The Gobi Desert of Central Asia is one of the earth's desolate places. Its million square kilometers of sand dunes, sculpted badlands and saw-toothed mountains are alternately scorched by summer's high-latitude sun and frozen by winter's Siberian winds. It is not a place to explore unprepared: crossing vast, uninhabited areas between a sprinkling of oases requires careful planning akin to the siege tactics for scaling a Himalayan peak or traversing the Antarctic continent. There are few maps, and satellite navigation is of limited help to a traveler trying to choose among deeply rutted, wildly crisscrossing roads that wander as unpredictably as the nomadic settlements they connect. Even a modern expedition runs the risk of water, fuel and food shortages. Getting lost is not merely frustrating but a matter of serious danger.

Yet the Gobi is a paradise for paleontologists. Its eroding terrain exposes nearly complete skeletons of creatures hitherto known only through painstaking reconstructions from a few scattered bones. Our expeditions, jointly sponsored by the Mongolian Academy of Sciences and the American Museum of Natural History, have excavated dinosaurs, lizards and small mammals in an unprecedented state of preservation. Freshly exposed skeletons sometimes look more like the recent remains of a carcass than like an 80-million-year-old fossil. The skeletons and skulls we have found are often complete or nearly complete, in sharp contrast to the fragments typically recovered elsewhere.

No one knows why fossils in the Gobi are so well preserved. In other specimen-rich areas, such as the one that became the Rocky Mountains, streams or rivers carried animal remains to fossil sites, disarranging them along the way. The Late Cretaceous environment in the Gobi, however, may have been much as it is today: open valleys of sand dunes and cliffs, sparsely watered by small, seasonal lakes or streams. Indeed, indications of ancient sand dunes can be observed in rock sections there.

It is also apparent that the animals were buried very soon after their death, before scavengers or weather had much time to get at them. Poorly sorted layers of sandstone in the Cretaceous rock formations suggest deposits of the kind one would expect in violent sandstorms. In the early 1990s Tomasz Jerzykiewicz, then at the Geological Survey of Canada in Calgary, and his colleagues studied fossil beds in Chinese Inner Mongolia and found that vertebrate fossils are often embedded in these layers. Such storms might not have merely buried carcasses but killed animals as well. Entombed in a matter of minutes or hours, their remains emerge some 80 million years later, almost undisturbed.

Stumbling on Protoceratops
Mongolia was not always recognized for its bounty of prehistoric material. During the late 19th and early 20th centuries, the Rocky Mountains region of western North America was the mecca for vertebrate paleontologists. Then, in 1922, Roy Chapman Andrews, a scientist from the American Museum of Natural History, led an expedition into the heart of the Gobi and changed the geography of the fossil world. Andrews chronicled his five expeditions in a remarkable narrative entitled The New Conquest of Central Asia. The romance and excitement of the enterprise foreshadowed the exploits of the movie character Indiana Jones. En route, the explorers were challenged by trackless dune fields, raging sandstorms and marauding bandits. Andrews's caravan of camels and spindly-wheeled Dodge motorcars was a logistic nightmare across the moonscape of the Gobi.

One of the most important discoveries in the history of scientific exploration came in the midst of such difficulties. Late in the first field season of 1922, the expedition got off track. Team photographer J. B. Shackleford wandered toward an unassuming rock rim at the edge of a field. There he was startled to find a fantasy of red cliffs and spires—and fossils.

Within 10 minutes he had uncovered the first known skull of the Protoceratops, a parrot-beaked, shield-headed dinosaur that has since become a reference fossil of the Late Cretaceous of Central Asia. The crew recovered more bones and even a small egg, which they mistook for that of a bird. They returned the next summer to find an extravagance of dinosaurs, ancient mammals and other vertebrates, as well as the first known cluster of dinosaur eggs. Their findings, particularly the eggs, became front-page news. Andrews named the place the Flaming Cliffs, inspired by their magnificent red-orange glow in the late afternoon sun.

By the beginning of the 1930s Andrews, frustrated by a volatile political scene in Mongolia, gave up his exploration. The Gobi was inaccessible to Western interests for more than 60 years, leaving Soviet-bloc scientists to extend the work Andrews had begun. Between 1946 and 1949, joint Russian-Mongolian expeditions penetrated the Nemegt basin and uncovered rich badlands of Cretaceous and Cenozoic fossils there.

