A reprint from American Scientist the magazine of Sigma Xi, The Scientific Research Society

This reprint is provided for personal and noncommercial use. For any other use, please send a request to Permissions, American Scientist, P.O. Box 13975, Research Triangle Park, NC, 27709, U.S.A., or by electronic mail to [email protected]. ©Sigma Xi, The Scientific Research Society and other rightsholders The Origins of Larvae

Mismatches between the forms of adult animals and their larvae may reflect fused genomes, expressed in sequence in complex life histories

Donald I. Williamson and Sonya E. Vickers

ince the earliest days of the field—when a Syoung Charles Darwin worked on his beloved barnacles, shelled shrimplike creatures, cemented to rocks, which lie on their backs and kick food into their mouths—evolutionary biologists have been fascinated by life’s myriad odd forms. The rigorous naturalist confronted with unexplained peculiari- ties of form, life history or behavior is compelled to search doggedly for a scientific explanation. Just such a search some 150 years ago gave the world The Origin of Species, which remains the founda- tional scientific explanation of how plants and ani- mals have changed through time. More animals are hatched or born than possibly survive to have offspring in their environment, Darwin wrote. Thus pressured to adapt, populations change gradually through time; they “descend with modification.” Darwin emphasized the gradual nature of change in living form that today we call evolution: the accumulation of changes, or mutations, through heredity. Not all modern theorists have accepted that change is gradual. Indeed, nature presents forms that can only be explained by sudden change. One of us (Williamson) has focused on such a case: D. P. Wilson/FLPA larvae, the distinctive young forms of many ani- Figure 1. Luidia sarsi, a starfish found in the North Sea and along the southern mals. These forms can differ so markedly from the and western shores of the British Isles, develops from a fertilized egg into a adults into which they develop that an observer larva with bilateral symmetry, inside which grows a juvenile form with the ra- is tempted to classify them as different species. dial symmetry of the starfish. In this unusual life history, the juvenile migrates And such an observer, we argue, might be right in to the outside and drops off the swimming larva (the translucent shape in this photograph); both continue to live independently for up to three months. a way. Williamson’s “larval transfer” hypothesis Author Williamson’s larval-transfer hypothesis explains this oddity as the ex- proposes that larvae, and the genes that specify pression of genomes fused when two ancestral marine animals hybridized, one them, have been transferred from one hereditary becoming the larva of the other. This argument is not only a radical proposal animal lineage to another by cross-species, cross- for solving open questions in the evolution of development but also one of genera and even cross-phyla fertilizations. We feel several proposals that suggests a bushier shape to the “tree of life” that Charles compelled to ask a question that is obvious to those Darwin proposed.

© 2007 Sigma Xi, The Scientific Research Society. Reproduction www.americanscientist.org 2007 November–December 509 with permission only. Contact [email protected]. organisms came to the same solution to problems such as dispersal and feeding. In this article we will tell a differ- ent version of the origin of larvae, dis- tinct from the “same stock” concept but consistent with Darwin’s theory of natural selection: Larva and adult began as different animals, each developing from its own type of egg. At a point in their evolu- tionary past, their ancestors interbred and produced offspring. Most hybrid offspring did not survive. Genomes so unlike each other had many difficul- ties in expression—the translation of genes into proteins. The few hybrids that did survive solved the dilemma by expressing their combined genomes sequentially rather than concurrently, first the larval genes and then the adult. The new combined animal survived and went on to reproduce specialized forms, usually highly advantageous in a b c the procurement of habitat and food, then spread their genes as adults. The Figure 2. From Darwin forward, classical theory has viewed life’s evolutionary history as taking legacy left by the patching together of place through continuing adaptation and mutation. Darwin’s tree of life may be seen as a smooth- dissimilar ancestral lines is the perilous ly branching tree in which each bifurcation indicates common ancestry, and truncated branches transition we call metamorphosis, the indicate extinctions (a). In 1972, Niles Eldredge and Stephen Jay Gould, noting spurts of activity stage at which the larva transforms into in the fossil record, introduced the concept of “punctuated equilibrium,” pointing out that species the adult. The outcome is frequently appear fully formed (b). Williamson’s larval-transfer theory introduces another wrinkle: the notion not transformation but death. that one animal can become the larva of another, even distantly related animal by hybridization, thus establishing a connection between two branches of the tree of life (c). The diagrams trace the ancestry of a hypothetical starfish (red), one of several species that may have acquired larvae The Tangled Tree of Life through hybridization (blue arrows). (Adapted from a diagram by Sonya E. Vickers.) Darwin’s Origin gave science a pow- erful metaphor: the tree of life. The not trapped in conventional evolution- In 21st-century scientific language, we Darwinian view of life’s history is ary thinking: Could animals with larval would say that a portion of one ani- a tree whose central trunk is root- forms be hybrids, the products of suc- mal’s complement of genetic material, ed in common ancestry. The tree’s cessful fusions of genomes that are ex- or genome, was acquired by another, branches—main limbs from which pressed in sequence during the animal’s creating a chimeric organism. further branches diverge, each bifur- life history? Following Darwin, most biologists cation indicating common ancestry— Larvae, familiar as immature stages today assume that a larva and its adult represent the diversity of life that arose in many animal life histories, are espe- began as a single individual and that over time. Most branches fail to reach cially common in marine plankton. The over time, the young gradually became the top (the present), as more than 99 caterpillar larva that spins the chrysalis more and more different from their adult percent of all past life on Earth is esti- from which an adult butterfly emerges forms. This can be labeled the “same mated to belong to extinct species. lives on land, but most dramatic lar- stock” or “direct filiation” assumption. A new concept, “punctuated equi- val-adult transformation takes place in Until recently, there was no alterna- librium,” was introduced by Niles El- the sea. Clams, starfish and sea urchins tive theory and no need to defend the dredge and Stephen Jay Gould in 1972 cast their eggs and sperm into the sea, presumption. There were unexplained to replace Darwin’s gradually branch- where they merge in fertilization. The anomalies, but these were regarded as ing tree. Eldredge and Gould pointed larval-transfer hypothesis proposes that mere curiosities. Conventional think- out that the fossil record contradicts all larvae transferred into their present- ing offers convergent evolution as the the “numerous, successive, slight mod- day lineages from other distantly relat- dominant explanation for the similarities ifications” described by Darwin. “A ed animal groups by cross-fertilization. of many larvae, speculating that many species does not arise gradually by the steady transformation of its ancestors,” they argued; rather, “it appears all at Until he retired in 1997, Donald I. Williamson was Reader in Marine Biology at the University of Liv- once and fully formed.” The concept erpool in the United Kingdom. He specialized in marine plankton, particularly marine larvae. Most of his publications are on crustacean larvae. He is the author of Larvae and Evolution: Toward a New of punctuated equilibrium explains the Zoology (Chapman and Hall, 1992) and The Origins of Larvae (Kluwer, 2003). Sonya E. Vickers is fossil record by spurts of activity fol- a teacher and naturalist. She has traveled extensively within the U.S. and around the world to sites of lowed by stasis. The Eldredge-Gould natural interest. Address for Williamson: 14 Pairk Beg, Port Erin, Isle of Man IM9 6NH, UK. Internet: replacement for Darwin’s tree looks [email protected] like a candelabra.

