First archaeological evidence of banana cultivation in central Africa during the third millennium before present

Journal of Archaeological Science (2000) 27, 151-162 ®doi:1O.1006/jasc.1999.0447, available online at http://www.idealibrary.com on 1DEtl

Evidence for Banana Cultivation and Animal Husbandry Duringthe First Millennium Be in the Forest of Southern Cameroon

1 Christophe M. Mbida

University of Yaoundé I, Bp 755, Yaoundé, Cameroon

Wim Van Neer, Hugues Doutrelepont and Luc Vrydaghs

Royal Museum of Central Africa, B-3080 Tervuren, Belgium

(Received 9 March 1998, revised manuscript accepted 15 March 1999)

The faunal and botanical data from the first millennium Be site of Nkang, Southern Cameroon, are presented in thispaper. The analysed material, retrieved from refuse pits, comprises charcoal, phytoliths, molluscs and animal bones,which allow a reconstruction of the former environment. ln addition, the site provides new insights into the emergenceof food-producing communities in the African rainforest. Food procurement strategies at the site involved gathering,hunting, fishing, as weil as smalllivestock keeping and ban ana cultivation. This is the earliest evidence for such practicesin Central Africa. © 2000 AcademiePress

Keywords: ARCHAEOZOOLOGY, ARCHAEOBOTANY, RAINFOREST, PALAEOECOLOGY,FOOD PRODUCTION, CHAR COAL, PHYTOLITHS, CENTRAL AFRICA.

Introduction

The transition from foraging to food producingcommunities in Central Africa has receivedincreasing attention in recent years. The main

issue has been to understand the processes thatinitiated agriculture and animal breeding and to assesstheir social, economie and ecological repercussions.

ln almost every central African country a limitednumber of archaeological excavations have beencarried out. Although in several cases they wererestricted to a few test pits, caves and rockshelters asweil as open air settlements have been explored. lngeneral, caves and rockshelters tend to offer betterconditions for the conservation of organic materials,which rarely survive in the archaeological record ofopen air sites, due to the acidity of equatorial soils(Phillipson, 1985: 135; Van Neer, 1990: 195; Eggert,1993: 325; Schwartz & Lanfranchi, 1993: 38; Iliffe,1995: 16).Archaeological data from the sites of Matupi(Van Noten, 1977; Van Neer, 1989), Ishango (Brooks& Smith, 1987; Peters, 1990), Ngovo (de Maret, 1986)in Congo (Kinshasa), Shum Laka and Abeke (deMaret, Clist & Van Neer, 1987), Fiye Nkwi, Mbi(Asombang, 1988) in Cameroon, Tchissanga (Denbow,1990) and Ntadi Yomba in Congo-Brazzaville (VanNeer & Lanfranchi, 1986) and Otumbi in Gabon(Oslisly, 1992) have permitted sorne palaeoecologicaland palaeoeconomic reconstruction.

0305-4403/00/020151 + 12 $35.00/0

The latest research into the available evidence con-cerning early human activities in Central Africapresents a picture of foraging communities. Theearliest traces of animal husbandry occur only in verylate archaeological contexts dated to the secondmillennium AD (Van Neer, 1990). Although numeroussites occupied in the second and first millennium Beshow signs of exploitation of palm nuts (Elaeïsguineensisy and canarium nuts (Canarium schwein-furthiîy in southern Cameroon (de Maret, 1985a) andGabon (Clist, 1995: 154), no remains of cultivatedcrops or domestic animais have so far been retrievedfor that period.

The purpose of this paper is to present the archaeo-zoological and archaeobotanical data from the Nkangsite in Southern Cameroon. This site yields accurateinformation on human activities and allows an assess-ment of past human adaptation to the rainforest.Many aspects of the Nkang site shed new light on thelifestyle of ancient villages in Central Africa. lnaddition, they also contribute to the study of regionalpalaeoecology.

Site Description and ExcavationNkang (110 19'E, 40 16'N) is a rural settlement locatedabout 10 km east of Monatelé, the administrative townof the Lekié division, and 70 km to the north-west of

151© 2000 AcademiePress

152 C. M. Mbida et al.

Sa'a•~Emana.•. -,

• NKANG<, \ • Obala

\\

/ ~~ ~- - --.....~ Matomb ~ _ --- -e YAOÛNDE ~ - - - - - - - - " ~

<, __ --- v-..........: -- .,...... "\

1111

\1

/(

1\\

Ebolowa J-_J

~ 0 lI'i ao 4fi km/ ~I__ ~I __ ~I __ ~I

'\, / ~ <, A_

r <, <~hQh-.\ /' ".".'-- .

Ndikiniméki Bafia \,1,\

Ngambé

•

Eseka•

•NangaEboko

Nyong

\\/

1\ -,

\\\

Figure L Location of the Nkang site. (- - -) roads; (-) rivers.

Yaounde, the capital of Cameroon (Figure 1).Nowadays, the local inhabitants of the area are theEton who live in permanent villages along the roads.They are farmers, growing cacao, plantain, groundnut,cassava and maïze, and practising poultry farming andovicaprine breeding. Hunting, fishing and collectingare secondary activities.

