Content uploaded by Ethan E Cochrane
Author content
All content in this area was uploaded by Ethan E Cochrane on Apr 04, 2018
Content may be subject to copyright.
https://doi.org/10.1177/0959683617714599
The Holocene
1 –12
© The Author(s) 2017
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0959683617714599
journals.sagepub.com/home/hol
Introduction
The initial migration of humans into Remote Oceania approxi-
mately 3000 years ago is of singular importance in Oceanic and
Polynesian prehistory. After at least 40,000 years in the islands
around New Guinea, voyagers sailed east and colonized Remote
Oceania, from Vanuatu in the west to Samoa and Tonga in the east,
islands within about 2,000,000 km2 of ocean and previously with-
out human inhabitants (Figure 1a). This migration was exceed-
ingly fast, occurring within a few hundred years (Sheppard et al.,
2015) and resulting in over 130 known archaeological sites
(Anderson et al., 2001; Bedford and Sand, 2007). These popula-
tions made an intricately decorated pottery called Lapita, and the
eponymous Lapita peoples (Kirch, 1997) were the likely ancestors
of the people who later colonized the archipelagos from Hawaii
in the north, Rapa Nui in the east, and New Zealand in the south
(cf. Addison and Matisoo-Smith, 2010; Skoglund et al., 2016).
The process that explains both the timing and choice of Lapita
site locations in Remote Oceania is largely unknown or, perhaps
more accurately, untested. Aside from a focus on small islands
and beach ridges with reef resources (Dickinson, 2014; Lepofsky,
1988; Nunn and Heorake, 2009), it is not clear whether groups
privileged coastlines that provided arable land or whether groups
instead focused on areas with easily exploited marine resources
(Burley, 2012). It is also possible that initial settlements were cho-
sen to be easily accessible to other archipelagos or islands (Hunt,
1988) or that settlement decisions were influenced by rapidly
filled niches, such that later arrivals had to settle in less desirable
locations (Kennett et al., 2006). The distribution of Remote Oce-
anic Lapita settlement locations is likely explained by a variable
combination of these possibilities, and determining which are
most likely in specific instances requires precise and accurate
reconstruction of the paleolandforms available to Lapita migrants.
In particular, knowledge of the timing of beach ridge formation,
the size of coastal landforms, the relative contribution of terrige-
nous and marine sediments, and the potential availability of near-
shore marine taxa is necessary to explain chronological and
spatial variation in Lapita settlement locations.
In this paper, we contribute to the explanation of Lapita settle-
ment locations by examining how the mid- to late-Holocene sea-
level retreat (Mitrovica and Peltier, 1991) may have influenced
the timing of coastal progradation, the development of habitable
island landscapes, and the initial settlement of a small Fijian
Archaeological and sedimentological
data indicate Lapita settlement on a
newly formed coastal plain: Tavua Island,
Mamanuca Group, Fiji
Alex E Morrison,1,2 Ethan E Cochrane,1 Timothy Rieth2 and Mark
Horrocks1,3
Abstract
The timing and choice of initial settlement location are examined on the small island of Tavua in Fiji’s Mamanuca Group. The mid- to late-Holocene sea-
level retreat influenced the island’s coastal landforms through the acceleration of coastal progradation and the production of habitable land. Archaeological,
sedimentological, and chronological data are integrated to better understand the island’s settlement and geomorphological history. These datasets are
then compared with regional and modeled sea-level curves for Fiji in order to constrain the time period for the onset of coastal regression. The results
indicate that Tavua was initially settled around 3000 years ago, within a few centuries of the formation of the coastal plain. Integrating archaeological,
sedimentological, and sea-level datasets helps produce a more precise understanding of the relationship between sea-level change and the timing of
settlement on small islands in Oceania.
Keywords
Bayesian models, coastal geomorphology, Fiji, geoarchaeology, landscape change, Oceanic Islands, sea-level change
Received 9 December 2016; revised manuscript accepted 13 April 2017
1Anthropology, School of Social Sciences, The University of Auckland,
New Zealand
2International Archaeological Research Institute, USA
3Microfossil Research Ltd, New Zealand
Corresponding author:
Alex E Morrison, Anthropology, School of Social Sciences, The
University of Auckland, Private Bag 92019, Auckland 1142, New
Zealand.
Email: alex.morrison@auckland.ac.nz
714599HOL0010.1177/0959683617714599The HoloceneMorrison et al.
research-article2017
Research paper
2 The Holocene 00(0)
Island, Tavua, in the Mamanuca Group (Figure 1). Tavua’s first
colonists arrived nearly 3000 years ago (Cochrane et al., 2007,
2011), during a time period when sea level was likely over 1 m
higher than present (Lal and Nunn, 2011; Morrison and Cochrane,
2008; Nunn, 2005, 2009; Nunn and Peltier, 2001). We integrate
archaeological, sedimentological, elevational, and chronological
data to better understand geomorphological evolution at Tavua,
and we address the timing of the formation of the Tavua coastal
plain upon which Lapita colonists initially settled. We compare
these results against regional and modeled sea-level data in order
to constrain the time period for the onset of coastal regression.
Our results provide an example of how archaeological, sedimen-
tological, and sea-level data can be integrated to better understand
geomorphological response to sea-level retreat and its possible
relationship with the settlement of small islands in Oceania. In the
next section, we briefly review Fiji’s regional sea-level history
which provides context for the subsequent study.
Sea-level history in Fiji
The Fijian Islands are located in the southern tropics of the Pacific
Ocean within an intricate geologic zone framed by the boundaries
of the Pacific and Indo-Australian plates. A number of micro-
plates are also present in the region, and as a result of this tectonic
diversity, different islands in the Fijian archipelago display paleo-
sea-level histories related to changes in eustatic sea level and
local tectonic processes (e.g. Dickinson, 2001; Miyata et al.,
1990; Nunn, 1998; Nunn and Peltier, 2001; Nunn et al., 2002).
Over the past three decades, a variety of field studies and sim-
ulation models indicate that sea level in the equatorial Pacific
Basin between 5000 and 3000 cal. BP was 1–3 m higher than
present mean sea level (MSL; Dickinson, 2001; Fletcher and
Jones, 1996; Goodwin and Harvey, 2008; Grossman et al., 1998;
Nunn and Peltier, 2001). Although the timing and magnitude of
sea-level fluctuation may have varied within and across island
groups (Grossman et al., 1998; Woodroffe et al., 2012), research
from the Fijian archipelago indicates that while local tectonic
subsidence and uplift influenced relative paleosea level in some
settings, in general, the Mamanuca Group has been tectonically
stable during the Quaternary (McKoy et al., 2010: 62; Nunn,
1990, 1998), Additionally, mid-Holocene sea-level reconstruc-
tions support elevation estimates from the ICE-4G model (Peltier,
1995) which predicts a sea-level maximum of ~+2.1 m in Fiji
around ~4000 cal. BP (Nunn and Peltier, 2001).
