Rates, Patterns, and Processes of Landscape Transformation and Extinction: Madagascar as an Experiment in Human Ecology
David A. Burney, Fordham University
Inferring cause and effect from the fossil record, as you might
have gathered, is not always a wholly satisfying enterprise.
The evidence is a bit stale, shall we say. Perhaps it's even
a bit crusty, as in this Archaeolemur skull. Many useful details
are missing, perhaps never to be found. Sequential events may
be collapsed together, inverted, or mixed with evidence from other
times. If there's any hope to ever straighten out this mess,
I believe our best bet is on remote islands, where the shallow
human time depth promises that the evidence will be fresher.
As the last place on earth -- or nearly last place -- to experience
a taxonomically diverse extinction event of the type we are discussing
today, and one of the last habitable land masses to attract human
colonizers, Madagascar can be thought of as perhaps the ultimate
natural experiment in human ecology.
But how shall we proceed in evaluating the results of this great
experiment? Well, honestly, I'm not sure how to proceed -- because,
in fact, this is a talk I've never given before; in fact, probably
nothing like it, and so we'll have to work it out together. First
of all, one of the things I had to face up to was that if I talked
about all the things that I and my colleagues have been trying
to do in Madagascar for the last 15 years, that, of course, mean
old Ross will drag me off kicking and screaming. So I decided,
instead, we would see what we could do in the way of narrowing
it down a little bit this afternoon to address this question of
how we might do this as a kind of semi-controlled experiment.
I think there are a couple of things that might be helpful in
that regard, a couple of areas that yield rich concepts. One
of these is the sort of emerging discipline of landscape ecology;
the other is to apply the criteria of experimental design as rigorously
as we can in a natural circumstance.
And so, first of all, to use the techniques of the landscape ecologist
a little bit, a group of people who are, I think, commendably
obsessed with scale and resolution, I think it's worthwhile for
us to ask ourselves: What are the things that we can actually
see in the fossil record? What can we measure, and what kinds
of evidence are useful for these measurements, and what sort of
scales do they exist on?
And so you can see here some of the things we've played around
with in recent years in Madagascar, for instance. Obviously,
the sort of conventional area for this kind of thing is looking
at climate change and human activities -- the subject of most of
the talks today -- utilizing things like excavations and sediment
cores. This is generally a thing which is effective within what
sometimes is referred to as "Libby time" -- that is,
the radiocarbon record time period -- on the order of tens of
thousands of years. Unfortunately, most of the time the resolution
for this type of thing is really not as good as we'd like for
it to be, but, under ideal circumstances, perhaps decadal resolution
can be obtained.
However, my interest, as a paleoecologist, is to try to look
at some of the other interesting things that we see in ecology
today, and see what goes on in the past in that regard, and so
I've made some effort, with some of my colleagues, to try to get
better resolution utilizing things like laminated sediments from
deep crater lakes, with, hopefully, annual resolution. And recently
we've been playing around with tree rings a little bit, to see
what we can get along those lines -- in order to get a better
handle on records of disturbance -- that is, fire-ecology events;
various kinds of disturbance events involving either the animals
or the climate, particularly looking at the issue of abrupt climate
Well, as the old song goes: That's nice work if you can get it
. . . But, in fact, obviously, those kinds of sites are fairly rare
in the tropics, or, perhaps, anywhere else. The other thing we
would like to do more, and I'm not going to say much more about that
today, but I think it's an important thing that several speakers
have addressed today, is to get the longer record, to help us
establish background levels of extinction and disturbance. And
in Madagascar what's worked well for us in that regard -- we're
looking at glacial-interglacial cycles and evolutionary aspects,
such as evidence for origination and extinction. Speleothems,
which can be dated back more than 10 times as far as the radiocarbon
record, utilizing uranium-dating series, have been helpful, as have,
of course, conventional excavation of Pleistocene sediments and
bones, and longer sediment cores.
