Coevolution

Although the idea that interacting organisms shape one another's evolution can be traced to Charles Darwin's observations on orchids and their insect pollinators in the 1860s, the formal concept of coevolution was introduced by Paul R. Ehrlich and Peter H. Raven in their landmark 1964 paper "Butterflies and Plants: A Study in Coevolution," published in the journal Evolution. Through a comprehensive survey of butterfly larval host-plant associations, they demonstrated that the diversification of butterfly lineages was closely tied to the chemical defenses of their food plants, and proposed that reciprocal selective pressures between the two groups had driven their correlated evolutionary histories. The term was subsequently refined by Daniel Janzen in 1980, who argued for a strict definition restricting coevolution to cases of demonstrably reciprocal evolutionary change between two interacting populations through mutual adaptation—as opposed to broader, looser usages that merely referred to parallel or coincident evolutionary patterns.

Coevolution is broadly classified according to the number of species involved and the nature of their interaction. Pairwise (specific) coevolution describes tight reciprocal evolutionary change between exactly two interacting species, such as a specific parasite and its host. Diffuse (guild) coevolution involves multiple interacting lineages that collectively influence one another's evolution, as when an assemblage of insect herbivores exerts reciprocal selective pressure on a community of host plants. Interactions may further be classified by their ecological character. Antagonistic coevolution occurs between species with conflicting fitness interests, such as predators and prey, hosts and parasites, or herbivores and their food plants. Mutualistic coevolution occurs between species that derive mutual benefit from their interaction, such as plants and their pollinators, or mycorrhizal fungi and their host plants.

One of the most vivid manifestations of coevolution is the evolutionary arms race, a concept formalized by Richard Dawkins and John Krebs in 1979. In an arms race, one species evolves an enhanced offensive or defensive trait, which in turn selects for a counter-adaptation in the interacting species, leading to escalating cycles of adaptation and counter-adaptation. A classic example is the coevolution between the rough-skinned newt (Taricha granulosa) and the common garter snake (Thamnophis sirtalis) in western North America, studied extensively by Edmund Brodie III and colleagues. The newt produces tetrodotoxin (TTX), one of the most potent neurotoxins known, while certain garter snake populations have evolved remarkable resistance to TTX through mutations in their sodium channel genes. Populations vary geographically in the level of newt toxicity and snake resistance, creating a geographic mosaic of coevolutionary escalation. In some populations, snake resistance appears to have outpaced newt toxicity, while in others the arms race remains ongoing.

Dawkins and Krebs also introduced the "life–dinner principle" to describe the asymmetry of selection pressures in predator–prey arms races: the prey is running for its life, while the predator is merely running for its dinner. This asymmetry predicts that prey species generally experience stronger selection, potentially allowing them to "win" the arms race—though empirical evidence shows the situation is more complex and context-dependent.

The Red Queen hypothesis, proposed by Leigh Van Valen in 1973, provides a macroevolutionary framework intimately connected to coevolution. Van Valen observed that the probability of extinction for a given taxon appeared roughly constant regardless of how long that taxon had already existed. He explained this by proposing that species must continuously evolve simply to maintain their relative fitness against other evolving species—analogous to the Red Queen in Lewis Carroll's Through the Looking-Glass, who must keep running just to stay in place. Research since Van Valen's original paper has identified three distinct modes of Red Queen dynamics: (1) the Fluctuating Red Queen (FRQ), driven by oscillating selection at few genetic loci, where exploiters track common host genotypes and rare genotypes gain a temporary advantage; (2) the Escalatory Red Queen (ERQ), characterized by directional selection and escalation of polygenic traits in an arms race; and (3) the Chase Red Queen (CRQ), in which local directional selection drives coevolutionary chases through multidimensional phenotype space. Empirical studies have shown that these modes can operate simultaneously, at different loci within a genome, at different stages of a coevolutionary interaction, and across different spatial scales.

John N. Thompson developed the Geographic Mosaic Theory of Coevolution (GMTC), articulated most fully in his 2005 book The Geographic Mosaic of Coevolution. This theory recognizes that coevolution does not proceed uniformly across the entire range of interacting species. Instead, coevolutionary dynamics vary geographically due to three key components: (1) geographic selection mosaics, where the direction and strength of reciprocal selection differ among populations; (2) coevolutionary hotspots and coldspots, where hotspots are populations in which reciprocal selection is active and coldspots are populations where one or both species experience little or no reciprocal selection; and (3) trait remixing, the continual reshuffling of coevolved traits across populations through gene flow, genetic drift, and local extinction and recolonization. The GMTC has been supported by empirical studies of numerous systems, including the newt–snake arms race, crossbill–lodgepole pine interactions, and wild flax–flax rust pathogen associations.

Ehrlich and Raven's original 1964 paper also laid the groundwork for what is now known as the escape-and-radiate model of coevolution, later explicitly named by John N. Thompson in 1989. Under this model, a plant lineage evolves a novel chemical defense that allows it to "escape" from its current suite of herbivores. Freed from herbivore pressure, the plant lineage undergoes adaptive radiation into multiple new species. Eventually, an herbivore lineage evolves the ability to overcome the novel defense, gaining access to the newly diversified plant clade as an unexploited resource, and the herbivore lineage itself then diversifies. This alternating pattern of escape and radiation in both plants and herbivores can generate bursts of speciation and is considered one mechanism contributing to the extraordinary species richness observed in plant–herbivore systems. Recent phylogenetic studies have tested this hypothesis at macroevolutionary timescales with mixed results—while some clades show patterns consistent with escape-and-radiate dynamics, adaptation and speciation appear to be partially decoupled in many systems.

