Detectability of Forest Birds from Stationary Points in Northern Wisconsin1 Amy T

Estimation of avian densities from point counts requires information about the distance at which birds can be detected by the observer. Detection distances also are important for designing the spacing of point counts in a regional sampling scheme. We examined the relationship between distance and detectability for forest songbirds in northern Wisconsin. Like previous investigators, we found that some birds can be heard from much greater distances than others. Within the same species, some individuals (or the same individual under different circumstances) can be heard from greater distances than others. In general, this within-site variation in detectability is similar to variation in detectability among individuals in different forest types. Knowledge about the relationship between distance and detectability can be used to approximate the area sampled from a stationary point. This information can then be used to estimate the density of vocalizing birds. Even under ideal field conditions with accurately measured distances, detectability does not follow a simple threshold relationship. On the basis of our empirical data, we use a statistical probit analysis to describe the attenuation of detection with distance; the resulting sigmoidal function can be used to approximate the effective sampling area. Complications arise because individual birds become increasingly difficult to distinguish from conspecifics at greater distances from the observer. Coupled with variation caused by habitat structure, wind conditions, observer bias, and other factors, we conclude that data from point counts can give only a crude picture of avian density. Nevertheless, such estimates might be the best available, and the costs or ambiguities associated with alternative procedures might outweigh the disadvantages of the point count method. Point counts are attractive for avian population monitoring programs because they are methodologically straightforward and permit sampling of numerous geographic sites. Results from unlimited-radius point counts are generally accepted as valid indices of abundance for a single species under some circumstances (Blondel and others 1981), but between-species comparisons of point counts are complicated by species-specific differences in delectability. Even within-species comparisons are based on the untested assumption that detectability is consistent among different sites. Of course, virtually any method for sampling bird populations is burdened with weaknesses (Ralph and Scott 1981). Despite its shortcomings, the point count method is probably the least subjective of today's widely used procedures. Unlimited-radius point counts are probably the simplest of all approaches because, unlike the variable-radius point count method (Reynolds and others 1980) or the fixed-radius method (Gates, in these Proceedings), observers do not need to estimate the distance of each bird from the observer. Interpretations of unlimited-radius point counts, on the other hand, are more sensitive to differences in detectabil1 An abbreviated version of this paper was presented at the Workshop on Monitoring Bird Population Trends by Point Counts, November 6-7, 1991, Beltsville, Maryland. 2 Graduate Student, Department of Natural and Applied Sciences, University of Wisconsin-Green Bay, Green Bay, WI 54311, present position: PhD Candidate, Graduate Group in Ecology, University of CaliforniaDavis, Davis, CA 95616; Associate Professors, Department of Natural and Applied Sciences, University of Wisconsin-Green Bay, Green Bay, WI 54311 USDA Forest Service Gen Tech. Rep. PSW-GTR-149. 1995 ity among individuals of the same or different species. In this paper we provide an objective analysis of the relationship between distance and detectability of forest songbirds in northern Wisconsin. We focus particular attention on the maximum distance from which an observer can detect a singing bird. This distance dictates the area covered by an unlimited-radius count and has obvious implications for designing and interpreting point count procedures. We also consider how empirically derived relationships between detectability and distance might be used in conjunction with point count data to generate estimates of relative (or perhaps absolute) avian abundance. Emlen and DeJong (1981) pointed out that detection of bird vocalizations is a threshold phenomenon--a bird is either heard or it is not heard. If detectability thresholds (the maximum distances from which birds can be heard) are species-specific and reasonably consistent among the habitats of interest, then the area covered by an unlimited-radius point count theoretically can be determined empirically (area = where = the detectability threshold) (Reynolds and others 1980). The number of vocalizing individuals detected within this area then can be used to calculate bird density. Estimates of detectability thresholds also may provide insight into other issues considered in this volume, notably the ideal distance between census points needed to assure independence of sample points. 2 t D π t D