Printer Friendly

Savanna anthropogenesis in the Mariana Islands, Micronesia: re-interpreting the palaeoenvironmental data.


This paper argues that human actions had nothing to do with creating tropical Pacific island savannas, which likely arose during the Pleistocene, and that geographic factors such as soils, climate, and fire are responsible for their distribution and persistence in the Holocene. Palaeontological observations from the southern Mariana Islands, including charcoal particles, pollen, and spores in palaeosediments from Guam and Saipan, cited by archaeologists as evidence for human-caused savannas, are re-interpreted as a natural outcome of geo-climatic conditions. Archaeological and ethnographic findings, past climate proxies, and field studies in soil science are also brought to bear on the issue. The data and arguments presented in favor of natural causation of the Marianas savannas motivate a re-examination of proposals that purport to explain the presence of savannas elsewhere in the tropical Pacific. Implications for future research are drawn.

Keywords: Mariana Islands, Micronesia; savanna origins; palaeosediments


'... to kill an error is as good a service as, and sometimes even better than, the establishing of a new truth or fact.' Charles Darwin (1809-1882)

Given the continuing debate over whether the Pacific Islands grasslands, or savannas, are anthropogenic, this paper provides reasons for archaeologists to abandon the notion that prehistoric human actions were responsible for the development of grasslands, or savannas, in the Mariana Islands of Micronesia (Fig. 1), and it calls into question the appropriateness of anthropogenic savannas elsewhere in Oceania. It considers palaeoenvironmental data (fossil pollen, spores, and charcoal particles in wetland sediments) that have been cited as evidence for this view and argues that in order to include such palaeoenvironmental observations in models of the human past, extra care is needed to warrant their use as human behavioral proxies. The paper shows that such care is lacking in the proposal by Athens and colleagues that deliberate firing of 'pristine forests', presumed to have covered the southern Mariana Islands upon human entry, resulted in the rise and spread of savannas beginning by c. 4300calBP or even earlier (Athens and Ward 2004a, 2004b; Athens et al. 2004). A geoclimatic alternative to the Athens model of Marianas savanna anthropogenesis is offered below, supported by palaeosediment data generated during fieldwork by Athens and colleagues, as well as by observations from archaeology, ethnography, geography, and soil science with which the Athens model does not conform.

The paper is organized as follows. First, current explanations for savanna formation, derived from mid-20th century theories of Pacific botanists ER. Fosberg and J. Barrau and applied to the Pacific Islands, are presented and their problems indicated. Next, a recent proposal by archaeologist J.S. Athens and colleagues that also invokes prehistoric human actions to account for the savannas of the southern Marianas is reviewed and critiqued. A geoclimatic alternative then is outlined, with attention to its ability to account for observations that have been problematic for human-impact theories. The paper ends with suggestions for future research.


Fosberg and Barrau

Influential Pacific botanists Fosberg (1960) and Barrau (1961) each proposed human-impact explanations of what they viewed as the anomalous occurrence of savannas in the tropical western Pacific islands. A puzzle for Fosberg and Barrau was that annual rainfall appeared sufficient for complete forest cover in islands where there are sometimes extensive savannas. How did these non-forested areas come about? The answer to these researchers was humanly set fires. In an analogy with dry-season fires that prevent tree seedlings from becoming established in grasslands today, they proposed that non-forested areas were created by prehistoric human actions such as 'slash-and-burn' farming and that the treeless areas so created were maintained by seasonal fires, also believed to be humanly set.

Fosberg (1960) invoked the prehistoric human use of fire to account for an increase in aerial extent, rather than for the origin, of the Micronesian savannas. Referring to Guam, where savannas make up a large proportion of the island's vegetation and cover nearly all of the southern uplands (Fig. 2), Fosberg (1960:73) opined, 'It is considered probable that the present savanna vegetation has spread, under the influence of man, from small relict patches on areas of impeded drainage and has been augmented by introduced plants.' Importantly for this conclusion, Fosberg (1960:32-33) had observed during his extensive field work on the island that Guam's forests are resistant to the dry-season wildfires that help maintain the grasslands.


Today, savanna wildfires are still considered unnatural; reasons include accidents while burning debris, and arson by deer hunters; arson is thought to account for over 80% of these seasonal conflagrations (Neill and Rea 2004; Minton 2005), although inadequate data on arson frequencies are available. The fact of Guam's forests' fire-resistance favored the idea that enlargement of presumed small patches of savanna prior to human advent in the islands could only have been accomplished by newcomers deliberately firing forests where now there are grasslands. Fosberg offered no reason for the early islanders intentionally doing this but his contemporary, Jacques Barrau, provided one: primitive gardening practices. Barrau (c. 1966) illustrated the deforestation process by over-intensive agriculture, entitling his schema 'An Aspect of Man's Impact on the Vegetation of Malayo-Oceania' (Fig. 3):


Barrau describes the original forests of Malayo-Oceania as a 'climatic climax community' since tropical temperatures were not limiting and annual rainfall was considered ample (seasonality in rainfall, characteristic of many islands where savannas occur, was not considered in the schema). These climatic climax forests were stable until the first human occupants began to make gardens in the forests by clearing and burning; over time, their primitive agricultural methods led to replacement of the tropical rain forest by secondary forest; continued intensified clearing and burning resulted in the replacement of secondary forest by high grassland, such as those dominated by Miscanthus. Further intensified burning and cultivation produced a 'plagioclimax (or disclimax)' of shorter grasses. Earlier, in his comprehensive treatise on the subsistence agriculture of Polynesia and Micronesia, Barrau (1961:10), offered a detailed theory of the human role in the rise of savannas:
   It is probable that the original agricultural systems, at
   least on the high islands, were either shifting agriculture
   or agriculture with bush-fallowing rotation. Burning was
   used for clearing space for gardens. Increased
   populations and the use of primitive agricultural
   techniques were probably responsible for deterioration of
   both vegetation and soils on the majority of high islands.
   With the decrease in land fertility it became necessary to
   employ semi-permanent forms of agriculture with
   artificial fertilization of the soil. It was this need,
   apparently, that led to the development of taro-growing
   on low-lying, hydromorphic soils and on irrigated
   terraces in the valleys, techniques which may have been
   introduced by the migrations [of later peoples with more
   advanced agricultural techniques, RHA].

Problems with the Barrau theory

Use of soils and terrain. Barrau's schema predicts a process of soil degradation from over-intensive farming because shortened fallows would not allow the soils to rest and 'regenerate'. However, savanna soils were probably just as nutrient-poor prior to human arrival as they are today in the islands. Yap's and Palau's grasslands coincide with nutrient-poor volcanic soils (Johnson et al. 1960; Falanruw et al. 1987; Cole et al. 1987), as do savannas in other islands of volcanic origin, such as Viti Levu and Totoya in Fiji (noted in Nunn 1994). Guam's savannas generally coincide with the distribution of volcanic soils, which occur mainly in the southern half of the island (Donnegan et al. 2004). In contrast, limestone substrates with soils of low or no acidity typify the three large islands north of Guam (Tracey et al. 1964; Young 1988). In Rota, Tinian and Saipan, grasslands cover 3125 ha, c. 15%; 2872 ha, c. 11%; and 3237 ha, c. 11%, respectively (Falanruw et al. 1989).

The notion that over-intensive farming by early prehistoric settlers had degraded the interior soils upon which savannas now occur is contradicted by soil scientists and other experts, who attribute nutrient-poor savanna soils, sometimes with high levels of bauxite, to long geological periods of tropical weathering (e.g. Bridge and Goldich 1948; Tracey et al. 1964; Young 1976; Sanchez 1976; Demeterio and DeGuzman 1988; Morrison 1999; and see Mueller-Dombois and Fosberg 1998:203). Since these soil conditions already prevailed in island interiors upon human entry, they were likely avoided for crops, or were the last rather than the first to be utilized, as Hunter-Anderson (1983) has argued for Yap.

Which farming techniques were first used in the Pacific Islands and whether relatively labor-intensive taro cultivation was one of them have been considered by Leach (1999), who concluded that both extensive and intensive methods likely were part of island settlers' agricultural repertories and both may have been used initially, even under low population density and consumer demand, depending upon local circumstances. Zan and Hunter-Anderson (1988), Hunter-Anderson (1991) and Falanruw (1990) show habitat-sensitive variation in traditional agricultural techniques is typical of Micronesian farming systems, and there seems no reason to doubt that intimate environmental knowledge was part of earlier cultural repertories in these islands.

Noting the impracticality of gardening the steep slopes of the southern Marianas, Fosberg (1960:29) generalized about locally adapted arboriculture in some of these areas as he had observed it:
   Although most of this land is too steep to be suitable for
   ordinary agriculture, because of erosion and leaching, an
   indigenous type of cultivation has been evolved that is
   eminently well suited to such terrain. This is a mixed
   planted forest of coconuts and breadfruit. From a distance
   such a stand resembles natural forest rather than planting.
   Spacing and arrangement are not at all regular, but are
   more suited to the terrain. Bananas, citrus, papayas, dry-land
   taro, yams, and Alocasia are commonly raised on the
   slopes beneath the trees, along with other minor
   crops ... In general the ravines seem to be planted more
   completely to breadfruit, while the ridges are given over
   to coconuts.

It is reasonable to assume that such accommodations to local terrain and other geographic factors were made by prehistoric farmers in Guam and neighboring islands. Rice cultivation, probably at interior wetland margins (Hunter-Anderson et al. 1995), was part of the mix, and considering the preference for coastal settlement throughout the prehistoric period, the rich alluvium of the lower valleys and better-watered back-strand areas were also utilized, possibly with relatively labor-intensive methods. In fact persistent clear-cutting of primary or secondary forests, implicit in the Barrau schema, has nowhere been documented in island farming systems.

Sediment deposition histories. Also implied by Barrau's schema is a distinctive lowland sediment deposition history reflecting deforestation and terrigenous erosion that began early in the human era, whenever that time was on each island (see related discussion in Nunn 1991:8). Marianas palaeosediment data show a different sediment deposition pattern than expected by the schema. For example, in the IARII Laguas core, raised near the mouth of the Laguas River on Guam's southwest coast (Athens and Ward 1999, 2004a), there are no pulses of sediment that might signify early interior deforestation and terrigenous erosion. It shows that sediment accumulation rates rose and fell radically twice during the Holocene but these fluctuations appear unrelated to human activities. For example, Table 3 in Athens and Ward (2004a) indicates two stepped periods of rapid accumulation (the format of radiocarbon dates cited here and below is that used by Athens and Ward, based upon age/depth interpolations derived from calibrated radiocarbon determinations).

The first period began c. 6574 calBP and peaked c. 5217 calBP, well prior to archaeologically established human advent c. 3500 years ago, and these mid-Holocene sediments are marine in character. The second period of rapid sediment accumulation at the Laguas coring site began c. 3558 calBP and peaked c. 2679 calBP, and thus falls within the archaeologically established prehistoric human era. However, sediments here are also marine rather than terrigenous. In the IARII Laguas core, then, terrestrial erosion from deforested interior lands is not evident in the sediment accumulation record.

