Bigger kill than chill: The uneven roles of humans and climate on late Quaternary megafaunal extinctions

Abstract

Starting around 50,000 years ago, most large terrestrial animals went extinct in most continents. These extinctions have been attributed either to climatic changes, impacts of human dispersal across the world or a synergy among both. Most studies regarding these extinctions, however, have focused on particular continents or used low-resolution analyses. We used recent advances in fossil dating and past climatic models in a high-resolution quantitative analysis, comparing the explanatory power of the hypotheses at global scale. The timing of human arrival to each region was the best explanation for the extinctions. Climatic effects, where present, were additive rather than synergistic with human arrival. While climatic variation was a contributory cause that helped explaining the process, anthropogenic impacts were the necessary cause that drove it.

Full text

Bigger kill than chill: The uneven roles of humans and climate on late
Quaternary megafaunal extinctions
Bernardo B.A. Araujo
a
,
*
, Luiz Gustavo R. Oliveira-Santos
a
,
c
, Matheus S. Lima-Ribeiro
b
,
1
,
Jos
e Alexandre F. Diniz-Filho
b
, Fernando A.S. Fernandez
a
a
Laborat
orio de Ecologia e Conservaç~
ao de Populaç~
oes, Departamento de Ecologia, Universidade Federal do Rio de Janeiro, C.P. 68020, Rio de Janeiro, RJ,
21941-902, Brazil
b
Laborat
orio de Ecologia Te
orica e Síntese, Universidade Federal de Goi
as, C.P. 131, Goi^
ania, GO, 74001-970, Brazil
c
Laborat
orio de Ecologia, Departamento de Ecologia, Centro de Ci^
encias Biol
ogicas, Universidade Federal de Mato Grosso do Sul, Campo Grande, MS, Brazil
Keywords:
Megafauna
Pleistocene
Holocene
Quaternary
Extinction
Human impacts
abstract
Starting around 50,000 years ago, most large terrestrial animals went extinct in most continents. These
extinctions have been attributed either to climatic changes, impacts of human dispersal across the world
or a synergy among both. Most studies regarding these extinctions, however, have focused on particular
continents or used low-resolution analyses. We used recent advances in fossil dating and past climatic
models in a high-resolution quantitative analysis, comparing the explanatory power of the hypotheses at
global scale. The timing of human arrival to each region was the best explanation for the extinctions.
Climatic effects, where present, were additive rather than synergistic with human arrival. While climatic
variation was a contributory cause that helped explaining the process, anthropogenic impacts were the
necessary cause that drove it.
©2015 Elsevier Ltd and INQUA. All rights reserved.
1. Introduction
Since the 19th century, when science became aware of the
sudden and geologically recent disappearance of many large-
bodied animals, the late Quaternary Extinctions (LQE) have
remained a great and controversial matter (Grayson, 2008). Start-
ing around 50,000 years ago, about two thirds of all large terrestrial
animal genera went extinct in a sequence that affected most con-
tinents (Koch and Barnosky, 2006). For a long time, two main hy-
potheses eattributing these extinctions either to climatic changes
during the last glacial event or to the impacts of modern man's
dispersal across the world ehave divided the academic commu-
nity. Many researchers also came to defend a synergy between both
factors as a more plausible scenario for the extinctions (Barnosky,
2004; Nogu
es-Bravo et al., 2008; Lorenzen et al., 2011; Prescott
et al., 2012; Lima-Ribeiro and Diniz-Filho, 2013), although contro-
versies about the balance of climate and humans as extinction
drivers still remain (Lima-Ribeiro et al., 2012; Prescott et al., 2012).
The late Quaternary megafaunal extinctions were a global
phenomenon and we believe that a global approach is the best way
to understand causal mechanisms. This would bring the full range
of temporal and geographical variation in extinction times to bear,
allowing one to disentangle the overall signal from regional trends.
