Aridity and hominin environments · Aridity and hominin environments Scott A. Blumenthala,1, Naomi E. Levinb, Francis H. Brownc, Jean-Philip Brugald, Kendra L. Chritze, John M. Harrisf,

Aridity and hominin environmentsScott A. Blumenthala,1, Naomi E. Levinb, Francis H. Brownc, Jean-Philip Brugald, Kendra L. Chritze, John M. Harrisf,Glynis E. Jehlec, and Thure E. Cerlingc

aResearch Laboratory for Archaeology, University of Oxford, Oxford OX1 3QY, United Kingdom; bDepartment of Earth and Environmental Sciences,University of Michigan, Ann Arbor, MI 48109; cDepartment of Geology & Geophysics, University of Utah, Salt Lake City, UT 84112; dAix-Marseille University,CNRS, UMR 7269, Laboratoire Méditerranéen de Préhistoire Europe Afrique, 13094 Aix-en-Provence Cedex 2, France; eNational Museum of Natural History,Smithsonian Institution, Washington, DC 20013; and fNatural History Museum of Los Angeles County, Los Angeles, CA 90007

Edited by James O’Connell, University of Utah, Salt Lake City, UT, and approved May 25, 2017 (received for review January 11, 2017)

Aridification is often considered a major driver of long-term ecolog-ical change and hominin evolution in eastern Africa during the Plio-Pleistocene; however, this hypothesis remains inadequately testedowing to difficulties in reconstructing terrestrial paleoclimate. Wepresent a revised aridity index for quantifying water deficit (WD) interrestrial environments using tooth enamel δ18O values, and use thisapproach to address paleoaridity over the past 4.4 million years ineastern Africa. We find no long-term trend in WD, consistent withother terrestrial climate indicators in the Omo-Turkana Basin, and norelationship between paleoaridity and herbivore paleodiet structureamong fossil collections meeting the criteria forWD estimation. Thus,we suggest that changes in the abundance of C4 grass and grazingherbivores in eastern Africa during the Pliocene and Pleistocene mayhave been decoupled from aridity. As in modern African ecosystems,other factors, such as rainfall seasonality or ecological interactionsamong plants and mammals, may be important for understandingthe evolution of C4 grass- and grazer-dominated biomes.

oxygen isotopes | terrestrial paleoclimate | human evolution | mammals |Africa

Acentral challenge of human evolutionary studies is un-derstanding the role of climatic change in shaping early

hominin environments and selective pressures (1, 2). Aridity influ-ences the distribution and abundance of vegetation in African en-vironments (3), and changes in aridity over both long and short timescales have been suggested to drive changes in hominin environ-ments leading to adaptation, dispersal, speciation, and extinction (2,4, 5). The notion that aridity may have driven certain adaptationshas been fundamental to discussions of hominin evolution since1925 (6), and continues to feature prominently in studies addressingchanges in hominin locomotion, body proportions, thermoregula-tion, food acquisition, tool use, and social organization (7–10).Changes in African climate are driven principally by changes in

Earth’s orbital geometry, which has been documented in the geo-logic past using marine and continental sedimentary records (4, 11–15). Marine core records of dust, leaf wax biomarkers, pollen, andsapropels indicate long-term aridification across Africa since thelate Miocene (4, 12, 14, 16–18), which has been linked to globalcooling (19), changes in ocean circulation and temperature gradi-ents (20), high-latitude glaciation (4), low-latitude atmosphericcirculation (14), and tectonic uplift (21). Increasing aridity has beenthought to drive the origin and subsequent expansion of C4 plants(grasses and sedges) (22). The long-term increase in the abundanceof C4 plants throughout the Pliocene and Pleistocene has been welldocumented in eastern Africa using carbon isotope ratios in ped-ogenic carbonates and leaf wax biomarkers (23, 24) and coincideswith an increasing reliance on C4-based resources among mam-mals, including hominins and other primates (25, 26). Variation inthe timing of vegetation change across basins indicates that existingcontinental- and regional-scale climate records are not sufficient tounderstand the drivers of basin- and local-scale ecological change,and do not reflect local hominin environments (23, 27). Evidencefor vegetation changes with precession-scale timing suggests directclimate forcing of such changes over thousands of years (28, 29),

but drivers of environmental change might not be equivalent atshort vs. long time scales and also may vary over time.Uncertainties in the relationships between climate and hominin

environments stem in part from difficulties in reconstructingterrestrial aridity. Terrestrial climate indicators commonly used ineastern Africa, including the isotopic composition of pedogeniccarbonates (21, 27, 30), mammal taxonomy (31–33), and mor-phology (34), provide valuable insight into past environments, butare sensitive to multiple environmental and evolutionary changes,making it difficult to identify the specific role of aridity. In ad-dition, existing faunal records (31–34) typically combine fossilsfrom multiple sites and may integrate relatively long (but varying)time periods. Other climate proxies, such as the deuterium iso-tope composition of leaf wax biomarkers (17) and fossil leafmorphology (35), have not been widely applied in Pliocene-Pleistocene sequences in Africa.In the present study, we address paleoaridity using oxygen iso-

tope ratios (δ18O) in herbivore tooth enamel. Our goal is to in-vestigate the role of climate in shaping hominin environments overthe past 4.4 million years, concentrating on individual stratigraphichorizons associated with hominin fossil and archaeological mate-rial. We focus on the Omo-Turkana Basin, where sediments pre-serve abundant evidence of early hominin evolution and associatedenvironments throughout the Pliocene and Pleistocene. The envi-ronmental history of this basin is not necessarily representative ofeastern Africa (1), but nonetheless provides a useful study systemfor investigating interactions between climate and ecology. A majorbenefit of analyzing herbivore tooth enamel is the possibility ofcomparing paired oxygen and carbon isotope records from thesame fossil collections in which hominin specimens or stone toolshave been found, providing indicators of climate and ecology at

Significance

Oxygen isotopes in modern and fossil mammals can provide in-formation on climate. In this study, we provide a new record ofaridity experienced by early hominins in Africa. We show that pastclimates were similar to the climate in eastern Africa today, andthat early hominins experienced highly variable climates over time.Unexpectedly, our findings suggest that the long-term expansionof grasses and grazing herbivores since the Pliocene, a majorecological transformation thought to drive aspects of homininevolution, was not coincident with aridification in northern Kenya.This finding raises the possibility that some aspects of homininenvironmental variability might have been uncoupled from aridity,and may instead be related to other factors, such as rainfall sea-sonality or ecological interactions among plants and mammals.

