Coevolution

Coevolution is the process of reciprocal evolutionary change between two or more interacting species, in which heritable genetic change in one species alters the directional, stabilizing, or frequency-dependent natural selection acting on a second species, which in turn evolves a counter-adaptation that alters selection back on the first species. Although the concept was clearly articulated by Charles Darwin in On the Origin of Species (1859) ("coadaptations of organic beings to each other") and in his 1862 monograph on orchid pollination, the term "coevolution" was coined in 1964 by entomologist Paul R. Ehrlich and botanist Peter H. Raven in their landmark study of butterfly larvae and host-plant secondary defensive chemistry.

Strict evolutionary definitions formulated by Daniel H. Janzen (1980) distinguish true specific (pairwise) coevolution—simultaneous or step-wise reciprocal genetic change between two specific interacting lineages—from diffuse (guild) coevolution, in which an entire guild of species (such as multiple vertebrate frugivores or generalist insect herbivores) collectively exerts selective pressure on a suite of plant species, and from sequential evolution or ecological fitting, in which a colonizing species tracks pre-existing host resources without inducing reciprocal evolutionary change in the host.

Across the biosphere, coevolution operates along an ecological continuum spanning two primary modes: antagonistic coevolution (predator–prey, herbivore–plant, and host–parasite interactions), which generates escalatory trait matching or perpetual Red Queen allele-frequency cycling; and mutualistic coevolution (plant–pollinator, seed-disperser, mycorrhizal, and intracellular symbioses), which drives trait convergence, morphological complementarity, and, at its macroevolutionary limit, metabolic integration as seen in endosymbiotic theory. At the landscape scale, John N. Thompson's (1994, 2005) Geographic Mosaic Theory of Coevolution (GMTC) demonstrates that these interactions are spatially structured across selection mosaics, coevolutionary hotspots, and coldspots connected by gene flow.

Darwin's Angraecum sesquipedale Prediction

In his 1862 treatise On the Various Contrivances by which British and Foreign Orchids are Fertilised by Insects, Charles Darwin examined the endemic Malagasy star orchid Angraecum sesquipedale, whose pale green flowers possess a whip-like nectar spur measuring 25–35 cm in length, with nectar secreted only in the bottom 2–4 cm. Darwin hypothesized that this extreme floral tube evolved through a reciprocal coevolutionary ratchet with a specialized sphingid hawkmoth: orchids with deeper spurs forced moths to press their heads firmly against the viscidium to reach the nectar, maximizing pollinia attachment, while moths with longer proboscises gained higher foraging energy intake and reproductive output. In 1903—forty-one years after Darwin's prediction—Walter Rothschild and Karl Jordan described the predicted Malagasy hawkmoth subspecies Xanthopan morganii praedicta, possessing an uncoiled proboscis averaging 22–28 cm.

Ehrlich and Raven's (1964) Escape-and-Radiation Model

Analyzing the phylogenetic distribution of Papilionoidea butterflies across angiosperm families, Ehrlich and Raven (1964) formulated the escape-and-radiation coevolutionary hypothesis to explain why plants and phytophagous insects comprise the majority of terrestrial eukaryotic species:

1. A plant lineage evolves a novel secondary defensive metabolite through mutation and recombination—such as furanocoumarins in Apiaceae, cardenolide cardiac glycosides in Asclepias milkweeds, or glucosinolates (mustard oil glycosides) in Brassicaceae—that renders its foliage toxic or unpalatable to generalist herbivores.
2. Released from herbivore grazing pressure, the chemically defended plant clade enters a novel adaptive zone and undergoes rapid speciation and ecological radiation.
3. Subsequently, an insect herbivore lineage evolves a physiological detoxifying mechanism—such as specialized cytochrome P450 monooxygenases (CYP6B in Papilio swallowtails feeding on furanocoumarins), target-site insensitivity in Na⁺/K⁺-ATPase (Danaus plexippus monarch caterpillars feeding on milkweed cardenolides), or nitrile-specifier proteins (Pieris cabbage white butterflies disarming glucosinolates)—allowing the insect clade to colonize the unexploited plant radiation and diversify in parallel.

