Effects of elevation and site temperature on heat tolerance and leaf traits
T50 and TCrit both decreased significantly with elevation (F1,150=35.1 and 46.1, for T50 and TCrit, respectively,p <0.001) (Fig. 1). Including CAM and C4plants in the analysis did not change the slopes, but decreased the variance explained—all CAM and C4 plants were collected from lowland sites and heat tolerance was higher in CAM plants, and lower in the C4 plant than in C3 plants. T50 and TCritdecreased by 0.26°C and 0.39°C per 100 m, respectively (0.28°C and 0.41°C per 100 m when including non-C3 species). The lapse rate across the study sites was 0.63°C decrease in mean annual temperature (MAT) per 100 m increase in elevation.
Heat tolerance increased significantly with MAT (F1,150=30.0 and 38.0, for T50 and TCrit, respectively, p <0.001) (Fig. 1). For every 1°C increase in MAT, T50 increased by 0.41°C and TCrit increased by 0.60°C. T50 and TCrit also correlated significantly with the mean maximum temperature of the warmest month and with the mean minimum temperature of the coldest month (Fig. S4). The b parameter, the steepness of the decline in Fv/Fm around T50, significantly decreased with elevation and increased with MAT, so Fv/Fm declined less steeply in cooler, higher elevation species than in lowland species. However, elevation and MAT only explained a small amount of variance in b (r2≤0.06).
Consistent with the patterns across species, within species there was a tendency for heat tolerance to decrease with elevation and increase with site temperature (irrespective whether minimum, mean, or maximum) (Fig. 2). Among lowland sites (between 0 and 200 m elevation, with MAT differences <1°C) patterns in heat tolerance with elevation and MAT were not consistent and the confidence intervals of the two values tended to overlap (Fig. 2). T50 and TCrit decreased with elevation by 0.4 and 0.7°C per 100 m, respectively, based on a regression weighted by elevation difference between sites—to reduce the weight of species measured at two lowland sites. T50 and TCrit increased with site temperature by 0.9 and 1.5°C per °C MAT, respectively.
To enable comparison of heat tolerance among plant categories we standardized T50 and TCrit to sea-level values using the trendlines in Fig. 1. Standardized T50and TCrit were not significantly different between native and non-native species, or between evergreen and deciduous species (Fig. S5). Gymnosperms tended to have moderately higher T50 than angiosperms (p <0.1, two-sample t-test) because of the high heat tolerance of three Zamiaspecies, but the sample size of gymnosperm species (n=5) was insufficient to draw meaningful biological conclusions from these apparent differences. For similar reasons, cycads (i.e., Zamiaspecies in our study), had higher T50 than lianas (n=6), shrubs (6), and trees (134), and higher TCrit than lianas (Fig. S5). When including the CAM species Agave americanaand Furcraea cabuya , forbs had higher T50 than most other functional groups (not shown).