new evidence and revised interpretations of early...

Res

ear c

h

839

New evidence and revised interpretationsof early agriculture in Highland NewGuineaTim Denham,1 Simon Haberle2 & Carol Lentfer3

This review of the evidence for early agriculture in New Guinea supported by new data from KukSwamp demonstrates that cultivation had begun there by at least 6950-6440 cal BP and probablymuch earlier. Contrary to previous ideas, the first farming in New Guinea was not owed to South-East Asia, but emerged independently in the Highlands. Indeed plants such as the banana wereprobably first domesticated in New Guinea and later diffused into the Asian continent.

Keywords: Mid-Holocene, New Guinea Highlands, agriculture, Musa bananas

Introduction

Several areas in the Highlands of New Guinea, in particular the Upper Wahgi Valley, aresignificant for interpreting the emergence of agriculture during the early to mid-Holocene.Of these sites, Kuk Swamp (hereafter referred to as “Kuk”) is the most important because ithas been investigated in greatest detail, provides the longest chronology and is the “type-site”for the investigation of prehistoric plant exploitation in the interior of New Guinea (Denham2003a; Denham et al. 2003; Golson 1977a, 1982, 1985, 1990, 1991a, 1991b; Golson &Hughes 1980; Hope & Golson 1995). The international significance of Kuk derives fromthe presence of successive phases of drainage for agriculture dating from 6950-6440 cal BPor earlier (Golson 1977a, 1991a, 1991b; Golson & Hughes 1980; cf. Denham 2004; Denhamet al. 2003; Table 1, see http://antiquity.ac.uk/ProjGall/denham). The evidence at Kuk hasprovided a basis for the interpretation of early and independent agricultural origins in theHighlands of New Guinea.

In this paper, new multi-disciplinary findings on early and mid-Holocene plant exploitationat Kuk, including archaeological, archaeobotanical and palaeoecological research, are presentedand interpreted with respect to the prehistory of the Upper Wahgi Valley. These results endorseprevious interpretations that New Guinea was a centre of independent agricultural origins,but question previous ideas that agriculture was necessarily developed in the lowlands andbrought into the Highlands with climatic amelioration and stabilisation at the beginning ofthe Holocene (Hope & Golson 1995: 818-819, 827-8). According to our alternative argument,

1 Department of Archaeology, School of Humanities, Flinders University, GPO Box 2100, Adelaide SA 5001, Australia.(Email: [email protected])

2 Resource Management in Asia-Pacific Program, Research School of Pacific and Asian Studies, Australian NationalUniversity, Canberra ACT 0200, Australia. (Email: [email protected])

3 Centre for Geoarchaeology and Palaeoenvironmental Research, School of Environmental Science and Management,Southern Cross University, PO Box 157, Lismore NSW 2480, Australia (Email: [email protected])

Received: 25 November 2003; Revised: 4 May 2004; Accepted: 9 May 2004

840

New evidence and revised interpretations of early agriculture in Highland New Guinea

agricultural practices arose in the Highlands of New Guinea during the early to mid-Holocenethrough the transformation of pre-existing plant exploitation strategies (following Denham& Barton in press; Haberle 1993: 299-305); the nature and timing of the earliest innovationsrequire clarification through further research.

The early and mid-Holocene remains, corresponding to Phases 1 and 2 at Kuk, are directlyrelevant to understanding the origins of agriculture in New Guinea. The early and mid-Holoceneremains pre-date any known Southeast Asian influence on New Guinea after its initial colonisationby at least c. 40 000 cal BP (Groube et al. 1986). Mid-Holocene Southeast Asian influences areusually associated with the expansion of Austronesian language-speakers to Melanesia at 3500-3300 cal BP (see Bellwood 1997: 210-54), or at c. 3200-3300 cal BP (Spriggs 2001: 240). Theinfluence of Austronesian diffusion on mainland New Guinea, marked by the Lapita culturalcomplex in island Melanesia (Green 1991), is unclear because there is a near absence of Lapitapottery and other well-dated traits on the island of New Guinea (see Terrell & Welsch 1997 forthe exceptions). On linguistic grounds, Austronesian influence on New Guinea is considered topost-date the onset of Lapita culture by at least 1000 years (Bellwood 1996: 487).

Site location

The wetland archaeological site at Kuk Swamp is located approximately 12 kilometres (km)north-east of the town of Mount Hagen in the Upper Wahgi Valley, Western HighlandsProvince, Papua New Guinea (Figure 1). Kuk is located in the Highlands, i.e., land above1200 metres above mean sea level (m), at an altitude of c.1560 m. Kuk Swamp forms part ofextensive wetlands carpeting the floor of the Upper Wahgi Valley, which is part of one of thelargest inter-montane valleys in the interior of New Guinea. The archaeological site is locatedon the former Kuk Agricultural (originally Tea) Research Station that covered 280 hectares(ha). Approximately 100 ha of the southeast corner were subject to multi-disciplinaryinvestigation (Golson 1977a: 610).

Like other wetlands in the Highlands of Papua New Guinea, Kuk was artificially drainedin the late 1960s and early 1970s for planting. Prior to drainage some parts of the wetlandwere underwater and surface gradients were generally “less than 1 in 200” (Golson 1977a:612), except for scattered low mounds of older, tephra-mantled lahar deposits over and aroundwhich the wetland had accumulated (Pain et al. 1987: 268).

The Upper Wahgi Valley has a lower montane humid climate with an average annualtemperature of 19°C and annual rainfall of c. 2700 mm (Hughes et al. 1991: 229). Climatein the Upper Wahgi Valley is dominated by local orographic effects (Powell et al. 1975: 2). Aslight dry season occurs in the Upper Wahgi Valley between May and June, although soilwater content does not usually limit plant growth (McAlpine et al. 1983: 74, 137).

Problems with previous research

Previous claims and evidence for early and independent origins of agriculture in New Guineahave been treated cautiously (Harris 1996: 568-9; Richerson et al. 2001: 400; Smith 1998:142-3), or been ignored (Harlan 1995). Doubts have been fostered by problems with thelimited archaeological, archaeobotanical and palaeoecological evidence presented. Only limitedarchaeological evidence associated with Phases 1 and 2 at Kuk was previously published (Bayliss-

Res

ear c

h

841

Tim Denham, Simon Haberle & Carol Lentfer

Figure 1 Map of New Guinea showing the location of important archaeological and palaeoecological sites and those mentionedin the text. The inset depicts wetland sites in the Highlands (amended with permission from Mountain 1991: Figure 2.1).

Smith 1996: 508; exceptions being Golson 1976, 1977a: 613-4, 1991a, 2000; Golson &Hughes 1980) and interpretations of the Phase 1 evidence changed through time (Bayliss-Smith 1996: 508-9). Consequently, the antiquity, mode of formation and function ofpalaeochannels and constituent features on the Phase 1 and 2 palaeosurfaces were uncertain(after Smith 1998: 142-3 and Spriggs 1996: 528). Findings contemporary with Phase 2 werereported at other sites in the interior, with some published in detail, e.g. Mugumamp (Harris& Hughes 1978), although all were interpreted in terms of the Kuk sequence (Denham 2003b).

