19.13.01 · eco-evo-bio / coevolution

Coevolution

shipped3 tiersLean: nonepending prereqs

Anchor (Master): Thompson, J. N. — The Geographic Mosaic of Coevolution (2005); relevant primary literature

Intuition Beginner

No species evolves in isolation. Every organism interacts with other species -- as predators, prey, competitors, hosts, parasites, pollinators, or mutualists. When two species each exert selective pressure on the other, and each evolves in response to the other's evolution, the result is coevolution: reciprocal evolutionary change between interacting species.

The simplest form is a predator-prey arms race. Predators evolve better weapons (sharper claws, faster speed, improved senses), and prey evolve better defenses (thicker shells, faster escape, camouflage). Each improvement by one side selects for a counter-improvement by the other, driving escalating adaptation. Cheetahs (the fastest land predator) and gazelles (their swift prey) have each evolved extraordinary speed because only the fastest cheetahs catch gazelles, and only the fastest gazelles escape cheetahs.

Host-parasite coevolution follows a similar pattern but with a twist: parasites typically evolve faster than their hosts because they have shorter generation times and larger populations. This is the basis of the Red Queen hypothesis (named after the Red Queen in Lewis Carroll's Through the Looking Glass, who says "it takes all the running you can do to keep in the same place"). The hypothesis proposes that organisms must constantly evolve just to maintain their fitness relative to evolving opponents (parasites, competitors). Sex and recombination may have evolved as a strategy for hosts to generate novel genetic combinations that stay ahead of rapidly evolving parasites.

Not all coevolution is antagonistic. Mutualistic coevolution occurs when both species benefit from the interaction and each evolves to enhance the partnership. Flowers and their pollinators are a classic example: flowers evolve colors, shapes, and nectar rewards that attract specific pollinators, while pollinators evolve mouthparts, behaviors, and sensory preferences that efficiently extract food from those flowers. The long tubular flowers of certain orchids and the correspondingly long proboscis of their hawk moth pollinators are a dramatic example of coevolutionary matching.

Coevolution can also be diffuse, involving multiple species simultaneously rather than strict pairwise interactions. An herbivorous insect may face defensive chemicals from many different plant species, and a plant may be attacked by many different herbivores. The coevolutionary dynamics in such networks are complex and can produce community-wide patterns of adaptation.

Visual Beginner

Types of coevolution:

Type	Interaction	Example
Predator-prey arms race	Antagonistic	Cheetah-gazelle speed escalation
Host-parasite (Red Queen)	Antagonistic	Human immune system vs. influenza virus
Competitive coevolution	Antagonistic	Character displacement in Darwin's finches
Mutualism (pollination)	Mutualistic	Orchids and hawk moths
Mutualism (mycorrhizae)	Mutualistic	Fungi and plant roots (nutrient exchange)
Mutualism (coral-algae)	Mutualistic	Reef-building corals and zooxanthellae
Diffuse coevolution	Multiple species	Plant defensive chemistry vs. insect herbivore community

Coevolutionary outcomes:

Pairwise coevolution:
  Species A evolves --> selects for change in Species B --> selects for change in Species A --> ...

Diffuse coevolution:
  Plant P defends against Herbivores H1, H2, H3 simultaneously
  H1, H2, H3 each evolve countermeasures
  The total selective pressure on P reflects ALL herbivores, not just one

Coevolutionary alternation:
  Parasite P specializes on Host H1
  H1 evolves resistance
  P shifts to alternate host H2
  H2 evolves resistance
  P may return to H1 if resistance has costly trade-offs

Worked example Beginner

The coevolution of rough-skinned newts (Taricha granulosa) and common garter snakes (Thamnophis sirtalis) is a textbook example of a predator-prey arms race at the molecular level.

The newt produces tetrodotoxin (TTX), one of the most potent neurotoxins known (the same toxin found in pufferfish). A single newt contains enough TTX to kill approximately 10 adult humans. The toxin blocks voltage-gated sodium channels in nerve and muscle cells, causing paralysis and death.

Garter snakes that prey on these newts have evolved resistance to TTX through amino acid substitutions in their sodium channel genes that reduce TTX binding. The snakes can eat highly toxic newts that would kill any other predator.

But the arms race continues. In some populations, newts have evolved even higher toxin levels, selecting for even greater resistance in snakes. The geographic pattern tells the story: in regions where newts and snakes co-occur, both toxicity and resistance are high. In regions where snakes are absent, newts have lower toxin levels (there is no selective pressure to maintain expensive toxin production). And in regions where newts are nontoxic, snakes have low resistance.

The cost of resistance is key: TTX-resistant sodium channels have slightly altered function that reduces the snake's locomotor performance. In the absence of toxic prey, non-resistant snakes actually perform better, explaining why resistance is not universal.

Check your understanding Beginner

Exercise (easy, multiple choice).

The Red Queen hypothesis proposes that organisms must constantly evolve primarily to:

A. Adapt to changing physical environments B. Keep up with the evolutionary advances of their biological opponents (parasites, competitors) C. Increase their body size over generations D. Develop more complex behaviors

Hint

The Red Queen said "it takes all the running you can do to keep in the same place." What does "running" represent in evolutionary terms, and who is the opponent?

Answer

Option B. The Red Queen hypothesis proposes that organisms must continually evolve to maintain their relative fitness against coevolving opponents, particularly parasites and pathogens. Like the Red Queen in Carroll's story, species must "run" (evolve) as fast as they can just to "stay in the same place" (maintain their current fitness). This is because their opponents (parasites, with short generation times and large populations) are also evolving rapidly. The hypothesis has been invoked to explain the persistence of sexual reproduction (sex generates novel genetic combinations that may stay ahead of parasites) and the prevalence of recombination and immune system diversity.

