PALEOECOLOGY

Plio-Pleistocene decline of Africanmegaherbivores: No evidencefor ancient hominin impactsJ. Tyler Faith1,2*, John Rowan3,4, Andrew Du5, Paul L. Koch6

It has long been proposed that pre-modern hominin impacts drove extinctions and shaped theevolutionary history of Africa’s exceptionally diverse large mammal communities, but thishypothesis has yet to be rigorously tested.We analyzed eastern African herbivore communitiesspanning the past 7 million years—encompassing the entirety of hominin evolutionaryhistory—to test the hypothesis that top-down impacts of tool-bearing, meat-eating homininscontributed to the demise of megaherbivores prior to the emergence of Homo sapiens.Wedocument a steady, long-term decline ofmegaherbivores beginning ~4.6million years ago, longbefore the appearance of hominin species capable of exerting top-down control of largemammal communities and predating evidence for hominin interactions with megaherbivoreprey. Expansion of C4 grasslands can account for the loss of megaherbivore diversity.

Africa is home to more species of large-bodied mammalian herbivores than any-where else today (1). Because most of theworld’s large-bodied vertebrates becameextinct toward the end of the Pleistocene

(2), present-day African faunas serve as modelsystems for understanding the ecology of largemammal communities (3) and the impact ofmassive megaherbivores (>1000 kg) on ecosys-tems (4). Such knowledge feeds directly intoconservation biology on a global scale by high-lighting the ecological consequences of ongoinglarge mammal diversity loss (1) and illustratingthe potential consequences of their rewilding (5).There is a perception that Africa’s exceptional

large herbivore diversity is due to the continent

being spared the extinctions that occurred else-where as modern humans (Homo sapiens) dis-persed across the world in the past 100,000 years(2). This anomaly has been thought to reflectcoevolution of hominin hunters and their prey(6) or perhaps a long history of extinctions pre-cipitated by pre-modern hominins (7) in Africa.Indeed, for decades it has been suggested thathominins drove extinctions and shifts in thefunctional structure of large mammal commu-nities throughout the Pleistocene (7–13). Many ofthese hypotheses posit that top-down control ofmammal communities by tool-bearing, meat-eating hominins contributed to the demise oflarge-bodied herbivores (e.g., the formerly di-verse Proboscidea) long before the emergence

of Homo sapiens (7, 9, 11, 12). Other versions ofthe “ancient impacts” hypothesis propose thatencroachment of Early Pleistocene Homo intothe carnivore guild led to the demise of severalcarnivoran lineages (8, 10), perhaps leading toenvironmental changes through relaxed preda-tion pressure on large-bodied herbivores (10).Both scenarios imply that ancient homininsplayed a key role in shaping African ecosystems,and by extension the environmental settings thatinfluenced our own evolutionary history.Despite decades of literature asserting ancient

hominin impacts on African faunas, there havebeen few attempts to test this scenario or to ex-plore alternatives. Here, we tested the hypothesisof top-down hominin impacts on mammal com-munities through the analysis of megaherbivorediversity over the past 7 million years (Ma) ineastern Africa. Our focus on this region reflectsits rich andwell-dated late Cenozoic fossil record,coupled with the fact that the earliest knownmembers of the hominin clade are from easternAfrica, which therefore provides the longest well-documented history of hominin-mammal com-munity interactions in the world. On the basis ofprevious hypotheses (7–13), we expect declines inmegaherbivore community richness to followor temporally coincide with hominin expansion

RESEARCH

Faith et al., Science 362, 938–941 (2018) 23 November 2018 1 of 4

1Natural History Museum of Utah, University of Utah, SaltLake City, UT 84108, USA. 2Department of Anthropology,University of Utah, Salt Lake City, UT 84112, USA.3Institute of Human Origins and School of HumanEvolution and Social Change, Arizona State University,Tempe, AZ 85282, USA. 4Department of Anthropology,University of Massachusetts, Amherst, MA 01003, USA.5Department of Organismal Biology and Anatomy,University of Chicago, Chicago, IL 60637, USA.6Department of Earth and Planetary Sciences, Universityof California, Santa Cruz, CA 95064, USA.*Corresponding author. Email: jfaith@nhmu.utah.edu

