Variance Inflation Factor (VIF) Analysis

The tables below show the results of variance inflation factor (VIF) analysis for three Bombus species: B. franklini, B. occidentalis, and B. vosnesenskii. The VIF values indicate the degree of multicollinearity among environmental variables, with values above 10 indicating problematic collinearity requiring variable removal.

All species show identical patterns in variable selection. Eleven variables were removed due to high multicollinearity (VIF > 10), including temperature extremes and most precipitation measures. Nine variables were consistently retained across all species, suggesting these represent a core set of non-redundant environmental predictors.

Among the retained variables, NLCD Land Cover Class shows the lowest VIF values (1.23-1.26) across all species. The highest variation occurs in Precipitation Seasonality (bio_15), with B. franklini showing the lowest value (6.104) compared to B. occidentalis (6.449) and B. vosnesenskii (6.556).

B. franklini generally exhibits lower VIF values across most retained variables, while B. occidentalis and B. vosnesenskii show nearly identical patterns. This consistency indicates robust variable selection for species distribution modeling, with minor differences likely reflecting species-specific habitat associations.