Pleistocene Extinctions: Chronology, Non-analog Communities, and Environmental Change

Russell W. Graham, Denver Museum of Natural History
Thomas W. Stafford, Jr., University of Colorado
Holmes A. Semken, Jr., University of Iowa

Real Audio Recording   

presented by Russell Graham

Thank you, Dave, ladies and gentlemen. Well, it's a pleasure for me to be invited here, but I was a little perplexed, because I think many of you know me. I've argued for environmental change as the cause of extinction -- and, of course, the title of this is "Humans and Other Catastrophes." But when I got here last night it became clear to me why I am here. As Ross McPhee gave us an orientation, he said he wanted to spark controversy and discussion; so what better way to do that than to take some environmental red meat and throw it to the lion pit of the human-impact people? Of course, being a basic coward, I didn't come by myself, and my colleagues Holmes Semken and Tom Stafford are here.

Let's start with the first slide. . . . The question is: How do we go from one landscape, where all these animals were at, to another landscape, where they do not occur? The first part of this is to talk about the chronology of the extinction. This is a project that Holmes Semken and Tom Stafford and I have been in for a number of years now, and it really gets down to bone-dating. Because one of the ways to best determine extinction is to date the bones of the extinct animals themselves.

This is a slide to show you the dating of one bone specimen from the Dent site, a Clovis site in Colorado, by Dr. Tom Stafford, and it shows you the different types of dates you get based upon the techniques you use, as well as improvements in the accelerator. That is, that gelatin and collagen are the basic dates you would get from a commercial lab, and you can see where they range. What Tom Stafford has done is to refine dating, by looking at a purification of the collagens -- the XAD collagens -- and you can see that the dates from the same bone shift. But the main thing is that, in these early dates -- and this relates to the mass spectrometer itself -- there's a pretty high standard error of about 400 years here. With improvements, we've actually narrowed this standard error to about plus or minus 50 or 60 years. So we can produce very high-precision bone dates today.

This raises two questions. One is that, basically, dates after 1982 were the best types of dates we're getting. Dates before that really probably overlapped the extinction event itself. And, therefore, many of the dates that have been run earlier probably should be rerun with new techniques to understand the actual age of the specimens.

What we've done is select a variety of extinct taxa, and the selection process was to try and locate the highest stratigraphic occurrence of these taxa, and then date them. We've dated over 100 specimens now, 17 different genera, and seven of these are new. And what you see is the extinction and the latest occurrences are right in this range here. And the range of the extinction, as we see it -- the names are over here -- these are a series of cats; can't read them all myself, but a variety of extinct taxa. But the range of extinction, as we've determined it from our data, date from about 10.8 to 11.2 radiocarbon years BP. And these are a certain set; here's the other set. You'll notice the outlier here. This is the Wrangel Island mammoth. Tom redated that and is concurrent with the Russian and Arizona dates, so it does look like they really did survive to about 4,000 years. But, again, you see right along in here is a very tight window of extinction.

The interesting aspect of this, though, is that maybe there were actually two waves of extinction, which were very tight, and that is the proboscidians versus the non-proboscidians. And what we've done is graph them here for you. And here are the proboscidians -- these are the mammoths and the mastodons. And it looks like, from our data, these are the last animals to go extinct -- they're the last survivors -- and that many of the other animals went extinct before them.

And this has three important implications. First, it would tend to refute the keystone species hypothesis of Norman Owen-Smith, based upon proboscidians being the keystone species. If another species actually served as the keystone, then we haven't refuted the concept in general. But it looks like the mammoths and the mastodons really are the last ones to go out, whereas they should have been the first ones to go out if that was true.

The other important point, which is very interesting, is that, as Gary Haynes and Paul Martin have indicated, most of the Clovis sites are associated with mammoths or mastodons, and not other taxa. So we are now beginning to wonder if a good part of the extinction was not over by the time Clovis people actually got here. And that the last animals they had to deal with were these mammoths and mastodons -- and that's why we find the really high association. Also, I think, any environmental model would tend to predict that the mammoths and mastodons would be the last to go extinct because of their large size, because many of these large animals are buffered by environmental change.

