Global biodiversity models exhibit low predictive accuracy on local scales.

Global alpha- and beta-diversity models show low predictive accuracy on local scales, even when land-use impacts match prior findings, revealing a gap between inferred effects and local forecasts.

• Why it matters: Conservation needs spatially explicit biodiversity indicators, but predictive performance of global human-impact models has not been systematically tested against local outcomes.
• What they did: Using representative global data, the study evaluated alpha- and beta-diversity models and found mixed models with study attributes as random effects perform poorly, while ecologically structured models using biome, realm, and taxonomic parameters improve only modestly.
• The result: When predictions are extended to new contexts, accuracy remains low and is linked to distribution shifts in environmental conditions and biodiversity responses, implying current data granularity limits reliable local indicators.

AI summary of the abstract · may contain errors · not medical advice · read the paper.