Filling Monitoring Gaps for Data-deficient Species Using Annual Occupancy Predictions from Co-occurrence Models

Hyun Yong Chung , Dae Kyung Lee, John Losey

Fragmented surveys and limited monitoring have excluded most invertebrates from conservation policy. We present a fill-in framework that uses species distribution models (SDMs) to reconstruct missing annual trends—not to extrapolate trends, but to fill them in. Instead of filtering data-sparse regions or years or relying on static environmental variables, we used co-occurrence patterns (COP) as variables, to capture year-to-year assemblage shifts. COP variables enabled annual prediction at all recorded sites from multisource, presence-only, sparse data. When applied to four rare native ladybugs across North America (2007–2021), COP models exceeded reliability benchmarks (Accuracy ≥ 0.70, AUC ≥ 0.70, Kappa ≥ 0.40, Brier ≤ 0.25) across standard 7:3 splits, cross-source and cross-period validations. Annual predictions were robust to temporal biases from variation in data volume and source composition. Multiple regression indicated negligible effects of those biases on reconstructed trends. Predicted decadal declines (9–31%) closely aligned with independent regional long-term monitoring, operationalizing IUCN Red List classification (from least concern to vulnerable) in the absence of standardized monitoring. By converting fragmented observations—primarily from citizen science—into reliable annual trend estimates, the fill-in approach extends extinction-risk assessment to data-deficient taxa long excluded from conservation frameworks.