Around 11,000 years ago, at the end of the Pleistocene epoch, North America witnessed an extinction that claimed its mammoths, giant ground sloths, camels and numerous other large-bodied animals. Exactly what happened to these megafauna is unknown. Indeed, researchers have puzzled over their disappearance for decades. Traditional explanations hold that either dramatic climate shifts, or human hunting (overkill) extinguished these species. But in recent years a new hypothesis has emerged. According to Ross D. E. MacPhee, curator of mammalogy at the American Museum of Natural History in New York City, extremely lethal disease, brought over by humans unwittingly when they arrived in the New World, may have wiped out those Ice Age giants.
Scientific American writer Kate Wong discussed this hyperdisease hypothesis with MacPhee last year. The edited transcript that follows falls into four sections. In the first part MacPhee talks about the shortcomings of the climate and overkill models. In the second part he provides examples of recent extinctions caused by disease and describes how the first Americans might have introduced hyperdisease when they came to North America. Megafaunal extinctions followed human arrival in Australia, New Guinea, the West Indies and Madagascar, too. The same pattern does not apply to Africa and southern Eurasia, however. MacPhee explains how his model accounts for these exceptions and ponders the surprising survival of certain North American megafauna in part three. So far MacPhee does not have empirical evidence for his hypothesis, but he and his colleagues hope to find it in mammoth remains. In part four he describes their search for signs of lethal microbes in ancient tissue and DNA.
SA: I'd like to start by asking you to explain how, in your opinion, the climate and overkill models are flawed.
RM: The climate has changed radically at times when there was no extinction, and extinctions have occurred when the climate, at least roughly speaking, should have been benign. There is no question that there were catastrophic changes in temperature and probably in precipitation on many occasions in the past 100,000 years. We know that, for entirely natural reasons, temperature excursions of seven to 12 degrees Celsius occurred within that time period in the space of a century or less, which is basically 12 times the maximum rate of change in the last century of "global warming." If changes like that are meaningful for extinction, then you would expect to see a correlation-how could it be otherwise? If climate change of that radical a caliber has occurred in the past there should have been losses. And the point is that there are no correlations. So all of that, as far as I'm concerned, puts climate, insofar as it's understood what we mean by climate change, out of the picture.
There is a strong correlation between arrivals of people in places where people haven't lived before and sudden spikes in the extinction rate so that you get sudden disappearances--particularly of large-bodied animals--in a period within decades or centuries of first human arrival. So it's easy to see why people would assume that these losses had something to do with the arrival of humans, and since we think of humans as being red in tooth and claw, that they must have provoked these extinctions by doing something nasty like hunting at a rate that they shouldn't have. The trouble with that particular argument is that the archaeological record does not support it in any of the places where these extinctions occurred. Of course there are cases where projectile points have been found embedded in mammoth bones. But when you take a look at the number of instances, you can barely come up with a dozen for the relevant time period in North America--between 11, 000 and 12,000 years ago. In other words, although people were clearly hunting, it is not a demonstration by that evidence alone that they were hunting on a scale that would have made any difference to the survival of species.
Image: CLARE FLEMMING
SA: So, in your opinion, even if the first Americans were highly skilled hunters, could their population sizes and the population sizes of these animals have been such that overkill would even be plausible?
RM: The answer is no, by any scenario. I don't care how early you want people to get into the New World, theres absolutely no evidence of a positive sort that they were there in huge numbers. In fact, it has to be the opposite, whereas the animals, in some cases, had distributions that were continent wide. Some of the ground sloths, for example, are known from as far south as Mexico and from as far north as the Yukon. The notion that people in whatever numbers and with whatever intent could have come in and slaughtered enough sloths in every possible habitat where they lived in numbers sufficient to cause their extinction--this is unbelievable to me.
Archaeologists say in looking at Clovis sites and similar ones in Monte Verde in South America that theres nothing at these sites that suggests anything other than band-level organization. What we know from modern ethnographic examples is that the individual family groups that compose a band tend to cooperate only for very specific objectives. As soon as that economic objective is met, it's over--they don't keep a high level of organization once there's no need. So if we're talking about Clovis people being at essentially the band level economically, how can it be that they would stay together for common purposes at the level necessary to cause these extinctions? Youd have to be killing things all the time, and you'd have to be doing it for some purpose, even if the purpose was just to kill. And it's just unimaginable to me that the people concerned would be interested merely in killing, especially large, dangerous animals like mammoths. You take out the one or two a year that you need, and then you go off gathering roots and tubers, which is in fact how most of these outfits keep goingits not by direct hunting. You can't look at the first Americans as being basically people like us who didn't wear suits--that their objectives would be similar, that their worldview would be similar, and all the rest of it. In fact, most assuredly they were not, if ethnographic comparisons mean anything.
