A 125-million-year-old volcanic mystery — and the fifteen-year scientific relay, across two generations, that finally solved it.
Dingshan Deng, Sanzhong Li*, Xianzhi Cao*, Nicolas Flament, Yanhui Suo, Liming Dai, Heda Wang, Qiuling Liu, & LiangLiang Wang
Published in Nature Geoscience
About 125 million years ago, the floor of the Pacific Ocean caught fire. Over a few tens of millions of years, the eruptions paved an area roughly the size of Australia — nearly six million square kilometers, with fresh volcanic crust. The plateaus they left behind, places like Ontong-Java and Shatsky Rise, remain the largest volcanic structures on Earth (Figure 1). The event was so vast that the rest of the planet moved with it: a global sea-level high, a major ocean anoxic event, and Earth's longest interval of stable magnetic polarity all cluster in the same narrow window. Yet for decades, no one could agree on what had lit the match.
Figure 1 | The crime scene. Present-day topography of the central Pacific, with the major Cretaceous Large Igneous Provinces (Ontong-Java, Manihiki and others) outlined. Together they cover an area roughly the size of Australia, and most were built in a single ~30-million-year pulse during the Early Cretaceous.
01 A question born at sea
Our story does not begin with a model. It begins on a ship. In 2009, while preparing the drilling proposal that would become IODP Expedition 324 to Shatsky Rise, our group's leader, Professor Sanzhong Li, framed a question that would set the group's course for the next fifteen years. The basaltic plateaus of the western Pacific erupted in a feverish pulse between 145 Ma and 125 Ma. Yet thousands of kilometers away, on the eastern margin of Eurasia, magmas of an entirely different temperament — explosive, granitic magmas broke out at almost exactly the same time, in the event Chinese geologists call the Yanshanian movement. Two great outbursts, oceanic and continental, mafic and felsic, at opposite ends of one plate system, in the same geological instant. Could they share a single, deep origin?
When Professor Li sailed with Expedition 324 in 2009, the question had no answer — the conditions did not yet exist. No one at Ocean University of China (OUC) was then running global mantle-convection codes or reconstructing the lost plates. The question was set aside as a seed, waiting for the right instruments.
02 When the tools caught up
What followed was the deliberate building of an armoury. In 2014, Professor Li opened a collaboration with Professor Dietmar Müller's group at the University of Sydney — a world centre for quantitative plate reconstruction, home of the GPlates software, and a wellspring of the mantle-flow modelling expertise. Over the next decade, he sent students and postdocs abroad to master the crucial skills: accurate global plate reconstruction and deep mantle flow modelling.
In parallel, Professor Li built a team at OUC devoted to these challenges, pioneering conceptual frameworks like the "microplate" tectonics theory — a perspective that views downwelling materials (whether subducted oceanic slabs or delaminated continental lithospheric mantle) as dynamic "mantle microplates" interacting within the deep Earth. By 2022, Professor Xianzhi Cao had returned from Sydney; together, they began training a new team fluent in both plate reconstruction and geodynamic modelling.
Therefore, when Xianzhi pointed to a previously published curve showing that the global subduction flux — the rate at which Earth's cold, downwelling slabs descend into the mantle — had peaked sharply between 130 and 120 million years ago, it was not a lucky accident. It was, rather, the deduction fifteen years of preparation had been waiting to make: if the cold, sinking limb of mantle convection had truly peaked then, its hot, rising limb should have peaked with it somewhere in the deep Pacific — and the Cretaceous volcanic pulse should be its surface signature. The question first raised at sea was ready, at last, for a four-dimensional Earth.
03 Ten months in the dark
The torch now passed to a new generation of Ph.D. students, trained deliberately for this moment. We built five global mantle-convection models — varying starting times, basal densities, and the convective vigor. They all told the same story: between roughly 130 and 120 million years ago, the heat carried upward by the Pacific's hot upwellings spiked, in lockstep with the subduction flux measured at the surface (Figure 2). Five models, one verdict. The correlation was beautiful.
Figure 2 | The curves that finally lined up. Through the Mesozoic, the circum-Pacific subduction flux, the dynamic pressure difference between the subduction girdle and the central-Pacific LLSVPs (magenta dashed), and the heat carried upward by the Pacific's hot upwellings (coloured lines) all peak together near 125 Ma — exactly when the histogram of Pacific LIP eruptions spikes. Removing the slabs, or the deep hot piles, flattens that peak.
But correlation is not mechanism. For the next ten months, we could not find the physical bridge between those two curves. Why should slabs sinking at the rim of the Pacific make upwellings rise faster in its centre? Ideas looked promising at lunch and were dead by dinner. For Dingshan — then a second-year master's student — these were months of restless nights and a quiet fear that the work might never come together in time, months when he had to talk himself into opening his computer each morning.
"Five models, one verdict — and yet we still could not say why. That gap, between a beautiful correlation and a convincible mechanism, is where most of the real work in our field lives."
The breakthrough came from two directions. First, we looked at the dynamic pressure field of the deep mantle. Beneath the ring of subduction zones encircling the Pacific, the fast-sinking slabs were quietly building a region of high pressure; beneath the central Pacific, the continent-sized hot pile known as the LLSVPs sat in low pressure. The mid-Cretaceous surge in subduction acted like a syringe — pressing deep material inward and then upward, forcing the hot piles into unusually vigorous outflow that fed the plumes igniting the Pacific (Figure 3). Second, we needed to connect these deep plumes to the surface. It was here that Dr. Nicolas Flament provided a crucial insight. He suggested a method to quantitatively analyze the interaction between plumes and mid-ocean ridges. We discovered that the ridges intersected the central Pacific upwellings exactly when hot upwellings were strongest, acting as the final trigger for the massive LIP eruptions. Neither cold slabs, superplumes, nor migrating ridges could light the fire alone — they needed each other.
Figure 3 | The mechanism, in motion (still frame at 120 Ma; the full animation between 165-65 Ma, Supplementary Animation 1 can be viewed through the web version of the original article). Plumes (warm colours) rise from the edges of the Pacific hot piles and feed an intense cluster of upwellings beneath the central Pacific; reconstructed LIPs (blue polygons) sit directly above the strongest of them. Arrows show converging flow in the lowermost mantle — the deflected, top-down push that quietly powered the bottom-up volcanic pulse.
04 The lost oceans, and the next chapter
So who set the Pacific on fire? Sinking slabs did — but only because a hot, yielding pile of ancient mantle was waiting to receive them, because mid-ocean ridges were positioned to extract the melt, and because a fifteen-year-old question had been patient enough to wait for the tools to answer it. The Cretaceous pulse was the surface signature of a vigorous episode of whole-mantle convection, where the planet's top and bottom behaved as a single, coupled machine.
And the machine has not stopped. We are now turning the same toolbox back toward the very region where Professor Li's question began: Mesozoic-Cenozoic East Asia. With a particle-by-particle tracking module, we are following the lost oceanic plates — Izanagi, and the early Pacific — into today's mantle beneath the East Asian margin, asking how those sunken slabs still shape the surface above. From a question raised on a ship in 2009 to the answers a new generation is writing now, this is less a single paper than a fifteen-year relay across two scientific generations. The next lap has already begun.
Article: Deng, D., Li, S., Cao, X. et al. Vigorous mantle convection triggered the Cretaceous Pacific large igneous provinces. Nat. Geosci. (2026). https://doi.org/10.1038/s41561-026-02016-y