Systematic Botany (2004), 29(3): pp. 569–586 q Copyright 2004 by the American Society of Plant Taxonomists A Morphological Cladistic Analysis of Olacaceae VALÉRY MALÉCOT,1,4 DANIEL L. NICKRENT,2 PIETER BAAS,3 LEEN DANIELLE LOBREAU-CALLEN1 VAN DEN OEVER,3 and Equipe d’Accueil 3496 Classification Evolution et Biosystématique, Laboratoire de Paléobotanique et Paléoécologie, Université Pierre et Marie Curie, 12 rue Cuvier, 75005 Paris, France; 2 Department of Plant Biology, Southern Illinois University, Carbondale, Illinois 62901-6509; 3 National Herbarium, Leiden University, P.O. Box 9514, 2300 RA Leiden, Netherlands; 4 Author for Correspondence. Present address: UMR A 462 SAGAH, Institut National d’Horticulture, 2 rue Le Nôtre, 49045 Angers Cedex 01, France (valery.malecot@inh.fr) 1 Communicating Editor: Gregory M. Plunkett ABSTRACT. A cladistic study based on morphological characters is presented for all 28 genera of Olacaceae as well as 26 representative genera from five other families of Santalales: Loranthaceae, Misodendraceae, Opiliaceae, Santalaceae, and Viscaceae. The data matrix consists of 80 macro-morphological, palynological, and anatomical characters. The phylogenetic trees obtained show a paraphyletic Olacaceae with four main clades. Some of these clades are congruent with previously recognized tribes, but all of subfamilies are para- or polyphyletic. Examination of character transformations confirms several assumptions of evolutionary trends within Olacaceae and Santalales, but others appear to be more complex than expected. Optimization of trophic mode on the consensus tree shows that root hemiparasitism had a single origin in Santalales. Whatever the optimization procedure used, the basal-most clade of Olacaceae consists of 12 genera, among which five are known to be autotrophs, whereas the remaining three clades (15 genera) contain four genera known to be root parasites. Olacaceae (ca. 28 genera and 200 species) constitute a small family of tropical, woody, autotrophic or root hemiparasitic plants (Breteler et al. 1996). Two genera, Erythropalum and Octoknema, are sometimes considered to belong to this family or to represent the monogeneric families Erythropalaceae and Octoknemaceae. Together with Loranthaceae, Misodendraceae, Opiliaceae, Santalaceae (including Eremolepidaceae, Nickrent et al. 1998), and Viscaceae, Olacaceae belong to the order Santalales and are considered the basalmost family in this order (Engler and Gilg 1924; Fagerlind 1948; Kuijt 1968, 1969). Such a basal position is supported by molecular phylogenies for Santalales proposed by Nickrent and Franchina (1990), Nickrent and Duff (1996), Nickrent et al. (1998), Nickrent and Malécot (2001), and in phylogenies of all angiosperm (Savolainen et al. 2000; Soltis et al. 2000). In the first edition of ‘‘Natürlichen Pflanzenfamilien’’ (Engler 1897), the family included three subfamilies and six tribes and since this publication, modification of the infrafamilial classification consists primarily of newly described genera as listed by Sleumer (1984b) and Breteler et al. (1996) (Table 1). The main distinction between the subfamilies Anacolosoideae (formerly known as Dysolacoideae), Olacoideae, and Schoepfioideae, was the presence or absence of ovule integuments and the micropyle position, but according to Sleumer (1984b), the sampling on which these observations were made was too limited to justify the use of these characters at such a level of classification. Moreover, Bouman and Boesewinkel (in Breteler et al. 1996) showed that numerous observations of integument number in the literature were doubtful, and probably erroneous. Tribes as defined by Engler (Anacoloseae, Aptandreae, Couleae, Heisterieae, Olaceae, Ximenieae) are distinguished mainly by the type of endosperm reserve (starch or lipid) and by the degree of fusion of the stamens. But as with ovule integument number, the endosperm reserve character (composed of just two states, either starch or lipid) is too simplistic, plus this feature is not known for about 20% of the genera in the family. Given these facts, previous authors have regularly expressed doubts on the validity of this classification (Gagnepain 1910; Reed 1955; Sleumer 1984b), and some have suggested that Olacaceae be split (Kuijt 1969) or have actually split it into several families (van Tieghem 1896; Gagnepain 1910; Hutchinson 1969; Takhtajan 1997). Specific studies of palynology (Feuer 1977; Lobreau-Callen 1980), leaf anatomy (Baas et al. 1982), and wood anatomy (van den Oever 1984), resulted in groupings that differed considerably from older classifications and highlighted the huge diversity of characters among the various genera of this family. More recently, molecular phylogenetic studies of Santalales, using nuclear small-subunit (SSU) ribosomal DNA (rDNA) sequences alone (Nickrent and Duff 1996) or together with the chloroplastencoded gene rbcL (Nickrent et al. 1998), revealed the polyphyletic nature of the family. These studies sampled relatively few genera, hence no classification was suggested for the entire family. A broader sampling (17 genera of Olacaceae and 58 species of Santalales) was used by Nickrent and Malécot (2001), in a phylogenetic study based on SSU rDNA and rbcL, but incomplete sampling similarly prevented proposing a classification for the entire family. 569 570 [Volume 29 SYSTEMATIC BOTANY TABLE 1. Infrafamilial classification of Olacaceae based on Engler (1897) with modification and additions by Sleumer (1984b) and Breteler et al. (1996). Subfamily Anacolosoideae Airy-Shaw Tribe 1. Couleae Engl. Coula Baill. Maburea Maas Minquartia Aubl. Ochanostachys Mast. Tribe 2. Heisterieae Dumort. Chaunochiton Benth. Heisteria Jacq. Tribe 3. Anacoloseae Engl. Anacolosa (Blume) Blume Brachynema Benth. Cathedra Miers Diogoa Exell and Mendonca Engomegoma Breteler Phanerodiscus Cavaco Scorodocarpus Becc. Strombosia Blume Strombosiopsis Engl. Tetrastylidium Engl. Tribe 4. Ximenieae Engl. Ximenia L. Subfamily Olacoideae Sond. Tribe 5. Olaceae Horan. Curupira G.A.Black Douradoa Sleumer Dulacia Vell. Malania Chun and S.K.Lee Olax L. Ptychopetalum Benth. Tribe 6. Aptandreae Engl. Aptandra Miers Harmandia Pierre ex Baill. Ongokea Pierre Subfamily Schoepfioideae Engl. Tribe 7. Schoepfieae Miers Schoepfia Schreb. Dubious affinities Erythropalum Blume Octoknema Pierre Distinguishing genera in the family is relatively straightforward owing to the monographic studies of Sleumer for Asia (Sleumer 1980), Malaysia (Sleumer 1984a) and Latin America (Sleumer 1984b), as well as those of African Olacaceae by Michaud (1962, 1966). Among the 28 genera,13 are monospecific, and only six consist of more than ten species. Moreover, four genera have been described within the last 20 years: Malania from China (Lee 1980), Douradoa from Brazil (Sleumer 1984b), Maburea from Guyana (Maas et al. 1992), and Engomegoma from Gabon (Breteler et al. 1996). Some genera show disjunct distributions (afroamerican, afro-asiatic or asiatic-american) and consideration of morphological, anatomical, and chemical characters suggest an origin of the family during the Lower Cretaceous (Ling 1982) or Upper Cretaceous (Sleumer 1984b; Haron and Ping 1997). This biogeographic hypothesis is reinforced by the presence of fossil pollen from the Maestrichtian assigned to the pollen form genus Anacolosidites, which is similar to pollen of the current genera Anacolosa, Cathedra, and Phanerodiscus (Krutzsch 1989; Askin 1989; Malécot 2002). The goal of this work is to use a cladistic approach to examine the phylogeny of Olacaceae and putatively related taxa using morphological, palynological, and anatomical (leaf and wood) characters. To take into account the risk associated with the study of a probable paraphyletic or polyphyletic group (Nickrent and Duff 1996; Nickrent et al. 1998; Nickrent and Malécot 2001), members of other families of Santalales (Loranthaceae, Misodendraceae, Opiliaceae, Santalaceae, and Viscaceae) have been included. A revised classification of the family will not be presented here but will be published after inclusion of results from molecular phylogenetic analyses (see Malécot 2002; Nickrent and Malécot, in prep.). A second objective is to study evolutionary trends in selected characters of the family and order inferred in previously published works. MATERIALS AND METHODS Taxa. The studied taxa include all genera that have at one time or another been classified in Olacaceae (Table 1), with the exception of Brachynema which is only doubtfully assigned to Santalales. Placentation in Santalales consists of a few ovules inserted on the top of a free-central column (or as a ‘mamelon’ in Opiliaceae, Loranthaceae, and Viscaceae). Placentation in Brachynema is axile in a 4–5 loculate ovary, i.e. clearly not santalaceous. In addition, other morphological characters such as glandular toothed leaves, apiculate stamen connectives, and convolute aestivation, are unknown in Santalales. For these reasons, Brachynema was excluded from this analysis and may be best placed among Ericales (Malécot 2002). In addition to Olacaceae, six genera of Opiliaceae, one genus of Misodendraceae, four genera of Loranthaceae, 11 genera of Santalaceae, and four genera of Viscaceae were included. For a few genera where some characters show variability between species (e.g., Heisteria, Phoradendron, Viscum), two species have been included as terminal taxa for this cladistic analysis. When character state variability (polymorphism) existed among species of a genus, both states were recorded in the data matrix. Recent molecular phylogenetic analyses (Savolainen et al. 2000; Soltis et al. 2000) have suggested various putative sister clades for Santalales, in particular Saxifragales and Caryophyllales. Therefore, Daphniphyllum, a primitive member of Saxifragales, and Rhabdodendron from Caryophyllales were chosen as outgroups. Characters. Macromorphological characters for all taxa were scored from specimens housed in various herbaria (e.g., K, KUN, L, MO, P), or taken from previously published descriptions (e.g., Malania, Douradoa, and Curupira). Voucher information is included in Lobreau-Callen (1980, 1982), Baas et al. (1982), and Kooek-Noorman and van Rijckevorsel (1983). Palynological characters were taken from Feuer (1977), Lobreau-Callen (1980, 1982, in Sleumer 1984b), as well as new observations by Lobreau-Callen. Leaf anatomical characters for Olacaceae have been obtained from Baas et al. (1982) and Baas and Koel (1983), whereas those for Opiliaceae were taken from Koek-Noorman and van Rijckevorsel (1983). Wood anatomical characters were obtained from publications by van den Oever (1984) and van den Oever et al. (1993) for Olacaceae, and from Koek-Noorman and van Rijckevorsel (1983) for Opiliaceae. Most of the data (morphology, palynology, leaf and wood anatomy) for Engomegoma and Maburea were taken from the very comprehensive papers by Breteler et al. (1996) and Maas et al. 2004] 571 MALÉCOT ET AL.: OLACACEAE PHYLOGENY (1992), respectively. Data for other Santalales and the outgroup genera were obtained from herbarium samples or were gleaned from the literature (particularly from Metcalfe 1935; Reed 1955; Metcalfe and Chalk 1957; Puff and Weber 1976; Bhatnagar and Garg 1977; Carlquist 1982, 1985, 2001; Zavada and Dilcher 1986; Norverto 1993; Huang 1996). The 80 characters used for this cladistic analysis are described in Appendix 1. Data matrices are available on TreeBASE (study accession number S1022, matrix accession number M1730). Parasitism was not included as a character in this cladistic analysis because the available data for Olacaceae were too poor and because we wished to study this character by optimizing it on the consensus tree obtained. Data Analysis. Cladistic analysis of the data was conducted using PAUP* (Swofford 2000). Heuristic searches for most parsimonious trees used 100 random addition sequences and TBR branch swapping. Characters were unweighted and character state transformations of binary and multistate characters were unordered. Node specific support was assessed using bootstrap analysis (Felsenstein 1985) with 1,000 replications and ‘‘Bremer support’’ (Bremer 1988, 1994; Donoghue et al. 1992) was calculated using Autodecay (Eriksson 1999). MacClade (Maddison and Maddison 1997) and Winclada (Nixon 1999) were used to optimize character state distributions. RESULTS A total of 80 parsimony informative characters was identified