Botanical Journal of the Linnean Society, 2009, 159, 268–279. With 19 figures A morphometric analysis of Daniellia (Fabaceae – Caesalpinioideae) MANUEL DE LA ESTRELLA*, CARLOS AEDO and MAURICIO VELAYOS Real Jardín Botánico, CSIC, Plaza de Murillo 2, 28014 Madrid, Spain Received 2 January 2007; accepted for publication 5 June 2008 A multivariate morphometric study of Daniellia, an endemic genus of tropical and subtropical Africa, indicates that nine species may be recognized: D. alsteeniana, D. klainei, D. oblonga, D. ogea, D. oliveri, D. pilosa, D. pynaertii, D. soyauxii and D. thurifera. In our study we found that some characters, not previously studied in detail, were significant in species delimitation: petiole indumentum, petiole width, number and position of glands on the lower surface of the leaflets and presence or absence of glands at the insertion of each pair of leaflets. The rare and scattered material of D. pilosa and D. soyauxii made their classification uncertain, although some qualitative characters support their differentiation. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279. ADDITIONAL KEYWORDS: Africa – Leguminosae – morphometry – savannah – taxonomy – tropical forest. INTRODUCTION Daniellia Benn. (Fabaceae: Caesalpinioideae) is a genus of nine species of medium to large trees found in tropical and subtropical areas of Africa. The different species grow from sea level to 1500 m, from swampy areas to seasonally dry forest (Mackinder, 2005). The highest concentration of species and morphological variation is found in the Guineo-Congolian region. Related genera and species are easily distinguished from Daniellia by its leaf rachis with a minute pair of glands below the insertion of the petiolules (sometimes in each pair of leaflets) visible at low magnification, by its flowers with four imbricate sepals, five petals, 10 stamens, all free or nine shortly connate, and its 1-seeded fruits, with a long funicle and the seed dispersed with one valve of the fruit (Cowan & Polhill, 1981). Bennett (1854) described Daniellia, named after Dr W. F. Daniell, collector of the type specimen of Daniellia thurifera Benn., in Sierra Leone. Hutchinson & Dalziel (1928) published the first work that included *Corresponding author. E-mail: mestrella@rjb.csic.es 268 keys for the eight accepted species known in West Tropical Africa. Baker (1930), in The Leguminosae of Tropical Africa, presented a key for the then 12 accepted species and proposed the subgeneric division currently accepted with some changes in nomenclature; subgenus Daniellia (as subgenus Eudaniellia) includes all known species except D. oliveri (Rolfe) Hutch. & Dalziel, which alone forms subgenus Paradaniellia. Léonard (1950), in his Étude Botanique des Copaliers du Congo Belge, tried to establish a classification of the different species within Daniellia, a genus which is easily recognizable from other Caesalpinioideae, but which has a complicated species delimitation (J. Léonard, pers. comm.). Scattered and poor materials for some species were some of the biggest difficulties encountered Keay (1954), prior to the second edition of Flora of West Tropical Africa Keay, 1958 checked the species of Daniellia, placed five previously accepted species into synonymy and established guidelines for the currently accepted specific subdivision, followed in most respects by Aubréville (1968, 1970). According to Bruneau et al. (2001), Daniellia is a monophyletic group sister to the remainder of Detarieae sensu stricto. During our ongoing taxonomic revision of Daniellia, we realized © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 DANIELLIA: MORPHOMETRIC ANALYSIS that the taxon described by Léonard (1949) as a variety of D. soyauxii (Harms) Rolfe had sufficient consistent differences to be recognized as a different species (Estrella, Aedo & Velayos, 2007). The aim of the present work is to carry out a morphometric study, based mainly on numerical analysis, as a preliminary step in