Exercise (easy, multiple choice).

Which of the following is an example of mutualistic coevolution?

A. Lions evolving stronger jaws to kill zebras B. Acacia trees producing nutritious Beltian bodies that ants eat, while ants defend the tree from herbivores C. Two species of birds competing for the same nesting sites D. A fungus killing a tree

Hint

Mutualistic coevolution involves two species both benefiting from the interaction and evolving to enhance the partnership. Which option describes a mutually beneficial relationship?

Answer

Option B. The acacia-ant mutualism is a classic example of coevolved mutualism. Acacia trees have evolved hollow thorns (nesting sites for ants), Beltian bodies (nutritious structures at leaf tips that ants harvest), and extrafloral nectaries (sugar sources for ants). In return, ants have evolved aggressive defense behavior: they attack herbivorous insects, remove fungal spores, and even剪除 nearby competing vegetation. Both species benefit, and each has evolved traits specifically to maintain and enhance the partnership.

Exercise (easy, true/false).

Coevolution can only occur between two species at a time.

Hint

Consider a plant that produces defensive chemicals. How many herbivore species might be affected by those chemicals? Can multiple species exert reciprocal selective pressures?

Answer

False. While pairwise coevolution (two species reciprocally evolving) is the simplest form, diffuse coevolution involves multiple interacting species simultaneously. A plant may be attacked by dozens of herbivore species, each exerting selection on the plant's defensive chemistry, and the plant's defenses affect all herbivores. Similarly, the pollinator community of a plant may include many species, and the plant's floral traits reflect the aggregate selective pressure from all pollinators. Diffuse coevolution is probably more common than pairwise coevolution in natural communities.

Formal definition Intermediate+

Coevolution is defined as reciprocal evolutionary change between two or more interacting species, where a genetic change in species A selects for a genetic change in species B, which in turn selects for further change in species A.

Janzen (1980) emphasized that coevolution requires specificity (the interaction is particular to the pair of species), reciprocity (both species evolve), and simultaneity (the evolutionary changes are concurrent). These criteria distinguish coevolution from simpler cases where one species adapts to another without reciprocal change.

Types of coevolution

Specific (pairwise) coevolution: Two species reciprocally evolve in response to each other. Example: the yucca-yucca moth mutualism, where the yucca is pollinated exclusively by the yucca moth, and the moth larvae feed exclusively on yucca seeds.

Diffuse (guild) coevolution: Multiple species in a guild (e.g., all herbivores of a plant, or all pollinators) exert reciprocal selective pressures. Example: Ehrlich and Raven (1964) showed that the diversification of butterflies and their host plants was correlated, with shifts to new plant families driving butterfly speciation -- but the interaction involves many butterfly species and many plant families.

Escape-and-radiate coevolution: One lineage evolves a key innovation that frees it from its coevolutionary antagonist, allowing rapid diversification, followed by the antagonist eventually "catching up." Example: the evolution of cardiac glycoside production in milkweed plants allowed an escape from generalist herbivores and a radiation of milkweed species; subsequently, monarch butterflies evolved resistance to the glycosides and radiated onto milkweeds.

Antagonistic coevolution: mathematical framework

For a host-parasite system, the coevolutionary dynamics can be modeled using a matching-alleles model (MAM) or a gene-for-gene model:

In the MAM, infection occurs when the parasite's "attack" alleles match the host's "defense" alleles. The change in allele frequency in the host (to escape infection) and the parasite (to match the host) produces oscillatory dynamics:

$\frac{d p _{H}}{d t} = - s_{H} p_{H} (1 - p_{H}) (p_{H} - p_{P})$

$\frac{d p _{P}}{d t} = s_{P} p_{P} (1 - p_{P}) (p_{H} - p_{P})$

where $p_{H}$ and $p_{P}$ are the frequencies of a matching allele in host and parasite, and $s_{H}$ and $s_{P}$ are selection coefficients. This system produces negative frequency-dependent selection: rare host alleles have an advantage (because parasites have not evolved to match them), maintaining genetic polymorphism at host defense loci.

Coevolution and chemical defense

The evolution of chemical defenses in plants and the counter-adaptations of herbivores is one of the richest areas of coevolutionary research. Plants produce an enormous diversity of secondary metabolites (alkaloids, terpenes, phenolics, cyanogenic glycosides) that are toxic, repellent, or antinutritive to herbivores. Herbivores, in turn, evolve detoxification enzymes (cytochrome P450 monooxygenases, glutathione S-transferases), behavioral avoidance, and sequestration mechanisms (storing plant toxins for their own defense).

Key results Intermediate+

Result 1 (Ehrlich and Raven, 1964). The diversification rates of butterflies and their host plant families are correlated. Clades of butterflies that shifted to feeding on a new plant family experienced accelerated speciation, because the evolutionary innovation required to detoxify a new class of plant defensive chemistry opened access to an underexploited resource. This "escape-and-radiate" pattern has been documented in multiple plant-herbivore systems and suggests that coevolutionary innovations are major drivers of biodiversity.

Result 2 (Red Queen dynamics in natural populations). Long-term studies of host-parasite systems have confirmed Red Queen dynamics. The Daphnia-Pasteuria system in freshwater ponds shows that parasite genotypes that are rare in one year become common in the next (negative frequency-dependent selection), while host genotypes that were common become rare. This dynamic cycling maintains genetic diversity at both host resistance loci and parasite infectivity loci.

Exercise 1

Exercise 1 (medium, short answer).

Explain how the geographic mosaic theory of coevolution differs from the traditional view of pairwise coevolution. What evidence supports the geographic mosaic theory?

Hint

The traditional view treats coevolution as a uniform process across a species' range. Thompson's geographic mosaic theory proposes that coevolutionary dynamics vary across the landscape.

Answer

The geographic mosaic theory of coevolution (Thompson, 2005) proposes three processes that structure coevolution across landscapes:

Geographic variation in selection: The strength and direction of coevolutionary selection vary among populations because the ecological context differs (different community composition, different abiotic conditions).
Coevolutionary hotspots and coldspots: Some geographic areas have strong reciprocal selection (hotspots), while others lack one or both interactors or have weak selection (coldspots).
Trait remixing: Gene flow, drift, and extinction-colonization dynamics continually remix coevolved traits across populations.

Evidence comes from studies like the Greya moth and Lithophragma plant system in western North America, where some populations show reciprocal coevolution (mutualistic pollination), others show only antagonistic interactions (larvae eat seeds without pollinating), and still others lack the interaction entirely. The geographic mosaic predicts that local maladaptation should be common (because gene flow brings in traits selected under different conditions), and this has been confirmed: many host-parasite systems show geographic patterns where parasites are better adapted to local hosts in some populations but poorly adapted in others.

Exercise 2

Exercise 2 (medium, symbolic).

In a simple matching-alleles model of host-parasite coevolution, the host has two alleles (A, a) and the parasite has two alleles (B, b). Infection occurs when the parasite allele matches the host allele (A-B and a-b are matching pairs). If the initial frequency of A in the host is 0.8 and B in the parasite is 0.5, explain why negative frequency-dependent selection will cause allele frequencies to oscillate over time.

Hint

At the start, which host allele is more common? Which parasite allele will be favored? Once that parasite allele becomes common, which host allele will then be favored?

Answer

At the start: host allele A has frequency 0.8 (common), allele a has frequency 0.2 (rare). Parasite allele B has frequency 0.5.

Parasite adapts to common host: The parasite allele B (which matches host allele A, the common host genotype) is favored by selection because it can infect the majority of hosts. Parasite B frequency increases.
Rare host allele is favored: As parasite B becomes common, hosts with allele A are increasingly infected. Host allele a (rare) is now at an advantage because the matching parasite allele b is still rare. Host allele a frequency increases.
Parasite counter-adapts: As host allele a becomes more common, parasite allele b (matching a) is now favored and increases.
The cycle repeats: As parasite b becomes common, host A is again favored. This produces oscillating allele frequencies -- a Red Queen dynamic.

This negative frequency-dependent selection maintains genetic polymorphism indefinitely: rare alleles always have an advantage because the matching parasite/infectivity allele is also rare. The system oscillates around 0.5 for both loci, with the amplitude and period depending on the relative strengths of selection in host and parasite.

Advanced treatment Master

The formal theory of coevolution has developed along several complementary lines, integrating population genetics, community ecology, and phylogenetic comparative methods.

Gene-for-gene versus matching-alleles models. The two primary genetic architectures for host-parasite coevolution make different predictions. In the gene-for-gene model (originally developed for plant-pathogen systems by Flor, 1956), host resistance (R) genes are dominant and specific: an R gene provides resistance against a pathogen carrying the corresponding avirulence (Avr) gene, but not against pathogens lacking that Avr gene. In the matching-alleles model (MAM), infection requires an exact match between host and parasite alleles at all loci, and there is no directional dominance. The MAM produces negative frequency-dependent dynamics and is more appropriate for systems where specificity is high (e.g., Daphnia-Pasteuria, snail-schistosome). The gene-for-gene model is more appropriate for systems where resistance is directional and dominance matters (e.g., wheat-rust fungus).

The geographic mosaic and community-level consequences. Thompson's geographic mosaic theory has profound implications for understanding biodiversity maintenance. If coevolutionary hotspots generate locally adapted traits that are then exported to coldspots via gene flow, the mosaic structure can maintain genetic diversity across the entire metapopulation. This connects coevolution to metacommunity theory: the regional diversity of traits is a product of local coevolutionary dynamics filtered by dispersal and drift.

Coevolutionary consequences for speciation. Coevolution can drive speciation through several mechanisms. Reproductive character displacement occurs when two species that share a pollinator evolve divergent floral traits (different flowering times, different flower colors, different floral shapes) to reduce costly hybrid pollen transfer. This process, documented in Phlox species by Levin and Kerster (1967), is coevolutionary because the plant species are each exerting selection on the other through their effects on pollinator behavior. Host race formation in parasites is another pathway: when a parasite population shifts to a new host species, assortative mating on the new host (because mates are encountered on the host) creates pre-zygotic reproductive isolation between the parasite populations on different hosts. This mechanism of sympatric speciation has been proposed for Rhagoletis fruit flies, where a host shift from hawthorn to apple approximately 150 years ago has produced partially reproductively isolated fly populations.

Evolutionary cascades. Coevolution can trigger ecological cascades that propagate through multiple trophic levels. The introduction of a novel predator can select for behavioral changes in prey, which alters the prey's feeding behavior on plants, which changes plant community composition, which affects soil nutrient cycling. Such coevolutionary cascades demonstrate that the evolutionary dynamics of a single species pair can reshape entire ecosystems. The coevolution of myxoma virus and European rabbits in Australia is a canonical example: the virus evolved reduced virulence (allowing the rabbit to survive longer and transmit more), the rabbits evolved resistance, and the resulting equilibrium reshaped Australian plant communities that had been devastated by overabundant rabbits.

