“Fish” is not a genuine biological group. In the framework that modern biologists use to classify living things, a valid group must include an ancestor and all of its descendants. “Fish” fails that test spectacularly: a lungfish is more closely related to a cow than it is to a shark, and a salmon shares a more recent common ancestor with you than it does with a lamprey. The word “fish” describes a grab bag of aquatic vertebrates united mainly by living in water and having fins, not by shared evolutionary heritage. That distinction sounds pedantic until you look at the evidence, which reveals just how misleading the category is.
What Counts as a Real Group
Biologists since the mid-twentieth century have increasingly insisted that the only meaningful way to classify organisms is by their evolutionary branching history. A group is considered natural if it contains one ancestor and every single lineage that descended from that ancestor. Mammals qualify: you can point to the ancestor of all mammals, and every descendant of that ancestor is still called a mammal. Birds qualify for the same reason. “Fish” does not, because the ancestor of everything we casually call a fish also gave rise to amphibians, reptiles, birds, and mammals. To make “fish” a valid group, you would have to include all land vertebrates inside it. At that point, the word stops meaning what anyone thinks it means.
The technical term for a group that excludes some of its descendants is “paraphyletic,” and the history of taxonomy is littered with debates about whether such groups should still be formally recognized. Some researchers have argued that paraphyletic groups can still be useful shorthand, but the dominant view in systematics is that only groups containing all descendants of a common ancestor belong in a formal classification.
The Family Tree That Breaks the Category
The problem becomes vivid when you look at how the major groups of “fish” actually relate to each other. Start at the base of the vertebrate tree. Lampreys and hagfish split off first. Then cartilaginous fish like sharks and rays branch away. After that, the ray-finned fish (think trout, tuna, goldfish) and the lobe-finned fish (lungfish and coelacanths) diverge. But here is the catch: land vertebrates, including frogs, lizards, and people, sit nested inside the lobe-finned branch. The lobe-finned fish did not produce a separate lineage of land animals and then go extinct as a side note. They are our closest “fish” relatives, and genomic analysis of the coelacanth genome confirmed that lungfish, not coelacanths, are the closest living relatives of all four-limbed vertebrates.
1PubMed Central. The African coelacanth genome provides insights into tetrapod evolutionFossil evidence tells the same story. A 410-million-year-old sarcopterygian fish from China, described as bridging the gap between early lobe-finned fish and the lineages that would eventually produce both lungfish and the first four-limbed animals, shows that these transitions were gradual and deeply interconnected.2Nature. A primitive fish close to the common ancestor of tetrapods and lungfish So when we say “fish,” we are drawing a line around some members of this tree and leaving out others, purely because the ones we left out happen to walk on land. That is like defining a group called “non-elephant mammals” and pretending it is a natural category.
The Blurry Border Between Water and Land
If “fish” were a real group, you would expect a clear boundary between fish-like creatures and land-dwelling ones. No such boundary exists. The fossil record is full of animals that blur the line so thoroughly that calling them “fish” or “not fish” becomes arbitrary.
The most famous of these intermediates is Tiktaalik roseae, a roughly 375-million-year-old animal discovered in the Canadian Arctic. Tiktaalik had fins with wrist-like joints, a flattened head, and a neck, features associated with life at least partly out of water. But it also retained a feeding system that mixed aquatic suction-based prey capture with a biting mechanism more like what land animals use. Detailed study of its skull showed that Tiktaalik could still expand its cheek and palate sideways, a motion used during suction feeding, even though its skull was already restructured to support biting and snapping. The researchers concluded that this limited skull flexibility was probably inherited by the earliest limbed vertebrates, meaning the “fish” feeding style did not simply switch off when animals started walking.3PubMed Central. The feeding system of Tiktaalik roseae: an intermediate between suction feeding and biting
Animals like Tiktaalik are exactly what the evolutionary framework predicts: gradients of change, not clean breaks. There is no moment where “fish” ends and “land animal” begins, which is itself an argument against treating “fish” as a distinct natural category.
Lungs Came Before Land
One reason people assume fish and land animals are fundamentally different is breathing. Fish have gills; we have lungs. But this intuition is historically backward. Evidence strongly suggests that air-breathing organs evolved in the ancestors of bony fish while those animals were still fully aquatic. The primitive organ was probably a modification of the posterior gill pouches, served by the sixth gill artery. Over time, the upper part of this structure may have developed into the swim bladder that many modern fish use for buoyancy, while paired lower extensions evolved into true lungs.4Comparative Biochemistry and Physiology Part A: Molecular & Integrative Physiology. Which came first, the lung or the breath?
