2026-10-14 https://api.researchdata.se/oai-pmh

oai:researchdata.se:doi-10-17044-scilifelab-31661575/0 2026-10-14 subject:ssif:3 principal:slug:uppsala-university

info:eu-repo/semantics/other http://purl.org/dc/dcmitype/Dataset https://doi.org/10.17044/SCILIFELAB.31661575 Data for the article Global biodiversity models exhibit low predictive accuracy on local scales https://orcid.org/0009-0008-0734-0936 Medical and Health Sciences Medicin och hälsovetenskap This data contains all the inputs to run the models used in the article Global biodiversity models exhibit low predictive accuracy on local scales (in press). These represent the outputs of the earlier data processing and feature engineering steps of the model pipeline. They can be found in the output folder. It also contains the outputs of all models analyzed in the article, in the folder saved_runs. Source code and instructions for how to run models and analyze outputs can be found in this GitHub repository (https://github.com/j-nystrom/biodiversity-impacts) . Below is an abstract describing the study: Amid the global biodiversity crisis, there is high demand for spatially explicit biodiversity indicators. Global models that quantify human impacts on biodiversity provide important insights for conservation, but systematic testing of predictive performance has been lacking. When evaluating alpha and beta diversity models using representative global data we find that, despite land-use impacts in line with previous research, there is a challenging gap between effect size inference and local prediction. The results show that mixed models with study attributes as random effects, common in biodiversity meta-analysis and indicators, exhibit low predictive accuracy, driven by reliance on highly averaged fixed effects. Ecologically structured models that replace study random effects with biome, realm, and taxonomic parameters, show improved but still modest results in sampled contexts. Further, performance when extending predictions to other contexts remains low, related to distribution shifts in environmental conditions and biodiversity responses. Our results highlight a tension between high-resolution predictor availability and the granularity at which biodiversity responses can be reliably estimated. While models are essential for informed conservation efforts, their applicability is fundamentally constrained by current data availability, which includes geographic and taxonomic biases. Countries with extensive data can build higher-fidelity national indicators, but accelerated and systematic data collection is needed to support data-poor regions with localized, actionable biodiversity insights. https://creativecommons.org/licenses/by/4.0/ Uppsala University Uppsala universitet 2026-04-01T00:00:00Z