Araliaceae) with Systematic and Ecological Implications

IAWA Journal, Vol. 33 (2), 2012: 163–186

Wood anatomy of Cussonia and Seemannaralia (Araliaceae) with systematic and ecological implications

Bernard J. De Villiers1, Alexei A. Oskolski1, 2, Patricia M. Tilney1 and Ben-Erik Van Wyk1, *

SUMMARY The wood structure of two related African genera, Cussonia Thunb. (15 of 21 species) and the monotypic Seemannaralia R.Vig. (Araliaceae) is examined. The considerable diversity in wood anatomical characters within these taxa is mostly related to environmental factors; taxonomic groupings or phylogenetic relationships seem to be less important. The shortening of vessel elements and fibres, an increase in vessel number per group, a decrease in vessel diameter and a reduction in the number of bars of perforation plates, are associated with the more temperate species. The changes in vessel grouping show a significant correlation with rainfall. The placement of the simple-leaved Cussonia species in the subgenus Protocussonia and the isolated position of C. paniculata Eckl. & Zeyh., the only member of the subgenus Paniculatae, are supported. Many Cussonia species share a very low fibre to vessel element length ratio. Despite the basal position of Seemannaralia relative to Cussonia revealed by molecular data (Plunkett et al. 2004), its wood structure is more specialised in terms of the Baileyan major trends in wood evolution. This discrepancy may be the effect of a long-term adaptation of tropical ancestors of Seemannaralia to drier biomes. Key words: Africa, fibre/vessel element length ratio, latitudinal trends, phylogenetics, taxonomy.

INTRODUCTION

The Araliaceae are relatively poorly represented in Africa, with five indigenous genera and one naturalised (Klopper et al. 2006); only two of them, namely Cussonia Thunb. and Seemannaralia (Seem.) Vig. are endemic to this continent. Cussonia comprises 21 species (one of which is currently undescribed) that are evergreen or deciduous trees or shrubs (occasionally caudiciform geophytes) showing a considerable range of leaf types: both simple and compound mono- and bi-digitate, as well as ternate (Cannon 1970, 1978; Strey 1973, 1981; Bamps 1974a, 1974b; Reyneke 1981, 1982). The genus

1) Department of Botany and Plant Biotechnology, University of Johannesburg, P.O. Box 524, Auck- land Park 2006, South Africa. 2) Botanical Museum, V.L. Komarov Botanical Institute of the Russian Academy of Sciences, Prof. Popov St., 197376 St. Petersburg, Russia. * Corresponding author [E-mail: [email protected]].

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 164 IAWA Journal, Vol. 33 (2), 2012 is distributed in sub-Saharan Africa, Yemen (the Arabian Peninsula) and the Comoro Islands (Frodin & Govaerts 2003). It is widespread within a range of biomes from tropical moist broadleaf forests to montane grasslands and Mediterranean wood- and shrublands (fynbos) covering tropical, dry and temperate climatic zones (Olson et al. 2001). The species occur in all seven centres of endemism of Africa, as defined by Linder (2001). Two species have a wide distribution, namely Cussonia arborea Hochst. ex A.Rich. and C. spicata Thunb. The range of C. arborea is from Senegal, east to Ethiopia and south to Zimbabwe while C. spicata is distributed from South Africa, along the eastern parts of Africa, to Sudan (Bamps 1974a). Seemannaralia gerrardii (Seem.) R.Vig., the single species of this monotypic genus, is a small to medium-sized, deciduous tree with palmately-lobed leaves. The geographical distribution is restricted to a few localities in the eastern provinces, KwaZulu-Natal and Mpumalanga, of the Republic of South Africa (Burtt & Dickison 1975). Seemann- aralia differs from Cussonia in its imbricate petal aestivation and dry, laterally-com- pressed fruits, as opposed to the valvate petal aestivation and fleshy, globose fruits in the latter genus. Moreover, Seemannaralia is distinctive in displaying pseudoparacarpy, i.e. the formation of the central cavity in the fruit by mechanical rupture of the septum between two ovarian locules, which has not been reported to date for indehiscent fruits in any other taxa (Oskolski et al. 2010). Nevertheless, a close relationship between these taxa has been suggested by various authors (Viguier 1906; Harms 1914; Strey 1973, 1981; Burtt & Dickison 1975; Reyneke 1981). Recently Plunkett et al. (2004) has shown that this relationship is supported by molecular data. A few infrageneric divisions of Cussonia have been attempted. Strey (1973, 1981) divided the genus into three subgenera, namely Cussonia, Paniculatae Strey and Proto- cussonia Strey based on leaf and inflorescence morphology. Reyneke (1981, 1982) placed more emphasis on the inflorescence morphology and proposed some changes to the system of Strey (1973, 1981), such as elevating the rank of the section Capitatae Strey to subgenus based on simple umbels, and the creation of two sections within the subgenus Capitata (Strey) Reyneke, namely Sessiliflora Reyneke and Pedicellata Reyneke. The section Sessiliflora is characterised by sessile flowers while the section Pedicellata has stalked flowers. These two systems were based mainly on the southern African members of the genus, which may raise criticism. Previous wood anatomical investigations have contributed to the understanding of infrageneric relationships in a number of genera of the family (Rodriguez 1957; Oskolski 1996, 2001), especially Schefflera (Oskolski 1995) and Meryta (Oskolski et al. 2007). It is therefore expected that a thorough examination of Cussonia species will increase the understanding of infrageneric relationships within the genus. Data on the wood anatomy of Cussonia were scanty with only five (or perhaps six) species having been examined previously, namely C. angolensis (Seem.) Hiern, C. arborea, C. holstii Harms, C. spicata, C. thyrsifloraThunb., and an unidentified species (Record & Hess 1944; Metcalfe & Chalk 1950; Rodriguez 1957; Oskolski 1994, 1996; InsideWood 2004). The wood structure of Seemannaralia gerrardii was described by Burtt and Dickison (1975) and the same single sample then re-examined by Oskolski (1994, 1996).

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 165

The present study, which also contributes to a general survey of the wood anatomy of Araliaceae (Oskolski 1994, 1995, 1996, 2001; Oskolski & Lowry 2001; Oskolski et al. 2007), surveys the wood structure of 15 Cussonia species and Seemannaralia gerrardii. The results are interpreted in terms of the systematic relationships and environmental features of these genera.

MATERIAL AND METHODS

Most of the wood samples examined were collected by the authors during field work in South Africa in 2007, or obtained from the wood collections at the Bundesforsc- hungsanstalt für Forst- und Holzwirtschaft, Hamburg (RBHw), the Centre Technique Forestier Tropical, Montpellier (CTFw), the Royal Botanic Gardens, Kew (Kw), the Nationaal Herbarium Nederland, Utrecht (Uw), and the Musée Royal de l’Afrique Centrale, Tervuren (Tw). Voucher specimens are deposited at JRAU, MO, and various other institutions, as shown in Table 1. Wood samples of Cussonia holstii and C. gamtoosensis Strey were obtained from P.P. Lowry II (Missouri Botanical Garden, St. Louis, and Musée national d’Histoire naturelle, Paris). The samples were taken mostly from stems with a secondary xylem radius of more than 10 mm, i.e. the distance from the pith where the average length of vessel elements in Araliaceae is likely to have reached mature values (Baas 1976). The sample of C. holstii may be juvenile as information from the voucher specimen is insufficient for making a clear determination. A list of the samples examined is given in Table 1, together with authors for names, which are not repeated from here on. Data on the mean average rainfall and biome type are based on Ernst & Walker (1973), Lovett & Pócs (1993), Linder (2001), Olson et al. (2001), Burger (2002) and Woodward et al. (2004). Standard procedures for the study of wood structure were employed to prepare sections and macerations for light microscopic studies (Carlquist 1988). Descriptive terminology follows Carlquist (1988) and the IAWA List of Microscopic Features for Hardwood Identification (IAWA Committee 1989). Principal Components Analysis (PCA) was used in an effort to establish an integra- tive view on the variation of wood features and, in particular, to differentiate among between-sample variation patterns in wood anatomy and environmental characters. A total of 16 variables, in which the variation appears to be more or less independent of one another, were selected for the PCA. These are the radius of the wood sample, average length of vessel elements, average length of fibres, maximum number of bars per perforation plate, percentage of simple perforation plates, average number of vessels per mm2, percentage of solitary vessels, average diameter of vessel lumina, average vertical size of intervessel pits, maximum width of multiseriate rays (in cells), average height of multiseriate rays, average number of uniseriate rays per 1 mm, average number of multiseriate rays per 1 mm, average tangential size of ray cells, latitude and annual rainfall. Ecological and latitudinal trends in the variation of wood features have been estimated by an analysis of variance (ANOVA). The programme package STATISTICA 7.0 was used to perform the PCA and ANOVA test.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 166 IAWA Journal, Vol. 33 (2), 2012 (mm) Rainfall 560 1329 1100 866 560 650 1130 650 1100 800 1100 709 (continued on next page)

