Early and mid Holocene tool-use and processing of taro (Colocasia esculenta), yam (Dioscorea sp.) and other plants at Kuk Swamp in the highlands of Papua New Guinea

Journal of Archaeological Science 33 (2006) 595e614 http://www.elsevier.com/locate/jas

Early and mid Holocene tool-use and processing of taro (Colocasia esculenta), yam (Dioscorea sp.) and other plants at Kuk Swamp in the highlands of Papua New Guinea Richard Fullagar a,*, Judith Field b, Tim Denham c, Carol Lentfer d a Department of Archaeology, University of Sydney, Sydney, NSW 2006, Australia Australian Key Centre for Microscopy and Microanalysis & The School of Philosophical and Historical Inquiry, University of Sydney, Sydney, NSW 2006, Australia c School of Geography and Environmental Science, Monash University, Clayton Campus, Wellington Road, Clayton, VIC 3800, Australia d School of Social Sciences, University of Queensland, St. Lucia, QLD 4072, Australia b

Received 15 November 2004; received in revised form 7 April 2005; accepted 22 July 2005

Abstract Recent multidisciplinary investigations document an independent emergence of agriculture at Kuk Swamp in the highlands of Papua New Guinea. In this paper we report preliminary usewear analysis and details of prehistoric use of stone tools for processing starchy food and other plants at Kuk Swamp. Morphological diagnostics for starch granules are reported for two potentially significant economic species, taro (Colocasia esculenta) and yam (Dioscorea sp.), following comparisons between prehistoric and botanical reference specimens. Usewear and residue analyses of starch granules indicate that both these species were processed on the wetland margin during the early and mid Holocene. We argue that processing of taro and yam commences by at least 10,200 calibrated years before present (cal BP), although the taro and yam starch granules do not permit us to distinguish between wild or cultivated forms. From at least 6950 to 6440 cal BP the processing of taro, yam and other plants indicates that they are likely to have been integrated into cultivation practices on the wetland edge. Ó 2005 Elsevier Ltd. All rights reserved. Keywords: Starch grains; Taro (Colocasia esculenta); Yam (Dioscorea sp.); Banana (Musa spp.); Phytoliths; Plant processing; Usewear; Residues

1. Introduction Over the last decade, starch grain analysis in archaeological contexts has become a highly significant new direction for reconstructing plant utilisation, processing technology and food production systems. In many parts of the world, research has targeted the identification of food and other economic plants, specifically, starch in roots, tubers [5,28,45,57,58,67e69,79, 80,81] and seeds [29,41,66,82], and also in palm pith and arboreal fruit [2]. In the Pacific, a suite of starch-rich plants has been fundamental to subsistence [85] in both agricultural and hunter-gatherer societies. Kuk Swamp, Papua New Guinea,

* Corresponding author. Tel.: þ61 2 9351 6794; fax: þ61 2 9351 5712. E-mail address: [email protected] (R. Fullagar). 0305-4403/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2005.07.020

has been central to debate about agricultural origins in the Pacific [23], and in this study we focus on evidence for the exploitation of taro (Colocasia spp.) and yams (Dioscorea spp.) at this site (Fig. 1). Debate about when agriculture commenced at Kuk Swamp has hinged on interpretations of palaeochannels and palaeosurfaces, and the identification of potentially cultivated plants [22,21]. Although several plants had previously been suggested as likely candidates for cultivation, recent studies have now identified particular taxa from archaeobotanical remains [23]. The aim of this paper is to provide the detailed evidence from the lithic usewear and residue study that first established the identification of the yam and taro starch on a selection of artefacts from Phases 1 and 2 of human use of the wetland margin at Kuk Swamp [23]. Although the lithic study is not yet complete, the preliminary usewear results are

596

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

Fig. 1. Site location map. (After Denham et al. [23].)

significant for interpreting the early history of plant exploitation at Kuk Swamp, and broaden interpretations of the earliest cultivation practices. First, we outline the archaeological context and selection of artefacts. Second, we describe methods of analysis, and the characteristics of taro and yam starch granules from modern botanical specimens. Third, we summarise preliminary interpretations of usewear, and the starch and phytolith identifications. Discussion focuses on methodology and the significance of yam and taro starch granules. 2. Kuk Swamp and stone artefacts Kuk Swamp is located at 1560 m above mean sea level in the upper Wahgi Valley, which is a large intermontane valley in the New Guinea highlands (Fig. 1). The Kuk vicinity has a mean annual temperature of 19
C and mean annual rainfall of c. 2700 mm [21]. The archaeological site at Kuk is located on a wetland margin comprising a low-gradient alluvial fan deposited after 21,500 calibrated years before present (cal BP). Investigations into the antiquity of agriculture at the site began in 1969, with major field seasons in 1972e1977 [33,37,44] and 1998e1999 [17e19]. Recently published evidence from Kuk confirms New Guinea was a primary centre of agricultural development by at least 6950e6440 cal BP [21], well before the introduction of Southeast Asian domesticates associated with Austronesian expansion after 3500 cal BP [10]. The archaeological finds at Kuk represent several phases of prehistoric utilisation and drainage of the wetland margin for plant exploitation and cultivation. The earliest use of the wetland margin (Phase 1) dates to c. 10,220e9910 cal BP, although interpretations vary as to whether this represents agriculture [33,37], or an undetermined form of plant exploitation ([17,19,23]; see review in [21]). Researchers agree that the earliest unequivocal evidence of agriculture dates to 6950e6440 cal BP and comprises mounded cultivation on the wetland margin ([17,33]; see [18] for a review of evidence from other sites). Subsequent

phases consist of ditch networks constructed for drainage and cultivation, although network designs vary [7e9,17,33]. Until recently, the plants cultivated, exploited and present during each phase were not well-documented ([17,23]; cf. [71,84]). In particular, there was no information on root-crops, which are the staple crops of agriculture in the Highlands today. Consequently, evidence for the processing of root crops during the early and mid Holocene at Kuk, which has been presented in summary previously [23], accords generally with interpretations of plant exploitation and early agriculture based on archaeological evidence [17e19,21]. Additionally, the exploitation of root crops from the early Holocene accords with genetic interpretations of their independent domestication in the New Guinea region [52]. The stone artefacts examined in this study were collected during field programmes directed by Allen [3], Golson [33,34] and Denham [17]. Cursory studies since 1990 by the late T.H. Loy (recently University of Queensland) and others have reported residues, retouch and other traces of use on the selection of artefacts that are analysed here. Consequently, the artefacts reported here do not constitute a random sample and all were expected to show signs of use. Subsequent functional analyses were undertaken in two stages at the Archaeology Laboratory, University of Sydney between 1999 and 2004. The aim of the Stage 1 study was simply to determine the nature of residue preservation and the range of tool functions (from all levels) for a selection of artefacts from Kuk (n ¼ 55), i.e., how tools were used (scraping, cutting, etc.), the broad class of contact materials (wood, bone, etc.), and the presence of starch. The aim of the Stage 2 study was to look specifically for yam and taro starch, and other identifiable residues on artefacts (n ¼ 12) with precise early-to-mid Holocene provenance. Residue samples were extracted from the twelve tools associated with early and mid Holocene phases of wetland manipulation, as well as from the intervening ‘grey clay’ stratigraphic unit. All potential traces of use were re-examined, and starch granules and phytoliths were compared with modern reference collections for identification.

3. Methods and procedures The principles of usewear and residue analysis are based largely on traces of use that are documented experimentally [26]. Processing of starchy plants is implied by the presence, frequency and distribution of starch granules1 on stone artefacts usually (but not necessarily) in association with other residues (such as cellulose) and usewear, which may indicate hardness (or type of material) and direction of tool motion [49,50]. Biogeography and ethnography provide further information on the function of stone tools, and the kinds of tasks and available materials likely to have been utilised [39,74,83]. The use of stone artefacts to process a variety of 1

Although ‘starch granule’ and ‘starch grain’ are often used interchangeably, we use ‘granule’ when specifically referring to individual (not composite) grains.

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

starchy plants in the Papua New Guinea-Australia region was common until very recently. Pounding and grinding stones were used to make flour and mash tubers, nuts, seeds and fruits; and flaked stone tools were used as knives to slice and peel tubers, and as spoons to scoop baked or roasted roots and tubers [64,65,76]. Ethnographic studies report the use of both flaked stone and pounding implements (mortar and pestle) to process taro in Papua New Guinea [76]. On the island of New Britain (Fig. 1), after European contact, bottle glass fragments replaced obsidian flakes to scrape raw tubers before and during cooking on an open fire (Fig. 2). The flaked tools were used to peel and scrape outer surfaces of tubers several times during cooking in order to remove physical irritants (raphides), unpalatable bitter compounds (acrid principle) and charred surfaces [11,75]. Stone and wooden pounding tools were also used to mash tubers. Ethnographic accounts of food and plant processing provide direct analogues for interpreting the function of archaeological stone artefacts [70,71]. Studies also show that ethnographic tools now in museum collections do bear starch granules (e.g. [24,25,31]).

