Increasingly frequent severe coral bleaching is among the greatest threats to coral reefs posed by climate change. Global climate models (GCMs) project great spatial variation in the timing of annual severe bleaching (ASB) conditions; a point at which reefs are certain to change and recovery will be limited. However, previous model-resolution projections (~1 × 1°) are too coarse to inform conservation planning. To meet the need for higher-resolution projections, we generated statistically downscaled projections (4-km resolution) for all coral reefs; these projections reveal high local-scale variation in ASB. Timing of ASB varies >10 years in 71 of the 87 countries and territories with >500 km2 of reef area. Emissions scenario RCP4.5 represents lower emissions mid-century than will eventuate if pledges made following the 2015 Paris Climate Change Conference (COP21) become reality. These pledges do little to provide reefs with more time to adapt and acclimate prior to severe bleaching conditions occurring annually. RCP4.5 adds 11 years to the global average ASB timing when compared to RCP8.5; however, >75% of reefs still experience ASB before 2070 under RCP4.5. Coral reef futures clearly vary greatly among and within countries, indicating the projections warrant consideration in most reef areas during conservation and management planning.

Phytoplankton production drives marine ecosystem trophic-structure and global fisheries yields. Phytoplankton biomass is particularly influential near coral reef islands and atolls that span the oligotrophic tropical oceans. The paradoxical enhancement in phytoplankton near an island-reef ecosystem—Island Mass Effect (IME)—was first documented 60 years ago, yet much remains unknown about the prevalence and drivers of this ecologically important phenomenon. Here we provide the first basin-scale investigation of IME. We show that IME is a near-ubiquitous feature among a majority (91%) of coral reef ecosystems surveyed, creating near-island ‘hotspots' of phytoplankton biomass throughout the upper water column. Variations in IME strength are governed by geomorphic type (atoll vs island), bathymetric slope, reef area and local human impacts (for example, human-derived nutrient input). These ocean oases increase nearshore phytoplankton biomass by up to 86% over oceanic conditions, providing basal energetic resources to higher trophic levels that support subsistence-based human populations.

Human impacts can homogenize and simplify ecosystems, favoring communities that are no longer naturally coupled with (or reflective of) the background environmental regimes in which they are found. Such a process of biophysical decoupling has been explored little in the marine environment due to a lack of replication across the intact‐to‐degraded ecosystem spectrum. Coral reefs lacking local human impacts provide critical baseline scenarios in which to explore natural biophysical relationships, and provide a template against which to test for their human‐induced decoupling. Using 39 Pacific islands, 24 unpopulated (relatively free from local human impacts) and 15 populated (with local human impacts present), spanning 45° of latitude and 65° of longitude, we ask, what are ‘natural’ biophysical relationships on coral reefs and do we see evidence for their human‐induced decoupling? Estimates of the percent cover of benthic groups were related to multiple physical environmental drivers (sea surface temperature, irradiance, chlorophyll‐a, and wave energy) using mixed‐effects models and island mean condition as the unit of replication. Models across unpopulated islands had high explanatory power, identifying key physical environmental drivers of variations in benthic cover in the absence of local human impacts. These same models performed poorly and lost explanatory power when fitted anew to populated (human impacted) islands; biophysical decoupling was clearly evident. Furthermore, key biophysical relationships at populated islands (i.e. those relationships driving benthic variation across space in conjunction with chronic human impact) bore little resemblance to the baseline scenarios identified from unpopulated islands. Our results highlight the ability of local human impacts to decouple biophysical relationships in the marine environment and fundamentally restructure the natural rules of nature.

Coral reef ecosystems are exposed to a range of environmental forcings that vary on daily to decadal time scales and across spatial scales spanning from reefs to archipelagos. Environmental variability is a major determinant of reef ecosystem structure and function, including coral reef extent and growth rates, and the abundance, diversity, and morphology of reef organisms. Proper characterization of environmental forcings on coral reef ecosystems is critical if we are to understand the dynamics and implications of abiotic–biotic interactions on reef ecosystems. This study combines high-resolution bathymetric information with remotely sensed sea surface temperature, chlorophyll-a and irradiance data, and modeled wave data to quantify environmental forcings on coral reefs. We present a methodological approach to develop spatially constrained, island- and atoll-scale metrics that quantify climatological range limits and anomalous environmental forcings across U.S. Pacific coral reef ecosystems. Our results indicate considerable spatial heterogeneity in climatological ranges and anomalies across 41 islands and atolls, with emergent spatial patterns specific to each environmental forcing. For example, wave energy was greatest at northern latitudes and generally decreased with latitude. In contrast, chlorophyll-a was greatest at reef ecosystems proximate to the equator and northern-most locations, showing little synchrony with latitude. In addition, we find that the reef ecosystems with the highest chlorophyll-a concentrations; Jarvis, Howland, Baker, Palmyra and Kingman are each uninhabited and are characterized by high hard coral cover and large numbers of predatory fishes. Finally, we find that scaling environmental data to the spatial footprint of individual islands and atolls is more likely to capture local environmental forcings, as chlorophyll-a concentrations decreased at relatively short distances (>7 km) from 85% of our study locations. These metrics will help identify reef ecosystems most exposed to environmental stress as well as systems that may be more resistant or resilient to future climate change.