Science funding Climate regulation Human rights

Pernambuco modeledcarpel. refugiaWhen in black. (A)Nixon H. enclosed Kevin and their inner structures, allowing the fossil to camps, depending on how the evolutionary refugium albomarginatus, (B) H. semilineatus, colleagues at Cornell University compared remain intact while Friis peers inside it trees were26 constructed. Evidence for Ecological Speciation and Its Alternative (C) H. faber. Note the absence of large 1171 www.sciencemag.org SCIENCE VOL 323 27 FEBRUARY 2009 its stable traits regions with those same traits in 173 living from many angles ( Science , 7 December In the mid-1980s, Peter Crane, now at Dolph Schluter Bahia refugium in the southern portion plants, Archaefructus came out as a sister to 2007, p. 1546). We can get fantastic resothe University of Chicago in Illinois, pro* * Science 6 February 2009 323: 737-741 of the forest (south of the Bahia and living closer the com- lution, says Friis. Its really exciting. But posed a solution, the anthophyte hypothesis. So angiosperms Paulo refugia)and relative to to the mon ancestor than even Amborella . so far, the flowers Friis finds are Using several evidenceSpecies and noting central and northern areas. Asterisks 30 lines The of Bacterial S o Paulo refugium Challenge: Making Sense of Genetic Archaefructus s distinction that both Bennettitales and denote refugia inferred beyond was the short- too diverse to trace back to a Gnetales and Ecological Diversity current ranges Within of the target species. lived, however. months, better dat- particular ancestor. From organize their male 5.4% and Christophe Fraser, together Eric J. Alm, Martin F. Polz, et al. Symbols localities sampled for found these fossils, we cannot ing of the indicate sediments in which it was female organs Science 6 February 2009 323: 741-746 molecular analysis. Scale bar, 400 km. yielded younger dates, putting this f irst say what is the basic in what could be con(Bottom) The 50% majority-rule con- fossil form, she says. flower squarely with other early strued as a preflower, sensus Bayesian phylogenetic trees, 7% 36 Stability Predicts Genetic flower parts, about 125 million years old. he considered them, Diversity in the Brazilian Atlantic 7.8% rooted with sequences from the othAlso, a 2009 phylogenetic analysis of Before flowers along with angiosperms, Forest Hotspot er two congeneric species studied 67 (root taxanot by shown). Doyle Thick and Peter Endress they have yet as comprising a single Ana Carolina Carnaval, Michael J. Hickerson, Clio F. B. Haddad, et al. page Although 36 internodes de- of the 5.3 University ofwith Zurich, Switzerland, angiosperm entity called note clades posterior probability placed to find the oldest fossil Science 6 February 323: 785-789 4% 2009 Larger than 5.8% life. Although merely thegreater fossilthan in with liliesindicate rather than at flowers, researchers anthophytes. For the next 90%.water Percentages 2.2 millimeters in diameter, this 3D corrected distances between theTamura-Nei base of the angiosperms, although this assume that the ancesdecade, most family trees fossil flower shows that grasses date Around the world, governments turn to AAAS as an objective, multidisciplinary scientic authority to clades (20). conclusion is contested. tral angiosperm evolved based image: on morphology supback to 94 million years ago. Cover Tui De Roy/Minden Pictures/FLPA 5.6% educate public ofcials and judicial gures on todays most pressing issues. Our goal is to promote Inset: George Richmond/Bridgeman Art Library, London (Superstock) informed policy decisions that benet society. And this is just one of the ways that AAAS is committed to www.sciencemag.org SCIENCE VOL 324 3 APRIL 2009 29 advancing science to support a healthy and prosperous world. Join us. Together we can make a difference. aaas.org/plusyou/policy

CREDITS (TOP TO BOTTOM): COURTESY OF STEPHEN MCCABE, UC SANTA CRUZ; PHOTO BY JENNIFER SVITKO, COURTESY OF WILLIAM L. CREPET, CORNELL UNIVERSITY

19th-century idealistic morphologists such natural-theological assumptions about a per- gram as one not dominated by a typological as Carl F. Kielmeyer and J. F. Meckel sonified God who had created a perfectly and linear-recapitulationist mindset but 0403NewsFocus.qxp 3/30/09 5:23 that PM Page 29 retained their teleology, their typological adapted nature. Bronns translation, though it rather as continuing to wrestle with the need emphasis on form, and their linear recapitula- altered key ideas to make Darwin comprehen- to account for variability and unpredictable tionism. This story, emphasizing the long per- sible to a German academic audience, was not change in terms of mechanistic laws of sistence of a German transcendental approach a conservative throwback. It represented the natureamong which Haeckel included, at to nature, has been deeply entrenched in the dynamic engagement of a leading paleontolo- the top of his list, natural selection. Haeckels Introduction history of biology. gist who had also long been working on many Darwinism thus shows continuity with early Gliboff challenges this history right from of the questions Darwin claimed as his own 19th-century concerns, mediated through Stockholm. When we started, 2 Darwins Inspiration, Darwins Legacy the beginning. ascription of simple theThe search profile was bigger, linear a critical yet generous equal, who saw himself Bronn. But those concerns were always more Andrew Sugden a magnolia she recapitulationism to the[flower], views of Romantic as moving science forward through the modi- flexible than has been acknowledged, and recalls. But 30 years ago,to she embryologists, he notes, owes much a carica- fications he made to Darwins flawed theory. their articulation changed over time. Of and by others discovered ture developed Karl Ernst vontiny Baer in a Bronns death in 1862 afforded him little course Haeckels Darwinism was not Articles ancient flowers by sieving polemical context, then adopted uncritically by chance to steer the conversation further. Darwins own, but it was not an aberration or through sand and clay sedi3 of influential historians such as E. S. a distortion some true theory, any On the Origin of Life onmore Earth ments. With this technique, Russell and they Stephen Jay collected Gould. hunthan any other post-Darwinian additions or Carl Zimmer have now Science 9 January 2009 323: 198-199 Gliboffs fresh reading the origadjustments were. It was science moving on. dreds of of millimeter-size some preserved in inal sources flowers, interprets Kielmeyer Gliboff s overall picture of scientific dimensions, from Por5 contrast On the of Art and Symbolism and Meckelthree as far less rigidly advance, in to Origin Richardss emphasis on tugal and other locations andMichael Balter typological in their orientation and with charisma passion, is one of scientists Cretaceous deposits 70 milScience 6 February 2009 323: 709-711 much more lion attentive to natures building and innovating incrementally, workto 120 million years old. variability than has been seen ing with what their predecessors have handed This fossil diversity 8 On the Origin of Photosynthesis before. Bothshows for these them and sculpting it into something new yet that early-19thangiosperms were Mitch Leslie century naturalists and for their understandable to those around them. His senthriving, with several groups Science 6 March 2009 323: 1286-1287 well-established, by 100 milintellectual heirs, Gliboff argues, sitive reading allows Out us of tothe see post-1859 past. page 10 lion years ago. In some, the Tiny Amborella sits the critical issue was to understand German evolutionists as rational actors rather 10 On theat Origin of of Flowering Plants flower parts are whorled the the natures manifold variety while like than irrationally stuck inbottom some early-19th Elizabeth Pennisi angiosperm family tree. those of modern flowers; in seeking out underlying strict natucentury moment with unmodern commitothers they are spiraled, con Science 3 April 2009 324: 28-31 ral laws to account for it. ments. By challenging the very foundations of sidered by some researchers This provides newprimitive starting arrangement. Some the standard narrative of von German morpholas the a more from one 14 of the nonflowering seed plants Alexander Humboldt and the General Physics point for analyzing Darwins first ogy, this careful, compelling account does at flower fossils have prescribed numbers of We are realizing that this or gymnosperms, whose heyday was 200 of the Earth petals, anotherpaleonmodern feature, whereas in ago. Modern gymnosperms translator, the prominent least asyears much as Richardss to undermine the Stephen T. Jackson huge diversity is probably million others the petal count include conifers, ginkgoes, and the cycads, tologist H. G. Bronna figure lit-varies. association of 19th-century German Science 1 May 2009 324:Darwin596-597 Inin 1998, geologist Ge Sun of with their a stout trunks and large fronds. tle attended to the Chinese standard ism with dangerously exceptional view of the result of one innovaJilin University in Changchun, China, came Before angiosperms came along, these story but the lynchpin of Glinature. But two books offer very different 16 the Making German Evolution: Translation and Tragedy across what seemed to be a much older tion piled on top of plants were much more diverse and boffs. Intriguingly andfossil, plausibly, reads. Is cycadlike scientific progress a matter of perLynn K. Nyhartsuch flower. The called Archaefructus, was included species, as the Gliboff argues that Bronns use sonal anguish and triumph, or of intellectual Science 27 February 2009 323: 1170-1171 an aquatic plant that looked to be 144 mil- another innovation. extinct Bennettitales, and many woody REPORTS of terms like vervollkommnet chugging Our concept it should be lion years old. By 2002, Sun and David plants called ofof which along?Gnetales, population expansion (23Originality ) areboth. found the in the sites located a presence two lineages that co-occur in the ad- 6.2% divergence). In contrast, Peter Crane, of of the Florida of Natural few representatives, including 18 Darwins (perfect) as Dilcher translations for Dar- Museum capacious enough to include unstable for H.today albomarginatus and H. faber, the refugia are genetically joint jacent refugia. In all species, average nethad nucle- outside (south of) History (FLMNH) in Gainesville University of Chicago firs, area survive (see family tree, Peter J. Bowler wins improved or favored well as incommon the Bahia in refugium area for were H. faber otide differences across from localities (22 )flowreflects more similar to each other, although to a lesser p. as described an entire plant, roots to 31). Also the Jurassic References and Notes Science 9 January 323: 223-226 were not abouthigh dragging Darwin within refugia (2.6 to extent in H. faber (0.1 to 1.6%). Signatures of and H. semilineatus . The lack gone; of 2009 signature geographic ers, entombed on astructure slab of rock unearthed in These fossils often spark debate because seed ferns, a group now long their of 1. E. Haeckel, Generelle Morphologie der Organismen backward into a German teleopage 16 Liaoning in northeastern China. specimens tend to be imperfectly preserved most(Georg famous member is Caytonia, which Reimer, Berlin, 1866). 22 The Red Queen and the Court Jester: Species Diversity logical view of nature (as has in putative 2. Theto reviewer previously served as astructures. press reader for both In one Archaefructus wasnt much and leave room for interpretation. To help seems have precarpel-like Fig. 2. sense, Genetic diversity C A B and the Role of Biotic and Abiotic Factors Through Time books at the manuscript stage. been claimed by those who have (stable) unstableplant areas before remedy that, Friis and her colleagues have These g roups perceived relevance to to refugial look at. Its versus a flowering Michael J. Benton in the Brazilian Atlantic rainforest. there flowers, Dilcher notes. It lacked paid attention towere Bronn at all). begun to examine flowers using synchroflower evolution and their relationships to A painter, too. Haeckels oil landscape of highlands in Java, from (Top ) and Species-specific stability maps; Science 6 February 2009 323: 728-732 petals sepals, it did have an tron radiation to generate a 3D image of angiosperms have ping-ponged between 10.1126/science.1169621 Instead, Gliboff asserts, Bronnsbut Wanderbilder (1905).

Contents

ORIGINS

CREDIT: ERNST HAECKEL/FROM WANDERBILDER (W. KOEHLER, GERA-UNTERMHAUS, 1905)

2009 by The American Association for the Advancement of Science. All rights reserved.

NEWSFOCUS

Darwins Inspiration, Darwins Legacy

The day has passed delightfully. Delight itself, however, is a weak term to express the feeling of a naturalist who, for the first time, has wandered by himself in a Brazilian forest. The elegance of the grasses, the novelty of the parasitical plants, the beauty of the flowers, the glossy green of the foliage, but above all the general luxuriance of the vegetation, filled me with admiration. A most paradoxical mixture of sound and silence pervades the shady parts of the wood. The noise from the insects is so loud, that it may be heard even in a vessel anchored several hundred yards from the shore; yet within the recesses of the forest a universal silence appears to reign. To a person fond of natural history, such a day as this brings with it a deeper pleasure than he can ever hope to experience again. Charles Darwin, The Voyage of the Beagle, Feb 29th [1832]

Introduction

On the Origin of of

Life on Earth
in some warm little pond, with all sorts of ammonia and phosphoric salts, light, heat, electricity, etc., present, that a protein compound was chemically formed ready to undergo still more complex changes, at the present day such matter would be instantly devoured or absorbed, which would not have been the case before living creatures were formed. Scientists today who study the origin of life do not share Darwins pessimism about our ability to reconstruct those early moments. Now is a good time to be doing this research, because the prospects for success are greater than they have ever been, says John Sutherland, a chemist at the University of Manchester in the United Kingdom. He and others are addressing each of the steps involved in the transition to life: where the raw materials came from, how complex organic molecules such as RNA formed, and how the first cells arose. In doing so, they are inching their way toward making life from scratch. When I was in graduate school, people thought investigating the origin of life was something old scientists did at the end of their career, when they could sit in an armchair and speculate, says Henderson James Cleaves of the Carnegie Institution for Science in Washington, D.C. Now making an artificial cell doesnt sound like science fiction any more. Its a reasonable pursuit.

The collection reprinted here is a sample of the articles published in 2009 by Science magazine in celebration of the Darwin bicentenary. We start with four of the essays from our On the Origin of series, prepared by Sciences news writers; further essays in this series are appearing monthly in Science throughout the year. A Perspective by Stephen Jackson then considers the legacy of Alexander von Humboldt, for whom, like Darwin, the South American tropics were a critical inspiration, and who died 150 years ago in the year of the publication of Darwins Origin. (Humboldts Personal Narrative of his tropical explorations was acknowledged by Darwin as far exceed[ing] in merit anything I have read on the subject.) A book review by Lynn Nyhart explores two recent volumes on Ernst Haeckels work, his interpretations of Darwin and his contributions to evolutionary thought. In the first of four Review articles reprinted here, Peter Bowler analyzes the originality of Darwins contribution to the understanding of the diversity and diversification of the living world. Michael Benton

Finally, with a focus on conservation, a Report by Ana Carnaval et al., who model evolutionary processes in endemic tree-frog species in the Brazilian Atlantic Forest, the very biodiversity hotspot that so inspired Darwin on his South American landfall, and that is now reduced to a collection of small fragments scattered along the coast. Darwin returned to the Brazilian coast on his final homeward leg, more than four years after his first landfall there. His enthusiasm for the tropical forested landscape was undiminished. In my last walk I stopped again and again to gaze on these beauties, and endeavoured to fix in my mind for ever, an impression which at the time I knew sooner or later must fail they will leave, like a tale heard in childhood, a picture full of indistinct, but most beautiful figures. Charles Darwin, The Voyage of the Beagle, August 1836 Andrew Sugden, Deputy Editor
CREDIT: KATHARINE SUTLIFF/SCIENCE

DARWIN

Step 1: Make RNA An RNA molecule is a chain of linked nucleotides. Each nucleotide in turn consists of three parts: a base (which functions as a

198

9 JANUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

CREDITS (TOP TO BOTTOM): KATHARINE SUTLIFF/SCIENCE; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

ike many other scientists raised in temperate latitudes, Charles Darwin was enthralled by his first glimpse of the tropical rain forest. His Beagle diary entry conveyed those immediate and thrilling first impressions, but the encounter with the Brazilian Atlantic Forest had an enduring influence on the development of his ideas over the following decades, with resounding echoes even today in 21st century evolutionary science.

examines the extent to which biotic and abiotic factors have shaped species diversity in the fossil record. Dolph Schluter reviews how research on speciation has shifted in focus from morphological evolution to reproductive isolation, tracing the links between Darwins ideas and current thinking. Christophe Fraser and colleagues discuss the contentious area of microbial species formation, an issue that would surely have vexed Darwin horribly had the bewildering diversity of microbes been known in his day.

AN AMAZON OF WORDS FLOWED FROM Charles Darwins pen. His books covered the gamut from barnacles to orchids, from geology to domestication. At the same time, he filled notebooks with his ruminations and scribbled thousands of letters packed with observations and speculations on nature. Yet Darwin dedicated only a few words of his great verbal flood to one of the biggest questions in all of biology: how life began. The only words he published in a book appeared near the end of On the Origin of Raw ingredients Species: Probably all the organic beings which Lifeor at least life as we know itappears to have ever lived on this earth have descended have emerged on Earth only once. Just about all from some one primordial form, into which life organisms use double-stranded DNA to encode was first breathed, Darwin wrote. genetic information, for example. They copy Darwin believed that life likely emerged their genes into RNA and then translate RNA spontaneously from the chemicals into proteins. The genetic code it is made of today, such as carbon, THE YEAR OF they use to translate DNA into pronitrogen, and phosphorus. But he teins is identical, whether they are The English did not publish these musings. emus or bread mold. The simplest naturalist had The English naturalist had built explanation for this shared biology his argument for evolution, in is that all living things inherited it built his argularge part, on the processes he from a common ancestor ment for evocould observe around him. He did namely, DNA-based microbes that lution, in large lived more than 3.5 billion years not think it would be possible to see life originating now because ago. That common ancestor was part, on the the life thats already here would already fairly complex, and many processes he prevent it from emerging. scientists have wondered how it In 1871, he outlined the prob- This might have evolved from a simpler essay is the first could observe lem in a letter to his friend, botanist in a monthly series, with predecessor. Some now argue that more on evolutionary around him. . Joseph Hooker: But if (and Oh! roots membrane-bound cells with only online at blogs. what a big if!) we could conceive sciencemag.org/origins RNA inside predated both DNA

and proteins. Later, RNA-based life may have evolved the ability to assemble amino acids into proteins. Its a small step, biochemically, for DNA to evolve from RNA. In modern cells, RNA is remarkably versatile. It can sense the levels of various compounds inside a cell and switch genes on and off to adjust these concentrations, for example. It can also join together amino acids, the building blocks of proteins. Thus, the first cells might have tapped RNA for all the tasks on which life depends. For 60 years, researchers have been honing theories about the sources of the amino acids and RNAs building blocks. Over time, they have had to refine their ideas to take into account an ever-clearer understanding of what early Earth was like. In an iconic experiment in 1953, Stanley Miller, then at the University of Chicago, ignited a spark that zapped through a chamber filled with ammonia, methane, and other gases. The spark created a goo rich in amino acids, and, based on his results, Miller suggested that lightning on the early Earth could have created many compounds that would later be assembled into living things. By the 1990s, however, the accumulated evidence indicated that the early Earth was dominated by carbon dioxide, with a pinch of nitrogentwo gases not found in Millers flask. When scientists tried to replicate Millers experiments with carbon dioxide in the mix, their sparks seemed to make almost no amino acids. The raw materials for life would have had to come from elsewhere, they concluded. In 2008, however, lightning began to look promising once again. Cleaves and his colleagues suspected that the failed experiments were flawed because the sparks might have produced nitrogen compounds that destroyed any newly formed amino acids. When they added buffering chemicals that could take up these nitrogen compounds, the experiments generated hundreds of times more amino acids than scientists had previously found. Cleaves suspects that lightning was only one of several ways in which organic compounds built up on Earth. Meteorites that fall to Earth contain amino acids and organic carbon molecules such as formaldehyde. Hydrothermal vents spew out other compounds that could have been incorporated into the first life forms. Raw materials were not an issue, he says: The real hurdle is how you put together organic compounds into a living system.

EVOLUTIONARY ROOTS
letter in a genes recipe), a sugar molecule, and a cluster of phosphorus and oxygen atoms, which link one sugar to the next. For years, researchers have tried in vain to synthesize RNA by producing sugars and bases, joining them together, and then adding phosphates. It just doesnt work, says Sutherland. This failure has led scientists to consider two other hypotheses about how RNA came to be. Cleaves and others think RNA-based life may have evolved from organisms that used a different genetic materialone no longer found in nature. Chemists have been able to use other compounds to build backbones for nucleotides (Science, 17 November 2000, p. 1306). Theyre now investigating whether these humanmade genetic molecules, called PNA and TNA, could have emerged on their own on the early Earth more easily than RNA. According to this hypothesis, RNA evolved later and replaced the earlier molecule. But it could also be that RNA wasnt put together the way scientists have thought. If you want to get from Boston to New York, there is an obvious way to go. But if you cant get there that way, there are other ways you could go, says Sutherland. He and his colleagues have been trying to build RNA from simple organic compounds, such as formaldehyde, that existed on Earth before life began. They find they make better progress toward producing RNA if they combine the components of sugars and the components of bases together instead of separately making complete sugars and bases first. Over the past few years, they have documented almost an entire route from prebiotic molecules to RNA and are preparing to publish even more details of their success. Discovering these new reactions makes Sutherland suspect it wouldnt have been that hard for RNA to emerge directly from an organic soup. Weve got the molecules in our RNA, producing the first protocells. The goal sights, he says. is to have something that can replicate by itself, Sutherland cant say for sure where these using just chemistry, says Szostak. reactions took place on the early Earth, but he After 2 decades, he and his colleagues notes that they work well at the temperatures have come up with RNA molecules that can and pH levels found in ponds. If those ponds build copies of other short RNA molecules. dried up temporarily, They have been able to they would concentrate mix RNA and fatty Now making an the nucleotides, making acids together in such a conditions for life even way that the RNA gets artificial cell doesnt more favorable. trapped in vesicles. The Were these Darwins vesicles are able to add sound like science warm little ponds? It fatty acids to their might just be that he fiction any more. Its membranes and grow. wasnt too far off, says In July 2008, Szostak Sutherland. a reasonable pursuit. reported that he had figured out how protoHENDERSON JAMES CLEAVES, cells could eat and Step 2: The cell CARNEGIE INSTITUTION FOR SCIENCE bring in nucleotides to If life did start out with RNA alone, that RNA build the RNA. would need to make copies of itself without All living cells depend on complicated help from proteins. Online in Science this channels to draw nucleotides across their week (www.sciencemag.org/cgi/content/ membranes, raising the question of how a abstract/1167856), Tracey Lincoln and Ger- primitive protocell membrane brought in these ald Joyce of the Scripps Research Institute in molecules. By experimenting with different San Diego, California, have shown how that recipes for membranes, Szostak and his colmight have been possible. They designed a leagues have come up with protocells leaky pair of RNA molecules that join together and enough to let nucleic acids slip inside, where assemble loose nucleotides to match their they could be assembled into RNA, but not so partner. Once the replication is complete, old porous that the large RNA could slip out. and new RNA molecules separate and join Their experiments also show that these with new partners to form new RNA. In 30 vesicles survive over a 100C range. At high hours, Lincoln and Joyce found, a population temperatures, protocells take in nucleotides of RNA molecules could grow 100 million quickly, and at lower temperatures, Szostak times bigger. found, they build RNA molecules faster. Lincoln and Joyce kept their RNA moleHe speculates that regular temperature cules in beakers. On the early Earth, however, cycles could have helped simple protocells surreplicating RNA might have been packed in the vive on the early Earth. They could draw in first cells. Jack Szostak and his colleagues at nucleotides when they were warm and then use Harvard Medical School in Boston have been them to build RNA when the temperature investigating how fatty acids and other mole- dropped. In Szostaks protocells, nucleotides cules on the early Earth might have trapped are arranged along a template of RNA. Strands of RNA tend to stick together at low temperatures. When the protocell warmed up again, the heat might cause the two strands to pull apart, allowing the new RNA molecule to function. Now Szostak is running experiments to bring his protocells closer to life. He is developing new forms of RNA that may be able to replicate longer molecules faster. For him, the true test of his experiments will be whether his protocells not only grow and reproduce, but evolve. To me, the origin of life and the origin of Darwinian evolution are essentially the same thing, says Szostak. And if Darwin was alive today, he might well be willing to write a lot Protocell. Researchers at more about how life began. Harvard are trying to make
simple life forms, shown here in a computer image.

EVOLUTIONARY ROOTS ORIGINS

letter in a genes recipe), a sugar molecule, soup. Weve got the molecules in our RNA, producing the first protocells. goal Venus, phallus, orThe pebble? and a cluster of phosphorus and oxygen sights, he says. is to have something thatknow can replicate by itself, I dont much about Art, but I know what atoms, which link one sugar to the next. For Sutherland cant say for sure where these using just chemistry, says Szostak. I like, quipped the humorist and art critic years, researchers have tried in vain to synthe- reactions took place on the early Earth, but he After 2 decades, and his colleagues Geletthe Burgess back in 1906. For archaeosize RNA by producing sugars and bases, notes that they work well at the temperatures have come up with RNA molecules that logists, distinguishing art can from nonart is still joining them together, and then adding phos- and pH levels found in ponds. If those ponds build copies of quite otherashort RNA Take molecules. challenge. the 6-centimeter-long phates. It just doesnt work, says Sutherland. dried up temporarily, have been able to piece They of quartzite known as the Venus of This failure has led scientists to consider they would concentrate mix RNA and fatty Tan-Tan. Found in Morocco in 1999 next to a Now making an two other hypotheses about how RNA came to the nucleotides, communicate making acidsof together suchestimated a meaning, whether they be the rich trove stone in tools to be be. Cleaves and others think RNA-based life conditions for life even way that the RNA gets words that make up our languages, the musibetween 300,000 and 500,000 years old, it artificial cell doesnt may have evolved from organisms that used a more favorable. cal sounds that convey emotion, or the dra- resembles trapped in vesicles. The a human figure with stubby arms different genetic materialone no longer Were these Darwins vesicles areBednarik, able to add matic paintings that, 30,000 years after their and legs. Robert an independent sound like science found in nature. Chemists have been able to warm little ponds? It caused the discoverers of the Chau- archaeologist fatty acids to in their creation, based Caulf ield South, use other compounds to build backbones for might just be vet that he to break membranes grow. Cave down in tears.more. Its Australia, insistsand that an ancient human fiction any nucleotides (Science, 17 November 2000, wasnt too far off, While says sites like Chauvet might be vivid deliberately In Julymodified 2008, Szostak the stone to p. 1306). Theyre now investigating whether Sutherland. reported that had a what reasonable pursuit. examples of some researchers still make it look more likehe a person. these humanmade genetic molecules, called figured how protoconsider a creative explosion that began If so, this objetout dart is so old JAMES CLEAVES, PNA and TNA, could have emerged on their Step 2: The cell cells couldnot eat and when modern humans HENDERSON colonized Europe that it was created by our CARNEGIE INSTITUTION FOR SCIENCE own on the early Earth more easily than RNA. If life did start out with bring inwhich nucleotides about 40,000 years ago, an increasing numown species, first to According to this hypothesis, RNA evolved RNA alone, that RNA build the RNA. ber of prehistorians are tracing our sym- appears in Africa nearly later and replaced the earlier molecule. would need to make copies of itself without All living cells depend on complicated bolic roots much further back in timeand 200,000 years ago, but But it could also be that RNA wasnt put help from proteins. Online in Science channels to draw nucleotides across their in some cases, to speciesthis ancestral to Homo by one of our ancestogether the way scientists have thought. If week (www.sciencemag.org/cgi/content/ membranes, raising question sapiens. Like modern humans themselves, tors, the perhaps theof how a you want to get from Boston to New York, abstract/1167856), Tracey Lincolnseems and Gerprimitive protocell membrane brought symbolic behavior to have its origins large-brained H. in these there is an obvious way to go. But if you cant ald Joyce of thein Scripps Research in have molecules. with Africa. Recent Institute excavations turnedBy experimenting heidelbergensis , different get there that way, there are other ways you San Diego, California, have shown how that recipes for membranes, Szostak up elaborate stone tools, beads, and ochre thought by someand his colcould go, says Sutherland. He and his col- might have been possible. They designed a leagues have come up with protocells leaky dating back 100,000 or more years ago, anthropologists to leagues have been trying to build RNA from pair of RNA molecules that join togetherare andstill enough to let nucleic acids slip inside, where although researchers debating be the common simple organic compounds, such as formaldeloose nucleotides to f match their demonstrate they could be assembled RNA, but not so Since their discovery by Frenchassemble spelunkers which of these inds really ancestor into of modhyde, that existed Earth before life began. replication is complete,But old theres porous that the large RNA could slip out. in on 1994, the magnificent lions, partner. horses, Once and the symbolic expression. wideern humans and They find they rhinos make better progress toward andwalls new RNA molecules separate andthe join Their experiments also show that seem to leap from the of spread agreement that building blocks Neandertals. That that these producing RNA if they combine the compo- France with new partners to form newpreceded RNA. In 30 vesicles art. survive over a 100C range. At high Chauvet Cave in southern have of symbolism full-blown would mean that nents of sugars reigned and the components of bases hours,paintLincoln and Joyce found, a population take in nucleotides as the worlds oldest cave When we talk about beads andtemperatures, art, we are protocells art is an extremely together instead of separately comofochre RNA and molecules could growabout 100 million quickly, and at ancient lower temperatures, Szostak ings. Expertly making composed in red actually talking material technologies part of the Homo plete sugars andblack bases charcoal, first. times bigger. for symbolic expression that certainly found, they build RNA molecules faster. the vivid drawings demonpostrepertoire. Ignoring the Symmetry in stone. Over the past few years, they have docuLincoln Joyce kept their RNA mole- thought He speculates regularwe temperature strate that the artistic gift stretches backanddate the origins of symbolic and fewthat specimens have Some stone tools mented almost an entire route from years. prebiotic cules in beakers. On the early Earth, however,by a cycles simple protocells surmore than 30,000 These paintings communication, potentially verycould wide have ofhelped very early paleoart, require a mental molecules to RNA and aresure preparing to pub- replicating RNAmargin, might have been packed in the Dietrich vive on the early Earth. They could draw in to create. are almost to be mentioned in any artisays archaeologist Stout explaining them away, image lish even more cle details of their success. Dis- art. first cells. Jack of Szostak and his colleagues at nucleotides when were warm and then use or paper about the earliest But what University College London. orthey rejecting them out covering these new reactions SutherHarvard School in Boston of have been them build RNA when the temperature do they reallymakes tell us about the originsMedical of The evolution symbolism wasto once of hand does not serve this discipline well, land suspect it wouldnt have been that hard investigating how fatty acids and other protocells, nucleotides artistic expression? thought to have been asmolerapid as dropped. flicking In on Szostaks Bednarik wrote in a 2003 analysis of the for RNA to emerge directly from an organicwho cules on the early Earth might have trapped Clive are arranged a template of RNA. StrandsAnthropology. The prehistoric humans decorated a light switch, as archaeologist Gamble along Venus of Tan-Tan in Current of RNA tend to stick together at low tempera- are skeptical, Chauvets walls by torchlight arrived at the of the Royal Holloway, University of LonYet many archaeologists tures. When again, resemblance the cave with their artistic genius already in full don, put it some years ago. But given newthe protocell arguing warmed that theup stones to a heatappears might cause the two strands to pull flower. And so, most researchers agree that evidence that symbolic behavior human figure might be apart, coincidence. Indeed, the new RNA molecule to Tan-Tan function.figurine is remthe origins of art cannot simply long before cave allowing paintings, the debate over the Now Szostak is running to be pegged to the latest discovery THE YEAR OF Gamble now says that his muchiniscent of aexperiments similar controversy over a Recent bring protocells closer to life. He is develof ancient paintings or sculpcited comment needs to behis modsmaller stone discovered in 1981 at the site of new forms of RNARam that may beIsraeli-occupied able to ture. Some of the earliest art ified: Its a dimmer oping switch now, Berekhat in the Golan excavations molecules him, likely perished over the ages; a stuttering candle. replicate longer Heights. Tofaster. someFor archaeologists, this have turned the true test of his experiments will be much remains to be found; and As they more precisely pin250,000-year-old object resembles a woman, whether his protocells only grow and up elaborate archaeologists dont always point when symbolic behavior but othersnot argue that it was shaped by natural reproduce, but evolve. agree on how to interpret what is began, scientists are hoping they forces, and, in any case, looks more like a stone tools, To me, the penguin origin ofor life and the origin of an exhaustive unearthed. As a result, instead of might one day crack the tougha phallus. Even after beads, and Darwinian evolution are essentially the same chasing after arts first appearest question of all: What was its microscopic study concluded that the thing, says AndRam if Darwin was alive ance, many researchers seek to evolutionary advantage toSzostak. Berekhat object had indeed been etched ochre dating today, he mightwith wellabe willing to write what a lot some consider understand its symbolic roots. humans? Didat symbols, as many tool to emphasize Protocell. Researchers back 100,000 more about how life began. After all, art is an aesthetic This essay continues researchers suspect, serve as a its head and arms, many researchers have Harvard are trying to make our or more series. See more ZIMMER simplesocial life forms, shown expression of something more monthly glue that helped tribes of rejected it as aCARL work of art. For some, proof of on humans evolutionary here in aearly computer image. to survive Carl Zimmer is the symbolic author of Microcosm: E.requires coli and the fundamental: the cognitive abil- journey humans and behavior evidence that the years onlineago. at blogs. New Science of Lifesymbols . ity to construct symbols that sciencemag.org/origins. reproduce? had a commonly understood mean-

On the Origin of of

Art and Symbolism

CREDITS (TOP (LEFTTO TOBOTTOM): RIGHT): KATHERINE SUTLIFF/ SCIENCE; THE BOXGROVE PROJECT CREDITS KATHARINE SUTLIFF/ SCIENCE ; THE BOXGROVE PROJECT; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

DARWIN

CREDIT: JANET IWASA

CARL ZIMMER

Carl Zimmer is the author of Microcosm: E. coli and the New Science of Life.

CREDIT: JANET IWASA

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

199

www.sciencemag.org SCIENCE VOL 323 9 JANUARY www.sciencemag.org SCIENCE VOL 323 2009 6 FEBRUARY 2009

199

5 709

ORIGINS
tools, which Wynn and many others argue are clear examples of an imposed form based on a mental template. Some have even argued that these skillfully crafted hand axes had symbolic meanings, for example to display prestige or even attract members of the opposite sex. The half-million-year mark also heralded the arrival of H. heidelbergensis, which had a much larger brain than H. erectus . Not long afterward, our African ancestors began to create a wide variety of finely crafted blades and projectile points, which allowed them to exploit their environment in more sophisticated ways, and so presumably enhance their survival and reproduction. Archaeologists refer to these tools as Middle Stone Age technology and agree that they did require mental templates. The tools tell us that the hominid world was changing, says Wynn. ing and were shared within groups of people. ity to hold an abstract concept in ones head As one moves forward in time, humans For example, the hundreds of bone and stone and, in the case of the tool, to impose a pre- appear able to imagine and create even more Venus figurines found at sites across Eura- determined form on raw material based on an elaborate tools, sharpening their evolutionsia beginning about 30,000 years ago were abstract mental template. That kind of ability ary edge in the battle for survival. By skillfully carved and follow a common motif. was probably not needed to make the earliest 260,000 years ago, for example, ancient They are widely regarded not only as sym- known tools, say Wynn and other researchers. humans at Twin Rivers in what is now Zambia bolic expression, but full-fledged art. These implements, which date back 2.6 mil- could envision a complex finished tool and Thus many researchers are reluctant to lion years, consist mostly of rocks that have put it together in steps from different compoaccept rare, one-off discoveries like the Tan- been split in two and then sharpened to make nents. They left behind finely made blades Tan or Berekhat Ram objects as signs of simple chopping and scraping implements. and other tools that had been modified symbolic behavior. You can imagine [an Then, about 1.7 million years ago, large, usually by blunting or backing one edge ancient human] recognizing a resemblance teardrop-shaped tools called Acheulean hand to be hafted onto handles, presumably made but [the object] still hav[ing] no symbolic axes appeared in Africa. Likely created by of wood. These so-called backed tools have meaning at all, says Philip Chase, an anthro- H. erectus and probably used to cut plants and been widely regarded as evidence of sympologist at the University of Pennsylvania. butcher animals, these hand-held tools vary bolic behavior when found at much younger Thomas Wynn, an anthropologist at the greatly in shape, and archaeologists have sites. This flexibility in stone tool manufacUniversity of Colorado, Colorado Springs, debated whether creating the earliest ones ture [indicates] symbolic capabilities, says agrees: If its a one-off, I dont think it required an abstract mental template. But by archaeologist Sarah Wurz of the Iziko Musecounts. Its not sending a message to anyone. about 500,000 years ago, ancient humans were ums of Cape Town in South Africa. creating more symmetrical Late Acheulean Similar cognitive abilities were possibly Tools of the imagination required to make the famous Given how difficult it is to detect 400,000-year-old wooden spears If its a one-off, I dont think the earliest symbolic messages in from Schningen, Germany. One the archaeological record, some recent study concludes that these it counts. Its not sending a researchers look instead for proxy spears creatorsprobably membehaviors that might have bers of H. heidelbergensiscarmessage to anyone. required similar cognitive abiliried out at least eight preplanned ties, such as toolmaking. Charles steps spanning several days, THOMAS WYNN, UNIVERSITY OF COLORADO, Darwin himself saw an evoluincluding chopping tree branches COLORADO SPRINGS tionary parallel between toolwith hand axes and shaping the making and language, probably spears with stone flakes. the most sophisticated form of The idea that sophisticated symbolic behavior. To chip a toolmaking and symbolic flint into the rudest tool, Darwin thought require similar cognitive wrote in The Descent of Man , skills also gets some support demands a perfect hand as well from a surprising quarter: brainadapted to that task as the vocal imaging studies. Stouts team ran organs are to speaking. positron emission tomography To many researchers, making scans on three archaeologists sophisticated tools and using Symbolic start. Some scientists argue that this 77,000-year-old engraved ochre all skillful stone knappersas symbols both require the capac- shows symbolic capacity. they made pre-Acheulean and
A roaring start. Researchers agree that Chauvet Caves magnificent paintings, including these lions, are full-blown art.

ORIGINS
Late Acheulean tools. Both methods turned on visual and motor areas of the brain. But only Late Acheulean knapping turned on circuits also linked to language, the team reported last year. Color me red At Twin Rivers, its not just the tools that hint at incipient symbolic behavior. Early humans there also left behind at least 300 lumps of ochre and other pigments in a rainbow of colors: yellow, red, pink, brown, purple, and blue-black, some of which were gathered far from the site. Excavator Lawrence Barham of the University of Liverpool in the United Kingdom thinks they used the ochre to paint their bodies, though theres little hard evidence for this. Most archaeologists agree that body painting, as well as the wearing of personal ornaments such as bead necklaces, was a key way that early humans symbolically communicated social identity such as membership in a particular group, much as people today declare social allegiances and individual personalities by their clothing and jewelry. Yet while the Twin Rivers evidence is suggestive, its hard to be sure how the ochre was actually used. Theres little sign that it was ground into powder, as needed for decoration, says Ian Watts, an independent ochre expert in Athens. And even ground ochre could have had utilitarian uses, says archaeologist Lyn Wadley of the University of Witwatersrand in Johannesburg, South Africa. Modern-day experiments have shown that ground ochre can be used to tan animal hides, help stone tools adhere to bone or wooden handles, and Eye of the beholder. Archaeologists debate whether this modified stone was meant to represent a woman. even protect skin against mosquito bites. We simply dont know how ancient people used ochre 300,000 years ago, Wadley consider diagnostic elements of symbolic says. And since at that date the ochre users behavior came together. And in work now were not modern humans but our archaic in press, the Blombos team reports finding ancestors, some experts are leery of assign- engraved ochre in levels dating back to ing them symbolic savvy. 100,000 years ago ( Science , Yet many archaeologists are 30 January, p. 569). willing to grant that our species, There are other hints that the H. sapiens , was creating and sciencemag.org modern humans who ventured Hear author using certain kinds of symbols out of Africa around this time Michael Balter by 75,000 years ago and per- discuss the roots of art at might also have engaged in symhaps much earlier. At sites such www.sciencemag.org/ bolic behavior. At the Skhul rock as Blombos Cave on South darwin. shelter in Israel, humans left Africas southern Cape, people 100,000-year-old shell beads left sophisticated tools, including elabo- considered by some to be personal ornarately crafted bone points, as well as perfo- ments (Science, 23 June 2006, p. 1731). At rated beads made from snail shells and the 92,000-year-old Qafzeh Cave site pieces of red ochre engraved with what nearby, modern humans apparently strongly appear to be abstract designs. At this single preferred the color red: Excavators have site, a number of what many archaeologists studied 71 pieces of bright red ochre associated with human burials. Some researchers argue that this represents an early case of color symbolism, citing the universal importance of red in historical cultures worldwide and the apparently great lengths to which early humans went to gather red ochre. There is very strong circumstantial evidence for the very great antiquity of the color red as a symbolic category, says anthropologist Sally McBrearty of the University of Connecticut, Storrs. These finds of colorful ochre, fancy tools, and beads have convinced many researchers that the building blocks of symbolism had emerged by at least 100,000 years ago and possibly much earlier. But why? What selective advantages did using symbols confer on our ancestors? To some scientists, the question is a no-brainer, especially when it is focused on the most sophisticated form of symbolic communication: language. The ability to communicate detailed, concrete information as well as abstract concepts allowed early humans to cooperate and plan for the future in ways unique to our species, thus enhancing their survival during rough times and boosting their reproductive success in good times. What aspects of human social organization and adaptation wouldnt benefit from the evolution of language? asked Terrence Deacon, a biological anthropologist at the University of California, Berkeley, in his influential book The Symbolic Species: The Coevolution of Language and the Brain . Deacon went on to list just some of the advantages: organizing hunts, sharing food, teaching toolmaking, sharing past experiences, and raising children. Indeed, many researchers have argued that symbolic communication is what held groups of early humans together as they explored new environments and endured climatic shifts. As for art and other nonlinguistic forms of symbolic behavior, they may also have been key to cementing these bonds, by expressing meanings that are difficult or impossible to put into words. In that way, artistic expression, including music, may have helped ensure the survival of the fittest. This may also explain why great art has such emotional force, because the most effective symbols are those that convey their messages the most powerfully something the artists at Chauvet Cave seem to have understood very well.
MICHAEL BALTER

CREDITS (TOP TO BOTTOM): FRENCH MINISTRY OF CULTURE AND COMMUNICATION/DRAC RHONE-ALPES/DEPARTMENT OF ARCHAEOLOGY; CHRIS HENSHILWOOD AND FRANCESCO DERRICO

Online

710

CREDIT: FRANCESCO DERRICO AND APRIL NOWELL

6 FEBRUARY 2009

VOL 323

SCIENCE

www.sciencemag.org

www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

711

ORIGINS

On the Origin of of

Photosynthesis
piece together how and when organisms first began to harness lights energy. Although most modern photosynthesizers make oxygen from water, the earliest solar-powered bacteria relied on different ingredients, perhaps hydrogen sulfide. Over time, the photosynthetic machinery became more sophisticated, eventually leading to the green, well-oxygenated world that surrounds us today. In the lab, some biochemists are recapitulating the chemical steps that led to this increased complexity. Other researchers are locked in debates over just when this transition happened, 2.4 billion years ago or much earlier. Looking so far into the past is difficult. The geological record for that time is skimpy and tricky to interpret. Eons of evolution have blurred the molecular vestiges of the early events that remain in living organisms. But its a terribly important problem, says biochemist Carl Bauer of Indiana University, Bloomington, one well worth the travails.

