Predictive modelling of mesophotic habitats in the northwestern Gulf of Mexico

Abstract Effective management of marine resources requires an understanding of the spatial distribution of biologically important communities. The northwestern Gulf of Mexico contains diverse marine ecosystems at a large range of depths and geographic settings. To better understand the distribution of these marine habitats across large geographic areas under consideration for marine sanctuary status, presenceonly predictive modelling was used. Results confirmed that local geographic characteristics can accurately predict the probability of occurrence for marine habitat types, and include a novel technique for assigning a single, most likely habitat in areas where multiple habitats are predicted. The highest resolution bathymetric data (10 m) available for the region was used to develop raster layers that represent characteristics that have been shown to influence species occurrence in other settings. A georeferenced historical photo record collected via remotely operated vehicle was classified according to six commonly found mesophotic habitats across the 18 reefs and banks under consideration for Flower Garden Banks National Marine Sanctuary boundary expansion. Using maximum entropy modelling, the influence of local geographic characteristics on the presence of these habitats was measured and a spatial probability distribution was developed for each habitat type across the study area.

Predictive modelling of mesophotic habitats in the northwestern Gulf of Mexico Thesis