Table 1. A comparison of plausible models of the relationship between body temperature and running speed in isopods from four thermal treatments.

Treatment	Model	K	AIC	Δi	Relative Likelihood	wi
10°C	Beta	6	152	0	1.000	0.952
	Gaussian	4	274	122	3.221·10−27	3.069·10−27
	Quadratic	4	286	134	7.985·10−30	7.606·10−30
	Mod. Gaussian	5	237	85	3.487·10−19	3.322·10−19
	Exp. Mod. Gaussian	6	158	6	0.049	0.047
20°C	Beta	6	164	0	1.000	0.993
	Gaussian	4	255	91	1.736·10−20	1.725·10−20
	Quadratic	4	249	85	3.487·10−19	3.464·10−19
	Mod. Gaussian	5	210	46	1.026·10−10	1.019·10−10
	Exp. Mod. Gaussian	6	174	10	0.006	0.006
Stochastic	Beta	6	273	0	1.000	0.970
	Gaussian	4	346	73	1.407·10−16	1.366·10−16
	Quadratic	4	351	78	1.155·10−17	1.121·10−17
	Mod. Gaussian	5	317	44	2.790·10−10	2.708·10−10
	Exp. Mod. Gaussian	6	280	7	0.030	0.029
Predictable	Beta	6	183	0	1.000	0.993
	Gaussian	4	264	81	2.577·10−18	2.560·10−18
	Quadratic	4	261	78	1.155·10−17	1.147·10−17
	Mod. Gaussian	5	229	46	1.026·10−10	1.019·10−10
	Exp. Mod. Gaussian	6	193	10	0.006	0.006

For all treatments, the beta model provided the best fit to the data. For each model, we report not only the AIC but also the differential AIC (Δ_i), which is the difference between a given model's AIC and the lowest AIC. We also report the Akaike weight (w_i), which is the normalized likelihood that the model is the best one in the set.