Version classiqueVersion mobile

Settling the World

 | 
Léna Sanders

Chapter 5: Transition 2: Modelling the peopling of an already inhabited territory: the case of expanding Bantu populations and Central African forest foragers

Christophe Coupé, Jean-Marie Hombert, Florent Le Néchet, Hélène Mathian et Lena Sanders

Texte intégral

1The second transition studied in the context of the programme TransMonDyn took place on the African continent. It concerns the expansion of the Bantu-speaking populations, which began between 4,000 and 5,000 years ago. This expansion is connected with the development of agriculture and saw an encounter between the Bantu and communities of Central African forest foragers who lived in the ‘rainforest’.

2Our objective is better to identify the temporal and spatial dynamics of this peopling transition, and we especially wish to evaluate the impact that the first relations between Bantu farmers and Central African forest foragers had on the migratory routes taken by the former across or around the forest in their geographic expansion at that time.

3We first describe the Bantu migrations and their geographic and environmental context. We next supply elements regarding the Central African forest foragers, and we reflect on the first contacts between these populations and their possible consequences. This permits us to be more precise about the transition. We then propose a model which takes up and extends that of the ‘out of Africa’ expansion of Homo sapiens nearly 70,000 ago (chapter 4). The different components of the model are detailed, prior to setting out an analysis of the simulation results and offering some interpretations in the light of known archaeological and linguistic facts.

The expansion of the Bantu populations

The Bantu migrations and their geographic and environmental context

  • 1 Hombert Jean-Marie, Hyman Larry (Eds.), Bantu Historical Linguistics: Theoretical and Empirical Per (...)

4Equatorial and austral Africa is today the site of coexistence of different human communities. The most important of these is the Bantu population, which covers the entire part of the continent extending to the south of an imaginary line joining southern Cameroon and northern Kenya, with the exception of the south-west tip of the continent, where the Khoisan populations live. The Bantu population is highly diverse in terms of social and political structures, but is unified by the use of some 450 languages termed Bantu, whose common origin is today well accepted1. Besides their linguistic affiliation, the Bantu populations also communally practice cultivation of different tubers and cereals such as yam and millet, which they supplement by hunting, fishing and gathering. Historically, they have also practiced metallurgy for the production of different tools.

  • 2 Li Sen, Schlebusch Carina, Jakobsson Mattias, ‘Genetic Variation Reveals Large-Scale Population Exp (...)
  • 3 Grollemund Rebecca, Branford Simon, Bostoen Koen, Meade Andrew, Venditti Chris, Pagel Mark, ‘Bantu (...)
  • 4 Vansina Jan, ‘New Linguistic Evidence and “the Bantu Expansion”’, The Journal of African History, 3 (...)

5The present distribution of Bantu populations is the result of a vast migratory movement begun between 4,000 and 5,000 years ago2. Its initial centre was situated in the ‘grasslands’ of Cameroon, a savannah region today densely peopled and situated in the west of the country. This is inferred, among other sources, from archaeological sites, genetic analyses or indeed linguistic reconstructions which suggest the routes of migration at different historical periods3. Conceptions concerning this peopling process have evolved over the years. The idea of a powerful wave of migrants, spreading out thanks to agriculture, subsequently metal-working, and obliterating the native populations in their passage, has given way to a reading considerably more complex and fragmented, featuring advances but also retreats, and comprising numerous secondary processes of expansion possessing their own dynamics4. The dates of the first migrations outside the Bantu heartland have also been pushed backward in time and been dissociated from the mastery of agriculture and metallurgy.

  • 5 Ehret, Christopher, ‘Bantu Expansions: Re-Envisioning a Central Problem of Early African History’, (...)
  • 6 Plaza Stéphanie, Salas Antonio, Calafell Francesc, Corte-Real Francisco, Bertranpetit Jaume, Carrac (...)
  • 7 Alves Isabel, Coelho Margarida, Gignoux Christopher, Damasceno Albertino, Prista Antonio, Rocha Jor (...)
  • 8 Bostoen Koen, Clist Bernard, Doumenge Charles, Grollemund Rebecca, Hombert Jean-Marie, Muluwa Josep (...)

6One of the most recent scenarios proposes a first migration, putatively taking place more than 4,000 years ago, in a southerly direction along the Atlantic coast, scarcely going beyond the region of Yaounde. These early Bantu-speaking populations are not supposed to have made use of agriculture and metallurgy. It is only about 2,500 years ago that far more substantial migrations are supposed to have occurred, this time with populations able to cultivate the land and produce iron tools. The classic scenario involves two major routes of diffusion, connected with the distinction between two significant branches of the tree of Bantu languages5. The first route is located more towards the west and takes a southerly direction. The second passes to the east of the rainforest after deviating around it by the north. It is increasingly clear today that these two migratory routes were not independent of each other and came close together6 7. The expansion begun 2,500 years ago would be better described as a large colonisation front with interconnected populations, rather than a succession of episodes of fission of small independent groups8.

  • 9 Ngomanda Alfred, Neumann Katharina, Schweizer Astrid, Maley Jean, ‘Seasonality change and the third (...)

7Among the mechanisms capable of explaining the causes of the migrations and the routes taken, environmental constraints are prominent. Paleo-climatic reconstructions, derived, for instance, from the analysis of pollen-containing sedimental deposits at the bottoms of lakes, give us a precise picture of the evolution of highly local situations. Even if it is difficult to extrapolate on this basis, the most recent data suggest the beginning of a climatic crisis about 4,000 years ago, and a consequent reduction of the area occupied by the rainforest, to the benefit of the savannahs. A more sharply defined change seems then to have occurred 2,500 years ago, with an amplification of seasonality and the development of mixed environments, halfway between so-called mature rainforest and pioneer rainforest9. These two major historical events correspond exactly to the migratory phenomena previously mentioned.

  • 10 Some articles suggest the possibility of an accentuation, even an inducement, of climatic change by (...)
  • 11 Bostoen Koen, ‘Pearl Millet in Early Bantu Speech Communities in Central Africa: A Reconsideration (...)
  • 12 Neumann Katharina, Bostoen Koen, Höhn Alexa, Kahlheber Stefanie, Ngomanda Alfred, Tchiengué Barthel (...)

8The changes of 2,500 years ago would thus have created favourable conditions for the development of agriculture10. Linguistic reconstructions attest to the cultivation of pearl millet at such historical periods, and this plant cannot grow except during a dry season11. During the migrations, the cutting down of trees of the pioneer rainforest, easier to clear than mature forest, would have permitted creation of spaces for agriculture. The latter would have further been accompanied by strategies for exploiting different wild plants12.

  • 13 Schwartz Dominique, ‘Assèchement Climatique Vers 3000 B.P. et Expansion Bantu En Afrique Centrale A (...)
  • 14 Bostoen Koen, Grollemund Rebecca, Muluwa Joseph Koni, ‘Climate-Induced Vegetation Dynamics and the (...)
  • 15 Ngomanda Alfred, Chepstow-Lusty Alex, Makaya Mvoubou, Favier Charles, Schevin Patrick, Maley Jean, (...)
  • 16 Maley Jean, ‘La destruction catastrophique des forêts d'Afrique centrale survenue il y a environ 25 (...)

9Despite past evolutions, observation of the current geography and in particular of the emplacement of the rainforest strongly suggests that the latter was a challenge on the route of the first Bantu farmers. The forest does not lend itself to cultivation, and it is possible that the farmers of former times, like those of today, feared to penetrate into places considered inhospitable and dangerous. The first farmers may thus have adopted strategies for avoiding the forest, whether by the west or the north. The climatic change nevertheless opened the way to passages through the forest13 which could have been used to advantage, as is suggested by linguistic studies of the names of pioneer vegetal species14. More specifically, the appearance of savannah along the Atlantic coasts 4,000 years ago 15could have facilitated the first migrations. A large corridor, known as the Sangha River Interval, subsequently opened up in the forest a little less than 3,000 years ago, separating one forest massif to the west from another16. The Bantu farmers would have been able to use this passage, as can be seen at the centre of figure 1.

 

  • 17 Russell Thembi, Silva Fabio, Steele James, ‘Modelling the Spread of Farming in the Bantu-Speaking R (...)

10If numerous communities of farmers probably circumvented forested areas when they moved, the question can be asked as to whether others did not cross those areas more directly. To do so, they could have, in particular, progressed over or along different water-courses. Such a hypothesis has consequences for the rhythm of the migrations and the relationship of the different Bantu populations17.

