The triple-dip La Niña that gripped the Pacific from 2020 to 2023 was no fluke. It was the latest expression of a trend 7,000 years in the making.

New evidence combining ancient coral records with climate simulations reveals that multi-year El Niño–Southern Oscillation events—those lasting two consecutive years or longer—have increased dramatically since the mid-Holocene. The ratio of multi-year to single-year events climbed from roughly one-in-ten to nearly one-in-two. A fivefold increase.

The finding matters because multi-year events don't just last longer. They hit harder. When an El Niño or La Niña stretches across multiple years, it compounds damage to agriculture, intensifies droughts or floods, and leaves entire regions swinging from one climate extreme to another. Australia has endured persistent wet conditions during extended La Niña phases. Parts of the southern United States, equatorial Africa, and southeast China have faced prolonged dry spells.

Understanding why these events became more frequent requires looking beneath the ocean's surface.

Reading the coral archive

The research team analyzed oxygen isotope signatures preserved in fossil corals from Kiritimati and Fanning Atolls in the central equatorial Pacific. These corals recorded monthly climate variations across more than 1,200 years of scattered intervals throughout the Holocene epoch, which began approximately 11,700 years ago.

Oxygen isotopes in coral skeletons respond to both sea surface temperature and salinity. During El Niño events, the central Pacific warms and receives more rainfall, leaving a distinct chemical fingerprint. La Niña events produce the opposite signal. The monthly resolution of these records allowed researchers to trace individual events from initiation through peak to decay.

From this archive, they identified 87 multi-year El Niño events and 108 multi-year La Niña events, alongside hundreds of single-year cases. Each multi-year event showed a characteristic pattern: after the first-year peak, conditions weakened briefly, then strengthened again to produce a second peak roughly twelve months later.

The corals also revealed something unexpected. Multi-year La Niña events consistently outnumbered multi-year El Niño events—a pattern confirmed by modern observations since the mid-19th century.

Models confirm the shift

To verify these findings and explore underlying mechanisms, the team turned to an ensemble of eight climate model simulations covering the Holocene. Four models were selected based on their ability to realistically reproduce ENSO periodicity and multi-year event frequency.

While these models underestimated the absolute number of consecutive events compared to coral data, they qualitatively reproduced the increasing trend. More importantly, they provided a complete picture of ocean and atmosphere conditions throughout the period—something corals alone cannot offer.

The models pointed to a specific culprit: changes in the tropical Pacific thermocline.

The thermocline connection

The thermocline is the boundary layer where ocean temperature drops sharply with depth. In the tropical Pacific, this layer sits roughly 150 meters below the surface and plays a central role in ENSO dynamics.

Analysis of the climate simulations revealed that the thermocline was deeper and warmer during the early-to-mid Holocene than today. This seemingly small difference had cascading effects on ENSO behavior.

A deeper thermocline weakens the so-called thermocline feedback—the mechanism by which wind-driven ocean adjustments amplify temperature anomalies at the surface. When the thermocline sits deeper, wind stress anomalies become less effective at tilting it and reinforcing existing El Niño or La Niña conditions.

The result? Shorter ENSO cycles. Faster transitions between warm and cold phases.

As the thermocline gradually shoaled over the past 7,000 years, ENSO periods lengthened from approximately 3.5 years to 4.1 years. This half-year extension created conditions favorable for events to persist into a second year rather than decaying during the spring "persistence barrier"—the season when ENSO events typically weaken and often terminate.

The researchers verified this relationship using a simplified mathematical model. By systematically reducing thermocline depth in the model, they demonstrated that ENSO periods lengthened proportionally. A 40% reduction in depth extended the cycle by roughly 21%.

Orbital forcing sets the stage

What drove the thermocline changes? Slow variations in Earth's orbit.

Orbital precession—the gradual shift in the timing of Earth's closest approach to the Sun—altered the seasonal distribution of incoming solar radiation. During the early-to-mid Holocene, the Southern Hemisphere received more sunlight during late winter and early spring.

This extra heating warmed surface waters in the subtropical South Pacific. Through a process called subduction, these warmer waters then propagated into the deeper layers of the tropical eastern Pacific thermocline. The warming was particularly pronounced from January through August, effectively deepening and weakening the vertical temperature gradient.

