The discussion of agricultural change in Tongan prehistory is the product of the direction of past research, rather than a paucity of evidence for this floral element of subsistence within the Tongan archaeological record. Microfossil research in other locations has demonstrated the applicability of this analysis towards identifying the presence of crops and the nature of their use within archaeological contexts (Fullagar et al. 1998, 2006; Horrocks et al. 2004; Horrocks, Grant-Mackie and Matissoo-Smith 2008; Loy et al. 1992; Piperno 2009; Summerhayes at al. 2010, among others). Similarly macrobotanical analysis has been used to provide direct evidence of plant cultivation and domestication in the New and Old Worlds, and the Indo- Pacific (Fuller 2007; Oliveira 2008; Paz 2001; Thompson 1994; Ugent et al. 1981).
Unlike archaeological investigations in the Old World, archaeobotanical techniques have not been consistently applied in the Pacific region. This is due to several factors. The first of these is that the preservation of plant remains in tropical climates is more variable than in the cooler or arid northern latitudes (Paz 2005). Second, the agricultural practices of most prehistoric Pacific communities revolved around the production and consumption of starchy tree nuts, roots and tubers, rather than small seeds such as wheat or rice which have been the traditional focus of archaeobotany (Hather 1994:51). Finally, many of the main domestic plants such as taro (Colocasia esculenta), bananas (Musa spp.), and breadfruit (Artocarpus spp.) are not genetically or phenotypically identifiable from wild populations (Fairburn 2005a). Consequently, archaeobotanical investigations in the tropics have often been based on chance finds (Paz 2005; Glover 1979). Recently, rock shelters, caves, and waterlogged sites have provided new site types for the preservation of plant remains in tropical climates (Glover 1979; Oliveira 2008; Paz 2001, 2005; Fairburn 2005a). Analysis of those remains that have been recovered have suggested that many tropical plant production systems are a poor fit within the traditional typologies of foraging or farming using wild or domestic species, and so the whole global system of agricultural classification has subsequently come into question (Fairburn 2005a).
These technical and theoretical issues have complicated the interpretation of plant use within the archaeobotanical record in the Pacific. Here, the goal is to review the antiquity of plant use within the Pacific economy, through focusing on the identification of macro- and microbotanical remains of edible root, tuber, and tree crops. Clearly domestication cannot be sufficiently discriminated from crop use through morphological or genetic variability within most of these longer-lived taxa, therefore the history of edible plant use from Island Southeast Asia, through Near Oceania to Remote Oceania will be assessed in terms of the archaeological contexts from which these plants were identified and the timing of their use. As much as possible archaeobotanical data, whether chance or systematic finds has been incorporated into
21 the review. Through this, a chronology of plant introductions in the Pacific economy will be developed.
Pleistocene: Independent origins for cultivation?
Of the suite of crops that now dominate agriculture in Tonga, including a number of aroids and yams, many have now been proven to have been first adopted for use within the terminal Pleistocene in Near Oceania, rather than brought during the mid-Holocene by Austronesian speakers from Island Southeast Asia. This review will therefore begin within New Guinea, discussing evidence for the incorporation of root, tuber and tree crops into subsistence before tracing plant use east to Remote Oceania. It will also be acknowledged that this was not a unilinear movement, and that some of these same crops were also incorporated into pre- agricultural diets further west, indicating that plants were also being carried in the opposite direction.