Furthermore, the extent of landscape change coincident with the early and mid-Holocenedrainage phases at Kuk and anticipated to accompany agro-ecosystems was unknown (Spriggs1996: 528). Geomorphological investigations indicated accelerated rates of deposition onthe wetland margin from the early Holocene, which were suggestive of erosion followingvegetation clearance for swidden-type dryland cultivation in the catchment (Golson & Hughes

842


1980: 296-7; Hughes et al. 1991: 233-234). Powell’s pollen diagrams from three other sitesin the Upper Wahgi Valley, at Draepi-Minjigina, Lake Ambra and Warrawau, indicated thatforest clearance had occurred on a regional scale by the mid-Holocene (Powell 1982b: 218);the commencement and intensity of these activities were unknown because there was anabsence of a terminal Pleistocene and early Holocene component in the relevant UpperWahgi Valley pollen diagrams. Deposition rates at Kuk, as well as Haberle et al.’s (1991)evidence of cumulative forest clearance pre-dating 7800 cal BP in the Baliem Valley, suggestedthe Wahgi clearances could plausibly date to the early Holocene and be synchronous withPhase 1 archaeological remains at Kuk (Hope & Golson 1995: 823-5).

In addition, there was a near-absence of published archaeobotanical evidence of former cropplants, domesticated or otherwise (Bayliss-Smith 1996: 508; Spriggs 1996: 528). Macrobotanicalremains of several edible plants and Musa banana phytoliths were documented for archaeologicalphases at Kuk (Powell 1982a; Wilson 1985). The associations of these botanical remains wereuncertain because the plants could have dispersed independently of intentional human practices.

Each of these problems has been redressed by the recent investigations at Kuk.

Archaeological evidence of cultivation practices

Over 200 trenches were excavated at Kuk to delimit field systems and house sites (see Figure 2).Most excavations investigated later ditch networks (Phases 3-6) and comparatively few explicitly

Figure 2 Plan of the south-east of Kuk Station showing the extent of excavations directed by Jack Golson in 1972-7 and byTim Denham and Jack Golson in 1998-9.

Res

ear c

h

843


targeted early to mid-Holocene remains (Phases 1-2). The excavations undertaken in 1998 and1999 were intended to clarify contentious archaeological finds from previous work and toobtain samples for multi-disciplinary analyses.

Phase 1 at 10 220-9910 cal BP

Archaeological evidence unique to Kuk was claimed to represent management of the wetlandmargin for agriculture at c.10 000 cal BP (Golson 1977a: 613-5; 1989: 679; 1991a; Golson& Hughes 1980; Hope & Golson 1995: 824). Formerly, the archaeology of Phase 1 consistedof “chronologically and perhaps functionally associated” (Golson 1981: 56) palaeochannel andpalaeosurface. The palaeochannel was 2 m wide and 1 m deep and traced for at least 500 macross the wetland (Golson 1977a: 613-5; 2000: 232). A human origin for the palaeochannelwas inferred from several characteristics including “directness of line between points of abruptchange in direction, one of them in a middle of a low hillock of volcanic ash” (Golson 1991a:485) and by analogy with later palaeochannels, which were similarly interpreted to be artificial(Golson 2000: 232). More recent research has challenged this interpretation, and here theearly palaeochannel is not considered to be artificial (Denham 2004: 47-53). The aggregatesand abundant macrobotanical remains, as well as pollen and microcharcoal signals, withinpalaeochannel fills are consistent with forest disturbance using fire and erosion within thecatchment (see below).

The palaeosurface, by contrast, consists of pits, runnels, stake- and post-holes associated withsparse artefacts, relatively heterogenous feature fills and an immature soil profile (Figure 3b).The artificiality of palaeosurface features has been determined from the morphology of featureswhich have well-defined, cut edges. The inferred functions of features are consistent with plantingand digging (pits), localised overland flow (runnels) and staking of plants (stake and post holes)within a cleared plot (after Powell et al. 1975: 24-6). The slightly heterogeneous feature fillsand pedogenesis are consistent with limited mechanical admixture and the formation of immatureA/C profiles within an alluvial setting (after Collins & Lavney 1983; Haywood 1985). Lastlythe artificial character of the palaeosurface is confirmed by the presence of stone tools, whichdisplay evidence for the processing of starchy and woody plants and include a flake used toprocess taro (Colocasia esculenta) (Denham et al. 2003: 191).

Based on excavations to date, features are restricted to higher ground adjacent to thepalaeochannel. This area was only dry enough to utilise for a short period, perhaps onlyonce, prior to abandonment. In contrast to previous portrayals, these features need notrepresent specialised wetland management, but may be the spatial extension of plantexploitation practices from adjacent dryland slopes.

The plant exploitation practices represented by early Holocene features are unclear. Featurescould be associated with relatively adventitious plant management and use, although from adifferent perspective they could be consistent with practices in an inter-cropped, swidden-type plot on the wetland edge. Without additional investigations, it is not possible to determineif these features represent “agriculture”, as previously claimed, or other activities.

844


Figu

re 3

Arc

haeo

strat

igra

phic

sum

mar

y of

Pha

ses 1

and

2 a

t Kuk

(m

odifi

ed w

ith p

erm

issio

n fr

om D

enha

m e

t al.

2003

: Fig

ure

2). F

ield

pla

ns w

ere

draw

n by

Art

hur

and

Che

rie

Rohn

.Ph

otog

raph

s wer

e ta

ken

by E

.C. H

arri

s in

1977

and

repr

oduc

ed c

ourt

esy

of Ja

ck G

olso

n, R

ecor

ds o

f the

Aus

tral

ian

Mus

eum

(ph

oto

b) a

nd A

rcha

eolo

gy in

Oce

ania

(ph

oto

c).

Res

ear c

h

845


Phase 2 at 6950-6440 cal BP

Mid-Holocene finds at Kuk represent the earliest, unequivocal evidence for agriculture inNew Guinea (Denham et al. 2003; Golson 1977b: 48-9, 1989: 679, 1991a: 484, 489; Hope& Golson 1995: 823). Phase 2 at Kuk was formerly characterised as consisting of at leastfour successive palaeochannels that articulated with palaeosurfaces comprised of inter-connected networks, or more dispersed features (Golson 1977a: 615-9). As with the earlyHolocene palaeochannel, there is insufficient evidence to determine that the mid-Holocenewatercourses were human-made (Denham 2003b: 163-4). Palaeochannel characteristics,including slightly sinuous planforms, cross-sectional morphologies and long-profile gradients,are arguably consistent with unmodified watercourses.

The Phase 2 palaeosurface has well-connected (integrated) and less-organised (discrete)areas. The integrated areas have been studied in detail and consist of regular-to-semi-regularnetworks of cut features defining upraised areas, previously referred to as raised “island beds”(Golson 1977a: 616), against the underlying grey clay (Figure 3c). Similar mid-Holocenepalaeosurfaces are preserved at Warrawau (Golson 2002) and Mugumamp (Harris & Hughes1978) in the Upper Wahgi Valley, with proximal locales at Kana (Muke & Mandui 2003)and Ruti (Gillieson et al. 1985) providing more equivocal evidence (see Figure 1; reviewed inDenham 2003b: 169-74). The preserved “island beds” represent the truncated bases of formermounds constructed on the wetland margin for cultivation. At the time of initial formation,grey clay was probably a B-horizon, with the original topsoil only surviving as displacedremnants within feature fills. X-radiographs and thin sections depict greater heterogeneitywithin fills in comparison to adjacent stratigraphic units, which lends support to previousinterpretations of mound construction.

