Ochneae

Schneider, Julio V., Jungcurt, Tanja, Cardoso, Domingos, Amorim, André Márcio, Töpel, Mats, Andermann, Tobias, Poncy, Odile, Berberich, Thomas & Zizka, Georg, 2021, Phylogenomics of the tropical plant family Ochnaceae using targeted enrichment of nuclear genes and 250 + taxa, TAXON 70 (1), pp. 48-71 : 63-64

Ochneae .

— Ochneae as defined in Schneider & al. (2014) contains the African Lophirinae (with Lophira as the sole genus) as sister to a clade of the neotropical Elvasiinae (with Elvasia and Perissocarpa ) and the pantropical Ochninae ( Brackenridgea , Campylospermum , Idertia , Ochna , Ouratea , Rhabdophyllum ). This classification was supported in the present study, too, with all clades receiving strong support, including the sister relationship of Elvasia and Perissocarpa , which was analyzed here for the first time using molecular data.

Ochninae is the most species-rich clade of Ochnaceae , containing approximately two-thirds of the family’ s species. The associated radiations most likely benefitted from the emergence of the savanna biome in the Old and New World as inferred from the time frame of the divergence events and the actual species distributions. The split into the six currently accepted genera occurred over a very short period of about a maximum of 5–10 million years during the Miocene (Schneider & al., 2017), which is certainly also one reason for the hitherto unclear relationships among them. In the present study, all backbone nodes of Ochninae were recovered with strong support, thus resolving its relationships for the first time. The first split divides Ochninae into the neotropical Ouratea and the remaining genera, which are all palaeotropical.

Perhaps most remarkable within the palaeotropical clade is the polyphyly of Campylospermum , which was already revealed in Bissiengou (2014), but which is here recovered with strong support. Clade A of Campylospermum contains all Central and West African species of this genus (except C. elongatum (Oliv.) Tiegh. ) and forms a clade with the African Rhabdophyllum . Clade B of Campylospermum contains the Malagasy (here, only C. obtusifolium Tiegh. included) and East African species (plus C. elongatum ) and is sister to Brackenridgea , both being sister to the West to Central African Idertia . It is important to mention that both clades share their own set of characters: Clade A usually has terminal inflorescences, and the embryos are either accumbent or incumbent and similar or dissimilar in size, whereas clade B shares usually axillary inflorescences, accumbent embryos that are similar in size, and a distinct reticulate tertiary leaf venation ( Bissiengou, 2014). Sosef (2008) also noted that the leaf venation of Campylospermum lecomtei (Tiegh.) Farron and C. paucinervatum Sosef , both belonging to our clade A, resembles that of Rhabdophyllum , thus providing further support from morphology for the here inferred sister-group relationship between both groups. However, we are awaiting a modern revision of the Malagasy species ( Bissiengou, 2014 only treated the continental species) before making the necessary nomenclatural changes.

Brackenridgea originally comprised only species from SE Asia, Australia and Oceania ( Van Tieghem, 1902). The morphologically similar genus Pleuroridgea Tiegh. was erected by Van Tieghem (1902) to unite species that share embryos with laterally disposed cotyledons and lateral, deeply divided caducous stipules, initially all from continental Africa. This concept was followed by Perrier de la Bathie (1941) but with an expanded Pleuroridgea to include new Malagasy species. Subsequent authors ( Robson, 1963; Du Toit & Obermeyer, 1976; Verdcourt, 2005; Callmander & al., 2010), however, merged both genera under a broad concept of Brackenridgea , judging the morphological differences too small for maintaining them separate. However, it is remarkable that in our phylogeny the former Brackenridgea s.str. and Pleuroridgea form well-supported geographically distinct clades within a monophyletic Brackenridgea . One clade unites the Asian-OceanianAustralian species B. palustris Bartell., B. foxworthii Furtado (= B. palustris Bartell. ), B. nitida A.Gray (= B. nitida subsp. australiana (F.Muell.) P.O.Karis ) and B. forbesii Tiegh. , the second one unites the African-Malagasy species B. madecassa (H.Perrier) Callm. , B. arenaria (De Wild. & T.Durand) N.Robson and B. zanguebarica Oliv. , the latter itself divided into a Malagasy and a continental (African) clade. Thus, the disjunct African-Asian distribution, which is shared with Ochna and Campylospermum , most likely resulted from a single crosscontinental dispersal event, perhaps involving Madagascar as an intermediate step.

