Towards a natural classification of Sapotaceae subfamily Chrysophylloideae in the Neotropics

Mário  Terra

Generic limits of Chrysophyllum and Pouteria (Chrysophylloideae, Sapotaceae) have been found to be untenable. We here search for natural lineages in Neotropical Chrysophylloideae by sampling 101 terminals for molecular sequences of nuclear ribosomal DNA (external and internal transcribed spacer), the nuclear gene RPB2 and 17 morphological characters. Data were analysed with Bayesian inference and parsimony jackknifing. Morphological traits were finally optimized onto the tree to identify the most coherent characters. The resulting phylogenetic tree suggests that the limits of the well-known genera Chrysophyllum and Pouteria must be amended. Diploon, Ecclinusa and Elaeoluma can be maintained and Chrysophyllum sections Ragala section Prieurella and the satellite gen-era Achrouteria, Cornuella, Martiusella and Nemaluma merit generic resurrection. Lucuma may be restored if the type species belongs to the clade. The accepted genera Chromolucuma, Pradosia and Sarcaulus gain strong clade support, but are embedded in a core clade of Pouteria and may be relegated to the subgeneric level if morphological studies cannot provide evidence concurring with narrow generic concepts. Circumscriptions of Micropholis and Chrysophyllum sections Chrysophyllum and Villocuspis remain unclear and must be explored by using an extended taxon sampling. We predict that yet-to-be-analysed species of Pouteria sections Franchetella, Gayella, Oxythece and Pouteria and members of the currently accepted genera Chromolucuma, Pradosia and Sarcaulus will fall inside the core clade of Pouteria when analysed.

Botanical Journal of the Linnean Society, 2017, 185, 27–55. With 4 figures. Towards a natural classification of Sapotaceae subfamily Chrysophylloideae in the Neotropics APARECIDA DONISETE DE FARIA1*, JOSÉ RUBENS PIRANI2, JOSÉ EDUARDO LAHOZ DA SILVA RIBEIRO1, STEPHAN NYLINDER3, MÁRIO HENRIQUE TERRA-ARAUJO4, PEDRO PAULO VIEIRA5 and ULF SWENSON3 1 Departamento de Biologia Animal e Vegetal, Centro de Ciências Biológicas, Universidade Estadual de Londrina, Campus Universitário, Rodovia Celso Garcia Cid, PR 445, Km 380, 86057-970, Londrina, PR, Brazil 2 Departamento de Botânica, Instituto de Biociências, Universidade de São Paulo, Rua do Matão, 277, 05508-9070, São Paulo, SP, Brazil 3 Department of Botany, Swedish Museum of Natural History, PO Box 50007, SE-104 05 Stockholm, Sweden 4 Instituto Nacional de Pesquisas da Amazônia, Avenida André Araujo 2936, Coordenação de Pesquisas em Botânica, 69060-000, Manaus, AM, Brazil 5 Laboratório de Biologia Molecular, Fundação de Medicina Tropical, Av. Pedro Teixeira, 25, Dom Pedro, 69040-000, Manaus, AM, Brazil Received 17 January 2017; revised 15 May 2017; accepted for publication 2 June 2017 Generic limits of Chrysophyllum and Pouteria (Chrysophylloideae, Sapotaceae) have been found to be untenable. We here search for natural lineages in Neotropical Chrysophylloideae by sampling 101 terminals for molecular sequences of nuclear ribosomal DNA (external and internal transcribed spacer), the nuclear gene RPB2 and 17 morphological characters. Data were analysed with Bayesian inference and parsimony jackknifing. Morphological traits were finally optimized onto the tree to identify the most coherent characters. The resulting phylogenetic tree suggests that the limits of the well-known genera Chrysophyllum and Pouteria must be amended. Diploon, Ecclinusa and Elaeoluma can be maintained and Chrysophyllum sections Ragala section Prieurella and the satellite genera Achrouteria, Cornuella, Martiusella and Nemaluma merit generic resurrection. Lucuma may be restored if the type species belongs to the clade. The accepted genera Chromolucuma, Pradosia and Sarcaulus gain strong clade support, but are embedded in a core clade of Pouteria and may be relegated to the subgeneric level if morphological studies cannot provide evidence concurring with narrow generic concepts. Circumscriptions of Micropholis and Chrysophyllum sections Chrysophyllum and Villocuspis remain unclear and must be explored by using an extended taxon sampling. We predict that yet-to-be-analysed species of Pouteria sections Franchetella, Gayella, Oxythece and Pouteria and members of the currently accepted genera Chromolucuma, Pradosia and Sarcaulus will fall inside the core clade of Pouteria when analysed. ADDITIONAL KEYWORDS: BEAST – Chrysophyllum – classification – jackknife – Pouteria – South America. INTRODUCTION Sapotaceae are important wet forest components throughout the Neotropics, with some members also being found in drier biomes of the Brazilian Cerrado and Campo Rupestre. Important diagnostic characters *Corresponding author. E-mail: cidadefaria@uol.com.br of the family include the presence of latex, simple and entire leaves, malpighiaceous trichomes and flowers arranged in fascicles (except in Sarcosperma Hook.f.). The number of species has been estimated at c. 1250 (Pennington, 1991; Govaerts, Frodin & Pennington, 2001), but it is steadily increasing in tropical America (Pennington, 2006, 2007), Africa (Gautier et al., 2016) and the Pacific region (Swenson, Munzinger & © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 27 28 A. D. DE FARIA ET AL. Bartish, 2007b; Munzinger & Swenson, 2009; Swenson & Munzinger, 2012, 2016). A new subfamilial classification was proposed by Swenson & Anderberg (2005), recognizing Chrysophylloideae (similar in circumscription to Chrysophylleae), Sarcospermatoideae and Sapotoideae, with the last now including four tribes (Isonandreae, Sapoteae, Sideroxyleae and Tseboneae; Smedmark, Swenson & Anderberg, 2006; Smedmark & Anderberg, 2007; Gautier et al., 2013). The number of genera in the family has fluctuated according to the adopted generic concepts, from Aubréville’s 122 (Aubréville, 1964a), to Baehni’s 63 (Baehni, 1965) and Pennington’s 53 (Pennington, 1991), all depending on the morphological features chosen to be indicative of natural relationships. Generic boundaries in the family are still not entirely clear, particularly in the Neotropical Chrysophylloideae (Swenson, Richardson & Bartish, 2008a). Pennington (1990, 2006, 2007) recognized approximately 350 species in Chrysophylleae in the New World, distributed in nine genera: Chromolucuma Ducke (five species), Chrysophyllum L. (45 species), Diploon Cronquist (one species), Ecclinusa Mart. (11 species), Elaeoluma Baill. (four species), Micropholis (Griseb.) Pierre (39 species), Pouteria Aubl. (c. 200 species), Pradosia Liais (25 species) and Sarcaulus Radlk. (five species). He adopted wide generic concepts for Chrysophyllum and Pouteria, resulting in pantropical distributions. Phylogenetic reconstructions of Sapotaceae have, however, clearly demonstrated that these two genera are polyphyletic, that the type species are placed in clades restricted to the New World and that features traditionally used for their delimitation show extensive homoplasy that lead to untenable classifications (Swenson & Anderberg, 2005; Triono et al., 2007; Swenson, Bartish & Munzinger, 2007a; Swenson et al. 2008a; Swenson, Nylinder & Munzinger, 2013). Thus, species of the non-staminodial genus Chrysophyllum in Malesia and New Caledonia (Vink, 1958) are better placed in Amorphospermum F.Muell., Niemeyera F.Muell. or Pycnandra Benth. (Swenson et al., 2013) and are only distantly related to Chrysophyllum spp. in the New World (Swenson et al., 2008a). Similarly, several Australasian taxa were included in Pouteria by Baehni (1942), transferred to Planchonella Pierre (van Royen, 1957), transferred back to Pouteria (Baehni, 1965), returned to Planchonella (Aubréville, 1967) and again united with Pouteria (Pennington, 1991). These changing taxonomic positions reflect results of taxonomic work done before molecular phylogenetic tools became widely used. Molecular phylogenetic analyses of the Australasian taxa clearly group species into monophyletic lineages concurring with Planchonella, Pleioluma (Baill.) Baehni, Sersalisia R.Br. and Van-royena Aubrév. (Swenson et al., 2007a, 2013). The African Pouteria satellite genus Donella Pierre ex Baill. has been recently reinstated (Mackinder, Harris & Gautier, 2016), but Aningeria Aubrév. & Pellegr., Gambeya Pierre and Malacantha Pierre are also in need of resurrection; however, their circumscriptions have not yet been investigated (Swenson et al., 2008b). Similarly, circumscription of the genera of Chrysophylloideae in the New World remains unclear. Recent phylogenetic analyses of molecular data indicate that Neotropical Chrysophylloideae probably form a clade that could have originated in South America in the Palaeocene, c. 59 Mya (Bartish et al., 2011). Some of the diagnostic characters used by Pennington (1990) to delimit the Neotropical genera are presence vs. absence of stipules, staminodes and endosperm, with the shape of the corolla (rotate/cup-shaped or cyathiform/tubular). Also, Pennington’s (1990) dichotomous generic keys indicate that the genera could be unnaturally delimited; for example, the majority of Pouteria spp. possess staminodes, but some lack them, a character pertinent for other genera. Stipules, a diagnostic feature for Chromolucuma and Ecclinusa, are also found in P. congestifolia Pilz, P. flavilatex T.D.Penn. and P. stipulifera T.D.Penn., leading Alves-Araújo & Alves (2012a) to transfer P. congestifolia to Chromolucuma. Chrysophyllum sensu Pennington comprises species with pentamerous flowers, no staminodes (or with fewer than the corolla lobes) and copious endosperm, a combination of characters that is also suggestive of some Pouteria spp. We therefore do not exclude the possibility that some of the Neotropical species of Chrysophyllum and/or Pouteria belong to other genera. Since many new sapotaceous species have been recently described (Alves-Araújo & Alves, 2011, 2012a, b; Morales, 2012; Terra-Araujo, Faria & Vicentini, 2012a; Santamaría-Aguilar, Chaves-Fallas & Aguilar, 2017), deficiencies in the diagnostic delimitation are of great concern and there is a scientific desire to reach nomenclatural stability. Pennington’s (1990, 1991) infrageneric classification of Chrysophyllum and Pouteria recognized six sections in Chrysophyllum (Aneuchrysophyllum Engl., Chrysophyllum, Donella Pierre ex Baill. Prieurella Pierre, Ragala Pierre and Villocuspis (A. DC.) Aubrév. & Pellegr.) and nine in Pouteria (Aneulucuma, Antholucuma, Franchetella, Gayella, Oligotheca, Oxythece, Pierrisideroxylon, Pouteria and Rivicoa). Five sections of Chrysophyllum and eight of Pouteria are present in the Neotropics. Most of them stem from taxa established in the late 1800s by Pierre (1890, 1891) and Baillon (1891), taxa that Aubréville (1964a) still recognized as distinct genera, but were reduced to sections by Pennington (Table 1). The objectives of this study are to investigate the phylogenetic relationships of the Neotropical Chrysophylloideae based © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE 29 Table 1. Overview of Pennington’s classification of Neotropical Chrysophylloideae (Sapotaceae) with satellite genera including year of publication, type species and equivalent name in current classification Genus or section, Type species Pennington’s classification Chromolucuma rubriflora Ducke Chromolucuma rubriflora Ducke Publication year and segregate genus Chromolucuma Ducke (2: 2) Chrysophyllum section Aneuchrysophyllum (11: 3) 1891 Chloroluma Baill. Chloroluma gonocarpa (Mart. & Eichler) Baill. ex Aubrév. 1891 Cornuella Pierre Cornuella venezuelanensis Pierre 1891 Martiusella Pierre Martiusella imperialis (K.Koch & Fintelm.) Pierre 1936 Achrouteria Eyma Achrouteria pomifera Eyma Chrysophyllum section Chrysophyllum (17: 2) 1753 Chrysophyllum L. 1794 Nycterisition Ruiz & Pavon 1838 Guersentia Raf. Chrysophyllum section Prieurella (5: 5) 1891 Prieurella Pierre Chrysophyllum section Ragala (4: 3) 1891 Ragala Pierre Chrysophyllum section Villocuspis (6: 1) 1961 Villocuspis (A.DC.) Aubrév. & Pellegr. Diploon Cronquist (1: 1) Ecclinusa Mart. (11: 3) Elaeoluma Baill. (4: 3) Micropholis* (Griseb.) Pierre (38: 6) 1890 Crepinodendron Pierre 1891 Meioluma Baill. 1891 Platyluma Baill. 1962 Paramicropholis Aubrév. & Pellegr. Pouteria section Aneulucuma (21: 0) 1890 Urbanella Pierre Pouteria section Antholucuma (13: 4) 1890 Radlkoferella Pierre Chrysophyllum cainito L. Nycterisition ferrugineum Ruiz & Pav. Chrysophyllum gonocarpum (Mart. & Eichler) Engl. Chrysophyllum venezuelanense (Pierre) T.D.Penn. Chrysophyllum imperiale (K.Koch & Fintelm.) Benth. & Hook.f. Chrysophyllum pomiferum (Eyma) T.D.Penn. Guersentia oliviformis (L.) Raf. Chrysophyllum cainito L. Chrysophyllum argenteum subsp. ferrugineum (Ruiz & Pav.) T.D.Penn. Chrysophyllum oliviforme L. Prieurella cuneifolia (Rudge) Aubrév. Chrysophyllum cuneifolium (Rudge) A.DC. Ragala sanguinolenta Pierre Chrysophyllum sanguinolentum (Pierre) Baehni Villocuspis flexuosa (Mart.) Aubrév. & Pellegr. Diploon cuspidatum (Hoehne) Cronquist Ecclinusa ramiflora Mart. Elaeoluma schomburgkiana (Miq.) Baill. Micropholis rugosa (Sw.) Pierre Crepinodendron crotonoides Pierre Meioluma guianensis Baill. Chrysophyllum flexuosum Mart. Platyluma calophylloides (Pierre) Baill. Paramicropholis acutangula (Ducke) Aubrév- & Pellegr. Diploon cuspidatum (Hoehne) Cronquist Ecclinusa ramiflora Mart. Elaeoluma schomburgkiana (Miq.) Baill. Micropholis rugosa (Sw.) Pierre Micropholis crotonoides (Pierre) Pierre Micropholis venulosa (Mart. & Eichler ex Miq.) Pierre Micropholis venulosa (Mart. & Eichler ex Miq.) Pierre Micropholis acutangula (Ducke) Eyma Urbanella procera (Mart.) Pierre Pouteria procera (Mart.) K.Hammer Radlkoferella multiflora (A.DC.) Pierre Pouteria multiflora (A.DC.) Eyma © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 30 A. D. DE FARIA ET AL. Table 1. Continued Genus or section, Type species Pennington’s classification Labatia sessiliflora Sw. Franchetella tarapotensis Pierre Pouteria sessiliflora (Sw.) Poir. Pouteria tarapotensis (Eichler ex Pierre) Baehni Pouteria dictyoneura (Griseb.) Radlk. Pouteria gardneri (Mart. & Eichler) Baehni Pouteria sagotiana (Baill.) Eyma Pouteria ramiflora (Mart.) Radlk. Pouteria engleri Eyma Publication year and segregate genus Pouteria section Franchetella (62: 29) 1788 Labatia Sw. 1890 Franchetella Pierre 1890 Paralabatia Pierre 1891 Discoluma Baill. 