Searching for the emergence of stone tool making in eastern Africa

Modern humans rely entirely on technology for their subsistence, and the interaction between technological and biological adaptations has played a key role in the evolution of our lineage. When exactly technology emerged in the fossil record and thus started to shape primate evolution has appealed to researchers since the beginnings of paleoanthropology as a discipline and, as Braun et al. (1) show in PNAS, is still today a source of important discoveries.

Our closest relatives at present, the chimpanzees, have a rich material culture that varies geographically (2) and comprises the use of a wide variety of tools, including leaves, clubs, sticks, stones, and others. Although our ancestors most likely also used an array of organic materials, stones are usually the only artifacts preserved in the record and, therefore, are key to trace the emergence of technology in human evolution. For nearly a century, the Oldowan defined by Louis Leakey (3) at Olduvai Gorge in Tanzania has remained the most solid evidence of earliest stone tool making in our lineage. Originally thought to be around half a million years old, the age of the Oldowan has been pushed back steadily (4) since the 1960s, with the Gona sites in Ethiopia at 2.58 million years ago (Ma) being, until Braun et al.’s (1) discovery, the oldest evidence of this technology (5). The new site of Bokol Dora 1 (BD 1) at the paleoanthropological area of Ledi-Geraru, also in Ethiopia, is slightly older than Gona at 2.61 to 2.58 Ma (1) and contributes to establishing ∼2.6 Ma as a firm date for the earliest Oldowan.

But the Oldowan might have not been the first archaeologically visible technology in the paleoanthropological record. In recent years, two discoveries—Dikika and Lomekwi—have sparked fresh debate about the potential existence of stone tool technologies further back in time, at ∼3.4 Ma. At Dikika, Ethiopia, some bones bear traces interpreted as cut marks (6), which would evidence that Pliocene hominins were using stone tools to process mammal carcasses. Although no stone tools have yet been reported in Dikika, the site of Lomekwi 3 (Kenya) has yielded several lithic artifacts that are attributed to 3.3-Ma sediments (7), thus complementing the indirect evidence of stone tool use proposed for Dikika.

While the evidence >3 Ma is not exempt of controversy (see recent reviews in refs. 8 and 9), Braun et al. (1) compare the lithic assemblage of BD 1 with that of Lomekwi and conclude that differences are significant enough to maintain a separation between the Lomekwian culture—the name given to the new discovery in Kenya (7)—and the Oldowan, for which BD 1 is now the earliest site. Acceptance of this scenario would imply reconsidering some long-standing paradigms in paleoanthropology.

For instance, an influential view established coalescence among the emergence of the genus Homo, the earliest stone tools, and the climatic change toward aridification at 2.9 to 2.6 Ma (10). The earliest fossils of Homo have been recently discovered in the same area of Ledi-Geraru as the Oldowan site reported by Braun et al. (1) and, although such human remains are 200,000 y older (11) than BD 1, given the fragmentary character of the paleoanthropological record, they do not necessarily challenge the scenario associating climate change with speciation and cultural responses. Accepting the validity of archaeological evidence at ∼3.4 Ma, however, changes substantially the scenario and decouples such association, as no significant climate changes have been detected in that particular time span, and hominins in both the Dikika (Australopithecus afarensis) and Lomekwi (Kenyanthropus platyops) areas are less derived than early Homo.

Another intriguing implication is the ∼0.7-Ma gulf between Lomekwi and BD 1. Basic lithic technologies such as the Lomekwian and the Oldowan may have been discovered and then disappeared and reinvented again in the course of human evolution (12)—in a fragmented sequence of cultural gains and losses that could potentially explain why no archaeological evidence is found between 3.3 and 2.6 Ma. In my view, however, a scenario of independent inventions, disappearance, and rediscovery of stone tool flaking does not fit well with the evidence available for the Oldowan record from 2.6 Ma onward, because the tight chronological clustering among Ledi-Geraru (1), Gona (5), and all of the eastern African sites that follow in the temporal sequence (Fig. 1) indicates continuity in the archaeological record and consistency in the patterns involved.

Fig. 1.

