For decades, coral reefs throughout the Caribbean have been suffering from disease, pollution, overfishing and rising sea temperatures, yet most have continued to grow – until now. In 2023 and 2024, surface temperatures climbed to record highs in the world’s oceans, and a marine heatwave of unprecedented length and intensity spread across the tropics . Satellites from the US National Oceanic and Atmospheric Administration detected heat stress that could cause corals to bleach across more than 80% of the planet’s reef areas . During these periods of extreme stress, corals expel the symbiotic algae that give them their colour and most of their food – turning them stark white and leaving them vulnerable to starvation, diseases and eventually death . Across the North Atlantic , including the Caribbean, the heat stayed for months, with heat stress two-to-three times higher than reefs had ever experienced. Heat stress, the phenomena of high temperatures putting fragile ecosystems under pressure, can permanently alter their ability to function. This triggered what is now recognised as the fourth global coral bleaching event, the most severe one that has been documented . Widespread coral bleaching during the 2023 marine heatwave. Coral reefs are among the most productive ecosystems on Earth, and their importance to people is fundamental. They feed hundreds of millions through small-scale fisheries, underpin tourism across the Caribbean, and serve as natural breakwaters that protect the coast from storms and reduce flooding events. Caribbean reefs are eroding fast In a new study , we found that across the Caribbean, the 2023 marine heatwave – combined with a deadly disease known as stony coral tissue loss disease – has pushed reefs over a threshold scientists thought was a decade or more away. They are now eroding faster than corals can rebuild them. We studied reefs in the Mexican Caribbean and the Gulf of Mexico, comparing data collected before the heatwave (2018–2022) with surveys after it (2023–24). At each reef, we counted live corals and organisms that break down the reef, like parrotfish and sea urchins. From those counts, we estimated how much reef-building (carbonate production) and reef-breaking (bioerosion) was happening, then calculated the net result – whether the reef was gaining or losing material. The results were stark: between 70% and 75% of our Caribbean sites had tipped from net growth into net erosion. They are now losing calcium carbonate faster than corals can add it. The threshold that earlier models had suggested might be crossed over during the next decade or so has already arrived . This shift was driven by the loss of fast‐growing, branching and plate‐forming corals, especially the Acropora species, which have very high growth rates and disproportionately contribute to reef building. One of our most unsettling findings is that the Caribbean reef sites that still had high coral cover and high carbonate production before the disease and heatwave were the ones that lost the most. Some lost up to 8 kilograms of calcium carbonate per square metre per year. A tale of two seas Our survey also revealed a striking contrast. While Caribbean reefs collapsed, reefs in the Gulf of Mexico largely held their ground. The great majority of Gulf sites remained net positive after the heatwave. The difference comes down to which corals are pre-eminent in each region. In the Gulf of Mexico, reefs are dominated by slow-growing, mound-shaped corals. They grow more slowly, but they are tougher when the heat kicks in. They bleached during the heatwave but mostly survived, keeping the reef’s carbonate budget positive. This is the balance between the constructing and eroding processes. When more is added than removed, the coral reef can grow. When that balance flips, the reef stops growing and may even erode. Moreover, sites in the Gulf of Mexico have not yet been affected by stony coral tissue loss disease, which preferentially kills the same massive, long-lived species that are keeping Gulf reefs alive. By the time the heat arrived, large parts of the Caribbean had already lost their most resilient corals because of the disease outbreak . What it started, the heatwave finished. Why reef erosion matters All the benefits reefs provide rely on a delicate balance between reef construction and erosion. Tropical reefs are essentially vast limestone structures, built slowly over centuries as corals deposit calcium carbonate skeletons. At the same time, waves and various reef organisms like parrotfish, sea urchins and boring sponges chip away at them. An eroding, flattening reef begins to lose its capacity to provide benefits to other species, and people. We did not expect to be documenting the moment at which a major region of the ocean crossed from growing to eroding. The fact that it happened this quickly, and at some of the most iconic and well-studied reefs in the Caribbean, suggests the timelines scientists have been using may be too optimistic. Main reef-builders in the Caribbean died as heat stress increased. Our findings may also force a rethink of how to approach coral restoration. Programmes across the Caribbean have invested heavily in replanting fast-growing branching species of coral, such as Acropora , because they rebuild structural complexity quickly. The 2023–24 heatwave wiped out many of these restored populations, along with wild ones. Restoration will have to diversify. Exploring approaches such as moving heat-tolerant genes between populations (assisted gene flow) and breeding corals that survive heat better (selective breeding) might be a promising path . But restoration alone will not be enough . Reversing the decline requires rapid cuts in greenhouse gas emissions to slow the frequency and intensity of marine heatwaves, alongside serious local action on pollution, nutrient runoff, sedimentation and disease – the stressors that weaken corals before the heat arrives. Chris Perry receives funding from the UK Natural Environment Research Council. Lorenzo Alvarez-Filip does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.