Effects of Sampling Strategy, Detection Probability, and Independence of Counts on the Use of Point Counts1

Many factors affect the use of point counts for monitoring bird populations, including sampling strategies, variation in detection rates, and independence of sample points. The most commonly used sampling plans are stratified sampling, cluster sampling, and systematic sampling. Each of these might be most useful for different objectives or field situations. Variation in detection probabilities and lack of independence among sample points can bias estimates and measures of precision. All of these factors should be considered when using point count methods. Sampling strategies, variable detection probabilities, and independence among counts are aspects of point count methodology often overlooked but require emphasis when planning point count projects or analyzing point count data. At least as much planning should be focused on sampling schemes and potential sources of bias as on survey logistics. The first consideration in planning a project with point counts is to explicitly state the objective. The three major objectives of point counts are to: (1) monitor trends, (2) assess habitat relationships, and (3) map bird distributions. Different sampling strategies best address each of these objectives; it might be impossible to design a single sampling plan that will provide data to do all three effectively. The effects of variable detection rates (e.g., birds are more easily detected at some time or place than at other times or places) and spatial correlation (counts at points close together are more similar than counts far apart even if in the same habitat) also differ among these objectives.