| This thesis examined the ability of landscape-level descriptors of habitat to explain the observed distribution of indicated breeding pairs (IBPs) of American Black Ducks (Anas rubripes) in the Canadian Maritime Provinces of New Brunswick and Nova Scotia, and whether these relationships could then be used to produce spatially explicit waterfowl population estimates. Models were also developed to explain trends in Black Duck IBPs during the period 1990--99.;These were landscape-level models with an extent of 125,000 km 2 and grain sizes of 25 km2 and 100 km2. Data were obtained from the Canadian Soil Information Survey, Maritime Wetlands Inventory, ecological land classifications, provincial soil surveys, digital elevation models and topographical maps. Data on the chemical limnology of surface waters in survey plots were obtained from over 1,000 water samples collected during 1996--98, and from Environment Canada databases.;Variation in chemical limnology due to survey plot location was orders of magnitude greater than variation due to aquatic habitat type, although both factors were significant. Chemical limnology of surface waters was correlated to soil chemistry and other landscape-level attributes. The nature and spatial scale of wetland data made ecological land classification and soil data minor components of models.;The best predictive model for Black Duck IBPs included eight variables and had R2 = 0.88. Black Duck IBPs were positively associated with maximum Golet Score, total number of lakes, total area of salt marsh, total phosphorous concentration of surface waters, stream density, and total number of buildings, and negatively associated with standard deviation of elevation and total length of roads. The Golet Score is a measure of a wetland's wildlife habitat value for a broad diversity of species. A univariate model based solely on Golet Score had R2 = 0.78. Data from plots surveyed during 1986--89, confirmed the importance of Golet Score as a predictor of Black Duck IBP distribution.;Increasing Black Duck populations in survey plots during the period 1990--99 were related to increasing stream density, decreasing variation in elevation, and hypothetically, greater numbers of beaver ponds. Logistic regression models were also developed to predict the presence of breeding Ring-necked Ducks (Aythya collaris), Common Mergansers (Mergus merganser ), and American Green-winged Teal (Anas crecca carolinensis ) in survey plots. |
