Table 2.

Proportional Overlap of Expansion of Cropland between 1992 and 2015 and Deviations from the Global Null Model of Cropland Extent in 1992 (Experimental Procedures; Null Model)

	Expansion of Cropland 1992–2015
	0.5% Crop	10% Crop	50% Crop
Positive Deviance 2SD	1.35	3.05	0.03
Positive Deviance 1SD	1.53	2.58	0.03
Negative Deviance 2SD	0.41	0.39	2.41
Negative Deviance 1SD	0.34	0.43	3.40

The predictor variables in the Null Model are the same as in Model 1 (Table 1), but the Null Model uses all data globally; Model 1 uses balanced samples of 500 presences and absences (Experimental Procedures). “Positive Deviance” and “Negative Deviance” refer to 2 or 1 or more positive (or negative) standard deviations (2SD and 1SD) from the Null Model. Ratios >1 indicate overrepresentation; ratios <1 indicate underrepresentation (more or less overlap of cropland and a given deviation threshold than would be expected if both are equally common across the land area they cover³⁹).