597

tool edges used to process a range of wood and other plants in Australia and Papua New Guinea [27]. Functional interpretations were made on the basis of comparative reference collections of traces that include about 400 tool-use experiments, in which stone was used to process a variety of materials from Australia and Papua New Guinea [25]. Due to ambiguous patterns of poorly developed wear traces on some artefacts from Kuk, interpretations of tool function were initially provided at three levels of confidence: possible, probable, and definite. Only those with developed usewear typical of specific worked materials were assigned to these particular classes: wood or woody tissue, skin, reed, and bone. Non-siliceous, soft plant processing and ochre were each indicated by probable and definite usewear in association with diagnostic residue traces: starch (identified by optical properties, see below); cellulose (identified by bright birefringence and cell structures); and smears of red pigment or ‘ochre’ (identified by opaque, bright red clay sized particles), which, however, may be natural concretions that form in wetland deposits, rather than artefacts of human activity. 3.2. Residues

3.1. Usewear All artefacts were examined under Olympus or Zeiss stereoscopic microscopes (magnifications 6 to 100) with oblique incident light. Most were also examined under OlympusÔ and ZeissÔ metallographic microscopes (magnifications ranging from 50 to 1000) with brightfield/darkfield and polarising filters specifically for the observation of use-polish, starch granules and other residues. The main forms of usewear, shown experimentally to indicate mode of use (for example, sawing, scraping, etc.) and contact material (for example, siliceous plant, wood, skin, etc.), include edge scarring, edge rounding, striations and smoothing or polish [25,48]. Experiments have shown that even a low amount of plant opal or silica can contribute significantly to distinctive polish development on stone

Fig. 2. Thomas is scraping partially cooked tubers with an obsidian flake next to the fire, West New Britain Province, PNG. (Photo: R. Fullagar and R. Torrence.)

Residue samples were extracted in ultra-pure water with a sonic cleaner or pipette. Artefacts examined in this study had been handled and cleaned in prior studies of the assemblage following initial excavation. However, original residues were firmly attached and were abundantly preserved; we are confident that they are not loosely adhering modern contaminants [6]. Microscope slides were prepared for each sample. Starch and phytoliths were mounted in KaroÔ and benzyl benzoate, respectively. Nomarski (differential interference contrast, DIC) microscopy was used to document starch granules at high magnification. Starch granule images were also recorded under polarised light and DIC using digital image capture (Zeiss Axiocam HrCÔ). Phytoliths were photographed under brightfield with colour transparency film (Olympus BH-2Ô). Identification of phytoliths with taxonomic certainty can be complex due to large variation of morphotypes within a single species and multiplicity and redundancy (i.e. multiple morphotypes can occur within a single species, and the same morphotypes can occur in several different species). Modern reference collections of phytoliths cover over 1000 species of tropical plants for Papua New Guinea and Southeast Asia [54e56]. Our modern reference collection of starch grains is not exhaustive but covers more than 100 species of starch-rich edible plants from the AustraliaeNew Guinea region, including grasses, palms, ferns, lilies, and fruit trees. Following a previous study of prehistoric starch granules [58], our reference collection also covers all the common aroids and yams used for food in the South Pacific. We include discussion of five species of Colocasia (Fig. 3), eight species of Dioscorea (Figs. 4 and 5), Musa banana (Fig. 6), and non-Colocasia aroids (Fig. 7). Musa bananas are also included because they are known to be present prehistorically at Kuk, and their starch granules appeared to be superficially similar to those of some Dioscorea yams. The reference specimens also comprise samples from various plant

598

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

parts including corm, flower, root, leaf, seed, stem and tuber [77]. Modern reference collections from Australia, Papua New Guinea and the southwestern Pacific islands are held at the University of Sydney, Australian Museum and Southern Cross University [77]. Taxonomic identification of archaeological starch granules is based on intrinsic variables such as size, shape, surface morphology, hilum features, and chemical properties; as well as on extrinsic variables such as clustering and co-occurrence with other tissue and particles (e.g. phytoliths and raphides) which can also have distinctive morphologies [40,57,58, 68,72,79,80,81]. Other studies have explored the discriminatory power of nominal and metric variables using large reference samples and different lighting conditions. For example, a recent study had significant success in discriminating some taxa, but cautioned that multivariate study ‘‘may have placed too much emphasis on characters visible under polarised lighting’’ [78, p. 10]. Previous studies have demonstrated that Colocasia and Dioscorea starch granules can be identified to genus and sometimes species levels [57,58,78]. In this study, we examined three dimensional shape and surface features of starch granules under DIC microscopy. The interpretation of small starch granules (<5 mm) can be complicated by apparent similarities with undiagnostic transient starch granules derived from leaves of some plants [42,57], although some forest tree starch granules, probably in transient form, appear to be distinct from Colocasia esculenta [78]. Although further study is required to fully characterise transient (or ‘transitory’) starches and their preservation in the archaeological record, Haslam [42] shows that some transient starch granules are flattened with ragged edges. In addition, we argue that larger (>5 mm) Colocasia esculenta tuber starch granules have distinctive surface features absent in transient starch. Furthermore, leaf starch from several species (Ipomoea batatas, Dioscorea alata, Colocasia esculenta) had much smaller granules scarcely visible at 500 magnification, whereas the tuber starches were routinely visible at 200. Moreover, leaf starch does not form dense deposits in vivo (as it does in storage organs) and is unlikely to be preserved in clusters (Peter Matthews, personal communication, 2005). 4. Taro and yams in the region, and their characteristic starch granules Our modern reference collection covers known taros, and yams in the region, and these are discussed in detail. Starch granules from both these groups are generally distinct from each other and from others in our current reference collections. Key descriptions for starch granules of each species are given in captions (Figs. 3e8). 4.1. Colocasia esculenta Initial cytological and genetic studies have suggested origins for taro as a crop in Southeast Asia [86], northeastern India [51] or Indonesia [53]. However, a wild type taro (Colocasia esculenta var. aquatilis) has been identified that

is indigenous to New Guinea and northern Australia [47,60,63]. Furthermore, endemic species-specific pollinators (Drosophilella sp.) [61, p. 110e1] and insect pests (Tarophagus spp.) [62] occur on wild type taro in New Guinea and are strongly suggestive of endemic C. esculenta in this region. Although Matthews [60] was uncertain if domestication occurred in New Guinea, independent domestication in Sunda and Sahul is probable based on a study of karyotypes [12], a review of isozyme and DNA markers [52, p. 624] and the presence of wild type taro in this region (after Matthews [61]). Haberle [38] identified C. esculenta pollen at c. 10,200 cal BP at a lowland site (Lake Wanum), which confirmed that the species was present in New Guinea at the time of the earliest agricultural practices [35,36]. It is rarely identifiable from pollen, and the primary evidence for actual utilisation of C. esculenta derives from microscopic residues on excavated stone tools reported here. Loy [57,58] reported Alocasia and Colocasia (from c. 28e 20,000 years ago) and Colocasia (from the earliest Holocene) starch grains on stone tools from Kilu Cave, Solomon Islands. He compared prehistoric starch granules with a reference collection comprising nine genera and 20 species, including seven Dioscorea yam species and four species of Colocasia taro. Loy [57,58] discriminated starch granules of C. esculenta and C. gigantea on the basis of size and the absence of distinctively shaped, long ‘whisker-like’ raphides in C. esculenta. Loy et al. [58] note that C. gigantea, C. affinis and C. fallax are not found in Melanesia; the latter two species are from India and Southeast Asia and have granules considerably larger than those of C. esculenta. Starch granules of Colocasia and Alocasia species are shown in Figs. 3 and 7. A comparison of tool-use experiments and botanical comparative reference specimens of taro tubers suggest that a combination of variables is needed to identify uncooked C. esculenta starch granules on tool edges and in extracted residues [30,57,58]. In this study, we used a greatly expanded modern reference collection and several attributes to identify starch granules to the species level (after Loy et al. [58]): 1. Starch granule size varies from the limits of resolution in light microscopy (w1 mm) to about 8 mm maximum dimension. 2. Shape of starch granules is roughly spherical, often with a distinctly faceted surface morphology. Faceted surface morphology, clearly visible on larger granules, is proposed here to be distinctive and discriminatory, perhaps being typical of the packing within cells, and distinguishes starch granules of C. esculenta from transient or transitory starch granules. 3. Clustering of small granules in sheets of tissue is common. 4. Co-occurrence of raphides is uncommon in reference slides (thin sections and mashed tissue), but is sometimes present. C. esculenta raphides sometimes have a squarerectangular cross-section, are tapered to a fine point (often at one end only) and squared or angled at the other end to sometimes form the ‘double bevel’ observed by Loy [58, p. 900].

