The ocean's significant warming in recent decades has led to more frequent, intense, and prolonged marine heatwaves (MHWs). These extreme warming events have devastating consequences for marine ecosystems and fisheries, exceeding the impacts of gradual mean warming. Understanding future MHW changes is crucial for developing effective mitigation strategies. Large Marine Ecosystems (LMEs), covering ~22% of the global ocean and responsible for 95% of global fisheries catches, are particularly vulnerable. However, previous studies using low-resolution climate models, which have inherent biases in coastal regions where LMEs are located, have limitations. This study aims to address this gap by employing a high-resolution Earth system model to project future MHW changes over LMEs, considering both a 'mean warming-inclusive threshold' and a 'future threshold' to isolate the effects of background warming and higher-order temperature statistics.

Satellite and field observations confirm that MHWs have intensified and increased in frequency and extent in recent decades. Regional and global model simulations project further intensification and increased incidence under future warming scenarios. Studies using a mean warming-inclusive threshold (defining MHWs relative to the historical mean climate) project almost permanent heatwaves by the end of the 21st century under high-emission scenarios. However, using a shifting baseline (future threshold) to isolate the impact of mean warming is beneficial. Low-resolution models struggle to accurately reproduce MHW characteristics, particularly in coastal regions, because they fail to capture small-scale processes like ocean eddies and coastal upwelling. High-resolution regional models show better fidelity in simulating observed MHW events. Existing research primarily focuses on global MHW characteristics, neglecting the crucial details of coastal regions and their impact on LMEs.

This study utilizes a high-resolution Earth System Model (CESM1.3) simulation running from 1850 to 2100, forced by historical forcings before 2005 and RCP 8.5 (high-emission scenario) thereafter. This high-resolution configuration (CESM-HR) resolves mesoscale eddies, improving the simulation of MHWs compared to low-resolution models. For comparison, results from a low-resolution version of CESM (CESM-LR) and the CMIP5 multi-model ensemble are also used. MHWs are defined using two thresholds: a mean warming-inclusive threshold (90th percentile of daily SST over a 30-year period) and a future threshold (90th percentile of daily SST in the future period). The study focuses on LMEs, which are categorized into six groups based on their continental location. The frequency, duration, and intensity of MHWs are analyzed for both historical and future periods, using NOAA Optimum Interpolation Sea Surface Temperature (OISST) and Group for High Resolution Sea Surface Temperature Multi-product Ensemble (GMPE) data for validation. The impact of mean warming on MHW changes is assessed using a pseudo-scenario, in which the future mean SST warming is added to historical daily SSTs.

The high-resolution CESM-HR model shows a better representation of historical MHW frequency and intensity compared to lower-resolution models (CESM-LR and CMIP5), especially in LMEs. Using the mean warming-inclusive threshold, CESM-HR projects substantial increases in MHW days and intensity by the end of the 21st century under RCP 8.5. Many areas will experience near-permanent MHWs. The increase in MHW days is more pronounced in the equatorial and subtropical regions than the North Atlantic and western boundary currents (WBCs). When using the future threshold, the projected increases in MHW days and intensity are substantially smaller, highlighting the dominant role of mean warming. However, even with the future threshold, most LMEs still experience increased MHW days and intensity. The increase in MHW intensity is correlated with changes in SST variance, particularly in WBC regions, where a meridional dipole pattern is observed with stronger increases poleward. LMEs with higher fishing capacity show larger increases in MHW days and intensity than those with lower capacity, indicating potentially more significant impacts on high-production areas. The strong correlation between historical and future MHW intensities (using the future threshold) suggests that LMEs currently under stress will remain so in the future, with the added pressure of mean warming and increased intensity.

The study's findings address the research question by demonstrating the significant threat posed by future MHWs to LMEs, even under the optimistic assumption of complete adaptation to mean warming. The use of a high-resolution model is crucial for capturing realistic MHW characteristics, particularly in coastal regions where LMEs are predominantly located. The contrasting results obtained using mean warming-inclusive and future thresholds highlight the importance of considering both mean warming and higher-order temperature variability in assessing future MHW impacts. The strong correlation between historical and future MHW intensity (under future threshold) underscores the continued vulnerability of already stressed LMEs. The findings have broad implications for fisheries management, conservation efforts, and the development of adaptation strategies. The results emphasize the disproportionate impacts of climate change on LMEs with high economic value.

This study demonstrates that increased MHW intensity and frequency are projected for most LMEs, even when accounting for potential adaptation to mean warming. High-resolution modeling is essential for accurate prediction of these changes. The findings highlight substantial risks to LMEs, calling for urgent adaptation and mitigation strategies. Future research should focus on expanding high-resolution multi-model ensembles under diverse climate scenarios to improve uncertainty assessment and provide timely warnings.

The study relies on a single high-resolution model (CESM-HR) and a specific emissions scenario (RCP 8.5). While this model shows improved accuracy over lower-resolution models, the results should be considered in the context of model uncertainties. The study focuses primarily on regions within 70° N/S, potentially limiting the generalizability of findings to higher latitudes. The impact of interactions with other stressors (e.g., ocean acidification, pollution) is not explicitly considered.