In set-net fisheries, it is difficult to assess the presence of fish schools within the net until the gear is hauled, which sometimes results in fishing operations that yield insufficient catch relative to the effort. This study aimed to classify the expected catch into two categories (whether the catch would justify a fishing operation or not) using echo images from fish finders installed on the set-net. We developed a convolutional neural network (CNN)-based estimation model constructed using multiple image selection methods, achieving a classification accuracy of 75–82％ for the fish-finder echo images. In addition, we conducted a detailed analysis of how the image cropping method affected classification accuracy and investigated the causes of misclassification cases.

　The amount of feed required for farmed fish is determined based on the biomass within the cage. While stereo cameras have been established as a practical tool for estimating body weight, no simple method for estimating the number of individuals has yet to be established. Recently, a method that combines a fish finder, simulation, and machine learning has been developed to estimate the number of farmed yellowtail. This method involves reproducing the farming cage, using virtually generated echo images, to train a machine learning model to associate fish finder echo images with the number of individuals, and then applying this model to estimate the number of individuals from real fish finder echo images obtained from the farming cage. The advantage of this method is that it uses a commercially available fish finder instead of specialized observation equipment. However, to date, this method had only been applied to yellowtail in small cages, and it had not been verified whether it could be applied to different cage sizes or species. In this study, we applied this method to Pacific bluefin tuna Thunnus orientalis reared in large cages. We tested the accuracy of estimating the individual number by varying the installation positions of the fish finder in four net cages, each containing 814 to 957 Pacific bluefin tuna, and found that it could be estimated with an error margin of within 5％.

　Predation by the sea stars Asterias amurensis and Distolasterias nipon is a major threat to sea ranching of Japanese scallops (Mizuhopecten yessoensis) on bottom-seeded grounds in northern Japan, yet species-specific quantitative impacts remain poorly understood. We analyzed multi-year rearing experiments with hierarchical Bayesian models to estimate how temperature, substrate, body size, and the densities of scallops and sea stars affect predation by each species. Predation by both species declined steeply with increasing scallop size, consistent with size-based physical defenses, whereas predation by D. nipon increased strongly with scallop density; temperature increased predation rates in both species. Field-scale simulations parameterized by the models indicated that approximately 60％ of cumulative predation occurs in year 1 and peaks in the warm season. To maintain at least 90％ survival after 3 years, sea star densities must be kept below 0.012 (A. amurensis) and 0.005 (D. nipon) individuals m^－2. Seeding larger scallops increases survival against both species, whereas high seeding densities increase the predicted number of scallops consumed where D. nipon is abundant, providing quantitative targets for species-specific sea star control and seeding design.

　In the fresh processing of scallops, reducing the labor required for manual shucking remains a major challenge. This study examined the relationship between shell-opening status and vitality indices of scallops in processing plants and evaluated the effectiveness of shell-opening induction using magnesium salt. No significant differences in vitality indicators were observed between opened and closed individuals. In contrast, immersion in magnesium salt solutions increased the shell-opening rate in a concentration-, temperature-, and time-dependent manner and reduced the time required for shucking. Furthermore, the treatment did not significantly affect the physicochemical or sensory characteristics of the adductor muscle. These results suggest that magnesium salt treatment enables more efficient shucking while maintaining product quality through optimized shucking procedures.

　This study investigated the characteristics of fishy odor formation in chub mackerel Scomber japonicus and spotted mackerel Scomber australasicus and examined the effects of three cooking methods (grilling, frying, and boiling) and ginger powder addition on odor compound generation. Thawed samples were minced and cooked under each condition. Volatile compounds were analyzed by GC–MS using a solid-phase microextraction (SPME) fiber. Lipids were extracted using the Bligh and Dyer method, and fatty acid composition was determined by GC after methyl esterification. Spotted mackerel produced higher levels of aldehydes associated with fishy odor than chub mackerel, due to differences in lipid content and fatty acid composition. Fried resulted in the lowest levels of odor compounds among the cooking methods. Ginger powder addition suppressed odor compound formation immediately after cooking and after 3 days of storage at 5°C. These results suggest that fried and ginger addition are effective for reducing fishy odor in cooked mackerel.