Species distribution modelling and kernel density analysis of benthic ecologically and biologically significant areas (EBSAs) and other benthic fauna in the Maritimes Region / by L. Beazley, E. Kenchington,...: Fs97-6/3204E-PDF

"We present random forest species distribution models and kernel density estimation (KDE)-derived significant areas for benthic Ecologically and Biologically Significant Areas (EBSAs) and other benthic taxa in Fisheries and Oceans Canada’s Maritimes Region: Horse Mussel (Modiolus modiolus) Beds, Stalked Tunicate Fields, Sand Dollar Beds, Soft Coral Gardens, and Flabellum cup corals. Using a suite of sixty-six environmental variables derived from various sources and native spatial resolutions, random forest was employed to predict the probability of occurrence and biomass distribution of these taxa using data collected from DFO multispecies trawl surveys, DFO scallop stock assessment surveys, and targeted in situ benthic camera and/or video surveys. Random forest presence-absence models had excellent predictive capacity, with cross-validated Area Under the Receiver Operating Characteristic Curve (AUC) values ranging from 0.868 to 0.965. Areas of suitable habitat generated from the presence-absence models were compared to the KDE-derived significant area polygons with relatively good congruence. Such a process could be formally conducted in the future to refine the outer boundaries of these significant areas. Significant concentrations (as defined by KDE) of these taxa are considered conservation priorities in the Scotian Shelf Bioregional Marine Protected Area (MPA) Network Design Strategy, and are currently being used in exploratory conservation planning analyses"--Abstract, p. v.

Permanent link to this Catalogue record:
publications.gc.ca/pub?id=9.833084&sl=0

• MARC XML format
• MARC HTML format

Request alternate formats