Large-scale surveys to estimate animal abundance can be useful for monitoring population status and trends, for measuring responses to management or environmental alterations, and for testing ecological hypotheses about abundance. However, large-scale surveys may be expensive and logistically complex. To ensure resources are not wasted on unattainable targets, the goals and uses of each survey should be specified carefully and alternative methods for addressing these objectives always should be considered. During survey design, the impoflance of each survey error component (spatial design, propofiion of detected animals, precision in detection) should be considered carefully to produce a complete statistically based survey.
Failure to address these three survey components may produce population estimates that are inaccurate (biased low), have unrealistic precision (too precise) and do not satisfactorily meet the survey objectives. Optimum survey design requires trade-offs in these sources of error relative to the costs of sampling plots and detecting animals on plots, considerations that are specific to the spatial logistics and survey methods. The Horvitz-Thompson estimators provide a comprehensive framework for considering all three survey components during the design and analysis of large-scale wildlife surveys.
Problems of spatial and temporal (especially survey to survey) heterogeneity in detection probabilities have received little consideration, but failure to account for heterogeneity produces biased population estimates. The goal of producing unbiased
population estimates is in conflict with the increased variation from heterogeneous detection in the population estimate. One solution to this conflict is to use an MSE-based approach to achieve a balance between bias reduction and increased variation. Further research is needed to develop methods that address spatial heterogeneity in detection, evaluate the effects of temporal heterogeneity on survey objectives and optimize decisions related to survey bias and variance.
Finally, managers and researchers involved in the survey design process must realize that obtaining the best survey results requires an interactive and recursive process of survey design, execution, analysis and redesign. Survey refinements will be possible as further knowledge is gained on the actual abundance and distribution of the population and on the most efficient techniques for detection animals.
|Title||Horvitz-Thompson survey sample methods for estimating large-scale animal abundance|
|Authors||M.D. Samuel, E.O. Garton|
|Publication Subtype||Journal Article|
|Series Title||Transactions of the North American Wildlife and Natural Resources Conference|
|Record Source||USGS Publications Warehouse|
|USGS Organization||National Wildlife Health Center|