Counting dead birds: examination of methods

We studied three methods (line transect, circular quadrat, complete count) for estimating density of dead birds, using models of sparrows and meadowlarks placed at a density of 50 birds of each type/ha. Line transects with a 500-m search line were used in cultivated pasture and native prairie habitats. The number of birds found by individual searchers in line transects varied markedly, particularly, in pasture habitat. More birds were found, and birds were detected at a greater distance, in prairie than in pasture. More meadowlarks than sparrows were found in both habitats but the mean estimated density of meadowlarks was greater than that of sparrows only in prairie. The number of birds found during most searches was less than that suggested for estimating density accurately and longer search lines were required. Density estimates obtained using circular plots to sample 10% of the prairie area ranged from 20 to 80 birds/ha for meadowlarks and from 10 to 60 birds/ha for sparrows in prairie habitat. A complete search by 25 volunteers spaced at 4 m intervals detected 90% of meadowlarks but only 62% of sparrows in pasture habitat. Mean (SD) time for a single search, including searchers and recorders, was 3.0 (0.8), 1.5 (0.3), and 7.5 person-hours, for line transect, circular plot and complete search, respectively.