Navigation – Plan du site

AccueilNuméros17Archaeological Evidence for Popul...

Archaeological Evidence for Population Rise and Collapse between ~2500 and ~500 cal. yr BP in Western Central Africa

Preuve archéologique de l’augmentation et de l’effondrement de la population entre ~2500 et ~500 ans cal. BP en Afrique centrale occidentale
Geoffroy de Saulieu, Yannick Garcin, David Sebag, Pascal R. Nlend Nlend, David Zeitlyn, Pierre Deschamps, Guillemette Ménot, Pierpaolo Di Carlo et Richard Oslisly
p. 11-32
Traduction(s) :
Preuve archéologique de l’augmentation et de l’effondrement de la population entre ~2500 et ~500 ans cal. BP en Afrique centrale occidentale [fr]

Résumés

Des études paléoenvironnementales antérieures ont montré que des changements majeurs de la végétation et de l’environnement se sont produits en Afrique centrale à partir de l’Holocène moyen (ex. Maley & Brenac 1998). Plusieurs d’entre elles mettent en évidence une origine humaine et supposent que les grandes migrations de population, les innovations techniques (par exemple, la technologie de la fonte du fer) et/ou de nouveaux choix dans les pratiques agricoles, conduisant à la déforestation et au défrichement, sont les moteurs de ces changements. Cependant, à ce stade, l’absence de reconstitution démographique ne permet pas de soutenir pleinement ces hypothèses. Notre étude utilise une base de données archéologiques géoréférencées pour déduire la dynamique des populations et l’évolution des pratiques culturelles en Afrique centrale occidentale au cours des 5000 dernières années. Cette base de données comprend 1139 dates calibrées au 14C provenant de 425 sites – localisés dans le sud du Cameroun, au Gabon, en République du Congo, en Guinée équatoriale et dans la partie occidentale de la République démocratique du Congo –, remontant à un maximum de 5000 ans cal. BP. La modélisation des données indique une possible croissance de la population entre ~2500 et ~1500 ans cal. BP, coïncidant avec l’apparition à l’échelle régionale de techniques et de pratiques spécifiques. L’augmentation concomitante des fosses dépotoirs, des vestiges d’utilisation de palmier à huile Elaeis guineesis, l’apparition de rares restes de millet Pennisetum glaucum et la montée en puissance des vestiges de métallurgie du fer ont eu lieu pendant la seconde moitié du Néolithique, à partir d’environ 2800 ans cal. BP. Dans les régions côtières, la croissance de la population concerne le Néolithique et le début de l’âge du fer (2500-2000 ans cal. BP et 2000-1500 ans cal. BP), tandis que dans l’Hinterland cette croissance semble légèrement plus tardive (2400 et 1300 ans cal. BP). Il n’est pas possible d’identifier un phénomène commun de diffusion à partir d’un seul centre. Les innovations techniques et les nouvelles pratiques semblent plutôt s’être répandues à travers un large réseau d’interactions culturelles qui a favorisé la formation des sociétés d’Afrique centrale occidentale au cours du troisième millénaire avant notre ère.

Haut de page

Texte intégral

Introduction

1Through the development of rescue archaeology and the funding of international programmes (e.g. Lavachery et al. 2010), archaeological research in Western Central Africa (henceforth WCA), comprising southern Cameroon, Gabon, the Republic of Congo, Equatorial Guinea, and the western part of the Democratic Republic of Congo, has made qualitative and quantitative progress since the 1990s. On the one hand, archaeology-only studies now include both site-specific analyses (e.g. Eggert & Seidensticker 2016) as well as regional syntheses (e.g. Bostoen et al. 2015; Bostoen 2018). On the other hand, there are a number of projects integrating archaeological data with palaeological (palaeoecological, palaeoclimatological, palaeoenvironmental), linguistic and genetic approaches (Quintana Murci et al. 2008; Berniell-Lee et al. 2009; de Filippo et al. 2012; Grollemund et al. 2015; Patin et al. 2017; Garcin et al. 2018; Lipson et al. 2020). The archaeological data accumulated in recent years cover several specific issues, including information on the first cultigens (palm oil, millet) (e.g. Neumann et al. 2012; Kahlheber et al. 2014), the evolution of regional cultures/ceramic styles (e.g. Wotzka 1995) and/or technological developments, such as the appearance, mastery and diffusion of iron metallurgy at a regional scale (e.g. Clist 2012).

2Although the archaeological sites investigated are unequally distributed in WCA, the accumulated dataset of 14C dates associated with archaeological evidence is now sufficient to infer regional population dynamics and to provide answers to specific questions regarding cultural dynamics (e.g. on ceramic styles, technology, etc.). Given the lack of human remains and ancient DNA in WCA, however, such an archaeological approach cannot easily distinguish between the spatial dynamics necessarily involving the movement of people (migration, colonization) and those without such a necessity (expansion/retreat of cultural styles, techniques and practices). Notwithstanding, through geochronological dating techniques (e.g. 14C), archaeology can distinguish between various phenomena. First of all, there are centrifugal dynamics (of people, of practices or both) that, for convenience, will be called in this paper diffusion. For instance, the diffusion of the Neolithic culture in Europe is illustrated by a clear 14C age distribution indicating the movement of pioneer populations (Ammerman & Cavalli-Sforza 1971, 1979; Bocquet-Appel et al. 2009; Dubouloz 2017). At the regional scale, a time lag between various related occurrences arranged on a gradient in both time and space would strongly suggest diffusion of populations or cultural traits.

3By contrast, a lack of spatial dynamics is more difficult to account for. Firstly, we cannot rule out that the methods available until now have been too imprecise to reveal any time lag. Secondly, it can also suggest that the modality of the cultural phenomenon in question is based on social interactions that are not yet understood. In this case, only local qualitative analyses may help to disentangle between diffusion and migration/colonization processes (e.g. Eggert and Seidensticker 2016; Wotzka 1995).

4Working from both demographic and cultural perspectives, this study compiles a regional dataset based on a comprehensive review of the archaeological literature concerning WCA during the last 5000 years. The review aims to (i) identify the key periods of regional population shifts through archaeological data, (ii) place the inferred population dynamics within a chronology of cultural and technical innovations and (iii) test the synchronicity of these innovations at a regional scale to show evidence of possible cultural diffusion processes. As a result, this study will provide an initial archaeological synthesis, independent of, but complementary to, current linguistic and genetic data, of a period of African prehistory in which the interplay between environmental changes, climate variability and human activities is still debated by scholars (Bayon et al. 2012; Clist et al. 2018; Garcin et al. 2018; Giresse et al. 2018; Maley et al. 2012; Neumann et al. 2012).

The Western Central African Archaeological Database and Statistical Methods

Radiocarbon ages and use

  • 1 The ‘Late Holocene Rainforest Crisis’ is a vegetation disturbance that is recorded in palaeoenviron (...)

5This study uses a regional georeferenced database including 425 sites and 1139 published 14C dates from archaeological sites in WCA ranging between 5000 and 15 14C yr BP (Fig. 1). Compiled Pennisetum glaucum, Elaeis guineensis, iron metallurgy, and pit features, were either directly dated or we assumed that the age of a given material is equal to ages of nearby dated material. This body of work has already been partially published in a companion paper (Garcin et al. 2018) mainly focused on the respective roles of climate variability and human activity in the Late Holocene Forest Crisis1. Radiocarbon dates and associated errors were rounded according to the convention proposed by Stuiver and Polach (1977).

Figure 1 – Spatiotemporal distribution of sample sites covering the past 5000 years.

Figure 1 – Spatiotemporal distribution of sample sites covering the past 5000 years.

(A) Current spatial distribution of the rainforest shown in green, taken from the Collection 5 MODIS Global Land Cover Type product (www.landcover.org). Countries: AGO–Angola; CAR–Central African Republic; ­CMR–Cameroon; COG–Congo; DRC–Democratic Republic of Congo; GAB–Gabon; GNQ–Equatorial Guinea; NGA–Nigeria. Overlain are 14C-dated archaeological sites in WCA with dated and associated material.
(B) and (C) time-latitude and time-longitude distribution of the calibrated radiocarbon dates, respectively. Horizontal and vertical bars show the 95% ranges of analytical error on the 14C dates.
(D) Histogram of the calibrated radiocarbon age (median) distribution of archaeological sites shown with 100-year bins for WCA.