Here, we present the by far most comprehensive phylogenetic framework for Ochna , comprising about 50% of its species. While a modern taxonomic revision of this genus is still lacking, this framework serves as important baseline data for future infrageneric classification. Earlier attempts for an infrageneric classification included the one by Van Tieghem (1902). He segregated 15 genera using characters such as anther dehiscence (poricidal versus longicidal), carpel number or embryo morphology (iso- versus heterocotyledonous). More recent regional treatments on African ( Robson, 1963; Du Toit & Obermeyer, 1976; Verdcourt, 2005; Callmander & Phillipson, 2012) or Asian ( Kanis, 1968) species of Ochna did not follow Van Tieghem’ s narrow concept. In the present study, there is a remarkable congruence of our phylogeny with the infrageneric classification of Robson (1963), who divided the genus into three sections based on carpel shape and anther dehiscence (see also Verdcourt, 2005). Our clade A contains species belonging to Robson’ s Ochna sect. Ochna , which is characterized by bi-porous anthers. Clade B contains species of his O. sect. Schizanthera in which anthers open by longitudinal slits. In contrast, his species-poor O. sect. Renicarpus, which differs by having reniform drupelets that are attached near the centre of the long side of the drupelet, appears to be polyphyletic according to the placement of Ochna arborea Burch. ex DC. and O. pulchra Hook. However , a more comprehensive taxon sampling of this section is required for a sound evaluation of its status. Besides the two large clades, there are two smaller clades. One contains the morphologically close O. polycarpa Baker and O. louvelii (H.Perrier) Callm. & Phillipson , which are from Madagascar ( Callmander & Phillipson, 2012). The other one is sister to the rest of Ochna and comprises O. andravinensis Baill. , O. pulchra Hook. and O. latisepala (Tiegh.) Bamps. How these smaller clades are integrated into an infrageneric classification and how species of O. sect. Reniformis are accommodated, is, however, beyond the scope of the present study and needs to be assessed in the framework of a modern taxonomic revision of this genus. From an ecological perspective remarkable is the observation that several of the species for which a fire-adapted geoxylic growth form has been reported (which are elements of the so-called underground forests; see White, 1976) were positioned in a subclade of clade C. The geoxylic species include, for example, O. leptoclada Oliv. , O. katangensis De Wild. , O. confusa Burtt Davy & Greenway and O. pygmaea Hiern. From a biogeographical perspective noteworthy are our findings that the Asian species ( O. integerrima (Lour.) Merr. , O. obtusata DC. ) included in this study form a clade that is nested in clade D and that the Malagasy species (e.g., O. brachypoda Baill. , O. macrantha Baker , O. polycarpa , O. andravinensis ) are spread across three major clades of Ochna .