  • 18 Grollemund Rebecca, ‘Nouvelles Approches En Classification : Application Aux Langues Bantu Du Nord- (...)

Figure 1: Various possible migratory pathways for crossing the forest18.

Figure 1: Various possible migratory pathways for crossing the forest18.

The trees in green represent the extent of the forest during the first phases of the Bantu migration. The arrows represent different migratory routes that the farmers may possibly have taken.

11The communities in motion could have counted on their own resourcefulness to spread out across the continent. But they also could have been influenced by their encounter with the populations of forest foragers who occupied equatorial and austral Africa before them. In particular, the forest foragers of the rainforest could have helped them to profit more from the forest milieu and its resources. To clarify this, it is necessary to describe the living conditions of these forest foragers and their relations with the Bantu farmers.

The Bantu and their neighbours: the Central African forest foragers

  • 19 In order to simplify the text we will often use « forest forager » without specifying that we are r (...)
  • 20 Gowdy John, ‘Hunter-Gatherers and the Mythology of the Market’, in The Cambridge Encyclopedia of Hu (...)
  • 21 Ichikawa Mitsuo, ‘Mbuti’, dans The Cambridge Encyclopedia of Hunters and Gatherers, ed. R. B. Lee a (...)
  • 22 Hitchcock Robert K., ‘Introduction: Africa’, dans The Cambridge Encyclopedia of Hunters and Gathere (...)
  • 23 Among the different articles on which we have relied, we may mention: Bahuchet Serge, ‘Les Pygmées (...)
  • 24 Quintana-Murci Lluís et al., ‘Maternal Traces of Deep Common Ancestry and Asymmetric Gene Flow betw (...)
  • 25 Patin Etienne et al., ‘Inferring the Demographic History of African Farmers and Pygmy Hunter-Gather (...)
  • 26 Perry George H. et al., ‘Adaptive, Convergent Origins of the Pygmy Phenotype in African Rainforest (...)
  • 27 Verdu Paul et al., ‘Sociocultural Behavior, Sex-Biased Admixture, and Effective Population Sizes in (...)

12In many regions of equatorial Africa, the Bantu coexist and interact with the populations of forest foragers commonly called ‘African Pygmies’, although this term has been seen as derogatory and ‘Central African forest foragers’ for instance is more neutral.19For anthropologists, the term ‘Pygmy’ refers generally to groups of individuals in which the height of adult males is inferior, on the average, to 1m50 and who are found in different regions of the globe. Even when focusing on equatorial Africa, use of the term masks a great diversity in the populations and their ways of life, whether in terms of socio-economic conditions, of language, or of relations with the neighbouring populations of farmers. We rely in this and the following sections on the anthropological data presented by Gowdy20, Ichikawa21 and Hitchcock22, as well as Bahuchet23. Different genetic studies of the populations, by Quintana-Murci et al.24, Patin et al.25, Perry et al.26 and Verdu et al.27, serve to complete these studies and open useful diachronic perspectives.

13More than twenty groups can be distinguished and, according to Bahuchet, can be classified into three principal inclusive categories: i) the Aka, Baka, Asua, Sua and Efe (the last three grouped together under the name Mbuti); ii) the Kola et the Bongo; and iii) the Twa. The first grouping corresponds to populations still semi-nomadic, moving from one temporary camp to another over sizeable areas and differing notably in religious, political, linguistic and indeed musical traditions from their farmer-neighbours. This suggests that the associations with the latter are either recent or ancient but marked by a strong instability. By contrast, the two other groups have experienced important cultural evolutions which relate them closely to their neighbours and indicate stable relations, already ancient, associated with greater sedentarism. This pattern seems to be confirmed by those genetic studies suggesting more marked exchanges of genes. There exists, however, no linguistic trace of any ancestral languages for Central African forest foragers prior to contact with the agricultural populations. This attests to the profound influence exercised by the latter.

14The great majority of forest foragers, whose total number is difficult to estimateat most some hundreds of thousands of individuals – meet part of their dietary needs from hunting and gathering in the forest. The hunting, gathering and dietary practices vary according to the regions and the characteristics of the forest. Hunting, which can be collective, is carried out by certain groups with nets, something which often permits capturing enough to sell part of the meat. What is procured by hunting is supplemented by different harvested products – roots, mushrooms, leaves, insects, etc. The search for food supplies takes up a great part of their time, and the food may be shared (especially during collective hunts) or not.

  • 28 Diamond Jared M., ‘Anthropology. Why Are Pygmies Small?’, Nature, 354(6349), 1991, pp. 111–112.

15A marked cultural heterogeneity is therefore manifest in the populations of forest foragers of equatorial Africa. Beyond this, what traits do they share? The small size already mentioned is a primary element, but if plausible reasons for this morphological property have been proposed – adaptation to the heat and humidity, greater facility in moving in a dense forest milieu, limited food or high adult mortality28–, it is still not clear whether this shared trait results from an ancient heritage, from more recent convergent adaptations, or a mixture of the two. The principal cultural trait shared by all Central African forest foragers is their role as specialists of the forest, the centre of their intellectual and spiritual life and the fabric of socioeconomic relations that they maintain with the Bantu populations. Capital, in the world of forest foragers, is not something physical that can be measured and manipulated, but rather a shared knowledge accessible to all. In particular, there is no ownership of the land and the products of the forest, apart from nests of bees and termite colonies designated for the purpose.

The relations between Bantu farmers and Central African forest foragers

16As has already been emphasised, the relations between Bantu farmers and Central African forest foragers are especially important. It is possible to approach them by way of material exchanges, then to take a more global view, accentuating their profound asymmetry.

17Even for the most mobile groups of forest foragers, a Bantu village is always situated at one extremity of their territory and is an important place of exchange. The relations maintained with the families, indeed the individuals, of this village are exclusive and most often extend over several generations. The link is thus particularly stable and close, even if all the forest foragers are not connected with Bantu, or vice-versa.

18For one thing, the forest foragers produce neither pottery nor artefacts in iron. They therefore procure cooking pots, knives, assegai points, etc., from the farmers with whom they are connected, to which must also be added tobacco and alcohol. Moreover, if hunting and gathering are activities important for subsistence, a very significant percentage of their foodstuffs consists of agricultural products (manioc, plantain banana, cassava, etc.) also furnished by the farmers. Indeed, an important point is that no group of forest foragers lives autonomously in the forest by nourishing themselves only with its ‘wild’ resources. In exchange for the farmers’ products, the forest foragers furnish different resources originating in the forest, and especially meat. This meat plays an important role, for its quantity permits farmers to gather in large groups during important social events and to maintain their social structures.

19This economic relation can be viewed as symbiotic, and Bahuchet thinks that its complementarity, extremely beneficial for the two groups, suggests that it has been in place since very ancient times. It depends in a way on the very different attitudes of the Bantu and the forest foragers with regard to the forest: while the former does not like going there and consider it a hostile milieu, the latter are at ease and have a way of life adapted to it.

20Nevertheless, if the material exchanges allow one to think that the Bantu and forest foragers are on an equal footing, their relation is in fact marked by a profound asymmetry and images highly unfavourable to the latter. First of all, the traditional relation of material exchanges depends on the idea that the farmer plays the role of tutor to the forest forager. In exchange for the products and the assistance in kind furnished by the latter, the former cares for him by offering what he needs. If one cannot speak of slavery, it is reasonable to speak of dependence, for the forest foragers are completely in the power of the Bantu for obtaining agricultural products and, even more so, iron tools. This need is such that it seems to have unbalanced the exchange in favour of the Bantu and allowed an ideology of superiority of one over the other to become instilled. The current social and racial discrimination manifests itself in a sentiment of superiority and a devalorisation. The forest foragers are regarded as savages and naturally destined to be dominated. Nevertheless, these images present a certain complexity. The forest foragers are disdained but also admired and feared for their faculties and their proximity to the world of the forest and the spirits. The Bantu recognise, for example, the talents and bravery of their hunters. They also consider some to be powerful healers, whom they do not hesitate to appeal to. More profoundly, their mythology reflects the unique and important place that the forest foragers occupy for them: in it, the latter are not considered as men, nor as animals, but as intermediate beings who supposedly permitted man to pass from nature to culture, achieved through the gift of fire, the domestication of plants or the working of iron. They are also supposed to have guided the farmers through the forest, initiated them into it and furnished them with its products, permitting them, in the end, to exercise a certain control over that difficult environment. The forest foragers thus play a structuring role in the Bantu system of thought, something that explains why some rituals strongly link members of the two communities and why the Bantu consider that they must protect beings who guided them out of the state of nature, even if these beings have remained within it or been forcibly assigned to it.

