evolution for the good of the group

8/2/2019 Evolution for the Good of the Group

1/10

380 American Scientist, Volume 96

Featur e Articles

2008 Sigma Xi, The Scientific Research Society. Reproductionwith permission only. Contact [email protected].

Is evolution a team sport, or is thecontest for survival played outstrictly between individuals? Theres

no question that natural selection actson individual organisms: Those withfavorable traits are more likely to passalong their genes to the next genera-tion. But perhaps similar processescould operate at other levels of the bio-logical hierarchy. In this way naturalselection could perpetuate traits thatare favorable not to an individual butto a social unit such as a flock or a col-ony, or to an entire species, or even toan ecosystem made up of many spe-cies. The underlying question is: Can

biological traits evolve for the goodof the group?

Many early biologists accepted theidea of group selection without think-ing very critically about it. For exam-ple, herds of grazing animals might bedescribed as evolving to conserve theirfood supply over the long term. Herds

that exercised restraint would be morelikely to survive than those that quick-ly exhausted a critical resource. Other

biologists, looking more closely at sucharguments, found a flaw in the rea-soning. Prudently managing a sharedresource benefits all members of agroup, including any cheaters whoconsume more than their share. Genesassociated with cheating would there-fore spread through the group, and thepropensity for cooperative resourcemanagement would be undermined.The situation is all too familiar in hu-man experience; it is the phenomenonthat Garrett Hardin famously namedthe tragedy of the commons.

By the middle of the 1960s, ideasbased on group selection were in deepdisfavor, and the term itself was avoid-ed in textbooks and the scholarly lit-erature. When biologists observed be-havior that appeared to benefit groupsor species, they strove to explain it interms of strictly individualistic selec-tion. For example, animals might coop-erate because they have genes in com-mon (kin selection) or because of thelikelihood of reciprocal aid in the fu-ture. In this way, apparent altruism was

interpreted as enlightened self-interest.It became almost mandatory for au-thors to assure their readers that groupselection was not being invoked.

The time has come for a careful andforthright reassessment of group selec-tion in evolutionary thinking. The mostnaive form of group selectionwhichaxiomatically assumes that behaviorsevolve for the good of the groupisclearly untenable. Nevertheless, traitswith public benefits and private costsdo evolve by natural selection. Just be-

cause cheaters have a within-group ad-

vantage doesnt mean that they prevailin the total population. Within-groupselection is opposed by between-group

selection, and the final outcome de-pends on the relative strength of theseeffects. Rather than categorically reject-ing group selection and for the goodof the group thinking, we need toevaluate the balance between levels ofselection on a case-by-case basis.

Russian DollsTo think clearly about group selection,it is important to compare the survivaland reproduction of individuals inthe right way. The problem with forthe good of the group behaviors isthat they are locally disadvantageous.A prudent member of the herd mightgain from conserving resources, butcheaters within the same group gaineven more. Natural selection is basedon relative fitness. If solid citizens areless fit than cheaters within their owngroup, then something more is re-quired to explain how they can evolvein the total population. That somethingis a positive fitness difference at a larg-er scale. Groups of solid citizens aremore fit than groups of cheaters.

These interacting layers of compe-tition and evolution are like Russianmatryoshka dolls nested one withinanother. At each level in the hierarchynatural selection favors a different setof adaptations. Selection between indi-viduals within groups favors cheating

behaviors, even at the expense of thegroup as a whole. Selection betweengroups within the total population fa-vors behaviors that increase the relativefitness of the whole groupalthoughthese behaviors, too, can have nega-

tive effects at a still-larger scale. We

Evolution for the Good of the Group

The process known as group selection was once accepted unthinkingly,then was widely discredited; its time for a more discriminating assessment

David Sloan Wilson and Edward O. Wilson

David Sloan Wilson and Edward O. Wilson areevolutionary biologists who recently collaboratedon a major review article, Rethinking the Theo-retical Foundation of Sociobiology, in the Quar-terly Review of Biology. David Sloan Wilson is

professor of biology and anthropology at Bingham-ton University in New York. He earned his Ph.D.at Michigan State University. His most recentbook is Evolution for Everyone: How DarwinsTheory Can Change the Way We Think AboutOur Lives. Edward O. Wilson is professor andcurator of entomology at the Museum of Compar-ative Zoology of Harvard University, which is alsowhere he earned his Ph.D. He is a recipient of theNational Medal of Science, the Crafoord Prize andtwo Pulitzer prizes. His forthcoming book withBert Hlldobler is titled The Superorganism.Address for David Sloan Wilson: Biology Depart-ment, Binghamton University, Binghamton, NY

13902. Internet: [email protected]


2/10

2008 SeptemberOctober 381www.americanscientist.org 2008 Sigma Xi, The Scientific Research Society. Reproduction

with permission only. Contact [email protected].

can extend the hierarchy downward tostudy selection between genes withina single organism, or upward to studyselection between even higher-levelentities. The general rule is: Adapta-tion at level X requires a correspond-ing process of selection at level X andtends to be undermined by selection atlower levels.

This way of thinking about evo-lution is called multilevel selection(MLS) theory. Although the termmultilevel selection is newer thanthe term group selection, the Rus-sian-doll logic has been present fromthe beginning, going back to the worksof Darwin.

