Skip to main content

Table 1 Summary of model predictive skill statistics (R2, AUC, TSS) for blue whale and elephant habitat models, each model type, and each pseudo-absence generation technique. Biological realism was assessed using the predictions at simulated absences and true presences, with visual realism assessed by the full suite of authors based on skill within the Southern California Bight (blue whales) and Etosha salt pan (elephants). Figure panel is also included for Fig. 4 (blue whales) and 5 (elephants) to aid cross-referencing. The best performing model using 100% test and training is shown in red with the worst shown in blue. For R2, AUC, TSS, and Predictions at presences, high values indicate better performance. For Predictions at pseudo-absence, values closer to 0 indicate better performance. Bold values are the top 4 performing models in each category, with blue backgrounds representing the best performing in that category and red representing worse

From: Where did they not go? Considerations for generating pseudo-absences for telemetry-based habitat models