Artefacts collected from Phase 2 contexts are sparse, which is anticipated at an agriculturalsite in Highland New Guinea where wooden tools were more probably used (Golson 1977c).Lithic artefacts were used to process Colocasia taro, probable Zingiber gingers and otherunidentified plants (Denham et al. 2003: 191-2, Table S3). Additionally, one palaeosurfacefeature exhibits anomalously high Musaceae (banana) phytolith frequencies, including seedphytoliths of Eumusa, which are interpreted to represent deliberate planting within grasslandmaintained by burning (Denham et al. 2003: 191-2, Figure 3).

Previous interpretations suggested that mounded cultivation enabled the multi-cropping ofplants with different edaphic requirements (Golson 1977a: 617). Water-tolerant plants, e.g.Colocasia taro, could potentially have been planted within the damper features between mounds,whereas water-intolerant plants, e.g. sugarcanes (Saccharum spp.), Musa bananas, yams (Dioscoreaspp.), edible pitpit (Setaria palmifolia) and mixed vegetables could have been planted and stakedon the “island beds” (Golson 1977a: 616, 1981: 57-8; Powell et al. 1975: 42). Archaeologicaland archaeobotanical evidence is consistent with this agricultural interpretation.

Palaeoecology and landscape transformation

Palaeoecological studies in New Guinea rely on a series of criteria to distinguish humanimpacts from natural processes in a record of vegetation history. These primarily include theidentification of processes that did not previously occur in the palaeoecological record suchas indications of forest decline and burning, and increases in secondary forest and herbs

846


(Haberle 1994). Pollenrecords from Lake Ambra,Draepi-Minjigina andWarrawau (Manton’s)Plantation in the UpperWahgi Valley documentdisturbance to the primaryforest that pre-dated 6200-5700 cal BP (Powell 1970a:199, 1980: 19, 1981: 306,1982b: 218; Figure 4). LakeAmbra and Draepi-Minjigina signalled a changefrom undisturbed primaryforest in the late Pleistoceneto a disturbed environmentwith higher percentages ofwoody non-forest species bythe mid-Holocene (Zone III;Powell 1970b: 165-86, 1981:306). Similarly, the M1 corefrom Warrawau depicted theestablishment of secondaryforest from the beginning ofthe pollen diagram at 5500cal BP (Powell 1970b: 155-9; Powell et al. 1975: 43-4,46-8). Secondary forest wasrepresented by lightdemanding species includingTrema sp., Acalypha sp.,Macaranga sp. and Dodonaeasp.

These Wahgi Valley sitesrecord further declines inforest cover at different timesover the following 4000 years(Powell 1982b: 218),suggesting a mosaic andtime-transgressive pattern offorest clearance activitythroughout the mid and lateHolocene. Other studiesacross the Highlands also

Figure 4 Diagram of stratigraphy, inferred chronology and summary arboreal pollenwith Poaceae (pollen sum for all taxa based on total forest and woody non-foresttaxa) from three swamp sites in the Upper Wahgi Valley (redrawn and amendedwith permission from Powell 1982b: 219, Powell 1982b: 221 and Powell 1970b:Figure 8.7, respectively). Solid lines mark comparable pollen assemblages at the threesites based on pollen assemblages and correspond approximately to inferred ages givenin the original texts: Zone I (before 21 000 cal BP), Zone II (c. 21 000-6000 calBP) and Zone III (c. 6000 cal BP-present). Calibration of 14C ages after 19 200 BPare based on Stuiver and Reimer (1993) and before 19 200 BP are derived from alinear conversion following Bard et al. (1990).

Res

ear c

h

847


show evidence of mid-Holocene forest clearance(Haberle 1994: Figure 8.2).Of most significance, humanclearance of primary forest atKelela Swamp in the BaliemValley occurred prior toc. 7800 cal BP, withcontinued disturbance andthe establishment ofextensive grasslands by 3000cal BP (Haberle et al. 1991).

The early Holocene gap inthe pollen records from theUpper Wahgi Valley has beenfilled by recent research atKuk. Samples collected froma core extending from theLate Pleistocene to lateHolocene, as well as thosecollected from the fills ofpalaeosurface features andpalaeochannels, enabledetailed reconstructions ofpalaeoenvironments duringthe early and mid-Holocene(Figure 5). A compositediagram incorporating apreviously unpublishedpollen diagram from Kuk(Powell 1984) illustrates theswamp vegetation historyfrom before the Last GlacialMaximum (LGM) throughto the present (Figure 6).

Cold-adapted vegetationin the form of grass and sedgedominated wetlandspersisted through to the earlyHolocene at least partiallyunder the influence of fire(zone K-1). In the earlyHolocene (zone K-2), the

Figure 5 Composite pollen diagram showing selected pollen and spore taxa (per centbase on total pollen and spore sum excluding aquatics) plotted as sample numbersand ordered according to archaeostratigraphic associations (left of diagram). TheMusaceae phytolith frequency curve is derived from the per cent of total phytoliths(counted separately). Significant pollen zones (K-1 to K-4) are shown on the rightside of the diagram. Calibrated age ranges based on Denham et al. 2003: Tables S1and S2.

848


mixed-montane forest is replaced by a more open forest environment with Pandanus(probably P. antaresensis) and Schefflera as important swamp forest components. Between 10200 cal BP and 7400 cal BP (zones K-2 and K-3), and during deposition of a distinctivegrey-brown clay on the wetland margin, grasslands and fern flora increase at the expense offorest which is under the influence of periodic fire episodes. At the same time, the catchmentforest composition changes from dominance of montane canopy taxa, such as Nothofagus,Castanopsis and gymnosperms to subcanopy taxa, particularly Pandanus. At the beginning ofthe mid-Holocene, at around 7000 cal BP (zone K-4), there is a rapid loss of forest associatedwith increased burning and an open grass-sedge swampland is established.

The new palaeoecological evidence from Kuk, in conjunction with previous diagrams atDraepi-Minjigina, Lake Ambra and Warrawau, indicates that forest clearance, widespreaduse of fire and the establishment and maintenance of disturbed environments were regionalprocesses across the Upper Wahgi Valley from at least 7000-6500 cal BP. Taken in conjunctionwith archaeological evidence for prehistoric mounding at Kuk, Mugumamp and Warrawau,this vegetation history represents the establishment of an agricultural landscape in the UpperWahgi Valley by the mid-Holocene.

Figure 6 Palaeoenvironmental reconstruction for Kuk Swamp catchment based on a composite pollen diagram of archaeologicaland stratigraphic samples (see Figure 5) and a previously unpublished core in A10 f/g (amended with permission from Powell1984). The diagram correlates Powell’s pollen zones at Kuk and other Upper Wahgi Valley sites (I-III) with Haberle’s pollenzones (K-1 to K-4) at Kuk.

