mt. J. Plant Sci. 157(6 Suppl.):S58 S WOOD, BARK, AND STEM ANATOMY OF GNETALES: A SUMMARY Introduction

Transcription

1 mt. J. Plant Sci. 157(6 Suppl.):S58 S by The University of Chicago. All rights reserved /96/5706S-0004$02.OO WOOD, BARK, AND STEM ANATOMY OF GNETALES: A SUMMARY SHERWIN CARLQUIST Santa Barbara Botanic Garden, 1212 Mission Canyon Road, Santa Barbara, California Data from a series of eight papers, representg a survey of Gnetales at species level, are summarized order to develop concepts on relationship between anatomical data ecology, phylogeny. systematics. Most of characters character states newly reported have phylogenetic ecological correlations for Gnetales as a whole rar than systematic significance with genera. Wood features are sensitively related to habit, although strategies lianoid species of Ephedra differ from those lianoid Gnetu species. Data also bear close relationships to organography. Vessel details are given special attention because workers have questioned wher vessels origated Gnetales dependently of those angiosperms. thus wher or not vessel orig two groups is a synapomorphy. The evidence cited here favors orig of vessels Gnetales dependent of vessel origs angio sperms. Hirto unappreciated contrasts between vessels of Gnetales those of angiosperms are detailed: torus margo structure of pits Gnetales (both Ephedra Gnetu), aspiration ability of gnetalean pits represent modes of conduction promotion of conductive safety different from those angiosperms, shape difference (circular vs. scalariform), cited data matrices of cladograms. is secondary to different functional syndromes. The tercalation of circular bordered pits to helical secondary wall thickengs of primary xylem tracheary elements of Ephedra Gneturn (a feature found also conifers Gkgo but not angiosperms) is given new functional terpretations. Ephedra Gnetum conta both tracheids fiber-tracheids that co-occur a mode different from that angiosperms which Conta both cell types toger. Ray structure like that of Gnetales occurs widely with seed plants likely offers little conclusive formation about relationships. Wood data on Gnelalec are compatible with hyposis that a vesselless group of gymnosperms, such as Pentoxylales or Bennettitales, is closest sister group of angiosperms. Formation of successive cambia Gnetum differs from that Welwirschia: latter has phellem but no or bark. Because Welwiischia has only phellem, re are more numerous bark features common between Ephedra Gnetum than one might have expected. The three genera appear monophyletic on basis of wood bark. Infrageneric anatomical features of systematic importance are few, although arboreal Gnetum gneon differs from lianoid Gnetuni species significant ways. Introduction This article summarizes eight papers that attempt to survey wood, bark, stem anatomy of Gnetales at species level, reby concludes series. With respect to anatomy of wood bark, earlier erature (e.g., Martens 1971) is based on study of a very small number of species. Therefore, a new survey, resentg many more species than had been studied before, was itiated. This pot is made because ear lier conclusions about nature of wood bark of Gnetales are based on a small base of formation, many character states summarized here do not appear; refore, recent papers on relationships of Gnetales. The fact that much new formation has been acquired requires tegration of that formation various ways: (1) simple descriptions of features, (2) alterna lions of earlier descriptions, (3) tegration of new formation to patterns provided by earlier work ers, (4) analysis of phylogenetic significance of data now available, (5) consideration of eco logical implications of wood bark data. The task at h analyzg se data sets, refore, is much more than citation of how features of wood impge on hyposes of seed plant evolution. The organiza tion of this article cannot, refore, be a lear amation of past hyposes terms of current lit rep ex evi Address for correspondence reprts: 4539 Via Huerto, Santa Barbara, California Manuscript received October 1995; revised manuscript received February dence: new wood bark formation must be analyzed summarized diverse contexts. The studies on wood bark of Gnetales at species level was begun with a study of New World species of Ephedra (Carlquist 1989), Old World species of Ephedra (Cariquist 1992), phenomenon of near-vessellessness some high-al pe species of Ephedra (Carlquist 1988b). The iti ation of survey with study of New World spe cies of Ephedra was related to fact that material of those species was more readily available to me. Al though Welwitschia has been studied for many cades, important new structural data proved worthy of reportg (Carlquist Gowans 1995). One tree spe cies of Gnetum, Gnetum gnemon, is widespread (Markgraf 1930) was considered a separate per (Carlquist 1994) because this species differed so much from lianoid species of genus. Both Af rican lianoid species of Gnetum were available (Carl quist Robson 1995); all New World species cept a recently named one were studied (Cariquist 1996b). The series concluded with a study of Indoma lesian Asiatic lianoid species (Carlquist I 996a). Representation at species level was 92% Ephed ra 66% Gnetum. Although less than optimal coverage of Indomalesian Asiatic species of de pa ex Gnetum, survey provides a basele to which formation on species not cluded can readily be added as material becomes available. Previous studies on wood, bark, axis anatomy of Ephedra, Gnetum, Welwitschia were summa rized by Pearson (1929) Martens (1971). The de S58

2 CARLQUIST ANATOMY OF GNETALES S59 scription of G. gnernon wood by Greguss (1955) has not been cited widely. Information on wood anatomy was assembled on a floristic basis for Ephedra by Fahn et al. (1985) for Gnetum by ter Welle Detienne (1991). Muhammad Sattler (1982) presented some observations on certa Gneturn vessels as a way of promotg particular phylogenetic concepts. By surveyg Gnetales at species level, much new formation was uncovered. The followg tures were newly reported for Gnetales as a whole: (1) helical thickengs secondary xylem vessels cheids of Ephedra (Carlquist 1989, 1992); (2) near vessellessness Ephedra (Carlquist 1988b); (3) pres ence nature of fiber-tracheids as well as tracheids wood, distribution of se cell types with lation to each or, Ephedra (Carlquist 1989, 1992) Gnetum (Carlquist 1994, 1996a, 1996b); (4) pres ence nature of axial parenchyma Ephedra (Carl quist 1989, 1992), Gneturn (Cariquist 1994, 1996a, 1996b; Cariquist Robson 1995), Welwitschia (Cariquist Gowans 1995); (5) diver sity axial parenchyma distribution types Ephedra (Carlquist 1989, 1992) Gnetum (Carlquist 1994, l996a, 1996b; Carlquist Robson 1995); (6) mi nute calcium oxalate crystals lg tercellular spaces Ephedra (Carlquist 1989, 1992) Welwitschia (Carlquist Gowans 1995); (7) tyloses vessels of Gnetum (Carlquist 1996b); (8) tyloses laticifers of Gnetum (Cariquist 1996b); (9) storied structure of wood Ephedra (Cariquist 1989, 1992); (10) phel loderm orig of lateral menstem activity Weiwit schia (Carlquist Gowans 1995); (11) presence of various types of sclerenchyma crystals bark of Ephedra (Carlquist 1989, 1992) Gneturn (Carl quist 1994, 1996a, 1996b; Carlquist Robson 1995). In addition, some features were newly reported for particular genera of Gnetales, although y were ready reported for one or both genera not listed: (1) tori pits of tracheary elements Gneturn (Carlquist 1996a, l996b; Carlquist Robson 1995); (2) wood plan as or plants with secondary growth, rar than vascular strs organized on some or basis, Welwitschia (Carlquist Gowans 1995); (3) growth rgs Gnetum (Carlquist 1994); (4) pres ence of uniseriate as well as multiseriate rays Ephedra (Carlquist 1989, 1992); (5) branchg ticifers Gneturn (Carlquist 1996b); (6) secretory nals, addition to laticifers, Gneturn (Cariquist 1 996a); (7) nature of phloem parenchyma, with respect to secondary xylem cells, Gneturn (Cariquist 1 996a). Some of se features are of major importance phy logenetically. For example, demonstration that eral meristem activity origates from cortex Gne turn (La Rivière 1916; Carlquist 1994, 1996a, 1996b; Carlquist Robson 1995) but from phelloderm Welwitschia (Carlquist Gowans 1995) means that lateral meristems have had different origs two genera that presence of successive cambia fea tra re al la ca lat ( anomalous secondary growth ) two genera cannot be regarded as a synapomorphy but rar as two autapomorphies. On contrary, presence of mute tercellular calcium oxalate crystals Ephed ra Welwitschia (a feature orwise not reported vascular plants, to my knowledge), seems best terpreted as a symplesiomorphy. Perhaps phyletically most important new report is that of tori pits of tracheary elements of Gneturn, because this feature represents a clearly gymnospermous feature rar than one transitional to angiosperms. These examples show that new reports of structural features Gne tales take on significance when tegrated with our previous knowledge of wood, bark, stems of Gne tales. Thus, before conclusions with regard to ecology phylogeny are attempted below, a summary at generic level of axial anatomy is presented. Infrage neric diversity is cited terms of exceptions to com mon conditions with genera. Although salient features of wood bark of Gnetales are summarized below, many details presented earlier papers of this series are not repeated here. Likewise, details on methods, techniques, material, termology are not repeated here. Wood anatomy EPHEDRA Stems roots of Ephedra feature a sgle cam bium with secondary phloem formed externally secondary xylem, ternally. Rays, chiefly multiseriate, separate broad fascicular areas. Particular fascicular ar eas may experience addition of secondary xylem more rapidly than ors; this lack of synchronization comes evident diagonal orientation of ray cells that terconnect equivalent portions of adjacent fascicular areas (Cariquist 1989, figs. 8, 12; Carlquist 1992, figs ; Lev-Yadun Aloni 1993, figs. 3, 4). In rath er old Ephedra stems (diameter depends on species) distortion orientation of fascicular ray portions can occur (Lev-Yadun Aloni 1993). This phenom be enon should not be equated with lateral meristem tivity leadg to production of successive cambia, features characteristic of Gnetum Welwitschia. Vessels of Ephedra are essentially solitary; if con tacts between vessels occur, y appear attributable to rom vessel placement wood which vessel den sity is great. There is no tendency toward vessel group g as re is dicotyledons that have fiber-tracheids or libriform fibers as imperforate tracheary element type (Carlquist 1 984a). Ephedra, like or Gnetales, exemplifies hyposis that when tracheids are present as imperforate tracheary element type, groupg of vessels does not occur to any appreciable extent (Cariquist 1984a). The shrubby species of Ephedra show strongly marked growth rgs. Not only are vessels larger earlywood than latewood but vessels may be absent latewood some species ac (Carlquist 1989, 1992). In a few species from high elevations, vessels are extremely few number, re

