Group selection is the idea that natural selection can act on groups of organisms, not just on individuals or genes. In a population divided into distinct groups, some groups may outcompete others because of traits shared among their members, even when those traits are costly to individual carriers. The concept has been one of the most contentious in evolutionary biology for over half a century, dismissed as negligible by many prominent theorists and championed as essential by others. The dispute has softened considerably in recent years, with most researchers now accepting that selection can operate at multiple levels simultaneously, though they still argue about how often group-level effects matter in the wild.
The Basic Logic and Its Central Tension
Imagine a population split into separate groups, where some groups contain individuals willing to cooperate at a personal cost and other groups are full of selfish individuals. Groups with more cooperators may grow faster, survive longer, or produce more migrants. Over time, those cooperative groups contribute more descendants to the overall population. That is group selection in its simplest form.
The catch is that within every cooperative group, a selfish individual who takes the benefits of group cooperation without paying the cost will do better than its cooperative neighbors. Selfish types reproduce faster locally and tend to spread within their group. So there is a race: between-group selection favors cooperation, but within-group selection favors selfishness. George C. Williams argued influentially in the 1960s that selection among groups rarely overrides this within-group advantage, and for decades that view dominated mainstream evolutionary thinking.1PubMed Central. The group selection controversy The gene-centric perspective, popularized by Richard Dawkins, pushed the individual gene as the fundamental unit of selection, treating groups as too unstable and too porous to be meaningful evolutionary players.
What has changed since then is not a reversal of Williams’s core insight but a more nuanced appreciation of the conditions under which between-group selection can, in fact, be strong enough to matter. When groups are stable, when mixing between them is limited, and when mechanisms exist to suppress selfishness within groups, the balance can tip.
The Relationship Between Group Selection and Kin Selection
One of the most important developments in the debate is the recognition that group selection and kin selection are, mathematically, two ways of describing the same underlying process. Kin selection theory, summarized by Hamilton’s rule, explains cooperation through the relatedness between interacting individuals: you help relatives because they share your genes. Group selection theory explains cooperation through competition between groups that differ in their composition. The formal equivalence of these two frameworks has been known for decades.2PubMed. Group selection and kin selection: formally equivalent approaches
This equivalence means that in many cases the two approaches make identical predictions. They are different bookkeeping systems for the same evolutionary outcome. A population structured into kin groups can be analyzed with kin selection math (asking how relatedness affects the spread of a cooperative gene) or with group selection math (asking how group composition affects group success). The answer comes out the same.
So why do researchers keep arguing? Partly because different bookkeeping systems highlight different biological features. Kin selection naturally draws attention to relatedness and to the endpoint of adaptation, asking what the optimal behavior looks like. Multilevel selection approaches draw attention to the process itself, tracking how selection pressures at different levels push trait frequencies around in real time.3PubMed Central. Kin and multilevel selection in social evolution: a never-ending controversy? These complementary perspectives can illuminate different aspects of the same problem, which is why many researchers now advocate a pluralistic stance.4PubMed. Individuals and groups in evolution: Darwinian pluralism and the multilevel selection debate
Experimental Evidence From the Lab
One of the clearest demonstrations that group selection can work came from flour beetles in the 1970s. Michael Wade selected among laboratory populations of Tribolium castaneum for either increased or decreased adult population size, essentially picking the most productive groups as founders of the next generation. The response was rapid and dramatic: within three or four generations, group-selected lines differed from control populations by over 200% in adult population size. This happened even though individual-level selection was simultaneously pushing control populations in the opposite direction, causing their numbers to crash from over 200 adults to around 50.5PubMed Central. Group selections among laboratory populations of Tribolium
The mechanisms behind the population-size response turned out to involve changes in fertility, developmental speed, body weight, and cannibalism rates. Groups selected for large size evolved lower cannibalism, which meant more individuals survived to adulthood. The experiment showed that when group structure is imposed and between-group selection is strong, the response can overwhelm opposing individual-level pressures. Critics pointed out that laboratory conditions with enforced group boundaries are far stricter than anything found in most natural populations, but the experiment remains a landmark proof of principle.
