Patterns in the chaos: Scale and the spatiotemporal dynamics of coral reef fish assemblages on the Great Barrier Reef

In the Anthropocene, understanding and managing ecological communities requires the characterization of natural spatiotemporal ecosystem dynamics. Complex ecosystems may appear chaotic and unstructured, making long‐term monitoring programs with hierarchical sampling designs ideal for investigating ecological patterns at multiple scales. Here, we use a dataset spanning the entire Great Barrier Reef (GBR) and multiple decades to determine the spatial patterns of distribution and abundance, and how these change through time, in 233 reef‐associated fish species. Community composition was strongly structured by position across the continental shelf, with distinct inner and outer shelf assemblages. Latitudinal differences were smaller, except for the distinctive assemblages of the southernmost Swain and Capricorn‐Bunker regions. GBR‐wide summaries of total density and species richness did not show directional shifts, and it was only after analyzing these metrics at the subregional scale that a wider range of temporal oscillations was identified, indicating that responses of fish assemblages to perturbations require examination of smaller than GBR scales. Within most subregions (65%), fish species composition has undergone a clear and ongoing directional shift away from the community identified in the 1990s. These changes were generally due to a reduction in the numbers of coral‐dependent species and an increased dominance of grazers and generalists. Among the 233 species, there have been more “winners” than “losers” across the whole GBR, but this masks a tendency for inner shelf reefs of the central GBR to have a dominance of “losers” over time. Reef fish assemblages on the GBR are dynamic and show clear recovery potential from disturbance events. Despite some pervasive community‐level shifts in the last decades, the clear biogeographic characteristics of each subregion remain intact. We pose the question of whether it is reasonable to expect these highly dynamic assemblages to reach a relatively stable “climax community,” and posit that the answer is scale‐dependent and, on the GBR, is currently resolved at the subregional scale, which is the scale at which most stakeholders and decision‐makers operate.