3 / SDN3IDS LNV Id 1O 1VNf1Of IVNOLLVNJ3LNI 09S

4 CARLQUIST ANATOMY OF GNETALES S6 I suitg a near-vesselless condition (Cariquist 1988b). The lack of vessels iatewood dicotyle dons that have tracheids as an imperforate tracheary element type has been terpreted as representg max imal abundance of tracheids, which are more resistant to formation of air embolisms than are vessel elements (Cariquist 1988a). Woods are mostly rg porous. The vessel elements of Ephedra have perforation plates composed of foramate (circular) perforations similar to outles of pit cavities of lateral wall pits on vessels on tracheid-to-vessel or tracheid-to-tracheid pits (figs. 1 4). There is little variation se fora mate perforations (sometimes termed ephedroid perforations ). The number of series of perforations ranges from one, most slender vessels, to two (fig. 1) or three (figs. 2 4), but rarely more than three, series. There are borders on perforations some spe cies (fig. 1), but most borders are mimal (figs. 2 4). Both promently bordered mimally bordered perforations can characterize a sgle species (figs. 1, 2), although perforations tend to be more promently bordered Ephedra americana Humb. & Bonpi. Ephedra a Poepp. ex Mey. (Carlquist 1989). No fusion of perforations is evident, although a few species, such as Ephedra kokanica Regel, relatively large perforations placed closely tend to be somewhat polygonal outle rar than circular (fig. 4). Pits on lateral walls of vessels, like tracheid-to-tracheid pits, are circular fully bordered, have promi nent tori connected to margo threads (fig. 9). Helical thickengs occur ner surfaces of secondary lem vessels of majority of New World species (Cariquist 1989) some Old World species (Carl quist 1992); species with helical sculpturg ves sels, such sculpturg also occurs tracheids. lrnper forate tracheary elements Ephedra are clearly differentiated to tracheids, with large bordered pits without cellular contents, fiber-tracheids, which have pits bordered on contacts with vessels or tracheids, but pits simple or quite vestigially bordered on contacts with or fiber-tracheids. Fiber-tracheids of Ephedra are nucleate but not septate. The reason why fiber-tracheids of Ephedra are not termed axial parenchyma is that y are not strs of cells sep arated by lignified walls; such strs do occur a few species of Ephedra (Ephedra foliata Boiss., E. kokanica Regel; Carlquist 1992, fig. 31). By contrast, Gnetum characteristically has both fiber-tracheids axial parenchyma. Or arguments for use of this mology, not used throughout Gnetales prior to current series of papers, have been presented (Carl quist 1989, p. 421). Fiber-tracheids Ephedra are ranged diffuse or diffuse--aggregates fashion ( xy ter ar latter illustrated Carlquist 1992, fig. 24); occasional large fiber-tracheid aggregations are present. Rays are mostly multiseriate, but uniseriates occur a few species, which y are much less common than multiseriates. Rays are composed of upright, square, procumbent cells; upright cells predomi nate rays of smaller-diameter samples, whereas cumbent cells crease abundance as a stem creas es diameter (Cariquist 1989). Starch gras are common ray cells, as y are also fiber-tracheids. Mute calcium oxalate crystals occur among cheary elements ray cells. Although uncommon a few species of Ephedra, se mute crystals are abundant vast majority of Ephedra woods. Storyg has been reported Ephedra coryi Reed var. viscida Cutler (Carlquist 1989), Ephedra equise ta Bunge, E. foliata (Carlquist 1992). Storyg is more common dicotyledons with short fusiform cambial itials, so occurrence of storyg Ephedra, which has shorter fusiform cambial itials than do or gymnosperms (with exception of Welwitschia), is not surprisg (Cariquist 1992). Wood of lianoid or scent species of Ephedra has some distctive features. Lianoid species clude Ephedra pedunculata Engelmann ex Watson New World, Old World, species of section Pseudobaccatae, tribe Scentes: Ephedra altissima Desf., Ephedra aphylla Forssk., Ephedra campylopoda C. A. Mey., E. foliata, Ephedra fragilis Desf., E. kokanica. Compared to nonlianoid species, se spe cies have a much greater density of vessels per mm of transection, longer vessel elements, greater per foration diameter (table 1), as well as less thick walls on imperforate tracheary elements. Differences tween lianoid nonlianoid Ephedra species with spect to vessel diameter wall thickness of ves sels are negligible. WELwITscHIA pro tra 2 be The axis of Welwitschia consists of a long taproot a relatively short, wide stem. From vestigations to date, re is every reason to believe that same histology (except for organization of primary xylem) is found both roots stems. The vascular strs of stems are so contorted that secondary growth histology m cannot be readily demonstrated. Therefore, descriptions here are based primarily upon root material. In both stems roots, re is a first cambium that produces a limited amount of secondary xylem secondary phloem, not unlike second ary growth seedlgs of various vascular plants with secondary growth. This stage has been illustrated by Bower (1881), Sykes (1910), Pearson (1929), But- re Figs. 1 4 SEM photographs of perforation plates of Ephedra, from radial sections. Figs. 1, 2, Ephedra major Host, Staton Fig. 1, Perforation plate with clearly bordered perforations. Fig. 2, Perforation plates with mimal borders. Fig. 3, Ephedra termedia Schrenk, Elias. July 21, 1983; perforations of nearly circular outle, mimally bordered. Fig. 4, E. kokanica, Elias 9742; relatively large perforations somewhat polygonal outle. Bars at upper left each photomicrograph = 10 m.

