The role of niche boundaries and dispersal limitations in shaping range limits
The accuracy of our prediction of range limits suggests that they are revealing ecological niche boundaries to some extent. If other environmental factors (e.g. lack of prey or presence of competitors, predators or parasites) had been constraining range expansions, our inferred range would have extended beyond real range boundaries, and realized range edges would be explained by the existence of limitations to expansion before fulfilling all cells within the spatial projection of the species’ niche (Holt 2003, Soberón and Peterson 2005, Holt 2009). In fact, this happened only in the northeastern border of the range, where GEAM predicted the presence of P. algirus beyond the barrier imposed by the Rhône River delta (niche boundaries would actually allow the species to reach Italy).
Also, our results point to an apparent lack of demographical center-border effects such as gene swamping or increased homozygosity near the edge of the range (Herrera and Bazaga 2008; Polechová et al. 2009; Pironon et al. 2017). If these processes were acting, they would be limiting the persistence of marginal populations near range boundaries, and we would systematically infer false positives beyond range limits (Case and Taper 2000; Bridle and Vines 2007; Lee-Yaw et al. 2018). However, this line of reasoning may be wrong, given the likely polygenic nature of adaptations to temperature and humidity (e.g. Rodríguez et al. 2017), if our approach had missed many of the small-effect variants that contribute to such adaptations; these variants should be relatively prone to gene swamping, showing lower core-margin differentiation that the one we inferred (Lenormand 2002, Yeaman 2015). In that case, both allele fixation at a small number of major-effect loci (either the ones included in our model or others with similar differentiation patterns), and gene swamping at a large number of small-effect loci not included in our model, could explain the shape of range limits. In fact, we did wrongly infer a relatively large inland area of false positives on the northern side of the Pyrenees (Fig. 3 and Fig. 4), which could thus be interpreted as consistent with the hypothesis of gene swamping. However, this area was suitable for a small number of genotypes (< 3 out of 46), so that another plausible explanation for these false positives is the low probability that the few genotypes that could be adapted to these unoccupied areas were available in nearby marginal populations (Pujol et al. 2009; Dawson et al. 2010; Barton and Etheridge 2018).
The dispersal ability of the genotypes arising at range margins could play an important role in the colonization of new areas beyond range limits (Simmons and Thomas 2004; Hardie and Hutchings 2010). In our system, for example, the range of P. algirus would extend ca. 13% beyond its eastern European border if lizards were able to disperse across the Rhône River and no other factors were constraining its expansion. Furthermore, genetic diversity would be fostered by greater dispersion abilities (Duckworth 2008), which should facilitate the co-occurrence of adaptive allele combinations at range margins. For instance, the expansion of these small terrestrial ectotherms towards suitable areas north of the Pyrenees (false positives in our model) could be constrained by the synergistic effects of low dispersal rates and a low probability of finding adapted genotypes at nearby marginal populations. To clarify the relative role of dispersal ability vs. other possible constraints in the colonization of new areas beyond range limits (see above), our genotype-based modelling approach could be applied to species showing different dispersal abilities (Sanford et al. 2006; Dawson et al. 2010), or we could perform transplant experiments beyond range limits (i.e. manipulating species’ dispersal abilities; Hargreaves et al. 2014). By doing this, we should be able to discern between the role of dispersal ability per se and the genetic contribution of pre-adapted genotypes arising (or not) at marginal populations (Bridle and Vines 2007; Sexton et al. 2009; Phillipsen et al. 2015).