Sediment accumulation rates are unavailable for other Marianas cores and deep soil exposures. However, information about the timing of onsets of terrigenous sediment deposition is available from Guam's southwestern coast north of the IARII Laguas coring site (Athens and Ward 2004a:26-27). This information also points to a mismatch with expectations. There, soil deposition increased between the early first millennium BP and the late second millennium BE well after human advent and after deforestation was supposed to have begun in the Marianas. While these sediment pulses have been interpreted by Athens and Ward (2004a:26) as evidence for inland agricultural intensification, another, more plausible, interpretation can be entertained.

This period saw beaches widen as sea level declined (see Dickinson 2000, 2001; Amesbury 2007). With the lowering of stream base levels, sheet wash and stream down-cutting would have deposited sediments along the coast from surrounding hillslopes (see discussion in Nunn 1994:325). Indications of such a process north of the Laguas area include the 'heavy clayey sediments' found on top of the 'slightly raised reef' whose uppermost portion was dated to 2455-2298 calBP (Athens and Ward 2004a:26, citing Athens and Ward 1999). As these authors noted, this date marks the termination of growth of the reef, possibly from sea level decline, but is not necessarily the start date for terrigenous deposition, which could have been later. In fact, evidence for somewhat later onsets of terrigenous erosion on Guam, c. 2000 years ago, were reviewed in Dye and Cleghorn (1990; and see Hunter-Anderson 1989); both these works were cited by Athens and Ward (2004a) in support of their conclusion that 'considerable coastal infilling occurred during the past roughly 2,000 years' in the Laguas area and elsewhere in southern Guam. Given the known geo-physical response to sea level decline of stream downcutting and sheet erosion, lower sea level likely played a key role, possibly the only role, in effecting these onsets.

The lesson provided by the Marianas case suggests that other sedimentation histories should be examined for conformity with the expectations of Barrau's schema, since many archaeologists still maintain its general validity (e.g. for Mangaia and Futuna see Kirch 1994:272, 1996; for Easter Island, Mieth et al. 2002; for Yap, Dodson and Intoh 1999). At Mangaia, for example, the sediment deposition histories of the 21 cores raised in eight drainages (Ellison 1994) do not conform to Barrau's schema in two aspects: (1) non-conformity between deposition rates and the timing of human actions as inferred from dated changes in palynomorph frequencies and (2) asynchrony among the cores with respect to signs of prehistoric human actions such as onset of the deposition of clay over peat, thought to represent forest disturbance.

With respect to (1), according to data provided in Ellison (1994: Table 1 and Fig. 2), the 860-cm deep core TM7, the main exemplar in discussions of Mangaia's environmental history (e.g. see Kirch 1996; Kirch and Ellison 1994; Kirch et al. 1995), sediment deposition rates were highest in the deepest, oldest portion, at 0.20 cm/yr in the segment from 830 to 680 cm, dated between 7240 and 6480 calBP By the mid-Holocene, 6480-4500 calBP, sediment deposition rates had decreased to 0.12 cm/yr and continued to drop over time. They were at their lowest, 0.03 cm/yr, in the segment from 52 to 0 cm, dated between 1640 calBP and the present. This is the opposite pattern to that expected under the Barrau theory, where sedimentation should have increased through time as human activities continued to damage the forests, and particularly after 1650 calBP when 'widespread clearance of vegetation on the central cone' (Ellison 1994:13) is supposed to have occurred.

With respect to (2), asynchrony is a feature of the sediment deposition histories among the Mangaia cores. Some of the differences must be due to their geographic placement along the island perimeter and to different catchment sizes. Nonetheless, under the Barrau theory (and according to Ellison's conclusion of 'widespread clearance of the central cone' after 1650 calBP), human effects of forest clearance upon the central cone's vegetation should have been evident in these sediment records at approximately the same time. Instead, in each core the timing of the onset of sustained clay over peat deposition near the top of the cores, assumed by Ellison (1994:13) as the sign of 'major environmental disturbance', differs by thousands of years.

According to data in Ellison (1994: Figs. 1, 2 and Tables 1, 2), the oldest onset times for the sustained deposition of clay over peat are recorded in the three cores from swamps on the north and east portions of the island perimeter (KA3, 6260 calBP; KA4, 2980 calBP; and IV1,1930 calBP). Much younger onset times are recorded in cores from swamps on the southern portion of the island perimeter (VT5, 1830 calBP; TM7, 1640 calBP; and VT6, 610 calBP). In core TM7, with some of the most complex stratigraphy of the Mangaia cores, the 610 calBP timing of the onset of sustained clay over peat was interpreted not as an anomalously late indication of clearance of the central cone's vegetation but as 'a point in a trend of continuing infill of Lake Tiriara in Veitatei' (Ellison 1994:13). Why this case is not interpreted like the others, as a sign of the beginning of environmental disturbance in the upper Veitatei catchment, is not explained. Ellison summarized her findings as follows. 'Major forest destruction occurred at ca. 1650 yr BE coinciding with major clay deposition as seen in the upper sequence of cores in all basins.' As the data cited above show, there was no coincidence.

More detailed evaluation of the human-impact theory using Mangaian palaeosediment data is beyond the scope of the present paper, but the reader is invited to compare the pollen diagrams of TM7 and VT6, both cores taken from adjacent swamps on the southern side of the island (Ellison 1994), with those from the Marianas (cited above). Among the similarities in pollen and spore sequences is the presence of grasses and other savanna plants throughout the sequences and a general trend of vegetation adapted to a drier climate beginning in the mid-Holocene.

Archaeology. Turning to archaeology, the Barrau theory predicts a prehistoric settlement pattern shift from a mobile one dictated by swiddening schedules in the interior forests, to a sedentary one near coastal taro plots. The expectation of a settlement pattern shift from inland to coastal settlement has not been met, according to Marianas archaeological data obtained since the mid-20th century, beginning with Spoehr's (1957) work in Rota, Tinian and Saipan, as well as Reinman's 1965-1966 surveys on Guam (Reinman 1977; and see Kurashina 1986). The Marianas prehistoric occupation sequence is typical of other Pacific islands where coastline areas were available: the earliest sites are all found on old beaches, and only several centuries later did people begin to occupy the island interiors. In Guam's case, inland settlement expansion up the river valleys began c. 1500 years ago, fully two millennia after human arrival as determined archaeologically (Kurashina 1986; Hunter-Anderson and Butler 1995). Thus deforestation by intensive cultivation of the interior starting early in Guam's prehistory cannot have happened in the way proposed by Barrau.

Especially damaging for the notion that agriculture radically changed the southern Mariana Islands' interior vegetation during early prehistory is that the Pre-Latte artefact assemblages dating between c. 3500-2500 years ago contain no agricultural implements (in contrast with Latte Period assemblages of the ancestral Chamorro cultural system, beginning c. 1000 years ago; for summaries see Rainbird 2004). Also missing from these early sites are the remains of domesticated animals and of the commensal Rattus exulans that apparently accompanied the earliest Polynesians into remote Oceania (Spoehr 1957; Reinman 1977; Moore 1983; Haun et al. 1999; Rainbird 2004: 117, 247).

Archaeological surveys in interior southern Guam (e.g. Hunter-Anderson 1994; Henry et al. 1999) have found no prehistoric evidence for terracing or other signs of intensive farming methods. The earliest archaeological evidence for taro growing comes from coastal sites and is relatively late in the prehistoric sequence. For example, Colocasia starch residues have been identified on the interior surface of three pottery sherds from two sites in the Ague Cove area of northwest Guam (Moore 2002:46). The earliest residue evidence comes from a pottery sherd of a type (distinctive mat-impressed 'pans'; see Hunter-Anderson and Moore 1999) that appears in Guam sites occupied during the Transitional Period, c. 2500-1500 BE The other two sherds with Colocasia starch residue came from rock shelter deposits dated to the late Transitional through Latte Period (c. 1500-1000 BP). These two sites are within 100 metres of each other, and the taro starch residue dates suggest a longstanding practice of taro consumption in the area, perhaps from gardens nearby.

While the absence of evidence can never be conclusive, the lack of agricultural tools, domesticated animals, and remains of Rattus exulans in the Pre-Latte assemblages requires new thinking about possible changes in the human adaptive contexts in these islands over the three millennia spanned by the archaeological record.

In addition to Palau, Yap, Fiji, New Zealand (discussed in another section below), Mangaia's archaeological record is at odds with claims of early human presence inferred palaeontologically. Although palaeosedimentary data from its swamp cores have been cited as evidence for human presence as early as 2500 years ago, no Mangaian occupation sites have been found older than c. 1000 C.E. (Kirch et al. 1995).

Unchecked population growth. Implicit in Barrau's schema (and in more recent models of human environmental impacts, e.g. Flenley et al. 1991; Kirch 1994, 1996; Dodson and Intoh 1999; Mieth et al. 2002) is the notion of unchecked population growth favoring ever more intensive farming methods to meet increasing demand. The role of population growth and other factors in agricultural intensification have informed a large and sometimes contentious literature, for example, Boserup (1965), Smith (1972), Hughes (1975), Tainter (1988), Fairhead and Leach (1996), Stone and Downum (1999), Leach (1999), Decker and Reuveny (2005) and many others. Population 'overshoot' with radical repercussions in subsistence and settlement systems is questionable here, given ethno-demographic data from small island communities where family size and farming land allocation are carefully controlled through customary practices (Carroll 1975; Kertzer and Fricke 1997).

Underlying such customary behavior associated with land use may be the simple recognition of the law of diminishing returns for effort in agricultural intensification, particularly where soils are less than optimal: the costs of maintaining the stability of energy and nutrient flows into an agricultural plot rise exponentially (see discussion in Athens 1977:362-374). Given the nutrient-poor status of Guam's volcanic upland soils (Yap's and Palau's too), this lesson would be severe indeed, and prehistoric islanders would have very quickly learned that exerting ever more effort to increase production by intensification 'costs' too much. There is also the question of adequate labour available for intensification. While Yap and Palau rarely experience typhoons, and may have had higher human densities in the past (see Hunter-Anderson 1983), human densities in the Marianas were probably never very high given frequent droughts, typhoons, and earthquakes, a condition that would discourage investment in permanent agricultural plots and favor a relatively mobile settlement system. In ecological terms, these small, remote islands were likely human population 'sinks' (see Pulliam 1996) with a near-constant recruitment problem and a history of periods of extreme population deficit.

Athens and colleagues

Palaeoenvironmental studies by Athens and his colleagues have ranged widely within Micronesia but here the focus is on their work in the southern Marianas. Mindful of the lack of early archaeological evidence for interior settlement in these islands, and of the lack of evidence for early terrestrial deposition in coastal contexts, Athens and Ward (2004a) have proposed that instead of intensive farming, as in the Barrau theory, it was casual firing of the interior forests that set in motion a positive feedback loop between human actions and the upland forests, resulting in a permanent landscape change:
   To account for the formation of the savannas, it appears
   that dry season fires must have been intentionally set on
   occasion, perhaps by individuals making forays into the
   interior for wild tubers or other wild food resources.
   These fires might have been set with the intention of
   increasing the production of certain wild forest products,
   to facilitate travel through these areas, or for pure
   entertainment. The actual reason is probably not
   determinable. The result, however, was that with
   exposure to the sun and tropical rains, and continued
   firing at irregular intervals during dry seasons, the fragile
   soils of the upland landscape quickly became degraded
   and could no longer support forest vegetation (Athens
   and Ward 2004a:27).