Where Group-Level Dynamics Show Up in Nature
The strongest natural cases for group selection tend to involve organisms that already live in structured populations with limited mixing. Microbes are a rich example. Bacteria living in biofilms face a classic public-goods problem: some cells produce enzymes that break down nutrients in the environment, and all nearby cells benefit, whether they produce the enzyme or not. Non-producers (cheaters) save the metabolic cost of enzyme production and should, in theory, outcompete producers. But research on bacterial biofilms has shown that when biofilms are thick and dense, the nutrients released by enzyme producers barely diffuse out of the local neighborhood. Increasing biofilm thickness from a single cell layer to five layers causes a roughly hundred-thousand-fold drop in the amount of nutrient escaping the biofilm. In those conditions, producers effectively keep their own benefits, and cheaters cannot freeload. Producers outcompete non-producers at every starting frequency.6PubMed Central. Solutions to the public goods dilemma in bacterial biofilms
Pathogen virulence offers another window. Within a single host, parasite strains compete for resources, and faster-growing strains win. But rapid growth can kill the host, cutting off transmission to new hosts. This is a group-selection dynamic: within the host, selfish exploitation wins, but across hosts, restrained exploitation produces more transmission. The trade-off between within-host competition and between-host transmission is one of the central frameworks in the study of parasite virulence.7PubMed. Models of parasite virulence
The Eusociality Debate
No discussion of group selection is complete without the social insects. Ant colonies, honeybee hives, and termite mounds represent some of the most extreme cooperation in nature, with sterile workers sacrificing their own reproduction entirely. For four decades, the standard explanation was kin selection: workers in hymenopteran colonies (ants, bees, wasps) share unusually high relatedness with their sisters due to their sex-determination system, so helping raise sisters can spread a worker’s genes more efficiently than reproducing directly.
E. O. Wilson challenged this consensus in a series of papers, arguing that group selection is the primary force binding eusocial colonies together, while kin selection plays at most a weak supporting role. In Wilson’s view, close kinship within colonies may be more a consequence of eusociality than a cause of it: once a colony structure exists, only close relatives end up living together, but the colony structure itself was driven by group-level advantages.8PubMed Central. Eusociality: origin and consequences A 2010 paper by Nowak, Tarnita, and Wilson went further, arguing that standard natural selection theory applied to precise models of population structure is a simpler and better framework for understanding eusociality than inclusive fitness.9PubMed Central. The evolution of eusociality
That paper drew an enormous backlash. Over a hundred biologists signed a response defending kin selection. The controversy highlighted just how much the debate is about framing rather than facts: given the mathematical equivalence of the approaches, the disagreement is often about which framework provides more insight, not which is technically correct.
Major Evolutionary Transitions
One of the most compelling arguments for taking group selection seriously is the pattern of major transitions in the history of life. Genes grouped into chromosomes. Cells joined into multicellular organisms. Organisms merged into colonies with reproductive division of labor. At each step, formerly independent units became parts of a larger whole, losing their individual autonomy. This process looks a lot like group selection: groups of cooperating units outcompeted groups of loners, and mechanisms evolved to suppress cheating within the group until the group itself became the new “individual.”
Multilevel selection theory has become a standard framework for understanding these transitions. The emergence of the eukaryotic cell (with its mitochondria that were once free-living bacteria), multicellularity, and complex animal societies can all be analyzed as shifts in the balance between within-group and between-group selection.10PubMed Central. Toward major evolutionary transitions theory 2.0 In this view, the history of life is not just a story of competition between individuals but a repeated pattern of groups becoming so integrated that they function as new individuals at a higher level.
Cheater Suppression and What Keeps Groups Together
For group selection to work, something has to prevent selfishness from dissolving cooperation from within. In social insects, this is partly handled by policing: workers destroy eggs laid by other workers, enforcing the queen’s monopoly on reproduction. In the broader theoretical framework, policing and what researchers call “reproductive bribing” turn out to follow predictable rules. If it pays a group member to behave selfishly, its partner is almost always favored to offer a reproductive incentive large enough to remove the temptation.11PubMed. Reproductive bribing and policing as evolutionary mechanisms for the suppression of within-group selfishness
Relatedness and competition interact in interesting ways here. Higher relatedness strengthens between-group selection against competitive selfishness and also strengthens selection for policing. But higher competition within groups independently favors the evolution of costly policing behaviors, so even groups with lower relatedness can end up with strong policing when internal conflict is intense.12PubMed Central. The evolution of competition and policing: opposing selection within and among groups This helps explain why cooperation exists even in populations where relatedness is not particularly high.
Spatial Structure and Why Geography Matters
Whether group selection can get off the ground depends heavily on how organisms move around. Limited dispersal keeps relatives together, increasing the genetic similarity within groups and creating the variation between groups that selection needs. But limited dispersal also creates a problem: relatives who stay close compete with each other for resources. In the simplest theoretical models, these two effects cancel out perfectly, and limited dispersal alone does nothing for cooperation.13Evolution. Limited Dispersal, Budding Dispersal, and Cooperation: An Experimental Study
The key is what breaks that cancellation. Budding dispersal, where whole groups or fragments of groups colonize new patches together rather than individuals scattering independently, can tip the balance because it preserves group composition during colonization. Spatial structure with limited diffusion of resources also helps, because the benefits of cooperation stay local while the costs of competition get diluted.14PubMed Central. Group-selection via aggregative propagule-formation enables cooperative multicellularity in an individual based, spatial model These details matter: group selection is not a universal force but one that depends on specific population structures to operate.
