Wood and Stem Anatomy of Woody Aniaranthaceae S.S.: Ecology, Systematics and the Problems of Defining Rays in Dicotyledons

Botanical Journal of the Linoean Society, 2OO, 143, 1—19. With 30 figures

Wood and stem anatomy of woody Aniaranthaceae s.s.: ecology, systematics and the problems of defining rays in dicotyledons

SHERWIN CARLQUIST*

Santa Barbara Botanic Garden, 1212 Mission Canyon Road, Santa Barbara, CA 93105, USA

Received December 2002; accepted for publication May 2003

Wood and stem anatomy is studied for seven species of six genera (root anatomy also reported for one species) of Amaranthaceae es. Quantitative data on vessels correlate closely with relative xeromorphy of respective species, agreeing with values reported for dicotyledons without successive cambia in comparable habitats. Libriform fibre abundance increases and vessel diameter decreases as stems and roots of the annualAmaranthus caudatus mature. Long, thick-walled fibres in Boseo yervwnoro may be related to the upright nature of elongate semi-climbing stems. Non-bordered or minutely bordered perforation plates characterize Amaranthaceae, as they do most other Caryo phyllales. Amaranthaceae have idioblastic cells containing druses, rhomboidal crystals or crystal sand: these forms intergrade and seem closely related. Rays are present in secondary xylem of the Amaranthaceae studied. Cells inter mediate between ray cells and libriform fibres occur in Charpentiera elliptica. Degrees of diversity in rays and reports of raylessness in Amaranthaceae induce discussion of definition and identification of rays in dicotyledons; some sources recognize both rays and radial plates of conjunctive tissue in Amaranthaceae. The action of successive cainbia is described: lateral meristem periclinal divisions produce secondary cortex externally, conjunctive tissue internally and yield vascular cambia as well. Vascular cambia produce secondary phloem and secondary xylem, in both ray and fascicular zones, as in a dicotyledon with a single cambium. Identification of rneristem activity and appreciation of varied ray manifestations are essential in understanding the ontogeny of stems in Arnaranthaceac cwhich have recently been united with Chenopodiaceae. © 2003 The Linnean Society of London, Botanical Journal of/he Linnean Society, 2003, 143. 1—19.

ADDITIONAL — — KEYWORDS: cambial — variants Caryophyllales Chenopodiaceae conjunctive tissue —

— — ecological wood anatomy lateral meristems non-bordered perthration plates — raylessness — secondary cortex

— successive cambia — vascular rays.

INTRODUCTION dicotyledons as well as in Gnetales. Also important, as shown by the species in this study, are the variations The occurrence of successive cambia has been known within Amaranthaceae in wood and stem anatomy. For many for years (see Schenck, 1893; Solereder, 1908; example, Rajput (2002) and Rajput & Rao (1999, 2000) Pfeiffer, 1926). Successive cambia are characteristic of report raylessness in Amaranthaceae, although none Amaranthaceae; probably some species with limited of the species in this study is rayless. The findings in secondary growth may lack them. Amaranthaceae are the present paper call into question the definitions an important group in attempts to understand how and kinds of rays and raylessness. successive cambia originate and what their products The species that comprise this study are among the are. These issues are in need of a careful reassess ‘woodier’ in Amaranthaceae, judging from stem diam ment, both in ‘core Caryophyllales’ and elsewhere in eter, although an annual in the study. Ama ranthus caudatus, offers a useful contrast. The relationships between wood anatomy and habit are examined. The tE-mail: s.carlquist@verizonnet habitats of the Amaranthaceae in the present study

