27 APRIL 2012 VOL 336 SCIENCE www.sciencemag.org 422

PERSPECTIVES

institutes ( 4). The advent of satellite tech-niques ( 5) restored the ability to compare clocks remotely without degradation, and the balance remained essentially this way through the remainder of the 20th century. This ability enabled fi rst the coordination of national time signals ( 6) and eventu-ally the generation of International Atomic Time (TAI) and Coordinated Universal Time (UTC). These time scales are now the basis of international timekeeping, drawing on and linking together hundreds of atomic clocks around the world ( 7) and in turn sup-porting global applications from electricity distribution to fi nancial markets.

Today, the best atomic clocks tick at opti-cal frequencies (~1015 Hz), and are orders of magnitude more stable again than their microwave forebears (~1010 Hz). Two optical clocks in the same laboratory can be com-pared ( 8) with exquisite precision. General relativity predicts that if a clock on Earth is elevated, it runs faster (because of its higher gravitational potential), and this effect has been resolved for a change in height of only 33 cm or a fractional frequency shift at the 10−17 level ( 9). However, for distant clocks, even the best satellite comparisons stop well short of this resolution, by a factor of ~100 after 1 day of averaging ( 10). Without new techniques, we must return to the fl y-ing clock comparisons of 50 years ago, and await the development of a portable optical frequency standard, an undertaking being pursued in several laboratories.

Transfer of signals by optical fi ber is one possible solution. A number of links have been successfully demonstrated for dis-tances of around 100 km, and, with active stabilization of fl uctuations in length, can transfer the full performance of optical standards within practical averaging times. The goal is now to reach longer distances: 1000 km or more to link laboratories around Europe, several thousand kilometers to span the United States or Australia, and even lon-ger for intercontinental links. The work of Predehl et al. sets a new distance record (920 km), but more importantly, it shows that the practical challenges such as attenuation and remote operation can be met on this distance scale. Separately, it has also been shown that these comparisons can be performed in par-allel with data transfer over the same “lit” fi ber—that is, one already in use on other wavelength channels ( 11). This capability is important, as it extends the reach of net-works where “dark” fi ber (installed in antici-pation of future capacity increases) proves impractical or prohibitively expensive.

Together, these technologies are work-ing toward a new international network of time and frequency standards. This network is not just for metrologists, but supports a whole range of applications, just as did ear-lier networks linked by telegraph, radio, or satellite. One example is precision geod-esy, measuring Earth’s gravitational poten-tial to new levels of precision with optical clocks as “Einstein sensors.” Precise syn-

chronization is also needed for very long baseline interferometry in radio astronomy, with antenna systems such as the proposed Square Kilometre Array extending over con-tinental scales. Here the same fi ber network that transfers data can also transfer time; in future, the same can be true for research, industry, and even our homes, with unprec-edented access to accurate and reliable time.

In the 60 years or so since the develop-ment of the atomic clock, technological infrastructure from telecommunications networks to satellite navigation systems has come to depend intimately not only on the performance of these clocks but also on the ability to compare them. This history tells us to expect the revolution to continue during the next 50 years of timekeeping.

References

1. K. Predehl et al., Science 336, 441 (2012). 2. W. Markowitz, R. G. Hall, L. Essen, J. V. L. Parry, Phys. Rev.

Lett. 1, 105 (1958). 3. F. H. Reder, M. R. Winkler, C. Bickart, Proc. Inst. Radio

Eng. 49, 1028 (1961). 4. A. S. Bagley, L. S. Cutler, Hewlett Packard J. 15, 1 (1964). 5. J. McA. Steele, W. Markowitz, C. A. Lidback, IEEE Trans.

Instrum. Meas. 13, 164 (1964). 6. L. Essen, Nature 187, 452 (1960). 7. Bureau International des Poids et Mesures (BIPM),

Circular T, available at www.bipm.org/jsp/en/TimeFtp.jsp (February 2012).

8. D. Kleppner, Science 319, 1768 (2008). 9. C. W. Chou, D. B. Hume, T. Rosenband, D. J. Wineland,

Science 329, 1630 (2010). 10. A. Bauch et al., Metrologia 43, 109 (2006). 11. O. Lopez et al., Opt. Express 18, 16849 (2010).

Microbial Evolution in the Wild

EVOLUTION

Edward F. DeLong

A genetic study of microbes in an acid mine

drainage site provide evolutionary rate

estimates for wild microbe populations.

Microbial species thrive in every corner of Earth’s biosphere. Their rapid growth rates and promiscu-

ous gene swapping provide ample grist for the evolutionary mill over relatively short time spans. In ocean surface waters, for exam-ple, microbial doubling times of about 1 day result in an estimated production of around 1030 new cells per year ( 1). (This number exceeds that of all individual grains of sand on Earth by about 10 orders of magnitude.) These cells are not static entities, but rather the products and perpetuators of dynamic

evolutionary processes. But exactly how fast, to what degree, and by what mechanisms, do free-living microbes change and evolve over time in natural settings? On page 462 of this issue, Denef and Banfi eld ( 2) report evolu-tionary rate estimates from free-living micro-bial species in the wild.

Measuring evolutionary rates of indi-vidual species in complex natural microbial assemblages is a daunting task. Yet Denef and Banfi eld faced even more formidable chal-lenges. The hot, humid, acidic, low-oxygen conditions found at their acid mine drain-age (AMD) study site is, as the authors put it, “close to the limit of human endurance” ( 2). This would seem an unlikely place for con-ducting detailed evolutionary studies—but in

some ways it is ideal. The AMD site harbors a very simple ecosystem containing only a handful of microbial species ( 3). Leptospiril-

lum, a hardy and versatile bacterium that can live in sulfuric acid, eat iron, and fi x carbon dioxide, often dominates AMD biofi lms ( 3, 4). The very low diversity and complexity of the AMD biofi lm allowed Denef and Banfi eld to directly follow the succession of genotypes in Leptospirillum populations over time, and to track recombination events and the accu-mulation of genomic nucleotide substitutions in populations collected over the course of 5 years (see the fi gure).