The climatic pattern of the study area is equatorial'of the guinean type (Amou'ou Jam et al., 1985: 81).Annual precipitation is l360 mm, falling mostlybetween March and June, and September andNovember. Mean annual temperature ranges between23 and 25°C. The vegetation is a degraded rainforest.The landform is characteristic of the region, consistingof innumerable whale-back hills (Kadomura, 1977)around 500 m above sea level, gently sloping downtowards the Sanaga valley. Nkang village stretchesalong the top of a hill called Nkol Belibi Ndomo. Thecutting of a new tarred road exposed the substructureof ferralitic soils, resulting from in situ weathering of

a deep regolith, derived from garnet-mica-schistsmigmatized with two micas of the Precambrian era.

Roadwork cutting exposed many refuse pits on theembankment slopes (Elouga, 1985) which were notrelated to any other archaeological structure (Figure2). They were either bottle-like, ovoid or cylindrical inshape, and their volumes ranged from 3 to 15m'(Figure 3). A systematical study of the infilling pro-cesses of three pits was carried out. One of them, F9,yielded clear pedological and chemical evidence show-ing that it had initially contained standing water. Othershafts, close to F9, whose filling processes were notthoroughly analysed, shared sorne common character-istics. Their volumes were over 6 nr', their basaldeposits were bedded and reduced and the bonescontained within were weil preserved due to neutral orslightly alkaline pH. The primary uses of Fl3 and F14are subject to conjecture. Fl3 may have been a pit forthe processing of organic products and F14 mayhave been a trap (Mbida, 1996). When the primary

Banana Cultivation and Animal Husbandry in Southern Cameroon 153

W 30m1

201

101

o1

101

201

301

401

50 m1

® FOSSE (PIT)

---------=F10 •

~

®o 5mWill

tN

Figure 2. General view of the Nkang site. A, vertical section of the northern talus; B, plan of the site. Numbers refer to pits.

functions were abandoned, these features served asrefuse pits and were fiIled with local soil, brokenceramics and stone tools, iron slags, charcoal andfaunal remains.

The study material presented here cornes from theexcavated pit fill. The features were bisected along oneaxis, and both halves of the filling were removed inartificial strata of between 5 or 50 cm, depending onthe thickness of the horizons. All visible ceramic, lithicand faunal material was retrieved and stored in markedplastic bags and boxes. Charcoal fragments larger than2 mm were systematicaIly hand picked with pincers.Eight charcoal samples from six pits were taken forradiocarbon dating and soil samples were collectedfrom aIl visible horizons in the pit profiles.

Nkang possibly corresponds to the northern limit ofthe "Obobogo tradition" (Claes, 1985; de Maret,1991). The pottery from Nkang is characterized byspherical, ellipsoid and ovaloid vessels, the majority of

which have a coIlar or a neck. The rims are generallythickened in vessels with a neck. The proportion of thedifferent forms and the greater average thickness of thevessels distinguishes the Nkang pottery from that ofother sites of the Obobogo group in the region. Thelithic material consists mainly of fragmented objectsmade of quartz, quartzite, gneiss, granite, micaschistand dolerite. The recovered tools comprise grinders,grinding stones, hammerstones, whetstones and apolished axe. About 1 kg of iron slag was buried in pitsF7NF and F7bis, indicating that iron working was alsopractised during this period (Mbida, 1996).

Material and MethodsThe pit sediments comprise dumped refuse and runoffdeposits. Their pH, neutra1 to a1ka1ine, offers goodpreservation conditions for organic remains. Samp1es

E


--,-------------'\. - -/,,

F14

@@

o

- - -,---..••. - - - - - - -/ }

F7

o1

501

100 cm1

Figure 3. Cross-section of five pits from the Nkang site.

were processed by wet sieving(2mm-32 um) to recovermacro- and microbotanical and faunal remains.Contexts sampled for phytolith analysis included sedi-ments and charred residues on pottery fragments. Thelatter were cleaned with distilled water, scraped offand processed chemically to get rid of the organicmaterial. Sediment fractions between 150 and 75 um,between 75 and 32).lm and <32).lm were prepared forphytolith analysis by the classicalmethods as describedby Rovner (1971), Powers & Gilbertson (1987) and

Piperno (1988). Observations were made at magnifi-cations of 400 and 1000x .

Identification of the charcoal and phytolithmaterial was carried out with the aid of the modernreference collections and the identification keys of theLaboratory of Wood Biology and Xylarium (RoyalMuseum of Central Africa, Tervuren). For the bananaphytolith identifications, the reference material ofthe INIBAB Transit Center (Katholieke UniversiteitLeuven) was used. ln addition, several monographs


Table 1. Radiocarbon dates and 2 S.D. calibrations (OxCal version2·15)

Structure cal-BeDepth (cm) Lab. codeC-14 date(years BP)

Pit 6Pit 9Pit 9Pit 9Pit 3Pit 7 bisPit 13Pit 14

300-350250-300200-250

??

