Human evolution likely encouraged by a dry climate

human evolution

Editor's Introduction

Late Pliocene fossiliferous sedimentary record and the environmental context of early Homo from Afar, Ethiopia

annotated by

The earliest known fossil of the genus Homo was found in northern Ethiopia. Scientists performed a detailed study of the surrounding rocks and fossils by collecting tuffs, or hardened volcanic ash layers, and analyzing them to determine the age of the fossil (fossils this old cannot be dated themselves). By collecting a plethora of other animal fossils and synthesizing local geology, they found evidence that ancient landscapes in this part of Ethiopia were becoming drier than they were before, suggesting the evolution of hominins was coincident with, and likely driven by, environmental change. 

Paper Details

Original title
Late Pliocene fossiliferous sedimentary record and the environmental context of early Homo from Afar, Ethiopia
Original publication date
Vol. 347, Issue 6228, pp. 1355-1359
Issue name


Sedimentary basins in eastern Africa preserve a record of continental rifting and contain important fossil assemblages for interpreting hominin evolution. However, the record of hominin evolution between 3 and 2.5 million years ago (Ma) is poorly documented in surface outcrops, particularly in Afar, Ethiopia. Here we present the discovery of a 2.84–to 2.58–million-year-old fossil and hominin-bearing sediments in the Ledi-Geraru research area of Afar, Ethiopia, that have produced the earliest record of the genus Homo. Vertebrate fossils record a faunal turnover indicative of more open and probably arid habitats than those reconstructed earlier in this region, which is in broad agreement with hypotheses addressing the role of environmental forcing in hominin evolution at this time. Geological analyses constrain depositional and structural models of Afar and date the LD 350-1 Homo mandible to 2.80 to 2.75 Ma.


Surface exposures of fossiliferous sedimentary rocks dated between 3.0 and 2.5 million years ago (Ma) are rare throughout Africa, yet are of great interest because this interval overlaps with shifts in African climate (1–5), corresponds to faunal turnover (68), and represents an important gap in our knowledge of evolutionary events in the human lineage (9). The time period coincides with changing geologic conditions in eastern Africa, as rifting processes (10, 11) and extensive volcanism (12) altered the architecture of sedimentary basins (1315), controlling the paleogeography of hominin and other mammalian habitats. In tectonically active areas such as the lower Awash Valley (LAV), Afar, Ethiopia, rift-basin dynamics create spatially variable and often incomplete records of deposition. At other fossil sites in the LAV, the fluvio-lacustrine sediments of the Hadar Formation (~3.8 to 2.9 Ma) are separated from the younger fluvial sediments of the Busidima Formation (~2.7 to 0.16 Ma) by an erosional unconformity (1416). The Hadar region contains early Homo dated to ~2.35 Ma (9) and an excellent record of Australopithecus afarensis from 3.5 to 2.95 Ma (17). However, the absence of fossiliferous sediments in the Hadar region due to the unconformity has impeded efforts to document a continuous record of hominin and other faunal evolution, and limits our understanding of regional habitat change in the LAV. Recent field investigations and geochronological analysis of sedimentary deposits at Ledi-Geraru (LG), located northeast of Hadar, Gona, and Dikika (Fig. 1), confirm the presence of late Pliocene fossiliferous sedimentary rocks dated to the interval represented elsewhere in the region by the erosional unconformity (18). Here we present the geology, chronostratigraphy, and paleontology of the Lee Adoyta region of LG, where the LD 350-1 early Homo mandible (19) and 614 other mammal specimens were recovered from sediments dated 2.84 to 2.58 Ma (Fig. 1).

Fig. 1. Geographic and geologic setting of the Lee Adoyta hominin site. (A) The LAV (yellow square), Afar Depression (gray area), Ethiopia. RS, Red Sea; GOA, Gulf of Aden; MER, Main Ethiopian Rift. (B) LAV project areas and the approximate mapped extent of the Hadar Formation senso stricto. The Busidima Formation is largely exposed in the areas of Hadar, Gona, and Dikika. (C) Sediments and volcanic rocks in the eastern LG research project area are cut by two sets of faults striking NW and NNE, indicating the influence of both the RS and MER extensional systems, respectively. At Lee Adoyta, NW-trending faults are most significant and appear to cross-cut NNE faults. Regions referred to in the text are labeled in black.
Plate tectonics

The dotted lines in Fig. 1 section (A) mark boundaries between tectonic plates. The Red Sea (to the north) and the Gulf of Aden (to the northeast) both exist because the Arabian Plate is moving away from the African Plate.