Most studies, however, have focused on particular continents and
taxa (e.g. Alroy, 2001; Diniz-Filho, 2004; Johnson, 2006; Nogu
es-
Bravo et al., 2008). A few global analyses have been presented
(Lyons et al., 2004; Gillespie, 2008; Prescott et al., 2012; Sandom
et al., 2014); but innovative and insightful as these studies have
been, they carry some problems. While some works lack quanti-
tative analyses of the proposed extinction causes (e.g. Lyons et al.,
2004; Gillespie, 2008), others are based on crude and often unre-
alistic scenarios of human arrival and megafaunal extinction
(Prescott et al., 2012; see Lima-Ribeiro et al., 2012 for details). Both
Prescott et al. (2012) and Sandom et al. (2014) include non-
quantitative variables in their models, as their hominin paleoge-
ography variable is based on discrete human arrival scenarios. The
most recent global analysis (Sandom et al., 2014) is based on global
databases on extinct (and extant) mammals' distributions that are
bound to be incomplete and/or to contain a proportion of un-
trustworthy data (as shown by the inclusion on the analysis
of Africa and Southern Asia, regions with poor paleontological
*Corresponding author.
E-mail address: araujo.bernardo@yahoo.com.br (B.B.A. Araujo).
1
Laborat
orio de Macroecologia, Universidade Federal de Goi
as, Campus Jataí,
75804-020, Jataí, GO, Brazil.
Contents lists available at ScienceDirect
Quaternary International
journal homepage: www.elsevier.com/locate/quaint
http://dx.doi.org/10.1016/j.quaint.2015.10.045
1040-6182/©2015 Elsevier Ltd and INQUA. All rights reserved.
Quaternary International 431 (2017) 216e222

records). Additionally, their approach lacks a comparison of
extinction dates with human arrival and climatic change focused on
chronology (rather than geography).
Fossil dating allows the establishment of synchrony between
extinction events and their potential drivers. In the last years, a
growing number of dates have been published and reviewed
around the world (see Supplementary References). Improved cli-
matic models have been developed for the last 122,500 years
(Andersen et al., 2004). These advances made a once unfeasible
chronological global analysis of climatic changes, human arrival to
each region and extinction of megafaunal taxa a concrete possi-
bility, opening a promising path for resolving the extinction debate.
In light of these new chronometric advances, we performed an
exhaustive gathering of data for human first appearance dates
(HFADs) and last appearance of megafaunal genera (MLADs) on
nineteen regions across the globe, together with climatic variation
through the late Quaternary, to provide the first high-resolution
chronological analysis of the LQE extinctions. We tested the hy-
potheses that human arrival or climate variance would be
responsible for the extinction of megafaunal genera. This more
detailed approach should advance the extinction debate, providing
the first quantitative chronological test of the roles of anthropo-
genic impacts and climatic variation on the demise of the world's
megafauna.
2. Materials and methods
2.1. Data
The predictions of both hypotheses were compared in order to
evaluate them. The environmental hypothesis predicts that ex-
tinctions would have occurred during or following intense climatic
changes through the late Quaternary. The human impact hypoth-
esis, on the other hand, predicts that extinctions would have fol-
lowed human colonization of each landmass across the planet.
First, last appearance dates of megafauna (MLADs) species and
first appearance dates of anatomically modern humans (HFADs)
on several landmasses were gathered from all published scientific
sources that could be assessed (see Supplementary Tables 1 and
2). These landmasses included South America, North America,
Caribbean islands, Northern and Western Eurasia, Australia, Tas-
mania, Madagascar, New Zealand and Japan. Climate variation in
the North hemisphere through the last millennia of the Quater-
nary was assessed by the North Greenland Ice Core Project
(NGRIP) data on the variation of oxygen isotopic composition in
ice cores (Andersen et al., 2004). This database comprises
d
18
O
data from the last 122,500 years, with
18
O values for every 50
years. For the South hemisphere we used the European Project for
Ice Coring in Antarctica (EPICA) database, which comprises data
on the variation of deuterium concentrations (
d
2
H) at irregular
but frequent intervals along the last 800,0 00 years. We used EPICA
data for the last 122,500 years only, to cover an interval similar to
the one provided by NGRIP. Both
d
18
Oand
d
2
H are proxies for
temperature conditions for their respective hemispheres. Their
use in our analysis assumes that although changes along the
glacial cycle differed among regions, times of intense global
temperature variation within each hemisphere would be reflected
as regional changes of increased magnitude (Walker, 2005). We
opted for this approach, instead of assuming any finer regionali-
zation, because actual global reconstructions of past climatic
conditions are few and punctual across time, and do not neces-
sarily reflect periods when megafaunal extinctions took place.
