Investigating uncertainties created by camera improvement in an optical survey

Abstract

Fish are an essential source of global food. Fishery-independent data is critical to fisheries assessment and management. The uncertainties created by changing methods to collect this data must be understood. In the U.S. Northeast Atlantic, a drop camera survey of Atlantic sea scallops is a critical source of fishery-independent data. From 1999 to 2016 the primary sampling unit of the survey was an analog 2.8 m 2 image. In 2017 this image was replaced with a digital 2.3 m 2 image. Here the key uncertainties associated with this change, an increased ability to detect small scallops and altered estimates of size frequencies and densities due to the survey's methods in the context of the smaller image size, are explored. Scallop detection differences were quantified using a maximum likelihood model and paired analog and digital images. The proportion of scallops measured by size in actual and simulated data and different methods of counting scallops along the edges of images were compared to explore these uncertainties. More scallops were detected in the digital images until scallops were greater than 75 mm in shell height, indicating the models developed here are needed to adjust scallop counts for analysis incorporating juveniles and data from the two cameras. The proportion of scallops measured did not match the simulation and was not linked to quadrat size. There was no significant difference in scallop density with different counting methods. This investigation spurred by altering the methods of a well-established survey provided confidence in results with potential avenues for fine-tuning.