Date of Award

Fall 12-2020

Document Type


Degree Name

Master of Science (MS)


Biological Sciences



Committee Director

Eric L. Walters

Committee Member

Holly Gaff

Committee Member

Jeffrey J. Buler

Committee Member

Matthias Leu


Autumn migration is a time when billions of birds move from breeding grounds in North America to wintering grounds in Central and South America, with many individuals relying on stopover habitats en route for resting and refueling purposes. These stopover sites are critical to the survival of the hundreds of species of migratory landbirds that migrate annually, and thus identifying important stopover sites is a high priority for conserving such taxa. The Delmarva Peninsula; a coastal region of Delaware, Maryland, and Virginia along the mid-Atlantic flyway; consists of forested habitats with ample food and shelter that likely serves as quality stopover sites for many species during autumn migration. Determining both extrinsic and intrinsic factors that most influence migrant use of forested stopover sites during this period is a necessary step towards providing adequate protection for vulnerable species, and one requiring a multi-scale analytical approach. I assessed the influence of variables at the regional- (i.e. proximity to the coast, location latitudinally), landscape- (i.e. proportions of surrounding land cover types), and patch-scales (i.e. habitat structure and vegetative characteristics) on migratory landbird use of forested stopover sites at 48 forested areas located across Delaware, Maryland, and Virginia during autumn migration in 2013 and 2014. Using boosted regression tree modelling techniques, I conducted analyses to determine variable influence on forested site use for 13 migratory species, as well as season-wide and early- vs. mid-season analyses using all nocturnal migratory landbird species. For season-wide analyses, autumn migration was separated into four 21-day sampling periods (period 1 = 15 Aug – 4 Sep, period 2 = 5 Sep – 15 Oct, period 3 = 26 Sep – 16 Oct, period 4 = 17 Oct – 7 Nov).

Predictor variables were not consistent in influence across multiple spatial and temporal scales during the migratory season. For all season-wide analyses, including the grouped model and thirteen individual species models, time of sampling (sampling period) was the most influential predictor variable in explaining migrant density. At the regional-scale, latitude was the most consistently influential predictor variable in explaining migrant density, generally showing higher densities at sites located further north. At the landscape-scale, proportion of hardwood forest, shrubland, impervious surface, and permanent water surrounding stopover sites were all influential at predicting migrant bird density, although their degrees of influence and relationship to migrant density (positive or negative) varied greatly across models. At the patch-scale, densities of invertebrate food resources and understory vegetation were influential predictor variables across migrant models. Early in the migratory season (15 Aug – 4 Sep), proportion of surrounding land cover (low impervious surface and high shrubland and hardwood forest) and metrics associated with patch-scale habitat structure (high ground vegetation and shrub counts) were the most influential predictor variables of migrant density. Alternatively, during the middle of the migratory season (26 Sep – 16 Oct), latitude and food availability were far more influential in predicting migrant use. These results demonstrate how spatially and temporally variable migrant use of forested stopover sites can be. Using a multi-scale approach, while logistically difficult, is necessary to understand the complexity of migrant use of stopover sites.