Spatial Joint Species N-Mixture Models for Multi-Source Observational Data with Application to Wild Deer Population Abundance

TL;DR

This paper introduces a novel spatial joint species N-mixture model to accurately estimate wild deer populations by integrating multiple data sources with different spatial resolutions, accounting for species correlations and ecological factors.

Contribution

The study develops a new joint modelling approach for spatially misaligned multi-source data, specifically applied to wild deer populations, incorporating species correlations and ecological covariates.

Findings

Effective integration of diverse data sources improves population estimates.

Model captures correlations between deer species.

Enhanced understanding of ecological covariate impacts.

Abstract

Accurate predictions of the populations and spatial distributions of wild animal species is critical from a species management and conservation perspective. Culling is a measure taken for various reasons, including when overpopulation of a species is observed or suspected. Thus accurate estimates of population numbers are essential for specifying, monitoring, and evaluating the impact of such programmes. Population data for wild animals is generally collated from various sources and at differing spatial resolutions. Citizen science projects typically provide point referenced data, whereas site surveys, hunter reports, and official government data may be aggregated and released at a small area or regional level. Jointly modelling these data resources involves overcoming challenges of spatial misalignment. In this article, we develop an N mixture modelling methodology for joint modelling of species populations in the presence of spatially misaligned data, motivated by the three main species of wild deer in the Republic of Ireland; fallow, red and sika. Previous studies of deer populations investigated the distribution and abundance on a species by species basis, failing to account for possible correlation between individual species and the impact of ecological covariates on their distributions.