Fig. 1. Megaherbivore richness in modern and fossil communities.(A) Geographic distribution of the 203 modern (continental map) and 101fossil (inset map) herbivore communities. (B) Relationship between thetotal number of herbivore species and the proportion of megaherbivorespecies in modern African communities and eastern African fossilassemblages. The solid line represents the maximum proportion ofmegaherbivores that could coexist today, based on the empirical observationof at most five sympatric megaherbivore species. Fossil assemblages fallingabove the line are non-analog because they include a greater proportion of

megaherbivores than is observed today. (C) Megaherbivore richness residualsover the past 7 Ma, illustrating the long-term decline of megaherbivoresstarting ~4.6 Ma ago. Data points represent residuals from the least-squaresregression modeling the relationship of megaherbivore richness as a functionof total community richness in the modern communities (fig. S2).The solidgray line represents LOESS (locally estimated scatterplot smoothing)regression with smoothing factor = 0.75; 95% confidence limits are shown inlight gray. Horizontal dashed lines encompass the middle 95% range ofvariation in the modern communities.

on October 11, 2020

http://science.sciencem

ag.org/D

ownloaded from

into carnivore niche space. Specifically, propo-nents of the ancient impacts hypothesis typicallyplace the onset of anthropogenic diversity declinebetween 2 and 1 Ma ago (7, 8, 10–12). This en-compasses the earliest evidence for systematichominin predation upon large-bodied mammals(~2 Ma ago) (14) and megaherbivores (~1.95 Maago) (15), as well as the appearance of Homoerectus (~1.9 Ma ago), the first hominin specieswhose paleobiology is similar to later represent-atives of our genus and that consumed largeamounts of animal tissue (16). These evolutionarychanges are thought to account for an unprec-edented reduction of megaherbivore diversitycoupledwith collapse of the large carnivore guild(7, 8, 10–12).We used present-day and fossil herbivore com-

munity data to quantify long-term changes in therichness of eastern African megaherbivores (17).A dataset of more than 200 modern communi-ties from protected areas across Africa allowedus to establish a baseline for present-day var-iability in megaherbivore community richness(Fig. 1A, table S1, and data S1). In addition, wecompiled a fossil dataset that includes the pres-ence of herbivore taxa in 101 eastern African fossilassemblages spanning the past ~7 Ma (table S2and data S2), a period that encompasses the

earliest probable and definitive hominin speciesin eastern Africa. Our focus on individual fossilassemblages within a single region provides apertinent spatiotemporal scale for our researchquestion because it allows a more direct assess-ment of hominin impacts on ancient herbivorecommunities than is possible from analyses ofspecies occurrences at continental scales [e.g.,(9)]. Because hominin impacts need not be theonly driver of change in the eastern Africanmega-herbivore community, we also examined trendsin megaherbivore community richness relativeto independent records of climatic and environ-mental change. These include global atmosphericpartial pressure of CO2 (pCO2) (18), the percentageof C4 biomass (e.g., tropical grasses) inferredfrom the stable carbon isotope (d13C) compo-sition of soil carbonates in eastern Africa (19),estimates of paleo-aridity derived from the stableoxygen isotopes (d18O) of eastern African fossilherbivores (20), and the percentage of C4 grazersamong ungulate taxa in eastern African fossilassemblages (20). The eastern African proxy datacome from many of the same sites examined inour analysis of megaherbivore diversity.Our compilation of the eastern African fossil

record reveals substantial megaherbivore ex-tinctions through time. Over the past 7 Ma,

