Estimating Scallop Dredge Catch Efficiency Using HabCam, a Habitat Mapping Camera System

The Atlantic sea scallop fishery along the U.S. Northeast Continental Shelf is one of the most valuable fisheries in North America.  The annual fishery-independent sea scallop survey provides the basis for estimating abundance of scallop stocks and setting a Total Allowable Catch, and is dependent on estimates of catch efficiency of the dredge gear used in the survey.  Catch efficiency, potentially by size, can be estimated for some fishery-independent surveys by deploying paired gear types, where one is assumed to be 100% efficient.  A habitat mapping camera system (HabCam) was developed by a group of scientists, engineers and commercial fishermen as a non-invasive tool for rapidly surveying scallops with high accuracy, and provides a tool for assessing dredge efficiency.  HabCam is towed at ~ 5 kts over the seafloor while capturing 5 images per second with about 50% overlap.  Scallops and other organisms along with substrate type are classified and measured providing a robust count and size distribution of targets along the track.  During 2008 and 2009 the HabCam and a NOAA scallop survey dredge were deployed at 146 stations to obtain paired information on scallop abundance from each “gear.”  Thousands of images were captured by the HabCam at each station and a subset of those images were processed for counts of scallops, which resulted in a number of observations per station ranging between 60 and >4000.  We developed a statistical model for the paired HabCam and scallop dredge data to estimate the catch efficiency of the scallop dredge relative to the HabCam.  The general model assumes the counts for each HabCam image and scallop dredge tow arise from negative binomial distributions.  We fit a range of models that made different assumptions about mean dredge catch efficiency and variability in counts at each station.  Under the assumption that the HabCam is 100% efficient, our results suggest that the dredge catch efficiency differs between predominantly sand (44%, 1% SE) and gravel (38%, 1% SE) substrates. These efficiency estimates are similar to those from previous studies, but the high precision of these new estimates result in more reliable measures of abundance.