There is also a plate boundary in the middle of Ethiopia. Is this boundary compressive (pushing together) or extensional (pulling apart)?

Geologic formations

In Fig. 1 section (B) the approximate extent on the surface of the Hadar Formation is indicated by the dotted area. A geologic formation is a group of rocks with identifiable characteristics that can be consistently observed over a fairly large area. Typically, geologic formations represent specific time intervals. They can be used to create geologic maps, as shown in Fig. 2 section (A).

The Lee Adoyta region preserves an ~70–m-thick sedimentary sequence that is cut by multiple closely spaced NW-SE (320° to 340°)–trending faults that postdate deposition (Figs. 1 and 2). Geologic mapping documents drag folds and stratigraphic juxtaposition to define the normal sense of motion along the faults, which is consistent with faulting patterns oriented NW-SE associated with Red Sea rift extension (14). There are four major fault-bounded blocks, each of which comprises a discrete sedimentary package (Fig. 2).

Fig. 2. Geologic map and cross section of the Lee Adoyta hominin site. (A) Geologic map of the region surrounding the Lee Adoyta hominin site (yellow star) showing NW-SE–oriented faults dissecting sedimentary packages into discrete blocks. We mapped the 900 × 500–m area in the field using high-resolution (1 m) stereo imagery and Global Positioning System technology. (B) West-to-east cross section of Lee Adoyta. The older Bulinan and Gurumaha fault blocks are uplifted relative to the younger adjacent Lee Adoyta and Garsalu fault blocks.
Geologic mapping

Fig. 2 section (A) is a classic geologic map, which uses different colors to indicate the different rock units exposed at Earth’s surface. In this case the rock units are bounded by fault lines caused by extensional tectonic stresses. These faults are 3D cracks in the Earth whose surface expression is represented in 2D on the map.

Geologic maps are created by many national and state agencies and cover most areas where rocks are exposed. The United States Geologic Survey hosts a great database here.

Geologic cross section

Fig. 2 section (B) gives an idealized view of a single vertical slice of Earth below the surface along the line from A to A’ shown in section (A). The colors correspond to the same rock units shown in section (A). The faults that meet the surface can now be seen to be dipping (that is, not vertical). These faults are normal faults, caused by an extension of Earth’s crust.

This short video showing a 3D map and cross section of an area in Montana might help you to visualize the idea of a geologic cross section.

The Bulinan sedimentary package is 10 m thick and consists of lacustrine deposits (laminated silty claystone with dispersed gastropod shells) with five intercalated 2- to 12-cm-thick altered tuffs (Fig. 3). The crystal-rich Bulinan Tuff lies 4 m above the base of the section. It is 2 to 3 cm thick, light pink in color, and composed of altered volcanic glass with <15% subangular lithic fragments and feldspar grains. The Bulinan Tuff was dated by the laser single-crystal incremental heating (SCIH) 40Ar/39Ar technique on individual grains of phenocrystic Na-plagioclase feldspar from a single sample. Age “plateaus” as revealed in 39Ar release spectra indicate the characteristic age of the feldspars. The population of those ages yielded a weighted-mean result of 2.842 ± 0.010 Ma (1σ internal error; ± 0.014 Ma external error; n = 4 grains) (figs. S2 to S4 and table S2). Just four fossils have been recovered from this fault block, and therefore an inference of habitat based on fauna is precluded.

Fig. 3. Stratigraphy and magnetostratigraphy of the sedimentary packages at Lee Adoyta. The dip and sense of motion along each fault bounding the sediment packages are shown to separate the packages and thus establish relative ages. Section locations (numbered) are provided in Fig. 2.
Stratigraphic columns

As a rule, geologic units are deposited in horizontal layers, with younger rocks deposited on top of older ones. Studying a stack of rocks allows scientists to “read” the history of Earth through time to understand how it has changed. A stratigraphic column, like the four shown here, is simply a way of graphically representing the layered rock units found in an area.

All four of these columns are from northern Ethiopia. Are they all the same? What are some of the differences between them? What might that mean about differences in the geologic history of each place?

Sedimentary rocks

With the exception of a tuff, the rock types in this column (conglomerate, sandstone, siltstone, and claystone) are all sedimentary. This means they are made up of bits of older rocks that were eroded at the surface of Earth and moved away and deposited by wind or water, for example. A great deal of information about past environments can be learned from changes in rock type.

A claystone, for example, might mean that the area was once the bottom of a lake or sea, or possible a river floodplain. By carefully studying the claystone and the small fossils it might contain (like fish bones) scientists can determine the kind of environment the clay was deposited in.