Zofia Kielan-Jaworowska, a world-renowned fossil-mammal specialist at the Institute of Paleontology in Warsaw, led a highly skilled and energetic Polish-Mongolian team to the Nemegt and other areas between 1963 and 1971. She and her colleagues produced a series of classic monographs and a magnificent display of dinosaurs and other fossil vertebrates at the Natural History Museum in the Mongolian capital of Ulaanbaatar. Since the 1960s Mongolian paleontologists have conducted extensive fieldwork both independently and in collaboration with Soviet (now Russian) scientists.

Westerners first returned after the development of Mongolian democracy in 1990. That summer our colleagues at the Mongolian Academy of Sciences invited us for a reconnaissance that paved the way for more ambitious expeditions during succeeding years.

Rockies of the Desert
If anything, the contrast between the Gobi and other, more accessible fossil areas has increased since Andrews's time. A century ago, in the glory days of dinosaur hunting in the American West, prospectors encountered valleys and canyons where skeletons were exposed like corpses on a deserted battlefield, but today many prime dinosaur hunting grounds appear nearly exhausted.

The cumulative activities in Mongolia over the past 80 years, in comparison, do not approach those in the Americas. Erosion is still exposing a wealth of fossils even at sites well mined by Andrews and others. Moreover, the very difficulty and unexplored nature of the Gobi increases the chance that paleontologists may yet stumble onto wholly unexplored pockets of badlands.

Early in the 1993 season, with our Mongolian colleague, the now late Demberelyin Dashzeveg of the Mongolian Academy of Sciences, our field party struck out for an undistinguished set of red-brown sandstones on the north side of the Nemegt Valley, near the base of a jagged mountain range called Gilbent Uul. Previous expeditions, Dashzeveg said, had ignored this region in their rush to reach the more dramatic badlands of the western Nemegt Valley. We arrived at the area, struggled for a few kilometers along a wash, and established a bivouac where our heavy gasoline tanker and trailer buried itself in the sand.

The next morning we started prospecting the hills and gullies nearest camp. Within hours it was clear that we had come across one of the richest concentrations of fossils ever recovered from the age of the dinosaurs. In a basin less than two kilometers across, we found scores of dinosaur skeletons and egg nest sites weathering on gentle slopes. Intermixed with the dinosaur fossils were abundant smaller vertebrates—lizards and mammals—that were also key members of the ancient Cretaceous ecosystem.

The local name for the site of this bonanza is Ukhaa Tolgod (“Brown Hills”). Its natural amphitheater contained roughly 100 readily visible dinosaur skeletons, many of them in nearly pristine condition. During subsequent field seasons we selected the most desirable specimens. Among them are 25 skeletons of theropod dinosaurs. This group of agile carnivores runs the gamut from the enormous Tyrannosaurus and Allosaurus through fast-running dromaeosaurs such as Velociraptor (the villainous predator of Jurassic Park, a title some 60 million years out of date) to smaller birdlike creatures such as the oviraptorids. We also gathered an unprecedentedly rich collection of small vertebrates: more than 200 skulls of mammals—many with their associated skeletons—and an even greater number of lizard skulls and skeletons.

Cretaceous Treasure Trove
As the variety of our specimens makes clear, the flowering of terrestrial life during the Cretaceous of Central Asia was not limited to dinosaurs. The Gobi of 80 million years ago supported a wide variety of lizards, crocodilians and mammals. We have found specimens representing more than 30 species of lizard; some are extremely well preserved and display anatomical features that offer clues to the relations among major lizard families.

Probably the most spectacular of these is Estesia. Early one morning during a reconnaissance, we came across an exquisite, 20-centimeter-long skull with knife-edged teeth half embedded, like a bas-relief, in a vertical slab of sandstone. At the time we thought it belonged to a small carnivorous dinosaur, but later examination determined that the skull was that of a wholly new kind of large predatory lizard, closely resembling the Komodo dragon of today. We named the species after Richard Estes of San Diego State University, the world's foremost authority on fossil lizards before his death in 1990.

