Greenhouse of the Dinosaurs: Evolution, Extinction, and the Future of our Planet

Most people think that science is about planning your research carefully to achieve some specific goal. They are often not tolerant about “pure research” that doesn’t have a specific conclusion in mind, but is focused on finding out general facts about nature, whether they have practical uses or not. Even the scientific funding agencies operate this way, rewarding research that is conventional and “more of the same,” but seldom funding research that is a speculative gamble. Again and again, talking heads on television or in Congress ridicule “pure research” that doesn’t have a specific practical goal or application. Narrow-minded and poorly educated people occasionally manage to interfere with the well-established scientific review process and shut down research they don’t like, even though it was approved by established scientists.

The sad irony of the concept that “science must be practical and useful” is that most of the greatest discoveries in science happen by accident. More often than not, scientists who find a crucial new piece of evidence are not looking for it, but rather searching for something else and make their great discovery without planning to. The term serendipity was coined to describe this phenomenon. It comes from an old Persian tale, “Three Princes of Serendip,” about princes who made discoveries unexpectedly. However, in the case of science, serendipity works most often when the researcher is prepared to see the implications of some new, unexpected development. As Louis Pasteur put it, “In the field of observation, chance favors only the prepared mind.”

Examples of accidental discoveries in science are legion, especially in chemistry. Alfred Nobel accidentally mixed nitroglycerin and collodium (gun cotton) and discovered gelignite, the key ingredient for his development of TNT. Hans Von Pechmann accidentally discovered polyethylene in 1898. Silly Putty, Teflon, Superglue, Scotchgard, and Rayon were accidents, as was the discovery of the elements helium and iodine. Among drugs, penicillin, laughing gas, Minoxidil for hair loss, the Pill, and LSD were discovered by accident. Viagra was originally developed to treat blood pressure, not impotence. Most of the great discoveries in physics and astronomy were unexpected, including the planet Uranus, infrared radiation, superconductivity, electromagnetism, X rays, and many others. Two Bell lab engineers discovered the cosmic background radiation from the Big Bang when they were trying to eliminate the noise from their newly developed microwave antennas. Among practical inventions, inkjet printers, corn flakes, safety glass, Corningware, and the vulcanization of rubber were accidents. Percy Spencer accidentally came across the principle of microwave ovens while testing a magnetron for radar sets and finding that the candy bar in his lab coat pocket had melted.

Likewise, geologists often find things they are not looking for. In 1855, J. H. Pratt and George Airy were doing routine surveying for the British government in northern India. They noticed that the plumb line under the surveying tripod was not as gravitationally attracted to the Himalayas as they had expected and eventually discovered the evidence for the deep crustal roots of mountains like the Himalayas. The marine geologists who mapped the magnetic anomalies on the seafloor were not looking for the crucial evidence that proved plate tectonics, but were simply doing routine data collection of magnetic, bathymetric, and oceanographic data as their ships undertook regularly scheduled voyages of discovery. Maurice Ewing, the founder of Lamont-Doherty Geological Observatory (now Lamont-Doherty Earth Observatory) of Columbia University, had a standing order that each ship would take a deep-sea core at the end of the day, no matter where they were, and many of those cores turned out to have crucial evidence for the history of oceans, climates, and the evolution of life.

I can cite such examples for many more pages, but the point is clear: science is not always predictable, and scientific research cannot be restricted to straightforward results that are expected when the study begins. Shortsighted people such as the right-wing radio hosts and the politicians who ridicule pure research must not be allowed to destroy our scientific curiosity and creativity, or our scientific discoveries will come to an end. Isaac Asimov said, “The most exciting phrase to hear in science, the one that heralds new discoveries, is not ‘Eureka!’, but ‘That’s funny. ’”

One particularly revealing example of serendipity in science occurred in Italy in 1978. Walter Alvarez, a young geologist who was working on the structural geology of Italy, was out mapping, measuring, and describing sections in the Apennine Mountains, the chain of mountains that form the “backbone” of Italy. (I knew Walter when I was a grad student at Lamont, and he was not yet famous, but just an ordinary structural geologist working in Italy.) Near the town of Gubbio, Italy, Walter came across a road-cut exposure (see the photograph that opens this chapter) that had thick limestones containing latest Cretaceous microfossils (as identified by Isabella Premoli-Silva), then another thick limestone with early Paleocene microfossils. In between them was a distinctive clay layer that marked the boundary between the Cretaceous and Tertiary, and thus one of the greatest mass-extinction horizons in Earth history. The Cretaceous–Tertiary (KT) boundary marked the extinction not only of the dinosaurs, but also of many marine organisms, including many types of plankton, marine reptiles, and the supremely successful ammonites, which looked something like the chambered nautilus and are related to squids and octopuses.

Walter knew the significance of the layer (the “prepared mind”) but took a sample of the boundary clay thinking it would tell him something about how this great mass extinction had occurred. When he took it home to Berkeley and showed it to his father, Nobel Prize–winning physicist Luis Alvarez, they both puzzled over a way to unlock its secrets. Luis hit on the idea that rare elements that rain down on Earth from cosmic dust, such as iridium, might be useful. If, on the one hand, the sample showed a low level of iridium, then it would represent a relatively fast rate of sediment accumulation. If, on the other hand, it showed a high level, it would suggest a long accumulation of cosmic dust.

