Development of Hierarchical Spatial Models for Assessing Ungulate Abundance and Habitat Relationships
1. Most studies investigating animal abundances and their environmental drivers rely on animal count data generated by interactions between ecological and spatial processes of interest. However, these counts are strongly affected by imperfect detection in
- PDF / 827,892 Bytes
- 48 Pages / 439.37 x 666.142 pts Page_size
- 40 Downloads / 186 Views
Development of Hierarchical Spatial Models for Assessing Ungulate Abundance and Habitat Relationships
Abstract
1. Most studies investigating animal abundances and their environmental drivers rely on animal count data generated by interactions between ecological and spatial processes of interest. However, these counts are strongly affected by imperfect detection inherent to the observation methods and processes. Reliable models of animal abundance patterns, which can simultaneously deal with all these important sources of variation, are rarely employed by investigators in practice. We address this key methodological need by developing a hierarchical model for animal-habitat relationships, which can rigorously investigate abundance patterns by explicitly parameterizing ecological, spatial and observational processes. 2. We use a hierarchical formulation with two model components: one describing the ecological processes that determine ungulate abundance and the other addressing the observation process involved in the field survey employing the distance sampling method. We used a Bayesian Poisson regression model to establish the effects of a set of habitat-related factors (predictor variables) on ungulate abundance (response variable). We included a Gaussian Conditional Autoregressive (CAR) prior to account for spatial interaction effects. The hierarchical model permitted specification of covariate effects on abundances at both local and landscape levels. A standard half-normal detection function was used to model the observation process during line transect field surveys. The observation model included 'cluster size' as an individual covariate additionally affecting the detectability of animal groups counted. We specified a zero-truncated Poisson distribution for modeling variations in cluster size. This model was implemented in the programming language R using the package NIMBLE. 3. We demonstrate the application of our hierarchical spatial model to describe the variation in the local abundance of chital deer (Axis axis) in the Nagarahole— Bandipur landscape, based on line transect data from counts of ungulate species
© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2021 N. S. Kumar et al., Spatial Dynamics and Ecology of Large Ungulate Populations in Tropical Forests of India, https://doi.org/10.1007/978-981-15-6934-0_2
35
36
2 Development of Hierarchical Spatial Models for Assessing Ungulate Abundance…
obtained by observers walking along 77 line transect samplers (spatial replicates) across a 1400 km2 study area. The count data were accumulated over six temporal replications under rigid field protocols. These counts and associated distance data were used to estimate chital abundance within each cell of a 1-km2 grid superimposed over the landscape. 4. Habitat features such as forest vegetation type, forage availability, distance to water sources, topography, anthropogenic disturbances and effectiveness of law enforcement were considered impor
Data Loading...