S 14°; E 27° S 28° 21'; E 32° 32' 2° W N 6°; S 7°; E 15° S 33° 55'; E 20° 01' S 03° 17'; E 36° 55' S 24° 30'; E 30° 38' S 29° 50'; E 31° 00' S 25° 38'; E 30° 58' S 28° 23'; E 32° 34' S 25 °03'; E 28° 15' Coordinates S 7°; E 15° Seemannaralia . The coordinates and mean maximum rainfall (mm) Zambia Bhangazi Dam, St. Lucia, KwaZulu-Natal, Africa South Ghana Democratic Republic of the Congo Africa Port Elizabeth, South Gamtoos River, Arusha National Park, ) Ngurdoto Crater, (alt. 1600 m Tanzania Arumeru District, Africa Ohrigstad, Mpumalanga, South Natal Botanical Garden, Durban, Africa KwaZulu-Natal, South Africa. Mpumalanga, South Waterval-Onder, Park Offices, St. Lucia, KwaZulu-Natal, Africa South Gauteng, Park, Johannesburg, Weltevreden Africa South Locality Angola Benguela,

Cussonia and Wood collection sample number Wood Kw 10642 A.A.B.J. & Oskolski 105 Villiers de (JRAU); (Bdv105) RBHw 16615 CTFw 5864 Phillipson 5623 P.B. Phillipson, Lowry II, P.B. P.P. Simon 4986 (MO) V. A. Mkeya & & Tilney P.M. B.J.Villiers, de A.A. Oskolski 72 (JRAU); (BdV72) A.A. Oskolski & B.J.Villiers, de 98 (JRAU); (BdV98) Wyk B-E. van & Tilney P.M. B.J.Villiers, de A.A. Oskolski 82 (JRAU); (BdV82) A.A. & Oskolski B.J.Villiers de 100 (JRAU); (BdV100) (BdV1) 1 (JRAU); B.J.Villiers de Voucher specimen(s) Voucher R. Dechamps 1072, Uw 23502 Eckl. & Zeyh. Strey Strey Harms ex Engl. Wild. De A.Rich. Hochst. ex C. arborea C. arenicola & Pellegr. C. bancoensis Aubrév. C. brieyi C. gamtoosensis Strey C. holstii C. natalensis Sond. C. nicholsonii C. paniculata Strey C. sphaerocephala Thunb. C. spicata for each locality are also given. Table 1. Specimens Table used in the wood anatomical study of Species Cussonia angolensis (Seem.) Hiern

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 167 (mm) Rainfall 500 1100 508 350 355 210 950 950 950 950 ' ' ' ' ' ; E 39° 2 8 ; E 30° 0 3 ; E 30° 0 7 ; E 30° 0 7 ; E 31° 2 ' ' ' ' '

Coordinates S 35° 17'; E 25° 55' S 23° 53'; E 28° 57' S 33° 59'; E 22° 37' S 24° 07'; E 29° 08' S 10° 0 2 S 29° 5 4 S 29° 5 4 S 29° 5 4 S 29° 4 7 Tanzania Tanzania Locality Olifantskop, Port Elizabeth, Eastern Cape, Africa South Africa Magoebas Kloof, Limpopo, South Wilderness, Ebb and Flow Nature Reserve, Africa Cape, South Western Africa Potgietersrus, Limpopo, South Lindi, Lindi Rural, Kenya Africa Richmond, KwaZulu-Natal, South Roselands Farm, Richmond, KwaZulu- Natal, South Africa Natal, South Roselands Farm, Richmond, KwaZulu- Natal, South Africa Natal, South and Nongoma, Vryheid Between KwaZulu-Natal, South Africa KwaZulu-Natal, South

A. Oskolski 92 A. Oskolski & J. de Villiers & A . & Villiers J. de J. de Villiers, A . Villiers, J. de J. de Villiers, A.A. Oskolski & Villiers, J. de J. de Villiers & A.A. Oskolski 107 & Villiers J. de Tw 39164 Tw . . Voucher specimen(s) Voucher collection sample number Wood B.A.A. & Oskolski 60 Villiers J. de (JRAU); (BdV60) & B.Tilney P.M. Villiers, J. de A.A. Oskolski 70 (JRAU); (BdV70 B.A.A. Oskolski & Villiers, J. de 62 (JRAU); (BdV62) Wyk B-E. van B. & Villiers C. de Villiers, J. de 58 (JRAU); (BdV58) Villiers J.M. de H. J. Schlieben 5557 (P), RBHw 1865, B . (JRAU); (BdV92) B . 99 (JRAU); (BdV99) Wyk B-E. van B-E. van Wyk 97 (JRAU); (BdV97) Wyk B-E. van (JRAU); (BdV107) (Table 1 continued) (Table Species C. spicata . C. spicata . Thunb. C. thyrsiflora Reyneke W.F. C. transvaalensis C. zimmermannii Harms C. zuluensis Strey C. zuluensis B Seemannaralia gerrardii (Seem.) Harms B Seemannaralia gerrardii

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 168 IAWA Journal, Vol. 33 (2), 2012

Table 2. Wood anatomical characters of Cussonia and Seemannaralia.

1: Radius of wood sample (mm). – 2: Length of vessel elements (average and min-max, µm). – 3: Vessel frequency (per mm²). – 4: Tangential diameter of vessels (average and minimum-maximum; µm). – 5: Average and the greatest number of vessels in a vessel group. – 6: Solitary vessels (%). – 7: Number of bars per perforation plate (minimum- maximum). – 8: Percentage of simple perforation plates (%). – 9: Average length of libriform fibres (average and minimum-maximum; µm). – 10: Fibre to vessel element length ratio. – 11: Vertical size of intervessel pits (average and maximum; µm). – 12: Common and maximum number of marginal uniseriate rows by multiseriate rays (common and maximum). – 13: Kribs’ types of rays (Hom = Homogeneous, Het = Heterogeneous). –14: Width of multiseriate rays (average and maximum, cells). –15: Height of multiseriate rays (average and maximum; µm). – 16: Number of multiseriate rays per mm. – 17: Number of uniseriate rays per mm. – 18: Total number of rays per mm. – 19: Tangential size of ray cells (average and maximum; µm). – 20: Scores of factor 1. – 21: Scores of factor 2.

