Model predictions that can be tested with genetic data
Integrating ENMs and SDMs with genomic data can be used to test whether our inferences about species responses to climate change are accurate and relevant and hence, make better predictions of potential responses to future climate change. The independent assessment of habitat quality in ecological niche models constitutes an improvement since, in addition to the inferences about changes in the species’ distribution ranges, by analyzing changes in niche marginality between LGM and present day we can infer changes in population fitness and selection pressure within a species’ distribution (Figure 5a).
An example is the northern clade of Plestiodon (i.e.Plestiodon skiltonianus , highland species) which shows a contraction of its distribution from LGM to present (Figure 5b, Figure 5c), in addition to a decrease in habitat quality (i.e. increased marginality) at its northernmost distribution. Populations of this species at these locations should show signatures of a bottleneck along with increased selection pressure associated with higher temperatures and lower precipitation in present day climate (Figure 1, Figure 5d). This could be evidenced by adaptations in thermoregulation or water physiology, and positive selection should be focused to genes associated with these processes such as aquaporins (e.g. Araya-Donoso et al., 2021) or heat shock proteins (Chen et al., 2018). An example of a lowland desert adapted taxon is the southern clade of the packrat Neotoma bryanti (Figure 5e), which exhibits a geographical expansion towards the north from LGM to present (Figure 5f), associated with increased habitat quality (i.e. decreased marginality) in the central populations (Figure 5g). Stable populations between LGM and present for this species should reflect an increase in effective population size, and could show signatures of natural selection associated with LGM climate (Figure 5f, Figure 5g) while the northern part of the range may be expected to have lower diversity as a consequence of range expansion as well as surfing of deleterious alleles (Escoffier et al., 2008; Gilbert et al., 2018).
These predictions can be tested with genomic data, evaluating if the patterns of genetic variation reflect the expected changes in effective population size and signatures of selection predicted by our models. According to published genetic data, the mammals Chaetodipus spinatus (Álvarez-Castañeda & Murphy, 2014), Spilogale gracilis(Ferguson et al., 2017), and Otospermophilus becheeyi (Phuong et al., 2017) show population size reduction during LGM, which is in agreement with our prediction of reduced suitable area or increased marginality in the LGM distribution models (except for C. spinatus ). Whole genome sequencing data from populations across the peninsula would be required to evaluate the selection pressure predictions from this study. An example of this approach is Farleigh et al. (2021), who used genomic data to infer changes in population size and potential genes under selection for the lizard Phrynosoma platyrhinos across the North American deserts, formally testing previous hypotheses about demographic changes and adaptation to different climates (Jezkova et al., 2016).