One clear night, with the lake so flat you could see the Milky Way reflected on its surface, Marsh edged the boat close to a small rock overhang on the Nevada shoreline and cut the engine. He sank a small light into the water, and soon you could make out scores of tiny wiggling white creatures with little black dots for eyes. They were razorback sucker larvae, hatched just a few days earlier at this spawning ground, known as Tequila Cove.

Roundup participants had already captured several thousand of them here and delivered them to Willow Beach National Fish Hatchery, at Lake Mohave's north end, where they would spend two weeks in an aquarium dining on artificial plankton, spirulina algae, powdered brine shrimp and other delicacies before moving on to indoor and then outdoor raceways. Each year, Roundup participants collect up to 30,000 larvae and ferry them upstream to the hatchery. They remain there for four years, reaching about 19 inches in length. At the end of that time, the surviving fish -- between 6,000 and 10,000 of them -- are tagged with tiny microchips (not much bigger than a staple and $5 apiece) and returned to the lake.

Watching the hatchlings swim in Tequila Cove was both thrilling and heartbreaking. In spite of the fact that humans have entirely transformed the river, adult razorbacks were clearly still doing what they've done for millions of years. And yet these larvae, if left on their own, had virtually no chance of survival. Unless they were caught and taken to the hatchery, they would be eaten by non-natives in a matter of days.

Year after year, biologists and hatchery workers go to the enormous trouble of capturing, transporting, raising and releasing the razorbacks for one simple reason: If the razorback has a future, these Mohave hatchlings carry it in their genes. In the Lower Colorado River Basin -- the section below Glen Canyon Dam -- razorback suckers continue to hang on in Lake Mead, to the north of Lake Mohave, and in Lake Havasu, to the south. Those populations, though, are genetic subsets of the Mohave razorbacks, meaning that the genomes found in the other two lakes -- and everywhere else on the river -- are also found in Mohave, but not necessarily vice versa. "As you go further upstream," said Marsh, "you get less and less genetic diversity, and each population that has been examined is a subset of the Mohave genetic template." Lake Mohave, in other words, is the motherlode of razorback DNA.

A small group of scientists started to issue warnings about the river's fishes in the 1940s, and several seminal research papers were published in the 1960s. But the alarm bell for Lake Mohave's razorback suckers rang out in 1983. W.L. Minckley, an ASU biologist, Marsh's mentor, and one of the godfathers of aquatic conservation on the lower Colorado, predicted that the lake's suckers were destined to vanish unless something could be done to help the hatchlings survive and grow.

He was right: In the mid-1980s, between 60,000 and 75,000 razorback suckers lived in the lake. By the turn of this century, there were fewer than 3,000. Today, scientists believe only about 50 of those wild fish and their lake-raised offspring survive. Despite more than two decades of research and conservation efforts, the population continues to dwindle. Up to 1,500 "repatriates," the fish raised at Willow Beach from wild larvae, are probably also swimming in Mohave's waters -- but those amount to less than 1 percent of the 150,000 hatchery-raised razorbacks that have been returned to the lake since 1992.