Ten Thousand Years of Cultivation at Kuk Swamp in the Highlands of Papua New Guinea
10
Introduction
The study of plants in archaeology—archaeobotany—is key to discovering how and when people exploited, cultivated and domesticated plants in the past, influenced their dispersal and effected their present-day biogeographic distributions. Archaeobotanical study incorporates a complex of methodologies, often reliant on carefully planned and executed sampling strategies and dependent on good preservation of various plant remains (Pearsall 2000).
Traditionally, the method for studying plant remains in archaeological deposits has been the analysis of macro remains of hardy material such as seeds, wood and nutshell (Textbox 10.1). These can provide direct evidence for human/plant association. However, they are not always available for study, being best preserved in extremely dry and cold environments, as well as in anaerobic conditions such as waterlogged deposits. Generally, macrobotanical remains are poorly preserved in well drained, acidic environments, particularly in the wet tropics, unless they have been burnt and converted to charcoal; even after burning it is usually only the hardy types of material that are preserved. Consequently, macrobotanical remains are limited in what they can tell us about the finer details of changing environments, plant distributions, manipulation and human exploitation because preservation in the wet tropics, when it does occur, is inconsistent, favouring some plants and/or plant parts over others.
It is often the case, therefore, that other analytical techniques are required to complement and enhance the analysis of macroremains or fill the gap in instances where they are not preserved. As well as firmly established pollen/spore and microcharcoal analyses, a host of microscopic techniques have been developed and applied to tropical archaeobotany over the last three decades of the 20th century (see Hather 1994), involving plant fibres, parenchyma and other plant tissues, as well as phytolith, starch and raphide analyses. These techniques enable the identification of a broader array of soft tissues than is usually preserved at archaeological sites, including the microscopic remnants of fruits, roots, rhizomes and tubers. Besides microscopic analysis, a number of chemical, spectrographic and other biomolecular techniques have potential to identify plant residues like resins, proteins, enzymes, lipids and fatty acids preserved on stone tools, pottery and other artefactual material (see Denham, Atchison et al. 2009 for a review). Additionally, techniques have been developed for ancient DNA (aDNA) analysis of plant remains and are presently being developed for identifying plant residues in sediments and on artefacts (Schlumbaum, Tensen and Jaenicke-Despres 2008).
Table 10.1 Archaeobotanical table listing potential food plants documented from Pleistocene and Holocene contexts at Kuk. Only potential food plants are considered, although other uses may have been as important in the past (cf. Powell 1976a).
|
Species/Genus1 |
Exploited Form2 |
Edible Part(s)3 |
Archaeobotanical Evidence4 |
Earliest Record5 |
|
Abelmoschus sp.6 |
c |
l, sh |
s |
pre-P1 |
|
Acalypha sp. |
w, t |
l |
s, p |
pre-P1 |
|
Castanopsis sp. |
w, t |
n |
w, p? |
pre-P1 |
|
Cerastium sp. |
w |
p |
s |
pre-P1 |
|
Coleus sp. |
w |
l |
s |
pre-P1 |
|
Ficus cf. copiosa6 |
c, w |
f, l |
s |
pre-P1 |
|
Ficus spp. |
c, w |
l, f |
w |
pre-P1 |
|
Garcinia sp. |
w |
f, l, b |
w |
pre-P1 |
|
Hydrocotyle sp. |
w |
l? |
s |
pre-P1 |
|
Lycopodium spp. |
w |
sh |
p |
pre-P1 |
|
Maesa sp. |
w |
f |
s, w |
pre-P1 |
|
Musaceae |
c, w |
f, c |
ph, st |
pre-P1 |
|
Oenanthe javanica |
c, w |
l, sh |
p |
pre-P1 |
|
Pandanus cf. antaresensis |
w |
d |
p |
pre-P1 |
|
Pandanus cf. brosimos |
c, w |
d |
p |
pre-P1 |
|
Pandanus spp. |
c, w |
d |
s, p |
pre-P1 |
|
Parsonsia sp. |
w |
n |
p |
pre-P1 |
|
Phragmites karka |
w |
l, r, sh |
ph |
pre-P1 |
|
Pouzolzia hirta |
w |
l, st |
s |
pre-P1 |
|
Rubus moluccanus |
w |
f |
s |
