Prehistoric agricultural production on Easter Island (Rapa Nui), Chile.

STEVENSON, CHRISTOPHER M. ; WOZNIAK, JOAN ; HAOA, SONIA 等

Introduction

Traditional Pacific island societies are fundamentally agricultural populations and their production systems have been the focus of extensive ethnographic research (e.g. Barrau 1961; 1965; Brookfield 1972; Kirch 1990; 1994; Malinowski 1935) as well as at the central aspect of scores of archaeological investigations (e.g. Farrington 1985; Kirch 1994; Ladefoged 1990; Ladefoged et al. 1996; Spriggs 1986). The importance of understanding changes in agricultural production over time has been highlighted by the growing consensus that on islands, intensification of agricultural production is required to feed an expanding population and support social competition amongst the elite components of chiefdoms (Kirch 1994; Spriggs 1985; 1986). The form of these productive capabilities has been hypothesized to influence the impetus (or momentum) and level of social stratification in ranked society.

Until recently, information about pre-contact Easter Island (Rapa Nui) agricultural systems has been derived from European ship logs prior to AD 1900 (Beechey 1831; Forster 1777; Lisiansky 1814), early 20th-century descriptions (Routledge 1919; Metraux 1940), ethnobotanical studies (Yen 1988) and the archaeological investigations of a few stone garden features (Ferdon 1961; Skjolsvold 1961). Although numerous bibliographic references make casual observations about field systems, the abundance of crops and lack of irrigation, there is little known about the technology of production for the dryland agricultural system. We believe this is largely due to the difficulty of identifying prehistoric field systems on the contemporary landscape.

It is the goal of this paper to pull together the interim results of several research projects conducted by Wozniak (1996; 1998), Stevenson (1997), Stevenson & Haoa (1998) and Stevenson et al. (1996; 1997) and to present the first detailed, yet preliminary, model of Rapa Nui dryland agricultural production. To accomplish this we first ask the most fundamental of questions: How can agricultural gardens be identified? Second, we examine the role of dryland agriculture in the development of social complexity. Do the lithic mulch gardens on Rapa Nui represent innovations connected with the intensification of production? Lastly, is this a form of capital investment or a developmental feature of continued landscape use?

Hierarchical societies and dryland agriculture

The ethnographic present

In a discussion of contemporary agriculture on the islands of Futuna and Alofi in Western Polynesia, Kirch (1994) contrasts environmental effects, production costs and their social consequences for societies using either irrigation-based agriculture or dryland field systems that rely upon annual rainfall. Kirch's observations about the functioning of dryland agricultural systems form a temporal model that provides a general scenario, presuming a stable environment, for societies which have implemented dryland farming over long periods of time. Here, we summarize these points and present a set of expectations about the agro-political trajectory for Rapa Nui.

There are three fundamental aspects to dryland agriculture (Kirch 1994). First, upon entering a previously uncultivated environment early field systems will be land extensive in nature and the fallow period long in duration. However, with an expanding population the situation can change significantly. Using the supporting observations of Geertz (1963), Kirch (1994) indicates that as a result of over-population, dryland agricultural practices can degrade the landscape and reduce yields. A reduction of the fallow cycle and increased cultivation of the same area can result in significant soil erosion and exhaustion of soil nutrients.

Secondly, all agricultural practices face the requirement of generating a surplus to meet the demands of social production that arise from competitive feasting, alliance exchange and other corporate efforts of the lineage. These internal requirements put a demand on the subsistence base in addition to that connected with population growth. The need for increased production results in decreased fallow times, increased labour input, or territorial expansion. In Polynesia where land is scarce, societies may be aggressive and militaristic. In situations such as this highly differentiated and complex chiefdoms can emerge and are represented by societies in prehistoric Hawaii, Tonga and the Society Islands (Kirch 1994).

Lastly, it was proposed (Hommon 1986; cf. Kirch 1994) that dryland agricultural systems are susceptible to environmental stresses such as droughts. Irrigation systems can carefully manage the flow of water during periods of low rainfall. However, in areas where irrigation is not an option, sporadic rainfall can lead to unpredictable or reduced production that may in turn lead to famine and can trigger wars of expansion.

Expectations for Rapa Nui

In our study of Rapa Nui dryland agricultural economy we wish to examine and explain the relationship between agricultural production and cultural complexity and change. We are asking basic questions such as: What are the environmental constraints that shaped the strategies of agricultural production and what new technologies and management strategies were developed to meet the demands of population growth and social production? These are the historical particulars of the situation that will add detail to the more generalized model developed from the work of Kirch (1994).

