GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 265 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

12

The Ideal Free Distribution, Food Production, and the Colonization of Oceania

Douglas Kennett, Atholl Anderson, and Bruce Winterhalder

Islands in Oceania were some of the last habitable habitats, stimulating dispersal, and ultimately mi- land masses on earth to be colonized by humans. grations to more distant islands in Oceania. Current archaeological evidence suggests that these islands were colonized episodically rather than The processes involved in the development continuously, and that bursts of migration were of food production worldwide during the last followed by longer periods of sedentism and popu- 10,000 years were complex and spatially vari- lation growth. The decision to colonize isolated, able. At a minimum, they involved some combi- unoccupied islands and archipelagos was complex nation of the following set of factors: (1) the and dependent on a variety of social, technological expansion of diet-breadth during the late Pleis- and environmental variables. In this chapter we tocene and early Holocene, leading to the develop an integrative, multivariate approach to development of co-evolutionary relationships island colonization in Oceania based on a model between humans and potential domesticates from behavioral ecology known as the Ideal Free (Richards et al. 2001; Rindos 1984; Stiner et al. Distribution. This ecological model provides a 1999, 2000; Winterhalder and Goland 1997); framework that considers the dynamic character of (2) intensified exploitation of wild plants and an- island suitability along with density-dependent imals by some prehistoric foragers (Henry and density-independent variables influencing mi- 1989); (3) translocation of wild plants and ani- gratory behavior. Unique among existing models, mals by foraging groups and the management it can account for the episodic nature of certain as- or cultivation of these wild species in some in- pects of the colonization process. Within this con- stances (Piperno and Pearsall 1998); (4) the ini- text we critically evaluate the role of foraging, low- tial domestication of plants and animals in sev- level food production, and ultimately intensive eral independent centers (Cowan and Watson food production, as important contextual variables 1992; Price and Gebauer 1995a; Smith 1998); that influenced decisions to disperse. We argue (5) the adoption of these plants and animals by that intensive food production was one variable foragers living in adjacent regions, often in dif- that contributed to decreasing suitability of island ferent habitats; (6) subsequent experimentation

265 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 266 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

FIGURE 12.1. Map of Oceania showing the relevant islands and archipelagos.

leading to a reliance on food production or the sion of anatomically modern humans into pre- stability of mixed subsistence strategies (low- viously unoccupied territory, most notably the level food production; Smith 2001a; Tucker, this colonization of ever smaller and more remote volume); ( 7) continued transmission of new islands in the Pacific, Mediterranean, and and improved domesticates through exchange Caribbean (Diamond and Bellwood 2003; Kee- networks (Hastorf 1999), and (8) the ultimate gan and Diamond 1987; Kirch 2000; Kirch and emergence of more intensive food production Green 2001; Patton 1996; but see Anderson in certain locations (Smith 1998). Some of the 2003a). consequences of food production included lo- In this chapter we explore the dispersal of calized population growth, the spread of do- people into Oceania and the role that food pro- mesticated plants and animals along with agrar- duction may have played in this complex social ian knowledge and technology through and ecological process. In particular, we are in- exchange networks, the actual migration of food terested in the migration of people onto islands producers into regions occupied by foraging in Near and Remote Oceania (Figure 12.1).1 populations (Cavalli-Sforza 1996; Diamond and Near Oceania consists of several large islands Bellwood 2003), and a general increase in hu- in the Bismarck Archipelago, positioned man impacts on natural ecosystems (Bellwood 100–200 km to the northwest coast of New 2001; Diamond and Bellwood 2003; Redman Guinea, and the Solomon Islands, a series of 1999). The demographic expansion of farming smaller islands that stretch to the southeast. populations is linked to significant cultural, lin- Prior to sea-level rise during the late Pleis- guistic, and biological changes (Bellwood tocene and early Holocene, the Solomons 2001). It has also been argued that the emer- formed a single larger island known as Greater gence of food production fostered the expan- Bougainville. Vanuatu and New Caledonia

266 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 267 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

form the western boundary of Remote Oceania, particular islands—factors like position, size, which also includes 38 major archipelagos of and the distance between pairs along likely 344 colonized islands in West and East Polyne- routes of colonization. In this view, the likeli- sia (Kirch 1984). West Polynesia encompasses hood that an island will be colonized decreases the larger, aggregated archipelagos of Tonga with distance, as does the possibility of follow-up and Samoa, plus some smaller archipelagos. assistance once an island is occupied. However, In its early prehistory, Fiji is also regarded as colonization of distant islands may be promoted West Polynesian. Except for New Zealand, is- by configurational effects. For instance, archi- lands in East Polynesia tend to be smaller and pelagos consisting of larger aggregations of more dispersed. islands potentially provide greater resource di- All of the islands in Near Oceania lie within versity for colonists compared with individual the tropics, but several islands in Remote Ocea- islands. Island size also influences the probabil- nia are subtropical or are positioned farther to ity of successful colonization because larger is- the south, and have temperate climates; for in- lands offer a greater quantity and diversity of stance New Zealand lies between 35 to 45Њ habitats and resources. south. Little seasonality in rainfall or tempera- Although physical and geometric properties ture occurs close to the equator, but cooler tem- are important for understanding island colo- peratures and distinctive wet and dry seasons nization, purely biogeographical models have become more common to the south (Spriggs shortcomings. Based on the geometry of posi- 1997; Anderson 2001a). The initial colonization tion, distance, and size, they highlight the prob- of Remote Oceania involved a sixfold increase abilistic elements of “blindly” reaching a partic- in minimum voyaging distances over those at- ular island and surviving there. They do not tained in Near Oceania (200 km) and distances help to analyze the reasons for initiating migra- of up to 3700 km were crossed to reach New tion, nor the intentional or unintentional conse- Zealand and Hawaii. quences of settlement for an island’s resource The study of island colonization has a long potential, and thus for the long-term persist- history with a large body of literature developed ence of settlement. Although likely to be impor- during the last 30 years (Fitzpatrick 2004; tant, such factors are extraneous to biogeo- Keegan and Diamond 1987). Much of this re- graphic models. search was stimulated by MacArthur and Wil- In Oceania, explanations for island coloniza- son’s 1967 book entitled The Theory of Island Bio- tion can be grouped into push or pull models. geography, and by the recognition that islands Most push models invoke demographic pres- provide a well-bounded context for studying cul- sure as the primary causal force initiating dis- tural evolutionary processes. In the late 1980s, persal (Clark and Terrell 1978; Anderson 1996). Keegan and Diamond (1987) synthesized the It has been argued that population levels on literature on island colonization in various parts islands generally increase with agricultural in- of the world and concluded that biogeographi- tensification, and eventually the population ex- cal principles, particularly their physical and ceeds carrying capacity, stimulating segments geometrical properties, provided a useful of the population to move to adjacent islands. framework for understanding the colonization Pull models often propose a rapid dispersal of process. They argued that climatic, geological, people through Oceanic island chains, as op- and oceanographic differences among islands portunistic foragers skim off the highest-quality shaped their terrestrial and marine productivity resources (Clark and Terrell 1978; Anderson and influenced the ability of humans to colo- 1996; Davidson and Leach 2001) and quickly nize them. Superimposed on these ecological move on to the next propitious location. Al- qualities are geometric properties influencing though an improvement, the combination of bio- the likelihood that seafaring migrants will reach geographic patterning and push-pull variables

ideal free distribution and oceania 267 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 268 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

+ East Polynesia 35,000 BP S.E Asia Lapita Marginal South Polynesia W. Polynesia

h

g

i

HighH

w

o

Low L

_ Colonizing Mobility Colonizing

30,000 29,000 4000 3000 2000 1000 Years BP FIGURE 12.2. Colonization mobility in Oceania during the last 35,000 years (see Anderson 2001a).

does not capture the episodic nature of migratory increased the overall carrying capacity of many behavior evident amongst humans and other ani- remote island habitats. This particular point is mals (Diamond 1977), and evident in the archae- set within a more general argument: low level ological data from Oceania (Anderson 2001b; and ultimately more intensive food production Figure 12.2). Island colonization in Oceania also was one of several variables including population appears to be a dual phase process. Each episode growth, dynamic impacts of exploitation on frag- seems to have a sedentary phase, perhaps repre- ile island environments, technological develop- senting a period of population growth, and a ment, and the inherent ecological suitability of phase of high mobility and rapid dispersal. The various island groups in Oceania. The ability of speed of colonization during these unstable human behavioral ecology (HBE) to integrate episodes does not suggest incremental demo- multiple contextual variables with an emphasis graphic pressure, but is more reminiscent of on behavioral responses to changing ecological rapid dispersal, triggered by opportunistic forag- conditions make it an ideal framework to ex- ing behavior. It appears that a variety of contextual plore the causes and consequences of human variables are at work; colonization of Oceania can- dispersal onto remote islands in Oceania. not be explained by invoking single variables. In this chapter we develop an integrative, IDEAL FREE DISTRIBUTION multivariate model for the colonization of Ocea- AND HUMAN MIGRATIONS nia within the behavioral ecology framework of the ideal free distribution (Abrahams and Healey Two consequences of food production were lo- 1990; Fretwell and Lucas 1970; Sutherland calized population increase and the dispersion 1996). This set of ideas provides a simple frame- of agricultural populations into areas occupied work that considers the dynamic character of by hunter-gatherers or regions not previously habitat suitability along with density-dependent populated (Bellwood 2001; Diamond and Bell- and density-independent variables that might wood 2003). The migration of animals or people influence dispersion and habitat selection by into new habitats often entails a series of complex colonists in Oceania. In particular, we examine behavioral responses to changing social and envi- how food production might have influenced de- ronmental conditions. Both density-dependent cisions to disperse and colonize remote is- and density-independent influences may stim- lands. We argue that low-level food production ulate migration. Ultimately, an individual’s or (Smith 2001a), and later intensive food produc- group’s decision to migrate depends on the cost tion, contributed to more rapid decreases in and likely success of relocating, and on the habitat suitability through degradation, but also overall size, quality, and productivity of a home-

