Palaeoenvironmental evidence for human colonization of remote Oceanic islands.

Kirch, Patrick V. ; Ellison, Joanna

Not every first footstep on a virgin shore leaves enduring trace, nor every first human settlement an enduring deposit that chances to survive, and then chances to be observed archaeologically. Good environmental evidence from Mangaia Island, central East Polynesia, gives -- it is contended -- a fairer picture of the human invasion of remote Oceania than the short and sceptical chronology recently published in ANTIQUITY.

Since Willard Libby provided the first Polynesian radiocarbon date to Kenneth Emory in 1951, yielding an 'absolute' estimate of the age of initial habitation at Kuli'ou'ou rockshelter (O'ahu, Hawaiian Islands), archaeologists have continued to debate the chronology of human expansion into the islands of Remote Oceania.(1) From a wide range of archaeological, historical-linguistic and human-biological evidence, virtually all prehistorians agree that East Polynesia (including Hawai'i, New Zealand, Easter Island, the Societies, Marquesas, Tuamotus, Australs, Gambier and Cook archipelagos) was the last region within Remote Oceania to receive human settlers. However, just when this final expansion took place continues to be a matter of debate (e.g. Sinoto 1970: Irwin 1981; 1992; Kirch 1986; Sutton 1987; Hunt & Holsen 1991; Anderson 1991). This issue is not trivial, because whether the time depths of the prehistoric sequences of particular East Polynesian islands were relatively 'long' or 'short' has major implications for the processes of evolutionary divergence and social transformation (see Kirch & Green 1987).

In a recent ANTIQUITY, Spriggs & Anderson (1993) continued this debate, arguing that there has been a systematic bias towards early radiocarbon dates, and proposing to 'rein in the speculation' (1993: 201). Their argument extends a methodology of 'chronometric hygiene', previously applied by Spriggs (1989; 1990) to radiocarbon dates from island southeast Asia and Melanesia. Anderson (1991) has also used this methodology to argue for a short chronology for the human occupation of New Zealand. In brief, their method consists of setting out a 'protocol of acceptability' for 14C ages, whereby dates are accepted or rejected according to sample material, pretreatment conditions, stratigraphic context, cultural associations, and other criteria (see Spriggs & Anderson 1993: 207-8; Anderson 1991:782-3 for details). Applying chronometric hygiene to a suite of 147 dates from East Polynesian sites, Spriggs & Anderson conclude there is 'nothing to demonstrate settlement in East Polynesia earlier than AD 300-600', and then only in the Marquesas Islands (1993: 211).

Spriggs & Anderson recognize that evidence for human colonization of islands may derive not only from habitation sites, but from evidence for 'anthropogenic environmental changes, particularly forest disturbance' (1993: 210). Convinced by their hygienic cleansing of the radiocarbon corpus for habitation sites that East Polynesia was settled late, however, they dispute recent palaeoenvironmental evidence for considerably earlier human disturbance in the Hawaiian, New Zealand, and Mangaia ecosystems (1993: 210-11). In particular, they question recent reports by Kirch et al. (1991; see also Kirch et el. 1992) that pollen records derived from stratigraphic cores on Mangaia Island in the southern Cook group provide evidence for Polynesian activity by at least 1600 BP, if not earlier. Although they do not adduce evidence to support their position, they invoke 'enrichment of sampled sediments by ancient coralline carbon' and imply that 'the radiocarbon determinations from the [Mangaia] pollen core do not record accurately the period of colonization' (1993: 211).

Our aim in this paper is not to debate the methodology of 'chronometric hygiene' as applied to radiocarbon dates from habitation sites, even though we believe that some of the rejection criteria used by Spriggs & Anderson (1993) may be over-zealously applied. Rather, we address the paleoenvironmental evidence for human colonization in remote Pacific islands. We are of the opinion, evidently shared by Hunt & Holsen (1991: 158), that carefully designed paleoenvironmental research (including not only palynology, but geomorphic evidence for rates of erosion and deposition, palaeo-botanical evidence for introduced floral elements, faunal extinction sequences, etc.) may offer the best approach to determining the approximate timing of initial human colonization in Remote Oceania. To this end, we present extensive new data from Mangaia which strengthens the interpretation of human presence on the island well before any known archaeological indications (e.g., habitation sites, dated artefact-bearing contexts). We demonstrate that the radiocarbon chronology for this sequence of vegetation and geomorphic change is entirely consistent with stratigraphy, and that there is no evidence for contamination by carbonates or other sources. In contrast with Spriggs' & Anderson's claims, the Mangaian data strongly support a model of early and continuous colonization of East Polynesia, a model extensively developed by Irwin (1992). We preface our presentation of the Mangaia case with some general points concerning human impacts on the island ecosystems of remote Oceania.

