Fossils from Mille-Logya, Afar, Ethiopia, elucidate the link between Pliocene environmental changes and Homo origins
PMCID: PMC7237685
PMID: 32427848
Abstract
Several hypotheses posit a link between the origin of Homo and climatic and environmental shifts between 3 and 2.5 Ma. Here we report on new results that shed light on the interplay between tectonics, basin migration and faunal change on the one hand and the fate of Australopithecus afarensis and the evolution of Homo on the other. Fieldwork at the new Mille-Logya site in the Afar, Ethiopia, dated to between 2.914 and 2.443 Ma, provides geological evidence for the northeast migration of the Hadar Basin, extending the record of this lacustrine basin to Mille-Logya. We have identified three new fossiliferous units, suggesting in situ faunal change within this interval. While the fauna in the older unit is comparable to that at Hadar and Dikika, the younger units contain species that indicate more open conditions along with remains of Homo . This suggests that Homo either emerged from Australopithecus during this interval or dispersed into the region as part of a fauna adapted to more open habitats. Key events in human evolution are thought to have occurred between 3 and 2.5 Ma, but the fossil record of this period is sparse. Here, Alemseged et al. report a new fossil site from this period, Mille-Logya, Ethiopia, and characterize the geology, basin evolution and fauna, including specimens of Homo .
Full Text
The Mille-Logya Project (MLP) is a new paleoanthropological site, dated from ca. 2.9 to 2.4 Ma, at the northeast end of the well-known Plio-Pleistocene sites in the Awash Valley of the Afar Regional State, Ethiopia (Fig. 1). Research at Mille-Logya started in 2012 and our team conducted systematic geological and paleontological surveys in 2014, 2015, 2016, 2018, and 2019. Here, we provide the first report on the geological and paleontological content of this site.
Sedimentary exposures in the Mille-Logya area provide access to generally disconnected sections of up to ~60 m in total thickness. Between these discontinuous exposures, extensive colluvial cover of volcanic, boulder- to cobble-sized material obscures most outcrop. Furthermore, a number of post-depositional faults divide the exposures into disconnected fault-bounded blocks. Hence, our stratigraphic interpretations of relationships between sections are presently based on widespread marker beds, chemical groupings of interfingered basalts and tephras, nine new 40Ar/39Ar dates, and several magnetostratigraphic sections. These observations are sufficient to describe the overall stratigraphy, and to divide the sedimentary strata into three main fossiliferous intervals each exposed at one of the three main areas: Gafura, Seraitu, and Uraitele (Fig. 2). In this report, we designate these as informal stratigraphic units, with the aim of formalizing a regional lithostratigraphic terminology in future work, building on these presently informal terminologies.
The lowest stratigraphic unit, Gafura, (Fig. 2) begins with a sequence of thick, columnar-jointed basalt flows with intra-flow residual paleosols developed on the basalts. The Gafura sediments are poorly exposed, but occur along the southwestern flank of Iki-Ilu Ridge (Fig. 1), and are best represented by a section exposed at Sidiha Koma (section JGW15-1). The upper surface of the basalt flow at the base of this zone forms a broad low-lying surface, dissected by the Gafura River, extending into the base of the Daamé Valley. This sequence of basalts defines the GFB-I and GFB-II groups (GFB = Gafura Basalts); it underlies the main sedimentary sequence and is thus stratigraphically distinct from the overlying flows represented as the Afar Stratoid Series. Within the lowermost exposures of the Gafura Basalts, a normal to reverse magnetostratigraphic reversal is recorded (see Supplementary Table 5). Given the age constraints of overlying strata, this reversal must be equal to or older than the age of the base of the Kaena Chron (3.127 Ma).
The transition to overlying sediments of the Sidiha Koma area is marked by mudstones with ferruginized burrows interspersed with thin, poorly-sorted sands with a framework of basaltic lithic grains, occasionally containing abundant gastropods, and some bivalves. Near the top of Gafura sediments, additional basalt flows overlie the sediments locally, although these have not yet been attributed to one of the geochemically-defined groups (see Supplementary Figs. 1, 2 and Supplementary Table 1). The fossiliferous sediments of Gafura underlie a widespread diatomaceous unit, the Iki-Ilu Diatomite, which can be mapped in regionally extensive exposures along the southwestern flank of Iki-Ilu Ridge, across its southern tip, and into the floor of the Seraitu Valley, making this a practical stratigraphic boundary. The Hinti Mageta Tuff (2.914 ± 0.036 Ma; see Supplementary Figs. 3–5 and Supplementary Tables 4, 5), preserved within the Iki-Ilu Diatomite, provides an upper age limit to the Gafura sediments.
