Detecting seabird captures via video observation

An experimental assessment of video observation in the inshore (primarily snapper target) bottom longline fishery was undertaken from April to November 2015 with the key aim of assessing whether video observation is a valid approach for monitoring seabird captures. Known numbers of seabird proxies were deployed on longlines during setting and subsequently detected in the video footage of the line hauling operation.
The experimental approach was largely successful and allowed the detectability of the seabird proxies to be assessed as 88.7% (95% C.I. 77.0% - 95.7%) for the primary reviewer, and 94.3% (95% C.I. 84.3% - 98.8%) by using all unique proxy detections when multiple reviews were included. Reduced visibility due to spray affecting the camera lens affected the efficiency of review of some footage and contributed to some missed proxies.
The trial highlighted several features of a video observation programme that must be implemented for successful, on-going monitoring. Foremost amongst these are procedures at the programme and vessel level to ensure that good footage is continuously collected. The technical aspects of footage collection have largely been addressed by developments in the observation technology, and aspects such as routine camera cleaning simply require ongoing operational diligence. This clearly benefits from a commitment to facilitating observation by the vessel operator and crew.
During the trial a number of real seabird captures were detected. These provided evidence that the proxy chosen for the trial was appropriate and also the opportunity to evaluate the extent to which seabird species and status could be identified from the footage. Species was assigned for three captures and group level identifications for the others. Four of the six captured seabirds were released alive and apparently uninjured.
Overall, the experiment demonstrated that seabird proxies were reliably, but not perfectly observed in the footage. Detection rates can likely be improved by continuous improvement in operating procedures. Real seabird interactions were also detected. In general, captures of live seabirds were very apparent due to the movement of the bird. Video observation, implemented appropriately, therefore appears to be an appropriate approach to monitoring seabird interactions in this, and similar, fisheries.