The Northeast Regional Ocean Council and the Mid-Atlantic Regional Council on the Ocean (MARCO) recently collaborated to fund the development of maps of marine life to support ocean planning and management. Researchers at several institutions who work collaboratively as the Marine-life Data and Analysis Team (MDAT) assembled a collection of new maps that represents one of the largest known efforts globally to assemble and disseminate spatial data for multiple species and taxa of marine life. The Marine Geospatial Ecology Lab at Duke University led the overall MDAT effort and developed marine mammal maps for the entire Atlantic coast. The methods used to develop the marine mammal maps were published recently in the journal Nature Scientific Reports. The MarineCadastre.gov team has selected a subset of the fish layers to include in the National Viewer

Marine mammal individual species products

Mammal density model results are the predicted number of animals per 100 square kilometers for a specific temporal period. Source data used to create the models are from 1992 through 2014. Model resolution is 10km x 10km and models were generated with an original extent of approximately the entire US east coast EEZ plus a portion of southeast Canada. Species-specific seasons are based on patterns in the sightings and reports in the literature. Full documentation for every individual model can be accessed online.

Characterizations of model uncertainty

Several measures of model uncertainty are provided with each habitat-based density model. The percentile maps reflect the statistical uncertainty of the Generalized Additive Models that are predicting density from environmental predictors. The uncertainty at a given location relates mainly to how well the environmental conditions that occurred there were surveyed (via remote sensing), and how variable conditions are throughout the year.

5% Confidence Interval - This measure indicates that the density of animals predicted by the model exceeds what is shown on the map 95% of the time.

95% Confidence Interval - On the 95th percentile map, the density of animals predicted by the model exceeds what is shown on the map only 5% of the time.

Standard Error - Standard error estimates how close the estimated density is likely to be to the actual density, accounting for the number of sightings that were made, the modeled taxon, and how effectively density was modeled statistically from the environmental variables. The units of standard error are the same as density. The standard error estimate does not account for the uncertainty in either the detection functions (which model the probability of detecting the taxon given its distance from the survey trackline) or the estimates of availability or perception bias (the tendencies to fail to detect the animal because it is submerged and unavailable for observation, or because it displays cryptic behaviors, is small and hard to see, etc.)

Coefficient of variation - The CV is the ratio of the standard error to the estimated density, and helps inform users about the magnitude of variation in model predictions from one place to another. Values greater than 1, i.e. where the standard error is greater than the density estimate, indicate substantial uncertainty. When high CVs occur where the density estimate is very low, as is often the case, there is little cause for concern. But when high CVs occur where the density estimate is high, it suggests the model cannot predict density well there.

More information


Any use of the data should be accompanied by the following citations:

Roberts JJ, Best BD, Mannocci L, Fujioka E, Halpin PN, Palka DL, Garrison LP, Mullin KD, Cole TVN, Khan CB, McLellan WM, Pabst DA, Lockhart GG (2016) Habitat-based cetacean density models for the U.S. Atlantic and Gulf of Mexico. Scientific Reports 6: 22615. doi: 10.1038/srep22615.

Curtice, C., Cleary J., Shumchenia E., Halpin P.N. 2016. Marine-life Data and Analysis Team (MDAT) technical report on the methods and development of marine-life data to support regional ocean planning and management. Prepared on behalf of the Marine-life Data and Analysis Team (MDAT). Accessed at: http://seamap.env.duke.edu/models/MDAT/MDAT-Technical-Report-v1_1.pdf


Marine Geospatial Ecology Lab (MGEL) at Duke University. This product was developed by MGEL in collaboration with colleagues at the National Oceanographic and Atmospheric Administration (NOAA) National Marine Fisheries Service (NMFS), the University of North Carolina, Wilmington (UNCW), and the Virginia Aquarium & Marine Science Center (VAMSC). It was derived from habitat-based density models for marine mammals built from shipboard and aerial line transect surveys conducted at sea between 1992-2014 by the NMFS Northeast and Southeast Fisheries Science Centers, UNCW, VAMSC, and the New Jersey Department of Environmental Protection. The UNCW surveys were funded by U.S. Navy Fleet Forces Command and NOAA. The VAMSC surveys were funded by the Virginia Coastal Zone Management Program at the Department of Environmental Quality through Task 1 of Grant NA12NOS4190027 and Task 95.02 of Grant NA13NOS4190135 from NOAA, under the Coastal Zone Management Act of 1972, as amended. The density models were initially developed with funding from the National Aeronautics and Space Administration and U.S. Navy Fleet Forces Command, and further elaborated with funding from the Northeast Regional Ocean Council. For more information, please contact Jason Roberts (Duke University, jason.roberts@duke.edu).