Insights  /  Research

The dinosaur literature, genus by genus

The dinosaur research atlas covers every non-avian dinosaur genus, 1,322 of them, each placed by its fossil localities and linked to the indexed literature. Its clearest result is how unevenly that literature is spread: a small set of famous genera hold most of it, and the apparent geographic and temporal gaps turn out to track the naming clock and the reach of English-language indexing, not research need.

World map of non-avian dinosaur fossil localities: 4,010 Paleobiology Database collections, dense across western North America, Patagonia, East Asia, Europe and southern Africa.

What the atlas is

The dinosaur research atlas maps the indexed literature on non-avian dinosaurs, genus by genus. The census is every valid non-avian dinosaur genus in the Paleobiology Database, body fossils only, with egg and footprint form-taxa excluded by their type specimen: 1,322 genera, split 473 theropods, 456 ornithischians, 378 sauropodomorphs and 15 early or unplaced. The literature is harvested from OpenAlex and linked to a genus by name, with a paleontological context guard on one-word genus names so homonyms, surnames and minerals do not leak in: 5,600 indexed papers and 10,012 genus-paper links.

Each of the 4,659 papers with an abstract was read by a four-expert panel against a fixed codebook, and every tag was confirmed by a second, independent model against the abstract text; a blind audit scores 81 percent precision. Each genus is placed by its fossil localities, 4,010 Paleobiology Database collections carrying country, formation, geological interval and age; 1,287 of the 1,322 genera are placed.

The literature is steeply concentrated

This is the atlas’s clearest and most robust finding. An independent model classified all 10,012 genus-paper links as subject, mention, or unrelated; 5,554, that is 55 percent, are papers whose subject is the named genus. Counting those subject links, the busiest 1 percent of genera hold 21.9 percent of the research and the busiest 10 percent hold 57.1 percent. Tyrannosaurus is the subject of 301 papers, more than twice the next genus, Iguanodon, at 112. Its raw name-match count is higher still, 557, because nearly half the papers that name it only mention it in passing; the concentration barely moves when those are removed, so the bias is real, not an artefact of name-dropping.

What the research is about

Systematics and anatomy dominate: 48 percent of the tagged papers are about naming and placing taxa, 44 percent about anatomy and functional morphology. Paleobiology and behaviour (20 percent), paleoecology (15 percent) and growth and histology (11 percent) form a thinner band. Macroevolution and extinction sit at 3 percent, but that figure is partly an artefact of the method: clade-level synthesis and extinction papers often name no single genus, so they fall outside a genus-anchored corpus. The mix describes the shape of the genus-linked literature, not the field’s full attention to any topic.

Where the fossils are

China holds the most recorded genera (307), then the United States (250), Argentina (174) and Mongolia (96). Most named genera are Cretaceous (931), with fewer from the Jurassic (300) and the Triassic (87). The richest single rock units are the classic Late Cretaceous beds of western North America and Mongolia, the Late Jurassic Morrison of the American West, and the Early Cretaceous Yixian of China, source of the feathered-dinosaur record.

Why there is no neglect map

It is tempting to rank countries or geological periods by how little their taxa have been studied. We tested that and will not publish it, because the ranking measures two things that are not research need. Rank countries by the share of their genera with few indexed papers and France comes out among the most “under-studied,” even after recent taxa are set aside; France is not research-neglected, but OpenAlex under-indexes its older and French-language work, so the count measures English-language indexing, not research. The second driver is timing: countries that look thin, such as Argentina and Brazil, are producing many newly named genera that have had no time to accumulate literature. Geological time behaves the same way, where the thinnest intervals are the ones with the sparsest fossil-bearing rock. So the atlas reports fossil localities as geography and stops there.

How the figures were checked

Tag precision was measured by a blind audit on the three domain dimensions: 81 percent of a 120-tag sample were confirmed by a fresh model reading each abstract. The subject rate of the genus-paper links, 55 percent, was measured on the whole corpus, not a sample. The census was cross-checked against the Paleobiology Database’s own reference counts, and the genera that appear there but not in our census turned out to be exactly the egg and footprint taxa and junior synonyms the body-fossil rules exclude. Recall is not yet estimated, and tagging covers only papers that carry an abstract, so every count describes the indexed, largely English-language record, never “all research.”

How to use it

The atlas is organised as sections that each answer one question: the census by clade, the concentration of the literature, the record by country, formation and time, what the research is about, and the full tag vocabulary. The interactive map plots every fossil locality and lets the corpus be filtered by genus, formation, geological period and research type; click a locality to read the papers whose subject is a genus found there.

Read the dinosaur atlas →

COMPASS ID makes research discoverable by what it covers, and by the gaps it has left dark. Browse the evidence atlases →