Development of a remotely operated vehicle based methodology to estimate fish community structure at artificial reef sites in the northern Gulf of Mexico

A remotely operated vehicle (ROV) based methodology was developed to estimate reef fish community structure at artificial reef sites off Pensacola, Florida in the northern Gulf of Mexico. The method is based on the visual census technique developed by Bohnsack and Bannerot (1986), with a key departure being that sampling was conducted with a micro ROV instead of divers. A VideoRay Pro III ROV equipped with a red laser scale (distance between lasers = 10 cm) was employed to sample fish communities and estimate the size distribution of fishes at study sites. Pool experiments were conducted to examine the effect of distance from target (1, 2.5, and 5 m) and laser angle of incidence (0o, 5o, 10o, 15o, 20o, and 30o from perpendicular) on the accuracy of estimating fish length with the laser scale. Results indicate that fish length estimated with the laser scale was accurate (i.e., mean absolute error < 5%) for distances < 5 m and angles of incidence ≤20o from perpendicular. In the field, the ROV was used to sample a 15-m wide cylinder around artificial reefs from the seafloor through the water column. Two readers independently analyzed video samples (n = 24) in which all fishes were identified to the lowest taxonomic level possible and counted. Average percent error between readers among all samples was 7.4%; the correlation coefficient between taxa-specific counts was 0.997 (p < 0.001). Overall, results suggest that micro ROVs can be used to estimate reef fish community structure precisely at artificial reef sites, as well as to estimate size distributions accurately of fishes present.