Printer Friendly

Development of an agroforest on a Micronesian high island: prehistoric Kosraean agriculture.

Prehistoric Micronesian agriculture

Prehistoric Micronesian agriculture has been seldom considered as a subject for investigation in its own right. Notwithstanding several discussions based on very limited data (e.g. Ayres & Haun 1990; Parker & King 1981; also recent review by Rainbird 1994), there has been no systematic treatment of prehistoric agriculture for any island. Rather, traditional agricultural practices as recorded ethnographically are generally assumed to have typified prehistoric practices, and differences between islands are regarded as having been present in the prehistoric past as well (see Hunter-Anderson 1991 for review of traditional Micronesian agriculture). This is hardly a satisfactory state of affairs. How do we know such continuity between the ethnographic present and the past is really true? There could have been major developmental changes in agriculture during the course of an island's prehistory as, for example, has been suggested by Parker & King (1981) for Chuuk. Did agriculture as we know it ethnographically just start from day one as each island was first colonized?

In the many archaeological reports concerning Micronesia during the past two decades, structural remains related to agriculture tend to be few or non-diagnostic, and there are few obvious remains of cultigens or diagnostic vegetal parts in the archaeological record. In this article we suggest a methodology for studying prehistoric Micronesian agriculture that involves pollen and charcoal sequences from both non-site and archaeological locations, and the identification of intensively collected archaeological charcoal. This has proven very effective in our archaeological research on Kosrae, a high volcanic island in the Eastern Caroline Islands of Micronesia.

The investigations were conducted during three field projects between 1982 and 1989. Most of the data derive from the Kosrae Wastewater Project, conducted over a two month period in 1989 with a 'landscape archaeology' focus (Athens 1995: 12-15). Fieldwork involved palaeo-environmental sediment coring in wetlands, excavations at off-site locations for geomorphological and palaeoenvironmental data and systematic recovery of carbonized plant remains from archaeological sites. For laboratory analysis, attention was given to radiocarbon dating, pollen analysis, charcoal/macrobotanical analysis and faunal analysis.


Kosrae is a small, tropical, lushly vegetated, high volcanic island in the Eastern Caroline Islands of Micronesia (Merlin et al. 1993). It is c. 550 km east-southeast of Pohnpei, the closest neighbouring high island, and 5 [degrees] latitude north of the equator [ILLUSTRATION FOR FIGURE 1 OMITTED]. The small coralline atolls of Pingelap and Mokil are situated between Kosrae and Pohnpei. East of Kosrae, the nearest atoll of the Marshall Islands, Ebon, is some 630 km distant.

Kosrae, with a land area of 109 sq. km (42 sq. miles), consists primarily of a rugged mountainous interior with a narrow coastal plain [ILLUSTRATION FOR FIGURE 2 OMITTED]. The mountains rise to a maximum 629 m (2064 ft) above sea level. Nearly 70% of the island has steep slopes and 15% of the land area is made up of foot slopes, alluvial fans and bottom lands, with most of the remaining 14% mangrove swamps (Whitesell et al. 1986).

Kosrae is tropical, with high rainfall (5000 mm annually with higher amounts in the island's interior) and high humidity all the year round. The island is at the edge of the tradewind belt to the east; it exerts only a slight effect on Kosrae's climate from February to April, when rainfall decreases somewhat and offshore winds become more noticeable. Kosrae rarely experiences typhoons, which tend to have their origins to the west off the coasts of Pohnpei and Chuuk and then move westward. At present approximately 63% of the island is forested, leaving 23% in agroforest (primarily breadfruit, coconut and banana), 11% in secondary vegetation and 3% non-forested (Whitesell et al. 1986). Along the eastern and southern coastline of the island a berm of coralline sand and beach deposits lies between the fringing reef and the mangroves that are attached to the main island mass. A radiocarbon determination indicates that this herm formed c. 2860[plus or minus]70 b.p. with the lowering of the high sea stand of the middle Holocene (Athens 1995: 239-51).(1) In many areas freshwater swamps are located inland of the mangroves. Shallow lagoons formerly existed where the mangroves are presently located. Most filled with organic sediments about 1500-900 years ago after the lowering of the sea to modern levels cut them off from fresh sea water (see Athens 1995: 239-51, 321-3, 360).

Besides the main volcanic land mass, there is a smaller 70-ha volcanic island, Lelu, adjacent to a natural deep-water harbour on Kosrae's east side [ILLUSTRATION FOR FIGURE 2 OMITTED]. The east side of Lelu consists of a large and steeply sloping volcanic hill (Finol Poro), 111 m high. The western half of Lelu, encompassing some 27 ha and entirely man-made, is the location of the impressive and well-known megalithic ruins of a late prehistoric chiefdom centre (Cordy 1985; 1993; Morgan 1988: 86-115). The reef flat here was purposefully filled with lagoonal sediments to raise the surface above the high-tide level for construction of residential, ritual and mortuary compounds for the paramount and high chiefs and their retainers. Though filling of the reef flat began about 750 years ago, megalithic construction did not start until c. 550 or 600 BP. Expansion of the Lelu compounds on the reef flat continued up to about 150 BP (Cordy 1985; 1993: 228-33; Athens 1995: 79-80).