establishing a comprehensive taxonomic treatment as part of a monograph of the genus. MATERIAL AND METHODS This study is based on 90 herbarium specimens (see SUPPORTING INFORMATION, Appendix), belonging to the nine recognized species of Daniellia. Some additional sheets examined during the study were not analysed numerically but contributed to validating our final conclusions (Fig. 1). The following herbaria were consulted: A, AAU, B, BM, BR, BRLU, C, COI, E, G, H, HBG, K, L, LISC, M, MA, MO, NY, P, U, UPS, US, WAG and Z. Digital photographs from FI 269 were also examined. The specimens were selected in order to characterize the morphological and geographical range of each species. Some of them were poorly represented in the herbaria and in such cases we studied all the available samples. They were considered as operational taxonomic units (OTUs). Seventy quantitative characters were recorded and measured using a Mitutotyo CD-15CD digital calliper (see Fig. 2 for some leaf and flower measurements). When possible, at least 10 specimens were measured, a number which has been considered by other taxonomists as representative for recording the morphological range of the species (Rico & Bachman, 2006). Each character was analysed for its mean and median values, range, standard deviation and significance, using the STATISTICA (http://www.statsoft.com) package. To represent the variability of each descriptor within species, box plots containing medians and percentiles were prepared and most informative width/length ratios were also plotted. Box plots showing the variability of the 14 most discriminant Figure 1. Generic distribution of Daniellia in Africa. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 270 M. DE LA ESTRELLA ET AL. Figure 2. Scheme of characters meassured in Daniellia and floral diagrams. A, leaf total length. B, largest leaflet width. C, largest leaflet petiolule length. D, apical leaflet width. E, inflorescence lateral branch length. F, flower pedicel length. G, receptacle width. H, sepal length. I, lateral petal length. J, stamen filament union. K, anther length. L, pod length. M, floral diagram of D. thurifera. N, floral diagram of D. oliveri. characters are shown in Figs. 3–16. The descriptors were: leaf total length (LTL); largest leaflet width (LLW); largest leaflet petiolule length (LLPL); apical leaflet width (ALW); inflorescence lateral branch length (ILBL); flower pedicel length (FPL); receptacle width (RW); sepal length (SL); sepal hairs length (SHL); lateral petal length (LPL); stamen filament union length (SFUL); anther length (AL); ovary stipe length (OSL); pod length (PL). These characters were used to perform the multivariant analyses using the STATISTICA package (see below). The most operative ones should be used to build the species key in our future monograph in conjunction with some of the 74 qualitative characters that we also studied and which could aid in species delimitation (Table 1), but these were not used in statistical analyses. Daniellia oblonga Oliv. was excluded from the multivariant analysis because of the scarcity of available samples. Daniella oliveri has numerous qualitative characters which made its inclusion in a multivariant analysis superfluous (see RESULTS). Principal component analyses (PCA) was carried out using the matrix of the standardized descriptors. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 DANIELLIA: MORPHOMETRIC ANALYSIS 271 Figures 3–16. Box plots representing the variability of the most discriminant quantitative characters in Daniellia. Several combinations were analysed and those plots which showed groups of OTUs in accordance with qualitative data were selected. The correlation matrix was obtained from the initial matrix, eigenvectors were extracted and the OTUs plotted (Fig. 17). Although these analyses are not designed for clustering, a general tendency between main groups could be traced. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 272 M. DE LA ESTRELLA ET AL. Figures 3–16. Continued. The relationships between the different groups were investigated by way of discriminant analysis (DA; Sneath & Sokal, 1973), which requires the a priori assignment of OTUs to groups and allows the determination of whether the recognized groups are statistically definable entities or whether there is too much variation within groups to permit classification. For DA, the raw matrix was obtained, the results © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 DANIELLIA: MORPHOMETRIC ANALYSIS 273 Figures 3–16. Continued. were assorted in to discrete groups and calculations were carried out. RESULTS Daniellia oliveri is the most widespread species in lowland savannah and it is easily distinguished from other species as it has one big petal (lateral, Fig. 2N) and the other four reduced [rarely two big petals (lateral ones) and three reduced]; filaments glabrous (rarely few hairs on basal extreme), free between them; whereas the other species (subgenus Daniellia) have two big petals, one medium sized and two reduced (Fig. 2M); filaments pubescent to villous at least 1/3–2/3 of their length, with nine ± united into a tube and one free (Table 1). This combination of qualitative characters was used by Baker (1930) to establish the subgeneric division of the genus. Daniellia oblonga is probably the most poorly understood species, known only from a few incomplete specimens from Cameroon, Equatorial Guinea and Gabon. The description and species delimitation will be improved when more samples become available. Daniellia oblonga is similar to D. ogea (Harms) Rolfe ex Holland, but the former has a glabrous ovary and stipe (only two or three hairs have been found at the insertion of the stipe with the ovary), whereas in D. ogea the ovary is densely villous to pubescent. The same indumentum differences are seen in the sepals: those from D. oblonga are glabrous, but have a ciliate margin and a tuft of hairs at the top; in contrast, D. ogea has densely pubescent sepals (Table 1). One last difference is the presence of a ring of hairs around the glands at the insertion of each pairs of leaflets in D. ogea, whereas in D. oblonga they are completely glabrous. All other species (subgenus Daniellia except D. oblonga) were included in the PCA and the scatter plot of two principal components axes is presented in Fig. 17, where 75.73% of variance was accounted for by the first two eigenvectors. The third eigenvector only accounted for 5.79% of variance, so it was not represented. The descriptors used for this analysis were LTL, LLW, LLPL, RW, LPL, AL and OSL. The highest loading on the first principal component corresponded to characters RW, LPL and OSL. This component is related to the flower characteristics. The variables with the highest loadings on the second principal component are LTL, LLW, AL and LLPL. These variables are more related to leaf structure. Although we did not find any discrete group (Fig. 17), we observed two main tendencies: first, specimens with bigger flowers and leaves on the left side of the scatter plot, which correspond with D. alsteeniana Duvign. and D. klainei Pierre ex A.Chev. are bordered by some specimens of D. pynaertii De Wild. and D. thurifera; second, specimens with smaller flowers and leaves, D. soyauxii (Harms) Rolfe, D. pilosa (Léonard) Estrella and D. ogea are also associated with some specimens of D. pynaertii and D. thurifera. Two DAs were carried out to study the separation between these groups using the descriptors listed above in MATERIAL AND METHODS. In DA1 we include the four species found in the first group in our PCA, D. alsteeniana, D. klainei, D. pynaertii and D. thurifera. The plot of root 1 against root 2 shows a significant separation between the species OTUs. Daniellia pynaertii is placed near D. thurifera and D. klainei, whereas D. alsteeniana is placed closest to the latter (Fig. 18). The characters contributing most to this separation were LLPL, LL, PL, SFUL and ALW (Table 2). © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 274 © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 Pairs of leaflets Number of main lateral veins Petiole indumentum Rachis glands Leaflet indumentum (midrib on lower surface) Leaflet glands (on lower surface) Inflorescence indumentum Inflorescence lateral branches Pedicel indumentum Flower bud width (mm) Sepal indumentum Stamen filament indumentum Ovary indumentum D. alsteeniana D. klainei D. oblonga D. ogea D. oliveri D. pilosa D. pynaertii D. soyauxii D. thurifera 5–9 4–7 7–8 6–9 6–11 8–10 5–10 7–9 6–9 9–18 16–26 10–16 8–16 9–17 9–16 10–18 8–16 10–25 Glabrous– pubescent With glands basally Tomentose– villous Glabrous Glabrous Glabrous Pubescent Glabrous With glands With glands Pubescent– glabrous With glands Pubescent With glands basally Glabrous Glabrous– pubescent With glands Glabrous Glabrous Pubescent Without glands Glabrous With glands basally Glabrous One gland in the midrib One gland in the midrib One gland in the lamina Pubescent– glabrescent 5–12 Glabrous–slightly pubescent 7–9 Slightly pubescent 7 One or two glands in the lamina Tomentose Pubescent– glabrous (6.7–)7.4– 8.7(-9.5) Pubescent margins and apex Pubescent at least at 1/3 of its length Glabrous to pubescent in margins Glabrous Glabrous– glabrescent 7.3–9.2 (6.9–)7.4– 8.8(-9.6) Glabrous except margins Pubescent at least at 2/3 of its length Glabrous Ciliate margin and apex Pubescent at least at 2/3 of its length Glabrous 5–12 Velvety pubescent (4.5–)5.2– 6.2(-7.5) Pubescent Pubescent at least at 2/3 of its length Densely villous Without glands Midrib pubescent Midrib pubescent Two glands in the lamina One gland in the lamina One gland in the lamina One gland in the lamina Two glands in the lamina Glabrescent– tomentose 6–16 Very long velvety 9–12 Slightly pubescent 4–11 Very long velvety 5–6 Glabrous Glabrous Velvety pubescent 4–5.5 Glabrescent Velvety pubescent 3.2–4.5 Glabrous 6–11 Glabrous, ciliate margin Glabrous Glabrous Velvety pubescent Pubescent at least at 2/3 of its length Densely villous (4–)5.4–8(-9) Pubescent margins and apex Pubescent at least at 2/3 of its length Pubescent– glabrescent Ciliate margin and apex Pubescent at least at 2/3 of its length Few hairs along sutures 6–10 (6.2–)8–9.4 Glabrous, but ciliate margin Pubescent at least at 2/3 of its length Glabrous M. DE LA ESTRELLA ET AL. Table 1. Qualitative and quantitative characters in Daniellia DANIELLIA: MORPHOMETRIC ANALYSIS 275 Figure 17. Plot of first two axes of the principal component analyses (PCA). 1, Daniellia alsteeniana; 2, D. klainei; 3, D. ogea; 4, D. pilosa; 5, D. pynaertii; 6, D. soyauxii; 7, D. thurifera. Figures 18, 19. Plots of the discriminant analyses (DAs) in Daniellia. In DA2, D. ogea, D. pilosa, D. soyauxii, D. thurifera and D. pynaertii were included (Fig. 19). The first four appear clearly separated from each other; D. pynaertii is placed among D. ogea, D. pilosa and D. thurifera. The most discriminant characters were LLW, LLPL, LL, SL and PL (Table 2). The OTUs used in DA1 were correctly classified in all cases, except one of D. alsteeniana that was misclassified as D. klainei and one of D. thurifera that was misclassified as D. pynaertii. In DA2, some predicted classifications were erroneous. One OTU of D. ogea was misclassified as D. pynaertii, one OTU of D. pilosa as D. pynaertii, two OTUs of D. pynaertii as D. ogea and two OTUs of D. thurifera as D. pynaertii (Table 3). DISCUSSION Hutchinson & Dalziel (1928) published the first work that included keys for the eight accepted species found in West Tropical Africa; they used the number of lateral nerves as a significant character in species © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 276 M. DE LA ESTRELLA ET AL. Table 2. Standardized coefficients obtained in discriminant analyses (DAs) for canonical variables DA1 Leaf length Largest leaflet width Largest leaflet petiolule length Apical leaflet width Inflorescence lateral branch lenght Pedicel