Oued Beht, Morocco: a complex early farming society in north-west Africa and its implications for western Mediterranean interaction during later prehistory

Abstract

The Maghreb (north-west Africa) played an important role during the Palaeolithic and later in connecting the western Mediterranean from the Phoenician to Islamic periods. Yet, knowledge of its later prehistory is limited, particularly between c. 4000 and 1000 BC. Here, the authors present the first results of investigations at Oued Beht, Morocco, revealing a hitherto unknown farming society dated to c. 3400–2900 BC. This is currently the earliest and largest agricultural complex in Africa beyond the Nile corridor. Pottery and lithics, together with numerous pits, point to a community that brings the Maghreb into dialogue with contemporaneous wider western Mediterranean developments.

Introduction

The Maghreb combines an immense Saharan interface with a Mediterranean environmental zone as large as Iberia's and the shortest maritime crossings between Africa and southern Europe. The region's credentials as a location for major cultural developments and inter-continental connections in the past are therefore exceptional. Research has shown that this potential was realised during the Palaeolithic (Hublin et al. 2017; Barton et al. 2020; Sehasseh et al. 2021) and from the Phoenician Iron Age to Islamic times (Papi 2014; Mederos Martín 2019; Bokbot 2020; Fenwick 2020). Yet the intervening nine millennia of Holocene prehistory remain far less explored—in particular the period c. 4000–1000 BC. For the earlier Holocene, a spectrum of hunting and foraging societies (exemplified by the Capsian of the eastern Maghreb) have been identified, followed in the sixth to fifth millennia BC by expansions of pastoralism from the east and elements of a Mediterranean agro-pastoral Neolithic in the north-west, closely linked to developments across the Gibraltar strait (see Broodbank & Lucarini 2019; Carríon Marco et al. 2022). After this, however, there is a lacuna, both in terms of reliable data and interpretative models (Lucarini et al. 2021). It is implausible that this reflects a genuine lack of activity, particularly given the contemporaneous emergence in north-east Africa of pharaonic Egypt, as well as the dynamic, interconnected Copper and Bronze Age societies of the Levant, Aegean, Central Mediterranean and Iberia (Broodbank 2013: 257–444). Critically, this gap in the evidence also leaves the role of north-west Africa in the emergence of Mediterranean and wider African long-term history untested and potentially radically underestimated.

In this article, we present results of new fieldwork in Morocco, demonstrating that one part of the Maghreb does provide remarkable evidence for farming, aggregation, long-range connectivity and, probably, storage within this timeframe, attributes that collectively suggest complex social relationships. The north-west Maghreb's position between the Atlantic and Mediterranean weather systems, together with its proximity to the Atlas mountain range, ensures adequate rainfall and significantly ameliorated the aridification observed over much of northern Africa from the fourth millennium BC onwards (Roberts et al. 2019), enabling it to remain fully viable for agriculture. Here, moreover, among other isolated, poorly dated finds, a long-known distribution of mid- to late-third-millennium BC pottery and metalwork of south-west European Beaker types, together with numerous finds of African ivory and ostrich eggshell in Copper Age Iberia, manifestly demand wider investigation (Bokbot 2005; Schuhmacher et al. 2009; Schuhmacher 2016). Recently, this scant archaeological evidence has been complemented by genetic analyses revealing a mixed population ancestry of local hunter-gatherers, Neolithic Iberian farmers and Saharan pastoralists (Fregel et al. 2018; Simões et al. 2023), while also identifying an individual of African descent buried at the Late Copper Age (Beaker period) site of Camino de las Yeseras in the Iberian interior (Olalde et al. 2019). So far missing, however, has been any substantive understanding of the Maghrebian societies involved. To date, most information derives from caves and cemeteries with problematic dating; prioritising open living sites has the potential to transform this situation.

Oued Beht (Khémisset province, Morocco)

The site of Oued Beht occupies a limestone, marl and conglomerate ridge 170–210m above sea level on the rolling Zemmour plateau, overlooking the eponymous perennial river (Figure 1). It lies in a Mediterranean-type, semi-arid bioclimatic zone (Hilmi et al. 2022), currently extensively farmed for cereals. Today the site is approximately 100km inland but maritime access may once have been closer, as it is likely that the Gharb drainage to the north formed an indented embayment or estuarine wetland during later prehistory.