Darwin would not have been mo-tivated to think about group selectionwere it not for the existence of traitsthat are selectively disadvantageouswithin groups. In a famous passage

from Descent of Man, he notes that

morally upright people do not have anobvious advantage over less-uprightpeople within their own tribe, butthat tribes of morally upright peoplewould robustly outcompete othertribes. He concluded by saying ... andthis would be natural selection. Dar-win was clearly employing the Rus-sian-doll logic of MLS theory in this

passage. He did not comment on theirony that morality expressed withingroups can become morally problem-atic in between-group interactions, buthis hypothetical example perfectly il-lustrates the general rule stated above,which makes adaptations at one levelpart of the problem at higher levels.

Darwins idea was elaborated byother evolutionists during the firsthalf of the 20th century, notably thethree founders of population genet-ics: Ronald Fisher, J. B. S. Haldane and

Sewall Wright, who gave the idea a

mathematical foundation. Their mod-els differed in detail, but all embodiedthe Russian-doll logic of MLS theory.Unfortunately, many biologists wereunaware of these models and thoughtnaively about group selection, as wehave already noted.

The events that led to the wide-spread rejection of group selection

make an interesting story. In the late1950s the University of Chicago wasa hotbed of naive group selectionism.George C. Williams, a newly arrivedpostdoctoral associate, attended a sem-inar by Alfred E. Emerson, a termite

biologist who likened all of nature to atermite colony. Williams was disgust-ed by the naivete of Emersons claimsand resolved to write a book clarifyingthe use and misuse of thinking aboutadaptation and natural selection. AsWilliams was writing, the Scottish bi-

ologist Vero C. Wynne-Edwards pub-

Figure 1. Sports such as bicycle racing illustrate some of the same conflicts between group and individual interests seen in biological evolu-

tion. A small group of cyclists can sustain a higher speed if they coordinate their efforts, with each rider taking a turn at the front of the group,where wind resistance is greatest, and then resting in the wake of the others. Each rider is competing as an individual and so has an incentive

to conserve energy until the last moments of the race. On the other hand, if too many riders shirk their duty at the front, the entire group will

surely be overtaken by the pelotonthe main mass of riders. Analogous situations arise in evolutionary biology, where individuals competewithin a group while groups compete against one another. The race shown above is the fifth stage of the 2006 Tour de France. A breakawaygroup of eight riders took an early lead, but all of them were caught by the peloton well before the finish of the 225-kilometer stage.

Franck Fife/AFP/Getty Images


3/10

382 American Scientist, Volume 96 2008 Sigma Xi, The Scientific Research Society. Reproduction


lished a book titled Animal Dispersion

in Relation to Social Behavior. Wynne-Edwards interpreted a vast array ofanimal social behaviors as adaptationsto prevent the overexploitation of re-sources. He was aware that such pru-dence might be selectively disadvan-tageous within groups but assumedthat it would evolve by group selec-tion. Wynne-Edwardss overreachingclaims were widely discussed and crit-icized. When Williams published his

bookAdaptation and Natural Selection in1966, evolutionary biologists were in amood to reach a consensus.

Williams forcefully asserted for abroad audience what a smaller groupof experts already knew: Traits thatare for the good of the group mightevolve, but only by a process of group-level selection that is strong enough toovercome selection within groups. Akey passage makes this point:

It is universally conceded by thosewho have seriously concernedthemselves with this problem ...that such group-related adapta-tions must be attributed to thenatural selection of alternativegroups of individuals and thatthe natural selection of alternativealleles within populations will beopposed to this development. Iam in entire agreement with thereasoning behind this conclusion.Only by a theory of between-group selection could we achievea scientific explanation of group-related adaptations.

But Williams did not let the matter

rest there. Having insisted that group

selection is required to explain the evo-

lution of group-level adaptations, hewent on to claim that group selectionis almost invariably weak, comparedto selection within groups. In his ownwords: Group-related adaptations donot, in fact, exist.

To summarize, Wynne-Edwards andWilliams both accepted the Russian-doll logic of MLS theory and both madeextreme claims that one level of selec-tion routinely prevails over the other.Evolutionists found Williams more per-suasive, leading to the widespread re-

jection of group selection. Generationsof students were taught that group-lev-el adaptations can evolve in principle,

but do not evolve in practice, makingfor the good of the group thinking

just plain wrong. Behaviors that appearto be for the good of the group must beexplained in ways that are consistentwith individual self-interest, such as kinselection or reciprocity.

In hindsight, it has become clear thatboth claims were too extreme. The bal-ance between levels of selection can tiltin either direction. Between-group selec-

tion is sometimes a weak evolutionaryforce, as Williams supposed, but it canalso be very strong, enabling groups toevolve into veritable superorganisms.There is no single formula; answers must

be worked out on a case-by-case basis.Some examples of group-level adapta-tions will reinforce this point.

Marine Invertebrate ColoniesA good place to begin is with organ-isms that seem to blur the boundary

between the individual and the group.

