Trait-environment linkages in fragments vs. contiguous forest
Elevation played a key role in driving trait-mediated abundances in contiguous forest. Higher elevations had species with thicker leaves, shorter stature, less dense wood and smaller seeds. The increase in thicker-leaved species was not in response to higher elevations being wetter and colder. Similarly, the increase in short-statured species at higher elevations was not due to more rainfall because taller-statured species increased in sites with more rain (Climate axis 1), consistent with expectations that drier climates favor shorter species (Méndez-Toribio et al., 2020). Notably though, water deficit had a stronger influence than Climate axis 1 on height, with water-limited sites (more negative CWD) having taller species. While this pattern runs contrary to expectation, it may explain the increase in shorter-statured species at higher elevations where water deficit decreased, albeit moderately.
Given these opposing patterns and unclear trends in SLA and height, observed trait-abundance associations may have resulted from stronger selection for another trait in contiguous forest. Seed size was the strongest driver of changes across elevation, and increase in smaller-seeded species at higher elevations would also lead to shorter, thicker-leaved species (PCA1, Fig. S1). Seed size-elevation linkages may have resulted from a decline in larger-seeded species at sites with lower water deficit. In fragments however, neither CWD nor climate axis 1 explained seed size mediated changes in abundance. Additionally, factors unmeasured here could drive compositional change, e.g., greater wind speed at higher elevation may strongly favor shorter-statured species (Umaña and Swenson, 2018), that in this landscape were associated with smaller seeds (Osuri and Sankaran, 2016b). While traits do seem to explain compositional change along the elevational gradient (Krishnadas and Osuri, 2020), the mechanisms clearly need evaluation (Loranger et al., 2018).
Compared to contiguous forest, the seed size-elevation linkage changed in fragments. Larger-seeded species increased at higher elevations, perhaps driven by larger-seeded species being more abundant at cooler sites. Species with higher SLA increased as water deficit decreased, suggesting that sites with lower water stress supported resource-acquisitive strategies (Méndez-Toribio et al., 2020). The relationship with water stress is consistent with patterns across larger-scale gradients of CWD in the Western Ghats, where low SLA species were more likely to occur in water-stressed sites (Krishnadas et al., 2021). Surprisingly though, higher precipitation correlated with an increase in low SLA species, albeit CWD being the stronger driver. This pattern may be attributed to greater shade in sites with higher rainfall; lower SLA values correlate with better shade tolerance (Reich, 2014; Wright et al., 2010). Overall though, trait-environment interactions alone (i.e., without random effects) explained at most ~16% variation in species abundances across the landscape as a whole.