Lelu Island, the prehistoric political centre of Kosrae, continued as the island's administrative centre after first European contact in 1824 and on through the whaling, German, Japanese, and American periods (see Cordy 1993). Its 2500 people compares to approximately 1200-1700 people estimated for the late prehistoric period (Cordy 1985: 256; 1993: 164-5). For all Kosrae the late prehistoric population is estimated to have been perhaps 3000-6000 people (Cordy 1985: 255-6; 1993: 11).

Traditional Kosraean society and agriculture

The earliest western visitors to Kosrae - the French and Russian exploration ships La Coquille and the Senyavin in 1824 and 1827, respectively (see Ritter & Ritter 1982) - described the island as politically unified; a population of no more than 3000 people was divided into four distinct social strata (note the population estimate by Cordy of 3000-6000; also Ritter 1981). At the top, the paramount chief was both the secular and sacred head of Kosraean society, and held ultimate title to all land. Under the paramount were about 10 high chiefs, usually male relatives that were appointed by him. The high chiefs, obliged to live in Lelu with the paramount, controlled named land units, facl, which ran from the high mountains to the reef. There were about 50 facl in Kosrae. The paramount also had his own facl. Below the high chiefs, 40 or 50 low chiefs were the land managers or overseers of the facl; they resided on the land for which they were responsible. Agricultural production was primarily the responsibility of commoners, the lowest social strata. Low chiefs saw to it that food, labour and tribute were provided to the paramount and high chiefs in Lelu.

Cordy (1985: 256-7) describes Lelu as

consist[ing] of about 100 walled compounds (dwellings, two royal burial compounds, and 17 sacred compounds) connected by sea piers, an internal canal system, and a paved network of streets. The dwelling compounds of the four social strata differed greatly. The ruler's and high chiefs' dwelling compounds were clustered in central Leluh and had massive basalt walls as high as 6 m. Within were multiple houses, with a large feast-house near the main entrance. Here the ruler entertained his visitors and retinue. Earthen ovens were located near this feast-house, and numerous food-pounding and seka (Piper methysticum)-pounding stones were present within the house.

In regard to traditional Kosraean subsistence, Rene Lesson (Ritter & Ritter 1982: 46), medical officer of the Coquille, observed:

The tree which furnishes the principal base of the existence of these islanders is the breadfruit tree [Artocarpus altilis]. It is so common that the fruit of this tree litter the ground.

A widespread zone of agroforest, with breadfruit the chief constituent, is suggested by the early descriptions. Other cultigens identified in the early historic literature included the giant swamp taro (Cyrtosperma chamissonis), taro (Colocasia esculenta), bananas, yams, coconuts and sugarcane (see Ritter & Ritter 1982; Hunter-Anderson 1991). Lesson also specifically mentions Arum macrorrhizon (Alocasia macrorrhiza), though his description suggests that he may not have distinguished it from Cyrtosperma chamissonis (see Ritter & Ritter 1982: 46; Athens 1995: 11-12). Friedrich Kittlitz, a naturalist on the Senyavin expedition, refers to 'both of these giant aroids' (Ritter & Ritter 1982: 177), verifying that both Cyrtosperma and Alocasia were present.

Dogs and pigs were absent from Kosrae at contact, and birds scarce except for pigeons and chickens.(2) Fyedor Lutke, Captain of the Senyavin, noted that 'fish and crayfish are the only animal food that they [Kosraeans] eat' (Ritter & Ritter 1982: 123).

Field investigations

Archaeological investigations prior to 1989 mostly concentrated on the functioning and chronology of chiefdom organization at Lelu (Cordy 1982; 1985; 1993; Morgan 1988) and settlement patterns around the island (especially Athens et al. 1983; Cordy et al. 1985; Welch et al. 1990; see review by Athens 1995: 20-35). The 1989 project, prompted by the construction of wastewater facilities, provided the opportunity to undertake archaeological studies with a landscape focus, complementing wetland coring undertaken for a road survey in 1988 (Welch et al. 1990). With respect to this latter project, the Okat-2 core is of special significance for the present discussion. Relevant excavations of the 1989 project include those at Katem and Finpea compounds in Lelu, the Malsu TP-1 and TP-2 test trenches near the north shore in Tafunsak, and the Mutunfunalu TP-3 test trench in Utwa. All water-screening was undertaken with a 1/8-inch screen.