1891 1891 Eremoluma Baill. Microluma Baill. 1891 Nemaluma Baill. 1891 Podoluma Baill. 1891 Pseudocladia Pierre 1962 Sandwithiodoxa Aubrév. & Pellegr. Peteniodendron Lundell 1976 1983 Piresodendron Aubrév. & Pellegr. Pouteria section Gayella (8: 1) 1890 Gayella Pierre Paralabatia dictyoneura (Griseb.) Aubrév. Discoluma gardneri (Mart. & Eichler) Baill. Eremoluma sagotiana Baill. Microluma parviflora (Benth. ex Miq.) Baill. Nemaluma engleri (Eyma) Aubrév. & Pellegr. Podoluma catocladantha (Eichler) Aubrév. Pseudocladia lateriflora (Benth. ex Miq.) Pierre Sandwithiodoxa egregia (Sandwith) Aubrév. & Pellegr. Peteniodendron belizense Lundell Piresodendron ucuqui (Pires & R.E.Schult.) Le Thomas Pouteria gardneri (Mart. & Eichler) Baehni Pouteria ramiflora (Mart.) Radlk. Pouteria egregia Sandwith Pouteria durlandii (Standl.) Baehni Pouteria ucuqui Pires & R.E.Schult. 1891 Myrtiluma Baill. 1925 Barylucuma Ducke Gayella valparadisaea (Molina) Pierre Myrtiluma eugeniifolia (Pierre) Aubrév. Barylucuma decussata Ducke Pouteria section Oligotheca (8: 3) 1925 Syzygiopsis Ducke Syzygiopsis oppositifolia Ducke Pouteria oppositifolia (Ducke) Baehni Caramuri opposita (Ducke) Aubrév. & Pellegr. Oxythece leptocarpa Miq. Pseudoxythece ambelaniifolia (Sandwith) Aubrév. Pouteria opposita (Ducke) T.D.Penn. Pouteria elegans (A.DC.) Baehni Pouteria ambelaniifolia (Sandwith) T.D.Penn. Pouteria guianensis Aubl. Guapeba laurifolia Gomes Pouteria guianensis Aubl. Pouteria caimito (Ruiz & Pav.) Radlk. Pouteria gomphiifolia (Mart. ex Miq.) Radlk. Pouteria guianensis Aubl. Pouteria lucens (Mart. & Miq.) Radlk. Pouteria pariry (Ducke) Baehni Pouteria section Oxythece (11: 4) 1961 Caramuri Aubrév. & Pellegr. 1961 1972 Neoxythece Aubrév. & Pellegr. Pseudoxythece Aubrév. Pouteria section Pouteria (48: 19) 1775 Pouteria Aubl. 1812 Guapeba Gomes 1891 Gomphiluma Baill. Gomphiluma martiana Baill. 1891 1891 Krugella Pierre Leioluma Baill. 1962 Eglerodendron Aubrév. & Pellegr. Krugella hartii Pierre Leioluma lucens (Mart. & Miq.) Baill. Eglerodendron pariry (Ducke) Aubrév. & Pellegr. Pouteria splendens (A.DC.) Kuntze Pouteria eugeniifolia (Pierre) Baehni Pouteria decussata (Ducke) Baehni © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE 31 Table 1. Continued Genus or section, Type species Pennington’s classification Pseudolabatia psammophila (Mart.) Aubrév. Pouteria psammophila (Mart.) Radlk. Lucuma bifera Molina Pouteria lucuma (Ruiz & Pav.) Kuntze Pouteria macrophylla (Lam.) Eyma Pouteria speciosa (Ducke) Baehni Pradosia lactescens (Vell.) Radlk. Sarcaulus brasiliensis (A.DC.) Eyma Publication year and segregate genus 1962 Pseudolabatia Aubrév. & Pellegr. Pouteria section Rivicoa (10: 5) 1782 Lucuma Molina 1890 Richardella Pierre 1891 Englerella Pierre Pradosia Liais (25: 5) Sarcaulus Radlk. (5: 1) Richardella macrophylla (Lam.) Aubrév. Englerella macrocarpa Pierre Pradosia glycyphloea (Casar.) Liais Sarcaulus brasiliensis (A.DC.) Eyma Numbers in parentheses refer to accepted species in the Flora Neotropica Monograph (Pennington, 1990), followed by the number of sampled accessions. Asterisks (*) indicate conserved names. Species sampled in this study appear in bold type in the right column. on nuclear ribosomal [external and internal transcribed spacer (ETS, ITS)] and nuclear (RPB2) DNA sequence data and a set of morphological characters using Bayesian inference and parsimony jackknifing. Our primary goals are to test (1) the monophyly of the Neotropical genera in Chrysophylloideae; (2) whether Pennington’s (1990, 1991) sections in Chrysophyllum and Pouteria are natural groups; (3) whether they correspond to groups recognized by Aubréville (1964a); and (4) whether it is possible to establish diagnostic morphological character sets for the recovered clades. MATERIAL AND METHODS NOMENCLATURE AND TAXON SAMPLING Pennington’s (1990, 1991) classification of Neotropical Chrysophylloideae was used for the taxa we sampled. Accepted names are available from the family checklist (Govaerts et al., 2001) and online at the World Checklist of Selected Plant Families, Royal Botanic Gardens, Kew (http://apps.kew.org/wcsp). The number of samples in this study has been expanded from 22 (Swenson et al., 2008a) to 101. Terminal taxa, voucher information and GenBank accession numbers are reported in the Appendix. In addition to Chrysophyllum and Pouteria, our sample includes members of all presently accepted genera in the New World, i.e. Chromolucuma, Diploon, Ecclinusa, Elaeoluma, Micropholis, Pradosia and Sarcaulus. Phylogenetic estimates suggest that most form a group confined to the New World (Swenson et al., 2008a; Bartish et al., 2011). There are also indications that Micropholis is sister to all other sampled taxa (Bartish et al., 2011), making it the most appropriate outgroup for the study of Neotropical Chrysophylloideae. The sample includes the type species of all genera (except Pradosia), three out of eight sections of Pouteria and three out of five sections of Chrysophyllum present in the Neotropics (Pennington, 1991). MOLECULAR DATA Broad molecular phylogenetic studies in Sapotaceae have revealed plastid DNA regions to contain low numbers of informative sites (1–3%) for phylogenetic estimates (Swenson et al., 2013). In contrast, molecular sequences of nuclear ribosomal DNA (rDNA) ITS1 and ITS2 (including complete 5.8S and parts of 18S and 26S), the ETS and the low copy nuclear gene RPB2 (Oxelman & Bremer, 2000) have proved useful for phylogenetic inference in the family (Bartish et al., 2005; Swenson et al., 2007a, 2008a, 2013). We have here focused on these molecular markers and publish for the first time 80 ITS sequences, 96 ETS sequences and 100 RPB2 sequences. Total DNA was extracted from leaves dried in silica gel. Extraction, amplification and primers for ETS, ITS and RPB2 followed the protocol described by Swenson et al. (2013), but also those of Bartish et al. (2005) for ITS and Swenson et al. (2008b) for ETS. Purified products of rDNA were sequenced with an ABI3130xl Automated DNA Sequencer (Applied Biosystems, Foster City, CA, USA). One drawback of ITS and ETS is that multiple copies of them occur in a typical plant genome and, depending on which copy is amplified and sequenced, inaccurate phylogeny could be reconstructed (Álvarez © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 32 A. D. DE FARIA ET AL. & Wendel, 2003; Poczai & Hyvönen, 2010; Naciri & Linder, 2015). To identify samples with multiple copies, we carefully checked for double peaks in the proofreading procedure, with PCR products being subsequently cloned, using the TOPO-TA Cloning Kit for Sequencing (Invitrogen, Carlsbad, CA, USA) and following the manufacturer’s instructions. DNA from clones were amplified by PCR using the specific plasmid M13F and M13R primers following the TOPO-TA Cloning Kit manual, with subsequently purified products sequenced using the M13F primer in an ABI313xl Automated DNA Sequencer, yielding between four and ten slightly different repeats in 13 species. DATA MATRIX CONSTRUCTION Sequences of ETS, ITS and RPB2 were partitioned into separate matrices and aligned in MAFFT v.6.818b (Katoh, Asimenos & Toh, 2009) using the l-insi predefined parameter setting. Resulting alignments were checked for similarity (Simmons, 2004), with subsequent minor manual adjustments to reduce potentially false homologies. Inferred gaps were coded as additional binary characters following the guidelines of Simmons & Ochoterena (2000). Gaps were assigned a simple substitution model allowing unconstrained reversible gains/losses of characters. The sequence data include introns (ETS, ITS1, ITS2) and exons (5.8S, 18S, 26S, RPB2) and it is possible that jModelTest (Posada, 2008) could identify less parameter-rich substitution models for phylogenetic inference. We therefore constructed a second matrix in which introns and exons, respectively, were combined in two partitions. The original matrices (ITS, ETS, RPB2) with gap coding are available as Supplementary data on the journal website. Accessions of 13 species with variable repeats (copies) of ITS or ETS were handled in two ways. The first matrix (called incomplete) was aligned with all variable repeats, but only the first repeat of each taxon was concatenated with the other unique marker, with all other copies concatenated with question marks. The second matrix (called complete) was a duplicate of the first, but the question marks were substituted with the available sequences of ITS or ETS. For example, for Chromolucuma cespedisiiformis J.F.Morales, four ETS copies and only one ITS sequence are available. Therefore, in the incomplete matrix it had four ETS entries, combined with one ITS sequence, and three entries with question marks. In the complete matrix, it was represented by four different ETS entries and four identical ITS sequences. Such matrices of 101 samples yielded 189 terminals, distributed in four types of matrices: (1) incomplete based on markers; (2) complete based on markers; (3) incomplete based on introns/exons; and (4) complete based on introns/exons. PHYLOGENETIC ANALYSES Phylogenetic relationships were estimated with Bayesian inference (Rannala & Yang, 1996; Yang & Rannala, 1997) and parsimony jackknifing (Farris et al., 1996). We used MrBayes v3.2.1 (Ronquist & Huelsenbeck, 2003) and the BEAST 1.7.5 package (Drummond et al., 2012) for phylogenetic reconstruction. To identify the best performing model for each separate partition we examined the relative fit of various nucleotide substitution models for ETS, ITS and RPB2 and those for introns and exons. Our selection was based on the Akaike information criterion (AIC; Akaike, 1974), implemented in jModelTest (Posada, 2008). Also, since ETS and ITS are part of the same transcription unit and not considered independent datasets (Baldwin & Markos, 1998), we concatenated these and ran them against RPB2 to reveal plausible supported incongruence. Phylogenetic inference in MrBayes was run for ten million generations with Markov chain Monte Carlo (MCMC), starting from random trees and flat priors (default). We used three heated chains and a single cold chain, with trees sampled every 5000th generation, producing 2001 trees per run. Markov chains were assumed stationary when the log-likelihood values reached a stable equilibrium (Huelsenbeck & Ronquist, 2001), with standard deviation decreased to < 0.05 and parameters gained effective sample size (ESS) values > 200. Majority-rule consensus and posterior probabilities (PP) for nodes were assembled from all post burn-in sampled trees. Phylogenetic reconstructions for each dataset were estimated using three independent runs to confirm that they converged on similar stationary parameter estimates. Aligned partitions of full-length sequences, respectively exons and introns, were prepared with BEAUti (part of the BEAST package) for an output xml-file for phylogenetic inference in BEAST. Substitution models (Posada, 2008) were set by manual modification of the rate parameters in the xml-file. The BEAST package was used primarily to derive a tree topology and not for divergence time estimates under a molecular clock assumption. The molecular clock was therefore unconstrained and the root was fixed by using a normal prior with an arbitrary mean (100) and a narrow standard deviation (0.1). The tree prior was set to a birth–death process (Gernhard, 2008). To ensure independent convergence on all parameters (ESS > 200), MCMCs were set to run five times, each for 30 million generations and sampling trees every 15 000 generations. Convergence and chain mixing were checked © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE using Tracer v.1.5 (Rambaut & Drummond, 2009). A proportion of samples in each run were discarded as burn-in, with the posterior set of trees summarized in TreeAnnotator (part of the BEAST package). The resulting maximum clade credibility (MCC) tree was then visualized with FigTree v1.3.1 (Rambaut, 2009). To obtain jackknife values (JK) for clades, parsimony jackknifing, as implemented in PAUP* 4.0 (Swofford, 2002), was performed using the following settings: 1000 jackknife replicates with a single random addition sequence; TBR branch swapping, saving a maximum of ten trees; collapsing branches if minimum length was zero; and steepest descent not in effect. The fraction of excluded characters per replicate was set to 37%. Support values for nodes are given as posterior probability (PP) and parsimony JK values. PP values of 0.80–0.94 were considered to be weak to moderate and those of 0.95–1.00 to be strong indicators of node support. JK values of 50–74% were considered weak, 75–89% moderate and 90–100% strong. Nodes with support < PP 0.8 and JK 50% were considered too weak and collapsed in the phylogenetic tree. MORPHOLOGICAL DATA Selecting useful diagnostic characters of morphology represents a major challenge, since high levels of morphological homoplasy in Sapotaceae have been revealed in earlier studies (Swenson & Anderberg, 2005; Swenson et al., 2007a, 2008a, b, 2013, 2015). Observed homoplasy may stem from poorly understood morphology, delimitation of character and character states, inconsistent terminology, inadequately known or cryptic species or a combination of these reasons. From