While the timing of the mid-Holocene sea-level retreat is not
precisely known and likely varies at local spatial scales, based on
these paleosea-level reconstructions, it is probable that the initial
Lapita colonizers who arrived in Fiji at ~3000 cal. BP would have
encountered very different island settings from today. Nunn and Pel-
tier (2001) compare extensive field evidence of the timing and mag-
nitude of mid-Holocene paleosea level from over 40 locations across
the Fijian archipelago. The majority of these measurements are
made on coral microatolls (Porites sp.) which provide precise and
accurate indications of past sea level, often with ranges as low as 3
cm (Allen et al., 2016; Goodwin and Harvey, 2008; Smithers and
Woodroffe, 2000; Woodroffe et al., 2012). After taking locally vari-
able tectonic subsidence and uplift into consideration, these data
indicate that paleosea level in the region potentially reached a single
maximum height of +2.19 m, between 5650 and 3200 cal. BP. It is
also possible that Fiji experienced two paleosea-level maxima, one
of +0.75–1.85 m (~6100–4550 cal. BP) and another of +0.90–2.46
m between 3590 and 2800 cal. BP (Nunn and Peltier, 2001: 203).
Although research on paleosea level and coastal geomorphology has
been limited in Fiji’s Mamanuca islands, McKoy et al. (2010), docu-
ment the formation of reef platform islands close to Tavua. Their
research demonstrates that the formation of low-lying cays occurred
only after the retreat of the mid-Holocene sea-level high stand, thus
demonstrating the profound impact that sea-level drawdown had on
coastal island landscapes in the area.
Site description
The archaeological deposits on Tavua Island’s main coastal
plain are directly behind the single inhabited village on the Island
(Figure 1b). The site, called simply Tavua Village (K27-4 in the
Fiji site registration scheme), is on a moderately sized beach flat
Figure 1. (a) Tavua Island in the Mamanuca Group, Fiji. (b) Tavua Village (Site K27-4). The red dot shows the approximate location of test unit
excavations and the extent of Lapita ceramics.
Morrison et al. 3
(550 m × 300 m) surrounded by steep hillsides on the north, east,
and south sides. Four other beach flats, separated by headlands,
ring the island. The Tavua Village site was located after noting the
presence of subsurface burned sediment, prehistoric pottery, and
shell midden eroding from the sidewall of a sand-mining pit dug
by people in the village. In 2006, a test unit (TU1) was placed
approximately 2 m from the sand pit. In 2009, a coring program
was initiated (see below) and four more excavation units were
placed in a rough line stretching approximately 20 m west from the
original 2006 excavation. Currently, the island is surrounded by
reef flat on the west, south, and east sides, extending up to approxi-
mately 600 m on the west side, in front of Tavua Village. Behind
the village there is a low-lying back beach area that is often inun-
dated by brackish water during the wet season, November to April.
Methods
Archaeological excavations
In total, five test units were excavated with shovels and trowels in
up to 10 cm thick arbitrary levels within natural stratigraphic units
(Figure 3). All sediment was passed through 3 mm (1/8 inch) screens
with cultural materials removed from screens. Cultural materials,
including lithic artifacts, midden shell, bone, and ceramics, were
recovered from all test units, from the surface up to a maximum
depth of 175 cm below the surface (cmbs). Ceramics with Lapita
designs were recovered as deep as 154 cmbs in some excavation
units (Cochrane et al., 2007, 2011). In general, Lapita designs com-
prise an intricate set of repeated motifs, typically produced by den-
tate stamps and found throughout the Bismarck Archipelago near
New Guinea, the Solomon Islands, Vanuatu, New Caledonia, Fiji,
Tonga, and Samoa (Kirch, 1997). In the islands of Remote Oceania,
from Vanuatu to Samoa (Figure 1a, inset), similar Lapita designs are
shared across archipelagos indicating interaction between local pop-
ulations (Cochrane and Lipo, 2010; Green, 2003). In Fiji, these
designs were made for about 500 years between 3000 and 2500 cal.
BP (Cochrane, in press), roughly the same duration as in Vanuatu
and New Caledonia, but a few hundred years longer than in Tonga
and Samoa.
The site stratigraphy (Figure 2, Table 1) consists of a calcare-
ous sand basal deposit with more recent strata containing higher
proportions of terrigenous grains. Strata are separated by bound-
aries of varying distinctness indicating abrupt to gradual transi-
tions between deposits. Post molds are clearly visible, excavated
into Layer V, and excavated at the Layer III–II transition. Test
Units 3 and 4 were placed approximately 10 m from Test Units 1,
2, and 5. Test Unit 3 contained a series of cultivation features that
were not encountered in the other test units.
Figure 2. Elevation, description, and soil horizon characteristics of stratigraphic units identified in Test Unit 2. Particle grain size descriptions
are provided in Table 3. Sediment descriptions are presented in Table 1.
Table 1. Description of stratigraphic units encountered in test excavation units 1–5.
Unit Color Sediment texture Structure (grade,
size, shape)
Consistence
(moist)
Non-cultural clasts* Cultural materials
1 7.5 YR 2.5/1; black Sandy clay loam Weak, very fine,
subangular, blocky
structure
Firm Minor percentage of
gravels and pebbles
Ceramics, shell,
fauna remains
2 7.5 YR 4/2; brown Sandy clay loam Weak, very fine,
granular structure
Loose Minor percentage of
cobbles and pebbles
Ceramics, shell,
fauna remains
3 10 YR 6/4; light yellowish-brown Loamy sand Weak, very fine,
granular structure
Loose Minor percentage of
pebbles
Ceramics, shell,
fauna remains
4 10 YR 5/2; grayish-brown Loamy sand Weak, very fine,
granular structure
Friable Minor percentage of
pebbles
Ceramics, shell,
fauna remains
5 2.5 YR 8/4; pale yellow Coarse sand Structureless, fine,
granular structure
Loose Minor percentage of
gravels and pebbles
Ceramics, shell,
fauna remains
6 2.5 YR 8/4; pale yellow Very coarse to
coarse sand
Structureless, fine,
granular structure
Loose High percentage of coral-
line gravels and pebbles
None
*The term ‘clast’ refers to a fragment of rock or carbonate material broken down from a larger rock or carbonate source.
4 The Holocene 00(0)
In addition to the five archaeological test units, a single 2 ×
1 m trench (Inland Trench 1) was excavated inland of the pri-
mary excavation locale (see Figure 3). Inland Trench 1 was
placed in an area characterized by a depression between 0.50
and 1.0 m lower than the surrounding topography that cur-
rently becomes inundated during the rainy months. The main
reason for conducting the inland excavation was to investigate
the subsurface of the depressed area for archaeological materi-
als and to also characterize the sedimentary deposits.
Test excavation units 1–5 revealed a sequence of six strati-
graphic units (Table 1) consistent with a transition from strata
dominated by calcareous sand-sized particles to the incorpora-
tion of higher percentages of terrigenous silts and clays. Layers
I–VI were identified in the field as different depositional units
based on color, texture, artifact density, and artifact content.
Layer VI represents culturally sterile very coarse to coarse
grain beach sand with a high percentage of gravel-sized coral
and mollusk fragments. There is no evidence for pedogenic
Figure 3. (a) Digital elevation model (DEM) with the location of auger cores and excavation units (note that elevation error is estimated at
15 cm). (b) A generalized profile of the depositional sequence at Tavua Beach.
Morrison et al. 5
processes, and based on the origin and size of the particles and
the lack of cultural material, the deposit formed as a result of
ocean-derived transport. Layer V is the initial stratum contain-
ing evidence of human activity in the form of Lapita ceramic
sherds and other deposited cultural material. The sedimentary
characteristics are similar to Layer VI and the majority of par-
ticles are coarse-grained coralline sand. Layer IV is loamy sand
containing an increase in silt- and clay-sized particles. Layer IV
contains abundant ceramics, shell and bone, and other artifacts.