Part 1

Taxa 1 2 3 4 5 6 7 8 9 Ave±ste min–Max ave±ste min–Max ave max ave±ste min–Max

Cussonia angolensis 55 927±46.6 258–1447 17.5 133±5.5 24–191 1.7 8 28.5 3–22 0 1262±54.0 29–1906 arborea 50 1109±60 349–1583 9.9 90±3.3 25–135 1.8 9 30.3 4–46 0 1223±36.5 360–1740 arenicola 11 978±25.3 708–1267 25.0 58±2.7 26–100 1.8 6 25.9 2–14 0 987±20.2 665–1551 bancoensis 60 1187±35 844–1640 12.2 90±4.8 13–149 1.7 12 31.5 5–31 0 1201±25.3 724–1641 brieyi 55 1043±59 413–1561 6.0 158±3.9 116–201 1.2 6 67.6 4–16 0 1072±20.5 940–1334 gamtoosensis 16 641±30.9 296–949 19.4 61±2.4 30–92 2.8 12 12.5 0–18 9 704±19.6 352–1077 holstii 14 702±26.2 280–983 21.7 73±3.4 24–175 2.9 12 10.5 0–6 3 731±37.7 208–1353 natalensis 30 858±20.3 546–1188 13.3 71±3.6 32–118 2.2 7 17.7 0–11 19 888±19.1 506–1390 nicholsonii 36 826±35.2 318–1312 22.4 59±2.1 30–82 1.8 7 28.1 1–6 0 778±31.9 250–1340 paniculata 50 579±35.5 141–1117 27.9 59±2.5 22–95 3.2 14 7.9 0–16 21 682±26.0 320–1404 sphaerocephala 15 1253±38.2 683–1616 7.6 99±5.6 56–146 1.8 5 31.2 3–16 0 1217±28 545–1609 spicata (BdV1) 30 641±32.2 256–1080 13.9 93±4.1 69–131 2.8 11 12.2 0–22 25 697±4.8 57–1391 spicata (BdV60) 42 932±39.0 221–1471 17.3 61±3.2 26–104 2.0 7 16.7 2–36 0 1009±28.6 537–1308 spicata (BdV70 33 1032±32.2 1112-1277 16.9 72±4.1 39–135 2.1 8 16.8 0–35 12 1048±47.4 348–1532 thyrsiflora 19 711±26.4 328–1040 8.9 79±5.2 24–140 1.7 5 31.4 0–15 3 706±25.8 239–1385 transvaalensis 62 636±32.2 262–1159 21.8 127±5.0 57–197 2.9 12 10.5 0–14 39 817±34.2 305–1264 zimmermannii (Schlieben 5557) 28 1114±36.5 698–1416 20.8 118±6.5 50–194 1.9 6 25.5 3–11 0 1158±38.9 406–1688 zimmermannii (Tw 39164) 60 1052±33 747–1360 17.4 110.5±4.2 45–152 1.81 8 27.72 2–14 0 1126±37.2 411–1666 zuluensis (BdV92) 42 906±31.6 547–1323 6.8 92±8.5 43–273 2.0 6 17.1 1–17 0 906±30.4 595–1375 zuluensis (BdV99) 52 916±34.0 403–1340 8.9 53±3.2 14.3–83.5 2.2 6 16.9 1–18 0 992±21.8 573–1346

Seemannaralia gerrardii (BdV97) 26 448±62.8 159–755 21.8 59±2.9 26.2–84.1 2.9 12 10.5 0–3 70 738±28.0 397–1037 gerrardii (BdV107) 22 577±27.0 33–941 20.9 58±3.0 39–90 2.9 12 10.2 0–8 50 765±18.0 413–1083

(see also next page)

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 169

Part 2

Taxa 10 11 12 13 14 15 16 17 18 19 20 21 ave max com max ave max ave max ave max

Cussonia angolensis 1.36 8.5 9.7 1 2 Het IIB 4.9 6 804 1289 3.2 0.2 3.4 21.8 30.8 -1.19 -0.44 arborea 1.10 6.7 12.1 1 3 Het IIB 4.3 6 574 1066 2.2 0.7 2.9 14.6 24.5 -0.81 0.38 arenicola 1.01 6.4 8.6 1 2 Het IIB 2.4 4 297 699 3.1 0.5 3.5 13.4 27.5 0.54 1.35 bancoensis 1.01 6.2 8.3 2 6 Het IIA 5.6 9 793 1332 3.6 0.8 4.4 18.0 29.8 -1.81 -0.30 brieyi 1.03 6.3 8.4 1 2 Het IIB 4.4 9 692 1740 1.3 0.4 1.7 12.4 21.3 -2.09 1.84 gamtoosensis 1.10 6.5 10.17 1 2 Het IIB 2.3 4 389 923 2.5 0.4 2.8 17.3 32.5 1.18 0.47 holstii 1.04 6.9 10.7 0 1 Hom I 3.6 6 433 950 4.0 0.1 4.2 21.6 31.7 0.61 0.14 natalensis 1.03 7.2 10.31 0 1 Hom I 5.4 10 532 1004 2.3 0.2 2.5 27.2 44.8 0.19 0.13 nicholsonii 0.94 8.2 10.1 1 3 Het IIA 2.6 4 562 2047 2.7 0.7 3.4 70.1 113.2 0.58 -0.67 paniculata 1.18 8.2 11.2 2 5 Het IIA 4.8 10 837 2139 3.5 0.9 4.4 26.3 36.4 0.64 -2.08 sphaerocephala 0.97 6.4 9.8 1 4 Het IIA 3.0 6 342 1252 3.8 0.7 4.5 18.3 30.3 -0.47 0.72 spicata (BdV1) 1.09 8.8 12.3 1 3 Het IIA 3.8 6 408 2059 3.0 0.1 3.1 12.2 20.8 0.76 -1.19 spicata (BdV60) 1.02 8.9 10.9 1 3 Het IIA 3.7 6 651 1545 3.2 0.7 4.0 16.2 24.4 -0.07 -1.42 spicata (BdV70) 1.01 8.7 11.9 1 3 Het IIA 3.8 6 747 1496 2.8 0.4 3.1 18.4 25.5 -0.25 -0.90 thyrsiflora 0.99 9.0 11.5 0 1 Hom I 3.1 6 475 1157 2.6 0.6 3.2 17.0 26.2 0.45 0.003 transvaalensis 1.28 7.7 10.5 1 3 Het IIA 5.0 8 584 1183 2.4 0.2 2.6 22.9 43.0 0.37 -0.74 zimmermannii (Schlieben 5557) 1.04 8.1 9.8 1 4 Het IIA 4.0 6 635 1149 2.6 0.1 2.7 17.0 22.4 -0.54 0.46 zimmermannii (Tw 39164) 1.07 7.2 9.4 1 2 Het IIA 4.3 6 1015 1871 2.7 0.3 3.0 18.9 27.9 -1.10 -0.23 zuluensis (BdV92) 0.99 11.1 13.3 1 3 Het IIA 5.0 7 814 1440 2.2 0.2 2.4 24.3 39.3 -0.38 -0.90 zuluensis (BdV99) 1.01 6.0 7.7 1 5 Het IIA 3.7 6 554 1605 1.4 0.1 1.5 13.6 24.2 0.06 0.85

Seemannaralia gerrardii 1.64 7.2 9.0 0 2 Hom I to 3.3 5 285 445 1.5 0.2 1.7 8.9 12.7 1.57 1.26 (BdV97) Het IIB gerrardii 1.33 6.3 8.5 0 2 Hom I to 3.0 4 218 413 2.2 0.2 2.3 11.9 21.7 1.76 1.26 (BdV107) Het IIB

RESULTS

Cussonia (Table 2; Fig. 1–10, 13, 14, 16–18). Growth rings mostly indistinct [but absent or very indistinct in C. bancoensis (Fig. 7) and C. transvaalensis] and marked by differences in fibre wall thickness between latewood and earlywood [C. arborea (Fig. 1) and C. sphaerocephala], or by tangential bands of radially flattened fibres (Fig. 3, 5), or sometimes in combination with interrupted lines of marginal axial parenchyma [C. natalensis (Fig. 9)]; fibres of varying wall thickness in earlywood and latewood in C. natalensis (Fig. 9), C. nicholsonii and C. zuluensis (BdV99). Vessels solitary, in clusters and in radial multiples of 2 to 6 (up to 12 in C. bancoensis, C. gamtoosensis, C. holstii and C. transvaalensis, and up to 14 in C. paniculata). Ves- sel frequency from 7 per mm2 in C. zuluensis (BdV92) to 25 per mm2 in C. arenicola.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 170 IAWA Journal, Vol. 33 (2), 2012

