Ecological niche and phylogeography elucidate complex biogeographic patterns in Loxosceles rufescens (Araneae, Sicariidae) in the Mediterranean Basin

Abstract

Background: Understanding the evolutionary history of morphologically cryptic species complexes is difficult, and made even more challenging when geographic distributions have been modified by human-mediated dispersal. This situation is common in the Mediterranean Basin where, aside from the environmental heterogeneity of the region, protracted human presence has obscured the biogeographic processes that shaped current diversity. Loxosceles rufescens (Araneae, Sicariidae) is an ideal example: native to the Mediterranean, the species has dispersed worldwide via cohabitation with humans. A previous study revealed considerable molecular diversity, suggesting cryptic species, but relationships among lineages did not correspond to geographic location.

Results: Delimitation analyses on cytochrome c oxidase subunit I identified 11 different evolutionary lineages, presenting two contrasting phylogeographic patterns: (1) lineages with well-structured populations in Morocco and Iberia, and (2) lineages lacking geographic structure across the Mediterranean Basin. Dating analyses placed main diversification events in the Pleistocene, and multiple Pleistocene refugia, identified using ecological niche modeling (ENM), are compatible with allopatric differentiation of lineages. Human-mediated transportation appears to have complicated the current biogeography of this medically important and synanthropic spider.

Conclusions: We integrated ecological niche models with phylogeographic analyses to elucidate the evolutionary history of L. rufescens in the Mediterranean Basin, with emphasis on the origins of mtDNA diversity. We found support for the hypothesis that northern Africa was the center of origin for L. rufescens, and that current genetic diversity originated in allopatry, likely promoted by successive glaciations during the Pleistocene. We corroborated the scenario of multiple refugia within the Mediterranean, principally in northern Africa, combining results from eight atmosphere-ocean general circulation models (AOGCMs) with two different refugium-delimitation methodologies. ENM results were useful for providing general views of putative refugia, with fine-scale details depending on the level of stringency applied for agreement among models.

Figure 1

Map of sampling localities. Green circles represent localities used in ENM analyses; pink stars indicate additional sampled localities used in the phylogenetic analyses.

Figure 2

Refugium delimitation methods. Refugial areas were identified as the intersection of the three time slices. Method 1 sought consensus among the eight AOGCMs in each time slice by requiring at least (a) one to agree, (b) 4/8 to agree, (c) 6/8 to agree and (d) all 8 to agree. Method 2 sought the intersection of the three time slices for each AOGCM independently, subsequently requiring the AOGCMs to agree in the same fashion as M1: at least (e) one to agree, (f) 4/8 to agree, (g) 6/8 to agree and (h) 7 to agree (8/8 was not possible).

Figure 3

Maximum likelihood tree based on single representatives of each cox1 haplotype. Node circles represent maximum likelihood bootstrap and Bayesian posterior probabilities, as shown in the legend. Green indicates the A clade, and blue the B clade. Each column on the right indicates a different delimitation method, and delimited lineages are represented with colored bars. Abbreviations: MA (Morocco), IP (Iberian Peninsula), GR (Greece), IT (Italian Peninsula), TR (Anatolian Peninsula), BI (Balearic Islands), TN (Tunisia), CR (Crete), SC (Sicily), LE (Levant).

Figure 4

Distribution map and haplotype network for each lineage. Colors on haplotype networks correspond to colored areas on the maps to the left (a-f). Haplotype networks are not to the same scale among lineages. Each circle represents one haplotype, and colors correspond to frequencies of region of origin for the haplotype (n = number of individuals). Note the two contrasting phylogeographic patterns, with lineages A1 to A5 restricted to one or a few well-structured populations, whereas the lineages distributed across the Mediterranean Basin generally lack geographic structure, which is likely a consequence of human-mediated dispersal.

Figure 5

Refugial maps developed via two delimitation methods. Maps a, b, c and d were obtained with Method 1, and maps e, f, g and h were obtained using Method 2, as described in Figure 2.