Wood anatomy of Begoniaceae, with comments on raylessness, paedomorphosis, relationships, vessel

diameter, and ecology

Sherwin Carlquist

B U L L E T I N O F T H E T O R R E V B O T A N 1 C A L C L U B

Vol.. 112. No. 1, pp. fi'l-69 JAM AKV-M SKI H. I<>85

B U L L E T I N O F T H E T O R R E V B O T A N I C A L C L U B Vol . 1 12. NO. 1. pp. yi-(>9 JAM AKV-MAKCH. 19H'>

Wood anatomy of Begoniaceae, with comments on raylessness, paedomorphosis, relationships, vessel

diameter, and ecology

Sherwin Carlquist

CARLQUIST, S. (Rancho Santa Ana Botanic Garden and Department of Biology. Pomona College, Claremont, CA 91711). Wood anatomy of Begoniaceae. with comments on raylessness. paedomorphosis, relationships, vessel diameter, and ecology. Bull. Torrey Boi. Club 112: 59—69. 1985.—Six collections of Tour species of Begonia ranging from moderately' to markedly woody (for the genus) are analyzed for qualitative and quantitative wood features. Begonia wood has vessels mostly solitary and oval or circular in outline, scalariform lateral wall pitting on vessels, tyloses in vessels, libriform fibers which are sometimes septate, vasicentric scanty axial parenchyma, tall and wide multiseriate rays (uniseriates absent) composed of erect cells (near-rayless in one collection), and storied structure. These features match those of Datiscaceae closely: features which do not may be related to the distinctive habit of woody Begoniaceae: semisucculent canelike stems of finite duration produced by a rhizome. As with similar stems of Piperaceae (which is not closely related to Begoniaceae), rays have erect tells predominantly or exclusively, rays alter little during ontogeny, and vessels have scalariform lateral wall pitting almost exclusively. These three features seem related to paedomorphosis, as does the presence of occasional scalariform perforation plates in early secondary xylem of Begonia. Quantitative features of Begonia wood mark all species as mesomorphic. Wood mesomorphy of B. pannflora increases with age of stem, suggesting greater transpiration from taller stems. Increase in vessel diameter with age is common in mam dicotyledon trees and shrubs, and may relate to ability of progressively larger root systems to tap greater water supplies, as well as to the tendency for a larger and more exposed crown to transpire more. Some annual ami short-lived perennial dicotyledons tend to have a progressive decrease in vessel diameter with age, which can be correlated with progressive drying of soil in the habitats of these species.

Key words: Begoniaceae, wood anatomy, paedomorphosis, juveniliMii. vessel diameter, eco­logical wood anatomy, Datiscaceae, Violates.

A study by Hildebrand (1859) provides most of what has been reported on wood anatomy of Begoniaceae. The accounts of Solereder (1908), Irmscher (1925), and Metcalfe and Chalk (1950) derive in large measure from Hildebrand's essay- A new investigation of wood anatomy of Begon­iaceae is offered at this time because liquid-preserved material of six collections of four species furnishes a more abundant rep­resentation of secondary xylem than has hitherto been available. Two of the collec­tions were made in the field in Peru in 1982 (Begoniaparviflora Poepp. & Endl. and B. peruviana A. DC). The material of two other Peruvian species (B. duchartrei Andre and B. maculata Raddi) was derived from cultivated specimens. These species have a range in degree of woodiness: stems of B. parviflora (Carlquist 7050) were as much as 9 cm in diameter, with the woody cylinder

Received for publication September 7, 1984 and in revised form February 14, 1985.

59

itself up to 2 cm in thickness. The woody cylinder oi B. peruviana (most of the thick­ness of which is shown in Fig. 1) was much less, although stems in that species exceeded 1 m in height.

The various degrees of woodiness in the Begonia species studied permit exami­nation of the wood with respect to juve-nilism. The tendency of Begonia wood to demonstrate aspects of paedomorphosis (termed neoteny by some authors) has been noted elsewhere (Carlquist 1962), and an examination of this phenomenon based on more extensive material oi Begonia has been needed. Begonia wood also casts light on the phenomenon of raylessness.

A related phenomenon worth study in terms of wood anatomy is the growth form of woody Begonia species: little-branched canelike stems, semisucculent in texture, branched from a rhizome system. In all of the species studied, a feature like that of woody Piperaceae or some Berberidaceae (Nandina) is evident: the stems have limited

© 1985 Torrey Botanical Club

60 1U 111 TIN <T Ti l l IOKKM liOTANICALCI.LB [VOL. 112

duration, and thus secondary xylem is never extensive. The stems of B. parviflura, al­though commonly 5 m tall in mature plants, are not truly self-supporting and will bend if supporting vegetation is removed.

