At the edge of a sprawling grassland, a pair of hoofed grazers resembling horses, an antelopelike notoungulate and a ground sloth feed quietly, oblivious to their impending doom. Equally unaware are the chinchilla and the tiny, mouselike marsupial nibbling seeds nearby. Suddenly, one of the jagged, snow-covered volcanoes on the horizon explodes catastrophically, sending a flood of muddy ash down its steep slopes. Soon after, this roiling slurry bursts across the flatter lowlands, entombing the unsuspecting animals in its path.

As devastating as this volcanic torrent was for the creatures it buried, it would become a boon for paleontology. Tens of millions of years after the mammals' untimely deaths, the exhuming forces of mountain building and subsequent erosion exposed remnants of their fossilized skeletons to the light of day high in the Andes Mountains of central Chile. Our team discovered the first of these bones in 1988 while searching for dinosaur remains in an alpine valley of the Tinguiririca River, near the border with Argentina. The initial finding of mammal bones proved so fruitful that we have returned to the region nearly every year since. So far we have uncovered more than 1,500 fossils of ancient mammals from dozens of sites in the central Chilean Andes.

Painstaking laboratory analysis of our growing collection has yielded major revelations about the history of South America's ancient mammals. To our astonishment, the Chilean fossils range from 40 million to 10 million years old--much younger than anything we expected to find there. Indeed, many of the specimens represent the only mammal remains from segments of that time interval found anywhere in South America. Some of these unique fossils illuminate a previously opaque period in the history of the continent's native mammal lineages; others help to resolve long-standing debates about the origins of key immigrant groups. Together they have revised understanding of when certain ecosystems--and the mountains themselves--appeared in this part of the world.

Tantalizing Discovery
Most of what scientists know about South America's ancient mammals is based on clues unearthed in the continent's far southern reaches, mainly Patagonia. Those regions have abundant outcrops of typical fossil-bearing rocks--shale, sandstone and other hardened sediments from rivers and their floodplains. Before our first visit to Chile, researchers had not searched systematically for land animal fossils in that country's mountainous areas, because most rocks there are volcanic. (The standard assumption is that lava and other erupted materials are too hot and violent to preserve organic remains.)

We decided to take a chance that the Tinguiririca Valley might harbor fossils when we learned of a report about dinosaur footprints. The rocks were the right age--geologists then assumed that most rocks along the main spine of the Chilean Andes dated back at least 65 million to 100 million years, to the latter part of the Mesozoic era, when dinosaurs reigned supreme. We knew that any sediments preserving footprints might also contain bony remnants of the track makers. If we were extremely lucky and kept our eyes close to the ground, we might even find a fossil of one of the dinosaurs' tiny mammalian contemporaries, which were no bigger than shrews.

On the last day of a one-week reconnaissance trip in 1988, our team of four split up to prospect the precipitous slopes flanking each side of the Tinguiririca River. Almost immediately, the pair working north of the river reached the layer of ancient sediments that bore the dinosaur tracks, then continued up the valley in search of more potential fossil-bearing deposits. To their dismay, however, the only fossils they recovered were from fish, ammonites and other ocean-dwelling creatures--no reptiles or mammals. Meanwhile the team members working on the south side of the river were having a similarly frustrating day. Late in the afternoon, though, their spirits soared when they spied a few fossilized scraps of bone and teeth eroding out of a large patch of reddish-brown volcanic sediments nearly 1,000 meters above the valley floor. A closer look revealed that the fossils were land-dwelling vertebrates about the size of a small horse.

At first, we tried to shoehorn these fragments into the prevailing view about the age of the rocks--animals this size must be peculiar dinosaurs or other odd Mesozoic beasts. But the complex, differentiated teeth with the high-crowned, flat-topped and multifaceted molars that are unique to some mammals told a different story. These mammals were clearly too large--and too advanced--to have lived anytime before about 50 million years ago. Apparently geologists had been a long way off in their estimate of the age of these rocks. Indeed, later analyses confirmed that the new fossils came from well within the Cenozoic era, the ongoing period of the earth's history that began when the nonavian dinosaurs went extinct 65 million years ago. (Birds are now known to be theropods, thereby representing a living group of dinosaurs.)