268 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 269 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

land region relative to alternatives elsewhere. other forms of despotic behavior (e.g., cannibal- The overall productivity and suitability of a re- ism; Kantner 1999) also create social instability gion can change for a variety of environmental and may stimulate the dispersal of people well and social reasons. Short- and long-term climatic below the actual carrying capacity of a habitat changes may alter the distribution and availabil- (Kennett and Kennett 2000). The net result of ity of subsistence resources, as will changing these despotic behaviors is reduced habitat suit- subsistence practices and technology such as for- ability leading to more rapid emigration. aging or food production, fluctuations in the density of the human occupants, habitat degra- THE IDEAL FREE DISTRIBUTION (IFD) dation due to unsustainable exploitation, and The IFD model provides an explanatory changes in social cohesion or conflicts. Dense framework for predicting when individuals populations often deplete resources rapidly, but will disperse or migrate to a new habitat based low density use can affect the availability and dis- on density-dependent changes in the suitability tribution of other plant and animal species on of the habitats available to them. Habitats are which a population depends (Sutherland 1996). ranked by their quality, as assessed by the fit- Resource exploitation also can entail mutualistic ness of the initial occupant. Typically, fitness- and beneficial relationships between humans, related measures such as production of young animals, and plants, for instance by enhancing or rate of food intake are used to measure qual- the suitability and resource richness of an envi- ity or suitability (see Winterhalder and Kennett, ronment (Rindos 1984). Chapter 1, this volume; Figure 12.3). Quality is Another consequence of larger populations density dependent and declining with increas- is competition that lowers the overall suitability ing population density due to competition. of a resource patch or habitat. Conspecifics de- Competitors may use up resources directly, for plete shared resources; interference can result instance by occupying living sites or by consum- from fighting, stealing, or control of resources ing and depleting food resources, or they may or patches by individuals (Sutherland 1996). indirectly make resources harder to find or cap- Territorial or despotic behavior by individuals ture; for instance, by stimulating their dispersal or groups may also affect the suitability of a re- or elevated wariness; or render them less desir- gion and can stimulate the movement of peo- able by contaminating or fighting over them. ple into adjacent, less desirable areas. Warfare The former is known as depletion competition, and despotic behavior would be expected in re- the latter as interference competition. Suther- gions that were environmentally or socially cir- land (1996, 9) gives this example: “drinking a cumscribed (Carneiro 1970, 1978, 1988). In this pub dry would be depletion whilst a crowd context, the largest populations would aggregate around the bar hindering access would be inter- in the most productive locations. Population in- ference.” In the case of the subsistence transi- creases, from endogenous growth or in-migra- tion from foraging to food production, depletion tion, and community fission would result in the would include density-dependent decreases in infilling of more marginal zones. In some in- game animals; interference would encompass stances this would result in environmental increases in erosion and the depletion of soil nu- “packing” and decreases in habitat suitability trients associated with more intensive land use. (Binford 1968, 1983). In addition, the territories For IFD purposes, a habitat is defined partly of some groups may extend well beyond their by scale—it is larger than the multiple patches immediate needs, thus forcing disenfranchised that would be encountered in a single foraging individuals to colonize more marginal habitats. trip and equal to or smaller than the whole range Contests for smaller, more circumscribed sec- available to a group—and partly by economic tions of arable land would predictably become characteristics—it is a relatively homogeneous more frequent within this context. Warfare and zone of production with respect to the resources

ideal free distribution and oceania 269 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 270 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

Best Habitat H1 Best Habitat + A + B B H2

H3

Worst Habitat SUITABILITY SUITABILITY

A - - Worst Habitat

0 A B 0

+ - POPULATION DENSITY + - POPULATION DENSITY

FIGURE 12.3. A) Ideal free distribution (after Fretwell and Lucas 1970, 24; Sutherland 1996, 5); B) Allee’s principle (after Fretwell and Lucas 1970, 25; Sutherland 1996, 11).

available and a characteristic method of extract- and that they are free or unrestricted in their ing them. Key to this definition: we expect each movement to effect that choice. They are com- habitat to be characterized by a unique curve petitors of equal ability and access to resources. representing its suitability, as a function of in- Under these conditions, habitat distribution creasing population density and exploitation. will work out in the following manner. Colo- For instance, habitats in Oceania would include nizing individuals will locate first in the best inland (initially forested), riverine, and coast habitat available. With increasing density due (e.g. rocky shoreline, beach and reef). This defi- to immigration or to in situ growth suitability nition recognizes that we may wish to analyze there drops. When it is diminished to the qual- population distribution over contemporaneous, ity level of the second ranked habitat, further but spatially separate habitats. Or, the IFD can population growth stimulates immigration and also be used to compare temporal shifts among populations will become apportioned between the habitats that emerge on the same landscape them. Because each individual is ready to relo- if, for instance, climate change significantly al- cate if another habitat offers an edge in suit- ters its resource characteristics or a technological ability, the population distribution will equalize innovation provides a new means of exploiting marginal qualities across all occupied habitats. the resources found there. The former would al- This is an equilibrium distribution, a conse- low for the analysis of how a population distrib- quence of the marginal equalization of habitat utes itself over a set of extant habitats; the latter suitability. At the IFD no individual has an in- would analyze choice with respect to “move- centive to relocate. ment” to changing environmental or technologi- The IFD makes two general predictions: cal possibilities. Combinations are possible, and (1) the extant population distribution over settlement of Oceania probably entailed both available habitats will reflect an equilibrium possibilities. In our very general IFD application that equalizes their marginal suitabilities; and, we model two habitats defined by environment; (2) the chronological sequence of habitat occu- small versus large islands, and two defined by pation and use; changing densities in a particular shifting between modes of resource extraction; habitat; changes in the variety of habitats foraging versus food production (see below). occupied will follow the pattern predicted by a We assume that individuals will elect to re- particular form of the IFD curves (see Sutherland side in the ideal or best habitat available to them, 1996, 1–14). In each case, empirical tests of

270 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 271 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

quantitative predictions provide stronger results overall survival rate may be low because of the than qualitative assessments. Our application difficulties associated with finding mates or focuses mainly on a corollary of (2): continuous problems with inbreeding depression (Allee change in overall population size will result, by et al. 1949; Sutherland 1996). Increasing density IFD predictions, in a process of habitat settle- improves the suitability of the habitat for subse- ment and migration with important discontinu- quent arrivals. Likewise, the subsistence system ous properties. In particular, migration from may be affected by positive economies of scale, larger islands should be more episodic than from where scale is determined by density. A growing smaller, and migration from agricultural popula- population might increase the density of desir- tions more discontinuous than from foraging able resources by more completely maintaining populations. Our evidence is meager and mainly forest cover in early stages of succession. It qualitative, but the effort is interesting because might facilitate technological improvements, no other model of comparable generality makes from seed distribution to irrigation. The suit- predictions consistent with the episodic charac- ability of marginal areas might be improved ter of human occupation of Pacific islands. once colonized, as forests were cleared and fields were prepared through plowing and ter- THE DESPOTIC VARIANT racing. Greater density may also offer protec- The ideal despotic distribution (IDD) is a variant tion from intruders or enemies. of the IFD highlighting differential access to re- Figure 12.3b shows the distributional conse- sources. If interference arises among competi- quences of the Allee effect. As before, initial tors of unequal abilities, or if by establishing settlement populates the highest ranked habi- territories, initial or superior competitors can tat, A. Whether from in situ growth or external protect themselves from density dependent immigration, increasing density eventually habitat deterioration by successfully defending spills over into habitat B. However, because the better resource opportunities, then the inferior suitability of habitat B increases with each addi- competitors and those without territories are tion, it draws population from habitat A, reduc- pushed to poorer habitats. Compared to the IFD, ing density there. If the apex of the suitabil- a despotic distribution will equilibrate with dis- ity/density curve for habitat B is higher than proportionate numbers or densities in the lower- that for habitat A, and B is sufficiently spa- ranked habitats. This makes intuitive sense: by cious, the Allee effect conceivably will empty garnering disproportionate resources in the best habitat A for a period, as individuals seeking a habitats, the better competitors push their infe- more salubrious habitat quickly migrate to this riors into habitats of lesser suitability. Because of new and improving zone. Habitat B would this, the use of lower-ranked resource patches show a rapid increase in population; the de- has been documented as a buffering strategy cline in habitat A could be quite dramatic. With among a variety of bird species (Brown 1969; an Allee effect, individuals might abandon ar- Meire and Kuyken 1984; Moser 1988). In fact, in eas that previously provided adequate payoffs, a empirical studies the ideal free distribution pattern consistent with settlement unconfor- sometimes serves as a null hypothesis to meas- mities evident in several parts of the world as ure the effects of interference competition and agricultural populations replaced or subsumed unequal resource access (Sutherland 1996). hunter-gatherer populations (Bellwood 2001; Renfrew and Boyle 2000). THE ALLEE EFFECT VARIANT Whatever form it takes, the IFD shows how There also may be density-dependent effects an incremental quantitative change in one vari- within habitats that make their suitability in- able such as population density or habitat suit- crease over some range of increasing population ability may lead to qualitative changes in another; density. At very low population densities the the range of habitats occupied; their relative

ideal free distribution and oceania 271 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 272 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