The Red Queen and the maintenance of sex. One of the most significant contributions of coevolutionary theory is the Red Queen hypothesis for the maintenance of sexual reproduction. The paradox of sex is that asexual reproduction is theoretically twice as efficient as sexual reproduction (an asexual female passes 100% of her genes to offspring, while a sexual female passes only 50%). Why sex is maintained in the vast majority of eukaryotic species despite this "twofold cost of sex" has been a central question in evolutionary biology. The Red Queen hypothesis proposes that sex is maintained because it generates novel genetic combinations through recombination, allowing hosts to stay ahead of rapidly evolving parasites. In a coevolutionary arms race with parasites, asexual hosts produce genetically identical offspring, all of which are vulnerable to the same parasite genotypes. Sexual hosts produce genetically diverse offspring, some of which may carry rare resistance alleles that are effective against currently common parasite genotypes.

Empirical support for the Red Queen hypothesis comes from several sources. Comparative studies show that species with higher parasite loads are more likely to reproduce sexually. Experimental evolution studies with Caenorhabditis elegans nematodes found that sexual populations outcompeted asexual populations when exposed to coevolving parasites, but not in parasite-free environments. Geographic studies of freshwater snails (Potamopyrgus antipodarum) in New Zealand show that the frequency of sexual individuals is higher in populations with high parasite prevalence, exactly as predicted by the Red Queen. These results demonstrate that host-parasite coevolution is a major selective force maintaining sexual reproduction in natural populations.

Fig-wasp mutualism: the most specialized coevolution. The fig-fig wasp mutualism is one of the most ancient and tightly coevolved interactions known, with fossil evidence dating to approximately 65 Ma. Each of the approximately 750 fig species (Ficus) is pollinated by its own specific species of agaonid wasp, and each wasp species reproduces only within the figs of its host plant. The fig's inflorescence (the syconium) is a closed, urn-shaped structure lined with hundreds of tiny flowers on the inside. The female wasp enters through a small opening (the ostiole), often losing her wings and antennae in the process. She pollinates the internal flowers (either actively, carrying pollen in specialized pockets, or passively, carrying pollen on her body) and lays eggs in the ovaries of some flowers. The wasp larvae develop within the fig's ovules, and the next generation of wingless male wasps emerge first, mate with females still inside their galls, and then chew exit tunnels through which the mated, pollen-carrying females escape to find new figs.

This extreme specificity creates a powerful coevolutionary dynamic: the fig's reproductive success depends entirely on its specific wasp, and the wasp's reproductive success depends entirely on its specific fig. The fig enforces cooperation through "sanctions": if a wasp fails to pollinate (carries no pollen), the fig can abort the seeds in that syconium, killing the wasp's offspring. This sanction mechanism aligns the evolutionary interests of both partners, preventing the wasp from evolving into a pure parasite. Phylogenetic studies show that fig and wasp phylogenies are largely congruent (cospeciation), meaning that when a fig species splits into two through speciation, its wasp typically splits as well, maintaining the one-to-one matching. However, some cases of host switching and duplication (multiple wasp species on one fig) have been documented, adding complexity to the coevolutionary picture.

Coevolution of mating signals and receiver psychology. Sexual selection and coevolution intersect in the evolution of mating signals. Male display traits (bird song, firefly flashes, butterfly wing patterns) coevolve with female preferences for those traits. The sensory drive model proposes that signals evolve to exploit the pre-existing sensory biases of receivers: if females have a sensory bias for a particular color or pattern, males that display that trait will be favored. Over time, the female preference and the male signal coevolve, potentially producing elaborate traits that are costly to produce but attractive to females (the handicap principle). This coevolutionary process can drive rapid divergence in mating signals between populations, contributing to speciation. The cichlid fish of Lake Victoria, discussed in the context of macroevolution 19.08.01, provide a dramatic example: male breeding coloration (blue versus red) coevolves with female visual sensitivity, which is tuned to the ambient light at different depths. This coevolution of signal and receiver creates reproductive isolation by depth, contributing to the explosive sympatric speciation observed in this group.

Coevolution in the fossil record. Detecting coevolution in the fossil record is challenging because it requires demonstrating reciprocal evolutionary change in two interacting lineages through time. However, several compelling examples exist. The coevolution of nacreous (mother-of-pearl) shell microstructure in ammonites and the crushing bite force of marine reptiles (mosasaurs) during the Late Cretaceous shows a progressive escalation: ammonite shells became thicker and more complexly ornamented, while mosasaur jaws became more robust and their teeth more specialized for crushing. The parallel increase in drilling predation by gastropods on bivalves through the Cenozoic (the "escalation" documented by Vermeij, 1987) provides another example: as drilling predators became more effective, bivalves evolved thicker shells, more ornamentation, and behavioral defenses (deeper burrowing), which in turn selected for more sophisticated drilling strategies. Vermeij's "escalation hypothesis" proposes that the long-term increase in the frequency and intensity of predation over geological time has been a major driver of morphological evolution in marine invertebrates, making predation-driven coevolution one of the dominant macroevolutionary forces in the history of life.

Coevolutionary networks and community stability. Modern coevolutionary research has expanded beyond pairwise interactions to study entire networks of interacting species. Plant-pollinator networks, host-parasite networks, and predator-prey food webs are structured by coevolutionary history. A key finding is that these networks are typically nested: specialists (species that interact with few partners) tend to interact with the most connected generalists, while generalists interact with both specialists and other generalists. This nested structure is thought to promote community stability because it minimizes competition for shared partners and ensures that the loss of a single species does not cascade through the network. However, nested networks are also vulnerable to the loss of highly connected generalist species, which support many specialists. The coevolutionary processes that generate nested structure -- particularly the tendency for specialists to evolve to exploit the most abundant and reliable partners -- are an active area of research.

