Long-term, not short-term, temperatures predict timing of egg-laying in European Starling (Sturnus vulgaris)

Temperature is thought to be an important supplemental cue used by female birds to time breeding. It is widely assumed that late spring temperatures, within ~1 month of egg-laying, are most predictive of laying date, consistent with the idea that temperature is used to fine-tune timing of breeding to year-specific local conditions at the time of laying. Here we show that a relatively broad, long-term, temperature window (Jan 2 – Apr 4, 92 days; r2 = 0.73) best predicted timing of egg-laying in European Starlings (Sturnus vulgaris). A “mid-winter” temperature window was also strongly correlated with laying date (r2 = 0.58) but there was no support for an influence of short-term temperatures immediately before egg-laying. We assessed the relationship between ambient temperature and timing of egg-laying using three complimentary approaches: a) an “unconstrained”, exploratory analysis, b) a traditional sliding window approach, and c) specific, biologically-informed temperature windows. Our results contrast markedly with the widely held view, in the avian breeding phenology literature, that immediate, pre-breeding, temperatures best predict short-term variation in laying because they allow birds to fine-tune timing of breeding to local conditions around the time of egg-laying. This means that mechanisms that allow integration of long-term temperature information must exist in birds – perhaps most parsimoniously involving direct effects of temperature on growth of the bird's ectothermic insect prey– even though these are currently poorly characterized.