Breeding Population Inventories and Measures of Recruitment
IV. Sampling
B. Stratification
Waterfowl productivity and breeding density can vary
greatly in any survey region, and stratification usually
increases the precision of sample estimates (Rutherford and
Hayes 1976). A number of criteria have been used for
delineating strata. The cooperative breeding ground
surveys use large strata selected on the basis of prior
knowledge of breeding waterfowl density and habitat
characteristics. The strata are bounded by political
boundaries, states and provinces, and are subdivided
according to latitude, pond density, and physiography.
Isakov (1970) divided the Soviet Union into districts and
stratified these based on prior knowledge of waterfowl
density. Stewart and Kantrud (1972) used geographic
variation in species composition within physiographic
regions as a basis of stratification in their studies of
breeding waterfowl in North Dakota. In their first studies of
breeding waterfowl in the prairie pothole region, Stewart and
Kantrud (1973) used a simple stratified random sample. That
sample proved inefficient because of the amount of travel
between sample plots. Later they (Stewart and Kantrud
1974) used two-stage cluster sampling. Plots (64.8 ha) were
drawn at random from randomly drawn townships (93.2 km2)
. Reed and Changnon (1987) stratified their study area by
using habitat maps and aerial photographs. They created
three strata representing perceived potential for production
of greater snow geese and then used optimum allocation to
assign plots to the strata. Chamberlain and Kaczynski (1965)
stratified their sample of aerial transects after the fact based
on phenological events such as the date at which ponds
and lakes became free of ice.
Previous Section -- Sampling--Sample Plots Return to Contents Next Section -- Sampling--Sample Design and Error