Coevolution and the human microbiome. The human body is home to trillions of microorganisms (the microbiome) that coevolve with their host. The gut microbiome, in particular, is a coevolved community that provides essential services: digestion of complex carbohydrates, synthesis of vitamins, immune system development, and protection against pathogens. The relationship between humans and their gut bacteria is a combination of mutualistic coevolution (both partners benefit) and conflict (bacteria may evolve to exploit the host if the opportunity arises). The rapid evolution of gut bacteria (generation times of minutes to hours, compared to decades for humans) means that the coevolutionary dynamics are strongly asymmetric: the microbiome can evolve on timescales that are effectively instantaneous from the host's perspective. This has implications for health and disease: dysbiosis (disruption of the normal microbiome) may reflect the breakdown of coevolved mutualisms due to modern diets, antibiotic use, or other environmental changes. The coevolutionary perspective suggests that restoring a healthy microbiome may require more than simply introducing beneficial bacteria; it may require recreating the environmental conditions under which the coevolved mutualism was stable.

Connections Master

Macroevolution 19.08.01. Coevolutionary arms races and escape-and-radiate dynamics are macroevolutionary processes that drive lineage diversification. The correlation between plant and herbivore diversification rates documented by Ehrlich and Raven (1964) is a macroevolutionary pattern produced by coevolutionary microevolution.
Ecosystem ecology 19.11.01. Mutualistic coevolution (mycorrhizae, coral-algae symbiosis, nitrogen-fixing bacteria) is foundational to ecosystem function. Mycorrhizal fungi associate with approximately 80% of plant species and are critical for phosphorus uptake; the plant-fungus nutrient exchange is a coevolved mutualism that underlies terrestrial ecosystem productivity.
Biogeography 19.12.01. The geographic mosaic of coevolution is inherently spatial: the variation in coevolutionary dynamics across populations is a biogeographic pattern. The distribution of hotspots and coldspots, and the gene flow between them, are shaped by the same processes (dispersal, vicariance, environmental gradients) that structure species distributions.
Conservation biology 19.14.01. Coevolved mutualisms are vulnerable to disruption. The decline of pollinators (bees, butterflies, bats) threatens the reproduction of coevolved plants. Coral bleaching (breakdown of the coral-algae mutualism due to ocean warming) is a coevolutionary relationship pushed past its environmental tolerance. Conservation strategies must account for coevolutionary dependencies.
Community ecology 19.10.01. Coevolution shapes the structure of ecological communities. Diffuse coevolution among multiple interacting species produces community-wide patterns of trait matching and mismatching. Trophic cascades, keystone species effects, and the stability of food webs all depend on the coevolutionary history of the interacting species. The concept of "ecological fitting" -- where species interact based on pre-existing traits rather than coevolved matching -- provides a null model for understanding when community patterns reflect coevolution versus incidental interaction.
Microevolution 19.07.01. Coevolution is microevolution applied to the case where the selective environment is itself evolving. The population genetic models that describe allele frequency change in single species (selection, drift, gene flow) are the building blocks of coevolutionary theory. The matching-alleles and gene-for-gene models extend single-species population genetics to two-species systems by coupling the fitness landscapes of the interactors. Understanding coevolution therefore requires a firm grounding in microevolutionary mechanisms.
Immunology 18.10.01. The vertebrate immune system is a coevolutionary product of the host-parasite arms race. The extraordinary diversity of the major histocompatibility complex (MHC) -- the most polymorphic gene region in vertebrates -- is maintained by negative frequency-dependent selection driven by pathogens. Pathogens evolve to evade the most common MHC alleles, favoring hosts with rare alleles. This is a direct molecular manifestation of Red Queen dynamics at the organismal level. The evolution of adaptive immunity itself, with its somatic recombination mechanisms generating millions of novel antigen receptors, can be interpreted as an evolutionary strategy for generating the genetic diversity needed to keep pace with rapidly evolving pathogens.

Historical & philosophical context Master

Coevolution became a named framework when biologists began studying reciprocal adaptation rather than one-sided selection. Plant defenses and herbivore counter-defenses, host-parasite arms races, pollination mutualisms, mimicry systems, and predator-prey dynamics showed that species often form each other's selective environments. The concept is powerful because it replaces static adaptation with moving targets. A trait can be advantageous only relative to another lineage's current traits, and that lineage may respond in turn.

The concept of coevolution was formally introduced by Ehrlich and Raven in their 1964 paper "Butterflies and plants: a study in coevolution," which documented the correlation between butterfly phylogenetic diversification and shifts onto new host plant families. Their insight was that the chemical arms race between plants (evolving novel defensive compounds) and butterflies (evolving detoxification mechanisms) was a primary driver of biodiversity in both groups. This paper established the escape-and-radiate model and made coevolution a central concept in evolutionary ecology.

The Red Queen hypothesis, proposed by Leigh Van Valen in 1973, extended coevolutionary thinking to explain broad patterns in the fossil record. Van Valen observed that the probability of extinction for a taxonomic group is approximately constant and independent of how long the group has existed. He interpreted this as evidence that extinction is driven by the biotic environment (other species) rather than the physical environment, because the biotic environment is itself constantly evolving. The Red Queen hypothesis provided a framework for understanding why species must continually evolve to maintain their fitness, and why evolutionary stasis is rare in the face of coevolving antagonists.

Daniel Janzen's 1980 paper "When is it coevolution?" provided important conceptual clarification by arguing that the term should be reserved for cases where there is specificity (the interaction is particular to the pair), reciprocity (both species evolve), and simultaneity (the changes are concurrent). Many cases of apparent coevolution, Janzen argued, are better described as sequential adaptation, where one species adapts to another without reciprocal change. This clarification forced coevolutionary biologists to provide more rigorous evidence for reciprocal evolutionary change.