For a spectacular example consider

the elegant marine life forms calledsiphonophores, which include the Por-tuguese man-of-war. Many marine in-vertebrates live as colonies of individualorganisms that are physically attachedto one another. In some cases, such ascorals, the members of a colony are un-differentiated and appear to functionas autonomous units. A siphonophore,

in contrast, is a colony made up of in-dividuals with specialized forms andfunctions. Some members of the colonyprovide locomotion; others are chargedwith stinging and capturing prey; oth-ers do the work of digestion and as-similation. There is even a rudimentarynervous system. The siphonophores, inother words, have created a new kind oforganism by turning simpler organismsinto organs. Can their specializations

be interpreted as for the good of thecolony in the same way that organs

can be interpreted as for the good ofthe individual?Most evolutionists would say yes,

but it is instructive to review the rea-soning that enabled an example suchas this to be regarded as compatiblewith the rejection of group selection.Insofar as siphonophore colonies grow

by asexual reproduction, their mem-bers are genetically identical. Of coursethe colony can be regarded as an adap-tive unit, similar to a multicellularorganism! What does this have to dowith group selection?

To answer this question, we merelyneed to employ the Russian-doll log-ic of MLS theory. If all members of acolony are genetically identical, therecan be no within-colony selection. Ad-aptations can evolve only by virtue ofcausing some colonies to survive andreproduce better than other colonies.The example represents a case of pure

between-colony selection, not a denialof between-colony selection. Moreover,there is no warrant for assuming thatmembers of a single colonyor the

cells of a multicellular organismaregenetically identical. Mutations arisewith every cell division, creating apotential for within-colony selection.Highly sophisticated adaptationsevolved to suppress selection amonggenes and cell lineages within multicel-lular organisms and presumably alsofor siphonophore colonies. The bot-tom line is that division of labor andother design features of siphonophorecolonies evolved by virtue of between-colony selection. If the example fails to

impress, it is because it seems so obvi-

Figure 2. Multilevel selection theory describes a hierarchy of evolutionary processes organizedlike nested Russian dolls. At the innermost level, within a single organism, genes contend

with each other for a place in the next generation; within a group of organisms, selection actson the relative fitness of individuals; groups within a population also differ in their collective

survival and reproduction. Adaptation at any given level tends to be undermined by selectionat lower levels. At even higher levels(not shown), populations, multispecies communities andwhole ecosystems can be subject to selection.


4/10



ous. Why, then, did evolutionary biolo-gists accept the categorical statementthat higher-level selection is invariablytrumped by lower-level selection?

Cellular Slime MoldsAnother species on the borderline

between individual and group is thecellular slime mold Dictyostelium dis-

coideum. For most of its lifetime Dictyo-stelium consists of single-celled amoe-

bae that forage in the soil, feeding onbacteria and other protists. When foodgrows scarce, thousands of amoebaecome together to form a body called aslug, which migrates toward light overdistances of up to 20 centimeters, thendevelops into a ball of reproductivespore cells held aloft by a nonrepro-ductive stalk. Cells in the stalk eventu-ally die, but the spores are dispersed

(usually by sticking to a passing inver-tebrate) to start a new generation.

In the 1950s Dictyostelium became amodel organism for the study of de-velopment, but the fact that the stalkcells appear to sacrifice their lives forthe good of the group attracted littleattention. Only later did Dictyosteliumalso become a model organism for the

study of multilevel selection.Before we take up the spectacular

example of suicidal stalk formation,consider the subtler issues of slug for-mation, movement and orientation.Creating a slug requires the cells to se-crete a polysaccharide matrix. Forwardmovement of the slug is accomplished

by spiral motion of the cells within thematrix. Orientation toward light re-quires communication to coordinatemovement. These are all examples of

public goods that provide the samebenefits for everyone but require in-dividual effort. Even if the effort isnearly cost-free, as might be the casewith orientation, the benefits are stillshared (all individuals within a slugmove to the same place), and so the fit-ness differences required to explain theadaptation reside at the group level

(some slugs move to better places thanothers). There is no plausible way toexplain these collective adaptationson the basis of within-group selection;they are group-level adaptations andcan be accepted at face value as beingfor the good of the group.

Stalk formation appears to be theultimate in self-sacrifice for the ben-efit of others. Individual cells have apowerful incentive to become repro-ductive spores rather than inert com-

Figure 3. A siphonophore is a colonial association of simpler organisms, which have differentiated to perform functions that benefit the entire

colony. In this specimen ofMarrus orthocanna the uppermost structure is a float, or pneumatophore; the semitransparent, pitcherlike append-ages below the float are nectophores, which provide locomotion; the curly tentacles at the lower right are attached to feeding polyps; the bright

streaks of orange at the upper right are stinging cells. Most of the structure is transparent; color comes from ingested food. In siphonophores

natural selection acts both within the colony and between colonies. Photograph courtesy of Casey Dunn, Brown University.


5/10



ponents of the stalk; and yet, if all thecells succeeded in becoming spores,there would be no stalk to promotedispersion. Selfish strains have beenobserved both in the laboratory and inthe field. The problem is to show how

between-group selection can be strongenough to counterbalance what ap-pears to be an extreme selective advan-tage for selfishness within groups.

One possibility is that cell recogni-tion enables amoebae to aggregate onlywith others that carry identical genes,so that each slug is genetically uniform.In this case natural selection would bepurely at the group level, except formutations, as with siphonophores.Another possibility is a lottery processanalogous to the human social conven-tion of drawing straws. In this case,

some individuals make the ultimatesacrifice to become stalk cells, but theyare a random sample, so there is nogenetic selection within a slug. Cur-rent research indicates that neither ofthese explanations is entirely correct.Within-group selection does operate toa degree, favoring traits that are mal-adaptive at the group level. It would be

just plain wrong to interpret these traitsas being for the good of the group.