140-170190-210

Lv-1940Lv-1944Lv-1943Lv-1942Lv-1939Lv-1941Lv-1945Lv-1946

2580 ± 1102490 ± 1102490 ± 802400 ± 602420 ± 702340 ± 702310 ± 902170±80

850-410840-370790-400770-350770-350800-150800-100390-AD 1

and articles with figured material and identificationkeys were consulted (Metcalfe & Chalk, 1950;Metcalfe, 1960-1971; Twiss, Suess & Smith 1969;Piperno, 1988; Pearsall & Dinan, 1992; Rapp &Mulholland, 1992; Runge, 1996)

The faunal remains were identified through com-parison with the reference collections of the RoyalMuseum of Central Africa and facilitated by the use ofliterature on the postcranial osteomorphology andosteomorphometry of African mammals (Walker,1985; Peters, 1988; 1989; Van Neer, 1989). The body

size of the fishes was reconstructed through compari-son of the fish bones with modern skeletons of knownlength.

ResultsTable 1 lists the radiocarbon dates that were obtainedfrom the eight charcoal samples. Calibrated ages(2 S.D.) situate the occupation in the last millenniumBe (Mbida, 1996: 474). The results of the anthracologi-cal analysis are listed in Table 2 and the phytolithidentifications are given in Table 3. The hand-collectedfaunal remains are listed in Table 4, whereas the bonesretrieved through sieving are given in Table 5.

DiscussionPalaeoecology

Numerous plant taxa found at the site are indicatorsof a semi-evergreen, guineo-congolian rainforestwith mean rainfall ab ove 1200 mm per year: Albiziaferruginea, Anthostema senegalensis, Carapa procera,Chrysophyllum pruniforme, Entandrophragma sp.,Lannea antiscorbutica, Newtonia sp., Piptadeniastrum

Table 2. Plant taxa Jrom Nkang identified by charcoal analysis

PitF 1 F7NF

Plant taxaF3 F6 F9 F 14F13F7

AnacardiaceaeAntrocaryon micrasterLannea antiscorbuticaLannea welwitschii

ApocynaceaeStrophantus intermedius

ArecaceaeElaeis guineensis

BurseraceaeCanarium schweinfurthii

EuphorbiaceaeAnthrostemma senegalensisSpondianthus preussiiUapaca sp.

FlacourtaceaeCaloncoba welwitschii

MeliaceaeCarapa proceraEntandrophragma sp.Trichilia prieuriana

MimosaceaeAlbizia ealaensisAlbizia ferrugineaNewtonia sp.Piptadeniastrum aJricanum

OchnaceaeOchna multiflora

SapindaceaeChytranthus macrobotrys

SapotaceaeChrysophyllum pruniformis

x x

xx

x

x

x x x x x x

x x x x xx xx

xx

x

x

xx

x

xx

xx

x

x

x


Table 3. Plant taxa from Nkang identified by phytolith analysis

PitF7 F F7bis F9

Plant taxa

Monocot familiesCommelinaceae x xCyperaceae xMusaceae (Musa sp.) x xPalmae x x x xPoaceae x x x xZingiberaceae x x

Dicot familiesBorraginaceae x xMoraceae x xUlmaceae x x

africanum and Trichilia prieureana. Antrocaryonmicraster, Canarium schweinfurthii, Lannea welwitschiiand Strophantus intermedius occur in old secondaryforest, whereas Albizia ealaensis and Caloncobawelwitschii grow in pioneer secondary forest. Thesesecondary forest species suggest a more open habitat.Sorne of the trees, Ochna multifiora, Spondianthus

Table 5. Faunal remains retrieved from sieved samples in pit F9.Figures indicate number of specimens

F 13 Alestesl BrycinusCyprinidae indet.Perciformes indet.Unidentified fishUnidentified smalllizardsUnidentified small rodents

231

1327

preussii and Uapaca sp. prefer a more humid environ-ment (swamps or riparian forest). Chytranthus macro-botrys is found either in waterlogged sites or in uplandforest. The taxonomie diversity of wood species variesfrom pit to pit (Table 2). Most species occur in Fl4 andF9 whereas FI, F2, F6, F7, F7NF and F13 containonly a limited number of taxa. ln addition to theiroccurrence as charcoal, Elaeïs guineensis and Canariumschweinfurthii are represented in aIl the pits by charrednuts. AIl identified species are typical of the semi-evergreen lowland and gallery forests that may beobserved today in the region of Nkang.

The overall pattern of phytolith occurrence in theNkang samples (Table 3) suggests an open air habitat

Table 4. Animal taxa identified among the hand-collected faunal remains from Nkang. Figures indicate number ofspecimens

Animal taxa

PitF 1 F 3 F 5 F 6 F 7 F7bis F7NF F 9 F 13 Total

Freshwater gastropodsLanistes libycusPotadoma cf. freethii

Terrestrial gastropodsAchatina sp.Limicolaria sp.

Marine gastropodTrochidae indet

BivalvesAspatharia sp.

CrustaceansFreshwater crab (Decapoda indet.)

