Table 5. ANOVA using burst speed as dependent variable, and temperature, mesocosm, predator cues and Roundup as categorical predictors, with mesocosm nested within the interaction of predator cues and Roundup, for Hyla versicolor.

	SS	d.f.	MS	F	p
Temperature	0.891	5	0.178	32.17	<0.001
Predator	0.106	1	0.106	19.16	<0.001
Roundup	0.002	1	0.002	0.38	0.537
Predator*Roundup	0.070	1	0.070	12.65	<0.001
Mesocosm (Predator*Roundup)	0.127	12	0.010	1.92	0.03
Predator*Temperature	0.023	5	0.005	0.83	0.528
Roundup*Temperature	0.017	5	0.003	0.62	0.683
Predator*Roundup*Temperature	0.009	5	0.002	0.33	0.903
Error	3.085	546	0.006

Univariate tests of significance for burst speed. We used Sigma-restricted parameterization and Type III sum of squares.