Environmental proxies with high spatial resolution, including
phytophysiognomical reconstructions based on pollen data, are
available for just a few regions across the world, which precludes
their use in global models (Gill et al., 2009, 2013; Rule et al., 2012).
Considering such limitations, we believe that high-resolution
chronological data for each hemisphere can be more informative
than a crude and possibly misleading interpolation of past cli-
matic scenarios in a geological period when climate undergone
many rapid changes.
To allow comparisons between the hypotheses' predictions,
data reliability was assessed through a scoring system. Paleonto-
logical and archaeological dates are sensitive to methodological
errors (Walker, 2005). Sample contamination, poor materials,
stratigraphic misinterpretations, inadequate dating methods and
other problems can seriously jeopardize a date's accuracy. To
identify reliable data, many authors have used different quanti-
tative scales based mainly on sample material, stratigraphic as-
sociations and the type of equipment and logistics used in a given
study (Mead an d Meltzer, 1984;Burney et al., 2004; Barnosky and
Lindsey, 2010; Iwase et al., 2012). Dates from articles and books
that passed through such scrutiny were collected without further
appraisal. In most cases, however, dates lacked any sort of accu-
racy determination, making data filtering a necessity. For radio-
carbon based dates, this filtering was achieved using the Mead-
Meltzer Scale (Mead and Meltzer, 1984)modified by Barnosky
and Lindsey (2010), applying strict criteria: for paleontological
and archaeological dates to be accepted, they had to reach at least
ranks 11 (out of a maximum rank of 12) and 13 (out of a maximum
rank of 17) respectively (following Barnosky and Lindsey, 2010).
Still, most datings performed in Oceania over the extinctions
period are based on different methods, mainly U/Th (Uraniume-
Thorium dating), OSL (Optically Stimulated Luminescence dating)
and ESR (Electron Spin Resonance dating). As there are no scoring
systems capable of evaluating the accuracy of dates obtained by
these methods, ranked scales along the lines of the Mead-Meltzer
Scale were designed to assess the reliability of U/Th and OSL dates
(Supplementary Table 3). The new scales do not include ranks
associated with archaeological remains, because human dates
were always based on radiocarbon methods. ESR dating involves a
more complex set of techniques, making its dates harder to fitinto
a simple scoring system. So, only sources that utilized CSUS-ESR
(Closed System U-Series ESR), a more accurate variant of the
ESR method, were considered in the following analyses (Grün
et al., 2008, 2010).
After the data filtering, date calibration was performed. Radio-
carbon datings are based on the
14
C/
12
C ratio of tested samples; as
base concentrations of both isotopes fluctuate through time in the
atmosphere, calibration is necessary to transform ‘radiocarbon
years’on actual ‘years before present’. Dates were calibrated using
the software Calib 6.0, using the IntCal09 curve for every sample.
Even though this calibration curve was originally designed for the
northern hemisphere, it is the only one that encompasses the
whole span of the extinction event.
As a last precaution, we tested bootstrapping corrections over
the paleontological dates of South America (using the Cueva del
Milodon, in Argentina, as the well sampled site) to avoid possible
biases caused by the Signor-Lipps effect, following the methodol-
ogy established by Barnosky and Lindsey (2010). This method has
been criticized by Johnson et al. (2013) for not accounting
adequately for the uncertainties and biases that affect the estima-
tion of MLAD and HFAD. Regarding the nature of the expected bias,
using uncorrected MLAD and HFAD would underestimate the
coexistence between humans and megafauna. Anyway, the use of
corrected data did not significantly affect the results, thus we opted
for using uncorrected data to perform all analyses described in the
following section, keeping in mind that this could make our ana-
lyses conservative against finding an association between human
arrival and megafaunal extinction.
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222 217

2.2. Statistical analysis
All analyses were based on genera rather than species to avoid
taxonomical noise, because fossils are not always identified to the
specific level or such identifications are often controversial. As both
times of megafauna extinctions and human colonization vary
considerably within large landmasses, the world was divided into
19 regions. These divisions were based mainly on great geograph-
ical barriers (e.g. Andes, Ural Mountains etc.) and temporal gaps on
human arrival (e.g. Mediterranean vs. northern Europe, islands vs.
adjacent landmasses) (Fig. 1). The time of human arrival was set by
the HFADs in each region, whereas the time of extinctions was set
by the MLADs of each megafaunal genus. The climate changes were
estimated based on
d
18
O and
d
2
H, as proxies for temperature
variation for the North and the South hemisphere respectively, as
described above. Finally, the relative importance of human arrival
and climate changes on LQE was assessed in two ways.
