Roads constrain movement across behavioural processes in a partially migratory ungulate

Table 3 Summary of the full model showing model coefficients (β) with 95% confidence intervals, the standard deviation of the random effects, and the number of observations

Variable	Categories	β	95% C.I.	ΔAIC_REMOVED
log(Step length)	–	− 0.407	[− 0.425, − 0.389]	4697
Forest density	–	0.347	[0.324, 0.371]	1784
Forest density : Time of day	Day (Ref.)	0.347	[0.324, 0.371]	1738
	Night	− 0.787	[− 0.825, − 0.749]
	Twilight	− 0.423	[− 0.486, − 0.36]
Road crossings	–	− 1.379	[− 1.505, − 1.252]	1076
Road crossings : log(Step length)	–	0.231	[0.205, 0.257]	284
log(Step length) : Sex	Female (Ref.)	− 0.407	[− 0.425, − 0.389]	142
	Male	0.143	[0.119, 0.168]
Slope	–	0.107	[0.081, 0.133]	129
Slope²	–	− 0.076	[− 0.092, − 0.06]	98
Road crossings : Forest density	–	0.068	[0.046, 0.09]	44
Road crossings : Time of day	Day (Ref.)	− 1.379	[− 1.505, − 1.252]	29
	Night	0.110	[0.065, 0.156]
	Twilight	0.047	[− 0.034, 0.129]
Elevation difference²	–	− 0.014	[− 0.02, − 0.007]	28
Elevation difference	–	0.003	[− 0.015, 0.021]	26
NDVI	–	0.056	[0.005, 0.107]	24
Road crossings : Sex	Female (Ref.)	− 1.379	[− 1.505, − 1.252]	16
	Male	− 0.066	[− 0.111, − 0.02]
NDVI²	–	− 0.023	[− 0.047, 0.002]	11
Road crossings : Movement mode	Resident (Ref.)	− 1.379	[− 1.505, − 1.252]	7
	Disperser	0.010	[− 0.061, 0.081]
	Migrant	0.030	[− 0.023, 0.084]

Random effects	Standard deviation
Individual animals (N=79)	0.041
Populations (N=6)	0.020
Number of events: 26892 observed + 268920 random steps

“Ref.” indicates the category of reference for categorical variables. For each variable, we also report the ΔAIC_REMOVED, i.e., the difference in AIC between the full model and the full model without that variable. Hence, the variables with the highest ΔAIC_REMOVED are the most important because, when removed, they caused the highest increase in AIC. Variables are listed in order of importance. The AIC of the full model including all variables was 129,748.

ISSN: 2051-3933