By comparing assembled genomes from different genotypes against total population DNA, the authors found that the AMD bio-

Departments of Biological Engineering and of Civil and Envi-ronmental Engineering, Massachusetts Institute of Technol-ogy, Cambridge, MA 02138, USA. E-mail: delong@mit.edu

10.1126/science.1220826

Published by AAAS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

www.sciencemag.org SCIENCE VOL 336 27 APRIL 2012 423

PERSPECTIVES

fi lm populations contained six distinct Lep-

tospirillum genotypes (types I to VI). The time series was dominated by Leptospiril-

lum genotype III, which appeared to be a recombinant hybrid of genotypes I and VI. Time-series population DNA sequencing of Leptospirillum genotype III biofi lms sam-pled over the course of 5 years yielded a sin-gle-nucleotide substitution rate of 1.4 × 10−9 substitutions per nucleotide per generation. This rate estimate, along with evolutionary history reconstructions, suggested that the six Leptospirillum genotypes diverged from a common ancestor in a matter of decades. The data also indicated that the major recombinations refl ected single events that led to the domination of specifi c recombi-nant genotypes. Population genomic anal-yses suggested that Leptospirillum geno-type evolution (see the fi gure) consisted of a complex menu of immigration, horizontal gene transfer (HGT), homologous recom-

bination, fi xation, and selective sweeps that successively generated and replaced differ-ent genotypes and, presumably, phenotypes.

The AMD Leptospirillum nucleotide substitution rate of 1.4 × 10−9 per nucleo-tide per generation is at the high end of pre-vious estimates, which have ranged from 7.2 × 10−11 to 4.0 × 10−9 ( 5– 7). One reason may be the use of a different approach. In perhaps the most elegant and comprehen-sive study of its kind to date, Denef and Banfield performed cultivation-indepen-dent population genomic analyses. In addi-tion, they leveraged a unique combina-tion of data that included in situ proteomic data sets ( 8), three distinct Leptospirillum population genotypes (assembled directly from natural populations) and a population genomic time series.

Unlike Denef and Banfi eld’s study, other estimates of bacterial evolutionary rates have relied on isolation, propagation, or

comparison of pure cultures or laboratory populations ( 5, 7, 9). These approaches may be selective and may not refl ect native com-munity complexity. On the other hand, the AMD habitat is itself unique; its unusual conditions and low biological complexity may affect the evolutionary rate estimates. Denef and Banfi eld point out, however, that their estimates are consistent with indepen-dent theoretical predictions of a universal mutation rate ( 10).

A central goal in microbial population biology is to identify key genomic changes

Early colonizer

Late colonizer

Recombination

Recombinant 1

Recombinant 2

1Initial colonization by an

early colonizer genotype

12Proliferation of late

colonizer genotype

3DNA transfer from late colonizer

to early colonizer

4Recombination in early colonizer

to create Recombinant 1

8Recombination in Recombinant 1

creates Recombinant 2

10

Accumulation of single

nucleotide polymorphisms

in Recombinant 2

5Proliferation of

Recombinant 1

9 Proliferation of

Recombinant 2

6Transient immigration of

late colonizer genotype

7DNA transfer from late

colonizer to Recombinant 1

11Immigration of late

colonizer genotype

2

Proliferation of the early colonizer,

followed by transient immigration

of a late colonizer genotype

Microbial microevolution in a biofi lm. This hypo-

thetical scenario of evolutionary events in AMD bio-

fi lms illustrates some of the processes observed and

inferred by Denef and Banfi eld. The white circles do

not necessarily represent the same biofi lm. The dot-

ted black arrows do not imply direct succession in

the same biofi lm; rather, different amounts of time,

and different spatial locations, may be involved in

the transitions, which could take years to decades.

Published by AAAS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

27 APRIL 2012 VOL 336 SCIENCE www.sciencemag.org 424

PERSPECTIVES

Complex Protostellar Chemistry

ASTRONOMY

Joseph A. Nuth III and Natasha M. Johnson

Numerical models show that protoplanetary

nebulae are sites of chemical activity even

in the cold outer disk.

that lead to adaptation and selection—the “differences that make a difference” ( 11–

13). Denef and Banfi eld speculate that phe-notypic selection in the Leptospirillum pop-ulations may be due to differences of only a few nucleotides, but there are some cave-ats. Their approach captured major homolo-gous recombination events, but nonhomolo-gous gene acquisitions from HGT may have gone undetected, because the Leptospiril-

lum population analyses relied on compari-sons of genes present in reference genomes. Without complete and contiguous genome assemblies from the population genomes, large blocks of newly acquired foreign genes could be invisible to this approach.

Recent studies suggest that HGT and ille-gitimate recombination can be common and frequent in native microbial populations ( 13–15).

It turns out that “genome” is a verb, not a noun—a process, not a product. Current and emerging microbial population genomic studies that capture this dynamic ( 2, 6, 7, 9,

13–15) promise to help paint a more inte-grated motion picture of microbial ecology and evolution in action.

References

1. W. B. Whitman, D. C. Coleman, W. J. Wiebe, Proc. Natl.

Acad. Sci. U.S.A. 95, 6578 (1998). 2. V. J. Denef, J. F. Banfi eld, Science 336, 462 (2012). 3. G. W. Tyson et al., Nature 428, 37 (2004).

4. G. W. Tyson et al., Appl. Environ. Microbiol. 71, 6319

(2005). 5. J. E. Barrick et al., Nature 461, 1243 (2009). 6. N. A. Moran, H. J. McLaughlin, R. Sorek, Science 323,

379 (2009). 7. L. Yang et al., Proc. Natl. Acad. Sci. U.S.A. 108, 7481

(2011). 8. V. J. Denef et al., Environ. Microbiol. 11, 313 (2009). 9. T. D. Lieberman et al., Nat. Genet. 43, 1275 (2011). 10. J. W. Drake, Proc. Natl. Acad. Sci. U.S.A. 88, 7160 (1991). 11. G. Bateson, Steps to an Ecology of Mind: Collected Essays

in Anthropology, Psychiatry, Evolution, and Epistemology (Univ. of Chicago Press, Chicago, 1972).

