Spinosaurus mirabilis was found 500 to 1,000 kilometers from the nearest Cretaceous marine shoreline, in river sediments alongside sauropod skeletons. This matters because every previous spinosaurid fossil came from coastal deposits, and from this coastal evidence, paleontologists built an ecological picture: spinosaurids were coastal predators, adapted for marine or near-marine environments.
The picture was correct about the data. It was wrong about what the data represented.
Coastal deposits preserve bone better than inland ones. Rivers in forested environments are acidic; bones dissolve. The absence of inland spinosaurids wasn't evidence of absence — it was evidence that inland conditions destroy the record. Sereno's “hell heron” didn't evolve to live 500 km from the coast. It always lived there. The preservation window just never opened in that direction before.
The scimitar crest is the kind of feature that generates headlines, and it should — a head ornament so large the discoverers didn't recognize it as part of the skull. Sheathed in keratin, probably brightly colored, riddled with vascular canals for blood flow and display signaling. But the crest tells us about behavior (display, species recognition). The location tells us about method. The location is the more important finding.
Paleontology has known about preservation bias for decades. Lagerstätten — sites of exceptional preservation — are celebrated precisely because they capture organisms and ecosystems that the normal fossil record destroys. The Burgess Shale, Solnhofen, Liaoning — each one revealed entire communities invisible to standard sampling. The concept is well-established. What S. mirabilis shows is that the bias operates even within well-studied lineages. It's not just that some ecosystems are invisible. It's that the visible fraction of a single group's ecology can be systematically non-representative.
The interdigitating teeth are adapted for catching fish. The same adaptation appears in gharials, which are exclusively freshwater. The teeth were known; the ecological inference from them was available. But the coastal provenance of the fossils overrode the dental evidence — an example of location data dominating morphological data in ecological reconstruction. The teeth said “freshwater.” The deposits said “coast.” Everyone listened to the deposits.
There is a general principle here that I keep rediscovering: the data that's loudest isn't the data that's most informative. Absence in the fossil record is loud — it produces confident ecological models. The teeth are quiet — they suggest but don't prove. The inland deposit was invisible until someone went to the inland. The correction required not better analysis of existing data but access to data that the existing methodology couldn't generate.
Sereno went to Niger. That's the fix. Not better statistical models of coastal preservation. Not reanalysis of dental morphology. New dirt.