This just is a summary slide of talking about the associations. These are all Clovis sites that are stratified or buried. We're missing one here, the Aubrey site. In almost all of these sites you can see the association is with mammoth and mastodon. At these sites you find other taxa periodically -- horses and camels -- but there's a real question about their association. And, in fact, we're beginning to wonder now a question about their age; and I think people are going to have to go in and start doing species-specific dating of individual skeletal elements to determine contemporaneity, and I'll show you that in a later slide.

The question about climate change has been raised by several people, and this is the record from the ice core. And one of the things that we have to realize, or to demonstrate with climate change, is that there was a significant climate change at the end of the Pleistocene during the time of extinction. If we cannot show that, then we're in serious trouble. The other part of that question, which has been raised by a number of speakers here: Was the climate change at the end of the Pleistocene significantly different than any other climate changes?

What we see here is the ice-core record, and there are a couple things to be important. These are depths and not ages. And one reason I'm using this is because here's one of the triggers, we think, the Younger Dryas. The age of the Younger Dryas is debatable. In the abstract that I published for the symposium, I indicated the age of the Younger Dryas at 10,000 to 10,600 years. This is based on a paper published by Svante Bjork and his colleagues in Science last fall. If that's the case -- if that is the age of the Younger Dryas -- then the Younger Dryas cannot be the trigger for the extinction, because the extinction had already occurred.

We could then back up and look at some of these others -- the Allerod and the Bolling, which are major warming events. And, if we shift to that time scale, these would be concurrent with the extinction. However, there is a paper in press right now, in Nature, by Hugen, Overpeck and others, where they've recalibrated the radiocarbon curve, and they're suggesting that the date of the Younger Dryas is actually between 10,600 and 11,200. If they're right, that then is coincident and could be the trigger factor for the extinction event.

The other two important points to note here is that you can see -- these are almost annual increments from the ice core -- you can see a lot of fluctuation in climate throughout the Pleistocene. And then, bang, a major change -- and then, through the Holocene, a much more stable climate. So the Holocene is very different from the Late Glacial climate here. And I think that this fluctuation in climate has a lot to do with the heterogeneity, or the patchiness, that I'll talk about towards the end -- things that Gary Haynes referred to.

The other important point to be made -- this is a stratigraphic profile at the Murray Spring site; Paul Martin showed you slides of that -- this is a classic Clovis site. But to again point out that climate change is coincident with a Clovis interval. At this site what you see are Clovis artifacts on a surface with mammoth tracks and fire pits, dated at about 11.2. But also at that site, at the same time period, Haynes has interpreted the stratigraphic sequence to indicate a major drought event, which he's called the "Clovis drought," and thinks is very widely spread throughout North America. I think that's debatable. But, clearly, at least in the Southwest, and maybe the Great Plains, there is a major climate change coincident with Clovis. And I think it's going to be very difficult to separate environmental models versus human overkill models on dates alone, because they're two really coincident factors here in North America. We may need to go to other continents to run those types of tests.

The FAUNMAP as a database -- and this is what we see here -- of Late Quaternary sites, dating from about 40,000 to 500 years ago -- it's a database that was assembled by myself and Ernie Lundelius, and a variety of other scientists -- of mammal sites in the United States. And you can see, basically, the geographic coverage that we have here. What this does, though -- it's a relational database linked with a geographic information system -- and we can look at changes in the distribution of species through time.

And there are two competing models of how organisms -- plants and animals -- respond to environmental change. One I've termed the "community unit," and there's a variety of terms that have been applied to this. And what you see are four species that are together at time one. We get some sort of environmental change. At time two we get the same four species, but they've basically shifted their geographic distribution to track the climate.

In the other model, which has been referred to as the "individualistic," you see the same environmental change, but in this case the species respond individualistically, in accordance to their own tolerance limits. They migrate in different directions, they migrate at different rates, they migrate at different times. We can actually test these models by looking at the FAUNMAP data and see what happens. And so what we've done is map a series of individual species with different time periods. By "Full Glacial" I'm referring to a time period of 15 to 20,000; by "Late Glacial" I'm talking about the time period from 10 to 15,000. This is Phenacomys intermedius; this is the heather vole. It's one of those species that John Alroy mentioned as a non-analog. You can see at Full Glacial it's much further south. By Late Glacial time it's beginning to retreat north. This is its modern distribution today, and here's the ice-sheet boundary during maximum glaciation. This species is gone from the United States after Late Glacial -- so it's a species that's moved north, moved north rather rapidly.