The earliest evidence for the adoption of root, tuber and tree crops that comprised significant components of the Pacific economy prior to European contact is in the New Guinea highlands. Dioscorea yam starch residues found on stone artefacts and charred Pandanus have been recovered from Joes Garden, Kosipe Mission and dated to over 40,000 BP using Themoluminescence (TL) dating (Summerhayes et al. 2010). Large amounts of archaeobotanical data have also been collected from years of excavations at Kuk Swamp. Within the various phases of occupation at the site, microfossils in the forms of phytoliths, starch and pollen of known edible cultigens have been identified and interpreted as evidence for the beginning of cultivation and manipulation of these taxa (Denham 2007; Denham et al. 2003, 2004; Donohue and Denham 2009; Golson 2007; Wilson 1985). Whether these starch-rich plants were incorporated into an existing highland subsistence system or introduced from lower elevations as part of a new system during the early Holocene remains unproven due to a lack of contemporary data from the New Guinea lowlands (Golson 2007). The Kuk sequence has been divided into several phases, the first two of which have been dated to the Pleistocene and early Holocene.
Phase 1 (10,200-9910 BP) represents the earliest evidence of plant exploitation within the upper Wahgi Valley at the wetland margin (Denham 2007:79). A palaeosurface consisting of pits, runnels, stake and post holes arguably representing wetland management for cultivation of corms and tubers (Denham 2007; Denham et al. 2003, 2004; Hope and Golson 1995) is contemporaneous with stone flakes and grinding stones that contain starch residues of taro (Colocasia esculenta) and a variety of yam (Dioscorea sp.) (Fullagar et al. 2006). Additionally, seed phytoliths of the Eumusa section of bananas, particularly Musa acuminata morphotypes, were extracted from sediments below Phase 1, suggesting the presence of this wild variety of Musaceae within the landscape prior to human occupation (Denham 2007; Denham et al. 2003;
22 Wilson 1985). Phase 2 (6950-6440 BP) has been argued to represent the first unequivocal evidence for deliberate planting at Kuk Swamp. This comprised of multi-cropping mounded cultivation of Colocasia taro and Dioscorea sp., most likely Dioscorea alata or Dioscorea pentaphylla, with Musa spp. and was confirmed through further residue analysis of artefacts combined with locally elevated frequencies of Musa phytoliths in sediments (Denham 2007; Denham et al. 2003; Fullagar et al. 2006). In total, over 30 edible plants were identified from archaeobotanical and palaeoecological investigations in the upper Wahgi Valley. The early to mid Holocene data collected from excavations at Kuk Swamp have since been interpreted to indicate that Musa acuminata sp. banksii was first domesticated or deliberately cultivated in New Guinea, along with the independent domestication of Colocasia esculenta and Dioscorea alata (Denham 2007; Denham et al. 2003, 2004; Donohue and Denham 2009; Golson 1989, 2007).
Other early archaeobotanical data from Island Southeast Asia, New Guinea and Near Oceania appears to support these arguments for pre-agricultural transport of species, or alternatively that the natural dispersal of the immediate ancestors of crops such as various aroids or Canarium predated human settlement of the region (Yen 1993). Haberle (1994, 1995) identified Colocasia pollen from a sediment core in Lake Wanum in lowland New Guinea at around 9000 BP, but argues that this only identifies the presence of this species within the landscape rather than indicating gardening activity. Similarly, Colocasia and possibly Alocasia
taro has been identified within starch, calcium oxylate crystal and cellulosic tissue in residues on stone tools from Kilu Cave in the Solomon Islands from 28,000 BP (Loy et al. 1992), indicating exploitation of these taxa in the Pleistocene. Loy and others (1992:910) argue that the implications of this and other data suggest that the northern Solomons and Australia should be included within the natural distribution of these aroids. Further west in Malaysia, research at Niah Cave has demonstrated that Dioscorea alata and cf. Dioscorea hispida were exploited around c. 40,000 BP through fragments of charred parenchyma and starch granules (Barton and Paz 2007:60-62). Fragments of charred rhizomes of aroids were also found in sediments dated to 23,850-23,020 cal BP, along with starch deriving from the Alocasia (Longiloba complex) or
Cyrtosperma merkusii that could also be associated with even older deposits. It is argued that Niah Cave deposits represent a relatively broad spectrum subsistence base that included some toxic plants such as taro which required higher energetic costs in food processing, but proving that people were deliberately manipulating the distribution of favourable plants such as these is another matter (2007:72). It is still therefore assumed that the archaeobotanical record of Niah represents evidence for rainforest foraging of naturally distributed plant taxa. Looking beyond these starch-rich taxa, a large number of other fruit and nut species including candlenut (Aleurites sp.), coconut (Cocos nucifera), Pacific almond or galip nut (Canarium spp.), island
23 lychee (Pometia pinnata), Pandanus spp., and Pangium sp. have been identified from deposits within the Sepik-Ramu region of New Guinea at around 5500BP (Swadling et al. 1991).