pre-P1 |
|
Rubus rosifolius |
w |
f |
s |
pre-P1 |
|
cf. Setaria palmifolia |
c, w |
s |
ph |
pre-P1 |
|
Solanum nigrum |
c, w |
l, sh |
s |
pre-P1 |
|
Syzygium sp. |
w |
f |
w |
pre-P1 |
|
Wahlenbergia sp. |
w |
p |
s, p |
pre-P1 |
|
cf. Zingiberaceae |
c, w |
r, l, sh |
ph |
pre-P1 |
|
Colocasia esculenta |
c |
c, l |
st |
P1 |
|
Dioscorea sp. |
c |
t |
st |
P1 |
|
Elaeocarpaceae |
w |
n |
p |
P1 |
|
Ipomoea sp. |
w |
sh |
p |
P1 |
|
Musa section bananas7 |
c, w |
f, c |
ph |
P1 |
|
Ingentimusa section7 bananas |
w |
f, c? |
ph |
P1 |
|
Phragmites sp. |
w |
st |
ph |
P1 |
|
Typha sp. |
w |
st |
p |
P1 |
|
Solanum sp. |
c, w |
f, l, sh, t |
s |
pre-P2 |
|
Commelina sp.6 |
c, w |
l, sh |
s |
P2 |
|
Drymaria cordata6 |
w |
l? |
s |
P2 |
|
Floscopa sp.6 |
c, w |
l, sh |
s |
P2 |
|
Viola arcuata |
w |
l |
s |
P2 |
|
Amaranthus sp. |
c |
l, p |
s |
P2/P3 |
|
Bidens pilosa |
w |
s |
p |
P2/P3 |
|
Polygonum chinense |
w |
l |
s |
P4/P6 |
|
Finschia sp. |
w |
s |
s |
P5 |
|
Saccharum officinarum |
c |
st |
pa |
P5/6 |
|
Bambusa sp. |
c, w |
sh |
w |
P6 |
|
Ipomoea batatas8 |
c |
t |
t |
P6 |
|
Elaeocarpus sp. |
w |
n |
w |
P6 |
|
prob. Ipomoea batatas |
c |
t |
pa |
P6 |
|
Polyscias sp. |
w |
l |
w |
P6 |
|
Oxalis corniculata |
w |
l, sh |
s |
post-P6 |
Source: Updated version of Denham et al. (2003: Table S3) and Denham, Haberle and Lentfer (2004: Table 2).
Notes:
1 List of edible species at Kuk based on ethnographically documented use of plants in New Guinea (Powell et al. 1975: 15–39; Powell 1976a: 108–112; Powell with Harrison 1982: 57–86; French 1986; M. Bourke 1989 and pers. comm., 2002; Haberle 1995; G. Hope, pers. comm., 2002). Edible species have been reported from other early and mid Holocene archaeological sites in the highlands.
2 Exploited form: c = cultivated, w = wild (used as supplementary food), t = transplanted.
3 Edible part(s): b = bark, c = corm, d = drupe, f = fruit, g = gourd, l = leaf, n = nut, p = plant, r = rhizome, s = seed, sh = shoot, st = stem, t = tuber.
4 Evidence: p = pollen (by Simon Haberle), pa = parenchyma (by Tara Lewis), ph = phytolith (by Carol Lentfer), s = seed (by Jocelyn Powell and Laurie Lucking), st = starch (by Richard Fullagar, Judith Field and Michael Therin), t = tuber (by Jon Hather), w = wood (by Jocelyn Powell and Laurie Lucking). Only those cases relevant for the earliest recorded occurrence are listed.
5 Earliest record are given for Pleistocene contexts (pre-P1), Phase 1 contexts (P1), contexts postdating Phase 1 and predating Phase 2 (pre-P2), Phase 2, 3, 4, 5 and 6 contexts (P2-6), and postdating Phase 6 (post-P6).
6 Identification to genus or species level should be considered provisional because it is based on a single seed.
7 A review of the sections within the genus Musa (Wong et al. 2002), which is not universally accepted (De Langhe et al. 2009), has reclassified the five former sections into three: section Musa contains the former Eumusa and Rhodochlamys; section Callimusa contains the former Callimusa and Australimusa; and section Ingentimusa remains the same. Previous identifications of Callimusa (formerly Australimusa) phytoliths from Phase 2 and 4 contexts at Kuk (Wilson 1985: Table 3) are excluded because they were based on a single phytolith each and no such phytolith morphotypes were documented from Phase 1–3 contexts at Kuk targeted during recent research (Denham et al. 2003: 192).
8 Provisionally identified as Dioscorea esculenta by Jon Hather (pers. comm. to Jack Golson), although reanalysis suggests probably Ipomoea batatas (Lewis, Denham and Golson 2016).
Interdisciplinary study at Kuk since the early 1970s has made several significant contributions to archaeobotanical research in Melanesia. A combination of techniques has obtained evidence for the presence of a suite of food plants at Kuk from the Pleistocene to the present (Table 10.1). The focus here is on food plants, although plants were used for a variety of different purposes including medicines, ornamentation, rituals, construction, tools and so on (Powell 1976a).
