Williams also found a strong correlation between water stress and the forested acreage killed by beetles and wildfire in the past 30 years. "Even if we think of a couple degrees of warming as relatively minor," Williams says, "forests notice a couple of degrees, and they express it by dying."

Because the atmosphere's sponginess is so strongly dictated by temperature, climate models can help predict how fast it will climb. "I considered a scenario where we begin curbing emissions significantly yesterday," Williams says. "Even in that most optimistic scenario, we're looking at megadrought conditions by the 2070s." In other words, even if we began to aggressively control carbon pollution tomorrow, the heat guaranteed by past and ongoing emissions could still devastate Southwestern conifers.

"By 2050, it doesn't matter if it's wet or dry, it's just too damn hot out," McDowell explains. The sense of inevitability that accompanied Williams' conclusions changed how McDowell views his work. At first, he was intrigued by the novel scientific questions involved in tree mortality. "Now I feel like I have a moral obligation to speak up," he says. "We're not just going to lose a bunch of trees, we're going to lose most of them in the Southwest. By 2050, we could be looking at Albuquerque vegetation in Los Alamos," a landscape now surrounded by forests. "Albuquerque has grass and creosote bush."

Such radical changes are unlikely to be confined to the Southwest. A newer modeling effort that Williams and McDowell participated in estimates that the Pacific Northwest could lose 60 percent of its conifers to heat-induced water stress by 2100 – an especially sobering finding for McDowell, whose love of forests was lit at an early age by the old Doug firs on Puget Sound and the Olympic Peninsula. "Can you imagine the Olympic Peninsula without trees?" he asks.

Such a future is hard to imagine. Many Western forests still look healthy, with plump, verdant canopies. But even some of the healthiest-looking stands may already be stressed.

In mid-July, U.S. Geological Survey forest ecologist Nate Stephenson drove me to a long-term forest-monitoring plot in Sequoia National Park, a few hours north of Los Angeles. As we left the shrubby foothills, where one could break a sweat standing still at 9 a.m., the temperature dropped 20 degrees, shadows painted the pavement, and giant sequoia appeared – the titans of the Sierra Nevada. The plot itself was blanketed with ferns, and full of soaring sequoias and lichen-covered sugar pines.

Stephenson helped establish the network in 1982, measuring off the first plots with string. He is wildly passionate about the Sierra Nevada: In graduate school, he designed a thesis project that allowed him to hike 500 "glorious" miles a summer in Sequoia's backcountry. After he earned his Ph.D., he returned to Sequoia with no promise of permanent employment. He wasn't interested in going where the jobs were. Stephenson has now studied this place for 34 years. But it can still surprise him. When he expanded the plot network across different elevations in the early '90s to study how climate affects forests, he says, "It didn't occur to me that by the mid-2000s, we would already be able to detect an increase in tree mortality."

Around that time, Phil van Mantgem, a scientist who worked in Stephenson's shop, began analyzing growth and mortality in the plots. He expected dull results – birth and death rates usually reach equilibrium in old growth – but something peculiar appeared in his data: Background mortality rates – the rate at which trees die in a healthy forest – had doubled. "We thought we did something wrong," Stephenson says. "We tried to make it go away. We couldn't." The only possible cause they couldn't eliminate was the average temperature, which had risen almost 2 degrees F since the 1980s.

Stephenson and van Mantgem ran the same analysis for old-growth forests West-wide. They found the same pattern: At many high-, mid- and low-elevation plots, from California to Idaho, Arizona and Colorado –– even in Washington's Hoh Rainforest –– conifers were dying at double the rate they used to.

"Every year, you expect some people to die in your hometown," Stephenson analogizes. "If that death rate started to creep up slowly, it doesn't create a dead landscape all at once, but you would sit up and go, 'Oh my gosh, what's happening?' "

As in so many ecological stories, what's happening is complicated. "There is something tied to temperature that is probably responsible for what we're seeing," says van Mantgem. But exactly what that something is may vary from forest to forest. At mid-elevations in the Southern Sierra, where the sugar pines and sequoias live, the increase in mortality seems to be tied primarily to a temperature-induced increase in the atmosphere's demand for water – the same thing Park Williams expects to happen more in the Southwest. But at higher elevations and in wetter forests, like the Hoh, warmer temperatures may instead be favoring the fungi and insects that attack trees.

What the uptick in background mortality ultimately portends is also uncertain. But the forests' response to mild temperature increases, van Mantgem says, indicates their vulnerability. "(The results) might be telling us that they have chronic stress as things get warmer. Then if you get an acute stress, like a severe drought, it might be something that hits you over the head." That is, it might be something that takes out a centuries-old forest in a year, or two – or, in the case of a forest fire, overnight.