Lightning Strikes Twice: Blitzkrieg, Hyperdisease, and Global Explanations of the Late Quaternary Catastrophic Extinctions
Ross D.E. MacPhee, AMNH
Preston A. Marx, Aaron Diamond AIDS Research Center
Paul Martin observed, in the review that he wrote for the paper that Preston Marx and I recently published on hyperdisease, that this was the best thing he had ever read on the least likely cause of Pleistocene extinctions. I would like to return that compliment today -- by observing that it must be very nice for Paul to be in a room in which Pleistocene extinctions are being considered, and that he does not have the weirdest ideas.
About 11,000 years ago, as you've now learned in detail, the continental parts of the New World -- South America, North America and Central America, but not the West Indies -- suffered a bout of mammalian extinctions unlike any other. It is argued that these losses did not begin to occur until shortly after the first human arrival, although the timing and number of these immigration events continue to be controversial.
Less controversial, however, are some other points, like the matter of scale. It is beyond argument that some 85 genera -- roughly 135 species, by my count -- disappeared during this event. It is also not controversial that most -- albeit not all -- were of large body size. There is somewhat more controversy about the rapidity with which these losses occurred, and we have had some discussion on that point today. The most recent radiocarbon dates -- like those reported by Russ [Graham] and his colleagues -- seem to indicate that the event was almost laserlike -- a point source in time; that this event in the Americas might have gone to completion in a period of only four centuries.
If Russ's figure is applicable to all of the continental New World -- and this is, of course, open to question; the data right now are all North American -- this would yield a rate of loss of one species every three years during the course of the extinction event, which, I think we can all agree, is a prodigious rate, indeed. Paul Martin's somewhat longer time span for the extinction of 1,000 years would yield a rate of one species every eight years, still very high.
Now, to put this in context, compare these figures with the empirical rate for mammal losses in the continental New World in the ensuing 10,000 years -- by that I mean the entire Holocene, save for the most recent period, the last 500 years. By "empirical" I mean bones out of the ground -- not a gratuitous application of the species area curve in order to divine the number of losses that should have happened. Using this criterion, the rate is very easy to compute -- it's zero. Except for the most recent period, continental New World mammals seem not to have suffered any species-level extinctions in the Holocene -- at least, none that have been dated or detected by good-quality radiometric dating.
Now, what about the empirical rate for the last 500 years? It's hard to say, because there's a number of possibly extinct, maybe extinct, and so on and so forth, type references in the literature, but the number that Clare [Flemming] and I came up with -- which is justifiable by any real evidence -- is very low: somewhere between 10 and 1. Don't get me wrong -- there were losses, there were enormous numbers of losses in the New World. But, for the past 10,000 years, they have been almost exclusively concentrated in the West Indies and other islands, not the mainlands -- despite extraordinary levels of endangerment caused by recent human societies.
We've already heard today -- that is a Megaladapis, not in its usual form -- one of the sloth lemurs . . . We've heard today about similar patterns of loss in prehistory in Madagascar, Australia and elsewhere, although few of these instances are so well dated as the American, and none matches them in the sheer number of disappeared.
The examples could be multiplied, but in each one I've looked into, I'm impressed by only two things. First, at least some of the extinctions occurred immediately subsequent to first contact -- that is, when humans came into new lands, places where they had never been previously resident -- and, secondly, the better-dated ones seem to have begun and ended with dramatic rapidity, leaving in their wake a multitude of losses, connected in the case of mammals, at least, only by the tendency for large things to die out.
If humans had something to do with these losses, as this slide suggests, why was extinction pressure on species not maintained into later times? Paul would argue, I think, in the case of the Americas, that the explanation lies in changing cultural practices -- indigenous peoples may have given up big-game hunting, because game grew very scarce. Perhaps they gave it up because they increasingly turned to crops and other forms of food procurement, or for other reasons not obvious in the archeological record.
Whether or not humans arrived much earlier than Clovis times, this transition -- this cultural transition, if it existed -- had to have happened in the space of a few centuries, at most, to account for the pattern, and it must have occurred virtually simultaneously from the North Slope in Alaska, down to Tierra del Fuego. In the case of Madagascar, this kind of argument, I think, is even more hard-pressed. If the ancestors of the Malagasy laid waste to the interior of the island and thereby precipitated at least two dozen vertebrate extinctions, as some argue, then why have their modern inheritors -- widely, if somewhat unfairly, regarded as the most destructive of island dwellers -- failed to precipitate a single lemur extinction in the last 400 years? Surely, it cannot be from either means or motive.
