Modeling distributions of cold-water corals in the Newfoundland and Labrador region using MaxEnt

Species distribution models (SDMs) are increasingly used in ecology and conservation to predict species distributions based on benthic habitat requirements. SDMs help when designing conservation approaches for regions lacking detailed data on species requiring protection; however, such approaches are limited by the quality of the training data, and by the modeling approaches used. In order to support the design of marine protected areas that can protect cold-water corals and sponges in the Newfoundland and Labrador region, Eastern Canada, we used MaxEnt to model the distributions of 19 coral species belonging to four functional groups: large gorgonians, small gorgonians, sea pens, and soft corals. Coral distribution data from 2004-2011, obtained from the Department of Fisheries and Oceans Canada (DFO) multi-species trawl surveys, were used to generate SDMs. These data encompass the south coast of Newfoundland, the Grand Banks, the Flemish Pass, Northeast Newfoundland Shelf, and Labrador shelf to 60N. Environmental datasets used to generate models included: GEBCO bathymetry, six terrain attributes derived from bathymetry, bottom temperature and salinity, and surface chlorophyll A concentrations. Modeled distributions predicted greatest coral abundance along the continental shelf break and upper slope throughout the region. When analyzed at the functional group level, large gorgonians were predicted to be most abundant off northern Labrador, off the central Labrador shelf, and in parts of the SW Grand Banks. Small gorgonian distributions were most concentrated along the Labrador Shelf break, the NE Newfoundland Shelf break, as well as along the shelf break of the SW Grand Banks. Sea pen distributions were greatest along the SW Grand Banks and the south coast, particularly within the Laurentian Channel. In contrast, soft corals were most abundant on the bank tops. Agreement between observed and modeled distributions was assessed using area under the curve (AUC), true skill statistics (TSS), and comparisons with independent datasets. Findings consistently illustrated high model accuracy (e.g. AUC >0.9). Analysis of predicted distributions at the species level found large differences among species within functional groups, particularly among the large gorgonians. Results indicate that coral functional groups based on taxonomy do not group species sharing similar environmental preferences, and that SDMs should instead be performed on individual species of interest. Furthermore, previous models for the region based upon Random Forest modeling have predicted sponge and coral habitats extend into deep water areas of the continental rise. Our study does not confirm those predictions, and limited knowledge of the bathyal and abyssal depths in these regions do not suggest appropriate bottom types for these species. In contrast, our results predict coral distributions are concentrated along the continental slope and shelf break, having a high level of agreement with the training dataset, and with predicted global distributions of the same taxa. Our models suggest that conservation efforts for corals should be concentrated at the shelf-break and upper continental slope, where they are known to occur, rather than in deep-water areas where they might occur, and where there are fewer competing human uses than on the shelf break and upper slope.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.