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

599

Fig. 3. Starch granules from tubers in the comparative reference collections under DIC (A,C,E) and polarising filters (B,D,F). (A,B) Starch granules of Colocasia esculenta tuber (AusMus 1/258, PNG). Granules of C. esculenta are generally very small, the largest granules usually not exceeding 6 mm in maximum dimension. These granules are circular to hexagonal in shape, depending on the degree of faceting, some appearing to have scalloped edges. Granules were observed in sheets, possibly membrane bound and are highly faceted. (C,D) Starch granules of Colocasia fallax tuber (AusMus 1/266, source unknown). Granules are very irregular in shape ranging from circular to ovate and many of the large granules have arms produced through irregular growth. The sizes range from <10 mm to around 35 mm in maximum dimension. The smaller granules tend to be more regular in shape with some faceted granules occurring. Raphides are common and have truncate ends. (E,F) Starch granules of Colocasia gigantea tuber (AusMus 1/265, Sri Lanka). Granules are very small, ranging from around 5 mm down to 1 or 2 mm in maximum dimension. The majority of the granules are in the lower end of the size range and are at the limits of resolution of the light microscope. Most granules are faceted as a function of cell packing and granule growth. (Photos: J. Field, ZeissÔ Axioskop 2 microscope with an AxioCam HRcÔ.)

4.2. Dioscorea spp. The pan-tropical, domesticated yams (Dioscorea alata and D. esculenta) were traditionally thought to be Southeast Asian introductions to Sahul [1, p. 417; 43, p. 71, cf. 13, p. 79]. Minor food species were considered to be indigenous including D. bulbifera, D. nummularia, D. pentaphylla and D. hispida [31,32,70, p. 119]. From the plurality of forms, Coursey

proposed New Guinea as a secondary centre of dispersal for D. alata and D. esculenta [14,15, p. 226; 16, p. 71]. Phylogenetic studies have not been able to differentiate D. alata cultivars on the basis of morphology, isozymes or physico-chemical characteristics [52, p. 624e5]. However, New Guinea has been proposed to be the place of origin given it is the centre of greatest genetic diversity [52, p. 625] and has many primitive types of D. alata [59, p. 10]. Separate domestication of Pacific

600

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

Fig. 4. Starch granules from tubers in the comparative reference collections, under DIC (A,C,E) and polarising filters (B,D,F). (A,B) Starch granules of Dioscorea alata tuber (AusMus 1/75, Bontok). The starch granules of D. alata are generally large, elongated, elliptical to oblong in shape, and have an obtuse to truncate end distal to the hilum. However there is also evidence of twinning of smaller sub-rounded granules, which develop facets. These granules are generally <12 mm in size. Some D. alata exhibit a contraction of the granule around an eccentric hilum, the latter being a feature of the Dioscorea genus. The size range of D. alata elongated granules is about 25e60 mm. (C,D) Starch granules of Dioscorea bulbifera tuber (AusMus 1/271; Vanuatu). Dioscorea bulbifera granules were observed to have two different morphologies. Small granules (7e15 mm maximum dimension) are circular to oblate, while larger granules (25e35 mm) were typically either oblong or triangular as seen in other Dioscorea sp. In profile they are elongate with a pinched ridge on the long axis. Lamellae were visible and in some cases granules with truncate ends showed a longitudinal rippling on the edge distal to the hilum. (E,F) Starch granules of Dioscorea rotundata tuber (AusMus 1/279; Vanuatu). Granules are typical of the Dioscorea genus in shape and hilum position (as for D. trifida). Size range is greater for D. rotundata than D. trifida, with some granules up to 60 mm in maximum length. (Photos: J. Field, ZeissÔ Axioskop 2, AxioCam HRcÔ.)

and Asian forms of D. bulbifera has been proposed based on genotypic differences [52, p. 625]. Loy et al. [58, p. 909] describe Dioscorea yams as having: ‘‘. large individual and/or composite starch grain size and the usually flattened, ovoid and often multi-faceted shape of the yam starch grains .’’. Although it may be difficult to discriminate species within the genus [58,46], Torrence et al. [78] found that starch granules of Dioscorea could be discriminated to the species level with variable degrees of confidence.

Mean and standard deviations for maximum dimensions of yam starch granules vary considerably but suggest that the small, highly faceted D. esculenta starch granules may form a distinct size population [57,58,73, pp. 214e5]. Our observations of starch granules from various Dioscorea species accord with previous studies (Table 1, Figs. 4 and 5). Apart from D. esculenta, Dioscorea starch granules are generally elongate, ovoid to triangular, and multi-faceted with an asymmetrical hilum location, smooth surfaces, and

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

faint concentric rings indicating the amylose and amylopectin layers under polarised light. However, there are differences in surface morphology and shape between species. For example, D. bulbifera has a distinctly flat base with a generally triangular shape, and an asymmetrical bend near the hilum. Larger granules of D. pentaphylla have a distinctly sculptured facet near the hilum, although this may be absent in smaller or immature granules of this species. 4.3. Musa spp. Initial study suggested that the large starch granules typical of Musa were similar in size and shape to starch granules of some Dioscorea species. Our modern reference samples include Musa ingens specimens (known from phytolith evidence at Kuk by at least 6950 to 6440 cal BP [23]) and recent cultivated banana varieties AAA and ABB (Fig. 6). 4.4. Non-Colocasia aroids We also sampled starch granules of non-Colocasia species of Araceae that are now commonly grown in PNG, including Yautia (Xanthosoma sagittifolium L. Schott), swamp taro (Cyrtosperma chamissonis Schott Merr.), giant taro (Alocasia macrorrhizos L. Schott) and elephant foot yam (Amorphophallus campanulatus Blume). The starch granules of these species are mostly greater than 10 mm (Fig. 7).

5. Results Both sets of results are presented here, including the preliminary study to assess the survival of usewear and residues and general classes of tool-use; and the more detailed analysis for identifying materials worked and specifically, the presence of distinctive Colocasia taro or Dioscorea yam starch granules. 5.1. Tool-use and residue preservation at Kuk In the Stage 1 study, 55 artefacts from Kuk were examined, and 50 had traces of usedmostly clear traces of woodworking, including six adze and adze fragments (Table 2). The use-polish on three tools overlapped with experimental polishes from woodworking as well as from processing skin, reeds and bone respectively. The usewear in combination with starch and phytolith residues suggest many artefacts were multi-functional tools, used on two or more materials. Seven artefacts had traces indicating soft plant processing, and nearly all of these had starch residues related to use (Fig. 8). One cobble pounding implement had traces of possible ‘ochre’, starch granules and crushed phytoliths, probably indicating at least two functions. Twenty-two artefacts were definitely used (that is, with distinct usewear) but the contact material was uncertain because developed polishes or taxonomically diagnostic residues were absent. The function of these tools could represent low intensity usage in a range of

601

tasks (including butchering), but probably excludes siliceous plant processing, which typically results in rapid formation of distinctive use-polish. The Stage 1 study demonstrated the potential for more detailed analysis of starch granules, phytoliths and associated usewear.

5.2. Tool function and identification of starch granules, phytoliths In the Stage 2 study, 12 well-provenanced artefacts from early contexts were examined in more detail (Figs. 9e12). Residues were removed and starch granules clearly observed on 11 of these artefacts, and phytoliths were found on eight artefacts (Table 3). Starch granules typical of Colocasia esculenta were identified on three artefacts. Starch granules indistinguishable between Dioscorea alata and D. pentaphylla were identified on two artefacts. Grass phytolith morphotypes were found on four artefacts, with three of those assemblages consisting of a group of morphotypes typical of Saccharum and Themeda species. Globular echinate phytoliths, more likely of palms but also present in gingers, were identified on two artefacts. Phytoliths were mostly found in association with starch.

5.2.1. K75/S178 Artefact K75/S178 is from a Phase 1 channel. It is a pestle or pounding/grinding stone (Fig. 10), has smoothing on most surfaces, though one side is slightly more polished and has a green-brown residue. There is minimal polish development, and no phytoliths were observed. Impacted residue was removed from one end by sonic cleaning and contained abundant starch granules, several of which are the size and shape typical of Dioscorea spp., including some with maximum dimensions above 50 mm (Fig. 8E,F). The size and elongated ovoid shape are typical of both D. alata and D. pentaphylla. Some modern reference D. pentaphylla starch granules have a very distinct sculptured end (Fig. 5C,D). This sculptured feature appears to be absent on smaller starch granules of this species, so we cannot eliminate D. pentaphylla. Initial study suggested that one starch granule (Fig. 8G,H) was similar in shape to starch granules of some Musa ingens specimens (Fig. 6E,F), but its size (above 70 mm) was much larger. Although similar in size and shape to some starch granules of modern cultivated triploid banana (AAA), the latter are generally smaller. The larger granules on K75/S178 are beyond the size range of Musa specimens, some of which are similar in shape to some Dioscorea species (Figs. 4 and 5), though far less regular. On the basis of size and shape we have eliminated D. esculenta; D. bulbifera (distinctive flat base, triangular shape and is bent at the hilum end); D. rotundata (a known modern introduction); and D. trifida (much smaller, and a known modern introduction). D. nummularia starch granules are more symmetrical and generally smaller. Other starch granules

602

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

603

Fig. 6. Starch granules from comparative reference collections of Musa. Brightfield images were taken with (B,D,F) and without (A,C,E) polarising filters. (A,B) Musa variety AAA; (C,D) Musa variety ABB; (E,F) Musa ingens. (Photos: C. Lentfer.)