For these and similar reasons, Preston and I conclude that something fundamental is being overlooked. What if the cause of the Late Quaternary extinctions was not what humans were purposely doing in the new environments they came into, but had something to do with what they were passively bringing in? Something which did not affect them appreciably, but was spectacularly lethal to the naïve, unhabituated faunas that they encountered. In short, let's consider disease as a possible cause of extinction.
In several regards, what we know of many first-contact extinctions is highly reminiscent of conventional disease processes . . . The spikelike loss of species throughout their range, within a narrow period, is very similar to the ballistic rise and fall that you see in a standard growth-and-decay curve, as Doug [Siegel-Causey] just illustrated for you, for infection rates in panzootics . . . or pandemias, in the case of people. Once successive mortality has pushed the interbreeding population size below a sustainable level, extinction looms -- just as Norman [Owen-Smith] reminded us this morning. That even though the processes that we envisage for extinction might be radically different from one another, the way in which they actually have to happen would seem to be the same. You have to collapse the population, and you have to do it quickly. You have too low a replacement rate, you get a bottleneck; . . . they don't pull through, you get extinction.
In short, this is overkill by other means. Indeed, the parallel that must now be acknowledged is that Martin's blitzkrieg is essentially descriptive of the epidemiology of a disease of unusual lethality. Blitzkrieg's basic argument is that we should believe that humans acted in the manner of a pathogen. Well, let's deny the metaphor and assume the pathogen was real.
Can we explain some of the agreed-upon facts of Late Quaternary extinctions any better? For example . . . why were there so few later Quaternary extinctions in Asia and Africa during the relevant time period? The overkill hypothesis proposes that the Afro-Asian megafauna developed behavioral mechanisms to deal with the evolving human predator. As a result, they did not experience losses at nearly the same scale as did naïve faunas elsewhere in the world. I would argue that the mechanisms were, instead, genetic and immunological -- living in the same environments over millions of years, humans and other mammals were exposed to, and coadapted to, the same disease pool.
Why were large species differentially affected? In most disease contexts, it is the youngest as well as the oldest members of a population or species that are at risk. In first-contact situations, high wastage of immunologically immature newborn and young individuals might have had very severe effects on population survival, especially if the disease could repeatedly arise from reservoirs -- as Doug illustrated with HN influenza strains. If it could repeatedly arise, then the populations could be hit again and again, until all susceptible individuals were gone. This scenario, I suggest as well, would be more serious for large taxa, which characteristically have lower reproductive rates, longer gestations, and smaller progeny sizes than small-bodied taxa.
Now, why did mammal extinctions level off after 10,000? Anthropogenic blitzkrieg assumes that later Indians had a reduced interest -- of some sort, at least -- in excessive hunting, but offers no plausible reason why their forebears induced dozens and dozens of losses in the mammal fauna, while they themselves did not prompt a single identifiable one in the next 10,000 years. Pathogenic blitzkrieg removes the apparent paradox by offering the alternative explanation -- that, after catastrophic first-contact losses had subsided, what was left of the original megafauna -- the musk oxen, the bears, the deer, the elk, and so on -- consisted of nonsusceptible species and the taxa that managed to pull through disease-induced bottlenecks. Given this scenario, there is no need to assume anything at all about aboriginal hunting practices as a cause of extinction; they were probably irrelevant to the process. That, in a nutshell, is the outline of our argument for the hyperdisease hypothesis -- that hypervirulent, hyperlethal disease may have been a factor in precipitating certain first-contact extinctions.
For this hypothesis to deserve a place at the table of apparent or possible causes, it's got to meet certain criteria, and it has to have a plausible testing procedure -- so let's look at some possibilities. Since we have no idea what the hyperdisease pathogen or pathogens might have been at this stage, let's compare known diseases -- using the same criteria -- to illustrate the fact that many known pathogenic entities have at least some of the characteristics of hyperdisease.
Now, . . . a listing of four "hyperdisease criteria," as Preston and I call them, which I will now go through sequentially. And all that I really want you to notice is that the pattern of crosses, of X's, is full and complete for the first couple of criteria -- but, as we move over towards the distance here, the far side, you'll notice that the number of X's fall out. There are not many diseases, in other words, that have a full complement of hyperdisease criteria.
Let's just go through them. Criterion number one -- there's got to be a reservoir species with a stable carrier state. What does this mean? A stable carrier state . . . is simply a reference to the fact that the pathogen is or can be borne by some other species as an inapparent infection -- in some form transmissible, in the case of viruses, by shedding, so that something else can get it. Since we're tying in extinction and disease with the human diaspora, it follows that either humans were the carriers themselves, or animals that ran with humans -- or the parasites of one or both -- and, for the reasons that Doug's already given, humans are extremely unlikely. Presumably, the hyperdisease pathogen wouldn't cause lethal disease in its vector, as that would end its career right there. So that is one kind of criterion.
