Review
TRENDS in Ecology and Evolution
Vol.18 No.10 October 2003
531
Untangling Oceanic settlement:
the edge of the knowable
Matthew E. Hurles1, Elizabeth Matisoo-Smith2,3, Russell D. Gray4 and David Penny3,5
1
Molecular Genetics Laboratory, McDonald Institute for Archaeological Research, University of Cambridge, Downing Street,
Cambridge CB2 3ER, UK
2
Department of Anthropology, University of Auckland, Auckland 92019, New Zealand
3
Allan Wilson Centre for Molecular Ecology and Evolution, Institute of Molecular BioSciences, Massey University, Palmerston North,
New Zealand
4
Department of Psychology, University of Auckland, Auckland 92019, New Zealand
5
Institute of Molecular BioSciences Massey University, Palmerston North, New Zealand
Human expansion into the far reaches of the Pacific has
occurred within the past 3000–4000 years. This is so
recent that it is arguably the best opportunity to test
models of the origin and dispersal of human groups and
their domesticated plants and animals, cultural and linguistic evolution, human impacts on a pristine environment, and the lower limits for a long-term sustainable
population. Multidisciplinary research is essential
because these models must account for archaeological,
ecological, cultural, historical, social, linguistic and
(both mitochondrial and nuclear) genetic data. This synthesis has not yet been achieved for any settlement in
the world, but there has been considerable progress
recently on integrating these disciplines with respect to
the settlement of Polynesia.
The puzzle of Oceanic settlement, given the vast distances
from Southeast Asia to eastern Polynesia (Figure 1), has
long fascinated scholars. Synthesizing the main records of
the past – genetics (humans and their domesticated plants
and animals), linguistics, history, archaeology and paleoecology – is a multidisciplinary enterprise that must
consider the complementary strengths and weaknesses
of these data types. Archaeology provides the most precise
dates, whereas linguistics and genetics both infer ancestral patterns from modern diversity (although ancient
DNA fills some gaps). Recent advances in interpreting
records of the past, especially in Polynesia, make a review
of the current state of knowledge timely because, now more
than ever, communication between disciplines is vital to
ensure continued improvement in our understanding of
human settlement into the Pacific.
A brief background to the region
The archaeological record gives an excellent background to
the arrival of modern humans into Australia and New
Guinea [1,2] by 55 000– 60 000 before present (BP ).
Although Australia and New Guinea were connected
during periods of lowered sea levels, the first settlers
must still have crossed open ocean from southeast Asia. By
29 000 BP, people had colonized the Bismarck and Solomon
Corresponding author: David Penny (D.Penny@massey.ac.nz).
Islands, which together with New Guinea, form ‘Near
Oceania’. Reaching some of these islands required sizeable
sea voyages, much longer than those being undertaken in
the Mediterranean during the same period. There were
local exchange networks by 18 000– 20 000 BP [2], and by
9000 BP tree crops and other plants were being cultivated,
making the region an important early center for plant
domestication. However, population density was not high
and pottery is absent [2].
Sometime around 3300– 3500 BP, a new culture –
Lapita – appears in the archaeological record, and in a
previously unoccupied coastal niche (their stilt houses
were often built over beach reefs or shallow lagoons). The
ultimate geographical extent of this culture is shown in
Figure 1c. Lapita is defined characteristically by a
decorated pottery style and is named after an excavation
site in New Caledonia in Island Melanesia. Lapita culture
introduced new features, including permanent villages, a
range of horticultural crops, domesticated animals (pigs,
dogs, chickens and rats), fishhooks for inshore and open
ocean fishing, fishing nets, sea-going canoes, stone adzes,
anvils and shell bracelets. Under Green’s triple-I model
[3], the genesis of this culture requires the ‘intrusion’ of a
new culture, followed by ‘integration’ between the
cultures, and local ‘innovation’ in technology. For example,
the cultures combined aspects of an agricultural package
found in slightly earlier archaeological sites from island
southeast Asia with tree crops grown previously in Near
Oceania, as well as making significant local technological
developments [2]. Within island southeast Asia, the
archaeological sites associated with agriculture form a
temporal gradient. The simplest model of agricultural
dispersal is a southeastward movement from continental
Asia, possibly as a result of population expansion following
the development of agriculture [4]. The expansion continued through Taiwan (Ta-p’en-k’eng culture 4500–
5000 BP ) and the Philippines (e.g. Luzon 4000– 4500 BP )
to Wallacea (the biogeographical zone lying between
Borneo and New Guinea; Figure 1) and Near Oceania ([2]).
Although the earliest Lapita sites are found within
Near Oceania, they are found within another 200 years in
some parts of previously uninhabited ‘Remote Oceania’
http://tree.trends.com 0169-5347/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0169-5347(03)00245-3
Review
532
(a)
TRENDS in Ecology and Evolution
Vol.18 No.10 October 2003
Formosan
Malayo-Polynesian (MP)
Formosan
Micronesia
WMP
SHWNG
Central/Eastern MP
Western MP (WMP)
Nuclear Micronesian
Oceanic
CMP
Madagascar
Eastern MP
Central MP (CMP)
Polynesian
South Halmahera West
New Guinea (SHWNG)
Oceanic
Wallacea
M
e
la
ne
si a
esia
Polyn
(b)
Micronesia
Wallacea
M
e
la
ne
si a
esia
Polyn
(c)
~6300 BP
~4800 BP
~4500 BP
Wallacea
Hawaii
1200–1400 BP
29000–35
000 B Lapita culture
P
2900–3500 BP
Marquesas
1400–1700 BP
Cook Islands
800–1200 BP
60 000 BP
Easter Island
1000–1400 BP
New Zealand
800–1000 BP
Near Oceania
Remote Oceania
TRENDS in Ecology & Evolution
http://tree.trends.com
Review
TRENDS in Ecology and Evolution
(Eastern Melanesia, Micronesia, and Polynesia – including sites in Vanuatu, New Caledonia, Fiji, Tonga and
Samoa). New Lapita sites are still being discovered [5].
From the archaeological evidence, there appears to be a
pause of 500 to 1000 years before permanent settlement of
the eastern-most islands of Polynesia. The archaeological
dates given in Figure 1c represent conservative estimates
for permanent settlement [6]. Models of migration [7]
predict that earlier exploration and temporary settlement
were likely. The central islands of Eastern Polynesia
(Cooks, Society Islands and Marquesas Group) were
colonized initially, then the more peripheral islands
(Hawaii, Rapanui/Easter Island and Aotearoa/New Zealand). However, by this time, the characteristic Lapita
pottery was no longer being made, even though at least
some islands had suitable materials [2]. In contrast to
Near Oceania, population densities are thought to have
been high in much of Polynesia (estimates of 60–100
people km22 are common). This might reflect the lower
density of pathogens in Remote Oceania [2], but whatever
the reason, the high population density led to the
landscape being highly modified from its original
condition.
