Table 1. Alternative models for Batrachochytrium dendrobatidis infection loads (in logarithm) in six different sites.

	AICc	Wi	Wt / W	model R2	Min T	Max T	abund	develop	%var sites
with temperature-2 days	181.82	0.444	264284	0.523	-0.780	0.082	-0.013	-0.037	0.053
with temperature-5 days	182.08	0.389	231832	0.522	-0.622	-0.079	-0.009	-0.039	0.052
with temperature-10 days	184.25	0.131	78112	0.506	-0.392	-0.289	-0.024	-0.029	0.050
with temperature-15 days	187.54	0.025	15085	0.480	-0.281	-0.378	-0.024	-0.024	0.050
with temperature-22 days	189.85	0.008	4766	0.462	-0.106	-0.529	-0.029	-0.006	0.049
with temperature-30 days	192.50	0.002	1268	0.440	-0.067	-0.546	-0.024	0.020	0.048
without temperature	206.79	0.000		0.242			-0.451	0.087	0.038
weighted averages				0.519	-0.648	-0.047	-0.013	-0.036	0.052

The first six models include the average minimum and maximum water temperature (T) in six different time spans prior to tadpole sampling (i.e., two, five, ten, 15, 22 and 30 days), taking into account other three predictor terms: tadpole abundance (abund), tadpole development (develop) and site. Sample size is 66 (six sites x 11 months). AICc: AIC corrected for small sample sizes. W_i: model weights. W_t / W: quotient of strength of evidence dividing the weight (W_t) each model containing both maximum and minimum temperatures with the model without temperatures (W). Figures below Min T, Max T, tadpole abundance and development are standardized regression coefficients (β) obtained in mixed general linear models (β values inform about the magnitude and sign of the partial relationships of the predictor variables). Weighted averages: multimodel inference of standardized β regression coefficients considering the model weights W_i. The AICc figure for the null model (i.e., not including any effect) is 208.04.