Lapita colonisers: Innovation and integration
Thus far, this review has been concerned with late-Pleistocene to mid-Holocene plant exploitation in the Pacific, culminating in the earliest evidence for agricultural and arboricultural systems which likely integrated endemic cultigens such as bananas, aroids and yams. During the latter half of the Holocene, it is argued that a new intrusive culture, linguistically and genetically distinct from the indigenous people of Near Oceania, arrived in the West Pacific. These people belonged to the Austronesian-speaking Lapita culture, and were the first to colonise Remote Oceania, bringing with them a number of cultigens that have been labelled by some as the ‘transported landscape’ (Kirch 1984). Whether the origins of the Lapita culture can be traced within Island Southeast Asia, or arose after occupation and interaction within Papua New Guinea is still under debate. The archaeobotanical record indicates that, within Near Oceania at least, the Lapita culture integrated some of these taxa that had already been cultivated in New Guinea into broad spectrum subsistence systems (Gosden 1992; Kirch 1987, 1988, 1989; Matthews and Gosden 1997). These cultigens such as aroids and bananas were then subsequently incorporated these into the suite of crops brought to Remote Oceania after 3000 BP.
Within Papua New Guinea, a number of Lapita sites contain preserved micro- and macrobotanical remains of indigenous and introduced taxa. Two sites are water-logged and anaerobic conditions have preserved macrobotanical remains of fruits and nuts from a range of tree crops. Remains from a mid- to late-Holocene site on Arawe Island are identified as deriving from eight genera and six species including Aleurites sp., Cocos sp., Canarium sp., Cordia sp., Cycas circinalis, Dracontomelon sp., Pandanus spp., and Terminalia sp.(Mathews and Gosden 1997:124). Matthews and Gosden (1997) concede that the botanical remains could be the result of natural beach drift, but point out that the best evidence for human involvement is modification through extraction processes. Extraction can be seen from the fragmentation patterns of the Canarium sp.and Terminalia sp., and charring of the Cycas remains(1997:128). Clearly Canarium continued to play an important role as a food source throughout the Holocene, as further evidence of the exploitation of this genera was found at another water- logged site on Mussau Island (Kirch 1987, 1988, 1989). Over 5000 anaerobically preserved seeds and seed cases representing at last 19 taxa, including Canarium sp., were recovered from Mussau (Kirch 1988:337). These included coconut (Cocos nucifera), Tahitian chestnut (Inocarpus fagifer), Corynocarpus caribbeanus, Dracontomelon dao, vi apple (Spondias dulcis), Pometia pinnata, Pangium edule, and tropical almond (Terminalia catappa).
Additionally, a number of Pandanus spp.fruit segments were recovered from contemporaneous deposits and shell peelers and scrapers that were argued to indicate the preparation of starchy
24 tubers and corms such as those of Dioscorea and Colocasia (Kirch 1987:177, Kirch 1988:338). Kirch (1987, 1988, 1989) argued that this data represented the first direct evidence that Lapita communities cultivated a wide variety of tree crops that are still used today in Melanesia, indicating that Lapita culture possessed a full component of arboricultural species at 3200-2800 BP.
