Integration of multiple data types for improved coral reef monitoring

The impending coral reef crisis demands time-sensitive, effective monitoring and early warning techniques, yet most current systems and methods rely on one or two isolated data types for coral reef health assessments. The integration of multiple data types can provide a more complete picture about reef community and biodiversity changes—for example, passive acoustics can detect and inform about cryptic species, while visual surveys provide context for non-soniferous species. We assessed three reefs of varying coral cover and fish communities, plus a nearby seagrass site, off St. John, US Virgin Islands with multiple integrated data types—passive acoustics (including band levels, automatically detected fish calls, and unsupervised clustering of call types), dissolved oxygen, temperature, microorganisms, nutrients, and visual fish and benthic surveys. The predictability of these integrated data on the reefs is examined on daily and monthly scales, in association with atmospheric variability, and compared to the seagrass bed. Initial evidence suggests that trends in acoustics and oxygen vary by site, although the stability of each is affected by storms and animal behavior—such as single-species chorusing events. Our findings aim to inform multi-modal monitoring systems for coral reefs and identify the most effective data types and time scales.