Evaluation of optimized variable area transect sampling using totally enumerated field data sets

• Published in: Environmetrics (London, Ont.) Vol. 16; no. 7; pp. 767 - 772
• Main Authors: Engeman, Richard M, Nielson, Ryan M, Sugihara, Robert T
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.11.2005; Wiley
• Subjects: Animal, plant and microbial ecology; Biological and medical sciences; density estimation; distance sampling; Earth sciences; Earth, ocean, space; Engineering and environment geology. Geothermics; environmental models; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Methods and techniques (sampling, tagging, trapping, modelling...); Monte Carlo method; optimization; plant damage; Pollution, environment geology; population density; population dynamics; sampling; simulation models; spatial distribution; spatial pattern; vertebrate pests; wildlife management; Natural population; distance sampling; Monte Carlo analysis; field studies; simulation; sampling; spatial distribution; quality; spatial pattern; spatial variations; populations; density estimation; Population density; Distance; errors; computers; Field study
• ISSN: 1099-095X; 1180-4009; 1099-095X
• DOI: 10.1002/env.736

Plotless density estimators (PDEs) can be efficient alternatives to quadrat sampling for estimating the density of stationary objects. Variable area transect (VAT) sampling had been identified, and optimized relative to effort, in previous Monte Carlo simulated population studies as a low-labor field method that demonstrated superior estimation properties among many PDEs considered. However, natural populations tend to be much more complex and less predictable in spatial distribution than computer generated populations. Therefore, we carried out a Monte Carlo simulation study that used 17 fully enumerated field populations rather than simulated populations. These natural populations represented a variety of population densities and spatial patterns. We focused on assessing the effect of the number of population members searched for along each transect (r), sample size, and transect width on estimation quality. Using relative root mean-squared error and relative bias as criteria, the optimal range for r was 5 to 7 population members encountered from each start point. Sample size was best if n > 20, but returns in estimation quality diminished by n = 40. Transect width was previously uninvestigated and found to be the most important design factor affecting estimation quality. Field studies should strive for transects as wide as logistically reasonable.