Island Menagerie
Finding fossils of any kind would have been great news for us. That they were mammals--and unexpectedly recent ones at that--was more than enough motivation to focus our next field season on that single area. We returned to the Tinguiririca Valley in the austral summer of 1989, after the high mountain snows melted enough for local authorities to reconstruct the small access road, which washes out almost every spring. This time we reached the fossil site on a clear, sunny January morning with a crew of seven scientists and complete ?expedition gear. Quickly unloading the pack animals, we set up camp near a small creek and started hunting.

To our delight, exquisite shards of bones and teeth popped into view within minutes of our beginning to comb the hillside. Protruding out both ends of one potato-size nodule of rock was a skull that was unmistakably mammalian, as indicated by, among other characteristics, its single pair of lower jawbones. (The lower jaws of reptiles are made up of many separate bones.) Later, we would formally describe that creature as a new species of notoungulate, a diverse group of hoofed herbivores that ranged from rabbit- to hippo-size and became extinct less than one million years ago. This new species probably looked a bit like an antelope. The teeth we found the previous year turned out to be from a rhinolike notoungulate. In all, during our first three seasons of fieldwork in Tinguiririca, we would turn up more than 300 specimens, includ?ing marsupials, an early sloth, arm?adillos and a chinchillalike rodent.

The full significance of our finds would take years to unfold, but we knew right away that we were onto something big. These new fossils clearly had much to say about the history of South America's distinctive living mammals--sloths, monkeys, anteaters and chinchillas among them. The ancestors of these animals, including many of the new creatures we were finding in the Tinguiririca Valley, were among those that evolved when South America was an island continent. For most of the past 80 million years, following the breakup of the supercontinent Pangaea and its southern portion, Gondwana, tectonic plate motions kept South America separated from other landmasses. This period of isolation fostered the evolution of native mammals uniquely adapted to that island's conditions and every bit as bizarre as those indigenous to modern islands such as Australia (like the platypus and koala) and Madagascar (famous for its lemurs). Unusual predecessors of South America's modern groups include hopping marsupials; saber-toothed marsupial "pseudocats"; armadillo cousins equipped with massive, spike-studded tail clubs; bear-size rodents; sloths as big as elephants; and sloths that swam in the sea.

Knowledge of the ancestors of South America's living mammals had been gleaned from earlier fossil discoveries in Patagonia and elsewhere, but critical information about many of these forebears remained elusive. Paleontologists knew, for instance, that sloths and anteaters got their start prior to 40 million years ago, as did several exotic lineages that are now extinct (including certain marsupials and notoungulates). But no fossils representing the transition to the second phase of South American mammal history--from about 40 million to 30 million years ago--had ever been discovered. Most exciting for us during those first few years, then, was the growing realization that the animals we found at Tinguiririca lived during this period of previously unknown history.

Paleontologists had suspected that during this mysterious gap in the fossil record many of the continent's unique lineages underwent explosive diversification. Indeed, our collection includes the earliest record of several groups of notoungulates and represents at least 25 mammal species, nearly all of them new to science. The era also saw the arrival of rodents and primates, neither of which were among South America's original mammal inhabitants.

Seafaring Rodents
One of the most significant of our Tinguiririca discoveries was a fossil of the earliest known South American rodent, a find that lends powerful evidence to a debate over the origins of today's living capybaras and chinchillas. Known as caviomorph rodents, these creatures and their immediate kin make up South America's most ancient rodent lineage (and are distinct from the younger rodent line of rats, mice and related creatures that arrived from the north about 3.5 million years ago, when the Isthmus of Panama first reconnected the two Americas). Paleontologists agreed that the first caviomorph rodents arrived sometime within the broad time span between 55 million and 25 million years ago, while South America was still an island. A few younger caviomorph fossils hinted that the predecessors came from Africa, but many researchers found it easier to imagine that the immigrant rodents made the shorter trip from North America, possibly via a chain of Caribbean islands.

