THERE'S no such thing as society. If you accept that evolution is all aboutselfish genes, then the group has no role to play. Survival of the fittest
means survival of the fittest DNA. Which makes you and I mere vehicles in which our genes are hitching a lift on the road to posterity. Or maybe not?
    For forty years this rather bleak, reductionist view of life has reigned supreme. Biologists ridiculed the idea that groups of organisms might gain
a survival advantage over other groups because they shared some beneficial trait. Now that is changing. We are starting to understand that evolution happens
on a variety of levels. Natural selection may favour certain genes, but it can also favour particular societies. Provided a group of individuals can cooperate
without any cheats trying to sneak an unfair advantage, then it may evolve as a single entity.
    In the most recent breakthrough, the high priest of group selection, David Sloan Wilson from Binghamton University in New York state has found that
even groups made up of many different species can possess traits that are passed from one generation to another. To some it looks like the biological
equivalent of quantum weirdness, but this is not just an academic oddity. Wilson's finding could allow biologists to create designer ecosystems that increase plant
productivity, change the "p"H of water or break down pollutants. In fact, the implications of group selection apply to any situation where individuals
congregate  from hens in the farmyard to parasites inside a human body.
    Darwin first suggested group selection to explain why animals sometimes behave in ways that reduce their reproductive success. In an extreme case,
a worker bee is unlikely to produce any offspring at all. How can this be, if natural selection favours those individuals who send more of their progeny
into the next generation than their rivals ? The objection applies to altruistic behaviour of any kind, and Darwin addressed it in his discussion of the
evolution of humans, who show quite unnatural charitable tendencies. The anomaly can be explained, he wrote, if you imagine natural selection
operating among groups of organisms, as well as among individuals. A group of people who are kind and helpful to each other may not do so well individually, but as
a team they may do better than other groups of people, and so the tendency to work as a team spreads through the population.
    For almost a century, the group selection idea went unchallenged. But in the 1960s, some very influential biologists raised serious objections.
Their main argument was that within groups of generous individuals anyone who behaved selfishly would have a huge advantage: so cheats and usurpers would always take over the groups from the inside. Put another way, group selection is a very weak force, compared with selection at the individual level. Over the next
couple of decades, other theories emerged to explain the evolution of altruistic behaviour. They involved helping relatives, or helping only
those who reciprocate the favour, and they were backed up by mathematical models. Mainstream biologists rejected group selection. It was swept under the
carpet and forgotten.

Natural progression

    But the idea is experiencing a revival. In part this is due to the efforts of mavericks like Wilson and his colleague Elliott Sober from the
University of Wisconsin at Madison, who never gave up on group selection. But, perhaps surprisingly, the idea's new respectability can also be traced to a shift
in thinking on the part of biologists who in the past have opposed group selection.
    The newly emerging view of evolution, proposed by John Maynard Smith from the University of Sussex and Eoers Szathmary from the University of
Budapest, Hungary, describes the entire development of life as a series of major transitions in which successively more complex levels of organisation have
become dominant. Each transition was a point when individual entities began working together in a group and natural selection kicked in at a higher
level. When cells joined forces to make multicellular organisms, for example, cells that cooperated fared better than cells that exploited the resources of
the group, because all the cells in an organism have a single, sealed fate. In this new "multilevel selection" view of life, group selection is a natural
progression.

