Genomic, Phenotypic, and Ecological Data Provide Decisive Support for Multiple Species in the Seriphium plumosum Complex (Gnaphalieae: Asteraceae)

Parallel advances in next-generation sequencing and the development of analytical tools that use sequence data from multiple, unlinked loci have, over the past decade, shifted species discovery and delimitation firmly into the genomic sphere. While most recent analytical applications are underpinned by the Multi-Species Coalescent (MSC), the assumption that species conform to Wright-Fisher populations compromises the ability of this model to resolve species which are loosely knit, metapopulation-like entities that display significant spatial genetic structure. In this paper, we therefore employ a suite of analytical methods (MSC-based species discovery, DISSECT/STACEY; sparse Non-negative Matrix Factorisation, sNMF; discriminant analysis of principal components, DAPC) to thousands of massively parallel-sequenced loci, generated via Genotyping-by-Sequencing, for the purpose of resolving metapopulation-like species in the taxonomically recalcitrant Seriphium plumosum L. (Asteraceae; Gnaphalieae) species complex. These methods are complementary in illuminating different characteristics of the entities being resolved (e.g., monophyly, gene pool exclusivity) and so testing different aspects of a species hypothesis. In our study, these methods display remarkable congruence in terms of the groups they reveal, together yielding a set of 12 consensus population clusters which correspond to putative taxonomic species. Using linear discriminant analyses and tests of isolation-by-distance, we further assess whether these entities are phenotypically and ecologically distinct and, where possible, whether they describe a common spatial field of gene exchange. Based on our results, we find evidence of at least eight independent species in the S. plumosum complex, most of which are not truly cryptic, and at least five of which maintain their genetic distinctness in sympatry and so qualify as biological species. Our findings corroborate previous work suggesting a need for diverse tools to properly circumscribe the species-level diversity of South Africa’s Cape flora.