Table 1. Summary of results for 8 stable isotope mixing models explaining variation in diet for 64 wolves in British Columbia.
| Dirichlet | CLR | |||||||||
| Model | Region | Pack | Individual | Residual | DIC | Region | Pack | Individual | Residual | DIC |
| 1 | N | N | N | N | 1142.30 | N | N | N | N | 1342.74 |
| 2 | N | N | N | Y | 586.20 | N | N | N | Y | 585.05 |
| 3 | Y | N | N | N | 692.22 | Y | N | N | N | 693.10 |
| 4 | Y | N | N | Y | 512.40 | Y | N | N | Y | 512.12 |
| 5 | Y | Y | N | N | 501.16 | Y | Y | N | N | 502.21 |
| 6 | N | N | Y | N | 347.78 | N | N | Y | N | 334.57 |
| 7 | Y | N | Y | N | 338.91 | Y | N | Y | N | 332.51 |
| 8 | Y | Y | Y | N | NA | Y | Y | Y | N | 325.56 |
Models may include variation among regions, packs (social groups), individuals, or residual error. The Deviance Information Criterion is used to evaluated data support, with smaller values signaling stronger support from the data. Two approaches to dealing with compositional data (Dirichlet mixtures or CLR transformed data) yielded similar results (NAs represent models with convergence issues).