EcoSta 2023: Start Registration
View Submission - EcoSta2023
A0585
Title: Bayesian multi-species N-mixture models for large scale spatial data in community ecology Authors:  Michele Peruzzi - University of Michigan (United States) [presenting]
Abstract: Community ecologists seek to model the local abundance of multiple animal species while considering that observed counts only represent a portion of the underlying population size. Analogously, modelling spatial correlations in species' latent abundances is essential when attempting to explain how species compete for scarce resources. A Bayesian multi-species N-mixture model with spatial latent effects to address both issues is developed. On the one hand, the model accounts for imperfect detection by modelling local abundance via a Poisson log-linear model. Conditional on the local abundance, the observed counts have a binomial distribution. On the other hand, a directed acyclic graph restricts spatial dependence is let to speed up computations and recently developed gradient-based Markov-chain Monte Carlo methods are used to sample a posteriori in the multivariate non-Gaussian data scenarios in which it is interested. The model is illustrated on synthetic data and data from the North American Breeding Bird Survey.