Pacific languages tell a similar story, and are well
integrated with archaeological studies [8]. Austronesian is
a well defined family of languages that is the world’s
largest and the most widely distributed [9]. Its ,1200
languages are classified into ten subfamilies, nine of which
are spoken only by indigenous Taiwanese (Formosans). By
contrast, languages of the tenth subfamily, MalayoPolynesian, are spoken from Madagascar (478 east) to
Easter Island (1098 west), and their comparative similarity
suggest that they share a recent common origin. Moreover,
non-Austronesian languages in Oceania are extremely
diverse and are expected consequently to be much older.
Although non-Austronesian languages are often lumped
together in a heterogeneous catchall ‘Papuan’ group, they
might be classified eventually into at least 12 major family
groupings [2]. These linguistic relationships can be
interpreted as predicting two major genetic groups in
the Pacific: the older Papuan lineages and a more recent
Austronesian group. However, language replacement
(where one language replaces another) is well known in
linguistics, and has occurred in Oceania [10]. Thus, we do
not always expect a one-to-one relationship between
languages and gene frequencies. In addition, borrowing
between languages could be substantial [11], especially in
Remote Oceania where contacts continued across hundreds of kilometers of ocean [12].
The traditional classification of Pacific islands into
Polynesia, Melanesia and Micronesia (Figure 1) does not
represent the complexity of the settlement of these islands.
Lumped together under the term ‘Melanesian’ are
Papuan-speakers from New Guinea, a land settled at
Vol.18 No.10 October 2003
533
least 50 000 years ago, with Austronesian-speakers from
Remote Oceania, who only settled within the past 3100
years. Here, we use ‘indigenous Melanesians’ to mean the
pre-Lapita occupants of Near Oceania.
Models for the initial settlement of Remote Oceania
The two oft-cited models for the origins of Remote Oceania
peoples – the ‘express train’ [13] and the ‘entangled bank’
(Box 1) – lie at the opposite ends of a spectrum. The former
predicts a strong phylogenetic (tree-like) signal reflecting
initial settlement pattern, whereas the latter suggests a
reticulate model with no phylogenetic signal remaining.
The situation is complicated because the models are
neither mutually exclusive, nor complete. One form of
data (e.g. genetics) might be tree-like, whereas another
(e.g. language) could be reticulate. Nor do the models
account for later effects, such as continued migration. In
addition, authors interpret the models differently. Each
model is complex such that a dataset might disagree with
one aspect of the model, but be concordant with others. It is
helpful to represent the spectrum of views (Box 1) as:
strong entangled bank $ weak entangled bank $ slow
train $ express train.
A general warning from the entangled-bank model is
important. Classic phylogenetic programs always output a
tree for morphological, molecular, or language data, even if
a tree is not appropriate. Although the programs enable
estimates of the fit of the tree to the data (using consistency
indices, g-statistics, etc.), they are not designed to evaluate
trees versus networks. Ignoring any genetic admixture
between populations can lead to inaccurate inferences of
the historical processes.
The models discussed here lie along a continuum, but
other versions are possible (Box 1). These models are
testable because they make predictions about genes,
cultures and languages. Although it sometimes appears
that authors assume that models stand or fall on a single
dataset, testing predictions requires that consideration be
given to all classes of data: molecular genetic, linguistic
and archaeological. During major migrations to previously
uninhabited lands, we might expect the simultaneous
transmission of genes, language and culture. By contrast,
we might expect the decoupling of biology, culture and
language during other periods. Thus, correspondence
between the different forms of data needs to be evaluated
rather than assumed.
What does recent linguistic research tell us about
Oceanic origins?
Linguistic analyses are a vital part of unraveling the
patterns of human migration [4]. The express-train model
of Pacific settlement predicts considerable tree-like signal
in Austronesian languages. Consequently, a tree of
Austronesian languages should reflect this pattern of
Figure 1. Map of Near and Remote Oceania, with ice-age sea levels, language families, Polynesian paternal and maternal lineages, and archaeological dates. (a) The distribution of Austronesian subfamilies and their phylogenetic relationships. The Malayo-Polynesian (MP) languages include Western (WMP), Central and Eastern (CEMP), Central (CMP) and Eastern (EMP) Malayo-Polynesian, as well as South Halmahera/Western New Guinea (SHWNG) and Oceanic. The Oceanic subfamily includes Nuclear
Micronesian and Polynesian languages. Shading indicates the approximate coastline during the last glacial maximum. (b) Distribution of the predominant paternal (blue)
and maternal (red) Polynesian lineages among modern populations. (c) Age of the earliest evidence for permanent settlement for Neolithic (black) and post-Neolithic (blue)
archaeological sites throughout the region. Particularly on islands, exploration could be earlier.
http://tree.trends.com
534
Review
TRENDS in Ecology and Evolution
Vol.18 No.10 October 2003
Box 1. Models of Oceanic origins: express trains and entangled banks
Express train
Entangled bank
The express-train model describes the initial spread of Polynesian
ancestors (Austronesians), first into Near Oceania and then into Remote
Oceania. The model is as follows.
(1) Austronesian languages arose northwestward of Near Oceania
(probably in Taiwan, but ultimately from China).
(2) The associated culture (including agriculture, horticulture, fishing, pottery, weaving and long-distance sailing) arose in the same
region northwest of Near Oceania.
(3) The Austronesian people differed genetically from the first
indigenous Melanesians (who spoke non-Austronesian
languages).
(4) The Austronesian people (with their languages) moved relatively
rapidly southward through the islands of southeast Asia and
subsequently eastward into Remote Oceania.
(5) There were no significant breaks between leaving Taiwan and
reaching western Polynesia.
(6) There was only limited genetic mixing between Austronesians
and indigenous Melanesians during this initial expansion, and no
large-scale replacement of indigenous Melanesians.
In its simplest form, the entangled-bank model [70] asserts that there
has been so much on-going interaction between adjacent populations
that representing them (or their cultures or languages) as tree diagrams
(phylogenies) is at best inappropriate and at worst seriously misleading.
