Abstract Summary
A major obstacle for reef conservation is the identification of stress tolerant corals and habitats at large scales. We present a novel method to overcome this limitation, detecting tolerant corals using prior information on the natural bleaching response of individual colonies and airborne imaging spectroscopy. The Carnegie Airborne Observatory (CAO) collected high-fidelity spectral data for all hardbottom habitats within Kaneohe Bay, Oahu, Hawaii, an area of approximately 45km2. These data were analyzed via machine learning with prior bleaching observations for Montipora capitata and Porites compressa to classify bleaching phenotype at the coral colony scale. This strategy can discriminate bleaching phenotype using imaging spectroscopy data, even for colonies that show no visual signs of current bleaching or paling. Spectral data can also resolve chlorophyll content and Symbiodinium counts at the organismal level. The capacity to accurately predict species-specific response to thermal stress highlights spatial patterns at landscape scales, including clustering of stress tolerant corals on some reefs and the distribution and margins of species aggregates. These patterns show the importance of the interaction between habitat and the coral holobiont for successful conservation. The capability to detect phenotype from spectral data could be scaled to include entire reef ecosystems or down to hand-held, in situ instruments to achieve similar predictive capabilities where prior data or future monitoring can provide biological context.