A good overview (with pictures) of most of these rock types is available here.

Tuff stuff

Tuff is an igneous rock formed from volcanic material (pumice, rocks, minerals) ejected during explosive volcanic eruptions. Because this region has many volcanoes, one would expect tuff layers to be present in the rock layers. Tuffs are critical for the absolute dating of geologic units because they contain feldspar minerals needed for radiogenic dating techniques.

For pictures of tuff and more information, check here.


The study of ancient paleomagnetic orientations preserved in rocks is often referred to as “paleomag.” Over geologic time, Earth’s magnetic field has switched directions many times. That is, at many intervals during the past, a compass would have pointed toward the South Pole rather than the North Pole (referred to as reversed polarity)

When some rocks form, certain minerals that are magnetic (such as magnetite) will naturally orient themselves with the current magnetic field. Such rocks preserve a record of what the magnetic field was when they formed, either normal or reversed. Normal polarity is what we are in today, where the arrow on your compass points to the North Pole. When Earth experiences reversed polarity, the north arrow on your compass would point south. The ability to measure these ancient magnetic fields in rocks has a variety of applications and even played an integral role in the discovery of plate tectonics. Efforts by many other geologists to date the timing of switches in magnetic field direction (from reversed-normal or vice versa) now allows for age constraints using these magnetic measurements of rocks. In this study, the authors measured the paleomagnetic orientations of their rocks as a way of dating them.

You can watch a geologist explain how to collect a real paleomagnetism core sample here.

The Gurumaha sedimentary package, which yielded the LD 350-1 hominin, is ~21 m thick and dips 3° to 5° E-SE. Gurumaha sediments coarsen up-section and include laminated mudstone with thin, fine sandstones, siltstone, and coarse cross-bedded sandstone with pebble lags. The package is capped by a fluvial sequence composed of a carbonate nodule–rich, cross-bedded pebble conglomerate and overlying sands with minimal basal scour. The Gurumaha Tuff is a crystal-rich lapilli tephra-fall deposit that contains pumice (table S5) and forms a continuous white to light gray stratigraphic marker 8 to 10 cm thick (Fig. 2). The Gurumaha Tuff was dated by the SCIH technique applied to four samples containing anorthoclase to Na-plagioclase feldspar phenocrysts. A weighted mean of the plateau ages yielded 2.822 ± 0.006 Ma (± 0.015 Ma external error; n = 23 grains) (figs. S2 to S4 and table S2). This age falls within chron C2An.1n (Gauss) of the astronomical polarity time scale (20), consistent with the normal paleomagnetic polarity measured for the entire Gurumaha sequence (Fig. 3). The age on the tuff provides a maximum age for the LD 350-1 hominin fossil, which was recovered from a vertebrate fossil–rich silt horizon 10 m conformably above the Gurumaha Tuff and 1 m below the base of the capping pebble conglomerate (19). Based on local and regional sedimentation rates, a refined age estimate of 2.80 to 2.75 million years is calculated for the fossiliferous horizon (19).

Ecological community structure analysis based on mammalian fauna recovered from the Gurumaha fault block indicates a more open habitat (mostly mixed grasslands/shrublands with gallery forest) that probably experienced less rainfall than any of those reconstructed for the members of the Hadar Formation (6). The landscape was similar to modern African open habitats, such as the Serengeti Plains, Kalahari, and other African open grasslands, given the abundance of grazing species and lack of arboreal taxa, although the presence of Deinotherium bozasi and tragelphins probably indicates a gallery forest (fig. S6). The existence of Kobus sigmoidalis, aff. Hippopotamus afarensis, crocodiles, and fish in this package reflects the presence of rivers and/or lakes. Approximately one-third of the mammalian taxa present are shared with those in the youngest Hadar Formation (~3 Ma), whereas one-third are first appearances of these taxa in the LAV (Table 1). The remaining one-third of mammals recovered can only be identified to the genus level.

Table 1. Preliminary faunal lists for the Gurumaha (GU), Lee Adoyta (LA), and Garsalu (GA) sedimentary packages from the Lee Adoyta Drainage. Key: X, present; 0, absent; *, taxa shared with the Kada Hadar submember 2; +, species previously unrecorded in the LAV, including chronospecies.