5 ities (Bierhorst 1960; Carlquist Gowans 1995) (1910) shows thickengs Welwitschia. Perforation represent lateral extensions of ner elliptical pit cav rar than thickengs, as Ephedra, although Sykes plates of vessel elements consist of one, rarely two, bordered circular perforations (Bierhorst 1960). Tori Martens (1971). Careful examation with both light microscopy SEM has thus far revealed only mem ures by Bierhorst (1960) suggest pit membranes ellip tical sectional view; se figures were copied by branes that are flat sectional view, with no evidence cheary elements present; no fiber-tracheids were ob served. woody cylder at least trees of moderate age; Gnetum gnemon has a sgle cambium a sgle bases of older trees, additional xylem segments may although some way new vascular tissue formed need of furr vestigation. be found (Rao Keng 1975). These segments orig ate from cambia formed bark (Carlquist 1994), from se cambia must be terconnected with xylem phloem of ma cylder. This is a topic vessels per group, 1.09 a number likely close to what rgs of an conspicuous sort have been observed narrow vessels near primary xylem, foramate per rom placement of vessels would dictate. Growth foration plates occur occasionally (figs. 5 7). The nar rowest of se vessels have perforation plates that consist of a row of well-bordered perforations (Carl quist 1994, fig. 15). Perforation plates somewhat wider vessels, such as those illustrated here (figs. 5 perforation shapes sizes suggest enlargement mergg of foramate perforations response to Se- 7), can be characterized as foramate-coalescent; are absent pits of tracheary elements, although fig are predomantly simple (fig. 8, portion shown). In a few collections (Carlquist 1994). Perforation plates Vessels are solitary or nearly so mean number of GNETuM GNEMON of tori. Tracheids are only type of imperforate tra 1995). All of parenchyma tissues produced by of lianoid species of Gnetum (Carlquist Gowans ler et al. (1973). Subsequent stages have been less fig outside of first cylder is formed from successive cambia (Martens 1971). Although first cambia out bia are exhausted before long. Subsequent cambia origate with radial files of parenchyma formed from a lateral meristem. These radial files are tercontuous with phelloderm, thus have a mode of orig different from that of radial files of paren chyma volved lateral meristems stems roots sue. Conjunctive tissues consist of starchy parenchy embedded relatively large calcium oxalate crystals. lateral meristem, or than parenchyma cells formed latous walls near outer surface of which are Many tercellular spaces conjunctive tissue are led with mute calcium oxalate crystals (Carlquist quist 1989). ured, although all authors agree that vascular tissue The successive cambia of Welwitschia produce rel atively large amounts of secondary phloem, as di cated by abundance of gelatous phloem fibers, com pared with secondary xylem cambia produce (Cariquist Gowans 1995). Neverless, sec ondary xylem, despite use of vague term vas Gowans 1995). Some rays beg as uniseriate but rap cular str above, has a secondary xylem plan much Mute calcium oxalate crystals le tercellular spac side of origal cylder arise cortical parenchy ma, cortical parenchyma sites for orig of se cam vascular strs, are termed conjunctive tis lem: fascicular ray areas are present (Carlquist es among some ray cells. Axial parenchyma terrupts are only 30% 35% as numerous as tracheids (origal ma, secretory cavities, large fibrosciereids with ge Gowans 1995) much like those of Ephedra (Carl like that of or vascular plants with secondary xy some of bs of fascicular xylem, which are one idly widen to multiseriate. All ray cells are upright. data). Vessel elements tracheids have grooves that to three tracheary elements wide. The vessel elements Asiatic Gnetum species clude Indomalesian species. 5, thickness, tm; CA = mean conductive area per mm2 transection, mm ; 6, TL = mean length of tracheids, tim; 7, TWT = mean wall thickness of tracheids. rim. Data on Welwitschia is origal. per mm2 transection; 3, VL = mean length of vessel elements, tm; 4, VW = mean vessel wall Key to columns: 1, VD = mean lumen diameter of vessels, tm; 2, VM mean number of vessels Sources. Data from Carlquist (1992, 1994, l996a, l996b); Carlquist Robson (1995). Gnetum, all lianas averaged Gnetum, Asiatic lianas Gnetum, African lianas Gnetum, New World lianas Gnetum gnemon (trees, shrubs) Welwjtschja Ephedra, Old World nonlianoid Ephedra, Old World lianoid Group (1) (2) (3) (4) (5) (6) (7) VD VM VL VW CA TL TW Table 1 WOOD CHARACTERISTICS OF GNETALE5 S62 INTERNATIONAL JOURNAL OF PLANT SCIENCES

6 CARLQUIST ANATOMY OF GNETALES S63 lective pressure for greater conductive area with perforation Lateral wall pits with plate. pits a of vessels have circular bordered cavity diameter much smaller than ameter of perforations. These pits bear vestures near 8) that occur vesturg at or is lackg with mouth of cavity. pit (fig. di pit aperture, Imper forate tracheary elements consist of both fiber-trache ids walled tracheids. The fiber-tracheids clearly bordered are relatively th usually septate; y have vestigially but that relatively small distributed romly with respect pits are to tracheids are (fig. 13). Tracheids have thicker walls than fiber-tracheids, lack have large circular bordered contents very close pits walls of vessels. on septa, No tori G. pit membranes of Paratracheal axial parenchyma cavity diameter those lateral have thus far been observed to on gnemon tracheary elements. is vasicentric rel atively abundant gnemon; apotracheal parenchy scarce diffuse. Strs consist of four have th but lignified walls, ma is 10 cells, G. is starch sometimes calcium oxalate crystals 13). are all walls. ray cells. Rays cells procumbent, except rays The most of both uniseriate cells; simple rays. are pits at ray on common multiseriate, to conta with (fig. al tips of multiseriate cells have lignified uniseriate Bordered pits predomate elsewhere Calcium oxalate crystals of various starch gras are common with ray cells. LIAN0ID SPECIES OF GNETuM The lianoid Gnetum species have that forms a stem, tangential walls a first cylder of secondary xylem with without a pith root. on ray sizes cambium pith Subsequent vascular cylders arise from successive cambia, parenchyma between se vascular cylders is termed conjunctive tissue composed of parenchyma that contas starch is here. a Conjunctive tissue, near rays, calcium oxalate crystals. The conjunctive tissue may conta laticifers Secretory cavities were observed Brongn. to The strs of phloem (fig. 17) some species. conjunctive tissue external second perhaps several successive cambia origate cortical parenchyma. Cortical parenchyma would termg be used tion of se cambia eventually, but what a cells (figs. 17, 18). site occur. of Gnetum gnemonoides more up I am of forma lateral meristem here forms additional parenchyma cortical region through production of radial Pearson (1929) hyposized that more than one files of orig for lateral meristem of orig activity might for orig of lateral meristem activity was demonstrat ed Gnetum by confirmed However, a cortical site La Rivière (1916), this has been each of three ma taxonomic geo graphical groupgs of lianoid Gnetum species (Carl quist Robson Cariquist 1995; 1996a, 1996b). files by I am first Cambia origate with radial termg lateral meristem. The orig of se cambia secondary phloem iclal divisions formed per ible Gnetum 1995). secondary xylem son Cambial action much more active y yield africanum Welw. is nate rays origate from sluggish fascicular fascicular areas, biseriate origate (fig. 16). se narrower rays, tracheids, are most clearly vis (Carlquist Rob future ray areas zones. Wide but multise successive cambia. With uniseriate rays or also The terrelationships between fiber-tracheids lianoid species have not been stated clearly lier literature. attempt be found G. An to clarify this situation ear can Carlquist (1996a). The vessel elements of lianoid species of Gnetum gnemon; that species, as are like only a few ma is pits pits. much vessels have foramate perforation plates, jority of perforation plates iously represented species no In addition is to on are simple. Vesturg lateral wall vesturg wholly absent vessel elements, those of narrower var of vessels, but se fascicular sec ondary xylem of lianoid Gnetum species consists of tracheids, fiber-tracheids, The features except that diffuse diffuse--aggregates axial parenchyma 14, 15). differ from each or more abundant than axial by are as G. gnemon, parenchyma which se three G. gnemon. cell (figs. types are The large circular bordered pits of vessel elements of tracheids have pit membranes that bear were observed hoizianum Engler (Carlquist Robson tori, at species. Tori G. Gnetum 1995), Tul. 1996b), Gnetum 1996a). all was tori a to Tori by SEM, some SEM is less or files (fig. 16) to Vasil 1961), y are not to sieve cells as cells by has is cell files files (fig. 16, with files are rays, as by as seen zones. However, study few such too few to files cells as seen (fig. 16). The cells lie files few to be or a ids Gnetum leyboldii pidatum Blume, (Carlquist relationship (Carlquist africanum, least some buch cus Gnetum microcarpum Blume Liquid-preserved material of se species mode of sample preservation. were demonstrated both light microscopy available. Lack of although ation between torus margo Ephedra conifers. Radially oriented might bear of photomicrographs, differenti marked than of parenchyma cells sec ondary phloem of lianoid species of Gnetum companion cells (Maheshwari have been likened but xylem sister cells were accurately designated Esau (1969). What not been clarified relationship between se parenchyma extensions of se to parenchyma secondary bottom). The radial orientation of xy lem cells tercontuous suggests that y might be xylem rays, confirmed fact some of m of uniseriate biseriate rays radial sections of se fascicular phloem parenchyma presence tangential of se tangential radial sections reveals relatively account for numerous radial transections radially ori ented rows of between radial possibly lem rays, of narrow that secondary xy of tra cheids prove mostly axial parenchyma strs. Although fiber-trache were not mentioned for African species of Gnetum fiber-tracheids

7 9 - I L / a I 4 I1 4r 1 I i d i I I IL S3DNI3S INV Id do 1VNIflOf 1VNOI1VNLLNI 179S

