RIASSUNTO
Seasonal patterns in immunity are frequently observed in vertebrates but are poorly understood. Here, we focused on a natural piscine model, the three-spined stickleback (Gasterosteus aculeatus), and asked how seasonal immune allocation is driven by physical variables (time, light, and heat). Using functionally-relevant gene expression metrics as a reporter of seasonal immune allocation, we synchronously sampled fish monthly from the wild (two habitats), and from semi-natural outdoors mesocosms (stocked from one of the wild habitats). This was repeated across two annual cycles, with continuous within-habitat monitoring of environmental temperature and implementing a manipulation of temperature in the mesocosms. We also conducted a long-term laboratory experiment, subjecting acclimated wild fish to natural and accelerated (×2) photoperiodic change at 7 and 15°C. The laboratory experiment demonstrated that immune allocation was independent of photoperiod and only a very modest effect, at most, was controlled by a tentative endogenous circannual rhythm. On the other hand, experimentally-determined thermal effects were able to quantitatively predict much of the summer–winter fluctuation observed in the field and mesocosms. Importantly, however, temperature was insufficient to fully predict, and occasionally was a poor predictor of, natural patterns. Thermal effects can thus be overridden by other (unidentified) natural environmental variation and do not take the form of an unavoidable constraint due to cold-blooded physiology. This is consistent with a context-dependent strategic control of immunity in response to temperature variation, and points to the existence of temperature-sensitive regulatory circuits that might be conserved in other vertebrates.