Table 3. Explanatory capabilities of previous year rainfall and population size for variance in births and deaths as indicated by model selection procedures.
| Males | Births | Deaths | ||||||
| Candidate model | AIC i | ∆ i | L i | w i | AIC i | ∆ i | L i | w i |
| Rainfall + Population size | -45.07 | - | 1.00 | 0.76 | -46.10 | 1.95 | 0.38 | 0.19 |
| Rainfall | -42.36 | 2.71 | 0.26 | 0.20 | -48.05 | - | 1.00 | 0.51 |
| Population size | -39.43 | 5.63 | 0.06 | 0.05 | -46.93 | 1.12 | 0.57 | 0.29 |
| Females | Births | Deaths | ||||||
| Candidate model | AIC i | ∆ i | L i | w i | AIC i | ∆ i | L i | w i |
| Rainfall + Population size | -48.91 | - | 1.00 | 0.90 | -44.90 | 1.93 | 0.38 | 0.22 |
| Rainfall | -44.22 | 4.69 | 0.10 | 0.09 | -44.88 | 1.95 | 0.38 | 0.21 |
| Population size | -40.62 | 8.27 | 0.02 | 0.01 | -46.84 | - | 1.00 | 0.57 |
AIC i is the Akaike Information Criterion value for candidate model i. For each variable, we considered three candidate models that included combinations of rainfall and population size. ∆ i represent the difference in AIC i with the minimum AIC i noted for candidate models. Models have a high likelihood (L i) and weight of evidence (w i) when ∆ i>2. We highlight in bold candidate models with greatest weight of evidence.