6The analysed samples straddle the Equator, extending from ~10°N to ~5°S, a zone which experiences seasonal variations in atmospheric CO2. Consequently, the 14C calibration curve to apply can be either the IntCal20 (Reimer et al. 2020) for the Northern Hemisphere, or the SHCal20 (Hogg et al. 2020) for the Southern Hemisphere, or a mix of both curves depending on the sample location. The distinction between the Northern and Southern Hemisphere atmospheres is not trivial and the boundary between them is set by the thermal Equator or the Intertropical Convergence Zone (ITCZ) at the time when the sample was formed (i.e. during the growing seasons in question), rather than by the geographic Equator (McCormac et al. 2004; Hogg et al. 2020). In the absence of any precise paleo-ITCZ reconstruction for the region, all dates were calibrated with SHCal20, based on the recommendation to use this curve for areas south of ITCZ in December-February (Hogg et al. 2020), which currently dominate WCA. The long-term hemispheric offset is 36 ± 27 14C yr (SHCal20 minus IntCal20), which may create additional uncertainties in 14C calibration (Hogg et al. 2020). Radiocarbon calibration and summed probability distributions (SPDs) were computed using the freely available R statistical computing package, Bchron 4.7.5 (Parnell 2015), and all ages are shown as calendar years before AD 1950 (cal. yr BP).

7Before calculating the Summed probability distributions (SPDs) and other statistical analyses (see below), 14C dates were binned to correct for investigator bias and oversampling within sites (Crema et al. 2016; Shennan et al. 2013; Timpson et al. 2014). To account for the highly variable sampling intensity across the study area, closely distanced radiocarbon dates were binned in space, using an arbitrary 10-km radius, with the ArcGIS 10.2.1 (ArcGIS, 2010) spatial analyst toolbox. This spatial binning before calculating SPDs has been done to avoid the overrepresentation of sites and regions having been thoroughly excavated and studied (like cities and their vicinity). Actually, the impact of such binning is fairly limited: the amplitude of SPDs deviations is slightly less important than what it is without binning. Radiocarbon dates were further binned in time by clustering the mean 14C yr using a threshold of 200 yr.

8Human activity was inferred from the SPDs of 14C-dated and/or associated material, including occurrences of Pennisetum glaucum (pearl millet), Elaeis guineensis (palm oil), iron metallurgy and pit features. The age of any associated material was assumed to be the same as that of the nearby dated material.

Statistical Methods

9Summed 14C dates (Fig. 1D) have often been used as a proxy for ancient demography, including Central Africa (Oslisly et al. 2013; Wotzka 2006), based on the idea that a larger amount of archaeological material and sites in an area reflects the presence of more people at the time (Oslisly et al. 2013). However, this method has evident weaknesses, mostly related to sampling bias, taphonomic loss and fluctuations in the radiocarbon calibration curve, such as that caused by the presence of a plateau (e.g. the Hallstatt plateau, Van der Plicht 2005), all of which reduce the reliability of demographic inference. To avoid the evident limitations of previous attempts and to produce a new and more credible representation of the region, we used recently developed demographic proxies based on the SPDs of calibrated 14C dates. This approach has been used in similar studies for Europe (Shennan et al. 2013; Timpson et al. 2014), Western Africa (Manning & Timpson 2014), South America (Goldberg et al. 2016; Riris & Arroyo-Kalin 2019), South West Asia (Roberts et al. 2018) and East Asia (Crema et al. 2016; Oh et al. 2017; Zahid et al. 2016). The method is now well established and was previously applied to Western Central Africa (Garcin et al. 2018). Seidensticker et al. (2021) recently used this method focusing on the post-1600 cal. yr BP period. In contrast, we focus here on the pre-1600 cal. yr BP period corresponding to the initial settlement history.

10The usage of SPDs as a population proxy has been widely debated (see for example Contreras & Meadows 2014; Torfing 2015; Williams 2012; Lupo et al. 2018). If the shape of the SPDs is not overemphasized, this method provides valuable insight into palaeodemographic dynamics (Timpson et al. 2015; Crema & Bevan 2021).

11Population dynamics were reconstructed from the SPDs of 14C dates compared with formal statistical models using the hypothesis-testing approach introduced in Shennan et al. (2013), further developed in Timpson et al. (2014) and provided in R (R Core Team 2017), statistical computing language, by Crema et al. (2016). In summary, the SPDs are first fitted to a generalized linear model. Then, random calendar year dates are created via Monte Carlo simulation. Finally, random calendar year dates are “back calibrated” into radiocarbon dates to generate null SPDs. This method provides both statistically significant local deviations (compared to the null model) and a global significance test (Timpson et al. 2014; Crema et al. 2016).

12Here, the SPDs of 14C dates compiled in the database were calculated after correction for oversampling biases (see above). The results of the preparative binning process are shown in Table 1. In order to avoid edge effects related to the abrupt decline of radiocarbon date intensity toward the present day, as other dating techniques such as historical documents become available, the simulation results, which started from 5150 cal. yr BP were trimmed at 400 cal. yr BP.

Table 1 – Data used for the SPD analysis of each of the three regions of Western Central Africa

Samples (n)

Sites (n)

Binned sites (n)

Binned dates (n)

North Coast

526

157

84

215

Hinterland

241

106

66

137

South Coast

372

175

96

226

Table 2 – Statistical significance for the exponential model and pair-wise permutation tests (values in bold are significant at 0.01)

Pairwise permutation test

Exponential null model test

Vs. North Coast

Vs. Hinterland

Vs. South Coast

North Coast

<0.0001

-

0.0019

0.4174

Hinterland

<0.0001

0.002

-

0.1595

South Coast

<0.0001

0.3401

0.12

-

Table 3 – Data used for the construction of Fig. 1

North Coast

Hinterland

South Coast

Millet

Samples (n)

16

2

0

Sites (n)

4

2

0

Binned sites (n)

3

1

0

Binned dates (n)

4

1

0

Elaeis guineensis

Samples (n)

118

121

42

Sites (n)

46

44

20

Binned sites (n)

28

27

13

Binned dates (n)

58

55

25

Iron metallurgy

Samples (n)

89

18

83

Sites (n)

50

13

63

Binned sites (n)

35

6

40

Binned dates (n)

52

7

60

13To assess whether the SPDs of 14C dates showed statistically relevant fluctuations, the SPDs were compared with an exponential model (based on 10,000 Monte Carlo simulation) used as a conservative null hypothesis reflecting long-term prehistoric human population growth and taphonomic losses, as already proposed elsewhere (Crema et al. 2016; Shennan et al. 2013; Timpson et al. 2014). This method resolves numerous biases related to sampling error and fluctuations in the 14C calibration curve. Significant deviations (negative or positive) from the SPDs reflect changes in the population dynamics that were significantly stronger than in other periods, although these deviations are not interpreted in terms of absolute demographic change. To examine population changes in different geographic sub-regions, a nonparametric extension of this method was performed, specifically a permutation test, allowing the statistical comparison of two or more sets of 14C dates (Crema et al. 2016). In this method the null hypothesis is that the SPDs are equal in the regions compared if the sample dates are generated from identically shaped population curves (Fig. 6 and Table 2). Three regions were defined (North Coast, South Coast and Hinterland) based on their geographical coordinates and using a hierarchical clustering analysis in Euclidean space calculated with Ward’s minimum variance (Table 1, Fig. 5 and R code in the Supplementary Materials). Had further regions been defined, the number of dates per region would have dropped significantly (values before and after binning are North Coast: 526–215, Hinterland: 241–137, South Coast: 372–226, respectively). This consequent reducing may hamper the statistical significance of 14C dataset. Timpson et al. (2014) however showed that the method remains robust even for small sample sizes, although it may not be fully suited to detecting rapid demographic processes at the scale of a few hundred years with reduced datasets. Supplementary Materials including Supplementary Tables and Figures, the archaeological database and the R codes used in this study are publicly available in the online repository DataSuds (https://doi.org/​10.23708/​XB8LHG).

Results

SPD deviation of archaeological remains through time

14In WCA, the SPD deviation of archaeological finds increased from 2800 cal. yr BP until around 1500 cal. yr BP when the curve suddenly fell, confirming the empirical observations of Oslisly et al. (2013) (Fig. 2B). Based on the SPDs of calibrated 14C dates for the entire WCA dataset (at 95% confidence interval), a significant change in the archaeological record is inferred, highlighting a major positive deviation in the SPD curve from 2500 to 1500 cal. yr BP. This positive deviation is marked by two distinct increases (Fig. 2B) with maximum values at 2300 cal. yr BP (interval ~2500–2000 cal. yr BP) and 1700 cal. yr BP (2000–1500 cal. yr BP).

Figure 2 – Archaeological synthesis of Western Central Africa during the past 5000 years.

Figure 2 – Archaeological synthesis of Western Central Africa during the past 5000 years.

