Modeling fine-scale abundance dynamics: a dual frequentist and Bayesian approach applied to common birds

Abstract: Understanding how animal populations grow and change at a local scale is key to studying ecosystems and supporting conservation efforts. In this study, we explore two complementary methods to analyze data from the French Common Birds Monitoring Program (STOC).

The first method uses a statistical model that views population changes as driven by births and deaths, influenced by how individuals interact with each other and by environmental factors like climate and landscape. One major challenge is that these interactions often depend on spatial location and that the data are partly random. To address this, we develop new ways to estimate birth and death patterns and use a mathematical theory (called stabilization) that assumes interactions mostly happen at a local scale. This helps ensure that our estimates are accurate and reliable when predicting changes in bird numbers at specific locations.

The second method uses a Bayesian spatio-temporal model, estimated with a technique called INLA (Integrated Nested Laplace Approximation). This model takes into account both space and time, helping us measure how environmental variables affect bird populations over time. It includes spatial patterns modeled through differential equations, time trends using autoregressive effects, and different responses depending on habitat type.

Mathematics

Audience: general audience

Gothenburg PhD seminar

Series comments: Rooms and times may vary, please check the latest update. In-person only.