settlement densities. Moving from the IFD to or Samoa and the Cook Islands, and in addition the IDD and Allee variants portends qualitative there were several very long passages to the changes of increasing magnitude. As with most marginal islands of Hawaii, Easter, and New HBE models, there are few limits on what kinds Zealand. Clearly the costs of relocation in these of variables one might accommodate in the IFD cases, especially to the margins, must have been and its variants. For instance, climate change higher than was common within the main is- might shift the relative suitability (vertical posi- land band. How much higher depends in part tion, thus relative ranking) of the curves. Habi- on what view is adopted of maritime technol- tats or subsistence practices highly susceptible ogy. If Polynesian vessels and navigation were to density dependent degradation will have of a high order of capability (Finney 1979; Irwin steep downward slopes; those which generally 1992; Lewis 1994), then relocation costs to the are not so sensitive to population density will margins must have been substantially lower have more shallow slopes. Economies of scale than if the technology was relatively undevel- in subsistence practice may cause the slope of oped (Anderson 2000a). On the other hand, the the curve to be positive over certain ranges of issue is also a perceptual one. Groups consider- density. The consequences of territoriality, so- ing relocation in circumstances where the pre- cial inequality, and economic exploitation for ferred choice was to sail off into the unknown dispersion and habitat use can be represented had no way of estimating the relative cost of in despotic versions of the model. Manipulation reaching a new island, and it is possible that, of these and other elements can be used to gen- within the initial period of dispersal, the per- erate hypotheses about population distribution ceived relocation cost flattened to virtual invari- and migration based on a wide range of poten- ance everywhere beyond the islands of an al- tially causal conditions. ready-settled archipelago. Of course, when The IFD model does not explicitly include voyaging occurred after the period of initial dis- the cost of relocation, assuming that this is neg- persal, by which time some sense of Oceanic ligible when compared to the benefits of opti- geography may have developed, then relocation mizing long-term habitat choice. This simplifi- costs could more reliably be factored into cation is expedient for analytical purposes, but choices. An additional issue is the degree to it may seem an especially unrealistic assump- which relocation might have generally coin- tion in the case of initial spread over broad ex- cided, or was not prompted by, periods of wind panses of the Pacific. The issue here, however, reversal on a millennial scale; in which case a is relative rather than absolute costs of reloca- high frequency of downwind sailing would also tion which can be thought to constitute a con- have reduced the relative costs of relocation tinuum broadly divisible in two. First, most of (Anderson, n.d.b). These considerations are, as the islands in Remote Oceania, like those in yet, inadequately researched, but relative in- Near Oceania, lie at distances which could be significance of relocation costs can be assumed covered in a week or less of sailing, and it can in the interim for the sake of exposition here. also be argued (Anderson 2003b) that pre- As will become evident in the following sec- dictability of finding new islands was quite high tions, the islands of Oceania were colonized within the main island band of the tropical over a long period of time by people using a south Pacific. Given the same sailing technol- wide range of subsistence practices, from forag- ogy and a choice of favorable sailing conditions, ing to intensive food production. The size and the relative cost of relocation was small. Second productivity of these islands also varies greatly, however, there were some passages that were from the large, more tightly clustered islands of unusually long within the main band of Remote Near Oceania to the smaller, more dispersed is- Oceanic islands, as between Vanuatu and Fiji, lands of Remote Oceania. As a starting point for

272 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 273 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

Foraging Food Production + +

Large Island

Large Island SUITABILITY SUITABILITY

Small Island - - Small Island

0 0

+ - POPULATION DENSITY + - POPULATION DENSITY

FIGURE 12.4. Hypothetical IFD models for mobile foraging and food production on small and large islands.

generating predictions, we distinguish between more likely to be mobile, and the pace of colo- IFD curves for foragers and food producers, and nization likely to be more regular and more de- those for large and small islands (Figure 12.4). pendent upon island size. Food producers are The hunter-gatherer curves are negatively likely to be less mobile, and the pace of colo- sloped reflecting population-dependent impacts nization more episodic. This is because the on habitat productivity. On small islands they Allee effect creates a period of “stickiness” to a decline steeply and approach the x-axis at fairly new settlement, in which things are actually im- low population sizes; on large islands they de- proving with increases in population size, cline more gradually and approach the x-axis negating any tendency for significant emigra- at significantly larger population sizes. By con- tion in a new round of colonization for some trast, the food-producer curves evidence an Allee time period. That delay will be longer the larger effect due to economy of scale and peak, before the island. Regardless of island size and mode of declining. On small islands the peak is fairly production, one of the most robust predictions narrow and the plunge to the x-axis occurs at low of this model is that if habitats vary in terms of population size; on large islands the peak is high average rewards, then migratory behavior is ex- and broad and the approach to the x-axis at quite pected to be episodic rather than continuous. large population sizes. Consistent with the ob- Because this appears to be the pattern of migra- servation that the earliest forms of food produc- tion in Oceania, a pattern not easily accommo- tion were probably less efficient than foraging, dated in alternative explanations, we believe the at its best, we have ranked the initial suitability IFD may have explanatory merit. Our argument of food-production “habitats” below those for requires that we turn to Oceania, first to the foraging in the same environment whether large specifics of island colonization and second to or small island. We do not depict despotic be- temporal changes in subsistence practices. havior leading to resource defense, but it would put pressure on weaker competitors to emigrate COLONIZATION OF OCEANIA by promoting more rapid plunges towards the x- axis, thus stimulating rapid dispersal. Archaeological evidence shows that the first The proposed set of IFD models allows us to pulse of colonization into Oceania was initiated more precisely predict episodes of colonization during the Pleistocene Epoch between 35,000 as a function of population density and mode of and 29,000 years ago (see Figure 12.2).2 production. Specifically, hunter-gatherers are These early colonists likely departed from the

ideal free distribution and oceania 273 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 274 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

northeastern coast of New Guinea to occupy the islands in the Bismarck archipelago except New Britain and New Ireland, the largest islands for Manus. This island is ϳ230 km from New in the Bismarck Archipelago (Allen et al. 1988; Ireland and required a blind crossing of 60–90 Allen et al. 1989). Adjacent islands exhibiting km (Spriggs 1997). However, it is unclear how evidence for Pleistocene age settlement include early in the Pleistocene Manus was colonized Buka, on the northern end of the Solomon Is- due to the absence of datable material at the land chain, and Manus, positioned 200 km base of Pamwak, the only Pleistocene-age site northeast of New Ireland (Fredrickson et al. known on the island (Fredericksen et al. 1993). 1993; Wickler and Spriggs 1988). During much The island of Buka, in the Solomon archipelago of the Pleistocene, sea level was substantially to the south, was also colonized early (29,000 lower than today and New Guinea was con- B.P.; Wickler and Spriggs 1988) and required a nected to Australia forming a super-continent partially blind crossing of 175 km from New known as Sahul. Anatomically modern humans Ireland. At this time, Buka was the northern colonized this landmass from Southeast Asia by extent of a single island known as “Greater at least 40,000 years ago (Allen 1994; Allen and Bougainville” that extended southeast through Holdaway 1995) and perhaps as early as 55,000 the modern day Solomon Island chain. There is B.P. (Roberts et al. 1994; Thorne et al. 1999). currently no evidence for colonization of Vanu- The crossing from Southeast Asia (also a land- atu or New Caledonia, or other more remote mass less insular and more extended by lowered archipelagos, until after 3300 B.P. (Anderson sea levels, known as Sunda) to Sahul suggests et al. 2001a). that people had some rudimentary seafaring The next significant episode of island colo- technology at this early time (Anderson 2000a; nization in Oceania appears to be associated Clark 1991; Erlandson 2001). with the expansion of agricultural populations Colonization of the Bismarck archipelago south from Taiwan through the islands of and adjacent islands by hunter-gatherers de- Southeast Asia between 5000 and 4000 years pended on the appropriate maritime technolo- ago. This spread of people is inferred from the gies and knowledge (Irwin 1992). Given the close widespread distribution of archaeological as- geographic proximity of these islands to New semblages containing Asian domesticates—the Guinea, however, relatively simple boats, even ubiquitous rice, pigs, and dog; red-slipped or bamboo rafts, some of which occur naturally after paddle impressed pottery; ground stone adzes; floods, could have been used (Anderson 2000a). distinctive shell and bone ornaments; and bark- It is very unlikely that sails were employed as cloth beaters (Bellwood 1975, 1978, 1985, 1996, Horridge (1987) has proposed, because sailing 2001). Biological and linquistic data suggest technology worldwide has a mid-Holocene ori- that the changes evident archaeologically were gin, including in China, one of the more likely related to an expansion of Austronesian-speaking sources of Pacific technology (McGrail 2003). people that ultimately replaced or swamped the The accessibility and suitability of some islands existing hunter-gatherer populations on each was also enhanced by lower sea-level stands in island (Diamond and Bellwood 2003). The the Pleistocene. Between 35,000 and 29,000 appearance of ceramics in archaeological as- years ago sea level was 40 and 70 m below cur- semblages in the Bismarcks at 3300 years ago rent levels (Thorne and Raymond 1989). New has also been interpreted as an extension of this Britain, New Ireland, and Manus were larger, expansion (Bellwood 1978; Shutler and Marck but the water gaps between New Guinea and 1975; Spriggs 1990)—an intrusion into island the Bismarcks were essentially the same. New Melanesia of new people with different subsis- Britain is visible across the 90 km gap sepa- tence regimes, settlement practices, and sociopo- rating it from New Guinea and there is two litical organization, known as the Lapita cultural way intervisibility (Ͻ100 km) between all of complex in Melanesia and Polynesia. However,

