An Evaluation of the Strip Transect Method for Censusing Bird Communities in Forests

Abstract

The effects of varying transect width and census duration on the number of birds counted, the density estimate, number of species detected and the percentage of unidentified birds were examined in rainforest and wet sclerophyll forest at Mount Nardi in northern New South Wales. The nine combinations of three strip widths (40, 60 and 80 m) and three durations of census (24, 18 and 12 min) were compared in 200-m-long transects in each forest. The census of birds was more sensitive to changes in census duration than in strip width, and the effects were greater in the rainforest than the wet sclerophyll forest. Both the precision of the density estimates and the number of species detected were highest for the narrowest strips censused for the longest time. The bias of the density estimate varied with the treatments both within and between forests, and thus the strip transect could not be used to compare them. We suggest that variation in bias between sites may be a problem common to all transect counts of birds.