CONCLUSIONS AND FUTURE PERSPECTIVES
The combination of genome-wide data with the use of coalescent models of demographic history has enabled us to reject both extreme morphological polymorphism and hybridisation to explain mitochondrial discordance and lack of nuclear sorting in nominal species of Hogna spiders in the Madeiran Archipelago. Instead, our results identified a third independent evolutionary lineage, for which the name Hogna biscoitoi is available, on the small and highly disturbed island of Porto Santo. With this species, the number of endemic Hogna species in Madeira rises to eight.
Our study constitutes a further example of the perils of inferring evolutionary processes based on the evidence provided by the results of population structure analyses. With the introduction of genomics and genome-wide markers, many studies have claimed the existence of hybrid populations or introgressive hybridisation, without further evaluating alternative scenarios compatible with the recovered patterns. We encourage future studies to consider alternative evolutionary scenarios to distinguish between ILS and gene flow and resolve challenging taxonomic situations. In this study, we implemented a straightforward protocol to fully disentangle between genetic signals in recently diverged species, which will contribute to improving the correct assessment of biodiversity in similar situations across biogeographic regions. Although our data have greatly helped to elucidate one of the major challenges in the current taxonomy of an island lineage of spiders, some of which are of conservation concern, we encourage further research to increase both the number of markers and, mainly, the individuals and population sample size to test more specific evolutionary events, such as bottlenecks associated with the colonisation of new islands. Equally important, reconciling the genetic evidence with a quantitative assessment of morphological data will allow for the future identification of the drivers of speciation in this exceptional island model.