The fossil record provides several lines of evidence relevant to coevolution, particularly in the relationship between herbivorous dinosaurs and their food plants. During the Mesozoic Era, the dominant vegetation consisted of gymnosperms (conifers, cycads, ginkgoes) with an understory of ferns. Large sauropod dinosaurs, which fed at many levels and probably consumed whole plants, were the dominant herbivores. Bruce Tiffney and colleagues have proposed that the enormous size and destructive feeding behavior of sauropods created intense disturbance regimes that may have exerted selective pressure on plant communities. The radiation of angiosperms (flowering plants) during the mid-Cretaceous—beginning approximately 135 million years ago—coincided with the diversification of ornithischian dinosaurs such as hadrosaurs and ceratopsians, which fed closer to the ground where angiosperms grew as shrubs and low vegetation. Some researchers have suggested that ornithischian diversification and angiosperm radiation represent a form of diffuse coevolution, though direct evidence for tight plant–dinosaur coevolutionary interactions remains debated. A 2009 study by Butler and colleagues found that, with one possible exception, diversity patterns for major groups of herbivorous dinosaurs were not positively correlated with angiosperm diversity, suggesting the relationship is more complex than a simple coevolutionary narrative.

After the end-Cretaceous mass extinction (~66 million years ago), the loss of large dinosaur herbivores dramatically altered plant–animal interactions. The absence of megaherbivores for roughly 25 million years appears to have slowed the evolution of new plant species, and plant defensive features such as spines regressed during this period. The subsequent radiation of mammalian herbivores established a fundamentally different coevolutionary regime characterized by smaller-bodied, organ-level herbivory and the evolution of new mutualistic interactions such as fruit consumption and seed dispersal.

Host–parasite interactions represent one of the most intensely studied arenas of coevolution. Hosts evolve defenses such as immune resistance, tolerance, and behavioral avoidance, while parasites counter-evolve mechanisms to evade or overcome these defenses. Theoretical models of host–parasite coevolution have been developed since the 1950s, beginning with Mode (1958) and inspired by Haldane's remarks on the impact of infectious diseases on natural selection (1949) and Flor's discovery of complementary resistance and infectivity genes in flax and flax rust (1956). Two major genetic frameworks describe the infection interface: the matching-alleles (MA) model, where infection requires a genetic match between parasite and host, producing rapid fluctuating selection; and the gene-for-gene (GFG) model, where variation in the breadth of resistance and infectivity creates a more complex range of outcomes including stable polymorphism and slower oscillations.

Host–parasite coevolution has been implicated in several fundamental biological phenomena, including the evolution and diversification of immune systems, the maintenance of genetic diversity within and among populations, the evolution of sexual reproduction (via the Red Queen hypothesis), and the evolution of mate choice and sexual selection.

Plant–pollinator coevolution is among the most celebrated examples of mutualistic coevolution. The coevolution of angiosperms and their insect pollinators likely began in the Early Cretaceous, with fossil evidence of specialized beetle–angiosperm pollination preserved in Burmese amber (approximately 99 million years old). Over time, the mutual dependence between flowers and their pollinators produced extraordinary morphological matches, such as the deep nectary of the Malagasy star orchid (Angraecum sesquipedale) and the correspondingly long proboscis of its hawkmoth pollinator (Xanthopan morganii praedicta)—a relationship famously predicted by Darwin in 1862. Another well-studied example involves Central American Acacia species, which have evolved hollow thorns and extrafloral nectaries that serve as exclusive nesting sites and food sources for mutualistic ants, which in turn defend the plant against herbivores.

Recent theoretical advances have expanded the quantitative genetic framework of coevolution to incorporate interspecific indirect genetic effects (IIGEs). In a 2022 paper published in Evolution, McGlothlin and colleagues adapted the theory of intraspecific social evolution to model how the genotype of one species can influence trait expression in another. Their model shows that reciprocal IIGEs can create feedback loops that amplify or dampen the evolutionary response to selection, drive patterns of correlated evolution between species even when selection does not covary between them, and in extreme cases produce coevolutionary change even when only one species possesses genetic variance for the relevant trait. These findings highlight that the phenotypic interface of coevolution is more complex than simple fitness effects, and that the pathways through which species influence one another's trait expression must be considered alongside the pathways through which they influence one another's fitness.

Coevolution is widely regarded as one of the primary engines of biodiversity. By generating reciprocal selective pressures, it drives the evolution of novel traits, promotes diversification through escape-and-radiate dynamics and geographic mosaics of selection, and maintains genetic variation through frequency-dependent selection. Coevolution shapes not only pairwise species interactions but also the structure of entire ecological communities and food webs. Understanding coevolutionary dynamics is increasingly recognized as essential for addressing applied challenges in agriculture (crop–pest interactions), medicine (pathogen–host evolution and antimicrobial resistance), and conservation biology (disrupted mutualisms in fragmented ecosystems).