Surface cooling in the easternmost Pacific during parts of the year, combined with subsurface warming, reduced ocean stratification. The effect persisted throughout most of the calendar year, creating background conditions that shaped ENSO behavior on millennial timescales.

Three climate models with available subsurface data all showed consistent patterns of reduced vertical temperature gradients during the mid-Holocene compared to recent centuries.

Multiple mechanisms at play

While thermocline depth emerged as the primary driver, ENSO amplitude also played a contributing role.

Coral records show that ENSO variability reached a minimum around 5,000 years ago and gradually intensified thereafter. Stronger ENSO events create larger ocean heat content variations, which can take longer to dissipate through the slow recharge-discharge process that governs the cycle.

Statistical analysis revealed a moderate correlation between ENSO amplitude and the frequency of multi-year events. However, the connection between multi-year events and ENSO period proved stronger across both proxy data and model simulations.

The persistence barrier—the springtime weakening of ENSO events—also became less pronounced over the Holocene. Coral data showed a significant negative correlation between persistence barrier strength and multi-year event frequency. As the barrier weakened, events were more likely to survive into a second year.

All these factors appear to work in concert, with thermocline changes playing the lead role.

Implications for the future

The findings carry weight for climate projections. Historical observations already document increased probability of multi-year La Niña and El Niño events in recent decades. Five of six La Niña events since 1950 became multi-year cases. Five multi-year El Niño events occurred since 1950, with two developing in the past decade alone.

The Holocene trend suggests these recent events may be occurring against a background state already predisposed toward persistence. The upper ocean has continued to stratify—the thermocline has continued to shoal—creating conditions even more favorable for prolonged ENSO events than at any time in the past 7,000 years.

Anthropogenic factors now layer additional complexity onto this orbital forcing. Increased carbon dioxide concentrations are expected to further modify tropical Pacific ocean structure, potentially accelerating the trend toward more persistent ENSO impacts.

Climate models with realistic representations of CO2-induced ocean responses will be essential for robust projections. The Holocene record demonstrates that ENSO duration responds to slowly varying background conditions. Models that fail to capture these subtle ocean changes may underestimate future shifts in ENSO behavior.

The research also highlights a critical gap. While models generally captured the qualitative trend, they underestimated the magnitude of change seen in coral records. This suggests persistent biases in how models represent ENSO physics—particularly the feedbacks that control event duration and timing.

A long view

The Holocene has been relatively stable compared to glacial periods. Yet even within this stability, the seasonal redistribution of sunlight was enough to reorganize tropical Pacific ocean structure and fundamentally alter ENSO character.

The transition wasn't abrupt. No dramatic shift appears in the records. Instead, a steady accumulation—more multi-year events per century, fewer single-year cases, a gradual drift toward longer periods between peaks.

That gradual nature makes the change easy to miss in short observational windows. The instrumental record captures barely 170 years. Internal climate variability can mask or amplify trends over such brief intervals. Only by extending the view across millennia does the signal emerge clearly.

And it points unambiguously forward. The same ocean processes that responded to orbital forcing will respond to anthropogenic forcing. The thermocline will likely continue shoaling. Stratification will likely strengthen. Multi-year events will likely continue increasing in frequency.

Combined with projected increases in overall ENSO amplitude, this creates a scenario where extreme conditions persist longer and strike harder—precisely the combination the Holocene record warns against.

The work underscores an uncomfortable reality: the ENSO system has been shifting for thousands of years toward a state that favors prolonged impacts. Recent multi-year events aren't anomalies. They're the latest chapter in a story written in ocean heat and orbital mechanics, now accelerating under human influence.

Climate mitigation efforts take on added urgency when viewed through this lens. The long-term trend is already unfavorable. Additional anthropogenic forcing will compound a shift already well underway.

Credit & Disclaimer: This article is a popular science summary written to make peer-reviewed research accessible to a broad audience. All scientific facts, findings, and conclusions presented here are drawn directly and accurately from the original research paper. Readers are strongly encouraged to consult the full research article for complete data, methodologies, and scientific detail. The article can be accessed through https://doi.org/10.1038/s41561-025-01670-y