In the 1990s, an agricultural scientist named William Muir set out to breed better egg-laying hens. His approach seemed obvious: pick the most productive hen in each cage and use her to seed the next generation. After a few cycles, he had his answer. What he’d actually bred was a strain of hyper-aggressive birds that achieved their productivity by pecking their cage-mates into submission, and overall egg output at the cage level had dropped.
Muir tried a different approach. He selected the most productive cages and bred all the hens within them. The result was startling. A docile, cooperative strain that within five generations had increased productivity at the cage level by 160 per cent.
The chicken experiment has become something of a landmark in a long and surprisingly bitter debate in evolutionary biology. The standard view of natural selection focuses on the individual: a trait spreads because it helps one organism out-compete its rivals. But what Muir had demonstrated, with feathers and egg counts rather than mathematical models, is that evolution can also act on groups. That selecting whole cages rather than star performers could reshape not just productivity but behaviour, temperament, the social fabric of the flock. It’s the kind of finding that seems obvious once you know it, yet for decades the mere suggestion that selection could work this way was enough to end careers or get you banned from certain classrooms.
That climate may be shifting.
A bibliometric review published this month in Frontiers in Ecology and Evolution has combed through nearly 3,000 indexed scientific articles and identified 280 studies providing solid empirical support for what researchers call multilevel selection, the idea that natural selection operates simultaneously at multiple levels of biological organisation. From viruses to eusocial insects to humans, across more than a century of research, the evidence has apparently been there all along. Somewhat hidden, perhaps, by the cultural politics of the field itself.
“If you measure the average increase in the frequency of a trait over generations and then say it’s favoured by natural selection, you’re not wrong,” says Anne Clark of Binghamton University, a co-author of the review. “But if I ask you: ‘What’s the mechanism for the slow increase in that trait over here and the rapid increase over there?’ you’re not going to be able to tell me. Whereas, if you had looked at different levels, you might see that group competition is more important in one place, or cooperation within groups in another.”
Clark is a behavioural ecologist who has spent much of her career at Binghamton, which has been a centre for multilevel selection theory since 1988. The resistance she describes isn’t merely academic. Since the 1960s, certain influential figures in evolutionary biology have viewed claims of group selection as naïve, sloppy thinking – a vestige of pre-rigorous science. Some banned discussion of it outright. The core objection, broadly, was that selfish individuals within any group would always outcompete altruists, so any group-level benefit would be rapidly undermined from within. Evolution, under this view, was essentially a war of each against all, dressed up in the mathematics of population genetics.
The new review argues this was too narrow a lens. Take cancer. It seems at first like the perfect emblem of individual-level selection run amok: cells that abandon cooperation with the body for their own runaway replication. But Clark notes the picture is considerably stranger than that. “In some cases, cancer cells act as a cooperative group in their own right; the ways they spread are strategic. You can also get competition between diseases for host resources.” Selection, in other words, isn’t just happening inside the tumour. It’s happening between the tumour and other pathogens competing for the same body, between populations of hosts, between transmission contexts that favour virulent or more benign strains.
The same logic applies to infectious disease more broadly. A fast-replicating pathogen in a densely connected population may thrive precisely because it can spread before killing its host. In an isolated population, that same virulence strategy becomes a dead end. Selection between groups of hosts – between epidemiological contexts – can push in exactly the opposite direction to selection within a host.
Michael Wade, the final author on the review and an evolutionary biologist at Indiana University, conducted some of the earliest experimental work on group selection in the 1970s using flour beetles. He created populations, allowed them to reproduce, then selected either the smallest or largest groups as parents for the next generation. Group size diverged rapidly between treatments. The process took only a handful of generations. Crucially, it required no altruism – in the small-group treatments, cannibalism of eggs actually evolved within groups. The group-level effect was real regardless of what was happening at the individual level inside each container.
More recently, researchers working with yeast have demonstrated something even more striking: the de novo evolution of macroscopic multicellularity after 600 rounds of artificial selection for larger group size, with yeast in one treatment becoming around 20,000 times larger than at the start. The emergence of multicellular life from single-celled ancestors is one of the biggest transitions in the history of life on Earth. Watching it happen in a laboratory, under selection pressures acting on groups rather than individuals, rather changes how one thinks about what evolution can do.
The practical applications are harder to dismiss than the theory. Group selection breeding is now standard practice in commercial animal and plant breeding programmes. Research in north-western China has argued that a century of unconscious wheat breeding led to smaller, less competitive root systems – root traits that benefit the crop stand as a whole rather than individual plants hoarding nutrients. Muir’s chickens were far from a fluke.
Clark thinks the implications extend further, into how we design human institutions. “We’ve been encouraged to think about our classroom as a set of diverse individuals,” she says. “What are we rewarding at the classroom level, and what aren’t we rewarding? If we looked at different levels, we would understand mechanisms and what’s really going on under the hood.” A classroom that rewards whoever raises their hand fastest selects for rapid responders. A workplace that promotes whoever grabs individual credit selects for a rather different kind of behaviour than one that rewards collective performance.
The review’s authors are fairly direct about the field’s reluctance to reckon with this. “The debate on the units of selection has gone on for too long,” they write. “It is high time to move on and focus on the empirical evidence and data.” Whether evolutionary biology’s mainstream will agree is, one suspects, still a matter of which group you happen to belong to.
Study link: https://www.frontiersin.org/journals/ecology-and-evolution/articles/10.3389/fevo.2026.1752597/full
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.