All archaeological data presented rely on the SPDs of calibrated 14C dates that were binned in space using 10-km radii and time using 200-yr intervals. SPDs were plotted with a 200-yr moving average to prevent over-interpretation of smaller scale variability.
(A) Evidence of human activity in the WCA inferred from the occurrence of remains of millet (Pennisetum glaucum), palm oil (Elaeis guineensis), iron metallurgy and pit features in archaeological context.
(B) SPD-inferred population dynamics: SPD of 14C dates (thick line) was compared against an exponential model used as a conservative null hypothesis (see text for explanation). The dark grey area represents the 95% confidence interval for the null model. Red and blue areas represent intervals with significant positive and negative deviations from the exponential model, respectively. Regional cultural timeline (2) shown at the bottom (LSA–Late Stone Age; Neolithic S.–Neolithic Stage; EIA–Early Iron Age; LIA–Late Iron Age).
(A) and (B) are adapted from (14). The blue and red vertical lines indicate the limit of the Late Holocene Rainforest Crisis (LHRC) at Lake Barombi (SW Cameroon).

15Conversely, between 1200 and 400 cal. yr BP, the SPD curve shows clearly a global negative deviation, which has been documented in all archaeological contexts as shown in previous studies (Oslisly et al. 2013; Saulieu et al. 2017). Since the dataset has been binned to avoid over- and under-representation, and since this decrease is documented across the whole region, at whatever scale of analysis, sampling bias can be ruled out.

16However, it is important to note that the decrease of the values initiated earlier, from 1500 up to 1200 cal. yr BP. After this decline, a stabilization of the curve occurred between 1200 and 800 cal. yr BP. Finally, values recovered from 800 to 400 cal. yr BP without reaching the exponential null model.

Archaeobotanical finds and changes in cultivation practices

17Throughout the period considered, the archaeological contexts remained homogeneous, consisting mostly of pit features (Fig. 2A). These features are about one metre in diameter and more than a metre depth and are generally filled with black earth containing charcoal and calcined fruits, sherds, stones and sometimes iron tools. The purpose of these features is still debated (Eggert et al. 2006; Mbida 2003; Mbida & Mvondo Ze 2016; Saulieu et al. 2017), but there are several lines of evidence that pits are associated with domestic sites, and likely correspond to refuse chutes (Oslisly et al. 2013; Saulieu et al. 2017).

18While pottery and pits in the domestic context in WCA first appeared around 2800 cal. yr BP (Figs. 2 and 3), recurrent archaeological elements—such as traces of Pennisetum glaucum, Elaeis guineensis nuts and iron metallurgy—arose simultaneously in different sites around ~2500 cal. yr BP and then increased sharply afterwards (Fig. 2A). Iron metallurgy is probably older in other parts of Africa (Clist 2012), notably in Nigeria (Eggert 2014). The expansion seen here therefore appears to be related to its initial introduction into the region.

19The occurrence of cultigens in WCA reveals different chronological patterns. Remains of both Pennisetum glaucum (charred grains) and Elaeis guineensis (charred nuts) appeared during the first part of the positive deviation (~2500–2000 cal. yr BP). Elaeis guineensis, a sun-loving tree, is fundamental to the diet in Central Africa for the proteins it provides and the fermented drink made from it. But the origin of its use is problematic, exactly the same way as for the other important tree in the region, Canarium schweinfurthii, present in archaeological contexts from the early Holocene. In archaeology, it is difficult to prove that a plant that is not domesticated (in the genetic sense of the term) was really cultivated. The remains of its charred endocarp that frequently appear in sites from the second half of the Holocene are therefore not necessarily proof of true domestication. Thus, there is a long-standing debate between those who interpret the visible variations of Elaeis in pollen diagrams, such as that of Lake Barombi (Maley & Brenac 1998), as natural phenomena, notably climatic (idem, Maley & Chepstow-Lusty 2001), and those who maintain that this tree was rationally exploited and disseminated by humans and was even a marker of the anthropization of landscapes (Warnier 1984; Sowunmi 1998). The statistical results presented here do not allow the debate to be resolved, but combined with the other trends, they could show how landscapes seem to be increasingly anthropised.

20Rare Pennisetum glaucum remains, a cereal domesticated in northern Mali and Mauritania about 4900 years ago (Burgarella et al. 2018), are only present for a short time period prior disappearance (Kahlheber et al. 2014). Noteworthy and as expected, its occurrence in WCA postdates the one in central Nigeria (the Nok Culture) by the first half of the first millennium BC (Champion & Fuller 2018). Unfortunately, instances are too scattered and scarce to retrace the timing of its spread in WCA. This low occurrence of Pennisetum glaucum remains can be partly attributed to the lack of studies using dedicated isolation methods (e.g. flotation method) (Fig. 2A) (Wotzka 2019). Moreover, the populations of the Congo Basin, although theoretically cultivators, only sporadically consumed this cereal, and in a very variable way depending on the region (Bleasdale et al. 2020). Despite its low occurrence, the use of this non-local cereal remains an important cultural fact.

Timing of archaeological occurrence and cultural periods

21Twenty-one well-described ceramic traditions were compared using SPDs, for which more than five reliable dates were available (Fig. 3). These have been chosen from the contexts that seem most representative (no apparent major bioturbation, homogeneous archaeological material) and have all been binned in space and time in the same way than the complete radiocarbon dataset (see p. 14). We chose to use SPDs for cultural chronology in coherence with the method chosen for our population modelling. The positive SPD deviation curve corresponds with a specific cultural period. The first half of the positive deviation, between ~2500 and ~2000 cal. yr BP, takes place within a ceramic horizon, which lasts from ~2800 to ~2000 cal. yr BP (Figs. 3 and 4). The ceramic styles predominant in this first period share numerous formal and decorative peculiarities present at a large regional scale. This period has different names in the literature: Neolithic or Neolithic Stage (Denbow 1990; Neumann et al. 2012; Oslisly et al. 2013), Stone to Metal Age in de Maret (1982, 1994) or Ceramic Later Stone Age (Denbow 2012, 2013). To avoid entering into a terminology debate that is beyond the aims of this article, the period is referred to here, for convenience, as Neolithic.

Figure 3 – Pottery traditions and main Periods of Western Central Africa.

Figure 3 – Pottery traditions and main Periods of Western Central Africa.

(A) Data shown in time. All archaeological data presented rely on the SPDs of calibrated 14C dates that were binned in space using 10-km radii and time using 200-yr intervals. SPDs were plotted with a 200-yr moving average to prevent over-interpretation of smaller scale variability and are shown as a greyscale (black represents maximum probability; white represents null probability). Coloured dots refer to sites on the map in (B). Regional cultural timeline shown at bottom (LSA–Late Stone Age; Neolithic S.–Neolithic Stage; EIA–Early Iron Age; LIA–Late Iron Age). The blue and red vertical lines indicate the limit of the Late Holocene Rainforest Crisis (LHRC) at Lake Barombi (SW Cameroon).
(B) Location and age of sample sites. Larger dots represent older samples and smaller dots represent younger samples. Colours refer to pottery traditions highlighted in (A).

22Despite some strong similarities, at least six distinct ceramic styles (Dibamba E, Malongo, Obobogo, Epona, Okala, Nya Zanga) occurred between ~2500 and ~2000 cal. yr BP. First described in Obobogo (Yaoundé, Cameroon; Fig. 3), ceramics are characterised by simple shapes: ovoid or spherical short-necked jars, usually with flat bases (Oslisly et al. 2013; Saulieu et al. 2017). Techniques, decorative arrangements and decorative motifs are dominated by comb or shell impressions (from vertical and/or rocker stamping) and shallow transverse linear tracings. In general, the decoration is not very neat, and the firing atmosphere generally seems to have been conducive to oxidizing. The lips of containers are often fluted (Fig. 4, p-s). Various coeval ceramic styles in the Republic of Congo (Denbow’s Ceramic Later Stone Age 1 & 2, Denbow 2013) and in the Democratic Republic of Congo (Imbonga, in Wotzka 1995) share occasional similarities and seem to be part of a common cultural area (Fig. 4, v & w).

Figure 4 – Chronology of Western Central Africa with Pottery traditions.

Figure 4 – Chronology of Western Central Africa with Pottery traditions.

Late Iron Age: (a, b) Dibamba A–Cameroon; (c) Lopé region from 11th to 18th centuries–Gabon, (d, e) Dibamba B–Cameroon, (f) Oku–Cameroon, (g) Dibamba C–Cameroon, (h, i) Lobéké–Cameroon, (j) Kovifem–Cameroon.
Early Iron Age: (k, l) Campo–Cameroon, (m) Akom near Mintom–Cameroon, (n) Lalara–Gabon, (o) Otoumbi–Gabon.
Neolithic: (p) Dibamba E–Cameroon, (q) Mpolongwé–Cameroon, (r, s) Nya Zanga–Cameroon, (t, u) Epona–Gabon, (v) Imbonga–Democratic Republic of Congo (w), CLSA from Madingo-Kayes–Republic of Congo.

© J. Denbow (w), © R. Oslisly (c, q, o, n, t, u), © G. de Saulieu (a, b, d-m, p, r, s), and © H.-P. Wotzka (v).