21The representations of the place of the farmers and that of the forest foragers are reflected in the prevailing intermarriage rules. Marriages between forest forager men and Bantu women are prohibited, and if marriages between Bantu men and forest forager women are possible, they are often explained by the latters’ lesser dowry or by the desire to construct an alliance with a renowned hunter or healer.

22Do the present relations between the forest foragers and the Bantu farmers reflect the conditions of their encounter? Different arguments make it possible to conceive in detail the response to this question. It is necessary, by way of introduction, to emphasise that the relation between the forest foragers and their Bantu ‘patrons’ has not been static over the centuries. If it may be considered relatively stable and economically balanced for several centuries prior to the colonial period, it has subsequently been profoundly affected by a series of socio-economic transformations which have progressively modified the traditional relationship. These changes have led the Bantu farmers to take control of their ‘clients’, especially as a source of agricultural labour, in order to face external demands and the requirements of local market economy. This has modified the mobility patterns of the forest foragers, increased the proportion of agricultural production in their diet, fostered individualism and social and health problems, or indeed alcoholism.

  • 29 Indeed, if a mutation occurs in an individual, it has a greater chance of prevailing in the group a (...)

23Looking behind these recent changes, there is hardly any doubt that the encounter between the two communities occurred during the expansion of the Bantu populations from their expansion base. Different genetic studies involving autosomal or mitochondrial DNA suggest an ancient separation between the ancestral population of forest foragers and an ancestral population of central Africa between 50,000 and 90,000 years ago, depending on the studies. The former would subsequently have evolved in equatorial Africa, with a separation between the forest foragers of the western rainforest and those of the east about 20,000 to 30,000 years ago. This is in basic agreement with the proposition of Bahuchet, who suggests that a macro-group living in the forest was divided into three groups between 30,000 and 18,000 years ago in response to the splitting of the forest into as many parts. The isolation of the three groups of forest foragers would have been maintained even after the reunification of the three forest massifs into a single one 15,000 years ago, and other local evolutions would have taken place, reflected in the great genetic differentiation in the present populations (linked also to the small size of the groups29). This conception of the history of the forest foragers of equatorial Africa can be termed ‘pre-Bantu era divergence’. It opposes the hypothesis of ‘recent divergence’, rejected by a majority today, according to which the forest foragers of the west were the result of a hybridisation between the ancestors of the forest foragers of the east and the Bantu farmers 2,000 or 3,000 years ago.

24The situation of the Central African forest foragers when the Bantu arrived is open to question. On the one hand, a previous anthropological reading of their way of life led to a view of them as a people ‘without a history’, living in isolation in the forest and in a manner unchanged from very ancient times. On the other hand, there are arguments strongly contradicting the idea of a strict isolation. First, study of the mitochondrial DNA of different populations of Bantu farmers and of forest foragers reveals an asymmetrical and long-standing flow of maternal genes from the ancestors of the forest foragers to the ancestors of the farmers, which began at least 40,000 years ago and persisted until some several thousand years ago. Moreover, the debate already referred to as to the possibility of surviving in isolation in the forest suggests that the ancestors of the present forest foragers may have lived differently from their descendants, more at the edge of the forest, either with exploitation of it but also of the savannah’s resources, or according to a mode of subsistence even more disconnected from the forest. This presence outside the forest could explain the contacts over the long term revealed by the genetic analysis of the maternal lines of descent.

25A recent study indicates that the gene-mixing between farmers and forest foragers took place in a significant way only over the course of the last several thousands of years, that is, at least 1,000 years after the first contacts between the two populations. The authors therefore propose that the first contacts did not lead to immediate genetic exchanges, but that these occurred only later, possibly after the establishment of a differentiation between the two populations linked to their respective adaptation to their different ways of life. Such a hypothesis might thus lead to the conclusion that the forest foragers became ‘professionals of the forest’ only recently and as a consequence of the arrival of the Bantu farmers. This hypothesis is rejected by Bahuchet, in keeping with his evolutive scenario of the populations of forest foragers, presented above, on the basis of linguistic comparisons, particularly the very slight percentage of basic words connected with agricultural life in the current languages of the forest foragers.

26The nature of the first contacts between Bantu and Central African forest foragers is difficult to be sure of and depends on the way of life at the time of the ancestors of the present forest foragers. The loss by the latter of their original languages in favour of Bantu languages suggests that their early contacts were important. The genetic and cultural diversification of the populations of the western Central African forest foragers might equally be a sign of the constraints imposed by the circumstances of contact, even if the mechanisms remain uncertain.

27Following the track of the legends and myths already cited, Bahuchet proposes a situation very different from the present one at the time of the initial contacts. Study of the shared vocabulary between the languages of the forest foragers and those of their agricultural neighbours lends credence to the idea that the farmers acquired knowledge of the forest from the forest foragers, as well as technical capacities associated with it. These elements suggest a form of dominance at the time of the forest foragers over the farmers. Anthropology also encourages us to pay attention to different conceptions of right to the land between the Bantu and the forest foragers. Whereas individual property hardly exists for the forest foragers, it is an active force for the Bantu and attributed to the first individual who arrived in the locale. It is possible on this basis that the first Bantu to enter into contact with the forest foragers attributed the possession of the lands to them and asked them for the right to exploit these.

28Numerous uncertainties remain: were the genetic exchanges modified shortly after the first encounter between forest foragers and farmers, or only after a certain period of living side-by-side, and if so, why? Might this correspond to a scenario in two phases, with a first period of communal life and the borrowing by the Central African forest foragers of the language of their Bantu neighbours, then a second phase following migrations and the formation of new alliances, this time without borrowing the language? Was the first phase ‘to the forest foragers’ advantage’, unlike the next one? Did the ancestral populations of the forest foragers already live in close relation to the forest when the farmers arrived? If this was not the case, was the specialisation in the forest milieu compelled by the pressure of the Bantu, or rather a free choice with the progressive establishment of a symbiotic relationship?

29For the model presented in the second part of this text, we have accordingly had to make a certain number of choices, fully aware that other scenarios would equally have deserved study by way of modelling. We have therefore opted for the idea of the forest foragers as already evolving in the forest milieu, and of a relation from the start beneficial to the farmers and more neutral for the forest foragers.

30To answer the preceding questions is important for understanding the influence that the forest foragers could have had on the rhythm and structure of the Bantu migrations. One might imagine that the relations developed with the ancestors of the forest foragers could initially have facilitated the clearing of certain zones of the forest. But the migrations could also have been slowed by the relations of dependence then created with the forest foragers, more inclined to mobility in a limited territory than to the ‘expansive’ mobility of farmers impelled by the need for new lands. It is to test these different possibilities that a computer model proves valuable.

Specification of the transition

31Two roughly stationary regimes can be defined before and after the transition: the first is that characterised by the presence of the first Bantu solely in present-day Cameroon, contemporary with the existence of other populations, and especially communities of forest foragers, in other regions of the African continent. The second regime is that of the current, or rather pre-colonial, situation, featuring maximum expansion of the Bantu populations and their contact with different communities of forest foragers in the equatorial region.

  • 30 Diamond Jared, Bellwood Peter, ‘Farmers and Their Languages: The First Expansions’, Science, 300(56 (...)

32The question of causality respecting the transition is complex. If agriculture is viewed as the main driving force of the ‘Bantu transition’, the reasons for its development are themselves subject to debate. Priority has been claimed for various causes: environmental, demographic (population increase), sociocultural (evolution of social relations within human groups, association of individuals for religious reasons). Moreover, these potentially apply in different ways according to the geographical regions under consideration. In fact, agriculture appeared on the surface of the planet many times and independently, beginning 11,000 years ago30. In this context, the appearance of agriculture in central Africa four or five thousand years ago is a recent event, and it is characterised by different environmental and sociocultural conditions.

33If the causes of the development of agriculture are themselves difficult to identify, its consequences are more evident and can be traced along several axes.

  • 31 Nettle Daniel, ‘Language Diversity in West Africa: An Ecological Approach’, Journal of Anthropologi (...)