274 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 275 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

other scholars interpret the appearance of Lapita a 200–400 year lag between the initial settle- as the product of in situ developments within Is- ment of Near Oceania and the first dispersal of land Melanesia (Allen and White 1989; Gosden people into portions of Remote Oceania. It also 1992; Gosden and Specht 1991; Terrell 1989; appears that colonization accelerated from west White et al. 1988; see below). to east and that the persistence of the Lapita cul- Lapita-age settlements of 3300 to 2300 B.P. tural complex was more fleeting in Remote are identified by the presence of dentate-stamped Oceania (Anderson 2001a). When the overall or incised pottery (Anderson et al. 2001a; Kirch number of sites is considered in Near and Re- 1997). A recent inventory of archaeological de- mote Oceania, settlements were established posits containing dentate-stamped pottery in- much more rapidly in Remote Oceania (22–27 cludes approximately 184 locations extending per century) when compared to Near Oceania 4500 km from the Bismarck Archipelago, (6–10 sites per century) (see Anderson 2001a southeast to Fiji, Tonga, Samoa, and Wallis in and Anderson et al. 2001a).3 the South Pacific (Anderson et al. 2001a; see Recent archaeological studies suggest a Figure 12.1). The appearance of Lapita pottery long pause of perhaps 1500 years between the east of the Solomon Islands represents the earli- expansion of Lapita peoples into West Polyne- est known colonization of Remote Oceania sia and the colonization of more remote is- (Kirch and Hunt 1988). Sites throughout the lands and archipelagos in East Polynesia Lapita range are most commonly found in (Anderson 2002; Anderson and Sinoto 2002). coastal contexts with overall densities on larger New work in the Society Islands suggests colo- islands being lower than smaller islands (Ander- nization no earlier than 1000 B.P. (Anderson son 2001a). If the frequency of Lapita sites is et al. 1999) and the current data from the Mar- standardized using land area by island size, the quesas indicate early settlement dating to number of Lapita sites in Near Oceania (1.74 around 900 B.P. (Rolett and Conte 1995; Rolett per 1000 km2) is virtually identical to Remote 1998). On the remote fringes of East Polyne- Oceania (1.69 per 1000 km2), but within Remote sia, Easter Island was likely colonized by 1000 Oceania the overall density of sites increases B.P. (Steadman et al. 1994), Hawaii at about from west to east (1.24 per 1000 km2 in Vanuatu/ the same time (Athens et al. 1999), and New New Caledonia to 2.45 per 1000 km2 in Fiji and Zealand (Anderson 1991, Higham et al. 1999) West Polynesia; Anderson 2001a). along with several other south Polynesian islands Although there are similar densities of (Anderson and White 2001; Anderson and Lapita sites in Near and Remote Oceania, the O’Regan 2000; Johnson 1995) by 800 B.P. We chronological range of these sites throughout currently are reevaluating the early settlement the region is quite different. Recent dates from history of Rapa in the Austral Islands, but it the Bismarck Archipelago suggest the appear- is likely that this remote island was settled ance of Lapita pottery at around 3300 B.P. with no earlier than 800 B.P. (Kennett et al. 2003; the cessation of production occurring as late as Walczak 2001). Colonization of East Polynesia 2000 B.P. (Anderson 2001a; Green and Anson also includes the 25 “mystery islands” (e.g., 2000; Specht and Gosden 1997; Torrence and Christmas, Norfolk, and Pitcairn Islands), col- Stevenson 2000). Dates range between 3100 onized after 1000 B.P., then abandoned before and 2600 in the Reef/Santa Cruz Archipelago; European contact (Anderson 2001b, Anderson 3000 to 2700 B.P. for Vanuatu (Bedford et al. et al. 2002; Bellwood 1978). Thus, the data in- 1998; Spriggs 1997); 3000 to 2700 for New dicate that the colonization of East Polynesia Caledonia (Sand 1997, 1999; 2000); 2900 to was late and, despite vast geographic expanse, 2600 B.P. for Fiji (Anderson and Clark 1999); relatively rapid. and 2850 to 2650 B.P. for West Polynesia These new settlement data do not support the (Tonga, Samoa and Wallis). These data suggest traditional view that eastward voyaging and

ideal free distribution and oceania 275 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 276 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

colonization were relatively continuous once of subsistence strategies and the emergence of they were initiated by Lapita peoples (Kirch 1997, food production in Oceania. 2000; Irwin 1992). The idea that colonization was continuous has been supported by de- SUBSISTENCE CHANGE AND THE creases in indigenous tree pollen and increases ORIGINS OF FOOD PRODUCTION in charcoal sediments in a paleoenvironmental sequence collected from Mangaia in the Cook The first people to colonize Near Oceania be- Islands interpreted as anthropogenic forest tween 35,000 and 29,000 years ago were prob- clearance starting as early as 2500 B.P. (Kirch ably relatively mobile hunter-gatherers whose 1996, 1997, 2000). Similar types of sequences subsistence strategies were structured by the have been used to argue for early settlement on availability and distribution of wild foods. Ter- Easter Island (1500 to 1200 B.P.; Flenley 1996; restrial fauna were restricted to a narrow range Parkes 1997; Parkes and Flenley 1990) and of edible species that included lizards, snakes, New Zealand (2000 to 1500 B.P.; Sutton 1987). rats, bats, and birds (Spriggs 1997), but coastal However, recent studies in New Zealand sug- habitats like reefs and lagoons offered a rich ar- gest that ancient soil carbons are often washed ray of marine foods (Flannery 1995). The tropi- into lake sediments and return anomalously old cal forests covering these islands provided little radiocarbon dates (McGlone and Wilmshurst in the way of edible plant foods, but disturbed 1999). Therefore, the lake core chronologies patches along rivers and coastlines, particularly from Mangaia and elsewhere are suspect and wetlands, would have afforded staples familiar should be carefully reevaluated. Furthermore, to early colonists (taro-Colocasia esculata; and there is often a dichotomy between lake core sago palm, Metroxylon) (Spriggs 1997; Yen 1985, data suggesting anthropogenic landscape modi- 1995). Forest disturbance and habitat manipu- fication and the earliest tangible archaeological lation during early settlement would have pro- evidence for colonization. For instance, Kirch moted the growth of these wild plants (Groube (1996) excavated seven rock shelter sites on 1989; Yen 1995). Mangaia and has a well-established chronology Evidence for settlement in Near Oceania (40 radiocarbon dates) for human activities on prior to 20,000 years ago is relatively scarce. the island starting at approximately 1000 B.P. Only four sites can clearly be assigned to this Interestingly, the earliest levels of the Tangatatau early period (Spriggs 1997). Three of these sites Rock shelter site (ϳ1000 B.P.) contained the are pericoastal rock shelters or caves that con- bones of several extinct landbirds and these tain stratigraphic evidence for sporadic use dur- species do not appear in more recent strata at ing this early period (Allen 1991; Gosden and this or other sites on the island, suggesting Robertson 1991). Simple flaked stone tool as- the prehistoric extirpation of these species semblages at these sites also suggest high mo- (Steadman and Kirch 1990). The presence and bility. Marine shells and fishbone are the most rapid extirpation of large landbirds (ϳ200–300 common constituents in these deposits. At the years after colonization; see below) is consistent sites of Buang Meraback and Matenkupkum with other early settlement sites in East Polynesia (New Ireland), the large size of shells suggests (Anderson 2002). Based on these data we would that the early inhabitants of this cave were for- argue that the deposits at Tangatatau Rock shel- aging in relatively pristine shell beds (Spriggs ter represent some of the earliest settlement 1997). The fishbone found in the lowest levels of on Mangaia. Again, this is consistent with new Matenkupkum is also some of the earliest evi- evidence from elsewhere in East Polynesia dence for marine fishing in the world (ϳ35,000 (e.g., Anderson and Sinoto 2002) and our central B.P.; Allen 1993). Useware and starch grains on thesis that colonization of the Pacific was stone tools from Kilu Cave on Buka Island episodic. We now turn to a more detailed analysis (Solomons) suggest that maritime foraging was

276 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 277 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

coupled with plant exploitation, including the ued exploitation of near-shore reef habitats but use of Colocasia taro, a plant that later became there is little evidence for intensified fishing an important agricultural staple in Melanesia after 10,000 B.P. A small number of shark teeth and Polynesia (Loy et al. 1992). The importance found in Holocene deposits (8400 B.P.) at the of plants to these early colonists is also sug- Balof 2 rock shelter in New Ireland provide gested by the open air site of Yambon, posi- weak evidence for fishing beyond the reef tioned in the forested interior of New Britain (White et al. 1991). There is substantial evi- (Pavlides 1993; Pavlides and Gosden 1994). dence for deliberate movement of wild animals Changes in settlement and subsistence in from New Guinea to the Bismarcks and Solomon Near Oceania after 20,000 years ago are sig- Islands during the late Pleistocene, followed by naled by a break in the occupation of pericoastal more intensive hunting of these animals during rock-shelters. The caves and rock-shelters used the terminal Pleistocene and early Holocene after 20,000 B.P. are positioned 3–5 km from (Gosden 1995; Flannery and White 1991). Evi- the coast at a slightly higher elevation (Spriggs dence from several rock shelter sites indicate 1997). Marine resource exploitation continued, the late Pleistocene introduction of a possum but changes in shellfish assemblages are evi- species (Phalanger orientalis) to New Ireland dent, particularly between sites occupied during from New Guinea (Flannery 1995). A different the late Pleistocene and early Holocene. Some possum species (Spilocuscus kramer) was intro- of these changes are easily attributed to habitat duced to the more remote island of Manus changes coincident with sea level rise or the for- (Bismarck Archipelago) and evidence for the in- mation of coastal estuaries associated with the troduction of the bandicoot (Echymipera kalubu) stabilization of sea level between 8000 and and a small wallaby (Thylogale browni) comes 6000 B.P. (Spriggs 1997). Other changes in from caves in New Ireland (Flannery 1995). In marine shell assemblages are attributed to in- addition, a rat species native to New Guinea tensified human exploitation strategies, their (Rattus praeto) has been discovered in 13,000- impact on intertidal resources, and subsequent year-old deposits at the site of Panakiwuk, New adjustments of foraging strategies. For in- Ireland (Flannery 1995). What appears to be the stance, when the Holocene (10,000 to 2000 BP) deliberate translocation of animal species to assemblages at Matenkupkum (New Ireland) Near Oceania suggests that people may have are compared to the pre-20,000 deposits, a been compensating for population density- much wider range of small and large species are dependent resource depletion with the intro- evident and the quantities of the largest species duction of new species (Anderson 2001a). are significantly reduced (Gosden and Robert- A similar pattern is evident in the floral son 1991). A similar pattern is evident at Buang records from early sites in Near Oceania. The Merabak where an increased variety of smaller presence of native almond seeds (Canarium shell taxa are evident after 10,800 BP (Balean indicum) in late Pleistocene and early Holocene 1989; cited in Spriggs 1997). These patterns are sites in Near Oceania indicate the transplanta- consistent with heavy exploitation and depletion tion of this species from New Guinea where it of high ranking resources and the consequent is naturally widespread (Yen 1990), although expansion of diet breadth, also documented by whether by people or the flying fox, remains un- Anderson (1981, 1983) in New Zealand, certain (Anderson n.d.a). Seeds in terminal Broughton in late prehistoric Northern California Pleistocene (13,000 to 10,000 B.P.) deposits at (Broughton 1999), Janetski in the Great Basin Pamwak (New Ireland) and Kilu (Solomons) (Janetski 1997), and others (Butler 2000; might suggest early arboriculture, in essence Nagaoka 2001, 2002). the deliberate planting, tending, and harvesting The fishbone collected from late Pleistocene of trees (Spriggs 1997). Evidence for short-term and early Holocene deposits suggests contin- forest clearance during the Early Holocene is

