A Comparison of Groundfish Species Composition, Abundance, and Density Estimates Derived from a Scientific Bottom-Trawl and a Small Remotely-Operated Vehicle for Trawlable Habitats

Despite a long history of being used to assess the abundance and distribution of benthic marine fishes, otter trawls are acknowledged to have several shortcomings. One major problem is the systematic exclusion of species and/or specific life history stages. Over the past five decades, a diversity of visually based survey tools, such as manned submersibles, video landers, and remotely-operated vehicles (ROVs), have been developed that allow assessment of benthic fish in complex habitats where trawling is not practicable. Few studies have been conducted, however, that directly compare the species composition observed by concurrent trawl and visual surveys on trawlable habitat, let alone the density estimates derived from such sampling. The increasing use of visual survey methods presents the opportunity for supplementing, or perhaps replacing, trawl surveys, but before this can occur comparisons of sampling bias and catchability/detection will be crucial if contemporary abundance estimates based on visual surveys and historic trawl-based estimates are to be merged into a seamless index for resource management purposes. Here, we report the results of paired benthic marine fish surveys using an observation-class ROV and an Eastern otter trawl. We conclusively demonstrate that on trawlable habitats in northern Puget Sound, WA, catch composition and density estimates differ substantially between these two survey methods for several key fish taxa and one targeted invertebrate. We conclude that sampling trawlable habitats with an ROV provides information for several small-bodied fish species that are otherwise undersampled by a benthic trawl, but provides an insufficient assessment of harvestable species due to the narrow field of view, limited capacity to sample individuals occurring >2m off the bottom, and propensity for buried flatfishes to go undetected. Our results suggest that ROV-collected data from trawlable seafloor habitats can provide complementary information to the trawl survey that could be used to develop a more complete model of Puget Sound ecosystem structure. Further, we recommend that any groundfish survey of an area consisting of heterogeneous habitats be conducted using a combination of trawl and ROV rather than by eliminating benthic trawling. Statistical methods must be developed to combine data for species that occur on both trawlable and untrawlable habitats that adequately account for methodological variation in detection rate and other sampling biases.