Figure 1. a) The north-western Maghreb, showing Oued Beht and other locations mentioned; b) the Oued Beht ridge and river, from the south-east (map and photograph by Toby Wilkinson).

The site first came to attention in the 1930s, when colonial French building work revealed a prodigious number of polished stone axes/adzes and macrolithic grinding artefacts, as well as nearby large linear features taken to be the walls of a prehistoric fortified enclosure (Ruhlmann 1936). Ever since, the site has been associated with an abundance of such axes/adzes (Souville 1973: 156–60). At least 1388 examples are currently curated in Rabat, at the Musée de l'Histoire et des Civilisations (hereafter Rabat Museum) and the Institut National des Sciences de l'Archéologie et du Patrimoine (INSAP), while many more are rumoured to have left Morocco via colonially sanctioned or subsequent illicit collection. Archaeological investigation was initiated in 2005 (by Youssef Bokbot), with an initial focus on Ifri n'Amr o'Moussa cave at the ridge's southern tip. There, levels associated with Epipalaeolithic, Early to Middle Neolithic (including burials and a radiocarbon date of 5210–4952 BC on domesticated barley), Beaker period and much later activity have been revealed (Bokbot 2005; Ben-Ncer et al. 2017). In 2013 a French-Moroccan team shifted attention to open areas beyond the cave and in 2016–2017 excavations were opened within the zone of abundant lithic surface finds, after which this collaboration ceased; the results remain unpublished.

Since 2021 the British-Italian-Moroccan Oued Beht Archaeological Project (OBAP) has undertaken new fieldwork, with four aims: to establish temporal and spatial patterns of activity through radiocarbon dating and intensive surface survey, complemented by drone-based photogrammetry; to understand the nature of the sub-surface archaeology, through excavation and geophysical prospection; to identify indicators of subsistence including faunal and macrobotanical remains; and to obtain a quantifiable sample of material culture sufficient for morphological, chronological, functional and materials analysis. Given how little is otherwise known about how people lived in this time and place, in combining the results of these strands, we place a strong emphasis on inference-building from the data up. Comparison with other societies of the Mediterranean and Africa remains an important goal, but one that also presents dangers of premature explanation by analogy.

Dating and extent

So far, we have generated 13 radiocarbon dates, all on charcoal or seeds recovered from our own excavations, OBAP's cleaning of earlier trenches and from material recovered by the earlier French-Moroccan excavations (provided by Youssef Bokbot) (Table 1). Most of the samples derive from deep pits that are widely encountered in trenches across the northern half of the ridge. The dates, from widely separated pits, consistently indicate a range of 3400–2900 BC (at 95.4% confidence), with the exception of two earlier fourth-millennium BC dates, and one with a slightly later end date (c. 2700 BC). This result matches the stylistic homogeneity of much of the prehistoric pottery recovered from the site. Given these dates and distinctive associated cultural traits, as well as uncertainties around the definition of the antecedent regional Middle and Late Neolithics (Linstädter 2016; Martínez Sánchez et al. 2018), we propose that this phase of activity at Oued Beht be referred to as a regional Final Neolithic, thereby filling one of the hitherto most obscure phases of Maghrebian prehistory.

Table 1. Radiocarbon dates from Oued Beht (calibration curve IntCal 20).

* From Lucarini et al. (2021).

During intensive surface survey in 2022 we collected pottery, chipped stone, axes/adzes and macrolithics at a 10 × 10m resolution over 19.5ha, generating the first fine-grained definition of the extent of the site and the distribution of material classes within it (Figures 2–5). The resultant total of 19 626 finds comprises 16 258 pottery fragments, 2947 chipped stone pieces, 50 axes/adzes and 371 macrolithics. The main focus of Final Neolithic activity emerges as a 9–10ha zone of high artefact density across the north of the site, with convincingly delineated edges save for losses to cliff erosion in the east. The extent of surface material is compatible with the currently known distribution of pits (Figure 2), though somewhat smaller than a previously published estimate prior to OBAP's fieldwork (Lucarini et al. 2021: 152). A smaller discrete concentration of prehistoric material near the large built features (OBAP's Walls B1, B2 and C) to the south also appears to be of Final Neolithic and/or slightly broader prehistoric date (possibly including the Beaker period). The concentration of pottery and lithics at Oued Beht is of a size unprecedented at this date on the African continent outside the Nile corridor and its immediate vicinity, and is also exceptional in Mediterranean terms.