for use in this cladistic analysis of Santalales. The data matrix consisted of 58 binary and 22 multistate characters. Of these, 32 were derived from anatomy, 14 from palynology, and 34 from floral and vegetative morphology, with 9.1% of the cells coded as missing. The data matrix used for this cladistic analysis is presented in Appendix 2. Maximum parsimony analysis resulted in 33 most parsimonious trees (length 5 457 steps, consistency index (CI) 5 0.33, retention index (RI) 5 0.69). The strict consensus of those trees is presented in Fig. 1. Relatively few nodes showed bootstrap (BS) values above 50%. Bremer support values above one are commonly associated with nodes that had bootstrap values above 50%. Santalales are monophyletic (BS 90%), with a clade of 13 Olacaceae genera sister to the remaining Santalales. As a family, Olacaceae are paraphyletic and four lineages (Clades 1–4 on Fig. 1) can be identified. Clade 1 is composed of 13 genera including members of tribe Couleae (Coula, Ochanostachys, Minquartia, Maburea), Heisterieae (Heisteria), Anacoloseae sensu Breteler et al. (1996) (Diogoa, Engomegoma, Strombosia, Strombosiopsis, Tetrastylidium), as well as two problematic genera Erythropalum and Octoknema. Clade 2 consists of four genera: Ximenia (tribe Ximenieae), Curupira, Douradoa, and Malania, the latter three genera doubtfully placed in tribe Olaceae by Sleumer (1984b). Clade 3 is composed of ten olacaceous genera from tribes Olaceae, Aptandreae, Anacoloseae (in part) and Chaunochiton. This clade is sister to one with Schoepfia and the remaining Santalales families: Misodendraceae, Opiliaceae, Loranthaceae, Santalaceae, and Viscaceae. Clade 4 consists solely of the genus Schoepfia which appears as sister to the other families of Santalales. A clade consisting of Misodendraceae plus Op- iliaceae is sister to a clade that includes Loranthaceae, a paraphyletic Santalaceae, and Viscaceae. Opiliaceae are monophyletic (BS 64%), as are Loranthaceae (BS 85%). DISCUSSION In general, the relationships obtained from this cladistic analysis of Santalales are quite similar to those obtained following analyses of SSU rDNA and rbcL sequences (Nickrent and Malécot 2001), despite differences in taxon sampling. This study also demonstrates the paraphyly of Olacaceae as first suggested by molecular analyses (Nickrent and Duff 1996; Nickrent et al. 1998; Nickrent and Malécot 2001). As traditionally circumscribed, Olacaceae is paraphyletic, but four groups (clades) can be recognized (Fig. 1). A Templeton test (as implemented in PAUP*) in which the monophyly of the four clades of Olacaceae was constrained shows that this hypothesis was significantly worse than the most parsimonious trees. Comparison to Previous Subfamilial and Tribal Delimitations. Neither of the two multigeneric subfamilies recognized by Sleumer (1984b) and Breteler et al. (1996), Anacolosoideae and Olacoideae (Table 1), are monophyletic and indeed Olacoideae are polyphyletic. Three genera of tribe Olaceae, Curupira, Douradoa, and Malania, are placed in a clade with Ximenia of tribe Ximenieae (subfamily Anacolosoideae). The remaining six genera of subfamily Olacoideae (Aptandra, Dulacia, Harmandia, Olax, Ongokea, Ptychopetalum), together with Chaunochiton (tribe Heisterieae) and Anacolosa, Cathedra, and Phanerodiscus of tribe Anacoloseae, constitute Clade 3. As with Olacoideae, subfamily Anacolosoideae is thus polyphyletic and is not the basalmost subfamily as suggested by Ling (1982). Most members of Anacolosoideae form a clade at the base of Santalales, including the two genera with doubtful affinities: Erythropalum and Octoknema. Only some of the tribes defined by Engler (1897) appear to be monophyletic, such as Couleae (Coula, Minquartia, Ochanostachys), Aptandreae (Aptandraceae of Takhtajan 5 Aptandra, Ongokea, Harmandia), and Olaceae (Olax, Ptychopetalum, Dulacia). However, other genera that were included within this last tribe by Sleumer (1984b), e.g., Malania, Curupira, and Douradoa, or Maburea, placed in tribe Couleae by Maas et al. (1992), are excluded. Tribes Ximenieae (Ximenia) and Schoepfieae (Schoepfia) were both monogeneric, but Ximenieae are included here in a larger clade with Malania, Douradoa, and Curupira. Tribe Heisterieae, composed of Heisteria and Chaunochiton, is polyphyletic, as already suggested by Reed (1955). Members of tribe Anacoloseae are scattered in various positions on the tree with members in both Clades 1 and 3 (Fig. 1). Baas et al. (1982), Lobreau-Callen (1980), and van den Oever (1984) have already highlighted this heterogeneity, 572 SYSTEMATIC BOTANY [Volume 29 FIG. 1. Phylogenetic relationships of Olacaceae and Santalales based on anatomical and morphological characters. Branch support is given above the branches in the following format: decay index / bootstrap percentage (1000 replications). Circled numbers indicate major clades of Olacaceae (see text). For nutritional mode character state transformations (using either ACCTRAN or DELTRAN optimizations): black 5 hemiparasitism absent; grey 5 root hemiparasitism; black dashed 5 aerial hemiparasitism. For trophic conditions: ? 5 trophic status unknown; 0 5 autotrophs; R 5 root hemiparasites; A 5 aerial hemiparasites. The subfamilial and tribal names for Olacaceae correspond to the infrafamilial classification shown in Table 1. which is confirmed here. Relationships within and among the clades as shown on Fig. 1 will be discussed in the following section. Clades of Olacaceae. The basalmost member of Clade 1 is Erythropalum, a genus with doubtful olacaceous affinity based on anatomy (Reed 1955; Baas et al. 1982) but whose embryology clearly place it in Santalales (Fagerlind 1946, 1948). Its wood structure (van den Oever, unpublished) and pollen morphology suggest an affinity with Santalaceae (Lobreau-Callen 1982). Ling (1982) considered this genus as the most derived subfamily of Olacaceae (Erythropaloideae). The remaining genera of Clade 1 form a polytomy of three poorly to well-supported clades. One of these consists of Octoknema, which is frequently placed in its own family, Octoknemaceae. Fagerlind (1948) showed that the genus has typical santalaceous placentation and proposed a relationship with tribe Couleae. From its wood anatomy, Reed (1955) linked it to Heisteria and Couleae. The same relationship has been proposed 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY by Feuer (1977) on the basis of palynology, whereas, according to Lobreau-Callen (1982), its pollen is similar to Opiliaceae. The second clade of the polytomy received strong support (BS 85%, Bremer support 5 3) and is composed of Diogoa, Engomegoma, Tetrastylidium, and Strombosiopsis (hereafter the Diogoa clade). Unambiguous changes for this clade are the occurrence of epidermal druse crystals and a constricted furrow in the pollen ectoaperture. In most previous classifications (Reed 1955; Feuer 1977; Lobreau-Callen 1980; Baas et al. 1982), a group was recognized that included not only these genera (minus Engomegoma, then undescribed) but also Strombosia and several other genera in tribe Anacoloseae. In another case (van den Oever 1984), two groups were distinguished, one with Diogoa and Strombosiopsis, the other with Tetrastylidium and Strombosia. Molecular analyses of Nickrent and Malécot (2001) showed that Diogoa and Strombosiopsis are associated with Strombosia and Scorodocarpus, with strong support (BS . 90%). This relationship is not supported here, but nodes between these genera are also not well supported. Breteler et al. (1996), who described the genus Engomegoma, indicated rather ambiguous affinities within Olacaceae near Diogoa, Strombosia, Strombosiopsis, and Tetrastylidium. This relationship is supported here. The third subclade of the polytomy of Clade 1 is composed of seven genera. For the genera or clades included in this polytomy, different patterns of relationships have been proposed based on different organs or on molecular data. For Maburea, wood anatomical data link it to Heisteria (Maas et al. 1992), leaf anatomy to Scorodocarpus and Brachynema (Maas et al. 1992) and the pollen apertures (analyzed by Hiepko and Lobreau-Callen, in Maas et al. 1992) to Minquartia, Coula, and Heisteria. On the basis of both palynology and anatomy, Strombosia is placed near Diogoa and Scorodocarpus (Feuer 1977; Lobreau-Callen 1980; Baas et al. 1982; van den Oever 1984). Molecular analyses by Nickrent and Malécot (2001) indicated a strong relationship (BS . 90%) among Diogoa, Strombosia, Strombosiopsis, and Tetrastylidium. According to wood anatomy, Scorodocarpus is closest to Couleae (van den Oever 1984). Leaf anatomy links it to members of the Diogoa clade and Strombosia, but also to Heisteria (Baas et al. 1982). Palynology suggests a relationship with Strombosia (Feuer 1977; Lobreau-Callen 1980). The genus Heisteria has been considered the most primitive member of Olacaceae by Reed (1955), mainly on the basis of wood anatomical features. In his wood anatomical study of the family, van den Oever (1984) placed this genus in the same group as Brachynema, likely because of shared ancestral features. The rather homogeneous tribe Couleae (Coula, Minquartia, Ochanostachys) has long been recognized (van Tieghem 1899a, 1899b; Stauffer 1961). Fagerlind (1948) suggested a basal po- 573 sition of this tribe within the family, a relationship that was supported by Michaud (1962, 1966) but not the present study. Couleae are unique among all dicotyledons because they possess both laticifer and secretory cavities in their leaves. Unambiguous changes in this tribe also include the acquisition of libriform fibers (also in Scorodocarpus), two or more cycles of stamens, and thyrsoid inflorescences. The composition of Clade 2 (Ximenia, Malania, Curupira, and Douradoa; BS 5 63%, Bremer support 5 2) is exactly the same as proposed by van den Oever (1984). According to Sleumer (1984b), Ximenia constitutes a monogeneric tribe (Ximenieae) within subfamily Anacolosoideae, whereas Curupira, Malania, and Douradoa were doubtfully placed within tribe Olaceae of subfamily Olacoideae. An affinity between Malania and Ximenia corresponds with the proposal made by Lee (1980) in his generic description, and was found by Nickrent and Malécot (2001) in their molecular study. This association was supported by Ling (1982), but was never pointed out in the later synthesis by Sleumer (1984b) and Breteler et al. (1996). Curupira was placed close to Ptychopetalum because of the absence of staminodes or an accrescent calyx (Black and Murça Pires 1948). In the present study, the clade composed of Ximenia and Malania as well as the two Brazilian genera Curupira and Douradoa can be attributed to their shared umbellate inflorescences, flowers with two stamen cycles, lipid reserves in the fruit, and several wood anatomical features. Clade 3 received less support (,50% BS, Bremer support 5 1) but is here composed of ten genera, arranged in three major subclades. Three members of one subclade (Olax, Ptychopetalum, and Dulacia) have long been recognized for their morphological (Sleumer 1984b), anatomical (Reed 1955; Baas et al. 1982; van den Oever 1984), and palynological (Feuer 1977; Lobreau-Callen 1980) homogeneity. This group was strongly supported (100% BS) in the molecular analysis of Nickrent and Malécot (2001). According to van Tieghem (1896), this group is equivalent to family Olacaceae sensu stricto. Unambiguous character changes for this clade include heterogeneous type I rays, a pollen foot-layer in masses, and the loss of endexine in the mesocolpium of the pollen. Bootstrap support for a clade composed of the three genera coinciding with tribe Aptandreae (Harmandia, Aptandra, and Ongokea) was strong (BS 95%, Bremer support 5 6). The homogeneity of this group has long been known, and indeed van Tieghem (1896), Pierre (1897), and Gagnepain (1910) all suggested that these genera be recognized at the family level (Aptandraceae). Several morphological features link these three genera, particularly the very peculiar fused androecium and the anthers that dehisce by means of flaps. Molecular data (Nickrent and Malécot 2001) provide strong support 574 SYSTEMATIC BOTANY (BS 5 82%) for a relationship between two representatives of this group (Aptandra and Ongokea). The third component of Clade 3 (Chaunochiton, Phanerodiscus, Cathedra, and Anacolosa) received BS support below 50% but Bremer support 5 3. A clade minus Chaunochiton