“Devolutionary” sounds like it should describe evolution running in reverse, organisms sliding backward toward some simpler state. Biologists almost universally reject the term, because evolution has no forward or backward, but the phenomena people reach for it to describe are real and fascinating. Organisms routinely lose complex traits: eyes shrink, limbs vanish, entire metabolic pathways shut down. These losses are not mistakes or decay. They are the product of the same evolutionary pressures that build complexity in the first place, and understanding how they work reveals something deeper about how life changes over time.
Why “Devolution” Is Not a Real Biological Concept
The idea that organisms can devolve rests on a ladder model of life, where bacteria sit at the bottom, humans perch at the top, and evolution is a long climb upward. This picture was popular in the nineteenth century, and it lingers in pop culture, but it does not reflect how evolution actually works. Natural selection has no target. It favors whatever helps an organism survive and reproduce in its current environment, even if that means discarding something elaborate. A blind cave fish is not less evolved than a sighted river fish; it is differently adapted.
The word “devolutionary” does show up occasionally in academic writing, but almost always in cultural or literary contexts rather than biological ones. A University of Washington study, for instance, used “the devolutionary imagination” to describe a theme in postwar British speculative fiction, where collapse and regression became narrative tools for exploring the end of empire. In biology, though, the preferred terms are “regressive evolution,” “trait loss,” or “reductive evolution,” all of which describe the phenomenon without implying that evolution has a preferred direction.
Relaxed Selection and the Drift Toward Loss
One of the most straightforward ways organisms lose traits is through relaxed selection. When a structure stops being useful, the genes that build it are no longer held in shape by natural selection. Mutations accumulate in those genes over generations, and the trait gradually degrades. This has been studied in detail in bats. Echolocating bat species that navigate using sound rather than sight show clear signs of relaxed selection on their color-vision genes. One study found that the gene for medium- and long-wavelength vision was under significantly weaker selection in all echolocating bats compared to their non-echolocating relatives, even though the gene had not yet broken down entirely.1PubMed Central. RELAX: Detecting Relaxed Selection in a Phylogenetic Framework Meanwhile, a separate short-wavelength vision gene had already become a pseudogene (a broken, non-functional copy) in cave-roosting and echolocating lineages, and selection on that gene had essentially ceased.2PubMed Central. RELAX: Detecting Relaxed Selection in a Phylogenetic Framework
Relaxed selection is a slow process. The trait does not vanish overnight. Instead, the gene accumulates neutral or mildly harmful mutations over thousands of generations, gradually losing its ability to produce a functional product. This pattern shows up repeatedly across the tree of life: genes that stop mattering stop being maintained.
Cavefish and the Genetics of Going Blind
The Mexican cavefish, Astyanax mexicanus, is the most intensely studied example of regressive evolution. Populations of this species live in dark limestone caves in northeastern Mexico, and over roughly the last few million years, cave-dwelling populations have independently lost their eyes and much of their body pigmentation. Surface-dwelling populations of the same species still have perfectly functional eyes. This makes Astyanax a natural experiment: researchers can compare two versions of the same fish and ask what changed.
Genome sequencing has identified a long list of candidate genes involved in eye loss. Several genes known to be critical for eye and lens development in other fish, including ones involved in lens formation, retinal patterning, and eye-cup shape, show significantly reduced expression in cavefish embryos compared to surface fish. In many cases the gene’s protein-coding sequence is nearly identical between the two forms; the difference lies in how much the gene is turned on during development.3Nature Communications. The cavefish genome reveals candidate genes for eye loss This points to regulatory changes, alterations in the switches that control gene activity, rather than destruction of the genes themselves.
Epigenetics adds another layer. Research has shown that excess DNA methylation, a chemical modification that silences genes, promotes eye degeneration in cave populations. The most heavily methylated and downregulated genes in cave morphs turned out to overlap with genes linked to human eye disorders, suggesting that the developmental toolkit for building eyes is deeply conserved across vertebrates.4PubMed Central. An epigenetic mechanism for cavefish eye degeneration The eyes are not simply rusting away from disuse. Multiple molecular mechanisms are actively reshaping the developmental program.