We propose a five-phase temporal model that describes the trajectory of the cultural system from its colonization as possibly as early as AD 500 until the demise in the middle AD 1800S. We propose that prehistoric Rapa Nui society experienced five phases of agro-political development:

1 Land extensive, long-fallow field systems from the time of colonization until the population settled over the entire island.

2 Environmental degradation induced by initial clearing and subsequent shorter fallow periods in an attempt to feed a growing population.

3 The appearance of labour-intensive innovations that increase productivity and reduce risk.

4 Rapid and marked hierarchical development and a high level of social production.

5 Predatory expansion by island polities to obtain additional agricultural land.

In this short paper we will concentrate upon evaluating two aspects of the model. We present data to argue that Rapa Nui experienced a marked increase in social complexity beginning around AD 1400 and peaked in the late AD 1500s. We will also present the results of our field system investigations to argue for the development of elite managed social production and the creation of innovations that increased productivity and reduced risk.

Rapa Nui dryland agricultural field systems

We identify two agricultural phenomena that will assist in evaluation of the model. The first of these are labour intensive agricultural features which were designed with the idea of reducing agricultural risk. Here, we identify the attributes and possible function of gardens and fields in three parts of the island: Te Niu, La Perouse, and Maunga Tari (FIGURE 1). Secondly, we discuss associations of cultural features used in surplus production and correlate these with hierarchical development. The patterned distribution of households, religious structures and specific agricultural features are used to argue for the presence of social production on Rapa Nui.

[Figure 1 ILLUSTRATION OMITTED]

Agricultural fields and gardens

Eight types of agricultural features were present within the three areas based upon variations in the size and distribution of surface rock distributions and the topographic setting.

Pu refers to a depression in the ground, or in this case, a depression within a rock concentration. The rock concentrations were located at the base of steep slopes and were estimated to be 50 to 60 cm thick. Contained within the stone layer are multiple circular depressions with a diameter of approximately 50 to 60 cm that extend to the soil layer below. These depressions would have held a single plant.

Household gardens consist of moderate densities of surface rock distributed along the base of linear basalt outcrops. The gardens were associated with domestic features such as house pavements, caves, garden enclosures, and earth ovens at their margins. Smaller household gardens that were round or ovoid in shape are found in dissected terrain formed by basalt outcrops or in protected areas formed by ahu platforms.

Lowland fields are larger versions of household gardens but lacked the direct association of domestic houses. They consist of moderate densities of surface rock and were located within the coastal area.

Slope fields are found in the ravines and at the base of Maunga Puko Puhi (La Perouse area). These gardens were found in association with pu which were constructed at the edges of ravines. Much of the rock in these areas appeared to be larger in diameter compared to that of household gardens. It is proposed that slope gardens were established since the terrain acts as a buffer against the prevailing south wind.

Hilltop fields are located on the top of Maunga Puko Puhi (La Perouse area) and consisted of an extensive rock distribution estimated from air photographs to be 50 hectares in size. It was formed by a dense concentration of stone that completely obscured the surface of the ground. Small cleared areas are frequently present on low rises within the field and these areas occasionally contained evidence of habitation.

Manavai or garden enclosures are present throughout the lowland coastal area and consist of the above-ground free-standing stacked stone structure as well as the subterranean type (McCoy 1976). These features currently support taller plants such as bananas. However, it is hypothesized that these structures may have been used as propagation beds for young plants.

Boulder gardens are natural alignments of large boulders (larger than 2 by 2 m) that form a wind-break and serve as support for growing vines. A surface lithic cover was often found at the base of these boulders. Occasionally semi-circular walls of medium sized rocks were built against the boulders.

Planting circles are formed by rings of small stone 1.0-1.5 m in diameter and define the perimeter of a planting pit. The soils within the circle are deep (30-50 cm) and have been thoroughly mixed. Excavated circles have identified the basal portion of a single plant with a dendritic root system in the centre of the planting pit.