Katem Compound, Lelu

At Katem compound, slightly inland of the south central shore of Lelu, a unit of 5 sq. m was excavated through compound fill and into the underlying reef deposit to a maximum depth of 2.55 m below a surface elevation of 2.92 m a.s.l. This general area, close to the original natural island, and probably formerly having a stream with a small coastal plain, would have been ideal for early pre-compound settlement (Cordy 1993: 229). With the water-table 26-69 cm below the surface, a 3-inch water pump was needed for excavation. The deepest sediments - designated Layer VI and clearly below the compound fill layer - contained pottery sherds, shell artefacts, some faunal bone remains and abundant charcoal and marine shell midden. The pottery, not previously documented on Kosrae, suggested a relation to early dates for pottery on the other central Micronesian islands of Pohnpei (Athens 1990a; 1990b) and Chuuk (Shutler 1984). All excavated sediments were bulk-bagged for water-screening. The residue, after drying, was then entirely sorted in the field laboratory, ensuring excellent recovery of remains. The Layer VI charcoal is summarized in FIGURE 3.

The lowest three levels (7, 8, and 9) in Layer VI produced closely matched charcoal dates c. 1900 BP (determinations in FIGURE 3; calibrations for all dates from CALIB 3.0.3 [Stuiver & Reimer 1993] with 1[Sigma] ranges). A level immediately above, having almost no pottery (the uppermost of Layer VI), produced a determination of 1516-1315 BP, which suggests pottery manufacture had essentially ceased by this time. The prehistoric compound fill layers - Layer IV and possibly V - are thought to date c. 600 BP [ILLUSTRATION FOR FIGURE 3 OMITTED]. This date apparently coincides with the formation of the Lelu chiefly centre and the initial use of megalithic architecture (see Cordy 1993: 260).

Finpea Compound, Lelu

Located near the south central shore, a single 1x2-m unit was excavated in Finpea compound to test for possible deeply buried cultural deposits and to obtain geomorphological information. The unit was placed near the base of the natural Lelu hill (Finol Poro) approximately 90 m from the present shore-line and at 2.82 m a.s.l. A depth of 232 cm below ground surface was reached, ending in culturally sterile marine sand and gravel below 199 cm. The water-table was 10 cm below the present land surface.

Between the surface and 132-141 cm, the sediments consisted of clay loam (Layer I), gleyed clay (Layer II) and reddish-brown clay loam mixed with irregular lenses of gleyed clay and becoming more gravelly with depth (Layer III). These sediments obviously derived from erosion of the adjacent hill slope. Artificial compound fill was absent, as were recognizable midden deposits. Modern trash, however, was present from the surface to a depth of 55 cm, with occasional trace amounts below. A high concentration of charcoal was present in upper Layer III at 55-77 cm [ILLUSTRATION FOR FIGURE 4 OMITTED], which may be late prehistoric. The sediments in Layer IV, 153-197 cm below surface, changed to a peaty loam, and the marine deposits of Layer V were below. Botanical remains - the fruit pods of the sedge, Scirpodendron ghaeri - from Layer IV yielded a radiocarbon determination of 1523-1350 BP. Although the peaty layer was not a midden deposit, three pottery sherds were present along with abundant dispersed charcoal [ILLUSTRATION FOR FIGURE 4 OMITTED], perhaps deposited as a result of erosional transport.

Malsu TP-1, Tafunsak

Situated in the Malsu land unit [ILLUSTRATION FOR FIGURE 2 OMITTED], this 1x2-m test excavation was placed on the gentle footslopes at the eastern margin of a small valley sharply defined by steep slopes on three sides and the coastal shore-line on its fourth and northern side (4.27 m a.s.l.). Its purpose was to provide geomorphological and palaeo-environmental information; no archaeological deposits were known or suspected.

Reaching a maximum depth of 2.40 m, the stratigraphy of the TP-1 excavation consisted of clayey sediments with weathered basalt gravel and rock increasing with depth. The water-table was 80 cm below surface. The lowest layer (Layer V), in addition to colluvial sediments, contained macrobotanical remains. Two bulk sediment samples (0.96 and 0.048 cu. m) yielded 8.5 and 13.2 g of charcoal for Layers IV and V, amounting to densities of 22.1 and 273.6 g/cu. m, respectively. The Layer V charcoal produced a radiocarbon determination of 1520-1350 BP.

Malsu TP-2, Tafunsak

This 1x2-m excavation unit was placed in the central part of the Malsu valley (3.6 m a.s.l.). The flat, marshy ground surface had a thick cover of tall grass (possibly California grass - Brachiaria mutica) and scattered banana, breadfruit, coconut and citrus trees. Like Malsu TP-1, the purpose of TP-2 was to provide geomorphological and palaeoenvironmental information; archaeological deposits were not known or suspected.

