This is a Preprint and has not been peer reviewed. The published version of this Preprint is available: https://doi.org/10.24072/pcjournal.337. This is version 3 of this Preprint.
You must log in to post a comment.
Inferring the strength of species interactions from demographic data is a challenging task. The Integrated Population Modelling (IPM) approach, bringing together population counts, capture-recapture, and individual-level fecundity data into a unified model framework, has been extended from single species to the community level. This allows to specify IPMs for multiple species with interactions specified as links between vital rates and stage-specific densities. However, there is no evaluation of such models when interactions are actually absent-while any interaction inference method runs the risk of producing false positives. We investigate here whether multispecies IPMs could output interactions where there are in fact none, building on an existing predator-prey IPM. We show that interspecific density-dependence estimates are centered on zero when simulated to be zero, and therefore their estimation is unbiased. Their coverage probability, quantifying how many times credible intervals include zero, is also satisfactory. We further confirm that adding random temporal variation to multispecies density-dependent link functions does not alter these results. This study therefore reaffirms the potential of multispecies IPMs to infer correctly how biotic interactions influence demography, although future studies should investigate model misspecifications.
Biostatistics, Dynamic Systems, Ecology and Evolutionary Biology, Population Biology, Statistical Methodology, Terrestrial and Aquatic Ecology
integrated population model, data assimilation, Species Interactions, predation, density-dependence
Published: 2022-12-15 13:36
Last Updated: 2023-12-11 06:15
Conflict of interest statement:
Data and Code Availability Statement:
All code and generated data and outputs are publicly available and deposited on OSF (https://doi.org/10.17605/osf.io/xfa6e), linked to a GitHub repository (https://github.com/MatthieuPaquet/multi_species)