Clearing confusion in Stylosanthes taxonomy: 1. S. seabrana

Stylosanthes seabrana was first formally described as a new species in 2002 following extensive morphological and agronomic characterization, accompanied by genetic and molecular studies. Since then it has been proposed as a synonym of Stylosanthes scabra Vogel. This paper refutes this synonymization and indicates the indisputable evidence that S. seabrana, a diploid, is a likely putative progenitor of the allotetraploid S. scabra.


Introduction
Since recognition in Australia of the forage value of the adventive species, Stylosanthes humilis Kunth, in the early 20th century, there has been continuing focus on the genus, Stylosanthes, to determine the commercial pasture potential of other species within the genus. Of the 40 species of Stylosanthes currently accepted by the US National Plant Germplasm System (GRIN), 7 have been demonstrated to have commercial agricultural merit. Large collections of a number of species were assembled by CIAT in Colombia and CSIRO in Australia, including shrubby stylo (Stylosanthes scabra) that was found to have potential in the acid, infertile soils of subhumid and semi-arid northern Australia. The most recent addition to the list of commercial species in the genus, S. seabrana B.L. Maass & 't Mannetje, has proven well-adapted to the slightly acid to alkaline, more fertile clay and clay-loam soils in the same region, but extending into the subtropics.

Taxonomy of Stylosanthes seabrana
While characterizing the S. scabra collection held by CIAT in Colombia, Maass (1989) identified a group of plants from Bahia state in Brazil that shared a number of morphological characteristics with S. scabra but were morphologically and agronomically different from S. scabra and other known species of Stylosanthes. Following the provisional name given to this form by plant collectors, she referred to the group in her classification as "cf. scabra-Type". This promising phenotypic group was subsequently referred to as S. sp. aff. scabra by Jansen and Edye (1996), and eventually as "Stylosanthes seabrana" by Edye et al. (1998), accepting that S. sp. aff. scabra was indeed a different species from S. scabra. The name was selected in reference to the town of Seabra in the region of Bahia state, where the earliest accessions of the species were collected. The scientific name, Stylosanthes seabrana, was formalized by Maass and Mannetje (2002). The common name, Caatinga stylo, was adopted in Australia, referring to the xerophytic Caatinga vegetation type in northeastern Brazil on the medium-to heavy-textured soils on which the species is largely found. Vanni and Fernandez (2011) disputed the conclusion of Maass and Mannetje (2002), claiming instead that S. seabrana is a synonym of S. scabra, a claim that is hereby rejected based on a comprehensive assessment of all relevant information/evidence.

Morphology
To help direct future plant evaluation in the subhumid/semi-arid tropics of Australia, and to obtain a clearer picture of the taxonomic and agronomic boundaries of a number of promising Stylosanthes species, morphological and agronomic classificatory experiments involving large numbers of entries of S. scabra, S. hamata (L.) Taub. and S. sp. aff. scabra (= S. seabrana) were conducted at CSIRO Lansdown Research Station, north Queensland (Jansen and Edye 1996;Date et al. 2010). Each used a numerical classification program, PATN (Belbin 1995) that, at the 5 group level, separated entries largely into homogeneous groups, with S. sp. aff. scabra separated from S. scabra. Shapes of the terminal leaflet and the terminal leaflet apex, the presence or absence of leaflet and stem hairs, the presence or absence of inflorescence bristles, the presence or absence of stipule horn lateral bristles and stipule horn terminal bristles were the most useful attributes defining groups. Maass and Mannetje (2002) used the most consistent of these and other observations to develop a key to distinguish the 3 species morphologically. Many collections and studies of Stylosanthes scabra have been conducted since Vogel (1838) described the specimen from Serra da Moeda, Minas Gerais, Brazil and Mohlenbrock (1957) reviewed the genus, Stylosanthes. On this basis, it can be presumed that the Edye and Topark-Ngarm (1992) description based on research experience and the description of Costa and Ferreira (1984) might be more comprehensive than earlier keys. Vanni and Fernandez (2011) provide what they call a "standard description" of S. scabra, which differs from those of Vogel (1838), Mohlenbrock (1957) and Costa and Ferreira (1984) /Edye and Topark-Ngarm (1992), all varying somewhat in their choice of descriptors. However, some characteristics provided in the various keys help to further distinguish S. seabrana from S. scabra morphologically. A characteristic not used in the Maass and Mannetje (2002) key is the length of the axis rudiment, 7-8 mm in their description of S. seabrana and 4-5 mm in S. scabra (Mohlenbrock 1957;Edye and Topark-Ngarm 1992).