The TP-2 excavation reached a depth of 2.27 m; the water-table was at 50 cm. A metal probe determined the presence of marine deposits beginning at 3.19 m. Including the unseen marine deposits, there were five sedimentary layers: Layer I is a clay loam from 0-30 cm; Layer II is a sandier clay loam with irregular oxidation and gleying from 30-140 cm; Layer III is a gleyed silty clay from 140-170 cm; Layer IV is a peaty humic loam from 170 to an estimated 319 cm where the marine deposits begin. A bulk sediment sample of 0.145 cu. m was taken from Layer IV at 1.70 to 2.00 m below surface for water-screening. No archaeological remains were present, though 1.7 g of charcoal was found (density of 11.74 g/cu. m). Radiocarbon determinations of 1264-1150, 1997-1709 and 2427-2214 BP were obtained on sediment from intervals at 1-50, 1.70-2.00 and 2.00-2.27 m respectively.(3)

Mutunfunalu TP-3, Utwa

Mutunfunalu is a curving narrow island of land north of the footbridge across the Finkol River estuary, which separates it from the village of Utwa on the south central coast of Kosrae [ILLUSTRATION FOR FIGURE 2 OMITTED]. Extensive mangrove and freshwater swamps are present to the north and east, with mangrove swamps all around the island except near the estuary. TP-3, a 0.5x1.45-m controlled excavation (below 50 cm, narrowed to 50x60 cm), was situated near the northwestern extremity of Mutunfunalu, placed to bisect an archaeological stone alignment. Eight levels were excavated to a depth of 1.60 m, revealing 6 depositional strata: Layer I is a sandy loam; Layer II is a sandy loam with basalt sand and rounded gravel; Layer III is a sandy loam but with somewhat more clay than Layers I and II and not as much gravel as Layer II; Layer IV is a sandy loam with gravel and small pebbles (similar to Layer II); Layer V is a sandy loam that is much more organic than the upper layers (there was wood and leafy material in addition to a small amount of coarse rounded basalt gravel); Layer VI consists of fine to coarse rounded basalt sand and some gravel. The water-table was at 56 cm. Charcoal flecking was noted in all layers except Layer VI. Sediments from each excavation level were bulk-bagged and water-screened for laboratory sorting. FIGURE 5 illustrates the distribution of charcoal by level; other archaeological remains were virtually absent. There was also 25.8 g of uncharted wood in Layer V. Radiocarbon determinations of 295-modern and 1407-1292 BP were obtained; respectively from charcoal in a level just below the wall feature in Layer I, and wood from Layer V.

The sediments represent alluvial deposits originating from the Finkol River and transported from upstream locations. Apparently the island of Mutunfunalu formed over a period of at least 1500 years, though underlying lagoonal sediments were not reached. Human occupation of the island may not have occurred until Layer I formed late in prehistoric times, perhaps because until then the island was still subject to tidal inundation from the estuary. None of the excavated sediments have a lagoonal origin.

Okat Core 2

This sediment core is located within the large mangrove swamp fringing the shoreline of Okat Harbour on the west side of Kosrae [ILLUSTRATION FOR FIGURE 2 OMITTED], southwest of the confluence of the Okat and Melo Rivers, approximately 0.65 km from the open water of the Harbour. Vegetation at the coring location, which becomes inundated at high tide, consists almost entirely of mangroves (Bruguiera gyrnnorhiza, Rhizophora apiculata and Sonneratia alba) with some Nypa fruticans palms.

A 6.10-m core consisting primarily of black humic silty clay sediments was recovered. Radiocarbon determinations of 2712-2351 and 3963-3644 BP were obtained, respectively, from the 385-395 cm the 605-610 cm intervals. Using the surface as a third point (0 years) for constructing a sediment accumulation graph (computer program by Maher 1992), a near-straight line was formed, suggesting a regular rate of sediment accumulation (Ward 1995: 322). A total of 14 pollen samples were processed and ages were interpolated for them.(4)

Macro and micro charcoal concentrations

Graphs of macro charcoal concentrations for Finpea and Mutunfunalu are shown in FIGURES 4 & 5, and of micro charcoal (microscopic particles on pollen slides) for Malsu TP-2 and the Okat-2 in FIGURES 6 & 7. The deposits from which the charcoal was obtained are not specifically cultural deposits. By contrast the Katem charcoal relates to highly localized cultural activities, e.g. burning wood for cooking or other activities in proximity to living areas. As there is no charcoal in the earlier part of the Okat-2 core or the other early Kosrae cores (Ward 1988; 1995), all charcoal appears to be ultimately the result of anthropogenic activities.

The early spike of macro charcoal for Finpea reaches a density of 168.9 g/cu. m while at Mutunfunalu it is only 57.5 g/cu. m. The Malsu TP-2 micro charcoal spike reaches a density of 1.82 (area sq. mm/pollen sum), Okat-2 Core spike only 0.20 (area sq. mm/pollen sum). These differences appear to reflect the size of the catchment or watershed. The Finpea swamp, with a localized catchment, received erosional debris directly off Finol Poro hill; Mutunfunalu received sediments from a huge catchment deposited far from their point of origin. The same processes of concentration and dilution would have operated on the micro samples in the small Malsu valley and the very large Okat swamp. What is of interest is less the absolute quantity of charcoal than the changes in amounts over time at the same locations. We suggest a definite pattern on Kosrae.