12. C. Fraser, E. J. Alm, M. F. Polz, B. G. Spratt, W. P. Hanage, Science 323, 741 (2009).

13. B. J. Shapiro et al., Science 336, 48 (2012). 14. J. H. Hehemann et al., Nature 464, 908 (2010). 15. C. S. Smillie et al., Nature 480, 241 (2011).

10.1126/science.1221822

Two decades ago, our understanding of the chemistry in protostars was sim-ple—matter either fell into the cen-

tral star or was trapped in planetary-scale objects. Some minor chemical changes might occur as the dust and gas fell inward, but such effects were overwhelmed by the much larger-scale processes that occurred even in bod-ies as small as asteroids. The chemistry that did occur in the nebula was relatively easy to model because the fall from the cold molecu-lar cloud into the growing star was a one-way trip down a well-known temperature-pressure gradient; the only free variable was time. However, just over 10 years ago it was sug-gested that some material could be processed in the inner nebula, fl ow outward, and become incorporated into comets ( 1, 2). This outward fl ow was confi rmed when the Stardust mis-sion returned crystalline mineral fragments ( 3) from Comet Wild 2 that must have been processed close to the Sun before they were incorporated into the comet. On page 452 of this issue, Ciesla and Sandford ( 4) demon-strate that even the outermost regions of the solar nebula can be a chemically active envi-ronment. Their finding could have conse-quences for the rest of the nebula.

Outward fl ow in the nebula is the natural result of conserving the total angular momen-tum of the star-forming system as mass con-tinues to accrete onto the rotating disk feed-

ing the growth of the central star ( 5– 7). As mass flows inward along the flared accre-tion disk, additional mass fl ows slowly out-ward along the midplane (see the fi gure). As an additional complication, because the mid-plane is hotter than the outer boundary of the disk, convection will also mix materials verti-cally in the nebula. Ciesla and Sandford show that ice-coated dust grains, moving outward and subject to convection, will be exposed to cosmic radiation that is suffi cient to cause the same chemical effects seen in dark cloud cores—that is, the conversion of simple car-

bon- and nitrogen-containing molecules into more complex organic species—and so will have consequences for nebular chemistry.

It has been nearly 40 years since the discov-ery that the oxygen in solid bodies throughout the solar system is mass-independently frac-tionated ( 8). It appears as if 16O were added or subtracted from the bulk composition of the dust independently of either 17O or 18O. While the fi rst explanation for this observation was the addition of 16O-rich supernova grains to the forming solar system ( 8), the currently favored explanation is known as chemical

Solar System Exploration Division, NASA’s Goddard Space Flight Center, Greenbelt, MD 20771, USA. E-mail: joseph.a.nuth@nasa.gov

10,000 1000 100 10 1 0.1 0

Distance from protosun (astronomical units, logarithmic scale)

Protosun

Dense accretion disc

In-falling gas and dust

Collimated

bipolar

outflow

0.1 1 10 100 1000 10,000

Complex reactions. Large-scale motion driven by conservation of angular momentum, together with more local convective cells above and below the hotter nebular midplane, dynamically mix products of chemical reactions from many different environments throughout the nebula.

Published by AAAS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

biomineralization activity may be sought in theform of Ba- and Sr-rich carbonates. These Ba- andSr-rich phases therefore add to the list of poten-tial mineral biosignatures in the fossil record.

References and Notes1. A. Herrero, E. Flores, Eds., The Cyanobacteria: Molecular

Biology, Genomics and Evolution (Caister Academic,Wymondham, Norfolk, UK, 2008).

2. C. Jansson, T. Northen, Curr. Opin. Biotechnol. 21, 365(2010).

3. R. Riding, Geobiology 4, 299 (2006).4. W. Altermann, J. Kazmierczak, A. Oren, D. T. Wright,

Geobiology 4, 147 (2006).5. J. Farquhar, A. L. Zerkle, A. Bekker, Photosynth. Res.

107, 11 (2011).6. G. Arp, A. Reimer, J. Reitner, Science 292, 1701

(2001).7. M. Merz, Facies 26, 81 (1992).8. N. Planavsky, R. P. Reid, T. W. Lyons, K. L. Myshrall,

P. T. Visscher, Geobiology 7, 566 (2009).9. C. Dupraz et al., Earth Sci. Rev. 96, 141 (2009).

10. M. Obst et al., Geochim. Cosmochim. Acta 73, 4180 (2009).11. M. R. Badger, D. Hanson, G. D. Price, Funct. Plant Biol.

29, 161 (2002).12. E. Kupriyanova et al., Microbiology 153, 1149 (2007).13. See supplementary materials on Science Online.14. J. Ka!mierczak et al., Facies 57, 543 (2011).15. E. Couradeau et al., PLoS ONE 6, e28767 (2011).

16. K. Benzerara et al., Geobiology 2, 249 (2004).17. M. E. Fleet, X. Y. Liu, Am. Mineral. 94, 1235

(2009).

18. G. E. Bath et al., Geochim. Cosmochim. Acta 64, 1705(2000).

19. M. R. Krejci et al., J. Struct. Biol. 176, 192 (2011).20. D. C. Dominguez, Mol. Microbiol. 54, 291 (2004).21. L. Addadi, S. Raz, S. Weiner, Adv. Mater. 15, 959 (2003).22. P. Rez, A. Blackwell, J. Phys. Chem. B 115, 11193 (2011).23. A. E. Walsby, P. K. Hayes, R. Boje, Eur. J. Phycol. 30, 87

(1995).24. R. Rippka, J. Waterbury, G. Cohen-Bazire, Arch.