The Lee Adoyta sedimentary package is ~22 m thick and approximately horizontal. Two tuffs separated stratigraphically by ~8 to 10 cm approximate the base of the Lee Adoyta package. The lower tuff is a 5- to 6-cm-thick basaltic ash typically altered to a yellowish bentonite. The upper unit is a 4- to 5-cm-thick light gray vitric-crystal tuff (table S6). The Lee Adoyta Tuffs are encased in brown fissile mudstone that directly overlies a green Vertisol. The overlying sedimentary units include brown mudstone, basalt-rich sandstone, and a pebble conglomerate. A 1.5-m-thick, cross-laminated, unnamed glassy tuff caps the section (table S6). Na-plagioclase phenocrysts from the upper Lee Adoyta Tuff have a weighted-mean SCIH age of 2.669 ± 0.011 Ma (± 0.03 Ma external; n = 5 grains) (figs. S2 to S4 and table S2). Paleomagnetic measurements record a transition from normal to reverse polarity ~12 m above the Lee Adoyta Tuffs (Fig. 3), which is probably the Gauss/Matuyama reversal at 2.581 Ma (20). The date and stratigraphic position of the Lee Adoyta Tuffs are consistent with its assignment to the C2An1.n chron (Gauss), and provide a minimum age constraint for the LD 350-1 fossil. The Lee Adoyta fault block yielded mammalian fauna with ~80% taxonomic overlap with the Gurumaha fauna, and the ecological community structure also reconstructs an open habitat (fig. S6 and Table 1).

The Garsalu sedimentary package encompasses strata exposed along the margins of the Lee Adoyta drainage (Fig. 2). We correlate these packages based on the similarity of sedimentary facies and downfaulting against adjacent fault blocks. These fluvial deposits are ~26 m thick, gently dipping, and include paleosols, sandstones, siltstones, and conglomerates (Fig. 3). The Garsalu sedimentary package is the youngest (<2.58 Ma) in the Lee Adoyta region, based on faulting relationships and the presence of Connochaetes gentryi.

The combined 70-m-thick section at Lee Adoyta is placed within the chronostratigraphic framework of the LAV in an interval previously undocumented in the regional sedimentary record (fig. S8). In general, the sedimentary deposits at Lee Adoyta coarsen upward and represent a variety of depositional environments. At Lee Adoyta, a paleolake (~2.84 Ma) extended at least 6 km north to Ambare and Mafala (Fig. 1C), where lacustrine deposits are present in similarly aged strata (18). The depositional environment progressed to a nearshore delta plain by 2.82 Ma, as indicated by sandy channel bodies and the presence of crocodiles, fish, and mammals in the Gurumaha sediments. At present, sedimentary deposits between the top of the Gurumaha sedimentary package (2.80 to 2.75 Ma) and the Lee Adoyta Tuffs (2.67 Ma) have not been observed. The Lee Adoyta fault block strata (<2.67 Ma) are coeval with a portion of the Busidima Formation and similarly capture a fluvial record probably deposited by tributaries to the ancestral Awash River system (16).

Geological investigations at Lee Adoyta allow us to place constraints on regional basin models. The presence of deposits dated to 2.8 Ma in eastern LG is consistent with continued deposition in the Hadar Basin as a result of northeastern migration of paleo Lake Hadar during the late Pliocene to early Pleistocene (142123). Sometime between 2.95 and 2.7 Ma, changes in base level associated with Main Ethiopian Rift extension eroded Hadar Basin sedimentary deposits in the areas of Gona, Hadar, Dikika, and central and southern LG, creating an erosional unconformity (Fig. 1B and fig. S8) (141518,24). The preservation of 2.8-Ma sediments in eastern LG, but not elsewhere in the lower Awash, suggests that the unconformity did not extend as far east as Lee Adoyta (or at least was not as long in duration). This may be related to the proximity of Lee Adoyta to border faults, spatial variability of base level changes, or localized downfaulting of eastern LG before erosion. Lee Adoyta lies beyond the proposed eastern margin of the Busidima half-graben (1415). Therefore, the <2.7-Ma deposits at Lee Adoyta were either deposited in a different basin, or the Busidima half-graben was larger and more variable than proposed. After ~2.6 Ma, NW-SE trending faults that cross-cut all sedimentary packages (Fig. 1C) indicate that Red Sea rifting was the dominant extensional regime.