West Coast Rapa Nui: Te Niu

Te Niu, located on the northwestern flanks of Mt Terevaka (FIGURE 1), has a stepped terrain. Flat land surfaces alternate with rocky outcrops or sloping swales. The elevation rises over the distance of 1.5 km from 100 m at the coastal cliff where the ceremonial platform Ahu Te Niu is located, to 500 m at the top of Mt Terevaka. Rock overhangs, lava tube caves and large boulders reflect a volcanic landscape. Soil depths vary from shallow (30 to 70 cm) on the sloped areas to deep (over 2 m) on flat topography. No springs or streams are present within the Te Niu area. Contemporary vegetation cover consists of long and short grasses.

Surface archaeological features including 31 residential sites, in various stages of preservation, indicate settlement was more concentrated between the coastal cliff to 900 m inland (FIGURE 2). Using radiocarbon and obsidian hydration dating techniques from stratigraphic horizons within the gardens, a rock-shelter, and the ahu, Te Niu was occupied from c. AD 1200 to approximately AD 1850. No obsidian date later than AD 1831 [+ or -] 32 years has been established in the Te Niu area.

[Figure 2 ILLUSTRATION OMITTED]

Garden identification and settlement distribution

Agricultural features present in the Te Niu survey area (FIGURE 2) include: household gardens adjacent to residential features, manavai garden enclosures that are either subterranean or free-standing above-ground features incorporating a stone chicken house, boulder gardens along surface lava flows, and a rock-covered lowland field covering 10,000 sq. m on a gentle slope.

A systematic sub-surface survey made within the Te Niu area yielded soil samples for pollen and phytolith analysis and geomorphological studies, stratigraphic profiles and soil characteristics such as Munsell colour, structure and texture for each stratigraphic level of the 40 shovel test pits and 16 excavations within or near the ceremonial platform and household sites. Rock-covered areas and rock-free areas in the same geomorphologic settings were also excavated to determine if anthropogenic soils were present, and if so, their extent and content. It was in these last sets of excavations that it was realized that the rock cover served as a lithic mulch, which protected the underlying soils.

Residential areas are commonly located adjacent to areas of flat terrain having a rock cover. Under the 30- to 50-cm thick layer of rocks, generally 5 to 20 cm in diameter, is a deep layer consisting of dark anthropogenic soil (FIGURE 3: Layer A2 (Ap)). The soil under the rock cover consists of a homogeneous organic soil of darker brown colour and friable texture or pit features of such soil within a yellowish brown soil. These anthropogenic soils, especially those observed in the pit features, often contain fragments of charcoal, red scoria, basalt and obsidian flakes, and occasionally, pieces of coral or shellfish carapace.

[Figure 3 ILLUSTRATION OMITTED]

Several elliptical elite houses are found in the project area of Te Niu. One elite house is located near the coastal ahu, one is on a step of flat terrain encompassing more than 12,000 sq. m of potential garden space located 300 m inland of the coast, and a third elliptical house, a boat-shaped foundation without the typically carved paenga foundation stones, is located on another step 700 m from the coastal ahu. The remaining 28 residential features have no stone foundations; these represent lower-ranking residential houses. The lower-ranked house features are immediately adjacent to lithic mulched gardens while the elite houses are not. This suggests that the elites may have been managing rather than taking an active role in agricultural production in the Te Niu area.

North Coast Rapa Nui: La Perouse

La Perouse Bay is located on the north coast of Rapa Nui (FIGURE 1). The terrain from the coast to the base of the hill at the foot of Maunga Puko Puhi is nearly level. Puko Puhi is a flat-topped ridge with a 30 [degrees] side slope that is dissected by a number of wide swales separated by smooth and bare ridges. The top of the hill is uniformly fiat and vegetation cover is limited to a short and sparsely distributed grass.

Garden identification at La Perouse

Pedestrian survey within the La Perouse area resulted in the identification of six forms of agricultural features. These included pu, household gardens, lowland fields, slope fields, hilltop fields and manavai (FIGURE 4).

[Figure 4 ILLUSTRATION OMITTED]

The prehistoric settlement pattern at La Perouse

The prehistoric settlement pattern in the La Perouse area consists of two components: a domestic component of households and specialized activity areas, and a religious or ideological component made up of elite households, ceremonial platforms (ahu), stacked stone platforms, crematoria, petroglyphs and cairns (pipi horeko). These surface features are distributed amongst the gardens and assist in the identification of landscape use.