Gosden (1992:63) argues that the nature of almost identical plant remains from sites in the Arawe and Mussau Islands indicate that these archaeobotanical records are not representative of formative stages of agricultural development during the Lapita period, but instead suggest well-developed arboricultural systems. Therefore, the transition to vegeculture through integration of local and introduced root crop taxa within subsistence still needs to be pinpointed. More recently, Lentfer and Green (2004) analysed the microbotanical record at the Reber-Rakival Lapita site on Watom Island in PNG, with a focus on the phytolith assemblages from three contexts. The presence of both introduced Eumusa and native Australimusa
morphotypes indicate that the Austronesians could have brought bananas with them, and possibly at this same stage used the newly encountered cultivars in conjunction with their own varieties, merging these two streams of banana domestication (Lentfer and Green 2004:85).
As Lapita populations expanded into Remote Oceania, further archaeobotanical evidence indicates the continued integration of roots, tubers and tree crops into colonizing subsistence regimes. In Vanuatu, microfossil evidence from the Lapita-associated sites of Teouma, Vao and Urupiv point to the cultivation of aroids such as Cyrtosperma merkusii, a range of yams including D. esculenta, D. nummularia and D. pentaphylla, and bananasby 3000 BP (Horrocks and Bedford 2004, 2010; Horrocks et al. 2009, 2014). It has been argued by numerous researchers that the brackish conditions in beach back-swamps near many early Lapita settlements would have enabled the initial cultivation of saline-resistant crops such as
Cyrtosperma prior to the establishment of more labour-intensive irrigation required for other aroids such as Colocasia or Alocasia (Kirch and Lepofsky 1993; Kirch and Yen 1982; Yen 1973a, 1993). Contained within this argument is the assumption that Lapita populations did not have intensive cultivation techniques, beyond an understanding that these aroids grow best in wet conditions. The late materialization of irrigation within island sequences can be seen to suggest that this technology was not transferred, but rather reinvented over generations or millennia (Kirch and Yen 1982:267). Others such as Spriggs (1990, 2002) instead suggest that the similarities between pond-field, island bed and taro pit wetland cultivation techniques across the Pacific indicate that these were not independent innovations, and therefore derived from prior knowledge and experience within founding populations. Moving east to Fiji, microbotanical data derived from Bourewa on Viti Levu (Horrocks and Nunn 2007) contained evidence for the starch, calcium oxylate crystals and xylem vessels of taro (Colocasia esculenta) and the lesser yam (Dioscorea esculenta) at 3000-2500BP. These identifications would
25 therefore appear to support Spriggs’ (1990, 2002) argument but additional work is required to further establish this link.
Micronesia: A different branch
Archaeobotanical research in Micronesia demonstrates the transfer and utilisation of crops east of Island Southeast Asia within the late Holocene. On Kosrae, Athens and others (1996) have collected both macrobotanical and microfossil data upon the antiquity of plant use within archaeological and palaeoecological sequences. Wood charcoal of breadfruit (Artocarpus spp.),
Thespesia and Cordyline, along with the charred endocarp and seeds of coconut (Cocos nucifera), Morinda citrifolia, Pandanus, Terminalia and Inocarpus, and storage parenchyma of
Alocasia taro and cf. Dioscoreaceae were recovered from excavations at Katem (1996:843). The pollen record from Tafunsak contained abundant giant swamp taro (Cyrtosperma merkusii) pollen coinciding with the earliest period of occupation at 1997-1709 BP, while breadfruit pollen was only present in the latest interval at 1264-1150 BP (1996:843; Pickersgill 2004). This information combined with the micro-charcoal record from the same core indicated that a lowland agroforest was established on the island at a very early date, and further burning was not needed as this would only damage existing crop trees (Athens et al. 1996:843).
