Northern Prairie Wildlife Research Center
Much effort has been devoted to assessing the status of the continental mallard populations through extensive surveys (Pospahala et al. 1974, Martin et al. 1979) and to annual survival estimates based on banding (Anderson 1975). Recruitment has been less intensively studied, especially on local areas, although several earlier studies have contributed substantially to our understanding of breeding biology of the species (e.g., Sowls 1955, Keith 1961, Smith 1971, Stoudt 1971, Dzubin and Gollop 1972). Our study was designed to furnish current recruitment information for a large study area in central North Dakota. In addition, radiotelemetry enabled us to avoid some problems encountered in previous studies and to obtain detailed data that were used to evaluate a previously developed model (Cowardin and Johnson 1979). That model related recruitment and survival of mallards to change in population size.
In the present paper we present estimates for the recruitment parameters derived from a sample of radio-marked hens. This technique has advantages over the more traditional estimate of recruitment derived from censuses of breeding populations and young produced. The marked birds are free to select their nesting site, and, if the hens represent the breeding population and are marked before nest site selection, an unbiased sample of nests can be obtained. To obtain an unbiased sample with unmarked birds, it is necessary to sample habitats in proportion to their occurrence (e.g., Duebbert and Kantrud 1974, Higgins 1977) and to find nests in the sampled areas. Recruitment can be estimated from radio-marked hens by determining the number of young fledged by each hen in the sample. The hen is the sample unit, and sample hens can be obtained over large areas. Recruitment estimates based on census data are derived from sample plots and require estimates of breeding population and total young produced on each plot. Breeding population estimates are difficult to interpret (Dzubin 1969a), and brood counts for the secretive mallard are highly unreliable (Stoudt 1971). Furthermore, the study plot is the sample unit, and replication of plots over a large area is time consuming and expensive.
The study had the following objectives: (1) to obtain unbiased estimates of the use of various habitats for nesting and to compare use of habitat with availability, (2) to estimate nest success, hen success, and brood survival and to develop an estimate of recruitment rate based on these data (3) to relate variation in recruitment to variation in pond conditions and weather; and (4) based on these data, and previously published survival estimates, to forecast future trends in mallard populations breeding in central North Dakota.