SA: Enter hyperdisease.
RM: Hyperdisease has its own extremely large explanatory defects. But the notions that you cant get cross-species infections, or that you can't get huge mortalities that might lead to extinction, are not among them. In fact, there are such examples. There is a group of birds native to Hawaii called the Hawaiian honeycreepers, several species of which have gone extinct probably within the last 100 years. Whereas when Europeans were first going to Hawaii at the beginning of the 19th century, honeycreepers were known at lower elevations. Nowadays, however, the surviving populations all live at high altitude. Why should this be? Researchers figured out in the 1960s that the distribution of the surviving birds is mediated by how far up in altitude avian malaria-carrying mosquitoes can go. The mosquitoes that we're talking about, representing a species of Culex, were introduced from tropical North America probably in the 1850s or 1860s. And in all probability what happened was that some boat going from San Francisco or Mexico over to Honolulu had freshwater in its bilge and female mosquitoes were laying eggs there. Some of the larvae survived after the bilge was dumped, they started biting the native birds, and some of the larvae had the [malaria-causing] protozoan Plasmodium in their system, so they inoculated the birds, and the birds died in droves. To repeat the experiment and thereby document what happened, what the researchers did was take some of these individuals from surviving populations up on the high mountains, bring them down to the lab at sea level, and then introduce them to Culex individuals that were known to be carrying avian malaria. The exposed birds died without exception--100 percent mortality.
What I make of this is obvious: the distribution of the surviving birds is disease-controlled. I also think that the populations or species that could not survive at high altitude--the birds that have gone extinct--became extinct because there was nowhere for them to go. Everywhere they went they were greeted by buzzing mosquitoes who were carrying the disease, and the birds dropped out. I think that is a pretty convincing case. And all of the leading list-makers--like IUCN, Nature Conservancy, U.S. Fish and Wildlife--when they attribute cause of extinction to those particular birds use the word disease, as opposed to habitat clearance, as opposed to introduced species, persecution, all of these other things that are believed to cause endangerment and extinction. So at least in some quarters it is accepted that those bird extinctions are due to disease per se.
The other great example is golden toad extinction in Panama and apparently huge declines in certain frog species in places like Queensland [Australia], all due to a fungal infection, chytridiomycosis. Up until the mid '90s in Panama the census numbers were consistently high. Then something happened to the golden toad. Within the space of a year or two the numbers of sightings and sound recordings and so on dropped from the average level established over many decades to zero, essentially overnight. For the past five years there have been no sightings of that particular frog in Panama and Costa Rica, which were the places where it lived. It looks like the populations have just sunk to nothing.
Peter Daszak, a parasitologist at the University of Georgia, and several other investigators got interested in this problem because they were given some autopsy specimens of golden toads to examine. What they found was that a particular chytrid was consistently present in autopsy specimens from areas in which the populations were known to have sunk to zero in the manner that I described. And what they found on further investigation was that the specific cause of death seems to have been that this fungus, which is an epidermal infection, caused local thickening of the skin--especially over an area known as the drink patch down in the pelvis, which is what these frogs use for osmoregulation, for balancing out their water. With any problem with that area of skin, the frog in effect suffocates or drowns. Interestingly, tadpoles had no epidermal infection, but they did have the chytrid present in their mouthparts. Presumably, as they metamorphosed, the chytrid infection became general on the epidermis and they died out as well. So what you've got, it seems to me, is the worst of all possible cases, which is a universal infection--that every individual either has it or could potentially have it. It's something that is obviously very easily passed on or distributed within the environment. There would seem to be no escape for the populations that suffered from the chytrid infection: it wasn't like the chytrid would pass through the adult group and the next generation would be okay.
