MEPS 506:145-161 (2014)  -  DOI: https://doi.org/10.3354/meps10816

ABSTRACT: The niche of many deep-sea species remains poorly resolved despite decades of seafloor exploration. Without better information on the distribution and habitat preference of key species, a complete understanding of the ecology of deep-sea communities will remain unattainable. It is increasingly apparent that cold-water corals are among the dominant foundation species in the deep sea, providing both structurally complex habitat and significant ecosystem services. In this study, the niche and distribution of the cold-water coral Lophelia pertusa in the Gulf of Mexico was evaluated using the maximum entropy (Maxent) approach. Ecological niche models were constructed for a broad region of the northern Gulf of Mexico using data gridded at a spatial resolution of 25 m, including bathymetry, substrate type, export productivity, and aragonite saturation state at depth. Fine-scale models were constructed at a resolution of 5 m using only remotely sensed bathymetric and surface reflectivity data. The broad-scale model performed well, with an area under the curve (AUC) of 0.981. All fine-scale models performed well when verified using training data (average AUC of 0.963) and when validated using independent occurrence data from a new geographic region (average AUC of 0.937). The distribution of L. pertusa in the Gulf of Mexico was found to be controlled primarily by depth, local topography, and availability of hard substrate. While these factors have long been associated with the success of cold-water corals, their relative importance has never been quantified in the Gulf of Mexico, making it historically difficult to precisely delineate L. pertusa’s niche and predict its distribution in unexplored regions. Given these results, we suggest that future expeditions combine remotely sensed data with niche modelling techniques to increase the efficiency of deep-sea exploration.

KEY WORDS: Bathymetry · Cold-water corals · Gulf of Mexico · Lophelia pertusa · Niche modelling · Maxent