Joint species distribution models with imperfect detection for high‐dimensional spatial data

Abstract

Determining the spatial distributions of species and communities is a key task in ecology and conservation efforts. Joint species distribution models are a fundamental tool in community ecology that use multi‐species detection–nondetection data to estimate species distributions and biodiversity metrics. The analysis of such data is complicated by residual correlations between species, imperfect detection, and spatial autocorrelation. While many methods exist to accommodate each of these complexities, there are few examples in the literature that address and explore all three complexities simultaneously. Here we developed a spatial factor multi‐species occupancy model to explicitly account for species correlations, imperfect detection, and spatial autocorrelation. The proposed model uses a spatial factor dimension reduction approach and Nearest Neighbor Gaussian Processes to ensure computational efficiency for data sets with both a large number of species (e.g., >100) and spatial locations (e.g., 100,000). We compared the proposed model performance to five alternative models, each addressing a subset of the three complexities. We implemented the proposed and alternative models in the spOccupancy software, designed to facilitate application via an accessible, well documented, and open‐source R package. Using simulations, we found that ignoring the three complexities when present leads to inferior model predictive performance, and the impacts of failing to account for one or more complexities will depend on the objectives of a given study. Using a case study on 98 bird species across the continental US, the spatial factor multi‐species occupancy model had the highest predictive performance among the alternative models. Our proposed framework, together with its implementation in spOccupancy, serves as a user‐friendly tool to understand spatial variation in species distributions and biodiversity while addressing common complexities in multi‐species detection–nondetection data.

Implications of the scale of detection for inferring co‐occurrence patterns from paired camera traps and acoustic recorders

“…We used JSDMs to quantify species–habitat associations and co‐occurrence patterns between wild ungulates and disturbance factors. Specifically, we used the spOccupancy package 0.5.2 (Doser, Finley, Kery, et al., 2022) for the R statistical programing language (R Core Team, 2022) to fit a multispecies occupancy model that accounted for the effects of environmental covariates, residual species correlations, imperfect detection, and spatial autocorrelation (Doser, Finley, & Banerjee, 2022). Species‐specific occurrence probabilities ($$\psi $$) were modeled as a logit‐linear combination of site‐specific predictors and latent variables with corresponding species‐specific factor loadings.…”

“…These latent variables could represent missing predictors, and factor loadings represented species‐specific responses to these latent variables (Tobler et al., 2019; Warton et al., 2015). The correlation between species‐specific factor loadings can be used to identify species that frequently (or infrequently) co‐occur due to factors not explained by the site‐level predictors included in the model (Doser, Finley, & Banerjee, 2022). Imperfect detection was accounted for by using an observational model with additional vegetation covariates thought to be associated with detection probabilities (Doser, Finley, Kery, et al., 2022; MacKenzie et al., 2002) due to interference with sound propagation and visibility of camera traps.…”

“…This filtering resulted in a total of 312 surveyed days. We then created 3‐dimensional, daily detection–nondetection histories, where the dimensions represented the number of species, number of sites, and number of survey replicates (Doser, Finley, & Banerjee, 2022). The cells in the array were assigned a value of 1 when a species was detected and 0 when it was not detected.…”

Assembly processes inferred from eDNA surveys of a pond metacommunity are consistent with known species ecologies

“…More technically, eDNA sampling makes it more feasible to collect multiple sample replicates, which would allow combining a JSDM with a detection model to account for observation error (Diana et al 2022;Doser et al 2023;Guillera-Arroita et al 2017;Hartig et al 2024;Tobler et al 2019). Also, only two species in our dataset showed strong signals of dispersal limitation (Figure 2), but this low number could be because near-neighbour ponds were not sampled in our dataset, removing the possibility of detecting fine-scale spatial autocorrelation and thereby possibly reducing the relative importance of dispersal that would support source-sink relations among closely adjacent ponds.…”

Scenario planning and multispecies occupancy models reveal positive avian responses to restoration of afforested woodlands

“…Multispecies occupancy modeling is a flexible approach that estimates both community‐ and species‐level relationships to environmental variables across multiple spatial scales (Dorazio & Royle 2005; Devarajan et al 2020). Recent extensions of multispecies occupancy models to accommodate species interactions and spatial autocorrelation may provide improved insights on occurrence patterns of diverse communities across local to continental scales (Tobler et al 2019; Doser et al 2023). For predictor variable inputs into these models, new remote sensing datasets such as the Rangeland Analysis Platform (RAP) can provide historical ranges of variation in environmental variables (e.g.…”