Coral reefs harbor high biodiversity and support the livelihoods of millions of people worldwide. Yet they, like numerous other ecosystems, can undergo large shifts in species composition to a less desirable state, such as a shift from coral dominance to macroalgal dominance. These shifts in the benthos can reverberate throughout the system, influencing the reef fishes that play an important role in the ecology of the ecosystem and in the economy of local communities. Previously, spatial heterogeneity and the arrangement of physical reef structure that reef organisms use for habitat has not been adequately considered in these dynamics, in part due to the difficulty in collecting the necessary spatial data in the subtidal marine environment. This research harnesses recent advances in computer vision and photogrammetry to address these knowledge gaps about the spatial ecology and benthic state shifts of coral reef ecosystems. The first aspect of this study involved the development of a novel methodology to conduct paired, georeferenced fish and benthic surveys in shallow coral reef systems. The increased efficiency of having one researcher conduct both surveys simultaneously allows much greater area to be surveyed, while maintaining high spatial resolution in the data. To process the large volume of data collected, benthic photographs were automatically annotated using the machine learning algorithm, CoralNet. At the scale of a one minute transect segment (roughly 10 m in length representing 50 m2 surveyed), the computer vision algorithm provided more accurate estimates of the percent cover of benthic organisms than expert human annotators due to its ability to process many more photographs per segment. Using this methodology, the fish and benthic communities the lagoons of Moorea, French Polynesia were surveyed in 2018 and different locations were surveyed again in 2019. The 2018 surveys were used to construct models to determine whether coral reef fishes were more x associated with static (e.g., location in the reef, depth, availability of physical structure) or dynamic (e.g., coral versus macroalgal cover) environmental variables. The performance of these models was tested using the new data collected in 2019, and the models maintained high predictive power. Most fish groups were influenced more by static compared to dynamic variables, suggesting that they may be somewhat resilient to a change in the reef state over short periods. Finally, the macroalga that is responsible for shifts in the benthic state in Moorea lagoons, Turbinaria ornata, is resistant to herbivory but has a low propagule dispersal distance, with most dispersal occurring < 1 m away from adults. To investigate whether the arrangement of physical habitat influences the growth of T. ornata, individual patches of T. ornata and the reef structure emerging from the sand (“bommies”) that they settle upon were mapped through time. Orthomosaics, large images created from overlapping many small images while preserving distances, were created in fifteen sites within the lagoon of Moorea in 2019 and again in 2021, and the amount of T. ornata on each bommie was recorded in both years. T. ornata remained highly stable through over two years, with little change in its cover over time, but there was evidence that most of the change in the area occupied by T. ornata was driven along edges of established patches, rather than through long distance dispersal across bommies. However, this study appeared to capture a time period when T. ornata switched from expanding to contracting across the lagoon, therefore a longer timescale (greater than two years) would be needed to fully understand these dynamics. Taken together, this research utilizes novel methodological advances to provide new insights into how coral reef ecosystems are structured spatially in a heterogenous environment and provides a foundation for further studies focused on the spatial dynamics of phase shifts in these systems. This can increase understanding and predictive ability as we try to manage coral reefs in a changing environment.