Wood anatomy provides a very good tool for examination of relationships of Begoniaceae. The affinity of Begoniaceae is not controversial: Cronquist (1981), Dahl-gren (1980) and Thorne (1983) place Be­goniaceae in Violales near Datiscaceae. Takhtajan (1980) places Begoniaceae in Begoniales along with Datiscaceae, but con­cedes that Begoniales is close to and derived from Violales. Thorne pairs Begoniaceae with Datiscaceae in a suborder, Begoniinae. The idea that Begoniaceae is related to Datiscaceae can be traced to Lindley (1846). A study by Davidson (1976) on wood and phloem anatomy of Datiscaceae provides an ideal basis for comparison with the results of the present study in assessing the merits of this alleged relationship.

Begoniaceae are characteristically plants of moist understory sites. Thereby, this family is an ideal one to use in assessing whether quantitative criteria for wood me-somorphy are widely applicable.

Materials and Methods. All samples were preserved in liquid. Samples of cul­tivated materials were fixed directly in for­malin-acetic acid-alcohol. The stem por­tions of Begonia species collected in Peru were placed in plastic bags to which par­aformaldehyde powder was added. After these specimens were transported to Clare-mont, they were transferred to 50% ethyl alcohol.

Woods of Begonia can be sectioned on a sliding microtome, but results are often poor when vessels are wide, ray cells or fibers are thin-walled, or the woody cylinder is thin. These problems were obviated by the use of a recently-devised softening method (Carlquist 1982). Sections were stained with safranin; most sections were also counterstained with fast green. Ma­cerations were prepared by means of Jef­frey's fluid and stained with safranin. Quan­titative data are based upon 25 measurements per feature, except for wall thickness of various cell types and except for diameter of imperforate tracheary elements; in these, a typical condition was selected. Ray width

measurements were intended to represent the widest point of rays. However, rays in Begonia are very tall, so that there is no way of discerning whether the widest por­tion of a ray is included within any partic­ular section or not. Therefore, the means given for this feature in the descriptions below are undoubtedly too low.

Specimens documenting the wood sam­ples have been deposited in the herbarium of the Rancho Santa Ana Botanic Garden. T h e plant designated as Begonia corrinea Hook. (Carlquist 1962) has been re-iden­tified as B. maculata. Thanks are due Dr. Bernice G. Schubert, who determined my specimens from Peru.

Results and Discussion. ANATOMICAL DESCRIPTIONS.

Begonia duchartrei (Carlquist 15805). Growth rings absent. Vessels circular or oval in transectional outline. Mean number of vessels per group, 1.13. Mean vessel diameter, 48 u,m. Mean number of vessels per mm2, 34. Mean vessel wall thickness. 1.2 u,m. Mean vessel-element length, 283 u,m. Vessels with simple perforation plates (a few scalariform plates seen in metaxylem and early secondary xylem). Lateral wall pitting of vessels scalariform; vessel-par­enchyma pits with much wider apertures than vessel-vessel pits. Libriform fibers nonseptate. Mean libriform fiber diameter at widest point. 18 u,m. Mean libriform fiber wall thickness, 1.7 u,m. Mean length of libriform fibers, 403 u.m. Axial paren­chyma is of the vasicentric scanty type, an incomplete sheath of cells 1—2 layers thick around vessels or vessel groups. Paren­chyma in strands of four cells typically. Rays multiseriate only. Rays exceed 5 mm in mean height (most rays not included in length of a section). Mean ray width, 33 cells or 850 |xm. Rays consist of erect cells (vertically more elongate cells at margins of rays). Mean ray cell wall thickness, 0.7 |xm. Ray cell walls with simple pits; starch observed in rays. Storied structure observed in most libriform fibers; some ray cells storied. Droplets of unidentified com­pounds occasional in parenchyma cells.

Begonia maculata (Carlquist, s.n.). Growth rings absent. Vessels round or oval in transectional outline. Mean number of

1985] CARLQUIST: WCXJD ANATOMY OF BEGOMAC.t.AK t i l

Figs. 1-6. Wood sections oi Begonia peruviana (Carlquist 7131). Fig. 1. Transection, cambial side at left, ray area above. Fig. 2. Tangential section; fascicular area to left of vessel, ray area to right. Near-ravless condition evident. Fig. 3. Fiberlike ray cells from radial section. Fig. 4. Tyloses in vessel from radial section. Fig. 5. Tangential section portion showing vessel-axial parenchyma pits. Fig. 6. Vessel-vessel pitting from tangential section. Figs. 1-2, magnification scale above Fig. 1 (finest divisions = 10 jjtm). Figs. 3-4, scale above Fig. 3 (divisions = 10 y.m). Figs. 5-6, scale above Fig. 5 (divisions = 10 y.m).