Estesia is a very primitive animal and as such is significant for understanding the family tree of the varanoid lizards (the group that includes the Komodo). The skull has an unusual series of canals at the base of the teeth that suggests Estesia injected poison into its prey. This lethal weapon is not common to living varanoids but is found in the Gila monster of the southwestern U.S. and northern Mexico.

We have since uncovered fragments of Estesia in other sites where smaller lizards, tiny mammals and dinosaur eggshells are common. Modern varanoids are noted for their voracious and wide-ranging appetites. It is likely that Estesia ate smaller vertebrates, small dinosaurs and possibly dinosaur eggs.

Although much of the Cretaceous Gobi was dry, water must have been abundant in at least a few places and times. We found occasional fossils of turtles, usually associated with aquatic habitats.

Some of the greatest treasures of the Cretaceous Gobi are easy to miss when scanning the slopes and gullies: the tiny skulls and skeletons of mammals. These fossils represent important precursors of the great mammalian radiation that followed the extinction of the dinosaurs at the end of the Mesozoic.

The bulk of scientific information on these earlier mammals comes from North American fossils, which are mostly fragmentary jaws and teeth. In fact, there are virtually no complete skulls of these Cretaceous mammals from North America. As a result, the Gobi assemblage, including our finds and those of earlier expeditions, surely represents the world's reference collection for Late Cretaceous mammals.

A small block recovered from Ukhaa Tolgod revealed six shrewlike placental mammals, each only a few centimeters long. Amazingly, the fossils consist of complete skulls attached to skeletons; such tiny bones are usually found separated or broken. These small creatures were probably buried rapidly after they had died.

We have found two basic groups of mammals. The first is the multituberculates, or “multis,” as paleontologists call them. They are a curious array of animals with long front incisors and molars with a complex of bumps (tubercles) on the tooth crowns. The Mongolian Cretaceous multis offer by far the best accumulation of skeletal material for examining the family trees of these creatures.

Multis can be thought of as the rodents of their time, even though they are in fact only distantly related to modern groups of mammals. Their rodentlike adaptations are a sign of convergent evolution with the rats, mice and squirrels familiar today. The multis thrived through the first several million years of the Paleogene period, after the dinosaurs had died out. They then dwindled in number and disappeared, replaced by more recent groups of similar habits.

The second group is the therians, ancestors of both marsupials and modern placental mammals (a category ranging from whales to bats, aardvarks and humans). These early therians consist of half a dozen shrewlike forms whose traits offer clues to the origins of later members of the group.

Mammals, lizards and other vertebrates are crucial to reconstructing the past environment of the Gobi and to tracing the main lines of evolution. But dinosaurs still occupy center stage in the public eye. The Cretaceous Gobi is unquestionably one of the world's great dinosaur hunting grounds. The fossils range from complete skeletons of Tarbosaurus—a fierce carnivore closely related to the North American Tyrannosaurus—to giant sauropods, duck-billed dinosaurs, armored ankylosaurs, frilled ceratopsian dinosaurs such as Protoceratops and a magnificent assemblage of smaller carnivores. Birdlike oviraptorids and dromaeosaurs such as Velociraptor are better represented in the stratified rocks of the Gobi than anywhere else in the world.

These remains have given rise to controversies but also to some definitive conclusions. Artists often depict Velociraptor hunting in packs like African wild dogs, for example, but there is no definitive proof that it was capable of such cooperative behavior. The predator's taste for Protoceratops, however, is more than a matter of speculation. In the late 1960s a group of Polish and Mongolian scientists at Tugrugeen Shireh, a white sandstone escarpment about 80 kilometers west of the Flaming Cliffs, excavated one of the most remarkable pair of specimens in the history of paleontology. Two nearly intact skeletons—a Protoceratops and a Velociraptor—are preserved locked in mortal combat. Velociraptor clutches the lowered head of Protoceratops with its forelimbs and raises the killing hooks of its hind claws high against its prey's flanks. The “fighting dinosaurs,” which may have met their end in one of the Gobi's sandstorms, are one of the great exhibits of the Natural History Museum in Ulaanbaatar.

Velociraptor skeletons are not only fascinating for the image they convey of intelligent, swift and lethal terrors. They offer clues to the evolutionary connection between birds and dinosaurs. Velociraptor and its relatives have many birdlike features, including the construction of the bony case of the brain and the design of the elongated limbs and digits. A nearly complete skeleton of Velociraptor unearthed at Tugrugeen in 1991 has a nearly complete braincase; in its details the architecture of the braincase is surprisingly similar to that of modern birds.