First, we created two sets of null models, one of climate changes
and another of human arrival, to investigate the chronological as-
sociation of periods of intense climatic change and of HFADs with
the MLADs. For the each hypothesis, a thousand dates were
randomly drawn from the 122,500 years of climatic data available
for each extinct genus on each region. In respect to the climatic
hypothesis, each true last appearance date had the
18
Oor
2
H vari-
ance calculated for the time interval comprising its dating error
plus another thousand years into the past. This same time span
(true dating error plus a thousand years) was used to calculate the
18
Oor
2
H variance of all generated random dates. This step assumes
that the effects of climatic changes on regional megafaunal ex-
tinctions would be apparent within a 1000 years interval; this we
regard as a conservative approach to accommodate a delayed
response by the extinct genera. Anyhow, we repeated this proce-
dure using 3000 years intervals and the results remained almost
unaltered. We then estimated the level of significance of climatic
Fig. 1. (A) The diaspora of modern man through the planet, showing the 19 regions in which the world wasdivided for this study. Arrows indicate the approximate direction of each
major colonization event. (B) The extinction of megafaunal genera at different parts of the world, according to calibrated reliable dates. Points are slightly jittered to minimize
overlap. The color scales indicate the timing of human arrival and extinctions in each place, from the oldest (cold colors) to the most recent (hot colors).
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222218

effects on each megafaunal genus in each region by considering the
proportion of random variances equal to or greater than that based
on true MLADs. This metric of climatic instability was chosen as the
environmental variable rather than extreme values because there is
only a single apex to the last glaciation(or two, if the Younger Dryas
is considered), thus climatic variance would be a better candidate
than climatic extremes as a predictor for the extinctions.
A similar method was employed to assess the second hypoth-
esis, only this time the chronological distance between each date
(random or true) and the corresponding regional HFAD was
measured. We then estimated the level of significance of anthro-
pogenic effect on each megafaunal genus in each region by
considering the proportion of random chronological distances
equal to or lesser than that based on true dates. On every occasion
where the climatic or anthropogenic effects were statistically sig-
nificant ei.e. when extinctions were more closely related to cli-
matic changes or to human arrival than expected by chance ethat
specific cause was considered responsible for the extinction of a
given megafauna genus. When both hypotheses showed significant
p-values, the result was considered “entangled”, meaning that the
models could not discern between causes for that particular
extinction event.
In a second approach, we used a generalized linear mixed model
(GLMM) to test, in a single model, the effects of both climate
changes and human arrival (fixed effects) on the extinction rates of
megafauna genera, as well as the effect of the interaction between
these hypothesized explanatory variables. As somegenera survived
several millennia after most of their concurrent taxa, to avoid ef-
fects of time-lags between the change of state of a given variable
(human arrival or climatic instability) and total megafaunal demise
within a region, 25% of the most recent extinction dates from every
region were removed from the generalized linear mixed models.
Within each region, the last 60 thousand years of data were divided
into bins of 41 different sizes, ranging from 1000 to 5000 years with
100 year increments. The number of extinct genera recorded in
each bin was obtained; therefore bins were our sample units. The
use of bins of different sizes allowed us to evaluate the robustness
of the results to variations in size, which is an arbitrary choice. The
time since human arrived in each region (the first explanatory
variable) was measured as the amount of years from the beginning
of each bin to the true HFAD on that region. The climate change (the
second explanatory variable) was quantified as
18
Oor
2
H variance
within each bin.
The GLMM models were fitted using a Poisson distribution
because the number of extinct genera in each bin is a typical
discrete count variable. The regions' identities were included in the
models as random effects (allowing random intercept estimation)
and temporal autocorrelation was controlled between successive
bins within each region using a first-order autoregressive correla-
tion structure. The models were validated through the checking of
both normality and absence of temporal autocorrelation of the
residuals. The GLMM and null models were run in R software (R
Core Team, 2012) by using the function glmmPQL from the pack-
age MASS (Venables and Ripley, 2002).
