A Four-Winged Dragon From the Cretaceous: How a Fossil Fragment From Northwest China Is Reshaping the Flight Origins Debate

For more than a decade, paleontologists working in the Changma Basin of northwestern China's Gansu Province have pulled hundreds of ancient bird fossils from the Xiagou Formation — along with the gnawed, crushed remnants of bird bones compacted into pellet-like masses, the calling card of a predator no one could identify [1]. In June 2026, researchers finally gave that predator a name: Jian changmaensis, a barn-owl-sized, feathered dinosaur whose long pennaceous feathers on both forelimbs and hindlimbs gave it the appearance of a creature with four wings [2].

The find, published in the Annals of Carnegie Museum, does more than fill a gap in one Chinese fossil bed's food web. It extends the known range of microraptorines — the group of small, gliding dromaeosaurids most closely associated with the "trees-down" hypothesis of flight evolution — well beyond their stronghold in northeastern China's Jehol Biota, and into a geologic window that may make Jian the youngest known member of its lineage [3]. For a field still divided over whether birds learned to fly by gliding from trees or sprinting on the ground, the specimen arrives as a pointed piece of evidence — and a reminder that the fossil record can still surprise.

The Specimen: What a Shoulder and Arm Can Tell You

Jian changmaensis is known from a single specimen — catalogued as GSGM-D050 and permanently housed at the Gansu Geological Museum — consisting of an articulated pectoral girdle (the scapulocoracoid, or fused shoulder blade and coracoid bone) and a partial forelimb including the humerus, radius, and ulna [4]. No skull, no pelvis, no tail. In paleontology, partial specimens are the rule, not the exception, and the diagnostic value of Jian rests on a set of specific anatomical features, called autapomorphies, that distinguish it from every other known microraptorine.

Three stand out. First, the ratio of the coracoid to the humerus is larger in Jian than in any other described microraptorine — a proportion that suggests a more robust shoulder joint, possibly indicating stronger muscle attachments for the flight stroke [4]. Second, the distal condyles of the humerus (the rounded knobs at the far end of the upper arm bone, where it meets the forearm) sit primarily on the bone's cranial (front-facing) surface, rather than in the more typical lateral orientation seen in relatives like Microraptor zhaoianus [5]. Third, the proximal end of the radius displays a prominent foramen — a hole in the bone visible from the ventral (underside) view — not documented in other microraptorines [5].

The coracoid also preserves the supracoracoid fenestra, a large opening that nearly bisects the shoulder bone. This feature is shared with all known microraptors but is rare outside the group, serving as a reliable indicator of subfamily membership [5].

What the specimen does not preserve are soft-tissue impressions of feathers. The inference that Jian was four-winged comes from phylogenetic bracketing: because every other known microraptorine with preserved integument shows long pennaceous feathers on both forelimbs and hindlimbs, and because Jian nests within that clade, the same condition is inferred for this species [2][3].

How Jian Differs From Velociraptor

The popular imagination groups all dromaeosaurids together as "raptors," but Jian changmaensis and Velociraptor mongoliensis occupied opposite ends of the family's ecological spectrum. Velociraptor, from the Late Cretaceous of Mongolia (roughly 75–71 million years ago), was a ground-dwelling cursorial predator with relatively flat, moderately curved pedal claws suited for terrestrial locomotion [6]. Claw curvature studies on extant birds and fossil theropods consistently place Velociraptor among ground-dwelling taxa, with claw geometry matching that of modern ratites and galliforms [7].

Microraptorines, by contrast, show claw morphologies more consistent with climbing and perching. Their smaller body size — Microraptor gui weighed roughly one kilogram and measured about 77 centimeters in length [8] — combined with hindlimb flight feathers that would have hampered terrestrial running [9], points to an arboreal or semi-arboreal lifestyle. Computer simulations of Microraptor gui's flight performance suggest it was adapted for undulatory "phugoid" gliding between trees, using a biplane wing configuration where the horizontal feathered tail provided additional lift and pitch control [9].

Jian changmaensis appears to have been larger than the typical Microraptor specimen. Its upper arm bone fragment measures about four inches (roughly 10 cm), and researchers estimate the complete animal would have had a wingspan of approximately four feet — comparable to a barn owl [1]. Jingmai O'Connor, associate curator of fossil reptiles at Chicago's Field Museum and a coauthor on the study, described it as "one of the biggest microraptor specimens that has ever been found" [10].