Solitary vessels rounded (Fig. 1), rarely [or more commonly in C. paniculata (Fig. 3), C. arenicola (Fig. 5), C. brieyi and C. spicata (BdV60)] angular in outline, rather narrow (tangential diameter < 85 µm) in C. nicholsonii and C. zuluensis (BdV99) to wider in other species [tangential diameter up to 200 µm in C. brieyi, and 270 µm in C. zuluensis (BdV92)]. Vessel walls 2–5 µm thick [up to 8 µm in C. holstii, C. natalensis and C. zuluensis (BdV92)]. Tyloses in C. paniculata and C. spicata (BdV1, BdV60). Vessel elements usually medium in length [(142–)580–702(–1080) µm long] in C. paniculata, C. spicata (BdV1), C. gamtoosensis, C. holstii and C. transvaalensis, and longer in other species examined, especially [(280–)1080–1250(–2000) µm long] in C. arborea, C. zimmermannii (RBHw1865), C. bancoensis and C. sphaerocephala (BdV100). Perforation plates scalariform, reticulate (Fig. 13, 14) and simple in more or less oblique end walls. Simple perforation plates common in C. natalensis and C. transvaalensis, less common in C. gamtoosensis, C. holstii, C. paniculata, C. spicata (BdV1, BdV70), C. thyrsiflora and C. zuluensis, and not observed in other species. Scalariform perforation plates with wide to narrow [mostly narrow in C. angolensis, C. brieyi, C. gamtoosensis and C. zimmermannii (RBHw 1865)] bars of various numbers [up to 39 in C. spicata (BdV60)]. Double perforation plates occur (Fig. 16). Intervessel pits mostly scalariform to transitional and alternate in C. arenicola (Fig. 16), C. spicata, C. sphaerocephala and C. transvaalensis, or mostly alternate (sometimes scalariform and transitional) in other taxa (Fig. 13), 4–9(–13) µm in vertical diameter, with rounded and polygonal in outline [mostly rounded in C. sphaerocephala, C. spicata, C. thyrsiflora, C. transvaalensis, C. zimmermannii (RBHw 1865) and C. zuluensis] with slit-like apertures. Vessel-ray and vessel-axial parenchyma pits similar to intervessel pits in size and shape, half-bordered, with indistinct borders (Fig. 17), but occasionally distinct in C. arborea, C. natalensis, C. nicholsonii (Fig. 18), C. paniculata, C. sphaerocephala, C. spicata, and C. zuluensis. Helical thickenings absent throughout. Vascular tracheids not found. Fibres libriform, thin-walled in C. arborea, C. transvaalensis, C. zimmermannii and C. zuluensis [1.3–4(–6) µm], and also moderately thick-walled, fibre walls 2–6(–9) µm thick in other species, with few simple to minutely bordered pits, with slit-like apertures in radial walls. Septate fibres relatively few inC. angolensis, C. bancoensis, C. gamtoosensis, C. holstii and C. zimmermannii (Tw39164), and numerous in other species studied (Fig. 13). Axial parenchyma scanty paratracheal [mostly solitary strands near vessels in C. arborea, C. arenicola, C. bancoensis, C. gamtoosensis, C. nicholsonii, C. spicata (BdV1, BdV60), C. transvaalensis, C. zimmermannii and C. zuluensis, and in incomplete (sometimes complete in C. paniculata) sheaths in other species], and marginal (forming interrupted tangential lines in C. natalensis). Strands of (3–)5–7(–17) cells. Total number of rays per mm 1.5–4.5, rays uni- and multiseriate of 2–3(–4) cells in width in C. arenicola (Fig. 6), C. gamtoosensis and C. nicholsonii, and wider in other species [up to 8 cells in C. transvaalensis, up to 9 cells in C. bancoensis (Fig. 8) and C. brieyi, and up to 10 cells in C. paniculata (Fig. 4) and C. natalensis (Fig. 10)]. Ray height commonly < 1 mm, but some rays exceed 1 mm in all species studied except in C. arenicola, C. gamtoosensis, C. holstii and C. transvaalensis.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 171

Figure 1–4. Species of Cussonia. – 1: TS of C. arborea showing a growth ring boundary (arrow) marked by differences in fibre wall thickness between latewood and earlywood. Soli- tary vessels rounded in outline. – 2: TLS of C. arborea showing the multiseriate rays composed mainly of procumbent cells with 1–2 marginal rows of upright and square cells. Radial canals present. – 3: TS of C. paniculata showing the growth ring boundaries (arrows) marked by tangential bands of radially flattened fibres. Solitary vessels angular in outline. – 4: TLS of C. paniculata showing the multiseriate rays with 1–4 marginal rows and incomplete sheaths of square and upright cells. Radial canals present. — Scale bars = 300 µm.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 172 IAWA Journal, Vol. 33 (2), 2012

Figure 5–8. Species of Cussonia. – 5: TS of C. arenicola showing the growth ring boundaries (arrows) marked by tangential bands of radially flattened fibres. Solitary vessels mostly angular in outline. – 6: TLS of C. arenicola illustrating the narrow multiseriate rays composed of procumbent cells with 1–2 marginal rows of upright and square cells. – 7: TS of C. bancoensis, in which growth rings are absent. Solitary vessels mostly rounded to slightly angular in outline. – 8: TLS of C. bancoensis, with multiseriate rays with 1–5 marginal rows of square and upright cells. Sheath cells (black arrow). Radial canal present. — Scale bars = 300 µm.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 173

Figure 9–12. Species of Cussonia and Seemannaralia. – 9: TS of C. natalensis showing the growth ring boundaries (arrows) marked by tangential bands of radially flattened thin-walled fibres in combination with interrupted lines of marginal parenchyma. Solitary vessels mostly rounded to slightly angular in outline. – 10: TLS of C. natalensis with multiseriate rays composed almost exclusively of procumbent cells. Sheath cell (black arrow). – 11: TS of S. gerrardii (BdV97) showing the growth ring boundaries marked by tangential bands of radially flattened fibres, and by differences in vessel frequency and diameter between latewood and earlywood. Solitary vessels rounded in outline. – 12: TLS of S. gerrardii (BdV97) with multiseriate rays composed mostly of procumbent cells; occasional upright and square cells occasionally occur in 1–2 marginal rows. Sheath cells (black arrow). Radial canals absent. — Scale bars = 300 µm.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 174 IAWA Journal, Vol. 33 (2), 2012

Figure 13–18. RLS of selected species of Cussonia and Seemannaralia. – 13: Cussonia arborea: scalariform perforation plates with few bars. Alternate intervessel pitting. Septate fibres. – 14: C. bancoensis: reticulate and scalariform perforation plates with few and numerous bars. – 15: Seemannaralia gerrardii [BdV97]: simple perforation plates. Scalariform perforation plate with two bars. Alternate intervessel pitting. Scalariform ray-vessel pits. – 16: C. arenicola: double scalariform perforation plate of two perforation plates with single bars. Scalariform to opposite and alternate intervessel pitting. – 17: C. arenicola: ray-vessel pitting. Pits mostly with indistinct borders. – 18: C. nicolsonii: ray-vessel pitting. Pits mostly with distinct borders. — Scale bars for 13, 14 = 200 µm, for 15 = 100 µm, for 16–18 = 20 µm.

Multiseriate rays with procumbent body cells; upright and square cells in a few rays forming 1 marginal row in C. holstii and C. natalensis (Fig. 10) (Kribs’ type Homo- geneous I), or in nearly all rays forming 1 to 2 marginal rows (Fig. 2), or 1–4 (up 6) marginal rows (Fig. 4, 8, Table 2) (Kribs’ types Heterogeneous IIb to Heterogeneous IIa). Upright and square cells occur as solitary sheath cells [exclusively solitary in C. angolensis, C. bancoensis (Fig. 8), C. brieyi, C. holstii, C. natalensis (Fig. 10) and C. thyrsiflora], or as incomplete sheaths, sometimes complete sheaths in C. gamtoosensis, C. nicholsonii, C. paniculata (Fig. 4), C. transvaalensis and C. zuluensis (BdV99)]. Uniseriate rays mostly of square and upright cells.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 175

Radial canals in all species examined, rather narrow (15–40 µm in diameter), bordered by few small, thin-walled epithelial cells (Fig. 2, 4, 6, 8, 10); frequency varies considerably between samples, and shows no clear relationship to taxonomy or ecology. No mineral inclusions observed in any of the species examined.