received higher support (BS 5 77%, Bremer support 5 5), thus in agreement with the assignment of the three genera to tribe Anacoloseae sensu stricto as recognized by Reed (1955) and Baas et al. (1982). Floral features of this smaller clade include petals inserted on the top of a disk and poricidal stamens. Chaunochiton has been more commonly associated with Aptandreae (‘‘Chaunochitoneae’’ of Fagerlind 1948), based on ovary morphology. The molecular analysis by Nickrent and Malécot (2001) also supported a clade composed of Chaunochiton and Aptandreae (BS 5 81%). Relationships between Anacoloseae s.s. and Aptandreae have been proposed by Baas et al. (1982) on the basis of leaf anatomy. Despite an early association with tribe Heisterieae (Engler 1897), affinities of Chaunochiton are clearly not with Heisteria as stated by Fagerlind (1948) and subsequent workers (Feuer 1977; LobreauCallen 1980; Baas et al. 1982; van den Oever 1984, 1990). Engler (1894) placed the genus Schoepfia in its own tribe in Olacaceae (Schoepfieae) and later (Engler 1897) proposed a monogeneric subfamily (Schoepfioideae). This genus was previously assigned to Loranthaceae (de Candolle 1830). Van Tieghem (1896) and Gagnepain (1910) both proposed raising this genus to the family level (Schoepfiaceae). According to wood anatomy, Reed (1955) noted that this genus was the most specialized among all Olacaceae, noting further that its pollen was comparable to that of some Santalaceae. In molecular phylogenetic analyses (Nickrent and Duff 1996; Nickrent et al. 1998; Nickrent and Malécot 2001), Schoepfia was separated from other Olacaceae and placed in a clade together with Misodendrum and Loranthaceae. In our study, this genus occurs in Clade 4 and is sister to all the remaining Santalales families. Compared to the traditional infrafamilial classification of Olacaceae, these results suggest that relatively few characters can be used to distinguish subfamilies Olacoideae and Anacolosoideae, such as integument number (1 or 2 for Anacolosoideae, 0 for Olacoideae) and an accrescent calyx (absent in Anacolosoideae, mostly present in Olacoideae) (Sleumer 1984b). Moreover, some genera have been assigned to subfamily based on erroneous data (e.g., Bouman and Boeswinkel in Breteler et al. 1996, comments on ovule integument number). Aside from those genera described after 1948, tribal characteristics appear more pertinent, except for tribes Anacoloseae and Heisterieae, in which genera were grouped based on non-homologous characters. For the genera described during the second part of the 20th century, some tribal assignments may have [Volume 29 been the consequence of the absence of data (e.g. Curupira placed in Olaceae because it lacks an accrescent calyx and staminodes). Moreover, some genera were classified based on data not initially used for delimitating the tribes, such as Douradoa and Malania, which were placed in Olaceae because they are morphologically close to Curupira, and Maburea, which waas placed in Couleae based on flower morphology. Other Santalalean Families. Of the major clades sister to Schoepfia, the first is Opiliaceae, a monophyletic and moderately well-supported family (BS 64%, Bremer support 5 2). This family was recognized as distinct from Olacaceae by Valeton (1886) and it is commonly divided into two tribes: Agonandreae and Opilieae. According to Sleumer (1935), the first tribe contains Agonandra and Gjellerupia, whereas Hiepko (1984) includes only Agonandra therein, placing Gjellerupia with Urobotrya and Lepionurus. Anatomical data presented by Koek-Noorman and van Rijckevorsel (1983) place these latter three genera together in one group and Champereia, Melientha, Opilia, and Rhopalopilia (and possibly Cansjera) in another group; Agonandra showed uncertain affinities. Molecular results support a monophyletic Opiliaceae but the topology differs from the above treatments. According to the combined analysis of SSU rDNA and rbcL by Nickrent and Malécot (2001), Lepionurus is basalmost in the family followed by Agonandra and the four other genera in the family (Champereia, Cansjera, Opilia, and Pentarhopalopilia). Although additional sampling is called for, the presence of Agonandra near the base for the family is supported by this morphological cladistic analysis because this genus has the highest number of plesiomorphies. Dividing the family into two tribes seems premature at this time, even though six characters state transformations (characters 2, 42, 50, 75, 76, 78) separate Agonandra from the remaining Opiliaceae. Misodendrum of the monogeneric mistletoe family Misodendraceae is placed here as sister to Opiliaceae. The genus has a large number of autapomorphies, such as wood features (peculiar/highly modified rays, fibers and parenchyma), or palynological characters (polyporate, echinulae) (Skottsberg 1914; Metcalfe and Chalk 1957; Feuer 1981; Carlquist 1985). This position as sister to Opiliaceae is not stable as indicated by a BS value of less than 50% (Bremer support 5 1) for this clade. Molecular data (Nickrent and Soltis 1995; Nickrent and Duff 1996; Nickrent et al. 1998; Nickrent and Malécot 2001) always link Misodendrum and Schoepfia, a relationship also recovered in some morphological cladistic analyses (data not shown) where fewer genera of Santalaceae and Loranthaceae were used. The four genera sampled from Loranthaceae (Dendrophthoe, Ligaria, Tripodanthus, and Tupeia) are monophyletic and this well-supported (BS 85%, Bremer sup- 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY port 5 2) clade is sister to Santalaceae and Viscaceae. Santalaceae appear paraphyletic in this study. One clade consists of members of tribe Thesieae of Pilger (Acanthosyris, Scleropyrum, Pyrularia) and is relatively similar in composition to Santalaceae clade c of Nickrent and Malécot (2001). Okoubaka, also considered to belong to this tribe based on habit, fruit morphology, wood anatomy, and palynology (Stauffer 1957; Lobreau-Callen 1982; Hallé 1987), is here placed as sister to Viscaceae, an anomalous position supported by a single character (37, long ray cells). The remaining Santalaceae form a poorly-supported clade that includes members of Santalaceae clades d, e and f of Nickrent and Malécot (2001), mainly from tribes Santaleae and Antholobeae. The four genera of Viscaceae are monophyletic, but with low BS support Bremer support 5 1, thus few intergeneric relationships are resolved. Character Transformations. This discussion of character transformations will concentrate mostly on Olacaceae, although some characters more particularly concern the other families. The unambiguous character optimizations are presented in Fig. 2. Alternate (character 1) and developed (character 3) leaves are plesiomorphic within Santalales, and opposite and/or squamate leaves appear independently derived several times in non-olacaceous families (Loranthaceae, Misodendraceae, Santalaceae, Viscaceae). Such evolution appears consistent with morphological reduction associated with the parasitic habit. The presence of tomentum, whether on twigs (character 2) or leaves (characters 7, 8), is a derived feature. Dendritic leaf hairs is a synapomorphy of tribe Couleae. Regarding venation, the variability of the states for the outgroups prevent any decision regarding plesiomorphic and apomorphic features as secondary (character 4) and tertiary (character 5) venation distinguish Clade 1, (with camptodromous leaves and reticulate tertiary venation) from the remaining clades (Clades 2, 3, 4). Mesh shape (character 6) is primarily polygonal and flabellate meshes are a synapomorphy for Coula and Ochanostachys. Evolution of leaf anatomical features commonly follows the assumptions of Baas et al. (1982), with derived features being sclerification or lignification of epidermal cells (character 9), lignification of guard cells (character 10), wide lumina of stomata (character 11), and the presence of cuticular ledges on guard cells (character 12). Occurrence of paracytic (character 13) or cyclocytic (character 15) stomata separate Clade 1 (cyclocytic stomata) from the remaining Santalales (with paracytic stomata). However, according to Baas et al. (1982), the presence of paracytic stomata may be plesiomorphic. Regarding stomatal features, Couleae strangely show a reversal compared with other members of Clade 1. Contrary to Baas et al. (1982), the presence of laticifers (character 17) is a derived feature, 575 whereas schizogeneous cavities (character 16) are a synapomorphy of Couleae, in agreement with them. The presence of druses in epidermal cells (character 18), silicified walls of mesophyll and epidermal cells (character 19), mesophyll sclereids (character 20), and leaf cystoliths (character 21) are all apomorphic within Santalales. The presence of mechanical tissues of the vascular system, e.g. sclerenchyma fibers (character 22) and astrosclereids (character 23) in the petiole and median vein, appear to be plesiomorphic within Santalales, in agreement with Baas et al. (1982); the absence of such mechanical tissue is a specialization achieved through reduction. General trends for wood anatomical features mostly conform to the hypotheses put forth by Bailey (1957), but occurrence of the derived state does not imply a close relationship. For example, ray organization (character 36) follows Bailey’s hypothesis, with heterocellular rays being plesiomorphic, whereas homocellular rays are apomorphic. Clade 1 consists only of taxa with heterocellular rays with several rows of erect or square cells. Clades 2 and 3 consist of taxa with heterocellular rays with a single row of erect or square cells. For the remaining santalalean families the picture is not so clear and several reversals from homocellular to heterocellular rays (whether with one or several rows of erect or square cells) have occurred in the Opiliaceae and Exocarpos/Santalum clades. Presence of wood cystoliths (character 38) is a possible synapomorphy for two genera of Opiliaceae (Champereia and Opilia), and the presence of silica bodies (character 39) is a synapomorphy for the three genera of tribe Olaceae (Ptychopetalum, Olax, Dulacia). Tetramerous (character 44) flowers with oval buds (character 43), partially fused sepals (character 45), accrescent calices (character 46), fused petals (character 47) and pubescent petals (character 48) occur commonly in a few unrelated clades, thus suggesting that parallel morphological evolution may be linked with pollination syndromes. The primitive androecial condition appears to be five (character 49), epipetalous (character 51) stamens organized in a single (character 50) cycle. Such an organization contradicts the hypothesis of Michaud (1962, 1966) who proposed that the two-cycled androecium as occurs in Heisteria and tribe Couleae is primitive. Duplication of the androecium occurs independently in Clade 1 (Couleae, Heisteria, Scorodocarpus, Maburea), Clade 2 (all genera except Douradoa), and Clade 3 (tribe Olaceae). The presence of staminodia (character 52), of fused filaments (character 53), or of a modified anther connective (character 54) are all synapomorphies of small clades, commonly regarded as tribes within Clades 1 or 3. Anthers are primitively basifixed (character 55) with a transformation to dorsifixed in Clade 1, and to oblique in Clade 3. Anther dehiscence mode (character 56) and 576 SYSTEMATIC BOTANY [Volume 29 FIG. 2. Optimization of morphological characters on cladogram given in Fig. 1. Open boxes indicate homoplastic characters, filled boxes non-homoplastic characters. orientation (character 57) appear to be linked, with longitudinal and introrse (or lateral) anther dehiscence as plesiomorphies. In contrast, extrorse dehiscence is almost always associated with porous or valvate dehiscence, as occurs in Chaunochiton and tribes Aptandreae and Anacoloseae of Clade 3. The accrescent disk (character 58) is a synapomorphy of Anacolosa and Cathedra within Clade 3. The glandular tissue (character 59) is primitively located between the stamens and ovary. Its location between the stamens and petals is a synapomorphy for members of Aptandreae. The absence of glandular tissue occurs in several independent lineages. A short cylindrical style (character 60) seems to be plesiomorphic within Santalales. Style shape modifications are common in Clade 1, but are uninformative. Conversely, 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY long conical