| description:
|
This data layer contains the draft output of the habitat suitability models for the taxon group Aphrocallistes vastus offshore California, Oregon, and Washington. Depicted is a “robust” very high likelihood prediction that indicates grid cells for which all the models fit to the bootstrap samples were classified as having a very high likelihood of containing suitable habitat.
NCCOS obtained the April 5, 2018 version of the NOAA Deep Sea Coral Research and Technology Program national database of deep-sea corals and sponges and extracted presence locations of deep-sea corals and sponges offshore California, Oregon, and Washington and shallower than the 1200m depth contour. The project science team identified 44 taxa of deep-sea coral and sponges with sufficient presence locations to develop habitat suitability models.
For each taxon, the R package ‘maxnet’ was used to fit models of habitat suitability using the values of the environmental covariates at locations of deep-sea coral or sponge occurrence. A stepwise model selection procedure was performed to select a final model for each taxon that balanced model predictive performance with model complexity. The set of environmental covariates in the final model was used to fit models to ten bootstrap samples to generate spatially explicit predictions of the relative likelihood of habitat suitability. Outputs from these predictions include the mean prediction across the models fit to the ten bootstrap samples, the coefficient of variation across the predictions, a classified prediction created by reclassifying the mean prediction into classes of increasing likelihood of habitat suitability, and a “robust” very high likelihood prediction that indicates grid cells for which all the models fit to the bootstrap samples were classified as having a very high likelihood of containing suitable habitat. |