Microbiol. 100, 419 (1974).25. L. I. Falcón, S. Magallón, A. Castillo, ISME J. 4, 777

(2010).26. E. Bittarello, D. Aquilano, Eur. J. Mineral. 19, 345 (2007).

Acknowledgments: E.C. was recipient of a fellowship from theFrench Ministère de la Recherche et de l’EnseignementSupérieur. This project was financed by the CNRSinterdisciplinary program “Environnements planétaires etorigines de la vie” (PID EPOV) and Institut National desSciences de l’Univers (INSU) program “InteractionsTerre/Vie”(InterrVie). The SEM facility of the Institut de Minéralogie etde Physique des Milieux Condensés is supported by RégionIle de France grant SESAME 2006 I-07-593/R, INSU-CNRS,INP-CNRS, University Pierre et Marie Curie, Paris. Support wasalso provided by NSF and the U.S. Environmental ProtectionAgency under cooperative agreement EF-0830093 through theCenter for Environmental Implications of Nanotechnology.Advanced Light Source (ALS) Molecular EnvironmentalScience beamline 11.0.2 is supported by the Office ofScience, Office of Basic Energy Sciences, Division of ChemicalSciences, Geosciences, and Biosciences and MaterialsSciences Division, U.S. Department of Energy, at the LawrenceBerkeley National Laboratory. Beamline 5.3.2 at ALS issupported by the Office of Basic Energy Sciences of the U.S.Department of Energy under contract DE-AC02-05CH11231.We thank R. Tavera (Universidad Autonónoma de México) forproviding help as a field guide and the Laboratoire deGéochimie des Eaux at IPGP for providing the facility for Granmeasurements. The sequence of the 16S rRNA gene ofCandidatus Gloeomargarita lithophora has been deposited inGenBank under accession no. JQ733894. Data and enrichmentcultures are available upon request.

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6080/459/DC1Materials and MethodsSupplementary TextFigs. S1 to S4Tables S1 to S8References (27–38)

3 November 2011; accepted 14 March 201210.1126/science.1216171

In Situ Evolutionary Rate MeasurementsShow Ecological Success of RecentlyEmerged Bacterial HybridsVincent J. Denef1* and Jillian F. Banfield1,2†

Few data are available on how quickly free-living microorganisms evolve. We analyzed biofilmscollected from a well-defined acid mine drainage system over 9 years to investigate the processesand determine rates of bacterial evolution directly in the environment. Population metagenomicanalyses of the dominant primary producer yielded the nucleotide substitution rate, which weused to show that proliferation of a series of recombinant bacterial strains occurred over the pastfew decades. The ecological success of hybrid bacterial types highlights the role of evolutionaryprocesses in rapid adaptation within natural microbial communities.

Microbial communities, which drive Earth’sgeochemical cycles (1), can rapidly re-spond to change, but the proportion of

this response that can be attributed to evolution-ary processes, rather than species compositionor gene expression shifts, remains an unresolved

Fig. 3. Phylogenetic position of the intracellularly calcifying cyanobacterium Candidatus Gloeomargaritalithophora. A maximum likelihood phylogenetic tree of cyanobacterial small subunit ribosomal DNAs wasconstructed from 1191 unambiguously aligned positions. Sequences obtained from various enrichment cultureswere 100% identical (pink) and very closely related to the environmental sequence Alchichica_AQ1_1_1C_35CyanoOTU02, abundantly retrieved from cultured microbialites. Numbers at nodes indicate bootstrapvalues. The scale bar indicates the number of substitutions per site for a unit branch length.

27 APRIL 2012 VOL 336 SCIENCE www.sciencemag.org462

REPORTS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

question (2). Most evolutionary rate estimates areavailable for nucleotide substitution rates andderive from laboratory measurements (3, 4). Itis difficult to know how relevant these rates arefor geochemical environments, because studieson natural populations have been restricted topathogens (5–7) and endosymbionts (8).

Our approach to measuring evolutionaryrates for free-living bacterial populations involvedtracking genome change in a time series of nat-ural microbial community samples. Few systemsare suitable for such a study; requirements includethe availability of discrete, well-defined, and rela-tively homogeneous microbial community samplesfrom a location where a reproducible communityforms over time, with very restricted microbial in-put from other regions. The microbial communi-ties catalyzing the formation of acidmine drainage(metal-rich, low-pH solutions) in the RichmondMine (IronMountain, CA) provide such a system.Biofilms that develop at the air-solution interfaceon standing pools and slow-flowing undergroundstreams have been used for more than a decade tostudy evolutionary processes and ecological com-plexity in nature (9). Key to the success of thisenvironment as a model system has been its lowspecies richness, which enables detailed culture-independent community genomics analyses ofthe genetic structure and dynamics of natural pop-ulations (10–13). Most biofilms within this sys-tem are dominated by the chemolithoautotrophicLeptospirillum group II. Previous metagenomicstudies on two biofilms led to the reconstructionof genomes for Leptospirillum group II, types Iand VI, the genes of which share ~94% averagenucleotide identity (10, 13, 14). Inferences basedon proteomics data indicate that large-scale ho-mologous recombination events occurred betweentwo parental Leptospirillum group II types, re-sulting in six recombinant (hybrid) genotypes(14, 15). One genotypic group, comprising typesII through VI, predominates during initial coloni-zation, whereas type I becomes abundant in latersuccessional stages (16).