nearly 50 surveyed characters, we arrived at 17 that (1) were frequently used to delimit Neotropical genera and sections of Chrysophyllum and Pouteria, (2) were the least ambiguous and (3) showed the least homoplasy across the molecular tree (Supporting Information Table S1). The data were collected from the literature (Pennington, 1990, 1991, 2006; Ribeiro et al., 1999; Roosmalen & Garcia, 2000; Morales, 2012) and checked against available live and/or herbarium material deposited at FUEL, HPL, IAC, INPA, S, SPF and UEC, abbreviations according to Index Herbariorum (Thiers, continuously updated). The morphological terminology follows that of Harris & Harris (1997). Character 1. Stipules are scattered across Sapotaceae, but unusual in subfamily Chrysophylloideae. Large, persistent stipules are present in Chromolucuma, Ecclinusa and in some Pouteria spp. Character 2. Leaf venation was used by Pennington (1990) as a rich source of characters in his 33 classification. The most common patterns are eucamptodromous and brochidodromous. However, combinations of them are frequently observed, when the lower half of the blade is eucamptodromous and a portion of the upper half is brochidodromous. Such mixed venation patterns, described by Ellis et al. (2009), are here called eucampto-brochidodromous. Character 3. Tertiary veins generally form five patterns (reticulate, oblique, admedial, striate and horizontal; Pennington, 1990). Reticulate tertiaries form an irregular pattern; oblique tertiaries usually cross between and anastomose with the secondaries at an angle of c. 90°; admedial tertiaries descend from the leaf margin and usually do not reach the midrib; striate tertiaries are formed by several closely spaced veins parallel to the secondaries; and horizontal tertiaries are perpendicular to the midrib and fuse with it. Character 4. Areolate venation is sometimes present, usually in the form of small round areas between the tertiaries (Munzinger & Swenson, 2009). However, in some cases the tertiaries are nearly impossible to distinguish and instead an areolate venation is directly formed as a higher order of venation. Character 5. White latex is one of the important diagnostic features for Sapotaceae, but in rare cases it can be cream or yellowish (Pennington, 1990; Ribeiro et al., 1999) or even bluish, as in some Pycnandra spp. (Swenson & Munzinger, 2016). As far as is known, latex colour does not change due to oxidation. Characters 6 and 7. Flowers of Neotropical Chrysophylloideae are actinomorphic and usually isomerous. The calyx is uniseriate, with four, five or six sepals (rarely more), whereas the corolla has four, five, six or seven lobes. These characters were important in Aubréville’s (1964a) classification, but since the flower parts can vary in a single species, they were down-weighted by Pennington (1990). Given that the number of sepals and corolla lobes is inconsistent with natural groups in New Caledonia (Swenson et al., 2008b), we explore the possibility that this ratio shows another pattern in South American Chrysophylloideae. Four general flower types may be distinguished: (1) tetramerous; (2) pentamerous; (3) four sepals with six petals; and (5) five sepals that are outnumbered by the petals. Character 8. Accrescent calyx refers to the continued growth of the calyx post-anthesis, resulting in a fruit with a persistent large calyx. This is a rare feature in Chrysophylloideae, found only in some species. Character 9. The corolla of Sapotaceae is sympetalous, with a tube and free corolla lobes. Corolla form played an important role in Pennington’s © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 34 A. D. DE FARIA ET AL. (1990) circumscription of different taxonomic groups. Flower characterization is difficult and subjective, but, in this study, we use the types proposed by Swenson et al. (2013), with some modifications. Depending on the length of sepals, corolla tube and lobes and the general shape of the corolla, flowers may be rotate, dome- or cup-shaped, tubular, urnshaped or campanulate. Rotate flowers have a short calyx and corolla tube, with the corolla lobes spread to form almost flat flowers (Fig. 1A). Rotate flowers should not be difficult to separate from domeand cup-shaped ones, which have slightly longer Figure 1. Flowers and corolla types of Neotropical Chrysophylloideae (Sapotaceae). A, Pradosia cochlearia, a large tree from the Amazonian with rotate flowers. B, Pouteria minima, a species with green flowers and dome-shaped corolla. C, Pouteria reticulata, one of a few species with white, dome-shaped flowers. D, Pouteria campanulata with cup-shaped flowers. E, Pouteria caimito, a widely distributed species in South America with tubular flowers. F, Micropholis caudata with urn-shaped flowers. G, Chrysophyllum oliviforme with campanulate flowers. Photographs: A–F by Aparecida de Faria, G by Hugh Nicholson. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE calyces usually reaching the tube orifice and the lobes forming a kind of dome (Fig. 1B, C), or somewhat spreading (Fig. 1D), usually hiding the pistil. Tubular flowers have a cylindrical corolla tube that is longer than the sepals and usually have erect corolla lobes (Fig. 1E). Urn-shaped flowers (not present in the sampled taxa) are similar to tubular flowers, differing from them in having rather short calyx lobes, a prominent corolla tube that is broadest at the mid-section and has a contracted mouth and short recurved lobes (Fig. 1F). Campanulate flowers are similar to tubular and/or urn-shaped flowers, but their corolla tube is broadest at the tube orifice, forming a small bell (Fig. 1G). Some taxa with unisexual flowers are difficult to code since the pistillate flowers (female) are tubular and the staminate flowers (male) are cup-shaped, as in Chrysophyllum sparsiflorum Klotzsch ex Miq. Character 10. Corolla margin in most species is glabrous, but is ciliate (and resembling trichomes) in some taxa and papillate in others. Character 11. Anther filaments are inserted at various levels in the corolla tube. They can be inserted in the tube orifice, just below the tube orifice, in the mid-section of the tube or near the base. Pennington (1990, 1991) considered this character of low taxonomic value due to its wide homoplasticity in his recognized groups. In contrast, Swenson et al. (2007a, 2013), using molecular data, have demonstrated it to be quite consistent in the clades recovered in their analyses and it is therefore explored here. Character 12. Pollen morphology of Sapotaceae has been surveyed for 398 species and classified into 12 pollen types (Harley, 1991). These types were used by Pennington, but numbered differently in his Flora Neotropica (1990) and The Genera of Sapotaceae (1991). Pollen types vary in the genera sensu Pennington (1991), with Chrysophyllum, for example, represented by four and Pouteria by eight pollen types. Pollen types in Chrysophylloideae, in a phylogenetic context, were discussed for the first time by Swenson et al. (2008a), who found some patterns consistent with clades. We repeat that exercise here, following Harley’s (1991) classification, even though data are missing for 48% of the sampled taxa. Character 13. Pennington (1990, 1991) used absence of staminodes as a diagnostic character to separate Chrysophyllum from Pouteria. However, Swenson et al. (2008a) demonstrated that staminodes had been lost multiple times across the subfamily and therefore they represent a character that is not as useful as commonly believed. Nevertheless, with an expanded sampling of Neotropical species, their distribution (presence, vestigial, absence) could still be diagnostic. Staminodes are inserted in the 35 corolla lobe sinuses and are scored as vestigial, as long as one (or more) is missing (or reduced). Character 14. Fruits of Sapotaceae are usually classified as berries with a leathery or somewhat woody outer pericarp and an inner fleshy endocarp that serves as the pulp (Pennington, 1991). However, fruits in Pradosia have been called drupes, but drupaceous is a better term since the endocarp differentiates and the inner layer becomes jelly-like, partially transparent, but never hard as in a true drupe (Terra-Araujo et al., 2013). Fruits are distinguished here as berries and drupaceous. Character 15–17. Three seed characters have been frequently used in the systematics of Sapotaceae: distinction between foliaceous or plano-convex cotyledons; radicle included in the cotyledons or exserted below the commissure; and presence or absence of an endosperm (Pennington, 1990, 1991). Phylogenetic analyses have confirmed that these characters covary in the phylogeny of Sapotaceae (Swenson et al., 2007a, 2008b, 2013; Gautier et al., 2013) and result in three types: foliaceous cotyledons, with the radicle extending below the commissure, and the presence of endosperm (Type 1); plano-convex cotyledons, with the radical not exserted, and no endosperm (Type 2); and planoconvex cotyledons, with radicle exserted, and no endosperm (Type 3). These morphological characters were added to the molecular dataset and mapped as discrete units as implemented in BEAST (Lemey et al., 2009). However, this approach was not feasible (see Results) and morphological characters were mapped on the molecular MCC tree obtained from the BEAST analyses using MacClade 4.0 (Maddison & Maddison, 2005). RESULTS PARTITIONS AND MULTIPLE COPIES The number of aligned characters is 2664, including 424 base pairs of ETS, 918 of ITS, 1230 of RPB2 and gaps. If the sequences are partitioned as exons and introns, there are 938 and 1634 nucleotides, respectively. The relative fit of various models of nucleotide substitution and aligned partitions are reported in Table 2. Multiple copies of ETS and ITS were never found in the same species. Multiple copies of ETS were found and cloned in ten species: Chromolucuma cespedisiiformis (four copies), Pouteria campanulata (Kunth) Baehni (seven copies), P. cuspidata (A.DC.) Baehni subsp. cuspidata (seven copies), P. eugeniifolia (Pierre) Baehni (six copies), P. freitasii T.D.Penn. (five copies), P. minima T.D.Penn. (ten copies), P. pallens T.D.Penn. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 36 A. D. DE FARIA ET AL. Table 2. Characteristics of nuclear sequences for each data partition and substitution model based on the Akaike information criterion (Akaike, 1974) in this phylogenetic analysis of Neotropical Chrysophylloideae (Sapotaceae) Data ETS ITS RPB2 Exons Introns Aligned length (bp) 424 918 1230 938 1634 Number of characters Constant Uninformative Potentially informative 75 293 619 550 394 45 195 332 242 335 304 (71.7%) 430 (46.8%) 279 (22.7%) 146 (15.6%) 905 (55.4%) (six copies), Pradosia decipiens Ducke (eight copies), P. schomburgkiana (A.DC.) Cronquist (three copies) and Sarcaulus brasiliensis (A.DC.) Eyma (nine copies). Similarly, four species contained multiple copies of ITS and were cloned: Chrysophyllum prieurii A.DC. (eight copies), Pouteria ambelaniifolia (Sandwith) T.D.Penn. (nine copies), P. fulva T.D.Penn. (ten copies) and Pouteria sp. from Tree183 in Picinguaba (Ubatuba, SP, Brazil) (ten copies) (Appendix). Separate analyses of the ETS+ITS and RPB2 recovered weak support for the backbone of the phylogenetic trees, with no supported (PP > 0.8) incongruence between major clades, except for two minor cases. Analyses of the ETS+ITS recovered Pouteria anomala T.D.Penn. and P. engleri Eyma, apart from Micropholis, as sister to all other taxa, whereas analysis of the RPB2 recovered them in a moderately (PP 0.93) supported polytomy similar to what is reported in Figure 2. The other case is represented by a single species, P. ramiflora (Mart.) Radlk., which falls as sister to clade N, when ETS+ITS is used (PP 1.0), but inside clade P (PP 0.93) when RPB2 is used. As will be shown below, neither of these cases has an impact on the bigger picture reported in this paper and therefore we kept the entire sample and concatenated the ETS+ITS and RPB2 in other analyses. TREE TOPOLOGY All analyses (MrBayes, BEAST, Jackknife) using nuclear data partitioned either by loci or exons/introns identified the same backbone phylogenetic relationships with similar support values, but with slightly lower support and less resolved topologies when incomplete matrices were used. Pouteria laevigata (Mart.) Radlk. and P. maxima T.D.Penn. clearly belong to Micropholis and had to be moved to the outgroup in order to recover a monophyletic ingroup. Analyses of the incomplete matrices rendered most species monophyletic or as members of clades with no hard incongruence. Only Pouteria ambelaniifolia, P. eugeniifolia and two Pradosia spp. Gaps Model 25 51 16 1 91 TIM3+G TPM1uf+G TPM2uf+G TVM+G TVM+G proved to be exceptions to this pattern. We therefore analysed the data using the complete matrices, with question marks substituted by the corresponding ETS or ITS sequences. Once more, all species except P. ambelaniifolia and the two Pradosia spp. were recovered as monophyletic, but with an improved overall resolution and support (Fig. 2). Bayesian inference of molecular sequence data and morphological data mapped as discrete units in BEAST failed to reach ESS values > 200 for several critical parameters. We therefore reran the analyses excluding all but four fruit characters, with the same result for ESS values. This outcome is interpreted as evidence for the morphological model failing to provide an unambiguous signal in informing the compound likelihood of the analysis, which translates to extensive morphological homoplasy (Swenson & Anderberg, 2005; Swenson et al., 2007a, 2008a, b). Overall tree resolution may be envisaged to include 17 supported clades (A–Q), plus a few species of uncertain affinity. Our analysis recovered Chrysophyllum and Pouteria, with most sections of the latter, as polyor paraphyletic in their current circumscriptions. Chrysophyllum sensu Pennington (1990) is divided into five sections, of which sections Ragala (clade B), Prieurella (clade F) and Chrysophyllum (clade H) are mutually monophyletic, but not closest relatives to each other. Members of C. sections Aneuchrysophyllum and Diploon are indicated as affiliated, but only with weak branch support (clade G). Species currently classified in Pouteria are scattered throughout the phylogenetic trees, from the outgroup (clade A) to Pouteria section Oxythece (clade Q). Pouteria sections Oxythece and Rivicoa are the only monophyletic groups that correspond to Pennington’s classification. Sections Antholucuma, Franchetella and Pouteria are all polyphyletic. The small genera Elaeoluma (clade C), Ecclinusa (clade E) and Pradosia (clade K) are all monophyletic and recovered with strong support. Nevertheless, it must be noted that different copies of Pradosia decipiens and P. schomburgkiana from the same © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE 37 Figure 2. Phylogenetic tree obtained from Bayesian inference and parsimony analyses of nuclear sequences of Neotropical Chrysophylloideae (Sapotaceae). Sections of Pouteria are colour-coded according to the legend. Species containing several ETS or ITS copies are illustrated with a black triangle, followed by the name and the number of obtained copies. Posterior probabilities (above) and parsimony jackknifing (below) are indicated along the branches. The type species of genera recognized by Pennington (1990) are indicated in bold. Clades A–Q are discussed in the text. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 38 A. D. DE FARIA ET AL. source forced accessions to group in two clades. This suggests hybridization or incomplete lineage sorting. Evolutionary history is investigated elsewhere (TerraAraujo et al., 2015). Multiple accessions of Pouteria caimito (Ruiz. & Pav.) Radlk, P. macrophylla (Lam.) Eyma, P. manaosensis (Aubrév. & Pellegr.) T.D.Penn. and Ecclinusa guianensis Eyma. were included. All five accessions of P. caimito fell in clade N, but three accessions from the Atlantic forest grouped together with strong support, whereas one specimen from the same biome was recovered in close proximity to a specimen from the Amazon forest. All accessions of P. macrophylla and P. manaosensis grouped in clade I, but not as distinct species. The two accessions of E. guianensis grouped together (clade E). Phylogenetic relationships of Diploon cuspidatum (Hoehne) Cronquist, Chrysophyllum imperiale (Linden ex K.Koch & Fintelm.) Benth. & Hook.f., C. pomiferum (Eyma) T.D.Penn., C. venezuelanense (Pierre) T.D.Penn., Pouteria oblanceolata and Sarcaulus brasiliensis remained unresolved. OPTIMIZATION OF MORPHOLOGY To identify diagnostic character combinations (Figs 3, 4), 17 morphological characters were mapped on the molecular MCC tree obtained from the BEAST analysis. The majority of features were highly variable, concurring with previous studies (Swenson & Anderberg, 2005; Swenson et al., 2007a, 2008a, b). However, several characters were congruent with more inclusive clades. The number of sepals and corolla lobes and the three seed characters are presented in Figures 3F and 4F, respectively. Character distribution is discussed below. DISCUSSION Our phylogenetic analyses of 101 terminal taxa of Neotropical Chrysophylloideae unambiguously demonstrate that Pennington’s (1990, 1991) classification of Chrysophyllum and Pouteria is untenable (Fig. 2). Depending on clade circumscription, five to 17 clades (clades A–Q) are recovered with various success of morphological coherence (Figs 3, 4). Terminals of Chrysophyllum are distributed in four to five groups, with C. section Ragala corresponding to the genus Ragala (clade B), section Prieurella corresponding to the genus Prieurella (clade F) and section Chrysophyllum possibly corresponding to a narrow circumscription of Chrysophyllum (clade H). Petersen, Parker & Potter (2012) used ITS sequences to demonstrate that five to six species of C. section Chrysophyllum, including the resurrected C. bicolor Poir, are close relatives of the generic type C. cainito L. In that study, other species of the genus seemed to be distantly related to C. section Chrysophyllum, but their conclusions were limited by the inclusion of only two Pouteria spp. apart from those of Chrysophyllum. In our analyses, and depending on delimitation, accessions of Pouteria are recovered in some ten clades, with a topology that does not concur with Aubréville’s or Pennington’s classifications (cf. Appendix), but strongly supports recent conclusions based on plastid DNA, nrDNA and morphology (Swenson & Anderberg, 2005; Swenson et al., 2008a; Bartish et al., 2011) that both Chrysophyllum and Pouteria remain ‘catchall’ baskets, despite recent resurrections of several Australasian genera (Swenson et al., 2007a, 2013). Multiple copies of ETS or ITS were found in 13 species, all but three species being monophyletic. The three non-monophyletic species are Pouteria ambelaniifolia, Pradosia decipiens and Pradosia schomburgkiana, but accessions of each one are recovered near each other in three different subclades and do not challenge the overall topology, i.e. the taxa with multiple copies of ETS or ITS represent local problems, not hindering rebuilding of a natural classification of Neotropical Chrysophylloideae. UTILITY OF MAPPED MORPHOLOGICAL CHARACTERS Morphological characters in Sapotaceae have been demonstrated to be homoplastic and their use in phylogenetic analysis, until their homology is better understood, is bound to introduce noise (Swenson & Anderberg, 2005; Swenson et al., 2007a, 2008a, b, 2013, 2015). The present study supports that notion. Seventeen characters have been selected for evaluation of their usefulness for taxon characterization. Since the current generic classification to a large extent does not agree with the proposed tree topology, in the following discussion we therefore refer to clades A–Q instead of generic names. Stipules Small caducous stipules are scattered across Sapotaceae, e.g. being diagnostic for the Malagasy tribe Tseboneae (Sapotoideae; Gautier et al., 2013). They are less common in Chrysophylloideae and are present only in the recovered clades E and L, except for Pouteria williamii (Aubrév. & Pellegr.) T.D.Penn. of the latter clade. Our clade E corresponds to Pennington’s Ecclinusa, whereas clade L would accord with Chromolucuma, if three Pouteria spp. were transferred to it (Fig. 3A). In fact, Alves-Araújo & Alves (2012a) transferred Pouteria congestifolia to Chromolucuma precisely based on presence of stipules and yellow latex. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE 39 Figure 3. Seven morphological characters mapped on the phylogenetic tree obtained from Bayesian and parsimony analyses of Neotropical Chrysophylloideae (Sapotaceae). Clades A–Q are discussed in the text. Black branches represent ambiguous character states. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 40 A. D. DE FARIA ET AL. Figure 4. Eight morphological characters mapped on the phylogenetic tree obtained from Bayesian and parsimony analyses of Neotropical Chrysophylloideae (Sapotaceae). Note that three characters are amalgamated in F. Clades A–Q are discussed in the text. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE 41 Secondary venation Accrescent calyx Leaf venation is a useful field character, but has been found to be fairly homoplasious (Swenson & Anderberg, 2005; Swenson et al., 2008a). Eucamptodromous and the mixed eucampto-brochidodromous venation types are the two most common patterns of secondary venation, although with variable coherence with the recovered clades (Fig. 3B). All members of clades B, E, F and J have eucamptodromous venation, with only clades M and N having the exclusively mixed eucampto-brochidodromous type. An accrescent calyx is rare in Chrysophylloideae, but present in all species of Chrysophyllum section Ragala (clade B). In fact, for Pierre (1891) it was the key character when he described the genus Ragala. Chrysophyllum eximium Ducke (not included) has this feature too, but was placed in C. section Aneuchrysophyllum on the basis of leaf venation with a note that it may belong in section Ragala (Pennington, 1990). Corolla form Tertiary venation Tertiary venation has been found to covary and be diagnostic for groups of Australasian Sapotaceae (Swenson et al., 2013). Five patterns of tertiary venation (reticulate, oblique, admedial, striate and horizontal) were distinguished and found to fit the phylogenetic tree slightly better than those of the secondary venation. Reticulate tertiaries are consistent with clades C, D, M and P and nearly so with clade N (Fig. 3C). Oblique tertiaries are found throughout clades B, E and F. Admedial tertiaries are present in clades H and Q, whereas striate tertiaries are restricted to some taxa of clade A (Micropholis). Higher-order leaf venation may be characterized by areolate patterns that are mostly diagnostic for clades K, N and O, although always with some exceptions (Fig. 3D). Latex Latex in Sapotaceae is usually white, but one clade of Neotropical species has yellow or yellowish latex (Fig. 3E), diagnostic of Chromolucuma (Pennington, 1991; Alves-Araújo & Alves, 2012a). A few species may also have cream latex, one being P. eugeniifolia, occurring in clade Q. Sepals and corolla lobes Pennington (1991) stressed that American Pouteria spp. fall into several well-defined groups based on the number of flower parts. Isomerous flowers with the same number of sepals and corolla lobes (four or five), often with some variation, are the dominating types (Fig. 3F). The pentamerous flower corresponds to the symplesiomorphic state, reduced three times, in clades J, N and P, to a tetramerous flower through evolution. Four species of Pouteria section Antholucuma (clade I) have four sepals and six corolla lobes. These form a well-supported grade to the members of P. section Rivicoa, which all have pentamerous flowers. Thus, the number of flower parts agrees well with our tree and, in combination with others, should be considered a useful character, as suggested by Pennington (1991). Classification of corolla forms, guided by the above definitions, is admittedly subjective and difficult. The cup-shaped corolla with spreading lobes is the most common type in Neotropical Chrysophylloideae (Fig. 4A). It is sometimes hard to separate from the slightly rounder dome-shaped corolla, with corolla lobes turned inward, not opening up as much as the former. These two types combine well with our clades and only a few taxa embedded among them have clearly different corolla types. The two exceptions are Chromolucuma cespedisiiformis, with campanulate corollas, embedded in clade L, and Pouteria eugeniifolia, with rotate corollas, embedded in clade Q. A tubular corolla is nearly consistent in clades I, J and N, but, as usual, with occasional exceptions. The rotate corolla is characteristic of Elaeoluma (clade C) and Pradosia (clade K), but also found in Diploon (part of clade G). It should also be kept in mind that corolla form variation may be due to sexual dimorphism. For example, Chrysophyllum venezuelanense (clade G) has cup-shaped pistillate flowers and campanulate staminate flowers, whereas C. sparsiflorum (in clade H) has tubular pistillate flowers and cup-shaped staminate flowers. Corolla margin Corolla margins in Chrysophylloideae are usually glabrous, without trichomes or papillae (Fig. 4B). However, all members of clade I have papillate corolla margins. Short cilia or trichomes are present in Chrysophyllum imperiale (of clade G), clade J and several taxa of clades M, N and O. Stamen insertion Stamens in Sapotaceae are opposite the corolla lobes and the filaments are inserted to the corolla tube at different levels. In addition, at the point of insertion, the filament can be completely fused with the corolla tube tissue or run above the corolla tube like a keel. The view of these features has varied, being considered taxonomically important (Aubréville, 1964a) or of © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 42 A. D. DE FARIA ET AL. little value (Pennington, 1991). In a phylogenetic context, the level of insertion is diagnostic for groups in Australasia (Swenson et al., 2007a). We believe that the level of insertion is quite consistent and useful across the South American clades (Fig. 4C). Stamens inserted in the tube orifice seem to be diagnostic for Micropholis (clade A), Elaeoluma (clade C), Chrysophyllum section Chrysophyllum (part of clade H), Diploon (part of clade G), clade I [except Pouteria multiflora (A.DC.) Eyma], Pradosia (clade K) and Sarcaulus and for some scattered species in clades O–Q. Stamens inserted just below the tube orifice are found in ten clades and consistent for Chrysophyllum section Ragala (clade B), Ecclinusa (clade E), C. section Prieurella (clade F) and, to a large extent, for clades P–Q. Stamens inserted in the mid-portion of the corolla tube are consistent only for clade N, J and, with some exceptions, Chromolucuma (clade L) and clade O. Taxa with stamens inserted near the corolla tube base are found in clades D, G, H and L, usually with some exceptions. In some rare cases scoring of this character is difficult; for example, in Pouteria laevigata, with a corolla tube only 0.5–1.5 mm long, stamen insertion could be interpreted either as just below the tube orifice or near the base of the corolla tube. Pollen types Harley (1991) explored pollen types in Neotropical Sapotaceae, but many species remain to be investigated. Eight pollen types were represented in this study and Harley’s numbering system is followed here, with Pennington’s (1991) numbering cited in