A similar chytrid--perhaps even the same chytrid--has appeared in Queensland, Australia. It is not known that it has been responsible for any complete extinctions, but it has certainly been responsible for massive depressions in population in the frog groups that it affects. It has also turned up in the southern part of South America. Why this particular chytrid at this particular time? Who knows? Perhaps it's because of people. Because you can get anywhere on the planet's surface nowadays within about 48 hours, the chances for pathogen pollution--in other words, bringing pathogens from one area to another where they might be able to take off--is incredibly enhanced over what it was even a few decades ago. From that point of view it's perhaps not even all that remarkable that we'd start to see diseases that are essentially pan-global. Not because they're getting distributed better by winds or currents, or whatever, but by people moving around.
To me these are startling examples of what diseases can do, and the only reason people havent heard about it is because theyre affecting species that are not all that visible, living in obscure places. It's the scientists who are ringing the Klaxon right now. What I think would focus people's attention would be if one of these diseases that are now emerging in African wildlife right now caused an outright extinction. African wild dogs in the Serengeti have been essentially wiped out by canine distemper transferred from domestic dogs. The wild dogs still exist in small numbers elsewhere in central Africa, but for a very large part of their original range they're bye-bye. Its a clear and present danger, and when you start putting these individual examples together, they in fact amount to something.
What relevance does this have for the Pleistocene? If diseases are emerging at a greater rate now, thanks to translocation due to people, then doesn't it make sense that when you have people beginning to translocate from Africa and south Asia, where Homo sapiens is ancient, that theyd be carrying new things with them, in the form of their biological baggage? All kinds of organismsincluding pathogens that they may or may not have known aboutcould have been brought to places where humans had never been previously resident. If these examples Ive been talking about are meaningful, when immunologically nave species are faced with the sudden introduction of pathogens that they have had no experience with whatsoever, a very typical outcome seems to be very high levels of mortality--even to the level of complete extinction. Generalize outward from that and there is nothing else in nature that we know about, short of a cometary impact, that could take out the number and kind and distribution of species in North and South America 11,000 years ago that weve been talking about.
If the North American end-Pleistocene extinction was a focal extinction, if it happened only along the western seaboard of North America where people came in first, I could accept that people were completely responsible--I wouldn't see the need to have an alternative explanation. If the extinction event affected single groups--like all the elephants all over the planet dying out--I'd think, Well, that's kind of peculiar, but maybe people, for one reason or another, are elephant mad and decided they didn't want any more elephants around. But when you look at the continental extinctions in the New World you have upward of 130 species disappearing in a time period of maybe half a millennium or less, all the way from north slope of Alaska down to Tierra del Fuego and in every kind of environment in between. Is there anything on Earth that we know about that could provoke losses on the scale necessary in such a limited time period, to affect populations wherever they existed? Disease is the only thing I know of that is part of the natural landscape that could possibly do it.
In imagining such diseases, ones that can spread very widely and are very host tolerant, I'm going to the edge of what we know is possible. But we do know from the recent past that there are such diseases. The rinderpest epidemic in East Africa at the beginning of the 20th century took out millions of individual African bovids--this would include wildebeest, hartebeest, bongos, the whole range. Ecologists have been looking at this and have decided that there were certain kinds of ecological change that were induced by the catastrophic loss of bovids around 1900 from which the forests and other landscapes of East Africa have still not recovered--which is amazing when you think what that means in terms of the loss of individuals, and the slowness with which the species have bounced back. A number of other diseases show much the same tolerance of hosts; rinderpest is not unique.
SA: North America is one of many places where megafaunal extinctions coincided with human arrivals. Yet this did not occur in Africa and southern Eurasia. Do you think the animals in these places were somehow resistant to these diseases?
RM: Yeah, the answer involves borrowing another page from Paul Martin's book [editor's note: Martin developed the overkill model]. The reason that overkill did not occur in Africa and south Asia, he would argue, is because the hominids and the local mammals evolved together. For every improvement in the hominid toolkit, the local animals came up with appropriate behavioral responses. So they were not nave, in a behavioral sense, and were able to deal with human predation on an ongoing basis. In the New World you have the opposite state of affairs, obviously. But I would argue that the navete was immunologic and genetic, rather than behavioral--which I actually think is extremely plausible. Just on first principles, Ive never really been able to get past the argument that animals are so persistently nave they just stand around and let themselves be butchered, especially in continental settings. But with disease you can imagine a situation, especially for herding animals, where the pathogen could be passed through a population in a matter of days, and they wouldn't be any the wiser. They would just be falling all over the place, without any clear threat on hand.
SA: A certain number of these big animals did survive. Are there any that you might have expected to perish if disease were as rampant as you think it must have been?