Well, to address the other part of this, which is to try to make
this a bit more of an experiment and a little less of just an
observation. I want to lay out right now, at considerable peril
to myself, the various hypotheses that have been proposed to
explain the extinction of about two dozen mammals, birds and reptiles
in Madagascar over the last couple of thousand years -- all of
these, or most of these, larger than their living relatives.
There's a long history for some of these hypotheses.
For instance, Madagascar is the one place in the world that I
can think of where the issue of fire has been often brought forward
as a primary cause of extinction, as a possible way of rapidly
transforming the environment. And this is an old theory in Madagascar.
And what I've tried to do in this table here is to try to divide
up the expectations of these hypotheses in terms of the rate at
which the thing would occur over the landscape, and also the instantaneous
rate on a site-by-site basis. Also to look at the pattern --
that is, does this extinction and transformation move as a wave
across the landscape, or does it break out all over? Or is it
patchy in some way? And, of course, to address the process, to
the extent that we can, whereby the extinctions occurred.
So we have the old hypothesis about fire being essentially introduced by humans, spreading over the landscape, and really transforming it so rapidly that the trees couldn't grow back fast enough -- and the presumably arboreal animals then went extinct. Of course, there are many problems with that, if you've ever heard my talks before -- and I'll try to stay off of that subject today, because, frankly, I'm a little bored with it and tired of it. But some of these animals are definitely not arboreal creatures, and so that was one problem. The other problem was the fact that we have plenty of evidence showing that some environments in Madagascar experienced fire, to a considerable degree, before the arrival of humans.
A second hypothesis is one that I call the "great drought,"
for lack of a better name. It was proposed by Mahé and Sourdat
-- and unlike Humbert's hypothesis about fire, this was really
a hypothesis that wasn't concerned primarily with the interior
as a whole, but was for the southern region -- which was the only
area that is conceivably dry enough for a great drought to have
probably done the job.
The expectations, obviously, here for fire would be that it had
to happen pretty rapidly, and it seems that, from the literature,
the notion was that it happened more or less simultaneously throughout
the island, or very quickly. Whereas, with the drought, you can
imagine this being a slow-moving thing, as we've heard this morning,
in regards to various climatic scenarios -- and in the southern
region only, as I mentioned.
Of course, Paul Martin's blitzkrieg hypothesis -- and, in many
ways, Madagascar looks pretty good for blitzkrieg. He would
specify -- bless his heart, he's not afraid to stick his neck
out, and people like that make these meetings much more interesting
than people like me, who tend to stay in the middle of the road. He specified that in a place like Madagascar this would have
had to have been pretty rapid. If the same thing went on as went
on in North America, for instance -- and only took, say, 400 years
or a thousand years in North America -- imagine how quickly it might
have happened in a place like Madagascar, that's only about the
size of Texas.
Likewise, as you recall, the blitzkrieg hypothesis would predict,
then, that the pattern of this extinction would be as a kind of
wave. And, most likely, in this case, it would be a wave that
would start in the coastal areas and spread into the interior.
One would imagine that any place you could dock a boat would
be good for starting this sort of wave, and then the interior
-- the deep part of the interior -- and the highest elevations
would probably be the last places where this kind of event would
A fourth one -- in a sense, a new theory . . . although, I guess,
Cuvier and various people also put across some notions about disease
in regards to extinctions -- MacPhee and Marx. And I'm not going to
say too much about it now, because I'm sure you're going to hear
plenty about it this afternoon. And the expectations of that,
as I read them, are that this would probably be very rapid. If
you take, for instance, rinderpest in Africa as a model -- the
rinderpest was probably introduced in, I guess, in the 1880s in
Somalia, and within 10 years it had killed millions of hoof stock,
all the way down to the Cape. So one would imagine, if it's like
that, it's being a very rapid event. It would appear in most
fossil sites as being essentially simultaneous throughout the
island, I would think.