Seemannaralia (Table 2; Fig. 11, 12, 15) Growth rings distinct, marked by tangential bands of radially flattened fibres, by differences in vessel frequency and diameter between latewood and earlywood (Fig. 11). Solitary vessels rounded, rarely angular in outline, rather narrow (tangential diameter < 90 µm). Vessel frequency c. 20 per mm². Vessels solitary and in radial multiples and clusters of 2 to 6 (up to 12). Vessel walls 2–4 µm thick. Tyloses not observed. Average vessel element lengths medium [(160–)450–580(–940) µm long]. Perfora- tion plates mostly simple, sometimes scalariform with few (up to 8 in BdV107) wide or narrow bars (Fig. 15), in more or less oblique end walls. Forked bars on scalariform perforation plates. Intervessel pits with alternate arrangement, 4–9 µm in vertical size, mostly with polygonal or rarely rounded outline and slit-like apertures. Vessel-ray and vessel-axial parenchyma pits scalariform (Fig. 15) to alternate, half-bordered, with indistinct to narrow borders. Helical thickenings absent throughout. Vascular tracheids not found. Fibres libriform, thin- to thick-walled, fibre walls 2–6(–7) µm, with few simple to minutely bordered pits, with slit-like apertures in radial walls, almost exclusively septate. Axial parenchyma scanty paratracheal in incomplete sheaths near vessels. Strands of axial parenchyma of 5 to 8 cells. Total number of rays per mm 1.5–3, rays uni- and multiseriate with 2–5(–6) cells in width. Ray height < 0.5 mm. Multiseriate rays of procumbent body cells; upright and square cells in a few rays forming 1 to 2 (to rarely 3) marginal rows (Kribs’ types Homogeneous I to Heterogeneous IIb), and rarely also as solitary sheath cells (Fig. 12). Uniseriate rays mostly of square and upright cells. Radial canals absent. Crystals not found in ray cells.

Numerical analysis The loadings of the four factors (i.e. coefficients of correlation between the factors and the characters examined) with the greatest influence in the Principal Components Analysis (PCA) are shown in Table 3. Together they explain 67.8% of the total variation, with the first factor accounting for almost one third of the total variation. The scores of the two most important factors for each of the species under study are given in Table 2. The pattern of within-sample variation of wood characters examined and of the environmental variable (summarised by the distribution of the scores of Factors 1 and 2) is presented in Figure 19. Factor 1, which explains 33.2% of the total variation, shows a high negative correlation with length of vessel elements, length of fibres, and diameter of vessels, and a high positive correlation with percentage of solitary vessels, and a weaker correlation with the percentage of simple perforation plates and latitude. To a certain degree, Factor 1 correlates negatively with the height of multiseriate rays, and with annual rainfall. As

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 176 IAWA Journal, Vol. 33 (2), 2012

Table 3. Loadings of the four factors with the greatest influence in the Principal Components Analysis (correlation coefficients, r > 0.7 noted in bold).

Factor 1 Factor 2 Factor 3 Factor 4

Radius of wood sample 0.601 -0.333 -0.537 0.050 Average length of vessel elements 0.833 0.153 0.436 0.026 Average vertical size of intervessel pits 0.015 -0.660 -0.040 0.162 Average number of vessels per sq.mm -0.597 -0.237 -0.015 0.249 Average diameter of vessels 0.671 0.086 -0.339 -0.003 Percentage of solitary vessels 0.752 0.437 -0.051 0.137 Maximum number of bars per perforation plate 0.476 -0.431 0.240 -0.519 Percentage of simple perforation plates -0.676 0.089 -0.534 -0.246 Average length of fibres 0.814 0.174 0.269 -0.039 Maximum width of rays, cells 0.456 -0.309 -0.538 -0.128 Average height of multiseriate rays 0.676 -0.537 -0.234 0.234 Number of multiseriate rays per mm 0.131 -0.556 0.526 -0.103 Number of uniseriate rays per mm -0.059 -0.471 0.026 -0.487 Average tangential size of ray cells -0.048 -0.371 0.124 0.779 Annual rainfall 0.684 0.307 -0.199 0.042

Percentage of total variation 32.8 14.6 11.3 9.1

2.5

2.0

1.5

1.0

0.5

0.0

Factor 2 -0.5

-1.0

-1.5

-2.0

-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 Factor 1 Figure 19. Pattern of within-sample variation of wood characters of species of Cussonia and Seemannaralia summarised by the distribution of the scores of Factors 1 and 2 from the Principal Components Analysis (PCA). The samples collected from low latitudes (nearly equatorial zone between N 20° and S 20°) are symbolised in black, the samples from higher latitudes (S 20°– S 35°) in white. Strey’s (1973) subgeneric groupings within Cussonia are marked as: square (/) = subgenus Paniculata; triangles (:, Δ) = subgenus Protocussonia, circles ($, %) = subgenus Cussonia. The samples of Seemannaralia are marked by diamonds (2).

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 177 the interrelated variation of these wood features is unlikely to be the effect of a common morphogenetic cause, we suggest that Factor 1 shows an influence of environment on wood formation. To test this suggestion, a statistical analysis (ANOVA) of latitudinal and ecological trends for wood features was performed and is discussed below. The influence of Factors 2, 3, and 4 on total variation is considerably lower than that of Factor 1. Factor 2 shows a weak negative correlation with the size of intervessel pits. As for Factors 3 and 4, their interpretation remains obscure because they show no significant correlation with any of the analysed features. The scores of the two most important factors for each of the species under study are given in Table 2. The pattern of within-sample variation of wood characters and of the environmental variable examined (summarised by the distribution of the scores of Factors 1 and 2) is presented in Figure 19. The samples form a broad cluster stretching across both axes (Fig. 19). Two samples, both of S. gerrardii, are located at the edge of this cluster along the axis of Factor 1. The single sample of C. paniculata, whose subgeneric rank was recognised by Strey (1973, 1981) and Reyneke (1981, 1982), is separated from all others along the axis of Factor 2. The sub-clusters of the samples belonging to the two other subgenera of Cus- sonia from Strey’s (1973, 1981) classification (namelyProtocussonia and Cussonia), cannot be distinguished within this cluster. Nevertheless, two somewhat overlapping sub-clusters can be discerned, each comprising the species either from low latitudes (between the nearly equatorial zone: N 20°–0° and S 0°–20°, in Fig. 19), or from higher latitudes southwards (S 20°–S 35°, in Fig. 19). These sub-clusters can also be distinguished with some overlap in the projections of sample variation on Factors 1 and 3, and on Factors 1 and 4 (not presented here). As the wood samples examined were taken from stems of different sizes (including rather slender ones), the effect of stem radius on the quantitative wood features was estimated by an analysis of variance (ANOVA) using the F-test. Variability of every wood character listed in Table 2 was compared between and within three groups of wood samples with different radii (10–20 mm, 21–40 mm, and 41–62 mm). The results of the ANOVA revealed a statistically significant increase in the average height of multiseriate rays (F = 7.19; p < 0.047; Fig. 20) and in the average width of multiseriate

Figure 20 & 21. Quantitative characters of multiseriate rays plotted against the radius of the wood sample (without bark). – 20: Average height of multiseriate rays. – 21: Average width of multiseriate rays (in cells).