34First of all, the relation to the environment is, of course, radically modified, since individuals deriving their subsistence from agriculture transform their relation to plant species and to the soil. If a certain temporality is already present in the way of life of forest foragers – movements governed by the seasons, exploitation of the nearby environment – farmers have their dependence on seasonal and non-seasonal changes of climate reinforced. This dependence and the at least partial sedentarisation that accompanies it are compensated for by the resulting gains derived in terms of subsistence. The short-term fluctuations in the production of agricultural resources (droughts, floods) can be made up for in part by the storage of harvests, recourse to varied sources of food and the implementation of long-distance networks of exchange31.

35Then, with regard to the modes of occupying space and spatial interactions, the passage to agriculture is commonly associated with sedentarisation. This, however, needs to be relativised – first, because itinerant farming systems exist, and then because the transition between hunting and gathering and agriculture has been much more liable to occur gradually than suddenly.

36On a macroscopic level, the need for irrigation – more or less important according to the type of agriculture – has resulted in intensive settlement in the valleys of great waterways, whether in sub-Saharan Africa or in other regions of the world. Agricultural regions have also seen an important increase in population densities. The distinction between nomad and partially sedentarised communities may have lent itself to translation as a complementary distribution of inhabited zones – in particular, forest for some, cleared areas for others. Between these zones of differentiated settlement, but also within them, circulatory networks of foodstuffs or goods (connected with agriculture or not) could develop.

37Forms of habitation also changed along with agriculture: as part of sedentarisation, housing conceived for longer periods of time developed. The forms of houses evolved progressively, and community buildings, intended for various uses, likewise made their appearance after the transition.

38Finally, the preceding evolutions were accompanied by profound social, technical and cultural transformations. The development of tools and agricultural techniques, storage for the harvests or indeed the domestication of animals were the visible aspects, which were underpinned by modifications and specialisations of the roles of members of the group with respect to the production of nourishment (agriculture/hunting). With the increase in density of the population and the possibility of storing food, new activities made their appearance: exchange and protection of surpluses, irrigation, fabrication of containers, etc.

39With the general increase in population density in the cultivated areas and the concentration of human presence in specific locations, new forms of power, economic exchanges or symbolic exchanges could become established. Social relations were transformed accordingly, both between and within groups, which could become integrated into social structures more complex and hierarchised. Religious and metaphysical ways of representation could develop easily.

40These different evolutions, which separate the pre-transition and post-transition regimes, allow for better understanding of the migratory and cultural dynamic of the Bantu populations, as they might do with other great migrations linked to agriculture – that of the Indo-Europeans, for example. They do not enlighten us, however, concerning the migration routes taken by the Bantu, which depended as much on the capacities of their communities as on their environment. In order more precisely to characterise the transition of Bantu peopling, we can therefore try to place it within its (paleo)ecological and social context. As with the first peopling transition, ‘Out of Africa’ (chapter 4), what matters then most is not to identify the causes and driving forces of the transition, but rather to explore the factors underlying the spatiotemporal structure of the migrations over more than 2,000 years.

The HU.M.E.-2 32 model: Modelling migrations in a context of interactions between groups

  • 32 Since the model for exploring the expansion of Bantu peopling was constructed on the basis of the H (...)
  • 33 Li S., ‘Genetic Variation …’, op. cit.
  • 34 Lake, Mark W., ‘Trends in Archaeological Simulation’, Journal of Archaeological Method and Theory, (...)

41The objective of the modelling is to contribute to the debate about the routes taken by the Bantu in their expansion to the south 33, exploring in a highly stylised manner the possible consequences of different types of interactions between Bantu and forest foragers for the form and the rhythm of that expansion. In contrast with an approach through statistical models, the point is not to analyse empirical data already assembled, but to produce, by simulation, artificial data corresponding to different hypotheses and initial situations34. By simulating alternative histories and comparing the evolutionary trajectories obtained according to different forms of interactions between groups of farmers and forest foragers, we hope to produce hypotheses concerning the processes involved during the Bantu expansion.

Modelling the interaction between agents in a context of competition for resources: a rapid overview of the literature

  • 35 Hazelwood Lee, Steele James, ‘Spatial Dynamics of Human Dispersals Constraints on Modelling and Arc (...)
  • 36 Parisi Domenico, Antinucci Francesco, Natale Francesco, Cecconi Federico, ‘Simulating the expansion (...)
  • 37 Barton Michael C., Riel-Salvatore Julien, ‘Agents of change: modeling biocultural evolution in Uppe (...)
  • 38 Such a search may extend over several days with the nights spent in sites somewhat removed from the (...)

42Two essential mechanisms intervene in migratory dynamics – the search for resources and the interactions between groups. The literature dealing with the dynamics of peopling associated with constraints of resources has been taken up in chapters 3 and 435 36. Other studies focus on the effects of interactions between agents on the evolution of the peopling system. Concentrating on the disappearance of the Neanderthals in favour of Homo sapiens, researchers37 have, for example, formalised an agent-based model for testing different hypotheses concerning the role played by climatic cooling. Two types of mobility figure in the practices of hunter-gatherers: (1) movements within their catchment area; (2) changes of base camp when the resources of their food-supply area are exhausted. Faced with a lack of resources, the groups of hunter-gatherers can thus choose between two strategies: enlarging their food-supply area by going further from their base camp during their search for resources38 or changing their base camp. The model developed by these researchers rests on the hypothesis that the climatic cooling led the groups of hunter-gatherers to enlarge their food-supply area, and the simulations show that the result is an increase in potential interactions between groups of Neanderthals and groups of Homo sapiens, interactions which might have sufficed to result in the assimilation of the former by the latter without the need to mobilise hypotheses relating to conflicts.

43The model developed here (HU.M.E.-2) to explore the expansion of the Bantu to the south also attributes driving force to the interactions between new arrivals and native groups. By contrast, since the challenge is to simulate the routes of migration, it is the scale of migrations (with a change of base camp) and not that of movements in the catchment area which is considered. In this sense, this model differs from that dealing with the expansion of Homo sapiens referred to above, as well as from the model ColoDyn described in chapter 3. The model HU.M.E.-2 was constructed in two stages: the first consisted in adapting the HU.M.E. model to respond to the problematic of Bantu expansion in a heterogenous environment of savannah and forest, the second in specifically formalising the form of interactions between groups exhibiting different behaviour.

First step in the construction of the HU.M.E.-2 model: adapting the parameters of the HU.M.E. colonisation model

  • 39 Sapiens, in most of the places where he travelled (with the notable exception of Australia and the (...)

44Here we very briefly recapitulate the functioning of the HU.M.E. model presented in chapter 4. Conceived in order to simulate the colonisation of an essentially39 empty space, the model includes ‘group-agents’ that move within a space. This is formalised by a grid composed of cells heterogenous in terms of resources. The group agents arrive in successive waves at the beginning of the simulation. Their number subsequently varies, one group-group-agent having the capacity of splitting in two (birth of a new group-group-agent) or of disappearing if there is no longer enough ‘energy’. Energy is an attribute of the group-agents. Its value increases with the time of remaining within a cell (as long as this offers sufficient resources) and diminishes with the movements effected. The group-agents exploit the resources according to their technical level, and their production increases in proportion to this level. The resources diminish when they are consumed and are renewed according to certain temporal intervals. The element triggering a migration is the insufficiency of resources. When these are lacking, the group-agent migrates following a random walk to one of the neighbouring cells. The model is stochastic, and each behavioural action of a group-agent (acquisition of a new more advanced technique, migration) is associated with a probability.

45In the HU.M.E. model, the group-agents are ontologically undifferentiated, in the sense that they are described by the same variables and act according to the same processes. The first adaptation of this model therefore consists in formalising the difference between two types of group-agents, taking specifically as a model the groups of Bantu and forest foragers (FF). The Bantu are farmers, who exploit their local resources more intensively and present a dynamic of strong demographic growth. This leads to rapid migrations toward the south of the African continent. On the contrary, the FF groups are hunter-gatherers, who exert lesser pressure on the environment and have a stable total population over the period. As a result, the frequency of their migrations is lower. The overall structure of agents and cells of the HU.M.E. model has been preserved, as well as the properties and processes belonging to the group-agents (cf. figure 2 of chapter 4). The agents’ capabilities of action in the HU.M.E.-2 model are thus broadly the same as for the HU.M.E. model. Four adaptations, however, have been applied in order to take into account the context of the Bantu expansion and some differences between the two types of group-agents:

  • The first adaptation relates to the simulation grid (a virtual continent), into which two environments of different natures have explicitly been introduced: the savannah and the forest (figure 2.a), as well as the notion of ‘biomass’ to quantify the state of the resources in each cell. Each milieu corresponds to a certain initial quantity of biomass, more important in the forest than in the savannah. This distinction between types of milieu makes it possible to take account of the different relations of the Bantu and FF groups to the environment, a difference which constitutes a driving force in the evolution of the system of Bantu peopling. The Bantu group-agents penetrate into the grid in successive waves at the outset of the simulation, whereas the FF group-agents are initially distributed in random fashion in the forest (figure 2.b). At t0 there are only two values of biomass: a small value for the savannah (bs) and a large value for the forest (bf). As the resources are consumed and renewed, the cells are differentiated, and one observes the emergence of a continuum of values.