Archaeobotanical investigations were undertaken on samples collected from archaeological and stratigraphic contexts, with subsequent identification of micro- and macrobotanical remains to genus and species level, where possible. The results of macrobotanical work undertaken in the 1970s were complemented during the 1990s by employing microfossil techniques, primarily phytolith analysis of stratigraphic samples (Textbox 10.2) and starch granule analysis of tool residues (Textbox 10.3) to yield evidence of plants that do not regularly produce pollen or seeds under cultivation, specifically bananas (Musa spp.), taro (Colocasia esculenta) and yams (Dioscorea spp.). The 1970s investigations targeted all archaeological phases and major stratigraphic units at the site, whereas the 1990s investigations targeted early and mid Holocene contexts only.
|
Textbox 10.3 Tool usewear and residues Richard Fullagar The study of how ancient tools were used is as old as archaeology. Over the last 50 years, new techniques have been developed to identify diagnostic wear patterns and organic/inorganic traces that can be related to how stone artefacts were used in the past. The concept of function refers to a range of specific aspects of implement use that usually requires an understanding of design, motion or mode of use (e.g. sawing, scraping), holding, hafting, properties of worked materials (e.g. wood, bone, grass), stages of manufacture and edge reduction and the type of stone (such as obsidian, chert and basalt). The techniques of usewear and residue studies are thus grounded in material sciences like biology. Some wear patterns are highly distinctive, like the polished surface on stone sago pounders (Fig. 10.T3.1). However, experiments are vital to evaluate variables like edge angle and scarring type and to test interpretations of residue forms (cf. Barton 2007). |
|
Figure 10.T3.1 Above: Use-polish on an ethnographic sago pounder from PNG collected by Ross Bowden, La Trobe University, Melbourne (Fullagar 1989: Plate 10.10). Below: Sago starch grains from the same artefact. Metallographic microscope, scale divisions are 0.0068mm (Fullagar 1989: Plate 10.11). Source: Photographs by Richard Fullagar, reproduced with permission. |
|
Tool-use experiments are particularly important for usewear studies because different kinds of stone, like obsidian and basalt, will sustain usewear according to differences in hardness and mineralogical composition. Tool-use experiments are particularly important in the study of residues on tool edges because tiny traces, like feather particles or plant fibres, might not be part of processing or manufacturing particular things, but instead be related to hafting, decoration or storage (e.g. as in a knife sheath). It is vital to integrate the results of usewear and residue analyses with broader information regarding stone tool technology and other data from the archaeological site. Residue studies require specific comparative reference collections for identifying the diagnostic particle shapes of pollen grains, starch granules and phytoliths. Starch from stone tool surfaces has shown the presence and use of Colocasia taro and a yam in the highlands during the early Holocene (Figs 10.T3.2 and 10.T3.3 here; Denham et al. 2003; Fullagar et al. 2006). Earlier use of tubers in Island Melanesia is indicated by starch grains on stone artefacts dating to over 28,000 years ago from Buka (Loy, Spriggs and Wickler 1992). Microscopy and lighting conditions are critical for observing particular forms of usewear. In the usewear studies reported (see Chapter 20), metallographic microscopes, with vertical incident light, and stereomicroscopes, with a point source of oblique incident light, were used to observe stone tool surfaces and edges. The former provides high magnification (well over 200x) but has a flat image, with little depth of focus, which is useful for examining smooth or polished surfaces. The latter provides images of surface features like striations and scarred or beveled edges in clear relief. |
|
Figure 10.T3.2 Above: Retouched flake (K/76/29B) collected from beneath the grey clay fill of a basin-like Phase 1 feature (c. 10,000 cal. BP) at Kuk (Chapter 20). Scale bar is 10 mm. Below: Starch grains of taro (Colocasia esculenta) from the working edge of the retouched flake (left: transmitted light; right: cross-polarised light). Source: Photographs by Richard Fullagar. |
|
Residues are observed under reflected light microscopes, but tiny particles, organic structures and fibres are best viewed on mounted glass slides through transmitted light. Diagnostic shapes of phytoliths and starch grains found on artefacts and in sediments provide some of the earliest definitive evidence of plant food exploitation and cultivation of tropical crop plants, such as banana and yam (Lentfer, Therin and Torrence 2002; see Textbox 10.2). In Melanesia, microscopic studies have identified the use of flaked stone tools for a range of distinct tasks other than plant food processing. For example, woodworking in the Jimi River Valley is indicated by a distinctive polish on retouched blade-like flakes (Fullagar 1989), while the usewear on stone drill bits indicates the drilling of shell for making fishhooks at Motupore Island, near Port Moresby (Allen, Holdaway and Fullagar 1997). Evidence for butchering, bone working and skin working is less commonly found (but see Kononenko 2008). |
|
Figure 10.T3.3 Above: Grinding stone (pestle, K/75/S178) recovered from the Phase 1 palaeochannel base of channel 101 (Kundil’s Baret) dating to about 10,000 cal. BP. Below: Large starch grain of a yam (Dioscorea sp.) from the residue extraction (left: transmitted light; right: cross-polarised light). Source: Photographs by Richard Fullagar. |
Food plants are presumed to have been continuously present in the upper Wahgi region after their earliest record at Kuk. Table 10.2 lists archaeobotanical finds at other sites in the upper Wahgi: gourds at Warrawau (the Manton site), as well as the middle Wahgi site of Kana; and pandanus at four Manim Valley sites (Donoghue 1988). Of the three other sites listed Yuku (bamboo and, problematically, sugarcane) is just north of the upper Wahgi in the upper Yuat; Aibura (bamboo) is at the upper reaches of the Lamari River in Eastern Highlands Province near Kainantu; and the Ivane Valley (pandanus) is in the Papuan highlands to the southeast.
