USGS Home
Contact USGS
Search USGS

The surveyed area encompasses 3.03 million km² (Figure 13-2). About 78-85% of the continental Mallard population breeds within the area (Pospahala et al. 1974, R. S. Pospahala pers. comm.). The surveys are conducted from low-flying aircraft. Hawkins et al. (1984) related much of the colorful history of development of aerial survey techniques. Several authors (Crissey 1957, Martinson and Kaczynski 1967, Henny et al. 1972) have traced the development of these surveys from experimental work in 1947 to the system currently in use. The sampling frame, methods used, and data bases created were described in detail by Martin et al. (1979). Johnson (1986) described the computations necessary for estimating numbers of ducks, reviewed some of the shortcomings of the long-term data set derived from the surveys, and suggested an alternative method of estimating duck numbers from the same data set.

Figure 13-2. Strata and transects used in the North American cooperative breeding ground surveys. Lines represent transects where ducks are counted from an aircraft. These counts are expanded to an estimate for each stratum shown by enclosing polygons.

Our purpose is not to reiterate the description of the surveys, which has been covered in the works cited above. We merely use them as an illustration of a large and complex operational system for estimating breeding population and obtaining indices to recruitment. These surveys are continental in scope and, therefore, do not require some of the difficult decisions concerning whether birds on a transect are resident or transient. They do require that individuals observed on more than one transect not be included in the breeding population more than once. There is no practical way to determine whether this requirement has been met. The surveys are initiated from south to north as migration progresses in order to minimize the chances of double counting.

A more important source of bias arises because the surveyed area does not include the entire continent. There is no good way to estimate the number of ducks in the unsurveyed area on an annual basis, and the proportion outside the boundaries likely varies by species. For the Mallard, various periodic state and federal surveys were available for deriving the average of 78-85% of the breeding Mallards estimated to be in the surveyed area. Based on the limitations of the individual surveys, however, the validity and representativeness of this range of values can be questioned. If the actual proportion of the continental population in the surveyed area varies greatly from year to year, then the continental population size estimated from the survey data would be biased. The only solution would be to expand the surveyed area to the entire continent, a procedure whose cost might not warrant the gain in accuracy. Not only is there potential for movement of the population in and out of the surveyed area, but also there is considerable movement within the surveyed area (Hansen and McKnight 1964, Johnson and Grier 1988). This movement can cause additional biases by causing birds to be missed or counted twice.