface, midrib faintly marked on both surfaces, margin slightly revolute, leaves smaller at the begining and end of a sequence of growth with one sequence often continuing into a subsequent growth unit; non- foliage leaves on main branches scale-like, bifacially flattened, long tri- angular, 1.5—-2.5 mm. long, keeled on the back and broadly decurrent, more crowded at the base of a foliage branch and on the peduncle of the seed cone; pollen cones terminal, 5 mm. long and 2 mm. in diam.; micro- sporophylls triangular, acute; seed cones lateral or terminal on a foliage branch, with a short (5 mm. ) scaly peduncle, the cone formed by a fleshy warty receptacle involving two bracts, the uppermost being fertile with a single inverted ovule partly covered by a broad epimatium; seed becom- ing nearly erect, rounded and elongated into a conical point, with the epimatium fused along one side, its margin forming a fringe about half- way to the micropyle, mature need not known. DistrIBuTION. Known only from two isolated mountains on either side of Viti Levu in dense low forest 800 to 1,050 meters in elevation, where it is locally common. Fiji. Vitr Levu: Mt. Vakarogasiu (Namosi), Gillespie 3273 s 900 m. (a-holo- type; K-isotype), Koroiveibau 14598 2 2,600 ft. (kK). Mt. Koroyanitu (Mt. Evans Range), Smith 4122 $ 950-1,050 m. (A, BRI, ILL, K). Without loc., Horne Sn. S (K). Intustration. Bucuuotz, J. T., & N. E. Gray, Jour. Arnold Arb. 28: ¢. IA. 1947. This rare species is of interest because its only relative, Acmopyle pancheri, occurs in New Caledonia, not a common combination. Decussocarpus de Laubenfels, gen. nov. Type species: Decussocarpus vitiensis (Seemann) de Laubenfels. Nageia Gaertner, De Fruct. et Sem. 191. 1788. Type species: Nageia japonica, nomen illeg. (description confused). Folia opposita, decussatim vel spiraliter inserta, lanceolata vel rotun- 1969 | DE LAUBENFELS, PODOCARPACEAE 341 data, ad basim contracta, uni- vel multinervata. Strobili masculi solitarii vel fasciculati. Strobili feminei pedunculati; pedunculi cum squamis (vel foliis); semina saepius singula, globosa, inversa, squama fertilia cum ovulo conjuncta. The new genus Decussocarpus is composed of three sections formerly treated as a part of Podocarpus. A group of characters unite these three sections while distinguishing them from Podocarpus. Ovules are pro- duced subterminally on a scaly shoot not divided into a naked peduncle micropylar end of the inverted seed extends distinctly downward (towards the base of the fertile complex) so that the mature seed appears to be attached at an angle on the end of the fertile shoot. As a result the seed displays a projecting curved beak in contrast with all related taxa. As- sociated with the elongated attachment is the tendency for the fruit to fall with the fertile shoot still attached. In contrast with Podocarpus, a cluster of five or more pollen cones may occur in some species on a single shoot. The leaves of Decussocarpus have a number of distinguishing character- istics. Opposite decussate leaves are found throughout the genus with the exception that in the section AFRocARPUS some branches have spirally placed leaves (herbarium specimens therefore may lack this character which, nevertheless, can be readily found on any mature living specimen of § Arrocarpus). A unique leaf orientation further occurs in all sections of the genus, although it may be absent in a few species of the section Dammaroiners. The distichous leaves being amphistomatic, instead of making unequal twists on opposing sides of the branch to bring the axial surface of the leaf upwards at all times, always turn in the same manner with respect to the axis so that on the left side the abaxial surface is uppermost. This can be seen in section AFROCARPUS even on branches without decussate leaves. Unlike Podocarpus, the leaves of Decusso- carpus have no accessory transfusion tissue and unlike Prumnopitys they do have a hypoderm. Also unlike Prumnopitys, the leaves are not linear, but oval or lanceolate. Most species of Decussocarpus are large trees, some of which are valuable timber trees and others are in demand for decorative planting in the warmer parts of the world. The genus is divided into three sections based on the relative width and venation of the leaves. Section DECUSSOCARPUS has single-veined but relatively wide leaves compared to section AFRO- CARPUS whose leaves are more than ten times as long as they are wide. Section Dammaroues has broad multiveined leaves. Section DEcUSSOCARPUS. Podocarpus section Polypodiopsis Bertrand, Ann. Sci. Nat. V. 20: 65. Type species: Podocarpus vitiensis Seemann [ Decussocarpus vitiensis (See- mann) de Laubenfels]. 342 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Trees with opposite decussate leaves which are ovate or lanceolate, ses- sile, sharply narrowed to a decurrent base, single veined, amphistomatic, and not more than about five times as long as wide; pollen cones sessile, solitary or grouped on a special scaly shoot; seed cones in the form of a scaly or leafy shoot with one or rarely two fertile subterminal bracts; ovule inverted and covered by the seed scale which makes an apical crest over the inverted base of the ovule; seed large, globular, blunt at one end but elongated into a curved beak at the micropylar end (at the base of the fruit) and covered by the fleshy seed scale. An account of this section is given in Wasscher (1941), who shows some uncertainty about how to treat it. Once its characters were com- pletely known, there was agreement that it is most closely related to section Dammaroies (Nageia of most authors). The difference in leaf size and venation, however, made it advisable to separate these two taxa into different sections which have until now been treated as a part of the extensive genus Podocarpus. The new genus, Decussocarpus, is here be- ing proposed to accommodate the two groups and the more recently named section ArrocarPus in order to recognize the considerable morphological differences that previously have been merged in the one genus Podocarpus. Section Decussocarpus extends from eastern Indonesia to South America, and fossil specimens of it from Chile were once the basis of reports of Sequoia in the Southern Hemisphere (Florin, 1940). There are four species. KEY TO THE SPECIES OF SECTION DECUSSOCARPUS 1; Sa heal branches bearing scales. Scales on non-foliage branches appressed, thin; eps leaves with a sharp narrow midrib; mature pollen cones elongated. .............-------: 7 GE Lk RS ea a 31. D. vitiensis. 2. Scales on non-foliage branches spreading, thick; mature pollen cone globu- lar iad elongate e foliage leaves with a raised central band narrower than the ad leaf margins; forest tree. ................ 32. D. comptonii. 3. Mature foliage leaves with a broad raised central band broader than adjacent leaf margins; small tree at water’s edge. .......------ 007° 1. Both primary and secondary branches bearing leaves. .... (D. 17 ospigliosii). 31. Decussocarpus vitiensis (Seemann) de Laubenfels, comb. nov. Podocarpus vitiensis Seemann, Jour. Bot. 1: 33. ¢. JJ. 1863. Type: Seemann , Fiji. Podocarpus filicifolius N. E. Gray a Patt), Jour. Arnold Arb. 43: 74. 1962. Type: Kostermans in 1949, Morot Tree to 43 m. high; bark brown to red brown, weathering to blackish or gray, fibrous, fissured and peeling in short vertical strips; foliage branches opposite or alternate on non-foliage branches and subtended by a short 1-2 cm. scaly base, sometimes with both lateral and terminal 1969 | DE LAUBENFELS, PODOCARPACEAE 343 foliage branches together on the same base, the foliage branch not nor- mally branching again; foliage leaves distichous and equally twisted at the base, lanceolate with a small blunt tip, a narrow but distinct rib marking the vascular bundle on both surfaces, juvenile leaves up to 40 mm. long by 8 mm. wide, adult leaves 15-25 mm. long and 3—5 mm. wide; non-foliage branches with appressed and thin scale leaves which are broadly decurrent and dispersed, 1-2 mm. long; pollen cones single and terminal or grouped with terminal and lateral cones together, either one to three at the apex of a foliage branch or one to three at the apex plus opposite pairs of groups of one to three along a scaly branch, cylindrical, 10-24 mm. long and 1.8—2.2 mm. in diam., microsporophylls triangular, about 1 mm. long; one or two ovules subterminal on a scaly shoot, 6-10 mm. long (which may be terminal or axillary on leafy or scaly branches and solitary or grouped); ovule inverted with the micropyle lying close to the attachment of the seed complex with the micropyle at the end of an elongated beak that may extend more than 2 mm. below the attach- ment, the fertile scale completely enveloping the ovule and forming over the young seed an apical crest which sometimes persists on the mature fruit; mature seed globular, pear shaped, 13-16 mm. long including the curved beak, 8-10 mm. in diam., covered by the deep red fleshy scale and usually accompanied when it falls by the fertile shoot on which some of the scales may still persist. DIstTRIBUTION. Scattered and locally common in a discontinuous se- ries of regions from Morotai to the Fiji Islands in rainforests, from near sea level to 1,800 meters. Map 12 Moluccas. Morotai, Kostermans (1949) j (t8-holotype of Podocarpus filici- folius; (a, K- isotypes). New Guinea. WESTERN HatF: Wissel Lakes, Mt. Barara, Eyma 5155 j (Lt). Wissel Lakes, Motito, Vink & Schram BW 8730 j 1,800 m. (L). Barnhard Camp (Idenburg R.), Brass & Versteegh 12534 s 1,200 m. (A, B), Brass 12787 2 1,200 m. (A, BM, K, L), 12787a j (A, L), 12912 2 (L). Cycloop Mts., Versteegh BW 913 2 1,100 m. (K, L, LAE), Van Royen & Sleumer 6073 s 1,220 m. (kK, z). Papua: Koroba Station, Pullen 2840 2 5,300 ft. (LAE). Alola, Carr 14160 8 6,000 ft. (a, BM, L, NY). Lala R., Carr 15666 6 5,000 ft. (A, BM, L). New Britain. Mt. Tangis, ‘Talasea Dist., Frodin NGF 26292 j 3,500 ft. . LAE), NGF 26917 s 2,400 ft. (L). Benim, Kandarian Dist., Henty & Frodin NGF 27359 8 1,000 ft. (L, Lar). Fullerborn Harbor (Kandarian), Hammermaster & Sayers NGF 21842 & 100 ft. (L). Santa Cruz Is. Vanikoro, Walker BSIP 1580 j (L). Fiji. Vitt Levu: Nandarivatu, Degener 14483 j 750-900 m. (A, K, L, NY, US), 14496 9 (a, Ny), Gillespie 3863 2 (K, NY, US), Gibbs 674 2, 6 (BM, K), Vaughn 3254 j (am, K), Mead 1964 j (x), 1974 é (K), 1982 s (kK). Nausori, Damanu NH15 s (x). Namboutini (Serua), de Laubenfels P309 j 1,000 ft. (a, RSA), Damanu R10 s (k), R15 s (K), R32 s (K), Qoro & Kuruvoli s.n. (k). Serua, Bola 10 s (k), Damanu NLS s (x), NL10 s (x), NL12 s (K), G7 s@, “The Leiden specimen is accompanied by an unattached seed of D. wallichianus, a Napa which was also collected in the area by Kostermans. Podocarpus ogee d on the presumed pairing of the alien seed with the accompanying leaves (de Linibetifeds, 1967). 344 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Naikorokoro, Damanu KU22 s (x). Galva Forest, Damanu 152 2 (kK). VANUA Levu: Mt. Kasi (Thakaundrove), Smith 1796 s 300-430 m. (a, K, US). Without loc.: Seemann 576 j (K-holotype of Podocarpus vitiensis; A, BM-isotypes), Horne 531 s (kK), Tothill 844 s (K), 845 s (kK), Graff 33 s (K). ILLUSTRATIONS, SEEMANN, B., Jour. Bot. 1: ¢. JJ. 1863; Fl. Vitiensis, t. 78. 1868, as Podocarpus vitiensis. While in most ways typical of the genus of which Decussocarpus vi- tiensis is the type species, the dimorphic foliage, shared by two other spe- cies in section Decussocarpus is rather unusual, found elsewhere in the family only in Dacrycarpus and Acmopyle. The compound clustering of specimens of D. vitiensis from other closely related species. It is of inter- est to note that D. vitiensis broadly overlaps D. wallichianus in its dis- tribution; the latter extends throughout New Guinea as well as further west. 32. Decussocarpus comptonii (Buchholz) de Laubenfels, comb. nov. Podocarpus comptonii Buchholz, Bull. Mus. Hist. Nat. Paris. II. 21: 284. 1949. Type: Buchholz 1684, New Caledonia, Mt. Mou. Tree to at least 30 m. high; bark tan to gray-brown, weathering to gray or dark gray, fibrous, becoming very rough and fissured on older trees, breaking off in short vertical strips or rough fragments; foliage branches opposite or alternate on non-foliage branches or one to several at the apex of an older foliage branch, subtended by one or two pairs of spread- ing scales; foliage leaves on young plants distichous and equally twisted at the base, lanceolate with a blunt tip, the midrib marked below by 4 sharp narrow ridge and above by a slight groove, up to 30 mm. long by 6 mm. wide; adult leaves becoming not distichous but still equally turned, coriaceous, the midrib marked by a raised strip narrower than the leaf margins, the edges of the strip when drying appearing as two parallel ridges on both leaf surfaces, ovate-lanceolate, 6-15 mm. long by 2.5-4 mm. wide; non-foliage branches with dispersed spreading scales which are coriaceous, rounded, 1-2 mm. long on young plants and up to 4 mm. long as reduced leaves on fertile specimens; pollen cones single in the axils of foliage leaves, or from one to five or more at the apex of a foliage branch, or in terminal or lateral groups on non-foliage branches (not in compound groups), ovate, 4-6 mm. long (rarely to 12 mm. ) an 2.5—3 mm. in diam., microsporophylls short triangular with large spreading edges to the open spore sacs; seed complex terminal on foliage branches or on lateral scaly branches and involving 2—3 decussate pairs of spread- ing scales or bracts followed by two unequal bracts one of which is fertile, or rarely both are fertile and equal; micropyle of the inverted ovule at the 1969] DE LAUBENFELS, PODOCARPACEAE 345 end of an elongated beak extending about 2 mm. below the spreading fertile bract, the fertile scale completely enveloping the ovule and form- ing an apical crest which sometimes persists on the mature fruit; mature eaten by some bird), the surface of the seed with low scallops and ridges. Distr1BuTION. In rainforests throughout New Caledonia mostly from 750 to 1,450 meters, but also lower where lower rainforests occur. Prob- ably the most common conifer in New Caledonia but always scattered in the forest. ew Caledonia. Ignambi, Compton 1524 s (BM), 1587 & (BM), Foster 160 j (P, z). Mt. Panié, McKee 15594 2 1,000-1,400 m. (p), 15639 j 800 m. (Pp). Mt. Tchingou, Hirlimann 1220 j 1,250 m. (Pp, z). Mt. Paéoua, McKee 17032 j 900—- 1,100 m. (P), 17056 2 (P), Bernardi 10131 s (Pp, z), 10149 s 900 m. (P, z). Mt. Boulinda, McKee 17354 j 1,150-1,300 m. (P), 17357 ¢ (P), Veillon 120 j 1,100 m. (P). Mt. Me Maoya, McKee 13037 s 1,350 m. (P), 13492 j 1,400-1,450 m. (Pp). Ridge W. of Col des Rousettes (Me Maoya), McKee 9886 2 800-900 m. (K, P). Bourail, below Téné, Balansa 1381 2 (xk, Pp). Mt. Nekandi (Thio), McKee 17908 j 1,200 m. (Pp). Dent de St. Vincent, oy 11 j (»). Mt. Hum- boldt, Schlechter 15331 & 1,400 m. (BM, K, P, Z), 15332 2 (P), Buchholz 1578 $ 1,300 m. (ILL), Baumann-Bodenheim 15393 s ee m. eo z), 15411 s (z). Mt. 2), 11200 5 (p, z), 11301 s (P, z), Bernardi 9879 s (P, z), Blanchon 341 s (r). Couvelée, Brousmiche 697 s (p). Mt. Dzumac, de Laubenfels P153 2 (sBT), P415 2 760 m. (A, K, RSA), Baumann-Bodenheim & Guillaumin 12725 j (P, Z), ries 2 800-900 m. (p). Upper Ouinné Valley, Bernier 267 j 750 m. (P), 268 Ae! Baumann-Bodenheim & Guillaumin 12815 s 700 m. ie 12843 s 700 m. 1573 j 530 m. (Pp, z), Bernardi 9445 9 600 m. (Pp, z). Bois de Mois de Mai (Walker’s Place), Bernier 203 2, j (P), 269 j ), ie s (P), Buchholz 1350 s ILL, es hy 1350A j (ILL, aul 1359 j (ILL, z=, &), 7 a ae P), 1367 s — P Bodenheim 15028 s (p, z). Inland from Bay of Pirogues, White 2120 j (a). Without loc. Sarlin 228 j = a. 552 j (P). ILLusTRATION. SARLIN, P., Bois et Foréts de la Nouvelle-Calédonie, t. 26. 1954, as : Restor comptonii. 346 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The juvenile form of Decussocarpus comptonii has a great deal in com- mon with the adult form of D. vitiensis and the ecology of the two is identical. The adult form of D. comptonii is in a number of ways differ- ent from its juvenile form. The fact that D. comptonii is strictly endemic to New Caledonia while D. vitiensis extends for several thousand miles and both to the east and to the west of New Caledonia is a clear illustra- tion of the curiously isolated flora of New Caledonia. There are many closely related species of conifers between the two areas mentioned, but none are common to the two. 33. Decussocarpus minor (Carriére) de Laubenfels, comb. nov. Nageia minor Carriére, Traité Conif. ed. 2. 641. 1867. Type: Vieillard 1275, New Caledonia, Lake Arnaud. Podocarpus minor (Carriére) Parlatore in DC. Prodr. 16(2): 509. 1868. Podocarpus palustris Buchholz, Bull. Mus. Hist. Nat. Paris. II. 21: 284. 1949. Type: Buchholz 1421, New Caledonia, Plaine des Lacs. Small tree or shrub 2-3 m. high; bark tan to dark brown (often stained with iron oxide from flood waters), very rough, fissured, fibrous, slightly scaly, breaking off in short thick vertical strips or ragged frag- ments; foliage branches opposite or alternate on non-foliage branches or single to grouped at the apex of an older foliage branch, subtended by one or two pairs of spreading scales; juvenile leaves distichous, equally twisted at the base, not crowded, lanceolate, the midrib marked by a broad raised area that may appear as three ridges when dry, up to 39 mm. long by 3-4.5 mm. wide, on young plants smaller, more crowded, not distichous but still equally twisted; mature foliage leaves almost imbricate and crowded but still (in part) with a slight equal turning, coriaceous, the midrib marked by a broad raised area, wider than the not raised margins (the raised area upon drying either irregularly wrinkled or appearing as three ridges), ovate, blunt, 7-20 mm. long by 2.5—-5 mm. wide; non- foliage branches with dispersed spreading scales which are thick, rounded, 1.5-2.5 mm. long, smaller on juvenile specimens; pollen cones solitary or clustered up to five or more, terminal and lateral in the axils of spread- ing scales on a deciduous shoot at the apex of foliage branches, ovate, 4-8 mm. long by 2-2.5 mm. in diam., microsporophylls triangular with an elongated point; seed complex terminal on foliage branches and in- volving 2-3 decussate pairs of crowded spreading scales or bracts fol- followed by two unequal bracts one of which is fertile (or rarely both fertile and equal); micropyle of the inverted ovule at the end of a blunt beak which extends up to 2 mm. below the fertile bract, the fertile scale completely enveloping the ovule and forming an asymmetrical apical crest which persists on the mature fruit; mature seed globular pear- shaped, about 20 mm. long including the curved beak and 11-12.5 mm. in diam., covered by the glaucous fleshy scale which sometimes becomes deep red when ripe and after drying tends to crack and flake off the seed (which may persist on the tree for some time or break off anywhere from 1969] DE LAUBENFELS, PODOCARPACEAE 347 the base of the fertile bract to the base of the fertile shoot), surface of the seed rough and porous (making it very buoyant) DistrisuTion. Along lake and river banks in shallow water over soils derived from serpentine, in the headwaters of the Yaté River and along small streams closer to the coast in the southernmost part of New Cale- sa at low elevations (up to 200 meters). ew Caledonia. Pirogues R., White 2261 @ (A, K, P, Us). R. Blanche (Mois de MD, McMillan 5120 s 600 ft. (A, K, P), Baumann- Bodenheim 13923 s (P, Z). Upper Yaté R. (22 km. station), Bernier 204 S,j (), 265 9, 5 @), 257 4, 3 (Pp), Buchholz 1347 s (Pp), 1348 9 (ILL, K, P), 1421 9 (111-holotype of Podo- carpus palustris; K, P-isotypes), 1705 j (ILL, K, P), Sarlin 73 2 (P), de Lauben- fels P112 2, 8, j (sz), P160 2 (ssr), Foster 200 8 (p). Marais Kiki (Yaté R.), McKee 1118 s (A), 1119 2 (a, Pp), Baumann-Bodenheim 6370 j (z), Hiirli- mann 3157 } (z). Creek Pernod, Guillaumin 8339 j (z), 8345 2 (z), Blanchon 1160 8 (Pp). R. des Lacs, bridge, Thorne 28565 2 (GH, Pp), Baumann-Boden- heim & — 6511 5 (z), 6580, j (P, Z), 6766 & (P, z), Hiirlimann 3113 s (z). R. acs (near Madeleine Falls), Bernier 125 j (Pp), 246 s (P), 249 s (P), 250 3 es Le Rat 2587 2 (pe), Buchholz 1474 2 (ILL), 1719 2 (ILL), 1729 2 (ILL, kK, p), Ddniker 228 2 (p, z), 228a 2 (z), de Laubenfels P340 Q (A, RSA), P340A 8 (A, RSA), Baumann-Bodenheim & Guillaumin 11749 s (P, Zz), 11811 s (P z), Stauffer 5807 2 (Pp, z), Blanchon 736 s (P). Plaine des Lacs in general, Le Rat 607 @ (BM, P), 751 2 (P), 1040 % (K, P), 2621 2 (P), McMillan 5139 & 600 ft. ee K, P), Baumann- Bodenheim & Guillaumin 6582 J (Pp, z), 6594 s (Pp, z), Aubré- (ear), McKee 3382 2 (a, Pp, us), Rohrdorf 178 2 (z), Bernardi 9369 4 (P, Z). bi nd ao ‘ee 658 s (A, ?). Plaine des Lacs, lake bank, Franc 207 s (a, “n K, P, z). L. Arnaud, Vieillard 1275 4 (p- holotype of Nageia minor). La Chate, Pa 208 s (P). Prony Bay (B. du Sud), Vieillard 1275 (apparently not the same as the previous collection of the same number) ¢ (A, BM, GH, K, See Sartin, P., Bois et Foréts de la Nouvelle-Calédonie, t. 27. 1954, as ps ened palustris. The peculiar habit of this species which grows in shallow water with a swollen base to the trunk (somewhat like a bald cypress), immediately sets it apart. In general morphology it strongly resembles Decussocarpus comptonii, but close inspection reveals a different leaf morphology and slight differences in the pollen cone and the seed. This general similarity Prevented the recognition of D. comptonii as a species for many years. ADDITIONAL SPECIES: Decussocarpus rospigliosii (Pilger) de Laubenfels, comb. nov. Podocar pus rospigliosii Pilger, Notizbl. Bot. Gard. Berlin 8: 273. 1923. Type: Esposto 556, Peru, Oxapampa (not seen, A photo. 348 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Section Dammaroides (Bennett) de Laubenfels, comb. nov. Podocarpus section Dammaroideae Bennett ex Horsfield, Pl. Jav. Rar 1838. Type species: Podocarpus latifolia Wallich [Decussocarpus sie ianus (Presl) de Laubenfels Podocarpus section Nageia Endlicher, Syn. Conif. 207. 1847. Type species: Podocarpus nageia R. Br. [Decussocarpus nagi (Thunb.) de Laubenfels]. Leaves opposite decussate or subopposite, multiveined with the veins converging to the apex, obovate to elliptic, acute, distichous, rather large; terminal buds small, bud scales acute; pollen cones linear, solitary or grouped in the axils of leaves; seed cone on a scaly shoot with one or two subterminal bracts fertile; ovule inverted and enveloped by the fleshy fertile scale; seed globular with a slight protrusion at the micropylar end close to the spreading fertile bract but on the opposite side from the fruit attachment. The multiveined leaves immediately distinguish section DAMMAROIDES from the remainder of the genus but, without fruit, trees of this group are very similar to the genus Agathis of the Araucariaceae. These are distinguished by their globular terminal buds quite different from the acute scales of buds in the section DammaromeEs. It has been cus- tomary for some time to call this section Nageia, ignoring Bennett’s name apparently because he used an improper ending. Bennett described his section quite adequately and both Pilger (1903) and Wasscher (1941) refer to his name without comment. Gordon (1858) and others treated this section as a genus, Nageia Gaertner, a name which if accepted would have priority not only for the genus being proposed here, but also in the genus Podocarpus (if the section were to be retained in that genus). In- deed, Kuntze (1891) did substitute Nageia for Podocarpus but Pilger (1903) pointed out that the original description of Nageia confused its characters with those of Myrica (“stam. quattuor et styl. duo.’”) and, therefore, the name must be rejected. There are five species differing in the presence or absence of a fleshy receptacle, the number and position of the pollen cones, and the orientation and size of the leaves. KEy TO THE SPECIES OF SECTION DAMMAROIDES 1. Seed with a fleshy receptacle. 2. Pollen cones grouped on a peduncle; leaves at least 6 ag ie gos Saas 4 Dd. 0 llichianus. 1 EE Ct BES FE at te Sa U7 isa mn eg ea 35. D. motleyi. Seed lacking a fleshy receptacle. 3. Leaves amphistomatic and equally turned, 20-31 cm. long. .....-----°° 36. D. maximus. 3. Leaves hypostomatic and unequally turned, less than 20 cm. long. 4. Pollen cone cluster sessile; leaves at least CM TON eos Pe a9. fleuryi. 1969] DE LAUBENFELS, PODOCARPACEAE 349 4. Pollen cone cluster on a peduncle; leaves usually less than 6 cm. long er Re ne Ga cis ee ee 38. D. nagi. 34. Decussocarpus wallichianus (Presl) de Laubenfels, comb nov. Podocarpus latifolia Blume, Enum. PI. Javae 1: 89. 1827, nomen illegit., non unb. 3 e: Blume s.n., Java, Mt. Salak. Podocarpus latifolia Wallich, Pl. As. Rar. 26. 1830, nomen illegit. Type: Wallich 6050, India, Mt. Sillet. Podocarpus wallichianus Presl, Bot. Bemerk. 110. 1844, based on Podocarpus latifolia Wall., which is a later homonym. Podocarpus blumei Endlicher, Syn. Conif. 208. 1847, based on Podocarpus latifolia Blume. Podocarpus agathifolia Blume, Rumphia 3: 217. 1849, based on Podocarpus latifolia Blume, Nageia blumei (Endl.) Gordon, Pinetum 135. 1858. Nageia latifolia (Wall.) Gordon, ibid. 138. Nageia wallichiana (Presl) Kuntze, Rev. Gen. Pl. 800. 1891. Podocarpus latifolia Blume forma ternatis De Boer, Conif. Archip. Ind. 14. 1866. Type: Teysmann s.n., Moluccas, Ternate. Tree up to 48 m. high; bark smooth, peeling in large, thin, very irregular plates, tan to brown within, weathering to dark brown or gray and develop- ing scattered large lenticels and irregular longitudinal markings; leaves decussate, distichous, amphistomatic, equally turned so that the lower surface is exposed on the left and the upper surface is exposed on the right side of the branch, many parallel vascular veins, elliptic, acute to acuminate, sometimes abruptly narrowed to the short (5-10 mm.) petiole, mostly 9-14 cm. long by 3—5 cm. wide but sometimes smaller to 6 cm. long and 2 cm. wide or (particularly for juvenile and shade leaves) up to 23 cm. long and 6.8 cm. wide, the extreme sizes (both smaller and larger) mixed with more usual sizes on the same tree (both extreme width and extreme length not usually together, the ratio of length to width varying from 2 to 6, so that the narrowest leaves are not usually the shortest while the longest are not usually the widest); terminal buds often 2-3 mm. beyond the last pair of leaf bases (lateral buds sessile), abruptly but slightly wider than the stem and then tapering, bud scales acute-acuminate and erect; pollen cones 1—7 on an axillary scaly peduncle 2-10 mm. long with one cone terminal and the remainder in decussate pairs, cylindrical, 8-18 mm. long by 3-4 mm. in diam., microsporophylls lanceolate, 2-3 mm, long; seed cone on an axillary scaly peduncle, 8-20 mm. or more long, the lanceolate scales deciduous as on the pollen cone peduncle; receptacle enlarged and becoming blackish and very fleshy upon maturing, 7-18 mm. long, composed of 4 to 7 sterile bracts, the curled ends of which protrude from the receptacle, and 1 or 2 subterminal fertile bracts with inverted ovules; seed smooth, globular with a small beak next to the point of attachment, completely covered by the thin fertile scale which accompanies the ripe seed, 15-18 mm. in diam. DistriputTion. In rainforests from eastern India to Normanby Island 350 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 east of New Guinea, often as a common forest element at low elevation and extending as high as 1,575 meters (one collection at 2,100 meters). Map 13. India. Mt. Sillet, Wallich 6050 4 (A-isotype of Podocarpus latifolia Wallich). E. o Griffith 3005 6 (GH, P). Khasia, Hooker & Thompson s.n. s 3,000 ft. (a, Apan 191 2, & (GH). Assam, King sn. & (a, L, Us). Thailand. Temscein Falconer s.n.s (Lt), Kao Lu as sa Stritamurat, Kerr 15445 s 600 K). Tamtieng, Ranaung, Kerr 11770 2 200 m. (kK). Kapor, Ranaung, Kerr pe 2 50 m. (K). Kumpuam, Ranaung, — {O86 2 50 m. (K). Kuabun, Ranaung, Kerr 16351 s 50 m. (K). Lem Dan, Kaw Chang, Rabil 19 s (K). Klaung Non Si, Kaw Chang, Kerr 9324 s 10 m. (K). Kao Kuap, Kuat, Kerr 17714 s 500 m. (K). Kao Ri Yai, Kanburi, Kerr 10400 s 1,400-1,500 m. (kK). Baw Rai Kinat, Kerr 9509 s 300 m. (x). Adang, Sulut, Kerr 14132 s 500 m (K). Khas Yai, 105 km. E. of Saraburi, King 555A 2 780 m. (L). Laos. Pak Mu- nung, Wengchau, Kerr 21215 & 1,400 m. (x, Pp). Cambodia. Phnom San Kas, Miiller 499 s (p). Elephant Mts., Poilane 23216 s 200 m. (Pp). Kre-dek, Poilane oe s 600 m. (Pp). Cochin China. Phu Quoc I., Pierre 5529 s (A, NY, P), 5530 9 (A, P). Annam. Vinh, Linhcam, Chevalier 38234 s (Pp); Ke Bon, ‘Chevalier 38127 s (P). Quang-nam (S. of a Nang), Poilane 31558 s 500 m. (p). Massif Ngok Guga near Dakto, Poilane 35675 s 1,000 m. (p). Massif du Brian near Djiring, Poilane 24234 s (Pp), ye s 1,500 m. (Pp). Phan Rang (Can-Wa), Poilane 5963 2 900 m. (a, P). W. of Ca Na, Evrard 2422 s 1,200 m. (A, NY, P): Malaya. Jerai Reserve, Kedah, (Mal.) 17848 s (K). Kledang Saiang Reserve, Perak, Mead 12828 2 (x), Noakes 20133 9 (k), 22147 s (x). Kinta, Perak, Low 64 @ (k). = Scortechini sn. j (A), Kebal Ayam, Kurantan, Pahang, Loh 15065 s (x). Soga, Johore, Ridley 11223 s (Kk). Labis, Sinclair 38991 s (L). Johore, G. ra Sinclair 10578 j (K, L, NY). Sumatra. Mt. Sibajak, E. Kisaran, Peete 238 j (A, ¥, US). E. Coast, Asahan Yates 2554 2 (A, NY). ogg? + 4A, 2) pao 8108 s (us). Tapanuli, Sibolga, NJFS 6b1357 j m. oh pea Angkola, NIFS bb31536 s 600 m. (L). Benkulen, Redjang, Sots bbE1084 s 800 m. (1), Renwarin bb2450 2 (L), NIFS bb8842 s 900 m. (L). Palembang, Paaioann, NIFS bbE1106 s 15 m. (1). Palembang, Pasemah land, NIFS TB200 s 1,200 m. (1). Mig srg: G. Pakiwang (Ranau L.), Van Steenis 3754 s 1575 m. (L), Java. G. Lajung, Koorders 1261 s 150-250 m. (L). G, Salak (Batavia), Anon. (1845) s (L), resis 24181 s 1,000 m. (x). Pre- anger, Parakan Salak, Koorders 39402 s (K, L), 39403 s 1,000 m. (A), 39404 s (L), 39406 s 1,100 m. (A), 39407 s 1,350 m. (a), 39409 s 1,000 m. (£1), 39413 s 1,000 m. (L), 39415 s (a). G. Sys gears Junghuhn s.n.s (L). G. Panga- ee Junghuhn sn, 2 3,000 ft. (L). G. Gedeh, Anon. 204 s (LZ), hate 346 $ (1). Takokak, Koorders 1262 s (t), 1264 s (L), 11909 s (1), 39592 s 1,200 m. (A), 39596 s (a, L). Preanger, Tjipatudja, Backer 8866 s 450 m. (L). W With- out loc., Blume sn. 2 (t-lectotype, Podocarpus latifolia Blume), 5.7. § (u), Hasskarl sm. j (L), Junghuhn sn. 5 (L), Miquel sn. 8 (1). Karimata. Sung Tajan, Teysmann 11598 g (1). Billiton. Riedel (1876) @ (#1). Sarawak. Lundu, G. Gading, Anderson 15391 2 2,600 ft. (k, L). B. Mersing, in from Tatau, Luang $2217 6 2 900 m. (x). Bintulu, Merurong Plateau, Brunig S9999 s 820 m. (L), Lawas, Brunig S$12083 s 1,000 m. (x). Mt. Majan, Clemens 21822 } faa North Borneo. Penampang, Leaio- Castro 5989 s, 5,000 ft. (K, L). 1969 | DE LAUBENFELS, PODOCARPACEAE sot Kinabalu, Clemens s.n. j 7,000 ft. (Ny), Chew & Corner RSNB 4878 2 5,000 ft. (kK). Elopura, Sandakan, Keith A7 s 10 ft. (K). Gompa, Kudat, Balajadia 4055 s sea level (K). Tawau, Ampon A1652 2 (K, us), Martyn SAN 18453 2 50 ft. (K, L, Ny), Meijer SAN 19547 s 30 ft. (L). Without loc. Wood 1244 i (A, ©), sm. 8 (US), Burbridge s.n. j (K). Borneo. Tidung’s Land (SE. Borneo), NJFS bb18217 s 4m. (a, L). G. Beratus (Balikpapan), Kostermans 7464 s 900 m. (A, L), 7486 2 900-950 m. (A, K, L). Samarinda, Kostermans (1956) 8 low (tL). Mahakam, Amdjah 51 j (L). Mt. Palimasen (Cent. Kutei), Kostermans (1954) 3 900 m. (K, L). Philippines. Cagayan, Curran 16738 2 (us), 17200 2 (Ny, US). Mt. Sulu (Apayao Subprov.), Fenix 28348 2 (A, Ny, us). Mt. Moises (Isabella), Ramos & Edano 46333 2 (A, NY). Baguio, Williams 1035 s (K, Ny, Us). Lamao R., Mt. Mariveles (Bataan), Williams 399 s 2,200 ft. (Ny, US), 624 2 3,000 ft. (NY), 752 2 2,000 ft. (wy), 753 2 1,800 ft. (A, K, NY, US), Barnes 147 @ (kx, NY, US), 194 2 (xk, Ny, US), Copeland 244 2 (K, Ny, us), Whitford 1353 2 (K, NY, us), Curran 17616 s (L). Mt. Giting-Giting, Sibuyan I., Elmer 12360 @ (A, K, L, NY, us). Celebes. Usu (Malili), NIFS Cel/III-80 s (A, K, L), Cel/III-143 s 10 m. (A, L), Cel/III-146 s 25 m. (kK). Tebetano (Malili), NJFS bb24489 j 450 m. (a, L). Tawingana (Malili), N/FS bb25541 s 120 m. (Lt). Tamborano (Malili), NJFS bb9696 s 600 m. (L). Gorontalo Buladu (Manado), N/FS bb15602 s 400 m. (Lt). Poso, Majoa (Manado), NJFS 6b31500 s 700 m. (tL). Manado, Kolonodale, Bakomtefe, N/FS bb31506 s 100 m. (a, L). Lapo Lapo, SE. of Kandari, Bec- cart (1874) s (FI). Moluccas. Obi, Ahasrip 118 s (K, L, NY). Ternate, Teys- mann s.n, } (L-holotype of forma ternatis). Morotai, Kostermans 1660 s 5 m. (A, L). W. Ceram, NJFS bb19647 s (L). New Guinea. VoGELKoP: Warsam- son Valley (E. of Sorong), Moll BW 11652 s 30 m. (L). Kebar Valley, Tobie, Schram BW 7951 s 740 m. (L), Smit BW 2314 s 650 m. (L), Sijde BW 5579 s 750 m. (L). Nertoi, Kebar Valley, Mangold BW 2350 2 750 m. (t). Kebar Valley, Van Royen 5058 2 550-700 m. (kK, L). Sidai (W. of Manokwari), Schram BW 1785 s (L), BW 6174s 5 m. (L), Koster BW 6705 s (L), BW 6760 2 10 m. (K, 1), BW 6922 s 10 m. (Lt), BW 6977 s 5-10 m. (L). Manokwari, Menusefer BW 8180 2 140 m. (z). Baru (Teminabuan), Hallewas BW 944 S 10m. (L). Beriat (S. of Teminabuan), Schram BW 6021 2 10m. (L). SERUI Biak, NIFS 6b30717 s 50 m. (A, L), 6b30779 s 80 m. (A, L), 6b30813 @ 50 m. (A, L), bb30887 s 50 m. (L), Moll BW 9574 s 35 m. (L), BW 9589 s 35 m. (L). Mios Num. I., NJFS 6630939 s 200 m. (A, L), 6630947 s 200 m. (A, L), 6630961 S 200 m. (A, L). Japen, Sumberaba, Koster BW 11121 s 8 m. (L). Japen, Aisao, Schram BW 10596 s 200 m. (L). Japen, NIFS 6630260 s 300 m. (L). West: Armina (Babo), Moll BW 12968 s (Lt). Agondo (Babo), Lundguist bb32983 (264) s 20 m. (K, L). Esania, Borowai (Fak Fak), Stefels BW 3147 s 20 m. (L). Tiwara, Arguni Bay, Telussa BW 5158 s (L). Tairi, Kaimana, Lou- 500 ft. (k, x). Gabensis (Lae), NGF W41 j (a). Papua: Fly R., D’Albertis (1877) 5 (Ft). L. Daviumbu (Middle Fly), Brass 7909 s (A, L). Sibium Range, Pullen 5932 s 2,300 ft. (t). Oriomo R., Hart 5005 @ (A, K, L), Brass 5878 s (A, K, L, ny, us), 5880 j (A, Ny, US), 5906 s 5-10 m. (a, Ny), White & Gray 352 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 (Cent. Div.), Brass 3962 j 500 m. (A, Ny). Milne Bay Dist., Womersley NGF 19298 2 1,200 ft. (K, L). Normanby I., Brass 25824 j 600 m. (a, L). ILLUSTRATIONS. BLUME, C. L., Rumphia 3: ¢. 173. 1849 (as Podocarpus agathifolia): Prrcer, R., Pflanzenreich IV. 5 (Heft 18): fig. 9A (as Podo- carpus wallichianus) & fig. 9B (as Podocarpus blumei). 1903; Nat. Pflanzenfam. ed. 2. 13: fig. 1344 (as Podocarpus wallichianus) & fig. 134B (as Podocarpus blumei). 1926; Koorpvers, S. H., & TH. VALETON, Atlas der Baumarten von Java 3: t. 588. 1915 (as Podocarpus blumet). Podocarpus wallichianus and P. blumei have generally been treated as distinct species but are here being united under the name Decussocarpus wallichianus, Wasscher (1941), who treated Podocarpus blumei, analyzed the two taxa and concluded they were probably the same. The differences reported, as thickness of leaf and acuminate leaf tip, are variations that may occur even on a single plant depending on age and exposure. Further, considerable variation in leaf dimensions are frequently found on single herbarium sheets. By comparing the predominant sizes within any re- gion, it appears that this species is, in fact, rather consistent throughout its considerable range. 35. Decussocarpus motleyi (Parlatore) de Laubenfels, comb. nov. Dammara motleyi Parlatore, Enum. Sem. Hort. Bot. Mus. Florent. 26. 1862. Type: Motley 1300, Borneo, Bandjarmasin. Podocarpus beccarii Parlatore in DC. Prodr. 16(2): 508. 1868. Type: Bec- cari 2649, Sarawak, near Kuching. Nageia beccarii (Parlatore) Gordon, Pinetum 2: 186. 1875. Agathis motleyi (Parlatore) Warburg, Monsunia 1: 185. 1900. Podocarpus motleyi (Parlatore) Diimmer, Jour. Bot. 52: 240. 1914. Tree up to 40 m. high; leaves opposite or subopposite, amphistomatic, distichous, coriaceous, elliptical, acute or with a small acuminate tip, nat- rowed at the base to a short thick petiole 2-3 mm. long, somewhat variable in shape, usually 3-5 cm. long and 15-22 mm. wide but reaching 7.5 cm. long and 28 mm. wide; terminal buds compact, tapering at first, the scales rounded to lanceolate, acute to acuminate, 3-5 mm. long; pol- len cones axillary, solitary, sessile, cylindrical, 15-20 mm. long when mature and 5—6 mm. in diameter; microsporophylls lanceolate, keeled, 2 mm. long; seed cone on an axillary scaly peduncle 2-5 mm. long with 3-4 pairs of decussate deciduous scales, receptacle becoming fleshy (on some specimens with nearly fully developed seed there is no enlargement) , 8-12 mm. long, composed of 5 to 9 spreading sterile bracts, the single subterminal fertile bract with an inverted ovule covered by the fertile scale; seed smooth, globular, with a small beak at the micropylar end near the point of attachment, 13-16 mm. in diam. 1969] DE LAUBENFELS, PODOCARPACEAE 353 DistrIBuTION. Mostly in low poorly drained areas but extending to 500 meters elevation in enna from Sumatra and Malaya to the southern part of Borneo. Map. Malaya. G. Tebu (Trengganu), Sinclair & Salleh SFN 40798 s (z). Lumut Dindings (Perak), Hadden 16554 2 (xk), Symington 27841 s (K). Legari Melin- tang, Dindings, Strugnell 16568 s (K). Johore, S. Kayu, Mawai-Temulang Road, Corner 21341 s (kK). Sumatra. Barus (Sibolga), Tapanuli, NIFS 6bb29532 s 25 m. (A, L), bb31596 s 1 m. (A, L). Near Banjunglintjir, Palembang, N/JFS m. (L). Arch., Karimon, NJFS bb17229 s 150 m. (A, L). Belimbing, NJFS 5628495 S 6m. (A, L). Sarawak. Near Kuching, Beccari 2649 2 (r1-holotype of Podo- carpus beccarii; K-isotype). G. Perigi (Lundu), Anderson 13304 & 1,000 ft. (A, K). Simunjan, Drahman $0316 s sea level (L). Without loc., Foxworthy 353 2 (kK). Borneo (SE. part). Tidung’s Land, S. Lebakis, NIFS 6b18328 SS i, (A, 2). ale noe Motley 1300 s (k-isotype of Dammara motleyi). Puruk Tjahu, Tahudjan, NJFS 6621151 2 500 m. (A, L). ILLUSTRATION. WaSSCHER, J., Blumea 4: ¢. 4, fig. 11. 1941, as Podo- carpus motleyi. The extremes of leaf sizes in Decussocarpus motleyi and D. wallichianus approach each other and the latter occurs throughout the range of the former so that the possibility of confusion exists. The pollen cones are diagnostic and the terminal buds lying directly above the last leaf attach- ment help to differentiate the two. By observing the most common leaf Sizes of a tree (or even a specimen), however, the species can be readily Separated. The lack of enlargement of the receptacle on some specimens (see NIJFS 12T1P185) merits further observation. 