Table 3.
Model selection results from generalized additive mixed modeling of Laysan and black-footed albatross habitat preference. Final models were arrived at by forward selection based on AIC using a subset of individuals, followed by cross-validation using the remaining individuals
| Laysan albatross | Model terms [% deviance explained by coefficient] | % deviance explained by model | |||||
|---|---|---|---|---|---|---|---|
| Incubation | s(SST) | s(dc) | s(depth) | SSHa | s(dTZCF) | EKE | 39.8 % |
| [60.6] | [30.6] | [5.34] | [1.79] | [0.90] | [0.75] | ||
| Brooding | dc | PP | 26.3 % | ||||
| [82.0] | [18.0] | ||||||
| Chick-rearing | s(SST) | s(dc) | s(depth) | s(PP) | s(dTZCF) | 39.2 % | |
| [45.7] | [45.1] | [4.08] | [2.64] | [2.52] | |||
| Black-footed albatross | Model terms [% deviance explained by coefficient] | % deviance explained by model | |||||
| Incubation | s(SST) | s(dc) | SSHa | 38.8 % | |||
| [75.9] | [21.9] | [2.18] | |||||
| Brooding | s(SST) | dc | 11.9 % | ||||
| [60.5] | [39.5] | ||||||
| Chick-rearing | s(SST) | s(dc) | s(dTZCF) | s(EKE) | s(PP) | depth | 28.3 % |
| [45.5] | [19.8] | [12.2] | [11.2] | [6.44] | [4.84] | ||
Cubic regression splines retained in the final models are represented by s( ) [84]. SST: sea surface temperature. dc: distance to the breeding colony. SSHa: sea surface height anomaly; dTZCF: distance to the Transition Zone Chlorophyll Front; EKE: eddy kinetic energy; PP: primary productivity