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 178 IAWA Journal, Vol. 33 (2), 2012 rays (F = 7.57; p < 0.038; Fig. 21) as the stem radius increased whereas other wood characters showed no significant effect in relation to stem size. In order to clarify the environmental influence on wood structure, comparison between two groups of Cussonia and Seemannaralia wood samples collected in low (0°– S 20°) and higher (S 21°–S 35°) latitudes, and also between two groups from dry (with annual rainfall less than 800 mm) and humid habitats (annual rainfall 800–2000 mm) was carried out using the data listed in Table 2 (the latitudes of collection localities are given in Table 1). The effects of annual rainfall and latitude on the variability of the wood features were also estimated by the one-way ANOVA using the F-test (Table 4). These analyses revealed latitudinal trends such as decreases in the annual rainfall, in the mean length of the vessel elements (Fig. 22) and of the fibres (Fig. 23), in the percentage of solitary vessels (Fig. 24), the average number of perforation plate bars (Fig. 25), the average height of multiseriate rays (Fig. 26), and the average number of vessels per group (Fig. 27) which, in higher latitudes, were found to be statistically significant (with p < 0.5). Moreover, a decrease in the average number of vessels per group, as well as increases in the percentage of solitary vessels (Fig. 28), and in the average number of bars per perforation plate (Fig. 29) show statistically significant relations to the increase in rainfall. In addition, the variability of the above listed wood features showing statistically significant trends in relation to annual rainfall and/or latitude was estimated by two- factor ANOVA in order to distinguish between the environmental influence from the effect of stem size on wood structure (Table 5). Two size groups of wood samples (with the radius < 35 mm, and the radius 35 mm and more) were analysed in combination with paired groups of samples either from low (0°–S 20°) and higher (S 21°–S 35°) latitudes, or from dry (with annual rainfall less than 800 mm) and humid habitats (annual rainfall 800–2000 mm). This test confirmed that decreases in the mean length of the vessel elements and of the fibres, the percentage of solitary vessels, and the average number of perforation plate bars in higher latitudes as well as decrease in the average number of vessels per group, and increase in the percentage of solitary vessels are not significantly influenced by the radius of wood samples and, therefore, these trends may be considered as representative of environmental influences. The latitudinal trend in variation of the average height of multiseriate rays seems to be, however, an effect of the stem size rather than of the latitude. Moreover, no statistically significant effects of annual rainfall and sample radius on the variation of the average number of perforation plate bars were revealed by the two-factor ANOVA.

→ Figure 22–29. Latitudinal and ecological variation in some wood anatomical characters of Cussonia and Seemannaralia. The wood characters are plotted against the latitude in Fig. 22–26, in Fig. 27–29 against annual rainfall. Small wood samples (radius less than 35 mm) are marked by diamonds (2), larger ones (radius 35 mm and more) by black triangles (:). – 22: Average length of vessel elements. – 23: Average length of fibres. – 24: Percentage of solitary vessels. – 25: Average number of bars per perforation plate. – 26: Average height of multiseriate rays. – 27: Average number of vessels per group. – 28: Percentage of solitary vessels. – 29: Average number of bars per perforation plate.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 179

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 180 IAWA Journal, Vol. 33 (2), 2012

DISCUSSION

The Cussonia species and Seemannaralia samples studied share certain wood characters that are typical for a number of other Araliaceae genera, including the presence of both scalariform and simple perforation plates, septate fibres and scanty paratracheal axial parenchyma (Metcalfe & Chalk 1950; Rodriguez 1957; Oskolski 1994, 1996, 2001). The range of wood structural diversity within these genera is, however, considerable for such a relatively small group. The results of the numerical analyses suggest that the diversity in wood anatomical characters within these taxa is mostly related to environmental factors and, to a lesser extent, to taxonomic groupings or phylogenetic relationships. In some cases, however, we cannot distinguish between ontogenetic (related to sample radius) and latitudinal effects on the variability of wood features because of insufficient sampling.

Table 4. Effects of annual rainfall and latitude on the variability of the wood features estimated by an analysis of variance (ANOVA) using the F-test. Two groups of wood samples collected in low (0°–S 20°) and higher (S 21°–S 35°) latitudes, and also two groups of wood samples from dry (with annual rainfall less than 800 mm) and humid habitats (annual rainfall 800–2000 mm) were analysed. Degrees of freedom: between-group = 1, within-group = 20. Statistically significant effects (p < 0.05) noted in bold.

Wood feature Annual rainfall Latitude F P F P Radius of wood sample 1.367 0.256 3.210 0.088 Length of vessel elements 2.169 0.156 5.965 0.024 Number of vessels per sq.mm 0.357 0.557 0.384 0.542 Tangential diameter of vessels 1.724 0.204 11.592 0.003 Average number of vessels per group 7.737 0.012 4.094 0.057 Maximum number of vessels per group 1.429 0.246 0.001 0.972 Percentage of solitary vessels 7.659 0.012 6.954 0.016 Average number of bars per perforation plate 5.719 0.027 12.300 0.002 Maximum number of bars per perforation plate 0.087 0.771 0.003 0.957 Percentage of simple perforation plates 3.298 0.084 3.903 0.062 Vertical size of intervessel pits 0.288 0.597 1.167 0.293 Length of fibres 1.998 0.173 11.336 0.003 Average width of multiseriate rays (μm) 0.493 0.491 2.181 0.155 Maximum width of multiseriate rays (μm) 0.355 0.558 0.003 0.957 Average width of multiseriate rays (cells) 0.528 0.476 3.415 0.079 Maximum width of multiseriate rays (cells) 0.044 0.835 0.751 0.396 Height of multiseriate rays 1.426 0.246 4.880 0.039 Number of multiseriate rays per 1 mm 0.442 0.514 1.440 0.244 Number of uniseriate rays per 1 mm 0.039 0.846 1.304 0.267 Total number of rays per 1 mm 0.081 0.779 0.780 0.387 Tangential diameter of ray cells 2.584 0.123 0.374 0.548

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 181

Table 5. Effects of wood sample size on the variability of the wood features showing statistically significant trends related to annual rainfall and latitude (see Table 4) estimated by the analysis of variance (two-factor ANOVA) using the F-test. Two size groups of wood samples (with the radius <35 mm. and the radius 35 mm and more) were analysed in combination with either two groups of samples collected in low (0°–S 20°) and higher (S 21°–S 35°) latitudes. or two groups of wood samples from dry (with annual rainfall less than 800 mm) and humid habitats (annual rainfall 800–2000 mm). Degrees of freedom: between-group for latitude or annual rainfall = 1. between-group for wood sample radius = 1; within-group = 19. Statistically significant effects (p < 0.05) noted in bold.

Annual Radius of Latitude Radius of Wood feature rainfall wood sample wood sample –––––––––––– –––––––––––– ––––––––––– –––––––––––– F P F P F P F P

Length of vessel elements 1.177 0.291 0.336 0.569 4.568 0.046 0.283 0.601

Tangential diameter of vessels 0.645 0.432 0.927 0.348 9.016 0.007 0.541 0.471

Average number of vessels per 5.541 0.029 0.097 0.758 2.859 0.107 0.582 0.455 group Percentage of solitary vessels 5.403 0.031 0.128 0.724 5.222 0.034 0.483 0.495

Average number of bars per 3.255 0.870 0.816 0.377 9.333 0.006 1.090 0.309 perforation plate Length of fibres 0.786 0.386 0.975 0.336 8.731 0.008 0.653 0.429

Height of multiseriate rays 0.031 0.862 15.523 0.001 2.623 0.121 14.781 0.001

Latitudinal and ecological trends Distinct latitudinal and sometimes also ecological trends are associated with the variation of such important wood characters as the length of tracheary elements, the type of perforation plates, the grouping of vessels and the diameter of vessel lumina as well as the height of multiseriate rays (Tables 4 & 5, Fig. 22–29). Among them, the shortening of vessel elements and fibres, as well as a decrease in vessel diameter, have been established as environmental trends for many other woody dicotyledonous genera with wide geographical distributions (Baas 1973, 1986; Van der Graaff & Baas 1974; Van den Oever et al. 1981; Noshiro & Baas 2000; Lens et al. 2003). The lengths of tracheary elements are linked to the size of the cambial fusiform cells, but the adaptive or functional significance of these parameters and their changes remain obscure (Noshiro & Baas 2000). Larger vessel groupings (i.e. a smaller percentage of solitary vessels) in higher latitudes have been reported for some plant taxa (Carlquist 1966; Van der Graaff & Baas 1974), but exceptions have also been noted (Baas & Schweingruber 1987; Baas et al. 1988). Apparently, the tendency towards large vessel groupings is a result of adaptations to dry environments which are not necessarily correlated to high latitudes (Carlquist 1966; Lindorf 1994; Alves & Angyalossy-Alfonso 2000). This suggestion is supported by our results for Cussonia and Seemannaralia, where the parameters of

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 182 IAWA Journal, Vol. 33 (2), 2012 vessel grouping (both the number of vessels per group and the percentage of single vessels) are also correlated with annual rainfall. As far as the types of perforation plates are concerned, our results are seemingly the opposite to the findings of other authors (Baas & Schweingruber 1987; Lenset al. 2003), who reported that scalariform perforation plates occur more commonly at higher latitudes. In contrast to their findings, theCussonia species with an essentially equatorial distribution have almost exclusively scalariform perforation plates whereas simple perforation plates occur mainly in those species from the southern regions of Africa. The frequency of simple perforation plates can reach c. 40% in C. natalensis and up to 70% in S. gerrardii. Similar tendencies were reported by Alves and Angyalossy-Alfonso (2000) where a higher frequency of scalariform perforation plates was noticed in humid environments (compared to semi-arid and arid). Wheeler and Baas (1991) suggested that this tendency towards simpler perforation plates in xeric habitats is an adaptation to the need for increased conductivity in narrow vessels and therefore supporting higher rates of transpiration. In the present study, the reduction in the number of bars on the perforation plates in Cussonia and Seemannaralia correlates not only with an increase in latitude but also with a decrease in rainfall. We, however, cannot distinguish between ontogenetic (related to sample radius) and environmental effects on the variability of bar numbers on perforation plates because of insufficient sampling.