Agronomy
There are clear agronomic differences between S. seabrana and S. scabra. Early research in the 1960s and 1970s to identify other Stylosanthes species to extend the range of S. humilis identified the potential of S. scabra and the tetraploid form of S. hamata (= S. hemihamata nom. nud.), resulting in the release of cultivars of each. However, while these were very effective in the light, acid infertile soils of northern Australia, they were not adapted to the heavier, more fertile clay soils in the region. Attention was then turned to the group of Stylosanthes sp. aff. S. scabra that were collected on broadly similar soils in Brazil (Edye and Maass 1997). These proved welladapted to heavy-and medium-textured alkaline soils in Australia, and unlike S. scabra, were also adapted to the more frost-prone environment of southern Queensland (Edye and Hall 1993;Jansen and Edye 1996). CSIRO applied for Plant Breeders Rights for the 2 most promising lines in 1996 (granted in 1997) as "Caatinga Stylo (Stylosanthes sp. nov. aff. S. scabra) cvv. Primar and Unica" to provide a legume base for forage systems on neutral to alkaline soils of central and southern Queensland.
Early evaluation highlighted another important difference between the 2 species. While S. scabra is promiscuous in its root nodule bacterial requirements, nodulating effectively on native strains of Bradyrhizobium in Australia or the broad spectrum CB 756 commercial strain (Date 1997), this was not the case for Caatinga stylo. During field evaluation at a range of sites in Queensland in the 1990s, Caatinga stylo accessions nodulated poorly and ineffectively and frequently failed to nodulate at all (Edye 1994;Edye et al. 1998). Most accessions grew well for 1 or 2 years, before beginning to show classical signs of nitrogen deficiency. Success of the new cultivars was contingent on discovery of an effective and persistent strain of inoculum. Accordingly, nodules were collected during germplasm collections in Brazil, and strains of Bradyrhizobium were isolated, tested and released prior to release of cvv. Primar and Unica (Date 2010;2016).

Ploidy
A major part of the argument advanced by Vanni and Fernandez (2011) revolves around their finding both diploid and tetraploid specimens in the roots of seedlings grown from a sample of commercial seed of S. seabrana cv. Unica from Australia. In their Introduction, they make the following confusing statement: "In addition, they (referring to Maass and Mannetje 2002) reported different levels of ploidy in S. scabra, 2n = 40 chromosomes and S. seabrana, 2n = 20 chromosomes." The ploidy cited for the 2 species is correct; however it in no way supports their contention of dual ploidy in S. scabra. Rather, Vanni and Fernandez (2011) use this confusing statement to support their claim that: "ploidy levels are not valid criteria for species distinction in the genus Stylosanthes, as S. scabra has been reported to be one of the few species with diploid (2n = 20) and tetraploid (2n = 40) genotypes (Cameron 1967)." This is not the case. In fact, Cameron (1967) determined the chromosome number for a single accession of S. tuberculata (presumably Stylosanthes tuberculata S.F. Blake syn. S. scabra Vogel), which he found to be tetraploid (2n = 40) only. Since then a number of workers (Battistin and Martins 1987;Liu et al. 1999;Lira 2015) have reported tetraploidy in S. scabra. No report of diploidy in the species exists in the published literature.
'Unica' was derived from CPI 110361, which has been shown to be diploid (Liu and Musial 1997), so the question arises: how could there have been the 2 ploidy levels in the sample tested by Vanni and Fernandez (2011)? The answer lies in the fact that the seed lot on which Vanni and Fernandez (2011) based their taxonomic revision was a commercial sample. Since seed crops of both S. scabra and S. seabrana are grown in the same general area in north Queensland, it is probable that a commercial sample of seed may contain both species, either from contamination in the crop (S. scabra is now naturalized in the region), in the harvester from a previously harvested crop of S. scabra or during post-harvest handling. There is no seed certification scheme for this cultivar in Australia and postharvest cleaning procedures for harvesting machinery are not as stringent for standard commercial crops as for certified crops.

Phylogeny
Until relatively recently, morphological characters were the only means of describing species, but they have not always provided the level of resolution required to categorically define interspecific and intraspecific differences. Vanni and Fernandez (2011) consider that the form of leaflets, the absence or presence of bristles and hairs on stipules and leaflets and their venation are not sufficient to separate species. Whether or not this is valid is debatable. However, the evidence provided from genetic and molecular studies is indisputable. As discriminatory methodologies improved with the development of molecular technologies, so did the evidence to more clearly define relationships within and between taxonomic groups.
It has been shown that S. scabra is an allotetraploid with S. viscosa Sw. as one of the putative diploid progenitors (Stace and Cameron 1984;Vander Stappen et al. 2002). The identity of the other diploid progenitor is not so clearcut. Stace and Cameron (1984) postulated that, since S. scabra bears an axis rudiment on the loment, a characteristic governed by a dominant gene, and S. viscosa lacks an axis rudiment (section Stylosanthes), the other parent must bear an axis rudiment (section Styposanthes). Working with chloroplast DNA, Gillies and Abbott (1996) proposed S. hamata sensu stricto as the section Styposanthes progenitor, while Liu and Musial (1997) provided evidence that the other putative progenitor was Stylosanthes sp. aff. S. scabra (= S. seabrana). These 2 species fell into the same basal genome group A, determined by restriction fragment length polymorphisms (RFLP) and sequence-tagged-sites (STS) analyses by Liu et al. (1999). In the same study, S. viscosa fell into basal genome group B and S. scabra into group AB. More recent work (Tewari and Chandra 2008; Chandra and Kaushal 2009;Marques et al. 2018)

Conclusion
Stylosanthes seabrana is clearly morphologically, agronomically, rhizobially, cytologically and phylogenetically different from S. scabra (Appendix I), and taxonomic logic dictates that it must be treated as a separate species. It is no more conspecific with S. scabra than is its other putative progenitor, S. viscosa. Similar confusion is faced by practitioners in relation to 2 other Stylosanthes diploid-allotetraploid derivative pairs, S. hamata -S. hemihamata nom. nud. and S. macrocephala -S. capitata, that will be dealt with in subsequent papers in this series.