Table 2.
Best fitting parameterizations of the Conditional Arnason–Schwarz and POPAN Models
| Model | # of parameters | Deviance1 | AIC2 | ΔAIC3 | AIC weights4 |
|---|---|---|---|---|---|
| Conditional Arnason–Schwarz modela | |||||
| Φ(.)p(to+t)Ψ(f * to + t) | 21 | 487.1455 | 529.1455 | 0 | 0.4210 |
| Φ(.)p(t)Ψ(f * to + t) | 20 | 490.8493 | 530.8493 | 1.7038 | 0.1796 |
| Φ(f)p(to + t)Ψ(f * to + t) | 22 | 487.1309 | 531.1309 | 1.9854 | 0.1560 |
| Φ(f)p(t)Ψ(f * to + t) | 21 | 489.683 | 531.683 | 2.5375 | 0.1183 |
| Φ(t)p(t)Ψ(f * to + t) | 28 | 480.4241 | 532.4241 | 3.2786 | 0.0817 |
| Φ(.)p(to + t)Ψ(t) | 20 | 495.8178 | 535.8178 | 6.6723 | 0.0150 |
| Φ(t)p(to + t)Ψ(f * to + t) | 29 | 477.9171 | 535.9171 | 6.7716 | 0.0143 |
| Φ(.)p(to + t)Ψ(i) | 12 | 511.9563 | 535.9563 | 6.8108 | 0.0140 |
| Φ(f * t)p(to * t)Ψ(f * to * t) | 54 | 460.0165 | 546.0165 | 16.871 | 0 |
| Φ(.)p(.)Ψ(.) | 3 | 581.9903 | 587.9903 | 58.8448 | 0 |
| POPAN modelb | |||||
| Φ(.)p(t)pent(t) | 14 | − 1028.251 | 417.6716 | 0 | 0.9962 |
| Φ(t)p(t)pent(t) | 20 | − 1030.5198 | 428.8351 | 11.1635 | 0.0038 |
| Φ(t)p(.)pent(t) | 15 | − 1004.0417 | 444.0828 | 26.4112 | 0 |
| Φ(.)p(.)pent(t) | 9 | − 962.111 | 473.0152 | 55.3436 | 0 |
| Φ(.)p(.)pent(.) | 3 | 30,658.918 | 32,081.5399 | 31,663.8683 | 0 |
| Φ(t)p(t)pent(.) | 15 | 30,659.738 | 32,107.8919 | 31,690.1903 | 0 |
Italicized terms are AIC values indicating the best fit models
1Deviance is a measure for how well the model fits the data
2Akaike’s information criterion
3Change in AIC from the single best fit model
4AIC weight is the model probability within the candidate model set
aProbabilities estimated are survival (Φ), capture (p), and disease state change (Ψ); and the variables examined to influence Φ, p and Ψ were time in months (t), Bd state at previous capture (to), Bd state of capture (f) and no variable (.)
bProbabilities estimated are Φ, p, and probability of entry into the population (Pent); these were modeled as either dependent upon time (t) or no variable (.)