Res

ear c

h

849


Archaeobotanical evidence of plants

Archaeobotanical investigations at Kuk employed macrofossil (e.g. seed, wood) and microfossil(e.g. phytolith, pollen, starch grain) techniques. The archaeobotanical work documents awide range of traditional food plants in the Kuk vicinity at various times in the past (Table 2,at http://antiquity.ac.uk/ProjGall/denham). Residue analysis of the used edges of stone toolsfrom early and mid-Holocene contexts indicates that Colocasia taro, cf. Zingiber gingers, andother starchy and woody plants were being exploited (Denham et al. 2003: 191).

The most significant early Holocene additions to the suite of edible plants present in theKuk vicinity are Colocasia taro and Eumusa bananas (Denham et al. 2003: 191-2; cf. Wilson1985: Table 3, at http://antiquity.ac.uk/ProjGall/denham). Colocasia taro has been identifiedfrom starch residues on the cutting edge of a flake collected from the fill of an early Holocenefeature, as well as on artefacts collected from within grey clay and the fill of a Phase 2 feature(Denham et al. 2003: 191). Eumusa section banana phytoliths are present at the base of greyclay sealing a 10 000 year-old palaeochannel (see Figure 5). Intact Musa phytolith chainswithin grey clay contexts are significant because they signify a local origin. Articulated chainsof Musa phytoliths are fragile and are suggestive of local in situ decomposition and reflectlimited post-depositional disturbance and translocation. Peaks within early Phase 2 contextsare, given their archaeological associations, more important and show banana phytoliths,including those of Eumusa section, accounting for up to 9 per cent of all phytoliths. Giventhe low productivity of phytoliths in bananas relative to grasses, elevated frequencies in Phase2 contexts have been interpreted to represent the cultivation of bananas on the wetlandmargin within a landscape degraded to grassland (Denham et al. 2003: 191-2).

In New Guinea, Colocasia taro is presumed to be a lowland plant due to greater phenotypicvariation among contemporary wild populations at lower altitude (Yen 1991: 77, 1995:835). The present-day altitudinal range of Colocasia taro is assumed to be a product ofagronomic selection (Yen 1995: 843). Similarly, wild Eumusa banana populations only occurat lower altitudes than Kuk in New Guinea (Argent 1976; Simmonds 1962), although likeColocasia taro, they may have spread to higher altitudes independently of human action aspost-LGM climates ameliorated.

Although the mechanisms for the diffusion of Colocasia taro and Eumusa section bananasto the Highlands are unclear, these plants were potentially important sources of starch forinhabitants in the early Holocene. Earlier occurrences of these and other plants may yet befound given the absence of artefacts and the lack of detailed multi-disciplinary investigationsof stratigraphy dating to the Late Pleistocene at Kuk and other Highland sites. The lack ofcorrelation between archaeobotanical remains and archaeological phases at Kuk indicatesthat plant exploitation was not restricted to the wetland margin during the early and mid-Holocene. Most edible plants in the Highlands are traditionally cultivated asexually fromwild populations in the vicinity and, therefore, many are anticipated to be continuallyrepresented in the archaeobotanical record.

Colocasia taro and Eumusa bananas, along with breadfruit (Artocarpus altilis), sago(Metroxylon sagu), sugarcanes (Saccharum spp.) and some yam species (Dioscorea spp.), arethought on phytogeographic and genetic grounds to be indigenous to, and domesticated in,the New Guinea region (De Langhe & de Maret 1999; Lebot 1999; Matthews 1991, 1995,2003b). Indeed, Lebot states (1999: 626): “New Guinea and some parts of Melanesia have

850


been areas of domestication and still are important centres of diversity for species that also exist inAsia”. The archaeobotanical findings from Kuk and other sites for Colocasia taro (Haberle1995; Loy et al. 1992) provide strong corroborating evidence for these interpretations andsuggest that the New Guinea region was a major centre of plant domestication.

Reconsidering the emergence of agriculture in New Guinea

Timing: early or mid-Holocene?

Early evidence of occupation at several sites in the Highlands dates to the terminal Pleistoceneand early Holocene (Golson 1991b: 88-9). Although, in part, this trend may be apparentand reflect a “temporal-decay effecy”, i.e. comparatively fewer open sites survive with increasingage, it may also reflect greater populations of mobile groups living in the Highlands. Basedon the inventory at Kuk, a broad range of edible plants was present in the Highlands duringthe Late Pleistocene (see pre-P1 in Table 2), which together with hunting and foraging forinsects and grubs could have facilitated permanent occupation of higher and lower altituderegions of the interior following initial colonisation of the island (cf. Groube 1989).Throughout this period, mobile groups may have ranged across the interior via the expansiveglacial grasslands that connected inter-montane valleys (Hope & Hope 1976) and utilisedlowland resources, such as sago (Metroxylon sagu), as well as those in the Highlands.

At the end of the LGM and prior to stabilisation in the early Holocene, Highland climatesare considered to have fluctuated considerably and to have been subject to greater interannualvariability than is experienced in the current El Niño-related climate regime (Haberle et al.2001). As suggested for other parts of the world (Bellwood 1996), fluctuating climates afterthe LGM may have stressed existing plant management practices in Highland New Guineaand necessitated more interventionist (i.e. greater ground preparation, tending and weedingof plants) and more extensive (i.e. greater disturbance of forest canopy to increase habitatdiversity) plant exploitation strategies in order to maintain yields and a broad-spectrum diet(Haberle 1993: 305-8). Such strategies may be represented by the asynchronous and non-cumulative records of human-induced forest disturbance using fire in the Highlands in theLate Pleistocene (e.g. Haberle 1998; Hope 1983; Powell 1984). Additional “forcing” occurredduring the early and mid-Holocene as people sought to obtain food in increasingly degradedand resource-poor environments on the floors of the larger inter-montane valleys. Thesignatures of more extensive and interventionist strategies are apparent during the early andmid-Holocene at Kuk, as well as at other archaeological and palaeoecological sites at wetlandsin the Highlands.

During the early Holocene (10 200-9900 cal BP) at Kuk, and perhaps elsewhere in theUpper Wahgi Valley, people used fire to increasingly disturb and modify the mixed montaneforest and soils on the valley walls. The resultant landscape comprised a continually changingmosaic of primary and secondary forest, regrowth and grasslands on the dryland slopes andvariously disturbed riparian and wetland environments. Several high-caloric and nutritiousplants were present in the Kuk vicinity including Castanopsis spp., Colocasia taro, Musa bananas(including identifiable members of Eumusa section with the onset of grey clay deposition),Pandanus spp., cf. Setaria palmifolia, and cf. Zingiberaceae. These plants, supplemented byhunting and foraging, could have been sufficient to sustain broad-spectrum diets for permanent

Res

ear c

h

851


occupants of the Highlands. At this time, people were adventitiously exploiting slightly raisedand drier ground adjacent to a palaeochannel at Kuk, where there is evidence suggestingdigging and staking within a plot and the exploitation of Colocasia taro and other starchyand woody plants.

By the mid-Holocene (6950-6440 cal BP), the forests on the floor of the Upper WahgiValley had been altered to grassland or to secondary regrowth. Within relatively nutrientpoor grassland environments, people gradually increased their degree of intervention in themanagement of significant food plants, especially high-energy staples. Most plants could stillbe collected or transplanted from primary and secondary forest and regrowth habitats in thevalley. At this time, mounds were made within cleared plots along the wetland edge andplanted with bananas and other inter-cropped plants. Colocasia taro, cf. Zingiber gingers andother starchy and woody plants were processed, presumably for consumption.