  • 2 Positive deviation: when the statistical curve is greater than the null model. The deviation is neg (...)

23The second half of the positive deviation2, occurring between ~2000 and ~1500 cal. yr BP, coincides with what some authors call the Early Iron Age. This period is characterised by a considerable amount of iron artefacts in archaeological contexts along the coastal regions (Table 3). Yet iron artefacts in the eastern regions of WCA are rare and of later date (Fig. S1C and Table 3) (Wotzka 1995, 2006). Regardless, the ceramic styles between ~2000 and ~1500 cal. yr BP (Fig. 4, k-o) are diverse but they share numerous similarities (Saulieu et al. 2017). They comprise flat-bottomed vessels and containers with composite shapes. These usually high-necked, high-shouldered pots, sometimes carinated, and various other containers such as bowls with direct or composite walls, are characterised by a particularly rich decoration, reminiscent of the earlier Imbonga ceramic style, made with combs (sometimes broadly grooved) and sticks. The patterns (parallel lines, hatching, dots, herringbone) are very dense and often deeply imprinted in the clay. In some areas appliqué was used. The decorations cover the whole exterior of the jar and, often, also the inner surface of its flared edges, always with great care. Thus, ceramic styles from this second period (Figs. 3, 4) share many formal and decorative traits at a wider regional scale.

Spatial occurrence of archaeological remains

24At the scale of the studied region (1500 km by 1500 km), the dates of the Neolithic ceramic styles (Fig. 3) do not seem to be distributed in a clear spatiotemporal gradient that would be indicative of diffusion processes. Indeed, in the case of a diffusion from a homeland, we may expect to have a well-delimited region characterized by the oldest dates, with younger and younger dates found when moving away from this centre. However, in order to avoid subjective bias in the search for a possible spatial dynamic, a clustering analysis of geographical coordinates was undertaken. The dated items were split into the three distinct regions already mentioned, namely ‘North Coast’, ‘South Coast’, and ‘Hinterland’ (Fig. 5). Comparisons using regional SPDs for each region (Fig. 5) and permutation tests of SPDs between regions (Fig. 6) allow two observations.

Figure 5 – Regional variability in population dynamics of Western Central Africa.

Figure 5 – Regional variability in population dynamics of Western Central Africa.

(A) Map of Western Central Africa. Dots indicate archaeological site locations and colours delineate the three regions used to estimate population changes.
(B) Definition of the three regions based on a hierarchical clustering analysis in Euclidean space.
(C) SPD of the three regions (thin line) against an exponential model. The thick lines show the 200-yr rolling mean and the grey area represents the 95% confidence interval for the model. Red and blue vertical bands represent periods with significant positive and negative deviations, respectively.

25Firstly, the permutations demonstrate that the North and South Coast regions are statistically indistinguishable for both cultural periods: the positive deviations are synchronous with the inceptions (~2500 cal. yr BP), terminations of significant archaeological occurrences (1500 cal. yr BP) and present similar general shapes. This suggests that the North and South Coast actually formed a single homogeneous cultural region between ~2500 and ~1500 cal. yr BP.

26Secondly, the Hinterland shows a different pattern. The 14C SPD curve does not show the typical bimodal distribution of both coastal areas. On the contrary, it shows a single principal peak whose inception shows a slight lag of ~170 yr compared with the coastal regions, which, however, is not statistically significant. This peak is stronger and lasts longer (Fig. 6), overlapping the second peak identified in the coastal zones. Finally, the 14C SPD curve dips drastically at ~1300 cal. yr BP, ~200 years later than for the coastal regions (Figs. 5, 6). These identified differences between the coastal regions and the Hinterland (discernible from metallurgy, the use of Elaeis guineensis and from the shapes of regional 14C SPD curves, see Fig. 5C and Fig. S1) occurred during the Early Iron Age.

Figure 6 – Comparison of SPDs of the three regions based on permutation tests.

Figure 6 – Comparison of SPDs of the three regions based on permutation tests.

Each row shows an observed SPD (black thick line with its model in grey) of a region compared against another in the column. Red and blue vertical bands indicate periods with significant positive and negative deviations, respectively, from the model of the combined set of each of pair.

Discussion

Evidence for a possible population growth between ~2500 and ~1500 cal. yr BP

27The SPD curves (Fig. 2) highlight one positive deviation between ~2500 and ~1500 cal. yr, followed by a negative deviation between ~1500 and ~ 400 cal. yr BP. This study, therefore, suggests a significant population growth (between ~2500 and ~1500 cal. yr BP) followed by a widespread demographic decline (between ~1200 and ~400 cal. yr BP) in WCA.

28Aside from the main trends in population dynamics identified, there are also regional differences. The first part of the population growth (~2500–2000 cal. yr BP) appears marked in the two coastal regions, while the second part (~ 2000–1500 cal. yr BP), affecting all regions, is more prominent in the Hinterland (Fig. 5).

Cultural and technical changes tied to the population growth

29The population growth also seems to correspond with cultural and technical changes, which can be summarized in three points:

  1. The first part of the population growth (~2500–2000 cal. yr BP), present only in the coastal regions, happens during the second half of the Neolithic period (~2800–2000 cal. yr BP) and coincides with the appearance of the cultigen / iron metallurgy package.

  2. The second part of the population growth (~2000–1500 cal. yr BP) corresponds with the Early Iron Age in all three regions. It is worth highlighting that this period is characterised by a general change in ceramic styles in the two coastal regions.

  3. In the coastal regions, the two parts of the population growth are associated with the cultural periods of the Neolithic and the Early Iron Age respectively. Conversely, in the Hinterland, the main population increase occurred during the Early Iron Age. These observations are consistent with archaeological data from the Democratic Republic of Congo showing the important diffusion of related ceramic styles eastwards along the tributaries of the Congo River and suggesting a colonization movement (Wotzka 1995,  2006).

Unresolved question of the environmental impact of the population growth

30The first part of the population growth (~2500–2000 cal. yr BP) seems coeval with environmental and vegetation changes recorded in sedimentary archives from WCA, at least at the local scale. Indeed, various palynological records show that roughly between 3000 and 2000 cal. yr BP, rainforests were replaced in many places by a forest-savannah mosaic (e.g. Vincens et al. 1999). The timing of this vegetation disturbance, referred to as the Late Holocene Rainforest Crisis (LHRC), is poorly constrained, mostly because of the low-scale time resolution of the sedimentary archives that were investigated and the presence of sedimentary hiatuses (Bonnefille 2011). A recent study (Garcin et al. 2018) confirms a prominent and abrupt appearance of C4 plants (i.e. some light demanding plants such as poaceae relying on the Hatch–Slack (C4) carbon fixation pathway) in the Lake Barombi catchment, at ~2600 cal. yr BP, followed by an equally sudden return to rainforest vegetation at ~2020 cal. yr BP. Based on changes in carbon and hydrogen isotope compositions of plant waxes, it also shows that there was no simultaneous hydrological change during this event. This local and well-dated record suggests the concomitance between the LHRC and the first part of the SPDs positive deviation interpreted as a regional population growth (~2500–2000 cal. yr BP). However, the question of the spatial scale of this impact remains open. Kiahtipes (2019) outlined that local human disturbances near the Atlantic coast might have amplified a long-term forest fragmentation during the Late Holocene, when the climate was drier than in the earlier Holocene.

The Bantu expansion issue

31It is questionable whether these results can be straightforwardly integrated with current data on genetics and historical linguistics, notably with the spread of Bantu languages. This expansion is generally assumed to have taken place between 5000 and 2000 cal. yr BP (Bostoen 2018; Bostoen et al. 2015; de Maret 2013; Vansina 1984, 1990). The population growth highlighted from ~2500 cal. yr BP onward is difficult to link with the expansion of languages in Central Africa for two main reasons. The first is that there is no ancient DNA for the populations concerned by this demographic phenomenon. The available DNA data (Lipson et al. 2020) are from the Shum Laka site located in the grassland of Western Cameroon, the supposed homeland of the Bantu languages and concern the immediately preceding period (8000–3000 cal. yr BP). Genome-wide DNA analyses clearly indicate that there is no direct link between the current Bantu populations and the four individuals recovered in the Shum Laka rockshelter (Ribot et al. 2001; Lavachery 2001). The second reason is that the statistical analysis of the archaeological data does not prove that this possible population growth was accompanied by a large migratory movement throughout the region. Indeed, the diffusion of the cultigen/iron metallurgy package co-occurring with the archaeological contexts dominated by refuse pits and accompanied by a population growth (from ~2500 to 1500 cal. yr BP) is very similar in the coastal regions. Here, it is difficult to argue in favour of diffusion because there is not one central region where all these phenomena appear first and that could play the role of a homeland. This can be explained in several ways. The first possibility is that the method of analysis is not precise enough to identify diffusion. It should be remembered that comparison of regional curves for periods of less than two hundred years (the threshold used to avoid chronological bias), is not statistically significant. Even if we could detect diffusion over such a time scale, it would not solve the interpretation problem since there could have been demic (i.e. resulting from population displacements) or non-demic (without population displacement) dissemination of cultural traits or techniques (or both). A demic diffusion from the North Coast to the South Coast region would imply the displacement of sedentary populations over 1000 km in less than 200 years. Knowing the quantity of archaeological remains involved, and taking seriously into account the distances (without a navigable rivers), this does not seem to be a realistic hypothesis: the shifts would be detectable. Conversely, the dissemination of technical practices or cultural traits over such distances in less than 200 years without population displacement is conceivable under certain conditions. Both social anthropology (e.g. Kopytoff 1987) or analytical sociology (e.g. Hedström 2006) can provide models or examples of transmission along social networks. However, these are currently impossible to demonstrate based on available archaeological evidence. Nonetheless, if this hypothesis is retained, then it follows that the social networks that enabled the rapid diffusion of these technical practices and/or cultural traits were already in place and mature before 2500 cal. yr BP.

