Table 2. Model selection for the tawny owl roadkill data based upon Akaike information criterion (Mod #: Model number; Variables in the model: variables included in each model; df: degrees of freedom; ΔAICc: AICc differences; Model prob (w_I): model probabilities; Evid ratio: evidence ratios of each model; Adj R²: adjusted R² of each model; VIF: variance inflation factor of each model; The evidence ratio provides a measure of how better is each model relatively to the null model (model 0); see Table 1 for variable codes).

Mod #	Variables in the model	df	ΔAICc	Model probies (wi)	Evid ratio	Adj R2	VIF
3	ROAD+HABITAT+HQ5	5	0	0.37	4.11	0.281	1.45
5	ROAD+HABITAT+HQ100	5	1.92	0.14	1.55	0.263	1.41
4	ROAD+HABITAT+HQ10	5	1.93	0.14	1.55	0.263	1.41
0	ROAD+HABITAT	4	2.90	0.09	1.00	0.241	1.35
2	ROAD+HABITAT+HQ2	5	3.81	0.06	0.67	0.24	1.38
9	ROAD+HABITAT+F10	5	4.32	0.04	0.44	0.239	1.37
1	ROAD+HABITAT+HQ1	5	4.43	0.04	0.44	0.24	1.37
10	ROAD+HABITAT+F100	5	4.58	0.04	0.44	0.236	1.36
8	ROAD+HABITAT+F5	5	5.16	0.03	0.33	0.230	1.35
7	ROAD+HABITAT+F2	5	5.18	0.03	0.33	0.230	1.35
6	ROAD+HABITAT+F1	5	5.20	0.03	0.33	0.230	1.35