The largest genus of Ochnaceae is Ouratea , which is among the larger woody genera of the New World with about 200 ( Amaral & Bittrich, 2014) to 310 accepted species (Schneider, unpub. data), comparable in size to other well-known radiations in tropical plant families such as Inga Mill. (ca. 300 spp., Fabaceae ; Richardson & al., 2001), Ocotea Aubl. (ca. 300 spp., Lauraceae ; Madriñán, 2004), Clusia L. (ca. 300 spp., Clusiaceae ; Gustafsson & al., 2007), or Guatteria Ruiz & Pav. (ca. 265 spp., Annonaceae ; Erkens & al., 2007). The lack of a comprehensive taxonomic revision of the genus hampers the interpretation of the here presented phylogeny. However, different infrageneric classification systems are available. First attempts for an infrageneric classification of this genus were largely based on unreliable characters such as leaf indumentum and texture ( Erhard, 1849). Later, Engler (1876) erected the two series, Ouratea ser. Cardiocarpae and ser. Oocarpae, using fruit characters for the then known 76 species. The first differs by its obcordate drupelets from the second, which is characterized by mostly ovoid to obovoid drupelets. With Van Tieghem’ s (1902) classification, the number of subdivisions increased drastically. He distinguished 22 genera for his subtribe Orthospermeae (which corresponds to modern Ouratea ; see Sastre & Offroy, 2016) using differences in inflorescence type and insertion, indumentum, number of styles, embryo shape or the persistence of stipules, among others. Subsequent authors disregarded Van Tieghem’ s system (e.g., Gilg, 1925). Perhaps the more satisfying classification of Ouratea was introduced by Sastre (1988). He subdivided Ouratea into the six sections O. sect. Cardiocarpae (Engl.) Sastre, sect. Kaieteuria (Dwyer) Sastre, sect. Polyouratea (Tiegh.) Sastre, sect. Ouratea , sect. Ouratella (Tiegh.) Sastre and sect. Persistens Sastre. This classification was superseded by his more recent one ( Sastre, 1995), in which O. sect. Caducae Sastre was newly erected, whereas sect. Persistens was placed in the synonymy of sect. Ouratea . Characters of diagnostic value for the different sections are, for example, the possession of 2–4 fused sepals in O. sect. Kaieteuria, fruits with the drupelets borne horizontally in sect. Cardiocarpae. Other important characters are the number of carpels (e.g., 6–10 in sect. Polyouratea), fruits with persistent sepals (sect. Ouratea ) and the position of the inflorescence (terminal versus axillary, the first characteristic of sect. Caducae).

Here, based on our phylogenetic analysis, including roughly half the species reported for Ouratea , we tentatively defined five major clades for the ease of discussion, all except clade D with maximum support. Mapping Sastre’ s (1995) sections for species for which we took the relevant information from his publications ( Sastre, 1988, 1995, 2001, 2007) revealed that all sections were polyphyletic (suppl. Fig. S5). In particular, Ouratea sect. Kaieteuria and sect. Ouratella were spread across most of the clades. Topological artefacts as resulting from methodological issues and sometimes associated with either the concatenation or the coalescent approach can be ruled out as an explanation for the lack of congruence between our phylogeny and the sectional classification because all five major clades are recovered identically with both methods of phylogenetic inference.

We also evaluated Van Tieghem’ s (1902) classification but did not find any congruent pattern. His genus Trichouratea Tiegh. is the only one which coincides in part with a subclade of our clade E. However, it is also polyphyletic and intermingled with species from two other genera he had defined. Therefore, many of the characters used for the infrageneric classification are supposedly homoplastic. This is most likely due to the explosive radiation during the early history of Ouratea as inferred from the short branches along its backbone, and which also might explain why previous morphology-based infrageneric classifications in such speciose angiosperm genera often fail in defining monophyletic groups (e.g., Simon & al., 2011; Goldenberg & al., 2018; Moonlight & al., 2018). Only at the level of smaller subclades, we discern some congruence with character combinations that defined Sastre’ s sections. For example, in clade D, we found a subclade uniting O. guildingii (Planch.) Urb. , O. grosourdyi (Tiegh.) Steyerm. , O. mexicana (Bonpl.) Engl. and O. pseudoguildingii Sastre , which all share the set of characters of Ouratea sect. Ouratella . Another subclade of clade D unites species belonging to O. sect. Kaieteuria ( O. steyermarkii Sastre , O. thyrsoidea Engl., O. clarkia Sastre, O. arbobrevicalyx Sastre ), and there is also a clade uniting the Caribbean species O. illicifolia (DC.) Baill. , O. agrophylla (Tiegh.) Urb. , O. elliptica (A.Rich.) M.Gómez , O. lenticellosa Urb. , O. laurifolia (Sw.) Engl. and O. jamaicensis (Planch.) Urb. , some of which have been included in Van Tieghem’ s (1902) Camptouratea. In clade E, we observe a subclade uniting species of Ouratea sect. Polyouratea ( O. scottii Sastre , O. discophora Ducke , O. decagyna Maguire ). Although we may find consistent character combinations for smaller species groups, it might be difficult to find such combinations for the major clades in view of the short branches during early divergence events creating a scenario of incomplete lineage sorting at deeper time ( Xu & Yang, 2016) and many clades with not yet well-fixed traits. However, morphological characters of diagnostic value that describe the here retrieved major clades will have to be searched and evaluated in an urgently needed taxonomic revision of that genus.