28 megaherbivore lineages became extinct (tableS3), leading to present-day communities thatare depauperate in megaherbivores. For exam-ple, modern African herbivore communitiesinclude only up to five sympatricmegaherbivores(data S1), including all of the extant species(Giraffa camelopardalis,Hippopotamusamphibius,Ceratotherium simum, Diceros bicornis, andLoxodonta africana). In contrast, the fossil re-cord documents paleocommunities that were con-siderably richer in megaherbivores, with someassemblages documenting the co-occurrence ofas many as 10 megaherbivore species (data S2).Although time-averaging of fossil assemblagescan produce associations of species that neverspatiotemporally co-occurred, it cannot accountfor the exceptional richness of megaherbivoreshere (17) (fig. S1). Controlling for communityrichness, a variable that increases as a functionof time-averaging and sampling effort, manyfossil assemblages over the past ~7 Ma fall out-side the modern range of variation because theyinclude both a greater proportion and a greaterabsolute number of megaherbivore species (Fig. 1,B and C).To illustrate temporal trends (Fig. 1C), we

calculated residuals for fossil assemblages asdeviations fromexpectedmegaherbivore richness

Faith et al., Science 362, 938–941 (2018) 23 November 2018 2 of 4

Fig. 2. The declineof megaherbivorerichness relative toclimatic andenvironmentalproxies. (A) GlobalpCO2. (B) Percentageof C4 vegetationinferred from d13C ofeastern Africanpaleosol carbonates.(C) Estimates of waterdeficit (aridity) foreastern African fossilsites based on d18O ofherbivore toothenamel. Error barsrepresent SE of themean water deficitestimates. (D) Per-centage of C4 grazersamong herbivoretaxa (Artiodactyla-Perissodactyla-Proboscidea) fromeastern African fossilsites, calculated as thepercentage of taxawithmean enamel d13Cvalues > –1 per mil.The gray line in allpanels represents theLOESS regression(95% confidence limitsin light gray) for mega-herbivore richnessresiduals, as in Fig. 1C.

RESEARCH | REPORTon O

ctober 11, 2020

http://science.sciencemag.org/

Dow

nloaded from

modeled as a linear function of herbivore com-munity richness based on our modern com-munity data (fig. S2). Residual analysis highlightsthe exceptional richness of fossil megaherbivorecommunities—many are well outside the mod-ern range of variation—and documents a steadyand long-term decline in richness beginningin the early Pliocene, with fossil assemblagesconsistently falling within the modern rangeof variation only after ~0.7 Ma ago (Fig. 1C).Breakpoint analysis places the onset of themegaherbivore decline at ~4.6 Ma ago (95%confidence interval, 3.3 to 5.9 Ma ago). After the~4.6Ma breakpoint, the rate of diversity declinethrough time does not change following theappearance of Homo erectus (1.9 Ma ago) or ateither end of the hypothesized window of ac-celerated anthropogenic diversity decline, 2 to1 Ma ago (table S4). Instead, the long-term de-cline in megaherbivore richness closely tracksglobal variation in atmospheric pCO2 as well asthe expansion of C4 grasslands and C4 grazersin eastern Africa, although there is no associa-tion with paleo-aridity (Fig. 2).The loss of large-bodied mammals is thought

to be a hallmark of anthropogenic extinctions(6, 7, 9), in part because large-bodied speciesare likely to have been preferentially targetedby hominin hunters and because their slowerlife history profiles (e.g., delayed reproductivematurity, prolonged gestation, low populationgrowth rates) render them more susceptible toextinction. Our analyses show that the diversityof megaherbivores in eastern African fossil as-semblages has undergone a steady decline since~4.6 Ma ago (Fig. 1), beginning long before

the proposed timing of anthropogenic impacts(2 to 1Ma ago) linked to encroachment ofHomoerectus into the carnivore guild (Fig. 3). Theantiquity of the megaherbivore decline effec-tively eliminates top-down hominin impactsas a plausible mechanism for setting it in mo-tion, as it would have had to involve small-bodied australopiths or pre-australopiths (e.g.,Australopithecus,Ardipithecus), whichwere func-tionally equivalent to bipedal apes. Like extantchimpanzees, these hominin taxa may havepreyed upon vertebrate species smaller thanthemselves, but they almost certainly did nothunt megaherbivore prey (21). Although we can-not rule out the possibility that the subsequentappearance of more derived hominin species,new stone-tool technologies, and increased car-nivory may have incidentally contributed to thedemise of megaherbivores, the steady decline ofmegaherbivores beginning ~4.6 Ma ago (Fig. 3and table S4)—in contrast to proposed extinctionpulses linked to major shifts in hominin evolu-tion (7)—implies that the primary driver was de-coupled from hominin evolution.We propose that the expansion of C4 grass-

