Description: This dataset depicts the predicted likelihood of suitable habitat for deep-sea corals in the order Pennatulacea in the U.S. Northeast Atlantic and Mid-Atlantic as coral habitat suitability likelihood classes derived from a categorical reclassification of the logistic output of a maximum entropy (MaxEnt) model derived from coral presence locations and environmental predictor variables at 370.65m resolution. The reclassification used calibrated thresholds based on different hypothetical scenarios of the false positive:false negative cost ratio, that is, the relative cost of false positive versus false negative errors. Thresholds were chosen to minimize error rates when tested on data left out of the model-fitting process (cross-validation datasets), after weighting false positive and false negative error rates by the indicatd cost ratio. The thresholds were then used to create coral habitat suitability likelihood classes depicting several categories of predicted likelihood of habitat suitability. Coral habitat suitability likelihood classes can be qualitatively described as follows: Low Likelihood (Category 0, FP:FN cost ratio 1:1 but less than 2:1), Medium Likelihood (Category 2, FP:FN cost ratio > 2:1 but less than 5:1), High Likelihood (Category 5, FP:FN cost ratio > 5:1 but less than 10:1), Very High Likelihood (Category 10, FP:FN cost ratio > 10:1). MaxEnt is a type of statistical model. The outputs depicted here represent statistical model predictions, and as such may vary greatly in their accuracy, and may or may not reflect actual conditions at any given location. This information is provided as-is for research and evaluation purposes.
Copyright Text: Brian Kinlan, Matthew Poti, Dan Dorfman, Chris Caldow (NOAA/NOS/NCCOS/CCMA Biogeography Branch); Amy Drohan, Dave Packer, Martha Nizinski (NOAA/NMFS Northeast Fisheries Science Center)