‘7:4/ji(,7 t;/?t .2-TJ

JslnbnIuva s WOOD AND STEM ANATOMY OF AMARANTHACEAE 5

with a highly oblique angle (or with tails) are rela thickness of libriform fibres is relatively uniform, but tively unusual, and occur mostly in narrow vessels fibres are notably thick walled in Bosea yervainora (Fig. 14). (Table 1, column 7). The mean F/V ratio (ratio of Vessel wall thickness (Table 1, column 5) is least in length of libriform fibres to length of vessel elements) the annual Ainaranthus caudatus (Figs 1, 3) and is 3.81 (Table 1, column 8), a ratio unusually high for greatest in Gel osia floribunda (Fig. 12). Pits on lateral dicotyledons. In dicotyledons in general, the mean ves waLls of vessels (Fig. 15) have pit cavities almost uni sel element length and fibre length figures would give formly 5 jim in diameter (as measured in an axial an F/V ratio of 2.03 see figures in Metcalfe & Chalk, direction). 1950: 1360). Occasional nuclei were observed in libriform fibres, Qualitative vessel features although an extensive search was not made for living Vessel restriction patterns occur in Aniaran thus call fibres. Because all of the species of Amaranthaceae datus, notable for the occurrence of most vessels in the examined by Rajput (2002), a worker experienced in last-formed portions of vascular bands of stems observation of living fibres, were shown by him to have (Fig. 1); this species shows absence of vessels from living fibres, living fibres can be assumed to occur last-formed vascular bands of roots (Fig. 3). In all spe widely within Amaranthaceae. However, a number of cies of Amaranthaceae studied, vessels do not contact species in the present study have stems of large diam rays and are mostly confined to central strips within eter, and many libriform fibres in inner portions of fascicular areas, clearly a vessel restriction pattern these stems, which can be ten or more years old, can (Figs 1, 3. 5. 7. 9, 11, 12, 19. 21, 23, 24, 27, 28, 30). be assumed to lack living nuclei. Even taking a liberal view of ray presence, this distri bution is characteristic of the family. Axial parenchynma Vessels are circular to oval in transectional outline in all of the Amaranthaceae studied, axial paren (Figs 1,3,4, 7,9, 11, 12, 19, 21, 23, 24, 27, 28 and 30) chyma is scanty paratracheal (vasicentric). Axial in Amaranthacene. No vessels could be characterized parenchyma is not obvious in any of the photographs as truly angular in shape. Vessel wall thickness herewith (except for perhaps Fig. 5). Walls of axial (Table 1, column 5) is relatively uniform in Amaran parenchyma ceLls are similar in thickness to those of thaceae, and only slightly less than the wall thickness libriform fibres in most species, so that discrimination of the libriform fibres in the respective species. between these two cell types in transections requires Perforation plates are non-bordered (Fig. 16) or very care. In longisections, axial parenchyma is more slightly bordered (Fig. 17). This feature can be readily identified because one can detect strands in observed with either light microscopy or SEM. most species. However, axial parenchyma in although a border is more clearly rendered by SEM. Arthraeruu leubnitziae has axial parenchyma cells not Lateral wall pits have pit cavities circular to some subdivided into strands. Strands of two cells predom what oval in outline (Fig. 15). Pit cavities are narrow inate in the remaining species studied, except for and slit-like (Figs 15, 17). Although some pit aper Bosea yervamora, in which strands consist mostly of tures are longer than the diameter of the pit cavities four cells. they overlie, they do not form what could be called grooves extending to any appreciable extent beyond Vascular rays the pit apertures. The description of rays in Amaranthaceae poses a sig nificant problem because definition of rays and ray litiperforate tracheary elements lessness. discussed later, is not the same as in families All imperforate tracheary elements in the Amara.n of dicotyledons with a single cambium. However. mul thaceae studied are non-septate libriform fibres with tiseriate rays can readily be seen (Figs 1, 2 [transi simple non-bordered pits. Imperforate tracheary ele tional to conjunctive tissue at right and below], 3,4, 7— ments in the species studied range from 474 to 11, 13, 14, 22, 23—26, 29 [transitional to conjunctive 1528 jim (Table 1, column 6), which parallels lengths tissue, below right]). These multiseriate rays are, in of vessel elements in the respective species. Wall many cases, extensions of primary rays. That they are

Figures 1—4. Sections of Amaranthus caudatus stem (1,2) and root (3, 4). Fig. 1. Transection, showing all vascular tissue external to pith; vessels decrease in size toward outside of plant. Scale bar = jim. Fig. 2. Tangential section; rays at left, ray parenchvma transitional to conjunctive tissue at lower right; storied cells in both rays and conjunctive tissue. Scale bar = 50 jim. Fig. 3. Transection, showing all vascular tissue external to pith; very few vessels formed in outer bands of vascular tissue. Scale bar 50 mi. Fig. 4. Transection, showing vascular cambium (larger arrows at left and right) and a multiseriate ray (vertical arrow, bottom). Scale bar = 20 jim.

JSIflWIIIVO S 9 WOOD AND STEM ANATOMY OF AMARANTHACEAE 7

produced by cambial activity is evident in such a pho Figure 26, both lignified and non-lignified ray cell tograph as Figure 4, in which a vascular cambium is walls may be seen. In Figure 25, the non-lignified cell present across the width of the photograph. Secondary walls are thin and stain poorly and appear faded out’. phkem (with identifiable sieve-tube elements and In these rays, the cells are upright and block-like, with companion cells) is present external to areas of libri blunt top and bottom ends. There are also rays in this form fibres in Figure 4. However, only (phloem ray) species that appear to consist of fusiform cells (Fig. 25; parenchyma external to the cambium is present in right third of Fig. 26); some of these ray cells are sto areas internal to which (xylem) ray parenchyma cells ried, and some are subdivided. (with relatively thin but lignified walls) are present in Are fusiform ray cells to be interpreted as ray cells the multiseriate ray present, left of centre. These or as libriform fibres? In fact, even transitions observations correspond with the examples of rays between blunt tips and pointed tips on ray cells occur commonly cited for dicotyledons Uniseriate rays are (Fig- 25, rays left of centre), so denying ray status to common. Only a few are present in Arthraeru-a leub short pointed cells seems arbitrary and more con nitziae (Figs 5, 6). cerned with definitions than are useful in describing A species with diverse manifestation of multiseriate the conditions present. There are true libriform fibres rays. Charpentiera elliptica, deserves careful analysis. (e.g. dark strips on either side of the vessel in Fig. 26) In Figures 23, 24, strands of secondary phloem are that are much longer than the pointed ray cells and evident: these are sheathed externally by conjunctive never subdivided into strands. A case can be made tissue converted into sciereids. Internally, the strands that in the adaxial portions of vascular bands of of secondary phloem contact secondary xylem (consist Arthraerua leuhnitziae, multiseriat.e rays are absent ing of libriform fibres and vessel elements). Similar because the cells in potential multiseriate ray areas formations can be seen in C. densiflora (Fig. 21). These are not distinguishable from libriform fibres (Fig. 6). radially orientated strips of secondary xylem, capped In Charpentiera elliptica transections, zones of what by strands of sclerenchyma-sheathed phloem, are very appear to be fibres (they may be short fibres, but are similar to the radial fascicular tissue plates in Bosea still definable as fibres) occur as strands among the yervatnora (Fig. 7). In Charpentiera, as in Bosea, the lignified parenchyma cells in the adaxial portions of tissue between these radial plates must be considered multiseriate rays, sometimes as strands, sometimes in vascular rays, derived from vascular cambia (see other arrangements (Figs 23, 24). Fig. 4 or Fig. 8 for cambium portions that produced The ray cells of the Arnaranthaceae studied are all these vascular rays). or almost all upright, and so the rays qualify as Pae If one views what I am terming the wide multiser domorphic Type II (Carlquist, 2001), except for the iate ray areas of Charpentiera. one sees that abaxial uniseriate rays of Arthraerua leubnitziae (Fig. 6), portions of these rays consist of parenchyTna cells with which fall into the unusual Paedomorphic Type III. lignified walls (alternatively, this could be viewed as The presence of storeying in some rays (Figs 2 25, 26. thin-walled scierenchyma, but ray cells with lignified 29) indicates radial longitudinal divisions that widen secondary walls are usually termed ray parenchyma). the rays, an action that seems to substitute for origin Note that thin-walled cells occur in the abaxial por of rays as uniseriate or narrow multiseriate rays, as is tions of multiseriate rays of Arthraerua (Fig. 5), but typical in woody dicotyledons. adaxial cells in these regions consist of fibres indistin guishable from the libriform fibres of fascicular areas (Fig. 6). In Charpentiera elliptica, there is a variety of NON-XYLARY STEM AND ROOT HISTOLOGY histological conditions in the multiseriate ray areas. If one views tangential sections of C. elliptica vascular Cortex bands (Figs 25, 26), one sees multiseriate rays. In the Stem cortical parenchyma cells are seen along the tops ray near right in Figure 25 and in the ray at left in of Figures 1,4, 8, 10—12, 19, 20. Cortical cells are tan-