Global climate change at ~2.8 Ma and resultant increases in African climatic variability and aridity are hypothesized to have spurred cladogenetic events in various mammalian lineages, including hominins (127). The faunal changes evident at Lee Adoyta appear to be in accord with these hypotheses, because the 2.8-Ma record shows a mammalian species turnover that includes first appearance datums and the dispersal of immigrant taxa previously unknown in Afar. Additionally, mammal communities in the Gurumaha and Lee Adoyta sedimentary packages indicate open habitats, with most vegetation cover consisting of grasses or low shrubs, a pattern that contrasts with the older, Australopithecus afarensisbearing, Hadar Formation. Although the Lee Adoyta data provide enticing evidence for a correlation between open habitats linked to African aridification and the origins of the genus Homo, evidence from other sites in eastern Africa shortly after 3 Ma does not show a uniform transition toward open habitats (82527). Ongoing research efforts in the eastern LG continue to explore previously undocumented sedimentary exposures that may allow us to test the hypothesis that the Lee Adoyta record samples a drier habitat of a larger, more variable ecosystem or represents a distinct arid phase in Afar during the late Pliocene.

Supplementary Materials

Materials and Methods
Figs. S1 to S8
Tables S1 to S6
References (2836)

References and Notes

  1. P. B. deMenocal, Earth Planet. Sci. Lett. 220, 3–24 (2004).

  2. P. B. deMenocal, Science 270, 53–59 (1995).

  3. J. G. Wynn, Am. J. Phys. Anthropol. 123, 106–118 (2004).

  4. T. E. Cerling, Palaeogeogr. Palaeoclimatol. Palaeoecol. 97, 241–247 (1992).

  5. A. C. Ravelo, D. H. Andreasen, M. Lyle, A. Olivarez Lyle, M. W. Wara, Nature 429, 263–267 (2004).

  6. K. E. Reed, J. Hum. Evol. 54, 743–768 (2008).

  7. E. S. Vrba, in Paleoclimate and Evolution, with Emphasis on Human Origins, E. S. Vrba, G. H. Denton, T. C. Partridge, L. H. Burckle, Eds. (Yale Univ. Press, New Haven, CT, 1995), pp. 385–424.

  8. R. Bobe, A. K. Behrensmeyer, Palaeogeogr. Palaeoclimatol. Palaeoecol. 207, 399–420 (2004).

  9. W. H. Kimbel et al., J. Hum. Evol. 31, 549–561 (1996).

  10. E. Wolfenden, C. Ebinger, G. Yirgu, A. Deino, D. Ayalew, Earth Planet. Sci. Lett. 224, 213–228 (2004).

  11. T. Kidane et al., J. Geophys. Res. 108, 2102 (2003).

  12. P. Lahitte, P.-Y. Gillot, T. Kidane, V. Courtillot, A. Bekele, J. Geophys. Res. 108 (B2), 2123 (2003).

  13. N. J. Hayward, C. J. Ebinger, Tectonics 15, 244–257 (1996).

  14. J. G. Wynn et al., Spec. Pap. Geol. Soc. Am. 446, 87–118 (2008).

  15. J. Quade et al., Spec. Pap. Geol. Soc. Am. 446, 1–31 (2008).

  16. J. Quade et al., Geol. Soc. Am. Bull. 116, 1529–1544 (2004).

  17. W. H. Kimbel, L. K. Delezene, Yearb. Phys. Anthropol. 140, 2–48 (2009).

  18. E. DiMaggio, thesis, Arizona State University, Tempe, AZ (2013).

  19. B. Villmoare et al., Science 347, 1352–1355 (2015).

  20. F. M. Gradstein, J. G. Ogg, M. D. Schmitz, G. M. Ogg, The Geologic Time Scale 2012 (Elsevier, Oxford, 2012).

  21. J. Aronson, M. Taieb, in Hominid Sites: Their Geologic Settings, G. J. Rapp, C. F. Vondra, Eds. (Westview Press, Boulder, CO, 1981), pp. 165–196.

  22. J. E. Kalb, Palaeogeogr. Palaeoclimatol. Palaeoecol. 114, 357–368 (1995).

  23. C. Campisano, C. Feibel, Spec. Pap. Geol. Soc. Am. 446, 179–201 (2008).

  24. C. J. Campisano, J. Hum. Evol. 62, 338–352 (2012).

  25. F. Bibi, A. Souron, H. Bocherens, K. Uno, J.-R. Boisserie, Biol. Lett. 9, 20120890 (2013).

  26. J. Kingston, in Late Cenozoic Environments and Hominid Evolution: A Tribute to Bill Bishop, P. Andrews, P. Banham, Eds.(Geological Society, London, 1999), pp. 69–84.

  27. N. E. Levin, F. H. Brown, A. K. Behrensmeyer, R. Bobe, T. E. Cerling, Palaeogeogr. Palaeoclimatol. Palaeoecol. 307, 75–89 (2011).