When they measured the sample, the iridium level was way off the charts and far too abundant to be the product of slow accumulation of cosmic dust. Luis eventually hit on a model that might explain this high level: the impact of an asteroid 10 kilometers (6 miles) in diameter, which had scattered debris rich in iridium all over the world and blocked out the sunlight with its dust cloud, causing mass extinction. Walter, Luis, and the two nuclear chemists who analyzed the samples at Berkeley, Frank Asaro and Helen Michel, finally wrote up their outrageous idea, and it was published in Science in 1980 (Alvarez, Alvarez, et al. 1980). It has since become one of the most famous and cited discoveries made in geology in the past 30 years.

Their original scenario went something like this: 65 million years ago a giant asteroid about 10 to 15 kilometers (6 to 9 miles) in diameter plummeted to earth. It was traveling at cosmic speeds of 20 to 70 kilometers per second (45,000 to 156,000 miles per hour). Such an enormous mass traveling so fast packs an enormous amount of energy, approximately the equivalent of 100 million megatons of TNT, or more energy than all the nuclear weapons in the world at the peak of the Cold War. When the impact occurred, it would have generated a shock wave that should have leveled everything within 1,000 kilometers (620 miles) of the impact site, causing the world to burst into flames. In the imagined scenario, it would excavate a crater 15 to 20 kilometers (9 to 12 miles) deep and at least 170 kilometers (105 miles) wide. If the asteroid landed near or in the ocean, it would cause huge tsunamis. Approximately 100 cubic kilometers of rock and debris would blow from the crater and rise as high as 100 kilometers (62 miles) into the stratosphere. Most of this debris would fall back immediately, but some would generate a huge smoke and dust cloud that would blanket Earth for months and shut off all light to plants. In some scenarios, Earth would freeze over. In others, the impact would generate huge clouds of sulfuric acid that would be devastating for all of life.

After the 1980 paper announcing the iridium anomaly and the asteroid impact hypothesis, the scientific community was at first skeptical, and rightly so. Some wondered whether the iridium level was truly off scale and what the background level in the rest of the layers was like, so it took a great deal of work to establish that the KT layer was indeed anomalous. Some scientists suggested that it might be due to the properties of marine clays in the Gubbio boundary layer. Clays are notorious for their ability to concentrate unusual trace elements, so this criticism was legitimate. When the same iridium anomaly was discovered in terrestrial rocks near Hell Creek, Montana (figure 5.1), however, the marine clay concentration hypothesis was ruled out. There was also one embarrassing gaffe in the original research: some samples were contaminated by the iridium from the platinum band of a technician’s wedding ring. Even the high levels of the iridium in the boundary clay is less concentrated than your average piece of platinum-bearing jewelry.

By this point, many other scientists were jumping on the bandwagon of this exciting discovery. The extinction of the dinosaurs and the ammonites had long been one of the greatest unsolved puzzles of paleontology. Until 1978, no hard data existed that could test various wild speculations about supernovas, impacts, disease, sea-level retreat, and climatic change. The Alvarez hypothesis gave all sorts of people—such as geochemists, geophysicists, and planetary scientists with limited training in fossils and stratigraphy—an opportunity to plunge into the debate. As more and more people looked at KT boundary layers around the world, they found not only iridium, but also quartz grains that had undergone the effects of an enormous shock (previously known only from nuclear bomb craters), glassy blobs of crustal rock that had been melted and thrown into the atmosphere, and apparent tsunami deposits in the Gulf Coast and Caribbean. The impact bandwagon was rolling very fast among geochemists and planetary geologists, but not everyone bought into the idea.

The anti-impactors pointed out that there were other possible sources of iridium, such as deep-mantle-derived volcanoes like Kilauea. And the KT boundary was marked by one of the biggest mantle-derived flood lava eruptions in Earth history. Known as the Deccan traps, these eruptions are located in what is now western India and parts of eastern Pakistan (figure 5.2). They produced more than 10,000 cubic kilometers (2,400 cubic miles) of lava flows, with individual flows as thick as 150 meters (492 feet), and totaling at least 2,400 meters (7,874 feet) in thickness. Such enormous mantle-derived eruptions would have filled the atmosphere with thick clouds of ash-bearing iridium as well as with gases such as carbon dioxide that would have changed atmospheric and ocean chemistry.

Then yet another joker popped up in the deck. The latest Cretaceous was marked by a major drop in sea level, which drained the great epicontinental seas that once covered the High Plains from Hudson Bay to the Gulf of Mexico. The Cretaceous Interior Seaway had turned some places into a gigantic shallow ocean full of marine reptiles such as mosasaurs, plesiosaurs, gigantic turtles, huge fish longer than 7.0 meters (21 feet), gigantic clams 1.5 meters (5 feet) across, and an incredible array of ammonites and other mollusks. This marine ecosystem was supported by a huge bloom of plankton whose skeletons are so numerous that they form the Cretaceous chalk that is widespread from the White Cliffs of Dover to western Kansas. Clearly, the drop in sea level and the drying of this seaway devastated shallow marine organisms and had an indirect effect on land life as well.

