Adaptive Cluster Sampling: Estimating Density of Spatially Autocorrelated Larvae of the Sea Lamprey with Improved Precision

Abstract

Adaptive cluster sampling (ACS) provides researchers with an alternative technique to estimate the abundance of rare or spatially clustered organisms, but its application in field investigations has been limited to relatively few studies. We used ACS to estimate parameters of a spatially autocorrelated population of larval sea lampreys, Petromyzon marinus, in Wilmot Creek, a Lake Ontario tributary. When compared with simple random sampling (SRS), ACS significantly increased catch per sample as well as the percentage of samples that contained larvae. Although ASC-generated samples are spatially biased, the use of established formulae enabled us to calculate unbiased estimators of mean larval density and variance. With ACS, variance was reduced, improving the precision around estimates of mean density, however; increased precision came at the price of greater sampling effort. When variance was adjusted for effort, ASC provided equal or greater efficiency than SRS in 33% of sampling events, with no apparent site or seasonal bias. Based on the knowledge that larval sea lampreys are spatially aggregated, we anticipated that ACS would result in higher precision for a greater proportion of sampling events. Nonetheless, we consider ACS to be a useful technique for evaluating larval sea lamprey populations and anticipate increased application for investigating other spatially over-dispersed species.