Fungi are abundant and ecologically important at a global scale, but little is known about whether their thermal adaptations are shaped by biochemical constraints (i.e. the Hotter is Better Model, HBM) or evolutionary tradeoffs (i.e., the Specialist Generalist Model, SGM). We tested these hypotheses by generating thermal performance curves (TPCs) of fungal cultivars farmed by six species of Panamanian fungus-farming 'attine' ants. These fungi represent evolutionary transitions in farming strategies as four cultivars are farmed by ants belowground at stable temperatures near 25°C and two cultivars are farmed aboveground at variable temperatures. We generated TPCs using a common garden experiment confining fungal isolates to different temperatures and then used a Bayesian hierarchical modeling approach to compare competing temperature sensitivity models. Some thermal performance traits differed consistently across farming strategies, with aboveground cultivars having: 1) higher tolerance to low temperatures (CTLmin) and 2) higher maximum growth rate at the optimal temperature (rmax). However, two core assumptions shared by the HBM or SGM were not supported as aboveground cultivars did not show systematic increases in either their optimal temperature (Topt) or thermal tolerance breadth. These results harness ant farming systems as long-term natural experiments to decouple the effects of environmental thermal variation and innate physiological temperature sensitivity on fungal thermal evolution. The results have clear implications for predicting climate warming induced breaking points in animal-microbe mutualisms.