Eyes Lost, Senses Gained
The cavefish story gets more interesting when you ask why eyes were lost rather than simply left to degrade. One influential hypothesis involves antagonistic pleiotropy, the idea that genes influencing eye development also affect other traits, so that selection for a beneficial change in one trait can drive a harmful change in another. Cavefish exhibit a behavior called vibration attraction, where they swim toward water disturbances that might signal food. Surface fish do not do this. Genetic mapping revealed that some of the same genomic regions containing genes for eye size also harbor genes for this vibration-sensing behavior.5PubMed Central. Eye regression in blind Astyanax cavefish may facilitate the evolution of an adaptive behavior and its sensory receptors
If those genetic regions work as a package deal, where boosting vibration attraction means shrinking eyes, then natural selection in a dark cave would favor the trade-off. The eyes are not just drifting away neutrally; they may be actively being selected against because losing them comes bundled with a sensory upgrade that matters in total darkness. This reframes “devolution” as something more nuanced: what looks like degradation from one angle is optimization from another.
How Snakes Lost Their Legs
Limb loss in snakes is another dramatic example of regressive evolution, and the genetic story is surprisingly precise. Researchers identified a long-range regulatory element called the ZRS that controls the activity of Sonic hedgehog, a gene essential for limb development across vertebrates. In snakes, this enhancer has accumulated mutations that cripple its function.6PubMed Central. Progressive Loss of Function in a Limb Enhancer During Snake Evolution
The evolutionary trajectory is visible across snake species. Basal snakes like pythons and boas, which still retain a vestigial pelvic girdle and rudimentary hindlimbs, have a ZRS enhancer that shares about 80% identity with the same region in limbed lizards. Advanced snakes like vipers, rattlesnakes, and cobras, which have lost all skeletal limb structures, show a dramatically accelerated rate of mutations in the ZRS that distinguishes it from other limb-related enhancers.7Cell. Progressive Loss of Function in a Limb Enhancer during Snake Evolution When researchers replaced the mouse version of this enhancer with the python version, the resulting mice developed severely reduced limbs. Restoring just a single transcription factor binding site that had been lost in the snake lineage was enough to rescue full enhancer function.8PubMed Central. Progressive Loss of Function in a Limb Enhancer During Snake Evolution A tiny change in a regulatory switch, not the destruction of a whole gene, was sufficient to eliminate an entire body plan feature.
Limb reduction has happened independently dozens of times across squamate reptiles, including in lizard lineages that have nothing to do with snakes. This parallel pattern tells biologists that losing limbs is not a rare catastrophe but a recurring evolutionary solution when body plans shift toward burrowing or serpentine locomotion.9PubMed. A farewell to arms and legs: a review of limb reduction in squamates
Island Birds on the Path Toward Flightlessness
Flight is metabolically expensive. It demands large breast muscles, lightweight bones, and a body plan oriented around getting airborne. On islands with few predators, that investment becomes less critical. A study of 366 bird populations across Caribbean and Pacific islands found that birds on smaller, more predator-poor islands had consistently smaller flight muscles and longer legs compared to their mainland relatives. These shifts in body proportion mirrored the anatomical changes seen in fully flightless species, suggesting that island populations drift along a trajectory toward flightlessness even if most never get all the way there.10PubMed Central. Predictable evolution toward flightlessness in volant island birds
This is a case where the environment reshapes the cost-benefit calculation. When predators are around, flight is non-negotiable, so selection fiercely maintains the machinery for it. Remove the predators, and maintaining flight muscles becomes a caloric luxury. Resources get redirected toward legs, foraging, and reproduction. The rails, a family of birds notorious for losing flight on islands, are the poster children for this pattern, but the underlying shift shows up across unrelated bird groups exposed to the same ecological conditions.
