Nested sampling: an improved visual-census technique for studying reef fish assemblages

Our aim was to devise visual sampling protocols to estimate densities in a morphologically and behaviorally diverse assemblage of reef fishes. The focus was on a design that is cost-effective and produces relatively accurate and precise density estimates for the largest number of species. We evaluated the 2 most widely used shapes of sampling units (strip transects and stationary cylinders) of several dimensions on a fringing reef at Fernando de Noronha Archipelago, tropical West Atlantic. Small sampling units produced the best density estimates for small fishes, while large ones produced the best estimates for large fishes, regardless of sampling unit shape. The sampling unit with the lowest cost was the stationary cylinder. We further compared accuracy and precision of density and richness estimates from several combinations of stationary cylinders and fish size. A simple improvement in stationary censuses, i.e. a sampling unit composed of 2 nested cylinders (2 m and 4 m radii, respectively), enhanced the accuracy and precision of estimates. Counts of small fish (<= 10 cm) were made in the smaller cylinder and counts of larger fish (>10 cm) were made in the larger. Such nested cylinders reduce the bias towards large and conspicuous species and allow for separate density estimates for adults and juveniles of large species.