Multi-disciplinary lines of evidence provide an impression of plant exploitation on thewetland margin in the early Holocene. Similar palaeoecological signals, but withoutaccompanying archaeological remains along the wetland edge, persist for the next 3-4000years during grey clay deposition. Although suggestive of swidden-type cultivation (Denham2004), the evidence is currently insufficient to determine whether these early Holocenepractices are comparable to contemporary agricultural practices in the region (cf. Bourke2001; Brookfield & Hart 1971: 94-124; Powell & Harrison 1982; Powell et al. 1975). Bythe mid-Holocene, a clearer picture of agricultural practices emerges and their effects areclearly documented in the cultivation and use of plants (e.g., bananas and Colocasia taro,respectively), some of which are considered to have undergone domestication in New Guinea,and in the dramatic transformation of the environment with the degradation of forest tograssland. Archaeological remains of prehistoric mound preparation on the wetland edgeprovide the foundation for interpretations of the archaeobotanical and palaeoecologicalevidence. Taken together, the multi-disciplinary lines of mid-Holocene evidence arecomparable to contemporary agricultural practices and their effects in the Highlands of NewGuinea.

Location: Highlands or lowlands?

Lowland environments and inhabitants were not subject to as severe climate-induced stressesas the Highlands during the terminal Pleistocene and the Holocene. Lowland inhabitantshad continuous access to a broad range of fauna and nutritious and high-caloric food plants.Of most significance are a range of indigenous nut-bearing (e.g., Artocarpus altilis, Barringtoniaspp., Canarium spp., Cocos nucifera, some Pandanus spp., Terminalia spp.) and starch-bearingtrees (e.g., Metroxylon sagu), which have restricted lowland distributions (Bourke 1996; Yen1995: 833-40). Additionally, a broad range of terrestrial, riverine, lagoonal and littoral fauna,including numerous freshwater fish, reptiles and shellfish, have distributions restricted to thelowlands. The relative abundance of edible tree resources led to an early focus on arboriculturein the lowlands, possibly from the terminal Pleistocene (Yen 1996: 41). A focus on managedstands of trees supplemented by hunting, fishing, foraging and garden horticulture createddiverse practices some of which are not readily classified as agricultural or hunter-gatherer(Dwyer & Minnegal 1991; Roscoe 2002), but are “ambiguous” in terms of such traditionalterminology (Terrell 2002; Terrell et al. 2003).

852


The diversity of present and, by extension, past practices blurs any clear distinction betweentypes of Highland and lowland subsistence. In the Highlands, however, there is a greater relianceon garden horticulture based on the vegetative propagation of vegetables and root-crops. Gardenhorticulture is practiced more intensively and forms the subsistence base in the Highlands,whereas in many lowland locales it supplements arboriculture and other activities.

Multiple lines of archaeological, archaeobotanical and palaeoecological evidence suggestthat agriculture in New Guinea, manifest as extensive (swidden) or intensive (semi-permanentand short fallow wetland and raised bed) cultivation, originated in the Highlands. Currently,however, there is a lack of comparable evidence for lowland subsistence, particularlyarboriculture, to test this proposition (for exceptions see Swadling et al. 1991; Yen 1996).There is little archaeological and palaeoecological evidence to indicate the expansion ofagriculturalists from the lowlands brought plants and horticultural practices into the Highlandsduring the terminal Pleistocene or early Holocene. Although the altitudinal range of Colocasiataro and some Musa bananas lies dominantly within the lowlands implying a lowland originfor these plants, the upper altitudinal limits lie within or close to the Highland valleys. Theappearance of these starch-rich plants at Kuk at the beginning of the Holocene may relate tonatural dispersal processes (i.e. seed-dispersed under warming climate conditions) as muchas to human agency (i.e. brought by mobile groups engaged in an extensive form of plantexploitation based on vegetative propagation).

Conclusion

Recent multi-disciplinary investigations have filled out previous, largely speculativechronologies of prehistoric plant exploitation (Harris 1995) and confirm previousinterpretations that agriculture arose independently in New Guinea (Table 3). New multi-disciplinary investigations at Kuk have advanced previous interpretations of prehistoric plantexploitation in the Highlands of New Guinea in two significant ways.

First, multi-disciplinary lines of evidence (including archaeological, archaeobotanical,palaeoecological and stratigraphic) show that agriculture was practiced in the Upper WahgiValley by at least 6950-6440 cal BP and probably much earlier. The evidence for earlier plantexploitation practices at Kuk is insufficient to be definitive of agriculture, as previously claimed(Golson 1977a: 613-5, 1991a; Golson & Hughes 1980; Hope & Golson 1995: 824). Thetiming of the emergence of agriculture in the Highlands of New Guinea requires clarificationthrough greater interpretative resolution of the Phase 1 record at Kuk and the excavation ofadditional sites with evidence of early Holocene plant exploitation.

Second, agriculture may have emerged in the Highlands of New Guinea as opposed to thelowlands. Golson suggested that agriculture originated in the lowlands and spread withexpanding populations into the Highlands as climates ameliorated at the beginning of theHolocene (Golson 1991b: 88-9; Hope & Golson 1995: 827-8). It is argued here thatagriculture emerged from broad-spectrum plant exploitation practices in the Highlands whichhad enabled permanent occupation of the interior during the Late Pleistocene. Climatic(Late Pleistocene) and environmental (early to mid-Holocene) forcing of existing plantexploitation practices led to the development of more interventionist and extensive strategiesin the Highlands. Although the specific mechanisms remain uncertain, an increasing focus

Res

ear c

h

853


on major sources of starch, including Colocasia taro and Eumusa bananas, were central to theemergence of agriculture in the Highlands.

As with an earlier debate concerning arboriculture (compare Kirch 1989 to Swadling et al.1991 and Yen 1996), agriculture in New Guinea was traditionally viewed as being of SoutheastAsian origin (e.g. Sauer 1952). The archaeological, archaeobotanical and palaeoecologicalfindings at Kuk corroborate phytogeographic and genetic interpretations of independentplant domestication in Melanesia and demonstrate that agriculture emerged in New Guineaindependently of any South-east Asian influence by at least 6950-6440 cal BP. Indeed, theearly Holocene Eumusa section bananas at Kuk solidify previous interpretations for the earlydiffusion of domesticated plants from New Guinea and their subsequent influence on thedevelopment of agriculture in Southeast Asia (De Langhe & de Maret 1999: 378-82) andAfrica (Mbida et al. 2000, 2001). Such findings open up new possibilities for interactionbetween mainland Southeast Asia and New Guinea during the early and mid-Holocene (i.e.prior to Austronesian dispersal into Indo-Malaysia and Melanesia) and require us to rethinkthe origins and spread of agriculture in the Pacific, Southeast Asia and beyond.

