Table 3.
Most supported (∆AICc <2) occupancy models for each scale-dependent landscape variable, using the detection structure p(TIME + RH)
| Model | AICc | ∆AICc | wi | K | -2 L |
|---|---|---|---|---|---|
| Impervious cover | |||||
| ψ(250) | 79.48 | 0.0 | 0.79 | 5 | 66.48 |
| Tree cover | |||||
| ψ(500) | 90.25 | 0.0 | 0.32 | 5 | 77.26 |
| ψ(1500) | 91.16 | 0.90 | 0.20 | 5 | 78.16 |
| ψ(2000) | 91.35 | 1.10 | 0.19 | 5 | 78.36 |
| ψ(1000) | 91.60 | 1.35 | 0.16 | 5 | 78.60 |
| Road density | |||||
| ψ(500) | 88.21 | 0.0 | 0.43 | 5 | 75.22 |
| Number of ponds | |||||
| ψ(1500) | 85.04 | 0.0 | 0.75 | 5 | 72.04 |
| Wetland cover | |||||
| ψ(2000) | 90.83 | 0.0 | 0.65 | 5 | 77.82 |
AICc is a second-order Akaike’s information criterion, for small sample sizes; ∆AICc is the difference in AICc value from top-ranked model; wi is AICc model weight; K is the number of estimated parameters in the model; −2 L is twice the negative log-likelihood; ψ is probability of occupancy