USGS Home
Contact USGS
Search USGS

Many investigators (e.g., Stewart and Kantrud 1972, 1973, 1974, Brewster et al. 1976, Kaminski and Prince 1984) have used equal-size, randomly selected plots of land. These plots are usually based on legal subdivisions, which in North America are often bounded by roads or fence lines and thus are easy to locate on the ground as well as on maps or photographs. Stewart and Kantrud (1972) used 64.8-ha sample plots. They found this size practical for logistic reasons, and they were able to see birds from one part of the study plot land in another part of the plot, thus minimizing the problem of "roll up" discussed earlier. Brewster et al. (1976) repeated their sampling procedure. An alternative method is to draw a sample of individual wetland basins. This procedure is best suited to areas where wetlands are large and dispersed but becomes impractical where there are numerous small wetlands, as in the glaciated prairies of North America.

Aerial surveys have usually used belt transects for estimating breeding populations and for obtaining indices to recruitment. Transects have the advantage of minimizing the amount of flying between sample areas. Transects oriented in an east-west direction are used in cooperative breeding ground surveys, and similar procedures have been used in many other inventories. The aerial transect is well suited as a sampling unit in areas with many small wetlands, but it becomes inefficient when waterfowl are found on larger water bodies, which may be separated by landforms without waterfowl habitat. Haapanen and Nilsson (1979) compared the use of belt transects and square plots laid out in transects for aerial surveys. They found that square plots were more efficient as a sampling unit than transects in northern Fennoscandia. Dennis (1974) also found that square plots were more efficient than transects for aerial surveys in Ontario.

If the purpose of the survey is to estimate recruitment rate, then the size of the sample plot is important. When an estimate of recruitment rate is required on the individual plots, the nests or broods counted on the plot should be derived from the pairs that were counted on the plot. Cowardin et al. (1983) recommended that the plot size should be equal to or greater than the size of the home range of the species being studied. For the Mallard they used 10.4-km 2 plots.

For estimating recruitment rate, telemetry, especially when combined with the use of aircraft (Gilmer et al. 1981), has a decided advantage. In telemetry studies, the radio-equipped hen rather than a plot of land becomes the sample unit. In general, the procedure permits much larger sample size for estimating recruitment rate, and the estimate should be unbiased if a representative sample of hens can be obtained. In addition, the sample can be distributed over a large area. Telemetry studies, however, are extremely expensive and not practical for many applications.