Fig 1.
A. The evolutionary relationships, external appearance, western Mediterranean distribution and ecological preferences for Polyommatus icarus, P. eros and P. celina. Polyommatus icarus and P. celina are almost identical ecologically and morphologically, but P. icarus is phylogenetically closer to P. eros, a morphologically differentiated mountain species that feeds on a more restricted number of host plants. B. The evolutionary relationships, external appearance, western Mediterranean distribution and ecological preferences for Aricia agestis, A. artaxerxes, A. montensis and A. cramera. Aricia agestis is a habitat generalist phylogenetically the closest to a pair of specialist species:Aricia artaxerxes (boreo-montane) and A. montensis (montane). Aricia cramera is phylogenetically more distant to A. agestis, although both species are almost identical ecologically and morphologically.
Fig 2.
Genetic landscapes obtained for the sequenced specimens of the two pairs of cryptic species.
The color gradient represents residuals of COI genetic p-distances. A. Polyommatus icarus and P. celina. B. Aricia agestis and A. cramera.
Fig 3.
Haplotype networks and evaluation of dispersal events over the sea.
Inferred over-sea dispersal events are numbered in grey rectangles, and ambiguous cases that were excluded from the analyses are highlighted in pink squares. The lengths of the dispersal events for the study area have been measured according to the-50m isobath and are displayed in histograms (cyan blocks), together with minimal dispersal events based on the distribution of the same species in the eastern Mediterranean (red blocks) and Canary Islands (blue blocks). The width of the key sea-straits Sardinia-Corsica, Italy-Sicily, Tunisia-Sicily and Ibero-Balearic are indicated by blue, red and green arrows, respectively.
Fig 4.
Projection over the study area of the logistic values obtained by MaxEnt analyses (H model) for the four species.
The color gradient indicates the logistic probabilities of species occurrence, increasing from white (very low probability) to red (very high probability). The maximum value of the lowest class (white) is represented by the maximum training sensitivity plus specificity logistic threshold returned by MaxEnt. The key areas where the species are predicted to occur but are not present despite geographical proximity are highlighted.
Fig 5.
Projection of the climatic variable precipitation in the driest quarter and logistic responses for each species.
A. The color gradient on the map indicates the precipitations (in mm) over the study area. B. The logistic responses of the precipitation in the driest quarter tested alone for the four studied species (P. icarus, P. celina, A. agestis, A. cramera from left to right). The response showed a clear and recurrent threshold between 50 and 100 mm of precipitation.