FishCatfish 1 (Chrysichthys sp.)Catfish 2 (Clariidae)Nile perch (Lates niloticus)

Wild mammalsSmall rodentsCane rat (Thryonomys sp.)Hippopotamus (Hippopotamus amphibius)Bushbuck (Tragelaphus scriptus)Waterbuck (Kobus ellipsiprymnusyKob (Kobus kob)Medium-sized duikers (Cephalophus sp.)Forest buffalo (Syncerus caffer nanus)

Domestic mammalsGoat (Capra aegagrus f. hircus)Sheep (Ovis ammon f. aries)Sheep or goat

Total identifiedUnidentified gastropodsUnidentified mammals

21

253

2 lû52

72

2 3

4 3 7311

7 85 5

2 7 101 2 4

12 21 1 27 10 4 28 3 17 3 74

1 1 3 58 15 1 25


in a secondary rainforest environment, from a qualita-tive as weIl as from a quantitative point of view. Theabundance of the phytoliths varies between less than 2and 2% on the basis of the table of visual estimation ofthe mineraI component quantities (Bullock et al.,1985). These 10w quantities point towards a forestedenvironment (Twiss, Suess & Smith, 1969; Twiss,1992). Botryoidal concretions (opal grains) are alsoindicative of a heavily wooded environment whereas anumber of other phytoliths point towards the presenceof open spaces. This is the case for the Zingiberaceae,Musa sp., the Palmaceae and the Poaceae identifiableas panicoid grasses.

Although less indicative than the botanical evidence,the faunal data from Nkang also contribute to theenvironmental reconstruction. The fauna compriseselements typical of closed and open environments andis therefore in accordance with the botanical data.Species typical of wooded areas include the land snailsAehatina, forest buffalo (Syneerus eaffer nanus), forestduikers (Cephalophus sp.), and bushbuck (Tragelaphusscriptus). Waterbuck (Kobus ellipsiprymnus) typicallyinhabits woodlands and clearings, whereas kob(Kobus kob) prefers savanna country and floodplains(Halthenorth & Diller, 1980). Both species usuallyoccur close to water.

PalaeoeeonomyLooking at the overall pattern of botanical andzoological evidence at the Nkang site, a few observa-tions can be made on the subsistence strategies and theimpact of human occupation on the vegetation and thefauna.

The plant remains that were studied comprise mainlycharcoal fragments and carbonized endocarps whichended up as refuse in the pits. The charcoal indicatesthat a wide variety of species were used for fuel andpossibly also for construction. Several plants wereexploited for their edible fruits, as is indicated by thephytolith evidence for Musa sp. and by the endo-carp finds of Antroearyon mieraster, Canariumsehweinfurthii, Chrytranthus macrobotrys and Elaeïsguineensis. Judging from their consistent presence inaIl the pits, it seems that the fruits of Canariumsehweinfurthii and Elaeïs guineensis were regularlyconsumed. ln addition, it could be demonstrated thatcanarium nuts were used in the decoration of pottery.Impressions of both the distal and the proximal sideswere found on pot sherds from pits F9 and F14.Certain plants may also have been used for purposesother than those mentioned. However, such practices,known from the ethnological record, can in the caseof Nkang not be proven archaeologically and theusages listed below remain therefore in the field ofspeculation. The seeds or the barks of Carapa proeera,Chrosophyllum pruniforme and Oehna multifloratraditionally have a medicinal function, while Carapaprocera, Triehilia prieureana and Oehna multiflora are

used as dye-plants, providing colourings for the humanbody or textiles. Strophantus intermedius produceswell-known poisonous fruits and Piptadeniastrumafricanum a poisonous sap. The trunk of canariumtrees exudes an inflammable resin, and the palm treenot only provides oil from the fruits and the kernelsbut is also a source of sugary sap, edible palm-heart,fibres and salt. An edible spinach grows on the palmtree and a decaying palm trunk harbours the ediblelarvae of beetles (Rhynehophorus phoenicis, Oryetesowariensisi which are traditionally exploited (Nkouka,1987; Linares, 1993).

The most important discovery in the archaeo-botanical assemblage is the Musa sp. phytoliths in thecarbon deposit of a pottery fragment (F7 NF C15) andin Pit F9 (horizons 2 and 7) (Figure 4). Stratigraphicalobservations in Pit F9 exclude major reworking due tobioturbation and the find adhering to the potteryfragment is very conclusive. The Musa phytolith camefrom a crust of charred organic matter. The surfaceof the crust was removed before further laboratorytreatment, thus excluding contamination of the crustcontent with postdepositional material. Firm morpho-logical differences between modern Musa and Ensetephytoliths were established using optical and scanningelectron microscopy (Doutrelepont et al., 1996). Theoccurrence of Musa sp. at Nkang is the first archaeo-logical indication of a cultivated crop for such an earlyperiod in Central Africa. It is not surprising that noother evidence for Musa has thus far been found, sinceits archaeological visibility is low. The ban ana plantdoes not produce pollen in Africa and its absence inpollen cores is therefore not significant. The tissues ofthe stipes or roots are not lignified and have never beenfound as macrobotanical remains, due to the extremelypoor preservation chances. The oldest historical sourcementioning the species in Africa dates to the 6thcentury AD and refers to the port of Adulis at theEthiopian coast (Vansina, 1991: 77). No wild ancestorsof Musa sp. occur on the African continent, meaningthat it must have been introduced (De Langhe, 1995).The origin of the wild banana and its centre ofdomestication is thought to be the region that stretchesfrom India to Papua New-Guinea and includesMalaysia and Indonesia. The edible bananas couldonly have been propagated in other parts of the worldthrough human intervention (De Langhe, 1995: 6). Thedomestication of Musa is a complex process thatinvolved seed sterility, parthenocarpy, interspecifichybridization and polyploïdization. This led to theformation of varieties and cultivars. AlI the bananas onthe African continent are seedless and triploïd, indicat-ing that it was the domesticated form that was intro-duced. The history of bananas and plantains and theirintroduction into Africa has so far been poorly docu-mented. It is believed that they were first introduced tothe continent via the eastern and north-eastern coast.The occurrence of Musa phytoliths at Nkang is clearevidence of cultivation in the last millennium Be and