parentheses. Pollen morphology seems to convey a phylogenetic clue (Fig. 4D). Thus, all taxa in clades A, B, D, E, F, M and N have (or may have) pollen type A1 (7). Elaeoluma (clade C) is the only lineage with pollen type A7 (10). The monotypic genus Diploon (clade G) is the only taxon in the family with the spiny pollen type A9 (12). Pollen type A3 (3) was found in clades G, H, K and P, such as in Chrysophyllum imperiale (clade G) and Pradosia (clade K), a genus also with other pollen types (Harley, 1991). Pollen type A6 (5) is rare but distinctive for clade I, and may be also indicative for clade J. Chromolucuma and Sarcaulus have the pollen types A8 and A10, respectively (combined by Pennington into his type 11), and are not known elsewhere in the family, except for three anomalous collections of Pouteria. Pollen type A5 (9), usually with entire and striate surface, is also rare and found exclusively in clade Q. Overall, pollen type A1 (7) is the most widespread in the subfamily and appears to be symplesiomorphic. If the so-far uninvestigated taxa in our sample were surveyed for pollen types, we predict that they would be largely compatible with the clades recovered in this study. Staminodes Loss of staminodes was shown to have occurred several times in the subfamily (Swenson et al., 2008a) and therefore their absence in Chrysophyllum or presence in Pouteria cannot be used as a synapomorphy. Our study, although supporting the above conclusions, also indicates that their loss is an ongoing process, with vestigial staminodes present, for instance, in Elaeoluma (clade C), Pouteria engleri (clade D) and C. venezuelanense (clade G) (Fig. 4E). Furthermore, in unisexual flowers of clade Q (Pouteria section Oxythece), they are absent in male (staminate) flowers but present in female (pistillate) flowers. We therefore consider their presence, reduction and absence as a useful morphological character. Seed characters Distribution of the three seed character state types shows strong congruence with the recovered tree topology (Fig. 4F). Foliaceous cotyledons, radicle exserted below the cotyledon commissure and presence of endosperm, suggested to be a symplesiomorphic character state combination (Swenson et al., 2008a), was found in clades A, B, F, G and H. The second type, with plano-convex cotyledons, included radicle and absent endosperm, concur with being an advanced and widely distributed character state combination, consistent for clades D, E, I, J and L–Q. The third type is similar to the second, but with the radicle exserted (vs. included) below the cotyledon commissure, a combination of character states diagnostic for Elaeoluma (clade C) and Pradosia (clade K). This latter type is rare in the subfamily, found only in the New Guinean genus Magodendron Vink and in some taxa of the poorly known assemblage of Synsepalum (A.DC.) Daniell in Africa (Swenson et al., 2013). Pennington (1991) had assumed Chromolucuma (clade L) to be of the latter type, but he knew fruits of only one species, with a slightly exserted radicle. Our re-examination of Chromolucuma seeds could not confirm presence of an exserted radicle. In summary, several characters including floral merosity, corolla form, corolla margin, level of stamen insertion and seed features show acceptable to good congruence with the recovered clades. We hope that combinations of these features will be useful in future endeavours of circumscribing natural groups in Neotropical Chrysophylloideae. TOWARDS A REVISED GENERIC CLASSIFICATION Our phylogenetic study of Chrysophylloideae in South America recovered an overall well-resolved topology with high posterior probabilities (PP t 0.95) and usually moderate (JK 75–89%) to strong (JK 90–100%) © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE jackknife support. Diagnostic character combinations are required to describe genera in Sapotaceae (Swenson et al., 2007a, 2008a, 2013; Gautier et al., 2013) and we believe we have identified some of the most useful characters in this study. Broad concepts of Chrysophyllum and Pouteria (Pennington 1991) have been found untenable and it is therefore better to consider clades. In the discussion below, we have opted for clades that covary with morphology and often refer to clades A–Q. MICROPHOLIS Micropholis, a genus of c. 40 species (Govaerts et al., 2001), with ephemeral flowers (Terra-Araujo et al., 2012b), was instated in 1890 by Pierre and recognized by Aubréville (1964a) and Pennington (1990, 1991), with the latter dividing it into two sections. The generic type, M. rugosa (Sw.) Pierre (M. section Micropholis), unfortunately remained unavailable for this study, but its overall morphology suggests that it should be recovered in clade A. Micropholis splendens Gilly ex Aubrév., the type of section Micropholis sect. Exsertistamen, T.D. Penn. was found embedded in clade A with strong support (Fig. 2). If Pouteria laevigata and P. maxima were transferred to Micropholis, the genus would be rendered monophyletic. Pennington (1990, 2006) placed P. laevigata and P. maxima in P. section Oligotheca on the basis of a pentamerous flower with staminodes and seeds with foliaceous cotyledons, exserted radicle and endosperm (Pennington 1990, 2006). However, P. section Oligotheca is without doubt an unnatural assemblage, as shown in several previous studies (Bartish et al., 2005; Swenson & Anderberg, 2005; Swenson et al., 2007a, 2008a), with its Australasian and Pacific members distributed among Planchonella, Pleioluma and Sersalisia (Swenson et al. 2013). The seed characters of P. laevigata and P. maxima concur with those of Micropholis and differ from the seed characters of other sections of Pouteria, which have plano-convex cotyledons, included radicle and no endosperm. Pouteria maxima is furthermore characterized by alternate distichous leaves, a feature of Micropholis sensu Pennington (1990, 1991). A more thorough phylogenetic analysis of Micropholis is therefore needed to test its monophyly and find all members of the group. ECCLINUSA, ELAEOLUMA, CLADE B AND CLADE D Clades B–E Our phylogenetic analyses show clades B–E, recovered with maximum Bayesian and moderate jackknife support, to be sister to the remaining taxa (clade F–Q). Clades B–E include five strongly supported lineages in a polytomy. No particular morphological character or character combination is diagnostic for the 43 larger clade. However, each subclade is possible to characterize. C lade B c ir c um s c r ibe s t hr e e ac c e s s io ns of Chrysophyllum, corresponding to the genus Ragala sensu Aubréville (1964a), or C. section Ragala sensu Pennington (1990), with C. sanguinolentum (Pierre) Baehni as the type species. Pierre (1891) originally established this genus on the basis of its leaf and stomata anatomy and affiliated it with Ecclinusa (clade E). Ecclinusa, established by Martius (1839), includes 11 species of trees and shrubs (Govaerts et al., 2001). The two lineages share several morphological characters, including pollen type (A1), but differ in others. If Ragala were to be reinstated, it would be defined by the absence of stipules, eucamptodromous venation, oblique tertiaries, stamen filaments inserted just below the tube orifice, absence of staminodes, seeds with foliaceous cotyledons, exserted radicle and the presence of the endosperm. One additional important character, rarely reported elsewhere for Neotropical Chrysophylloideae, is an accrescent calyx, subtending the fruit. Ecclinusa differs from the above character combination by the presence of stipules, absence of an accrescent calyx and seeds with plano-convex cotyledons, included radicle and the absence of the endosperm (Figs 3, 4). In addition, flowers of Ecclinusa are sessile, a character rare for the subfamily in South America and known only in a few species of Micropholis (two species) and Pouteria (eight species) (Pennington, 1990). Elaeoluma (clade C), established by Baillon (1891), includes four species distributed from Panama to central Brazil (Aubréville, 1964a; Pennington, 1990). Its monophyly has previously been suggested (Swenson & Anderberg, 2005; Swenson et al., 2008a) and in our analyses it was recovered with maximum support. The genus is readily recognized by the densely and minutely punctuated lower leaf surfaces, pale green colour and reticulate tertiaries. Apart from its exclusive A7 pollen type, it differs from the rest of the clade B–E lineages by the presence of vestigial staminodes in the pistillate flowers and an exserted radicle, notwithstanding the plano-convex cotyledons and the absence of endosperm. Pouteria anomala and P. engleri (clade D), two species recovered with maximum clade support, are closely related to their congener P. oblanceolata, but not to the majority of Pouteria spp. or even the type species (Sarcaulus–clade Q). Lacking staminodes, P. anomala has been associated with Chrysophyllum, whereas P. engleri is the generic type of Nemaluma (Pennington, 1991), a monotypic genus established by Baillon (1891) and recognized only once since then (Aubréville, 1961a). Aubréville characterized the genus by a poorly developed tertiary venation, pentamerous flower, stamens fused to the base of the corolla tube © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 44 A. D. DE FARIA ET AL. and vestigial staminodes. Our observations agree with those of Aubréville, except that the tertiaries have been found to be reticulate and the staminodes either absent or vestigial. Pouteria oblanceolata is one of c. 160 species of P. section Oligotheca sensu Pennington (1990, 1991), with about 150 of them distributed in the natural genera Planchonella, Pleioluma, Sersalisia and Vanroyena of Oceania and Southeast Asia (Swenson et al., 2013). Pouteria section Oligotheca is represented in the Neotropics by ten species, three of them sampled in this study. Two of these (P. laevigata and P. maxima) are best transferred to Micropholis (clade A), but P. oblanceolata occurs as a sole member in the polytomy B–E. It has pentamerous, cup-shaped, unisexual flowers, and shares characters with clades B (Figs 3B, C, 4F) and E (Fig. 3B, C). Filaments are inserted in the mid-section of the corolla tube (Fig. 4C) and well-developed staminodes in both pistillate (c. 0.5 mm long) and staminate (1.7–2.5 mm long) flowers (Fig. 4E) are the only odd characters for this clade. If its ITS sequence is blasted on the European Nucleotide Archive web search, the sequence is most similar to the Australasian genus Pleioluma, with which it shares several features, including the presence of staminodes, filaments inserted in the mid-section of the corolla tube and seed characters. However, it lacks areolate venation and campanulate flowers diagnostic for Pleioluma (Swenson et al., 2013). It thus represents an odd member of the subfamily of uncertain sister relationships and with a unique combination of characters; it may deserve generic recognition. To summarize, clades B–E, although with strong molecular support, have practically no clear subclade relationships. Since amalgamating two or more lineages in any combination would create heterogeneous genera, we suggest that Ragala (clade B), Elaeoluma (clade C), Nemaluma (clade D) and Ecclinusa (clade E) are recognized as separate genera and that Pouteria oblanceolata is included in a broader analysis in order to search for possible closer relatives. RESURRECTION OF PRIEURELLA Clade F Five samples of Chrysophyllum are grouped in clade F with maximum support and are only distantly related to the generic type C. cainito. Clade F corresponds to the genus Prieurella established by Pierre (1891), recognized with five species by Aubréville (1964a), but reduced to a section of Chrysophyllum by Pennington (1990). Aubréville (1964b) distinguished Prieurella based on eucamptodromous leaf venation, oblique tertiaries, pentamerous and globose (dome-shaped) flowers, stamen filaments inserted just below the tube orifice, absence of staminodes, seeds with foliaceous cotyledons, exserted radicle and endosperm. This character combination fully agrees with our observations of the included species and with that of the generic type P. cuneifolia (not validly published), which was not included here due to its presumed hybrid origin (Swenson et al., 2008a). Bark and crushed leaves of plants of this group are also usually characterized by the odour of bitter almonds (hydrocyanic acid), a useful field character. Since these taxa are clearly not close relatives to Chrysophyllum s.s. and the clade is rather easy to recognize, Prieurella deserves to be resurrected. CLADE G AND CHRYSOPHYLLUM POMIFERUM Clade G is a group of three species (Diploon cuspidatum, Chrysophyllum imperiale and C. venezuelanense) with a relationship support of only PP 0.80 and JK 56. These three taxa have been historically treated as separate genera or associated with Chrysophyllum. The monotypic Diploon, established by Cronquist (1946), was based on C. cuspidatum. Because of an odd character combination (unilocular ovary, basi-lateral seed scar and the absence of endosperm), Cronquist was unsure if it belonged in Sapotaceae, but eventually decided to keep it in the family. Aubréville (1964a) and Pennington (1991) accepted Diploon, but its phylogenetic affiliations remain uncertain (Swenson & Anderberg, 2005; Swenson et al., 2008a). The brochidodromous leaf venation, admedial tertiaries, rotate flowers, absence of staminodes, plano-convex cotyledons and the spiny type (A9) of the pollen grains (Fig. 4D), known only in one other species in the family, Micropholis retusa (Spruce ex Miq.) Eyma (not sampled in this study), readily distinguish Diploon from other genera in Sapotaceae. Chrysophyllum imperiale from the Atlantic forest in Brazil and C. venezuelanense from Central America and the north-western corner of South