The charcoal graphs all start from a point of zero [ILLUSTRATION FOR FIGURES 5 AND 7 OMITTED] or very nearly zero charcoal [ILLUSTRATION FOR FIGURE 4 AND 6 OMITTED]. Then about 1550-1350 BP there is a sudden spike of charcoal. The Okat-2 core [ILLUSTRATION FOR FIGURE 7 OMITTED] provides the best long-term and continuous record. The charcoal [TABULAR DATA FOR TABLE 1 OMITTED] collected as macro remains from Finpea and Mutunfunalu show a similar pattern [ILLUSTRATION FOR FIGURES 4 AND 5 OMITTED], though in addition they have a second spike near the top of the excavation, perhaps a result of their proximity to site areas dating to the late prehistoric period. The Malsu TP-2 unit, while not sampled intensively, is well dated, and its samples show the same trend [ILLUSTRATION FOR FIGURE 6 OMITTED]. Even the Malsu TP-1 unit is suggestive of this trend: Layer IV (200-220 cm below surface) yielded 22.1 g/cu. m of charcoal, and Layer V (220240 cm below surface) 273.6 g/cu. m of charcoal (charcoal was not observed in the upper layers during excavation, though it is possible that trace amounts were missed since these strata were not water-screened). Layer V, with its heavy concentration of charcoal, dates to 1520-1350 BP, which is the same time-frame as other units having an early burst of charcoal.

These data show widespread burning early in Kosrae's prehistory. Charcoal first appears with initial settlement (at least by 1900 BP, perhaps as early as 2000 BP); micro and macro charcoal then become plentiful in non-site areas c. 1550-1350 BP. This suggests the lowland forests were systematically burned as seen from four widely separated points around the island (Finpea, Malsu, Mutunfunalu and Okat - [ILLUSTRATION FOR FIGURE 2 OMITTED]). Presumably the burning was primarily related to agriculture, though the expansion of settlements and farmsteads from a growing population also could have been factors. Neither gradual nor persisting through time, the burning seems to have subsided almost as quickly as it began. We see in this the establishment of a lowland agroforest around the island at a very early date; further burning was not needed and only would have damaged existing crop trees (e.g. coconut, breadfruit, banana). If swidden agriculture was ever practised on Kosrae, it would have been primarily during the earliest settlement and before an agroforest was established.

Charred botanical remains

Five selected samples of charcoal from the Layer VI deposits at Katem compound (dating to c. 1900 BP) were analysed prior to separating out samples for radiocarbon dating (Murakami 1995). This amounted to 352.8 g of a total of 631.3 g of recovered charcoal. Of this 231.6 g was actually sorted into taxonomic categories (TABLE 1).

Among the identified cultigens are breadfruit (Artocarpus altilis), taro (Alocasia macrorrhiza) and coconut (Cocos nucifera). In addition, the early Katem deposits contain Thespesia populnea and Cordyline fruticosa, and probable seeds of Morinda citrifolia (from KW-106 - see Allen 1995). These latter taxa, though not food plants, may have been prehistoric introductions to Kosrae as they were for a number of Polynesian islands (Whistler 1991). There was a possible yam (either wild or domestic). Small amounts of seeds and fruits of Terminalia, Pandanus and Inocarpus fagifer were identified, plants commonly utilized for food on other Pacific islands (e.g. Kirch 1989). A number of other taxa native to Kosrae were identified, and there were 14 unknowns. Along with coconut nutshell, Bruguiera gymnorhiza (a mangrove tree) was the most common taxon.

Pollen analysis

Pollen analysis was undertaken on samples from the peaty Layer IV deposits of Finpea, the Malsu TP-2 unit and the Okat-2 core (in addition to samples from other cores not directly relevant to the present discussion - see Ward 1995).

Pollen analysis of two samples from Finpea dating to 1523-1350 BP revealed relatively common taro (Colocasia esculenta), breadfruit and coconut pollen, and a single grain of the giant swamp taro (Cyrtosperma chamissonis) pollen. Apparently the early prehistoric inhabitants of Lelu took advantage of this (presumably natural) wetland to cultivate taro. Breadfruit and coconut were likely grown on the near-by slopes of Finol Poro and adjacent coastal areas.

For the main island, analysis of three stratigraphic samples from the Malsu TP-2 unit in Tafunsak revealed abundant Cyrtosperma chamissonis pollen in the middle interval dating 1997-1709 BP and corresponding to the period of earliest human settlement. Breadfruit pollen occurred only in the latest interval, 1264-1150 BP. A single grain of coconut pollen was in the earliest sample interval, 2427-2214 seemingly too early for a human introduction; it was more common in the later intervals where preservation was much better.

