Caterpillars transform into butterflies and moths via a radical process known as metamorphosis, where their bodies virtually turn to soup and develop anew.

Since Darwin, biologists have believed that the larval and the adult forms of insects evolved from a common ancestor. Indeed, the evolution of metamorphosis is thought to have fueled the incredible diversity of insects today, allowing them to exploit different habitats at different life stages.

Now, a lone scientist claims that the phenomenon arose when two very different creatures accidentally mated.

In the current issue of Proceedings of the National Academy of Sciences (PNAS), Donald Williamson, a wheelchair-bound 87-year-old zoologist from the University of Liverpool in England, suggests that the ancestors of modern butterflies mistakenly fertilized their eggs with sperm from velvet worms, also known as onychophorans. "People have been trying to find one solution that covers all of metamorphosis," Williamson says. "I say it's a change in taxon during development."

Velvet worms, which fall between worms and insects on the tree of life, have soft-bodies and superficially resemble caterpillars, particularly the larvae of an early butterfly relative known as Micropterix. Velvet worms have evolved a variety of elaborate fertilization procedures. Males are known to place sperm packets not on the female's genital opening, but rather on skin tissue, which the sperm penetrates before migrating to the ovaries.

Williamson believes that an ancient insect accidentally picked up that sperm, and butterflies now contain two developmental programs so they live half their life as velvet worms and half their life as winged butterflies.

"Animals have been able to hybridize since they invented sex," Williamson says. "With external fertilization, there's always the possibility that some sperm will fertilize the wrong egg."

However, scientists asked to comment on Williamson's theory were taken back by it and surprised it made it into such a prestigious journal. For example, from insect paleontologist Conrad Labandeira of the Smithsonian National Museum of Natural History in Washington, D.C.: "You're kidding!"

After looking over the paper, Labandeira pointed out more substantial criticisms. Hybridization between closely related species sometimes occurs in the animal kingdom, but it is highly unlikely that the sperm of a velvet worm could fertilize a distantly related insect egg and produce a viable embryo. He also raises the question of where the genetic program controlling metamorphosis would come from.

"If I was reviewing [this paper] I would probably opt to reject it," he says, "but I'm not saying it's a bad thing that this is published. What it may do is broaden the discussion on how metamorphosis works and…[on]…the origin of these very radical life cycles."

Insect developmental biologist Fred Nijhout of Duke University took a less diplomatic view of the article saying it would be better suited for the "National Enquirer than the National Academy."

"The paper is hypothetical and speculative and not a single bit of evidence supports the idea," he says. He points out that the developmental pathways that connect larvae to adults are well-established and that many structures in caterpillars map to adult butterflies.

This publication is not the first time Williamson has broken rank with the scientific community. Since his retirement in the late 1980s, he has also been arguing that marine invertebrate larvae, such as those from clams and barnacles, arose through hybridization, but those claims were also met with consternation.

One champion of his ideas has been National Academy of Sciences member Lynn Margulis of the University of Massachusetts Amherst, a strong proponent of the controversial idea that genetic variation in organisms has been driven by the horizontal transfer of genes among unrelated organisms. (As LaBandeira puts it, "She has a fondness for weird theories.")

Margulis recalls Williamson first approaching her with his ideas 20 years ago and saying: "I am 68-years-old, from a shortlived family, and on a straight-lined course for posthumous recognition." Williamson later slipped and fell while collecting marine larvae for his studies, which compromised his mobility and further isolated him at his residence on the Isle of Man.*

Margulis initially tried to get this work published in PNAS in the early 1990s, where as an Academy member she serves as an editor, selecting peer reviewers and making publication decisions. However, too many of those reviews were negative to merit publication. One reviewer suggested that Margulis's reputation would be jeopardized by the association with Williamson's work.

Eventually, Margulis convinced Williamson to focus on caterpillars because, she explains, "Everybody knows what a caterpillar is, and it doesn't look anything like a butterfly."  She says it took "6 or 7" peer reviews before she had the "2 or 3" positive ones necessary to make a case for its publication.

One of those reviewers was Robert Higgins, who retired from his position as curator of invertebrate zoology at the Smithsonian in 1993 and now lives in Asheville, North Carolina.  "I’m probably the only one who gave a favorable review to it," he chuckles. "It wasn't that I believed what [Williamson] had in the way of evidence because I don't know that much about every group of invertebrates.  I just look at it as a hypothesis that should be tested."

And so, after 22 years of persistence -- and with Margulis's assistance -- Williamson is finally being heard, for better or worse. "My career took off when I retired," he says.

*Clarification (8/27/09): This sentence has been added since the original publication