Table 2. Stress-induced changes in biodiversity effects.

	Dominance effect				Trait-dependent complementarity effect				Trait-independent complementarity effect
	DF	t value	Estimate (s.e.)	P value	DF	t value	Estimate (s.e.)	P value	DF	t value	Estimate (s.e.)	P value
Model 1
Intercept	91	0.94	0.0014 (0.002)	0.35	92	−0.91	−0.0008 (0.0009)	0.36	93	−2.078	−0.00127 (0.0006)	0.04
LDiv	29	−1.47	−0.0037 (0.003)	0.15
Day	91	−2.54	−0.0002 (0.0001)	0.013	92	2.07	0.0001 (0.0001)	0.04
LDiv × Day	91	4.28	0.0005 (0.0001)	<0.0001
Model 2
Intercept	29	2.40	0.022 (0.009)	0.02	29	0.94	0.011 (0.01)	0.35	29	3.46	0.044 (0.01)	0.002
Tol	27	−2.41	−0.097 (0.04)	0.02	27	−3.02	−0.164 (0.05)	0.006	29	3.60	0.005 (0.001)	0.001
Inter	27	2.34	0.003 (0.001)	0.03	27	0.94	0.001 (0.001)	0.36
Tol × Inter	27	−2.48	−0.012 (0.005)	0.02	27	−3.06	−0.020 (0.006)	0.005

Model 1: Mixed effects model estimating the effect of log₁₀ diversity (LDiv) on stress-induced changes in biodiversity effects over the course of the experiment (Day). Model 2: Mixed effects model estimating the effect of the weighted mean species stress tolerance (Tol) and per capita strength of interspecific interactions (Inter) on changes in biodiversity effects at day 21 and 28. Means were weighted for the relative species abundance. Tolerance was calculated as the ratio between the species monoculture yield at 250 μg l⁻¹ atrazine and in control conditions. The strength of interspecific interactions was based on the parameter estimates of the community model (see the ‘Methods' section). s.e. is the standard error of the estimated fixed effects.