Try to picture the world without photosynthesis. Obviously, youd have to strip away the greenerynot just the redwoods and sunflowers, but also the humble algae and the light-capturing bacteria that nourish many of the worlds ecosystems. Gone, too, would be everything that depends on photosynthetic organisms, directly or indirectly, for sustenancefrom leaf-munching beetles to meat-eating lions. Even corals, which play host to algal partners, would lose their main food source. Photosynthesis makes Earth congenial for life in other ways, too. Early photosynthesizers pumped up atmospheric oxygen concentrations, making way for complex multicellular life, including us. And water-dwellers were able to colonize the land only because the oxygen helped create the ozone layer that shields against the suns ultraviolet radiation. Oxygen-producing, or oxygenic, photosynthesis was the last of the great inventions of microbial metabolism, and it changed the planetary environment forever, says geobiologist Paul Falkowski of Rutgers University in New Brunswick, New Jersey. Given its importance in making and keeping Earth lush, photosynthesis ranks high on the top-10 list of evolutionary milestones. By delving into ancient rocks and poring over DNA sequences, researchers are now trying to

To catch a photon Over more than 200 years, researchers have ironed out most of the molecular details of how organisms turn carbon dioxide and water into food. Chlorophyll pigment and about 100 other proteins team up to put light to work. Plants, some protists, and cyanobacteria embed their chlorophyll in two large protein clusters, photosystem I and photosystem II. And they need both systems to use water as an electron source. Light jump-starts an electrical circuit in which The electron thief electrons flow from the photosystems Either way, it took some fancy fiddling to through protein chains that convert the primitive reaction make the energy-rich molecules THE YEAR OF centers to oxygen-generating Given its ATP and NADPH. These molephotosystems. Oxygenic photocules then power the synthesis synthesis was a huge upgrade, importance in of the sugars that organisms leading to a land of plenty, says making and depend on to grow and multiply. biochemist John Allen of Queen keeping earth Photosystem IIthe strongest Mary, University of London. naturally occurring oxidant Water is everywhere, so the lush, photoregains its lost electrons by organisms never ran out of elecsynthesis ranks swiping them from water, genertrons. They were unstoppable. ating oxygen as a waste product. But water clings to its elechigh on the However, some bacteria trons. With its oxidizing power, top- 10 list of dont rely on water as an elec- This essay is the third photosystem II can wrench them monthly series. More tron source, using hydrogen sul- in aevolutionary away, but the reaction centers in on evolution online at fide or other alternatives. These blogs.sciencemag.org/ nonoxygenic photosynthesizers milestones. nonconformists, which today origins. cannot. Biochemists James

live in habitats such as scalding hot springs, dont generate oxygen. Their photosynthetic proteins huddle in relatively simple reaction centers that may have been the predecessors of the two photosystems. Envisioning the steps that led to this complex biochemistry is mind-boggling. Similarities between proteins in photosynthetic and nonphotosynthetic bacteria suggest that early microbes co-opted some photosynthesis genes from other metabolic pathways. But protophotosynthesizers might also have helped each other piece these pathways together by swapping genes. Biochemist Robert Blankenship of Washington University in St. Louis, Missouri, and colleagues say theyve uncovered traces of these lateral gene transfers by comparing complete bacterial genomes. For example, their 2002 study of more than 60 photosynthetic and nonphotosynthetic bacteria ( Science , 22 November 2002, p. 1616) suggested that bugs had passed around several photosynthesis genes, including some involved in synthesizing the bacterial version of chlorophyll. Gene-sharing might also explain the puzzling distribution of the photosystems, Blankenship says. A cell needs both photosystems to carry out oxygenic photosynthesis. Yet modern nonoxygenic bacteria have the presumptive predecessor either of photosystem I or of photosystem II, never both. To explain how the two protein complexes wound up together, Blankenship favors a large-scale lateral [gene] transfer or even a fusion of organisms carrying each photosystem. However, other researchers remain skeptical, arguing that one photosystem evolved from the other, possibly through the duplication of genes, creating an ancient cell with both. No one knows for sure.

Allen (no relation to John Allen) and JoAnn Williams of Arizona State University, Tempe, and colleagues are working out how a bacterial reaction center could have evolved photosystem IIs appetite for electrons. Taking a hands-on approach, they have been tinkering with the reaction center of the purple bacterium Rhodobacter sphaeroides to determine if they can make it more like photosystem II. First they targeted bacteriochlorophyll, the bacterial version of chlorophyll thats at the core of the reaction center, and altered the number of hydrogen bonds. Adding hydrogen bonds hiked the molecules greed for electrons, they found. The water-cleaving portion of photosystem II sports four manganese atoms that become oxidized, or lose electrons. So the team equipped the bacterial reaction center with one atom of the metal. In this modif ied version, the added manganese also underwent oxidation, the researchers reported in 2005. James Allen says that their creations arent powerful enough to split water. But eventually, they hope to engineer a reaction center that can oxidize less possessive molecules, such as hydrogen peroxide, that would have been present on the early Earth. Even if the researchers never replicate photosystem II, if we def ine the intermediate stages, weve accomplished a lot, he says.

Catching rays. Long before plants got in on the act, photosynthetic cyanobacteria living in pools like this one in Yellowstone National Park were changing the composition of the atmosphere.

DARWIN

says astrobiologist Roger Buick of the University of Washington, Seattle. These hints could push the origin back 600 million years or more. One line of evidence is oil biomarkers that researchers think are the remains of cyanobacteria. Theyve turned up in rocks that are up to 2.7 billion years old. And in western Australia, thick shale deposits that are 3.2 billion years old hold rich bacterial remains but no traces of sulfur or other possible electron sources, suggesting that the microbes were using water to make energy. Geologist Euan Nisbet of Royal Holloway, University of London, and colleagues found additional support for an early origin when they went searching for traces of RuBisCO, a key photosynthetic enzyme. RuBisCO feeds carbon dioxide into the reactions that yield sugars. The enzyme version found in oxygenic photosynthesizers Oxygenic photosynthesis was the plays favorites: It prefers carbon dioxide that contains the carbonlast of the great inventions of micro12 isotope over the bulkier carbon-13. In 2007, Nisbet and bial metabolism, and it changed the his colleagues found disproportionately low carbon-13 values planetary environment forever. indicative of RuBisCO activity Paul Falkowski, Rutgers University when they analyzed organic matter in rocks from three sites Something in the air about 2.4 billion years ago, geologists see in Zimbabwe and Canada that are between How the photosystems got their start is cru- the first unmistakable signs of significant, 2.7 billion and 2.9 billion years old. Nisbet cial for understanding the origin of photo- sustained levels of atmospheric oxygen. concludes that oxygen-making photosynthesis. But the question thats drawn the These signs include red beds, or layers synthesis began at least 2.9 billion years ago. most attentionand provoked the most tinged by oxidized iron, i.e., rust. Further The early-origin case isnt ironclad. For wranglingis when photosynthesis began. support that the GOE marks an atmospheric example, a 2008 paper that has some Most researchers accept that nonoxygenic revolution comes from a technique that researchers fuming claims that the oil biophotosynthesis arose first, probably shortly detects skewed abundances of sulfur iso- markers are contaminants that seeped in after life originated more than 3.8 billion topes that occur if the air lacks oxygen. from younger rocks. Advocates also have to years ago. Life needs an energy source, and These imbalances persisted until the GOE, explain why it took hundreds of millions of the sun is the only ubiquitous and reliable when they vanished. years for oxygen to build up in the air. energy source, says Blankenship. Hard-liners construe these data to mean Although the last word on the origins of The sharpest disputes revolve around that oxygenic photosynthesis could not have oxygen-making photosynthesis isnt in, when organisms shifted to oxygenic photo- emerged until shortly before the GOE. But researchers say they are making progress. One synthesis. At issue is how to interpret a other scientists disagree. We are finding thing is for certain, however: Without this watershed in the fossil record known as the more and more hints that oxygenic photo- innovation, Earth would look a lot like Mars. MITCH LESLIE great oxidation event (GOE). In rocks from synthesis goes deeper into the fossil record,
www.sciencemag.org SCIENCE VOL 323 6 MARCH 2009

8 1286

CREDIT: NEALE CLARK/GETTY IMAGES

CREDIT: K. SUTLIFF/SCIENCE

6 MARCH 2009

VOL 323

SCIENCE

www.sciencemag.org

9 1287

NEWSFOCUS

ORIGINS
embryo that serves as its food supply. Darwin was perplexed by the diversity of flowering plants; they were too numerous and too varied, and there were too few fossils to sort out which were more primitive. Throughout much of the 20th century, magnolia relatives with relatively large flowers were leading candidates for the most primihow flowers got startedand from which tive living flowers, although a few ancestor. Today, researchers have analytical researchers looked to small herbs instead. tools, fossils, genomic data, and insights that In the late 1990s, molecular systematics Darwin could never have imagined, all of came to the rescue, with several reports prewhich make these mysteries less abom- senting a fairly consistent picture of the inable. Over the past 40 years, techniques lower branches of the angiosperm tree. An for assessing the relationships between obscure shrub found only in New Caledonia organisms have greatly improved, and gene emerged as a crucial window to the past. sequences, as well as morphology, now help Amborella trichopoda, with its 6-millimeter researchers sort out which angiosperms greenish-yellow flowers, lives deep in the arose early and which arose late. New fossil cloud forests there. In multiple gene-based finds and new ways to study themwith assessments, including an analysis in 2007 synchrotron radiation, for examplepro- of 81 genes from chloroplast genomes vide a clearer view of the detailed anatomy belonging to 64 species, Amborella sits of ancient plants. And researchers from var- at the base of the angiosperm family tree, ious fields are figuring out genomic changes the sister group of all the rest of the that might explain the amazing success of angiosperms. this fast-evolving group. Given that placement, Amborellas tiny These approaches have given researchers flowers may hint at what early blossoms a much better sense of what early flowers were like. Its one of the most similar living were like and the relationships among them. flower[s] to the worlds first flower, says But one of Darwins mysteries remains: the James Doyle of the University of Californature and identity of the angiosperm ances- nia, Davis. The petals and sepals of its sintor itself. When flowering plants show up in gle-sex flowers are indistinguishable and the fossil record, they appear with a bang, vary in number; so too do the numbers of with no obvious series of intermediates, as seed-producing carpels on female flowers Darwin noted. Researchers still dont know and pollen-generating stamens on male which seed- and pollen-bearing flowers. The organs are spirally organs eventually evolved into THE YEAR OF arranged, and carpels, rather the comparable flower parts. than being closed by fused tisFor more on Were a bit mystif ied, says sue as in roses and almost all botanist Michael Donoghue of familiar flowers, are sealed by a flower origins, Yale University. It doesnt secretion. listen to a appear that we can locate a close Most genetic analyses showed relative of the flowering plants. that water lilies were the next podcast by branch up the angiosperm tree, author Elizabeth Seeking the first flower followed by a group represented One of two major living groups by star anise, which also has a Pennisi at of seed plants, angiosperms have primitive look about it, says covered seeds that develop This essay is the fourth Doyle, though each of these www. encased in a protective tissue in a monthly series. has deviations from the ancesFor more on evolutionary sciencemag. called a carpel (picture a bean topics tral type. online, see the at pod). Thats in contrast to the Origins blog org/ nonflowering gymnosperms, blogs.sciencemag.org/ Fossil records origins. For more on multimedia/ such as conifers, which bear flower Although some fossil pollen origins, listen by author naked seeds on scales. An to a podcast dates back 135 million years, no podcast. Elizabeth Pennisi at angiosperms carpel sits at the www.sciencemag.org/ credible earlier fossil evidence center of the flower, typically multimedia/podcast. exists. In Darwins day, and for surrounded by pollen-laden stamany decades afterward, palemens. In most flowers, the carpel and stamens obotanists primarily found leaves or pollen are surrounded by petals and an outer row of but almost no fossil flowers. They had the leaflike sepals. Seeds have a double coating wrong search image, says Else Marie Friis of as well as endosperm, tissue surrounding the the Swedish Museum of Natural History in Stockholm. When we started, the search profile was bigger, a magnolia [flower], she recalls. But 30 years ago, she and others discovered tiny ancient flowers by sieving through sand and clay sediments. With this technique, they have now collected hundreds of millimeter-size flowers, some preserved in three dimensions, from Portugal and other locations with Cretaceous deposits 70 million to 120 million years old. This fossil diversity shows that angiosperms were thriving, with several groups well-established, by 100 million years ago. In some, the flower parts are whorled like those of modern flowers; in others they are spiraled, considered by some researchers as the more primitive arrangement. Some flower fossils have prescribed numbers of petals, another modern feature, whereas in others the petal count varies. In 1998, Chinese geologist Ge Sun of Jilin University in Changchun, China, came across what seemed to be a much older flower. The fossil, called Archaefructus, was an aquatic plant that looked to be 144 million years old. By 2002, Sun and David Dilcher of the Florida Museum of Natural History (FLMNH) in Gainesville had described an entire plant, from roots to flowers, entombed on a slab of rock unearthed in Liaoning in northeastern China. In one sense, Archaefructus wasnt much to look at. Its a flowering plant before there were flowers, Dilcher notes. It lacked petals and sepals, but it did have an enclosed carpel. When Kevin Nixon and colleagues at Cornell University compared its traits with those same traits in 173 living plants, Archaefructus came out as a sister to living angiosperms and closer to the common ancestor than even Amborella. Archaefructus s distinction was shortlived, however. Within months, better dating of the sediments in which it was found yielded younger dates, putting this f irst flower squarely with other early fossil flower parts, about 125 million years old. Also, a 2009 phylogenetic analysis of 67 taxa by Doyle and Peter Endress of the University of Zurich, Switzerland, placed the fossil in with water lilies rather than at the base of the angiosperms, although this conclusion is contested.

On the Origin of of

Flowering Plants

CREDIT: REPRODUCED WITH THE KIND PERMISSION OF THEPERMISSION DIRECTOR AND THE DIRECTOR BOARD OFAND TRUSTEES, ROYAL BOTANIC GARDENS, KEW CREDITS (TOP TO BOTTOM): REPRODUCED WITH THE KIND OF THE THE BOARD OF TRUSTEES, ROYAL BOTANIC GARDENS, KEW; GEORGE RICHMOND/BRIDGEMAN ART LIBRARY, LONDON (SUPERSTOCK)

Out of the past. Tiny Amborella sits at the bottom of the angiosperm family tree.

IN 1879, CHARLES DARWIN PENNED A LETTER

to British botanist Joseph Dalton Hooker, lamenting an abominable mystery that threw a wrench into his theory of evolution: How did flowering plants diversify and spread so rapidly across the globe? From rice paddies to orange groves, alpine meadows to formal gardens, prairies to oakhickory forests, the 300,000 species of angiosperms alive today shape most terrestrial landscapes and much of human life and culture. Their blooms color and scent our world; their fruits, roots, and seeds feed us; and their biomass provides clothing, building materials, and fuel. And yet this takeover, which took place about 100 million years ago, apparently happened in a blink of geological time, just a few tens of millions of years. The father of evolution couldnt quite fathom it. Darwin had an abhorrence that evolution could be both rapid and potentially even saltational, writes William Friedman in the January American Journal of Botany , which is devoted to this abominable mystery. Throughout his life, Darwin pestered botanists for their thoughts on the matter, but they couldnt give him much help. Now, 130 years later, evolutionary biologists are still pestering botanists for clues about what has made this plant group so successful, as well as when, where, and

DARWIN

28

10

CREDITS (TOP TO BOTTOM): COURTESY OF STEPHEN MCCABE, UC SANTA CRUZ; PHOTO BY JENNIFER SVITKO, COURTESY OF WILLIAM L. CREPET, CORNELL UNIVERSITY

We are realizing that this huge diversity is probably the result of one innovation piled on top of another innovation.

from one of the nonflowering seed plants or gymnosperms, whose heyday was 200 million years ago. Modern gymnosperms include conifers, ginkgoes, and the cycads, with their stout trunks and large fronds. Before angiosperms came along, these plants were much more diverse and included cycadlike species, such as the extinct Bennettitales, and many woody plants called Gnetales, of which Peter Crane, a few representatives, including the University of Chicago joint firs, survive today (see family tree, p. 31). Also common in the Jurassic were These fossils often spark debate because seed ferns, a group now long gone; their specimens tend to be imperfectly preserved most famous member is Caytonia, which and leave room for interpretation. To help seems to have precarpel-like structures. remedy that, Friis and her colleagues have These g roups perceived relevance to begun to examine flowers using synchro- flower evolution and their relationships to tron radiation to generate a 3D image of angiosperms have ping-ponged between their inner structures, allowing the fossil to camps, depending on how the evolutionary remain intact while Friis peers inside it trees were constructed. from many angles ( Science , 7 December In the mid-1980s, Peter Crane, now at 2007, p. 1546). We can get fantastic reso- the University of Chicago in Illinois, prolution, says Friis. Its really exciting. But posed a solution, the anthophyte hypothesis. so far, the flowers Friis finds are Using several lines of evidence and noting too diverse to trace back to a that both Bennettitales and particular ancestor. From Gnetales organize their male these fossils, we cannot and female organs together say what is the basic in what could be conform, she says. strued as a preflower, he considered them, Before flowers along with angiosperms, Although they have yet as comprising a single to find the oldest fossil angiosperm entity called Larger than life. Although merely flowers, researchers anthophytes. For the next 2.2 millimeters in diameter, this 3D assume that the ancesdecade, most family trees fossil flower shows that grasses date tral angiosperm evolved based on morphology supback to 94 million years ago.
SCIENCE VOL 324 3 APRIL 2009

3 APRIL 2009

VOL 324

SCIENCE

www.sciencemag.org

www.sciencemag.org

11

29

ORIGINS

ORIGINS
such a causal relationship is not settled. Later, animals that ate fruit and SEED PLANT Paleozoic seed ferns dispersed seeds likely helped evolvPHYLOGENY ing species expand quickly into new Asterids territory. Some think the answer lies Eudicots in genes: duplications that gave the angiosperm genome opportunities to Rosids try out new floral shapes, new chemical attractants, and so forth. This Monocots flexibility enabled angiosperms to exploit new niches and set them up for long-term evolutionary success. Magnolias My own view is that in the past, we have looked for one feature, says Crane. Now, we are realizing that Water lilies this huge diversity is probably the result of one innovation piled on top ? of another innovation. Archaefructus The latest insights into diversification come from gene studies. From Amborella 2001 to 2006, Pamela Soltis of the FLMNH and Claude dePamphilis ? of Pennsylvania State University, Caytonia University Park, participated in the Floral Genome Project, which ? Bennettitales searched for genes in 15 angiosperms. Now as a follow-up, the Ancestral Angiosperm Genome Project looks at gene activity in five early angioCycads sperms and a cycad, a gymnosperm. DePamphilis and his colleagues Ginkgoes matched all the genes in each species against one another to deter? Gnetales mine the number of duplicates. They then looked at the number of differences in the sequences of each gene Conifers pair to get a sense of how long ago Extinct taxa the duplication occurred. In most early angiosperms, including water lilies and magnolias, they saw many Shifting branches. As this simplified family tree shows, gene studies have helped clarify the relationships of many simultaneous duplicationsbut not living angiosperms, but fitting in extinct species is still a challenge, and some nodes are hotly debated. in Amborella , they reported in the January 2009 American Journal of Botany, The Floral Genome Project also looked are not as well-defined as they are in Araconfirming earlier reports. The data suggest to see whether the genetic programs guiding bidopsis. This sloppiness may have made that a key genome duplication happened flower development were consistent development flexible enough to undergo after the lineage leading to Amborella split throughout the angiosperms. We found that many small changes in expression patterns off but before water lilies evolved. Were there are fundamental aspects that are con- and functions that helped yield the great beginning to get the idea that polyploidiza- served in the earliest lineages, says Soltis. diversity in floral forms. tion may have been a driving force in creat- But there are differences in how the genes In his letter to Hooker, Darwin wrote ing many new genes that drive floral devel- are deployed. that he would like to see this whole probopment, dePamphilis says. Take the avocado, a species on the lower lem solved. A decade ago, Crepet thought Others have noted that a duplication branches of the angiosperm tree. In most Darwin would have gotten his wish by now. occurred in the evolution of grasses, and the angiosperms, the flower parts are arranged in That hasnt happened, but Crepet is optiFloral Genome Project confirms that yet concentric circles, or whorls, around the mistic that he and his colleagues are on the another duplication paved the way for eudi- carpels, with stamens innermost, then petals, right track, as analyses of various kinds of cots, the group that includes apples, roses, and finally sepals. Each tissue has its own data become more sophisticated. We are beans, tomatoes, and sunflowers. There are distinct pattern of gene expression, but not less likely to go around in circles in the next some real hot spots in angiosperm evolu- in the avocado. Genes that in Arabidopsis 10 years, he says. I believe a solution to tionary history, says dePamphilis, who is are active only in, say, the developing petals the problem is within reach. The mystery working to fully sequence the genome of spill over in avocado to the sepals. Thus in is solvable. ELIZABETH PENNISI Amborella with his colleagues. the more primitive plants, petals and sepals

Angiosperms

Flowers, food, fuel. at the the diversity diversityof ofangiosperms. angiosperms. Given that they represent nine in 10and land plants, its no surprise that they as fuel. Darwin Darwin marveled marveled at Given mainstays of both our welfare sense of beauty. Clockwise from left:serve aspens, that they represent nine in 10 land its no surprise that they serve as aspens, orchids, grasses, sunflowers, tulips, apples, mainstays of both our welfare and plants, sense of beauty. Clockwise from top left: orchids, grasses, sunowers, tulips,walnuts. apples, walnuts.

ported this idea. Crane and others carefully dissected and described fossils of these groups, looking for the precursors to carpels, the seeds double coat, and other distinctive angiosperm traits. But they have run into problems. We do not really know how to compare them because the structures are very differentlooking; figuring out whats homologous is quite a difficult thing, says Crane. He and his colleagues argue, for example, that the seeds in the Bennettitales have two coverings, which may be a link to angiosperms. But in the January American Journal of Botany, Gar Rothwell of Ohio University, Athens, and two colleagues disagree, saying that what Crane calls the outer layer is the only layer, and f ind fault with the hypothesis in general. To make matters worse for anthophyte proponents, genebased evolutionary trees break up this grouping, pulling the Gnetales off any angiosperm branch and placing them among or next to the other gymnosperms. The molecular work points in one direction; the paleobotanical

work points you in another direction, Crane says. And if the molecular work is correct, then the field doesnt know in which direction to turn, because in most analyses the genetic data dont place any living plant close to angiosperms. The angiosperms group together, the living gymnosperms group together, and theres nothing in between. The nonangiosperm ancestor just isnt there, says paleobotanist William Crepet of Cornell. Im starting to worry that we will never know, that it transformed without intermediates. Seeds of success The angiosperms ancestor may be missing, but what is very clearand was quite annoying to Darwinis that the angiosperm prototype so readily proved a winner. Seed ferns and other gymnosperms arose about 370 million years ago and dominated the planet for 250 million years. Then in a few tens of millions of years, angiosperms edged them
Inside and out. Synchrotron radiation helped produce a 3D rendering (gold) of this fossil male flower (right) and insights into its internal structure.

out. Today, almost nine in 10 land plants are angiosperms. The exact timing of the angiosperms explosion and expansion is under debate, as is the cause. At least one estimate based on the rate at which gene sequences changethat is, the ticking of the molecular clock pushes angiosperm evolution back to 215 million years ago. There appears to be a gap in the fossil record, says Donoghue, who also notes that molecular dating methods are still in their infancy and, thus, could be misleading. He and others think that flowering plants lingered in obscurity for tens of millions of years before radiating toward their current diversity. Whatever the timing, there was something special about the angiosperm radiation. During the 1980s and again in 1997, Cornells Karl Niklas compiled a database showing the first and last occurrences of fossil plants. When he and Crepet used that and more recent information to look at species appearances and disappearances, they found that new angiosperms appeared in bursts through time, whereas other plants, such as gymnosperms, radiated rapidly only at first. Moreover, angiosperms proved less likely to disappear, somehow resisting extinction, says Crepet. Once the angiosperms arrived, how did they diversify and spread so quickly? Darwin suspected that coevolution with insect pollinators helped drive diversification, though

Living gymnosperms

12 30

CREDITS CREDITS (TOP (TOP TO TO BOTTOM): BOTTOM): PHOTOS.COM; PHOTOS.COM; ELSE ELSE MARIE MARIE FRIIS FRIIS

CREDITS: (ARCHAEFRUCTUS IMAGE) DAVID DILCHER; (ALL OTHER IMAGES) PHOTOS.COM

3 APRIL 2009

VOL 324

SCIENCE

www.sciencemag.org

www.sciencemag.org

SCIENCE

VOL 324

3 APRIL 2009

13 31

TIVES PERSPECTIVES
HISTORY HISTORY OF OF SCIENCE SCIENCE HISTORY OF SCIENCE

von Humboldt and vonAlexander Humboldt and the General Physics of the Earth l Physics of the Earth
Stephen Stephen T. T. Jackson Jackson Stephen T. Jackson

In early In the the early 19th 19th century, century, Alexander Alexander von von In the early 19th century, Alexander von In the early 19th century, Alexander von Humboldt laid the foundations Humboldt laid the foundations for todays todays Humboldt laid the foundations for for todays Humboldt laid the foundations for todays Earth Earth system system sciences. sciences. Earth system sciences.

Earth system sciences.

RIGINAL COPY IN THE NATURAL HISTORY MUSEUM, PARIS

s s scientists scientists are are celebrating celebrating the the 200th 200th s scientists are celebrating the 200th anniversary of Charles Darwins anniversary of of Charles Charles Darwins Darwins birth birth anniversary birth rating the 200thand and the the 150th 150th anniversary anniversary of of the the pubpuband the 150th anniversary of the publication s Darwins birthof lication of his his On On the the Origin Origin of of Species Species, lication of his On the Origin of Species ,, Darwins rsary of the pub- ideas Darwins ideas continue continue to to shape shape and and enrich enrich Darwins ideas continue to shape and enrich the the sciences sciences (1 1). ). 6 6 May May 2009 2009 marks marks the the 150th 150th the sciences ( 1 ). 6 May 2009 marks the 150th gin of Species , ( anniversary of the death of another 19th-cenanniversary of the death of another 19th-cenanniversary of the death of another 19th-cenhape andtury enrich tury figureAlexander figureAlexander von von Humboldt Humboldt tury figureAlexander von Humboldt marks the 150th whose scientific whose scientific legacy legacy also also flourishes flourishes in in the the whose scientific legacy also flourishes in the 21st century. Humboldt helped create nother 19th-cen21st century. century. Humboldt Humboldt helped helped create create the the 21st the intellectual n Humboldt intellectual world world Darwin Darwin inhabited, inhabited, and and his his intellectual world Darwin inhabited, and his writings inspired Darwin to embark writings inspired Darwin to embark on writings on flourishes in the inspired Darwin to embark on H.M.S. Beagle. More pertinent to our time, H.M.S. Beagle. More pertinent to our time, H.M.S. Beagle. More pertinent to our time, lped create the Humboldt Humboldt established established the the foundation foundation for for the the Humboldt established the foundation for the habited, Earth and his system sciences: the integrated system Earth system system sciences: sciences: the the integrated integrated system system Earth of on to embark on of knowledge knowledge on which which human human society society may may of knowledge on which human society may depend in depend in the the face face of of global global climate climate change. change. ent to our time, depend in the face of global climate change. Darwin, Like Darwin, Humboldt Humboldt undertook undertook a a Like Darwin, Humboldt undertook a undationmajor forLike the voyage that would shape major voyage voyage that that would would shape shape his his ideas ideas major his ideas tegrated and system and thinking. thinking. Humboldt Humboldt spent spent 5 5 years years (1799 (1799 and thinking. Humboldt spent 5 years (1799 man society maywith to to 1804) 1804) with botanist botanistAim Aim Bonpland Bonpland explorexplorto 1804) with botanist Aim Bonpland exploring limate change. ing Venezuela, Venezuela, the the northern northern Andes, Andes, and and cencening Venezuela, the northern Andes, and central Mexico, with visits to Tenerife, Cuba, tral Mexico, Mexico, with visits visits to to Tenerife, Tenerife, Cuba, Cuba, and and dt undertook a with tral and the United the United States. States. They They collected collected botanical, botanical, the United States. They collected botanical, hape his ideas zoological, zoological, geological, geological, and and ethnological ethnological specspeczoological, geological, and ethnological specnt 5 years (1799 imens, made extensive atmospheric imens, made made extensive extensive atmospheric atmospheric and and geogeo- Intellectual imens, and geoIntellectual riches. riches. The The central central portion portion of of Humboldts Humboldts Physical Physical Tableau Tableau of of the the Andes Andes and and Neighboring Neighboring Intellectual riches. The central portion of Humboldts Physical Tableau of the Andes and Neighboring Bonplandphysical explor- measurements, physical measurements, and and recorded recorded the the Countries , published as part of ( 2 , 3 ), shows Chimborazo in profile, with vegetation zones, plant species, physical measurements, and recorded the Countries , published as part of ( 2 , 3 ), shows Chimborazo in profile, with vegetation zones, plant species, and and Countries , published as part of ( 2 , 3 ), shows Chimborazo in profile, with vegetation zones, plant species, and geographic Andes, and cen- location geographic location of of their their thousands thousands of of snowline geographic location of their thousands of snowline depicted depicted at at appropriate appropriate elevations. elevations. In In the the original, original, the the profile profile is is flanked flanked on on both both sides sides by by tables tables snowline depicted at appropriate elevations. In the original, the profile is flanked on both sides by tables specimens specimens and tens tens of of thousands thousands of of measuremeasure- describing specimens and tens of thousands of measuredescribing elevational elevational patterns patterns in in temperature, temperature, humidity, humidity, light light refraction refraction and and intensity, intensity, agriculture, agriculture, fauna, fauna, and and erife, Cuba, and and describing elevational patterns in temperature, humidity, light refraction and intensity, agriculture, fauna, and ments. Humboldt spent the next 22 years and other physical, chemical, and biological features. ments. Humboldt spent the next 22 years and other physical, chemical, and biological features. ments. Humboldt spent the next 22 years and other physical, chemical, and biological features. ected botanical, most most of of his his inherited inherited fortune fortune in in Paris, Paris, prepremost of his inherited fortune in Paris, prehnological specparing and publishing 45 volumes of paring and publishing 45 volumes of a a nevernever- distribution distribution of of incident incident radiation, radiation, the the transport transport serving serving as as both both an an index index of of climate climate and and a a proxproxparing and publishing 45 volumes of a neverdistribution of incident radiation, the transport serving as both an index of climate and a proxspheric and geo-report finished on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, finished report on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, finished report on his travels. of heat and materials in winds and ocean curimal control of microclimate, animal habitat, Intellectual riches. The central Humboldts Physical Tableau of the Andes and Neighboring Countries , published as part Intellectual riches. The portion centralofportion of Humboldts Physical Tableau of the Andes and Neighboring Of these the first was slim work rents, the of on and cultural practices ( ). d recorded the of (2, 3), shows Chimborazo in prole, with vegetation zones, plant species, and snowline depicted at appropriate elevations. Of these volumes, volumes, the first was a a as slim work rents, the influence influence of temperature temperature on plant plant and cultural practices (6 6 8 8 ). Of these volumes, the first was a slim work rents, the influence of temperature on plant and cultural practices ( 6 8 ). Countries , published part of ( 2 , 3 ), shows Chimborazo in profile, with vegetation zones, plant species, and entitled Essay on the Geography of Plants form, and the effect of latitude and continenHumboldts dream of systematic observaIn the original, the prole is anked on both sides by tables describing elevational patterns in temperature, humidity, light entitled Essay on the Geography of Plants form, and the effect of latitude and continenHumboldts dream of systematic observaentitled on the Geography of Plants form, and the effect of latitude and continenHumboldts dream of systematic observair thousands ofEssay snowline depicted at appropriate elevations. In the original, the profilefeatures. is flanked on bothacross sides the by tables refraction and intensity, agriculture, fauna, andon other physical, chemical, and biological ( 2 , 3 ). The modest title belies the intellectual tality mountain snowline. tional arrays globe took hold in ( 2 , 3 ). The modest title belies the intellectual tality on mountain snowline. tional arrays across the globe took hold in the the (2, 3). The modest title belies the intellectual tality on mountain snowline. tional arrays across the globe took hold in the nds of measuredescribing elevational patterns in temperature, humidity, light refraction and intensity, agriculture, fauna,the and richness He richness within. within. In In the the text text and and accompanying accompanying He expanded expanded this this vision vision in in the the succeeding succeeding 19th 19thcentury. century.Throughout Throughoutthe thecentury, century,countless countless richness within. In the text and accompanying He expanded this vision in the succeeding 19th century. Throughout century, countless ext 22 years and other physical, chemical, and biological color the Humboldt lays out years, establishing color plate plate (see (see the figure), figure), Humboldt lays out years, features. establishing international international cooperative cooperative Humboldt-inspired Humboldt-inspiredexplorations explorationswere werelaunched, launched, color plate (see the figure), Humboldt lays out years, establishing international cooperative Humboldt-inspired explorations were launched, a vision of a comprehensive general physics networks of meteorological and geomagnetic each involving systematic measurement and e in Paris, prea vision visions of a comprehensive general physics networks of meteorological and geomagnetic each involving systematic measurement and a of a comprehensive general physics networks of meteorological and geomagnetic each involving systematic measurement and scientists are celebrating the a 200th and geophysical measurements, inhabited, and his writings inspired isotherms Darwin to atmospheric of the Earth aimed at nothing less than synmeasurement stations, inventing mapping of physical, biological, and often culumes of a neverdistribution of incident radiation, the transport serving as both an index of climate and a proxof the Earth aimed at nothing less than a synmeasurement stations, inventing isotherms mapping of physical, biological, and often culof the Earth aimed at than a birth syn- embark measurement stations, isotherms mapping of physical, biological, and often culDarwins on H.M.S. Beagle.inventing More pertinent to our and of their anniversary ofnothing Charlesless recorded the geographic location thesis atmospheric, oceanic, geological, and other graphical devices to portray spatial tural of and oceans ( thesis of of atmospheric, oceanic, geological, and other graphical devices to portray spatial tural features features of landscapes landscapes and oceans (8 8 10 10). ). thesis of atmospheric, oceanic, geological, and other graphical devices to portray spatial tural features of landscapes and oceans ( 8 10 ). of150th heat anniversary and materials winds and ocean curimal control of microclimate, animal habitat, the foundation for thousands and the of thein publitime, Humboldt established of specimens and tens of thousands ecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, typecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, typecological, and cultural phenomena across the patterns, and noting that plant form is often These surveys were relentlessly inductive, typof his On the Origin of Species, Darwins system of Humboldt spent the next 22 the Earth on system sciences: thecultural integratedpractices cation measurements. was a slim work rents, the influence of temperature plant and (6 8producing ). globe. Humboldts obsession with geographibetter predicted by local environment than by ically detailed descriptive reports globe. Humboldts obsession withthe geographibetter predicted by local environment thandeby years icallyand producing detailed descriptive reports globe. Humboldts obsession with geographibetter predicted local environment than by ically producing descriptive reports andthe enrich sciences Paris, continue to shape of knowledge on by which human society may most of detailed his inherited fortune in raphy ofideas Plants form, and effect of latitude and continenHumboldts dream of systematic observacally referenced measurements and collectaxonomic affinity (a paradox resolved by with little integration within or among the comcally referenced measurements and collectaxonomic affinity (a paradox resolved by with little integration within or among the comcally referenced measurements and collectaxonomic affinity (a paradox resolved by preparing with little integration within among of the of pend of global climate change. ( 1). 6 May 2009 marks the 150th anniversary and publishing 45 or volumes acomnevin the face tions was central to his vision. He recognized Darwin). Humboldts genius lay in his geoponent entities. However, for a few intellectus the intellectual tality on mountain snowline. tional arrays across the globe took hold in the tions was central to his vision. He recognized Darwin). Humboldts genius lay in his geoponent entities. However, for a few intellectutions was central to his vision. He recognized Darwin). Humboldts genius lay in his geoponent entities. However, for a few intellectuthe death of another 19th-century figureAlexLike Darwin, Humboldt undertook a major er-finished report on his travels. that arrays of observations could be graphical vision, and in intuition that nimble participantsincluding Charles that spatial spatial arrays ofexpanded observations could be in graphical vision, and19th in his his intuition that ally ally nimble participantsincluding Charles d accompanying He this vision the succeeding century. Throughout the century, countless that spatial arrays of observations could be graphical and in his intuition that ally nimble participantsincluding Charles von Humboldtwhose scientific legacy his ideas and thinking. Of these volumes, the first was a slim work ander voyage thatvision, would shape aggregated to reveal patterns that would in Earths land surface, oceans, atmosphere, and Darwin, T. H. Huxley, Matthew Maury, Asa aggregated to reveal patterns thatHumboldt would in Humboldt Earths land surface, oceans, atmosphere, and explorations Darwin, T. H. H. Huxley, Matthewof Maury, Asa aggregated to reveal patterns that would in Earths land surface, oceans, atmosphere, and Darwin, T. Huxley, Matthew Maury, the establishing 21st century. the Geography PlantsAsa (2, flourishes in spent 5 years (1799 to 1804) with entitled Essay on umboldt also lays out years, international cooperative Humboldt-inspired were launched, turn reveal underlying processessuch as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin turn reveal underlying processessuch as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin turn reveal underlying processessuch as the inhabitants form an integrated whole, with Gray, C. Hart Merriam, and Peter Kropotkin Darwin intellectual richhelped create the intellectual world botanist Aim Bonpland exploring Venezuela, 3 ). The modest title belies the general physics networks of meteorological and geomagnetic each components involving systematic measurement anddata and experilinkages among the various ( these explorations provided linkages among theand various components (4 4, these explorations provided data and and experiexperilinkages among the various components ( 4 ,, ness these explorations provided data northern Andes, central Mexico, with within. In the text and accompanying color the g less than a syn- measurement stations, inventing isotherms mapping of physical, biological, and often cul5 ). Humboldts general physics of the Earth ence that spurred the development of biogeog5 ). Humboldts general physics of the Earth ence that spurred the development of biogeogCuba, and the United States. plate (see the figure), Humboldt lays out a vivisits to Tenerife, 5 ). Humboldts general physics of the Earth ence that spurred the development of biogeogBotany Department and Program in Ecology, University of Botany Department Department and and Program Program in in Ecology, Ecology, University University of of Botany envisioned climate as a control of raphy, oceanography, other envinic, geological, and other graphical devices to portray spatial tural features of landscapes and oceans (8 10). and Wyoming, envisioned climate as a major major control of sion raphy, ecology, oceanography, and otherof envigeological, general physics the They collected botanical, zoological, of ecology, a comprehensive envisioned climate as a major control of raphy, ecology, oceanography, and other enviWyoming, Laramie, Laramie, WY WY 82071, 82071, USA. USA. E-mail: E-mail: jackson@ jackson@ Wyoming, Laramie, WY 82071, USA. E-mail: jackson@ uwyo.edu Earth-surface phenomena, with vegetation ronmental sciences ( uwyo.edu Earth-surface phenomena, with vegetation ronmental sciences (11 11). ). omena across the patterns, and noting that plant and ethnological specimens, made extensive Earth aimed at nothing less than a synthesis of form is often These surveys were relentlessly inductive, typuwyo.edu Earth-surface phenomena, with vegetation ronmental sciences ( 11 ).

atmospheric, oceanic, geological, ecological, and cultural phenomena across the globe. Humboldts obsession with geographically referenced measurements and collections was central to his vision. He recognized that spatial arrays of observations could be aggregated to reveal patterns that would in turn reveal underlying processessuch as the distribution of incident radiation, the transport of heat and materials in winds and ocean currents, the influence of temperature on plant form, and the effect of latitude and continentality on mountain snowline. He expanded this vision in the succeeding years, establishing international cooperative networks of meteorological and geomagnetic measurement stations, inventing isotherms and other graphical devices to portray spatial patterns, and noting that plant form is often better predicted by local environment than by taxonomic affinity (a paradox resolved by Darwin). Humboldts genius lay in his geographical vision, and in his intuition that Earths land surface, oceans, atmosphere, and inhabitants form an integrated whole, with linkages among the various components (4, 5). Humboldts general physics of the Earth envisioned climate as a major control of Earth-surface phenomena, with vegetation serving as both an index of climate and a proximal control of microclimate, animal habitat, and cultural practices (68). Humboldts dream of systematic observational arrays across the globe took hold in the 19th century. Throughout the century, countless Humboldt-inspired explorations were launched, each involving systematic measurement and mapping of physical, biological, and often cultural features of landscapes and oceans (810). These surveys were relentlessly inductive,

typically producing detailed descriptive reports with little integration within or among the component entities. However, for a few intellectually nimble participantsincluding Charles Darwin, T. H. Huxley, Matthew Maury, Asa Gray, C. Hart Merriam, and Peter Kropotkin these explorations provided data and experience that spurred the development of biogeography, ecology, oceanography, and other environmental sciences (11). Unfortunately, the conceptual unification among the sciences of the Earth that Humboldt sought never developed in the century following his death. Disciplinary specialization played a large role in eclipsing Humboldts integration, as did 20th-century trends toward reductionism, experimentalism, and fine-scale processes in many disciplines. A new incarnation of Humboldts general physics of the Earth began to emerge with the plate tectonics revolution in the 1960s. Drawing on Humboldtian spatial arrays of observations, this theory provided a unified explanatory framework for disparate geophysical, geological, paleontological, and biogeographic phenomena. Today, a second, even broader manifestation of Humboldts vision aspires to understand the interactions and feedbacks among the components of the Earth system, encompassing the lithosphere, atmosphere, hydrosphere, cryosphere, and biosphere as well as human societies and economies. This effort is often referred to as Earth system science, but it could just as well be designated general physics of the Earth, using the early-19th century definition of physics as the study of the material world and its phenomena (which we now call science).

Global environmental change may be the greatest challenge faced by human societies since the advent of agriculture. Humboldt advocated for science that spoke to human needs and concerns (5). It is fitting that on the 150th anniversary of his death, we recognize his role in fostering the sciences that speak to the most profound human concernssustainability of human societies and the ecosystems on which they depend.
1. P. J. Bowler, Science 323, 223 (2009).[Abstract/ FreeFullText] 2. A. de Humboldt, Essai sur la gographie des plantes (Levrault, Schell & Co., Paris, 1807). 3. An English translation of (2) is currently in press at the University of Chicago Press, Chicago. 4. A. de Humboldt, Personal Narrative of Travels to the Equinoctial Regions of the New Continent, During the Years 17991804, by Alexander de Humboldt and Aime Bonpland; with Maps, Plans, &c. (Longman, Hurst, Rees, Orme, & Brown, London, 1814 to 1829), vols. 1 to 7. 5. A. von Humboldt, Cosmos: A Sketch of the Physical Description of the Universe (Johns Hopkins Univ. Press, Baltimore, 1997), vol. 1. 6. M. Nicolson, Hist. Sci. 25, 167 (1987). [ISI] 7. M. Nicolson, in Romanticism and the Sciences, A. Cunningham, N. Jardine, Eds. (Cambridge Univ. Press, Cambridge, 1990), p. 169. 8. M. Dettelbach, in Cultures of Natural History, N. Jardine, et al., Eds. (Cambridge Univ. Press, Cambridge, 1996), p. 287. 9. S. F. Cannon, Science in Culture: The Early Victorian Period (Dawson, New York, 1978). 10. W. H. Goetzmann, New Lands, New Men: America and the Second Great Age of Discovery (Viking, New York, 1986). 11. P. J. Bowler, The Norton History of the Environmental Sciences (Norton, New York, 1993).

