In particular, memory of previous best models does not guide future searching.
The procedure adopted to estimate the best model does not provide significantly different value of t on earthquakes above M 5.5 and M 6.0 (about 900 and 300 events, respectively).
The addition of Dfmin or body size to the best model did not result in a better supported model (Table 1, comparing model 2 with model 5: LRT χ21 = 1.82, P = 0.18; comparing model 3 with model 2: LRT χ21 = 2.27, P = 0.13).
In contrast, the best model did require different drift rate values.
† The best model did not include NND, but the second best, which had a close AICc value (difference of 0.71), included it.
Overall, however, the best model did not contain any hierarchical random effects (Model 3 DIC: 1281.52), implying that DUP was not associated with neighbourhood-level characteristics.
As the "best" model does not imply that it gives an adequate fit to the data, we need to assess the goodness-of-fit.
If the best model does not provide a good fit, or if it has parameters with very large confidence intervals, the lot can be augmented to include additional models with more or fewer parameters, respectively.
We find that the best model does not include linear additive genetic effects, and hence that the inherited part of breeding season success probably is very close to zero.
§, ¥ The best model included NS, but the second best, which had a close AICc value (difference of 0.32 and 0.29, did not. £ The best model did not include NS, but the second best, which had a close AICc value (difference of 0.66), included it.
On other hand, the hypothesis that human monkeypox occurrences depend primarily on climatic variables is not supported by our data for Sankuru, to the extent that our best model did not include either precipitation or temperature as predictors of human monkeypox cases.