Parasites That Gave Up Photosynthesis
Plants that become parasites undergo some of the most extreme trait loss in nature. Parasitic flowering plants that steal nutrients from host plants no longer need to photosynthesize, and their chloroplast genomes reflect this. The chloroplast, the organelle responsible for photosynthesis, carries its own small genome, and in parasitic lineages this genome shrinks dramatically as photosynthesis genes are shed.11PubMed. Photosynthetic evolution in parasitic plants: insight from the chloroplast genome
Rafflesia, the genus that produces the world’s largest flowers, may represent the extreme end of this process. Researchers searching for an intact chloroplast genome in Rafflesia lagascae found none. Instead, they recovered only small, broken fragments of what were once plastid genes, with no intact reading frames remaining.12Molecular Biology and Evolution. Possible Loss of the Chloroplast Genome in the Parasitic Flowering Plant Rafflesia lagascae (Rafflesiaceae) Another holoparasitic plant, Aeginetia indica, retains a plastome reduced to about 86,000 base pairs with almost all photosynthesis-related genes gone. Fragments of those lost genes were found transferred into the plant’s mitochondrial and nuclear genomes, but transcriptome analysis showed they were non-functional, just genetic debris.13PubMed Central. The loss of photosynthesis pathway and genomic locations of the lost plastid genes in a holoparasitic plant Aeginetia indica
A similar pattern plays out in bacterial endosymbionts, microbes that live permanently inside insect cells and provide essential nutrients in exchange for shelter. Their genomes can shrink to a fraction of the size of their free-living relatives, shedding genes that the host cell’s own machinery can supply.14PubMed Central. How do bacterial endosymbionts work with so few genes? Studies of early gene loss in these symbionts found that the genes shed first were, on average, less important to the organism’s fitness, which is what you would expect if natural selection was loosening its grip on genes that no longer earned their keep.15PubMed Central. Tempo and mode of early gene loss in endosymbiotic bacteria from insects
Can Lost Traits Come Back?
A principle called Dollo’s law, named after the nineteenth-century paleontologist Louis Dollo, holds that once a complex trait is lost in evolution, it cannot re-evolve in the same form. The logic is intuitive: after millions of years of mutational decay, the genetic instructions for building the trait should be too degraded to reassemble. But biology keeps turning up apparent exceptions.
The most striking case involves teeth in frogs. The ancestor of all modern frogs lost mandibular (lower jaw) teeth at least 230 million years ago. Yet in the genus Gastrotheca, a group of marsupial frogs, lower jaw teeth have reappeared within the last roughly 5 to 17 million years.16PubMed. Re-evolution of lost mandibular teeth in frogs after more than 200 million years, and re-evaluating Dollo’s law Detailed morphological analysis confirmed these are true teeth, with pedicels, bicuspid crowns, and a successional dental lamina indicating normal tooth replacement, not some rough approximation.17Evolution. Re-evaluating the morphological evidence for the re-evolution of lost mandibular teeth in frogs
How is this possible? The key seems to be that frogs never lost upper jaw teeth. Because the developmental pathway for making teeth remained under active selection in the upper jaw, all the genes, signaling networks, and cellular machinery stayed functional. In the right circumstances, those pathways could be reactivated in the lower jaw, where they had been dormant for hundreds of millions of years.18PubMed Central. Challenging and redefining Dollo’s law of evolution: re-appearance of lost structures A broad review of trait reversals across the tree of life found that this pattern holds generally: reversal is possible when some form of the lost trait’s developmental pathway persists elsewhere in the organism.19PubMed Central. Loss, persistence and reversal of phenotypic traits
Dollo’s law is not exactly wrong, but it needs refining. If every component of the developmental program for a trait has decayed, re-evolution is effectively impossible. But if related structures keep the genetic toolkit alive, the door stays open, sometimes for astonishingly long stretches of time. Research on felid dentition has suggested that silenced regulatory elements and individual protein-interaction motifs can be reactivated after periods far longer than an entire protein-coding gene could survive intact without selection.20bioRxiv. Return of a lost structure in the evolution of felid dentition revisited: A DevoEvo perspective on the irreversibility of evolution Some researchers have cautioned that phylogenetic methods used to detect trait reversals can produce false positives, so not every claimed reversal may hold up to scrutiny.21Evolution. On Phylogenetic Tests of Irreversible Evolution Still, the frog-teeth example, with direct morphological confirmation, stands as robust evidence that Dollo’s law has limits.
Smaller Brains Under Domestication
Domesticated animals consistently have smaller brains than their wild ancestors, a pattern that has been documented for over a century across species from dogs and pigs to sheep and cattle. A review of the evidence concluded that while the exact magnitudes of reduction vary and older measurements contain some error, the overall pattern of brain-size reduction in domestic forms is well supported.22PubMed Central. The mammalian brain under domestication: Discovering patterns after a century of old and new analyses
In cattle, researchers used fossil data to compare the brains of domestic Bos taurus with their wild ancestor, the aurochs (Bos primigenius), and found not only that domestic cattle have smaller brains overall, but that brain size varies significantly by breed, with some breeds having much smaller brains than others.23PubMed Central. Intensive human contact correlates with smaller brains: differential brain size reduction in cattle types Breeds that have more intensive daily contact with humans tend to show greater reduction, raising the possibility that tameness itself, the selection for docility, pulls brain size down as a correlated effect. This echoes the cavefish story: selection for one trait can drag along changes in another.