36. Decussocarpus maximus de Laubenfels, sp. nov. Arbor parva, 4.5 m. alta; folia magna, decussata, coriacea, elliptica, acuminata, basi magis rotundata, in petiolum perbrevem angustata, 20— 34 cm. longa, 6-9 cm. crassa; gemmae parvae, acuminatae; strobili fem- inei ramunculum axillarem 12 mm. longum formantes; squamae ra- musculi decussatae, 3-4 mm. longae, deciduae; receptaculum nullum; semen globosum, diametro ca. 16-18 mm. Holotypus: Anderson 3361/7 (L), Sarawak, Sibu. Fic. 10. DistriBUTION. In low elevation swamp forest of Sarawak and possibly Sumatra Sarawak. Sg. Assan, Naman F. R. (Sibu), Anderson 3361/7 @ 12 ft. (1- holotype ; K-isotype)./) Sumatra. Silo Maradja (Ashan), Bartlett 6805 s (K, L NY, Us). Between Ey Tombak and Taratak, Tanah Pe (ancien L). Decussocarpus maximus is unique in the genus in combining a lack of a fleshy receptacle with amphistomatic leaves, although, as mentioned above, some specimens of D. motleyi with very small leaves may also JOURNAL OF THE ARNOLD ARBORETUM [ VOL. wees iceml eat] Malis Sg eid HORA AW, teh yew hes Hee oF Depart, Sanaa at may? Dts ' “etoearrae + Sandie, pan se ond ; Lovalite Ste — Not Qegl tow, 26% ie peat Det Godelier Coll wt ier Deertuted be A. R,. Anderson 3361/7 (1) 50 egies: 10. Decussocarpus maximus de Laubenfels, photograph of the holo- type 1969 | DE LAUBENFELS, PODOCARPACEAE 355 combine these characters. The leaves of D. maximus are without ques- tion the largest of any conifer, being approached at their lower limits by juvenile leaves of D. wallichianus and of various species of Agathis. The largest leaves described above belong to a fertile specimen. Aside from leaf size, the leaf form and terminal buds correspond to D. wallich- ianus but the seed is produced on a shoot that is not fleshy, The range of D. maximus is included within the range of closely related species and the Sumatra specimens, being sterile, are distinguished by their leaf size only (Bartlett 6805: 15-20 by 6.5-9 cm. and blunt, possibly from damage; Bartlett 8226 and Teysmann (Bangka): 20-22 by 7-8 cm.). In having neither hypostomatic leaves nor a fleshy receptacle, D. maximus (and perhaps a part of D. motleyi) exemplifies best, among the species in section Dammaroides, the special characteristics of the genus. 37. Decussocarpus fleuryi (Hickel) de Laubenfels, comb. nov. Podocarpus fleuryi Hickel, Bull. Soc. Dendrol. France 75: 75. 1930. Lecto- type: Fleury 38017, Tonkin, Phu Tho.® Tree to at least 10 m. high; leaves opposite, decussate, coriaceous, hypostomatic with unequal turning so that the upper surface of the leaf is always uppermost, elliptic, acuminate, narrowed at the base more or less to a petiole, 8-18 cm. long and 3.5—5 cm. wide; terminal bud large, somewhat beyond the nearest leaves, tapering sharply, but scales lan- ceolate, acute, erect; pollen cones grouped in an axillary sessile cluster of three and subtended by several pairs of overlapping, broad, keeled scales, long cylindrical, about 3.5 cm. long and 4 mm. in diam.; micro- sporophylls small, triangular, acute; seed cone on an axillary scaly not en- larged peduncle, 15—20 mm. long; ovule inverted in the axil of a subter- minal bract; seed globular, 15-18 mm. in diam. Fic. 11. DistripuTION. In mountain forests from northern Annam to Kwang- tung. Map 14, Kwangtung. Naam Kwan Shan (Tseng Shing Dist.), Tsang 20123 ¢ (A, NA, NY, US), 25273 @ (a), Tonkin. Phu Tho, Chevalier 38017 2 (v-lectotype). Hoa Binh, Ste. forestier 8408 s (p-syntype). Phu Huo, Chevalier 37512 s (P). Annam. Vinh, Nghe An, Fleury 30180 s (p-syntype). Near Hue, Poilane 29808 2 1,300-1,400 m. (ILL, Pp). Mt. Bana (25 km. from Tourane), Clemens 4190 s (A, K, NY, P), The much larger leaves distinguish sterile specimens of Decussocar pus fleuryi from D. nagi, although both have hypostomatic leaves. Otherwise, the female peduncle of D. fleuryi is longer, the pollen cones are longer, and, particularly, the pollen cone cluster is sessile. The general form of the leaf corresponds with that of D. wallichianus and even Hickel, in the original description, included a specimen of the latter (Poilane 5963) among the specimens he listed. However, the stomatic condition and ® Chevalier is given as the collector for this specimen on the sheet in Paris. 356 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 FLORA OF RWANGT UNG Barta rts Petccobu ‘Waeicbioon> Jtae} ft. Bligh: fre needy Sommon; thioxet. . Ble Ne n Kwan Shan Hi 1 it f Viliag 4 H Pies 11. Decussocarpus fleuryi (Hickel) de Laubenfels, photograph of sang 252 (A). 1969] DE LAUBENFELS, PODOCARPACEAE 557 orientation of the leaves at once distinguish even sterile specimens. The sessile pollen cluster and the lack of a fleshy receptacle also separate these two species. 38. Decussocarpus nagi (Thunberg) de Laubenfels, comb. nov. Myrica nagi Thunb. Fl. Japon. 76. 1784. Type: ex herb Thunb., microfiche no. 23381 (AGH). Nageia japonica Gaertner, De Fruct. et Sem. 1: 191. 1788 (in part, nomen illeg., description confused). Podocarpus nag ela R. Br. ex Mirbel, Mem. Mus. Paris 13: 75. 1825 (based japonica). nec. ees Endl. Syn. Conif. 207. 1847. Hort. Nageia cuspidata (Endl.) Gordon, Pinetum 136. 1858. Nageia ovata Gordon, Pinetum, Suppl. 42. 1862. Type: Fortune in 1861, Japan, Yeddo (not seen). Podocarpus nageia R. Br. var. rotundifolia Maxim. Gartenflora 13: 37. 1864 (based on Nageia ovata Gordon Podocarpus nageia R. Br. var. angustifolia Maxim. ibid. Hor Podocarpus ovata (Gordon) Henk. & es et Nadelh. ae 1865. Dammara veitchii Henk. & Hoch. ibi H es japonica (Gaertner) Nelson, ge ee 1866, nomen illeg., non Sie Podo saree caesius Maxim. Meél. Biol. 7: 561. 1870. Hort. Nageia nagi (Thunb.) Kuntze, Rev. Gen. Pl. 798. 1891. Podocarpus nagi (Thunb.) Makino, Bot. Mag. Tokyo 17: 113 . P oo nagi (Thunb.) Makino var. rotundifolia (Maxim.) ee ibid. P ais nagi (Thunb.) Makino var. angustifolia (Maxim.) Makino, ibid. Podocarpus formosensis Diimmer, Gard. Chron, III. 52: 295. 1912. Type: Schmiiser 1357, Formosa, S. Cape (not seen, photo of type accompanies descript. Podocarpus “nankoensis aig, Ic. Pl. Formos. 7: 39. 1918. Type: Hayata in 1916, Formosa, Podocarpus nagi (Thunb.) oo var. koshunensis Kanehira, Trans. Nat. Hist. Soc. Formosa 21: 145. 1931. Syntypes: Mori 25075, Sasaki 25076, 25077, Kanehira 26078, 22239, Matsuda 2594 (not seen). ? Podocarpus koshunensis (Kanehira) Kanehira, Formosan Trees. Rev. ed. 36. 1936. Tree to 25 m. high; bark smooth, peeling in thin flakes, dark brown weathering gray; leaves decussate, distichous, hypostomatic, multiveined, elliptic, acuminate, to rounded at the tip, the apex often showing evidence of aborted growth, sometimes abruptly narrowed to a short wide petiole, often glaucous especially on the underside, 4.5—5 cm. long or sometimes longer, 10-20 mm. wide, somewhat variable in size and shape even on individual specimens; terminal bud often 1-2 mm. beyond the last pair of leaves, abruptly wider than the stem and then tapering to an acuminate apex, bud scales long lanceolate; pollen cones 1-5 on an axillary scaly peduncle 3-10 mm. long, subtended by a lanceolate scale up to 6 mm. long, cylindrical, 10-20 mm. long, the longer ones terminal in the cluster, 358 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 microsporophylls small, acuminate, widely spreading and not crowded, 1 mm. long; seed cone on an axillary peduncle with deciduous lan- ceolate scales, peduncle 5—10 mm. long, not enlarged; one or two seeds developing from inverted ovules (rarely there are 3 ovules) in the axils of subterminal bracts, completely covered by the seed scale, globular and elongated into a hooked beak at the micropylar end, smooth, glaucous, the seed itself 12-13 mm. wide and 15-16 mm. long, the fleshy bluish- black covering at least 2 mm. thick but drying on the seed and wrinkling, the seed often falling with the peduncle attached. DIsTRIBUTION, Scattered from southeastern China and Hainan to southern Japan in forests at low elevation, up to 800 meters in more southerly parts. Because of the high degree of disturbance of forests al- most throughout its range and its popularity in cultivation, it is most difficult to distinguish between plants naturalized from cultivated sources and truly native individuals. Probably some of the specimens cited are, in fact, cultivated even where not indicated as such. Map 15. China. Hatnan: Pak Shik Ling (Cheng Mai Dist.), Lei 745 2 (A, L, NY, US). Pak Shek Shan (Lam Ko & Cheng Mai Dists.), Tsang 681 (L.U. 17430) s (A, L, NY, Us). Nodoa, Sha Po Ling, McClure 8131 2 800 m. (A). Manning, How 73876 s 700 ft. (a, P). Kwancost: Ta Tse Tsuen, Yung Hsien, Steward & Cheo 728 2 380 m. (A, NY, P). Sup-man-ta Shan, Liang 69401 s (A). Wah Kong (Hing On Dist.), Chung (Tsoong) 83667 s (A), Kiancst: Tung Lei (Kiennan Dist.), Lau 3964 2 (a, us). FuKiEN: Hinghwa, Chung 924 @ (A). Yenping, Chung 2979 9 (A), 3570 s (A), Dunn 3523 Q (A). CHEKIANG: Ping- yang Hsien, Ho 1554 @ (a). Tsingtien, Keng 99 @ (a), 20-40 miles W. of Wenchow, Ching 1832 2 250-450 m. (a, Pp, us). Yentang Shan, Chiao 14685 s (A, NY, Us, z), Hu 241 2 (a). Without loc., Chen 4091 2 (a). Formosa. SOUTHERN: Koshun (=Hengchun), Kanehira 28 s (a), 29 s (A). E. Coast: Hualien, Kuntz 084 4 (us). Nanwo (Karenko Prov.), Wilson 11109 s (A, US). = sj ze 5 aN = = = a @ by S i~ wn iy 5S i © Nm mn ~~, fe | mn —_—Z —G S = S => i=] = & SS 8 Qa Q R @ aS = a 3 Ss nN Do © = mn & Kasugidani), Moran 5351 s 350 m. (us). ILLUSTRATIONS. PitcER, R., Pflanzenreich IV. 5 (Heft 18): fig. 9C-E. 1903; Nat. Pflanzenfam. ed. 2. 13: fig. 134C-E. 1926, as Podocarpus nagi; Dimmer, R., Gard. Chron. III. 52: ¢. 132. 1912, as Podocarpus formosensis; KANEHIRA, K., Formosan Trees, rev. ed. t. 4. 1936, as Podo- carpus koshunensis. 1969 | DE LAUBENFELS, PODOCARPACEAE 359 The large number of specific names which have been applied to De- cussocarpus nagi, including several horticultural names, result from long aquaintance with it in cultivation, and from its wide distribution. Most of the differences noted for the various units proposed are within the normal variation of a population or even of an individual. The variety rotundifolia (equals Nageia ovata) of Podocarpus nagi possibly is defen- sible as a distinct taxon, having leaves ovate, compared to the usual elongated outline. Section Afrocarpus (Buchholz & Gray) de Laubenfels, comb. nov. Podocarpus section Afrocarpus Buchholz & Gray, Jour. Arnold Arb. 2 Type species: Podocarpus falcatus (Thunb.) R. Br. stearate falcatus (Thunb.) de Laubenfels ]. Decussocarpus falcatus (Thunb.) de Laubenfels, comb. nov. Taxus falcata Thunb. Prodr. Pl. paige 117. 1800. Type: Thunberg in 1773-1774, Cape of Good Hope (not s ae Jateatu (Thunb.) R. Br. ex Mitb, Mém. Mus. Hist. Nat. Paris aaa elo Stapf, Fl. Trop. sue “= Prain] 6(2): 343. 1917. Type: Nelson 423, Transvaal, Houtschber Decussocarpus gracilior (Pilger) de Laubenfels, comb. nov. Podocarpus gracilior Pilger, sama IV. 5 (Heft 18): 71. 1903. Type: Schimper 1160, Ethiopia, Che Decussocarpus mannii (Hook.) de Laubenfels, comb. nov. Podocarpus mannii Hook. Jour. Linn. Soc. 7: 218. 1864. Type: Mann 1065, St. Thomas Island. Nageia mannii (Hook.) Kuntze, Rev. Gen. Pl. 800. 189 Podocarpus usambarensis Pilger, ae la 5 "Glett 18): 70. 1903. Lectotype: Holst 2467, Tanganyika, Usa Podocarpus dawei Stapf, Fl. Trop. Afr. red. one 6(2): 342. 1917. Type: Dawe 961, Uganda. LITERATURE CITED BucHuotz, J. T. Embryogeny of the Podocarpaceae. Bot. Gaz. 103: 1-37. 1941, Dr orig otis D. J. Parasitic conifer found in New Caledonia. Science 130: - 1959. ————.. The primitiveness of polycotyledony considered with special reference to the cotyledonary condition in Podocarpaceae. Phytomorphology 1 296-300. 1962. . Podocarpus vitiensis in the Moluccas (Taxaceae). Blumea 15: 440. Florin, R. Untersuchungen sur Stammesgeschichte der Coniferales und Cor- __ dhital les. Sv. Vet-akad. Handl. III. 10: 1-588. 58 pls. ————.. The Tertiary fossil conifers of south Chile and their phytogeographical significance, Ibid. 19: 1-107. 6 pls. 1940. 360 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Grsss, L. S. A oboe i to the phytogeography and flora of the Arfak Moun- tains. London. Gorpon, G. The satay London. 1858. Hooker, J. D. Podocarpus pectinata. bb5671, bb6737, bb7077 (12a): bb7708, bb8737 (21b); bb8842 (34): bb9003 (21a); bb9664 (17); bb9671 (1); bb9696 (34); bb10748 (12a) bb11803 (21a); bb13633 (26): bb14390, bb14519 (6a): bb15154. (1): bb15155 (21b) bb15504 (2la); bb17229 (35); bb17269 (21a); [voL. 50 bb17544 (17); bb18217 (34); bb18328 (35); bb18752, bb19559 (21b); 6619564 (1); bb19647 (34); bb19709 (6a); 6b19869, bb19870 (4a); bb20202 (21a); bb20270 (1); bb20535. (6a); bb20782 (1); bb21151 = (35); bb21294 (17); bb21509 (1) ; bb21511 (3); 0b24489 (34); bb24777 (1); 6624778 (17); 6b24779 (12a); bb24934 (6a); bb24951, bb24956 (21b); bb24957 (21a); bb24958 (26); bb24964 (3); bb25157 (17); bb25541 (34); bb26288 (3); bb26589, bb27736 (5a); bb28147 (21b); 6b28495 (35); 6b28751, bb28752, bb28753, bb28754 (5a); bb29195 (21a); 0bb29532 (359% bb30260 (34); 6b30321, bb30475 (6a); 6b30717, bb30779, bb30813, bb30887, bb30939, —-bb30947, bb30961, bb31500, bb 31506, bb31536 (34); bb31596 (35); 6632284, bb32434, bb33046 (5a) New Guinea Forestry Department (NGF), the following by anony- mous collectors: NGF-W41 (34); NGF 3128 (21c); NGF 4503 (34) Nicholson SAN 17292 (5a); SAN 17823 (1); SAN 17826 (14); SAN 17827 (13); SAN 39766 (25); Nicolson 1319 Noakes 20133, 22147 (34) Nur 10507 (17) Ocampo 27926 (24) Oillerings 175 (21a) Omar SFN 376 (Sb) Palmer & Bryant 988 (21a) Pancher 4 (22); 379 (18); 380 (8) Pascua 15692 - Petit 138 (33); 177 (8) Phengkhlai 568 nie 691 (2) Pickles 2991 (2) Pierre 1396 (2); 5528 (21b); 5529 5530 (34); se (2) Pitard 2090 Poilane 25 (2); 320 (21b); 1539 (2); 2147, 3387 (21b); 3455, 3782 (2); 4038 (21b); 4411 (2); 5963 (34); 1969] 6509 (21b): 7095 (2); 9103, 10293, 16092, 23118 (21b); 23216, et 24314 (34); 29808 (37); 29960 (21d); 31558 (34); 32825, 33351 (2); 35595 (21b); 35675 Poore 6228 (17) Posthumus 2175 mand i (21a) Pringo Atmodjo 82 (21 Pulle 663 (3): 801 oa 964 (26); 966 (3); 982, 1018, 1042 (1) Pullen 273A (1); 313, 313A (26); 338 (28); 2674, 2680 (6b); 2716, 2716A (23); 2840 (31); 5052 (26): 5111, 5138 (28); 5267 (26); 5914, 5930 (21c); 5930A (3); 5932 (34); 611 Purseglove P5006, P5553 (5a) Quisumbing & Sulit 82404 (1); 82481 (21c) Rabil 1 Rabor 20482 (17); 20485 (13) pate 19557 (24); 77401 (1) os & Edafio 26394 (17); 26501 Raap on 768 (21a) 9 (34 (12); 37757, 38738 (1); 45005 (24); 46333 3 (34) Ramos & Pascasio 34497 (5a) Rappard BW 697 (6b); BW 698 ) Rashid $9546 (5a) Rensch 1307 (21a) Renwarin 5b2436 (21b); 662450 (34) Richards 1058 (1); 1059 (12a); 1628 (1); 1768 (21a); 1808 (12a); 1834, 1836 (17); 1962 (2): ao (12a); 1997 (10); 2421, 2476 Ridley 5695 (17); "8636 «ib, 11223 34); 16026, 16178 Robbins 238 (26); 598 (3); 673 (26); 718 (8). 3112 (26); 3214 (27); 3266 Beene 7 Rossum 122, 784 (5a) Sadau 42890 (21a) DE LAUBENFELS, PODOCARPACEAE 367 Salverda 6622564 bb22576 (6b) Santos 31817 (21c) (1); 6622571, ) Sarlin 73 (33); 228 (32); 229 (18); 237 (22): 242 (15); 244, 341 (22) Saunders 708 (28); 804 (26); 823 (21c); 824 (1); 861 (21c); 1025 (1); 1048 (21c); 1088 (34) Sayers NGF 21613 (21c pen 1473, 1474 (21b); 1475 ations 15175, 13176 (8); 19331, Schmid 137 (29) Schodde 1561 (21c); 2014 (1); 2021, 26 Schram BW 1785, BW 6021, BW 6174, BW 6705, BW 6760, BW 7951 (34); BW 7972, BW 9271 (6a); BW 10596 (34) Seemann 573 (6a); 576 (31) Shockton 2699 (1 Sijde BW 5579 (34); BW 5596 (6a) Sinclair 10578, 38991 (34); 39094 (2) Sinclair & Kadim 9053 (1); 9146 (25); 10318 (5a) Sinclair & Salleh SFN 40798 (35) Sing JC/59 (Sa) Singh SAN 24336 (Sa) Skottsberg 202 (32) Smit BW 2314 (34) Smith, A. C. 1773 (6a); 1796 (31); 4122 (30); 4901 (21b); 5734, 6244 (6a); 6245 (21b); 7076 (31) Smith, L. S. NGF 1352 (6a) Smith, R. 66 (38) Smitinand 19058 ( S10601 S10607 2) (21c); Sonohara, Tawada, & Amano 6290 (38 Spurway 376 Stauffer 5651 (1); 5652 (21c); 5670 (28); 5729 (18); 5807 (33) Stauffer & Blanchon 5812 Stauffer, Blanchon & Boulet 5778 (22) Stauffer & Kuruvoli 5841 (6a) Stefels BW 2006 (21c); BW 2008, BW 2010 (1); BW 2014 (21c); 368 BW 2015 (3); BW 2031 (1); BW 2033 (3); BW 2038 (21c); BW 147 (34 Steiner 2032 (24); 2150 (1); 2207 24 Stern 2242 (21c) Stern & Rojo 2289, 2292 (21c) Steup bb23045 (6a ) Steward & Cheo 728 (38) Storck 906 (6a) Stresemann 125 (26); 133 (1); 158 (21b); 251, 276A (26); 354, 363 21b); 395 (3) Strugnell 16568 (35); 23931 (21b) Sulaiman 2 (5a Sulit 7586 (21c); (21b); 10052, es (1); (12a); 30051 (24) Surbeck 107 (12a); 532 (21b) Symington 27841 (35) 7669 (1): 9896 21694 Tagei 1795 (5b) Tang 438 (21b); 457 (5a) Telussa BW 5158 (34 Teysmann 169 (1 617 (Sa); 11598 (34); 11599 (5a): 21647 (2) Thailand Royal Forest Department 631 (2 Thorenaar 12713 (35) Thorne 28565 (33); 28568 (8); 28644 (18): 28704 (29): 28705 (18); 28734 (9) Toropai NGF 17153 (1) Tothill 553 (6a); 844, 845 (31); 854 6a Toxopeus 427 (3); 485 (21b) Tsang 681, L.U. 17430 (8); 20123, 25273 (37); 27332 (21b Tsang & Fung L.U. 18100 = Tuckwell W1553 (26) Van Romer 736 (26); 1233 (12c) Van Royen 3721 (1); 3857 (3): 3873 (1); 3895 (21c); 5058 (34); NGF 16182 (1); NGF 20289 (23); NGF 20309 (26) Van Royen & Sleumer 6073 (31 6246 (6a); 7219 (3); 7403 (1). 7948 2); 79484 (21c); 7948B (3): 8203A (19) JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Van Royen, Sleumer, & Schram 7791 (12 a Van Steenis 3754 (34); 8357 (12a); 23 (21d); 17544, 18267 (21a) Vaughn 3254 (31); 3258 (21b) Veillon 120 (32); 136 (29); 142 (22); 145, 511 Versteegh BW 248 (1); BW 250 (21c); BW 253 (1); BW 269 (3); W 7596, BW 10407 (1); BW 10411 BW 12610 (21b); 15249 (6a) Versteegh & Kalkman BW 5594 (6a) Versteegh & Koster BW 14 (21b) Vidal 623 (24); 3910 (13) Vieillard 1259 (18); 1260, 1261, 1262 (22); 1275 (33); 1277 (8); 1278, 3262 (7); 3264 (32); 3265 (9) Vink NGF 12430 (26); BW 15271 (6a); 17188, 17242 (26); 17499, 17500, 17501, 17502 (27) Vink & Schram BW 8620 (6b); BW 8667 (23); BW 8730 (31); si 8731 (21c); BW 8746 (3); 8764 (1); BW 8796, BW i (6b); BW 8945 (1) ae oh 9 (9); 10 (29); 37 (7); 8 (32); 39 (18); 40 (9); 152 fy. 187 (7); 206 (22); 400 (7); 469 (18); 658 (33) Wakau 4155 (2) Walker, F. S. BSIP 212 (31); BSIP 247 (13 Walker 70 (2); ANU 859, ANU 859A (1); 5649 (38); 7526 (21c) Wallich 6045 (2); 6050 (34) ang 33651 (Sa); 35591 (21b); 36532 (5a); 39608 (21b) Warburg 11119 (21a); 14721 (21b) White 2001 (7); 2033 (32); 2112 (22): 2120 (32); 2122 (8); 2238 (7); 2261 (33); 2285 (22) White & Gray NGF 10407 (6a); GF 10415 (34) Whitford 951 (24); 1353 (34) Whitmore 2368 (21b) 1969] Williams 399, 624, 752, 753, 1035 (34); 1298, 1299 (21c Wilson 6262, 8064, 10242, 11109 Winkler 512 (1); 1035 (21a); (1); 1037 (13); 1866 (21a) Womersley NGF 3704 (34); NGF 4420 (3); NGF 4428 (23); NGF 4483 (19); NGF 5338 (21c); NGF 10279, 1036 11067, NGF 11260 (21c); NGF 13922 (19); NGF 14018 (26); NGF 14253 (21c); NGF 17621 (34); NGF 17902 (1); NGF 17939 (23): NGF 19298 (34); NGF 24563 (21c); NGF 24569 (26); NGF 24928 (21b) Womersley & deLaubenfels 19460 (21c) Womersley & Floyd NGF 6138 (21c) NGF DE LAUBENFELS, PODOCARPACEAE 369 Womersley & a NGF 7680 (3); NGF 8324 Womersley & Sones NGF 14013 21¢ Wood 1244 (34); SAN 4172 (5b); SAN A4179 Wood & Wyatt- ea SAN A4493 (25) Wray 1028 (2); 1198 (21b); 3875 (12b); 3899 Wray & Robinson 5354, 5380 (2) Wyatt-Smith 71650, 80370, 80371 (1); 93115 Yapp 493 (12a) Yates 1987, 2148 (21b); 2554 (34) Zollinger 2262 (21a); 3025 (34) Zwart 6517 (21d) DEPARTMENT OF GEOGRAPHY SYRACUSE UNIVERSITY Syracuse, New York 13210 370 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 THE VASCULAR SYSTEM IN THE AXIS OF DRACAENA FRAGRANS (AGAVACEAE), 1. DISTRIBUTION AND DEVELOPMENT OF PRIMARY STRANDS. M. H. ZIMMERMANN AND P. B. ToMLINSON 1 WE HAVE OUTLINED THE STATE of existing knowledge of the vascular anatomy of monocotyledons with secondary growth in a previous article which serves as an introduction to our present studies (Tomlinson & Zimmermann, 1969). These have been concerned with a number of genera in the Agavaceae. There are quantitative and often diagnostic anatomical differences between the plants we have studied, but we believe that funda- mental principles of vascular distribution is the same i all of them. We have therefore restricted our description to one species, Dracaena fragrans (L.) Ker-Gawl., of which we had abundant living material available for a detailed study. The same species had also been used by earlier investi- gators of this problem (e.g., Cordemoy, 1894; Meneghini, 1836; von Mohl, 1824) although often under the older, now incorrect name of Aletris fra- grans L. Where necessary, however, we shall refer to other plant species. For convenience, our results are presented as two separate articles, devoted to primary and secondary tissues respectively, although such a separation is somewhat arbitrary. Indeed, we will have cause to show that the two types of vascular tissue are interdependent and often continuous. MATERIALS AND METHODS Sectioning. Specimens investigated were collected from a large clump cultivated at the Fairchild Tropical Garden. Material was either sec- tioned freshly or after fixation in FAA and subsequent washing. For the study of the course of vascular bundles sequential series of tranverse sections were cut on a “Reichert” sliding microtome from two lengths of distal mature shoots, each representing a sympodium. Selection of ma- terial needed some care because in some shoots the central tissue is easily torn during sectioning. For the investigation of the vascular anatomy of leaf-trace departure a complete series of sections 40» thick was prepared. For the analysis of longitudinal continuity of vascular traces sections 33p thick at intervals of 100. were used. These sections were stained in safranin and Delafield’s haematoxylin and mounted permanently in “Piccolyte.”’ For the study of vascular development in the crown, serial transverse and longitudinal sections 10% thick of shoot apices were cut on a rotary * Contribution to a study of the vascular system of monocotyledons by one of @ (P.B.T.), supported by N. S. F. grant GB-5762-X. 1969 | ZIMMERMANN & TOMLINSON, DRACAENA, 1 371 microtome from material embedded in “Paraplast.” Routine methods of embedding, staining and mounting were employed. Because of the wide diameter of these developing crowns, pieces of ribbon containing only four sections were mounted on each slide. Serial analysis. Cinematographic analysis of the three-dimensional vascular structure was carried out with the series of sections from the mature stem. These methods have been described in ample detail in previous papers (Zimmermann & Tomlinson, 1965, 1966). The method of plotting provascular strands in the series of sections from the meriste- matic crown has also been described in previous papers (Zimmermann & Tomlinson, 1967, 1968). The optical shuttle was employed for plotting, according to the procedure described in the paper on the vascular develop- ment of Prionium (Zimmermann & Tomlinson, 1968). A slight methodical variation was necessary because each slide in this series contained four sections. Most strands can easily be followed by matching the corre- sponding section on two successive slides with an interval of 3 sections between. In areas where provascular strands make sharp turns (immedi- ately below the apical meristem) and require the use of each section, optical alignment was achieved in the following way: 5d—6a, 5d—6b, Sd— 6c, 6c—7a, 6d-7a, etc., whereby the number indicates the slide, the letter the section on the slide, and the italics mark the sequence of photography. The reader can easily appreciate that alignment had to be achieved over an interval of two sections (e.g., 5d-6c above) once every four sections. With the Dracaena crown this was just possible without losing continuity of the strand which was plotted. There is no question that the procedure is easier to follow when each slide contains only a single section. GENERAL MORPHOLOGY AND ANATOMY Growth habit. Dracaena fragrans, a native of tropical Africa, is a com- mon ornamental in South Florida. It has apparently been known in cul- tivation in Europe at least since 1768 (Sims, 1808). In cultivation it forms a diffuse shrub or rarely a low tree. Basal shoots are straight and erect, but distal branches are often bent over by the weight of the terminal cluster of leaves. Leaves are lanceolate, up to 75 cm. long, and lack a petiole. Their insertion is open but broad and even overlapping. On vigorous shoots, leaves persist for a long time, so that leafy shoots up to 2 m. high may be present. Otherwise on suppressed and less vigorous shoots leaves form the distinct terminal cluster which is so common in many other woody monocotyledons. The longevity of leaves is of con- siderable anatomical significance, a point which will be discussed later. Each leaf subtends a minute axillary bud enclosed by its prophyll. Inflorescences are always terminal (Fic. 1) and branching is closely associated with flowering. Flowering and resultant branching begins in plants 2 to 3 years old when they are about 1 m. high. The transition from vegetative to reproductive state of the axis is marked by a gradual reduc- 372 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Fic. 1. Dracaena fragrans, Distal part of a flowering shoot, X 1/5. tion in leaf size associated with an abrupt elongation of internodes (Fics. 2-5). Bracts which subtend the first-order flowering branches are white and caducous although it is obvious from the transition upwards along the shoot that they are homologous with foliage leaves. The detailed structure of the flower-bearing branches has been described by Troll (1962). Branching is normally sympodial from a bud in the axil of one of the transitional leaves immediately below the inflorescence. At the time of anthesis this renewal bud is indistinguishable from other dormant buds (Fic. 4). After flowering it grows out rapidly (Fic. 7), pushing the terminal inflorescence into a pseudolateral position. Each axis branches in this way and is, therefore, a sympodium consisting of many successive growth units. The sympodium appears articulate both from the scars of the dried inflorescence stalks, and a swelling which marks each joint (Fic. 3). This articulation is most noticeable in distal, horizontal axes. After flowering there is an evident competition among a number of potential renewal buds, because the inhibition of more than one is always released (Fic. 7). One of them usually becomes dominant and re-imposes inhibition upon the others. However, two (rarely more) buds may grow out simultaneously. This leads to a branched axis with an inflorescence scar in the crotch of the fork. Plants in South Florida flower several times in one year and although it is obvious that different axes flower at dif- ferent times of the year it seems likely that a vigorous shoot can flower 1969] “aNas Eats Me%, S (, fi, u/”( G, Cy E Le | 4 ud ( | wd ( 3, rie, . 2-5. Dracaena fragrans. Habit details. 2, Terminal inflorescence, < 1/2. b ow ering shoot with all leaves detached, showing 3 units of sympodium and ae uation of vegetative axis, X 1/2. Letters Ciied i. & m distal unit shown in Fic. 3, their levels of insertion indi. €d in that figu - . besinning with prophyll (a) and ending with most distal transitional leaf ( (f), x 374 JOURNAL OF THE ARNOLD ARBORETUM [voL, 50 Fics. 6 and 7. Dracaena fragrans. Branching. 6, Erect axis with 2 branches, renewal shoots at A and B from axillary buds whose inhibition is released by destruction of apex of parent shoot at X, 2/3. 7, Normal renewal growth below old inflorescence axis, X 2. Buds developing in axils of two most distal coed ; prophylls conspicuous. This corresponds to Fic. 4 after lapse of 2 months. two or three times each year. This is indicated by a close succession of inflorescence scars. For a further morphological description of flowering in arborescent Liliiflorae the reader is referred to the detailed work of Schoute (1903, 1918). The release of apical dominance is normally the result of flowering but it may be induced in other ways. Decapitation releases from inhibition the dormant buds immediately below the injury (Fic. 6). Apical domi- nance is also released on the upper side of leaning stems where numerous dormant buds may grow out, much as in woody dicotyledons (see Fig. 15 in Tomlinson & Zimmermann, 1969). Erect, rapidly growing suckers commonly develop from the base of old plants, presumably for the same reason. The influence of these various methods of growth on the distribu- tion of vascular tissue is largely described in the second article of this series. Primary tissues (Fic. 8). Epidermis slightly thick-walled, covered by a thin but conspicuous cuticle. Periderm in hypodermal or subhypo- dermal layers developing early by etagen-like divisions of cortical cells, the outermost derivatives suberized. Cortex, 1-3 mm. wide, of uniform and fairly compact parenchyma; no independent cortical vascular system . \ on 1969] ZIMMERMANN & TOMLINSON, DRACAENA, 1 375 aa ; ot ey, A Eide ee oe = aR f+ fae oes i : ae we Cit Ag * F agrans, “ 36. All bundles in cortical area are leaf traces. Arrows point out vertical bundles immediately above point of branching from leaf trace. They appear small because they lack the fibrous sheath. G. 9 (BELow). Transverse section of mature stem of Dracaena fragrans, taken from the same ; w centimeters higher, below the sympodial branch. A small amount of secondary tissue has been formed by the cambium, xX 36. developed. Central cylinder delimited by compact, often lignified ground parenchyma and peripheral, congested vascular bundles. Central bundles More diffusely distributed among thin-walled ground tissue resembling 376 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 parenchyma of cortex. Vascular bundles each with a sheath of narrow compact angular cells, the sheathing cells thick-walled around phloem but sheath becoming more uniformly sclerotic around peripheral bundles. Vascular tissues collateral, including a wide strand of angular metaxylem elements, V-shaped in transverse section, usually with narrow protoxylem elements at the apex of the V and a single phloem strand in the angle of the V. Peripheral vascular bundles with little or no protoxylem, the metaxylem scarcely V-shaped. Leaf traces conspicuous in outer part of central cylinder, the xylem represented largely by abundant protoxylem. Xylem including fairly wide angular tracheids with indistinct end walls and scalariform pitting. Protoxylem elements rounded, with annular or spiral wall thickening. Metaxylem tracheids of the order of 5—10 mm. long, and overlapping extensively. Phloem including long sieve-tube ele- ments usually with transverse end walls and simple sieve plates, but sieve plates commonly compound on oblique or very oblique end walls. Raphide clusters common in otherwise unmodified parenchyma cells. Tannin cells infrequent. Secondary tissues (Fic. 9). This arises from an etagen cambium of the type which has already been described in the earlier review (Tomlin- son & Zimmermann, 1969). Ground tissue of compact tabular and radially- arranged cells about 120, long, with slightly thickened and lignified cells, the walls with abundant simple pits. Tannin deposits and raphide clusters frequent. Secondary vascular bundles always amphivasal. Central phloem strand including short sieve-tube elements with simple, more or less trans- verse sieve plates. Phloem separated from xylem by short, thin-walled parenchyma cells. Secondary tracheids conspicuously different from those of primary vascular bundles; of the order of 3.6 mm. long and with indefinite end walls; walls thick; bordered pits with crossed slit-like apertures, more or less parallel to the axis of the cell. Short xylem parenchyma cells infrequent. The difference in length between secondary ground tissue cells and secondary tracheids suggests that the latter undergo about a 30-fold extension during development since both arise from similar initials. COURSE OF PRIMARY VASCULAR BUNDLES The distribution of primary vascular tissue in Dracaena fragrans is similar to that of the palm Rhapis excelsa as described by us (Zimmer- mann & Tomlinson, 1965) with slight quantitative differences. Each leaf is supplied with a number of leaf traces which diverge from the stem at varying depths. Major bundles diverge from the center, minor bundles from near the periphery, and intermediate bundles from an intermediate area of the stem. Outgoing leaf traces produce a number of derivative bundles by branching. Most of these branches are short bridges which link in an upward direction with nearby vertical bundles. Axial continuity from each leaf trace is maintained by a continuing vertical bundle which 1969 | ZIMMERMANN & TOMLINSON, DRACAENA, 1 377 CENTRAL CYLINDER CENTRAL CYLINDER <—__. <— | DRACAENA RHAPIS Eat | | usually diverges from the leaf trace at the very periphery of the central cylinder. The newly released vertical bundle is normally very narrow if the stem does not contain secondary tissue (Fic. 8). In this very peripheral position it readily splits or anastomoses with similar neighboring bundles on its way up the stem. This contrasts with the situation in Rhapis where the vertical bundle is released near the stem center and accompanies the parent leaf trace, on its outwardly diverging path, almost all the way to the periphery of the central cylinder (Fic. 10). The upwardly continuing vertical bundle, as in Rhapis, then gradually approaches, over a distance of many internodes, the center of the stem whereupon the process of bundle branching is repeated again in association with another, more distal leaf. Major bundles have the longest, minor bundles the shortest istances between two such successive leaf contacts. Thus, the overall course of vascular bundles is similar to that illustrated for RAapis (Zim- mermann & Tomlinson, 1965; Fig. 3, right). In Dracaena, in contrast with Rhapis, the central bundles have no helical path. Major dorsal bundles merely describe a turn of about 120° in the center, as described 378 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 in the next section of this paper. This turn, as with the helical twisting in Rhapis, is in the direction of the phyllotactic spiral. This turn may be governed by the same developmental principle which causes phyllotaxis. The distribution of protoxylem changes throughout each bundle in the same manner as in Rhapis. This change is less conspicuous in Dracaena than in Rhapis because protoxylem and metaxylem elements are more nearly of the same diameter. As one follows a vascular bundle of Dracaena in a distal direction one finds protoxylem first not very far above its divergence from the leaf trace as a vertical bundle. Continuing upwards, the number of elements further increases to reach a maximum where the bundle passes out into another leaf. Metaxylem is continuous into bridges as well as the continuing vertical bundle but the leaf trace contains only protoxylem. In this respect Dracaena is identical with Rhapis although the “loss” of metaxylem from the outgoing leaf trace is less obvious be- cause the two tissues are not so clearly distinguished. Irregularities in the course of bundles throughout the stem are somewhat more common than in Rhapis. The anastomosing tendency of the lower part of vertical bundles, at the periphery of the central cylinder, has been mentioned. If the stem consists of primary vascular tissue only, the vertical bundles are quite small in their lowermost portion, at the periphery, where they come off the leaf trace and without a well-developed fibrous sheath (Fic. 8). In places where the primary vascular cylinder is covered by a mantle of secondary tissue, the same vertical bundles are larger and more conspicuous, because the fibrous sheath is better developed (Fic. 9). Another irregularity which has been observed is the occasional forking leaf traces. When such a bundle is followed upwards in the stem center, the two branches diverge along two different radii. From these obser- vations it appears that developmental processes are somewhat less rigid in Dracaena than in Rhapis. The important topic of the relation between primary and secondary vascular bundles is reserved for the second article in this series. DEVELOPMENTAL PATTERN OF THE PRIMARY VASCULAR SYSTEM Observations. General aspects of the anatomy of the meristematic crown are shown in the photomicrographs, FicurEs 11 and 12. Leaves and leaf primordia are arranged in a phyllotactic spiral with a divergence between 1/3 and 2/5, as can be seen from Ficure 11. The approximately median longitudinal section through the crown shows the usual mono- cotyledonous organization (Fic. 12). It is obvious from this longitudinal section that primary thickening growth involves re-orientation of tissue through about 90° as we have described for Rhapis and Prionium (Zim- mermann & Tomlinson, 1967, 1968). The developing vascular system of the meristematic crown is far too complex to be demonstrated in individual microtome sections. Provascular strands were, therefore, followed throughout a series of transverse sections and their radial distance from the stem center plotted on graph paper 45 1969 | ZIMMERMANN & TOMLINSON, DRACAENA, 1 379 Fic, eee crown of oo fragrans at the level of the apical meristem, X 2 Showing the phyllotactic arrangements of the leaves. Note the symmetrical ar- rangement of the major leaf traces on the dorsal side. 1G. 12 (Bexow). Approximate median longitudinal section through the meristematic crown of Dracaena fragrans, X 26. Because of their c ge three- dimensional path, none of the indi vidual inh nr strands can be er more than a very short distance. A thorough knowledge of their path, ~ ine S$ quite easily. Note the sha arp turns of the major strands below the apica meristem. Note also the minor leaf trace on the far right of the photograph. 