Author contributions: S.A.B., N.E.L., F.H.B., and T.E.C. designed research; S.A.B., N.E.L.,F.H.B., J.-P.B., K.L.C., J.M.H., G.E.J., and T.E.C. performed research; S.A.B. and N.E.L. ana-lyzed data; and S.A.B. and N.E.L. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1700597114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1700597114 PNAS | July 11, 2017 | vol. 114 | no. 28 | 7331–7336

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

spatial scales directly relevant to hominin environments. We can-not address short-term orbital scale environmental variability,however, owing to discontinuous sedimentation associated withterrestrial vertebrate fossil collections.Our geochemical approach for quantifying aridity in tropical

African ecosystems relies on differing oxygen isotopic effectsamong taxa that are evaporation-sensitive (ES) or evaporation-insensitive (EI) (36, 37). This method, which builds on earlier workthat focused on oxygen isotope variation among individual taxa(38–42), relies on a comparison of multiple taxa and simulta-neously accounts for isotopic variation related to changes in bothclimate and environmental water. This proxy has advantages overpreviously used paleoaridity indicators because it is largely in-sensitive to changes in (i) vegetation, which control mammal tax-onomic abundances, diet, and carbon isotopic records from toothenamel, soil carbonates, and leaf wax biomarkers; (ii) moisturesource, soil temperature, and elevation, which influence oxygenisotopic records reflecting meteoric water, such as soil carbonatesand leaf wax biomarkers; and (iii) mammal physiology and be-havior, which affect oxygen isotopic records of individual species.Previous applications of this approach to the African fossil recordhave been hampered by uncertainties in the selection of appro-priate taxa and the unavailability of appropriate fossil collections.Aridity is expressed as water deficit (WD), which describes the

annual difference (in mm/y) between water loss (evaporation andtranspiration) and water gain (precipitation) and is a useful in-dicator of water availability in African ecosystems (43, 44). δ18Ovalues in large mammalian herbivore tooth enamel are in equi-librium with body water, which reflects oxygen inputs from food,drinking water, and air, and ultimately relate to meteoric(precipitation-derived) water (45). In the tropics, the oxygen iso-topic composition of meteoric water is related to rainfall amount,elevation, and moisture source (46). Evaporation enriches theremaining water in the heavy isotope 18O relative to source water,such that aridity can be quantified by comparing one isotopic re-cord that tracks meteoric water with another that tracks evapora-tive enrichment (36, 37).The aridity index (36) is based on regressions between the WD

and the oxygen isotopic enrichment between tooth enamel andlocal meteoric water (eenamel-mw). Mammalian herbivore taxa forwhich eenamel-mw increases with WD are classified as ES, and taxafor which eenamel-mw does not change with WD are classified asEI. Meteoric water cannot be measured directly in the fossilrecord; therefore, these relationships can be extended to thefossil record to predict WD by using δ18O values of EI taxa torepresent meteoric water, because eES-EI and eES-mw both trackaridity (36). Applying the aridity index to the fossil record re-quires the assessment of appropriate taxa, geological context,diagenetic alteration, and sample size (SI Appendix).To revise the aridity index, we present a compilation of new and

previously published δ18O values (n = 1,224 in 57 species) mea-sured on tooth enamel from modern mammalian herbivores from37 locations in eastern and central Africa (Fig. 1), along with cli-mate data and WD estimates for each location and δ18O values inmeteoric water (n = 161) from 33 of these locations (SI Appendix,Table S1 and Datasets S1 and S2). Our compilation significantlyexpands on a previously published dataset (36) and includes δ18Odata from many more locations and taxa, and also expands the WDscale owing to the correction of a mathematical error in calculatingpotential evapotranspiration (SI Appendix, Fig. S1). To addresspaleoaridity in eastern Africa, we present a compilation of newand previously published mammalian herbivore δ18Oenamel values(n = 273) from 26 fossil collections (Fig. 1) ranging in age from∼4.4–0.01 Ma, chosen based on their potential for addressingpaleoaridity and their association with hominin fossil and archaeo-logical material (SI Appendix, Dataset S3). We use a subset ofδ18Oenamel values (n = 160) from 11 fossil collections in the Omo-Turkana Basin, including specimens from the Kanapoi, Koobi Fora,

Nachukui, and Kibish Formations (Fig. 1 and SI Appendix, Fig. S3and Tables S2 and S3), that meet the criteria for application of thearidity index to evaluate long-term changes in paleoaridity in thisbasin. We also estimate paleoaridity using previously publishedδ18Oenamel values from two eastern African fossil collections out-side the Turkana Basin that meet the criteria for applying thismethod, including Aramis, Ethiopia (4.4 Ma), and Kanjera South,Kenya (2.0 Ma). Finally, we investigate the relationship betweenaridity and ecosystem structure in eastern Africa using a compi-lation of previously published modern mammalian herbivore tis-sue δ13C values (n > 1,600) (25) and a compilation of new andpreviously published fossil tooth enamel δ13C values (n = 658)from fossil collections with paleoaridity estimates.