Taxonomic implications Although the two genera studied show an overall similarity in wood structure, they differ markedly in several features, and also in the results of the PCA (Fig. 19). In contrast to Cussonia, Seemannaralia is distinctive in having predominantly simple perforation plates, an absence of radial secretory canals, and also a relatively high (1.3–1.6) fibre to vessel length (F/V) ratio. At the same time, the presence inSeemannaralia of homogeneous rays and alternate intervessel pits that are polygonal in outline link this genus to the species of Cussonia belonging to subgenus Protocussonia (see below). In his infra-generic classification, Strey (1973, 1981) recognised the subgenera Cussonia (comprising most of the species), Protocussonia, and Paniculatae. An isolated position of C. paniculata, the single member of the subgenus Paniculatae [its subgeneric rank was also recognised by Reyneke (1981, 1982)], is supported by the results of the PCA (Fig. 19). We found, however, no wood anatomical characters for clearly distinguishing between C. paniculata (i.e., the subgenus Paniculatae) and the subgenus Cussonia. Within the subgenus Cussonia, C. arenicola, C. nicholsonii and C. gamtoosensis are distinctive in their narrow rays that do not exceed five cells in width. A close relationship between C. arenicola and C. nicholsonii was recognised by all the students of Cussonia, and Reyneke (1981, 1982) even proposed that these two taxa be considered as varieties of C. zuluensis. This opinion cannot be supported by wood anatomical data: C. arenicola and C. nicholsonii differ clearly enough from C. zuluensis and from each other by ray width, the number of marginal rows on the multiseriate rays and the type of intervessel pitting. No relationship of C. arenicola and C. nicholsonii to C. gamtoosensis has been suggested to date.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 183

It is noteworthy that many Cussonia species (with the exception of C. angolensis, C. arborea, C. gamtoosensis, C. spicata and C. transvaalensis) share very low F/V ratios that are nearly equal to one. In other words, the vessel elements in these species are nearly the same length as the libriform fibres, a condition that could be the result of intrusive growth suppression during the differentiation of fibres. This low F/V ratio is quite uncommon for Araliaceae and has been recorded for only a few species, such as Tupidanthus calyptratus Hook.f. & Thoms., Dendropanax chevalieri (Vig.) Merr. and Aralia searelliana Dunn etc. (Oskolski 1994, 1996), and is uncommon as well as in other families of dicotyledons (Carlquist 2003). In the case of Cussonia, we found this condition not only as a sporadic peculiarity of a particular species but as a common tendency shared by members of the genus. This feature could be of taxonomic importance and merits further investigation. Molecular analyses showed that Seemannaralia represents a branch which is basally diverged from the well-supported subclade comprising all Cussonia species (Lowry et al. 2004; Plunkett et al. 2004). Despite this basal position the wood structure in Seemannaralia is more specialised in terms of the major trends of secondary xylem evolution established by the Baileyan school (Frost 1930; Bailey 1944) than in any species of Cussonia, as the shorter vessel elements and predominantly simple perforation plates suggest. As the members of the Cussonia-Seemannaralia clade are widespread within a range of humid and arid habitats, both in tropical and subtropical zones, the short fusiform cambial cells and the numerous simple perforation plates in Seemann- aralia are more likely a result of a long evolutionary history of expansion into the dry grass- and shrubland biomes of southern Africa, rather than as the ancestral conditions (Bredenkamp et al. 2002). A similar explanation can also be proposed based on another case of apparent reversals in major evolutionary trends within Araliaceae, viz. for the temperate genus Tetrapanax (K. Koch) K. Koch. As molecular data suggest, this taxon is placed in a basal position to the tropical clade comprising Heteropanax Seem. and the Asian Schefflera species (Lowry et al. 2004; Plunkett et al. 2005). However, it has a more specialised wood structure than that of the latter two taxa (Oskolski 1994, 1996). Generally, accurate analyses of inter-relationships between the major evolutionary trends and latitudinal or/and ecological trends in diversity of the same wood features in other plant taxa on the frameworks of their molecular trees could be of great interest.

ACKNOWLEDGEMENTS

The National Research Foundation (NRF, South Africa), the Russian Foundation of Basic Research (RFFI, grants # 06-04-48003 and 09-04-00618 held by A.A. Oskolski) and the Ministry of Education and Science of the Russian Federation (contract #16.518.11.7071 for A.A. Oskolski) are thanked for funding the project. The University of Johannesburg is thanked for the use of its facilities. The authors would like to thank Johan Hurter (SANBI, Lowveld Botanical Garden), Hugh Glen (SANBI, Natal Herbarium) and Ernst van Jaarsveld (SANBI, Kirstenbosch Botanical Garden) for permission to col- lect material in the Gardens. Pete Lowry (Missouri Botanical Garden, St. Louis, and Musée national d’Histoire naturelle, Paris), Hans-Georg Richter (Bundesforschungsanstalt für Forst- und Holzwirtschaft, Hamburg), Pierre Détienne (Centre Technique Forestier Tropical, Montpellier), Peter Gasson (Royal Botanic Gardens, Kew), Pieter Baas (Nationaal Herbarium Nederland, Leiden) and Hans Beeckman

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 184 IAWA Journal, Vol. 33 (2), 2012

(Musée Royal de l’Afrique Centrale, Tervuren, Tw) are also thanked for their great assistance in ob- taining some of the wood samples. The authors also acknowledge the help of various officials from the different provinces in South Africa for granting collecting permits to B. J. de Villiers.