Figure 2. Plan of Oued Beht, showing known pits and linear structures (figure by Toby Wilkinson).

Figure 3. Distribution of prehistoric pottery and Final Neolithic painted sherds (figure by Toby Wilkinson).

Figure 4. Distribution of macrolithic tools by function (figure by Toby Wilkinson).

Figure 5. Distribution of polished axes/adzes and chipped stone tools (figure by Toby Wilkinson).

Subsurface archaeology and evidence of domesticates

Small-scale targeted excavations in 2021–2022 prioritised cleaning and extending the French-Moroccan trenches within the Final Neolithic site. OBAP's trenches 1 and 2 revealed a shallow stratigraphy in the first 0.1–0.4m, comprising activity surfaces, postholes and other features, fragments of structural daub, and the typically narrow mouths of deep pits (of approximately 1–2m³ capacity), whose bellying, roughly bell-shaped form is exposed in the deeper sections. The associated cultural material relates overwhelmingly to the Final Neolithic. The pits were encountered in every trench and average approximately one per 10m²; should this density prove typical across the entire site—a question for upcoming geophysics and test trenching—there are significant volumetric consequences. The constricted mouths of the pits may suggest an initial use as storage silos, potentially for crops (a traditional North African practice until recently; Morales et al. 2014), although this hypothesis requires further investigation given that the pits’ fills probably do not reflect their primary use but, instead, formal or informal waste disposal (see below). This issue finds parallels in the debate around the storage, refuse and/or ritual functions of similarly shaped and numerous pits in contemporaneous southern Iberia, a resonance to which we will return (Jiménez-Jáimez & Suárez-Padilla 2020; Armenteros-Lojo & Jiménez-Jáimez 2024).

One strategic aim in 2022 was to obtain an initial insight into Final Neolithic subsistence practices. Pit 222 in trench 2 was accordingly excavated, revealing a rapidly deposited lower fill containing large sherds of fine Final Neolithic vessels and fragmented grinding tools, and upper fills from gradual natural processes, with pottery of comparable date and no later intrusions (Figure 6). Wet-sieving of the pit's contents produced charred seeds of domesticated naked barley (Hordeum vulgare var. nudum) and, more rarely, wheat (Triticum sp.) and pea (Pisum sativum), alongside wild olive (Olea europaea subsp. oleaster) and pistachio (Pistacia atlantica/terebinthus), probably representing food-processing residues (Table S1 in online supplementary material (OSM), Figure 7). Samples of the first three species have been radiocarbon dated (Table 1). Analysis of faunal remains confirms the presence of domestic goats, alongside sheep, cattle, pig and a single equid tooth, with all the main carcass parts represented (except for equids) and cut marks suggestive of butchery on site (Figure 8). These results match initial observations of the site's overall faunal profile, which is dominated by domesticated rather than wild species (although given that wild bovids and suids were indigenous to the Holocene Maghreb, some diversity of status within these taxa cannot at present be ruled out). The untapped potential of Maghrebian later prehistory is exemplified within this one modestly sized pit, which has alone produced the most reliable evidence for agriculture in the region for almost two millennia on either side of the site's late-fourth-millennium floruit.

Figure 6. Section profile from trench 2 showing pit 222, stratigraphic units and associated radiocarbon dates (drawing by Alessia Brucato & Lucy Farr).

Figure 7. Plant macro-remains from trench 2, pit 222: a) Hordeum vulgare var. nudum, naked barley; b) Triticum sp., wheat; c) Pisum sativum, pea; d) Pistacia atlantica/terebinthus, wild pistachio; e) Olea europaea subsp. oleaster, wild olive. Scale bar = 1mm (photographs by Jacob Morales).

Figure 8. Faunal remains from trenches 1 and 2: a) relative proportion (percentage number of identified specimens) of the total identified mammals; b) caprine mandibular condyle fragment with cut marks, and close-up detail (trench 2, pit 222, context 201.08); c) caprine astragalus with cut marks, and close-up detail (trench 2, pit 222, context 201.16) (photographs by Hassan Hachami).

