Wetlands of the Prairie Pothole Region of North America provide habitat for over 50% of the continent's breeding waterfowl, but most of the region's wetlands have been lost or degraded through intensive agricultural development. Despite widespread wetland losses in much of the Canadian prairies, there is little information about trends in degradation of remaining wetlands. Using habitat data collected for ~10,500 wetlands across the Canadian prairies during annual waterfowl surveys, 1985-2005, I employed multistate models in Program MARK to estimate rates of impact and recovery of wetlands resulting from agricultural activities. Then, I characterized the incidence of agricultural degradation to these wetlands. Rates of impact to wetland margins (natural vegetation around flooded basins) declined over time, likely due to a decreasing percentage of unaffected wetlands; recovery rates for margins were always lower than impact rates, suggesting increased cumulative degradation of wetlands over time. Unlike margins, impact and recovery rates for basins fluctuated with spring pond densities. Shallow ephemeral wetlands located in agricultural fields had the highest impact and lowest recovery rates. Multistate modeling could also be used to estimate rates associated with other landscape processes.<p><p> My second objective was to determine whether physical characteristics of prairie Canada wetlands could be used to predict breeding duck abundance. First, I sought to determine how pre-existing models developed in the Dakotas (USA) performed when predicting breeding duck abundances on Canadian prairie wetlands. I related duck pair abundance to pond area, and then compared observed to predicted duck abundance. The Dakota models performed reasonably well in predicting numbers of blue-winged teal (<i>Anas discors</i>), gadwall (<i>A. strepera</i>), and northern pintail (<i>A. acuta</i>), but predicted fewer mallards (<i>A. platyrhynchos</i>) and northern shovelers (<i>A. clypeata</i>) than were observed on wetlands. Pond area was an important predictor of duck abundance in all models, but results were less biased and more consistent in models developed specifically for Canadian wetlands. Spatiotemporal variation in the relationship of breeding duck abundance and wetland characteristics was also affected by regional duck and pond densities. Overall, the new applications and models developed and validated in this study will be useful for wetland and waterfowl management in the Canadian prairies.

Wetlands of the Prairie Pothole Region of North America provide habitat for over 50% of the continent's breeding waterfowl, but most of the region's wetlands have been lost or degraded through intensive agricultural development. Despite widespread wetland losses in much of the Canadian prairies, there is little information about trends in degradation of remaining wetlands. Using habitat data collected for ~10,500 wetlands across the Canadian prairies during annual waterfowl surveys, 1985-2005, I employed multistate models in Program MARK to estimate rates of impact and recovery of wetlands resulting from agricultural activities. Then, I characterized the incidence of agricultural degradation to these wetlands. Rates of impact to wetland margins (natural vegetation around flooded basins) declined over time, likely due to a decreasing percentage of unaffected wetlands; recovery rates for margins were always lower than impact rates, suggesting increased cumulative degradation of wetlands over time. Unlike margins, impact and recovery rates for basins fluctuated with spring pond densities. Shallow ephemeral wetlands located in agricultural fields had the highest impact and lowest recovery rates. Multistate modeling could also be used to estimate rates associated with other landscape processes.<p><p> My second objective was to determine whether physical characteristics of prairie Canada wetlands could be used to predict breeding duck abundance. First, I sought to determine how pre-existing models developed in the Dakotas (USA) performed when predicting breeding duck abundances on Canadian prairie wetlands. I related duck pair abundance to pond area, and then compared observed to predicted duck abundance. The Dakota models performed reasonably well in predicting numbers of blue-winged teal (<i>Anas discors</i>), gadwall (<i>A. strepera</i>), and northern pintail (<i>A. acuta</i>), but predicted fewer mallards (<i>A. platyrhynchos</i>) and northern shovelers (<i>A. clypeata</i>) than were observed on wetlands. Pond area was an important predictor of duck abundance in all models, but results were less biased and more consistent in models developed specifically for Canadian wetlands. Spatiotemporal variation in the relationship of breeding duck abundance and wetland characteristics was also affected by regional duck and pond densities. Overall, the new applications and models developed and validated in this study will be useful for wetland and waterfowl management in the Canadian prairies.