Another species that moves north is the red-backed vole. Again, the same time periods here, and you can see it's much further south during the Full Glacial. By the Late Glacial, it's moved further north. And then, by the Late Holocene -- which is from about 500 to 4,000 years ago -- you can see it's moved almost into its geographic range, but there are a few sites outside. So it's a species that's moved north, but at a very different rate than Phenacomys. So this fits with this individualistic idea of species dispersing at different rates.

Also we can look at different directions of dispersal. That is, here's the least shrew, Cryptotis parva, and here's the northern pocket gopher, Thomomys talpoides with their modern distributions. Here are their fossil distributions, again, for the different time periods. You will notice that Thomomys, the northern pocket gopher, has extended its range much further east, and Cryptotis, the least shrew has extended much further west during the Late Pleistocene. Also of significance is the area of overlap of these two species. They overlap in a broad range, basically running down the front range of the Rocky Mountains. By the Late Holocene, you can see that they have basically retreated into their modern distributions, and overlap only here in the Dakotas today.

So these are two species that again have moved in different directions, east to west, and then there are a series of species that just remained where they were -- and many of these were southern species. If you apply the community unit model, you would expect these to be pushed down into refugia, down here in the southeastern United States. But here are the fossil sites that date between 10 and 15,000 years ago, and you can see it's as far north then as it is today.

What does this create? This creates what we call "non-analog faunas," or "disharmonious faunas." These are faunas that are in the fossil record -- and these were, again, talked about by John Alroy -- where you can see a fossil site here in western Iowa. We have a radiocarbon date on charcoal at about 24,000 years, and the remains of all of these animals -- these are their modern distributions -- co-occur in the same stratigraphic unit at that site. They're called "non-analogs" because there's nowhere in the United States -- or, actually, in the world -- you can go today to find all three of those animals co-occurring. It's, again, the reason of individualistic response. Those species occurred here 24,000 years ago. Environmental change caused them to disperse in different directions, and some of them stayed right where they were.

The question of this, however, for it's been for a long time, is that you can get these types of associations artificially. There are many different taphonomic processes that can cause mixing of faunal remains. Or time averaging of faunal remains. So that you could get the remains of these animals in the same stratigraphic horizon, but they would be of different ages, so they would have no biological meaning. The other explanation is the explanation that they represent communities and habitats that no longer exist.

It's been only until the last decade or so that we could actually directly test this. We've argued for this for a long time; most of these arguments have been inferential. They basically occur in the Pleistocene; we don't see them very much in the Holocene; they're worldwide; they're not restricted to specific taphonomic settings.

But now, with the new methods of radiocarbon dating bone that have been developed by Tom Stafford, we can date individual specimens as small as rodent jaws or rodent teeth. And so we can select these from the same stratigraphic level, and here's what we end up with. Here is the Cheek Bend site, which includes an Archaic component in Tennessee with the modern distributions of two animals -- the plains pocket gopher, Geomys bursarius, a great plains steppe species, and the yellow-cheeked vole, Microtus xanthognathus., a north boreal forest-dwelling species. These two are separated by over 600 miles and neither ranges into Tennessee today. But here are the radiocarbon dates on individual specimens -- actually, individual teeth of those two animals. One dated 14,120, plus or minus 70; the other 14,120 plus or minus 80. There's no question in my mind that these animals lived at that site at that time period.

We've done this for a number of sites now -- not just Cheek Bend. We've got other taxa involved in this. We've done it for Russian sites as well as other North American sites, and there's no question that these non-analog faunas did exist. There are problems with taphonomy, we've also found that -- because, at a number of sites, we have found mixing to be a problem -- and so it's going to be a case-by-case, site-by-site type of analysis to really sort this out. But the question is these did exist. They're not only restricted to mammals -- there's a vegetational record. A paper published in Geology, by Overpeck and others, summarized this. They occur for molluscan faunas, and they occur for insects, as well. So I think what we're seeing are actually non-analog communities -- non-analog ecosystems -- during the Pleistocene that no longer exist today. And the significance of this is that we see, then, the destruction of these habitats.