Quantitative Genetics of Escalatory Coevolutionary Antagonism

Consider a predator (or parasite/herbivore) species with mean offensive trait z̄_x and a prey (or host) species with mean defensive trait z̄_y. Following Russell Lande (1976) and Scott L. Nuismer (2017), under a trait-escalation (phenotype-difference) model, the outcome of an encounter depends on the difference z_x − z_y, whereas under a phenotype-matching model, the parasite succeeds only when its phenotype closely matches the host phenotype (z_x ≈ z_y). Given additive genetic variances G_x and G_y, the coupled single-generation evolutionary trajectories of the two interacting species are:

Δ*z̄*_x = G_x β_x(z̄_x, z̄_y) and Δ*z̄*_y = G_y β_y(z̄_x, z̄_y)

where the selection gradients β_x = ∂ ln w̄_x / ∂*z̄*_x and β_y = ∂ ln w̄_y / ∂*z̄*_y couple the fitness landscapes of both species:

β_x = α_x(z̄_y − z̄_x) − γ_x(z̄_x − θ_x)

Here α_x measures the strength of biotic interaction selection and γ_x measures stabilizing ecological selection toward each species' physiological baseline optimum θ_x (reflecting the metabolic cost of defense or offense). Escalation continues until the marginal survival benefit of stronger defense is balanced by its physiological cost (β_x = β_y = 0).

Tetrodotoxin Escalation in Taricha Newts and Thamnophis Snakes

The textbook empirical model of escalatory coevolutionary antagonism is the predator–prey interaction between the rough-skinned newt (Taricha granulosa) and the common garter snake (Thamnophis sirtalis) across western North America, investigated over four decades by Edmund D. Brodie Jr. and Edmund D. Brodie III (1990, 1999, 2002). Taricha granulosa concentrates the potent neurotoxin tetrodotoxin (TTX) in its dorsal skin glands, which binds to the outer pore of voltage-gated sodium channels (Na_V1.4) in vertebrate skeletal muscle and nerves, blocking Na⁺ ion influx and causing death by respiratory paralysis.

In co-occurring populations, Thamnophis sirtalis has evolved high physiological resistance to ingested TTX through specific amino acid substitutions in the domain IV pore loop (p-loop) of the skeletal muscle sodium channel gene SCN4A (encoding Na_V1.4). However, these pore mutations alter channel gating kinetics, imposing a measurable physiological cost in reduced crawl speed and thermoregulatory performance. Consequently, across the Pacific Northwest landscape, newt toxicity and snake resistance vary across three orders of magnitude in tight spatial lockstep.

Host–Pathogen Gene-for-Gene vs. Matching-Alleles Dynamics

In host–parasite coevolution, reciprocal selection operates on discrete molecular recognition loci via two primary genetic architectures:

1. Flor's Gene-for-Gene (GFG) Model: Discovered by plant pathologist Harold H. Flor (1942, 1955) in cultivated flax (Linum usitatissimum) and the rust fungus Melampsora lini, infection fails (incompatible defense activation) if and only if the host carries a dominant resistance allele (R, typically encoding an intracellular nucleotide-binding leucine-rich repeat NBS-LRR receptor) that specifically recognizes a corresponding dominant avirulence effector allele (Avr) in the pathogen. Pathogens carrying a mutated recessive virulence allele (avr) evade detection and infect both R and r hosts, which would lead to fixation of avr and loss of R unless both virulence (avr) and resistance (R) carry intrinsic pleiotropic fitness costs (c_R, c_avr > 0), generating cyclical Red Queen oscillations.
2. Matching-Alleles (MA) Model: Prevalent in invertebrate innate immunity (such as the freshwater snail Potamopygus antipodarum and its castrating trematode parasite Microphallus, studied by Curtis M. Lively) and vertebrate MHC antigen presentation, infection succeeds only when the parasite's surface antigen alleles exactly match the host's self/receptor genotype (A₁ infects H₁, A₂ infects H₂). Because any common host genotype becomes the prime target for rapidly evolving parasites, rare host genotypes enjoy a fitness advantage, driving continual negative frequency-dependent selection and preserving sexual reproduction and genetic polymorphism.