10e20 mm in diameter are common, but have not been identified. The functional traces on this artefact indicate use as a pestle for mashing soft, non-siliceous, raw starchy plants including Dioscorea sp.

5.2.2. K76/S28 Artefact K76/S28 is a chert fragment, probably the distal end of a broken flake (Fig. 10C,D), from the edge of a Phase 1 feature. The dorsal edge of the break has scarring, rounding, striations (mostly perpendicular to the edge) and smooth

Fig. 5. Starch granules from tubers in the comparative reference collections under DIC (A,C,E,G) and polarising filters (B,D,F,H). (A,B) Starch granules of Dioscorea esculenta tuber (AusMus 1/258; PNG). Granules are 3e10 mm in maximum dimension. They appear in sheets or as clusters (compound grains), and individual granules are highly faceted often with scalloped margins and ‘ridges’ from the hilum to the margin. In some cases they appear square in plan view. D. esculenta granules are irregular in shape and no circular granules have been observed. (C,D) Starch granules of D. pentaphylla (AusMus 1/167, unknown source). Granules exhibit a range of shapes (including elongate, elliptical and tear drop) with the narrow end always at the hilum. As for other Dioscorea species, it has an eccentric hilum. In many cases, the end of the granule, near the hilum, is clearly sculptured and appears to be distinctive for this species. The larger granules are up to 100 mm in maximum dimension, and smaller granules are commonly above 30 mm. (E,F) Starch granules of D. nummularia (AusMus 1/281, Vanuatu). The starch granules of D. nummularia are generally triangular in shape with a width:length ratio of w0.6. Some granules are oblong to elliptical in shape. The hilum is eccentric. D. nummularia is distinguished by fine, apparently lightly incised lines that run down the centre of the granule from the hilum to the base. These lines have not been observed in granules that are elliptical in shape where the maximum width is w30% of the maximum length. (G,H) Starch granules of D. trifida tuber (AusMus 1/278, Vanuatu). Granule shapes are circular, elliptical or ovate (the latter being the dominant shape). The surface of the granules is smooth, and lamellae are not apparent. The hilum is eccentric, the granule profile is elliptical with a pinched ridge running from the hilum down the centre of the granule. The sizes for D. trifida range from approximately 25 to 50 mm. (Photos: J. Field, ZeissÔ Axioskop 2, AxioCam HRcÔ.)

604

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

605

5.2.4. K76/S29B Artefact K76/S29B (Fig. 9B) is a retouched chert fragment (probably the distal end of a flake) from the edge of a Phase 1 feature. Usewear includes scarring and reticular polish typical of woodworking. Alignment of polish and perpendicular striations on the edge opposite the retouch indicate scraping. There is also abundant mashed plant tissue with fibres oriented at right angles to the edge, indicating a scraping motion. Residues from extractions include abundant small starch granules present both singly and in sheets (Fig. 8A,B). The granules range in size from less than 3 mm to approximately 8 mm, but are commonly 5 mm diameter, and roughly spherical with faceting on the surface, which is most obvious on larger granules. These features are typical of C. esculenta starch granules in modern reference specimens (Fig. 3A,B). Raphides with square ends were also observed. Phytoliths include non-diagnostic epidermal long cell and psilate elongate, common to several plant groups including grasses, palms and gingers. Given the nature of the use-polish, we think palm working is most likely. The usewear and retouch indicate that the tool was used for scraping hard silicious woody tissue, that it had been sharpened at least once, and was subsequently used to scrape or slice soft, starchy, raw taro (Colocasia esculenta).

Unifacial flaking is present around 75% of the edge, which also has usewear (marked rounding, scarring and polish) in association with impacted green-brown tissue and red coloured tissue (possibly resin impregnated fibres). Initial extractions from this artefact had abundant small starch granules similar in size to Colocasia esculenta (Michael Therin 2001, personal communication). These granules display the typical faceting of C. esculenta (cf. Figs. 3A,B and 8A,B). Subsequent extractions provided fewer small starch granules (with faceting, but not in clusters). In addition to the smaller starch granules, several large (up to 60 mm diameter) ovoid starch granules (Fig. 8C,D,G,H) were recovered and are typical in size and shape of Dioscorea spp. (Figs. 4 and 5). We have eliminated all except D. alata and D. pentaphylla (see also discussion above under K75/ S178). Some Musa ingens starch granules are similar in size and shape to those of Dioscorea bulbifera, although the former are far more heterogeneous in our modern reference samples (see Fig. 6). Further analysis of Musa starch granules is required, although any abundance on stone artefacts would be surprising. Our modern reference sample of D. bulbifera is probably from a cultivated variety, and the homogeneity of starch granule form may provide an indicator of cultivation (Dolores Piperno, 2002, personal communication). None of the tool starch granules from Kuk have the highly distinctive form of D. bulbifera. Other starch granules observed do not match either Colocasia esculenta or Dioscorea spp., and further study of modern reference collections is ongoing. Trichome and epidermal long cell morphotypes typical of Saccharum were also identified on K75/S179 (Fig. 11GeI). However, the morphotypes, comprising two trichomes and an epidermal long cell identified on K75/S179 are also found in Themeda. Artefact K75/S179 was probably used as a core for flake production and also as an implement for chopping soft woody tissue and other plants. Rounding and other usewear is concentrated on three parts of the retouched edge with high frequency of starch granules. The association of usewear (specifically marked edge rounding) with identified starch granules indicate use to mash Dioscorea and Colocasia.

5.2.5. K75/S179 Artefact K75/S179 (Fig. 9E) is a large (1238.1 g) volcanic pebble core found within grey clay between Phases 1 and 2.

5.2.6. K98/217 Artefact K98/217, from the grey clay, is retouched chert flake with at least three utilised edges. Unidentified starch

use-polish suggesting plant processing. The dorsal surface has cellulose, a residue film (probably plant exudate), phytoliths and bright polish. Extractions from this surface had starch granules, which have not yet been identified. The tool was probably used once for scraping and slicing starchy and siliceous plant. 5.2.3. K76/S29A Artefact K76/S29A (Fig. 9C) is a chert flake with a diffuse bulb of percussion. Three discrete edges have scarring (fine retouch on one edge), rounding, striations and grainy to very smooth polish. Where the polish is most developed it has a reticular pattern typical of use-polish on experimental wood scraping tools. Impacted fibres were visible, and unidentified starch granules were recorded in residue extractions from the distal utilised edge. No phytoliths were observed.

Fig. 7. Starch granules from tubers in the comparative reference collections under DIC (A,C,E) and polarising filters (B,D,F). (A,B) Starch granules of Alocasia macrorrhizos tuber (AusMus 1/274; unknown source). Granules are circular to ovate in shape and range from 8 to 20 mm in maximum dimension. They are faceted and twinning is common with up to 8 granules observed in clusters. Damage to the hilum in these images is an artefact of the mounting media. (C,D) Starch granules of Cyrtosperma chamonissonis tuber (AusMus 1/272; Fiji). Granules are circular to ovate in shape with most granules exhibiting faceting on one or more surfaces. The hilum is centric and the sizes range from around 8 to 17 mm. Twinning is common and some granules that have faceting on many surfaces appear to be hexagonal in shape. (E,F) Starch granules of Amorphophallus campanulatus tuber (AusMus 1/276; PNG). Granules are circular to ovate and exhibit a high degree of faceting. Granules have a centric hilum with size range about 8e20 mm. They appear in sheets or as clusters (compound grains) and individual granules are highly faceted often appearing to have scalloped margins with ‘ridges’ from the hilum to the points of the scallops. In some cases they appear square in plan view. (G,H) Starch granules of Xanthosoma sagittiflora tuber (AusMus 1/182 Tonga). Granules of X. sagittiflora are circular to ovate, and have a high incidence of faceting on one or more surfaces. Twinning is common though 3 or more granules are generally joined together. The hilum is centric, and in these images the pitting at the hilum is an artefact of the mounting media used in preparation (Euparol). Granule size range is 3e20 mm, mostly 10e16 mm. Small granules are circular. (Photos: J. Field, ZeissÔ Axioskop 2, AxioCam HRcÔ.)

606

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

Fig. 8. Starch granules from Kuk Swamp artefact extractions, under DIC (A,C,E,G) and polarising filters (B,D,F,H). (A,B) Artefact K76-29B, Phase 1; (C,D) Artefact K75-179, Phase 1-2; (E,F) Artefact K75-S178, Phase 1; (G,H) Artefact K75-179, Phase 1-2. (Photos: J. Field, ZeissÔ Axioskop 2, AxioCam HRcÔ.)