Anthropogenic impacts on Remote Oceanic ecosystems: theoretical considerations

Certain conditions make the islands of the central and eastern Pacific (Green's (1991) 'Remote Oceania') particularly suitable locations in which to apply a paleoenvironmental approach to human colonization chronologies. These aspects of island ecosystems may be unfamiliar to readers of ANTIQUITY who work in continental situations. Island biogeographers and ecologists have long recognized that the isolated island ecosystems of the Pacific are marked by such characteristics as limitation in biotic diversity (especially at family levels and higher), reduced inter-species competition (in part due to the absence of large vertebrate predators), protection from outside competition, and the 'preservation of archaic, bizarre, or possibly ill-adapted forms' (Fosberg 1963a: 5). As a consequence, 'the thing that most distinguishes islands, at least oceanic islands ... is their extreme vulnerability to disturbance' (Fosberg 1963b: 559; see also Carlquist 1974; Culliney 1988; Williamson 1981: 50). Fosberg elaborates (1963a: 5) on the stability of island ecosystems prior to the arrival of humans:

It is likely that, before the advent of man, many or most of the older island ecosystems had reached such relative stability that changes were mostly very slow. In most respects organisms present had evolved into an effective equilibrium with their environments. Closed biotic communities had developed that made difficult the unaided invasion of new organisms.... It is clear enough that the arrival of man has invariably increased, to some extent, the degree of instability in these systems.

Indeed, humans were the first large terrestrial vertebrates -- other than birds and fruit bats -- to invade Remote Oceania. Moreover, the human populations that expanded beyond the Melanesian archipelagos of the Bismarcks and Solomons beginning about 3500 BP were horticulturalists who carried with them an entire 'portmanteau biota' (the term comes from Crosby (1986)) including an extensive root-tuber-tree crop complex, domestic animals (pigs, dogs, fowl), and a range of commensal or synanthropic species (e.g. the Pacific rat (Rattus exulans), various terrestrial gastropods and insects). With a horticulturally-based economy -- in which shifting cultivation involving the use of fire for forest clearance was a major component -- these colonizing groups precipitated unprecedented rates of ecological change and disturbance in the isolated and vulnerable island ecosystems they invaded. To be sure, such habitat modifications may have been spatially restricted to the catchments surrounding initial settlement localities, given relatively small founding human populations. However, the majority of inner Pacific Islands are also small in scale, so that even human populations on the order of 50-500 persons could have created significant disturbance within a century or two of first colonization.

The role of fire is especially relevant to determining the timing of initial colonization based on paleoenvironmental evidence for ecological disturbance. Spriggs & Anderson (1993: 200-201) rightly point out that in the Hawaiian Islands, volcanism must be taken into account as a probable ignition source for fires (the same is true for parts of New Zealand). But in the Hawaiian Islands, only Hawai'i and Maui have been volcanically active within the Holocene. The vast majority of remote Oceanic islands have had no volcanic activity within the Holocene. Moreover, most are situated within the humid tropics, with climax rainforest vegetation. Fire was simply not a common occurrence in these humid forests until the advent of humans, as is increasingly evident from determination of microscopic charcoal particles in sediment cores taken from Pacific Island sites (see the Mangaia case, below). Thus, in most Pacific Islands when volcanism can be ruled out as an ignition source, the appearance of substantial and sustained quantities of charcoal particles in alluvial and colluvial sediments, and of microscopic carbon particles in pollen slides, can reasonably be taken as a proxy measure of human-induced disturbance within the immediate catchment area.(2) Naturally, such evidence is greatly strengthened when matched by independent measures of vegetation change, increased rate of erosion, faunal extinctions, and so forth.

There is substantial evidence, both linguistic (French-Wright 1983) and archaeological, that early Austronesian-speaking peoples who migrated eastwards from Melanesia into Polynesia practised shifting cultivation, or 'slash-and-burn' horticulture, as a major component of their subsistence strategies. The ethno-botanist Barrau opined, 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' (1961: 18). Several decades of archaeological research on prehistoric Polynesian agriculture have generally substantiated Barrau's hypothesis (see Kirch 1991 for a summary). Such agricultural activity should predictably generate several kinds of environmental signals. First, shifting cultivation with burning will generate substantial quantities of charcoal, ranging from larger chunks of burned wood which become incorporated into colluvial sediments, to fine air- and water-borne particles that are deposited in alluvial and lacustrine sediments. Second, because new gardens are cleared from forest each year, the area covered is extensive and the disturbance to natural vegetation associations can be substantial; such changes are predictably indicated in changing pollen spectra. Such increases in charcoal influx to sediments will also be sustained. Third, on old, stable high island ecosystems, removal of the climax rainforest vegetation and exposure of the thin, organic soil cover may result in substantially increased rates of erosion, readily measurable where sedimentary sequences are preserved in depositional basins.

Several decades of inter-disciplinary collaboration between Oceanic archaeologists and natural scientists have convincingly demonstrated that prehistoric humans wrought major ecological changes and disruptions to the fragile island ecosystems of the inner Pacific (e.g. Kirch 1982: Olsen & James 1984; Spriggs 1986; Flenley et al. 1991; Steadman 1989; Bayliss-Smith et al. 1988). In this context, a research strategy focused on palaeoenvironmental indicators of human-induced disturbance is a reasonable approach to assessing the chronology for colonization of islands. Indeed, and here we must concur with Hunt & Holsen (1991: 158), such a paleoenvironmental approach strikes us as generally more efficient than attempting to locate archaeologically the first habitation sites on an island (especially an island of any size). Not only were initial settlement sites limited in number and extent, they are the most likely (by virtue of their age) to have been destroyed, eroded, or buried through later alluviation or natural shoreline progradation (see Kirch (1986) for further discussion of these issues with regard to East Polynesia). On the other hand, paleoenvironmental indicators such as charcoal and pollen can be expected to be widely distributed throughout an island environment, a continually-generated background signal of human activity. The 'chronometric hygiene' of Spriggs & Anderson relies primarily on dates from habitation sites, and distinctly downplays what they refer to as 'the currently popular tactic of searching for signs of earlier human habitation in pollen cores' (1993: 211). As we hope to demonstrate through the Mangaia case, we believe they are misguided in this viewpoint.