We have collected biofilms within the Rich-mond Mine over approximately 9 years. Sam-pling required field expeditions to locations whereambient conditions are often close to the limitof human endurance (e.g., 100% humidity, ~48°C,with depressed O2 levels) at sites only accessibleat times of the year when flow rates are low anddanger from underground collapse is minimized.Working within these constraints, we collectedmicrobial biofilms (both time series and spatiallyresolved) from six underground locations betweenMarch 2002 and December 2010 at intervals of3 weeks to more than 1 year (Fig. 1, table S1, and

figs. S1 and S2) (17).We selected 13 samples fromthe C75 location, a site indicated by proteomics-based inferences to be typically dominated by theLeptospirillum group II, type III genotype (15).This allowed us to analyze single-nucleotide sub-stitution accumulation over 5 years, and therebyto estimate the substitution rate. On the basis ofthe same proteomics-based inferences, we alsoselected nine biofilms that provided multiplesamplings of most of the other Leptospirillumgroup II genotypes (types I, III, IV, V, and VI).Together with the two sequence data sets fromwhich the type I and type VI reference genomeshad been reconstructed, this allowed us to recon-struct a lineage history (fig. S1).

We generated community genomic data setscomprising ~63 billion base pairs of sequencefrom 24 samples, each of which contained ~108

Leptospirillum group II cells (10). Using se-quences obtained from the first five C75 loca-tion biofilms, for which longer read lengths wereavailable, we reconstructed a type III genome denovo (17). The genome is indeed a recombinanthybrid of the previously reconstructed type I andtype VI genomes, as deduced from proteomicanalysis (15). All identified recombination pointswere located within genes (table S2 and fig. S3).All Leptospirillum group II sequences were ac-counted for (fig. S4) (17), excluding the possi-bility of other genotypes in these five samples.

The sequencing reads from each of the 13 C75samples collected over 5 yearswere recruited to thepreviously reconstructed Leptospirillum group II,type I and type VI genomes (13, 18) to identify

the recombinant block structure of the genomeat each time point at this location. Only alignmentof reads to the sequence with the highest similaritywas permitted. Twelve biofilms contained only thetype III genotype (>99% of cells); the remainingone contained ~84% type III and ~16% type I geno-type (Fig. 1 and figs. S5 and S6A). Because 1.7million to 8million Illumina sequencing reads wereobtained per sample (table S1), and because eachread is an independentmeasurement of population-level sequence variation, we could unambiguouslyidentify high-frequency single-nucleotide polymor-phisms (SNPs) (>90% of all reads).

Nucleotide by nucleotide, we compared thetype I–like segments of the type III genome to thetype I genome sampled in 2002. Similarly, typeVI–like regions of the type III genome werecompared to the type VI genome sampled in2005. In total, relative to the reference sequences,64 mutations were fixed in all populations sam-pled at the C75 location (fig. S7A).

We then tracked genome change at the C75location between August 2006 and December2010. In the 11 C75 samples taken after August2006, we identified up to six additional high-frequency substitutions relative to the 2006 type IIIgenome (fig. S7A). Notably, the number of high-frequencymutations increased over time (fig. S8,R2 = 0.78). Using this information, we calculateda rate of 1.4 ! 10!9 (SE, T0.2 ! 10!9) substitutionsper nucleotide per generation (table S3). This num-ber is in the higher range of previously reportedbacterial genome-wide short-term substitutionrates [7.2 ! 10!11 to 4.0 ! 10!9 substitutions per

1Department of Earth and Planetary Science, University ofCalifornia, Berkeley, CA 94720, USA. 2Department of En-vironmental Science, Policy, and Management, University ofCalifornia, Berkeley, CA 94720, USA.

*Present address: Department of Ecology and EvolutionaryBiology, University of Michigan, Ann Arbor, MI 48109, USA.†To whom correspondence should be addressed. E-mail:jbanfield@berkeley.edu

Fig. 1. Leptospirillumgroup II genotype dis-tribution shows disper-sal across the RichmondMine and type III strainturnover at the C75 loca-tion. (A) Richmond Mineschematic map, with piecharts indicating geno-type proportions in 24samples, estimated on thebasis of read recruitment(figs. S5 and S9). (B) Acidmine drainage flow rate(measured at the mineentrance) and communitycomposition at the C75location, measured by flu-orescence in situ hybrid-ization (Arc, Archaea; L3,Leptospirillum group III;L2, Leptospirillum groupII). Leptospirillum group II,type III strain transitions,revealed by SNP analysis,are indicatedbyanasteriskinside the Leptospirillumgroup II FISHdata. (C) Flowdata, 2001 to 2011. Shad-ing indicates period dis-played in (B).

A5-way

UBAI III IV IVa V VI

AB20m (06/06)

C10

C75

80

40

0

Arc L3 L2 late colonizer (I) L2 early colonizer (II-VI)

Apr

-06

Jun-

06

Aug

-06

Nov

-06

Feb

-07

May

-07

Aug

-07

Nov

-07

Feb

-08

Jun-

08

Oct

-08

Feb

-09

Sep

-10

Nov

-10

Dec

-10 *

AB muck

%

* * * * *

Feb

-10

03/02

06/05

06/06 08/06 11/06 05/07

08/07 11/07

12/10

09/10 11/10

06/08V1

B

06/08V2

06/08V3

10/08

08/06 08/07

05/07

08/07

DS0.1 DS1

DS2 DS3

11/06

AMD flow direction

C

0

1000

2000

3000

L/min0

1000

2000

3000

‘01 ‘03 ‘05 ‘07 ‘09 ‘11

www.sciencemag.org SCIENCE VOL 336 27 APRIL 2012 463

REPORTS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

nucleotide per generation (4, 7, 8)]. It is similar tothe rate of 1.3 ! 10!9 substitutions per nucleotideper generation predicted for the type III genomesize using the universal mutations-per-genomerate suggested by Drake (3). Although the highestrates have been reported for endosymbionts andpathogens, which have small effective popula-tion sizes, correspondence with Drake’s predic-tion may indicate that genetic drift acceleratedby population bottlenecks is not the main factoraffecting the substitution rate in Leptospirillumgroup II.