As the debate raged on through the 1980s, the biggest problem for the impact advocates was the absence of a crater of the right age. Then in 1990 planetary geologist Alan Hildebrand looked up some old oil company reports written by Glen Penfield in 1980. The oil companies had been drilling in the northern Yucatán at a place called Chicxulub (a Mayan word), where they had found a huge gravity anomaly that suggested some massive structure beneath the jungle. Some of their drill samples produced unusual rocks, such as shattered gypsum and possible impact debris. But they didn’t find oil and weren’t interested in the newly announced KT impact hypothesis, so their reports were quietly shelved. Hildebrand’s “prepared mind,” though, saw these findings as the “smoking gun” because all the strongest evidence for impact was around the Caribbean and Gulf of Mexico. Sure enough, when the impact advocates drilled much deeper under the jungle floor, they found an impact crater that had been filled in and buried, complete with abundant impact debris and argon-argon dates (analyzed by my friend Carl Swisher) that fit perfectly for the KT boundary.

By 1990, the debate was no longer whether the impact occurred—that was clearly established. Nor were the Deccan traps under any question—these immense eruptions occurred at just the right time. Nor was there any doubt about the great drop in sea level. With all these catastrophes happening in a short time window at the end of the Cretaceous, it was clearly a bad time to be alive on planet Earth. So the real issue was, Which of these three possible factors was most important? Did the impact all by itself cause the extinction, as the impact advocates argue? Or were the Deccan volcanics more important, possibly combined with the effects caused by sea-level change? The only way to answer these questions is to look at the fossils themselves.

If we’re going to get beyond the name calling and wild speculation that has pervaded the research on the KT extinctions, we need to look closely at the evidence. Which organisms were affected? Which ones were not? Does the extinction show a gradual protracted pattern over the Late Cretaceous, or do most of the extinctions cluster right beneath the horizon bearing the impact debris and iridium? Fortunately, the intense interest in this extinction horizon has generated a great deal of research that answers just these kinds of questions. What do the fossils say?

First, let’s look at the marine realm (figure 5.3). Five groups of plankton have readily fossilizable skeletons and make up the bulk of our marine record. The smallest are the tiny algae known as coccolithophorids, which are surrounded by a series of button-shaped calcareous plates only a few microns across, known as coccoliths. Another kind of golden brown algae are the diatoms, which secrete tiny plates that look like delicate Petri dishes made of silica. Yet a third group of algae, the dinoflagellates, have organic-walled shells propelled by tiny flagella. Feeding on these algae are amoebalike protistans that secrete a skeleton (unlike the amoebas familiar from your pond). These shells include the multiple bubblelike chambers of the calcareous shells of the foraminifera and the delicate siliceous “Christmas ornament” shells of the radiolaria.

Analysis of the plankton gave mixed results (MacLeod et al. 1997; Popsichal 1997). The tiny coccolithophorids did experience a severe extinction (as befitting a plant that requires light), but other algae, such as the dinoflagellates and diatoms, did not. Impact advocates argue that dinoflagellates and diatoms could have survived in resting spores on the seafloor until the bad times ended, but this theory does not apply to the radiolaria, which also did not experience a significant extinction. In fact, the richness of radiolarians actually increased across the KT boundary, suggesting an improvement in oceanic circulation and productivity. Finally, the extinction in the foraminifera has been the most controversial over time. The bottom-dwelling benthic foraminifera show almost no extinction. Most micropaleontologists argue that the planktonic foraminifera suffered a significant and rapid extinction, but other researchers, such as Norman MacLeod and Gerta Keller (1995), argue that the extinction was protracted over 300,000 years around the KT boundary.

Up the food chain, the corals were in decline long before the KT event, and most “extinct” species of the latest Cretaceous reappeared in the Paleocene, so there is no clear evidence of a mass extinction of the coral reefs (Rosen and Turnsek 1989; Rosen 2000). Among the marine mollusks, there were the huge inoceramid clams and the cone-shaped rudistid clams that dominated the Cretaceous seafloor, but both groups were in decline long before the end of the Cretaceous and gone before the impact occurred (Kauffmann 1988; MacLeod 1994). The rest of the marine clams and snails show only a minor extinction, with about 35 percent of the snail species and 55 percent of the clam species dying out. Every study has shown that their extinction was gradual across the KT boundary (Hansen et al. 1987; Bryan and Jones 1989; Hansen, Farrell, and Upshaw 1989; Zinsmeister et al. 1989).

The most abundant group of mollusks to vanish were the ammonites, which had evolved and flourished after each of the previous mass extinctions, but did not survive the KT. Most studies have shown that they gradually disappeared through the Late Cretaceous, with only a few species surviving to witness the impact (Ward, Kennedy, and MacLeod 1991; Zinsmeister and Feldmann 1993). And their close relatives the nautiloids went right through the KT boundary with no extinction whatsoever and are still found in the South Pacific today. The squid-like belemnites, which once left thousands of shells shaped like large-caliber bullets, also declined throughout the later Cretaceous, with only one species around at the end to witness the impact (MacLeod et al. 1997).

There was an abrupt extinction in the brachiopods, or lamp shells (Surlyk and Johansen 1984), but not in their close relatives the bryozoans, or moss animals (Hakansson and Thomsen 1979). Nor was there much extinction in the echinoderms, including the crinoids (sea lilies), the sea stars, brittle stars, or the sea urchins and heart urchins (Birkelund and Hakansson 1982; Smith and Jeffrey 1998, 2000).

Finally, what about the marine vertebrates? More than 90 percent of the fish families survived, although their fossil record is too incomplete to determine how many species vanished right at the KT boundary. The marine reptiles, such as the dolphinlike ichthyosaurs and the long-necked plesiosaurs, were in decline long before the KT boundary and gone before the impact. The huge seagoing monitor lizards known as mosasaurs, however, were flourishing in the Late Cretaceous. With the huge sea-level retreat and the lack of shallow marine deposits at the KT boundary, it’s hard to know whether they vanished abruptly at the impact horizon or not.

