Citizen science for predicting spatio-temporal patterns in seabird abundance during migration.

Area of Science:

• Ornithology
• Marine Ecology
• Conservation Biology

Background:

• Pelagic seabirds are difficult to track, making migration distribution studies challenging.
• Long-term data on seabird migration is crucial for understanding population dynamics and conservation needs.

Purpose of the Study:

• To conduct the first long-term study on the spatial distribution of migrating shearwaters using citizen science data.
• To apply machine learning to predict shearwater abundance and distribution patterns during migration.

Main Methods:

• Utilized daily abundance data (2005-2017) for Balearic shearwaters from Trektellen and eBird citizen science databases.
• Employed Random Forest regression models with 15 environmental predictors to forecast shearwater abundance.
• Developed separate models for pre-breeding and post-breeding migration periods.

Main Results:

• Models explained over 52% of the variation in shearwater abundance and predicted distributions with approximately 70% accuracy.
• Variable importance differed between pre-breeding and post-breeding migration models, indicating distinct environmental drivers.
• Demonstrated the successful application of citizen science data and machine learning for describing migratory spatio-temporal patterns.

Conclusions:

• Citizen science data combined with high-resolution satellite imagery offers an accurate method for monitoring seabird migration.
• Predictive modeling provides a cost-effective tool for long-term monitoring of sensitive marine species and identifying conservation areas.
• Modeling approaches are vital for detecting the impacts of climate change and global shifts on marine species.