Table 4. β estimates from the logit link function based on best and univariate models for red panda occupancy probability (Psi).

Model	Intercept	${\widehat{\mathit{\beta }}}_{\mathit{A}\mathit{E}}\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{E}\mathit{L}\mathit{E}}\left]\right)$	${\widehat{\mathit{\beta }}}_{\mathit{N}\mathit{D}\mathit{V}\mathit{I}}\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{N}\mathit{D}\mathit{V}\mathit{I}}\left]\right)$	${\widehat{\mathit{\beta }}}_{\mathit{H}\mathit{A}\mathit{B}}\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{H}\mathit{A}\mathit{B}}\left]\right)$	${\widehat{\mathit{\beta }}}_{\mathit{B}\mathit{A}\mathit{M}}\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{B}\mathit{A}\mathit{M}}\left]\right)$	${\widehat{\mathit{\beta }}}_{\mathit{D}\mathit{N}\mathit{S}}\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{D}\mathit{N}\mathit{S}}\left]\right)$	${\widehat{\mathit{\beta }}}_{\mathit{D}\mathit{W}\mathit{S}}$ $\left(\mathit{S}\widehat{\mathit{E}}\right[{\widehat{\mathit{\beta }}}_{\mathit{D}\mathit{W}\mathit{S}}\left]\right)$
A priori relationship		-	+	+	+	+	-
Best Model (w = 0.81)	-0.46 (0.33)	1.22(0.30)			0.75(0.30)		0.57(0.27)
Univariate Model		0.98(0.25)	-0.37(0.19)	-0.45(0.47)	0.67(0.25)	0.68(0.31)	0.43(0.22)

Covariates considered: ELE: Average elevation; NDVI: Normalized differential vegetation index; HAB: Available habitat; BAM: Bamboo cover in each grid cell; DNS: Distance to nearest settlement; DWS: distance to nearest water sources. Bold indicates strong or robust impact, that is 95% confidence intervals as defined by $\widehat{\mathrm{\beta }}\pm 1.96\times \mathrm{SE}$ not overlapping at 0; Italics indicate opposite from a priori prediction. Figures in the parenthesis denotes standard error (SE).