Figure 2.

The hypothesized (a) vs observed (b–d) difference between T_crit-hot and T_crit-cold for each treatment for species in each biome relative to their respective reference treatment. Based on the resource cost of increased tolerance, our expectation was that shifts in tolerance would represent different trade-offs, depending on treatment. We had hypothesized that hot days would shift species up towards higher heat tolerance on the y-axis with no effect on the x-axis (panel a, upper two quadrants), while we expected cold nights to improve cold tolerance on the x-axis with no effect on the y-axis position (left-hand quadrants). We did not expect any treatment to lead to reduction in both heat and cold tolerance (lower right quadrant). Finally, we expected the combination treatment would be additive, where species responses would reside in the upper left-hand quadrant. We found that alpine species (b) tended to become increasingly heat tolerant at the cost of cold tolerance, regardless of treatment. The desert species (c) showed the additive effect that we expected for the combination treatment, but for all treatments. The temperate species (d) became increasingly heat tolerant but did not deviate much in their cold tolerance.