Results and DiscussionTerrestrial Aridity Proxy. Across modern localities, WD increasesnonlinearly with decreasing mean annual precipitation (logarithmicregression, R2 = 0.7546, P < 0.0001) and increasing mean annualtemperature (quadratic polynomial regression, R2 = 0.6829, P <0.0001) (SI Appendix, Fig. S1). eenamel-mw values for Hippopot-amidae, Elephantidae, and Rhinocerotidae do not vary with WD,and these taxa are classified as EI (Fig. 2A). eenamel-mw values forGiraffidae, Hippotragini, and Tragelaphini increase with WD, andthese taxa are classified as ES (Fig. 2B). The slopes of WD-eES-mwregressions vary significantly from one another (P < 0.05, F test),except for Giraffidae and Hippotragini (P > 0.05, F test). Despitepreliminary observations to the contrary (36), the addition of morelocations in this dataset reveals that Antilopini, Bovini, and Neo-tragini should be excluded from the ES category (Fig. 2C). Othersampled bovids, suids, and equids (Fig. 2C) have no significant re-lationship with WD (P > 0.05), except Cephalophini (P < 0.05),which are not considered further owing to eenamel-mw values that aremore variable and/or span a restricted WD range. Additional dataare needed to address the variability in eenamel-mw values across bovidgenera, although many bovid fossils are identifiable only to tribe.We use a body water model to identify possible physiological and

behavioral mechanisms driving the relationship between eenamel-mwand WD among EI and ES taxa (SI Appendix). A static oxygenbudget, in which body water is influenced solely by changes in theisotopic composition of oxygen influxes rather than by changes inthe balance of influxes, is inconsistent with the eenamel-mw values ofeither ES or EI taxa (SI Appendix, Fig. S2 C and D); therefore, 18Oenrichment of leaf water in arid environments is insufficient toexplain the relationship between eenamel-mw values and WD amongES taxa. Instead, sensitivity to evaporation is likely related to dif-ferences in drinking behavior and associated changes in the balanceof oxygen influxes as the environment varies. Predicted eenamel-mwvalues suggest that EI taxa reflect meteoric water as aridity

10°N

0°

20°E 30°E

SHUNGURA FM.

KOOBI FORA FM.

NA

CH

UK

UI F

M.

Kanapoi

Lothagam

LO7

KU1

LO4

LO8LO3LO6

LO7KU3

KU2LO1/2

LO10

LA1

KS2KS1

KK

NY2NY3

KL3&6

KL2&6KL1

NC3NC1

KI1

LO5

KG1&2

37°E36°E

4°N

3°N

LK

105-1311105-0208FxJj1&3

FxJj82FwJj20

Allia Bay104

NY4

5 kmA B C

Fig. 1. Map of the study area. (A) Detailed map of fossil exposures (red areas)and sites (red circles) and drainages associated with the Nachukui Formation,west of Lake Turkana. (B) Fossil collection sites and formations in the Omo-Turkana Basin. (C) Map of Africa with sampling locations for modern teethand meteoric water (white circles) and fossil sites (red circles).

7332 | www.pnas.org/cgi/doi/10.1073/pnas.1700597114 Blumenthal et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

increases owing to a balance between increasing drinking water anddecreasing food water (SI Appendix, Fig. S2C), and ES taxa trackincreasing aridity owing to a balance between decreasing drinkingwater and increasing intake of food water and O2, both of whichare sensitive to evaporation (SI Appendix, Fig. S2D).The aridity index reflects the relationship between WD and

the enrichment between ES and EI taxa (eES-EI). SignificantWD-eES-EI regressions (P < 0.05) that can be used to estimatepaleoaridity include eGiraffid-Hippopotamidae, eTragelaphini-Hippopotamidae,eHippotragini-Hippopotamidae, eTragelaphini-Elephantidae, and eTragelaphini-Rhinocerotidae (Fig. 3 and SI Appendix, Table S4). The SEs of theseregression models are relatively low (±193 to ±478.1 mm/y) andcoefficients of determination are high (R2 = 0.81–0.92), and thusthese models have sufficient predictive power to estimate paleo-aridity in the fossil record. Slopes are different among WD-eES-EIregressions (P < 0.05, F test); therefore, a pooled or commonslope, as suggested previously (36), is not appropriate. We use themean of WD values calculated with WD-eES-EI regressions for allavailable ES-EI pairs from each fossil collection. Uncertainty inWD estimates (∼800 mm/y) corresponds to ∼20% of the WDrange in modern eastern African environments, sufficient to de-tect long-term trends in Turkana (SI Appendix). Other WD-eES-EIregressions are not significant (P > 0.05) and should not be used toestimate paleoaridity.

Paleoaridity. Our oxygen isotope analyses of fossil tooth enamelfor paleoaridity estimation were restricted to fossil collections withwell-defined stratigraphic and sedimentological context and the

preservation of appropriate ES and EI taxa. Among all fossil col-lections used to evaluate paleoaridity, we find highly variable condi-tions, including both mesic (WD < 0) and arid (WD > 0) climatesthat fall within a WD range (∼−550–1,700 mm/y) encompassing∼61% of the modern range (Fig. 4 and SI Appendix, Fig. S4). Themean WD estimated by fossil tooth enamel is 471 mm/y, similar tothe present-day mean WD in eastern and central Africa (251 mm/y;P > 0.05) (SI Appendix, Figs. S1 and S4 and Tables S1 and S3).Among fossil collections from the Omo-Turkana Basin, we detect nolong-term trend inWD between∼4.2 and 0.01Ma (P > 0.05) (Fig. 4).The paleoaridity record (Fig. 4 and SI Appendix, Fig. S4 and

