American Scientist

To the Editors:

Donald I. Williamson's unconventional idea ("The Origins of Larvae," November-December) about hybridization and the evolution of complex animal life cycles has had more than 15 years to grow from a newborn heresy into a mature hypothesis. Why has this idea been ignored by almost all other researchers? Perhaps because it is resistant to critical testing.

I published such a test in Evolution in 1996, but Dr. Williamson and his coauthor Sonya E. Vickers mischaracterize the result in their article as an unsuccessful "search for sea-squirt DNA." In fact, the test was highly successful, but produced strong evidence against larval transfer. 

Many evolutionary biologists would agree with W. Ford Doolittle (not "Fred," as mistakenly written in the article) that hybridization and lateral gene transfer are important processes at the roots and the tips of the tree of life, and at least a few (including me) would like to take the cold-fusion hypothesis seriously.

But I doubt that anyone will do so until Dr. Williamson and his advocates agree to offer sensible and testable predictions and to accept the evidence from those tests. This seems to me the only way for Dr. Williamson's contributions to achieve the level of recognition they deserve by either entering the mainstream or by being laid to rest.

Michael Hart
Simon Fraser University
Burnaby, British Columbia, Canada

To the Editors:

Dr. Williamson's hypothesis on origins of larvae raises more difficulties than it solves. In his article with Ms. Vickers, he has continued to ignore these problems and to dismiss alternative hypotheses which do not present such obstacles.

Dr. Williamson has failed to demonstrate three major points: larval-adult incongruities that force one to invoke unusual or unlikely evolutionary processes; developmental processes that indicate the proposed hybridization; and evidence that the hybridization between distantly related animals has occurred or can occur.

On the first point, he presents no incongruities that could not be plausibly explained by convergent evolution. Dr. Williamson ignores studies of functional morphology, ecology and developmental biology of larvae that can explain both metamorphosis and similarities among larval forms of distantly related animals.

On the second, he provides no evidence that his hypothesis excludes alternative explanations of larval development or even that his hypothesized transfers are consistent with what is known of development. As an example, it is unclear how Dr. Williamson's hypothesis could account for the early and simultaneous specification of body axes and structures shared by larval and postlarval mollusks.

On the third, Dr. Williamson's experimental evidence was that he obtained echinoid plutei by fertilization of ascidian eggs with echinoid sperm. Dr. Hart found that mitochondrial DNA sequences in the alleged hybrids were echinoid, not ascidian. The mitochondria are maternally inherited in these animals. The evidence indicates that the plutei were from sea urchin eggs, not ascidian eggs.

This is not to say that horizontal gene transfer is unimportant for the evolution of animals. There is evidence that animals have acquired genes from their symbionts. There are greater obstacles to gene transfer by hybridization between distantly related animals. Such a transfer would be very interesting if it has occurred, but Dr. Williamson has not provided evidence for it. There is as yet no evidence for transfer of larval forms between distantly related animals by hybridization.

Richard R. Strathmann
Friday Harbor, WA

To the Editors:

In their article Dr. Williamson and Ms. Vickers summarize their arguments for the fusion of larval and adult genomes as an explanation for complex life cycles in marine invertebrates. While we agree that the explosion of research into evolution and development has generated "lively debate" about larval evolution, we find the "larval transfer" hypothesis to be highly speculative with little supporting evidence.

Revolutionary ideas in science are often met with intense skepticism and we are loathe to dismiss the larval transfer hypothesis (or any other hypothesis) simply on the grounds that it is heretical. However, after nearly 20 years of research into this hypothesis, Dr. Williamson has provided few experimental data to support the idea that the transfer of genomes by interphylum hybridization explains the discord between the morphology of larval and adult forms of marine invertebrate animals. In fact, the only experimental data subject to external corroboration (through DNA analysis of "hybrid" offspring) were found by Dr. Hart to provide no support at all for interphylum hybridization.

Given Dr. Williamson's claim that "all larvae transferred into their present-day lineages from other . . . animal groups" and that "this activity was not limited to the distant past," there should be ample opportunity for biologists to confirm the existence of interphylum hybridization as a mechanism of evolution. However, despite the ease with which Dr. Williamson's hybridization experiments could be replicated (ascidians and urchins are common members of marine communities worldwide) and the fact that a successful confirmation of this type of hybridization would be truly groundbreaking, there have been no published confirmations of Dr. Williamson's results.

In 2002 one of us attempted to repeat Dr. Williamson's experiments in interphylum hybridization using eggs of the ascidian Boltenia villosa and sperm of the echinoid Dendraster excentricus with no success. We suspect there have been many similar failed attempts to confirm this radical hypothesis that remain unpublished.

Dr. Williamson and Ms. Vickers suggest that the larval transfer hypothesis can explain "oddities" of development such as the persistence of both larval and juvenile stages following metamorphosis in sea stars such as Luidia clathrata. What the authors fail to mention is that it has been known for almost 30 years that members of the genus Luidia frequently undergo asexual reproduction during the larval stage. Therefore the retention of a larval form following metamorphosis may be a functional adaptation to increase offspring numbers through budding. It should also be recognized that asexual reproduction through budding is a trait shared by both larval and adult echinoderms (particularly ophiuroids and asteroids), one of many traits that suggest shared ancestry between the different stages of the life cycle.