styles appear as a clear synapormophy for Clade 2 and a long cylindrical style as a synapomorphy for Clade 3. The inferior ovary (character 62) is a synapomorphy of non-olacaceous families and Schoepfia, but within Olacaceae, semi-inferior ovaries arose several times independently. The ovary locule number of two (character 63) is plesiomorphic within Clade 1, and increases (to four or five) in the Diogoa clade. For the remaining Santalales (Clades 2, 3 etc.), the plesiomorphic state is a single locule, with an increase of up to two in Anacoloseae, or to three in Olax. Such patterns contradict common assumptions on the prevalence of reduction in locule number evolution in Santalales (Sleumer 1984b), but new observations may be necessary for some taxa as some reports date to Valeton (1886). Breviaxial pollen (character 67) is a synapomorphy for Clade 3, whereas heteropolarity (character 68) has occurred independently in various groups. A concave mesocolpium (character 69) and concave apocolpium (character 70) are synapomorphies for members of Aptandreae, the former occurring also in a few unrelated genera. Apertures are primitively three or four and zonoaperturate (character 71); duplication of the number occurs independently in Ptychopetalum and in Anacoloseae s.s. (Anacolosa, Cathedra, Phanerodiscus). Round or elliptic ectoapertures (character 73) are a synapomorphy for Clade 3 (with a reversal for Chaunochiton), as stated by Lobreau-Callen (1980). The occurrence of granules on the apertural membrane (character 73) does not constitute a synapomorphy for any group of Olacaceae, but granules in the endoaperture (character 76) are synapomorphies of Clade 3. The endoaperture and ectoaperture of identical size (character 75) are mainly synapomorphies of Clade 3, with few independent occurrences in other Santalales. Mesocolpium ornamentation (character 77) is primitively smooth or microperforate, but derived states occur several times independently. The presence of granules in the infratectum (character 78) is a synapomorphy for members of Couleae and Opiliaceae, but this feature appears to be independently derived in these two groups. Formation of a columella in the infratectum (character 79) also appears independently in a few clades (Aptandreae and Anacoloseae s.s.; Opiliaceae p.p.) suggesting these features are not always homologous. A smooth foot layer (character 80) is plesiomorphic in Santalales, but contrary to the assumptions of Lobreau-Callen (1980), an irregular foot layer surface is not homologous in all Olacaceae. An irregular foot-layer is a synapomorphy for two groups, Couleae (Coula, Ochanostachys) and Olaceae (Ptychopetalum, Dulacia, Olax). Parasitism. The occurrence of root hemiparasitic and autotrophic taxa in Olacaceae has been known for decades, but the evolution of this ecological specialization remains poorly understood. Only four genera of 577 Olacaceae are known to contain species that possess haustoria on their roots: Ximenia (Heckel 1899), Olax (Barber 1907), Schoepfia (Piehl 1973; Werth et al. 1979), and Ptychopetalum (Anselmino 1932). The absence of haustoria has been documented only for Heisteria (Kuijt 1969), Ochanostachys, Strombosia, Scorodocarpus, and Erythropalum (Ping 1997). No data are available for the remaining 19 genera in the family. Haustoria have been recorded for eight of the ten genera of Opiliaceae. Gjellerupia and Pentarhopalopilia are the only genera for which parasitism has not yet been confirmed, but all Opiliaceae are assumed to be root parasites (Kubat 1987). All Santalaceae are considered to be either aerial or root hemiparasites, including Okoubaka, a 40 meter high African tree whose parasitic mode was only recently reported (Veenendaal et al. 1996). Hemiparasitism has been documented in Acanthosyris (Barroso 1969), Colpoon (Visser 1981), Pyrularia (Leopold and Muller 1983), Scleropyrum (van Rheede 1688; Arnott 1836; Nicolson et al. 1988) and for Santalum, Thesium, Exocarpos, Osyris, and Buckleya (see references in Kuijt 1969). Among the 11 genera of Santalaceae used in this study, documentation of parasitism is lacking only for Rhoiacarpos. The three genera of Loranthaceae included here are all aerial hemiparasites (mistletoes), as is Misodendrum and all Viscaceae. On the basis of molecular results, Nickrent (2002) showed that aerial hemiparasitism arose independently in Santalales at least five times, but the evolution of root hemiparasitism was not addressed. Positive documentation of hemiparasitism and nonparasitism have been plotted on the tree obtained from this cladistic analysis (Fig. 1). When those taxa with an unknown parasitic state are coded with ‘‘?’’, root hemiparasitism arises only once in Santalales, in the common ancestor of Clade 1 (Olacaceae) and its sister group (remaining Santalales). This result is obtained using both ACCTRAN (accelerated transformation) and DELTRAN (delayed transformation) optimization strategies. Aerial hemiparasitism appears in three independent lineages among the taxa studied here: Misodendraceae, Loranthaceae and Viscaceae (an aerial parasite from Santalaceae was not included). Moreover, an additional appearance of root parasitism can be assumed at the base of Loranthaceae (Atkinsonia, Gaiadendron, and Nuytsia, none of which were sampled here). If taxa coded as unknown are considered root parasites or autotrophs (‘‘R’’ or ‘‘0’’), ancestral states for Santalales are ambiguous, thus leaving the possibility open that root parasitism arose numerous times. From the topology of the tree shown in Fig. 1, it may be assumed that, in addition to the five olacaceous genera documented as lacking haustoria, eight other genera are autotrophs: Coula, Diogoa, Engomegoma, Minquartia, Maburea, Octoknema, Strombosiopsis, Tetrastylidium. Furthermore, eleven genera not known to be 578 SYSTEMATIC BOTANY parasites may, based on our optimizations, be root hemiparasites: Anacolosa, Aptandra, Cathedra, Chaunochiton, Curupira, Douradoa, Dulacia, Harmandia, Malania, Ongokea and Phanerodiscus. If the tree topology is correct, and any of these genera are later found to be autotrophic, such cases would represent reversals from the parasitic mode. Such reversals, however, have never been reported for any parasitic plant and seem unlikely given the selective advantage conferred by this trophic mode. Perspectives. Several previous works based on morphological or molecular data suggested a paraphyletic or polyphyletic Olacaceae. The present study reaches the same conclusion, but the positions of genera such as Maburea, Strombosia, and Scorodocarpus remain uncertain when compared with available molecular data. This cladistic analysis demonstrates that the infrafamilial classification of the family is in need of revision. The existing classification is based on the work of Engler from the 19th century (Engler 1894, 1897). Subsequent work during the last century mainly provided amendments, such as the description of new taxa, but the overall structure of the classification remained relatively unchanged. In the present analysis, the subfamilies Olacoideae and particularly Anacolosoideae are polyphyletic and only a few of the named tribes appeared monophyletic. A revised classification of the family will not be presented here but will be published after consideration of results from molecular phylogenetic analyses (Nickrent and Malécot, in prep.). Regarding parasitism, if states are inferred for unknown taxa, this study suggests a single origin of root hemiparasitism within Santalales, whereas aerial hemiparasitism appeared independently in at least three of the lineages sampled here. The evolution of hemiparasitism appears to have been a major event in Olacaceae, one that can be used to distinguish two groups in the family. Character optimization for parasitism allows a prediction, for the first time, of the nutritional mode for nineteen genera of Olacaceae. ACKNOWLEDGEMENTS. We thanks the curators and staff of K, KUN, L, MO and P for making specimens available for study through visit or loans. The first author (V. Malécot) is indebted to Jean Broutin and staff of the Equipe Classification Evolution et Biosystématique at University Pierre & Marie Curie, Paris, for support and advice during the preparation of his PhD dissertation from which much of this work is derived. We are also grateful to P. Herendeen (as editor and reviewer), G.M. Plunkett (as editor) and an anonymous reviewer for helpful comments on the manuscript. LITERATURE CITED ANSELMINO, E. 1932. Die Stammpflanzen der Drogue Muira-puama. Notizblatt des Botanischen Gartens und Museums zu BerlinDahlem 11: 623–626. ARGAWAL, S. 1963. Morphological and embryological studies in the family Olacaceae–I. Olax L. Phytomorphology 13: 183–196. ARNOTT, G. A. W. 1836. Observations on some new or obscure [Volume 29 species of plants no. II. Magazine of Zoology and Botany 2: 543– 552. ASKIN, R. A. 1989. Endemism and heterochroneity in the Late Cretaceous (Campanian) to Paleocene palynofloras of Seymour island, Antarctica: implications for origins, dispersal and palaeoclimates of southern floras. Pp. 107–119 in Origins and evolution of the antarctic biota, ed J. A. Crame. London: The Geological Society. BAAS, P. 1986. Terminology of imperforate tracheary elements – in defense of libriform fibres with minutely bordered pits. IAWA Bulletin, new series 7: 82–86 ——— and R. KOOL. 1983. Comparative leaf anatomy of Heisteria (Olacaceae). Blumea 28: 367–388. ———, E. VAN OOSTERHOUD, and C. J. L. SCHOLTES. 1982. Leaf anatomy and classification of the Olacaceae, Octoknema, and Erythropalum. Allertonia 3: 155–210. BAILEY, C. D. 1957. The potentialities and limitations of wood anatomy in the study of the phylogeny and classification of angiosperms. Journal of the Arnold Arboretum 38: 243–254. BARBER, C. A. 1907. Studies in root parasitism. The haustorium of Olax scandens. Memoirs of the Department of Agriculture in India, Botanical Series 2: 1–47. BARROSO, G. M. 1969. Acanthosyris Paulo-alvinii – una nova espécie de Santalaceae. Anais do Congresso da Sociedade de Botânica do Brasil (XIX Congresso Nacional de Botânica Fortaleza, 21–29 january 1968): 107–109. BHATNAGAR, A. K. and M. GARG. 1977. Affinities of Daphniphyllum—palynological approach. Phytomorphology 27: 92–97. BLACK, G. A. and J. MURÇA PIRES. 1948. Dois gêneros novos Curupira e Froesia, cinco espécies novas, uma nova combinaçâo, chaves e observaçôes sobre plantas da regiâo amazônica. Boletim Técnico do Instituto Agronômica do Norte 15: 16–19. BREMER, K. 1988. The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803. ———. 1994. Branch support and tree stability. Cladistics 10: 295– 304. BRETELER, J. F., P. BAAS, F. D. BOESEWINKEL, F. BOUMAN, and D. LOBREAU-CALLEN. 1996. Engomegoma Breteler (Olacaceae) a new monotypic genus from Gabon. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 118: 113– 132. CAPURON, R. 1968. Matériaux pour l’etude de la flore forestière de Madagascar: Olacaceae, Opiliaceae et Santalaceae arbustives ou arborescentes de Madagascar. Antananarivo: Centre Technique Forestier Tropical section de Madagascar. CARLQUIST, S. 1982. Wood anatomy of Daphniphyllaceae: ecological and phylogenetic considerations, review of Pittosporalean affinities. Brittonia 34: 252–266. ———. 1985. Wood and stem anatomy of Misodendraceae: systematic and ecological conclusions. Brittonia 37: 58–75. ———. 2001. Wood and stem anatomy of Rhabdodendraceae is consistent with placement in Caryophyllales sensu lato. IAWA Journal 22: 171–181. DE CANDOLLE, A. P. 1830. Prodromus systematis naturalis regni vegetabilis, sive enumeratio contracta ordinum, generum, specierumque plantarum hucusque cognitarum, juxta methodi naturalis normas digesta. Paris, Strasbourg, London: Treuttel et Würtz. DONOGHUE, M. J., R. G. OLMSTEAD, J. F. SMITH, and J. D. PALMER. 1992. Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345. EICHLER, A. W. 1878. Blütendiagramme. Leipzig: Wilhelm Engelmann. ENGLER, A. 1894. Olacaceae. Pp. 231–242 in Natürlichen Pflanzenfamilien, ed. A. Engler. Berlin: Borntraeger. ———. 1897. Olacaceae. Pp. 144–149 in Natürlichen Pflanzenfamilien Nachträge zu Teil II-IV, eds A. Engler and K. Prantl. Berlin: Borntraeger. 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY ——— and E. GILG. 1924. Syllabus der Pflanzenfamilien. Berlin: Borntraeger. ERIKSSON, T. 1999. Autodecay, version 4.0. Stockholm: Bergius Foundation, Royal Swedish Academy of Sciences. FAGERLIND, F. 1946. Gynöceummorphologische, Embryologie und systematische Stellung der Gattung Erythropalum. Svensk Botanisk Tidskrift Utgifven af Svenska Botaniska Föreningen 40: 9– 14. ———. 