Compare the genomes of humans and chimpanzees and you find sequences that are nearly 99 percent identical. Yet when geneticists mapped where crossing over happens during meiosis, they found something surprising. The recombination hotspots of the two species barely overlap. The places where chromosomes swap segments have been reshuffled in the few million years since our lineages split.

Recombination is the process that breaks up combinations of alleles and creates new ones each generation. We usually treat it as background machinery. But its rate is itself a trait. It varies between individuals, between sexes, between species, and between neighbouring stretches of the same chromosome. Like any heritable trait, it can evolve.

That raises a mechanistic question. If recombination can be dialled up or down, what determines the setting? Answering it means looking at three layers: the molecular machinery that places crossovers, the ecological pressures that might reward shuffling, and the population genetics of the genes that control it.

Hotspots and Coldspots

Recombination across the human genome is wildly uneven. Most crossovers fall within tens of thousands of hotspots. Each is typically only one to two kilobases wide, and they are spaced tens of kilobases apart. Roughly 80 percent of recombination occurs in about 10 to 20 percent of the sequence. At the other extreme sit coldspots, such as regions near centromeres, where crossing over is almost entirely suppressed. A gene's evolutionary fate depends heavily on which neighbourhood it lives in.

In humans, mice, and many other mammals, a protein called PRDM9 largely determines where hotspots fall. Its zinc finger array binds a specific DNA motif. It then marks nearby histones with a chemical tag (H3K4me3), flagging the site for the double-strand breaks that initiate recombination. Change the zinc fingers and you change where crossovers happen across the entire genome.

This creates a system that destroys itself. When a break forms at a hotspot, the cell repairs the broken chromosome by copying its homologue. Suppose the homologue carries a version of the motif that PRDM9 binds less well. Then the strongly binding motif gets overwritten, a form of biased gene conversion. Over generations, hotspots erase themselves. This so-called hotspot paradox helps explain why PRDM9's zinc finger region is among the fastest-evolving sequences in mammalian genomes. New variants that recognise fresh motifs restore recombination where the old targets have eroded.

Species without functional PRDM9 offer a revealing contrast. Dogs and birds lack it. Their hotspots cluster at gene promoters and CpG islands, and they stay in place over long evolutionary periods. So the rapid hotspot turnover in humans and chimpanzees is not a universal feature of recombination. It is the signature of a particular molecular arms race playing out inside the genome itself.

Takeaway

The recombination landscape is not a fixed map but a moving target. The very mechanism that places crossovers also destroys its own targets, forcing perpetual renewal.

Red Queen Dynamics

Recombination carries real costs. It breaks apart allele combinations that have already proven successful, a cost sometimes called recombination load. So what benefit could outweigh it? John Jaenike and William Hamilton developed an influential answer between the late 1970s and 1980, and it points to parasites. The idea takes its name from Lewis Carroll's character, who must keep running just to stay in place. The Red Queen hypothesis proposes that hosts must keep changing simply to keep pace with coevolving enemies.

The logic runs through frequency-dependent selection. Parasites evolve to exploit the most common host genotypes, because those are the most abundant targets. The combination of resistance alleles that succeeds today becomes the one most heavily attacked tomorrow. Selection therefore fluctuates. An allele combination favoured in one generation can be disfavoured a few generations later. Recombination generates rare combinations that parasites have not yet adapted to, giving offspring a moving defence.

Field and laboratory evidence supports parts of this picture. Curtis Lively studied mud snails in New Zealand and found that sexual individuals were more common where trematode parasites were abundant. Levi Morran and colleagues tested the idea in experimental populations of the nematode Caenorhabditis elegans. Self-fertilising populations exposed to coevolving Serratia bacteria were driven extinct, while outcrossing populations persisted. In fruit flies and flour beetles, infection has even been shown to raise recombination rates within a single generation.

The theory has limits. Models by Sarah Otto and Scott Nuismer showed that host-parasite coevolution favours higher recombination only under fairly specific conditions. Selection must be strong, and its fluctuations must occur on the right timescale. Across many parameter ranges, coevolution actually favours less recombination. The Red Queen is a plausible driver, but not a guaranteed one. Its strength must be measured case by case rather than assumed.

Takeaway

When the environment is another evolving species, yesterday's winning genotype becomes tomorrow's target. That makes the ability to generate novelty valuable in itself.

Modifier Gene Selection

To understand how recombination rate evolves, population geneticists track modifier genes. These are alleles that change how often crossing over happens without directly affecting fitness themselves. Such genes are real. Variants of RNF212 influence genome-wide crossover counts in humans and cattle, and PRDM9 alleles shift where crossovers fall. The question is when selection favours a modifier that increases recombination and when it favours one that suppresses it.

The crucial feature is that selection on modifiers is indirect. A modifier that increases recombination gains no immediate advantage. It benefits only if the new allele combinations it creates happen to be fitter, and it rides along with them through linkage. This hitchhiking is weak. It depends on the modifier staying associated with the combinations it produced, which is why tightly linked modifiers respond most strongly. The framework builds on Ronald Fisher's argument that recombination speeds adaptation by bringing together beneficial mutations that arise in different individuals.

Theory delivers a striking baseline result. Consider a population at equilibrium under constant selection. In that setting, modifiers that reduce recombination are generally favoured, a result known as the reduction principle, developed by Marcus Feldman and colleagues. If selection has already built good combinations, shuffling them only breaks them up. Increased recombination is favoured only when something disrupts equilibrium. Candidates include directional selection, the interaction of drift and selection in finite populations (the Hill-Robertson effect), and fluctuating selection of the Red Queen kind.

Both outcomes appear in nature. Artificial selection experiments in Drosophila have often produced increased recombination as a side effect, consistent with directional selection favouring shuffling. Meanwhile, chromosomal inversions that suppress recombination lock coadapted alleles together into supergenes. Examples include the genes controlling wing patterns in Heliconius numata butterflies and social organisation in fire ants. Selection does not push recombination in one direction. Whether it favours more or less depends on whether a population needs to build new combinations or protect existing ones.

Takeaway

Recombination rate evolves not because it helps the individual carrying the modifier, but because of what it does to the genetic company that modifier keeps.

Recombination rate is not a fixed property of genomes. Several forces shape it at once. Molecular machinery like PRDM9 places and erodes hotspots. Ecological pressures like coevolving parasites can reward shuffling. Indirect selection acts on the modifier genes that set the rate.

What unites these levels is a tension between building and preserving. Recombination creates new allele combinations but dismantles existing ones. Whether selection favours more or less of it depends on whether the environment, internal or external, is changing.

Seen this way, the uneven map of crossovers across a chromosome is a record of evolutionary history. It is a snapshot of a trait still evolving, shaped by the same forces it helps to direct.