Tables S2 and S3) begins in the Pliocene with a highly uncertainestimate of arid conditions associated with Ardipithecus ramidusfossils from the Lower Aramis Member, Sangatole Formation(∼4.4 Ma) (47). In the Omo-Turkana Basin, we find arid condi-tions at Kanapoi (∼4.16 Ma) and a highly uncertain estimate ofmesic conditions at Allia Bay (∼4.0 Ma), both associated withAustralopithecus anamensis (48), and pedogenic carbonate δ13Cvalues indicative of vegetation ranging from woodland/bushland/shrubland to wooded grassland (49). Mid- to late-Pliocene (∼3.5–2.8 Ma) fossil collections in Turkana indicate variable conditionsthat include arid (Kangatukuseo KU1) and arid to mesic(Lomekwi LO4/5) climates, associated with pedogenic carbonateδ13C values indicating woody cover, including woodland/bushland/shrubland (Fig. 4) (49). This time interval in Turkana includesfossils of the hominin genera Kenyanthropus and Paranthropus. Theearly Pleistocene (∼2.5–1.5 Ma) is represented in the TurkanaBasin by fossil assemblages in the upper Burgi Member of theKoobi Fora Formation as well as the Kaito Member of theNachukui Formation, which indicate arid (FwJj20 and KalochoroKL3/6, Naiyena Engol NY2/3), nearly balanced (KokiseleiKS2 and Kangatukuseo KU2/3), and mesic (Kokiselei KS1) con-ditions. Pedogenic carbonate δ13C values indicate that relativelyopen vegetation, including wooded grasslands, became moreprevalent during this time interval in Turkana (49), which includes

Equidae

SuidaeBovidae

A B

E. burchellii E. grevyi

Cephalophini

p = 0.04044r2 = 0.92

Alcelaphini

Antilopini

Reduncini

Phacochoerus Potamochoerus

Hylochoerusmeinertzhageni

Hippotragini

p = 0.002041r2 = 0.71

Neotragini

Giraffidae

p = 0.002126r2 = 0.71

Tragelaphini

p < 0.001r2 = 0.82

C

Hippopotamidae Elephantidae

Rhinocerotidae

0 2000

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

WD (mm/y)

Aepycerotini

30

35

40

45

Bovini

enam

el-m

wen

amel

-mw

enam

el-m

wen

amel

-mw

enam

el-m

wen

amel

-mw

WD (mm/y) WD (mm/y)

WD (mm/y)

WD (mm/y)

WD (mm/y)

WD (mm/y)

enam

el-m

wen

amel

-mw

enam

el-m

wen

amel

-mw

enam

el-m

w

WD (mm/y)

WD (mm/y)

WD (mm/y)

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

30

35

40

45

-1000 1000

0 2000-1000 1000

0 2000-1000 1000

0 2000-1000 1000

0 2000-1000 1000

0 20001000

0 20001000

0 20001000

0 20001000

0 20001000

Fig. 2. Isotopic enrichment between enamel and meteoric water (eenamel-mw)among eastern and central African herbivores. (A) EI taxa. (B) ES taxa. (C) Otherbovids, equids, and suids. Error bars represent propagated SE of eenamel-mw val-ues. Data are compiled in SI Appendix, Datasets S1 and S2.

0

5

10

15

0

5

10

15

0

5

10

15

WD (mm/y) WD (mm/y) WD (mm/y)

0 1000 2000-10000 2000-10000 1000 2000-1000

p = 0.188

p = 0.0005727r2 = 0.92

p = 0.106

p = 0.2557

p < 0.0001r2 = 0.8911

p = 0.05341

p = 0.00711r2 = 0.87

p < 0.0001r2 = 0.88

p = 0.002208r2 = 0.81

ES

-EI

ES

-EI

ES

-EI

Gira

ffida

eTr

agel

aphi

niH

ippo

trag

ini

1000

RhinocerotidaeElephantidaeHippopotamidae

Fig. 3. Isotopic enrichment (eES-EI) between modern ES and EI taxa. eES-EI wascalculated using mean δ18Oenamel values of sampled ES taxa (rows) and EI taxa(columns) from eastern and central Africa. Dashed lines indicate significantWD-eES-EI regressions (SI Appendix, Table S4). Error bars represent propagatedSE of eES-EI. Calculated from values provided in SI Appendix, Dataset S2.

Blumenthal et al. PNAS | July 11, 2017 | vol. 114 | no. 28 | 7333

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

an abundant fossil record of Homo and Paranthropus. Mesic con-ditions prevailed at Kanjera South KS-2 in southwestern Kenya,associated with an open grassland ecosystem (50). Archaeologicaloccurrences at Kanjera South, as well as in the Nachukui andKoobi Fora Formations in the Turkana Basin, demonstrate thatOldowan tool-making hominins inhabited mesic and arid envi-ronments. The late Middle Pleistocene to Holocene (∼0.2–0.01Ma) is represented in the Turkana Basin by fossil assemblages inthe Kibish Formation, which indicate arid conditions in Member4 and arid to mesic conditions in Member 1. Fossils identified asHomo sapiens (Omo I and Omo II) are from Member 1, and otherhuman specimens are derived from either Member 3 or Member 4.WD estimates in Members 1 and 4 are substantially lower thanthose in Turkana today (modern WD = 2,386 mm/y; SI Appendix,Table S1), consistent with deposition during relatively humidperiods associated with high lake levels and sapropel formationintervals (51).

Relationships Between Climate and Ecology. To understand thesignificance of aridity in shaping hominin environments in easternAfrica, we further consider the relationship between climate andecology in modern African ecosystems. Vegetation in Africa isshaped by complex interactions between multiple abiotic (e.g.,rainfall amount and seasonality, fire, atmospheric pCO2) andbiotic (e.g., herbivory) factors, and the relative importance of thesefactors is contingent on the ecological history of each area (52–54).Although woody cover is constrained by aridity (55), vegetationdoes not respond in a direct or continuous manner to changes inannual rainfall, and each biome (e.g., forest, savanna, grassland) isdistributed over a wide rainfall range (1,000–3,000 mm/y) (52, 56,57). We find that among modern eastern and central Africanecosystems, the proportion of C4 grazers increases with WD (R2 =0.262, P = 0.00536), and the proportion of C3 browsers decreaseswith WD (R2 = 0.1884, P = 0.021) (Fig. 5). These correlations areweak, however, and during the Pliocene-Pleistocene forests wererare in the Turkana Basin (25, 27, 49, 58) and elsewhere in easternAfrica (23). After excluding forests, we find no relationship be-tween WD and the proportional abundance of each diet guild(Fig. 5). Thus, although the abundances of C4 plants and C4grazing herbivores are often used as an indicators of aridity (21, 30),variation in C4 biomass among nonforest biomes can be decoupledfrom aridity.Paleoaridity records from δ18O of tooth enamel provide a