To assess potential biases in the measuredsubstitution rate introduced by spatial variationwithin the contiguous biofilms covering the C75acid mine drainage pool, we sampled three loca-

tions ~1 m apart in June 2008. Samples 1 and 3were dominated by the same genotype (with fourSNPs relative to the type III genome), whereassample 2 was dominated by a variant of this geno-type lacking one high-frequency polymorphism(fig. S7A). The three samples also had only 93 (T8)replicated SNPs at frequency greater than 0.05(average frequency 0.08). The low levels of var-iation, both within populations and across space,provide confidence that nucleotide changes inpopulations at the C75 location over time can beused to calculate the substitution rate.

Read recruitment of community genomicdata from sites located across the undergroundsystem (five-way, UBA, C75, C10, AB muck,and AB20 locations) to types I and VI yielded

six genotypes (Fig. 1 and fig. S9): type I, type VI,and four other types (III, IV, IVa, and V) that eachconsist of a mosaic of type I and type VI genomeblocks tens to hundreds of kilobases in length(Fig. 2, A andB, and figs. S6B and S9). This againconfirmed previous strain-resolved proteomicanalysis–based inferences (14, 15). The presenceof exactly the same transition points between therecombinant blocks in each of the genotypes inevery sample indicates that each major recombi-nation event occurred in a single cell, whose de-scendants rose to fixation (Fig. 2). Each genotype’sfixed mutations were identified by read recruit-ment to type III (Fig. 3 and fig. S7B) (17) andwere used to construct a phylogenetic tree (Fig.2C). Because read recruitment was performed

8049

8692_8135

8027_8524

7931

8241_8062

III C75 10/08

III C75 06/08 V2

III C75 06/08 V3

III C75 06/08 V1

III C75 12/10

III C75 11/10

III C75 09/10

III C75 11/07III C75 08/07

III C75 05/07

III C75 11/06

III C75 08/06

III C75 06/06

III ABmuck 05/07

IV ABmuck 08/07

IVa AB20 06/06

V C10 08/06

V C10 08/07VI UBA 06/05

VI AB20 06/06

I 5-way 03/02

93

100

100

99

96

68

96

88

66

99

100

10 fixed variant loci

20061989-1990

198119681963End to century of mining(main mining of Richmond ore body between 1930-1950)

First under-ground cleanup

1995First scientificsampling

2001-2003Major underground cleanupScientific sampling beyond 5-way

First C75m sampling~50,000 BP

VI

IVa

V

IV

III

B

C

A

III_C75

III_ABmuck(+V+VI+I)

IV(+I)

IVa(+VI+I)

V(+I)

VI (+I)

2003

Early colonizer

Late colonizer I

C

A

BD

E

FG

H

I

J

K

L

M

N

O

P

Q

R

S

T

U V

D*

Fig. 2. Reconstruction of the recent evolutionary history of Leptospirillum groupII. (A) The outer circle plots the counts of peptides unique to the type I (red) andtype VI (blue) reference genomes at each protein locus (dots) for the C75 type IIIpopulation [proteomics-inferred genotyping (PIGT); data from (15)]. Inner circlesshow Illumina read recruitment to the reference genomes, revealing theprevalence of type III, IV, IVa, V, and VI recombinants. (B) Correspondence ofPIGT and recruitment data, highlighting the identical recombination point ingenotypes III to V. (C) Evolutionary history of the sampled genotypes, based on

the variant loci (table S7), inferred using the maximum parsimony method. Thepercentage of replicate trees in which the associated taxa clustered together inthe bootstrap test is shown next to the branches. Dotted arrows indicate re-combination events; circle schematics represent the regions affected. Thetimeline indicates the calculated time ranges of recombination events (tableS3) as well as historical events (fig. S2) (18, 21). Branch D* is presented astwo strains, each assigned half the total number of UBA 06/05 SNPs, becauselow incidence of SNPs precluded their linkage. BP, years before the present.

27 APRIL 2012 VOL 336 SCIENCE www.sciencemag.org464

REPORTS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

relative to the type III genome, positions withintype I–like regions not shared by type IV, V, orVI genotypes were designated as gaps for treeconstruction, and these gaps were treated asmissing data in the calculation of tree branchlength. Notably, the tree topology indicates thatthe larger the fraction of type I–like sequencerecombined into the typeVI background, themorerecently the specific genotype emerged (Fig. 2C).

Using the substitution rate calculated from theC75 time series and assuming equal generationtimes for the different genotypes, we estimatedthat the times to coalescence for the suite of re-combinant genotypes ranged between 2 and 44years (Fig. 2 and table S4). The causes of the suc-cess of these hybrid genotypes are hard to establish.In Bacteria and Archaea, little is known aboutthe importance of hybridization in adaptation tochange. However, one recent study hypothesizedthat agricultural practices led to proliferation of hy-brids of Campylobacter jejuni and Campylobactercoli formed by homologous recombination (19).In Eukarya, hybrids are often reported to have littleecological success (20). However, increased fit-ness relative to parental types has been observedwhen the adaptive landscape changes as a result

of natural- or human-induced environmental al-terations (21, 22). The Richmond Mine site is sub-ject to both natural (seasonal and year-to-yearvariation; Fig. 1C) and human perturbations(fig. S2) (17), a subset of which may have pro-vided opportunity for hybrid proliferation.

