High Quality
Open the downloaded document, and select print from the file menu (PDF reader required).
1331
disappears at 2 T, even stronger supercon-ductivity reappears at the astonishinglylarge field of ~10 T.Field-induced superconductivity had been seen before but could be explained ina rather conventional way, with the exter-nal field simply canceling the internalfield that exists in a ferromagnet. Thanksto their neutron-scattering work, Lévy
et al.
can give convincing arguments for anentirely different, quantum critical originfor what they are seeing. A common fea-ture of novel quantum order near QCPs isthat the new states are fragile and are rap-idly destroyed if the crystals are imperfectand contain disorder. This is also the casehere, so in addition to performing difficulthigh-precision physical measurements,Lévy
et al.
had to use state-of-the-art crys-tal growth to obtain material of the requi-site purity. Their work is an experimentaltour de force.In a more general sense, the impor-tance of these advances is the construc-tion—arguably for the first time in thisfield—of a framework for discovery.Careful work producing and then investi-gating QCPs is leading to a series of experimental breakthroughs. This doesnot, however, mean that we now under-stand everything. For example, it is widelyassumed that quantum critical supercon-ductivity is driven by magnetic fluctua-tions. Although this is very likely to betrue, further work will be needed beforethe statement can be made with absolutecertainty. Coupling between magnetic and structural properties ensures that phononswill also still play some role, and thestrength of this coupling has not yet beendetermined in most cases.A broader question is whether oneneeds a fluctuation-driven description atall. QCPs are indisputably accompanied by strong quantum fluctuations, but themean-field free energy landscape also becomes very flat in their vicinity,increasing the susceptibility to phasechanges in which the role of fluctuationsis much less clear. The fact that these new phases need not always be superconduct-ing gives promise for advances that willcontinue to surprise. Progress is rapid and exciting, but there is still much to beachieved.
References
1.F.Lévy
et al
.,
Science
309
,1343 (2005).2.P.Coleman,A.J.Schofield,
Nature
433
,226 (2005).3.N.D.Mathur
et al
.,
Nature
394
,39 (1998).4.H.Q.Yuan
et al
.,
Science
302
,2104(2003).5.S.S.Saxena
et al
.,
Nature
406
,587 (2000).6.S.A.Grigera
et al
.,
Science
306
,1154 (2004).7.K.H.Kim
et al
.,
Phys. Rev. Lett
.
93
,206402 (2004).8.D.Aoki
et al
.,
Nature
413
,613 (2001).9.S.A.Grigera
et al
.,
Science
294
,329 (2001).10.1126/science.1117436
E
xploring microbial diversity is becom-ing more like exploring outer spacewith soil representing a “final fron-tier” that harbors a largely unknown micro- bial universe. There are more than 10
16
prokaryotes in a ton of soil compared to amere 10
11
stars in our galaxy. Astronomershave wisely inferred the population of celestial objects by mathematical infer-ence. Now microbiologists are followingsuit, adopting a similar strategy to estimatethe number of prokaryote taxa in soil. Asshown by Gans
et al.
on page 1387 (
1
), theinferred diversity is staggering—higher than previously thought by almost threeorders of magnitude.The extent of prokaryote diversity has been hotly debated and rightly so.Microbial communities are central tohealth, sustainable cities, agriculture, and most of the planet’s geochemical cycles.Prokaryote communities are also reser-voirs for the discovery of new drugs and metabolic processes. As with any reser-voir, its size is important.Measuring the reservoir of prokaryoticdiversity is not a trivial task. There is broad agreement that the key is to eschew theorganisms themselves and to focus instead on their DNA. If DNA from a single organ-ism is purified and heated, the strands of the double helix separate or “melt.” If youthen slowly cool the DNA, the strands willreassociate or reanneal, and the rate atwhich this happens is affected by the sizeand complexity of the DNA. Big and com- plex DNA reanneals slowly. This fact has been used for the past four decades to esti-mate the size and complexity of genomesfrom individual organisms. Around 15years ago, Torsvik
et al.