lands (Fig. 2B) played a critical role in mega-herbivore decline. Analysis of our fossil datasetindicates that the long-term decline of mega-herbivores is primarily due to the loss of mega-herbivore browsers and mixed feeders thatconsumed C3 plants, including trees, shrubs, andherbs (fig. S3). Unlike megaherbivore grazers,temporal declines in the richness of mega-herbivore browsers and mixed feeders are inclose agreement with the overall megaherbi-vore diversity loss (fig. S3). Loss of C3 consum-

ers is also evident in a compilation of d13C datafrom tooth enamel of eastern African mega-herbivores (fig. S3 and data S3). Their declinein fossil assemblages is likely related to the Plio-Pleistocene expansion of C4 grasslands (Fig. 2B),which reduced the availability of C3 forage onthe landscape. Because larger-bodied species aremore strongly limited by forage availability thansmaller-bodied species (4), a decline in mega-herbivore species dependent on C3 forage isexpected. At the same time, decreasing atmo-spheric CO2 concentrations (Fig. 2A) would havediminished the advantage of massive body size,which allows megaherbivores to consume lower-quality foods than their smaller-bodied counter-parts (4). Especially among C3 plant species,plant tissue grown at lower CO2 concentrationsprovides higher-quality forage because such plantshave higher N content (lower C:N ratios) andfewer secondary compounds (22). This wouldallow smaller-bodied species—which are restrictedto high-quality forage (3)—to exploit a greaterrange of increasingly rare C3 resources, withthe outcome being less food available to mega-herbivore browsers and mixed feeders. Someor all of these mechanisms are likely to haveplayed a role in driving large herbivore extinc-tions elsewhere. For example, although the timingof C4 expansion and the nature of its ecologicalconsequences varied globally as a result of differ-ences in climate and historical contingencies (23),the C4 expansion in the Siwalik Group (Pakistan)from ~8.5 to 6.0 Ma ago is associated with con-siderable losses among C3 consumers and the lastappearances of many massive herbivores, includ-ing five rhino species, a proboscidean, a chalico-there, and a sivathere (24).There is a long history of debate concerning

the mechanisms responsible for the expansionof C4 grasslands in eastern Africa (19). This phe-nomenon is often linked to increased aridityafter the onset of Northern Hemisphere Glacia-tion, ~2.8 Ma ago (25). However, dust flux re-cords from marine sedimentary archives thatreflect eastern African climate indicate substan-tial long-term increases in aridity only after~1.5 Ma ago (26), long after the onset of C4

expansion (Fig. 2B). In addition, recent analy-ses show that terrestrial proxies for aridity andvegetation in eastern Africa vary independentlythrough the Plio-Pleistocene, implying that otherabiotic or biotic mechanisms likely underpinhabitat change (20). Because C4 grasses arefavored at lower CO2 concentrations (27), thePlio-Pleistocene CO2 decline (Fig. 2A) is likelyan important abiotic mechanism (19). For ex-ample, at high temperatures, such as those in-ferred from Turkana Basin paleosol carbonates(>30°C) (28), C4 grasses are expected to expandwhen atmospheric CO2 concentrations fall below~450 parts per million (27). The expansion of C4grasslands and associated decline of megaherbi-vores are consistent with long-term decline ofCO2 (Fig. 2), likely facilitated by episodes ofaridity (i.e., positive water deficit) that occurredacross the interval (Fig. 2C). The persistentexpansion of C4 grasslands is remarkable in

Faith et al., Science 362, 938–941 (2018) 23 November 2018 3 of 4

Fig. 3. The decline of megaherbivore richness relative to milestones in hominin evolution.These include the appearance of novel technologies and behaviors, as well as the observed temporalranges of eastern African hominin taxa. The vertical dashed line indicates the breakpoint denotingonset of megaherbivore decline (4.6 Ma ago), with red shading representing the 95% confidenceinterval (3.3 to 5.9 Ma ago).