1947. Gynöceummorphologische und embryologische Studien in der Familie Olacaceae. Botaniska Notiser 1947: 207–230. ———. 1948. Beitrage zur Kenntnis der Gynaceummorphologie und Phylogenie der Santalales-Familien. Svensk Botanisk Tidskrift Utgifven af Svenska Botaniska Föreningen 42: 195–229. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. FEUER, S. M. 1977. Pollen morphology and evolution in the Santalales s.s. a parasitic order of flowering plants. Ph.D. Dissertation, University of Massachusetts, Amherst. ———. 1981. Pollen morphology and relationships of the Misodendraceae (Santalales). Nordic Journal of Botany 1: 731–734 GAGNEPAIN, F. 1910. Comment faut- il comprendre la famille des Olacacées? Bulletin de la Société Botanique de France 57: 373– 380. GEORGE, A. S. 1984. Olacaceae. Pp. 13–25 in Flora of Australia, Volume 22, Rhizophorales to Celastrales, ed. A. S. George. Canberra: Australian Government Publishing Service. HALLÉ, N. 1987. Okoubaka Pellegrin & Normand est vraiment un genre de Santalaceae. Bulletin du Muséum National d’Histoire Naturelle. Section B, Adansonia: Botanique Phytochimie 9: 355– 363. HARON, N. W. and S. T. PING. 1997. Distribution and taxonomic significance of flavonoids in the Olacaceae and Icacinaceae. Biochemical Systematics and Ecology 25: 265–263. HECKEL, E. 1899. Sur le processus germinatif dans la graine de Ximenia americana L. et sur la nature des écailles radiciformes propres à cette espèce. Revue Générale de Botanique 11: 401– 408. HEGNAUER, R. 1966. Chemotaxonomie des Pflanzen, Volume 4. Basel, Stuttgart: Birchauser Verlag. ———. 1969. Chemotaxonomie des Pflanzen, Volume 5. Basel, Stuttgart: Birchauser Verlag. ———. 1973. Chemotaxonomie des Pflanzen, Volume 6. Basel, Stuttgart: Birchauser Verlag. HERENDEEN, P. S. and R. B. MILLER. 2000. Utility of wood anatomical characters in cladistic analyses. IAWA Journal 21: 247–276. HIEPKO, P. 1984. Opiliaceae. Pp. 31–52 in Flora Malesiana, ed. C. G. G. J. van Steenis. La Hague, Boston, London: Martinus Nijhoff pub. ———. 2000. Opiliaceae. Pp. 1–53 in Flora Neotropica Monograph 82, ed. Organization for Flora Neotropica. Bronx: New York Botanical Garden. HUANG, T.-C. 1996. Notes on taxonomy and pollen of malesian Daphniphylum (Daphniphyllaceae). Blumea 41: 231–244. HUTCHINSON, J. 1969. Evolution and phylogeny of flowering plants. London: Academic press. KOEK-NOORMAN, J. and P. VAN RIJCKEVORSEL. 1983. Wood and leaf anatomy of Opiliaceae. Willdenowia 13: 147–174. KRIBS, D. A. 1935. Salient lines of structural specialization in the wood rays of dicotyledons. Botanical Gazette 96: 547–557. ———. 1968. Commercial foreign woods on the american market. New York: Dover publication. KRUTZSCH, W. 1989. Paleogeography and historical phytogeography (paleochorology) in the Neophyticum. Plant Systematics and Evolution 162: 5–61. KUBAT, R. 1987. Report of the first investigations of parasitism in Opiliaceae (Santalales). Pp. 489–492 in Parasitic flowering plants. Proceedings of the 4th ISPFP, eds H. C. Weber and W. Forstreuter. Marburg: Kempkes Offset. 579 KUIJT, J. 1968. Mutual affinities of santalalean families. Brittonia 20: 136–147. ———. 1969. The biology of parasitic flowering plants. Berkeley: University of California Press. LEE, S.-K. 1980. Malania, a new genus of oil- yelding plant. Bulletin of the Botanical Laboratory of the North-East Forestry Institute 6: 67–71. LEOPOLD, D. J. and R. N. MULLER. 1983. Hosts of Pyrularia pubera Michx. (Santalaceae) in the field and in culture. Castanea 48: 138–145. LING, Y.-R. 1982. A revision and addition of Olacaceae from China and a primitive discussion for the taxonomy and floristics of the family. Bulletin of Botanical Research 2: 7–36. LINNÉ, C. 1753. Species plantarum. Holmiae: Impensis Laurentii Salvii. LOBREAU-CALLEN, D. 1980. Caractères comparés du pollen des Icacinaceae et des Olacaceae. Adansonia, série 2 20: 29–89. ———. 1982. Structures et affinités polliniques des Cardiopterygaceae, Dipentodontaceae, Erythropalaceae et Octoknemataceae. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 103: 371–412. MAAS, P. J. M., P. BAAS, F. D. BOESEWINKEL, P. HIEPKO, D. LOBREAUCALLEN, L. VAN DER OEVER, and B. J. H. TER WELLE. 1992. The identity of ‘‘unknown Z’’: Maburea Maas, a new genus of Olacaceae in Guyana. Botanische Jahrbücher für Systematik, Pflanzengeschichte und Pflanzengeographie 114: 275–291. MADDISON, W. P. and D. R. MADDISON. 1997. MacClade: analysis of phylogeny and character evolution, version 4.0. Sunderland: Sinauer Associates. MALÉCOT, V. 2002. Histoire, classification et phylogénie des Olacaceae R. Brown (Santalales). Unpublished Ph.D. Dissertation, Université Paris VI, Paris. ———, G. E. SCHATZ, and J. BOSSER. (2003). Révision synoptique du genre Phanerodiscus à Madagascar. Adansonia, série 3 25: 119–128. METCALFE, C. R. 1935. The structure of some sandalwoods and their substitutes and some other little known scented woods. Kew Bulletin of Miscellaneous Information 1935: 165–195. ——— and L. CHALK (eds). 1957. Anatomy of the dicotyledons. Oxford: Clarendon Press. MICHAUD, M. 1962. Contribution à l’étude des Olacacées d’Afrique tropicale. DESS Dissertation, Université Paris VI, Paris. ———. 1966. Contribution à l’étude des Olacacées d’Afrique tropicale. Mémoires de l’Institut Fondamental d’Afrique Noire 75: 157– 290. MOUTON, J.-A. 1972. Contribution de la morphologie foliaire à la phylogénie des angiospermes. Pp. 199–209 in Comptes rendus du 93e Congrès national des sociétés savantes, Tours 1968, Section sciences, Volume III Biologie Végétale. Paris: Gauthier-Villars et Cie. NICKRENT, D. L. 2002. Mistletoe phylogenetics: current relationships gained from analysis of DNA sequences. Pp. 48–57 in Proceedings of the Forty-eighth Western International Forest Disease Work Conference, ed. P. Angwin. Redding: USDA Forest Service. ——— and R. J. DUFF. 1996. Molecular studies of parasitic plants using ribosomal RNA. Pp. 28–52 in Advances in parasitic plant research, eds M. T. Moreno, J. I. Cubero, D. Berner, D. Joel, L. J. Musselman, and C. Parker. Cordoba: Junta de Andalucia, Direccion General de Investigation Agraria. ———, R. J. DUFF, A. E. COLWELL, A. D. WOLFE, N. D. YOUNG, K. E. STEINER, and C. DEPAMPHILIS. 1998. Molecular phylogenetic and evolutionary studies of parasitic plants. Pp. 211– 241 in Molecular systematics of plants II. DNA sequencing, eds D. Soltis, P. Soltis, and J. Doyle. Boston: Kluwer Academic Publishers. ——— and C. R. FRANCHINA. 1990. Phylogenetic relationships of 580 [Volume 29 SYSTEMATIC BOTANY the Santalales and relatives. Journal of Molecular Evolution 31: 294–301. ——— and V. MALÉCOT. 2001. A molecular phylogeny of Santalales. Pp. 69–74 in 7th International parasitic weed symposium, eds A. Fer, P. Thalouarn, D. M. Joel, L. J. Musselman, C. Parker and J. A. C. Verkleij. Nantes: Faculté des sciences de Nantes. ——— and D. A. SOLTIS. 1995. A comparison of angiosperm phylogenies from nuclear 18S rDNA and rbcL sequences. Annals of the Missouri Botanical Garden 82: 208–234. NICOLSON, D. H., C. R. SURESH, and K. S. MANILAL. 1988. An interpretation of van Rheede’s ‘Hortus Malabaricus’. Königstein: Koeltz Scientific Books. NIXON, K. C. 1999. Winclada (beta) version 0.9.9. Ithaca: Published by the author. NORVERTO, C. A. 1993. Wood anatomy and relationships of Santalaceae. I. Acanthosyris, Jodina, and Myoschilos. Aliso 13: 499– 511. PATIL, D. A. and R. M. PAI. 1984. The floral anatomy of Olax scandens Roxb. (Olacaceae). Indian Botanical Reporter 3: 10–14. PIEHL, M. A. 1973. Root parasitism in Schoepfia chrysophylloides (A. Rich.) Planchon and Ximenia americana L. (Santalales, Olacaceae). ASB Bulletin 20: 75. PIERRE, J. B. L. 1897. Sur le genre Ongokea et la famille des Aptandracées. Bulletin Mensuel de la Société Linnéenne de Paris, série 2 2: 1313–1315. PING, S. T. 1997. Root hemi-parasitism in malayan Olacaceae. Gardens’ Bulletin Singapore 49: 7–13. PUFF, C. and A. WEBER. 1976. Contributions to the morphology, anatomy, and karyology of Rhabdodendron, and a reconsideration of the systematic position of Rhabdodendronaceae. Plant Systematics and Evolution 125: 195–222. PUNT, W., S. BLACKMORE, S. NILSSON, and A. LE THOMAS. 1994. Glossary of pollen and spore terminology. Utrecht: LPP foundation. REED, C. F. 1955. The comparative morphology of the Olacaceae, Opiliaceae and Octoknemaceae. Memorias de Sociedade Broteriana 10: 29–79. SAVOLAINEN, V., M. W. CHASE, S. B. HOOT, C. M. MORTON, D. E. SOLTIS, C. BAYER, M. F. FAY, A. Y. DE BRUIJN, S. SULLIVAN, and Y.-L. QIU. 2000. Phylogenetics of flowering plants based on combined analysis of plastid atpB and rbcL gene sequences. Systematic Biology 49: 306–362. SKOTTSBERG, C. 1914. Myzodendraceae. Pp. 1–16 in Das Pflanzenreich, ed. A. Engler. Berlin: Borntraeger. SLEUMER, H. O. 1935. Olacaceae. Pp. 1–32 in Die natürlichen Pflanzenfamilien, eds A. Engler and K. Prantl. Berlin: Borntraeger. ———. 1980. A taxonomic account of the Olacaceae of Asia, Melanesia, and Adjacent Areas. Blumea 26: 145–168. ———. 1984a. Olacaceae. Pp. 1–29 in Flora Malesiana ser. 1, ed. C. G. G. J. van Steenis. La Hague, Boston, London: Martinus Nijhoff. ———. 1984b. Olacaceae. Pp. 1–160 in Flora Neotropica Monograph 38, ed. Organization for Flora Neotropica. Bronx: New York Botanical Garden. SOLTIS, D. E., P. S. SOLTIS, M. W. CHASE, M. E. MORT, D. C. ALBACH, M. ZANIS, V. SAVOLAINEN, W. H. HAHN, S. B. HOOT, M. F. FAY, M. AXTELL, S. M. SWENSEN, L. M. PRINCE, W. J. KRESS, K. C. NIXON, and J. S. FARRIS. 2000. Angiosperm phylogeny inferred from 18S rDNA, rbcL, and atpB sequences. Botanical Journal of the Linnean Society 133: 381–461. STAUFFER, H. U. 1957. Zur stellung der gattung Okoubaka Pellegrin et Normand. Berichte der Schweizerischen Botanischen Gesellschaft/Bulletin de la Société Botanique Suisse 67: 422–427. ———. 1961. Santalales studien VIII – Zur morphologie und taxonomie des Olacaceae – Tribus Couleae. Vierteljarsschrift der Naturforschenden Gesellschaft in Zürich 106: 412–418. SWOFFORD, D. L. 2000. PAUP*: Phylogenetic analysis using parsi- mony (* and other methods), version 4.0b. Sunderland: Sinauer Associates. TAKHTAJAN, A. J. 1997. Diversity and classification of flowering plants. New York: Columbia University Press. VALETON, T. 1886. Critisch Oversicht der Olacineae B. et H. Doctor in Plant- en Dierkunde Dissertation, Rijks Universiteit, Gröningen. VAN DEN OEVER, L. 1984. Comparative wood anatomy of the Olacaceae. Pp. 177–178 in Proceedings of the Pacific Regional Wood Anatomy Conference, ed. S. Sudo. Tsukuba: Tsukuba University. ———. 1990. Phylogenetic wood anatomy of the Olacaceae and related families. IAWA Bulletin, new series 11: 133. ———, B. J. H. TER WELLE, and J. KOEK-NOORMAN. 1993. Olacaceae woods and timber. Pp. 44–64 in Flora of Guiana 14, ed. A. R. A. Gorts-van Rijn. Koenigstein: Koeltz scientific books. VAN RHEEDE, H. A. 1688. Hortus indicus malabaricus. Amsterdam: Joannis van Someren. VAN TIEGHEM, P. 1896. Sur les phanérogames a ovule sans nucelle, formant le groupe des innucellées ou santalinées. Bulletin de la Société Botanique de France 43: 543–577. ———. 1899a. Sur les Coulacées. Journal de Botanique 13: 69–79. ———. 1899b. Deux genres nouveaux pour la famille des Coulacées. Bulletin du Muséum d’Histoire Naturelle 5: 97–100. ———. 1905. Sur le genre Octocnème considéré comme type d’une famille distincte, les Octocnémacées. Journal de Botanique 19: 45–58. VEENENDAAL, E. M., I. K. ABEBRESE, M. F. WALSH, and M. D. SWAINE. 1996. Root hemiparasitism in a west african rainforest tree Okoubaka aubrevillei (Santalaceae). New Phytologist 134: 487–493. VISSER, J. 1981. South African parasitic flowering plants. Capetown: Juta and Co., Lmt. WERTH, C. M., W. V. BAIRD, and L. J. MUSSELMAN. 1979. Root parasitism in Schoepfia Schreb. (Olacaceae). Biotropica 11: 140– 143. ZAVADA, M. S. and D. L. DILCHER. 1986. Comparative pollen morphology and its relationship to phylogeny of pollen in the Hamamelidae. Annals of the Missouri Botanical Garden 73: 348– 381. APPENDIX 1 List of characters used in this cladistic analysis of Olacaceae with discussion of character state assignments. Stem 1. Phyllotaxy: alternate (0): opposite (1). All species of Olacaceae have alternate phyllotaxy, but various Santalaceae and Loranthaceae have opposite leaves. In rare cases (e.g., Okoubaka), phyllotaxy varies near the proximal ends of shoots within an individual, thus such species were coded as polymorphic. 