John Thompson's geographic mosaic theory of coevolution, developed over several decades and synthesized in his 2005 book, represents the most significant recent advance. Thompson recognized that coevolutionary dynamics vary across the geographic range of interacting species, creating a spatial mosaic of hotspots (strong reciprocal selection) and coldspots (weak or absent selection). Gene flow between populations, local extinction and recolonization, and geographic variation in community composition continually remix coevolved traits, preventing any single coevolutionary outcome from dominating the entire species pair. This theory resolved the paradox that local populations often appear maladapted (because gene flow brings in traits selected under different conditions) while the overall species interaction remains stable. The geographic mosaic perspective has become the dominant framework for empirical coevolutionary research.

The philosophical implications of coevolution are significant. Coevolution demonstrates that adaptation is relational rather than absolute: a trait's fitness value depends on the traits of other species, which are themselves evolving. This means that evolution has no fixed endpoint and no single optimal solution. The metaphor of the "evolutionary arms race" captures this dynamic, but coevolution also produces cooperation, mutualism, and stable equilibria. The interplay between antagonistic and mutualistic coevolutionary forces -- and the conditions that favor each -- remains a central question in evolutionary biology. The tension between the "selfish gene" perspective (organisms are vehicles for gene replication, and cooperation is always ultimately self-serving) and the mutualistic perspective (genuine cooperation between species can be evolutionarily stable) reflects deeper philosophical questions about the nature of biological organization and the levels at which natural selection operates.

The practical importance of coevolution extends to agriculture, medicine, and conservation. Agricultural pests evolve resistance to pesticides through the same coevolutionary dynamics as herbivores adapting to plant defenses. Pathogens evolve resistance to antibiotics through arms races with the human immune system and medical interventions. Understanding these dynamics -- and predicting when and how resistance will evolve -- requires the coevolutionary framework developed over the past six decades. The geographic mosaic theory, in particular, has practical implications for managing resistance: spatial variation in selection pressures, and gene flow between treated and untreated areas, can slow or accelerate the evolution of resistance in agricultural pests and human pathogens. As anthropogenic pressures reshape species interactions worldwide, the coevolutionary framework provides essential tools for predicting and mitigating the consequences.

Bibliography Master

Campbell, N. A. & Reece, J. B. Biology, 12th ed. (Pearson, 2020). Ch. 54.
Thompson, J. N. The Geographic Mosaic of Coevolution (University of Chicago Press, 2005).
Ehrlich, P. R. & Raven, P. H. "Butterflies and plants: a study in coevolution." Evolution 18 (1964) 586-608.
Van Valen, L. "A new evolutionary law." Evol. Theory 1 (1973) 1-30.
Janzen, D. H. "When is it coevolution?" Evolution 34 (1980) 611-612.
Flor, H. H. "The complementary genic systems in flax and flax rust." Adv. Genet. 8 (1956) 29-54.
Vermeij, G. J. Evolution and Escalation: An Ecological History of Life (Princeton University Press, 1987).
Bull, J. J. "Virulence." Evolution 48 (1994) 1423-1437.
Cook, J. M. & Rasplus, J.-Y. "Mutualists with attitude: coevolving fig wasps and figs." Trends Ecol. Evol. 18 (2003) 241-248.
Lively, C. M. "Parasite virulence, host life history, and patterns of infection and reinfection in a natural population of the freshwater snail Potamopyrgus antipodarum." Oecologia 162 (2010) 101-111.
Vermeij, G. J. "The escalation hypothesis." In Key Topics in Paleobiology (Cambridge University Press, 2002).
Bergstrom, C. T. & Lachmann, M. "The Red King effect: when the slowest runner wins the coevolutionary race." Proc. Natl. Acad. Sci. 100 (2003) 593-598.
Sachs, J. L. & Simms, E. L. "Pathways to mutualism breakdown." Trends Ecol. Evol. 21 (2006) 585-592.
Lively, C. M. "Evidence from a New Zealand snail for the maintenance of sex by parasitism." Nature 328 (1987) 519-521.
Pellmyr, O. "Yuccas, yucca moths, and coevolution: a review." Ann. Missouri Bot. Gard. 90 (2003) 35-55.
Mode, C. J. "A mathematical model for the co-evolution of obligate parasites and their hosts." Evolution 12 (1958) 158-165.
Estes, J. A. et al. "Trophic downgrading of planet Earth." Science 333 (2011) 301-306.
Bronstein, J. L. "Our current understanding of mutualism." Q. Rev. Biol. 69 (1994) 31-51.
Schmid-Hempel, P. Evolutionary Parasitology: The Integrated Study of Infections, Immunology, Ecology, and Genetics (Oxford University Press, 2011).

Exercise 3

Exercise 3 (hard, analysis).

The fig-fig wasp mutualism is often cited as one of the most tightly coevolved interactions in nature, with approximately 750 fig species each pollinated by a specific wasp species. Explain how the fig enforces cooperation through "sanctions" and why this mechanism is more effective at maintaining mutualism than "partner fidelity feedback" alone. What happens to the mutualism when sanctions are experimentally removed?

Hint

Consider: what prevents the wasp from cheating (laying eggs without pollinating)? If the fig cannot detect cheating, what maintains the mutualism? Sanctions are a punishment mechanism; partner fidelity feedback is a shared-fate mechanism.

Answer

The fig sanctions mechanism works as follows: when a female wasp enters a fig syconium, she pollinates the internal flowers and lays eggs in some of them. The fig can detect whether pollination has occurred (through unknown chemical or mechanical cues). If a wasp enters but fails to pollinate (a "cheater" wasp), the fig aborts the developing seeds in that syconium. Because the wasp larvae develop inside the fig's ovules, aborting the seeds also kills the wasp's offspring. The fig effectively punishes non-pollinating wasps by terminating their reproductive success.