To summarize, the marine record sends a confusing mix of signals. Some extinctions, such as those of the planktonic foraminifera, the coccolithophorids, and possibly the brachiopods, are consistent with the idea that the impact was the dominant kill mechanism. Other animals—such as the benthic foraminifera, dinoflagellates, diatoms, radiolarians, and most of the clams, snails, nautiloids, echinoderms, and bryozoans—show relatively little effect at the KT boundary. And still other animals—such as the inoceramids, rudistids, belemnites, most ammonites, and the marine reptiles—were clearly in decline long before the KT impact and were not alive to witness it. No matter how you plead for the impact scenario, the facts show that it cannot have been as hellish as some advocates suggest, or there would have been a much more severe effect on the oceans.

What about the record on land? Once again, the biological record is a confusing mix of signals that cannot be explained by one simplistic mechanism. As J. David Archibald and Laurie Bryant show in their review of extinctions (Archibald and Bryant 1990; Archibald 1996), the dinosaurs are practically the only major victims of the KT asteroid impact. A number of scientists, however, have shown that the dinosaurs were already slowly declining through the Late Cretaceous, with only Tyrannosaurus and Triceratops vanishing at the end—and their youngest bones are 3 meters (10 feet) below the iridium anomaly in the Hell Creek beds in Montana (see figure 5.1). Out of 111 species documented in the Late Cretaceous, about 65 percent of the species survived the impact. In addition to the dinosaurs, there were significant losses in the sharks, marsupials, and lizards. Even more striking is which animals were not affected. The Late Cretaceous was rich in crocodilians, some of which were larger than the smaller dinosaurs, and they were relatively unaffected. Likewise, the turtles show no effects of the KT extinction, nor do the bony fish and amphibians. Although the fossil record of birds and insects is not as good, nearly all the Late Cretaceous lineages survived, suggesting that they were not decimated either (LaBandeira and Sepkoski 1993; Chiappe 1995). Douglas Robertson and colleagues (2004) tried to salvage the impact scenario by suggesting that all the survivors were able to burrow or seek shelter in lakes and rivers or in the ocean, but that explanation won’t work with insects or birds, which survived, or with sharks, which didn’t survive. No matter how one modifies the extreme effects of the “nuclear winter” version of the KT impact, it does not account for the majority of either extinctions or survivals at the KT boundary and must have been much less catastrophic than claimed. Instead, the slow decline of many of the species argues that extreme conditions, possibly triggered by the Deccan trap eruptions, may have been more important. Archibald (1996) looked at the known ecological characteristics of victims and survivors and showed that the sea-level drop actually explains the extinctions (especially of sharks) better than the more popular mechanisms. If anything, the impact was probably just the coup de grâce that finished off some of the survivors of this hellish time.

Another related scenario has been ruled out entirely. Several groups of scientists argued that the KT extinctions occurred because huge amounts of acid rain were produced when the impacting body hit the sulfur-rich basement rocks of the northern Yucatán Peninsula. But this idea can be dismissed because one of the groups to survive with almost no extinctions was the amphibians even though today the slight amounts of acid rain caused by human pollution are wiping out frog and salamander populations worldwide. If the huge acid rain bath scenario were true, there would not be an amphibian alive on the planet now (Weil 1984). Similarly, tropical bees cannot survive more than a few days if the climate becomes too cold or the flowers disappear, yet they did not vanish after the impact (Kosizek 2003).

Most paleontologists are still very skeptical because the paleontological evidence is not consistent with an impact that wiped out animals in the pattern that had been predicted. In addition, many geologists could think of other explanations. Through the 1980s and 1990s, Charles Officer, Charles Drake, Gerta Keller, Norman MacLeod, and many others kept the debate going. I vividly remember the heated and often bitter scientific sessions at the annual GSA meeting when these two sides squared off against each other again and again. The arguments soon became very nasty and personal. The published literature was full of direct attacks on opponents’ scientific credibility and competence, and the name-calling at meetings was even worse. Some people’s careers were ruined because they were on the wrong side of the debate, and many others suffered enormously as the big boys battered it out. Luis Alvarez remarked: “I don’t want to say bad things about paleontologists, but they’re really not very good scientists. They’re more like stamp collectors” (qtd. in Browne 1988). On the opposite side, Bob Bakker told a reporter: “The arrogance of these people is simply unbelievable. They know next to nothing about how real animals evolve, live, and become extinct. But, despite their ignorance, the geochemists feel that all you have to do is crank up some fancy machine and you’ve revolutionized science. The real reasons for the dinosaur extinctions have to do with temperature and sea level changes, the spread of diseases by migration and other complex events. In effect, they’re saying this: we high-tech people have all the answers, and you paleontologists are just primitive rockhounds” (qtd. in Browne 1985).