The creation of these non-analog faunas is mainly by the addition of northern taxa to resident taxa. And I think this is a point that John Alroy made very clear, and that he saw very few, the percentage of actual numbers of species that would be involved in these would be small. Because it's mainly these northern taxa -- the voles and all of the northern forms -- coming down, but not displacing southern forms, but being integrated with them to create these. But what we do see is that these non-analog faunas, non-analog vegetation, non-analog insect assemblages are geographically widespread. Not only here -- you'll find them in Australia, you'll find them in Europe, Eurasia, and also in South America and, to an extent, in Africa, but not as much. So what we're seeing, then, is this ice physically displaces these northern forms, plus the cooler climate sends them further south, where they integrate with resident faunas.

Now, I think the driving force behind the extinctions is habitat destruction and habitat loss -- what we're talking about today with animals. I think it's the same thing, but it's done naturally. What we can do is take the FAUNMAP database, and we can categorize living animals by different habitats that they occur in. Mammals are pretty eclectic -- very few of them are really restricted to one or two habitats; they occupy a variety of habitats. What we've done is score them, based upon their modern distributions, on the types of habitats they prefer, and rank them by the most preferred to the least preferred. And then we've cut it off with the three different habitats. We can then create maps showing the different habitats.

And one of these is that I'm showing you the prairie -- or, basically, the Great Plains. What we've done is calculate the percentage of prairie taxa in the faunas -- and you can see the distribution of faunas here -- and then contoured those. So at this line here is the 20% boundary. It turns out that line actually is the boundary between the grasslands and the forest today. Tom Webb has done the same thing for vegetation -- generated the same types of maps -- and it turns out, very interestingly, he gets the same sort of correspondence, a 20% line. The interesting thing right here is this bulge, because that bulge is known as the "prairie peninsula," a Middle Holocene phenomenon. This is the Late Holocene.

The other important thing to see is right here in the core of the great grasslands -- you see 60% and 80%; also in the northern part of the grasslands -- 60% and 80%. So very high concentrations of prairie taxa in the shortgrass prairies. What did the Pleistocene look like? Here's the Pleistocene. You don't see any 60% or 80% -- the highest percentage is 40%. And the reason for that is that I think parkland environments and savanna environments, a greater mix of woodland and grassland prevailed during the Late Pleistocene than we see today. Essentially, the Great Plains region is in evolution during the Holocene.

Another type of environment is the boreal forest. The boreal forest of the Pleistocene was different than the boreal forest we see today. This is a little time machine that we had at the Illinois State Museum, so we could go back and capture what it looked like for you. It's a painting, though based upon pollen reconstructions from western Missouri. And what you see is a boreal forest, but a very open type of boreal forest -- not a closed, dense, spruce forest like we see in the boreal regions today. And here's one of our favorite animals, the mastodon -- and this is the animal that's associated with this type of pollen structure throughout the Midwestern United States.

So what happens to this habitat through time? These are a series of maps that were generated by George Jacobson, Eric Grimm and Tom Webb in 1987, and what they're doing is mapping different types of vegetation. And what you see is the yellow represents a certain percentage of pollen of sedges, the pink represents a certain percentage of pollen of spruce, and this orange color is that spruce-sedge mixture.

Now, it's interesting to look and see what happens to this orange during a time series; and this is the sort of preferred habitat, to an extent, of the American mastodon, that's associated here in Missouri. And here's the Full Glacial at 18,000, 14,000, 12,000. And you can see what's happening is, this orange strip is shrinking. Here's 10,000 -- it's just a little dab here. By 8,000 it's gone completely. So you see the greatest change between 10,000 and 12,000 years ago -- so, again, this is the time of the extinction. So, basically, what we're seeing is a shrinking of the habitat that this animal preferred during the Late Pleistocene -- with its disappearance. But it then reemerges during the Late Holocene, but it's a very small area that you see. So, again, it's this environmental change, and these are climatically driven. The vegetation's not being changed by Clovis people in any way that I know of.