Some of the issues that arise insiphonophores and in Dictyostelium

have also been explored in a labora-

tory setting in experiments with thebacterium Pseudomonas fluorescens.When this species is cultured in an un-stirred broth, the cells soon consumemost of the oxygen in the bulk of themedium, so only a thin layer near thesurface remains habitable. A spontane-ous mutation called wrinkly spreadercauses cells to secrete a cellulosic poly-mer that forms a mat and helps themcolonize the water surface. Productionof the polymer is metabolically expen-sive, which means that nonproducingcheaters have the highest relative fit-ness within the mat; they get the ben-efit of the mat without contributing toits upkeep. However, if the propor-tion of cheaters grows too high, theyare undone by their own success. Themat disintegrates, and the entire group

sinks into the anoxic broth. Experi-ments by Paul B. Rainey and KatrinaRainey have shown that the wrinklyspreader trait is maintained in thepopulation by group selection, eventhough it is disadvantageous withinany one group.

Meiotic DriveThe three examples described so farillustrate that between-group selectioncan be a significant evolutionary forceand that group-level adaptations can

be accepted at face value as for the

good of the group. Now lets use MLStheory to think about selection amonggenes within individual organisms.

Meiosis is the reduction divisionthat separates pairs of chromosomesin gametes. It is usually a fair process:The two genes at each locus on eachchromosome have an equal chance of

being represented in the next genera-tion. The fairness of meiosis suppressesnatural selection among genes withinan individual and concentrates selec-tion at the individual level (or above).That is why individuals are so func-tionally organized that they deservethe term organism.

Some genes find ways to break therules of meiosis, however, gaining awithin-individual advantage througha process called meiotic drive. In the

fruit fly Drosophila, a complex of genesknown as segregation distorter, or SD,perpetuates itself even though it is det-rimental to some of the flies that carryit. In males that are heterozygous forthe SD genes, sperm cells bearing thegenes secrete substances toxic to spermlacking SD; as a result, almost all thesurviving sperm are SD-positive, andthe gene complex is overrepresentedin the male flys offspring. This is ad-vantageous to the genes but not to theflies: Males that receive SD genes from

both parents tend to be infertile.

Figure 4. The cellular slime moldDictyostelium discoides offers another instance in which individuals compete within a group and cooperate toperpetuate the group as a whole. The single-celled amoebae ofDictyostelium assemble to form a multicellular slug, which develops into a fruit-ing body. Only the spore cells in the head of the fruiting body pass their genes to the next generation, but the spores can successfully disperseonly if other cells sacrifice their chance to reproduce by forming a stalk. The scanning electron micrograph at left shows various stages in the

formation of the slug and fruiting body. The cross section at right shows stalk cells dying as they fill with vacuoles, whereas cells destined to be-come spores remain viable. Images courtesy of Larry Blanton of North Carolina State University and Mark J. Grimson of Texas Tech University.


6/10



This example shows how the Russian-doll logic of MLS theory can be applied toall levels of the biological hierarchy. Mei-otic drive is for the good of the gene

but it would be futile and erroneous tocall it for the good of the individual. Itsimply isnt. On the other hand, it is fullywarranted to call an adaptation such as aturtles shell for the good of the individ-

ual because it evolved by a process ofselection among individuals. The samecriteria that establish the legitimacy offor the good of the group thinking alsoestablish the legitimacy of for the goodof the individual thinking.

Group Selection or Selfish Genes?Our examples show how MLS theorycan be used to evaluate the balance be-tween levels of selection on a case-by-case basis, rather than declaring cat-egorically that group selection always

prevails (Wynne-Edwards) or neverprevails (Williams). Why wasnt thisjudicious middle ground achieved longago? The answer comes in part from asubtle shift in perspective that departsfrom the usual logic of MLS theory.

Consider the turtles shell, a standardexample of an individual-level adapta-tion. Genes that contribute to this adap-tation are not more fit than alternativegenes within any single turtle, but theyare more fit than the alternative genes inthe total population. That is just anotherway of saying that they evolve. Thuswe can say that turtle shells evolve bygene-level selection, defined as the fit-ness of genes, all things considered. It isimportant to stress that nothing aboutthe biological example has changed. Theinformation for making the Russian-dollcomparisons is still available; we aresimply choosing to ignore the fact thatthe gene achieves its success by virtue of

between-individual selection rather thanwithin-individual selection.

The same gambit can be employedfor individuals within groups. The

wrinkly spreader strain is not more fitthan the cheater strain within any mat,

but it is more fit in the total populationwhenever mats evolve. In this fashion,all adaptations that evolve at level X,according to the Russian-doll compari-sons, can be portrayed as lower-leveladaptations. Ultimately, everything thatevolves can be portrayed as adaptiveat the gene level, regardless of wherethe fitness differences are located in the

biological hierarchy.This genes eye view of evolution

has always played a role in population

genetics, where it is called averageeffects because it involves averagingthe effects of alternative genes acrossall contexts to determine what evolvesin the total population. Williams ex-pounded on the concept in Adaptationand Natural Selection, where he aptlycalled it a bookkeeping method. Mostpeople, however, know average effects

under the name selfish genes, fromthe influential book of that title pub-lished by Richard Dawkins in 1976.