The Mangaia case-study

Mangaia Island, most southerly of the Cook Islands in central East Polynesia, was selected by Kirch and D.W. Steadman in 1989 as the locus for an inter-disciplinary study of human-induced changes to a remote island ecosystem. Two seasons of intensive fieldwork, involving geomorphological, palynological, palaeontological, zooarchaeological and palaeoethno;botanical approaches have been completed (preliminary results in Dawson 1990; Ellison 1994; Hather & Kirch 1991; Kirch et al. 1991; Kirch et al. 1992; Lamont 1990; Steadman & Kirch 1990). The reader is referred to these and forthcoming papers for full details of our Mangaia Project. Of specific relevance to the theme of this paper -- and to the debate on East Polynesian chronology -- is the stratigraphic and palynological evidence for human disturbance in the Mangaian ecosystem. (We stress that our evidence is not confined to the pollen record, but also includes comparative swamp stratigraphy, and geochemical analysis of the cores.) It is therefore the pollen-stratigraphic record that we summarize below, noting that full details of this work are published elsewhere (Lamont 1990; Kirch et al. 1992; Ellison 1994), and will not be repeated here.

Steadman and Kirch specifically chose Mangaia because the island's geology and geomorphology offered especially promising conditions for recovering a finely detailed environmental record for the Holocene. Geologically, Mangaia consists of a highly weathered central volcanic cone (c. 17-20 million years age as determined by K/Ar dating), with a radial drainage pattern, completely encircled by a ring of elevated limestone of reef origin (Marshall 1927; Stoddart et al. 1985). The volcanic interior is heavily degraded, covered in a terminal, pyrophytic vegetation dominated by the fern Dicranopteris linearis, by scrub Pandanus tectorius and by ironwood Casuarina equisitifolia (Merlin 1991). The limestone rim, called makatea by the Polynesians, has been weathered through solution to an escarpment along much of its interior margin. The makatea effectively blocks the radial stream drainage pattern at the lower ends of the valleys (although there is some drainage through subterranean caverns), producing a series of more-or-less closed depositional basins. In historic times, these depositional basins have been filled with swampy terrain that was intensively utilized by the Mangaians for irrigated cultivation of taro (Colocasia esculenta). These swampy depositional basins, Kirch & Steadman reasoned, would provide excellent field sites for stratigraphic coring, including pollen analysis. To this end, we invited the collaboration of J. Flenley during the 1989 field season. Assisted by F. Lamont and S. Dawson (who completed the analysis of the TIR-1 core), Flenley obtained three deep cores from the lower portions of the Veitatei Valley depositional basin. Radiocarbon dates from these cores, and a summary of the pollen analytical results from core TIR-1, were presented by Kirch et al. (1991; see also Kirch et al. 1992), who argued that dramatic vegetation changes occurring after c. 1600 BP could only have been associated with human activities.

Although the pollen analysis by Lamont (1990) and geochemical analysis by Dawson (1990) of the TIR-1 core provided compelling evidence for human presence on Mangaia by c. 1600 BP, Kirch and Steadman were concerned that evidence from a single core might be questioned by those inclined to a 'late chronology' for East Polynesian prehistory. We therefore invited J. Ellison to join our 1991 field team, to extend and amplify the stratigraphic and palynological sequence for the island. Among our 1991 aims were:

1 to extend the coring operations beyond Veitatei Valley to encompass every major drainage basin on the island;

2 to perform pollen analyses on two additional cores, as a check on the sequence outlined by Lamont (1990); and

3 to count microscopic charcoal particles in the pollen samples, a procedure not performed for the TIR-1 core.

During the 1991 season, Ellison recovered 21 cores from eight drainage basins in all parts of the island, using Livingstone and Hiller corers. Cores ranged in depth from 4.5 to 12 m, but most were in the range from 8 to 12 m, providing finely detailed stratigraphic sequences. Cores were X-rayed to check for possible disturbance (none was indicated), and 18 peat samples from 7 cores were submitted for 14C age determination. Along with the 8 samples from 1989 cores (Kirch et al. 1991), this makes a total of 26 radiocarbon dates for the Mangaia stratigraphic record.(3) Following a detailed study of core stratigraphy, two cores (TM7 from Tamarua Valley, and VT6 from Veitatei Valley, see FIGURE 1) were selected for pollen analysis. The pollen analytical work followed standard methods (Faegri & Iverson 1975), but included two important improvements over the 1990 analysis of core TIR-1: the determination of absolute microfossil concentrations (rather than simply relative abundances); and the counting of carbonized particles. Because the 1991 study is described fully by Ellison (1994), we confine ourselves to the key results relevant to the chronology of human occupation and impact on the environment of Mangaia. In particular, we address the claims of Spriggs & Anderson (1991: 211) that the Mangaian pollen record does not document early Polynesian colonization.