Population decline between ~1200 and ~ 400 cal. yr BP

32The negative deviation of the curve (1200–400 cal. yr BP) corresponding to a possible global decline in occupation levels is widespread and follows a millennium long trend of population growth and probable technologically-driven modification of the environment through, for example, iron metallurgy. This regional decrease may not be clearly reflected in all local archaeological records. Even if in some regions human settlements seem to be maintained (Lupo et al. 2018; Oslisly et al. 2013), there seems to be a gap in the cultural chronologies of the Lopé region of central Gabon (Oslisly 1993; Oslisly 2001; Assoko Ndong 2002) and the Dibamba site on the Cameroonian coast (Saulieu et al. 2017).

33Changes in population dynamics in both coastal regions of WCA and in the Hinterland were asynchronous (Fig. 5). The permutation test (Fig. 6) indicates that during the first population growth the Hinterland response lagged behind the coastal regions by ~170 years (which is not statistically significant), and the population decline arrived ~200 years later. This confirms the particular status of the Hinterland region, in contrast to the two coastal regions. The latter are actually very similar as demonstrated by the permutation test.

34It must also be emphasised that the negative deviation of the curve observed between ~1200 and 400 cal. yr BP is coeval with the proposed substantial gene admixture between the rainforest hunter-gatherers (‘pygmies’) and the ancestors of the modern Bantu-speaking peoples, estimated between 1000 and 800 cal. yr BP (Patin et al. 2014, 2017). This raises a multitude of questions regarding when these different populations settled in Central Africa, as well as when and how the very particular relations that they maintain today came about. The well-known complementarity between these different populations (Bahuchet 1991; Testart 1982) could therefore have been established at that time, attributing to the pygmies the activities related to the deep forest.

35The cause of this population decline is unknown. A sanitary or epidemic regional crisis could be envisaged, similar to the hypothesis of the second plague pandemic impacting societies in the Ife political sphere of influence, and many other urban sites in West and East Africa starting in the 14th century (Chouin 2013, 2018; Green 2018). The demographic decline should probably be seen as the result of multiple rather than single factors, as it is deep and long term. What initiated the decline and what prolonged it? The chronological framework we present coincides with the chronology of the ‘First plague pandemic’ (540–850 CE) as already mentioned in Saulieu et al. (2017). This hypothesis seems to be unsubstantiated for the moment. Moreover, as the demographic crisis is very long, it is not impossible that this event is one of the other factors in the depopulation, particularly environmental ones (e.g. Sebag et al. 2016). The environmental transformations triggered by these sedentary human communities could have shifted epidemiological boundaries. We could also make the hypothesis that human communities could have adopted new practices that brought them into contact with particularly virulent pathogenic vectors. However, the insufficient accuracy of the statistical method and the absence of human remains do not allow further insight.

Conclusion

36The statistical analysis of summed probability distributions of 14C dates from archaeological data in Central Africa indicates a population growth between ~2500 and ~1500 cal. yr BP. The concomitant increase of refuse pits, Elaeis guineensis and iron metallurgy along with rare remains of Pennisetum glaucum took place during the second half of the Neolithic, beginning around 2800 cal. yr BP. Regional differences were demonstrated by splitting the data into to three geographical clusters (‘North coast’, ‘South coast’ and ‘Hinterland’). In the coastal regions, the population growth appears in two successive parts during the Neolithic and the Early Iron Age (~2500–2000 cal. yr BP and ~2000–1500 cal. yr BP) while in the Hinterland the shape and timing of the curve is slightly different (~2400–1300 cal. yr BP). However, it is not possible to identify a common diffusion phenomenon from a homeland. While the specificities of the SPD curve for the Hinterland region can probably be explained by colonization in the Congo Basin (Wotzka 1995, 2006), the two coastal regions evolve in an identical way, to the extent that it seems unlikely that the archaeological proxies and the SPD curve are here indicators of a population displacement. On the contrary, it suggests a wide-ranging interaction network, already in place when new innovations arrived around 2500 cal. yr BP, causing population growth. A decline in settlements at the regional level after ~1500 cal. yr BP is also evident and detailed at a sub-regional scale. This decline can be compared with the long-term local archaeological chronology observed at the Lopé (Oslisly 1993, 2001; Assoko Ndong 2002) or Dibamba (Saulieu et al. 2017) sites. However, the causes are yet to be elucidated.

37This study establishes for the first time a plausible picture of the population dynamics and the evolution of cultural traits and technical innovation in WCA over the last five millennia, based on a comprehensive and statistically-binned archaeological database. These results provide a robust archaeological timeframe that may contribute to the refinement of phylogenetic and linguistic models in the area. Together with ancient DNA data (Lipson et al. 2020), this study aims to contribute also to the debates regarding the cultural and linguistic history of Central African populations as well as the long-term changes in land use that affected this region during the Late Holocene.

Acknowledgments

38We acknowledge the French National Research Institute for Sustainable Development (IRD) and the LMI DYCOFAC for their support. This study was supported by the ANR project TAPIOCA. Many 14C dates were carried out by the National Platform LMC14 (LSCE (CNRS-CEA-UVSQ)-IRD-IRSN-MC) with the support of IRD and ANPN.

39Author contributions: G.S., Y.G., D.S. and R.O. conceptualized the paper. R.O., G.S., Y.G., and P.N. built the archaeological dataset. Y.G., D.S. and G.S. performed analyses. All authors assisted with data interpretation and paper preparation. D.Z. and P.C. provided feedback on the concepts, data interpretation and paper preparation. G.S., D.S. and Y.G. wrote the paper.

Haut de page

Bibliographie

Ammerman A.J. & Cavalli-Sforza L.L. (1971) – Measuring the Rate of Spread of Early Farming in Europe. Man, 6(4): 674–688.

Ammerman A.J. & Cavalli-Sforza L.L. (1979) – The Wave of Advance Model for the Spread of Agriculture in Europe. In: C. Renfrew & K.L. Cooke (eds), Transformations. Mathematical Approaches to Culture Change, Elsevier: 275–293. https://doi.org/10.1016/B978-0-12-586050-5.50023-3

Assoko Ndong A. (2002) – Synthèse des données archéologiques récentes sur le peuplement à l’Holocène de la réserve de faune de la Lopé, Gabon. L’Anthropologie, 106 : 135-158.

Bahuchet S. (1991) – Les Pygmées d’aujourd’hui en Afrique centrale. Journal des Africanistes, 61 : 5-35.

