Four bones from Colombia’s La Venta beds carry puncture wounds, crushed cortex, and a second set of serrated grooves. A new Journal of Vertebrate Paleontology paper says the first attacker was Purussaurus. The second was likely a land-walking sebecid cleaning the carcass.
 |
| Reconstruction of the faunal assemblage of La Venta, showing the hypothetical predation event with Purussaurus neivensis attacking Pericotoxodon platignathus; also pictured in the foreground are Anhinga cf. A. grandis and Caiman sp. Image credit: Miguel Hernandez. |
Key Takeaways by Planet Today
The claim: Oscar Wilson and colleagues report the first direct physical evidence that giant crocodylians preyed on large ungulates in Middle Miocene Colombia, based on bite traces on four bones from three herbivore species.
The animals: The prey are Pericotoxodon platignathus, a rhinoceros-sized toxodont, and two astrapotheres, Xenastrapotherium kraglievichi and Granastrapotherium snorki. Reconstructed masses in the paper run from about 616 kg to about 1,822 kg.
The predator: Size, shape, and crushing of the punctures point to Purussaurus neivensis, a caiman relative estimated near 7 meters and about 1,800 kg. A second trace type on one jaw is assigned to scavenging by Langstonia huilensis.
The force: Some puncture reconstructions approach or exceed 40,000 newtons per tooth. The authors note that at least one figure, on thin frontal bone, is probably an overestimate.
The setting: La Venta, in today’s Tatacoa Desert of Huila, Colombia, sampled a wetland world when South America was still an island continent. Large placental cats and dogs had not yet arrived across Panama.
The implication: If the traces are predation rather than only scavenging, reptiles — not marsupial sabertooths or terror birds — were the main check on the biggest plant-eaters in that ecosystem.
What is still open: Four specimens, found incidentally, cannot measure how often such hunts happened. Healing is absent, so death-by-bite is inferred from location and damage, not filmed. Other large crocodyliforms lived in the same wetlands.
A crime scene in a museum drawer
Paleontology rarely catches a kill in the act. Bones lying next to other bones only say that two species shared a landscape. Tooth marks say something stricter: this mouth closed on that body. That is why ichnology — the study of traces — matters more than another reconstruction of a giant caiman standing in a swamp.
On 16 August 2026, the Journal of Vertebrate Paleontology published “Physical evidence of crocodylian predation on ungulates in the Middle Miocene of Colombia.” Lead author Oscar E. Wilson, a postdoctoral researcher at the University of Helsinki, worked with colleagues in Colombia, France, Argentina, Canada, and the United States. The University of Helsinki issued the embargoed summary the same day through EurekAlert. Sci.News followed on 17 August. ScienceDaily posted the Helsinki account on 19 August, which is the latest widely circulated write-up as of this article.
The sample is small and specific. Four fossils. Three species. A skull, a distal femur, and two mandibles. All come from the La Venta assemblage in Huila Department, the best-sampled Neotropical vertebrate community of the Middle Miocene. The marks are circular and bisected punctures, pits, crushed cortical bone, and, on one jaw, a row of parallel grooves.
“There was so much prey available that mammalian predators couldn’t have eaten it all. Our new study shows that crocodylians in particular were the region’s most important predators, especially when it came to hunting large prey.” — Oscar Wilson, University of Helsinki
That sentence is the paper’s public argument. The technical argument is narrower: these traces match Purussaurus teeth and modern crocodile feeding damage better than they match the other large predators on the same floodplain.
What the four bones show
Pericotoxodon platignathus was a toxodont, a native South American ungulate built on a heavy, rhinoceros-like plan. A partial cranium, catalogued MPV 826, preserves a large bisected circular puncture assigned to the ichnotaxon Nihilichnus hastarius. Maximum diameter is 29.2 millimeters. That sits inside the range of 22 isolated Purussaurus teeth measured in the Villavieja collections, which run from 19.6 to 43.7 millimeters at the base, with a mean of 27.8 millimeters. A distal femur referred to the same species carries a series of semicircular punctures and smaller pits. Body-mass reconstructions for these two Pericotoxodon specimens sit near 616 kg for the femur and roughly 930–980 kg for the skull.