8 CARLQUIST ANATOMY OF GNETALES S65 (Carlquist Robson 1995), careful study of slides of those species revealed fiber-tracheids, albeit small numbers. Therefore, all of lianoid species of Gne turn, as well as G. gnernon, are now claimed to possess fiber-tracheids as defed for Gnetum by Cariquist (1994). Bark Bark of Ephedra (Cariquist 1989, 1992) has been shown to conta followg cell types: (1) sieve cells, phloem parenchyma, phloem ray parenchyma (obliterated to various extents dependg on age of bark); (2) sclereids derived from axial parenchyma cells; (3) suberized phellem; (4) phelloderm (th walled or sometimes scierenchymatous), often associated with mute tercellular calcium oxalate crystals; (5) sclereids derived from phloem ray cells; (6) gelatous phloem fibers; addition, younger stems reveal presence of (7) gelatous fibers dif fusely distributed cortex (fig. 20). The diffuse cortical fibers were not mentioned earlier accounts of bark because y were not evident once cortex was lost. The diffuse fibers are noteworthy because similar fibers occur cortex bark of lianoid Gne turn species. Not all of above seven features may be present all Ephedra species, accurate data on systematic distribution of features with genus are not available. In Gnetum gnernon bark, followg cell types occur (Carlquist 1994): (1) th-walled (nonscleren chymatous) phellem; (2) starch-rich parenchyma; (3) nests of brachysclereids, some of which conta large calcium oxalate crystals; (4) gelatous phloem fibers. In lianoid Gneturn species, all of se cell types occur, but addition, one may fd (5) occa sional sciereids phloem (Cariquist 1996b); (6) ge latous fibers scattered cortex (fig. 19); (7) a cylder of brachysclereids, outside of parenchy matous ner cortex but side of nests of bra chysciereids (fig. 17, top; fig. 18, top); (8) laticifers (easily visible some species, such as Gneturn schwackeanurn Taub. (fig. 17); (9) sclerenchyma tous phelloderm (Carlquist 1 996a). The sclerenchy matous cylder sclerenchymatous phelloderm are absent underground roots stems of lianoid spe cies (fig. 19; Cariquist 1996a). There is no bark ordary sense Weiwit schia, only phellem, which contas a few of giant gelatous fibrosciereids walls of which large crystals are embedded. Secondary phloem Welwit schia vascular strs does conta large numbers of gelatous phloem fibers of sort seen Ephed ra Gneturn bark, however. The roster of features bark of Gnetum es pecially diffuse gelatous fibers presence of sclerenchymatous phelloderm is remarkably sim ilar to that for Ephedra. At least some of se features seem likely to be symplesiomorphies. Rar than con cludg that similarities between two genera mark m as more closely related to each or than eir is to Welwitschia, one should consider pos sibility that its anatomical structures, as well as its bizarre gross morphology, Welwitschia has departed from a hypotical common ancestor reby contas a number of autapomorphies or two genera lack. Condusions on organography, habit, ecology ORGAN0GRAPHY In Ephedra, relationships of vessel features to or ganography were detailed (Carlquist 1989, p. 447) be cause collections of particular portions were made field. These data, sum, show that vessels are wider roots underground stems than upright aerial stems, reby agreeg with those dicotyle dons (Patel 1965). However, reverse trends occur Gneturn, admittedly on basis of a small samplg of lianoid species (Carlquist 1 996a). Vessel diameter roots is less than it is stems not only lianoid Gneturn species but Gneturn gnernon as well (Carl quist 1994), perhaps because roots studied are smaller than stems (although not dramatically so; see data Cariquist 1994). In dicotyledons, vessels ordi narily crease diameter with crease stem di ameter (Cariquist l984b). Increase vessel diameter with crease stem diameter was evident Ephedra (Ephedra boelckii Roig, Ephedra trifurca Torr., Ephedra viridis Coville). Relationships between quan titative vessel features organography deserve de tailed furr study Gnetum. HABIT In Ephedra, habit ranges from small shrubs to large shrubs to small trees to lianas or sprawlg shrubs. To be sure, lianas sprawlg species of Ephed ra are not a uniform group with respect to habit. The lianas Gneturn vary somewhat but seem amply dis tct habit from tree species Gnetum gnemon Gneturn costaturn K. Schum., judgg from descriptions of Markgraf (1930). However, Markgraf (1930, p. 436) claims that species one usually sees as a tree, G. gnernon, can be a shrub or a liana as well. The nature of lianoid dividuals of G. gnernon is need of vestigation, because such lianas have not Figs. 5 8 Perforation plates of Gnetum gnenson, Cariquist 8088; from radial sections (figs. 5, 6, 8) illustrated with SEM from a maceration (fig. 7) illustrated with light microscopy. Fig. 5, Perforation plate showg little alteration from a foramate condition. Fig. 6, Perforation plate with fewer, larger perforations. Fig. 7, Perforation plate with three perforations, one partially traversed by wall material. Fig. 8, Simple perforation plate (vesturg visible apertures of lateral wall pits of vessels). Fig. 7, Magnification scale above photograph (divisions = 10 p.m). Figs. 5, 6, 8, Bars, upper left, = 10 aim.

9 (Metcalfe Chalk 1950; Dickison et al. 1978). Wid foration plates dicotyledons (Carlquist 1975). That trend arboreal Gneturn collections as well as ference is similar to values shown for vessel diameter The lianoid Gneturn species have appreciably great gnemon) collections (82 jim; table 1, col. 1). This dif (table 1, col. 2) between lianoid nonlianoid Gne were excluded obtag figures for vessel density measurement: conjunctive tissue wide rays section of lianas compared with nonlianas Gneturn woody samples of dicotyledons (Carlquist 1975, p. groups of lianas probably because size of plants 206). The African species of Gnetum do not have for or lianas. One noteworthy feature that emerges from liana quantitative features comparable to those of or available for African species is much smaller than have thicker-walled vessels (Cariquist 1975, p. 181). nonliana comparisons Gneturn is that vessel walls lection for wall thickness to promote mechanical tendency for lianoid species of dicotyledons to Tracheids lianoid species of Gnetum have th ner walls (3.6 jim) than do G. gnernon collections strength a tree than a related liana. Data on this pot are frequently collected dicotyledons would be desirable. Tracheid walls lianoid Quantitative vessel tracheid data have not hith than those of shrubby species (3.1 jim), an ter estg confirmation of trend Gneturn. comparable to that of a desert shrub, a stem succulent, selected to develop data for Welwitschia table transectional areas of xylem are so small scattered comparable areas by rom scanng methods be accord with concept that re is greater se erto been presented for secondary xylem of Weiwit schia. Analyzg significance of se data is dif or any of various or xerophytes. Although collec tions undoubtedly differ, a typical root transection was usually used is applicable. Neverless, quanti 2) may a range of dicotyledonous growth forms (Carlquist 1975, p. 206). The apparent similarity vessel density lianas. Still, conductive area per mm such large areas of conjunctive tissue that fdg 2, col. 2) is essentially an estimate, because er vessels also have accelerated simplification of per lianoid ones is, refore, understable. er vessel diameter (124 jim) than do arboreal (G. be part an artifact of conventions adopted for are appreciably thicker Gnetum lianas than ficult, because wood of Welwitschia is not really 2 tran XER0M0RPHY (4.8 jim), as shown table 1, column 7. This would sels per mm 1 984b) by Hibbertia Tetracera Dilleniaceae 1. The figure for vessel density (mean number of ves shown by Piptocalyx Trimeniaceae (Cariquist Ephedra species (table 1, col. 7) are thner (2.2 jim) G. gnernon (table 1, col. 4). This, too, is accord with is similar to difference between lianas mesic turn collections (both close to 20 vessels per mm dicotyledonous families with scalariform perfora to wood of shrubby species (table 1), one fds that shrubs. Old World species were used exclusively table 1 for liana sample because re were 11 mention. received comment or study sce Markgraf s brief When one compares wood of lianas Ephedra re is little difference vessel diameter (col. 1) but much more difference vessel density (col. 2): lianas have a vessel density more than twice that of collections of seven species available tribe Scan nonlianoid Old World species (10.4 m). The conduc dentes (Cariquist 1992). Perforation diameter of se lianas (13.9 rim) is appreciably greater than that of tive area per mm times mean number of perforations times mean num ber of vessels per mm2; table 1, col. 5), of lianoid (for Ephedra) latewood vessels lianoid species cies. The perforation area, although a subsidiary factor stem surfaces of Ephedra are so limited, about nificant. The difference conductive area between li anas nonlianas Ephedra is terestg because same lianas nonlianas. Possible correlations species is more than three times as large as that of istics of lianas shrubs need to be vestigated. The nonlianoid species. The numerous relatively large account for most of this difference. Vessels are few to absent rar narrow latewood of shrubby spe differentiatg lianas nonlianas, is statistically sig between habit ecology vessel character greater length of vessel elements lianoid versus non tistically significant. This trend would be expected if lianoid Old World species (table 1, col. 3) is also sta shrubby species are more xeromorphic, sce shorter vessel elements characterize xeric species dicotyle When comparg lianoid to nonlianoid Gneturn spe studied present series of papers), differences quantitative vessel features are much more dramatic perforations per perforation plate is thus only a frac 2 transection (mean perforation area dons (Carlquist 1975). cies (table 1; figures based on all Gneturn collections by far majority are simple. The mean number of tion above 1.0 eir arboreal or lianoid Gneturn a broad-leaved liana. In Ephedra, it is very tion that lianoid habit relates to simplification of than Ephedra. Perforation plates of Ephedra are al ways foramate, whereas only a few perforation plates any stem or root of Gneturn are foramate; collections. Evidently, selective pressure for simple likely that microphylly reduces transpiration so con siderably that greater vessel density a slightly larg er mean perforation diameter suffice to accommodate perforation plate Gnetum is observation that amate perforation plate. Reforcg terpreta tion plates that have both shrubby/arboreal lianoid tion plate is much lower lianoid species. This is perforation plates is great both a broad-leaved tree creased flow rates lianas, re is no selective pressure for modification of morphology of for broad-leaved species, number of bars per perfora S66 INTERNATIONAL JOURNAL OF PLANT SCIENCES