Stratigraphy and dating of cores

Independent of the pollen record, the Mangaian stratigraphic core sequences reveal significant, island-wide changes in the depositional regimes of the valley bottoms. Deep cores from Veitatei, Tamarua, Ivirua, and Karanga valleys all indicate a period of peat deposition associated with higher lake levels, beginning c. 7000 BP (reflecting the rapid increase in post-Pleistocene sea levels, to a central Pacific high stand of +1-1.5 m (e.g. Pirazzoli & Montaggioni 1988)). After about 4000 BP a shift occurs from lake peat to reed peat, indicating a lowering of lake levels, again probably correlated to sea-level fluctuations. However, the most radical change in depositional regime occurs in the upper portions of the cores, where the peats are rapidly replaced with massive alluvial clays, reflecting significantly heightened rates of erosion of the interior volcanic cone and in-filling of the depositional basins. This abrupt transition from peat to clay has been dated to 2980[+ or -]80 BP (Beta-47723) in core KA4, to 1930[+ or -]60 BP (Beta-52829) in core IV1, to 1830[+ or -]80 BP (Beta-47734) in core VT5, and to 1640[+ or -]80 BP in core TM7 (Beta-47727). Although some thin clay lensing occurs at earlier time periods (most probably representing El Nino-Southern Oscillation (ENSO) events), as in the TM7 core, the major peat-clay transition is dramatic and sustained. It cannot be explained by any conceivable natural phenomenon alone, such as an ENSO event or cyclonic disturbance. Thus, the stratigraphic sequences revealed by the Mangaia cores are in themselves highly suggestive of human-induced changes to the island's erosional/depositional cycle by at least 2000 BP, independent of pollen results.

The TIR-1 core, situated in the lowest portion of the Veitatei drainage where Lake Tiriara has not yet been in-filled with alluvium, was geochemically analysed by Dawson (1990; see summary in Kirch et al. 1992: figure 9). Several major and sustained geochemical changes are evidenced beginning somewhat before 1640[+ or -]50 BP (Beta-38314): an appreciable decline in organic content of the sediment; a rise in the quantity of free iron; and significant increases in the oxides Si[O.sub.2] and [Al.sub.2][O.sub.3]. These are strong proxy indications that the island's central volcanic cone, consisting of deeply weathered laterites, had been exposed to major and continuing erosion. The sustained nature of these changes strongly argues that they were not induced by short-term natural phenomena, such as ENSO events or cyclones. Thus, again independent of pollen data, the geochemical analysis of the TIR-1 core supports the interpretation that human effects on the island's environment were occurring by at least 1600 BP (Dawson 1990).

At this point it is relevant to comment on the radiocarbon corpus for the Mangaian stratigraphic cores. We have 14C-dated 26 peat samples from these cores, not a trivial number. All the determinations are in correct stratigraphic order; this corresponds with the X-ray analysis of cores indicating a lack of disturbance, and in one case very fine varve-like layering. Moreover, as we had previously demonstrated for the TIR-1 core, a regular age-depth progression is evident (Kirch et al. 1991: figure 4), as it is in the well-dated TM7 core from the west Tamarua Valley. Such extremely regular and consistent age-depth progressions provide prima facie evidence that the radio-carbon dates on Mangaia cores are not -- as feared by Spriggs & Anderson (1993: 211) -- contaminated or enriched. Indeed, there is no good geological or hydrological reason to suspect carbonate enrichment, because the Mangaian streams flow over weathered basaltic terrain, and do not come into contact with the carbonate makatea until after they have passed through the depositional basins which were the focus of our cores. Indeed, Stoddart et al. (1985: 128, table 2, figure 7) report that the Mangaian streams are 'considerably undersaturated with respect to calcium carbonate'. Further, all of the core dates are on peat samples which were thoroughly dispersed in hot acid to remove carbonates before dating; the dates are on organic vegetative matter alone.

In sum, the stratigraphic and geochemical sequences from the Mangaian cores offer strong indications of dramatic changes in the island's environment as early as 2000 BP, changes that cannot be plausibly accounted for on the basis of natural phenomena. With this background, we now turn to the palynological evidence.

Vegetation history

We will not burden the reader with extensive details of the three cores analysed for pollen content, as these and the full diagrams are available elsewhere (for core TIR-1 see Lamont 1990; Kirch et al. 1991; and Kirch et al. 1992; for cores VT6 and TM7 see Ellison 1994). It is sufficient to present here a summary version of core TM7 from Tamarua Valley, and to discuss its correlation with cores TIR-1 and VT6.

Core TM7, with a depth of 8.6 m, has a basal age of 7240[+ or -]100 BP (Beta-47730), making it the oldest dated core from the island, slightly older than TIR-1.(4) Six 14C age determinations, all in consistent age-depth progression, provide excellent chronological control. One peat sample (Beta-55630, 2480[+ or -]60 BP) was submitted after the completion of pollen analysis to date the onset of the major anthropogenic effects detected through pollen analysis, and described below.

The summary pollen diagram shows a dramatic change in core TM7 at about 2.4 m depth. Below this depth there is a total absence of charcoal particles, and the pollen spectra are dominated by forest taxa including Palmae, Malvaceae, Guettarda speciosa, Weinmannia samoensis, Sophora sp., Erythrina sp. and Hernandia sp. Spores of forest ferns such as Asplenium nidus and Cyathea sp. are also highly represented. Above 2.4 m depth (at which a distinct clay band occurs), charcoal first occurs in the core, initially at a concentration of 10,000 grains/[cm.sup.3], rising to a maximum of 300,000 grains/[cm.sup.3] at 2.0 m depth. Between 1.9 and 1.3 m depth charcoal particles range between 25,000 and 60,000 grains/[cm.sup.3], and at 1.1 m (the base of the overlying clay alluvium), the concentration is 113,000 grains/[cm.sup.3]. Correlating with this sudden and massive appearance of charcoal in the sediments are marked changes in the pollen spectra. Three key taxa, present only in limited concentrations below 2.4 m depth, display dramatic increases: the fern Dicranopteris linearis, which dominates the degraded volcanic interior today; Pandanus tectorius, also a dominant of the interior ridges, and the fern Cyclosorus interruptus. At the same time, the various forest taxa which dominated the sediments below 2.4 m depth undergo dramatic reductions in concentration; some disappear completely. We stress that these changes in pollen spectra represent real changes in vegetation composition, because they are based on absolute pollen concentrations, and not on relative changes in frequency.