Material culture

The sample of material culture recovered from Oued Beht is substantial (Figures 9 & 10). Here, we report only the prehistoric material, though the surface survey (unlike the excavations) also retrieved considerable quantities of medieval and later pottery, along with rotary querns of volcanic stone. Some 1295 prehistoric sherds were recovered from stratified contexts in trenches 1 and 2. A range of round-based jars, bowls and other open shapes are mainly made of a buff-firing clay, variably (sometimes finely) levigated (Figure 9a), while shallow cooking plates are typically of a red-brown fabric (Figure 9b). Some vessels are large, with massive tunnel lugs or handles. Approximately 25 per cent of the pottery, mainly in the buff fabric, is slipped (red-brown or dark) and burnished, sometimes to a high lustre. In addition, dark-on-light painted pottery constitutes 1–7 per cent of counts per context (Figure 9c). This otherwise little-known type has rare, broadly contemporaneous parallels at Ghar Cahal in north-west Morocco (Becerra et al. 2021), and at multiple sites in southern Iberia (Aranda Jiménez et al. 2012: 49–55; Carrasco Rus et al. 2012; Mederos Martín et al. 2023a). Overall, the degree of ceramic technological sophistication is striking relative to other early Maghrebian pottery. The survey pottery complements this picture (Figure 3), confirming the extent of Final Neolithic material, including painted sherds, as well as adding hints of an earlier Neolithic presence in the north and patchy post-Final Neolithic activity to the north and south, near the linear built features.

Figure 9. Final Neolithic pottery from Oued Beht: a) buff fabric bowls, jar (with post-firing hole), large tunnel lugs and round base; b) red-brown fabric cooking vessels; c) painted sherds (scale bar = 50mm) (photographs by Rafael Laoutari, Rafael Martínez Sánchez & Moad Radi).

The macrolithic assemblage is overwhelmingly prehistoric. Most examples were collected by the surface survey (371 artefacts), with only 13 from excavation (including five grinding tools in pit 222). The vast majority (340) of the surface finds are associated with the Final Neolithic focus in the north (Figure 4); only 31 are from the south, mainly close to the built features. Most (53.6%) consist of grinding implements, probably for cereal processing (Figure 10a & b). Many tools were also reused as anvils (including some pierres à cupules; Figure 10c), which, together with those solely used for hammering, testify to extensive on-site manufacturing, including the first stages of lithic artefact knapping by bipolar percussion. Tools for pounding, abrading, burnishing and polishing, as well as possible palettes, are attested in smaller numbers. A single pick (pièce à gorge) parallels those previously reported at the site by Souville (1973: 157–60, figs. 77–80) and presently curated at Rabat Museum (Figure 10d). The prehistoric macrolithics are mostly made from sandstones and quartzitic sandstones, mainly sourced from cobbles and boulders locally available along the river bed or in the conglomerate exposures on the ridge. Rarer raw materials include basalt, schist, breccia, limestone and granitoid rocks. Traces of manufacture (knapping, pecking and grinding) can be detected on several artefacts, especially the grinding implements.

Figure 10. Stone tools from Oued Beht: a & b) lower grinding stones; c) upper grinder reused as anvil (pierre à cupule); d) pick (pièce à gorge); e & f) polished axes; g) axe/adze preform; h) serrated sickle element; i) rectilinear sickle element; j) circular endscraper; k) product from bipolar percussion (a & c = Aït Siberne authority office; b, d–f = Rabat Museum; g-k = OBAP 2022 survey. Scale bar: a–d = 200mm; e–k = 50mm) (photographs by Lorena Lombardi & Moad Radi).

Intriguingly, given Oued Beht's reputation as a provenance of axes and adzes, only 50 examples were recovered in 2021–2022, all as surface finds and almost all spatially associated with the main Final Neolithic site (Figure 5). This modest number compares to some 1388 from earlier, mostly twentieth-century, collections (Figure 10e & f). The discrepancy may be due to depletion of (near-)surface finds over decades of collection or could imply concentrations of such objects in contexts that we did not encounter. Most of the 50 examples are made from quarzitic sandstone or basalt. In form they are triangular or oval (or rarely trapezoidal or rectangular) and approximately 50–120mm in length. Most are finished or almost finished tools, some exhibiting traces of use, but others are apparently unused. Recovery of several pre-forms (Figure 10g) and numerous manufacturing flakes (the latter from survey (Figure 5) and excavation contexts) confirm on-site axe/adze production.

