Spinosaurus Went Deep: What AI Says About Returning to the Water

One of paleontology's longest-running arguments is whether the giant sail-backed predator Spinosaurus was a true diver or just a waterside stalker. A new study out of Yale University, published in Current Biology, comes down hard on the diving side — and the way the researchers got there is as interesting as the conclusion itself. For Lowcountry collectors, the story is a useful reminder that the hardest question in vertebrate paleontology isn't always "what is this animal?" but "how did it actually live?"

What the Yale Team Did

Lead author Caleb Gordon, who finished his doctorate at Yale and is now at the Florida Museum of Natural History, worked with senior author Bhart-Anjan Bhullar and colleagues to assemble an enormous comparative dataset. According to the summary, they examined hundreds of fossil and living specimens from the Yale Peabody Museum and dozens of other institutions, gathering more than 11,000 measurements along with photographs and CT scans.

Rather than lean on a single anatomical clue, the team trained machine-learning models on modern animals whose lifestyles are known, then asked those models to estimate how extinct species used the water. Because the method learns the relationship between skeletal proportions and real-world behavior, it can even predict soft-tissue features — such as flippers — that almost never fossilize. The researchers report that forelimb proportions proved especially informative, letting their models guess with better than 90% accuracy whether an animal had soft-tissue flippers and whether it was highly or fully aquatic. They also folded in a statistical technique that, oddly enough, traces back to World War II naval work.

The Verdict on Spinosaurus

Spinosaurus lived roughly 113 to 94 million years ago in what is now North Africa. For years the debate has run between two camps: one picturing a seal- or penguin-like pursuit predator that chased prey underwater, the other imagining a heron-style wader picking fish from the shallows. Gordon's team lands firmly on the aquatic interpretation, concluding that the animal "spent the vast majority of its time submerged." As the researchers themselves note, this doesn't resolve every detail of how the dinosaur hunted, but it strengthens the case that it was a water-dweller far more than a shoreline tourist.

Just as striking is the flip side of the analysis. Mesosaurs — small marine reptiles from about 290 to 274 million years ago — have often been read as fully aquatic because they bore live young. The Yale models disagree, suggesting their limbs stayed suited to moving on land, much like a modern alligator. The team goes further, proposing that all Paleozoic marine reptiles kept coming back to shore, with even the most water-adapted among them remaining amphibious. In short, live birth alone doesn't prove a creature cut ties with land.

Why Convergence Matters to South Carolina

Here's where the story comes home. The same evolutionary phenomenon driving this research — secondary aquatic adaptation, in which land animals re-invade the sea — is written all over the fossil record of the Charleston embayment. South Carolina may not produce Spinosaurus, but our marls and river gravels are packed with animals that made exactly the "U-turn" the Yale team studied.

Consider the Oligocene Chandler Bridge and Ashley formations around Charleston. These units have yielded some of the most scientifically important early toothed and baleen whales in the world, including small, archaic forms that document cetaceans in the middle of their own return to the water. Whales are the textbook case of secondary aquatic evolution — descendants of hoofed land mammals that eventually became fully marine. The dolphins, whale ear bones, and vertebrae that turn up in Lowcountry diving and creek material are pieces of that same grand experiment in convergence, the one that independently produced streamlined bodies and flipper-shaped limbs again and again.

That connection is more than poetic. The method Gordon developed is, at its core, a way to squeeze behavioral information out of bones when direct evidence is missing — and incomplete bones are the daily reality of Lowcountry collecting. Our fossils arrive as isolated, water-worn elements in a gravel lag: a single whale vertebra here, a chunk of rostrum there, rarely an articulated skeleton. Tools that can read lifestyle from forelimb or limb-bone proportions are precisely the kind of approach that could someday wring more meaning out of the fragmentary cetacean material the Ashley River system keeps producing.

A Caution Worth Keeping

The mesosaur result carries a lesson every collector and reader should absorb: a single "aquatic" trait can mislead you. Live birth looked like proof of a sea-bound life, and it wasn't. In our own backyard, it's tempting to assume any robust, dense bone pulled from a blackwater river belonged to a committed swimmer. But dense, heavy bone — pachyostosis — shows up in shoreline waders and bottom-walkers too, and the Yale study is a reminder that behavior is best reconstructed from the whole animal, not one convenient clue. The researchers are careful to hedge where the data are ambiguous, and that restraint is exactly the mindset to bring to a mystery bone in the sorting tray.

The Bigger Picture

What makes this work exciting is that it offers a repeatable, quantitative way to ask whether an extinct animal truly lived in the water — moving the conversation past storytelling and toward testable prediction. For Spinosaurus, that means a stronger case for a genuinely aquatic giant. For the rest of us, it means the analytical net is tightening around one of evolution's most repeated tricks: leaving the land and going home to the sea.

The next time you turn over a whale earbone or a worn cetacean vertebra from a Cooper or Ashley River lag, it's worth remembering that you're holding evidence from one of those return journeys — and that researchers are building ever-better tools to read the details locked inside it.