62 BULLETIN OF THE TORRF.Y BOTANICAL CLUB [VOL. 112

vessels per group. 1.40. Mean vessel di­ameter, 55 jim. Mean number or vessels per mm2, 52. Mean vessel wall thickness, 2.0 |xm. Mean vessel element length, 411 fj.m. Perforation plates simple, a few sca­lariform plates in metaxylem and in early secondary xylem. Lateral wall pitting of vessels scalariform, apertures wider on ves­sel-parenchyma pits than on vessel-vessel pits. Tyloses present, thin-walled. Imper­forate tracheary elements are mainly li-briform fibers with simple pits (Fig. 14); a few vestigial borders were observed on pits of some imperforate tracheary elements. Libriform fibers and fiber-tracheids once or twice septate and with numerous starch grains (Fig. 17). Mean diameter of imper­forate tracheary elements at widest point. 25 jim; mean wall thickness, 3.2 u,m; mean length, 645 (xm. Axial parenchyma is of the vascientric scanty type, an incomplete sheath usually one cell thick around vessels and vessel groups. Axial parenchyma typ­ically in strands of four cells. Rays multi-seriate only. Mean ray height exceeds 5 mm. Mean ray width 12 cells or 300 u.m. Mean ray cell wall thickness, 2.2 |im. Ray cells predominantly erect, but a few procumbent cells present in central portions of mulii-seriate rays (Fig. 16). Starch grains present in ray cells. Ray cell walls lignified, bordered pits on tangential walls of some cells (Fig. 15), simple pits otherwise present. Storied structure is apparent in some imperforate tracheary elements. Droplets of unidenti­fied compounds occasional in parenchvma cells.

Begonia parviflora (Carlquist 7050, pe­riphery of woody cylinder). Growth rings absent (Fig. 7). Vessels round or oval (ra­dially widened) in transectional outline (Fig. 12). Mean number of vessels per group. 1.37. Mean vessel diameter. 84 u,m. Mean number of vessels per mm2. 27. Mean vessel wall thickness, 2.3 |xm. Mean vessel element length, 405 u,m. Perforation plates simple. Lateral wall pitting scalariform, vessel-pa­renchyma pits with wider apertures than those on vessel-vessel pits (Fig. 10, 11). Thin-walled tyloses present, abundant (Fig. 13). Mean libriform fiber diameter at widest point, 28 \xm. Mean wall thickness of lib­riform fibers, 2.3 |xm. Mean length of libriform fibers, 630 |xm. Libriform fibers

nonseptate, pits 1.5-2 jxm in diameter (Fig. 10, right). Axial parenchyma is of the vas-icentric scanty type (Fig. 12), forming in­complete sheaths one or two cells thick around vessels or vessel groups. Axial pa­renchyma typically composed of strands of four cells. Rays multiseriate only. Mean height of multiseriate rays exceeds 5 mm. Mean multiseriate ray width 17 cells or 322 u.m. Ray cells markedly erect (Fig. 8, 9). Mean ray cell wall thickness, 1.2 u,m. Ray cell walls thin but lignified. with simple pits. Ray cells somewhat fusiform in shape as seen in tangential section (Fig. 13) but not radial section (Fig. 9). Storied structure present in fibers (Fig. 8, 13). Large droplets of an unidentified compound common in rays and axial parenchyma (Fig. 9, 10, 11). Vascular bundles present in pith. These pith bundles are amphivasal, with phloem eccentrically placed. A limited amount of secondary growth is evident in the pith bundles.

Begonia parviflora (Carlquist 7062, woody cylinder thinner than in the above collection, plant younger). Qualitative fea­tures the same as in the above collection except as noted. Mean number of vessels per group, 1.40. Mean vessel diameter, 40 (xm. Mean number of vessels per mm2, 66. Mean vessel wall thickness, 2.0 u.m. Mean vessel element length, 306 u,m. Perforation plates predominantly simple, but scalari­form in some metaxylem and early sec-ondary xylem vessels. Mean diameter of libriform fibers at widest point, 18 u,m. Mean wall thickness of libriform fibers, 2.3 u.m. Mean length of libriform fibers, 502 u-m. Mean height of multiseriate rays ex­ceeds 5 mm. Mean width of multiseriate rays 17 cells or 400 u.m. Mean wall thickness of ray cells, 0.8 u.m.