© 2007 Sigma Xi, The Scientific Research Society. Reproduction 510 American Scientist, Volume 95 with permission only. Contact [email protected]. Larval transfer is one of a number of phenomena that imply a bushier sort of tree, one whose branches occasionally fuse. Although less tidy, the concept is not new. In sexual animals, fertiliza- tion routinely fuses genomes, usually those of members of the same spe- cies. But genome fusion can even take place between biological kingdoms, as in the formation of the composite life forms called lichens. These gray-green patches on rocks and trees represent fu- sions between a green alga in the protist kingdom (or often a cyanobacterium in 2 1 the prokaryote kingdom) with specific ascomycetes in the fungus kingdom. And as this article was going to press, 4JQVODVMB "OOFMJEB .PMMVTDB SPUJGFS the genome of the fruit fly Drosophila ananassae was found to have encapsu- Figure 3. Rotifers have simple life histories, but these small marine and freshwater animals may have lated within it the entire genome of the contributed a larval stage to the life histories of other animals, explaining the scattering of so-called bacterium Wolbachia, each generation of trochophore larvae through unrelated phyla. In the above scenario, reading forward in evolutionary insect inheriting the parasite’s genome time from the bottom, a polychaete worm hybridized with a rotifer, acquiring a trochophore larva (1); from its parents. The finding, as W. Fred this part of the polychaete’s genome was acquired by a sipunculan worm in a second hybridization Doolittle of Dalhousie University put it, (2). Further hybridizations with rotifers gave trochophore larvae to the ancestors of today’s clam-like and snail-like mollusks. Their close relatives, the octopuses and squid, lack larvae. In conventional “establishes the widespread occurrence thinking , larval forms arose over time as young and adult forms within a species became more and and high frequency of a process that we more different. The similarities among larvae in distantly related species are thus conventionally would have dismissed as science fiction explained by convergent evolution—many organisms developing larval stages to solve problems until just a few years ago”—lateral gene such as dispersal and feeding. (Adapted from diagram by Sonya E. Vickers.) transfers involving higher organisms. Elsewhere Geosiphon pyriforme adults mens into snail shells, which they carry It gradually became apparent to form by the fusion of a fungus with around with them. (They are not true me that the hypothesis could be ap- Nostoc, a cyanobacterium, every six to crabs.) I compared them with sponge plied to all larvae, and that all larvae eight weeks to form an organism that crabs, true crabs that carry pieces of have (or once had) an adult coun- looks like a small bulbous moss plant. sponge on their backs. These adult ani- terpart—an animal that does not And Lynn Margulis of the University mals are very different from each other, metamorphose. The whole genome of of Massachusetts at Amherst has docu- hardly related at all. The larvae of the this animal is transferred, but the hy- mented perhaps the ultimate genome two groups, however, look like mysid brid uses only part. fusion: when bacteria that respired at- shrimps and are strikingly similar. It was mospheric oxygen merged, perhaps as if sponge crabs had acquired shrimp­ The Caterpillar and the Snail two billion years ago, to produce aero- like larvae from hermit crabs—an unex- Imagine taking a larva’s view of the bic protists that permanently contain plained curiosity and, according to the organization of life. To us landlubbers, mitochondria. The novelty of the larval- thinking of the time, an impossibility. the most familiar larvae are the cat- transfer hypothesis is not the perma- How, I wondered, could a familiar erpillars of lepidopterans—butterflies nent merger of two different genomes, shrimp metamorphose into two dif- and moths—so we can begin there. but the fact that each oversees a sepa- ferent types of crabs? I began to notice Caterpillars are larvae with three pairs rate portion of the animal’s life history. other anomalies. Radially symmetrical of legs extending from the thorax and A complete “changing of the guard” starfish and bilaterally symmetrical a variable number of small extra legs, takes place during metamorphosis. acorn worms also have very similar or prolegs, attached to the abdomen. If an adult animal and its larva are a larvae. The conventional explanation, An expandable “eating machine,” a chimera evolved from the fusion of two advanced by Ernst Haeckel in 1866, caterpillar can only crawl, never fly. very different animals, the resulting pat- imagined a common ancestor with bi- Caterpillar larvae, however, are not tern of life’s history should be depicted lateral symmetry. Ancestral starfish, confined to the order Lepidoptera. They more like a network than a tree. The the story goes, anchored themselves to also occur in scorpionflies in the order larval-transfer hypothesis also provides a substratum and gradually developed Mecoptera and in woodwasps and one possible mechanism for Eldredge a radial symmetry more efficient for sawflies in Hymenoptera. Other hy- and Gould’s punctuated equilibrium. fixed forms, while the free-floating lar- menopterans, including ants, bees and vae retained the primitive symmetry. wasps, have legless grubs as larvae. The Larval-Transfer Hypothesis I questioned this far-fetched tale and If you were to classify the types of Williamson recalls the origin of the hy- the conventional wisdom that it was insect larvae, in fact, you would come pothesis this way: impossible to transfer larvae or their up with a pattern quite independent It all began with sponge crabs and genetic recipes. I eventually decided of the classification of the adults. Al- hermit crabs. I first studied hermit crabs, that cross-fertilization, or hybridiza- though this pattern is problematic for so called because they tuck their abdo- tion, was the method of transfer. the theorist looking for a common an-