Figures 5—8. Stem sections of Arthraerua ieuhnitziae (5, 6) and Bosea yertaniora (7, 8). Fig. 5. Transection, showing vascular bands; apparent indentations in fibre bands (vertical bands) are thin-walled multiseriate rays (vertical arrow); a few uniseriate rays are present in vascular bands. Scale bar 50 im. Fig. 6. Tangential section of vascular bands to show fibres; a short uniseriate ray is indicated by the vertical arrow Scale bar 50 jIm. Fig. 7. Transection; typical vascular bands are composed of fascicular areas separated from each other by multiseriate rays laterally; conjunctive tissue separates vascular the tissue of successive bands. Scale bar = 50 jim. Fig. S. Transection, to show (from top to bottom) fibres and sciereids in outer cortex; randomly organized cells of inner cortex; secondary cortex (radial rows of thin-walled cells); a lateral meristem in which, at left, a vascuiar cambium portion is active and has given rise to a strand of fibres; and, at- bottom, phloem and xylem of a mature vascular band. Scale bar = 20 tim.

JsinWiuvo 8 9 WOOD AND STEM ANATOMY OF AMARANTHACEAE 9 gentially stretched and often subdivided, evidence of rower closer to the pith, wider further out in Nototri accommodation to increase in stern circumference, chiurn sandwicense (compare Figs 27 and 30); the Strands of fibre-like sclereids are a feature that reverse pattern is true inAmaranthus caudatus stems demarcates the primary cortex from secondary cortex (Fig. 1) and roots (Fig. 3). in several species (Figs 8, 11, 12: near tops of As seen in tangential section, conjunctive tissue is photographs). composed of axially elongate cells much like ray cells (Fig. 2, right third; Fig. 29, bottom right). These cells Pith are often storied, a pattern that agrees with that of Pith bundles were observed in Amaranth us caudatus rays in the two figures cited, and indicates radial lon (Fig. 3, lower left) and Nototrichium sandwicense gitudinal divisions either in lateral meristem cells or (Fig. 30). Some secondary growth probably occurs in internal derivatives of them. these bundles, judging from presence of compressed primary phloem and secondary phloem. Lateral meristems Divisions do not occur continually in the secondary cor Conjunctive tissue tex. When periclinal divisions do occur, these may be Cells of conjunctive tissue are relatively thin walled; termed lateral meristem. Active divisions can be the walls are generally not lignified (Figs 1, 3, 5, 7, 9, traced in tangential lines (Figs 11, 12) but not without 11, 12, 19, 21,23,24,27,28). Conjunctive tissue bands exception (Fig. 10). Consequently, the lateral meristem range mostly from three to eight (commonly four) cells is not restricted to a single layer, as it is in the non- in radial thickness. Variation in a single stem is not diffuse lateral meristem of Barbeuit-z (Carlquist, unusual. Conjunctive tissue bands are mostly nar 1999a). The lateral meristems of Amaranthaceae qual

Tablet. Wood features of Amaranthaceae

1 2 3 4 5 6 7 8 9 Species VG VU VM VL VW FL FW FV ME

A,naranthus caudatus 1.57 22 .52 137 2.2 462 3.0 3.4 58 Arthraerua leubnitziac 2.07 15 110 124 2.9 347 2.9 3.2 17 Boscayeruatnora 1.81 29 78 215 3.1 1528 5.2 7.1 80 Celosia floribunda 3.56 34 131 179 4.0 654 3.8 3.6 46 Charpentiera densiflora 1.67 43 24 148 2.9 442 4.1 3.0 265 Charpcntiera elliptiea 1.75 52 15 171 3.6 477 3.6 28 581 Nototrichium sandwicense 2.90 40 30 134 3.9 474 3.0 3.9 179 Above species, averaged 2.19 34 63 158 3.1 633 3.1 3.8 175

Key: 1 (VG), mean number of vessels per group: 2 (VD), mean vessel lumen diameter; 3 (VM), mean number of vessels mm2 of transection excluding cortex and pith; 4 (VL), mean vessel element length, Am; 5 (VW), mean vessel wall thickness, (Am; 6 (FL), mean libriform fibre length. cm; 7 (FW). mean libriform fibre wall thickness, pin: 8 (F’V). F/V Ratio (mean libriform fibre length divided by mean vessel element length); 9 (ME), Mesomorphy Ratio (mean vessel lumen diameter times mean vessel element length divided by mean number of vessels mm2). Additional information in Material and Methods.

Figures 9—12. Stem sections of Celosia floribunda. Fig. 9. Transection of outer stem to show mature vascular band below, vascular band in active growth above, and, at extreme top, tangentially stretched cortical parenchyma cells. Scale bar = 50 pm. Figs 10—12, sections of outer vascular band in active growth to show nature of meristematic activity. Scale bars = 20 tm. Fig. 10. Vascular cambium (large arrows left and right) actively producing fascicular xylem and phloem (right, centre) and two multiseriate rays (the large ray indicated by a vertical arrow; periclinal divisions in secondary cortex indicated by small arrows. Fig. 11. Vascular cambium (large arrows left and right) actively producing fascicular xylem (left and right) and a multiseriate ray (centre, vertical arrow); numerous periclinal divisions, constituting a lateral meristem, occur between the tips of the two smaller horizontal arrows- Fig. 12. Vascular cambium (large arrows left and right) is producing secondary xylem and secondary phloem; a vascular ray is indicated by vertical arrow, below; numerous periclinal divisions in the secondary cortex, constituting a lateral merstem, are present in the secondary cortex (small honzoatal arrows).

LLSlflb’12:IVD 8 01 WOOD AND STEM ANATOMY OF AMARANTHACEAE 11 ify as diffuse, as in Stegnosperina (Carlquist, 1999b). CONCLUSIONS In some instances, lateral merist.em activity ceases, as in the mature plant of Amaran thus caudatus (Fig. 1); MERISTEMATIC PROCESSES, RAYS AND RAYLESSNESS the maturity of this plant was indicated by the fact I interpret the material studied here as follows: after that all branches had terminated in inflorescences, the first cylinder of stem bundles, which constitute and no active terminal or lateral buds remained. Lack primary vascular tissue, has been formed, a lateral of divisions in the lateral meristem zone of some stems meristem forms in the inner cortex. Divisions in this (Fig. 19, top) may indicate periods of quiescence that lateral meristem are often not confined to a single alternate with periods of activity (Fig. 20). Offsets in layer, and thus are termed diffuse. The lateral mer secondary xylem accumulation (Figs 27. 28) are prob istem produces radial files of secondary cortex to the ably the result of greater activity in some lateral mer outside and radial files of conjunctive tissue to the istem segments or possibly differential rates of inside; meristematic cells that do not mature into sec division in vascular cambium segments. The lateral ondary cortex or conjunctive tissue become vascular meristem produces radial rows of cells internally cambia. A vascular cambium, as elsewhere in dicoty (Fig. 20); these become conjunctive tissue. To the out ledons, is a single layer of cells that produces second side of the meristem, more secondary cortical cells are ary phloem (both axial phloem and phloem rays) to the produced in radial files. Vascular cambia are also outside and secondary xylem (both axial xylem and derived from the lateral meristem; vascular cambia xylem rays) to the inside. This can be shown clearly are the residual meristematic cells that do not mature (Fig. 4). If rays are not interpreted as products of a into secondary cortex or conjunctive tissue. A vascular vascular cambium, one must say that they are radial cambium tFig. 20, bottom) produces xylem and phloem plates of conjunctive tissue that interconnect with con for a finite period of time, but usually remains active centric bands of conjunctive tissue, a concept difficult while a new vascular cambium is formed outside of the to imagine ontogenetically. This concept, however, is vascular cambium and its products. mentioned as a possibility in Amaranthaceae by Met calfe & Chalk (1950) and Rajput (2002). Metcalfe & Crystals Chalk (1950) state that vascular rays are present in Crystal sand was observed in Amaranthus caudatus some Amaranthaceae, but that other species are ray- (inner cortex, conjunctive tissue, rays), Bosea yerva less but possess radial plates of conjunctive tissue. If mont cortex, pith), Ce! osia floribunda (cortex, Fig. 5; envisaged three-dimensionally, the designation of pith) and Nototrichiurn sanduicense (cortex, conjunc radial parenchyma sheets as conjunctive tissue tive tissue, pith). Druses were observed in Arthraerua becomes unlikely, even if theoretically possible. In tan leubnitziae (gigantic druses in conjunctive tissue. gential sections of these species, one sees fusiform vas Fig. 5; also in multiseriate ray&, Charpentiera dean cular rays, some long and some short (Figs 2,6, 13, 14, flora (cortex, conjunctive tissue, pith) and IVototrich 22, 25, 29). The histology of these rays is similar to turn sandwicense (cortex, conjunctive tissue, pith). that of the concentric bands of conjunctive tissue (ver Crystal-containing cells occur in scattered fashion and tically elongate cells, storied in some Amaranthaceae: are larger than neighbouring cells; therefore, they see Figs 2, 29). The similarity in histology of vascular may be termed idioblastic. rays and of conjunctive tissue is not an indication of identity in origin. Rather, this similarity probably Starch results from the fact that both vascular cambia and Starch grains were observed in parenchyma cells of conjunctive tissue are derived from a lateral mer three species: Amaranthus caudatus (cortex, rays, istem. If both ray parenchyma and conjunctive tissue conjunctive tissue, pith), Bosea yervainora (interfas parenchytna are ultimately products of lateral mer cicular rays and rays that originate within fascicular istems, one would expect that their cell sizes and even areas) and Charpentiera densiflora (cortex, conjunc wall thicknesses (both are parenchyma) would be sim tive tisue, rays, and pith). ilar. Moreover, one would be pressed to explain how

4 Figures 13—18. Sections of Celosia floribunda. Figs 13—is, portions of tangential sections. Fig. 13. Tall vascular rays alternate with typical axial secondary xylem. Scale bar = .50 m. Fig. 14. Multiseriate ray (centre, top to bottom) consists mostly of upright cells; numerous narrow vessels to right of the ray, Scale bar 20 jim. Fig. 15 Lateral wall pitting in a narrow vessel consists of circular bordered pits with slit-like apertures. Scale bar = 10 jim. Figs 16, 17, SEM photographs of perforation plates from tangential section. Scale bars 10 jim. Fig. 16. A non-bordered perforation plate; no border is evident where the two vessel elements join. Fig. 17. A minimally bordered perforation plate, with a small groove-like border between the plates of the two vessel elements. Fig. 18. SEM photograph of a cortical cell from a transection of a stem, to show an idioblastic cell containing tetrahedral crystal sand crystals. Scale bar = 5 jim.