3. Results
A total of 2088 dates for 67 genera of extinct megafauna (58
mammals, 8 birds and 1 reptile) were considered reliable, totalizing
126 independent sampling units across 19 regions (Supplementary
Table 1). Similarly, 762 human dates fulfilled the reliability re-
quirements, and were used in the analyses (Supplementary
Table 2).
Null models showed that most extinctions (85/126, or 67.4%;
Fig. 2 and Supplementary Table 4) took place around the time of
human arrival in each region, as it can be seen by the close temporal
gap between high-resolution MLADs and HFADs across the globe.
This pattern emerged despite the use of uncorrected dates, which
would underestimate the coexistence between humans and
megafauna. A similar correlation of extinction dates with times of
intense climatic variation did not occur: only 17.5% (22/126) of the
MLADs happened in periods of intense
18
Oor
2
Hfluctuation.
Among these, only 1.6% (2/126) genera disappeared in periods
linked only to great climatic variance, and not to human arrival. On
the other hand, 65 (51.6%) were closer than expected by chance
only to human arrival. Twenty cases (15.9%) were associated to both
events (entangled). In the remaining 39 cases no association could
be found.
Forty one GLMM models were generated encompassing the
range of time bins between 1000 and 5000 years. If either tem-
perature proxy were used to represent climatic variation for the
whole planet, Antarctic
2
H detected a stronger climatic effect than
Greenlandic
18
O, but human arrival had the strongest effect in both
cases (see Supplementary Table 5). Adopting our approach of using
each proxy to represent climatic variation in its own hemisphere,
the time of most intense climatic variation was significantly related
to the extinction of megafauna in only 1 of the 41 bins. In sharp
contrast, the date of human arrival was the best predictor of the
extinction of megafauna, with a significant effect in 40 of the 41
bins. These models had an average R
2
of 0.525. In the models
considering the interaction between both factors, the interaction
term was significant in only 4 of the 41 bins (Supplementary
Table 5). Thus GLMM provided stronger evidence for anthropo-
genic than for climatic effects, and little evidence of a synergistic
action of the two factors.
4. Discussion
Overall, our results indicate that human arrival was a necessary
factor for the extinctions, whereas climate variation was a
contributory one, enhancing regionally the effects of anthropogenic
impacts in additive rather than synergistic ways. This conclusion
builds upon the previous findings of the previous global analyses by
Prescott et al. (2012) and Sandom et al. (2014), but it rests on a finer
data base and it clarifies the causal relation between the two fac-
tors. The fact that Africa escaped the strong global extinction
pattern reinforces the interpretation of human-driven causes
because there the necessary cause was missing (Klein, 1984): one
would expect stronger anthropogenic impacts in first contacts with
faunas without any previous evolutionary contact with Homo sa-
piens (Diamond, 1984). Besides, the interpretation that climate
variation was not a necessary cause for the massive LQE is consis-
tent with the pattern that the Pleistocene accommodated over 30
glacial cycles, many as intense as the Last Glacial Maximum
(Barnosky, 2004; Walker, 2005), all of them with few or no asso-
ciated extinctions (Cione et al., 2003; Barnosky, 2004). It has been
argued that the last glacial cycle would have had a greater variance
of temperature than earlier glacial periods eat least in maximum,
rather than the minimum, temperature ein Sahul (Wroe et al.,
2013), but globally climatic variance was not a strong extinction
predictor in our results.
In relation to previous global analyses, our findings clarify
regional differences, highlighting the limitations of single-
continent studies to make inferences about the causes of the
global process of LQE. For example, while the anthropogenic signal
was straightforward for Australia and the Americas, strong regional
patterns can be observed when northern Eurasia (except Bering) is
viewed apart from the world. In the sampling units within West
Siberia, Central Russia, Japan and Europe, only two of 28 (7%) ex-
tinctions were associated to human arrival and another two
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222 219

extinctions (7%) to climatic variation (the only ones in the analysis),
leaving 24 cases unexplained. On Eurasia, extinctions covered a
longer period than anywhere else, encompassing MLADs from 40 to
10 thousand years ago. Therefore, most of them could not be
explained by cold temperature peaks of the LGM or any stadials.
