In the summer of 2015, a wildfire scorched 280,000 acres along the border of Idaho and Oregon. The spark came from lightning. The tinder came from cheatgrass, an invasive plant from Asia whose dry, straw-like stalks are almost too perfect as kindling.

“After cheatgrass invades,” says Matthew Germino, an ecologist at the United States Geological Survey, “spaces are much more prone to wildfires.” And cheatgrass has been an enormously successful invader, crowding out native plants and leading to ever more and bigger wildfires.

This fall, Germino is part of an experiment to get rid of cheatgrass where the 2015 wildfire burned. His tools are a combination of the conventional (herbicides) and the starkly unconventional (soil bacteria). This bacteria, called Pseudomonas fluorescens, could prevent cheatgrass from putting down its long roots in the ground—and ultimately quell wildfires.

The idea that soil microbes impact plants is, in a way, an old one. Farmers have long planted legumes like alfalfa and clover, which have nitrogen-fixing bacteria in their roots that make fertilizer. And more recently, ag companies big and small have been jumping into the emerging market for “microbials,” bacteria or bacterial extracts that kills pests or boost plant growth. Microbials are the much-buzzed natural alternative to synthetic chemical pesticides and fertilizers.

Until now, no one has tried to use bacteria to tip the entire complex ecology of Western rangelands. Crop fields are relatively simple; they’re monocultures. And the soil bacteria that microbiologists know the most about tends to be soil bacteria in crop fields. “There’s a huge area of semi-arid rangeland and it’s not been studied nearly as well,” says Germino. He’s not sure if the bacteria will work on rangelands, but that’s the whole point of doing the experiment.

The data suggesting these bacteria do suppress cheatgrass come from eastern Washington. The weather there was overcast and cool this week, when I called Ann Kennedy, a soil scientist with the USDA’s Agricultural Research Service. “This is my kind of weather,” she said, by which she means it’s P. fluorescens’ kind of weather. Too much sun and ultraviolet rays kill the bacteria; too hot and they go dormant. Her team was just out spraying bacteria this week. Kennedy has spent about 30 years studying the effects of these bacteria on test plots in Washington.

Kennedy first discovered P. fluorescens, appropriately enough, when she was studying fields of winter wheat. The wheat did not grow well in some areas of eastern Washington and the cause, she found, was bacteria. Some strains of P. fluorescens affected the roots of wheat. Yet other strains affected other plants, including invasive ones like cheatgrass and medusahead. After testing 25,000 strains, her team eventually found about a dozen that disproportionately cut down on the weeds.

The bacteria appear to work by attacking cheatgrass’s main advantage. In the fall, cheatgrass puts down roots that keep growing through winter. So by the time other grasses emerge, it already has a well-established network of roots to outcompete the suckers. P. fluorescens lodges itself just inside the cheatgrass’s roots and prevents them from growing long and deep. After three years, the bacteria cut the amount of cheatgrass in test plots by half. It takes about five years to get rid of them completely.

The key, though, is that the bacteria take time to work. P. fluorescens is not a panacea, but a complement to herbicides. In the Idaho test plots, the hope is that the herbicides can knock out cheatgrass right away and the bacteria will keep them from coming back over several years. “We still need herbicides because bacteria can’t do it all by itself,” says Kennedy, “but that does reduce our herbicide use.”

If the bacteria works and cheatgrass goes away and wildfires die down, another set of consequences come into play. One of the casualties of frequent wildfires is the greater sage grouse, a bird that has been a candidate for listing under the Endangered Species Act. If things get worse for the sage grouse, a listing could close off its habitats to grazing or recreation, which are common in rangelands. “It would greatly decrease the ability for humans to work in these rangelands,” says Germino. These tiny bacteria can have a big effect on land in the West.