Layer III, a loamy sand, is not present in all test units, but
appears to be partially a product of freshwater flooding perhaps
from the back beach area, anthropogenic deposition of organic
material, and soil formation. The artifact content in Layer III is
much lower (in absolute and volumetric terms) compared with
Layers IV and II (Cochrane et al., 2011). Layer II, a sandy
clay loam, contains abundant ceramics, shell, and vertebrate
remains. Layer I represents an O horizon, dominated by decom-
posing organic material and formed through modern gardening
activities.
Geological coring
A primary goal of the 2009 fieldwork was to acquire a better under-
standing of the environmental context for the Lapita deposit dis-
covered in 2006. In addition to the test excavation units, 39 auger
cores were excavated along the paleobeach ridge behind the village
and in the vicinity of the Lapita deposit, as well as in select loca-
tions across the coastal plain (Figure 3). These cores were exca-
vated using a hand-driven auger outfitted with a 10-cm bucket
designed for the collection of loose sandy sediments. Each 10 cm
sediment sample was analyzed for sediment texture, grain mor-
phology, size, color, and composition (following USDA Kellogg
Soil Survey Manual, 2014; Soil Survey Staff 2014). Formal litho-
logical and pedological characteristics were identified for all auger
cores and stratigraphic changes, and boundary depths were recorded
relative to surface elevations. Representative sediment samples
from each auger core were photographed in the field and collected
for further laboratory analyses. Sediment particle size analysis was
conducted to determine the transport agent and depositional history
at key locations along the coastal plain. Sediment samples were
first mechanically sieved using a 6-mm (¼ inch) screen and pan.
The samples were vortexed using a MT19 Vortex machine and a
pipette was used to squirt the soil from the samples into a Master-
sizer 2000 particle analyzer.
The entire suite of auger core samples was evaluated to
reconstruct the sequence of sediment source, transport agent,
environment of deposition, and post-depositional alterations
(following Stein, 2001). The geographical coordinates and ele-
vation information for each auger core were collected with a
Leica electronic distance measurement (EDM) total station with
an estimated accuracy of ~±15 cm based on repeat measure-
ments on a stationary datum. The vertical error is largely a result
of continuous movement of the total station in order to ensure a
clear line of sight through vegetation.
Acquisition of modern topographic information for the Tavua
coastal plain in 2009 was conducted through the collection of ele-
vation points distributed across the coastal plain and back beach
area. The topographic survey transects stretched from the modern
low watermark to approximately 500 m inland. Since the tidal
relationship between Tavua and a local tidal reference gauge is
unknown, each subsurface stratigraphic unit recorded in every
core was then adjusted relative to low-tide recorded at the site on 4
August and 12 August 2009. These tidal positions were then
adjusted to MSL as recorded at the Suva Tide Station, Suva Harbor
(Tide Tables, 2009). The post-processing procedures resulted in a
set of 39 adjusted auger core profiles and six excavation unit pro-
files with adjusted surface and stratigraphic elevation heights. The
adjusted elevation data were used to generate a digital elevation
model (DEM) using a geostatistical kriging approach (Figure 3a).
Results
Site chronology
Seven radiocarbon dates (Table 2) were obtained from the Tavua sub-
surface excavations: four dates from TU1, one date each from TUs 2
and 3, and a single date from Core 26 (see also Cochrane et al., 2011).
Samples from these proveniences were chosen to provide estimates
for the initial colonization of Tavua Island and to infer geomorphic
response to sea-level retreat on Tavua. While coastal plain sediments
cannot be used to establish precise relationships with past sea level,
subsurface carbonates are valuable for determining the chronology of
shoreline regression when paired with local and regional sea-level
curves (following Kane et al., 2017). These chronometric determi-
nations were used to create a Bayesian model in Oxcal 4.2 (Bronk
Ramsey, 2009, 2013). Following Nunn and Petchey (2013), terres-
trial samples were calibrated using the Northern Hemisphere calibra-
tion curve (IntCal13; Reimer et al., 2013) because of Fiji’s location
within the South Pacific Convergence Zone. The marine samples
were calibrated using Marine13 (Reimer et al., 2013) with a local
delta R value of 11 ± 26 (Petchey et al., 2008).
The model uses Oxcal’s Sequence, Phase, Boundary, and Dif-
ference commands and has three phases: (1) a pre-cultural phase
(Layer VI), (2) an initial Lapita deposit phase (Layer V), and (3)
a final Lapita deposit phase (Layer IV) (code provided as
Table 2. Radiocarbon dates from geological and archaeological investigations on Tavua, organized by Bayesian calibration model phases.
Bayesian calibration
model phase
Location Material δ13C 14C age BP Lab no.
Un-modeled K27-4, Core 26, Site Layer VI,
155–165 cmbs
Non-abraded coral clast – 2816 ± 20 Wk-40061
Un-modeled K27-4, TU1, Site Layer II, level 4,
37–47 cmbs
Trochus sp. (hinge) 2.4 2832 ± 30 Wk-20392
Final Lapita K27-4, TU1, level 8, Site Layer IV,
77–87 cmbs
Unidentified charcoal residue
adhering to ceramic
−24.3 2536 ± 35 AA-73317
Initial Lapita K27-4, TU 3, Site Layer V, level 16,
163 cmbs
Small-diameter wood charcoal -27.6 2762 ± 28 Wk-37135
Initial Lapita K27-4, TU 1, Site Layer V, level 11,
92–106 cmbs
Unidentified wood charcoal −24.8 2693 ± 39 AA-73315
Initial Lapita K27-4, TU 1, Site Layer V, level 13,
116–126 cmbs
Unidentified wood charcoal −27.8 2850 ± 36 AA-73316
Pre-cultural K27-4, TU 2, Site Layer VI, level 17,
140 cmbs
Non-abraded coral clast – 3294 ± 21 Wk-40062
6 The Holocene 00(0)
Supplementary Material, Appendix A, available online). The
phases are placed in contiguous order because there is no strati-
graphic or artifact evidence of a hiatus between the strata.
Absolute chronometric information for the pre-cultural
phase is based on one radiocarbon date (Wk-40062) obtained
from a non-abraded coral fragment (Figure 4a) collected from
sterile sediment (Layer VI) immediately below the earliest cul-
tural stratum (Layer V). Assuming there is a close relationship
between the timing of the death of the coral and its deposition,
the associated radiocarbon date is indicative of a period when
the area near the test units was within the tidal range and there-
fore tracks the timing of coastal beach progradation. The initial
Lapita phase includes three determinations from unidentified
wood charcoal samples (AA-73315, AA-73316, and Wk-37135)
obtained from TU 1 and 3 Layer V. The final Lapita phase has
a single determination from organic residue on a sherd (AA-
73317) that was collected from TU 1 Layer IV.
Wk-40061 (Figure 4b) is a pre-cultural non-abraded coral
fragment that was not included in the Bayesian model because
the relative vertical relationship of the stratum from which it
was collected cannot be confidently correlated with the other
excavation units. However, the date provides supplementary
information regarding the timing of coastal progradation. An
additional date, Wk-20392, obtained from a Trochus sp. shell
collected from TU 1 Layer II is significantly too old for this
stratum based on stratigraphy and pottery analysis and was
therefore not included in the model.