Bayon G., Dennielou B., Etoubleau J., Ponzevera E., Toucanne S. & Bermell S. (2012) – Intensifying weathering and land use in Iron Age Central Africa. Science, 335: 1219–1222. DOI: 10.1126/science.1215400

Berniell-Lee G., Calafell F., Bosch E., Heyer E., Sica L., Mouguiama-Daouda P., Van der Veen L., Hombert J.-M., Quintana-Murci L. & Comas D. (2009) – Genetic and demographic implications of the Bantu expansion: insights from human paternal lineages. Molecular biology and evolution, 26: 1581–1589. DOI: 10.1093/molbev/msp069

Bleasdale M., Wotzka H.-P., Eichhorn B., Mercader J., Styring A., Zech J., Soto M., Inwood J., Clarke S. & Marzo S. (2020) – Isotopic and microbotanical insights into Iron Age agricultural reliance in the Central African rainforest. Communications biology, 3: 1–10. https://doi.org/10.1038/s42003-020-01324-2

Bocquet-Appel J.-P., Naji S., Linden M.V. & Kozlowski J.K. (2009) – Detection of diffusion and contact zones of early farming in Europe from the space-time distribution of 14C dates. Journal of Archaeological Science, 36: 807–820. https://doi.org/10.1016/j.jas.2008.11.004

Bonnefille R. (2011) – Rainforest responses to past climatic changes in tropical Africa. In: M. Bush & W. Gosling (ed.), Tropical Rainforest Responses to Climatic Change. Heidelberg, Springer-Verlag Berlin: 125–184. DOI: 10.1007/978-3-642-05383-2

Bostoen K. (2018) – The Bantu Expansion. In: Oxford Research Encyclopedia of African History. Oxford, Oxford University Press. https://doi.org/10.1093/acrefore/9780190277734.013.191

Bostoen K., Clist B., Doumenge C., Grollemund R., Hombert J.-M., Muluwa J.K. & Maley J. (2015) – Middle to Late Holocene Paleoclimatic Change and the Early Bantu Expansion in the Rain Forests of Western Central Africa. Current Anthropology, 56: 354–384. https://doi.org/10.1086/681436

Burgarella C., Cubry P., Kane N.A., Varshney R.K., Mariac C., Liu X., Shi C., Thudi M., Couderc M. & Xu X (2018) – A western Sahara centre of domestication inferred from pearl millet genomes. Nature ecology & evolution, 2: 1377-1380. https://doi.org/10.1038/s41559-018-0643-y

Champion L. & Fuller D.Q. (2018) – New evidence on the development of millet and rice economies in the Niger river basin: archaeobotanical results from Benin. In: A.M. Mercuri, A.C. D’Andrea, R. Fornaciari & A. Höhn (eds), Plants and People in the African Past. Springer International Publishing, Cham: 529–547. https://doi.org/10.1007/978-3-319-89839-1_23

Chouin G. (2013) – Fossés, enceintes et peste noire en Afrique de l’Ouest forestière (500-1500 AD). Réflexions sous canopée. Afrique : Archéologie & Arts, 9 : 43-66. https://doi.org/10.4000/aaa.284

Chouin G. (2018) – Reflections on plague in African history (14th–19th c.). Afriques. Débats, méthodes et terrains d’histoire, 9. https://doi.org/10.4000/afriques.2228

Clist B. (2012) – Vers une réduction des préjugés et la fonte des antagonismes : un bilan de l’expansion de la métallurgie du fer en Afrique sub-saharienne. Journal of African Archaeology, 10: 71–84. https://doi.org/10.3213/2191-5784-10205

Clist B., Bostoen K., de Maret P., Eggert M.K., Höhn A., Mindzié C.M., Neumann K. & Seidensticker D. (2018) – Did human activity really trigger the late Holocene rainforest crisis in Central Africa? Proceedings of the National Academy of Sciences, 115: E4733–E4734. https://doi.org/10.1073/pnas.1805247115

Contreras D.A. & Meadows J. (2014) – Summed radiocarbon calibrations as a population proxy: a critical evaluation using a realistic simulation approach. Journal of Archaeological Science, 52: 591–608. https ://doi.org/10.1016/j.jas.2014.05.030

Crema E.R. & Bevan A. (2021) – Inference from large sets of radiocarbon dates: Software and methods. Radiocarbon, 63(1): 23–39. DOI: https://doi.org/10.1017/RDC.2020.95

Crema E.R., Habu J., Kobayashi K. & Madella M. (2016) – Summed Probability Distribution of 14C Dates Suggests Regional Divergences in the Population Dynamics of the Jomon Period in Eastern Japan. PLoS ONE, 11: e0154809. https://doi.org/10.1371/journal.pone.0154809

De Filippo C., Bostoen K., Stoneking M. & Pakendorf B. (2012) –Bringing together linguistic and genetic evidence to test the Bantu expansion. Philosophical Transactions of the Royal Society B, 279: 3256–3263. https://doi.org/10.1098/rspb.2012.0318

De Maret P. (1982) – From the Stone Age to the Iron Age: the “Neolithic” problem in the West and South. In: F. van Noten (ed.), The Archeology of Central Africa, Akademische Drück- und Verlagsantstalt, Graz: 59–65.

De Maret P. (1994) – Pits, pots and the far-west streams. Azania: Archaeological Research in Africa, 29: 316–323. https://doi.org/10.1080/00672709409511690

De Maret P. (2013) – The Archaeologies of the Bantu expansion. In: P. Mitchell & P. Lane (eds), The Oxford Handbook of African Archaeology. Oxford, Oxford University Press: 319–328. DOI: 10.1093/oxfordhb/9780199569885.013.0043

Denbow J. (1990) – Congo to Kalahari: data and hypotheses about the political economy of the western stream of the Early Iron Age. African Archaeological Review, 8: 139–175.

Denbow J. (2012) – Pride, prejudice, plunder and preservation: archaeology and the re-envisioning of ethnogenesis on the Loango coast of the Republic of Congo. Antiquity, 86(332): 383–408. DOI: https://doi.org/10.1017/S0003598X00062839

Denbow J. (2013) – The archaeology and ethnography of Central Africa. New York, Cambridge University Press. https://doi.org/10.1017/CBO9781139629263

Dubouloz J. (2017) – Modélisation et simulation de la colonisation néolithique de l’Europe tempérée par la culture à céramique linéaire. In : D. Garcia & H. Le Bras (dir.), Archéologie des migrations, Paris, La Découverte : 111-124. https://doi.org/10.3917/dec.garci.2017.01.0111

Eggert M. (2014) – Early Iron in Wess and Central Africa. In: P. Breunig (ed.), Nok. African sculpture in archaeological context, Franckfurt, Goeth-Universität Franckfurt and Africa Magn Verlag: 51-59.

Eggert M.K., Höhn A., Kahlheber S., Meister C., Neumann K. & Schweizer A. (2006) – Pits, graves and grains: archaeological and archaeobotanical research in southern Cameroun. Journal of African Archaeology, 4(2): 273–298. DOI: 10.3213/1612-1651-10076

Eggert M.K. & Seidensticker D. (2016) – Campo: archaeological research at the mouth of the Ntem River (south Cameroon). Africa Praehistorica 31, Köln, Heinrich-Barth-Institut.

Garcin Y., Deschamps P., Ménot G., Saulieu G. de, Schefuß E., Sebag D., Dupont L.M., Oslisly R., Brademann B., Mbusnum K.G., Onana J.-M., Ako A.A., Epp L.S., Tjallingii R., Strecker M.R., Brauer A. & Sachse D. (2018) – Early anthropogenic impact on Western Central African rainforests 2,600 y ago. Proceedings of the National Academy of Sciences 115 (13): 3261-3266. https://doi.org/10.1073/pnas.1715336115

Giresse P., Maley J., Doumenge C., Philippon N., Mahé G., Chepstow-Lusty A., Aleman J., Lokonda M. & Elenga H. (2018) Paleoclimatic changes are the most probable causes of the rainforest crises 2,600 y ago in Central Africa, Proceedings of the National Academy of Sciences: 115-29: E6672-E6673. https://doi.org/10.1073/pnas.1807615115

Goldberg A., Mychajliw A.M. & Hadly E.A. (2016) – Post-invasion demography of prehistoric humans in South America. Nature, 532: 232–235. https://doi.org/10.1038/nature17176

Green M.H. (2018) – Putting Africa on the Black Death map: Narratives from genetics and history. Afriques. Débats, méthodes et terrains d’histoire, 9. DOI : https://doi.org/10.4000/afriques.2125

Grollemund R., Branford S., Bostoen K., Meade A., Venditti C. & Pagel M. (2015) – Bantu expansion shows that habitat alters the route and pace of human dispersals. Proceedings of the National Academy of Sciences, 112: 13296–13301. https://doi.org/10.1073/pnas.1503793112

Hedström P. (2006) – Explaining social change: An analytical approach. Papers: revista de sociologia, 80: 73–95. DOI: 10.5565/rev/papers/v80n0.1770

Hogg A.G., Heaton T.J., Hua Q., Palmer J.G., Turney C.S.M., Southon J., Bayliss A., Blackwell P.G., Boswijk G., Bronk Ramsey C., Pearson C., Petchey F., Reimer P., Reimer R. & Wacker L. (2020) – SHCal20 Southern Hemisphere Calibration, 0–55,000 Years cal BP. Radiocarbon, 62: 759–778. DOI: https://doi.org/10.1017/RDC.2020.59

Kahlheber S., Eggert M.K., Seidensticker D. & Wotzka H.-P. (2014) – Pearl millet and other plant remains from the Early Iron Age site of Boso-Njafo (inner Congo Basin, Democratic Republic of the Congo). African Archaeological Review, 31: 479–512. DOI: https://doi.org/10.1007/s10437-014-9168-1

Kiahtipes C.A. (2019) – Anthropogenic impacts in Africa’s rain forests: State of the art. In: B. Eichhorn & A. Höhn (eds), Trees, Grasses and Crops. People and Plants in Sub-Saharan Africa and Beyond. Frankfurter Archäologische Schriften, Frankfurt Archaeological Studies 37, Bonn, Verlag Dr. Rudolf Habelt GmbH: 257–269.