    Maynard Smith is convinced that group selection has been important at certain points in the history of life, particularly between human groups
in more recent times. "There is no doubt that we were way too hasty in trashing group selection," agrees Joel Peck, also at the University of Sussex. "The
theoretical models of the 60s and 70s were very oversimplified and should be taken with a pinch of salt." Peck points out that along with the
theoretical shift, a growing number of experiments demonstrate that selection between groups does occur and that it can be more powerful than selection between
individuals.
    In the early 1980s, for example, David Craig of the University of Illinois, Chicago, conducted artificial evolution experiments with communities of
flour beetles living in glass vials. Each vial was allowed to produce a batch of young beetles, some of which were selected and redistributed among new
vials to form the next generation. Craig tried selecting for and against the tendency to leave the vial, which benefits the rest of the vial community because
there is extra space for those left behind.
    When he selected only individual beetles that stayed in their vials, the proportion of beetles leaving the vials did not change, even after 14
generations. But when he selected any group of beetles from populations where there was a high proportion of emigrants, the number of beetles climbing
out of each vial doubled in 14 generations. "In these experiments, individual selection is weak, but group selection is very effective," says Peck.
"Which is exactly the opposite of what we have all been taught."
    Craig's findings give clues about when and why group selection occurs. "The early models which rejected group selection assumed that altruistic traits
were controlled by a single gene, with two possible forms, that was directly inherited from one generation to the next," says Peck. "Environmental
variation did not exist in those models." But group selection seems to be strong when a trait is controlled not just by genes, but by interactions with the
environment and with other organisms. In genetic jargon, this is a trait with "low heritability".

    The importance of interactions between individuals is shown by a group selection experiment that could shake up the poultry industry. If you try
to improve the egg production of chickens kept in cages by selecting individual birds that are the most fertile, you will find that productivity actually
goes down. This is because you are unwittingly selecting hens that are also aggressive, who will snatch food from other birds in their cage. Put these
birds together and they will clash horribly, lowering overall fertility. In 1996, William Muir and his team from Purdue University, Indiana, selected
for high egg production by picking out whole cages of birds that did well, rather than selecting individual birds. They managed to increase annual egg
production by 160 per cent, and the chickens lived so well together that they no longer had to suffer the indignity of having their beaks cut off  a standard
practice in battery farming.

    Group selection worked in this case because the trait "egg production" was not just a property of the individual bird, but depended on the
interaction of many other birds in the group. Wilson and his research student William Swenson wondered whether they could take this a step further. If selecting at
group level works on chickens, why shouldn't it work on whole ecosystems, which have many characteristics that are functions of the interactions between
organisms ?

    To test this possibility, Wilson and Swenson used soil  an ecosystem made up of hundreds of thousands of individuals, from many different species of
fungi, bacteria and protozoa. They placed samples from a nearby forest in several transparent containers and grew plants in these artificial
microcosms. After 35 days they selected those that had grown the most plant biomass, and created a second generation of microcosms using soil from them. After 16 generations, the communities under selection were producing three times as much plant biomass as other communities. The researchers conclude that some
feature of the soil ecosystem that enhances plant growth passes from one soil community to another and is open to group selection.
    "People said it would never work," Wilson says. But it did. "The concept of selecting ecosystems without knowing about the actual organisms involved
is incomprehensible to some microbiologists, who think that the only good science is to work with carefully isolated organisms . . .We would love to know
what the actual mechanism is. It's likely to be something to do with interactions between different organisms, but at the moment it's a black box."
    Wilson and Swenson have also found other ecosystems in which group selection appears to work, for example the community of organisms living
in pond water. They found they could select for ecosystems that decreased the acidity of the water. And in another experiment, they dramatically
improved the ability of soil ecosystems to digest a common chemical pollutant called 3chloroaniline. This method could have extensive commercial applications,
especially because the breakdown of contaminant chemicals often involves more than one type of organism. "What we are doing here is developing designer
ecosystems to do certain tasks," says Wilson.

Inside evolution

    If it is simple to do in the lab, then who's to say that many features of ecological communities have not evolved by group selection in nature ?
There are countless examples of natural populations that are structured in discrete groups  from communities living on microscopic particles in the sea to
patchy populations of plants. The best example is parasites, which live in groups of hundreds of thousands confined inside the bodies of their hosts.
    "Much of the current interest in group selection is in parasitology," says Peck. Disease organisms often evolve towards nonvirulence, with each
individual parasite restraining its own reproduction so that the host can survive. This means the rest of the parasite group can survive and disperse effectively.
"Evolution of nonvirulence cannot be discussed without invoking group selection," he adds.