Geography and Geology: Why the Changma Basin Matters

Until Jian, every undoubted microraptorine specimen came from northeastern China, primarily from the Yixian and Jiufotang Formations that together constitute much of the Jehol Biota — a fossil assemblage dating to roughly 133–120 million years ago [11]. These formations, in Liaoning Province, are famous for their exceptional preservation, including soft tissues, feather impressions, and gut contents. The Jehol deposits have yielded Microraptor zhaoianus, M. gui, M. hanqingi, Sinornithosaurus, Changyuraptor, Zhongjianosaurus, and Wulong [12].

The Xiagou Formation of the Changma Basin sits roughly 2,000 kilometers to the west and represents a contemporaneous but distinct depositional environment. The formation dates to the Aptian stage of the Early Cretaceous, approximately 124–120 million years ago [4]. Matt Lamanna, the Mary R. Dawson Curator of Vertebrate Paleontology at the Carnegie Museum of Natural History and corresponding author of the study, has led fieldwork in the Changma Basin since 2004 [3].

The geographic expansion matters because it suggests microraptorines were not confined to the humid lakeside forests of the Jehol but ranged across a far wider swath of Early Cretaceous China. If Jian proves to be among the geologically youngest microraptorines, it would also extend the group's temporal range, challenging the notion that four-winged dromaeosaurids were a brief evolutionary experiment.

The Growing Roster of Four-Winged Dinosaurs

The description of Jian changmaensis brings the count of described four-winged non-avian dinosaur taxa to approximately eight, depending on how species boundaries are drawn [12]. The trajectory of discovery has been steady since Xu Xing's landmark 2003 description of Microraptor gui as a "four-winged dinosaur" [13].

Source: Compiled from paleontological literature

Data as of Jun 6, 2026CSV

A persistent complication is taxonomic lumping versus splitting. The three named Microraptor species — M. zhaoianus (2000), M. gui (2003), and M. hanqingi (2012) — have been argued by some researchers to represent variation within a single species, properly called M. zhaoianus, with differences attributable to ontogenetic (age-related) or individual variation rather than species-level divergence [12]. If the lumpers are correct, the true count of distinct four-winged genera drops. Skeptics could ask whether Jian itself might fall within the range of variation for known Microraptor — a question the authors address by pointing to the distinctive coracoid-to-humerus ratio and humeral condyle orientation as features outside the known variation in any Microraptor specimen [5].

The Flight Origins Debate: Trees Down or Ground Up?

Every four-winged dinosaur discovery feeds into one of paleontology's most durable arguments: did avian flight evolve from arboreal ancestors that glided downward ("trees-down"), or from cursorial ground-dwellers that developed flapping thrust through running ("ground-up")?

The trees-down model, championed in various forms since the 19th century, received its strongest modern support from the Jehol microraptorines. The reasoning is straightforward: animals with long flight feathers on all four limbs would have been poor runners but effective gliders, and the four-winged configuration could represent a transitional stage between non-flying dinosaurs and two-winged birds [9][13].

The ground-up model draws support from the observation that most non-avian theropods, including small feathered deinonychosaurs and Archaeopteryx, show limb proportions and claw geometries that cluster with fully terrestrial modern mammals rather than arboreal ones [14]. A 2011 study in PLOS ONE testing arboreal adaptations in bird antecedents concluded that results "reject an arboreal capacity for the avian stem lineage," lending no support for the trees-down model [14]. More recent experimental work has shown that active flapping motion could develop naturally in running dinosaurs without requiring a prior gliding phase [15].

No formal survey has quantified the split among active paleornithologists, but the ground-up model appears to hold a plurality of support in the field, particularly at institutions like the American Museum of Natural History, where it has been the favored framework for exhibit interpretation [16]. The trees-down camp, however, has gained ground with each new four-winged taxon described — and with biomechanical modeling that demonstrates the aerodynamic plausibility of microraptorine gliding [9].

Jian changmaensis does not settle the debate, but it complicates the ground-up narrative by showing that four-winged morphology persisted across a wide geography and potentially a longer time span than previously documented. If four-winged flight were a brief dead end, one would expect it to be geographically and temporally constrained. The Changma Basin find suggests otherwise.

Convergence or Common Ancestry?