The time period over which this process operated was from c. 4300 calBP (or earlier) to about 2300 calBP (Athens and Ward 2004a:25-26). Inserted into the sequence at c. 3900 calBP was the notion of 'small-scale gardening activities' in Guam's uplands, so that at this point the Barrau and Athens theories converge.

Problems with the Athens theory

Scientific method. Stating that 'the actual reason' for 5th millennium, non-agricultural settlers lighting fires in the upland forests is probably not determinable and thus likely a waste of time to pursue, minimizes the importance of the scientific requirement to provide reliable warrants for citing this behavior in an explanation. Arguments and data that make expectable both the suggested behavior and its purported results of soil degradation and erosion and replacement of forests by savanna are precisely what are needed to make this theory acceptable, indeed testable. Without these, the theory is weakened by simply invoking behavioral plausibility. In addition, a theory needs to be comprehensive, accounting for as many aspects of the phenomenon being explained as possible. The Athens theory lacks comprehensiveness, first in not accounting for the subsistence system change to inland agriculture after several centuries of presumably marine-focused subsistence, and second in its inattention to a critical implication of Fosberg's finding that Guam's forests are highly resistant to fire.

Regarding the first problem, no facts or arguments are presented to show how the southern Marianas could have supported several centuries of human settlement without agriculture, when sea level had not yet declined from the mid-Holocene highstand (Dickinson 2000). Island land masses were smaller, coastal approaches more difficult, and protective reefs not yet developed (see Nunn 1994). Regarding the second, natural fire resistance of Guam's forests implies that burning must have been intensive and frequent in order to effect such a major vegetation change as wholesale conversion of forests to grasslands, in contrast with seemingly less effective burning 'at irregular intervals' over a long time span. No arguments or data are presented to warrant this complex proposition.

Soil degradation and loss. The Athens theory predicts soil degradation and loss through erosion from the interaction of burning, sun and tropical rains, once people arrived in the southern Marianas and began to burn the forests. Ample quantities of sun and rain probably have been important features of southern Marianas climate since at least the end Pleistocene, and clearly they contribute to terrestrial erosion by alternately drying and saturating exposed soils that make them vulnerable to wind and water scouring and slumping. The issue here is what role burning down trees, in addition to the operation of sun and rain, has in this process. The Athens theory assumes that trees are better than grasses at holding 'fragile soils of the upland landscape' in place, and that if trees are removed, severe erosion in these areas will ensue.

This assumption appears unfounded: recent field experiments under natural sun and rainfall conditions on sloping plots on the volcanic soils of Guam's south-central savanna indicate that virtually no soil loss takes place after a burn of the natural vegetative cover of grasses, shrubs, and small trees (Golabi et al. 2005a, 2005b). The least soil loss occurred on plots planted to Vetiver grass (an introduced species), while the most severe soil loss in these experiments occurred on plots in which all vegetation had been removed prior to burning. It would appear from these results that the large plumes of sediment that have been recorded in bays in southwestern Guam (e.g. see Minton 2005) probably derive from the 'badlands' sections within adjacent grasslands, where saprolite is exposed and erodibility is highest (Young 1988).

Palynomorphs and charcoal particles. The above facts notwithstanding, Athens and colleagues have preferred to interpret pollen, spores, and charcoal particles in palaeosediments as indications of human landscape manipulations. For example, from the timing of the initial appearance of charcoal particles in wetland cores, followed shortly by rises in pollen and spore indicators of ecosystem disturbance (listed below) in the early 5th millennium BE these authors argue that the charcoal particle increases are prima facie evidence for deliberate burning of Guam's interior forests that started the process of savanna creation. In this interpretation, the deforestation had accelerated c. 2900 calBP and was complete by c. 2300 calBP, when 'only remnant patches of the native forest remained' (Athens and Ward 2004a:26).

These authors dismissed a climatic cause for the appearance of 5th millennium charcoal in palaeosediment cores as follows:
   It might be argued that charcoal particles were derived
   from natural burning during drier conditions. However,
   any climatic change towards more arid conditions should
   show a broader scale impact, such as an increase in firing
   in [Guam's] interior. No such increase is visible, nor is
   any evidence of increased long-term drought seen
   elsewhere in the global paleoclimatic record (Athens and
   Ward 2004a:25).

Data affirming broad-scale impacts regionally and globally of drier conditions during the mid-Holocene will be presented in another section. Here the palynomorphs cited in support of the deforestation process envisioned by Athens and Ward are considered in detail. They include palynological indicators of savanna communities such as Gleichenia linearis, Lycopodium cernuum and Poaceae (grasses; the pollen of individual grass species is not usually identifiable), which make their first 'significant' appearance in the Pago and IARII Laguas cores at about the same time as the earliest charcoal particles, between 4800 and 3300 calBP (Athens and Ward 2004a:Figs. 7, 8; Athens et al. 2004:Table 1). However, it is also the case that tree and shrub pollen from numerous genera and species was observed throughout the two cores, and that pollen and spores from savanna indicator plants were also observed down the cores, albeit in low frequencies, and well below samples containing charcoal (Athens and Ward 2004a:Fig. 6; Ward 1994:Fig. 9.B.2., Table 9.B.3). The base of the Pago core is estimated at greater than 10,450 calBP (Ward 1994:Fig. 9.B.2.), and that of the IARII Laguas core at greater than 9242 calBP (Athens and Ward 2004a:Fig. 6). Thus, contrary to the Athens theory in which human-created savanna vegetation first appears on Guam c. 4800 and 3300 calBP, these data suggest a savanna history that is considerably older than that, even older than the first palaeoenvironmental signs of fire in Guam ecosystems.

Turning to Saipan, where there are palynomorphs in palaeosediments that date well prior to the mid-Holocene, I now consider Trench 98-4, a >4 m-deep soil exposure at the Kagman sinkhole located in the karstic southern part of the island (Cummings 1996; Cleghorn 1998; Dega and Cleghorn 2003; Athens et al. 2004). In contrast to the Guam cores, the Kagman profile exhibited relatively high Poaceae pollen frequencies throughout, suggesting grasses were present in the vicinity by at least 7900 calBP (the base of the deposit was not reached; basal soil sampling began at 420 cm below the surface; see Cummings 1996:2). Judging from the kinds and frequencies of palynomorphs in the lower level samples (dating from c. 7900 calBP up through levels dating to c. 4520 calBP), Athens et al. (2004:26) characterize the vegetation as 'mixed forest and grassland'. By about 4520 calBP (beginning at c. 250 cm below the surface), a sudden shift in palynomorph frequencies indicates a more open landscape had developed. The shift is evidenced by a steep rise in spores of the fern Lycopodium cernuum, lower counts of Pteris and other fern spores that had dominated lower levels, and a reduction in palm pollen counts (Athens et al. 2004:Fig. 6). Cummings (1996:4) summarized the change in pollen and spore frequencies as a removal of the 'taller elements of the vegetation community', i.e. in upper levels, Pandanus pollen disappears, there is less palm pollen, fewer fern spores and more grass pollen.

Athens and colleagues aver that this shift in pollen and spore frequencies is the clear sign of human intervention in ecosystem processes, and note the consistent presence of charcoal particles at this location in the profile as further evidence of human presence. However, it is the case that charcoal particles were observed throughout the Kagman sinkhole soil exposure but were not quantified in samples from lower than 250 cm below the surface (these samples are presently curated in Honolulu and could be analyzed and dated; M. Dega pers. comm. 2009). According to the logic of Athens and colleagues, if the deeper (and presumably older) charcoal in samples from Trench 98-4 is as valid an indicator of fires as that seen in the samples from 250 cm and above, then people were present in Saipan as early as 7900 calBE If the deeper and presumably older charcoal is considered non-human in origin (no one so far has claimed people were present in Saipan by 7900 calBP), then why not the younger charcoal, still a millennium older than archaeological evidence for human presence in Saipan, in this core? The parsimony standard would suggest the latter conclusion is preferable, especially given the discrepancy in timing of the earliest archaeological evidence.

Another Saipan palaeosediment core is available for examination for signs of Holocene vegetation changes, that raised at the Susupe marsh near the west coast. This 6 m-deep core spans nearly eight thousand years of sediment deposition (Athens and Ward 2004b:Table 5). As in the Kagman Trench 98-4 profile, the Susupe core indicates a pre-mid-Holocene open landscape. According to Ward's analysis, Poaceae pollen occurs throughout the Susupe core as well do as low counts of Gleichenia linearis spores, including in a sample estimated to date c. 7680 calBP Low counts of Lycopodium cernuum spores, a savanna indicator, start c. 4840 calBP in the same sample as the earliest charcoal and continue up the core. This addition does suggest a vegetation change, but not one caused by wholesale deforestation.

Savanna endemics. Evolutionary theory posits that endemic species become reproductively differentiated by isolation over long time periods. Under this understanding, the savanna plant endemics presently found in the Marianas, such as Glochidion marianum, Myrtella benningseniana, Timonius albus, Ischmaemum longisetum, Hedyotis fruticosa, Geniostoma micranthum, Phyllanthus saffordii, and Dimeria chloridiformis (Mueller-Dombois and Fosberg 1998:232), indicate great antiquity of their preferred habitat. Since they are shade-intolerant, these plants cannot have developed or lived under forest, as Fosberg (1962) noted (one of the reasons he allowed that 'man' did not create the savannas but only expanded them). Palynological evidence for the antiquity of Saipan's savanna includes Ward's identification in the Susupe core of Timonius nitidus? pollen in a sample estimated to date c. 7821 calBP. The pollen of T. nitidus? was also identified in two other samples, estimated to date c. 4027-4168 calBP (Athens and Ward 2004b:Table 6). If these identifications are correct, savanna was present in Saipan well before the mid-Holocene as well as after the mid-Holocene (and over four thousand years before the archaeological record began in the Marianas).

Manner et al. (1999:100) have remarked that the existence of savanna plant endemics in Micronesia, including the southern Marianas, does not necessarily preclude human causation, since they could have evolved elsewhere (e.g. in the far northern islands of the archipelago which have extensive grasslands) and then 'invaded the [southern Marianas] savannas after they were formed' (citing Fosberg 1960). However, this travelling endemics scenario should also have to include some prescient (and very fast-migrating) bird populations, because the bones of grassland-dependent rails (Gallirallus sp.) have been found in Saipan in early Pre-Latte cultural deposits dating to c. 3500 years ago (Steadman 2006:273). This finding implies that the preferred habitat of these birds was available to them in the southern Marianas when people arrived in the archipelago. Linguistic hints of a long interaction between savanna biota and Mariana Islanders include the native Chamorro terms for grassland birds (for species and preferred habitats see Pratt et al. 1987) such as the Guam rail (Rallus owstoni; koko) and the short-eared owl (Asio flammeus; mommo), as well as names for savanna plants such as the Gleichenia linearis fern (man, mana) and the endemics Glochidion marianum (chosgo), Hedyotis fruticosa (paode 'do'), Phyllanthus saffordii (gaogao uchan), and Timonius nitidus (maholok layu); and the tall sword grass Miscanthus floridulus (nette, tupon nette) that burns so well in wildfires. Although Spanish has heavily influenced the Chamorro language, these still-used terms are clearly indigenous (Topping et al. 1975).