(i)


(a) (b)

Figure 4. Lateral view of (a) a phytolith from pit F9 and (b) a modern Musa phytolith (reference number BS754).

gives further insight into the propagation of the culti-vars in Africa. This archaeological finding is consistentwith the hypothesis of a very long history of plantain inAfrica which is substantiated by the study of its uniquediversity in the rainforest and by the collective nameapplied to plantain in many Bantu language zones (DeLanghe, Swenen & Vuylsteke, 1995). The finds fromNkang indicate that agricultural practices in the rain-forest are much older than previously assumed. Theearly presence of banana in Cameroon could explainthe observed increase in village density and shouldimprove our understanding of the early stages of Bantuexpansion.

The majority of the faunal remains can be con-sidered human food refuse. Exceptions inc1ude thesmall rodents, the lizards and some of the shells. Therodent and smalllizard finds of pits F3, F9 and F13comprise in each case several bones of single indi-viduals indicating that we aredealing with animals thatare intrusive. They may represent individuals that fellinto the pits while these were still in use, or animalsthat burrowed in the archaeological layers and diednaturally in their burrows. Although Limicolaria canbe considered edible, there are no indications that thesesnails served as food at Nkang. They are often found athuman habitation sites which they colonize after theirabandonment (Gautier, 1983: 95). Since there is evi-dence that some of the pits had contained standingwater (Mbida, 1996: 481), it is likely that the fresh-water snails Lanistes libycus also represent individualsthat lived and died in the pits. The marine gastropod isalso not considered to be food refuse.

The anthropogenic faunal remains reflect differenteconomie activities carried out by the inhabitants ofNkang. They practised harvesting of molluscs, fishing,hunting and stock breeding. Both Achatina andAspatharia are molluscs that are regularly encounteredin African archaeological sites. Aspatharia and Pota-doma cf. freethii may have been collected from thenearby small tributary of the Lekie river, whereasAchatina are terrestrial, forest species that may haveoccurred near the site. These molluscs are edible butAchatina and Aspatharia are also widely used as con-tainers or as raw material for the production of beads.Several perforated dises made of Achatina shells werefound at Nkang in a single pit (F9). On the basis ofethnological analogies all over Africa, it is supposedthat they served either as pendants or as payment unitsin social exchanges (de Maret, 1985b: 166). A marinemollusc shell (Trochidae sp.) with a hole drilled nearthe axis of its basal aperture occurs in the same pit.Both the hole and the basal aperture show signs ofwear, probably resulting from rope friction. Themarine snail likely had similar uses as the Achatina,and was acquired through contacts with the Atlanticcoast about 200 km to the west, as the crow flies.

Despite the 10w number of remains, it is likely thatfishing was an important activity. The hand-collectedmaterial comprised only a few bones of large speciesbut the sieved samples from F9 all yielded fish remains.More extensive sieving would no doubt have signifi-cantly increased the number of fish bones. The avail-able hand-collected material comprises a caudalvertebra of a c1ariid catfish measuring about 1m

(0.


A B

Figure 5. Goat humerus from pit F6. (a) Dorsal view; (b) medial view. Scale bars are 1cm.

standard length (SL), as well as remains of a 60-70 cmlong Chrysichthys catfish and of a 70-80 cm long Nileperch (Lates niloticus). The latter species requires deepand well-oxygenated water, conditions which are notmet by the small tributary along which the site islocated. The Sanaga river, at about 10 km west ofNkang, is the closest locality where Lates occurs. Theinhabitants of Nkang may have practised fishing thereor may have obtained the Nile perch through exchangewith people living along the Sanaga. The clariid andChrysichthys catfish, on the other hand, are able tosurvive in shallow water and may therefore have beencaptured locally. A pincer of a freshwater crab foundin pit F7b further indicates that crustaceans were alsocollected from the river. The sieved samples from pitF9 yielded several bones of small cyprinids, a percoidfish and characids (Table 4) which may have a localorigin. The reconstructed sizes of the corresponding

fish are around 5 cm SL. The fish bones were discov-ered during the treatment of the sediment samples andoccurred inside concretions which were rich in calciumand phosphate. They most probably correspond to thecontents of human or animal excrements.

The majority of the faunalremains are derived fromwild mammals (Table 4). No archaeological objectsrelated to hunting have been found at Nkang. Smallerspecies, which may have been captured by snares ortraps, include cane rat and forest duikers, althoughthey may also have been caught through active hunt.Larger ungulates include bushbuck, waterbuck,kob,forest buffalo and hippopotamus and may have been

. captured in pits or hunted by groups using nets andwounding gear.