Secondly, to cause extirpation of all populations of a species, the disease process has to work fast, before attenuated strains or general genetic and immunological resistance begins to appear, and that genotype becomes more common, and the population pulls through. The best way to do this is for the pathogen to affect, at least potentially, any and all age groups. Notice that diseases like AIDS would be highly unlikely as killer plagues in this sense, because AIDS is normally transmitted by sexual contact, implying that, for the transmission sequence to work, you're talking only about a small fraction of the population -- . . . those that are sexually active.
Criterion number three . . . Third criterion -- excessive mortality -- and I mean excessive. Along the lines of Spanish flu 1918, squared. If you take a look over on the chart on my right here, you'll see that the field is now noticeably thin -- we're now into the third category. There's not a lot of them that have a lethality rate that is so high that it would stand some chance of beating the replacement rate by several orders of magnitude.
Well, there are some near examples, however, and I can mention just a couple of these . . . just to give you a taste for both the problems and prospects of this line of thought. The myxomatosis panzootic that was introduced purposely in Australia in the 1950s took advantage of the fact that cottontails in the New World have endemic within them, carrying as an infection, the myxoma virus, which causes in them no more than benign tumors. But within European rabbits, a different genus, it causes very quick, very lethal disease. The hope was that, by introducing myxoma into Australia at the right time of year . . . that you would induce a nemesis plague and get rid of your rabbit problem.
Well, Aussies being Aussies, the problem was that the disease got out a little bit too early, at the wrong time of year. . . . The myxoma virus is largely carried by arthropods. And, consequently, when you get seasonal lows of arthropod populations, you get more rabbits surviving . . . and so on and so on. Then all the bets are off.
A different effort was tried in Australia. . . . This time they're trying to introduce another highly lethal disease -- this one hemorrhagic fever, produced by rabbit calicivirus disease. Again, the virus got out early, at the wrong time of year. The bets are that the calicivirus project is going to be ultimately ineffective as well. People from Richard Holdaway's country are apparently a good deal more careful about this sort of thing, if you saw last week's Science. The leading scientific body there condemned what the Australians were doing, and they don't want to do it in New Zealand. So people clearly are worried about the problem of hyperlethal disease.
Now, I'm going to try and get my fourth criterion . . . there we are. The worst of all, the hardest -- the ability to infect multiple species. The way that Doug is approaching his equivalent theory of hyperdisease is to question whether or not you can actually have a hyperlethal agent, because it's going to burn itself out. Preston and I are making a slightly different hypothesis, one which we think, in fact, is a sine qua non, if this is to have any applicability for something like the Late Pleistocene extinctions in North America. That is, that in order for this notion to work, you have to have multiple, simultaneous, epizootics in a large number of species. They have to be sharing the disease-provoking organism in a way that is unprecedented. You have to have a disease that crosses species lines with incredible ease.
Now, once again, on the basis of analogy with what we know about modern diseases, are there any prospects? Well, there are a few -- but, again, they have limits. Canine distemper in East Africa right now is raging through lions, bat-eared foxes, African wild dogs, causing immense concern -- particularly with respect to the last species. How did these animals get it? Probably from infection out of domestic animals, particularly dogs.
Rinderpest, another morbillivirus, as David Burney briefly mentioned, caused enormous losses in native ungulates in eastern and southern Africa when it was introduced via Asian cattle, introduced . . . by the British in the 1890s while they were fighting the Mahdi in the Sudan.
The article over on the far left here, in the Tuesday Science Times, concerns the spread of tuberculosis in Kruger National Park, in South Africa, where right now you have numbers of buffalos, cheetahs, lions, kudus and even baboons dying of this disease. Tuberculosis is, in fact, not plausible as a hyperdisease according to our criteria, because it doesn't kill enough, and it doesn't kill fast enough. But it does have the feature -- the different strains and species of Mycobacterium -- evidently could potentially affect virtually every mammal and bird.
Well, all of this is well and good, but how about some really serious reality checks? I'm only going to go through a couple, because I have to adhere to the schedule, just like everyone else. Let's ask the question: Is there any empirical evidence that hyperlethal diseases could maintain themselves in nature? One objection that any evolutionary biologist out there should be thinking of is: How could natural selection possibly favor the emergence of hyperlethal pathogens, since, in killing their new hosts, the pathogens would undoubtedly go extinct as well? It's usually said in textbooks that well-adapted or successful parasites have evolved into a state of benign correlationship with their host, because there is no selective advantage in killing them. However, as Robert May has pointed out, the notion of a selective drive toward the benign state is far too general. At any one time, selection might favor increasing virulence, increasing convergence toward benignity, or any combination of these and other factors.
