Models of Production Rates in American Black Duck Populations

Abstract

Understanding the relationship between the annual reproductive success and changes in environment is important for appropriate waterfowl management. We developed predictive models of American black duck (Anas rubripes) production rates as a function of biotic (black duck and mallard [A. platyrhynchos] abundance) and abiotic factors (spring precipitation and temperature) across predefined breeding areas, from 1990 to 2001. We used male age ratios in the fall population, estimated from wing samples of harvested black ducks corrected for differential vulnerability via band-recoveries, as the index to annual reproduction. Information criteria suggested that a model containing predictors for density-dependence, competition with mallards, spring precipitation, and temperature and stratum-specific coefficients was the best model of black duck production rates. However, coefficients of this model were highly imprecise, leading to relatively poor predictive ability, possibly due to multicollinearity among predictors and the relatively short time span of analysis. We fit several models that included only black duck and mallard abundance as predictors; of these, models with constant slopes and stratum-specific intercepts performed best. Model-averaged parameter estimates supported inverse relationships between black duck and mallard abundance and age ratios, with stronger relative effects for black duck density-dependence. Both effects have implications for adaptive harvest management, in that harvest potential for black ducks may differ greatly depending on combinations of number of each species and the relative belief in alternative hypotheses about the impacts of mallards. Much variability in age ratios remained unexplained by our models, some possibly due to the lack of habitat explanatory variables but also apparently due to random factors. Model improvement could be achieved by incorporating recent developments in the modeling of random effects, especially via Markov Chain Monte Carlo methods. More research is also needed to incorporate recently acquired habitat predictors into predictive modeling for black ducks and other ducks breeding in eastern North America. These results provide critical input for models of adaptive harvest management, currently under consideration as an approach for developing an international (Canada–U.S.) harvest strategy for black ducks.