380 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 had been done for Rkapis and Prionium. The results are shown in FIGURE 13. The three-dimensional arrangement of provascular strands in the crown is difficult to represent on paper, our representation is therefore simplified as follows. All radii are shown in a single plane. All leaves are rotated into the same plane. This eliminates the 120° turn of the major bundles. In order to reconstruct the three-dimensional pattern of the crown from Ficure 13, the reader has to go through a mental exercise which first involves rotating leaves back to their proper position, and second involves re-establishing the 120° turn of the major bundles. This process of simplification is very similar to the one used in our description of the Rhapis crown. It has the advantage that one can more easily appreciate the re-orientation of a major dorsal bundle during successive developmental stages. FicurE 13 shows the major dorsal leaf trace in leaf primordia P 1 to P 17, P 1 being the youngest visible primordium. The pattern of vascular development appears to be the same as the one found in Rhapis and Prionium. Vertical bundles originate from major leaf traces of P 17. Approximately below the base of P 14 they fuse into the meristematic cap into which all blind-ending vertical bundles converge (cf. Zimmer- mann & Tomlinson, 1967, 1968). From the diagram one can extrapolate that the leaf-contact distance for a major bundle is about 20 to 25 inter- nodes, although the series of sections was too short to show this directly. If the section series had been longer and had included the insertions of older (lower) leaves, the major leaf trace of P 1 would have been seen origin- ating as a vertical bundle from a leaf trace diverging into a leaf at about the level of P 20-25. A rather unusual type of vertical-bundle branch was found in two major leaf traces to P 11. Both vertical bundles ended distally immediately below the apical meristem in what might be a leaf primordium younger than P 1 and represented by an indistinct ridge. If this interpretation is correct there would be a leaf-contact distance of 11 internodes between P 0 and P 11. Only two such centrally located vertical bundle branches were found and one of them is shown in Ficure 13. The developmental meaning of this rare type of vertical bundle is unknown. Ficure 13 shows some further irregularities which are of no funda- mental significance, such as the apparent crossing over of the lower portions of the major leaf traces of P 1 and P 2, P 5 and P 6, P 7 and P 8. They could have resulted by comparing bundles on different radii of the stem (the crown is not perfectly circular in transverse section), from slight irregularities of development, or indeed, from the process of plotting. The meristematic cap is similar in position and extent to that described in the apices of Rhapis and Prionium. It is recognized as the umbrella- shaped meristematic area, below the shoot apex proper, into the periphery of which the blind-ending vertical bundles fuse. It is pierced by lea traces already connected to vertical bundles. f The primary vascular connection between an axillary bud and the vascular system of the central cylinder has also been traced in this series 1969] ZIMMERMANN & TOMLINSON, DRACAENA, 1 381 : OXY Wir OS ae Ve WR So bg / Al is J £ 2 3 4 MILLIMETERS FROM STEM AXIS DOWN Fic. 13. The path of major leaf traces from leaf primordia 1 through 17, obtained from plotting each individual bundle through a series of transverse bu : : bundle. The dashed line marks the approximate level of leaf insertion. INSET BELOW. The 120° turn of the major leaf traces in the stem center, as seen in successive transverse sections. of sections. The provascular connection between axillary bud and stem was established only in P 17 and older leaves. In P 15 the axillary bud meristem was apparent but still entirely without discernible procambial Strands. This suggests that vascular continuity between axillary buds and main axis is established late, in the manner of minor leaf traces. A more detailed discussion of the development of the vascular system of axillary buds will follow in the second paper of this series. Developmental inferences. The sequence of vascular development is thought to be as follows. Leaf traces link up with a potential vertical bundle in the cap, then differentiate out below the cap. Leaf traces which develop early, i.e., those arising in a position near the center of the cap, €come major bundles; those developing further out, near the cap Periphery, become minor bundles. For comparison both a major and a minor trace from P 7 are shown in Ficure 13. 382 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 These developmental processes have been discussed in detail in our articles on the developmental pattern of the vascular system of Rhapis and Prionium. We may merely point out here that there are no cortical bundles in Dracaena, a fact which is of utmost significance as we shall see in the next paper of this series. DISCUSSION In a previous review (Tomlinson & Zimmermann, 1969) we have noted that von Mohl (1824) equated the primary vascular system of “Aletris fragrans” and other species which he studied with that of a palm, in so far as he understood the course of vascular bundles in the palm stem. Von Mohl’s contemporaries and all subsequent investigators who studied ar- borescent Liliiflorae at first hand claim to have confirmed his observations (e.g. Meneghini, 1836; Millardet, 1865; de Cordemoy, 1894; and others). However, our own more recent investigation of the palm stem (Zimmer- mann & Tomlinson, 1965) has shown that von Mohl’s understanding was incomplete, because he overlooked the axial continuity of vascular bundles which is so important in long-distance transport. We have already given our historical interpretation of the topic (Tomlinson & Zimmermann, 1966) and need not discuss it any further. The present study of Dracaena fragrans has confirmed that von Mohl was right in principle. The primary vascular anatomy of the axis of this plant does indeed correspond in all essentials with that of a palm, but we now have a much more complete understanding of the anatomy of the palm and its development. Dracaena conforms in the pattern of primary vascular differentiation, a pattern which we believe is fundamental for monocotyledons as a whole (Zim- mermann & Tomlinson, 1965, 1967, 1968). Our cinematographic analyses have included other species and genera of arborescent Liliiflorae. Analysis of the mature axis of Cordyline ter- minalis, Dracaena marginata and Pleomele (Dracaena) reflexa confirms the course of vascular bundles described for Dracaena fragrans. Single sections which we have prepared from the stems of several other genera and species can also be interpreted according to our three-dimensional analysis. This additional evidence puts our interpretation on firm ground. SUMMARY The primary vascular system of arborescent Liliiflorae was thought by von Mohl and subsequent investigators to be equivalent in principle to that of palms. An analysis of the system in the vegetative axis of Dracaen@ fragrans with the aid of cinematographic methods confirms this. In addi- tion, however, it also shows that axial continuity of the palm type, OV&T looked by these early anatomists, which has only recently been demon- strated, also occurs in Dracaena and related plants. The origin of the primary vascular system has been traced by plotting the course of provascular strands in the developing crown. We regard, on the basis of pee pene —_— 1969 | ZIMMERMANN & TOMLINSON, DRACAENA, 1 383 similar studies of other plants, this pattern as fundamental for the mono- cotyledons. The basis has thus been laid for a future investigation of secondary vascular tissues in these plants. LITERATURE CITED Corpemoy, H. J. pe. 1894. Recherches sur les Monocotylédones a accroissement secondaire. Thesis. Paris. pp. 108. 3 pls. MENEGHINI, G. 1836. Ricerche sulla struttura del caule nelle piante Monoco- tiledoni. pp. 110. 10 pls. Minerva, Padua. MItarbet, A. 1865. Sur l’anatomie et le rigour du corps ligneux dans les genres Yucca et Dracaena. Mém. Soc. Sci. Nat. Cherbourg 11: 1-24. Mont, H. von. 1824. De palmarum structura. In: K. F. P. von Martius, His- toria Naturalis Palmarum 1: pp. I-LII. 16 is. sais var = C. 1903. Die Stammesbildung der Monokotylen. Flora (Jena) 92: ei Uber die Verastelung bei monokotylen Baumen. III. Die Veras- telung einiger baumartigen Liliaceen. Rec. Trav. Bot. Néerl. 15: 263-335. Sims, ae oo Dracaena fragrans, Sweet-scented Dracaena. In: Bot. Mag. 28: 081. TROLL, . 1962. Uber die “Prolificitat” von Chlorophytum comosum. Neue Hefte Morphologie 4: 9-68. Tomtinson, P. B., _H. ZIMMERMANN, 1966. Vascular bundles in ee stems Saithedy bibliographic evolution. Proc. Am. Philos. Soc. 110: 1. 1969. Vascular anatomy of monocotyledons Nan peal rowth — an ‘introduction, Jour. Arnold Arb. 50: 159-179. ,& ————__, excelsa, I. Mature vegetative axis. Jour. Arnold Arb. 46: & 1966. Analysis of aia vascular ie in ren Optical shuttle ies peviecant 72, ——— . Anatomy of os palm Rhapis excelsa. IV. Vascular de- i ie dh in aes of vegetative aérial axis and rhizome. Jour. Arnold Arb. 48: Siu a ee & —. 1968. Vascular construction and development in the aérial stem of Prionium (Juncaceae). Am. Jour. Bot. 55: 1100-1109. [M.H.Z.] BT. HARVARD UNIver RSITY FAIRCHILD TROPICAL GARDEN Casot FounDATION 10901 OLtp CuTLER Roap PETE M Miami, FLorma 33156 MASSACHUSETTS 01366 384 JOURNAL OF THE ARNOLD ARBORETUM [VvoL. 50 COMPARATIVE MORPHOLOGICAL STUDIES IN DILLENIACEAE, Iv. ANATOMY OF THE NODE AND VASCULARIZATION OF THE LEAF WILLIAM C. DICKISON IN A CONTINUING EFFORT to provide comprehensive anatomical infor- mation which might prove useful in elucidating taxonomic and phylo- genetic relationships of the Dilleniaceae, an extensive investigation of nodal and leaf vasculature was undertaken. Aside from remarks pertaining to ovular structure by Cordemoy (1859), and an occasional reference to internal structure by various other workers, the earliest comprehensive anatomical investigations on Dilleniaceae are the contributions of Baillon (1866-67, 1871) and Hitzemann (1886, cited by Ozenda, 1949). The first comparative morphological studies on the family to appear were those of Parmentier (1896), who found the leaf to contain charac- ters of diagnostic value, and Steppuhn (1895) who made an extensive in- vestigation of stem, leaf, and root of some one hundred fifty dilleniaceous species. Solereder (1908) and Metcalfe and Chalk (1950) published additional anatomical information, but contributed little to help clarify the phylo- genetic position of the group. The most recent study on comparative vegetative anatomy of the family was by Ozenda (1949) whose observa- tions on seedling, nodal, and leaf anatomy were scattered among seven genera. All researches referring to the Dilleniaceae, therefore, are either in- complete, or else were produced in the last century and thus warrant re- investigation. This paper describes heretofore unreported anatomical data of both taxonomic and phylogenetic significance. MATERIALS AND METHODS Material of over one hundred dilleniaceous species was examined. Speci mens studied were received from, or are housed in: the Arnold Arboretum, Harvard University (a); State Herbarium of South Australia, Adelaide (ap); Arizona State University, Tempe (Asu); Botanic Museum and Her- barium, Brisbane (pri); Commonwealth Scientific and Industrial Research Organization, Canberra (caANB); Royal Botanic Garden, Edinburgh (£); Gray Herbarium, Harvard University (crt); Royal Botanic Gardens, Kew (k); Botanical Survey of India, Southern Circle, Coimbatore (aH); Mis souri Botanical Garden, St. Louis (mo); Animal Industry Branch, North- ern Territory Administration, Alice Springs (Nr); Western Australian 1969 | DICKISON, DILLENIACEAE, IV 385 Herbarium, Perth (pertH); Rancho Santa Ana Botanic Garden, Clare- mont (RSA); Sarawak Museum, Kuching (sar); Botanic Gardens, Singa- pore (sinc); University of California, Berkeley (uc); and the United States National Museum, Washington (us). The assistance of the cura- tors of these collections is gratefully acknowledged. I also wish to thank Doctors R. D. Hoogland, H. Keng, and C. R. Metcalfe for providing seed used in this study. The study of lamina vascularization was accomplished entirely through the use of cleared leaves. Clearing was carried out using the standard NaOH method followed by safranin stain. Dried materials were initially re-expanded in 5 percent NaOH prior to fixation and sectioning. Nodes were serially sectioned and stained with a combination of safranin-fast green. Petiole vasculature was followed by obtaining sections throughout the length of the petiole as well as midway through the midrib. NODAL ANATOMY No detailed, comprehensive study of nodal anatomy in the Dilleniaceae has previously been undertaken. Sinnott (1914) attempted to utilize the nodes of several genera within the family (sensu Gilg, 1893) as evidence to support his idea that the trilacunar node was primitive; furthermore, that the unilacunar and multilacunar node was derived by reduction or amplification. This author listed six genera of the Dilleniaceae (sensu stricto) as having tri- or pentalacunar nodes. Ozenda (1949) after an examination of Hibbertia, Dillenia, Schuma- cheria, Tetracera, C uratella, and Davilla also concluded that the mature nodes of the family were tri- or multilacunar; however, he was of the opinion that the multilacunar condition was the primitive pattern. The primitive nature of the multilacunar node in the Dilleniaceae has also been advocated by Meeuse (1966, p. 49). As a result of comparative morphological data from both fossil and extant plants, in addition to ontogenetic considerations, the primitive na- ture of the trilacunar node was questioned by Marsden and Bailey (1955) and Canright (1955). These authors suggested that the unilacunar two- trace system represented the primitive condition. The unilacunar two- trace node is characteristically described as having two vascular traces which arise from independent primary bundles and, therefore, do not Tepresent the dichotomy of a single median trace. Pant and Mehra (1964) re-evaluated nodal anatomy in many Pterop- sida, and concluded that the statements of Marsden and Bailey (loc. cit.) concerning nodal patterns in fossil ferns and gymnosperms were not always substantiated, They then advised caution in accepting the unilacunar two- trace node as primitive for all Pteropsida. Results from a study of devel- opmental patterns in stem primary xylem indicated to Benzing (1967a, b) that the odd-numbered trace, unilacunar one-trace or trilacunar, was more likely to be primitive in angiosperms. A recent paper by Namboodiri and 386 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Beck (1968b) supports this view and regards the unilacunar one-trace node as the primitive condition in the Coniferales. My observations reveal that mature nodes of Dillenia (Fic. 8), Dides- mandra, and Schumacheria (Fic. 7) are exclusively multilacunar.t | The numerous leaf traces are associated with a corresponding number of pa- renchymatous gaps in the cauline stele. The number of traces in Schuma- cheria and Didesmandra was found to be stable at nine and seventeen respectively. The number in Dillenia, however, varies from as few as seven (D. puichella) to as high as twenty-seven (D. suffruticosa). Variability is also evident within a single species, the number of traces apparently reflecting the age and size of the node. The manner in which bundles de- part the stele was found generally to be correlated with the presence or absence of sheathing leaf bases. If the petiole does not sheath the stem, all traces tend to depart simultaneously. If the leaves are amplexicaul, the median trace passes out initially, with laterals departing in succession at higher levels. The leaves examined of the semi-herbaceous, rhizomatous genus Acro- trema were supplied by three traces; i.e., the node was trilacunar (Fic. 4). The median trace departs first with the resulting gap remaining open above the level at which the two lateral gaps close. Large-leaved species (e.g., A. arnottianum) should be studied when available to determine to what extent leaf size affects the nodal pattern in this extremely variable genus. In contrast to the information presented by Ozenda (1949), all hib- bertias are not uniformally trilacunar. Numerous species with reduced, needle-like leaves possess unilacunar nodes (Fic. 2). In these cases, the primary stele is composed of a continuous cylinder of vascular tissue with no discrete bundles discernible. At the unilacunar node, a single trace passes directly into the leaf. All broad-leaved hibbertias from New Cale- donia and Fiji are trilacunar. Leaf size is not always indicative of nodal patterns, however. Trilacunar Hibbertia huegelli (Fic. 45) and H. mono- gyna (Fic. 46), for example, possess smaller leaves than H. nitida (Fic. 47) which is unilacunar. The most reduced leaves in the family are encountered in the genus Pachynema. The small, scale-like, lateral appendages were found to be vascularized by a single prominent trace with a well defined gap in the stele. From the flattened stem of P. dilatatum, leaves may be secondarily supplied by weak cauline traces (Fic. 34). The New World genera Curatella, Davilla (Fic. 6), and Doliocarpus are mostly pentalacunar; but seven-trace nodes occur in Davilla aspera, and Doliocarpus major is trilacunar. Trilacunar, three-trace nodes are also uniform throughout the genus Tetracera where special effort was made to examine representative species from the Old and New World tropics. In Tetracera, three bundles are associated with three widely separated “The report by Benzing (1967a) of unilacunar one-trace nodes in the mature stems of Dillenia indica is in error. I have personally examined the sections used ™ this study and conclude that they were not taken from any member of the Dilleniaceae- —— ae 1969 | DICKISON, DILLENIACEAE, IV 387 gaps (Fic. 3). The lateral bundles arc up and through the cortex where they enter the petiole. This contrasts with the condition in the trilacunar hibbertias with sheathing leaf bases, where the laterals enter the leaf di- rectly from the stele (Fic. 5). The seedling anatomy of Dillenia indica, Tetracera indica, Hibbertia dentata, and H. scandens was examined. The cotyledonary nodes in the first two species are of the 2:1 type, viz., two traces departing from a single gap (Fics. 1, 40). My observation of the 2:1 cotyledonary nodal pattern in Dillenia indica once again contradicts the information presented by Ozenda (1949) who illustrated a single cotyledonary trace. Particular attention was paid to the double traces at subnodal levels and in all in- stances they originated from independent parts of the stele. In Dillenia indica, a species with multilacunar nodes in the mature stem, the first formed seedling leaves possess a trilacunar node (Fic. 41). Numerous examples can be found in other dicotyledonous families of a similar pro- gression, as in Magnoliaceae, Degeneriaceae, etc. The cotyledonary node of Hibbertia dentata and H. scandens differs by being of the unilacunar one-trace type (Fic. 12). No evidence of double- ness could be observed in the single, broad strand of vascular tissue which passes into the cotyledon. The occurrence of a 1:1 cotyledonary node in Hibbertia is of particular interest, since it is a genus with trilacunar mature nodes, and is generally considered to be more primitive in its characters than either Dillenia or Tetracera. The question again arises whether an even or odd number of nodal traces represents the primitive condition. A thorough study of the cotyledonary node in other Dilleniaceae would be worthy of careful attention. It is perhaps significant, that more than one case was observed among the seedling and mature nodes of Dillenia where an even number of traces prevailed. This condition resulted from suppression of one of the lateral bundles with the result that an even number of traces departed the stele. Although this might be dismissed as abnormal, the fact that it was observed more than once indicates that it may be of some significance. Although the majority of plant families exhibit a combination of uni- lacunar and trilacunar, or trilacunar and multilacunar nodes, it is relative- ly uncommon for a single family to possess all three types (Bailey & Nast, 1944). The Dilleniaceae are, therefore, unusual in possessing four pat- terns: unilacunar two-trace, unilacunar one-trace, trilacunar, and multi- lacunar. Bailey (1956) points out that transitions from trilacunar to uni- lacunar nodes may occur as the result of anatomical specialization in re- sponse to the environment. The reduction of leaf size in Hibbertia is such an adaptation. Thus, the mature foliage nodes in Dilleniaceae (sensu stricto) demonstrate two distinct trends of specialization: (1) secondary reduction and elimination of the lateral strands of the trilacunar nodes; and (2) amplification of the trilacunar node by the addition of laterals. Bailey and Howard (1941) note that trends of specialization in nodal anatomy are not infrequently correlated with specializations elsewhere in the plant (e.g., wood). No such direct correlations were noted in the dil- 388 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 lenias. In fact, those genera having the least advanced wood generally pos- sess the more highly evolved nodes. It is not possible to construct relationships within the family solely on the basis of nodal structure due to the presence of a similar anatomy in both Old and New World genera. Likewise, nodal anatomy is of limited value in determining relationships beyond the family. When the nodes of putatively related families are compared, it is evident that they are all essentially tri- or multilacunar. There is, accordingly, no basis for ac- cepting or rejecting alliances from this information alone. A significant exception to the above generalizations is found in the Theaceae where the node is uniformly characterized by a broad trace which departs from a single gap. Keng (1962) agreed with Canright (1955) in considering this pattern to be the result of phylogenetic fusion of several separate traces. If Canright’s (Joc. cit.) suggested trends of nodal spe- cialization are accepted, it leads one to the conclusion that the nodal anatomy in the two families represents the culmination of distinct lines of evolution. Therefore, although the wood and pollen of these groups is similar (Dickison, 1967a, b), nodal anatomy suggests they may in fact be only distantly related. A similar conclusion might be reached regarding the predominantly unilacunar, one-trace nodes of Ericaceae; however, the report by Philip- son and Philipson (1968) of trilacunar nodes in Rhododendron gives cause for re-evaluation. PETIOLE VASCULARIZATION An attempt to define the range of variability in petiolar anatomy of Dil- leniaceae disclosed the following major patterns: Species with Unilacunar Nodes (1:1). (1) A single, slender, unbranched trace enters the lamina: numerous hibbertias (Fic. 27). Species with Trilacunar Nodes (3:3). (1) Traces fuse and form a flattened arc: Hibbertia quadricolor. (2) Traces fuse and form “V” shaped arc: Hibbertia coriacea (FIc. 17): (3) Traces fuse and form cylindrical, flattened, or concave vascular ring, either confluent or slightly dissected: Doliocarpus major; D. olivaceus; * Hib- (4) Traces form a closed cylindrical ring with one or more medullary bundles produced by invagination: Hibbertia lucida (Fic. 13). ; (5) Traces form an abaxial arc of fused or dissected collateral bundles with a separate adaxial trace derived from the inrolling and/or division of the lateral bundles. The adaxial trace may be lost in the lamina: Acrotrema * Nodes were not examined. 1969 | DICKISON, DILLENIACEAE, IV 389 sp.; A. bullatum; A. gardneri; A. lanceolatum; A. uniflorum; A, walkeri; Hibbertia banksii; H. pancheri; H. wagapii: Tetracera boiviniana; T. masuiana Traces form an abaxial arc of fused or dissected collateral bundles with a separate adaxial trace derived from division of the median bundle, The adaxial trace may be lost in the lamina: Hibbertia scandens; Tetracera indica. = Nn — Species with Multilacunar Nodes (five to many traces from an equal number of gaps). (1) Traces remain free forming a ring of widely dissected collateral bundles (bundles are often of unequal sizes): Acrotrema costatum;* Didesmandra (Fic. 15); Dillenia excelsa; D. luzoniensis; Schumacheria angustifolia. (2) Traces fuse to form confluent or only slightly dissected ring, often “U” or “V” shaped in outline: Davilla (Fic. 22); Dillenia bolsteri (Fic. 19); D. eximia; D. indica; D. ovata; D. pentagyna; D. salomonensis; D. suf- fruticosa; D. turbinata., Traces fuse to form confluent or only slightly dissected ring with an arc (rarely superimposed) of fused or dissected medullary bundles: Curatella americana (Fics. 10A,B,C); Dillenia alata; D. beccariana (Fic. 20); D. castaneifolia (Fic. 18); D. megalantha; D. papuana; D. philippinensis; D. reifferscheidia. Traces form an abaxial arc of fused or dissected collateral bundles with an adaxial enclosed siphonostele. The adaxial ring may subsequently open laterally or invaginate to produce additional free bundles: Doliocarpus coriaceus; D. dentatus; D. guianensis (Fic. 21); D. rolandri. ~~, Ww wa “-o-~ ae le It is evident that petiole vascularization in Dilleniaceae is quite diverse both between and within genera. In the present study, subtle deviations in vascularization pattern, the general outline of the vascular cylinder, and small, adaxial, subsidiary wing traces were ascribed little importance. Despite acknowledged incompleteness, I feel the descriptions outlined above will prove useful in future comparative studies relating to the family. It should be emphasized, that the descriptions presented are not based entirely upon observations from a single “characteristic” region. Wher- ever possible, sections were examined throughout the petiole and midrib aS suggested by Howard (1962). The importance of determining the 390 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 liocarpus, D. major and D. olivaceus are distinguished by their pubescent ovaries and fruits. An absence of medullary bundles in the petiole was also found to separate these taxa readily from all other species examined in the genus. Hunter (1966) considers Doliocarpus rolandri Gmel. to be a synonym of D. major Gmel. I have studied a collection from Brazil cited as D. rolandri (Pires & Cavalcante 52254, us) and found the petiole to possess medullary bundles, a character which is not encountered in D. major. A re-examination of this genus taxonomically might yield addi- tional basis for separation of the species. Other specific variation is found in Acrotrema (where A. costatum, from Thailand and Malaya, is quite dis- tinct from the Ceylonese species), Hibbertia, and Dillenia, though much more material must be studied before the true value of these data can be realized. Petiole structure cannot be used to separate the Dilleniaceae into sub- families or tribes. Moreover, there is little or no correlation between petiole vascularization and nodal anatomy. When considered as a whole, the vascular pattern in species with trilacunar nodes cannot be considered more primitive than that in species with multilacunar nodes. In plants with multilacunar nodes, petioles with widely dissected cyl- inders tend to be correlated with slender venation lacking massive bundle tive condition in multilacunar dillenias. Subsequent evolutionary pro- gression has produced fusion of traces and the formation of more complex medullary bundle patterns. These specializations have apparently oc- curred more than once, since the same apparent trends are also evident in species with trilacunar nodes. These ideas of nodal and petiolar evolution in Dilleniaceae do not agree with the conclusions of Decker (1967) who worked on the Luxemburgieae (Ochnaceae). Within the Luxemburgieae, Decker considers the multila- cunar node more primitive than the trilacunar, and petioles with numer- ous, unfused bundles (some of which may be medullary), more primitive than petioles with fused traces devoid of medullary bundles. In view of the frequent derivation of the Ochnaceae from the Dilleniaceae, such contrasting opinions are of special interest. A foliar character of debatable morphological derivation is the presence of petiolar wings in some Old World dilleniaceous species. Hooglan (1952) attaches taxonomic importance to the presence of completely am- plexicaul petiolar wings in certain species of Dillenia. Morphologically, these wings are frequently considered to be stipules. Hoogland (/oc. cit. does not accept this interpretation for the following reasons: (1) there is often no sharp distinction between the petiolar wings and lamina, (2) stipules of the usual morphological type do not occur in the dillenias, (3) in caducous wings, separation from the petiole begins from the base of the petiole and not from the apex as one would expect, and (4) Ozenda (1949) describes the wings as being weakly vascularized in contrast to the situation 1969 | DICKISON, DILLENIACEAE, IV 391 in the Magnoliaceae where the stipules receive a separate trace from the cauline stele. I have found vascularization of the wings in Dillenia to vary from weak (e.g., D. albiflos) to rather strong (e.g., D. philippinensis: D. suffruticosa). In the latter case the venation is highly reticulate. In either situation the wings are never supplied by independent traces from the cauline stele. Although I do not have any original interpretation for these structures, they do not appear comparable to true stipules. VASCULARIZATION OF THE LAMINA Major Venation. Although the prevailing type of major foliar vena- tion in the Dilleniaceae is pinnate, with the secondary veins proceeding to the margin of the blade, wide variation in leaf size, shape, and vasculariza- tion is encountered in the genus Hibbertia. A study of leaf vasculature in this genus showed that three basic venation patterns can be recognized: (1) pinnate leaves in which the numerous, strong, parallel, lateral veins extend diagonally outward from the midvein toward the margin of the lamina where they are interconnected by curved peripheral venation (Fic. 43); (2) pinnate leaves in which the principal lateral veins are fewer in number, irregular in their occurrence, more tenuous, and tend to sweep upward upon departure from the midvein (Fics. 42, 44); and (3), a pat- tern where two or more strong, terminal, lateral veins reflex back after de- parture from the midrib to terminate, often very massively, at the leaf ase. A varying number of prominent lateral veins may connect the mid- rib with the reflexed lateral (Fic. 49). This specialized venation pattern is exclusively associated with those hibbertias with reduced, needle-like leaves. The physiological significance of this type of vasculature is not clear. Concomitant in Hibbertia with a general trend toward reduction in leaf size as a response to xerophytic conditions, is a trend in reduction of leaf vascularization. Theoretically, this specialization commences with the pro- mination of the midrib was also observed in the mature leaves of the family. 392 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 The first-formed seedling leaves of Dillenia indica, with their strong, paral- lel, pinnate veins and serrate margins are sharply distinguished from the cotyledons (Fic. 41). Minor Venation. In addition to noteworthy features of major vena- tion, the pattern and diameter of the minor veins, in association with bundle sheathing, is often of diagnostic and perhaps of taxonomic signifi- cance in the Dilleniaceae. The occurrence of bundle sheaths around the veins is almost a universal feature of dilleniaceous leaves. Sheathing is noticeably absent only in some hibbertias and Acrotrema. When present, the sheath cells are either unlignified and parenchymatous in nature, or lignified, pitted, sclerenchy- matous elements. Parenchymatous sheaths typically surround both the major veins and terminal veinlets. These sheaths usually consist of cells elongated parallel to the vascular bundles; however, occasionally they become considerably lobed and oriented at right angles to the veins (Fic. 33). Sclerotized bundle sheathing is recognized by the presence of lignified, extensively pitted cells. When sclerenchyma occurs, it may form massive sheaths over the veins and veinlets as in Hibbertia (Fic. 30), Tetracera (Fic. 32), and some species of Doliocarpus. The formation of sclerified bundle sheaths enclosing the terminal tracheids is an uncommon feature in di- cotyledonous leaves (Esau, 1965). Of more frequent occurrence in the family is sclerenchyma around the major veins, but with veinlets devoid of sheathing or possessing only an incomplete sheath. The most striking pattern is seen in Hibbertia banksii where the mature leaves exhibit an interrupted sclerenchymatous sheath (Fic. 29). arenchymatous sheath cells were observed in Curatella (Fic. 31) and all species of Dillenia (Fic. 23), except D. philippinensis and D. reiffer- scheidia where pitted elements are found. Also, the presence of lobed parenchymatous sheathing around the terminal veinlets in Doliocarpus dentatus (Fic. 33) and D. rolandri readily distinguishes them from all other species of the genus. The variation present in Doliocarpus in the node, petiole, and minor venation warrants further study. Distinctions can also be made between genera and species on the basis of the diameter of veins and veinlets. Very slender venation is present in Acrotrema, Didesmandra (Fic. 26), Schumacheria, and some hibbertias (H. scandens, H. dentata, H. tetrandra, etc.). Associated with slender vascularization is weak bundle sheathing or its complete absence. Only in Hibbertia is massive venation sometimes devoid of sheathing (Fic. 28). There appears, nevertheless, to be in the family a rather distinct trend toward increased vein size accompanied by intensification of the amount of vein sheathing. A restricted trend was observed in Dillenia toward the formation of vein islets devoid of free vein endings. It is possible to trace this progres sion from species with slender veins and numerous free vein endings (e.g. 1969] DICKISON, DILLENIACEAE, IV 393 D. salomonensis and Fic. 23) through species with an intermediate pat- tern (e.g. D. quercifolia, D. ovalifolia, D. nalagi) to a pattern illustrated by D. papuana (Fic. 24) where free vein endings are scarce. The terminal condition in this sequence is seen in the massive, closed venation of D. schlechteri (Fic. 25). A taxonomic correlation of minor venation patterns in Dillenia is illus- trated by similar closed venation types occurring in D. papuana, D. cyclo- pensis, and D. schlechteri, all of which are considered closely related by Hoogland (1959) on the basis of floral structure. The only other species which were observed to possess comparable vasculature were D. beccariana, and D. turbinata, On the basis of leaf venation, I was not able to segregate Wormia as a distinct genus from Dillenia. The leaves of the Dilleniaceae appear to display a rather distinct phylo- genetic trend of specialization toward more massive vascularization, ac- companied by an increase in bundle sheathing. The same fundamental trends have also been described for the Winteraceae (Bailey & Nast, 1944). When the venation pattern, size, type, and degree of bundle sheathing, as well as petiole vasculature, are considered together, they offer excellent diagnostic leaf characters at the family, genus, and in some instances, species level. Additional material in all stages of maturity will have to be examined to understand fully the taxonomic significance of this infor- mation. TERMINAL IDIOBLASTS The occurrence of specialized terminal-veinlet elements in several wide- ly diverse dicotyledonous families has been well established. In a recent review of the literature, Tucker (1964) describes their presence in the Magnoliaceae. The occurrence of terminal idioblasts is now reported for the first time in the Dilleniaceae. ; Specialized terminal cells were observed only in relatively few species of Hibbertia. The diversity in vein endings is thus in accordance with variation in leaf shape and venation. The terminal cells are all of the basic tracheoid type (see Foster, 1956). Employing the classification of Tucker (loc. cit.), one can recognize tracheoidal elements, viz. scalari- form or scalariform-reticulate pitted cells, and dilated tracheids. Leaves of H. scandens (cult. K, s.n.) and H. dentata (cult. K, s.m.) were found to contain terminal elements which closely resemble tracheary cells in general morphology. The elements in H. dentata (Fic. 36) tend to occur singly and have exclusively scalariform pitting whereas the idioblasts of H. scandens (Fic. 35) often occur in clusters where they assume more irregular shapes and have reticulate pitting. The terminal elements of Hibbertia pachyrhiza (C. L. Wilson 861 j Fic. 38), although of the basic tracheoid type, differ considerably in their morphology. The latter cells are thick-walled, pitted to a much less degree, and are characteristically spherical in outline. In comparison with sur- rounding parenchyma these elements are significantly larger (75-140 » in 394 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 diameter). They occur singly or in clusters of three to four on each vein ending. Terminal elements in Hibbertia huegelli (C. L. Wilson 777), H. mono- gyna (Maiden s.n.), H. elata (Ingram 19852), H. billardieri (Clemens 42584a), and H. linearis (White 8580) tend to be intermediate between the elongated tracheoid element and the spherical one (Fic. 37). Dilated tracheids were found in Hibbertia scandens (Fic. 39), H. den- tata, H. nymphaea (Morrison s.n., a), and H. amplexicaulis (Pritzel 531). In H. scandens and H. dentata they were often in the same leaf with tracheoidal elements, and several veins were noted where the two types were present at the same vein endings. Generally, however, dilated tracheids seem to occur rather sparsely throughout the leaf and are not present at every vein terminus. The taxonomic usefulness of terminal idioblasts in the Dilleniaceae appears limited in view of their rather infrequent occurrence. Phylo- genetically it is of interest that the most diverse vein endings are found in Hibbertia, which on the basis of other criteria, is considered rather primitive. A similar situation has been reported in the Magnoliaceae (Tucker, loc. cit.). The full phylogenetic value of terminal idioblasts still remains to be developed; however, the trend toward the formation of specialized terminal cells appears to be a distinct one, subsequently lead- ing toward the reduction in size of the elements and in the amount of pitting on the wall surface. SUMMARY A comprehensive study of nodal and leaf vascularization in Dillenia- ceae has led to the following fundamental conclusions: cotyledonary node is unilacunar two-trace or unilacunar one-trace. Evidence from nodal anatomy appears to discredit a close relation- ship between the Dilleniaceae and Theaceae. (2) The petiolar anatomy of the family shows considerable diversity. De- scriptions of major venation patterns reveal that, in general, vascular cylinders composed of widely dissected bundles are more primitive than petioles with fused bundles and more complex medullary traces. (3) The vascularization of the lamina displays fundamental phylogenetic trends of specialization in both major and minor venation. Bundle sheath cells are either parenchymatous or sclerenchymatous and may enclose the terminal tracheids. Slender venation patterns lacking bundle sheathing are less specialized than coarser-veined leaves with massive bundle sheathing. (4) When considered together, nodal anatomy and foliar vasculature are of excellent diagnostic value and frequently of taxonomic and phylo- genetic significance in the Dilleniaceae. 