When Humans Lost the Ability to Make Vitamin C
Humans, along with other primates, guinea pigs, and some bats, cannot synthesize vitamin C. Most mammals produce it internally, but in our lineage the gene for the final enzyme in the synthesis pathway (L-gulonolactone oxidase) became a pseudogene tens of millions of years ago. On its face, losing the ability to make an essential nutrient looks like a textbook example of harmful degeneration.
But the story may be more nuanced. One hypothesis focuses on a glucose transporter called Glut-1 that appeared on primate red blood cells around the same time the synthesis gene broke. Glut-1 is far more efficient at recycling vitamin C from the blood than the transporter it replaced. The result is that species that lost the ability to synthesize vitamin C need dramatically less of it, roughly 100-fold less per kilogram of body weight per day than species that still make it internally.24PubMed Central. Glut-1 explains the evolutionary advantage of the loss of endogenous vitamin C-synthesis: The electron transfer hypothesis If efficient recycling was already in place, the metabolic cost of running a whole synthesis pathway became unnecessary. Losing it was not degeneration; it was streamlining.
Trait Loss and Small-Population Risk
Not all trait loss is adaptive. In very small, isolated populations, the machinery of natural selection weakens, and random genetic drift gains the upper hand. Mildly harmful mutations that would be weeded out in a large population can accumulate when the population is too small for selection to act effectively.25Biological Conservation. Genetics and extinction This is not the orderly, environment-driven trait loss seen in cavefish or snakes. It is genetic erosion, and it can push species toward extinction.
Simulations have shown that populations with a history of being large are especially vulnerable when they suddenly shrink. Large populations accumulate recessive strongly harmful mutations that remain hidden because each individual carries only one copy. When inbreeding in a small population exposes these mutations in double doses, fitness can crash rapidly.26PubMed Central. Strongly deleterious mutations are a primary determinant of extinction risk due to inbreeding depression For conservation biologists, understanding this kind of non-adaptive “devolutionary” process, the degradation of functional traits through drift rather than selection, is critical for managing endangered species. The distinction between adaptive trait loss and drift-driven decay matters enormously when deciding how to intervene.
Epigenetics and the Speed of Change
Classical genetics focuses on changes to the DNA sequence itself, but epigenetic modifications, chemical tags on DNA or its associated proteins that alter gene expression without changing the underlying code, can also drive trait changes across generations. The cavefish eye-loss story already showed DNA methylation at work, but the broader implications reach further. Environmentally induced epigenetic changes have been observed to persist across multiple generations, meaning organisms can sometimes acquire and transmit altered traits faster than traditional mutation-and-selection would allow.27PubMed Central. Generational stability of epigenetic transgenerational inheritance facilitates adaptation and evolution
For trait loss specifically, this raises an intriguing possibility. If an environmental shift makes a trait unnecessary, epigenetic silencing could begin dampening the relevant genes within a few generations, well before mutational decay would degrade them. The trait might appear to vanish quickly, with the genetic sequence remaining intact underneath. Whether this kind of rapid, epigenetically driven trait suppression can become permanent over evolutionary time, essentially paving the way for later genetic decay, is an active area of research. It adds a wrinkle to the standard relaxed-selection model and suggests that what looks like abrupt “devolution” in the fossil record might sometimes reflect epigenetic rather than genetic change.
Enhancer Loss as an Evolutionary Tool
Many of the examples discussed here involve changes not to genes themselves but to their regulatory elements, the enhancers and promoters that determine when, where, and how much a gene is expressed. Enhancers are modular: an organism can lose one enhancer and thereby shut off a gene in one tissue while leaving its expression elsewhere untouched. This modularity makes enhancer loss a surprisingly clean evolutionary mechanism. A gene critical for eye development can be silenced in the eye without disrupting its role in the brain, for instance.28PubMed Central. Enhancer modularity and the evolution of new traits
The snake limb story is a vivid illustration: the Sonic hedgehog gene did not disappear; its limb-specific enhancer degraded. The gene presumably continues doing its many other jobs in snake development. Enhancer-level changes may explain why so many “devolutionary” shifts seem precise rather than catastrophic. Organisms do not lose entire genetic programs so much as selectively disconnect specific outputs, like unplugging one appliance from a circuit rather than cutting the power to the house.