  • The second adaptation consists in introducing different forms of behaviour for the two types of group-agents and associating different parameter values with them: those characterising the potential for innovation are high for the Bantu group-agents and low for the FF group-agents; the parameters relating to energy are set as neutral for the FF group-agents, the hypothesis being that the latter have little or no recourse to the capitalisation of energy (storage does not correspond to the practices of the Central African forest foragers); finally, the value of the demographic parameter (i.e., the probability that a group can divide at each time-step and create a new group) is virtually nil for the FF group-agents, whereas it manifests a tendency towards growth for the Bantu groups.

  • A third adaptation concerns the interactions between the group-agents and their environment. For a FF group-agent the quantity of resources it can extract in a single time-step increases in linear fashion with the quantity of biomass (figure 3.b). By contrast, the most favourable environments for the Bantu group-agents, as farmers, are intermediate environments in terms of biomass (rich, but in the savannah, i.e., outside of the forest, where cultivation is difficult). To manifest all these hypotheses together in the model, we have recourse to the notion of utility, derived from the economic sciences and resting on the hypothesis that it is possible to quantify on a linear scale what an economic agent, faced with a choice among several options, will gain from each of the possible ones (with the idea that the option having great ‘utility’ for the economic agent will most often be the one chosen). In the HU.M.E.-2 model, the utility for the Bantu group-agents of moving into this or that cell is expressed by a function of discontinuous utility (figure 3.a), the point of discontinuity corresponding to the threshold of biomass at which the forest begins (signified as bf in figure 3). Beyond this threshold, the capacity of Bantu group-agents to exploit the biomass, fB(b1), falls drastically. The lower the value of this threshold, the higher the repulsive effect of the forest on the Bantu group-agent.

  • The fourth adaptation consists in anticipating the direction of the group-agents’ movements: they have the capacity to know the resources and the human occupation of the neighbouring cells. This knowledge conditions their movement, for the group-agents seek to optimise their future potential to exploit the resources. This capacity of anticipation is a realistic hypothesis for the period and the spatial scale considered, if one accepts the plausible idea that the potential destinations were explored by scouts. At each increment of time t, the group-agent evaluates the resources of the cell in which it is located, and if these are insufficient, it explores the potential of the resources of the four neighbouring cells. Migration is more likely to take place towards the cell which offers the greatest potential resources (cf. figure 3.b). One can underline the fact that for the FF group-agents, the principal driving force for the migrations is the insufficiency of resources.

Figure 2: Simulation grid of the HU.M.E.-2 model.

Figure 2: Simulation grid of the HU.M.E.-2 model.

2.a: Introduction of two types of environment, savannah and forest (the grid thus corresponds roughly to that of the HU.M.E. model, with the addition of the rainforest).
2.b: Initial situation, with the FF group-agents in the forest and the flow of Bantu migrants penetrating into the grid by the north-west corner.
2.c: Situation after several time-steps.

Figure 3: Differences in behaviour between Bantu and FF group-agents

Figure 3: Differences in behaviour between Bantu and FF group-agents

3.a: Differential capacities of the Bantu and FF group-agents to exploit the biomass
3.b: Evaluation of the resource potential of the destination zone for each type of group-agent.

Second step in the construction of the HU.M.E.-2 model: formalising the interaction between groups with different behaviour

46An interaction between group-agents occurs when both occupy the same cell. The form of this interaction varies, depending on whether the two group-agents are of the same type (in which case the formalisation is the same as in HU.M.E.) or whether the two group-agents are of different types. In the latter case, the rules have been formulated to account for a relation of reciprocal, but asymmetric, ‘dependence’ between FF and Bantu groups: the former furnish resources to the latter, thus facilitating their survival. The objective is not to model the process that has led to the emergence of this relation (which has likely evolved over time), but to explore what role such interactions have had in the rhythms of Bantu migrations: for instance, did the exchanges with the FF groups contribute to a more rapid, and/or more secure, passage of the Bantu groups toward the south? The H.U.M.E.-2 model should permit exploration of such a hypothesis, and the conditions of such a result could then be interpreted as factors facilitating the colonisation of new territories.

47In the model, this relation of ‘dependence’ between Bantu and FF group-agents is formalised by a ‘link’ which appears or disappears according to certain conditions. The link is created when a Bantu group-agent finds itself in the neighbourhood (in the same cell or a nearby one) of a FF group-agent. Let us suppose that the Bantu group-agent represented in figure 4a has too few resources in the cell in which it finds itself. Therefore, it must move and evaluate the potential of neighbouring cells. The cells to the east and south offer the same potential in terms of exploitable resources, superior to that of the forested cell to the north (to apply the function of utility described in figure 3). In this circumstance, a random choice is made which leads, in the example represented, to opting for the cell to the east. The existence of a FF group-agent in the neighbouring cell then leads to a link between the two group-agents (figure 4.b). This link was not anticipated by any agent; it appears in this context of spatial proximity. The existence of this link has two consequences:

  • 40 For the FF agent-group, everything takes place as if it had to exploit the forest for a more import (...)

48at the level of the resources of the Bantu group-agent: the resources exploited in the cell in which it is located are augmented by the resources exploitable by the forest foragers group-agent with which it is linked (figure 4.b). The quantity of resources transferred from the FF group-agent40 to the Bantu group-agent is determined by a parameter (Ω). All things being equal, this cell will therefore be more ‘inhabitable’ for the Bantu group-agent, and the consequence is that it will be able to stay there longer.

  • 41 In reality, at the present time, the Bantu have need of the FF chiefly for meat and for working in (...)

49at the level of the future movements of the group-agents: the link is formalised using a mathematical function working as an ‘elastic’, in the sense that a ‘tension’ is created when one of the two group-agents moves, and this tension applies a force which will influence the future choice of destination of the group-agents (figure 4.c). In fact, each group-agent seeks, when it moves, to minimise the distance from the group with which it is linked. The existence of the link thus plays a role in determining the direction of the movement when the agent must move, and this is equally the case for the FF as for the Bantu. Beyond a certain distance (which value is a parameter of the model), the link is broken (figure 4.d). This rule manifests the ‘reciprocal’ character of the relation, insofar as each group-agent is attracted by the group-agent with which it is linked. This new rule complexifies the interaction between Bantu and FF and as a result, the relationship between Bantu and FF cannot be described simply as a hierarchical relationship. Indeed, empirical studies show that the FF have an interest in the proximity of the Bantu, who can furnish them, for example, with tools facilitating exploitation of the resources in the forest41.

  • 42 The Cobb-Douglas function is used in economics to express, in a multiplicative manner, the effects (...)

50To formalise the effect of this link on the choice of a destination cell by the group-agent that needs to move, a utility function combining two factors is used: 1) the attraction of the potential of the future cell; and 2) the minimisation of the distance to the group-agent with which it is linked. The following function, known as that of Cobb-Douglas42, has been used to formalise the combined roles of these two factors:

UC,T(D) = fC(T,bD) × g(L)1-a

51Where

  • UC,T(D) is the utility of the destination cell D for a group of culture C (Bantu or FF) and of technical level T,

  • bD is the quantity of biomass found in the destination cell D,

  • fC(T,bD) is the capacity to exploit a quantity of biomass bD for a group of culture C and technical level T (cf. figure 3.a),

  • g(L) is the function expressing the effect of the link (of length L, calculated by Euclidean metric), when this exists, on the utility. The function g(L)= 1/(L+1) has been used, expressing a diminution of this effect when the length of the link (i.e., the distance between the two group-agents concerned) increases,

  • a is the parameter which determines the relative weight of each of the two factors f and g at the time of choosing the destination cell. Let us consider the case of the Bantu group-agent in figure 4c. If a = 1, only the potential capacity to exploit the biomass intervenes in the choice of the new destination, which will then be either the cell ‘south’ or the cell ‘west’. Inversely, if a = 0, the only intervention will be that of the minimisation of the distance from the FF group-agent, and the movement will be to the north or the east. In the intermediate cases, the two elements will intervene in the calculation of the best destination cell. It is therefore the combination of the greater or lesser value of a and the difference in the exploitable resources in the possible destinations for the Bantu group-agent that will determine its choice. This situation is illustrated by the passage from figure 4.c to figure 4.d: if the weight given to the resources is predominant, the Bantu group-agent will move to the south (the case of scenario 2); otherwise, it will choose to move to the east (scenario 1).