1969 | DICKISON, DILLENIACEAE, IV 395 (5) The presence of specialized terminal idioblasts in the leaves of Hib- bertia is a character of which the importance is yet to be determined. MATERIAL EXAMINED Acrotrema sp. Ceylon: Thwaites CP3899 (us). A. bullatum Thw. Ceylon: Thwaites CP239 (us), A. costatum Jack. Thailand: Smitinand 2999 (us). A. gardneri Thw. Ceylon: Thwaites CP253 (us). A. lanceolatum Hook. Ceylon: Thwaites CP2660 (us). A. uniflorum Hook. Ceylon: Thwaites CP1014 (us). A. walkeri Wight. Ceylon: Thwaites CP694 (us). Curatella americana L. Brazil: Irwin 5470 (Ny); Nicaragua: Van der Sluijs s.n. (preserved material). Davilla aspera (Aubl.) Benoist. Trinidad: Howard 10502 (cu); Brazil: N.T. Silva 16. D. multiflora (DC). St. Hil. Panama: Dodge & Allen 17360 (mo). D. rugosa Poir. Brazil: A. de Mattos Filho s.n. (preserved material). Davilla sp. Brazil: Irwin 5570 (Ny). Didesmandra aspera Stapf. Sarawak: Burtt & Woods B.2540 (gE); S.18297 (sar); Native collector (sar) s.2. (preserved material). Dillenia alata (R.Br. ex DC.) Mart. New Guinea: P. van Royen 4677 (A, US). D. albiflos (Ridl.) Hoogl. Malaya: Corner SING F 29369 (a). D. beccariana Martelli. Sarawak: SAR 16272 (a). D. biflora (A. Gray) Martelli ex Dur. & Jacks. Fiji: Gillespie 2182 (GH); A. C. Smith 8762 (us). D. bolsteri Merr. Philippines: Wenzel 3112 (cH). D. castaneifolia (Miq.) Martelli ex Dur. & Jacks. New Guinea: Womersley NGF 3768 (a). D. cyclopensis Hoogl. New Guinea: van Royen & Sleumer 5812 (a). D. excelsa (Jack) Gilg. North Borneo: Ramos 1379 (A). D. eximia Miq. Borneo: NIFS bb 16830 (A). D. indica L. Australia: Cult. BRI 5.2. (preserved material); India: Sastri s.n. (preserved material); Cult. E C4388. D. luzoniensis (Vidal) Martelli ex Dur. & Jacks. Philippines: J. V. Pancho s.n. (preserved material). D. megalantha Merr. Philippines: Quezon. M. Q. Lagrimas s.n. (preserved material). D. monantha Merr. Philippines: Herre 1010 (a). D, montana Diels. New Guinea: Hoogland & Pullen 6265 (a). D. nalagi Hoogl. Papua: Hoogland & Taylor 3438 (A). D. ochreata (Miq.) Teysm. & Binn. ex Martelli. Celebes: NJFS bb 18085 (a). D. ovalifolia Hoogl. New 16655 (a). D. salomonensis (White) Hoogl. Solomon Islands: Walker & White 145 (a). D. schlechteri Diels. New Guinea: Womersley & Millar NGF 7000 (A). D. suffruticosa (Griff.) Martelli. Singapore: Cult. sinc s.2. (preserved material) ; Canright 978 (asv). D. turbinata Fin. & Gagnep. Hainan: How 72058 (A). Doliocarpus coriaceus (Mart. & Zucc.) Gilg. British Honduras: Gentle 2892 (us); Colombia: Cuatrecasas 16556 (us). D. dentatus (Aubl.) Standl. Bolivia: Krukof 396 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 10088 (uc). D. guianensis (Aubl.) Gilg. Surinam: uc 947180. D. lasiogyne Benoist. Brazil: Klein 1.281 (us). D. major Gmel. Panama: von Wedel 2860 (mo); J. M. Johnston 1694 (mo). D. olivaceus Sprague & Williams ex Stand. Panama: Stern et al. 11 (us). D. rolandri Gmel. Brazil: Pirés & Cavalcante 52254 (us). a acicularis (Labill.) F. Muell. Australia: Queensland. Clemens aa (A); C. T. White 9466 (a). H. altigena Schlechter. New Caledonia: H. S. M 3709 (aA). H. amplexicaulis Steud. Australia: Pritzel 531 (a). H. aspera De. aseetiga New South Wales. Constable 42837 (a). H. aurea Steud. Aus- C. L. Wilson 843 (us). H. aun oe Australia: Aston 359 ©. “HL. banksti Benth. Papua: L. J. Brass 1 (a). H. baudouinii Brongn. & Gris. New Caledonia: tee aril al ie? (a). H. bracteata (R.Br.) Benth. Australia: New South Wales, C. T. White 5012 (a). H. billardieri F. Muell. Australia: Queensland. Clemens 42584a (us). H. brongniartii Gilg. New Cale- donia: Thorne 28580 (rsa). H. cistiflora Wakefield. Australia: New South Wales. Helms 1290 (a). H. cistifolia R.Br. Australia: Specht 843 (us). H. coriacea (Pers.) Baill. Madagascar: Humbert 5866 (us). H. crenata Andr. Australia: C. L. Wilson 851 (us). H. cuneiformis (Labill.) Gilg. Australia: Cult. K, s.1. (pre- served material); E. H. Wilson 297 (us). H. dealbata Benth. Australia: Specht Australia: Royce 5760 (us). H. ebracteata Bur. ex Guillaum. New Caledonia: H. S. McKee 3697 (a). H. elata Maiden & Betche. Australia: New South Wales. Ingram 19852 (us). H. exutiacies Wakefield. Australia: Eichler 17965 (av). H. fasciculata R.Br. ex DC. Australia: Aston 387 (a). H. furfuracea Benth. Aus- ia: C. T. White 5382 Hi: mane Diels. Australia: C. L. Wilson 856 (us). H. pi eee F. Muell. Australia: Perry 5379 (us). H. gracilipes Benth. Aus- tralia: Royce 5792 (us). H. huegelli F. Muell. Australia: C. L. Wilson 777 (us). H. hypericoides (DC.) Benth. Australia: E. H. Wilson 454 (a). H. pos i Benth. Australia: C. L. Wilson = pee H. linearis R.Br. ex DC. Australia Australia: C. L. Wilson 740 (us). H. microphylla Steud. Australia: C. T. White 5317 (A). H. miniata Gard. Australia: C. L. Wilson 782 (us). H. monogyna R.Br. ex DC. Australia: New South Wales. J. H. Maiden s.n. (GH). H. montana C. L. Wilson 861 oom i. pancheri (Porch. & sie Briq. New Caledonia: Thorne 28585 (rsa). H. patula Guillaum. New Caledonia: H. S. McKee 3543 (A). H. procumbens DC. Australia: Long 209 (a). H. paces geo New 1969} DICKISON, DILLENIACEAE, IV 397 H. scandens (Willd.) Dryand. Australia: Cult. prt, s.2. (preserved material); K, sm. (preserved material). H. sericea (R.Br.) Benth. Australia: Muir 855 (A). H. serrata Hotchkiss. Australia: C. L. Wilson 855 (us). H. stirlingii C. T. White. Australia: C. L. Wilson 757 (us). H. stricta (DC.) R.Br. ex F. Muell. Australia: Hoogland 8420 (cans). H. subvaginata (Steud.) Ostenf. Aus- tralia: C. L. Wilson 764 (us). H. tetrandra (Lindl.) Gilg. Australia: C. L. Wil- son 848 (us); Cult. K, s.2. (preserved material); Cult. £, C3544. H. tomentosa R.Br. Australia: Specht 638 (A). H. tontoutensia Guillaum. New Caledonia: McMillan 5060 (a). H. trachyphylla Schlechter. New Caledonia: Hiirlimann 846 (A). H. uncinata (Benth.) F. Muell. Australia: E. H. Wilson 155 (a). H. vaginata (Benth.) F. Muell. Australia: C. L. Wilson 859 (us). H. vestita A, Cunn. Aus- ia: New South Wales. NSW 55998 (a). H. wagapii Gilg. New Caledonia: Thorne 28266 (GH). Pachynema dilatatum Benth, Australia: Northern Territory. NT 6129. P. jun- ceum Benth. Australia: Northern Territory. NT 6750. Schumacheria castaneifolia Vahl. Ceylon: Abeywickrama s.n. (preserved ma- terial). S. angustifolia Hook. f. & Thoms. Ceylon: us 597415. Tetracera akara (Burm. f.) Merr. Borneo: Elmer 21314 (a). T. arborescens Jack. Sumatra: Toroes 5293 (a). T. asiatica (Lour.) Hoogl. Hainan: Lau 3875 (a). T. asiatica (Lour.) Hoogl. ssp. asiatica Hoogl. China: Liang 69507 (a). T. boiviniana Baill. Tanganyika: Tanner 2548 (uc). T. daemeliana F. Muell. Aus- material). T. volubilis L. Mexico: Purpus 7647 (MO). LITERATURE CITED Bartey, I. W. 1956. Nodal anatomy in retrospect. Jour. Arnold Arb. 38: 269- 287 ——, & R. A. Howarp. 1941. The comparative morphology of the Icacin- aceae. I. Anatomy of the node and internode. Jour. Arnold Arb. 22: 125-132. —, & C. G. Nast. 1943. The comparative morphology of the Winteraceae. II. Ca : ; rb. 24: 472-481. : —— & cali ve er fanharscntn morphology of the Winteraceae. IV. Anatomy of the node and vascularization of the leaf. / bid. 25: 216-221. — Barton, H. 1866-67. Observations sur l’anatomie des Dilléniacées. Adansonia : 88-93, ———.. 1871. The Natural History of Plants. Vol. 1. (Transl. by M. M. Harteg). L. Reeve & Co., London. : Benzinc, D. H. 1967a, Developmental patterns in stem primary xylem of — Ranales. I. Species with unilacunar nodes. Am. Jour. Bot. 54: 805-815. 398 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 . 1967b. Developmental patterns in stem primary xylem of woody Ra- nales. II. Species with trilacunar and multilacunar nodes. /bid. 813-820. Canricut, J. E. 1955. The comparative morphology and relationships of the Magnoliaceae. IV. Wood and nodal anatomy. Jour. Arnold Arb. 36: 119-140. Corpemoy, C. J. pe. 1859. Note sur les ovules de deux genres de Dilléniacées. Bull. Soc. Bot. France 6: 409-411, 449-450. DEcKER, J. M. 1967. Petiole Ve of Luxemburgieae (Ochnaceae). Am. Jour. Bot. 54: 1175-1 Dicxison, W. C. 1967a. eee Pon aa studies in Dilleniaceae. I. Wood anatomy. Jour. Arnold Arb. 48: 1-29. . 1967b. Comparative neon ae ra studies in Dilleniaceae, II. The pollen. /bid. 231-240. Esau, K. 1965. Plant Anatomy, 2nd ed. John Wiley & Sons, Inc., New York. Foster, A. S. 1956. Plant idioblasts: remarkable examples of cell specialization. Protoplasma 46: 184-193. Gite, E. 1893. Dilleniaceae. Nat. Pflanzenfam. III. 6: 100-128. HiITzeMANN, C. 1886. Beitrage zur vergleichenden Anatomie der Ternstroemia- ceen, Dilleniaceen, Dipterocarpaceen und Chlaenaceen. (Inaug. Diss.) Univ. Kiel. cageager R. D. 1952. A revision of the genus Dillenia. Blumea 7: 1-145. 959, Additional notes on Dilleniaceae 1-9. Ibid. 9: 577-589. How en A. 1962. The vascular structure of the petiole as a taxonomic aoe. Jn: Garnaud, Advances in horticultural is and their appli- cations. Vol. III. pp. 7-13. Pergamon Press, New York. Keno, H. 1962. Comparative morphological studies in Theaceae. Univ. Calif. Publ. Bot. 33: 269-384. Marspen, M. P. F., & I. W. Battey. 1955. A fourth type of nodal anatomy in dicotyledons, illustrated by Clerodendron trichotomum Thunb. Jour. Arnold Arb. 36: 1-51. MeeusE, A. D. J. 1966. Fundamentals of Phytomorphology. The Ronald Press Co., New York. Mercatre, C. R., & C. CHALK. 1950, Anatomy of the Dicotyledons. 2 Vols. The Clarendon Press, Oxford. Nampoonrrr, K. K., & C. B. Beck. 1968b. A comparative study of the primary vascular system of conifers. II. ea with opposite and whorled_ phy!l- lotaxis. Am. Jour. Bot. 55: 458-4 OzenpA, P. 1949. Recherches sur i aiid: apocarpiques. Publ. Lab. YEcole Normal Supérieure, Ser. Biol. Fasc. II. Paris. Pant, D. D., & B. MenRa. 1964. Nodal anatomy in retrospect. Phy ‘tomorphology 14: 384-387. PaRMENTIER, M. P. 1896. Contribution a l’étude de la famille des Dilléniacées. Compt. Rend. Assoc. francaise. Avanc. Sci. [Sess. 24] 1895. pt. 2: 626-630. Puivipson, W. R., & M. N. Puriipson. 1968. Diverse nodal types in Rhodo- dendron. Jour. Armold Arb. 49: 193-224. Stnnott, E, W. 1914. Investigations on the phylogeny of the angiosperms. * The anatomy of the node as an aid in the classification of the angiosperms. Am. Jour. Bot. 1: 303-322. SOLEREDER, H. 1908. Systematic Anatomy of the Dicotyledons. (Engl. transl. by oodle & Fritsch.) Vol. I. Oxford Univ. Press, London mete H. 1895. Beitrage zur vergleichenden anatomie der Dilleniaceen. Beih. Bot. Centralbl. 62: 337-342, 369-378, 401-413. 1969] DICKISON, DILLENIACEAE, IV 399 Tucker, S. C. 1964. The terminal idioblasts in magnoliaceous leaves. Am. Jour. Bot. 51: 1051-1062. DEPARTMENT OF BIOLOGY VIRGINIA POLYTECHNIC INSTITUTE BLACKSBURG, VIRGINIA 24061 EXPLANATION OF PLATES PLATE . Fics. 1-8. Dilleniaceae, nodal anatomy. Tetracera indica (seed received from H. Keng, Sin gapore), transverse Ae of cotyledonary node showing unilacunar range condition (c.t., cotyledonary trace), X 32. 2, Hibbertia pungens (Royce 7640), transverse section of unilacunar node, X 32. 3, Tetra- cera boiviniana (Tanner 2548), transverse section of trilacunar node illustra ating widely separated gaps, X 13. 4, Acrotrema sp. (Thwaites CP3899), transverse through section through rhiz f leaf trace (It) an entitious root de- parture, e scandens (cult. BRI, s.m.), transverse section of node illustrating trilacunar condition with widely separated gaps t pet- , ' vill gosa (de Mattos Filho s.m.), tranverse section of pentalacunar node, . 7, Schumacheria castaneifolia (Abeywi a S$.M.), ransverse section of weorree node. Note the rap trace departs paint aii (It, leaf trace), 13, illenia ovata (cult. SING, 5.7.), transvers: PLATE II Figs. 9, 10. Dilleniaceae, petiolar anatomy. 9A, B, C, D, Acrotrema (Thwaites CP3899), transverse sections of the petiole and midrib illustrating omaha n o abaxial arc and adaxial bundle, &* 32. 10A, B uratella amer. chia oe rw 5470), eens sections of petiole illustrating formation of ull vee 7 beanies, xX 1 transverse section of petiole at e amina showing same medullary indies Craps d by arrows), complete fusion of vascular cylinder, and extraxylary fibers, X 30. PLATE I Fics. 11-14. Dilleniaceae, pears nd Bee anatomy. All figures < 30. 11A verse section of petiole at base of lamina showing bundles by invagination. 14, H. patula (McKee 3543), “transverse section petiole at base of lamina depicting confluent — cylinder. Note ae ber PLATE IV Fics. 15-22, Dilleniaceae, petiolar anatomy. 15, Didesmandra aspera (Sara- wak, s.m.), transverse section of transverse section of petiole at base 0 s ected ose @) 32 ‘ ey coriacea (Humbert pon trans- verse section of petiole at base 0 g vascular tissue, < of lamina 18, Dillenia "castoneifotia: (B (Womersley N GF 3768), transverse section of. petiole at base of lamina showing arc of medullary bundles, X 15. 19, D. bolste 400 * JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ge 3112), transverse section of Sag aes at base of lamina rapes confluent — S pte “sa vascular cylinder, x 3 beccariana (SAR 16272), trans- verse section of petiole at base Ste ‘showing superimposed medullary bu a Gndicated by arrows), X 17. 21, Doliocarpus guianensis (uc 947180), transverse section at base of petiole showing abaxial arc of dissected bundles and adaxial siphonostele, X 17. 22, Davilla aspera (Howard 10502), transverse section of petiole at base of lamina 7 ai nearly complete vascular cylinder. Note abundant sclereids in cortex, & 3 PLATE V . 23-26. Dilleniaceae, minor venation. All figures X 25. 23, Dillenia . Q. Lagrimas not ley & Millar NGF 7000), note complete absence of free vein endings accom- panied by massive venation and bundle sheathing. Sheath cells extend into vein islets. 26, Didesmandra aspera (SAR S.18297), note weak, slender venation and incomplete bundle sheathing. PLATE VI Fics. 27-30. Dilleniaceae, minor venation. 27, Hibbertia pepe (Eichler 17965), a2 of leaf showing single leaf oe and termination of reflexed lateral veins, 5. 28, H. subvaginata (C. L. Wilson 764), note massive cya lack- ing bundle sheathing, X 35. 29, H. b anki ew 8431), venation showing char- acteristic interrupted oo sheat 25. 30, H. wagapii (Thorne 28266), note terminal veinlets are ee enclosed by i es oer bundle sheathing, « 2 PLATE VII Fics. 31-34. Dilleniaceae, minor venation. 31, Curatella americana ({rwin 5470), minor venation showing abundant parenchymatous bundle sheathing, X 25. 32, Tetracera macrophylla (Canright 1127), note ctr See bundle sheathin ng completely surrounds terminal tracheids, X 25. 33, Doliocarpus den- tatus se f 10088), terminal veinlet with abundant parenchymatous ape ing. sheath cells often orientated at right angles to vein, x 54. Pachynena aia (NT 6129), scale-like leaf eae by weak jae traces, X 5 PLATE VIII isc 35-39. Terminal veinlet — in mtg a. 35, B. scandens Co s.n.), X 130. 36, H. dentata (cult. K, s.n.), X 1 7. fi. “nuegelli cL son ihy, x 100. 38, H. pachyrhiza (C. L. Wilson a S130. 39, H. scandens (cult. K, s.m.), K 1 PLATE IX Fics. 40-44. Dilleniaceae, major venation. 40, Dillenia indica (seed received from H. Keng, rin ata pegs donary node and vascularization n of cotyle edon. 41, the — vasculariza of first foliage ge nd leaf. 42, route age ta (cult. K, $.m.), natur. “ cae 43, ‘ tontoutensia (McMillan 5060), x1 4, H. cisnesformis (Wilson 297), X 2 PLATE X f Fics. 45-50. Leaf vascularization in Hibbertia. Due to i apeceteem ps — size, magnifications 4. H. huegelli (C. L. Wilson 777), X 3.5. . 46, H. mo yna (Maiden ns. ; 41, H. nitida (Fl. Novae Holl, 141), % §. 48, H. vestita (NSW 55098), x 13. 49, H. exutiacies (Eichler 17965), x > oe H. see tas tase 387), X Jour. ArNotp Ars. VoL. 50 PLaTE I Dick1son, DILLENIACEAE, IV Pirate II Jour. ARNOLD Ars. VOL. 50 DickIson, DILLENIACEAE, IV * Jour. ARNOLD Ars. VoL. 50 12 Dick1son, DILLENIACEAE, IV Prate Iii Jour. ARNOLD ArB. VOL. 50 PraTE IV Dickison, DILLENIACEAE, IV Jour. ARNOLD Ars. VoL. 50 PLATE V DickIson, DILLENIACEAE, IV aot Jour. ARNOLD Ars, VoL. 50 PraTe VI Dickison, DILLENIACEAE, IV Jour. ARNOLD Ars. VoL. 50 Piate VII DickIsON, DILLENIACEAE, IV | | t Jour. ARNOLD Arp, VoL. 50 PiaTeE VIII Dickson, DILLENIACEAE, IV fis tae nna Jour. ARNOLD Ars. VoL. 50 Dick1son, DILLENIACEAE, IV PLaTE IX — 1969 | UHL, NANNORRHOPS RITCHIANA 411 ANATOMY AND ONTOGENY OF THE CINCINNI AND FLOWERS IN NANNORRHOPS RITCHIANA (PALMAE) ! NATALIE W. UHL THE LARGE, TERMINAL, compound inflorescence of Nannorrhops ritchiana (Palmae-Coryphoideae) is composed of unspecialized branch systems (Tomlinson & Moore, 1968) which may serve as a model for the deriva- tion of more specialized types of palm inflorescence. Observations on the inflorescence of Nannorrhops ritchiana are continued here with a descrip- tion of the anatomy and some aspects of the ontogeny of the rachillae, of the ultimate flowering units, and of the flowers. Nannorrhops is especially important because completely sheathing and vasculated bracteoles are present throughout the ultimate flowering unit. Detailed studies confirm Tomlinson and Moore’s tentative designation of this unit as a cincinnus and reveal basic constructional principles that apply to many, if not all, of the varied flowering units found in palms; e.g. the triad of a pistillate and two staminate flowers, where interpretation has been difficult because bracts are absent or lack vasculature (Uhl, 1966). The form and anatomy of the carpel may also illustrate some primitive features for palms. MATERIAL AND METHODS Inflorescence branches from plants at the Fairchild Tropical Garden, Miami, Florida, were available in various stages of development from the following collections: Moore 6009, Read 735, and Tomlinson 14.X1.63 and 14.X1.66. These were fixed in formalin-acetic acid-ethanol, desilicified for 1 to 2 weeks with approximately 1/3 commercial strength hydrofluoric acid, and embedded in Paraplast. Serial sections of flowers and rachillae were made at 5, 7, 10, and 15 microns and were stained with safranin and fast Sreen or safranin and aniline blue. Cincinni and flowers were also cleared as described previously (Uhl, 1966), the number of cleared flowers ex- amined exceeding 50. Two films were prepared for cinematographic anal- ysis (Tomlinson & Zimmermann, 1965) of rachillae and mature flowers. Some observations and photographs (Fics. 8-19) were made in polarized light. Since growth is continuous, but not uniform, dimensions of the ma- terial examined are included below. RACHILLAE Morphology. Structural patterns are simple despite the large size of the inflorescence in Nannorrhops (Tomlinson & Moore, 1968, Fig. 42). Up i *From work supported by National Science Foundation Grant GB-7758; principal investigator, Harold E. Moore, Jr 412 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 to five orders of branches are formed monopodially. The visible flower- bearing axes or rachillae are mostly branches of the fourth order, but whatever the order, they are similar in size and in the number of flower- clusters or cincinni produced. Rachillae taper slightly in diameter (from 1.5 mm. to 0.75 mm.) and are indeterminate in length and in potential number of flower clusters. Fully expanded rachillae in the material ex- amined range from 5 to 12 cm. in length and bear from about 20 to 45 cincinni. A few distal cincinni are usually abortive. Development. Maturation of flowering axes within the inflorescence is complex. Four different patterns can be recognized: one with reference to the inflorescence as a whole, a second in the sequence of development of lateral branches, a third on individual rachillae, and a fourth within each flower cluster. 4 Fics. 1-7. Fics. . 1-3, Three successive developmental stages of a third yee branch, am mee indicate sequence, further explanation in text. Fic. 1 immature upper with one petal and one stamen removed, X 5; Fic. 6, part t of a cleared rachilla to show bundles in —. pabcaslies stalks of first flowers of cincinni, 1969 | UHL, NANNORRHOPS RITCHIANA 413 Maturation is basipetal in the inflorescence as a whole. Upper first and second order branches produce the first flowers (Tomlinson & Moore, 1968, Fig. 40). The further expansion of specific lateral branches is not uniform but can be related to the order of the branch. Third order branches mature acropetally, but development of fourth order branches is irregular. Ma- ture flowers are produced on some fourth order axes when others are still in early stages of development as illustrated by Fics. 1-3. In early Stages of growth, some fourth order branches are equal in size to the main branch (third order axis) on which they are borne (Fic. 1). The result is a digitate configuration, which may be useful in interpreting digitate branching in mature inflorescences elsewhere in palms. In a later stage (Fic. 2) some of the fourth order branches have matured while others are still undeveloped. Flowers may mature irregularly on a specific rachilla. Those in cincinni at the middle of the rachilla often develop before those in cincinni nearer the base or the apex (Fic. 3), but in general the order of development is acropetal. Within each cincinnus there is still another acropetal series in the maturation of individual flowers (Fic. 4). Anatomy. The vascular system in rachillae is composed of a central or subcentral group of about 10 (8-12) large vascular bundles with a num- ber (ca. 13) of intermediate and smaller bundles peripheral to them (Fics. 8, 9, 12). The peripheral bundles represent strands which supply cin- cinni, and they vary in number and position depending on the proximity of the level examined to a cincinnus. Each large bundle has 1 to 4 large vessels (Fic. 12) and a complete fibrous sheath which is approximately 5 to 7 cells wide over the xylem. Most commonly there are two large ves- sels per bundle, but bundles about to branch have three large vessels and small branch bundles only one. The narrow cortex is of small unspecialized parenchyma 6 to 8 cells in width, and the epidermis is of smaller iso- diametric cells. BRACTS Morphology. Two types of bracts may be present on rachillae. On first, second, and third order branches there is usually an irregularly bi- carinate prophyll which is inserted basally in an adaxial position. This bract is commonly empty on first and second order branches but on third order branches it subtends the first lateral branch. Flower clusters are subtended not by prophylls but by irregularly fun- nel-shaped bracts with attenuate dorsal tips. Similar but larger bracts borne on one axis and subtending branches of the next order occur through- out the inflorescence and may be arranged in a reduction series from a foliage leaf (Tomlinson & Moore, 1968). Bracts subtending cincinni are the smallest of the series and are all equal in size and shape at maturity. On fully expanded rachillae, each bract is about 3 mm. long, the sheathing Part extending for ca. 2 mm. of this. 414 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 —13. Successive transections of a cincinnus, taken in peter i oS at leve origin of pet ae be ey on second flower, X 36; Fic. *': transection at level of o r trace to prophyll on sec cond flower, X 3 Fic. 12, transection of "Pchatle. An stalks of first, second, third, and fourth 1969 | UHL, NANNORRHOPS RITCHIANA 415 Anatomy. Each bract subtending a cincinnus is supplied by five vas- cular bundles and by a large number of fibrous strands (Fics. 6, 8,9). The vascular bundles, which may be designated as a midvein and two pairs of lateral bundles, originate as small branches of peripheral stelar strands. The continuing vertical bundles (Zimmermann & Tomlinson, 1965) from which the midvein and first pair of lateral bundles originate usually enter the stalk of the first flower. Vertical bundles providing the second pair of laterals, however, continue in the rachilla. Lateral vascular bundles branch and anastomose distally in the bract (Fic. 6). The numerous fibrous strands (Fics. 6, 8) are wide tangentially and also branch and anas- tomose distally. They are tapered somewhat proximally but are not con- nected to the vascular cylinder of the rachilla. FLOWERING UNITS Morphology. With the initiation of the first flower, the growth pattern of the inflorescence shifts from monopodial to a sympodial elaboration of clusters, each consisting of five or six successively younger flowers (Fic. 4). The bract on the rachilla subtends the first flower. The stalk of this flower in turn bears an adaxially situated bracteole which is completely sheathing and has two subequal adaxial tips, thus differing from the bract subtending the first flower and definable as a prophyll. The prophyll sub- tends the second flower of the cluster. The stalk of the second flower bears a similar bracteole which subtends the third flower. This pattern is repeated up to five or six times in Nannorrhops (Fics. 8-13). Each floral primordium is initiated on the opposite side of the appropriate floral stalk and at an angle of approximately 75°. Although five or six buds are present in the cluster, only three flowers usually mature. Because flowers are successively younger and pedicels elongate successively during maturation, the two-rowed condition of a cincinnus, though structurally Present, is not readily evident macroscopically. Left-handed and right- handed cincinni occur, depending on whether the second flower is initiated on the left or right side of the first floral axis. A specific rachilla usually bears predominantly left- or right-handed clusters — e.g. on a right-handed rachilla only one or two basal and one or two median cincinni are left- handed, Anatomy. Bundles which supply the first flower of a cincinnus origi- nate as branches of major axial bundles in the rachilla. The first such branch originates at about the level of insertion of the second cincinnus flowers, x 18; Fic. 13, transection through all flowers of a cincinnus, X 36 F . Detatts: br, bract subtending first flower of a cincinnus; fl 1 to fl 6, successive owers of a cincinnus; fs, fibrous bundles of bract; mv br, midvein of bract; lv, On axis of the first flower; pr 2, prophyll of second flower, arrow points to lower trace; pr 3, prophyll of third flower; pr 4 and pr 5, prophylls of fourth and fifth flowers respectively; ra, rachilla. 416 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 below. About three more branches are derived from axial bundles at high- er levels, and further branching of these provides the complete supply to the first flower. At the level of origin of the midvein of the subtending bract, this supply consists of a group of about 16 bundles. The exact number of bundles is somewhat subjective unless the level is carefully in- dicated, since bundles are frequently small, especially near their origin, and fibrous bundle sheaths are often confluent for some distance. Anatomically as well as morphologically prophylls are different from other bracts in the inflorescence. The main vascular complement of each prophyll is two vascular bundles, one supplying each tip (Fics. 9, 11, 13). These traces are derived as small branches of marginal stelar bundles of the floral stalk. The bracteole is obliquely inserted and irregularly bi- carinate, one tip being slightly longer than the other. The trace to the longer tip originates at a slightly lower level than that to the shorter (Fic. 11), and is a somewhat larger bundle which often branches distally. Un- connected fibrous strands are also present in the prophyll (Fic. 11) and occasionally a third vascular bundle (Fic. 11) is seen. Above the origin of the traces to the first bracteole, the stelar bundles of the first flower provide the vascular supply to the second floral stalk (Fics. 8, 9, 20). Two of the ensuing bundles produce small branches, each supplying one tip of the second prophyll (Fics. 10, 11), and the pat- tern is repeated until up to five or six floral primordia are formed (Fics. 8-13, 20). Thus anatomically each flower, its axis, and bracteole are identical to the others making up the cincinnus. Transections of the floral axes of the first, second, third, and fourth flowers may be compared in Ficures 10, 12, and 13 and their similarity noted. The pattern of origin of the vascular supply to each floral stalk is also similar as can be seen in Ficure 20 which is a camera lucida drawing of the major bundles in a cleared cincinnus. THE FLOWER AT ANTHESIS Morphology. Among the palms, approximately 165 genera are monoe- cious, about 39 are dioecious, and some 34 genera bear perfect flowers. Nannorrhops belongs among the last, having perfect flowers with three sepals, three petals, six stamens, and a tricarpellate gynoecium. Open flowers (Fic. 5) are approximately 6 mm. long. The sepals are 3 mm. long and are connate for two-thirds this length forming a sheath, above which the membranaceous tips are free. Petals are ca. 5 mm. long, ovate, somewhat fleshy, shortly imbricate near the base and then valvate. Stamen- filaments are wide and fleshy basally (Fics. 5, 18), but taper to the at- tachment of the versatile anthers which are subequal, basally divergent, and laterally dehiscent. The three carpels are free in young stages, but 1m mature flowers are connate by ventral faces through the ovarian and stylar regions. Thus at anthesis the gynoecium is syncarpous with definite eX- ternal grooves showing the limits of each carpel. Each carpel has a dis- tinct stalk, an ovoid fertile part, and a long attenuate style through which 1969 | UHL, NANNORRHOPS RITCHIANA of mature flowers, a in polarized light. Fic. 14, Fic Sections tangential Jongisction with two carpels, x 20; . 15, transection through sepal a ty) axis at level of origin of petal eae X 36; Fic. 16, transection through base of flower above Fic. 15 3 ; : Fic. 17, transection at higher level where Carpel stipes are distinct, < 36; Fic. 18, transe ction of ovarian part of ae he carpels connate, « 36; Fic. 19, transection of anthers and style, ETAILS: ca s, carpel stipe; mi, micropyle; pe, petal; se, sepal tube; st, stamen trace; sty, style. - 418 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 a locular canal extends to open distally. There is no connection (com- pitum, Carr & Carr, 1960) between locular canals of adjacent carpels. No definite stigmas are present. Papillose stigmatoid tissue is apparently pres- ent at anthesis around the stylar opening but is not developed until the flower opens. An anatropous ovule is attached ventrally and basally in each locule and is turned so that the micropyle is lateral rather than dorsal in respect to the funiculus. At anthesis about one-third the length of the flower consists of a tapered solid basal part (Fic. 5) sheathed by the sepal tube and representing the region of insertion of petals, stamens, and carpels. A very short petal- stamen tube surrounds the free carpel stalks (Fics. 14, 17) but since all organs are free just below the ovary in the mature flower, the short petal- stamen tube does not seem to justify the term “perigynous.” Anatomy. Floral anatomy in Nannorrhops ritchiana has been described by Morrow (1965) and Gupta (1960). The present study confirms most of the observations of these authors and provides further details of carpel anatomy, organogeny, and histogenesis. The general outlines of the floral vascular system can be seen in Fic. 7 which is a cleared preparation of the central part of a flower. Just below sepal insertion, bundles present in the floral stalk enlarge, extend peripherally, and branch forming a group of bundles which provide traces to the floral organs. Further details of this pattern are presented in a radial plot of one of the large axial bundles (Fic. 23). The pattern is irregular in that traces to floral organs are branches originating near the insertion of the organ or at a lower level. Gupta (1960) reports two rings of bundles in the floral pedicels: an outer of 11 or 12 and an inner of three larger ones. Morrow (1965) states that 9 (8 to 10) strands enter the base of the flower. Three central strands do mature first in floral stalks and are often larger (Fics. 12, 13). In mature pedicels both large and small bundles are present with a gradual transition in size. The number of bundles is somewhat subjective because of the difficulty of getting exactly comparable levels. In the material I studied, 10 to 15 bundles were present, five or six showing birefringent xylem (Fics. 10, 12, 13). Just below sepal insertion, larger bundles of the stalk extend toward the periphery, become larger, and branch (Fic. 7). Smaller bundles may fuse with larger ones or also branch. The floral stele, at the level of sepal insertion, consists, therefore, of about 20 to 25 medium to small bundles, arranged in a thick ring, the larger bundles toward the center. Fifteen (14 to 18) small sepal traces originate as branches of peripheral bundles of this stele. Sepal traces near their origin consist of a few sieve elements and two to four xylem elements and are very easily overlooked; but slightly higher in the sepal-tube, fibrous caps are present on these bundles and unconnected fibrous bundles are present between vascular strands. Thus a ring of approximately 28 bundles is present in a transection of the sepal-tube (Fic. 15). Five vascular bundles with four to six interspersed == 1969 | UHL, NANNORRHOPS RITCHIANA 419 TR 20 Fic. 20. Wash drawing of a cleared cincinnus, done with Wild MS stereo- microscope and drawing attachment, how major vascular rg to flowers. ike a two he fl 1 and fl 2, abscissed; younger flowers, 3 to fl 5, in bud; pr 420 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 fibrous strands represent the supply to each sepal, a midstrand and two lat- erals reaching the tip. There are six to nine traces, situated in a median row, in the base of each petal (Fic. 16). These may be stelar bundles extending directly into the petal or they may be branches of a stelar bundle. The number of traces varies slightly. Morrow (1965) reports three and Gupta (1960) five from receptacular strands and one from a perianth-stamen bundle. A single median procambial strand develops first in each petal followed by three strands at a later stage in ontogeny. The two laterals from this group of three divide very near the central stele and with the midvein and one or more small traces from the receptacle form the seven major strands (Fics. 16, 17). These often produce parallel branches at higher levels. Traces to stamens arise in two whorls in the same manner as petal sup- plies by the branching or direct conversion of a vertical bundle into a stamen trace. Antipetalous traces may arise as a branch of the same verti- cal bundle which formed the median petal trace, or as a branch or conver- sion of an adjacent bundle. Stamen traces are large bundles which divide in the base of the filament (Fics. 17, 18) into two traces which are oriented xylem to xylem in the filament with the phloems lateral in position, but which reunite in the distal part of the filament. In the receptacle below the gynoecium, about ten large vertical bundles (bright spots, Fic. 16) are arranged in a central ring with smaller strands external to them. Slightly higher, all stelar bundles are divided into three groups, one of which supplies each carpel stalk. Some 14 bundles are present in a close group in the lower part of each stalk. One of the larger bundles becomes the dorsal bundle of each carpel and the others form the lateral and ventral bundles. There are usually four major pairs of lateral bundles and two ventrals (Fic. 21). The latter may be distinguished by position and by their extension with the dorsal bundle higher into the style. Other small bundles are aligned along the ventral face of the locule and at anthesis extend about one-half the length of the ovary. Branches of the ventral bundles and the dorsal bundle extend into the style while lateral bundles and branches of the ventrals and the dorsal vascularize the ovary wall around the locule (Fics. 21, 22). Two or three small bundles from the carpellary stele remain in median positions and, with a branch from one ventral bundle, form the ovular supply. In the funiculus these bundles are nearly confluent but divide into separate bundles in the cha- lazal region (Fics. 21, 22). ORGANOGENY The value of broadening surveys of floral anatomy to include organ- ogeny and histogenesis has been emphasized recently (Tepfer, 1953; Esau, 1965; Kaplan, 1968). Gupta (1960) includes a brief description of or- ganogeny in Nannorrhops, but floral histogenesis has not previously been done for a palm. The difficulty of obtaining suitable stages for such ' 1969] UHL, NANNORRHOPS RITCHIANA 421 le mc 22. Wash drawings of cleared gynoecia, prepared as for Fic. 20. . 21, entire gynoeci sho nly bundles of the pram carpel in dorsal F esponding lateral bundles omitted for clarity. Detatms: db, dorsal bundle; ib, lateral bundles; ov, ovule; ov s, ovular supply studies in most palms is obvious. In Nannorrhops, however, particularly : age sequences in maturation of both inflorescence branches and c that results in mature flowers over a long period of time, necessary material for ontogenetic a of flowers up to anthesis may be found on a single inflorescence bran 422 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 2000 ca 16004 st 1200+ pe| FP pe pe es 8004 é © 400+ Po E £ = D c * 100+ : 180 220 Distance from center of axis in microns 23 Fic. 23. Plot of the radial path of a major bundle of the floral axis. DETAILS: ca, carpel trace; pe, petal trace; se, sepal trace(s); st, stamen trace. Above the insertion of the bracteole and subtended floral axis, floral organs arise in acropetal succession on the flanks of the apex. The floral apex is relatively long and is broadly ovate in outline; the one illustrated in FiguRE 24 is ca. 50u long and 60 wide. Floral organs are similar 1n shape in earliest stages and are developed in whorls of three, but each whorl is actually a low spiral since no three organs are at exactly the same eve Sepals are essentially triangular in outline and slightly narrower than 1969 | UHL, NANNORRHOPS RITCHIANA 423 Fics. 24-27. Hrstocenests. Fic. 24, near-median longisection of a floral apex; Fic. 25, transection showing the ep ordium of a floral se oaed and the pro- ing it; F of petal primordia ee. 27, transe tion f an older floral apex showing young tee of the pa whorl of stamens. All enema to scale, Fic. 24; scale equals 50u, Detarts: fl p, floral borconpeeietae pe, petal; pr, prophyll; se, ‘sepal: st 1, stamen of ie whorl; t 1 and t 2, first and sions! tunica layer other appendages. After initiation, the separate sepal primordia increase I size (Fic. 5), closed petals protrude ait from the sepals reaching a 424 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 length of 5 to 6 mm. in late bud. The thickened apical regions of the petals mature first. Later elongation is by a basal meristematic region (Fic. 37). Stamen primordia are initiated in two whorls of three and are elliptic to triangular in outline. FicurE 27 shows the lower whorl in early stages and Ficure 28 a later stage of the upper whorl. Initial growth is by apical and marginal meristematic areas. Anther sacs develop in adaxial and lateral positions (Fic. 30). Development of other aspects of the anther is similar to but not as regular as that described by Boke (1949) for Vinca rosea (= Catharanthus roseus). Sporogenous cells are formed by divisions of primary parietal cells. The tapetum develops later and is one to two cells wide (Fic. 36). In mature stages the endothecium is a single layer of large cells (Fic. 19). The three carpels are separate in origin (Fics. 28, 30) and show the familiar crescentic shape illustrated for developing carpels by other authors (Tepfer, 1953; Esau, 1965). In early stages carpels resemble stamens in size and shape (Fics. 28, 29). Marginal and adaxial growth (Fic. 30) provide the horseshoe-shaped primordium with a solid base and develop what has been called an adaxial lip (Tucker, 1959). The ovule primordium arises ventrally on one side in the base of the shallow cup-like lamina. Directly above the insertion of the ovule, ventral sutures of the carpels are open (Fic. 33). The submarginal position of the ovule can be seen in a young carpel (Fic. 33). Fusion of the three carpels is ontogenetic and begins in the style. FIc- URES 32 to 34 are a series of transections of a gynoecium 230, in height. Only the upper 130, of the styles are connate. Ficure 34 is the first sec- tion (proceeding distally) which shows connation. Fusion is by meri- stematic activity along the appressed ventral faces of the carpels. Initially epidermal cell walls become pointed and interlock (Fic. 35). Subsequent cell divisions produce a solid zone of tissue with no evidence of epidermal layers (Fic. 18). This zone closes the ventral suture of each carpel and joins the three carpels. Fusion progresses gradually toward the base of the gynoecium so that in the flower at anthesis, stylar and ovarian parts are connate but stipes are still separate (Fics. 17, 18). HISTOGENESIS The floral apex. Esau (1965) states that the amount of zonation of a floral apex may depend on its “determinateness,” zonation being lost oF obscured in more determinate apices. This applies well to the floral apex of Nannorrhops which is relatively indeterminate and shows distinct zona- tion. The apex (Fic. 24) is zonate with a two-layered tunica, a centra group of large corpus initials, and a rib meristem, Barnard (1960) states that two-layered tunicas are relatively common in both floral and vegeta- tive apices of monocotyledons and lists them in the Gramineae, Cypeta- ceae, Juncaceae, and Liliaceae. Rohweder (1963) has since demonstrated two-layered tunicas in the floral apices of Commelinaceae. 