The colonization of Eurasia was the only moment in human
dispersal when paleolithic humans euntil then, a fully tropical
species ewere forced to move against increasingly colder envi-
ronments as they expanded their populations toward high lati-
tudes. The HFADs from Central Russia and Europe are over 45
thousand years old, about 30 thousand years older than those from
Bering and almost 20 thousand years older than the first human
Fig. 2. Geographic distribution of the results of the null model investigating the effects of two factors on megafaunal extinctions across the world. P-values in red show cases were
extinctions happened in (a) periods of high climatic variation as expressed by
18
Oor
2
H concentrations in ice cores and (b) closer than expected by chance to human arrival dates.
Histograms show the distribution of p-values under the two hypotheses. The lower image (c) shows the geographical distribution of cases where the extinctions were better
explained by climatic variations, human arrival, both factors (entangled) and none. Points are slightly jittered to minimize overlap.
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222220

dates across the arctic (Pitulko et al., 2004). Accordingly, Surovell
et al. (2009), after correcting for taphonomic bias, found evidence
that mammoths suffered a long process of decline toward extinc-
tion, correlated to increased human densities. Thus, the long pro-
cess of peopling northern Eurasia against the cold gradient may
have allowed a greater temporal gap between HFADs and MLADs,
generating a weaker signal in the null model.
Another possible explanation for why extinctions were slower
in Eurasia is the coexistence of megafauna with early Homo for
several thousand years prior to the arrival of modern humans
(Sandom et al., 2014). H.heidelbergensis and H.neanderthalensis had
been present there for hundreds of thousands of years before the
diaspora of modern humans to Eurasia (Stringer and Andrews,
2005). So, as in Africa, coevolution possibly granted the large ani-
mals a greater resilience to hunting by H.sapiens. One must keep in
mind that the true baseline for the behavior of megafauna at first
contact with hominids is more likely to be represented by the “is-
land naivety”recorded for vertebrates in historical first contacts
than by modern behavior of animals that have coexisted with
humans for millennia (Diamond, 1984).
Our work is based on few assumptions, including that biological
impacts of climatic variation would be apparent within one to five
thousand year intervals (see Materials and methods), and that
humans had significant interactions with the megafauna. Some
authors have questioned the latter assumption, arguing that co-
incidences of HFADs and MLADs are not enough to establish cau-
sality, and that direct evidence that humans hunted megafauna is
scarce (Grayson and Meltzer, 2003). However, the extinct species
were the demographically most vulnerable ones (Johnson, 2002,
2006), and the frequency and geographical distribution of the
existing humanemegafauna associations are consistent with the
expected given the short coexistence between H.sapiens and large
animals in any given locality (Barnosky, 2004; Surovell and Grund,
2012).
5. Conclusion
The present study allowed a widened quantitative perspective
on the relative roles of climate variation and human impacts in the
LQE, with human impacts as a much stronger determinant of the
number of extinct genera than climate variance. The extinction of
hundreds of species across the whole planet is such a complex
process that one could hardly expect a single factor to completely
explain every aspect of it. It is highly likely that causal factors like
human impacts and climate changes both acted, in different ways
in different places, to produce the final outcome, as our findings
highlight. However, as recent analyses are making increasingly
clear, this does not necessarily mean that both factors had equal
importance. Many Late Quaternary megafaunal extinctions
occurred in the absence of any relevant climatic change, but
seldom, if ever, they occurred independently from human arrival.
Thus, climatic variations were a contributory cause, which some-
times helped determining where and which species went extinct,
while anthropogenic impacts were the necessary cause, without
which probably nobody would be talking today about the extinc-
tion of the megafauna.
Acknowledgements
We thank Joaquín Hortal, Leonardo

Avila, Leopoldo Solbeizon,
the late Paul S. Martin, the members of the Laborat
orio de Ecologia
e Conservaç~
ao de Populaç~
oes eUFRJ and the members of the First
Peopling of the Americas UNESCO Symposium at Puebla for dis-
cussions. We also thank Caio Kenup for his help with R program-
ming, and Chris Johnson for his constructive commentaries on the
manuscript. All authors except M.S.L.-R. were supported by per-
sonal grants from CNPq (Conselho Nacional de Desenvolvimento
Científico e Tecnol
ogico eBrazil) during this research; M.S.L.-R.
received a graduated fellowship from FAPEG (Fundaç~
ao de
Amparo 
a Pesquisa do Estado de Goi
as).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.quaint.2015.10.045.