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614 Table 1 Size variation mm in Dioscorea spp. starch granules Species

Loy et al. [58]

Seidemann [73]

Dioscorea esculenta D. nummularia D. trifida

3e10 25e45 27e40

1e5

D. D. D. D.

25e50 24e55 19e75 40e60

5e45 5e50

bulbifera alata pentaphylla rotundata

Notes

Modern introduction to Pacific

5e60

Modern introduction to Pacific

granules are also present with very diffuse extinction crosses, suggesting gelatinisation from heating (Fig. 12B). Usewear includes scarring, striations, edge rounding and developed reticular polish (Fig. 12A) typical of working very siliceous woody tissue, possibly grass, reeds or palm. Trichome and epidermal long cell morphotypes typical of Saccharum and Sorghum were identified (Fig. 11GeI). Globular phytoliths of 10 mm diameter and regular echinate decoration were also identified (Fig. 11L). These echinate morphotypes are characteristic of Palmae and Zingiberaceae, being common in the leaves, fruit and wood of numerous palms including the Calamus, Actinorhytus, Areca, Calyptrocalyx, Daeminodoropis, Heterospathe, Licuala, Metroxylon, Phtychosperma species and gingers, in particular, Hornstedtia, Amomum and Kaempferia species (Fig. 11MeR). Although it is difficult to differentiate between the two families, preliminary analyses comparing palm and ginger assemblages indicate that phytoliths comparable to the residue samples (i.e. with regular echinate decoration and 10 mm diameter) are rare in Zingiberaceae but occur commonly in many palm species [54]. On this basis, therefore, the phytoliths found on both artefacts are considered more typical of Palmae than Zingiberaceae.

Table 2 Tool functions of Kuk artefacts (n ¼ 55) by contact material for all phases Contact material

No. of artefacts

Wood Wood/skin Wood/reeds Wood/bone Soft plant Ochre (?) Uncertain Not used Total

17 1 1 1 7 1 22 5 55

No. of artefacts with starch granules 7 1 1 6 1 2 18

‘Wood’ is identified by the dominant usewear pattern. ‘Wood/skin’, ‘Wood/ reeds’ and ‘Wood/bone’ indicate overlapping usewear patterns. ‘Soft plant’ is indicated by a combination of usewear consistent with plant processing together with abundant starch. Ochre identification is discussed in the text. ‘Uncertain’ indicates that usewear is not distinctive of material worked, and that the residues (if present) may not be associated with utilised edges.

607

5.2.7. K76/36 Artefact K76/36 (Fig. 9D) is a chert flake from a Phase 2 palaeosurface. There is usewear (scarring and rounding) on the distal ventral edge. Residues include red films, cellulose and resin impregnated fibres. Polish is not well-developed, and obscured by labelling glue. The usewear and residues suggest scraping hard woody tissue. Starch granules have not been identified. 5.2.8. K77/S4 Artefact K77/S4 is a retouched chert fragment with slight rounding and polish (Fig. 12C). Starch granules, which are larger than 6 mm, have not been identified (Fig. 12D). Phytoliths were observed under reflected light, but have not been identified. It was probably used to scrape soft-medium siliceous plant tissue. 5.2.9. K77/S9 Artefact K77/S9 is a probable grinding stone fragment from a Phase 2 feature, with striations and distinct smoothing with low polish development (Fig. 10). Starch granules have not been identified. A trichome found on K77/S9 is typical of panicoid grasses including Saccharum, Setaria, Themeda, Sorghum and Cymbopogon, the elongated morphotype with echinate margins is less diagnostic, being found in several other modern plant groups as well as grasses (Fig. 11J,K). 5.2.10. K77/S20 Artefact K77/S20 (Fig. 9F) is a chert fragment, retouched along one edge with one small step scar on the opposite surface, and slight edge rounding and reticular polish, similar to that found on wood working tools. Small starch grains were observed under reflected light but they have not been identified. 5.2.11. K77/S17 Artefact K77/S17 (Fig. 10) is a heavily weathered volcanic pebble collected from an early Phase 2 context. There is smoothing and striations suggesting it was used for grinding. A distinct groove on the surface appears to have been naturally formed. Residues are impacted on one surface and include plant tissue and possible red ochre. Small starch granules were found in sheets of tissue, with clusters of spherical faceted granules about three to four microns diameter within the typical size range of C. esculenta. Although this size range also overlaps with C. gigantea, and the sheets of tissue are typical of modern reference slides of that species (Fig. 3F,G), C. gigantea is not found naturally in Melanesia [58]. Consequently and given their age, the starch granules are thought to represent C. esculenta. Numerous starch granules between 10 and 20 mm diameter were also found but have not yet been identified. A non-diagnostic polyhedral phytolith and a trichome, were identified on K77/S17. Both morphotypes are common to several plant groups including grasses and gingers. Further investigation of the unidentified starch granules found on this artefact may assist with more definitive determination of

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

608

B

A

1 cm

C

1 cm

D

1 cm

E

1 cm

F

5 cm

1 cm

G H

1 cm

1 cm

I

1 cm

Extent of usewear Retouch

Fig. 9. Kuk Stone artefacts: (A) K76 S28; (B) K76 S29B; (C) K76 S29A; (D) K76 36; (E) K75 S179; (F) K77 S20; (G) K98 217; (H) K77 S4; (I) K76 S19. (Illustrations: R. Fullagar.)

source of these phytoliths. The residues and usewear indicate grinding a variety of starchy and siliceous plants. 5.2.12. K76/S19 Artefact K76/S19 (Fig. 9I), from a Phase 3 feature fill, is a small retouched chert fragment with slight rounding, small step scars and a band of undulating polish with pitting intermediate to that found on bone working and some wood working tools. No starch granules were found. Phytolith

morphotypes including a bilobate epidermal short cell, epidermal long cells, a tracheid and a trichome (Fig. 11AeF). All six morphotypes identified in the assemblage are characteristic of grasses and are typical of Saccharum in particular. While each of the morphotypes is also found in other grasses there is no evidence that they all co-occur in any of the other 59 modern grass genera examined in this analysis. Therefore, although it is possible that multiple species from other grass genera (e.g. Sorghum, Setaria, Cymbopogon and Themeda) contributed to

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

609

Fig. 10. Kuk Stone artefacts: (Upper) K75 S178, scale bar is 1 cm; (Middle) K77 S17, scale bar is 1 cm; (Lower) K77 S9 with grinding surface shown at far left, scale bar divisions are 1 cm. (Photos: D. Page.)

the residue phytolith assemblage, the most likely single source is considered to be from one of the four species of Saccharum including the two cultivated species S. edule and S. officinarum, and the two wild species S. spontaneum and S. robustum. Globular phytoliths of 10 mm diameter and regular echinate decoration were identified on K76/S19 (as for artefact K98/ 217, Fig. 11L). On the basis of size (see above under K98/ 217) the phytoliths are considered more typical of Palmae than Zingiberaceae. Moreover, the polish is similar to experimental usewear on palm-working tools. 6. Discussion The Stage 1 study reveals excellent preservation of residues at Kuk, and a range of tool activities with evidence of

grinding, pounding and scraping of siliceous, woody and soft, starchy plants. Wear patterns indicate multiple tasks including skin and bone working tools that were also used for plant processing. Starch granules are well preserved, and were removed in high abundance by sonication. Some larger granules are typical of Dioscorea. Smaller starch granules (less than 10 mm) were most common in sheets of tissue, typical of Colocasia esculenta. The presence of starch typical of known food plants provides evidence not just for the use of tools in food processing, but it adds support to previous arguments that these plants were early crops or domesticates in the New Guinea Highlands. Colocasia esculenta starch granules were found on two flaked artefacts and a probable pounding/grinding stone or pestle. Dioscorea spp. starch was found on one flaked

610

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

Fig. 11. Phytoliths. Scale bar ¼ 10 mm. (AeF) Typical Saccharum phytoliths from Artefact K76/S19. (A) Bilobate epidermal short cell. (BeD) Epidermal long cells. (E) Tracheid. (F) Trichome. (GeI) Grass phytoliths from Artefact K75/S179 common in Saccharum and Themeda species: (G) epidermal long cell; (H,I) trichomes. (JeK) Phytoliths from Artefact K77/S9: (J) trichome typical of panicoid grasses including Saccharum, Setaria, Themeda, Sorghum and Cymbopogon species on Artefact K77/S9; (K) non-diagnostic elongated morphotype with echinate margins on Artefact K77/S9. (L) Echinate globular morphotype from Artefact K76/S19. (MeR) Examples of echinate globular phytoliths from modern reference samples of palms (MeO) and gingers (PeR): (M) Calamus sp. WNB519lf2; (N) Calyptrocalyx sp. WNB521spth3; (O) Areca catechu; (P) Hornstedtia scottiana; (Q) Amomum sp.; (R) Kaempferia gilbertii. (Photos: M. Therin and Carol Lentfer, Olympus BH-2Ô microscope.)

artefact and a pestle. These artefact types suggest taro and yam were sliced or scraped and also mashed or pounded. Our sample size is insufficient to indicate the relative importance of different plants to prehistoric inhabitants, and there are many unidentified starch residues that require further analyses and comparison with modern reference collections. Further research is needed to determine the effects of different kinds of tuber processing on the morphology and optical properties of starch granules [4]. Identification of phytoliths from grasses, palms and possibly gingers provides further evidence for use of additional plant resources and multifunctionality, although quantitative analyses are needed for more definitive results. Preliminary analysis suggests that artefacts K76/S19 and K98/217 were

used primarily on hard siliceous woody material perhaps palm, and robust culms of tall grasses (possibly Saccharum or Sorghum). Artefact K76/S29B was apparently used on Dioscorea, Colocasia and robust grasses or palm wood. Panicoid grass phytoliths on artefacts K77/S9 and 75/S179 (neither with usewear indicative of working siliceous material) indicate low intensity use or possible contamination. Artefact K77/S17 has Colocasia starch, and possibly other useful plants such as grasses and gingers. Although grasses, palms and gingers are utilised as food, Zingiberaceae ginger is used mainly for ritual and medicinal purposes; Saccharum and other canes are used for housing, fencing and arrow shafts; and palms have multiple uses in housing and numerous artefacts [70].