© 2007 Sigma Xi, The Scientific Research Society. Reproduction www.americanscientist.org 2007 November–December 511 with permission only. Contact [email protected]. To take another example, the rotifers or “wheel animalcules” are a phylum of small marine and freshwater animals that have cilia and a simple life cycle. Trochosphaera is a rotifer with a marked resemblance to trochophore larvae. We are convinced that clams, snails and polychaete worms and some sipun- culans acquired trochophore larvae by hybridizing with rotifers. No oc- topuses, squids, earthworms or other sipunculans ever hybridized. Paired with natural selection, the hy- pothesis allows the distribution of lar- vae to be explained by the occurrence or nonoccurrence of hybridization, but also by the loss of larvae in species that once hybridized. Some sea snails and polychaete worms that lack larvae but Peter Parks/imagequestmarine.com Peter Parks/imagequestmarine.com are closely related to species with larvae Figure 4. Echinoderms are radially symmetrical marine animals, including starfish, sea ur- have probably lost their larvae. The pat- chins and sea cucumbers. Although the radial symmetry of adults is traditionally thought to tern of cell division within the egg in have evolved in response to their sedentary lifestyle, the evidence is mixed. Radial adults can these animals is similar to that in species be found deep in the fossil record; adult sea cucumbers show a mix of bilateral and radial fea- with larvae. The pattern of cell division tures, and their auricularia larvae (left) more closely resemble the tornaria larvae of creatures such as the acorn worm (right) than the larvae of other echinoderms. Williamson proposes in octopuses, squids and earthworms is that the acorn worm acquired its larva by hybridization with an ancient planctosphere, then quite different, in line with the view that hybridized with a sea cucumber. these animals never had larvae. The competing explanation un- cestor, it is explicable if larvae were Within these groups one sees a curi- der the same-stock theory is that the later additions to life histories. ous distribution of larval forms. Most adult rotifers are “persistent larvae”— Williamson’s larval-transfer hypoth- clams and sea snails develop from small, descendants of forms that matured in esis holds that the original caterpillar translucent trochophore larvae, charac- the larval state. Were this so, we should larvae were transferred from adults re- terized by one or more bands of hair- expect the genes of the lost adult to be sembling present-day velvet worms of like appendages (cilia) and sharing no preserved as the “junk DNA” found the genus Peripatus. This is the adult morphological traits with the adults into in many genomes; but in fact rotifers counterpart for all caterpillar larvae. which they grow. Octopuses and squids, have remarkably few junk genes. This worm lives in the organic-rich soils also Mollusca, entirely lack larvae. of tropical America, South Africa and Many polychaete worms have Stars on the Bottom of the Sea Australia. These worms-with-legs thus trochophore larvae similar to those The echinoderms are a phylum of radi- belong to the phylum Onychophora, of clams and snails. Yet in the same ally symmetrical animals that includes entirely separate from insects or even phylum, earthworms have no larvae. starfish, brittle stars, feather stars, sea earthworms. When hybridization be- Some but not all members of several urchins and sea cucumbers. Echino- tween different insects and velvet lesser-known marine phyla, including derm larvae have a symmetry different worms took place, the surviving chi- some sipunculan worms (the “peanut from adults. Not only are they bilater- mera enjoyed the best of both worlds: worms”), also have trochophore lar- ally symmetrical, but with their mix- a larval form specialized for feeding vae. The conventional explanation of ture of convoluted and straight bands and a flying adult adept at spreading its the link between these phyla is that of cilia, some also resemble the tornaria genes. Both onychophorans and flying they all descended from ancestors with larvae found in the Hemichordata, the insects have survived rigorous natural trochophore larvae. Following this phylum of acorn worms. The widely selection for millions of years, yet hy- logic, groups such as octopuses, earth- accepted explanation of this anomaly, bridization generated the novelty that worms and some sipunculans evolved the one advanced by Haeckel, states neither could manage alone. by loss of larvae. that the bilateral larvae of echinoderms Broadening one’s focus from cater- The larval-transfer hypothesis im- and hemichordates evolved from a pillars to look at other larval types up- plies, then, that similar larvae will turn common ancestor with tornaria larvae. ends conventional views of life’s orga- up in the life histories of distantly re- An ancestor of modern echinoderms nization. The Mollusca (mollusks) are lated and very different animals—and evolved radial symmetry, he claimed, a large phylum that includes clams, that closely related animals will have in response to sedentary life. snails, octopuses and squid. Annelids quite different larvae or diverging life Fossils and larvae described since are another large phylum of segment- histories because only some acquired Haeckel do not support his views on ed worms that includes polychaetes larvae. It also implies that there must the origins of echinoderms. Radial and earthworms. Members of these be adult forms similar to larvae. echinoderm adults can be found in the two major groups are as distantly re- And indeed, in addition to the vel- fossil record going back at least 540 mil- lated as are rabbits and butterflies. vet worm mentioned above, there are. lion years. Some hemichordate larvae