means to investigate links between climate and ecology in homininenvironments, but also are highly discontinuous owing to the in-completeness of the terrestrial fossil record, compounded in this

case by the scarcity of fossil assemblages meeting the criteria forapplying the aridity index. We address this problem in three ways.First, we examine the fidelity of long-term environmental recordsderived from the fossil collections used for paleoaridity analy-sis. The long-term increase in the proportion of C4 grazersamong Artiodactyla-Perissodactyla-Proboscidea over thePliocene-Pleistocene in the Omo-Turkana Basin is similar(P > 0.05, F test) when calculated using tooth enamel δ13Cvalues from paleoaridity fossil collections (R2 = 0.3868, P =0.02324) or from a larger fossil tooth enamel δ13C dataset di-vided into long time bins (>100 ka) (R2 = 0.7391, P = 0.0006911)(SI Appendix, Fig. S5). Therefore, it is possible to recover first-order environmental trends using fossils from these discontinuousdepositional intervals.Second, we compare our WD record with previously published

geological and faunal-based reconstructions of terrestrial paleo-climate in Turkana with varying time representation and analyticalbiases. There are no trends in paleoclimate based on paleosol

A

-2000 0

Climate HerbivoresVegetation

E

D

5

4

3

2

1

0

Age(Ma)

Water Deficit (mm/y)

25 50 75−10 −5 0

closed openmesic arid

C4 grazers (%)among APP taxa

Temperature (°C)

30 35 40

CB

Homo erectus

Paranthropus

early Homo

Plioceneaustralopiths

butchery?Lomekwian

Oldowan

omnivoryAcheulean

habitualbipedality

Hominin behavior and taxa

Homo sapiensMiddle stone age

δ13Cpc (‰)

2000

F

5

4

3

2

1

0

Fig. 4. Compilation of data indicating aspects of climate and ecology over the past 5 million years in the Omo-Turkana Basin. (A) Paleoaridity estimates, witherror bars indicating age uncertainty and propagated SE of mean WD estimates using all available combinations of ES and EI taxa (SI Appendix, Table S3). (B)Deep lake intervals (62). (C) Paleosol carbonate clumped-isotope temperatures (63). (D) Carbon isotope values of pedogenic carbonates (δ13Cpc) (64). There is atrend toward increasing δ13C values over time (R2 = 0.2442, P < 0.0001). (E) Percent C4 grazers among Artiodactyla-Perissodactyla-Proboscidea (APP). There is atrend toward including the proportion of C4 grazers over time (R2 = 0.7391, P < 0.001). (F) Schematic timeline showing the appearance of major homininbehaviors and taxa in eastern Africa (SI Appendix).

100

80

60

40

20

0

Gra

zer

(%)

Mix

ed fe

eder

(%

)

Bro

wse

r (%

)

excluding forestsall biomes

grasslandssavannasforests

A

B100

80

60

40

20

0

WD (mm/y)

100

80

60

40

20

0

100

80

60

40

20

0

Gra

zer

(%)

Mix

ed fe

eder

(%

)

Bro

wse

r (%

)

−2000 0 2000

0 2000

−2000 0 2000 −2000 0 2000

WD (mm/y)

WD (mm/y) WD (mm/y)WD (mm/y)

0 2000 0 2000

100

80

60

40

20

0

100

80

60

40

20

0

WD (mm/y)

Modern

Fossil

Fig. 5. WD (mm/y) and the proportion of C4 grazers, C3-C4 mixed feeders, andC3 browsers calculated using the average δ13C value of each taxon. (A) Moderncollections in eastern and central Africa. (B) Fossil collections in eastern Africa.Modern data (51, 65) and fossil data are summarized in SI Appendix, Table S5and compiled in SI Appendix, Dataset S4.

7334 | www.pnas.org/cgi/doi/10.1073/pnas.1700597114 Blumenthal et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

calcic depth, mammal hypsodonty and lophedness (k-nearest-neighbor model), or bovid tribe abundance (SI Appendix, Fig.S6). Estimated precipitation decreases over time based on mam-mal community structure (R2 = 0.2295, P = 0.03258) and mammalhypsodonty and lophedness (linear regression model) (R2 =0.0583, P = 0.004477), but these trends are weak and based onproxies influenced by evolutionary and dietary changes, re-spectively, that might not be related to aridity (SI Appendix,Fig. S6). Taken together, evidence for marked long-term aridifi-cation in the Turkana Basin is weak.Third, we examine the relationship between WD and ecology in

the fossil record irrespective of time. There are no relationships(P > 0.05) between WD and the proportion of C4 grazers, C3-C4mixed feeders, or C3 browsers (Fig. 5 and SI Appendix, Table S4).Owing to the lack of suitable fossil collections for the application ofour tooth enamel aridity proxy, we do not address climate before∼4 Ma. The Late Miocene appears to have been more humid inTurkana and elsewhere (34), although aridification before ∼4 Mapredates the long-term increase in C4 vegetation and C4 grazingmammalian herbivore that continued throughout the Pliocene andPleistocene (25, 27).Despite the coarse time resolution associated with the tooth