There was also a deep cultural divide between the geochemists and planetary geologists on one side, who were used to fairly simple testable explanations based on data from big machines, and paleontologists on the other side, who were aware of the complexity of biological systems and especially the peculiar extinction pattern at the KT boundary. With such high stakes involved, each side in the debate had a great deal riding on sticking with its position and defending it to the end—not only scientific prestige, but also access to publications, grant money, lab space, tenure, and many other tangible and intangible rewards. No wonder the debate became so bitter and personal! It was and is a classic example of the sociology of science, where a deeply divisive scientific debate is largely propagated by the differences in scientific perspective as well as by stubbornness and unwillingness to consider opponents’ arguments. Even though the arguments at the professional meetings have largely died down, the major players have not changed their minds and keep on publishing contrasting points of view some 29 years after the issue began (e.g., Keller 2005).

The geochemists and planetary geologists have claimed victory and dismissed all those who disagree with them as crackpots and fringe scientists. But that assessment cannot be valid when whole areas of the profession do not go along with the dominant view. In 1985, reporter Malcolm Browne of the New York Times took an informal poll of the paleontologists at the SVP meeting in Rapid City, South Dakota. (I was there, but somehow he missed me, probably because I was a young scientist with no publications on the topic.) The impact hypothesis was already five years old, but, according to the poll, the vast majority of scientists at that meeting found it unconvincing, and only 4 percent thought it was responsible for the KT extinctions. Even though some people have declared the impact hypothesis the winner and the matter settled, a number of books by well-respected paleontologists (Archibald 1996; Hallam and Wignall 1997; Dingus and Rowe 1998) beg to differ. The survey by MacLeod and his colleagues (1997) of a distinguished panel of 22 British paleontological specialists in nearly every group of marine fossils also came out against the impact scenario as the cause of marine extinctions. Then in 2004 another poll (Brysse 2004) was conducted of the SVP membership. Of those surveyed, 72 percent felt that the KT extinctions were caused by gradual processes followed by an impact. Only 20 percent felt that the impact was the sole cause. The other 8 percent had no opinion or questioned whether it was a mass extinction at all. Twenty-four years of arguments by the impact advocates apparently did not really change the opinions of the people who know the fossils the best.

During the early, heady days of the KT impact idea, scientists soon investigated the notion that impacts might cause other mass extinctions. They looked at the other major mass-extinction boundaries and tried to find iridium and other impact indicators. Some claimed there was evidence for impact at several of the other mass extinctions, although these reports proved to be premature. Nevertheless, I recall sitting in the audience at the International Geologic Congress in Washington, D.C., in 1989 and hearing Canadian paleontologist Digby McLaren (who once advocated a supernova as the cause of the Late Devonian extinctions) say that all mass extinctions were caused by impact, whether there was evidence of impact in the fossil record or not! I vividly remember the audience’s stunned reaction at this blatantly untestable and unscientific statement. Ever the gadfly and provocateur, David Raup (1991) wrote that all extinctions (even normal background extinctions) might be caused by impacts. With such statements, why bother with data anymore? Impacts occurred, and extinctions occurred; therefore, all extinctions were caused by impacts.

Despite objections, the bandwagon for the generality of the impact-extinction hypothesis was rolling, and many scientists jumped on for the ride. As Keith Thomson put it, “With most subjects there is a silly season, usually of unpredictable duration and with an intensity correlated with the status of the acceptance of the new idea, [including] proposal of ideas even more far out than the original one” (1988:59). Paper after paper began to appear in Nature and Science claiming to find iridium or shocked quartz at many of the other extinction horizons. These papers were briefly peer reviewed, but were published quickly, often with great fanfare in the popular scientific media. Then a year or two later another group of scientists would try to reevaluate the data or replicate the results, and the “great discovery” would turn out to have been an illusion. The rebuttals were almost never published in these high-profile journals, however, nor did they get any publicity, so people only remember the first mention of the “discovery” and never realize that it was discredited.

In 1984, David Raup and Jack Sepkoski published a paper that claimed there was a 26-million-year periodicity in mass extinctions. Within weeks after it was published, numerous papers (based on the unpublished, unreviewed preprint of the article, according to Raup [1991]) came out trying to explain this “periodicity” in terms of periodic comet showers, the oscillation of the solar system through the galactic plane, an unknown Planet X, and even an undetected companion star to the Sun dubbed “Nemesis.” Still others tried to explain the “periodicity” in terms of pulses of mantle volcanism. Within a few years, however, the idea itself was dead because the “periodicity” was not real, but rather an artifact of the methods that Raup and Sepkoski had used, along with a compilation of garbage data, bad statistics, fossil species that were not real, and bad timescale estimates (see the detailed review in Prothero 2003:chap. 6). In 1989, Sepkoski published a last-gasp defense of the idea, but Steve Stanley (1990) proposed a much simpler explanation for the apparent spacing of mass extinctions at 20 to 30 million years or longer. In a truly major mass extinction, life is decimated and reduced to a low-diversity world of a few “weedy” opportunistic survivors. It takes a full 15 to 20 million years for all the extinction-prone highly specialized species to reevolve. If some major crisis happens too soon after a major mass extinction, it has very limited effects because the world is inhabited by extinction-resistant survivors and has few or no vulnerable species.