Earliest archaeological sites in eastern Africa (Upper) and techniques used to produce sharp-edged stone flakes (Lower). Block-on-block technique: the core is struck against a stationary anvil. Bipolar technique: a core resting over a stationary anvil is struck with a hammerstone. Freehand technique: a handheld core is knocked off with a hammerstone.

And which are those patterns? All early archaeological assemblages contain clusters of stone tools in lake margins or riverine settings, which often correspond to more open environments than those inhabited by earlier hominins and the majority of nonhuman living primates. Most early archaeological assemblages also yield fossils, although in many, like the one reported by Braun et al. (1), their association with lithic artifacts is only spatial and not necessarily contextual (e.g., in the form of anthropogenic marks on fossils demonstrating an effective link between stone tools and bones). In some other assemblages, lithics are abundant while fossils are not present (Fig. 1), but their absence may be due to taphonomic bias against bone preservation. In some rare—and controversial (8)—instances, reported cut marks in bones without a clear association with lithic assemblages (e.g., Dikika and Bouri) serve to emphasize the main purpose of early stone tools, which is deemed to be the production of cutting edges (13), most likely to aid early humans accessing animal carcasses.

In addition to producing cutting edges, stones were most likely used in other tasks; percussive activities involved in processing organic materials are common among various primates (14) and may potentially have preceded stone tool knapping during the initial stages of hominins’ exploitation of animal resources (9). However, most early archaeological sites from Fig. 1 contain few battered artifacts other than those related to flake production, with such production being the primary goal in all assemblages. As Braun et al. (1) emphasize, nonetheless, technological variability associated with flake production may have been considerable. When the Gona assemblages were thought to be the earliest evidence of stone tool knapping, researchers were puzzled about the high quality of flaking production, as parsimony dictated that there would be an incremental process of skill acquisition through the Oldowan. Braun et al. (1) indicate that the slightly older assemblage of BD 1 is also slightly less technologically sophisticated than Gona, which makes it tempting to return to a gradualist view of technological evolution during the Oldowan. As shown in Fig. 1, 3 main techniques were employed to produce flakes in the earliest archaeological sites. The block-on-block technique has only been reported in Lomekwi, involves the production of much larger flakes than the ∼2- to 5-cm-long products typical of the Oldowan, and is one of the features put forward to differentiate the Lomekwian from the Oldowan (7). However, bipolar and freehand techniques are found across the early Oldowan record (Fig. 1), and the quality of flaking products may be strongly influenced by raw-material constrains (15) and site particularities (16). Thus, for now, it might be safest to highlight that there is indeed technological variability among the earliest Oldowan sites, without attributing time trends to such variability.

From that perspective, the overarching features shared by most of the earliest sites in Fig. 1 would be the clustering of artifacts (and often also bones) obtained through conchoidal fracture. Mastering of the conchoidal fracture has long stood as a defining character of human technology (17), and artifact clustering has been associated with behavioral changes by early hominins that enable archaeological visibility of their activities (18).

But are these features truly unique to hominins and their earliest archaeological sites? Findings in the brand-new discipline of primate archaeology (19) seem to say otherwise; for example, chimpanzee nut-cracking activities are now observed to produce significant concentrations of artifacts, in many ways mimicking spatial patterns typical of the Stone Age record (20). Even more surprisingly, capuchin monkeys have recently been shown to obtain conchoidally fractured flakes through recurrent use of hammerstones (21). Even though capuchins produce such lithic assemblages unintentionally, the technological behavior of these distant relatives in our evolutionary lineage indicates that large brains and humanlike hands are not, after all, indispensable requirements for stone knapping.

Primate archaeology findings are blurring the line of what is exclusive to early human technology and opening new perspectives into the comparative analysis of archaeological assemblages made by hominins and nonhuman primates. It now seems clear that lithic technology has appeared independently in various lineages of primates and may potentially have been discovered and disappeared several times through the Pliocene and Early Pleistocene; by pushing the emergence of the Oldowan a bit further back in time, Ledi-Geraru (1) contributes both to exploring the roots of lithic technology and to characterizing the internal variability of early stone tool assemblages. It is hardly an exaggeration to state that today, the field remains as vibrant as when Leakey (3) defined the Oldowan nearly a century ago.