Table 1.

Generalized linear mixed models selected based on AICc for the different response variables (models with ΔAICc ≤ 2 were considered to be equivalent). * Means significant effect.

Response variable	Model	Predictor variable	Random	Error distribution	Link function	χ 2	df	P
Plant visitation rate	1	Pollinator	Plant ID/census ID	Gamma	Log	201.09	3	<0.001*
		Year				8.03	1	0.005*
		Pollinator × year				34.40	3	<0.001*
	2	Pollinator	Plant ID/census ID	Gamma	Log	201.05	3	<0.001*
		Year				8.10	1	0.004*
		Plant sex				0.13	1	0.722
		Pollinator × year				34.37	3	<0.001*
Seed set	1	Treatment	Plant ID	Binomial	Logit	1670.9	2	<0.001*
	2	Treatment	Plant ID	Binomial	Logit	1670.88	2	<0.001*
		Plant width				1.28	1	0.258
	3	Treatment	Plant ID	Binomial	Logit	1670.89	2	<0.001*
		Plant height				0.63	1	0.426
Seed weight	1	Treatment	Plant ID	Gaussian	Identity	5.10	2	0.080
Germination %	1	Treatment	Plant ID	Binomial	Logit	3.25	2	0.197
		Seed weight				28.95	1	<0.001*
		Treatment × seed weight				12.65	2	0.002*
	2	Treatment	Plant ID	Binomial	Logit	3.01	2	0.222
		Seed weight				31.47	1	<0.001*
		Plant width				2.55	1	0.111
		Treatment × seed weight				12.01	2	0.003*
	3	Treatment	Plant ID	Binomial	Logit	3.20	2	0.202
		Seed weight				29.30	1	<0.001*
		Plant height				0.39	1	0.533
		Treatment × seed weight				12.60	2	0.002*