I studied the influence of local- and landscape-level habitat factors on breeding-bird communities in a prairie landscape in southwestern Oklahoma, U.S.A. Local habitat characteristics were sampled at study plot transects. Landscape variables within 100 m × 1500 m belt transects were computed from digitized aerial photography. Multiscale ordination revealed that landscape characteristics were most pronounced at a spatial scale of 8 ha (i.e., 100 m × 800 m). Grassland bird species were associated with large homogeneous habitat patches, while woodland species occurred in small heterogeneous patches. Bird species categorized on the basis of migratory status showed considerable variation along gradients reflecting the degree of heterogeneity and fragmentation. Mean patch size was an important factor in discriminating among groups of bird species classified on the basis of habitat, nesting height, and nest type.

Habitat regression models using local and landscape habitat components explained 79% of the variation in bird species richness and 41% of the variation in abundance. Bird species richness was best explained by landscape habitat variables, while overall bird abundance was related to a combination of local and landscape variables. Abundance of woodland and grassland/savanna species was accounted for by local habitat variables representing vegetation and a slope gradient. Models for typical woodland species--Tufted Titmice (Parus bicolor), Red-eyed Vireos (Vireo olivaceus), Great Crested Flycatchers (Myiarchus crinitus), and Northern Cardinals (Cardinalis cardinalis)--were explained primarily by local vegetation characteristics (r\sp2=0.52−.74). A combination of local and landscape habitat features accounted for the variation (local, 42-100%; landscape, 0-60%) in abundance of four typical grassland/savanna species--the Field Sparrow (Spizella pusilla), Eastern Meadowlark (Sturnella magna), Grasshopper Sparrow (Ammodramus savannarum), and Dickcissel (Spiza americana). In general, avian-community indices were related both to landscape and local habitat characteristics. Ecological groups of birds and individual species were accounted for primarily by local habitat characteristics.

I evaluated the degree of fragmentation of habitats used by grassland and savanna bird species in a prairie landscape. Habitat fragmentation was estimated using perimeter-area fractal dimension computed at 13 spatial scales ranging from 10 to 1,000 ha. Habitats occupied by grassland/savanna birds were significantly different from randomly-chosen sites at spatial scales of 50 and 75 ha. Mourning Doves, Dickcissels, Field Sparrows, and Eastern Meadowlarks occurred in habitats that exhibited a higher level of fragmentation than random sites. Natural fragmentation of the grasslands by woodlands associated with intermittent streams has a strong influence on the distribution of grassland/savanna birds. (Abstract shortened by UMI.)