This sanction mechanism is more effective than partner fidelity feedback (the idea that cooperation is maintained because both partners' fitness is linked through repeated interactions) for several reasons:

Immediate detection and punishment. Sanctions detect cheating within a single interaction and punish it immediately, while partner fidelity feedback requires multiple interactions over time to detect and respond to cheating.
No memory or repeated interaction required. Partner fidelity feedback requires that the same individuals interact repeatedly, allowing the detection of cheating over time. In the fig-wasp system, each wasp interacts with only one fig in her lifetime (she dies after entering), so there is no opportunity for repeated interaction. Sanctions solve this problem by enforcing cooperation in a single interaction.
Cost-effective enforcement. The fig's sanction (aborting one syconium) is relatively inexpensive compared to the cost of being exploited by a cheater. The fig can afford to sacrifice one syconium to enforce cooperation across all syconia.

When sanctions are experimentally removed (by preventing pollination while allowing oviposition), the results confirm the theory: wasps that do not pollinate still successfully reproduce, and their offspring are more likely to carry non-pollinating genotypes. Over time, without sanctions, the mutualism breaks down as cheater wasps increase in frequency. This demonstrates that the mutualism is actively maintained by the fig's enforcement mechanism rather than being an evolutionarily stable strategy maintained solely by the benefits of cooperation.

Exercise 4

Exercise 4 (medium, short answer).

Explain the concept of "evolutionary escalation" as proposed by Geerat Vermeij. How does it differ from a simple predator-prey arms race, and what evidence supports it?

Hint

Escalation is a long-term, directional trend: defenses and weapons become more elaborate over geological time. It is not just about two species, but about the overall level of predation pressure in the ecosystem.

Answer

Vermeij's escalation hypothesis (1987) proposes that the long-term history of life has been characterized by a directional increase in the frequency and intensity of predation, driving corresponding increases in the elaborateness of defensive adaptations. Unlike a simple two-species arms race, escalation operates at the community level: as predators become more effective overall, all prey species face increased selection for defenses, and as defenses become more common, all predator species face increased selection for counter-adaptations. The result is a community-wide trend toward more sophisticated weapons and defenses over geological time.

Evidence for escalation comes from the marine fossil record:

Drilling predation. The frequency of drill holes in bivalve shells (made by predatory gastropods) has increased steadily from the Mesozoic to the present, indicating that drilling predation has become a more important selective force over time.
Shell architecture. The proportion of gastropod species with heavily armored shells (thickened, spiny, or narrowly aperture) has increased from the Paleozoic through the Cenozoic, while thin-shelled, open-aperture forms have declined.
Repair scars. The frequency of repaired shell damage (indicating failed predation attempts) has increased through time, showing that predators are attacking more frequently and that prey are surviving attacks through improved defenses.
Crushing versus swallowing. Mesozoic marine reptiles (mosasaurs, plesiosaurs) evolved increasingly powerful crushing bites capable of breaking thick-shelled prey, while their prey evolved progressively thicker and more ornamented shells.

Escalation differs from a simple arms race in that it is a community-level, multi-species phenomenon that proceeds directionally over geological time, rather than a pairwise oscillation between two species. It predicts that the overall "investment" in offense and defense should increase through time, which is supported by the fossil evidence.

Exercise 5

Exercise 5 (hard, analysis).

The Red Queen hypothesis predicts that sexual reproduction should be more common in populations with high parasite pressure. Describe the design and results of the New Zealand freshwater snail (Potamopyrgus antipodarum) study that tested this prediction. What are the strengths and limitations of this natural experiment?

Hint

P. antipodarum has both sexual and asexual individuals in the same lakes. If Red Queen is correct, sexual individuals should be more common where parasites are more prevalent. Consider also the potential confounding variables.

Answer

The New Zealand freshwater snail Potamopyrgus antipodarum is unusual in that both sexual and asexual individuals coexist within the same populations. Asexual females produce clonal daughters, while sexual females produce genetically diverse offspring through cross-fertilization with males. This makes the species a natural laboratory for testing the Red Queen hypothesis.

Study design: Lively (1992) and subsequent researchers sampled P. antipodarum populations from numerous lakes and streams across New Zealand, recording: (a) the proportion of sexual versus asexual individuals in each population, and (b) the prevalence of infection by trematode parasites (particularly Microphallus), which castrate the snail and are transmitted trophically (when the infected snail is eaten by a waterfowl, the parasite's definitive host).

Results: The proportion of sexual individuals was significantly positively correlated with parasite prevalence across populations. Lakes with high Microphallus infection rates had a higher frequency of sexual snails, while lakes with low parasite prevalence were dominated by asexual clones. Furthermore, the most common asexual clone in each lake was disproportionately infected by parasites, indicating that parasites had adapted to infect the most common host genotype -- exactly as predicted by negative frequency-dependent selection. Temporal studies showed that asexual clones that were common and uninfected in one year became common and heavily infected in subsequent years, while previously rare clones increased in frequency, confirming the dynamic cycling predicted by Red Queen dynamics.

Strengths:

The coexistence of sexual and asexual individuals in the same habitat controls for environmental variables that might otherwise confound the comparison.
The trematode parasite is a severe selective force (it castrates the host), providing strong coevolutionary pressure.
The geographic replication across many lakes provides statistical power and generality.

Limitations:

Correlation does not establish causation: other factors that covary with parasite prevalence (lake depth, productivity, waterfowl abundance) could be responsible for the pattern.
The "microcosm" nature of lake populations may not generalize to other systems or to organisms without mixed reproductive modes.
The timescale of observation (years to decades) may not capture the full coevolutionary dynamics, which can operate over longer timescales.

Despite these limitations, the P. antipodarum system remains one of the strongest pieces of empirical support for the Red Queen hypothesis and has become a model system for studying host-parasite coevolution in nature.

Exercise 6

Exercise 6 (medium, short answer).