An unexpected discovery at Tugrugeen further amplified the proposed connection between dinosaurs and birds. We found a delicate skeleton that was identical, except for its smaller size, to one discovered by Mongolian scientists some years earlier. The animal, roughly the size of a turkey, has a remarkably gracile frame with long legs. In addition, the keel of the breastbone is extremely well developed. In modern birds, strong pectoral muscles that power the downstroke of the beating wing attach to this keel. Instead of long wing bones, however, this creature has stubby, massive forelimbs somewhat like those of a digging mole. The end of the arm and hand is appointed with a single, very large claw; hence, the scientific name bestowed on the animal is Mononykus—literally, “single claw.”

Mononykus is a bizarre creature. Although it has no wings, it has features that suggest a closer relation to modern birds than the famous primitive bird Archaeopteryx. A detailed analysis of Mononykus favors the view that this creature was a flightless relative of modern birds.

That argument has drawn some criticism. Certain specialists claim Mononykus is simply a small dinosaur whose birdlike features are a product of convergent evolution. The weight of the evidence, however, does not favor convergence. The history of birds is marked by species (such as ostriches) that have lost their capacity for flight. Our Mononykus fossils do not show evidence of feathers, but it is only by some miracle of preservation that the fine Jurassic limestone entombing Archaeopteryx leaves impressions of tiny feathers. Mononykus, like most fossils, is not preserved in such unusual rock.

Egg Hunter or Incubator?
Eggs of both dinosaurs and birds, found in various parts of the Gobi, add another dimension to the fossil record. Some of the eggs contain small embryonic skeletons of the bird Gobipteryx, and others preserve the skeletal remains of a small embryonic dinosaur. In some places, several nests may be clustered on a hillside, and we infer that these nests mark a congregation of dinosaurs, much like a colony of seabirds today.

At Tugrugeen, we found 12 jumbled skeletons of Protoceratops on a flat not much larger than a putting green. A Sino-Canadian team also reported such accumulations of Protoceratops in Cretaceous rocks of northern China.

The Protoceratops sample includes several growth stages, providing a glimpse of this largely unknown aspect of dinosaur biology. Adults typically measure two meters long; in 1994 our team recovered some Protoceratops less than nine centimeters long. These skeletons are obviously those of very young individuals, possibly newborns.

As we make such discoveries, however, the picture of dinosaur life that emerges becomes more complex. Because Protoceratops is the most common dinosaur fossil in the region, paleontologists have long assumed that the many shells and egg aggregates found at the Flaming Cliffs and elsewhere belong to it. Yet evidence for this supposition has been unsatisfactory. None of the hundreds of dinosaur eggs collected have clearly identifiable Protoceratops embryos within them. Even the tiny skulls of Protoceratops we discovered cannot be positively linked with an egg of a particular type.

A find from Ukhaa Tolgod suggests that this assumption may have been wrong. The examination of a clutch of eggs containing dinosaur embryos found on our first day there revealed that an oblong, somewhat wrinkly egg usually attributed to Protoceratops held a nearly perfect oviraptorid skeleton. It appears likely that many of the eggs found at Ukhaa Tolgod (and possibly elsewhere) belong to these small carnivores rather than to the parrot-beaked, herbivorous Protos.

The Ukhaa Tolgod “nest” also contained two tiny skulls of a troodontid (possibly Byronosaurus) in the clump of eggs; bits of oviraptorid eggshell were associated with their bones. This curious coincidence of eggs, an oviraptorid embryo, and two very young or newborn troodontids has several plausible explanations.

Perhaps the young troodontids were honing their skills at an early age by raiding dinosaur nests. Alternatively, the parent oviraptorid may have been feeding the troodontids to her offspring, or the troodontids might have been interlopers, their eggs placed in the oviraptorid nest in much the same way that cuckoo birds place their eggs in the nests of other bird species.

This discovery puts an ironic twist on nomenclatural history. The Andrews expeditions applied the name Oviraptor to a skeleton at the Flaming Cliffs because it was found atop a clutch of eggs. They assumed that the eggs belonged to the common Protoceratops and that Oviraptor (literally, “egg hunter”) was raiding a nest. Our find demonstrates that Oviraptor may not have been devouring eggs but rather incubating them.

