Statistical Power Comparison of Two Sampling Protocols for Riverine Snails
Fore, Leska S.
Clark, William H.
MetadataShow full item record
We compared the statistical power of two alternative sampling designs to detect changes in threatened and endangered snail species populations in the Mid-Snake River (Idaho). Our goal was to determine which sampling approach would have the best chance of detecting a change associated with different hydroelectric project management scenarios. We summarized the data as 1) the average number of snails collected across quadrats (density/m2) and 2) the proportion of quadrats that had snails present. We calculated the minimum detectable difference that each measure could detect with a two-sample t test. The density measure was highly variable and even a complete loss of snails failed to represent a statistically significant change for most sites. The precision improved somewhat when density was log-transformed, the number of replicate quadrats was increased, and larger sampling quadrat used; however, statistical power to detect change remained low. In contrast, proportion measures were much more precise and could detect a 34% reduction in the proportion of quadrats with snails present. When the number of quadrats was increased to 30, a 24% change could be detected and for 50 quadrats an 18% change. Proportion of quadrats with snails present was also highly correlated with the average density of snails (Pearson’s r = 0.91). In addition to being a more sensitive indicator, the proportion measure is quicker to observe for each sample which means that a larger area can be surveyed during the same amount of time.