It is important to note that since the Bayesian model is par-
tially based on a number of unidentified charcoal samples, date
ranges may be influenced by in-built age on the order of 100
years or more and the modeled date ranges may be too old by a
similar magnitude. For example, the Lapita motifs on the Tavua
sherds are typical of Late Lapita assemblages in Fiji and West
Polynesia with simplified designs lacking the densely packed
Figure 4. (a) Sample Wk-40062 is an un-abraded coral fragment collected below the depth of initial Lapita occupation of Tavua Island. The
modeled date of 3181–2917 cal. BP (95.4%) provides a temporal estimate for when ocean energy was still actively depositing material in the
future location of the Lapita settlement. (b) Sample Wk-40061 is a coral fragment with limited abrasion recovered from Core 26 along the
Tavua coastal plain. The un-modeled radiocarbon date of 2689–2423 cal. BP (95.4%) provides a temporal estimate for when the coastal plain
had prograded 38 horizontal meters from the Lapita settlement.
Morrison et al. 7
impressions of earlier Lapita assemblages (Burley and Con-
naughton, 2010; Kirch, 1997).
Figure 5 presents the structure of the Bayesian model and the
posterior density estimates. The model shows good agreement
between the radiocarbon dates, the stratigraphy, and phasing
(Amodel = 100.4). Modeled results are italicized in the text to distin-
guish these results from un-modeled dates. The modeled chronol-
ogy differs slightly from Cochrane et al.’s (2011) initial
presentation of the site’s chronology, which was based on simple
calibration of individual dates.
Based on the age determinations from the coral sample and
earliest culturally associated date, the model estimates that the
paleobeach around TUs 1 and 3 had stabilized by 3133–2871 cal.
BP (95.4%; Figure 5 ‘Boundary end pre-cultural/start Lapita’)
and probably by 3061–2915 cal. BP (68.2%). This corresponds
with the earliest use of the coastal plain at this location as identi-
fied by the presence of dentate stamped and plain ceramic sherds,
charcoal, and a small amount of midden recovered from TU 1
and 3 Layer V. This estimate places the Tavua Village deposit as
slightly older than or roughly contemporaneous with Naigani
(VL 21/5) and older than Bourewa (Sheppard et al., 2015). How-
ever, this relative placement of Tavua within Fiji’s Lapita coloni-
zation sequence requires confirmation through the dating of
identified short-lived charcoal specimens and/or appropriate
marine shell samples, a point also recently stressed by Nunn and
Petchey (2013). Dated Lapita activity (i.e. production of dentate
stamped ceramics and Lapita vessel forms) ended during 2751–
2026 cal. BP (95.4%; Figure 5 ‘Boundary end Lapita’) and prob-
ably 2739–2507 cal. BP (68.2%). Lapita occupation spanned
41–457 years (95.4%) or more likely 135–337 years (68.2%).
During the interval 2689–2423 cal. BP (95.4%) and probably
2646–2496 cal. BP (68.2%) (based on the un-modeled coral date
from Core 26; Wk-40061), the coastal plain had prograded at
least 38 m established by the horizontal location of the sample
along the Tavua coastal plain.
Geological coring results
In general, there are three to four depositional units present across
the coastal plain; however, the characteristics of these strata differ
slightly depending on the location of the cores with more seaward
cores containing less developed cultural deposits and soils. Strati-
graphic units were correlated between all auger cores and excava-
tion units where possible. Figure 3b depicts the morphological
cross section at Tavua. Stratigraphic Unit 3 (TU 2, Layer VI), the
basal beach sediment, is well-sorted, very coarse to
Figure 5. Probability distributions of dates from Tavua. Modeled probability distributions are solid black, while the un-modeled calibrated
dates are in outline.
8 The Holocene 00(0)
coarse, yellow sand deposited by high-energy waves with a minor
contribution of fine fraction and no cultural material. Strati-
graphic Unit 2 (TU 2, Layers IV–V) is medium to coarse grain
yellow sand deposited by a combination of wave- and wind-
derived energy. Finally, Stratigraphic Unit 1(TU 2, Layers I–II)
consists of predominately sandy loam sediments and modern
developed soils.
Figure 3b shows the location of the two dated coral samples
deposited within the coastal plain, which provide evidence of
the general timing of coastal plain progradation revealed by the
withdrawal of the mid-Holocene sea-level high stand. Sample
WK40062 (3294 ± 21 BP) is located below the initial Lapita
deposit at Tavua at an elevation of approximately 1.75 m above
MSL (±15 cm). While the dated coral clasts are not in growth
position and therefore cannot provide a precise and reliable
proxy for sea level, the vertical position, chronology, and sedi-
mentological characteristic are generally consistent with the
current understanding of mid- to late-Holocene sea level in the
Fijian Islands and the South Pacific and its effect on the devel-
opment of coastal plains (Kane et al., 2017; Allen et al., 2016).
Based on the date of the sample and its position subordinate to
the Lapita deposit, within a few centuries of 3181–2917 cal. BP,
sea level had fallen sufficiently such that the coastal deposit had
stabilized and Lapita colonizers established a settlement on the
island. Prior to this period, the nature of the sample indicates
that the area surrounding the excavation units was close to an
active marine environment. While it is possible that evidence of
earlier settlement could still exist, perhaps deeply buried inland
of the excavated area, additional auger cores, and exploratory
trenches excavated inland of the test units failed to uncover evi-
dence of earlier settlement, and in fact the back beach area is
devoid of artefactual remains.
An additional corroborating date useful for tracking the rela-
tionship between sea-level retreat and coastal plain progradation
at Tavua is provided by sample WK40061 (2816 ± 20 BP) recov-
ered from Core 26. The sample of non-abraded coral was recov-
ered at an elevation of 1.9 m (±15 cm) above MSL and provides
an un-modeled date range of 2689–2423 cal. BP (95.4%). The
location of the sample approximately 38 m seaward of Test Unit 2
and its date relative to the chronometric estimate of initial occupa-
tion of the coastal plain, as well as coral sample WK40062, indi-
cate that the coastal plain continued prograding during the
intervening ~250–650 years.
Particle size analysis results
To formally examine the depositional change in the vicinity of
the Lapita deposit, we used particle size analysis to document
the source, agent of deposition, and depositional environment
of the sediment recovered in Test Unit 2 and an inland geologi-
cal trench (Inland Trench 1; Table 3). Layer V, the initial deposit
with Lapita pottery, represents coarse to very coarse marine
sand similar in composition to Layer VI, the underlying sterile
deposit. Approximately 87% of Layer V consists of very coarse
to coarse calcareous sand, 10% is medium sand, and around 3%
is made up of fine-grained silts and clays likely deposited by
anthropogenic activity and/or soil forming processes. The sand
fraction of the deposit is calcareous and was derived from a
marine source.
Layer IV is approximately 54% medium- to coarse-grained
sand with only a ~9% contribution of very coarse-grained sand.
Very fine and fine sand increases when compared with Layer V
to ~12%. The overall percentage of silt and clay increases sig-
nificantly to nearly 25%. The majority of Layer IV came from
a marine source; however, the increase in medium to fine-
grained sand indicates that the calcareous marine component of
the deposit was likely deposited by wind and humans rather
than solely by wave and tidal energy. Furthermore, the high
percentage of silts and clays points to a transformation of a
predominately marine environment to more input from terrige-
nous sources.