And then, finally, just, I guess, for the sake of discussion,
I also propose my own vague notion -- and I don't claim to know
what happened, frankly -- I'll tell you that upfront. But I do
think that what we're seeing in a lot of our sites is evidence
for several things going on at once. And what Melanie was saying
just now about that certainly strikes a resonant chord, and several
other speakers today have mentioned the possibility of interacting
factors -- that is, several of the other factors, perhaps all
the factors we've been talking about -- interacting at a crucial
time, in a crucial way, to accomplish these extinctions.
Now, I would -- you know, just to be fair and say what I predict
with this hypothesis -- I would predict that this would be a relatively
slow event, if, in fact, synergy was what was required, because
you have to think of it in a sense as the biota perhaps being
poised until the right combination appears. And I would also
predict that, in terms of a pattern, it would be a mosaic pattern:
not breaking out simultaneously everywhere; not spreading as
a wave, necessarily; but, probably, as conditions are met in each
individual region. We would then see what we call a "single
So we now have a methodological problem -- the problem being,
of course, that we now have to integrate various spatiotemporal
scales, and we have various kinds of data to deal with. Because
I tend to be interested in all of this, and because the evidence
from Madagascar is so fresh, compared to these other places, it
raises the opportunity of getting a lot of different kinds of
So I propose a couple of solutions we might test out a little
bit this afternoon -- and I'm not positive it'll work -- would
be, first of all, to try to visualize our data, from a subset
of our data, as a series of sites that are in a transect -- a
transect whereby we can control for certain variables and only
allow one or two things to really be changing at a time, and see
what that does. The other thing that I think, in the long run,
may be more helpful -- and, certainly, is equally helpful, is
what I call "integrated sites" -- which is the idea
of finding places where we can measure many different kinds of
things together at the same time, in the same stratigraphy, and
then compare these different kinds of evidence.
So, as I said, I'm tempted to talk about 24 different radiocarbon-dated
sites, because that's what we have for our paleoecological synthesis,
but I'm already eating up my time rapidly. So, instead, what
I'm doing this afternoon is, I've made a kind of transect for
you, here in the coarse stipple, where I'm just going to talk about
a group of sites along the west coast of Madagascar. And the
reason I choose this as my transect is, we can almost, in a sense,
throw out a lot of the variables in the equation here. It happens
that all of these sites are either directly on the coast, or just
a short distance up a navigable river, or they're at least less
than a day's walk into the interior. They're also all at very
low elevation -- just a little above sea level -- and they have
in common many features in terms of the type of sediments that
they offer us to study, and so forth and so on. So I'm going to
look at those this afternoon, and tip my hand a little and show
you one of the things we're going to find, and then I'm going to try
to show you that you should believe this.
First of all, if you look at our transect of sites here, what
we see is that these dates are our first evidence for humans
in each of these places -- and I'll explain in subsequent slides
what the evidence is; it varies from one place to another, to
some extent. But you'll see that it's about 2,000 years ago,
starting here at Andolonomby in the southwest. Oh, I neglected
to mention what the independent variable is in the physical factors
here. It happens that this transect goes from the driest part
of Madagascar to one of the wettest. So, in addition to any sort
of effect from humans, the other thing we're looking at here is:
Does precipitation and the resulting vegetation play some role
in this story? So we have 400 millimeters of rain here, and 2,400
up here in this rainforest. So we start with a semidesert and
we end in a rainforest.
Two thousand years ago, we have evidence for people here. We have similar
ages for these other dry sites. Here are Belo-sur-mer at 600 millimeters,
Lake Komango at 900 -- and, lo and behold, just up here, in places that
are just as accessible by boat, if not more so, in somewhat wetter climates,
we have much later evidence in sites that have received at least as much
study of the type we're doing here about 890 and 1,130, and finally 710 years ago,
up here inside the rainforest.
