Northern Prairie Wildlife Research Center
MINNESOTA WILDLIFE REPORT NO. 12
Minnesota Department of Natural Resources
Section of Wildlife
Wildlife Populations and Research Unit
St. Paul, Minnesota
A stochastic simulation model synthesizing past mallard (Anas platyrhynchos) research is available to assist management planning. From 1990-92, we studied nesting mallards at 3 sites to assess model performance in southwest Minnesota. We compared field estimates of mallard parameters to various model predictions for the study sites. Summer hen mortality, probability of a hen being killed if her nest was destroyed, habitat-specific nest occurrence, habitat-specific nest survival rates, nesting attempts per hen, average clutch size, hatch rate, hen success, and number of hatched nests were estimated. We executed 2 models for comparison with field estimates. The first (default model) used previously suggested Minnesota inputs, and the second (customized model) used inputs modified to more accurately describe habitat conditions during the study.
Predictions of important endpoints (i.e., nests per hen, hen success, and successful nests) from both models exceeded our field estimates. Summer hen mortality was underpredicted with both models. Differences were substantial and could inflate predictions of recruitment and population growth. Customized model predictions deviated slightly less from field estimates than did default model predictions.
True model assessment should be predicated on a single alternative that predictions depart significantly from true values due to an incomplete or misspecified model. Two other alternatives existed: 1) one or more model inputs could have been inappropriate or 2) field estimates of true values could have been biased. Certain field estimates caused us to question appropriateness of key input values. Additionally, unequal detection of nests and influences of radiotransmitters on hens may have biased field estimates. Therefore, comparisons of model predictions and field estimates could not be interpreted unambiguously as model success or failure. However, we believe inappropriate values for key inputs were primarily responsible for lack of agreement between model predictions and field estimates. Consequently, we executed an ad hoc model using inputs derived from our field data. After changing inputs for nest mortality in Conservation Reserve Program cover, hen age ratios, and daily hen mortality, endpoint predictions of hatch rate, hen success, average clutch size, and hatched nests fell within field estimate confidence intervals.
Appropriate model inputs are essential. The model is landscape-based and most data needs focus on gaining a better understanding of habitats being managed. Current date- and habitat-specific estimates of daily nest survival, knowledge of attractiveness for nesting of a number of habitats, and information about hen body mass and seasonal survival are needed. Data are most lacking for Minnesota Prairie Pothole Joint Venture target areas where the preponderance of breeding mallards occur.
Key words: Anas platyrhynchos, mallard, modeling, model assessment, management planning, Prairie Pothole Joint Venture, landscapes
Zicus, Michael C. and David P. Rave. 1998. Assessment of a mallard model in Minnesota's prairie coteau. Minnesota Wildl. Rep. 12. 43 pp.This resource should be cited as:
Zicus, Michael C. and David P. Rave. 1998. Assessment of a mallard model in Minnesota's prairie coteau. Minnesota Wildl. Rep. 12. 43 pp. Jamestown, ND: Northern Prairie Wildlife Research Center Online. http://www.npwrc.usgs.gov/resource/birds/malmodel/index.htm (Version 22JUL99).