We maintain that the dramatic and sustained disruptions and transformation of vegetation evidenced by core TM7 can only be explained as a result of human actions, specifically forest clearance aided by fire. That naturally ignited fires were not common in the Mangaian environment prior to the arrival of humans is indicated by the absence of charcoal particles below 2.4 m. When evidence for fire in the form of microscopic charcoal first appears, the dramatic reduction in forest tollen and attendant expansion of pyrophytic ferns and Pandanus commences.

Core VT6 (see Ellison 1994) reveals a nearly identical vegetation history. In this case, the transition occurs at 8.5 m depth, with a corresponding 14C age of 2570[+ or -]90 BP (Beta-52830). Again, charcoal particles are absent in the lower sediments, and the same changes in forest taxa and pyrophytic species occur. This replication of results from two cores in different drainage basins indicates that the changes recorded are correct, reproducible, and of island-wide scale.

Core TIR-1 (Lamont 1990) also displays essentially the same record of vegetation change, although in this case charcoal particles were not determined. It is important, however, to modify earlier statements that the onset of major vegetation changes interpreted as of anthropogenic origin occurred at c. 1600 BP, as stated in Kirch et al. (1991), and thence quoted by Spriggs & Anderson (1993). After completing her preliminary pollen analysis of the TIR-1 core, Lament divided the sequence into zones with a break at 3.5 m (upper heavy line in Kirch et al. 1992: figure 9). This break was dated at 1640[+ or -]50 BP (Beta-38314). However, an examination of the pollen diagram reveals that Lamont was conservative in her delineation of the transition from a forest zone to the anthropogenically disturbed zone; the decline in tree taxa, increases in ferns, and corresponding changes in core geochemistry actually commence slightly deeper, at about 4.0-4.2 m. A peat sample from just below this, at 4.6 m, was dated to 2450[+ or -]80 BP (Beta-33063). Thus, the onset of anthropogenic changes as reflected in the TIR-1 core can be bracketed to a period between 2450[+ or -]80 BP and 1640[+ or -]50 BP.

In sum, three stratigraphic cores from two separate drainage basins have been analysed by three palynologists (Flenley, Lamont, Ellison) working independently. All three cores have revealed essentially identical sequences of vegetation change, with a dramatic and sustained transformation commencing at points calibrated by radiocarbon ages of 2450[+ or -]80, 2480[+ or -]60 and 2570[+ or -]80 BP. These ages all overlap at a single standard deviation, and can be pooled for a mean calibrated age of 2495[+ or -]41 BP. In the opinions of the palynologists, the vegetation changes which commenced at this time, and which in the case of cores TM7 and VT6 correspond with the first appearance of charcoal particles in the stratigraphic record, can only be explained as the result of human activities. If this extensive, carefully-documented evidence does not provide a 'smoking gun' for Polynesian presence in the southern Cook Islands by c. 2500 BP, it at least shifts the burden of proof to Spriggs & Anderson to justify their claim of no evidence for human colonization in the region prior to AD 300-600.

This palaeoenvironmental evidence poses a challenge to archaeology, given that no habitation sites have yet been discovered on Mangaia dating to earlier than c. 1000 BP (Kirch et al. 1992). This absence of evident early sites should not surprise us, however, given the stratigraphic evidence for massive alluvial in-filling of the Mangaian valleys. Our cores revealed between 1 and 6 m of clays covering the old land surface (palaeosol) upon which the initial Polynesian colonizers would likely have established their habitations, adjacent to the formerly more extensive freshwater lakes. Thus, locating early archaeological sites on Mangaia will require innovative methods of sub-surface prospecting and remote sensing, a daunting task.

We also must respond to Spriggs & Anderson's comments on the Mangaian pollen record in relation to faunal evidence for avian extinctions (1993:211), the latter derived from excavations in the Tangatatau rockshelter (MAN-44) and other sites dating after c. 1000 BP. They believe it 'implausible' that a suite of bird species would have survived on the island for some centuries after human colonization. Steadman & Kirch (1990: 9607-8) specifically address this issue, noting that Mangaia differs from most other Pacific islands with known avifaunal extinction records in the presence of the extensive karst makatea (some 56% of the island's land surface). In many areas, the makatea is extremely difficult to penetrate, and would have provided a likely refuge area for many bird species. Moreover, Spriggs & Anderson's contention that in other archipelagos 'wholesale extinctions occurred more rapidly' (1993: 211) simply does not concur with recent palaeontological evidence, such as that determined by Helen James (1991) for Hawai'i, where extinctions continued to occur over the entire time-span of Polynesian occupation, and on into the post-European contact era. Thus, the Mangaia case is not substantially out-of-line with evidence from elsewhere in Polynesia.

