A Gobi fossil overturned how scientists read a whole branch of the mammal family tree

A chipmunk-sized skeleton from the eastern Gobi Desert has forced paleontologists to move an entire group of Cretaceous mammals from one branch of the family tree to another. The animal, named Tamirkhan balcarceli, measures about 6 to 7 inches from nose to tail and is the most complete specimen of the extinct group called zhelestids ever recovered.

For nearly 40 years, zhelestids were known mostly from isolated teeth and fragments. Those teeth were low and rounded, built for eating plants, and they led researchers to argue over whether the animals were early relatives of today’s placental mammals, an unknown lineage of hoofed mammals, or something else entirely. A single skeleton now points to a different answer: according to the researchers, zhelestids belong with the shrew-like zalambdalestoids, a group of small insect-eating mammals, and not with the placentals or the hoofed mammals.

A museum find from 2004 and the animal it turned out to be

The specimen was collected in 2004 by Andres Giallombardo, who was then a graduate student on a Museum-sponsored expedition, according to a ScienceDaily report of the findings from the American Museum of Natural History. He later became the lead author of the paper that formally describes it. In the Museum’s own release, Giallombardo called it “the thrill of a career to find a new species so completely preserved.”

The fossil comes from the Khugenetsavkhlant badlands of the eastern Gobi, in the Upper Cretaceous of Mongolia, and it preserves a nearly complete skull and skeleton rather than a handful of molars. Museum curator Michael Novacek, a coauthor who has co-led annual Gobi expeditions since 1990 with the Mongolian Academy of Sciences, described the Gobi as one of the only places where such remarkably complete Cretaceous mammals are routinely recovered, and said fossils like this “continue to transform” understanding of mammalian evolution.

Kazakhstan, Kyrgyzstan and Uzbekistan also have zhelestid fossil sites, so the group was never confined to Mongolia. Paleontologists have informally called zhelestids “Cretaceous rabbits” because of their rabbit-like look and long hind limbs, a nickname that the new skeleton complicates, since the animal’s closest relatives are now the shrew-like zalambdalestoids.

Long incisors and slender legs that gave the game away

The animal still has the rounded, plant-grinding molars that defined zhelestids What the older material rarely showed was the rest of the body; the new skeleton supplies the skull, the incisors and a partial hind limb adapted for running, all from one individual. Tamirkhan has long, continuously growing incisors, several distinctive skull features, and long, slender hind legs with specialized ankle bones, a package the researchers identify with the shrew-like zalambdalestoids, a group of small insect-eating mammals.

Shawn Zack, a paleontologist at the University of Arizona and a coauthor, put it bluntly in the same report: “Tamirkhan’s hind legs are long and slender with distinctive ankles, traits that are unmistakably zalambdalestoid.” The Nature article adds specifics, naming elongate first lower incisors, translacrimal canals and fused lower hind limbs with long metatarsals as skeletal features of that clade.

The paper’s title states the conclusion directly: “Cretaceous zhelestid mammals are zalambdalestoids”. Its abstract describes the zhelestid molar pattern as a dental variant within Zalambdalestoidea, not a sign of a plant-eating radiation close to placentals. The specimen is catalogued as holotype PSS-MAE 618, and the authors placed it in the family tree using the MorphoBank database of anatomical characters, according to the journal page.

Why teeth alone misled researchers for decades

The reclassification rests on convergent evolution, the process by which unrelated animals independently develop similar traits. On the team’s reading, zhelestids and the herbivorous mammals they had been compared with arrived at similar tooth shapes separately, so molars alone pointed in the wrong direction. Maureen O’Leary, a paleontologist at Stony Brook University and a research associate at the Museum, said the work shows that teeth cannot always reveal how a whole animal looked, and the Museum’s account links her remark to Georges Cuvier’s claim, made more than 200 years ago, that a single tooth could predict an animal’s full anatomy.

The finding also touches a live argument about when placental mammals arose. Molecular clock studies support an origin well before the end of the Cretaceous, and some researchers had proposed that zhelestids were among the earliest placentals, which would have offered fossil support for that timing. According to the Museum’s release, the new analysis argues against a close placental relationship, so the fossil evidence cited for an early placental radiation is no longer supported by this group. The release does not say that the molecular estimates are wrong; it says only that the new analysis argues against zhelestids being close placental relatives.

Paúl Velazco, a comparative biologist at Arcadia University and a coauthor, said fossils supply the evidence needed to test predictions made by molecular models. The Nature paper went online on 29 July 2026 and appeared in the 24 September issue of Nature, volume 657, pages 973 to 980. The study is also indexed in the PubMed record.

The authors’ own claim is narrower than a general verdict on tooth-based classification. It is that one specimen, catalogued as PSS-MAE 618, carries zhelestid molars on a zalambdalestoid body, and that the group of zhelestids sits inside the zalambdalestoids. The sources do not say how many other Cretaceous mammals known only from teeth might be misplaced in the same way, and that remains an open question the new skeleton cannot answer by itself.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview