Inferring reproductive phenology and success from proportions of juveniles in population monitoring

1. Phenological shifts caused by climate change are increasingly documented in wild populations by the widespread collection of datasets on reproductive timing and success. These phenological events are often inferred by examining changes in population abundance and age structure throughout the breeding season. However, the quantitative relationship between the observed proportion of juveniles over time and the underlying distribution of breeding times (e.g., onset of reproduction) and average reproductive success is often not explicitly addressed. In addition, potential biases introduced by selection on reproductive phenology or by the sampling design can affect our inference of reproductive phenology and success. 2. In this study, taking the example of bird monitoring, we proposed an analytical model to relate the proportion of juveniles in counts (e.g., mist-net captures) to the distribution of fledging dates and mean reproductive success in the population. We then infer laying dates from fledging dates, accounting for putative selection through fertility and/or juvenile survival to fledging. We simulated varying levels of variance, selection strength, and sampling effort. 3. Our simulation results show that our approach is able to recover the true mean and variance of laying dates and the mean reproductive success under ideal conditions (large sampling effort, no selection). It notably corrects for the fact that the mean fledging time lags behind the inflection point in the proportion of sampled juveniles, all the more so as laying date variance and reproductive success are high. Selection for earlier breeding systematically biases the estimates of mean laying dates, but we show how this bias can be corrected if information on selection strength is available. Multi-site analyses reveal that low sampling effort and high within-site variation can introduce further biases, but also suggest that four sampling sessions with reasonable effort per year provides reasonable estimates. 4. These findings underscore the importance of explicitly modeling the population processes (including possibly selection), and carefully planning sampling designs, to improve the accuracy of phenological estimates from population monitorings, and better interpret climate-driven changes in wild populations.