In the early 20th century, George Burrell and other chemists with the Bureau of Mines tested canaries to see if they could reliably signal dangerous levels of carbon monoxide in coal mines. And, as was the way of science at the time, the men also cheerfully gassed themselves, recording how much carbon monoxide it took to produce a walloping headache or leave a man disoriented for days. The canaries looked ill before the men felt ill, making them a useful tool to gauge the quality of mine air. They were the first "indicator species," creatures monitored to determine the health of an environment.

But decades later, when "canary in the coal mine" had become a cliche and the question was not the effects of underground blasting, but aboveground clearcuts, the issue became thornier. How do you take the pulse of a forest? Can you look at one bird species, check its population, its breeding success, the extent of its territory, and learn something meaningful about a patch of old growth or a stand of lowland conifers?

Under the 1982 guidelines for the National Forest Management Act, the Forest Service was supposed to use indicator species to determine if activities such as logging were damaging its lands. The ideal indicator would be dependent on a specific habitat. By tracking its population, land managers could get a glimpse of the health of the habitat and the other animals that relied on it. The northern spotted owl was linked to Northwestern old growth, the Merriam's wild turkey to piñon-juniper forests, the mountain quail to chaparral. Some forests used five indicator species, all birds. Others, like the Klamath National Forest in California and Oregon, used dozens, ranging from the tailed frog to the northern water shrew.

From the beginning, some researchers were skeptical about this approach. These days, when even threatened and endangered species rarely receive the monitoring they need, tracking indicators is a significant expense. (Monitoring just one indicator in western Montana's Lolo National Forest -- the flammulated owl -- costs almost $30,000 a year.) Often scientists have only scraps of data about many indicator species, and baseline information about historical population numbers is usually absent. The kinds of studies Burrell did in his laboratory no longer exist. It's frustratingly inexact.

But indicator species, along with a requirement that managers maintain viable populations of native vertebrates, made the 1982 rule a powerful tool for environmental groups seeking to stop forest plans they believed to be flawed. The courts have repeatedly determined that the rule demands wildlife surveys, and the Forest Service has repeatedly said it doesn't have the time or money. In 2002, the Utah Environmental Congress was able to stop a salvage logging project on the Manti-La Sal National Forest, because the Forest Service hadn't surveyed for blue grouse for at least 10 years. In 2006, a judge halted a grazing plan for Idaho's Sawtooth National Forest because of insufficient monitoring of two indicator species, pileated woodpeckers and bulltrout.

Now, in the face of tight budgets and mounting criticism from foresters and managers, the Forest Service wants to abandon indicator species. The method's critics celebrate the decision and see it as a chance to find sounder alternatives. But environmentalists and some wildlife biologists worry that animals will take a huge hit.