TABLE 2.

Summary of discrimination accuracy (AUC, TSS, and Kappa) and functional accuracy metrics calculated for all EWM occurrence SDMs with and without a spatial autocovariate. Numbers with a downward arrow subscript indicate drop in discrimination accuracy and functional accuracy estimates after including spatial autocovariate. Functional accuracy shown here is calculated as the Pearson correlation coefficient (r _all and r _nonzero) and Spearman correlation coefficient (⍴ _all and ⍴ _nonzero) of EWM abundance–suitability relationship (see Methods for details)

Model type	Discrimination accuracy												Functional accuracy
	Nonindependent			Quasi‐independent			Spatially blocked			Combined average			Abundance–suitability correlation
	AUC	TSS	Kappa	AUC	TSS	Kappa	AUC	TSS	Kappa	Avg. AUC	Avg. TSS	Avg. Kappa	rall	rnonzero	⍴all	⍴nonzero
Presence–absence	0.83	0.56	0.54	0.82	0.44	0.43	0.76	0.43	0.44	0.80	0.48	0.47	0.58	0.39	0.68	0.40
Presence–absence + Autocovariate	0.88	0.65	0.63	0.90	0.72	0.70	0.86	0.54	0.53	0.88	0.69	0.62	0.59	0.40	0.69	0.43
Presence‐only (random pseudoabsence)	0.88	0.66	0.65	0.87	0.63	0.62	0.86	0.54	0.54	0.87	0.61	0.60	0.47	0.28	0.62	0.26
Presence‐only (random pseudoabsence) + Autocovariate	0.91	0.69	0.67	0.86↓	0.58↓	0.57↓	0.89	0.59	0.62	0.89	0.62	0.62	0.51	0.36	0.65	0.33
Presence‐only (distant pseudoabsence)	0.98	0.87	0.86	0.87	0.61	0.59	0.91	0.73	0.73	0.92	0.74	0.73	0.40	0.26	0.61	0.20
Presence‐only (distant pseudoabsence) + Autocovariate	0.99	0.99	0.99	1	1	1	0.99	0.98	0.99	0.99	0.99	0.99	0.26↓	0.24↓	0.62	0.21
Presence‐only (proximal pseudoabsence)	0.76	0.43	0.43	0.86	0.62	0.61	0.64	0.32	0.35	0.75	0.45	0.46	0.48	0.25	0.64	0.30
Presence‐only (proximal pseudoabsence) + Autocovariate	0.79	0.48	0.47	0.81↓	0.53↓	0.52↓	0.76	0.34	0.36	0.79	0.45	0.45↓	0.51	0.36	0.65	0.41

The final estimates of average discrimination accuracy measures and correlation derived functional accuracy measures are shown in bold.