For example, John Terrell states, ‘that these islands form an enormous
geographic array of local and island populations that, in all likelihood,
kept more or less in touch with one another ever since the first arrival of
people at least 45 000 years ago’ [70]. The strong version of the model is
that, because of extensive interactions in the past, there can never be
tree-like structure in these types of data, whether in Oceania or
elsewhere. By contrast, the weak version does not deny that there can
(sometimes) be a phylogenetic signal in the data; it is a general-purpose
null model that assumes no tree-like signal. Any phylogenetic signal
must be demonstrated, rather than assumed.
The first three points emphasize the ‘Out of Asia’ aspect, the last three
the ‘express train’. The spread of Lapita culture is the fastest in the
archaeological record [68], presumably because it was largely into
unoccupied niches and/or islands. However, the model says nothing
about any subsequent migration and is thus incomplete, although it is
usually assumed [69] that there was later admixture of indigenous
Melanesians and Austronesians at least as far as Fiji and probably
Samoa and Tonga. Thus the final point is:
(7) There is ongoing genetic admixture in Near Oceania over the past
3000 years.
dispersal. However, given the relatively fast travel across
Island Melanesia to Tonga/Samoa, the model would not
predict a robust signal (e.g. high bootstrap values) for this
section of the tree. Very rapid dispersals do not allow
sufficient time for languages to develop the novel linguistic
innovations that define subgroups. But given the longer
pause before the settlement of eastern Polynesia [2], there
should be more robust support for this subgroup of
languages if the founding population was from a relatively
homogenous source. Testing such hypotheses requires a
large database of linguistic information with a good
sampling of Austronesian languages.
Fortunately, Robert Blust at the University of Hawaii
has compiled such a dataset, the Austronesian Comparative Dictionary (unpublished), with over 5000 cognate sets
(homologous words) for over 200 Austronesian languages.
Gray and Jordan [14] converted this information into a
matrix with languages as taxa and cognate words as
binary characters. A variety of tests, such as skewness of
tree lengths, demonstrated that there is significant treelike signal in these data. The most parsimonious tree for
the data (Figure 2) matches closely the geographical
migration sequence postulated by the express-train model
(Box 1). These results contradict the entangled-bank
model in both its weak (null) and strong forms.
The bootstrap values, as predicted from the archaeological data, are high for the separation of the eastern
Polynesian clade, but lower for the early Oceanic branches.
Although there is significant phylogenetic signal in these
http://tree.trends.com
Other models of Oceanic origins
Alternatives to the express-train and entangled-bank models include
several intermediate models. The ‘slow-train’ version of the model
differs from the express version in predicting considerable genetic
admixture between Austronesians and the indigenous inhabitants of
the transit zones before the settlement of Remote Oceania. Both fastand slow-train models accept that some Austronesians stopped in
island Melanesia and coastal New Guinea. An ‘indigenous Melanesian’
model would derive Polynesian ancestry solely from the original
settlers from 50 000 BP . A recent ‘slow-boat’ model [20] is based on
genetic evidence and argues not only for a proximate Polynesian origin
within Wallacea, but also for an Austronesian origin in island southeast
Asia, rather than Taiwan/China. The proponents of this model consider
that a genetic signal of recent dispersal (, 10 000 BP ) can be traced back
from Polynesia only as far as Wallacea, although the ultimate origin for
such lineages in continental Asia is not in doubt.
data, there is also evidence for borrowing between
languages (e.g. the consistency index has a relatively low
value of 0.26). This is not surprising, given that the
Austronesians were expert navigators who dispersed as
far as Madagascar and Easter Island. A well-studied
example from South America of such a borrowing is
described in Box 2. Recent borrowings do not erase all
traces of phylogenetic signal from the initial migrationdriven tree. Phylogenetic methods such as split decomposition [15] do not assume a pure tree model, and thus can
be used to investigate these conflicting signals. Figure 3
shows a parsimony tree for 11 Polynesian languages and
their corresponding splits graph. The splits graph shows
separation between western and eastern Polynesian
clades, but substantial borrowing within these groups.
For example, there is signal linking Hawaii to both the
Marquesian and Tahitic language groups.
Linguistic data might also be used to distinguish
between the express-train and the slow-train models.
According to the slow-train model, there was sufficient
time for substantial borrowing from ‘Papuan’ languages
into Oceanic ones. This means that numerous words of
Papuan origin should be found throughout Oceanic
languages, but should not occur in the earlier branches
of the Austronesian tree (Figure 2). To the best of our
knowledge, these ancient loan words are not found, but the
question would repay further investigation. A third way of
using language data to discriminate between hypotheses is
to estimate dates of linguistic divergence. Languages do
Review
TRENDS in Ecology and Evolution
90
Box 2. Kumala – sound changes help reveal linguistic
origins
Sound changes enable inferences about population histories to be
made because they often change in regular ways. For example,
words in the Paamese language of Vanuatu have lost typically both
/pk/ from the proto-Oceanic form and /pa/ at the end of words [71]. In
Table I, words inferred to be in a proto-language are asterisked.
Words not following these changes are likely to be more recent
borrowings. The Polynesian sweet potato (kumala) illustrates this.
The striking similarity of the word in many parts of the Pacific (Maori,
/kuumara/: Rarotongan, /kuumara/; Samoan, /’umala/; Tahitian,
/’umara/) might be taken as evidence for its antiquity among protoOceanic speaking peoples. However, if kumala was an old inheritance from proto-Oceanic into Paamese, then the initial /k/ and the
final /a/ should have been lost, and the form should be /umal/. The
fact that it is /kumala/ is consistent with it being a more recent
borrowing. Botanical evidence [72] suggests an origin for the sweet
potato in South America and archaeological evidence suggests that
kumala cultivation has spread through Polynesia relatively recently,
although before European contact [73]. Correspondingly, in the
Quechua language of Peru, the term for sweet potato is strikingly
similar – /kumar/. This suggests that Polynesians voyaged to South
America and returned with this root crop in recent prehistory,
although this possibility requires further testing.
Table I. Observed word changes between proto-Oceanic
and Paamese
Proto-Oceanic form
Paamese
Meaning
p
a:i
ai
ahi
tree
fish
Malay apple
a kai
p
a ika
p
kapika
diverge with time, but Swadesh’s proposal [16] that this
change is sufficiently clock-like to be able to date language
divergences is contentious [17]. If recent methods for
dating trees with sequence data (in the presence of rate
variation) can be applied to linguistic data, then this would
give a significant improvement in our ability to discriminate between complex settlement scenarios. To summarize
this section:
(1) Language data generally support an express-train
model with respect to both the origin of the Austronesian language family and the rate of Austronesian
movement across Island Melanesia.
(2) Split decomposition analyses of lexical data also
reveal evidence of considerable post-settlement
contact.