When people arrived in Madagascar, as any place that we're talking about
today, is very critical to evaluating these hypotheses. So, over the years,
I've developed four ways of doing that, and none of these are definitive
by themselves. It's good to have more than one of these that agree --
that's reassuring. It's also good to have paleontological and/or
archeological evidence that supports these conclusions. But it helps,
because it allows us to do, as the landscape ecologists would tell us
we have to do -- get a fine grid of sites over the landscape.
So these ways that we might find these people, then, are through
the appearance in our cores of exotic pollen types -- that is,
plants that they might have brought, evidence for these plants.
The best one we have so far for the Malagasy, as opposed to
the Europeans -- which, of course, brought a number of other plants,
which nicely mark post-European contact sediments. For the Malagasy,
though, it happens there's one very well-dispersed pollen type
that's known to have been introduced to Africa -- and, presumably,
Madagascar -- from Asia, which is easily identified and is a plant
that spreads like a weed on its own. And we also know, from
ethnobotanical research, is spread by maritime peoples. And whenever
I tell about this, I always get a few snickers around the room.
This is the pollen of Cannabis. It happens that Cannabis is
a very heavy pollen producer and is brought by sailors to places.
It's useful for making sails and ropes, and caulking their boat,
and possibly other things as well. You thought these were palm
trees growing down here, didn't you?
Anyway, it happens that, throughout the island, wherever we look,
we find Cannabis pollen at 2,000 BP. And this is not only in
places where there's evidence for people doing something locally,
but also well ahead of that. As if this plant, once it was introduced,
was sort of like Holdaway's rat -- it moved off on its own and
really is a kind of bow wave ahead of the human diaspora.
Another method that we've used is to look at an increase in ruderal
pollen types. These can be native plants, but these are plants
adapted to disturbance. Of course, this is a fairly conventional
way of doing this. The third way is to look at a sudden increase
in microscopic charcoal particles. That is, not the kind of charcoal
you dig up in big chunks out of archeological sites, but the kind
that's deposited as a kind of palynomorph -- that is, something
in the size range of pollen grains -- in sediment cores, and it
can be analyzed along with the pollen. I'm going to be talking about
that a lot, so I'll leave that one for the moment.
And then, finally, the use of paleolimnological evidence for
cultural eutrophication -- and, by that, speaking of a number
of different methods utilizing diatoms, algae, skeletons, various
kind of biochemical techniques, to infer that people are pumping
up the nutrient load in the waters. So these all serve as kind
of proxy indicators of human arrival in various places in Madagascar,
and it's worked for us in some other places, as well, such as
So the first site I want to talk about, then, is Andolonomby,
which is the dry lake bed at 400 millimeters of rainfall down
in the semidesert, and this is a bad old faded picture from over
10 years ago, with Ross MacPhee right there, back in, perhaps,
younger and more carefree days. I don't know about the carefree
part. But he was at Duke University, as a professor, and I was
doing a postdoc with him, and we refound the old fossil site of
Ambolisatra, which had yielded vast quantities of bones of many
of these extinct giant lemurs, and hippos, and elephant birds,
and big tortoises and so forth -- and so we went back and reopened
We didn't find a wonderful lot of stuff; we found plenty for dating.
But it happens that a lot of the best stuff that Ross got out
of this, he actually found in a dusty drawer at the Museum d'Histoire
Naturelle in Paris, which had been dug up by Grandidier and
his friends back around the turn of the century, some of them
from this site. And these are some of these wonderful modified,
human-modified hippo bones of the extinct pygmy hippo of Madagascar.
You can see these distinctive cut marks here -- and as best Ross
and I can tell, you can't really dig up a bone that's several
hundred years old, and cut it, and get these kinds of nice smooth
cuts. Maybe that's debatable. But, at any rate, in dating these,
we came up with an initial horizon for humans on this basis --
which, by the way, is the oldest direct archeological evidence
for humans in Madagascar -- of about 2,000 BP. And these bones
occur in at least three sites on the coast of Madagascar -- mostly
hippo bones. There's one Aepyornis bone -- 1,860 BP -- from one
of these sites.
