2026 American Samoa ROV + Mapping Exploration

Much of the deep Pacific remains difficult to observe directly. Deep-sea corals occur hundreds to thousands of meters below the surface, often on steep slopes, seamounts, ridges, and other complex seafloor features. These habitats provide three-dimensional structure for fishes and invertebrates, but ship time and remotely operated vehicle (ROV) surveys can cover only a small fraction of this vast region, meaning that our understanding of where deep-sea corals are found based on imaging alone remains limited.

Deep-sea corals provide habitat for a wide variety of organisms. This squat lobster was observed living in the branches of a coral during Dive 01 of the 2026 American Samoa ROV + Mapping Exploration expedition. Image courtesy of NOAA Ocean Exploration, 2026 American Samoa ROV + Mapping Exploration. Download largest version (jpg, 593 KB).

Scientists at NOAA Fisheries’ Pacific Islands Fisheries Science Center are using species distribution models to help fill those gaps. The work is part of the Pacific Islands Deep-Sea Ecosystems of Sponges and Corals: Exploration, Investigation and Technology, or DESCENT, Initiative. One of DESCENT’s core goals is to develop broad-scale distribution and habitat-suitability models that improve understanding of deep-sea coral and sponge communities and support ecosystem-based management.

How Species Distribution Models Work 

In building and applying species distribution models, we start by combining observations collected over decades from multiple sources, including NOAA’s national deep-sea coral and sponge records, the Global Biodiversity Information Facility, the Ocean Biodiversity Information System, and the University of Hawaiʻi’s Deep-sea Animal Research Center. For roughly 30 deep-sea coral species, we match known occurrence locations with environmental information describing conditions such as depth, temperature, salinity, dissolved oxygen, nutrients, chlorophyll, and ocean currents.

Within this dataset, the species distribution model looks for patterns to estimate where each species is most likely to live and overlays those patterns. The models allow scientists to estimate where each coral species is likely to occur by relating observed locations to the environmental conditions the coral was experiencing at the time it was observed . These species-level predictions can highlight broader patterns in expected coral richness and identify potential biodiversity hotspots. The resulting maps are predictions rather than direct inventories of the seafloor and are intended to complement field observations. The predictions provide researchers with limited ROV time on the seafloor information needed to target areas to further investigate.

Preliminary results suggest clear spatial patterns for the Main Hawaiian Islands. Higher modeled coral richness follows many slopes, banks, and deep features surrounding the islands, while smaller areas emerge as the highest-ranking hotspots. In the map shown here, highlighted cells represent areas with the highest modeled hotspot values, with the most concentrated areas identified as “core” hotspots. These areas provide precise target zones for upcoming seafloor expeditions.

Predicted deep-sea coral species richness and biodiversity hotspots across the Main Hawaiian Islands. Expected richness was estimated by combining species-level habitat suitability predictions from species distribution models. Cooler-to-lighter colors indicate increasing expected richness, while highlighted cells identify areas with the highest modeled richness, including core hotspots. The dashed contour indicates the 3,000-m depth limit. These predictions represent modeled patterns of potential deep-sea coral diversity rather than direct observations and can help identify priority areas for future surveys and model validation. Image courtesy of Tanaka & Lumsden, In prep. Download largest version (jpg, 466 KB).

The connection between modeling and exploration is central to DESCENT. The models can help scientists prioritize where new observations would be most informative, identify under-sampled areas, and evaluate how environmental conditions shape deep-sea coral distributions. The American Samoa map below shows modeled coral richness and hotspot areas across the archipelago. As additional records and validation data become available, the models will continue to be refined. Data collected during the 2026 American Samoa ROV + Mapping Exploration expedition will be integrated into these models, helping us to better understand the distribution of deep-sea corals in this region.

Predicted deep-sea coral species richness and biodiversity hotspots across American Samoa. Expected richness was estimated by combining species-level habitat suitability predictions from species distribution models. Cooler-to-lighter colors indicate increasing expected richness, while orange and red cells identify the global top 5% and top 1% hotspot classes. The dashed contour indicates the 3,000-meter (1.86-mile) depth limit. These predictions represent modeled patterns of potential deep-sea coral diversity rather than direct observations. Data collected during the 2026 American Samoa ROV + Mapping Exploration expedition will be integrated into these models as they are refined. Image courtesy of Tanaka & Lumsden, In prep. Download largest version (jpg, 317 KB).

In a region as large and underexplored as the Pacific Islands, predictive modeling provides a way to turn decades of scattered observations into a more complete spatial picture, helping direct future exploration toward places where deep-sea coral diversity may be greatest.

By Kisei Tanaka and Beth Lumsden, NOAA Fisheries’ Pacific Islands Fisheries Science Center