Generating spatial predictions of species distribution is a central task for research and policy. Currently, correlative species distribution models (cSDMs) are among the most widely used tools for this purpose. However, cSDMs fundamental assumption of species distributions in equilibrium with their environment is rarely met in real data and limits their applicability for dynamic projections. Process-based, dynamic SDMs (dSDMs) promise to overcome these limitations as they explicitly represent transient dynamics and enhance spatio-temporal transferability. Software tools for implementing dSDMs become increasingly available, yet their parameter estimation can be complex.
Here, we test the feasibility of calibrating and validating a dSDM using long-term monitoring data of Swiss red kites (Milvus milvus). This population has shown strong increases in abundance and a progressive range expansion over the last decades, indicating a non-equilibrium situation. We construct an individual-based model with the RangeShiftR modelling platform and use Bayesian inference for model calibration. This allows the integration of heterogeneous data sources, such as parameter estimates from published literature as well as observational data from monitoring schemes, with coherent assessment of parameter uncertainty. Our monitoring data encompass counts of breeding pairs at 267 sites across Switzerland over 22 years. We validate our model using a spatial-block cross-validation scheme and assess predictive performance with a rank-correlation coefficient.
Our model showed very good predictive accuracy of spatial projections and represented well the observed population dynamics over the last two decades. Results suggest that reproductive success was a key factor driving the observed range expansion. According to our model, the Swiss red kite population fills large parts of its current range but has potential for further increases in density.
We demonstrate the practicality of data integration and validation for dSDMs using RangeShifteR. This approach can improve predictive performance compared to cSDMs. The workflow presented here can be adopted for any population for which some prior knowledge on demographic and dispersal parameters as well as spatio-temporal observations of abundance or presence/absence are available. The fitted model provides improved quantitative insights into the ecology of a species, which may greatly help conservation and management actions.
This submission uses novel code which is provided in an external repository. All data and code required to replicate the presented analyses are provided as private-for-peer review via a public GitHub repository under the following link: https://github.com/UP-macroecology/Malchow_IBMcalibration_2023
Upon acceptance, this repository will be archived and versioned with Zenodo and its DOI provided.
For this study, a tagged development version of the R package RangeShiftR was used that is available at: https://github.com/RangeShifter/...

10.1101/2022.09.26.509574