Table 10.2 Macrobotanical remains of food plants reported at other archaeological sites in the highlands.
|
Plant |
Site |
Approx. antiquity (years ago)4 |
Laboratory code |
References |
|
Bamboo |
||||
|
Bambusa sp. |
Yuku |
11,600–10,600 |
ANU-358 |
Bulmer 1975: 31 |
|
Bambusa sp. |
Aibura |
after 4450–3900 |
GaK-623 |
White 1972: 57, 61 |
|
Gourds |
||||
|
Lagenaria siceraria or Benincasa hispida |
Warrawau (Manton) |
5900–5050 |
ANU-288 and 2086 |
Powell 1970b: 144–145; Golson 2002: 70–71; Denham 2003b: 170; 2007a: 88 |
|
Benincasa hispida |
Kana |
2950–2000 |
ANU-9487 |
Matthews 2003; Muke and Mandui 2003: 184 |
|
Pandanus1,2 |
||||
|
Pandanus spp. |
Ivane Valley |
36,300–34,600 |
Wk-17261 |
Fairbairn et al. 2006: 379; Summerhayes et al. 2010: Table S1 |
|
Pandanus spp. (wild) |
Etpiti |
after 2350–2000 |
ANU-1323 |
Christensen 1975: Table 1; Donoghue 1988: 85–88 |
|
Kamapuk |
from 2750–2350 |
ANU-1326 |
Donoghue 1988: 78–80 |
|
|
Tugeri |
after 2700–2350 |
ANU-1321 |
Donoghue 1988: 88–90 |
|
|
Pandanus antaresensis |
Kamapuk |
after 2750–2350 |
ANU-1326 |
Donoghue 1988: 81 |
|
Manim 2 |
from before 11,800–10,300 to just after 7000–6350 |
ANU-1375 |
Christensen 1975: 31; Donoghue 1988: 71–73 |
|
|
ANU-1373 |
||||
|
Pandanus brosimos |
Tugeri |
after 1700–1400 |
ANU-1322 |
Donoghue 1988: 88–90 |
|
Pandanus iwen |
Etpiti |
after 2350–2000 |
ANU-1323 |
Donoghue 1988: 85 |
|
Kamapuk |
after 2750–2350 |
ANU-1326 |
Donoghue 1988: 78 |
|
|
Pandanus julianettii |
Etpiti |
after 1350–1050 |
ANU-1324 |
Donoghue 1988: 85 |
|
Kamapuk |
modern? |
stratigraphic inference |
Donoghue 1988: 81 |
|
|
Tugeri |
after 1700–1400 |
ANU-1322 |
Donoghue 1988: 88 |
|
|
Sugarcane |
||||
|
Saccharum officinarum3 |
Yuku |
5650–4550 |
GX-3111B |
Bulmer 1975: 31 |
Source: Data collated by Lentfer and Denham.
Notes:
1 Abundant archaeobotanical material besides pandanus was collected from the Manim Valley sites and is subject to ongoing investigation. Seeds collected as part of palaeoecological studies at wetlands in the highlands are excluded here because of their uncertain associations with humans.
2Pandanus conoideus, identified at Yuku and formerly referred to as Pandanus spp. (Bulmer 1975: 31), was subsequently claimed as approximately 14,000 years old (Bulmer 2005: 392–393). Recent redating by Denham (2016) of pandanus material recovered during the original Yuku excavations has yielded modern radiocarbon dates and it should be considered likely that all P. conoideus at Yuku is modern. Consequently, these finds are excluded from consideration here.