ideal free distribution and oceania 277 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 278 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

visible in pollen records from New Ireland strategy with some degree of localized intensifi- (Allen et al. 1989). Colocasia taro residues are cation and sedentarization during the Holocene evident on tools from late Pleistocene and early (Spriggs 1997). Holocene deposits and yam (Dioscorea) residues By contrast, substantial changes in settlement are present on two tools from the Holocene lev- and subsistence are evident in Island Melanesia els at Balof 2 (New Ireland; Barton and White after 3300 B.P. (Kirch 1997; Summerhayes 1993). This indicates continued low-level use of 2000a). Pottery, including dentate stamped starchy plants from the Pleistocene into the Lapita forms, appears in the record for the Holocene. first time (Kirch 1997, 2000; Spriggs 1997; There is also some evidence in Island Summerhayes 2000a). New settlements were Melanesia for the intensification of subsistence established on small, offshore islands and stilt strategies during the Holocene. Pollen and houses were sometimes constructed over coastal charcoal evidence from several locations indi- lagoons. These communities were commonly cate short-term local clearance of forest and larger than previously occupied residential bases continuous, low-level burning in some areas. (Anderson 2001a). Accelerated erosion evident Shell and bone accumulations increased during in the geomorphological records surrounding the Holocene at the sites of Pamwak on Manus these settlements is consistent with intensified and formal artifact types, including shell and gardening activities on adjacent hillslopes. The edge-ground stone axes, become more common first undisputed appearance of domesticated an- (Spriggs 1997). A similar pattern is evident at imal bone of pig, chicken, and dog also occurs in the sites of Matenbek (New Ireland) where these deposits (Spriggs 1997). The changes evi- larger amounts of imported obsidian are symp- dent in the archaeological record are interpreted tomatic of intensified trade (Gosden 1995). by some as a culmination of intensified subsis- Heightened exploitation of possum is also evi- tence practices in Island Melanesia during the dent at several locations during the Holocene Holocene (Allen and White 1989; Gosden 1992; possibly providing the foundation for more sta- Gosden and Specht 1991; Terrell 1989; White et ble, sedentary settlement at some inland loca- al. 1988). Others (Green 1991; Kirch 1997, tions (Marshall and Allen 1991; White et al. 2000; Spriggs 1997) hypothesize that these 1991). Several cave sites were also abandoned changes represent an influx of people and cul- between 6000 and 5000 B.P. perhaps signal- ture from island Southeast Asia, broadly linked ing reduced mobility and the consolidation of with the Austronesian expansion out of China populations on the landscape. Wood structures (Bellwood 2001, Diamond and Bellwood 2003). found in the waterlogged deposits at the Apalo This hypothesis is supported by a variety of ar- site (Arawe Islands) suggest a settlement stabil- chaeological, linguistic, and genetic evidence ity at some locations between 4250 and 4050 (Bellwood 2001; Kirch 1997; Spriggs 1997); B.P. (Gosden and Web 1994). These deposits however it is likely that cultural developments also contained a large variety and quantity of during this time were complex involving intru- seeds from trees (Canarium, coconuts, and oth- sion, accommodation, integration, and innova- ers; Hayes 1992, cited in Spriggs 1997). Whether tion (Green 1991). that suggests only an accumulation of shoreline Spriggs (1997; Jones and Spriggs 2002) has flotsam in the waterlogged site (Matthews and described the early Lapita populations in Near Gosden 1997), systematic collection of naturally Oceania as “full blown” agriculturalists, but the grown resources, or more formal arboriculture importance of food production is debatable still remains uncertain. The predominant sub- based on the available data—particularly during sistence-settlement pattern evident in the the early stages of this cultural period. Lapita age records from across Near Oceania from 20,000 sites (3300 to 2200 B.P.) in Near Oceania are to 3300 B.P. is that of a mobile hunter-forager commonly positioned on the coast near lagoons

278 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 279 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

or close to natural openings in island-fringing the hypothesis that Lapita peoples had a well- reefs. Shells and fish bones preserved in these developed food production economy (Spriggs sites suggest the exploitation of fish and shellfish 1997). It is true that terrestrial resources on from a variety of marine habitats: coastal lagoons, islands east and south of the main Solomons are reefs, and open ocean. Artifact assemblages con- depauperate compared with the islands of tain fishhooks, lures, and netweights, the tech- Melanesia that were colonized during the Pleis- nology of relatively sophisticated fishing practices tocene. Indigenous land mammals are absent (Summerhayes 2000a). Food production is in- except for several species of bat. Large flightless ferred from geomorphological studies showing birds were common on many of these more re- increased erosion rates near Lapita age settle- mote islands, but the overall diversity of avifauna ments (Spriggs 1997), presumably associated was reduced. Plant diversity is also restricted in with deforestation and field preparation for taro remote Oceania, and wild taro (Colocasia or Cyr- and yams in hillside gardens, but such data are tosperma), yams (Dioscorea), and bananas are ab- difficult to interpret because forest clearance sent (Green 1991; Spriggs 1997; van Balgooy can occur in the absence of food production 1971). With respect to Vanuatu, Spriggs (1997, (e.g., clearing for settlement; Anderson 1995, 41) argued that, in the absence of domesticated 2000b). Beyond these data, building a case for plants and animals, human settlement may intensive food production is difficult due to have been impossible, from which it was in- preservation problems in tropical environ- ferred that “transported landscapes” composed ments. Waterlogged sites in Island Melanesia— of taro, bananas, native almonds, breadfruit, Mussau and Arawes Islands (Kirch 1989; Gosden pigs, chickens, and dogs carried in the canoes 1992)—provide some evidence for the utiliza- of early colonists were vital and allowed them tion of nuts, and possibly of arboriculture based to replicate their homeland economies (Kirch on the presence of Canarium and coconuts, but 1997). evidence for taro, yams, breadfruit, and bananas The importance of food production to are absent. Forest clearance and a form of these early colonizing populations in Remote “swidden” horticulture may be indicated by in- Oceania is largely hypothetical and not se- creased charcoal concentrations in sediment curely demonstrated (Anderson 2000b; n.d.a; cores after 3000 B.P. Changes in pollen assem- Anderson and Clark 1999, Burley et al. 2001; blages also indicate a decline in tree taxa and the Clark and Anderson 2001). Lapita settlements Lapita adze kit may have been used to ring trees in Remote Oceania are generally positioned on and clear forest. Pig, chicken, and dog bones are old beach surfaces close to fringing reefs and la- all found in Lapita faunal assemblages, but all goonal environments that provided a wide domestic animal bone occurs in much lower pro- range of marine resources and clearly supplied portions relative to undomesticated species— the bulk of the diet (Kirch 1988; Burley 1998; particularly fish. For instance, at the waterlogged Galipaud 1996). The size of these communities site of Talepakemalai eighteen pig bones were in Remote Oceania generally is smaller than identified in a vertebrate assemblage consisting equivalent sites in Near Oceania and settlement of 14,148 bones (Kirch 1997). mobility appears to have been relatively high Several scholars have argued that the colo- (Anderson 2001a; Clark 1999; Bedford et al. nization of more remote archipelagoes in the 1998, 1999). Similarly to Lapita assemblages in Pacific—Vanuatu, New Caledonia, Fiji, Tonga, Near Oceania, marine vertebrates and inverte- and Samoa—was significantly dependent on brates are common. Diverse kinds of fishing agriculture (Kirch and Green 2001; Spriggs tackle, such as fishhooks and lures, point to the 1997). Indeed, evidence for rapid colonization importance of maritime subsistence activities of more remote islands between 3200 and 2700 (Kirch 1997). Early Lapita deposits also contain B.P. is used as one line of evidence supporting terrestrial birds and reptiles that rapidly went

ideal free distribution and oceania 279 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 280 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