length Receptacle width Sepal length Sepal hair length Lateral petal length Stamen filament union length Anther length Ovary stipe length Pod length Eigen values Total cumulative proportion DA2 Root 1 Root 2 Root 1 Root 2 -0.64914 -0.24551 1.11395 -0.04235 0.49774 0.36537 0.54468 -0.03427 0.16673 -0.16856 -0.11682 0.37161 0.25591 0.63612 6.34754 0.76864 0.52263 -0.30511 0.00474 0.51413 -0.10209 -0.05699 0.06549 -0.49918 -0.27217 0.05097 -0.71725 -0.27287 0.46709 0.05870 1.05295 0.89615 -0.79948 1.08569 0.90353 -0.62804 -0.00955 0.19058 0.28071 0.61141 -0.00421 -0.52099 0.43291 0.15073 -0.10173 -0.34432 3.54284 0.49164 -0.44039 0.35456 -0.89617 0.56807 0.35357 -0.74487 -0.35115 0.78235 -0.17748 0.20015 0.00132 0.32697 0.32651 0.56595 2.25548 0.80463 Table 3. Correct classifications and values of P obtained in discriminant analyses (DAs) of Daniellia D. D. D. D. D. D. D. D. D. alsteeniana klainei pynaertii thurifera ogea pilosa pynaertii soyauxii thurifera DA No. of OTUs Correct predicted classifications (%) Incorrect predicted classifications P 1 1 1 1 2 2 2 2 2 12 14 11 14 15 5 11 4 14 91.7 100 100 92.9 93.3 80 81.8 100 85.7 1 OTU = D. – – 1 OTU = D. 1 OTU = D. 1 OTU = D. 2 OTU = D. – 2 OTU = D. 0.23529 0.27451 0.21569 0.27451 0.30612 0.10204 0.22449 0.08163 0.28571 klainei pynaertii pynaertii pynaertii ogea pynaertii OTU, operational taxonomic unit. delimitation, which we found in our study to be of little value (Table 1). They also used the number of pairs of leaflets, but as shown in Table 1 these values are normally overlapping between species. The ovary indumentum and shape of leaflets were also used. We found that these characters are quite variable within a species; for example, in D. thurifera the leaflets are reported to have different sizes and shapes on depending of the tree age (Léonard, 1950). Baker (1930) used the same characters for species delimitation. Léonard (1950) used almost the same characters in species delimitation as previous authors, but he also introduced two new characters that have been found relevant to our study: pedicel length (Fig. 8) and the length of petiolules in some species (although he did not state which leaflet was measured). Keay (1958) established the guidelines for the currently accepted species limits. These were followed largely by Aubréville (1968, 1970), who recognized nine species and one variety. Both authors used some characters that we consider too variable between species, such as number of leaflets, leaflet shape, ovary indumentum and flower colour. However, they also used some characters that appear to be significant, e.g. pedicel length (Fig. 8), size and indumentum of sepals (Figs 10, 11) and indumentum on the lower surface of the midrib of the leaflets (Table 1). In our study we found that some characters, not previously studied in detail, were significant in © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 DANIELLIA: MORPHOMETRIC ANALYSIS species delimitation: petiole indumentum, petiole width, number and position of glands on the lower surface of the leaflet and presence or absence of glands at the insertion of each pair of leaflets. Daniellia alsteeniana is the most southern species of the genus. Our analyses show that D. alsteeniana and D. klainei overlap to some extent. The former is easily distinguished from the majority of Daniellia species by its comparatively long lateral branches in the inflorescence (Fig. 7) and the presence of one gland on the lower surface on the midrib 1–3 cm from the petiolule. These two characters are shared with D. klainei. Most specimens of D. alsteeniana have some pubescence on the petiolules, rachis, leaflets and inflorescence, which allows differentiation from D. klainei, a completely glabrous species (Table 1). Additionally, D. alsteeniana has larger sepals and petals (Figs 10, 12); the leaflets are usually larger in D. klainei and in this species we frequently found petiolules thickened by glands. Daniellia alsteeniana filaments are slightly