Spatial inferences from adaptive cluster sampling of Gulf of Alaska rockfish

National Marine Fisheries Service trawl surveys result in more variable biomass estimates for long-lived Gulf of Alaska rockfish than researchers expect. Adaptive cluster sampling (ACS), a technique not yet widely used in fisheries studies, may provide improved estimates for aggregated rockfish populations. In August of 1998, the first of two sampling cruises tested ACS for rockfish, specifically Pacific ocean perch (POP, Sebastes alutus) and shortraker and rougheye rockfish (SR/RE, Sebastes borealis, Sebastes aleutianus). Two study areas east of Kodiak Island were selected for the experiment. The POP area was subdivided into four strata and the SR/RE into two strata based on habitat and geographic location. In each stratum, simple random sampling was conducted initially followed by ACS around the top one to three stations from the random sample. A stopping rule prevented the sampling from continuing indefinitely. ACS resulted in more precise estimates for POP compared to the initial random sample, but not with equal sample sizes. ACS combined with stratification provided the best estimates for POP, suggesting that the spatial distribution has both fine-scale and habitat-scale patterns. Results suggested that ACS worked better for POP than for SR/RE. Variogram analysis suggested that the distribution of POP was more aggregated than SR/RE, but not as aggregated as we expected. Both species were highly restricted to specific depths. The highest POP catches occurred during the early morning hours. While this experiment did not resolve whether ACS should be used for rockfish surveys, further investigation of this approach should continue.