52Thus, a ‘linked’ Bantu group-agent will be able to remain longer in a cell, thanks to the contribution of the FF group-agent, and that link will therefore have the effect of reducing the rate of migration. The existence of the link will, additionally, influence the direction of that trajectory when a movement eventually takes place.

Simulating the crossing of the rainforest by Bantu populations: some simulation results

53Once constructed, the model makes it possible to explore the role of the different parameters regarding the way in which the Bantu expansion towards the south took place, and notably the overcoming of the obstacle constituted by the forest. The point, then, is to study in the domain of the model, the chief thematic questions developed earlier:

  • How often, in what forms, and with what rhythms was the forest crossed by the Bantu groups?

  • What role did the forest foragers groups play in the course of that transition?

  • 43 Following Heckbert (2013), we use this term to designate an analysis which has not systematically e (...)

54A first step consisted in elaborating the indicators making it possible to characterise the output of the simulations. An experiment plan was then designed to explore the results of the model for different configurations of parameters. An informal analysis of sensitivity43 was realised by causing the key parameters to vary. Thirty repetitions were run for each set of parameters, and the results of the simulations are represented with the aid of graphs and maps to characterise the situation at time step t=4000, which offers a good basis for a comparison of the results of the different simulations. Here we present the first results obtained.

Construction of indicators to characterise the passage through the forest of the Bantu group-agents

55In order to characterise the trajectories of the Bantu group-agents, three geographic zones were distinguished in the simulation grid (Figure 5.1): zone 1, occupied by the savannah to the north, zone 2 corresponding to the forest, with two savannah corridors situated at the centre and to the east of the grid, and zone 3, occupied by the savannah to the south of the forest. To characterise the overall results associated with each simulation, indicators were constructed which describe the way in which the passage through the forest was accomplished during the Bantu expansion to the south. Four general indicators were used:

  • Two indicators concern the ‘attempt’ and the ‘success’ of the crossing of the forest. Note that here we use the term ‘attempt to cross the forest’, even though the Bantu group-agents do not have a ‘global’ objective but only a local one (movement to a neighbouring cell which permits them to optimise their exploitation of available resources). Successfully crossing the forest and expanding beyond it are therefore emerging features of the peopling process. To calculate the number of attempts to cross the forest, each entry into zone 2 by a Bantu group-agent is interpreted as the beginning of an attempt to cross. Three outcomes are then possible: 1) the Bantu group-agent eventually lacks resources and disappears (figure 5.2); 2) the Bantu group-agent returns to zone 1 (figure 5.2); 3) the Bantu group-agent ‘succeeds’ in crossing the forest and arrives in zone 3 (figure 5.3). In the first two cases, the crossing is interpreted as a failure, in the third as a success. The two indicators correspond respectively to a count (the number of crossings attempted) and a calculation of percentage (the proportion of crossings accomplished).

  • The average duration of the crossing of the obstacle is calculated based on the total number of cumulated time steps spent in zone 2 by all the groups that have succeeded in passing from zone 1 to zone 3 (figure 5.3), divided by the number of such groups, in the course of a simulation. This number of time-steps is interpreted as the average duration of the forest crossing.

  • A final indicator describes the way of crossing the forest, i.e., the routes preferred. It calculates the percentage of time spent respectively in each of the two milieux, forest and savannah corridors, by the totality of the groups that have succeeded in crossing the obstacle (figure 5.3).

Figure 4: Formalisation of a reciprocal but asymmetrical ‘link’ between Bantu and FF group-agents: point of view of the Bantu group-agents

Figure 4: Formalisation of a reciprocal but asymmetrical ‘link’ between Bantu and FF group-agents: point of view of the Bantu group-agents

(the left-hand column illustrates the situations before movement of the Bantu group-agent and the right-hand one the situation after the movement).

56The experiments related to the following processes:

  • the deterrent effect of the forest, as modelled using the parameter determining the capacity to exploit the forest (value fB(bf) in figure 3);

  • the effect of the interactions with the FF group-agents, using the parameter ‘a’ described above.

Figure 5: Basic principles for calculating the indicators of obstacle crossing.

Figure 5: Basic principles for calculating the indicators of obstacle crossing.

Impact of the deterrent effect of the forest on the expansion of Bantu group-agents

57The parameter tested here corresponds to the capacity of the Bantu group-agents to exploit the biomass in the forest milieu. This capacity function fB decreases abruptly beyond the threshold ‘bf ‘ (figure 3). Three different values of bf were tested: each corresponds to a level of loss of capacity to exploit resources by a Bantu group-agent finding itself in the forest, compared with what it might exploit in the savannah milieu (the values tested correspond respectively to losses of 78%, 97% and 99.5%, see figure 6). The results of the simulations show that the Bantu group-agents attempt to overcome the obstacle less often when the incapacity to exploit a high biomass increases. The number of attempts to cross the forest is thus five times greater for the highest value of this parameter than for the lowest. The percentage of successes in crossing, by contrast, is stable and close to 10%. Therefore, once it has penetrated into zone 2 (cf. figure 5.1), the group-agent has one chance in ten of ‘succeeding’. Finally, the means of crossing differs, as is to be expected, according to the values of the parameter, the savannah corridors being more often used by the Bantu group-agents as their incapacity to exploit the forest biomass increases (figure 6). On the other hand, the greater the incapacity to exploit the biomass, the more rapid the crossing of the forest, with an increased preference for the use of the corridor. The Bantu crossing of the forest thus takes place at different rates of speed and following different itineraries, but takes place in any case regardless of the level of avoidance of the forest.

Figure 6: Variety of crossing times (for the group-agents having crossed the forest) as a function of the capacity to exploit the forest biomass.

Figure 6: Variety of crossing times (for the group-agents having crossed the forest) as a function of the capacity to exploit the forest biomass.

Effects on the Bantu expansion of interactions with the FF group-agents

58The effect of the link between a Bantu group-agent and a FF group-agent figures by way of two parameters:

  • parameter ‘a’, which expresses the relative weight attributed to the resources and to the ‘link’ at the point of choice of destination by a Bantu group-agent.  This parameter varies between 0, in the case where the only element that matters is the minimisation of the distance from the FF group-agent with which a link exists, and 1, in the case where only the evaluation of resources is taken into account;

  • parameter ‘Ω ‘, which measures the contribution of a FF group-agent to the subsistence of a linked Bantu group-agent in terms of resources: the higher this parameter, the greater the quantity of biomass furnished by the former group-agent to the latter . This parameter varies between 0, in the case where no transfer takes place, and 1, in the case where the FF group-agent furnishes a quantity equivalent to what it exploits for itself.

59The effect of each of these two parameters was tested separately.

60The results of the simulation show that, the weaker ‘a’ is, the more frequent are the attempts at crossing (division by 6 between a=0.2 and a=1) (figure 7.a). Accordingly, the durations of the crossing are longer, and the relative portion of time spent in the forest rather than in the savannah corridors is greater (figure 7.b). This result suggests that for those situations where the effect of the link is predominant, the ways of crossing are close to a ‘conditional random walk’ through the forest, with Bantu and FF group-agents advancing in pairs. One observes, however, that the percentage of success is independent of the value of this parameter, but that the total number of crossings will finally be superior, given the increase in the number of attempts.

Figure7: Sensitivity of the simulation results to variations in the parameter ‘a’

Figure7: Sensitivity of the simulation results to variations in the parameter ‘a’

7.a: Variation of the number of crossing attempts as a function of ‘a’
7.b: Variation of the duration of the crossing as a function of ‘a’

61Three values of parameter Ω relating to the transfer of resources between a FF and a Bantu group-agent were then tested. The number of attempted crossings appears stable for the different values. This effect might appear contrary to intuition, insofar as FF assistance might in theory attract the Bantu group-agents into the forest. This ‘gift’ of resources to a Bantu group-agent situated on the edge of the forest contributes, however, to the fact that it has less need to move. It seems, therefore, that these two effects balance each other.

