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

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

I am currently developing a joint species distribution model with imperfect detection that incorporates residual species correlations as well as spatial autocorrelation. I am leveraging a spatial factor model framework to allow for computationally efficient modeling of large communities, as well as Nearest Neighbor Gaussian Processes for computational efficiency for large spatial data sets. I am applying this framework to a community of approximately 100 bird species across the continental US. I am implementing the framework in the spOccupancy R package. The development version of the package currently has implementations of the modeling framework, which will be included in the CRAN version sometime in spring 2022.

Avatar
Jeff Doser
Postdoctoral Research Associate