(
2
) reasoned that pooled genomic DNA from a microbialcommunity might reanneal like the DNAfrom a large genome. Indeed, they showed that DNA extracted from soil reassociated slowly—so slowly that it resembled agenome that was 7000 times as large as thegenome of a single bacterium. It followsthat there could have been at least 7000 dif-ferent prokaryote taxa in the sample of soilthat they analyzed. At the time, this wasconsidered a mind-boggling number. Evenecologist E.O. Wilson speculated that“microbial diversity was beyond practicalcalculation” (
3
).There is, however, another way to esti-mate prokaryotic diversity in the environ-ment. A biological community has a char-acteristic abundance distribution of itsmember species. The observation and con-templation of these distributions have arich literature in conventional ecology thatis helping rescue microbial ecology fromthe conundrum of how to estimate diver-sity. In principle, if you know the shape of the taxa abundance distribution curve, youknow the diversity. But there is a catch:Typically, for large organisms, speciesabundance distributions have been deter-mined by assessing the abundance of almost all of the species in a sample,which means that you must already knowthe number of species. In the absence of such information, one still can draw uponcertain theoretical considerations (
4
),assume that a particular species distribu-tion pertains, and then make an estimate(
5
). Alternatively, you can fit a curve to thedata you have to make an estimate (
6, 7
).The latter approach has great merit, butgathering enough data to make a sensibledecision about the underlying species dis-tribution pattern is problematic. At pres-ent, most microbiologists attempt to esti-mate diversity by looking at a gene thatoccurs in all cellular life forms. They infer diversity from the number of differentvariants that can be cloned from a sampleof environmental DNA. Unfortunately, thenumber of clones analyzed is typicallysmall (tens to hundreds) compared to thenumber of individual microbes being ana-lyzed (billions or trillions). This is likerandomly sampling a bus load of peopleand then trying to infer the diversity of all people in the world. You would not expectto find many Lithuanians.Gans
et al.
(
1
) and others (
8
) realized that the pattern of DNA reassociation
MICROBIOLOGY
Exploring Microbial Diversity—A Vast Below
T.P.Curtis and W.T.Sloan
T.P.Curtis is at the School of Civil Engineering andGeosciences,University of Newcastle Upon Tyne,Newcastle Upon Tyne,NE1 7RU,UK.E-mail:tom.curtis@ncl.ac.ukW.T.Sloan is at the Departmentof Civil Engineering,University of Glasgow,Glasgow,G12 8LT,UK.E-mail:sloan@civil.gla.ac.uk
www.sciencemag.orgSCIENCEVOL 30926 AUGUST 2005
P
E R S P E C T I V E S
Published by AAAS
1332
kinetics reflects the underlyingdistribution of similar sequen-ces, and hence likely reflectsgenomic diversity. However, theauthors have gone further and show that there is probablyenough data in published DNAreassociation curves for bacter-ial communities to allow dis-crimination between different possible species abundancecurves. By applying new mathe-matical treatment of data, theauthors generate abundancecurves, the most plausible of which suggests that there could be 10
7
distinct prokaryote taxa in10 grams of pristine (free of chemical contaminants) soil (seethe figure). Moreover, rare organ-isms comprise most of this diver-sity. They further determine thatmost of these rare organismscould be wiped out by heavymetal pollution of the soil.The jargon and mathematicalnotation of taxa abundance dis-tributions can be obscure.However, when presented graph-ically the curves are as simpleand as useful as an outline of anunexplored continent (see thefigure). Thus, power law distri- butions (like the zipf distribu-tion) simply describe exponen-tially increasing numbers of species at exponentially decreas-ing abundances. On the other hand, lognormal distributions suggest thatat lower abundances, the number of rarespecies start to decrease. It is no mysterytherefore that wiping out rare species, as inthe case with heavy metal soil pollution,diminishes the ability to distinguish between the two different situations. Gans
et al
. (
1
) point out that the log-Laplace dis-tribution (see the figure) is theoreticallyattractive because it derives from anensemble of lognormal distributions but,as such, it can be made to look a little bitlike either a lognormal or a zipf and there-fore (unsurprisingly) fits well in all cir-cumstances.The work of Gans
et al.