The Okat-2 core provides a long and continuous sedimentary record and generally high pollen counts ([ILLUSTRATION FOR FIGURE 8 OMITTED]; see Ward 1995 for counts and other details). Three pollen zones were recognized. Zone A (c. 3800-3247 BP) is dominated both by Metroxylon palms and fern spores, though the proportion of Metroxylon is significantly richer in the basal sample. Metroxylon, no longer found on Kosrae, apparently had been extirpated by earliest historic contact (Ward 1988: 265-6). The endemic palm, Ptychosperma (not shown in the diagram), 24% of the Zone A assemblage, appears to have been common in the coastal lowlands; now it is restricted to the highest elevations.

In Zone B, c. 3247-1772 BP, Metroxylon continues to be dominant, though the mangrove taxa show steady increases. The expansion of the mangroves may be related to the drop in sea level after about 3000 BP (Athens 1995: 23951), which would have left broad coastal flats subject only to tidal inundation. Cocos makes its appearance with a single pollen grain at 2256 BP, again suggesting this palm may have been on Kosrae before human colonization. Ptychosperma virtually disappears from the record after about 3170 BP. At the top of Zone B Metroxylon begins to plummet. This trend is clearly present at the 1772 BP interval which signals the start of Zone C (this date interpolation represents an average between the lower sample at 2035 years BP, which shows no significant change in littoral forest types (principally Metroxylon), and the next higher interval at 1508 BP, which registers a drastic decline in the littoral forest).

Zone C (c. 1770 BP to present) shows Metroxylon declining from c. 50% in the previous zone to c. 10% throughout Zone C, except the uppermost near-surface sample where it falls to c. 4%. Since Metroxylon is not present on Kosrae today, the presence of its pollen in near-surface sediment may be interpreted as deriving from soils re-worked from lower levels, or, more likely, from the re-deposition of older sediments eroded from inland slopes. Other cores also document the decline and disappearance of Metroxylon (Ward 1988: 266; 1995: 324, 327).

The record of plant introductions by prehistoric human settlers is seen in Zone C with the recovery of breadfruit (1281 BP), and single-grain pollen occurrences of Cordyline fruticosa (826 BP) and possibly Thespesia populnea (1508 BP).

The abrupt change of the Metroxylon-dominated littoral forest after 1772 BP is consistent with the proposition that a lowland agroforest developed at this time. Also, it is at 1508 BP that the number of charcoal particles surges in the Okat-2 core record [ILLUSTRATION FOR FIGURE 7 OMITTED], which is the same approximate time the excavation samples show an abundance of charcoal.

Early Kosraean agriculture

The Kosrae study demonstrates that the earliest settlers arrived on the island with cuttings and corms of essentially all the food plants (except perhaps coconut) they are known to have had at the time of contact; the main exceptions are banana and sugarcane, plants difficult to identify.(5) Other economically useful plants may have been brought as well.

This is, to the best of our knowledge, the first time that agricultural remains, except coconut, have been identified archaeologically anywhere in Micronesia. While the identified food plants are what most investigators would have expected, their documentation is an important demonstration of the apparently unchanging character of Kosraean subsistence for c. 2000 years, and of the importance of agriculture from the time of earliest settlement. It is likely that Kosrae was settled on a purposeful voyage of discovery; why else would the earliest settlers have arrived in Kosrae with the full complement of plant cuttings and corms, which could only be to continue the traditional subsistence and agricultural practices of their former homeland?

The natural lowland forest all around Kosrae was transformed into an agroforest by 1550-1350 years BP. The destruction of the native lowland forest, not slow or gradual, seems to have been deliberate and rapid; it may even have been a purposeful goal of the early settlers. Although swidden agriculture is usually regarded as a precursor to intensification, it apparently was used only briefly - if at all. Indeed, the history of Kosraean agriculture seems to show stability and sameness from virtually the time of initial colonization to historic contact, with the development of an agroforest a key to this success. The rise of social complexity as evidenced by the Lelu megalithic ruins was not accompanied by any perceptible change in cultigens or agricultural practices.

Extensive/intensive agricultural systems

There are few prehistoric agricultural sequences in the Pacific readily comparable to what we have described for Kosrae. In part this must be due to the fact that data on prehistoric agricultural sequences are relatively few and of uneven quality. Another factor is that we expect to see a great deal of variation in agricultural systems and sequences as a result of the tremendous geographical variation of islands throughout the Pacific. However, we also expect to see some similarities, and the Tikopia case (Kirch & Yen 1982) comes most readily to mind. Here we have an ethnographically well-developed arboriculture system on a small, remote, high volcanic island. Kirch & Yen (1982: 350-61) propose a prehistoric sequence of early swidden cultivation and late arboriculture, with charcoal from agricultural sediments plentiful in the early record and sparse late in time.