REFERENCES

Alves, E.S. & V. Angyalossy-Alfonso. 2000. Ecological trends in the wood anatomy of some Brazilian species I: Growth rings and vessels. IAWA J. 21: 3–30. Baas, P. 1973. The wood anatomical range in Ilex (Aquifoliaceae) and its ecological and phylogenetic significance. Blumea 21: 193–258. Baas, P. 1976. Some functional and adaptive aspects of vessel member morphology. Leiden Bot. Series 3: 157–181. Baas, P. 1986. Ecological patterns of xylem anatomy. In: J. Givnish (ed.), On the economy of plant form and function: 327–352. Cambridge University Press, Cambridge. Baas, P., P.E. Esser, M.E.T. van der Westen & M. Zandee. 1988. Wood anatomy of the Olea- ceae. IAWA Bull. n.s. 9: 103–182. Baas, P. & F.H. Schweingruber. 1987. Ecological trends in the wood anatomy of trees, shrubs and climbers from Europe. IAWA Bull. n.s. 8: 245–274. Bailey, I.W. 1944. The development of vessels in angiosperms and its significance in morpho- logical research. Amer. J. Bot. 31: 421–428. Bamps, P. 1974a. Contribution á l’étude des Araliacées africaines. Bull. Jard. Bot. Nat. Belg. 44: 101–139. Bamps, P. 1974b. Araliaceae II: 1. Flora d’Afrique Centrale. Zaïre-Rwanda-Burundi, Sperma- tophytes. Jardin Botanique National de Belgique, Brussels. Bredenkamp, G.J., F. Spada & E. Kazmierczak. 2002. On the origin of northern and southern hemisphere grasslands. Plant Ecol. 163: 209–220. Burger, A-M. 2002. A Study: Using WORLDMAP: of distributions of African savanna plants. Syst. Geogr. Pl. 71: 301–312. Burtt, B.L. & W.C. Dickison. 1975. The morphology and relationships of Seemannaralia (Ara- liaceae). Edinb. J. Bot. 33: 449. Cannon, J.F.M. 1970. Araliaceae. In: A.W. Exell, A. Fernandes & E.J. Mendes (eds.), Conspectus Florae Angolensis, Vol. 4: 360–363. Instituto de Investigação Científica de Angola, Lisbon. Cannon, J.F.M. 1978. Araliaceae. In: E. Launert (ed.), Flora Zambesiaca, Ed. 4: 621–632. Carlquist, S. 1966. Wood anatomy of Compositae: a summary, with comments on factors con- trolling wood evolution. Aliso 6: 25–44. Carlquist, S. 1988. Comparative wood anatomy. Springer Publishers, Berlin. Carlquist, S. 2003. Wood anatomy of Aextoxicaceae and Berberidopsidaceae is compatible with their inclusion in Berberidopsidales. Syst. Bot. 28: 317–325. Ernst, W.H.O. & B.H. Walker. 1973. Studies on the hydrature of trees in Miombo woodland in South Central Africa. J. Ecol. 61: 667–673. Frodin, D.G. & R. Govaerts. 2003. World checklist and bibliography of Araliaceae. The Royal Botanic Gardens, Kew, London. Frost, F.H. 1930. Specialization in secondary xylem in dicotyledons. I. Origin of vessels. Bot. Gaz. 91: 88–96. Harms, H. 1914. Araliaceae. Die natürlichen Planzenfamilien. Nachträge 4: 219. IAWA Committee. 1989. IAWA list of microscopic features for hardwood identification. IAWA Bull. n.s. 10: 219–332. InsideWood. 2004-onwards. Published on the Internet. http://insidewood.lib.ncsu.edu/search [31.10.2009].

Downloaded from Brill.com10/02/2021 07:52:57PM via free access De Villiers et al. — Cussonia and Seemannaralia (Araliaceae) wood 185

Klopper, R.R., C. Chatelain, V. Bänninger, C. Habashi, H.M. Steyn, B.C. de Wet, T.H. Arnold, L. Gautier, G.F. Smith & R. Spichiger. 2006. Checklist of the flowering plants of sub-Saharan Africa. An index of accepted names and synonyms. South African Biodiversity Network Report no. 42. SABONET, Pretoria. 88 pp. Lens, F., P. Gasson, E. Smets & S. Jansen. 2003. Comparative wood anatomy of Epacrids (Styphelioideae, Ericaceae s.l.). Ann. Bot. 91: 835–856. Linder, H.P. 2001. Plant diversity and endemism in Sub-Saharan Tropical Africa. J. Biogeogr. 28: 169–182. Lindorf, H. 1994. Eco-anatomical wood features of species from a very dry tropical forest. IAWA J. 15: 361–376. Lovett, J.C. & T. Pócs. 1993. Assessment of the condition of the Catchment Forest Reserves, a botanical appraisal. Catchment Forestry Report 93.3. Lowry, P.P. II, G.M. Plunkett & J. Wen. 2004. Generic relationships in Araliaceae: looking into the crystal ball. S. Afr. J. Bot. 70: 382–392. Metcalfe, C. R. & L. Chalk. 1950. Anatomy of the dicotyledons. Vol. 2. Clarendon Press, Oxford. Noshiro, S. & P. Baas. 2000. Latitudinal trends in wood anatomy within species and genera: Case study in Cornus s.l. (Cornaceae). Amer. J. Bot. 87: 1495–1506. Olson, D., M.E. Dinerstein, E.D. Wikramanayake, N.D. Burgess, G.V.N. Powell, E.C. Under- wood, J.A. D’Amico, I. Itoua, H.E. Strand, J.C. Morrison, C.J. Loucks, T.F. Allnutt, T.H. Ricketts, Y. Kura, J.F. Lamoreux, W.W. Wettengel, P. Hedao & K.R. Kassem. 2001. Terrestrial ecoregions of the world: a new map of life on earth. BioScience 51: 933–938. Oskolski, A.A. 1994. Wood anatomy of Araliaceae. Komarov Botanical Institute, St. Petersburg [in Russian]. Oskolski, A.A. 1995. Wood anatomy of Schefflera and related taxa (Araliaceae). IAWA J. 16: 191–215. Oskolski, A.A. 1996. A survey of the wood anatomy of the Araliaceae. In: L.A. Donaldson, A.P. Singh, B.G. Butterfield & L.J. Whitehouse (eds.), Recent advances in wood anatomy: 99–119. New Zealand Forest Research Institute, Rotorua. Oskolski, A.A. 2001. Systematic and phylogenetic wood anatomy of Apiales. Edinb. J. Bot. 58: 201–206. Oskolski, A.A., E.L. Kotina, I.V. Fomichev, F. Tronchet & P.P. Lowry II. 2007. Systematic implications of wood and bark anatomy in the Pacific Island genus Meryta (Araliaceae). Bot. J. Linnean Soc. 153: 363–379. Oskolski, A.A. & P.P. Lowry II. 2001. Wood anatomy of Schefflera and related taxa (Araliaceae). II. Systematic wood anatomy of New Caledonian Schefflera. IAWA J. 22: 301–330. Oskolski A.A., D.D. Sokoloff & B.-E. Van Wyk. 2010. False paracarpy in Seemannaralia (Araliaceae): from bilocular ovary to unilocular fruit. Ann. Bot. 106: 29–36. Plunkett, G.M., P.P. Lowry II, D.G. Frodin & J. Wen. 2005. Phylogeny and geography of Schefflera: pervasive polyphyly in the largest genus of Araliaceae. Ann. Missouri Bot. Gard. 92: 202–224. Plunkett, G.M., J. Wen & P.P. Lowry II. 2004. Infrafamilial classifications and characters in Araliaceae: Insights from the phylogenetic analysis of nuclear (ITS) and plastid (trnL-trnF) sequence data. Plant Syst. Evol. 245: 1–39. Record, S.J. & R.W. Hess. 1944. Timbers of the New World. Yale School of Forestry, New Haven. Reyneke, W.F. 1981. ’n Morfologies-taksonomiese studie van die familie Araliaceae in suidelike Afrika. D.Sc. thesis, University of Pretoria, Pretoria. Reyneke, W.F. 1982. Infrageneriese taksonomie van die genus Cussonia Thunb. (Araliaceae). Dendrologiese Tydskrif 2(3&4): 121–131.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access 186 IAWA Journal, Vol. 33 (2), 2012

Rodriguez, R.L. 1957. Systematic anatomical studies on Myrrhidendron and other woody Um- bellales. Univ. Calif. Publ. Bot. 29: 145–318. Strey, R.G. 1973. Notes on the genus Cussonia in South Africa. Bothalia 11: 191–201. Strey, R.G. 1981. Observations on the morphology of the Araliaceae in southern Africa. J. Dendrol. 1(3&4): 66–83. Van den Oever, L., P. Baas & M. Zandee. 1981. Comparative wood anatomy of Symplocos and latitude and altitude of provenance. IAWA Bull. n.s. 2: 3–24. Van der Graaff, N.A. & P. Baas. 1974. Wood anatomical variation in relation to latitude and altitude. Blumea 22: 101–121. Viguier, R. 1906. Recherches anatomiques sur la classification des Araliacées. Ann. Sci. Nat. Bot. Sér. 9, 4: 1–210. Wheeler, E.A. & P. Baas. 1991. A survey of the fossil record for dicotyledonous wood and its significance for evolutionary and ecological wood anatomy. IAWA Bull. n.s. 12: 271–332. Woodward, F.I., M.R. Lomas & C.K. Kelly. 2004. Global climate and the distribution of plant biomes. Philosophical Transactions: Biological Sciences Vol. 359, No. 1450, Plant Phylogeny and the Origin of Major Biomes (Oct. 29, 2004), pp. 1465–1476.

Downloaded from Brill.com10/02/2021 07:52:57PM via free access