(5 of 7)
But that doesn't mean it has always oscillated, as it does today, roughly every two to seven years. Fairbanks, for one, thinks the present pattern may have switched on between 14,000 and 9,000 years ago, when rising sea levels swamped a landmass that included present-day Australia and Indonesia. Such a continent could have stabilized atmospheric pressure, keeping El Ninos from ever getting started. Whether this hypothesis is correct, no one yet knows. But the bits of coral Fairbanks and his team wrested from the submerged reefs around Christmas Island nicely bracket the period in question.
Trees too can preserve evidence of long-past climate patterns. David Stahle of the University of Arkansas Tree Ring Lab recently presented data derived from teaks in Java and firs in Mexico and the American Southwest that date back to 1706. The thicker the trees' growth rings, the more rain fell that year. According to Stahle, "it looks like a substantial shift occurred after 1880." After that date, the rainfall patterns typical of El Nino start to recur on average every 4.9 years instead of every 7.5, while patterns typical of La Nina show up at 4.2-year intervals versus once a decade.
An even longer-term perspective comes from paleogeologist Lonnie Thompson of Ohio State University, who specializes in extracting climate histories from mountain ice. Like trees and corals, ice grows in distinct layers whose thickness depends on the snowfall in a given year. Drilling into the Quelccaya ice cap in the Andes of Peru, Thompson has detected the short-term precipitation swings typical of El Nino and La Nina.
No one of these climate records is perfect. Storms in the Andes may hit or miss a particular location regardless of the overall weather pattern. Trees provide more comprehensive geographical coverage, but not always in the right locations. The palm trees that grow on tropical islands, for example, do not have rings that are useful for dating.
Out of all of this information, though, a crude picture begins to emerge. It appears the El Nino cycle is considerably more variable than scientists previously imagined, subject to protean swings of mood that last anywhere from decades to hundreds or even thousands of years. To account for this eccentric behavior, many scientists invoke the science of chaos, which says slight differences now--a barely perceptible increase in wind speed, for example--can lead to a dramatic change down the road. According to this scenario, the El Nino cycle resembles a chaotic pendulum whose swings never retrace the same path. Yet there is also a rhythm to the swings, like a jazzman's improvisations, endlessly circling a central theme. Chaos routinely pops up, in fact, when Lamont-Doherty climate modelers Stephen Zebiak and Mark Cane run computer simulations of the El Nino cycle. All the virtual El Ninos resemble one another, says Zebiak, but out of thousands of simulations, no two evolve in precisely the same way.