The effects of uncertainty about age at transition on bias in the Mayfield family of estimators

Recent theoretical advances in nest survival estimation have demonstrated the importance of knowledge of nest age in improving nest survival estimates. However, data on the age of nests may be difficult to collect because nests may vary in developmental rate, handling eggs or nestlings may be too intrusive, or candling eggs and measuring nestlings may be infeasible due to nest location. We use a Mayfield Markov chain formulation to explore how bias in nest survival estimates arising from age-specific heterogeneity in transition probabilities can be managed through careful timing of nest visits. In general, the magnitude of bias is minimized when the dates of transitional events (hatching and fledging) are known exactly through daily monitoring. However, researchers can maintain similarly low levels of bias by concentrating nest visits around anticipated hatching and fledging dates, and allowing longer intervals between visits when no transition is imminent. This can reduce the number of required nest visits by about one third, depending on sample size. Accurate characterizations of the joint probability distributions of hatching and nest age, and fledging and nest age, can also reduce bias in daily survival estimates and are particularly important for overall success estimates.