The other two animals are astrapotheres — the so-called lightning beasts, tapir-like to elephant-like herbivores with a short trunk. A left mandible of Xenastrapotherium kraglievichi (UCMP 38115) is the busiest specimen. It has large punctures with associated fractures, assigned to Nihilichnus fissuratus, plus crushing labeled Brutalichnus, and ten linear parallel grooves on the buccal side assigned to Knethichnus parallelum. The paper’s mass for that individual is about 1,687 kg. A Granastrapotherium snorki mandible (VPPLT 1798) also shows multiple punctures and deformation; its reconstructed mass is about 1,822 kg. Popular summaries rounded the set to “between roughly 600 and 1,800 kg.” The paper’s specimen-level numbers are more useful.
In every case the authors interpret Purussaurus as the maker of the large punctures, because of tooth diameter, carinated (keeled) tooth edges that can leave vertices on a bisected pit, and the amount of cortical collapse. They do not claim a unique smoking-gun tooth still stuck in the bone. They claim the best available match.
UCMP 38115 is also the specimen that records a second feeder. The parallel grooves are the kind of mark left by ziphodont — serrated, blade-like — teeth. The likely maker is Langstonia huilensis, a terrestrial sebecid crocodyliform. The sequence the authors favor is a fatal or near-fatal Purussaurus attack, then later scavenging by the smaller land-dweller. That is an interpretation of two trace types on one bone, not two eyewitnesses.
How hard a bite has to be
Sci.News reported that some of the bite forces needed to make the punctures approached or exceeded 40,000 newtons per tooth, in line with earlier Purussaurus estimates. The paper calculates minimum forces from the area of crushed green bone, using a compressive-failure approach. It also warns that the figure attached to the thin frontal bone of MPV 826 is probably too high. Dense mandible and femur are better mechanical analogues than a plate of skull roof.
Readers will see larger numbers elsewhere. A 2015 biomechanical study of Purussaurus, often cited in this literature, produced very high whole-jaw estimates, and some news stories this week mixed those older, larger-species figures with P. neivensis. The Colombian animal in this paper is not the later, even bigger Purussaurus brasiliensis of the Amazon. Helsinki and the journal describe P. neivensis as about seven meters long and about 1,800 kilograms — truck-sized, not bus-sized. Keeping the species straight is the difference between reporting and collage.
Modern saltwater and Nile crocodiles remain the living comparison the authors want. Those animals ambush at the water’s edge, seize the head or neck of large prey, and hold or roll. Three of the four La Venta specimens are marked on the head or jaws. That anatomical concentration is the main reason the team prefers active predation over a random scavenger walking over a dry carcass. It is also why they say Miocene giants can be read, with care, through living crocodylian behavior.
An island continent without lions
A reminder that belongs in any account of this fauna: in the Middle Miocene, South America was still isolated. The Isthmus of Panama had not closed. That land bridge is generally dated to around three million years ago, after which the Great American Biotic Interchange mixed northern and southern mammals. Fourteen million years ago, and still at twelve, the big placental carnivores of later savannas — cats, dogs, bears — were not part of this food web.
The mammalian hunters that did exist were mostly sparassodonts, the borhyaenoid marsupials, some of them saber-toothed. Terrestrial predators also included giant flightless phorusrhacid birds and sebecids such as Langstonia, crocodyliforms with laterally compressed, tyrannosaur-like teeth. The Helsinki release lists anacondas and saber-toothed mammals among the shoreline threats. None of those animals, in the authors’ view, occupied the large-prey niche as completely as the giant aquatic caimans.
South American fossil assemblages have long shown a low ratio of mammalian predators to prey. That observation is older than this paper. What was missing at La Venta, for the largest ungulates, was a bone with the right holes in it. Previous Purussaurus traces had been reported on a mylodont sloth tibia from Peru, on the giant turtle Stupendemys, on another crocodylian, and, in 2025, on a terror-bird tibiotarsus from La Venta itself. The new work extends that list to toxodonts and astrapotheres in the 600-to-1,800-kilogram class.
Wilson’s remark about surplus prey is ecological, not romantic. If herbivores are abundant and mammalian carnivores are few, something else has to eat the surplus or the plant-eaters would be limited only by food and disease. Giant crocodylians, sitting at the water’s edge of a continent-scale wetland, are a plausible something else. The traces make the plausibility empirical for these three species.