Further south, Di Piazza (1998) identified evidence for introduced taxa on Kiribati through analysis of wood charcoal and endocarp within an earth oven dated to around 1430 to 1645 AD. A number of species were interpreted as fuel sources including Cocos nucifera, Cordia subcordata, Pemphis acidula, Guettarda speciosa, and Morinda citrifolia, as well as edible remains such as Pandanus tectorius that had cooked in the oven. The pounding and eating of the drupes of Pandanus have been recorded ethnographically (Di Piazza 1998), as well as the utilization of the uneaten portions as the preferred fuel to ensure fires last long enough to cook pigs or Cyrtosperma. Ethnographic and historic accounts highlight the importance of the giant swamp taro within subsistence throughout Micronesia, as this saline-resistant crop can be grown relatively easily within atoll environments. Although it has recently been discovered that
Cyrtosperma is actually endemic to Micronesia (Athens and Stevenson 2001), pollen of this species has also been recorded at 4500 BP within the Ngerchau core in Palau (Athens and Ward 2001) but there is no supporting archaeological evidence for occupation.
Island Melanesia and Polynesia: Evidence for extensive and intensive cultivation
It is often assumed that early human arrival and colonisation in Remote Oceania introduced a number of core cultigens within the taro-yam complex, and arboricultural taxa into a large variety of island environments. As the socio-politics and material culture of these colonizing populations evolved into what is sometimes termed Ancestral Polynesian Society (APS) within Western Polynesia and then expanded east to the Society Islands, Tahiti, Marquesas, Easter Island, Hawaii and New Zealand, plant cultivation also evolved to suit local growing conditions.
26 The difficulty of transporting crops by canoe over large stretches of ocean and plant translocation led to the exclusion of some species from localised subsistence strategies. Post- Lapita archaeobotanical evidence from islands within Remote Oceania has elucidated the timing of these introductions, as well as the contexts within which these taxa were cultivated or exploited.
This section of the review begins with an assemblage of microbotanical remains that are not associated with Ancestral Polynesian Society, but instead with Podtanean post-Lapita deposits in New Caledonia that are synchronous with this cultural transition in Polynesia. Horrocks, Grant-Mackie and Matissoo-Smith (2008) identified cultigens including C. esculenta, D. esculenta and one or more undifferentiated species of Dioscorea from identification of starch granules and calcium oxylate crystals within archaeological deposits dated to 2700-1800 BP in Me` Aure` Cave. The presence of these taxa was interpreted to fill a crucial gap in establishing “a continuous spatio-temporal record of ceramic-age agriculture across the western Pacific (2008:179). Further to this, Dotte-Sarout (2010; Dotte-Sarout et al. 2013) investigated whether arboriculture could be identified from the anthracological record at precolonial Kanak sites in New Caledonia, dated after 1300 AD. Results indicated a continuance of a forest gardening tradition focusing on Ficus sp., Syzygium malaccense, Aleurites moluccana, Hibiscus tiliaceus, Cordia subcordata, Calophyllum inophyllum, Artocarpus altilis, Cocos nucifera, Terminalia catappa among others, originating from the Austronesian/Indo-Pacific sphere of interaction (2013:133).
The Polynesian triangle is bound by the island groups of Hawaii in the north, New Zealand in the south, and Rapanui to the east. This boundary defines the culturally and genetically distinct Polynesians from the Melanesians of Fiji, Vanuatu, New Caledonia and islands further west. The botanical record of Hawaii is important for investigating the expansion of cultivars to the northern extreme of Polynesia. Along with Tonga, the Hawaiian archipelago represents one of only a small number of cases where dryland agriculture was adopted (Yen 1982). It is often presumed that where possible, irrigation for taro will be developed and that dryland techniques are a secondary choice (Spriggs 1982; Yen 1982). Both macro- and microbotanical remains demonstrating the nature and extent of dryland crop utilisation have been recovered from a temporally and geographically diverse range of archaeological sites. Douglas Yen was the first to apply modern identification techniques to assemblages of plant remains from dry rockshelters and open sites on Kaho’lawe Island, Molokai, Oahu and Hawaii Island (Allen 1984). From these sites, a number of primary (Dioscorea alata, Ipomoea batatas,