In sum, this review has shown some problems between the human-impact theories of savanna formation proposed by Fosberg, Barrau, and Athens and colleagues and data from a variety of sources. Now it is time to turn to a more realistic, geo-climatic alternative.

A geo-climatic alternative to anthropogenesis

While explanations of botanical phenomena such as Pacific Island savanna origins are not usually part of an archaeologist's intellectual kitbag, it seems necessary to provide at least an outline of one here. Added to the above arguments and data, relevant findings from Pleistocene palaeo-geography, Holocene climate studies, fire history and contemporary wildfire documentation help form a more comprehensive interpretive framework for understanding how savannas arose and have persisted into modern times. These findings and their implications for better understanding the Marianas savannas, and perhaps savannas elsewhere in the Pacific, are presented below as a series of propositions.

A Pleistocene 'Savanna Corridor' included the Marianas

The idea of ancient, pre-human savannas in the tropical western Pacific finds support in a recent study (Bird et al. 2005) that evaluated a large body of evidence for the consistent presence of a 'savanna corridor' during the coldest phases of the Pleistocene. Data from geomorphology, palynology, and biogeography have been incorporated in climate modeling and in constructions of past vegetation. In this construction, forests were restricted to refugia primarily in Sumatra, Borneo and the continental shelf beneath the modern South China Sea, which extended through the continent of Sundaland (modern Indonesia and Malaysia). While the study did not consider the Pacific Islands, there is no reason to doubt that the vegetation of the southern Marianas responded similarly during Pleistocene cold phases (for other cases, see also Spriggs 1981; Southern 1986 and discussion in Nunn 1991:8, 1994:316-317).

Given prolonged periods of cooler and drier Pleistocene climate, savanna plant endemics likely evolved within the Mariana archipelago. Smaller forested areas, confined to moister and less exposed locales, with extensive open grasslands on ridges and hillsides, may have characterized Guam, Rota, Tinian and Saipan during these times. Since the Pleistocene was not a climatically stable period in Earth's history, these forest/savanna boundaries probably fluctuated with oscillations in temperature and moisture, some of which were gradual and others relatively abrupt (e.g. Stocker 2000; Stevens et al. 2008).

Early Holocene conditions may have reduced the areal extent of savannas

With the trend toward a warmer and wetter early Holocene, beginning c. 10,000 years ago (see Roberts 1998; Mackay et al. 2005), island savannas again may have decreased in areal extent, as forests expanded under more favorable growing conditions for trees. Current savanna plant endemism and faunal preferences for open grassy habitats thus can be understood as related to the Pleistocene origin of savanna vegetation, and the survival of savanna endemics as related to the survival of savannas, albeit in smaller proportions of total vegetation.

During the early Holocene, the composition and structure of island savannas likely changed in response to more favorable growing conditions but were different than today. For example, higher annual precipitation may have enabled more vigorous plant growth in the savannas where soil conditions precluded major encroachment of forests, particularly where this condition favored endemics. Island faunas likely changed as well; for example, water-loving bird species may have increased due to the new availability of interior wetlands and of estuaries and mangroves that developed with sea transgression.

Mid-Holocene aridity and higher temperatures dried vegetation and reduced plant biomass

In contrast to climate in the early Holocene, during the mid-Holocene aridity and temperature were higher in the Pacific/Indonesia region and elsewhere (see Gagan 1998). Evidence for these changes has been accumulating over recent decades (Tsukada 1966; Haberle et al. 2001; McGregor et al. 2002; Morrill et al. 2003; Gagan et al. 2004; Anderson et al. 2007; Wei et al. 2007). As early as the 1960s (Tsukada 1966), mid-Holocene warmth and aridity in Taiwan had been deduced from changes in fossil pollen and other palaeontological data. More recently new analytical techniques have enabled McGregor et al. (2002; and see McGregor and Gagan 2004) to recover fossil coral evidence related to this period. They detected a significant rise in sea surface temperature beginning c. 7000 years ago and peaking c. 6000 years ago. Similar findings from fossil coral retrieved from off Hainan Island are reported by Wei et al. (2007).

In the Marianas, a stalagmite from Guam was analyzed by Sinclair et al. (2008), who found a good match in isotopic signatures of Mg and Sr between the Guam speleothem and several other Pacific speleothems. According to the authors, 'Both elements show a broad peak centered on 5-6 kybp, which may indicate a dry period in the western Pacific during the mid-Holocene.'

Taken together, these disparate data seem to affirm that 'increased long-term drought in the global palaeo-climatic record' occurred during the mid-Holocene in the western Pacific. Climatic conditions during this period apparently affected local and regional biochemical processes, as recorded by Pacific Islands speleothems, coral records, and palaeosediments.

Mid-Holocene climatic effects upon the vegetation of the Marianas likely included a decline in abundance and/or specific adaptive changes in the plants accustomed to greater moisture and cooler temperatures--in short, a drying-out of existing plant communities, with displacements and replacements of some forms. In the islands north of Guam, such as Saipan, the mid-Holocene climate may have been more effective in maintaining Pleistocene-originated open vegetations than in Guam, since annual rainfall amounts are lower; with increases in latitude in this archipelago, temperatures decline as does annual rainfall (Lander 1994). This geographic difference may be reflected in the Saipan palaeosediment data from Susupe and Kagman, which record grassland forms on the island by c. 7900 calBP.

The drier, hotter conditions of the mid-Holocene may have added a new element in island ecology, fire. Evidence that post-mid-Holocene climatic change is linked to fires is found in palaeosediment data from Guam and Saipan. Regarding the denial that a 'broader scale impact' of drier conditions, 'such as increased firing in the interior' is not 'visible' (Athens and Ward 2004a:25), there are only two interior records of ancient fires that can be consulted so far. The visible evidence comes from the Pago core in south-central Guam (Ward 1994; Athens and Ward 2004a) and from Trench 98-4 in the Kagman sinkhole in southern Saipan (Dega and Cleghorn 2003; Athens et al. 2004).


At Pago, the earliest charcoal particles were dated to c. 4800 calBP, five hundred years before the earliest charcoal particles in the coastal Laguas core, suggesting that fires occurred in Guam's interior prior to those recorded in the lowland sediments. The complete fire record of Kagman Trench 98-4 is unknown because while charcoal particles deposited between c. 4129-3933 and 4863-4653 calBP were counted in five samples from the middle of the 4 m-deep exposure (basal date, 7900 calBP), charcoal particles present in the lower samples were not counted (Athens et al. 2004a:26; Dega and Cleghorn 2003:31, 39; Dega pers. comm. 2009). Thus, fire antiquity in interior Saipan is very likely to be even greater than reported. In the coastal Susupe marsh core in Saipan, charcoal particles first appear c. 4860 calBP (Athens and Ward 2004b; Athens et al. 2004:Table 1). If future analysis of charcoal occurrence in the lower samples from Trench 98-4 confirms even earlier fires at Kagman, then Saipan followed Guam's pattern of fires occurring in the interior first and the coast later.

Strengthening the proposition that post-mid-Holocene climates brought vegetation biomass adjustments to drought in the Marianas, are total concentrations of palynomorph and charcoal particles in the Laguas and Pago cores. These measures can serve indirectly as climate proxies as they relate to hydrological conditions including erosion rates. In the IARII Laguas core (Fig. 4), a precipitous drop in pollen and spore concentrations was recorded at 23 m depth in the core, at c. 7134 calBP. Charcoal particle concentrations started to be calculable at 12 m depth, at c. 4324 calBP. If the interpolated date is correct for the drop in pollen concentrations around 7000 years ago, then the vegetation contributing pollen and spores to this record may have been affected by mid-Holocene warming and aridity, with standing biomass declining accordingly. The pollen record indicates no increase in mangrove pollen at this coastal site, which one might expect with a sea transgression, and we are left with the possibility that it was hot and dry at this time and biomass production was relatively low in the Laguas catchment.

In the interior upland Pago Valley core (Ward 1994), sediment accumulated much more rapidly here than in the IARII Laguas core on the west coast. This may be due to the relatively localized transport of sediments by slope wash at the Pago coring site compared with the long and winding course of the Laguas. The small number of radiocarbon assays (four, on bulk soil) and relatively wide sampling intervals in the Pago core preclude a detailed comparison with the IARII Laguas core. Nonetheless, under the linear interpolation model used by Ward for these dates, it appears that between 16.9 m and 12.5 m depth, palynomorph concentrations decline (and are sustained) in a similar pattern to that seen at Laguas. The interpolated date of the beginning of the drop in pollen and spores at Pago is around 6500 calBP, while charcoal particle concentrations begin to be calculable at around 4800 calBP.

Similar arguments to those offered for the Laguas core pertain to the Pago core. If aridity were the cause of low palynomorph inputs at the Laguas River mouth, one would expect its effects to be seen in an interior upland setting like Pago as well, and this is evident here. In line with an interpretation of marked aridity and high temperatures during the mid-Holocene, tree pollen at Pago declines began c. 7000 calBP, exhibiting the trend toward more open vegetation that one would expect during arid periods, and by 4500 calBP, palynomorphs from plants typical of today's savanna/forest mosaic vegetation dominate the Pago profile (Ward 1994:Fig. 9.B.2).

Post-mid-Holocene fires are related to seasonality and El Nino droughts

That fires burned in the southern Marianas toward the end of the mid-Holocene is strongly indicated by palaeosediment data, specifically charcoal particles c. 4860cal BP at Susupe in Saipan (and possibly earlier at Kagman sinkhole), and at Pago in interior Guam. This was the time when modern climates and associated weather patterns were becoming established in the western Pacific (see Rodbell et al. 1999; Grove and Chappell 2000; Woodroffe et al. 2003). These patterns include relatively frequent El Nino droughts that now occur every four to seven years; El Nino phases are followed by periods of heavy rains and high tides associated with La Nina phases (see Pidwirny and Vranes 2007). Could these fires have been a natural consequence of a post-mid-Holocene climate shift toward more seasonal regimes linked to periods of vigorous plant growth as well as periodic fires?

One way to answer this question is to determine whether there is evidence for fires elsewhere in the tropical Pacific at this time, well prior to human entry as established archaeologically. If so, then the Marianas case is not unique and a regional phenomenon such as more seasonal climates would be indicated.

Charcoal particles at the base of the Ngerchau palaeosediment core raised in a fallow taro field off the northeast coast of Babeldaob Island, Palau indicate that fires occurred in this vicinity by 4572 calBP, and possibly even earlier, according to Athens and Ward (2001:168). The pollen sum and the charcoal concentration (in counts expressed as x[10.sup.2] grains/cc) at this depth (641-643 cmbs) in the Ngerchau core are low, and counts and concentrations increase higher in the core, for example, at 290-292 cmbs, dated c. 4291 calBP. The earliest archaeological evidence for human presence in Palau, from Ulong Island, has been dated to c. 3000 calBP (Clark 2005:Table 1), indicating a thousand-year gap between archaeological evidence and palaeosediment observations of charcoal.

