Table 2.

Parameter estimates from the best-fit generalized linear model (binomial errors, log link) of winter deer persistence (number of months of deer presence January-March over 3 winters, 0–9), modelled against landscape features.

Parameter	β estimate	std. error	t-value	p-value
(Intercept)	−1.44	0.22	−6.58	<0.0001
Upland deciduous forest	0.02	0.01	3.75	0.0002
Forest harvest blocks	0.11	0.02	6.30	<0.0001
Petroleum well sites	0.14	0.08	1.79	0.0734
All trails (unpaved roads)	−1.50	0.33	−4.58	<0.0001
Industrial block features	0.57	0.09	6.02	<0.0001

*Null deviance = 295.9 on 60 df; residual deviance = 160.27 on 55 df; AIC = 270.35.