The presence of domestic sheep and goat is attestedby five bones only, suggesting that stock breeding wasa subsidiary activity. They were found in the following


Figure 6. Third phalanx of sheep from pit F3. (a) Peripheral view;(b) distal view. Scale bar is 1 cm.

pit structures: F3 (2 S.D. calibrated dates between 770and 350 cal Be), F6 (dated between 850 and 410 cal BC)and F7 bis dated to 800-150 cal BC. A goat humeruswith its proximal end unfused was found in F6 (Figure5). Its measurements (Bd 24·2 mm; BT 23·2 mm, takenaccording to the methods of von den Driesch (1976)) aswell as the overall size of the bone, indicate a rathersmall breed, comparable in size to the dwarf goatsfound in equatorial Africa today (Epstein, 1971: 211).The presence of sheep is attested in pit F3 by a distalfragment of a metatarsal belonging to a subadultspecimen, and by a complete third phalanx (Figure 6)with the following measurements: DLS 21·6 mm; Ld18·0mm. Both specimens belong to a small breedcomparable to the dwarf breed of thin-tailed hair sheepwhich is widely distributed today in the tropical forestof Africa (Epstein, 1971: 48). The domestic ovicaprinesfrom Nkang are the earliest evidence of stock breedingin Central Africa, but there are very few other archaeo-zoological data in the region allowing a precise recon-struction of the propagation of these domesticates(Van Neer, 2000). As a result of the increasing aridityof the late Holocene, a southward migration ofpastoralists from the Sahara occurred between roughly3000 and 2000 BC. This migration, as well as the furthersouth ward propagation of domesticates in the savannabelt, is well documented (Gautier, 1987). Small live-stock animals dated to the second half of the firstmillennium BC have been found at Daima in Nigeria,close to the Cameroonian border (Connah, 1976,1981). More extensive and more securely dated ma-terial is available from Gajiganna, Nigeria (Breunig,

1995; Breunig, Neumann & Van Neer, 1996), where theoldest sheep and goat remains occur in levels datedbetween about 1800 and 1500 cal-sc. Early goat boneshave also been reported from Ntereso and KintampoRockshelter 6 in Ghana (Carter & Flight, 1972). Thelevels from which these remains were retrieved datebetween about 2100 and 1300 BC at Ntereso, andKintampo Rockshelter 6 would compare closely indate to Ntereso (Stahl, 1985). The goat remains ofbothsites were described as a dwarf breed but firm morpho-metric evidence confirming this identification is notgiven in the original descriptions.

Conclusions

The botanical and zoological remains at the Nkang siteyield clear evidence of a food producing economy, withthe occurrence of ban ana or plantain phytoliths (Musasp.) and bones of domestic goat and sheep. Fishing,hunting and collecting remained important sources offood. The inhabitants maintained contacts and ex-changes within the immediate vicinity and with regionsas distant as the Atlantic coast. Both botanical andzoological evidence indicate a forest environment withopen spaces. The vegetation that people of the lastmillennium BC would have encountered at Nkang wasto a large extent similar to that which can be observedtoday, but the faunal environment has undergone amore drastic change. Many of the wild mammalsidentified in the archaeological record, such as hippo-potamus, forest buffalo, waterbuck, kob and forestduikers, have radically declined in number in theregion or have even become extinct locally. More siteswill be needed to further document the propagation ofdomestic species and to obtain additional informationon diet, ecology and human adaptation to the Africanrainforest over the last 3000 years.

AcknowledgementsThis paper is an excerpt from a Ph.D. thesis directed byProfessor Dr Pierre de Maret and completed in May1996 at Brussels Free University (Belgium) by the firstauthor. The contribution of Wim Van Neer to thisarticle represents research results from the Belgianprogramme on Interuniversity Poles of Attractioninitiated by the Belgian State, Prime Minister's Office,Federal Services. Many thanks are due to Dr JanMoeyersons, Dr Ir Hans Beekman, and Pr Dr E. DeLanghe for their cooperation. We are grateful to MrsGilberte Vendemmia for secretarial assistance andYvette Paquay for drawing the illustrations. Thephotographs for Figures 5 and 6 were taken by HansDenis (IAP). We are also indebted to Dr RaymondAsombang and Gina Griffith for comments on anearlier version of this paper. Dr Varsha Pilbrow kindlycorrected the English.


ReferencesAmou'ou Jam, J.-P., Melingui, A., Mounkam, J. & Tchepannou, A.

(1985). Géographie. Le Cameroun. Paris: Armand Colin.Asombang, R. N. (1988). Bamenda in Prehistory (The evidence from

Fiye Nkwi, Mbi crater and Shum Laka Rockshelters). Ph.D.Thesis, University of London.

Breunig, P. (1995). Gajiganna und Konduga. Zur frühen Besiedlungdes Tschadbeckens in Nigeria. Beitràge zur Allgemeinen undVergleichenden Archâologie 15, 3--48.

Breunig, P., Neumann, K. & Van Neer, W. (1996). New research onthe holocene settlement and environment of the Chad Basin inNigeria. The African Archaeological Review 13, 111-145.

Brooks, A. S. & Smith, C. (1987). Ishango revisited: newage determinations and cultural interpretations. The AfricanArchaeological Review 5, 65-78.

