The Power of Pleistocene Hunger-gatherers: A Forward and Backward Search for the Evidence about Mammoth Extinction
Gary Haynes and B. Sunday Eiselt, University of Nevada
presented by Gary Haynes
We've known for 60 years that these well-made chipstone spear
points, called Clovis points, are associated with some animals
in the fossil record in North America. We also know that there's
a large number of these spread out through a range of about 3,000
miles east-west, and almost 3,000 miles north-south. We know
that there's about 15 sites where these points, these spear points,
are associated with mammoth bones -- and maybe, in a couple of
cases, with mastodon bones -- which is not very many. There might
be 55, 60, 70 different individual mammoths that died, or were
scavenged, by Clovis hunters at these sites.
The question has been raised: Is this a good enough fossil record
to indicate overkill or hunting by humans? Or is this scavenging
by Clovis people -- finding animals dead during a period of major
climatic change? This is sort of the conventional wisdom amongst
us archaeologists. There may not be too many of us here. Dave
Hurst Thomas is one, and I'm one, and we tend to try to sound
a little bit more scientific, maybe, than we should.
But the conventional wisdom is -- or at least it was when I was
in graduate school -- that this was a period of great stress for
some of the big mammals, or most of them, in North America, and
it's in this period that Clovis abruptly appeared -- either dispersing
quickly, or technology was introduced amongst a preexisting population.
And this was good enough to drive these mammoths and mastodons
over the edge into extinction. I'm going to try to limit myself
only to mammoths and mastodons here, and not the rest of the fauna.
Some of the ideas, though, that were part of this conventional
wisdom have been questioned in the last decade or so, or probably
in the last five years. One of the problems with this model is,
there are no ethnographic examples of people hunting animals to
extinction. They also don't do a lot of other things that we
assume Clovis people did. This is what has been called "ethnographic
snapshot" -- the use of ethnographic analogy to sort of limit
what you can say about the archeological record. This is one
criticism that's been leveled at the idea that Clovis people did
the hunting to extinctions. There's been others, including some,
I think, misapplied examples of foraging theory. This is one example -- marginal value theorem -- which is oftentimes
probably misunderstood by archaeologists as the concept that people
stop hunting certain animals when they get hard to find. So that,
in other words, as animals are becoming extinct, people would
no longer select those animals as prey. And I'll come back to
these two ideas in a few minutes, after I go through a few more
Recent revisions of Clovis, now, have us to believe that, in fact,
mammoth hunting was not a major subsistence activity amongst Clovis
people. In fact, big-game hunting at all was probably not a major
thing amongst Clovis people. I've written this as . . . revisions
as two hyphenated words, because, to me, these are only visions;
these are not scientific arguments about Clovis. But I want to
run through these.
Clovis mammoth sites are supposed to be rare. I think we can
refute that pretty easily. I think Paul Martin originally, in
his talk here, said something about something that I've told him,
that there have been thousands and thousands of elephants killed
-- not only in clearing of land for agricultural use, but in culling
operations. In 1984, '85 and '86 I saw 9,000 elephants culled,
and I cannot find a single site anymore of their bones -- within
10 to 15 years they're all gone. And I saw this happening within
three months of every year, over those three-year periods.
So saying that Clovis sites are rare, to me, is actually counterintuitive.
In fact, 15 sites with 50 mammoths dead in them to me is an extremely
rich, enormously rich, archeological record, of something going
on over a very brief period. If it took several hundred years,
and there's 75 mammoths, I can understand that being rich, because
the three years of culling that produced 9,000 dead elephants
had left no fossil, or future fossil, record. And the rest of
these ethnographic snapshots -- we cannot find any examples of
ethnographic people being observed hunting animals to extinction.
We can't find any examples of them doing anything other than
generalized foraging. So, therefore, according to this reasoning,
Clovis people must have been generalized foragers who did nothing
but generalized foraging, and did not bother hunting large animals.
And the rest of these -- the fact that there's variability in
fluted points has suggested to some archaeologists that all the
people making fluted points adopted a technology, but they all
different ancestors -- they were already here long before the
extinction event occurred.
Now, let me go to my Zimbabwe work, to my own fieldwork, because
that's the viewpoint I can tell you the best, or the most about.
Most of what I have produced seems to be used to support the
climate-change-only model of explaining extinction. Now, admittedly,
a lot of it actually does point that way very clearly. For example,
if you look at the mammoth sites in the fossil record, and you
look at modern elephant sites -- the ones that have been documented
during culling or other noncultural death events -- they look
very much alike. The same kinds of age profiles are represented
in the fossil and the modern elephants; the same kind of geomorphic
locations -- not at big rivers or big waterholes, or whatever
-- these are at headwater locations, rather small streams.
