Species-specific responses to island connectivity cycles: refined models for testing phylogeographic concordance across a Mediterranean Pleistocene Aggregate Island Complex.

The contribution of Pleistocene sea level changes to diversification patterns in archipelagos around the world, and specifically whether the repeated cycles of island connectivity and isolation acted as a 'species pump' is debated. The debate has been perpetuated in part because of the type of evidence used to evaluate the species-pump hypothesis. Specifically, existing tests of the 'Pleistocene Aggregate Island Complex' (PAIC) model of diversification interpret the lack of concordant divergence times among multiple codistributed taxa as a rejection of the PAIC model. However, the null expectation of concordance disregards taxon-specific ecological traits and geographic characteristics that may affect population persistence and gene flow among islands. Here, we study the factors affecting population divergence in thirteen flightless darkling beetle species (Coleoptera: Tenebrionidae) across the PAIC system of the Cycladic plateau in the Aegean archipelago. Based on isolation-by-resistance analyses, hierarchical amova and the degree of genealogical sorting on individual islands, we identify a major effect of bathymetry and habitat stability on the levels of genetic divergence across the PAIC, with island size and body size playing a secondary role as well. We subsequently use bathymetric maps and habitat association to generate predictions about the set of islands and group of taxa expected to show phylogeographic concordance. We test these predictions using hierarchical approximate Bayesian computation and show how our interpretations regarding the role of PAICs as drivers of divergence change when relying on a null expectation of concordance compared to a refined model that takes geography and ecological traits into account.