CHARLES DARWIN UNDERSTOOD that hybridization is real, and it deeply confused him.

In The Origin of Species, he devoted a chapter to hybrids, but their existence was a riddle he never really solved. Hybrid animals like mules, Darwin noted, are usually sterile. He deemed it a "strange arrangement" that nature would afford two species the "special power" to create hybrids but then prevent these offspring from propagating. He offered squishy theories about why this is so—nobody knew anything about genes then—and on how hybrids fit into his overarching theory of natural selection.

Nine years later, in a book that examined variation in domesticated animals, Darwin explored hybrids more closely—there was even a mention of ligers, which had first been bred in England in 1824. Darwin asserted that hybrids might inadvertently push back the evolutionary clock, resurrecting traits that were better left behind. He used the mixing of human racial groups as an example, stating that foreign travelers frequently remarked on "the degraded state and savage disposition of crossed races of man."

Darwin's hybrids-are-bad dictum became orthodoxy during evolutionary biology's "modern synthesis," in the 1930s and '40s, which firmly connected genetics to natural selection. Harvard ornithologist Ernst Mayr, a leading neo-Darwinist, set the tone by dismissing hybrids as an evolutionary dead end.

Mayr's verdict involved a surprisingly contentious question: What exactly is a species? Darwin had seen it as an arbitrary designation for animals that have similar physical features. Mayr came up with a concrete definition known as the "biological species concept." A species, he declared, is a reproductively isolated group that can interbreed.

By this formula, a species was a fixed unit that was improved over time by forces like random mutation and mate selection, not by having "gene flow"—biologese for doin' the nasty—with other species. "Species were rocks," says Michael Arnold, an evolutionary biologist at the University of Georgia who studies hybridization in both plants and animals.

But maybe they aren't. Arnold is part of a growing camp that sees species as more liquid than solid, and he rejects the idea that hybrids are always evolutionary losers just because they often can't reproduce. "A lot of us have been hammering away on this for many years," says Arnold. "They used to call us the Mongol hordes at their gates, but now we're inside."

Arnold is pushing a profound reconceptualization of evolution, one in which hybrids are more than bit players. Forget the tree of life, with new species neatly branching off from a common ancestor. It's a web of life, and hybrids help genes flow in unexpected directions.

But what about their famous sterility? Some hybrids can reproduce, and Arnold stresses that rare events have an "overwhelming importance" in the evolutionary process. Hybrids often have desirable traits—some are more fit than either parent—and there have been instances when hybrids were able to find enough fertile hybrid partners to create a new species. This process appears to be under way right now in the U.S., involving a hybrid of the Pecos pupfish and the sheepshead minnow that has greatly multiplied and expanded its range. And some scientists contend that matings between gray wolves and coyotes thousands of years ago created an entirely new species: the red wolf.

More commonly, though, hybrids mate with one of their parent species, influencing the mix of what gets passed along to subsequent generations; essentially, they provide a bridge for genes to cross the species divide.

In a paper about hybridization and primate evolution that Arnold co-wrote last year for the journal Zoology, he offers several examples, including chimpanzees and bonobos. DNA studies suggest that these two great apes swapped genes sometime after separating from a shared ancestor at least 800,000 years ago. Arnold believes these ape cousins occasionally mated but that the resulting "bonanzees" did not establish a new species. Instead, they hooked up with either chimps or bonobos. Bonanzees ultimately vanished, but they left genetic footprints in the genomes of their descendants.

In addition to comparing genomes for evidence of unusual gene flow, scientists increasingly are using DNA analysis to confirm the existence of heretofore unknown natural hybrids, whose existence argues that this process still occurs. On April 16, 2006, a hunter in Canada's Northwest Territories shot a polar bear whose fur had an orangish tint. Research showed that this animal had a grizzly bear father, making it the first confirmed wild pizzly ever found. (Pizzlies had been bred before in captivity.) In 2003, DNA analysis done by the Forest Service confirmed that five odd-looking felines found in Maine and Minnesota were bobcat-lynx hybrids, dubbed blynxes. Other DNA-confirmed hybrid mammals reported since 1999 include the forest/savanna elephant in sub-Saharan Africa, minks-polecats in France, and a sheep-goat in Botswana.

How much, then, do hybrids contribute to evolution? Nobody really knows. That's what's so exciting about these new DNA discoveries: The story is still unfolding.