High on a windswept escarpment in southeastern Mauritania, where sheer cliffs drop more than 350 meters into sand-choked basins, lies one of West Africa’s least explored archaeological landscapes: hundreds of dry-stone villages, walled compounds and burial monuments left behind by the successors—and quite possibly the descendants—of the region’s first farming civilization. Now a team of German researchers has completed the first archaeological field survey on the Tagant escarpment in more than three decades, documenting hilltop settlements laid out on a grid, intact domed stone buildings, early iron-smelting furnaces and a stone slab carved with pecked glyphs related to the ancient Libyan script, a writing tradition never before reported in this part of the Sahara. The work, published as an open-access research report in the African Archaeological Review, does not yet rewrite the textbooks. But it lays the groundwork for answering a question that has lingered since the early 1990s: what became of the people of the Tichitt culture when their cliff-top world came to an end?

The Tichitt culture—known in French as the Culture Néolithique de Tichitt and in anglophone literature as the Tichitt tradition—flourished between roughly 3000 and 400 BCE along the escarpments bordering the Hodh depression, most famously at the cliffs of Tichitt and Oualata. It is the earliest known sedentary, food-producing complex society of Saharan and Sahelian West Africa, and its pivotal role in the region’s cultural development is difficult to overstate. Its inhabitants built dry-stone compounds in settlements of demonstrably different sizes, kept cattle, sheep and goats, and cultivated domesticated pearl millet, the first cereal grown in this part of the Sahara; plant impressions pressed into their pottery provided the key evidence for that crop. Yet a century of research has concentrated on the eastern escarpments, and both the culture’s full geographic extent and its internal chronology remain contested. Some researchers now argue that it stretched from the southern margins of the Mauritanian Adrar in the north to the Rkiz massif in the south, and possibly as far east as the Malian lakes region—making the question of who lived on its margins all the more pressing.

Between 1986 and 1992, the Mauritanian archaeologist Mohamed Ould Khattar undertook four field seasons along the Tagant escarpment, much of it by camel, documenting 273 sites across some two hundred kilometers of rough, largely uninhabited terrain. His 1995 doctoral thesis remains the foundation of everything known about the area. He identified one site type closely resembling the classic Tichitt assemblages—complete with pillar structures and a total absence of metallurgy—but argued that most Tagant settlements belonged to distinct local categories, marked by iron metallurgy, rectilinear “ditch” structures, and a virtual absence of chipped stone. Iron-working is conspicuously absent from the Tichitt heartland, which makes the Tagant a potential laboratory for the region’s transition from the Ceramic Late Stone Age into the Early Iron Age. No new fieldwork followed for three decades, although satellite-based studies have multiplied the known inventory: a remote survey published in 2021 recorded 276 funerary tumuli in eighteen clusters, and more recent mapping counts more than a thousand sites in all.

Satellite data alone, however, cannot resolve questions of chronology, subsistence or cultural affiliation—and the Tagant is among the hardest places in the Sahel to survey on foot. In places the elevation difference between the plateau top and the basin below, the baten, exceeds 350 meters; sheer cliff faces whose ledges are packed with wind-blown sand, and scree slopes of rolling dolerite boulders cut by deep ravines, make direct ascent a serious challenge even for experienced mountaineers. Because the ancient builders used those same dolerite boulders, their structures are often invisible from ground level until one is standing on them. In January 2025, Nikolas Gestrich of the Frobenius Institute and Maximilian Schulze of Goethe University Frankfurt therefore spent eight days documenting two contrasting stretches of the escarpment, with the support and advice of Mauritania’s Institut Mauritanien de Recherche et Formation en Patrimoine Culturel. The campaign was strictly non-invasive—no permission to excavate, no sampling—so every observation rests on surface architecture and artifacts alone.

The most extraordinary site was DT_998, a sprawling settlement complex that satellite researchers had flagged as unique in the region but which was apparently never visited by Ould Khattar. On the open parts of the slope its layout is astonishingly regular: straight pathways divide almost rectilinear compounds, each containing one or more circular domed dry-stone structures entered through a single low linteled doorway, a small rectilinear area about five by 1.5 meters, and an unbuilt but cleared and enclosed yard below. Rectilinear planning of this kind has never been described elsewhere on the Tagant. Higher up, among giant dolerite boulders, the builders cleared platforms, modified natural rockshelters and raised two domed structures that still stand intact, with an internal height of around 1.7 meters and living floors littered with potsherds and animal bone. The architecture recalls that attributed to the Gangara—believed to be Soninké people known for their dry-stone granaries and terracing further west and south—although the authors caution that the very notion of a coherent “Gangara culture” may not survive renewed fieldwork in the area.