2. Twig tomentum: none (0): simple hairs (1): stellate hairs (2). According to Reed (1955), Coula and Minquartia, as well as Octoknema, could be allied by the presence of branched hairs on the young branches. Diverse Opiliaceae also have pubescent young branches, albeit with simple hairs. We interpreted the pubescence on young growth among Opiliaceae and Couleae as non-homologous and coded the states accordingly. Leaf The majority of foliar anatomical characters result from work of Baas et al. (1982) who studied infrageneric variation in leaf anatomical characteristics of Olacaceae. They also proposed assumptions as for the homology and the taxonomic utility of these attributes while taking into account the other families of Santalales. For Opiliaceae, Koek-Norman and van Rijckvorsel (1983) did similar work. 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY 3. Leaf size: developed (0): squamate (1). 4. Secondary venation: camptodromous (0): brochidodromous (1). The venation of the leaves of Olacaceae shows some diversity. The secondary venation is always pinnate, either of a festooned brochidodromous, or a festooned eucamptodromous type (Mouton 1972). In the genera with three clear basal veins (Maburea, Curupira, etc.), the secondary veins are widely spaced at the base, and much closer towards the top, and the type of secondary venation remains identifiable starting from these final veins. 5. Tertiary venation: plagiodromous (parallel) (0): reticulate (1). The tertiary venation is of a simple plagiodromous type (parallel veins between them, at oblique angles to the secondary veins), or reticulate and irregular. 6. Quarternary venation (areolation): meshes flabellate (0): meshes polygonal (1). The meshes (areoleae) are either flabellate in a direction parallel with the tertiary veins or polygonal. 7. Leaf dentritic hairs: absent (0); present (1). Baas et al. (1982) recognized three types of hairs: dendritic, unicellular, and uniseriate. The simultaneous presence of hairs of two types in certain genera (e.g. Ochanostachys) led us to independently code for each hair type rather than to use only one character with four states (hairs absent, dendritic, unicellular, uniseriate). Moreover, according to Baas et al. (1982), the presence of uniseriate hairs does not make it possible to characterize any genus of Olacaceae, and this character varies between leaves of the same individual. Consequently, only two characters will be defined, one for dentritic hairs and a second one (character 8) for unicellular hairs. 8. Leaf unicellular hairs: absent (0); present (1). 9. Epidermal cell lignification: none (0); weakly lignified (1); lignified (2). The secondary walls of the ‘‘ordinary’’ epidermal cells (excluding guard cells and stomatic cells) can be sclerified—a very rare phenomenon in angiosperms (Baas et al. 1982). Such cells are found on the entire abaxial surface of the leaves of the genera of tribe Couleae, as well as occasionally on the leaves of Heisteria p.p., Harmandia, and Aptandra. For these last three genera, a very weak sclerification is observed, thus the phenomenon may not be homologous with that observed in Couleae. Thus weakly lignified epidermal cells was treated as distinct. 10. Guard cell lignification: not lignified (0); lignified (1). According to Baas et al. (1982), there is no link between the sclerification of the epidermal cells and the lignification of the guard cells. The majority (more than 60%) of the guard cells are lignified in the genera Aptandra, Cathedra, Maburea, and Phanerodiscus. Taking into account the small percentage (lower than 10%) of lignified guard cells mentioned for Heisteria p.p., Chaunochiton, Harmandia and Dulacia, these genera were scored with a question mark. 11. Lumina of guard cells: narrow (0); wide (1). Seen in cross section, the guard cells of the stomata are of very different form and size in Olacaceae. Two extreme types were defined by Baas et al. (1982): 1) guard cells with a broad lumen and equally thick walls above and below and 2) guard cells with a narrow lumen with the inner wall (facing the mesophyll) much thicker than the outer wall. They found that the guard cells with broad lumens are constantly associated with marked cuticular ledges, whereas the guard cells with narrow lumens generally occur (but not always) in the species with paracytic stomates. However, exceptions to this relationship exist. Thus, two characters will be defined, one corresponding to the size of the lumen of the guard cells, the other to the presence of cuticular ledges (character 12). 12. Cuticular ledges of guard cells: inconspicuous (0); thick (1). 13. Paracytic stomata: absent (0); present (1). Olacaceae show a great diversity of stomate types (Baas et al. 1982), and different types can be present on the same individual. Thus, in an effort to avoid using a polymorphic characters, each stomatal type was coded independently (characters 13–15) except anomocytic stomata which occur in all genera. 14. Cyclocytic stomata: absent (0); present (1). 15. Anisocytic stomata: absent (0); present (1). 581 16. Schizogenous cavities in leaf. absent (0); present (1). 17. Laticiferous channel in leaf: absent (0); present (1). The laticifers present in the leaves of certain species of Olacaceae are not ramified, and have a diameter ranging between 8 and 45 mm (Baas et al. 1982). Their presence has been used to delimit tribe Couleae (Sleumer 1935), however, they are also present in species of Heisteria, Chaunochiton, Harmandia, and Aptandra. 18. Druses in epidermal cells: absent (0); present (1). Druses are considered by Baas et al. (1982) to be an apomorphy for the genera Diogoa, Strombosia, Strombosiopsis, and Tetrastylidium. Other types of crystals listed by Baas et al. (1982) were recorded in only some species of the same genus, thus only druses will be used in this phylogenetic analysis. 19. Silicified walls of mesophyll and epidermal cells: absent or extremely infrequent (0); present (1). In various genera of Olacaceae and the majority of the other families of Santalales the walls of certain cells of the mesophyll and epidermis are silicified, more particularly those associated with the stomatal complex or vein endings. Baas et al. (1982) consider that this is a derived condition in Olacaceae. 20. Mesophyll sclereids: absent (0); present (1). Idioblastic sclereids are morphologically variable in the species of Heisteria (Baas et al. 1982; Baas and Kool 1983), but not so in the other genera of the family. Only the presence of this type of mesophyll sclereid will be coded here. 21. Leaf cystoliths: absent (0); present (1). Mesophyll cystoliths are observed only for Opiliaceae and could constitute an apomorphy for this family (Koek-Noorman and van Rijckevorsel 1983). 22. Petiole and median vein sclerenchyma fibers: absent (0); present (1). Baas et al. (1982) considered that the presence of sclerenchyma fibers is strongly related to the type of vascularization. The presence of such fibers is common in angiosperms, but the linkage with the type of vascularization is not so strong in other families. On the other hand, the absence of sclerenchymatous fibers (more exactly their substitution by collenchyma) as supportive tissue in the median vein is not very common in angiosperms (Baas et al. 1982). Moreover, this absence from the median vein does not automatically correspond to an absence near the other veins. Baas et al. (1982) identified a particular type of nonlignified fiber near the margin of the leaf in Strombosia, Strombosiopsis, and Diogoa. The difficulty in observing this type of fiber makes it impossible to determine if its absence in the other families of Santalales is real or simply a lack of observation. Thus, only the presence or absence of sclerenchyma fibers in the petiole and median vein will be coded, but not in other veins. 23. Petiole and median vein astrosclereids: absent (0); present (1). With the exception of Phanerodiscus, these sclereids are present in the majority of Anacoloseae s.l., such as Heisteria, Maburea, Engomegoma, Curupira, and Douradoa, and sporadically in Malania and Ximenia (Baas et al. 1982). These sclereids are morphologically close (groups of isodiametric and slightly ramified sclereids) and have an identical position in the central tissue of the petiole and the median vein (Baas et al. 1982). Vascularization of nodes and petiole 24. Vascularization of the nodes: unilacunar (0); trilacunar (1); pentalacunar or more (2). The genera for which information is available derives from Reed (1955), otherwise taxa were coded with missing data. A possible correlation (although unconfirmed) between the vascularization of the petiole and the median vein may exist (Baas et al. 1982). Thus, the unilacunar nodes would be generally associated with a simple type of vascularization of the petiole and principal vein. However, trilacunar nodes can be associated with a closed vascular cylinder or a simple arc. For Octoknema, the vascularization of the nodes was described by van Tieghem (1905). 25. Basal petiole vascularization: simple bundle (0); incomplete vascular cylinder (1); simple vascular cylinder (2). This type of 582 SYSTEMATIC BOTANY vascularization is constant in a genus and sometimes in a tribe (Baas et al. 1982). Variation exists, however, in the condition of the vascular system between the base of the petiole, the top of the petiole (character 26), and the median vein in the center of the lamina (character 27). 26. Distal petiole vascularization: simple bundle (0); complex vascular cylinder and one or more adaxial or enclosed strands (1); simple vascular cylinder (2). 27. Median vein vascularization: simple bundle (0); complex vascular cylinder and one or more adaxial or enclosed strands (1); simple vascular cylinder (2). Wood anatomy The diagnostic characteristics of wood in Olacaceae were outlined by van den Oever (1984). Moreover, Herendeen and Miller (2000) proposed a phylogenetic coding for characters traditionally used for wood anatomical data and this coding will be used here. 28. Vessel grouping: solitary (0); grouped in multiples (1). Vessels can be exclusively solitary, or can be grouped in multiples of two to four (and even up to 15 in Anacolosa). A frequency of solitary vessels higher than 60% was coded as wood with solitary vessels, following van den Oever (1984). 29. Perforation plate type: scalariform (0); simple (1). The type of perforation was studied by van den Oever (1984) on mature wood. Heisteria scandens, a lianescent species, has simple and scalariform perforations in young wood, then exclusively simple perforations in older wood (van den Oever in prep.). Herendeen and Miller (2000) proposed using a more complex coding that does not have utility in this study. The number of bars of the scalariform perforations could also be introduced into the analysis. Van den Oever (in prep.) mentions that scalariform perforations of the genera Coula, Minquartia, Ochanostachys, Scorodocarpus, and Engomegoma always have more than ten bars. 30. Vessel member length: below 900 mm (0); over 900 mm (1). Two distinct non-overlaping length classes for vessel elements exist in the wood of Olacaceae (van den Oever 1984; in prep.). 31. Intervascular pits (if present): alternate (0); opposite (1); scalariform (2). Although used by van den Oever (1984),this character may not be independent of character 28. Solitary vessels obviously never have intervascular pits. However, taking into account the occasional presence of two adjacent vessels (and coding ‘‘solitary vessels’’ when at least 60% of them are solitary), the type of intervascular pitting can be easily observed (van den Oever in prep.). This coding corresponds to that recommended by Herendeen and Miller (2000) when character states inapplicable in Santalales are removed. 32. Vascular tracheids associated with the vessels: absent (0); present (1). This character was defined by van den Oever (1984). These tracheids are rare but can be observed in certain cases in macerations. The coding used corresponds to that of Herendeen and Miller (2000). 33. Fibers: fiber-tracheids (0); libriform fibers (1). The terminology used for this character is that of Baas (1986), and coding is that proposed by van den Oever (1984, in prep.). Usually one observes either fiber tracheids or libriform fibers. An exceptional case is the septate libriform fibers in Octoknema, which were coded as libriform fibers. 