Similarly for Yap, there is a large time gap between the earliest archaeological deposits and a relatively high count of charcoal particles noted in a palaeosediment core raised in the Fool Swamp in south-central Yap Island by Dodson and Intoh (1999). Charcoal was present throughout the analyzed part of the core. The marked concentration of charcoal particles thought to indicate human forest-clearing activity was observed at 280 cmbs, estimated to date to c. 3300 BP. There are ambiguities in this inference, however. The palynomorph analysis of the Fool Swamp core ceased at 280 crabs, yet the authors obtained a date of 5230 70 years BP (uncalibrated) on a peat sample taken at 330-340 cmbs (Dodson and Intoh 1999:Table 1). Since no charcoal counts were performed along the core between 280 and 340 cmbs, it is unclear whether charcoal particles also occurred in that interval (see Fig. 4, the Fool Swamp profile, in Dodson and Intoh 1999). The earliest archaeological deposits in Yap date to c. 1800 years ago (Gifford and Gifford 1959 and see Intoh 1996). If older charcoal is present below 280 cmbs in the Fool Swamp core, then the gap between fire evidence and archaeological evidence at Yap would be even larger than is reported by Dodson and Intoh (see discussion of this issue in Rainbird 2004:78).

Nunn et al. (2001) have described a soil sedimentary sequence in the Sigatoka valley, Fiji, in which a thick lens of charcoal, apparently from a massive forest fire, was found in a deep sediment profile; the lens was radiocarbon dated to c. 5600-5050 calBP. The authors propose that the actual age range for the forest burning is more likely 5600-4650 calBP, 'well before the first unmistakable indication of the first human presence in Fiji about 2900 calBP' (950 B.C. Nunn et al. 2001:7 citing Anderson and Clark 1999). Other evidence for pre-human fires in Fiji (see Latham 1983) and elsewhere in the Pacific were noted by Nunn et al. 2001; they include accumulation of charcoal dated c. 5458-4436 calBP at Bonatoa Bog, Viti Levu (citing Southern 1986 and calibration by Spriggs 1996) and a high influx of charcoal in cores from Saint Louis Lac, New Caledonia, dated c. 4650-4350 calBP (citing Stevenson and Dodson 1995). Although not tropical, New Zealand's South Island experienced pre-human forest burning episodes dated c. 5800 and 3800 years ago (Nunn et al. 2001:7, citing Burrows and Russell 1990).

Judging from the archaeological and palaeosediment records in the Marianas, Fiji, New Caledonia and elsewhere, it appears that fire became a factor in at least some Pacific island ecosystems shortly after the mid-Holocene and in the absence of people. Nunn et al. (2001:8) suggest the Sigatoka valley fire was made possible by the slow drying out during the mid-Holocene of formerly woody environments (perhaps invigorated by favorable early Holocene conditions as suggested above), conditions which transformed the forest 'into a veritable "tinder box" of biomass', which then burned catastrophically under the severe drought conditions such as are occasionally produced during strong El Nino events.

Evidence for 5th millennium BP archaeological fires is questionable

While most of the data cited in support of an early 5th millennium BP date for human arrival in the Marianas are palaeoenvironmental, two archaeological cases have been argued as relevant (Athens and Ward 2004a:25). One of these is an age determination on a charcoal sample from a 'compacted floor-like surface' at the Achugao site in Saipan's west coast (Butler 1994: 21): 3866-3574 calBP (1 sd = 120 years, due to extra counting time for the small quantity of charcoal). The high standard deviation of this result places it well within the range of other early Pre-Latte dates and thus does not unambiguously support early 5th millennium human advent in the Marianas.

The other early find cited by Athens and Ward (2004a; also cited in Athens et al. 2004) came from the Matapang site at Tumon, on Guam's west coast. They state (Athens and Ward 2004a:25):
   In view of the paleoenvironmental evidence for a 4,300
   calBP date for the initial settlement of Guam, it is very
   interesting to reconsider an often dismissed or ignored
   archaeological radiocarbon determination obtained by
   Joyce Bath in the 1980s for the San Vitores Road Project.
   The charcoal date has a calibrated range of 4419-4150
   calBP (1 sd), which was derived from '... a dense firepit
   deposit and thus of cultural origin' (citing Bath 1986:41).

An important reason that Bath's early date at the Matapang site has been dismissed or ignored for over 20 years is the doubtful cultural status of Layer E, where the 'dense firepit deposit' was found. According to Bath's report (Bath 1986, 1987), the dated deposit in Layer E was recognized at the base of a backhoe trench, BT-1. The deposit was 'bulked out' to obtain dating material. Before removal the deposit was at least c. 20 cm wide but other dimensions and a description of the contents (e.g. whether the charcoal was coconut shell or other likely fuel) are not provided. No cultural remains were observed in Layer E in BT-1, nor were any cultural remains found in Layer E in a hand-excavated unit, EU 131, that was placed adjacent to the end of BT-1 where the dated deposit had been exposed (Bath 1987, Pt. 2: Table 12). The layer above Layer E, Layer D, contained several 'charcoal lenses or very dark sand lenses' but no 'identifiable firepits' (Bath 1986:39). Cultural materials were rare in Layer D, and no dates were obtained from this layer.

The problems of the early radiocarbon date from the 'firepit deposit' in Layer E at Matapang extend to ceramics and to the site location very close to the present shoreline, since early Pre-Latte deposits are usually found several metres inland from later prehistoric cultural deposits. This pattern of early cultural deposits farther from shore than later ones is consistent with an hypothesis of beach progradation over the last two millennia (see Amesbury 2007).

The pottery analyst D. Moore (in Bath 1986:41) found no evidence for a Pre-Latte occupation in the ceramics from any hand-excavated units at the Matapang site. She offered an alternative explanation for these facts:
   ... the Prelatte cultural deposit at Matapang may have
   been scoured and redeposited by storm waves. Certainly
   no Prelatte occupation floor was exposed during the
   excavations located on the inland, latte or beach zones of
   Matapang. Furthermore, the Prelatte sherds which were
   recovered during the excavation were scattered and often
   reduced to small fragments.

Given the sparseness of cultural materials in Layer D and the complete absence of them in Layer E, a more comprehensive and simple interpretation of the dated 'dense firepit deposit' in Layer E is that the charcoal lenses and very dark sand lenses represent natural fires that occurred in the vicinity of the strand before it was occupied during the Latte Period. The date argued to reflect very early human presence at the Matapang site may be derived from old wood, a possibility acknowledged by Athens and Ward (2004a:25).

As discussed above, charcoal particles have been observed in palaeosediment cores and other deep soil exposures in several Pacific Islands. While their archaeological significance has been suggested, these occurrences are more simply explained as the outcomes of regional environmental processes. It will be argued below that the re-organization of global climates after the mid-Holocene was a major factor favoring periodic fires.

The Post-mid-Holocene onset of modern climates brought more frequent El Nino droughts and fires

The evidence is strong for a post-mid-Holocene onset of modern climatic regimes in the Pacific (Markgraf et al. 1992; Markgraf 1993; Sandweiss et al. 1996) and for increased variability in climatic factors such as rainfall and drought after this time, including the establishment of current El Nino frequencies (Haberle and Ledru 2001). Thus, post-mid-Holocene charcoal in the Marianas palaeosediment records more economically can be explained as an indication that fires had begun to play a role in the maintenance of fire-adapted 'pyrophytic prairies' (as the Pacific savannas were aptly termed by botanist R. Porteres in 1962).

Contemporary wildfires in the southern Marianas reflect post-mid-Holocene climate, such as the fact that the southern islands are normally in drought condition for over half the year (Lander 1994). El Nino droughts occur every 4 to 7 years and are sometimes very severe. In part this causes the great variability in rainfall between years and on decadal scales seen in this region. The other part is the frequency of typhoons, which cause annual precipitation to jump far above the average. As can be seen in Fig. 5, heavy rainfall from a major typhoon in December 1990 had a major dampening effect upon wildfires in that and the following year. Regarding the effects of rainfall variability on grasslands, in long-term field ecological studies (Kaiser 2001), grasslands respond dramatically to fluctuations in rainfall, compared with deserts and forests, and it has been learned that wetter years have a much great effect on plant growth than dry spells. These are plant properties that enable grasses both to resist drought and quickly sprout new growth when well watered.

Students of Pacific climatology estimate that the Pacific El Nino/Southern Oscillation (ENSO) weather system has existed since at least the mid-Holocene and possibly much earlier; see Kilbourne (2003; Wei et al. 2007; Woodroffe et al. 2003; McGregor and Gagan 2004). Most authors agree that since at least c. 3500 years ago, modern patterns of ENSO occurrence had become established and that they are now more frequent and more variable in amplitude than during the mid-Holocene. In the western Pacific, during the dry phases (El Nino) of the ENSO system low tides and droughts prevail, and, during its wet phases (La Nina), high tides and abundant rains are characteristic. Due to consistently warm temperatures, La Nina rains enable high rates of plant growth while El Nino droughts, which can extend the normal dry season by several months, excessively dry out the plants, which shed excess stems and leaves, creating fuel for wildfires.


On Guam, recent records of El Nino droughts and wildfires in the grasslands are correlated, suggesting such a dynamic ecosystemic interaction (Fig. 5). That wildfires and climate are often related has been shown elsewhere in the tropical western Pacific (Schimel and Baker 2002). The extreme 1997-98 El Nino-associated fires in Indonesia is an example (Siegert and Hoffmann 2000) echoed in the Guam data (and see Haberle et al. 2001).

A causal mechanism for naturally occurring wildfires is unknown

While the ancient fires attested by charcoal particles in palaeosediments discussed above are clearly natural, a mechanism to initiate combustion in grassland settings in recent times has not been pursued. This may be because all wildfires are assumed to be human-initiated, either deliberately or by accident (Neill and Rea 2004). However, given the ancient, pre-human-era records of fires in the islands considered here, research aimed at solving this mystery is in order. Some natural ignition factors can be eliminated, such as lightning strikes and volcanic eruptions. In the southern Marianas, with pronounced wet and dry seasons, lightning strikes are an unlikely candidate since fires are absent when lightning can occur, i.e. during the storms in the rainy season months and during typhoons. Volcanic eruptions occur irregularly in the northern Mariana Island of Pagan, which has an active stratovolcano, and not all these events emit burning ash (Smithsonian National Museum of Natural History Global Volcanism Program 2009). Further making Pagan an unlikely ignition source for ancient southern Marianas fires is its distance from these islands; the fact that northeasterly and southwesterly wind directions in the archipelago do not favor carrying hot ash southward; and other than during its origin c. 10,000 years ago, Pagan's eruptions (dated lava flows) have occurred within the last 1,000 years (Trusdell et al. 2006). Nor are Pagan's eruptions synchronized with dry season or El Nino droughts, since fires break out on a near-annual basis.

The possibility that, in addition to mid-late-Holocene fires, Pleistocene-era fires also occurred in the Marianas cannot be discounted; however, palaeosediment records that might indicate this are not available. During the last glacial maximum, when climate in the Pacific Islands was relatively cool and dry, fires seem unlikely. Rather, a combination of high heat and dry fuel, as became typical after the mid-Holocene, appear to be necessary, but insufficient, as a cause; this is because wildfires do not always occur when these two conditions are met. Thus, the ignition mechanism that continues to operate during hot and dry weather in the southern Marianas has yet to be identified.

