Dunkleosteus: Smaller Than We Thought
Dunkleosteus terrelliA 2023 re-analysis put Dunkleosteus at an estimated 3.4 to 4.1 m, down from older 6 to 10 m figures, yet its skull alone reached 60 to 70 cm in length.
Dunkleosteus was a giant armoured fish that patrolled the shallow tropical seas of the Late Devonian period, roughly 382 to 358 million years ago. Despite a superficial resemblance to modern sharks, it was not a shark, not a bony fish and not anything alive today.
Dunkleosteus represented an entire vertebrate class called Placodermi, which ruled Devonian seas for over fifty million years before disappearing entirely during the Late Devonian mass extinction event, with no surviving descendants. This class is one of only a few vertebrate groups in the fossil record to have gone completely extinct.
This guide covers every aspect of Dunkleosteus biology, ecology and fossil history: its armoured head, its missing body, its toothless but terrifyingly effective jaws, its place in the Devonian food web, the mass extinction that killed it and the ongoing scientific argument about how big it actually was. It is a reference entry: so expect specifics: millimetres of armour, newtons of bite force, millions of years of geological time and named fossil localities.
Etymology and Classification
The genus name Dunkleosteus was coined in 1956 by American paleontologist Jean-Pierre Lehman in honour of David Dunkle, then curator of vertebrate paleontology at the Cleveland Museum of Natural History. The suffix -osteus is Greek for "bone," a reference to the massive bony plates of the head and thorax.
Before 1956 the animal was lumped into the older genus Dinichthys : literally "terrible fish", a name that had accumulated nearly every large arthrodire species found in Ohio since the 1870s. Modern work split that wastebasket taxon into several distinct genera, of which Dunkleosteus became the most famous.
The type species is Dunkleosteus terrelli , named for Jay Terrell, a hotel owner and amateur paleontologist who found the first fossils in cliffs along Lake Erie in Ohio in 1867. About ten species are currently assigned to the genus, though several of those assignments are contested and a revision of the genus is likely in coming years. Placement in the broader tree of life is as follows:
- Kingdom: Animalia
- Phylum: Chordata
- Class: Placodermi (extinct)
- Order: Arthrodira (extinct)
- Family: Dunkleosteidae (extinct)
- Genus: Dunkleosteus
- Species: D. terrelli (type), plus D. belgicus , D. denisoni , D. marsaisi , and others
Placoderms are one of the few entire vertebrate classes in the fossil record known to go completely extinct without leaving living descendants. Every other major vertebrate class alive during the Devonian, cartilaginous fish, bony fish, lobe-finned fish, tetrapods, still has living members today. Placoderms do not.
This makes Dunkleosteus and its relatives scientifically interesting far out of proportion to their geological lifespan: they are a complete evolutionary experiment, run to failure, preserved in stone.
Anatomy and the Missing Body
Dunkleosteus is one of the most visually iconic prehistoric animals precisely because its anatomy is so strange and so incomplete. The animal was armoured only across the head and the front portion of the body. Everything behind the thoracic shield was cartilaginous and therefore does not fossilise in normal conditions.
The head shield:
- A single dome-like carapace of bony plates covering the skull roof
- Small, forward-facing eyes set into reinforced orbits
- Large openings for water flow past the gills
- Articulation points that allowed the entire head shield to pivot upward
The thoracic shield:
- A saddle-like arrangement of plates covering the chest and shoulders
- Fused to the head via a pair of balanced ball-and-socket neck joints
- Openings through which the pectoral fins emerged
- The lower plates continued the armour across the belly
The posterior body (inferred):
- Probably shark-like and fusiform, narrowing to a heterocercal tail
- Paired pelvic fins
- Mostly cartilaginous internal skeleton
- No confirmed dorsal fin despite frequent reconstructions showing one
The key anatomical feature is the neck joint between head shield and thoracic shield. Unlike any animal alive today, arthrodire placoderms rotated the entire upper head shield upward at this joint during a bite. Effectively the mouth opened from both ends at once, lower jaw dropping, upper head rotating back, producing a gape out of all proportion to the skull's resting shape and generating strong inward suction.