pubescent for 1/3 of their length in contrast with D. klainei, the filaments of which are densely pubescent for 2/3 of their length. In our morphometric study, we found that D. pynaertii is the species with most overlap with the other Daniellia species included in the analyses (Table 3). This is probably a result of the great variability in leaflet texture and size found in D. pynaertii, which nevertheless is usually readily distinguished from other Daniellia species by means of its pubescent midribs (Table 1). Daniellia pynaertii shares this feature with D. pilosa; the former is distinguished by the presence of a pair of glands at the insertion of each pair of leaflets and its usually simple, paniculate inflorescence, whereas, in D. pilosa, rachis glands have not been found and the inflorescence is usually a compound panicle. Material of D. soyauxii in herbaria is limited. This species is poorly known and its classification will remain a problem until more collections are obtained. In our previous work (Estrella et al., 2007) several characters allowed us to present D. pilosa as a distinct species closely related to D. soyauxii. Daniella pilosa has a pubescent midrib on the lower surface (in some specimens a few leaflets were glabrous) and the sepals are velvety pubescent, whereas in D. soyauxii they have only a ciliate margin and a tuft of hairs on the top. Other features concern flower size or number of lateral branches in the inflorescence (9–12 in D. pilosa vs. 5–6 in D. soyauxii) which also distinguish these two species (Table 1). Daniellia thurifera is one of the species with greatest variability in leaflet shape, size and texture. Léonard (1950: 100) said that, according to field observations by Melville, this species had small leaflets in older trees. Daniella thurifera is the only 277 species in which the adaxial petal (medium sized, 12–16 mm long) is ± the same size as the lateral ones (big petals, 12–16 mm long), which, united with the characters presented in Table 1, allows the differentiation of this species. Daniellia ogea is widely dispersed in West and West Central Africa. This is the most problematic species with respect to its delimitation from D. thurifera. Daniellia ogea also shows great variability in leaflet shape, size and texture (as a result of the difference between young and mature leaflets). Daniellia ogea is distinguished from the remaining Daniellia species by its usually glabrous leaflets, a ring of hairs around the glands at the insertion of each pair of leaflets and flowers with densely pubescent sepals and a densely villous ovary. In view of these data, we conclude that nine species should be recognized in the genus Daniellia. We also reflect the need for future explorations to improve the species description and delimitation within this African genus. ACKNOWLEDGEMENTS The authors wish to thank the staff of the cited herbaria for their support in our visit and/or loan of selected material, and also J. L. Castillo and A. Martín for their technical support. We are indebted to J. J. Wieringa for his help and B. Mackinder for her advice and critical review of the manuscript. This work was financed by the Flora of Equatorial Guinea project (CGL 2006-01223). M. de la Estrella was funded by a Universidad Complutense de Madrid pre-doctoral grant and visited BR under FPVI European-funded Integrated Infrastructure Initiative grant SYNTHESYS, BE-TAF 2142 project. REFERENCES Aubréville A. 1968. Flore du Gabon 15: Légumineuses – Caesalpinoidées. Paris: Muséum National d’Histoire Naturelle. Aubréville A. 1970. Flore du Cameroun 9: Légumineuses – Caesalpinoidées. Paris: Muséum National d’Histoire Naturelle. Baker EG. 1930. The Leguminosae of Tropical Africa. Ostend: Unitas Press. Bennett JJ. 1854. Description of the Bungo, or Frankincense Tree of Sierra Leone. Pharmaceutical Journal and Transactions 14: 252–253. Bruneau A, Forest F, Herendeen PS, Klitgaard BB, Lewis GP. 2001. Phylogenetic relationships in the