Table 1.

Results from global step selection function models. We present the parameter estimates (β) ± standard errors (SE) and P values of covariates included in step selection function (SSF) models explaining the crossings and detours of migratory caribou from the Rivière-aux-Feuilles herd in the vicinity of the largest lakes in Northern Québec, Canada. The least parameterized model, which is also the most parsimonious (lowest AIC_c value), appears in black, whereas the more complex model with interactions (and a higher AIC_c value) appears in grey. Number of parameters (k), log-likelihood (LL), and AIC_c values are shown for each model, and validation results ($\overline{r_S}$) appear for the most parsimonious model only

Model	Covariates	β ± SE	P value
Model 1, k = 2, LL = −3809.01, AICc = 7622.03, $\overline{r_S}$ = 0.52
	Proportion of steps in water	−0.733 ± 0.363	0.04
	Proportion of steps on ice	0.475 ± 0.179	0.01
Model 2, k = 5, LL = −3808.28, AICc = 7628.57
	Proportion of steps in water	−0.641 ± 0.469	0.17
	Proportion of steps on ice	0.581 ± 0.343	0.09
	Prop. water × Type:reservoir	−0.270 ± 0.786	0.73
	Prop. water × Type:river	0.663 ± 1.863	0.72
	Prop. ice × Type:reservoir	−0.087 ± 0.407	0.83
	Prop. ice × Type:river	−0.923 ± 0.872	0.29

Statistically significant effects (P ≤ 0.05) are shown in bold