杏吧原创

Evolutionary success is all about looking out for number one 鈥� or so most biologists would tell you. The genes that do the best job of passing themselves along to the next generation, whether by brute selfishness or canny cooperation, are the ones that flourish 鈥� a view most memorably championed by more than 30 years ago in his bestselling book The Selfish Gene.

This relentless focus on the gene may not tell the whole story, however. A small but growing coterie of evolutionary biologists argue that it leaves us blind to crucial evolutionary processes at higher scales 鈥� among groups, species and even whole ecosystem. If they are right, the popular view of evolution and the biological world needs a radical shake-up.

Almost everyone agrees that the gene鈥檚-eye view works perfectly well most of the time. 鈥淚t鈥檚 dominated the field, and dominated for a long time,鈥� says Michael Ruse, a philosopher of science at Florida State University in Tallahassee. Indeed, many biologists think the selfish-gene concept can explain all the intricacies thrown up by evolution, and not just the obviously selfish ones.

Helping relatives

For example, the gene or genes that make worker ants devote themselves to helping their queen reproduce rather than reproducing themselves might appear altruistic but really these genes are promoting their own survival: helping a close relative is another way of passing on one鈥檚 own genes. As this example shows, 鈥渟elfish genes鈥� do not always favour self-centred, uncooperative behaviour, a common misreading of Dawkins鈥檚 position.

However, the consensus is that evolution never favours what might be called 鈥渟elfless鈥� genes 鈥� that is, adaptations that benefit a group of organisms or the species as a whole. An example would be a gene that restricts how many offspring a predator has, to avoid wiping out its prey. Such a gene should always lose out to selfish genes that maximise reproduction, the thinking goes, even if reproducing without restraint threatens the survival of the whole species.

Increasingly, though, this consensus is being challenged, and on several fronts. The least controversial of these is the notion that entire species themselves can have traits that, over geological time, make them more likely than others to escape extinction and branch off new daughter species. This can lead to evolutionary change that could not be predicted from individual adaptations alone.

Species selection

For example, has shown that, over millions of years, marine snails with small ranges have been more likely to go extinct than more widespread ones. 鈥淓ven small perturbations can take out a highly localised species, whereas a more widespread species will live to fight another day,鈥� says Jablonski. As a result, the geographic range of species in a lineage tends to increase over time, Jablonski has found 鈥� though this trend is muddied by periodic mass extinctions, which wipe out widespread species as well as those that occupy a more specialist niche.

This so-called 鈥渟pecies selection鈥� may help explain other puzzling observations. For example, larger individuals often outcompete smaller ones, so selection at the level of individuals would suggest that the average body size of mammals ought to increase over millions of years 鈥� yet in many groups it doesn鈥檛. A larger-bodied species, however, has a larger requirement for food and space, and so might run a greater risk of extinction. In this case species selection may oppose individual selection and so help keep body size constant, says Jablonski.

Even Dawkins agrees there is something to the idea of species selection. 鈥淲hat it takes to survive as an individual is different than what it takes to survive as a species,鈥� he says. Selection at the species level does not drive the evolution of new traits but it does help determine how such adaptations fare in the broad sweep of evolution, in changing environments over vast stretches of time. 鈥淪pecies selection may not build horns, but it can determine how many species have horns or how long horns persist,鈥� says Jablonski.

鈥淲hat it takes to survive as an individual is different to what it takes to survive as a species鈥�

Just how important is species selection, though? 鈥淗ow frequent is it, and how often does it operate counter to individual selection? We don鈥檛 have a good sense of that yet, because so few people are testing at multiple levels,鈥� says Jablonski.

Group selection

This is starting to change. Carl Simpson, a palaeobiologist at the Humboldt University of Berlin, Germany, recently looked at the body shape of fossil crinoids, organisms that resemble upside-down starfish on a stalk. Simpson used a mathematical description of evolution called to determine to what extent the complexity of crinoids鈥� armoured plates might be driven by species or individual selection. Species selection does play a role, he found, but its influence was negligible.

However, species selection could yet turn out to play a stronger role in other situations. We shouldn鈥檛 expect it to be important all the time, Simpson notes. After all, even gene or individual-level selection 鈥� which everyone agrees is a potent force 鈥� often produces little net change for long periods of time.

Even gene鈥檚-eye purists like Dawkins find species selection somewhat palatable, because species, like genes, are relatively discrete entities that are stable enough over time for selection to have some effect in shaping their characteristics. In contrast, most evolutionary biologists have difficulty swallowing the idea that natural selection can act at an intermediate level, on loose affiliations of unrelated members of the same species 鈥� a phenomenon known as group selection.

Hyperaggressive plants

Certainly, group selection can be a powerful force in artificial situations. Indeed, crop breeders rely on it, often without realising they are doing so. Choosing the most vigorous individual plants produces hyperaggressive plants that, when grown together in a field, interfere with each other so much that yields go down. Instead, crop breeders choose plants that get along well, by growing them in test plots and breeding from the plots that yield best 鈥� in effect, practising group selection.