Given the lack of empirical support for the larval transfer hypothesis, we are left to wonder why Dr. Williamson continues to ignore the robust literature on larval form and function that serves to support alternative explanations for larval evolution (such as descents with modification and convergent evolution). Larval forms possess traits well suited to life in plankton and serve adaptive functions related to dispersal, food collection and predator avoidance, as Dr. Strathmann reported in the Bulletin of Marine Science in 2007. Variation in larval forms among closely related species suggests that natural selection is acting on these traits as it would on any other trait. The best-studied examples of closely related species with disparate larval forms show key similarities in larval form (such as the presence of vestigial feeding structures in non-feeding forms) that are most parsimoniously explained by shared ancestry. In addition, some of the same genes are expressed during both embryonic and juvenile development in the same species. For example, as reported by Michael J. Ferkowicz and Rudolf A. Raff in Evolutionary Development in 2001, in two closely related species of sea urchins within the genus Heliocidaris the gene Wnt-5 is expressed in the blastula stage, in the larval ciliated bands and in the developing adult rudiment and nervous system. If the larval-transfer hypothesis were correct, we might expect that different genes (from distantly related genomes) would be involved in embryonic and juvenile development. Furthermore the fact that both the sequence and expression pattern of Wnt-5 are conserved between the two Heliocidaris species with divergent larval morphologies (one lecithotrophic, one planktotrophic) suggests that common ancestry underlies these forms.

Instead of selectively using developmental data to support the illogical conclusion that "even close relatives may not have a common ancestor," these data should be used to highlight the ways in which development can help explain mechanisms of metazoan evolution. Thankfully, the latter is the approach taken by the vast majority of biologists specializing in this field.

Jonathan D. Allen
Margaret Pizer
Brunswick, ME

Dr. Williamson responds:

Dr. Strathmann, in his review of my 1992 book, originated the suggestion that I may have unwittingly used eggs from hermaphrodite Echinus in my hybridization experiments in 1989 and 1990, and this was repeated by Dr. Hart, who investigated two genes from sea urchins produced by the 1990 experiment. In my 2003 book I show that eggs from these improbable hypothetical hermaphrodite sea urchins would not have hatched.

Neither the hermaphrodite Echinus hypothesis nor Dr. Hart's alternative suggestion of contaminated cultures explains the fact that two of the four Echinus that lived for more than a year were the only tetraradial urchins ever recorded, and that the great majority of the pluteus larvae in the culture that produced these urchins metamorphosed into spheroids, which are quite unlike any known form in echinoderm development.

I do not regard Dr. Hart's results as "an unsuccessful search for sea squirt DNA" but as a valuable investigation into the genes of hybrids. They show that, in this case, male mitochondrial DNA was inherited. I am aware that the male mitochondrion disintegrates in early cleavage in most mammals, but it survives in bivalve molluscs and probably in other invertebrates. This is the first time it has been investigated in a cross of a sea squirt and a sea urchin. Two colleagues and I have since successfully fertilized sea-urchin eggs with sea-squirt sperm (the reciprocal cross to that carried out in 1989 and 1990). This, too, produced plutei followed by spheroids, samples of which have been sent to Dr. Hart.

Drs. Hart, Strathmann, Allen and Pizer all call for predictions to test the larval-transfer hypothesis. I have already published several, including one relating to Michael Syvanen's 18S rRNA cladogram, mentioned in the American Scientist article. This shows the sequence acorn worm to sea cucumber to starfish to sea urchin, which is not the order in which these groups evolved but the order in which I claim they acquired larvae. I predict that, if and when the 18S genes of Planctosphaera and a brittle star are investigated, Planctosphaera will come at the beginning of this sequence and the brittle star at the end.

Dr. Strathmann seems to think that most larval anomalies can be explained by convergent evolution, the process that explains why many species of fast swimmers are streamlined. Most marine larvae, however, are planktonic, which means they swim slowly or drift. The almost infinite variety of planktonic shapes shows that there is no optimal shape for swimming slowly. Convergent evolution cannot explain why sponge crabs and hermit crabs have such similar larvae, and dozens of other larval anomalies.

Regarding the points raised by Drs. Allen and Pizer, larval transfer explains how animals acquired the basic forms of larvae. What happened subsequently to these larvae is, in many cases, a matter of Darwinian evolution. Both processes go on together. The fact that larval budding can take place in some species of Luidia does not detract from the remarkable degree of independence shown by the larva and juvenile of L. sarsi, consistent with the idea that they originated in different genomes.

I urge Drs. Allen and Pizer, and anyone with access to a laboratory, to try again to cross distantly related animals. Please pre-treat the eggs with acid seawater of pH 5 for 40 seconds and use a dilute sperm suspension. I have just submitted for publication an account of hybrids between Psammechinus miliaris eggs and Ascidiella aspersa sperm. The resulting larvae were again plutei which resorbed their arms to become spheroids, but these spheroids divided, and some developed into forms resembling juvenile ascidians but did not develop further.