Bullock, P., Fedoroff, N., Jongerius, A., Stoops, G. & Tursim, T.(1985). Handbook of Soil Thin Section Description. WaineResearch Publications, Albrighton, Wolverhampton.

Carter, P. L. & Flight, C. (1972). A report on the fauna from the sitesof Ntereso and Kintampo Rock Shelter 6 in Ghana with evidencefor the practice of animal husbandry during the second millenniumB. C.. Man 7, 277-282.

Claes, P. (1985). Contribution à l'étude des céramiques anciennes desenvirons de Yaoundé. Mémoire de Licence, Université Libre deBruxelles.

Clist, B. (1995). Gabon: 100000 ans d'histoire. Condé-sur-Noizeau:Sépia.

Connah, G. (1976). The Daima sequence and the prehistoricchronology of the Lake Chad region of Nigeria. Journal of AfricanHistory 17, 321-352.

Connah, G. (1981). Man and a lake. ln Le Sol, la Parole et l'Ecrit.Mélanges en Hommage à Mauny. Paris: Société Françaised'Histoire d'Outre-Mer, pp. 161-178.

De Langhe, E. (1995). Banana and plantain: the earliest fruit crops?INIBAP Annual Report Paper 1, 6-8.

De Langhe, E., Swenen, R. & Vuylsteke, D. (1995). Plantain in theearly Bantu world. Azania 29-30, 147-160.

Denbow, J. (1990). Congo to Kalahari: data and hypotheses aboutthe political economy of the western stream of the Early Iron Age.The African Archaeological Review 8, 139-176.

Doutrelepont, H., Vrydaghs, L., De Langhe, E., Swennen, R.,Mbida, C., Janssens, B. & de Maret, P. (1996). Banana in Africa.ln (A. Pinilla, M. J. Machado & J. J. Tresserras, Eds) FirstEuropean Meeting on Phytolith Research. Madrid, Spain, 23-26September 1996 p. 27.

von den Driesch, A. (1976). A Guide to the Measurement of AnimalBones from Archaeological Sites. Peabody Museum Bulletin 1.Cambridge, Massachusetts: Harvard University Press.

Eggert, M. K. H. (1993). Central Africa and the archaeology of theequatorial rainforest: reflections on sorne major topies. ln (T.Shaw, P. Sinclair, B. Andah & A. Okpoko, Eds) The Archaeologyof Africa: Food, Metals and Towns. London: Routledge,pp. 289-329.

Elouga, M. (1985). Prospection archéologique dans la Lékié et étudeparticulière du site de Nkometou (Mfomakap). Mémoire deMaîtrise, University of Yaoundé.

Epstein, H. (1971). The Origin of the Domestic Animais ofAfrica. Vol. II. ew York/London/Munich: Africana PublishingCorporation.

Gautier, A. (1983). Animal life along the prehistoric ile: theevidence from Saggai 1 and Geili (Sudan). Origini 12,50-115.

Gautier, A. (1987). Prehistoric men and cattle in North Africa: adearth of data and a surfeit of models. ln (A. E. Close, Ed.)Prehistory of Arid North Africa: Essays in Honor of Fred WendorfDallas: Southern Methodist University Press, pp. 163-187.

Haltenorth, T. & Diller, H. (1980). A Field Guide to the Mammals ofAfrica Including Madagascar. London: Collins.

Iliffe, J. (1995). Africans: The History of a Continent. Cambridge:Cambridge University Press.

Kadomura, H. (1977). Sorne aspects of geomorphology in the forestand savannah areas of Cameroon, with special reference to

south-north variation. ln (H. Kadomura, K. Nakamura, T.Kikuchi & N. Hori, Eds) Geomorphological Studies in the Forestand Savannah Areas of Cameroon. Tokyo: Tokyo MetropolitanUniversity, pp. 7-35.

Linares, O. F. (1993). Palm oil vers us palm wine. Symbolic andeconomie dimensions. ln (c. M. Hladick, O. F. Linares, H.Pagezy, A. Semple & M. Hadley, Eds) Tropical Forest, People andFood. Paris: Man and Biosphere Series, 13, 595-606.

de Maret, P. (1985a). Recent archaeological research and dates fromCentral Africa. Journal of African History 26, 129-148.

de Maret, P. (1985b). Fouilles Archéologiques dans la Vallée duHaut-Lualaba, Zaïre. II Sanga et Katongo, 1974. Tervuren:Annales du Musée Royal de l'Afrique Centrale, SciencesHumaines 120.

de Maret, P. (1986). The Ngovo group: an industry with polishedstone tools and pottery in Lower Zaire. The African ArchaeologicalReview 4, 103-133.

de Maret, P. (1991). La recherche archéologique au Cameroun. ln(P. Salmon & G. S. Symoens, Eds) La recherche en scienceshumaines au Cameroun. Bruxelles: Académie Royale des Sciencesd'Outre-Mer, pp. 37-51.

de Maret, P., Clist, B. & Van Neer, W. (1987). Résultats despremières fouilles dans les abris de Shum Laka et d'Abéké auNord-ouest du Cameroun. L'Anthropologie 91, 559-584.

Mbida, M. C. (1996). L'émergence de communautés villageoises auCameroun méridional. Etude archéologique des sites de Nkang et deNdindan. Ph.D. thesis, Université Libre de Bruxelles.

