Heat tolerance in relation to leaf thermoregulation
Regression models of T50 and TCrit were consistently better (based on AIC) when they included elevation rather than any of the temperature metrics (TMin, MAT, or TMax). Because temperature and elevation were tightly correlated, we did not include both in the same models. We tested the effects of various leaf traits associated with thermoregulation on heat tolerance. Leaf width did not correlate with T50 or TCrit. However, leaf area and elevation combined yielded a good model for TCrit(R2adj =0.27, F2,118=23.4, p <0.001). Very similar results were obtained for a model containing leaf length in addition to elevation, with R2adj of 0.29 (F2,118=25.5, p <0.001). Species with a higher τ, i.e., species with leaves that take longer to heat and cool, had higher TCrit when accounting for elevation (F2,118=22.6, p <0.001, R2adj=0.27), without an interaction effect of elevation—TCrit was greater at low than at high elevation (Two-sample t test, t = 6.27, P <0.001), whereas τ did not differ between sites (Fig. 5). When substituting MAT for elevation τ also significantly affected TCrit(F2,118=19.6, p <0.001, R2adj=0.24). T50 and b were not affected by τ, and elevation did not interact significantly with τ’s relationship with heat tolerance traits.