Chipped stone is abundant (3368 pieces, 2947 from the surface survey and 421 from stratified excavations), but only a small proportion are formal tools as opposed to expedient forms and flakes of low chronological diagnostic value. Most artefacts are of flint, followed by quartzite and, rarely, chalcedony, from locally exposed conglomerate nodules. The artefacts were knapped using both direct percussion, with a hard or soft hammerstone, and bipolar technique for splitting pebbles and detaching flakes (Figure 10k). A minority of the formal tools are manufactured using finer-grained flints. The most common type (19 examples) is a circular or arched front endscraper, mainly found in the north (Figures 5 & 10j). More chronologically informative, however, are eight flint sickle elements. Six are characterised by a serrated active edge and backing retouch, rendering the hafted edge semicircular or rectilinear in shape, and retaining traces of gloss from use (Figure 10h). These stand out as typologically similar to artefacts from the Beaker-period site of Nador Klalcha near Kenitra in north-west Morocco (Rodrigue 2012: 72, fig. 4, nos. 22 & 23) and also from sites from the later third to second millennia BC in Iberia (Early Bronze Age; Cabanilles 1985; Gibaja Bao 2003). The remaining two sickle elements are produced on blades and show a rectilinear working edge (Figure 10i). Most sickle elements were recovered from the south and north-west of the site (Figure 5), both areas with hints of possibly post-Final Neolithic pottery. In contrast to evidence for crop processing, consumption and potentially major storage, tools for harvesting are conspicuously lacking for the main Final Neolithic phase at Oued Beht. This could imply the use of a bare-handed technique, a traditional practice in arid regions (Lucarini 2014; Simm & Russell 2018), or that primary harvesting and/or the curation of sickles was based elsewhere.

Interpreting Oued Beht in its broader context

What we have learnt and what we still do not understand about Oued Beht are now finely balanced. Over a minimum of five centuries around the later fourth to earliest third millennium BC (and probably a couple more centuries on either side), a major concentration of activity and investment of labour and resources developed across an area of at least 9–10ha, focused on the northern part of the ridge. This included the digging and use of numerous pits, as well as above-ground activities and structures of undetermined nature. Pottery for food preparation, consumption and storage, some of it elaborately decorated, were in use alongside a chipped stone tool industry and the large-scale provision of grinding stones and polished stone axes/adzes, at least some of which were made on site. Food from a typically Mediterranean Neolithic suite of domestic animals and crops was processed and consumed, and, in the case of the latter, possibly stored in bulk. By contrast, unambiguous signs of the gathering and hunting of wild resources are rare, suggesting an economy primarily reliant on food production. Anecdotal reports of previous finds of human remains may point to a further dimension of activity, currently under investigation by ongoing excavations.

It is tempting to bypass the uncertainties of interpretation and assume that Oued Beht was a large Final Neolithic village. Such may prove to be the case and, if so, its size, the abundance of grinding stones and the potential storage capacity of the site's pits, would suggest that it was a populous one. Such a community would be unprecedented in the Maghreb and indeed anywhere in non-Nilotic Africa until the Dhar Tichitt sites of the second-millennium BC Sahel (Linares-Matás 2022). Along the northern flank of the later fourth- to third-millennium BC Mediterranean, Oued Beht is only significantly superseded in terms of extent by the extraordinary Iberian Copper Age mega-sites, themselves often studded with pits and of equally debated social significance (García Sanjuán et al. 2017). Of these, Valencina de la Concepción, near Seville, remains the most extreme example, at some 200ha during its third-millennium floruit (Mederos Martín et al. 2023b: 280, fig. 30, 289, tab. 3), but more strictly contemporaneous comparisons are furnished by sites of the later fourth-millennium BC Iberian terminal Neolithic to Copper Age transition, notably La Loma (near Granada), of uncertain extent but possessing both numerous pits and painted pottery (Aranda Jiménez et al. 2012), and the 20ha site of Perdigões in southern Portugal (Valera 2018). Further east, Oued Beht compares favourably in size with Early Bronze Age Aegean sites such as Troy and others on Crete and the Greek mainland (Whitelaw 2004; Jablonka 2016), locations long at the heart of that region's ‘emergence of civilisation’ narrative (Renfrew 1972).