Begonia peruviana (Carlquist 7331). Growth rings absent. Vessels circular or oval in transectional outline (Fig. 1). Mean number of vessels per group, 1.40. Mean vessel diameter, 74 p-m. Mean number of vessels per mm2, 42. Mean vessel wall thick­ness, 1.2 u,m. Mean vessel element length. 359 fxm. Perforation plates predominantly simple, but occasional scalariform plates in metaxylem and early secondary xylem. Lat­eral wall pitting of vessels scalariform. pit apertures wider in vessel-parenchyma pits

1985] CARLQUIST: WOOD ANATOMY OF BEGONIACEAE 63

Figs. 7—11. Wood sections of Begonia parvijlora (Carlquist 7050). Fig. 7. Transection", vessels relatively large in diameter. Fig. 8. Tangential section, ray to left of center; storying in fibers evident. Fig. 9. Ray from radial section: cells markedly upright. Fig. 10. Tangential section portion showing vessel-vessel pitting (left) and libriform fibers (right). Fig. 11. Vessel-axial parenchyma pitting from langeniial section; dropletsof unidentified compound evident. Figs. 7-8, magnification scale above Fig. 1. Fig. 9, scale above Fig. 3. Figs. 10—1 I, scale above Fig. 5.

64 BULLETIN OK THE TORREY BOTANICAL CLUB [VOL. 112

than in vessel-vessel pits (Fig. 4). Libriform fibers nonseptate (Fig. 2, left). Mean di­ameter of libriform fibers at widest point, 18 (im. Mean length of libriform fibers, 496 (xm. Axial parenchyma is of the vasi-centric scanty type, with one layer of cells forming an incomplete sheath around ves­sels or vessel groups. Axial parenchyma in strands of four cells typically. Primary ray areas converted into fibers (Fig. 1, above; Fig. 2, right; Fig. 3), these fibrous cells with lignified walls (mean wall thickness, 1.8 u.m) and with simple pits. The fibers formed in ray areas have tapered tips, typical of fibers rather than of typical ray cells (Fig. 3); fibers from ray areas have been included in the figure for length, since they are not readily distinguishable in macerations from fibers of the fascicular areas. A rayless condition may be said to exist in this spec­imen even though one can differentiate between the fibers of the fascicular areas and those of the ray areas on the basis of wideness (Fig. 2). Mean height of the ray areas exceeds 5 mm, and the mean width of these areas exceeds 25 cells. No areas corresponding to uniseriate rays are pres­ent. Storied structure is evident in fibers of the fascicular areas (Fig. 2, left) and in a few fibers of the ray areas (Fig. 2, right). Droplets of dark-staining unidentified com­pounds common in tyloses and in axial parenchyma (Fig. 4, 5, 6).

Begonia peruviana (Carlquist 7336). Qualitative features the same as those of the preceding collection unless otherwise stated. Mean number of vessels per group. 1.37. Mean vessel diameter, 58 u.m. Mean number of vessels per mm2, 33. Mean vessel wall thickness, 1.2 u.m. Mean vessel element length, 328 fxm. Mean libriform fiber di­ameter at widest point, 16 fxm. Mean lib­riform fiber wall thickness, 1.8 u,m. Mean length libriform fibers, 444 \xm. Rays mul-tiseriate only. Mean height of multiseriate rays exceeds 5 mm. Mean width of mul­tiseriate rays 26 cells or 718 ^.m. Ray cells are markedly erect but not as fiberlike as in the preceding collection. Mean ray cell wall thickness, 0.5 u.m.

RAYLESSNESS. Raylessness characterizes only a few dicotyledon groups, especially herbaceous groups which are secondarily woody (Barghoorn 1941; Carlquist 1970,

1981; Gibson 1978). If fusiform cambial initials are short, as they are in various groups of herbs, yet ray initials are relatively tall, as they often are in such groups (Carl­quist 1962), raylessness can result (Carlquist 1970). If raylessness does occur, older stems tend to develop rays eventually because of horizontal subdivision of cambial initials in the interfascicular areas of the cambium (Barghoorn 1941). To resemble imperfor­ate tracheary elements, cells in the ray areas must be as tall as imperforate tracheary elements in a particular wood sample and they must resemble them in wall thickness and have tapered tips typical of libriform fibers.

All of the above conditions are realized in one collection of.fi. peruviana (Fig. 2, 3). In this species, ray cells have walls of exactly the same thickness as those of the libriform fibers of fascicular areas. Only the fact that fibers in the ray areas are wider in diameter than those of the fascicular areas, and somewhat less storied, would lead one to regard fi. peruviana as an imperfect example of raylessness. Indeed, the difference be­tween fascicular areas and ray areas is obvious in transection (Fig. 1), and one would not guess from transections that wood of B. peruviana satisfies most if not all criteria of raylessness. The species of Begonia studied here, other than B. peru­viana, do not qualify as rayless (and only one collection of that species makes a close approach to the condition). Ray cells in the remaining Begonia species studied here are not as long as are libriform fibers. The stems of B. peruviana have relatively little secondary xylem. Therefore, the more ex­tended period of development in which typical ray cells could develop (Barghoorn 1941) has not elapsed in that species.