A subtler question is whether the four-winged body plan evolved once in a common ancestor of microraptorines or arose independently in multiple lineages. Xu Xing's original 2003 paper noted that four dromaeosaurid species possessed hindlimb flight feathers, raising the possibility that this was an ancestral condition for the broader group [13]. If so, ground-dwelling dromaeosaurids like Velociraptor would represent a secondary loss of hindlimb feathers, not the primitive state.

Some phylogenetic analyses have recovered topologies suggesting that hindlimb feathering appeared convergently in different dromaeosaurid lineages, a finding that, if confirmed, would mean four-winged flight was an evolutionary strategy that multiple lineages stumbled upon independently [12]. This convergence scenario weakens the argument that four-winged taxa represent a single linear stage in the path to modern birds. Instead, it paints four-winged flight as a recurring experiment — one that worked well enough to evolve more than once but never well enough to persist into the Late Cretaceous.

Either way, the four-winged lineages went extinct without leaving modern descendants. No living bird has flight feathers on its legs comparable to those of Microraptor. Whether this makes the four-winged clade an "evolutionary dead end" or a "stepping stone" depends partly on definition: if four-winged ancestors gave rise to two-winged descendants by losing the hindwing feathers, then the stage was transitional even if the four-winged forms themselves died out [9].

Research Publications and Ongoing Interest

Academic output on Microraptor and related taxa has remained robust. According to OpenAlex data, 423 research papers mentioning "microraptor" have been published since 2011, with a peak of 61 papers in 2018 [17]. Publication volume in 2026 stands at 14 papers so far, reflecting continued — though not accelerating — interest in the group.

Source: OpenAlex

Data as of Jan 1, 2026CSV

Provenance, Custody, and the Shadow of Forgery

The provenance of Chinese dinosaur fossils carries particular weight because of the field's history with forgeries. The most notorious case remains "Archaeoraptor," a composite fossil assembled from parts of at least two different species, purchased from a Chinese dealer, smuggled to the United States, and featured in National Geographic in 1999 before being exposed as a fraud [18]. CT scanning by paleontologist Timothy Rowe confirmed the forgery, which had been assembled to increase the specimen's commercial value [19]. The specimen was reportedly sold for $80,000 on the private market [18].

Jian changmaensis avoids many of the red flags associated with the Archaeoraptor scandal. The specimen was collected during licensed fieldwork that Lamanna's team has conducted in the Changma Basin since 2004 [3]. It is permanently accessioned as GSGM-D050 at the Gansu Geological Museum, a Chinese government institution, ensuring that it remains available for independent verification and future study [4]. The study was published in a peer-reviewed journal, with coauthors from both Chinese and American institutions — the Carnegie Museum, the Field Museum, and the Chinese Academy of Sciences' Institute of Vertebrate Paleontology and Paleoanthropology (IVPP) [3][10].

China's fossil protection laws, rooted in the 1982 Cultural Relics Protection Law and strengthened by subsequent regulations administered by the Ministry of Land and Resources, classify vertebrate fossils as state property [20]. Export of significant paleontological specimens without government approval can carry severe penalties, including imprisonment [21]. These regulations mean that the physical specimen will remain in China, and international researchers seeking to study it will need to work through established institutional agreements with the Gansu Geological Museum.

What This Find Does and Does Not Prove

Jian changmaensis is a fragmentary specimen — a shoulder and partial arm — from which researchers have extracted a remarkable amount of information. It establishes the presence of microraptorines in northwestern China for the first time, potentially extends the group's temporal range, and adds another data point to the fossil record of four-winged non-avian dinosaurs [3].

What it cannot do, on its own, is resolve the flight-origins debate. A single partial forelimb does not preserve the hindlimb feathers that would confirm four-winged gliding capacity; that inference rests on phylogenetic reasoning rather than direct observation. Nor can a lone specimen from one locality establish whether four-winged morphology evolved once or multiple times across dromaeosaurid phylogeny.

The specimen's greatest contribution may be its geographic context. Found among the remains of more than 100 Early Cretaceous birds — some of them apparently its prey — Jian changmaensis offers a glimpse of an ecosystem where feathered dinosaurs hunted feathered birds from the canopy, a scenario that only a few decades ago would have seemed improbable [1]. That such an animal existed far from the Jehol Biota's famous fossil beds suggests the four-winged body plan was more widespread, more successful, and more durable than the "evolutionary dead end" label implies.