Developing an AI-powered microbiome decision framework to predict compatible microbial inocula and improve seagrass restoration reliability off the U.S. East Coast.
Seagrass off the east coast of the United States
Team & Partners: Dr. Geoffrey Zahn & Talia Backman (William & Mary) & Dr. Ashley Bulseco (UNH), Trevor Mattera (Piscataqua Region Estuaries Partnership), Dr. Christopher Patrick (Virginia Institute of Marine Science), Great Bay National Estuarine Research Reserve, Virginia Coastal Resilience Collaborative, The Nature Conservancy, SeqCoast Genomics, LLC
Challenge: Eelgrass restoration often fails because introduced microbial inocula are incompatible with resident sediment microbiomes, a barrier that practitioners currently have no tools to predict.
Approach: Conduct experiments crossing donor and resident sediment microbiomes from NH and VA, train graph neural network (GNN) models to predict inoculum establishment and host performance, then validate model-guided decisions at active restoration sites in both states.
Anticipated outcomes: A validated, open-source decision-support framework for forecasting microbiome-restoration compatibility, paired with built-in practitioner adoption pathways, an archived microbiome biobank, and proof-of-concept field validation.