10 CARLQUIST ANATOMY OF GNETALES S67 tative wood data for Weiwitschia show it to have a remarkably xeromorphic wood. The mean vessel di ameter for a samplg of dicotyledonous desert shrubs is 29 m (Cariquist 1975, p. 206), whereas figure for Welwitschia is 15 m. To be sure, former sampie was obtaed measurg outside diameter of ves sels, so that for that sample lumen diameter (which was what was measured for Gnetales table 1) would be Ca. 24 rim, but that is still significantly wider than figure for Welwitschia. The conductive area per mm transection (based on number of vessels per mm is low for Welwitschia (.04) but close to figures for dicotyledonous desert shrubs (0.18), stem succu lents (.09), woody species with successive cambia (.06) (Carlquist 1975, p. 206; because vessel wall thickness was cluded vessel diameter se dicotyledon figures, figures based on lumen area as Gnetales data would be smaller by perhaps 10%). The conductive area for liana stems is, by comparison, 0.36 mm per mm transection (Cariquist, 1975, p. 206). However, tracheids are likely quite effective conduction Welwitschia are Ca. three times as numerous as vessel elements Welwitschia secondary xylem (tracheids were not cluded estimate of conductive area per unit transection, only vessels). The vessel element length of Welwitschia (508 lim; table 1, col. 3) does not seem short by comparison to that of dicotyledonous desert shrubs (218 aim; Carl quist 1975, p. 206). That length is shorter than vessel element length for shrubby species of Ephedra (585 I.tm; table 1, col. 3) is less than half vessel element length of any Gnetum species. The mean ves sel lumen diameter of dicotyledonous desert shrubs is ca. 24 aim; vessel lumen diameter of Ephedra is slightly larger (nonlianoid species, 32 m; lianoid spe cies, 38 pm), but order of magnitude is compara ble: few mesic dicotyledonous trees have vessel di ameter less than 100 jim (Metcalfe Chalk 1950). As Welwitschia, we cannot estimate likely con ductive capabilities of tracheids as compared to those of vessels Ephedra or Gnetum. Comparisons with conductive areas per unit transection of dicotyledon woods with tracheids, dicotyledon woods with libri form fibers or fiber-tracheids, or with conifer woods where conductive areas are concerned are premature until we know role tracheids play relative to ves sels conduction Gnetales. Welwitschia many species of Ephedra possess a feature that is considered dicative of xeromorphy: helical sculpturg on ner surfaces of vessels tracheids. In Ephedra, this takes form of helical thickengs on ner surfaces of vessels of tracheids of most of New World species (Carlquist 1989) a scatterg of Old World species (Carl quist 1992). In Welwitschia, grooves that represent lat eral extensions of pit apertures around cell occur vessels tracheids, providg texturally a grooved wall surface similar topographically, even if different phylogenetic orig, to helical sculpturg pres ent vessels tracheids of Ephedra. In dicotyle 2) dons, sculpturg vessels tracheids characterizes species of areas drier colder than those occupied by congeners that lack helical sculpturg (Carl quist 1975). Although re is uncertaty about physiological significance of helical sculpturg sec ondary xylem vessels tracheids of dicotyledons, re is little doubt that it represents a xeromorphic adaptation because of ecological systematic distribution of this phenomenon. Helical sculpturg occurs tracheids of some conifers, notably Taxaceae some Cupressaceae (Greguss 1955). The mesic ecology of Gnetum is evident low ves sel density: vessel density lianoid Gnetum spe cies is almost same as for a samplg of dicoty ledonous lianas (Cariquist 1975, p. 206). The mean diameter of vessels G. gnemon (82 jim) does not fall much short of that for a samplg of mesic dicot yledon woody species (108 jim), discrepancy can be explaed by clusion of a young stem sample a shrub sample G. gnemon survey. The vessel diameter of Gnetum lianas (124 jim) is less than outside vessel diameter for dicotyledon lianas (157 jim, with lumen diameter probably as much as 10 jim less than 157jim). However, aga takg to account some immature liana samples notably African Gnetum species G. gnemon seems to fit templates for mesic woody dicotyledons. Phylogenetic conclusions The uncerta nature of relationships of Gnetales to or seed plants remas a central terest. The value of data from morphology anatomy this respect has not lessened but has creased because at least sev eral extct groups of seed plants may be sister groups closest to Gnetales. We can contue to accu mulate anatomical formation both on se sister groups on Gnetales, but DNA data for extct groups are not available. The importance of this cir cumstance cannot be overemphasized, because exam ples of effect on phylogeny of clusion versus exclusion of DNA data cladograms based on several data sets can be cited. Such examples are shown cladograms of Albert et al. (1994), Doyle et al. (1994), Nixon et al. (1994) that exclude fossils. In clado grams based on DNA evidence alone, Gnetales variably appear closer to angiosperms, sometimes showg angiosperms as sister group closest to Gnetales. When DNA evidence macromorphology are combed (Doyle Donoghue 1992; Albert et al. 1994, pp ; Doyle et al. 1994, p. 426), Pen toxylales Bennettitales are shown to be closest sister groups of Gnetales, with Glossopteridales Caytoniales separated from Gnetales by relatively small quantitative cladistic values. VEssELs The most important item from vegetative anatomy attemptg to decipher relationships of Gnetales is nature of gnetalean vessel. Thompson (1918) Bailey (1944, 1953) claimed that angiosperm

11 Welwitschia Gnetum (see cladistic papers cited above). The modifications of ephdroid perforation Thompson-Bailey hyposis, as derived from an as semblage of circular bordered pits like those on lateral walls (fig. 9). The condition Ephedra is re garded this concept as little modified from primitive vessels of Gnetales, accordance with virtually plate would clude crease size of perforations branes. Gnetum vessels would represent furr mod universal cladistic placement of Ephedra as basal to xylem vessels close to primary xylem. The accelerated modification of Gnetum perforation plate, evident ifications, notably enlargement coalescence of perforations so as to result a simple perforation plate (figs. 5 8), but ephedroid condition can still be seen Gnetum, most clearly narrow secondary as compared to lateral wall pits loss of pit mem of perforation plates is considered to represent an Ephedra (figs. 1 4) is regarded, accordg to ley (1944, 1953) considered gnetalean vessels to have so that lateral wall pittg is not excluded from Sattler (1982). Vessel presence two major vessel gnetalean vessel had separate origs. groups as a symplesiomorphy would be a significant piece of evidence for closeness of two groups, angiosperms would favor idea that Gnetales arose origated separately from those of angiosperms lie extension of, or a modification of, lateral wall pittg, whereas homoplastic orig of vessels Gnetales from a vesselless group of gymnosperms. This concept has been questioned by Muhammad The central reasons why Thompson (1918) Bai argument. Thus, foramate perforation plate of nature of perforation plate. In turn, nature sense Oliver, AW s. n.; three pits from end wall of tracheid to show small size of pores pit membranes. Bars at upper left each = 10 p.m. three pits showg conspicuous torus center of each connected with margo threads, at least some of which are tact. Fig. 10, Tetracentron 9 10 I - k Figs. 9, 10 SEM photomicrographs of pits from tracheids as seen on radial sections. Fig. 9, Ephedra gerardiana Wallich, Freitag 1425;.j T S68 INTERNATIONAL JOURNAL OF PLANT SCIENCES