Indeed, a pathogen of unusually great lethality might survive, as the theorist P.W. Ewald has recently suggested, so long as its vector, or method of transmission, were so pervasive in the environment that it could afford to continue spreading, even while it's lethality was operating at an incredible pitch. This is the so-called "burning bridge strategy," and one such method would be aerosol transmission and simultaneous infection of multiple species, just as we suggested for hyperdisease.
Well, no one argues that plagues don't kill -- obviously, they do. The only relevant question, ultimately, is: Can they kill enough to cause extinction? Which leads me to my last two plausible objections -- right off the bat, at least; I'm sure I will hear others -- to hyperdisease. Is there any evidence that disease can cause complete extinction of species? Is there evidence from historical times, for example? Or that disease of this nature can propagate through multiple species at once?
Well, we've looked at some near misses, like myxomatosis and rabbit calicivirus in Australia, and so on. These are ultimately uncompelling, because they didn't do the job. Like Melanie [Stiassny], I feel a certain sadness in not being able to count these as having gone down, but, obviously, that is not entirely true. And if there's nothing in recent experience that we can point to, that at least comes close, then, very probably, Preston and I are wasting our time trying to dig out evidence of such an agent in very much earlier times.
Well, there are not many pertinent examples to point to, and I'm only going to go through very quickly a couple, just because of their natural historical interest, more than anything else. Two rats used to live in Christmas Island -- a small isolated island south of Indonesia that was not permanently populated by people until the 1880s. It was noted that, around the turn of the century, from a state of being extremely common -- virtually underfoot -- they disappeared utterly in the space of five or six years. Charles Andrews of The British Museum -- with no evidence to back him up -- said at the time: Maybe they died out because of an introduced trypanosome from ships rats.
Another possibility briefly referred to by Helen [James] and by Doug is the case of the loss of Hawaiian drepaniids. In fact, this might be, very possibly, one of the best examples of all. I'd like to make the point about the introduction of avian malaria, and whether or not some of these species died out extremely recently (and by that I mean within the last century or so) to make the point that the only way in which death would have come to these birds is when the vector -- the culex mosquito that is capable of carrying avian malaria -- was itself introduced. This is the reason, from our perspective, why it doesn't matter whether malarial birds were migrating in from North America to the Hawaiian Islands -- in them it was a dead-end infection; there was no way to spread it to the native birds . . . or, at least, to the endemic birds. Once the vector was in, however, all of the tables were turned.
Papers by Wood Jones and many others cite anecdotal evidence for the possibility that the thylacine, mainland populations of eastern quolls, and so on, met their end around the turn of the century, or slightly later, as the result of introduced disease. Indeed, Rounsevell has even suggested that the thylacine -- everybody's poster-child example of an animal brought down by severe overexploitation -- may have, instead, been finally killed off by a disease. However, to my knowledge, there is no backup in the form of evidence for any of this.
Right -- last point. Is hyperdisease testable? If by that is meant, Can we envisage a series of empirical discoveries that would be consistent with the hyperdisease model? -- yes, I believe it is testable. But until we've had the opportunity to start undertaking some of the necessary laboratory investigations, we're a long way from being able to present decent results. The best evidence, of course, would be genomic, using the modern techniques of molecular biology. Preston and I are currently trying to develop methodologies that would be appropriate for this purpose, and we've been vastly encouraged with the work that Doug has done, and other scientists have done -- for example, partially retrieving the genome of the Spanish flu 1918-1919, which was widely reported in the press. And we hope, within a fairly decent time, we will either have very productive results to present for you, or give it up as another half-baked idea.
I want to just conclude by making the point that the title of Paul's and Henry Martin's book in 1967 was Pleistocene Extinctions: Search for a Cause. Well, it's quite clear, I think, we're still searching. Whether or not hyperdisease joins the list -- whether it will edify the search or merely encumber it with another half-baked idea -- remains to be seen.
I would like to just say, in closing, that one way in which hyperdisease, I think, does a better job than blitzkrieg is that it offers a way to imagine how an extinction episode could pass through a large area -- a continental-sized area, for example -- with dramatic rapidity, without any reference to local topology, to vegetation cover, or the habitats in which people at least prefer to live.
And that's where I'm going to end it.