Cultural Group Selection in Humans
Perhaps the most active frontier for group selection thinking is human evolution. Humans cooperate with non-relatives on a scale unmatched by any other species, and explaining this has been a persistent challenge. Cultural group selection theory proposes that human social learning, combined with rewards and punishment, can stabilize almost any group behavior. Once different groups settle on different cultural norms, groups with more cooperative norms can outcompete less-cooperative ones through warfare, economic competition, or differential migration.15PubMed Central. Cultural group selection and human cooperation: a conceptual and empirical review
Evidence from contemporary small-scale societies supports at least part of this picture. Research across multiple populations has found that cooperation between groups is predicted by cultural similarity, suggesting that norms of cooperation have evolved under group-level selection on cultural variation.16Nature Communications. Human large-scale cooperation as a product of competition between cultural groups Cultural group selection operates faster than genetic group selection because cultural traits can change within a single generation, and conformist social learning can maintain group differences even with migration between groups. This makes the conditions for group selection easier to meet in cultural evolution than in purely genetic contexts.
Practical Uses in Agriculture
Group selection has moved from theoretical debate to commercial application in at least one domain: poultry breeding. Hens housed together in cages often engage in feather pecking and cannibalism, which cause welfare problems and economic losses. Traditional individual selection for egg production inadvertently favors more aggressive birds, because the highest-producing individuals in a cage may be the ones monopolizing food at their cagemates’ expense.
When breeders switched to group selection, choosing entire families based on total group survival and egg production rather than individual performance, the results were striking. In one study, hens from individually selected lines needed 287 replacements per 576 birds to maintain group size from 17 to 44 weeks of age, compared to only 46 replacements in the group-selected line. Group-selected birds also had better feathering and lower body weight.17Poultry Science. Group Selection for Adaptation to Multiple-Hen Cages: Beak-Related Mortality, Feathering, and Body Weight Responses Research continues into applying group selection techniques to reduce feather pecking in commercial flocks, essentially harnessing between-group selection to breed for traits that benefit the whole cage rather than the most aggressive individual.18World’s Poultry Science Journal. The prevention and control of feather pecking: application to commercial systems
Species Selection at Larger Scales
The logic of group selection extends beyond groups of organisms to groups of species. Species selection occurs when certain traits make some species more likely to speciate or less likely to go extinct. Over millions of years, lineages with those traits come to dominate, not because of any advantage to individual organisms, but because of differential survival and reproduction at the species level. This is increasingly recognized as a real macroevolutionary force.19PubMed. Species selection and random drift in macroevolution
Species selection can amplify or suppress microevolutionary change. If individual-level selection pushes a trait in one direction but species carrying that trait go extinct faster, the macroevolutionary trend may stall or reverse. Stable patterns in the fossil record lasting millions of years may reflect species selection neutralizing microevolutionary pressures. In studies of marine snails (archaeogastropods), researchers have argued that stabilizing species selection, driven primarily by differential rates of species origination rather than differential extinction, maintained certain shell forms over deep time.20Paleobiology. Stabilizing species selection in the Archaeogastropoda
Selecting Whole Ecosystems
The most radical extension of group selection thinking involves selection on entire ecosystems. Laboratory experiments have demonstrated that artificial selection can shape the properties of whole ecosystems. In these experiments, microbial communities are established, their collective function is measured (say, the rate at which they decompose a target compound), and the highest-performing communities are used to seed the next generation. Despite starting with thousands of species and millions of individuals, the ecosystem-level traits show heritable variation and respond to selection.21PubMed. Artificial ecosystem selection A growing body of work in applied microbiology is exploring whether this approach can be used to improve ecosystem functions for practical purposes, from soil health to waste treatment.22PubMed Central. The effectiveness of artificial microbial community selection: a conceptual framework and a meta-analysis
Why the Argument Refuses to Die
Given that kin selection and group selection are formally equivalent, and given that most researchers now accept multilevel selection as a legitimate framework, it is fair to ask why the debate has been so bitter and so durable. Part of the answer is scientific: the two approaches, though mathematically equivalent, are not equally useful in every situation. In populations where relatedness is easy to measure and interactions are pairwise, kin selection provides a cleaner analysis. In populations with complex group structure or cultural inheritance, multilevel selection can be more intuitive.
But another part of the answer is ideological. Historians of science have noted that the concept of “group” carries social and political baggage. The notion that groups can be units of selection resonates differently depending on whether you associate it with mutualism, nationalism, or collectivism, and researchers’ stances in the debate have been shaped by the broader ideologies they support or fear.23Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences. Shifting values partly explain the debate over group selection This does not mean the debate is purely political, but it helps explain why a question that is partly a matter of mathematical bookkeeping has generated such intense emotional investment for over 50 years.