Events

Date: 2 October 2025   Time: 14:00 - 15:00

Location: Hybrid: School of Mathematical Sciences, MB503 or via the Teams link

Abstract: Ecological surveys often track individuals from wildlife populations using methods such as traps or nets, aiming to estimate population size and to monitor processes such as migration patterns, behavioural change, or life state transitions. These processes are typically complex, latent, and vary over time and between individuals. Over the last decade, we have developed a suite of flexible Bayesian nonparametric models tailored to such data, building on the Dirichlet process (DP), the Hierarchical DP, the Dependent DP, and Pólya trees. These models account for complex observation mechanisms and enable inference on the latent ecological processes of interest. This talk will present models for several common survey types—including capture-recapture, count, and batch-mark surveys—and demonstrate how the models describe complex latent patterns and corresponding observation processes, solving open inference problems in the field.