Using GIS to Predict Mallard Nest Structure Occupancy
Knowing which type of nest structure to use and where to deploy them in a landscape should be important to waterfowl managers. Zicus et al. (2006a) studied mallard (Anas platyrhynchos) nest structure occupancy in an attempt to understand how landscape features affected structure use. They were interested in the effect of 5 covariates, and their final fitted model was complex, including 3 interactions and 1 main effect. More nests were initiated as the size of the open water area where structures were deployed increased. Simultaneously, cover influence interacted with period of the nesting season such that nesting probability was positively associated with cover height and density early in the season and negatively associated with cover height and density late in the season.
Nest success in structures is generally good (Eskowich et al. 1998) with early nests having higher nest success (M. Zicus, Minnesota Department of Natural Resources, unpublished data). Consequently, hen mortality associated with renesting (Sargeant et al. 1984) would be reduced for hens nesting in structures early in the year. Further, brood and duckling survival from early-hatched nests is believed to be greater than that of later-hatched nests (e.g., Rotella and Ratti 1992, Dzus and Clark 1998, Krapu et al. 2000). These understandings led Zicus et al. (2006a) to recommend that nest structures be deployed in larger wetlands where early-season residual cover in the surrounding uplands was most abundant within 1 km of the structure. Geographic Information System (GIS) models might provide powerful tools to help waterfowl managers decide where nest structure should be placed in complex landscapes.
A model was developed that used 3 predictors: 1) nest structure type, 2) 4 measures of the size of open water area containing the structure, and 3) a measure that described the mean aggregate visual obstruction of all residual cover during the early part of the nesting season in a buffer with a 1.6 km radius around each structure. We used a bootstrap procedure to obtain an unbiased measure of future predictive performance of the models that we fit. Unfortunately, we failed to produce a GIS model with much predictive power.
Statistical analysis suggests that the mean visual obstruction values are not likely to be useful for predicting the mean number of ducklings (across periods and years) produced in nesting structures and that the available measures of open water size (NWI and GAP) are of questionable value.
For additional information about nest structure occupancy predictions contact:
Habitat and Population Evaluation Team