There's lots of different species represented -- it's not only
elephant or mammoth. There's large masses of bone; there's bone
that looks like it could be broken artifactually, or broken to
make something out of. And another point that oftentimes is not
raised when comparing the modern elephant population depletion
during the ivory-hunting craze of 100 years ago is that, in spite
of the fact that many elephant populations were driven to about
zero -- at least to a virtual zero point -- during the ivory-hunting
phase of late 19th century southern Africa, they've recovered
to some of the highest densities anywhere in Africa within a hundred
years. So elephants can recover from overhunting. How could
Clovis people, with spear points, have hunted an entire population
of mammoths in North America to extinction, if people with high-powered
rifles couldn't do it in the late 19th century?
Now, I want to explain some things about the sites, the modern
elephant sites, if you're not familiar with what does go on in
southern Africa. There are several kinds of sites that I observed
being created, by elephants dying either through cultural or noncultural
means, and I want to run through these as quickly as I can. Some
of the noncultural sites are serial sites, where single animals
would die for one reason or another -- old age, starvation, disease,
whatever -- and there's quite a few of these on record. Another
kind of site, also noncultural -- these are animals dying natural
deaths -- are mass death sites. These are jawbones from animals
that starved to death in 1995 around one water source here that
I've collected to determine ages. And these are rather limited
in number, but there are several of them that are known and have
been documented for generations.
So these are sites I've been going back to and describing, and
trying to understand what death processes are occurring, and trying
to explain what the sites look like. This is a site here that
contains bones of animals that had died. These are elephant bones
in earlier generations, or earlier years of drought, returning
to find water. They're not running out of water, they're running
out of food nearby to the water.
These sites have some characteristics which are similar, in some
ways, to other kinds of sites, but very unique in other ways,
and one of the unique things is that the age profile is dominated
by subadults -- very young animals. Sixty to 80% of the age profile
will be in a noncultural mass die-off, subadults. And another
thing that happens is, there's so many dead bodies of elephants
around that the scavengers don't generally utilize them very heavily
-- so there's very little chewing and scattering and breaking
of bones by hyenas, lions, jackals or whatever. And here's an
example of a skeleton three years after death. Even the stomach
contents are still preserved in it.
Another kind of site are the mass cultural sites -- the sites
created by humans who are shooting whole herds of elephants.
And I'm not talking about poaching sites here -- I'm talking about
legal culling, which has gone through a phase of popularity, and
now, currently, unpopularity, in southern Africa. These are attempts
to reduce populations of elephants in national parks, where their
habitat is perceived to be undergoing change, which is not thought
to be something that is to be encouraged. So during certain years
in Zimbabwe, large groups -- large numbers of elephants, as I
mentioned earlier -- were shot, were destroyed together as herd
groups, without selectivity. These are groups ranging from 15
to 73 animals, generally shot by marksmen, and then the rest of
the carcasses are salvaged -- the meat, the skin and so on . . .
These have their own unique characteristics, too, when compared
to the noncultural sites. There's a lot of butchering, complete
butchering. There's complete salvaging of the carcasses going
on day to day. But it's very rare -- and, in fact, it's been
impossible for me to find any sign of that butchering. No cut
marks, for example; no chop marks. So the lack of these signs
at sites that humans were responsible for is not necessarily an
argument in the fossil record against human involvement.
Age profiles tend to be all over the place with elephant herds,
just as in any small sampling of humans -- for example, you might
get a lot of young and a lot of old in certain groups. But there
is a tendency for this kind of catastrophic sort of age profile
to show up over and over again, with a few unusual ones, such
as this one here. These are four groups that were culled in four
different times and places. Lots of young -- typical of a healthy
population. And one group, for example, here doesn't have any
young, but there is a general trend here towards this kind of
decreasing ages, or decreasing numbers at each successively older
There's also a lot of artifacts and fire features around these
sites. They're created, after all, by humans, who spend all day
butchering. This is a fire pit that's going to be full of ashes.
They're usually covered over when abandoned. There's also the
odd cigarette butt, the button that falls off of someone's clothes,
the lost garment, the lost hat, newspapers and so on. There are
plenty of artifacts around these, but I'll come back to this in
a minute. The presence of artifacts -- the presence of fire features,
for example -- which is quite common in an archeological site,
does not necessarily mean the animals were killed by humans.
