Site selection and nest success of ring-necked pheasants as a function of location in Iowa landscapes

Wildlife managers in the midwestern United States implicitly recognize that large-scale changes in land use have been a major factor in the nesting ecology of the ring-necked pheasant (Phasianus colchicus), but they have lacked models that quantify the relation between nest success and landscape variables at multiple scales. We used data from 288 nests of radiomarked female pheasants during the 1990-94 breeding seasons in Iowa to study nest-site selection and nest success. We quantified habitat and landscape metrics within the 485-m radius of a home range around nests by using aerial imagery and FRAGSTATS. We screened potential landscape variables by using principal component and classification and regression tree (CART) analyses before developing logistic regression models to predict nest-site selection and success as a function of landscape conditions. A 5-variable logistic regression model incorporating nesting patch size, mean grassland patch size, landscape core area, landscape shape index, and distance to edge predicted nest-site locations at a 77% posterior concordant rate. The CART analyses suggested nest success was best modeled by splitting nest observations into nests in patches ≤15.6 ha and >15.6 ha. For nests in patches ≤15.6 ha, a logistic regression model with site cover type and mean core area index predicted nest success with a 73% posterior concordant rate (P 15.6 ha, a model with site cover type, core area standard deviation, and distance from the nest to the edge predicted nest success best (concordant rate = 64%, P = 0.069), although the effect of distance to edge was very slight (conditional odds ratio = 1.003, 95% CI = 0.997-1.009). Our models suggest managers should strive to provide undisturbed grassland blocks ≥15 ha for nesting pheasants, but our observation was that success was highest in fields 4 times that size. Furthermore, cover in several large grassland blocks within the nesting home range is preferable to concentrating cover in 1 large block. Landscape models like ours could be used to project the consequences of changes in agricultural policy on ring-necked pheasant populations.