By itself, the genes eye view inter-prets everything that evolves as adap-tive at the gene level; however, evenselfish-gene theorists need a way to dis-tinguish a gene for meiotic drive from agene for a turtles shell. A concept calledvehicle of selection is added to dis-tinguish these cases. The individual isthe vehicle of selection in the case of theturtles shell because all of the genes in

the turtle are in the same boat as faras their fitness is concerned. It shouldbe obvious that the vehicle conceptwithin selfish-gene theory duplicatesthe Russian-doll logic of MLS theory.

The point is that the Russian-doll logicis not arbitrary. It is essential for makingsense of what we conventionally mean

by for the good of the individual and,by extension, adaptations at all levels ofthe biological hierarchy. Numerous theo-

ries have been proposed to explain theevolution of apparent altruism and othersolid-citizen behaviors without invok-ing group selection. When these theoriesare examined in detail, they are usually

based on comparisons that depart fromthe Russian-doll logic of MLS theory.When reformulated in terms of the Rus-sian-doll logic, they require group-level

selection after all.

Eusocial Insect ColoniesNowhere has the pendulum of scientificthought swung more widely than in thestudy of eusocial insects (ants, wasps,

bees and termites). Long before scienceexisted as a cultural practice, honeybeeswere lauded as an example from na-ture of individuals acting for the goodof the group. The great entomologistWilliam Morton Wheeler was the firstto describe eusocial insect colonies as

superorganisms in 1911.It might seem that if any biologicalphenomenon could survive the rejec-tion of group selection, it would bethe eusocial insects. Instead, eusocialinsects became a particular focus of at-tempts to reformulate all evolutionaryevents as consequences of individualmotives and acts.

The key to this program was kin se-lection, presented in the 1960s by W. D.

Figure 5. The bacterium Pseudomonas fluorescens illustrates tradeoffs between individualand group selection in experiments conducted by Paul B. Rainey and Katrina Rainey of the

University of Aukland in New Zealand. In an unstirred broth, Pseudomonas cells can surviveonly at the surface. Cells with a gene called wrinkly spreader (green) secrete a polymer thatforms a buoyant mat (left). Producing the polymer has a metabolic cost, which limits the cellsrate of growth. Nonsecreting mutants (yellow) can live as freeloaders, benefiting from theirneighbors exertions. The freeloader cells reproduce faster; when they become too numerous,however, the entire mat disintegrates and sinks

(right), in a tragedy of the commons.


7/10



Hamilton as an alternative to group se-lection. According to Hamiltons rule,an altruistic behavior evolves whenbrc > 0, where c is the cost to the al-truist, b is the benefit to the recipient,and r is the coefficient of genealogicalrelatedness, which ranges from 0 (forunrelated individuals) to 1 (for iden-tical twins). When this inequality issatisfied, an altruistic act increases theabsolute number of copies of the altru-istic gene. For example, in some eu-social insects r is as high as 0.75, so thata cost of 1 is worth paying if the benefitto a colony-mate is at least 1.33.

Hamilton later realized that increas-ing the number of copies of an altruisticgene is not the same as increasing thefitness of altruists, relative to non-altru-ists in the same group. Working withanother theoretical biologist, George

Price, Hamilton reformulated his theoryin the 1970s using the Russian-doll logicof MLS. To his surprise, he discoveredthat altruism is selectively disadvanta-geous within groups, even when thegroups are composed of kin; it evolvesin the total population only by virtue of

between-group selection. His rule stillpredicted when altruism would evolvein the total population, but the theorywas no longer an alternative to groupselection; on the contrary, it required be-tween-group selection. The coefficient

of relatedness, which was originally

interpreted as the probability of shar-ing genes identical by descent, couldnow be interpreted more generally asan index of genetic variation betweengroups. When r = 0, the groups are ran-domly formed. When r = 1, membersof each group are genetically identical,and natural selection is concentratedentirely at the group level.

These developments should haveled to the immediate revival of MLStheory in the 1970s, but intellectualmomentum is not so easily reversed.From the present perspective it is clearthat the vast majority of traits that en-able social insect colonies to functionas adaptive units can be accepted asproducts of between-colony selection.Some traits do evolve by within-colo-ny selection, but these tend to disruptcolony-level functional organization.

Insect Factories and FortressesOne of us (E. O. Wilson) has describeda social insect colony as a factory in-side a fortress. Both factories and for-tresses suggest group-level functionaldesign. The physical architecture ofa termite mound, the caste system ofan ant colony, the defensive responseof a wasp nest and the foraging sys-tem of a honeybee colony are all com-munal activities that can be evaluatedfor their collective efficacy in the same

way as the adaptations of individual

organisms. For example, when a ter-mite mound is damaged, workers rushto repair the breach, much as variouskinds of cells repair a skin wound inan individual organism.

How did the termite behaviorevolve? One possibility is that moundrepair reflects within-colony selection:Individual termites with a genotype

that predisposed them to make repairssurvived and reproduced better thancolony-mates with other genotypes.Far more plausible in this case is be-tween-colony selection: Colonies inwhich the mound-repair genotype wascommon were more likely to surviveand reproduce than other colonies.

