Dispersal limitations drive community structure across multi-hierarchical levels at fine geographical scales
To investigate whether dispersal limitations shape distance-decay relationships within a mountain, we performed an IBD and IBR analyses on Diptera and Collembola, which have contrasting dispersal capabilities (winged and unwinged, respectively). The unified neutral theory of biodiversity predicts that similarity in species composition decreases with distance due to dispersal limitation (Hubbell, 2001). To test this, we examined if DDRs at multi-hierarchical levels can be explained by the dispersion limitation in addition to distance. Specifically we analyzed which landscape variables (Figure S5) influenced DDRs in arthropods of contrasting dispersal abilities (Collembola and Diptera), considering our entire sampling in Nevado de Toluca (19 km max distance among sampling points) and finer geographic scales within the East (<5 km) and West (<2 km) subsets of our sampling. We found that decay of similarity of communities decreases with spatial distance at the level of haplotypes, all CLs (0.5 to 7.5% lineages) and putative molecular species by GMYC (Figure 5). This occurs both in Collembola and Diptera, but is more marked in Collembola whose dispersal abilities are more limited. Interestingly, for Collembola our results also hold both considering our entire sampling as well as finer (<2 km) geographic distances (Figure 5, 6) which is consistent with genetic studies within Collembola showing genetic differentiation over very short geographic distances (Cicconardi et al., 2013; Faria et al., 2019).
High dispersal ability is expected to enhance community similarity (Baselga et al., 2012). Our results support this, because the fit of the decay is higher in the wingless Collembola (r 2 = 0.704 andr 2 = 0.599 at the haplotype and GMYC levels, respectively; Table S3; Figure 5a) than in the winged Diptera (r 2 = 0.293 and r 2 = 0.195 at the haplotype and GMYC levels, respectively; Table S4; Figure 5c). Similar patterns of higher distance decay relationships at multi-hierarchical levels in poorly dispersing organisms than in better dispersers were found in European water beetles (Baselga et al., 2013), Iberian leaf beetles (Baselga et al., 2015) and European beetles (Gómez-Rodríguez & Baselga, 2018) at much larger (hundreds of km) geographical scales than here (but see Gómez-Rodríguez et al., 2019 where the pattern was not clear for terrestrial molluscs). Communities of good dispersers are more homogeneous not only because they can disperse larger distances, but also because they can more easily overcome geographical barriers between suitable habitat (Thompson & Townsend, 2006; Vellend, 2010). Thus, if dispersal abilities matter, then landscape features impeding dispersal may also play a role in structuring diversity, which can be explicitly tested including landscape features in analyses such as IBR.
IBR quantifies ‘effective distances’ between communities that may yield more biologically informative DDRs than Euclidean distance (McRae, 2006; McRae et al., 2008). Our results show positive significant correlation with different explanatory power depending on the surface used, with altitudinal differences better explaining similarity decay than distance alone (“flat” landscape), slope or vegetation type. The resistance surface “flat” (i.e., IBD) has slightly less explanatory power for Collembola (r 2 = 0.704 at the haplotype level,r 2 = 0.599 at GMYC; Table S3; Figure 5a) than “Altitude 3,000” (r 2 = 0.723 at the haplotype level, r 2 = 0.644 at GMYC; Table S3; Figure 5b), the best fitting resistance surface. This resistance surface corresponds to the elevation at which the Nevado de Toluca volcano massif begins (Figure S5; Table S2), suggesting that Collembola followed a pattern of IBD and that their limited dispersal is not impacted by landscape features. For Diptera, the highest explanatory power was provided by the resistance surface “Altitude B” (r 2 = 0.319 at the haplotype level,r 2 = 0.228 at GMYC; Table S4; Figure 5d). This resistance surface assumes maximum conductance at the mean altitude of our sampling and a gradual decrease until reaching altitudes outside of our sampling range, but still where Abies forest can be found (Table S2). This suggests that besides being able to disperse larger distances, Diptera moves through relatively unsuitable conditions (different altitudes) less efficiently. Therefore, for Diptera it is not distance alone that drives community structure, but also landscape features. Thus, although our sampling blocks are separated by short distances from 50 m to 19 km, connectivity among sites for Diptera depends upon the elevation model used to set the conductance values (Table S2; Figure S5). This is congruent with Janzen’s prediction of “mountain passes being higher the tropics” (Janzen, 1967), and adds to the recent empirical data (Polato et al., 2018) corroborating it. However, although our results show that landscape connectivity contributes to dispersal limitation, geographic distance seems to play a more dominant role both for both orders. This is consistent with dispersal limitation acting over evolutionary time, as has been suggested to explain the small spatial scale diversification ofScarelus beetles within tropical mountains (Bray & Bocak, 2016).
Distance decay patterns at the species level could reflect environmental heterogeneity spatially correlated (i.e., between western and eastern sides of the Nevado de Toluca). While some degree of environmental distance could impact on the obtained biodiversity patterns (but see above on the homogeneity of the sampling study habitat), our results on i) spatial patterns of community dissimilarity recurrently found for multiple hierarchical levels, including haplotypes which are expected to behave neutraly across environmental gradients; ii) high values of turnover and local endemicity at multiple spatial scales and iii) the consistent multihierarchical pattern of distance decay in community similarity at reduce scales (within mountain sites) for less dispersive species, while substantially diluted for the more dispersive ones, point to dispersal limitation within this single single sky-island as a major driver of community assemblage.
Multi‐hierarchical approaches are useful to assess whether variation in biological assemblages driven by dispersion follow a fractal geometry where the same neutral processes underlie the distribution of haplotypes and higher clustering levels (Baselga et al., 2013, 2015). Fractal patterns have been revealed in aquatic beetles, leaf beetles, and terrestrial molluscs, highlighting an important role for neutral processes in the spatial structuring of biodiversity (Baselga et al., 2013, 2015; Gómez-Rodríguez et al., 2019). Our results also reveal the existence of a fractal pattern for DDRs, with similarity decreasing with spatial distance from the level of haplotypes to 7.5% CLs (Table S6). These patterns represent a considerably finer geographic scale than that reported in previous studies (from 820 km to 4,500 km as in Baselga et al., 2013, 2015). DDRs decreased with distance at even finer geographic scales (<5 and <2 km) in Collembola (Table S5, Figure 6 and S6), revealing that for arthropods with low dispersal ability, DDRs can occur at very fine geographic distances and at all multi-hierarchical levels, within the geographic confines of a sky island. Further to this, we reveal that DDRs can also emerge within arthropod groups that are typically considered as good dispersers (Diptera), but at comparatively larger geographic distances (Figure 5). Our results align well with analyses performed in Iberian forest and grassland mesofauna, where DDRs were found at all hierarchical levels, also in less than 15 km, for soil taxa with low dispersal abilities (Arribas et al., 2020). Given these short distances, our findings are important not only for understanding evolution, but also for biomonitoring efforts aiming to detect changes in community assembly, even in relatively short distances among sampling points.