A comparison of the behavioural responses of fishes to a remotely operated vehicle and diver-based stereo-video sampling

Successful monitoring of fishes to inform adaptive management relies on accurate and robust data on a range of indicators including diversity, abundance, and length-based metrics. The methods for gathering this data have evolved from destructive techniques to underwater visual census (UVC), and now emergent methods such as divers or remote operated vehicles (ROVs) equipped with stereo-video systems. Stereo-video techniques can provide accurate information on traditional metrics such as abundance, length, and biomass, as well as metrics on fish behaviour. Fish wariness, measured by the minimum distance between fish and an observer or stereo-video system before the fish moves away (minimum approach distance; MAD) is an emerging metric for assessing changes in fishing pressure which can be more sensitive than traditional metrics. Differences in fish behaviour across methods can bias results and influence measurement accuracy and precision of stereo-video systems, because accuracy decreases as the distance of fish increases away from the cameras. We assessed the behaviour of six focal species towards a diver using a stereo-video system (stereo-DOV) and a stereo-video system mounted on a remotely operated vehicle (stereo-ROV). We also compared the fish lengths and the length frequency distribution of measurements from the two methods. We aimed to increase the understanding of the relative abilities of the two methods at gathering length-based data and the differences in fish behavioural responses towards the methods. There were no differences in the proportion of length measurements or the mean length, and small differences in the length frequency distribution sampled by the two methods. We found that MAD differed between the methods and the direction of this difference was species-specific. This was likely due to different trophic traits such as foraging modes and behavioural patterns between these species and potentially adaptations resulting from fishing pressure. Indicator species such as Pomacentridae species showed more aggressive behaviour towards SCUBA divers, likely due to their feeding regime, while fisheries targeted species such as Choerodon rubescens showed more wariness towards divers as a possible adaptation to fishing pressure. At the assemblage level, the relationship between MAD and fish length was weaker with the stereo-ROV, indicating that larger fish showed more wariness towards SCUBA divers than the ROV. The decreased wariness of larger fish and fisheries targeted species towards the stereo-ROV compared to stereo-DOV, as well as the similarities in the length frequency distributions sampled by the two methods suggest that stereo-ROV may provide a suitable alternative to stereo-DOV.