1969 | UHL, NANNORRHOPS RITCHIANA 425 Fics. 28-31. cerca yond continued. Fic. 28, transection of a floral apex ) flow showing carpel primordia; Fic. 29, longisection of a youn er, stamen, and carpel ptieedia posreinacs Ake equal in length; Fic. 30, transection of a young nie ristematic activity adaxial in two upper carpels, marginal in lower; , transection of base of an older flower showing adnate carpels, ee and petals. All referable to scale Fic. 29; scale equals 50”, DETAILS: ca, : Pc, procambial strand; pe, petal; se, sepal; st, stamen; st 1, stamen of Tex horl; st 2, stamen of upper whorl. hy a w ss Prophyll and floral primordium. The prophyll is Aen sion otc and is initiated first (Fic. 24, right). In early stages it is aes in outline. There appear to be oblique or periclinal divisions in the dermat- 426 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 : Fics, 32-35. DEVELOPMENT OF SYNCARPY, Fics. 32-34. Successive transec- tions through a young gynoecium ca. 2304 long. Fic. 32, ovarian part of gynoe- n 2 e . 32, op of ovules, carpels free, ventral sutures open; Fic. 34, 404 above Fic. 33, first section showing fusion of carpels; Fic. 35, transection of epidermal layers Ww 0 carpels in mat i dermal cells. All referable to scale Fic. 34; scale equals 504. DE dermis; ov, ovule; unlabeled arrow, Fic. 34, indicates area of fusion of carpel ogen in the initiation of the tips of the prophyll. This is the only place where periclinal divisions were observed in the first tunica layer. After initiation, each segment of the prophyll is extended by marginal growth, — vs 1969 | UHL, NANNORRHOPS RITCHIANA 427 the two extensions meeting to complete the abaxial sheathing part of the prophyll. The adaxial part of the sheath is adnate to the axis to a slightly higher level and apparently develops by intercalary growth. Floral organs. All floral organs are initiated by periclinal divisions in the second tunica and usually only one underlying corpus layer. Initiation of petals is illustrated in Fic. 26, stamens in Fic. 27, and carpels in Fic. 28. Only anticlinal divisions were observed in the first tunica layer dur- ing the development of floral organs. Procambium. The difficulties of determining direction of maturation of procambium are well recognized. In all organs studied for Nannorrhops, development of the first procambial strands appears to be acropetal. The first recognizable procambium in a floral stalk is in the form of three central strands. In all floral organs, a single median strand of procambium de- velops first. This is present in sepals when they are about 160, high. Stamens and carpels are about 40, in length when the median strand is recognizable. A single procambial strand is present in petals when they are about 250 long and three strands are developed when the petals are early elongation of these organs to enclose developing stamens and carpels which achieve more maturity before elongation. DISCUSSION The cincinnus. For obvious reasons the sometimes huge inflorescences of palms have not been readily available for detailed studies. Within the family much diversity is found in both major axes and ultimate flowering units. Evolution in the inflorescence of Nannorrhops appears to have re- sulted in complex patterns of maturation rather than in extreme conden- Sation and/or fusion. Consequently study of this genus is particularly helpful in understanding other genera where more reduction is present. The monopodial systems of major axes are described in a previous paper (Tomlinson & Moore, 1968). With the initiation of the first flower, growth in the inflorescence changes abruptly from monopodial to sym- podial. Designation of the flowering unit in Nannorrhops as a cincinnus is not readily evident macroscopically because the five to six flowers within each cluster are successively younger. Details of anatomy and ontogeny, how- ever, show that the basic unit in each flower cluster is a single flower bear- ing a distinctive bracteole on its axis. In the axil of the bracteole, a new floral primordium is initiated at an angle of approximately 75° on the alternate and abaxial side of each successive floral stalk. Thus the theo- retical main axis of the unit is reversed at each primordium and the result is a short scorpioid cyme or cincinnus (Rickett, 1955). Comparison of the ultimate units of Nannorrhops and Aristeyera (a 428 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Fics. 36, 37. Fic. 36, transection of part of anther, for magnification refer to 7 q scale Fic. 34; Fic. 37, near-median longisection of young flower; scale e uals 50u, DETAILs: ca, carpel; pe, petal; sp c, sporogenous cells; st, stamen; ta, tape- tum. triad of flowers, Uhl, 1966) suggests that the angle of divergence and position of the bracteole and its subtended primordium determine the shape and consequent definition of the flower cluster. In Avisteyera, each floral primordium is borne on the adaxial side of the axis rather than the abaxial as in Nannorrhops and the angle of divergence is approximately 25° to 45°. This type of analysis seems to be applicable to many of the diverse ultimate units in palms which are to be treated in detail elsewhere. Realization that the bract on the flower may be distinctive in shape and anatomy is also useful in interpreting other units. In Aristeyera no vas- cular bundles are present in the bracteole. The second bracteole is bi- carinate, however, suggesting a prophyll as in Nannorrhops. The signifi- cance of the prophyll, a bract which is morphologically and anatomically different from other bracts in the inflorescence, is not apparent at this time. The flower. Barnard (1955, 1957a,b, 1958) found that in the Gramineae, Cyperaceae, and Juncaceae, stamens were initiated in deeper layers of the floral meristems than other floral organs and, therefore, more closely resembled axial buds. Sharman (1960) also thought stamens (Gramineae) were cauline since they are more like buds in initiation and leaves in patterns of initiation and growth. Because of the nature of palm leaves, developmental patterns are obviously complex and cannot be com- 1969 | UHL, NANNORRHOPS RITCHIANA 429 pared to those of floral organs except in very earliest stages. Leaf pri- mordia in some palms as described by Periasamy (1962) seem similar to those of floral organs in Nannorrhops but histogenesis has not been studied. Evidence from organogeny and histogenesis in Nannorrhops suggests that all floral organs are homologous. Stamens and all other floral ap- pendages arise by periclinal divisions in the T, and one or more corpus layers. In early stages organs are similar in form and all receive an initial median procambial strand. Later growth patterns differ according to the whorl involved. Sepals develop rapidly and enclose other organs, but re- main separate from other floral whorls; while petals, stamens, and carpels become briefly adnate showing zonal growth for a short distance at the base of the flower (Fics. 14, 31). In addition to evidence from histogenesis, the shape and anatomy of the mature stamens suggest a laminar or foliar nature. The filaments are very wide at the base (Fic. 5). Further the large vascular bundle divides near the base of the filament suggesting the multiple trace condition of foliar stamens (Canright, 1952; Moseley, 1958) In general the vascular system of the Nannorrhops flower is similar to that of Rhapis (Uhl, Morrow, & Moore, 1969) and differs from that of the arecoid palms, Juania, Ravenea, and Ceroxylon (Uhl, in press). Ma- supply of the carpels. In Juania, Ravenea, and Ceroxylon, carpels are connate peripherally and ventral sutures are not completely closed at an- thesis; in Rhapis carpels are separate and those of Nannorrhops are con- nate by ventral faces. Ventral sutures are closed at anthesis in both the latter taxa. The ovular supply in the arecoid genera is a single bundle formed by fusion of a branch from each ventral bundle. In both Rhapis and Nannorrhops, a branch from one ventral and branches of several other bundles form the ovular supply. Within the palms the Nannorrhops flower is relatively unspecialized but within the Coryphoideae, it is one of a few in which extensive syncarpy is developed. Ontogenetic development of syncarpy would seem to re- late N. annorrhops to other Coryphoideae with separate carpels, and stylar origin of the fusion suggests further connection to a group of coryphoid genera in which the carpels are connate by the stylar regions only. A sec- ond type of fusion is seen in the sepals. The connate base arises as a unit with no evidence of union during ontogeny. Many coryphoid genera have connate sepals (Morrow, 1965), the significance of connation here is not understood at the present time. Much has been written and argued about the basic nature of the angio- sperm carpel (Eames, 1961; Tucker, 1959). Consideration of many as- pects of carpel structure in palms is beyond the scope of this paper and will be considered in a later survey. It is tempting, however, to point out here that certain features of carpels in the monocotyledons (Rhapis, Nannorrhops, Juania, Ravenea, Ceroxylon) seem equally and, perhaps, 430 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 more primitive than those of the Ranales (sensu Eames, 1961). In early stages, carpels of Nannorrhops are separate, stipitate, and conduplicate, with open ventral sutures. These are features considered primitive in carpels ( Bailey & Swamy, 1951; Baum, 1961). The ovule, in both Nan- norrhops and Rhapis, is attached basally and submarginally to one side of the laminate region. The large vascular supply to the ovule and its origin in both genera, when considered with the unspecialized form, lead to the surmise that a single ovule may possibly be primitive in palms. ACKNOWLEDGMENTS Acknowledgment is due to Professor Harold E. Moore, Jr., for valuable help during this study. Thanks are also extended to Mrs. Donald Ferguson for technical assistance. LITERATURE CITED Barney, I. W., & B. G. L. Swamy. 1951. The conduplicate carpel of dicotyle- dons and its initial trends of specialization, Am. Jour. Bot. 38: 373-379. Barnarb, C. 1957a. Floral histogenesis in the monocotyledons. I. The Gramin- eae. Austral. Jour. Bot. 5: 1-20. . 1957b. Floral histogenesis in the monocotyledons. II. The Cypera- ceae. Ibid. 115-129. . 1958. Floral histogenesis in the monocotyledons. III. The Juncaceae. Ibid. 6: 285-298. is . 1960. Floral histogenesis in the monocotyledons. IV. The Liliaceae. Ibid. 8: 213-225. Baum, H. 1952. Uber die “primitivste” Karpellform. Gsterr. Bot. Zeitschr. 99: 63 4. Boxe, N. H. 1947. Development of the adult shoot apex and floral initiation in Vinca rosea L. Am . Jour. Bot. 34: 433-439. . 1948. Development of the perianth in Vinca rosea L. Ibid. 35: 413-425. . 1949. Development of the stamens and carpels in Vinca rosea L. Ibid. 36: 535-547. CanricHt, J. E. 1952. The comparative morphology and relationships of the Magnoliacese I. Trends of specialization in the stamens. Am. Jour. ot. 39: 484— Carr, S. G. 146 & D. J. Carr. 1961. The functional significance of syncarPy- ma 11: 249-256. EAMES, A. J. 1961. Morphology of the angiosperms. McGraw-Hill Book Co., Esau, K. 1965. Plant Anatomy. John Wiley & Sons, Inc., N.Y. Gupra, S. C. 1960. Organogeny and floral coer of Nannorrhops ritchieana . Wendl. Jour. Res. Agra. Univ. 9: 103- KapLan, D.R. 1968. Histogenesis of the era and gynoecium in Downin- gia bocigalutés Am. Jour. Bot. 55: 933-950. KAussMANN, B. 1963. Pflanzenanatomie. Gustav Fischer. Jen ane Morrow, L. O. 1965. Floral morphology and anatomy of seitats Coryphoidea (Palmae). Ph.D. Thesis, Cornell Univ. 1969 | UHL, NANNORRHOPS RITCHIANA 431 MoseLey, M. F., Jr. 1958. Morphological studies of ns Nymphaeaceae — I. The nature of the stamens. Phytomorphology PERIASAMY, K. 1962. Morphological and ontogenetic ieee in Palms — 1. De- ve lopment of the plicate condition in the palm-leaf. sade ees 12: 54-64. Rickett, H. W. 1955. Materials for a page 8 of botanical terms — III. In- florescences. Bull. Torrey Bot. Club 82: 445. ROHWEDER, Otto. 1963. Anatomische und nacda Untersuchungen an Laubsprossen und Bliiten der Commelinaceen. Bot. Jahrb. 82: 1-99. SHARMAN, B. C. 1960. Developmental anatomy of the stamen and carpel pri- mordia i in oo odoratum, Bot. Gaz. 121: 192-198. TEPFER, S. S. 1953. Floral anatomy and ontogeny in Aquilegia formosa var. truncata and Ranunculus repens. Univ. Calif. Publ. Bot. 25: 513-648 ToMLINnson, P. B., & H. E. Moore, Jr. 1968. ass in Nannorrhops ritchiana (Palmae). Jour. Arnold Arb. 49: 16- Tucker, S. C. 1959. Ontogeny of the ie hihi oe the flower in Drimys winteri var. chilensis. Univ. Calif. Publ. Bot. 30: 257-366. UuL, N. W. 1966. Morphology and anatomy of the inflorescence axis and flow- ers of a new palm, Aristeyera spicata. Jour. Arnold Arb. 47: 9-22. . 1969. Floral anatomy of Juania, on and Ceroxylon (Palmae- Arecoideae). Gent. Herb. 10 (4): in p , L. O. Morrow, & H. E. Moore, a 1969. sored of the palm Rhapis excelsa, VII. Flowers. Jour. Arnold Arb. 50: 138-1 ZIMMERMANN, M. H., & P. B. TomLtInson. 1965. retnda of the i. Rhapis excelsa, I. a ae axis. Jour. Arnold Arb. 46: 160— L. H. Battey Hortortum CORNELL UNIVERSITY IrHaca, NEw Yorx 14850 432 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ASPECTS OF MORPHOLOGY OF AMENTOTAXUS FORMOSANA WITH A NOTE ON THE TAXONOMIC POSITION OF THE GENUS Hsuan KENG THE GENUS Amentotaxus was established by Pilger in 1916 (Bot. Jahrb. 54: 41), based on the type species, A. argotaenia (Hance) Pilger. This species, described from sterile material only, was originally desig- nated as a member of the genus Podocarpus. As soon as its compound staminate strobilus became known, it was transferred to Cephalotaxus, and finally to a separate genus, Amentotaxus. The genus Amentotaxus is endemic to eastern Asia. It was first col- lected from a small islet near Hongkong and also from southern Kwang- tung. Subsequently it was reported from southern Formosa, western Hupeh and Szechuan, and from southern Yunnan and northern Tonkin (see Fic. 1). Fossil remains have been recorded from Europe and west- ern America (Sporne, 1965). It was generally considered as a monotypic genus; Li (1952), however, recognizes that there are at least four dis- tinct entities (which he considered species) involved, based on color and relative width of the stomatal bands, and geographic distribution. Chuang and Hu (1963), on the other hand, point out that the characters of the stomatal band appear to be less constant, and maintain that there is only one species, namely A. argotaenia (Hance) Pilger. It is rather difficult to make a judgment on this controversial issue without thoroughly examining suitable materials with reproductive struc- tures, which unfortunately, are not available. For simplicity of nomen- clature, since all the materials used in this study are from a small locality in southern Formosa, the binomial Amentotaxus formosana Li is, ac cordingly, adopted. It would be interesting to have reports on the strobilate structures based on the materials from other parts of the geographical range of the genus. The plants of Amentotaxus are small to medium-sized, dioecious, evergreen trees. A limited number of them are perhaps in existence, and they grow in almost inaccessible places. Moreover, they are not repre sented in any botanical garden or arboretum in the world. Owing to the The stomatal and ovulate structures were reported by Florin (1931, 1938— 45); his interpretation of the latter, as indicated in a drawing reproduced in 1951, p. 375, fig. 64, was apparently based on poorly preserved her- barium material, and is inadequate. Only fragments of the embryonic 1969 | KENG, AMENTOTAXUS FORMOSANA 433 300 mis Ficure 1. Geographic ie ee of the genus Amentotaxus. 1, Lantao Is- land, near Hongkong; 2, Mt. Lo-fau-shan, Kovangtung: a, ni and S. Kaoh- siung, lg (Formosa a); oe isi ae -shan, Hupeh; 5, Mt. Omei-shan, Szechuan; , , Yunnan; 7, Cha Pa, Tonkin ( res on the ore ar specimens cited in on 1952 ?. development were given by Sugihara (1943); and his chromosome num- er, m = 11, on the basis of counts from the female gametophyte, is in- correct, as pointed out by Chuang and Hu (1963). The pollen morphology has been carefully investigated by Erdtman (1957). MATERIALS AND METHODS Material preserved in FAA (including leaves, staminate strobili, ovulate strobili, seeds, and seedlings), and dried material (including young ovulate strobili, seeds, and seedlings) taken from herbarium specimens, were received from Professor Ching-en Chang of the Pingtung Agriculture Col- lege, Taiwan. All the materials were collected by Professor Chang from near Shin-Huah Farm, Shaw-Jia, Dah-Wu, Taitung, between 1965 and 1968, Clearings of testeatiie were made with 5 percent NaOH at room temperature. Microtome sections 10 to 12 » thick were stained with a safranin-fast green combination. 434 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 vas. b. u.epl = pal t. spon. t. Ficure 2. Transverse section of a leaf, and surface ~e parce views of a stoma. a Diagram of the transverse section of a leaf; C, portions of A, enlarged, soar the cellular details; D, sectional view gee the axis) of a stoma; E, surface view of a stoma "(the broken lines marked guard ae’ : gua are drawn at a different focus). Nore: aper. = apertures; guard Cc. = cells; J. epi. = lower epidermis; pal. t. = palisade tissue; res. d. = resin ducts; Spon. t. = Fn gpg tissue; stom. b. = stomatiferous band; subs. ¢. = subsidiary cells: u. epi. = upper epidermis; vas, b. = vascular bundle. Leaves. Foliage leaves are persistent, spirally arranged on the branch- lets but twisted at the base into two rows in one plane. Internodes are o 7 mm. long (average). Each leaf consists of a lamina and a Very short oetiole, Laminae are coriaceous, bifacially flattened, with the adaxial a 1969 | KENG, AMENTOTAXUS FORMOSANA 435 surface upward. They are linear, often strongly falcate, acute or more often acuminate at apex and slightly oblique at base, 5—7 cm. long, 0.5—1 cm. broad. Two very prominent stomatiferous bands are present on the abaxial surface and run parallel to the elevated midrib, one on each side of it (see Plate I, a & b). Petioles are strongly decurrent on the branch- lets. Anatomically, each lamina possesses only one large, median vascular bundle with a resin canal beneath (see Fic. 2, A & B). The assimilatory tissues consist of one to two (near and at the midrib) rows of palisade cells and numerous polygonal, elongate, and dissipated spongy cells. The upper and lower epidermis are both well defined, the former with slightly more thickened cuticle layer. Stomata are arranged in longitudinal rows, their axes oriented more or less parallel to the midrib of the leaf. Each stoma (see PLATE II, b; Fic. 2, E) is encircled by 7 to 9 subsidiary cells. Strong papillae of the subsidiary cells surround the orifice of the stomatal apparatus like a wall, while the guard cells are also heavily cutinized. Sclereids are abundant, slender, branched or unbranched at one or both ends, and generally lying between the midrib and leaf-margins and per- pendicular to them (see PLATE II, b). Staminate strobilus. The compound staminate strobili are produced within the large winter bud which is borne on the top of the previous year’s branchlets. They are short-stalked, usually four (sometimes three, rarely two or five) together, subtended by four rows of imbricate bud-scales (see Piate I, a; Fic. 3, A). These scales are leathery, strongly keeled and more or less pointed. The true terminal bud of these branchlets is gen- erally in the center and is further protected by small, thin scales (Fic. , B); it remains dormant and resumes its activity only after expansion and withering of the surrounding compound staminate strobili. Each compound staminate strobilus is spike- or catkin-like, from which is derived the generic name Amentotaxus (Fic. 3, C). When fully ex- panded, the compound strobilus can reach a length of 2.5 to 3 cm. or more. It consists of approximately 20 to 30 globular staminate strobili somewhat decussately arranged (though not quite regular), growing along the main axis in four rows. These globular staminate strobili are clearly recognizable especially in the middle portion of the spike, since the distal Ones are overcrowded and fused, and the lowermost ones are sometimes adherent to the side (secondary) branches rather than being on the main axis itself. The staminate strobilus is globular or ovoid, 2.5 to 3 mm. in diameter in bud (see Fic. 3, D). It is composed of 9 to 12 closely compacted microsporangiophores,! which are peltate, with four or five (varying from f. 20 & 21) and Pseudotaxus (Florin, 1948 a, p. 389, f. 2), microsporangiophore is perhaps preferable, as a designation, to microsporophyll; although no trace of the subtending bracts has been found at the base of the stalk in Amentotaxus. 436 [voL, 50 JOURNAL OF THE ARNOLD ARBORETUM yt Py = = ry tem Be a Shes Y, fi WY Y Tt ko ry Ficure 3. Compound staminate strobilus, staminate strobilus and mic sporangiophore. A, External view of an un : of four (one is not seen) compound staminate strobili; B, the same, with ro- folded winter bud, showing 2 rage 1969 | KENG, AMENTOTAXUS FORMOSANA 437 two to eight) microsporangia hanging underneath in a semicircle and with a short stalk near the center (see PLate II, e; Fic. 3, E). The outline of the peltate microsporangiophores, as seen from the outer surface, varies from round to deltoid, to more commonly polygonal, due to mutual compression. At maturity the thickened outermost layer and one or two (in part) inner layers of the saerirtg walls are retained (see Plate II, d & e). The microspores are wingless Ovulate strobilus. The ovulate strobilus is globular to ovoid, flat- tened dorsiventrally (see Pirate I, b; Fic. 4, Ay, As, By, B en C2). These strobili are situated singly in the axils of foliage leaves. “The ovule is solitary, terminal on the strobilate axis, and subtended below by five (or six) pairs of opposite and decussate, sterile bracts; three pairs of which are lateral and prominently keeled, and the other two (or three) pairs are dorsiventral and only slightly curved (Fic. 4, Ay). The stalk of the strobilus is slender, about 1-1.5 cm. long, more or less flattened and narrowly winged. oung ovules, at the stage of about 3 mm. in length (excluding the sterile bracts) (Fic. 4, A,), possess an elongate conical nucellus, of which the upper part is loosely enveloped by a single layer of integument, the lower half, however, is seemingly associated only with the cupular aril- lus primordium. The integument (if it is interpreted as confined to the portion above the arillus only) and the arillus, at this stage, appear to be completely separated. The vascular supply to the ovule, as seen in cross section, consists of 8 to 10 normally oriented vascular bundles. They terminate at the end of the ovule far below the nucellus — neither the integument nor the arillus primordium is visibly vascularized. ee the slightly older ovules at the stage of about 5 mm. in length (Fic. B;), as a result of the enlargement of both the nucellus and the integ- evidently embedded in the cupular arillus. The vascular supply can be observed near the base of the ovule. In the still older ovules, at the stage of about 6.5 cm. in length (Fic. 4, C3; Fic. 5), the nucellus is enveloped by and fused with the integument except for the uppermost part which remains free. Nearly two-thirds of the integument, in turn, is embedded in and completely united with the arillus. Isolated tracheids may be found in the lower part of the ovule at a fairly high level in the region where the fusion of aril and integument occurs, of the bracts ae compound staminate strobili removed, showing the hidden terminal bud inside; C, a compound staminate strobilus, showing a number of ovoid to shies staminate strobili more or less decussately callgge ge on an axis; D, a globular staminate strobilus (taken from the me ian } larged; E, and E., two views of a spemnsinieor showing five microsporangia arranged in a semicircle below; F, and F,, two p scatelie te eal views of the sporangiophor 438 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 A general outline of the tissues in the largest ovule available is shown in Ficure 5. The arillus consists of 10 to 15 layers of parenchyma cells with rows of resinous cells lining the epidermis near the rim. The cuticle is thin. The lower portion of the integument is composed of 12 to 20 layers of small, partly closely packed, and partly loosely dissipated parenchyma cells. There is no clear distinction of the arillus from the integument below the level of fusion. The upper portion of the integu- ment is heavily cuticularized. The cells near the micropyle are enlarged, sclerenchymatous, and oriented horizontally. The nucellus is prominently beaked; the beak is hemispheric, and composed of numerous small polyg- onal cells with moderately heavy walls rather loosely arranged especially towards the micropylar end. A large portion of the nucellus at this stage, is digested and replaced by the multicellular megagametophyte. Isolated tracheids and short rows of tracheids are observed at the lower part of the peripheral region where the integument and arillus are merged. The ovular structure of Amentotaxus in general, as noted by several authors (e.g. Florin and others), is rather similar to that of Torreya; but its vasculature is very much simpler. In Amentotaxus, although there are 8 to 10 vascular strands entering the base of the ovule, only the isolated tracheids are present in the lower part of the ovule, in the region where the integument and arillus meet. In Torreya, however, there are two vas- cular strands running up inside the arillus nearly to the apex of the seed, each of which then sends a branch through a foramen in the stony layer of the integument; each branch forks, forming a loop which encircles the seed. Oliver (1903) proposed the “hyposperm theory” to explain this peculiar vascular structure. According to this theory, all the basal part (the “hyposperm”) of the ovule is an intercalated growth and phylo- genetically younger than the extreme tip (the “archisperm’’). The branching of the integumental vascular bundles inwardly is also reported in Austrotaxus (Saxton, 1934, p. 419, fig. 18) and Cephalotaxus (Singh, 1961, p. 160, fig. k). In the case of Amentotaxus, no traces of such branching are present. With the intercalary growth of the lower part of the integument and arillus concomitantly with the enlargement of the nucellus, the integument becomes evidently embedded in the arillus and fused with it. There seems to be no evidence to prove that the lower part of the Amentotaxus ovule is a “hyposperm,” or is phylogenetically youns- er than the upper part. Seed and seedling. The seed is ellipsoid-oblong, drupe-like (PLATE I, c), 3.2 to 3.6 cm. long, 1 to 1.2 cm. broad, and slightly flattened dorsi- ventrally. The outer part of the seed coat is completely covered and fuse with the arillus, except the extreme tip which is exposed (Fic. 6, A & B). The merged structure is soft-leathery in texture although the outer por- tion is easily blistered and disintegrated when soaked in water. The nucellus is almost entirely replaced by the ivory female gametophyte (“endosperm”) which has an entire rather than ruminate margin (as 7 Torreya). The embryo is linear, lying in the center of the gametophyte. Ficure 4. External and sectional views of three ovulate strobili. A, and As, two views of a young strobilus; A, and A,, longi- tudinal and transverse sections of the same strobilus; B, and B., two views of a slightly older strobilus; Bs, longitudinal section of the same; C, and C., two views of an older strobilus ; C;, longitudinal section of the same; Nore: micr. = micropyle; integ. = integument; ster. br. = sterile bract; vas. b, = vasc cular bundle. VNVSOWYOA SAXVLOLNAWNV ‘ONAN [6961 ocr 440 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 In common with other conifers, germination is of epigeal type. In the one year old seedling examined (Prater I, d), the cotyledons have dropped, but the two cotyledonary scars are clearly evident. The juvenile leaves are 3.5 to 4 cm. long, 3.5 to 4 mm. wide, with two glaucous stomatiferous bands underneath. Fundamentally of spiral arrangement, since the inter- nodes are of variable length, the juvenile leaves appear subopposite or rarely subverticillate. TAXONOMIC POSITION OF AMENTOTAXUS AND THE CLASSIFICATION OF THE CONIFERALES Pilger (1926) assigns Amentotaxus, together with Cephalotaxus, to the Cephalotaxaceae on the basis of the compound nature of their staminate strobili. Kudo (1931), after seeing the ovulate strobilus, hitherto un- known, maintains that “it (Amentotaxus) must be included in a new family Amentotaxaceae, or in a new subfamily or tribe of Taxaceae, but not in Cephalotaxaceae” (p. 311). As a result, a new family Amento- taxaceae was proposed by Kudo and Yamamoto (in Kudo, 1931). Koid- zumi (1932) strongly felt that the new family was not necessary. He, therefore, established a subfamily Amentotaxoideae (including both Amen- totaxus and Austrotaxus) within the Taxaceae. Later on, following his enumeration of various similarities and dissimilarities among the Taxaceae, Cephalotaxaceae, and Amentotaxus, he (1942), recognized that Amento- taxus and Cephalotaxus are in fact related, and moreover, suggested that Taxaceae and Cephalotaxaceae should be merged into one family and both reduced to subfamilial status. Florin (1948, 1951) emphasized the dif- ferences of ovulate strobili and stomatal structures between Cephalotaxus and Amentotaxus, and thus sustained the transference of Amentotaxus from Cephalotaxaceae to Taxaceae. Chuang and Hu (1965) report the chromosome number of Amentotaxus argotaenia (Hance) Pilger (or A. formosana Li) to be x = 7, which is different from those reported from Taxus (x = 12), Torreya (x = 11), and Cephalotaxus (x = 12). They, therefore, support Kudo and Yamamoto in maintaining Amentotaxus in a separate family, the Amentotaxaceae. The present writer is inclined to think that (1) Amentotaxus is probably better placed in the Taxaceae than in the Cephalotaxaceae or in a separate family; (2) the Taxaceae are not isolated, but are likely allied to the Cephalotaxaceae, probably through Amentotaxus. These two points are elaborated in the following paragraphs, Features which distinguish Amentotaxus from other members of the Taxaceae such as the spicate compound staminate strobili, the peculiar stomatal structure (with larger number of subsidiary cells, thickened papillae, etc.), etc., appear to be insufficient to warrant a separate family status. The difference in chromosome number is probably inadequate to be cited as a justification for the establishment of the Amentotaxaceae.” *For example, in a recent report (Hair & Beuzenberg, 1958) on the chromosome numbers of the Podocarpaceae, the following two closely related genera possess such 1969 | KENG, AMENTOTAXUS FORMOSANA 441 aS integ. Z Fah Z EO} Zee 2» GN nuc. b ca iS (a) ei ON aril , res. ¢ ese AN ‘\ rely WY i 6 a ribet Wie | a A trach. ee rte a () ot} Ch, FIGURE . ot ae sa an ovule, details of Figure 4C;. bangs muc. b. cellus beak; esin cells; muc. = nucellus; @ = megagametophyte: trach. = nae ids. On the other hand, the resemblance of Amentotaxus to the Taxaceae, es- pecially to the genus Torreya, in the general structure of staminate strobili, ovulate strobili, microsporangiophores, ovules, etc. is overwhelming. There- fore, Janchen’s (1949) treatment including both Amentotaxus and Torreya in the Tribe Torreyeae under the Taxaceae appears to be a logical one. 4 Tange of variation: Dacrydium (x = 15, 12, 11, 10, heaped & = 19, 18, 17, 13, 12, 11, 10); whereas other members of the fam o have various feent basic numbers: Acmopyle (x = 10), a eiveaiami (x = ae iscedaiae (a = 13), Phyllocladus (x = 9), and Saxagothaea (x = 12). 442 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Ficure 6. External and sectional views of a seed. A more or less similar view is held by Kudo (1931), Koidzumi (1932), Li (1952), and others. Florin in a series of papers (1948, 1951, 1954) strongly advocated sep- aration of Taxaceae (which includes the following five genera: Taxus, Amentotaxus, Torreya, Austrotaxus, and Pseudotaxus [ = Nothotaxus]) from the rest of the Coniferales to form a separate order, the Taxales, 4 view originally expressed by Sahni (1920) but modified by Florin with the exclusion of Cephalotaxus. A quite different scheme proposed by Buchholz (1934), was summarized by Chamberlain (1935, pp- 229; 230) as follows: The order Coniferales can be divided into two suborders: one the Pinineae (as Phanerostrobilares or Pinares) with an obvious cone, includes the Pinaceae, Taxodiaceae, Cupressaceae and Araucariaceae; the other, Taxineae (as Aphanostrobilares or Taxares) without such an ob- vious cone, contains the Podocarpaceae, Taxaceae, and Cephalotaxaceae. Florin (1951, pp. 363, 364) fully endorsed Wilde’s (1944) postulation that in Podocarpus, the species with 1-ovulate strobili are independently derived from those with multiovulate strobili, and represent the ultimate stage of reduction. In addition, his own interpretation (1951, 1954) of the ovulate strobilate structures of the modern conifers as possibly evolved from a much more complicated structure of fossil groups such 1969} KENG, AMENTOTAXUS FORMOSANA 443 as found in the palaeozoic Lebachia, Ernestiodendron, Walchia, and Pseudovoltzia, has been widely appreciated. Paradoxically, he insists that the l-ovulate strobilus of the Taxaceae is a primitive rather than a derived condition; therefore the family Taxaceae is of entirely different origin from the rest of the other Conifers. This is mainly because of his emphasis on the finding of 1-ovulate Palaeotaxus in the Triassic and Taxus jurassica in the Jurassic rocks. “Because of its high geological age” he noted (1951, p. 349), “Palaeotaxus can hardly derive from any cone- bearing type.” It seems he does not realize the possible existence of the exceptionally fast rate of evolution, designated by Simpson (1944) as tachytelic evolution. Many authors, such as Chamberlain (1935, p. 439), Pulle (1937), Takhtajan (1953, p. 34), etc. express their notions that the single ovulate strobilus of taxads is most likely derived from the multi- ovulate cones. The present writer (Keng, 1963) also points out that the evolution of the ovulate strobili in the genus PAyllocladus (belonging to the Podocarpaceae, or according to some authors, the monogeneric family, Phyllocladaceae) might indicate the possible mode of how the single, pseudo-terminate ovule of taxads could have been achieved. Incidentally, Phyllocladus is somewhat intermediate between the Taxaceae and Podo- carpaceae; on morphological ground it is probably correct for it to be placed in the Podocarpaceae (Maheshwari, 1962). Although, as discussed above, Amentotaxus should be better classified in the Taxaceae rather than Cephalotaxaceae, it does not mean that the Taxaceae and Cephalotaxaceae are totally unrelated as suggested by Florin. The present writer agrees with Saxton (1934), Pulle (1937), Koidzumi (1942), and many others that these two families are in fact related. In this connection, it is rather interesting to mention the views of Singh (1961) who has contributed an excellent account on the life history of Cephalotaxus drupacea Sieb. & Zucc. In his discussion of the relation- ships of the Cephalotaxaceae and Taxaceae, he pointed out a number of similarities between these two families and noted that they “resemble each other in wood structure, pollen structure, and to some extent embryogeny”” (p. 193). He was, unfortunately, dominated by Florin’s misconception that the Taxaceae are isolated and reached the contradictory conclusion that “it appears best to regard the Taxaceae and the Cephalotaxaceae as unrelated” (p. 193). If we accept the general view that the compound staminate strobilus is a primitive condition (Wilde, 1944), that the peltate sporangiophore is more antiquated than the dorsiventral ones (Florin, 1948), and that the one-ovulate strobilus is derived from a multiovulate strobilus (Pulle, 1937), and also if we assume that the Taxaceae and Cephalotaxaceae are phylogenetically affiliated, then an ideal ancestral form of Taxus-Amento- taxus-Cephalotaxus complex would hypothetically possess the following Synthetic strobilate features. rae taminate or microsporangiate strobili—a cluster of spike-like com- pound strobili surrounding a terminal bud and enveloped by numerous bud-scales (cf. Amentotaxus); each compound strobilus composed of 444 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 many ovoid or globular strobili; each strobilus consisting of many peltate, spirally arranged sporangiophores with a number of sporangia on the undersurface around the stalk (cf. Taxus, Pseudotaxus); each peltate sporangiophore further subtended by a leafy bract (cf. Pseudotaxus, see Florin, 1948a, p. 389, fig. 2; or Austrotaxus, see Saxton, 1934, p. 423, figs. 20 & 21). Ovulate or megasporangiate strobili—a strobilus composed of many imbricate ovuliferous scales each with several to two (or one) ovules on its upper surface (cf. Cephalotaxus); ovules erect, with only one integu- ment and surrounded by a cupular arillus but free from it (cf. Taxus, Pseudotaxus); the integument supplied by a number of lengthwise vascu- lar bundles, each of which gives a horizontal branch in the middle of the integument, toward the inner part of the ovule to supply the nucellus and gametophyte (cf. Torreya, see Oliver, 1903; Austrotaxus, see Saxton, 1934, p. 419, fig. 18; Cephalotaxus, see Singh, 1961, p. 160, fig. *). To summarize, firstly, since the resemblance of Amentotaxus to Torreya (Taxaceae) is so overwhelming, it seems logical to include Amentotaxus in the Taxaceae; secondly, since the family Taxaceae is intricately affiliated to the Cephalotaxaceae on the one hand and possibly to the Podocar- paceae on the other, Buchholz’s scheme of classification of the Coniferales, therefore, appears to be sound. ACKNOWLEDGMENTS I am most grateful to Professor Ching-en Chang, who kindly made several trips to the southern part of Taitung District, Formosa, to collect the materials for this study, and to Dr. H. L. Li and Dr. Gloria Lim for reading the manuscript. My thanks are also due to Professor B. Y. Yang, Dr. Ding Hou, and Mr. M. C. Kao for supplying some of the literature; to Dr. C. C. Hsii for translating several paragraphs from the Japanese literature; and to Mr. D. Teow for taking the photomicrographs. LITERATURE CITED Bucuuotz, J. T. The classification of Coniferales. Trans. Illinois Acad. Sct. 25: 112, 113. 