Not that long ago, tens of millions of Bay checkerspots likely fluttered over the Bay area. Then the Spanish arrived, bringing the seeds of Mediterranean pasture grasses. These fast-growing annuals quickly spread everywhere, avoiding only the poorest soils, particularly those derived from serpentine, a class of greenish minerals for which the state rock of California is named. Naturally low in nitrogen, these soils protected the checkerspot until well into the 20th century, when highways, housing developments and shopping malls invaded the serpentine, chopping it into ever smaller, more vulnerable patches.

Like other insects, checkerspots are prone to boom-bust population cycles. A hillside that supports 20,000 caterpillars one year may harbor less than 3,000 a year or two later. Heavy spring rains knock egg masses off stems and prevent adults from feeding. Hot dry springs cause dwarf plantain and flowery nectar sources to shrivel early, leading to mass starvation. Stressed by multiple years of bad weather, small, isolated populations can easily go extinct, as happened at Stanford University's Jasper Ridge Biological Preserve in the late 1990s, around a decade after the Bay checkerspot was officially listed as threatened.

By that point, Weiss was concerned about the fate of much larger Bay checkerspot populations. In the early 1990s, he started monitoring the ecological health of six Coyote Ridge properties. All had extensive stands of dwarf plantain, seemingly healthy checkerspot populations and, as Weiss went on to establish, similarly high rates of nitrogen deposition -- between 10 and 20 pounds per acre per year, up to five times the rate measured at a less polluted location.

There was, however, one key difference. At two of the properties, near the northern end of the ridge, grazing had recently stopped. While Weiss looked on, both turned into dense swards of Italian rye grass; as the native plants faded away, so did the Bay checkerspot butterflies.

In 2001, checkerspots vanished from another swath of ungrazed serpentine, this time at Edgewood Park, a nature preserve that borders Interstate-280. Once again, the population crash coincided with the disappearance of dwarf plantain under a smothering cover of grasses. Once again, Weiss measured a high rate of nitrogen deposition, including a big dose of ammonia produced by catalytic converters. Edgewood Park, he says, was a "drive-by extinction."

Alarmed by the rapid downward spiral, Weiss began inviting local conservationists and political leaders to come up and visit Coyote Ridge. One of his goals was to dispel the widespread impression that cows are always bad for conservation. While ill-controlled herds have damaged landscapes across the West, Weiss says, well-managed herds can help preserve native ecosystems, including these flowery grasslands.

Weiss' efforts have had an impact. In 2005, when the Santa Clara Valley Transportation Authority established a 548-acre conservation set-aside on Coyote Ridge, grazing continued as before. Just last year, the Silicon Valley Land Conservancy invited fourth-generation rancher Justin Fields to sic his cows on a grass-choked meadow at nearby Tulare Hill. "This is new, environmentalists telling us we're doing the right thing," says Fields, who also grazes his cows on Coyote Ridge.

Weiss angles his way along a slope so steep that, here and there, he holds out his arms for balance. For the past couple of hours, he and two assistants have been conducting a series of caterpillar counts. Timed by a stopwatch, eyes on the ground, they walk parallel to each other along imaginary transect lines. Exactly 200 seconds later, when the stopwatch beeps, they record the number of caterpillars they've collectively spotted.

Weiss has worked out a formula: Each caterpillar found per 10 person-minutes can be extrapolated into 100 caterpillars per acre. Given an expanse of more than 5,000 acres -- the size of the serpentine at Coyote Ridge -- these annual counts can reach impressive numbers. In good years like 2003 and 2004, Weiss says, there may be as many as 2 million caterpillars; in bad years like 1991 or 1996, there may be only 100,000.