References
Alroy, J., 2001. A multispecies overkill simulation of the end-Pleistocene megafaunal
mass extinction. Science 292, 1893e1896.
Andersen, K.K., et al., 2004. High-resolution record of northern hemisphere climate
extending into the last interglacial period. Nature 431, 147e151.
Barnosky, A.D., 2004. Assessing the causes of late Pleistocene extinctions on the
continents. Science 306, 70e75.
Barnosky, A.D., Lindsey, E.L., 2010. Timing of Quaternary megafaunal extinction in
South America in relation to human arrival and climate change. Quaternary
International 217, 10e29.
Burney, D.A., Burney, L.P., Godfrey, L.R., Jungers, W.L., Goodman, S.M., Wright, H.T.,
Jull, A.J.T., 2004. A chronology for late prehistoric Madagascar. Journal of Human
Evolution 47, 25e63.
Cione, A.L., Tonni, E.P., Soibelzon, L., 2003. The broken zig-zag: late Cenozoic large
mammal and tortoise extinction in South America. Revista del Museo Argentino
de Ciencias Naturales 5, 1e19.
Diamond, J.M., 1984. Historic extinction: a Rosetta stone for understanding pre-
historic extinctions. In: Martin, P.S., Klein, R.G. (Eds.), Quaternary Extinctions: a
Prehistoric Revolution. Univ. Arizona Press, Tucson, pp. 824e862.
Diniz-Filho, J.A., 2004. Macroecological analyses support an overkill scenario for
Late Pleistocene extinctions. Brazilian Journal of Biology 64, 407e414 .
Gill, J.A., Williams, J.W., Jackson, S.T., Lininger, K.B., Robinson, G.S., 2009. Pleistocene
megafaunal collapse, novel plant communities, and enhanced fire regimes in
North America. Science 326, 1100e1103 .
Gill, J.A., McLauchlan, K.K., Skibbe, A.M., Goring, S., Zirbel, C.R., Williams, J.W., 2013.
Linking abundance of the dung fungus Sporormiella to the density of bison:
implications for assessing grazing by megaherbivores in palaeorecords. Journal
of Ecology 101, 1125e1136 .
Gillespie, R., 2008. Updating Martin's global extinction model. Quaternary Science
Reviews 27, 2522e2529.
Grayson, D.K., 2008. Holocene underkill. Proceedings of the National Academy of
Sciences USA 108, 4077e4078.
Grayson, D.K., Meltzer, D.J., 2003. A requiem for North American overkill. Journal of
Archaeological Sciences 30, 585e593.
Grün, R., Wells, R., Eggins, S., Spooner, N., Aubert, M., Brown, L., Rhodes, E., 2008.
Electron spin resonance dating of South Australian megafauna sites. Australian
Journal of Earth Sciences 55, 917e935.
Grün, R., Eggins, S., Aubert, M., Spooner, N., Pike, A.W.G., Müller, W., 2010. ESR and
U-series analysis of faunal material from Cuddie Springs, NSW, Australia: im-
plications for the timing of the extinction of the Australian megafauna. Qua-
ternary Science Reviews 29, 596e610.
Iwase, A., Hashizume, J., Izuho, M., Takahashi, K., Sato, H., 2012. Timing of mega-
faunal extinction in the Late Pleistocene on the Japanese Archipelago. Quater-
nary International 255, 114e124.
Johnson, C.N., 2002. Determinants of loss of mammal species during the late
Quaternary ‘megafauna’extinctions: life history and ecology, but not body size.
Proceedings of the Royal Society of London B 269, 2221e
2227.
Johnson, C.N., 2006. Australia's Mammal Extinctions: a 500 00 Year History. Cam-
bridge Univ. Press, Cambridge.
Johnson, C.N., Bradshaw, C.J.A., Cooper, A., Gillespie, R., Brook, B.W., 2013. Rapid
megafaunal extinction following human arrival throughout the New World.
Quaternary International 308e309, 273e277.