Layer III is approximately 68% coarse to very coarse sand
and is therefore similar in composition to Layer V. An addi-
tional 24% of Layer III is made up of medium sand and there is
only a very small (~8%) contribution from fine and very fine
sand, silt, and clay. The high percentage of coarse-grained sand
suggests that Layer III may represent an intrusive high wave
energy event that inundated the area. Furthermore, Layer III is
only present in TUs 2 and 5 suggesting that it was a fairly local-
ized depositional event.
Layer II is approximately 56% medium- to coarse-grained
sand and 18% very coarse-grained sand. Very fine and fine sand
comprises nearly 7%, while silts and clays contribute 19% to the
deposit. Layer II is similar to Layer IV and is likely derived from
wind energy and anthropogenic deposition of silt and clay parti-
cles into the calcareous sediment.
Layer I is over 50% medium- to coarse-grained sand with
very coarse sand contributing approximately 22%. Very fine
and fine sand is approximately 5%, while silts and clay are over
21%. The contribution of different grain sizes is very similar to
Layer II.
While the progradation and formation of a sandy coastal ter-
race seaward of the Lapita deposit undoubtedly protected the
settlement from tidal inundation and periodic destruction from
extreme tides and storms, perhaps more significant were the con-
sequences for the back beach area which is a locus for the island’s
horticultural activity today. The 2 × 1 m trench excavated approx-
imately 10 m inland from the test units allows us to examine how
regional sea-level decline affected the formation of the back
beach area. The surface of the inland trench resides at 2.80 m (±15
cm) above MSL in an area of depressed elevation relative to the
location of the Lapita deposit. Sediment samples from three strati-
graphic units uncovered in the trench were also analyzed using a
laser particle size analyzer.
The sediment particle size for Layer III of the inland trench
indicates that at some point during the mid Holocene, this area
Table 3. Sediment particle size analyses results for Test Unit 2 and Inland Trench 1 (percentage contributions per phi size).
Class Clay Very fine silt Fine silt Medium silt Coarse silt Very fine sand Fine sand Medium sand Coarse sand Very coarse sand
ϕ scale 10 to 8 8 to 7 7 to 6 6 to 5 5 to 4 4 to 3 3 to 2 2 to 1 1 to 0 0 to −1
TU2 Layer I 4.35 5.07 5.3 4.3 3.26 2.56 2.38 14.51 35.98 22.23
TU2 Layer II 3.29 3.44 4.08 4.06 4.14 3.41 3.38 19.05 36.71 18.4
TU2 Layer III 1.16 1.21 1.18 0.75 0.73 0.65 2.22 23.55 45.73 22.76
TU2 Layer IV 6.94 5.17 5.39 4.11 3.25 3.25 8.88 25.99 28.29 8.67
TU2 Layer V 0.498 0.54 0.61 0.54 0.031 0 0 10 49.28 38.48
ITU Layer I 13.5 13.26 13.84 12.57 9.57 5.98 3.9 6.56 13.27 7.54
ITU Layer II 20.01 8.03 6.64 4.2 2.39 2.14 0.54 4.82 30.63 20.6
ITU Layer III 0.63 0.45 0.42 0.23 0.01 0.72 1.37 16.72 49.1 30.35
Morrison et al. 9
was inundated with marine water. Layer III comprises predomi-
nately (~80%) coarse to very coarse sand with only a minor
contribution (>2%) of fine-grained silts and clays. The size and
composition of Layer III are nearly identical to Layer VI in Test
Unit 2; however, Layer III ends at approximately 1.30 m above
MSL which is somewhat lower than the termination of the ster-
ile very coarse sand layer in the excavation units. Additional
particle size analysis for Layer II of the inland trench docu-
ments a decline in the relative contribution of very coarse- and
coarse-grained sand particles and an increase in fine- to very
fine-grained silt and clay fractions. This pattern is indicative of
the infilling of a marine environment concomitant with retreat-
ing sea level, the formation of the beach ridge, and coastal pro-
gradation. It is important to note that the depth of Layer III
~1.30 m above MSL, the depressed surface elevation of the sur-
rounding area, and the lack of cultural material recovered from
excavation trenches and auger cores in the vicinity all suggest
that the back beach area probably infilled with terrestrial-based
sediment sometimes after the formation of the coastal plain
upon which the Lapita deposit resides. This scenario would be
consistent with the presence of an embayment near the Lapita
settlement; however, this hypothesis needs to be examined
through further chronometric analysis.
Discussion
Together, the archaeological and sedimentological data indi-
cate that coastal progradation on Tavua Island had commenced
by 3133–2871 cal. BP based on the Bayesian model estimate
for the boundary between the end of the pre-cultural coral frag-
ment date and the beginning of the initial Lapita date. Although
there is an unknown lag between the death of the coral and its
deposition, the overlying radiocarbon date recovered from the
Lapita deposit constrains it, ensuring that its deposition could
not have occurred any later than the estimate from the radio-
carbon sample above. This is one of the advantages of integrat-
ing coral fragment dates with archaeological dates in a formal
Bayesian model.
A second radiocarbon date on a coral sample approximately
38 m seaward of Test Unit 2 provides an independent estimate of
2690–2420 cal. BP depicting a general increasing seaward–inland
age relationship among the cores. In order to understand the tim-
ing of coastal progradation and its relationship with sea-level his-
tory, the radiocarbon samples from both coral samples, the suite
of cultural radiocarbon dates, and subsurface carbonate sediment
dates from a number of nearby low-lying reef islands in the
Mamanuca Group (McKoy et al., 2010) are compared with mod-
eled and regional sea-level curves for Fiji (Figure 6).
Nunn’s (1990) sea-level curve and the ICE-4G curve of Peltier
(1995) both predict that the mid- to late-Holocene sea-level high-
stand in Fiji reached its maximum height around 4000 years ago
(Figure 6). The archaeological and sedimentological data demon-
strate that the coastal plain began prograding no later than 3133–2871
cal. BP, although it is possible that progradation had commenced
earlier. The timing of coastal progradation and its relationship with
sea-level retreat are also confirmed by additional sedimentary data
reported in McKoy et al. (2010) who found that nearby low-lying
sandy cay islands in the Mamanuca Group did not begin to form until
approximately 2260–2110 BP at the earliest, after the retreat of the
mid-Holocene sea-level high stand (Figure 6). While this date is
800–600 years later than the date for commencement of progradation
on Tavua, for low-lying lying reef islands the process of coastal
regression would have likely required a more significant drop in sea
level than for basalt islands like Tavua. When viewed in concert,
these data indicate that the island landscapes of the Mamanuca Group
were altered by the decline in mid-Holocene sea level in ways that
would have made them attractive to human settlement.
Based on the chronologic and stratigraphic relationships
between the progradation of the coastal plain and evidence of ini-
tial human occupation, settlement commenced very shortly after
the beach in the vicinity of the excavation units began to emerge.
For example, inspection of Figure 6 shows that cultural dates for
the Lapita deposit on Tavua overlap slightly with the earliest evi-
dence of coastal regression. However, the stratigraphic relation-
ship between the sedimentary and archaeological samples confirms
that the earliest date for settlement of the island occurred close on
the heels of the formation of the coastal plain. Moreover, dating of
a non-abraded coral clast recovered approximately 38 m seaward
of the primary excavations indicates that from 2689 to 2423 cal.
