A new Galápagos coral reconstruction found a major industrial-era increase in eastern Pacific ENSO variability. That does not mean every El Niño, storm or drought is now 36% worse. The study is good. The slogan is the problem.
A new study in Science, led by Julia Cole at the University of Michigan, reconstructs eastern Pacific ENSO behaviour across the last millennium from Galápagos corals. It reports that temperature variability in that region has risen about 36.5% above the preindustrial baseline, with the sharpest acceleration in the past four decades.
It is a serious piece of work, and the headlines have already stretched it. Within days the number was circulating as "El Niño Has Become 36% More Extreme" and "Climate Change Has Made El Niño 36 Percent Stronger." The paper itself is titled Recent strengthening of eastern Pacific ENSO in the last millennium paleorecord, and the quantity it reconstructs is, in the authors' words, "interannual variability of sea surface temperature in the eastern equatorial Pacific."
The gap between those two framings is where this article lives.
Coral skeletons record sea surface temperature as they grow. Stack enough overlapping colonies, living and dead, and you get a continuous temperature series reaching back centuries before thermometers existed. Cole's team assembled 28 separate series from five Galápagos islands, with some corals extending roughly 900 years.
The Galápagos sit in the far eastern equatorial Pacific, in the Niño 1+2 region. That location is the study's strength and its boundary: it is the best possible place to measure eastern Pacific ENSO, and it is not a measurement of the Pacific as a whole.
The team then compared the modern era against 12 preindustrial climate model simulations, effectively asking how often a climate without human greenhouse warming could produce what the corals show. The models struggled to reproduce it. That is the finding that makes this publishable, and it deserves to be taken seriously.
A change in variability is not a multiplier on every event. The authors attribute the modern increase primarily to stronger El Niño events, so "stronger El Niños" is not wholly wrong. The abstract states it directly: "This increase parallels the rise in global temperature and results from stronger El Niño events." What is wrong is interpreting 36.5% as meaning every El Niño is now 36.5% warmer, stronger or more damaging. The statistic compares variability across two multidecadal periods, not individual events.
One precision note for anyone citing the figure. The abstract describes the quantity as interannual variability of sea surface temperature but carries no percentage; the 36.5% appears in reporting and in the body of the work. We checked the authors' open preprint and could not establish from it whether that percentage describes variance, standard deviation or another variability calculation. The distinction changes what the number implies, so it is worth confirming against the paper before repeating it.
ENSO is not only El Niño. The quantity reconstructed is ENSO variability, which includes La Niña. Reporting on the study indicates the increase is driven mainly by El Niño behaviour rather than La Niña changes, which is a meaningful result, but the headline statistic is not an El Niño-only measurement.
Eastern Pacific is not everywhere. This is the constraint the coverage dropped entirely. The finding is explicitly about the eastern Pacific. Coral records from the Line Islands in the central Pacific show a similar average increase but with considerably more scatter in the data, which is a weaker result. An eastern Pacific finding is genuinely relevant to the event now developing, which is strongly east-weighted, and it says much less about central Pacific events like 2015-16.
Here is the detail that did not survive the trip into the headlines. The 36.5% is measured against a preindustrial window running from 1000 to 1850 CE. Against the early instrumental period of 1851 to 1982, the reported increase is 16.2%.
| Comparison baseline | Reported increase in variability |
|---|---|
| Preindustrial, 1000 to 1850 CE | 36.5% |
| Early instrumental, 1851 to 1982 | 16.2% |
Both are real results from the same study and they answer different questions. The first asks how today compares with a world before industrial greenhouse forcing. The second asks how today compares with an early modern comparison period, much of which people alive today have partly lived through.
The headline took the larger number, which is a defensible editorial choice on its own. What is not defensible is reporting it without the baseline attached. "36% stronger" with no reference period is not a claim anyone can check.
There is a live methodological debate here, and it would be dishonest to write this piece without it.
In April 2026, Dolman and colleagues published in Communications Earth & Environment a finding that individual coral strontium-to-calcium and oxygen-isotope records can exaggerate reconstructed temperature variance by a factor of two to seven, equivalent to standard deviations 1.4 to 2.6 times too high. The proposed cause is non-climate noise from vital effects, the physiology of how corals incorporate these elements as growth rates and symbionts change.
That sounds devastating. It is not, and here is why we are not using it as one.
The Dolman result concerns individual records and focuses on decadal variability. ENSO variance is interannual, a different band. Using 28 series should reduce some record-specific noise, although the degree to which it resolves systematic coral-proxy biases depends on how the records and their uncertainties were combined. The appropriate reading is that proxy variance reconstructions carry a known inflation risk that must be handled, not that this one is wrong.
But it does mean a coral-derived variance increase is a harder measurement than a thermometer reading, and the honest version of the story says so. Two papers in the same year, one finding unprecedented modern ENSO variance from corals and one finding that corals can overstate variance, is not a scandal. It is what an active field looks like.
The tempting move right now is to connect this study to California's emergency proclamation and conclude that the coming winter will be 36% worse. That inference fails at every step.
A millennium-scale change in the statistical distribution of eastern Pacific temperature says nothing about which event you get in a particular year, and nothing whatever about rainfall. CPC's own caution applies with full force: the strength of an event does not necessarily correspond with the strength of its impact. Nothing in a coral record changes where the subtropical jet sets up in January.
What the study does support is a background statement: the distribution this event is drawn from appears to have shifted, and the eastern Pacific end of it has shifted most. Given that the developing event is unusually east-weighted, with Niño 1+2 near +3.9°C while Niño 4 sits at zero, that is a relevant piece of context. It is context, not a forecast, and not a damage multiplier.
Would strengthen the finding: independent reconstructions from other eastern Pacific sites reproducing the increase at similar magnitude, and improved climate models that generate the observed variance change under historical forcing rather than failing to.
Would weaken it: a demonstration that the vital-effect noise identified by Dolman and colleagues operates at interannual as well as decadal timescales at a magnitude large enough to account for the trend, or central Pacific records that fail to reproduce any increase once their scatter is handled properly.
What will not settle it: the 2026-27 event. A single El Niño, however large, is one draw from a distribution. It cannot confirm or refute a claim about the distribution itself, and anyone who tells you this winter proved or disproved the paper has misunderstood what the paper claims.
The paper may well have identified a real industrial-era change in the Pacific. But 36.5% describes a reconstructed climate pattern in one region, measured against a thousand-year baseline. It is not the strength of any individual El Niño, and it is not the damage from any disaster that carries the name.