extinct (Anderson 2002). Early colonists in Van- 1997), charcoal levels in sedimentary cores in- uatu focused on harvesting naturally occurring creased after about 3000 B.P. and the diversity of foods that resulted in localized resource deple- tree pollen was reduced. However, large-scale de- tion and frequent site relocation (Bedford et al. forestation is not evident in these records until 1998). In New Caledonia, the early animals tar- 2500 B.P. (Stevenson 1999; Stevenson and Dod- geted and extirpated included a large megapode son 1995). A similar pattern is evident at several (Sylviornus neocaledoniae), a crocodile (Mekosuchus locations within the Fijian archipelago, some- inexpectus), and a horned tortoise (Meiolania sp.) times coincident with evidence for increasing (Balouet and Olsen 1989; Balouet 1987; Sand erosion (Anderson 2002). 1996a, 1997). Similar faunal extinctions are evi- Therefore, the case for early Lapita food pro- dent in Fiji where there was another giant duction rests solely upon linguistic reconstruc- megapode (Megavitiornis altirostris) and another tion (Kirch and Green 2001). Linguists have genus and species of mekosuchid crocodile traced the root of modern Polynesian languages (Volia athollandersoni; Anderson et al., 2001b, back to a proto-Oceanic language. Proto-Oceanic Molnar et al. 2002, Worthy 2000; Worthy et al. words for many of the domesticated plants, in- 1999). Likewise in Tonga, there were large, now cluding taro, yam, banana, and breadfruit, and extinct, iguanids (Pregill and Dye 1989; Stead- for aspects of the swidden agricultural system man 1993). Animal extinctions were a product both suggest that food production has a long his- of direct hunting and habitat destruction, but all tory in Oceania (Kirch 1997). These linguistic of the known species seem to have gone extinct data, although intriguing, do not carry sufficient before there could have been any serious com- chronological precision, however, to validate the petition with introduced domestic and non- idea that early Lapita colonists practiced food domesticated (e.g., rats) animals. production and carried domesticated plants and Only chicken bone appears in reasonably animals during the first push into Remote Ocea- early Lapita contexts in Remote Oceania, and nia. Evidence for the movement of obsidian, even there it might not have arrived with the first adzes, and sometimes pottery (Burley and Dick- settlers (Steadman et al. 2002). Carbonized plant inson 2001; Green and Kirch 1997; Weisler and remains are rare in Remote Oceanic deposits, Kirch 1996; Weisler and Woodhead 1995) over and no cultigens are known from early Lapita vast areas during the Lapita Period shows that contexts. The presence of garden snails (Lamel- the full suite of Oceanic domesticates could have laxis gracilis) could provide early evidence for become established in Remote Oceania through food production because it is a species that was continued contact with populations in Island probably translocated to Remote Oceania with Melanesia following the initial colonization of taro planting stocks and associated soils (Kirch more remote islands. 1997). However, there is no evidence to show The importance of food production for early how early that might have occurred in the Lapita Lapita colonists is far from resolved and further era. In addition, recently acquired and more field study will be required to test several alter- firmly dated pollen and charcoal records from is- native propositions. We offer the following lands in Remote Oceania also suggest that food testable hypotheses in lieu of a solid statement production played only a minor, if any, role ini- regarding the importance of food production to tially (Anderson 2002). These newer records are early Lapita populations. To start, due to the va- starting to show a lag of up to 500 years between garies of the archaeological record we cannot the appearance of burning and forest clearing ac- completely rule out the hypothesis that early tivities for cultivation and the accelerated defor- Lapita populations transported domesticated estation indicative of more intensive food pro- plants and animals to more remote islands in duction. In New Caledonia, where early Lapita Oceania as a package. This remains an alterna- sites date to between 3000 and 2900 B.P. (Sand tive hypothesis, although we find no compelling

280 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 281 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

archaeological, linquistic, or biological evidence pottery across the landscape suggests the ex- in its support at this time. Alternatively, we pres- pansion of people into interior locations as late ent two additional hypotheses. The first, favored as 1000 years ago. Rapid increases in charcoal by Anderson (n.d.a), is that the earliest Lapita frequencies also occur in a core from the Rewa colonists were effectively foragers who skimmed delta on the southeast coast of Viti Levu, Fiji, by the highest ranked marine and terrestrial re- about 2300 B.P. (Anderson 2002). This is con- sources as they dispersed through Vanuatu, sistent with the post Lapita phase record from New Caledonia, Fiji, and into Western Polynesia Fiji indicating a depletion of easily gathered nat- (Anderson, n.d.a). The implication of this hypoth- ural foods and an increased reliance on agricul- esis is that early Lapita colonization outran the ture between 2300 and 1900 B.P. (Clark 1999). movement of most food production into Remote Inland areas started to be colonized late in the Oceania and that domesticates, other than the Lapita period, when settlements covered the chicken, were introduced into the Lapita econ- landscape by 1000 B.P. Group conflict over ter- omy during the continuing migration process, ritory is indicated by the establishment of forti- probably in a piecemeal fashion, rather than as a fied villages on the landscape by 1200 B.P. (Field package (Anderson n.d.a). The second hypothe- 2004). In New Caledonia, intensified agricul- sis, favored by Kennett, is that the early Lapita tural strategies are inferred from new sediment economy combined a low-level food production core data suggesting deforestation after about package, as defined by Smith (2001a), of select 2500 B.P. (Stevenson 1999; Stevenson and domesticates: chicken and possibly taro, mar- Dodson 1995), and population expansion into itime foraging for shellfish and fish, and the ex- interior areas at 2000 B.P. that culminated ploitation of the most easily obtainable terres- around 1000 B.P. (Sand 1996b; Galipaud 1996). trial foods (Canarium nuts, large birds, eggs, The first fortifications on the Loyalty Islands at etc.). This mixed production strategy may have 1800 B.P. are attributed to infilling of the land- been similar to the low-level food production scape and territoriality (Sand 1996b). This con- practiced by the Mikea of western Madagascar text is more consistent with the despotic variant (Tucker 2001, this volume). In both proposi- of IFD and would have stimulated emigration tions, foraging for wild foods was the most im- more rapidly. portant strategy initially, but food production in- The evidence for rapid colonization of East creased in importance as people impacted the Polynesia (ϳ1000 B.P.), if correct, suggests that availability of wild resources, and the abundance migrations were initiated from West Polynesia of easy prey diminished on each island. in the context of: (1) increasing population den- Current archaeological data suggest a pause sity, (2) decreases in habitat suitability caused in colonization activities once Lapita settlements by erosion and soil degradation after an initial were established in the Fijian, Tongan, and increase in habitat suitability for agriculture Samoan archipelagos. After this time (ϳ3000 due to forest clearance and terracing; and (3) B.P.), settlement pattern data for these island heightened interference due to territoriality and groups suggest: (1) increased number of settle- warfare. Once colonized, subsistence and settle- ments in coastal locations and other previously ment strategies varied between islands in East unoccupied islands within each archipelago; Polynesia, but early colonists generally com- (2) reductions in settlement mobility (Clark bined the hunting of ecologically naïve flight- 1999); (3) the expansion of populations into the less birds, large reptiles, and other easy prey, interiors of larger islands (Clark 1999; Hunt with maritime foraging/fishing and low-level 1987; Sand 1996b); and, (4) intensified agricul- food production. The dominant faunal con- tural practices, inferred from inland expansion stituents in early East Polynesian assemblages and the development of terracing and irrigation are near shore reef fish (e.g., 90.4% of assem- systems. On Vanuatu, the presence of Malakulan blage at Tangatatau Rock shelter on Mangaia;

ideal free distribution and oceania 281 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 282 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

Steadman and Kirch 1990). However, these latter increase through time as the frequency of early faunal assemblages also contain the bones native landbird species decreases (Steadman of flightless birds, which were often extirpated and Kirch 1990). The 25 mystery islands of East within the first few hundred years of occupation Polynesia appear to have been abandoned after of each island. The rapid extinction of twelve the collapse of indigenous fauna and before agri- species of Moas in New Zealand is well-known cultural intensification (Anderson 2001b, Ander- (Anderson 1989) though it was probably not as son et al. 2002). In several of these instances it rapid as proposed by Holdaway and Jacomb appears that only one or two domesticated (2000; see Anderson 2000c). A wide range of plants or animals were successfully transported other animals was also extirpated (40 species of or propagated on these islands. birds, 1 bat, 3–5 frogs; Anderson 1997, 2002; Some temperate islands, particularly the Worthy 1997) there and elsewhere in the archi- South Island of New Zealand, were also outside pelagos of East Polynesia. Heavy intertidal pre- the range of successful cultivation of tropical dation pressure, indicated by decreasing shell cultigens. However, on islands that lacked such size through time, is also evident in shellfish as- environmental limitations, or were not aban- semblages from several islands (Steadman and doned, intensified agricultural strategies appear Kirch 1990). This ubiquitous East Polynesian to develop much more rapidly (ϳ100–200 years) pattern is evidence for early and rapid extinction, when compared to West Polynesia (ϳ500–1000 extirpation, or reduction in the largest, or most years) (Anderson 2002). This parallels evidence accessible, animals on each island (Anderson for increases in sociopolitical complexity, terri- 1981, 1984, 1988; Kirch 1996; Steadman 1989; toriality, and warfare (Kirch 2000). Evidence for Steadman and Kirch 1990; Steadman et al. 1994; anthropogenic environmental changes of defor- Weisler 1995). estation and erosion appear earlier and were Intensive agricultural strategies were well more rapid in East Polynesia (Anderson 2002). developed in West Polynesia prior to the colo- This was related, in part, to the small size of nization of East Polynesia. The available records these islands, but was also linked to the more indicate that early colonists carried economi- developed nature of food production at this late cally valuable plants and animals into East Poly- date. nesia. For instance, the early cultural strata at Tangatatau Rock shelter (ϳ1000 B.P.) contain a SUMMARY AND DISCUSSION rich carbonized plant record that includes taro (Colocasis esculenta), other root crops (Cyrtosperma Hundreds of islands were colonized in Oceania chamissonis), banana (Musa), breadfruit (Artocar- after 35,000 years ago, and each island provided pus altilis), Tahitian chestnut (Inocarpus fag- a new set of opportunities and constraints to po- iferus), sugarcane (Saccharum officinarum), ti tential colonists. Basic foraging models (Winter- (Cordyline terminalis), and the sweet potato (Ipo- halder 2001) predict that considerable temporal moea batatas) (Hather and Kirch 1991; Kirch and spatial variability in colonizing behavior et al. 1995; Kirch 1996). Chickens (Gallus gal- would have existed. Beyond the occupational lus), pigs (Sus scrofa), and dogs (Canine) are also histories of individual islands, there does appear evident in East Polynesian records relatively to be temporal and spatial structure in the early (Steadman and Kirch 1990). However, an process of colonization. Current archaeological immediate commitment to intensive food pro- data strongly suggest that the migration of peo- duction is not evident in the available records. ple to smaller, more remote islands and archi- Turning back to the Tangatatau Rock shelter ex- pelagos was episodic and not continuous. Bursts ample, the earliest levels contain large concen- of colonization activity were followed by longer trations of native land birds and very few domes- periods of local population growth, environmen- ticated animal bones (chickens or pigs), but the tal infilling, and intensification of subsistence

