TABLE 2.

Spatial capture–recapture model selection from analysis of conservative and lenient detection histories for the reintroduced river otter population in the Upper Rio Grande dendritic network (2018)

Model	K a	AIC b	AIC c c	ΔAIC c d	logLik e	Deviance f	w g
Conservative
D~1 λ 0~Sex σ~1	5	527.03	529.53	0.00	–258.17	516.34	0.31
D~1 λ 0~bk + Sex σ~1	6	526.92	530.56	1.04	–257.14	514.28	0.18
D~Lat λ 0~Sex σ~1	6	527.48	531.13	1.60	–257.48	514.96	0.14
D~1 λ 0~bk × Sex σ~1	7	527.57	531.23	1.69	–257.79	515.58	0.10
Lenient
D~1 λ 0~Sex σ~1	5	545.16	549.02	0.00	–265.58	531.16	0.19
D~1 λ 0~bk × Sex σ~1	7	547.56	549.49	0.47	–268.78	537.56	0.15
D~1 λ 0~1 σ~Sex	5	548.18	550.11	1.09	–269.09	538.18	0.11
D~1 λ 0~1 σ~1	4	549.47	550.72	1.70	–270.73	541.46	0.08

Primary model parameters were population density (D), baseline detection rate (λ ₀), and the spatial scale of detection (σ). Models were fit in which D followed a homogenous Poisson point process (1) or spatially varied as a log‐linear function of latitude (Lat). Models also considered a latrine‐specific behavioral response (bk) that was either shared between sexes (1) or was sex‐specific (Sex), via both additive (+) and interaction (×) effects, and considered σ that was either sex‐specific (Sex) or shared between sexes (1). For brevity, only competing models (≤2 ΔAIC_c) are presented; Tables B1 and B2 in Appendix S2 provide the complete model selection for both detection histories.

^a Number of model parameters.

^b Akaike's information criterion.

^c AIC corrected for small sample size.

^d Difference between AIC_c of model and AIC_c of top‐ranked model.

^e Log‐likelihood.

^f −2 × log‐likelihood.

^g Model weight.