IAWA Bulletin n.s., Vol. 12 (4),1991: 389-417 WOOD ANATOMY OF THE BIGNONIACEAE, WI'IH A COMPARISON OF TREES AND LlANAS by Peter GBS80n 1 and David R. Dobbins2 Summary The secondary xylem anatomy of trees Hess 1943; Jain & Singh 1980; see also Greg­ and lianas was compared in the family Big­ ory 1980 and in preparation). Den Outer and noniaceae. General descriptions of the family Veenendaal (1983) have compared Bignonia­ and the six woody tribes are provided. Lianas ceae wood anatomy with that of Uncarina belong to the tribes Bignonieae, Tecomeae (Pedaliaceae) and Santos has recently describ­ and Schlegelieae, and most have ve.ssels of ed the New World Tecomeae for an M.S. two distinct diameters, many vessels per unit thesis (Santos 1990) and Flora Neotropica area, large intervascular pits, septate fibres, (Santos in press). large heterocellular rays often of two distinct The family Bignoniaceae has a wide dis­ sizes, scanty paratracheal and vasicentric axial tribution from about 400N to 35°S, encom­ parenchyma and anomalous growth. Conver­ passing North and South America, Africa sely, trees, which belong to the tribes Coleeae, south of the Sahara, Asia, Indonesia, New Crescentieae, Oroxyleae and Tecomeae gen­ Guinea and eastern Australia. It is mainly erally have narrower vessels in one diameter tropical, with most species in northern South class, fewer vessels per unit area, smaller America, and consists of lianas, trees and intervascular pits, non-septate fibres, small shrubs with very few herbs. Estimates of the homocellular rays, scanty paratracheal, ali­ number of genera and species vary: 650 form or confluent parenchyma, and none ex­ species in 120 genera (Willis 1973) or 800 hibits anomalous growth. The majority of species in 110 genera (Takhtajan 1987), al­ both trees and Hanas possess growth rings, though Gentry (1973) considers that the num­ are diffuse-porous, have non-solitary vessels ber of genera is too high. which lack helical thickenings, and few have We have examined specimens from 35 tree apotracheal parenchyma or storied structure. and 27 liana genera from six of the eight All species have alternate intervascular pitting tribes recognized by Gentry (1980), Cronquist and simple perforation plates. (1981), and Takhtajan (1987). Two tribes, Key words: Lianas, trees, wood, anomalous both neotropical and monogeneric, Tourret­ structure, xylem, Bignoniaceae. tieae and Eccremocarpeae, are not covered in this paper since they are herbaceous vines (Gentry 1980). Much of the world distribu­ Introduction tion of the family is also covered. An overall This paper has two purposes: to present comparison of trees and lianas has been made, descriptions of the secondary xylem anatomy and anatomical descriptions of the six tribes of the Bignoniaceae, and to compare the wood are presented. The tribes Coleeae, Crescen­ anatomy of trees and Hanas within the family. tieae and Oroxyleae consist entirely of trees Metcalfe and Chalk (1950) have surveyed the and shrubs, whereas all members of the wood anatomy of this family, but there is no tribes Bignonieae and Schlegelieae are lianas. comprehensive publication, although many The Tecomeae are mainly trees and shrubs, genera and species are covered in papers on but Campsis and Tecomanthe are climbers. particular geographical regions (Record & The trees come from a wide range of habitats 1) Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3DS, U.K. 2) Biology Department, Millersville University, Millersville, Pennsylvania 17.5.51, U.S.A. Downloaded from Brill.com10/04/2021 01:59:21PM via free access 390 IAWA Bulletin n.s., Vol. 12 (4),1991 and climates, e.g. temperate (Catalpa), warm (lAWA Committee 1989) as a guide, with temperate and subtropical (Chilopsis), mon­ emphasis on those characters most pertinent soon forest (Millingtonia), secondary tropical to the comparison of trees with lianas in this forest (Oroxyiurn), arid tropics (Catophrac- family. These are: growth ring definition; tes), dry tropical forest (Tabebuia, Tecoma, vessel distribution, number, diameter, element Cybistax, Godmania), seasonal savanna (Ki- length and intervascular pit size; fibre type gelia), mangrove (Dolichandrone spathacea), (i.e. septate or non-septate) and wall thick­ and moist tropical forest (many genera). The ness; axial parenchyma distribution; ray size lianas tend to be mostly from moist tropical and composition; storeying of tissues; pres­ forest, although some do grow in drier con­ ence or absence of anomalous secondary ditions, e. g. Cydista and Macfadyena. thickening. Lianas are relatively rapidly growing clim­ In order to compare the frequency of these bers which produce very long stems and have features in trees and lianas, Tables 1 and 2 limited vascular cambial activity (Schenck (pages 398-407) were prepared. In the ten 1893; Dobbins 1971; Dobbins & Fisher 1986) tree genera in which we examined three or whereas trees and shrubs grow more slowly more species, some features were of variable and have greater cambial activity