Oyster Hatchery Production and Machine Learning

The demand for reliable spat production from oyster hatcheries has risen in recent years with the onset of Chesapeake Bay oyster restoration and the rise of the Bay’s oyster aquaculture industry.

Oyster hatcheries are facilities where adult oysters are spawned, and spat is grown and later sold for use in the aquaculture industry, fisheries augmentation, or restoration efforts. Despite highly skilled and experienced staff running hatcheries, there are periods of poor larval growth and uneven production levels (also termed ‘crashes’). These crashes affect hatchery profitability and become a bottleneck in the supply chain for private oyster growers, consumers, and state entities that manage natural resources. In most cases, the causes of crashes and their potential remedies are unidentified.

This project is being performed by the University of Maryland Center for Environmental Sciences researchers Drs. Vyacheslav Lyubchich, Matthew Grey and Greg Silsbe. It aims to process large amounts of hatchery and environmental data to identify conditions leading to hatchery inefficiencies and strategies for mitigating their impacts on production. The hatcheries are heavily influenced by environmental conditions in the coastal zones where they are located. The water is taken from the environment and, after more or less thorough processing (such as filtering, heating, or cooling), is used for growing oyster larvae. This project proposes to use machine learning techniques to identify patterns in the data; between hatchery production outcomes and factors such as water quality, atmospheric conditions, and implementation of agricultural fertilizers and herbicides upstream from the hatchery. This extensive data analysis will help in finding the causes of the crashes and will allow managers to stay informed about the potential success of the production. With this knowledge, the managers will be able to focus on manipulating specific variables to improve production outcomes.

Read About More Research