References and Notes

CREDIT: CREDIT: REPRODUCED REPRODUCED FROM FROM AN AN ORIGINAL ORIGINAL COPY COPY IN IN THE THE NATURAL NATURAL HISTORY HISTORY MUSEUM, MUSEUM, PARIS PARIS CREDIT: REPRODUCED FROM AN ORIGINAL COPY IN THE NATURAL HISTORY MUSEUM, PARIS

In the early 19th century, Alexander von Humboldt laid the foundations for todays Earth system sciences.

with geographi14 596 596 ents596 and collec-

better predicted by local environment than by ically producing detailed descriptive reports 1 2009 VOL SCIENCE www.sciencemag.org 1 MAY MAY 2009 resolved VOL 324 324 by SCIENCE www.sciencemag.org 1 MAY 2009 VOL 324 SCIENCE www.sciencemag.org taxonomic affinity (a paradox with little integration within or among the com-

15

BOOKS ET AL.
HISTORY OF SCIENCE

BOOKS ETAL.
mer students who challenged his ideas as they gained intellectual independence, and debated the pro-evolution (but anti-Haeckel) Jesuit priest and entomologist Erich Wasby Robert J. Richards mannthe list could go on University of Chicago and on. These were not isoPress, Chicago, 2008. duced important books. Robert lated episodes but rather 579 pp. $39, 27. J. Richards, the director of the moments in a lifelong camISBN 9780226712147. University of Chicagos Fishpaign to advance his philosobein Center for the History of phy, which was accompanied H. G. Bronn, Science and Medicine and a by a bitter hostility to orgaErnst Haeckel, and much-published author on Darwin nized religion. the Origins of and German Romantic biology, Richards does not negGerman Darwinism has written a biography of Haeclect Haeckels science proA Study in Translation kel. Sander Gliboff, a professor per, treating us to fascinatand Transformation in Indiana Universitys Departing and original discussions ment of History and Philosophy of his pathbreaking systemby Sander Gliboff of Science, places Haeckel at the atic and phylogenetic work MIT Press, Cambridge, MA, end of a study that examines the on radiolaria and other marine 2008. 271 pp. $35, 22.95. larger process through which organisms, the importance ISBN 9780262072939. Darwins words were translated, of linguistic analysis to his and his ideas modified, in the phylogenetic trees of the context of German biology. Both illuminate races of humans, and his remarkable experithe twists and turns that evolutionary thought mental work with siphonophores. These contook in Germany, but they do so in dramati- stitute important contributions to our undercally different ways. standing of the technical development of Richardss book, though over twice as long evolutionary biology. as Gliboffs, is the more entertaining read of The big picture here, however, is an arguthe two. In his characteristically rich and ment about the power of personalityat least rolling prose, Richards weaves a compelling one personalityto shape the course of scistory of a life marked by tragedy and of an ence. In Richardss presentation, German evointense, larger-than-life figure whose passions lutionism was profoundly shaped by both drove his scientific research and philosophy. In Haeckels charisma and his combativeness. Richardss rendering, the scientific Haeckel Perhaps the late-19th-century opposition of cannot be understood separately from the evolutionary science to Christianity would not mans personality and private circumstances. have been so fiery, he suggests, had Haeckel His love of nature was surpassed only by his not continually fanned its flames. And love for his first wife, Anna Sethe, who died in although Richards absolves Haeckel of perabdominal agony on his 30th birthday. Over sonal responsibility for fascism and Nazism, the next year, he wrote his way through the in part by situating him firmly in his time and despair that enveloped him, producing his place, he does show how the scientists ardent foundational work, Generelle Morphologie temperament led him to the occasional intem(1). Although he remarried, the union was not perate statement that could be taken up by happy, and passionate love would elude him extreme thinkers. One cannot leave this book until his sixties, when he had a secret affair that without a deep appreciation for Haeckel as a ended tragically with the death of his lover. tragic figure and for the force of personality in Science remained his salvation and refuge. shaping the direction science may take. His professional life was also filled with Gliboffs account is of a completely differdrama, much of which centered on his philos- ent order. His is not a story of personalities or ophy of evolutionary monisma science- private lives (although he mentions salient centered faith that became one of the most details), but of German academics seeking to successful alternatives to the Judeo-Christian live up to the highest (if changing) ideals of religion among those searching for a secular Wissenschaft and of the ways in which spirituality. Haeckel could not turn down a Darwins theory was translated into this envifight: He battled the physician-statesman ronment. He thus situates Haeckel at the end Rudolf Virchow over the role of evolution in of a revised intellectual history of 19ththe schools (Haeckel argued that it should century German evolutionism. Central to his replace religious education), sparred with reli- account is the idea of translation, which he giously conservative scientists and with for- uses both synchronically, especially in treatThe Tragic Sense of Life Ernst Haeckel and the Struggle over Evolutionary Thought VOL 323 SCIENCE www.sciencemag.org

Making German Evolution: Translation and Tragedy

Lynn K. Nyhart

CREDIT: ERNST HAECKEL/FROM WANDERBILDER (W. KOEHLER, GERA-UNTERMHAUS, 1905)

n this year of Darwin anniversaries (the 200th year of his birth and the 150th anniversary of On the Origin of Species), The Tragic Sense of Life and H. G. Bronn, Ernst Haeckel, and the Origins of German Darwinism remind us that the history of evolutionary thought in the 19th century extended well beyond Darwin himself. Darwin did not launch his theory onto an unprepared public and scientific community, nor was the evolutionism that developed after 1859 a mere extension of his viewsit was not even one thing. How, then, should we think about the history of evolution in the 19th century? What sorts of accounts best help us understand the reception of Darwins theory, its relations to earlier ideas about nature, the directions that evolutionary investigation subsequently took, and the relations of all of these to the broader social, cultural, and religious concerns scientists shared with their contemporaries? These questions become especially pointed when one considers German Darwinism, and especially Germanys best-known follower of Darwin, Ernst Haeckel. Most often remembered by biologists as the author of the biogenetic law (ontogeny recapitulates phylogeny), Haeckel has also been accused of promoting European fascism via his monistic philosophy and of presenting a eugenic, biologically determinist vision of humanity that led to Hitlers final solution. Can one scientist be responsible for so much? Most historians would say no, arguing that it takes a community, rather than an individual, to make a movement; that single-cause explanations are insufficient to account for something as broad as fascism; and that an individual cannot be held responsible for the ways in which others (such as Hitler) took up his ideas and molded them to new agendas after his death. But that still leaves open the questions of how to write responsibly about what Haeckel actually believed and how we should situate him in the history of evolutionary thought. The historians under consideration here have chosen two radically different strategies to understanding Haeckels place within German evolutionism, and both have proThe reviewer is at the Department of the History of Science, University of Wisconsin, Madison, WI 537061393, USA. E-mail: lknyhart@wisc.edu

ing the translation of Origin of Species into translations involved an attempt to recast existGerman, and (more intriguingly) diachroni- ing German terms in a newer, more up-to-date cally, as scientists reworked older words such mode that encompassed selection yet tamed as perfection and type to lend them new Darwins emphasis on unpredictability to meet meanings. Gliboffs own clear, crisp prose is the more rigorous requirements of a German key to the success of this analysis, as he deftly academic scientists understanding of a law leads his reader through dense philosophical of organic nature. Simultaneously, Bronn and terminological thickets with nary a thorn sought to translate Darwins ideas about selecscratch. This is some of the best close reading tion into a language without an exact equivaI have seen. It also represents a profound chal- lent for the term, and for an academic audience lenge to our standard picture of 19th-century lacking the gentlemanly traditions of breeding German biology. pigeons and dogs so central to Darwins expoThe old story, crudely put, is that Haeckels sition. The selection metaphor was further version of evolution was a Darwinism in name fraught with an anthropomorphism foreign to only, best understood as an update on early- Germans, who were not brought up on British 19th-century idealistic morphologists such natural-theological assumptions about a peras Carl F. Kielmeyer and J. F. Meckel that sonified God who had created a perfectly retained their teleology, their typological adapted nature. Bronns translation, though it emphasis on form, and their linear recapitula- altered key ideas to make Darwin comprehentionism. This story, emphasizing the long per- sible to a German academic audience, was not sistence of a German transcendental approach a conservative throwback. It represented the to nature, has been deeply entrenched in the dynamic engagement of a leading paleontolohistory of biology. gist who had also long been working on many Gliboff challenges this history right from of the questions Darwin claimed as his own the beginning. The ascription of simple linear a critical yet generous equal, who saw himself recapitulationism to the views of Romantic as moving science forward through the modiembryologists, he notes, owes much to a carica- fications he made to Darwins flawed theory. ture developed by Karl Ernst von Baer in a Bronns death in 1862 afforded him little polemical context, then adopted uncritically by chance to steer the conversation further. influential historians such as E. S. Russell and Stephen Jay Gould. Gliboffs fresh reading of the original sources interprets Kielmeyer and Meckel as far less rigidly typological in their orientation and much more attentive to natures variability than has been seen before. Both for these early-19thcentury naturalists and for their intellectual heirs, Gliboff argues, the critical issue was to understand natures manifold variety while seeking out underlying strict natural laws to account for it. This provides a new starting point for analyzing Darwins first translator, the prominent paleontologist H. G. Bronna figure little attended to in the standard story but the lynchpin of Gliboffs. Intriguingly and plausibly, Gliboff argues that Bronns use of terms like vervollkommnet (perfect) as translations for Darwins improved or favored were not about dragging Darwin backward into a German teleological view of nature (as has been claimed by those who have paid attention to Bronn at all). A painter, too. Haeckels oil landscape of highlands in Java, from Instead, Gliboff asserts, Bronns Wanderbilder (1905).
www.sciencemag.org SCIENCE VOL 323

And so, finally, we come to Haeckel. Gliboff s key insight here is that Haeckel originally read Bronns translation of Darwin, not Darwin in the original. Gliboff shows Haeckel as both echoing and responding to Bronns concerns, rather than either reflecting directly on Darwins original writing or reaching directly back to the Romantic embryologists. (Although Gliboff acknowledges the centrality of monism to Haeckels thought, he focuses on the working evolutionary theorist, not the popular ideologue.) Like Bronn himself, Haeckel made further amendments both terminological and intellectual, and Gliboff rereads Haeckels research program as one not dominated by a typological and linear-recapitulationist mindset but rather as continuing to wrestle with the need to account for variability and unpredictable change in terms of mechanistic laws of natureamong which Haeckel included, at the top of his list, natural selection. Haeckels Darwinism thus shows continuity with early19th-century concerns, mediated through Bronn. But those concerns were always more flexible than has been acknowledged, and their articulation changed over time. Of course Haeckels Darwinism was not Darwins own, but it was not an aberration or a distortion of some true theory, any more than any other post-Darwinian additions or adjustments were. It was science moving on. Gliboff s overall picture of scientific advance, in contrast to Richardss emphasis on charisma and passion, is one of scientists building and innovating incrementally, working with what their predecessors have handed them and sculpting it into something new yet understandable to those around them. His sensitive reading allows us to see post-1859 German evolutionists as rational actors rather than irrationally stuck in some early-19thcentury moment with unmodern commitments. By challenging the very foundations of the standard narrative of German morphology, this careful, compelling account does at least as much as Richardss to undermine the association of 19th-century German Darwinism with a dangerously exceptional view of nature. But the two books offer very different reads. Is scientific progress a matter of personal anguish and triumph, or of intellectual chugging along? Our concept of it should be capacious enough to include both.
References and Notes
1. E. Haeckel, Generelle Morphologie der Organismen (Georg Reimer, Berlin, 1866). 2. The reviewer previously served as a press reader for both books at the manuscript stage. 10.1126/science.1169621

1170

16

27 FEBRUARY 2009

27 FEBRUARY 2009

1171

17

up, in explaining this case by in- distribuby something (later identified as genetic mutascience, present shaped by the organisms own activities (the so- vancing through a predetermined sequence of ulation became divided to terms oceanic islands. tions), but it was effect), not aimed inthis anytoo onepermitted direction stages within each family, driven by force derived dependent acts of migration opments that would push other naturalists toward tions in of past migrations, called Lamarckian but Here, Darwin followed Lyell in seeing that Darwin bioand, thus, left adaptive evolution essentially an evolutionary vision during the years he worked extinctions and (for but not multiple vectors of change. Evolution hadopento be from individual development. REVIEW opments that would push other naturalists toward geography must become a historical ended. He allowed a limited role for variation in isolation. By the late 1850s, the idea of profor Lyell) evolutionary adaptations. depicted as a branching tree in which each act of an evolutionary vision during the years he worked science, explaining present distribushaped by the organisms own activities (the sogressive evolution was widelyto recognized, andorderly the divided by geographical branching was thevariants result of aa more or less was unhis theory predict an pattern of proposals of the individual variants in a population was the relations among species. Several in By the late 1850s, the idea of protions in Populations terms past migrations, called Lamarckian effect), but this too permitted was widely recognized, It has been of argued that Darwins move of isolation. progressive evolution of the individual in population positive role of individual competition was being opments that would push other naturalists toward barriers will develop independently predictable migration of organisms to a new locarelations. away essentially undirected ruled out any possibility available in the 1830s deflected attention opments that would push other naturalists toward extinctions and (for Darwin but not multiple vectors of change. Evolution had to be gressive evolution was widely recognized, and the was inspired and the positive role of individual competition essentially undirected ruled out any possibilto a more historical viewpoint articulated by thinkers such as argued Herbert Peter J. Bowler that evolution could be shaped by a predeter- from the model of the branching an evolutionary vision during the years was he Spencer worked as each adapts to its new environtion.depicted At the same time, Darwin s undermined Itby has been Darwins move to a tree (11 ). for Lyell) evolutionary adaptations. as a branching treeshaped intheory which each act of an evolutionary vision during the years he worked positive role of individual competition being was being articulated thinkers such asthat Herbert could be by a predeby German romanticism [e.g. (12)], but a ity that evolution (Fig. 1). But keymore aspects of the viewpoint Darwinian vision in isolation. By the late 1850s, the idea of proment in its own way, and the posthe old idea that species were idealized types, Populations divided by geographical branching was the result of a more or less unhistorical was inspired by German William Sharp Macleay s quinary or circular mined developmental trend. There was no obin isolation. By the late 1850s, the idea of proarticulated by thinkers such as Herbert Spencer Peter J. Bowler But key aspects of the Darwinwas no incentive was provided by Spencer (Fig. 1). termined developmental trend. There more practical were truly original and wouldrecognized, not have occurred Charles Darwins theory of natural has been hailed of theof most innovative supposed barriers will develop predictable migration of organisms a new order. locagressive evolution was widely and the sibility thatindependently barriers be crossed fixed elements in a clearly defined to natural romanticism [e.g. (12)], but a more practical system classification that vious goal toward selection which it was aimed, and as it one gressive evolution was widely recognized, and the can (Fig. 1).every But key aspects of the Darwinian vision and would not have on the ian vision were truly original obvious goal toward which it was aimed, and the biogeographical insights gained as each adapts to its new environtion. At the same time, Darwin s theory undermined to any be other naturalist at the time. Here, Wallace contributions todid modern science. an When first pattern proposed in 1859, however, it was widely rejected positive role of individual competition was being occasionally allows for the branchSpecies had to be treated as populations of varyincentive was provided by the biogeographical genus contained five species that could arnot produce orderly of relations positive role of individual competition was being were truly original and would not have occurred Charles Darwins theory of natural selection has been hailed as one of the most innovative Here, produce anspecies orderly pattern of relations occurred to any comparison: other naturalist at the time. Beagle voyage (183136). The Galapagos it did not ment in its own way, and the posthe old idea that were idealized types, provides a good He too moved toward by his contemporaries, even by those who accepted the general idea of evolution. This article articulated by thinkers such as Herbert Spencer ingHerbert process Spencer of evolution that Darwin individuals, with no fixed limit on the range Peter J. Bowler insights gained onHere, the Beagle voyagePeter (1831 rangedit inwas a circle; each family five genera, so species. When The accusation that the theory however, articulated by thinkers such provided as J.36). Bowlering to any and other naturalist at the time. Wallace contributions to between modern science. first proposed in 1859, widely rejected accusation that theorder. theothe most obvious example Wallace provides a good comparison: He too between species. in The species sibility that barriers can be crossed fixed elements a clearly defined natural the idea ofs branching evolution drivenprovided by local identifies those depended aspects of Darwin s work that led him to develop this revolutionary theory, hierarchy. (Fig. 1). a But key aspects of He the too Darwinian vision conceived in vision the late 1830s. It was of possible variation. The Galapagos species the most obvious on of through the taxonomic Chambers on random variation indicated con- idea (Fig. 1). But key aspects of the Darwinian provides good comparison: moved toward by his contemporaries, even by those who accepted the the general evolution. This article occasionally allows forwithin the branchSpecies hadon to be treated as variation populations of varythe idea of branching evolution group can be ry depended random indicated of how the relations moved toward adaptation, but even he did not share Darwin s including his studies of biogeography and animal breeding, and his recognition of the role played were truly original andevolution would not have occurred Charles Darwin theory of natural has been hailed as one of the most innovative by approaching the a problem of the example of how the relations within a group can Vestiges of the Natural of idea Creation cerns ofof his opponents onthat this score. As Darwin were truly original and would not have occurred Charles Darwins theory of natural selection has beenthe hailed oflocal branching driven local identifies thoses aspects Darwin s selection work led him to develop this revolutionary theory, History the ing process of that Darwin individuals, with no fixed limit rangeas one of the most innovative but even heby did not the ing concerns of his opponents onon this score. driven by adaptation, explained byevolution supposing that an original The Tree of Life insight that the work of the animal breeders throws by the struggle for existence. to any other naturalist at the time. Here, Wallace contributions to modern science. When first proposed in 1859, however, it was widely rejected origin of new species through a study be explained by share supposing that pop- to modern evolution terms of parallel lines adhimself made clear, variation was certainly caused to any other naturalistconceived at the time. Here, contributions science. When first proposed in 1859, however, it was widely rejected adaptation, but even he did not Darwin s an original including his studies of biogeography and animal breeding, and hisdepicted recognition of thein role played in the late Wallace 1830s. Itup, wasin this case possible variation. made clear, variation was became divided share Darwins insight that the work of the As of Darwin himself population light on the process of natural provides a good comparison: He selection. too moved toward by the his contemporaries, even by those who accepted the general of evolution. article oftoo biogeography that Darwin was One innovation the heart of Darwin s theory ulation became divided up, in this case by invancing through This a predetermined sequence of by existence. something (later identified as genetic muta- idea provides a good comparison: He moved toward by his contemporaries, even by at those who accepted the general idea of evolution. This article by approaching the problem of the insight that the work of the animal breeders throws by struggle for identified toopenoceanimal breeders throws light ondriven the process of certainly caused by something (later by independent acts of his migration The theory was both original and disturbing. pre-existing ones in a progressive lead- by light he those publication of s in On Theso Tree of Life the idea of process branching evolution by local identifies aspects ofCharles Darwin Darwin s work that led him to develop this revolutionary theory, led driven to construct model of obvious it led is hard us to this revolutionary theory, dependent acts of migration to oceanic islands. stages within eachsequence family, driven force derived tions), but it was not aimed any one direction of new species through a study the idea of branchingorigin evolution by local identifies those seems aspects of Darwin today s workthat that him for to develop on the of natural selection. as genetic mutations), but it was not radical aimed in anic islands. Here, Darwin followed Lyell selection. It was not just that the idea offollowed natural selection ing up opento and humans (5).individual But if the general idea of natural thehis Origin of thus, Species in 1859 evolution is widely adaptation, but even he did not share Darwin s including studies of biogeography and animal breeding, his recognition of the role played ended, divergent evolution. Wallace appreciate just how new and how it of biogeography that Darwin was One innovation at the heart of Darwin s theory Here, Darwin Lyell in seeing that biofrom development. and, left adaptive essentially including his studies of biogeography and animal breeding, and his recognition of the role played adaptation, but even he did not share Darwins The theory theory was was both both original original and disturbing. pre-existing ones in a progressive sequence leadhe publication of Charles Darwin s On he publication of Charles Darwins On vi- insight Thethat and disturbing. one direction and, thus, left adaptive evomust become of evolution was entirely new, Darwins any seems in seeing that biogeography challenged thework beliefof that the world was designed evolution was notnot entirely new, Darwin s vision supposed to existence. have initiated aarevolution led toanimal construct his model of openso time. obvious today that it isproposed hard for us to the animal breeders throws by thethe struggle for developed a similar model and tested at the Lamarck had that geography must becomeby a historical ended. He allowed limited role ing for variation insight that the work of the breeders throws the struggle was for existence. was not just that thethe idea of natural selection up to humans (5). But if the general idea of It Origin of Species in 1859 is widely in 1859 is widely that the idea of natural selection open-ended. He allowed a Origin of Species sion of how the process worked certainly was. It was not just lution essentially a historical science, explaining present ended, divergent evolution. Wallace appreciate just how new and how radical it by a wise and benevolent God. There was a wider of how the process worked certainly was. Alboth inthe science and in Western culture. Yet there light on the process of natural selection. it during his explorations in South there might be natural processes adapting spescience, by the organisms own activities (the was so- not entirely new, Darwins vision challenged the belief that the world light on the process of natural selection. wasexplaining designed present distribuevolution supposed shaped to have initiated a revolution developed a similar model and testedArchipelago at the time. Lamarck had proposed that theory was eventually paralleled the world was designed theDarwin organdistributions in terms of past migrations, supposed to have initiated a revolution belief that limited role for variation shaped Although the elementtheory of the teleology or goal-directedness almost though the theory eventually paralleled by challenged have been frequent claims that Darwinism was was both original and disturbing. pre-existing inwas aworked progressive sequence publication of Charles Darwin s On America and the Malay cieswas to of changes in their environment. But tions in terms of past migrations, called Lamarckian effect), but this too permitted The theory was both original and disturbing. pre-existing ones in a progressive sequence leadhe publication Charles Darwin s On by wise and benevolent God. There was a wider of how the ones process certainly was.leadAl- by aThe both inhe science and in Western culture. Yetthere it during his explorations in South there might be natural processes adapting speinthe science and in Western culture. Yet there its basic outa Darwin but not for by Wallace, Darwin had conceived by a wise and benevolent God. There was isms own activities (the so-called Lamarckian extinctions and (for both universally accepted at the time. Most thinkers Wallace, Darwin had conceived its basic outline somehow in the air at the time, merely waiting It was not just that the idea of natural selection ing had up to 5). But if the general ideaby of element ofclaims Species in Darwinism 1859 is widely (modern Indonesia). was perhaps the first to realize that ing if adaptation extinctions and (for Darwin but not multiple vectors of change. Evolution to humans be It was not just that the idea of natural selection up to humans ( 5 ). But if the general idea of the Origin of Species in 1859 is widely of teleology or goal-directedness almost though the theory ( was eventually paralleled have been Origin frequent that was America and the Malay Archipelago cies to changes in their environment. But Darwin claims that Darwinism was but this too vectors have been frequent line inlate thewas latenot 1830s, after his Darwin return from the wider element of teleology or goal-directedness adaptations. including Jean-Baptiste Lamarck and Chambers ineach the 1830s, after his return from the voyage for someone to put few as readily available points challenged the belief that the world was designed evolution entirely new, soutline vision supposed to a initiated a revolution Adrian Desmond andPopulaJames toto the local environment was themultiple onlyifmechanism for Lyell) evolutionary adaptations. a branching tree in which act of that the evolutionary world was designed was not entirely new, Darwins vision challenged the belief Lyell) supposedeffect), have initiated a permitted revolution universally accepted at the time. Most thinkers Wallace, Darwin had conceived its basic somehow in thedepicted air have at the time, merely waiting (modern Indonesia). was perhaps the first to realize that evolution adaptation took it for granted that the development of life on by geographical ofor H.M.S. Beagle . Beagle He worked on it in relative merely waitvoyage of H.M.S. . He worked on itAlin almost accepted at the was time. Most had tomajor be depicted as a process worked certainly was. Al- by a wise and benevolent divided byhave geographical barriers will somehow in the air at the time, of change. Evolution together in the right way [for instance (1)]. The universally tions by a wise and benevolent God. There a wider of how the process worked certainly was. both in science and in Western culture. Yet there Moore recently proposed that of evolution, there would be implications Populations divided branching was the result of a more less unGod. There was a wider of how the both in science and in Western culture. Yet there for someone to put a few readily available points in the late 1830s, after his return from the voyage including Jean-Baptiste Lamarck and Chambers Adrian Desmond and James to the local environment was the only mechanism earth represents the unfolding of a Lamarck coherent plan isolation over the over next 20 years, until the arrival of took fact for that Alfred Russel Wallace (Fig. 1) indepennext 20 years, the and branching as adapts to its ing someone to claims put few readily available relative isolation thinkersincluding Jean-Baptiste branching tree in which each act of independently element of teleology or goal-directedness almost though theory eventually paralleled by have been that Darwinism was to Darwin s hatred ofeach slavery prompted for the whole system by which species are the clasbarriers will develop predictable migration of organisms aH.M.S. new the locaor goal-directedness almost though theory was eventually paralleled by element of teleology develop have been frequent claims that Darwinism was Moore have recently proposed that of evolution, there would be major implications it for granted that the development of life on independently of Beagle . was Hethe worked on it in until relative together in frequent the right waya[for instance ( 1 )]. The aimed at a predetermined goal. (This assumption Wallace s paper in 1858 precipitated the flurry of universally dently together formulated athe theory natural selection in right way [for instance in 1858 precipitated arrival of Wallaces paper Chamberstook it unfolding for granted that the developwas result of aDarwin more or less unpredictable in itsprompted own way, and(13 the points accepted at the time. Most thinkers its new Wallace, Darwin had conceived its basic outline somehow in the air at theof time, merely waiting move toward evolutionism ). sified into groups. s mentor in geology, as each adapts to environtion. At the same time, Darwin sin theory undermined Darwin shis hatred of thinkers slavery the whole system by which species are clasat theenvironment time. Most Wallace, Darwin had conceived its basic outline universally accepted new somehow in the airfor the at the time, merely waiting represents the of a coherent plan isolation over the next 20 years, until the arrival of earth fact that Alfred Russel Wallace (Fig. 1) indepenis still preserved in the very term evolution ; the activity leading to theleading publication of publication the Origin . aimed 1858 is taken as evidence forthat this position. But idealized Alfred Russel Wallace (Fig. of life on earth represents the unfolding of of organisms to a new location. At possibility that barriers can be crossed oc( 1 )]. The fact that the flurry of activity to the ment migration including Jean-Baptiste Lamarck and Chambers in the late 1830s, after his return from the voyage for someone to put a few readily available points Because many slaveholders argued Charles Lyell, had shown how his uniformitarian his move toward evolutionism ( 13 ). sified into groups. Darwin s mentor in geology, ment in its own way, and the posthe old idea species were types, including Jean-Baptiste Lamarck and Chambers in the late 1830s, after his return from the voyage for someone to put a few readily available points at a predetermined goal. (This assumption dently formulated a theory of natural selection in Wallaces paper in 1858 precipitated the flurry of Latin evolutio refers to the unrolling of a barriers scroll.) Historians quarried Darwin s in notebooks Darwin had created the outlines the theory Because many slaveholders argued Charles Lyell, had shown how his uniformitarian took itpreserved for granted that the development of life on cantogether of H.M.S. Beagle He worked on relative of of the Origin . have aimed at a predetermined undermined for the branching process 1) independently formulated theory a coherent plan the same time, Darwins theory casionally allows together in the right way [for (1natural )]. The activity that the black race was separately theory would allow the biogeographer to the resibility that be crossed fixed elements in aof clearly defined natural order. the development of life on of H.M.S. Beagle. He worked on it in relative took it for granted that in the right way [for instance (1)]. The is still in the very term evolution goal. ; the leading to . the publication ofit the Origin . 1858 is taken as evidence forainstance this position. But that the created black race was plan separately theory would allow the biogeographer to re- the next 20 years, until the arrival of earth represents the unfolding The represents explanatory framework centered on the for fact and letters tothe establish the complex process by 20 years earlier, and there were significant dif- isolation earth the unfolding of a coherent plan over next 20 years, until the arrival of fact that Alfred Russel Wallace (Fig. 1) indepenfrom the white, Darwin Historians have quarried Darwins notebooks in the very in the selection in 1858 is taken as evidence for this (This assumption is still preserved old idea that species were idealized types, fixed of evolution that Darwin conceived construct the migrations of species on an everoccasionally allows the branchSpecies had to be treated as populations of varyof a coherent isolation over that Alfred Russel Wallace (Fig. 1) indepenHistorians have quarried Darwins notebooks Latin evolutio refers to the unrolling of a scroll.) Darwin had created the outlines of the theory created from the white, Darwin construct the migrations of species on an everaimed at a predetermined goal. (This assumption Wallace s paper in 1858the precipitated flurryby of The dently formulated a theory natural selection in wanted toby show that all races share changing earth. Populations couldWallace sometimes elements in a clearly defined natural order. Spe - in 1858 precipitated the flurry of aimed at a predetermined late 1830s. It was approaching the probing process of evolution that Darwin individuals, with no fixed limitand on the range goal. (This assumption s paper dently formulated a theory of natural selection in explanatory framework centered on the letters to establish complex the process 20 years earlier, ing and there of were significant difwanted to show that ancestry, all races share changing earth. Populations could sometimes is still preserved in the very term evolution ; the 1858 is taken asof evidence this position. But activity leading to the publication of the Origin. a common and he through realized become by position. geographical barriers, soleading that to the publication of the Origin. be treated as populations of varying origin of new species cies haddivided to for lem ofterm the conceived in the late 1830s. possiblefor variation. is still preserved in the very evolution ; the activity 1858 It iswas taken as evidence this But a common ancestry, and he realized become divided by geographical barriers, so that of a scroll.) Historians have quarried Darwins notebooks Latin evolutio refers to the unrolling Darwin had created the outlines of the theory that this could defended what was once a single could split into limit on the Historians range of have quarried Darwins notebooks Latin evolutio refers a was individuals, with fixed study of biogeography that be Darwin by approaching the problem of the to the unrolling ofclaim a scroll.) Darwin had created the outlines of the species theory that this claim could be defended what was once no a single species could split into Life significant dif- and letters to establish the complex process by The explanatory frameworkorigin centered on the 20 years earlier,The and Tree thereof were by extending the throughout multiple branches adapting to separate environof new species through a study variation. led construct model of open-ended, centered on theidea and environletters to establish the complex process by The explanatory framework 20 years earlier,possible and there were significant by to extending the his idea throughout multiple branches adapting difto separate the animal kingdom. As a basis for ments (10( ). Evolution would become divergent of biogeography that Darwin was One innovation at the heart of Darwins theory Wallace developed a divergent evolution. the animal kingdom. As a basis for ments 10 ). Evolution would become a a divergent his thinking, this thesis to process, with some branches over and and led to construct his model of openseems so obvious today that it is hard for us to his thinking, this thesis is sure to is sure process, with some branchessplitting splitting over The Tree of Life similar model and tested it during his exmuch controversy, but if over again, whereas others to dead end Fig. generate generate much controversy, but if the Malay over again, whereas others came to a dead end ended, divergent evolution. Wallace appreciate just how new and how radical it One innovation at the heartcame of Darwins theory in South America and plorations Fig. Tree of Life, from Darwin s notebooks 2.2. Tree of Life, from Darwin s notebooks (22). (22). accepted it would emphasize the through extinction. accepted it would emphasize the through extinction. developed a similar model and tested was at the time. Lamarck had proposed that seems so obvious today that it is hard for us to Archipelago (modern Indonesia). crucial crucial role played his move toward a model These rigidly structured models of taxoimage of the tree oflife lifehad had appeared appeared in roleby played by his toward a model These rigidly structured models of taxoTheThe image of the tree of in it during his explorations in South there might be natural processes adapting spehow radical it was The idea of common descent now seems so Adrian Desmond andmove James Moore have appreciate just how new and of branching evolution based on based geographical relations and evolution made good good sense sense Darwin s notebooks of thelate late 1830s 1830s (Fig. 2) of branching evolution on geographical nomic relations evolution made Darwin s notebooks ofhad the (Fig. 2) nomic America and the Malay Archipelago cies to changes in their environment. But Darwin wonder what alternative hatred of slavat the time. Lamarck proposed that there obvious that weand might recently proposed that Darwins diversity. to anyone embedded in a vision of nature as a and was proposed independently by Wallace in diversity. to anyone embedded in a vision of nature as a and was proposed independently by Wallace in (modern Indonesia). was perhaps the first to realize that if adaptation might be natural processes species to predictable, models could have been proposed account for This ery prompted move evolutionism model was so his radical that toward many late orderly system governed by to a divine a paper published in 1855. adapting Both realized that it This modelmany was so radical thatargued many that late predictable, orderly system governed by a divine a paper published in 1855. Both realized that it plan. Adrian Desmond and James to the local environment was the only mechanism But Darwin was Because slaveholders changes in their the relations among Several (13). evolutionists were unable to accept Such a world view species. made it difficult toproposals ac- 19th-century explained whyenvironment. naturalists were able to arrange 19th-century evolutionists were unable to accept plan. Such a world view made it difficult to acexplained why naturalists were able to arrange Moore have recently proposed that of evolution, there would be major implications it in full. Mayr argued that the theory of from the that the of life on earth attention might be away species into groups within groups, using descent to cept inhistory the 1830s deflected created perhaps the first to realize that if adaptation theErnst black race was separately available it in full. Ernst Mayr argued that that the theory of cept that the history of life on earth be white, species groups ancestor within using Darwins hatred of slavery prompted for the whole system by which species are clascommon descent was one of Darwin s greatest and unpredictable, dependant frominto a common to explain the descent under- essentially branching tree (might 11). Wilthe local environment wasgroups, the only mechanism from theirregular model of the Darwin wanted to show all races common descent was one of Darwin s greatest essentially irregular and unpredictable, dependant from a common ancestor to explain the underhis move toward evolutionism (13). sified into groups. Darwins mentor in geology, achievements, addition ancestry, to natural selection the Sharp hazards of migration, isolation, and local lying similarities. Closely be related species have di- on of evolution, there would major implications system share ain common and he realized that liam Macleays quinary or circular achievements, in addition to natural of migration, isolation, and local lying similarities. related ancestor, species whereas have di- on the hazards itself (14 ). So it was, but I think Mayr over- selection Darwin was led toward his alternaverged recently Closely from a common Because many slaveholders argued Charles Lyell, had shown how his uniformitarian system by which species are clas- adaptation. for the whole of classification supposed that every genus con- this claim could be defended by extending the estimated the ( rapidity with whichbut other model in Darwin part because was more his interthe recently ancestry of more distantly related forms must tive itself 14). So it was, I naturalthink Mayr overadaptation. was he led toward alternaverged from a common ancestor, whereas that the black race was separately theory would allow the biogeographer to rementor in geology, five species that could be arranged in a were As a basified groups. Darwins idea throughout animal kingdom. tained Fig. 1. Charles construct Darwin, Alfred Russel Wallace, and on Herbert Spencer. ists converted to rapidity thethe theory. Many of the adaptation thanbecause cosmic teleology, be into traced further down the family tree to ested estimated the with which other naturaltivein model in part he was thanks more interthe ancestry of moreback distantly related forms must created from the white, Darwin the migrations of species an everfive genera, and so on through Charles Lyell, had shown how his uniformitar- tocircle; each family sis for his thinking, this thesis is sure to genernon-Darwinian theories of evolution proposed the influence of William Paley s natural thefind the common point of origin. be traced further back down the family tree to ested in adaptation than cosmic teleology, thanks ists were converted to the theory. Many of the wanted to show that all races share changing earth. Populations could Herbert sometimes Fig. 1. Charles Darwin, Alfred Russel Wallace, Spencer. Darwin had created the outlines ofandand thetheory complex process by term would allow the biogeographer to ology. Vestigesduring of the if late accepted it would letters to establish evolution; the Latinchallenged evolutio refers to the the taxonomic hierarchy. Chamberss position. ian theory atemuch eclipsecontroversy, of Darwinism but in the 19th proposed Natural selection replaced divine The idea of point common descent now seems so theory of natural selection this vision which he developed his (69 ). Darwin ferences But between the ways in by which he and non-Darwinian theories of evolution to the influence of William Paley s benevnatural thefind the common of origin. a common ancestry, and he realized become divided geographical barriers, so that century were introduced with the aim of subvertolence as an explanation of adaptation. Unlike obvious that we might wonder what alternative and there were sigDarwin unrolling of a scroll.) The explanatory framework of species on an everrole played by his move the theory 20 years earlier, which he developed his theory ( 6 9 ). reconstruct the migrations the Natural History of Creation depicted evoluemphasize the crucial of relations. a highly creative thinker who synthesized a of nature as an orderly pattern Wallace formulated their ideas. In this essay, I wassplit during the eclipse of Darwinism in the late 19th selection replaced divine benevThe idea of common descent now seems so ology. Natural that thisthis claim could be defended what was once single species of on natural selection challenged vision which heinto developed his theory (6synthesized 9). Darwin theory ferences between the ways in a which he and could ing the toward implications of the principle of common and Chambers, Darwin not expect models earth. could have been proposed to account for Macleay the theory of natural selection chalof parallel lines did advancing through based on nificant differences between the ways in which was a highly creative thinker who centered changing Populations could sometimes a model of branching evolution tion in terms Darwin s world view was profoundly differnumber of key insights, some derived from his argue that Darwin was truly original in his thinkFig. 1. Charles Darwin, Alfred Russel Wallace, and Herbert Spencer. olence as an explanation of adaptation. Unlike century were introduced with the aim of subvertweRussel might Wallace, wonder what alternative by extending the idea Fig. throughout 1. Charles obvious Darwin, that Alfred and Herbert Spencer. multiple to Iseparate environnature as vision an orderly pattern of relations. was a highly creative thinker who synthesized a of Wallace formulated their branches ideas. In adapting this essay, he and Wallace formulated their ideas. In this esof key insights, some derived from his as an orderly pattern become divided by geographical barriers, so sequence of stages within each lenged this of nature a predetermined geographical diversity. a number adaptation of pop- As a basis for work and others from currents circulat- ent because he argued that thethe ing, and I support this(10 claim by addressing the scientific models could have been proposed to account for Macleay and Chambers, Darwin did not expect ing the implications of the principle of common animal kingdom. ments ). original Evolution become a divergent Darwins world view was profoundly differofwork key insights, some derived from his of relations. argue Darwin was truly in would his thinkthat Darwin was truly original inand his number from currents circusplit family, This model was so radical that many late say, I that argue scientific and others that what was once a single species could driven by force derived from individual ulations to their local environment was the solethesis ing over in his cultural environment. Few would now ent related issue of defining just why the theory was 224 9 JANUARY 2009 his VOL 323 theory of natural selection challenged this vision which he developed his theory ( 6 9 ). Darwin ferences between the ways in which he his thinking, this is sure to process, with some branches splitting and theory of natural selection challenged this vision he developed theory (6SCIENCE 9). Darwinwww.sciencemag.org ferences between the ways in which he and which because heworld argued that the adaptation of popwork and others from currents circulating, and Iand support this claim by addressing the scientific this claim by addressing Darwins view was profoundly difthinking, I support lating in his cultural environment. Few would into multiple branches adapting to separate en- development. 19th-century evolutionists were unable to accept cause of transmutation. Many people found it accept the claim that evolution by natural seso disturbing to his contemporaries. of nature as an orderly pattern of relations. was a highly creative thinker who synthesized a Wallace formulated their ideas. In this essay, I generate controversy, butformulated if over again, whereas others was came to a in dead Wallace their ideas. In this essay, I was a highly creative thinker who synthesized a of nature as an orderly pattern of relations. ulations to their local environment wasmuch the sole ing his end cultural environment. now s related issue of defining just why the theory Fig. 2. Tree ofFew Life,would from Darwin notebooks (22he ). why theory now accept the claim evolution by natural that the agent adaptation of emphasize the These rigidly structured models ofwww.sciencemag.org taxonomic it in full. Ernst Mayr argued that the theory of the Darwin related issue of defining just ferent argued a 224 vironments (10). Evolution would become 9 JANUARY 2009 VOL 323 SCIENCE hard tobecause seenatural selection as the ofdiffereither lection was in insights, the air.that Darwin approached the cause wasthrough certainly not the first to sug- number Darwin s world viewMany was profoundly of claim key some derived from his argue that Darwin was truly original inthe his thinkaccepted it would extinction. Darwins world view was profoundly differargue that Darwin was truly original in his think- number of key insights, some derived from his of transmutation. people found it accept the that evolution by natural seso disturbing to his contemporaries. in the air. Darwin approached the was the process, with some branches splitting made good sense to any- common descent was one of Darwins greatest was so disturbing to his contemporaries. selection was populations to their local environment relations and evolution divergent divine benevolence or of a rationally structured subject in a way that was significantly different gest the idea of evolution as an alternative to ent because he argued that the adaptation of pop- toward scientific work from currents circulating, and I support this the had crucial role played by his move model These rigidly structured models of taxoTheclaim image of addressing the tree life appeared inin and ing,a and I support this claim by addressing the scientific work and others from currents circulat- ent because he argued that the adaptation of pophard to see natural selection as the agent of either lection was the others air. Darwin approached the Darwin was certainly notby the first toof sugwas certainly not the first to sug significantly different subject in cultural a way that was sole cause of transmutation. Many people found over and over again, whereas others came to a one embedded in a vision of nature as a predict- achievements, in addition to natural selection itcosmic teleology. Selection adapted species to anon geographical from any of the nomic other efforts being made to extheDarwin creation of defining species by God. J. B. Lamarck s 1830s ulations to their local environment was the sole ing in his environment. Few would now related issue just why the theory was of branching evolution based relations and evolution made good sense Darwin s notebooks of the late (Fig. 2) related issue of defining just why the theory was ing in his cultural environment. Few would now ulations to their local environment was the sole gest the idea of evolution as an alternative to subject in a way that was significantly different divine benevolence or of a rationally structured as an alternative to accept made to exto see natural selection as the agent of think Mayr gest thepublished ideato of evolution from any of the other being end through extinction. able, orderly system by a cause divineof plan. self (14). Many So it was, but I found it hardof ever-changing and itpeople did so by killing plain the history ofthat lifeefforts on earth. He had unique theory, incontemporaries. 1809, had been widely discause transmutation. Many found it the claim evolution bymade natural se- of so disturbing his diversity. anyone embedded ina a nature as a environment, and was independently Wallace in transmutation. people it overestimataccept the claim that evolution by governed natural seso disturbing to dead his contemporaries. cosmic teleology. Selection adapted species to an from any of the to other efforts being tovision exthe creation of species byproposed God. J. B. Lamarcks by or of a rationally The image of the tree of life had appeared a world view made it difficult to accept the creation of species by God. J. B. Lamarcks plain the history of life on earth. He had a unique either divine benevolence ed the rapidity with which other Such off useless variations in a ruthless struggle for combination of scientific interests that alerted cussed, although generally rejected ( 2 4 ). Robert hard to see natural selection as the agent of either lection was in the air. Darwin approached the Darwin was certainly not the first to sugThis model was late was certainly not the first to sug- lection was in the air. Darwin approached the hard to see natural selection as the agent of either naturalists were predictable, orderly system governed by a divine environment, paper published inwidely 1855. disBoth realized that it of Darwin ever-changing and it did soso byradical killing that many plain the history life on earth. He had a unique theory, publisheda in 1809, had been theory, published in 1809, had been widely discombination ofignored scientific interests thatdifferent alerted of the late (Fig. earth might be essenstructured teleology. Selection Darwins as notebooks that the history of life on converted to the theory. Many of the non-Darexistence. cosmic This 19th-century did notof seem the kind structured ofadapted process him to topics by significantly other naturalists. Heit difficult Chambers s Vestiges of the Natural History of able divine benevolence or a rationally subject in a way that was gest the idea of evolution as an alternative to evolutionists were unable to accept plan. Such a world view made to acexplained why naturalists were to arrange divine benevolence or of a rationally structured subject in 2) a way that was significantly different gest the idea ofin evolution an alternative to 1830s cussed, although generally rejected (24). Robert combination of scientific interests that alerted off useless variations in a ruthless struggle for generally rejected (2 4the ). groups, Robignored by other naturalists. He species to an ever-changing environment, and it the Wallace in thetially on winian of evolution proposed during the cussed, although him to topics and was proposed independently and unpredictable, theories thatmight would be instituted by a benevolent God, certainly drew on ideas widely discussed at exthe Creation of 1844 sparked a debate over poscosmic teleology. adapted species to an from any of the other efforts being made to the creation species God. J. B. Lamarck s using itSelection in full. Ernst argued that theory of of species cept that history of life on earth species into groups within descent cosmic teleology. Selection adapted species to an from any of otherirregular efforts being made to ex- dependant the creation by God. J. B. Lamarcks by existence. be This did not seem the Mayr kind of process him to topics ignored by the other naturalists. He Chambers s of Vestiges ofby the Natural History of especially because its essentially selfish nature time, but was forced by his scientific interests to sibility that new species were produced from the Natural History discussed at the off useless variations in a ruthadlate 19th century ert Chamberss Vestiges of certainly drew on ideas widely did so by killing a paper published in 1855. Both realized that it the hazards of migration, isolation, and local eclipse of Darwinism in the ever-changing environment, and it did so by killing plain the history of life on earth. He had a unique theory, published in 1809, had been widely discommonby descent was oneGod, of Darwin s greatest essentially irregular and unpredictable, a common explain the undertheory, published in 1809, had been widely dis- plain the history of life on earth. He had a unique ever-changing environment, and it did so by killing would be instituted a benevolent certainly drew on ideas widely discussed at the thatdependant Creation of 1844from sparked a debateancestor over the to posmeant that a parasitic way of life was anot perfectly use those sources of inspiration in migration, a that highly origthe This did seem explained why naturalists were able to arrange the of Creation of 1844 sparked a debate over time, but was forced by his scientific interests less struggle for existence. aptation. Darwin was led toward his alternative were introduced with the aim of subverting off useless variations in a ruthless struggle for combination of scientific interests alerted cussed, although generally rejected ( 2 4 ). Robert achievements, in addition to natural selection on the hazards of isolation, and local lying similarities. Closely related species have dioff useless variations in a ruthless struggle for combination of scientific interests that alerted cussed, although generally rejected ( 2 4 ). Robert sibility that new species were produced from time, but was forced by his scientific interests to especially because its essentially selfish nature natural adaptive response in some circumstances. inal way. to use those sources ofby inspiration in a led highly within groups, because he wasHe more existence. interestedin possibility that new species produced from the kind of process that would be instituted species the principle of common descent model in part existence. aThis did not seem theit kind process him to topics ignored other He Chambers s Vestiges ofrecently thewere Natural ofancestor, itself (14 ). was, but I by think Mayr overadaptation. Darwin was toward his alternaverged from aHistory common whereas This implications did not seemof the kind of process him descent to topics ignored by other naturalists. Chambers s Vestiges ofinto the groups Natural History of using School of Philosophy and Anthropological Studies, The Queen's meant that parasitic way ofSo life was aof perfectly use those sources of inspiration innaturalists. a highly origMore seriously for the idea of cosmic teleTo some Darwin may have at been ones in a Road progressive sequence explain the underto thebe instituted American neo-Lamarckians Edward original way. a more benevolent God, especially its essenfrom a common ancestor adaptation than discussed cosmic teleology, pre-existing (15). Theby that would be instituted by abecause benevolent God, certainly drew extent, on ideas widely discussed thewas Creation of 1844 sparked a debate over the posUniversity of Belfast, University Belfast, Belfast, Northern estimated the rapidity with which other naturaltive model in part because he interthe ancestry of more distantly related forms must that would a benevolent God, certainly drew on ideas widely at the thanks Creation of 1844 sparked a debate over theto posnatural adaptive response in some circumstances. inal way. School of BT7 Philosophy and Anthropological The Queen's ology, Darwin s ists supposition that the production merely ahead of his time, anticipating develIreland, 1NN, UK. E-mail: p.bowler@qub.ac.uk especially because its essentially selfish nature time, but was forced by his scientific interests to sibility that species were produced from if Studies, the general idea To some extent, Darwin may have been nature meant that a parasitic way have proposed that leading up tonew humans (5). But lying similarities. Closely related species influence of William Paleys natural theology. Drinker Cope and Alpheus Hyatt tially selfish were converted to the theory. Many of the ested in adaptation than cosmic teleology, thanks be traced further back down the family tree to especially because its essentially selfish nature time, but was forced by his scientific interests to sibility that new species were produced from More seriously for the idea of cosmic teleTo some extent, Darwin may have been University of Belfast, University Road Belfast, Belfast, Northern meant that a way of that life was a response perfectly use those sources inspiration in a William highly origdeveladaptive Natural selection replaced divine benevolence merely ahead of his time, of life Darwin was a parasitic perfectly natural diverged recently from a common the evolution of each group should be seen as a theories of evolution proposed toof the influence of Paleys natural thefind thep.bowler@qub.ac.uk common point of origin. merely meant that a parasitic way of life was a perfectly use ancestor, those sources of inspiration in a highly origology, s non-Darwinian supposition the production ahead of his time, anticipating anticipating develIreland, BT7 1NN, UK. E-mail: natural adaptive response insome circumstances. inal way. that to- divine related as an explanation of adaptation. Unlike Macleay opments other naturalists in some circumstances. of parallel moved through the same whereas the ancestry of more distantly during the eclipse of Darwinism in the late 19th ology.push Natural selection replaced benevThe idea of Studies, common descent now seems so would natural adaptive series response in some lines circumstances. inal way. School of Philosophy and Anthropological The Queen's School of Philosophy and Anthropological Studies, The Queen's www.sciencemag.org SCIENCE VOL 323 9 been JANUARYMore 2009seriously for the idea of cosmic tele- 223 Toan some extent, Darwin may have More seriously for the idea of cosmic teledown the his theory of developmental stages, an updated ward evolutionary vision the years forms must be traced further back and Chambers, Darwin did not expect hierarchy century were introduced with the aimUniversity of subvertolence as anduring explanation of he adaptation. Unlike obviousRoad thatBelfast, we might wonder alternative More seriously for the idea of cosmic teleTo some extent, Darwin may have been University of Belfast, University Belfast, Northern what of Belfast, University Road Belfast, Belfast, Northern ology, Darwins ing supposition that the the of production merely ahead of his time, anticipating develIreland, BT7 1NN, UK. E-mail: p.bowler@qub.ac.uk 223of www.sciencemag.org SCIENCE VOL 323 the 9Darwin JANUARY 2009 the implications the principle common Macleay and Chambers, not expect models could have been proposed to account for isolation. By the late 1850s, idea did that production ology, Darwins supposition family to find the common point of origin. to predict an pattern of relations. of the suggested in Chamberss worked in ology, Darwinsversion supposition thatidea the production merely ahead of his time, orderly anticipating develIreland, BT7 1NN, UK. E-mail:tree p.bowler@qub.ac.uk

