A Proof-of-Concept for Acoustic Monitoring of Breeding Phenology and Success in Cavity-Nesting Migrants

Understanding how migratory bird populations respond to environmental change requires linking migratory behaviour to reproductive outcomes. For long-distance migrants such as the Pied Flycatcher (Ficedula hypoleuca), timely arrival at breeding
grounds is critical for reproductive success, yet models predicting population dynamics lack sufficient empirical data on how breeding phenology and success vary between migration strategies.

This thesis presents a proof-of-concept for using passive acoustic monitoring to detect breeding behaviour and estimate breeding phenology and success metrics in cavity nesting migrants. Traditional methods, including colour ringing, tracking devices, and
in-person nest box monitoring, are limited by costs and labour demands. Audio recorders offer a minimally invasive and scalable alternative, though their application to nest box monitoring remains largely untested.

To address this, two complementary studies were conducted. First, the feasibility of detecting nest box behaviours was evaluated using audio recorders and custom-built classifiers in Kaleidoscope Pro, validated using in-field observations. Song was detected with higher accuracy compared to calls and visitations which required greater manual validation. Comparisons with BirdNET indicated that custom classifiers captured more detections but required additional development and processing time. Second, Bayesian changepoint models were applied to behavioural detections over time to assess whether shifts in frequency aligned with in-field estimates for breeding stage transitions and success metrics. Vocalisation-derived changepoints reflected arrival timing, while visitation detections aligned with onset of chick-begging and fledging. Behaviour-derived estimates were consistent across full-day and reduced recording schedules, highlighting the potential for reductions in processing time and data storage. While visitation rates showed limited correlation with brood size, binary nest success or failure may be inferred.

These findings demonstrate that audio recorders provide reliable estimates of breeding phenology and represent a scalable complement to traditional monitoring. By connecting behaviour to reproductive phenology, acoustic monitoring offers novel empirical data to parameterise migration models, advancing predictive capacity under environmental change.