Because habitat characteristics may influence rates of brood parasitism by brown-headed cowbirds (Molothrus ater) and nest predation, we investigated the effects of nest-patch and habitat-patch characteristics on brood parasitism and predation of Bell's vireo (Vireo bellii) nests in grassland-shrub habitats in central Missouri, USA. We developed multivariate logistic regression models explaining parasitism and predation from variables measured at 142 and 130 nests, respective. We ranked models based on akaike's Information Criterion (AIC(y)) and reported models that differed by <1 AIC(y) unit from the minimum AIC(y) value as candidates for the best model. We identified 5 candidate models explaining parasitism that included 9-11 variables and R-2 = 0.358-0.386. Nest-patch width, territory shrub cover, distance to crop field, site, year, site x territory shrub cover, nest height x side concealment, patch width x length, and habitat type x patch length were in all models. We identified 5 candidate models explaining predation with 3-5 variables and R-2 = 0.615-0.199. Parasitism status, patch width, and patch length or patch length x site appeared in all models. Overall, our models provide support for hypotheses that increased nest concealment, amount of nest cover, and nest height affect on nest predation and parasitism. Nest-patch width was a factor in most models but in the opposite direction predicted by these hypotheses. Interactions among some variables also produced habitat and site-specific exceptions to predictions by these hypotheses. We hypothesize that the best overall method to increase nest success in existing grassland shrub habitats is to increase the density of large shrub patches.