Kin selection can be insightful in itsown right, but it is profoundly mis-leading when interpreted as an alter-native to between-colony selection. Italso misleads us about the nature of

variation among groups. Kin selectionassumes that behavioral similarity isproportional to genetic similarity; theonly way for a group to be behaviorallyuniform, for example, is to be genetical-ly uniform. However, colony-level ad-aptations in the eusocial insects usuallyrely on principles of self-organization,whereby relatively simple behavioralrules at the individual level interact toproduce complex adaptive behaviors atthe colony level. In self-organizing sys-tems, small changes in the lower-levelrules often produce large changes inthe properties of the system as a whole.As a result, modest genetic variationamong social insect colonies can pro-duce substantial behavioral variation atthe colony level, providing the raw ma-terial for colony-level selection. A corre-lation between genetic and phenotypicvariation is required, but a high degreeof genetic variation between groups(high values of r) is not. Even randomgenetic variation (r = 0) can suffice.

Incipient social behavior is observedeven in some solitary insects that do

not share the peculiar kinship struc-ture of the eusocial species. Bees insome species of the genera Certatinaand Lasioglossum normally live alone,

but when individuals are forced to livetogether, they spontaneously divideup tasks such as foraging, tunnelingand nest guarding. Each bee is alreadyprogrammed to perform these taskson its own, switching among themas needed. But when a bee within agroup encounters another bee that isalready performing a task, it moves on

to other tasks. Another pre-adaptation

adults

gametes

+

+++ ++ + +++ ++ +

+

+

+

+

+

+

+

Figure 6. Meiotic drive reveals a conflict of interest between individual organisms and the

genes they contain. In the fruit fly Drosophila melanogastergenes that form a complex calledSD (for segregation distorter) act to increase their own frequency even though the genes aredetrimental to some of the flies in which they occur. In males that are heterozygous for SD,half the spermatozoa carry the SD genes (marked + in the upper panel) and half do not ().But the + sperm cells produce substances that kill cells, so that almost all the viablesperm are +. Egg development in females is unaffected by SD. Because few sperm cellssurvive, the SD genes are overrepresented in the offspring generation. The trait persists eventhough males homozygous for the SD genes (++) tend to be infertile.


8/10



for social life is dominance behavior,which results in a prompt division oflabor into reproductive and non-repro-ductive roles as soon as solitary beesare forced together. These adaptationsin solitary insects that build and provi-sion nests are spring-loaded to be-come adaptations for colonial life withminor modifications.

The Group MindThe concept of a group possessing acollective mind might seem like sci-ence fiction, but it follows directly fromMLS theory and has been documentedin impressive detail in the eusocial in-sects. Consider the parallel examplesof a rhesus monkey making a decisionand a honeybee colony making a deci-sion. The monkey has been trained towatch a number of dots moving left orright on a screen and to turn its head in

the direction that most of the dots aremoving. Inside its brain, some neuronsfire in response to right-moving dotsand others fire in response to left-mov-ing dots. A decision is made when oneof these classes of neurons fires abovea threshold rate. The honeybee colonyhas evolved to evaluate potential nestsites during the swarming phase ofits life cycle. Higher-quality nest sitesattract more scouts, and a decision ismade when a threshold number ofscouts is exceeded for one of the nestsites. The decision-making processis similar in the two cases, with indi-vidual bees playing the same role asmonkey neurons.

Many other examples of distrib-uted information processinggroupmindscould be cited for the eusocialinsects. Even more exciting, groupminds are probably not restricted tothe eusocial insects. As we have justshown, between-group selection doesnot necessarily require extreme geneticvariation between groups. In the caseof information processing, the collec-

tive benefits of making a wise deci-sion can be great and the within-groupcosts can be low. Even individuals whoare competing with one another canprofit by making a wise collective deci-sion about where to forage, providinga bigger pie to fight over. Evolutionary

biologists are provisionally identifyinggroup minds in species as diverse asAfrican buffalo and American toad tad-poles. Theoretical models of consensusdecision-making are often described interms of self-interest because everyone

in the group stands to gain from mak-

ing a good decision. However, thereare no fitness differences in a win-winsituation. Employing the Russian-dollcomparisons of MLS theory clearly re-veals that collective decision-makingevolves by between-group selection.

Group Selection in the LaboratoryGroup selection can easily be studied

in the laboratory. One merely createsa population of groups and selects fora group-level trait, in the same waythat individual-level traits have beenartificially selected for centuries. Theseexperiments almost invariably dem-onstrate a response to group-levelselection. Often the magnitude of theresponse is greater than anticipated

because of the way self-organizing sys-tems amplify phenotypic variation.

William Muir of Purdue Universitycompared two kinds of selection for egg

productivity in hens. The hens were keptin cages, with several hens per cage. Inthe first experiment, the most produc-tive hen within each cage was selected to

breed the next generation (within-groupselection). In the second experiment, allhens within the most productive cageswere used to breed the next generation(between-group selection). In the firstexperiment, the most productive henin each cage achieved her productivitylargely by bullying the other hens. After

six generations, a hyper-aggressive strainhad been produced, with hens pluckingeach others feathers in incessant attacksthat were sometimes fatal. Egg produc-tivity plummeted over the course of theexperiment, even though the most pro-ductive hens had been chosen in everygeneration. In the second experiment,group-level selection resulted in a docile

strain of hens, and egg productivity in-creased 160 percent in six generations.