However, given all that we still do not know, and the likely co-existence of farming and pastoralism across the wider Maghreb, we prefer to keep alternative models of aggregation constructively open, acknowledging the possibility of different, and less familiar, trajectories that could have resulted in the complex phenomenon that is Oued Beht. One consequence of shedding light on Oued Beht is that it emphasises, by contrast, the almost total lack of knowledge of its surrounding context. Temporally, we still know virtually nothing of the immediate background from which the site emerged and little about its succession, beyond a few hints of earlier Neolithic activity, and potentially a post-Final Neolithic presence, both suggestively matched at Ifri n'Amr o'Moussa cave at the southern end of the ridge. Spatially, aside from a handful of under-investigated prehistoric sites, and a probable salt mine, within a day's return walk (Ruhlmann 1936, 1937; Souville 1991; Ben-Ncer et al. 2017), we know nothing of the wider social landscape. Bearing in mind the paucity of Final Neolithic harvesting tools, coupled with the fact that the potential storage capacity of perhaps hundreds of pits is likely to have exceeded the needs of any plausible resident group, it is conceivable that Oued Beht acted as a hub for a widespread, so far invisible population. In this context the presence or absence of donkeys (regionally likely by this date but locally still undetermined) may prove informative in terms of the ability to transport and mobilise crops and other goods (Lucarini et al. 2021: 152–53). Extended survey in Oued Beht's vicinity is therefore vital, and at a more ambitious spatial scale might reveal how typical or exceptional this site is at a regional level.

Panning out further, the centuries either side of 3000 BC saw the emergence in southern Iberia of, at least superficially, similar large terminal Neolithic to Copper Age sites with many pits—the so-called ‘silo culture’ extending from western Andalusia to the Portuguese Alentejo (Márquez Romero & Jiménez Jáimez 2010; Martínez Romero 2018). It is possible that travellers from across the sea will in future be identified at Oued Beht and their presence help to explain the thought-provoking parallels between the two regions. Yet it is also manifest that Africa has long endured a notorious colonial experience of exogenous explanation for supposedly unprecedented local developments. In this instance, the relatively early dates emerging from Oued Beht give pause for thought; it may ultimately prove impossible to explain fully Iberian Copper Age developments—not least the strengthening evidence for engagement in far-flung western Mediterranean exchange networks—without regard for contemporaneous north-west African agency. At present, for example, Oued Beht boasts the largest site-specific corpus of a painted pottery tradition, much of it plausibly locally produced, that is more thinly (if also more widely) attested north of the strait. Likewise, it remains to be established how far the axes manufactured at Oued Beht may have travelled. Ivory was worked in the north-west Maghreb long before its first, later fourth-millennium BC, appearance north of the strait (Schuhmacher 2016) and, although there is as yet no evidence for ivory at Oued Beht, its surrounding region was still roamed by elephants (and probably ostriches) as late as the Roman period. Speculatively, the first Iberian donkeys (Bernáldez-Sánchez et al. 2023) could constitute another Maghrebian import. It is therefore crucial to consider Oued Beht within a wider co-evolving and connective framework embracing peoples both sides of the Mediterranean-Atlantic gateway during the later fourth and third millennia BC—and, for all the likelihood of movement in both directions, to recognise it as a distinctively African-based community that contributed substantially to the shaping of that social world.

Conclusion

Addressing the long-known but little studied site of Oued Beht, through fieldwork and related analyses, produces both new insights into Maghrebian society and many further questions. Prominent among the former are the undeniable presence of an established farming lifestyle in the late fourth millennium BC Maghreb, and of complex local communities reciprocally engaged with their contemporaries in southern Iberia. Our hope is that, however challenging it may prove to resolve some of the outstanding questions, Oued Beht and the north-west Maghreb will henceforth occupy an integral and, in explanatory terms, profoundly revisionary place in the later prehistory of the Mediterranean and Africa that they so richly deserve, and that for far too many years has gone unrecognised.