Three weeks into our first fossil-hunting expedition in Madagascar in 1996, we were beginning to worry that dust-choked laundry might be all we would have to show for our efforts. We had turned up only a few random teeth and bones—rough terrain and other logistical difficulties had encumbered our search. With our field season drawing rapidly to a close, we finally stumbled on an encouraging clue in the southwestern part of the island. A tourist map hanging in the visitor center of Isalo National Park marked a local site called “the place of animal bones.” We asked two young men from a neighboring village to take us there right away.
Our high hopes faded quickly as we realized that the bleached scraps of skeletons eroding out of the hillside belonged to cattle and other modern-day animals. This site, though potentially interesting to archaeologists, held no promise of harboring the much more ancient quarry we were after. Later that day another guide, accompanied by two dozen curious children from the village, led us to a second embankment similarly strewn with bones. With great excitement we spotted two thumb-sized jaw fragments that were undoubtedly ancient. They belonged to long-extinct, parrot-beaked cousins of the dinosaurs called rhynchosaurs.
The rhynchosaur bones turned out to be a harbinger of a spectacular slew of prehistoric discoveries. Since then, the world's fourth-largest island has become a prolific source of new information about animals that walked the land during the Mesozoic era, from 252 million to 66 million years ago, when both dinosaurs and mammals were making their debut. We unearthed the bones of what seemed like primitive dinosaurs, which we suspected were older than any found previously. We have also stirred up controversy with the discovery of a shrewlike creature that seems to defy a prominent theory of mammalian history by being in the “wrong” hemisphere. These specimens, and others collected over five field seasons, enabled us to begin painting a picture of ancient Madagascar and shaped our strategy for a sixth expedition in 2003.
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Much of our research over the past three decades has been aimed at unraveling the history of land-dwelling animals on the southern continents. Such questions have driven other paleontologists to fossil-rich locales in South Africa, Brazil, Antarctica and India. Rather than probing those established sites for additional finds, we were lured to Madagascar: the island embraces vast swaths of Mesozoic- age rocks, but until recently only a handful of terrestrial vertebrate fossils from that time had been discovered there. Why? We had a hunch that no one had looked persistently enough to find them.
Persistence became our motto as we launched our 1996 expedition. Our team consisted of a dozen scientists and students from the U.S. and the University of Antananarivo in Madagascar. In addition to scientific and educational benefits, our partnership with the country's leading university facilitated the acquisition of collecting and exporting permits—requisite components of all paleontological fieldwork. Before long, however, we ran into logistical obstacles that surely contributed to earlier failures to find ancient fossils on the island. Mesozoic deposits in western Madagascar are spread over an area roughly the size of California. Generations of oxcarts and foot travel have carved the only trails into remote areas, and most of them are impassable by even the brawniest four-wheel-drive vehicles. We had to haul most of our food, including hundreds of pounds of rice, beans and canned meats, from the capital. Fuel shortages sometimes seriously restricted mobility, and our work was even thwarted by wildfires, which occur frequently and rage unchecked. New challenges arose unexpectedly, requiring us to adjust our plans on the spot.
Ancient Luck
Perhaps the most daunting obstacle we faced in prospecting such a large region was deciding where to begin. Fortunately, we were not planning our search blindly. The pioneering fieldwork of geologists such as Henri Besairie, who directed Madagascar's ministry of mines during the mid-1900s, provided us with large-scale maps of the island's Mesozoic rocks. From those studies we knew that a fortuitous combination of geologic factors had led to the accumulation of a thick blanket of sediments over most of Madagascar's western lowlands—and gave us good reason to believe that ancient bones and teeth might have been trapped and preserved there.
At the dawn of the Mesozoic era 252 million years ago, it would have been possible to walk from Madagascar to almost anywhere else in the world. All of the planet's landmasses were united in the supercontinent Pangaea, and Madagascar was nestled between the western coast of what is now India and the eastern coast of present-day Africa [see map above]. The world was much warmer than at present—even the poles were free of ice. In the supercontinent's southern region, Gondwana, enormous rivers coursed into lowland basins that would eventually become the Mozambique Channel, which today spans the 250 miles between Madagascar and eastern Africa.
These giant basins represent the edge of the geologic gash created as Madagascar began pulling away from Africa more than 240 million years ago. This seemingly destructive process, called rifting, is an extremely effective way to accumulate fossils. (Indeed, many of the world's most important fossil vertebrate localities occur in ancient rift settings.) The rivers flowing into the basins carried with them mud, sand, and occasionally the carcasses or bones of dead animals. Over time the rivers deposited this material as a sequence of vast layers. Continued rifting and the growing mass of sediment caused the floors of the basins to sink ever deeper. This depositional process persisted for nearly 100 million years, until the basin floors thinned to the breaking point and molten rock ascended from the planet's interior to fill the gap as new ocean crust.
Up to that point nature had afforded Madagascar three crucial ingredients required for fossil preservation: dead organisms, holes in which to bury them (rift basins), and material to cover them (sand and mud). But special conditions were also needed to ensure that the fossils were not destroyed before they could be found millions of years later. Again, geologic circumstances proved fortuitous. As the newly separated landmasses of Africa and Madagascar drifted farther apart, their sediment-laden coastlines rarely experienced volcanic eruptions or other events that could have destroyed buried fossils. Also key for fossil preservation is that the ancient rift basins ended up on the western side of the island, which today is dotted with dry forests, grasslands and desert scrub. In a more humid environment, such deposits would have eroded away or would be hidden under dense vegetation like the kind that hugs much of the island's eastern coast.
Initially Madagascar remained attached to the other Gondwanan landmasses: India, Australia, Antarctica and South America. It did not attain islandhood until it split from India about 90 million years ago. Sometime since then, the island acquired its suite of bizarre modern creatures, of which lemurs are the best known. For more than a century, researchers have wondered how long these modern creatures and their ancestors have inhabited the island. Illuminating discoveries by another team of paleontologists indicate that almost all major groups of living vertebrates arrived on Madagascar since sometime near the end of the Mesozoic era 66 million years ago [see box on page 36]. Our own probing has focused on a more ancient interval of Madagascar's history—the first two periods of the Mesozoic era.
Pay Dirt
One of the joys of working in little-charted terrain has been that if we manage to find anything, its scientific significance is virtually assured. That is why our first discoveries near Isalo National Park were so exciting. The same evening in 1996 that we found the rhynchosaur jaw fragments, then University of Antananarivo student Léon Razafimanantsoa spotted the six-inch-long skull of another interesting creature. We immediately identified the animal as a peculiar plant eater, neither mammal nor reptile, called a traversodontid cynodont [see illustration on page 35].
The rhynchosaur jaws and the exquisite traversodontid skull—the first significant discoveries of our ongoing U.S.-Malagasy project—invigorated our expedition. The first fossil is always the hardest one to find; now we could do the detailed collecting necessary to begin piecing together an image of the past. The white sandstones we were excavating had formed from the sand carried by the rivers that poured into lowlands as Madagascar unhinged from Africa. Within these prehistoric valleys rhynchosaurs and traversodontids, both four-legged creatures ranging from three to 10 feet in length, probably grazed together much the same way zebras and wildebeests do in Africa today. The presence of rhynchosaurs, which are relatively common in coeval rocks around the world, narrowed the date to sometime within the Triassic period (the first of three Mesozoic time intervals), which spans from 252 million to 201 million years ago. And because traversodontids were much more diverse and abundant during the first half of the Triassic than during the second, we thought initially that this scene played out before about 230 million years ago.
During our second expedition, in 1997, a third type of animal challenged our sense of where we were in time. Shortly after we arrived in southwestern Madagascar, one of our field assistants, a local resident named Mena, showed us some bones that he had found across the river from our previous localities. We were struck by the fine-grained red rock adhering to the bones—everything we had found until that point was buried in the coarse white sandstone. Mena led us about half a mile north of the rhynchosaur and traversodontid site to the bottom of a deep gully. Within a few minutes we spotted the bone-producing layer from which his unusual specimens had rolled. A rich concentration of fossils was entombed within the three-foot-thick layer of red mudstones, which had formed in the floodplains of the same ancient rivers that deposited the white sands. Excavation yielded about two dozen specimens of what seemed at first to be dinosaurs. Our team found jaws, strings of vertebrae, hips, claws, an articulated forearm with some wrist bones, and other assorted skeletal elements. When we examined the jawbones more closely, we realized that we had uncovered remains of a new species that resembled an enigmatic animal from Morocco called Azendohsaurus, at the time considered a prosauropod dinosaur. (Prosauropods, which typically appear in rocks between 225 million and 190 million years old, are smaller-bodied precursors of the long-necked sauropod dinosaurs.)