Caesalpinioideae (Leguminosae) as inferred from chloroplast trnL intron sequences. Systematic Botany 26: 487–514. Cowan RS, Polhill RM. 1981. Detarieae. In: Polhill RM, Raven PH, eds. Advances in Legume systematics. Part 1. Kew: Royal Botanic Gardens, 117–134. © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 278 M. DE LA ESTRELLA ET AL. Estrella M, Aedo C, Velayos M. 2007. Daniellia pilosa (J. Léonard) Estrella, comb. & stat. nov. (Leguminosae). Annales Botanici Fennici 44: 149–150. Hutchinson J, Dalziel JM. 1928. Flora of West Tropical Africa 1(2). London: The Crown Agents for the Colonies. Keay RWJ. 1954. Revision of the ‘Flora of West Tropical Africa’ – V. Kew Bulletin 4: 487–492. Keay RWJ. 1958. Flora of West Tropical Africa 1(2), 2nd ed. London: The Crown agents for Oversea Governments and Administrations. Léonard J. 1949. Notulae Systematicae IV (CaesalpinaceaeAmherstieae africanae americanaeque). Bulletin du Jardin Botanique de l’État 19: 383–408. Léonard J, 1950. Étude botanique des copaliers du Congo Belge. Publications de l’ Institut National pour l’Étude Agronnomique du Congo Belge, Série Scientifique 45: 1–158. Mackinder B. 2005. Detarieae. In: Lewis G, Schrire B, Mackinder B & Lock M, eds. Legumes of the World. Kew: Royal Botanic Gardens, 69–109. Rico ML, Bachman S. 2006. A taxonomic revision of Acaciella (Leguminosae, Mimosoideae). Anales del Jardín Botánico de Madrid 63: 189–244. Sneath PH, Sokal RR. 1973. Numerical taxonomy. San Francisco: W.H. Freeman. APPENDIX LIST Species OF SPECIMENS MEASURED AND USED FOR THE MORPHOLOGICAL AND NUMERICAL STUDIES Collections studied Daniellia alsteeniana Duvign. ANGOLA. Lunda Norte: Mungo (Carumbo), confluencia do Luxico com o Luéle, 14.ix.1927, Carrisso & Mendoza 537 (COI); entre Maludi e Chiafua, 11.viii.1965, Mendes dos Santos 1605 (LISC). Lunda Sul: Saurimo, Gossweiler 14088 (BM). Moxico: Moxico, R. Cassai, between R. Cassai and Vila Luso, v.1937, Exell & Mendoza 1478 (COI); Luena (Vila Luzo), Cassai, v.1937, Gossweiler 11282 (LISC). DEMOCRATIC REPUBLIC OF CONGO. Katanga: Mwene-Djungu, Lamanga, iv.1958, François 48 (BR). Kasai-Oriental: Tshibombo, territoire Bakwanga, 16.xi.1956, Liben 1930 (BR), Bakwanda (au N de Katabaie, Territoire Mwene-Ditu), 12.v.1957, Liben 2938 (BR); Mwene-Ditu, savane de Kabwele, 26.v.1951, Simon 30B (BR). Kinshasa: Kahemba-Kwanbo, Mikondo, 30 iv 1955, Devred 1849 (BR). Sud-Kivu: Masisi, Kahemba, ix.1949, Dubois 1491 (BR) GABON. Doudou Mountains, ca. 35 km SW of Doussala, 27.viii.1985, Reitsma 1414 (MA). Daniellia klainei Pierre ex A.Chev. ANGOLA. Cabinda: Maiombe, Buco-Zau, 14.x.1916, Gossweiler 6746 (BM); Buco Zau, entre Chion e Chiaca, 22.ix.1958, Monteiro, Santos & Murta 337 (LISC). DEMOCRATIC REPUBLIC OF CONGO. Bas-Congo: Luki, Mayumbe, 1959, Hombert 568 (BR); Léopoldville, Boma, Luki, 23.x.1948, Madoux 91 (BR); Luki, Mayumbe, 1959, Wagemans 2440 (BR). GABON. 10.xii.1900, Klaine 1925 (BR). Estuaire, environs Libreville, x.1900, Klaine 1440 (G); Mbilagoné (Bilagone), 27.viii.1938, Thomson 12 (K). Moyen-Ogooué: environs d’Adouma, sur l’Orimbo, affluent de l’Ogooué, 29.vii.1912, Fleury & Chevalier 26540 (P). Ngounié: Mission de St. Martin, 7.ix.1938, Walker s.n. (P). Ogooué-Lolo: Makande surroundings, about 65 km SSW of Booué. Makande, 26.ii.1999, Breteler, Caballé, Issembe, Moussavou & 15223 (WAG). Ogooué-Maritime: Nguessi, 24.viii.1918, Le Testu 2283 (BM). Tchibanga, 14.ix.1914, Le Testu 1784 (BM). Daniellia oblonga Oliv. CAMEROON. Yaoundé (Jaúnde), i.1914, Mildbraed 8031 (K); Likomba pflanzung, 15–35 km NE von Victoria, xii.1928, Mildbraed 10759 (A). GABON. 