282 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 283 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

strategies. Smaller, more remote islands with inundated portions of Greater Bougainville to decreasing resource potential were colonized af- form the modern-day distribution of islands in ter periodic delays. the Solomons (Fairbanks 1989; Thiel 1987). The largest landmasses (New Britain, New Coastal wetlands would have become increas- Ireland, and Greater Bougainville) in Near ingly productive with the stabilization of sea Oceania were first colonized by hunter-gatherers level after 7000 years ago and may have con- between 35,000 and 29,000 B.P. Archaeologi- tributed to intensified use of maritime re- cal evidence for this early time is limited, but sources after this time. suggests that foragers complimented the use of Despite several decades of work in Remote protein-rich intertidal resources with wild plant Oceania, there is no evidence for occupation foods from wetland habitats. This included the east of the Solomon Islands prior to 3300 B.P. exploitation of Colocasia taro, a plant species (Anderson et al. 2001a). The ideal free distribu- that later became an important Oceanic domes- tion model predicts that these more remote is- ticate (Yen 1995). The periodic use of caves and lands would have been colonized only when rock shelters suggests that populations were average subsistence returns there were equal highly mobile using the landscape extensively. to those in the Bismarck Archipelago or the After initial colonization the record suggests a Solomon Islands. Delayed colonization of these long period of stasis with some evidence for remote islands (ϳ30,000 years) suggests that small-scale increases in population after 20,000 (1) population in the Bismarcks and Solomon years ago and again during the Holocene Islands were not experiencing sharp decreases (ϳ10,000–3500 B.P.; Spriggs 1997). This pe- in habitat suitability, but perhaps the opposite, riod saw the infilling of different environmental an Allee effect; (2) initial suitability of these re- zones on larger islands and the colonization of mote islands was low; or (3) the dispersal of pop- smaller adjacent islands in the Bismarck ulations to these islands was restricted because Archipelago and Solomon Islands (e.g., Manus; of environmental, technological, or social barri- Spriggs 1997). As populations increased, habi- ers (e.g., territoriality). A combination of these tat suitability would have decreased due to re- factors likely contributed to delayed coloniza- source depletion or interference, stimulating tion of more remote islands in the Pacific. intensification or migration. Other factors may have impeded coloniza- The presence of non-native species of trees tion as well. Clark and Kelly (1993) have argued and animals (e.g., chestnut and marsupial pos- that endemic malaria, common in the region to- sum) in late Pleistocene deposits in Near Ocea- day, caused high infant mortality and kept early nia (ϳ13,000 years ago) suggests that people populations in Near Oceania relatively low. It is were actively manipulating the landscape and also probable that resource availability in New possibly compensating for exploitation-induced Caledonia and Vanuatu, as well as other remote depression in the availability of naturally occur- islands, was comparatively low due to their ring foods (Anderson 2001b). Intensified sub- limited terrestrial plant and animal diversity sistence strategies are evident at interior loca- (Anderson 2001a; Spriggs 1997). In addition, tions where possum were aggressively targeted these islands are all south of the equator and are (Marshall and Allen 1991; White et al. 1991), in subject to greater seasonal differences in tem- Holocene shellfish assemblages suggesting in- perature and rainfall (Spriggs 1997). In the creases in diet breadth, and perhaps in the de- absence of agriculture and storage, seasonal velopment of arboriculture by about 4000 B.P. resource shortfalls—likely to have been more pro- Population increases and the intensification of nounced during Pleistocene glacial conditions— subsistence strategies occurred in the context of would have constrained the viability of coloniza- global sea level rise after 18,000 B.P. that re- tion. Current data suggest that these islands duced the size of islands in the Bismarks and were unoccupied until the late Holocene so

ideal free distribution and oceania 283 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 284 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

impediments due to resistance of earlier settlers as early as 3300 B.P. Reductions in habitat were certainly not a problem, but environmental suitability due to resource depression and inter- and technological barriers may have inhibited col- ference were probably exacerbated by interfer- onization. For example, the water gaps between ence from hostile indigenous populations. It islands were greater (over 300 km) and may have was in this context that, following a lag of sev- required more specialized maritime technology. eral hundred years, voyaging and colonization The appearance of Lapita settlements in the activities increased and the more remote parts Bismarck Archipelago was a major threshold in of Oceania—New Caledonia, Vanuatu, Fiji, Oceanic prehistory that represents either an Tonga and Samoa—were colonized between outgrowth of indigenous developments (Allen 3000 and 2800 B.P. (Anderson 2001a). and White 1989; Gosden 1992), an intrusion of As the vigor of eastward colonization began to Austronesian peoples from Southeast Asia wane 2700 years ago, and Lapita pottery started (Bellwood 2001; Spriggs 1997), or a combina- dropping out of the archaeological record, there tion of the two (Green 1991). Lapita settlements is increased evidence for reduced settlement were strategically positioned near or over, by mobility and intensified food production in Fiji, means of stilthouses, coastal lagoons that were New Caledonia, Tonga, and Samoa (Clark 1999; often on small islets adjacent to larger islands Sand 1996b). Intensive food production in Re- occupied by indigenous, non-Lapita, hunter- mote Oceania is signaled by large-scale forest horticulturalists. The placement of settlements clearance, intensive erosion, and the movement on smaller islands suggests that other island of soils from hillsides to valley bottoms (Anderson habitats were full or that Lapita peoples selected 2002). This culminated in the settlement of in- locations to avoid hostilities or endemic malaria terior locations and the terracing of hillsides to that likely plagued indigenous communities in contain erosion and maximize the amount of Near Oceania at the time (Clark and Kelly 1993). cultivated land. Territoriality and social circum- The frequency and size of Lapita settlements scription are suggested by the appearance of suggest that populations increased more rapidly fortified settlements in lowland and upland set- relative to contemporaneous hunter-horticultural tings (e.g., Best 1993; Field 2004), contributing communities on adjacent islands. The economic to decreases in habitat suitability that would engine for increased population growth probably have promoted emigration. The emergence of consisted of intensive maritime foraging coupled more intensive strategies varies between is- with low-level food production that intensified lands, but they were well established through- through the interval. It is also possible that rapid out West Polynesia by 1000 years ago. Several population increase was partially related to re- forms of intensive food production were evident duced susceptibility of Austronesian popula- by this time, including terracing, pond-field cul- tions to endemic malaria (Kelly 1999). Regard- tivation of taro, and irrigation agricultural sys- less, population increase in Lapita communities tems (Clark 1999). ultimately outpaced indigenous populations If our current estimates for the colonization which were swamped or replaced. of East Polynesia are correct, then the popula- Lapita migrants in Near Oceania entered tion increases, environmental infilling, and agri- landscapes that were depleted of larger game an- cultural intensification evident in West Polynesia imals and other easy prey (Allen 1996). The pro- occurred during a pause in eastward coloniza- ductivity of intertidal resources like shellfish was tion activity that lasted for more than 1500 years also reduced, at least in some areas, because of (Anderson 2002). Many of the islands in East sustained exploitation for thousands of years Polynesia were colonized rapidly after 1000 (Spriggs 1997). Lapita faunal assemblages indi- years ago and New Zealand was colonized as late cate a clear maritime focus and improvements as 800 B.P. Except for New Zealand, all of the in seafaring with fishing technologies available islands in East Polynesia are relatively small and

284 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 285 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

are not complex ecologically. The biological di- time and enough repetitions of migration events versity on these islands is low, the productivity that patterning becomes visible at a scale con- of marine habitats declines from west to east, comitant to that of the model. and seasonal variations in temperature and Nevertheless, the episodic nature of colo- rainfall become accentuated particularly on nization now evident in the archaeological subtropical islands. Rapid extinctions of the record could be the result of other historical largest landbirds and reptiles, evident archaeo- processes. For instance, colonization episodes logically (Anderson 1989; Steadman and Kirch could have been triggered by periodic advances 1990), reduced the biological diversity of these in seafaring technology, both of boats and navi- islands further still. gational and other sailing skills (Anderson Human ability to discover and colonize is- 2000a). Direct archaeological evidence for boats lands was dependent on factors of access, such is uncommon and makes this hypothesis diffi- as island size and remoteness, prevailing cli- cult to test. The neotraditional assumption is matic and environmental conditions in relation that celestial navigation techniques were well to maritime activities, and the availability of sea- developed and that large double-hulled canoes faring technology and knowledge (Anderson existed prior to the colonization of Remote 2000a; Erlandson 2001; Irwin 1992; Spriggs Oceania (Irwin 1992). Yet, the late Holocene ap- 1997). However, the data from Oceania are gen- pearance of Lapita assemblages coincides with a erally consistent with the predictions of the IFD sixfold increase in voyaging range into the pre- model: the long interval between the initial col- vailing wind and likely signals the first arrival of onization of Near Oceania (35,000 B.P.) and Re- the sail and possibly the outrigger (Anderson, mote Oceania (3300 B.P.) consistent with the 2000a, 2001c, n.d.a). Linguistic data, further- IFD curve for foragers living on large islands; more, suggest that the large double-hulled the more rapid succession of colonization canoes were a relatively late development, cer- episodes in Remote Oceania after the establish- tainly after the colonization of Fiji and West ment of Lapita populations in West Polynesia Polynesia (Blust 1997; Anderson n.d.a). The more consistent with IFD curves for food pro- colonization of extremely remote islands in East ducers on small islands. The 200–400 year Polynesia was probably contingent on the devel- pause in colonizing vigor after the first arrival of opment of the double-hulled canoe (Anderson Lapita people in Near Oceania and the 1500 2000a, 2001c). Developments in maritime year pause in West Polynesia prior to the colo- technology, therefore, were one important com- nization of East Polynesia, if upheld archaeolog- ponent in the episodic nature of colonization, ically, is also consistent with the infilling of pop- particularly for the discovery and assessment of ulations and the intensification of subsistence remote islands. Given the probable nature of strategies in Near Oceania and West Polynesia early sailing technology, particularly the likely respectively. absence of an upwind capacity (Anderson Thus, there appears to be a close match be- 2000a, 2001c), the larger numbers of voyages tween the scale of this HBE model and the avail- may have occurred during periods of wind re- able archaeological data (c.f. Smith, this volume). versal related to millennial scale changes in cli- IFD predictions are ones of central tendency, the matic and associated oceanographic conditions probabilistic equilibrium result of a large num- (Anderson n.d.b). Long-term climatic change ber of individual decisions made over a time in- may turn out to be another key element in ex- terval that allows for several rounds of adjust- plaining the episodic nature of the colonization ment in habitat choice. Likewise, the overall of Oceanic islands. history of colonization in Oceania, as recorded in An important assumption of the IFD model the archaeological record, smoothes highly local- employed here is that the costs of relocating were ized decisions over sufficiently long periods of small enough to be ignored. Setting sail across

