Group selection

Group selection is the process by which natural selection operates on groups of organisms — populations, colonies, or species — rather than on individuals alone. A trait that reduces the fitness of individuals within a group may nonetheless spread if groups containing the trait outcompete groups lacking it, particularly when groups are small, genetically related, and subject to frequent extinction and recolonization. The concept has been one of the most contentious in evolutionary biology, with critics arguing that individual-level selection will almost always overwhelm group-level effects, and defenders pointing to cases — from the evolution of multicellularity to human cooperation — where group-level processes appear indispensable.

The modern resolution, articulated through multilevel selection theory and inclusive fitness theory, is that selection operates at multiple nested scales simultaneously. The question is not whether group selection exists but under what conditions it is strong enough to produce traits that cannot be explained by individual selection alone.

The Historical Debate

Group selection was first proposed by V.C. Wynne-Edwards in 1962 as an explanation for animal behavior that appeared to limit population growth — territoriality, dominance hierarchies, and reproductive suppression. Wynne-Edwards argued that these behaviors evolved because groups that limited their growth survived better than groups that overexploited their resources and crashed. The argument was attractive because it seemed to explain altruistic behavior without invoking kinship: individuals sacrificed their own reproduction for the good of the group.

The response, led by George C. Williams in his 1966 book Adaptation and Natural Selection, was devastating. Williams showed that group selection requires conditions that are rarely met in nature: groups must be small, isolated, and subject to frequent extinction; the genetic variation among groups must be large; and the rate of within-group selection against the altruistic trait must be slow. In most natural populations, individuals disperse among groups, gene flow homogenizes group differences, and individual selection against altruism is strong. Under these conditions, group selection is negligible.

Williams's critique was so effective that "group selection" became a term of abuse in evolutionary biology — a label for sloppy thinking that invoked higher-level processes to explain what could be explained by individual-level mechanisms. For three decades, the concept was marginalized, and altruistic behavior was explained exclusively through kin selection and reciprocal altruism.

The Revival: Multilevel Selection Theory

The revival of group selection began in the 1990s with the development of multilevel selection theory, particularly the work of David Sloan Wilson and Elliott Sober. The key insight was mathematical: the Price equation partitions evolutionary change into within-group and between-group components. A trait can increase in frequency in the total population even if it decreases within every group, provided that groups with the trait contribute disproportionately to the next generation.

The Price equation makes the conditions for group selection precise:

Var(Var()) > Cov(w_i, z_i)

where Var(z) is the variance in trait value among groups, Cov(w_i, z_i) is the covariance between individual fitness and trait value within groups, and the inequality states that group-level selection is stronger than individual-level selection. The condition is demanding: the variation among groups must be large relative to the within-group selection against the trait.

When does this condition hold? The empirical evidence points to several contexts:

The evolution of multicellularity. The transition from single-celled to multicellular organisms is the most dramatic group-level event in evolution. A multicellular organism is, in effect, a group of cells that have surrendered their individual reproductive capacity for the benefit of the group. The somatic cells of your body are altruists: they work, divide, and die without passing on their genes. The only cells that reproduce are the germ line. This arrangement is stable because the cells are genetically identical (clonal), eliminating within-group selection against altruism. Group selection — selection among multicellular organisms — is the only force that matters.

Colonial organisms. Social insects (ants, bees, termites) live in colonies that function as superorganisms. The workers are sterile; they sacrifice their own reproduction to raise the queen's offspring. This arrangement is explained by kin selection (workers share genes with the queen's offspring), but it is also a case of group selection: colonies with more cooperative workers outcompete colonies with less cooperative workers, and the variation among colonies is maintained by the reproductive monopoly of the queen.

Human cooperation. Human societies exhibit cooperation among non-kin on scales that no other species achieves. The explanation is contested, but group selection is a leading candidate: human groups compete for resources, territory, and survival, and groups with norms and institutions that sustain cooperation outcompete groups without them. The genetic evidence is mixed, but the cultural evidence is strong: cooperative norms spread between groups by cultural group selection, even if the genetic variation is small.

The Price Equation and Levels of Selection

The Price equation is the mathematical foundation of multilevel selection theory. It states that the change in the average value of a trait in a population is equal to the covariance between the trait and fitness, plus the expectation of the change in trait value within individuals. The equation is exact — it is an identity, not an approximation — and it can be partitioned in multiple ways.

The standard partition separates within-group and between-group components:

Δz̄ = Cov(w, z) = Cov_B(w, z) + E[Cov_W(w, z)]

where Cov_B is the covariance between group fitness and group mean trait value (between-group selection), and Cov_W is the covariance between individual fitness and trait value within groups (within-group selection). The equation shows that total evolutionary change is the sum of between-group and within-group selection.

This partition is not merely a mathematical trick. It reveals that the debate between group selectionists and individual selectionists is often a debate about accounting, not about causation. Both sides agree on the mathematics. They disagree about which level provides the most parsimonious explanation for a given trait.

Group Selection and the Major Transitions

The most important application of group selection theory is to the "major transitions" in evolution — the origins of chromosomes, cells, multicellularity, and societies. Each transition involves the formation of a new level of individuality from a group of lower-level individuals. The lower-level individuals must surrender their autonomy for the higher-level entity to function as a unit of selection.

John Maynard Smith and Eörs Szathmáry identified the common feature of these transitions: the suppression of within-group competition. Before the transition, the group is a collection of competing individuals. After the transition, the group is an individual, and competition within it is suppressed by mechanisms that align the interests of the lower-level units with the interest of the higher-level unit. In multicellular organisms, this alignment is achieved by genetic relatedness (clonality) and by germ-soma separation. In social insects, it is achieved by haplodiploidy (which increases relatedness among sisters) and by queen pheromones that suppress worker reproduction.

Group selection is the force that drives these transitions. Without it, the lower-level individuals would continue competing, and the higher-level entity could not function as a unit. The transition is a shift in the level at which selection operates — from the lower level to the higher level — and it is irreversible because the higher-level entity, once formed, suppresses the lower-level competition that would dismantle it.

The Systems View

From a systems perspective, group selection is an instance of a general pattern: selection operating at multiple nested scales simultaneously. The pattern appears not only in biology but in social systems, economic systems, and technological systems. Markets select among firms; firms select among employees; employees select among strategies. Each level has its own selection criteria, and the dynamics at each level are coupled to the dynamics at other levels.

The key insight is that the level of selection is not given by nature but emerges from the structure of the system. In a well-mixed population with no group structure, selection operates at the individual level. In a population with strong group structure — small groups, limited dispersal, frequent group extinction — selection operates at the group level. The level of selection is a property of the population structure, not a fundamental feature of the evolutionary process.

This has implications for how we think about cooperation in human societies. If cooperation is a group-level adaptation, then its maintenance depends on the preservation of group structure — on mechanisms that limit within-group competition and enhance between-group competition. Institutions that enforce cooperation (laws, norms, religions) are, in effect, group-selection mechanisms: they suppress the free-riding that would undermine group-level benefits and they amplify the differences between cooperative and uncooperative groups.

Group selection was rejected too thoroughly and revived too cautiously. The truth is that selection operates at every level of biological organization, and the question is never "which level is correct?" but "which level dominates for this trait, in this population, under these conditions?" The gene, the individual, and the group are all real levels of selection, and evolution is the interplay among them. The organisms that survive are not those that optimize any single level. They are those that maintain the calibrated balance among levels — the nested architecture that allows fast-level innovation without slow-level collapse.

See Also

• Multilevel selection theory
• Kin selection
• Selfish gene
• Inclusive fitness
• Price equation
• Species selection