When a bandwagon gets going, though, it is hard to stop. Planetary geologists and geochemists may not know much about biology or paleontology, but they are more than happy to jump in and take a few samples of a key boundary and make pronouncements. Only after time has passed and enough skeptical geologists with better training in stratigraphy and paleontology have reexamined the data do we really know if the initial discovery is real or not. For example, the “mother of all mass extinctions” was the Permo–Triassic event at 251 million years ago, which ended the Paleozoic era and may have wiped out 95 percent of all species on the planet. It is the greatest mass extinction to hit in the past 600 million years, so many geologists have tried to explain it. Various models based on global sea-level change, assembly of Pangea, and global cooling have been proposed and then shot down as better data are collected. In 2001, Luann Becker and colleagues argued that the extinction was caused by an impact and claimed to have discovered evidence at the Permo–Triassic boundary of some unusual forms of helium contained in fullerenes, which are the 60-carbon molecules also known as “buckyballs” after their geodesic structure and in honor of Buckminster Fuller. According to Becker and her coauthors (2001), these fullerenes with helium were products of a Permo–Triassic impact. These researchers then went on to identify Bedout Crater in Australia as the impact site (Becker et al. 2004). But Kenneth Farley and Sujoy Mukhopadhyay (2001) were not able to replicate Becker’s “fullerenes” and found no unusual forms of helium that might indicate an impact. Several researchers (Glikson 2004; Koerberl et al. 2004; Renne et al. 2004; Wignall et al. 2004) quickly shot down the idea of the Bedout Crater, especially because it is the wrong age to have anything to do with the Permo–Triassic extinctions.

The same goes for all other mass extinctions that have been blamed on impact. The fifth biggest extinction in Earth history occurred between the Triassic and Jurassic periods of the Mesozoic, which eliminated 48 percent of the marine genera and was part of the process that shifted the dominant land vertebrates from the synapsids (formerly but incorrectly called the “mammal-like reptiles,” even though they are not reptiles) to the early dinosaurs. In the 1980s, many geologists tried to tie this mass extinction to impacts as well. My friend Paul Olsen claimed there was evidence that the huge Manicouagan Crater in Quebec was the culprit (Olsen, Shubin, and Anders 1987; Olsen et al. 2002). This monstrous hole, which shows up as a huge ring on the satellite images of Quebec, is about 100 kilometers (62 miles) in diameter, not much smaller than Chicxulub. But recent redating of the crater debris puts its age at 214 million years ago, nowhere near the Triassic–Jurassic boundary at 201 million years or near the age of any other mass extinction (Palfy, Mortensen, and Carter 2000). Shocked quartz and iridium have also been claimed for this boundary, but further scrutiny has shown that their concentration was so small as to be unlikely to cause extinction (Hallam 1990, 2004; Hallam and Wignall 1997; Tanner, Lucas, and Chapman 2003). The third greatest extinction was in the Late Devonian (375 million years ago), when 75 percent of the marine species died out. Once again, impacts were blamed at first, and iridium anomalies were reported. However, closer scrutiny shows that these iridium anomalies are at the wrong time, and the evidence for impacts (if it is real) is not correlated with the several pulses of geochemical changes and extinctions in the Late Devonian (McGhee 1996).

The most striking example of scientists jumping on the impact bandwagon and getting ahead of their data concerned my own area of expertise, the Eocene–Oligocene transition. It is not one of the “Big Five” mass extinctions, but as we have already seen, it was a major event nonetheless. Raup and Sepkoski (1984, 1986) tried to fit it in their periodic-extinction model, although the “periodicity” predicted that the extinction should have occurred at 40 million years ago. In fact, there were two extinction pulses at 37 and 33 million years ago—nowhere even close! The Berkeley impact gang (Asaro et al. 1982; Alvarez, Asaro, et al. 1982) and several others (Ganapathy 1982; Glass, DuBois, and Ganapathy 1982) went looking for iridium. Sure enough, they found it near the Eocene–Oligocene boundary, so they crowed loudly in the scientific press about how they had “solved” the mystery of the Eocene–Oligocene extinctions. But paleontologists and stratigraphers who really knew the late Eocene (see the papers in Prothero and Berggren 1992 and in Prothero, Ivany, and Nesbitt 2003) dug in and did the hard detective work that the impact gang had completely neglected. The details that came to light showed that the impact was in the middle of the late Eocene at 35.5 million years ago, too late for the extinctions at 37 million years ago and too early for the extinctions at 33 million years ago. Except for a few species of radiolarians, there were no extinctions in any other group of organisms associated with these impact layers. “Close enough” may work in horseshoes and hand grenades, but not for precisely dated events such as those of the Eocene-Oligocene transition, where 1 to 2 million years of strata lie between each extinction horizon and the iridium anomaly.

In fact, this horizon has turned out to be an embarrassment for the impact advocates and a striking falsification of their simplistic view of the world. The sites of the middle to late Eocene impacts are now well known (Poag et al. 1992; Poag 1999; Poag, Mankinen, and Norris 2003). They include two big craters, one underneath Chesapeake Bay and the other at Toms Canyon on the Atlantic continental shelf, plus an additional impact site at Popigai in northern Siberia (figure 5.4). The Chesapeake Bay crater is huge, almost 100 kilometers (62 miles) in diameter. It is full of impact debris at the bottom, and since the impact occurred, it has filled with sediments from Chesapeake Bay over the past 35 million years. Popigai is only slightly smaller (90 kilometers [56 miles] in diameter). These craters are not as big as the 180-kilometer (112-mile) Chicxulub crater in Yucatán that is blamed for the KT extinctions, but they are close. Yet detailed studies of the crater and the strata formed in the aftermath show no evidence of extinction or of the global cooling predicted from the dust clouds, but actually a short global-warming event (Poag, Mankinen, and Norris 2003). Some last-ditch efforts (Poag 1999; Coccioni et al. 2000; Vonhof et al. 2000; Fawcett and Boslough 2002; Poag, Mankinen, and Norris 2003) have been made to salvage the impact-extinction scenario by blaming impact-induced climatic changes for the mass extinctions that happened 2.5 million years later, but these effects don’t last in the atmosphere or oceans that long.

