Evaluating a New Video Metric for Enumerating Reef Fish in the Southeastern US

The selection of an appropriate method to monitor reef fish populations is an integral part of developing a fishery-independent survey.  Over the last twenty years, underwater video has become an increasingly important tool to assess reef fish populations in southeastern US waters. A central issue in the analysis of video data is the selection of an appropriate metric to quantify the number of individuals in an area.  The most widely used method to index reef fish abundance from videos is the ‘MinCount’ approach, calculated as the maximum number of individuals observed in a single video frame at one time.  However, recent simulations suggest that the MinCount approach may not track linearly with true abundance, instead reaching an asymptote when true abundance is high.  These simulations showed that an alternative video metric, the ‘MeanCount’ (i.e., the mean of fish counts at a number of intervals throughout a video segment), tracked linearly with true abundance with only slightly reduced precision compared to MinCount.  The current study sought to empirically groundtruth the findings of these previous simulations by relating MinCount and MeanCount from videos collected during fishery-independent surveys in southeastern US waters.  Data for a variety of reef fish species, representing different behaviors and ecologies, were analyzed.  A nonlinear relationship between MinCount and MeanCount was documented for most species, particularly for those species reaching relatively high abundance levels.  These results are consistent with previous simulations and suggest that MeanCount more accurately reflects changes in true abundance than MinCount.  The results of our study should lead to improvements in the utility of video-based fishery-independent surveys, as well as the ability of stock assessment scientists to accurately measure changes in reef fish populations.