Development and simulation testing for a new approach to density dependence in species distribution models

Density dependence is included in many population-dynamics models, but few options exist within species distribution models (SDMs). One option for density-dependence in SDMs proceeds by including an independent time-series of population abundance as covariate using a spatially varying coefficient (SVC). We extend this via three alternative approaches that replace the independent time-series with information available within the SDM. We recommend the intermediate complexity approach that estimates a SVC responding to median abundance in each time; this SVC indicates whether a given location has a smaller- or greater-than-average sensitivity to changes in median abundance. We next develop a reaction-advection-diffusive simulation model, wherein individuals avoid habitats that exceed a threshold in local density. This movement model results in an estimated SVC that is negatively correlated with the average spatial distribution. Finally, we show that a SVC can be identified using bottom trawl data for four species in the eastern Bering Sea from 1982 to 2019. We conclude that the common basin-model for animal movement results in an ecological teleconnection, wherein population depletion or recovery at one locations will affect resulting dynamics at geographically distant habitats, and that this form of density dependence can be detected using SDMs.

Automated summary

This study demonstrates the importance of density dependence in species distribution models and proposes several new approaches to address density dependence. They find that the impact of species abundance on habitats is spatiotemporally varied, and this relationship can be detected using SDMs.