America are two large tree species placed by Pennington (1990) in C. section Aneuchrysophyllum. Molecular and morphological phylogenetic analyses, both in this study and elsewhere (Swenson et al., 2008a; Bartish et al., 2011), have shown C. section Aneuchrysophyllum to be an unnatural assemblage of African and American lineages, with the American members related to each other but of uncertain affinity in the subfamily. Chrysophyllum imperiale and C. subspinosum Monach. (not sampled in this study) are the only family members with a spinous-serrate leaf margin, a feature that led to the initial placement of the former in Theophrastaceae (= Primulaceae sensu APG IV, 2016), with an eventual transfer to Sapotaceae and Chrysophyllum by Bentham & Hooker (1876). © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE Pierre (1891) used these two species to establish two closely associated genera, Martiusella Pierre and Cornuella Pierre, respectively, but neither was recognized by Aubréville (1964a) or Pennington (1991). If accepted as genera, both are readily distinguished from other members of the subfamily by a combination of eucamptodromous leaf venation, pentamerous flowers and stamen filaments inserted near the base of the corolla tube. In combination with oblique tertiaries, bisexual flowers and absence of staminodes, the spinous-serrate leaf margin is a distinguishing character of Martiusella. Cornuella differs from Martiusella in the entire leaves, more or less horizontal tertiaries, unisexual flowers and vestigial staminodes. Chrysophyllum pomiferum was originally described as the monotypic genus Achrouteria Eyma, a name referring to the endospermous seeds of Achras (Manilkara), or Chrysophyllum, and staminodes of Pouteria, with no known intermediates at the time (Eyma, 1936). Pennington (1990) transferred it to Chrysophyllum, despite the presence of well-developed staminodes, placing it in section Aneuchrysophyllum, with C. gonocarpum (Mart. & Eichler) Engl., C. lucentifolium Cronquist and C. venezuelanense. Our study indicates a weak relationship between C. pomiferum and clade G. Apart from the common pollen type (A1) and pentamerous flowers, C. pomiferum shows a combination of characters typical of several different clades, such as reticulate tertiaries (a widespread feature), the presence of staminodes (clades I, J and L–P) and endospermous seeds (clades A, B, F and H). As far as we know, only one other species, C. durifructum (W.A.Rodrigues) T.D.Penn. (not sampled in this study), has the same combination of characters. In summary, the three species of clade G and C. pomiferum are impossible to maintain in Chrysophyllum. If they were to be united in a single genus, based on a certain phylogenetic affinity, a heteromorphic group would be created (Figs 3, 4). On the other hand, by retaining Diploon, reinstatements of Achrouteria, Cornuella and Martiusella would be justified, with Achrouteria and possibly Martiusella containing two species each and Cornuella and Diploon being monotypic. However, Chrysophyllum gonocarpum, the type species of Chloroluma (Table 1), remains to be analysed in a phylogenetic context. It is a species with an unusual character combination and may be a candidate for generic recognition. If future phylogenetic analyses group C. gonocarpum with C. pomiferum, Chloroluma would have priority over Achrouteria (McNeill et al., 2012), if a narrow generic concept were used. For the meantime, we recommend that existing names of Achrouteria, Cornuella and Martiusella be used, simply because that is a better solution than forcing them into an unnatural assemblage of Chrysophyllum. 45 CHRYSOPHYLLUM S.S. AND VILLOCUSPIS Clade H receives maximum support (PP 1.0, JK 100%) and groups two sections of Chrysophyllum as sisters, C. sections Chrysophyllum (C. cainito, C. oliviforme L.) and Villocuspis (C. sparsiflorum), sections that, respectively, include 17 and six species (Pennington, 1990, 1991). Phylogenetic analysis groups these two sections with some Pouteria spp. instead of members of Chrysophyllum. In any event, Aubréville and Pellegrin (in Aubréville, 1961a) recognized Villocuspis at the generic level using the pubescent anthers as a cardinal character, but Pennington emphasized the homogeneity of floral and endospermous seeds and returned to a sectional concept in Chrysophyllum. Our phylogenetic analyses and examination of morphology of terminals in clade H show that flowers have variable numbers of sepals and corolla lobes, usually a sericeous corolla on the outer surface, absence of staminodes and endospermous seeds, but other characters are different. For example, C. section Chrysophyllum has admedial tertiaries, stamens inserted in the tube orifice, and glabrous anthers, contrasting with C. section Villocuspis with laxly reticulate tertiaries, stamens inserted near the corolla base and pubescent anthers (cf. Pennington, 1990). Despite this study including only three species of the two sections, we suspect that an expanded phylogenetic analysis of these two groups may support recognition of two clades corresponding to Chrysophyllum s.s. and Villocuspis. RESURRECTION OF LUCUMA? Clade I, recovered with maximum support (PP 1.0, JK 100%), includes all of the sampled species of Pouteria section Antholucuma and of section Rivicoa sensu Pennington (1990) (Fig. 2). The type species of these two sections, P. multiflora and P. macrophylla, respectively, were recognized by Pierre (1890) as the genera Radlkoferella Pierre and Richardella Pierre, the former on the basis of flowers with four sepals and usually six petals and the latter with pentamerous flowers (Fig. 3F). However, members of section Antholucuma form a well-supported grade to section Rivicoa in our analysis, making recognition of two sections or two genera untenable. Another option would be to unite all members of clade I in a single genus, recognized on two synapomorphies, the papillate corolla margin (Fig. 4B) and the A6 pollen type with unusually thick walls in proportion to the size of the grain (Fig. 4D; Harley, 1991). In addition, this clade would further be recognized based on a series of rather constant characters: oblique tertiaries [except in P. dominigensis (C.F.Gaertn.) Baehni]; bisexual flowers; filaments inserted in the corolla tube orifice (except in P. multiflora); the presence of staminodes; and non-endospermous seeds with plano-convex cotyledons and an © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 46 A. D. DE FARIA ET AL. included radicle. The oldest name in section Rivicoa is Lucuma Molina (Table 1), based on Lucuma bifera Molina [= P. lucuma (Ruiz & Pav.) Kuntze]. Although Pouteria lucuma was not sampled in this study, its morphology fully agrees with the above combination of characters (pollen type unknown), making Lucuma a good candidate to represent a monophyletic group, with its circumscription remaining to be further investigated. A NEW CIRCUMSCRIPTION OF POUTERIA? Our phylogenetic analyses, based on ribosomal and nuclear sequence data, group Sarcaulus–clade Q with maximum Bayesian (PP 1.0) and weak jackknife (JK 64%) support in one large clade. It includes all species of Pouteria not discussed above, plus three widely accepted genera, Chromolucuma, Pradosia and Sarcaulus (Aubréville, 1964a; Pennington, 1990, 1991, 2006, 2007; Swenson & Anderberg, 2005; AlvesAraújo & Alves, 2012a; Terra-Araujo et al., 2013, 2015). Different character combinations readily distinguish these three genera, but in our analyses, they are embedded among the clades of Pouteria. Terminals of clades J and N are usually placed in Pouteria section Pouteria, have nearly identical character distributions and are expected to form Pouteria s.s. with the type species P. guianensis. However, Chromolucuma, Pradosia and part of P. section Franchetella render the two clades of section Pouteria polyphyletic with strong branch support. We foresee two possible solutions: (1) circumscribing Sarcaulus–clade Q as Pouteria, with several recognized subgenera; or (2) assigning generic rank to each clade. Neither of these alternatives is unproblematic, but most of the taxa in the clade Sarcaulus–clade Q are still identified by a simple character combination of a non-papillate corolla margin, presence of staminodes and seeds with plano-convex cotyledons, included radicle and absence of endosperm. However, clade K (Pradosia) is still an exception with similar seed characters, but with an exserted radicle and absence of staminodes. Sarcaulus includes five species with white to yellowish, dome-shaped, carnose flowers that have a strong jasmine-like scent. The stamens are borne on short filaments inserted in the tube orifice and the fruits have plano-convex cotyledons with an included radicle and no endosperm. Its pollen is of type A10, not found elsewhere in the family, a type that has provided evidence to uphold Sarcaulus as a distinct genus (Harley, 1991). Monophyly of Sarcaulus has not been tested. Pradosia (clade K) includes 23 species of trees or shrubs and has been united with Pouteria (Eyma, 1936) and Chrysophyllum (Baehni, 1965), but for some time it has been recognized at the generic level (Aubréville, 1964a; Pennington, 1991; Swenson & Anderberg, 2005; Terra-Araujo et al., 2012a, 2015; Terra-Araujo, Faria & Swenson, 2016). Monophyly of the group receives maximum branch support. Pradosia is possibly the easiest group to identify of all Chrysophylloideae in the Neotropics on the basis of the drupaceous fruit, a synapomorphy for the genus, with a jelly-like (cartilaginous) endocarp that surrounds the single seed with plano-convex cotyledons and an exserted radicle (Terra-Araujo et al., 2013, 2015, 2016). Chromolucuma is paraphyletic in its present circumscription because it forms a grade with three species of Pouteria section Franchetella (clade L). Alves-Araújo & Alves (2012a) expanded Chromolucuma with two species on the basis of presence of stipules, so the group would now circumscribe eight species occurring in the Amazon, the Atlantic forest and north to Costa Rica. The clade is readily distinguished by persistent stipules (absent or small in Pouteria williamii) (Fig. 3A), yellow latex (Fig. 3E) and pollen type A8, if it is unique to the group (Fig. 4D). The group needs to be rendered monophyletic either by transferring three Chromolucuma spp. to Pouteria or another three Pouteria spp. to Chromolucuma. Clades J, M–N and O–Q include members of sections Franchetella, Gayella, Oxythece and Pouteria, supported with high-to-maximum Bayesian and parsimony values. None of these can be satisfactorily distinguished by a combination of characters analysed in this study. For example, tetramerous flowers correlate well with clades J, N and P, but seem to have evolved in Pouteria at least three times (Fig. 3F). Other floral characters, such as the corolla form (Fig. 4A) and stamen insertion in the corolla tube (Fig. 4C), are homoplastic, providing no consistent synapomorphies. There is some indication of correlation in the secondary leaf venation (Fig. 3B), with clade J differing from clades M–N and O–Q, but the evidence is not strong enough to merit a generic recognition for any of the clades. Other reviewed characters, proven useful for the Australasian taxa (Swenson et al., 2013), such as pubescence of the inner surface of sepals, or trichomes on the corolla, have shown no significantly useful pattern. CONCLUSIONS AND PERSPECTIVES Searching for a natural classification of Neotropical Chrysophylloideae proved to be more complicated than first expected. In retrospect, we realize that our sampling is not sufficiently exhaustive for overall conclusions, but our subfamilial phylogenetic assessment provides new insights for a multitude of conclusions and suggests hypotheses to be tested in the future. The monophyly of Neotropical Chrysophylloideae with one or two establishments in the region should be tested. Broad generic concepts of Chrysophyllum and Pouteria © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE are untenable and unpractical and their lineages must be explored, delimited, recognizable and named. Early diverging lineages, often with aberrant morphological character distributions, are easier to distinguish than more recent and diverse radiations. We conclude that the usually recognized genera Diploon, Ecclinusa and Elaeoluma should be maintained and that the satellite genera (or sections) Ragala (clade B), Nemaluma (clade D), Prieurella (clade F), Achrouteria, Cornuella and Martiusella (clade G) merit resurrection. The presumed intercontinental hybrid origin of Chrysophyllum cuneifolium and its status as the generic type of Prieurella must be clarified. Monophyly of Micropholis, here used to orientate the tree, needs to be tested in a broader analysis. Likewise, the true members of Chrysophyllum, i.e. sections Chrysophyllum and Villocuspis, must be explored with a larger taxon sampling. Phylogenetic relationships of Chrysophyllum durifructum, C. eximium, C. flexuosum, C. gonocarpum, C. subspinosum, Pouteria lucuma, P. oblanceolata and the remaining seven species of section Oligotheca are all of particular interest, not the least because some are types. For example, the relationship of P. lucuma is important for a possible resurrection of Lucuma. It is recommended that multiple accessions are included of species with disjunct distributions or that are divided into several subspecies. Then there is the challenge to circumscribe the assemblage of Pouteria, including the relationships and monophyly of Sarcaulus. An evaluation of the alternatives mentioned above must include the type species (Table 1), a more complete taxon sampling and a further pursuit of useful morphological characters. Micromorphological leaf characters were recently successfully used to separate morphologically