1934. CHAMBERLAIN, C. J. Gymnosperms, structure and evolution. xi + 484 PP» gs. Univ. of Chicago Press. 1935. Cuuane, T. L, & W. W. L. Hu. Study of Amentotaxus argotaenia (Hance) Pilger. Bot. Bull. Acad. Sinica, II. (Taipeh) 4: 10-14. 1963. : ErptMAN, G. Pollen and spore morphology/plant taxonomy. 147 pp., fronits., pls. Almqvist & Wiksell, Stockholm. 1957 FLorin, R. Untersuchungen zur Stammesgeschichte der Coniferales und Cor- daitales, I. Sv. Vet-akad. Handl. III. 10: 1-588. 58 pls. 1931. ———. Die Koniferen des Oberkarbons und des unteren Perms. -Vill. Palaeontographica 85B: 1-729. 1938-45. [Cited in Florin, 1951.] 1969 | KENG, AMENTOTAXUS FORMOSANA 445 On the eee and relationships of the Taxaceae. Bot. Gaz. 110: 31-39. 1948 . On Nothotaxus, a new genus of the Taxaceae from eastern China. Acta Horti Berg. 14: 385-395. 1948a. - Evolution in Cordaites and conifers. Ibid. 15: 285-388. 1951. . The female reproductive organs of conifers and taxads. Biol. Rev. 29: 367-389, 954 Harr, J. B., & E. _ BEUZENBERG. rage i evolution in the Podocarpaceae. Nature 181: 1584-1586, (June) 19 JANCHEN, E. Das System der aie ‘Akad. Wien. Sitz-ber. 158: 155-262. 1949. [Cited in Li, 1952. Kenc, H. Taxonomic position of Phyllocladus and the classification of Conifers. Gard. Bull. Singapore 20: 127-130. 1963. Koizumi, G. Notes on Amentotaxaceae. [In Japanese.] Acta Phytotax. Geo- bot, i: 185. 1932. . Further notes on Amentotaxaceae Kudo. [In Japanese.] Jbid. 11: 135, 136. 1942. Kupo, Y., & Y. Yamamoto. Amentotaxaceae. In: Kupo, Mater. Fl. Formosa IV. Jour. Soc. Trop. Agric. (Taihoku) 3: 110, 111. 1931. Li, H. L. The genus Amentotaxus. Jour. Arnold Arb. 33: 192-198. 1952. MAHESHWARI, P. The overpowering role of morphology in taxonomy. Bull. Bot. Surv. India 4: 85-94. 1962. Otiver, F. W. The ovules of the older gymnosperms. Ann, Bot. 17: 451-476. 1903. PILcer, R. Cephalotaxaceae. IN: ENGLER & PRANTL, Nat. Pflanzenfam. ed. 2. 13: 267-271. 1926. Putte, A. Remarks nc as system of the Spermatophytes. Med. Bot. Mus. Utrecht 43: 1-17. SAHNI, B. On certain ede features in the seed of Taxus baccata, with re- marks on the antiquity of the Taxineae. Ann. Bot. 34: 117-133. 1920. Saxton, W. T. Notes on Conifers VIII. The morphology of Austrotaxus spi- cata Compton. Ann. Bot. 48: 412-427. 1934. Smupson, G. G. Tempo and mode in evolution. Columbia Univ. Press. New York. 1944, SINGH, H. The life history and systematic ag oe oe drupacea ieb. et Zucc. Phytomorphology 11: 153- SPoRNE, K. R. The morphology of ioe tia Univ. Library. London, 1965. SucrHaRA, Y. Notes on Amentotaxus. [In Japanese.] Bot. Mag. Tokyo 57: 404, 405. 1943. TAKHTAJAN, A. L. — principles of the system of higher plants. Bot. Rev. 19; 1-45, Wipe, M. H. A new gamma of coniferous cones. I. Podocarpaceae (Podocarpus). Ann. Bot. II. 8: 1-41. 1944. Yamamoto, Y. Cephalotaxaceae. Suppl. Icon. Pl. Formos. 3: 1-5. 1927. eae, Amentotaxaceae. Ibid. 5: 7-11. 1932. DEPARTMENT OF BOTANY UNIVERSITY OF SINGAPORE Bukit Trmau Roap SINGAPORE, 10 446 JOURNAL OF THE ARNOLD ARBORETUM [VoL. 50 EXPLANATION OF PLATES PLATE I (Scale in each figure in 1 mm. divisions.) Amentotaxus formosana Li. a, Cluster of compound staminate strobili from an unfolded winter bud borne on the tip of a branchlet (cf. Fic. 3A); b, solit pd ovulate strobilus borne in the axil of a leaf (which has dropped off) (cf. 4, C.); c, one fully mature and three young seeds (cf. Fic. 6A); d, celine of which the two cotyledons (on the first node) ae dropped off. PLATE II (Scales: in a, 150 w; in b and c, eat & in d and e, 200 uw; in f, 2 mm.) Amentotaxus formosana. a, Transverse section of a leaf showing the midrib region (cf. Fic. 2B); b, lower (abaxial) surface of a leaf (after ee show- ing the stomata and sclereids : c, stomata in transverse eo (cf. Fic. 2D); d, transverse section of a microsporangiophore (ef, Pe. 3, .F,)i e, kon gitudinal —. of a BP coms aa f, longitudinal section of an ovulate ennai 44. (cf. Jour. ARNOLD ArB. VOL. 50 PLATE I KEenc. AMENTOTAXUS FORMOSANA Jour. ARNOLD Ars. VOL. 50 PLATE II KENG, AMENTOTAXUS FORMOSANA 1969} RUDENBERG & GREEN, LONICERA, II 449 A KARYOLOGICAL SURVEY OF LONICERA, II Lity RUDENBERG AND PETER S. GREEN * IN THE FIRST PAPER presenting the results of this survey, all the chromo- some numbers recorded for the genus Lonicera, to that date, were as- sembled, together with many new counts. Since that time the study of Lonicera has continued, but to bring the investigation to a conclusion all the additional counts that have been made using the Arnold Arboretum collections are presented below (together with three further records that have appeared in the literature). Cytological methods, documentation and nomenclature used here fol- low those of the first paper, to which reference should be made An attempt was made to note differences in karyotype morphology and, certainly, differences in the overall size of chromosome complements were observed between different species. Also, variation in individual chro- mosomes, their size, centromere position, and the presence and size of satellites were noted, but considering the relatively large number of spe- cies in the genus and the few individuals investigated, it has not proved possible to compare and correlate these differences, and their groupings, with the infrageneric classification proposed by Rehder (1903). At metaphase the chromosomes, in many cases, were so contracted that two satellites were not always visible. Thus, it was not possible to determine whether or not Lonicera modesta had a satellited chromosome pair. More details of morphology could be observed at late prophase. In some cells, pretreatment with oxyquinoline (Tjio & Levan, 1950) caused a Structural differentiation of the chromosomes by revealing positively and negatively heteropycnotic segments. Homologues of similar size could then be identified by the location of the centromere and by the individual distribution of these segments. A comparable pattern has been observed in several homologues of different species of Lonicera. Ficures 1 to 10 present examples which were encountered of nuclei in mitosis (most ex- amples taken from species in different subsections of Rehder’s classifica- tion). A few comments may be made. In four cases both diploid and tetra- ploid plants have been recorded within the same species. In Lonicera ferdinandii Franch., the earlier undocumented counts and all the plants at the Arnold Arboretum appear to be diploid, except for one (AA 21595) which is tetraploid. This particular bush is an old one, raised from seed of Rock 13519 collected in S.W. Kansu, China, in 1925, yet phenotypical- * In this survey, the cytological investigations have been eae out by one of us (L. R.), and the complementary taxonomy by the other (P.S “Part I was published in Jour. Arnold Arb. 47: 222-247. 1966. 450 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 ly it does not appear to differ significantly from the diploid. In L. alpigena L., Poucques (1949, pp. 129 & 186) has recorded » = 9 and 2n = 18, both of which numbers were confirmed by counts on a plant in the Arnold Ar- boretum (AA 91-60) which, unfortunately, died before an authenticating herbarium specimen was collected. However, in this species, the tetraploid number, 2” = 36, has been found in two plants of f. mana (Carr.) Zabel (see below). In L. maximowiczii (Rupr.) Maxim. var. sachalinensis Fr. Schmidt we can now document a tetraploid (7 = 18 and 2n = 36), in contrast to the diploid number of 2n = 18 recorded for the species by Janaki Ammal & Saunders (1952, p. 540). The plant on which their count was based does not appear to have been documented and it is now impossible to know which variety may have been involved, or to confirm its identity. Lastly, in our first paper we recorded a plant of L. modesta Rehd. var. modesta as diploid (n = 9 and 2n = 18) and of var. lusha- nensis Rehd. as tetraploid (n = 18 and 2n = 36), both plants having been raised from seed sent from the Lushan Botanic Gardens in China. Here, however, there is need for taxonomic reassessment, as we have pointed out (Riidenberg & Green, 1966, p. 225). Available herbarium material has proved inadequate to enable one to come to a sound con- clusion, but it may well prove that two species are involved where diagnostic distinctions need careful delineation. It is, perhaps, worth drawing attention to the fact that in the whole of both subsections TATARICAE and OcHRANTHAE, including many culti- vars and hybrids, but with one exception, no polyploid plants have been observed. The exception is Lonicera floribunda Boiss. & Buhse (AA 341-44) which is tetraploid. Within and between these subsections hy- bridization takes place readily, yet meiosis in most of these diploid hybrids is, with the exception of some plants with bridges, perfectly normal. A few of the plants studied at the Arnold Arboretum form bridges at ana- phase I, especially L. x bella; meiosis was, therefore, checked the next year to determine its constancy and whether or not the frequency of these bridges could be correlated with the seasonal variation in climate. It was found that the number of cells showing bridges was not the same for the two years. It was smaller after the more normal spring, in contrast to one with especially cold nights and periods of drought. LITERATURE CITED JANAKI Amat, E. K., & B. SauNnpERS. 1952. Chromosome numbers in spe- cies of Lonicera. Kew Bull. 1952: 539-541. Love, A. 1968. IOPB Chromosome number report XIX. Taxon 17: 573-577. Love, A., & D. Loéve. 1966. Cytotaxonomy of the alpine vascular plants of Mount Washington. Univ. Colorado Studies, Ser. Biol. 24: 1-74 Poucques. M.-L. pe. 1949. Recherches caryologiques sur les Rubiales. Revue Gén. Bot. 56: 5-27, 74-138, 172-188. [Lonicera pp. 84-95, 129, 186.] REHDER, A. 1903. Synopsis of the genus Lonicera. Ann. Rep. Missouri Bot. Gard. 14: 27-232. 1969 | RUDENBERG & GREEN, LONICERA, II 451 | RUDENBERG, L., & P. S. GREEN. 1966. A karyological survey of Lonicera, I. Jour. Armold Aah 47: 222-247. | sinks R. L., & G. A. MULLIGAN. 1968. Flora of the Queen Charlotte Islands, ol ree & A. Levan. 1950. The use of oxyquinoline in chromosome anal- ysis. Anal. Estac. Exp. Aula Dei 2: 21-64. | TABLE. Additional chromosome numbers in Lonicera csp DOCUMENTATION GENERAL SPECIES n 2n AND COLLECTOR DISTRIBUTION Subgenus Lonicera (Subgen. Chamaecerasus (L.) Rehd.) Sect. IsoxyLOSTEUM Reh Subsect. MrcrostyLaE Rehd. L. angustifolia Wall. ex DC. 9 See Mehra & Gill in Himalayas 1 1291 (PUNJAB), Simla, W. Himalayas *D. syringantha Maxim. 18 AA 405-35, Palmer, North & West China 1 June & 26 Aug. 1936 *var. wolfii Rehd. 18 36 AA 4992-2, Allen, West China 1 June 1927, also Dudley & Dodd, 28 May 1965 *cv. Grandiflora 36 AA 1089-61, agree 18 May 1 Sect. IstKA (Adans.) Rehd. Subsect. CAERULEAE Rehd. L. villosa (Mich.) Roem. & Schult. 18 See Love & Love (1966, p.51). Northeastern Based on Love & Love North America 7496 & 7591, Mt. Wash- ington, New Hampshire WOLAAOTAV CIONUYV AHL JO TWNYNOL OS “10A] eo -— all 2 y Subsect. PILEATAE Rehd. *L. pileata Oliv. 18 AA 151031-B, Dudley & Central and Dodd, 28 May 1965 western China 9 AA 225-28-E, Gre 4 Nov. 1965 and ae 225-28) Kobuski & Roush, 14 Sept. 1931 *L. nitida Wils. 18 AA 923-49, Green, Western China 4 Nov. 1965 Subsect. VESICARIAE (Komar.) Rehd. L. ferdinandii Franch. 18 AA 21595 (Rock 13519, Northern China Kansu, 1925), Kreps, 25 May 1964 Subsect. BRACTEATAE (Hook. f. & Thoms.) Rehd. L. altmannii Reg. & Schmalh. *var. pilosiuscula Rehd. 18 AA 14999, Rehder, Turkestan 5 May 1927 Subsect. DisTeGIar (Raf.) Rehd. L. involucrata (Richards.) Banks ex Spreng. See Taylor & Mullig, Northern ei 9 18 (1968, p. 109). “ee on CTS and south into 35077 & CT 35434, Rocky om Graham Is., British Colombia Subsect. ge Rehd. L. alpigen f. nana ae Zabel 36+ AA 14994-1, Allen, Central and southern 13 August 1927 European Mts. 36 AA 803-35, Green, 26 May 1965 Il ‘VYAOINOT ‘NAGA ® OUAANACOAA [6961 ¥ * This i is ee first fants etal ofa dsecmerted count for this taxo } Due to an error 2n 18 was incorrectly recorded for this tas in part I, p. 234. 453 “pez ‘d ‘7 yaed ur yueyd sty) 10) papsooas Apoa1s09UI SUM gL = MZ 10119 UB 0} ancy |. “UOXE} SITY} 10J JUNOD pajyuauINoOp & Jo UOTVBITGNd ys1y wae st SL * RUDENBERG & GREEN, LONICERA, II 1969] ‘Sv. uvadoingy u19Y4}NOS puUP [eI]UID ‘SI A*—DOY 0jUI YyyNos puv BoLaUIy U1ay}ION ubysoyIn [, eUuIy.) WIIyION BUIyS U19}S9\4 Buy) U1a}saMm pure [erjuag S96I ABW 9Z ‘uaad ‘SE-EO8 VV L761 ysnany ¢T ‘uanlV ‘I--66b1 WV BIqUIO]OD wre * ‘syT Weyer ‘PEPSE LD ® LLOSE SLO Uo paseg *(60T “d ‘g96T) URSTTINJA WY A0[AR]T, 9S L761 Av ¢ ‘4apyay ‘666¢1 VV $961 AP SZ ‘sday ‘(S261 ‘nsury ‘OLSEL Y90OY) S6STZ VV S961 ‘AON + ‘uaad) ‘6b-£76 VV T¢61 "ydag pI “ysnoy @ 1ysnqoy (82-S22 si pue S96 “AON + ‘uaad) “A-87-S7Z VV S961 APL 8z ‘Ppod & Kajpnq ‘q-IeOIst VV 9¢ +9¢ [PqeZ CadvD) vuvu *} ‘rT Duastdip “7 ‘pyey AVNAIATY “asqns 6 ‘Suaids xa syurg (‘spavyoryy) Dypsonjoaut “7 ‘pyey (Jey) AVIOALSIC “Wasqns ‘pyay vynasnisopid *IwA,y “YPUIEIS ‘Bay wuuvw7yv “7 ‘pyoy (‘SuUIOYL ® '} “YOoH) AVLVALOVAg “JOasqnsg 8T ‘youely mpunurpsaf “7 ‘pyay (CAvWOY) AVIAVOIsaA “}asqns “STEM Optiit “Tx ‘AIO 0700/14 “Tx ‘PYIY AVLVATI “}2asqng JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 454 8961 APY, OL ‘Saaquapny “69-69 VV 6 ueysayxin TL S9OI APPL IE 0} adoiny u1ayseq ‘Uaad) ‘Y-Tp-8872 VV 8I ‘T[ DI14DID} “'T ‘pyay AVOIIVLY | “yesqns CPYPY wnajsojkx0]a0) "}99S) VAAIINO'T “39S OP6l aunt ¢T ‘dad ‘ZOLST VW 81 pyay (ddiq) oyofrsuo) 1A, b96T aunf Z ‘Maas ‘(pEoy ‘AdyINT, ‘9S9T SVJ) bE-956 VV 6 euIy.) U19]SoM CO6I OuNf + 0} IOUT, BISY ‘Kaypnqd ‘W-8£-102 WV SI ‘Wey sipDpuarso “7 996T Av sz ‘S4aquapny pure ‘cg6t ounf + ‘apn ‘q@-8E-86S VV 8T 9961 ABW Sz ‘Suaquapny valoy pur ulpeyseg uosttM) J-ZO1OT WV 9¢ IPIWIYIS “IY SPUaUIDYIDS "IWAy ‘wIxeyy (idny) uzomouwmpu “7 QE6T ABI LZ ‘samog vaI0oy F ‘(ETOT ‘eUIyD ‘ogeor eulyZ) U19y}I0N 4akayy) {-br-LI VV 8T “UNIXBYAL MMOULIDIDI “Ty ‘PYPA CuIXep,) AVHLNVdOH Y *"Wasqns NOILNETALSICT YOLOATIOZ) INV uz u Salads TVAANA‘) NOILVINAWNO0d (panuzjuo)) vsa2uoy] uy sxaquinu awosowory> [eUORIppy “a‘Tav | JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 456 BUIYS) UI9}S9 AA uonraning uedef pue evisy JsvayJION ‘SIA reqTy 07 edoing uoneaAniny 8961 ABW OT ‘Z4aquapny ‘r9-7E9 VV 9961 API 81 ‘S4aquapny ‘bS—989 VV S961 APIA 02 ‘uaad ‘PS-LBS VV S96T APIA TE “‘uaady) ‘Y-LE-prOl WV 8961 APIA OT ‘S4aquapny ‘79-929 VV $961 API 92 ‘sdaay ‘1-99 VV 8961 AP LT ‘wosqty) ‘79-85 VV 9961 API 9Z ‘Saaquapny ‘pe-S9OL VV +961 AP SZ ‘sda4y ‘Y—-9€-881 VV 8961 ACW OT ‘Saaquapny ‘S9-f611 WV 8961 APN OL ‘gaaquapny ‘b9-6L VV 8961 API OT ‘gaaquapny ‘29-b1¢1 WV ‘pyay Dunauyaoy “'T ‘pyay xo unvig DYyJuDsKayd-opnasd XK “Tx ‘Pury CUYysTYy) Dunyadas *} ‘ZIIN], DYIUDSAAYI 'T UPN ‘AD PYPA (T989yY) sypoue “Fx "TT uinaysojax “7 ‘OP Vd 40204; uvx “J ‘pyay sisuauapuanu X< “7 IVaaNar) NOILAGIMLSIC AOLOATIOR ANY NOILVLINAWAI0G uz saloads (panutyuo)) es921UO] UF S19qQuINU aWOSOUIOIYD [euORIppy “aTAV.L ‘UOXE} SIY} 1OJ JUNOD payuauINdOp & Jo UOTJBIT|GNd ys1y Oy} St SIU, » 457 RUDENBERG & GREEN, LONICERA, II 1969] bss Aine ZT RISy Ul9}Sey ‘quad avg ‘I-¢S6 VV 8T “WIXBY (J99MS) VOOLNIN "3995 9961 aunt 6 ‘aajjazuadd py ‘q-6S-€12 VV 8st vAReUTEY 9y} S961 eunf 12 0} uv}sTULYysTY ‘uaady ‘W-6S-E17 VV 8 9961 "3das 61 juny pur ‘sg6t aunt gz ppod & gee FE, A j So ee he, 8 eB #: i - % « . * &. , RUDENBERG & GREEN, LonicerA, II Jour. ARNOLD Ars. VoL. 50 Piate II +! “ . = eal BAT on , 42 . 14 e ° 4 A Re ; we 2 oy Per f * : ~ of tay. = 4-0? < “ane 2 i 10 RUDENBERG & GREEN, Lonicera, II 462 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 NOTES ON WEST INDIAN ORCHIDS, I LESLIE A. GARAY DuRING THE couRSE of routine identifications of collections from various parts of the West Indies, several new species as well as a number of nomenclatorial changes have been noted. A study of the flora of the West Indies is currently under way by Dr. Richard A. Howard of Har- vard University, which will document both the distribution and diversity of all orchid species known in that floristic region. In the mean time, notes, similar to this one, will be published seriatim. Habenaria Dussii Cogn. in Urb. Symb. Antill. 6: 307. 1909. There is a flower from the holotype of H. Dussii Cogn. given by Pro- fessor Cogniaux to the collections of the Orchid Herbarium of Oakes Ames. Since then the type specimen has been destroyed in Berlin dur- ing World War II. This single flower enabled me to identify the fol- lowing two collections reported for the first time outside the island of Guadeloupe. Puerto Rico: Sierra de Luquillo, open grass-sedge savannah, wet, in cloud forest along El Toro trail, south side of El Yunque, R. A. Howard & G. Taylor 18701 (AMES). St. Vincent: St. David Parish, Soufriére Mountain, in tundra-like growth at elevation of 2800 ft. Entire plant green, G. R. Cooley 8446 (AMES). Cryptophoranthus erosus Garay, sp. nov. Fic. 1a-d. Epiphytica, caespitosa, usque ad 3 cm. alta; radicibus crassiusculis, elongatis, satis profusis, flexuosis, glabris; caulibus secundariis erectis, atropurpureis; sepalo postico spathulato-rhombeo, valde concavo, 3- nervio, dorsaliter apicem versus carinato mucronatoque, 14 mm. longo, 6.5 mm. lato; sepalis lateralibus usque ad apicem in synsepalo conniven- tibus, valde concavis, dorsaliter carinatis, acutis, 15 mm. longis, mises se 6 mm. latis; petalis carnosis, subfalcato-lanceolatis, acuminatis, ae nerviis, 4 mm. longis, 1 mm. latis; labello breviter angusteque unguicu- lato, deinde suborbiculari expanso, margine valde eroso; disco utrinque carnoso carinato in medio, antice pectinato, 4.5 mm. longo, 3 mm. lato; —~- 1969] GARAY, WEST INDIAN ORCHIDS, I 463 columna clavata, late alata, clinandrio lacero; ovario cylindrico, verru- coso, 2 mm. longo Dominican Republic: in the vicinity of Constanza. Flowers deep purple. Collected by Rev. Donald Dod and cultivated by him for Bro. Alain H. Liogier 13508 (NY, type!). This new species vegetatively resembles C. sarcophyllus (Rchb.f.) Schltr. from Venezuela, but the latter has broader, entire leaves, as well as dissimilar petals and lip. Pleurothallis Dodii Garay, nom. nov. Basionym: Pleurothallis cryptantha Cogn. in Urb. Symb. Antill. 7: 176. 1912, not Barb. Rodr. 1877. A recent collection by Rev. D. Dod, s.n. (Ny), of this rare species in the Dominican Republic: Las Abejas, Cabo Rojo, has shown that the disc of the lip is covered with fine, but sparsely distributed, hairs as are the margins. This character, although not mentioned in the original description by Cogniaux, is present on the holotype which I have re- cently examined in Bruxelles. Lepanthopsis Dodii Garay, sp. nov. Fic. 2e-f. Epiphytica, caespitosa, usque ad 8 cm. alta; radicibus filiformibus, flexuosis, glabris; caulibus secundariis erectis, gracilibus, vaginis satis distantibus, adpressis, sursum dilatatis hispidulisque omnino obtectis, usque ad 4 cm. longis; foliis tenuibus, ellipticis, acutis vel obtusiusculis, margine muricato-denticulatis, usque ad 2 cm. longis, 6 mm. latis; in- florescentiis capillaribus, subdense multifloris, usque ad 4 cm. longis; bracteis infundibuliformibus, acuminatis, 1 mm. longis; floribus tenuibus, diaphanis, patentibus, glabris; sepalo postico ovato-lanceolato, acuminato, uninervio, 2 mm. longo, 1 mm. lato; sepalis lateralibus inter se usque ad medium connatis, ovato-lanceolatis, acuminatis, uninervis, 2 mm. longis, inter se 1.5 mm. latis; petalis ellipticis vel subrhombeis, acutis vel ob- tusis, 1 mm. longis, 0.5 mm, latis; labello carnoso, triangulari-cordato, 3- nervio, 1 mm. longo latoque; columna humili, crassa, apoda; ovario pedi- cellato 2 mm. longo. Dominican Republic: Polo, epiphytic on trees, Rev. D. Dod 43 (ames, type!). Lepanthopsis Dodii Garay differs from L. acuminata Ames in having smaller flowers, proportionately shorter and broader sepals, and dif- ferently shaped petals. Since my revision of the genus in The Orchid Journal 2: 467-469. 1953, I have examined the types of all West Indian Pleurothallis species. Among these species the following need to be transferred to the genus Lepanthopsis. 464 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Lepanthopsis barahonensis (Cogn.) Garay, comb. nov. Basionym: Pleurothallis barahonensis Cogn. in Urb. Symb. Antill. 7: 177. 1912. Lepanthopsis blepharophylla (Griseb.) Garay, comb. nov. Basionym: Pleurothallis blepharophylla Griseb. Cat. Pl. Cub. 260. 1866. Lepanthopsis dentifera (L. O. Wms.) Garay, comb. nov. Basionym: Pleurothallis dentifera L. O. Wms. in Ceiba 1: 227. 1951. Lepanthopsis Fuertesii (Cogn.) Garay, comb. nov. Basionym: Pleurothallis Fuertesii Cogn. in Urb. Symb. Antill. 7: 178. 1912. Brachionidium ciliolatum Garay, sp. nov. Fic. 3g-j. Epiphytica, parvula, ascendenti, usque ad 7 cm. alta; radicibus fili- formibus, glabris; rhizomate ascendenti, cauliformi, vaginis scariosis, in- fundibuliformibus imbricantibusque omnino obtecto; caulibus secun- dariis vix ullis, monophyllis; foliis pergameneis, oblongo-ellipticis, acutis, subpetiolatis, usque ad 2 cm. longis, 5 mm. latis; inflorescentiis singulis, unifloris; pedunculo capillari, in medio univaginato, usque ad 3 cm. longo; bracteis infundibuliformibus, ovariis pedicellatis aequilongis; floribus pro genere satis parvulis, ciliolatis; sepalo postico ovato-lanceolato, subacu- minato, 3-nervato, margine ciliolato, 7 mm. longo, 4 mm. lato; sepalis lateralibus usque ad apicem connatis, ibi bidentatis, ellipticis, obtusis, 4- nervatis, margine ciliolatis, 6 mm. longis, 4 mm. latis; petalis ellipticis, apice subito in apiculo triangulari-subfalcato, acuminato productis, 3- nervatis, margine ciliolatis, 6 mm. longis, 4 mm. latis; labello carnoso, € cuneata basi subsigmoideo, antice triangulari, acuto, 3-nervato, margine valde ciliolato; disco callo pulvinari, antice exciso ornato; toto labello 3 mm. longo, 2.5 mm. lato; columna humili, crassa, vix 1 mm. alta; ovarlo pedicellato ca. 2 mm. longo. Puerto Rico: Pico del Oeste, Sierra de Luquillo, 1020 m. alt. Epiphytic or chid, plants with 3-4 leaves; flowers yellow-green, apparently do not open. Study trail area. R. A. Howard & L. I. Nevling 16929 (ames, type!). This new species closely resembles B. parvum Cogn. both in size and in general appearance. It differs, however, in the shape of the floral segments which are not caudate. Both B. tetrapetalum (Lehm. & Kral.) Schltr., and B. simplex Garay, although similar in appearance to B. cilio- latum Garay, have dissimilar and eciliate lips. Epidendrum isochilum var. tridens Rchb. f. in Ber. Deutsch. Bot. Ges. 3: 277. 1885. Syn.: Epidendrum belvederense Fawc. & Rendle in Jour. Bot. 47: 123. 1909. 1969] GARAY, WEST INDIAN ORCHIDS, I 465 Ficures 1-4, West Indian orchids. Fic. 1, a-d, Crypto phoranthus erosus Garay; y Fic. 2, e-f, Pig Wass Dodii Garay; Fic. 3, g-j, Brachionidium cilio latum Gar aray; Fic. 4, k-n, Campylocentrum constanzense Garay. All greatly magnified. 466 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 There appears to be no distinction between Epidendrum belvederense Fawc. & Rendle and E. isochilum var. tridens Rchb. f. as a study of the holotypes indicates. Judging from the number of specimens which I have examined of this species from the Dominican Republic, this variety seems to be much more common than the typical variety, which is de- scribed as having an entire lip. Epidendrum neoporpax Ames in Bot. Mus. Leafl. Harvard Univ. 2: Liz, 1934. Basionym: Epidendrum Porpax Rchb. f. in Flora 48: 278. 1865 not Rchb. f. Syn: Epidendrum vestitum Ames in Sched. Orch. 4: 51. 1923. Epidendrum Porpax var. domingensis Cogn. in Urb. Symb. Antill. 7: 181. 1912. This rather rare Cuban species has been found recently in Costa Rica, and rediscovered by Mr. Ariza Julia, s.n., in the Dominican Republic: Sabaneta de Yasica, Puerto Plata Province. An examination of the type of Epidendrum Porpax var. domingensis Cogn. in the Bruxelles herbarium convinces me that it is identical with E. neoporpax Ames. Epidendrum Sintenisii Rchb. f. in Ber. Deutsch. Bot. Ges. 3: 241s Syn.: Epidendrum monticolum Fawc. & Rendle in Jour. Bot. 47: 124. 1909. Recently I had the opportunity to examine and to compare the holo- types of E. Sintenisii Rchb. f. and E. monticolum Fawc. & Rendle. As a result of this study, I am convinced that they are conspecific. Epi- dendrum Sintenisii is now recorded from Puerto Rico and Jamaica. Stellilabium minutiflorum (Krzl.) Garay, comb. nov. Basionym: Telipogon minutiflorus Krzl. in Ann. Nat. Hist. Mus. Wien 33: 14. 1919. Syn.: Telipogon Lankesteri Ames Sched. Orch. 3: 23. 1923. Stellilabium Helleri L. O. Wms. in Brittonia 14: 443. 1962. This rather rare Costa Rican species has recently been found in the Dominican Republic: Casalito Bonao by Rev. D. Dod, s.n. (NY): This is also a new record for the West Indies. Stellilabium Helleri L. O. Wns., of which I also have studied the holotype, agrees in every respect with Kraenzlin’s type material which I examined in Vienna. Telipogon Lankesteri Ames likewise, does not offer any criterion by which it could be kept separate from S. minutiflorum (Krzl.) Garay. Polyradicion Garay, gen. nov. Pfitzer in describing the genus Polyrrhiza stated that it consists of four West Indian species. Of these four he mentioned only one in making BIB = een ance 1969] GARAY, WEST INDIAN ORCHIDS, I 467 an Official transfer, namely P. funalis (Sw.) Pfitz. Thus, the genus Poly- rrhiza is typified by this species. In Flora of Jamaica, Fawcett and Rendle regard P. funalis (Sw.) Pfitz. to be a synonym of Dendrophylax funalis (Sw.) Benth., a judgment which I consider to be correct. Since Poly- rrhiza automatically becomes a synonym of Dendrophylax through this transfer, it leaves the other species without a validly published generic name. Since there are only two species involved I reject the idea of con- servation in favor of a new name which I propose here with the same etymological meaning as was used by Pfitzer. The genus is, thus, charac- terized as follows: Sepala petalaque simillima, aperta, lanceolata; labellum maximum, 3- lobum, lobi laterales quam lobum intermedium multoties breviores, basi in calcari valde evolutum producta; columna humilis, crassa, apoda, basi labellum adnata; clinandrium humile; anthera incumbens, opercularis; pollinia 2, stipiti nudi, distincti glandulae affixa. Plantae epiphyticae, aphyllae; radices crassae, valde evolutae; caules vix ulli; pedunculi laterales, graciles, arcuati, abbreviati, semper uniflori; flores majusculae. Species 2, Indiae Occidentalis incolae. Typus: Angraecum Lindenii Lindl. Polyradicion Lindenii (Lindl.) Garay, comb. nov. Basionym: Angraecum Lindenii Lindl. in Gard. Chron. 135. 1846. Syn.: Aeranthus Lindenii Rchb. f. in Walp. Ann. Bot. Syst. 6: 902. 1864. Dendrophylax Lindenii Benth. ex Rolfe in Gard. Chron. ser. 3. 4: 533. 1888. Polyrrhiza Lindenii Cogn. in Urb. Symb. Antill. 6: 680. 1910. Distribution: Florida, Cuba. Polyradicion Sallei (Rchb. f.) Garay, comb. nov. Basionym: Aeranthus Sallei Rchb. f. in Walp. Ann. Bot. Syst. 6: 902. 1864. Syn.: Dendrophylax Sallei Benth. ex Rolfe in Gard. Chron. ser. 3. 4: 533. Polyrrhiza Sallei Cogn. in Urb. Symb. Antill. 6: 680. 1910. Distribution: Dominican Republic, Haiti. Dendrophylax gracilis (Cogn.) Garay, comb. nov. Basionym: Polyrrhiza gracilis Cogn. in Urb. Symb. Antill. 6: 679. 1910. An examination of the holotype, Wright 3300, in the Orchid Herbarium of Oakes Ames has shown clearly that it is referable to the genus Dendro- phylax Rchb. f. It is closely allied to D. hymenantha Rchb. f., differing in its shorter, 1-flowered peduncle and in the size of its flowers which are twice as large. Dendrophylax hymenantha Rchb. f., however, has been united with D. varius (Gmel.) Urb., but this decision requires further study. 468 JOURNAL OF THE ARNOLD ARBORETUM [VvoL. 50 Campylocentrum constanzense Garay, sp. nov. Fic. 4k-n. Epiphytica, caespitosa, aphylla, usque ad 4 cm. alta; radicibus nu- merosis fasciculatis, filiformibus, flexuosis, glabris; caulibus nullis vel vix ullis; inflorescentiis numerosis, fasciculatis, erectis, capillaribus, sim- plicibus vel dichotome ramosis, supra laxe plurifloris, omnino setaceo- hirsutis, usque ad 4 cm. longis; bracteis ovato-cucullatis, acutis vel ob- tusiusculis, extus setaceo-hirsutis, 1 mm. longis; floribus minimis, hyalinis; sepalo postico ovato, acuto vel obtusiusculo, uninervio, extus sparse setaceo- hirsuto, 1.5 mm. longo, 1 mm. lato; sepalis lateralibus oblique ovatis, obtusiusculis, extus setaceo-hirsutis, 3-nerviis, 2 mm. longis, 1 mm. latis; petalis subfalcato-ovatis, obtusiusculis, 3-nerviis, glabris, 1.5 mm. longis, 1 mm. latis; labello anchoriformi-lobato, antice breviter apiculato, basi calcarato, calcari cylindrico obtuso, setaceo-hirsuto; disco in medio lon- gitudinaliter carinato, antice setaceo-hirsuto; toto labello 3 mm. longo, antice 1.5 mm. lato; columna humili, crassa, vix 1 mm. alta; ovario cylindrico, muricato-hispidulo, cum pedicello 2 mm. longo. Dominican Republic: Constanza, epiphytic on trees, Rev. D. Dod 66 (AMES, type!). This species is quite unique in the Section DENDRopHYLopsis Cogn. be- cause of the anchor-shaped lip and a distichously branching, setaceous in- florescence. BoraNnicaL MusEuM HARVARD UNIVERSITY 1969] SMITH, POLLEN OF AFRICAN VERNONIA 469 POLLEN CHARACTERISTICS OF AFRICAN SPECIES OF VERNONIA C, Ear LE SMITH, JR. DURING A TAXONOMIC sTUDY of the species of section STENGELIA of the Composite genus, Vernonia, pollen of a number of species was examined. Whenever possible, a floret from a specimen of the type collection was dissected and the anthers macerated in lacto-phenol and methylene blue. A single grain from each slide was photographed. Size was determined by measurements of ten grains from each slide, after scanning to ascertain whether the grains measured fell into more than one size class. Measure- ments were made to the outside of the reticula, but did not include the length of spines. Obviously, not all of the specimens examined belong to section STEN- GELIA, although all of the species have been assigned here because of thin terminal appendages on otherwise firm or chartaceous phyllaries. Per- haps a future student of the genus will find corollary characters on which the sections of the genus can be more firmly based. The large number of species involved in Africa alone precludes this in my short-term ex- amination of the section STENGELIA. Pollen sizes range from an average of 29.1 » for Vernonia praecox Welw. ex O. Hoffm. to 69.5 » for grains of V. wittei Hutch. & Burtt (Fic. 6). The average pollen diameter is 51.9 ». The largest number of species with a similar pollen size fall into the next highest size class, 53.3 ». A total of seven species have an average pollen size of 51.7 ». Twenty-six species fall into larger pollen-size classes and 24 species fall into smaller pollen- size classes than the 15 species in the median groups. Thus, except for the few plants having either very large pollen grains or very small pollen grains in relation to the average pollen size for this group of species, the species are well clustered with the greatest number falling centrally. No attempt was made to study the anatomy of the pollen grains of this group. Morphologically, all of the grains examined are similar. All are nearly spherical and are evidently tricolporate, although this some- times is difficult to determine (Fic. 4). In all of the pollen examined, the outside of the grain is marked by a raised reticulum. T his may be thin, but in the majority of the species the surface reticulum is moderately to heavily thickened. Often, the pattern of the reticulum is very regular with polar alveoli surrounded by a ring from which radiate, at regular intervals, a series of bars. These are crossed on the sides of the grains at regular intervals by bars of equal size. On one side of the grain a longitudinal alveolus extends from pole to pole and one of the pores oc- curs in this at the equator (Fic. 1). Frequently, the polar rings are 470 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Ficures 1-5, pollen grains of Vernonia species. Fic. Vernoma guineensis var. poi average pollen meter 51.74; the seed grain, in polar view, shows the regularity of the reticulation common to many of the upright or shrub- 1969 | SMITH, POLLEN OF AFRICAN VERNONIA 471 broken by this longitudinal opening so that if flattened, the total open area would resemble a dumbbell. On the sides of the grain, the alveoli do not appear to be geometrically balanced in any of the species. Where the reticulation follows a similar pattern in all of the grains examined, I have called it regular in spite of a lack of an exact geometric pattern. In many species, polar areas may be defined or not, but the reticulation over the remainder of the surface of the pollen grain appears to be randomly placed (Fic. 5). I have called this type of reticulation irregular. In most of the species, the reticulation is further decorated by spines arising from the top and sides, The spines all appear to be simple conical protuberances of the same material from which the reticulum is formed. V. albo-violacea De Wild. has almost no spines on the reticulum. The spines on pollen grains of other species vary from small to large, but they are uniform in size for a species. Only V. bojeri Less., V. gerberiformis Oliver & Hiern, V. mandrarensis Humbert, and V. prolixa S. Moore lack spines completely. The reticulation of V. gerberiformis is unusual among the species ex- amined. The reticulation is relatively narrow and produced upward from the surface of the grain in a wing-like projection (Fic. 3). I shall make no attempt to formulate an adaptive advantage for this deviation from the usual pattern among the Vernonia pollen grains studied. With an average pollen diameter of 66.3 » (measured at the outside of the reticu- lum), this is the second largest grain examined. The surface of the pollen grains was examined under oil immersion (485 &) in order to study the surface details. In none of the grains was a regular pattern or design seen. Some of the reticula and pollen surfaces are not smooth. The roughness is not readily discernible and does not appear to be produced in a regular pattern. However, the reticulation on some of the species (for example, V. adoensis Sch.-Bip. ex Walp., V. ger- beriformis) is not completely contiguous with the surface of the pollen grain. When the reticulum is heavy, it frequently stands above the grain surface on a series of pedestals which may be relatively far apart or rather close together (Fic. 2). The presence of the pedestals negates the possibility that this is an artifact created when the surface of the om shrinks away from the reticulum. In the pollen grains of other species with an equally heavy reticulum, the elevation of the reticulation above the surface does not seem to be present. ll m median section. Fic. 5, Vernonia castellana, this shows the rosette of leaves. 472 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 On the basis of the gross morphological characteristics of specimens and on a detailed study of the achenes and flowers, the species have been grouped into clusters which show similarities. Pollen grain data have been regrouped (TABLE 1) on this basis (excluding species which apparent- ly do not belong to section SrENGELIA). In general, the characteristics of the pollen grains seem to support the groupings by overall morphology. For instance, the first group of three species, V. /asiopus O. Hoffm. in Engl., V. brownii S. Moore, and V. albo-violacea, varies relatively little in average pollen grain diameter; the grains have reticula of medium thick- ness with short or nearly absent spines. Number 5 of 4 Species 3 54 3; 4 433 Sis 34 64 695 Average Diameter of Pollen Grains Ficure 6. Graph illustrating the number of species of Vernonia, section STENGELIA, with pollen falling into each size class. Note that pollen size for most of the species falls near the median, 51.9z. The second group of species, V. polyura O. Hofim., V. filigera Oliver & Hiern, V. longipetiolata Muschler, and V. oxyura O. Hoffm. in Engl., again agree well in pollen characteristics as well as in overall morphology, €X- cept for grain size in one species. The pollen grain size of three of the species ranges from an average diameter of 40.4 » to 46.9 ». The average pollen grain size of grains of V. filigera is 56.6 uw. It is hardly desirable to exclude the species from this grouping on this one feature alone, but it does necessitate another careful look at the specimens to be included here. The break in size observed in the example cited above is perhaps better illustrated in the group of species clustered around the type species of the section, V. adoensis Sch.