Table 3. Model selection to assess the effects of age, mate replacements and diet on the productivity of breeding successful pairs.

Model definition	ΔAICc	AICcw	R2 GLMM(m)	R2 GLMM(c)
1. Age+Replacement	0.000 *	0.215	0.036	0.109
2. Age+Replacement+PSi+PSi2	0.301	0.185	0.107	0.206
3 Age+Replacement+H′	1.774	0.088	0.042	0.114
4. Age+Replacement+(OC+AR)	2.074	0.076	0.038	0.111
5. Age+Replacement+OC	2.129	0.074	0.037	0.105
6. Age+Replacement+PSi	2.149	0.073	0.037	0.108
7. Age+Replacement+AR	2.173	0.072	0.037	0.119
8. Age+Replacement+H′+H′2	3.331	0.041	0.058	0.118
9. Age+Replacement+(OC+AR)+H′	3.693	0.034	0.048	0.122
10. Age+Replacement+OC+H′	3.852	0.031	0.045	0.132
11. Age+Replacement+AR+H′	4.078	0.028	0.042	0.110
12. Age+Replacement+(OC+AR)+PSi	4.320	0.025	0.039	0.109
13. Age+Replacement+OC+PSi	4.362	0.024	0.038	0.104
14. Age+Replacement+AR+PSi	4.399	0.024	0.038	0.115
15. Age+Replacement+H′+PSi+(H′*PSi)	6.185	0.010	0.047	0.120

Model definition enumerates the fixed effects considered in the GLMMs. All models included territory and year nested by population as random effects.

*Best model AIC_c  = 174.899.

Note: OC = rabbit consumption; AR = partridge consumption; H′ = diet diversity; PS_i = prey consumption specificity. Parameters’ interactions are denoted by (*), while (²) indicates a quadratic effect. ΔAIC_c refers to the difference in the corrected Akaike Information Criteria (AIC_c) between model i and the model with the lowest (AIC_c) (i.e. the best model). Models with ΔAIC_c <2 are shown in bold type. The Akaike weights (AIC_cw) explains the probability that a given candidate model is the best of the proposed set, so the sum of all the models is 1.0. R² _GLMM(m) estimates model fit using fixed effects only, while R² _GLMM(c) estimates model fit including both fixed and random effects.