PAEDOMORPHOSIS. An assemblage of features in wood anatomy of dicotyledons has been cited as indicative of paedomor­phosis (Carlquist 1962). These include moderate woodiness in a plant with her­baceous or succulent aspects; tendency of tracheary elements to decrease in length or stay about the same in length as sec­ondary growth proceeds (rather than ex­hibit a gradual increase in length as in typically woody plants); tendency for erect cells to predominate in rays and for pro­cumbent cells to be scarce or absent. All

1985] CARLQU1ST: WOOD ANATOMY OF BEGOMACEAE 65

Figs. 12-17. Wood sections of Begonia. Figs. 12—13. B. parviflora (Carlquist 7050). Fig. 12. Transection; axial parenchyma evident around some vessels. Fig. IS, Tangential section portion with tyloses (right) and storied fibers (left). Fig. 14—17. B. maculate (Carlquist, s.n.). Fig. 14. Libriform fibers from radial section, simple pits at left in sectional view. Fig. 15. Ray cells from radial section, borders evident in sectional view of pits. Fig. 16. Ray cells from radial section, showing procumbent cells (center right) and starch content of cells. Fig. 17. Septate fibers containing starch from radial section. Fig. 12, 13, 16, magnification scale above Fig. 3. Fig. 14, 15, 17, scale above Fig. 5.

6 6 BULLETIN OF THE TORREY BOTANICAL CLUB (VOL. 112

of these features are exemplified in the woods of Begonia.

Stress should be laid on the fact that the scalariform lateral wall pitting of vessels in dicotyledon groups which show pae-domorphosis is something not to be ex­pected in terms of earlier work in evolution because one normally expects scalariform lateral wall pitting to accompany scalari­form perforation plates and other primitive features (Frost 1930). In woods with pae-domorphosis, scalariform lateral wall pit­ting is normally accompanied by simple perforation plates. This and other aspects of paedomorphosis have been reviewed recently by Carlquist (1980) and Cumbie (1983).'

Scalariform lateral wall pitting of vessels occurs in metaxylem of species which have scalariform perforation plates in secondary xylem—thereby validating Bailey's (1944) idea that primary xylem is a refugium of primitive features. Bierhorst and Zamora (1965) list 76 species which show the con­dition just mentioned, and these represent only a fraction of what more intensive investigation would uncover. The theory of paedomorphosis accounts for how sca­lariform lateral wall pitting tends to be carried over into secondary xylem of certain species (herbs, succulents, woody herbs, secondarily woody rosette trees), rather than restricted to metaxylem. The theory of paedomorphosis in wood is thus quite different from Bailey's (1944) refugium concept, despite the fact that Bierhorst and Zamora (1965, p. 701) incorrectly equated the two concepts. Bailey's refugium concept describes how specialized xylem features (presumably those adapted to rapid con­duction with relatively low safety) origi­nated in secondary xylem and proceeded into primary xylem evolutionarily. Primi­tive vessel element features were eliminated in both secondary and primary xylem of groups with the greatest xylary speciali­zation. Primitive features were left in pri­mary xylem (e.g., scalariform perforation plates) of other groups, somewhat less spec­ialized with respect to xylem, in which simple perforation plates occur in second­ary xylem. The concept paedomorphosis describes how the primitive features rem­nant in primary xylem can spread evolu­tionarily into secondary xylem in such mod­

erately specialized groups. Those not familiar with the intricacies of wood in predomi­nantly herbaceous groups containing sec­ondarily woody species may be unfamiliar with paedomorphosis and its manifesta­tions, and thus the criteria for the concepts have been restated here.

One might expect that scalariform per­foration plates might, by paedomorphosis, occasionally be carried over (in the mor-phogenetic sense) to secondary xylem in groups with scalariform perforation plates in primary xylem and (basically) simple perforation plates in secondary xylem. Ap­parently the selection for conductive effi­ciency is so great that this ordinarily does not occur. Only a very few cases seem to exemplify this tendency: Pentaphragma hor.s-fieldii (Miq.) Airy-Shaw (Carlquist 1975a) and Patrinia villosa Juss. (Carlquist 1983). These plants have very limited secondary xylem (thus development is insufficient to lose juvenilism) and they grow in moist places (understory in the case of Penta­phragma) which probably induce moderate conductive rates such that scalariform per­foration plates are tolerable in secondary xylem. Even in Patrinia villosa, simple plates outnumber scalariform ones. This is true in the Begonia species studied here also. Note should be taken of the fact that sca­lariform perforation plates in Begonia be­come less frequent as secondary xylem proceeds; this is in accordance with the idea of paedomorphosis. At the periphery of wood of large stems of Begonia parviflora, no scalariform perforation plates were ob­served.