(
1
) simply repre-sents a rough “map” of the whole microbialcommunity of a sample of soil, which may be far more useful right now than an exqui-site description of just part of it. One could argue that the estimates might be affected by reassociation of sequences common tomany taxa. And the curves presented by theauthors certainly constitute tremendousextrapolations (though they are supported by simulations). However, such quibbles areimmaterial if the overall picture is evenapproximately correct. An imperfect, sim- ple map of an entire region can guide anexplorer with more certainty than a perfectrepresentation of one creek. The explorer,thus guided, will produce better maps thatwill better guide more explorers, and so on.Dispensing with the map is like “wildcat-ting” or “swashbuckling” for diversity:exciting and profitable if you make a strike, but ultimately subject to diminishingreturns and the inevitable disinterest of your sponsors. (Sir Walter Raleigh, a 16th-cen-tury English swashbuckler, failed to find ElDorado on the Orinoco and was later beheaded. Today’s sponsors are a little moreunderstanding.) Rational plans and costs for exploring the microbial frontier require a proper mathematical framework for thistask. Indeed, we cannot even resolve somevery basic questions.Thus, exponentially increasing num- bers of taxa at exponentially decreasingabundances means that, in random sam- ples of diversity, a few abundant taxa canturn up again and again. Resolvingwhether the recurrence of abundant taxa isan artifact of sampling (the equivalent of the mapless explorer goingaround in circles) or a genuinereflection of low diversity (
9
) isimportant in resolving thedebate on the extent of prokary-otic diversity.The Gans
et al.
(
1
) report is part of a wider shift in the studyof microbial communities. At present, the science is primar-ily observational. This study,together with other work (
4–7,10
), shows that our powers of observation can be vastlyenhanced by sensible mathe-matical techniques. But obser-vation per se will never beenough to explain the microbialworld, any more than we canexplain the universe by justlooking at the stars. A recentflurry of papers on taxa-arearelationships (
11–14
) may pointto a way forward. These papersare also observational and can-not explain the microbial world either. However, the emergenceof common patterns in datafrom a number of environmentshints at deeper and perhaps uni-versal, underlying processesthat could be expressed for-mally as explicit mathemati-cally supported theory.In the world of microbialecology, we need theory very badly. Almost any consequentialmicrobial community will have10
10
to 10
17
bacteria that could composemore than 10
7
differing taxonomic groupsand countless functional groups. It seemsremarkable that we should even contem- plate trying to understand such vast sys-tems without recourse to some form of the-ory. At the very least we may hope to notonly “substitute one theory for many facts”(
15
) (and we are accruing facts at anastounding rate), but also quantitativelyguide our exploration, facilitate the appli-cation of the methods we have, and develop new tools. We might ultimatelyhope to develop greater powers of predic-tion. This would be a revolutionary devel-opment, but not necessarily an easy one.The findings of Gans
et al.
place a specialduty on theorists in this area of ecology.The challenge is to offer mathematicalguidance in a world we can barely perceiveand not to explore ultimate causes for pat-terns we can readily observe. We thereforeneed to develop simple (
16
) models thatwe can calibrate and then improve. Likethe “map” outlined by Gans
et al.
, simpletheories will mark the beginning, not theend, of the rational exploration of this fron-
Mapping microbial diversity.
The relative abundance of microbial taxacan be described with abundance distributions.The total number of species is the area under the taxa plot line.The precise shape of anygiven curve will depend on the parameters selected.In all cases,themajority of the biomass that we can most readily observe (TerraFrequentata) make up a minority of the diversity.Most taxa are veryhard to find by random sampling (Terra Incognita).
26 AUGUST 2005VOL 309SCIENCEwww.sciencemag.org
P
E R S P E C T I V E S
Published by AAAS
Add a Comment