62By contrast, the average duration of the forest crossing varies according to the values of Ω. It increases with the amount of support brought by the FF group-agent to that of the Bantu. The consequent increase in the length of time spent in zone 2 illustrates a ‘braking’ effect applied by the interaction with the FF group-agent. The transfer of resources permits the Bantu group-agent to survive more easily in this milieu, and for this reason it is less often impelled to move. With respect to itineraries, the time spent in the forest compared with that spent in the savannah corridors diminishes when Ω increases. This last result may seem counter-intuitive: one might in fact have imagined the Bantu group-agents remaining longer in the forest, since the resources given by the FF group-agents permitted them to survive in it. It seems, however, that an effect of interaction combines with the preference for the savannah milieu to favour trajectories at the interface between forest and savannah in the corridors in the centre and to the east of the forest. The Bantu group-agents remain on the edge of the forest, in their privileged milieu, and remain there longer, thanks to the supplementary resources supplied by the FF groups.

Discussion

  • 44 Alves Isabel et al. ‘Genetic homogeneity across Bantu-speaking groups …’, op.cit.
  • 45 The alternative hypothesis to the ‘early split’ is that of a direct crossing through the forest by (...)

63The results of the simulations show that the interactions of the Bantu group-agents with the FF group-agents tend to favour the attempts to cross the forest and overall to slow the rate of that crossing. Thus, the introduction of specific rules for the interaction between Bantu and FF, with an effect of reciprocal attraction, throws into relief the surmounting of the obstacle by passage through the middle of the forest, as is compatible with the ‘early-split’ hypothesis proposed by Alves et al.44 – that is, the existence of a ‘separation of the population into two major routes’, one passing directly to the south across the forest, and the other by-passing the forest by the north-east45. It is to be noted, however, that no intentionality is introduced in the model. The rules of the model concern movements solely dictated by proximal needs: the lack of resources in a cell entails a movement to a neighbouring one. There is no intention to cross the forest to occupy the territories of savannah more to the south. In fact, there is no awareness even of the existence of these spaces. The crossing of the forest by a collection of groups and the southward expansion are emergent properties of the model. This result shows that the crossing of the forest is possible without the existence of any plan to traverse this environment, considered an obstacle. In the domain of the model, the presence of the FF group-agents in the forest has two consequences: to render more probable the penetration of this milieu; to increase the time spent there before arriving at the savannahs of the south or moving back to the north of the forest.

64The difference between the highly generic HU.M.E. model and the HU.M.E.-2 model associated with a precise case-study reflects the difference in generality between transitions 1 and 2. The first is associated with the departure from Africa and the conquest of the world by Homo sapiens without precision as to either the temporal and spatial scales or the geographical places concerned. The second concerns the expansion of a specific peopling (that of the Bantu) within a given geographical framework, and especially in a context of particular interactions (between Bantu and forest forager populations). The challenge, however, was to formalise an interaction which would be sufficiently general to carry across a period of 3,000 years, when in the course of ‘real’ history the relations between Bantu and forest foragers have likely evolved, without the mechanisms of these changes being clearly understood by specialists. The point was to find the common denominator in the different forms of interaction between Bantu and forest foragers over 3,000 years. The metaphor of the ‘link’ to designate and formalise these interactions responds to this challenge. It permits recognition of a dependence in the movements without having to specify a precise form of domination or subjection between the groups being modelled. Consequently, this model could be generalised to all objects which co-evolve within space.

Notes

1 Hombert Jean-Marie, Hyman Larry (Eds.), Bantu Historical Linguistics: Theoretical and Empirical Perspectives, Stanford, CSLI Publications, 1999.

2 Li Sen, Schlebusch Carina, Jakobsson Mattias, ‘Genetic Variation Reveals Large-Scale Population Expansion and Migration during the Expansion of Bantu-Speaking Peoples’, Proceedings of the Royal Society of London B (Biological sciences), 281, 2014, p. 20141448.

3 Grollemund Rebecca, Branford Simon, Bostoen Koen, Meade Andrew, Venditti Chris, Pagel Mark, ‘Bantu Expansion Shows Habitat Alters the Route and Pace of Human Dispersals’, Proceedings of the National Academy of Sciences of the USA, 112(43), 2015, pp. 13296–13301.

4 Vansina Jan, ‘New Linguistic Evidence and “the Bantu Expansion”’, The Journal of African History, 36(2), 1995, pp. 173–195.

5 Ehret, Christopher, ‘Bantu Expansions: Re-Envisioning a Central Problem of Early African History’, The International Journal of African Historical Studies, 34(1), 2001, pp. 5–41.

6 Plaza Stéphanie, Salas Antonio, Calafell Francesc, Corte-Real Francisco, Bertranpetit Jaume, Carracedo Ángel, Comas David, ‘Insights into the Western Bantu Dispersal: mtDNA Lineage Analysis in Angola’, Human Genetics, 115(5), 2004, pp. 439–447.

7 Alves Isabel, Coelho Margarida, Gignoux Christopher, Damasceno Albertino, Prista Antonio, Rocha Jorge, ‘Genetic homogeneity across Bantu-speaking groups from Mozambique and Angola challenges early split scenarios between East and West Bantu populations’, Human Biology, 83(1), 2011, pp. 13–38.

8 Bostoen Koen, Clist Bernard, Doumenge Charles, Grollemund Rebecca, Hombert Jean-Marie, Muluwa Joseph Koni, Maley Jean, ‘Middle to Late Holocene Paleoclimatic Change and the Early Bantu Expansion in the Rain Forests of Western Central Africa’, Current Anthropology, 56(3), 2015, pp. 354-384.

9 Ngomanda Alfred, Neumann Katharina, Schweizer Astrid, Maley Jean, ‘Seasonality change and the third millennium BP rainforest crisis in southern Cameroon (Central Africa)’, Quaternary Research, 71(3), 2009, pp. 307–318.

10 Some articles suggest the possibility of an accentuation, even an inducement, of climatic change by the practices of deforestation.

11 Bostoen Koen, ‘Pearl Millet in Early Bantu Speech Communities in Central Africa: A Reconsideration of the Lexical Evidence’, Afrika und Übersee, 89, 2006, pp. 183–213.

12 Neumann Katharina, Bostoen Koen, Höhn Alexa, Kahlheber Stefanie, Ngomanda Alfred, Tchiengué Barthelémy, ‘First farmers in the Central African rainforest: A view from southern Cameroon’, Quaternary International, 249, 2012, pp. 53–62.

13 Schwartz Dominique, ‘Assèchement Climatique Vers 3000 B.P. et Expansion Bantu En Afrique Centrale Atlantique : Quelques Réflexions’, Bulletin de la Société géologique de France, 3, 1992, pp. 353–361.

14 Bostoen Koen, Grollemund Rebecca, Muluwa Joseph Koni, ‘Climate-Induced Vegetation Dynamics and the Bantu Expansion: Evidence from Bantu Names for Pioneer Trees (Elaeis Guineensis, Canarium Schweinfurthii, and Musanga Cecropioides)’, Comptes Rendus - Geoscience, 345(7-8), 2013, pp. 336–349.

15 Ngomanda Alfred, Chepstow-Lusty Alex, Makaya Mvoubou, Favier Charles, Schevin Patrick, Maley Jean, Fontugne Michel, Oslisly Richard, Jolly Dominique, ‘Western equatorial African forest-savanna mosaics: a legacy of late Holocene climatic change?’, Climate of the Past Discussions, 5(1), 2009, pp. 341–367.

16 Maley Jean, ‘La destruction catastrophique des forêts d'Afrique centrale survenue il y a environ 2500 ans exerce encore une influence majeure sur la répartition actuelle des formations végétales’, Systematics and Geography of Plants, 71, 2001, pp. 777-796.

17 Russell Thembi, Silva Fabio, Steele James, ‘Modelling the Spread of Farming in the Bantu-Speaking Regions of Africa: An Archaeology-Based Phylogeography’, PLoS ONE, 9(1), 2014.

18 Grollemund Rebecca, ‘Nouvelles Approches En Classification : Application Aux Langues Bantu Du Nord-Ouest’, Thèse de doctorat de l’Université Lumière Lyon 2, 2012.

19 In order to simplify the text we will often use « forest forager » without specifying that we are referring in all cases to the Central African forest foragers. In the second part of the chapter concerning the development of a model, we will even use the shorter abbreviation FF.