resulting in occurrence within the genus. For these genera, larger amounts of xylem and phloem. Because an index is given to denote the proportion of of their climbing habit, lianas are subject the genus with the feature (i.e., if three spe­ to different stresses than trees and shrubs. cies out of five have a feature, the index is Lianas form an extensive canopy, and the dis­ 0.6). The lianas are treated separately in tance water and solutes have to be transport­ Table 2. The indices give an indication of the ed is often greater than for trees (Ewers & variability of a feature within a given genus, Fisher 1989). Moreover, the control of devel­ and are used to calculate the overall percent­ opment and subsequent anatomical features age frequencies for the comparison of trees in Hanas are demonstrably different from and Hanas, which are on a genus basis. those of trees and shrubs (Schenck 1893), Information on whether the trees are de­ and this paper documents some of these ana­ ciduous or evergreen was rarely available on tomical differences in the Bignoniaceae. herbarium sheets or in the floras consulted (see the Appendix), except for Van Steenis Materials and Methods (1977). Where available this information is Apart from some Hanas which were col­ given in Table 1. It is assumed that most of lected fresh from the Fairchild Tropical Gar­ the Hanas are evergreen. den, Florida, U.S.A. (FTG), all the material At least 25 measurements were made of examined was from the following institutions: each quantitative feature per sample, and Jodrell Laboratory, Kew (KJw); museum these were bulked to obtain the means for collections, Kew (Kw); Institute of Sys­ species represented by more than one sample. tematic Botany, University of Utrecht, the For species with vessels of two distinct Netherlands (Uw); Rijksherbarium, Leiden, diameters, only the wider vessels were mea­ the Netherlands (Lw); Universidad de Los sured. Statistical analyses of mean vessel Andes, Merida, Venezuela (MERw); Forest diameter (Fig. 22), mean vessel element Products Laboratory, Madison, Wisconsin, length (Fig. 23) and intervascular pit border U.S.A. (MADw and SJRw). The slides la­ diameter (Fig. 24) were made using Kolmo­ belled FHOw are in the Jodrell Laboratory gorov-Smirnov Chi-square tests. These data, collection ( KJw), but originated from Ox­ plus number of vessels per square mm were ford Forestry Institute. The abbreviations for examined using an ANOVA. The relationship wood collections follow those in Stern (1988). between mean vessel diameter in J.LM and A complete list of the specimens examined is vessel number per square mm is shown in given in the Appendix (pages 415-417). Figure 25. The statistical results are given Microscopical observations of sectioned with the captions for Figures 22-25. Quan­ material were interpreted using the IAWA titative data are incomplete because some List of Features for Hardwood Identification samples provided too few measurements for Downloaded from Brill.com10/04/2021 01:59:21PM via free access Gasson & Dobbins - Trees versus lianas in Bignoniaceae 391 inclusion in the tables and graphs; however, some species and genera is not uniform, i.e. a sufficient number of samples was measured some genera have more than one feature state, to compare trees and lianas in Figures 22-25 the total percentage of genera with a feature (see pages 411 & 412). may exceed 100% (e.g. in trees 48% have distinct. 55 % indistinct and 17 % absent De8criptions growth rings which totals 120%. The same The descriptions are in tribal order for applies to axial parenchyma, where particular trees: Co1eeae, Crescentieae, Oroxyleae and distribution patterns like scanty paratracheal, Tecomeae, followed by lianas: Tecom<;:ae, confluent and aliform often occur together Bignonieae and Schlegelieae. The Tecomeae and are not mutually exclusive). The final trees and lianas are described separately, one description outlines the features found in the after the other for easy reference. These tribal family and refers to the tribes to which certain descriptions provide more detail and qualify characters are restricted. some of the information given in Tables 1 (trees) and 2 (lianas). Since the anatomy of (text continued on page 397) Legends of Figures 1-13 (trees) and 14-21 (lianas and climbers): Figs. 1-4. Trees. - 1: Catalpa bignonioides TS. Ring-porous (scale as no.3). - 2: Jacaranda copaia TS. Diffuse-porous, winged aliform axial parenchyma (scale as no.3). - 3: Kigelia afri- cana TS. Diffuse-porous, lozenge aliform and eonfluent parenchyma (scale line = 500 111Il). - 4: Crescentia cujete TS, diffuse-porous, initial, confluent and aliform parenchyma (scale as no.3). Figs. 5-9 Trees. - 5: Catalpa longissima TS. Diffuse-porous, confluent parenchyma (scale line = 500 J.Lm). - 6: Paratecoma peroba, the only tree with a tendency towards vessels of two dis­ tinct sizes (scale as no.5). -7: Oroxylum indicum TS, showing a vessel with a foraminate per­ foration plate and confluent parenchyma (scale line = 500 J.Lm).