12 S. CARLQUIST

v nva isae rrac nz...-ew asj t. S K — ia

fl I

I I

319 20

‘ ‘4WLi; 4AT1 a:- nrr i u 4 !4!C U

C 2003 The Linnean Society of London, Botanical Journal of the Linnean Society, 2003, 143, 1—19 WOOD AND STEM ANATOMY OF AMARANTHACEAE 13 primary rays can be present in a primary stem, but There is no known reason why the change in histol why one should then be forced to designate all subse ogy occurs. One could speculate that lignified walls on quent rays identical in appearance to the primary rays ray cells and presence of fibres in potential ray areas as radial bands of conjunctive tissue. offer enhanced mechanical strength (Charpentiera is Rajput (2002) claimed that only three of the nine a genus of small trees), but that later in the develop genera of Amaranthaceae he studied have rays. ment of these potential ray cells, thin-walled paren whereas all of the species in the present study have chyma cells (thus ‘true’ ray cells) are added, perhaps rays according to my definition, which corresponds to to enhance starch- or even water-storage capacity the definition for vascular rays ‘xylem rays and These features are visible most clearly in C. chip/ice, phloem rays produced by a vascular cambium) by the but may also be seen in C. densiflora (Fig. 21). In IAWA Committee on Nomenclature (1964) and innu Arthraerua (Fig. 5), there is a sudden change in merable textbooks. I believe that most workers would potential ray areas from fibres to thin-walled ray agree that rays are radially orientated plates of paren cells. In all of these instances, the zones of thin- chyma in secondary xylem (and secondary phloem), walled ray cells must be termed multiseriate rays. derived from vascular cambia and fusiform in shape There is no sharp demarcation between this ray as seen in tangential sections of stems and roots. A parenchyma and the conjunctive tissue parenchyma tangential section of Nothosuerna brachiata Wight in outside the ray, but the very slow addition of ray cells Rajput (2002: 3D) very clearly shows thin-walled to the parenchyma makes identification of cambial upright ray cells as well as libriform fibres, and at activity and delimitation of ray from conjunctive tis least the beginnings of rays are evident in the transec sue difficult in most cases when the tissues are tion of Gomphrena globosa L. (Rajput, 2002: 3A). It mature (for an instance of clear delimitation, see would have been useful to see more extensive areas of Fig. 4). Within the zones of fibres in vascular bands of sections in these and other species that Rajput (2002) Arthraerucz, howevei-, occasional uniseriate rays do and Rajput & Rao (1999, 2000) report as rayless, occur (Fig. 6), so at no point does Arthraerua qualify although the secondary xylem may well be predomi as rayless. The histological conditions of rays, poten nantly rayless. A single tangential section at relatively tial ray areas, and cells transitional between fibres high magnification does not suffice to demonstrate the and ray cells in Charpentiera pose interesting cases intricacies of this situation to the reader. because one must concede that the presence or Chnrpentiera elliptica does show zones of fibres in absence of rays cannot always be easily defined if potential ray areas (these fibre zones are free from cells intermediate between typical ray parenchyma vessels: in rayless dicotyledons with a single cam and fibres occur. Similar problems in ray histology, bium, vessels are present randomly). These potential leading one to question whether rays occur, can be ray areas are changed markedly in aspect as the vas cited in Lee/orEs: compare the smaller rayless stems cular cambium adds thin-walled parenchyma to such in Carlquist (1964) with the clearly ray-bearing older potential ray areas, and thereby easily defined wide stems studied in Carlquist (1990). Likewise, Misoden multiseriate rays result (Figs 23, 24). The appearance draceae offer some curious situations with regard to of these transections at first glance is unusual, and ray definition, presence and characteristics (Carl seems to show flanges of secondary xylem, each quist, 1985a). Such instances of histological interme tipped by a strand of phloem, extending into paren diacy between ray cells and fascicular xylem cells are chyma and separated from each other laterally by uncommon. Xylem and phloem rays in species with parenchyma; however, the explanation given above is successive cambia may contain relatively few cells the one that can be demonstrated in terms of ontoge and may feature some radial elongation of cells, but if netic sequences. In the adaxial portions of these rays, one can demonstrate origin of what appears to be a there are ray cells with lignified walls and strands or xylem ray or a phloeni ray from a vascular cambium, bands, variously distributed, of fibres (Figs 23, 24). one surely is justified in using the term secondary ray

Figures 19—22. Stem sections of Charpentiera densiflora. Fig. 19. Transection near outside of stem to show tangentially stretched cortical cells at extreme top, a vascular band in active growth, and, lower half of photograph. a mature vascular band portion of another at bottom); the two arrows denote the boundary between latewood (narrow vessels) and earlvwood (wider vessels) midway in the vascular band. Scale bar = 50 pm. Fig. 20. Section from outer stem to show primary cortex, above, delimited from secondary cortex (which has radial rows of cells) by strands of fibres; a lateral meristem is actively dividing, yielding a vascular cambium, centre; xylem and phloem of a mature vascular band at bottom. Scale bar = 20 pm. Fig. 21. Transection of mature vascular bands; fibres occur in earlier portions of potential ray areas of vascular bands, but are followed by thin-walled cells of multiseriate rays (for more extreme examples. see Figs 23, 24). Scale bar = 50 tim. Fig. 22. Tangential section of vascular band, showing typical vascular rays. Scale bar = 50 pm.