What does recent molecular genetic work tell us about
Oceanic origins?
Although Oceania is arguably the best place to reconstruct
human migration, the task is full of possible pitfalls because
modern genetic diversity results from a mixture of signals:
(1) a founding population (a subset of a larger population);
(2) genetic drift and selection (in both the founding and
ancestral populations);
(3) subsequent migration in prehistoric times (again into
both the founding and the ancestral populations, and
which might be different for males and females); and
(4) ‘post-contact’ admixture from a variety of populations,
including Europeans.
http://tree.trends.com
535
Vol.18 No.10 October 2003
100
21
22
64
Polynesia
Micronesia
Remote
Oceania
Solomon
Islands
Near Oceania
WMP Outliers
CMP
South
Sulawesi
Borneo
Indonesia
WMP
N. Borneo
N. Sulawesi
Philippines
Taiwan
TRENDS in Ecology & Evolution
Figure 2. Parsimony tree for 77 Austronesian languages. The tree is rooted with
two Formosan (indigenous Taiwanese) languages. Bootstrap values are shown
above some of the branches in the Oceanic clade. Consistent with the suggestion
of a pause in settlement, the Eastern Polynesian subgroup has strong bootstrap
support (90%). Abbreviations are as in Figure 1.
A primary problem is separating the extent of admixture between dispersing (Austronesian) and indigenous
populations during the initial movement across Near
Oceania and Wallacea from that occurring over the
subsequent 3500 years. Genetic drift and post-settlement
migration render analyses difficult because current gene
frequencies do not represent ancestral gene frequencies.
Although the models agree in inferring that Continental
Asia is the ultimate origin of genetic lineages now in
Remote Oceania, they differ in the timing of the migration
movements. This leads to a focus on non-recombining loci,
specifically from the maternally inherited mitochondrial
DNA (mtDNA) and the paternally inherited Y chromosome. Both loci provide opportunities for extracting
chronological information from genetic data for comparison with archaeological dates. These two important loci
also have the advantage of lower effective population sizes
than the rest of the genome, resulting in higher genetic
drift and greater population differentiation.
Mitochondrial DNA studies in the Pacific show a high
frequency of a nine bp deletion in the COII/tRNALys region
of the genome. The distribution of this deletion has been
catalogued extensively from Madagascar to eastern Asia to
Easter Island [10]. The deletion, together with three
characteristic mutations in the D-loop, forms a ‘Polynesian
motif,’ and it is useful to examine its distribution (together
536
Review
TRENDS in Ecology and Evolution
Vol.18 No.10 October 2003
Figure 3. Detecting potential borrowing between Polynesian languages. (a) Parsimony tree – data derived from the Pollex database [67]. (b) Splits graph for the same data
as in (a), showing evidence of reticulation (borrowing) within the western (blue) and eastern (red) Polynesian clades. (c) A sample of cognate (homologous) lexical items
for these languages. Noncognate terms are in an italic font.
with its predecessors with just one or two of the
characteristic D-loop mutations). The full motif has its
highest frequencies in Polynesia (where it predominates),
and is confined largely to speakers of the central/eastern
Malayo-Polynesian subgroup of Austronesian languages
spoken in Wallacea and the Pacific (Figure 1). Stepping
backwards phylogenetically, the ancestral lineage with
two of the three D-loop mutations is found across the entire
range of Austronesian language speakers from Madagascar to Easter Island. However, the earlier lineage (with
only one of the three mutations) is confined to the central
part of this range. Thus, this pattern of a directional series
of changes provides a relative chronology of dispersal with
ultimate origins (undated) in Asia. Similarly, in a dataset
of 53 complete human mitochondrial genomes [18], the
closest relative of the single Polynesian (Samoan)
sequence is Korean, with which it shares the nine bp
deletion. These two sequences then join with a Chinese
sequence and these three ‘Asian’ mitochondrial genomes
are closer (phylogenetically) to those of Europeans than to
those of indigenous Melanesians. Overall, the distribution
of mitochondrial lineages is consistent with an ‘Out of Asia’
model, whether express-train, slow-train or slow-boat, but
the distribution of nine bp deletions was taken initially as
support for the express train [19].
Chronological estimates of the time to the most recent
common ancestor (TMRCA) of a set of sequences depend on
accurate parameters and assumptions about mutation
rate, generation time and prehistoric demographics, and
come consequently with wide confidence intervals. The
diversity of the Polynesian motif in island southeast Asia
(Wallacea) has led to an estimated TMRCA of mtDNAs
carrying the Polynesian motif at , 17 000 BP [20],
http://tree.trends.com
although this date still requires a full sensitivity analysis
[21] and appears to contradict the linguistic data
(Figure 2). If this date does hold up, then there might be
a pre-agricultural origin of the final motif within Wallacea;
only after the arrival of agriculture would there be
migration into Remote Oceania. However, the paucity in
Remote Oceania of common highland New Guinean
lineages means there was little maternal genetic admixture between the dispersing population and indigenous
Melanesians.
Y-chromosome studies enrich the picture across Polynesia and island Melanesia. There is clear evidence for
male-biased European admixture among some Polynesian
populations [22,23], but without a comparable gene flow of
European mtDNA. In one case, careful historical work was
required to trace Native American Y chromosomes back to
the Polynesian slave trade [24]. In other cases, it was
possible to detect post-settlement gene flow; for example,
post-Lapita gene flow of Y chromosomes from island
Melanesia into Tonga [25], and mtDNA into the Santa
Cruz Islands [26]. Oral histories suggest subsequent
migrations into Fiji [27]. Several recent studies report
two Y-chromosomal lineages predominating in Polynesia
[25,28,29], one of which is absent from the current
population of indigenous Taiwanese. However, ancient
highland New Guinean paternal lineages, which presumably represent indigenous Melanesians, are scarce in
Polynesia [30]. Although both major Polynesian lineages
appear to predate the earliest agricultural dates in island
southeast Asia, the higher-frequency lineage appears in
Wallacea and is absent further north in island southeast
Asia (Figure 1), mirroring the distribution of the mtDNA
Polynesian motif. Given their antiquity, the absence of
Review
TRENDS in Ecology and Evolution
these lineages from the supposed origin of Austronesian
languages and agriculture in Taiwan and southeast China
does not conform to the simple express-train model.
