Table 2.
Predictor combinations in statistical modelling with more than 5% support in Akaike weights
| Measure of diversity /Factor combinations | d.f. | SSres. | % explained variance | AIC | Akaike weight | |
|---|---|---|---|---|---|---|
| HE | N = 80 | |||||
| bar+exp | 2 | 1.676 | 92.5 | -76.22 | 0.56 | |
| bio+exp | 2 | 1.685 | 92.4 | -75.79 | 0.24 | |
| bio+exp+marg | 3 | 1.654 | 92.6 | -75.28 | 0.09 | |
| bar+exp+lim | 3 | 1.660 | 92.6 | -74.99 | 0.05 | |
| A | N = 80 | |||||
| bio+exp | 2 | 115.06 | 90.1 | 262.10 | 0.88 | |
| bio+exp+marg | 3 | 113.87 | 90.2 | 340.09 | 0.09 | |
| Hmt | N = 66 | |||||
| bar+exp+lim | 3 | 273.23 | 79.9 | 289.06 | 0.95 | |
| s | N = 80 | |||||
| marg+exp+lim+size | 5 | 4.06 | 49.0 | 122.1 | 0.97 |
All possible predictor combinations were tested with Generalised Linear Modelling and the support of all models by the data inferred by the Akaike Information Criterion (AIC). Akaike weights were calculated on the basis of all models tested. HE = expected heterozygosity, A average number of rarefied alleles per locus, Hmt number of rarefied mitochondrial haplotypes, s estimate of population selfing rate.