similar Pouteria spp. (Popovkin, Faria & Swenson, 2016) and here might provide useful features that concur with recovered clades. Making a qualified prediction, we would be surprised if yet-to-be-analysed species of Pouteria sections Franchetella, Gayella, Oxythece and Pouteria and members of the currently accepted genera Chromolucuma, Pradosia and Sarcaulus would fall outside of clade Sarcaulus–Q. ACKNOWLEDGMENTS Igor Bartish, Laurent Gautier, one anonymous reviewer and, especially, the Associate Editor Lars Chatrou are thanked for many constructive comments on the manuscript. We thank the herbarium curators Carlos Henrique Franciscon (INPA), Roseli Torres (IAC) and Ana Odete Santos Vieira (FUEL) for making specimens available for this study. Hugh Nicholson is thanked for his contribution of field images. ADF is grateful to the National Institute for Amazonian 47 Research (INPA) by supporting the work in the Amazon, with special thanks to the Postgraduate Course in Plant Biology and the Biological Dynamics of Forest Fragments (PDBFF). Cynthia de Oliveira Ferreira, Laboratory of Molecular Biology at the Hospital for Tropical Biology, State of Amazonas, and Mattias Myrenås and Bodil Cronholm, Swedish Museum of Natural History, are acknowledged for their invaluable laboratory assistance. José Ribamar Mesquita Ferreira and Everaldo da Costa Pereira provided invaluable field assistance by sampling most of the Amazonian specimens. For productive collaboration in the field, we thank the team of Flora do Uatuma Project in the State of Amazon, Lidyanne Aona in the State of Bahia and João Carlos Galvão, André Rochelle and Catia Urbanetz in the State of São Paulo. This work was supported by grants to ADF and JRP from the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP – 06/55507-9 and 06/55508-5) and to US from the Swedish Research Council (VR 621-20075845). JRP is also supported by a CNPq grant. REFERENCES Akaike H. 1974. A new look at the statistical model identification. IEEE Transactions on Automatic Control 19: 716–723. Álvarez I, Wendel JF. 2003. Ribosomal ITS sequences and plant phylogenetic inference. Molecular Phylogenetics and Evolution 29: 417–434. Alves-Araújo A, Alves M. 2011. Two new species of Pouteria (Sapotaceae) from the Atlantic Forest in Brazil. Systematic Botany. 36: 1004–1007. Alves-Araújo A, Alves M. 2012a. Two new species and a new combination of Neotropical Sapotaceae. Brittonia 64: 23–29. Alves-Araújo A, Alves M. 2012b. Pouteria ciliata, P. confusa, P. nordestinensis and P. velutinicarpa spp. nov. (Sapotaceae) from Brazil. Nordic Jounal of Botany 30: 399–406. APG IV. 2016. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG IV. Botanical Journal of the Linnean Society 181: 1–20. Aubréville A. 1961a. Notes sur les Sapotacées Africaines et Sud-Américaines. Adansonia, n.s. 1: 6–38. Aubréville A. 1961b. Notes sur des Poutériées Américaines. Adansonia, n.s. 1: 150–191. Aubréville A. 1964a. Les Sapotacées: taxonomie et phytogéographie. Adansonia, série 2, Mémoire 1: 1–157. Aubréville A. 1964b. Notes sur des Sapotacées. III. Adansonia, n.s. 4: 367–391. Aubréville A. 1967. Flore de la Nouvelle Calédonie et dépendances, Vol. 1, Sapotacées. Paris: Muséum National d’Histoire Naturelle. Baehni C. 1942. Mémoires sur les Sapotacées II. Le genre Pouteria. Candollea 9: 147–476. Baehni C. 1965. Mémoires sur les Sapotacées III. Inventaire des genres. Boissiera 11: 1–262. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 48 A. D. DE FARIA ET AL. Baillon H. 1891. Histoire des plantes, Vol. 11. Paris: Librairie Hachette. Baldwin BG, Markos S. 1998. Phylogenetic utility of the external transcribed spacer (ETS) of 18S–26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Molecular Phylogenetics and Evolution 10: 449–463. Bartish IV, Swenson U, Munzinger J, Anderberg AA. 2005. Phylogenetic relationships among New Caledonian Sapotaceae (Ericales): molecular evidence for generic polyphyly and repeated dispersal. American Journal of Botany 92: 667–673. Bartish IV, Antonelli A, Richardson JE, Swenson U. 2011. Vicariance or long-distance dispersal: historical biogeography of the pantropical subfamily Chrysophylloideae (Sapotaceae). Journal of Biogeography 38: 177–190. Bentham G, Hooker JD. 1876. Genera plantarum 2. London: Reeve & Co. Cronquist A. 1946. Studies in the Sapotaceae — VI: miscellaneous notes. Bulletin of the Torrey Botanical Club 73: 465–471. Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Molecular Biology and Evolution 29: 1969–1973. Ellis B, Daly DC, Hickey LJ, Johnson KR, Mitchell JD, Wilf P, Wing SL. 2009. Manual of leaf architecture. Ithaca: The New York Botanical Garden Press and Comstock Publishing Associates. Eyma PJ. 1936. Notes on Guiana Sapotaceae. Recueil des Travaux Botaniques Néerlandais 33: 156–210. Farris JS, Albert VA, Källersjö M, Lipscomb D, Kluge AG. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12: 99–124. Gautier L, Naciri Y, Anderberg AA, Smedmark JEE, Randrianaivo R, Swenson U. 2013. A new species, genus and tribe of Sapotaceae, endemic to Madagascar. Taxon 62: 972–983. Gautier L, Lachenaud O, van der Burgt X, Kenfack D. 2016. Five new species of Englerophytum K. Krause (Sapotaceae) from central Africa. Candollea 71: 287–305. Gernhard T. 2008. The conditioned reconstructed process. Journal of Theoretical Biology 253: 769–778. Govaerts R, Frodin DG, Pennington TD. 2001. World checklist and bibliography of Sapotaceae. Kew: Royal Botanic Gardens. Harley MM. 1991. The pollen morphology of the Sapotaceae. Kew Bulletin 46: 379–491. Harris GJ, Harris MW. 1997. Plant identification terminology: an illustrated glossary, 1st edn. 5th printing. Springer Lake: Springer Lake Publishing. Huelsenbeck JP, Ronquist F. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754–755. Katoh K, Asimenos G, Toh H. 2009. Multiple alignment of DNA sequences with MAFFT. In: Posada D, ed. Bioinformatics for DNA sequence analysis. Totowa: Human Press, 39–64. Lemey P, Rambaut A, Drummond AJ, Suchard MA. 2009. Bayesian phylogeography finds its roots. PLoS Computational Biology 5: e1000520. Mackinder B, Harris DJ, Gautier L. 2016. A reinstatement, recircumscription and revision of the genus Donella (Sapotaceae). Edinburgh Journal of Botany 73: 297–339. Maddison DR, Maddison WP. 2005. MacClade 4: analysis of phylogeny and character evolution, version 4.0. Sunderland: Sinauer. Martius CFP. 1839. (19.) Ecclinusa ramiflora Mart. Flora 22 (Beiblatt 1): 2. McNeill J, Barrie FR, Buck WR, Demoulin V, Greuter W, Hawksworth DL, Herendeen PS, Knapp S, Marhold K, Prado J, Prud’homme van Reine WF, Smith GF, Wiersema JH, Turland NJ, eds. 2012. International Code of Botanical Nomenclature for algae, fungi, and plants (Melbourne Code): adopted by the Eighteenth International Botanical Congress Melbourne, Australia, July 2011. Regnum Vegetabile 154. Königstein: Koeltz. Morales F. 2012. Nuevas especies de Sapotaceae para Cosa Rica. Darwiniana 50: 107–113. Munzinger J, Swenson U. 2009. Three new species of Planchonella Pierre (Sapotaceae) with a dichotomous and online key to the genus in New Caledonia. Adansonia, séries 3, 31: 175–189. Naciri Y, Linder HP. 2015. Species delimitation and relationships: the dance of the seven veils. Taxon 64: 3–16. Oxelman B, Bremer B. 2000. Discovery of paralogous nuclear gene sequences coding for the second-largest subunit of RNA polymerase II (RPB2) and their phylogenetic utility in gentianales of the asterids. Molecular Biology and Evolution 17: 1131–1145. Pennington TD. 1990. Flora Neotropica Monograph 52: Sapotaceae. New York: New York Botanical Garden. Pennington TD. 1991. The genera of Sapotaceae. Kew: Royal Botanic Gardens. Pennington TD. 2006. Flora da Reserva Ducke, Amazonas, Brasil: Sapotaceae. Rodriguésia 57: 251–366. Pennington TD. 2007. Flora of Ecuador 80. 152. Sapotaceae. Gothenburg: Department of Plant and Environmental Sciences, Göteborg University. Petersen JJ, Parker IM, Potter D. 2012. Origins and close relatives of a semi-domesticated neotropical fruit tree: Chrysophyllum cainito (Sapotaceae). American Journal of Botany 99: 585–604. Pierre L. 1890. Notes botaniques: Sapotacées, pp. 1–36. Paris: Librairie des sciences Paul Klincksieck. Pierre L. 1891. Notes botaniques: Sapotacées, pp. 37–67. Paris: Librairie des sciences Paul Klincksieck. Poczai P, Hyvönen J. 2010. Nuclear ribosomal spacer regions in plant phylogenetics: problems and prospects. Molecular Biology Reports 37: 1897–1912. Popovkin AV, Faria AD, Swenson U. 2016. Pouteria synsepala (Sapotaceae: Chrysophylloideae): a new species from the northern littoral of Bahia, Brazil. Phytotaxa 286: 39–46. Posada D. 2008. jModelTest: phylogenetic model averaging. Molecular Biology and Evolution 25: 1253–1256. Rambaut A. 2009. FigTree version 1.3.1. Computer program available at: http://tree.bio.ed.ac.uk/software/figtree/ (last accessed November 2015). Rambaut A, Drummond AJ. 2009. Tracer version 1.5. Computer program available at: http://tree.bio.ed.ac.uk/software/tracer/ (last accessed November 2015). © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 NEOTROPICAL CHRYSOPHYLLOIDEAE Rannala B, Yang Z. 1996. Probability distribution of molecular evolutionary trees: a new method of phylogenetic inference. Journal of Molecular Evolution 43: 304–311. Ribeiro JELS, Hopkins MJG, Vicentini A, Sothers CA, Costa MAS, Brito JM, Souza MA, Martins LHP, Lohmann LG, Assunção PACL, Pereira EC, Silva CF, Mesquita MR, Procópio LC. 1999. Flora da Reserva Ducke: guia de identificação das plantas vasculares de uma floresta de terra firme na Amazônia Central. Manaus: INPA. Ronquist F, Huelsenbeck JP. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. Roosmalen MGM, Garcia OMCG. 2000. Fruits of the Amazonian forest. Part II: Sapotaceae. Acta Amazonica 30: 187–290. Santamaría-Aguilar D, Chaves-Fallas JM, Aguilar R. 2017. Two new species of Chrysophyllum (Sapotaceae) endemic to Costa Rica. Brittonia 69: 1–9. Simmons MP. 2004. Independence of alignment and tree search. Molecular Phylogenetics and Evolution 31: 874–879. Simmons MP, Ochoterena H. 2000. Gaps as characters in sequence-based phylogenetic analyses. Systematic Biology 49: 369–381. Smedmark JE, Anderberg AA. 2007. Boreotropical migration explains hybridization between geographically distant lineages in the pantropical clade Sideroxyleae (Sapotaceae). American Journal of Botany 94: 1491–1505. Smedmark JE, Swenson U, Anderberg AA. 2006. Accounting for variation of substitution rates through time in Bayesian phylogeny reconstruction of Sapotoideae (Sapotaceae). Molecular Phylogenetics and Evolution 39: 706–721. Swenson U, Anderberg AA. 2005. Phylogeny, character evolution, and classification of Sapotaceae (Ericales). Cladistics 21: 101–130. Swenson U, Munzinger J. 2012. Revision of Pichonia (Sapotaceae) in New Caledonia. Australian Systematic Botany 25: 31–48. Swenson U, Munzinger J. 2016. Five new species and a systematic synopsis of Pycnandra (Sapotaceae), the largest endemic genus in New Caledonia. Australian Systematic Botany 29: 1–40. Swenson U, Bartish IV, Munzinger J. 2007a. Phylogeny, diagnostic characters, and generic limitation of Australasian Chrysophylloideae (Sapotaceae, Ericales): evidence from ITS sequence data and morphology. Cladistics 23: 201–228. Swenson U, Munzinger J, Bartish IV. 2007b. Molecular phylogeny of Planchonella (Sapotaceae) and eight new species from New Caledonia. Taxon 56: 329–354. Swenson U, Richardson JE, Bartish IV. 2008a. Multi-gene phylogeny of the pantropical subfamily Chrysophylloideae (Sapotaceae): evidence of generic polyphyly and extensive morphological homoplasy. Cladistics 24: 1006–1031. 49 Swenson U, Nylinder S, Munzinger J. 2013. Towards a natural classification of Sapotaceae subfamily Chrysophylloideae in Oceania and Southeast Asia based on nuclear sequence data. Taxon 62: 746–770. Swenson U, Lowry PP 2nd, Munzinger J, Rydin C, Bartish IV. 2008b. Phylogeny and generic limits in the Niemeyera complex of New Caledonian Sapotaceae: evidence of multiple origins of the anisomerous flower. Molecular Phylogenetics and Evolution 49: 909–929. Swenson U, Munzinger J, Lowry PP II, Cronholm B, Nylinder S. 2015. Island life – classification, speciation and cryptic species of Pycnandra (Sapotaceae) in New Caledonia. Botanical Journal of the Linnean Society 179: 57–77. Swofford DL. 2002. PAUP*: Phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sunderland: Sinauer. Terra-Araujo MH, Faria AD, Vicentini A. 2012a. A new species of Pradosia (Sapotaceae) from Central Amazonia. Brittonia 64: 139–142. Terra-Araujo MH, Faria AD, Ribeiro JELS, Swenson U. 2012b. Flower biology and subspecies concepts in Micropholis guyanensis (Sapotaceae): evidence of ephemeral flowers in the family. Australian Systematic Botany 25: 295–303. Terra-Araujo MH, Faria AD, Alves-Araujo A, Alves M. 2013. Pradosia restingae sp. nov. from the Atlantic forest, Brazil. Nordic Journal of Botany 31: 437–441. Terra-Araujo MH, de Faria AD, Vicentini A, Nylinder S, Swenson U. 2015. Species tree phylogeny and biogeography of the Neotropical genus Pradosia (Sapotaceae, Chrysophylloideae). Molecular Phylogenetics and Evolution 87: 1–13. Terra-Araujo MH, Faria AD, Swenson U. 2016. A taxonomic update of Neotropical Pradosia (Sapotaceae, Chrysophylloideae). Systematic Botany 41: 634–650. Thiers B. (continuously updated). Index herbariorum: a global directory of public herbaria and associated staff. New York Botanical Garden’s Virtual Herbarium. Available at: http://sweetgum.nybg.org/science/ih/ (last accessed March 2017). Triono T, Brown AHD, West JG, Crisp MD. 2007. A phylogeny of Pouteria (Sapotaceae) from Malesia and Australasia. Australian Systematic Botany 20: 107–118. van Royen P. 1957. Revision of the Sapotaceae of the Malaysian area in a wider sense. VII. Planchonella Pierre. Blumea 8: 235–445. Vink W. 1958. Revision of the Sapotaceae of the Malaysian area in a wider sense XIII. Chrysophyllum L. Blumea 9: 21–74. Yang Z, Rannala B. 1997. Bayesian phylogenetic inference using DNA sequences: a Markov Chain Monte Carlo Method. Molecular Biology and Evolution 14: 717–724. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 50 Taxon Pennington (1990,1991) Aubréville (1961b,1964a) Origin: Voucher ETS ITS RPB2 Chromolucuma Ducke *Chromolucuma cespedisiiformis J.F.Morales Chromolucuma rubriflora Ducke Chrysophyllum L. Chrysophyllum section Aneuchrysophyllum C. imperiale (K.Koch & Fintelm.) Benth. & Hook.f. C. pomiferum (Eyma) T.D.Penn. C. venezuelanense (Pierre) T.D.Penn. Chrysophyllum section Chrysophyllum C. cainito L. C. oliviforme L. Chrysophyllum section Prieurella C. amazonicum T.D.Penn. C. amazonicum T.D.Penn. C. colombianum T.D.Penn. C. manaosense (Aubrév.) T.D.Penn. *C. prieurii A.DC. Chromolucuma unknown at the time Costa Rica: Arne Anderberg & al. 20 (S) KJ453560–63* EF558614 KJ453710 Chromolucuma Chrysophyllum Chloroluma Baill. Chloroluma Brazil: Vicentini & al. 1229 (INPA) NS KJ399340 KJ453711 Brazil: Pennington s.n. (S) KJ453567 EF558615 KJ453715 Chloroluma Chloroluma Chrysophyllum Chrysophyllum Chrysophyllum Prieurella Pierre Prieurella Prieurella Prieurella Prieurella Prieurella Brazil: Assunção & Silva 651 (INPA) Ecuador: Ståhl & al. 5755 (S) KJ453570 KJ453576 KJ399345 DQ246673 KJ453718 KJ453724 Mexico: Petersen 94 (UCD) Cuba: Gutiérrez & Nilsson 1 (S) KJ453565 KJ453569 KJ399342 DQ246670 KJ453713 KJ453717 Brazil: Assunção & Pereira 207 (INPA) French Guiana: Poncy 1745 (P) Brazil: Vicentini & Pereira 913 (INPA) Brazil: Ribeiro & al. 1246 (INPA) Brazil: Nascimento & al. 778 (INPA) KJ453564 NS KJ453566 KJ453568 KJ453571 KJ399341 DQ246690 KJ399343 KJ399344 KJ399346– 53* KJ453712 KJ453791 KJ453714 KJ453716 KJ453719 Chrysophyllum section Ragala C. sanguinolentum (Pierre) Baehni subsp. balata (Ducke) T.D.Penn. subsp. spurium (Ducke) T.D.Penn. C. ucuquiranabranca (Aubrév. & Pellegr.) T.D.Penn. Chrysophyllum section Villocuspis Ragala Pierre Ragala Ragala Ragala Ragala Brazil: Vicentini & Silva 379 (INPA) Brazil: Nascimento & al. 549 (INPA) Brazil: Nascimento & al. 777 (INPA) KJ453572 KJ453573 KJ453575 KJ399354 KJ399355 KJ399357 KJ453720 KJ453721 KJ453723 Brazil: Sothers & Silva 927 (INPA) KJ453574 KJ399356 KJ453722 French Guiana: Pennington & al. 13843 (U) KJ453577 DQ246676 KJ453725 C. sparsiflorum Klotzsch ex Miq. Diploon Cronquist Diploon cuspidatum (Hoehne) Cronq. Villocuspis Aubrév. & Pellegr. Villocuspis Diploon Diploon A. D. DE FARIA ET AL. Appendix. Accessions of Neotropical Chrysophylloideae sampled in this study, following Pennington’s (1990, 1991) classification compared to that of Aubréville (1961b, 1964a). Type species for genera or sections are in bold type. Sequences published for the first time are prefixed with KJ and multiple accessions with asterisks (*). © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 APPENDIX Taxon Pennington (1990,1991) Aubréville (1961b,1964a) Ecclinusa Mart. Ecclinusa Ecclinusa guianensis Eyma Ecclinusa Ecclinusa guianensis Eyma Ecclinusa Ecclinusa ramiflora Mart. Ecclinusa Elaeoluma Baill. Elaeoluma Elaeoluma glabrescens (Mart. & Eichler Elaeoluma ex Miq.) Aubrév. Elaeoluma nuda (Baehni) Aubrév. Elaeoluma Elaeoluma schomburgkiana (Miq.) Baill. Elaeoluma Micropholis Pierre section Micropholis Micropholis Micropholis casiquiarensis Aubrév. Micropholis Micropholis guyanensis (A.DC.) Pierre Micropholis Micropholis guyanensis (A.DC.) Pierre Micropholis Micropholis venulosa (Mart. & Eichler ex Micropholis Miq.) Pierre Micropholis williamii Aubrév. & Pellegr. Micropholis Micropholis section Exsertistamen not defined at the time Micropholis splendens Gilly ex Aubrév. Micropholis Pouteria Aubl. Pouteria Pouteria section Antholucuma Radlkoferella Pierre P. dominigensis (C.F.Gaertn.) Baehni Radlkoferella P. multiflora (A.DC.) Eyma Radlkoferella P. oxypetala T.D.Penn. unknown at the time P. venosa (Mart.) Baehni Radlkoferella Pouteria section Franchetella Franchetella Pierre P. andarahiensis T.D.Penn. unknown at the time P. anomala (Pires) T.D.Penn. ? P. bilocularis (H.Winkl.) Baehni Labatia Sw. *P. campanulata Baehni Pouteria P. cladantha Sandwith Neoxythece Aubrév. & Pellegr. P. coriacea (Pierre) Pierre Pouteria P. durlandii (Standl.) Baehni Paralabatia Pierre P. engleri Eyma Nemaluma Baill. P. erythrochrysa T.D.Penn. unknown at the time P. flavilatex T.D.Penn. unknown at the time *P. fulva T.D.Penn. unknown at the time Origin: Voucher ETS ITS RPB2 Brazil: Assunção & al. 162 (INPA) Brazil: Ducke Reserve 05-906 (K) Suriname: Irwing & al. 55081 (S) KJ453578 KJ453579 KJ453580 KJ399358 DQ246677 DQ246678 KJ453726 KJ453727 KJ453728 Costa Rica: Anderberg & al. 33 (S) KJ453581 EF558616 KJ453729 Brazil: Souza & al. 409 (INPA) Brazil: Keel & Coelho 243 (S) KJ453582 KJ453583 KJ399359 DQ246679 KJ453730 KJ453731 Brazil: Nascimento & al. 770 (INPA) Brazil: Hopkins & al. 1475 (INPA) Puerto Rico: Taylor 11691 (MO) Brazil: Assunção 122 (U) NS KJ453584 NS KJ453586 KJ399360 KJ399361 DQ246682 DQ246683 KJ453732 KJ453733 KJ453734 KJ453736 Brazil: Ribeiro & al. 1199 (INPA) KJ453587 KJ399363 KJ453737 Brazil: Assunção & Pereira 13 (INPA) KJ453585 KJ399362 KJ453735 Cuba: Gutiérrez & Nilsson 13 (S) Ecuador: Villa & Rivaz 257 (BM) Brazil: Bertoni & Geremias 293 (IAC) Brazil: Faria & Ribeiro 2007/55 (INPA) KJ453615 NS KJ453658 KJ453682 AY552106 DQ246693 KJ399417 KJ399445 KJ453754 KJ453777 KJ453780 KJ453800 Brazil: Faria & Ribeiro 2008/09 (SPF) Brazil: Sothers & Pereira 1068 (INPA) Brazil: Nascimento & al. 508 (INPA) Brazil: Assunção & al. 673 (INPA) Brazil: Assunção & Pereira 616 (INPA) KJ453589 KJ453590 KJ453592 KJ453598–604* KJ453606 KJ399373 KJ399374 KJ399376 KJ399382 KJ399383 KJ453739 KJ453740 KJ453742 KJ453748 KJ453750 Brazil: Faria & Ribeiro 2007/07 (INPA) Brazil: Ribeiro & al. 1904 (INPA) Brazil: Faria & Ribeiro 735 (INPA) Brazil: Ribeiro & al. 1785 (INPA) Brazil: Ribeiro & al. 1906 (INPA) Brazil: Faria & Ribeiro 2007/52 (INPA) NS. KJ453616 KJ453617 KJ453619 KJ453628 KJ453634 KJ399384 KJ399387 KJ399388 KJ399390 KJ399394 KJ399396– 405* KJ453751 KJ453755 KJ453756 KJ453758 KJ453762 KJ453764 NEOTROPICAL CHRYSOPHYLLOIDEAE © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 Appendix. Continued 51 52 Aubréville (1961b,1964a) Origin: Voucher ETS ITS RPB2 P. gardneri (Mart. & Miq.) Baehni Podoluma Baill. KJ453674 KJ399437 KJ453792 P. jariensis Pires & T.D.Penn. *P. minima T.D.Penn. *P. pallens T.D.Penn. P. pentamera T.D.Penn. P. platyphylla (A.C.Sm.) Baehni P. ramiflora (Mart.) Radlk. P. reticulata (Engl.) Eyma P. retinervis T.D.Penn. P. rostrata (Huber) Baehni P. stipulifera T.D.Penn. P. subsessilifolia Cronquist P. vernicosa T.D.Penn. P. vernicosa T.D.Penn. P. virescens Baehni P. williamii (Aubrév. & Pellegr.) T.D.Penn. P. Faria13 P. Faria51 *P. Tree183 ? unknown at the time unknown at the time unknown at the time Pouteria Paralabatia Franchetella Pierre unknown at the time Pouteria unknown at the time ? unknown at the time unknown at the time Pouteria Eremoluma Baill. ? ? ? Brazil, Picinguaba: Tree 992 PJ sub50 (INPA) Brazil: Oliveira & al. 215 (INPA) Brazil: Ribeiro & Pereira 1876 (INPA) Brazil: Ribeiro & al. 1905 (INPA) Brazil: Faria & Ribeiro 2007/54 (INPA) Brazil: Faria & Ribeiro 2007/39 (INPA) Brazil: Faria & Ribeiro 2008/5 (SPF) Brazil: Assunção & al. 728 (INPA) Brazil: Ribeiro & al. 1936 (INPA) Brazil: Faria & Ribeiro 2007/40 (INPA) Brazil: Ribeiro & Pereira 1956 (INPA) Brazil: Faria & Ribeiro 2008/11 (SPF) Brazil: Sothers & Pereira 380 (INPA) Ecuador: Villa, Velez & Rivaz 1304 (BM) Brazil: Martins & al. 49 (INPA) Brazil: Souza & al. 477 (INPA) Brazil: Faria & Ribeiro 2007/13 (INPA) Brazil: Faria & Ribeiro 2007/51 (INPA) Brazil, Picinguaba: Tree 183 PJ sub11 KJ453639 KJ453646–55* KJ453659–64* KJ453665 KJ453667 KJ453669 KJ453671 KJ453672 KJ453673 KJ453677 KJ453679 KJ453683 KJ453684 KJ453685 KJ453686 KJ453593 KJ453618 KJ453668 KJ399408 KJ399414 KJ399418 KJ399419 KJ399421 KJ399432 KJ399434 KJ399435 KJ399436 KJ399440 KJ399442 KJ399446 DQ246694 KJ399447 KJ399448 KJ399377 KJ399389 KJ399422– 31* KJ453769 KJ453776 KJ453781 KJ453782 KJ453784 KJ453786 KJ453788 KJ453789 KJ453790 KJ453795 KJ453797 KJ453801 KJ453802 KJ453803 KJ453804 KJ453743 KJ453757 KJ453785 Gayella Pierre Myrtiluma Baill. Brazil: Faria & Ribeiro 2007/38 (INPA) KJ453620–25* KJ399391 KJ453759 Labatia unknown at the time ? Brazil: Vicentini & al. 762 (INPA) Brazil: Vicentini & al. 1203 (INPA) Brazil: Ribeiro & Silva 1373 (INPA) KJ453640 KJ453645 KJ453656 KJ399409 KJ399413 KJ399415 KJ453770 KJ453775 KJ453778 Pseudoxythece Aubrév. Brazil: Ribeiro & al. 1895 (INPA) KJ453588 KJ399364– 72* KJ453738 Pouteria Neoxythece Caramuri Aubrév. & Pellegr. Brazil: Faria & Ribeiro 2007/21 (INPA) Brazil: Sothers & al. 200 (INPA) Brazil: Martins & al. 45 (INPA) KJ453607–13* KJ453614 KJ453657 KJ399385 KJ399386 KJ399416 KJ453752 KJ453753 KJ453779 Pouteria section Gayella *P. eugeniifolia (Pierre) Baehni Pouteria section Oligotheca P. laevigata (Mart.) Radlk. P. maxima T.D.Penn. P. oblanceolata Pires Pouteria section Oxythece *P. ambelaniifolia (Sandwith) T.D.Penn. P. cuspidata (A.DC.) Baehni *subsp. cuspidata subsp. dura (Eyma) T.D.Penn. P. opposita (Ducke) T.D.Penn. A. D. DE FARIA ET AL. Taxon Pennington (1990,1991) © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 Appendix. Continued Taxon Pennington (1990,1991) Origin: Voucher ETS ITS RPB2 Brazil: Tree D1906, Cananéia KJ453591 KJ399375 KJ453741 P. caimito (Ruiz & Pav.) Radlk. Pseudolabatia Aubrév. & Pellegr. Labatia KJ453594 KJ399378 KJ453744 P. caimito (Ruiz & Pav.) Radlk. Labatia KJ453595 KJ399379 KJ453745 P. caimito (Ruiz & Pav.) Radlk. Labatia KJ453596 KJ399380 KJ453746 P. caimito (Ruiz & Pav.) Radlk. Labatia KJ453597 KJ399381 KJ453747 P. caimito (Ruiz & Pav.) Radlk. P. filipes Eyma P. fimbriata Baehni *P. freitasii T.D.Penn. P. gardneriana (A.DC.) Radlk. P. guianensis Aubl. P. hispida Eyma P. hispida Eyma P. petiolata T.D.Penn. P. resinosa T.D.Penn. P. subcaerulea Pierre ex Dubard P. torta (Mart.) Radlk. subsp. glabra T.D.Penn. subsp. torta Labatia Pseudolabatia Pouteria unknown at the time Pouteria Pouteria Pouteria Pouteria unknown at the time unknown at the time Pseudolabatia Brazil: Manaus: Assunção & Silva 649 (INPA) Brazil, Guariva: Bertoni & Geremias 11 (IAC) Brazil, Penápolis: Bertoni & Geremias 391 (IAC) Brazil, Ubatuba: Bertoni & Geremias 295 (IAC) Brazil, Cananéia: Tree C1913 (IAC) Brazil: Faria & Ribeiro 2007/43 (INPA) Brazil: Ribeiro & al. 1908 (INPA) Brazil: Assunção & Silva 770 (INPA) Argentina: Schwarz 8216 (UPS) Brazil: Ribeiro & al. 1902 (INPA) Brazil: Assunção & Silva 197 (INPA) French Guiana: Mori & al. 25432 (NY) Brazil: Sothers 946 (INPA) Brazil: Faria & Ribeiro 2007/33 (INPA) Brazil: Faria & Ribeiro 2008/3 (SPF) KJ453676 KJ453626 KJ453627 KJ453629–33* KJ453635 KJ453636 KJ453637 KJ453638 KJ453666 KJ453670 KJ453678 KJ399439 KJ399392 KJ399393 KJ399395 DQ246689 KJ399406 KJ399407 DQ246691 KJ399420 KJ399433 KJ399441 KJ453794 KJ453760 KJ453761 KJ453763 KJ453765 KJ453766 KJ453767 KJ453768 KJ453783 KJ453787 KJ453796 KJ453680 KJ453675 KJ399443 KJ399438 KJ453798 KJ453793 Pouteria Richardella Pierre Richardella Richardella Richardella Richardella Brazil: Ribeiro & al. 1310 (INPA) Brazil, Picinguaba: Tree 0412 PJ sub21 (INPA) Brazil: Faria & Ribeiro 2008/2 (SPF) KJ453681 KJ399444 KJ453799 Taiwan (cultivated): Wang 798 (HAST) Brazil: Vicentini & Pereira 771 (INPA) Bolivia: Seidel & al. 5905 (K) Brazil: Nascimento & Pereira 571 (INPA) KJ453605 KJ453642 KJ453641 KJ453644 DQ246688 KJ399410 DQ246692 KJ399412 KJ453749 KJ453772 KJ453771 KJ453774 Richardella Brazil: Nascimento 607 (INPA) KJ453643 KJ399411 KJ453773 Pradosia Ecclinusa Ecclinusa Pradosia Brazil: Lindeman 6743 (U) Brazil: Brito & al. 29 (INPA) Brazil: Brito & al. 1950 (INPA) KJ453687 KJ453688 KJ453689–96* AY552158 KJ399449 KJ399450 NS KJ453805 KJ453806 Pouteria section Pouteria P. beaureparei (Glaz. & Raunk.) Baehni subsp. torta Pouteria section Rivicoa P. campechiana (Kunth) Baehni P. macrophylla (Lam.) Eyma P. macrophylla (Lam.) Eyma P. manaosensis (Aubrév. & Pellegr.) T.D.Penn. P. manaosensis (Aubrév. & Pellegr.) T.D.Penn. Pradosia Liais Pradosia brevipes (Pierre) T.D.Penn. Pradosia cochlearia (Lecomte) T.D.Penn. *Pradosia decipiens Ducke Pouteria Pouteria 53 Aubréville (1961b,1964a) NEOTROPICAL CHRYSOPHYLLOIDEAE © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 Appendix. Continued 54 Taxon Pennington (1990,1991) Aubréville (1961b,1964a) Origin: Voucher ETS ITS RPB2 *Pradosia schomburgkiana (A.DC.) Cronquist Pradosia surinamensis (Eyma) T.D.Penn. Sarcaulus Radlk. *Sarcaulus brasiliensis (A.DC.) Eyma Pradosia Brazil: Faria & Ribeiro 2007/09 (INPA) KJ453697–99* KJ399451 KJ453807 Pradosia Guyana: Harris 1076 (U) KJ453700 AY552157 KJ453808 Sarcaulus Brazil: Martins & al. 48 (INPA) KJ453701–09* KJ399452 KJ453809 NS, not sampled. *Species with variable repeats in either ETS or ITS. A. D. DE FARIA ET AL. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017 Appendix. Continued NEOTROPICAL CHRYSOPHYLLOIDEAE 55 SUPPORTING INORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web-site: Table S1. Characters sampled for this phylogenetic study of Neotropical Chrysophylloideae (Sapotaceae). An aligned data matrix. © 2017 The Linnean Society of London, Botanical Journal of the Linnean Society, 2017, 185, 27–55 Downloaded from https://academic.oup.com/botlinnean/article-abstract/185/1/27/4100608/Towards-a-natural-classification-of-Sapotaceae by Swedish Museum of Natural History user on 01 September 2017

RELATED PAPERS

RELATED TOPICS

Log In

Towards a natural classification of Sapotaceae subfamily Chrysophylloideae in the Neotropics

Towards a natural classification of Sapotaceae subfamily Chrysophylloideae in the Neotropics

Related Papers

RELATED PAPERS

RELATED TOPICS