World Aquaculture December 2020

50 DECEMBER 2020 • WORLD AQUACULTURE • WWW.WA S.ORG from juvenile through harvestable size. Some of the species currently being investigated are primarily cobia, olive flounder, red snapper and Florida pompano. Furthermore, the UMAquaculture Program has been working closely with industry leaders such as Open Blue Sea Farms, Panama, and Martec S.A., Costa Rica, to improve feed efficiencies while maximizing fish performance. In addition to work with these groups, the United Soybean Board, the Illinois Soybean Association, and the US Soybean Export Council (USSEC) have been strong supporters of nutrition research at the UM over the years. These partnerships led to the development of nutritional requirements, apparent digestibility of different ingredients and the replacement of fishmeal and fish oil in diets for cobia. We are continuously seeking and routinely developing collaborations with feed companies in the US and abroad to conduct digestibility studies of the ingredients used in the formulation and manufacture of specialized diets for commercially important marine finfish species. Research and development collaborations have also been made with multiple partners (UJAT of Mexico and the US-based company Aquaculture Intelligence) to promote fish health through the use of probiotics, essential oils and organic acids. Through this collaboration, a repository of tropical fish enzymes of high commercial value has been set up to evaluate the enzymatic activity of commercial ingredients and diets. Additional nutrition research is underway to assess the nutritional profile and possible use of black soldier fly meal and oil for the manufacturing of commercial diets for marine fish and shrimp. Based on a preliminary review of studies, it appears that different meal and oil extraction methods will modulate the nutritional profile of the meal and oil. Additionally, black soldier fly meal seems to be comparable to plant-based meals rather than animal-based meals. Furthermore, proximate analysis of the meal can be altered depending on the food source of the fly larvae, usually reflecting the profile of the food source. In summary, nutrition is key at broodstock, larval, live feeds, fingerling, juvenile and grow-out stages. Performance studies are required for understanding the selected species’ nutritional requirements and digestibility of proteins, amino acids, lipids and energy of the various ingredients used for diets at all stages of the production cycle, while maximizing the use of high-quality ingredients and additives. These are crucial for aiding in the formulation of specific diets, reducing fish-in-fish-out ratios, reducing use of fishmeal and fish oil with alternative sources of animal and plant protein. Knowledge of the nutritional requirements and digestibility of nutrients of large fish is necessary to optimize growth and minimize waste in commercial marine finfish farms. This will be more rapidly attained with synergistic collaborations among all stakeholders in the productive chain, from suppliers of rawmaterials and aquafeed manufacturers to commercial farmers and research institutions. The overall goal it to improve the economic and ecological efficiency of farming operations and nutrition research by the UMAquaculture Program plays an important role in meeting this goal. Machine Learning and Artificial Intelligence In recent years, machine learning and artificial intelligence (AI) has been introduced into commercial aquaculture. AI has the potential to play a role in critical processes of the aquaculture production cycle, from spawning to grow-out stages, to improve efficiency and decrease operational costs. The UMAquaculture Program is engaged in research activities dedicated to innovative approaches to improve aquaculture hatchery technologies with state-of-the-art AI techniques. Current efforts focus on developing machine learning-based computer vision pipelines for autonomous rotifer ( Brachionus sp.) culture system (ARCS). The procedure is to use a systematic combination of object detection, object tracking, convolutional neural networks and sequential neural networks to empower the machine vision to accurately detect, recognize and classify rotifers under the microscope and estimate rotifer motility and ciliate concentration. The AI-driven computer Feeding Artemia to cobia larvae at UMEH production tanks. Passive transfer is used in larval production of all species at UMEH. Goggle-eye fry.