Study Helps Scientists and Managers Predict Fish and Coral Distributions
Researchers from NOAA’s National Centers for Coastal Ocean Science recently published a study on the utility of combining airborne LiDAR (light detection and ranging) bathymetry and in situ biological data for predicting fish and coral distributions across coral reef ecosystems off southwestern Puerto Rico. This study demonstrates the utility of LiDAR-derived bathymetry and spatial predictive modeling as a novel and cost-effective technique to improve our knowledge of faunal distributions over complex coral reef seascapes, to support ecosystem-based management and marine spatial planning. The techniques developed may also have broader applications for predicting the occurrence of deepwater corals and for predicting the geographic shifts of species under varying climate change scenarios.
Seafloor topographic complexity was quantified using seven surface metrics applied within a geographical information system (GIS) at multiple spatial scales. Predictive models for nineteen fish metrics and two coral metrics were developed using the new statistical-learning technique of stochastic gradient boosting applied to regression trees. The models explained high proportions of the variance in fish and coral biomass and species richness using predictors from a single remote sensing device, indicating that the technique will provide a cost-effective tool for conservation planning. The most significant environmental driver was the magnitude of slope change at relatively local spatial scales (15-100 m radii), with herbivorous fish responding to topographic complexity at spatial scales smaller than those for piscivorous fish. The study appears in a Fall 2009 special issue of the Journal of Coastal Research.