Pt1tj, V

ISInIUvD S PT WOOD AND STEM ANATOMY OF AMARANTHACEAE 15

(see Fig. 4). The slow pace of divisions in ray areas of HABIT vascular cambia in species with successive cambia complicates making this determination. Regardless, Bosea is distinctive among Amaranthaceae in being a this is a concern than cannot be resolved here, semiscandent shrub. It can grow as an independent because examples from as many families and genera large shrub or small tree when individuals occur iso with successive cambia as possible must be exam lated from other vegetation, but when Bosca shrubs ined, especially in ontogenetic terms. These compari grow close to larger shrubs or small trees, some sons will be offered in a future review of the upright stems of a Bosea shrub tend to extend upward, phenomenon of successive cambia and how the xylem leaning on branches of the neighbouring tree, and and phloem produced by successive cambia differ bending or branching of these stems occurs when they from interxylary phloem produced by a single reach canopy levels. Wood of Bosea reflects this habit cambium. in two respects. Libriform fibres are notably elongate Variations in successive cambia and their products (the F/V ratio of 7.1 is unusually high for a dicotyle in Caryophyllales will be reviewed in a paper summa don). The greater length (and relatively great walL rizing wood and stem anatomy in this order. The thickness) of libriform fibres in Basea may be related present paper is a contribution to this planned syn to the elongate nature of major branches and the thesis. Differences and discrepancies among workers necessity that such branches be self-supporting until with respect to the description and interpretation of they encounter branches of other shrubs or trees on successive cambia, their origin and products, are which they can lean. The wide multiseriate rays and numerous. Reasons for this diversity appear to lie wide bands of conjunctive tissue, composed of thin- mainly in that most workers have been able to exam- walled parenchyma. suggest flexibility in stems of iDe only a small number of species, and have not been Bosea; this feature is often found in lianas (Cariquist, able to review, using preparations with histological 1985b). detail and good sequences of maturation of derivatives The wood of annuals, still little investigated, offers from meristematic cells. sections of dicotyiedons that features correlated with the short life cycle, relatively possess successive cambia. Pfeiffer (1926) has recently small plant size and probably also the fact that mois surveyed such a wide range of material, but lacked ture near the soil surface must suffice for growth of certain good microtechnical preparations; lack of fine annuals. In the mature stem and root of Amaranthus sections may account for some of the variation in caudatus, vessel diameter and abundance decrease as description and interpretations seen in various papers the plant matures. These features may be genetically (also, drawings instead of photographs are often used controlled because the cultivated plants studied were to illustrate tissues with thin walls in meristematic watered steadily until they were harvested. The pro areas in families with successive cambia). Instances in gressive decrease in vessel diameter in this species is which numerous vascular bands are scattered appar like that seen in a single growth ring of an ordinary ently randomly throughout a conjunctive tissue back woody dicotyledon. ground (e.g. at least some species of Pisonia) have Also of interest in the stem and root of Ainaranthu.s proved particularly confusing to some workers. Stu caudatus is the tendency for vessels to be not only dents of comparative wood anatomy seek definitions smaller but sparser toward the outside of the stem. based on mature structure, definitions in which onto- The vascular bands are also closer together nearer to genetic studies are not required. In my experience, the stem surface. These tendencies may relate to mature structure reflects ontogeny readily, and one acquisition of greater mechanical strength as the can therefore rely on mature structure when applying upper portions of the plant, with their relatively definitions. weighty inflorescences, form. Increase in mechanical

Figures 23—26. Stem sections of Charpentiera elliptica. Figs 23, 24, portions of vascular bands from a transection to show nature of fascicular hand. Scale bar 50 urn. Fig 23. Ray, centre, begins with parenchyma cells with lignified walls (above vertical arrow) and to the left of them, fibres; abaxial to these cells are thin-walled ray cells; note sclerenchyma-sheathed phloem. Fig. 24. Potential ray areas that begin with fibres inght half of photograph); at left is a ray that begins with parenchyma cells with lignified walls, among which a strand of fibres is present (above tip of vertical arrow). Fig. 25. Tangential section with vascular rays; rays in left half of photograph are composed of short fibre-like cells; ray in right half of photograph is wider and has more typical ray cells, with lignified walls in centre of ra3ç but with non-lignified walls that stain poorly at top and bottom. Scale bar = 50 tim. Fig. 26. Tangential section portion to show variations in ray cells; in ray at left, cells are like typical upright ray cells (some with non- lignified walls, others with lignified walls) aLthough at margin of ray (facing the fibres that sheath the vessel, centre) are more fibre-like ray cells: in ray at right, cells are variously fibre-like, some very short and subdivided transversely. Scale bar = 20 Mm.