Although we cannot connect specific eventsto emergence of the hybrid genotypes, we canevaluate the role of neutral versus selection-basedevolutionary processes in lineage divergence.For example, extreme population bottleneckssuch as flushing events that remove biofilms dur-ing the rainy season may purge diversity indepen-dent of the fitness of particular strains. However,there are several lines of evidence indicating thathybrid genotypes arose as the result of selectiveprocesses. We detected evidence for positive se-lection (McDonald-Kreitman test, P < 0.05) formutations specific to one type VI population(branch C; see Fig. 2), type V (branch F), and theancestor of types III to V (branch E) (table S5).Selection was also supported by the dispropor-tionately high number of signal transduction genes(Fig. 3, functional category T) and transcriptionalregulation genes (Fig. 3, category K), globalregulators in particular, that were affected by

fixedmutations (Fig. 2, branches E, F, D*, andH;see also tables S5 and S6). Previous observa-tions in laboratory settings (23–25) and a naturalsystem (16) similarly identified evolution of geneexpression as an important factor underpinningearly ecological divergence. Finally, an earlierstudy analyzing the distribution of the differentLeptospirillum group II genotypes in the contextof geochemical conditions indicated selectionamong type I through type VI recombinants (15).Evidence for selection also emerged from thecurrent study, where two sites located ~140 mapart along a single flow path were compared(C10 and C75). The type III population, whichdominated all upstream C75 biofilms, was foundin only one of three biofilms sampled at the sametime at the downstream C10 location. Even whenthe type III population was present at both sites,the genomes were different (Fig. 1 and fig. S10).This result suggests selection between genotypesthat differ by only a few nucleotides—a findingconsistent with prior laboratory studies showingincreased fitness due to adaptive SNPs (4, 25).It also indicates that lack of diversity is not theexplanation for dominance of many biofilms bya single genotype.

Fig. 3. Overview of the Leptospiril-lum group II high-frequency variants,based on read recruitment to the C75June 2006 type III genotype. Dots in-dicate variant nucleotides relative tothe last common ancestor of types IIIto VI. Functional categories (table S6)are indicated only for nonsynonymoussubstitutions and only for the first oc-currence, starting from the most re-cent genotype (outside circle). Shadingin inner circles indicates regions thatwere excluded for SNP analysis in a par-ticular genotype because of recombi-nation or indel events.

Leptospirillum Group IIstrain C75m 06/06

(2,590,985 bp)

Non-synonymous substitutionSynonymous substitutionIntergenic substitution

Substitution in 16S rRNA

Fixed variants relative to last common ancestor:Insertion/deletion

Homologous recombination

Other events (shading in inner circles):T, L, ... Functional category (see Table S6)

insertionType III relative to reference genomes (outer circle): Type I (03/02) Type VI (06/05)

1

2

3

4

5

6

7

8

9

10

11 12 13 14 15

16

17 18

19

20

2122

23

2425

2627

28

29

30

31

3233

34

3536

37

38

39

40

41

4243

4445

46

47

48

49

50515253545556

57

58

59

60

61

62

63

64

65

6667

68

69

7071

72

737475

76

77

78

79

80

81

8283848586

no

nono

O

O

S

no

no

no

no

MM

M

H

Q

O

G

R

C

S

S

S

P

S

VV

S

C

S

K

T

TT

T

T

L

E

L

L

L

no

no

no

no

Tr

M

Tr

M

G

GG

U

UQ

K

S

T

EE

L

L

F

J

no

no

no

no

M

M

M

O

G

P

K

K

S

T

T

T

T

L

L

E

E

J

J

no

no

N

P

I

no

no

no

noM

M

Tr

M

O

R

K CK

S

K

K

C

C

S

T

T

T

TT

T

FE

L

L

E

E

F

F

www.sciencemag.org SCIENCE VOL 336 27 APRIL 2012 465

REPORTS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from

Our population genomic analyses provideddata on the accumulation of genome change infree-living populations that underpin geochemicalcycles. Application of these rates revealed that aseries of important divergence events occurredover a time scale of years to decades within thenatural model system of the Richmond Mine.We attribute these to major selection episodes,although we cannot conclude whether they weremediated by natural or human factors. Rapidadaptive evolution may have assisted in the main-tenance of Leptospirillum group II as both thedominant primary producer and most active ironoxidizer responsible for acid mine drainage for-mation. Our results contribute to the developmentof a predictive understanding of how microbialsystems respond to both natural and anthropo-genic change (2).

References and Notes1. P. G. Falkowski, T. Fenchel, E. F. Delong, Science 320,

1034 (2008).2. A. Reid, M. Buckley, Microbial Evolution (American

Academy of Microbiology, Washington, DC, 2011).3. J. W. Drake, Proc. Natl. Acad. Sci. U.S.A. 88, 7160 (1991).4. J. E. Barrick et al., Nature 461, 1243 (2009).5. G. Morelli et al., PLoS Genet. 6, e1001036 (2010).

6. U. Nübel et al., PLoS Pathog. 6, e1000855 (2010).7. L. Yang et al., Proc. Natl. Acad. Sci. U.S.A. 108, 7481

(2011).8. N. A. Moran, H. J. McLaughlin, R. Sorek, Science 323,

379 (2009).9. V. J. Denef, R. S. Mueller, J. F. Banfield, ISME J. 4, 599

(2010).10. G. W. Tyson et al., Nature 428, 37 (2004).11. E. E. Allen et al., Proc. Natl. Acad. Sci. U.S.A. 104, 1883

(2007).12. J. M. Eppley, G. W. Tyson, W. M. Getz, J. F. Banfield,

Genetics 177, 407 (2007).13. S. L. Simmons et al., PLoS Biol. 6, e177 (2008).14. I. Lo et al., Nature 446, 537 (2007).15. V. J. Denef et al., Environ. Microbiol. 11, 313 (2009).16. V. J. Denef et al., Proc. Natl. Acad. Sci. U.S.A. 107, 2383

(2010).17. See supplementary materials on Science Online.18. D. S. A. Goltsman et al., Appl. Environ. Microbiol. 75,

4599 (2009).19. S. K. Sheppard, N. D. McCarthy, D. Falush, M. C. J.

Maiden, Science 320, 237 (2008).20. T. Hatfield, D. Schluter, Evolution 53, 866 (1999).21. B. R. Grant, P. R. Grant, Ecology 77, 500 (1996).22. A. P. Hendry et al., Proc. Biol. Sci. 273, 1887 (2006).23. S. F. Elena, R. E. Lenski, Nat. Rev. Genet. 4, 457 (2003).24. T. F. Cooper, D. E. Rozen, R. E. Lenski, Proc. Natl. Acad.