Concluding remarks

The recent palaeo-environmental work undertaken by Athens, Ellison, and many others in the Pacific region have increased our ability to characterize past environments, an essential part of most archaeological explanations. More critical to archaeology's effectiveness as the anthropology of the past, however, is obtaining reliable knowledge of the factors that shape cultural responses to environmental variations; in short, we need to build knowledge in the ecology of cultural systems. Thwarting this goal is applying a 'one size fits all' model of culturally mediated behavior, such as the Barrau schema and versions of it discussed above, if for no other reason than because this practice can mask interesting archaeological variation that could provoke new understandings of the human past.

An example is inattention to the early Marianas archaeological record, with its curious lack of agricultural implements and domesticated and commensal fauna, in marked contrast with the assemblages associated with Polynesian radiation into Remote Oceania. Even when such disparities are recognized, as by Athens and Ward (2004a) when eschewing the over-intensive farming aspect of Barrau's schema, at least in the initial stages, no new knowledge was gained because the preferred explanatory narrative continued to privilege palaeo-environmental data at the expense of relevant archaeological data. By providing a detailed critique and an alternative, this paper has attempted to curb this wayward trend toward reliance upon palaeosediments as if they have the same epistemological status as archaeological observations; clearly they do not.

The critique developed here has shown that the charcoal particle, pollen, and spore frequencies--taken for signs of ancient human actions affecting the landscapes of Guam and Saipan ('the data are about as unambiguous as it is possible for palaeoenvironmental data to be'--Athens and Ward 2004:25)--are indeed ambiguous and that they more clearly indicate local geological and climate-related changes in vegetation and other natural processes also taking place in the tropical western Pacific region during the Holocene. This is not to deny past human effects upon island landscapes; the ethnographic and archaeological records of Oceania amply testify to this, for example, the ecologically focused work of numerous ethnographers, linguists, demographers, and geographers who documented a rich diversity of Oceanic (sensu lato) customs and practices that are effective in helping people cope with locally variable environmental conditions. These include land reclamation, soil enrichment, mobile settlement systems, and wide demographic networks (e.g. Akimichi 1978; Brookfield and Hart 1971; Clarke 1971; Lewthwaite 1964; Rappaport 1968; for Micronesia, Alkire 1965, 1977, 1978; Carroll 1975; Marck 1986; and see Terrell 1986:180-181,264). Archaeological fieldwork has extended the dating of many of these practices far into prehistory (for summaries see Bellwood 1978; Jennings 1979; Kirch 2000; Rainbird 2004; Lilley 2006).

As all these studies make clear, human existence in islands would have been impossible without this knowledge and the skills manifested in human-altered landscapes, as most recognize (see, for example, Crumley 1994; Kirch and Hunt 1997; Balee 1998; Bayliss-Smith et al. 2003). Inadvertent landscape change, however, as implied in current anthropogenic models, appears unlikely. Furthermore, scientific standards of parsimony and goodness-of-fit require rejection of human causation of savannas as an overly complex hypothesis that depends upon unsupported assumptions about past human behavior, assumptions that have been discredited by ethnographic research. Also, an hypothesis of human causation is significantly weakened by the large time gap between archaeological evidence for human presence and the purported onset of alleged human-caused vegetation disturbance inferred from palaeosediment records. A further weakness of savanna anthropogenesis is that its independent confirmation appears elusive; simply citing as validation additional cases that conform to the pattern of early palaeosedimentary evidence of fire and vegetation shifts followed much later by archaeological evidence of human advent, as Athens et al. (2004:26) have done, is not a test. The more cases of a phenomenon that are recognized, the more comprehensive and detailed must be their explanation.

Unlike anthropogenesis based upon human behavioral inferences from palaeosediment data, the geo-climatic alternative favored here accounts for the palaeontological record more comprehensively, is in no conflict with the archaeological record, and can be evaluated along independent dimensions. Confidence in this alternative will be strengthened if confirmed through independent investigations into Pleistocene and Holocene climate histories in the relevant islands using such potential sources of palaeoclimatic information as speleothems, fossil corals, and mollusks.

Implications for future research

The propositions and discussions comprising the 'geoclimatic alternative' are offered as reasons for archaeologists to abandon the notion of anthropogenic savanna formation in the Mariana Islands, and by implication in other Pacific Islands where savannas occur but there are similar discrepancies with the archaeological record. The question of savanna origins is a subject that intersects with archaeology, due to long-standing interest in human landscape effects but is not inherently the purview of archaeology. Therefore a more comprehensive treatment of the issues raised by this debate is better left to experts in geography, botany, geology, palaeoclimatology and other pertinent scientific fields, should they be so inspired.

One issue with the potential to advance knowledge in the natural sciences implied by the geo-climatic alternative is the need to discover an environmental mechanism that initiates wildfires in locales such as the Marianas grasslands (and elsewhere where similar conditions obtain). Not only would such a finding be important for understanding the persistence of these formations, it could contribute to more effective public policy. While arson has been assumed responsible for most of the Marianas wildfires (Neill and Rea 2004), proof is lacking. Field and laboratory studies of plant physiological responses to stress from sustained high heat and aridity may hold a key to this puzzle. If, through such studies, wildfires are found to be 'natural,' as the correlation among wildfire frequency, acres burned and El Nino droughts suggests, and given that grasses have been found to be more effective than trees at preventing soil erosion in the Marianas savannas, then government efforts would be spent more profitably elsewhere than in fire suppression and tree planting in these 'pyrophytic prairies'.


Akimichi, Tomoya. 1978. Ecological aspects of Lau (Solomon Islands) ethnoichthyology. Journal of the Polynesian Society 87:301-326.

Alkire, W.H. 1965. Lamotrek Atoll and Inter-island Socioeconomic Ties. Illinois Studies in Anthropology 5, Univ. of Illinois Press, Urbana.

Alkire, W.H. 1977. An Introduction to the Peoples and Cultures of Micronesia. Second edition. Cummings Publishing Company, Menlo Park.

Alkire, W.H. 1978. Coral Islanders. AHM Publishing Corporation, Arlington Heights, Illinois.

Amesbury, J.R. 2007. Mollusk collecting and environmental change during the Prehistoric Period in the Mariana Islands. Coral Reefs 26:947-958.

Anderson, D.G., Kirk A. Maasch and Daniel H. Sandweiss (eds) 2007. Climate change and cultural dynamics: a global perspective on mid-Holocene transitions. Elsevier, New York.

Athens, J.S. 1977. Theory building and the study of evolutionary process in complex societies. In L.R. Binford (ed.) For Theory Building in Archaeology. Academic Press, New York. Pp. 353-384.

Athens, J.S. and J.V. Ward 1999. Palaeoclimate, vegetation, landscape change on Guam: The Laguas core. In B. Dixon, J.S. Athens, J.V. Ward, T. Mangieri, and T. Rieth (eds) Archaeological Inventory Survey of the Sasa Valley and Tenjo Vista Fuel Tank Farms, Piti District, Territory of Guam, Mariana Islands. Report prepared for Dept. of the Navy, Pacific Division, Naval Facilities Engineering Command, Pearl Harbor, Hawaii. International Archaeological Research Institute, Inc., Honolulu.

Athens, J.S. and J.V. Ward 2001. Palaeoenvironmental evidence for early human settlement in Palau; The Ngerchau core. In C.M. Stevenson, G. Lee, and EJ. Morin (eds) Proceedings of the Fifth International Conference on Easter Island and the Pacific. Easter Island Foundation and Bearsville Press, Los Osos. Pp. 164-177.

Athens, J.S. and J.V. Ward 2004a. Holocene vegetation, savanna origins and human settlement of Guam. Records of the Australian Museum, Supp. 29. Pp. 15-30.

Athens, J.S. and J.V. Ward 2004b. Holocene Palaeoenvironment of Saipan: Analysis of a Core from Lake Susupe. Report prepared for Division of Historic Preservation, Dept. of Community and Cultural Affairs, CNMI, Saipan. International Archaeological Research Institute, Inc., Honolulu.

Athens, J.S., M.F. Dega and J.V. Ward 2004. Austronesian colonization of the Mariana Islands: The palaeoenvironmental evidence. Indo-Pacific Prehistory Association Bulletin 24(2):23-30.

Balee, W. (ed.) 1998. Advances in Historical Ecology. Columbia University Press, New York.

Barrau, J. 1961. Subsistence Agriculture in Polynesia and Micronesia. B.P. Bishop Museum Bull. 223.

Barrau, J. c. 1966. An Ethnobotanical Guide for Anthropological Research in Malayo-Oceania (preliminary draft). UNESCO Science Cooperation Office for Southeast Asia. Pacific Collection, Micronesian Area Research Center, Univ. of Guam.

Bath, J.E. 1986. The San Vitores Road Project, Part 1: Final Report. Prepared for Maeda Pacific Corporation and Maeda Road Construction Company, Ltd., A Joint Venture. On file, Micronesian Archaeological Research Services, Guam.

Bath, J.E. 1987. The San Vitores Road project, Part 2: Final Report. Prepared for Maeda Pacific Corporation, Maeda Road Construction Co., Ltd., A Joint Venture. On file, Micronesian Archaeological Research Services, Guam.

Bayliss-Smith, T., E. Hviding and T. Whitmore 2003. Rainforest composition and histories of human disturbance in Solomon Islands. Ambio 32(5):346-352.

Bellwood, E 1978. Man's Conquest of the Pacific. Collins, Auckland.

Bird, M.I., D. Taylor and C. Hunt 2005. Palaeoenvironments of insular Southeast Asia during the Last Glacial Period: A savanna corridor in Sundaland? Quaternary Science Reviews 24:2228-2242.

Boserup, E. 1965. The Conditions of Agricultural Growth: The Economics of Agrarian Changes under Population Pressure. Chicago: Aldine-Atherton, Inc.

Bridge, Josiah and S.S. Goldich. 1948. Preliminary Report on the Bauxite Deposits of Babeldaup Island, Palau Group. U.S. Geological Survey: Tokyo.

Brookfield, H. with D. Hart. 1971. Melanesia: A geographical interpretation of an island world. Methuen, London.

Butler, B.M. 1994. Early Prehistoric settlement in the Mariana Islands: New evidence from Saipan. Man and Culture in Oceania 10:15-38.

Carroll, Vern (ed.) 1975. Pacific Atoll Populations. An East-West Center Book, Univ. of Hawaii Press, Honolulu.

Clark, G.R. 2005. A 3000-year culture sequence from Palau, western Micronesia. Asian Perspectives 44(2):349-380.

Clarke, W.C. 1971. Place and People: An Ecology of a New Guinean Community. Univ. of California Press, Berkeley.

Cleghorn, Paul L. 1998. Archaeological Investigations at Kagman Airfield, Kagman, Saipan, Commonwealth of the Northern Mariana Islands. Report prepared for U.S. Army Corps of Engineers, Pacific Ocean Division, Ft. Shafter, Hawaii.

Cole, T.G., M.C. Falanruw, C.D. MacLean, C.D. Whitesell and A.H. Ambacher. 1987. Vegetation survey of the Republic of Palau. U.S. Dept. of Agriculture Resource Bull. PSW-22.

Crumley, Carole L. (ed.) 1994. Historical Ecology: Cultural Knowledge and Changing Landscapes. School of American Research Press, Santa Fe, NM.

