Ostrich eggshell beads reveal 50,000-year-old social network in Africa

Abstract

Humans evolved in a patchwork of semi-connected populations across Africa^1,2; understanding when and how these groups connected is critical to interpreting our present-day biological and cultural diversity. Genetic analyses reveal that eastern and southern African lineages diverged sometime in the Pleistocene epoch, approximately 350-70 thousand years ago (ka)^3,4; however, little is known about the exact timing of these interactions, the cultural context of these exchanges or the mechanisms that drove their separation. Here we compare ostrich eggshell bead variations between eastern and southern Africa to explore population dynamics over the past 50,000 years. We found that ostrich eggshell bead technology probably originated in eastern Africa and spread southward approximately 50-33 ka via a regional network. This connection breaks down approximately 33 ka, with populations remaining isolated until herders entered southern Africa after 2 ka. The timing of this disconnection broadly corresponds with the southward shift of the Intertropical Convergence Zone, which caused periodic flooding of the Zambezi River catchment (an area that connects eastern and southern Africa). This suggests that climate exerted some influence in shaping human social contact. Our study implies a later regional divergence than predicted by genetic analyses, identifies an approximately 3,000-kilometre stylistic connection and offers important new insights into the social dimension of ancient interactions.

Fig. 1. Locations of sites included in this study and palaeoclimate records.

Base map modified from Natural Earth. a, Kakapel Rockshelter (1); Enkapune ya Muto (2); Mumba Rockshelter (3); Panga ya Saidi (4); Daumboy 3 Rockshelter (5); Kisese II Rockshelter (6); Mlambalasi Rockshelter (7); Magubike Rockshelter (8); White Paintings Shelter (9); Geduld (10); Lower Numas Cave (11); Lower Orabes Shelter (12); Leopard Cave (13); Eros (14); Wortel (15); Bushman Rockshelter (16); Border Cave (17); Apollo 11 Cave (18); Wonderwerk Cave (19); Dikbosch 1 Shelter (20); Sehonghong (21); SK2001.026 (22); Rooiwal Hollow/Midden (23); Varsche Rivier 003 (24); Paternoster (25); Grassridge Shelter (26); Witklip (27); Kasteelberg A + B (28); Geelbek Dunes (29); Voelvlei (30); Nelson Bay Cave (31). b, Representative OES beads from sites in eastern Africa. c, Representative OES beads from sites in southern Africa.

Fig. 2. OES bead diameter, thickness and aperture diameter distribution through the past 50 kyr in eastern and southern Africa.

a, Generalized additive model plots to show bead characteristic evolutionary trajectories. The mean (curved lines) and 95% confidence interval are shown for each parameter. b, Split violin plots of bead parameters. The violins represent the kernel density of the frequency distribution, and the points are presented as mean values ± one standard error. Statistical results are shown in Supplementary Tables 5–7, 11. The asterisks denote the significance of region differences (*P < 0.01 and ***P < 0.001). When no significance between two regions was found, P values are presented.

Fig. 3. Comparison of bead characteristics between eastern and southern Africa during the past 50 kyr.

a, Principal component analyses (PCA) of OES bead metric parameters for phases III–V. Diameter, aperture diameter and thickness account for more than 89% of the variation, separating eastern and southern Africa into distinct groups. PCA clustering for two regions for phase II and phase I is not possible due to insufficient data. b, Paired diameter and aperture diameter for each phase. Newly reported data: collected by authors, reported as individual beads; published data: drawn from published metrics, reported as individual beads; average data: drawn from published metrics, reported as averaged values.

Fig. 4. Bead-derived social connection and comparison climate proxies from eastern and southern Africa.

a, Annual mean rainfall map of Africa (map modified from Wikimedia Commons under the CC BY-SA 4.0 licence), showing the position of palaeoclimate data (yellow squares) used in c: Lake Tanganyika; Lake Malawi; Core GIK16160-3 (ref. ); Core 64PE304-80 (ref. ); Zambezi River catchment area (red) and major tributaries (white). The two black boxes indicate the modelled climate data area for eastern Africa (22.5–40° E, −9 to 9° N) and southern Africa (8–35° E, 20–35° S) from the LOVECLIM transient climate model. The black dashed lines indicate the location of present-day Intertropical Convergence Zone (ITCZ) in January and July. b, Latitudinal north to south sequence of climate data series (from top to bottom): total precipitation for eastern Africa (22.5–40° E, −9 to 9° N) derived from the LOVECLIM transient climate model; moisture availability derived from the δD of leaf wax record from Lake Tanganyika is overlaid with the June–July–August (JJA) mean insolation at 0–20° N (red dashed line); lake level is derived from PCA analyses (PC1) in Lake Malawi; moisture availability derived from the δD of leaf wax record (ice volume corrected) from the Zambezi River catchment area overlaid with the December–January–February (DJF) mean insolation at 0–20° S; the neodymium (Nd) isotope signature of lithogenic fractions in marine sediments from the Mozambique Shelf margin; the total precipitation for southern Africa (8–35° E, 20–35° S) derived from the LOVECLIM transient climate model. HE3, Heinrich Event III; LGM, Last Glacial Maximum; VSMOW, Vienna Standard Mean Ocean Water. c, Social connections between eastern and southern Africa, derived through bead styles, over the past 50 kyr.