RESEARCH | REPORTon O

ctober 11, 2020

http://science.sciencemag.org/

Dow

nloaded from

light of the decline of megaherbivore diversity.Megaherbivore browsers and mixed feeders,especially elephants (L. africana) and blackrhinoceros (D. bicornis), promote grassland ex-pansion through consumption of woody (C3)vegetation, toppling and ringbarking of trees,and synergistic effects with fire (4, 29). With allother factors held constant, we would expectthe considerable decline of such taxa—whichimplies a reduction in megaherbivore biomass(fig. S4)—to drive an expansion of woody coverthrough the Plio-Pleistocene. That the exactopposite occurred (Fig. 2B) suggests a domi-nance of abiotic mechanisms in driving the C4

expansion.We note that the CO2-driven expansion of

C4 grasslands and the associated extinction ofmegaherbivores can account for other changes ineastern African mammal communities that havebeen attributed to ancient hominin impacts. Asnoted earlier, the loss of richness and functionaldiversity among eastern African carnivoransthrough the Pleistocene has been attributed tothe encroachment of Homo into the large car-nivore guild (8, 10–12). However, some extinctPleistocene carnivorans lacking modern ana-logs (e.g., sabertooth felids) are known to havespecialized on juvenile megaherbivores (30).Given the close association between large carni-vore diversity and herbivore diversity (31), theloss of megaherbivore prey likely contributeddirectly to the extinction of attendant carnivores.Thus, in the absence of detailed consideration ofeastern African herbivores, it is premature toinvoke hominin impacts as an important driverof carnivoran diversity loss and associated eco-

system change. Together with our observationsindicating that bottom-up processes can accountfor the loss of megaherbivores, it follows that inthe search for anthropogenic impacts on ancientAfrican ecosystems, we must focus our attentionon the one species known to be capable ofcausing them: Homo sapiens over the past300,000 years.

REFERENCES AND NOTES

1. W. J. Ripple et al., Sci. Adv. 1, e1400103 (2015).2. P. L. Koch, A. D. Barnosky, Annu. Rev. Ecol. Evol. Syst. 37,

215–250 (2006).3. R. H. V. Bell, Sci. Am. 225, 86–93 (July 1971).4. R. N. Owen-Smith, Megaherbivores: The Influence of Very Large

Body Size on Ecology (Cambridge Univ. Press, 1988).5. J.-C. Svenning et al., Proc. Natl. Acad. Sci. U.S.A. 113, 898–906

(2016).6. P. S. Martin, in Quaternary Extinctions: A Prehistoric Revolution,

P. S. Martin, R. G. Klein, Eds. (Univ. of Arizona Press, 1984),pp. 354–403.

7. S. K. Lyons, F. A. Smith, J. H. Brown, Evol. Ecol. Res. 6,339–358 (2004).

8. L. Werdelin, M. E. Lewis, PLOS ONE 8, e57944 (2013).9. F. A. Smith, R. E. Elliott Smith, S. K. Lyons, J. L. Payne, Science

360, 310–313 (2018).10. M. Fortelius et al., Philos. Trans. R. Soc. B 371, 20150232

(2016).11. C. N. Johnson et al., Science 356, 270–275 (2017).12. Y. Malhi et al., Proc. Natl. Acad. Sci. U.S.A. 113, 838–846

(2016).13. P. S. Martin, Nature 212, 339–342 (1966).14. J. V. Ferraro et al., PLOS ONE 8, e62174 (2013).15. D. R. Braun et al., Proc. Natl. Acad. Sci. U.S.A. 107,

10002–10007 (2010).16. S. C. Antón, R. Potts, L. C. Aiello, Science 345, 1236828 (2014).17. See supplementary materials.18. L. B. Stap et al., Earth Planet. Sci. Lett. 439, 1–10 (2016).19. N. E. Levin, Annu. Rev. Earth Planet. Sci. 43, 405–429 (2015).20. S. A. Blumenthal et al., Proc. Natl. Acad. Sci. U.S.A. 114,

7331–7336 (2017).