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

611

Fig. 12. (A) Use polish on Artefact K98/217, showing smoothing, bright reflectivity and alignments perpendicular to the edge. The extent and brightness of the polish is typical of scraping very siliceous plants, possibly reeds or palm. Width of field: 0.3 mm. (B) Starch granules (under crossed polarising filters) on Artefact K98/217 along the polished edge. Note that the extinction crosses are very weak, suggesting gelatinisation probably from heating or cooking. Width of field: 0.2 mm. (C) Use polish on artefact K77S4. Width of field: 0.3 mm. (D) Starch granules (under polarising filters) on artefact K77S4 along the utilised edge. Width of field: 0.2 mm. (Photos: R. Fullagar.)

The history of food processing techniques at Kuk (1560 m above mean sea level) spans periods before, during and after the construction of elaborate field systems. Wild yams in New Guinea are gathered as food up to only 100 m above mean sea level [85, p. 836] and wild forms of taro grow in the Wahgi Valley [85, p. 835]. Further study may help us to distinguish wild and cultivated food plants on the basis of starch granule morphology, following the research of Ugent [79e81] on South American potatoes. The presence of Colocasia esculenta and Dioscorea sp. during the early-to-mid Holocene at Kuk is consistent with previous interpretations of these plants being indigenous to New Guinea. At present, it is unclear how these plants reached Kuk, and presumably other Highland locales. They could have dispersed naturally as seeds, given that C. esculenta and Dioscorea spp. would have been within their altitudinal ranges during the warmer and wetter climates of the earlyto-mid Holocene. Alternatively, the plants could have been deliberately brought by people engaged in plant exploitation or nascent agriculture, or inadvertently disseminated following the discard of regenerative plant parts [20]. Although yam and taro were being used for food during the early and mid Holocene, it is unclear whether the starch granules are derived from wild or domesticated varieties, or whether they were being cultivated. Circumstantially, the early Holocene use of these two root crops, which are major staples across New Guinea, lends support to arguments for the longterm targeting and selection of these plants, practices that

ultimately led to their domestication in the region [52]. Additionally, the processing of these plants, in a location subject to wetland manipulation for plant exploitation (Phase 1) and agriculture (Phase 2) is suggestive of deliberate targeted resource use and cultivation, respectively. In summary, usewear and residue analyses on 55 excavated artefacts from Kuk, Papua New Guinea, indicate that 50 had traces of use including 18 artefacts with starch granules and 8 with phytoliths. Identifiable phytoliths of grasses, palm and/or ginger were noted on extractions from some artefacts. Twenty artefacts had distinct usewear indicative of woodworking. Starch granules were observed on artefacts associated with early (Phase 1) and mid (Phase 2) Holocene plant exploitation on the wetland margin. Numerous starch granules of Colocasia esculenta were identified on the basis of shape, size, surface morphology, clustering in sheets of tissue, and in one case the co-occurrence of a raphide fragment. No other starch granules in our modern reference collection share all these characteristics. Several starch granules (c. 60 mm maximum dimension) were found on one artefact from Phase 1, and are consistent in morphology and size with the largest Dioscorea starch granules. We can eliminate all but two known species, D. alata and D. pentaphylla, on the basis of size and shape. Numerous unidentified starch granules intermediate in size occur on early Holocene artefacts (Phase 1) and will be the subject of further work. Starch granules from palm pith and

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

612

Table 3 Starch granules and phytoliths on Kuk artefacts (n ¼ 12) from Phases 1e3 Artefact no.

Figures

Phase

Provenience

K75/S178

8E,F, 10

1

Phase 1 channel

K76/S28

9A

1

K76/S29A

9C

1

K76/S29B

8A,B, 9B

K75/S179

Age cal BP

Starch granules

Phytoliths

Artefact type and contact material indicated by usewear and residues

10,220e9910

Yes, D

No

Fill of a Phase 1 feature Edge of a Phase 1 feature

10,220e9910

Yes

Yes

10,220e9910

Yes

No

1

Edge of a Phase 1 feature

10,220e9910

Yes, C

Yes, ND

8C,D, 9E, 11GeI

1e2

Grey clay between Phase 2 and Phase 1

w10,000e7300

Yes, C, D

Yes, G, Sac? Th

K98/217

9G, 12A,B

1e2

Grey clay between Phase 2 and Phase 1

w10,000e7300

Yes

Yes, P/Z, G, Sac? Sorg?

K76/36

9D

2

Phase 2 palaeosurface

6950e6440

Yes

No

K77/S4

9H, 12C,D

2

6950e6440

Yes

Yes

K77/S9

10, 11J,K

2

Beneath R þ W ash in Phase 2 feature fill Beneath R þ W ash in Phase 2 feature

6950e6440

Yes

Yes, G

K77/S20

9F

2

6950e6440

Yes

No

K77/S17

10

2

6950e6440

Yes, C

Yesa, ND

K76/S19

9I, 11AeF,L

3

Associated with Phase 2 palaeosurface Associated with Phase 2 palaeosurfacedwithin R þ W ash Phase 3 feature fill

No

Yes, P/Z, G, Sac?

Altered volcanic cobble (greenstone): pestle or pounding/grinding stone, starchy plant Broken chert flake: starchy plant ?Retouched chert flake: hard woody tissue, starchy plant Retouched chert flake: hard woody tissue, starchy plant Volcanic cobble core: uncertain usewear, starchy plant Retouched chert flake: hard woody tissue, starchy siliceous plant Chert flake: hard woody tissue Retouched chert fragment: starchy siliceous plantb Volcanic cobble: probable grinding stone fragment, starchy plant Retouched chert fragment: hard woody tissueb Volcanic cobble: probable grinding stone, starchy plant, ochre Retouched chert fragment: hard woody tissue; bone?b

pre-32602800

C, Colocasia esculenta starch granules; D, Dioscorea sp. starch granules; G, grass phytolith morphotypes (Sac ¼ Saccharum, Sorg ¼ Sorghum, Th ¼ Themeda); P/Z, echinate globular phytolith morphotypes found most commonly in palms (Palmae) and gingers (Zingiberacae); ND, non-diagnostic phytolith morphotypes. a The majority of the phytoliths on tool K77/S17 were fractured and incomplete, consistent with grinding and pounding. b These small fragments lack definitive flake attributes. They are small tabular pieces w1 cm thick with steep retouch. Technologically, they are cores (Fig. 11).

wood are a particular focus of future work to determine if genus and species can be distinguished.

7. Conclusions We have confirmed the presence of taro and yam at Kuk Swamp, during Phase 1 and Phase 2. Use wear and residue analyses also provide the first direct evidence for the processing of these plants at Kuk Swamp. The study demonstrates the use of at least two starch-rich plants, taro (Colocasia esculenta) and a yam (Dioscorea sp.), as well as other plants, in the Highlands of New Guinea during the early and midHolocene. Taro and yam were first exploited over 10,000 years ago, and are likely to be among those first cultivated by 6950e 6440 cal BP. Tool functions include both slicing and scraping, as well as pounding or mashing, which are all processing techniques known ethnographically. Our study of the starch

granules does not permit the distinction of wild and domestic forms. The age of taro (Colocasia esculenta) at Kuk is slightly older than similar remains found on early Holocene stone tools from Kilu Cave, Solomon Islands, where Colocasia was also found on c. 28e20,000 year old artefacts [58]. Although little is known about the origins and palaeoecology of these plants, their presence from the early Holocene demonstrates the botanical and behavioural foundations for the emergence of agriculture in the Highlands. The evidence also establishes the possibility of a long time-depth for the domestication of taro and yams. At present, however, we are only at the beginning of our archaeobotanical investigations into the domestication of yams, taros, Musa banana and other taxa, in the New Guinea region. Our detailed analysis is mostly restricted to a sample of Phase 1 and Phase 2 artefacts. We anticipate that further analyses will provide new information on tool-use, plant processing and lithic technology during the Holocene transition to agriculture.

