Evaluation of adaptive cluster sampling for remotely-operated underwater vehicle surveys of inshore rockfish (Sebastes spp.)

Remotely-operated vehicle surveys of inshore rockfish (Sebastes spp.) are low-cost and non-lethal, but frequently result in density estimates with high variance. Adaptive cluster sampling may result in more precise abundance estimates when the distribution of organisms is rare and clustered. Using computer simulations, I investigate the ability of adaptive cluster sampling to reduce sampling variance for remotely-operated vehicle surveys of rockfish. In general, adaptive cluster sampling was less precise than simple random sampling in surveys of equivalent sampling effort. Adaptive cluster sampling was only more precise (for equivalent sample size) for highly clustered rockfish populations, large initial sample sizes, and small numbers of adaptively sampled units, conditions which are not likely met for ROV surveys of rockfish, based on my analyses. Adaptive cluster sampling is not recommended for remotely operated vehicle surveys of rockfish and other strategies should be investigated to increase the precision of abundance estimates.