What mainstream coverage emphasized
University and agency channels stayed close to the fossils. Helsinki, EurekAlert, and ScienceDaily stressed dominance of the food chain and the 23-foot, 1,800-kilogram caiman. Sci.News, on 17 August, led with “first direct physical evidence” and listed the four bones. AAAS Science’s news brief used the pickup-truck comparison and noted a 2025 study in which crocodylians appear to have fed on a terror bird. Nautilus and other features lingered on head-first ambush and leftovers for scavengers.
None of those outlets invented a controversy that the paper does not contain. They did, in a few places, flatten P. neivensis into a generic “giant crocodile” and borrow size records from other Purussaurus species. That is ordinary popularization error, not an alternate theory.
There is, as of 20 August 2026, no substantial alternative-media story claiming the marks are fake, misdated, or planted. The skeptical position that exists is the one already written into the paper’s limitations. Trace fossils are not photographs. Predation and scavenging can look alike once healing is absent. Four incidental specimens are not a census of diet. Other large crocodyliforms at La Venta — including the gharial-like Gryposuchus and the strange, floppy-snouted Mourasuchus — have been discussed in earlier work as possible hunters of big terrestrial prey on size alone. This paper argues that snout shape, tooth shape, and the new traces make Purussaurus the better specialist on large ungulates.
That is a debate among measurements, not a debate among worldviews. Readers who want the argument in the authors’ own words should use the journal article, not the headline that says a crocodile “ruled” a continent.
La Venta as a wetland, not a desert
Today the Tatacoa is dry. In the Middle Miocene the same rocks record lakes, rivers, and the northern edge of the vast Pebas wetland system that covered much of western Amazonia. The reconstructed scene in the paper — and in Miguel Hernandez’s illustration — puts Purussaurus neivensis on Pericotoxodon, with a darter (Anhinga) and a smaller Caiman in the foreground. It is labeled hypothetical, which is the correct label. The fossils are real; the lunge is inferred.
Stratigraphic ages cited for the analyzed fossils span about 10.5 to 16 million years. The classic La Venta, or Laventan, fauna is often placed near 13 to 12 million years ago. Sci.News used “roughly 12 million years ago.” Both the broad window and the familiar midpoint appear in the coverage. Either way, the animals lived millions of years before the Panama land bridge, in a fauna that still belonged to South America’s long isolation.
Collections used in the study include the Museo Paleontológico de Villavieja, the Museo de La Tatacoa, the University of California Museum of Paleontology, and the Muséum National d’Histoire Naturelle in Paris. Methods included photography, photogrammetry, CT scanning, caliper measurements, dental and cranial mass regressions, and comparison of puncture diameters with actual Purussaurus teeth. That last step is the one that keeps the story from being only “big holes, therefore big crocodile.”
How far the conclusion should travel
Wilson said the frequency of marks likely made by Purussaurus suggests it was an important predator in these tropical ecosystems. Frequency, here, means that among the damaged ungulate bones they examined, Purussaurus-type traces show up more than once and on more than one species. It does not mean a percentage of all La Venta herbivores. The authors note that the traces were recognized incidentally, not through a systematic survey of every large bone in every drawer. A systematic search could raise or lower the apparent rate.
They also cannot always separate a killing bite from a first-come scavenger on a fresh carcass. The head-and-jaw pattern, the lack of healing, and the analogy to living crocodiles are their reasons for calling the large traces predation. A cautious reader can accept the traces, accept Purussaurus as the best match, and still reserve judgment on whether every animal was taken alive. The paper itself treats that distinction as difficult.
Body-mass estimates from teeth can run high. Ziphodont sebecids have no precise living analogue, so Langstonia’s “predation zone” is harder to reconstruct than a caiman’s. Missing parts of a skull may have been bitten away, or weathered away, or collected incompletely. Those caveats are not hostility to the result. They are the difference between four well-described specimens and a complete food-web model.
If later surveys find the same puncture geometry on sloths, glyptodonts, and more astrapotheres, the case that Purussaurus regulated large herbivore numbers gets stronger. If the marks remain rare, the animal can still have been capable of such hunts without being the statistical king of the floodplain. Capability and dominance are not the same sentence.