Small above-ground or below-ground garden enclosures and household features form clusters on the landscape (FIGURE 5). In La Perouse, the domestic settlement pattern extends from the shore to the base of Maunga Puko Puhi. The settlement concentration near the coast ([is less than] 400 m of the shoreline) and in the vicinity of Ahu Heki'i is thought to represent very late occupation which occurred after AD 1700 (Stevenson 1997). Recent dating of a household in this sector supported this hypothesis (Martinsson-Wallin & Wallin 1998).

[Figure 5 ILLUSTRATION OMITTED]

At a distance of about 400-1200 m from the ocean, a strip of dispersed households represents a stable and long-term use of the coastal plain. The households are adjacent to small gardens located at the base of rises or irregularities in the landscape. Excavations at a single household (31-98) revealed an occupational history ranging from approximately AD 1300 to 1850 (Stevenson et al. 1996). At a distance of 1200 m from the coast the settlement pattern changes abruptly. There are few signs of occupation on the slopes of Maunga Puko Puhi. On the top of the hill few lowland household types are represented. Instead, small earth ovens, the occasional house pavement stone and lithic debitage represent activity in cleared areas within the upland field system.

The religious elements of the settlement pattern are found in three areas (FIGURE 6). In the near coastal region the larger ahu are found. Radiocarbon dates from two of the larger ahu show a terminal 17th-century date for construction. Ahu Heki'i, the principal centre of the territory, is estimated to have been constructed around AD 1257-1387 (Ua1170, 2 sigma) and last expanded in AD 1500-1600 (Martinsson-Wallin 1998). Ahu Ra'ai, an ahu of lesser size located to the east, has also been dated to the same period as Ahu Heki'i (Martinsson-Wallin & Wallin 1998). Small gardens are present around these structures but they are presumed to have been late developments that used the ahu walls for protection against the wind once desanctification of the coastal area had occurred (Stevenson 1997). Other features in this region include smaller stacked stone platforms, cairns, isolated crematorium, stacked stone towers (tupa) and petroglyphs. The age of all these features remains undetermined.

[Figure 6 ILLUSTRATION OMITTED]

In the central portion of the study area is a cluster of well-preserved features: cairns, stacked stone platforms, and a group of cooking ovens (FIGURE 6), suggesting that they are late (post AD 1700) constructions. Located on the crest of Maunga Puko Puhi is a small ahu, measuring 30 m in length (FIGURE 6). The construction and form were consistent with other ahu dating to before AD 1650. Situated an additional 100 m to the south was a rebuilt elite dwelling (hare paenga) located on a small elevated clearing at the edge of the hilltop field system. An earth oven rake-out midden and a destroyed chicken house were associated with this elite dwelling. Excavations in the activity area adjacent to this structure produced an obsidian hydration date range of AD 1400 to the early AD 1700s at which time the field system was abandoned. Test pit excavations were also conducted within the field system and demonstrated the occurrence of a anthropogenic Ap horizon created through the long-term preparation of planting pits (Stevenson & Haoa 1998).

Central Rapa Nui: Maunga Tari

Maunga Tari is the name for a small cinder cone located in the centre of Rapa Nui (FIGURE 1). On the rim of the cone and on the surrounding slopes are a variety of domestic dwellings and specialized agricultural features. A portion of these sites were tested to determine the type of agricultural activity, the social context and duration of occupation (Stevenson 1997).

Garden features at Maunga Tari

At the time of field investigation the attributes of lithic mulched gardens and field systems such as those found in Te Niu and La Perouse had not been defined. These features were overlooked in the initial survey by the University of Chile and in our follow-up investigation in 1987-89. However, a reinspection of photographs from the time do show large lithic mulched gardens at the base of Maunga Tari.

Agricultural surface features at Maunga Tari were varied. They consisted of small, 1-m diameter, planting pits which occurred in clusters within a stone perimeter (10-242) or as isolated features. In addition, a water diversion channel (10-243) appears to have directed surface runoff in to field systems which we now know were in the area. Associated with these features were several habitation sites (10-240, 10-241, 10-244). Deep storage pits, hearths, numerous earth ovens lacking stone linings, and possible larger roasting-pit features were present at 10-241. Obsidian use-wear indicated that plant processing (e.g. sweet potato) was the dominant activity. Radiocarbon dates placed the use of the site in middle to late AD 1400s and the obsidian hydration dates extended the occupational range to the early AD 1200s. Such an association of features was interpreted to represent a site where shorter term task specific groups were engaged in the processing of tubers and other plants, their cooking and storage.