These are also present in noncultural sites.
And, finally, the last type of site that I've been describing
are these transported sites, where people have created the sites.
They've killed the animals, but they've taken the bones back
somewhere to do something with them.
Now, what I wanted to do was find a way of determining how you
tell these apart, the different types of sites: the mass sites,
the serial sites, the transported sites, the in situ sites, the
cultural and the noncultural sites. And if you can tell them
apart with elephants, years and years after the elephants have
been killed, you might be able to go to the mammoth bone record
and tell them apart the same way.
Some of the characteristics are very similar. You can run down
this list here. This is the noncultural die-offs of modern African
elephants, and here the characteristics of sites that were created
by humans shooting them all. But some things are different.
For example, the age profile here, is a bit different.
There must be some other things that were different, I had decided,
and we've run through these and tried to create a scoring system,
by assigning points to four different variables.
The variables are the way carnivores use the carcasses -- the
utilization of the carcasses by hyenas, lions and jackals, and
so on. One point would be very light utilization -- very little
chewing and gnawing and disturbance -- and three points would be
very heavy. And I scored three other variables the same way --
one, two or three points: weathering; bone representation --
whether some bones have been selectively removed or not -- and age
profiling. You've got a list here of the cultural sites characteristics,
and the noncultural mass site characteristics -- and there's some
slight difference in the total: four points for cultural, seven
for noncultural, which is semi-inspiring that there may be some
way of separating out these different kinds of sites.
And I did the same thing for serial sites; I did the same thing
for transported, as well. I'm not going to show you those slides.
But these totals come to some higher numbers, and there is slight separation between cultural and noncultural. And it did become possible -- now, I have to admit that this
is sort of a first approximation -- but it did become possible
to separate out cultural from noncultural, based on these kinds
of scores. Very low numbers were always cultural amongst the
African elephant sites that I knew how they were created, and very high numbers were also cultural, and everything in between
was noncultural. It also became possible to figure out whether
the animals died together, en masse, or one at a time, in series
-- with numbers that range, as you can see here in this slide.
Now, the thing to do, and which we did, was to go to the fossil
record and try to apply these kinds of analyses to actual mammoth
bone accumulations. Here's a selection of sites that have artifacts
with them, Clovis points, and a minimum number of individuals
of at least five animals -- we tried to have big collections, big
assemblages -- and then there's some sites that didn't have any
Now, I've called this "cultural" and "no artifacts,"
but, basically, I'm not trying to make the assumption that just having artifacts means it's cultural,
and not having artifacts means it's noncultural. Here's one example
of a noncultural site -- or some would argue it's cultural. This
is Lamb Spring, in Colorado, that had at least 35 Columbian mammoths
in it, and a few other species. There aren't any really good
clear associations with artifacts, but there are a few scattered
around that may be associated with it. But, basically, it has
no Clovis points, so it's questionable whether it is cultural
or not. But I'm presuming, at the outset, it probably isn't;
and I'm going to let the scoring tell me whether it is or not.
And another one -- the Hot Springs site, in South Dakota, which
has at least 50 -- I think 55 -- Columbian mammoths, and a few
woolly mammoths in it. There's no artifacts here. This is much
older. Lamb Spring, the last slide, was about 11,000 to 13,000 years
old; this is somewhere around 20,000 to 26,000 years old.
And the Waco, Texas, site -- this is not very well-reported yet -- but there's at least 22 mammoths in this, (and it's still growing
every year) including this one bull, and the rest of them look
like females and young, with absolutely no artifacts or features
associated with it. So that's at least three that don't seem
to be cultural at the outset.
The scores -- well, there are some surprises and some results that
just are not surprises. These three up in here are the ones without
the artifacts, and the numbers look like noncultural. It's a
little bit hard to tell whether they're mass deaths or serial
deaths. But the two down here at the bottom, Dent and Lehner,
are Clovis sites. Dent has maybe 14 or 15 mammoths in it, and
only two or three Clovis points. Lehner has 13 mammoths, and
I think 13 Clovis points. And the numbers here suggest that not
only are they cultural -- which you might expect, because they
have Clovis points -- but they're also mass deaths, which means
they died together. A herd of 15 mammoths, and a herd of 13 mammoths,
killed by people using stone spears, according to these numbers,
if these numbers work as well for mammoths as they do for modern
African elephants. So I'll put that up there one more time:
They're in situ mass deaths -- Clovis people killing mammoths
en masse, and they're definitely cultural.
