Using Integrated Species Distribution Models to Improve Range Size Metrics for IUCN Red List Assessments

Aim Estimates of a species' Area of Occupancy (AOO) and Extent of Occurrence (EOO) are key metrics for Red List assessments. Presence‐only species distribution models (SDMs) with arbitrary thresholding are commonly used for this task even when data could support more rigorous analyses. We evaluate the advantages of integrated SDMs (iSDMs) over presence‐only SDMs for estimating these metrics, focusing on threshold selection and uncertainty quantification. Location Patagonia, South America, including Islas Malvinas/Falkland Islands. Methods We modelled distributions for five raptor species ( Glaucidium nana , Strix rufipes , Buteo ventralis , Phalcoboenus australis , P. albogularis ), fitting iSDMs using a hierarchical Bayesian framework with spatial random effects in spOccupancy and comparing AOO and EOO estimates to those from Maxent presence‐background SDMs. To assess accuracy against a known truth, we then simulated data from an iSDM subject to detection and sampling error, comparing the true AOO (the metric most sensitive to thresholding) to estimates from iSDM and SDM under different thresholding rules. Results Empirically, both approaches predicted similar occupancy patterns ( r = 0.48–0.90), but AOO and EOO estimates diverged. Threshold choice in Maxent produced variation comparable to iSDM 90% credible intervals, with AOO and EOO ranging over thousands and hundreds of thousands of km 2 , respectively. Only one species showed overlapping EOO estimates between methods, and three overlapping AOO. The iSDM estimates suggested no species qualifies as threatened under Red List Criterion B, though Striated Caracara approached Near Threatened. In simulations, SDM estimates of AOO correlated poorly with the truth regardless of threshold ( r = 0.09), whereas iSDM estimates correlated strongly ( r = 0.78) with nominal coverage. Main Conclusions When detection/non‐detection data are available, iSDMs provide uncertainty quantification and avoid arbitrary thresholds, two critical advantages over presence‐only SDMs for estimating range‐size metrics. We recommend iSDMs for taxonomic groups with such data, rather than degrading data to fit presence‐only SDMs.