12 CARLQUIST ANATOMY OF GNETALES S69 coalescent-foramate conditions (figs. 5, 6) as well as simple near-simple (figs. 7, 8) conditions, is easily explaed terms of flow rate. With mute foliar surface, selective pressure for simplification of perforation plate of Ephedra is mimal, whereas greater probable transpirational rates of broad leaves of Gnetum favor simplification of perfora tion plate to accommodate peak flows vessels. The perforation plate of Welwitschia is not generally entered to this sequence, presumably because vessels that species are so different from those Ephedra Gnetum. Also, consequence of considerg Ephedra basal genus of Gnetales means that determg relationships of Gnetales, nature of vessel Ephedra is more important than that of Wel witschia. Muhammad Sattler (1982), however, concentrate on attemptg to show a morphological contuum between vessels of selected angiosperms those of Gnetum rar than a contuum between Ephedra or Welwitschia vessels those of angio sperms. Muhammad Sattler (1982) designate cer ta perforation plates of Gnetum as scalaroid (e.g., fig. 7), apparently present se plates because y are considered to be suggestive of what vessels termediate between those of Gnetales those of primitive angiosperms might be like. These plates do not appear to me to be scalariformlike but, rar, can be more accurately categorized as foramate coales cent, exactly what would be expected a phyloge netic series which an ephedroid perforation plate is undergog simplification response to adaptation to higher peak flow rates. Muhammad Sattler (1982) have chosen for comparisons with Gneturn a series of angiosperms that are not at all basal with angiosperms terms of contemporary phylogenies, eir those that are based on cladistics (e.g., Qiu et al. 1993) or those that use more traditional methods (e.g., Thorne 1992). Vessels of Juglaceae, Myricaceae, Salicaceae are used by Muhammad Sattler (1982) for purposes of comparison, although those families seem, on basis of our current ideas of ir phylogenetic placement, unlikely to reta perforation plates like those of an cestral angiosperms. However; additional features vessel phylesis with regard to Gnetales are worthy of consideration have not been discussed to date to any appreciable extent at all. The circular bordered pits on lateral walls of Ephedra, Gnetum, or Welwitschia vessels are mark edly unlike those on lateral walls of angiosperm ves sels considered to be primitive. In turn, primitive gnetalean tracheid from which gnetalean vessels are to be derived bears little resemblance to tracheids widely considered to be primitive angiosperms. This con trast is stressed by juxtaposition of figs but needs additional explanation. A difference shape of perforations or shape of pits of lateral walls between Gnetales angiosperms has been considered by many authors, especially recent seed plant cladists (e.g., Doyle et al. 1994; Nixon et al. 1994). However, should shape be central issue? The circular shape of gnetalean pits is contrasted by se authors with scalariform type of pittg wide ly thought to be primitive angiosperms. However, if examed terms of adaptations to function that underlie phylesis, shape may not be primary con sideration. The gymnospermous tracheary element pit, as exemplified by those of Ephedra (fig. 9), contas a thick central torus, suspended by margo threads. The function of torus (Frey-Wysslg 1976) is closure of pit aperture durg aspiration, when a tracheid that has filled with air is sealed off, by displacement of tori so as to close pit apertures, from tracheary el ements that are still functiong. Moreover, margo strs feature wide spaces, spaces wide enough to accommodate passage of colloidal gold particles (Frey-Wysslg 1976) or or objects about 0.2 im diameter (Pansh De Zeeuw 1980). These spaces are much larger than micropores pit membranes of angiosperms (fig. 10; pits from an end wall of a Tetracentron tracheid are shown re, but lateral wall pits would be quite similar this respect). The margo strs, by ir slender, flexible strlike nature way y suspend torus, allow for movement of torus thus closure of gym nospermous pit durg aspiration. The angiospermous pit (fig. 10), on or h, lacks a torus is, any case, generally too rigid to function aspiration. Exclusion of potential air entry from a deactivated tra cheid of this type to a functiong water-filled tra cheid is prevented by excessively small size of micropores, through which air bubbles cannot pass (Carlquist 1988a, p. 108; Jarbeau et al. 1995). A cir cular shape is irrelevant to primitive angiosperm pit because a movable torus is not present; stead, scalariform shape covers lateral width of wall efficiently, maximizg pit area on a tracheid con tact with adjacent cells. Thus, water hlg con ductive safety mechanisms of circular gymnosper mous pit scalariform angiospermous pit represent two alternative functional modes, na ture of this difference should not be described primar ily terms of pit shape. The different shape is, as we have seen, a by-product of differences physiological mechanisms, not a cause for m or a fundamental phylogenetic feature its own right. The discovery of tori tracheary elements of Gne turn (Carlquist Robson 1995; Carlquist 1996a, 1996b) is of significance because thus pits of Gne turn tracheids are not transitional from those of cycads to those of conifers, as thought by Frey-Wysslg (1976, p. 61), simplified perforation plates of Gnetum that allegedly look transitional ( scalaroid ) to scalariform angiosperm plates must be viewed as modifications of lateral wall pittg that, Gnetum, is not at all like that of angiosperms. The fact that tra cheary elements of Gnetum are not fundamentally dif ferent from those of Ephedra can be considered evi dence for monophyly of Gnetales. Thus, a transitional position for Gneturn between more primitive Gnetales

13 conifers, Gkgo, Gnetales is that circular bordered been noticed for many years (Bierhorst 1960). Such pits tercalated to helical secondary wall bs not yet been considered, largely because emphasis has se circular pits is suggested by ir structure tercalated bordered pits are not present pri mary xylem of angiosperms. Thus, those who visual been placed on this feature from a descriptive or tax onomic pot of view. One functional possibility for of primary xylem tracheary elements of conifers, of angiosperms. communication), ultrastructure of sieve elements of angiosperms is not supported. There is similar ev idence phloem: accordg to Ray F Evert (personal Gnetales is like that of gymnosperms rar than that of Gnetales also offers evidence about phyloge vessel elements of Gnetum (fig. 11, 12) Ephedra, ements. A special feature of helical bs of sec bordered helical thickengs, composed of lignified secondary wall material, are attached to th pri mary walls, just as many vascular plants. As has been described extensively, gyres of helical thick engs such metaxylem elements can stretch out, accommodatg crease length by tracheary el ondary wall material metaxylem tracheids of S70 INTERNATIONAL JOURNAL OF PLANT SCIENCES Figs. 11, 12 SEM photographs of walls of primary xylem tracheary elements from radial sections of stems of Gnetum leyboldii, McPherson Fig. 11, Pit membranes lackg circular pitlike structures, top bottom, which refore are perforations. Fig. 12, Smaller circular bordered pits tercalated to helical thickengs; a pit membrane is clearly present pit at left, neir of pitlike structures is likely a perforation. Bars at upper left = 10 p.m. A highly significant formation primary xylem netic position of Gnetales. In metaxylem tracheids pits are tercalated to se bs. This situation has ize Gnetales as transitional to angiosperms are faced with explag presence of this neglected feature Gnetales. I hyposize tha presence of circular bordered Gkgo, Gnetales is related to functions that have figure 11, which two circular pits are, fact, 2 A I