ideal free distribution and oceania 285 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 286 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

the Pacific without prior knowledge of habitable ses in response to the traditional assumption islands would have been risky and potentially that food production was one of the primary costly, but it might not have been perceived in contextual changes that stimulated the coloniza- that light. In any event, we have assumed, for the tion of Remote Oceania (Kirch 2000; Spriggs sake of simplicity, that the locations of surround- 1997), a hypothesis that is currently not well ing, uninhabited islands were known to emi- supported by the available archaeological data. grants, their position and relevant agro-ecological One proposition is that early Lapita peoples features having been scouted prior to coloniza- were mainly foragers who skimmed the high- tion. To this extent, the decision to disperse est-ranked resources from pristine island en- would have been an informed one, and of rela- vironments in Remote Oceania. The second tively low risk and cost in transportation when proposition is that these people combined low- considered relative to the lifetime scale of its con- level food production of taro and chicken with sequences. In some cases this assumption may foraging for wild food, both terrestrial and ma- be unrealistic because of the distances traveled to rine. Regardless, it is clear that food production Fiji, Easter Island, and Hawaii and the ability of was not essential for the initial dispersal of peo- early island explorers to return to their home is- ple into Remote Oceania in a strict economic lands given prevailing winds and limitations im- sense. The known early Lapita assemblages in posed by maritime technologies (Anderson Near and Remote Oceania are dominated by 2003b, n.d.a). marine resources and contain few, if any, do- Given these limitations we suspect that the mesticates. If early Lapita people were foragers colonization of islands involved a dispersal (Hypothesis #1) then the decision to disperse phase and a migratory phase (Anderson, n.d.a), was not stimulated by the perceived advantages the former being the outward, one-way initial of food production on smaller, more remote is- occupation of islands, the latter a two-way lands that would not have sustained foraging process involving the movement of people back economies, increased productivity, and decreased and forth between island groups along with subsistence risk. This would suggest that early food and goods. Dispersal would have been an Lapita peoples sailed away from low-level food exploratory phase when the resource potential production and that they only brought in do- of islands was assessed given available subsis- mesticated plants and animals as the suitability tence practices and technologies. This phase of island habitats in Remote Oceania decreased would have been riskier, and therefore more with faunal collapse, one form of resource costly, than the migratory phase when more in- depression. formation was available regarding target is- Alternatively, the combination of low-level lands. Therefore, we envision the colonization food production and intensive maritime forag- of islands as a process rather than an event. Re- ing in Near Oceania may have contributed to gardless, population-dependent decreases in is- decreases in island habitat suitability during the land suitability would likely stimulate dispersal 200–400 year period that separates the first ap- and migration in much the same way, but the pearance of Lapita peoples and their dispersal costs of dispersal would have been considerably into Remote Oceania, in combination with higher relative to the follow-up migration long-term decreases in habitat suitability associ- process and would have contributed to the long ated with a sustained, long-term occupation by pauses between colonizing episodes. hunter-gatherers. This mixed subsistence strat- The distinction between dispersal and mi- egy would also have changed the perception of gration phases is important when considering the overall resource potential of smaller, more the role that food production played in the ini- remote islands while decreasing subsistence tial colonization of Remote Oceania by Lapita risk. Therefore, domesticates may have played a peoples. We have proposed alternative hypothe- significant role in the decision to disperse even

286 ideal free distribution and oceania GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 287 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

if the role of food production was minor ini- for this is also available (Kirch 1996). The most tially, due to the local availability of wild re- productive island habitats were selected, and sources. In the absence of competing popula- early colonists combined the hunting of ecologi- tions, the earliest settlement on these islands cally naïve landbirds and reptiles with continued would be expected in optimal locations for col- maritime foraging and agriculture. The rapid lecting marine and terrestrial resources and extinction of larger fauna caused by over- low-level food production. Less optimal loca- exploitation was compensated by increased food tions in the interior are expected to develop production, which intensified in parallel with in- later. Deforestation, sediment loading in valley creased population density. In some cases, fau- bottoms, and terracing could have created an nal collapse on islands with little agricultural po- Allee effect where larger populations were pos- tential resulted in abandonment (e.g., Pitcairn, sibly supported prior to environmental degrada- Norfolk and 23 others; Anderson 2001b). On tion. More despotic/territorial behavior would other islands rapid extinction resulted in agricul- be expected with habitat infilling as would the tural intensification, rapid population growth, intensification of agricultural strategies. Both the formation of large villages, territoriality, and would have contributed to more rapid emigra- more centralized political systems founded upon tion. Evaluation of this hypothesis is dependent hereditary leadership (Kirch 2000). upon larger-scale excavations in Early Lapita sites in Near and Remote Oceania coupled with CONCLUSION new technology for detecting domesticated plants in environments unfavorable for the The proposed Ideal Free Distribution model preservation of organic material, such as starch provides a framework that considers the dy- grain and phytolith analysis. namic character of island habitat suitability Food production appears to have been well along with density-dependent and density-inde- developed before the colonization of East Polyne- pendent variables influencing migratory behav- sia and certainly contributed to relatively rapid ior. The archaeological data from Near Oceania decreases in habitat suitability that played an im- is consistent with the IFD curves for hunter- portant role in the decision to emigrate. The gathers living on large islands. Evidence for the availability of domesticated plants and animals to more rapid colonization of Remote Oceania is augment the depauperate environments of re- consistent with IFD curves for food producers mote islands probably influenced the decision to living on small islands. The model also ac- emigrate. However, in a similar fashion to the counts for the episodic nature of island colo- Lapita colonization of West Polynesia, the initial nization. The initial generations of colonists in generations of colonists were subsidized by naïve Near and Remote Oceania were subsidized by and easily captured game. Successful and per- relatively dense and easily captured populations sistent settlement of these ecologically impov- of naïve game. This phase lasted several genera- erished islands was often dependent upon a tions, the bounty and persistence of this wind- rapid increase in agricultural production. fall a function of island size. Declines in wild Pollen and charcoal records from different is- resources appear to have been offset by intensi- lands indicate that extensive forest clearing fied foraging and the translocation of wild ani- and burning started within a century of colo- mals (Near Oceania) or efforts to replace these nization (Anderson 2002). Intensive agricultural resources with expanding food production production involving terracing, irrigation, and (Remote Oceania). In Remote Oceania the tran- pond-field cultivation developed rapidly suggest- sition from dependence on an ephemeral local ing that domesticated plants and animals, along bounty of foraged resources to a stable and with extensive agrarian knowledge, were carried fairly productive regime of cultivation was the by early colonists. Direct archaeological evidence point of greatest risk for successful colonization,

ideal free distribution and oceania 287 GRBQ084-2272G-C12[265-288]. qxd 11/28/2005 19:48 Page 288 pinnacle SUBHAKANT:GRBQ084:Chapters:Chapter-12:

while it is clear from the abandonment of sev- islands in East Polynesia late in prehistory when eral islands after faunal collapse that it was not options for emigration were limited. We argue always successful (Anderson 2003b). The popu- that this scenario is plausible, and testable. lation that made it through this period of vul- However, we note that others might be devised nerability, then experienced an Allee effect of within the framework of the IFD, perhaps giving increasing economies of scale in food produc- greater attention to social stratification and re- tion. Local habitat suitability grew as an effective source inequalities, and thus to a despotic vari- system of agro-ecological production was devel- ant of the IFD. oped. This phase was one in which population Our use of the IFD model also is qualitative was locally tethered; it lasted for a relatively long and general, as is our assessment of its fit to the interval before overpopulation, circumscription, available archaeological data. A more robust ap- and possibly environmental degradation began plication would develop out of independent to reduce suitability, leading to a new round of quantitative information on available habitats, emigration if suitable, uninhabited islands were including measurement of their ranking (qual- available. The length of each phase in this cycle ity) and the response of their suitability to in- was a function of island size, which would mean creasing human populations. The latter requires that, once a colonization episode was initiated, careful determination of the shape of the suit- small islands in East Polynesia would have been ability/density curve. The information on migra- colonized rapidly and relatively continuously. tion and settlement required to test this model The rapid development of social stratification in would include the sequence and timing of intra- East Polynesia after colonization also acceler- island settlement of habitats, and inter-island ated the tendency for emigration. migration, both relative to the population history Our formulation of this model is based on and its density in particular locales. It also would fundamental HBE principles, the IFD model in require observations on the socioeconomic con- particular, and our interpretations of the avail- ditions pertinent to “free” or “despotic” regimes able archaeological data in Oceania. The model of resource competition. It is encouraging that predicts that colonization of Oceania would this type of data is archaeologically accessible, have been episodic and not continuous. New and will eventually become available in suffi- radiocarbon chronologies throughout Oceania cient detail to evaluate the explanatory potential should continue to confirm the episodic nature of the IFD in this setting in a more quantita- of colonization starting 35,000 years ago. The tively rigorous fashion. model also predicts that a relationship exists be- tween population density and habitat suitability. NOTES Therefore, substantial population growth and 1. The islands of Micronesia are not considered in reductions in habitat suitability should be this paper, but the ideas explored here could eas- clearly evident in the palaeoenvironmental and ily be extended to that vast region. archaeological records in source archipelagos 2. All Pleistocene and early Holocene dates are prior to the next episode of emigration. More reported in radiocarbon years before present and rapid reductions in habitat suitability are ex- dates in the late Holocene are calibrated years be- fore present. pected with intensified food production strate- gies in that emigration would be expected prior 3. This pattern could also be caused by differences in the amount of archaeological fieldwork com- to large-scale agricultural intensification. The pleted in each region, but sustained work in both most intensive food-producing strategies such Near and Remote Oceania for the last 40 years as pondfields or terracing should be evident on would suggest that this is not the case.

288 ideal free distribution and oceania