Discussion

The Mangaian case is significant both in its demonstration of the importance of palaeoenvironmental data for assessing the chronology of human colonization of islands, and in certain specific issues of Polynesian culture history. We will discuss these in turn.

We maintain that the Mangaian case challenges Spriggs & Anderson's claim that 'the currently popular tactic of searching for signs of earlier human habitation in pollen cores is fraught with various technical and interpretational problems' (1993: 211). To be sure, palynological data must be interpreted carefully, and in consort with stratigraphic and geochemical analyses, and sequences must be backed up by extensive suites of radiocarbon dates. But there is nothing inherently suspicious in this research 'tactic.' To the contrary, and for the good theoretical reasons outlined earlier, carefully designed palaeoenvironmental research may be a more productive approach to dating the colonization of remote Pacific Islands than an elusive search for the 'first colonization site,' which becomes something of a Holy Grail.

The Mangaia palaeoenvironmental results also suggest that it may be premature to dismiss, as do Spriggs & Anderson (1993: 210), some recent palaeoenvironmental evidence from other East Polynesian archipelagos. While we are the first to ask for continued and indeed more intensive pollen analysis from Hawai'i, such indications as the seed cases of candlenut (Aleurites moluccana), a probable Polynesian floral introduction, in early deposits in Kahana Valley, O'ahu (Beggerly 1990), might suggest that more work in that locality is warranted. We would even be so bold as to suggest, following Sutton (1987), that archaeologists and palynologists not completely close their minds to the possibilities of pre-AD 1000 anthropogenic disturbances to the New Zealand vegetation, particularly in North Island. Recent and continuing work by Elliot, Flenley & Sutton (in press; see also Elliot et al. 1993) in the Northland region has yielded evidence for anthropogenic forest disturbance radiocarbon dated to c. 1400 BP. In the Society Islands, Lepofsky's (1994a; 1994b) geomorphological and palaeobotanical studies of the Opunohu Valley on Mo'orea indicate anthropogenic impacts on the valley's vegetation and erosion rates as early as c. 1300 BP -- some centuries older than the earliest dated habitation sites in the Society Islands.

It is surprising that Spriggs & Anderson (1993) take a negative position with regard to palynological evidence for initial human colonization of islands, given that Spriggs relied on precisely these kinds of data in his earlier study of Aneityum Island in Melanesia (Hope & Spriggs 1982; Spriggs 1986). Although Spriggs' archaeological survey had failed to reveal direct evidence for human presence on Aneityum during the period of Lapita expansion in Melanesia, Hope & Spriggs wisely concluded that pollen evidence for vegetation clearance could 'provide independent evidence which can provide a check on archaeological sequences' (1982: 88). A pollen sequence from Anawau Swamp -- remarkably similar to that from Mangaia in terms of the evidence for forest decline, increases in ferns and grass, and sudden appearance of abundant charcoal particles -- was interpreted as offering 'unequivocal' evidence for human presence. This was an important pioneering study in the use of palaeoenvironmental signals to date the timing of human colonization, and we believe that the methodology is as applicable in East Polynesia as in Vanuatu.

Our second major point concerns the anthropological implications of respective 'early' and 'late' chronologies for East Polynesian colonization. Spriggs & Anderson note that their proposed late chronology 'compels the retention of an apparent 1300-1600 year standstill in Pacific colonization' (1993: 211). This, in turn, necessitates a model of virtually explosive population growth and voyaging expansion within a very short time-span (c. AD 600-1000) once the 'standstill' ended and people again moved eastwards. But as Irwin (1992) and others (e.g. Kirch 1986; Terrell 1986) have pointed out, there are many compelling reasons to doubt that a pause or 'standstill' of such long duration actually occurred. Spriggs & Anderson's disingenuous suggestion that 'colonists presumably outran the push of population growth in older settled areas of West Polynesia' (1993: 211-14) does not square with what can be inferred regarding population growth rates during the immediately preceding Lapita phase.

A more carefully reasoned model for East Polynesian settlement is that proposed by Irwin (1992: 88-9, figure 30). In Irwin's construction, there was no 'systematic delay' in West Polynesia, although some time was required for continued population growth. Many voyaging attempts were needed 'before both nearer and farther central East Polynesia' were settled, and learning was involved as 'exploration adds geographical knowledge to navigational knowledge' (1992: 88). Thus, Irwin's model predicts a steady and continuing colonization process into East Polynesia, requiring as much as a millennium to complete from 'nearer East Polynesia' (e.g. the southern Cooks) to 'marginal East Polynesia' (e.g. New Zealand). In our view, this is a far more logical model than the sort of 1600-year 'standstill' followed by explosive population expansion put forward by Spriggs & Anderson.

Irwin has hypothesized dates for this continuous settlement model, suggesting that 'the first part of central East Polynesia was discovered and possibly settled by 500 BC, and probably before, and the Marquesas, which are the least accessible part of it, by around AD 0' (1992: 89). FIGURE 4 presents a modification of Irwin's log/log transformation map of the central-eastern Pacific, with time lines for human settlement based on an 'early' interpretation of available radiocarbon dates. As can be seen, Irwin's projection for colonization of the southern Cook Islands, c. 2500 BP or slightly earlier, is now remarkably well attested by the Mangaian palaeoenvironmental evidence reviewed above.