To help settle the debate, we compared the anatomical details of the Tinguiririca animal to rodent remains found elsewhere in the world. Most informative was the shape of the tiny teeth still rooted in the lower jaw (the upper jaw and molars have yet to be found). That shape implied that the Tinguiririca animal's upper molars had five distinct crests--as did the upper molars of African rodents from the same period. In contrast, North America's ancient rodent species had only four crests on their upper molars. These comparisons strongly suggest that the Tinguiririca rodent is more closely related to animals in Africa. The absence of plausible caviomorph forebears in older North American fossil beds also supports the out-of-Africa theory.

Presumably the original caviomorph colonists traveled from Africa to South America on floating logs or other rafts of vegetation--scientists' best guess for how various unusual plants and animals made their way to many geographically isolated regions [see "Madagascar's Mesozoic Secrets," by John J. Flynn and André R. Wyss; Scientific American, February 2002]. The idea of such an incredible transoceanic journey may seem far-fetched, but it is more plausible in the context of the global environment before about 32 million years ago. At that time, the South Atlantic was only about 1,400 kilometers at its narrowest point--half as wide as it is today--and east-to-west ocean currents in the tropics were strengthening sporadically.

These conditions would have permitted a journey in about two weeks, and the animals may have gone into torpor (inactivity and greatly reduced metabolism during times of stress). Moreover, sea level was dropping at the time (because of the formation of ice sheets on and around Antarctica), so one or more volcanic "stepping stone" islands, now submerged, may have made the crossing easier.

Emerging Ecosystems
Using a new and extremely precise dating method that analyzes trace amounts of argon gas trapped inside crystals of the fossil-bearing rocks, we determined that the rodents and other Tinguiririca mammals were between 33 million and 31.5 million years old. Growing Antarctic ice sheets and other phenomena indicate that global climate at that time was becoming cooler and drier. Knowing that a major shift in climate occurred just as the Tinguiririca mammals were thriving led us to test whether the animals and their environment might have been responding to those changes.

Several lines of indirect evidence have enabled us to reconstruct the habitat of the Tinguiririca mammals even though we have never found plant fossils from the same rocks. Our early analysis of their teeth told us that the Tinguiririca animals must have lived in a very different ecosystem than their immediate forebears did. Most of the older South American mammals known from about 65 million to 34 million years ago were herbivores that browsed on typical forest greenery, such as tree leaves and herbs. (Indeed, plant fossils do confirm that lush forests probably covered much of the continent during that period.) Mammals that eat these soft foods, including humans, typically have low-crowned teeth with a thin cap of protective enamel covering each tooth only to the gum line.

In sharp contrast, most of the Tinguiririca herbivores possessed extremely high crowned teeth with enamel extending past the gum line nearly to the root tips, a condition known as hypsodonty. The additional enamel (much harder than the dentin in tooth interiors) makes hypsodont teeth more wear-resistant than low-crowned teeth are. The Tinguiririca herbivores almost certainly evolved such teeth in response to abrasive particles within the foods they ate, as did cows, antelopes, horses and other animals that eat gritty grasses in open prairies and savannas elsewhere in the world. It is also significant that two thirds of species of the entire Tinguiririca fauna were hypsodont. The proportion of hypsodont taxa relative to other dental types generally increases with the amount of open habitat, and the Tinguiririca level of hypsodonty surpasses even that observed for mammals living in modern, open habitats such as the Great Plains of central North America.