An exciting prospect is that mitochondrial, Y-chromosomal and autosomal data [31,32] could reveal sex-specific
differences in prehistoric demography, including levels of
endogamy and/or migration patterns. One example of sexbiased endogamy is a matrilineal society where females
tend to stay locally, but accept outside males into the group
(e.g. females in early 19th century New Zealand accepted
external males into the reproductive group [33]). Linguistic evidence [34] suggests proto-Oceanic societies were
both matrilineal and practiced matrilocal residence,
whereby husbands moved into their wife’s village. Marital
residence patterns can generate differences in patterns of
male and female lineage diversity [35]. Historians go
further [36] and report that chieftain status was inherited
maternally from Madagascar to Polynesia, with highranking women being very influential. Population genetic
modeling needs to take these factors into account.
At a more microgeographic level, the clinal decrease in
genetic and linguistic diversity from west to east across the
Pacific [21,37] emphasizes the role of successive founder
events in shaping patterns of variation in Remote Oceania.
Estimating the size of the colonizing populations becomes
crucial to understanding this process. Thus far, this has
been done only for New Zealand Maori, where the estimate
for females of a founding population of 50– 100 individuals
agrees well with oral histories [21]. However, nobody as yet
has presented a unifying model of Oceanic settlement
consistent with all molecular, linguistic, historical and
archaeological evidence.
The ecological impact of Oceanic settlement
In Remote Oceania, evidence of initial human activity is
easier to identify than it is in the rest of the world. This is a
result of several factors, from the relatively short time
frame of human occupation, to the decreases in general
biodiversity as one moves eastward across the Pacific (new
introductions are more obvious and identifiable). Similarly, the more fragile island ecosystems are disturbed
more easily and obviously by human and human-associated activity such as burning and other habitat modification, hunting, and the introduction of new predators or
competitors.
Introduction of commensal and domesticated species
One of the greatest impacts to island ecosystems is the
introduction of novel flora and fauna. Pacific colonists
introduced a range of plant and animal species to the
relatively depauperate island environments. Humanassisted dispersal of animals dates as far back as the
Pleistocene in Near Oceania, with the translocation from
New Guinea to New Ireland of the possum-like cuscus
(Phalanger orientalis) occurring possibly as early as
19 000 BP, and of a wallaby (Thylogale brunii) occurring
perhaps around 7000 BP. Some archaeological evidence
suggests that pigs (species uncertain) were introduced to
New Guinea from somewhere in southeast Asia by
6000 BP ; some evidence suggests it might be as early as
10 000 BP, although these early dates for pigs are debated
http://tree.trends.com
Vol.18 No.10 October 2003
537
fiercely. Linguistic evidence suggests that proto-Oceanic
speakers introduced the pig , 4000– 3500 BP [38].
Despite the debates regarding early pig introductions,
there is little doubt that Lapita colonists introduced
several animals into Remote Oceania, including dogs
(Canis familiaris), pigs (assumed to be Sus scrofa), jungle
fowl (Gallus gallus) and the Pacific rat (Rattus exulans). It
is unclear whether other animals such as geckos, skinks
and land snails were accidental or deliberate introductions
(human viruses would also be accidental introductions
[39]). Not all animals made it to all locations. For example,
only the rat and the dog were introduced successfully to
New Zealand, and the rat and chicken to Easter Island.
Sometimes, the appearance of dogs and pigs on an island
occurs early, thereafter disappearing from the archaeological record, only to be re-introduced later [40]. The
Pacific rat was the only commensal animal introduced to
all islands that show evidence of Polynesian presence.
Mitochondrial DNA lineages shared between Pacific rat
populations identify homeland regions, and disjunct
variation within island rat populations suggests separate
introductions from several sources [41]. This is consistent
with archaeological evidence that post-settlement contact
and long-distance exchange were significant in Polynesian
prehistory, even in the geographical extremes of Hawaii,
Easter Island and New Zealand [42]. Other domesticated
animals are less straightforward in that the pigs, dogs and
fowl introduced by the Lapita colonists are the same
species as those introduced later by Europeans and,
therefore, the respective populations are subject to more
recent admixture. Thus, earlier and later lineages must
be separated, a task for which DNA sequences,
including ancient DNA studies, are ideal. The commensal
approach has also been applied to extant populations of
Lipinia noctua, a lizard native to New Guinea, to show
that it was also transported from Near Oceania into
Remote Oceania [43].
Just as with studies of humans, we need to know how
well modern animal populations represent the genetic
variation present in prehistory. Animal bones (rats in
particular) are found in large numbers in archaeological
sites throughout the Pacific, allowing ancient and modern
populations to be compared. Working with animal remains
is less fraught with the ethical implications of handling
human remains! Thus, ancient DNA analyses of commensal animals are valuable for understanding issues in
Pacific prehistory, and as a model for the evolution of
island populations in general.
In one case, archaeological and modern mtDNA of the
Pacific rat from the Chatham Islands (an isolated group to
the east of New Zealand with a relatively simple
colonization history) show that modern populations
represent fairly early populations. By contrast, results
from New Zealand itself are more complex. Whereas
archaeological remains of rats from the South Island have
the same mtDNA lineage as extant rats, two distinct
lineages appear in the North Island. One is identical or
related closely to extant populations, yet the second is
distinct and not yet found in extant populations in New
Zealand or East Polynesia [44]. Loss of diversity is not
surprising. Because of competition from introduced
Review
TRENDS in Ecology and Evolution
rodents (the brown rat Rattus norvegicus, the house rat
Rattus rattus, and the house mouse Mus musculus), the
Pacific rat is extinct on the North and South Islands except
for a few pockets in Fiordland. Extant populations on
offshore islands are remnants and do not represent the full
original variation. Current work extends to Asian and
western Pacific rat bones [45] and further analyses of
Pacific pigs and dogs. In general, this sampling of remnant
animal populations is exactly what researchers are facing
in most human studies of population variation of origins
and dispersal, and we need to consider lineage extinction
in our interpretations.
Holdaway [46] reports another application of the study
of commensal animals. Introduced small animals are often
prey to large birds or other animal predators, and their
bones are often present in natural deposits found in caves
(e.g. in raptor pellets). If these remains pre-date archaeological dates for successful human settlement, then we
have valuable markers of pre-settlement contact that
otherwise might not show up in the archaeological record.
In this way, we can tease apart potentially the different
types of human contact (discovery, visitation, early
settlements and later successful colonization [7]) and,
therefore, better understand the total impact and history
of prehistoric people in the Pacific.
