Download the publication
Summary
To understand the threat posed by incidental fisheries captures, accurate identification of seabirds captured in Aotearoa New Zealand fisheries is necessary. Historically, at-sea identification has been undertaken by Fisheries Observers with expert verification taking place during review of observer photographs and/or necropsy of bycaught individuals, however with the rollout of cameras on inshore commercial vessels, experts are required to assess and verify records of seabird interactions captured via camera footage to identify species to the lowest possible taxonomic level. This project aimed to assess the extent to which seabirds can be identified (i.e., taxonomic resolution) from camera footage, and to identify barriers to identification, such as in different fishing methods. The first set of images and videos were reviewed over three days in February 2025 by two seabird experts.
A total of 492 clips were provided for review. The first 42 events were reviewed independently by both experts to determine accuracy of the video footage and system, as well as consistency with seabird identification, and following this, the remaining footage was split by date, and each timeframe (each alternative day) was reviewed by one expert (with consultation where required).
Within these 492 clips, a total of 777 seabirds were recorded. An additional 15 clips were missing the interaction detail. Positive identification was confirmed in 97.0% (477) of all clips. Most birds were identified as being adults (97.8%; 760). A total of 26 species of seabird were recorded in these interactions. The larger number of interactions relative to the necropsy programme over the same timeframe reflects the broader coverage of the on board cameras programme.
Sometimes fishers presented seabirds to the cameras, which aided in identification. In addition, location (Fisheries Management Area) and date of capture information were used to assist with the identification of species, especially where visually similar species are expected in different locations. It is possible that further improvements in species identification may be possible if latitude and longitude of the captures are provided (as well as FMA), as well as if fishers were encouraged to present seabirds to the cameras.
Of the 777 seabirds reviewed, 42.2% had been identified to species level by the MPI camera review team, and of these 94.8% were identified correctly. The MPI camera review team also sorted 54.2% of the seabirds to the correct groups, but many could only be sorted into high level groups which limited the usefulness of the data for detailed analysis. Expert review of the camera imagery confirmed that 41.4% (322 seabirds) aligned to the species or species group that had been assigned by the MPI camera review team.
Of the 777 seabirds identified as part of this project, DOC had also provided identifications through the expedited expert identification process for 227 (29.2%) birds. There was a 76.7% (174/227) agreement to species level between the identification provided by the expert review and those provided by the DOC experts. For the events where there was disagreement between identification, 10.6% (24/227) was due to differing levels of taxonomic resolution.
The project demonstrated that the use of experts with experience in seabird identification can ensure rapid and accurate identification of seabirds from camera footage.
Publication information
Bell, E.; Bell, M. (2026). Identification of seabirds following review of footage from cameras on Aotearoa New Zealand commercial fishing vessels. New Zealand Aquatic Environment and Biodiversity Report No. 373. 17 p.