Much to our surprise, however, as the other bones were later cleaned from the rock and as we found additional bones over subsequent field seasons, we realized we were dealing with a perplexing early lineage of reptiles rather than a true dinosaur. This mysterious animal has teeth and some skull features that look just like those of typical prosauropods, but the rest of its skeleton is strikingly more primitive than any known dinosaur. We have not yet completely analyzed the unexpected combination of bones and teeth, but this evolutionary chimera—mixing characteristics of distantly related reptile groups—has led us to question whether certain features long believed to characterize dinosaurs might have evolved much earlier. If so, the new Malagasy fossil, Azendohsaurus madagaskarensis, and other animals long assumed to be dinosaurs are likely something else entirely.
When we discovered that Azendohsaurus was foraging alongside rhynchosaurs and traversodontids, it became clear that we had unearthed a collection of fossils not known to coexist anywhere else. In Africa, South America and other parts of the world, traversodontids are much less abundant and less diverse once dinosaurs appear. Similarly, the most common type of rhynchosaur we found, Isalorhynchus, lacks advanced characteristics and thus is inferred to be more ancient. What is more, the Malagasy fossil assemblage lacks remains of several younger reptile groups usually found with the earliest dinosaurs, including the heavily armored, crocodilelike phytosaurs and aetosaurs. The mixture of ancient kinds of animals found alongside our enigmatic reptile, plus the lack of younger creatures, suggests that the Malagasy fossil deposit is as old as any dinosaur ever discovered, if not older.
Just one early dinosaur site—at Ischigualasto, Argentina—contains a rock layer that has been dated directly; all other early dinosaur sites with similar fossils are thus estimated to be no older than its radioisotopic age of about 228 million years. (Reliable radioisotopic ages are obtainable only from rock layers produced by contemporaneous volcanoes. The Malagasy sediments accumulated with no volcanoes nearby.) Based on the fossils present, we have tentatively concluded that our fossil-bearing rocks slightly predate the Ischigualasto time span. And because different major branches of the dinosaur evolutionary tree are already present at Ischigualasto, the common ancestor of all dinosaurs must be older still. Rocks from before about 245 million years ago have been moderately well sampled around the world, but none of them has yielded dinosaurs. That means the search for the common ancestor of all dinosaurs must focus on a relatively poorly known and ever narrowing interval of Middle Triassic rocks, between about 240 million and 230 million years old.
Mostly Mammals
Dinosaurs naturally attract considerable attention, being the most conspicuous land animals of the Mesozoic. Less widely appreciated is the fact that mammals and dinosaurs sprang onto the evolutionary stage at nearly the same time. At least two factors account for the popular misconception that mammals arose only after dinosaurs became extinct: Early mammals were all chipmunk-sized or smaller, so they don't grab the popular imagination. In addition, the fossil record of early mammals is sparse. To our delight, Madagascar has once again filled in two mysterious gaps in the fossil record. The traversodontid cynodonts from the Isalo deposits reveal new details about close mammalian relatives, and a younger fossil from the northwestern side of the island poses controversial questions about where and when a key advanced group of mammals got its start.
The Malagasy traversodontids, the first known from the island, include some of the best-preserved representatives of early cynodonts ever discovered. (Cynodontia is a broad group of land animals that includes mammals and their nearest relatives.) Accordingly, these bones provide a wealth of new anatomical information. These cynodonts are identified by, among other features, a simplified lower jaw dominated by a single bone, the dentary. Some specimens include both skulls and skeletons. Understanding the complete morphology of these animals is crucial for resolving the complex evolutionary transition from the large cold-blooded, scale-covered animals with sprawling limbs (which dominated the continents prior to the Mesozoic) to the much smaller warm-blooded, furry animals with an erect posture that are so plentiful today.
Many kinds of mammals, with many anatomical variations, now inhabit the planet. But they all share a common ancestor marked by a single, distinctive suite of features. To determine what these first mammals looked like, paleontologists must examine their closest evolutionary relatives within the Cynodontia, which include the traversodontids and their much rarer cousins, the chiniquodontids (also known as probainognathians), both of which we have found in southwestern Madagascar. Traversodontids almost certainly were herbivorous because their wide cheek teeth are designed for grinding—evidence of which we and our colleagues documented with scanning electron microscope analyses of microwear features on the teeth. The chiniquodontids, in contrast, were undoubtedly carnivorous, with sharp, pointed teeth. Most paleontologists agree that some chiniquodontids share a more recent common ancestor with mammals than the herbivorous traversodontids do. The chiniquodontid skulls and skeletons we found in Madagascar will help reconstruct the bridge between early cynodonts and true mammals.