BP, the coastal plain continued to prograde. Presumably, coastal
progradation increased over the course of the next millennium
until sea level stabilized; however, the nature of geomorphological
evolution seaward of Core 26 (see Figure 3) is not well understood
partially as a result of significant historic disturbances associated
with the construction of modern Tavua village.
Comparative research in Fiji at the Natunuku Lapita site on Viti
Levu by Nunn (2005) suggests that when the initial Lapita occupants
arrived on the island approximately 2800 cal. BP, sea level had
Figure 6. Sea-level curve (adapted from McKoy et al., 2010). The black dotted line represents the modeled ICE-4G sea-level curve from Nunn
and Peltier (2001) for the local region for 17°450′S, 178°45′E. The gray envelope represents Nunn’s (1990) sea-level curve for the Fijian Islands
adapted from McKoy et al. (2010). Black squares are dated subsurface sand sample from nearby cay islands in the Mamanuca (data from McKoy
et al., 2010), red squares are dated subsurface cultural samples from Tavua, and white squares are dated coral fragments from Tavua.
10 The Holocene 00(0)
retreated nearly 0.85 m from the mid-Holocene maximum. One con-
sequence of this sea-level decline was an increase in habitable coastal
lowlands. By 2400 cal. BP, the coastal lowlands had expanded sig-
nificantly which Nunn suggests, ‘encouraged a greater use of horti-
culture by the areas inhabitants’ (p. 19). The results from Natunuku
mirror closely both the temporal and spatial sequence of settlement
and coastal progradation documented along the Tavua beach coastal
plain. Additionally, microfossil analysis from the Tavua excavations
indicates an increase in vegetation disturbance and a greater amount
of cultivated plant use at the site (Appendix 1).
The Tavua Village cultural and geomorphological results are
also consistent with a number of recent Pacific Island research
projects that point to the importance of beach ridges as locations
of initial settlement. For example, Dickinson (2014) notes,
Many of the earliest known archaeological sites on multiple
islands were located atop beach ridges where habitations
could be established close to offshore reef resources and
handy for water travel by boat, and were commonly first
occupied close on the heels of the construction of the beach
ridge. (p. 264)
Similarly, Allen et al. (2016) generated a locally derived sea-
level curve from uranium–thorium (U/Th) dating of fossil micro-
atolls, in combination with dated archaeological deposits and
elevation information. They conclude that many locations on
Aitutaki’s (Southern Cook Islands) mainland and offshore islets
would have been unsuitable for colonization until at least AD
450–750 when regional sea level declined past the crossover
date, a time when high-tide levels fell below mid-Holocene low-
tide levels (Dickinson, 2003). The archaeological evidence sup-
ports this conclusion, as the early unambiguous evidence of
settlement on Aitutaki can be found on a newly formed beach
ridge located along the west coast of the mainland, only after
regional sea level had declined.
Archaeological and geological investigations on the south
coast of ‘Upolu in Samoa by Cochrane et al. (2016; see also Kane
et al., in press) suggest that much of the Satitoa coastal plain had
not formed until after ~1000 cal. BP. Like Tavua and Aitutaki,
higher than present sea level, combined with tectonic subsidence
in the Satitoa example, limited island habitability and it was not
until sea level began to decline that beach ridges and coastal ter-
races expanded enough to support human settlement. Although
the timing and magnitude have varied across the equatorial
Pacific basin (Grossman et al., 1998), sea-level fall and its con-
comitant influence on local coastal geomorphology have clearly
played an important role in the colonization chronology of oce-
anic islands (Dickinson, 2003).
Conclusion
People living in small island environments are particularly vul-
nerable to sea-level change because of the proximity of settle-
ments to the coast, and in many cases the low elevation and
limited land mass of island terrestrial environments. For example,
the majority of the coralline atolls of Tuvalu, in the south Pacific,
sit between 1 and 2 m above MSL and the islands are forecasted
to be completely submerged by the end of the 21st century.
Undoubtedly, future sea-level change associated with climate
variability will have a substantial impact on the life-ways of peo-
ple living on these atolls. Some of the most serious risks associ-
ated with elevated sea level include coastal erosion, increased
flooding, and salt water intrusion into critical freshwater resources
(Rahmstorf, 2012). The first colonists of Fiji and other Oceanic
islands would have faced similar ecological contexts when they
arrived at their new island homes during times of higher sea level.
Reconstruction of the geomorphology, chronology, and sedi-
mentology of the Tavua coastal plain indicates a sequence of
coastal progradation and beach ridge formation, beginning around
3000 cal. BP, and progradation associated with continued sea-
level drawdown. While this sequence of sea-level fall, beach
ridge formation, and coastal progradation on the heels of initial
settlement has been demonstrated for a number of Remote Oce-
anic islands, few similar studies relative to the 130 or more
Remote Oceanic Lapita deposits have been carried out. When
analyzed in combination, the archaeological materials, sedimen-
tological data, and chronology make it possible that Tavua was
settled very soon after it could be settled.
The Tavua geological and archaeological sequence also pro-
vides important information regarding broader substantive topics
in Fijian prehistory and the Lapita cultural horizon generally. The
Lapita migration occurred rapidly across a geographically expan-
sive region; however, within several centuries, locally divergent
ceramic styles developed in Fiji, nearby Vanuatu, and New Cale-
donia, and Lapita decorative styles ceased to be produced. This
suggests that interaction between populations living across dispa-
rate island groups became less frequent (Cochrane and Lipo,
2010). We propose that sea-level fall and the associated formation
of coastal plains, especially on small islands, played an important
role in this decreased interaction and the subsequent development
of local cultural diversity. The combination of a newly stable
coastal landform would have reduced the risk associated with per-
manent settlement of this previously dynamic island, negating the
need for continued contact with other social groups. Testing the
applicability of this scenario requires more research on landform
evolution, paleoenvironments, and the chronology of Lapita
deposits throughout Remote Oceania.
Acknowledgements
The authors thank Dave Burley, Julie Field, Chip Fletcher, Haunai
Kane, Patrick Nunn, and Peter Sheppard for providing insight-
ful comments on earlier drafts of this manuscript. This research
was also greatly improved through numerous conversations with
Melinda Allen, Mike Evans, Drew Lorrey, Helen McGregor, and
Steve Athens. The authors also thank the residents of Tavua Vil-
lage for graciously hosting them during their stay on the island.
Funding
Financial support for this research was provided by the British
Academy (grant SG-52445) and The University of Auckland
Arts Faculty Performance Based Research Fund.
References
Addison DJ and Matisoo-Smith E (2010) Rethinking Polynesians
origins: A West-Polynesia triple-I model. Archaeology in
Oceania 45: 1–12.
Allen MS, Morrison AE, Lorrey AM et al. (2016) Timing, mag-
nitude and effects of late Holocene sea level drawdown on
island habitability, Aitutaki, Cook Islands. Archaeology in
Oceania 51: 108–121.
Anderson A, Bedford S, Clark G et al. (2001) An inventory of
Lapita sites containing dentate-stamped pottery. In: Clark
GA, Anderson AJ and Vunidilo TS (eds) The Archaeology of
Lapita Dispersal in Oceania. Canberra: Pandanus Books, pp.
1–14.
Bedford S and Sand C (2007) Lapita and Western Pacific set-
tlement: Progress, prospects and persistent problems. In:
Bedford S, Sand C and Connaughton SP (eds) Oceanic Explo-
rations: Lapita and Western Pacific Settlement. Canberra:
ANU Press, pp. 1–16.