© 2007 Sigma Xi, The Scientific Research Society. Reproduction 512 American Scientist, Volume 95 with permission only. Contact [email protected]. resemble trochophores, not tornarias. the hypothesis views successful hy- Of all echinoderm larvae, the sea cu- bridization as a very rare event—in cumber larva most closely resembles a one experiment more than 3,000 eggs tornaria. Adult sea cucumbers show a hatched and produced pluteus larvae, mixture of bilateral and radial features. the typical larval shape of the pater- A Haeckelian interpretation suggests nal sea urchin, whose slender arms that sea cucumbers are the nearest liv- are supported by calcareous rods. Well QMBODUPTQIFSF ing echinoderms to the ancestral form. over 90 percent of these resorbed their Paleontology, however, tells us that sea pluteal arms to become spheroids, cucumbers evolved comparatively late which lived for more than a month in echinoderm phylogeny. Williamson but did not develop further. The other proposes, then, that larvae were later 7 to 9 percent of the plutei metamor- additions to this branch on the tangled phosed into sea urchins, four of which tree of life. The adult from which echi- survived for more than a year. Two noderm tornaria larvae evolved, in this survivors had the fivefold radial sym- view, is Planctosphaera pelagica. This metry of a typical sea urchin, but the BDPSO
XPSN UPSOBSJB is the only known planctosphere—a other two displayed fourfold symme- spherical planktonic animal (hence the try. The three largest survived beyond name) that is up to 25 millimeters in four years until the seawater circula- diameter and propelled by convoluted tion failed in the laboratory, all of them bands of cilia. producing eggs in their later life. This animal is classified as a hemi- Hart, reviewing this evidence later, TFB
DVDVNCFS BVSJDVMBSJB chordate because of its resemblance proposed that if the survivors were hy- to a tornaria larva, but we regard it brids, genetic analysis should detect mi- as an adult member of the group that tochondrial DNA from the female par- gave rise to tornaria larvae by hybrid ent, the sea squirt A. mentula, as well as transfer. That is, an ancestor of Plancto- nuclear DNA from both parents. Three sphaera hybridized with an acorn worm years after the death of the larvae, he to produce an acorn worm with a tor- extracted DNA from frozen tube feet of naria larva. This type of larva was then the three surviving urchins for compar- spread by cross-fertilization between ison with the DNA of wild individuals TUBSGJTI CJQJOOBSJB an acorn worm and a sea cucumber. of the two species involved. Further hybridizations between sea Hart’s comparison of aligned nucleo- cucumbers and starfish, starfish and tide sequences and a portion of the nu- sea urchins, and sea urchins and brittle clear 28S ribosomal RNA gene did not stars, would explain the larval forms turn up evidence of sea-squirt genetic found among these echinoderms. material in the survivors of the experi- The doliolaria larva of one group ment. Hart speculated that the experi- GFBUIFS
TUBS EPMJPMBSJB of echinoderm, the feather stars, can- ment might have been contaminated not be traced back to a planctosphere. with sea-urchin eggs, despite the precau- And such connections are not neces- tions taken. Along with Richard Strath- sary with larval transfer, which sug- mann of the University of Washington, gests that the evolutionary dispersal of he wondered whether there might have tornaria-like larvae should be indepen- been hermaphrodites—known to be dent of adult echinoderm evolution. rare—in the sea-urchin population used. Williamson totally rejects these explana- TFB
VSDIJO QMVUFVT Testing the Hypothesis tions and gave his reasons in his 2003 As Michael W. Hart of Simon Fraser book, The Origins of Larvae. University pointed out in 1996, the lar- Although the search for sea-squirt val-transfer hypothesis is so heretical DNA in Williamson’s tissue samples was that a single positive example would unsuccessful, molecular-biological evi- bolster it. As a test, Williamson devised dence for larval transfer is now appear- CSJUUMF
TUBS QMVUFVT an experiment aimed at inducing hy- ing. The Canadian zoologist Ernest W. bridization in the laboratory. In 1990, MacBride, who in 1914 tackled the prob- Figure 5. Dissimilarities between adults (left) he fertilized eggs of a sea squirt (a uro- lem of echinoderm forms, recognized a and larvae (right) of the acorn worm (a hemichor- chordate), Ascidia mentula, with sperm sequence in the appearance of larvae in date) and the echinoderms are striking. All the of a sea urchin, the echinoderm Echinus the evolutionary record, from tornaria larvae shown may have been acquired through esculentus. The results were reported (acorn worms), to auricularia (sea cucum- a series of hybridizations that began when the acorn worm acquired an ancient planctosphere in Williamson’s 1992 book Larvae and bers), to bipinnaria (starfish) and finally (top). In this drawing adults are reduced in size Evolution: Toward a New Zoology. to pluteus (sea urchins and brittle stars). and larvae enlarged. Planctosphere and tornaria What happened? Although, as ex- This same sequence, consistent with lar- redrawn from Barnes et al. 2001. All others re- pected, the Ascidia eggs failed to divide val transfer, has now appeared in the dis- drawn from Williamson 2003, with permission in the majority of such experiments— tribution of a ribosomal gene. of Springer Science and Business Media.

