Bennett, A. F., R. E. Lenski, and J. E. Mittler. 1992. Evolutionary adaptation to temperature. I. Fitness responses of Escherichia coli to changes in its thermal environment. Evolution 46:16-30.

Abstract:
We used bacteria to study experimentally the process of genetic adaptation to environmental temperature. Replicate lines of Escherichia coli, founded from a common ancestor, were propagated for 2,000 generations in 4 different thermal regimes as 4 experimental groups: constant 32, 37, or 42-degrees-C (thermal specialists), or a daily alternation between 32 and 42-degrees-C (32/42-degrees-C: thermal generalists). The ancestor had previously been propagated at 37-degrees-C for 2,000 generations. Adaptation of the groups to temperature was measured by improvement in fitness relative to the ancestor, as estimated by competition experiments. All four experimental groups showed improved relative fitness in their own thermal environment (direct response of fitness). However, rates of fitness improvement varied greatly among temperature groups. The 42-degrees-C group responded most rapidly and extensively, followed by the 32 and 32/42-degrees-C groups, whose fitness improvements were indistinguishable. The 37-degrees-C group, which experienced the ancestral temperature, had the slowest and least extensive fitness improvement. The correlated fitness responses of each group, again relative to the common ancestor, were measured over the entire experimental range of temperatures. No necessary tradeoff between direct and correlated responses of fitness was apparent: for example, the improved fitness of the 42-degrees-C group at 42-degrees-C was not accompanied by a loss of fitness at 37-degrees-C or 32-degrees-C. However, the direct fitness responses were usually greater than the correlated responses, judged both by comparing direct and correlated responses of a single group at different temperatures and by comparing direct and correlated responses of different groups at a single temperature. These comparisons indicate that the observed adaptation was, in fact, largely temperature specific. Also, the fitness responses of the generalist group across a range of temperatures were less variable than those of the thermal specialist groups considered as whole.