3 The identification of Saccharum officinarum at Yuku is uncertain (Yen 1998: 168) and given problems of dating at the site, both botanical identification and age are problematic (Denham 2016).
4 The radiocarbon calibrations are derived as follows: for Yuku and Aibura from Sutton et al. (2009: Table S1, supplementary information); for Warrawau inferred from chrono-stratigraphic interpretations of the site (references in table); for Kana and the Ivane Valley derived from the listed publications; and for the Manim Valley sites using Calib 6.0.0 (Stuiver and Reimer 1993) and IntCal04 calibration curve (Reimer et al. 2004). Only the highest probability distributions at two standard deviations are given. Calibrated date ranges below 10,000 years are rounded to the nearest 50 years and those above 10,000 years are rounded to the nearest 100 years.
Suite of techniques: Suite of food plants
Through the combined application of a suite of macrobotanical and microbotanical techniques, it was possible to identify in the archaeological and stratigraphic records a suite of plants that could have been used for food during various periods in the past (Table 10.1). The suite of food plants recovered from the archaeological record at Kuk, comprising starch-rich staples and supplements, vegetables and food-bearing trees, is similar to that harvested wild and cultivated in gardens across the Wahgi Valley today (Powell et al. 1975; Powell 1976a; Denham 2016). Many food plants were present from the Pleistocene, although several highly significant examples first appear in the early, mid and late Holocene. Analyses have provided evidence for the use and cultivation of some plants, but most techniques indicate no more than the presence of specific plants. The availability of a wide range of edible plants opens up the possibility for broad-spectrum subsistence systems persisting up to the early Holocene (Powell 1982a: 211).
Starch-rich staples and supplements
The most significant new archaeobotanical evidence is for the use of taro (Colocasia esculenta) and a yam (Dioscorea sp.) and the presence of Musa spp. from the early Holocene (Phase 1). These plants provide energy-rich sources of food and could have functioned as staples in the Wahgi Valley because they provide aseasonal, high-caloric resources in parts of New Guinea today (e.g. Powell 1976a; Gagné 1982a: 236; Dwyer and Minnegal 1991; Yen 1991a, 1995).
Starch granules of taro and yam were recovered from the used edges of stone tools dating from the early and mid Holocene (Fullagar et al. 2006, 2008; see Chapter 20 here). These finds, together with the appearance of other plant residues on stone tools, are indicative of plant processing, most probably prior to consumption. The presence of these two starch-rich plants is significant, because taro and some yam species are indigenous to and were domesticated in the New Guinea region (Lebot 1999; see Chapter 3 here). There is earlier evidence that starch-rich tuberous plants were being exploited in lowland sites of Island Melanesia (Loy, Spriggs and Wickler 1992; Barton and White 1993).
High frequencies of Musaceae phytoliths, including types of seed morphotypes specific to the Musa (formerly Eumusa) section, recur throughout the Holocene (Denham et al. 2003; Lentfer 2003a). Importantly, the high frequencies of banana phytoliths in some Phase 2 contexts (dating from 7000–6400 years ago) are suggestive of deliberate planting in multi-cropped plots within a grassed landscape managed by fire (Textbox 10.2). In New Guinea generally, bananas and possibly Ensete could have been exploited for their fruit, flowers, leaves and/or for their seeds and the edible corm at the base of the pseudostem during the early Holocene or earlier (Michael Bourke, pers. comm., 2002; Lentfer 2003b).
Banana, yam and taro starch sources could have been supplemented by edible pitpit (Setaria palmifolia), which may have been present in the vicinity during the Pleistocene (see Table 10.1), and potentially sugarcane (Saccharum officinarum), although mid Holocene claims for the latter at Yuku are unlikely (Bulmer 1975: 31; cf. Yen 1998: 168). Both these grasses have starch-rich pith. Saccharum officinarum could also have been a staple as reported ethnohistorically for the highlands (Daniels and Daniels 1993). Gingers (cf. Zingiberaceae) could have been exploited for their starch from the early Holocene, even though they are only minor contributors to diet in the highlands today.
In addition to these known starch sources, wild tubers are eaten in the highlands today, particularly during famine conditions (Watson 1964b; Waddell 1973; Powell et al. 1975: 15–32). Wild forms of Colocasia taro (swamp), Dioscorea yams (forest), Pueraria sp. (forest) and yam-like tubers (possibly Dioscorea but unknown species of the forest) grow in highland valleys today and are used for food (Michael Bourke, pers. comm., 2002; Powell et al. 1975: 21, 24–25, 36; Yen 1991b: 77, 81). Other wild varieties of the same species are known but never eaten or have poorly developed tubers or corms (Powell et al. 1975: 21, 24–25). Although many of these plants are generally associated with dryland forests, they may have grown, or been grown, at Kuk during drier periods and could have been significant sources of starch in the past. At present, the taxonomy, ecology, nutritional value and human exploitation of these wild plants are poorly understood and consequently they may have limited archaeobotanical visibility.
