When Michael Wade published his experiment in PNAS in 1976, he did something most evolutionary biologists thought was either impossible or pointless. He bred flour beetles in groups, selecting entire populations based on their collective traits rather than individual performance. The result? Measurable evolutionary change driven not by competition between organisms, but by competition between groups. For decades, this kind of work languished at the margins of evolutionary biology. Most textbooks barely mentioned it. Many scientists dismissed group selection as a theoretical curiosity with little real-world relevance.

Then in February 2026, a team led by César Marín published a review that changed the conversation. After screening nearly 3,000 scientific articles, they identified 280 empirical studies demonstrating multilevel selection across the entire spectrum of life—from viruses to ecosystems. Nearly 50 years of evidence had been hiding in plain sight.

The Hierarchy Problem

Natural selection doesn't just happen at one level. Genes compete within genomes. Cells compete within bodies. Organisms compete within groups. Groups compete within ecosystems. At each level, traits that increase fitness at that scale can spread through populations.

The catch: what's good for one level isn't always good for another. A gene that copies itself aggressively might damage the organism carrying it. A cell that divides without restraint becomes cancer. An individual that hoards resources might doom its group to extinction. Evolution operates simultaneously across these nested hierarchies, with selection pressures pushing in different directions.

This is multilevel selection. It occurs when natural selection acts on two or more biological levels at once, creating evolutionary dynamics that can't be explained by looking at individuals alone.

What the Evidence Shows

Of the 280 studies Marín's team identified, 64% were controlled laboratory experiments. Researchers didn't just observe multilevel selection in nature—they created it, manipulating selection pressures to watch evolution unfold across different scales.

The experiments span an impressive range. Scientists have bred yeast aggregates, selecting for collective traits of microbial communities. They've grown plants in pots as mini-ecosystems, applying selection to entire assemblages. They've studied ant colonies, bacterial biofilms, and human cultural practices. The remaining 36% documented multilevel selection in wild populations, measuring how group-level traits influence survival and reproduction in natural settings.

One pattern emerges clearly: 90% of studies focused on group-level selection, examining how collections of organisms—colonies, herds, communities—evolve as units. This makes sense. Groups are where multilevel selection becomes most visible and consequential.

When Levels Collide

The relationship between selection at different levels isn't fixed. Sometimes individual and group interests align. A trait that helps an organism survive might also benefit its group. But often they conflict.

Consider cooperation. Helping others typically costs the helper something—energy, resources, risk. From an individual perspective, the best strategy is often to cheat: enjoy the benefits of others' cooperation without paying the costs yourself. Within a group, cheaters should win. But groups full of cooperators outcompete groups full of cheaters. Group selection can favor cooperation even when individual selection opposes it.

This explains phenomena that puzzled Darwin himself. How did sterile worker castes evolve in ants and bees? Why do organisms sometimes sacrifice themselves for others? Multilevel selection provides answers: traits that harm individuals can spread if they benefit the groups containing them strongly enough.

The math matters here. Researchers use contextual analysis to partition fitness effects, separating what comes from individual traits versus group context. The balance determines which level dominates. Change the ecological conditions—how groups form, how often they compete, how much individuals move between them—and you change which level of selection wins.

From Molecules to Ecosystems

The evidence doesn't cluster at one scale. Multilevel selection appears across the entire biological hierarchy.

At the molecular level, genetic elements compete within genomes. Some genes boost their own transmission at the expense of the organism—"selfish" DNA that spreads despite harming its host. Selection among organisms then favors mechanisms to suppress these molecular cheaters.

Cells within multicellular bodies face similar tensions. The origin of multicellularity itself represents a major transition where lower-level units (cells) became integrated into higher-level individuals (organisms). This required suppressing cell-level selection—preventing cells from reverting to independent reproduction. Cancer shows what happens when that suppression fails.

At larger scales, the logic continues. Microbiomes evolve as communities, with selection acting on collective properties like nutrient processing or pathogen resistance. Plant communities in experimental pots show heritable variation in ecosystem-level traits. Even human cultural evolution involves group-level processes, where practices spread between societies based on group success.

Why the Silence?

Given this evidence base, why has multilevel selection remained controversial? Why do biology textbooks still gloss over it?

Part of the answer is historical. In the 1960s and 70s, evolutionary biologists fought bitter debates about group selection. Many concluded it was theoretically weak and empirically rare. That verdict stuck, even as evidence accumulated otherwise. Scientific narratives, once established, prove hard to shift.

The other issue is conceptual. You can often reframe group selection in terms of individual selection by carefully defining "individual fitness" to include group effects. This mathematical equivalence led some researchers to dismiss multilevel selection as just a different way of describing the same thing. But that misses the point. The question isn't whether you can describe evolution using only individual selection—it's whether doing so provides insight. When selection acts on genuinely collective traits, when group structure shapes evolutionary outcomes, multilevel thinking becomes essential.

Evolution's Nested Games

The 280 studies Marín's team identified don't just validate a theory. They reveal how evolution actually works: as nested games played simultaneously at multiple scales, with outcomes depending on the complex interplay between levels.

This matters for more than academic debates. Understanding multilevel selection helps explain the origin of life's major innovations—multicellularity, eusociality, even consciousness. It informs practical applications from agriculture (where group selection breeding improves livestock) to medicine (where cancer represents selection at the wrong level). It shapes how we think about human cooperation, cultural change, and collective action problems.

The evidence has been building for half a century. Perhaps now we'll finally pay attention.