Bronk Ramsey C (2009) Bayesian analysis of radiocarbon dates.
Radiocarbon 51(1): 337–360.
Morrison et al. 11
Bronk Ramsey C (2013) Oxcal 4.2. Available at: http://c14.arch.
ox.ac.uk/oxcal.
Burley DV (2012) Exploration as a strategic process in the Lapita
settlement of Fiji: The implications of Vorovoro Island. Jour-
nal of Pacific Archaeology 3(1): 22–34.
Burley DV and Connaughton J (2010) Completing the story: A
late Lapita dentate stamped pot from Sigatoka, Fiji. Archaeol-
ogy in Oceania 45: 130–132.
Cochrane EE (in press) Ancient Fiji: Melting pot of the South-
west Pacific. In: Cochrane EE and Hunt TL (eds) The Oxford
Handbook of Prehistoric Oceania. Oxford: Oxford Univer-
sity Press.
Cochrane EE and Lipo CP (2010) Phylogenetic analyses of Lap-
ita decoration do not support branching evolution or regional
population structure during colonization of Remote Oceania.
Philosophical Transactions of the Royal Society of London
Series B: Biological Sciences 365(1559): 3889–3902.
Cochrane EE, Matararaba S and Nakoro E (2007) Lapita and
later archaeology of the Malolo and Mamanuca Islands,
Fiji. The Journal of Island and Coastal Archaeology 2(2):
245–250.
Cochrane EE, Rivera-Collazo IC and Walsh E (2011) New
evidence for variation in colonization, cultural transmis-
sion, and subsistence from Lapita (2900 BP) to the historic
period in Southwestern Fiji. Journal of Pacific Archaeology
2(1): 40–55.
Cochrane EE, Kane H, Fletcher C et al. (2016) Lack of suitable
coastal plains likely influenced Lapita (~2800 cal. BP) settle-
ment of Sāmoa: Evidence from south-eastern ‘Upolu’. The
Holocene 26(1): 126–135.
Dickinson WR (2001) Paleoshoreline record of relative Holocene
sea levels on Pacific islands. Earth-Science Reviews 55(3):
191–234.
Dickinson WR (2003) Impact of mid-Holocene hydro-isostatic
highstand in regional sea level on habitability of islands
in Pacific Oceania. Journal of Coastal Research 19:
489–502.
Dickinson WR (2014) Beach ridges as favored locales for human
settlement on Pacific Islands. Geoarchaeology 29(3): 249–
267.
Fletcher CH and Jones AT (1996) Sea-level highstand recorded
in Holocene shoreline deposits on Oahu, Hawaii. Journal of
Sedimentary Research 66(3): 632–641.
Goodwin ID and Harvey N (2008) Subtropical sea-level history
from coral microatolls in the Southern Cook Islands, since
300 AD. Marine Geology 253(1): 14–25.
Green RC (2003) The Lapita horizon and traditions – Signature for
one set of oceanic migrations. In: Sand C (ed.) Pacific Archae-
ology: Assessments and Prospects (vol. 15, Les Cahiers de
l’Archaéologie en Nouvelle Calédonie). Nouméa: Service des
Musées et du Patrimoine de Nouvelle-Calédonie, pp. 95–120.
Grossman EE, Fletcher CH III and Richmond BM (1998) The
Holocene sea-level highstand in the equatorial Pacific: Analy-
sis of the insular paleosea-level database. Coral Reefs 17(3):
309–327.
Horrocks M (2005) A combined procedure for recovering phy-
toliths and starch residues from soils, sedimentary deposits
and similar materials. Journal of Archaeological Science 32:
1169–1175.
Hunt TL (1988) Graph theoretic network models for Lapita
exchange: A trial application. In: Kirch PV and Hunt TL
(eds) Archaeology of the Lapita Cultural Complex: A Critical
Review, vol. 5. Seattle, WA: Burke Museum, pp. 135–156.
Kane HH, Fletcher CH III, Cochrane EE et al. (2017) Coastal
plain stratigraphy records tectonic, environmental, and
human habitability changes related to sea-level drawdown,
‘Upolu, Samoa’. Quaternary Research 87(2): 246–257.
Kennett DJ, Anderson A and Winterhalder B (2006) The ideal
free distribution, food production, and the colonization of
Oceania. In: Kennett DJ and Winterhalder B (eds) Behavioral
Ecology and the Transition to Agriculture. Berkeley, CA:
University of California Press, pp. 265–288.
Kirch PV (1997) The Lapita Peoples: Ancestors of the Oceanic
World. Oxford: Blackwell.
Lal KK and Nunn PD (2011) Holocene sea levels and coastal
change, South-west Viti Levu Island, Fiji. Australian Geog-
rapher 42(1): 41–51.
Lepofsky D (1988) The environmental context of Lapita settle-
ment locations. In: Kirch PV and Hunt TL (eds) Archaeology
of the Lapita Cultural Complex: A Critical Review. Seattle,
WA: Burke Museum, pp. 33–48.
Loy TH, Spriggs M and Wickler S (1992) Direct evidence for
human use of plants 28,000 years ago: Starch residues on
stone artefacts from the northern Solomon Islands. Antiquity
66: 898–912.
McKoy H, Kennedy DM and Kench PS (2010) Sand cay evolu-
tion on reef platforms, Mamanuca Islands, Fiji. Marine Geol-
ogy 269(1): 61–73.
Mitrovica JX and Peltier WR (1991) On postglacial geoid sub-
sidence over the equatorial oceans. Journal of Geophysical
Research: Solid Earth 96(B12): 20053–20071.
Miyata T, Maeda Y, Matsumoto E et al. (1990) Evidence for a
Holocene high sea-level stand, Vanua Levu, Fiji. Quaternary
Research 33(3): 352–359.
Moore PD, Webb JA and Collinson ME (1991) Pollen Analysis.
2nd Edition. London: Blackwell Scientific.
Morrison AE and Cochrane EE (2008) Investigating shellfish
deposition and landscape history at the Natia Beach site, Fiji.
Journal of Archaeological Science 35(8): 2387–2399.
Nunn PD (1990) Coastal processes and landforms of Fiji: Their
bearing on Holocene sea-level changes in the south and west
Pacific. Journal of Coastal Research 6: 279–310.
Nunn PD (1998) Pacific Island Landscapes: Landscape and Geo-
logical Development of Southwest Pacific Islands, Especially
Fiji, Samoa and Tonga. Suva: Institute of Pacific Studies, The
University of the South Pacific.
Nunn PD (2005) Reconstructing tropical paleoshorelines using
archaeological data: Examples from the Fiji Archipelago,
southwest Pacific. Journal of Coastal Research 42(Special
Issue): 15–25.
Nunn PD (2009) Geographical influences on settlement-location
choices by initial colonizers: A case study of the Fiji Islands.
Geographical Research 47(3): 306–319.
Nunn PD and Heorake TA (2009) Understanding the place prop-
erly: palaeogeography of selected Lapita sites in the western
tropical Pacific Islands and its implications. In: Sheppard PJ,
Thomas T and Summerhayes GR (eds) Lapita: Ancestors
and Descendants. Monograph 28. Auckland: New Zealand
Archaeological Association, pp. 235–254.