Not only are Madagascar's Triassic cynodonts among the best preserved in the world, they also sample a time period that is poorly known elsewhere. The same is true for the youngest fossils our expeditions have uncovered—those from a region of the northwest where the sediments are about 165 million years in age. (That date falls within the middle of the Jurassic, the second of the Mesozoic's three periods.) Because these sediments were considerably younger than our Triassic rocks, we allowed ourselves the hope that we might find remains of an ancient mammal. Not a single mammal had been recorded from Jurassic rocks of a southern landmass at that point, but this did nothing to thwart our motivation.
Once again, persistence paid off. During our 1996 field season, we had visited the village of Ambondromahabo after hearing local reports of abundant large fossils of the sauropod dinosaur Lapparentosaurus. Sometimes where large animals are preserved, the remains of smaller animals can also be found—though not as easily. We crawled over the landscape, eyes held a few inches from the ground. This uncomfortable but time-tested strategy turned up a few small theropod dinosaur teeth, fish scales and other bone fragments, which had accumulated at the surface of a small mound of sediment near the village. These unprepossessing fossils hinted that more significant items might be buried in the sediment below. We bagged about 200 pounds of sediment and washed it through mosquito-net hats back in the capital, Antananarivo.
During the subsequent years back in the U.S., while our studies focused on the exceptional Triassic material, the tedious process of sorting the Jurassic sediment took place. A dedicated team—Dennis Kinzig, Ross Chisholm and Warren Valsa, all then volunteers at the Field Museum in Chicago—spent many a weekend sifting through the concentrated sediment under a microscope in search of valuable flecks of bone or teeth. We did not think much about that sediment again until 1998, when Kinzig relayed the news that they had uncovered the partial jawbone of a tiny mammal with three grinding teeth still in place. We were startled not only by the jaw's existence but also by its remarkably advanced cheek teeth. The shapes of the teeth document the earliest occurrence of Tribosphenida, a group encompassing the vast majority of living mammals. We named the new species Ambondro mahabo, after its place of origin.
The discovery pushes back the range of this group of mammals by more than 25 million years and offers the first glimpse of mammalian evolution on the southern continents during the latter half of the Jurassic period. It shows that this subgroup of mammals may have evolved in the Southern Hemisphere rather than the Northern, as is commonly supposed. The information does not resolve the debate, but it does point out the precarious nature of long-standing assumptions rooted in a fossil record historically biased toward the Northern Hemisphere [see box on page 32].
Planning Persistently
Although our team has recovered a broad spectrum of fossils in Madagascar, scientists are only beginning to describe the Mesozoic history of the southern continents. The number of species of Mesozoic land vertebrates known from Australia, Antarctica, Africa and South America is probably on an order of magnitude smaller than the number of contemporaneous findings from the Northern Hemisphere. Clearly, Madagascar now ranks as one of the world's top prospects for paleontologists.
Often the most significant hypotheses about ancient life on earth can be suggested only after these kinds of new fossil discoveries are made. Our team's explorations provide two cases in point: some fossils long mistaken for early dinosaurs actually belong to a distantly related lineage of reptiles, and the existence of the tiny mammal at our Jurassic site implies that tribosphenic mammals may have originated in the Southern, rather than Northern, Hemisphere. The best way to bolster these proposals (or to prove them wrong) is to go out and uncover more bones. That is why the primary goal of our August-September 2003 expedition was the same as it had been for our previous five expeditions: find as many fossils as possible.
Our agenda typically includes digging deeper into known sites and surveying new regions, blending risky efforts with sure bets. No matter how carefully formulated, however, our plans are always subject to last-minute changes. In 2003 our most daunting challenge was the appearance of frenzied boomtowns.
During our first three expeditions, we never gave a second thought to the gravels that overlay the Triassic rock outcrops in the southwestern part of the island. Little did we know that those gravels contain sapphires. By 1999 tens of thousands of people were scouring the landscape in search of these gems. The next year all our Triassic sites fell within sapphire-mining claims. Paleontologists are now required to get permission for entry from both the claim holder and the government. In addition to that extra set of hurdles, the miners' presence has created security problems.
Even without such logistical obstacles, it would require uncountable lifetimes to carefully survey all the island's untouched rock exposures. But now that we have seen a few of Madagascar's treasures, we are inspired to keep digging.