Darwins Originality Darwins Originality Darwins Originality

REVIEW REVIEW

REVIEW

Darwins Originality

T T T T

CREDIT (LEFT STAPLETON TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS CREDIT (LEFT TO RIGHT): COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS CREDIT (LEFT TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

CREDIT (LEFT TO RIGHT): STAPLETON COLLECTION/CORBIS; HULTON-DEUTSCH COLLECTION/CORBIS; MICHAEL NICHOLSON/CORBIS

18

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

224

www.sciencemag.org SCIENCE VOL323 323 SCIENCE 9 JANUARY 2009 9 JANUARY 2009 VOL www.sciencemag.org

223

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

223

19

CREDIT: SYNDICS OF CAMBRIDGE UNIVERSITY LIBRARY

REVIEW
Vestiges. The similarities linking the species in directed variation among individual organisms. tion thinking, and although he may have exa genus were due not to a recent common ances- Although convinced that the degree of variabil- aggerated the extent to which Darwin himself try, but to parallel trends independently reaching ity was artificially enhanced under domestica- made the conceptual transition, the subsequent the same stage of development. Like Chambers, tion, Darwin, nevertheless, accepted that there development of the selection theory brought this they endorsed the recapitulation theory (ontog- must be some equivalent variability in every implication out more clearly. In the debates that followed the publication eny recapitulates phylogeny, in the terminology wild population. The analogy with artificial seintroduced by Ernst Haeckel) and saw evolution lection then allowed him to depict natural selec- of On the Origin of Species, the analogy with role as the addition of preordained stages to ontog- tion as a parallel process in which a few variant artificial selection continued to play a key REVIEW eny. Adaptation was not crucial once the basic individuals, in this case with characters useful to by forcing even Darwins critics to think about descent (15). The American neo-Lamarckians scale and that could be investigated directly. interbreeding individuals. Traditionally, species character of the group was established, and the species rather than the human breeder, sur- the problems of heredity and variation in a new Edward Drinker Cope and Alpheus Hyatt pro- There was a well-developed network of breeders were treated as idealized types with a fixed esof the of group andbe reproduce. with harmful characsuch Fleeming Jenkin, the linear, orthogenetic evolution vive way (18 ). Opponents sence, any variation fromas the norm being trivial by this Those time, and although their ideas about posed that the evolution each group should byand thevariation struggle for existence, on large variations might eventually generate bizarre nonadaptive ters through are eliminated who saw selection working and impermanent. The breeders knew that they heredity were distinctly pregenetseen as a series of parallel lines moved could produce huge changes in structure by still acical (like Darwin own), they a very clear the same hierarchy of developmental an the up- breeder to extinctionthe the- stages, any had animal just as will notspermit or sports of nature, were, nevertheless, characters as a prelude variations over a number of appreciation of have how they produced changes in cumulating dated version of the idea suggested Chambers s if Darwin could make no into he working ory of racial senility. reproduce it does not the character within normal the framework defined by this Vestiges. The similarities linking the species in a their artificially small populations. The insight generations. When Darwin linked this informasense of the theory proposed by Cope and Hyatt, wants. It was the breeders who taught Darwin analogy. For supporters such as Francis Galton, genus were due not to a recent common ancestry, that they worked by selection may have been tion with his conviction that species could change is not directed preor- among helped to was clarify the nature because he could not imagine an evolutionary artificial selection variation indefinitely over time, he driven toward a (this is toward the pointsome of contention but to parallel trends independently that reaching the important trends. But Like the Chambers, build s on his exist- but of paving the popway dainedthey goal, allowing him to heredity selection, process driven by predetermined new form of and species concept in which the experts studying Darwin notebooks), theboth same stage of development. ulation becomes paramount. natural rangegeof breeders certainly evolution taught himmust one be thing. for He the endorsed the in recapitulation theory (ontogeny flourished the late 19th fact that such theories ing conviction that adaptive revolutionary impact The of Mendelian variability becomes of the species character, that in process. a domesticated population there phylogeny, the terminology intro- realized the in theory The notion of part hard heredity was incentury demonstratesrecapitulates just how radical an open-ended, branching netics. duced by Ernst Haeckel) and saw evolution as is always a fund of apparently purposeless and not the result of accidental deviations from a At the same time, the breeders attitude to- troduced in opposition to the soft form of of open-ended, divergent evolution was to the the addition of preordained stages to ontogeny. undirected variation among individual organisms. fixed norm. This is what Mayr called the transiDarwin toward the view ward pushed inheritance implied by the Lamarckian process. naturalists of the time. tion from typological thinking to population thinkconvinced that the degree of variabilAdaptation was not crucial once the basic variation char- Although of domestication, inter- The undirected naturehe ofmay variation was clarified that species is artificially just a population ing, and although have exaggerated the ity was enhanced under acter of the group was established, and the the linear, extent to the which Darwin himself made the conDarwin, nevertheless, accepted that there must be through might even- individuals. Artificial Selection orthogenetic evolution of the group breeding study of large populations by Traditionally, species both ceptual transition, the the subsequent of some equivalent variability every wild tually generate bizarre nonadaptive characters These non-Darwinian models were ultimately idealized types with in a fixed es- popustudies of the Galton and through breeding development were treated as as a prelude to extinctionthe theory of racial lation. The analogy with artificial selection then the selection theory brought this implication out synthesis of the selection sence, any variation from the norm being trivial geneticists. Although it took some time for the marginalized by thesenility. Darwin could make no sense of the allowed him to depict natural selection as a par- more clearly. century. geneticists to accept their studies theory and genetics theory in the early and impermanent. The breeders knew that they In the debates the that situation, followed the publication in which a few variant individuals, proposed20th by Cope and Hyatt, because he allel process ultimately endorsed Darwins Genetic mutations seemed toimagine be essentially plu- process could produce changes in structure by of mutation of On the Origin of Species, the analogyclaim with inhuge this case with characters useful to acthe species could not an evolutionary driven artificial selection to play a key role rather than the human breeder, survive re- the by providing predetermined trends. But the fact that such normal just the source variations over a number of and that could affect only way the continued environment ralistic and undirected, cumulating by forcing was evenby Darwin s critics to thinkevoluabout Those with harmfulthis characters are elimtheories flourishedmechanism in the late 19th century dem- produce. that Darwins linked inforof random variation generations. When Darwin the population selection. Modern onstrates just how radical the theory of open- inated by the struggle for existence, just as the the problems of heredity and variation in a new required as its raw ended, material. This later devel- mation with his conviction that species could tionary developmental biology has reopened the divergent evolution was to the naturalists breeder will not permit any animal to reproduce if way (18). Opponents such as Fleeming Jenkin, over he was driven evolution can opment highlights the importance of another change indefinitely of whether variation saw selection working and on large variations it does not have time, the character he wants. It wasquestion the who of the time. species in variation which is be form of as open-ended aswere, Darwin and his still folinsight gained by Darwin in the late 1830s, his toward a newbreeders or sports of nature, nevertheless, who taughtconcept Darwin that not quite Artificial Selection working within the framework defined by this directed toward some preordained goal, allowing of the animal the population The natural believed. But the non-Darwinian vision decision to investigate the work becomes paramount. lowers analogy.unfolding For supporters such as Francis Galton, him to build on hispart existing conviction that adaptThese non-Darwinian models ultimately theirwere to an orderly, predictable breeders (Fig. 3) and his recognition that range of variability becomes of the species of evolution marginalized by the synthesis of the selection ive evolution must be an open-ended, branching artificial selection helped to clarify the nature method of artificial theory selection offered a useful character, not the result of accidental deviations plan has been essentially marginalized by accepof both heredity and selection, paving the way and genetics in the early 20th century. process. how the equivalent natural norm. This is what Mayr called the from a fixed tance of the key insights onimpact which Darwin based way of understandingGenetic for the revolutionary of Mendelian At the same time, the breeders attitude mutations seemed to be essentially pluexact role played by Dar- just to populaof natural selection. process operated. The transition typological genetics. The notion of hard heredity was toward variationthinking pushed Darwin toward his the theory ralistic and undirected, providing the sourcefrom of random variation that Darwins mecha- view that the species is just a population of introduced in opposition to the soft form of of his wins study of breeding in the formulation nism by required as its raw material. theory is much debated historians (16 17), This later development highdoubt of how important but there can be little lights the importance of another artificial andby natural selec the analogy between insight gained Darwin in the In this tion became in his later thinking. late 1830s, his decision tocase, investigate because the workeven of the animal Darwin was truly unique, Wallace breeders (Fig. 3) and his recogdid not take this step and dissociated himself nition that their method of artifiselection expressed from the link with artificial cial selection offered a useful way in Darwins later writings. of understanding how the equivDarwin turned to alent the breeders in search of a natural process operated. The exact role played by Darwins could be changed clue as to how a population study ofwhere breeding in the formulamodifications here at least was a situation tion of his theory is much debated were actually being produced on(16 a human by historians 17), buttime there scale and that could be directly. can be investigated little doubt of how impornetwork of breedThere was a well-developed tant the analogy between artificial natural selection became in his their ideas about ers by this time, andand although later thinking. In this pregeneticase, Darwin heredity and variation were distinctly was truly unique, because even they a very clear cal (like Darwins own), Wallace did had not take this step and they produced in appreciation of how dissociated himself changes from the link The expressed insight their artificially small populations. with artificial selection Darwins later writings. that they worked byinselection may have been Darwin turned to theamong breeders of contention important (this is the point in search of a clue as to how a notebooks), but the experts studying Darwins population could be changed He realbreeders certainly taught him one thing. here at least was a situation where ized that in a domesticated population there beis modifications were actually ing produced on a human time Fig. 3. Pigeons (23). purposeless and unalways a fund of apparently

interbreeding Traditionally, species scale and that for could be investigated directly. descent (15). The neo-Lamarckians TheAmerican Struggle for Existence that transition in the late 19th century. It individuals. the Men Who Discovered It (Doubleday, New were treated as idealized types with a fixed esThere was a well-developed network of breeders Edward DrinkerOne Cope and Alpheus Hyatt proYork, 1958). of the most disturbing aspects of Darwins did, however, highlight the harsher aspects of 2. P. Corsi, Age of Lamarck: sence, any variation from theThe norm being trivial Evolutionary by for this existime, and their of ideas about The posed that the evolution of each group should to the be struggle struggle. potential theory was its appeal the although consequences Theories in France, 17901830 and impermanent. The breeders knew that they (Univ. of heredity were distinctly pregenetseen as a series tence of parallel moved through that equates with and the variation clearly as thelines natural process implications were drawn out even more California Press, Berkeley, 1988). could produce huge changes in structure by ical (like Darwin s own), they had a very clear the same hierarchy of developmental stages, an upbreeders activity as a selecting agent. This very when Galton argued that it would be necessary 3. A. Desmond, The Politics acof Evolution: Morcumulating normal variations over a number of appreciation of how they produced changes in dated version of harsh the idea suggested in Chambers s vision of nature certainly threatened the to apply artificial selection to the human race phology, Medicine and Reform in Radical generations. When Darwin linked this informatheir artificially small populations. The insight Vestiges. The similarities linking the species in a London (Univ. of Chicago Press, Chicago, traditional belief in a benevolent Creator. The in order to prevent unfit individuals from that species could change genus were due not to a recent common ancestry, that they worked by selection may have been tion with his conviction 1989). term struggle for existence occurs in Thomas reproducing and undermining the biological 4.time, P. J. Bowler, Evolution: The a History of an he was driven toward but to parallel trends independently reaching the important (this is the point of contention among indefinitely over Robert Malthuss An Essay on the Principle health of the population. This was the eugenics (Univ. California Press, Berkeley, ed. concept inof which the popsame stage of development. Like Chambers, they experts studying Darwins notebooks), but the new form of speciesIdea of Population, although used in the context of program, and in its most extreme manifestation 3, 2003). The natural range of endorsed the recapitulation theory (ontogeny breeders certainly taught him one thing. He ulation becomes paramount. tribal groups competing for limited resources. at the hands of the Nazis, it led not just to the 5. J. A. Secord, Victorian Sensation: The Exof the species character, recapitulates phylogeny, in the terminology intro- realized that in a domesticated population there variability becomes part traordinary Publication, Reception, and Secret Darwin saw that population pressure would lead sterilization but also to the actual elimination of deviations duced by Ernst Haeckel) and saw evolution as is always a fund of apparently purposeless and not the result of accidental Authorship of Vestigesfrom of theaNatural History to competition between individuals and was per- those unfortunates deemed unfit by the state. Did Mayr called the transi- Press, Chicago, the addition of preordained stages to ontogeny. undirected variation among individual organisms. fixed norm. This is what of Creation (Univ. of Chicago the first to realize that it might represent Darwins emphasis the natural tion elimination from typological 2000). thinking to population thinkAlthough convinced that the degree on of variabilAdaptation was haps not crucial once the basic charby which the population could change variants help to create a 6. heD. a means of maladaptive Kohn, Ed., The Darwinian ing, and although may have exaggerated the Heritage: A acter of the group was established, and the linear, ity was artificially enhanced under domestication, processDarwin, workednevertheless, by climate through (19, 20 ). The of opinion in must which atrocities Centennial Retrospect (Princeton extent to which Darwin himself made the con- Univ. Press, accepted that there be such orthogenetic evolution oftime the group might evenNJ, 1985). theequivalent popu- became possible? the least fit variants within subsequent development of some variability in every wild popu- ceptual transition, thePrinceton, tually generate eliminating bizarre nonadaptive characters 7. J. Browne, Charles Darwin: Voyaging (Jonahas to be admitted that,then by making death it- theory lation and allowing theof better adapted to survive the selection brought this implication out lation. The analogyItwith artificial selection as a prelude to extinction the theory racial than Cape, London, 1985). breed. Thisno was whatof the philosopher in nature, Darwin introduced self a creative and more clearly. 8. J. Browne, Charles Darwin: The Power of allowedHerhim to depict natural force selection as a parsenility. Darwin could make sense the refer he to as the survival into Spencer would later a new and profoundly disturbing insight In thethe debates that followed the Cape, publication allel process in which a few variant individuals, theory proposedbert by Cope and Hyatt, because Place (Jonathan London, 2002). selection of an the fittest. Strictly speaking, an insight seems to have 9. ofM. J. S. Hodge, G. Radick, Eds., The Camof resonated On the Origin Species , the analogy with in this case withworld, characters useful that to the species could not imagine evolutionary process drivennatural bridge Companion Darwin with the thinking of many not underrequires differential artificial selection continued to play ato key role (Cambridge than the human breeder, survive andwho re- did by predetermined trends.only But the fact that reproduction such ratheramong Univ. Press, Cambridge, 2003). variants, thought that the pressure of with his theory. Darwinstand or accept the details by forcing even Darwin s critics to think about produce. Those harmful characters areof elimtheories flourished in the but lateDarwin 19th century dem10. M. J. S. Hodge, Stud. Hist. Biol. 6, 1 (1982). was necessary to makeinated it effective. Darwinism competition ism was not responsible for social the problems of 11. heredity and variation in a new Naturalists: by the struggle for existence, just as the onstrates just how radical the theory of open P. F. Rehbock, The Philosophical input frombreeder Malthus, henot permit Itevolution seems that without or eugenics in any way. all, some way (18 ). Opponents such as Fleeming Jenkin, will any animal tosimple reproduce if After ended, divergent was to thethe naturalists Themes in Early Nineteenth-Century British geneticists endorsed eugenics by analogy would not have come up with the theory. early saw selection working on large variations it does not have the character he wants. It was the who of the time. Biology (Univ. of Wisconsin Press, Madison, 1983). The idea of struggle was pervasive in thewho lit- taught even whileordismissing with Darwin animal that breeding sports of nature, were, nevertheless, still breeders variation is not J. Richards, The Meaning Artificial Selection exploited in some evolution. erature of the period, but could be directed natural selection as the mechanism working within 12. the R. framework defined by thisof Evolution: The toward preordained goal, allowing of Morphological Construction For supporters such as Francis Galton,and Ideological to build existing conviction that Spencer had on his many different In the 1850s, him And the Nazis wanted toadaptpurify a analogy. fixed racial These non-Darwinian models ways. were ultimately Reconstruction of Darwins Theory (Univ. of artificial selection helped toPress, clarify the nature ivebe evolution bewhich an open-ended, branching marginalized by the synthesis of competition the selectioncould they certainly did not want to admit already seen how turned must type, Chicago Chicago, 1992). of both heredity and selection, paving the way process. less had evolved gradually from an ape ancestry. theory and genetics the different, early 20th century. into ain very and in some respects But 13. A. Desmond, J. R. Moore, Darwins Sacred for primarily the revolutionary impact Mendelian same the breeders attitude Genetic mutations seemed to be essentially Cause: Race, of Slavery and the Quest for Humechanism of pluprogress (At 21).the For disturbing, bytime, proposing that evolution worked man (Allen Lane,was London, 2009). genetics. The notion ofOrigins hard heredity variation pushed toward the variants, ralistic and undirected, providing just the source individuals Spencer, the interaction betweentoward through the Darwin elimination of useless Dar14. E. Mayr, One Long Argument: introduced in opposition to the soft form of Charles Darview that the species is just a population of of random variation that Darwin s mechastimulated their efforts to adapt to the chang- win created an image that could all too easily be win and the Genesis of Evolutionary Thought nism required asing its raw material. social and physical environment. He then exploited by those who wanted the human race (Harvard Univ. Press, Cambridge, MA, 1991). This later development high- concept of the inheritance to conform to their own pre-existing ideals. In 15. P. J. Bowler, The Eclipse of Darwinism: Antiinvoked Lamarcks lights the importance of another of acquired characteristics to explain how these the same way, his popularization of the struggle Darwinian Evolution Theories in the Decades insight gained by Darwin in the accumulated over many gen- metaphor focused attention onto the individualAround 1900 (Johns Hopkins Univ. Press, self-improvements Baltimore, MD, 1983). late 1830s, his decision inves- to biological evolution and so- istic aspects of Spencers philosophy. leading erations,to 16. R. J. Richards, Stud. Hist. Philos. Sci. 28, 75 tigate the work of the animal Modern science recognizes the importance cial progress. Spencers self-improvement mod(1997). breeders (Fig. 3) his recogel and of progress became immensely popular in the of Darwins key insights when used as a way 17. M. Ruse, J. Hist. Ideas 36, 339 (1975). nition that their later method of artifi19th century, and because it too seemed to of explaining countless otherwise mysterious 18. J. Gayon, Darwinisms Struggle for Survival: cial selection offered a useful wayas the motor of change, it was aspects of the natural world. But some of those Heredity and the Hypothesis of Natural Sestruggle rely on of understanding howconfused the equivlection (Cambridge Univ. Press, New York, often with the Darwinian mechanism. insights came from sources with profoundly dis1998). alent natural process operated. In fact, Spencer thought that all humans will turbing implications, and many historians now 19. P. J. Bowler, J. Hist. Ideas 37, 631 (1976). The exact role played by Darwins eventually acquire the faculties needed to inter- recognize that the theory, in turn, played into 20. A. Desmond, J. R. Moore, Darwin (Michael study of breeding in the formulaact harmoniously with one another. But his occa- the way those implications were developed by Joseph, London, 1991). tion of his theory is much debated sional use of highly individualistic language al- later generations. This is not a simple matter of 21. M. Francis, Herbert Spencer and the Invention by historians (1617), but there of Modern Life (Acumen, Stocksfield, UK, lowed him to be perceived as the apostle of free science being misused by social commentacan be little doubt of how impor2007). Much of what later became known tors, because Darwins theorizing would almost 22. C. Darwin, Transmutation Notebook B, from tant the analogy enterprise. between artificial was, in fact, Spencerian certainly have been different had he not drawn as social Darwinism Natural Selection portfolio p. 36 (Cambridge and natural selection became in his social Lamarckism expressed in the terminology inspiration from social, as well as scientific, inUniv. Library, Cambridge, 1838); P. H. Barlater thinking. In this case, Darwin rett, P. J. Gautrey, S. Herbert, D. Kohn, S. with of struggle popularized by Darwin. fluences. We may well feel uncomfortable was truly unique, because even Smith, transcribers and Eds., Charles DarThis point is important in the context of those aspects of his theory today, especially in Wallace did not take this step and wins Notebooks p. 180 (British Museum of by modern opponents of of their subsequent applications to human the charge raised light dissociated himself from the link Natural History, Cornell Univ. Press, Ithaca, Darwinism that the theory is responsible for the affairs. But if we accept sciences power to upwith artificial selection expressed NY, 1987); available at the Darwin Digital Library, http://darwinlibrary.amnh.org/. in Darwins laterappearance writings. of a whole range of unpleasant social set the traditional foundations of how we think on struggle. Darwin exploited the about the world, we should also accept its po- 23. G. Neumeister, Das Ganze der Taubenzucht policies Darwin turned to the based breeders (B. F. Voigt, Weimar, 1876). idea as of to the struggle in search of a clue how a for existence in a way that tential to interact with moral values. nearly was unique until paralleled by Wallace population could be changed years later. Their theory certainly fed into here at least was 20 a situation where that led toward various kinds of References and Notes the movements modifications were actually benot the only vehicle 1. L. Eiseley, Darwins Century: Evolution and social Darwinism, it was Fig. 3. Pigeons (23). ing produced on a human time but www.sciencemag.org SCIENCE VOL 323 9 JANUARY 2009

20

CREDIT: AMERICAN MUSEUM OF NATURAL HISTORY

CREDIT: AMERICAN MUSEUM OF NATURAL HISTORY

www.sciencemag.org

SCIENCE

VOL 323

9 JANUARY 2009

225

225

21

Speciation
REVIEW

Red Queen and the Court Jester: ies Diversity and the Role of Biotic Abiotic Factors Through Time

disentangle key aspects of clade histories. Clades are monophyletic, including all descendants of an ancestor, whereas taxa may be monophyletic or vestigations, provide an outline of how of these paraphyletic, excluding some descendants the and other Comparative studies correspond to the predictions ancestor. macroecological studies disentangle key aspects of clade histories. Clades of the Red Court Jester, and multilevel add rigor toQueen, analyses showing that sister clades REVIEW are monophyletic, including all descendants of an (Table 1), and outline some phymixed models may vary whereas in rate of taxa evolution, timing of increases ancestor, may be monophyletic or of spelogenetic studies of themorphospace macroevolution in species richness and occupation, paraphyletic, excluding some descendants of the cies diversity. of evolutionary novelties across and distributions ancestor. Comparative macroecological studies lineages and Here, I that will sister explore the add rigor to subclades. analyses showing clades The Global global, Pattern of Diversification largest-scale taxic investigations, provide may vary in rate of evolution, timing of increases Michael J. Benton Through an outline Time of how these and other studies correin species richness and morphospace occupation, A key about the of modern spond toquestion the predictions of theorigin Red Queen, Court and distributions of evolutionary novelties across Evolution may be dominated by biotic factors, as in the Red Queen model, or abiotic factors, todays 10 million species biodiversity is how Jester, and multilevel mixed models (Table 1), lineages and subclades. Here, I will exploreand the as in the Court Jester model, or a mixture of both. The two models appear to operate ofthe microbial arose from a single ultimate species outline some phylogenetic studies of macrolargest-scale global, taxic investigations, provide predominantly over different geographic and temporal scales: Competition, predation, and other Michael J. Benton life 3500 of million ago 1).correTwo evolution species diversity. an outline of howyears these and (Ma) other (Fig. studies biotic factors shape ecosystems locally and over short time spans, but extrinsic factors such as models for global diversification are termed the spond to the predictions of the Red Queen, Court disentangle key aspects of clade histories. Clades The climate and oceanographic and events as shape larger-scale patterns regionally and Global Pattern ofmodel Diversification (57) and the exEvolution may be dominated by tectonic biotic factors, in the Redare Queen model, or abiotic factors, of an saturation/equilibrium monophyletic, including all descendants Jester, and multilevel mixed models (Table 1), and globally, through millions years. Paleobiological studies that species Through model Time (811). The equilibrium model as in the and Court Jester thousands model, or and a mixture of of both. The two ancestor, models whereas appear tosuggest operate taxa may be monophyletic or pansion outline some phylogenetic studies of the macrodiversity is driven largely by abiotic factors such as climate, landscape, or food supply, and paraphyletic, excluding some descendants of the A key question about the origin of modern biopredominantly over different geographic and temporal scales: Competition, predation, and other at has prevailed, among marine paleobiologists evolution of species diversity. ancestor. Comparative macroecological studies comparative approaches into clade but dynamics. diversity is how today sand 10 million species arose biotic factorsphylogenetic shape ecosystems locallyoffer and new over insights short time spans, extrinsic factors such as least, for a long time, represents a classic add rigor to analyses showing that sister clades from a single ultimate ofitmicrobial life climate and oceanographic and tectonic events shape larger-scale patterns regionally and The Global Pattern of species Diversification viewpoint because implies priQueen may vary in rate of evolution, timing of increases Red 3500 million years ago (density (Ma) (Fig. 1). Two modportant not export organism-level processes to marily here are two ways of viewing evolution, globally, andare through thousands and millions of years. Paleobiological studies suggest that species Through Time into species richness and morphospace occupation, here two ways of viewing evoluscales, and it is likely that evolution operates in dependence) on biotic controls and distributions offood evolutionary els fordiversity. global diversification termed biothe regional or global scales, and it isnovelties likely that global through the largely spectacles of either the Red diversity is driven abiotic factors such as climate, landscape, or supply, and across of either way (3). tion, through the by spectacles a pluralistic A key question about the originare of modern lineages and Here,way I will explore the saturation/equilibrium model (57) and the exevolution operates insubclades. a pluralistic (3 ). studQueen or Queen the Court The Red Queen comparative phylogenetic approaches offer new insights into clade dynamics. There are two broad methodologies for There is are two versions of the species equilibrium the Red or Jester. the Court Jester. The diversity how today s 10 million arose largest-scale global, taxic investigations, provide pansion ( 811 The equilibrium model has There are two broad methodologies for studies model ( 1 ) stems from Darwin, who viewed evoenton taxic correand model, the time when the global diversity differing in ). Red Queen model (1) stems from Darwin, who ies of species from a model single ultimate species of microbial life an outline of howthrough these and time, other studies prevailed, among marine paleobiologists at least, of species diversity through time, taxic and phylution as primarily a balance of biotic pressures, spond to the predictions of the Red Queen, Court of bi- phylogenetic involves viewedhere evolution asways primarily a balance 4). The taxic approach ecosystem saturated. 3500 million yearsbecame ago (Ma) (Fig. 1).Sepkoskis Two modportant not to ( export organism-level processes to marine are two of viewing evolution, y be dominated by biotic factors, as in the Red Queen model, or abiotic factors, and multilevel mixed models 1), and for a long time, and represents a classic Red Queen logenetic (Jester, 4).global The taxic approach treatmost notably competition, and it was characterels for global diversification are termed the regional species, or scales, and itinvolves is (Table likely that coupled the spectacles of either Red otic notably competition, it treating genera, or families as indelogistic model (5) identified three urt Jester model, orpressures, a through mixture ofmost both. The two models appear the toand operate outline some or phylogenetic studies of the macro- viewpoint because it implies primarily biotic coning species, genera, families as independent ized by the Red Queen s statement to Alice in saturation/equilibrium model ( 5 7 ) and the exevolution operates in a pluralistic way ( 3 ). Queen or temporal the Court Jester. The Red Queen and by thescales: Red Queens statement and counting their occurrences in the Cambrian, most of the was characterized pendent equilibria, y over different geographic and Competition, predation, other entities evolution of species diversity. trols (density dependence) global diversity. entities and counting their occurrences Through the Looking-Glass that but it takes allevothe shape ecosystems locally and over short time spans, extrinsic factors as pansion model (8perhaps 11). Theaon equilibrium model has There areand twoother broad methodologies foragainst studies Paleozoic, model (1in ) stems from Darwin, who viewed that it such factors. The phylogenetic beginning in to Alice Through the Looking-Glass against time and third, oceanographic and tectonic events shape larger-scale patterns regionally and There are two versions of the present equilibrium time and other factors. The phylogenetic aprunning you can do, to keep in the same place. The Global Pattern oftime, Diversification prevailed, among marine paleobiologists at (Fig. least, of speciesuses diversity through taxic and phylution all as primarily a balance of biotic pressures, trees to takes the running you can do, to keep in approach cladograms or molecular the Pliocene and continuing to the through thousands and millions of years. Paleobiological studies suggest that species Through Time differing in the time when the proach uses or molecular trees to model, The Court Jester model (2 ) isit that evolution, for a long time, and represents a classic Red global Queen logenetic (4key ).cladograms The taxicof approach involves treatmost notably competition, and was characterplace. The Court Jester model (2supply, ) is and disentangle aspects clade histories. Clades same riven largely the by abiotic factors such as climate, landscape, or food A key question about the origin of modern bio- 2A). These three equilibrium levels correspond speciation, and extinction rarely happen except in viewpoint because it implies primarily biotic coning species, genera, or families as independent ized by the Red Queen s statement to Alice in phylogenetic that approaches offer new insights into clade dynamics.rarely and extinction are monophyletic, allmillion descendants of to three sets of phyla, the Cambrian, Paleozoic, evolution, speciation, diversity is including how todays 10 species arose response to unpredictable changes in the physical trols Modern, (density dependence) on global diversity. entities counting their occurrences against Throughexcept the Looking-Glass that takes all the an from a single ultimate species of microbial life and whereas taxa may be monophyletic happen in response to it unpredictable ancestor, that interacted and successively Fig. 1. and Operation of Red Queen environment, recalling the capricious behavior 3500 million years ago (Ma) (Fig. 1).A Two modre two ways of viewing evolution, portant not to export organism-level to There each are two versions of the time and other factors. The phylogenetic ap- replaced running you can do, to keep in the same place. processes changes in the physical environment, recalling some descendants other through the equilibrium Cambrianor paraphyletic, excluding (biotic causation) and Court Jester els for global diversification are termed the h the spectacles of Court either the Red regional or scales, and it is likely that uses of the licensed fool of Medieval times. Neither model, differing in the time when the global proach cladograms or molecular trees Genus to Ordovician The Jester model (2 ) global is that evolution, (abiotic causation) models at differbehavior of the licensed fool of intervals, the capricious of the ancestor. Comparative macroecological and Permo-Triassic saturation/equilibrium model (57) and the life ex-span or the Court Jester. Thewas Red Queen evolution operates in a and pluralistic way (3). model proposed asrarely exclusive, both speciation, and extinction happen except in ent geographic and temporal scales (3-6 My) was proposed as for studies rigormodel to analyses showing that sister Medieval times. Neither model pansion (811). The equilibrium model has reaching higher equilibrium levels after each ms from Darwin, who viewed evoThere are two broad methodologies studies add Darwin and Van Valen (1 ) allowed extrinsic Court response unpredictable changes infor the physical prevailed, among marine paleobiologists at least, marily a balance of biotic to pressures, species diversity through time, taxic clades and ( A).phyThe Red Queen may prevail at timing vary in rate of evolution, of long-term replacementJester event. The second model exclusive, and both of Darwin and Van Valen (1) may Fig. 1. Operation of Red Queen influences on evolution in their primarily biotic, a long time, and represents a classic Red Queen competition, environment, and it was characterlogenetic (capricious 4). The taxic approach involves treat- for recalling the behavior organismic and species level on short Species A allowed for extrinsic influences on evolution in and morphospace equilibrium level from increases in species richness ( 6 , 7 ) identifies a single (biotic causation) and Court Jester Queen views. viewpoint because it implies primarily biotic conRed Queens Red statement to Alice in species, genera, or families as independent life span of the licensed fooling of Medieval times. Neither time scales, whereas the Court Jester Genus biotic, Red Queen views. their primarily occupation, and distributions of evolutionary the early Paleozoic, perhaps 400 Ma, to the trols (density dependence) on global diversity. Looking-Glass that it takes all the entities and counting their occurrences against (abiotic causation) models at differ(1-2 My) Species in as a Red Queen world de- holds his own on larger scales. The life span model wasdiversity proposed and both There aretemporal two versions of the equilibrium can do, to keep in the same place. time other factors. The phylogenetic ap- across Species diversity in and a exclusive, Red Queen world novelties lineages andscales subclades. Here, present (Fig. 2B). In both models, the equilibria ent geographic and (3-6 My) pends primarily on intrinsic factors, such as body Red stippled green shape shows an time areawhen the global Darwin Van Valen (1)uses allowed for extrinsic model, differing inprevail the ester model ( 2) is that and evolution, proach cladograms or molecular trees to Court Apparent Milankovitch intrinsic factors, such as I the largest-scale global, to biodiversity saturation in which depends primarily on will explore correspondQueen? (A ). The Red Queen may at taxic insize, breadth of physiological tolerance, or adaptRed Queen effects might be d extinction rarely happen except in Jester Red Queen influences on evolution in their primarily biotic, where Scale (100 ky) organismic and species level on short Species tolerance, body in size, breadth of a physiological from scale effect ability to hard times. In Court Jester world, spenpredictable changes the physical erroneously, but these are Red Queen views. Fig. 1. Operation of Red Queen identified life span time whereas the Court Jester Fig. 1. Operation of Red Queen (biotic causation) or adaptability to hard times. In a Court Jester likely recalling the capricious behavior A scales, cies diversity depends on fluctuations in climate, the result of spatial averaging of Locality Average Biotic Entire (1-2 My) and world Court Jester Species diversity(biotic in a causation) Red Queen de- holds his own on larger scales. The ed fool of Medieval times. Neither and Court Jester (abiotic causation) at difGenus world, species diversity depends onmodels fluctuations species region/ models continent landscape, and food supply. In reality, of as course, responses to climate change (abiotic causation) at body differ- regional primarily on intrinsic factors, such life span proposed as pends exclusive, and both Red and stippled green shape shows an area range climate ferent geographic temporal scales ( A ). The Red landscape, food supply. Inand realin climate, entand geographic and temporal scales Apparent (3-6 My) Milankovitch both aspects might prevail in different ways at and other complex physical perturzone Van Valen (1) size, allowed for extrinsic Queen? breadth of physiological tolerance, or adaptCourt be Queen and species level where Red Queen effects might Red Queen (A). The Red Queen may prevail at Scale (100 ky) may prevail at organismic might prevail in difity, course, both aspects bations that may be in opposite what could perhaps be called the Jesterdievolution in different theirof primarily biotic, fromthe scale effect ability totimes, hard times. In a Court Jester organismic and speciesworld, level onspeshort Species on short time scales, whereas Court Jester holds identified erroneously, but these are Geographical scale iews. could ferent ways and at different times, the what and so cancel each other, life span multilevel mixed model. Traditionally, biologists time scales, whereas Court Jester rections, cies diversity depends on fluctuations in climate, his own on larger scales. The stippled green shape likely (1-2 theMy) result of spatial averaging of Locality Average Biotic Entire iversity in a Red Queen world deholds his own onmixed larger scales. The suggesting no controlling effect of the perhaps be called the multilevel model. have tended to think in a Red Queen, Darwinian, species region/ continent shows an area where Red Queen effects might be food supply. In reality, of course, regional responses to climate change ly on intrinsiclandscape, factors, such and as body Red stippled green shape shows an area environment on evolution. range climate Apparent B biologists have tended to think in Traditionally, Milankovitch intrinsic, biotic factors way, and geologists in a physical identied erroneously, but these are likely the reof physiological tolerance, or adaptboth aspects might prevail in different ways and at and other complexQueen? physical perturwhere Red Queen effects might be Physical-environmental Red Queen zone Scale (100 ky) disruptions biotic faca Red Queen, Darwinian, intrinsic, Court Jester, extrinsic, physical factors way. from scale d times. In a Court Jester world, speerroneously, but these are bations that may be in opposite di- effect different times, whatidentified could perhaps be called the Genus sult of spatial averaging of regional responses to elicit biotic responses along the Geographical scale depends on fluctuations inand climate, likely the result of spatial averaging of may tors way, geologists in a Court Jester, exclimate change and other complex physical perturlife span Much of the divergence between the Red Locality Average Biotic Entire rections, and so cancel each other, multilevel mixed model. Traditionally, biologists species and region/ continent (3-6 My) red line separating Red Queen d food supply. In reality, of course, regional responses to climate change bations that may be in opposite directions, and so Queen and Court Jester world views may depend suggesting no controlling effect physical factors way. trinsic, range ofclimate the have tended to think in a Red Queen, Darwinian, might prevail in different ways and at and other complex physical pertur- Court Jester outcomes (B). The usage zone cancel each other, suggesting no controlling effect Court on scale ( 2 ) (Fig. 1): Biotic interactions drive Much of the divergence between the Red physical environment on evolution. bations that maygeologists be in opposite s, what could intrinsic, perhaps be biotic called the factors way, and in dia is the microevolutionary Red Queen,scale B Species of the physical environment Jester on evolution. Physicalmuch the local-scale success or failure of in- here and so cancel each other, Physical-environmental Geographical disruptions world views may Queen and Court rections, Jester xed model. Traditionally, biologists Court of Jester, extrinsic, physical factors way. life span as opposed to the macroevolutionary Genus environmental disruptions may elicit biotic resuggesting no controlling effect of the o think in a Red Queen, Darwinian, dividuals, populations, and species (Red Queen), may elicit biotic responses along the (1-2 My) depend on of scale (2 ) (Fig. 1): Biotic interactions life span Much the divergence between Red Red Queen that posits constant exphysical environment onthe evolution. sponses along the red line separating Red Queen ic factors way, and geologists in a processes B line separating Red Queen and (3-6 My) red but perhaps these are overwhelmed by drive much of Jester the local-scale success or tinction Queen and Court world views may depend Physical-environmental disruptions risk, a view that has been largely and Court Jester extrinsic, physical factors way. Red outcomes (B). The usage here is Genus Milankovitch Court Jester outcomes ( B ). The usage substantial tectonic and climatic processes atdrive time may elicit biotic responses along the rejected populations, and species failure of (individuals, Court on between scale 2 ) (Fig. 1): Biotic interactions life span the divergence the Red (31). Illustration based on (2). the ky) microevolutionary Red Queen, as opposed to Queen Scale (100 5 (3-6 My) red line separating Red Queen and Species here is the microevolutionary Red Queen, scales above 10 but years (Court Jester). It is imJester ourt Jester world views may depend (Red Queen), perhaps these processes much of the local-scale success or failure of inthe macroevolutionary Red Queen that posits conCourt Jester outcomes ( B ). The usage life span as opposed to the macroevolutionary Court (Fig. 1): Biotic interactions drive by substantial tectonic and are overwhelmed stant extinction risk, a view that has More been largely Less dividuals, populations, and species (Red Queen), Species (1-2 My) Jester local-scale success or failure of in- here is the microevolutionary Red Queen, Red Queen 5 life span that posits constant exDepartment of these Earth Sciences, University of Bristol, Bristol as opposed to the macroevolutionary rejected ( 31 ). Illustration based on ( 2 ). at time scales above 10 climatic processes but perhaps processes are overwhelmed by pulations, and species (Red Queen), Environmental scale (1-2 My) a view that has been largely tinction risk, Red Queen that posits constant exBS8 1RJ, UK. tectonic E-mail: mike.benton@bristol.ac.uk Red Milankovitch not to export years (Court Jester). It is important substantial climatic processes at time rejected (31). Illustration based on (2). hese processes are overwhelmed by and tinction risk, a view that has been largely Queen Scale (100 ky) Red 5 Milankovitch ctonic and climatic processes at time organism-level to regional scales above 10processes years (Court Jester).or It global ison imrejected ( 31). Illustration based (2). Queen Scale (100 ky) 5

