The effectiveness of the semi‐automated Identifin software to accurately fast‐track dorsal fin photographic identifications of sharks

Abstract

Abstract For some marine species, photo‐identification (photo‐ID) data can be used to generate population estimates critical for conservation efforts. For the white shark ( Carcharadon carcharias ), different photo‐ID methods have been proposed: (1) the traditional manual matching method based on non‐organized databases, (2) the notches code method based on organized databases and (3) the semi‐automatic software, Identifin. However, the effectiveness of these different methods and the applicability to other shark species have not been adequately explored. Eleven inexperienced and two experienced investigators used the three different photo‐ID methods to record the accuracy and time taken to identify individual white sharks from photo‐ID images. Both the traditional and notches code methods allowed for the accurate matching of individuals to the databases provided. For all investigators, the notches code method used significantly less time (14 min ± 0.14 min) than the traditional matching method (19 min ± 1.34 min). The semi‐automated method, Identifin, further shortened the identification time to 2 min (± 0.10 min) and provided a significantly higher matching accuracy (96% correct matches) compared with traditional manual matching (78% correct matches) and the notches code method (70% correct matches). The use of Identifin was tested to identify bronze whaler sharks ( Carcharhinus brachyurus ), a species with finer notches patterns when compared with white sharks. Identifin showed less efficiency in matching individual photographs to the database, but when very‐high‐quality (VHQ) photographs were available, matching improved substantially. Identifin software shows great potential in assisting researchers to accurately identify individual sharks in large databases, and this can contribute significantly to the conservation of shark species. The software and automated process need to be adapted to be functional on a variety of different shark species, but it proves to be an effective tool in managing large photo‐ID databases.