AcknowledgementsThe authors are indebted to Emeritus Professor Jack Golson for his assistance, encouragement and enthusiasmduring every stage of the project. Denham acknowledges scholarships and a fieldwork grant from the School ofArchaeology and Anthropology, Australian National University (ANU) and grants from the Australia and PacificScience Foundation (to J. Golson in 1998 and 1999) and the Pacific Islands Development Program (in 1999).Radiocarbon dates were provided by Centre for Archaeological Research, Australian Nuclear Science andTechnology Organisation and Research School of Earth Sciences, ANU (courtesy of John Chappell and AbazAlimanovic). Haberle undertook his research during an Australian Research Council QEII Fellowship. Lentferacknowledges a scholarship from Southern Cross University and grants from the Pacific Biological Foundation(2002) and Australian Museum. The authors are indebted to the contributions of Richard Fullagar, Judith Field,Michael Therin (residue analysis, all University of Sydney), Barbara Winsborough (diatom analysis, WinsboroughConsulting, Texas) and Nick Porch (insect analysis, Monash University) – although their work does not appear inthis paper, they are integral members of the research team. The authors thank Kay Dancey, Rob Patat, Art andCherie Rohn and Resource Management in Asia-Pacific Program, ANU for assistance with figures. Golson andJocelyn Powell kindly granted permission to publish the A10f/g pollen diagram from Kuk and Powell grantedpermission to reproduce and modify pollen diagrams from sites in the Upper Wahgi Valley. Atholl Anderson,Soren Blau, Jack Golson and two anonymous reviewers are thanked for comments on earlier manuscripts. Thisresearch was made possible with the permission and assistance of the Papua New Guinea National Museum andArt Gallery, National Research Institute, Western Highlands Provincial Government and the Kawelka at Kuk.

BibliographyARGENT, G.C.G. 1976. The wild bananas of Papua

New Guinea, Notes from The Royal Botanic Garden,Edinburgh 35: 77-114.

BARD, E., B. HAMELIN, R.G. FAIRBANKS & A. ZINDLER.1990. Calibration of the 14C timescale over thepast 30 000 years using mass spectrometric U-Thages from Barbados, Nature 345: 405-410.

BAYLISS-SMITH, T. 1996. People-plant interactions in theNew Guinea highlands: agricultural hearthland orhorticultural backwater? in D.R. Harris (ed.): 499-523.

BAYLISS-SMITH, T. & J. GOLSON. 1992a. Wetlandagriculture in New Guinea Highlands prehistory,in B. Coles (ed.), The wetland revolution inprehistory: 15-27. Exeter: The Prehistoric Societyand Wetland Archaeological Research Project.

–1992b. A Colocasian revolution in the New GuineaHighlands?: insights from Phase 4 at Kuk,Archaeology in Oceania 27: 1-21.

–1999. The Meaning of Ditches: Deconstructing theSocial Landscapes of Drainage in New Guinea,Kuk, Phase 4, in C. Gosden and J. Hather (eds.):199-231.

854


BELLWOOD, P. 1996. The origins and spread ofagriculture in the Indo-Pacific region: gradualismand diffusion or revolution and colonisation? inD.R. Harris (ed.): 465-98.

–1997. Prehistory of the Indo-Malaysian Archipelago.Honolulu: University of Hawaii Press.

BOURKE, R.M. 1996. Edible indigenous nuts in PapuaNew Guinea, in M.L. Stevens, R.M. Bourke andB.R. Evans (eds.): 45-55.

–2001. Intensification of agricultural systems in PapuaNew Guinea, Asia Pacific Viewpoint 42(2-3): 219-236.

–n.d. Altitudinal limits of 220 economic crop species inPapua New Guinea. Unpublished manuscript onfile at Research School of Pacific and Asian Studies(RSPAS), Australian National University (ANU).

BROOKFIELD, H.C. & D. HART. 1971. Melanesia.London: Methuen.

BULMER, S. 1975. Settlement and economy inprehistoric Papua New Guinea: a review of thearchaeological evidence, Journal de la Société desOcéanistes 31(46): 7-75.

COLLINS, J.F. & F. LAVNEY. 1983. Micromorphologicalchanges with advancing pedogenesis in some Irishalluvial soils, in P. Bullock and C.P. Murphy (eds.),Soil micromorphology: 297-307. Berkhamstead: ABAcademic.

DE LANGHE, E. & P. DE MARET. 1999. Tracking thebanana: its significance in early agriculture, in C.Gosden and J. Hather (eds.): 377-96.

DENHAM, T.P. 2003a. The Kuk morass: multi-disciplinary investigations of early to midHolocene plant exploitation at Kuk Swamp, WahgiValley, Papua New Guinea. Unpublished PhDthesis, Australian National University.

–2003b. Archaeological evidence for mid-Holoceneagriculture in the interior of Papua New Guinea: acritical review, in T.P. Denham and C. Ballard(eds.): 159-76.

–2004. Early agriculture in the Highlands of NewGuinea: an assessment of Phase 1 (c. 9000 bp) atKuk Swamp, in V.J. Attenbrow and R. Fullagar(eds.), A Pacific Odyssey: archaeology andanthropology in the Western Pacific. Papers in honourof Jim Specht: 47-57. Records of the AustralianMuseum Supplement 29, Sydney: AustralianMuseum.

–In press. Agricultural origins and the emergence ofrectilinear ditch networks in the Highlands of NewGuinea, in A. Pawley, R. Attenbrough, J. Golsonand R. Hide (eds.), Papuan pasts: Studies in thecultural, linguistic and biological history of thePapuan-speaking peoples. Adelaide: Crawford House(to be published late 2004).

DENHAM, T.P. & C. BALLARD (eds.). 2003. Perspectiveson prehistoric agriculture in New Guinea.Archaeology in Oceania Special Issue.

DENHAM, T.P. & H. BARTON. In press. The emergenceof agriculture in New Guinea: continuity from pre-existing foraging practices, In D.J. Kennett and B.Winterhalder (eds.), Behavioral ecology and thetransition to agriculture. Washington: SmithsonianInstitution Press (to be published late 2004).

DENHAM, T.P., S.G. HABERLE, C. LENTFER, R. FULLAGAR,J. FIELD, M. THERIN, N. PORCH & B.WINSBOROUGH. 2003. Origins of agriculture atKuk Swamp in the Highlands of New Guinea.Science 301: 189-193.

DWYER, P.D. & M. MINNEGAL. 1991. Hunting inlowland, tropical rain forest: Towards a model ofnon-agricultural subsistence, Human Ecology 19:187-212.

FRENCH, B.R. 1986. Food plants of Papua New Guinea:a compendium. Privately published book.

GILLIESON, D., P. GORECKI & G. HOPE. 1985.Prehistoric agricultural systems in a lowlandswamp, Papua New Guinea, Archaeology inOceania 20: 32-7.

GOLSON, J. 1976. Archaeology and agricultural historyin the New Guinea Highlands, in G. de G.Sieveking, I.A. Longworth and K.E. Wilson (eds.),Problems in economic and social archaeology: 201-20. London: Duckworth.

–1977a. No room at the top: agriculturalintensification in the New Guinea Highlands, in J.Allen, J. Golson and R. Jones (eds.), Sunda andSahul: prehistoric studies in southeast Asia, Melanesiaand Australia: 601-38. London: Academic Press.