Where scalariform perforation plates do occur in Begonia, they often may be found only on one end of a vessel element, with a simple plate present at the other end. This condition was noted bv Hilde-brand (1859) and Metcalfe and Chalk (1950). A peculiar feature of some perforation plates of Begonia deserves special descrip­tion. In end walls of some Begonia vessel elements from early secondary xylem, sim­ple perforation plates are present, but the end wall also may bear what appear to be scalariform perforations (they actually may be pits: presence of membranes could not be determined accurately with the light microscope). The co-occurrence on a single end wall of a simple perforation plate and

1985] CARLQUIST: WOOD ANATOMY OF BEGONIACEAE 67

scalariform perforations (or pits) in this fashion was figured by Hildebrand (1859); Hildebrand's figure was reproduced by So-lereder (1908, p. 405). This peculiar per­foration plate or end wall morphology may be attributed to paedomorphosis, but it might also be attributed to a deeply-seated genetic basis for production of the scalar­iform lateral wall pitting which is so char­acteristic of Begonia vessels.

Bierhorst and Zamora (1965) list two species of Begonia in which scalariform perforation plates occur in primary xylem, although simple perforation plates char­acterize secondary xylem. Such a config­uration is exactly what one would specify as a precursor to the occasional scalariform plates seen in wood of early secondarv xylem, as in the present study, if paedo­morphosis has been the mechanism for the production of occasional scalariform plates.

The large rays of Begonia change little during secondary growth. Most of the rays are unaltered extensions of primary rays and, as in Piperaceae, little ray breakup occurs. The tendency for rays to remain intact to a large extent may be considered an aspect of juvenilism in Begonia woods.

The occurrence of scalariform lateral-wall pitting in vessels of B. maculata was cited as an indication of paedomorphosis (Carlquist 1962). The wide apertures of the vessel-ray pitting and vessel-axial par­enchyma pitting may relate to diminished mechanical strength of woods showing pae­domorphosis (Carlquist 1975a): the large rays of Begonia also seem to betoken a lowered mechanical strength.

Vessels in radial rows are reported in Begonia by Metcalfe and Chalk (1950). Ra­dial clusters (vessels actually in contact in radial series) were not observed in the present material. Radial clusters area often observed in primary xylem of dicotyledons in which radial clusters do not occur in secondary xylem. Study of material which consists largely of primary xylem (as is true in most Begonia stems) could account for reports of radial rows or clusters.

RELATIONSHIPS. Characteristic features of Begoniaceae include the following: ves­sels mostly solitary, round to oval (radially widened) in transectional outline; vessels with simple perforation plates (occasional scalariform plates near primary xylem);

lateral wall pitting of vessels scalariform with wide apertures on the vessel-paren­chyma pits; imperforate tracheary elements are all libriform fibers (rarely with vestigial borders on pits, the elements therefore fiber-tracheids); libriform fibers septate and with starch or nonseptate; axial paren­chyma of the vasicentric scanty type, 1-2 cells thick and forming a sheath (unusually incomplete) around vessels or vessel groups; rays multiseriate only, composed of erect cells (procumbent and square cells infre­quent, the erect cells in one B. peruviana collection simulating fibers and therefore producing a rayless condition); libriform fibers storied (fewer storied fibers in earlier-formed secondary xylem) and some ray cells storied; droplets of unidentified com­pounds common in ray and axial paren­chyma cells.

The resemblance between the compos­ite description for Begoniaceae above and the features reported by Davidson (1976) for Datiscaceae is very close. Most of the features whereby the two families differ can be attributed to the herbaceous nature (semisucculent canes branching from a rhi­zome system) and the corresponding an­atomical paedomorphosis (juvenilism, neo-teny) in Begoniaceae. Thus, lateral wall pitting of vessels is scalariform in Begon­iaceae, but consists of alternate pits, circular to oval in outline, in Datiscaceae (elongate pits occasional in Datisca glomerata (Presl) Baill.). The occasional scalariform perfo­ration plates in Begoniaceae can be ex­plained as a result of paedomorphosis, as noted above. The wide, tall rays relate to the semisucculent habit, much as do the rays of this nature in Piperaceae, in which a similar habit occurs. Similarities between Piperaceae and Begoniaceae do not con­note any phylogenetic relationship, how­ever.

The nature of parenchyma in Begon­iaceae as compared to that of Datiscaceae shows some degree of difference. Axial parenchyma is vasicentric in both families. The abundant vasicentric parenchyma of Octomeles and Tetrameles is not matched in Begoniaceae, but axial parenchyma is rel­atively scanty in Datisca glomerata. Thus, the difference is not great if one takes Datisca into account. One should note that the libriform fibers of Begonia, by virtue

68 BULLETIN OF THE TORREY BOTANICAL CLUB [VOL. 112

of increased longevity (as evidenced by septum formation and by starch contents), may be in the process of forming a phys­iological substitute for axial parenchyma.