In mutualistic coevolution, both interacting species experience net positive fitness increments (w₁₂ > w₁, w₂), whether through diffuse guilds (such as fleshy-fruited angiosperms coevolving with frugivorous birds and bats) or tightly coupled obligate brood-site pollination mutualisms such as figs (Ficus, >800 species) and agaonid fig wasps, or yuccas (Yucca) and prodoxid yucca moths (Tegeticula). In the Yucca–Tegeticula system, female moths possess unique coiled maxillary tentacles found in no other lepidopteran family, which they use to scrape pollen from anthers, pack it into a subcephalic pellet, and actively stuff it into the stigmatic cavity of a flower in which they have just oviposited. Because moth larvae consume a fraction of the developing seeds, theoretical stability against "cheater" moths that lay excessive eggs without pollinating is enforced by host sanctions: as demonstrated by Olle Pellmyr and Chad J. Huth (1994), Yucca plants selectively abscise (abort) flowers that receive high egg loads or low pollen loads.

Synthesizing antagonistic and mutualistic systems across heterogeneous landscapes, John N. Thompson (1994, 2005) formulated the Geographic Mosaic Theory of Coevolution (GMTC) around three testable principles:

1. Selection mosaics: The fitness outcome of an interspecific interaction varies qualitatively across environments (G × G × E interactions). For instance, the woodland star Lithophragma parviflorum and its floral moth Greya politella interact as mutualists in habitats lacking co-pollinators, but shift to commensalism or antagonism where bombyliid bee flies provide abundant free pollination.
2. Coevolutionary hotspots and coldspots: Reciprocal selection occurs actively only within local hotspots (such as coastal Oregon for Taricha–Thamnophis), which are embedded within a broader matrix of coldspots where one species is absent or interaction selection is unilateral or zero.
3. Trait remixing: Gene flow, random genetic drift, and local metapopulation extinction-recolonization continually redistribute coevolved alleles between hotspots and coldspots, preventing species-wide fixation and maintaining spatial genetic variation.

Related concepts include Gene–Culture Coevolution and Lichen.

What is the difference between pairwise coevolution and diffuse coevolution?

Pairwise (specific) coevolution occurs when two individual species exert reciprocal natural selection directly on each other, such as the rough-skinned newt (Taricha granulosa) and the common garter snake (Thamnophis sirtalis). Diffuse (guild) coevolution occurs when reciprocal selection operates among entire suites of interacting species—such as multiple insect herbivores feeding on a family of toxic plants, or a guild of bees and hummingbirds pollinating tubular wildflowers.

How did Ehrlich and Raven explain the high diversity of plants and insects?

In their 1964 "escape-and-radiation" model, Paul Ehrlich and Peter Raven proposed that when a plant lineage mutates a novel toxic secondary metabolite (such as milkweed cardenolides or mustard glucosinolates), it escapes herbivory and radiates into a diverse clade. When a specialized insect lineage subsequently evolves biochemical detoxification against that chemical, it colonizes the unexploited plant clade and undergoes a parallel adaptive radiation.

What are coevolutionary hotspots and coldspots in the Geographic Mosaic Theory?

Formulated by John N. Thompson (1994, 2005), coevolutionary hotspots are local geographic populations where two interacting species exert strong reciprocal natural selection on each other's traits. Coevolutionary coldspots are populations within the species' ranges where the interacting partner is absent, rare, or rendered ecologically neutral by third-party species, resulting in no reciprocal selection; gene flow continually mixes traits between hotspots and coldspots.

1. Ehrlich, P. R., & Raven, P. H. (1964). "Butterflies and plants: A study in coevolution." Evolution, 18(4), 586–608. https://doi.org/10.1111/j.1558-5646.1964.tb01674.x
2. Janzen, D. H. (1980). "When is it coevolution?" Evolution, 34(3), 611–612. https://doi.org/10.1111/j.1558-5646.1980.tb04849.x
3. Brodie, E. D., III, & Brodie, E. D., Jr. (1999). "Predator-prey arms races: Asymmetrical selection on predators and prey." BioScience, 49(7), 557–568. https://doi.org/10.2307/1313476
4. Thompson, J. N. (2005). The Geographic Mosaic of Coevolution. University of Chicago Press, Chicago. https://doi.org/10.7208/chicago/9780226118697.001.0001