Similarly, a host of lab experiments over the past several decades have shown that group selection can lead to evolutionary change. These experiments are done in controlled situations, though. In the real world of nature, group selection may not have such an easy time, because cooperative groups are vulnerable to takeover by cheaters. These selfish invaders 鈥� a fast-growing wheat plant, for example 鈥� pay none of the costs of cooperation, yet reap all the benefits of being in a cooperative group. In the lab, or in breeders鈥� test plots, researchers can parry this influx of selfish genes. Since nature lacks such oversight, most biologists doubt group selection can be important. 鈥淭he interesting question,鈥� says Dawkins, 鈥渋s whether any adaptation of a wild animal or plant is interpretable as group selection. I don鈥檛 think it is.鈥�

Even the proponents of group selection agree that only a few potential examples have been identified so far, such as the small size of some annual plants that grow together and reduced virulence in some parasites (to keep their hosts alive). 鈥淭hat鈥檚 a serious problem,鈥� says Charles Goodnight, an evolutionary biologist at the University of Vermont in Burlington. 鈥淲e don鈥檛 know how common it is in nature. The reason we don鈥檛 know is that we haven鈥檛 looked for it.鈥�

The slime test

That may be changing, though, as fresh ideas emerge that give group selection some theoretical traction. In particular, in New York state has shown that cheaters will not prosper if groups frequently break up and reform again with new members. With each fresh start, the groups that happen to have the least cheaters thrive while those with lots of cheaters perish.

Microbial biofilms are one of the best test cases for this hypothesis. Biofilms are colonies of bacteria living within a matrix of slime that they secrete. They enrich the slime with molecules that help them extract nutrients, such as iron, from the environment. 鈥淪ingle bacteria produce these compounds that scavenge the iron, but once this molecule is produced, it鈥檚 a common good, and you can get cheats arising,鈥� says Alexandra Penn, an evolutionary biologist at the University of Southampton in the UK.

The characteristics of the slime also determine how big the biofilm can get and how long it persists before breaking up. That sets the stage for group selection through the formation and break-up of groups.

Ecosystem selection

Penn鈥檚 models suggest that not only should there be group selection, but that evolution can make the conditions needed for group selection more likely in the future. In other words, if cooperative biofilms do better than selfish ones, the result will be slime that persists just long enough to favour group selection. Penn is beginning experiments to see whether the bacteria鈥檚 slime does indeed evolve towards the consistency her work suggests is ideal for group selection.

To begin, Penn鈥檚 biofilms will include just one species. Eventually, however, she hopes to include multiple species to study whether selection operates at an even higher level, that of whole ecosystems. The idea is that just as group selection may sometimes favour the interests of a group over any of its constituent individuals, so ecosystem selection might act to shape an entire ecosystem over the interest of its constituent species.

For example, if 鈥渦nbalanced鈥� patches of rainforest or coral reefs 鈥� dominated by just a few species 鈥� were more vulnerable to being wiped out in harsh years, ecosystem-level selection might favour 鈥渃ollegial鈥� species that temper their own growth to maintain a suitable balance. According to this hypothesis, just as the cells within our bodies sometimes sacrifice themselves to ensure the body as a whole remains healthy, so individual species within some ecosystems may on occasion make sacrifices to ensure that the whole ecosystem survives.

Microcosms

This idea remains anathema to most mainstream evolutionary biologists. According to the conventional view, individual species in an ecosystem should be the equivalent of cancerous cells in a body, perhaps cooperating with some other species but growing as aggressively as possible heedless of the cost to the whole ecosystem.

Finding out whether there is ecosystem selection or not isn鈥檛 easy. A decade ago, for example, Sloan Wilson and his colleagues grew microcosms containing hundreds of species of soil microbes. With each 鈥済eneration鈥� they tested the microcosms to see which could support the greatest plant biomass, and used the soil from the winning microcosm to found new microcosms. After 16 generations, the selected soil ecosystems could support three times the biomass of similar, unselected soils. Since they were selecting for a property of the whole ecosystem, not of any individual microbial species, what was happening was ecosystem selection, they argued.

But the experiment had a key shortcoming. 鈥淭here was no way they could rule out the possibility that they just got an ecosystem that happened to have a good batch of individual species in it,鈥� says , an evolutionary modeller at the University of East Anglia, UK. So Williams set to work to repeat the experiment using a detailed computer simulation instead of actual organisms. In Williams鈥檚 model, digital organisms take in nutrients, metabolise them and excrete wastes, altering their environment. They grow, reproduce and evolve 鈥� with the key difference that, with the click of a computer mouse, Williams can turn off individual selection to see whether a separate, ecosystem-level evolution is also occurring.

Context is all

Sure enough, when Williams and his colleague Timothy Lenton simulated the effects of selecting ecosystems that approached some ecosystem-level target (such as acidity levels), they found that ecosystem-level selection best explained their results. 鈥淲e found that you couldn鈥檛 decompose the response we observed at the community level to a lower-level response. There was no single species that could do the job on its own,鈥� says Williams.

Penn, too, has seen evidence of ecosystem-level selection in experiments using soil microcosms, and she expects to see something similar within her microbial biofilms, where multiple species may work together to produce an environment that maximises their collective survival and reproduction. 鈥淚t鈥檚 meaningless to say you鈥檝e just got individual selection in that case,鈥� she says. 鈥淚f you looked at each species individually, you couldn鈥檛 predict what they鈥檇 be doing. You have to look at them in the context of the ecosystem.鈥�

If ecosystem-level selection is the norm, it could prompt a major shake-up in our view of the microbial world and, by extension, the macroscopic world, too. 鈥淚t鈥檚 only in the last 5 or 10 years that people realised that the majority of bacteria live in multispecies collectives,鈥� says Penn. 鈥淏acteria are driving the basic processes of the biosphere, so if their evolution is in this higher-level context, it鈥檚 going to be very different to the way we鈥檝e thought about it previously, and their responses to climate change could be very different than we would expect from thinking about them individually.鈥�

It is still too early to know whether group, species and ecosystem-level selection are major evolutionary forces or merely minor curiosities 鈥� baroque ornaments on the central edifice of individual or gene-level selection. But the dominance of the 鈥渟elfish gene鈥� in evolutionary thought is facing its strongest challenge in many years.

鈥淚f you just look at genes, you鈥檙e not going to see why evolution happens鈥