Population genetic structure shaped by isolation-by-distance and historical dynamics
The population genetic pattern that we observe is partly consistent with isolation-by-distance (IBD). The Mantel test showed a significant, positive correlation between pairwise geographic distance and genetic distance although the effect size is very small when individuals with high kinship values are removed from the dataset. Individuals located within 1 km have significantly lower pairwise genetic distances than expected by chance, and this significance drops off by 2 km (Figure S4). Additionally, pairwise F ST between the non-adjacent MT summit and MK clusters is greater than the value between neighboring clusters (Table 2a). High rates of gene flow between adjacent demes across the landscape, with relatively short dispersal distances as suggested by the correlogram and ANOVA (Figure S4, Table 1), could have generated the clinal pattern we observe (Figure 5); however, this does not explain the distinctiveness of the summit MT cluster. The Euclidean distance between the lowest and highest sampled points on MK (ca. 13.5 km) is greater than the distance between the lowest and highest sampling points on MT (ca. 4.5 km), yet there is more population genetic differentiation on MT. This indicates that the structure we observe is not due to isolation-by-distance or isolation-by-elevation alone, and that genetic similarity decays with geographic distance at unequal rates in this landscape (Figure S5a).
Historical population dynamics, in addition to IBD, likely contributed to the observed population genetic structure. Without data from other locations in Borneo, it is difficult to determine what process(es) generated the pattern. However, we suggest a plausible scenario given known information about the relative ages of MT and MK and the degree of divergence between the mountain treeshrew and its sister species.MT reached its current elevation earlier (ca. 11–7 Mya) than MK (ca. 1 Mya) (Collenette, 1964; Hall, 1998; Liew et al. 2010). This suggests that MT was available for colonization prior to the split of mountain treeshrews from ruddy treeshrews ca. 4 Mya (Roberts et al., 2011). If mountain treeshrews were resident on MT prior to the major uplift of MK, and a second colonization event occurred later, this would explain the signature of two population clusters. We find higher-than-average genetic diversity among individuals at high elevation MT despite its smaller habitat area (Figures 1a & S5b), which is consistent with our hypothesis that this region maintained a relatively stable, or recently reduced, effective population size over time relative to MK. Lower gene flow upslope to high elevation MT relative to gene flow towards MK may have preserved the signature of this cluster.
In the MIGRATE-N analysis, the estimated rate of migration upslope to high elevation MT was lower than migration downslope and toward MK (Table S6). We suggest that reduced gene flow to high elevation MT may be related to shifts in the plant community that occur between 1450 masl and the summit (van der Ent, Cardace, Tibbett, & Echevarria, 2018). Supporting this hypothesis, trapping success of mountain treeshrews and other small mammals is low from 1500 to 1800 masl, and increases above 2000 masl (Camacho-Sanchez et al., 2019).
By contrast, lack of differentiation across MK could have been influenced by an upslope shift at the mountain treeshrew’s upper elevational limit, enabled by climate warming and upslope shifts in montane forest since the Last Glacial Maximum (LGM). During late Quaternary glacial cycles, MK’s summit alternated between ice-free (during glacial minima) and ice-covered (during glacial maxima) periods (Hall et al., 2008). After the LGM ca. 20,000 ybp, the ice began to melt and by 9,200 ybp the summit was likely ice-free (Hall et al., 2008). Upslope shifts in montane forest during this period of warming could have enabled range expansion at high elevations by mountain treeshrews. Mountain treeshrews on MT likely did not experience a concurrent upslope range shift since MT has a much lower summit than MK which was never covered in ice (Hall et al., 2008). The lack of a population expansion signature in the mountain treeshrew mitogenome data could be explained by unrestricted gene flow between adjacent areas during expansion (Pierce, Gutierrez, Rice, & Pfennig, 2017). As predicted for a recent expansion, we find lower-than-average genetic diversity among high elevation MK individuals (≥1600 masl) using estimated effective migration surface modeling (Petkova, Novembre, & Stephens, 2016) to visualize genetic diversity on the landscape (Figure S5b).