https://orcid.org/0000-0002-8950-9895
-
Views
-
Cite
Cite
Jeffrey W Doser, Krishna Pacifici, Discussion on “Continuous-space occupancy models” by Wilson J. Wright and Mevin B. Hooten, Biometrics, Volume 81, Issue 2, June 2025, ujaf056, https://doi.org/10.1093/biomtc/ujaf056
Close - Share Icon Share
Extract
1 INTRODUCTION
Congratulations to Wilson J. Wright and Mevin B. Hooten for this insightful contribution and thank you to the Biometrics editors for the opportunity to discuss this paper. Spatial occupancy models are an increasingly common framework used to model species distributions while accounting for false negatives in data collection and residual spatial autocorrelation in the ecological process. Spatial autocorrelation is typically accommodated within an occupancy modeling framework through the use of discrete conditionally autoregressive terms (Johnson et al., 2013) or with continuous spatial processes (Doser et al., 2022) despite the observed data being collected within areal units. Wright and Hooten argue that such misalignment between the observed data and modeling of spatial structure in the ecological process can result in inferior inferences regarding the proportion of area occupied by a species of interest. The authors propose an elegant solution to this problem based on a clipped Gaussian process (De Oliveira, 2000) and change of support methods (Cressie, 1996) that they implement using an efficient Markov chain Monte Carlo (MCMC) algorithm.
