Flowers bloom and buds pop to herald the arrival of spring, but it is much harder to mark the natural start of autumn. The spectacular color changes in fall foliage take place gradually and vary geographically. Ecologists struggle to model the timing of current autumn seasons, let alone forecast onsets a century from now. But achieving the latter goal could enable predictions about seasonal shifts expected to have an effect on future climate.

The customary approach to predicting autumn's start date relies on two variables: temperature and day length. Trevor Keenan of Macquarie University in Sydney, Australia, and Andrew Richardson of Harvard University knew the model was too simple, however, because ecologists frequently obtain poor results from it. So they analyzed approximately 20 years of tree foliage observations from the Harvard Forest in Massachusetts and the Hubbard Brook Experimental Forest in New Hampshire in combination with 13 years of foliage satellite data for the eastern U.S. The data revealed a new trigger for fall's arrival: the timing of spring. The analysis, published online in Global Change Biology, found that if spring began one day earlier for a particular tree, defined as when leaves emerged, then autumn arrived 0.6 day earlier on average for that same tree, defined as when leaves changed color. “It's quite an unexpected finding,” says Boston University ecologist Richard Primack, who was not involved in the study.

Keenan and Richardson do not know why autumn's arrival would depend on the onset of the preceding spring. “What's actually happening underneath the hood—the processes that initiate autumn—are quite complicated and not well understood,” Keenan says. Leaves may be programmed to drop once they reach a certain age, causing their senescence to shift earlier in the year if spring arrives sooner. Or perhaps an early spring means trees siphon more groundwater from soil, which could stress the water supply later in the growing season and kill leaves prematurely.

The spring-autumn link most likely is bad news for humans. Under the old model, ecologists had predicted that a century from now, autumn would start two weeks later given a seven degrees Celsius warmer climate—an outcome that would mitigate global warming because a longer summer allows trees to capture more carbon dioxide from the atmosphere. But Keenan and Richardson's model indicates spring's earlier arrival on a warmer planet will tug fall's start date forward. In such a scenario, autumn would be delayed only a few days, so trees would not capture much more carbon.

This new thinking is hardly the final word on autumnal alterations, but the study paves a route for future research. “This paper is going to stimulate a lot of interest,” Primack says. “People all over the world will read it and immediately go back to their own data sets and start reanalyzing them.”