Sci. U.S.A. 100, 1072 (2003).25. A. Giraud et al., PLoS Genet. 4, e2 (2008).

Acknowledgments: We thank T. W. Arman (president,Iron Mountain Mines Inc.) and R. Sugarek (U.S. Environmental

Protection Agency) for site access; R. Carver and M. Jones foron-site assistance; R. Carver for Richmond Mine flow data;J. Sickles (U.S. EPA) for providing site history documentation;Banfield laboratory members for their contributions tosample collection; C. Miller, C. Sun, and B. Thomas(University of California, Berkeley) for assistance withnext-generation sequencing data analysis; and C. Miller andK. Vogel (University of Arizona) and three anonymousreviewers for constructive criticism. Supported by the Office ofBiological and Environmental Research, U.S. Department ofEnergy, through the Genomic Sciences (DE-FG02-05ER64134)and Carbon-Cycling (DE-FG02-10ER64996) programs. Theauthors declare no conflict of interest. All 454 and Illuminasequencing reads generated for this study have been depositedin the NCBI sequence reads archive (accession numberSRA047370). The Leptospirillum group II C75 strain genomehas been deposited at DDBJ/EMBL/GenBank under theWhole Genome Shotgun project accession numberAIJM00000000. The version described in this paper,containing the assembled and annotated contigs, is the firstversion, AIJM01000000.

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6080/462/DC1Materials and MethodsFigs. S1 to S11Tables S1 to S7References (26–48)

23 December 2011; accepted 9 March 201210.1126/science.1218389

Origins and Genetic Legacy ofNeolithic Farmers andHunter-Gatherers in EuropePontus Skoglund,1* Helena Malmström,1 Maanasa Raghavan,2 Jan Storå,3 Per Hall,4

Eske Willerslev,2 M. Thomas P. Gilbert,2 Anders Götherström,1,5*† Mattias Jakobsson1,5*†

The farming way of life originated in the Near East some 11,000 years ago and had reachedmost of the European continent 5000 years later. However, the impact of the agriculturalrevolution on demography and patterns of genomic variation in Europe remains unknown.We obtained 249 million base pairs of genomic DNA from ~5000-year-old remains of threehunter-gatherers and one farmer excavated in Scandinavia and find that the farmer is geneticallymost similar to extant southern Europeans, contrasting sharply to the hunter-gatherers, whosedistinct genetic signature is most similar to that of extant northern Europeans. Our results suggestthat migration from southern Europe catalyzed the spread of agriculture and that admixture inthe wake of this expansion eventually shaped the genomic landscape of modern-day Europe.

The transition from the hunter-gatherer life-style to a sedentary farming economyspread throughout Europe from the south-

east, starting ~8400 years before the present (yrB.P.) (1). Early population genetic studies ar-gued that clines in genetic variation within Eu-rope favored a model in which this expansionoccurred in concert with substantial replacementof resident hunter-gatherer populations (2–4), con-tradicting models emphasizing the culturally me-diated spread of farming economy (5). Currentresults from extant populations are inconclusive(4, 6); however, ancient DNA analyses (7–10)present tentative evidence for population replace-ment but suffer from uncertainties associatedwith single-locus studies of the Y chromosomeor the mitochondrion. Population genomic anal-ysis of ancient human remains (6, 11) shows

promise, but poses technical difficulties due tolow DNAyield and the risk of present-day humancontamination (6). In addition, geographical andtemporal differences between ancient samplesmake observations of genetic differentiation dif-ficult to interpret. Thus, more robust interpretationsof ancient DNA might be gained by analyzingsamples from cultural complexes occurring inthe same region and during the same time period.In this study, we focused on the Neolithic era ofnorthern Europe, where the relatively late arriv-al of farming (~6000 yr B.P.) was followed bymore than 1000 years of coexistence betweenhunter-gatherer and farming cultures (12).

We obtained genomic DNA sequences fromthree samples (Ajv52, Ajv70, and Ire8) from ahunter-gatherer context [“Neolithic hunter-gatherers”associated with the Pitted Ware Culture (PWC)],one sample from a farming context [“Neolithicfarmer” associated with the Funnel Beaker Cul-ture or Trichterbecher kultur (TRB)], and twoanimal remains as contamination controls. Thehuman remains were chosen from a larger pan-el on the basis of their molecular preservation andpreviously yielded reproducible single-locusgenetic data (8, 13, 14). The Neolithic farmersample (Gök4) was excavated from a mega-lithic burial structure in Gökhem parish, Swe-den, and has been directly 14C-dated to 4921 T50 calibrated yr B.P. (cal yr B.P.), similar to theage (5100 to 4900 cal yr B.P.) of the majority ofother finds in the area (15). There were no in-dications from the burial context suggesting thatGök4 was different from other TRB individuals(15, 16), and strontium isotope analyses indicatethat Gök4 was born less than 100 km from the

1Department of Evolutionary Biology, Evolutionary BiologyCentre, Uppsala University, Uppsala, Sweden. 2Centre forGeoGenetics, Natural History Museum of Denmark, Universityof Copenhagen, Copenhagen, Denmark. 3OsteoarchaeologicalResearch Laboratory, Department of Archaeology and Clas-sical Studies, Stockholm University, Stockholm, Sweden. 4De-partment of Medical Epidemiology and Biostatistics, KarolinskaInstitutet, Stockholm, Sweden. 5Science for Life Laboratory,Uppsala University, Uppsala, Sweden.

*To whom correspondence should be addressed. E-mail:pontus.skoglund@ebc.uu.se (P.S.); anders.gotherstrom@ebc.uu.se (A.G.); mattias.jakobsson@ebc.uu.se (M.J.)†These authors contributed equally to this work.

27 APRIL 2012 VOL 336 SCIENCE www.sciencemag.org466

REPORTS

on

May

10,

201

2w

ww

.sci

ence

mag

.org

Dow

nloa

ded

from