–1977b. The making of the New Guinea Highlands, inJ.H. Winslow (ed.), The Melanesian environment:45-56. Canberra: Australian National UniversityPress.

–1977c. Simple tools and complex technology:Agriculture and agricultural implements in theNew Guinea Highlands, in R.V.S. Wright (ed.),Stone tools as cultural markers: change, evolution andcomplexity: 154-61. New Jersey: Humanities PressInc.

–1981. New Guinea agricultural history: a case study,in D. Denoon and C. Snowden (eds.), A time toplant and a time to uproot: 55-64. Port Moresby:Institute of Papua New Guinea Studies.

–1982. The Ipomoean revolution revisited: society andsweet potato in the upper Wahgi Valley, in A.Strathern (ed.), Inequality in New Guinea Highlandsocieties: 109-36. Cambridge: CambridgeUniversity Press.

Res

ear c

h

855


–1985.Agricultural origins in Southeast Asia: a viewfrom the east, in N. Misra and P. Bellwood (eds.),Recent advances in Indo-Pacific prehistory: 307-14.Leiden: E. J. Bill.

–1989. The origins and development of New Guineaagriculture, in D.R. Harris and G.C. Hillman(eds.): 678-87.

–1990. Kuk and the development of agriculture in NewGuinea: retrospection and introspection, in D.E.Yen and J.M.J. Mummery (eds.), Pacific ProductionSystems: Approaches to Economic Prehistory: 139-47.Canberra: Australian National University.

–1991a. Bulmer Phase II: early agriculture in the NewGuinea Highlands, in A. Pawley (ed.), Man and ahalf: essays in Pacific anthropology and ethnobiologyin honour of Ralph Bulmer: 484-91. Auckland: ThePolynesian Society.

–1991b. The New Guinea Highlands on the eve ofagriculture, Bulletin of the Indo-Pacific PrehistoryAssociation 11: 82-91.

–2000. A stone bowl fragment from the Early MiddleHolocene of the Upper Wahgi Valley, WesternHighlands Province, Papua New Guinea, in A.Anderson and T. Murray (eds.), AustralianArchaeologist: collected papers in honour of Jim Allen:231-48. Canberra: Coombs Academic Publishing,ANU.

–2002. Gourds in New Guinea, Asia and the Pacific, inS. Bedford, C. Sand and D. Burley (eds.), Fiftyyears in the field. Essays in honour and celebration ofRichard Shutler Jr.’s archaeological career: 69-78.New Zealand Archaeological Journal Monograph25, Auckland: Auckland Museum.

GOLSON, J. & P.J. HUGHES. 1980. The appearance ofplant and animal domestication in New Guinea,Journal de la Société des Océanistes 36: 294-303.

GOSDEN, C. & J. HATHER (eds.). 1999. The prehistory offood: appetites for change. London: Routledge.

GREEN, R. 1991. The Lapita cultural complex: currentevidence and proposed models, Bulletin of the Indo-Pacific Prehistory Association 11: 295-305.

GROUBE, L. 1989. The taming of the rainforests: amodel for Late Pleistocene forest exploitation inNew Guinea, in D.R. Harris and G.C. Hillman(eds.): 292-304.

GROUBE, L., J. CHAPPELL, J. MUKE & D. PRICE. 1986. A40,000 year-old human occupation site at HuonPeninsula, Papua New Guinea, Nature 324(4):453-5.

HABERLE, S.G. 1993. Late Quaternary environmentalhistory of the Tari Basin, Papua New Guinea.Unpublished PhD thesis, Australian NationalUniversity.

–1994. Anthropogenic indicators in pollen diagrams:problems and prospects for late Quaternarypalynology in New Guinea, in J. G. Hather, ed.,Tropical archaeobotany: applications and newdevelopments: 172-201. London: Routledge.

–1995. Identification of cultivated Pandanus andColocasia in pollen records and the implications forthe study of early agriculture in New Guinea,Vegetation History and Archaeobotany 4: 195-210.

–1998. Late Quaternary change in the Tari Basin,Papua New Guinea, Palaeogeography,Palaeoclimatology, Palaeoecology 137: 1-24.

HABERLE, S.G., G.S. HOPE & Y. DE FRETES. 1991.Environmental change in the Baliem Valley,Montane Irian Jaya, Republic of Indonesia, Journalof Biogeography 18: 25-40.

HABERLE, S.G., G.S. HOPE, & S. VAN DER KAARS. 2001.Biomass burning in Indonesia and Papua NewGuinea: natural and human induced fire events inthe fossil record, Palaeogeography, Palaeoclimatology,Palaeoecology 171: 259-268.

HARLAN, J. 1995. The living field: our agriculturalheritage. Oxford: Oxford University Press.

HARRIS, D.R. 1995. Early agriculture in New Guineaand the Torres Strait divide, Antiquity 69 (SpecialNumber 265): 848-54.

–1996. The origins and spread of agriculture andpastoralism in Eurasia: an

–(ed.). 1996. The origins and spread of agriculture andpastoralism in Eurasia. London: UCL Press.

HARRIS, D.R. & G.C. HILLMAN (eds.). 1989. Foragingand Farming: the evolution of plant exploitation.London: Unwin Hyman.

HARRIS, E.C. 1977. Hed mound: a New Guinea housesite. Unpublished manuscript on file at RSPAS,ANU.

HARRIS, E.C. & P.J. HUGHES. 1978. An earlyagricultural system at Mugumamp Ridge, WesternHighlands Province, Papua New Guinea, Mankind11 (4): 437-45.

HAYWOOD, M. 1985. Soil development in Flandrianlandscapes: River Severn case-study, in J.Boardman (ed.), Soils and Quaternary landscapeevolution: 281-99. Chichester: John Wiley.

HOPE, G.S. 1983. The vegetational changes of the last20,000 years at Telefomin, Papua New Guinea,Singapore Journal of Tropical Geography 4: 25-33.

HOPE, G.S. & J. GOLSON. 1995. Late Quaternarychange in the mountains of New Guinea, Antiquity69 (Special Number 265): 818-30.

856


HOPE, J.H. & G.S. HOPE. 1976. Palaeoenvironmentsfor man in New Guinea, in R.L. Kirk and A.G.Thorne (eds.), The Biological Origins of theAustralians: 29-54. Canberra: Australian Instituteof Aboriginal Studies Publication.

HUGHES, P.J., M.E. SULLIVAN & D. YOK. 1991. Humaninduced erosion in a Highlands catchment inPapua New Guinea: the prehistoric andcontemporary records, Zeitschrift fürGeomorphologie Suppl. 83: 227-39.

KIRCH, P.V. 1989. Second millenium B.C. aboriculturein Melanesia: archaeological evidence from theMussau Islands, Economic Botany 43(2): 225-40.

LEBOT, V. 1999. Biomolecular evidence for plantdomestication in Sahul, Genetic Resources and CropEvolution 46: 619-28.

LOY, T., M. SPRIGGS & S. WICKLER. 1992. Directevidence for human use of plants 28,000 years ago:starch residues on stone artefacts from northernSolomon Islands, Antiquity 66: 898-912.

MATTHEWS, P.J. 1991. A possible wild type taro:Colocasia esculenta var. aquatalis, Indo-PacificPrehistory Association Bulletin 11: 69-81.