Extinction of native species
As expected, there have been major extinctions throughout
Polynesia following the arrival of humans [47,48]. As
usual, there is disagreement whether this is because of
physical (climate change) or biological (direct and indirect
human impacts) factors. The overview is clear in the
Pacific when broadened to Australia and North America,
as well as Near Oceania, Fiji, Tahiti and New Zealand
(Figure 4). Extinction follows human contact [49], whether
in the middle of the last Ice Age in Australia [1], possibly
around the glacial maximum (20 000 BP ) in the rest of
Near Oceania, around the end of the Ice Age (, 11 000–
12 000 BP in North America [50]), 3000– 4000 Bp in island
Melanesia [51], or in the last millennium in New Zealand
[52 – 54]. As we become more aware of the diversity of
mechanisms of human impact, the problem of accepting
human impact disappears; the direct effect of ‘climate
change’ on extinction is not an option for species that have
already survived many Glacial/Interglacial cycles.
Hunting is just one impact, and many sites are known.
A recent study of over 8400 bones from one site [55] found
species that were hunted more frequently (as assessed by
their presence in middens) were more likely to go extinct.
For the giant flightless ratites (moa), extinction might
have been fast [52]. Modeling shows that birds with slow
rates of reproduction were particularly susceptible and
this conclusion has now been extended to mammals [56].
Accounts by early explorers of the naı̈veté of the native
fauna (to humans) have been collected [57] and show how
easy hunting could be. Thus, comparisons between Europe
and North America [50] are not relevant; humans have
been in Europe for hundreds of thousand of years. It is
standard evolutionary theory that animals evolve defense
mechanisms (including behaviors) only for existing dangers. However, direct predation is only one mechanism. As
http://tree.trends.com
Vol.18 No.10 October 2003
summarized earlier, human impacts also include habitat
modification (including land clearance and effects of fire)
and indirect effects via introduced biota. Effects will
depend on many factors, including human population
density [2], island size and diversity of landforms, and
climate and vegetation types (e.g. Polynesian fires
removed forests on the drier east coast of New Zealand,
but not on the wetter west coast).
Ecological evidence on settlement processes
New Zealand is the last major area in Polynesia to have
permanent human settlement; carefully calibrated 14C
dates have shown no confirmed settlements before 1200
AD . However, bones of the Pacific rat several hundred
(possibly up to 1000) years older than this ‘imply an early,
transient, human contact’ [46]. This claim has been
controversial [58] and all early dates are eliminated
under the guise of ‘chronometric hygiene’. However, as
mentioned earlier, studies of human migrations [7] predict
a standard pattern of:
† Exploration well before outpost settlement,
† Some early settlements, then
† Migration and permanent settlement, and
† Ongoing contacts.
Was there release of commensal rats before settlement,
and, if so, how long before settlement? It was common
practice by European explorers to release food animals
(e.g. goats, sheep, and cattle) on uninhabited islands, and,
given that the Polynesians were the world’s most advanced
explorers of their time, we expect them to have done the
same. Oral tradition in Polynesia records exploration
before settlement, although the time scale is difficult to
60 000
50 000
Impact (yr BP)
538
40 000
30 000
20 000
10 000
0
0
10 000 20 000 30 000 40 000 50 000 60 000
Human arrival (yr BP)
TRENDS in Ecology & Evolution
Figure 4. Testing ‘climate change’ versus ‘human impacts’ (direct and indirect) for
extinction of a naı̈ve biota. Under a global climate-change model, extinctions
should be independent of human arrival times: if extinctions are human induced
(direct or indirect), then they cannot occur before human arrival (i.e. not in the top
left-half of the graph). Conversely, under the human-impact model, the changes
will be post discovery: if extinctions are human induced, they must have occurred
after human arrival (i.e. either will appear on the dashed line or in the bottom
right-half of the graph). In practice, impacts on a naive fauna are simultaneous
with, or after human discovery (diamonds). Effects might be slower on larger
and/or more diverse landmasses; conversely, they should be faster in smaller,
more homogenous islands. Some indirect effects (such as from the release of commensal animals) could occur between initial discovery and permanent settlement.
Review
TRENDS in Ecology and Evolution
determine. If rats were released early, then extinctions of
some invertebrates and small vertebrates should precede
the extinction of moa and other large taxa by habitat
modification and hunting. There is indeed some evidence
[54] for early extinctions before human settlement. Any
release of commensal animals before permanent settlement would help separate direct and indirect ‘human’
impacts. The full range includes effects of commensal
animals, hunting/foraging, habitat modification (including
fire), and loss of prey. There is probably nowhere else in the
world where we can separate so many individual factors.
Finally, early European explorers in the Pacific found
13 unpopulated islands where there was evidence of
earlier human settlement [59]. All are small islands, and
would not have had a large population. Some were at the
lower temperature limits for growing food plants or had
unsuitable coral soils. Given the intense interest in
conservation biology for the sustainability of different
population sizes, these case studies are potentially highly
informative. In addition to ecological and psychological
stresses, small populations result in high demographic
stochastic variation with dire consequence for mate
availability and long-term population survival.
Conclusions and future directions
Oceanic prehistory is not simple, but it is tractable. There
is a strong phylogenetic signal, together with some
reticulation, for both languages and genetics. The ultimate
origins of Austronesians and the more proximate origins of
Polynesians and other Remote Oceanic populations will
not be settled until all disciplines are better able to
characterize pre-Lapita populations in Wallacea and postLapita population movements in Remote Oceania. Greater
interaction between disciplines must occur and we
consider a few examples here. First, archaeo-linguistic
reconstructions of ancestral societies [60] (the historical
record of male/female roles will help in this respect [33,36,
61]) should help resolve differences between mtDNA and
Y-chromosomal data. Second, archaeological evidence of
pauses in dispersal can co-occur both with linguistic clades
exhibiting high bootstrap values and with regions of
heightened genetic admixture between indigenous and
dispersing peoples given the additional time for interaction. Improved models of language evolution will help
estimate times of divergence. If there is some selection
against language change, then language in a large
community might evolve more slowly than in a smaller
one. By contrast, under a neutral model of language
change, the rate would be independent of population size.
Genetics offers many opportunities for increased
precision. Complete mtDNA genomes are much more
powerful than D-loop sequences [18], although many more
samples are required. New techniques [62] are increasing
resolution for Y-chromosomal data. A new opportunity
comes from the detection of haplotype blocks – relatively
long stretches of nuclear DNA where recombination has
not occurred for many generations. These blocks contain
information about population history and are potential
markers of selection and health [63,64]. Increased genetic
study of commensal and domesticated species, such as
chickens and pigs as well as for domesticated plant species
http://tree.trends.com
Vol.18 No.10 October 2003
539
such as species of taro and breadfruit, is necessary. We
need more genetic data about the sweet potato (Box 2) and
bottle gourd [65], which are thought to have South
American origins. Models of long-distance voyaging
make getting to South America and back realistic for
early Polynesians [66].
