Statistical Perspectives and Experimental Design When Counting Birds on Line Transects

We used data from 87 km of line transects in northern Michigan and northern Wisconsin in June 1985 to determine the optimum length of replicate transects required to assess bird populations. Data are from a study comparing bird populations in areas affected by electromagnetic fields (treatments) to those in areas unaffected (controls). Transects were subdivided into six different lengths varying from 100 m to 1,000 m. With equal effort, we were able to detect smaller differences in bird counts between control and treatment areas with short transects and large sample sizes than with long transects and small sample sizes. Transects shorter that 350 m required the smallest amount of effort to detect a 15% difference between means for number of individuals and species. The most efficient transect length for detecting a 25% difference between means was not consistent for individual species but was positively correlated with relative density. The shortest transect (100 m) was best for detecting differences for the Ovenbird (Seiurus aurocapillus), Red-eyed Vireo (Vireo olivaceus), and Nashville Warbler (Vermivora ruficapilla) the most abundant species; a 250-m transect was best for the Black-throated Green Warbler (Dendroica virens); and a 500-m transect for the least abundant species, the White-throated Sparrow (Zonotrichia albicollis).