In her U.S. Forest Service laboratory in Provo, Utah, ecologist Susan Meyer wends through piles of petri dishes. Each holds a fungus she calls the Black Fingers of Death, which grows on cheatgrass seeds, poking out like grasping fingers. Aptly named, it has the power to mass-murder those seeds.

Meyer discovered this power while trying to solve a restoration problem. Typically, after a fire, teams would spray a pre-emergent herbicide in the fall, to kill cheatgrass when it sprouted after autumn rains. At the same time, they'd seed natives to sprout the following year, after the cheatgrass was killed off.

But the grass is clever; not all its seeds germinate with that first rainfall, and the leftovers, often as many as 10,000 per square meter, go dormant, usually until the rains of the following fall. By then, the herbicide is gone, and spraying again isn't possible, because it would kill germinating natives. "It's common to have a reasonably successful seeding that then gets overwhelmed by cheatgrass the next year," because of those second string germinators, says Meyer.

But Meyer's fieldwork also showed her that some of those second-year seedlings were dying off, even without herbicide. "We found this fungus that was eating 60 to 90 percent of the seeds."

That fungus, which hitched a ride across the ocean attached to cheatgrass, was Black Fingers of Death, or BFOD. In its natural state, BFOD is not potent enough to kill all the late germinating seeds. Meyer thought she could get around that by seeking out the most deadly strains and culturing them, so that all those late-germinating cheatgrass seeds "won't be there next year to haunt your seeding." Meyer's achieved a pretty consistent 90 percent seed death rate, with 100 percent in a few cases. She has applied for a patent, and has been contacted by a couple of biocontrol companies.

In order to make Black Fingers as deadly as possible, Meyer's collaborators have gone so far as to change how it reproduces. In the lab, it typically does this clonally, but biologist Julie Beckstead is working on getting the fungus to "do sex," says Meyer, which would allow the researchers to breed the most effective BFOD strains together.

When used in concert with an herbicide that takes out the first-year crop of cheatgrass seedlings (because they germinate so quickly, the fungus doesn't work on those early seeds, although Meyer is researching that problem), Meyer's work may create the perfect conditions for restoration. She's tested to see if it kills native grasses, and has found little evidence that it would affect them.

There is a catch, though. Even if a biological control like Meyer's succeeds, land managers are still left with an open desert landscape. It's up to her fellow scientists to keep cheatgrass -- or something worse -- from seizing the moment yet again.

Lack of rain is the main culprit in most Great Basin restoration failures. A wet year makes everyone a good range manager, as the saying goes. Yet even though land managers know their seedings fail more often in tough, dry years, they do not know exactly why. Jeremy James, a range ecologist with the University of California Agriculture and Natural Resources Program, reasoned that if he knew exactly what caused plants to die, he could help them survive.

In 2007, James experimented with tracking seeds planted after four different fires in Great Basin ecosystems in eastern Oregon. To his surprise, the ecologist found that it wasn't that seeds failed to germinate. "In most years, we can get 60, 70, 80 percent germination in the field," says James. "But we also found that about 90 percent of those seeds that have initiated germination never make it. They never actually get above the soil surface."

To understand the many ways seeds can germinate but still fail, think about Beth Leger's Poa. Say one of its seeds drops down onto that hard pebbly dirt and makes its way into the soil. Come spring, the 3.5 millimeter oblong kernel might swell with water as the ground warms and snow melts. Then it stretches out its first root, and sends the beginnings of a leaf upward, towards the soil surface. But in a light snow year, the seed may stop there, lacking the moisture to continue. Maybe there's a late hard frost, and the tiny root freezes and withers. Perhaps a fungus in the soil devours the seed, or the seedling simply cannot push its way through the hard, dry soil crust.

In a normally functioning ecosystem, none of these failures would really bother the Poa. Sure, its seeds might not make it one year, but they'd succeed the next, or the one after that. But while Poa might not be in a hurry, rangeland managers lack the luxury of time. They need natives to establish quickly, before cheatgrass takes over again.

In James' next experiment, a three-year study that he is conducting with Leger and a few other scientists, he will try to narrow down the causes behind seed failure. What he finds could radically change BLM practices after a fire. If those native seeds are mostly becoming pathogen fodder, the BLM could order seed coated in fungicides. If the problem is lack of moisture, seeds could perhaps be planted to have better seed-soil contact, especially in dry years. James' colleague, ecologist Matthew Madsen, has been working on a technique called "bundling," where many native seeds are coated and clustered together, like clumps of granola. In his trials, this has helped them retain moisture, emerge through thick soil crusts and better survive dry winters.

James is also working with a plant physiologist and modeler, Stuart Hardegree of the USDA Agricultural Research Service in Boise. The models Hardegree is developing, based on historic weather patterns, will help land managers predict weather conditions months into the future. That way they'll know, when seeding after a fire, if they should give seeds extra help, depending on whether that spring will be drier or colder. In some cases, if the model is pessimistic about restoration conditions, land managers might elect to plant something like crested wheatgrass, which is more successful at establishing than native plants, says Hardegree. Even planting a field of non-native crested wheatgrass might be better than doing an expensive seeding of natives and having them fail, followed by a return to flammable cheatgrass.