In conclusion, several starch-rich plants were potential food sources for people in the highlands at various times from the beginning of the Holocene. Most significantly, the remains of three starch-rich plants, Musa bananas, Colocasia taro and a Dioscorea yam, which have featured prominently in the development of Melanesian and Pacific agriculture, were first present at Kuk in the early Holocene and were associated with the earliest evidence of plant manipulation along the wetland margin.
Vegetables
At an early stage of the Kuk Project, Jocelyn Powell began a programme of macrofossil analysis, using seeds recovered by flotation from sediment samples taken from stratigraphic columns and ditch fills (Powell 1982b). The aim was to obtain more information on the presence of food plants and gardening systems at the site than was provided by pollen analysis. The field and laboratory work was conducted by Powell and Laurie Lucking in 1974–76.
The results showed that many indigenous vegetables were present at Kuk in the early to mid Holocene or earlier (Table 10.1). Many of these plants are highly nutritious, although not major sources of calories. Several are cultivated in gardens today, including Amaranthus sp., Commelina sp., Floscopa scandens, Oenanthe javanica and Solanum spp. (Powell 1970a: 200; 1981: 296; 1982b: 31–32; Powell et al. 1975: 15–32), whereas others are harvested from wild sources or as garden weeds, e.g. Cerastium sp., Drymaria cordata, Hydrocotyle javanica, Oxalis corniculata, Rubus rosifolius and Viola sp. Although it is possible that these resources were all natural colonisers of the wetland margin at Kuk and other wetlands in the upper Wahgi Valley (Powell 1970b), their distributions may have been enhanced by clearance using fire and by the cultivation of plots. The similarity of the suite of vegetables and starchy staples available in the past at Kuk to contemporary garden assemblages in the area suggest intentional behaviour, i.e. management or deliberate planting in gardens (after Powell 1982b: 31).
Nut- and fruit-bearing trees
Two of the most important nut-bearing trees in the highlands, Pandanus spp. and Castanopsis sp., were present in the Kuk catchment from the Pleistocene (Table 10.1). Wood from an additional nut-bearing tree, Elaeocarpus, was present in Phase 6 contexts at Kuk, and was probably there at an earlier date. Castanopsis sp. pollen decreased markedly in the Holocene from Pleistocene levels, largely as a result of forest disturbance (Denham, Haberle and Lentfer 2004: 848).
Pollen records from Kuk (Denham et al. 2003; Denham, Haberle and Lentfer 2004: 845–848) show that Pandanus spp. were present in low frequencies in the Pleistocene and at the beginning of the Holocene (Phase 1). During grey clay deposition (approximately between 10,000 and 7000 years ago) the frequency of pandanus pollen increased dramatically. Two kinds of wild nut Pandanus were present at Kuk in the Late Pleistocene, P. cf. antaresensis and P. cf. brosimos, belonging to two distinct groupings in terms of pollen morphology, the antaresensis-type being spiky (echinate) and the brosimos-type being smooth (psilate) (Haberle 1995: 200–206). Pandanus brosimos has a mean usual altitudinal range of 2400–3100 m and P. julianettii, referred to as the cultivated form of P. brosimos (cf. Haberle 1995: 207), has a mean altitudinal range of 1800–2600 m. Pandanus antaresensis has a mean altitudinal range of 1000–2350 m (Table 4.4).
Pandans of the brosimos/julianettii complex have been previously characterised as a highly seasonal resource. Indeed, some authors have claimed that human presence in the highlands during the Pleistocene was based on hunting and the seasonal harvesting of nuts (White, Crook and Ruxton 1970: 168–169; Bulmer 1977: 69; Golson 1991b: 86–88; Fairbairn, Hope and Summerhayes 2006: 379). However, seasonality and hence reliability of production in these species have been shown to vary geographically with climate across the highlands (Bourke et al. 2004: 39–40, 154, 191; cf. Donoghue 1988: 50). Consequently, people must have relied upon a much more broad-based diet in those parts of the highlands where nut pandanus produced aseasonally (Denham 2007c).