Settlement distribution at Maunga Tari

The complete excavation of a rectangular house and pavement (10-244) located on an artificial terrace just below the hilltop revealed the presence of elite persons in the area (Stevenson 1997). The structure was defined by a large fitted stone pavement that surrounded a house foundation of shaped basalt slabs (paenga) and a deep storage pit, the top of which was lined with a flaring rim of worked stone placed end to end. No cooking facilities were associated with the initial occupation which was dated from the mid AD 1200s to the late AD 1500s. The structure was disturbed during a later occupation.

The elite dwelling was associated with a small ahu (10-246) situated on the eastern edge of the cone (FIGURE 7). Structural remains indicate that a 5 m x 5 m, 40-cm high platform had been constructed of earth and surrounded by a low retaining wall of stone. A statue was inferred based upon the recovery of a red scoria top knot which had been completely buried at the margins of the structure. A small crematorium containing the remains of an adult ([is greater than] 20 years), a young adult (18-20 years) and a child ([is less than] 12 years) were present. Obsidian dates from the platform fill reveal an AD 1170 to AD 1580 age range which was equivalent to that of the elite house on the slopes below.

[Figure 7 ILLUSTRATION OMITTED]

Discussion

Easter Island is situated in the dry zone of the Pacific Ocean. It has a subtropical to temperate climate with a seasonal growing season. Rainfall averages 50 inches (120 cm), little of which falls during the actual growing period. Basalt rocks weather into phosphate-rich soil which provides quite adequate conditions for plant growth. The soils are excessively drained and have little moisture retention capacity (Wright & Diaz 1962), Moisture retention was the limiting factor faced by Rapanui agriculturalists.

In order to increase productivity in an area with limited rainfall and no means of irrigation, a maximal growing situation must be sought. Labour-intensive agricultural features as lithic mulch constitutes a means by which moisture is conserved, excess temperature fluctuations are ameliorated, and garden soils are protected from wind erosion (Maxwell & Anschuetz 1992; Lightfoot 1994; MacFadgen 1980). Lithic mulch allows for a rapid infiltration of rainwater from the surface and retention of this moisture. A cover of rocks that contain a sufficient pore size between the rock faces tend to retain water (Fairbourn 1973). Evaporation from the soil is intense during the day and surface soil temperatures may reach 120 [degrees] F at the surface (Wright & Diaz 1962). With lithic mulch, ground temperatures are stabilized in situations where shade cover is limited, allowing the temperatures to rise at night and to be maintained lower during the day.

The implementation of a moisture-retention strategy took a variety of forms. In both the Te Niu area and La Perouse, household gardens and lowland fields were covered with a mulch of small rocks. Manavai were constructed. In Te Niu, natural alignments of large boulders forming a protective microenvironment were employed. At La Perouse, garden forms included deep accumulations of stone at the base of slopes (pu) and cultivated areas in swales on the slopes of the Maunga Puko Puhi hillside. Most notable, however, was the extensive Maunga Puko Puhi hilltop field system so densely covered with lithic mulch that virtually no weeds or plants grow on it today.

Was the Rapanui strategy of applying lithic mulch a consequence of landscape formation processes or a later innovation associated with the intensification of the agricultural system? An obvious archaeological correlate of agricultural innovation would be the discovery of early gardens where lithic mulch was not in use. Unfortunately these contexts have not been documented to date. However, we do argue that the use of lithic mulch was a later-period innovation for two reasons.

First, the subsurface testing at Te Niu revealed that anthropogenic soils with associated pit features are only found in those landforms heavily covered by a layer of rocks. The rocks constitute a lithic volume in the top 20-cm level greater than, or equivalent to, the rock volume of the underlying soils which may range from 50 to over 150 cm in depth. An excavation within a household garden at Te Niu yielded an equivalent of 340 litres per cu. m of small rocks (less than 20 cm in diameter) from Level 1 (the top 20 cm). The underlying soil between 20 cm and 100 cm deep contained an overall 69 litres per cu. m of small rocks, with the rock/soil ratio decreasing with depth. Soils containing evidence of the original Rapa Nui forest lie below 100 cm. These deep forest soils contained few rocks in all cases where they were detected at Te Niu.