A food web with too many mouths at the waterline
La Venta is famous for crocodyliform diversity, not just for one giant. Aquatic generalists, slender-snouted fish-eaters, duck-faced Mourasuchus, and terrestrial sebecids overlapped. That crowding is part of why diet reconstructions matter. If every large reptile ate the same tonne-class ungulates, they would have been in constant interference. Tooth shape suggests they did not. Purussaurus has the robust, blunt-to-conical anterior teeth of a bone-crushing generalist. Langstonia has slicing dentition better suited to flesh. Gryposuchus is built more like a gharial. The new traces put the crushing role on the caiman and the leftover-slicing role, in one documented case, on the sebecid.
Mammalian predators remain in the picture. They simply do not, on present evidence, leave this class of holes in this class of bone. Terror birds were real apex threats on land; a 2025 La Venta paper argued that a crocodyliform, possibly a medium-sized caiman, fed on one of them. Apex predators eating other apex predators is what a crowded wetland looks like. It is also a reminder that “top of the food chain” is a headline phrase. Chains at La Venta were webs, and the water’s edge was the dangerous edge.
For readers who follow how physical traces get turned into ecological stories — in laboratories as well as in rock — see Planet Today’s science desk at planet-today.com, including reporting on biotech changes in everyday food production and on archaeological claims that rest on bones rather than on rumor.
What to watch next
Three follow-ups would test the result without requiring a new miracle fossil.
First, a systematic ichnological pass through La Venta ungulate collections. If Purussaurus-type punctures are common on heads and rare on mid-shafts, the ambush reading holds. If they are scattered, scavenging rises.
Second, more sebecid and sparassodont comparisons. Langstonia is named here as a scavenger on one mandible. That does not prove sebecids never killed large prey. It proves that on this bone, two mouth types fed.
Third, tighter dates and field context for each specimen. A bone from a lake-margin deposit and a bone from a drier paleosol do not tell the same hunting story, even if the holes look similar.
Until then, the documented facts are these. Four Middle Miocene ungulate bones from Colombia carry crocodylian bite traces. The large punctures fit Purussaurus neivensis better than the other named suspects. One jaw also records a ziphodont scavenger, likely Langstonia. Head-focused damage resembles living crocodile attacks. The wider claim — that giant crocodylians were the most important predators of large prey in that ecosystem — is a reasonable reading of a low mammalian-predator ratio plus these traces. It is not a completed census.
That is enough to change how the La Venta mural should be painted. It is not enough to retire every marsupial sabertooth and every terror bird from the shoreline.
Sources
- Oscar E. Wilson, Jorge W. Moreno-Bernal, Catalina Suarez, Frank Cardozo Tovar, Andrés Vanegas, Rubén Vanegas, Ethan A. Shirley, Patricia A. Holroyd, John-Paul Zonneveld. “Physical evidence of crocodylian predation on ungulates in the Middle Miocene of Colombia.” Journal of Vertebrate Paleontology, published online 16 August 2026. DOI: 10.1080/02724634.2026.2701154
- Full text: https://www.tandfonline.com/doi/full/10.1080/02724634.2026.2701154
- University of Helsinki / EurekAlert, 16 August 2026: Giant crocodylians dominated the food chain in Miocene South America
- ScienceDaily, 19 August 2026: This 23-foot crocodylian may have ruled South America’s food chain
- Sci.News, 17 August 2026: Giant Crocodiles Once Preyed on Massive Mammals in South America
- Related prior trace evidence: Pujos & Salas-Gismondi, 2020, Purussaurus on a mylodont sloth; Link et al., 2025, crocodyliform traces on a La Venta terror bird.
- Reconstruction image credit: Miguel Hernandez.
Original source
Assignment seed and general-audience summary: Sci.News, “Giant Crocodiles Once Preyed on Massive Mammals in South America,” 17 August 2026, https://www.sci.news/paleontology/south-american-miocene-crocodylians-14996.html. Underlying research: Wilson et al., Journal of Vertebrate Paleontology, 16 August 2026. Latest institutional recirculation: University of Helsinki via ScienceDaily, 19 August 2026.
Disclaimer for fact-checkers: This article reports a peer-reviewed ichnological study and its university press materials. Bite-maker identifications are matches of puncture size, shape, and bone damage to known teeth, not a predator preserved in the act. Predation versus scavenging is inferred. Body-mass and bite-force numbers are reconstructions with stated uncertainties. Popular size records for other Purussaurus species should not be substituted for P. neivensis. No claim here depends on hidden data, and no claim is stronger than four described specimens.