What I got out of this was, it turns out that these were also
pretty good sediments for pollen analysis, and so I was able to
do a pollen study of the last 5,000 years from the same site
that contains these bones. And what I see there is evidence of
a climate change. I know you can't read all this back there,
but this is part of -- you know, I don't see very well, and this
is my revenge on the rest of the world. But I can tell you what
This is pollen of various kinds of dry-adapted trees and shrubs.
At 3,000 years ago, this disappears fairly quickly. It was replaced
by a palm savanna, with just palms and a few leguminous trees,
and quite a bit of grass. That lasted about a thousand years.
Other evidence, as well, has suggested that's a climatic change.
We know from many sites now that in Madagascar it seems that
about 3,000 years ago there was a drying of the climate. And
then, at about 2,000 we see a sudden great increase in grasses.
We could take that as a ruderal, perhaps -- there's some other ruderals
over here that go up at that time. At about 1,890 was the exact
date for that.
So here's the charcoal diagram from the same site -- and I mention
this as a kind of rapid bioassay. It happens now, after doing
this in many, many sites in Madagascar, that we've concluded that,
below about a 1,000 millimeters of rainfall, there's seldom any
charcoal -- or not very much charcoal -- in sediments before the
presumed arrival of humans. In sites with between about 1,000
and 2,000 millimeters of rainfall roughly, there's a lot of prehuman
fire. And just as a sort of caveat, for comparison -- here's
a site from that kind of regime in the interior -- highly seasonal
regime -- and there's actually more charcoal in the sediments
before the arrival of humans. Then it drops down during a phase
here in the Late Holocene, and then seems to pick up at about
1,200 BP at that site deep in the interior. And archeological
evidence suggests that's when people arrived there. But, at any
rate, this is from that site, and we see there's almost no charcoal
until 1,890 -- and then, suddenly, there's a great amount of charcoal
in the site.
Now, farther up the transect, in a place that is, indeed, between
1,000 and 2,000 millimeters of rainfall, you see that, in fact,
there was a lot of charcoal here at Lake Mitsinjo before the
arrival of humans. But it's almost all woody charcoal -- very
little grass-derived charcoal. And then we see evidence at about
1,000 BP, at that site -- actually, at 890 -- for an increase,
a great increase, in charcoal derived from both grass and wood.
So this set me to thinking. Here we had these various kinds of evidence,
all from one site -- and what really got me thinking about this, again,
recently, is that, in the interior, not far from here, dates on Megaladapis and Paleopropithecus -- two of the extinct giant lemurs -- two of the
biggest ones, were only 500 and 600 BP. And so I started thinking to
myself: Hmm . . . now this is pretty interesting. We have all these different
kinds of evidence, and they all seem to be, at this one site, converging
on the idea that people arrived there 2,000 years ago, but they didn't
really seem to finish the job, in terms of the extinctions, for perhaps
a thousand or 1,500 years. And it took them some time to really transform
the landscape completely. This type of site -- I've just got a
Venn diagram here to sort of advertise the idea of integrated sites --
what I mean by an integrated site is a site that has archeological, paleontological, and paleoecological information
all in the same site. This requires a rather special
kind of site in terms of sedimentation and biochemistry, and we all need
a lot more of these, I'm convinced.
So after feeling that I had some success with this, I decided
that I would look for another place that ought to be an integrated
site. So we went to Belo-sur-mer -- another place that's yielded
these modified hippo bones, up the coast -- also accessible to
the coast in the same way. We've been working there for several
years now. Some of the people who've been helping me -- Ramilisonira, who is the premier Malagasy archeologist, who's been working
with me for about 15 years; Bill Jungers, a primate anatomist;
here's Ann Yoder, working on ancient DNA of the giant lemurs.
Not pictured here is my wife, Lida Pigott Burney; and George
Brook, a geomorphologist who's worked with us a lot; Helen James,
an avian paleontologist. We've had a very good group of people
to work on these problems.
