HABITAT USE AND NEST SUCCESS OF OVERWATER NESTING DUCKS IN WESTCENTRAL MINNESOTA

Where several duck species coexist, managers need knowledge of species -specific similarities and differences in patterns of nest habitat use and nest success. We searched overwater habitat (i.e., rooted stands of cattail [Typha spp.], bulrush [Scirpus spp.], or phragmites [Phragm ites australis], floating mats of sedge [Carex spp.] and/or cattail, w et sedge/grass [Poaceae] meadows, and willow [Salix spp.] swamps) in w estcentral Minnesota and located 155 overwater nests of 5 duck species . Nests were in wetlands of 0.2 to 32.0 ha that were 20 to 100% covere d by overwater habitat. A canonical discriminant function analysis cla rified some similarities and differences among the 5 species in their nest habitat use. Some individuals of each species, but especially red heads (Aythya americana) and ruddy ducks (Oxyura jamaicensis), nested in relatively more open, larger, deeper type 4 wetlands with cover nea r nests dominated by cattails and lacking sedges. Most mallards (Anas platyrhynchos) and ring-necked ducks (Aythya collaris) nested in small er, shallower type 2 and 3 wetlands which had a high proportion of the basin covered by overwater habitat and at sites where overwater veget ation, particularly sedges, was dense and nests were well screened. Ma llards and ring-necked ducks were the only species nesting in type 2 w etlands or on floating sedge mats and except for 2 canvasback (Aythya valisineria) nests, were the only species to use floating cattail mars as nest sites. Canvasbacks typically nested in type 3 and 4 wetlands of intermediate size, depth, and proportion of basin covered by overwa ter habitat. Their nests were at sites of moderate vegetation density and were less well screened by vegetation than those of the other spec ies. Mallards had lower nest success (3.8%) than ring-necked ducks (34 .1%) or all diving ducks combined (26.5%) (alpha = 0.10). Of 105 nests that failed to hatch, 74.3% were depredated while 13.3% were flooded. Multiple logistic regression analyses indicated that none of the 11 h abitat characteristics measured at our nest sites predicted whether a nest would hatch or be depredated (all P > 0.20).