Laboratory experiments can also beused to explore selection at the level ofmulti-species communities. In the 1990sCharles Goodnight of the University ofVermont performed a series of ingeniousexperiments with two species of flour

beetle, Tribolium castaneum and T. confu-sum. Goodnight placed equal numbersof beetles of the two species in vials andallowed them to interact and reproducefor several weeks. He then ranked the

vialseach a two-species communityaccording to the population density ofcastaneum. Vials that scored highest wereused as parents to create a new gener-ation of vials. Even though the vials wereselected based on the density of one ofthe species (the community-level phe-notype), both species were transferredacross generations as part of the samecommunity, and each vial was startedwith equal numbers of each species. Inother words, the unit of selection in the

Figure 7. Cooperative and even self-sacrificial behavior by termites is most readily explainedby selection acting at the level of the colony rather than the individual. The image shows con-

struction of a tunnel linking a laboratory-maintained termite colonys nest to a food source.Soldier-caste termites, which are smaller and darker, take up sentry positions facing outward

along the new trail route, while workers (larger, with light-colored abdomens) extend thearched tunnel. Neither of these behaviors seems likely to enhance the survival of individual

termites compared with colony-mates that stay away from the dangerously exposed construc-tion zone, but colonies in which termites build covered galleries to protect them while forag-

ing have a clear advantage over colonies that lack this capacity. The termites are of the species

Nasutitermes corniger. Photograph courtesy of Barbara L. Thorne, University of Maryland.


9/10



experiment was the whole community.Goodnight observed a response to com-munity-level selection: The phenotypictrait (density of castaneum) increasedover the course of the experiment, justas if he had been selecting an individual-level trait. In a second round of experi-ments, new communities were formed

by combining the selected line of casta-

neum with confusum from the originalstock population. In these reconstructedcommunities the evolved trait failed toappear. In this way Goodnight was ableto show that community-level selectionhad produced genetic changes in bothspecies, which interacted to increase thedensity of castaneum.

At a lower level of the biological hi-erarchy, an experiment like this onewould be nothing out of the ordinary.

If you breed Drosophila for a trait suchas wing length, you can expect to seechanges in genes on two or more chro-mosomes. In Goodnights experimentthe community plays the role of theDrosophila individual, and the two

beetle species are like the interactinggenes on separate chromosomes.

With William Swenson, one of us (Da-

vid Sloan Wilson) has ratcheted the levelof selection still higher in experimentswith microbial ecosystems cultured intest tubes. Each test tube was inoculatedwith millions of microorganisms from asingle well-mixed source (pond water),so that initial genetic variation betweentest tubes was negligible. Nevertheless,the test tubes quickly diverged from oneanother in measurable traits such aspHor ability to degrade a toxic substance.

Both of these traits are properties of thephysical environment influenced by liv-ing processes. Thus we were selecting atrait associated with an entire ecosystem,not with any individual or even any onespecies. Otherwise, the procedure waslittle different from selecting an indi-vidual-level trait such as wing length inDrosophila. Ecosystems that scored high-

est on the phenotypic trait were selectedas parents to create a new generationof ecosystems. There was a response toselection, demonstrating that heritablephenotypic variation can exist at the lev-el of whole ecosystems.

An unintended version of this ex-periment appears to have taken placein the selection of strains of yeasts and

bacteria that produce kefir, a yogurt-likedrink. In this case the ecosystem-level

group selection for Tribolium castaneum(9 generations)

naive confusumnaive castaneum

co-evolvedconfusum

co-evolvedcastaneum

co-evolved confusumand castaneum

Figure 8. Selection at the community level was demonstrated in experiments done by Charles Goodnight of the University of Vermont. Twospecies of flour beetle, Tribolium castaneum(black) and T. confusum(red), were reared together in vials and selected for high population den-sity of T. castaneum. In a preliminary round of experiments (upper part of diagram), beetles from vials that attained the highest density becameparents of the next generation. After nine repetitions of this procedure, the propensity to live at high density was a heritable trait: When co-

evolved castaneum and confusum beetles from the selected populations were incubated together (center panel), they continued to reach highdensity. However, when co-evolved confusum beetles were placed with naive castaneum insects (left panel), no significant enhancement ofdensity was observed; likewise co-evolved castaneum and naive confusum exhibited no effect (right panel). The inference is that selectioncaused genetic changes that are expressed only in the dual-species community, not in either species alone.


10/10


with permission only Contact perms@amsci org

phenotypic trait is the taste of the kefirand its health benefits, causing some

batches to be favored over others. Notonly has this process selected a multi-species microbial community, but thecommunity has evolved to aggregateinto clusters held together by a polysac-charide matrix so that it can disperseacross batches as a single unit.

These experiments raise the excitingprospect of creating designer groups,communities and ecosystems that per-form useful functions such as degrad-ing toxins, using artificial selection pro-cedures that have been practiced onindividual organisms for centuries. Theyalso encourage us to look for examplesof natural selection at the level of com-munities and ecosystems.

Human EvolutionIndividualistic theories of human behav-

ior have been in vogue for the past half-century. Before then, it was common tothink of human societies as being likesingle organisms. Indeed, the tradition offunctionalism in the social sciences bearsa strong resemblance to naive group se-lection in biology. Of course the uncriti-cal application of for the good of thegroup thinking is no more justified inhuman affairs than elsewhere; between-group selection can prevail only when itis stronger than within-group selection.Having said this, however, we also as-sert that group selection has been a verypowerful force in human genetic andcultural evolution.

Converging lines of evidence suggestthat the key difference between humanancestors and other primate species wasthe suppression of fitness differenceswithin groups, concentrating selectionat the group level. Hunter-gatherer so-cieties are fiercely egalitarian. Meat isscrupulously shared; aspiring alphamales are put in their place; and self-serving behaviors are censured. Un-able to succeed at each others expense,

members of hunter-gatherer groupssucceed primarily by teamwork.

