Different species or genetically divergent populations? Integrative species delimitation of the Primulina hochiensis complex from isolated karst habitats

To consistently and objectively delineate species-level divergence from population subdivision has been a challenge in systematics. This is particularly evident in naturally fragmented and allopatric systems in which small population size often leads to extreme population structuring. Here we evaluated the robustness of the species delimitation methods implemented in BEAST, BPP, and iBPP in the Primulina hochiensis complex comprising four described and one candidate species (five taxa in total) distributed in karst landscapes of southern China. We analyzed levels of molecular and morphological divergence among species using multilocus sequence data (nine chloroplast loci and 10 nuclear loci), and morphological data (16 quantitative and 12 qualitative traits), for 124 individuals from 25 populations of the complex. Independent analyses of cpDNA and nDNA sequence data revealed high levels of genetic differentiation among the five taxa. Both BPP and iBPP delimited five candidate species, which correspond to the five genetic clusters recovered with population structure analysis. In contrast, morphological differences among populations were more limited, so that results from principal component analysis (PCA) recovered only three distinct clusters. We ruled out the possibility of morphologically cryptic species because reciprocally monophyletic groups were not supported among the morphologically undifferentiated taxa. Our results represent a case where extreme population genetic structuring leads to oversplit of species diversity by molecular data using the multispecies coalescent (MSC) methods. The observed congruence across multiple analyses corroborates the recognition of a new species P. lianpingensis and indicates its sister species relationship with P. yingdeensis. This study highlights the dangers of violating model assumption and the importance of incorporating multiple evidence into species delimitation of a particular system.