-Bip. ex Walp. On the basis of their gross morphology, these species fall readily into a group. An examination of the details of the achenes and flowers discloses no major discrepancy 10 the pattern. For the most part, the morphology of the pollen grains of these species supports the grouping. Average pollen diameters for most species of the group range between 56.6 » and 63.0 ». However, the average pollen grain diameters of V. shirensis Oliver & Hiern and V. woodii O. Hoffm. (which are now considered synonymous) are 46.9 p. 1969 | SMITH, POLLEN OF AFRICAN VERNONIA 473 The species of section STENGELIA fall into two distinct groups on the basis of plant habit. The bulk of the species are rank-growing upright sub-shrubs from a perennial base, or upright shrubby plants. A few may become tree-like. The pollen of many of these species has an average diameter of 50.1 » or more, except for the species grouped around V. polyura. The reticulum on the grains is generally heavy. The other group of species is distinguished by a rosette habit with flowers borne on a, usually, leafless scape. The scape may be unbranched and support a single head or it may support several heads. Many of the average pollen grain diameters for this group are less than 50 ». The reticulation on the grains is often thin and very irregular. However, more exceptions occur among the species with basal rosettes than among the species with an upright habit. For example, the pollen of V. gerberiformis, which was previously described, is very different from the usual pattern of grain size and morphology. The grains of V. myassae Oliv. in Hook., V. pumila Kotschy & Peyr. and V. anandrioides S. Moore have a heavy, regular reticulum. Furthermore, the grains of the last two have an average diameter of 53.3 ». The pollen grains of the species with a basal rosette are more variable in size and morphology than are the pollen grains of the other species assigned to section STENGELIA. In only three of the species examined, pollen grains of two size classes occurred. Both V. calvoana (Hook. f.) Hook. f. and V. insignis (Hook. f.) Oliver & Hiern have pollen grains similar in size and morphology. In both, the larger grains appeared to be normal. The smaller pollen grains appear to have been aborted. Because they were removed from herbarium specimens, it was impossible to apply germination tests for viability of the grains to confirm my assumption that the smaller grains are not functional. About half the pollen grains of V. achyrocepha- loides Hutch. & Bruce were also smaller than the grains measured, and appeared to be nonfunctional. So little is yet known about the biology and genetics of species of section STENGELIA that I can make no as- sumptions as to the cause of the difference in pollen grain sizes. It is, perhaps, significant that all of the species have average pollen diameters near the upper limit of pollen size for this group of 64 species. SUMMARY 474 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 achenes and floral morphology, pollen morphology and average pollen diameter confirms the groupings for the most part. The species traditionally included in section STENGELIA can be divided into a group with basal rosettes and heads borne on scapes versus a group of upright sub-shrubs or shrubs (rarely tree-like) with heads on the sides or ends of branches. In general the first group has smaller pollen grains with thin, irregular reticula. The second group generally has larger pollen grains with heavy, more regular reticula. The group with basal rosettes is less homogeneous than the other in regard to pol- len size and morpholo In only three species of the 64 examined were pollen grains of two size classes found. In all instances, the smaller grains appeared to be nonfunctional. All three species have pollen diameters in the upper size range. Insufficient knowledge of the biology and genetics of these species precludes an explanation. TABLE 1. Comparison of pollen characters with gross morphology of species of Vernonia section Stengelia POLLEN RETICULATION SPINES SPECIES SIZE, THICKNESS PATTERN LENGTH DISTRIBUTION PROVENIENCE V. albo-violacea $i7 Medium Regular Almost none Bequaert 492, Congo V. Ob i 56.6 Irregular Short Frequent Brown 2656, Uganda V. lasiopus Sa.3 ss Regular = Occasional Volkens 444, Tanzania V. oxyura 40.4 Thin + Regular Medium Numerous Buchanan s.n., Malawi V. longipetiolata 42.0 Medium Regular 7 Occasional Kassner 2746, Congo V. p 46.9 Thin Irregular Large si Goetze 866, Tanzania V. filigera 56.6 Medium . Medium Numerous Schimper 1530, Ethiopia V. nyassae 48.5 Heavy Regular Small Frequent Thomson s.n., Zambia? V. swynnertonii 51,7 ae Irregular Short +Numerous Swynnerton 1908, Rhodesia V. gerberiformis 66.3 Winged = None Schweinfurth 2688, Sudan? V. wittei 69.5 Medium i Short Frequent de Witte 543, Congo V. chthonocephala 35.6 Thin a ae Welwitsch 3886, ing V. subaphylla 38.8 ° is o Few Carson 10, Zam V. praemorsa 40.4 Medium 2 is Occasional Stolz 104, Hotaeny V. agricola 43.6 Thin si Medium Numerou Kassner 2136, Zambia V. castellana 48.5 sg 2 Gossweiler 2883, Angola V. anandrioides 53.3 Heavy Regular Short Occasional Gossweiler 2132, Angola V. pumila ae Medium + Regular ‘i Frequent Elliot 7037, Kenya? V. homilocephala 54.9 + Heavy Irregular ‘s Numerous Elliot 7058, Kenya? V. eat 50.1 Heavy " Short Ww Homblé 881, Congo V. pleiotaxo |e si fc : Frequent Quarré 2654, Congo V. procera 53.4 . ° Numerous Chevalier 7899, Congo? V. lancibracteata 58.2 6 Regular Ha Frequent Eyles 291, Zambia V. firma 50.1 Medium Regular Short Frequent Schweinfurth 3153, Sudan? V. vallicola 58.2 Heavy ‘s Medium a Gossweiler 3781, Angola VINONUAA NVOIMAV AO NATIOd “HLIWS 6961 SLY TABLE 1. Comparison of pollen characters with gross morphology of species of Vernonia section Stengelia (Continued ) POLLEN RETICULATION SPINES SPECIES S1zE yp THICKNESS PATTERN LENGTH DISTRIBUTION PROVENIENCE V. uni 51.7 Heavy Irregular Medium Numerous Braun 1979, Tanzania V. calvoana var microce phala S3:3 Medium Regular Short Frequent Lightbody 26259, 2 oC \3 - i | | i ANS Map 2. Distribution of Flindersia pimenteliana F. Muell. Morwood NGF 6204 (a, BO, BRI, K, L, LAE, NSW 99656, US), McAdam 291 (LAE), Ross NGF W1001 (a, BO, BRI, L, LAE, NSW 99660), Womersley & Brass N GF 11025 (A, BRI, K, L, LAE); W of Bulolo near Bulolo-Watut Divide, Frodin & Hill NGF 26355 (x, L, LAE, Nsw 99658), Upper Long Iskand Creek, near Bulo- lo, Havel & Henry NGF 17024 (a, BO, BRI, CANB, K, L, LAE), Upper Nauwata- Banda logging area, near Bulolo, Havel & Kairo NGF 11140 (BRI, K, 1, LAE), NGF 11142 (Bo, K, L, LAE); near Dengalu Village, Womersley NGF 19063 (kK, L, LAE, NSW 99668); Wau, White NGF 2529 (1, LAE, Nsw 99655); Edie Creek, 1969 | HARTLEY, THE GENUS FLINDERSIA 497 Streimann NGF 17476 (x, LAE); Kauli Creek, 5 miles S of Wau, Hartley 11513 (A, LAE), Henty NGF 14726 (x, L, LAE, NSW 9 9654), Millar NGF 14516 (k, L, LAE, NSW 99657), van Royen NGF 16302 (B0, L). Papua. CENTRAL DISTRICT: Lala River, Carr 15805 (cans, L), 15989 (kK, L); Isuarava, Carr 15475 (1), 15569 (L), 15969 (kK, Lt); Mafulu, Brass 5339 (a, Ny, US); Boridi, Carr 13152 (A, K, L, NY), 14408 (A, K, L, NY), 14808 (L), 14910 (A, K, L, NY); Mt. Obree to Laruni Spur, Lane-Poole 382 (BRI). NORTHERN District: Managalase area, S side of Hydrographers Range near Siarane, Pullen 6263 (CANB); Bariji- Managalase area, N side of Sibium Range, S of Toma, Pullen 6358 (CANB, LAE), 6387 (CANB, LAE). Queensland. Cook District: Cape York Peninsula, Mt. Finnegan, Brass 20322 (a); Great Dividing Range ca. 6 miles S of Mossman and near “Devil Devil Creek”, Smith 3953 (prt); Mt. Spurgeon, White 10596 (a, K); Danbulla, Jones 1117 (c ANB); Atherton Tableland, Juara Creek area, near Danbulla, Smith 3780 (ert); Atherton Tableland, Lake Barrine, Kajewski 1114 (A, BRI, Ny, P); Atherton, Mocatta, February 1913 (prt); Atherton Area, Webb E Yungaburra, Dreghorn, December 1935 (A, BRI, K); Gadgarra, Dreghorn 22E (BRI), January 9, 1934 (A, BRI, NY), White 1566 (A, BRI); Glenallan, Malanda, Hayes (srt); Paronella Park, on Mena Creek, ca. 14 miles S sel Innisfail, Smith, August 5, 1948 (srr). NortH KENNEDY District: Koolmoon Creek, ca. 11 miles SSE of Ravenshoe, Smith 4588X (srt); Evelyn, Bailey, oak 8, 1899 (BrI-holotype of Flindersia mazlini F. M. Bailey; x-isotype); Coast Range, Anonymous, February 1866 (BRI); oe Bay, Dallachy cee Geniece of Flindersia pimenteliana F. Muell.; , BO, K, NSW 99651), Anonymous (BRI, MEL); Mt. Macalister, Dallachy, Apri on 1869 (MEL); Mt. Fox, Clemens, September-December 1949 (GH, MICH C. T. White (1921) has previously placed Flindersia mazlini in the Synonymy of F. pimenteliana. The type collection of Flindersia chrysantha was made at 2300 m. in West New Guinea and is typical of a number of collections from mountain rain forests in New Guinea. The leaflets of these collections tend to be of thicker texture and have more prominent veins and less tapering bases and apices than typical F. pimenteliana. These are not sharply defined differ- ences, however, and probably are only environmental modifications. Mer- rill and Perry noted in their original description of F. chrysantha that the petals differed from those of F. pimenteliana in color (yellow vs. red) and pubescence (glabrous vs. pubescent abaxially). With the larger number of collections that are now available it can be seen that both of these characters are quite variable in typical F. pimenteliana from lower eleva- tions in New Guinea. Flower color ranges from red to pink, cream or white, and the petals range from pubescent to glabrous abaxially. F aes sex and the distribution of,sexes is extremely variable in this species. In some collections, such as Moll BW 9745 and Dreghorn, Decem- ber 1935, all of the flowers appear to be perfect. In others, such as Dreghorn, January 9, 1934, some appear to be perfect and many appear to be functionally staminate. A condition where all of the flowers on a specimen appear to be functionally staminate is found in a number of collections from New Guinea including Brass 29153 and Carr 14910. Still other collections, such as Brass 11128 and Hartley 12018 have a majority 498 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 of functionally carpellate flowers mixed with flowers that appear to be functionally neutral Insect galls are often formed from the flowers in New Guinea. These have the appearance of young fruits and were occasionally mistaken for them by collectors. Other than Flindersia unifoliolata, which seems to be very closely re- lated, F. pimenteliana does not appear to have any very close relatives. As is indicated above in the outline of species relationships, I think its having exclusively simple trichomes would place it closer to those species which share that character than to any of the others, but the leaf differ- ence and especially the difference in hypocotyl position makes any very close relationship seem unlikely 5. Flindersia unifoliolata Hartley sp. nov. Arbor usque 15 m. alta; ramulis, rhachidibus et laminis subtus glabris vel minute et sparse puberulis, pilis simplicibus. Folia opposita vel sub- opposita, unifoliolata vel interdum imparipinnata et unijuga, 3-9 cm. longa; petiolulis foliolorum lateralium 1—4 mm. longis, rhachidi ad apicem extensa usque 1 cm. longa foliolum terminale ferente; petiolis foliorum unifoliolatorum usque 1.8 cm. longis; laminis subcoriaceis, consperse pellu- cido-punctatis, ellipticis, equilateris vel parum inaequilateris, 3-8 cm. longis, 1.4~3.2 cm. latis, basi acuta usque cuneata plerumque aequilatera, venis primariis utrinque 8-16, apice obtuse acuminata. Capsula secedens in valvas distinctas maturite, elliptica, 7.8 cm. longa; exocarpio in sicco atro-rufescente, glabro, muricato, processibus inaequilongis, usque 2 mm. longis; endocarpio brunnescente et leviter ferrugineo-maculato. Semina 2 inquoque latere dissepimentorum, utrinque alata, 3.5—-4 cm. longa; hypo- cotylo laterali parum adscendente. Flores non visi. Holotypus: Sayer 136 (MEL). Queensland. Coox District: Mt. Bellenden Ker, alt. 5200 ft., Sayer 136 (MEL-holotype); Mt. Bartle Frere, in low scrub, 4000-5000 ft., Martin & Hy- land 1881 (pr). The localities of the above collections are about 15 miles apart about 40 miles south of Cairns, Queensland. The holotype is a fruiting branch while the Martin & H yland specimen is sterile. Closely related to Flindersia pimenteliana which appears to be restricted to lower elevations in this part of Queensland. The fruits and seeds of the two species appear roughly identical. 6. Flindersia amboinensis Poir. in Lam. Encycl. Suppl. 4: 650. 1816. Neotype: DeVriese & Teysmann, Moluccas, Ceram. Arbor radulifera Poir. in Lam. Encycl. 6: 58. 1804 (provisional name, based on plate and description by Rumphius, Herb. Amboin. 3: 201. ¢. 129. 1743). Flindersia earbet nce Spreng. Geschicht. Bot. 2: 76. 1818 se illegit.). 1969 | HARTLEY, THE GENUS FLINDERSIA 499 Flindersia oo Lane-Poole ex White & Francis, Proc. Roy. Soc. — 38: 232. t. 3. 1927. Type: Lane-Poole 362, Papua, Owen Stanley ange. Large trees to 45 m.; outer bark gray to brown, smooth or slightly roughened; inner bark usually yellow grading to white or yellow-brown toward the cambium; sapwood white, cream, or yellow; heartwood yellow- brown; branchlets, leaves and inflorescences glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves alternate, im- paripinnate or (occasional leaves) paripinnate, 18-57 cm. long; rachis glabrate to appressed- or soft-pubescent; petiolules of lateral leaflets 2-8 mm. long, terminal leaflet on an extension of the rachis 1-4.3 cm. long; leaflets 2-4(-5) pairs, chartaceous to subcoriaceous, with or without scattered pellucid dots, glabrous to appressed- or soft-pubescent below, glabrous or occasionally finely pubescent along the midrib above, elliptic to elliptic-lanceolate, usually strongly unequal-sided and_ occasionally falcate, 8-20 cm. long, 3-9.5 cm. wide, base acute to broadly rounded, usually oblique, main veins 10—20 on each side of the midrib, apex obtuse to acuminate. Inflorescence terminal, 18—30 cm. long, usually about as wide as long, axes and branches densely appressed- to soft-pubescent. Flowers bisexual, 3.5~5 mm. long; pedicels obsolete to 1.7 mm. long; sepals densely appressed-pubescent, ciliolate, broadly ovate to suborbicular, 0.7—1 mm. long; petals yellow-brown, cream, or red, glabrous to densely ap- pressed-pubescent abaxially, glabrous to subvillous at about the middle adaxially, elliptic-oblong, 3-4.5 mm. long; stamens declinate, 2.2—3.6 mm. long, filaments pilose subapically or (rarely) glabrate, anthers dorsifixed, obtuse, 0.7-1 mm. long; staminodes 0.7—1.8 mm. long; disc 0.7—1.3 mm. high; gynoecium about 2 mm. high and about 1.5 mm. wide, ovules 3 on each side of the placentae. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic to elliptic-oblong, 9-21 cm. long; exocarp drying light to dark brown, densely and minutely pubescent to glabrate, muricate, the excrescences thin and flattened laterally or thick and conical, 5-8 mm. long; endocarp reddish brown or cream. Seeds 3 on each side of the dissepiments, winged at both ends, 5—-8.5 cm. long; hypocotyl lateral, horizontal. ILLustraTion. Rumpuivs, G. E., ibid. Distr1BuTION. Ceram and Tanimbar Islands in the Moluccas eastward throughout New Guinea: rain forests from sea level to 1700 meters. See Map 3 Moluccas. CERAM: without definite locality, DeVriese & Teysmann, 1859- 1860 (L-neotype of Flindersia amboinensis Sa are bb 25869 (L). TANIM- BAR IsLanps: Otimmer, NJFS bb 24346 (L). t New Guinea (West Irian) and neighboring islands. AROE ISLANDS: aa Island, Dosimanalaoe, ste bb 25298 (1), NIFS bb 25322 (1); Trangan Island, Ngaibor, N/JFS bb 3 (L). Watceo IsLaNp: E bank of Majalibit Bay 8 km. NE of Waifor, begs sig Sal 5165 (CANB, L). SALAWATI IsLAND: Kaloal, Koster BW 1495 (tL). JAPEN IsLanp: Seroei Aisaoe, Sebosiari, Jwanggin BW 10054 (1); Seroei, 500 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 | | Map 3. Distributions of Flindersia amboinensis Poir. (dots) and F. acumt- nata C, T. White (half-filled circles). Mariattoe, NJFS bb 30424 (L). VocELKop Division: Sorong, van Royen 3505 , BRI, K, L, LAE); Warsamson Valley, E of Sorong, Jwanggin BW 5643 (Lt), Moll BW 11623 (1, Laz); Kebar Valley, Koster BW 7116 (1), Schram BW 7805 (L); pir Twanggin BW 5780 (L, LAE), Koster BW 4433 (BO, CANB, L, LAE), BW 4470 (cANB, L, LAE), BW 4474 (1), BW 4497 (cANB, L, LAE), BW 7062 (esx, L, LAE); Wa kd, ca. 50 km. W of Manokwari, Schram BW 7615 (L); i Plain, Koster BW 11059 (1, LAE), Schram BW 1813 (CANB, L), pilieorg BW 10444 (L); 8 km. NW of Manokwari, Koster BW 4350 (CANB, L, LAE); Lower Pami River, ca. 5 km. N of Manokwari, Koster BW 4359 (Bo, = a Manokwar Koster BW 11890 (1 L, LAE); Momi, Kostermans 321 (t), 337 (L), 8 (L); Wariap, Kostermans 481 (1); Inanwatan, Moetoeri (Steenkool), ies bb 32680 (L). Hotranpra Division: Hollandia, Schram BW 1678 (CANB, SouTHERN Division: Opka, ca. 10 km. NE of Ninati, Subdivision Moejoe, Kalkman BW 6454 (case, L, LAE). Territory of New Guinea. SEPIK DISTRICT: Wewak-Angoram Area, 3 miles E of Urimo, Saunders 975 (CANB, LAE); 3 miles N of Angoram, Pullen 1882 (A, L, LAE); without definite locality, Leder- mann 10406 (L). Mapanc District: Usino, Henty NGF 27500 (K, LAE); Hoogland 5035 (A, BM, BRI, K, L, LAE, US). MOoROBE DistRICT: Oomsis Creek, Bulolo, Dobson & Havel NGF 9116 (Lae). Papua. WeEsTERN District: Lake Daviumbu, Middle Fly River, Brass 7517 (A, BRI, L, LAE); Lower Fly River, E bank opposite re Island, Brass 8032 (a); Upper Wassi Kussa sig Brass 8634 (A, BRI, L, LAE); Oriomo River, Hart NGF 5018 (Bo, CANB, L, Nsw 99661), White . Gray NGF 10366 (A, 80, BRI, K, L, LAE, NSW 09662). 1969 | HARTLEY, THE GENUS FLINDERSIA 501 oa Brown River, Allan & Jones NGF 2751 (srt, CANB, L, LAE); near Karema, (A, BO, K, L, LAE, NSW 99663); Owen Stanley Range, Lane-Poole 362 (Brt- holotype of Flindersia macrocarpa Lane-Poole ex White & F rancis; K-isotype). NorTHERN District: Yodda River, Carr 13913 (a); Dobodura Area, NGF 2063 (LAE); ca. 1 km. W of Popondetta, Hoogland 3737A (CANB, LAE); Mana- galase Area, S side of Hydrographers Range near Siurane, Pullen 5584 (CANB). Cultivated. Java. Botanic Gardens, Bogor, Anonymous, January 1890 (Bo, L). Not, as the specific epithet implies, known from Ambon Island. In the original description in Herbarium Amboinense Rumphius gave Ceram as the locality. As delimited here, this is probably the most variable species in the genus. Among the variable features are: Petals glabrous to densely appressed-pubescent abaxially. Petals glabrous to subvillous adaxially. Stamens 2.2--3.6 mm. long. Filaments glabrate to pilose subapically. Capsules 9-21 cm. long. Exocarp of capsule glabrate to minutely pubescent. Excrescences of exocarp flattened laterally or thick-conical. Endocarp reddish brown or cream colored. OPI AKRON IT have not been able to find sufficient correlations among these characters to recognize more than a single taxon. The color of the endocarp, for example, does not always correlate with the pubescence of the exocarp and neither of these characters consistently correlate with the shape of the excrescences or the size of the capsule. In the flowers, the only tendency toward correlation is between petal pubescence and stamen length. Per- haps some definite correlations could be made if flowering and fruiting material could be studied for each variant. Where I have been able to do this, however, studying flowering and fruiting collections that were ob- viously the same morphologic type, it has seemed unlikely. Finally, it should be noted that each of the variable features of this species was found to be variable in one or more of the other species of the genus as well. Flindersia macrocarpa, originally described as having larger capsules than had previously been attributed to any other species of the genus, now grades into typical F. amboinensis. Obviously very closely related to Flindersia acuminata, which differs mainly in having smaller leaves, narrower, more acuminate leaflets, and shorter stamens. Neither of these species appears to be very closely re- lated to any of the other species of the genus. 502 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Flowers that were apparently insect-stung were noted in several collec- tions. They are atypical in having enlarged, almost cartilaginous petals, filaments, staminodes, and ovaries. Remains of insect larvae were found in the ovaries of some of the affected flowers. 7. Flindersia acuminata C. T. White, Queensl. Dept. Agr. Bot. Bull. 21: 5. ¢. 2. 1919. Type: Mocatta, July 3, 1915, Queensland, Ather- ton, Small to rather large trees to 33 m.; outer bark brownish, with shallow longitudinal fissures; inner bark brownish grading to yellowish brown or cream toward the cambium; sapwood whitish; heartwood yellowish; branchlets, leaves, and inflorescences glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves alternate, imparipin- nate or (occasional leaves) paripinnate, 12.5-35 cm. long; rachis gla- brous to finely pubescent; petiolules of lateral leaflets 4-10 mm. long; terminal leaflet on an extension of the rachis 1—3.1 cm. long; leaflets 3—5 pairs, chartaceous to subcoriaceous, sparsely to densely pellucid-dotted, glabrous to loosely pubescent below, glabrous or short-pubescent along the midrib above, elliptic to elliptic-lanceolate, usually unequal-sided and often subfalcate, 5-15 cm. long, 1.3—4.8 cm. wide, base acute to rounded, often oblique, main veins 10-17 on each side of the midrib, apex narrowly long-tapering to short and bluntly acuminate. Inflorescence terminal, 7— 14 cm. long, usually about as wide as long, axes and branches short-pubes- cent. Flowers bisexual, 3-4 mm. long; pedicels 0.7—1 mm. long; sepals sparsely appressed-pubescent, ciliolate, suborbicular, 1—1.2 mm. long; petals creamy yellow, sparsely appressed-pubescent abaxially, glabrous or with a few papillae adaxially, elliptic-oblong, 3-3.2 mm. long; stamens in- flexed apically, 1.5-2 mm. long, filaments glabrous, anthers basifixed, broadly rounded apically, about 0.5 mm. long; staminodes about 1 mm. long; disc about 1 mm. high; gynoecium about 1.5 mm. high and about 1 mm. wide, ovules 3 on each side of the placentae. Capsule sep- arating (or easily separable) into 5 distinct valves at maturity, elliptic- oblong, 9-12 cm. long; exocarp drying light to dark brown, densely and minutely pubescent to glabrate, muricate, the excrescences thick and conical, to 5 mm. long; endocarp very pale brown flecked with medium brown. Seeds 3 on each side of the dissepiments, winged at both ends, 5 cm. long; hypocotyl lateral, horizontal. ILLUSTRATIONS. WuitE, C. T., ibid. Francis, W. D., Australian Rain- forest Trees 427. 1951 DistrIBUTION. Cook District, Queensland; well-drained rain forests. See Map 3 Queensland. Cook Disrricr: Kuranda, Crothers, January 1926 (BRI); Forestry Reserve 607 ca. 10 miles W of Cairns, Smith 10121 (sri), Draper (BRI); Tinaroo Range ca. 15 miles NE of Atherton, Smith & Webb 3372 (BRI); 1969 | HARTLEY, THE GENUS FLINDERSIA 503 Atherton, Mocatta, July 3, 1915 (srt- holotype; MEL-isotype), Mocatta (srt, MEL), Jones 1284 (CANB); Gadgarra, Barnard 31 (CAN NB), Dreghorn 11 (prt), 20E (Bri), Smith 10144 (Bri), 10155 (prt), 10424 (pri), Volk 1411 (BRI), Webb 1661 (CANB), White 1567 (rt); Innisfail, Michael 403 (a, prt): head of Johnstone River, White, January 1918 (srr), eases definite locality: Wood Technology Dept. Queensland Forestry Service 56 (NY 8. Flindersia schottiana F. Muell. Frag. od el Austral. 3: 25. 1862. Lectotype: Bidwill 95, Queensland, Wide Bay Flindersia schottiana F. Muell. var. pubescens F. Muell. Frag. Phytogr. Austral. 5: 143. 1866. gt ie — Pua saris seca Bay. Flindersia pubescens F. M. Bailey, Queensl. Agr. Jou . 1898. Type: F. M. Bailey, ieee oe 1883, eer Sees ae. Large trees to 50 m.; outer bark pale brown or gray, generally quite smooth; inner bark white grading to light brown or yellow-brown toward the cambium: sapwood pale yellow; heartwood light brown; branchlets, leaves, and inflorescences glabrous to pubescent with predominantly stel- late, usually rust-colored trichomes. Leaves opposite, imparipinnate or (occasional leaves) paripinnate, 19-54 cm. long (the leaves of immature trees generally much larger); rachis glabrate to appressed- or soft- pubescent; petiolules of lateral leaflets usually obsolete, occasionally kag 2.5 mm. long, terminal leaflet on an extension of the rachis 1.1-2.7 c long; leaflets (4—)5-8 pairs, chartaceous to subcoriaceous, densely ade cid-dotted, glabrate to appressed- or soft-pubescent below, glabrous or short-pubescent along the midrib above, narrowly elliptic to oblong, usually unequal-sided and often falcate, 8-22 cm. long, 1.6~—6.3 cm . wide, bases of lateral leaflets narrowly to broadly rounded to cordate oblique, base of terminal leaflet acute to cuneate, main veins 15-22 on each side of the midrib, apex gradually tapering to subacuminate. Inflorescence terminal, to 40 cm. long, generally much wider than long, axes and branches appressed- to soft-pubescent. Flowers bisexual, 4-6 mm. long; pedicels obsolete to 1 mm. long; sepals minutely appressed-pubescent, ciliolate, broadly ovate to suborbicular, 1-1.5 mm. long; petals white, sparsely appressed-pubescent abaxially, as: in the basal one-third to one-half adaxially, elliptic-oblong, 4-6 mm. long; stamens declinate, about 3 mm. long, filaments pilose i fee or (rarely) glabrous, anthers basifixed, bluntly mucronulate, about 1 mm. long; staminodes 1-1.7 mm. long; disc about 1.5 mm. high; gynoecium 2-2.8 mm. high, about 1.5 mm. wide, ovules 3 on each side of the placentae. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic to elliptic-oblong, 8~13 cm. long; exocarp drying brown, glabrate to minutely and densely pubescent, muricate, the excrescences rather thick and conical, to 5 mm. long; endocarp reddish brown to light brown. Seeds 3 on each side of the dissepiments, winged at both ends, 5-6 cm. long; hypocotyl lateral, hori- zontal. ILLustratTions. Barey, F. M., Comprehensive Cat. Queens]. Pl. P/. 504 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 IV. 1913 (as Flindersia pubescens). Francis, W. D., Australian Rain- forest Trees 161, 162, 163. 1929; 179, 180, 181, 427 (as F. og 430 (as F. Sasbescons’. 1951. Mawen, J. H., Forest Fl. New S. Wales 2 . 69 & 70. 1905. ~ Map 4. Distribution of Flindersia schottiana F. Muell. DistRIBUTION. New Guinea and eastern Australia south to the Hastings i New South Wales; rain forests to 700 meters. See MAP 4 ew Guinea (West Irian). VocELKop Division: Sorong, Rmoe, Pleyte 703 beget gratin near Kp. Baroe, Pleyte 737 (Bo, L); Mlasoen Hill E of So- ‘€ 1969} HARTLEY, THE GENUS FLINDERSIA 505 rong, van Royen 3406 (A, L, LAE); Warsamson Valley E of Sorong, Schram BW 12354 (L, Lae); Kebar Valley, Koster BW 7121 (L), Moll BW 9531 (1, LAE), Schram BW 7921 (1); Sidai, Schram BW 1752 (CANB, L), BW 1756 (tL), BW 7607 (1). Papua. WESTERN District: Lower Fly River, E bank opposite Sturt Island, Brass 7991 (A, BRI, L, LAE); Oriomo Creek, mouth of Yakup Creek, 40 miles from sea, Womersley NGF 17729 (4, mO, K, 1, S 99681). NorTHERN DISTRICT: Bariji-Managalase Area, N side of Sibium Range S of Toma Village, Pullen 6383 (CANB E); 2 miles from Mafo L along Ibinamo River toward Mt. Suckling, Darbyshire 1164 (A, L, LAE, NSW 99603). Queensland. Coox Duisrrict: Cairns, Bailey (Nsw 99606), Betche, August 1901 (Nsw 99609, MEL); Trinity Bay, Hill, 1876 (BRI, MEL); Rocky Creek, Atherton District, Bailey, June 29, 1899 (BRI); Martintown, Bailey, June-July 1899 (Bri, MEL); Forest Reserve 185, Juara Creek, near Danbulla, Fraser 19 (prt), Smith 3781 (srt); Danbulla, Jones & Pedley 669 (CANB); Kairi, White, January 24, 1918 (prt); Atherton District, Mocatta (BRI, MEL); Forestry Reserve 191, Barron, Wongabel, Forestry Department 1 (srt), 2 (BRI), 3 (BRI, K). NortH KENNEDY District: Rockingham Bay, Anonymous MEL); Mount Dryander, Kilner & Fitzalan (BM, MEL); Dalrymple Heights and vicinity, Clemens, June-November 1947 (A, BRI), September-November 1947 (GH, MEL, MIcH), November 1947 (GH, MICH); Eungella Range, via Mackay, Francis, October 3-12, 1922 (prt); Cape Hillsborough, ca. 15 miles NW o Mackay, Bardsley, November 8, 1967 (BRI). Port Curtis DIsTRICT: Kal- power, Floyd, September 5, 1949 (LAE); Baffle Creek District, White, April 1920 (srt). BurRNeTtr District: Mt. Perry, Forestry Department, October 1921 (BRI). Wipe Bay District: near the Hummock, a few miles E of Bunda- berg, Smith 4100 (prt), 4101 (Bri); Bingera, ca. 10 miles WSW of Bundaberg, GH, MICH); Wide Bay, Bidwill 95 (x-lectotype of Flindersia schottiana F. Muell.); Fraser Island, Petrie 31 (srt); Bauple, Clemens, June 13, 1945 (cH), June 10-20, 1945 (micn); Amamoor, Moore in Swain 337 (prt); Kin Kin, Francis, May 1919 (prt); Imbil, Epps, August 1914 (srr), Webb 5019 (CANB); Brooloo, Webb SN 5339 (cANB), 1635 (caNB). Moreton District: Parish of Monsidale, Webb SN 5422d (cANB); Eumundi, 5 miles S of Cooroy, Bailey & Simmonds, November 1894 (rt); Blackall Range, Twine, February 1918 (BRI), White, April 1918 (srt); Sawpid Range, White, April 1918 (srt); Nor- ueensian n tt Gan 99600): 5 miles S of Brunswick Heads, Gray & Gray - (CANB); Potts Point, Robbins 2612 (cANB); Byron Bay Lighthouse Road, Con- stable 6505 (Nsw 83976); Richmond River, Henderson, November 1886 ae Nsw 99597); Clarence and Richmond (rivers), Northern Woods aa ou Wales 24 (x, MEL); Whian Whian, Jones 469 (cANB), 939 (cANB); Casino, 506 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 District Forester, December 1, 1905 (BRI, Nsw 99594), McAuliffe 8338/12 (nsw 99593, us); Tintenbar, Baeuerlen 633 (BM, MICH, NSW 99602); Clarence River, Anonymous (MEL), Moore 137 (MEL), Northern Woods New South Wales 61 (K, MEL), Selwyn (MEL); Clarence River, Grafton, Northern Woods New South Wales 63 (Nsw 99604), Squire, November 1937 (Nsw 99595); Mt. Yarrahappini, Briggs 70.05F (Nsw 99601), Smith 2 (Nsw 99592); lower slopes of Dorrigo Range, Wheen, August 29, 1945 (Nsw 99596); Taylor’s Arm, Wilshire, June 2, 1905 (Nsw 99605); Bellinger and McLeay Rivers, Anony- mous (K); McLeay River, MacDonald 183 (MEL), Woolls (MEL); Kempsey, MacDonald 212 (met), Rudder, May 17, 1891 (Nsw 99591); Hastings River, Tozer (Thozet?) (Nsw 99599). Cultivated. Java. Bogor Botanic Gardens, Anonymous, 1903 (us). Queensland. Cook District: Forestry Reserve 310, Gadgarra, Smith 10824 (prt). Burke District: Mt. Isa, Pedley 1060 (Rt). Port Curtis District: Rockhampton Botanic Gardens, Simmons 4 (BRI). Moreton District: Brisbane Botanic Gardens, Bailey (Brn), Blake 2688 (BRI), White (srt); Kangaroo Point, Brisbane, Francis (Bo), White, November 1912 (BRI); Wickham Reserve, ae os October 23, 1883 (BRI-holotype of Flindersia pubescens F. M. iley). w South Wales. Sydney Botanic Gar- dens, Boorman, December one (P). Flindersia pubescens is apparently recognized as a distinct species by many Australian botanists. It is considered to have a rather limited distribution centering on the Cairns area in northeast Queensland whereas F. schottiana, sensu stricto, is considered to be wider ranging, extending from New Guinea and north Queensland south to the Hastings River in New South Wales. The morphologic differences between the two are given in the following key from White (1921): Leaflets on flowering shoots subcoriaceous, somewhat falcate, 6.5-13 cm. long, 2-3.3 cm. broad, quite glabrous on the rachis and under surface clothed with very close and dense stellate, velvety tomentum, veins and Neos F. schottiana Leaflets on the flowering branches chartaceous, 12.5-23 cm. long, 4.5-6.5 cm broad, rachis densely clothed with comparatively long golden-brown stel- late hairs, under surface clothed with numerous but more or less scat- tered stellate hairs, the veins and veinlets prominent F. pubescens After studying a large number of collections from both Australia and New Guinea I am convinced that the differences between these two taxa all break down to a greater or lesser degree and that they represent environmental adaptations of sub-taxonomic significance. Their geographic distributions tend to substantiate this, I think, since both forms occur at widely disjunct localities in New Guinea. For example, Brass 7991, from the Fly River and Pleyte 737, from the Vogelkop Peninsula, are good matches for typical F. schottiana, whereas Darbyshire 1164, from the Northern District of Papua, and Schram BW 12354, from the Vogelkop Peninsula, are typical of F. pubescens. As indicated above, in the outline to species relationships, Flindersia schottiana, F. bourjotiana, and F. xanthoxyla appear to be more closely related to one another than to any of the other species of the genus. Beyond this, however, the three are mutually quite distinct. 1969 | HARTLEY, THE GENUS FLINDERSIA 507 9. Flindersia bourjotiana F. Muell. Frag. Phytogr. Austral. 9: 133. 1875. Type: Dallachy, Queensland, Rockingham Bay. Flindersia tysoni C. DC. Bull. Herb. Boiss. ser. 2. 6: 986. 1906. Type: Tryon, August 1901, Queensland, Mossman River. Large trees to 35 m.; outer bark gray or brown, rather smooth; inner bark pale yellow-brown; sapwood whitish; heartwood pale yellow-brown; branchlets, leaves, and inflorescences glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves opposite, imparipinnate or (occasional leaves) paripinnate, 9-33 cm. long; rachis glabrate to appressed- or velvety-pubescent; petiolules of lateral leaflets 1.5-4 mm. long, terminal leaflet on an extension of the rachis 1—2 cm. long; leaflets (1-)2-4 pairs, chartaceous to subcoriaceous, densely pellucid-dotted, glabrous to rather densely short-pubescent below, glabrous to sparsely appressed-pubescent above, elliptic to lanceolate, equal- or only slightly unequal-sided, 5.5-17 cm. long, 1.5—4.8 cm. wide, base acute to obtuse, main veins 12-17 on each side of the midrib, apex obtuse to acute or occasionally bluntly acuminate. Inflorescence terminal, to 20 cm. long, generally much wider than long, axes and branches sparsely to rather densely short-pubescent. Flowers bisexual or (a few to many flowers in an inflorescence) functionally staminate, 6-10 mm. long; pedicels obsolete to 1 mm. long; sepals glabrate to sparsely appressed-pubescent, ciliolate, broadly ovate to suborbicular, 1-2 mm. long; petals white or greenish white, sparsely appressed-pubescent abaxially, glabrous adaxially, elliptic, 5—9.5 mm. long; stamens declinate, 4—5.7 mm. long, filaments sparsely to rather densely pilose subapically, anthers dorsifixed, mucronate, 1—1.2 mm. long; staminodes 1-2 mm. long; disc thin in bisexual flowers and com- paratively thick in functionally staminate flowers, 1-1.5 mm. high; Synoecium in bisexual flowers 2-3 mm. high, about 1.5 mm. wide, ovules 3 on each side of the placentae; gynoecium in functionally staminate flowers poorly differentiated, narrowly conical, about 1 mm. high, without ovules. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic, 7-15 cm. long; exocarp drying blackish-reddish brown, glabrous, muricate, excrescences slender, often recurved, to 4 mm. long; endocarp light brown. Seeds 3 on each side of the dissepiments, winged at both ends, about 5.5 cm. long; hypocotyl lateral, ascending. ILLUSTRATION. Francis, W. D., Australian Rain-forest Trees 426. 1951. Distripution. Northeast Queensland, Cook and North Kennedy Dis- tricts; rain forests to 900 meters. See Map 5. Queensland. Cook District: Mossman River, Tryon, August 1901 (srI, NSw-isotypes of Flindersia tysoni C. DC.); near Ayton, Gittons 575 (BRI 35658); Bailey’s Creek Area, Smith 11556 (prt); Mt. Lewis, ca. 10 miles N of Mt. Malloy, Schodde 3326 (cANB); Kuranda, DuRietz, August 1927 (BRI); Atherton, Curry 5 (ny), Jones 1289A (CANB), 1289B (cANB), Mitchell, August 1911 (Nsw 99683), Tardent X230 (srt), Webb 2502 (cans); Atherton Table- 508 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 P 5. Distributions of Flindersia bourjotiana F. Muell. (dots) and F. xanthoxyla (A. Cunn. ex Hook.) Domin (half-filled circles). land, Tardent (a, BRI); ercnmne! Fraser 22 (BRI), Webb 5131 (cANB); Gad- garra, Dreghorn 12 (BRI), Kajewski 1140 (a, BM, BRI, NY, US), Trist 33 (NY); Lake Barrine, Barnard, June 15, 1941 (CANB) ; ’ Glenallan, Malanda, Tardent 188 (BRI); Ravenshoe, Manuell 31 (kK); Russel River, Anonymous (MEL); In- nisfail, Michael 132 (GH), Petrie (A); Johnstone River, Bancroft, 1885-1886 (BRI, MEL), Michael, May 1916 (prt); Bingil Bay Road, Gittons 575 (BRI 35542); Etty Bay, Webb 905 (cans). NortH KeNNepy District: Herberton, Mocatta, February 1917 (srr); Kirrama Range W of Kennedy, between So- ciety Flat and Yuccabine Creek, Smith 3205 (srt); Murray River, Anonymous, December 12, 1861 (meL); Rockingham Bay, Dallachy (met-holotype of Flindersia hausjoviana F. Muell.; Bo, MEL-isotypes) Flindersia tysoni was seneaiiialy placed in the synonymy of F. bour- jotiana by White (1921). 10. Flindersia nooner ae (A. Cunn. ex Hook.) Domin, Bibliot. Bot. 22 (89): 298. 1927. sree xanthoxyla A. Cunn. ex Hook. in Hooker’s Bot. Misc. 1: 246. t. 54. 830. Type: Cunningham 117, Queensland, Brisbane River 1969 | HARTLEY, THE GENUS FLINDERSIA 509 Flindersia oxleyana F. Muell. Frag. Phytogr. Austral. 1: 65. 1859 (nomen illegit., based on Oxleya xanthoxyla A. Cunn. ex Hook.). Medium to large trees to 40 m.; outer bark gray or gray-brown, fairly smooth; wood yellow; branchlets, leaves, and inflorescences glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves opposite, imparipinnate, 15-32 cm. long; rachis glabrous to appressed- or short-pubescent; petiolules of lateral leaflets obsolete to 6 mm. long, terminal leaflet on an extension of the rachis 0.8—2.8 cm. long; leaflets (2—)3-5 pairs, membranaceous to chartaceous, very brittle when dry, with or without scattered pellucid dots, glabrous to appressed- or soft-pubescent below, glabrous to sparsely appressed-pubescent above, elliptic to lanceo- late, usually unequal-sided and often falcate, (2.2-)4-13 cm. long, (0.6-) 1.3-3.2 cm. wide, base obtuse to cuneate, often oblique in lateral leaflets, main veins 11-15 on each side of the midrib, apex long-tapering, acute. Inflorescence terminal, to 25 cm. long, usually much wider than long, axes and branches appressed- to short-pubescent. Flowers bisexual, 4-5 mm. long; pedicels obsolete to 2 mm. long; sepals glabrous to sparsely ap- pressed-pubescent basally, ciliate, broadly ovate, about 1 mm. long; petals pale yellow, glabrous, elliptic-oblong, about 4.3 mm. long; stamens decli- nate (becoming straight after anthesis), about 3 mm. long, filaments sparsely pilose subapically, anthers dorsifixed, obtuse apically, about 1 mm. long; staminodes about 1.3 mm. long; disc 0.7-1.2 mm. high; gynoe- cium 1.5 mm. high, about 1 mm. wide, ovules 3 on each side of the placentae. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic-oblong, 6.5~11 cm. long; exocarp drying dark brown to pale gray-brown, densely and minutely pubescent, muricate, excrescences rather narrowly conical, unequal in length, to 4 mm. long; endocarp yellow-brown. Seeds (2—)3 on each side of the dissepiments, winged at both ends, 3.3-5 cm. long; hypocotyl lateral, horizontal. ILLUSTRATIONS. CUNNINGHAM, A., ibid. Francis, W. D., Australian Rain-forest Trees 164, 165. 1929 (as Flindersia oxleyana) ; 184, 185. 1951. Maren, J. H., Forest Fl. New S. Wales 2: t. 73 & 74. 1906 (as F. oxleyana). DistRIBUTION. Southeast Queensland and adjacent New South Wales; rain forests to 500 meters. See Map 5. Queensland. Burnetr District: Edenvale Hill, near Kingaroy, Michael 3106 (BRI), Wipe Bay District: Mary River Scrub, Gympie, Kenny (BRI); Imbil, McAdam 83 (prt, NY), 85 (A, BRI), 87 (BRI, NY), Weatherhead, July 1917 (srt). Moreton District: Parish of Monsidale, Webb SN 5422c (CANB); Yarraman, Clemens, August 4-15, 1944 (A, Ny), Floyd, August 29, 1949 (LAE), Webb 5143 (cans), SN 5337 (cans); South Pine River near Samford, Hub- bard 5941 (a, Bri, K); South Pine River near Bald Hills, White 7155 (a, BO, NY); Samford Range, Shirley & White, April 1918 (BRI); Samford, Tracey in Webb & Tracey 3392 (cans); Petrie, Blake 3079 (srt); Enoggera, Bailey (Nsw 99643): Three-mile Scrub, Enoggera Creek, Bancroft (Bri); Brisbane River below Breakfast Creek Bridge, Bailey (BRI, NSw 96644); Brisbane River, 510 JOURNAL OF THE ARNOLD ARBORETUM [ VoL. 50 Cunningham 109 (GH), 117 (BM-holotype of Oxleya xanthoxyla A. Cunn. ex Hook.; k-isotype), Hill (MEL); Sherwood, Brisbane River, Hubbard 5942 (x), Anonymous, December 18, 1930 (BRI); Tamborine Mountain, Longman & White, February 1917 (pr1). Without definite locality: Wood Technology Dept. Queensland Forest Service 80 (NY). New South Wales. Acacia Creek via Killarney, Queensland, Boorman, February 1905 (Nsw 99639), Dunn, No- vember 1905 (NSW 99642), Dunn 252 (NSW 99640); Tweed River, Anony- mous 59 (MEL), Moore 14125 (pm); Murwillumbah, Charles, January 9, 1905 (Nsw 96649); Whian Whian, Jones 943 (CANB), Webb 2457 (CANB), 5241 (CANB), White 12769 (srt); Wollongbar Experimental Farm, Lismore, Johnson & Constable, June 11, 1957 (Nsw 96650); Tintenbar, Baeuerlen, March 1892 (a); Richmond River, Fawcett (MEL), Henderson (MEL), Watts, 1902 (NSW 99641); Richmond River to the Tweed River, Moore (pm, GH, K); Hastings River, Boorman, August 1907 (Pp); Sandiland Ranges, Boorman, November 1904 (Nsw 96646). Without definite locality. Leichhardt (K, Nsw 99638). Cultivated. Queensland. Wipe Bay District: Gympie, Wickham and Channon Streets, Kenny, January 17, 1907 (BRI). New South Wales. Sydney ot Gardens, Boorman, February 1907 (NSW 96647), Camfield, January 1894 (N 96648), December 1896 (Ny), January 1898 (MEL, Us). 