A pattern recognition (PATREC) habitat capability model for the ovenbird (Seiurus aurocapillus) was tested and refined on the Mark Twain National Forest (NF) in the Missouri Ozarks. Breeding bird surveys were conducted on 48 stands in 1983, 31 different stands in 1984, and 26 different stands in 1985. The literature-based PATREC model in use on the forest was used as an operational hypothesis to guide stand selection. Habitat data were obtained from the timber management and wildlife management information system data bases maintained by the Mark Twain NF. Stepwise (Wilks selection criterion) and direct multivariate discriminant analyses were used to determine which habitat variables best separated stands where ovenbirds were observed from those where no ovenbirds were observed. The best discriminant function for the 1983 data set included 5 habitat variables: forest type, stand condition, total basal area, percent overstory crown cover, and percent ground cover. The best discriminant functions for the 1984 data set and the pooled 1983-1984 data set included the same 5 habitat variables. These 5 variables and the survey data from the 79 stands in the pooled data set were used to develop a PATREC habitat capability model. In an empirical test of this PATREC model on the source data set, the model correctly classified 85% of the stands used by ovenbirds as ovenbird habitat. When applied to the 1985 data set in a separate field trial, the PATREC model correctly classified 94% of the stands used by ovenbirds as ovenbird habitat.