Klein, R.G., 1984. Mammalian extinctions and Stone Age people in Africa. In:
Martin, P.S., Klein, R.G. (Eds.), Quaternary Extinctions: a Prehistoric Revolution.
Univ. Arizona Press, Tucson, pp. 553e573.
Koch, P.L., Barnosky, A.D., 2006. Late Quaternary extinctions: state of the debate.
Annual Review of Ecology, Evolution and Systematics 37, 215e250.
Lima-Ribeiro, M.S., Nogu
es-Bravo, D., Marske, K.A., Fernandez, F.A.S., Araujo, B.,
Diniz-Filho, J.A.F., 2012. Human arrival scenarios have a strong influence on
interpretations of the late Quaternary extinctions. Proceedings of the National
Academy of Sciences USA 109, E2409eE2410.
Lima-Ribeiro, M.S., Diniz-Filho, J.A., 2013. Modelos Ecol
ogicos e a Extinç~
ao da
Megafauna: Clima e Homem na Am
erica do Sul. Editora Cubo, S~
ao Carlos.
Lorenzen, E.D., et al., 2011. Species-specific responses of late Quaternary megafauna
to climate and humans. Nature 479, 359e365.
Lyons, S.K., Smith, F.A., Brown, J.H., 2004. Of mice, mastodons, and men: human
mediated extinctions on four continents. Evolutionary Ecology Research 6,
339e358.
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222 221

Mead, J.I., Meltzer, D.J., 1984. North American late Quaternary extinctions and the
radiocarbon record. In: Martin, P.S., Klein, R.G. (Eds.), Quaternary Extinctions: a
Prehistoric Revolution. Univ. Arizona Press, Tucson, pp. 441e450.
Nogu
es-Bravo, D., Rodríguez, J., Hortal, J., Batra, P., Araújo, M.B., 2008. Climate
change, humans, and the extinction of the woolly mammoth. Plos Biology 6,
e79.
Pitulko, V.V., Nikolsky, P.A., Girya, E.Yu, Basilyan, A.E., Tumskoy, V.E., Koulakov, S.A.,
Astakhov, S.N., Pavlova, E.Yu, Anisimov, M.A., 2004. The Yana RHS site: humans
in the Arctic before the last glacial maximum. Science 303, 52e56.
Prescott, G.W., Williams, D.R., Balmford, A., Green, R.E., Manica, A., 2012. Quanti-
tative global analysis of the role of climate and people in explaining late Qua-
ternary megafaunal extinctions. Proceedings of the National Academy of
Sciences USA 109, 4527e4531.
R Core Team, 2012. R: a Language and Environment for Statistical Computing.
Foundation for Statistical Computing, Vienna. URL. http://www.R-project.org/.
Rule, S., Brook, B.W., Haberle, S.G., Turney, C.S.M., Kershaw, A.P., Johnson, C.N., 2012.
The aftermath of megafaunal extinction: ecosystem transformation in Pleisto-
cene Australia. Science 335, 1483e1486.
Sandom, C., Faurby, S., Sandel, B., Svenning, J.-C., 2014. Global late Quaternary
megafauna extinctions linked to humans, not climate change. Proceedings of
the Royal Society B 281, 20133254.
Stringer, C., Andrews, P., 2005. The Complete World of Human Evolution. Thames,
Hudson, New York.
Surovell, T.A., Finley, J.D., Smith, G.M., Brantingham, P.J., Kelly, R., 2009. Correcting
temporal frequency distributions for taphonomic bias. Journal of Archaeological
Science 36, 1715e17 24 .
Surovell, T.A., Grund, B.S., 2012. The associational critique of Quaternary overkill
and why it is largely irrelevant to the extinction debate. American Antiquity 77,
673e688.
Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S. Springer, New
York.
Walker, M., 2005. Quaternary Dating Methods. John Wiley, Sons, New Jersey.
Wroe, S., Field, J.H., Archer, M., Grayson, D.K., Price, G.J., Louys, J., Faith, J.T.,
Webb, G.E., Davidson, I., Mooney, S.D., 2013. Climate change debate over the
extinction of megafauna in Sahul (Pleistocene Australia eNew Guinea). Pro-
ceedings of the National Academy of Sciences USA 110, 8777e8781.
B.B.A. Araujo et al. / Quaternary International 431 (2017) 216e222222