-While censusing coniferous forest birds in 30-m-radius circular plots, I suspected that frequent movement by birds into and out of the plots might be biasing my results. To quantify the extent of bias, I analyzed data from 142 censuses for three actively calling species that were unlikely to be missed if present on a plot for 5 min. Each census consisted of two consecutive 5-min counts. For the three species, 36 to 72% of all individuals detected were detected in only one of the two counts; thus at least that many moved across plot boundaries during a 10-min period. Density estimates based on a 10-min were 22 to 56% higher than those based on a 5-min period. My estimates for fixed circular plots were thus biased because they were derived from cumulative rather than instantaneous counts. Movement poses similar problems for variable circular plot, strip transect, and line transect methods. It is difficult to reduce bias due to movement without increasing bias from other sources. Recently proposed techniques for estimating bird densities have stimulated considerable debate over the relative accuracy of different methods (see especially Ralph and Scott 1981). Available methods include mark-recapture techniques (Seber 1973, Cormack et al. 1979), mapping and counting territories (Williams 1936, Kendeigh 1944, International Bird Census Committee 1970), and several methods based on counting individual birds (see Kendeigh 1944, Emlen 1971, 1977, Fowler and McGinnes 1973, Ramsey and Scott, 1979, 1981, Burnham et al. 1980, Reynolds et al. 1980). Counting methods are straightforward for stationary objects such as plants, but not for highly mobile organisms such as birds, which often move into and out of the census area during the period. Thus, to be strictly accurate, a must be instantaneous. In practice, however, it is impossible in most habitats for an observer to detect all birds present at one instant and most methods must rely on a of several minutes or more. In station counts, with a stationary observer, periods have ranged from 2-20 min (Scott and Ramsey 1981). In transect counts, with the observer moving along a line, the count period at any one point depends on the speed of the observer and the distance limits, ahead and behind, within which birds are recorded (Emlen 1971). A critical assumption for methods based on counts is that so little bird movement occurs during the that it does not affect density estimates appreciably. This assumption has received surprisingly little attention in the literature. Burnham et al. (1980) concluded that evasive movement away from the observer before detection is the only serious problem caused by bird movement in line transec surveys, but that it can result in substantial bias. Emlen (1971) pointed out that movement toward, as well as away from, the observer can be a problem with a few species such as hummingbirds. Random movement of birds into fixed plots during the was recognized as a problem by Seber (1973). Ramsey and Scott (1978) and Verner (1981) pointed out that random movement poses a similar problem for variable circular plots. Another source of bias is that moving birds are likely to be counted twice (Burnham et al. 1980, Reynolds et al. 1980). Apparently only three studies have provided empirical evidence for the amount of bias created by movement. Breckenridge (1935) and Emlen (1971) both noted that more birds were detected at intermediate than at short distances from transects, suggesting evasive movement by birds. R msey and Scott (1978), using variable circular plots, also found more birds at intermediate distances but attributed this to the random movement of distant birds to points within detection range. In the course of a study using small circular plots, I began to suspect that considerable movement into the plots was occurring during my counts, leading to cumulative rather than instantaneous totals and thus inflating my density figures. In this paper I estimate the degree to which such movement biased my density estimates.
Read full abstract
Dec 1, 1996
I Yu Khoma 
Open Access