Conclusion

We suggest that the important issue of establishing the chronology of human colonization and occupation of Remote Oceania will not be resolved solely by applying 'chronometric hygiene' to the available suite of radiocarbon dates from known habitation sites. Indeed, there are good reasons to believe that for many East Polynesian islands and archipelagos no sites dating to the initial colonization phases have yet been discovered (Kirch 1986). We maintain that the issue of 'long' versus 'short' chronologies in East Polynesian still requires intensive field and laboratory investigations. Fieldwork involving inter-disciplinary collaboration that focuses on a range of palaeoenvironmental indicators, backed up by extensive suites of radiocarbon dates, needs to be carried out in other key island groups in East Polynesia. For example, carefully selected field sites in the Austral and Society Islands could provide a critical test of Irwin's model of continuous settlement, as discussed above. Likewise, extensive laboratory dating programmes (using new and improved technology such as AMS dating and bone pre-treatment protocols) should be initiated for key sites which have already been excavated, and for which there are archived collections of faunal and floral materials. Only after much more work of this kind has been undertaken will Polynesian archaeologists be able to speak with confidence about the timing of human arrival in the vast eastern Pacific.

Acknowledgements. The Mangaia Project was supported by grants from the National Geographic Society (Grant 4001-89) and by the US National Science Foundation (BNS-9020750). We thank Kent Lightfoot, Marshall Weisler, Atholl Anderson. David Steadman and an anonymous reviewer for their comments on an earlier draft of this paper.

1 By the term 'Remote Oceania' we refer to the vast central and eastern Pacific region beyond the Bismarcks and Solomons which, unlike the latter archipelagos, was not settled by humans until the mid to late Holocene (Green 1991).

2 We stress the importance of sustained presence of charcoal, because while rare natural firings may have occurred, these will normally produce only short-term peaks in the charcoal record.

3 All BP dates cited in this paper were calibrated using the CALIB program (Rev. 2.0) of Stuiver & Reimer (1986).

4 The IV1 core, of comparable age with a basal date of 7260[+ or -]80 BP (Beta-47721), has not been analysed for pollen content.

References

ANDERSON, A. 1991. The chronology of colonization in New Zealand, Antiquity 65: 767-95.

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

BAYLISS-SMITH, T., R. BEDFORD, H. BROOKFIELD & M. LATHAM. 1988. Islands, islanders, and the world. Cambridge: Cambridge University Press.

BEGGERLY, P.E.P. 1990. Kahana Valley, Hawai'i: a geomorphic artifact: a study of the inter-relationships among geomorphic structures, natural processes, and ancient Hawaiian technology, land use, and settlement patterns. (Ph.D dissertation, University of Hawaii.) Ann Arbor (MI): University Microfilms.

CARLQUIST, S. 1974. Island biology. New York (NY): Columbia University Press.

CROSBY, A.W. 1986. Ecological imperialism: the biological expansion of Europe, 900-1900. Cambridge: Cambridge University Press.

CULLINEY, J.L. 1988. Islands in a far sea: nature and man in Hawaii. San Francisco (CA): Sierra Club Books.

DAWSON, S. 1990. A chemical and mineralogical study of a sediment core from Lake Tiriara, Mangaia, southern Cook Islands with special reference to the impact of early man, (B.Sc Hens. dissertation, University of Hull.)

ELLIOT, M.B., J.R. FLENLEY & D.G. SUTTON. In press. Late Holocene palaeoecology of the Lake Tauanui Catchment, Northland, New Zealand, Journal of Paleolimnology.

ELLIOT, M.D., B. STRIEWSKI, J.R. FLENLEY & D.G. SUTTON. 1993. Date of colonization of Northland, Part I, Year 1. Report on work done in 1992. Massey University.

ELLISON, J. 1994. Paleo-lake and swamp stratigraphic records of Holocone vegetation and sea-level changes, Mangaia, Cook Islands, Pacific Science 48: 1-15.

FAEGRI, K. & J. IVERSEN. 1972. Textbook of pollen analysis. New York (NY): Hafner.

FLENLEY, J.R., A.S.M. KING, J. JACKSON, C. CHEW, J.T. TELER & M.E. PRENTICE. 1991. The late Quaternary vegetational and climatic history of Easter Island, Journal of Quaternary Science 6: 85-115.

FOSBERG, F.R. 1963a. The island ecosystem. in F.R. Fosberg (ed.), Man's place in the island ecosystem: 1-7. Honolulu (HI): Bishop Museum Press.

1963b. Disturbance in island ecosystems, in J.L. Gressitt (ed.), Pacific Basin biogeography: 557-61. Honolulu (HI): Bishop Museum Press.

FRENCH-WRIGHT, R. 1983. Proto-oceanic horticultural practices. (Unpublished MA thesis, University of Auckland, New Zealand.)

GREEN, R.C. 1991. Near and remote Oceania--disestablishing 'Melanesia' in culture history, in A. Pawley (ed.), Man and a half: essays in Pacific anthropology and ethnobiology in honour of Ralph Bulmer: 491-502. Auckland: Polynesian Society.

HATHER, J. & P.V. KIRCH. 1991. Prehistoric sweet potato (Ipomoea batatas) from Mangaia Island, central Polynesia, Antiquity 65: 887-93.

HOPE, G.S. & M.J.T. SPRIGGS. 1982. A preliminary pollen sequence from Aneityum Island, Southern Vanuatu. Bulletin of the Indo-Pacific Prehistory Association 3: 88-94.

