Kernel Density Analysis and Mapping of Ecosystem Functions in the NAFO Regulatory Area

0

Inglés. In support of the 2021/2022 NAFO review of the closed areas to protect vulnerable marine ecosystems (VMEs) in the NAFO Regulatory Area, previously established kernel density estimation (KDE) methods were applied to four important ecological functions provided by benthic communities: A) Bioturbation; B) Nutrient cycling; C) Habitat provision; and D) Functional diversity (FRic), in order to evaluate significant adverse impacts of NAFO bottom-contact fishing on vulnerable marine ecosystems against the wider benthic contributions to those functions. Fish and invertebrate species recorded in the EU and Canadian surveys from 2011-2019 were classified a priori as contributing to each of bioturbation, nutrient cycling and habitat provision functions, using literature references. The resultant catch biomass data for each function were examined using K-S statistics and cumulative biomass distribution plots to determine whether data from the different surveys could be combined. With few exceptions the surveys were analyzed separately and the KDE polygons overlain a posteriori to produce combined polygon areas for each function. A suite of species were important contributors to the biomass of catches used to delineate each of the KDE polygons. For bioturbation, the sea cucumber Cucumaria frondosa and sea pens, both considered surficial modifiers, contributed most to the biomass. Nutrient cycling and habitat provision functions were delineated by catches where sponges dominated the biomass. Details of the analyses and the species that contributed to the delineation of the polygons are provided. Functional diversity was not assessed as information needed on a wide variety of traits and modalities was not completed. However published data from a survey in 2007 of Division 3M was used to run the KDE analyses with equivocal results. The KDE polygons generated matched published maps of FRic created using the same data and interpolated using random forest modeling. However the data were not sufficiently aggregated to allow for a clear KDE threshold to be determined. All other KDE analyses performed well and showed good congruence to the published maps of their corresponding functions.

Español; castellano. Versión del editor