This project aims to use recent advancements in machine learning to improve current mechanistic, ecosystem-based modeling approaches, benefiting managers and stakeholders alike

The National Oceanic and Atmospheric Administration awarded the University of Hawaii at Manoa a $238,015 cooperative agreement on August 1, 2026, under the Sea Grant Support program (CFDA 11.417) to develop machine learning modeling frameworks for Hawaiian reef fisheries and ecosystem management.

The recipient will create, train, and test universal differential equations models to address three management objectives: assess potential impacts of coral loss on Hawaiian reef fish assemblages, characterize constraints that reef fish assemblages exert on benthic communities, and forecast combined effects of rising temperatures and increased fishing pressure on reef ecosystems. The project will deliver three ready-to-use modeling frameworks designed to serve as templates for broader application across the Pacific and to validate the effectiveness of universal differential equations in ecosystem and fisheries management contexts.

Work will be performed in Honolulu, Hawaii, with a period of performance from August 1, 2026, through July 31, 2029. No sub-recipients are included in this award.

Activities

The primary activities to be performed for this project will be the development of a series of models to achieve three objectives relevant to the management of hawaiian reef fisheries and ecosystems: 1) assess the potential impacts of a loss in coral on hawaiian reef fish assemblages, 2) characterize the putative constraints that hawaiian reef fish assemblages exert on the benthic community, and 3) forecast the combined effects of rising temperatures and increased fishing pressure on reef ecosystems as a whole. This will include creating, training, and testing universal differential equations models for hawaiian reef ecosystems and fisheries.

Expected outcomes

The expected outcome of this project is to provide three ready-to-use modeling frameworks that will serve as the basis for future efforts to model hawaiian reef ecosystems and fisheries. These models can also be used as templates for broader use across the pacific. In addition, this project will serve as a comprehensive test for the effectiveness of universal differential equations when applied to a management context, potentially opening the door for broader implementation of machine learning methods to address ecosystem and fisheries management questions.

Intended beneficiaries

No subrecipients are included in this proposal.