14 CARLQUIST ANATOMY OF GNETALES S71 not pits, but perforations. Lack of pit membranes is well shown for se two circular areas, earlier studies by means of light microscopy were not able to demonstrate lack of pit membranes with any certaty. Presumably, perforations of this sort would occur primary xylem of Gnetum Ephedra but not Gkgo or conifers, which are vesselless. A second likely functional significance of ter calated circular bordered pits of primary xylem ele ments of Gnetales has to do with torus-margo or ganization of secondary xylem tracheary element pits. By formg circular bordered pits, conductive capabilities of large spaces that characteristically occur among margo strs are available. Also, pit aspiration mechanism, by virtue of presence of a torus, is also potentially present. The primary xylem elements evidently have negligible microporosities lateral primary wall between helical gyres of secondary wall material, accordg to my SEM pho tographs (figs. 11, 12). Thus, conduction capabilities of primary xylem tracheary elements with tercalated circular pits are greatly enhanced by availability se pits of spaces between margo strs. This potential mechanism for enhancement of conduction is available to Gkgo conifers, as well as Gnetales, has not received comment as a functionally valuable feature. OTHER XYLARY FEATURES Ephedra, Gnetum, Welwitschia were claimed, almost all earlier literature, to have only two types of tracheary elements: tracheids vessel elements. Axial parenchyma was claimed to be present Ephed ra Gnetum (Martens 1971), although drawgs by Greguss (1955) claim fiber-tracheids for Gnetum gne mon. Fiber-tracheids are not present Welwitschia, but fiber-tracheids are present all species of Ephedra Gnetum (fig. 13) studied thus far. The phyloge netic significance of this is terestg several re spects. Dimorphic tracheary elements tracheids co existg with fiber-tracheids a wood this stance do not appear to be basal with angiosperms. Among dicotyledons that appear at all close to beg basal recent phylogenetic treatments, dimorphism of this kd has been identified only Austrobaileyaceae (Carlquist 1 988c), such dimor phism has not been identified any gymnosperms or than Ephedra Gnetum. Axial parenchyma is present numerous gymnosperms primitive dicotyledons, it occurs Welwitschia (Cariquist Gowans 1995) all species of Gnetum (figs ) a few species of Ephedra (Carlquist 1992). In addition to differg from wood of or gymnosperms from woody dicotyledons or than Austrobaileya some more specialized phylads havg tracheid dimorphism, Gnetum Ephedra woods are furr distct distribution of fiber tracheids with respect to tracheids. In dicotyledons that have tracheid dimorphism, tracheids occur a vas icentric pattern, whereas fiber-tracheids are distal to vessels (Carlquist 1985). In Gnetum gnemon, fiber tracheids appear distributed romly with relation to tracheids; lianoid Gnetum species, fiber-trache ids tend to be radial files alternatg with tracheids (fig. 16). In Ephedra, fiber-tracheids are arranged dif fusely or tangential bs ( diffuse--aggregates is term applied to that distribution, but this term is normally applied to axial parenchyma, not to fiber-tra cheids). In no species of Ephedra or Gnetum was a vasicentric distribution of tracheids seen. The observation that both Ephedra Gnetum have fiber-tracheids toger with tracheids, a condi tion orwise unreported gymnosperms, suggests this feature may be plesiomorphic Gnetales, which case lack of fiber-tracheids Welwitschia would have to be terpreted as an autapomorphy. Gnetales are distctive havg both multiseriate uniseriate rays, although Ephedra, multiseriate rays predomate uniseriate rays are relatively scarce. In Welwitschia, rays are more difficult to ob Figs Light photomicrographs of Gnetum to show differentiation between cell types. Fig. 13, Gnetum gnemon, Carlquist 8088; radial section of root showg, bottom to top, a tracheid, two fiber-tracheids contag septa starch, a pair of cells from an axial parenchyma str, separated by a lignified wall, which is vertical this photograph. Fig. 14, Gnetum cuspidatum Blume, transection of stem, showg three cell types axial secondary xylem: tracheids have thick walls; fiber-tracheids have walls of termediate thickness; axial paren chyma cells (some of which conta crystals) have very th walls, some of which are not clearly deleated. Fig. 15, Gnetum schwackeanum, Farifio 425; transection of stem, showg a portion of secondary xylem which axial parenchyma cells (th gray walls) are relatively abundant. Fig. 16, Gnetum leyboldii, McPherson 10003; transection of juncture between secondary phloem (above) secondary xylem (below), to show that files of phloem parenchyma (narrower cells with th walls) are tercontuous with files of narrow cells secondary xylem. Wide cells secondary phloem = sieve cells; wide cells secondary xylem = tracheids. Furr explanation text. Figs. 13, 14, 16, Magnification scale above figure 7; fig. 15, scale above figure 15 (divisions = 10 p.m). Figs Light photomicrographs of Stem transections of Gnetum (figs ) Ephedra. Fig. 17, Gnetum schwackeanum, Faro 425; portion of outer stem to show (top to bottom), sclerenchymatous cylder, lateral meristem region, recently formed vascular str (five vessels visible secondary xylem), conjunctive tissue which several laticifers are present. Fig. 18, Gnetum cuspidatum, Carlquist 8091; portion of vascular str (phloem fibers are dark, near center), above which to right of which are radially oriented files of cells related to lateral meristem activity. Fig. 19, Gnetum latifolium Blume, Carlquist 8087; transection of outer portion of underground stem; a th gray b about one-fourth from top of photograph delimits phellem (above) from cortex; cortex, re are scattered gelatous fibers (lower right) a few sclereids. Fig. 20, Ephedra cal(fornica, cortex of primary stem; arrows pot to gelatous fibers, which are pale gray. Figs. 17, 19, Scale above figure 17 (divisions = 10 m); figs. 18, 20, scale above figure 15.

15 SZ NIDS INVId 1O 1VN1f1Of IVNOIIVNINI ZLS

16 CARLQUIST ANATOMY OF GNETALES S73 4: 1%i )f i F?;

17 Gnetales are rich autapomorphies with respect to tracheids, yet without vasicentric distribution of iracheids as all angiosperms that have tracheid dimor phism. The cladograms by Albert et al. (1994), Doyle good example of a feature restricted to Gnetum. Al though developmentally younger stages of Gnetum plants conta laticifers commonly (Martens 1971), la ticifers are not evident older stems roots of velopment all species of Gnetum. Vesturg pits perforation plates). Several bark features, such as pres planations may be entertaed, however: richness autapomorphies Welwitschia (which has phellem et al. (1994), Nixon et al. (1994) that dicate that likely sister group of Gnetales is a vesselless group stem xylem anatomy. Laticifers (fig. 17) are a some Gnetum species. Laticifers thus must be consid ered a feature related to development, absence cannot be claimed for any species of Gnetum without study of developmental stages; laticifers should be as sumed to be likely to be present at some stage of de Several wood features, especially some of those that but lacks or bark features) suggests it has departed cies studied was judged to lack vesturg altoger. ence of diffusely distributed gelatous fibers, suggest Among likely autapomorphies Welwitschia are lack of fiber-tracheids, vessels with simple perforation (rar than thickengs) on ner surfaces of vessels calcium oxalate crystals lg tercellular features. are newly reported this series of papers, suggest affity between Ephedra Gnetum (e.g., presence from an ancestral gnetalean stock numerous ways. plates like those of lateral walls of vessels, grooves tracheids, presence of secretory cavities, presence lateral meristem from phelloderm. One peculiar feature Although much formation valuable major phy not reported earlier decades appears to lk Ephedra This feature is orwise unknown vascular plants, haps occurrence of mute tercellular crystals lost ancestors of Gnetum as that phylad shifted to mesic localities. Gnetum has crystals with cells of wood, phloem, bark. logenetic ecological contexts emerged from survey of bark wood at a species level, very few features likely to be of taxonomic significance at tures at present. One dication that some features of liar ray astrosclereids Gnetum klossii Merrill. Gne terest may await discovery is presence of pecu of gymnosperms are reby supported by wood data. of vessels is also a likely autapomorphy for Gnetum. of large crystal-bearg fibrosclereids, orig of of both fiber-tracheids tracheids, foramate Ephedra Welwitschia is a plesiomorphy, a feature fragenenc levels were uncovered. In part, samplg of Ephedra Gnetum is too poor to delimit se fea As with laticifers, one can fd vessels some species at least form reported se two genera. Per Welwjtschja: occurrence of mute tercellular a resemblance between two genera. Alternative ex which vesturg is quite conspicuous, but no spe Relationships with Gnetales is volved, although time of orig of se new as claimed by La Rivière (1916) as confirmed of agg stem. The orig of lateral meristem ac of Bailey (1944, 1953) that vessels arose depen Gnetum has been cited as a pot of resemblance to my opion. The opions of Thompson (1918) The presence of both multiseriate uniseriate rays likely to be sister groups of Gnetales. Both uniseriate biseriate rays are present wood of Bennet at least a few of se may beg as uniseriate rays. serve, certaly multiseriate rays are present, but angiosperms, but it may be a plesiomorphic feature gymnosperms than has generally been appreciated. Gnetales. In addition, it may occur more widely with Presence of both types of rays is common cycads (Greguss 1968). More pertent is wher both mul tiseriate uniseriate rays are present gym nosperms that are placed close to Gnetales recent cladograms, especially those orders of gymnosperms titales (Wiel 1906). Pentoxylon has some biseriate nu-mittre 1957), Bucklia has rays uniseriate to tris multiseriate rays as well as uniseriate ones (Vish rays. Thus, presence of multiseriate rays termixed this feature is refore best not terpreted as a syn bark of G. gnemon is similar that cortical tissue eriate (Bose 1953), Doyle Donoghue (1986) score Caytoniales as havg at least some multiseriate nosperms commonly considered close to Gnetales, apomorphy that would lk Gnetales only with angio phylogenetic significance is presence of successive cambia (figs. 17, 18) that origate parenchyma de rived from a kd of lateral meristem activity. I do not place cambial modifications described by Lev-Ya with uniseriate rays appears widespread among gym sperms. The or wood or stem feature of major potential those conditions appear to be distortion of cambial ori entation owg to creasgly contorted conformation Cariquist 1996a, l996b). The orig of new cambia tivity all Gnetum lianoid species is ner cortex, cambia is evidently much delayed compared with or phelloderm rar than cortex. Therefore, varia synapomorphies or symplesiomorphies but a series of autapomorphies. dun Aloni (1993) for Ephedra this category; more recent studies (Carlquist Robson 1995; ig of those lianoid species. The lateral menstem files Welwitschia, however, are derived from tions cambial activity with Gnetales are likely not dently angiosperms Gnetales are supported nated from a vesselless gymnosperm group with both multiseriate uniseriate rays. This terpretation is reforced by demonstration of tori pits of at Wood stem features of Gnetales do not, re here, on basis of new concepts as well as previ ously cited evidence. Gnetales appear to have origi least five species of Gnetum demonstration that fore, dicate any synapomorphies with angiosperms, Gnetum Ephedra have both tracheids fiber S74 INTERNATIONAL JOURNAL OF PLANT SCIENCES