Multiseriate rays in Datiscaceae are much smaller than those in Begoniaceae. How­ever, breakup of rays in Datiscaceae is doubtless much more active, related to the habits of those species. Uniseriate rays are present in Datiscaceae but not in Begoni­aceae. This remains a distinction between the families, but uniseriate rays are not abundant in Datiscaceae, so elimination of uniseriate rays in a phylad leading to Be­goniaceae is not difficult to imagine. Pro­cumbent cells are abundant in rays of Da­tiscaceae, but only at the periphery of stems, not closer to the pith (Davidson 1976). Juvenilism can account for lack of procum­bent cells in rays of Begoniaceae.

Storied structure is a distinctive feature common to Begoniaceae and Datiscaceae.

Thus, there are few differences between Begoniaceae and Datiscaceae not attrib­utable directly or indirectly to differences in habit, and wood anatomy supports a relationship between the two families. Be­cause little doubt has been cast on rela­tionship between Begoniaceae and Datis­caceae (see introduction of this paper), there seems no value at present in com­paring wood anatomy of Begoniaceae to that of a wider range of families.

VESSEL DIAMETER: ECOLOGY. There are no growth rings in the species of Begonia examined here. However, increase in vessel diameter with distance from pith was clearly observed in B. parviflora: vessels from the periphery of the woodier stem Carlquist 7050 averaged 84 u,m diameter, whereas vessels from the less woody stem of Carl­quist 7062 averaged 40 u,m.

Although much comment has been de­voted to change in vessel diameter within growth rings in dicotyledons at large, rel­atively little note has been taken of trends in vessel diameter over longer periods of time. Gradual increase in diameter is per­haps a common trend, as in Octomeles su-matrana Miq. (Davidson 1976) or Trirnenia neocaledonica Bak. f. (Carlquist 1984a). In some woods, little fluctuation in vessel di­ameter is noted. However, by contrast one may cite species of short-lived perennials and even annuals in which decrease in vessel

diameter occurs: Centaurea americana Nutt. (Carlquist 1965). Lopezia grandiflora Zucc. (Carlquist 1975b), and Epilobium cokhicum Alboff (Carlquist 1977), for example. A logical hypothesis to explain this is that trees and many shrubs penetrate deeper soil levels and thereby secure steadily in­creased water supplies as they grow, a fact reflected in increased water transpiration and increased vessel capacity. Larger and more exposed crowns also form during growth and transpire more water; such foliage would not be formed, of course, unless roots simultaneously became more extensive and tended to tap deeper water supplies which would be more abundant and more reliable. Short-lived perennials and annuals, on the contrary, frequently experience decreasing water availability, since they tend to tap surface soil levels from which water evaporates readily; such plants may terminate under conditions of drought. This would be reflected in steady decrease in vessel diameter, much like what can be observed in a single growth ring of a ring-porous species (an annual can be said to have wood consisting of a single growth ring).

The marked increase in vessel diameter of'Begoniapan>iflora in accordance with stem diameter and plant size suggests that the plant steadily increases water conduction as stems become taller and leaves are con­comitantly more exposed to sunlight. There is a simultaneous decrease in number of vessels per mm2, but nonetheless there is a net increase in conductive area per mm2

(even if the conductive area remained the same, wider vessels would provide less re­sistance to flow). Presumably a larger Be­gonia plant taps greater water supplies than a small one.

I f we develop figures on "Mesomorphy" (Carlquist 1977), we can supply figures for the Begonia species. "Mesomorphy" is vessel diameter multiplied by vessel element length, the resulting product then divided by num­ber of vessels per mm2. Figures in Begonia are: B. duchartrei, 400; B. maculata, 435; B. parviflora (Carlquist 7050), 1260; B. par­viflora (Carlquist 7062), 184; B. peruviana (Carlquist 7331), 639; B. peruviana (Carl­quist 7336), 576. These figures tend to confirm the increase in wood mesomorphy with increase in plant size in B. parviflora.

1985] CARLQUIST: WOOD ANATOMY OF BEGONIACEAE 69

In addition, these Mesomorphy figures show that all the woods of Begonia must be de­scribed as mesomorphic. Much lower fig­ures, for example, can be seen in many Onagraceae; within that family, figures comparable to those for Begonia can be found in Fuchsia and Hauya (Carlquist 1977). This is hardly surprising to those familiar with the similar ecological regimes in which Begonia and Fuchsia may be found. The sensitivity of die Mesomorphy index is clear, but one must note that Begonia, Fuchsia, and Hauya are alike in all having leaves which are vulnerable to wilt; plants with leaves more resistant to wilt may deviate in Mesomorphy figures on that account.