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

Acknowledgements The authors are indebted to Jack Golson for initiating the usewear and residue studies at Kuk and for his assistance through every step of the investigations. For specific discussion we thank Jack Golson, Peter Matthews, the late Tom Loy, Dolores Piperno, Michael Therin, and Robin Torrence. References [1] J. Alexander, D.G. Coursey, The origins of yam cultivation, in: P.J. Ucko, G.W. Dimbleby (Eds.), The Domestication and Exploitation of Plants and Animals, Gerald Duckworth, London, 1969, pp. 405e425. [2] J. Atchison, R. Fullagar, Starch residues on pounding implements from Jinmium rock-shelter, in: R.Fullagar (Ed.), A Closer Look: Recent Studies of Australian Stone Tools, Sydney University Archaeological Methods Series 6, Archaeological Computing Laboratory, School of Archaeology, University of Sydney, Sydney, 1998, pp. 109e125. [3] J. Allen, Prehistoric agricultural systems in the Wahgi valleyda further note, Mankind 7 (1970) 177e183. [4] M. del, P. Babot, Starch grain damage as an indicator of food processing, in: D.M. Hart, L.A. Wallis (Eds.), Phytolith and Starch Research in the AustralianePacificeAsian Regions: The State of the Art, Terra Australis 19, Pandanus Books, Research School of Pacific and Asian Studies, Australian National University, Canberra, 2003, pp. 69e81. [5] H. Barton, J.P. White, Use of stone and shell artefacts at Balof 2, New Ireland, Papua New Guinea, Asian Perspectives 32 (1993) 169e181. [6] H. Barton, R. Torrence, R. Fullagar, Clues to stone tool function re-examined: comparing starch granule frequencies on used and unused obsidian artifacts, Journal of Archaeological Science 25 (1998) 1231e1238. [7] T. Bayliss-Smith, J. Golson, Wetland agriculture in New Guinea Highlands prehistory, in: B. Coles (Ed.), The Wetland Revolution in Prehistory, The Prehistoric Society and Wetland Archaeological Research Project, Exeter, 1992, pp. 15e27. [8] T. Bayliss-Smith, J. Golson, A Colocasian revolution in the New Guinea Highlands? Insights from Phase 4 at Kuk, Archaeology in Oceania 27 (1992) 1e21. [9] T. Bayliss-Smith, J. Golson, The Meaning of ditches: deconstructing the social landscapes of drainage in New Guinea, Kuk, Phase 4, in: C. Gosden, J. Hather (Eds.), The Prehistory of Food: Appetites for Change, Routledge, London, 1999, pp. 199e231. [10] P. Bellwood, Prehistory of the Indo-Malaysian Archipelago, University of Hawaii Press, Honolulu, 1997. [11] B.J.H. Bradbury, W.D. Holloway, Chemistry of Tropical Root Crops: Significance for Nutrition and Agriculture in the Pacific, Australian Centre for International Agricultural Research, Canberra, 1988. [12] D.J. Coates, D.E. Yen, P.M. Gaffey, Chromosome variation in taro, Colocasia esculenta: implications for origin in the Pacific, Cytologia 53 (1988) 551e560. [13] A. de Candolle, Origin of Cultivated Plants, Kegan Paul, London, 1884. [14] D.G. Coursey, Yams, Longmans, Green and Company, London, 1967. [15] D.G. Coursey, The civilizations of the yam: interrelationships of man and yams in Africa and the Indo-Pacific region, Archaeology and Physical Anthropology in Oceania VII (1) (1972) 215e233. [16] D.G. Coursey, Yams: Dioscorea spp. (Dioscoreaceae), in: N.W. Simmonds (Ed.), Evolution of Crop Plants, Longman, London, 1976, pp. 70e74. [17] T.P. Denham, The Kuk morass: multi-disciplinary investigations of early to midHolocene plant exploitation at Kuk Swamp, Wahgi Valley, Papua New Guinea, unpublished PhD thesis, Australian National University, Canberra, 2003. [18] T.P. Denham, Archaeological evidence for mid-Holocene agriculture in the interior of Papua New Guinea: a critical review, in: T.P. Denham, C. Ballard (Eds.), Perspectives on Prehistoric Agriculture in the New Guinea Highlands, Archaeology in Oceania 38 (2003) 159e176.

613

[19] T.P. Denham, Early agriculture in the highlands of New Guinea: an assessment of Phase 1 (c.9000 bp) at Kuk Swamp, in: V. Attenbrow, R. Fullagar (Eds.), A Pacific Odyssey: Archaeology and Anthropology in the Western Pacific, Papers in Honour of Jim Specht, Australian Museum, Sydney, Records of the Australian Museum Supplement 29, 2004, pp. 47e57. [20] T.P. Denham, H. Barton, The emergence of agriculture in New Guinea: continuity from pre-existing foraging practices, in: D.J. Kennett, B. Winterhalder (Eds.), Behavioral Ecology and the Transition to Agriculture, California University Press, California, in press. [21] T.P. Denham, J. Golson, P.J. Hughes, Reading early agriculture at Kuk (Phases 1e3), Wahgi Valley, Papua New Guinea: the wetland archaeological features (Phases 1e3), Proceedings of the Prehistoric Society 70 (2004) 259e298. [22] T.P. Denham, S.G. Haberle, C. Lentfer, New evidence and revised interpretations of early agriculture in Highland New Guinea, Antiquity 78 (2004) 839e857. [23] T.P. Denham, S.G. Haberle, C. Lentfer, R. Fullagar, J. Field, M. Therin, N. Porch, B. Winsborough, Origins of agriculture at Kuk Swamp in the Highlands of New Guinea, Science 301 (2003) 189e193. [24] J. Field, R. Fullagar, Grinding and pounding stones from Cuddie Springs and Jinmium, in: R. Fullagar (Ed.), A Closer Look: Recent Studies of Australian Stone Tools, SUAMS 6, University of Sydney Archaeological Computing Laboratory, Sydney, 1998, pp. 95e108. [25] R. Fullagar, The potential of lithic use-wear and residue studies for determining stone tool functions, in: P.P. Gorecki, D. Gillieson (Eds.), A Crack in the Spine: Prehistory and Ecology of the Jimi-Yuat Valley, Papua New Guinea, James Cook University of North Queensland, Townsville, 1989, pp. 209e223. [26] R. Fullagar, Residues and usewear, in: J. Balme, A. Paterson (Eds.), Archaeology in Practice: A Student Guide to Archaeological Analysis, Blackwell Publishing, Malden, 2006, pp. 207e234. [27] R. Fullagar, The role of silica in polish formation, Journal of Archaeological Science 18 (1991) 1e25. [28] R. Fullagar (Ed.), A Closer Look: Australian Studies of Stone Tools, SUAMS 6, University of Sydney Archaeological Computing Laboratory, Sydney, 1998. [29] R. Fullagar, J. Field, Pleistocene seed grinding implements from the Australian arid zone, Antiquity 71 (1997) 300e307. [30] R. Fullagar, T. Loy, S. Cox, Starch grains, sediments and stone tool function: evidence from Bitokara, Papua New Guinea, in: R. Fullagar (Ed.), A Closer Look: Australian Studies of Stone Tools, SUAMS 6, University of Sydney Archaeological Computing Laboratory, Sydney, 1998, pp. 49e60. [31] R. Fullagar, B. Meehan, R. Jones, Residue analysis of ethnographic plant-working and other tools from northern Australia, in: P. AndersonGerfaud (Ed.), Prehistory of Agriculture: New Experimental and Ethnographic Approaches, Institute of Archaeology, UCLA, Berkeley, 1999, pp. 15e23. [32] W.C. Gagne´, Staple crops in subsistence agriculture, in: J.L. Gressitt (Ed.), Biogeography and Ecology of New Guinea, vol. 1, Junk, The Hague, 1982, pp. 229e259. [33] J. Golson, No room at the top: agricultural intensification in the New Guinea Highlands, in: J. Allen, J. Golson, R. Jones (Eds.), Sunda and Sahul: Prehistoric Studies in Southeast Asia, Melanesia and Australia, Academic Press, London, 1977, pp. 601e638. [34] J. Golson, Simple tools and complex technology, in: R.V.S. Wright (Ed.), Stone Tools as Cultural Markers: Change Evolution and Complexity, Australian Institute of Aboriginal and Torres Strait Islander Studies, Canberra, 1977, pp. 154e161. [35] J. Golson, Kuk and the development of agriculture in New Guinea: retrospection and introspection, in: D.E. Yen, J.M.J. Mummery (Eds.), Pacific Production Systems, Occasional Papers in Prehistory 18, Department of Prehistory, Research School of Pacific Studies, Australian National University, Canberra, 1990, pp. 139e147. [36] J. Golson, D.S. Gardner, Agricultural and sociopolitical organization in New Guinea Highlands prehistory, Annual Review of Anthropology 19 (1990) 395e417.