This means lungs are not an invention of land animals. They are an ancient feature of bony fish that most ray-finned fish later repurposed or lost. A goldfish’s swim bladder and your lungs share an evolutionary origin. The divide we imagine between “gilled water creature” and “lunged land creature” collapses once you know that breathing air started underwater.
Fingers Were Foreshadowed in Fins
Hands and fingers feel quintessentially land-animal. But the genetic program that builds digits did not spring into existence the moment vertebrates crawled ashore. Researchers studying the Australian lungfish found that a gene called hoxa13, which marks the hand-and-finger-building region during limb development in land vertebrates, is already active during fin development in lungfish. The developmental pattern suggests that the digit program originated in the back edge of the fin and later expanded forward as limbs evolved.5PubMed Central. Sarcopterygian fin ontogeny elucidates the origin of hands with digits
The roots go even deeper. A study of Hoxd gene expression in shark paired fins found a second, distal phase of expression that had previously been thought to be unique to limbs with digits. The researchers proposed that this late burst of gene activity was already present in the common ancestor of sharks and bony fish, meaning the genetic groundwork for fingers existed hundreds of millions of years before any animal had a hand.6PLOS ONE. Biphasic Hoxd Gene Expression in Shark Paired Fins Reveals an Ancient Origin of the Distal Limb Domain Features we think of as defining land animals were quietly brewing inside “fish” for eons. The category boundary dissolves further the closer you look at the genetics.
Senses That Evolved Independently Across “Fish”
If all fish were truly one group, you might expect them to share a common sensory toolkit inherited from their shared ancestor. Some features do go back that far, but others tell a more complicated story. Electroreception, the ability to detect electrical fields in water, is one of the clearest examples. Many sharks, rays, and primitive bony fish can sense electrical signals through specialized organs called ampullary electroreceptors. These appear to be ancestral in vertebrates.
Among the ray-finned fish, however, the story fractures. The teleosts, the enormous group that contains the vast majority of living fish species, lost ancestral electroreception early in their history. Several teleost lineages later re-evolved ampullary electroreceptors independently, but these new organs work in a fundamentally different way from the originals: they respond to opposite electrical polarity, and the voltage-sensing mechanism sits on a different part of the cell membrane.7Journal of Experimental Biology. The evolution and development of vertebrate lateral line electroreceptors In other words, a catfish’s electroreceptor and a shark’s electroreceptor look like the same adaptation, but they were built from scratch using different biological machinery. Grouping these animals together as “fish” implies a unity that their actual sensory biology contradicts.
Eyes That Saw Land Before Bodies Walked On It
One of the more surprising discoveries about the water-to-land transition involves eyes. Measurements of fossil eye sockets show that vertebrate eyes nearly tripled in size just before the first animals started living on land. Simulations showed that this increase provided almost no benefit for seeing through water. In air, however, bigger eyes dramatically improved visual range. The researchers concluded that these animals were likely still aquatic but had begun looking up through the water’s surface into the air, in a lifestyle that resembled modern crocodiles: bodies in water, eyes scanning above it.8PubMed Central. Massive increase in visual range preceded the origin of terrestrial vertebrates
This finding reinforces the broader point. There was no sharp transition from “fish” to “land animal.” These creatures were adapting their sensory systems to an aerial world while they were still swimming. The eyes of the earliest land vertebrates were not newly evolved for land; they were modified fish eyes that had already been optimized for seeing through air. The transition was a long, overlapping process, not a clean jump across a category boundary.
Why Everyone Still Says “Fish” Anyway
If “fish” is not a real group, why does the word persist? Part of the answer is that our brains are wired to categorize animals by what they look like and where they live. Research on how people naturally classify living things shows that humans across very different cultures organize organisms into similar folk categories based on overall appearance and ecological role, not evolutionary history.9PubMed Central. Folk biology and the anthropology of science: cognitive universals and cultural particulars “Thing that lives in water, has scales, breathes through gills, and swims with fins” is a category that feels obvious because our brains evolved to notice exactly those surface patterns. Evolutionary relationships are invisible to the naked eye, so they never shaped everyday language.
Ancient classifications followed the same instinct. Roman categorization of aquatic animals, for instance, organized creatures by where they lived and what they looked like, not by any concept of shared descent.10BioOne / Anthropozoologica. The Roman classification and nomenclature of aquatic animals: an annotated checklist (with a focus on ethnobiology) This is not a flaw in human thinking. It is a perfectly sensible way to organize the world if your goal is catching dinner or avoiding danger. It just does not map onto the actual tree of life.
