Many fishery sampling designs estimate abundance based on a set of sampling sites which are sampled once each year for a number of years. These sampling designs only account for spatial and annual variability. In these designs, short term variation in abundance due to local movements are assumed to be small relative to the annual and spatial variation. We present data that shows that in at least some systems, short term variation in abundance can be large, and if left unaccounted for, can limit the power of a sampling design. We repeatedly measured abundances of sculpin (Cottus spp), longnose dace (Rhinichthys cataractae), brook trout (Salvelinus fontinalis), coho salmon (Oncorhynchus kisutch) and rainbow trout (Oncorhynchus mykiss). Abundances were sampled within both small-scale local habitats (~10 m long) and larger sites ( ~90m long) up to 4 times over 14 days. We estimated abundances using two complementary techniques repeated snorkeling surveys and repeated electrofishing surveys. We also conducted a simple power analysis, where short term variability was the only source of variation, to demonstrate how short term variability can obscure changes in population abundances.
Our data suggests abundances within sites and local habitats vary on daily time scales, and that these variations are large enough to obscure even large changes in population size. Site scale abundances had coefficients of variation that ranged from 11.9% for sculpin , and 61.6% for rainbow trout. The average coefficient of variation across species, sites and sampling techniques was 31.6%. At the local habitat scale abundances were generally more variable. At the local habitat scale, abundance coefficients of variation ranged from 3.7% for sculpin to 173% for longnose dace. The average coefficient of variation across sites and sampling techniques was 60.9%.
Our power analysis tested the ability to detect a change in abundance, given varying amounts of short term variation. We found that at a 10% coefficient of variation, approximately the smallest amount of site scale variation we observed, we had 100% power detecting a 50% change in abundance with 5 samples before and after the change. Power decreased to 37% when the amount of variation increased to 30%, which was approximately the average amount of site scale variation we observed.