The Red Queen and the Court Jester: Species Diversity and the Role of Biotic The Abiotic Red Queen andThrough the Court Jester: and Factors Time Species Diversity and the Role of Biotic and Abiotic Factors Through Time

Speciation

SPECIAL SECTION
new taxa could become established only by on diversity, and other correction regimes SPECIAL SECTION plants, insects, or vertebrates, because marine ecosystem became saturated. these be groups seem to have Sepkoski s coupled logistic model (5) as toradiated produce data in driving others to extinction. Key evidence may so complex explosively, without diversity plateaus, identified threeand equilibria, in the Camorigination extinction rates which geologic and signals are is that both plants, insects, orbiologic vertebrates, because marine ecosystem became saturated. particularly the past 100 million brian, most density-dependent of the Paleozoic, and pernot obviously separated. appear tos have been these in groups seem to have radiated Sepkoski coupled logistic model (5) (5 years (My) (10). haps a third, beginning in the Pliocene Life on land today may be as much as 7), limiting rises in diversity and promoting explosively, without diversity plateaus, identified three equilibria, in CamResolution between the equilibriand continuing to the the present (Fig. particularly in as the past 100 million brian, most 2A). of the Paleozoic, and per- levels after extinction events. life inand the sea, so it 25 asand diverse rapid recovery um times models, between these These three equilibrium years (My) (10 ).might seem haps a third,correspond beginning theglobal Pliocene expansionist models, toin three sets of phyla, the Alternative models for diversimay be wrong to generalize from marine straightforward, but thebetween solution Cambrian, Paleozoic, and (Fig. Modern, that Resolution the and continuing to the present allowing global paleontological studies to all delife.equilibriPerhaps fication are expansionist, pendsand on adequate assessment of the these interacted and successively replaced um models, and between 2A). These threeto equilibrium levels similar patterns of and exspecies diversity rise, with damping, but land sea show quality of the fossil record. The longeach other through the Cambrianexpansionist models, might seem correspond to three sets limit of phyla, the without a predictable ( 8 11 ). Den(8, ponential increase in species numbers term saturation model for global diverOrdovician and Permo-Triassic interCm O S D C P Tr J K Pg Ng straightforward, but the solution deCambrian, Paleozoic, and Modern, that of origination and extincsity dependence 9 , 11), or perhaps they differ in their key sification (Fig. 2, blue curve) arises vals, reaching higher equilibrium levels 500 400 300 200 100 0 pends adequate assessment the interacted replaced from extensive attempts to correct data as aof after successively each replacement event. with the sea acting giant tion rates and does notlong-term preclude expansionist rules ( 13, 14on ), Time (Ma) quality of error the fossil Thediverlongeach other through the Cambriansets for sampling (6 , 7),record. whereThe second model (6dampened , 7) identifies a models because they may be by Gaussian petri dish, where species as theterm multiple-equilibria and expansingle equilibrium level from the food early saturation model for global diverOrdovician and Permo-Triassic interof limiting factors such as shortage and density-dependent, 2. Patterns animal genus diversification the past sity is equilibrial O of S marine D C P Tr J K through Pg Ng originally used raw data, Paleozoic, perhaps 400 Ma, to the Fig. Fig. 2. Cm Patterns of marine animal genus diversication through sion models sification (Fig. 2, blue curve) arises vals, reaching higher equilibrium levels 530 My, the Phanerozoic. The two lines compare current estimates from by or space, or active predation, as well asthe and thecorrection land witnessing 400Phanerozoic. 300 The200 100 0 without (5, 811),continuing although (damppresent (Fig. 2B). In both models, the empirical past500 530(uncorrected) My, the two(red lines compare current from extensive attempts to correct data after each long-term replacement event. the Sepkoski database line) and samplingclimate andequilibria other physical factors. Further, ened) exponential rise in diversity more recent analyses return a some- as ever correspond to biodiversity estimates from the empirical (uncorrected) database Time (Ma) standardized (corrected) analysis of the Paleobiology Sepkoski Database (blue line). sets for sampling error ( 6 , 7 ), whereThe second model ( 6 , 7 ) identifies a what dampened butecospace congruent signal saturation in whichlogistic new taxamodel could The the coupled sectors of are conquered it seems that (red line) and sampling-standardized (corrected) analysisreached of the new empirical curve (red line) suggests that global marine diversity as the multiple-equilibria and expansingle equilibrium level from the early when corrections are imposed. Corbecome established only by driving a for global diversifica may be partly an artifact of taxonomic ( 9, 14 ). Any model Paleobiology Database (blue line). The empirical curve (red line) possible plateau through the Paleozoic (450 to 250 Ma) and has risen, Fig. 2. Patterns of marine animal genus diversification through the past sion models originally used Paleozoic, perhaps 400 it Ma, to the out rection for encompass sampling is the clearly es- raw data, others extinction. Key evidence is apparently exponentially, ever since. The sampling-standardized curve suggests that global marine diversity reached a possible plateau scale (Fig. 2, red to curve); was worked tion must independent evi530 My, the Phanerozoic. The two lines compare current estimates from (6, 7), and future investigation that both and the extinction (blue line) suggests that global marine diversity reached near-modern sentialwithout correction (5, 811),of although present (Fig. 2B). In origination both models, through the Paleozoic (450 to 250 Ma) and(red has risen, levels does not complexity organat ordinal and dence for increasing the empirical (uncorrected) Sepkoski database line) apparently and samplingappropriate indepenratesfamilial appear to to have but been densitylevels some 400 ever Ma and there hassampling-standardized been only modest increase since (blue then. must determine more recent analyses return a someequilibria correspond biodiversity exponentially, since. The curve specific work convincingly at generic or isms, increases in the occupation (corrected) analysis ofD, the Paleobiology Database (blue line). dent proxies for preservation and hu- of novel dependent ( 57 ), limiting rises in standardized di- Cm, Cambrian; C, Carboniferous; Devonian; J, Jurassic; K, Cretaceous; what dampened but congruent signal saturation in which new taxa could The empirical line) suggests that line) global marinethat diversity reached near-modern curve (red suggests marineS, diversity man error; some current proxies (such and promoting rapid recovery levels (10, versity 11), there are problems with Ng, explosive evolution within parNeogene; O, Ordovician; P, Permian; global Pg, Paleogene; Silurian;reached Tr, ecospace, whenofcorrections are imposed. Corbecome established only by driving levels some 400 Ma and there has been only modest increase as number fossiliferous localities) after extinction events. a possible plateau through Triassic. Based on (6). the Paleozoic (450 to 250 Ma) and has risen, assumptions (11, 12), and and addition of novel clades key numerical ticular clades, Cm, Cambrian; C, Carboniferous; D, Devonian; J, Jurection dependent for sampling is clearly esothers to extinction. Keymodels evidence is apparently are themselves on diversity, Alternative for a global di- since then. exponentially, ever since. The sampling-standardized curve (9, 11, 15), the background assumption of global without the loss of precursors rassic; K, Cretaceous; Ng, Neogene; O, Ordovician; P,red Permian; Pg, correction sential ( 6 , 7 ), and future investigation that both origination extinction allowing and other regimes may be so complex partly an artifact of taxonomic scale (Fig. 2, versification and are expansionist, global (blue line) suggests that global marine diversity reached near-modern carrying capacity is doubtful (8 , 10 , 11 ). but all data of which have happened many times in Paleogene; S, Silurian; Tr, Triassic. Based on ( 6 ). as to produce in which geologic and biologic curve); it was worked out at ordinal and familial species diversity to rise, with damping, withrates appear to have been density- levels some 400 Ma and there has been only modest increase since then. must determine appropriate indepenhard to export the loMy. Further, it has proved past 500 separated. not obviously out a),predictable limit (8di11 ). Density depen- levels but does not work convincingly at generic signals are the dent proxies for preservation and hudependent ( 57 limiting rises in Cm, Cambrian; C, Carboniferous; D, Devonian;problems J, Jurassic; K, Cretaceous; Life on land today may be as much as 25 or specific levels (10extensive , 11), there are with dence of origination and extinction rates does not terrestricurve) arises from attempts to corgistic model to the much more speciose versity and promoting rapid recovery Ng, Neogene; O, Ordovician; P, Permian; Pg, Paleogene; S, Silurian; Tr, man error; some current proxies (such as diverse as life in the sea, so it may be numerical assumptions , 12), and back- times preclude one expansionist models because they may Large-Scale on Species Diversity data sets for sampling(11 error (6 , 7the ), whereas al realm, whether considers plants, insects, rectkey as Controls number of fossiliferous localities) after extinction events. by Triassic. Based on (6). assumption of a global carrying capacity is wrong to generalize from marine paleontologground be dampened limiting factors such as shortage because these groups seem to and expansion models Taxic paleobiological studies have provided a the multiple-equilibria or vertebrates, arelife. themselves on diversity, Alternative models for global diPerhaps dependent land and sea of food or space, or active predation, as well as by doubtful (8, 10, 11). Further, it has proved hard to ical studies to all without diversity pla( 5 , evidence about controls, mainly have radiated explosively, originally used raw data, without correction great deal of and other correction regimes increase may bein so complex partly an artifact of taxonomic scale (Fig. 2, redshow versification are expansionist, allowing similar patterns of exponential export the logistic model to the much more climate and other physical factors.global Further, it years Biotic factors, such teaus, particularly in the pastdamping, 100 million 811 ), although more recent analyses return aspecies abiotic, on species diversity. numbers (8data , 9, in 11 ), or perhaps they speciose terrestrial realm, whether oneand considers seems to that the coupled logistic model may be as to produce which geologic and biologic curve); it was worked out at ordinal familial species diversity rise, with but withdampened but congruent signal when or ecologi(My) 10). Table 1.limit as body size, colonizing ability somewhat signals are notdiet, obviously separated. levels but does not the work convincingly at generic out a (predictable (811). Density depenMacroevolutionary phenomena and their support for either Red Queen (biotic, intrinsic) or Court Jester (physical, extrinsic) models. Many Resolution between equilibrium models, Correction for samcorrections are ( imposed. specialization, appear to have little effect on could fit either worldview, and so are noted multilevel mixed. on land today may be as much as 25 or specific levels 10, 11), there are problems with cal Life dence of origination andthe extinction rates does not as expansionist models, investhe diversity of as modern organisms, although and between these and pling is clearly essential (6( ,11 7), and future times as diverse life in the sea, so it may be key numerical assumptions , 12 ), and the backpreclude expansionist models because they may Red Queen Court Jester Multilevel mixed solution ground life-history characmight seem by straightforward, the tigationassumption must determine appropriate indepenand r-selected wrong to generalize from marine paleontologof a global carrying capacity is abundance be dampened limiting factorsbut such as shortage Interspecific competition Waxing and waning of clades in association with and humanVicariance and dispersal in major phylogenetic splits (17) litter size, for preservation error; depends on adequate assessment of the quality (short dent proxies ical studies to gestation all life. period, Perhapslarge land and sea (8, 10 , 11 ). Further, it has proved hard to teristics of food or space, or active predation, as well as by doubtful tectonic and oceanographic events (2, 17) record. The long-term saturation as number of fossome current proxies (such and short interbirth intervals) sometimes correof the fossil show similar patterns of exponential increase in export the logistic model to the much more climate and other physical factors. Further, it Character displacement Mass extinctions and smaller extinction events Latitudinal diversity gradient (2224) (Fig. may 2, triggered blue withnumbers high species 16). model that for global diversification siliferous localities) themselves dependent species (8, richness 9, 11), ( or perhaps they speciose whether one considers late seems the coupled logistic model be by extrinsicterrestrial causes suchrealm, asare eruptions,
3000 600 400

Number genera 2000 of 3000 (empirical; red)

Number of genera 400 600 (corrected; blue)

2000

Number of genera (empirical; red)

1000

tion

T T T

Temporal scale Temporal scale

Table 1. Macroevolutionary phenomena Coordinated and their support for either the Red Queen (biotic, intrinsic) or Court Jester (physical, extrinsic) models. Many Evolutionary arms races (1) turnovers, originations, and extinctions Occupation of new ecospace (25) could fit either worldview, and so are noted as multilevel mixed. in response to physical perturbations termed
Red Queen
of ecological InterspecificConstancy competition guilds through time (25) tectonic and oceanographic events (2, 17) Incumbency advantage Lack of evidence for a global carrying capacity and Character displacement Mass extinctions and smaller extinction events (3, 24) equilibrium levels (8, 10) triggered extrinsic causes such as eruptions, Lack ofby cohesiveness of the great evolutionary climate change, faunas (12)anoxia, impact (10, 11) Species turnovers, richnessenergy relationship (18, 19) Evolutionary arms races (1) Coordinated originations, and extinctions coordinated stasis or turnover pulse hypothesis (2, 29, 30) Court Jester Nonconstant probability of extinction (3, 11)with Waxing and waning of clades in association

climate change, anoxia, impact (10, 11)

Number of genera (corrected; blue)

1000

200 0

200

Multilevel mixed
Subdivision of niches/specialization (10, 25phylogenetic ) Vicariance and dispersal in major splits (17) Declining global extinction rates through time (1, 5)

Temporal scale

Latitudinal diversity gradient (2224)

Onshore-offshore patterns and disturbance (3) Resource use: stenotopes more speciose Occupation of new are ecospace (25) than eurytopes (29, 30)

Temporal scale Temporal scale

Constancy of ecological guilds through time (25) Incumbency advantage (3, 24)

in response to physical perturbations termed Inverse relationship between global temperature and coordinated stasis or turnover pulse hypothesis biodiversity (21) (2, 29 , 30 ) Lack of clear correlation of species richness with Nonconstant extinction (3 body probability size or other of biotic factors (16 ), 11) Lack of evidence for a global carrying capacity and www.sciencemag.org SCIENCE VOL 323 equilibrium levels (8, 10) Lack of cohesiveness of the great evolutionary faunas (12) Species richnessenergy relationship (18, 19) Inverse relationship between global temperature and biodiversity (21) Lack of clear correlation of species richness with body size or other biotic factors (16)

Subdivision of niches/specialization (10, 25) Declining global extinction rates through time (1, 5) 729 Onshore-offshore patterns and disturbance (3) Resource use: stenotopes are more speciose than eurytopes (29, 30)

6 FEBRUARY 2009

10 years (Court Jester). It is im-

728

6 FEBRUARY 2009
Bristol

Temporal scale

Department of Earth Earth Sciences, University of Bristol, Bristol Sciences, University of Bristol, BS8 1RJ, UK. E-mail: mike.benton@bristol.ac.uk mail: mike.benton@bristol.ac.uk

VOL 323 Less

SCIENCE

Less www.sciencemag.org More

More

Environmental scale

Environmental scale

728

22

6 FEBRUARY 2009

VOL 323

6 FEBRUARY 2009

SCIENCE

www.sciencemag.org

VOL 323

SCIENCE

www.sciencemag.org

23
www.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 2009

729

their continuing diversification in the Late Jurassic and Cretaceous was mainly indistinguishable from a random walk. In particular, dinosaurs did not participate in the Cretaceous Terrestrial Revolution, some 130 to 100 Ma, when flowering plants, leaf-eating insects, social insects, squamates, and many other modern groups radiated substantially.

shifts should mainly occur low in a clades history: Clade shapes vary from bottom-heavy to topheavy, and diversification shifts may be concentrated low (dinosaurs and bats) or high (insects and ants) in a clade (26). In the future, the identification of diversification shifts across numerous taxa may provide evidence

245.9

237

232.5 228

203.6 199.6

Anisian

Lad

Crn

Norian
PTEROSAURIA

Rh. EJ

Court Jester worldviews. If the majority of diversification shifts are coordinated, and associated with particular climatic, tectonic, and geographic drivers, then the Court Jester model of macroevolution would prevail. This would link most increases in species diversity to particular largescale radiation events, such as the Cretaceous Terrestrial Revolution (26), or recoveries after mass extinctions. If, on the other hand, the majority of diversification shifts are unique to particular clades, and not coordinated temporally with others, then the Red Queen worldview might be considered. Comparing Sister Taxa A powerful element of the comparative phylogenetic approach to species diversity relationships through time Fig. 3. Phylogenetic and moris the opportunity to compare sister taxa. Sisters phospace occupation for Triassic archosaurs. arose from a single ancestor, and so their trajectories occupy the same amount of time, and ( A) Framework phylogeny for Triassic cruthey started with the same genotype and pherotarsans scaled tosubsequent the Triassic time scale. notype. Any similarities in their evolution probably reflect this phylogenetic signal of Numbers at top refer to millions of years a common origin, but differences reflect independent aspects of their separate histories. before the present; gray bars represent the Comparisons of sister taxa have allowed tests observed durations of of the resource-use hypothesis (29), thatmajor general- lineages; verists are less speciose and have longer species tical dashed lines denote two extinction durations than specialists. Specialists divide the physical environment small patches, each events, at theinto Carnian-Norian and Triassicoccupied by a species, and each probably more Jurassic boundaries; arrowheads indicate subject to environmental crises than their generalist relatives. Classic examples in support of latter the lineages that survived the event. Lad, resource-use hypothesis come from studies of NeLadinian; Carnian; Rh, Rhaetian; EJ, Early ogene mammalsCrn, (29). For example, two antelope subgroups, the tribes Alcelaphini and Aepycerotini, Jurassic. ) Empirical morphospace for Late diverged 6 to ( 8B Ma. The former is now highly speciose, witharchosaurs, some 7 living and 25 extinct speTriassic based on the rst two cies, and the latter is represented by two species, principal coordinates. Large only one, the impala Aepyceros, surviving. The circles, dinoslowly evolving Aepycerotini consists of few saurs; ovals, pterosaurs; squares, poposauspecies at any time, and each of those is long lived, whereas the speciosephytosaurs; Alcelaphini consists roids; hexagons, stars, aetosaurs; of many short-lived species. The ecological crosses, crocodylomorphs; habits of both clades differ: The impala has a smaller black broad, generalist diet, whereas the speciose aldots, rauisuchids; larger black dots, nondicelaphines show more dietary specialization. In nosaurian dinosauromorphs, wider studies of many clades of Neogene African Scleromochlus. and South American mammals (30), the resourceBased onwas (28 ). and some subsidiary use hypothesis supported, predictions confirmed: Specialists are more common than generalists, carnivores include more generalists than herbivores, and there are more specialists in habitats that underwent recent environmental change (tropical rain forests and deserts). The resource-use model then stresses the role of climate and tectonic movements in determining species diversity rather than biological controls such as competition and predation.

Dinosauromorphs
ORNITHISCHIA

Dinosauria

SAUROPODOMORPHA THEROPODA PHYTOSAURIA

Crurotarsi

AETOSAURIA CROCODYLOMORPHA

Rauisuchids
POPOSAUROIDEA ORNITHOSUCHIDAE

B
0.12 0.06
Principal coordinate 2
0

-0.06 -0.12 -0.18 -0.24 -0.3 Crurotarsan morphospace -0.36 -0.24 -0.16 -0.08 0 0.08

Dinosaur morphospace

Pterosaur morphospace

0.16

0.24

0.32

and upland habitats of the later Paleozoic when land animals first burrowed, climbed, and flew, through the introduction of herbivory, giant size, endothermy, and intelligence among vertebrates, and the great blossoming of flowering plants (with associated vast expansions in diversity of plant-eating and social insects and modern vertebrates) during the Cretaceous Terrestrial Revolution 100 Ma (26). The other mode of species increase globally or regionally is by niche subdivision, or increasing specialization. This is hard to document because of the number of other factors that vary between ecosystems through time. However, mean species number in communities (alpha diversity) has increased through time in both marine (15, 25) and terrestrial (10) systems, even though niche subdivision may be less important than occupation of new ecospace in increasing biodiversity. Further, morphological complexity may be quantified, and a comparative study of crustaceans shows, for example, that complexity has increased many times in parallel in separate lineages (27).

Principal coordinate 1

Geographic and tectonic history has generated patterns of species diversity through time. The slow dance of the continents as Pangaea broke up during the www.sciencemag.org past 200 My has affected SCIENCE VOL 323 modern distribution patterns. Unique terrestrial faunas and floras, notably those of Australia and South America, arose because those continents were islands for much of the past 100 My. Further, major geologic events such as the formation of the Isthmus of Panama have permitted the dispersal of terrestrial organisms and have split the distributions of marine organisms. A classic example of vicariance is the fundamental division of placental mammals into three clades, Edentata in South America, Afrotheria in Africa, and Boreoeutheria in the northern hemisphere, presumably triggered by the split of those continents 100 Ma (17). Other splits in species trees may relate to dispersal events, or there may be no geographic component at all. Species richness through time may correlate with energy. The species richnessenergy relationship (18) posits correlations with evapotranspiration, temperature, or productivity, and studies of terrestrial and marine ecosystems have shown that these factors may explain as much as 90% of current diversity, although relationships between species diversity and productivity change with spatial scale (19). Over long time spans, there are strong correlations between plankton morphology and diversity and water temperature: Cooling sea temperatures through the past 70 My, and consequent increasing ocean stratification, drove a major radiation of Foraminifera, associated with increasing body size (20). More widely, there is close tracking be-

Fig. 3. Phylogenetic relationships and morphospace occupation for Triassic archosaurs. (A) Framework phylogeny for Triassic crurotarsans scaled to the Triassic time scale. Numbers at top refer to millions of years before the present; gray bars represent the observed durations of major lineages; vertical dashed lines denote two extinction events, at the Carnian-Norian and Triassic-Jurassic boundaries; arrowheads indicate lineages that survived the latter event. Lad, Ladinian; Crn, Carnian; Rh, Rhaetian; EJ, Early Jurassic. (B) Empirical morphospace for Late Triassic archosaurs, based on the first two principal coordinates. Large circles, dinosaurs; ovals, pterosaurs; squares, poposauroids; hexagons, phytosaurs; stars, aetosaurs; crosses, crocodylomorphs; smaller black dots, rauisuchids; larger black dots, nondinosaurian dinosauromorphs, Scleromochlus. Based on (28).

tween temperature and biodiversity on the globOutlook al scale for both marine and terrestrial organisms Paleontologists and evolutionary ecologists have debated species diversity and largely familial independently. richness were (21), where generic relatively low during warm greenhouse phas6 FEBRUARY 2009 731 es of Earth history, coinciding with relatively high origination and extinction rates. A much-studied manifestation of energy and temperature gradients is the latitudinal diversity gradient (LDG), namely the greater diversity of life in the tropics than in temperate or polar regions, both on land and in the sea. There are two explanations (22): (i) the time and area hypothesis, that the tropical belt is older and larger than temperate and polar zones, and so tropical clades have had longer to speciate, or (ii) the diversification rate hypothesis, that there are higher rates of speciation and lower rates of extinction in the tropics than elsewhere. There is geological and paleontological evidence for a mixture of both hypotheses (23, 24). Species diversity may increase by the occupation of new ecospace. The number of occupied guilds, that is, broad ecological groupings of organisms with shared habits, has increased in several steps through time, from 20 in the early Paleozoic to 62 in post-Paleozoic marine faunas (25). Further, marine animals have shown several step increases in tiering, the ability to occupy and exploit different levels in the habitat: At times, burrowers have burrowed deeper, and reef-builders have built taller and more complex reefs. Analogous, if even more dramatic, expansions of ecospace have occurred on land, with numerous stepwise additions of new habitats, from the water-margin plants and arthropods of the early Paleozoic to the forests

Phylogenetic Studies of Clade Histories Species are not randomly distributed; they have an evolutionary history, and so occur as twigs on a great phylogenetic tree. Studying species as members of clades is a fruitful approach to understanding the drivers and controls on speciation. Key questions include (i) Do species diversify early in a clades history? (ii) How do diversity and disparity (variance in characters or morphology) covary? (iii) Do major lineages within a clade follow similar, or different, patterns, and if different, why? (iv) Do evolutionary radiations follow the acquisition of new characters or emptying of ecospace? (v) How do major clades of apparent competitors interact over long spans of geologic time? and (vi) How do sister clades vary in species diversity and why? For such analyses, the ideal is a complete species tree, a phylogenetic tree that contains all species living and extinct, plotted accurately against geologic time (4). Simple to say; hard to achieve. More commonly, incomplete trees have been used, with the risk of error in calculations of evolutionary rates or comparisons of subclades. In paleontology, it has proven much easier to work with higher taxa such as genera or families because species fossil records are less complete than those of higher taxa, and yet it is not clear how higher-level patterns relate to those at species level. Many key questions can be tackled by comparing a real tree to a hypothetical tree that follows an equal-rate Markov (ERM) model, equivalent to tree growth after a random walk, where equal chances of speciation and of extinction are shared by all species (4). Major biotic replacements, where one clade replaces another, have been a focus of debate about the roles of competition and progress in

macroevolution, and dinosaurs provide a classic example. The standard view was that dinosaurs originated in the Late Triassic, some 230 Ma, by a process of competition in which they prevailed over their precursors, the crocodile-like crurotarsans and others, because of superior adaptations. A comparative phylogenetic study (28) shows, however (Fig. 3), that the Dinosauria expanded in two steps, one after an extinction event 225 Ma that removed dominant herbivores, and the second following the end-Triassic extinction 200 Ma that removed most of the crurotarsans. Dinosauria remained at moderate diversity and low disparity, and at lower disparity than the crurotarsans they supposedly out competed, during the 25 My between the events, suggesting that there was no insistent competition driving other groups to extinction but rather that the dinosaurs occupied new ecospace opportunistically, after it had been vacated. A further study on Dinosauria explored the subsequent evolution of the clade (26). Classic views that the dinosaurs arose with a flourish, and then finally gave way in the Cretaceous to the superior mammals, or that they dwindled to extinction because of racial senility, had long been abandoned. The dinosaurs seemed to be radiating actively in the Cretaceous, with many new clades appearing through their last 55 My, and especially in their final 15 My. The new study (26) shows that most diversification shifts (departures from ERM assumptions) fall in the first one-third of the history of the clade and that their continuing diversification in the Late Jurassic and Cretaceous was mainly indistinguishable from a random walk. In particular, dinosaurs did not participate in the Cretaceous Terrestrial Revolution, some 130 to 100 Ma, when flowering plants, leaf-eating insects, social insects, squamates, and many other modern groups radiated substantially. There is no geometric reason that diversification shifts should mainly occur low in a clades history: Clade shapes vary from bottom-heavy to top-heavy, and diversification shifts may be concentrated low (dinosaurs and bats) or high (insects and ants) in a clade (26). In the future, the identification of diversification shifts across numerous taxa may provide evidence for the relative importance of the Red Queen and Court Jester worldviews. If the majority of diversification shifts are coordinated, and associated with particular climatic, tectonic, and geographic drivers, then the Court Jester model of macroevolution would prevail. This would link most increases in species diversity to particular large-scale radiation events, such as the Cretaceous Terrestrial Revolution (26), or recoveries after mass extinctions. If, on the other hand, the majority of diversification shifts are unique to particular clades, and not coordinated temporally with others, then the Red Queen worldview might be considered.

Comparing Sister Taxa A powerful element of the comparative phylogenetic approach to species diversity through time is the opportunity to compare sister taxa. Sisters arose from a single ancestor, and so their trajectories occupy the same amount of time, and they started with the same genotype and phenotype. Any similarities in their subsequent evolution probably reflect this phylogenetic signal of a common origin, but differences reflect independent aspects of their separate histories. Comparisons of sister taxa have allowed tests of the resource-use hypothesis (29), that generalists are less speciose and have longer species durations than specialists. Specialists divide the physical environment into small patches, each occupied by a species, and each probably more subject to environmental crises than their generalist relatives. Classic examples in support of the resource-use hypothesis come from studies of Neogene mammals (29). For example, two antelope subgroups, the tribes Alcelaphini and Aepycerotini, diverged 6 to 8 Ma. The former is now highly speciose, with some 7 living and 25 extinct species, and the latter is represented by two species, only one, the impala Aepyceros, surviving. The slowly evolving Aepycerotini consists of few species at any time, and each of those is long lived, whereas the speciose Alcelaphini consists of many short-lived species. The ecological habits of both clades differ: The impala has a broad, generalist diet, whereas the speciose alcelaphines show more dietary specialization. In wider studies of many clades of Neogene African and South American mammals (30), the resource-use hypothesis was supported, and some subsidiary predictions confirmed: Specialists are more common than generalists, carnivores include more generalists than herbivores, and there are more specialists in habitats that underwent recent environmental change (tropical rain forests and deserts). The resourceuse model then stresses the role of climate and tectonic movements in determining species diversity rather than biological controls such as competition and predation. Outlook Paleontologists and evolutionary ecologists have debated species diversity largely independently. The realization that the Red Queen and Court Jester models may be scale-dependent, and that evolution may be pluralistic (3), opens opportunities for dialog. Taxic studies in paleontology continue to have great value in highlighting correlations between species richness and other factors, but comparative phylogenetic methods will illuminate questions about clade dynamics, species richness, and the origin of novelties. Further, methods are shared by paleontologists and neontologists, and this allows direct communication on the patterns and processes of

macroevolution. Viewed close up, evolution is all about biotic interactions in ecosystems (Red Queen model), but from further away, the large patterns of biodiversity are driven by the physical environment (Court Jester model). References and Notes

1. L. M. Van Valen, Evol. Theory 1, 1 (1973). 2. A. D. Barnosky, J. Vert. Paleont. 21, 172 (2001). 3. D. Jablonski, Evolution 62, 715 (2008). 4. A. Purvis, Annu. Rev. Ecol. Syst. 39, 301 (2008). 5. J. J. Sepkoski Jr., Paleobiology 10, 246 (1984). 6. J. Alroy et al., Science 321, 97 (2008). 7. J. Alroy, Proc. Natl. Acad. Sci. U.S.A. 105, 11536 (2008). 8. T. D. Walker, J. W. Valentine, Am. Nat. 124, 887 (1984). 9. M. J. Benton, Science 268, 52 (1995). 10. M. J. Benton, B. C. Emerson, Palaeontology 50, 23 (2007). 11. S. M. Stanley, Paleobiology 33, S1 (2007). 12. J. Alroy, Evol. Ecol. Res. 6, 1 (2004). 13. G. J. Eble, Geobios 32, 223 (1999). 14. M. J. Benton, Geol. J. 36, 211 (2001). 15. R. K. Bambach, A. H. Knoll, J. J. Sepkoski Jr., Proc. Natl. Acad. Sci. U.S.A. 99, 6854 (2002). 16. N. J. B. Isaac, K. E. Jones, J. L. Gittleman, A. Purvis, Am. Nat. 165, 600 (2005). 17. D. E. Wildman et al., Proc. Natl. Acad. Sci. U.S.A. 104, 14395 (2007). 18. G. Hunt, T. M. Cronin, K. Roy, Ecol. Lett. 8, 739 (2005). 19. G. G. Mittelbach et al., Ecology 82, 2381 (2001). 20. D. N. Schmidt, H. R. Thierstein, J. Bollmann, R. Schiebel, Science 303, 207 (2004). 21. P. J. Mayhew, G. B. Jenkins, T. G. Benton, Proc. R. Soc. London B. Biol. Sci. 275, 47 (2008). 22. G. G. Mittelbach, Ecol. Lett. 10, 315 (2007). 23. D. Jablonski, K. Roy, J. W. Valentine, Science 314, 102 (2006). 24. J. W. Valentine, D. Jablonski, A. Z. Krug, K. Roy, Paleobiology 34, 169 (2008). 25. R. K. Bambach, A. M. Bush, D. H. Erwin, Palaeontology 50, 1 (2007). 26. G. T. Lloyd et al., Proc. R. Soc. London B. Biol. Sci. 275, 2483 (2008). 27. S. J. Adamowicz, A. Purvis, M. A. Wills, Proc. Natl. Acad. Sci. U.S.A. 105, 4786 (2008). 28. S. L. Brusatte, M. J. Benton, M. Ruta, G. T. Lloyd, Science 321, 1485 (2008). 29. E. S. Vrba, Am. J. Sci. 293A, 418 (1993). 30. A. M. Bofarull et al., BMC Evol. Biol. 8, 97 (2008). 31. S. Finnegan, J. L. Payne, S. C. Wang, Paleobiology 34, 318 (2008). 32. Thanks to S. J. Braddy, P. C. J. Donoghue, D. Jablonski, A. Purvis, M. Ruta, D. N. Schmidt, and anonymous referees for comments and to S. Powell for drafting the figures. Supported by the Natural Environment Research Council and the Royal Society.

24

25

theories of the process (8, 11). I leave out discussion of sympatric and allopatric speciation but SPECIAL SECTION Under this the Darwinian instead identify likelihoodperspective, of ecologicallinkand was relatively ing speciation with adaptation mutation-order speciation when there is gene 61. R. C. Albertson, J. T. Streelman, T. D. Kocheer, J. Hered. 52. O. Seehausen et al., Nature 455, 620 (2008). 41. D. Jablonski, Evolution 62, 715 (2008). straightforward, requiring only test of whether flow. I, ignore reinforcement, aaspecial type of 94 291 (2003). 53. R. Lande, Genetics 128, 443 (1991). 42. R. J. Whittaker, K. A. Triantis, R. J. Ladle, J. Biogeogr. 35, 62. K. Schwenk, N. Brede, B. Streit, Philos.species Trans. R. Soc. 54. D. A. Joyce et al., Nature 435, 90 (2005). 977 (2008). phenotypic differences between were natural selection thought to favor stronger preLondon Ser. B 363 , 2805 (2008). 55. G. Fryer, Environ. Biol. Fishes 45, 109 (1996). 43. R. Gillespie, Science 303, 356 (2004). selection. For example, at the caused by natural mating reproductive isolation once postzygotic 63. D. Garant, S. E. Forde, A. P. Hendry, Funct. Ecol. 21, 434 56. J. W. Valentine, D. Jablonski, in Evolution, Time and 44. J. Todd Streelman, P. D. Danley, Trends Ecol. Evol. 18, Advancement of American Association for the isolation has evolved. I also ignore speciation by (2007). Space, R. W. Sims, J. H. Price, P. E. S. Whalley, Eds. 126 (2003). Dolph Schluter Science 1939 [the last polyploidy, evenspeciation though selection crucial 64. O. Seehausen, Curr. Biol. 16symposium , R334 might (2006).be (Academic Press, London, 1983), pp. 201226. 45. D. Ackerly, D. Schwilk, C. Webb, Ecology 87, S50 (2006). 65. C.early Darwin, Voyage on of the Boogle (Collier, Newthe York, 57. C. O. Webb, D. D. Ackerly, M. A. McPeek, M. J. Donoghue, in 46. C. E. Parent, B. J. Crespi, Evolution 60, 2311 (2006). the stages. symposium speciation before biomajor Natural drives the origin as Darwin initially Mechanisms of 1909), p. 383.concept (7)], an extensive comRev. Ecol. Syst. 33claimed. , 475 (2002). 47. J. A. selection Coyne, T. D.commonly Price, Evolution 54, 2166 (2000). of species, Annu. logical species speciation by D. selection fall into mutation-order. Under 66. We thankand A. Birand, R. Glor, A.from Harrison, L. Harmon, 58. ecological J. J. Wiens, C.and H. Graham, Annu. Rev. Ecol. Syst. 36, 519 Speciation 48. J. B. Losos, Schluter, Nature 408,two 847 broad (2000). categories: Adaptation inparative and biogeographic study showcased D Jablonski, L. Mahler, C. Marshall, and the anonymous 49. D. Lack,speciation, Darwins Finches (Cambridge is Univ. Press, by divergent(2005). ecological divergence driven natural selection between environments, Darwin to Dobzhansky stances in which derived morphological and life reviewers for useful comments. Supported by the NSF 59. S. Skulason, T. B. Smith, Trends Ecol. Evol. 10, 366 Cambridge, 1947). whereas under mutation-order speciation, divergence occurs when different mutations arise and (grant DEB-0519777) and thearisen NIH between (grant GM56693). (1995). 50. U. Dieckman, M. Doebeli, J. A. J. Metz, D. Tautz, Adaptive Appreciation of connection adaptaofthe fishes had over and over history forms are fixed in separate adapting to similar60. selection pressures. of parallel evolution H. D. Bradshaw, D. W.Tests Schemske, Nature 426 , 176 Speciation (Cambridgepopulations Univ. Press, Cambridge, 2004). tion and speciation began with Darwin when a type ( 12 ). The reagain from the same ancestral of reproductive trait-based assortative and reproductive isolation by active 10.1126/science.1157966 (2003). 51. U. K. Schliewen,isolation, D. Tautz, S. Pbo, Nature 368 , 629 (1994). mating, morphological concept of nonparasitic species largely prepeated, parallel origin of lamprey selection have demonstrated that ecological speciation is a common means by which new species vailed. In On the Origin of Species , Darwin wrote in streams from the same migratory, parasitic arise. Evidence for mutation-order speciation by natural selection is more limited and has been that I look at thethat term species, asagain one arbitrarily Again and the paraancestor showed best documented by instances of reproductive isolation resulting from intragenomic conflict. given forof the sake ofevolved convenience to aout setdisof nonparasitic sitic lampreys have into theories the process ( 8 , 11 ). I leave However, we still have not identified all aspects of selection, and identifying the underlying genes REVIEW individuals closely resembling each other... and forms...correlated with in small streams, cussion of sympatric and life allopatric speciation but for reproductive isolation remains challenging. The amount of difference is one very important where a identify suitable food supply inof the way of large instead the likelihood ecological and criterion in settling whether two forms should be or seasonal (12 ). When correlated fish is scarce mutation-order speciation when there is gene ries: ecological speciation and mutation-order tttook evolutionary biologists nearly 150 years, ranked as species or varieties ( 1 ). Under this flow. environmental I ignore reinforcement, a special type of took evolutionary biologists nearly 150 would nevertheless be advantageous in both of with repetition is factors, such speciation is thought defined as accumulation of Ecological speciation refers to the but at last can we agree with Darwin the speciation. natural selection to the favor stronger precan agree with that Darwin of view, years, butwe at last their environments. The relative importance unlikely to result from chance; environmental origin of species, that mystery of mysteries evolution of reproductive isolation between pop- sufficiently many differences between populamating reproductive isolation oncebe postzygotic pressures must therefore the cause that the origin of species, that mystery of these two categories of mechanism for the ori- selection (1), really does occur by means of natural selection ulations or subsets of a single population by ad- tions to warrant their Iclassification as separate isolation has evolved. also ignore speciation by mysteries (1), really does occur by means of gin of species in nature is unknown. of speciation. As a result of our recent studies Schluter (Dolph 25). Not all species appear to evolve by aptation to different environments or ecological taxonomic species. Darwin understood the impolyploidy, even though selection might be crucial In this review, I summarize progress in un- on being added to fishes...weight is constantly natural selection (25). Not all species appear to selection, but the evidence suggests that most of niches (2, 8, 9). Natural selection is divergent, portance of reproductive barriers between species in the early stages. but the evidence suggests features of speciation evolve by selection, derstanding the general the theory that speciation is...under the rigid Natural commonly drives the origin of species, initially claimed. Mechanisms of (1), but the study of speciation after the pubactingas in Darwin contrasting directions between environthem do.selection The effort leading up to this conclusion up to by selection. speciation control of the environment (12). However, this that most of do. The leading Iand do mutation-order. not differentiate speciation bythem selection fall effort into broad categories: ecological Under which drives the fixation of different lication involved many experimental and two conceptual ad- ments, of this work focused mainly on the evoSpeciation Adaptation from of referring to the origin morphothis conclusion involved many is experimental by sexual selection here becauseenvironments, natural selec- case is onlyand ecological speciation, divergence selection between alleles,natural each advantageous in one environment lution vances, including a revision of the driven notionby ofdivergent of species differences, particularly of Darwin to Dobzhansky advances, including a revision drives the divergence of mate preferences, logical species. and conceptual tion whereas under speciation, occurs when different mutations in the other. Following G. S. arise Maniand and morphological speciation itself, mutation-order 80 years after publication ofdivergence On but not traits but also of between behavioral and Appreciation of the connection adapta80 years after or mutation-order mechaThe turning point for speciation studies of the notion of Species speciation itself, byselection either ecological are fixed in populations adapting toon similar pressures. Tests of parallel evolution B. C. Clarke (10), I define mutation-order specia- other the Origin of separate the , to a definition based phenotypic traits. tion and speciation began with Darwin when a (8, 11). I came concept of speciation publication of On the instead Origin of the Species , to mating, nisms, in most theories of the process with the Darwinian modern of reproductive isolation, trait-based assortative reproductive isolation byisolation active tion as and the evolution of reproductive by reproductive isolation of morphological Under this linking morphological concept of perspective, species largely of prethe a definition based on reproductive isolation in leave out discussion of sympatric and allopatric Species separation is defined as a stage selection have demonstrated that ecological speciation is a common means which of new species speciation with adaptation was relatively straightthe chance occurrence andby fixation different differences (6, 7). vailed. In On the Origin of Species , Darwin wrote differences ( 6 , 7 ). speciation but instead identify the likelihood of evolutionary process at which physiological stead of morphological arise. Evidence for mutation-order speciation by natural selection is more limited and has been The main question today is how does selec- alleles between populations adapting to similar forward, requiring a species, test of whether that I look at theonly term as one phenotyparbitrarily The main question todayare is of how does selec-isolation and mutation-order speciation when ic mechanisms become developed ( 6) ecological isolating best documented by instances resulting from intragenomic conflict. selection pressures. Reproductive isolation evolves tion lead to speciation? What thereproductive mechanisms differences between species were caused by given for the sake of convenience to a set of What are the mechareinforcement, a speto mean tion lead to speciation? there is gene flow. I ignore (here, physiological is interpreted However, we still have not identified all aspects of selection, and identifying the underlying genes of natural selection, what genes are affected, and because populations fix distinct mutations that natural selection. For example, at the American individuals closely resembling each other... and favor nisms naturalat selection, what genes are af- would cial type of naturalbe selection thought evolved reproductive isolation between populafor reproductive isolation remains challenging. nevertheless advantageous in to both of Association how doof changes these genes yield the habitat, forof the Advancement of Science 1939 The amount difference is one very important from geographical barriers to fected, and mechanical, how do changes at these genes yield their stronger premating reproductive isolation once tions, as distinct environments. The relative importance of speciation behavioral, chemical, physiological, symposium [the last major symposium criterion in settling whether two forms should be chemical, these isolation has evolved. for I also species were deinterbreeding). Subsequently, the other habitat, behavioral, mechanical, postzygotic categories of mechanism the ignore origin on and that are the reproducspeciation before the varieties biological species concept ecological speciation and mutation-order t took incompatibilities evolutionary biologists nearly 150 years, ries: two ranked as species or ( 1). Under this that even though selection ( physiological, and other incompatibilities speciation by polyploidy, fined asextensive groups of interbreeding natural popuof species in nature is unknown. tive barriers between new species? As a start, the 7 )], an comparative and biogeographic but at last we can agree with Darwin that the speciation. Ecological speciation refers to the view, speciation is defined as the accumulation of that are instances reproductively isolated from are the reproductive between new spemight be crucial in the early stages. lations In this review, I summarize progress in undermany ways by which barriers new might arise by showcased in which derived morof reproductive isolation between pop- study origin of species, thatspecies mystery of mysteries evolution sufficiently many differences between populathe many ways by which new standing the general features of speciation by se- phological this of point on,had the other such and groups (7). From cies? As can a start, selection be grouped into two broad categolife history forms fishes (1), really does occur by means of natural selection ulations or subsets of a single population by ad- tions to warrant their classification as separate Speciation and Adaptation from Darwin can be grouped the evoluspecies might arise by selection study of speciation was the study of lection. I do not differentiate speciation by sexual arisen over and over again from the same ancestral (25). Not all species appear to evolve by aptation to different environments or ecological taxonomic species. Darwin understood the imto Dobzhansky into two broad categories: ecological speciation tion of reproductive isolation (3). Progress up to here natural selection the type (12 ).of The repeated, parallel origin of nonniches (2 , 8, because 9). Natural selection is drives divergent, selection, but the evidence suggests most of selection portance reproductive barriers between species Biodiversity Research Centre and Zoologythat Department, divergence of of mate by either eco- parasitic lamprey in streams from the same migraand mutation-order speciation. Ecological Appreciation the preferences, connection between adapthen in understanding link between morphoUniversity British Columbia, Vancouver, V6T spe1Z4, acting in contrasting directions between environthem do.of The effort leading up to thisBC conclusion (1), but the study of the speciation after the pubor mutation-order most tory, parasitic ancestor showed that was on Again and Canada. E-mail: schluter@zoology.ubc.ca toexperimental the evolution ofconceptual reproductive with Darwin when and adaptation largely ciation refers tation speciation logical speciation ments,and which drives began themechanisms, fixation of in different involved many and ad- logical lication of this work focused mainly the evoor subsets of a under isolation between populations a morphological concept of species largely forgotten, its contributions uncertain vances, including a revision of the notion of alleles, each advantageous in one environment lution of species differences, particularly the of On the Following Origin of Species , Darwin single population by adaptation to different concept. traits but also of behavioral and but not in In the other. S. Mani and new speciation itself, 80 years after publication of enOn prevailed. morphological www.sciencemag.org SCIENCE VOL G. 323 6 FEBRUARY 2009 737 look the mutation-order term species, as one other The biologicaltraits. species concept must surely vironments or ecological niches (2, 8, 9 ). Natuthat I(10 B. C. Clarke ), I at define speciathe Origin of the Species, to a definition based on wrote phenotypic divergent, acting contrasting arbitrarily difficultperspective, to investigate any given forof the sake of convenience made this it more ral selection is tion as the evolution reproductive isolation by have reproductive isolation instead of in morphological Under Darwinian linking selection. T. directions set of individuals resembling each link between speciation and natural thea chance occurrence closely and fixation of different differencesbetween (6, 7). environments, which drives to speciation with adaptation was relatively straightand The amount of difference is one Dobzhansky the The fixation of different alleles, each advanta - other... (13) only suggested that the genes unalleles between populations adapting to similar main question today is how does selecforward, requiring a test of whether phenotypin to one environment but not the other. very criterion in settling whether two derlying differences between populations in orimportant geous selection pressures. Reproductive isolation evolves tion lead speciation? What are thein mechanisms ic differences between species were caused by and B. C. Clarke (10and ), I forms species or varieties were unlikely to be the Following G. S. Mani should be ranked traits because populations fix as distinct mutations that dinary of natural selection, what genes are affected, natural phenotypic selection. For example, at the American the habitat, evolu- ( as the changed define mutation-order speciation as the 1). Under this view,be speciation is defined of reproductive isolation. He would nevertheless advantageous in both of basis how do changes at these genes yield Association for the Advancement oflater Science 1939 tion of reproductive isolation by the chance accumulation of sufficiently many differences mind, but at the time this viewpoint, and the their environments. The relative importance of his behavioral, mechanical, chemical, physiological, speciation symposium [the last major symposium and fixation of that different alleles be- between populations to mechanism warrant their - generally greater difficulty of studying reprooccurrence these two categories of forclassifica the origin and other incompatibilities are the reproducon speciation before the biological species concept of species in nature is unknown. tive barriers betweenadapting new species? As a start, the tion (7)], an extensive comparative and biogeographic to similar selecspecies. Darwin un- ductive morphology, must have tween populations as separate taxonomic isolation than In this review, I summarize progress in under- discouraged many ways by which new species might evolves arise by derstood study showcased instances in which derived mortion pressures. Reproductive isolation the importance of reproductive barriers many from pursuing the connecstanding species the general of speciation by se- tion. selectionpopulations can be grouped into two broad categophological and no life history effort forms followed of fishes that had because fix distinct mutations that between (1), features but the study of speciation Virtually research lection. do not differentiate speciation bymainly sexual tested arisen the over andof over again from the same ancestral of this work focused role adaptation in speciation. after theIpublication selection here because natural selection drives the type (12). The repeated, parallel origin of nonBiodiversity Research Centre and Zoology Department, on the evolution of species differences, particudivergence of mate preferences, by either eco- Models parasitic lamprey in streams from the same migraof Speciation by Selection of morphological traits but also of behavUniversity of British Columbia, Vancouver, BC V6T 1Z4, larly logical or mutation-order mechanisms, in most tory, parasitic ancestor showed Again and Canada. E-mail: schluter@zoology.ubc.ca The topic of natural selection that in speciation is ioral and other phenotypic traits.
REVIEW