Kopytoff I. (1987) – The African frontier: the reproduction of traditional African societies. Bloomington and Indianapolis: Indiana University Press.

Lavachery P. (2001) – The Holocene archaeological sequence of Shum Laka rock shelter (Grassfields, western Cameroon). African archaeological review, 18: 213–247. DOI: https://doi.org/10.1023/A:1013114008855

Lavachery P., MacEachern S., Bouimon T. & Mbida C. (2010) – De Komé à Kribi. Archéologie préventive le long de l’oléoduc Tchad-Cameroun, 1999-2004. Monograph Series 5. Frankfurt am Main, Africa Magna Verlag.

Lipson M., Ribot I., Mallick S., Rohland N., Olalde I., Adamski N., Broomandkhoshbacht N., Lawson A.M., López S. & Oppenheimer J. (2020) – Ancient West African foragers in the context of African population history. Nature, 577: 665–670. DOI: https://doi.org/10.1038/s41586-020-1929-1

Lupo K.D., Kiahtipes C.A., Schmitt D.N., Ndanga J.-P., Craig Young D. & Simiti B. (2018) – An elusive record exposed: radiocarbon chronology of late Holocene human settlement in the northern Congo Basin, southern Central African Republic. Azania: Archaeological Research in Africa, 53(2): 209–227. DOI: https://doi.org/10.1080/0067270X.2018.1471798

Maley J. & Brenac P. (1998) – Vegetation dynamics, palaeoenvironments and climatic changes in the forests of western cameroon during the last 28,000 years B.P., Review of Palaeobotany and Palynology, 99(2): 157–187. DOI: https://doi.org/10.1016/S0034-6667(97)00047-X

Maley J. & Chepstow-Lusty A. (2001) – Ealeis guineensis Jacq. (oil palm) fluctuations in central Africa during the late Holocene: climate or human driving foreces for this pioneering species? Vegetation History and Archeobotany, 10: 117–120.

Maley J., Giresse P., Doumenge C. & Favier C. (2012) – Comment on “Intensifying Weathering and Land Use in Iron Age Central Africa.” Science, 337: 1040–1040. https://doi.org/10.1126/science.1221820

Manning K. & Timpson A. (2014) – The demographic response to Holocene climate change in the Sahara. Quaternary Science Reviews, 101: 28–35.

Mbida C. (2003) – Essai d’interprétation spatiale des sites à fosses du Sud-Cameroun. In: A. Froment & J. Guffroy (dir.) Peuplements anciens et actuels des forêts tropicales, Colloques et Séminaires, Paris, IRD éd. : 103-112.

Mbida C., Mvondo Ze A. (2016) – Chemical analysis of soil samples from Campo-Eglise and Campo-Center. In: M. Eggert & D. Seidensticker (dir.), Campo: Archaeological Research at the Mouth of the Ntem River (South Cameroon), Köln, Heinrich-Barth-Institut: 147–150.

McCormac F.G., Hogg A.G., Blackwell P.G., Buck C.E., Higham T.F.G., Reimer P.J. (2004) – SHCal04 Southern Hemisphere calibration, 0-11.0 cal kyr BP. Radiocarbon, 46: 1087–1092. DOI: https://doi.org/10.1017/S0033822200033014

Neumann K., Bostoen K., Höhn A., Kahlheber S., Ngomanda A. & Tchiengué B. (2012) – First farmers in the Central African rainforest: A view from southern Cameroon. Quaternary International, 249: 53–62. DOI: https://doi.org/10.1016/j.quaint.2011.03.024

Neumann K., Eggert M.K.H., Oslisly R., Clist B., Denham T., de Maret P., Ozainne S., Hildebrand E., Bostoen K., Salzmann U., Schwartz D., Eichhorn B., Tchiengué B. & Höhn A. (2012) – Comment on “Intensifying Weathering and Land Use in Iron Age Central Africa.” Science, 337: 1040. DOI: https://doi.org/10.1126/science.1221747

Ngomanda A., Neumann K., Schweizer A., Maley J. (2009) – Seasonality change and the third millennium BP rainforest crisis in southern Cameroon (Central Africa). Quaternary Research, 71–3: 307-318. DOI: 10.1016/j.yqres.2008.12.002

Oh Y., Conte M., Kang S., Kim J. & Hwang J. (2017) – Population Fluctuation and the Adoption of Food Production in Prehistoric Korea: Using Radiocarbon Dates as a Proxy for Population Change. Radiocarbon, 59–6: 1761–1770. DOI:10.1017/RDC.2017.122

Oslisly R. (1993) – Préhistoire de la moyenne vallée de l’Ogooué, Gabon, Travaux et documents microfichés. Paris, ORSTOM éd.

Oslisly R. (2001) – The history of human settlement in the middle Ogooué valley (Gabon). In: W. Weber, A. Vedder & L.J.T. White, African rain forest ecology and conservation: 101–118.

Oslisly R., White L., Bentaleb I., Favier C., Fontugne M., Gillet J.-F. & Sebag D. (2013) – Climatic and cultural changes in the west Congo Basin forests over the past 5000 years. Philosophical Transactions of the Royal Society B, 368: 20120304. DOI:  http://dx.doi.org/10.1098/rstb.2012.0304

Parnell A. (2015) – Bchron: Radiocarbon dating, age-depth modelling, relative sea level rate estimation, and non-parametric phase modelling. R package version 4.7.5.

Patin E., Lopez M., Grollemund R., Verdu P., Harmant C., Quach H., Laval G., Perry G.H., Barreiro L.B., Froment A., Heyer E., Massougbodji A., Fortes-Lima C., Migot-Nabias F., Bellis G., Dugoujon J.-M., Pereira J.B., Fernandes V., Pereira L., Van der Veen L., Mouguiama-Daouda P., Bustamante C.D., Hombert J.-M. & Quintana-Murci L. (2017) – Dispersals and genetic adaptation of Bantu-speaking populations in Africa and North America. Science, 356: 543–546. https://doi.org/10.1126/science.aal1988

Patin E., Siddle K.J., Laval G., Quach H., Harmant C., Becker N., Froment A., Régnault B., Lemée L., Gravel S. (2014) – The impact of agricultural emergence on the genetic history of African rainforest hunter-gatherers and agriculturalists. Nature communications, 5: 3163. DOI: https://doi.org/10.1038/ncomms4163

Quintana-Murci L., Quach H., Harmant C., Luca F., Massonnet B., Patin E., Sica L., Mouguiama-Daouda P., Comas D., Tzur S. (2008) – Maternal traces of deep common ancestry and asymmetric gene flow between Pygmy hunter–gatherers and Bantu-speaking farmers. Proceedings of the National Academy of Sciences, 105: 1596–1601. DOI: 10.1073/pnas.0711467105

R Core Team (2017) – R: A language and environment for statistical computing. R Found. Stat. Comput. Vienna, Austria. URL http://www.R-project.org/., page R Foundation for Statistical Computing.

Reimer P., Austin W.E.N., Bard E., Bayliss A., Blackwell P.G., Bronk Ramsey C., Butzin M., Cheng H., Lawrence Edwards R., Friedrich M., Grootes P.M., Guilderson T.P., Hajdas I., Heaton T.J., Hogg A.G., Hughen K.A., Kromer B., Manning S.W., Muscheler R., Palmer J.G., Pearson C., van der Plicht J., Reimer R., Richards D.A., Scott E.M., Southon J.R., Turney C.S.M., Wacker L., Adolphi F., Büntgen U., Capano M., Fahrni S., Fogtmann-Schulz A., Friedrich R., Miyake F., Olsen J., Reinig F., Sakamoto M., Sookdeo A. & Talamo S. (2020) – The IntCal20 Northern Hemisphere radiocarbon age calibration curve (0–55 kcal BP). Radiocarbon, 62: 725–757. DOI: https://doi.org/10.1017/RDC.2020.41

Ribot I., Orban R., De Maret P. (2001) – The Prehistoric Burials of Shum La

Haut de page

Notes

1 The ‘Late Holocene Rainforest Crisis’ is a vegetation disturbance that is recorded in palaeoenvironmental proxies, mostly pollen, from several sites in Western Central Africa between 2800 and 2000 cal. yr BP (Ngomanda et al. 2009). It was originally described in the sediments of Lake Barombi in south-west Cameroon (Maley & Brenac 1998). Here, the crisis is recorded by a greater openness of the forest environment, an increase in the relative importance of poaceae and pioneer tree taxa, a relative decline in old-growth forest taxa, and an increase in pollen from the oil palm Elaeis guineensis—an heliophilic taxon commensal with humans.