enamel δ18O WD calculations, we suggest that the Pliocene-Pleistocene expansion in C4 plants and C4 grazing herbivores ap-pears to not be coincident with significant long-term aridification inthe Omo-Turkana Basin (Fig. 4 and SI Appendix, Fig. S6). Thepossibility of a smaller long-term increase in aridity, undetectedowing to uncertainty in WD estimates, cannot be discounted, butwould not necessarily have been a major environmental driver,given that ecological feedback in African biomes inhibits vegeta-tion responses to climate change (52–54). Thus, the cause of themajor long-term expansion of C4 biomass within Turkana andelsewhere remains unclear, but may be related to climatic andecological dynamics that are unrelated to annual WD and need notbe equivalent across basins or regions (1, 23, 25). Our results donot preclude the possibility of climate-driven change in homininenvironments generally, but highlight the need to address possiblevariability in the determinants of environmental change in differentareas, because basins do not necessarily respond in a straightfor-ward way to continental- and regional-scale aridification. Similarly,previous paleosol and leaf wax biomarker paleovegetation studiesdemonstrate that the timing and magnitude of the expansion ofC4 plants is not uniform across eastern and northeastern Africa(23, 27, 59). Thus, climatic and ecological dynamics appear tovary across basins, and regional-scale climate proxies must becontextualized by terrestrial, basin-scale environmental recordsmost relevant to hominin evolution.Our aridity record is consistent with the notion that climate in-

stability may be an important driver of hominin evolution (2, 30).Arid conditions were prevalent during two large lake intervals∼4.0 and 2.0 Ma (Fig. 4), consistent with climate variability in-cluding periods of increased aridity occurring within generally hu-mid periods characterized by widespread lake formation (30).Orbital-scale environmental change has been demonstrated usingleaf wax biomarkers from Early Pleistocene lake sediments atOlduvai Gorge, suggesting a direct link between rainfall andchanges in the balance of woody and grassy vegetation (28). Thiscase, and other episodes of climate-driven vegetation change (29,60), are entirely consistent with ecological dynamics in cases wherewoody cover is not constrained by other factors, or when drasticchanges in precipitation, particularly during periods of heightenedclimatic variability, exceed thresholds for stable biome statesotherwise maintained by herbivory, fire, or other factors (44, 53).

Aridity and Human Evolution. Our paleoaridity record demonstratesthat hominins were able to accommodate variable environmentsthroughout the Pliocene-Pleistocene in eastern Africa, and is

consistent with the notion that biological and behavioral changes inhominins, including upright posture, hair loss, sweating, and long-distance scavenging or running, may be related to thermophysio-logical challenges associated with surviving periodically arid con-ditions and high heat loads (58). The relative abundance of C3woody vegetation during the Pliocene (Fig. 4) is consistent with thenotion that early bipedal hominins could have relied on areas withshade-providing plants that may have reduced water and heatstress. Archaeological occurrences in Turkana during the earlyPleistocene (∼2.4–1.4 Ma) are preferentially associated with lowerδ13C values of paleosol carbonate compared with those fromnonarchaeological deposits, indicating that hominins concentratedtheir activities in more wooded areas (61). In contrast, archaeo-logical occurrences at Kanjera South in southwestern Kenya(2.0 Ma) demonstrate hominins repeatedly using an open grassland(50) when aridity was low (SI Appendix, Fig. S4). Thus, earlyhominin land use patterns were likely structured by the interplaybetween aridity and vegetation, such that the exploitation of in-creasingly open C4-dominated ecosystems may have been limitedduring periods of high aridity owing to constraints on the availabilityof water and shade.

ConclusionOur findings demonstrate how mammal tooth enamel δ18O valuescan be used to quantify paleoaridity directly associated with thehominin fossil and archaeological record. WD values estimatedfrom fossil tooth enamel δ18O values suggest that early homininsexperienced highly variable climatic conditions within the range ofpresent-day environments in the region, and could accommodatearid conditions as early as ∼4.2 Ma. The modern hyperarid climatein Turkana is not a useful analog for paleoaridity in the basin. Thelack of evidence for marked, long-term aridification, along with theabsence of any relationship between aridity and herbivore dietstructure, suggest that other abiotic or biotic determinants mayhave driven long-term ecological restructuring in the Omo-Turkana Basin. The complex interplay of ecology and behaviorsuggests that disentangling the influence of climate on the evolu-tion of humans and other mammals remains a significant chal-lenge. Future interbasinal and intrabasinal studies are needed toinvestigate relationships among changing basinal geometry, bio-geography, climate, depositional setting, ecology, and evolution.

Materials and MethodsAdditional details on isotopic and statistical methods, WD calculations, andmodels of leaf water, leaf cellulose, and body water δ18O, along with an ex-panded discussion on criteria for the application of the aridity index, are pro-vided in SI Appendix. Modern meteoric water samples were compiled from theliterature (SI Appendix, Dataset S1). Modern and fossil samples of mammaliantooth enamel were analyzed for δ18O using standard methods or were com-piled from the literature (SI Appendix, Datasets S2 and S3). Information on thegeological context of fossil specimens is provided in SI Appendix, Dataset S3.

ACKNOWLEDGMENTS. This study was made possible by geological andpaleontological fieldwork in the Omo-Turkana Basin over the last 50 y. Fossilcollection was done in collaboration with Anna K. Behrensmeyer, David R.Braun, Meave G. Leakey, and the West Turkana Archaeological Project. Wethank the National Museums of Kenya, particularly Idle Farah, Emma Mbua,Fredrick Manthi, Purity Kiura, and Mary Muungu, for providing support andfacilitating access to fossil specimens, and other researchers and staff for fossilpreparation. We also thank Tom Plummer for reading a previous draft of thismanuscript, and Faysal Bibi and Scott Jasechko for sharing data. Manyorganizations have offered assistance and access to collections, including theAmerican Museum of Natural History, the Centre de Recherche en ScienceNaturelles of the Democratic Republic of Congo, the Kenya Wildlife Service,Save the Elephants, the Turkana Basin Institute, and the Uganda WildlifeAuthority. This research was funded by the Geological Society of America, theLeakey Foundation, the National Geographic Society (Grant YEG 9349-13), theNational Science Foundation (Grants 0617010, 0621542, and 1260535), SigmaXi, and the Wenner-Gren Foundation (Grant 8694).