18 CARLQUIST ANATOMY OF GNETALES S75 turn gnernon appears rar different from lianoid species of Gneturn bark wood characters, but study of sister species of G. gnernon, Gneturn cos taturn K. Schum. is needed, much fraspecific diversity with G. gnernon remas to be explored. Attention should be focused on wher Markgraf s (1930) sections subsections which correspond with habit geography are confirmed by newer data. The placement of New World lianas same section as G. gnernon rar than section that cludes Indomalesian Asiatic lianoid Gneturn species is one feature of Markgraf s (1930) system that vites revestigation. Although wood bark data seem to offer a number of features of major phylogenetic ecological significance, molecular data will likely be of prime importance for frageneric systematics of Ephedra Gneturn. At present, we have few reliable hts about evolution speciation patterns with se two remarkably widespread divergent genera, even our concepts about what constitutes a species with se genera are tentative. Acknowledgments For grants that permitted collection of specimens of Gneturn Welwitschia, I express my appreciation to National Geographic Society American Philosophical Society. Collections by Gordon Mc Pherson were made possible through courtesy of Dr. Peter H. Raven. Dr. Jack Fisher shared material of Gneturn rnicrocarpurn. Livg material of Gneturn was provided by Stephen Morgan of Botanic Garden of University of California, Riverside. Dr. Dennis Stevenson of New York Botanic Garden aided with identifications material from herbarium of that stitution. The studies were begun at Rancho Santa Ana Botanic Garden, which aided by providg sup plies. Literature cited Albert VA, A Backlund, K Bremer, MA Chase, JR Manhart, BD Mishler, KC Nixon 1994 Functional constrats rbcl evi dence for l plant phylogeny. Ann Mo Bot Gard 81: Bailey 1W 1944 The development of vessels angiosperms its significance morphological research. Am J Bot 3 1: Evolution of tracheary tissue of l plants. Am J Bot 40:4 8. Bierhorst DW 1960 Observations on tracheary elements. Phyto morphology 10: Bose MN 1953 Bucklia sahnii sp. nov. from Jurassic of Rajmahal Hills, Bihar. Palaeobotanist 2: Bower FO 1881 On furr development of Welwitschia mirab ilis. Q J Microsc Res 21: Butler V, CII Bornman, RF Evert 1973 Welwitschia mirabilis, vas cularization of a one-year-old seedlg. Bot Gaz 134: Cariquist S 1975 Ecological strategies of xylem evolution. Univer sity of California Press, Berkeley Los Angeles. 259 pp. 1984a Vessel groupg dicotyledon woods: significance relationship to imperforate tracheary elements. Aliso 10: b Wood anatomy of Trimeniaceae. Plant Syst Evol 144: Vasicentric tracheids as a drought survival mechanism woody flora of sourn California similar regions. Aliso 11: a Comparative wood anatomy. Sprger Verlag, Ber l. 436 pp. 1988h Near-vessellessness Ephedra its significance. Am J Bot 75: c Tracheid dimorphism: a new pathway evolution of imperforate tracheary elements. Aliso 12: Wood bark anatomy of New World species of Ephedra. Aliso 12: Wood, bark, pith anatomy of Old World species of Ephedra summary for genus. Aliso 13: Wood bark anatomy of Gnetum gnemon L. Bot J Ln Soc 116: a Wood bark anatomy of lianoid Indomalesian Asiatic species of Gnetum. Bot J Ln Soc 12 1: b Wood, bark, stem anatomy of New World spe cies of Gnetum. Bot J Ln Soc 120: Carlquist S, DA Gowans 1995 Secondary growth wood his tology of Welwitschia. Bot J Ln Soc 118: Carlquist 5, AA Robson 1995 Wood bark anatomy of African species of Gnetum. Bot J Ln Soc 118: Dickison WC, PM Rury, GL Stebbs 1978 Xylem anatomy of Hibbertia relation to ecology systematics. J Arnold Arbor Harv Univ 59: Doyle JA, MJ Donoghue 1986 Seed plant phylogeny orig of angiosperms: an experimental approach. Bot Rev 52: Fossils seed plant phylogeny reanalyzed. Britton ia 44: Doyle JA, MJ Donoghue, EA Zimmer 1994 Integration of morpho logical ribosomal RNA data on orig of angiosperms. Ann Mo Bot Gard 81: Esau K 1969 The phloem. Hbuch der Pflanzenanatomie. 2d ed. Vol 2. Borntraeger, Berl Stuttgart. Fahn A, E Werker, P Baas 1985 Wood anatomy identification of trees shrubs from Israel adjacent regions. Israel Acad emy of Sciences Humanities, Jerusalem. 221 pp. Frey-Wysslg A 1976 The plant cell wall. Hbuch der Pflan zenanatomie. 2d ed. Vol 3, Pt 4. Borntraeger, Berl Stuttgart. 294 pp. Greguss P 1955 Identification of livg gymnosperms on basis of xylotomy. Akadémiai Kiado, Budapest. 263 pp Xylotomy of livg cycads. Akadémiai Kiado, Budapest. 260 pp. Jarbeau JA, FW Ewers, SD Davis 1995 The mechanism of waterstress-duced embolism two species of chaparral shrubs. Plant Cell Environ 18: La Rivière H 1916 Sur l anatomie et l epaississement des tiges du Gnetum moluccense Karst. An Jard Bot Buitenzorg 15: Lev-Yadun 5, R Aloni 1993 Variant secondary growth old stems of Ephedra campylopoda C. A. Mey. Bot J Ln Soc 112: Maheshwari E V Vasil 1961 Gnetum. Council of Scientific Industrial Research, New Delhi. 142 pp. Markgraf F 1930 Monographie der Gattung Gnetum. An Jard Bot Buitenzorg 10: Martens P 1971 Les Gnétophytes. Hbuch der Pflanzenanatomie. 2d ed. Vol 12, Pt 2. Borntraeger Berl Stuttgart. 295 Pp. Metcalfe CR, L Chalk 1950 Anatomy of dicotyledons. Clar endon, Oxford pp. Muhammad AF R Sattler 1982 Vessel Structure of Gnetum orig of angiosperms. Am J Bot 69: Nixon KC, WL Crepet, D Stevenson, EM Friis 1994 A reevaluation of Seed plant phylogeny. Ann Mo Bot Gard 8 1: Pansh AJ, C De Zeeuw 1980 Textbook of wood technology. 4th ed. McGraw Hill, New York. 722 pp. Patel RN 1965 A comparison of anatomy of secondary xylem roots stems. Holzforschung 19: Pearson HHW 1929 Gnetales. Cambridge University Press, Cam bridge. 194 pp. Qiu Y-L, MW Chase, DII Les, CR Parks 1993 Molecular phylo

19 Wiel GR 1906 The American fossil cycads. Carnegie Institution of Washgton, Washgton, D.C. angiosperms. Bot Gaz (Crawfordsville) 69: plants. Bot Rev 58: Series). India Pentoxyleae. Palaeobotanist 6: Thompson WP 1918 Independent evolution of vessels Gnetales Thorne RF 1992 Classification geography of flowerg Vishnu-Mittre 1957 Studies on fossil flora of Nipania (Rajmahal ARA Gorts-van Rijn, ed. Flora of Guianas. Vol 9. gne?non. Ann Bot 39: Rao AN, H Keng 1975 Anomalous secondary growth Gnetum Sykes MG 1910 The anatomy of Welwiischia,nirabili,c seed 2 lg adult states. Trans Ln Soc Loud, Ser Bot, 7: ter Welle BJH, P Detienne 1991 Wood timber. Pages genetics of Magnoliidae: cladistic analysis of nucleotide se quences of plastid gene rbcl. Ann Mo Bot Gard 80: S76 INTERNATIONAL JOURNAL OF PLANT SCIENCES