The Tikopia data, unfortunately, entirely derive from archaeological excavations rather than wetland sediment cores. Thus, their resolution is certainly coarser than otherwise might have been the case. What we would like to suggest is that, like Kosrae, the transformation to an arboriculture system on Tikopia may have been much earlier and accomplished much more rapidly than Kirch & Yen realize. Their actual evidence for shifting cultivation, other than charcoal in the sediments, is essentially non-existent.

Although only speculative for the Tikopia case, we would like to suggest the possibility that arboriculture, here as for at least some other islands, may have had a much longer history than presently supposed, and that swidden systems on these islands may have scarcely existed or been very limited both spatially and temporally. Thus, the commonly cited developmental sequence of early extensive swidden cultivation systems eventually becoming transformed to late intensive cultivation systems may not be applicable or fully represent the variability characterizing some Pacific island sequences. Further empirical research is needed, and we hope our Kosrae case will stimulate others to undertake similar investigations.

Acknowledgements. Support for archaeological investigations was provided by the Kosrae State Government for compliance with US federal and Kosraean historic preservation legislation in relation to construction of the RS-2 road segment and wastewater facilities. We are most grateful for this support and the opportunity to conduct archaeological investigations on Kosrae. We thank the Kosrae Historic Preservation Office for its valuable assistance throughout the project. We mourn the loss of its late director, Teddy John, a warm friend and colleague who did so much to promote an interest in Kosrae's cultural heritage. Finally, we thank Jan Hather for his help with the identification of Alocasia macrorrhiza.

1 Two radiocarbon determinations were obtained on coral by Athens in late 1993 from two remnant raised reef platforms on the northeast side of Kosrae. The lower and more seaward platform dated to 880[+ or -]50 b.p., and the higher, more shoreward platform 2860[plus or minus]70 b.p. (Beta-69785 and Beta-69786; uncalibrated with no correction for isotopic fractionation). The hater determination of 880 b.p. suggests that sea level at this time, while lower than the highstand at 2860 b.p, was still above modern levels. The 880 b.p. date is not out of line with some terminal highstand dates from South Pacific island groups (Pirazzoli & Montaggioni 1986; 1988).

2 The presence of prehistoric dog has been confirmed archaeologically (Athens 1995: 93-101).

3 Two of these radiocarbon determinations have not been previously published: KW-77b (Wk-3628) from silty clay sediment 150 cm below the surface, 1280[+ or -]50 b.p. [[Delta].sup.13]C adjusted, [[Delta].sup.13]C -19.9[plus or minus]0.2; and KW-78 (Wk-3629) from humic loam sediment 170-200 cm below the surface, 1930[+ or -]130 b.p. [[Delta].sup.13]C adjusted, [[Delta].sup.13]C -23.9-[+ or -]0-2.

4 Interpolated ages, while appearing to represent precise dates, should be understood to encompass a standard error range at least as great as the radiocarbon determinations on which they are based.

5 Banana plant or fruit remains, though unlikely to be preserved, can be identified through phytolith analysis (see Pearsall 1990). Two samples from Malsu TP-2 were analysed with negative results. Sugarcane can be hard to identify archaeologically without a large preserved or carbonized piece. Although the carbonized monocot stem #2 type identified in Layer VI of the Katem excavations could be sugarcane, it cannot be positively distinguished from other grasses.


ALLEN, M.S. 1995. Macrobotanical remains, Finpea and Katem, Lelu, in Athens 1995: 349-52.

ATHENS, J.S. 1990a. Kosrae pottery, clay, and early settlement, Micronesica Supplement 2: 171-86.

1990b. Nan Madol pottery, Pohnpei, Micronesica Supplement 2: 17-32.

1995. Landscape archaeology: prehistoric settlement, subsistence, and environment of Kosrae, Eastern Caroline Islands, Micronesia. Report prepared for the Kosrae State Government, F.S.M. Honolulu (HI): International Archaeological Research Institute, Inc.

ATHENS, J.S., R. CORDY, K. SHUN, C.F. STRECK & T. UEKI. 1983. Kosrae circumferential road survey: archaeological site descriptions and test excavations. Draft report submitted to the Office of Historic Preservation, Trust Territory of the Pacific Islands, Saipan.

AYRES, W.S. & A.E. HAUN. 1990. Archaeological perspectives on food production in eastern Micronesia, in I.S. Farrington (ed.). Prehistoric intensive agriculture in the tropics: 455-73. Oxford: British Archaeological Reports. International series S232 (Part ii).

CORDY, R. 1982. Lelu, the stone city of Kosrae: 1978-1981 research, Journal of the Polynesian Society 91(1): 103-19.

1985. Investigations of Leluh's stone ruins, National Geographic Research 1: 255-63.

1993. The Lelu stone ruins (Kosrae, Micronesia): 1978-81 historical and archaeological research. Honolulu (HI): University of Hawaii at Manoa, Social Science Research Institute. Asian and Pacific Archaeology Series 10.

