| Accuracy assessment of models is often overlooked, however it is critical when models are designed to assess wildlife conservation issues. The most rigorous form of accuracy assessment is performed by comparing the model(s) to an independent dataset. Agreement between independent fish distribution surveys and predicted distributions was generally better than chance (i.e., positive Kappa) at all scales assessed (drainages, stream sizes, valley segments, sites, and species), however agreement was variable with correct classification rates ranging from 0--100%. The GAP Analysis Program fish distribution models may have uses in their current form, but revisions are recommended before using the models for conservation planning. A common species modeled by the GAP Analysis distribution models was the white sucker Catostomus commersonii. White sucker population characteristics including length frequency, growth, condition, and precision of age estimates between scales, pectoral fin rays, and otoliths varied among six populations in the upper Missouri River Basin. |
