Assay conditions in laboratory experiments: is the use of constant rather than fluctuating temperatures justified when investigating temperature-induced plasticity?

Temperature is an important selective agent in nature. Consequently, temperature-induced plasticity which may help buffering detrimental effects of temperature variation has received considerable attention over recent decades. Laboratory studies have almost exclusively used constant temperatures, while in nature, temperature typically shows pronounced daily fluctuations. Using a factorial design with constant versus fluctuating temperatures and a higher versus a lower mean temperature, we here investigate in the butterfly Lycaena tityrus whether the use of constant temperatures is justified. Fluctuating compared to constant temperatures caused shorter development times, increased heat but decreased cold stress resistance, decreased heat-shock protein expression, and increased immunocompetence. Thus, overall, fluctuating temperatures were more beneficial to the butterflies compared to constant ones. However, despite substantial variation across temperature regimes, the ranking of trait values among treatments remained largely unaffected (e.g. lower constant as well as fluctuating temperatures caused increased pupal mass). Thus, we tentatively conclude that there is no general reason for concern about using constant temperatures in studies investigating phenotypic plasticity, which seem to comprise a fair proxy. However, substantial differences in mean values as well as interactive effects suggest that one needs to be cautious. We further demonstrate negative effects of high temperatures on butterfly immune function, which seem to result from a trade-off between the latter and the heat shock response.