5–30 km NNW of Ndjolé, 23.iv.1992, Breteler, Jongkind & Wieringa 11046 (WAG). EQUATORIAL GUINEA: Bioko (Fernando Po), Barter 2074 (K). Daniellia ogea (Harms) Rolfe ex Holland CAMEROON. Ndiki, xi.1938, Jacques-Félix 2536 (WAG). GHANA. Gold Coast, 27.iii.1909, Imperial Institute s.n. (K). IVORY COAST. Agnéby, Agboville, Aúbreville 2775 (K). LIBERIA. Nimba, New Bapa, 19.ix.1964, Adames 561 (UPS); Gbanka (Banga), 25.x.1926, Linder 1233 (K); Zle (Tiatown), 10 miles E of Tapeta, 8.x.1961, Voorhoeve 517 (WAG), Bong, Loma National Forest near Basiweng, 15.xii.1961, Voorhoeve 721, 735 (WAG). NIGERIA. Without locality, Kenedy 328 (BM). Central Province: Agogidigbo, 11.xii.1907, Unwin 179 (K). Lagos: Ibadan Forest Reserve, 25.xi.1900, Punch 115 (K). Mamu Reserve, 19.ii.1906, Foster 156 (K); Ijebu Lagos, 1895, Millen 191 (K). Benin, Okomu Forest Reserve, 15.i.1948, Brenan, Jones & Richards 8817 (K). SIERRA LEONE. Gola forest, 13.iii.1909, Unwin & Smythe 45 (K). © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279 DANIELLIA: MORPHOMETRIC ANALYSIS 279 APPENDIX Continued Species Collections studied Daniellia oliveri (Rolfe) Hutch. & Dalziel BURKINA FASO. Between Ramba & Laye on Ouaga-Ouhigouya road, 18.iii.1999, van Slageren & Lessina MSSL851 (K). CAMEROON. about 5 km S of Ngaoundéré, 5.xii.1964, Wilde & Wilde-Duyfjes 4618 (WAG). CENTRAL AFRICAN REPUBLIC. Haute-Kotto, Yalinga, 26.i.1923, Le Testu 4510 (BM); Bamingui-Bangoran, 29.i.1980, Spinage 310 (K). GHANA. Mole National Park, 30.xi.1995, Schmidt, Amponsah & Welsing 1870 (UPS). GUINEA BISSAU. Bissau, Brene, 24.i.1945, Espirito Santo 1675 (LISC). NIGERIA. Jamtari, on Jamtari-Kamari old motor road, 22.xii.1954, Latilo & Daramola 28931 (K). SENEGAL. Tambacounda, Parc National du Niokolo Koba, Gué de Damantan-Dabala, km 13, 22.i.1993, Bâ, Madsen, Sambou, Goudiaby, Traoré, Sa, 1246 (AAU); Fatick, Sine Saloum National Park, forêt Fathala, 26.i.1994, Madsen, Goudiaby & Traoré 3052 (AAU). SENEGAMBIA. Without locality, Heudelot 364 (K). WITHOUT LOCALITY. Mann 978 (K). Daniellia pilosa (J.Léonard) Estrella GABON. Nyanga: Mayumba (Mayoumba), 24.v.1915, Le Testu 2062 (LISC). Ogooué-Lolo: région de la ‘forêt des Abeilles’, campement rivière Makandé (2 km en amont de son embouchure dans l’Offoué), 8.iii.1999, Hallé 4610 (WAG); región de Lastoursville, Nzambi, 29.viii.1930, Le Testu 8292 (BR); 28 km NE of Lastoursville, CEB explotation, 23.viii.1992, Wieringa & van de Poll 1462 (WAG); forêt des Abeilles, 13 km SE of the confluence of Gongué-Offoué river, 2.viii.1993, Wilks & Dibata 2703 (MA). Daniellia pynaertii De Wild. DEMOCRATIC REPUBLIC OF CONGO. Équator: Équateur, Forestier Central, Eala, 13.iv.1934, Corbisier Baland 1032 (BR); Eala, 1931, Corbisier Baland 1176 (A); Eala, 1921, Goossens 1639 (K); Eala, 1936, Lemans 220 (BR); Eala, route de Coq, 3.xii.1946, Léonard 1090 (BR); Forestier Central, Dundusana, xii 1913, Mortehan 931 (BR); Eala, route de Coq, 1947, Poucet 1090b (BR); Eala, 20.xi.1906, Pynaert 679 (BR). Kakenge, 4.ix.1958, Dechamps 56 (BR), Bena-Lungu, 20.xi.1958, Dechamps 94 (BR). NIGERIA. Southern Province: Lagos Colony, v.1883, Moloney s.n. (K). Daniellia soyauxii (Harms) Rolfe GABON. Estuaire: Munda, Sibange Farm, 2.vi.1880, Soyaux 90 (K). Moyen-Ogooué: 26 km ENE of Lambaréné, 6 km ENE of Bellevue, 2.iv.1994, Wieringa & Haegens 2614 (WAG). Ogooué-Maritime: Rabi-Kounga, road to Divangui, 23.ix.1992, Wieringa & Epoma 1627 (WAG); Rabi, 11 km on road to Divangui, 29.ix.1994, Wieringa & Nzabi 2800 (WAG). Daniellia thurifera Benn. GUINEA. route Longuery, 1.xii.1905, Caille 14827 (P). GUINEA BISSAU. Tombali, Catió, 24.vi.1945, Espirito Santo 2099 (LISC). IVORY COAST. Lagunes, Forêt d’I.D.E.R.T. Along the Lagune Ebrié, near O.R.S.T.O.M., W of Abidjan, 3.viii.1963, Wilde 625 (WAG). LIBERIA. from vicinity of Firestone Plantation, along Dukwai River, Monrovia, 27.x.1928, Cooper 95 (K); Dukwai river, Monrovia, 5.iv.1929, Cooper 350 (A); Bong, Bong Range, 32 km N of Kakata, 7.i.1962, Voorhoeve 753 (WAG). SIERRA LEONE. Without locality, 12.x.1795, Afzelius s.n. (UPS); 1854, Daniell s.n. (BM); the evergreen ‘rain forest’ of Sierra Leone, Lane Poole s.n. (K). Eastern: Kenema, Nongowa chiefdom, 20.xii.1965, Samai 261 (K). Western Area: road to Kent from main road aroun the Peninsula, 6.xii.1963, Morton 197 (K). © 2009 The Linnean Society of London, Botanical Journal of the Linnean Society, 2009, 159, 268–279