Skip to main content
. 2016 Sep 21;118(1):21–30. doi: 10.1038/hdy.2016.77

Table 2. Fits of models to empirical data on changing colour frequencies (Figure 3) and comparisons between models.

Colour Model log likelihood (2LL) estimates Model comparisons
  Migration Directional selection Heterozyg. advantage Frequency dependent M vs DS M vs HA M vs FD DS vs HA DS vs FD HA vs FD
Parameters/test 2 2 3 3 ΔAIC χ12 χ12 χ12 χ12 ΔAIC
White-1 104.3 117.7 87.8 85.6 −13.4 16.4*** 18.6*** 29.8*** 32.0*** 2.2
White-2 75.6 70.0 70.0 69.6 5.5 5.5† 6.0* 0.0 0.5 0.5
White-3 73.7 72.2 72.2 71.4 1.5 1.5 2.3 0.0 0.9 0.9
White-4 53.9 58.5 52.4 51.7 −4.7 1.5 2.2 6.2* 6.8** 0.6
White combined 307.3 318.4 282.4 278.3 −11.1 24.9*** 29.1*** 36.0*** 40.1*** 4.1
Banded-1 99.5 98.8 98.8 98.4 0.7 0.7 1.1 0.0 0.4 0.4
Banded-2 80.7 80.7 80.7 78.9 0.1 0.1 1.9 0.0 1.8 1.8
Banded-3 86.9 87.3 86.3 86.1 −0.4 0.6 0.8 1.0 1.2 0.2
Banded combined 267.2 266.8 265.7 263.3 0.4 1.4 3.9 1.0 3.4 2.4
Red 111.6 116.8 108.0 108.6 −5.2 3.6† 3.0† 8.9** 8.2** −0.6

Abbreviations: AIC, Akaike information criterion; DS, directional selection; FD, frequency-dependent selection; HA, heterozygote advantage; Heterozyg., heterozygote; M, migration.

Negative values indicate that the latter of the two models in the comparison had the higher likelihood value. In χ2 tests; d.f.=1 for single population model comparisons, d.f.=4 for White combined and d.f.=3 for banded combined. Bold log-likelihood estimates indicate the best models.

P<0.1, *P<0.05, **P<0.01, ***P<0.001 (after sequential Bonferroni correction for repeated tests of white and banded skerries).