Blumenthal et al. PNAS | July 11, 2017 | vol. 114 | no. 28 | 7335

ANTH

ROPO

LOGY

EART

H,A

TMOSP

HER

IC,

ANDPL

ANET

ARY

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0

1. Levin NE (2015) Environment and climate of early human evolution. Annu Rev EarthPlanet Sci 43:405–429.

2. Potts R (2013) Hominin evolution in settings of strong environmental variability. QuatSci Rev 73:1–13.

3. Bond WJ (2008) What limits trees in C4 grasslands and savannas? Annu Rev Ecol EvolSyst 39:641–659.

4. deMenocal PB (2004) African climate change and faunal evolution during the Plio-cene-Pleistocene. Earth Planet Sci Lett 220:3–24.

5. Shultz S, Maslin M (2013) Early human speciation, brain expansion and dispersalinfluenced by African climate pulses. PLoS One 8:e76750.

6. Dart RA (1925) Australopithecus africanus: The man-ape of South Africa. Nature 115:195–199.

7. Bramble DM, Lieberman DE (2004) Endurance running and the evolution of Homo.Nature 432:345–352.

8. Ruxton GD, Wilkinson DM (2011) Avoidance of overheating and selection for bothhair loss and bipedality in hominins. Proc Natl Acad Sci USA 108:20965–20969.

9. O’Connell JF, Hawkes K, Blurton Jones NG (1999) Grandmothering and the evolutionof Homo erectus. J Hum Evol 36:461–485.

10. Ruff CB (1993) Climatic adaptation and hominid evolution: The thermoregulatoryimperative. Evol Anthropol 2:53–60.

11. Tiedemann R, Sarnthein M, Shackleton NJ (1994) Astronomic timescale for the Plio-cene Atlantic δ18O and dust flux records of Ocean Drilling Program Site 659.Paleoceanography 9:619–638.

12. Larrasoaña JC, Roberts AP, Rohling EJ, Winklhofer M, Wehausen R (2003) Threemillion years of monsoon variability over the northern Sahara. Clim Dyn 21:689–698.

13. Rossignol-Strick M (1983) African monsoons, an immediate climate response to orbitalinsolation. Nature 304:46–49.

14. Trauth MH, Larrasoaña JC, Mudelsee M (2009) Trends, rhythms and events in Plio-Pleistocene African climate. Quat Sci Rev 28:399–411.

15. Kingston JD, Deino AL, Edgar RK, Hill A (2007) Astronomically forced climate changein the Kenyan Rift Valley 2.7-2.55 Ma: Implications for the evolution of early homininecosystems. J Hum Evol 53:487–503.

16. Dupont LM, Rommerskirchen F, Mollenhauer G, Schefuß E (2013) Miocene to Pliocenechanges in South African hydrology and vegetation in relation to the expansion of C4

plants. Earth Planet Sci Lett 375:408–417.17. Liddy HM, Feakins SJ, Tierney JE (2016) Cooling and drying in northeast Africa across

the Pliocene. Earth Planet Sci Lett 449:430–438.18. Bonnefille R (2010) Cenozoic vegetation, climate changes and hominid evolution in

tropical Africa. Global Planet Change 72:390–411.19. Zachos J, Pagani M, Sloan L, Thomas E, Billups K (2001) Trends, rhythms, and aber-

rations in global climate 65 Ma to present. Science 292:686–693.20. Cane MA, Molnar P (2001) Closing of the Indonesian seaway as a precursor to east

African aridification around 3-4 million years ago. Nature 411:157–162.21. Sepulchre P, et al. (2006) Tectonic uplift and Eastern Africa aridification. Science 313:

1419–1423.22. Edwards EJ, et al.; C4 Grasses Consortium (2010) The origins of C4 grasslands: In-

tegrating evolutionary and ecosystem science. Science 328:587–591.23. Uno KT, Polissar PJ, Jackson KE, deMenocal PB (2016) Neogene biomarker record of

vegetation change in eastern Africa. Proc Natl Acad Sci USA 113:6355–6363.24. Feakins SJ, Levin NE, Liddy HM, Sieracki A (2013) Northeast African vegetation change

over 12 my. Geology 41:295–298.25. Cerling TE, et al. (2015) Dietary changes of large herbivores in the Turkana Basin,

Kenya from 4 to 1 Ma. Proc Natl Acad Sci USA 112:11467–11472.26. Levin NE, Haile-Selassie Y, Frost SR, Saylor BZ (2015) Dietary change among hom-

inins and cercopithecids in Ethiopia during the early Pliocene. Proc Natl Acad SciUSA 112:12304–12309.

27. Levin NE, Brown FH, Behrensmeyer AK, Bobe R, Cerling TE (2011) Paleosol carbonatesfrom the Omo Group: Isotopic records of local and regional environmental change inEast Africa. Palaeogeogr Palaeoclimatol Palaeoecol 307:75–89.

28. Magill CR, Ashley GM, Freeman KH (2013) Ecosystem variability and early humanhabitats in eastern Africa. Proc Natl Acad Sci USA 110:1167–1174.

29. Rose C, Polissar PJ, Tierney JE, Filley T, deMenocal PB (2016) Changes in northeast Africanhydrology and vegetation associated with Pliocene–Pleistocene sapropel cycles. PhilosTrans R Soc Lond B Biol Sci 371:20150243.

30. Maslin MA, et al. (2014) East African climate pulses and early human evolution. QuatSci Rev 101:1–17.

31. Bibi F, Kiessling W (2015) Continuous evolutionary change in Plio-Pleistocene mam-mals of eastern Africa. Proc Natl Acad Sci USA 112:10623–10628.

32. Bobe R (2006) The evolution of arid ecosystems in eastern Africa. J Arid Environ 66:564–584.

33. Hernández Fernández M, Vrba ES (2006) Plio-Pleistocene climatic change in the TurkanaBasin (East Africa): Evidence from large mammal faunas. J Hum Evol 50:595–626.

34. Fortelius M, et al. (2016) An ecometric analysis of the fossil mammal record of theTurkana Basin. Philos Trans R Soc Lond B Biol Sci 371:20150232.