The mechanism is called cranial kinesis, and while many modern fish have forms of it, the arthrodire version is its own evolutionary invention, unmatched in modern vertebrates.
The Size Debate
For most of the 20th century, the size of Dunkleosteus was estimated by extrapolating from the armoured head outward, using proportions inferred from sharks or from smaller, better-preserved placoderms. Those reconstructions produced figures of six to ten metres in length and body masses of several tonnes.
In 2023, Russell Engelman published a re-analysis in Diversity . Using several body-size proxies, including head length, the angle at which the body enters the head shield, and the position of the pectoral and pelvic girdles, mostly scaled from the smaller arthrodire Coccosteus, Engelman argued that adult Dunkleosteus terrelli was typically about 3.4 metres long, with rare exceptional individuals reaching 4.1 metres. That is roughly the length of a small great white shark.
Large Devonian arthrodires seem to have had relatively short, stocky bodies instead of the long, tapering tails that earlier reconstructions assumed. The head was truly enormous, with skulls reaching 60 to 70 centimetres in length, but it constituted a larger fraction of the entire animal than previously believed.
Of all the work done to date, this 2023 paper is the most comprehensive analysis, though the body remains unknown from complete specimens and estimates stay uncertain. Three conclusions are reasonably secure:
- Dunkleosteus was one of the largest known vertebrates of the Devonian, regardless of which size model is correct.
- The head and jaws were massive in absolute terms, making it a formidable predator.
- Older popular culture claims of a "10-metre, 4-tonne" Dunkleosteus are probably too large.
| Estimate source | Length | Approximate mass |
|---|---|---|
| Traditional (20th century) | 6-10 m | 1,000-4,000 kg |
| Engelman (2023) | 3.4-4.1 m | ~950-1,760 kg |
Jaws, Bite Force, and Speed
The most remarkable feature of Dunkleosteus biology is its feeding apparatus. The jaws lacked true teeth entirely. Instead, the upper and lower jaw bones carried long, blade-like projections of bone covered in dentine, the same material that forms the bulk of a mammalian tooth. These gnathal plates sheared past each other like a pair of scissors when the jaws closed.
Because the plates were made of continuously-growing dentine, they were self-sharpening. Every bite wore the opposing edges into sharper profiles. The plates grew from beneath as they wore above, so in theory the animal never lost its cutting edge. The trade-off was that Dunkleosteus could not shed and replace teeth the way sharks do: if a plate chipped or broke, the animal had to grow the damage out.
The 2006 biomechanical study:
In 2007, Philip Anderson and Mark Westneat published a mechanical model of the Dunkleosteus terrelli skull in Biology Letters . Treating the skull, lower jaw and suspensorium as a four-bar linkage, they calculated:
- Bite force at the blade edge: ~5,400 N (up to ~7,500 N in a later model)
- Bite force at the tip of the anterior fangs: ~4,400 N (up to ~6,200 N in a later model)
- Jaw opening time: ~20 milliseconds
- Jaw closing time: ~50-60 milliseconds
The authors considered these among the highest bite forces of any fish and among the highest of any animal. A 2025 study by Engelman and colleagues questioned both the bite force estimates and the suction-feeding interpretation, arguing that the skull contained more cartilage than previously assumed.
The modelled jaw opening time of about 20 milliseconds is comparable to modern fishes that use suction feeding to help capture prey.
This combination, shearing bone blades plus crushing force plus explosive opening speed, is unusual in vertebrate evolution. Most large predators specialise in one or two of those traits. Dunkleosteus had all three.
Diet, Hunting, and Cannibalism
Dunkleosteus was the apex predator of its ecosystem. There is no credible candidate for an animal that regularly preyed on adult Dunkleosteus, except other Dunkleosteus. Its diet included:
- Sharks and shark-like cartilaginous fish (Cladoselache, common in the Cleveland Shale)
- Other placoderms, including smaller arthrodires
- Other Dunkleosteus (suggested by puncture-marked specimens)
- Ammonoids and other cephalopods
- Large free-swimming invertebrates
Evidence of cannibalism:
One Dunkleosteus specimen (CMNH 5302) and a specimen of the related Titanichthys show puncture damage attributed to the bony fangs of other Dunkleosteus. Cannibalism is common in modern apex predators, and this damage is the fossil evidence suggesting it in Dunkleosteus.