Nowhere else in the world offers the same opportunity
to unravel the dramatic and complex effects of humans on
the environment and on indigenous plants and animals.
Everything we have learned confirms that Remote
Oceania is the best location to test our understanding of
human migration and impacts on a naive environment.
Acknowledgements
We thank many people for discussion and assistance with ideas, including
Robert Blust, Roger Green, Richard Holdaway, Kerry Howe, Matthew
Spriggs, David Steadman and Trevor Worthy.
References
1 Roberts, R.G. et al. (2001) New ages for the last Australian megafauna:
continent-wide extinction about 46,000 years ago. Science 292,
1888– 1892
2 Kirch, P.V. (2000) On the Road of the Winds: An Archaeological History
of the Pacific Islands before European Contact, University of California
Press
3 Green, R.C. (1991) The Lapita cultural complex: current evidence and
proposed models. Indo Pacific Prehist. Assoc. Bull. 11, 295– 305
4 Diamond, J. and Bellwood, P. (2003) Farmers and their languages: the
first expansions. Science 300, 597– 603
5 Burley, D.V. and Dickinson, W.R. (2001) Origin and significance of a
founding settlement in Polynesia. Proc. Natl. Acad. Sci. U. S. A. 98,
11829– 11831
6 Spriggs, M. and Anderson, A. (1993) Late colonisation of East
Polynesia. Antiquity 67, 200 – 217
7 Fix, A.G. (1999) Migration and Colonization in Human Microevolution, Cambridge University Press
8 Green, R.C. (1999) Integrating historical linguistics with archaeology:
insights from research in remote Oceania. Indo Pacific Prehist. Assoc.
Bull. 18, 3 – 16
9 Blust, R. (1999) Subgrouping, circularity and extinction: some issues in
Austronesian comparative linguistics. Symp. Ser. Inst. Linguist. Acad.
Sinica 1, 31– 94
10 Merriwether, D.A. et al. (1999) Mitochondrial DNA variation is an
indicator of Austronesian influence in island Melanesia. Am. J. Phys.
Anthropol. 110, 243– 270
11 Geraghty, P. Borrowed plants in Fiji and Polynesia: some linguistic
evidence. In Borrowing: A Pacific Perspective (Tent, J., and Geraghty,
P., eds), Pacific Linguistics (in press)
12 Weisler, M.I. and Kirch, P.V. (1996) Interisland and interarchipelago
transfer of stone tools in prehistoric Polynesia. Proc. Natl. Acad. Sci.
U. S. A. 93, 1381– 1385
13 Bellwood, P. (1998) The archaeology of Papuan and Austronesian
prehistory in the Northern Moluccas, Eastern Indonesia. In Archaeology and Language II: Correlating Archeological and Linguistic
Hypotheses (Blench, R. and Spriggs, M., eds) pp. 128 – 140, Routledge
14 Gray, R.D. and Jordan, F.M. (2000) Language trees support the
express-train sequence of Austronesian expansion. Nature 405,
1052– 1055
15 Huson, D.H. (1998) SplitsTree: analyzing and visualizing evolutionary
data. Bioinformatics 14, 68 – 73
16 Swadesh, M. (1952) Lexico-statistical dating of prehistoric ethnic
contacts: with special reference to North American Indians and
Eskimos. Proc. Am. Philos. Soc. 96, 452– 463
17 Renfrew, C. et al. (2000) Time Depth in Historical Linguistics (Vols 1
and 2), McDonald Institute Archaeological Research, University of
Cambridge
18 Ingman, M. et al. (2000) Mitochondrial genome variation and the
origin of modern humans. Nature 408, 708 – 713
19 Melton, T. et al. (1995) Polynesian genetic affinities with Southeast
Review
540
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
TRENDS in Ecology and Evolution
Asian populations as identified by mtDNA analysis. Am. J. Hum.
Genet. 57, 403 – 414
Oppenheimer, S.J. and Richards, M. (2001) Slow boat to Melanesia?
Nature 410, 166 – 167
Murray-McIntosh, R.P. et al. (1998) Testing migration patterns and
estimating founding population size in Polynesia by using human
mtDNA sequences. Proc. Natl. Acad. Sci. U. S. A. 95, 9047 – 9052
Underhill, P.A. et al. (2001) Maori origins, Y-chromosome haplotypes
and implications for human history in the Pacific. Hum. Mutat. 17,
271 – 280
Hurles, M.E. et al. (1998) European Y-chromosomal lineages in
Polynesians: a contrast to the population structure revealed by
mtDNA. Am. J. Hum. Genet. 63, 1793– 1806
Hurles, M.E. et al. (2003) Native American Y chromosomes in
Polynesia: the genetic impact of the Polynesian slave trade. Am.
J. Hum. Genet. 72, 1282 – 1287
Hurles, M.E. et al. (2002) Y chromosomal evidence for the origins of
Oceanic-speaking peoples. Genetics 160, 289 – 303
Friedlaender, J.S. et al. (2002) A cautionary tale on ancient migration
detection: mitochondrial DNA variation in Santa Cruz islands.
Solomon Islands. Hum. Biol. 74, 453– 471
Parry, J.T., (1997) The North Coast of Viti Levu – Ba to Ra. Air Photo
Archaeology and Ethnohistory, Bulletin of the Fiji Museum (No. 10)
Kayser, M. et al. (2001) Independent histories of human Y chromosomes from Melanesia and Australia. Am. J. Hum. Genet. 68, 173– 190
Capelli, C. et al. (2001) A predominantly indigenous paternal heritage
for the Austronesian-speaking peoples of insular Southeast Asia and
Oceania. Am. J. Hum. Genet. 68, 432 – 443
Su, B. et al. (2000) Polynesian origins: insights from the Ychromosome. Proc. Natl. Acad. Sci. U. S. A. 97, 8225 – 8228
Lum, J.K. et al. (1998) Mitochondrial and nuclear genetic relationships
among Pacific Island and Asian populations. Am. J. Hum. Genet. 63,
613 – 624
Hagelberg, E. et al. (1999) Molecular genetic evidence for the human
settlement of the Pacific: analysis of mitochondrial DNA, Y chromosome and HLA markers. Philos. Trans. R. Soc. Lond. Ser. B 354,
141 – 152
Belich, J. (1996) Making Peoples: A History of the New Zealanders,
Penguin
Hage, P. (1998) Was proto-Oceania society matrilineal? J. Polynes. Soc.