The middle stratigraphic unit is represented by the Seraitu lake beds, which often form bare, steep cliffs of largely mudstone outcrops representing lacustrine deposition. We use the base of the Iki-Ilu Diatomite as the lower boundary of this zone, although most sections cannot be mapped in continuity within a measured stratigraphic distance to the diatomite. The upper boundary of this zone may not be defined by a single widespread marker but can be locally taken as the stratigraphically lowest basalt flow with chemical composition characteristic of the UGB Group (Uraitele-Garsele Dora Group), which may consist of several different flow units. The UGB is frequently accompanied by an overlying, distinctive and widespread air-fall tuff with well-preserved glass and small lapilli-sized pumices, the Goyana Tuff (see Supplementary Table 2). Thus, the presence of one or both of these markers provides a working stratigraphic definition.
The sediments of the Seraitu lake beds are predominantly laminated clays which often contain abundant ostracods, gastropods and some bivalves, as well as plant fragments and fish remains. Diatoms in the Iki-Ilu Diatomite are somewhat recrystallized but identifiable to the genus Aulacoseira. Tephras are also numerous in the lake beds but characteristically thin (<30 cm), often air-fall occurrences, in which the primary glass is altered. Despite this, abundant feldspar crystal populations are preserved, providing two of the 40Ar/39Ar dates reported here. Besides the Hinti Mageta Tuff at the base of the Seraitu lake beds, two tuffs within the lake sediments provide precise ages: MLP14/SR-6 at 2.576 ± 0.008 Ma and MLP14/GOY-2 at 2.485 ± 0.018 Ma. In addition to the chronological information from these markers, two sections within the Seraitu lake beds zone record a magnetostratigraphic reversal which we interpret to be the Gauss/Matuyama, dated to 2.59 Ma (Fig. 1; section JGW14-10 and in section JGW14-6; see Supplementary Figs. 6–8, Supplementary Table 5).
The third unit, Uraitele, includes limited sedimentary exposures in-between extensive and thick basalt flows of the UGB, GYB-I, and GYB-II groups, which outcrop in the Goyana and Uraitele areas. The sediments include some lenticular sandstones interpreted as fluvial channels, but are predominantly laminated mudstones with occasional gastropod and bivalve bearing sandstones formed on surfaces of the UGB basalt or within the mudstones. The upper boundary of the Uraitele zone is as yet undefined, as the section continues in a thick sequence of numerous basalt flows that extend into the ridges of the Magenta Mountains at the northern extent of the area (mapped as the Afar Stratoid Series). The uppermost flows form two chemically-distinct groups, termed the GYB-I and GYB-II groups (Supplementary Table 1).
As with the Gafura area, the sediments of the Uraitele area contain a number of reworked vitric tuffs generally lacking large feldspar populations, but having distinctive chemical composition with no known correlates from the Awash Valley (see Supplementary Table 2). One of these vitric tephras, the Uraitele Tuff, has also produced populations of feldspars suitable for dating, and represents the most precise age of those presently analyzed from the Mille-Logya area: 2.443 ± 0.003 Ma.
The aforementioned stratigraphic setting provides a framework to interpret fossil data recovered from the three units. Fossil concentrations at Mille-Logya are sparse and relatively difficult to locate. Yet, after four field seasons, the fossil collection currently includes 2287 specimens, of which 1835 were collected while the rest were observed and documented on site (Table 1). Fossil collections at MLP followed a standardized protocol in order to minimize collection bias. The identifiable specimens comprise 62 Cercopithecidae, 4 Hominidae, 33 Proboscidea, 10 Camelidae, 165 Suidae, 135 Hippopotamidae, 36 Giraffidae, 944 Bovidae, 218 Equidae, 21 Rhinocerotidae, 20 Carnivora, 17 birds, and some rodents, fishes, turtles, and crocodiles. Below is a summary description of the major faunal elements.