A few kilometers away, in the recessed valley of Kheneg Levraiat, the team documented DT_942, one of the largest delimited sites on the entire escarpment: a roughly 35-hectare settlement enclosed by a dry-stone wall standing up to 1.8 meters high, with a single clear funnel-shaped entrance in the northeast and simple openings to the east and west. The wall, the researchers note, gives less the impression of a defensive rampart than of an enclosure built to keep things in—livestock, most likely. In its northwestern quarter stand two low mounds about five meters across, blanketed in iron slag and technical ceramics, including tuyères—furnace blowpipes—tempered with long grass stems and vitrified at their tips: the remains of what the team interprets as only a handful of small-scale smelting operations. Inside the enclosure, a single standing stone structure set within a large walled area echoes the Tichitt tradition’s celebrated “structures à piliers.” And at DT_1134, a group of four tumuli nearby, a flat rock carries pecked glyphs that are neither Arabic nor Tifinagh, but appear related to the ancient Libyan script—an unexpected first for a region otherwise known only for Tifinagh inscriptions.

The ceramic evidence, distilled from 181 recorded sherds, points to two distinct technological traditions coexisting on the escarpment. One is the classic Tichitt recipe: vessels tempered with large quantities of vegetal material, leaving recognizable impressions of chaff and millet grains in the clay. The other, by far the dominant strategy at the Tagant sites, relies exclusively on mineral temper—small gravels under half a centimeter embedded in the paste. Decoration spans the breadth of the Tichitt repertoire and beyond: cord-wrapped stick roulettes, twisted and knotted cord roulettes, rocker stamping, combed wavy lines unique to site DT_949, plastic nubbins, and incised geometric bands. Particularly diagnostic are the applied clay bands, or bandes appliqués, which Munson dated to the Arriane phase around 900–700 BCE, and high collared rims closely resembling his Type 16. Strikingly, the hallmarks of the medieval period—red-slipped and burnished wares, folded strip roulettes—are almost entirely absent, suggesting that unlike the neighboring Dhar Néma, the Tagant saw no significant medieval reoccupation.

The stone tools tell a subtler story still. Silcrete and quartzite scrapers—mostly short, wide flakes with steep unifacial retouch and almost no remaining cortex—turn up at nearly every site, yet without a single chip of manufacturing debris, implying the tools were knapped elsewhere and imported as finished objects. More surprising is what is missing: arrowheads. On the Tichitt and Oualata escarpments, bifacial projectile points are among the most characteristic finds; on the Tagant, the team documented only two candidate points in eight days of survey. Conversely, polished dolerite axes and gouges, which Ould Khattar had interpreted as rare, curated heirlooms, proved to be locally manufactured: at the lowland artifact scatter DT_949 the researchers found chipped roughouts alongside edge-damaged, transversely broken specimens showing heavy use. Together with abundant sandstone grinding basins and hand grinders, an amazonite bead and a polished quartz “marble” of a type typical of Tichitt sites, the assemblage sketches a genuinely transitional economy: chipped-stone technology fading, small-scale iron production emerging, and ground-stone tools still the workhorses of daily life.

Equally consequential, the researchers argue, are the sites without any stone architecture—landscapes that satellite surveys cannot see at all. At the foot of a dune northeast of the well of Eyriera, extensive artifact scatters cover thin, carbonate-hardened deposits that appear to be the former floors of a lake. Embedded within them are several articulated skeletons of Arius gigas, a giant catfish that once patrolled Holocene Saharan waters, alongside fossilized freshwater molluscs, one probably Bulinus truncatus—unmistakable evidence that this basin, today inhabited by a solitary goatherd, once held a permanent body of water. The artifacts on these dune sites are the most diverse of the entire survey: grinding basins and handstones, ground-stone axes, chipped quartz implements, a fragment of a polished stone arm ring, and red-slipped, burnished pottery unlike anything found on the cliff-top villages. Their variety hints at human activity that both predates and postdates the stone settlements, and the authors urge that future research give these unconstructed sites at least equal weight to their more spectacular neighbors.

Existing radiocarbon dates—few, and often imprecisely tied to their contexts—suggest that the Tagant was occupied, at least sporadically, across the entire span of the Tichitt tradition, but was most intensely settled in the five centuries either side of the start of the Common Era, precisely when the eastern escarpments show a major dip in occupation probabilities, before the area was abandoned abruptly. One long-standing hypothesis holds that the Tagant served as a refuge for populations emigrating from Tichitt and Oualata in response to climatic deterioration. A single pottery date from DT_949—4040 ± 65 BP, or 2871–2356 cal BCE—raises the further possibility that some surface scatters are palimpsests of human activity stacked over several millennia. Gestrich and Schulze are now planning a return: a full restudy of Ould Khattar’s collections, direct dating of the ceramic assemblages, and targeted excavation of the natural and anthropogenic sediment traps that could preserve stratified layers in a landscape where millennia of deflation have stripped most sites bare. The fate of West Africa’s first farmers, it seems, is still waiting in the cliffs of the Tagant.