© 2007 Sigma Xi, The Scientific Research Society. Reproduction www.americanscientist.org 2007 November–December 513 with permission only. Contact [email protected]. Figure 6. In a laboratory experiment, Williamson fertilized eggs of a sea squirt, Ascidia mentula, with sperm of a sea urchin, Echinus esculentus. Surviving three-year-old sea urchins from this experiment—two showing fivefold radial symmetry typical of a sea urchin, but the other two displaying fourfold symmetry—are shown in a 1993 photograph. At right is a pluteus larva from the experiment, photographed at one month. (Photographs courtesy of the author, republished by permission of Springer Science and Business Media. )

Ribosomal RNA genes are the most animal morphology, it is also compatible a new opening; several echinoderms conserved genetic material across all of with this molecular evidence. that have lost their larvae develop as life’s kingdoms, and they have come Syvanen’s RNA analysis, it should deuterostomes. If Kirk’s brittle star lost to be used in building “molecular be added, provides genetic evidence of its larva, it must have also adopted a phylogenies” that trace lineages and “widespread parallelisms” across major fundamentally different pattern of cell estimate the rates at which new spe- lineages of higher organisms. Parallel division in the embryo. We believe, cies diverge. Sometimes the diagrams patterns can often be interpreted as con- however, that Kirk’s brittle star has no that emerge are different from those vergent evolution; dolphins and fish, larva because none of its ancestors hy- based on other observations, such as for example, both have body plans suit- bridized with a sea urchin. The heart fossils, life histories or the body shapes ed for swimming. But Syvanen’s work urchin Abatus cordatus and the three of adult organisms. strongly suggests that genes are in fact known species of sea daisies also have Michael Syvanen, currently a medi- transferred across great taxonomic dis- no larvae and develop as protostomes. cal microbiologist at the University of tances in multicellular plants and ani- We suggest that this is the ancestral California, Davis, has been analyzing mals. Larval transfer is probably one of method of echinoderm development, large-scale patterns in the 18S ribo- several mechanisms of genome fusion and the pattern called deuterostomy somal RNA gene for more than two that will eventually be discovered. came with the transferred larvae. decades. Geneticists have deposited Two sea urchins, Lytechinus variega- about 4,500 18S sequences for multicel- The Star of the Larval-Transfer Story tus and Lytechinus verruculatus, are of lular animals into a gene database, and Once you start taking phylogenetic di- the same genus, but the similar adults Syvanen has shared with Williamson agrams apart and looking at them from each develop from very different plu- some of the phylogenetic patterns seen the larva’s viewpoint, other anomalies teus larvae. Such cases are difficult to in his analysis. A cladogram, or phylo- in echinoderm development begin to explain if larvae and their correspond- genetic diagram, derived from the 18S find explanation. Brittle stars and sea ing adults evolved from one common sequences shows the same time series urchins are very different as adults but ancestor. They are explainable, how- of emergence: acorn worm, sea cucum- share the unique pluteus form of larva. ever, if the larvae were acquired by hy- ber, starfish, sea urchin. Although this Why would similar larvae produce dis- bridization, and the two similar adults order of appearance is the same shown similar adults? Larval transfer propos- hybridized comparatively recently by the larvae of these groups, it bears es that the basic pluteus larva evolved with different species. no relation to the evolutionary history only once, in a sea urchin, and was Perhaps the strangest anomaly is the of the adults told by other evidence. retained in this sea urchin’s descend- starfish Luidia sarsi, which decorates The larval-transfer hypothesis ex- ants. An ancestor of most existing brit- the cover of Williamson’s 2003 book. plains this pattern simply: The 18S ribo- tle stars then acquired a pluteus larva As in other starfish, the fertilized egg somal RNA has been transferred several by hybridizing with a sea urchin. develops into a bilateral larva with a times between taxa. In the case of hemi- One member of this group, Kirk’s small radial juvenile inside. The juve- chordates and echinoderms, it appears brittle star, develops directly from nile then migrates to the outside of the to have been transferred at the same the fertilized egg, with no trace of a larva. In most starfish the larva would time as genes specifying larval form. Hy- bilateral larva, and its blastopore be- then settle and degenerate, leaving the bridization is the most plausible method comes a mouth. It is, therefore, a pro- juvenile to crawl away. In L. sarsi, how- for the simultaneous transfer of these tostome. All echinoderm larvae are in ever, the juvenile drops off the swim- ribosomal and other nuclear genes. Al- a different developmental class, the ming larva, and both continue to live though the hypothesis was founded on deuterostomes, in which the mouth is independently for months.