JSIflYIflVD 9 9T WOOD AND STEM ANATOMY OF AMARANTHACEAE 17

strength of roots as growth proceeds is of potential large shrub of dry areas, with relatively little climbing value for the same reasons. There is slightly more ray activity It is a shrub with only incipent lianoid ten and conjunctive tissue parenchyma in roots than in dencies. stems of Amaranthus caudatus. This would correlate Number of vessels per group is a feature correlated with the tendency for roots to store photosynthates, with xeromorphy in woods with libriform fibres (Car and possibly with the fact that roots are not upright lquist, 1984). Grouping of vessels confers safety self-supporting structures. because occurrence of embolisms in one vessel of a group does not disable the pathway to which it belongs. The highest mean numbers of vessels per ECOLOGY group in the woods surveyed here (Table 1, column 1) The Mesomorphy Ratio vaLues for the various species occur in the Namibian desert sand—dune shrub (Table 1, column 9) are significantly variable. If one Arthraerua leubnitziae, a small dryland tree from were to discount the ‘dilution’ of woody tissues by the southern Baja California, Mexico, Cetosia floribunda, presence of conjunctive tissue parenchyma. the figures and a dryland shrub from lower elevations in the for these ratios would be somewhat larger. However, Hawaiian Islands, Nototrichiurn sandu’icense. The inclusion of conjunctive tissue in the vessel density fig lowest mean numbers of vessels per group were reg ures, like inclusion of rays when reporting number of istered in the Hawaiian rain forest tree Charpentiera; vessels mm2, is defensible in that it more accurately the two species of this genus also have notably low reflects the conductive area per unit transectional numbers of vessels mm2 (Table 1, column 3), and thus area. As mentioned earlier, the mean conductive area these species qualify as mesomorphic in wood fea for the Amaranthaceae studied here is virtually the tures. same as the conductive area mm2 of transection for a sampling of species with successive cambia (Cariquist, 1975). The lowest figure for the Mesomorphy Ratio in S YSTE MAT IC S the species studied occurs in Arthraerurz leubnitziue. Although the number of species of Amaranthaceae for Such a low figure is appropriate for this species which wood and stem anatomy have been studied is because it is a desert shrub of very dry sand dunes. appreciable, and the papers dealing with these aspects The Mesomorphy figure for this species, 17, is close to of the family do not overlap in the species they study, the mean for the ratio in southern Californian desert the sampling for this family still is relatively small; shrubs, 20.9 (Carlquist & Hoekman, 1985). Note that there are about 850 species in 65 genera of Amaran the area of Namibia in which Arthraerua leubnitziae is thaceae s.s. (Cronquist & Thorne, 1992). Although native received virtually no rain, only nocturnal con there are some highly distinctive character states densation. Arthraerua leubnitziae may receive some reported in the anatomy of those species studied, the water from condensation on nearby roads, but even if taxonomic and systematic distribution of these cannot it does not, the value of narrow, short vessel elements be assessed at present. for prevention of embolism formation is just as great One feature seems probably diagnostic of Caryo as it is in plants of desert areas where there is a short phyllales s.s. and even Caryophyllales si.: presence of rainy season but in which soil moisture is negligible non-bordered or minimally bordered perforation during the dry season. The figures for the two species plates in vessels. Although a few families have not of Charpentiera (265 and 581, respectively) are similar been studied with i-espect to this feature, the list of to figures for southern Californian trees (486) or suc caryophyllalean families, especially in core Caryo culents (368). The conjunctive tissue in Charpentiera phyllales’, with non-bordered perforation plates con can be interpreted as a form of succulence. The rela tinues to grow (Carlquist, 1998a. b, 1999a, b, c, 2003). tively low Mesomorphy Ratio for Boseayervamora (80) Non-bordered perforation plates are uncommon in does not match the figures for southern Californian dicotyledons in general (unpubl. data), although this vines (441), but Bosea is not really a vine, but rather a feature has been neglected by most workers in wood

Figures 27—30. Stem sections of Nototrichium sondu’icense. Fig. 27. Transection typical of vascular bands in stem; bands of conjunctive tisue (relatively thin-walled) present, and ray tissue is limited. Scale bar 50 jam. Fig. 28. Portion of transection to show offset’ in vascular band: secondary phloem is further out in upper left of photograph, further in at right, where cells of conjunctive tissue are outside the secondary phloem. Scale bar = 20 jam. Fig. 29. Tangential section: typical vascular ray composed of upright cells at left, but also vascular rays in which the section shows a band of conjunctive tissue interconnection with ray tissue at right (especially lower right). Scale bar = 50 tm. Fig. 30. Pith aad first several vascular bands from stem transection: pith vascular bundle at lower left: relatively little conjunctive tissue present; and rays that appear composed of fibre-like cells present, as in the area above tip of vertical arrow. Scale bar 50 jam.

© 200aThe Linnean Society of Loadon, Botanical Journal of the Linneon Society, 2003, 143, 1—19 18 S. CARLQUIST anatomy and its systematic distribution remains to be eral ineristeins and successive cambial activity. 1/tWA Jour fully explored. nal 2th 149—163. The significance of successive cambia in Caryophyl Carlquist S. 1999c. Wood anatomy of Agdestis (Caryophyll lales is also of potential systematic significance. There ales): systematic position and nature of the successive cam is a tacit assumption in the literature that this feature hia. AIIm 18: 35—43. is an apomorph in dicotyledons in general (Metcalfe Cariquist S. 2001. Comparative wood anatomy. Heidelberg: & Chalk. 195O, and although this may be true, it is Springer Verlag. possible that it is a symplesiomorphy within Caryo Carlquist S. 2003. Wood anatomy of Polygonaceae: familial phyllales as. or even sd. Such an interpretation will be relationships; analysis of a family with exceptional wood diversity. Botanical Journal certified when a phylogenetic tree with strongly sup of the Linnean Society 141: 25— 51. ported clades becomes available, and one can then Cariquist 5, Hoekmau DA. 1985. Ecological wood anatomy compare the distribution of successive cambia to the of the woody southern Californian flora. IAWA Journal 6: branching of such a tree. 319—348. Costea M, DeMason DA. 2001. Stem morphology and anat