Speciation

Evidence for Ecological Speciation and Its Alternative

I I

Evidence for Ecological Speciation and Its Alternative

way becau parasitic lampreys have evolved ecologiinto non- A Gambusia Both models of speciation, once again receiving attention. The two most again the ulations d forms...correlated with life in small general hypotheses involving selection are parasitic cal and mutation-order, are theoretically mutations where a suitable food supply in the way ecological and mutation-order speciation. Eco- streams, plausible, and only data can determine order. Div of large fish is scarce or seasonal (12). When corimportance in nature. The logical speciation is defined as the evolution related their relative chastic bu with environmental factors, such repetition of reproductive isolation between populations is unlikely key is to out by which mechanism from gene to figure result from chance; environmental reproductive isolation first evolved (3 ). by divergent natural selection arising from dif- selection both smal pressures must therefore be the cause of infinite) p As aearliest result ofgenetic our recent studies on differences Once the ferences between ecological environments (2, speciation. can be ec is constantly being added to the 8, 9, 14). It predicts that reproductive isolation fishes...weight have accumulated between populations mutationspeciation is...under the rigid control of either process, subsequent mutations should evolve between populations adapting to theory by that ecology d the environment (12). However, this case is only contrasting environments but not between pop- referring might beorigin favored in one population gence as s to the of morphological species. and turning not the other because studies of epistatic ulations adapting to similar environments. The The evolution point for speciation came with genetic background basic idea has been around for a while (7), al- with interactions ductive iso the modern concept of speciation Species tion of eco is defined as a stage of the evoluthough it was tested only recently. The agents of separation (10). Hence, epistasis, including that postzygot process Dobzhansky-Muller at which physiologicalincomisolat- B Mimulus producing divergent selection are extrinsic and can include tionary Specia ing mechanisms become developed ( 6 ) (here, abiotic and biotic factors such as food resources, patibilities in hybrids between species tragenomi physiological is interpreted to mean evolved climate, habitat, and interspecies interactions reproductive (3), can result from either ecological or otic drive isolation between populations, as speciation. such as disease, competition, and behavioral in- distinct mutation-order sterility (F from geographical barriers to interbreedSpeciation can be rapid both terference. Ecological speciation can lead to the ing). Subsequently, mutationspecies were under defined as cause, by c of interbreeding populations alleles that are evolution of any type of reproductive isolation, groups speciation models,natural because tions cau reproductively isolated from other such to fixation by natural selection including premating isolation, hybrid sterility, are driven countering groups ( 7 ). From this point on, the study of and intrinsic hybrid inviability as well as extrin- in both cases. However, under the muthe same i speciation was the study of the evolution of sic, ecologically based pre- and postzygotic iso- reproductive tation-order process, the same alleles, Speciation isolation (3). Progress up to then in if present,the would be favored in every lation. Speciation by sexual selection is ecologi- understanding mutation-o link between morphological the early forgotten, stages of population, at least in cal speciation if ecologically based divergent se- speciation vergence and adaptation was largely gamete re uncertain under the new concept. lection drives divergence of mating preferences, its contributions divergence. For this reason, mutationFig. 1. (A) Example of ecological speciation. Repeatedly and fixation o The biological species concept when must surely 1. ( A ) Example of ecological speciation. Repeatedly and speciation is difficult there Fig. for example by sensory drive (15). order independently, the mosquito fish, Gambusia hubbsi, inhabiting more because difficult to investigate any link- independently, mosquito sh, Gambusia hubbsi , inhabitIn accordance with (10), mutation-order spe- have flow increas ismade geneitflow, gene blue holes in the the Bahamas has evolved a larger caudal region and geous mu between speciation and natural selection. T. ing blue holes the Bahamas evolved a larger caudal head in in the presence of has predators (top) than in their ulations, ciation is defined as the evolution of reproduc- Dobzhansky es the possibility that favorable muta- smaller (16). Div (13) suggested that the genes under- region and smaller head in the presence ofprobability predatorsof (top) absence (bottom) ( 29 ). In laboratory trials, the in one population will than in their absence (bottom) (29). In laboratory trials, two tive isolation by the fixation of different advan- lying tions occurring differences between populations in ordinary the other learn individuals mating was higher when they were from different tageous mutations in separate populations ex- phenotypic spread traits to other werepopulations, unlikely to be preventing the basis of probability of twothe individuals mating was higher (and when they havior und populations having same predation environment similar isolation. He). later changed his mind, were divergence (17, 18 Any process resultperiencing similar selection pressures. Whereas reproductive from different populations the same preda- tion, not m body shape) than when they were having from opposite predation at the viewpoint, theincluding generally tion environment (and similar shape)(29 than when they efficient s environments. [Photo credit: Brianbody Langerhans )]. (B ) Example flow, different alleles are favored between populations but ing in time low this levels of geneand from opposite predation environments. [Photo credit: is the cul difficulty of studying reproductivediverisola- were of reproductive isolation evolving under the mutation-order under ecological speciation, the same alleles greater selection, facilitates subsequent Langerhans (29)]. (left) (B) Example of reproductive isolation tion than morphology, must have discouraged Brian mechanism. Male-fertile and male-sterile (right) flowers of mutation-o would be favored in different populations under gence by the mutation-order process evolving under thean mutation-order mechanism. Male-fertile between Oregon population of monkey flowers (M. al scenario many from pursuing the connection. Virtually no F2 hybrids speciation mutation-order speciation. Divergence occurs research (19).effort In contrast, and male-sterile (right)male owers of F2 hybrids between guttatus ) having a cytoplasmic sterility element and nuclear Both followed ecological that tested the role of (left) anyway because, by chance, the populations do adaptation can proceed with or without gene flow, an restorer and population a closely related species owers (M. nasutus having neither Oregon of monkey (M.) guttatus ) hav- ecologica in speciation. (46,a47 ). Both flowers shown have M. guttatus cytoplasm. The are theore cytoplasmic male sterility element and nuclear restorer not acquire the same mutations or fix them in the although it is easiest when gene flow ing flower on the left also has the nuclear restorer, whereas the one on, only data of Speciation by Selection and a closely related species (M. nasutus ) having neither (46 same order. Divergence is therefore stochastic Models is absent. the). right, undeveloped anthers, lacks the cytoplasm. restorer. [Photo 47 Both with owers shown have M. guttatus The ative imp The topic of natural selection in speciation is once Experiments with laboratory popu- credit: Andrea but the process is distinct from genetic drift. It 47)] has the nuclear restorer, whereas the key is to on theCase left (also again receiving attention. The two most general ower can occur in both small and large (though not lations of Drosophila and yeast dem- one on the right, with undeveloped anthers, lacks the restormechanism hypotheses involving selection are ecological and infinite) populations. Selection can be ecologi- mutation-order onstrate the plausibility of ecological [Photoincluding credit: Andrea Case isolation, (47)] isolation, premating hybrid first evolved (3). Once the speciation. Ecological speciation er. when cally based under mutation-order speciation, but is defined speciation. In thoseofinstances as the evolution reproductive iso- sterility, and intrinsic hybrid inviability as well as ences have accumulated b between populations by postmating divergent natural ecology does not favor divergence as such. It lation measurable pre- and re- extrinsic, ecologically based pre- and postzygotic either process, subsequen isolation. by sexual is favored one populatio arising isolation from differences between ecoOdsh, JYAlpha ), and selection sexual isolation (ds2)in beevolved, it was greater be- (Speciation can lead to the evolution of any type of repro- selection productive speciation if ecologically basedMost diver- of because of epistatic inte environments (2, 8, 9, to 14). It predicts that ecological tween environments Drosophila species. these genes ductive isolation, with the exception of ecologi- logical tween lines subjected different reproductive isolation should evolve between gent selection drives divergence of mating background (10). Hence, e than between lines raised under homogeneous show molecular signatures of positive selection, cally based pre- and postzygotic isolation. producing Dobzhansky-M populations adapting to contrasting environments preferences, for example by sensory drive (15). Speciation resulting from intragenomic con- but conditions 20, 21). adapting Laboratory experiments that proving natural role (3), provided In accordance with (10), selections mutation-order spein hybrids between specie not between ( populations to similar under flict such as meiotic drive or cytoplasmic male environments. on various microbes maintained fixation occurred before complete reproductive ciation is defined as the evolution of reproductive either ecological or mutati The basic idea has been around for homoisolation rather afterward. sterility (Fig. 1B) is likely to be mutation-order a while geneous conditions for many generations have by the fixation of than different advanta- The top-down Speciation can be rapid (7), although it was tested only recently. isolation geous in separate populations expebecause alleles ar agents of divergent selection are extrinsic and with genetic divergence consistent the mutations identifying (i) themodels, phenotypic speciation because, by chance, the initial muta- Thedetected approach involves riencing selection pressures. Whereas(ii) natural include abiotic andprocess biotic factors such food on (22), butas effects re- similar those selection traits in both traits under divergent selection, tions causing drive and those countering it are can mutation-order resources, climate, habitat, and interspecies inter- different alleles are favored between populations the mutation-order proces unlikely to be the same in separate populations. productive isolation have not been explored. associated with reproductive isolation, and (iii) actions such as disease, competition, and behav- under ecological speciation, the same alleles present, would be favored Two approaches investigate Speciation by sexual selection is mutation-order ioral interference. the genes underlying traits and reproductive isoin different populations under least in the early stages o Ecological speciation the can mechanisms lead would be favored in nature. The lation. Step (iii) has been challenging undermutation-order both speciation if divergence of mate preferences or to the of speciation byany natural mutation-order speciation. Divergence occurs any- reason, spe evolution of type selection of reproductive gamete recognition occurs by the fixation of bottom-up approach involves (i) genetic map- approaches but is needed to understand how sealternative advantageous mutations in different ping of reproductive isolation between closely lection has led to reproductive isolation. 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 738 populations, as by sexual conflict (16). Diver- related species, (ii) testing whether discovered gence in song and other learned components genes exhibit a genomic signature of positive Ecological Speciation of behavior under purely social selection, not selection, and (iii) identifying the phenotype Evidence for ecological speciation has accumumolded by selection for efficient signal trans- and source of fitness effects of alternative alleles lated from top-down studies of adaptation and mission (5), is the cultural equivalent of the at selected loci. The approach has been hugely reproductive isolation [reviewed in (2, 8, 9)]. mutation-order process. Additional scenarios successful in identifying major genes implicated We now know of many real species that have, in hybrid inviability (Hmr, Lhr, Nup96), sterility at least in part, evolved by divergent natural seare elaborated in (5).

26

www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

737

27

y to mate if they are of nents of reproductive isolation lacking identifiable less of relatedness as causes (Fig. 2). The unidentified component of speciation, if built by selection and not genetic affinity. is also prematComponents divergent selection plants (28), Littorina marine snail ecotypes n which rentially inhabiting different zones of the intertidal 15 basis of (24), and mosquito fish inhabiting blue e under 10 holes with and without fish predators in the nt selecBahamas (29) (Fig. 1A). In these studies, it ortative 5 was shown that males and females are more insects, likely to mate if they are of the same ecoin fish, 0 or preftype, regardless of relatedness as indicated notypic by phylogenetic affinity. Components unknown cause flowerEcological speciation is also supported 15 be under by examples of premating reproductive environ10 isolation in which individuals choose or 30, 31)]. preferentially encounter mates on the basis vergent 5 directly of phenotypic traits that are under ecologiuced fitcally based divergent selection. Examples 0 he enviinclude assortative mating by host choice -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 migrant in insects, body size and coloration in fish, 31, 32)] Cumulative reproductive isolation beak size in birds, pollinator preferences for ybrids in of of the the magnitude of reproductive isolation Fig. 2. 2. Estimates Estimates magnitude of reproductive [extrin- Fig. specific phenotypic floral traits, and variaresulting from divergent selection components (top ), compared isolation resulting from divergent selection components 3)]. For tion in flowering timetraits inferred to be with other componentswith lacking identifiable causes (bottom ). (top ), compared other components lacking l perenDivergent selection components include those attributable to under divergent selection between environidentiable causes ( bottom ). Divergent selection s of the active selection on traits (immigrant inviability and extrinsic components include those attributable to active ments [see examples in (8, 30, 31)]. uttatus) postzygotic and to trait-based assortative and mating (habitat selectionisolation) on traits (immigrant inviability extrinsic Ecologically based divergent selection North preference, floral isolation, and breeding time). The assortative unattributed postzygotic isolation) and to trait-based s when components include intrinsic hybrid inviability, sexual selection has also been directly measured, as shown mating (habitat preference, oral isolation, and breeding of the against by reduced fitness of each ecotype in the enpollen competition, and reduced hybrid fecundity. time).hybrids, The unattributed components include intrinsic mple of Data were taken from (32, 31) (table S1). A negative value vironment of the other [immigrant inviabilhybrid inviability, sexual selection against hybrids, pollen ypic dif- indicates that hybrids had higher fitness than the parental species competition, and reduced hybrid fecundity. Data were ity; reviewed in (31, 32)] and by reduced fitutes di- for at least one component of postzygotic isolation. One data value taken from (32, 31) (table S1). A negative value indicates ness of hybrids in the parental environments because of 2.66 was left out of the bottom panel. that hybrids had higher tness than the parental species [extrinsic postzygotic isolation (33)]. For for at least one component of postzygotic isolation. One example, each of the coastal perennial and value of 2.66 was left out of the bottom panel. 739 VOL 323 data 6 FEBRUARY 2009
Number of studies

lection between environments. The connections between selection on ordinary phenotypic traits and reproductive isolation are often strong and straightforward. It follows that much of the genetic basis of reproductive isolation should involve ordinary genes that underlie differences in phenotypic traits. But we still know little about the genetics of ecological speciation. One line of evidence comes from tests of parallel speciation, whereby greater reproductive isolation repeatedly evolves between independent populations adapting to contrasting environments than between independent populations adapting to similar environments (20, 23). A major challenge in applying the test to natural populations is to eliminate the possibility that each ecotype has originated just once and has spread to multiple locales. This is difficult because gene flow of neutral markers between closely related but nearby populations can result in the false appearance of multiple independent origins of these populations when evaluated by phylogenies (3, 24). However, multiple origins are supported in several examples of parallel speciation, including the sympatric benthiclimnetic species pairs of threespine stickleback in young lakes of British Columbia (25, 26), the repeated origin of divergent marine and stream populations of threespine stickleback around the Northern Hemisphere (27), ecotypes of Timema walking stick insects living on different host

inland annual races of the monkey flower (Mimulus guttatus) along the west coast of North America has low fitness when transplanted to the habitat of the other (31). This is an example of active selection on phenotypic differences, and it also constitutes direct reproductive isolation because it is an evolved barrier to gene flow between parental populations. Multiple traits are probably involved, including flowering time and tolerance of salt and drought. This type of reproductive isolation is context-dependent and is weakened in intermediate environments. On the other hand, active selection favors the evolution of ever-greater differences between populations, which may strengthen the barrier to gene flow (20). It is unclear how much reproductive isolation typically evolves by ecologically based divergent selection in nature. We can approximate an answer from estimates of the combined contribution of active selection on traits and trait-based assortative mating, as compared with other forms of reproductive isolation (Fig. 2 and table S1). These estimates are incomplete because individual studies may lack data on components of reproductive isolation, separate components may not be independent, and the strength of barriers between species may not be symmetric (34). Nevertheless, compilation of the data shows that the amount of reproductive isolation attributable to active selection and traitbased assortative mating is at least as strong, on average, as the amount from components

of reproductive isolation lacking identifiable causes (Fig. 2). The unidentified component of speciation, if built by selection and not genetic drift, could be the result of either ecological or mutation-order mechanisms. These examples indicate a growing knowledge of the mechanisms of selection and its consequences for reproductive isolation. At this point, the most glaring deficiency is our knowledge of the impact of selection on genes. Optimistically, progress is being made with genetic mapping to identify quantitative trait loci (QTLs) and genes or regulatory control regions that affect individual phenotypic traits on which components of reproductive isolation depend. Examples include the yup QTL, which affects flower color differences between the monkey flowers, Mimulus cardinalis and M. lewisii (35). Swapping alleles of this QTL between the species with repeated backcrossing resulted in shifts in pollinator preference and, hence, indirectly affected premating isolation. Survival and salt tolerance of second-generation hybrids between the sunflowers Helianthus annuus and H. petiolaris transplanted to the salt marsh habitat of their hybrid descendent species (H. paradoxus) mapped strongly to a QTL identified as the salt tolerance gene CDPK3 (36). Another form of investigation involves the analysis of genome scans of ecologically different populations and species. These scans compare allelic variation within and between populations at many marker loci spaced throughout the genome (37). Markers that show excessive differentiation between populations (outliers) may indicate selection on nearby genes. The method is particularly informative when applied to populations with ongoing hybridization, because outlier markers may identify points in the genome that resist the homogenizing influence of gene flow, perhaps indicating genomic regions under divergent selection. However, sets of genes that diverged under a mutation-order process can produce the same pattern (17, 18), which makes analysis of such studies more difficult. Clues to whether ecologically based divergent selection is involved are gained if outliers at the same genomic locations turn up repeatedly in scans between populations that inhabit contrasting environments (38) and by identifying phenotypic traits under divergent selection that map to those locations in the genome (36, 39, 40). As genomic resources increase for more species, it will be possible to measure natural selection directly on genomic regions of interest by transplanting otherwise relatively homogenous experimental populations containing alternative alleles into the environments of the parent species (35).

of mutation-order divergence. It may be that the mutation-order process is more difficult to detect, or perhaps we have not looked hard enough at species with only small ecological differences (5). We still do not know much about the selective factors causing mutation-order speciation. Evidence for mutation-order speciation comes from instances in which reproductive isolation apparently evolved as a by-product of conflict resolution between genetic elements within individuals (intragenomic conflict), such as cytoplasmic male sterility in hermaphroditic plants (Fig. 1B), and genetic elements conferring meiotic drive. Under both mechanisms, a mutation arises that can distort representation in gametes and spreads in a selfish manner, even though such an element reduces overall fitness of the organism that bears it. This, in turn, places selection on mutations in other genes that counter the selfish elements effects and restore more equal genetic representation in gametes. Distorter and restorer mutations are unlikely to be the same in different populations regardless of environment; thus the process leads to divergence. The mismatch between the distorter in one population and the restorer in the other can result in hybrid sterility or inviability and, thus, reproductive isolation (3, 41). Numerous examples of selfish elements, such as those observed in cytoplasmic male sterility of plants, support these hypotheses (42, 43). In addition, partial reproductive isolation generated by meiotic drive has been identified in Drosophila [reviewed in (3, 41)]. Sexual conflict is also expected to lead to mutation-order speciation, but there are few compelling examples (3). The contribution by these mechanisms to speciation is still uncertain, however. The alleles responsible for meiotic drive and cytoplasmic male sterility may be prevented from spreading to fixation because selection on such elements is frequencydependent (43) and because restorer alleles arise and weaken selection on the distorter elements (44). Second, if divergent populations come into secondary contact, the alleles within each population causing cytoplasmic male sterility or meiotic drive (and the corresponding restorer alleles) will spread between the populations by gene flow, eliminating that component of reproductive isolation (43). Hence, for these mechanisms to contribute to speciation, the fitness of hybrids must be reduced to very low levels, or other incompatibilities must arise that interact with these genes to prevent their spread after secondary contact.

between genes could generate reproductive isolation (45). Mostly, I expect that he would be chuffed by mounting evidence for the role of natural selection on phenotypic traits in the origin of species. This is really what On the Origin of Species was all about. Between 1859 and the present, the general acceptance of the biological species concept altered the focus of speciation studies. Yet, the discovery that reproductive isolation can be brought about by ecological adaptation in ordinary phenotypic traits bridges Darwins science of speciation and our own. The most obvious shortcoming of our current understanding of speciation is that the threads connecting genes and selection are still few. We have many cases of ecological selection generating reproductive isolation with little knowledge of the genetic changes that allow it. We have strong signatures of positive selection at genes for reproductive isolation without enough knowledge of the mechanisms of selection behind them. But we hardly have time to complain. So many new model systems for speciation are being developed that the filling of major gaps is imminent. By the time we reach the bicentennial of the greatest book ever written, I expect that we will have that much more to celebrate. References and Notes

Mutation-Order Speciation Mounting evidence for divergent selection in speciation does not diminish the potential role

Conclusions Our understanding of the role of natural selection in speciation has come a long way since Darwins time. If he were here to witness, he would most likely be staggered by the discoveries of genes and molecular evolution and astonished at the prospect that evolutionary conflict

1. C. Darwin, On the Origin of Species by Means of Natural Selection (J. Murray, London, 1859). 2. D. Schluter, The Ecology of Adaptive Radiation (Oxford Univ. Press, Oxford, 2000). 3. J. A. Coyne, H. A. Orr, Speciation (Sinauer Sunderland, MA, 2004). 4. P. R. Grant, B. R. Grant, How and Why Species Multiply: The Radiation of Darwins Finches (Princeton Univ. Press, Princeton, NJ, 2008). 5. T. Price, Speciation in Birds (Roberts & Company, Greenwood Village, CO, 2008). 6. T. G. Dobzhansky, Genetics and the Origin of Species (Columbia Univ. Press, New York, 1937). 7. E. Mayr, Systematics and the Origin of Species (Columbia Univ. Press, New York, 1942). 8. D. Schluter, Trends Ecol. Evol. 16, 372 (2001). 9. H. D. Rundle, P. Nosil, Ecol. Lett. 8, 336 (2005). 10. G. S. Mani, B. C. Clarke, Proc. R. Soc. London Ser. B Biol. Sci. 240, 29 (1990). 11. M. Turelli, N. H. Barton, J. A. Coyne, Trends Ecol. Evol. 16, 330 (2001). 12. C. L. Hubbs, Am. Nat. 74, 198 (1940). 13. T. Dobzhansky, Am. Nat. 74, 312 (1940). 14. Additional references are available as supporting material on Science Online. 15. J. A. Endler, Am. Nat. 139, S125 (1992). 16. W. R. Rice et al., Proc. Natl. Acad. Sci. U.S.A. 102, 6527 (2005). 17. N. Barton, B. O. Bengtsson, Heredity 57, 357 (1986). 18. A. S. Kondrashov, Evolution 57, 151 (2003). 19. L. H. Rieseberg, personal communication (2008).

20. W. R. Rice, E. E. Hostert, Evolution 47, 1637 (1993). 21. J. R. Dettman, C. Sirjusingh, L. M. Kohn, J. B. Anderson, Nature 447, 585 (2007). 22. Z. D. Blount, C. Z. Borland, R. E. Lenski, Proc. Natl. Acad. Sci. U.S.A. 105, 7899 (2008). 23. D. Schluter, L. M. Nagel, Am. Nat. 146, 292 (1995). 24. R. K. Butlin, J. Galindo, J. W. Grahame, Philos. Trans. R. Soc. London Ser. B Biol. Sci. 363, 2997 (2008). 25. H. D. Rundle, L. Nagel, J. W. Boughman, D. Schluter, Science 287, 306 (2000). 26. E. B. Taylor, J. D. McPhail, Proc. R. Soc. London Ser. B Biol. Sci. 267, 2375 (2000). 27. J. S. McKinnon et al., Nature 429, 294 (2004). 28. P. Nosil, S. P. Egan, D. J. Funk, Evolution 62, 316 (2008). 29. R. B. Langerhans, M. E. Gifford, E. O. Joseph, Evolution 61, 2056 (2007). 30. S. Via, Trends Ecol. Evol. 16, 381 (2001). 31. D. B. Lowry, J. L. Modliszewski, K. M. Wright, C. A. Wu, J. H. Willis, Philos. Trans. Royal Soc. London Ser. B Biol. Sci. 363, 3009 (2008). 32. P. Nosil, T. H. Vines, D. J. Funk, Evolution 59, 705 (2005). 33. H. D. Rundle, M. C. Whitlock, Evolution 55, 198 (2001). 34. N. H. Martin, J. H. Willis, Evolution 61, 68 (2007). 35. H. D. Bradshaw, D. W. Schemske, Nature 426, 176 (2003). 36. C. Lexer, Z. Lai, L. H. Rieseberg, New Phytol. 161, 225 (2004). 37. M. A. Beaumont, Trends Ecol. Evol. 20, 435 (2005). 38. P. Nosil, D. J. Funk, D. Ortiz-Barrientos, Mol. Ecol. 18, 375 (2009). 39. S. M. Rogers, L. Bernatchez, Mol. Biol. Evol. 24, 1423 (2007). 40. S. Via, J. West, Mol. Ecol. 17, 4334 (2008). 41. H. A. Orr, J. P. Masly, N. Phadnis, J. Hered. 98, 103 (2007). 42. D. A. Levin, Syst. Bot. 28, 5 (2003). 43. L. H. Rieseberg, J. H. Willis, Science 317, 910 (2007). 44. D. C. Presgraves, Trends Genet. 24, 336 (2008). 45. D. C. Presgraves, Curr. Biol. 17, R125 (2007). 46. L. Fishman, J. H. Willis, Evolution 60, 1372 (2006). 47. A. L. Case, J. H. Willis, Evolution 62, 1026 (2008). 48. We thank M. Arnegard, R. Barrett, A. Case, H. Hoekstra, M. Noor, P. Nosil, S. Otto, T. Price, L. Rieseberg, S. Rogers, A. Schluter, S. Via, M. Whitlock, J. Willis, and a reviewer for assistance and comments. This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada and the Canada Foundation for Innovation.

Supporting Online Material

www.sciencemag.org/content/full/323/5915/737/ DC1 Tables S1 to S3 References 10.1126/science.1160006

28

29

values that place two bacterial isolates into the same or different species. The overall genetic SPECIAL SECTION relatedness of is isolates mayhuman be measured by the pneumoniae a major pathogen, S. extent of DNA hybridization between them, and mitis is a commensal bacteria with a history S. Rogers, A. Schluter, S. or Via,more M. Whitlock, J. Willis, and 40. S. Via, J. West, Mol. Ecol. 17, 4334 (2008). 32. P. Nosil, T. H. Vines, D. J. Funk, Evolution 59, 705 (2005). those that show 70% DNA hybridpseudoof taxonomic 11), and S. work a reviewer foruncertainty assistance and ( comments. This 41. H. A. Orr, J. P. Masly, N. Phadnis, J. Hered. 98, 103 33. H. D. Rundle, M. C. Whitlock, Evolution 55, 198 (2001). ization are defined as the same species ( 2 , 10 ). pneumoniae is aby recently described organism of was supported grants from the Natural Sciences and (2007). 34. N. H. Martin, J. H. Willis, Evolution 61, 68 (2007). Such cutoffs imply that sequences that cluster Engineering Research Council of Canada and the Canada 42. D. A. Levin, Syst. Bot. 28, 5 (2003). 35. H. D. Bradshaw, D. W. Schemske, Nature 426, 176 (2003). uncertain status that nonetheless corresponds to Foundation foraInnovation. 43. L. H. Rieseberg, J. H. Willis, Science 317, 910 (2007). 36. C. Lexer, Z. Lai, L. H. Rieseberg, New Phytol. 161, 225 together with certain of similarity data (12 ). There are a distinct cluster in theseamount 44. D. C. Presgraves, Trends Genet. 24, 336 (2008). (2004). must be from the same species, and moreover Supporting Online Material of sequence striking differences in the amount 45. D. C. Presgraves, Curr. Biol. 17, R125 (2007). 37. M. A. Beaumont, Trends Ecol. Evol. 20, 435 (2005). that this cutoff value is applicable to all groups www.sciencemag.org/cgi/content/full/323/5915/737/DC1 diversity observed within homologous house46.2L. Fishman, J. H. Willis, Evolution 60, 1372 (2006). 38. P. Nosil, D. J. Funk,1D. Ortiz-Barrientos, Mol. Ecol. 18, 2,3,4 1 1 Tables S1 to S3or archaea. Recent MLSA studies, of bacteria Christophe Fraser, * Eric J. Alm, Martin F. Polz, Brian G. William Hanage 47. A. L. Case, J. Spratt, H. Willis, Evolution 62,P. 1026 (2008). 375 (2009). genes in these named species, ranging keeping References which use the concatenated sequences of mul48. We thank M. Arnegard, R. Barrett, A. Case, H. Hoekstra, 39. S. M. Rogers, L. Bernatchez, Mol. Biol. Evol. 24, 1423 from 1.2% for S. pneumoniae to 3.0% for S. 10.1126/science.1160006 M. Noor, P. Nosil, S. Otto, of T. Price, Rieseberg, housekeeping genes to discern clustering The (2007). Bacteria and Archaea are the most genetically diverse superkingdoms life, L. and techniques tiple pseudopneumoniae and up to 5.0% for S. mitis. patterns among populations of closely related for exploring that diversity are only just becoming widespread. Taxonomists classify these The distance between two randomly selected organisms into species in much the same way as they classify eukaryotes, but differences in their taxa, suggest that species defined by taxonois similar to the average S. mitisthat genotypes values place two bacterial isolates into disthe mists in many cases correspond to well-resolved biology including horizontal gene transfer between distantly related taxa and variable rates of REVIEW S. pseudotance between S. pneumoniae and same or different The overall genetic clusters. species. However, these studies also homologous recombinationmean that we still do not understand what a bacterial species is. This sequence pneumoniae genotypes (5.1%) (2). This implies relatedness of isolates may be cutoff measured by the show that there is no universal or descripis not merely a semantic question; evolutionary theory should be able to explain why species the use of that a fixed level of sequence diver that extent of DNA hybridization between them, and tor of clusters characterizes a species. Furexist at all levels of the tree of life, and we need to be able to define species for practical species would tend gence for differentiating those that show 70% or more DNA hybridthermore, inspection of the clusters does not applications in industry, agriculture, and medicine. Recent studies have emphasized the need to to either rejoin S. pneumoniae and S.hierarchy pseudoization are defined as the same (2, 10). always clearly reveal which level species in the combine genetic diversity and distinct ecology in an attempt to define species in a coherent and break up sequences S. mitis so(Fig. that 1) nearly pneumoniae , or Such cutoffs imply that that cluster is more fundamental than any other (7). convincing fashion. The resulting data may help to discriminate among the many theories of of its own. This is every isolate was a species together with a certain amount of similarity As an example, Fig. 1A shows the relationprokaryotic species that have been produced to date. clearly unsatisfactory. must be frommultiple the same species, moreover ships among isolates of and three closely that this streptococcal cutoff value is species. applicable to all groups Streptococcus he species debate in microbiology is not reflect only a tiny subset of those characters that related 1 2,3,4 2 1 it traordinary variety of 1 microbial life, much he species debate in isbacnot Habitats and Ecological Differentiation of bacteria or archaea. Recent MLSA S. studies, Christophe Fraser, * Eric J.microbiology Alm, Martin F. Polz, Brian G. Spratt, Williamresources P. Hanage is a major human pathogen, mitis allow bacteria to use different inof the pneumoniae only about a human desire to catalog desiremanner, to catalog Beyond this, taxa too fraction similar of to is A clear natural criterion to identify clusters of uncultured (1). only diversity about a in human which use the concatenated sequences of taxomula commensal bacteria with a history of and only capture a small terial a consistent but environment, by rRNA se- nomic importance, we might want in consistent manbe distinguished and evolutionary tiple housekeeping topseudopneumoniae discern clustering The Bacteria anddiversity Archaea areabecause the most diverse superkingdoms life, and techniques uncertainty (11genes ), andwhich S. true diversity in circumscribed this of superkingdom of life. is also abacterial fundamental argument ofgenetically what it the patterns among of uncertain closely related butabout is also a fundamental argument because ecological features that quences have Taxonomists revealed further diversity through is to species, is populations to find ner, for exploring that diversity of are only just becomingMore widespread. classify these acall recently described organism status recently, molecular methods [particularly reveals our ignorance how evolutionary taxa, suggest that species defined bycluster taxonoabout our ignorance of way how analysis (MLSA) ( 2) and that Among of what it into reveals distinguish them from close relatives. multilocus organisms species much the same classifysequence eukaryotes, but differences in their nonetheless corresponds to a distinct in DNA-DNA hybridization and ribosomal RNA forces form, shape, and in extinguish bacterial ge-as they mists data in many correspond to differences well-resolved biology including horizontal gene transfer between distantly related taxa and variable rates of and extinguish diversity evolutionary forces form, shape, metagenomic studies (1 ), and pathogens, the ability tostriking cause a distinctive these (12 ).cases There are in (rRNA) sequencing] have helped to this define species, netic lineages, of the mechanisms of differensequence clusters. However, these studies also homologous recombination mean that we still not these understand what a by bacterial species This of dobut has been used to within define bacterial genetic lineages, of the mechanisms needs to be explained theory. Thus, is. practidisease amount ofhistorically sequence diversity observed methods have serious limitations and the tiation between subpopulations sharing common show that there is no universal cutoff ora descripis not merely abetween semantic question; evolutionary should beassign able of to explain why species lack and observations constitute only minute differentiation subpopulations sharing species, but pathogens cal difficulties, housekeeping genes in these named cannot reliably a theory, large collection of similar homologous descent, and of the process of adaptation to new theory tor of clusters that characterizes a species. Furexist at all levels of the tree of life, and we need to be able to define species for practical species, ranging from 1.2% for S. pneumoniae to and of the process of adaptaas yet unclassified microbial of of vast to amounts common descent, speciesof (e.g., rRNA sequences are too fraction of overall bacterial diversity. Mapping niches and changing environments. Animal spe- strains thermore, inspection of the clusters does applications in industry, agriculture, and medicine. Recent studies have emphasized therRNA need to for S. pseudopneumoniae and up to 5.0% not for to resolve similar species). se- 3.0% cies are defined by and theirchanging morphological and be- conserved environments. of how tion to new niches diversity have all fueled the controversy bacterial diversity onto environmental resources always reveal which level in randomly the combine traits genetic diversity distinct ecology attempt to define species in acoherent and S. mitis.clearly The between two sequence surveys have, however, revealed havioral and by defined their and ability or inability to indicates thatdistance closely related groups ofhierarchy bacteria Animal species are by their morpho - in an one defines a bacterial species (3 8). the extrais more fundamental than any other (Fig. 1)fine(7). convincing fashion. The resulting data may help to discriminate among the many theories of lected S. mitis genotypes is similar to the average ordinary variety of microbial life, much of it interbreed, but such categories cannot easily be can be ecologically divergent. For example, logical and behavioral traits and by their abilAsresource an example, 1A has shows the relationprokaryotic species that been produced to date. between S. Fig. pneumoniae and S.observed pseudouncultured (1). Beyond this, taxa too similar to be distance applied to the Bacteria orhave Archaea (or indeed to Genetic Clustering but such categories been ity or inability to interbreed, scale partitioning ships among multiple closely genotypes (5.1%) (of 2). three This implies and circumscribed rRNA se- pneumoniae many eukaryotic taxonomists or Ar- distinguished Darwin commented that all trueby classification cannot easily be microbes). applied toInstead, the Bacteria among coastal Vibrio isolates populations coexisting related streptococcal species. Streptococcus reflect only a tiny subset of those characters that he species debate in microbiology is not that the use of a fixed level of sequence divergence quences have revealed further diversity through have been forced to rely on biochemical tests and chaea (or indeed to many eukaryotic microbes). is genealogical [(9), p. 404]. Taxonomists have in the water column (13). Partitioning was pneumoniae is a major human pathogen, mitis allow bacteria to use different(MLSA) resources in the for only about a human desire to catalog bacdifferentiating species would tend toS.either sequence analysis (2cutoff ) and limited morphological characteristics for this taxonomists have been forced to purrely multilocus to define were collected from thus used sequence relatedness discovered because strains Instead, is a commensal bacteria with a history of taxoenvironment, and only capture a small fraction of terial diversity in a consistent manner, but rejoin S. pneumoniae and S. pseudopneumoniae , metagenomic studies ( 1 ), and this diversity needs pose. Naturally, biochemical characters have been on biochemical tests and limited morphological values that place two bacterial isolates into the distinct, ecologically informative samples, and nomic uncertainty ( 11 ), and S. pseudopneumoniae the true diversity in this superkingdom of life. is also a fundamental argument because of what it or break up S. mitis so that nearly every isolate was to be explained by theory. Thus, practical difselected for the convenience of taxonomists; they characteristics for this purpose. Naturally, bio- same or different species. The overall genetic the phylogenetic structure of the ecologically isspecies a recently described organism of uncertain status molecular methods [particularly reveals about our ignorance of how evolutionary More recently, of its own. This is clearly unsatisfactory. lack theory,may and observations ofof isolates be measured of byvast the a populations was superimposed chemical characters have been selected for the ficulties, relatedness differentiated that nonetheless corresponds to a distinct cluster in 1 DNA-DNA hybridization and ribosomal RNA forces form, shape, and extinguish bacterial geamounts of as yet unclassified microbial diversity Department of of Infectious Disease Epidemiology, taxonomists; they reflect Imperial only a extent of DNA hybridization between them, and on their habitats. Habitats were defined using convenience 2 Habitats Ecological these dataand (12). There are Differentiation striking differences in (rRNA) sequencing] have helped to define species, netic lineages, of the mechanisms of differenCollege London, London W2 1PG, UK. Department of Civil have all fueled the controversy of how one detiny subset of those characters that allow bac- those that show 70% or more DNA hybridiza- an empirical modeling approach. This analysis and Environmental Engineering, Massachusetts Institute of amount of sequence diversity observed within but these methods have(3serious and theclear tiation between subpopulations sharing common criterion to identify clusters of fines a bacterial species 8 ). limitations defined as the same species (2, 10). Such A revealednatural high levels of specialization for some teria to use different resources in 3the environtion are Department of Technology, Cambridge, MA 02139, USA. homologous housekeeping genes we in these named cannot reliably assign a large collection of similar evolutionary descent, and of the process of adaptation to new importance, which might want only capture a small fraction of the cutoffs imply that sequences that cluster together populations (e.g., V. ordalii is only found as ment, and Biological Engineering, Massachusetts Institute of Technolspecies, rangingis from 1.2% for S. pneumoniae to strains toClustering species (e.g., rRNA sequences are too to niches and changing environments. Animal spe- Genetic call species, to find ecological features that of MIT ogy, MAthis 02139, USA. 4Broad Institute superkingdom of life. More true Cambridge, diversity in with a certain amount of similarity must be from single free-swimming cells), whereas others are 3.0% for S. pseudopneumoniae and up to 5.0% for conserved to resolve that similar rRNA se- distinguish ciesHarvard are defined by their morphological and be- Darwin them from close relatives. Among commented allspecies). true classification and University, Cambridge, MA 02139, USA. DNAa wide recently, molecular methods [particularly the same species, and moreover that this more generalist (Fig. 1B) and can colonize S. mitis . The distance randomly sequence surveys revealed thecutoff extrahavioral traits and by their ability or inability to is the ability between to cause two a distinctive disgenealogical have, [(9), however, p. 404]. Taxonomists have pathogens, *To whom correspondence should be addressed. E-mail: DNA hybridization ribosomal RNA (rRNA) value is applicable to all groups of much bacteria or variety of surfaces, including organic particles lected S. mitis genotypes is similar to the average ordinary variety of microbial life, of it interbreed, but suchand categories cannot easily be thus c.fraser@imperial.ac.uk used sequence relatedness to define cutoff ease has historically been used to define species, helped define (or species, but Recent MLSA studies, which use zooplankton water column 13pseudo). Most sequencing] have distance between in S.the pneumoniae and ( S. uncultured (1). Beyond this, taxa too similar tothe be and applied to the Bacteria or to Archaea indeed to archaea. and can- concatenated of multiple Vibrio (5.1%) populations are implies deeply these have serious Instead, limitations the predicted pneumoniae genotypes (2). This distinguished sequences and circumscribed byhousekeeprRNA se- of many methods eukaryotic microbes). taxonomists among cases 741 not reliably assign large collection www.sciencemag.org of similar genes to SCIENCE discern clustering patterns divergent eachlevel other, and in many VOL 323 6 FEBRUARY 2009 that the usefrom of a fixed of sequence divergence quences have revealed further diversity through have been forced to a rely on biochemical tests and ing to species (e.g., rRNA sequences are too populations of closely related taxa, suggest that are congruent with named species; however, V. strains limited morphological characteristics for this pur- multilocus sequence analysis (MLSA) (2) and for differentiating species would tend to either similarcharacters species). have rRNA se- species defined by taxonomists many needs cases splendidus is a notable exception and splits into conserved to resolve rejoin S. pneumoniae and S. pseudopneumoniae , metagenomic studies (1), and this in diversity pose. Naturally, biochemical been ex- correspond sequence clusters. related groups with distinct quence have, however, revealed thethey to well-resolved or break up closely S. mitis so that nearly every isolate was to be explained by theory. Thus, practical dif- numerous selectedsurveys for the convenience of taxonomists; is ecological indicating However, these studies and alsoobservations show that there presumably a species of preferences, its own. This is clearly unsatisfactory. ficulties, lack of theory, of vast no universal cutoff or descriptor of clusters that recent ecological radiation from a sympatric 1 amounts of as yet unclassified microbial diversity Department of Infectious Disease Epidemiology, Imperial Habitats population and Ecological a species. Furthermore, inspection (13). Differentiation Thus, genetic clusters characterizes College London, London W2 1PG, UK. 2Department of Civil have all fueled the controversy of how one de- ancestral and Environmental Engineering, Massachusetts Institute of clearly reveal that ecology be discerned. of the clusters does not correlate with A clear natural criterion tocan identify clusters of fines a bacterial species (3always 8). Technology, Cambridge, MA 02139, USA. 3Department of What do the genetic data tell us about mechwhich level in the hierarchy is more fundamen- evolutionary importance, which we might want Biological Engineering, Massachusetts Institute of TechnolGenetic Clustering than any other (Fig. 1) (7). differentiation and that the anisms of population tal to call species, is to find ecological features ogy, Cambridge, MA 02139, USA. 4Broad Institute of MIT As an example, Fig. shows the relation- evolutionary queshistory of close the microbes distinguish them from relatives. in Among Darwin commented that1A all true classification and Harvard University, Cambridge, MA 02139, USA. pathogens, ability to organized cause a distinctive disis genealogical [(9), p.isolates 404]. Taxonomists have tion? of three closely ships among multiple That the bacteria are into genetic *To whom correspondence should be addressed. E-mail: c.fraser@imperial.ac.uk ease hasis historically been usedinteresting to define species, thus used sequence relatedness define cutoff clusters not, per se, a very obserrelated streptococcal species. toStreptococcus
REVIEW

Speciation
lyt icu s
r ua ia s i nu ge i/lo r e ch

The Bacterial Species Challenge: Making Sense of Genetic and Ecological Diversity

V. r
vibrio n Entero V. calvie
S. mitis 5% divergent

um
V. a

V. ordalii

orvegic

us

lg in o
st ae is V. V. f eri ch fis . V

oi en

si s

S. pseudopneumoniae S. pneumoniae 3% divergent 1.2% divergent

nsis

s V.

rs pe

teu

s a*

p V.

e an

cid

T T

The Bacterial Species Challenge: Making Sense of Genetic and Ecological Diversity

S. oralis 0.005

Vibrio splendidus

0.005

Fig. 1. Multilocus sequence analysis of closely related species. (A) Radial minimum evolution tree constructed using MEGA4, showing clusters among 97 isolates of four Streptococcus species identified as indicated. The tree was built using concatenates of six housekeeping loci, resulting in a total of 2751 positions in the final data set (2). Distances were calculated as the percentage of variant nucleotide sites. The mean distance within the clusters, calculated by MEGA4, is shown. To the right, the pneumococcal cluster is shown at larger scale, and putative subclusters are indicated in dark gray, purple, and green. (B) Ecological

associations of Vibrionaceae sequence clusters (13). Habitats (colored dots) were estimated as differential distributions of groups of closely related strains among samples (size fractions enriched in different environmental resources). Clusters associated with named species are evident, and in most cases species show a clear predilection for one of the habitats. The exception is V. splendidus, which breaks up into many closely related ecological populations. Asterisk denotes that trees based on additional loci indicate that the placement of V. panecida within V. splendidus may be an artifact of horizontal gene transfer at the Hsp60 locus.