Evidence of regurgitation:
The Cleveland Shale also preserves clusters of fragmented bone and armour plates that appear to be regurgitated material, the undigested hard parts of prey, coughed back up after the flesh was extracted. This is the same behaviour modern owls and some seabirds use to clear their crops of indigestible bone.
Hunting strategy:
No one has watched Dunkleosteus hunt, but its anatomy implies a strategy. Earlier studies proposed fast jaw opening and suction feeding, but the 2025 reassessment compared its blade-like dentition to that of predatory fish that pursue prey and bite it into pieces with a wide gape. Teeth of the fast-swimming chondrichthyan Orodus found with Dunkleosteus remains suggest it may have been fast enough to catch quick prey.
Habitat and Geographic Range
During the Late Devonian, the world's continents were arranged very differently than today. The supercontinent Gondwana occupied much of the southern hemisphere, while Laurussia (also called Euramerica) sat near the equator and was bordered by shallow tropical seas. Dunkleosteus lived in those shallow tropical seas, and its fossils are found in rocks that once lined the edges of Laurussia.
Major fossil localities:
- Cleveland Shale, Ohio, USA: the richest source, with fossils preserved in black anoxic mudstone
- Chagrin Shale and Huron Shale, Ohio: other Ohio Shale units that have yielded Dunkleosteus
- Pennsylvania and New York, USA: scattered material
- Belgium and Poland: European arthrodire-bearing formations
- Morocco: D. marsaisi and other species from North African shelf deposits
The Cleveland Shale is exceptional because the Devonian seafloor there was oxygen-starved. Scavengers could not survive at the bottom, and carcasses that sank into the mud were buried largely intact. That anoxic burial preserved the bony head and thoracic shield, and in at least one specimen the outline of a pectoral fin.
Life in the Devonian Seas
The Late Devonian is sometimes called the Age of Fishes. The first great radiation of jawed vertebrates was in full swing. The seas contained:
- Placoderms: dominant predators and grazers, including Dunkleosteus and relatives
- Early sharks: Cladoselache and other cladodonts, smaller and faster than today's sharks
- Acanthodians: small "spiny sharks," mostly filter-feeders and small predators
- Early bony fish: lobe-finned and ray-finned lineages, still diversifying
- Ammonoids and nautiloids: shelled cephalopods filling pelagic predator roles
- Trilobites. Past their peak, yet still to be found on the sea floor.
- Reef-builders: stromatoporoid sponges and rugose/tabulate corals
Early forests filled Laurussia and Gondwana with Archaeopteris and lycophyte trees. Tetrapods started moving from shallow water onto land. Warm climate and few ice caps prevailed as atmospheric carbon dioxide declined because the new forests were burying organic carbon.
Against this background, Dunkleosteus occupied the top of the marine food chain in the shallow tropical seas around Laurussia. Its ecological role is comparable to that of modern large sharks and killer whales combined, a fast, armoured, heavily-biting apex predator with no effective natural enemies as an adult.
Reproduction and Growth
Placoderms were among the first vertebrates known to reproduce by internal fertilisation. Several related placoderms (notably Materpiscis and Incisoscutum) have been found with preserved embryos inside the body cavity and with bony pelvic claspers used to transfer sperm.
Whether Dunkleosteus itself bore live young or laid fertilised eggs is not directly known, because the relevant pelvic anatomy of Dunkleosteus is not preserved. But given that internal fertilisation is ancestral for the wider placoderm group, Dunkleosteus probably fertilised internally as well.
Studies of the lower jaws of juvenile D. terrelli show they were proportionally as robust as those of adults, so juveniles could already produce high bite forces, unlike juvenile tetrapods, whose jaws are more slender than those of adults.