Hominidae: Hominins were recovered from four different localities and are represented by a left and right proximal ulnae (MLP-786 & MLP-1617 respectively (Fig. 3a, b): 2.6–2.8 Ma; from two different localities, thus not from the same individual), a calvarium fragment (MLP-1469 (Fig. 3c): 2.6–2.8 Ma) and a diagnostic and complete upper second molar crown (MLP-1549 (Fig. 3d); 2.4–2.5 Ma). The molar, found in two pieces that refit cleanly, measures 14 mm and 12.5 mm buccolingually and mesiodistally, respectively, falling within the known range of A. afarensis as well as early Homo as represented by A.L. 666-1 from the younger horizons at Hadar dated at 2.33 Ma. The buccolingual and mesiodistal dimensions overlap with those of early Homo and are closest to the mean values (Fig. 4 a, b; Supplementary Fig. 10). The occlusal outline, which is dominated by the two mesial cusps, is rhomboidal with the longest axis running from the distolingual to mesiobuccal corners. The distobuccal corner is truncated. The tooth is moderately worn with no cuspal dentine exposure. The lingual wear flattens the protocone and hypocone and polishes the lingual margin leading to a rather homogenized region of the lingual half of the tooth. In contrast, the paracone and metacone are not as worn and the buccal margin is still sharp. The distal marginal ridge and distal fovea are quite perceptible, but the mesial marginal ridge is largely worn down leaving only a hint of the mesial fovea. In occlusal view, running buccal to the protocone and distal to the paracone is a large buccal groove that dominates other grooves and is positioned mesiobuccal to a lingual groove that is much smaller.
Buccolingual and mesiodistal dimensions of MLP-1549 compared to values in Au.
anamensis, Au. afarensis, Au. africanus, P. boisei, P. robustus, H. habilis, H. erectus & A.L. 666-1(Homo sp. from Hadar dated to 2.33 Ma). a upper second molar bucco-lingual dimensions in mm; b upper second molar mesio-distal dimensions in mm. Source data are provided as a Source Data file.
An asymmetric and rhomboidal occlusal outline of the upper second molar has been reported to characterize Homo erectus and H. habilis but is rare in A. afarensis. The MLP M2 possesses these features but is buccolingually broad unlike Homo erectus. Based on size and average enamel thickness (Fig. 4 a, b; Supplementary Fig. 11) in addition to diagnostic occlusal features, we attribute it to Homo sp. With an age of 2.4–2.5 Ma, this molar represents one of the oldest specimens of this genus and expands the earliest Homo sample from the Afar, which currently includes only LD 350-1 from Ledi Geraru at 2.8 Ma and A.L. 666-1 from Hadar at 2.33 Ma.
We used all MLP specimens identifiable to genus to compute the Sørenson (also known as Dice) faunal dissimilarity index for each pairwise comparison among the faunal zones. The results indicate that Seraitu and Uraitele are more compositionally similar to one another (Sørenson = 0.31) at the genus level than either of these zones is to Gafura (Gafura-Seraitu Sørenson = 0.4, Gafura-Uraitele Sørenson = 0.44). See Table 1 for faunal abundances in the three MLP zones. We further conducted a correspondence analysis on taxon abundances in order to compare the MLP faunal zones with assemblages from the Hadar Formation at Hadar and Dikika (Fig. 5). We restricted our analysis to seven bovid tribes (Aepycerotini, Alcelaphini, Antilopini, Bovini, Hippotragini, Reduncini, Tragelaphini), the suid genera Notochoerus and Kolpochoerus, and all Equidae identified only to family. These relatively broad taxonomic categories were chosen to reduce the influence of inter-observer variation in taxonomic identifications. The correspondence analysis (Fig. 5) demonstrates that the Gafura assemblage is distinct from the Seraitu and Uraitele assemblages, with Gafura showing a high abundance of Notochoerus, and the Seraitu and Uraitele assemblages showing a high abundance of Alcelaphini and Antilopini, which are open-habitat indicator taxa.
In sum, the overall similarity of the Mille-Logya fauna, especially from the older Gafura sites, with that from the Hadar Formation is remarkable. Also, in spite of the apparent younger age, it contrasts with younger sites in the Middle and Lower Awash (Table 2). For instance, the bovid assemblage of Bouri Hata at ca. 2.5 Ma includes a long list of bovid taxa for which there is no evidence at Mille-Logya: Beatragus, cf. Numidocapra, cf. Rabaticeras, Megalotragus, Hippotragus, Oryx, and Tragelaphus strepsiceros. The nearby Ledi-Geraru area contains sediments whose ages are very similar to those of Mille-Logya, but their faunal assemblage also looks different, containing Beatragus, Syncerus and Ugandax, but lacking Kobus oricornus, the most common reduncini at Mille-Logya.
