The perils of inferring hybridisation from population structure patterns
Our genome-wide data revealed a population structure closely matching the two nominal species, Hogna insularum, and H. maderiana , but with a third component, formed by individuals from Porto Santo, first identified as H. insularum, of ambiguous interpretation. This pattern was consistently observed across both non-model-based (DAPC) and model-based clustering methods (structure and fineradstructure) (Figure 2). DAPC and fineradstructure analyses identified three genetic clusters, with slightly higher co-ancestry within each cluster than among them, suggesting a certain degree of genetic similarity across the three genetic clusters. While in general agreement, the structure analysis further supported this genetic continuity. At K=3 (as suggested by Evanno’s criterion), we observed varying levels of genetic similarity among the groups with similar dominant components across them, but different secondary components for those individuals outside Porto Santo. At K=2, individuals of H. insularum from Porto Santo exhibited an equal ancestry proportion of 50% from each species, suggesting ongoing hybridisation between them. At this point, it is essential to note that determining the correct number of genetic clusters using Evanno and Pritchard methods in STRUCTURE is challenging, as these approaches can either underestimate or overestimate the number of clusters, particularly in populations with complex evolutionary histories (Janes et al., 2017). In the case of the Hogna species complex, the identification of K=2 suggests that the H. insularum lineage from Porto Santo may result from continued hybridisation between H. insularum and H. maderiana in Porto Santo. Surprisingly, none of the samples collected from Porto Santo belonged to the H. insularum Madeira, Desertas, and Bugio (MDB) (Figure 2), which challenges the putative scenario of a potential hybrid zone between the two nominal species co-occurring in Porto Santo.