20 Gowdy John, ‘Hunter-Gatherers and the Mythology of the Market’, in The Cambridge Encyclopedia of Hunters and Gatherers, ed. R. B. Lee and R. Daly, Cambridge, Cambridge University Press, 2004, pp. 391-398

21 Ichikawa Mitsuo, ‘Mbuti’, dans The Cambridge Encyclopedia of Hunters and Gatherers, ed. R. B. Lee and R. Daly, Cambridge, Cambridge University Press, 2004, pp. 210-214.

22 Hitchcock Robert K., ‘Introduction: Africa’, dans The Cambridge Encyclopedia of Hunters and Gatherers, ed. R. B. Lee and R. Daly, Cambridge, Cambridge University Press, 2004, pp.175-184.

23 Among the different articles on which we have relied, we may mention: Bahuchet Serge, ‘Les Pygmées D’aujourd'hui En Afrique Centrale’, Journal des africanistes, 61(1), 1991, pp. 5–35 ; Bahuchet Serge, ‘Changing Language, Remaining Pygmy’, Human Biology, 84(1), 2012, pp. 11–43 ; Bahuchet Serge, Guillaume Henri, ‘Aka-Farmer Relations in the Northwest Congo Basin’, in Politics and History in Band Societies, ed. E. Leacock et R. Lee, Cambridge, Cambridge University Press, 1982, pp. 189-211

24 Quintana-Murci Lluís et al., ‘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 of the USA, 105(5), 2008, pp. 1596–1601.

25 Patin Etienne et al., ‘Inferring the Demographic History of African Farmers and Pygmy Hunter-Gatherers Using a Multilocus Resequencing Data Set’, PLoS Genetics, 5(4), 2009; Patin Etienne et al., ‘The Impact of Agricultural Emergence on the Genetic History of African Rainforest Hunter-Gatherers and Agriculturalists’, Nature communications, 5, 2014, p. 3163.

26 Perry George H. et al., ‘Adaptive, Convergent Origins of the Pygmy Phenotype in African Rainforest Hunter-Gatherers’, Proceedings of the National Academy of Sciences of the USA, 111(35), 2014, pp. E3596–E3603.

27 Verdu Paul et al., ‘Sociocultural Behavior, Sex-Biased Admixture, and Effective Population Sizes in Central African Pygmies and Non-Pygmies’, Molecular biology and evolution, 30(4), 2013, pp. 918-937 ; Verdu Paul et al., ‘Origins and Genetic Diversity of Pygmy Hunter-Gatherers from Western Central Africa’, Current Biology, 19(4), 2009, pp. 312–318.

28 Diamond Jared M., ‘Anthropology. Why Are Pygmies Small?’, Nature, 354(6349), 1991, pp. 111–112.

29 Indeed, if a mutation occurs in an individual, it has a greater chance of prevailing in the group as a whole if the latter is smaller rather than larger. Groups of smaller size will therefore diverge more rapidly from each other genetically.

30 Diamond Jared, Bellwood Peter, ‘Farmers and Their Languages: The First Expansions’, Science, 300(5619), 2003, pp. 597–603.

31 Nettle Daniel, ‘Language Diversity in West Africa: An Ecological Approach’, Journal of Anthropological Archaeology, 15, 1996, pp. 403–438.

32 Since the model for exploring the expansion of Bantu peopling was constructed on the basis of the HU.M.E (HUman Migration and Environment) model developed in chapter 4, the same acronym has been used, with the new version distinguished only by the number ‘2’.

33 Li S., ‘Genetic Variation …’, op. cit.

34 Lake, Mark W., ‘Trends in Archaeological Simulation’, Journal of Archaeological Method and Theory, 21(2), 2014, pp. 258–287.

35 Hazelwood Lee, Steele James, ‘Spatial Dynamics of Human Dispersals Constraints on Modelling and Archaeological Validation’, Journal of Archaeological Science, 31(8), 2004, pp. 669–679.

36 Parisi Domenico, Antinucci Francesco, Natale Francesco, Cecconi Federico, ‘Simulating the expansion of farming and the differentiation of European languages’, in Origin and Evolution of Languages Approaches, Models, Paradigms, ed. B. Laks, S. Cleuziou, J.-P. Demoule and P. Encrevé, London, Equinox Publishing, 2008, pp.192-214.

37 Barton Michael C., Riel-Salvatore Julien, ‘Agents of change: modeling biocultural evolution in Upper Pleistocene western Eurasia’, Advances in Complex Systems, 15(1-2), 2012.

38 Such a search may extend over several days with the nights spent in sites somewhat removed from the principal camp.

39 Sapiens, in most of the places where he travelled (with the notable exception of Australia and the Americas), encountered other human beings, and the spaces were therefore not actually empty, even if the population densities were very low. The HU.M.E. model simplifies the reality by supposing that the spaces to be colonised were uninhabited.

40 For the FF agent-group, everything takes place as if it had to exploit the forest for a more important group, without this having consequence for the existing resource. Such a rule is justified by the nature of the resource transferred (hunting and gathering), which consumes less biomass that the agricultural exploitation of the Bantu agent-group.

41 In reality, at the present time, the Bantu have need of the FF chiefly for meat and for working in the fields. The FF, however, are dependent on the Bantu in order to live in the forest: iron tools, food produced by agriculture, without which living in the forest would be difficult, if not impossible. The choice has been made here, on the one hand, to introduce the symmetrical effect of the relation in a general and implicit way (each being attracted by the other) and, on the other hand, to spell out the link of domination through a material transfer of resources. We will return to these choices later in the discussion.

42 The Cobb-Douglas function is used in economics to express, in a multiplicative manner, the effects of different factors when each is affected by a coefficient.

43 Following Heckbert (2013), we use this term to designate an analysis which has not systematically explored all the space of the parameters, but in which exploration is guided by the results of preceding simulations in an interactive form.

44 Alves Isabel et al. ‘Genetic homogeneity across Bantu-speaking groups …’, op.cit.

45 The alternative hypothesis to the ‘early split’ is that of a direct crossing through the forest by all the Bantu groups, then, in a second phase, of a spatial dispersion of groups to the south and the east of the forest.

Table des illustrations

Titre Figure 1: Various possible migratory pathways for crossing the forest18.
Légende The trees in green represent the extent of the forest during the first phases of the Bantu migration. The arrows represent different migratory routes that the farmers may possibly have taken.
URL http://books.openedition.org/pufr/docannexe/image/19795/img-1.png
Fichier image/png, 748k
Titre Figure 2: Simulation grid of the HU.M.E.-2 model.
Légende 2.a: Introduction of two types of environment, savannah and forest (the grid thus corresponds roughly to that of the HU.M.E. model, with the addition of the rainforest). 2.b: Initial situation, with the FF group-agents in the forest and the flow of Bantu migrants penetrating into the grid by the north-west corner. 2.c: Situation after several time-steps.
URL http://books.openedition.org/pufr/docannexe/image/19795/img-2.jpg
Fichier image/jpeg, 94k
Titre Figure 3: Differences in behaviour between Bantu and FF group-agents
Légende 3.a: Differential capacities of the Bantu and FF group-agents to exploit the biomass 3.b: Evaluation of the resource potential of the destination zone for each type of group-agent.
URL http://books.openedition.org/pufr/docannexe/image/19795/img-3.png
Fichier image/png, 51k
Titre Figure 4: Formalisation of a reciprocal but asymmetrical ‘link’ between Bantu and FF group-agents: point of view of the Bantu group-agents
Légende (the left-hand column illustrates the situations before movement of the Bantu group-agent and the right-hand one the situation after the movement).
URL http://books.openedition.org/pufr/docannexe/image/19795/img-4.jpg
Fichier image/jpeg, 353k
Titre Figure 5: Basic principles for calculating the indicators of obstacle crossing.
URL http://books.openedition.org/pufr/docannexe/image/19795/img-5.png
Fichier image/png, 78k
Titre Figure 6: Variety of crossing times (for the group-agents having crossed the forest) as a function of the capacity to exploit the forest biomass.
URL http://books.openedition.org/pufr/docannexe/image/19795/img-6.png
Fichier image/png, 30k
Titre Figure7: Sensitivity of the simulation results to variations in the parameter ‘a’
Légende 7.a: Variation of the number of crossing attempts as a function of ‘a’7.b: Variation of the duration of the crossing as a function of ‘a’
URL http://books.openedition.org/pufr/docannexe/image/19795/img-7.jpg
Fichier image/jpeg, 85k

Auteurs

Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search