21. C. Stanford, The Hunting Apes: Meat Eating and the Origins ofHuman Behavior (Princeton Univ. Press, 1999).

22. C. E. Owensby, R. C. Cochran, L. M. Auen, in Carbon Dioxide,Populations, and Communities, C. Körner, F. Bazzaz, Eds.(Academic Press, 1996), pp. 363–371.

23. C. Strömberg, Annu. Rev. Earth Planet. Sci. 39, 517–544 (2011).24. C. Badgley et al., Proc. Natl. Acad. Sci. U.S.A. 105, 12145–12149

(2008).25. P. B. deMenocal, Earth Planet. Sci. Lett. 220, 3–24 (2004).26. M. H. Trauth, J. C. Larrasoaña, M. Mudelsee, Quat. Sci. Rev. 28,

399–411 (2009).27. J. R. Ehleringer, T. E. Cerling, B. R. Helliker, Oecologia 112,

285–299 (1997).28. B. H. Passey, N. E. Levin, T. E. Cerling, F. H. Brown, J. M. Eiler,

Proc. Natl. Acad. Sci. U.S.A. 107, 11245–11249 (2010).29. R. M. Holdo, R. D. Holt, J. M. Fryxell, Ecol. Appl. 19, 95–109

(2009).30. C. W. Marean, C. L. Ehrhardt, J. Hum. Evol. 29, 515–547 (1995).31. C. Sandom et al., Ecology 94, 1112–1122 (2013).

ACKNOWLEDGMENTS

We thank J. O’Connell and members of the University of UtahArchaeological Center for helpful discussions and comments onprevious drafts of this manuscript, S. Blumenthal for providingdata used in Fig. 2C, and G. Hempson for providing data used infig. S4. Funding: J.T.F.’s early contributions to this project weresupported by an Australian Research Council DECRA Fellowship(DE160100030). Author contributions: J.T.F., J.R., and A.D.conceived the project; J.R. compiled the modern and fossil data;J.T.F., J.R., and A.D. analyzed the data; J.T.F., J.R., A.D., and P.L.K.interpreted the data; J.T.F. wrote the first draft and all co-authorsprovided feedback. Competing interests: Authors declare nocompeting interests. Data and materials availability: Data areavailable in the main text or the supplementary materials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/362/6417/938/suppl/DC1Materials and MethodsFigs. S1 to S4Tables S1 to S4Data S1 to S3References (32–149)

23 May 2018; accepted 28 September 201810.1126/science.aau2728

Faith et al., Science 362, 938–941 (2018) 23 November 2018 4 of 4

RESEARCH | REPORTon O

ctober 11, 2020

http://science.sciencemag.org/

Dow

nloaded from

Plio-Pleistocene decline of African megaherbivores: No evidence for ancient hominin impactsJ. Tyler Faith, John Rowan, Andrew Du and Paul L. Koch

DOI: 10.1126/science.aau2728 (6417), 938-941.362Science 

, this issue p. 938; see also p. 892Scienceexpansion of grasslands.million years. Instead, megaherbivore decline may have been triggered by declining atmospheric carbon dioxide and to decline about 4.6 million years ago, preceding evidence for hominin consumption of animal tissues by more than 1years (see the Perspective by Bobe and Carvalho). Megaherbivores (for example, elephants, rhinos, and hippos) began

7 million∼ challenge this view with an analysis of eastern African herbivore communities spanning the past et al.Faith Human ancestors have been proposed as drivers of extinctions of Africa's diverse large mammal communities.

Megaherbivore extinctions in Africa

ARTICLE TOOLS http://science.sciencemag.org/content/362/6417/938

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/11/19/362.6417.938.DC1

CONTENTRELATED http://science.sciencemag.org/content/sci/362/6417/892.full

REFERENCES

http://science.sciencemag.org/content/362/6417/938#BIBLThis article cites 133 articles, 23 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on October 11, 2020

http://science.sciencem

ag.org/D

ownloaded from