RM: I have wondered, as everybody has wondered, why elk, moose, musk ox, bison, among very few others, have managed to survive in a place like North America, or llamas down in South America, when all of them have close relatives that turned up their toes at the end of the Pleistocene during these big die-offs. From the disease scenario perspective, all I can imagine is something like the following: perhaps nearly every kind of mammal species was susceptible, but there would be groups that were affected but not devastated by the die-off, or individuals with the right genotypes to make it through. And then it just becomes a straight Darwinian selection problem. In some cases, enough immune individuals might be left to continue the species. In other cases, the die-off was so rapid that there was no possibility of coming back. This would be a good argument if there were a common feature that surviving megafauna sharedthat they all lived in one place or that they had a particular kind of birth spacing. Ive tried to check this out and theres nothing that I find very convincing. That's just the way it is. Which is not a very good argument, I know. But that is certainly one explanationthat what were seeing are survivors.
It is interesting, however, from a slightly different perspective, that at least half of the surviving megafauna of North America have Old World populations. This would apply to wolves, elk; this would originally have applied to musk oxen although they later became extinct in Asia; it would apply to moose also. My point here is that as catastrophic as these New World extinctions appear to have been, they may have been even worse. For example, the North American populations of something like moose could have completely disappeared at the end of the Pleistocene. The reason we have moose today, so this argument might go, is because North America was repopulated by Asian moose, who did not die out because they were much less susceptible to the pathogens. Possibly they'd had earlier hits and survived, or what have you. You can go crazy with all of these permutations, but it is interesting to me that a significant number of the megafaunal survivors have basically a holarctic distribution, which means a distribution across all the high latitudes. It's not just open and shut that you had significant survival of megafauna here in North America. We don't really know that. We know that some megafauna survived, but if you cut out the guys with holarctic distributions, then what remains is very, very limited indeed--pronghorns, llamas, tapirs, mountain goats and a couple of others, none very big.
SA: So far there is no empirical evidence to support the hyperdisease hypothesis. What are the chances that youre going to find a smoking gun--a microbe--in ancient megafaunal remains?
RM: I do not expect to be successful in the short term. There are both practical and theoretical reasons for that. The very practical one is that we're working with ancient DNA, which has a very poor reputation scientifically because of the enormous claims that were once made for retrieving DNA from dinosaurs and from inclusions in amber early in the '90s. None of these results could be replicated independently, and there's therefore a prevalent belief that most work in ancient DNA is so riddled with error and the possibility of contamination that it's not worth doing. Unfortunately it's one of the few approaches we can take in respect to trying to find evidence of pathogens, so were going to continue doing it. But what that means is that every experiment we do has to be cleaner than clean. We have to be able to stand behind every result that we think is good, which means endlessly redoing experiments and sending samples to independent labs to see whether they can replicate our results. As a result it's a slow operation, but theres no other way of doing it--we want to be right.
In principle (and here's the theoretical part) there is no reason, if pathogenic material is present in a well-preserved bone, that we shouldn't be able to visualize what's there by PCR [the polymerase chain reaction]. Because the technique is so sensitive, anything that is in your sample will be replicated. If you have the right fishing tools, in terms of primers, you should be able to get a good length of sequence and determine whether its one that youre interested in dealing with, which in our case would be a sequence from an existing virus, for example. The troubling detail is copy number. People have an easy time doing mitochondrial DNA, from the point of view of getting results, because whereas in the average cell you have only one nucleus and therefore only one copy of the complete nuclear genome, in any such cell you might have a thousand mitochondria. So just from a statistical point of view, the chances are that if anything gets replicated it'll be mitochondrial DNA. For a long time people didnt think it would be possible to get nuclear DNA from fossil material, but my colleague Alex Greenwood [at the American Museum of Natural History], who works with me on the disease hypothesis, was able to do exactly that with mammoth material a couple of years ago, and we have now consistently done it with mammoth specimens that were working on for the hyperdisease work. This means that our techniques are at least good enough to go out and get material of infectious organisms that might be there in very low copy number. In principle we think we should be able to do it. The techniques exist, at least primitively, and we have expectations that there will be great improvements in what we can do with ancient DNA in the next decade or so.