And so we get in these sites -- and in this case we moved over
10 cubic meters of sediment here in the last few years, and this
is what it basically looks like. A little over a meter of sediment
-- it's this kind of silty clay, and much darker here at the top
-- and I'll explain why in just a moment. And this is just chockablock
full of bones of these extinct animals.
Well, that in itself is interesting. But what's really got me
thinking now, about how long it took for people to do the job
is: If these dates -- which are acid alkali treated collagen
-- on extinct animals are good, and we're going to sure find out
and maybe Tom Stafford will be able to help me with this --
we see that what we really have is a long list of these things
in the human period. And some of them, such as Archaeolemur, spanning
nearly 1,000 years at this site. Some, like Pachylemur, surviving
till very late -- at a site that, presumably, humans arrived at
quite early, relatively early. So I'm just going to skip over that -- but that's the charcoal from that site, demonstrating the same sort of thing.
Lake Komango -- which is up, a slightly wetter place, 900 millimeters
-- and here we have charcoal. I'm going to have to start racing
ahead now; this always happens. But here we have this charcoal
peak, which is actually a large peak, at 1,750, suggesting, again,
early arrival here in the southwest.
Let's jump now to the rainforest in the north. This is Lake Amparihibe on the
offshore island of Nosy Be. We're getting our raft ready for coring. I'm
not paying much attention here -- I'm presenting my hindquarters to the
four-meter crocodiles -- but it came out okay.
On this lake we have the most wonderful sediment, because this
is a 50-meter-deep crater lake, and these laminated sediments,
which look good enough to eat, are really the right kind of place
to resolve these kinds of questions we're talking about. And
here you see a 2,000-year record, and this is the charcoal -- woody
charcoal and grass charcoal. And this represents many, many
hours of painstaking work by Lida Burney. And what we see is
essentially no charcoal. Now, I neglected to mention before that,
above 2,000 millimeters of rainfall, our record seemed to suggest
that, again, we don't have natural pyrogenic communities as you'd
expect -- it's too wet. But in this period between 2,000 and
about 1,000, we see no evidence for humans at this very accessible
coastal site. And then, at about 1,200 BP -- this is an 1,130
accelerator date on a plant macrofossil -- we see huge amounts
So let me just push the envelope a little farther and go into
the interior, about a half a day's walk into the rainforest, here
at Benavony, and what we see at Benavony is essentially no charcoal
in the sediments, until 710 BP -- and then this huge spike. This
corresponds to a time when a large walled town was built along
the coast there, and it appears that there was a lot of environmental
degradation in that area. And it also appears that, within a
couple hundred years, that dense population of humans disappeared.
This civilization seems to have collapsed about the right time
for the introduction of the plague, which was, of course, plaguing
much of the rest of the world at that time.
Well, Ross is going to drag me off here in a moment, so I'll just
have to wrap up and say that I don't claim to have all the answers
to this yet, but I think what we can say at this point is that
maybe a better question to ask than, What happened to these extinct
animals?, is to ask whether what is happening today is really very
different from what happened after first contact. We have, really,
four hypotheses that people have proposed,
that suggest that what happened in the past in Madagascar is different
than what we see happening there now.
The fifth hypothesis, my synergy notion, allows that the processes
invoked in the other hypotheses -- like fire and climate change, overhunting, biological invasions -- were probably ingredients
of the mix then as now, with the presence of humans as the key
A resounding refutation of any of these hypotheses will actually be difficult,
I admit, but it may be eventually possible, with enough radiometric dates
from the right places and times, and a more complete paleoecological
reconstruction at the landscape level, to narrow down this range of hypotheses.
It's possible that there are no distinct temporal or causal breaks between
earlier and later extinctions of the human period in Madagascar, and current
species endangerment there. If this were the case, it would be hard to
dispute the importance of environmental synergy, because we see it everywhere
we look in Madagascar today.