At La Perouse we are unsure as to the exact formation processes that created the dense lithic mulch on top of the Maunga Puko Puhi hill-top. The aerial photographs of the island show an amorphous-shaped lava flow that follows the contours of the terrain. Thus, the extensive accumulation of stone on the hilltop is likely to have formed as a result of natural weathering and decomposition of the basalt. The prehistoric farmers took advantage of this natural feature and may have added to the depth of the lithic mulch during the penetration of the subsoil while planting and bringing undecomposed basalt fragments to the surface. Regardless of exactly how the hilltop lithic mulch was created, the chronometric data shows the first use of this area did not begin until AD 1400. This late date of use suggests that lithic mulched contexts were not perceived as usable parts of the landscape until later in the prehistoric sequence.

In each of the case examples examined, we found evidence for the management of production by elites. In Te Niu elite dwellings are not only located close to the ceremonial structure (ahu) but also among the field systems and households yet spatially separated from these domestic features. At La Perouse the occurrence of an elite house at the edge of the large hilltop field system is more compelling. Here, the elite presence is emphasized by association of the house with a small ah u. This pattern is repeated in the central part of the island at Maunga Tari. Again, the settlement pattern consists of specialized agricultural features and short-term habitation/processing sites at the margins of lithic mulched field systems. On the highest part of the terrain, a small ahu with statue overlooks the gardens below. Close to this shrine is the house of a ranked individual that dates to same period. These repeated associations indicate that ranked persons were most likely ritually involved with agricultural production and used the position to manage the field systems under their authority.

The question remains as to whether the larger field systems were put into production to meet the annual obligations of feasting and exchange as well as the costs of ahu construction. Two lines of evidence suggest that surplus production was required. The first evidence comes from the documented construction episodes at some of the largest ahu located around the perimeter of the island. The construction of ahu is currently thought to have been initiated around AD 1000-1200 (Skjolsvold 1994; Martinsson-Wallin 1994; Stevenson 1997). The earlier structures would have been small and it is only later in the sequence do we see the physical movement of large statues, lithic materials and the investment in craftsmanship at a high level. For example, at Ahu Heki'i in La Perouse a large addition to the plaza and possibly the wings occurred in the 16th century. At other large and elaborate temples such as Vinapu 1, the best context at this finely crafted structure produced an age of AD 1310-1496 (T-5175). The third and most elaborate phase of Ahu Nau Nau at Anakena also produced a similar calibrated age range of AD 1305-1412 (Ua-617). Although not a principal centre like the above examples, the larger Ahu Akivi has been radiocarbon dated to the late AD 1400s or early AD 1500s (Mulloy & Figueroa 1978).

The second line of evidence comes from the dating of the field systems and elite dwellings themselves. As noted above, activity at the elite structure adjacent to the field system on top of Maunga Puko Puhi began in the early AD 1400s and ended in the early AD 1700s. At Maunga Tari the initial use of this area begins earlier and ranges from the AD 1200s to the end of the 16th century. Another important piece of data comes from the obsidian dating of numerous rectangular houses located further up the slopes of Mt Terevaka. Similar to the elite structure with a cut-stone foundation at Maunga Tari but lacking the large patio, these structures are scattered amongst the upland fields and fragile domestic habitations (e.g. circular house foundations of uncut stone, small house pavements without foundations, lithic scatters) and are interpreted as short-term residences for elite managers. It is interesting to note that the sites are most intensively used after AD 1425 and are abandoned in the early AD 1600s when the chiefdom collapsed (Stevenson 1997). It is unlikely that this expansion into new regions of the island is the result of a growing population that could not be sustained through household gardening, fishing and consumption of domesticated chickens. If such were the case we would see the expansion of households into these new regions and the construction of permanent architecture. This is clearly not the case.

In this first look at prehistoric agricultural production we have succeeded in defining the characteristics and distribution of gardens and field systems in three different parts of the island. Although the chronometric data are not voluminous and we do not have examples of early forms of Rapanui gardens, we propose that the introduction of lithic mulching was a technological innovation that was introduced possibly in the AD 1300s to increase the productivity of moisture-limited gardens. This innovation was incorporated into much larger elite-managed field systems which appeared early in the AD 1400s. This new production strategy is correlated with a sharp rise in the construction of religious structures located in the primary centres of island districts. This series of events lends support to Kirch's (1994) proposed model for the development of island chiefdoms dependent upon a dryland agricultural system.