Cummings, L.S. 1996. Pollen analysis of a large sinkhole, Kagman Peninsula, Saipan. Palaeo Research Labs Technical Report 9639. Prepared for Pacific Legacy, Kailua, Hawaii. On file, Micronesian Archaeological Research Services, Guam.

Decker, C.S. and R. Reuveny 2005. Endogenous technological progress and the Malthusian trap: Could Simon and Boserup have saved Easter Island? Human Ecology 33(1): 119-140.

Dega, M.E and P. Cleghorn. 2003. Anthropogenic landscape change on Saipan: The sedimentary and pollen record from Kagman Peninsula, Saipan, CMNI. Draft Report Delivery Order No. 0033. Prepared for US Army Corps of Engineers, Pacific Ocean Division, Ft. Shafter, Hawaii.

Demeterio, J.L. and B. DeGuzman (eds) 1988. Proceedings of the Third International Soil Management Workshop on the Management and Utilization of Acid Soils of Oceania. College of Agriculture, Univ. of Guam.

Dickinson, W.R. 2000. Hydro-isostatic and tectonic influences on emergent Holocene palaeoshorelines in the Mariana Islands, Western Pacific Ocean. Journal of Coastal Research 16(3): 735-746.

Dickinson, W.R. 2001. Palaeoshoreline record of relative Holocene sea levels on Pacific islands. Earth-Science Reviews 55: 191-234.

Dodson, J.R. and M. Intoh 1999. Prehistory and palaeoecology of Yap, Federated States of Micronesia. Quaternary International 59:17-26.

Donnegan, J.A., S.L. Butler, W. Grabowiecki, B.A. Hiserote and D. Limtiaco. 2004. Guam's Forest Resources, 2002. U.S. Forest Service Resource Bull. PNW-RB-243.

Dye, T. and P. L. Cleghorn 1990. Prehistoric use of the interior of southern Guam. In R.L. Hunter-Anderson (Ed.), Recent Advances in Micronesian Archaeology. Micronesica Supplement 2:261-274.

Ellison, J.C. 1994. Palaeo-lake and swamp stratigraphic records of Holocene vegetation and sea-level changes, Mangaia, Cook Islands. Pacific Science 48( 1): 1 - 15.

Fairhead, J. and M. Leach. 1996. Misreading the African Landscape. Cambridge: Cambridge Univ. Press.

Falanruw, M.V.C. 1990. Traditional adaptation to natural processes of erosion and sedimentation on Yap Island. In Proceedings of the Fiji Symposium Research Needs and Applications to Reduce Erosion and Sedimentation in Tropical Steeplands, June 1990. IAHS-AISH Publication No. 192:231-237.

Falanruw, M.C., C.D. Whitesell, T.G. Cole, C.D. McLean and A.H. Ambacher. 1987. Vegetation Survey of Yap, Federated States of Micronesia. U.S. Dept. of Agriculture Resource Bull. PSW-21.

Falanruw, M.C., C.D. Whitesell, T.G. Cole, C.D. McLean and A.H. Ambacher. 1989. Vegetation Survey of Rota, Tinian, and Saipan, Commonwealth of the Northern Mariana Islands. U.S. Dept. of Agriculture Resource Bull. PSW-27.

Flenley, J.R., A.S.M. King, M.E. Prentice, J. Jackson, and C. Chew. 1991. The late Quaternary vegetational history and climatic history of Easter Island. Journal of Quaternary Science 6(2):85-115.

Fosberg, ER. 1960. The Vegetation of Micronesia. American Museum of Natural History Bull. 119.

Fosberg, ER. 1962. Tropical Pacific grasslands and savanna. In Proceedings of the Ninth Pacific Science Congress of the Pacific Science Association (1957). Volume 4, Botany. The Secretariat, Ninth Pacific Science Congress, Dept. of Science, Bangkok, Thailand. Pp. 118-123.

Gagan, M.K. (1998). Temperature and Surface-Ocean Water Balance of the Mid-Holocene Tropical Western Pacific. Science 279 (5353): 1014-1018.

Gagan, M.K., E.J. Hendy, S.G. Haberle, and W.S. Hantoro. 2004. Post-glacial evolution of the Indo-Pacific Warm Pool and E1 Nino-Southern Oscillation. Quaternary International 118119:127-143.

Gifford, E.W. and D.S. Gifford. 1959. Archaeological Investigations in Yap. Anthropological Records 18(2): 149-224. Univ. of California Press, Berkeley.

Golabi, M.H., C. lyekar and M.J. Denney. 2005a. Challenges and actions regarding the rehabilitation of degraded lands: Case study from the Pacific island of Guam. In: The proceedings of International Symposium on Land Degradation and Desertification. Uberlandia - Minas Gerais, Brazil. Sponsored by the Commission on Land Degradation and Desertification COMLAN/UGI and published by Revista Sociedade & Natureza - Special Issue - May 2005. ISSN 0103-1570.

Golabi, M. H., D. Minton, C. Iyekar, C.L. Raulerson, and J.C. Drake. 2005b. Watershed Management to Meet Water Quality Standards by Using the Vetiver System in Southern Guam. Australian Journal of Technology 9 (1):63-69 (July 2005).

Golden Gate Weather Service, San Francisco. 2009,, accessed 1-24-09.

Grove, Richard H. and John Chappell (eds) 2000. El Nino: History and Crisis, Cambridge: White Horse Press.

Haberle, S.G. and M.-P. Ledru. 2001. Correlations among charcoal records of fires from the past 16,000 years in Indonesia, Papua New Guinea, and Central and South America. Quaternary Research 55:97-104.

Haberle, S.G., G.S. Hope and S. van der Kaars. 2001. Biomass burning in Indonesia and Papua New Guinea: Natural and human induced fire events in the fossil record. Palaeogeography, Palaeoclimatology, Palaeocology 171:259268.

Haun, A.E., J.A. Jimenez and M. Kirkendall. 1999. Archaeological Investigations at Unai Chulu, Island of Tinian, Commonwealth of the Northern Mariana Islands. Paul H. Rosendahl PhD, Inc. Hilo.

Henry, J.D., A.E. Haun and M.A. Kirkendall 1999. Phase I Archaeological Survey and Subsurface Testing, U.S. Naval Activities Ordnance Annex, Guam. Prepared for Commander, Pacific Division, Naval Facilities Engineering Command, Pearl Harbor, Hawaii. Report 1775.1-120199. Hilo, Hawaii: Paul H. Rosendahl Ph.D, Inc.

Hughes, J.D. 1975. Ecology in Ancient Civilizations. Albuquerque: Univ. of New Mexico Press.

Hunter-Anderson, R.L. 1983. Yapese Settlement Patterns: An Ethnoarchaeological Approach. Monograph No. 3, Pacific Studies Institute, Agana, Guam.

Hunter-Anderson, R.L. 1989. Archaeological Investigations in the Small Boat Harbor Project Area, Agat, Guam. International Archaeological Research Institute, Inc., Honolulu.

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

Hunter-Anderson, R.L. (ed.) 1994. Archaeology in Manenggon Hills, Guam. 4 vols. Report prepared for MDI Corp., Yona, Guam: Mangilao, Guam: Micronesian Archaeological Research Services.

Hunter-Anderson, R.L. and B.M. Butler 1995. An Overview of Northern Marianas Prehistory. Micronesian Archaeological Survey Rep. 31.

Hunter-Anderson, R.L. and D.R. Moore 1999. Pots and pans in the Intermediate Pre-Latte (2500-1600 bp), Mariana Islands, Micronesia. In J.-C. Gallipaud and I. Lilley (eds) The Pacific from 5000 to 2000 BP, Colonisation and Transformations. IRD Editions, Paris.

Hunter-Anderson, R.L., G.B. Thompson and D.R. Moore 1995. Rice as a prehistoric valuable in the Mariana Islands, Micronesia. Asian Perspectives 34(1):69-89.

Intoh, M. 1996. Multi-regional contacts of prehistoric Fais Islanders in Micronesia. Indo-Pacific Prehistory Association Bulletin 15, Vol. 2, pp. 111-117.

Jennings, J.D. (ed.) 1979. The Prehistory of Polynesia. Harvard Univ. Press, Cambridge.

Johnson, C.G., R.J. Alvis and R.L. Hetzler. 1960. Military Geology of Yap. Prepared under direction of Chief of Engineers, U.S. Army. Honolulu: U.S. Army Pacific Command.

Kaiser, J. 2001. How rain pulses drive biome growth. Science 291:413-414.

Kertzer, D.I. and T.E. Fricke (eds) 1997. Anthropological Demography, Toward a New Synthesis. Univ. of Chicago Press, Chicago.

Kilbourne, Kelly H. 2003. A Coral Window on Western Tropical Pacific Climate During the Pleistocene. Unpublished Master's Thesis, Univ. of South Florida.

Kirch, P.V. 1994. The Wet and the Dry, Irrigation and Agricultural Intensification in Polynesia. Univ. of Chicago Press, Chicago.

Kirch, EV. 1996. Late Holocene human-induced modifications to a central Polynesian island ecosystem. Proc. National Academy of Science USA 93:5296-5300.

Kirch, EV. 2000. On the Road of the Winds', An Archaeological History of the Pacific Islands before European Contact. Univ. of California Press, Berkeley.

Kirch, EV. and J. Ellison. 1994. Palaeoenvironmental evidence for human colonization of remote Oceanic islands. Antiquity 68(259):310-321.

Kirch, EV. and T.L. Hunt (eds) 1997. Historical Ecology in the Pacific Islands', Prehistoric Environment and Landscape Change. Yale Univ. Press, New Haven.

Kirch, P.V., D.W. Steadman, V.L. Butler, J. Hather, and M.I. Weisler 1995. Prehistory and human ecology in Eastern Polynesia: Excavations at Tangatutau Rockshelter, Mangaia, Cook Islands. Archaeology in Oceania 30:47-65.

Kurashina, Hiro. 1986. Prehistoric settlement pattern in Guam. Paper prepared for the annual meetings of the Society for American Archaeology. April 23-26, New Orleans.

Lander, M.A. 1994. Meteorological Factors associated with Drought on Guam. Water and Energy Research Institute, Univ. of Guam Technical Report 75.

Latham, M. 1983. Origine de la formation ?t talasiga. In M. Latham and H. Brookfield (eds) Iles Fidji Orientales: etude du milieu natural, de son utilisation et de son evolution sous l'influence humaine. General Report No. 3, UNESCO/UNFPA, Man and the Biosphere Project, ORSTOM, Paris. Pp. 129-141.

Leach, Helen M. 1999. Intensification in the Pacific: A critique of the archaeological criteria and their application. Current Anthropology 40(3):311-321.

Lewthwaite, G.R. 1964; Man and land in early Tahiti: Polynesian agriculture through European eyes. Pacific Viewpoint 5:11-34.

Lilley, Ian (ed.) 2006. Archaeology of Oceania: Australia and the Pacific Islands. Blackwell Publishing, Oxford.

Mackay, A.W., R.W. Battarbee, H.J.B. Birks and E Oldfield (eds). 2005. Global change in the Holocene. Hodder Arnold, London.

