<p>Coral topography is recognized as an important reef asset, yet reliable methods to characterize reef structure are needed to better evaluate the relation of physical attributes to ecosystem function, integrity and services. One opportunity arises from a long-term demographic coral assessment program in the Florida Reef Tract that documented the species, number, size and partial mortality of individual colonies from 2005 to 2020. The survey recorded 146,557 stony coral colonies in 6,016 10-m² transects. This robust dataset was previously employed to examine the physical attributes of individual colonies including density, height, footprint, volume, and surface area. However, relationships were developed for each attribute independently, not as an aggregate that more realistically describes community reef structure. Here, Bayesian network analysis was performed to compare the same nine input variables simultaneously and identify groupings (clusters) of transects with similar coral colony attributes. This resulted in five distinct clusters that incorporated all 6016 transects. Posterior probabilities were used to characterize attributes of individual clusters and optimization scenarios determined the sensitivity of site characteristics (subregion, zone, depth) to cluster assignment. Clustering affords an opportunity to relate a composite of reef physical variables to ecosystem functions, integrity and services. The clusters characterized here illustrate spatial and habitat patterns, identify priority areas for management protections, and set a baseline to document future changes to reef structure.</p>