Acknowledgements. The research at La Perouse and Maunga Tari was made possible by funding through the Earthwatch Institute and The Foundation for the Explorations and Research of Cultural Origins (FERCO). Research on Te Niu was carried out by Joan A. Wozniak of the University of Oregon, Eugene (OR), during the period of January to December, 1996, with funding from the Fulbright Commission for Study between United States and Chile, Santiago, Chile. We are very grateful to the Consejo de Monumentos Nacionales de Chile, Santiago, and the people of Rapa Nui for allowing us to conduct these research programmes. We would also like to thank William Ayres for his thoughtful review and suggestions for improving the draft manuscript. However, all errors or omissions are the responsibility of the authors.

References

BARRAU, J. 1961. Subsistence agriculture in Polynesia and Micronesia. Honolulu (HI): B.P. Bishop Museum Press. Bernice P. Bishop Museum Bulletin 223.

1965. L'humide et le sec: an essay on the ethnobiological adaptation to contrastive environments in the Indo-Pacific area, Journal of the Polynesian Society 74: 329-46.

BEECHEY, F.W. 1831. Narrative of a voyage to the Pacific and Beering's Strait ... in the years 1825, 1826, 1828. London: Henry Colburn & Richard Bentley.

BROOKFIELD, H.C. 1972. Intensification and disintensification in Pacific agriculture: a theoretical approach, Pacific Viewpoint 13: 30-48.

CRISTINO, C., P. VARGAS & R. IZUARIETA. 1981. Atlas Arqueologico de Isla de Pascua. Santiago: Facultad de Arquitectura y Urbanismo, Centro de Estudios, Isla de Pascua.

FAIRBOURN, M. 1973. Effect of gravel mulch on crop yields, Agronomy Journal 65: 925-8.

FARRINGTON, I.S. 1985. Prehistoric intensive agriculture in the Tropics. Oxford: British Archaeological Reports. International series 232.

FERDON, E.N., JR. 1961. Sites E-4 and E-5, in Heyerdahl & Ferdon (ed.): 305-11.

FORSTER, G. 1777. A voyage round the world, in His Britannic Majesty's Sloop, Resolution, commanded by Capt. James Cook, during the years 1772, 3, 4, and 5. London: B. White, J. Robson, P. Elmsley & G. Robinson.

GEERTZ, C. 1963. Agricultural involution: The processes of ecological change in Indonesia. Berkeley (CA): University of California Press.

HEYERDAHL, T. & E.N. FERDON (ed.). 1961. The archaeology of Easter Island, Reports of the Norwegian Archaeological Expedition to Easter Island and the East Pacific 1. Santa Fe (NM): University of New Mexico. Monographs of the School of American Research and the Museum of New Mexico, No. 24, Part 1.

HOMMON, R.J. 1986. Social evolution in ancient Hawai'i, in P.V. Kirch (ed.), Island societies: Archaeological approaches to evolution and transformation: 55-68. Cambridge: Cambridge University Press.

KIRCH, P.V. 1990. Production, intensification, and the early Hawaiian kingdom, in D.E. Yen & J.M.J. Mummery (ed.), Pacific production systems: 190-210. Canberra: Department of Prehistory, Research School of Pacific Studies, Australian National University.

1994. The wet and the dry: irrigation and agricultural intensification in Polynesia. Chicago (IL): University of Chicago Press.

LADEFOGED, T.N. 1990. Hawaiian dryland agricultural intensification and the Pacific economy, Pacific Studies 16: 119-31.

LADEFOGED, T.N., M.W. GRAVES & R.P. JENNINGS. 1996. Dryland agricultural expansion and intensification in Kohala, Hawai'i island, Antiquity 70: 861-80.

LIGHTFOOT, D.R. 1994. Morphology and ecology of lithic-mulch agriculture, Geographical Review 84: 172-85.

LISIANSKY, U. 1814. Voyage around the world, 1803-1806, in the ship `Neva'. London: Longmans.

MALINOWSKI, B. 1935. Coral gardens and their magic. London: George Allen & Unwin.

MARTINSSON-WALLIN, H. 1994. Ahu -- The ceremonial stone structures of Easter Island: analysis of variation and interpretation of meanings. Uppsala: Societas Archaeologica Upsaliensis. AUN 19.

1998. Excavations at Ahu Heki'i, La Perouse, Easter Island, in Stevenson, Lee & Morin (ed.): 171-7.

MARTINSSON-WALLIN, H. & P. WALLIN. 1998. Dating of ahu structures within the La Perouse area, Rapa Nui Journal 12: 85.