Extinction: The Late Devonian Mass Extinction
The Late Devonian mass extinction is counted as one of the five great mass extinctions of the Phanerozoic. It was not a single event but a series of extinction pulses spread over roughly 20 to 25 million years toward the end of the Devonian period. Two pulses are particularly important for placoderms:
- The Kellwasser event (~372 million years ago): at the Frasnian-Famennian boundary, devastated reef communities and coral-sponge ecosystems
- The Hangenberg event (~358 million years ago): at the Devonian-Carboniferous boundary, finished off the placoderms entirely
Proposed causes include:
- Ocean anoxia. Shallow seas lost oxygen widely, which directly killed marine animals and preserved black shales such as the Cleveland Shale.
- Global cooling: driven in part by the expansion of early forests drawing carbon dioxide out of the atmosphere
- Sea-level changes: multiple regressions and transgressions that disrupted the shallow-shelf habitats where placoderms thrived
- Possible volcanic activity: the Viluy large igneous province in Siberia is sometimes implicated in the Frasnian-Famennian pulse
- Meteorite impacts: some impact structures date to this interval, though their role is debated
By the end of the Hangenberg event, the entire class Placodermi was gone. Not a single placoderm lineage survived into the Carboniferous. Sharks diversified into much of the space left empty. Bony fish eventually radiated into the rest. But no placoderm ever evolved again. Dunkleosteus and its relatives are the most spectacular casualties of an extinction event that completely reshaped the vertebrate world.
Fossil History and Scientific Study
The first Dunkleosteus fossil was found in 1867 by Jay Terrell, a hotel owner and amateur paleontologist, in cliffs along Lake Erie near Sheffield Lake, Ohio. Terrell donated his fossils to John Strong Newberry of the Ohio Geological Survey. Newberry described the material in 1873 as the new genus Dinichthys , and in 1875 named the more common Cleveland Shale species Dinichthys terrelli . For the next eighty years, nearly every large arthrodire head from Ohio was funnelled into the genus Dinichthys .
The modern taxonomy began with Lehman's 1956 revision, which split Dinichthys and erected the new genus Dunkleosteus , named for David Dunkle. Subsequent work through the late 20th century added species, refined relationships and produced the first mechanical reconstructions of the jaw.
Key 20th and 21st century works:
- Newberry (1873, 1875): original descriptions of Dinichthys and Dinichthys terrelli
- Lehman (1956): erection of genus Dunkleosteus
- Carr and Hlavin (1995, 2010): revisions of dinichthyid and dunkleosteid arthrodires
- Anderson and Westneat (2007): biomechanical model, bite force and speed
- Engelman (2023): body-length re-estimation using several size proxies
The Cleveland Museum of Natural History, the Smithsonian and the American Museum of Natural History hold the largest collections.
Why Dunkleosteus Matters
Dunkleosteus is scientifically important for reasons that go beyond its size and jaws. It is a window into a vertebrate class that is entirely extinct. Everything placoderms tried, the armoured head, the shearing dentine blades, the rotating cranial joint, the internal fertilisation by claspers, is preserved only in fossils. Modern fish did not inherit those features.
Any direct study of placoderm biology is a study of a parallel evolutionary experiment that ran alongside our own ancestry and then ended.
It is also important as a case study in mass extinction. The Late Devonian extinction hit reef-builders and shallow warm-water marine groups especially hard, and placoderms were hit hard by the Kellwasser event before disappearing entirely in the Hangenberg event.
Understanding why such animals are selectively vulnerable in mass extinctions is directly relevant to predicting the consequences of modern ocean deoxygenation and warming.
Finally, Dunkleosteus is important to the public understanding of deep time. The animal is visually striking, the story of its extinction is dramatic and its jaws are mechanically extraordinary. For many people Dunkleosteus is the first prehistoric animal they learn about that is not a dinosaur and not a mammoth, a reminder that the history of life is far older and far stranger than the Mesozoic and Cenozoic alone.
Frequently asked questions
The size of Dunkleosteus has been revised downward. Traditional reconstructions extrapolated the missing cartilaginous body from the known armoured head and put adult Dunkleosteus terrelli at roughly 6-10 metres long and 1-4 tonnes in weight.