Actually, the lack of extinction for the second- and third-largest craters known after Chicxulub says something very different: impacts don’t cause mass extinctions except in extraordinary cases. Wylie Poag (1997) replotted the “kill curve” that Raup (1991) had once fit to a single data point, the Chicxulub impact. Raup’s “impact kill curve” predicted that a crater of 100 kilometers (62 miles) in diameter should wipe out 60 percent of the species on the planet (figure 5.5). But when the 90- to 100-kilometer Chesapeake and Popigai craters are added in, the curve suddenly takes a very different shape. Apparently, only craters of nearly 200 kilometers (124 miles) in diameter are correlated with mass extinctions, and anything much smaller has little or no effect on life. Now even hard-core impact advocates such as Peter Ward (2007) are admitting that the impact bandwagon was premature and that impacts (except for the KT) have no effect on life. At the 2006 GSA meetings in Philadelphia, where the impact advocates had once reigned supreme, talk after talk was about the failure of impacts to account for extinctions and why the KT event is the sole exception. Some blamed extinction on the impact target. The KT Chicxulub crater was blasted into sulfur-rich gypsum bedrock, supposedly producing sulfuric acid rain, whereas all the other meteors hit granitic or gnessic continents or basaltic seafloor, which are not chemically reactive. As we saw earlier, however, this excuse is weak because the evidence of the negative effect of “sulfuric acid rain” is debunked by the fact that we still have frogs and salamanders. And as I already pointed out, the KT impact may not have had much of an effect after all if the paleontological data are to be believed.

In my scientific career, I’ve seen the profession go from puzzlement about mass extinctions before 1980 to the erection and dismantlement of the impact bandwagon from 1980 to 2003, and now new ideas are coming along that may or may not better explain mass extinctions. The impact bandwagon beautifully demonstrates a salient fact about science: scientists are human and subject to social pressures and lured by attractive new ideas. But science is not like politics or philosophy, where one school of thought or idea can persist even after it has outlived its usefulness or in spite of much negative evidence. In science, we must measure our ideas against an external reality. Fads may come and go, and scientists may favor certain ideas for irrational reasons, but scientific hypotheses must stand the test of time and be corroborated by studies that may take years to finish. Our cherished ideas may ultimately turn out to be illusions, but as scientists, we cannot hang on to them and must move on. As Thomas Henry Huxley put it, this is “the great tragedy of Science—the slaying of a beautiful hypothesis by an ugly fact.” Or as H. L. Mencken said, “For every problem, there is a solution that is simple, neat, and wrong.”

FIGURE 5.5 The “impact kill curve” as plotted by Wylie Poag (1997). In David Raup’s (1991) original version (dashed line), the curve was fitted to only one point: the Cretaceous–Tertiary impact at Chicxulub. It predicted that a crater of 80 kilometers (50 miles) in diameter should produce a 40 percent extinction of species, a 145-kilometer (90-mile) crater should result in a 60 percent species extinction, and so on. But Wylie Poag (1997) refit the curve to the huge late Eocene craters at Chesapeake Bay and Popigai, and found that even craters of 97 kilometers (60 miles) in diameter produced almost no extinction. (From Poag, 1997, Fig. 4; courtesy SEPM)

In November 2002, I found myself flying “across the pond” from Los Angeles to London and then running as fast as I could through Heathrow Airport customs and security to catch my Lufthansa flight to Berlin. I had been invited to speak at a major conference at the Humboldt Museum für Naturkunde in Berlin. The conference was titled “Mesozoic-Cenozoic Bioevents: Possible Links to Impacts and Other Causes.” This huge old museum contains the five-story-tall Brachiosaurus skeleton from Tendaguru, Tanzania, featured in nearly every book on dinosaurs. Every time we stepped out of the conference room and looked out over the balcony, we could see its immense skull towering above us from our balcony view on the second floor. It was a striking reminder of the topic of extinctions that was the focus of the conference. The museum contains many other paleontological treasures, including the best-known and most complete of the 11 known specimens of Archaeopteryx (the “Berlin specimen”) as well as amazing dinosaurs and marine reptiles collected in the late nineteenth and early twentieth centuries, when Germany led the rest of the world in the field of paleontological exploration. This museum amazingly survived the bombing and shelling of Berlin during World War II, even though most other German museums were destroyed and lost countless irreplaceable paleontological treasures. During my first visit to Berlin as a high schooler on a two-month tour of Europe in 1971, I had no chance to visit the museum because it was on the other side of the Berlin Wall. We had to reach Berlin by train through then Communist East Germany and saw the sights of West Berlin. My high school tour group was allowed to pass through Checkpoint Charlie and briefly drive around some of East Berlin on a tour bus, mostly to witness the grim desolation of the Communist side of the city. In 2002, the Berlin Wall and Checkpoint Charlie were gone, replaced by Cold War museums and new streets, and Germany was unified and economically strong and thriving again.