614

R. Fullagar et al. / Journal of Archaeological Science 33 (2006) 595e614

[37] J. Golson, P.J. Hughes, The appearance of plant and animal domestication in New Guinea, Journal de la Socie´te´ des Oce´anistes 36 (1980) 294e303. [38] S.G. Haberle, Identification of cultivated Pandanus and Colocasia in pollen records and the implications for the study of early agriculture in New Guinea, Vegetation History and Archaeobotany 4 (1995) 195e210. [39] K. Hardy, P. Sillitoe, Material perspectives: stone tool use and material culture among the Wola, Papua New Guinea, Internet Archaeology (2003) 14. . [40] J. Hall, S. Higgins, R. Fullagar, Plant residues on stone tools, in: W. Beck, A. Clarke, L. Head (Eds.), Plants in Australian Archaeology, Tempus 1, University of Queensland, Anthropology Museum, 1989, pp. 136e160. [41] M. Haslam, Evidence for maize processing on 2000-year-old obsidian artefacts from Copan, Honduras, in: D.M. Hart, L.A. Wallis (Eds.), Phytolith and Starch Research in the AustralianePacificeAsian Regions: The State of the Art, Terra Australis 19, Pandanus Books, Research School of Pacific and Asian Studies, Australian National University, Canberra, 2003, pp. 153e161. [42] M. Haslam, The decomposition of starch grains in soils: implications for archaeological residue analyses, Journal of Archaeological Science 31 (2004) 1715e1734. [43] J. Hather, The Archaeobotany of subsistence in the Pacific, World Archaeology 24 (1992) 70e81. [44] G.S. Hope, J. Golson, Late Quaternary change in the mountains of New Guinea, Antiquity 69 (1995) 818e830. [45] M. Horrocks, G. Irwin, M. Jones, D. Sutton, Starch grains and xylem cells of sweet potato (Ipomoea batatas) and bracken (Pteridium esculentum) in archaeological deposits from northern New Zealand, Journal of Archaeological Science 31 (2004) 251e258. [46] A. Jackes, Spot the difference: analyses of starch granules of the Dioscorea genus piloting a method in the search for differentiating and identifying features, unpublished BEnvSc thesis, University of Wollongong, 1997. [47] R. Jones, B. Meehan, Plant foods of the Gidjingali: ethnographic and archaeological perspectives from northern Australia on tuber and seed exploitation, in: D.R. Harris, G.C. Hillman (Eds.), Foraging and Farming: The Evolution of Plant Exploitation, Unwin Hyman, London, 1989, pp. 120e135. [48] J. Kamminga, Over the Edge, in: Occasional Papers in Anthropology 12, Anthropological Museum, University of Queensland, 1982. [49] L. Kealhofer, R. Torrence, R. Fullagar, Integrating phytoliths within usewear/residue studies of stone tools, Journal of Archaeological Science 26 (1999) 527e546. [50] L. Kealhofer, D.R. Piperno, Opal phytoliths in Southeast Asian flora, in: Smithsonian Contributions to Botany, Number 88, Smithsonian Institution Press, Washington, DC, 1998. [51] K.M. Kurvilla, A. Singh, Karyotypic and electrophoretic studies on taro and its origin, Euphytica 30 (1981) 405e413. [52] V. Lebot, Biomolecular evidence for plant domestication in Sahul, Genetic Resources and Crop Evolution 46 (1999) 619e628. [53] V. Lebot, K.M. Aradhya, Isozyme variation in taro (Colocasia esculenta (L.) Schott) from Asia and Oceania, Euphytica 56 (1991) 55e66. [54] C.J. Lentfer, Plants, People and Landscapes in Prehistoric Papua New Guinea. A Compendium of Phytolith (and Starch) Analyses, unpublished Ph.D. thesis, Southern Cross University, Lismore, NSW, Australia, 2003. [55] C.J. Lentfer, Opal phytoliths of Papua New Guinea flora, Records of the Australian Museum, submitted for publication. [56] C. Lentfer, M. Therin, R. Torrence, Starch grains and environmental reconstruction: a modern test case from West New Britain, Papua New Guinea, Journal of Archaeological Science 29 (2002) 687e698. [57] T.H. Loy, Methods in the analysis of starch residues on prehistoric stone tools, in: J. Hather (Ed.), Tropical Archaeobotany, Routledge, London, 1994, pp. 86e114. [58] T.H. Loy, M. Spriggs, S. Wickler, Direct evidence for human use of plants 28,000 years ago: starch residues on stone artifacts from the northern Solomon Islands, Antiquity 66 (1992) 898e912. [59] F.W. Martin, A.M. Rhodes, Intra-specific classification of Dioscorea alata, Tropical Agriculture (Trinidad) 54 (1977) 1e13.

[60] P.J. Matthews, A possible wild type taro: Colocasia esculenta var. aquatilis, Indo-Pacific Prehistory Association Bulletin 11 (1991) 69e81. [61] P.J. Matthews, Aroids and Austronesians, Tropics 4 (2) (1995) 105e126. [62] P.J. Matthews, Taro planthoppers (Tarophagus spp.) in Australia and the origins of taro (Colocasia esculenta) in Oceania, in: T.P. Denham, C. Ballard (Eds.), Perspectives on Prehistoric Agriculture in the New Guinea Highlands, Archaeology in Oceania 38 (2003) 159e176. [63] P.J. Matthews, R. Terauchi, The genetics of agriculture: DNA variation in taro and yam, in: J.G. Hather (Ed.), Tropical Archaeobotany: Applications and New Developments, Routledge, London, 1994, pp. 251e262. [64] F.D. McCarthy, Australian Aboriginal Stone Implements, The Australian Museum, Sydney, 1967. [65] J.F. O’Connell, Spoons knives and scrapers: the function of Yilugwa in central Australia, Mankind 9 (1974) 189e194. [66] D.M. Pearsall, K. Chandler-Ezell, J.A. Zeidler, Maize in ancient Ecuador: results of residue analysis of stone tools from the Real Alto site, Journal of Archaeological Science 31 (2004) 423e442. [67] L. Perry, Starch analyses reveal the relationship between tool type and function: an example from the Orinoco valley of Venezuela, Journal of Archaeological Science 31 (2004) 1069e1081. [68] D.R. Piperno, I. Holst, The presence of starch grains on prehistoric stone tools from the lowland Neotropics: indications of early tuber use and agriculture in Panama, Journal of Archaeological Science 25 (1998) 765e776. [69] D.R. Piperno, A.J. Ranere, I. Holst, P. Hansell, Starch grains reveal early root crop horticulture in the Panamanian tropical forest, Nature 407 (2000) 894e897. [70] J.M. Powell, Vegetation, in: K. Paijmans (Ed.), New Guinea Vegetation, CSIRO and ANU Press, Canberra, 1976, pp. 23e105. [71] J.M. Powell, Plant resources and palaeobotanical evidence for plant use in the Papua New Guinea Highlands, Archaeology in Oceania 17 (1982) 28e37. [72] E.T. Reichart, The Differentiation and Specificity of Starches in Relation to Genera, Species, etc, Carnegie Institution of Washington, Washington, DC, 1913. [73] J. Seidemann, Sta¨rke-Atlas, Paul Parey, Berlin/Hamburg, 1996. [74] P. Sillitoe, K. Hardy, Living Lithics, Ethnography and archaeology in Highland Papua New Guinea, Antiquity 77 (2003) 555e566. [75] J. Specht, R. Fullagar, R. Torrence, N. Baker, Preliminary Report on Archaeological Research in West New Britain Province, Papua New Guinea, JulyeAugust, 1989, Unpublished report for West New Britain Cultural Centre, and Papua New Guinea National Museum and Art Gallery, 1989. [76] P. Swadling, Papua New Guinea’s Prehistory: An Introduction, National Museum and Art Gallery, Port Moresby, 1981. [77] M. Therin, R. Torrence, R. Fullagar, Australian Museum starch reference collection, Australian Archaeology 44 (1997) 52e53. [78] R. Torrence, R. Wright, R. Conway, Identification of starch granules using image analysis and multivariate techniques, Journal of Archaeological Science 31 (2004) 519e532. [79] D. Ugent, S. Pozorski, T. Pozorski, Prehistoric remains of the sweet potato from the Casma Valley of Peru, Phytologia 49 (1981) 401e415. [80] D. Ugent, S. Pozorski, T. Pozorski, Archaeological potato tuber remains from the Casma Valley of Peru, Economic Botany 36 (1982) 182e192. [81] D. Ugent, S. Pozorski, T. Pozorski, New evidence for ancient cultivation of Canna edulis in Peru, Economic Botany 38 (1984) 417e432. [82] P. Van Peer, R. Fullagar, S. Stokes, R. Bailey, J. Moeyersons, F. Steenhoudt, A. Geerts, T. Vanderbeken, M. De Dapper, F. Geus, The early to middle stone age transition and the emergence of modern human behaviour at site 8-B-11, Sai Island, Sudan, Journal of Human Evolution 45 (2003) 187e193. [83] J.P. White, Stone naip bilong timbuna: the living stone age in New Guinea, in: D. de Sonneville-Bordes (Ed.), La Prehistorie: probleme et tendences, Centre National de la Recherche Scientifique, Paris, 1968, pp. 511e516. [84] S.M. Wilson, Phytolith evidence from Kuk, an early agricultural site in New Guinea, Archaeology in Oceania 20 (1985) 90e97. [85] D.E. Yen, The development of Sahul agriculture with Australia as bystander, Antiquity 69 (1995) 831e847. [86] D.E. Yen, J.M. Wheeler, Introduction of taro into the Pacific: the indications of chromosome numbers, Ethnology 7 (1968) 259e267.