Phylogeny, character evolution, and classification of Sapotaceae (Ericales)

We present the first cladistic study of the largely tropical family Sapotaceae based on both morphological and molecular data. The data were analyzed with standard parsimony and parsimony jackknife algorithms using equally and successive weighted characters. Sapotaceae are confirmed to constitute two main evolutionary lineages corresponding to the tribes Isonandreae-MimusopeaeSideroxyleae and Chrysophylleae-Omphalocarpeae. The Sideroxyleae are monophyletic, Isonandreae are polyphyletic as presently circumscribed, and as suggested by the analyses, the subtribe Mimusopeae-Mimusopinae has evolved within the MimusopeaeManilkarinae, which hence is also paraphyletic. Generic limits must be altered within Sideroxyleae with the current members Argania, Nesoluma and Sideroxylon. Argania cannot be maintained at a generic level unless a narrower generic concept is adopted for Sideroxylon. Nesoluma cannot be upheld in a narrow or broad generic concept of Sideroxylon. The large tribe Chrysophylleae circumscribes genera such as Chrysophyllum, Pouteria, Synsepalum, and Xantolis, but the tribe is monophyletic only if the taxa from Omphalocarpeae are also included. Neither Chrysophyllum nor Pouteria are monophyletic in their current definitions. The results indicate that the African taxa of Pouteria are monophyletic and distinguishable from the South American taxa. Resurrection of Planchonella, corresponding to Pouteria section Oligotheca, is proposed. The African genera Synsepalum and Englerophytum form a monophyletic group, but their generic limits are uncertain. Classification of the Asian genus Xantolis is particularly interesting. Morphology alone is indecisive regarding Xantolis relationships, the combined unweighted data of molecules and morphology indicates a sister position to Isonandreae-Mimusopeae-Sideroxyleae, whereas molecular data alone, as well as successive weighted combined data point to a sister position to Chrysophylleae-Omphalocarpeae. An amended subfamily classification is proposed corresponding to the monophyletic groups: Sarcospermatoideae (Sarcosperma), Sapotoideae (Isonandreae-Mimusopeae-Sideroxyleae) and Chrysophylloideae (Chrysophylleae-Omphalocarpeae), where Sapotoideae circumscribes the tribes Sapoteae and Sideroxyleae as well as two or three as yet unnamed lineages. Morphological characters are often highly homoplasious and unambiguous synapomorphies cannot be identified for subfamilies or tribes, which we believe are the reason for the variations seen between different classifications of Sapotaceae. The Willi Hennig Society 2005. Sapotaceae are one member of the Ericales, a clade of morphologically variable angiosperm families where the relationships are not yet fully understood (Anderberg et al., 2002). The Sapotaceae are currently subdivided into five tribes with 53 genera and approximately 1250 species (Pennington, 1991; Govaerts et al., 2001). It consists of trees or shrubs with a world-wide distribution, although the highest species diversity is found in the tropical and subtropical regions of Asia and South America. Sticky and often white latex is found in cuts of bark, branches, leaves and fruits, although it often appears slowly in species growing in dry conditions. Leaves are generally alternate, simple, and entire, but exceptions of opposite leaves are present in Leptostylis, Pichonia, and Sarcosperma. Flower structure provides many useful characters for the circumscription of tribes and genera, and can be either simple or complex. Stamens are always opposite the corolla lobes, but many genera have staminodes in the corolla lobe sinuses. *Corresponding author. E-mail address: [email protected] The Willi Hennig Society 2005 Cladistics www.blackwell-synergy.com Cladistics 21 (2005) 101–130 Another family characteristic is the malpighian hairs on different organs, a structure of a small stalk and two branches, often of different length. Lam (1939), Aubréville (1964), Baehni (1965), and most recently Pennington (1991), proposed systems of classification based on morphological investigations that exhibited more or less contradictory results. Pennington accepted five tribes: Chrysophylleae, Isonandreae, Mimusopeae, Omphalocarpeae and Sideroxyleae, with Mimusopeae subdivided into three subtribes. Chrysophylleae are the largest tribe and traditionally recognized on 4–5-merous flowers with a calyx in a single whorl, imbricate or quincuncial sepals, and with corolla lobes, stamens and staminodes (if present) in the same number as sepals. Each corolla lobe is always entire. Sideroxyleae is characterized by features similar to those of Chrysophylleae, but differ in the usually divided corolla lobes. Mimusopeae and Isonandreae have a calyx in two whorls of 2–4 valvate sepals in each, though, never two in Mimusopeae or four in Isonandreae. Both tribes show exceptions to this pattern and rather have a calyx in a single whorl of 4–5 sepals. Corolla lobes, stamens and staminodes are usually of the same number as sepals in Mimusopeae whereas they are 2–3 times in Isonandreae, the latter consistently lacking staminodes. The two tribes also differ in having subdivided (Mimusopeae) or entire (Isonandreae) corolla lobes. In other words, both tribes have complex combinations of flower structures. The Omphalocarpeae are an assemblage of genera that are difficult to place. All members have several stamens opposite the corolla lobe, which is not unique to this tribe, but stamen position is not precise. As a result of the often complex distribution of characters and character states, Pennington (1991) circumscribed several genera in a wide sense. He revised and reduced Aubréville’s (1964) 122 accepted genera to 53. For instance, Bumelia, Dipholis, Monotheca and Mastichodendron, that had been recognized by earlier workers are now included in Sideroxylon. Another example is Aningeria, Calocarpum, Lucuma, Malacantha and Planchonella, which are all presently included in Pouteria. As a result, Chrysophyllum (81 spp.), Pouteria (304 spp.) and Sideroxylon (76 spp.) are now large groups that are difficult to recognize and lack apparent synapomorphies. A phylogenetic hypothesis based on cpDNA sequence data from the plastid gene ndhF was presented by Anderberg and Swenson (2003) in which three main evolutionary lineages in the family were identified, Sarcosperma (referred to as Clade 1), Isonandreae, Mimusopeae, and Sideroxyleae (Clade 2), and Chrysophylleae-Omphalocarpeae (Clade 3). While providing a robust basis for future work, this molecular study did not provide sufficient data to resolve the generic relationships within each of the two larger clades, and did not shed light on the distribution of stated morphological features to diagnose genera and tribes. Adding morphological data to our previous molecular data set will help to better understand the diagnostic value of the morphological features on which earlier taxonomists relied to form their opinions of generic relationships. In our previous paper (Anderberg and Swenson, 2003) we proposed that one problem with morphological characters was that character states combined in various ways, and that some states, such as a simple calyx, entire corolla lobes, and stamens equalling the sepals in number, may be symplesiomorphies, as they are present in Sarcosperma, the sister to all other Sapotaceae (Anderberg et al., 2002; Anderberg and Swenson, 2003). Other more complex features, such as stamens in one or two whorls, double calyx, and divided corolla lobes, were proposed to be synapomorphic. For example, the simple, often 5-merous flowers of Chrysophyllum could represent a symplesiomorphic type of flower, whereas the complex flowers of Manilkara could indicate a more derived relationship in the family. The present analysis increases the sampling of taxa and combines the DNA sequence data from the ndhF gene with a newly developed morphological data set. The primary goal is to investigate the monophyly of the two main evolutionary lineages and the large genera of Sapotaceae. We also attempt to evaluate the diagnostic value of the more important and often used morphological characters. Materials and methods