Twenty-seven samples from Mile-Logya basalt samples were selected for major and trace element concentration determinations using a PANalytical 2404 X-ray fluorescence (XRF) vacuum spectrometer at Franklin and Marshall College, Lancaster, PA, USA following the techniques outlined in Mertzman, as described in detail in Supplementary Note 1. In short, the analytical technique includes the determination of ferrous iron (FeO) by standard titration methods and total volatile content (LOI). We analyzed the chemical composition of volcanic glass from 25 tephra samples from the Mille-Logya region using an electron microprobe (EMP). Major-element abundances were analyzed by wavelength-dispersive spectrometry on a JEOL 8900 Superprobe housed at the Smithsonian institution’s Department of Mineral Sciences (2014 samples) or a JEOL 8900 Superprobe housed at the Carnegie Institution for Science’s Geophysical Laboratory (2015 samples). For each tephra sample, 9–11 shards were analyzed. Both instruments were run with 12 kV, a 10 nA beam current, and a 10 µm spot size, conditions ideal for reducing alkali loss while obtaining reliable counts for elements such as Fe. We compared analyses of tephra from Mille-Logya with published analyses of tephras from throughout eastern Africa. We use the Borchardt coefficient to identify potential correlates based on glass chemistry. For chemically-similar tephras (BC ≥ 0.85), we consider the degree of similarity, stratigraphic position and radiometric ages in evaluating tephra correlations.
Following a period of several weeks of radiological ‘cooling’ after irradiation, the feldspars were analyzed individually by the 40Ar/39Ar technique using single-crystal incremental heating (SCIH). In the SCIH method, individual phenocrysts are incrementally heated in 4–7 steps (depending on grain size and gas yield) at progressively increasing power to fusion, to better examine the argon release patterns, drive off surficial argon in early steps, and maintain fairly consistent gas yields for better reproducibility. These detailed analyses were conducted on a Nu Instruments Noblesse noble-gas mass spectrometer, featuring a high-efficiency ionization source and simultaneous multi-isotope measurement using all ion-counting electron multiplier detection systems. A total of 300 SCIH steps on 91 phenocrysts from the four samples were analyzed (Table 1). 48 of these phenocrysts were rejected as candidates for complete step-heating analysis after one low-power steps, due to an obviously old xenocrystic ages or high Ca/K content, whereas the remainder (43 grains) were carried to completion. All argon measurements were performed at the Berkeley Geochronology Center. Additional details of the 40Ar/39Ar dating method as applied herein are provided in ref. and are described in greater detail in Supplementary Note 1.
Paleomagnetic samples from the Gafura and Seraitu zones were collected in 2015. Sections were trenched for measurement, description, and paleomagnetic sampling. Paleomagnetic samples were drilled using a battery-powered 2.5 cm-diameter diamond-coated bit cooled with air which was applied using a handpump. Orientation of the samples were measured using a geological compass and inclinometer – no dip correction was made for the bedding as the dip was below 5 degrees. After cutting the samples to standard size the measurement of the natural remanent magnetization (NRM) of the specimens and the progressive demagnetization was carried out in the laboratory of Paleomagnetism and Rock Magnetism at the University of Oxford (England). A pilot set of specimens were subjected to stepwise alternating-field (AF) demagnetization at applied peak fields of 0, 5, 10, 15, 20, 25, 30, 35, 40, 50, 60, 70, and 80 mT. Thermal demagnetization was done using the following temperature steps: 20, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, 575, and 600 °C. Most measurements and demagnetization steps were performed using a 2G Enterprises DC-SQUID cryogenic magnetometer with an in-line, triaxial, alternating field (AF) demagnetizer in a shielded room at the University of Oxford (United Kingdom). One batch of samples were thermally demagnetized following temperature steps 20, 100, 150, 200, 250, 300, 350, 400, 450, 500, 550, and 600 °C at Fort Hoofddijk Paleomagnetic Laboratory of the Utrecht University (The Netherlands) on a 2G Enterprises DC SQUID cryogenic magnetometer. Thermal demagnetization was performed on an ASC thermal demagnetizer (residual field <20 nT). Natural remanent magnetization (NRM) intensities were typically several orders of magnitude higher than the instrument sensitivity (~3 × 10−12 Am2). The results of the demagnetization were interpreted to identify the Characteristic Remanent Magnetization (ChRM) directions using Paleomagnetism.org, an online open source tool for paleomagnetic data analysis. ChRM directions were calculated with a minimum of four consecutive steps (See Supplementary Table 5). Paleomagnetism.org uses a set of techniques to statistically interpret the results.
Further information on research design is available in the Nature Research Reporting Summary linked to this article.