107, 365 – 379
Oota, H. et al. (2001) Human mtDNA and Y-chromosome variation is
correlated with matrilocal versus patrilocal residences. Nat. Genet. 29,
20 – 21
Gunson, N. (1987) Sacred women chiefs and female ‘headmen’ in
Polynesian history. J. Pac. Hist. 22, 139– 171
Lum, K.J. et al. (2002) Affinities among Melanesians, Micronesians,
and Polynesians: a neutral, biparental genetic perspective. Hum. Biol.
74, 423 – 430
Blust, R. (1976) Austronesian culture history: some linguistic
inferences and their relations to the archaeological record. World
Archaeol. 8, 19 – 43
Harrison, G.L. et al. (2001) Hepatitis B virus genotypes: a South Pacific
perspective. Pac. Health Dialogue 8, 188 – 192
Allen, M.S. et al. (2001) Pacific ‘Babes’: issues in the origins and
dispersal of Pacific pigs and the potential of mitochondrial DNA
analysis. Int. J. Osteoarch. 11, 4 – 13
Matisoo-Smith, E. et al. (1998) Patterns of prehistoric human mobility
in Polynesia indicated by mtDNA from the Pacific rat. Proc. Natl.
Acad. Sci. U. S. A. 95, 15145 – 15150
Weisler, M.I., and Green, R.C., (2001) Holistic approaches to
interaction studies: a Polynesian example. In Australasian Connections and New Directions (Jones, M. and Sheppard P., eds), pp. 413 –
447, Research in Anthropology and Linguistics No 5, University of
Auckland
Austin, C.C. (1999) Lizards took express train to Polynesia. Nature
397, 113 – 114
Matisoo-Smith, E. and Allen, J.S. (2001) Name that rat: molecular and
morphological identification of pacific rodent remains. Int.
J. Osteoarch. 11, 34– 42
Matisoo-Smith, E.A. and Robins, J.H. (2003) The commensal model
http://tree.trends.com
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
Vol.18 No.10 October 2003
continues – Rattus exulans mtDNA variation from Island Southeast
Asia through to Polynesia. Am. J. Phys. Anthrop. 31 (Suppl.), 148
Holdaway, R.N. (1996) Arrival of rats in New Zealand. Nature 384,
225– 226
Steadman, D.W. (1995) Prehistoric extinctions of Pacific island birds:
biodiversity meets zooarchaeology. Science 267, 1123 – 1131
Steadman, D.W. et al. (2002) Rapid prehistoric extinction of iguanas
and birds in Polynesia. Proc. Natl. Acad. Sci. U. S. A. 99, 3673– 3677
Brook, B.W. and Bowman, D.M.J.S. (2002) Explaining the Pleistocene
megafaunal extinctions: models, chronologies and assumptions. Proc.
Natl. Acad. Sci. U. S. A. 99, 14624 – 14627
Grayson, D.K. (2001) The archaeological record of human impacts on
animal populations. J. World Prehist. 15, 1 – 68
Worthy, T.H. et al. (1999) Megafaunal expression in a land without
mammals – the first fossil faunas from terrestrial deposits in Fiji.
Sencken. Biol. 79, 237 – 242
Holdaway, R.N. and Jacomb, C. (2000) Rapid extinction of the moas
(Aves: Dinornithiformes): model, test, and implications. Science 287,
2250– 2254
McGlone, M.S. and Wilmshurst, J.M. (1999) Dating initial Maori
environmental impact in New Zealand. Quat. Int. 59, 5 – 16
Worthy, T.H. and Holdaway, R.N. (2002) The Lost World of the Moa,
Indiana University Press
Duncan, R.P. et al. (2002) Prehistoric bird extinctions and human
hunting. Proc. R. Soc. London B Biol. Sci. 269, 517 – 529
Johnson, C.N. (2002) Determinants of loss of mammal species during
the late Quaternary ‘megafauna’ extinctions: life history and ecology,
but not body size. Proc. R. Soc. Lond. Ser. B 269, 2221 – 2227
Flannery, T.F. (1994) The Future Eaters, Reed, Melbourne
Anderson, A. (2000) Differential reliability of 14C AMS ages of Rattus
exulans bone gelatin in South Pacific prehistory. J. R. Soc. N. Z. 30,
243– 261
Anderson, A. (2001) No meat on that beautiful shore. Prehistoric
abandonment of subtropical Polynesian islands. Int. J. Osteoarch. 11,
14 – 23
Kirch, P.V. and Green, R.C. (2001) Hawaiki, Ancestral Polynesia,
Cambridge University Press
Reid, A. (1995) Continuity and change in the Austronesian transition
to Islam and Christianity. In The Austronesians: Historical and
Comparative Aspects (Bellwood, P. et al., eds), pp. 314 – 331, Australian
National University
Paracchini, S. et al. (2002) Hierarchical high-throughput SNP
genotyping of the human Y chromosome using MALDI-TOF mass
spectrometry. Nucleic Acids Res. 30, e27
Gabriel, S.B. et al. (2002) The structure of haplotype blocks in the
human genome. Science 269, 2225 – 2229
Sabeti, P.C. et al. (2002) Detecting recent positive selection in the
human genome from haplotype structure. Nature 419, 832 – 837
Golson, J. (2002) Gourds in New Guinea, Asia and the Pacific. In Fifty
Years in the Field (Bedford, S. et al., eds), pp. 69– 78, N. Z.
Archaeological Assoc., Monograph 25
Irwin, G. (1992) The Prehistoric Exploration and Colonisation of the
Pacific, Cambridge University Press
Biggs, B. (1998) POLLEX: Proto Polynesian Lexicon, University of
Auckland
Bellwood, P. (2002) Early agricultural population diasporas? Farming,
languages and genes. Annu. Rev. Anthropol. 30, 181 – 207
Serjeantson, S.W. and Gao, X. (1995) Homo sapiens is an evolving
species: origins of the Austronesians. In The Austronesians: Historical
and Comparative Aspects (Bellwood, P., ed.), pp. 165– 180, Australian
National University, Canberra
Terrell, J. et al. (2001) Update from foregone conclusions – in search of
‘Papuans’ and ‘Austronesians’. Curr. Anthropol. 42, 97 – 124
Crowley, T. (1997) An Introduction to Historical Linguistics, 3rd edn,
Oxford University Press
Harvey, C.F. et al. (1997) Assessing relationships of kumara cultivars
by RAPD analysis. N. Z. J. Bot. 35, 479– 485
Hather, J. and Kirch, P.V. (1991) Prehistoric sweet potato (Ipomoea
batatas) from Mangaia Island, central Polynesia. Antiquity 65,
887– 893