Time Eludes
Nesting sites and skeletons of birds, dinosaurs, mammals and other vertebrates all make up a fairly detailed picture of life in the Gobi during the Late Cretaceous. The evidence contributed by the Mongolian Academy–American Museum expeditions has been gathered by logging thousands of kilometers over a wide stretch of the Gobi rather than concentrating for a prolonged period on a single location or on a few sites. This method not only increases the chance of finding new fossil sites, it conveys a better sense of the rock sequence through comparison of fossil-bearing strata over a broad area. Thus, we can try to determine whether assemblages of animals and sediments representing a particular environment and time interval are widespread or confined to isolated outcrops.

For example, paleontologists have generally believed that the community of fossils in the Djadokhta Formation (a Central Gobi bed of brilliant red sandstones named for the Flaming Cliffs) is slightly older than that of the Barun Goyot Formation (which gets its name from an ancient settlement in the Nemegt Valley) in the western Nemegt. Both our findings at Ukhaa Tolgod and our broad survey, however, suggest that the two formations preserve contemporary, virtually identical fauna. We found an extension of this community in the magnificent red and vermilion beds of Khermeen Tsav, an isolated set of badlands in the arid desert west of the Nemegt region that strongly resembles the canyon lands of southern Utah.

We have also found fossils from the Djadokhta community, including the familiar Protoceratops, in an area called Khugene Tsavkhlant, near the eastern railway. These discoveries are particularly significant because the sandstones there appear to be the result of stream or river action, a situation more typical of North American sites than of the Gobi. It is slowly becoming clear that the animal community once thought to be localized at the Flaming Cliffs may have occupied a range of habitats.

The wide geographical separation of many sites, however, impedes comparisons. In addition, Gobi rock sequences are entirely sedimentary, without even traces of volcanic rocks. Thus, geologists cannot determine the age of these strata by analyzing their proportions of radioactive isotopes. Estimates of the age of various formations must rely on the similarity of the vertebrates to those of reference faunas on other continents and on correlations with invertebrate fossils from Cretaceous marine rocks in Central and East Asia.

In yet another twist, the rocks of the Gobi appear to be missing precisely those strata that currently hold the greatest public interest: no sections found thus far include the Cretaceous-Paleogene boundary, when the dinosaurs became extinct. Although the Gobi is richly endowed with early Paleogene mammal faunas, there seems to be a gap of at least several million years between these and the Late Cretaceous dinosaur faunas. Whatever cataclysm wiped out the dinosaurs, its mark on Central Asia has apparently been erased. If a continuous sequence could be located somewhere in the desert's vastness, it would make a formidable contribution to our knowledge concerning the dinosaur extinction and the subsequent rise of mammals.

The notion of finding the mass extinction in the Gobi is not just wishful thinking. Satellite navigation has already made a tremendous difference in the effectiveness of our work. We can plot the precise location of fossil sites and the routes that lead to them. We have also used satellite images as a prospecting tool. After we returned from Ukhaa Tolgod in 1993, Evan Smith, then at the Yale University Center for Earth Observation, enhanced red and brown spectral bands on computer-based satellite images by matching colors from photographs of the rocks there. The result is a map that shows with high precision the extent and contours of fossil-bearing strata.

During the 1994 season, we used these images as a field guide and simply drove to the latitude and longitude of a telltale cluster of red pixels, and some of these spots proved productive. We now have a useful paleontological atlas in a region where detailed topographic or geologic maps are virtually lacking. We also have something that Andrews lacked for his most important but serendipitous discoveries: a fairly decent road map of the Gobi.

Despite our new technology and the decades of insights into the evolution of vertebrates, exploration of the Gobi has much the same quality that Andrews experienced 92 years ago. The Flaming Cliffs we encountered on that first joyful day in 1990 were as Andrews described them—imposing, brilliant red in color and replete with fossils. Sandstorms that engulfed the 1920s expeditions returned to wreak havoc on our fragile campsites.

When the sandstorms clear, one can see from the top of the cliffs the mauve, furrowed mountains of the Gurvan Saichan. Beyond the mountains are hundreds of square kilometers of fossil-rich badlands whose existence Andrews could only have imagined.

