Role of covariates in determining detection probability of leopard sign (Pt) on 2‐km‐long replicates of east Chure

Model	AIC	ΔAIC	w	Model Likelihood	K
$\widehat{\mathrm{\Psi }}$(·),p(R)	249.03	0	0.3824	1	18
$\widehat{\mathrm{\Psi }}$ (·),p(R+N)	249.99	0.96	0.2366	0.6188	19
$\widehat{\mathrm{\Psi }}$ (·),p(R+L)	249.99	0.96	0.2366	0.6188	19
$\widehat{\mathrm{\Psi }}$ (·),p(R+Samp_Eff)	251	1.97	0.1428	0.3734	19
$\widehat{\mathrm{\Psi }}$ (·),p(·)	262.17	13.14	0.0005	0.0014	17
$\widehat{\mathrm{\Psi }}$ (·),p(L)	262.27	13.24	0.0005	0.0013	18
$\widehat{\mathrm{\Psi }}$ (·),p(N)	263.75	14.72	0.0002	0.0006	18
$\widehat{\mathrm{\Psi }}$ (·),p(Samp_Eff)	263.96	14.93	0.0002	0.0006	18

$\widehat{\mathrm{\Psi }}$: model‐averaged leopard occupancy; p = replicate‐level detectability; AIC = Akaike's information criterion, ΔAIC = difference in AIC value between the top model and the focal model; w = AIC weight; Model likelihood is −2 logarithm of the likelihood function evaluated at maximum; k = number of model parameters; Covariates: R = terrain ruggedness averaged across each grid; N = nondifferent vegetative index averaged across each grid; L = livestock presence; Samp_Eff=sampling effort; + = covariates modeled additively; (·) = parameters are held constant. β‐coefficient estimates for R from the top detection model = 1.123 (SE 0.3).