There is a need to improve the efficiency and reduce the environmental footprint of northern shrimp (Pandalus borealis) fisheries. Northern shrimp are mainly harvested using demersal trawls, which are fuel-intensive and can impact the seabed. Although shrimp primarily occur near the seabed, they are known to migrate vertically at night, suggesting that pelagic trawling may be a viable alternative if shrimp can be acoustically detected in the water column at commercially relevant densities. To develop acoustic methods for shrimp detection and identification, reliable knowledge of target strength (TS), which describes the acoustic backscatter from an individual target, is needed. However, available TS data for shrimp are limited. In this study, we measured the frequency- (34–380 kHz) and tilt-angle-dependent TS of tethered shrimp. The measured TS generally increased with frequency and was highest at broadside incidence. The measurements were compared with predictions from DWBA and FEM–BEM scattering models using the same prescribed shrimp geometry and assumed homogeneous material properties. TS generally decreased away from broadside incidence, while frequency-dependent variability became more pronounced. Overall, there was good agreement between modelled and measured TS at broadside incidence, but poorer agreement away from broadside incidence. At off-broadside incidence, refining the spatial discretisation of the DWBA model substantially improved its agreement with FEM–BEM predictions. However, both models predicted a stronger reduction in TS away from broadside than observed in the measurements. Improved TS measurements and scattering modelling are therefore needed to support acoustic detection of shrimp and the development of more efficient, lower-impact harvesting methods.
The use of Artificial Intelligence (AI) to manage bycatch and discards is a growing interest in fisheries. However, implementing AI in fisheries faces challenges despite its promised benefits. This study evaluates the applications, challenges, and opportunities of adopting AI to reduce bycatch and discards using the Technology-Organization-External business environment (TOE) framework. We reviewed literature to identify projects using AI to manage bycatch and discards, conducted 17 in-depth interviews with fishery stakeholders, and supplemented these insights by attending two fishery innovation-related events. We identified several use cases of AI in managing bycatch and discards, including real-time species identification during fishing and catch registration, on-board automated sorting of species via conveyor belt and robotics, integration of AI with smart fishing gear, environmental-DNA sampling, and data-driven decision making in fishing. Several challenges hinder the adoption of AI-based tools in fisheries. First, most use-cases are at developmental, pilot, or prototype stage, suggesting the technology is not yet fully mature. Second, there is limited acceptance of AI-based tools among fishers due to concerns about trust, privacy, security, lack of incentives, and human replacement. Third, the need for high-quality big data and the complexity of modelling ecological systems make AI-models questionable. This study highlights three recommendations: (1) cross-sector collaboration to promote trust, support science-based policies and decisions, and make AI-based tools more feasible for fishers; (2) combining economic instruments on bycatch and discards with AI adoption to make it more attractive to fishers; and (3) using data management technologies to create trust in sharing data.
In fish dietary analysis, reducing the substantial time and effort required is essential for accelerating the accumulation of ecological knowledge. Recent advances in machine learning have enabled development of highly accurate object detection models. Here, we aimed to establish a foundational technology for automating stomach content analysis by developing an automated prey detection system based on YOLO, a general-purpose object detection model. Stomach contents from demersal fish collected by bottom trawling were photographed using a stereomicroscope, and a training dataset comprising 682 images (1655 instances) across 18 classes was constructed. The developed model successfully detected prey items across a wide range of digestion states, including highly degraded and amorphous prey that are difficult to identify visually. Despite the challenging nature of this task, the class-average precision, recall, and class assignment error rate were 41.3% ± 8.4%, 37.2% ± 7.6%, and 34.1% ± 7.2%, respectively. Among all classes, precision was highest for the ‘Fish(eye)’ class at 88.3%, whereas recall was highest for ‘Decapoda(shrimp)’ at 70.5%. In addition to these conventional metrics, we introduced a novel metric, Identification to Higher-level Taxon (IdHT), to account for taxonomic-level inconsistencies common in dietary analyses. The class-average IdHT was 19.5% ± 5.3%. Inter-class confusion occurred frequently, particularly among crustacean classes, likely due to texture bias inherent in convolutional neural networks. This study represents the first application of deep learning–based object detection to dietary analysis across multiple prey taxa, providing a foundational framework for efficiently conducting large-scale deep-sea food-web studies in the future.
Seabird bycatch mitigation in longline fisheries has been a focus of sustainable fisheries research for decades. One strategy is to increase the sink rate of hooks to reduce the window of opportunity for seabirds to take baits and become hooked. However, this requires an understanding of the different forces on hooks and branchlines when setting gear. Using vessels in Australia’s Eastern Tuna and Billfish Fishery, we explored four experimental trials assessing how gear and setting modifications affect sink rates of hooks and hook depth at the 75 m extent of tori lines. We examined sink rates of hooks with novel gear modifications including Procella hooks, Lumo leads, and Hookpods. Second, we explored how mainline tension interacts with current direction to influence hook sink rate. We also examined the sink rate of hooks in relation to port- versus starboard-deployed branchlines setting across different current directions. Finally, to inform live baiting practices, we explored whether nose- versus dorsal-hooked live bait influenced hook sink rate. We found sink rates of hooks were fastest i) setting across currents with a slack line, ii) set on the port side of the vessel on a port current, and iii) for live baits hooked by the nose. Differences in hook sink rates were also observed between vessels, potentially reflecting different operational conditions and vessel configurations. Across all four trials, none of the treatments achieved the controlled-condition sink rate of 0.5 m/s at 5 m depth outlined in the 2024 Agreement on the Conservation of Albatrosses and Petrels review, highlighting the difficulty of achieving this benchmark under active fishing conditions. Moreover, all trials were shallower than a dive-zone threshold of 5 m depth within the 75 m maximal extent of the tori line. Beyond this zone, baited hooks are left exposed to diving species like petrels and shearwaters.