MAXWELL, T.D. & K.F. ANSCHUETZ. 1992. The Southwestern ethnographic record and prehistoric agricultural diversity, in T.W. Killon (ed.), Gardens of prehistory: 35-68. Tuscaloosa (AL): University of Alabama Press.

McCOY, P. 1976. Easter Island settlement patterns in the Late Prehistoric and Proto-historic periods. Washington (DC): Easter Island Committee, International Fund for Monuments, Inc. Bulletin 5.

MCFADGEN, B.G. 1980. Maori plaggen soils in New Zealand, their origins and properties, Journal of the Royal Society of New Zealand 10: 3-19.

METRAUX, A. 1940. Ethnology of Easter Island. Honolulu (HI): B.P. Bishop Museum.

MULLOY, W. & G. FIGUEROA. 1978. The A Kivi-Vai Teka Complex and its relation to Easter Island architectural prehistory. Manoa (HI): Social Sciences Research Institute, University of Hawaii at Manoa. Asian and Pacific Archaeology Series 8.

ROUTLEDGE, K. 1919. The mystery of Easter Island. London: Hazell, Watson & Viney.

SKJOLSVOLD, A. 1961. Site E-2, a circular stone dwelling, in Heyerdahl & Ferdon (ed.): 295-303.

(Ed.). 1994. Archaeological investigations at Anakena, Easter Island. Oslo: Esset Grafisk as. The Kon-Tiki Museum Occasional Papers 3.

SPRIGGS, M. 1985. Prehistoric man-induced landscape enhancement in the Pacific: examples and implications, in Farrington (ed.): 409-34.

1986. Landscape, land use and political transformation in southern Melanesia, in P.V. Kirch (ed.), Island societies: archaeological approaches to evolution and transformation: 10-19. Cambridge: Cambridge University Press.

STEVENSON, C. 1997. Archaeological investigations on Easter Island; Maunga Tari: An upland agricultural complex. Los Osos (CA): Bearsville and Cloud Mountain Presses.

STEVENSON, C. & S. HAOA. 1998. Prehistoric gardening systems and agricultural intensification in the La Perouse area of Easter Island, in Stevenson, Lee & Morin (ed.): 205-13.

STEVENSON, C., S. HAOA, P. MILLS, J.M. RAMIREZ & C. GOMEZ. 1996. The La Perouse Archaeological Project, Report of the 1995 Field Investigation. Report submitted to the Consejo de Monumentos Nacionales, Santiago, Chile.

STEVENSON, C., S. HAOA, P. WALLIN, H. MARTINSSON-WALLIN & J.M. RAMIREZ. 1997. The La Perouse Archaeological Project, Report of the 1996 Field Investigation. Report submitted to the Consejo de Monumentos Nacionales, Santiago, Chile.

STEVENSON, C., G. LEE & F. MORIN (ed.). 1998. Easter Island in Pacific context. Proceedings of the Fourth International Conference on Easter Island and East Polynesia. Los Osos (CA): Bearsville and Cloud Mountain Presses. Easter Island Occasional Paper 4.

WOZNIAK, J. 1996. Settlement patterns and subsistence on the northwest coast of Easter Island. Paper presented at the 1996 Easter Island conference, Hanga Roa, Easter Island, Chile.

1998. Settlement patterns and subsistence on the northwest coast of Rapa Nui, in Stevenson, Lee & Morin (ed.): 185-92.

WRIGHT, C.S. & C. DIAZ V. 1962. Soils and agricultural development of Easter Island, Chile. Quarterly Report Supplement No. 1 (FAO).

YEN, D.E. 1988. Easter Island agriculture in prehistory: the possibilities of reconstruction, in C. Cristino, P. Vargas, R. Izaurieta & R. Budd (ed.), First International Congress: Easter Island and East Polynesia: 59-81. Santiago: Universidad de Chile.

CHRISTOPHER M. STEVENSON, JOAN WOZNIAK & SONIA HAOA, Stevenson, ASC Group Inc., 4620 Indianola Avenue, Columbus OH 43214, USA. obsidlab@aol.com Wozniak, Department of Anthropology, University of Oregon, Eugene OR 97403, USA. joanwoz@yahoo.com Haoa, Hanga Roa, Easter Island, Chile. furceley@interactiva.cl

Received 23 December 1998, accepted 19 February 1999, revised 17 March 1999.