According to Hyndman (1984: 296), writing of the Wopkaimin of the Star Mountains, Pandanus antaresensis produces continuously throughout the year, which would compensate for the difficulty of getting at the nuts inside the hard fused drupes (Haberle 1995: 197). The species is unimportant as a food source throughout the highlands today (Haberle 1995: 197), although a range of other uses has been documented (Hyndman 1984: 296–298; Majnep and Bulmer 2007: 313–314, 330–332, 362). The abundance of carbonised drupe fragments of P. antaresensis in Late Pleistocene and early Holocene contexts at Manim 2, a rockshelter off the upper Wahgi Valley (Christensen 1975; Donoghue 1988), suggests that the species may have been a more important food source in the past. The period of relatively intensive exploitation of P. antaresensis at Manim 2, from before 11,800–10,300 BP to just after 7000–6350 BP, broadly corresponds to the deposition of grey clay at Kuk (see Fig. 6.10, note C and D) and a peak in the frequency of pandanus pollen (see Chapter 9, section ‘The early Holocene’ and Fig. 9.4). Haberle (pers. comm., 2008) states that the pandanus pollen at Kuk is overwhelmingly of antaresensis-type, which is also the case for a Syzygium swamp forest dating to about 8000 years ago at Kelela Swamp in the Baliem Valley (Haberle 1995: 207).
Pandanus of the brosimos/julianettii complex and Castanopsis are producers of abundant highly nutritious nuts that are still, or were until recently, targeted by highlands populations (Bourke 1996: 49–50 and Bulmer 1964, respectively). For P. antaresensis, which is not an important food source, there is no comparable information regarding productivity and food value (Robin Hide, pers. comm., 2008). Bulmer has remarked that the kernels of antaresensis nuts ‘require considerable effort to extract, for a rather small return’ (Majnep and Bulmer 2007: 314; cf. Hyndman 1984: 296). Several other fruit- and nut-bearing species were present at Kuk during the Pleistocene and Holocene (Table 10.1). Although of minor importance in contemporary diets, these fruit- and nut-bearing trees, together with edible forbs, ferns, shrubs, vines and fungi in the understorey, could have been highly significant contributors to diet in the past (after Powell 1982a: 210–211).
Archaeological associations
The presence in prehistoric contexts of plants that are today exploited for food does not mean they were necessarily used for food in the past. The presence of food plants in prehistoric contexts that are today associated with agriculture is not necessarily a sign of agriculture in the past. Even today, in simple numerical terms, most kinds of food plant are wild varieties (Powell 1976: Table 3.1). There is currently insufficient archaeobotanical resolution to determine whether most prehistoric plants were wild or domesticated varieties, although there is evidence for the use of plants that were subsequently domesticated. More detailed corroboratory information on the use of plants is needed from occupation sites in the highlands, such as the archaeobotanical collection from Manim 2 (Christensen 1975).
Most food plants in archaeobotanical assemblages at Kuk occur in low frequencies and are not correlated to archaeological phases. The low frequencies suggest low numbers, wild plants and possibly adventitious colonisation of disturbed habitats. However, given the phenology, reproductive habits, infrequent flowering at altitude and harvesting prior to flowering of these plants under cultivation, their limited archaeobotanical signal is not entirely surprising (after Powell 1970a: 199). Overall, food plants have similarly low frequencies in seed assemblages from samples collected in undoubtedly agricultural contexts, i.e. Phase 4, 5 and 6 ditch fills and stratigraphy, which also had better preservation and generally larger assemblages of most taxa. The lack of correlation between food plant frequencies and archaeological phases in general suggests that the archaeological evidence reflects only a part of broader, continuous land-management practices. The wetland archaeological evidence has been buried and preserved, whereas that on dryland slopes within the catchment has been eroded.
Importantly, high frequencies of Musa spp., Pandanus spp. and Zingiberaceae phytoliths in early to mid Holocene contexts are suggestive of deliberate management. The presence of Musa section phytoliths, in particular, is significant, because two species of that section, Musa acuminata and Musa balbisiana, are central to the early domestication of Musa bananas and the subsequent development of hybrids and sterile triploids (Yen 1998; De Langhe and de Maret 1999; Lebot 1999).
Previous to the Kuk studies in the 1990s, it was thought that bananas of Musa (formerly Eumusa) section, as distinct from those of the former Australimusa section, were an Austronesian introduction to the region (Simmonds 1962; cf. Wilson 1985; see Table 10.1 note 7). However, recent biomolecular and genetic studies have shown that the wild subspecies of M. acuminata indigenous to New Guinea, M. acuminata ssp. banksii, is an early and significant contributor to the A genome of Musa section banana cultivars (De Langhe and de Maret 1999; Lebot 1999; Careel et al. 2002; Perrier et al. 2009, 2011). The genetic and biogeographic evidence, together with the high numbers of diploid Musa spp. still cultivated there, indicate the New Guinea region as a centre for the early domestication of Musa section bananas (Lebot 1999; Perrier et al. 2009, 2011; see Chapter 3 here). The presence of Musa spp. at Kuk in the early Holocene is therefore not necessarily a product of human agency, although wild populations of M. acuminata ssp. banksii and its close relative, M. schizocarpa, presently occur at much lower altitudes. In view of the current biogeographical and archaeobotanical evidence, the presence of Musa acuminata morphotypes at Kuk suggests natural dispersal into the highlands and/or intentional translocation of bananas from the lowlands by humans following climatic amelioration at the beginning of the Holocene. Irrespective of how they got there, cultivation of bananas at about 7000–6400 years ago at Kuk, including those of Musa section, corroborates and provides a minimum time depth for the domestication of bananas in New Guinea.