    Despite all the evidence, group selection remains unacceptable to some biologists. Richard Dawkins, author of "The Selfish Gene", has little time
for Wilson's latest work. "They are interesting experiments, but have no connection with group selection," he says. Dawkins accuses Wilson of trying to
resurrect an old biological heresy. "Enormous credit would accrue to anybody who could pull off the seemingly impossible and rehabilitate group selection," he
says. "But actually, such rehabilitation can't be achieved, because the great heresy really is wrong."
    Dawkins argues that group selection is just a kind of kin selection, because members of a group are always going to be related to one another,
so helping others means furthering the genes they have in common. "There is only a revival of group selection among people who have arbitrarily redefined kin
selection as group selection. It is particularly galling, since the term kin selection was originally invented to distinguish it from group selection,"
he says. "Let's get on with pushing evolutionary theory ahead, without this tiresome, timeconsuming, backwardlooking distraction."
    Wilson responds that you could just as well say that kin selection is a type of group selection. "It is all a question of perspective," he
comments, "and we need different perspectives because they hold different insights."
    While theoreticians bicker among themselves, the implications of group selection may be extending far beyond the purely biological. Computer
scientists designing artificially intelligent software are showing an interest in these techniques. "Imposing group selection on software agents or
robots, by selecting groups of components that work well for whatever reason, might just provide the quantum jump that is needed in software development," says
Peck. If he's right, group selection could be one of the most commercially successful ideas to have emerged from biology for a long time.

The roots of morality

    NOMADIC huntergatherer groups such as the Inuit have a system of egalitarianism in which everything is shared equally, and selfish
behaviour is severely punished. This extreme altruism is difficult to explain, because the groups include unrelated individuals, and strict reciprocity does not
operate. But Chris Boehm, of the University of Southern California, Los Angeles, believes he nows how the behaviour evolved.
    "Biologists agree that group selection is possible in theory, but it is such a weak force that the existence of free riders would cancel it out,"
says Boehm. Egalitarianism, he argues, has the effect of strengthening group selection and weakening individual selection. It is all to do with
variation the raw material on which evolution works. Within groups of egalitarian huntergatherers, decisions are made by consensus, so variation within a
group is severely constrained. But variation between groups is enhanced because different groups are likely to settle on different strategies. Survival of
the fittest becomes a competition among groups, favouring those that make the best decisions about lifeanddeath matters, such as where to migrate each year.
    Boehm believes, like Darwin before him, that morality evolved through group selection because it acts as social glue sealing the combined fate of the
group. "If we assume that prehistoric humans were behaving in a similar way to modern forager groups, then this process of selection between groups could
have been operating for 30 000 years," he says.

No freeloaders please

    THE key to changing the level at which selection operates lies in suppressing the interests of the individuals that make up the groups.
"What intrigues me is how this shift happens. What sort of conditions are required to make the transition to a new level where freeloaders can no longer prosper?"
says Paul Rainey, a geneticist at the University of Oxford.
    Rainey works on bacterial cells that form floating microbial mats. Individual bacterial cells in a mat produce a polymer that knits them
together in a group. The energetic cost of producing the polymer means that cells in the mats grow more slowly than those living independent lives, but they
benefit from floating at the water surface, where the oxygen supply is good. "This is a great system to study group selection in, because there are welldefined
costs and benefits to the cooperation, and the groups are open to invasion by cheats," he says.
    The cheats are cells that live in the mat, but don't produce the polymer. Rainey wonders whether he can impose an evolutionary transition on his
microbial mats, so that selection operating at the level of the group eliminates cheats by favouring mats with only honest citizens. He is
running test tube evolution experiments in which mats with a low proportion of cheats are artificially selected to form each successive generation. "The results
are looking good," he says. "We're really interested in what mechanisms might spontaneously emerge to counteract the invasion of cheats. At the moment
we have no idea."

Further reading: "Unto Others" by Elliott Sober and David Sloan Wilson (Harvard University Press, 1998)
"The Origins of Life" by John Maynard Smith and Eoers Szathmary (Oxford University Press, 1999)
"Artificial ecosystem selection" by William Swenson and David Sloan Wilson "Proceedings of the National Academy of Sciences" (forthcoming)