A study of benthic habitat classification schemes in the northern equatorial Pacific based on a Gaussian mixture model

Due to the complexity and extreme environment of the deep sea, related research is limited by the level of detection technology, and the overall understanding of this topic is limited. At present, the increasing number of disturbances in the deep sea, such as those caused by deep-sea mining, necessitates the urgent formulation of scientifically informed deep-sea environmental management decisions. In response to this challenge, a data-driven, large-scale, and probability-based benthic habitat classification scheme is proposed for the northern equatorial Pacific. The dominant environmental factors were identified as topographic features (depth and slope), nutrients (particulate organic carbon flux entering the seabed and phosphate), and water mass properties (dissolved oxygen and pH), and the importance assessment indicated that nutrient played a dominant role, followed by water mass properties, with topographic features having a relatively minor influence. The multiple classification evaluation parameters determined 5 clusters of benthic habitats in the northern equatorial Pacific using the Gaussian mixture model (GMM). Furthermore, this paper provides the first discussion of ecotones at the edges of habitats and suggests that ecotones are more cost effective for environmental management. In conclusion, the benthic habitat classification scheme could provide a consistent unit for habitat mapping and representative assessment of marine protected area network design, as well as a scientific basis for the protection and management of the international seabed area.