© 2007 Sigma Xi, The Scientific Research Society. Reproduction 514 American Scientist, Volume 95 with permission only. Contact [email protected]. These two very unlike organisms are 1MBODUPTQIBFSPNPSQIB the same individual, hatched from the QMBODUPTQIFSFT same fertilized egg! How can a single EFVUFSPTUPNF

UPSOBSJB )FNJDIPSEBUB individual split into dissimilar animals, BODFTUPS &OUFSPQOFVTUB both having the same genome? Accord- BDPSO
XPSNT ing to the larval-transfer hypothesis, USPDIPQIPSF the coexisting swimming larva and 1UFSPCSBODIJB crawling starfish of Luidia sarsi evolved QUFSPCSBODIT by the fusion of genomes of two ani- BVSJDVMBSJB mals that hybridized to produce one )PMPUIVSPNPSQIB animal with a larva. The two genomes, TFB
DVDVNCFST however, have retained a considerable EPMJPMBSJB degree of independence, and both are QMVUFVT expressed together during develop- &DIJOPNPSQIB TFB
VSDIJOT ment. This is an example of overlapping &DIJOPEFSNBUB metamorphosis, which also takes place CJQJOOBSJB in other echinoderms, in doliolid salps "TUFSPNPSQIB and in polychaete worms, although the TUBSGJTI overlap does not last as long as in Lu- idia sarsi. Because the larva and juvenile QSPUPTUPNF 0QIJVSPNPSQIB exist side by side, the larva does not BODFTUPS CSJUUMF
TUBST “develop into” the juvenile, as would QMVUFVT be expected if adults and larvae had $PODFOUSJDZDMPNPSQIB evolved from a common ancestor. TFB
EBJTJFT $SJOPNPSQIB Other Views on Larval Evolution EPMJPMBSJB TFB
MJMJFT If you haven’t tuned in to debates on the evolution-of-development (“evo-devo”) frontier lately, you might expect that the Figure 7. Although a test for genetic evidence of hybridization from the experiment in Figure 6 major questions about larval evolution did not find telltale sequences of sea-squirt DNA, more general molecular evidence supporting have accepted answers. In fact the larval- a sequence of larval transfers across hemichordate and echinoderm species comes from Michael transfer hypothesis is one of several pro- Syvanen’s analysis of large-scale patterns in 18S ribosomal RNA, which provides a way to trace posals in an area of biology where lively lineages and estimate divergence. The resulting cladogram, or tree of hypothetical relationships, indicates the same order of species appearance that Williamson’s proposed larval transfers, debate may continue for some time. shown by blue arrows in this diagram, would suggest. This can be explained if the 18S ribos- As mentioned above, most literature omal RNA was transferred at the same time as genes specifying larval form. More generally, on the evolution of animals with lar- Syvanen’s work strongly suggests that genes are transferred across large taxonomic distances vae assumes that an organism started in higher animals. Protostomes and deuterostomes are groups of organisms classified by their out without a larval stage, then through developmental pattern. In the former the larval blastopore becomes the animal’s mouth; in the evolution developed one as a means of latter the mouth is a new opening. (Adapted from diagram by Donald I. Williamson.) survival and propagation. For example, the nematode Caenorhabditis elegans is lar larvae of diverse groups look simi- ited in their ability to divide and unable capable of producing a slightly differ- lar because they face the same survival to differentiate) and patches of distinct ent body plan if stressed during de- tasks. But there is no perfect planktonic “set-aside cells” that will create the rudi- velopment. It will revert to the normal shape to converge to. Marine plankton ment of the adult body. In what is called body plan when the stress is removed. take on an amazing diversity of shapes, indirect development—a life history Genes have been located for this di- all apparently adapted to the environ- that essentially jumps the track to pass vergent body plan, and it has been hy- ment. One wonders how Luidia sarsi, through a larval stage—the larva pro- pothesized by Birgit Gerisch and her the little starfish that separates from its vides a life-support system for the rudi- colleagues at the Max Planck Institute larva and both continue to live, fits the ment, which develops within the larva for Molecular Genetics in Berlin and by paradigm of a single organism evolving or appended to it. After metamorphosis John Wang and Stuart K. Kim at Stan- a larva. This starfish seems to proclaim the larva-specific structures are lost. ford University that continued environ- its double parentage for all to see. Examining patterns of set-aside cells mental changes might cause the two Another hypothesis might be called and genes across several phyla, the body plans to diverge so much that one a larvae-came-first proposal. (It is more Caltech group concluded in 1997 that eventually becomes the larva that will commonly called a “larva-like ancestor” larval forms are widely homologous— later develop into the adult body plan. hypothesis because it is not accepted evolutionarily closely related—across Such plasticity reminds us of how flex- that larvae can be acquired through hy- several phyla, representing regula- ible animal genomes can be, and how bridization.) This proposal, from Kevin tory programs for development that much diversity is possible. J. Peterson, Eric H. Davidson and col- are still being used by their modern The most common explanation for leagues at the California Institute of descendants. Applying conventional the similarities among larvae of such Technology, arises from the fact that at evolutionary logic, their analysis sug- different phyla as mollusks and anne- the cellular level larvae can be seen to gests that larval forms were part of the lids is convergent evolution: The simi- be constructed of larval-only cells (lim- primitive ancestry of creatures from

