Table 2.

Results derived from LMMs (GLMMs for Floral Richness) based on the best supported models (i.e., AICc < 2; Supplementary Table 1) for each response variable. The direction of significant effects for Floral Density were positive. For consistency across models F-values are provided, however, Flower Richness was fitted using a GLMM and thus probability values are derived from model comparison using the Likelihood Ratio Test. Numerator and ranges for denominator degrees of freedom (which vary in mixed models for different response measures as they are based on estimated variance components) are also provided. Shading indicates fixed effects were not tested for a specific response variable. NS denotes that a specific fixed effect was included in confidence set but not found to be significant. Fixed effect parameter estimates are provided in Table S4 for flowers and S5 for pollinators.

Taxa	Measure of community structure	Year (1, 11–93)	Month (3, 36–37)	Region (1, 10–37)	Legume treatment (9, 84–108	Floral density (1, 73–93)	Month * Legume treatment (9, 108)	Region * Month (3, 18)
Flower	Abundance	5.16*	14.86***	5.18*	31.46***		16.80***
	Abundance		12.32***	6.97*	31.46***		16.80***
	Richness		NS	NS	7.19***
	Richness		NS		7.19***
	Richness				7.19***
	Richness			NS	7.19***
	Richness		NS	NS	7.19***			NS
Bumblebee	Abundance	12.31***			2.77**	43.42***
	Richness	13.05***			2.97**	37.67***
Hoverfly	Abundance			9.19*	10.57***
	Richness			7.93*	8.50***
	Richness				8.57***