Does fish behaviour bias abundance and length information collected by baited underwater video?

Baited remote underwater stereo-video systems (stereo-BRUVs) are commonly used to sample fish assemblages across areas of differing fish densities with little consideration of how intraspecific and interspecific behaviours may influence estimates of abundance and body-size distribution. To investigate these potential biases, the current study compared the abundances and body-size distributions of seven target carnivorous species and six lower trophic level non-target species, across sites with high and low densities of large-bodied target species, using Stereo-BRUVs. Samples were collected inside and outside of an area closed to fishing at the Houtman Abrolhos Islands, Western Australia. Densities of large-bodied target species were found to be higher inside the closed fishery area, compared to similar areas outside. The presence of large-bodied target species did not appear to influence the body-size distribution of conspecifics, or the abundance and body-size distribution of small-bodied non-target species throughout the deployments. The abundance of large-bodied target species was found to peak earlier in deployments within the closed area than the areas open to fishing. This difference may be due to the higher relative density with the closed area, which may result in shorter arrival times as fish move towards the baited video, and/or to behavioural differences, as fish within the closed area may approach the baited video more readily. This potential behavioural difference between areas closed and open to fishing has important implications for duration of baited video sampling times, and we suggest that shorter deployments times (< 15 min) are less likely to bias abundance estimates of fishery target species.