© 2007 Sigma Xi, The Scientific Research Society. Reproduction www.americanscientist.org 2007 November–December 515 with permission only. Contact [email protected]. bridization. This is not easy work to do; the picture is complicated by the for- BOUFSJPS
DPFMPN mation of serial chimeras, since larval JOWBHJOBUJOH BEVMU
SVEJNFOU transfer suggests that many present-day PSBM
FDUPEFSN animals came about through not just one hybridization, but perhaps many as NJEEMF
DPFMPN MBSWBM seen in the echinoderms. NJEHVU MBSWBM
HVU If the tree of life is not a tree but a tan- gled web of intertwining relationships, QPTUFSJPS QPTUFSJPS DPFMPN the DNA evidence might be as confusing DPFMPN to evolutionists as the 18S ribosomal RNA data. But powerful tools are now avail- QSPCPTDJT
EJTD QSPCPTDJT
EJTD able to apply to such questions. Just last year the sequencing of the genome of the DFQIBMJD
EJTD BNOJPUJD
DBWJUZ purple sea urchin was completed—a full CMBTUPDPFM
XJUI century after Theodor Boveri, using sea- NFTFODIZNF DFQIBMJD
EJTD urchin eggs, determined that normal em- DFMMT bryonic development requires that every DFSFCSBM
EJTD EFWFMPQJOH
BEVMU cell have a full complement of chromo- NVTDVMBUVSF somes. We can only hope that the search USVOL
EJTD GVTFE
DFSFCSBM for larval origins in this exquisite echino- USVOL
BOE derm will not take another century. NJEHVU EPSTBM
EJTDT EPSTBM
EJTD BNOJPO Problems Like the hotly debated hypotheses above, Figure 8. Among the recent proposals for answering questions in larval evolution is one involv- larval transfer raises many questions that ing so-called set-aside cells. Kevin J. Peterson, Eric H. Davidson and their colleagues point out do not yet have answers. Strathmann has that larvae are made up of larval-only cells and patches of distinct cells (the “set-aside cells”) pointed out the grave difficulties imped- that are capable of differentiating and will create the rudiment of the adult body. Here are ing hybridization. These obstacles include shown in color the areas of set-aside cells in cross-sectional diagrams of a sea-urchin larva (top) sperm-egg binding and the influence of and a nemertean worm (bottom). Peterson and colleagues suggest that patterns of set-aside cells and related genes support the idea that larvae are primitive, part of the deep ancestry of the egg cytoplasm on gene expression many phyla. Williamson sees these cells differently—in the case of the sea urchin, as a legacy during early development. Indeed the that came with the acquisition of tornaria-like larvae. (Adapted from Peterson et al. 1997, used mechanisms of fertilization and embryo- with permission of Wiley-Liss/John Wiley & Sons.) genesis differ from one organism to an- other, and animal genomes do not readily sea urchins to leeches to insects. In fies these as a legacy acquired with tor- fuse between taxonomic groups. 2000, Peterson and Davidson proposed naria-like larvae, transferred by hybrid- These barriers, like much else in evo- that the evolution of bilateral symme- ization from an acorn worm. Lacking lution, are unlikely to be insurmount- try must have followed “a long prior larvae, ancestral echinoderms would able, although they can certainly ex- history of bilateral micrometazoans have extended their five arms from the plain why it would be exceedingly similar in grade of organism to the pri- gastrula, a structure formed during an hard to achieve hybridization and mary larvae of modern bilaterians.” intermediate stage in embryonic devel- subsequent survival to adulthood in a In 2003, Belinda J. Sly and her col- opment, just as Kirk’s brittle star does laboratory experiment. Chance fertil- leagues at Indiana University presented today. Modern echinoderms with larvae ization between unlike animals is prob- a sharp critique of the larval-form-first substitute a larval coelomic sac of ap- ably infrequent, but it has taken place. idea and noted evidence showing that propriate size for a gastrula. The cells Unfertilized insect eggs are occasionally larvae appear over short spans of evo- from which a radial echinoderm grows spawned by sperm of different species. lutionary time. In Sly’s scenario, lar- were not “set aside” but resulted from a Onychophoran fertilization is likewise vae have evolved by co-opting genes chance hybridization between an acorn external. In the sea, eggs and sperm are from the adult genome to function in worm and an echinoderm. often cast into open water where fertil- the larva. Many of the same patterns A way to differentiate between these izations between unlike animals seem in genomes and in the evolutionary re- views and those that involve hybridiza- more probable. Most of the evidence cord that support Sly’s proposal could tion might be a further comparison of for the larval-transfer hypothesis comes be explained by larval transfer. the DNA sequences of larval and adult from these marine animals, and it is not Peterson and his colleagues devel- forms. Similar larvae of different adults hard to come up with an equally long oped their “set aside” hypothesis from would be expected to have similar genes list of problems with same-stock theo- the development of echinoderms. Wil- that they received from their common ries. As Brian K. Hall and Marvalee H. liamson explains echinoderm meta- ancestor during hybridization. Closely Wake wrote in introducing their 1999 morphosis quite differently. The cells in related adults that have different larvae, volume on larval forms, question are undifferentiated mesoderm on the other hand, should be expected to cells—stem cells—lining the coelomic sacs show a disparity in the genes expressed Despite a long history of research, (body cavities) in echinoderm larvae. in the larval stage if they indeed came however, even such fundamental The larval-transfer hypothesis identi- from two different sources through hy- questions as “what is a larva?” and