30

not, per se, The a very interesting observation; many or makes strict thereference accumulation of neutral diversity. The but pathogens only a minute fraction of diversity. of a population first proposed mechanism was unifying biological to the vation; many constitute or most models most models of a population with a principles first proposed was based on artificial overall bacterial diversity. of mutation bacterial based selection reproducing experiments with reproducing with a small Mapping amount of on artificial of mechanism selection and niche partitioning, amount of mutation will eventually produce selection experiments with bacteria grownas for diversity onto environmental resources indicates small will eventually produce populations consisting under the ecotype has rightly become popular a bacteria grown for extended periods populations consisting of clusters of related orga- framework extended periods under stable in chethat closely of related groups of bacteria can be ecorelated organisms, irrespective of stable which to conditions discuss bacterial of clusters conditions in chemostats, which showed within nisms, irrespective the details of the the evolutionmostats, which showedand repeated selective sweeps logically divergent. For example, fine-scale re- repeated forces or ecologiselective of sweeps in which whole evolution, the details of the evolutionary speciation, ecology. ary forces or ecological more in The which the whole was thought to source partitioningA has been observed among genome fixation ecotype modelgenome (4, 16, 24 ) predicts that cal differentiation. more substantial observawas thought todifferentiation. hitchhike to A is thatmutation there is very little common hitchhike ancestry to fixation along with an advantageous coastal Vibrio populations coexisting in the water substantial there is very little neutral diversity will be preserved among bacalong with observation an advantageous (periodic tion is that diversity many populations microbes, mutation (periodic selection) (22). Selective column (13 ). Partitioning discovered because of was microbes, from which neutral (22). in Selective sweeps of can purge terial (which sweeps should in many populations populations within niches selection) from which we may infer some features of the be canmonophyletic), purge almost alland genetic diversity the popstrains were collected distinct, ecologically in the population we may infer some from features of the selective almost all genetic diversity thus predictsinthat ecolandscape. Neutral diversity is the ulation and thus constitute a candidate informative samples,diversity and the phylogenetic struclandscape. Neutral is the amount of selective for types are coherent self-contained genemechapools. and thus constitute a candidate mechanism of polymorphism is evident in non- As nism for reducing variation that (23). ture of the ecologically differentiated populations in noncoding re- amount been suggested ecotypes polymorphism that is evident reducing neutral variationthat (23). a result, it has neutral was superimposed onsynonymous their habitats. substitutions. Habitats were coding regions or results in synonymous sub- should be considered as putative or actual spegions or results in Niches and Ecotypes stitutions. One common measure of neutral cies, defined using measure an empirical modeling approach. and Ecotypes One common of neutral diversity is the Niches depending on the level of genetic differdiversity is the effective population size N To extendfrom this model, one can population. consider multiple This analysis revealed high levels of specialization , dee the ancestral This effective population size Ne, defined as the size To extend this model, one can consider multiple entiation ecological niches characterized by of theproviding selective for some populations (e.g., V. ordalii is only found fined as the size of a population evolving in the of a population evolving in the absence of se- ecological niches characterized by the selective model therefore has the advantage advantages they confer to specific genes. This is as single free-swimming cells), others are of selection that would generate as much genes. This a lection that would generate aswhereas much neutral di- absence advantages they confer to specific mechanistic understanding of the evolutionmore generalist (Fig. 1B) and can colonize a wide neutral diversity as is actually observed. Esti- the ecotype model, where genes adapted to to ary processes, as well as an organizing principle versity as is actually observed. Estimates of Ne is the ecotype model, where genes adapted variety of surfaces, including organic particles and mates of Ne for bacteria range from 105 to 109 specific niches cause selective sweeps within cause selective sweeps within for classifying species, that is based on experifor bacteria range from 105 to 109 (1418). To specific niches zooplankton in the water column (13). Most of the (1418). To put this into context, the numbers of those niches but not in other niches. In this way put this into context, the numbers of Vibrio cells those niches but not in other niches. In this mental observations of bacterial populations. predicted Vibrio populations are deeply divergent Vibrio cells per cubic meter of seawater in tem- the population will undergo adaptation and difHowever, these observations of repeated per cubic meter of seawater in temperate coastal way the population will undergo adaptation from each other, and in 8 many cases are congruent perate coastal regions range from 108 to 109 (19), ferentiation while maintaining relatively low levels regions range from 10 to 109 (19), which sug- and differentiation while maintaining relatively selective sweeps were made in chemostats, of neutral diversity, as selective sweeps confined with named species; however, V. splendidus is a which suggests vast census population sizes census and population sizes (>1020closely ). This low of neutral diversity, as selective ungests vast levels natural environments are markedly to each ecotype regularly purge the population notable exception splits into numerous (>1020 ). This observation a mismatch of many whereas mismatch of many orders of regularly observationa sweeps confined to each ecotype stable and diverse. How would one detect the related groups with distinct ecological preferences, orders of magnitude between effective population of any diversity that might have accumulated size purge sweeps in diversity natural bactemagnitude indicating between effective population the census population of any diversity that selective (Fig. 2A). of Crucially, what neutral we do presumably recent ecological radiation size and population size (true of might most presence accumulated (Fig. Crucially, what examples and census population size (true of (most bac- have populations? The most conclusive observe is predicted to be associated with adaptfrom a sympatric ancestral population 13). Thus, bacteria studied to date) 2A). was originally used to rial tothat date)was originally used to neutral is argue predicted diversity we do observe not The from bacteria but from RNA viruses, teria studied ive traits. ability of such selective sweeps to genetic clusters correlate with ecology can be counter claims of neutrality and instead that come The mutate at much highersize rates than DNAcounter claims of neutrality and instead argue to associated with limit the effective population has been recdiscerned. all be genetic variation wasadaptive adaptive traits. (20, 21 ). which been established from thatWhat all genetic wastell adaptive (20 , 21). ability of there such are selective limit the based life It has ognized forforms. some time (17 , 23), and this model do thevariation genetic data us about mechHowever, several sweeps different to mechanisms population size has (Fig. been2). recognized sequences the However, are several different and mechacollected over many by years that and has been substantially developed Cohan anisms of there population differentiation the effective that can explain this mismatch explain thismicrobes mismatch 2). anddriving this model has population for Whatever some time (17, 23), are the human influenza nisms that can colleagues (structure 4, 16, 24).of Because it links patternsvi of evolutionary history of the in(Fig. question? mechanisms the differWhatever mechanisms are driving the driven by repeated selecbeen substantially developed by Cohan and rus is predominantly That bacteria are organized into genetic clusters is entiation of bacteria into clusters, they must re- genetic differentiation with adaptation, and makes differentiation of bacteria into clusters, they colleagues (4, 16, 24). Because it links patterns tive sweeps (25) and that the resulting effective must restrict the accumulation of neutral of genetic differentiation with adaptation, and population size Ne (<100) is very much smaller 742 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org

www.sciencemag.org

SCIENCE

VOL 323

6 FEBRUARY 2009

741

31

SPECIAL SECTION
A
E1 E2

cts that ng bacould be cotypes As a reshould ies, dentiation l therehanistic sses, as ssifying bserva-

* * * *
stable ecotype concept

* * *

colonization

old and void

D
20 Population size 15 10 5 0 0 20 40 Time 60 80 100 Bacteria Phage

ated sewhereas ble and ence of lations? ot from utate at orms. It ollected cture of y driven hat the 100) is acteria. genetic models karyotic alogies, iven by ngitudions, as shes in changes ological

(27). This structure well describes the situation does not predict an association between neutral than observed for bacteria. The use of longitu- Bottlenecks, Metapopulations, and Local for parasites, which can colonize a host but are diversity and adaptive traits. dinal ecological and genetic data to distinguish Extinctions then forced to move on because the host develops The relevance of the metapopulation model to between competing models of evolution has a The essential element of the ecotype model immunity or dies (17). It also describes any the species question is that, although highly ideneutral diversity long pedigree eukaryotic biology 26). On with to limiting situation wherein bacteria use a limited (resource alizedrespect and simplified, it may capture some is of not the the basis of these analogies, any inference of a niche per se, rather the intensively for short bursts, followed by dispersal effects adaptation of complexity andbut instability of effecactual structure driven by selective sweeps population bottleneck caused by structure. the replacement of to new resource patches (e.g., colonization of tive ecosystems on population Selective data from natupopulation by descendants from a would require good longitudinal the whole organic particles in seawater by Vibrio popula- sweeps are predicted to be inevitable in simple, extinction ral bacterial populations, as well as is observations individual and the resulting tions). This metapopulation model fundamen- single stable environments but not in complex metaof episodic crashes in diversity causally associall other [a lineages (Fig. 2A). Otherin mechatally different from the ecotype model because it of populations point partly addressed (28)]. ated with genetic changes and not associated nisms that induce or involve regular population with changes in ecological covariates. bottlenecks will also restrict neutral diversity. 743 ww.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 2009

del with ot niche bottlewhole le indill other induce ks will pulation ded into etween e popupatches acterial eria are ig. 2B)

Fig. 2. Different models of microbial evolution that lead to low values of Ne. (A) The ecotype model of bacterial population differentiation. The tree shows a single bacterial lineage that differentiates into two sublineages (E1 and E2) that differ in some aspect of their ecology. Periodic selection (a selective sweep) occurs at the points marked by asterisks and eliminates almost all of the diversity that has arisen since the last episode of periodic selection, which is shown by the dashed branches (diversity purged by periodic selection) or solid branches (existing diversity) on the tree. As the two populations are ecologically distinct (i.e., ecotypes), periodic selection in one sublineage does not influence diversity in the other sublineage and vice versa. Each ecotype can therefore diverge to become separate species. Reproduced from (24) with permission. (B) A metapopulation. Patches of varying size (gray circles) are vacant (empty) or may be colonized by a single genotype randomly acquired from another patch. Strains may diversify within a patch (as shown by different colors representing distinct genotypes), which may colonize empty patches as described above. A characteristic of this sort of metapopulation is patch turnover, in which patches occasionally become unable to support colonization and their inhabitants are removed (solid gray circles). (C) A neutral model with small population size. Different genotypes (different colors) arise by mutation or recombination and increase or decrease in the population by random drift. For some purposes, this simple model is an adequate effective description of the more complex processes represented in (A), (B), and (D), and of other more complex evolutionary models not described in this review. (D) Predator-prey dynamics and population bottlenecks. Regular population bottlenecks can drastically shrink the effective population size. In this case, bacteria-phage predator-prey dynamics are simulated with a classical Lotka-Volterra model, which can generate oscillations in population size of any amplitude. Population sizes and time axes are in arbitrary units for illustrative purposes only.

Choosing Between Models It has proven difficult to discriminate between models of population differentiation that focus on ecotypes or metapopulations. For example, the ecotypic structure of a soil Bacillus has been modeled to predict a priori which sequence clusters were ecotypes, and hence which ones should be associated with specific ecological properties (16). Some clusters are associated with certain phenotypic traits, such as a propensity to grow on shady north-facing slopes or sunny south-facing slopes. However, this model fitted no better (and in fact slightly worse) than a version of the model with several subpopulations and diversity generated only by neutral drift. This version of the model was dismissed because of its association with a very low estimate of population size (14). However, estimates of effective population size Ne are often grossly disconnected from census population sizes. It has proven very challenging to find models that successfully explain low estimated values of Ne while providing better predictions than models based on simple neutral drift. The analysis of Bacillus partly did this by predicting more ecotypes in the model than were observed

32

Relative r of of change distance between Relative rate r rate change in in distance between -9)-9) clusters per generation (x10 clusters per generation (x10

iples of ype has within ciation,

B
new and empty colonization

Metapopulation structure, in which the population is divided into patches and where individuals disperse between patches, can generate very low effective population sizes if patches turn over (i.e., if patches are only intermittently able to support bacterial growth, and if a small number of bacteria are dispersed to colonize empty patches) (Fig. 2B) (27). This structure well describes the situation for parasites, which can colonize a host but are then forced to move on because the host develops immunity or dies (17). It also describes any situation where bacteria use a limited resource intensively for short bursts, followed by dispersal to new resource patches (e.g., colonization of organic particles in seawater by Vibrio populations). This metapopulation model is fundamentally different from the ecotype model because it does not predict an association between neutral diversity and adaptive traits. The relevance of the metapopulation model to the species question is that, although highly idealized and simplified, it may capture some of the effects of complexity and instability of actual ecosystems on population structure. Selective sweeps are predicted to be inevitable in simple, stable environments but not in complex metapopulations [a point partly addressed in (28)]. A metapopulation may evolve, differentiate, and adapt without global selective sweeps. Diversity lost by a local selective sweep in one patch may be rescued and reintroduced from other patches. The ecotype model, with its predicted monophyletic relationship between niche and genotype, may therefore not be an appropriate model of speciation in complex ecosystems.

and diversity generated only by neutral drift. This ples shown in Fig. 3 differ only in the rate of to recombination will depend more on the accuto recombination will depend more on the accuples shown in Fig. 3 differbetween only inthe theclusters, rate of mulation and diversity only by neutral drift.of This of differences at neutral loci than at recombination version of thegenerated model was dismissed because its homologous mulation loci. of differences neutral lociathan at homologous recombination between the clusters, version of the model waslow dismissed because of its all The modelsat also assumed homoother parameters being held constant. As re- adaptive association with a very estimate of populaadaptive loci. The models also assumed a homoall other parameters being held constant. As reassociation with a very low estimate of population size (14). However, estimates of effective combination increases, we see a distinction be- geneous distribution of polymorphisms across tion size (14 ). N However, estimates of effective combination increases, we see a distinction be- geneous distribution of polymorphisms across population size e are often grossly disconnected population size Ne ecological are often grossly Recombination rate relative to mutation using established criteria, a hypothfrom census population sizes. It hasdisconnected proven very Recombination rate relative to mutation from census population sizes. has successfully proven very that can be esis challenging to tested. find models It that 0.10 clonal 3 challenging to find models that 0.10 clonal This problem of low power selection 3 explain low estimated values to ofdetect Nesuccessfully while pro0.53 threshold explain low estimated values Ne while proto reject neutrality) is a (or, more accurately, viding better predictions than of models based on 0.53 threshold viding better predictions than models on 2.00 sexual simple neutral drift. in The analysis of based Bacillus genetics that very general problem population 2 2.00 sexual simple neutral drift. The analysis of Bacillus 2 partly did this by more ecotypes in in does not negate thepredicting importance of adaptation partly did this by predicting more ecotypes in the model than were observed established evolution, but rather suggests using that more work theneeded model than were a observed using established 1 ecological criteria, hypothesis that can be model-based methods is if we want 1 ecological criteria, a hypothesis that can be tested. to discriminate among different biologically tested. This problem of low power to detectdata. selection plausible explanations of genetic In 0 problem of lowreject power to detect selection (or, This more accurately, neutrality) is a very a scheme for performing Table 1 we proposeto 0 (or, more accurately, to reject neutrality) is a very general problem in population that does analyses that could be used genetics to test, develop, general problem in population genetics that does not negate the importance of adaptation inmore evoand validate different competing models -1 not negate the importance ofthat adaptation in evolution, but rather suggests more work is -1 systematically. lution, but rather suggests that more work is "Speciation point" for needed if we want model-based methods to dis"Speciation point" for sexual species needed if among we want model-based methods to discriminate different biologically plausible -2 sexual species Homologous Recombination criminate among different biologically plausible -2 explanations genetic data. In Table 1 we proOne specific of challenge to models that invoke explanations of genetic data. In Table 1 we propose a scheme for performing analyses that could involves a feature of bacterial ecotypic structure pose a scheme for performing analyses could -3 be used to test, develop, and validatethat different evolutionhomologous recombinationthat -3 be used to test, develop, and validate different competing models more systematically. 0% 10% 20% 30% we have not yet discussed. Bacterial reproduccompeting models more systematically. 0% 10% 20% 30% Mean genetic distance between clusters the obligate reassortment tion does not involve Homologous Recombination Mean genetic distance between clusters of genetic material observed in most higher orHomologous Recombination 3. The dynamics of cluster divergence. The figure summarizes some key results from (15) in a One specific challenge to models that invoke Fig. Fig. 3. The dynamics of cluster The figure summarizes some key results from (15) in a ganisms. However, recombination does occur in phase-space plot of the geneticdivergence. dynamics of two populations, with recombination occurring beOne specific challenge to a models invoke ecotypic structure involves feature that of bacterial phase-space plot of the genetic dynamics of two populations, with recombination occurring between them at a rate that is varied for the three different simulations. The y axis shows the rate of archaea (29) and typically bacteria ecotypicand structure involves a feature of involves bacterial evolution homologous recombination that tween them at a rate that is varied for the three different simulations. The y axis shows the rate of change of genetic distance between the clusters as a function of the genetic distance itself (x axis). piece of DNA with the replacement of a shortrecombination evolution homologous that we have not yet discussed. Bacterial reproduction When change of rate genetic distance as a function of the genetic distance itself (x axis). the of change isbetween positive,the theclusters populations will diverge genetically; when negative, they strain. the homologous segment from another we have not yet discussed. Bacterial reproduction does not involve the obligate reassortment of converge. When the rate of change of is positive, the each populations will diverge genetically; when negative, they The direction change for scenario is shown by arrows color-coded to each Recombination becomes less probable with does not involve the obligate reassortment of genetic material observed in most higher orga- scenario. converge. For Thelow direction of change for each scenario is shown by arrows color-coded to each recombination rates, the populations are effectively clonal and always diverge sequence divergence between the increasing genetic However, material observed in most higher orgascenario. For As low recombination rates, populations are effectively clonal and alwaysslow diverge nisms. recombination does occur in (green line). the recombination rate the increases, the cohesive effects of recombination the nisms. However, recombination does occur in 31 ), which reduces donor and the recipient ( 30 , (green line). As the recombination rate increases, the cohesive effects of recombination slow the bacteria and archaea (29) and typically involves rate of divergence, until a threshold is passed (red line) and the populations become effectively bacteria and ) and typically involves but does not archaea eliminate recombination between rate of in divergence, threshold is passed (red line) and For the recombination populations become effectively the replacement of a(29 short piece of DNA with sexual the sense until that a the populations no longer diverge. rates above this the replacement of a short pieceanother of such DNA with sexual in the sense that the populations no longer diverge. For recombination rates above this closely related species. Because of interthe homologous segment from strain. level, the fate of the two populations will depend on how genetically distinct they are at the outset. the homologous segment from another strain. any given isolate within species recombination, level, of the populations will depend on how genetically distinct are at the outset. Recombination becomes less probable with in- If theythe arefate within thetwo speciation point, then recombination will cause themthey to merge. If they are Recombination becomes lessto probable with in- farther If they are within the speciation point, then recombination will cause them to merge. If they are at donor least a species is almost certain contain creasing sequence divergence between the away than this speciation point, they will continue to diverge from each other. These creasing sequence divergence the donor farther are away than using this speciation they continue to diverge from each other. These some genetic material that isbetween characteristic of curves derived the model point, described in will (15). and the recipient (30 , 31), which reduces but does and the recipient (30, species. 31), which reduces but does closely related Hence, whereas it curves are derived using the model described in (15). other was once thought that bacteria do not form spe6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 744 cies in the eukaryotic sense because they do not 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 744 recombine at all (32), one current view is that process that reduces the rate of recombination Illegitimate Recombination and Gene they do not form species because they recom- between themfor example, a period of allopa- Content Variation bine too much (5). try or ecological differentiation. The speciation Illegitimate recombination or gene acquisition In asexual clonal organisms, even in the point is the amount of divergence between clus- is another unusual feature of bacteria. In this absence of any selective pressure, clusters will ters that needs to accumulate to prevent them case, genes or clusters of genes are acquired spontaneously split into multiple lineages or from returning to a single cluster if the barriers that typically have no homolog(s) in the recipidaughter clusters (15). However, under cer- to recombination are removed. A recent study ent strain. The importance of this phenomenon tain circumstances recombination can prevent hypothesized that two related Campylobacter is evident in the clear and ubiquitous signature this, and we can hence divide the bacteria into species are currently undergoing this process of such events in the growing body of genomic sexual and nonsexual species. This effect, of merging into a single species as a result of data. These are identified by differences in the described at greater length elsewhere (15), is changes in their environment (33). characteristics of the acquired DNA and that of The above insights were reached using mod- the host strain, for example, in base composition summarized in Fig. 3, which shows the rate at which two clusters diverge over timethat els based on the assumption that genetic varia- or codon usage; in most cases, the donor of the is, the increase in the mean genetic distance tion is neutral. Although this is obviously not DNA in question is unknown. Gene acquisition between them. If this becomes negative, then always an appropriate assumption, it is plau- leads to genomes being punctuated by stretches the two clusters will stop diverging and instead sible that the number of loci explicitly involved of foreign DNA. The largest of these (which converge. The three examples shown in Fig. 3 in adaptive ecological differentiation will be may be many kilobases in length) were initially differ only in the rate of homologous recom- small, and thus that in an unstable landscape, termed pathogenicity islands, because the new bination between the clusters, all other param- genomic barriers to recombination will depend functions encoded by the imports were often eters being held constant. As recombination in- more on the accumulation of differences at neu- involved in virulence, but a better term is gecreases, we see a distinction between a clonal tral loci than at adaptive loci. The models also nomic islands as the phenomenon is far from organism in which clusters are predicted to di- assumed a homogeneous distribution of poly- limited to pathogens (36, 37). Although it is verge (the green line) and a sexual organism morphisms across the genome, and violation hard to quantify the selective impact of import(the blue line) in which they are predicted to be of this may alter the tempo and mode of these ing any given gene(s) into a new background, held together by recombination. For sexual processes (34, 35). the occasional ability to gain a new adaptation species, the divergence of clusters requires a in this fashionsuch as a new metabolic capa-

33

the clear and ubiquitous signature of such events tionary fate of such genes may hence be only acteristics, including the extent of variation in The importance of this phenomenon is evident in between them by mobile elements. The evolu- different genera on the basis of their specific charin the growing body of genomic data. These are loosely coupled with that of any particular gene content and recombination. In any case, no the clear and ubiquitous signature of such events tionary fate of such genes may hence be only acteristics, including the extent of variation in identified by differences in the characteristics of species or strain in which they are found, and biologist would deny the importance of ecology in the growing body of genomic data. These are loosely coupled with that of any particular gene content and recombination. In any case, no are maintained selection by and the biologist to what would we observe, but it may not be easy to the acquired DNA that of in the host strain, for of they species or strain in through which they are found, deny the importance of ecology identified by and differences the characteristics incorporate it in a fashion that example, in base composition or cobut it may not be to is the acquired DNA andof that of the host strain, for they are maintained through selection by the to what we observe, bility or a new mode transticular species or strain ineasy which convenient for taxonomists. Nonedon usage; in most cases, the donor of incorporate it inand a fashion is example, in base composition or cothey arethat mainmission for a pathogenmay they are found, theless, population geneticists may the DNA in question is unknown. convenient for taxonomists. Nonedon usage; in most cases, the donor of be of enormous importance in tained through selection by the have little choice geneticists but to tackle Gene the acquisition to genomes theless, population may the DNA inleads question is unknown. terms of speciation. habitat to which each host strain is question defining being Gene punctuated by leads stretches of have little of choice but bacterial to tackle species the acquisition to genomes Perhaps even more striking adapted. In the case of very mobile question of defining bacterial species punctuated by of stretches of or, at the very least, populations. foreignbeing DNA. The largest these elementsfor example, plasmids is the amount of variation in or, at the very least, populations. DNA. Thekilobases largest ofin theseEcological Whether are estimating effective (whichforeign may be many content revealed by mulencoding we resistance to antibiotics gene Ecological factor b Whether we size are estimating effective may be termed many kilobases in population from neutral diversity length)(which were initially pathofactor b tiple genomes from the same or heavy metalsthe ecological population size from neutral diversity length) were initially termed pathoor choosing an appropriate set of genicity islands, because the that new which implies determined by these species, specificity or choosing anfor appropriate set of at genicity islands, because the new strains to test positive selection functions encoded by the importsat were gene acquisition occurs a accessory loci may have no link strains to of test for positive selection at functionsin encoded by the imports were a locus interest, species definitions often involved virulence, but a bethigh frequency. Ita betobserve using surprisingly the of sequence a to locus interest, we species definitions often involved in virulence, but are implicit ingenes much(Fig. of the ter term is genomic islands speak as the is now to housekeeping 4).analytical are implicit in much of the analytical ter termcommonplace is genomic islands as the toolkit of population genetics. phenomenon is far from limited to genome, which of the core toolkit of population genetics. phenomenon is far from limited to Distinguishing among mechapathogens (36, 37 ). Although it is hard encodes funcIdentifying Mechanisms and Distinguishing among mechapathogens (fundamental 36 , 37). Although it is hard nisms population differentiation to quantify the selective impact of imtions shared all members Delineating Species nisms ofof population differentiation to quantify the by selective impact of imin bacteria ultimately comes down portingporting any given gene(s) a new What do ultimately we want comes from bacterial of a species (and, gene(s) it into should go in bacteria down to to any given into a new Ecological factor Ecological factoraa species? Do we theoretical without saying, otherability related testing the ability of different models background, the occasional to to testing the ability ofneed different models background, the occasional ability 4. Differences between core and schematic to 4. Differences between core andauxiliary auxiliary genes. genes. This schematic consistency evenvariable atvariable the expense species), onto which is bolted to explain highly patterns gain a new adaptation in this fashion Fig.Fig. explain highly patterns gain a new adaptation in this fashion illustrates the relationships between three species in ecotype space, illustrates the relationships between three species in ecotype space, within of taxonomic practicality, incorpothe auxiliary or accessory within and between genetic-ecological such as a new metabolic capability and between genetic-ecological such as a new metabolic capability shown here in two dimensions, and mobilegene gene common common to here in two dimensions, and a amobile to all all three. three. clusters clonal sexual genome, composed of genes rating both (Fig. 1).1). It It isand still unclear or a new mode of transmission clusters (Fig. is still unclear or a new mode of transmission for afor ashown The areas occupied by the species are shown as solid in red, blue, The areas occupied by the species are shown as solid lines in red, blue, whether populations into a single theoand operons that may or may these patterns areare mainpathogen may of enormous whether these patterns mainpathogen may be ofbe enormous im- im- and green. The part of the ecological space where the shared mobile gene and green. The part of the ecological space where the shared mobile gene tained framework? One unifying present all isolates. It reticalby notin be gene flow or or selection, portance in terms of speciation. tained by gene flow selection, portance terms of in speciation. is selected in each species shown by dashedpurple purple line line and is selected in each species is is shown by aadashed and overlaps overlaps and theoretical concept is consider seems likely that more such striking auxiliawhat the effect of to population Perhaps even is the and what the effect of population Perhaps even more striking is the all three species ranges. Examples of (circular)genomes genomes from from each species all three species ranges. Examples of (circular) each species structure is. The joint distribution of of amount of variation in gene content with and without the purple mobile element are also illustrated. Note as the arena within which genes help to the species ryof structure is. The joint distribution amount variation indetermine gene content that with and without the purple mobile element are also illustrated. Note that genetic and ecological data cancan be be revealed by multiple genomes from individuals are similar enough, specific ecological properties for each species, the locus is not selected for all isolates, and its evolugenetic and ecological data revealed by multiple genomes from for each species, the locus is not selected for all isolates, and its evolu- used, as described above that for Vibrio the same species, For which implies that enough, indiorganism. example, or interbreed of the tionary fate is uncoupled from that of each host species, because if one used, as described above for Vibrio the same species, which implies that tionary fate is uncoupled from that of each host species, because if one species 13), to genes define compete populations gene acquisition at a sur- undergoes vidual (variant dia group of related occurs Leptospirila selective sweep or goes extinct, the mobile gene may be gene acquisition occurs at a sur- undergoes a selective sweep or goes extinct, the mobile gene may be species (13), to define populations making a strong success. theoretprisingly high frequency. It is now reintroduced from one of the other species. Examples of such distributed without reproductive been hypothrectly for lum has recently without making strong theoretprisingly high frequency. It of isthe now reintroduced from one of the determinants other species. such distributed ical commitment to a either of these commonplace to speak core loci include drug resistance inExamples pathogensof (e.g., b-lactamase Practical advances building on esized to adapt to different ical commitment to either of these commonplace to speak of thefundamental core loci genes) include drug resistance determinants in pathogens (e.g., b -lactamase alternatives. One clear result from genome, which encodes and heavy metal resistance in environmental organisms. These this or other theoretical concepts areas of an acid mine drainage alternatives. One clear result from genome, which encodes fundamental genes) and heavy metal resistance in environmental of the studies hereare is functions shared by allof members genes may be transferred among strains and species by organisms. conjugative These plas- all will only come discussed when these system by shuffling chro- of a all of the studies discussed here functions shared by all members of a genes may be transferred among strains and species by conjugative plasthat the underlying ques- is species it should go without mids or other mobile elements (including transducing phage). developed into theoretical explicit models enriched in mosome(and, segments that the underlying theoretical quesspecies (and, it should go without mids or other mobile elements (including transducing phage). tions concerning species will not be saying, other related species), onto and model-based algorithms that noncore genes (38, 39). We tions concerning species will not saying, other related species), onto answered in the absence of more detailed genetichabitat to which each host strain is adapted. In which is bolted the auxiliary or accessory are tested and refined on a wide be should, however, be aware mapping. Moreover, some guidethe case of very mobile elements for examcomposed of genes genes and operons answered in the absence of more detailed geneticto which each host strain is adapted. In environmental which genome, is bolted the or accessory that it may be sensible that changes inauxiliary core may also lead to habitat tic of different levels of ecological specificity, range of data. Alternatively, lines for the types of ecological studies that will ple, plasmids encoding resistance to antibiotics or may or may not be present in all isolates. It seems environmental mapping. Moreover, some guidethe case of very mobile elements for examgenome, composed of genes and operons that ecological differentiation, a phenomenon well ranging from highly conserved core functions to suggest an ad hoc application of principles be most informative are emerging. Most imporheavy metals the ecological specificity deterlikely that such auxiliary genes help to determine lines for the types of ecological studies that ple, plasmids encoding resistance to antibiotics or may or may not be present in all isolates. It seems documented in experimental studies of bacteria that are essential for growth in all environments to different genera on the basis of their specificwill the ecological data be imporrelmined by these accessory loci may have no link the specific ecological properties ofdetermine the organism. be most informative arecollected emerging. heavy metals the ecological specificity deterlikely that suchin auxiliary genes help to including the extentmust ofMost variation growing structured environments (40 ). to loci that are involved with adaptation to a tant, characteristics, evant to the niche boundaries of the populations to the sequence clusters we observe using houseFor example, a group of related Leptospirillum tant, the ecological data collected be relby these accessory mayniche-specific have no link in the specific ecological properties of the In must any case, Estimates vary, depending on organism. the genomes mined Some loci narrow gene content and recombination. specific habitat. studied. if genetic groups do map exgenesclusters (Fig. 4). has recently been of hypothesized to adapt to dif- keeping evant toAnd the niche boundaries of not the populations sequence we observe using houseFor example, a group related Leptospirillum but as little as 40% of genes to the being no biologist would deny the importance of ecolthat are available, genes may be distributed across species, clusively onto sampling categories (as is likely ferent areas of an acid mine drainage system by keeping genes (Fig. 4). studied. And geneticbut groups do not has recently hypothesized to adapt to difto what weif observe, it may not bemap easy exgenomes of a transferred between them by mobile elements. ogy may bebeen present in all sequenced to be the case), more complex statistical models shuffling of chromosome segments enriched in Identifying Mechanisms and clusively onto sampling categories (as is likely in a fashion that is convenient fate of such genes may hence to incorporate it ). We may consider genes Delineating The evolutionary ferent named areas ofspecies an acid(41 mine drainage system by Species will be needed to identify and describe the undernoncore genes (38, 39). We should, however, be Identifying Mechanisms and population gea named species as being characteristaxonomists. Nonetheless, be only loosely coupled with that of any par- for within to be the case), more complex statistical models shuffling of chromosome segments enriched in aware that changes in core genes may also lead to What do we want from bacterial species? Do we lying niche structure. Longitudinal studies that Delineating Species neticists may have little choice but to tackle will be needed to identify and describe the the undernoncore genes (38 , 39). We should, however, be the dynamics of ecological associations ecological differentiation, a phenomenon well need theoretical consistency even at the expense measure question of defining bacterial species or, thethat lying niche structure. Longitudinal studies aware documented that changes in core genes may also lead to What do we want from bacterial species? Do we 18. H. Ochman, C. Wilson, in Escherichia coli andat time willA. also be helpful to determine how in experimental studies of bacteria and ofvalidating taxonomic practicality, incorporating Table 1. A proposed strategy for developing models of bacterial evolution both that over populations. we are estimatvery least, Salmonella typhimurium :Whether Cellular and Molecular Biology measure the dynamics of ecological associations ecological differentiation, a phenomenon well diversity need theoretical even at into the expense might eventually be used to classify genetic data and provide a firm foundation for a transient natural habitats are, and thus how likely, growing in structured environments (40 ). clonal and consistency sexual populations a single F. C. Neidhart, Ed. (ASM Press, Washington, DC, diver1987), size from neutral ing effective population over time will also be helpful to determine how documented in species experimental studies ofthe bacteria taxonomic practicality, both bottlenecks bacterial concept. are to result. Finally, whole-genome Estimates vary, depending on genomes of theoretical framework? Oneincorporating unifying theoretical pp. 16491654. to sity or choosing an appropriate set thus of transient natural habitats are, and how likely growing inare structured environments (40 ). of genes clonal and sexual populations into a within single sequences from et entire populations of strains environthat available, but as little as 40% concept is to consider species as the arena 19. J. R. Thompson al., Appl. Environ. Microbiol. 70, 4103 1. Collect according to systematic stratification. Focus on longitudinal studies, for positive a locuswhole-genome of interest, (2004). bottlenecks are selection to result.at Finally, Estimates vary, samples depending on the genomesecological theoretical framework? One unifying theoretical test geographical studies, and measurement of physical and chemical gradients affecting bacterial 20. M. Kimura, Trends Biochem. Sci. 1, N152 (1976). definitions are implicit in much the sequences from entire populations of of environthat are available, but as little as 40% of genes concept is to consider species as the arena within species 21. R. Milkman, Trends Sci. 1genetics. , N152 (1976). growth. Consider biotic factors www.sciencemag.org such as the presence of other competing bacteria or323 parasitic 745 SCIENCE VOL 6 phage. FEBRUARY 2009 of Biochem. population analytical toolkit

Speciation

22. K. C. Atwood, L. K. Schneider, F. J. Ryan, Proc. Natl. 2. For each isolate, sequence as much as possible and affordable (16S rRNA, MLSA, auxiliary Distinguishing among mechanisms of popAcad. Sci. U.S.A. 37, 146 (1951). genes, full genomes, etc.). in(1981). bacteria ultimately ulation www.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 2009 23. B. R. differentiation Levin, Genetics 99, 1 3. Use empirical classification algorithms that use genetic and ecological data to jointly map isolates. 24. F. M. Cohan, to E. B.testing Perry, Curr. Biol. 17, R373 differcomes down the ability of(2007). 25. models A. Rambaut al., Nature 453, 615 (2008). patterns 4. To guide model formulation, use population genetic tests on observed clusters, focusing on ent toetexplain highly variable 26. R. A. Fisher, E. B. Ford, Heredity 1, 143 (1947). tests for selection, population structure, and gene flow. and between genetic-ecological clusters within 27. M. Slatkin, Theor. Popul. Biol. 12, 253 (1977). 5. Generate evolutionary models and simulate populations. unclear whether these (Fig. It is still 28. J.1). Majewski, F. M. Cohan, Genetics 152 , 1459patterns (1999). 6. Test, then reject or adapt, evolutionary models according to agreement between simulations 29. maintained E. J. Feil, B. G. Spratt, Annu. Rev. Microbiol. 55, 561 (2001). and are by gene flow or selection, 30. J. Majewski, F. M. Cohan, Genetics 153, 1525 (1999). and real populations; if necessary, return to step 1. what the effect of population structure is. The 31. J. Majewski, P. Zawadzki, P. Pickerill, F. M. Cohan, 7. For successful models, develop model-based methods for interpreting pure genetic data jointC. distribution of genetic ecological data G. Dowson, J. Bacteriol. 182and , 1016 (2000). (without ecological covariates) and test on new data. as above for Vibrio specan 32. be S. T.used, Cowan, in described Microbial Classification, 12th Symposium 8. If one or more validated models emerge, use these to classify genetic data and to develop the for General Microbiology , G. C. Ainsworth, without making cies of (13 ),Society to define populations bacterial species concepts. P. H. A. Sneath, Eds. (Cambridge Univ. Press, Cambridge, a strong theoretical commitment to either of 1962), pp. 433455.

these alternatives. One clear result from all of the studies discussed here is that the underlying theoretical questions concerning species will not be answered in the absence of more detailed genetic-environmental mapping. Moreover, some guidelines for the types of ecological studies that will be most informative are emerging. Most important, the ecological data collected must be relevant to the niche boundaries of the populations studied. And if genetic groups do not map exclusively onto sampling categories (as is likely to be the case), more complex statistical models will be needed to identify and describe the underlying niche structure. Longitudinal studies that measure the dynamics of ecological associations over time will also be helpful to determine how transient natural habitats are, and thus how likely bottlenecks are to result. Finally, whole-genome sequences from entire populations of environmental bacteria will be useful in dissecting the roles of the auxiliary and core genome in ecological differentiation. If after this process it emerges that some model or models are consistently validated for different study systems, these would inevitably form a good basis for identifying fundamental levels of clustering, or species. In the foregoing we have emphasized ecotype and metapopulation models, but there are others that deserve considerationnotably the epidemic clonal model (42) and the impact of phage epidemics causing classic Lotka-Volterra boom-bust dynamics (43) illustrated in Fig. 2Dand it is possible, even likely, that more than one of these mechanisms may be relevant to any given problem in speciation and cluster formation. Distinguishing among these mechanisms is the bacterial species challenge (Table 1), described in 1991 by John Maynard Smith as follows: Ecotypic structure, hitch-hiking, and localized recombination can explain the observed patterns of variation. The difficulty, of course, is that the model is sufficiently flexible to explain almost anything. To test the hypothesis of ecotypic structure, we need to know the distribution of electrophoretic types [i.e., genotypes] in different habitats (17). Much research on bacterial species to date has come from studies on pathogens, where the correct identification of species is crucial for accurate clinical diagnoses. However, for pathogens the identification of the multiple ecological

niches within (for example) the nasopharynx or gut is difficult, and studies of the relationships between bacterial populations and ecology may be more fruitful for some environmental species where the categorization of niches is a more tractable enterprise. Hopefully, we will soon obtain richer data sets that map bacterial diversity onto ecology and provide a way to distinguish among various models of population differentiation and speciation, including those based on ecotypes or metapopulations.