Myxoma virus was introduced to Australia in 1950 to control the European rabbit population. Describe the coevolutionary trajectory of the virus and rabbit over the subsequent decades. Why did the virus evolve reduced virulence, and what does this tell us about the conditions that favor virulence versus avirulence?

Hint

Virulence is the harm done to the host. If a virus kills its host too quickly, it may not have time to transmit. What is the optimal virulence level from the virus's perspective?

Answer

The myxoma virus-rabbit coevolution in Australia is one of the best-documented cases of coevolution in real time. When the virus was first introduced in 1950, it had a case fatality rate of approximately 99.8%, killing rabbits within 10-12 days. The rabbit population crashed by over 95% within two years.

Viral evolution (reduced virulence): Over the next decade, the virus evolved reduced virulence. By the 1960s, the dominant viral strains had case fatality rates of 70-90%, killing rabbits in 20-30 days. By the 1980s, the most common strains had fatality rates of 50-70%. This evolution occurred because highly virulent strains killed their hosts so quickly that the hosts died before infectious vectors (mosquitoes) could bite them and transmit the virus. Moderately virulent strains, which allowed the rabbit to survive longer, had higher transmission rates and outcompeted the most lethal strains. This is a classic example of the trade-off model of virulence evolution: virulence is constrained by the need for transmission.

Rabbit evolution (increased resistance): Simultaneously, the rabbit population evolved increased genetic resistance to the virus. When exposed to the same viral strain, rabbits from later generations survived at higher rates than rabbits from the initial population. This resistance evolved through selection on the rabbit immune system and physiological tolerance mechanisms.

The coevolutionary equilibrium: The system has reached a dynamic equilibrium where both moderate virulence (virus) and moderate resistance (rabbit) are maintained. The virus cannot evolve to complete avirulence because some level of virulence is necessary for efficient transmission (the virus needs to produce sufficient viral particles in the rabbit's skin for mosquitoes to acquire it). The rabbit cannot evolve to complete resistance because resistance carries costs (immune system maintenance, reduced reproductive output).

This case demonstrates that virulence is not inevitably maximized but evolves to an optimum determined by the transmission-virulence trade-off. When transmission requires host survival (as in vector-borne diseases), moderate virulence is evolutionarily stable. When transmission does not require host survival (as in waterborne diseases or transmission by predation), higher virulence can evolve.

Exercise 7

Exercise 7 (hard, analysis).

Plant-pollinator networks are typically "nested": specialist plants are pollinated by generalist pollinators, while specialist pollinators visit generalist plants. Explain how coevolutionary processes generate this nested structure, and discuss its implications for community stability and vulnerability to species loss.

Hint

Specialists evolve to exploit the most reliable partners (which tend to be generalists). Generalists interact broadly. What happens to the network if a keystone generalist is lost versus a specialist?

Answer

Nested structure in plant-pollinator networks arises through several coevolutionary and ecological mechanisms:

Abundance-driven assembly. The most abundant plants and pollinators are likely to interact first and most frequently. Generalist pollinators, by definition, visit many plant species, and the most common plants are visited most often. Specialist plants, which have evolved to attract a narrow range of pollinators, benefit most from reliable, abundant generalist pollinators rather than rare specialists that may not be present when the plant is flowering.
Phenological reliability. Generalist pollinators are active across a longer flowering season, making them more reliable partners for plants with narrow flowering windows. Specialist plants that coevolve to depend on a single pollinator species risk reproductive failure if that pollinator's phenology does not match the plant's flowering time. By contrast, depending on a suite of generalist pollinators provides insurance against phenological mismatch.
Coevolutionary escalation toward specialization. Plants that specialize their floral morphology (long corolla tubes, specific color patterns, particular nectar compositions) exclude less effective pollinators and attract the most efficient ones. The most efficient pollinators tend to be generalists that have coevolved the ability to handle a wide range of floral morphologies. This produces the nested pattern: highly specialized plants interacting with highly generalized pollinators.

Implications for stability: Nested networks are more resistant to random species loss because the removal of a specialist species has minimal impact on the rest of the network (the specialist's generalist partners still have many other partners). However, nested networks are vulnerable to the loss of highly connected generalist species (keystone species), which support many specialists. If a keystone generalist pollinator is lost, many specialist plants that depend exclusively on it may lose their sole pollinator and face reproductive collapse.

Implications for conservation: The nested structure means that conservation priorities should focus on protecting highly connected generalist species, whose loss would propagate through the network. Climate change and habitat fragmentation may disproportionately affect specialist species, potentially eroding the nested structure over time and reducing network stability. Understanding coevolutionary network structure is therefore essential for predicting the community-level consequences of biodiversity loss.

Prerequisites

19.08.01 pending
19.10.01 pending

Tier anchors

beginner: Campbell Biology, 12th ed. (2020), Ch. 54; Khan Academy Biology
intermediate: Futuyma & Kirkpatrick — Evolution, 4th ed. (2017); Thompson — Interaction and Coevolution, 1982
master: Thompson, J. N. — The Geographic Mosaic of Coevolution (2005); relevant primary literature

References

Campbell Biology, 12th ed. (Pearson, 2020) · Ch. 54 Community Ecology
Thompson, J. N. — The Geographic Mosaic of Coevolution (University of Chicago Press, 2005) · Ch. 1-10
Ehrlich, P. R. & Raven, P. H. — Butterflies and plants: a study in coevolution, Evolution 18 (1964) 586-608 · Foundational coevolution paper
Van Valen, L. — A new evolutionary law, Evol. Theory 1 (1973) 1-30 · Red Queen hypothesis
Janzén, D. H. — When is it coevolution? Evolution 34 (1980) 611-612 · Clarification of coevolutionary criteria

Estimated time

beginner: 15m
intermediate: 35m
master: 55m