Legal Fish and Functional Fish
The gap between evolutionary reality and practical language shows up vividly in law. California’s Fish and Game Code defines “fish” to include not just what a biologist would recognize as fish but also mollusks, crustaceans, other invertebrates, and amphibians.11American Entomologist. Terms of Art and Terms of Arthropods Under this definition, an oyster is legally a fish. A frog is a fish. This is not biology; it is administrative convenience. The word “fish” in legal and regulatory contexts functions as shorthand for “aquatic wildlife we want to protect,” and that definition has survived court challenges precisely because the law does not need to reflect evolutionary classification.
Ecologists also use “fish” in ways that openly depart from evolutionary grouping. When marine researchers need to categorize species by ecological role rather than ancestry, they often create functional groups based on traits like body size, feeding strategy, and habitat. One such analysis of marine species found that three functional groups consistently emerged: reef and bottom-dwelling species including skates and rays, large open-ocean and deep-sea species, and sharks as a separate cluster.12PeerJ. Creating functional groups of marine fish from categorical traits These groupings are explicitly not evolutionary categories. They are practical tools for understanding ecosystems, and they work well for that purpose. Nobody involved is confused about the evolutionary relationships. The point is that what ecologists study day to day does not always require evolutionary groupings, so “fish” remains useful shorthand in those contexts even though it has no evolutionary validity.
The Ongoing Push Toward Honest Classification
Ichthyologists have been aware of the “fish problem” for decades, and formal classifications have been shifting to reflect it. A comprehensive review of bony fish classification noted that molecular data have been reshaping the field, revealing natural groupings that older systems based on anatomy alone never anticipated. But the review also found that progress toward a fully evolutionary framework has been slow, with many classifications still mixing well-supported evolutionary groups with arbitrary holdovers from older systems.13PubMed Central. Phylogenetic classification of bony fishes Old habits in taxonomy, as in language, die hard.
There have also been broader efforts to overhaul how biological names work. The PhyloCode, an alternative system of naming organisms, was designed specifically to define groups based on evolutionary branching points rather than traditional ranks like class or order. While it does not outright ban the recognition of groups that exclude some descendants, its architects have consistently championed naming practices that tie directly to the tree of life.14Oxford Academic (Systematic Biology). The PhyloCode and the Distinction between Taxonomy and Nomenclature In practice, most working biologists today avoid using “fish” as a formal taxon. You will see Actinopterygii (ray-finned fish), Chondrichthyes (sharks and rays), and Sarcopterygii (lobe-finned fish, including land vertebrates) treated as real groups. “Fish” survives in conversation, on menus, and in law, but not in the technical literature as a category that means anything evolutionary.
Jaws, and How They Reshape the Story
Even the jaw, a feature so fundamental to most vertebrates that the primary split in vertebrate classification is between jawed and jawless animals, tells a story of evolutionary innovation rather than clean categories. The jaw is thought to have arisen by modifying one of the front gill arches of an ancient jawless ancestor. That transformation established a new developmental program in which upper and lower biting structures formed from the same embryonic tissue that previously supported a gill.15PubMed Central. Evolution of the vertebrate jaw: comparative embryology and molecular developmental biology reveal the factors behind evolutionary novelty Every jawed vertebrate, from a great white shark to a sparrow to a human, inherited that same modified arch. When you chew your food, you are using repurposed gill anatomy from an animal that would have looked, to any casual observer, like a fish.
Understanding jaw evolution also illuminates why lampreys and hagfish feel so different from everything else we call a fish. They lack jaws entirely, having diverged from the vertebrate tree before that gill-arch modification occurred. Grouping them with salmon and sharks under “fish” obscures one of the most important events in vertebrate history. The category flattens a half-billion years of evolutionary divergence into a single word.
When the Category Does Real Harm
For most people, calling a trout a fish and leaving it at that causes no practical problems. But in science communication and conservation, the false equivalence embedded in the word can mislead. When researchers report that “fish populations are declining,” the statement lumps together groups with radically different biology, reproductive strategies, and conservation needs. Shark populations face different pressures than coral reef teleosts, which face different pressures than deep-sea species. The functional-group approach to marine ecology exists partly because treating “fish” as a single management unit leads to policies that miss these differences.
In evolutionary biology education, the word creates a subtler problem. Students who learn that “fish evolved into amphibians” absorb the idea of a clean progression: fish, then amphibians, then reptiles, then mammals. That ladder image is wrong. Life is a branching tree, not a staircase. Your lineage did not stop being “fish” at some point; it was always part of the same branching process. Lungfish alive today are not failed attempts at becoming land animals. They are a successful, fully modern lineage that happens to share a more recent ancestor with you than with a tuna. Dropping the word “fish” as a formal category helps make the tree-shaped reality of evolution easier to see.