1. J. Handelsman, Microbiol. Mol. Biol. Rev. 68, 669 (2004). 2. W. P. Hanage, C. Fraser, B. G. Spratt, Philos. Trans. R. Soc. London Ser. B 361, 1917 (2006). 3. M. Achtman, M. Wagner, Nat. Rev. Microbiol. 6, 431 (2008). 4. F. M. Cohan, Annu. Rev. Microbiol. 56, 457 (2002). 5. W. F. Doolittle, R. T. Papke, Genome Biol. 7, 116 (2006). 6. D. Gevers et al., Nat. Rev. Microbiol. 3, 733 (2005). 7. M. F. Polz, D. E. Hunt, S. P. Preheim, D. M. Weinreich, Philos. Trans. R. Soc. London Ser. B 361, 2009 (2006). 8. D. B. Rusch et al., PLoS Biol. 5, e77 (2007). 9. C. Darwin, The Origin of Species (Penguin Classics, London, 1985). 10. E. Stackebrandt et al., Int. J. Syst. Evol. Microbiol. 52, 1043 (2002). 11. R. Facklam, Clin. Microbiol. Rev. 15, 613 (2002). 12. J. C. Arbique et al., J. Clin. Microbiol. 42, 4686 (2004). 13. D. E. Hunt et al., Science 320, 1081 (2008). 14. See supplementary information in (16) for a statistical comparison of the ecotype model with an effective neutral model and an implicit estimate of Ne. 15. C. Fraser, W. P. Hanage, B. G. Spratt, Science 315, 476 (2007). 16. A. Koeppel et al., Proc. Natl. Acad. Sci. U.S.A. 105, 2504 (2008). 17. J. Maynard Smith, Proc. R. Soc. London Ser. B 245, 37 (1991). 18. H. Ochman, A. C. Wilson, in Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology, F. C. Neidhart, Ed. (ASM Press, Washington, DC, 1987), pp. 16491654. 19. J. R. Thompson et al., Appl. Environ. Microbiol. 70, 4103 (2004). 20. M. Kimura, Trends Biochem. Sci. 1, N152 (1976).

References and Notes

21. R. Milkman, Trends Biochem. Sci. 1, N152 (1976). 22. K. C. Atwood, L. K. Schneider, F. J. Ryan, Proc. Natl. Acad. Sci. U.S.A. 37, 146 (1951). 23. B. R. Levin, Genetics 99, 1 (1981). 24. F. M. Cohan, E. B. Perry, Curr. Biol. 17, R373 (2007). 25. A. Rambaut et al., Nature 453, 615 (2008). 26. R. A. Fisher, E. B. Ford, Heredity 1, 143 (1947). 27. M. Slatkin, Theor. Popul. Biol. 12, 253 (1977). 28. J. Majewski, F. M. Cohan, Genetics 152, 1459 (1999). 29. E. J. Feil, B. G. Spratt, Annu. Rev. Microbiol. 55, 561 (2001). 30. J. Majewski, F. M. Cohan, Genetics 153, 1525 (1999). 31. J. Majewski, P. Zawadzki, P. Pickerill, F. M. Cohan, C. G. Dowson, J. Bacteriol. 182, 1016 (2000). 32. S. T. Cowan, in Microbial Classification, 12th Symposium of the Society for General Microbiology, G. C. Ainsworth, P. H. A. Sneath, Eds. (Cambridge Univ. Press, Cambridge, 1962), pp. 433455. 33. S. K. Sheppard, N. D. McCarthy, D. Falush, M. C. J. Maiden, Science 320, 237 (2008). 34. A. C. Retchless, J. G. Lawrence, Science 317, 1093 (2007). 35. K. Vetsigian, N. Goldenfeld, Proc. Natl. Acad. Sci. U.S.A. 102, 7332 (2005). 36. A. Tuanyok et al., BMC Genomics 9, 566 (2008). 37. U. Dobrindt, B. Hochhut, U. Hentschel, J. Hacker, Nat. Rev. Microbiol. 2, 414 (2004). 38. P. Wilmes, S. L. Simmons, V. J. Denef, J. F. Banfield, FEMS Microbiol. Rev. 33, 109 (2009). 39. V. J. Denef et al., Environ. Microbiol. 11, 313 (2008). 40. E. Bantinaki et al., Genetics 176, 441 (2007). 41. R. A. Welch et al., Proc. Natl. Acad. Sci. U.S.A. 99, 17020 (2002). 42. J. Maynard Smith, N. H. Smith, M. ORourke, B. G. Spratt, Proc. Natl. Acad. Sci. U.S.A. 90, 4384 (1993). 43. K. H. Hoffmann et al., FEMS Microbiol. Lett. 273, 224 (2007). 44. We thank T. Connor and S. Deeny for useful discussions. Supported by University Research Fellowships from the Royal Society (C.F. and W.P.H.), a program grant from the Wellcome Trust (B.G.S.), grants from the U.S. Department of Energy Genomes to Life program (M.F.P. and E.J.A.), and the NSF/ National Institute of Environmental Health Sciences Woods Hole Centre for Oceans and Human Health, the NSF Biological Oceanography Program, and the Moore Foundation (M.F.P.).

745

34 mental bacteria will be useful in dissecting the

roles of the auxiliary and core genome in

References and Notes

1. J. Handelsman, Microbiol. Mol. Biol. Rev. 68, 669 (2004).

33. S. K. Sheppard, N. D. McCarthy, D. Falush, M. C. J. Maiden, Science 320, 237 (2008). 34. A. C. Retchless, J. G. Lawrence, Science 317, 1093 (2007).

35

is admitted and accepted as a member of a host society, it mimics adult ant acoustics (particularly queens) to advance its seniority toward the

s hiermimicry lack the and of. species li (12): studied nsals or with ants e acouslthough s attracts s to siga basal t expect nea. ecies of icularly Diptera, ays the nts hicue has ate and

Berlin,

nu. Rev.

9. J. A. Thomas, G. W. Elmes, J. C. Wardlaw, Proc. R. Soc. in persistence hence diversity, they lead to tropical biomesand have usefully employed post hoc London Ser. B 265, 1895 (1998). 10. G. W. Elmes, J. C. Wardlaw, K. Schnrogge, J. A. Thomas, phylogeographic predictions for both individual palaeoclimate models of species and habitats to Entomol. Exp. Appl. 110, 53 (2004). (codistributed taxa; species insights and assemblages provide about processes shaping genetic 11. Materials and methods are available as supporting driven by the model Fig. 1). Field sampling is and species diversity ( 5 , 7 ). Building on them, we material on Science Online. first map the palaeodistribution endemic and predictions to cover both predictedof refugia 12. P. J. DeVries, R. B. Cocroft, J. A. Thomas, Biol. J. Linn. Soc. 49, 229 (1993). species identify colonized) temporallyareas, stable (refugial) unstableto (recently particularly 1 2 Ana Carolina Carnaval, Michael Hickerson, Clio F. B. Haddad,3 13. F. Roces, J. Tautz, J. Acoust.*Soc. Am. 109J. , 3080 (2001). and unstable previously (recently colonized) regions for undersampled areas. If emphasizing 4 1 Miguel T. Rodrigues, Moritz 14. R. Hickling, R. L. Brown, Craig J. Acoust. Soc. Am. 108, 1920 species occurrence, which are then validated with current patterns the approach correctly predicts (2000). multispecies molecular Going beyond the of biodiversity at the data. regional scale, species 15. H. Markl, B. Hlldobler, Behav. Ecol. Sociobiol. 4, 183 Biodiversity hotspots, representing regions with high species endemism and conservation threat, (1978). traditional species-by-species the moshould consistently show (i) approach, higher genetic dihave been globally. Yet, biodiversity distribution data from within hotspots are too sparse 16. H. Markl,mapped Z. Vgl. Physiol. 60, 103 (1968). lecular analyses contrast the fit of assemblageand among populations in refugia versity within 17. effective H. Markl, Science 149, 1392in (1965). for conservation the face of rapid environmental change. Using frogs as indicators, level data to the spatially explicit demographic 18. D. A. Grasso, A. models Mori, F. Le Moli, M. Giovannotti, ecological niche under paleoclimates, and simultaneous Bayesian analyses of multispecies relative to unstable areas, because of long-term scenarios suggested by the climate-based A. Fanfani, Ital. J. Zool. 65, 167 (1998). genetic persistence and population structure; (ii)models. molecular data, we compare alternative hypotheses of assemblage-scale response to late 19. T. C. Scott-Phillips, J. Evol. Biol. 21, 387 (2008). We apply this approach to one of in theunstable worlds signature of population expansion Quaternary climate change. This reveals a Apollo hotspot within the Brazilian Atlantic forest hotspot. 20. K. G. Schurian, K. Fiedler, Nachr. Entomol. Vereins most species-rich, yet notoriously endangered and We show that the southern Atlantic forest was climatically unstable relative to the central region, areas, reflecting multispecies colonization from 14, 339 (1994). understudied ecosystems: the Brazilian Atlantic 21. C. J. Hill, J. Aust. Entomol. Soc. 32 , 283 (1993). which served as a large climatic refugium for neotropical species in the late Pleistocene. This sets adjacent refugial regions after the Last Glacial 22. D. R. Nash, T. D. Als, R. Maile, G. R. Jones, J. J. Boomsma, Maximum (LGM, 21 kybp); (iii) absence of new priorities for conservation in Brazil and establishes a validated approach to biodiversity 1 Science 319, 88 (2008). Museum patterns of Vertebrate Zoology, University of California, genetic of isolation-by-distance in unprediction in other understudied, species-rich regions. 2 23. P. J. DeVries, Am. Mus. Nov. 3025, 1 (1991). Berkeley, CA 94720 3160, Biology Department, areas, given that USA. colonization has been stable 24. K. Fiedler, B. Hlldobler, P. Seufert, Experientia 52, 14 Queens College, City University of New York, Flushing, NY 3 permit restoration of equilibrium too recent to (1996). Quaternary fluctuations refugia models have been dismissed because of 11367, USA. Departamento de Zoologia, Instituto de 25. M. ate A. Travassos, N. E. climate Pierce,climate Anim. Behav. 60, helped 13 being described (8, 9). Our ultimate goal is to Biocincias, ate Quaternary fluctuations and between migration genetic drift (15); UNESP, Rioand Claro, SP 3526-4100, Brazil. to shape present-day diversity in temper- conflicting evidence (2, 3) or circularity in iden- 4Departamento de Zoologia, Instituto de Biocincias, (2000). diversity in pinpoint regions for inventory work and habitat (iv) strong phylogeographic structure between helpedetto shape present-day 26. N. ate E. Pierce al ., Annu. Rev. Entomol. 47 , 733 (2002). and boreal systems (1), providing a tifying putative refugia (4), but historical pro- Universidade de So Paulo, SP 055008-090, Brazil. and boreal systems (1 ),, pro- protection before we lose a substantial fraction refugia, reflecting assemblage-wide, long-term 27. J. C.temperate Downey, C. Allyn, Bull. Mus. Entomol. 14 general contextA.for understanding current pat- cesses must be invoked to explain regions of whom correspondence be addressed. 1 (1973). in isolated areas. E-mail: population persistence should viding a general context for understanding cur- of described and undocumented diversity. The *To terns of endemism. In the tropics, Pleistocene high endemism (5, 6). Recent studies from sub- carnaval@berkeley.edu 28. We thank N. Elfferich and P. J. DeVries for introducing Pleis- approach differs from previous methods by diDistribution models developed under current rent patterns of endemism. In the tropics, us to ant-butterfly acoustics; G. W. Elmes, J. C. Wardlaw, tocene refugia models have been dismissed be- rectly modeling historical processes, as opposed climatic conditions accurately predict distribuV. La Morgia, M. B. Bonsall, and referees for of conflicting evidence (2, 3)for ordesigning circularity to observed biodiversity (10), with the tions of each of the target species along the At- 785 cause www.sciencemag.org SCIENCE patterns VOL 323 6 FEBRUARY 2009 comments and advice; and M. Charles the acoustical equipment. putative refugia (4), but historical aim of informing conservation. in identifying lantic rainforest domain [area-under-the-curve

Hypothesis Formulation

Stability Predicts Genetic Diversity in the Brazilian Atlantic Forest Hotspot

Diversity Distribution Modeling Diversity Model Validation

REPORTS

7. K. Schnrogge et al., J. Chem. Ecol. 30, 91 (2004). 8. T. Akino, J. J. Knapp, J. A. Thomas, G. W. Elmes, Proc. R. Soc. London Ser. B 266, 1419 (1999).

Audio S1 to S4 22 July 2008; accepted 28 November 2008 10.1126/science.1163583

L L

9). nndez,

4). nsect y, J. Rolff, 18. 004). , Proc. R.

in ot

hreat, sparse ators, species

pot. region, his sets ty

processes be invoked to explain regions Supporting must Online Material www.sciencemag.org/cgi/content/full/323/5915/782/DC1 of high endemism (5, 6). Recent studies from Materials and Methods subtropical biomes have usefully employed post SOM Text models of species and habihoc palaeoclimate Figs. S1 and S2 Table to S1 provide insights about processes shaping tats References and species diversity (5, 7). Building genetic Audio S1 to S4 on them, we first map the palaeodistribution of 22 July 2008; accepted to 28 identify November 2008 temporally stable endemic species 10.1126/science.1163583 (refugial) and unstable (recently colonized) regions for species occurrence, which are then molecular data. validated with multispecies tropical biomes have usefully employed post Gohoc ing beyond the traditional species-by-species palaeoclimate models of species and habitats to approach, the molecular analyses contrast the provide insights about processes shaping genetic data to the spatially exfit of assemblage-level and species diversity (5, 7). Building on them, we by the plicit demographic scenarios suggested first map the palaeodistribution of endemic climate-based models. species to identify temporally stable (refugial) apply this approach to one of regions the worlds andWe unstable (recently colonized) for yet notoriously endangered most species-rich, species occurrence, which are then validated with Brazilian and understudied ecosystems: the multispecies molecular data. Going beyond the Atlantic Originally extending for traditionalrainforest. species-by-species approach, the mokm2 along the Brazilian coast and 1,300,000 lecular analyses contrast the fit of assemblagethis biome reaching into and Argentina, level data to Paraguay the spatially explicit demographic (8). has been reduced to by less than 8% of its range scenarios suggested the climate-based models. Todays fragments harbor one of the perWe apply this approach to one of largest the world s of endemic species in the world, with centages most species-rich, yet notoriously endangered and understudied the of Brazilian Atlantic vertebrates still many species ecosystems: and even genera
1 Museum of Vertebrate Zoology, University of California, Berkeley, CA 947203160, USA. 2Biology Department, Queens College, City University of New York, Flushing, NY 11367, USA. 3Departamento de Zoologia, Instituto de Biocincias, UNESP, Rio Claro, SP 3526-4100, Brazil. 4 Departamento de Zoologia, Instituto de Biocincias, Universidade de So Paulo, SP 055008-090, Brazil.

ause of n idenal proions of m sub-

*To whom correspondence should be addressed. E-mail: carnaval@berkeley.edu

We use molecular genetic data from multiple, largely codistributed species to test whether spatial modeling of species-specific Late Quaternary refugia sheds light on historical processes and hence improves prediction of genetic endemism and diversity in tropical Brazil (11). We focus on three common species of tree frogs that are widely distributed along the Brazilian Atlantic forest: Hypsiboas albomarginatus, H. semilineatus, and H. faber. Given their life history traits, amphibians are useful indicators of environmental changes through time (12). Whereas H. albomarginatus and H. semilineatus occur in low and mid altitudes and are mostly restricted to the evergreen or semideciduous components of the Atlantic Forest in eastern Brazil, H. faber has a broader altitudinal range and also inhabits mixed and deciduous areas, occupying interior and coastal sites in the Atlantic Forest south to Paraguay and Argentina (figs. S1 and S2) (13). The comparative phylogeographic approach is a powerful test of assemblage-scale responses to former environmental change and thereby provides a means for critical assessment of the scenarios produced by modeling of species distributions under palaeoclimates (7). The palaeomodeling method intersects predicted species distributions under current conditions and climatic extremes of the Late Quaternary (6000 years before present, or 6 kybp, and 21 kybp) to predict areas of stability (regions in which species are predicted to occupy irrespective of time period) and unstable areas (7, 14). Because the stability maps raise specific hypotheses about regional differences

(AUC) values (16) 0.968, 0.989, and 0.994; maximum Kappa (17) 0.81, 0.925, and 0.94 in H. albomarginatus, H. faber, and H. semilineatus, respectively (fig. S2)]. Stability maps, depicting the intersection of distribution models for each taxon under current, 6 kybp, and 21 kybp climates, predict for all species a large central refugium throughout the Late Quaternary (Bahia refugium) (Fig. 2). A second, much smaller refugium is predicted in the northeasternmost portion of the forest (Pernambuco refugium). In H. faber, a third, southeastern refugium of intermediate size is also predicted (So Paulo refugium). This is not surprising, given that this species occupies a broader environmental niche. In contrast to the central and northern regions, populations south of the Bahia or So Paulo refugia appear much less stable, despite the more extensive (preclearing) range of the forest in southern and southeastern Brazil. We hypothesize that these areas received a significant influx of migrants from adjacent, large refugial populations after the LGM. These palaeomodel results are congruent with the fossil pollen record, which documents a replacement of forests by grasslands in the southern Atlantic forest during the LGM (14, 18) and suggests the occurrence of small forest refugia in the southernmost range of the putative Bahia refugium (19). The results also agree generally with forest models published previously (14), although the central refugium extends farther south in the frog-based models. Such differences are expected because the forest and its associated species may differ slightly in their climatic tolerances and realized

is higherlarger thanin the because of the small in the southernmost range by of magnitude critical assessment of the scenarios produced by test the other central species (Bahia) refugium which refugia documents a replacement of forests phylogeographic approach is a powerful of forest modeling of species distributions under assemblage-scale responses to palaeoformer environ- grasslands in the southern Atlantic forest during relative to the less stable (southern) portion of the climates (7). mental change and thereby provides a means for the LGM (14, 18) and suggests the occurrence of forest. Diversity of H. faber in this southern area species because of the the southernmost range of is higher than the other critical assessment of the scenarios preproduced by small forest refugia inSpecies The palaeomodeling method intersects Map of predicted Current & of species distributions under palaeooccurrence stable areas historical climate dicted speciesmodeling distributions under current condiclimates (7). data Inclimatic H. albomarginatus and H. Quaternary faber, the niches. (putative refugia) data tions and extremes of the Late palaeomodeling method intersects preSpecies Map of predicted Current & So Paulo refugium extension of the The predicted (6000 years before present, or 6 kybp, and 21 kybp) occurrence stable areas historical climate dicted species distributions under current condiwestward the neighboring Cerrado data (putative refugia) data to predict into areas of stability (regions which tions and climatic extremes of in thebiome Late Quaternary reflects overprediction (fig.irrespective S2) (14 ). and (6000 years present, or 6 kybp, species model are predicted tobefore occupy of21 kybp) Spatially explicit hypotheses Re: to predict areas ofthrough stability (regions Models ofand habitat stability -in which time period) unstable areas (7, 14). fluctuat Because species are predicted to occupy irrespective of predict patterns of phyloing climatesmaps correctly Spatially explicit hypotheses Re: the stability raise specific hypotheses about and unstable areas (7, 14). Because rainforest geography in time the period) Brazilian Atlantic regional differences in persistence and hence dithe stability maps raise specific hypotheses about spatial-temporal patterns (Fig. 2 and S3 to S5). In all species, high distribution of congruence versity, theyfigs. lead to phylogeographic predictions regional differences in persistence and hence diof colonization genetic diversity across taxa spatial-temporal patterns levels of individual divergence and population structure are for both species and assemblages (codistribution of congruence versity, they lead to phylogeographic predictions and/or vicariance of colonization genetic diversity across taxa forFig. both1). individual species and distributed taxa; Field sampling isassemblages driven (corefugia (Tamura-Nei corrected observed across and/or vicariance distributed taxa; Fig. 1). Field sampling is driven by the model predictions to coverBahia both predicted and Perdistances (20): 4 to 7% between by the model predictions to coverareas, both predicted Sampling across predicted stable and unstable areas refugia and unstable colonized) nambuco refugia, 1% (recently between the nearby Bahia Sampling across predicted stable and unstable areas refugia and unstable (recently colonized) areas, particularly emphasizing previously undersamSo Paulo refugia in H. faber ). Similarly, in and particularly emphasizing previously undersampled areas. Ifare the approach correctly predicts curthere multiple, divergent clades withall taxa pled areas. If the approach correctly predicts current patterns of biodiversity at the regional scale, rent patterns of biodiversity at the regional in the Bahia region, agreeing with model-based scale, Genetic tests of of Assemblage-scale Genetic tests Assemblage-scale should consistently show (i) higher species should consistently show (i) genetic Descriptive Descriptive this area. In genetic predictions of species a large refugium inhigher stability/expansion, hypothesis testingtesting stability/expansion, hypothesis phylogeography diversity within populations and among populations in refugia phylogeography diversity within and among in refugia divergence times (HABC) H. faber, divergent clades are also represented divergence times (HABC) relative to unstable areas, because of long-term perrelative to unstable areas, because of long-term perin the So Paulo region, of sistence andmatching population predictions structure; (ii) genetic sigsistence andrefugium population structure; (ii) sigin this area. Allgenetic taxa a mid-sized nature of population expansion in show unstable areas, Fig. 1. Proposed method of biodiversity prediction. Three stages are involved: biodiversity disnature of population expansion incolonization unstable areas, Fig. 1. Proposed method of model-based biodiversityhypothesis prediction. Three stages arehypothesis involved: biodiversity disreflecting multispecies from adjacent tribution modeling (top), formulation (middle), testing and southernmost low genetic diversity across the reflecting multispecies colonization from adjacent tribution modeling (top), model-based hypothesis formulation (middle), hypothesis testing and refugial regions after the Last Glacial Maximum model validation (bottom). range of the forest, an area predicted to be less refugial regions after the Last Glacial Maximum stable by the palaeomodels. Furthermore, mito- model validation (bottom). DNA (mtDNA) lineages found in this chondrial786 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org refugia (one in in and E). Using Bayes factor (25), we also detect scale, long-term persistence of populations region are shared with adjacent 786 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org H. albomarginatus and H. semilineatus, two in isolated refugial areas, as opposed to post-LGM evidence for stability in both areas under the noH. faber). migration model [B(Z2 = 0, Z2 > 0) = 4.89], as colonization of refugial regions. Metrics of genetic diversity confirm the To test for assemblage-wide colonization well as under a postisolation migration model above patterns (Table 1). In H. albomarginatus of predicted unstable areas, we group mtDNA [B(Z2 = 0, Z2 > 0) = 4.84]. Relative to nuclear loci, mtDNA data are and H. semilineatus, genetic diversity (21) is an sequences from the southernmost refugial sites order of magnitude larger in the central (Bahia) [population 1 (Fig. 3A)] and from localities in more variable and readily collected and often refugium relative to the less stable (southern) unstable areas south of the refugium [population provide key insights into biological response portion of the forest. Diversity of H. faber in 2 (Fig. 3A)] to contrast two alternative historical to environmental modification (1). Although this southern area is higher than the other spe- models across the three codistributed species, single-locus inference can be imprecise in the cies because of the presence of two lineages that while allowing the taxon-specific demographic face of coalescent variance and the possibility of co-occur in the adjacent refugia. In all species, parameters to vary. In H1, the long-term persis- selection (26), our method benefits from a mulaverage net nucleotide differences across locali- tence model, two contemporary populations split titaxon approach, while explicitly accounting ties (22) reflects high geographic structure with- from an ancestral population prior to the LGM for the stochasticity of a single-locus coalescent in refugia (2.6 to 6.2% divergence). In contrast, (120,000 to 1.2 million years before present, or across taxa. Combining data sets from several sites located outside (south of) the refugia are Mybp, Fig. 3A). In H2, the recent colonization codistributed groups into a single hierarchical genetically more similar to each other, although model, population 2 is modeled as being colo- Bayesian analysis allowed us to estimate conto a lesser extent in H. faber (0.1 to 1.6%). Sig- nized from refugial population 1 subsequent to gruence across species, while borrowing strength natures of population expansion (23) are found the LGM (0 to 20 kybp; Fig. 3A). The results from the full comparative phylogeographic samin the unstable area for H. albomarginatus and indicate that all three species colonized the ple (24). This can translate into higher analytical H. faber, as well as in the Bahia refugium area southern (unstable) areas after the LGM (Z2 = 3, power and be more informative than qualitative for H. faber and H. semilineatus. The lack of the number of species evolved under H2), even comparisons of species-specific analyses. By signature of population expansion in the south- when allowing for postisolation migration (Fig. capturing the historical signal that emerges from ernmost localities of H. semilineatus may reflect 3, B and C). When Bayes factor is used (25), larger, combined multispecies molecular data low statistical power because of the exception- there is strong support for recent colonization in sets, HABC will offer the possibility of looking ally low levels of diversity observed in this spe- all three species (Z2 = 3) under the no-migration at patterns of historical community assembly in cies. As predicted, isolation by distance is not model [B(Z2 = 3, Z2 < 3) = 35.16], and moderate codistributed nonmodel organisms for which observed in unstable regions, but is detected support under a postisolation migration model barcode-type DNA sequence information (e.g., within refugial areas for H. albomarginatus and [B(Z2 = 3, Z2 < 3) = 5.70]. mtDNA data) can be feasibly collected. Using the same framework to test for longCollectively, the results identify the central H. faber. The hierarchical approximate Bayesian com- term persistence of refugial populations, we region as a hotspot within the Atlantic rainforest putation (HABC) method (24) allows us to use compare mtDNA sequences between the pre- hotspot and a refuge for biodiversity during data from all three species at once to test for dicted Pernambuco refugium [population 1 (Fig. climatic extremes of the Late Pleistocene. assemblage-wide responses to Late Quaternary 3A)] and adjacent (northern) populations from This is not to say that southern areas entirely climate change. These analyses support both the Bahia refugium [population 2 (Fig. 3A)] to lacked forested habitats in the late Pleistocene: model-driven hypotheses of (i) simultaneous, contrast alternative historical models H1 and H2. The existence of species and genera endemic multispecies colonization of unstable areas from In this case, the HABC results infer long-term to the southern forests (27), as well as some adjacent refugial populations since the LGM, persistence of populations in isolated refugia for palaeoecological and genetic evidence (28), as opposed to long-term persistence of popu- all three species (Z2 = 0, i.e., Z1 = 3), even when offer evidence to the contrary. Rather, the lations in unstable areas, and (ii) assemblage- allowing for postisolation migration (Fig. 3, D phylogeographically validated palaeomodels
Model Validation Hypothesis Formulation Distribution Modeling

6 FEBRUARY 2009

36

785

37

refugial (stable) versus unstable areas unstable area H. albomarginatus and (south refugiaare aregenetically geneticallyunstable jacent refugia. all species, average net nuclearea for H. for albomarginatus and H. faber , H. faber, outside (south of) of) the the refugia jacent In refugia. In all species, average net nucle- outside in thedifferences Brazilian Atlantic rainforest. as well as in the refugium Bahia refugium area for H. faber similar although a lesseras well otide across across localities (22) (reflects as in the Bahia area for H. faber more similarto toeach each other, other, although to to a lesser otide differences localities 22) reflects more ( Top)geographic Species-specific stability maps; H. semilineatus . The lack of lack signature of extent H.faber faber (0.1 (0.1 to Signatures of ofand and high geographic structure within refugia H. semilineatus . The of signature of inin H. to 1.6%). 1.6%). Signatures high structure within refugia (2.6 (2.6 to to extent Pernambuco modeled refugia in black. (A) H. refugium albomarginatus, (B) H. semilineatus, Fig. 2. diversity Genetic diversity in putative C Fig. Genetic in putative A B ( C) H.2. faber . Note the absence of large C B refugial (stable) versus unstable areas A Bahia refugium refugial (stable) unstable areas stable regions inversus the southern portion in the Brazilian Atlantic rainforest. * in the the forest Brazilian Atlantic rainforest. * of of the Bahia andmaps; (Top(south ) Species-specific stability (Top)Paulo Species-specific stability maps; Pernambuco modeled refugia in black. (A) H. So refugia) relative to the refugium Pernambuco albomarginatus , (B) H. , modeled refugia in areas. black. (semilineatus A) H. central and northern Asterisks S o Paulo refugium refugium (C) H. faber . Note the absence of albomarginatus , inferred (B) H. semilineatus , large denote refugia beyond the Bahia refugium stable regions in the southern portion ( C) H. faber . Note the absence of large current ranges of the target species. 5.4% * * of the forest (south of the Bahia and Bahia refugium stable regions in the southern portion Symbols indicate localities sampled for So Paulo refugia) relative to the * * of the forest (south of the Bahia and molecularcentral analysis. Scale bar, areas. 400 km. and northern Asterisks S o Paulo refugium So Paulo refugia) relative the the (Bottom) The 50% majority-rule condenote refugia inferredto beyond central and northern areas. Asterisks current ranges of the target species. sensus Bayesian phylogenetic trees, 5.4% 7% S o Paulo refugium 7.8% Symbols indicatefrom localities denote with refugia inferred beyond the for rooted sequences thesampled othmolecular Scale bar, 400 km. current of analysis. the target species. er two ranges congeneric species studied 5.4% (Bottom) The 50%sampled majority-rule Symbols localities for con(root notindicate shown). Thick internodes de5.3 sensus Bayesian phylogenetic trees, 7% molecular analysis. Scale bar, 400 km. note clades withwith posterior probability 7.8% 4% 5.8% rooted sequences from the oth(Bottom) The 50% majority-rule congreater than Percentages indicate er 90%. two congeneric species studied sensus Bayesian trees, deTamura-Nei corrected distances (root not phylogenetic shown). Thick between internodes 7% 5.3 7.8% rooted(with the probability othclades withfrom posterior 4% clades 20note ). sequences 5.8% 5.6% greater than 90%. Percentages indicate er two congeneric species studied Tamura-Nei corrected distances between (root not shown). Thick internodes de5.3 clades (20 ). note clades with posterior probability 4% 5.8% 5.6% greater than 90%. Percentages indicate Tamura-Nei corrected distances between clades (20).
5.6%

presence of two lineages that co-occur in the adA

6.2% divergence). In contrast, B sites located

population expansion (23) are found in the C

Mybp, Fig. 3A). In H2, the recent colonization model, population 2 is modeled as being colonized from refugial population 1 subsequent to

coalescent variance and the possibility of selection (26), our method benefits from a multitaxon approach, while explicitly accounting for the

southern regions. This reassures us that the processes uncovered by the amphibian data may be generalized to and help to explain patterns of

Fig. 3. HABC analyses. (A) Simulated models H1 (long-term persistence) and H2 (recent colonization). In both cases, each species was modeled as two contemporary populations with mutation-drift parameters q1 and q2 that split from an ancestral population at a time t in the past. Ancestral population sizes are represented by (qt)1 and (qt)2; ybp, years before present. (B to E) Hyperposterior (bars) and hyperprior (dashed) densities of Z2 (number of species evolved under H2) given data from three codistributed frog species. (B) and (C) Models of refugial sites (population 1) and unstable, southern areas (population 2). (D) and (E) Models of Pernambuco refugium (population 1) and Bahia refugium (population 2). (B) and (D) Postisolation migration not included in model; (C) and (E) postisolation migration included in model.

788 model-based demographic hypotheses 6 FEBRUARY 2009 VOL 323 SCIENCE www.sciencemag.org 32. We thank U. Caramaschi and H. Zaher for 2008 (American Museum of Natural History, and
Table 1. Population genetic summary metrics used in model validation. n, Table 1. Population genetic summary metrics used in model validation. n, Sample size; S, number of segregating sites. sites. The diversity parameter qq and Sample size; S, number of segregating The diversity parameter andmean mean Da across D localities are given per base pair (bp). Hs test (23) is used to detect a across localities are given per base pair (bp). Hs test (23) is used to detect population expansion. BA, Bahia; SP, So refugia. Because population expansion. BA, Bahia; SP, Paulo So Paulo refugia. Becausepredicted predicted refugia were often larger than predicted unstable (recently colonized) areas,n n S, refugia were often larger than predicted unstable (recently colonized) areas, ,, S
SS n, Table 1. Population genetic summary metrics usednin n model validation. Species Species Area Area (min.; max.) (min.; max.) Sample size; S, number of segregating sites. The diversity parameter q(min.; and mean (min.; max.) max.)
albomarginatus (BA) (bp). Hs test 36 Da across H. localities are given per Stable base pair (23) is used to 207 detect H. albomarginatus Stable (BA) 36 207 (970 bp) (13; 23) (81; 155) population expansion. BA, Bahia; SP, So Paulo refugia. Because predicted (970 bp) (13; 23) (81; 155) Unstable 22 refugia were often larger than predicted unstable (recently27 colonized) areas, n, S,
a of the former were obtained not only from the total q, and average Da values numberofof samples, all possible combinations of spatially number samples, but but also also from from all possible combinations of spatially contiguous localities distributed within the geographic extension of the unstable contiguous localities distributed within the geographic extension of the unstable area. Parentheses encompass minimum and maximum values from subsamples. area. Parentheses encompass minimum and maximum values from subsamples. Pvalues values bold highlight statistical significance 0.05 probability level. P inin bold highlight statistical significance at 0.05 at probability level.

At a broader level, the congruence between

An Online Reference, version 5.2, 15 July

(2000).

q, and average D values of the former were obtained not only from the total

qq Mean Hs sMantel corr. a ofDthe former were obtained not onlycoef. from the coef. total q, and average Da values Mean D Hs Mantel s corr. a (min.; max.) (min.; max.) value) (P value) (P value) number of samples, but alsomax.) from(Pall possible of spatially (min.; max.) (min.; (P value)combinations 0.076 localities distributed 0.062 within 20.546 0.499 of the unstable contiguous the geographic extension 0.076 0.062 20.546 (0.001) 0.499 (0.034; 0.072) encompass (0.020; 0.082) (0.141) area. Parentheses minimum and maximum values from subsamples. (0.034; (0.020; 0.082) (0.141) 0.003 0.001 11.498 0.140 (0.001) P values in0.072) bold highlight statistical significance at 0.05probability level.

Unstable(south of BA) 27 22 0.003 0.001 11.498 (0.580) 0.140 (0.004) H. semilineatus Stable (BA) 71 0.031 0.036 0.054 (south of BA) (0.004) (0.580) n 28 S q Mean Da 17.778 Hs Mantel s corr. coef. Species Area (718 bp) (6; 13) (14; 58) (0.009; 0.034) 0.041) H. semilineatus Stable (BA) 28 71 0.031 0.036 (min.; max.) (min.; max.) (min.; max.) (0.007; (min.; max.) (0.029) ( P17.778 value) (0.460) (P0.054 value) Unstable 15 0.0030.034) 0.004 0.041) 0.114(0.029) 0.436 (0.460) (718 bp) (6; 13) (14; 9 58) (0.009; (0.007; H. albomarginatus Stable (BA) 36 207 0.076 0.062 20.546 (0.248) 0.499 (south of BA) (0.357) Unstable 15 9 0.003 0.004 0.114 0.436 H. faber Stable (BA) 28 0.0180.072) 0.026 0.082)38.111 0.803 (0.001) (970 bp) (13; 23) (81; 94 155) (0.034; (0.020; (0.141) (south of BA) (0.357) (0.0003) (0.248) (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.003) Unstable 27 22 0.003 0.001 11.498 0.140 H. faber Stable Stable (BA) (SP) 28 15 94 0.018 0.026 38.111 0.803 48 0.023 0.028 5.981 0.305 (0.580) (south of BA) (0.004) (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.115) (0.003) (0.221) (0.0003) H. semilineatus Stable (BA) 28 71 0.031 0.036 17.778 0.054 40 0.015 0.016 13.255 0.0001 0.305 Stable Unstable (SP) 15 18 48 0.023 0.028 5.981 (718 bp) (6; 13) (14; 58) (0.009; 0.034) (0.007; 0.041)(0.014) (0.029) (0.456) (0.460) (south of SP) (0.115) (0.221) Unstable 15 9 0.003 0.004 0.114 0.436 Unstable 18 40 0.015 0.016 13.255 0.0001 (south of BA) (0.357) (0.248) (south of SP) (0.014) (0.456) H. faber Stable (BA) 28 94 0.018 6 FEBRUARY 0.026 38.111 0.803 www.sciencemag.org SCIENCE VOL 323 2009 787 (771 bp) (13; 23) (42; 80) (0.012; 0.022) (0.001; 0.044) (0.003) (0.0003) Stable 48 related groups 0.023 0.028 in this region 5.981relative to the 0.305 much more distantly presented here show that the (SP) central region 15 of Atlantic estation more ex- 787 www.sciencemag.org SCIENCE VOL 323 6 FEBRUARY 2009 (0.115) (0.221) had much higher stability relative to the south. forest endemics. tensive forests in So Paulo and southern Brazil Unstable 40 efforts, molecular 0.015 studies, (0.016 13.255 0.0001 diversity Forest lizards (14, 29) and birds (30) also show 18 Because collection 9, 31). Not only could much unique (south of (0.014) (0.456)could high diversity in the central portion ofSP) the biome and conservation priorities have been heavily be lost, but ongoing habitat destruction relative to southern areas, and provide evidence biased toward southern and southeastern Brazil quickly erase the signature of the historical for population expansion in southern regions. (8, 9, 31), we predict that genetic diversity and processes that led to it, preventing a full underwww.sciencemag.org SCIENCE VOL corridor 323 6of FEBRUARY 2009 in the central the standing This reassures us that the processes uncovered of the mechanisms underlying local en- 787 narrow endemism by the amphibian data may be generalized to biome have been substantially underestimated. demism and, therefore, impeding more effective and help to explain patterns of diversity in other, This is serious, given the higher rate of defor- conservation measures.

joint, multispecies analyses of mtDNA diversity shows that palaeoclimatic niche models and assemblage-scale molecular genetic analyses can be used to forecast spatial patterns of diversity in poorly explored, highly threatened ecosystems. In a world of ever-accelerating environmental changes, this approach can help to guide research and conservation in other global hotspots or similarly complex tropical ecosystems.
1. G. Hewitt, Nature 405, 907 (2000). [CrossRef] 2. F. E. Mayle, D. J. Beerling, W. D. Gosling, M. B. Bush, Philos. Trans. R. Soc. London B Biol. Sci. 359, 499 (2004). 3. E. P. Lessa, J. A. Cook, J. L. Patton, Proc. Natl. Acad. Sci. U.S.A. 100, 10331 (2003). 4. B. W. Nelson, C. A. C. Ferreira, M. F. da Silva, M. L. Kawasaki, Nature 345, 714 (1990). 5. C. H. Graham, C. Moritz, S. E. Williams, Proc. Natl. Acad. Sci. U.S.A. 103, 632 (2006). 6. W. Jetz, C. Rahbek, R. K. Colwell, Ecol. Lett. 7, 1180 (2004). 7. A. Hugall, C. Moritz, A. Moussalli, J. Stanisic, Proc. Natl. Acad. Sci. U.S.A. 99, 6112 (2002). 8. L. P. C. Morellato, C. F. B. Haddad, Biotropica 32, 786 (2000). 9. M. T. Rodrigues, Conserv. Biol. 19, 659 (2005). 10. C. Kremen et al., Science 320, 222 (2008). 11. Materials and methods are available as supporting material on Science Online. 12. M. B. Arajo et al., Ecography 31, 8 (2008). 13. D. R. Frost, Amphibian Species of the World:

14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

References and Notes

28. 29. 30. 31.

New York, 2008); electronic database accessible at http://research.amnh.org/ herpetology/amphibia/index.php. A. C. Carnaval, C. Moritz, J. Biogeogr. 35, 1187 (2008). M. Slatkin, Evolution 47, 264 (1993). J. A. Hanley, B. J. McNeil, Radiology 143, 29 (1982). J. Cohen, Ed. Psych. Meas. 20, 37 (1960). H. Behling, R. R. B. Negrelle, Quat. Res. 56, 383 (2001). H. Behling, H. W. Arz, J. Ptzold, G. Wefer, Palaeogeogr. Palaeoclimatol. Palaeoecol. 179, 227 (2002). K. Tamura, M. Nei, Mol. Biol. Evol. 9, 678 (1993). F. Tajima, Genetics 143, 1457 (1996). M. Nei, Molecular Evolutionary Genetics (Columbia Univ. Press, New York, 1987). J. C. Fay, C. I. Wu, Genetics 155, 1405 (2000). M. J. Hickerson, C. P. Meyer, BMC Evol. Biol. 8, 322 (2008). R. E. Kass, A. E. Raftery, J. Am. Stat. Assoc. 90, 773 (1995). J. W. O. Ballard, M. Kreitman, Genetics 138, 757 (1994). C. F. B. Haddad, L. F. Toledo, C. P. A. Prado, Anfibios da Mata Atlntica (Atlantic Forest Amphibians) (Editora Neotropica, So Paulo, Brazil, 2008). M.-P. Ledru et al., Divers. Distrib. 13, 761 (2007). K. C. M. Pellegrino et al., Biol. J. Linn. Soc. London 85, 13 (2005). G. S. Cabanne, F. M. dHorta, E. H. R. Sari, F. R. Santos, C. Y. Miyaki, Mol. Phylogenet. Evol. 49, 760 (2008). J. M. C. da Silva, M. Tabarelli, Nature 404, 72

providing access to collections MNRJ and MZUSP; O. Peixoto, M. Gomes, A. Muri, R. Kautskyi, S. Lima, E. Santos, J. S. Filho, J. V. Filho, G. Barros, J. Queiroz, R. Arajo, L. Japp, H. Japp, J. Giovanelli, J. Alexandrino, L. Toledo, O. Arajo, G. Egito, J. Zina, D. Loebmann, D. Pavan, R. Amaro, V. Verdade, F. Curcio, M. Dixo, and J. Cassimiro for field work assistance; W. Monahan and R. Hijmans for discussions about the modeling work; L. Smith and D. Turong for DNA-sequencing assistance; R. Pereira, R. Damasceno, S. Rovito, J. Kolbe, S. Singhal, R. Puschendorf, and A. Pounds for discussions about earlier versions of the manuscript. Funding was provided by the NSF (awards DBI 0512013 to A.C.C., DEB 0743648 to M.J.H., DEB 416250 and DEB 0817035 to C.M.), Fundao de Amparo Pesquisa do Estado de So Paulo and Conselho Nacional de Desenvolvimento Cientfico e Tecnolgico (grants to C.F.B.H. and M.T.R.). Sequences are deposited in GenBank (FJ502639-FJ502822). A.C.C. and C.M. designed the study; A.C.C., C.F.B.H., and M.T.R. collected the data; A.C.C., C.M. and M.J.H. analyzed the data; A.C.C. wrote the paper. All authors discussed the results and commented on earlier versions of the manuscript.

Supporting Online Material

www.sciencemag.org/cgi/content/ full/323/5915/785/DC1 Materials and Methods Figs. S1 to S6 Tables S1 and S2 References 8 October 2008; accepted 9 December 2008 10.1126/science.1166955

38

39

Our New Darwin shirt yours FREE when you become a AAAS member.

Join Today!

Special offer: $99 Professional $50 Postdoc/Student*

Become a AAAS member today, at this special introductory offer and receive a free Darwin shirt (a $22.50 value free). Your membership gives you all of the personal membership benets of AAAS including 51 issues of Science magazine, career development tools, and much more all while supporting our international, public policy, and educational programs the ones that advance science and serve society.

Order online at promo.aaas.org/darwin24

Visit promo.aaas.org/giftdarwin to order now

*International orders will receive Science Digital edition - other membership categories are available online.