–1995. Aroids and Austronesians, Tropics 4 (2): 105-26.

–2003a. Identification of Benincasa hispida (wax gourd)from the Kana archaeological site, WesternHighlands Province, Papua New Guinea, in T.P.Denham and C. Ballard (eds.): 186-91.

–2003b. Taro planthoppers (Tarophagus spp.) inAustralia and the origins of taro (Colocasiaesculenta) in Oceania, in T.P. Denham and C.Ballard (eds.): 192-202.

MBIDA, C.M., H. DOUTRELEPONT, L. VRYDAGHS, R.L.SWENNEN, R.J. SWENNEN, H. BEECKMAN, E. DE

LANGHE & P. DE MARET. 2001. First archaeologicalevidence of banana cultivation in central Africaduring the third millennium before present,Vegetation History and Archaeobotany 10: 1-6.

MBIDA, C.M., W. VAN NEER, H. DOUTRELEPONT & L.VRYDAGHS. 2000. Evidence for banana cultivationand animal husbandry during the first millenniumBC in the forest of Southern Cameroon, Journal ofArchaeological Science 27: 151-62.

MCALPINE, J.R., G. KEIG & R. FALLS. 1983. Climate ofPapua New Guinea. Canberra: CSIRO and ANUPress.

MOUNTAIN, M.J. 1991. Highland New Guinea hunter-gatherers: the evidence of Nombe Rockshelter, Simbu,with emphasis on the Pleistocene. Unpublished PhDthesis, Australian National University.

MUKE, J. & H. MANDUI. 2003. Shadows of Kuk:evidence for prehistoric agriculture at Kana, WahgiValley, Papua New Guinea, in T.P. Denham and C.Ballard (eds.): 177-85.

PAIN, C.F., C.J. PIGRAM, R.J. BLONG & G.O. ARNOLD.1987. Cainozoic geology and geomorphology ofthe Wahgi Valley, central highlands of Papua NewGuinea, BMR Journal of Australian Geology andGeophysics 10: 267-75.

POWELL, J.M. 1970a. The history of agriculture in theNew Guinea Highlands, Search 1(5): 199-200.

–1970b. The impact of man on the vegetation of theMount Hagen region, New Guinea. UnpublishedPhD thesis, Australian National University.

–1976. Ethnobotany, in K. Paijmans (ed.), New Guineavegetation: 106-83. Canberra: CSIRO and ANUPress.

–1980. Studies of New Guinea vegetation history,Proceedings of the IVth International PalynologyConference, Lucknow (1976-7) 3: 11-20.

–1981. The origins of agriculture in New Guinea,Proceedings of the IVth International PalynologyConference, Lucknow (1976-7) 3: 295-310.

–1982a. Plant resources and palaeobotanical evidencefor plant use in the Papua New Guinea Highlands,Archaeology in Oceania 17: 28-37.

–1982b. The history of plant use and man’s impact onthe vegetation, in J.L. Gressitt (ed.), Biogeographyand ecology of New Guinea. Volume 1: 207-27. TheHague: Junk.

–1984. Ecological and palaeoecological studies at KukI: below the grey clay. Unpublished manuscript onfile at RSPAS, ANU.

POWELL, J.M. & S. HARRISON. 1982. Haiyapugua:aspects of Huli subsistence and swamp cultivation..Port Moresby: Department of Geography, UPNG(Occasional Paper No. 1, New Series).

POWELL, J.M., A. KULUNGA, R. MOGE, C. PONO, F.ZIMIKE & J. GOLSON. 1975. Agricultural traditionsin the Mount Hagen area. Port Moresby:Department of Geography, UPNG (OccasionalPaper No. 12).

RICHERSON, P.J., R. BOYD & R.L. BETTINGER. 2001.Was agriculture impossible during the Pleistocenebut mandatory during the Holocene? A climatechange hypothesis, American Antiquity 66(3): 387-411.

ROSCOE, P. 2002. The hunters and gatherers of NewGuinea, Current Anthropology 43(1): 153-62.

SAUER, C.O. 1952. Agricultural origins and theirdispersals. New York: American GeographicalSociety.

SIMMONDS, N.W. 1962. The evolution of the bananas.London: Longmans.

SMITH, B.D. 1998. The emergence of agriculture. NewYork: Scientific American Library.

Res

ear c

h

857


SPRIGGS, M. 1996. Early agriculture and what wentbefore in Island Melanesia: continuity or intrusion?in D.R. Harris (ed.): 524-37.

–2001. Who cares what time it is? The importance ofchronology in Pacific archaeology, in A. Anderson,I. Lilley and S. O’Connor (eds.), Histories of oldages: essays in honour of Rhys Jones: 237-49.Canberra: Pandanus Books, Australian NationalUniversity.

STEVENS, M.L., R.M. BOURKE & B.R. EVANS (eds.).South Pacific indigenous nuts. Canberra: ACIAR.

STUIVER, M & P.J. REIMER. 1993. Extended 14Cdatabase and revised CALIB radiocarboncalibration program, Radiocarbon 35:215-230.

SWADLING, P., N. ARAHO & B. IVUYO. 1991. Settlementsassociated with the inland Sepik-Ramu Sea,Bulletin of the Indo-Pacific Prehistory Association 11:92-110.

TERRELL, J.E. 2002. Tropical agroforestry, coastallagoons and Holocene prehistory in Greater NearOceania, in Y. Shuji and P.J. Matthews (eds.),Proceedings of the International Area StudiesConference VII: vegeculture in Eastern Asia andOceania: 195-216. Osaka: National Museum ofEthnology.

TERRELL, J.E., J.P. HART, S. BARUT, N. CELLINESE, A.CURET, T.P. DENHAM, H. HAINES, C.M. KUSIMBA,K. LATINIS, R. OKA, J. PALKA, M.E.D. POHL, K.O.POPE, J.E. STALLER & P.R. WILLIAMS. 2003.Domesticated landscapes: the subsistence ecologyof plant and animal domestication, Journal ofArchaeological Method and Theory 10(4): 323-68.

TERRELL, J.E. & R. WELSCH. 1997. Lapita and thetemporal geography of prehistory, Antiquity 71:548-72.

WILSON, S.M. 1985. Phytolith evidence from Kuk, anearly agricultural site in New Guinea, Archaeologyin Oceania 20: 90-7.

YEN, D.E. 1991. Polynesian cultigens and cultivars: thequestions of origin, in P.A. Cox and S.A. Banack(eds.), Islands, plants and Polynesians: anintroduction to Pacific ethnobotany: 67-95.Portland: Dioscorides Press.

–1995. The development of Sahul agriculture withAustralia as bystander, Antiquity 69, SpecialNumber 265: 831-47.

–1996. Melanesian aboriculture: historical perspectiveswith emphasis on the genus Canarium, in M.L.Stevens, R.M. Bourke and B.R. Evans (eds.): 36-44.

–1998. Subsistence to commerce in Pacific agriculture:some four thousand years of plant exchange, inH.D.V. Pendergast, N.L. Etkin, D.R. Harris andP.J. Houghton (eds.), Plants for food and medicine:161-83. Kew: Royal Botanic Gardens.

new evidence and revised interpretations of early...

Documents