11. Flindersia bennettiana F. Muell. ex Benth. Fl. Austral. 1: 389. 1863. Lectotype: Bidwill, Queensland, Wide Bay. Flindersia leichhardtii C. DC. Monogr. Phanerog. 1: 731. 1878. Type: Leichhardt, 1845, Queensland, Moreton Bay. Small to large trees to 43 m.; outer bark pale gray, to reddish brown, quite smooth; inner bark whitish; branchlets, leaves, and inflorescences glabrous to minutely pubescent with stellate trichomes. Leaves opposite, imparipinnate, 8.5-36(—45) cm. long; rachis glabrous to appressed- or rarely short-pubescent; petiolules of lateral leaflets 1-6 mm. long, terminal leaflet on an extension of the rachis 0.9-3.5 cm. long; leaflets 1-3 (—4) pairs, subcoriaceous to coriaceous, densely pellucid-dotted, glabrous or sparsely appressed-pubescent below, glabrous above, ovate to elliptic to elliptic-oblong, equal- or occasionally unequal-sided and subfalcate, 6—18.5 cm. long, 1.7—6.7 cm. wide, base obtuse to subacute, often slightly oblique, veins prominent above, main veins 10-30 on each side of the midrib, apex obtuse or occasionally acute. Inflorescence terminal or occasionally ter- minal and upper-axillary, to 25 cm. long, as wide or nearly as wide as long, axes and branches appressed- to short-pubescent. Flowers bisexual or (a few to many flowers in an inflorescence) functionally staminate, 3-6 mm. long; pedicels 0.3-3.5 mm. long; sepals sparsely to densely short- pubescent, ciliolate, ovate, 1-1.5 mm. long; petals white, sparsely ap- pressed-pubescent abaxially, glabrous stile, elliptic-oblong, 2.5—5 mm. long; stamens declinate, 2.5-4 mm. long, filaments glabrous, ae dor- sifixed, bluntly mucronulate, about 0.9 mm. long; staminodes 1-1.7 mm. long; disc about 1 mm. high; gynoecium in bisexual flowers 2—2.8 mm. high, about 2 mm. wide, ovules 2 on each side of the placentae; gynoecium in functionally staminate flowers poorly differentiated, conical, about 1 mm. 1969 | HARTLEY, THE GENUS FLINDERSIA 511 high, without ovules. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic to elliptic-oblong, 4-7 cm. long; exocarp drying medium to very dark reddish brown, glabrous, muricate, excres- cences narrowly conical, unequal in length to 4 mm. long; endocarp reddish brown. Seeds 2 on each side of the dissepiments, winged at both ends, 3—4.3 cm. long; hypocoty] lateral, horizontal. ILLUSTRATIONS. Francis, W. D., Australian Rain-forest Trees 168, 169. 1929; 186, 187. 1951. Maren, J. H., Forest Fl. New S. Wales 3: t. 77 & 78. 1906 DIstRIBUTION. Southeast Queensland and northeast New South Wales; rain forests to 300 meters. Queensland. Wipe Bay District: Fraser Island, Epps 30 (ny), Forestry Dept. 131 (prt), 132 (BRI), 133 (BRI), Hubbard 4403 (x), Petrie 30 (srt), Trist 6 (NY); Dundowran via Gympie, Tryon, July 1928 (srt); Maryborough District, Young, September 1916 (prt), Mt. Bauple, Kajewski, September 1922 (a, BRI); Wide Bay, Bidwill (x-lectotype of Flindersia bennettiana F. Muell. ex Benth.; cH, ny); Gympie, Hamilton-Kenny 145.05 (NSw 99582); Kin Kin, Francis (prt, Nsw 99586); Forest Survey Camp, Amamoor, Anony- mous, April 22, 1919 (prt); Imbil, Floyd, September 18, 1949 (LaE), McAdam 90 (A, BRI, NY), Weatherhead in Swain 346 (sri), Webb 5013 (CANB). More- TON District: Eumundi, Bailey, June 1893 (BRI, Nsw 99588), Simmonds, June 1895 (a); Candle Mountain, White, May 5, 1918 (prt); Moreton Bay, Cunningham (pm), Hill (xk, MEL), Leichhardt, 1845 (p-holotype of Flindersia leichhardtii C. DC.); Brisbane, Boorman, April 1899 (MEL, Nsw 99585), White, June 3, 1926 (a, BO, BRI); Mount Cotton, Scortechini (BRI, MEL); Tamborine Mountain, Clemens, March 1947 (prt); Beech Mountain, White 1907 (A, BRI, NSw 99584); Southport, Simmonds, June 1913 (a, Ny); Meyer’s Ferry, Nerang River, White, October 20, 1917 (BRI, NSW 99583). Without definite locality: Bennett (nsw 99587). New South Wales. Tweed River, Anonymous, November 1897 (MEL); Tweed River District, Grime, July 1900 (Nsw 99576); Cudgen, Murwillumbah District, McKee 9547 (BRI, CANB, NSW 99589); Hastings Point, Trapnell, June 7, 1960 (BRI); east foothills of Nightcap Range ca. 2-3 miles Scrub, near Byron Bay, Forsyth, November 1898 (NSW 99575); Whian Whian State Forest, Jones 937 (cans), Webb 5242 (cans); Richmond River, Baeuer- len 244 (MeL), Boorman, February 1899 (p), Fawcett (MEL), Henderson 22 (MEL), (BM, BO, MEL, Nsw 99573), Mrs. Hodgk (MEL); Dalwood, Richmond River, Graham in Watts 37 (MEL, NSW 99579); Richmond and Clarence Dis- tricts, Bennett (MEL); Minyon Falls, N of Lismore, de Beuzeville, July 16, 1936 (Nsw 99566); Dorroughby, 14 miles NE of Lismore, Byrnes, September 1953 (Nsw 99580); Wollongbar Experimental Farm, Anonymous, November 1897 (Nsw 99570), Manager Experimental Farm, Wollongbar 6 (Nsw 99577); Lismore, Forest Guard, March 1909 (Nsw 99568), Baeuerlen 350 (A, NSW 99574), 698 (Nsw 99581); Ballina, Baeuerlen 836 (MEL); 8 miles N of W ood- burn, Williams J43 (Nsw 99571); Clarence River, Beckler (MEL), Wilcox, 1875 (MEL); Iluka, Cameron 93 (Nsw 99567). Cultivated. New South Wales. Syd- ney Botanic Gardens, Boorman, September 1919 (BRI, NSw 99578, us), Cam- field, September 1896 (ny), February 1899 (Nsw 99569, us), Mueller (Nsw 512 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 99590); Sydney, Paddington, Vernon, March 11, 1879 (MEL). Victoria. Mel- bourne Botanic Gardens, Anonymous, October 23, 1934 (MEL) There is apparently no authentic material of Flindersia leichhardtii in the Australian herbaria and it has not previously been placed in the synonymy of F. bennettiana. As is indicated above, Flindersia bennettiana, F. collina, F. dissosperma, and F. maculosa comprise a group of related species. They appear to be related to one another in a linear sequence, beginning with a rain-forest species, F. bennettiana, and ending with a xerophyte, F. maculosa. The following outline shows the apparent relationships of these species as indicated by various morphologic features. pacer stellate; leaves imparipinnate, 8.5-36(45) a hes the rachis III sb. des. 2 10 Wd atin anraceto is else re a ae aed . F. benne ae Trichomes stellate and a leaves imparipinnate or age 5-14 crs TOG, AT YARIS WIE. cod os pow a es Keke de vee dnkas . F. collina. Trichomes predominantly lepidote; leaves “seeing trifoliolate, 1.5— G5 crm. Ini, the vachin Wine... 02. kek cen ecu . F. dissosperma. Trichomes predominantly lepidote; leaves 1-9 cm. long, simple. .........- Se iene ea te te ee Parsee or aired ore ead Gaia: 14. F. maculosa. 12. Flindersia collina F. M. Bailey, Queensl. Agr. Jour. 3: 354. 1898 (based on Flindersia strzeleckiana F. Muell. var. latifolia F. M. Bailey). Lectotype: Bailey, Queensland, Moreton District, Main Range. Flindersia see F. Muell. var. latifolia F. M. Bailey, Synop. Queensl. Fl. 1st suppl. 12. 1886. Medium to large trees to 40 m.; outer bark mottled, green and brown, exfoliating in thin roundish flakes leaving shallow depressions; inner bark reddish grading to cream toward the cambium; sapwood yellow to yellow- brown; heartwood yellow to pale brown; branchlets, leaves, and inflor- escences glabrous to lepidote or minutely pubescent with predominantly scale-like and stellate trichomes. Leaves opposite or subopposite, impari- pinnate or trifoliolate, 5-14 cm. long; rachis glabrous to lepidote or appressed-pubescent below, glabrous or short-pubescent above, narrowly to broadly winged laterally, the wings extending 0.3—2.4(—4) mm. on each side; leaflets sessile, 1-2(—3) pairs, chartaceous to coriaceous, with scat- tered pellucid dots, glabrous or sparsely lepidote or appressed-pubescent below, glabrous above, elliptic to obovate to broadly spatulate, 2-9 cm. long, 1-4.7 cm. wide, base obtuse or (in some terminal leaflets) attenuate, veins prominent above, main veins (10—)12—16 on each side of the midrib, apex obtuse to rounded, usually retuse or emarginate. Inflorescence ter- minal or terminal and axillary, to 17 cm. long, generally about as wide as long, axes and branches glabrate to short-pubescent. Flowers bisexual or (a few to many flowers in an inflorescence) functionally staminate, 4.7-5-3 nd 1969 | HARTLEY, THE GENUS FLINDERSIA 513 mm. long; pedicels 0.54.5 mm. long; sepals appressed-pubescent, ciliate, broadly ovate, about 1 mm. long; petals white, glabrate to appressed- pubescent abaxially, short-pubescent in the basal half adaxially, elliptic, 4-5 mm. long; stamens declinate, 3.5-4 mm. long, filaments glabrous, anthers subdorsifixed, subacute apically, about 1 mm. long; staminodes about 2 mm. long; disc about 0.75 mm. high; gynoecium in bisexual flowers about 1.75 mm. high, 1.25 mm. wide, ovules 2 on each side of the placentae; gynoecium in functionally staminate flowers poorly differen- tiated, conical, about 1 mm. high, without ovules. Capsule separating (or easily separable) at maturity into 5 distinct valves, elliptic to elliptic- oblong, 2.8-5 cm. long; exocarp drying dark brown, glabrous or glabrate, muricate, excrescences 1-2 mm. long; endocarp rather dark brown. Seeds 2 on each side of the dissepiments, winged at both ends, 1.5—2.5 cm. long; hypocotyl lateral, horizontal. ILLustrations. Francis, W. D., Australian Rain-forest Trees 170, 171. 1929; 188, 189. 1951. Maren, J. H., Forest Fl. New S. Wales 3: ¢. 81 & 82. 1907 Map 6. oe ae “ oo collina F. M. Bailey (half-filled genie F. dissosperma (F. Muell.) Domin (dots), and F. maculosa (Lindl.) B (open circles with eee line 514 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 DistRIBUTION. Southeast Queensland and adjacent New South Wales; rain forests and rather dry scrubs to 700 meters. See Map 6. Queensland. LEICHHARDT District: Taroom, Mobsby, October 1912 (BRI). BurNEtt District: Eidsvold, Bancroft, April 1911 (pri), April 1923 (a), April 1928 (Nsw 99671); Goodnight Scrub, ca. 40 miles SW of Bundaberg, Smith 9882 (BRI); West Wooroolin, NW of Kingaroy, Everist, March 23, 1961 (prt); Kingaroy, Smith 3115 (pri, Ny); Nanango, Grove 131 (BRI); Cooyar and Charlestown, Gympie Forest District, Swain, March 1917 (BRI); without definite locality (probably Mt. Perry fide L. S. Smith), Keys(?) 800 (BRI). Wipe Bay District: Bundaberg, Boorman, July 1912 (Nsw 96676); Kepnock, about 2 miles E of Bundaberg, Smith 4163 (prt); Childers, Helms, January 3, 1899 (BRI), Anonymous (NSW 99670); Bauple, Clemens, June 20, 1945 (GH, MICH); near Imbil, Smith & Webb 3129 (BRI, CANB). MARANOA District: about 20 miles W of Mitchell, Everist, March 7, 1950 (prt). Dar- LING Downs District: Chinchilla, Beasley 52 (prt); Bunga Mts., White, Octo- ber 1919 (srt); Rangemore School Area, Cooyar-Bunya Mountains Road, Smith 10260 (A, BRI, K, NSW 96677); Toowoomba, Longman, October 1910 (kK, Nsw 99675). Moreton District: Parish of Monsidale, Webb SN 54226 (CANB); Yarraman, Clemens, August 1944 (A, BRI, NY, US), Floyd, August 28, 1949 (LAE), Webb SN 5338 (cans); Kilcoy, English, October 1919 (A, BRI); Jimna, near Kilcoy, Webb 5248 (cans); Crow’s Nest, Clemens 43641 (A); Main Range, Bailey (srt-lectotype of Flindersia collina F. M. Bailey), Pente- cost 29 (BRI, NSW 99669); Flagstone Creek, ca. 8 miles SE of Toowoomba, 7 miles SW of Helidon, Smith, October 13, 1965 (BRI); Helidon, White 8765 (A); Fernvale, ca. 13 miles NW of Ipswich, Bevington, 1909 (srt); Rosewood, Anonymous, October 1908 (BRI); Brisbane | i abhi 164 (pm), Octo- ber 1828 (kK, oe Moreton Bay, pa 8 (BM, GH, NY), pee (K); bar Road, Everist & Webb 1413 (srt). Without definite locality: Bowman (MEL), Trist 19 (Ny). New South Wales. Acacia Creek, near Killarney, Queens- land, Boorman 15 (Nsw 99673), February 1905 (xk), Dunn, September 1905 (MEL, NSW 99672), October 1905 (Kk), February 1906 (Nsw 99674); Unumgar, Jones 2371 (CANB). Without definite locality. Hill 53 (MEL). 13. Flindersia dissosperma (F. Muell.) Domin, Bibliot. Bot. 22 (89): 298. 1927. Strzeleckya dissosperma F. Muell. in Hooker’s Jour. Bot. aig Gard. Misc. 9: 308. 1857. Type. Mueller, Queensland, Burdiken Riv Flindersia strzeleckiana F. Muell. Frag. Phytogr. Austral. 1: “6S. 1859 (nomen illegit., based on Strzeleckya dissosperma F. Muell.). Small trees to 10 m., developing from a divaricately branched shrub stage; outer bark mottled, dark gray and white, cream or salmon, rough- scaly on the trunk, smooth above; inner bark reddish grading to cream toward the cambium; branchlets, leaves, and inflorescences glabrous to lepidote or minutely pubescent with scale-like and stellate trichomes. Leaves opposite, imparipinnate or trifoliolate, or (rare occasional leaves) simple, (0.8—)1.5-6.3 cm. long; rachis sparsely lepidote below, winged 1969 | HARTLEY, THE GENUS FLINDERSIA 515 laterally, the wings extending 0.5—1.5 mm. on each side; leaflets sessile, 1—2 pairs, chartaceous to subcoriaceous, with scattered pellucid dots, sparse- ly lepidote below, glabrous above, elliptic, spatulate, oblong or sublinear, 0.6-3.7 cm. long, 0.2-0.7 cm. wide, base obtuse, main veins usually indis- cernible, 8-10 on each side of the midrib, apex rounded to acute, occa- sionally retuse. Inflorescence terminal or rarely terminal and upper- axillary, to 8 cm. long, usually nearly as wide as long, axes and branches sparsely to rather densely lepidote to appressed- or short-pubescent. Flowers bisexual or (a few to many or all of the flowers in an inflorescence) functionally staminate, 3-4 mm. long; pedicels 0.7-2 mm. long; sepals glabrous or glabrate, ciliate, suborbicular, 1-1.3 mm. long; petals white to cream, glabrous, broadly elliptic, 3-3.5 mm. long; stamens inflexed apically, about 2.5 mm. long, filaments glabrous, anthers subdorsifixed, bluntly mucronulate, about 1 mm. long; staminodes 0.5—1 mm. long; disc about 0.5 mm. high; gynoecium in bisexual flowers about 1.5 mm. high and 1 mm. wide, ovules 2 on each side of the placentae; gynoecium in functionally staminate flowers poorly differentiated, pulvinate, about 0.4 mm. high, without ovules. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic, 2-3 cm. long; exocarp drying dark reddish brown, glabrous, muricate, the excrescences 1-2 mm. long; endo- carp brown. Seeds 2 on each side of the dissepiments, winged at both ends, 1.5—1.8 cm. long; hypocotyl lateral, horizontal. ILLUSTRATION. Battery, F. M., Comprehensive Cat. Queensl. Pl. t. 73. 1913 (as Flindersia euieciionas. DistripuTion. East central Queensland; dry scrubs to 300 meters. See AP 6. Queensland. NortH KENNEDY District: Maryvale Station, Daintree (MEL); W of Charters Towers, Blake 14906 (prt); Charters Towers, Michael 1275 (BRI), Stephens North Queensl. Nat. Club 10468 (Nsw 99689); Millchester Hill, Stephens North Queensl. Nat. Club 9056 (srt); Burdiken, Kennelly 208 oe Mueller (met-holotype of Strzeleckya dissosperma F. Muell.; K-isotype); H bert’s Creek, Bowman 70 (MeL); near Bogie Range, lower Burdekin River, ca. 42 miles S of Ayr, Smith 4532 (prt); Cape River, Daintree (met), Fitzalan (MEL). SourH KENNEDY District: Head of Suttor River, Sutherland (MEL); Laglan, about 80 miles W of Clermont, Everist, October 12, 1960 (BRI, K); 18 March 1920 (srr). LercHHarpT District: Peak Downs, Mueller (MEL); 9 miles SW of Anakie, Adams 1281 (CANB, MEL, NS 99685): E of Emerald, Webb 2251 (cans); Emerald, Webb SN 5296 mene Blair Athol, Massey 30 (BRI); ca. 3 miles N of Clermont on road to Charters Towers, Smith 3160 (BRI); Clermont, Small, September 1912 (BRI, K, NSW 99621); Chirnside, ca. 6 miles S of Capella, Smith & Webb 3428 (BRI); 2 miles N of Emerald, Bisset E198 (BRI); 4 miles E of Girrah Homestead, 36 miles N of Blackwater Township, 1967 (prt); 7 miles N of Goowarra, Anderson, October 15, 1965 (BRI); about 3 miles E of Goowarra, Johnson 943 (A, BRI); 3 miles E of Parkes Homestead, 516 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Speck 1672 (CANB, NSW 99686); Dingo, O’Shanesy 2013 (MEL); 8.5 miles SW of Duaringa, Speck 1820 (BRI, CANB, NSW 99687); near Duaringa, Simmons, May 1938 (RI); between the Barcoo and Springsure, Bailey (MEL). Without definite locality: Bailey (BRI, NSw 99688), Fitzalan (MEL). 14. Flindersia maculosa (Lindl.) Benth. Fl. Austral. 1: 389. 1863. Elaeodendron maculosum Lindl. in Mitchell Trop. Austral. 384. 1848. Type: Mitchell, November 1846, New South Wales, Balonne River. Flindersia maculata F. Muell. Quart. Jour. Trans. Pharm. Soc. Victoria 2: 44. 1859 (nomen illegit., based on Elaeodendron maculosum Lindl.). Small to medium trees to 15 m., developing from a divaricately branched shrub stage; outer bark mottled, dark gray or brown and cream or white, smooth-scaly on the trunk, smooth above; wood yellow; branchlets, leaves and inflorescences glabrous to lepidote or minutely pubescent with scale- like and stellate trichomes. Leaves opposite, simple; petioles lepidote below, short-pubescent above, 1-12 mm. long; leaf blades chartaceous to subcoriaceous, with scattered pellucid dots, minutely and sparsely lepidote and appressed-pubescent below, glabrous or short-pubescent on the midrib above, narrowly elliptic-oblong, oblanceolate or sublinear, 1-8 cm. long, 0.25-1 cm. wide, base narrowly obtuse to attenuate, main veins usually indiscernible, 12-17 on each side of the midrib, apex rounded to obtuse, occasionally retuse. Inflorescence terminal or rarely terminal and upper- axillary, to 7.5 cm. long, usually about as wide as long, axes and branches sparsely to densely lepidote to appressed- or short-pubescent. Flowers bisexual, 4-4.5 mm. long; pedicels obsolete to 2.5 mm. long; sepals glabrous, ciliate, broadly ovate to orbicular, 1-1.5 mm. long; petals white to cream, glabrous, obovate, about 4 mm. long; stamens inflexed apically, about 3 mm. long, filaments glabrous, anthers subdorsifixed, obtuse to bluntly mucronulate, 1-1.2 mm. long; staminodes about 0.5 mm. long; disc about 0.75 mm. high; gynoecium about 1.5 mm. high and 1 mm. wide, ovules 2 on each side of the placentae. Capsule separating (or easily separable) into 5 distinct valves at maturity, elliptic, 2.3-2.7 cm. long; exocarp drying dark reddish brown, glabrous, muricate, the excrescences 1—1.5 mm. long; endocarp brown. Seeds 2 on each side of the dissepiments, winged at both ends, about 1.8 cm. long; hypocotyl lateral, horizontal. ILLUSTRATIONS. BAILey, F. M., Comprehensive Cat. Queensl. Pl. ¢. 73 bis. 1913. Maten, J. H., Forest Fl. New S. Wales 1: 213. t. 39. 1904. DIsTRIBUTION. Central Queensland south to southcentral New South Wales; dry, rather open places. See Map 6 Queensland. BurKE District: 57 miles W of Hughenden, McCray, Septem- ber 21, 1967 (BRI); Hughenden, Brass & White 63 (A, BO, BRI, K). MITCHELL Everist & White 134 (srt); Aramac, Paulton (mEL); Barcaldine, Francis, 1969 | HARTLEY, THE GENUS FLINDERSIA Si? March 1920 (srr); between Emerald and Longreach, Jarvis, October 1913 (srr); Blackall, Bailey (Nsw 99620), Everist 1562 (a, BRI), White 12387 (A, K, US). Grecory SoutH District: Thylungra, about 75 miles NW of Quilpie, Bverist 3787 (BRI). WarRrEGO District: about 19 miles N of Thargomindah, Sm 6346 (BRI); about 34 miles N of Charleville on Ward River Road, Smith 341 (A, BRI, K, MEL, NY); between Cunnamulla and Wyandra, Key, October 1940 (CANB); near Cunnamulla, White 11782 (Br1). MARANOA District: Maranoa, Anonymous (MEL); Bollon Area, Epps, June 26, 1953 (BRI); 10 miles N of Dir- ranbandi, Key, October 17, 1937 (cANB); Buckinbah, near St. George, Jones 215 (BRI): St. George, W edd, December 1893 (BRI); Nindigully District, Anon- ymous, November 1938 (cans); Warrie, Nindigully, Allen A547 (CANB), Roe R10 (cANB), October 1937 (cANB). DARLING Downs District: 36 miles W of Goondiwindi on Talwood Road, Webb SN 5319a (cans), SN 5319b (CANB), SN 5319c (cANB), SN 5319d (cans); Goondiwindi District, Jones C50 (CANB), Webb 1611 (CANB), 2498 (CANB). New South Wales. Yantara Lake, Anonymous 385 (MEL); E of Broken Hill, Pidgeon & Vickery, August 1939 (Nsw 99617); Koonenberry Mountains, ca. 62 miles SSE of Milparinka, Constable 4609 (Nsw 64979), V. E. Expedition (MEL); Mount Hope Station, 3 miles N of White Cliffs, Constable 4595 (Nsw 67368); Wonominta River, ae Jan- uary 1887 (MEL): Wilcannia, Johnson 547/90 (Nsw 5181), Kennedy, 1886 (MEL); about 12 miles S of Wilcannia, Hogan, October 1955 (MEL); between Wilcannia and Cobar, Campbell 109 (cans); 30 miles E of Wilcannia, Riek & Common 322 (cANB); above Morinda (Menindee) and Mt. Murchinson (Murchiston), Anonymous (MEL); Menindee District, Neila-garri Station, Constable, November 20, 1947 (Nsw 4879); Mt. Murchinson, Anonymous (MEL), Dallachy (Bm, GH, K, NY), Dallachy & Goodwin (MEL); Darling River, Anony- mous (MEL), Dallachy (pm), Kennedy, 1866 (MEL); about 1 mile E of Ivan- hoe, Johnson, May 6, 1955 (Nsw 99615); Mossgiel, Bruckner, October 1885 (mex), M ueller (GH); between the Lachlan River and Darling River, Bruckner, 1885 (MEL), Mueller (p); Lachlan River, Tucker (met); Upper Lachlan River, Curran 21 (MEL); 11 miles from Ivanhoe on Paddington Road, Whaite 1388 (Nsw 99616); between Kirriby and Lauradale, Warego River to Paroo River, Boorman, October 1912 (srt, NSW 99619, Us); near Morton Boolka, Morris 766 (Nsw 99618); Barringun, Foyster, 1884 (MEL); Bourke, Betche, Novem- ber 1887 (Nsw 99623), McDougall, 1901 (Nsw 99613), Wuerfel, 1884 (MEL); Clover Creek, Bourke, Mackay 112 (MEL); 20 miles SE of Bourke, Riek & N 99632); Dunlop Station, ca. 8 miles S of Louth, Etheridge 28 (Nsw 99622); Cobar, 4 mile, Wilcannia Road, Forestry Officer 24 (Nsw 99629); between the Bogan and Darling Rivers, Morton 77 (met); Bogan River, Anonymous (MEL); Bogan, Morton, 1880 (MEL); W of Nevertire, Clarke, June 14, 1944 (CANB); Jackson, October 1911 (Nsw 99628); Balonne River, St. George’s Bridge (ca. 25 miles W of Mungindi), Mitchell, November 1846 (k-holotype of Elaeoden- dron maculosum Lindl.: Nsw 99612-isotype); Brewarrina, Boorman, Novem- ber 1903 (GH, NSW 99630, ny, us); Yarrawin Station, Barwon River, 30 miles SE of Brewarrina, Froggatt 16 (nsw 99627); Carinda, ca. 40 miles SW of Walgett, Bucknell & Lowe, August 1965 (Nsw 99631); Pilliga, Rupp 13 — 99625); Pilliga Scrub, Anonymous, November 1932 (MEL); between Pilliga an Wee Waa, Bassett, November 14, 1947 (Nsw 99626); plains near Baradine Forsyth, October 1899 (NSW 99624); Ellenborough Falls(?), Boorman, 1904 (A, P, us); Murrumbidgee River, Bennett 3 (mEL); Cobar to Riverina Dis- 518 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 tricts, Anonymous (MEL); Riverina, Morton, 1880 (MEL). Without definite locality. Bidwill 74 (k), Anonymous (NSW 99614). 15. Flindersia ifflaiana F. Muell. Frag. Phytogr. Austral. 10: 94. 1877. Type: Hill, Queensland, Trinity Bay. Flindersia brachycarpa Merr. & Perry, Jour. Arnold Arb, 20: 332. 1939. Type: Brass 8389, Papua, Wassi Kussa River. Medium trees to 35 m.; outer bark gray-brown, thick, deeply longitudi- nally fissured; inner bark red, grading to cream toward the cambium; sapwood yellow or yellow-brown; heartwood yellow-brown or brown; branchlets, inflorescences, and capsules glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves opposite, paripinnate, 15-34 cm. long; rachis glabrous; petiolules 4-9 mm. long; leaflets 3-6 pairs, chartaceous to subcoriaceous, densely pellucid-dotted, glabrous, ovate-elliptic, elliptic, elliptic-lanceolate, or occasionally lanceolate, usually unequal-sided and often subfalcate, 6-13.5 cm. long, 2.8-5.2 cm. wide, base rounded to subcuneate, often slightly oblique, main veins 14-22 on each side of the midrib, apex rounded to narrowly obtuse, rarely subacu- minate. Inflorescence terminal or terminal and upper-axillary, 14-25 cm. long, usually as wide as long or somewhat wider, axes and branches glabrate to appressed- or short-pubescent. Flowers bisexual or (a few to many flowers in an inflorescence) functionally staminate, 2.5-3 mm. long; pedicels 0.5-2.5 mm. long; sepals sparsely appressed-pubescent, ciliolate, ovate-triangular, 1—1.2 mm. long; petals white, sparsely appressed-pubes- cent abaxially, glabrous or with a few papillae adaxially, elliptic to elliptic-oblong, 2—2.5 mm. long; stamens inflexed apically, about 1 mm. long, filaments glabrous, anthers subdorsifixed, bluntly mucronulate, about 0.5 mm. long; staminodes 0.5 mm. long; disc about 0.7 mm. high; gynoe- cium in bisexual flowers about 1.5 mm. high, about 1 mm. wide, ovules 2 on each side of the placentae; gynoecium in functionally staminate flowers poorly differentiated, turbinate, about 0.5 mm. high, without ovules. Capsule comparatively woody and heavy, separating to one-half or more of the length but not completely, rounded short-cylindric, 3.2—-5.5 cm. long; exocarp drying blackish brown to reddish brown, densely and minutely pubescent, often cracked with age, muricate, the excrescences rather widely spaced, to 2 mm. long; endocarp reddish brown. Seeds 2 on each side of the dissepiments, winged at the apical end only, 2.7-3.3 cm. long; hypocotyl lateral, ascending. ILLUSTRATION. FRANcis, W. D., Australian Rain-forest Trees 426. 1951. DistRIBUTION. Southern Papua south to the Atherton Tableland, Queensland; rain forests to 400 meters. See M . Papua. WESTERN District: Tarara, Wassi rege oe 8389 (a-holo- type of Flindersia brachycarpa Merr. & Perry; BRI, L, LAE-is s); Oriomo Creek, mouth of Yakup Creek, 40 miles from sea, yy ie ON GF 17728 (A, Map 7. Distributions of Flindersia ifflaiana F. Muell. (dots) and F. australis R. Br. (half-filled circles). BO, K, L, LAE, NSW 99667). Queensland. Cook District: Cape York Peninsula, Cape Grenville, Y oung 64 (prt); Endeavour River, Parsieh(?) (MEL); moun- tains near Mossman, Rosenstrom (prt); Great Dividing Range ca. 6 miles S Of Mossman, Smith 3964 (prt); Mt. Molloy, Crothers, 1934 (A, BO, BRI, NY); Barron River, Anonymous 46 (MEL); Forestry Reserve 1073, N of Kuranda, Dansie 1995 (srt, K), 1996 (BRI), 2008 (srt); Timber Reserve 315, Smith- field, Kuranda, Doggrell A36 (sri); Black Mountain near Kuranda, Jones 800 (CANB), 1511 (cANB), Webb 5189 (cANB); a few miles N of Kuranda, Smith 520 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 5309 (BRI); Trinity Bay, Hill (met-holotype of Flindersia ifflaiana F. Muell.; K-isotype); Cairns, Bailey (Nsw 99684); Freshwater Creek, Bailey (BRI); Kamerunga, Cowley 43 (prt); Edge Hill, near Cairns, Morris North Queensl. Nat. Club 4084 (srt); Redlynch, near Cairns, White 12813 (A, Ny); Atherton Area, Webb 2547 (cANB), 5208 (CANB). Without definite locality: Anonymous 10 (rt), Wood Technology Dept. Queensland Forest Service 91 (A, NY). Cul- tivated. Queensland: Cook District, Cairns, on esplanade, White, February 1918 (BRI). In the original description of Flindersia brachycarpa, Merrill and Perry stated that it strongly resembled F. ifflaiana but differed “.... . in having larger leaves with the leaflets strongly oblique at the base and more acute at the apex, and a little larger fruits.” These have proven to be quite variable characters in many species of Flindersia and now, with additional collections available, it is evident that they do not serve to distinguish F. brachycarpa from F. ifflaiana. As is pointed out above, Flindersia ifflaiana is apparently more closely related to F. australis than to any of the other species of the genus. The two do not appear to be particularly closely related, however. The leaves of F. iffaiana are opposite and paripinnate whereas those of F. australis are usually alternate and basically imparipinnate. Additional differences are given in the key to species. 16. si Seg australis R. Br. in Flinders’ Voyage 2: 595. t. 1. 1814. Type: Brown, September, 1802, Queensland, Broad Sound. Medium trees to 25 m.; outer bark gray to brown, smooth or with shallow longitudinal fissures, often exfoliating in flakes; inner bark reddish; branchlets, leaves, inflorescences, and capsules glabrous to pubescent with mostly minute, predominantly stellate trichomes. Leaves alternate, sub- opposite or opposite, in mature growth closely crowded at the branchlet apices, imparipinnate or (occasional leaves) paripinnate, (5.5—)9-34 cm. long; rachis glabrous to densely appressed-pubescent, near the base often narrowly crisped-winged laterally; petiolules of lateral leaflets obsolete to 3(—5) mm. long, terminal leaflet sessile or on an extension of the rachis to 3.2 cm. long; leaflets 2-4(—6) pairs, chartaceous to subcoriaceous, with scattered pellucid dots, glabrous to densely appressed-pubescent below, glabrous or sparsely short-pubescent on the midrib above, broadly to narrowly elliptic to occasionally lanceolate, equal- or slightly unequal- sided, (2.4-)3-12 cm. long, (0.8—)1.5-4.3 cm. wide, bases of lateral leaflets obtuse to cuneate, base of terminal leaflet sbiuke to attenuate, main veins 8-18 on each side of the midrib, apex obtuse to acute or subacuminate, occasionally rounded. Inflorescence terminal or terminal and upper-axil- lary, often on leafless branchlets, to 15 cm. long, about as wide as long, axes and branches rather sparsely to densely pubescent. Flowers bisexual or (a few to many flowers in an inflorescence) functionally staminate, 6.5-7.5 mm. long; pedicels 0.2-1 mm. long; sepals densely pees pubescent, pee ciliate, ovate-triangular, 2.2-2.5 mm. long; petals 1969] HARTLEY, THE GENUS FLINDERSIA 521 white to cream, densely appressed-pubescent (except for the margins) abaxially, sparsely short-pubescent in the basal one-half to two-thirds adaxially, elliptic-oblong, 6-7 mm. long; stamens declinate, 3.5—4.5 mm. long, filaments glabrous or with a few papillae and/or crisped trichomes subapically, anthers subdorsifixed, mucronate, 1—1.3 mm. long; staminodes -5 mm. long; disc in bisexual flowers rather thin, about 1.5 mm. high; disc in functionally staminate flowers comparatively thick, about 1 mm. high; gynoecium in bisexual flowers about 3 mm. high, and 1.4 mm_ wide, ovules 2 on each side of the placentae; gynoecium in functionally stami- nate flowers poorly differentiated, conical, about 0.6 mm. high, without ovules. Capsule comparatively woody and heavy, separating to one-half or more of its length but not completely, rounded short-cylindric, 4.6—9 cm. long; surface of exocarp drying blackish, reddish brown, or pale brown, densely and minutely pubescent, muricate, the excrescences rather densely crowded, often recurved, to 10 mm. long; inner part of exocarp very rough where exposed by dehiscence; endocarp cream, yellowish, or pale brown. Seeds 2 on each side of the dissepiments, winged at the apical end only, 3.4~5 cm. long; hypocotyl lateral, horizontal. ILLUSTRATIONS. Brown, R., ibid. FRANcis, W. D., Australian Rain- forest Trees 158, 159. 1929; 176, 177. 1951. Mawen, J. H., Forest FI. New S. Wales 2: t. 67 & 68. 1905. DistripuTion. Eastcentral Queensland south to northeastern New South Wales; rain forests and rather dry thickets. See Map Queensland. SoutH KENNEDY District: Pinevale via Mackay, Webb & Tracey 3319 (prt), LetcHHARDT District: Lake Elphinstone, Dietrich (MEL); Morambah Homestead, 43 miles SW of Nebo Township, Story & Yapp 121 (BRI, K, MEL); 4.6 miles SW of Duaringa, Speck 1670 (cANB); Coomooboolaroo, Berney 1919 (prt); Expedition Range, Byerly (MEL); 36 miles WSW of Theo- dore Township, Sean 6928 (BRI, K); 18 miles N of Taroom, Speck 1861 (BRI, CANB, K). Port CurTIs weiagse Broad Sound, Brown, September 1802 (k-holotype: BM, MEL, P-isotypes); r Ogmore, ca. 75 miles NW of Rock- hampton, Smith, ‘October 18, 1951 as hea ae Thozet (MEL). BURNETT District: Eidsvold, Bancroft, April 1912 (prt). Wipe Bay District: Kolan River, Smith’s Crossing, ca. 14 miles WNW of Bundaberg, Smith 4159 (BRI); Maryborough, Clemens, October 27, 1948 (BRI, GH, MIcH); Mt. Bauple, Clem- ens, June 10-20, 1945 (micH); Imbil, McAdam 84 (A, BRI); Cooroy, Douglas, November 2, 1962 (BRI). DARLING sina District: Chinchilla, Beasley 51 — Moreton Disreict: Yarraman, nig August 29, 1949 ens Coal ford, Meebold 7999 (nv); Brisbane River, (cae ee 60 (K), Mueller, July, 1855 hoor dowd November 1917 4 NSW 99565); tema Creek, near ue 572 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 Law (MEL); Unumgar, Jones 2367 (cANB); Murwillumbah, Charles 458.04 (nsw 99562); Whian Whian State Forest, Webb & Tracey 378 (CANB), 1953- 1958 (BRI); Rous, near Lismore, Cheel, September 1926 (Nsw 99560); Rich- mond River, Cameron (MEL), Fawcett (MEL), Henderson (MEL), Hodgkington (mEL); Clarence River, Moore (met), Beckler (MEL); Kempsey, MacDonald, October 3, 1895 (MEL). Cultivated. Queensland: Brisbane Botanic Gardens, Bailey, November 11, 1884 (srt), Blake 2797 (prt), Hubbard 4745 (x), White 2442 (kK). New South Wales: Sydney Botanic Gardens, Boorman, October 1920 (Ny, us), November 1920 (Nsw 99563). EXCLUDED NAMES FLINDERSIA GREAVESII C. Moore, Cat. Nat. Industr. Prod. New S. Wales 53. 1861 This name appeared in the “indigenous woods” section of a catalogue that accompanied an exhibit. Several other timber species were listed, each represented, apparently, by a wood sample. The description follows: A magnificent tree, the monarch of the northern forests, attaining a height of 150 feet, 3 to 6 feet in diameter, distinguishable from every other species of the genus by its dark brown and rough scaly bark, as well as by other charac- ters; timber used for house building purposes. Mountain brushes on the Clar- ence [River, New South Wales]. The wood sample has apparently been lost, but an herbarium specimen at Sydney (Nsw 99604), labelled Flindersia greavesii by Moore and bearing the number 63, which corresponds to the number of that species in the catalogue, could be designated as the type collection. The only really diagnostic statement in Moore’s description is that per- taining to the bark, which would almost certainly apply to Flindersia australis. The specimen mentioned above, however, clearly matches F. schottiana (published a year later), a species with relatively smooth bark. Maiden (Forest Fl. New S. Wales 2: 152. 1905) listed Flindersia greavesit as a synonym of F. australis and also designated it as a nomen nudum. He concluded, as I have, that the description referred to FP. aus- tralis and that the type collection referred to F. schottiana. In addition, he pointed out that a tree in the Sydney Botanic Gardens, labelled by Moore as F. greavesii, is really F. australis and that Mr. W. A. B. Greaves, after whom the tree was named, gave him a fruit which he stated was that of F. greavesii and that it actually was F. australis. Moore must have realized his error, too, since he did not mention Flin- dersia greavesii in his Census of the plants of New South Wales (1884) or in his Handbook of the flora of New South Wales (1893). Flindersia australis and F. schottiana were both mentioned in these publications. Although I do not think Flindersia greavesii can be discounted as a nomen nudum, I think this name should be excluded since the description and the type collection clearly refer to two different species. 1969} HARTLEY, THE GENUS FLINDERSIA 523 FLINDERSIA PAPUANA F. Muell. Descript. Notes Papuan Pl. 1(5): 84. 1877. Type: D’Albertis, Fly River, Papua (not seen). Mueller named this species from a single immature fruit and stated that it was “. . . just a temporary name for the Papuan plant to place it on the record until foliage and flowers can be obtained.” Thus it is a pro- visional name and is not validly published. LITERATURE CITED Atry-SHaw, H. K. Diagnoses of new families, new names, etc. for the seventh edition of Willis’ “Dictionary.” Kew Bull. 18: 249-273. 1965. BalLey, F. M. Meliaceae. Queensl. Fl. part 1. 225-243. 1899. BenTHAM, G., & F. Muetter. Meliaceae. Fl. Austral. 1: 378-390. 1863. CANDOLLE, C. DE. Meliaceae. Monogr. Phanerog. 1: 399-752. 1878. ENGLER, A. Rutaceae. Nat. Pflanzenfam. III. 4: 95-201. 1896. . Rutaceae. Nat. Pflanzenfam. ed. 2. 19a: 187~358. 1931. ErDTMAN, G. Pollen morphology and plant taxonomy. Angiosperms. xii + 539 pp. Chronica Botanica, Waltham, Massachusetts. 1952. Francis, W. D. Australian Rain-forest Trees. xi + 347 pp. Commonwealth of Australia. 1929 Australian os forest Trees. xvi + 469 pp. Commonwealth of Aus- tralia. 1951. Harrar, E. S. Notes on the genus Flindersia R. Br. and the systematic anat- om e important flindersian timbers indigenous to Queensland. Jour. Elisha Mitchell Soc. 53: 282-293. 1937 Hartiey, T. G. A revision of the Malesian species of Zanthoxylum (Ruta- ceae). Jour. Arnold Arb. 47: 171-221. 1966 . A revision of the genus Lunasia (Rutaceae). Jour. Arnold Arb. 48: 460-475. 1967 Mercatre, C. R., & L. CHALK. Anatomy of the Dicotyledons 1: Ixiv + 724 pp. Clarendon Press, Oxford. Price, J. R. The distribution of alkaloids in the Rutaceae. In: Chemical Plant Taxonomy, T. Swatn, Ed. 429-452. Academic Press, London and New York. , Ritcuir, E. oe of Flindersia species. Rev. Pure and Appl. Chem. 14: 47-56. SMITH- a : Chromosome numbers in the Boronieae (Rutaceae) and their bearing on the evolutionary development the tribe in the Aus- tralian flora. Austral. Jour. Bot. 2: 287-303. WELsH, M. a A ctsinsintae “maple” (Flindersia ah Sone, Woods No. 25: 18-23. 19 Waite, C. T. tes on the genus Flindersia (Rutaceae). Proc. Linn. Soc. New S. Wales 46: 324-329. 1921. Wittaman, J. J., & B. G. Scuusert. Alkaloid-bearing plants and their con- ta ined alkaloids. US. Dept. Agr. Tech. Bull. No. 1234. 247 pp. 1961. 524 JOURNAL OF THE ARNOLD ARBORETUM [voL. 50 INDEX TO EXSICCATAE text Adams, 1281, 1317 (13) Allan & Jones NGF 2751 (6) Allen A547 (14) Baeuerlen 244, 350 (11); 633 (8); 698, - (11) Bailey 10 (8) Balansa 163, ag (1) Barnard 31 Beasley 51 eae ge (it) Beckler 7620 (13 & 14) Bennett 3 Bidwill 74 oe 95 (8) Bisset E198 Blake 2688 (8) 2797 (16); 3079 (10); 14906 (13) Boorman 15 (12) Bowman 70 Brass 5339 (4); 5565, 7517 (6); 7991 (8); 8032 (6); 8389 (15); 8495, 8542 (2b): 8634 (6); 20322 (4); 28881, 28906 (2b); 29153 (4) Brass & — i 12535 (4) Brass & White 6. Briggs 75.05F @) Cameron 93 (11) Campbell 0109 (14) Carr 13152 (4); 13913 (6); 14408, 14808, 14910, 15475, 15569, 15805, 15969, 15989 (4) Charles 458.04 (16) Clemens 4739, 5086, 6825 (4): 43213, 43373 (3); 43641 (12) Collins W966 (4) Constable 4595, 4609 (14); 6505 (8) Cowley 43 (15 unningham 18 (12); 60 (16); 109, 117 (10); 164 (12) Dansie 1995, 1996, 2008 (15) Darbyshire 1164 (8 obson & Havel NGF 9116 (6) Docters van Leeuwen 10478, 10618, 10682 (4) The numbers in parentheses refer to the corresponding species and varieties in the Doggrell A35 (3); A36 (15) Dreghorn 11 Ai i2 (9); 13, 14 G@): 20E (7); ) Dunn 252 a Eddowes NGF 13086 (2b) Epps 30 (11); 31 (8) Etheridge 28 (14) Everist 1562, 5787 (14) Everist & Webb 1413 (12) Everist & White 134 (14) Floyd NGF 7472, NGF 7527 (4) Floyd & Morwood NGF 6204 (4) Forbes 421 (6) Forestry Department 1, 2, 3 (8); 237, 132, 433 (it) Forestry Officer 24 (14) Fournier & Sebert 22 (1) Franc 1738,