The organizers of the conference included my grad school buddy and frequent coauthor Dave Lazarus, who was now a curator of micropaleontology at the museum, and several meteorite specialists who were big fans of the impact model. The impactors presented their talks the first day of the meeting, but for the rest of the conference the paleontologists got up and presented the newer data that made the impact model obsolete. Many of the leading figures in the mass-extinction debates were there, so the questions were pointed and intense. I was practically the only speaker focusing on the Eocene–Oligocene transition, but I made a strong case that the impact model, the mantle-volcanism model, and many others were inadequate to explain the transition. This was first time I saw the impact-extinction hypothesis go into disfavor and other models start to replace it.

The impact-extinction model may be on the decline, but that does not stop scientists from looking for a common pattern among the great mass extinctions. The most popular alternative to the impact model has been the mantle-volcanism model, which argues that during times of gigantic flood basalt eruptions, huge amounts of gases are released from the mantle that may cause extreme climate change and even runaway global warming (Rampino and Stothers 1988; Courtillot 1999). This model receives support from the fact that three of the Big Five mass extinctions are definitely associated with huge mantle-derived volcanic eruptions. I have already mentioned the Deccan traps, which erupted just before the end of the Cretaceous. The Triassic–Jurassic extinctions correlate with huge eruptions from the Central Atlantic Magmatic Province, which took place when the North Atlantic began to rip open as Pangea broke apart.

The Permo–Triassic extinction was coincident with huge eruptions known as the Siberian traps, the largest such eruptions known in Earth history. These flows were up to 6,500 meters (7,100 yards) thick in 11 discrete eruptive sequences and covered a total of about 7 million square kilometers (2.7 million square miles), an area equivalent to that of the continental United States (Erwin 2006). The ages of these flows have been recently redated at 252.2 and 251.1 million years old, exactly the same as the dates from the Permian–Triassic boundary in China. It would be remarkable if eruptions so immense did not have an effect on late Permian life.

The problem with this attractive model is that none of the other major mass extinctions correlate with major volcanic episodes. In addition, many huge eruptions occurred that did not cause mass extinctions. No major volcanic episodes have been shown to correspond to the complex Late Devonian extinction or with the Late Ordovician extinction. Conversely, no mass extinctions are associated with the Columbia River flood basalts (15 to 16 million years old) that cover most of eastern Oregon and Washington. The eruption of the North Atlantic Tertiary province is dated between 61 and 56 million years ago, but it is not associated with any significant extinction either. Several researchers (Rampino and Stothers 1988; Courtillot 1999; Courtillot and Renne 2003) attempted to attribute the Eocene–Oligocene extinctions to flood basalts in Ethiopia and Yemen. Unfortunately, the latest dates on these lavas are between 29.5 and 31 million years, or during the middle Oligocene, which makes them several million years too young to have had anything to do with the extinctions at 37 and 33 million years ago.

But there may be an indirect link between some of these eruptions and another possible killer: atmospheric gases. The trendiest new idea is that some of the major mass extinctions (especially the Permo–Triassic, Triassic–Jurassic, and possibly the Paleocene–Eocene extinctions mentioned in chapter 1) were caused by unusually high concentrations of carbon dioxide and low concentrations of oxygen in the atmosphere (Ward 2006, 2007). The evidence for this explanation comes from the models of atmospheric gases developed by Bob Berner at Yale (Berner et al. 2003), which show unusually high levels of carbon dioxide and low oxygen at these three extinction horizons, as well as at the Late Devonian, Late Ordovician, and Late Cambrian extinctions (figure 5.6). Further evidence for this model comes from the geochemistry of seafloor sediments, which show unusual enrichment in light carbon that might have come from eruptions such as flood basalts. In fact, the current model for the Permo–Triassic extinction involves oversaturation of the oceans with excess carbon dioxide (called hypercapnia), which is fatal to most marine life. Likewise, the high carbon dioxide and low oxygen levels of the Permo–Triassic and Triassic–Jurassic boundaries would have been extremely stressful for most land animals and particularly favored those with more efficient respiration, such as the early dinosaurs (Ward 2006).

At the GSA meeting held in Denver in October 2007, this trendy new idea was promoted at the Pardee Symposium and given maximum publicity. It certainly seems plausible given what we know now and worthy of further examination. Nevertheless, it is not the be-all or end-all of explanations of mass extinction. For example, there is no evidence that the Eocene–Oligocene extinctions suffered from elevated carbon dioxide or low oxygen, either on the Berner curves or in any other data source. In fact, the transition from greenhouse climate to icehouse climate suggests that carbon dioxide was declining, not spiking, in abundance at the mass-extinction horizons. Some people have tested specific predictions made by Peter Ward (2006) and found that they contradict or even falsify his model (Holtz 2007), so it is premature either to declare it as the final explanation for mass extinctions or to rule it out either. As occurred with its predecessors, collecting enough data to evaluate it and either accept or reject it will take years. More likely, we may find that it fits certain extinction events, but not others.

Eocene–Oligocene extinctions are notably the joker in the deck that always seems to falsify the generalized attempts to explain mass extinctions, from the Raup and Sepkoski periodicity model to periodic volcanism to the impact scenario and now the “gas attack” model. No matter what we do to massage and tweak the data for these extinctions, they stubbornly refuse to go along with any single pattern deduced from any other mass extinction. The Eocene–Oligocene extinctions stand out, resistant to simple explanation, as the persistent falsifier that ruins any attempt to give a unified explanation to mass extinction. I didn’t realize it at the time I started research on the Eocene–Oligocene, but I could not have asked for a more interesting and important interval to build my career on. It’s just another example of serendipity.