Thirteen Galápagos corals put recent El Niños about 40% above their preindustrial strength

The Galápagos is the right place to take this measurement, and the reason is signal size. Earlier fossil-coral reconstructions came from the western and central Pacific ^[10]; Cole's team worked in the eastern tropical Pacific, which she describes as the heart of where El Niño has some of its strongest impacts ^[11]. In an ordinary event the water there runs one to two degrees C above normal, and above two degrees in a strong year ^[15]. A proxy sited where the swing is largest has the easiest job.

The chemistry is two thermometers rather than one. Corals build calcium carbonate skeletons and substitute strontium for calcium more readily when the water is cooler; separately, the ratio of oxygen isotopes in the skeleton varies with temperature ^[8]. Measuring both is the internal control, since an artifact in one system is unlikely to be mimicked by the other. Resolution follows from growth rate: the colonies add one to two centimetres a year ^[5] and the team sampled every millimetre ^[6], which comes to roughly ten to twenty samples per year ^[23], fine enough to catch a seasonal peak. That matters, because intensity is a peak measurement.

Now the limit. Thirteen cores ^[4], each chosen to span at least twenty years because extremes arrive only every few years ^[7], gives at least 260 coral-years of record; set against a thousand-year interval that is 26 percent even if no two cores overlap in time ^[24]. Cole calls them windows into points in the past ^[27]. The claim the design supports is correspondingly precise: no sampled window resembles the last forty to fifty years ^[22], which is how the team can speak about the preindustrial period with confidence for the first time ^[14].

What will get argued is the attribution, so it is worth separating which model does what. Simulations have given mixed signals on whether warming should strengthen El Niño at all ^[9], which is precisely why a coral record was worth years of beach-combing. The same class of tool then supplies the exclusion: last-millennium runs forced with volcanic eruption and solar variability histories show no comparable change ^[12]. Those two jobs are not equally hard. Asking whether known natural forcings can manufacture an amplitude jump of the observed size is bounded; asking a model to get ENSO's greenhouse response right is not. The exclusion is the more defensible half of the argument, and it is the half doing the work here.

The mechanism remains open. Cole points to ocean stratification, in which a warmer and rainier surface layer becomes less dense, more isolated from the cool water beneath, and more responsive when the trade winds slacken, while allowing that how warming strengthens the cycle is not yet clear ^[16].

The figure is amplitude at one location, and it doesn't speak to frequency. The published accounts still describe events arriving every two to seven years ^[17] without reporting a change in that spacing. Teleconnection strength is a separate matter that a Galápagos temperature series leaves untouched; the southward shift in US storm tracks, with a wetter desert Southwest and a drier Northwest ^[21], follows from the tropical signal through a chain this record does not close by itself.

My read: eastern Pacific El Niño amplitude now belongs in the observed-trend column, with two conditions attached. The trend is documented in the size of the extremes, not in their count, and the headline forecasts quoted alongside the study, three degrees above normal near the islands ^[15] and a global anomaly of 1.7 to 1.8 degrees C ^[20], are forecasts rather than paleorecord.