I wanted to do the same thing with woolly mammoths now -- not
to shortchange Eurasian's mammoth extinction event. And some
of these sites I have studied myself: Krakow's Spadzista Street
in Poland, for example. There's plenty of literature about these
sites, such as Mezhirich and Mezen in the Ukraine. So I know some of them firsthand; I could
actually do the scoring system firsthand, and I can get enough
from the literature to do some of the rest. But I couldn't do
all of these. I did a small number -- here's Mezhirich, for example,
in the Ukraine -- based on the literature and my firsthand study.
And, again, there's some surprises in here, to me. These are
surprises to me, because, as I said at the outset of this, 18
years of research has produced results that suggest the fossil
record could be explained only by climate change, not by human
The dwellings in the Ukraine, which have hundreds of mammoths
represented by bones, are both cultural and serial. Cultural
you would expect, because they're full of artifacts, but "cultural"
here means they were killed originally by those people who made
their houses out of the bones. Serial is also to be expected,
based on the Russian literature -- or the Ukrainian literature
-- because the bones are weathered to different degrees, but this
is at least some sort of -- call it vindication, I guess -- of
people who have thought that the Ukrainian hunters of 15,000 to 25,000
years ago could kill hundreds of mammoths to make houses out of
their bones. Krakow's Spadzista Street has 73 mammoths represented
by at least 8,000 bones excavated in a 150-square-meter area -- a tiny little bone midden full of mammoths that the numbers suggest were all killed together by people in the Upper Paleolithic.
And Berelekh, this is a big surprise, because I don't think there's
another living human being who would say this is a cultural site;
it's nearby a site, but even the Russians don't believe it
is. And, according to my numbers here, these are 166 mammoths
-- that's the maximum estimated population of this site -- 166
mammoths died together.
I'll put this one up here again, because this is the two-bone
dwellings. These are transported serial and cultural sites.
There's no doubt in my mind anymore that people were hunting mammoths,
as there had been about 15 years ago. Now the question then
becomes: Why would anybody hunt mammoths? Going back to the
ethnographic snapshots and the foraging theory models of what's
going on here, why would people have done this? Or why would
they have hunted all of them, or killed all of them? And
what we tried to do here is look at the way foraging theory either
can be used to support, or not to support, the idea that Clovis
people actually hunted animals to extinction.
There's one model called the "diet breadth model" that
has probably been misapplied in trying to argue against people
being the agency of extinction, and I'm not going to spend much time
on this. Like all foraging-theory discussions, there
comes a point where you're sort of engaged in priestly jargon
rather than communication about foraging theory, and I know it
gets to be very dull. But I do want to say something about specialization,
because the argument is often made: If Clovis people are hunting
mammoths, they must be specialized -- and, in fact, nobody specializes
in hunting big-game animals such as mammoth. Why would they have
done this? What needs to be explained here is that diet breadth -- the number of items that are included in the diet -- has nothing
to do with people selecting an animal because it's special to
them. It has everything to do with making a rational decision
based on the returns you get from procuring that particular resource.
So on this graph, we've got diet breadth here. These are
the high-rated items, and these are the low-rated items -- and
you could see, handling time is very low for a high-rated item.
If you kill a very big animal, you get an awful lot of return
for the time it takes you to process and eat it. Handling time
goes way up here, and that's why these items down here are low-rated.
This would be things like grass seeds. You can find grass seeds
anywhere, so the search time, obviously, is going to be low for grass
seeds, but it could take you a long time to make grass seeds into
a nutritious food. Search time might be high for mammoths, but
the handling time is low enough that you're getting so many returns
it's worthwhile hunting mammoths. And it's been shown -- and
it's actually rational to expect -- that the larger the body size
of an animal, the higher its ranking is going to be intrinsically.
Because it doesn't matter how rare it is -- everybody knows that
you get a high return when you've killed that animal.
So what happens in high and low ranking of resources is, it doesn't
matter what your diet breadth is -- every time you encounter a
high-ranked resource, you take it. You don't turn it down. If
you're eating grass seeds and all the rest of this stuff out in
here -- the low-ranked items -- when you encounter a mammoth,
you will kill it, because you understand that it is high-ranked,
because of its return.
