1. Evolutionary theory predicts that the capacity to acclimate should be favoured in variable environments. However, perfect compensation for thermal variation is rare and the capacity for thermal acclimation can vary considerably between individuals within natural populations. This variation may be explained by costs associated with acclimation, but it is not clear what these costs are. 2. We tested the hypothesis that oxidative stress is a cost of acclimation that could explain the variation between individuals in acclimation capacity. We acclimated individual mosquitofish (Gambusia holbrooki, n = 416) to 18oC and 28oC sequentially, and determined swimming performance at each temperature to evaluate their acclimation capacity. Fish were then acclimated to either cold (18oC) or warm (28oC) conditions, and we increased antioxidant capacities of a subset of fish experimentally by administering N-acetyl cysteine (NAC). We measured H2O2 production, catalase antioxidant activities, and oxidative damage to proteins and membranes. 3. We show that there is significant variation in acclimation capacity between individuals, and that there is a trade-off between acclimation capacity and swimming performance in warm conditions. Mean swimming performance across both acclimation temperatures increased with increasing acclimation capacity, but the increase was small biologically. Hence, greater thermal plasticity (high acclimation capacity) resulted in only minor performance benefits across acclimation conditions. We verified this rather surprising result in replicate populations grown in outdoor mesocosms. 4. ROS production, antioxidant activities, and oxidative damage were higher in cold-acclimated fish, and particularly in those fish with low capacity for acclimation. However, experimentally increasing antioxidant capacities with NAC alleviated these changes to oxidative status, suggesting a causal relationship. Hence, oxidative stress may be a cost that constrains the capacity for acclimation. Together, the performance trade-off and oxidative cost indicate that phenotypic plasticity is not always advantageous in variable environments, and instead bet-hedging may be a more beneficial strategy, particularly in short-lived species.

Data were collected in a laboratory experiment, and from a mesocosm experiment. Details are given in the main manuscript and in the Explanatory Notes tab of the Excel sheet providing the data.

Raw data and details of treatments are provided in the data Excel sheet, as well as Explanatory notes regarding data collection.