By Mark Norell and Michael J. Novacek

Many discoveries have been made since we wrote our 1994 article on the Gobi fossil hunt. Our summer 2013 expedition marked the 24th consecutive year of joint journeys by the Mongolian Academy of Sciences and the American Museum of Natural History. During the past two decades the important Ukhaa area has continued to yield many new and exciting fossils, including those in older rocks, in Lower Cretaceous beds in the north-central Gobi and, between 2002 and 2009, in Cretaceous sequences in the poorly explored reaches of the eastern Gobi near the Trans-Siberian Railway. This cumulative effort has revealed a broader and more graphic picture of life and death in Central Asia between 100 million and 75 million years ago.

Ukhaa Tolgod remains as perhaps the most significant discovery. It is a small, shallow drainage area of about four square kilometers, bordered by a serpentine line of low bluffs and adjacent flats where red rocks of the Djadokhta Formation are exposed. Geologic work has prompted reinterpretation of the accepted view that animals there were buried by gargantuan sandstorms. Based on detailed studies of sediments, our geologists David Loope and Lowell Dingus noted that the fossils at Ukhaa Tolgod are mostly restricted to beds that have no structure, suggesting that the dunes were stationary and possibly crowned with plants whose roots disrupted sand layers below the surface. This process of bioturbation was aided by the burrowing of small subterranean invertebrates, such as worms and insects, as well as larger burrowing mammals and lizards.

Loope and Dingus also observed that Ukhaa fossil beds contain streamers of pebbles too big to be windblown. Moreover, the sand has a high clay content, and the Mesozoic dunes are intercalated with water-resistant caliche layers. Hence, the dunes could not drain during heavy rainstorms; they instead acted as giant sandy sponges. At some point they became oversaturated and collapsed in large debris flows that covered sedentary animals, such as nesting dinosaurs, along with more active animals, making for phenomenal fossil preservation.

Many dinosaurs, lizards and mammals have been found at Ukhaa Tolgod. Most of them are fairly small, but there are hints (such as shed dinosaur teeth and footprints) that very large dinosaurs at least passed through. Some of the most common Ukhaa dinosaurs, such as ankylosaurs, were 4.5 meters long, and adult oviraptorids were a respectable three meters long. The remarkable dinosaurs also include groups of immature armored ankylosaurs, which may well have been gregarious because multiple skeletons are often found together.

Some of the most dramatic specimens are of the oviraptorid Citipati, which have been recovered sitting on their nests of eggs. Named after the Himalayan protector god of the funeral pyre, the creatures are the first nonavian dinosaurs known to exhibit avianlike brooding behavior. Additional specimens of the bizarre dinosaur Shuvuuia (previously called Mononykus) indicate that this animal is not a close relative of early birds but more primitive. Surprisingly, however, with one Shuvuuia specimen we found structural and biochemical evidence of feathers.

More than 600 mammal skulls, many with skeletons, have been recovered. This rich collection is critical to understanding the origins of modern mammal groups. Noteworthy is a form we named Ukhaatherium, which resembles the shrewlike species of the modern placental mammals but also retains primitive features such as the splintlike epipubic bones extending from the pelvis. The site has also yielded some of the best skeletons of Deltatheridium, an early relative of the marsupials, a group that includes opossums and kangaroos. Forms such as Deltatheridium and even more primitive species from the Early Cretaceous of northern China suggest that important branching events leading to the marsupial lineage occurred in the Mesozoic of Central Asia.

Between 2002 and 2009 we concentrated our efforts in the eastern Gobi, near our field operations of 1991. Additional study at Khugene Tsavkhlant, where we had earlier reported the ubiquitous Protoceratops, has indicated that this form is a much more primitive relative. We have also found interesting remains of mammals with low-crowned teeth. Right now we think this group is a distinct lineage that branched off just before the radiation of the modern placental mammals. In more recent years we have even ventured to the far western areas of Mongolia near the Chinese border to explore much older Jurassic beds.

Research progress has been aided by the technological trappings of 21st-century field exploration. Global Positioning System units, satellite phones and satellite imaging, portable computers, and digital cameras are standard equipment. We even have a solar-powered refrigerator to make camp life a little easier. The Gobi will doubtless continue to yield spectacular finds.