CORDY, R., J. BATH, K. SHUN & J.S. ATHENS. 1985. Archaeological data recovery in central Utwa, Kosrae Circumferential Road. Unpublished report; copy available at International Archaeological Research Institute. Inc., Honolulu.

HUNTER-ANDERSON, R. 1991. A review of traditional Micronesian high island horticulture in Belau, Yap, Chuuk. Pohnpei, and Kosrae, Micronesica 24(1): 1-56.

KIRCH, P.V. 1989. Second millennium B.C. arboriculture in Melanesia: archaeological evidence from the Mussau Islands, Economic Botany 43: 225-40.

KIRCH, P.V. & D.E. YEN. 1982, Tikopia: the prehistory and ecology of a Polynesian outlier. Honolulu (HI): Bishop Museum Press. Bernice P. Bishop Museum Bulletin 238.

MAHER, L.J., JR. 1992. Depth-age conversion of pollen data. INQUA - Commission for the Study of the Holocene, Working Group on Data-Handling Methods, INQUA Newsletter 7: 13-17.

MERLIN, M., R. TAULUNG & J. JUVIK. 1993. Sahk kap ac kain in acn Kosrae: plants and environments of Kosrae. Honolulu (HI): East-West Center. Program on Environment.

MORGAN, W.N. 1988. Prehistoric architecture in Micronesia. Austin (TX): University of Texas Press.

MURAKAMI, G.M. 1995. Identification of charred plant remains, in Athens 1995: 337-48.

PEARSALL, D.M. 1990. Application of phytolith analysis to reconstruction of past environments and subsistence: recent research in the Pacific, Micronesica Supplement 2: 65-74.

PIRAZZOLI, P.A. & L.F. MONTAGGIONI. 1986. Late Holocene sea-level changes in the northwest Tuamotu Islands, French Polynesia, Quaternary Research 25: 350-68.

1988. Holocene sea-level changes in French Polynesia, Paleogeogrophy, Paleoclimatology. Paleoecology 68: 153-75.

RAINBIRD, P. 1994. Prehistory in the northwest tropical Pacific: the Caroline, Mariana, and Marshall Islands, Journal of World Prehistory 8(3): 293-349.

RITTER, L.T. & P.L. RITTER. 1982. The European discovery of Kosrae Island: accounts by Louis Isidore Duperrey, Jules Sebastien Cesar Dumont D'Urville, Rene Primevere Lesson, Fyedor Lutke, and Friedrich Heinrich van Kittlitz. Saipan: Historic Preservation Office, Trust Territory of the Pacific Islands. Micronesian Archaeological Survey Report 13.

RITTER, P.L. 1981. The population of Kosrae at contact, Micronesica 17: 11-28.

SHUTLER, R., JR. 1984. Radiocarbon dating, in Y.H. Sinoto (ed.), Caroline Islands archaeology: investigations on Fefan, Faraulep, Woleai, and Lamotrek: 57-9. Honolulu (HI): Bernice P. Bishop Museum, Department of Anthropology. Pacific Anthropological Records 35.

STUIVER, M. & P.J. REIMER. 1993. Extended 14C data base and revised CALIB 3.0 14C age calibration program, Radiocarbon 35(1): 215-30.

WARD, J.V. 1988. Palynology of Kosrae. Eastern Caroline Islands: recoveries from pollen rain and Holocene deposits, Review of Palaeobotany and Palynology 55: 247-71.

1995. Sediment coring and palynology, in Athens 1995: 277-313.

WELCH, D.J., J.R. MCNEILL & J.S. ATHENS. 1990. Intensive archaeological survey of the RS-3 circumferential road corridor, Okat Valley. Kosrae. Eastern Caroline Islands, Micronesia. Report prepared for the Kosrae State Government, F.S.M. Honolulu (HI): International Archaeological Research Institute, Inc.

WHISTLER, W.A. 1991. Polynesian plant introductions, in P.A. Cox & S.A. Banack (ed.), Islands, plants, and Polynesians: an introduction to Polynesian ethnobotany: 41-66. Portland (OR): Dioscorides Press.

WHITESELL, C.D., C.D. MACLEAN, M.C. FALANRUW, T.G. COLE & A.H. AMBACHER. 1986. Vegetation survey of Kosrae, Federated States of Micronesia. Berkeley (CA): US Dept. of Agriculture, Forest Service, Pacific Southwest Forest and Range Experiment Station. Resource Bulletin PSW-17.
COPYRIGHT 1996 Antiquity Publications, Ltd.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Athens, J. Stephen; Ward, Jerome V.; Murakami, Gail M.
Date:Dec 1, 1996
Previous Article:4000 years of human impact and vegetation change in the central Peruvian Andes - with events paralleling the Maya record?
Next Article:People, process and the poverty-pew: a functional analysis of mundane buildings in the Nottinghamshire framework-knitting industry.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters