Northern Prairie Wildlife Research Center
A mallard (Anas platyrhynchos) management model (Johnson et al. 1987, Cowardin et al. 1988, Mack 1991) is available to assist managers with planning. The model is a mathematical synthesis of past mallard research in the prairies. Basically, it allows users to input descriptions about how different habitat conditions in a landscape might change with management and then predicts mallard recruitment and the expected proportional population change. Intermediate predictions pertaining to various aspects of mallard reproduction also are computed. Model performance has been examined (Johnson et al. 1986, Cowardin et al. 1988), and results suggest that predictions are reasonable. However, past evaluations lack rigor because of the expense associated with the needed long-term studies (Cowardin et al. 1988:24). There is suspicion that model performance is poor in parts of Minnesota, particularly in transition areas where woodland habitats become more common (D. Johnson, U.S. Geol. Survey, pers. commun.).
Assessing performance of the mallard model in Minnesota has become more important as the number and scope of PPJV projects increase. Johnson et al. (1986) acknowledged that true validation of the model was impossible because no methods exist to count actual numbers of mallards produced from a landscape. However, they suggested that studies could be conducted allowing comparison of field data with intermediate parameters predicted by the model, and we used this approach in our study. We assessed the mallard model's performance in southwest Minnesota by studying the nesting biology of mallards and then modeling mallard productivity in our study area. Specifically, we estimated summer hen mortality, the conditional hen mortality rate (i.e., probability that a hen would be killed if her nest was destroyed), habitat-specific nest occurrence, number of nests per hen, habitat-specific nest survival, hatch rate (i.e., number of hatched nests per initiated nest), hen success (i.e., proportion of hens that hatched a nest), average clutch size, and number of nests hatched by resident hens. We then compared these estimates to model inputs and predictions.