Another important aspect of this, again was brought up by Gary Haynes, is environmental heterogeneity -- is that we've argued for a long time that we think that Pleistocene environments are much more heterogeneous, and that's why you get these non-analog faunas. You get a greater mixture of habitat types in a greater mosaic at that time. With the FAUNMAP database, we've been able to test that statistically. What we did was to divide the country up into grid cells of about 100 square miles, put all the faunas in the grid cells -- so we had a faunal list for each grid cell -- and this was sort of averaging out some of the taphonomic factors that you see in site-specific databases. We then compared each grid cell to all other grid cells, using the dice-similarity coefficient, and then we calculated a mean average for the dice-similarity coefficient based upon distance. And distance is expressed here as a bin, and a bin is about 250 kilometers in size. So when you get out here, you're talking about 1,250 and then 2,500 kilometers out here.

What we then did is plot the mean dice-similarity coefficient against distance for two different time periods -- for the Late Pleistocene in this curve versus the Late Holocene in this curve. And what you see is that the similarities in the Late Holocene are greater than those of the Late Pleistocene. And they're statistically significant up to about this area -- because you could see the error bars here. And we interpret this to indicate that the Late Pleistocene environments were much more heterogeneous -- or, in quotes, "patchy" -- than the Late Holocene environments.

And we then checked this against randomness and ran a series of Monte Carlo, where we developed faunas, and ran the statistical Monte Carlo. And basically, the mean dice coefficient stays very similar, and there's complete overlap here. So we don't think this is some sort of artifact due to our sampling strategy; and we think it does represent a more heterogeneous environment.

Well, what's important about a heterogeneous environment? I think a heterogeneous environment allows species to have a much wider distribution. And as you reduce that heterogeneity, you then reduce the geographic distribution of these species. And as you reduce the geographic distribution of species, the probability of extinction goes up, for a wide variety of reasons.

And it's frequently thought: Well, during the Late Pleistocene, we see these megafauna just everywhere -- but that's not the case. I think they were very restricted during the Late Pleistocene, and I'll show you a series of taxa. Here's the giant beaver -- it's a beaver about the size of a black bear. And you can see its distribution here was basically in the eastern U.S. Here's one of Paul Martin's favorite animals -- Nothrotherium shastense ... it's a ground sloth. You can see it's basically a western species. Remember, now, we don't have the Mexican data, nor do we have the Canadian, so we're restricted here to the U.S., so it is somewhat biased. Also here are two different -- this is a peccary, Mybhyus which again is sort of a southeastern taxon during the Late Pleistocene -- and these are all extinct forms. Here's Navahoceros fricki, which is the mountain deer -- which, again, has a relatively restricted geographic distribution.

And then, finally, here are two animals -- Cervalces scotti, the stag moose, which occurs basically here in the Midwest; and it occurs in that environment that's very similar to the mastodon. Both the giant beaver and Cervalces scotti occur in a lot of these open boreal forest environments. And then here's a large armadillo, Holmesina septentrionalis, which is, again, sort of a Gulf coastal plains species. It actually lapped down, will go into Mexico, Central America, probably. But, again, these species were not really widespread. But if you'll look at maps, and you try to map this -- we haven't been able to do this, because we don't have the database yet. But if you do it for a few species, you can see that many of these species were widespread during earlier parts of the Pleistocene, and that throughout the Pleistocene what we're seeing -- or what we think we see -- is range reduction in these species.

And so what we would suggest is that you're having environmental change, you're having climate change throughout the Pleistocene -- that this climate change is continually sorting out this heterogeneous environment, and making it more homogeneous; that the ranges of these species are collapsing through time, and that we're seeing habitat destruction, in some cases, by these climatic changes. And that it basically reaches a critical level at the Late Pleistocene, is that these ranges have been reduced so much that the next big hit -- which would be the Younger Dryas or the Bolling-Allerod -- is basically the trigger at a threshold effect, which causes this extinction.

So it's not necessary, with this model, to argue that the Late Pleistocene, or the Pleistocene-Holocene transition, was a unique event. It doesn't have to be, and it may not have been. But it's an event that acts as a trigger. However, I think there still is a lot to learn about the Late Pleistocene-Holocene transition, because if you look at some of these earlier faunas, the faunas are different -- and I'm not convinced that the changes before that really were the same.

And, so, with that I'll conclude and say that the problems that plague the Pleistocene fauna are problems that plague us today -- but, clearly, humans have been more of an impact in the historic than I think in the prehistoric.

interview with Russell Graham | bio