Two-phase sampling to estimate river-wide populations of freshwater mussels

Abstract

Two-phase sampling provides a statistical framework for combining data from qualitative and quantitative sampling methods. It is a useful approach when the survey objective is a drainage-wide estimate of mussel density and the cost of qualitative sampling is small relative to the cost of quantitative sampling. This survey design has several advantages including: 1) no need for a priori classification of stream reaches into sampling strata, 2) allocation of sampling effort so that more time is spent sampling where mussels are at higher density and less time is spent sampling where mussels are not present, 3) the ability to evaluate the relationship between qualitative and quantitative estimates of density, and 4) efficient allocation of effort so that more stream reaches can be surveyed compared to quantitative-only sampling designs with similar effort. The survey design consists of sampling as many sites as possible during a qualitative 1st phase and sampling far fewer sites during a quantitative 2nd phase. In 1995, we used a 2-phase sampling design to estimate the distribution and abundance of freshwater mussels in riffle habitat in the Cacapon River, West Virginia. Our estimate of river-wide surface density of freshwater mussels in riffles was 0.59/m2 (SE = 0.14). We used resampling simulation based on our data and determined that the most effective 2nd-phase sampling strategy was to sample a low to moderate proportion of low-density sites and a high proportion of high-density sites.