Given the estimated timing of the introduction of Musa bananas (with New Guinea-type A genome) into Africa by 2500 years ago (Mindzie et al. 2001) and perhaps as early as 5200 years ago (Lejju, Robertshaw and Taylor 2006), it is likely that banana domestication processes were underway before they were introduced there. Such processes would have involved the deliberate selection of desirable traits, ultimately leading to the development of parthenocarpy, hybridisation and seed sterility as seen in the fully domesticated banana varieties grown throughout the world today. Domestication of starch-rich food crops, including bananas, may have been initiated in the New Guinea region during the early Holocene or even earlier. Colocasia taro and a Dioscorea yam were present in the highlands and were being exploited by at least 10,000 years ago (see Textbox 10.3). Like Musa bananas, both taro and yam are generally considered to be lowland plants (Yen 1995: 835) and introduced to the highlands. Their presence in the early Holocene at Kuk therefore gives additional support for human translocation, deliberate planting and incipient domestication of starch-rich food crops.
Conclusion
The archaeobotanical records at Kuk and other sites, taken together with the chronology of agriculture in the highlands, enable some general trends in plant exploitation to be elicited. Some minor starch-rich staples, most vegetables and some nut-bearing trees in Pleistocene contexts are still cultivated and gathered from wild resources in the Kuk vicinity today. The Pleistocene antiquity of these plants at Kuk suggests that they were exploited, gradually brought under cultivation and in some cases domesticated from wild sources in the highlands.
For some of the major starch staples, particularly Colocasia taro, Musa bananas and Dioscorea yams, the history of plant use may be more complex. The earliest records of these three plants at Kuk coincide during the early Holocene, following post–Last Glacial Maximum climatic amelioration. These three plant groups are thought by some to be native to the lowlands of New Guinea, to have been domesticated there first and to have been brought into the highlands by agricultural populations who moved upward as climates ameliorated at the beginning of the Holocene (Hope and Golson 1995; Yen 1995; Golson 2007: 115–117). The archaeobotanical evidence appears to support this scenario. Detailed microfossil research has not been undertaken on sufficient Pleistocene samples to determine if any of the plants in question occurred in the highlands before the Holocene, although Bourke (Table 4.1) would argue the opposite on grounds of their altitudinal limits at the present time.1 Even if these plants did diffuse to the highlands during the early Holocene, it is still not clear if people brought them as part of agricultural practices or as part of another form of plant exploitation (Denham and Barton 2006). Furthermore, it is quite possible that these three plants all dispersed naturally to the highlands from the lowlands as climates ameliorated.
In summary, from the early Holocene, most of the important plants of traditional highlands agriculture, including several starch-rich staples, were present in the upper Wahgi Valley. By taking the archaeobotanical finds in conjunction with other lines of evidence—primarily archaeological, palaeoecological and stratigraphic—specific landuses can be inferred during periods of wetland manipulation. During archaeological Phase 1 at Kuk, it is unclear whether people were engaged in agriculture (see Chapter 11, section ‘Agriculture at 10,000 years ago?’), although they were certainly practising a form of plant exploitation that included, among other things, digging, the likely tending of plants and the processing of tuberous plants. However, by at least 7000–6400 years ago (early Phase 2; see Chapter 12), people had begun to undertake mounded cultivation on the wetland margin, which is likely to have consisted of the intercropping of plants with different edaphic requirements. Subsequent periods of wetland cultivation (Phases 3–6; see Chapters 13–16) required the digging of ditches for drainage and to delimit plots, with cultivation on the intervening drained surfaces.
1 Horrocks, Bulmer and Gardner (2008: 300, cf. 293–294) say that starch evidence from Yuku rockshelter at 1280 m on the northern fall of the Sepik-Wahgi Divide suggests processing of the lesser yam, Dioscorea esculenta, possibly from before 14,000 years ago. The starch grain identifications were made on the basis of Horrocks, Grant-Mackie and Matisoo-Snith (2007). However, whatever the status of the identifications, the dating at Yuku is problematical and much of the stratigraphy is likely disturbed: see Table 10.1, notes 2 and 3 here; Denham (2016).
Previous Next


menu