2 Positive deviation: when the statistical curve is greater than the null model. The deviation is negative when the curve is smaller than the null model.

Haut de page

Table des illustrations

Crédits © Google Earth
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-1.jpg
Fichier image/jpeg, 551k
Titre Figure 1 – Spatiotemporal distribution of sample sites covering the past 5000 years.
Légende (A) Current spatial distribution of the rainforest shown in green, taken from the Collection 5 MODIS Global Land Cover Type product (www.landcover.org). Countries: AGO–Angola; CAR–Central African Republic; ­CMR–Cameroon; COG–Congo; DRC–Democratic Republic of Congo; GAB–Gabon; GNQ–Equatorial Guinea; NGA–Nigeria. Overlain are 14C-dated archaeological sites in WCA with dated and associated material. (B) and (C) time-latitude and time-longitude distribution of the calibrated radiocarbon dates, respectively. Horizontal and vertical bars show the 95% ranges of analytical error on the 14C dates. (D) Histogram of the calibrated radiocarbon age (median) distribution of archaeological sites shown with 100-year bins for WCA.
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-2.jpg
Fichier image/jpeg, 656k
Titre Figure 2 – Archaeological synthesis of Western Central Africa during the past 5000 years.
Légende All archaeological data presented rely on the SPDs of calibrated 14C dates that were binned in space using 10-km radii and time using 200-yr intervals. SPDs were plotted with a 200-yr moving average to prevent over-interpretation of smaller scale variability. (A) Evidence of human activity in the WCA inferred from the occurrence of remains of millet (Pennisetum glaucum), palm oil (Elaeis guineensis), iron metallurgy and pit features in archaeological context. (B) SPD-inferred population dynamics: SPD of 14C dates (thick line) was compared against an exponential model used as a conservative null hypothesis (see text for explanation). The dark grey area represents the 95% confidence interval for the null model. Red and blue areas represent intervals with significant positive and negative deviations from the exponential model, respectively. Regional cultural timeline (2) shown at the bottom (LSA–Late Stone Age; Neolithic S.–Neolithic Stage; EIA–Early Iron Age; LIA–Late Iron Age). (A) and (B) are adapted from (14). The blue and red vertical lines indicate the limit of the Late Holocene Rainforest Crisis (LHRC) at Lake Barombi (SW Cameroon).
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-3.jpg
Fichier image/jpeg, 176k
Titre Figure 3 – Pottery traditions and main Periods of Western Central Africa.
Légende (A) Data shown in time. All archaeological data presented rely on the SPDs of calibrated 14C dates that were binned in space using 10-km radii and time using 200-yr intervals. SPDs were plotted with a 200-yr moving average to prevent over-interpretation of smaller scale variability and are shown as a greyscale (black represents maximum probability; white represents null probability). Coloured dots refer to sites on the map in (B). Regional cultural timeline shown at bottom (LSA–Late Stone Age; Neolithic S.–Neolithic Stage; EIA–Early Iron Age; LIA–Late Iron Age). The blue and red vertical lines indicate the limit of the Late Holocene Rainforest Crisis (LHRC) at Lake Barombi (SW Cameroon). (B) Location and age of sample sites. Larger dots represent older samples and smaller dots represent younger samples. Colours refer to pottery traditions highlighted in (A).
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-4.jpg
Fichier image/jpeg, 303k
Titre Figure 4 – Chronology of Western Central Africa with Pottery traditions.
Légende Late Iron Age: (a, b) Dibamba A–Cameroon; (c) Lopé region from 11th to 18th centuries–Gabon, (d, e) Dibamba B–Cameroon, (f) Oku–Cameroon, (g) Dibamba C–Cameroon, (h, i) Lobéké–Cameroon, (j) Kovifem–Cameroon. Early Iron Age: (k, l) Campo–Cameroon, (m) Akom near Mintom–Cameroon, (n) Lalara–Gabon, (o) Otoumbi–Gabon. Neolithic: (p) Dibamba E–Cameroon, (q) Mpolongwé–Cameroon, (r, s) Nya Zanga–Cameroon, (t, u) Epona–Gabon, (v) Imbonga–Democratic Republic of Congo (w), CLSA from Madingo-Kayes–Republic of Congo.
Crédits © J. Denbow (w), © R. Oslisly (c, q, o, n, t, u), © G. de Saulieu (a, b, d-m, p, r, s), and © H.-P. Wotzka (v).
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-5.jpg
Fichier image/jpeg, 365k
Titre Figure 5 – Regional variability in population dynamics of Western Central Africa.
Légende (A) Map of Western Central Africa. Dots indicate archaeological site locations and colours delineate the three regions used to estimate population changes. (B) Definition of the three regions based on a hierarchical clustering analysis in Euclidean space. (C) SPD of the three regions (thin line) against an exponential model. The thick lines show the 200-yr rolling mean and the grey area represents the 95% confidence interval for the model. Red and blue vertical bands represent periods with significant positive and negative deviations, respectively.
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-6.jpg
Fichier image/jpeg, 471k
Titre Figure 6 – Comparison of SPDs of the three regions based on permutation tests.
Légende Each row shows an observed SPD (black thick line with its model in grey) of a region compared against another in the column. Red and blue vertical bands indicate periods with significant positive and negative deviations, respectively, from the model of the combined set of each of pair.
URL http://journals.openedition.org/aaa/docannexe/image/3029/img-7.jpg
Fichier image/jpeg, 285k
Haut de page

Pour citer cet article

Référence papier

Geoffroy de Saulieu, Yannick Garcin, David Sebag, Pascal R. Nlend Nlend, David Zeitlyn, Pierre Deschamps, Guillemette Ménot, Pierpaolo Di Carlo et Richard Oslisly, « Archaeological Evidence for Population Rise and Collapse between ~2500 and ~500 cal. yr BP in Western Central Africa »Afrique : Archéologie & Arts, 17 | 2021, 11-32.

Référence électronique

Geoffroy de Saulieu, Yannick Garcin, David Sebag, Pascal R. Nlend Nlend, David Zeitlyn, Pierre Deschamps, Guillemette Ménot, Pierpaolo Di Carlo et Richard Oslisly, « Archaeological Evidence for Population Rise and Collapse between ~2500 and ~500 cal. yr BP in Western Central Africa »Afrique : Archéologie & Arts [En ligne], 17 | 2021, mis en ligne le 03 novembre 2021, consulté le 02 novembre 2024. URL : http://journals.openedition.org/aaa/3029 ; DOI : https://doi.org/10.4000/aaa.3029

Haut de page

Auteurs

Geoffroy de Saulieu

geoffroy.desaulieu@ird.fr – Patrimoines Locaux Environnement et Globalisation UMR 208, IRD, MNHN, 57 rue Cuvier - Case Postale 51, 75231 Paris cedex05 (France)

Articles du même auteur

Yannick Garcin

garcin@cerege.fr - Aix Marseille Université, CNRS, IRD, INRAE, CEREGE, Aix-en-Provence (France)

David Sebag

davidsebag.ird@gmail.com – Normandie Université, UNIROUEN, UNICAEN, CNRS, M2C, 76000 Rouen, (France)/Institute of Earth Surface Dynamics, Geopolis, Université de Lausanne, Lausanne (Switzerland)/HSM, IRD, CNRS, Université de Montpellier, Montpellier (France) / IFP Énergies Nouvelles, Earth Sciences and Environmental Technologies Division, 1 et 4 avenue de Bois-Préau 92852 Rueil-Malmaison (France)

Pascal R. Nlend Nlend

pr_nlend@yahoo.fr – Université de Yaoundé I, Département des Arts et de l’Archéologie et Centre de Recherche et d’Expertise scientifique, Yaoundé (Cameroun)

David Zeitlyn

david.zeitlyn@anthro.ox.ac.uk – Institute of Social and Cultural Anthropology, School of Anthropology and Museum Ethnography, University of Oxford (UK)

Articles du même auteur

Pierre Deschamps

deschamps@cerege.fr – Aix Marseille Université, CNRS, IRD, INRAE, CEREGE, Aix-en-Provence (France)

Guillemette Ménot

guillemette.menot@ens-lyon.fr – ENSL, Université de Lyon 1, CNRS, LGL-TPE, F-69007 Lyon (France)

Pierpaolo Di Carlo

pierpaolodicarlo@gmail.com – Department of Linguistics, University at Buffalo, The State University of New York (USA)

Richard Oslisly

roslisly@parcsgabon.ga – Cellule Scientifique, Agence Nationale des Parcs Nationaux, BP 20379 Libreville (Gabon)/Patrimoines Locaux Environnement et Globalisation UMR 208, IRD, MNHN, 57 rue Cuvier, case postale 51, 75231 Paris cedex 05 (France)

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-4.0

Le texte seul est utilisable sous licence CC BY-NC 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search