One of the difficulties we've had with PCR work is that in order to be successful you've got to know what you're after before you do it, because it's all in the primer design. And we don't know for a verifiable fact what kinds of pathogens were even dealing with, so it's a real needle in a haystack kind of problem. To deal with this situation we are trying some other approaches that would help us prior to the actual PCR-ing. One of these would be to use immunological approaches, which are used diagnostically all the time in the case of human diseases. If you want to know whether someone has sleeping sickness, or syphilis, for example, there are immunological tests that you can utilize that will give you a very clear yes/no answer because of the specificity of what's known as antibody/antigen reactions. If you get a positive result, it is meaningful because of the specificity of these reactions. Or if you get nothing, that is also meaningful--it means that the antigen [and therefore probably the microbe] you are probing for is simply not there.
Our idea here is to develop appropriate tests for identifying whether or not, say, a highly specific protein known to be present in the capsular covering of certain pathogenic viruses is in a given fossil specimen. That's the antigen we will then probe for using custom-designed antibodies. If we come up with a positive, that the protein is there, then we have narrowed our search enormously, down to that set of viruses known to manufacture the protein of interest. That gives us a clue as to what genes we should then look for in primer design for use in PCR experiments. If we get corroboration back, in the form of an expected sequence, then we know that this particular pathogen or kind of pathogen was present in that mammoth at whatever time it lived. That's not a demonstration of hyperdisease, clearly. That's only saying that you can in fact retrieve genetic information on "fossil" exogenous viruses. But that would be an astounding breakthrough for our work, because once we know how to conduct the proper experiment in the first case, the procedure can be generalized to all cases.
Another possible approach is using electron microscopy. This seems very primitive, sort of harkening back to a different age. But for us, identification is everything. If we could get an initial impression of what kinds of pathogens are present in a fossil, then we could go to more sensitive modern techniques to learn a lot more about them. In the particular case of mammoths, our idea is that if we could get vascular tissues, which occur on the inside of well-preserved bones, we could search for organized particles by a process of filtration. A number of viruses have organized capsules that are morphologically distinctive and diagnostic. If we can identify such viruses in filtrates examined microscopically, then weve got something else we can use to establish whether this or that particular pathogen is present. The test then becomes to see whether we can complete the hat trick by fishing out a sequence for it using PCR.
SA: Are you looking at samples from mammoths from a particular area?
RM: We're in all of these cases trying to find individuals that were possibly parts of one or more "terminal populations." This is another one of the very hard things that we have to do. The logic of our argument is such that it has to be accepted that the killer pathogens were not present in the population until they were introduced. Once introduced, their effect was so massive that the population and thereafter the species died out in its entirety. If we're talking about extremely lethal, acute infections, then once the panzootic gets started species will presumably have to disappear very quicklyperhaps in a few tens to a few hundreds of years. The difficulty paleontologically is to find those specimens from terminal populations. It's really very difficult to do because you have no idea necessarily where to look. So we're doing the best that we can, and what that means in the case of mammoths is going to northern Asia and radiocarbon-dating as many young-looking specimens as we can--and we have had some success in this already. One of the places we've worked in, for example, is Wrangel Island, which is where mammoths survived up until about 4,000 years ago. Thats clearly a terminal population case. Another is the Taimyr Peninsula, which is where the youngest radiocarbon dates for continental Asia have come from. It looks like that was some kind of refugium for mammothsthey survived there up until about 10,000 years ago.
SA: Are you envisioning a single disease or a suite of diseases wreaking all this havoc?
RM: It could be either. It would be simplest, of course, if it were one disease. But it strikes me that while it might have been one disease in North America, it was almost sure to have been quite another that took down the giant lemurs of Madagascar less than 2,000 years ago. This would scarcely be remarkable. The picture that I get just from what we know about human disease and the introduction of new diseases into human populations is that literally anything can take you down. You may know that there's a huge controversy right now about disease being purposely introduced into Yanomam country in Amazonia. There was a major epidemic of measles there in the '60s, and people died of it. Yet for European-derived populations, dying from measles is extremely rare. So there's no reason at all to think that even garden-variety diseases that are prevalent today could not have that killer impact on nave populations. Alternatively, it is also plausible to think that slight genetic changes in relatively benign microbes might render them lethal. For a good example we need go no further than the type A flu that went global in 1918 at the end of the First World War. This was a true killer plague--the worst in recent times--and it caused the death of between 20 and 40 million people in about a year and a half. Yet this novel flu evidently gained its lethality through a couple of substitutions in a couple of its genes. The point is these things are going on in the disease pool we all share in all the time. I suggest that you should be very, very frightened by these facts.