Fisheries can profoundly affect bycatch species with ‘slow’ life history traits. Managing bait type offers one tool to control species selectivity. Different species and sizes of marine predators have different prey, and hence bait, preferences. This preference is a function of a bait’s chemical, visual, acoustic and textural characteristics and size, and for seabirds the effect on hook sink rate is also important. We conducted a global meta-analysis of existing estimates of the relative risk of capture on different pelagic longline baits. We applied a Bayesian random effects meta-analytic regression modelling approach to estimate overall expected bait-specific catch rates. For blue shark and marine turtles, there were 34% (95% HDI: 4–59%) and 60% (95% HDI: 44–76%) significantly lower relative risks of capture on forage fish bait than squid bait, respectively. Overall estimates of bait-specific relative risk were not significantly different for seven other assessed taxa. The lack of a significant overall estimate of relative capture risk for pelagic shark species combined but significant effect for blue sharks suggests there is species-specific variability in bait-specific catch risk within this group. A qualitative literature review suggests that tunas and istiophorid billfishes may have higher catch rates on squid than fish bait, which conflicts with reducing marine turtle and blue shark catch rates. The findings from this synthesis of quantitative and qualitative evidence support identifying economically viable bycatch management measures with acceptable tradeoffs when multispecies conflicts are unavoidable, and highlight research priorities for global pelagic longline fisheries.

A real-time tuna detection network on mobile devices is a common tool for accurate tuna catch statistics. However, most object detection models have multiple parameters, and normal mobile devices have difficulties in satisfying real-time detection. Based on YOLOv3, this paper proposes a Tuna-YOLO, which is a lightweight object detection network for mobile devices. Firstly, following a comparison of the performance of various lightweight backbone networks, the MobileNet v3 was used as a backbone structure to reduce the number of parameters and calculations. Secondly, the SENET module was replaced with a CBAM attention module to further improve the feature extraction ability of tuna. Then, the knowledge distillation was used to make the Tuna-YOLO detect more accurate. We created a small dataset by deframing electronic surveillance video of fishing boats and labeled the data. After data annotation on the dataset, the K-means algorithm was used to get nine better anchor boxes on the basis of label information, which was used to improve the detection precision. In addition, we compared the detection performance of the Tuna-YOLO and three versions of YOLO v5-6.1 s/m/l after image enhancement. The results show that the Tuna-YOLO reduces the parameters of YOLOv3 from 234.74 MB to 88.45 MB, increases detection precision from 93.33% to 95.83%, and increases the calculation speed from 10.12 fps to 15.23 fps. The performance of the Tuna-YOLO is better than three versions of YOLO v5-6.1 s/m/l. Tuna-YOLO provides a basis for subsequent deployment of algorithms to mobile devices and real-time catch statistics.