Another model is the patch-choice model. Many archaeologists
have argued that Clovis hunters would not have chosen the patches
where mammoths were all that often -- they would have hunted somewhere
else. The patch-choice model is a little bit more complicated
than what I'm going to tell you here, but I want to summarize this
as quickly as I can. The time you spend in a patch looking
for food and the returns are graphed here in three different ways. The time you spend here versus return in this patch type A is what
may be expectable in the ocean, for example. If you catch a fish
in the ocean, sooner or later another one's going to swim in, and
you can just stay there forever and keep on catching fish in the
ocean. This type here is also something you wouldn't expect to
find very often in a terrestrial ecosystem. The more time you
spend, the greater your return -- until you reach a point where
there's no more return. If you're going to, say, a nut patch and you eat all the nuts, you basically reach zero return. This
patch type down here is probably the most common kind of return
that foragers expect. You spend a lot of time in the patch, you
get more and more -- but eventually it levels off; it reaches
kind of asymptotic shape. And you don't really go to zero, but
you go to the point where the return rate is so low that you decide
to move to another patch.
Which brings up the question of: What were Pleistocene patches
like? I think this is probably the important thing, that many
archaeologists who argue against overkill don't really fully understand.
They weren't necessarily different in terms of return and time
-- they were different in terms of location. And location makes
all the difference for foragers, who were trying to get from one
patch to the other.
I'm going to suggest here that maybe these Pleistocene patches at
the time of Clovis occupation were actually shrinking little refuges
separated farther and farther apart by nonproductive areas. And,
in fact, they may have been crowded islands -- much like national
parks are today in Africa or anywhere else, with sterile areas
outside of them -- game sources versus game sinks, as was brought
up earlier. They may have been part of a complex mosaic environment,
which is what we, I think, understand, as archaeologists, the
Late Pleistocene must have been like. And that brings up another
series of questions of: What was a mosaic environment really
I've got four different suggestions here on how it may have been
on the ground a real mosaic: the overlap, the interlock, the
concentric and the successional. Overlap -- you can think of
this as species A, B, C, moving together so that they have
an overlapped area of high productivity. The others you get small
areas -- A, B, C, woods, grass, shrubs or whatever. It seems more
likely that this is the situation on the ground as it may have
We also know that mammoths were not all that unusual when compared
to living elephants, and they probably had very stereotyped diets.
And we know from isotopic ecology studies that maybe these patches
that were overlapping were rather large in size. And they weren't
a question of a few scattered trees here and there with a little
bit of grassland -- that they may have been rather large patches
of wood next to rather large patches of grass, and they had an
area of overlap.
And in these conditions, as had been seen in modern studies, it's
bad news for large mammals, in more ways than one. There's a
differential kind of reproductive success going on when there's
this kind of crowding, or creating little islands. There's delayed
maturation in mating -- and all the effects of all of these is
to make a large population size actually behave like it's much
smaller. There's so much differential reproductive success.
And there's even more bad news: It creates situations
where older females and very young animals will die first. The
older females would, in fact, be the repositories of all the knowledge
amongst the elephant herds, for example, or mammoth herds. And
trampling, of course -- and there's a number of other things that
would be bad news besides just trampling. This would reduce the
diversity of the soil biota -- so, from the bottom up, diversity's
Which would suggest something about how Clovis people may have
been timing themselves. They have been, actually, instead of
abruptly going and leaving one patch full of mammoths and going
to another, they may have been deciding to stay in them for a
long period of time. Now, this is also counterintuitive -- this
suggests that people are hunting animals which are rapidly disappearing
from the landscape at large, but not disappearing locally at these
refuge patches, and hunting them right to the bitter end.
The last slide here is also priestly jargon -- I don't think I'm
going to bother too much with this one. But this has to do with
marginal value theorem -- that the longer it takes you to get
from your patch to the next patch makes a difference in how long
you decide to spend in that patch. So people who have to walk
from one patch to the next a longer distance will stay longer
and create different archeological sites than people who walk
short distances from one patch to the next, and get their return,
and then go on to the next patch -- because they know the next
patch will be productive. This will make a big difference in
terms of the size and diversity of archeological remains in an
archeological site within the patch.
It won't take very long to find these patches. There's an easy
way to get from one place to an elephant die-off, and that's to look
for where the elephants are all going. I don't have any time
to develop this theme, but this is, in fact, not wasted time,
for a forager to follow the elephant trails. They are so easy
to see, and it's also quite clear, when you're following on a
trail of elephants which are going someplace under stress. And,
at the end of the trail, there may not be a big cliff, of course,
but there'll be some mammoths there -- and the more mammoths you
hunt, the higher your return is going to be, until all the mammoths
And I don't have a big finish. I do have a small finish. I would like to be
able to say this proves that Clovis people killed them, and killed them
all, and it wasn't climate. But, of course, we can't do that, because
we've been making this argument at each other for a hundred years. So
I'll just leave it with the little finish here.