Selection for teamwork probably be-gan very early in human evolution.Human infants spontaneously pointthings out to others, and not merely toget what they want, which chimpan-zees do not do at any age. Symbolicthought, language and the social trans-mission of information are fundamen-tally communal activities that rely ontrustworthy social partners. Exploita-tion, cheating and free riding do exist

in human groups, but what is most

remarkable is the degree to which theyare suppressed. They loom so large inour thoughts partly because we areprimed to suppress them, like a well-adapted immune system.

Teamwork enabled our ancestors tospread throughout Africa and beyond,replacing all other hominid speciesalong the way. While we remained a

single biological species, we diversifiedculturally to occupy hundreds of eco-logical niches, harvesting everythingfrom seeds to whales. The invention ofagriculture added new layers to the bio-logical hierarchy. We now live in groupsof groups of groups.

When we confront the panoramaof human genetic and cultural evolu-tion, are we permitted to think aboutadaptations as being for the good ofthe group? As soon as we employ theRussian-doll logic of MLS theory, the

answer becomes unambiguously yes.The idea that within-group selectioninvariably trumps between-group se-lection is as absurd for ourselves as itis for the eusocial insects.

Seizing the Middle GroundDebates about adaptation and naturalselection can become so contentiousthat they are called wars. In these wars,the most difficult ground to hold is themiddle ground. In the 1970s Stephen JayGould and Richard Lewontin initiated adebate about the importance of adapta-tions compared to by-products of evolu-tion (spandrels) that became needless-ly polarized. After the fighting was over,it was difficult to imagine what the fusswas about. Of course there are spandrelsin addition to adaptations. Both must beidentified on a case-by-case basis.

The controversy over levels of selec-tion started earlier and is not yet quiteover. InAdaptation and Natural Selection,Williams performed both a service anda disservice. The service was to force-fully assert that adaptation at any given

level requires a process of selection atthe same level and should not be in-voked otherwise. The disservice wasto adopt the extreme view that group-level adaptations do not, in fact, exist.

We think that the time has come todeclare a victory for the middle ground.Future evolutionists will look back andwonder what the fuss was about. Ofcourse natural selection operates at mul-tiple levels of the biological hierarchy.The Russian-doll logic of MLS theorymust be used to evaluate the importance

of each level on a case-by-case basis.

BibliographyBurt, A., and R. Trivers. 2006. Genes in Conflict:

The Biology of Selfish Genetic Elements. Cam-bridge, Mass.: Belknap Press of HarvardUniversity Press.

Goodnight, C. J. 1990. Experimental studiesof community evolution. I: The responseto selection at the community level. II: Theecological basis of the response to commu-nity selection. Evolution 44:16141636.

Hamilton, W. D. 1996. Narrow Roads of GeneLand: The Collected Papers of W. D. Hamilton .Oxford: W. H. Freeman/Spektrum.

Lopitz-Otsoa, F., A. Rementeria, N. Elguezabaland J. Garaizar. 2006. Kefir: A symbioticyeasts-bacteria community with allegedhealthy capabilities. Revista Iberoamericanade Micologa 23:6774.

Okasha, S. 2006. Evolution and the Levels of Se-lection. Oxford: Clarendon Press.

Price, G. R. 1970. Selection and covariance.Nature 227:520521.

Rainey, P. B., and K. Rainey. 2003. Evolutionof cooperation and conflict in experimentalbacterial populations. Nature 425:7274.

Sober, E., and D. S. Wilson. 1998. Unto Oth-ers: The Evolution and Psychology of UnselfishBehavior. Cambridge, Mass.: Harvard Uni-versity Press.

Swenson, W., J. Arendt and D. S. Wilson. 2000.Artificial selection of microbial ecosystemsfor 3-chloroaniline biodegradation. Environ-mental Microbiology 2(5):564571.

Visscher, P. K., T. Seeley and K. Passino. 2006.Group decision making in honey beeswarms.American Scientist 94:220229.

Williams, G. C. 1966. Adaptation and NaturalSelection: A Critique of Some Current Evolu-tionary Thought. Princeton: Princeton Uni-versity Press.

Wilson, D. S. 2002. Darwins Cathedral: Evolu-

tion, Religion, and the Nature of Society. Chi-cago: University of Chicago Press.

Wilson, D. S., and E. O. Wilson. 2007. Rethink-ing the theoretical foundation of sociobiol-ogy. The Quarterly Review of Biology 82:327348.

Wilson, D. S., M. Van Vugt and R. OGorman.2008. Multilevel selection and major evolu-tionary transitions: Implications for psycho-logical science. Current Directions in Psycho-logical Science 17:69.

Wilson, E. O. 1975. Sociobiology: The New Syn-thesis. Cambridge, Mass.: Belknap Press ofHarvard University Press.

Wilson, E. O. 2008. One giant leap: How in-

sects achieved altruism and colonial life.BioScience 58(1):1725.

Wynne-Edwards, Vero Copner. 1962. AnimalDispersion in Relation to Social Behaviour.New York: Hafner Pub. Co.

For relevant Web links, consult this

issue ofAmerican Scientist Online:

http://www.americanscientist.org/

issues/id.74/past.aspx

evolution for the good of the group

Documents