A low degree of vessel grouping is ev­ident in the Begonia species studied. Vessel grouping can occur in dicotyledon groups in which librifbrm fibers are the imper­forate tracheary element type; however, appreciable grouping (2.0 vessels per group or above) only occurs in those groups which occupy dry or seasonally dry areas (Carl­quist 1984b), and these habitats are not characteristic for Begonia species.

Li te ra ture Cited

BAILEY, I. W. 1944. The development of vessels in angiosperms and its significance in morphological research. Am. J. Bot. 31:421-428.

BAIU;HOORN, E. S. 1941. The ontogenetic and phy-logenetic specialization of rays in the xyleni of dicotyledons. III. The elimination of rays. Bull. Torrey Bot. Club 68:317-325.

BitRHORST, D. W. AND P. M. ZAMORA. 1965. Primary xylem elements and element associations of an­giosperms. Am. J. Bot. 52:657-710.

CARLQUIST, S. 1962. A theory of paedomorphosis in dicotyledonous woods. Phvtomorphology 12: 50— 45.

. 1965. Wood anatomy of Cynareae (Compositae). Aliso 6(1): 13-24.

. 1970. Wood anatomy of insular species of Plantago and the problem of ravlessness. Bull. Torrey Bot. Club 97:353-361.

. 1975a. Ecological strategies of xylem evolution. University of California Press. Berke­ley. Los Angeles, and London. 259 p.

. 1975b. Wood anatomy of Onagraceae, with notes on alternative modes of photosynthate

translocation in dicotyledon woods. Ann. Mis­souri Bot. Gard. 62:386-424.

. 1977. Wood anatomy of Onagraceae: additional species and concepts. Ann. Missouri Bot. Gard. 64:627-637.

. 1980. Further concepts in ecological wood anatomy, with comments on recent work in wood anatomy and evolution. Aliso 9:499-553.

. 1981. Types of cambial activity and wood anatomy of Styttdium (Stylidiaceae). Am. J. Bot. 68:778-785.

. 1982. The use of ethylenediamine in softening hard plant structures for paraffin sec­tioning. Stain Tech. 57:311-317.

. 1983. Wood anatomy of Calyceraceae and Valerianaceae. with comments on aberrant perforation plates in predominantly herbaceous groups of dicotyledons. Aliso 10:413-425.

. 1984a. Wood anatomv of Trimeniaceae. PI. Syst. Evol. 144:103-118.

. 1984b. Vessel grouping in dicotyledon wood: significance and relation to imperforate tracheary elements. Aliso 10:505-525.

CRONQLIST, A. 1981. An integrated system of clas­sification of flowering plants. Columbia Uni­versity Press, New York. 1262 p.

CUMBIE, B. G. 1983. Developmental changes in the wood of Bocconia vulcanica Donn. Smith. IAWA Bull, n.s., 4:131-140.

DAHLGREN, R. M. T. 1980. A revised system of clas­sification of the angiosperms. Bot. J. Linn. Soc. 80:91-124.

DAVIDSON, C. 1976, Anatomy of xylem and phloem of the Datiscaceae. Nat. His. Mus. Los Angeles Contrib. in Sci. 280:1-28.

FROST, F. H. 1930. Specialization in secondary xylem in dicotyledons. I. Origin of vessel. Bot. (.a/. 89:67-94.

GIBSON, A. C. 1978. Rayless secondary xylem in Halopkjtum. Bull. Torrey Bot. Club 105:39-44.

HILDEBRAND, F. 1859. Anatomische Studien Liber die Stamme der Begoniaceen. Thesis, Berlin. 34 p.

IRMSCHER, E. 1925. Begoniaceae. In A. Engler &: K. Prantl, [eds.], Die Natiirlichen Pflanzenfamilien, ed. 2, 21:548-588. Verlag von Wilhelm Engel-mann, Leipzig.

LINDLEY, J. 1846. The vegetable kingdom. Vol. 1. Bradbury & Evans, London. 908 p.

METCALFE, C. R. AND L. CHALK. 1950. Anatomv of the dicotyledons. Clarendon Press. Oxford. 1500 P-

SOI.EREDER, H. 1908. Systematic anatomy oi the di­cotyledons (trans, by L. A. Boodle and F. E. Fritsch). Clarendon Press, Oxford. 1182 p.

TAKHTAJAN, A. L. 1980. Outline of the classification of flowering plants (Magnoliophvta). Bot. Rev. 46:225-359.

THORNE, R. F. 1983. Proposed new realignments in the angiosperms. Nordic J. Bot. 3:85-117.