35. Peppe DJ, et al. (2011) Sensitivity of leaf size and shape to climate: Global patternsand paleoclimatic applications. New Phytol 190:724–739.

36. Levin NE, Cerling TE, Passey BH, Harris JM, Ehleringer JR (2006) A stable isotope aridityindex for terrestrial environments. Proc Natl Acad Sci USA 103:11201–11205.

37. Kohn MJ, Schoeninger MJ, Valley JW (1996) Herbivore tooth oxygen isotope com-positions: Effects of diet and physiology. Geochim Cosmochim Acta 60:3889–3896.

38. Ayliffe LK, Chivas AR (1990) Oxygen isotope composition of the bone phosphate ofAustralian kangaroos: Potential as a palaeoenviromental recorder. Geochim CosmochimActa 54:2603–2609.

39. Ayliffe LK, Lister A, Chivas AR (1992) The preservation of glacial-interglacial climaticsignatures in the oxygen isotopes of elephant skeletal phosphate. PalaeogeogrPalaeoclimatol Palaeoecol 99:179–191.

40. Luz B, Cormie AB, Schwarcz HP (1990) Oxygen isotope variations in phosphate of deerbones. Geochim Cosmochim Acta 54:1723–1728.

41. Huertas AD, Iacumin P, Stenni B, Chillión B, Longinelli A (1995) Oxygen isotope var-iations of phosphate in mammalian bone and tooth enamel. Geochim CosmochimActa 59:4299–4305.

42. Cormie A, Luz B, Schwarcz H (1994) Relationship between the hydrogen and oxygenisotopes of deer bone and their use in the estimation of relative humidity. GeochimCosmochim Acta 58:3439–3449.

43. Thornthwaite CW (1948) An approach toward a rational classification of climate.Geogr Rev 38:55–94.

44. Lehmann CER, Archibald SA, Hoffmann WA, Bond WJ (2011) Deciphering the distri-bution of the savanna biome. New Phytol 191:197–209.

45. Kohn MJ (1996) Predicting animal δ18O: Accounting for diet and physiological ad-aptation. Geochim Cosmochim Acta 60:4811–4829.

46. Levin NE, Zipser E, Cerling TE (2009) Isotopic composition of waters from Ethiopia andKenya: Insights into moisture sources for eastern Africa. J Geophys Res 114:2–13.

47. White TD, et al. (2009) Macrovertebrate paleontology and the Pliocene habitat ofArdipithecus ramidus. Science 326:87–93.

48. Leakey MG, Feibel CS, McDougall I, Walker A (1995) New four-million-year-oldhominid species from Kanapoi and Allia Bay, Kenya. Nature 376:565–571.

49. Cerling TE, et al. (2011) Woody cover and hominin environments in the past 6 millionyears. Nature 476:51–56.

50. Plummer TW, et al. (2009) Oldest evidence of tool-making hominins in a grassland-dominated ecosystem. PLoS One 4:e7199.

51. Brown FH, McDougall I, Fleagle JG (2012) Correlation of the KHS Tuff of the KibishFormation to volcanic ash layers at other sites, and the age of early Homo sapiens(Omo I and Omo II). J Hum Evol 63:577–585.

52. Staver AC, Archibald S, Levin S (2011) Tree cover in sub-Saharan Africa: Rainfall andfire constrain forest and savanna as alternative stable states. Ecology 92:1063–1072.

53. Lehmann CER, et al. (2014) Savanna vegetation-fire-climate relationships differamong continents. Science 343:548–552.

54. Oliveras I, Malhi Y (2016) Many shades of green: The dynamic tropical forest–savannahtransition zones. Philos Trans R Soc Lond B Biol Sci 371:20150308.

55. Sankaran M, et al. (2005) Determinants of woody cover in African savannas. Nature438:846–849.

56. Guan K, et al. (2014) Terrestrial hydrological controls on land surface phenology ofAfrican savannas and woodlands. J Geophys Res Biogeosci 119:1652–1669.

57. Hirota M, Holmgren M, Van Nes EH, Scheffer M (2011) Global resilience of tropicalforest and savanna to critical transitions. Science 334:232–235.

58. Passey BH, Levin NE, Cerling TE, Brown FH, Eiler JM (2010) High-temperature envi-ronments of human evolution in East Africa based on bond ordering in paleosolcarbonates. Proc Natl Acad Sci USA 107:11245–11249.

59. Lüdecke T, et al. (2016) Persistent C3 vegetation accompanied Plio-Pleistocene hom-inin evolution in the Malawi Rift (Chiwondo Beds, Malawi). J Hum Evol 90:163–175.

60. Bonnefille R, Potts R, Chalié F, Jolly D, Peyron O (2004) High-resolution vegetationand climate change associated with Pliocene Australopithecus afarensis. Proc NatlAcad Sci USA 101:12125–12129.

61. Quinn RL, et al. (2013) Pedogenic carbonate stable isotopic evidence for wooded habitatpreference of early Pleistocene tool makers in the Turkana Basin. J Hum Evol 65:65–78.

62. Lepre CJ, et al. (2011) An earlier origin for the Acheulian. Nature 477:82–85.63. Harris JM, Brown FH, Leakey MG, Walker AC, Leakey RE (1988) Pliocene and pleis-

tocene hominid-bearing sites from west of Lake Turkana, Kenya. Science 239:27–33.64. Robinson JR, Rowan J, Faith JT, Fleagle JG (2016) Paleoenvironmental change in the

late middle Pleistocene–Holocene Kibish Formation, southern Ethiopia: Evidencefrom ungulate isotopic ecology. Palaeogeogr Palaeoclimatol Palaeoecol 450:50–59.

65. Brown FH, Fuller CR (2008) Stratigraphy and tephra of the Kibish Formation, south-western Ethiopia. J Hum Evol 55:366–403.

7336 | www.pnas.org/cgi/doi/10.1073/pnas.1700597114 Blumenthal et al.

Dow

nloa

ded

by g

uest

on

Nov

embe

r 17

, 202

0