omy in An1aranthus L. (Amaranthaceae) — taxonomic signif ACKNOWLEDGEMENTS icance. Journal of the Torrey Botanical Society 128: 254— The kindness of Dr David Lorence in providing mate 281. rial of Charpentiera densiflora and Nototrichium Cronquist A, Thorne HF. 1992. Nomenclatural and taxo sandwicense is greatly appreciated. Travel to Namibia nomic history. In: Behnke H-D, Mabry TJ, eds. Cars-vphyl was made possible by a grant from the American lales. Evolution and systematics. Heidelberg: Springer Philosophical Society. Verlag, 5—25. Downie SR, Katz-Downie DS, Cho K-J. 1997. Relationships in the Carvophyllnles as suggested by phylogenetic analyses REFERENCES of partial chloroplnst DNA 0RF2280 homolog sequences. American Journal of Botany 84: 253—273. BorschT, Clemants 5, Mosyakin S. 2001. Symposium: biol Downie SR, Palmer J. 1994. A chloroplast DNA phylogeny of ogy of the Amaranthaceae—Chenopodiaceae al]iance. Jour the Caryophyllales based on structural and inverted repeat nal of the Torrey Botanical Society 128: 234—235. restriction site variation. Systematic Botany 19: 236—252. Cariquist S. 1964. Morphology and relationships of Lactori Eichler AW. 1876. Syllabus der Vorlesungen uber Phaneroga daceae. Aliso 5: 421—435. menkunde. Kiel: Schwerische Buchhandlung. Cariquist S. 1975. Ecological strategies of xylem evolution. Engel T, Barthlott W. 1988. Micro:norphology of epicuticuLar Berkeley: University of California Press. waxes in centrosperms. Plant Systematics and Evolution Cariquist S. 1982. The use of ethylene diamine in softening 161: 71—85. hard plant structures for paraffin sectioning. Stain Technol Goldblatt P, Manning J. 2000. Cape plants. A conspectus of ogy 57: 311—317. the Cope Flora of South Africa. Pretoria: National Botanical Cariquist S. 1984. Vessel grouping in dicoty]edons woods: sig Institute and St Louis: Missouri Botanical Garden. nificance and relationship to imperforate trachenry ele Gregory M. 1994. Bibliography of the systematic wood anat ments. Aliso 10: 505—525 omy of dicotyledons. JAWA Journal Supplement 1: 1—265. Cariquist S. 1985a. Wood and stem anatomy of Misoden IAWA Committee on Nomenclature. 1964. Multilingual draceae: systematic and ecological conclusions. Brittonia 37: glossary of terms used in wood anatomy. Winterthur: Ver 58—75. lagsbuchanstalt Konkordia. Carlquist S. 1985b. Observations on functiona] wood histol Joshi AC. 1937. Some sa]ient points in the evolution of the ogy of vines and lianas: vessel dimorphism, tracheids, vasi secondary vascular cylinder of Amnranthaceae and Chenop centric tracheids, narrow vessels, and parenchyma. Aliso 11: odiaceae. American Journal of Botany 24: 3—9. 139—15 7. Metcalfe CR, Chalk L. 1950. Anatomy of the dicotyledvns. Cariquist S. 1990. Wood anatomy and relationships of Lac Oxford: Clarendon Press. toridaceae. American Journal of Botany 11: 1498—1505. Pfeiffer H. 1926. Das abnorme Dickenwachstum. Handbuch Cariquist S. 1998a. Wood and stem anatomy of Petiveria and der Pflanzenanatomie, Vol. 9 (21. Berlin: Gebruder Borntrae Rivina. IAWA Journal L9: 383—391. ger, 1—272. Carlquist S. 1998b. Wood anatomy of Portulacaceae and Hec Rajput KS. 2002. Stem anatomy of Amaranthaceae: rayless torellaceae: ecological, habital, and systematic implications. nature of xylem. Flora 197: 224—232. Aiiso 16: 137—153. Rajput KS, Rao KS. 1999. Structural and developmental Carlquist S. 1999a. Wood anatomy; stem anatomy; and cam studies on eambial variant in Pupalia lappacea Amaran bial activity of Barbeuia (Caryophyllales). JAWA journal 20: thaceae). Annales Botanicae Fennicoe 36: 137—141. 431—440. Rajput KS, Rao KS. 2000. Secondary growth in the stem of Cariquist S. 1999b. Wood and stem armtomy of Stegnosperma some species ofAlternanthero and .kchyranthes aspera Ama Caryophyllales): phylogenetic relationships; nature of lat ranthaceael. L4W4 -Journal 21: 417—424.

Rettig .114, Wilson HD, Manhart JR. 1992. Phylogeny of Solereder H. 1908. Systematic anatomy of the dicotyledons

the Caryophyllales — gene sequence data. Thxon 41: 201— (trans. Boodle & Fritach), 2 vols. Oxford: Clarendon Press. 209. Soltis DE, Soltis PS, Chase MW, Mon ME, Albach DC, Rodman JE, Oliver MK. Nakamura K, McClammer JU, Zanis M, Savolainen V, Hahn WH, Hoot SB, Fay MF, Bledsoe All. 1984. A taxonomic analysis and classification Axtell M, Swensen SM, Prince EM, Kress WJ, Nixon of Centrospermae. Systematic Botany 9: 297—323. 14C, Farris JS. 2000. Angiosperm phylogeny inferred from Schenck H. 1893. Beitrage zur Biologic und Anatomic der 188 rDNA, rbcL, and atpB sequences. Botanical Journal of Lianen. II. Schimpers Botanischen Mittheilen der Tropischen the Linnean Society 133: 381—461. 5: 1—271. Williams SE, Albert VA, Chase MW. 1994. Relationships of Schweingruber FH. 1990. Anatomic europaischer Ebizer. Droseraceae: a cladistic analysis of rbcL sequence and mor Anatomy of European Woods. Stuttgart: Verlag Paul Haupt. phological data. American Journal of Botany 81: 1027—1037.