Metcalfe, C. R. & Chalk, L. (1950). Anatomy of the Dicotyledons.Vols 1 & 2. Oxford: Clarendon Press.

Metcalfe, R. (Ed.) (1960-1971). Anatomy of the Monocotyledons(5 vols.). Oxford: Clarendon Press.

Ikouka, E. (1987). Les insectes comestibles dans les sociétés del'Afrique centrale. Muntu 6, 171-178.

Oslisly, R. (1992). Préhistoire de la moyenne vallée de l'Ogoué(Gabon). Ph.D. thesis, Université de Paris I.

Pearsall, D. M. & Dinan, E. H. (1992). Developing a phytolithclassification system. ln (G. Rapp Jr. & S. C. Mulholland, Eds)Phytolithic Systematics: Emerging Issues. New York: PlenumPress, pp. 37-64.

Peters, J. (1988). Osteomorphological features of the appendicularskeleton of African buffalo, Syncerus caffer (Sparrman, 1779) andof domestic cattle, Bos primigenius f. taurus Bojanus, 1827.Zeitschrift für Sdugetierkunde 53, 108-123.

Peters, J. (1989). Osteomorphological features of the appendicularskeleton of gazelles, genus Gazella Blainville 1816, bohor reed-buck, Redunca redunca (Pallas, 1767) and bushbuck, Tragelaphusscriptus (Pallas, 1766). Anatomia, Histologia, Embryologia 18,97-113.

Peters, J. (1990). Late Pleistocene hunter-gatherers at Ishango(Eastern Zaïre): the faunal evidence. Revue de Paléobiologie 9,73-112.

Phillipson, D. W. (1985). African Archaeology. Cambridge:Cambridge University Press.

Piperno, D. R. (1988). Phytolith Analysis: an Archaeological andGeological Perspective. San Diego: Academie Press.

Powers, A. H. & Gilbertson, D. D. (1987). A simple preparationtechnique for the study of opal phytoliths from archaeologicaland quaternary sediments. Journal of Archaeological Science 14,529-535.

Rapp, G. Jr. & Mulholland, S. C. (1992). Phytolith Systematics:Emerging Issues. Advances in Archaeological and Museum Science.Vol. 1. New York: Plenum Press.

Rovner, I. (1971). Potential of opal phytoliths for use in paleo-ecological reconstruction. Quaternary Research 1, 343-359.

Range, F. (1996). Opal Phytolithe in Pflanzen aus dem humiden undsemi-ariden Osten Afrikas und ihre Bedeutung für die Klima- undVegetationsgeschichte. Botanische Jahrbücher für Systematick 118,303-363.

Schwartz, D. & Lanfranchi, R. (1993). Les cadres paléoenvironne-mentaux de L'évolution humaine en Afrique centrale atlantique.L'Anthropologie 97, 17-50.


Stahl, A. B. (1985). Reinvestigation of Kintampo 6 rock shelter,Ghana: implications for the nature of culture change. The AfricanArchaeological Review 3, 117-150.

Twiss, P. C. (1992). Predicted world distribution of C3 and C4 grassphytoliths. ln (G. Rapp Jr. & S. C. Mulholland, Eds) PhytolithSystematics. Emerging Issues. Advances in Archaeological andMuseum Science. Vol J. New York, London: Plenum Press,pp. 113-128.

Twiss, P. c., Suess, E. & Smith, R. M. (1969). Morphologicalclassification of grass phytoliths. Soi! Science Society of America,Proceedings 33, 109-115.

Van Neer, W. (1989). Contribution to the Archaeozoology of CentralAfrica. Tervuren: Annales du Musée Royal de l'Afrique Centrale,Sciences Zoologiques 259.

Van Neer, W.' (1990). Les faunes de vertébrés quaternaires enAfrique centrale. ln (R. Lanfranchi & D. Schwartz, Eds) Paysagesquaternaires de l'Afrique centrale atlantique. Paris: ORSTOM,pp. 195-200.

Van Neer, W. (2000). Domestic animais from archaeological sites inCentral and West-Central Africa. ln (R. M. Blench & K. C.MacDonald, Eds), The Origins and Development of AfricanLivestock. London: UCL Press, pp. 163-190.

Van Neer, W. & Lanfranchi, R. (1986). Une association de faune etd'outillage Tshitolien (Age récent de la Pierre, 7000 B.P.) dansl'abri de Ntadi Yomba (Région du Niari) en R.P. du Congo.Eléments nouveaux pour un essai de reconstitution du paysagecongolais à cette époque. Comptes Rendues de l'Académie desSciences, Paris 302, 831-834.

Van Noten, F. (1977). Excavations at Matupi Cave. Antiquity 51,35-40.

Vansina, J. (J 991). Sur les sentiers du passé en forêt. Les chemine-ments de la tradition politique ancienne de l'Afrique équatoriale.Louvain-la-Neuve: Centre d'Histoire de l'Afrique, UCL.

Walker, R. (1985). A Guide 10 the Post-cranial Bones of East AfricanAnimais. Norwich, U.K.: Hylochoerus Press.

First archaeological evidence of banana cultivation in central Africa during the third millennium before present

Documents