34. Axial parenchyma frequency: abundant (0); rare or absent (1). The abundance of axial parenchyma was recorded by van den Oever (1984; in prep.). This type of tissue is very rare in Heisteria and Maburea and some other genera. 35. Axial parenchyma strand width: less than seven cells (0); more than seven cells (1). Two classes of axial parenchyma strand width can be distinguished in Olacaceae: less than seven cells broad (narrow bands of axial parenchyma), and greater than seven cells broad (broad bands of axial parenchyma) (van den Oever 1984, in prep.). Taxa with rare or absent axial parenchyma (character 34 state 1) have been coded as inapplicable ‘‘-’’. This coding [Volume 29 corresponds to one of the four characters suggested by Herendeen and Miller (2000) to code parenchyma. 36. Ray type: all cells procumbent (homogeneous) (0); heterocellular, one row of erect cells (1); heterocellular, several rows of erect or square marginal cells (2). Various types of rays can be distinguished that correspond to the heterogeneous rays I, II, III types (van den Oever 1984, in prep.) and with the homogeneous rays of Kribs (1968). Herendeen and Miller (2000) recommend coding the composition of the rays in the following way: first, homogeneous rays, all cells procumbent (homogeneous I II and III of Kribs 1935 5 homogeneous of Kribs 1968): second, heterocellular rays, a file of erect cells (heterogeneous I of Kribs 1935 and Kribs 1968): third, ray heterocellular, only one row of erect or square marginal cells (heterogeneous IIA and IIB of Kribs 1935 5 heterogeneous II and III of Kribs 1968). Such coding will be used here. 37. Ray height: long (over 1,000 mm) (0); short (below 1,000 mm) (1). The height of the rays varies from 94 mm to 4300 mm, and van den Oever (1984) determined a threshold value equal to 1 mm, making it possible to distinguish two non-overlaping groups within Olacaceae. 38. Wood cystoliths: absent (0); present (1). Various genera of Opiliaceae have cystoliths in the wood rays, that are apparently independent of their presence in leaves (Koek-Noorman and van Rijckevorsel 1983). It is possible that this character could be used to distinguish groups within Opiliaceae. 39. Silica bodies in ray cells: absent (0): present (1). van den Oever (1984) mentions only their presence in three genera, Olax, Ptychopetalum, and Dulacia. Inflorescence and flowers 40. Inflorescence bracts: absent (0); present (1). 41. Floral and inflorescence trichomes: absent (0); present (1). 42. Flower sexual condition: bisexual (0); unisexual (1). Perianth 43. Floral bud shape: spherical (0); oval (1). In Olacaceae, floral buds are either spherical, measuring 2–3 mm in diameter (e.g., Coula), or lengthened, and reaching more than 5 cm in length (e.g., Chaunochiton). The length and width of floral buds before anthesis were measured on rehydrated flowers from herbarium specimens. The flowers for which the length/width ratio was between 0.9 and 1.1 were regarded as spherical, whereas those with values higher than 1.1 were regarded as oval. 44. Flower merosity: five (0); four (1); six (2). The number of parts per floral cycle was determined directly or obtained from the literature (Douradoa). The number of stamens was not used to code this character because of the simultaneous and variable presence, in certain genera, of stamens and staminodes. Within a genus, the number of parts per cycle does not vary, except for Phanerodiscus and Olax spp. Sometimes the number of floral parts per flower varies on the same branch. Olax is often regarded as having only three petals (Linné 1753; Sleumer 1984a, 1984b), but the anatomical studies of Argawal (1963) and Patil and Pai (1984), the observations of Capuron (1968), and our own observations of herbarium samples indicate that these flowers are pentamerous with varying degrees of fusion between the more or less distinct petals. 45. Sepals: developed, partially fused (0); developed, completely fused (1); calyculus (2); absent (3). Among genera of Olacaceae, the petals are more or less fused. In Opiliaceae, Santalaceae, Viscaceae, and Loranthaceae various and contradictory interpretations of the floral parts have been proposed. Here we scored Misodendraceae and Opiliaceae as lacking a calyx (a ‘torus’ is present in Opiliaceae according to Hiepko 2000), whereas we interpreted Santalaceae and Viscaceae as having only a developed calyx (Eichler 1878; Kuijt 1968). In Loranthaceae, a calyculus is generally present. Among Olacaceae, Schoepfia and Octoknema were regarded 2004] MALÉCOT ET AL.: OLACACEAE PHYLOGENY as lacking a calyx, although very small sepals are present on the female flowers of Octoknema orientalis. 46. Accrescent calyx: absent (0); present (1). Identification of an accrescent structure around the fruit was based on the study of herbarium samples of ripe and unriped fruits. Taking into account the problems of homology and the difficulty in considering the development relative to certain organs, each tissue that can be accrescent was coded independently, and only the accrescence of the calyx and of the disc/receptacle (character 58) were taken into account. In Erythropalum, one observes no accrescent structure around the fruit, but at maturity the mesocarp tears into four to five parts, revealing a black, sclerified endocarp, and a bright red, internal mesocarp face. This structure has sometimes been regarded as an accrescent calyx (particularly given its radial pseudosymmetry), but it is not at all homologous. In santalaceous families treated here as lacking a calyx (character 45 state 3) this character was coded as inapplicable. 47. Corolla connation: apopetalous (0); sympetalous at base only (1); sympetalous with a floral tube (2). If one considers that Santalaceae have only one cycle of floral parts (monochlamydous) and this is interpreted to represent the calyx, the present character cannot be coded for that family, and was scored as inapplicable. For our purposes, we considered petals fused by less than half their length to be connate at their base, whereas those fused by more half their length to have a floral tube. In the case of variation in the degree of fusion between petals of the same flower (Olax), we took account of the measured minimal value. 48. Petal pubescence: petal glabrous (0); petal with hairs in upper third (1); petal with 1 row of hairs (2). Pubescence is particularly variable on the external as well as the internal surfaces of the petals. External pubescence was very variable on the same individual at anthesis and was thus not taken into consideration. The exact nature of the trichomes occupying the internal face was not established with precision but in Anacolosa, Cathedra, and Phanerodiscus they appear as two types. Those adjacent to the stamens are likely glandular (but were not coded here), whereas the uniserial trichomes were coded. Androecium 49. Stamen number: three (six including staminodes) (0); four (eight including staminodes) (1); five (ten including staminodes) (2); twelve to twenty (3). 50. Stamen whorls: one (0); two (1); three (2). The number of staminal cycles was calculated by taking into account possible staminodes and the merosity of the flowers. In the case of Olax, the total number of stamens and staminodes is in general not proportional to the number of petals (five), but always lies between seven and ten. The flowers of this genus were regarded as having an androecium composed of two cycles. 51. Stamen position relative to corolla: alternate and opposite (0); opposite only (1). For Olacaceae we tried to identify epipetalous stamens and alternipetalous stamens, but, in various genera this distinction proved to be impossible to realize in the absence of fresh material and a study of floral development. Thus, for flowers of Scorodocarpus, two stamens are fused with each petal and are placed at equal distance from the edges of the petal. For Olax, a similar phenomenon is observed, which becomes complicated because of the probable absence of certain parts of the androecium. In these last two cases, stamens were regarded as only opposite. 52. Staminodia: absent (0); present (1). The presence of staminodia provides a means of distinguishing the genera of tribe Olaceae. Other genera (e.g. Ongokea) have at the base of the staminal column either staminodia or a divided disc, depending upon interpretation. Similarly, the condition in Opiliaceae can be interpreted as staminodia (Kuijt 1969) or a divided disc (Hiepko 2000). In all such cases, our coding follows the latter interpretation. 53. Filament: free (0); fused (1). The mutual fusion of the fila- 583 ments of stamens (synadelphous or monadelphous) is characteristic of a small number of genera: Aptandra, Harmandia, and Ongokea. 54. Anther connective: typical (0); prolonged to a point or thicken (1). The top of the stamen connective is generally blunt and without excrescenses, but for Anacolosa and Cathedra it is pubescent, whereas for Diogoa, Tetrastylidium, and Engomegoma it is pointed. 55. Anther attachment: dorsifixed (0); basifixed (1); oblique (2); other (3). The state ‘‘oblique’’ was applied to the very short anther of Chaunochiton whereas ‘‘other’’ was applied to members of tribe Aptandreae where stamens are fused by their filaments. 56. Anther dehiscence: slit (0); pores (1); flaps (2); disaggregation (3). Anther dehiscence is generally longitudinal but poricidal for tribe Aptandreae. For Anacoloseae and Chaunochiton, dehiscence is valvate. Curupira has a curious short longitudinal dehiscence that does not appear to be the result of mechanical tearing but a disaggregation of the tissue that exposes the interior of the thecae. 57. Direction of anther dehiscence: introrse or lateral (0); extrorse (1). Gynoecium 58. Accrescent disk: absent (0); present (1). The disc corresponds either to glandular tissue surrounding the ovary (Strombosia), or located below it (Olax), or with a cup into which the petals and stamens are inserted (Anacolosa, Cathedra, Phanerodiscus). Thus, the disc is a descriptive term used to indicate an often glandular structure of the flower (see also character 59). Accrescence of the ‘‘disk’’ applies only to a cup into which the corolla and androcium are inserted. For Phanerodiscus, Capuron (1968) and Malécot et al. (2003) showed that neither the calyx nor the disk are accrescent, but a structure indiscernable in the flower at anthesis. 59. Glandular tissue (disk): between stamens and ovary (0); between stamens and petals (1); none (2). In this analysis, we code only glandular tissue that correspond to structures of reduced size, that areseparate from the ovary (not fused as in Strombosiopsis), and that are not being used as support for the stamens and the petals (as in Anacolosa, Cathedra, and Phanerodiscus) (see also character 58). 60. Style shape: long conical (0); cylindrical short (1); cylindrical long (2); short conical (3). Although style shape may be influenced by the position of the ovary (inferior or superior), the relative length of the style was established by taking into account the relationship between the distance from the top of the locules of the ovary to the stigma and the diameter below the stigma. When this ratio is lower than six, the style was regarded as short. The values used were measured on rehydrated floral material at anthesis (long-styled flowers in the case of the genera with heterostylous flowers). Heterostyly is mentioned or supposed for several genera of Olacaceae (Michaud 1966; George 1984; Sleumer 1984a, 1984b), in which case only longistylous material was used for coding. 61. Stigma length: small globular (0); elongated (1). The length of the stigmas was measured on the same material used for coding character 60. The elongated character state occurs only in the outgroup (stigmas of more than 1 mm in length). The micromorphology of the stigmas was not studied. 62. Ovary position: hypogynous (0); epigynous (1); half inferior (2). The position of the ovary was observed on rehydrated flowers obtained from herbarium samples or was determined from the literature (Douradoa). Schoepfia, sometimes described as having a half inferior ovary, was regarded as having an inferior one. 63. Ovary locule number: five (0); four (1); three (2); two (3); one (4). The number of locules at the base of the ovary is based on the data available in the literature, in particular Fagerlind (1946, 1947, 1948) and Sleumer (1984a, 1984b). The reliability of this information is doubtful for some and unknown for other genera. 584 SYSTEMATIC BOTANY 64. Integument number: two (0); one (1); none (2). Given varying accounts in the old literature, only the indications provided by Bouman and Boesewinkel (in Breteler et al. 1996) were used in coding. The genera for which variation exists (cf. Sleumer 1984b) were coded as polymorphic. Fruit 65. Starch in fruit: absent (0); present (1). The nature of the reserves of the fruit was coded according to the data available in the literature (Michaud 1966; Hegnauer 1966, 1969, 1973, in Sleumer 1984a, 1984b). This character was used by Engler (1897) to distinguish the subfamilies of Olacaceae. In general, starch and/or lipid quantity is not known and, according to Sleumer (1984b), varies in the same genus and even in the same species. Thus some taxa that were recorded as lacking either starch or lipid may contain these, as well as other types of seed reserves. Despite the absence of data for some taxa, the presence/absence of starch and lipid (character 66) was coded. 66. Lipids in fruit: absent (0); present (1). Pollen 67. Pollen shape: equiaxial (0); longiaxial (1); breviaxial (2). The ratio of the distance between the poles and the equatorial diameter of the pollen grains results from the observations of Lobreau-Callen (1980, 1982). The values used to distinguish the three character states are those recommended by Punt et al. (1994). 68. Pollen symmetry: isopolar (0); heteropolar (1). Work by Feuer (1977), Lobreau-Callen (1980, 1982, in Sleumer 1984b, in Breteler et al. 1996) and Hiepko and Lobreau-Callen (in Maas et al. 1992) showed that the pollen of certain Olacaceae was asymmetrical. 69. Mesocolpium shape: flat or convex (0); concave (1). According to Lobreau-Callen (1980), the pollen of Aptandreae has a concave mesocolpium, a unique feature in Santalales. 70. Apocolpium shape: convex (0): concave (1). Concave apocolpium is a distinctive feature of tribe Aptandreae. 71. Aperture number: three to four, zonoaperturate (0); six, diploporate (1); more than six, periaperturate (2). In Santalales, the number and position of the apertures are linked: pollen grains with three or four apertures are always zonoaperturate, those with six apertures are always diploporous, and those with more than six apertures (Misodendrum) are periaperturate. In some genera, the number of apertures varies between three or four for pollen derived from the same anther, but these are zonoaperturate pollens. 72. Ectoaperture shape: ellongate (furrow) (0); round (porous) [Volume 29 or elliptic (1). The ectoaperture of Olacaceae can either be lengthened perpendicular to the equator (furrow), or circular or elliptic. 73. Apertural membrane: smooth or scabrous (0); granular or verucose (1). 74. Endoaperture shape: elliptic (0); circular (1). The endoaperture can be lengthened perpendicular to the equator, or circular. The absence of endoapertures for certain Opiliaceae prevented coding this character for some taxa, and they were scored as inapplicable. 75. Relative size of the endo- and ectoaperture: ecto- and endoaperture of distinct size (0); ecto- and endoaperture of identical size (1). 76. Endoaperture granules: none and endoaperture smooth (0); present and endoarpeture smooth (1); none and endoaperture with endosculpture (2). Lobreau-Callen (1980, 1982) showed that the pollen endoapertures of Olacaceae could be ornamented. Presence of granules in the endoaperture was coded but not other modifications of the nexine (such as endosculpted nexine instead of a true endoaperture). 77. Exine in mesocolpium: smooth or microperforate tectum (1); reticulate exine (1); echinulate tectum (2); other (3). In the mesocolpium and apocolpium the exine can correspond to a continuous tectum with rare perforations, with a microperforate or perforated tectum or with a network (Lobreau-Callen 1980, 1982). However, the transition between these various types is almost continuous and combines two elements, the size of the perforations and the density of the perforations. The terminology used here corresponds to extremes but is not easily applicable to certain intermediate forms. 78. Granules in infratectum: absent (0); present (1). According to Lobreau-Callen (1980), the structure of the infratectum makes it possible to distinguish various groups of Olacaceae. In certain genera this layer consists of individualized grains laid out in several layers, in others only one layer of grains exist, which tend to form columellae (Anacolosa, Aptandra, Ptychopetalum), or individualized columellae (Chaunochiton). Homology among these features is not clear and two characters were defined, one for the presence of granules whatever the number of layers, and one for the coding presence of columellae-like structures (character 79). 79. Columella in infratectum: absent (0); present (1). 80. Foot layer surface: smooth (0); irregular with masses (1). Lobreau-Callen (1980) use the foot-layer as a distinctive feature to separate two main set of Olacaceae. Despite the relatively low number of taxa for which this information was available, it appears useful for characterizing some groups among Olacaceae. Various other structural characteristics of the plate and the endexine were highlighted by Feuer (1977) and Lobreau-Callen (1980). These characters are available only for the small number of genera that have been studied using transmission electron microscopy. 2004] APPENDIX 2. Data matrix analyzed in this study. Abbreviations: A 5 0 or 1; B 5 1 or 2; C 5 0 or 2. 2222222222 0123456789 3333333333 0123456789 4444444444 0123456789 5555555555 0123456789 6666666666 0123456789 77777777778 01234567890 Viscaceae Korthalsella complanata Notothixos leiophyllus Phoradendron californicum Phoradendron serotinum Viscum album Viscum articulatum 101???002 100011002 10A011002 10A011002 10A011002 10A011002 00?1000010 00?1000010 00?1000011 00?1000011 00?1000010 00?1000010 ?0??000000 ?0??000000 ?011000000 ?011000000 ?011000000 ?011000000 10?1000000 10?1000000 10?01-0000 10?01-0000 10?1001000 10?1001000 ??10110000 ??10110000 ??10110000 ??10110000 ??1011000A ??1011000A 01000?0002 01000?0002 01000?0002 01000?0002 01000?0002 01000?0002 ?014201000 ?014201000 ?014201000 ?014201000 ?014201000 ?014201000 0001?002??? 0001?002??? 0001?002??? 0001?002??? 0001?002??? 0001?002??? Santalaceae Acanthosyris asipapote Buckleya distichophylla Colpoon compressum Exocarpos bidwillii Okoubaka aubrevillei Osyris lanceolata Pyrularia pubera Rhoiacarpos capensis Santalum album Scleropyrum pentandrum Thesium humifusum 000111002 100011012 100???002 001011102 100011002 000???002 000111012 100???002 100A11002 000???002 000?11002 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 00?1000010 ?011000000 ?011000000 ?0??000010 ?0??000010 ?0??000010 1011000010 ?0??000000 ?0??000010 ?0??000010 ?0??000010 10??000010 11001-A101 11101-010? 1?101-110? 1?101-1100 1?001-0000 1?101-1100 10001-210? 1?101-110? 1?101-1100 1?001-210? 1?101-110? ??00010002 ??10010001 ??00010001 ??10010002 ??10010012 ??1001000A ??10010002 ??0001000B ??00010001 ??00010002 ??0011001B 01000?0000 01000?0002 01000?0002 01000?0002 01100?0000 01000?0002 01000?0002 01000?0002 01000?0002 01000?0002 01000?0002 ?014201000 ?014201000 ?014201000 ?014201100 ?014201000 ?014201100 ?014201010 ?014201000 2014201100 ?014201000 ?014201011 0011?100??? 0001?005??? 0001?001??? 0001?005??? 0000100010? 0001?000??? 00A1?100??? 0001?000??? 00110104??? 00A1?100??? 0011?101??? Opiliaceae Agonandra macrocarpa Champereia manillana Opilia amentacea Pentarhopalopilia marquesii Rhopalopilia umbellata Urobotrya siamensis 010011000 000011000 010011000 000011000 000011010 000011010 0011000000 0011000001 0011000000 0011000000 0011000001 0011000001 0100?00010 0100?00010 0110?00010 0100?00010 0100?00010 0100?00000 1000001100 1000001110 1010001110 1000001100 1000001100 1000001100 ??1013--01 ??0003--02 ??0003--02 ??0003--02 ??0013--01 ??0013--02 01000?0000 01000?0000 01000?0000 01000?0000 01000?0000 01000?0000 1014201000 1014201000 1014201000 1014201000 1014201200 1014201200 00010?20000 0001-001010 00011001010 0001-001010 00010021010 00010001010 Loranthaceae Dendrophthoe lepidota Ligaria cuneifolia Tripodanthus flagellaris Tupeia antarctica 100011002 100???002 100???002 100011002 00?1000010 00?1000010 00?1000010 00?1000010 1000000000 ?0??000000 ?0??000000 1000000000 1001000100 1001000100 100?000100 1????????0 ??01020202 ??01020202 ??0102020? ??1??20?0? 01000?0002 01000?0002 01000?0002 01??0??002 ?014201200 ?014201201 ?014201201 ?014201200 0101?00???? 0001?005??? 0001?002??? 0101?00???? Misodendraceae Misodendrum brachystachyum 000011010 00?1000001 0110?00000 1101000100 ??1013--00 01100??002 1014201000 0211211200? Olacaceae Anacolosa frutescens Aptandra tubicina Cathedra acuminata Chaunochiton kappleri 000111010 000111001 000111010 000111010 1001000001 0001000101 1111000001 0111000101 0011000000 0000000000 1011000000 0011000000 101101A000 1010001000 101101A000 1011001100 0000010112 0001110001 A000010012 1001011112 0100111012 0101032101 0100101012 0100022102 20031?1200 2003B11211 2003110200 2003B?1200 01111110110 10111110110 0111111010? 00000003010 585 1111111111 0123456789 MALÉCOT ET AL.: OLACACEAE PHYLOGENY 123456789 586 APPENDIX 2. Continued 1111111111 0123456789 2222222222 0123456789 3333333333 0123456789 4444444444 0123456789 5555555555 0123456789 6666666666 0123456789 77777777778 01234567890 Coula edulis Curupira tefeensis Diogoa zenkeri Douradoa consimilis Dulacia candida Engomegoma gordonii Erythropalum scandens Harmandia mekongensis Heisteria concinna Heisteria parvifolia Maburea trinervis Malania oleifera Minquartia guianensis Ochanostachys amentacea Octoknema dinklagei Olax subscorpioidea Ongokea gore Phanerodiscus perrieri Ptychopetalum petiolatum Schoepfia schreberi Scorodocarpus borneensis Strombosia javanica Strombosiopsis tetrandra Tetrastylidium peruvianum Ximenia americana 010000102 000111000 000001000 000111000 000111010 000???000 000001000 000111001 000001000 000AA1000 010001000 000111000 010001102 000000102 020001000 000111010 000111000 000111000 000111000 000111010 000001000 000001000 000001A00 000001000 000111000 0011001100 0001000001 0110110010 0001000001 0011000001 0111110010 1000010000 0101000101 0110110100 0110110100 1010110000 0110000001 0111001100 0111001100 0010010000 0000000001 0001000001 1001000000 0001000001 0001000001 0110110000 0110100010 0110110010 0110110010 0001000001 0011121101 0001110010 0011?21111 0001?10010 0000100010 0011?21111 0011?22001 0000000010 1011122211 0011122211 0011?21111 0001?10010 0011122201 0011121101 0011222201 0000000000 0000000010 0000?00000 0000100000 0000000000 0011222201 1011122201 1011121101 1011121101 0001110010 0001012000 1010001100 0100012100 10101-1000 1010001101 0?00012000 A00A112000 1011001100 0B00112000 0B00112000 0000012000 1010001100 0001012000 0001012000 0101112000 1011001101 1010001100 101101A100 1011001101 1011000100 0001012000 0101012000 0101012000 0101012000 10001-1100 1100010003 1100110011 1000010002 1100110011 1101211100 100011000B 1000000A12 1011110201 0000001A22 0000001A22 1100000112 1100100001 0100010112 1100110001 ?110020002 1101211120 1?01110002 1100C00012 1001010012 ??00020222 1101010012 1101010012 1001110101 0000110101 1001110011 B000000002 1000013002 0100010002 0000010?02 0110000002 0100110002 0100010002 0101032101 1000000002 1000000002 1000000002 1000000002 1000000002 B000000000 0100000000 1010010000 0111032101 0100111102 100001000? 01000?0000 1100010000 0100000002 0100110000 0100100002 1000010002 3002011010 0001?01000 30001??000 000?101000 2002B1A200 30011??0A0 1002101000 2004B01211 3002011??? 3002011010 10020??010 000???1210 3001010000 3002010010 3002A??000 2002B01200 2004B?1211 2003??1201 2004201200 1012101010 2002A10010 2024110010 30211??000 1001101000 00040010A0 0000-000001 0001-000??0 00010001100 00010001100 00111110101 00010001010 00010000101 10111110110 00????????? 00010000100 00010001??? 00000000??? 00000000000 0001-000001 0001-110110 00111110101 10111110111 01111110110 01111110101 00011000100 00010000100 00010010101 00010001100 00010000100 00010000111 outgroups Daphniphyllum pentandrum Rhabdodendron amazonicum 000???100 020???101 0000001001 0000001001 10??2???11 1011222210 02001-1000 0210002000 ?????00??? 0000010003 ?00?0??00? 2000110002 ?1000????? 11140??000 00????????? 0001???1??? SYSTEMATIC BOTANY 123456789 [Volume 29