Model selection with AIC.

For each model, the AIC value was computed using bias-adjustment for small sample sizes according to: AICc = n*ln(RSS/n)+2*K+(2*K*(K+1))/(n−K−1), where n is the number of data, RSS the residuals sum of squares and K the number of parameters (25). The plausibility of each model is assessed by its corresponding Akaike weight Wi which was obtained by normalizing the relative likelihoods exp (−0.5*Δi), with Δi the difference between the AICc of the model i and the lowest AICc. The plausibility of model 3 versus model 2 is given by Wi(model 3)/Wi(model 2) = 0.999/7.15 10−4 = 1400.