© 2007 Sigma Xi, The Scientific Research Society. Reproduction 516 American Scientist, Volume 95 with permission only. Contact [email protected]. “what is metamorphosis?” occupy mollusks and annelids acquired trocho- Peterson, K. J., R. A. Cameron and E. H. David- us today as they occupied natural- phore larvae. Rotifers, planctospheres son. 1997. Set-aside cells in maximal indirect development: Evolutionary and develop- ists, zoologists, marine biologists, and onychophorans are not persistent mental significance. BioEssays 19:623–631. and evolutionary biologists over a larvae; rather, they have been the source Sly, B. J., M. S. Snoke and R. A. Raff. 2003. century and a half ago. of larvae acquired by other organisms. Who came first—larvae or adults? Origins of bilaterian metazoan larvae. International What Might Larval Transfer Imply? Bibliography Journal of Developmental Biology 47:623–632. Strathmann, R. R. 1993. Larvae and Evolution: If, as Williamson proposes, larvae were Abouheif, E., R. Zardoya and A. Meyer. 1998. Toward a New Zoology (book review). The later additions to life histories, the ear- Limitations of metazoan 18S rRNA sequence Quarterly Review of Biology 68:280–282. liest animals cannot have had larvae. data: Implications for reconstructing a phyl- When a successful hybridization oc- ogeny of the animal kingdom and inferring Wang, J., and S. K. Kim. 2003. Global analysis the reality of the Cambrian explosion. Journal of dauer gene expression in Caenorahbditis curred, the resulting chimera had the of Molecular Evolution 47:394–405. elegans. Development 130:1621–1634. benefits that each animal had acquired Balfour, F. M. 1880–81. A Treatise on Comparative Williamson, D. I. 1988. Incongruous larvae and through years of natural selection, along Embryology. 2 vols. London: Macmillan. the origin of some invertebrate life-histories. Progress in Oceanography 19:87–116. with the new benefits of an early feed- Darwin, C. 1859. The Origin of Species by Means ing stage coupled with a later reproduc- of Natural Selection. London: John Murray. Williamson, D. I. 2001. Larval transfer and the origins of larvae. Zoological Journal of the tive stage. These benefits gave it the po- Barnes, R. S. K., et al. 2001. The Invertebrates: A Linnean Society 131:111–122. tential to make many more of its kind, Synthesis. Oxford, U.K.: Blackwell Science. Williamson, D. I. 2002. Larval transfer in evolu- resulting in the diversity seen today of Eldredge, N., and S. J. Gould. 1972. Punctu- tion. In Horizontal Gene Transfer, ed. M. Sy- animals with larvae. Occasional success- ated equilibria: An alternative to phyletic vanen and C. Kado (2nd edition), pp. 395–410. gradualism. In Models in Paleobiology, ed. ful hybridizations between adults oc- London and New York: Academic Press. curred in the seas where egg and sperm T. J. M. Schopf, pp. 82–115. San Francisco: Freeman, Cooper. Williamson, D. I. 2003. The Origins of Larvae. were randomly distributed, but this ac- Dordrecht: Kluwer. Gerisch, B., C. Weitzel, C. Kober-Eisermann, tivity was not limited to the distant past. V. Rottiers and A. Antebi. 2001. A hormonal Williamson, D. I. 2006. Hybridization in the Brittle stars in the same genus display- signaling pathway influencing C. elegans evolution of animal form and life-cycle. ing different larval forms indicate a re- metabolism, reproductive development, Zoological Journal of the Linnean Society cent hybridization. This would indicate and life span. Developmental Cell 1:841–858. 148:585–602. that even close relatives may not have Haeckel, E. 1866. Generelle Morphologie der Or- Williamson, D. I. 2006. The origins of crusta- a common ancestor. The web, not tree, ganismen: Algemeine Grundzügge der orga- cean larvae. In Treatise on Zoology, The Crus- nischen Formen-Wissenschaft, mechanisch be- tacea, Vol. 2, ed. J. Forest and J. C. Vaupel of life has fusing branches deep in time gründet durch die von Charles Darwin reformirte Klein, pp. 461–482. Leiden: Brill. that involve many phyla, and fusions Descendenz-Theorie. Berlin: Georg Reimer. Williamson, D. I., and A. L. Rice. 1996. Lar- that have occurred recently that sepa- Hall, B. K., and M. H. Wake, eds. 1999. The Ori- val evolution in the Crustacea. Crustaceana rate even species within a genus. gin and Evolution of Larval Forms. San Diego 69:267–287. Larval transfer shows how larvae and London: Academic Press. originated, and it explains the distri- Hart, M. W. 1996. Testing cold fusion of phyla: bution of types of larvae in the animal Maternity in a tunicate × sea urchin hybrid For relevant Web links, kingdom. This distribution is indepen- determined from DNA comparisons. Evolu- tion 50:1713–1718. consult this issue of dent of the phylogeny of adults, so the MacBride, E. W. 1914. Text-Book of Embryology. American Scientist Online: phylogenetic trees that have been almost Vol. 1, Invertebrata. London: MacMillan. universally accepted since the early 20th http://www.americanscientist.org/ Margulis, L. 1993. Symbiosis in Cell Evolution: Issue TOC/issue/1021 century are fatally flawed. Because the Microbial Communities in Archean and Prot- larvae were later additions to their life erozoic Eons. San Diego: W. H. Freeman. histories, mollusks, annelids and other so-called trochophorate phyla did not, in this revised view, evolve from a com- mon ancestor with trochophore larvae, and, similarly, echinoderms and chor- dates did not spring from a common ancestor with tornaria larvae. We can now answer the question posed by Sly and her colleagues: “Who came first—larvae or adults?” We are convinced that larvae were later ad- ditions to life-histories, so there were mollusks and echinoderms before ei- ther had a larva. The basic features of larvae, however, must have evolved long before animals with larvae existed, just as Margulis pointed out that “the functions now performed by cell organ- elles are thought to have evolved long before eukaryotic cells existed.” In the case of larvae, there were rotifers before

© 2007 Sigma Xi, The Scientific Research Society. Reproduction www.americanscientist.org 2007 November–December 517 with permission only. Contact [email protected].