In 2023 Russell Engelman published a re-analysis using several body-size proxies and concluded that typical adults were about 3.4 metres long, with rare exceptional individuals reaching about 4.1 metres, roughly the length of a small great white shark. Even at the downsized figure it was among the largest vertebrates of the Late Devonian, with skulls up to 60-70 centimetres long.
No. Dunkleosteus had no true teeth. Instead its jaws were lined with gnathal plates, blade-like projections of bone covered in dentine, that sheared past each other like scissors. The edges were self-sharpening as the plates ground against one another during feeding.
The downside was that Dunkleosteus could not shed and replace its plates the way a shark replaces teeth, so a chipped or broken plate had to be grown out.
A 2007 study by Anderson and Westneat modelled the jaw mechanics of Dunkleosteus terrelli using a four-bar linkage and estimated a bite force of roughly 4,400 newtons at the tip and 5,400 newtons at the blade edge. A later model gave up to about 6,200 and 7,500 newtons. The authors considered these among the highest bite forces of any animal.
The same model suggested the jaws could open in about 20 milliseconds. A 2025 study by Engelman and colleagues questioned both the bite force estimates and the suction-feeding interpretation.
Dunkleosteus lived during the Late Devonian period, roughly 382 to 358 million years ago. Its fossils have been found in the United States, Canada, Poland, Belgium, Morocco and Russia.
Dunkleosteus and every other placoderm disappeared during the Late Devonian mass extinctions. Placoderms were hit hard by the Kellwasser event and died out completely in the Hangenberg event at the end of the Famennian, around 358 million years ago.
The Late Devonian extinction had multiple pulses, including the Kellwasser event around 372 million years ago and the Hangenberg event around 358 million years ago. Placoderms were hit hard by the Kellwasser event and finished off by the Hangenberg event.
Proposed causes include ocean anoxia (widespread oxygen loss in shallow seas), global cooling linked to the expansion of early forests drawing down atmospheric carbon dioxide, and sea-level changes that disrupted shallow-shelf habitats.
Placoderms are one of the few vertebrate classes to have gone entirely extinct, leaving no living descendants, though they are generally considered close to the ancestry of modern jawed vertebrates.
Dunkleosteus was an apex predator of its Devonian ecosystem. Its likely prey included sharks and other cartilaginous fish, other placoderms, ammonoids and possibly large free-swimming invertebrates. Teeth of the fast-swimming chondrichthyan Orodus have been found with Dunkleosteus remains.
Fossils of Dunkleosteus are frequently found with clusters of fish bones that appear to be regurgitated, and puncture damage on a Dunkleosteus specimen has been attributed to the fangs of other Dunkleosteus, suggesting cannibalism.
That is one of the open questions in Devonian paleontology. Only the head and the thoracic shield of Dunkleosteus were heavily armoured with bony plates, and those are what fossilise. The rest of the body was cartilaginous, so it is mostly unknown. Reconstructions of the posterior body rely on better-preserved smaller arthrodires such as Coccosteus.
The 2023 Engelman analysis argued that the body was probably shorter and stockier than the classic long-tapered reconstructions suggested.
Placoderms occupy a key position near the base of jawed vertebrates (Gnathostomata). Placoderms are thought to be paraphyletic, consisting of several groups that are outgroups or sister taxa to all living jawed vertebrates, which originated among their ranks. However, Dunkleosteus itself was a specialist that left no descendants.
Placoderms are better described as stem jawed vertebrates: close relatives of our ancestors, but not the direct line and the Arthrodira, including Dunkleosteus, left no living descendants when the class went extinct.
References
- Extinction-event stratigraphy follows the Geologic Time Scale 2020 and the International Commission on Stratigraphy Devonian-Carboniferous boundary framework. Cleveland Shale taphonomy follows work by the Cleveland Museum of Natural History research collections team.
- Engelman, R. K. (2023). A Devonian Fish Tale: A New Method of Body Length Estimation Suggests Much Smaller Sizes for Dunkleosteus terrelli (Placodermi: Arthrodira). Diversity 15(3): 318. View source