The findings implied that the Tinguiririca herbivores grazed in open grasslands rather than in forests, but the teeth are not alone in pointing to this conclusion. Our former graduate student Darin Croft, now a professor at Case Western Reserve University, provided two other independent deductions about the annual rainfall and vegetation that dominated the ancient Tinguiririca ecosystem. Statistical analyses of the number of species within different body-size categories and their ecological attributes--so-called cenogram and macroniche analyses--revealed that the Tinguiririca animals most closely resembled modern fauna that live in dry grasslands with patchy woodlands, such as parts of the savanna of Africa or the Caatingas and Chaco habitats of South America.

Our conclusion that the ancient Tinguiririca habitat was open, relatively dry and contained abundant grasses came as a bit of a surprise, considering that all previous evidence suggested that the first open grasslands on other continents did not appear until 18 million years ago. The apparent emergence of the Tinguiririca grasslands some 15 million years earlier could have resulted from the global trend toward aridity and cooling around that time--perhaps accentuated by the rain shadow cast by the rising Andes Mountains. Grasslands would have been better suited to a cooler, drier climate than the lush forests of previous millennia were. So far, though, suspicion that grasslands resulted from global cooling needs more scrutiny. Further testing a direct causal connection is an avenue of future research.

No Monkeying Around
Having uncovered such a rich trove of paleontological and environmental information at several sites in the Tinguiririca Valley, we began to consider whether this part of the Andes was a paleontological "one-hit wonder." Soon after, an unexpected obstacle led us to an answer. In the spring of 1994 reconstruction of the road leading into the Tinguiririca Valley was especially slow, but we did not learn of this impediment until arriving on the scene. Turning frustration into an opportunity, we began explorations outside the valley.

We searched other major valleys sporting outcrops of the same volcanically derived sedimentary rocks, which are exposed across thousands of square kilo?meters of mountainous terrain. Over the ensuing years of fieldwork, we determined that fossil mammals were indeed not restricted to the Tinguiririca Valley and that the volcanic torrent that inundated that ancient landscape was not a one-time, isolated cataclysm. Rather, seen over a time frame of millions of years, such devastating events occurred with some frequency. Each time, countless layers of material from additional eruptions buried the older deposits (and the bones they contained) ever more deeply. Ultimately, this layered stack of sediment (now transformed into rock) and lavas measured upward of three kilometers thick. Later, collision of converging tectonic plates squeezed this unruly pile skyward.

Our ongoing analyses of the numerous faunas, which range between 10 million and 40 million years in age, are revealing fresh insights into the region's history. One of our most significant newer finds--from a site about 100 kilometers north of Tinguiririca, in the drainage basin of the Cachapoal River--is the most complete skull of an early New World monkey yet discovered. The five-centimeter-long skull, with both eye sockets and every tooth in the upper jaw intact, came from a petite monkey weighing about one kilogram at the most. Named Chilecebus carrascoensis, this creature resembled modern New World monkeys, such as marmosets and tamarins. As with the caviomorph rodents, experts had long debated whether New World monkeys originated in North America or Africa. But anatomical details of the Chilecebus skull and teeth argue for its common heritage with a group of primates originating in Africa. Like the caviomorph rodents, it seems the ancestors of Chilecebus somehow made the Atlantic crossing from Africa.

Beginning with the Tinguiririca fauna and culminating with the New World monkey and other ongoing discoveries across central Chile, volcanic deposits once ignored for fossils have turned out to contain superbly preserved bones and are now recognized as a premier archive of South American mammal evolution. Over the years, we have developed a keen sense of what auspicious rocks look like, sometimes even being able to spot them from many kilometers away. These fossils are hard-won, though, given the precipitous topography and the remoteness of many of the localities. Some sites lie within a few kilometers of gravel roads or dirt trails, but most can be reached only by long approaches on foot, horse or even helicopter. We jokingly refer to "Andy's Rule" (dubbed in honor of Wyss), which says that the difficulty of getting there is proportional to the quantity and quality of fossils we will find.

Together the Cenozoic fossils of the Chilean Andes are helping to elucidate mammalian evolution and environmental transformations in South America, a continent whose long history of isolation represents a splendid natural experiment for investigating large-scale evolutionary phenomena.