HUNT, T.L. & R.M. HOLSEN. 1991. An early radiocarbon chronology for the Hawaiian Islands: a preliminary analysis, Asian Perspectives 30: 147-61.

IRWIN, G. 1981. How Lapita lost its pots: the question of continuity in the colonization of Polynesia, Journal of the Polynesian Society 90: 481-94.

1992. The prehistoric exploration and colonisation of the Pacific. Cambridge: Cambridge University Press.

JAMES, H. 1991. Prehistoric extinctions of Hawaiian birds: an update. Paper presented in the symposium Environmental and Landscape Change in Prehistoric Oceania. XVIIth Pacific Science Congress, Honolulu (HI).

KIRCH, P.V. 1982. The impact of the prehistoric Polynesian on the Hawaiian ecosystem, Pacific Science 36: 1-14.

1986. Rethinking East Polynesian prehistory, Journal of the Polynesian Society 95: 9-40.

1991. Polynesian agricultural systems, in P.A. Cox and S.A. Banack (ed.), Islands, plants, and Polynesians: an introduction to Polynesian ethnobotany: 113-34. Portland (OR): Dioscorides Press.

KIRCH, P.V., J.R. FLENLEY & D.W. STEADMAN. 1991. A radio-carbon chronology for human-induced environmental change on Mangaia, southern Cook Islands, Radiocarbon 33: 317-28.

KIRCH, P.V., J.R. FLENLEY, D.W. STEADMAN, F. LAMONT & S. DAWSON. 1992. Ancient environmental degradation: prehistoric human impacts to an island ecosystem: Mangaia, Central Polynesia, National Geographic Research and Exploration 8: 166-79.

KIRCH, P.V. & R.C. GREEN. 1987. History, phylogeny, and evolution in Polynesia, Current Anthropology 28: 431-56.

LAMONT, F. 1990. A 6,000 year pollen record from Mangaia, Cook Islands, South Pacific: evidence for early human impact. (B.Sc (Hons) dissertation, University of Hull.)

LEPOFSKY, D. 1994a. Prehistoric human-induced ecosystem changes and agricultural production in the Opunohu Valley, Society Islands. Paper presented at the Annual Meeting of the Society for Ethnobiology.

1994b. Prehistoric agriculture in the Society Islands. (Ph.D dissertation, University of California, Berkeley (CA).)

MARSHALL, P. 1927. Geology of Mangaia. Honolulu (HI): Bernice P. Bishop Museum. Bulletin 36.

MERLIN, M.D. 1991. Woody vegetation on the raised coral limestone island of Mangaia, southern Cook Islands, Pacific Science 45: 131-51.

OLSON, S.L. & H.F. JAMES. 1984. The role of the Polynesians in the extinction of the avifauna of the Hawaiian Islands, in P.S. Martin & R.G. Klein (ed.), Quaternary extinctions: 768-80. Tucson (AZ): University of Arizona Press.

PIRAZZOLI, P.A. & L.F. MONTAGGIONI. 1988. Holocene sea-level changes in French Polynesia, Paleogeography, Paleoclimatology, Paleoecology 68: 153-75.

SINOTO, Y.H. 1970. An arclaeologically-based assessment of the Marquesas Islands as a dispersal center in East Polynesia. in R.C. Green & M. Kelly (ed.), Studies in Oceanic culture history 1: 105-32. Honolulu (HI): Bernice P. Bishop Museum. Pacific Anthropological Records 11.

SPRIGGS, M. 1986. Landscape, land use and political transformation in southern Melanesia. in P.V. Kirch (ed.), Island Societies: archaeological approaches to evolution and transformation: 6-19. Cambridge: Cambridge University Press.

1989. The dating of the island Southeast Asian Neolithic: an attempt at chronometric hygiene and linguistic correlation, Antiquity 63: 587-613.

1990. Dating Lapita: another view, in M. Spriggs (ed.), Lapita design, form and composition: proceedings of the Lapita Design Workshop, Canberra, 1988: 6-27. Canberra: Department of Prehistory, Australian National University. Occasional Papers in Prehistory 19.

SPRIGGS, M. & A. ANDERSON. 1993. Late colonization of East Polynesia. Antiquity 67: 200-17.

STEADMAN, D.W. 1989. Extinction of birds in Eastern Polynesia: a review of the record, and comparisons with other Pacific Island groups, Journal of Archaeological Science 16: 177-205.

STEADMAN, D.W. & P.V. KIRCH. 1990. Prehistoric extinction of birds on Mangaia. Cook Islands, Polynesia, Proceedings of the National Academy of Sciences 87: 9605-9.

STODDART, D.R., T. SPENCER & T.P. SCOFFIN. 1985. Reef growth and karst erosion on Mangaia, Cook Islands: a reinterpretation, Zeitschrift fur Geomorphologie 57: 121-40.

STUIVER, M. & P. REIMER. 1986. A computer program for radiocarbon age calibration, in M. Stuiver & R.S. Kra (ed.), Proceedings of the 12th International 14C Conference, Radiocarbon 28(2B): 1022-30.

SUTTON, D.G. 1987. A paradigmatic shift in Polynesian pre-history: implications for New Zealand, New Zealand Journal of Archaeology 9: 135-55.

TERRELL, J. 1986. Prehistory in the Pacific islands. Cambridge: Cambridge University Press.

WILLIAMSON, M. 1981. Island populations. Oxford: Oxford University Press.