Nunn PD and Peltier WR (2001) Far-field test of the ICE-4G
model of global isostatic response to deglaciation using
empirical and theoretical Holocene sea-level reconstruc-
tions for the Fiji Islands, southwestern Pacific. Quaternary
Research 55(2): 203–214.
Nunn PD and Petchey F (2013) Bayesian re-evaluation of Lapita
settlement in Fiji: Radiocarbon analysis of the Lapita occupa-
tion at Bourewa and nearby sites on the Rove Peninsula, Viti
Levu Island. Journal of Pacific Archaeology 4: 21–34.
Nunn PD, Ollier C, Hope G et al. (2002) Late quaternary sea-
level and tectonic changes in northeast Fiji. Marine Geology
187(3): 299–311.
Peltier WR (1995) VLBI baseline variations from the Ice-4G
model of postglacial rebound. Geophysical Research Letters
22(4): 465–468.
12 The Holocene 00(0)
Petchey F, Anderson A, Hogg A et al. (2008) The marine reser-
voir effect in the Southern Ocean: An evaluation of extant and
new? R values and their application to archaeological chro-
nologies. Journal of the Royal Society of New Zealand 38(4):
243–262.
Rahmstorf S (2012) Modeling sea level rise. Nature Education
Knowledge 3(10): 4.
Reimer PJ, Bard E, Bayliss A et al. (2013) IntCal13 and Marine13
radiocarbon age calibration curves 0–50,000 years cal BP.
Radiocarbon 55(4): 1869–1887.
Sheppard PJ, Chiu S and Walter R (2015) Re-dating Lapita move-
ment into remote Oceania. Journal of Pacific Archaeology
6(1): 26–36.
Skoglund P, Posth C, Sirak K et al. (2016) Genomic insights into
the peopling of the Southwest Pacific. Nature 538: 510–513.
Smithers SG and Woodroffe CD (2000) Microatolls as sea-level
indicators on a mid-ocean atoll. Marine Geology 168(1):
61–78.
Soil Survey Staff (2014) Soil Survey Field and Laboratory Meth-
ods Manual. Soil Survey; Kellogg Soil Survey Laboratory;
National Soil Survey Center. Available at: https://www.nrcs.
usda.gov/Internet/FSE_DOCUMENTS/stelprdb1244466.pdf.
Stein JK (2001) Archaeological sediments in cultural environ-
ments. In: Stein JK and Farrand WR (eds) Sediments in
Archaeological Context. Salt Lake City, UT: University of
Utah Press, pp. 1–28.
Sunell L and Healey P (1979) Distribution of calcium oxalate
crystal idioblasts in corms of taro (Colocasia esculenta).
American Journal of Botany 66(9): 1029–1032.
Tables AT (2009) Admiralty Tide Tables. Taunton: UKHO.
Thompson S (1982) Cyrtosperma chamissonis (Araceae): Ecol-
ogy, distribution, and economic importance in the South
Pacific. Journal d’Agriculture Tropicale et de Botanique
Appliquee 24: 185–203.
Whistler WA (2009) Plants of the Canoe People: An Ethnobotani-
cal Voyage through Polynesia. Kalaheo, HI: National Tropi-
cal Botanical Garden.
Woodroffe CD, McGregor HV, Lambeck K et al. (2012) Mid-
Pacific microatolls record sea-level stability over the past
5000 yr. Geology 40(10): 951–954.
Appendix 1
Plant microfossils from Tavua
Two samples (ID 1376 and 1377) recovered from the controlled
excavations were analyzed for plant microfossils to provide a his-
tory of vegetation, environments, and human activity. The sam-
ples were analyzed for pollen, phytoliths, starch, and other plant
material.
The samples were prepared for pollen analysis by the stan-
dard acetolysis method (Moore et al., 1991). Both samples con-
tained insufficient pollen for meaningful counting. Slides were
scanned, however, and occasional pollen types were noted.
Microscopic fragments of charcoal are extracted along with pol-
len during preparation, providing evidence of fire. The samples
were prepared for phytolith analysis by density separation (Hor-
rocks, 2005). At least 150 phytoliths were counted for each
sample, and slides were scanned for types not found during the
counts. Starch and other plant material were prepared for
analysis by density separation, and presence/absence was noted
(Horrocks, 2005).
Two sediment samples were examined for plant microfossils to
generate preliminary data on economic plants, vegetation changes
possibly associated with human disturbance of the landscape, and
natural geomorphological change. Sample 1376 was recovered
from Layer IV (TU 2, 96–103 cmbs), which is estimated to have
begun forming between 2863 and 2607 cal BP (95.2%) with depo-
sition ceasing sometime 2751–2026 cal BP (95.4%). This micro-
fossil sample dates to the end of the Lapita occupation. Sample
1377 was recovered from Layer II (TU 2, 47–50 cmbs), above the
Lapita deposits. There are no clearly associated and informative
radiocarbon samples to provide a date range for this sample, but
associated ceramics (Cochrane et al., 2011) indicate that it most
likely dates to the end of the Fijian plainware phase or early Navatu
phase, around 1800–1600 BP (Cochrane, in press).
Abundant microscopic fragments of charcoal were found in
both plant microfossil samples, reflecting human activity at the
site, namely, hearth fires and burning of vegetation. Pollen
and spores were extremely sparse, insufficient for meaningful
counting. The few found comprised Pandanus tectorius in both
samples, cf. coconut (Cocos nucifera) in sample 1377 and two
types of undifferentiated fern spore in sample 1376. Coconut and
Pandanus are thought to be indigenous to Fiji, although cultivars
of both species were introduced to Polynesia (Whistler, 2009).
Many of the phytoliths in the samples could not be identified
because of corrosion and fragmentation. These phytoliths were
assigned to the ‘degraded’ category.
The abundant spherical verrucose phytoliths in sample
1376, from Layer IV, are from dicotyledons, possibly trees
(Figure 7). If this is the case, the relatively very small amount
of this phytolith type coincident with the large amount of grass
phytoliths in sample 1377, Layer II, suggests an increase in
vegetation disturbance over the time period represented by the
samples. The banana leaf phytolith in sample 1376 indicates
use of this introduced crop at the site.
Two types of starch, and some other plant material, were iden-
tified in both samples. The first type comprised individual
medium-sized grains, sub-spherical to ovate, with flattened pres-
sure facets. Based on this morphology, there are three species of
introduced aroid (Araceae) starch crops that are potential con-
tributors, namely, Alocasia macrorrhiza, Amorphophallus paeo-
niifolius, and Cyrtosperma merkusii (Loy et al., 1992). Starch
grains of the corms of these species can be difficult to differenti-
ate. Given the beach site location, however, C. merkusii is the
most likely source since it is the most tolerant of brackish water
environments (Thompson, 1982). The other introduced Oceanic
aroid, taro (Colocasia esculenta), can be ruled out as its starch
grains are distinctly different from those identified in this study.
Other plant material identified in the samples included degraded,
fragmented calcium oxalate crystals (raphides and druses). This
presence supports the starch grain evidence of aroids, as members
of this family have very high concentrations of calcium oxalate
crystals in all tissues relative to most other plants (Sunell and
Healey, 1979).
The other type of starch, comprising, relatively large, ovate
individual grains, is consistent with the tuber of introduced spiny
yam (Dioscorea nummularia). In addition, xylem cells consistent
with this genus were identified in both samples.