Manner, H.I., D. Mueller-Dombois and M. Rapaport 1999. Terrestrial ecosystems. Chapter 8 in M. Rapaport (ed.) The Pacific Islands, Environment and Society. Bess Press, Honolulu. Pp. 93-108.

Marck, Jeff. 1986. Micronesian dialects and the overnight voyage. Journal of the Polynesian Society 95(2):253-258.

Markgraf, V. 1993. Climatic history of Central and South America since 18,000 yr BP: Comparison of pollen records and model simulations. In Wright, H.E., Jr., J.E. Kutzbach, T. Webb III, W.F. Ruddiman, EA. Street-Perrott and RJ. Bartlein (eds). Global Climates since the Last Glacial Maximum. Univ. of Minnesota Press, Minneapolis. Pp. 357-385.

Markgraf, V., J.R. Dodson, A.P. Kershaw, M.S. McGlone and N. Nicholls. 1992. Evolution of late Pleistocene and Holocene climates in the circum-South Pacific land areas. Climate Dynamics 6:193-211.

McGregor, H.V., M.K. Gagan, M.T. McCulloch and J. Chapell 2002. High resolution coral proxy records of an abrupt shift in mid-Holocene climate in the Western Pacific Warm Pool. Earth Environmental Group Annual Report 2002. 1 9.html. Accessed 7-8-05.

McGregor, H.V. and M.K. Gagan 2004. Western Pacific corel 8180 anomalous Holocene variability in the E1 Nino-Southern Oscillation. Geophysical Research Letters 31, L11204.

Mieth, A., H.-R. Bork and I. Feeser 2002. Prehistoric and recent land use effects on Poike Peninsula, Easter Island (Rapa Nui). Rapa Nui Journal 16(2):89-95.

Minton, D. 2005. Fire, erosion, and sedimentation in the Asan-Piti watershed and War in the Pacific NHR Guam. Report prepared for the National Park Service. PCSU Technical Report 150. Pacific Cooperative Studies Unit (PCSU), Department of Botany, Univ. of Hawaii. Honolulu.

Moore, D.R. 1983. Measuring Change in Marianas Pottery: The Sequence of Pottery Production at Tarague, Guam. Unpublished Master's Thesis, Univ. of Guam.

Moore, D.R. 2002. Guam's Prehistoric Pottery and its Chronological Sequence. Report prepared for Dept. of the Navy, Pacific Division, Naval Facilities Engineering Command, Pearl Harbor, Hawaii. Micronesian Archaeological Research Services, Mangilao, Guam.

Morill, C., J.T. Overpeck and J.E. Cole 2003. A synthesis of abrupt changes in the Asian summer monsoon since the last deglaciation. The Holocene 13(4):465-476.

Morrison, R.J. 1999. Soil. Chapter 5 in M. Rapaport (ed.) The Pacific Islands, Environment and Society. Bess Press, Honolulu. Pp. 56-65.

Mueller-Dombois, D. and ER. Fosberg. 1998. Vegetation of the Tropical Pacific Islands. Ecological Studies 132, Springer-Verlag, New York.

Neill, C. and J. Rea. 2004. Territory of Guam Fire Assessment. Pacific Southwest Region Fire Management, U.S. Forest Service. pdf, accessed 1-13-09.

Nunn, ED. 1991. Keimami sa vakila na liga ni Kalou (Feeling the Hand of God): Human and Non-human Impacts on Pacific Island Environments. Occasional Paper 13, Environment and Policy Institute, East-West Center, Honolulu.

Nunn, ED. 1994. Oceanic Islands. Blackwell, Oxford.

Nunn, ED., R.R. Thaman, L. Duffy, S. Finikaso, N. Ram and M. Swamy 2001. Age of a charcoal band in fluvial sediments, Keiyasi, Sigatoka Valley, Fiji: Possibly indicator of a severe drought throughout the Southwest Pacific 4500-5000 years ago. South Pacific Journal of Natural Science 19:5-10.

Pidwirny, M. and K. Vranes 2007. E1 Nino, La Nifia and the southern oscillation. In C.J. Cleveland (ed.). Encyclopedia of Earth. Environmental Information Coalition, National Council for Science and the Environment, Washington, D.C.,_La_Nifia and the so uthern_oscillation First published online in the Encyclopedia of Earth October 24, 2006, last revised October 24, 2007, accessed 3-2-09.

Porteres, R. 1962. The problem of the origin of the "savannahs" of the islands of the Pacific. In Proceedings of the Ninth Pacific Science Congress of the Pacific Science Association (1957). Volume 4, Botany. The Secretariat, Ninth Pacific Science Congress, Dept. of Science, Bangkok, Thailand. Pp. 86-88. Pratt, H.D., EL. Bruner and D.G. Berrett 1987. A Field Guide to the Birds of Hawaii and the Tropical Pacific. Princeton Univ. Press, Princeton.

Pulliam, H.R. 1996. Sources and sinks: empirical evidence and population consequences. In O.E. Rhodes, Jr., R.K. Chesser and M.H. Smith (eds). Population dynamics in ecological space and time. The University of Chicago Press, Chicago. Pp. 45-69.

Rainbird, E 2004. The Archaeology of Micronesia. Cambridge Univ. Press, Cambridge.

Rappaport, Roy A. 1968. Pigs for the Ancestors: Ritual in the Ecology of a New Guinea People. Yale University Press, New Haven.

Reinman, E 1977. An Archaeological Survey and Preliminary Test Excavations on the Island of Guam, Mariana Islands, 1965-66. Miscellaneous Publication 1, Micronesian Area Research Center, Univ. of Guam.

Roberts, N. 1998. The Holocene: An Environmental History. Blackwell Publishers Inc., Oxford.

Rodbell, D.T., G.O. Seltzer, D.M. Anderson, M.B. Abbott, D.B. Enfield, and J.H. Newman. 1999. An -15,000-Year Record of E1 Nino-Driven Alluviation in Southwestern Ecuador. Science 283 (5401):516-520.

Sanchez, P.A. 1976. Properties and Management of Soils in the Tropics. John Wiley and Sons, New York.

Sandweiss, D.H., J.B. Richardson III, E.J. Reitz, H.B. Rollins and K.A. Maasch 1996. Geoarcheological evidence from Peru for a 5000 years B.E onset of E1 Nino. Science 273:1531-1533.

Schimel, D. and D. Baker. 2002. The wildfire factor. Nature 420:29-30.

Siegert, E and A.A. Hoffmann 2000. The 1998 forest fires in East-Kalimantan (Indonesia); A quantitative evaluation using high resolution, multitemporal ERS-2 SAR images and NOAAAVHRR hotspot data. Remote Sensing of Environment 72: 64-77.

Sinclair, D.J., J.L. Banner, EW. Taylor, T.M. Quinn, J.W. Jenson and J.E. Mylroie 2008. Deglacial climate dynamics in the western Pacific Ocean measured in a stalagmite from Guam. Abstract of paper presented at Goldschmidt 2008, Vancouver, BC.

Smith, P.E.L. 1972. Changes in population pressure in archaeological explanation. World Archaeology 4(1): 5-18.

Smithsonian National Museum of Natural History Global Volcanism Program. 2009. Pagan, Index of Monthly Reports, accessed 3-7-09.

Southern, W. 1986. The Late Quaternary Environment of Fiji. Ph.D. dissertation, Australian National University.

Spoehr, A. 1957. Marianas Prehistory: Archaeological Survey and Excavations on Saipan, Tinian and Rota. Fieldiana: Anthropology 48.

Spriggs, M. 1981. Vegetable Kingdoms: Taro Irrigation and Pacific Prehistory. Unpublished Ph.D. dissertation, Australian National University, Canberra.

Steadman, David W. 2006. Extinction and Biogeography of Tropical Pacific Birds. University of Chicago Press, Chicago.

Stevens, T., H. Lu, D.S.G. Thomas and S.J. Armitage 2008. Optical dating of abrupt shifts in the late Pleistocene East Asian monsoon. Geology 36(5): 415-418.

Stocker, T.E 2000. Past and future reorganizations in the climate system. Quaternary Science Reviews 19:301-319.

Stone, G.D. and C.E. Downum 1999. Non-Boserupian ecology and agricultural risk: Ethnic politics and land control in the arid southwest. American Anthropologist 101 : 113-128.

Tainter, J. 1988. The Collapse of Complex Societies. Cambridge: Cambridge Univ. Press.

Terrell, John. 1986. Prehistory in the Pacific Islands. Cambridge University Press, Cambridge.

Topping, D.M., EM. Ogo and B.C. Dungca. 1975. Chamorro-English Dictionary. Univ. of Hawaii Press, Honolulu.

Tracey, J.I. Jr., S.O. Schlanger, J.T. Stark, E.B. Doan and H.G. May. 1964. General Geology of Guam. Prof. Paper 403-A-1 Washington, D.C: United States Government Printing Office. United States Dept. of Army.

Trusdell, EA., R.B. Moore and M.K. Sako. 2006. Preliminary Geologic Map of Mount Pagan Volcano, Pagan Island, Commonwealth of the Northern Mariana Islands U.S. Geological Survey Open-File Report 2006-1386, Version 1.0.

Tsukada, M. 1966. Late Pleistocene vegetation and climate in Taiwan (Formosa). Proceedings of the National Academy of Science USA, Pp. 543-548.

Ward, J.V. 1994. A Holocene pollen record from the Pago River Valley, Guam. In R.L. Hunter-Anderson (ed.) Archaeology in Mangenggon Hills, Yona, Guam, Vol. III, Results of Analyses. Chapter 9B, pp. 9.34-9.52. Report prepared for MDI Guam Corporation, Yona, Guam. Micronesian Archaeological Research Services, Guam.

Wei, G., W. Deng, K. Yu, X. Li, W. Sun and J. Zhao. 2007. Sea surface temperature records in the northern South China Sea from mid-Holocene coral St/Ca ratios. Palaeoceanography 22: PA3206, doe: 10.1029/2006PA001270.

Woodroffe, Colin D., Matthew R. Beech, and Michael K. Gagan. 2003. Mid-late Holocene E1 Nino variability n the equatorial Pacific from coral microatolls. Geophysical Research Letters 30(7):1358.

Young, A. 1976. Tropical Soils and Soil Survey. Cambridge Univ. Press, Cambridge.

Young, EJ. 1988. Soil Survey of Territory of Guam. U.S. Dept. of Agriculture, Soil Conservation Service: Washington, D.C.

Zan, Y. and R.L. Hunter-Anderson 1988. On the origins of the Micronesian "savannahs": An anthropological perspective. In J.L. Demeterio and B. DeGuzman (eds), Proceedings of the Third International Soil Management Workshop for the Management and Utilization of Acid Soils of Oceania, Belau, Feb. 2-6, 1987. Univ. of Guam, Mangilao. Pp. 18-27.

Dept. of Anthropology, Univ. of New Mexico, Albuquerque, NM, USA 87106. Email:
COPYRIGHT 2009 Blackwell Publishing Limited, a company of John Wiley & Sons, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hunter-Anderson, Rosalind L.
Publication:Archaeology in Oceania
Geographic Code:8MICR
Date:Oct 1, 2009
Previous Article:Archaeology of Taumako. A Polynesian outlier in the Eastern Solomon Islands.
Next Article:Another burning question: hunter-gatherer exploitation of Macrozamia spp.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters