bs_bs_banner
Archaeology in Oceania, Vol. •• (2015): ••–••
DOI: 10.1002/arco.5050
Human ecodynamics in the Mangareva Islands: a stratified sequence from
Nenega-Iti Rock Shelter (site AGA-3, Agakauitai Island)
PATRICK V. KIRCH, GUILLAUME MOLLE, CORDELIA NICKELSEN, PETER MILLS,
EMILIE DOTTE-SAROUT, JILLIAN SWIFT, ALLISON WOLFE and MARK HORROCKS
PVK, CN, JS, AW: University of California, Berkeley; GM: Université de la Polynésie Française; PM: University of
Hawai’i; ED-S: University of Western Australia; MH: Microfossil Research Ltd; MH: University of Auckland
ABSTRACT
The Gambier Islands (French Polynesia) are noted for their extreme deforestation and low biodiversity in the post-European contact
period. We report on the archaeological and palaeoecological investigation of a stratified rock shelter (site AGA-3) on Agakauitai
Island, revealing a sequence of environmental transformation following Polynesian colonisation of the archipelago. Radiocarbon dates
indicate use of the rock shelter from the 13th to the mid-17th centuries, followed by a sterile depositional hiatus, and then final early
post-contact use (late 18th to early 19th century). Zooarchaeological analysis of faunal remains indicates rapid declines in local
populations of seabirds, especially procellariids, as well as later increases in numbers of the introduced, commensal Pacific rat (Rattus
exulans). Macro- and micro-botanical evidence documents transformation of the island’s flora from indigenous forest to one dominated
by economic plants and fire-resistant taxa. A multi-causal model of dynamic interactions, including nutrient depletion due to seabird
loss, most likely accounts for this dramatic ecological transformation.
Keywords:
Gambier Islands, Polynesian colonisation, seabird extinctions, fishhooks, adzes, Pacific rat.
RÉSUMÉ
Les îles de l’archipel des Gambier, en Polynésie française, sont notamment connues pour leur déforestation extrême et leur faible
biodiversité dont témoignaient déjà les écrits des premiers occidentaux. Nous présentons ici les résultats d’un projet de recherche
portant sur l’étude archéologique et paléoécologique d’un abri-sous-roche stratifié (site AGA-3) sur l’île d’Agakauitai. Nos travaux
démontrent une séquence de transformations environnementales qui débutèrent immédiatement après la colonisation Polynésienne de
l’archipel. Les datations radiocarbone indiquent que cet abri fut utilisé continuellement entre le 13ème et le milieu du 17ème siècle AD.
Suite à une interruption de l’occupation marquée par un dépôt stérile, l’abri est finalement réutilisé au cours de la période
post-européenne (fin 18ème – début 19ème siècle). L’analyse zooarchéologique des restes fauniques démontre un rapide déclin des
populations locales d’oiseaux marins, en particulier les procellariidés, ainsi qu’une augmentation plus tardive du nombre de restes de
rat Pacifique (Rattus exulans), une espèce commensale introduite par les Polynésiens. L’étude des restes micro- et macrobotaniques
documente également la transformation de la flore de l’île en démontrant une évolution du couvert végétal passant d’une forêt indigène
à un environnement dominé par les plantes à valeur économique ainsi que des taxons résistants au feu. Le modèle d’interactions
dynamiques à plusieurs facteurs ici proposé, intégrant notamment la perte de nutriments consécutive à la disparition des oiseaux
marins, constitue à présent l’explication la plus plausible à cette dramatique transformation écologique.
Mots-clés:
Iles Gambier, colonisation polynésienne, extinction de l’avifaune, hameçons, herminettes, rat pacifique.
Correspondence: Patrick V. Kirch, Department of Anthropology, University of California, 232 Kroeber Hall, Berkeley,
CA 94720, USA. Email: kirch@berkeley.edu
The Mangareva, or Gambier, Islands lie at the
south-eastern extreme of French Polynesia (23°07’S,
134°58’W), with the Acteon Group of the Tuamotu
Archipelago 180 km north-west, and the
Pitcairn–Henderson islands 540 km south-east. Mangareva
was investigated by pioneering Bishop Museum
archaeologist Kenneth P. Emory in 1934 (Emory 1939),
© 2015 Oceania Publications
and by Roger Green in 1959 (Green & Weisler 2000,
2002, 2004; Suggs 1961a), yet remains one of the least
known Polynesian archipelagoes. Renewed archaeological
investigations began in Mangareva in 2001–2005
(Anderson et al. 2003; Conte & Kirch 2004; Kirch &
Conte 2008; Kirch et al. 2010; Orliac 2003). Here we
report on 2012 excavations at the Nenega-Iti Rock Shelter
2
(site 190-12-AGA-3) on Agakauitai Island, documenting a
stratified sequence reflecting significant anthropogenic
landscape and biotic changes.
THE MANGAREVA ECOSYSTEM: BACKGROUND
AND HYPOTHESES
European explorers and naturalists alike have stressed the
deforested and biotically depauperate nature of the
Mangareva terrestrial environment. Captain Wilson (1799:
118) of the Duff observed in 1797: “The tops of the hills,
to about half way down, are chiefly covered with sun-burnt
grass; and in some places there are spots of reddish soil, as
on the middle grounds of Otaheite.” Harold St John,
botanist of the 1934 Bishop Museum Mangarevan
Expedition, wrote: “Mangareva Islands are desolated; their
natural flora is more completely exterminated than that of
any other part of the world that I have seen” (1935: 57).
The 1934 Expedition’s leader, malacologist C. Montague
Cooke Jr, observed that “all the endemic forests have
disappeared . . . except on the precipitous southern slope
of Mount Mokoto, where some of our party found a small
remnant of native forest near the base of the cliff. A few
scattered native shrubs and small trees were growing on
the ledges above” (1935: 41).
Butaud (2013; see also Huguenin 1974) enumerates 584
species of ferns and higher plants in the Gambier Islands,
including 92 indigenous taxa and nine endemics. Of the
492 introduced plants, 60 are considered to have been
Polynesian introductions. Key aspects of the Mangarevan
vegetation are: (1) the absence of native forests on the
volcanic hills, dominated by pyrophytic fernlands of
Dicranopteris (Gleichenia) linearis on ferralitic soils, and
canelands of Miscanthus floridulus on vertisols; and, (2)
the strongly anthropogenic character of the coastal and
lowland vegetation, dominated by economically useful
Polynesian introductions (e.g. Cocos nucifera, Artocarpus
altilis and Cordyline fruticosa). We hypothesise that these
pyrophytic fern- and canelands developed in response to
anthropogenic burning and forest clearance on slopes with
old, nutrient-poor soils. One objective of our excavations
at AGA-3 was to test this hypothesis through
palaeobotanical analysis of macroscopic charcoal and
microfloral plant remains (pollen, opal phytoliths) from
stratified contexts.
The terrestrial fauna of Mangareva is as impoverished
as the flora. Cochereau’s (1974) faunal inventory is
dominated by invertebrates, particularly insects, among
which there are only a few endemic species (e.g.
Zimmerman 1936). Terrestrial molluscs are represented by
just six taxa, three of which are widely dispersed
pulmonates transported inadvertently by the Polynesians
(Tornatellinops variabilis, Elasmias apertum and
Lamellidea oblonga; see Kirch 1984: 137). Yet subfossil
deposits identified during the 1934 Mangarevan Expedition
(Cooke 1935) have yielded several endemic genera and
species of endodontid landsnails (Solem 1976), all of
which are now extinct.
Human ecodynamics in Mangareva
The avifauna of Mangareva was surveyed by Lacan and
Mougin (1974) and more recently by Thibault and Cibois
(2012) and by Waugh et al. (2013), the list being
dominated by 15 species of seabirds. There is a native
kingfisher (Halcyon gambieri) of a species found also in
the Tuamotus; the only other land birds are a reed-warbler
(listed as Conopoderas caffra) and the common rock dove
(Columba livia). However, Thibault and Cibois (2012,
table 1) note that a number of other land bird species were
present in the 19th and early 20th centuries. Lacan and
Mougin (1974: 537) stress the uneven geographical
distribution of seabirds among the volcanic islets and coral
motu, especially the nesting and reproducing populations,
which are heavily concentrated on three small,
difficult-to-access and uninhabited islets in the southern
part of the lagoon (Makaroa, Manui and Motu Teiku).
Hiroa (1938: 9) lists Mangarevan bird names including a
kuku (pigeon) and moho (probably a rail), both said to
have been extinct by 1934.
In sum, observations over the past two centuries point to
an ecosystem impacted by deforestation and terrestrial
biotic impoverishment. Research on other Eastern
Polynesian islands, including Mangaia (Kirch & Ellison
1994; Kirch et al. 1995) and Rapa Nui (Flenley et al. 1991;
Hunt 2007; Steadman et al. 1994) has revealed that human
colonisation and land use on remote and vulnerable island
ecosystems often led to irreversible changes in island
biodiversity. A hypothesis of anthropogenically driven
changes to Mangareva’s vegetation and biota, however,
must be empirically tested; this was a principal objective of
renewed archaeological investigations initiated in 2001. In
particular, we hypothesised that a reduction in land and
seabird populations resulted from the direct or indirect
impacts of human colonisation, including predation, habitat
clearance and modification, and the introduction of the
Pacific rat (Rattus exulans). The potential role of rats in
modifying island biotas has been the subject of much recent
discussion (e.g. Athens et al. 2002; Brooke et al. 2010;
Hunt 2007; Mieth & Bork 2010). In addition, significant
reductions in seabird populations are likely to have affected
nutrient cycling within the Mangareva ecosystem, in turn
limiting the regrowth of native vegetation after Polynesian
land clearance.
AGAKAUITAI ISLAND AND
THE NENEGA-ITI SITE
Agakauitai is one of the smaller of ten volcanic islets
within the Gambier Islands “near-atoll” surrounded by a
single barrier reef and lagoon ecosystem (Figure 1). With
a land area of 0.7 km2, the island’s slopes rise steeply to
a summit at 139 m. The AGA-3 Rock Shelter is situated
near the island’s north-western end, about 100 m from
the present shoreline, at the back of a small valley called
Nenega-Iti. This part of the island faces Taravai Island,
across a channel with a sandy beach terrace that would
have been suitable for habitation, and for canoe landings
and launchings. In contrast, much of the rest of
© 2015 Oceania Publications
Archaeology in Oceania
Agakauitai is ringed with steep cliffs (Figure 2a).
Shifting tidal flows through the channel between
Agakauitai and Taravai make this an excellent location
for net fishing.
The rock shelter is formed by a volcanic dyke in a cliff
face about 20 m high; the softer breccia surrounding the
dyke has eroded away, leaving the dyke as the rear wall of
the rock shelter (Figure 2b). The shelter has a length of
about 16 m, and a depth of between 2 and 2.5 m from the
dripline to the rear wall, with a protected floor area of
about 35 m2. Vegetation in the shelter’s vicinity is
thoroughly anthropogenic. Upslope of the overhanging
cliff is Miscanthus grassland, while the valley floor
fronting the site is dominated by young mango (Mangifera
indica) and coconut, along with Hibiscus tiliaceous and
Java plum (Syzigium cumini).
FIELD AND LABORATORY METHODS
Excavations at AGA-3 followed methods used in
previous fieldwork in the Gambiers. A metric grid was
established over the shelter floor (Figure 3), while
vertical control was referenced to a fixed datum
(a chiselled “X” on a large boulder), with depths taken
using a Nikon level and stadia rod. Excavation followed
natural stratigraphy, with strata subdivided into levels
(usually 5 cm thick) for finer control. Sediment was dry
screened through 3 mm mesh, but in some levels also
wet-sieved through 2 mm mesh to improve the recovery of
small bones. Bone, shell, charcoal and lithics were bagged
by level for laboratory analysis; artefacts found in situ
were point plotted.
Vertebrate and invertebrate faunal materials were
analysed at the Oceanic Archaeology Laboratory (OAL) in
Berkeley, using reference collections of the OAL and of
the UC Berkeley Museum of Vertebrate Zoology (MVZ).
Fish bone was identified with the OAL collection aided by
the Manual of Hawaiian Fish Remains (Dye &
Longenecker 2004), according to methods described by
Leach (1997) and Dye and Longenecker (2004), using five
distinct cranial elements (premaxilla, maxilla, dentary,
angular and quadrate) and additional special bones
(pharyngeal plates, tangs, vomer, dorsal spines for certain
taxa). Bird bones were identified by comparison with
archaeological specimens previously identified from the
Onemea (TAR-6) and AGA-3 sites (Worthy & Tennyson
2004) and with reference specimens in the MVZ. Marine
molluscs were identified with reference to Salvat and
Rives (1991).
A continuous sediment column with 10-cm sampling
intervals was taken from the face of unit F9; samples did
not cross natural stratigraphic boundaries. After
sterilisation at 155°C in a laboratory oven, the sediment
samples were weighed, and subsamples removed for
microfossil analysis. The Munsell colour was recorded
(dry), while pH was determined using an Oakton Acorn
Series pH 5 meter. The organic content was measured by
loss on ignition (Dean 1974), by heating the samples in a
© 2015 Oceania Publications
3
Thermolyne 30400 muffle furnace at 560°C for 1 h. Grain
size was determined by wet sieving through nested
geological screens with mesh sizes from −4Φ to 3Φ
(Wentworth scale); dried screen contents were weighed
with an Ohaus Sc4020 digital scale. Microscope slides
were prepared from the 0 Φ fraction (1 mm, very coarse
sand) and examined under a stereo microscope for
lithology and composition.
Four subsamples from the sediment column were
analysed for pollen and phytoliths. The samples were
prepared for pollen analysis by the standard acetylation
method (Moore et al. 1991). At least 150 pollen grains and
spores were counted and slides were scanned for types not
found during the counts. Samples were prepared for
phytolith analysis by density separation. At least 150
phytoliths were counted per sample and the slides were
scanned for types not found during the counts (Horrocks
2005).
To assess temporal change in reconstructed Pacific rat
diet, 37 Rattus exulans bone samples were selected for
δ13C and δ15N analysis. Sample selection was conducted
by calculating minimum number of individuals (MNI) for
each level to minimise resampling from the same
individuals (Table S1). Specimens were sonicated with
ultrafiltered water for 4 h, dried, abraded to remove surface
contaminants and cut into ∼1 mm chunks. Bone collagen
was isolated following a procedure modified from
Ambrose (1990; and see Sealy et al. 2014); samples over
20 mg were treated with 1 ml 0.5 M HCl and samples
under 20 mg with 1 ml 0.25 M HCl for 48 h, with fresh
HCl applied after 24 h. All samples were then treated with
1 ml 0.1 M NaOH for 24 h to remove humic contaminants,
then freeze-dried for 48 h. Samples were analysed for C
and N isotope ratios at the Center for Stable Isotope
Biogeochemistry Laboratory, UC Berkeley using a
CHNOS elemental analyser and an IsoPrime 100 mass
spectrometer. All samples possess C:N ratios between 3.2
and 3.5 and wt%C, wt%N and wt% collagen within
appropriate ranges for samples from tropical environments
(Pestle & Colvard 2012).
Macroscopic charcoal was retained from all excavation
units during dry screening (down to 3 mm). Charcoal
samples from three strata within unit F9 were sorted and
identified using Dotte-Sarout’s (2010) Pacific Wood
Anatomy Database and reference collection. Anatomical
features were described after Wheeler et al. (1989).
Sampling and interpretations are based on anthracology
principles (Théry-Parisot et al. 2010). The sample from
non-cultural Layer I is of reduced size (48 identifiable
fragments), while the assemblages from cultural
Layers II and IV each comprise 100 identifiable
fragments.
Basalt adzes from AGA-3 were analysed at the
Geoarchaeology Laboratory of the University of Hawai’i,
Hilo, using a ThermoNoran QuanX energy-dispersive
X-ray fluorescence (EDXRF) spectrometer, following the
non-destructive methodology of Lundblad et al. (2008,
2010). Results were compared with EDXRF analyses of
4
Human ecodynamics in Mangareva
Figure 1. An aerial photograph of Agakauitai Island, showing the location of site AGA-3 and the trench in the colluvial
fan at AGAK-2. The inset shows the location of Agakauitai Island within the Gambier group.
Figure 2. (a) A view of Agakauitai Island from the west; (b) the AGA-3 Rock Shelter as seen from the north during
excavation; (c) the block excavation in AGA-3; (d) a view of the main trench in AGA-3 after completion of excavations.
(Photographs by P. Kirch).
© 2015 Oceania Publications
Archaeology in Oceania
Figure 3. A plan of the AGA-3 site, showing the
locations of the excavated units. Contour interval 50 cm.
Eiao basalt adze quarry source material in the Marquesas
Islands (Charleux et al. 2014), debitage from Mo’orea in
the Society Islands (Kahn et al. 2013) and published
values for the Tautama basalt quarry, Pitcairn (Sinton &
Sinoto 1997: 201). Two geological specimens from Mt
Duff on Mangareva were also analysed, as well as a USGS
geological standard from Kı̄lauea caldera, Hawai’i Island
(BHVO-2).
EXCAVATIONS AND STRATIGRAPHY
The AGA-3 Rock Shelter was discovered in 2003 by
Conte and Kirch (2004: 87-91), who excavated a 1 m2 test
pit into a cultural deposit extending to 65 cm, terminating
at a charcoal-flecked reddish soil. The test pit yielded nine
pearl shell fishhooks or hook fragments, 11 coral abraders
and several other artefacts. A calibrated radiocarbon date
of AD 1260–1290 was obtained from the base of the
cultural deposit, while a sample from 10 cm below the
© 2015 Oceania Publications
5
surface yielded a calibrated age of AD 1430–1460 (Kirch
et al. 2004, Tables 4.1 and 4.2). In 2005, Conte and Kirch
started to reopen excavations at AGA-3, but were forced to
abandon the effort when a heavy rainstorm flooded the
shelter floor, bringing cobbles crashing down off the cliff,
making it too dangerous to continue.
Returning to the site in 2012, it was apparent that
erosion had continued since 2005, with a thick rocky
colluvium covering the shelter floor, obscuring previously
visible surface features. The exact location of the 2003 test
pit could not initially be discerned, although this was
uncovered during the renewed excavations.
Due to the increased erosion, since 2003 the sediment
within the rock shelter has become repeatedly
water-saturated, with extensive leaching and chemical
decomposition of calcium-carbonate materials, including
shell midden and pearl shell artefacts. The 2003 test pit
yielded a total of 2.21 kg of shell midden from Layer II.
When excavated in 2012, adjacent units D13 and E14
yielded only 0.36 and 0.11 kg of shell respectively, all of it
chalky and heavily decomposed. Thin-walled shells such as
those of Cellana taitensis (with 0.50 kg in the 2003 test pit)
were completely lacking from the adjacent units in 2012,
with only thick-walled species such as Turbo setosus and
Lambis truncata surviving. This taphonomic situation also
affected the survival of pearlshell fishhooks, of which nine
were found in the single meter test square in 2003, while
only a single, large, very eroded pearl shell fishhook was
recovered in the much more extensive 2012 excavation.
Thus within less than a decade, the majority of the shells
and artefacts made of shell within the shelter seem to have
entirely decomposed. This situation did not affect the
preservation or recovery of bone; comparison of bone
concentrations between the 2003 and 2012 excavations
reveal no significant differences (e.g. 1709 NISP bone in
TP1 from 2003 versus 1602 NISP from unit D13 in 2012).
The 2012 excavations commenced in a trench (units
D9–F9) extending out from the rock shelter’s wall to the
dripline (Figure 3). After removing the recent colluvium
(Layer I), we encountered the Layer II cultural deposit,
taking this down to the reddish clay with charcoal
flecking, which Conte and Kirch (2004) identified as an
original ground surface. We then opened a 6-m2 block
excavation, in order to expose a greater area of the cultural
deposit. Figure 2b shows the rock shelter viewed from the
north, with the D9–F9 trench and excavations in progress
in the block excavation. The block excavation was taken
down to the base of the Layer II cultural deposit
(Figure 2c), exposing a small combustion feature along
with nine complete stone adzes, arrayed near to a large
boulder (Figure 4).
Although Conte and Kirch (2004) interpreted Layer III
as a natural soil, we wanted to confirm that no deeper
cultural deposits were present. After digging through
25–30 cm of compact clay in the D9–F9 trench, a deeper
cultural deposit (Layer IV) appeared. We then halted work
in the block excavation, concentrating on the D9–F9
trench through the 60 cm thick Layer IV cultural deposit.
6
Human ecodynamics in Mangareva
Figure 4. A plan of the block excavation at AGA-3,
showing the Feature 3 hearth and the locations of the
basalt adzes found in situ. Note the location of the 2003
TP1.
Because the rock shelter wall slopes inward, this opened
up part of an additional unit, G9, in the deepest part of the
trench. Layer IV yielded a sizeable faunal assemblage,
including numerous seabird bones near at the interface
with the original ground surface (Layer V), similar to the
situation at the Onemea site on adjacent Taravai Island
(Kirch et al. 2010).
The stratigraphy within Layer IV was complicated by a
concentration of large boulders and a slab of calcareous
beach rock in unit D9; these had been transported to the
shelter and placed there, possibly to make a wall across
the front of the shelter. The emplacement of these boulders
resulted in disturbance of the bottom of Layer IV.
Time restrictions did not allow us to carry the block
excavation down to Layer IV, a task that was deferred for
a future field season. After covering the floor of the block
excavation with heavy plastic tarps the excavation was
backfilled.
A stratigraphic section through the rock-shelter deposits
in the D9–G9 trench was described in the field as follows
(Figure 5):
Layer IA. Dark reddish brown (5 YR 3/25 YR 3/2) clay
loam, penetrated by small rootlets. Angular peds, very
compact. Some subangular basalt clasts, most in 1–3 cm
size range. Contact with Layer IB diffuse. This
represents the most recent phase of erosion into the
rock shelter, largely accumulating since 2003.
Layer IB. Dark reddish grey (5 YR 4/2) clay. Very
compact. Structure single grain, very fine sand to clay.
Almost completely lacking in clasts. Contact with IC
diffuse. Very hard when dry.
Layer IC. Dark reddish brown (5 YR 3/2) clay loam.
Similar to Layer IA, with subangular peds. Large
number of small subrounded volcanic clasts
(0.2–2.0 cm size range) giving this layer a gravelly
texture. The clasts are very weathered (saprolite).
Contact with Layer II sharp, slightly irregular.
Layer II. Very dark grey (5 YR 3/1) sandy loam. Upper
cultural deposit, with charcoal inclusions throughout.
Numerous angular dykestones (manuports). The
sediment has a sandy texture. Lower contact with Layer
III slightly diffuse.
Layer III. Reddish brown (5 YR 4/3), very compact
clay with numerous subrounded weathered volcanic
clasts (saprolite). Clasts range in size from 0.2–4.0 cm.
Occasional flecks of charcoal. Layer III appears to be
an in-wash deposit from upslope, possibly a single
major depositional event. Contact with Layer IV sharp,
somewhat irregular.
Layer IV. Black (5 YR 2.5/1) cultural deposit. Soft,
“fluffy” texture, with a relatively low clay content.
Numerous charcoal inclusions with finely dispersed
carbon throughout the deposit. Some dykestone
inclusions present, especially near the top of the
deposit. A few light-coloured (whitish) as lenses visible
near the base of the deposit. Contact with Layer V
somewhat diffuse, mixed at the interface.
Layer V. Reddish brown (5 YR 4/4) clayey gravel, full
of subrounded clasts (1–5 cm size range). No charcoal
observed. Quite soft, not compact. Pre-cultural floor of
the rock shelter.
GEOARCHAEOLOGICAL ANALYSIS OF THE
AGA-3 SEDIMENTS
Figure 6 summarises key sedimentological characteristics
through the AGA-3 stratigraphic sequence, based on
laboratory analysis of the F9 sediment column. Layer V is
poorly sorted, with a substantial component of
pebble-sized (−4φ) grains. The lower cultural deposit,
Layer IV, is mostly made up of very fine sand to silt (>4
φ) particles, although there is a pulse between 120 and
130 cm with a greater influx of larger particles. Layer III,
mostly non-cultural sediment that accumulated during a
period of little or no occupation, is dominated by medium
to very fine-grained sands and silts (1–4 φ), which
originated from the steep slopes above the shelter. The
upper cultural deposit, Layer II, shows a marked increase
in larger clasts (−3 to −4 φ), probably due to cultural
activities, while the post-occupation Layer I is dominated
by finer-grained sand to silt and clay-sized particles
derived from the denuded slopes above the site.
© 2015 Oceania Publications
Archaeology in Oceania
7
Figure 5. A stratigraphic section through the D9–F9 trench. The upper diagram shows the south and west walls, while the
lower diagram shows the east wall of the trench. Note the large boulders and the beachrock slab in unit D9, probably
reflecting wall construction across the front of the rock shelter.
Figure 6. The composition of the 0 φ size fraction from the F9 sediment column in AGA-3.
© 2015 Oceania Publications
8
Human ecodynamics in Mangareva
Microscopic analysis of the 0 φ fraction reveals marked
differences between the stratigraphic units, reflecting
substantial temporal changes in site use and deposition.
The deepest layers, V and IV, have the highest percentages
of rock grains (subangular basalt and andesite grains),
while the percentage of saprolitic grains (subangular
particles of strongly oxidised and degraded rock) increases
dramatically in Layers III and I; this change probably
reflects deforestation and increased exposure of weathered
saprolite on the steep slopes above the rock shelter. The
earlier cultural deposit, Layer IV, displays high quantities
of charcoal (20–35%) and bone (10–25%) fragments,
while the upper cultural deposit, Layer II, has only minor
quantities of bone and charcoal (10 and 6%). These
differences may indicate more intensive cooking and food
processing activities in the shelter during the earlier period
of use. Layer II is also marked by a high frequency (74%)
of subangular peds of grey, ashy sediment containing
microscopic charcoal fragments. The formation processes
of these peds are not certain, but they were resistant to
soaking and wet sieving in the laboratory; they may have
formed through repeated wetting and drying of the Layer
II sediment.
was calibrated using the Marine13 curve (Reimer et al.
2013), since isotopic data indicate that this seabird was
subsisting on a marine diet (δ15N/δ14N = +18.7‰); a
marine reservoir offset (ΔR value) of 54 ± 20 was used,
based on Welch et al. (2012), who determined this value
from 28 pre-bomb Pacific petrel bones from museum
collections. All other samples were calibrated using the
ShCal13 calibration curve adjusted for the Southern
Hemisphere (Hogg et al. 2013). CalAD ages are reported
at two standard deviations. Figure 7 is an Oxcal plot of the
seven 2012 samples, with Gaussian probability
distributions shown at two standard deviations.
Samples Beta-332243 and -374442 derive from the base
of Layer IV, in unit F9 towards the rear of the shelter.
These dates on a carbonised Pandanus key and on the
humerus of a Pseudobulweria petrel (situated at the
interface between Layer IV and the underlying
pre-occupation Layer V) are in strong agreement, placing
initial use of the shelter between calAD 1221 and 1375 in
the case of the Pandanus, and between calAD 1196 and
1328 in the case of the bird bone. The next four samples
(Beta-330525, -330526, -332244 and -332245) are from
Layer IV, associated with the large cobbles and beachrock
slab in units D9 and E9, marking construction of a wall
across the front of the shelter. These samples range in age
from calAD 1413–1482 to calAD 1501–1662, indicating
that the upper part of Layer IV was deposited in the
15th–16th centuries. Following the deposition of the
inwashed clay Layer III, renewed use of the shelter is
indicated by sample Beta-330524. This sample has
multiple possible intercepts (with none earlier than calAD
1674), but Layer II contains no artefactual evidence for
sustained use of the site after regular European contact in
the mid-19th century. The highest probability intercept of
calAD 1719–1826 is therefore the most likely estimate of
the final permanent use of the site.
RADIOCARBON DATING AND CHRONOLOGY
As noted earlier, two radiocarbon dates from the 2003 test
pit in AGA-3 (Kirch et al. 2004) suggested that deposition
of Layer II spanned the late 13th to the 15th centuries. In
order to further refine the site’s chronology and occupation
sequence, an additional six samples of short-lived,
botanically identified carbonised plant remains and one
sample of seabird bone were radiocarbon-dated by Beta
Analytic, Inc. (Table 1).
Calibration of the 14C ages was performed with Oxcal
4.2 (Bronk Ramsey 2009). The procellarid seabird bone
Table 1. Radiocarbon age determinations from the 2012 site AGA-3 excavations.
Lab. no.,
Beta-
Provenience
330524
D9, L-3-3, Layer II-1
330525
D9, L-20-1, Layer IV-2
330526
D9, L-21-1, Layer IV-2
332243
F9, L-13-2, Layer IV-3
332244
E9, L-14-3, Layer IV,
stone wall
E9, L-15-1, Layer IV,
stone wall
F9, L-13-3-1, Layer IV-4
332245
374442
Measured
C age (BP)
δ13C (‰)
Carbonised endocarp,
Cocos nucifera
160 ± 30
−25.2
160 ± 30
Carbonised endocarp,
Aleurites moluccana
Carbonised fruit key,
Pandanus tectorius
Carbonised fruit key,
Pandanus tectorius
Carbonised fruit key,
Pandanus tectorius
Carbonised husk,
Cocos nucifera
Left humerus,
Procellariid, cf.
Pseudobulweria sp.
470 ± 30
−23.9
490 ± 30
1664–1707 (16.7%)
1719–1826 (47.4%)
1832–1884 (12.6%)
1914 (18.6%)
1404–1450 (95.4%)
330 ± 30
−25.8
320 ± 30
1483–1646 (95.4%)
810 ± 30
−26.8
780 ± 30
1210–1281 (95.4%)
370 ± 30
−23.3
400 ± 30
380 ± 30
−23.5
400 ± 30
970 ± 30
−12.2
1180 ± 30
1436–1522 (76.4%)
1575–1625 (19.0%)
1436–1522 (76.4%)
1575–1625 (19.0%)
1196–1328 (95.4%)
Material
14
Conventional
C age (BP)
14
Calibrated age range
AD (2σ)†
†Calibrations performed with Oxcal 4.2.
© 2015 Oceania Publications
9
Archaeology in Oceania
Figure 7. The Oxcal plot of calibrated radiocarbon dates from the 2012 AGA-3 excavations.
Figure 8. Portable artefacts from site AGA-3. Upper left: pearl shell fishhook and worked pearl shell pieces. Lower left:
two grooved stone sinkers. Top right (a–d): worked pig bone, probable thatching needles. Lower right: metal medallion.
CULTURAL CONTENT AND ARTEFACTS
Portable artefacts
Fishhook manufacture and fishing equipment
The 2012 AGA-3 excavations yielded a single fishhook
and two worked pieces of pearl shell (Pinctada
margaritifera), in comparison with the nine fishhooks and
47 pieces of worked pearl shell recovered by Conte and
© 2015 Oceania Publications
Kirch from TP1 in 2003. This is a consequence of recent
water-saturation of the rock-shelter deposits, resulting in
rapid chemical decomposition of the calcium-carbonate
shell materials. The broken fishhook E13-4-5 from layer
II-2 is heavily eroded, retaining no trace of manufacture
(Figure 8). In contrast with the “rotating” fishhooks from
AGA-3 (Weisler et al. 2004: 128), this is of the “jabbing”
type (Emory et al. 1959). The hook was quite large, with a
preserved shank section 43 mm long.
10
Two pieces of worked pearl shell were recovered from
Layer II-3. Specimen E14-6-2 is abraded on both faces,
and was probably intended for a large-sized hook.
Specimen E13-6-2 is degraded and fragile but shows a
perforation on one edge that could indicate the shaping
phase of the hook. The excavations also yielded 14
branches of degraded Acropora sp. coral, 11 of which
exhibit evident use-wear on the tip, indicating their use as
abraders or files. Their presence in Layers II and IV attest
to fishhook manufacture.
Other fishing gear includes two sinkers from Layer II in
the block excavation, both made of dense slightly vesicular
basalt (Figure 8). Their shape is irregularly spherical, with
a groove along the longitudinal axis, a common type in
East Polynesia (Lavondès 1971: 351). Specimen F12-3-1
has a diameter of 85 mm and weighs 559 g, while
E13-3-1 is smaller (39 mm in diameter) and weighs
67 g. They could have served as weights for either
fishing lines or nets.
Bone needles
The excavations yielded four mammal rib bones, probably
pig (Sus scrofa), which were cut and shaped to a point at
one end (Figure 8); these are similar to a specimen
discovered in 2003, interpreted as a possible thatching
needle (Weisler et al. 2004: 129). The lengths of the bones
range from 48 to 93 mm.
Basalt adzes
Nine adzes as well as flakes and adze flakes were
recovered from Layer II in the block excavation
(Figure 9). Based on visual inspection, the majority of
the lithic material is of dykestone, veins of which occur
both on Agakauitai and on the neighbouring island of
Taravai. However, the basalt grain texture varies
indicating different sources, with some adzes probably
brought from other localities in the Gambier group. The
nine adzes from Layer II vary in size and morphology,
even though the majority exhibit a trapezoidal
cross-section (Table S2). The length of the blades ranges
from 50 to 129 mm, suggesting that these adzes were
used for different functional tasks. Following Weisler and
Green’s (2001) typology for Mangarevan adzes (see
Weisler et al. 2004), adzes of type 1 are the most
common. The small dimensions of some of these pieces
are due to the process of reshaping, indicated by the
chipping marks and the flakes, which led to a size
reduction of the initial blade, as in specimen E13-4-4.
Adze E12-2-2, made of fine-grained basalt, is polished
on the bevel and inner and outer surfaces, with a
trapezoidal section, having the pronounced apex towards
the back. It is the only adze of type 5 (Weisler and Green
2001: 419), of which few specimens have been discovered
in Mangareva, though this type spans the entire cultural
sequence in Marquesas. Adzes E13-4-3 and E13-6-1 can
be distinguished by their fine-grained basalt texture, along
with their narrow shape and oval to circular cross-sections.
These adzes are well-polished on the entire surface,
Human ecodynamics in Mangareva
though E13-6-1 shows a slight reduction of the butt,
indicating an incipient tang. These adzes fit into the
Hatiheu type described by Suggs (1961b: 110) for the
Marquesas, although the specimens from AGA-3 are
smaller.
Two other adzes suggest an opportunistic and rapid
process of manufacturing. E13-4-2 and E14-4-1 are
preforms on dykestone that could have been used as hafted
tools. The cutting edges were reshaped (but unfinished for
E13-4-2) before being discarded, probably due to the poor
quality of the material.
Historic medallion
An oval, metal medallion from Layer II-1 has engraved
motifs on both faces, too eroded to be recognisable
(Figure 8). This piece probably indicates some brief
visitation at the AGA-3 site in the post-contact period.
Geochemical sourcing of adzes
Nine adzes recovered from the block excavation, along
with two geological specimens from Mangareva Island,
were analysed at the UH Hilo Geoarchaeology Laboratory
(see Methods, above for analytical technique). All nine
adzes as well as the geological specimens display
geochemical characteristics of tholeiitic basalts, with
relatively low trace element values (Figure 10) that do not
match well with any of the known major central Eastern
Polynesian quarries documented for the Marquesas,
Society Islands or Pitcairn (Allen & McAlister 2013;
Charleux et al. 2014; Kahn et al. 2013; Rolett 1998; Rolett
et al. 1997; Sinton & Sinoto 1997). These results suggest
that either some fine-grained basalts were available within
the Mangareva group for adze manufacture, or that an
unknown tholeiitic fine-grained source was being exploited
elsewhere.
Weisler (1996, 1997) noted that Mangareva is
predominantly composed of tholeiitic basalt that is
typically produced during shield-building phases of Pacific
Island volcanoes, and that very little alkali basalt is
present. No prehistoric quarry sites are presently known in
Mangareva, and few examples of fine-grained basalts that
would be suitable for adze manufacture. Weisler inferred
from the presence of a few tholeiitic basalt flakes found on
Henderson that Mangarevans may have engaged in
“opportunistic use of local, but inferior lithic sources for
adze manufacture” (Weisler 1997: 164).
The similarity of the geological standard from Kı̄lauea
Volcano (BHVO-2) to the nine adzes from AGA-3 and two
geological samples from Mangareva points out the
difficulty in using the narrow range of trace elements
accessible by EDXRF to discriminate local and non-local
sources (especially with tholeiitic rocks). Our preliminary
geochemical data demonstrate that the AGA-3 adzes do
not bear any geochemical similarity to known quarry
sources in the Marquesas, Society Islands or on Pitcairn,
while the presence of cobble cortex points to the
likelihood of expedient local manufacture. Nonetheless, we
cannot rule out the possibility that an as yet unknown
© 2015 Oceania Publications
Archaeology in Oceania
Figure 9. Basalt adzes from site AGA-3.
© 2015 Oceania Publications
11
12
Human ecodynamics in Mangareva
Figure 10. The geochemistry (Sr/Zr ratio) of adzes and geological samples from Mangareva compared with known
Eastern Polynesian sources.
Table 2. Bird bones (NISP) from site AGA-3.
I
Procellariid sp. cf. Pseudobulweria
Procellariid sp. cf. Puffinus
Procellariid sp. cf. Pterodroma
Procellariid sp.
Phasianidae sp. cf. Gallus gallus
Unknown, but potentially identifiable
Unidentifiable, small fragments
Total
II-1
II-2
2
II-3
III
2
1
IV-1
4
1
1
1
1
0
12
16
fine-grained tholeiitic source was moving between islands;
this possibility could be addressed with additional isotopic
analyses.
Faunal remains
Birds
A total NISP of 312 bird bones was recovered from the
D9–G9 trench (Table 2); 91% of the bird bones are from
Layer IV, especially from the lower part of that deposit.
Half of the specimens consisted of small fragments, not
identifiable to taxon. Of the identifiable specimens, 86%
are of procellariid seabirds (petrels and shearwaters),
including species referred to the genera Pseudobulweria,
Puffinus and Pterodroma. The majority of the identifiable
procellariids are a species of Pseudobulweria, quite likely
P. rostrata (the Tahiti petrel) which still exists in
Mangareva in very small numbers (Thibault & Cibois
2012; Waugh et al. 2013). Several of the Puffinus bones
are probably P. nativitatus (Christmas Island shearwater),
although more than one species may be presented;
P. nativitatus, P. pacificus and P. lherminieri are all known
from Mangareva in recent times, again in low numbers.
1
3
6
3
4
9
13
IV-2
IV-3
IV-4
IV/V
Total
14
19
1
2
8
15
2
21
45
8
48
91
38
5
1
22
1
6
55
128
82
6
1
46
2
19
156
312
2
4
8
Only one specimen of Pterodroma was identified; both
P. heraldica and P. ultima are known from the Gambiers
today. An additional 19 bones are potentially identifiable
but could not be referred to specific taxa with the
reference collections available.
Two tibiotarsus specimens were tentatively identified as
chicken (Gallus gallus), although the bones are smaller
and more gracile than skeletons of domestic chickens in
the MVZ collection. Green and Weisler (2004: 36)
reported the presence of three chicken bones from the
GK-1 and -2 sites on Kamaka Island.
Rat
Bones of Polynesian-introduced Rattus exulans and of
European-introduced R. rattus were identified from
AGA-3, the latter present only in Layers I and II-2. Rattus
exulans is present from the beginning of the sequence, but
becomes especially prevalent in the Layer II cultural
deposit (Table 3).
Pig and dog
Pig (Sus scrofa) was ethnographically reported to have
been exterminated in Mangareva prior to European
contact, possibly as a result of trophic competition with
© 2015 Oceania Publications
13
Archaeology in Oceania
Table 3. Mammal bone from site AGA-3.
Layer
I
II1
II2
II3
III
IV1
IV2
IV3
IV4
IV/V
Total
Rattus exulans
Rattus rattus
Rattus sp.
Sus scrofa
Canis familiaris
(teeth)
Miscellaneous
mammal bone
Total
NISP
Weight
NISP
Weight
NISP
Weight
NISP
NISP
Weight
NISP
NISP
Weight
3
0.33
59
247
160
11
21
6
8
4
26
542
2.16
11.47
7.74
0.6
0.64
0.22
0.28
0.19
1.19
24.49
0.11
0.02
0.2
1
10
277
4
0.21
8.65
74.16
1.82
2
1.56
7
9.18
1
10.35
300
104.37
7
72
599
167
11
29
14
13
4
28
944
0.65
12.39
104.09
11.07
0.6
10.63
1.87
12.26
0.19
1.93
155.68
1
4
0.08
0.41
3
1
8
12
0.33
humans for the islands’ limited food supply (Hiroa 1938:
194-5; Kirch 2001). Green and Weisler (2004) reported
rare pig bones or teeth in Green’s 1959 rock-shelter
excavations, while Conte and Kirch (2004: 117) found a
single pig tooth in AGA-3 during the 2003 test excavation.
In 2012, 300 NISP Sus scrofa bones (104.37 g) were
found, predominantly in Layer II (Table 2). A large
proportion (N = 256, 61%) of these come from the remains
of a single juvenile pig in units E9 and F9, Layer II-2; the
total pig MNI for the site consists of just one juvenile and
one adult. Two teeth of Canis familiaris were identified
from Layer II-1, but these may represent a post-contact
introduction of dog.
Additionally, 83 NISP of miscellaneous medium
mammal (23.7 g) were identified; these consist of bone
fragments that are not large enough to differentiate to
species and could include pig, dog or human.
Sea turtle
Thirty-nine NISP of sea turtle were identified (29.7 g)
including plastron fragments, one leg bone and other
fragments. Six turtle bones come from the top of Layer IV
in the main trench, while the remainder were excavated
from Layer II.
Fish
Out of a total NISP of 26537 fishbones (weighing
2159.2 g), 1231 NISP were identified to 15 families and
19 genera (Table S3). A few taxa strongly dominate the
assemblage, especially groupers (Serranidae) and
parrotfish (Scaridae, predominantly species in the genus
Scarus), which occupy the first two ranks in order of
frequency, followed more distantly by surgeonfishes
(Acanthuridae), wrasses (Labridae), and squirrelfish
(Holocentridae). This is similar to rank order of dominant
taxa reported by Weisler and Green (2013) for
Mangarevan sites excavated by Green in 1959. All of the
taxa represented occur on the reef or lagoon habitats found
close to Agakauitai Island; no pelagic taxa are present.
Measurements of fish vertebrae widths from all layers
© 2015 Oceania Publications
Weight
2
1.56
Weight
66
3
18.18
1.51
1
8
4
0.81
1.65
1.63
2
83
0.74
23.71
showed no statistically significant differences throughout
the stratigraphic sequence, suggesting the absence of
marine resource depression.
Marine molluscs
Recent periodic flooding of the AGA-3 deposits has led to
much of the rock shelter’s shell content being dissolved,
with only larger shell pieces remaining in a degraded and
chalky condition. The drastic reduction in the quantity of
shell remaining is evident by comparing the quantity of
shell reported by Howard and Kirch (2004) from TP1
excavated in 2003. Layer II in this pit yielded 2586.1 g of
shell, whereas the greatly expanded 2012 excavation in
total yielded only 4064.21 g.
Marine shell at AGA-3 totalled 2587 NISP (Table S4).
Due to its extremely degraded condition, in many cases
the shell could only be identified to genus. The assemblage
is dominated by Turbo sp. and by pearl shell (Pinctada),
with much of the latter probably representing detritus from
fishhook manufacture.
Stable isotope analysis of Rattus exulans
Figure 11 displays preliminary results of isotopic analysis
with average δ13C and δ15N values for Rattus exulans bone
collagen from AGA-3 by stratigraphic unit, providing
evidence for impacts of rats on the local environment and
avifaunal populations (see also Table S1). 13C/12C ratios are
effective in distinguishing C3 from either terrestrial C4 or
marine resources, while 15N/14N provides an indication of
trophic level, and aids in distinguishing terrestrial versus
marine resources. Most individuals from the deeper
sub-layers of Layer IV cluster together within a range of
δ13C values from −19 to −17 and δ15N values between 12
and 13. Layer IV-1 contains individuals with the highest
δ15N values and most negative δ13C values in the sequence.
Moving into later layers, average values follow a temporal
trend of decreasing δ15N and less negative δ13C values.
Although Layer III does not fit within this overall trend,
this is probably due in part to the small sample of
individuals from Layer III (N = 3). In addition to these
14
Figure 11. δ13C and δ15N values for Rattus exulans bones
from the AGA-3 site.
Human ecodynamics in Mangareva
availability. As R. exulans is an omnivorous and
commensal species with a limited home range
(Spennemann 1997), reconstructed rat diet might be a
reliable proxy indicator of human diet and resource use at
the site. These implications will be explored in future
research.
Palaeobotanical remains
trends, rat diet becomes more variable overall later in the
sequence, especially with respect to δ13C values.
There are several possible explanations for these trends.
Extinction and extirpation of avifauna following
Polynesian colonisation may have eliminated a high-order
protein contribution to R. exulans diet, lowering δ15N
values by one to two trophic levels. If the decline in δ15N
was primarily a result of the extirpation of seabirds, one
would expect a corollary trend of more negative δ13C
values. However, the opposite trend is evident, indicating
that if birds were a key component of early rat diet, these
would probably have been primarily terrestrial species.
This may support Green and Weisler’s (2004) suggestion
that Gallus gallus disappeared on Mangareva due to rat
predation, although the limited presence of Gallus gallus
in Mangarevan sites is too sparse to corroborate a strong
early dietary reliance on chicken.
The impacts of avifaunal extinctions and extirpations on
nutrient cycling and nitrogen availability must also be
considered. Small-island ecosystems relying on
marine-derived nutrient subsidies (such as seabird guano)
tend to have higher baseline δ15N values than other regions
(Anderson & Polis 1999; Briggs et al. 2012; Fukami et al.
2006; Mizota & Naikatini 2007). The apparent decline in
trophic level of R. exulans over time at AGA-3 may be a
reflection of changes in nitrogen availability and baseline
15
N/14N on Agakauitai. Finally, some correlation exists
between rat dietary isotope values and the fauna recovered
for each stratigraphic unit. The decline of δ15N through
time correlates with the steep temporal decline in avifaunal
remains. Additionally, quantities of fish bone show a
general increase in NISP over time, which correlates with
the decreasingly negative δ13C values in rat diet. It is
therefore possible that the primary influence on rat diet at
the site was changes in human diet and resource
Macroscopic charcoal
Macroscopic charcoal collected during dry sieving of unit
F9 was analysed from five levels, representative of
Layers I, II and IV (Table 4). The deepest two levels,
from Layer IV-4 and IV-3, are dominated by Thespesia
populnea, an indigenous coastal shrub or small tree, and
by Hibiscus tiliaceous, another indigenous shrub that
ranges from littoral to inland valley habitats and is
regarded as invasive in disturbed environments. Other
indigenous taxa present in these oldest samples include
the large strand tree Barringtonia asiatica, Pandanus sp.
(probably P. tectorius, widely distributed in Polynesia),
Fagraea berteroana, a coastal to lowland tree,
Heliotropium foertherianum, a small littoral tree, and
Ficus cf. prolixa, a fig tree, possibly being the banyan
tree. All of these indigenous taxa have socio-economic
uses in Polynesia (Butaud et al. 2008; Elevitch 2006).
Butaud et al. (2008: 113) suggest that F. berteroana, a
species indigenous to most of the Pacific (Elevitch 2006),
is a recent introduction to the Gambiers, where it is now
only found cultivated, a conclusion that is contradicted
by our evidence. Well represented in medium to low
frequencies are Polynesian introductions or cultivars, such
as the medicinally important Morinda citrifolia, the
breadfruit tree (Artocarpus altilis), the naturalised
candlenut (Aleurites moluccana), the cultivated paper
mulberry (Broussonetia papyrifera) and the coconut
(Cocos nucifera). Both M. citrifolia and C. nucifera are
likely to be indigenous species from which cultivars were
dispersed by Polynesians (Butaud et al. 2008; Kahn et al.
in press). The Layer IV anthracological assemblage
suggests that, at the time of its earlier occupation, the
vegetation surrounding the AGA-3 site was already
largely anthropogenic, albeit comprising more indigenous
cultivated taxa than Polynesian introductions. Indeed,
taxa dominating the sample in both number and
frequencies are important Polynesian cultigens, while in
parallel, indigenous coastal and lowland trees and shrubs
are more represented than confirmed Polynesian
introductions.
The samples from Layers II-3 and II-2 are heavily
dominated by three taxa: Hibiscus tiliaceous, Pandanus
sp. and breadfruit; as noted above, H. tiliaceous is
invasive in disturbed habitats, while the breadfruit tree
was a dominant source of staple starch in Mangareva
(Hiroa 1938). Thespesia and Barringtonia are absent,
although two coastal trees important for timber are
represented: Calophyllum inophyllum and Cordia
subcordata. The first is nowadays considered as a
© 2015 Oceania Publications
15
Archaeology in Oceania
Table 4. Identified charcoal from unit F9 of site AGA-3.
Taxon
Level 2
Layer I
Level 5
Layer II-2
Level 6
Layer II-3
12
15
1
5
1
Level 12
Layer IV-3
Historical introductions
Mangifera indica
Psidium guayava
Syzygium cf. cumini
Polynesian introductions
Artocarpus altilis
Syzygium malaccense
Morinda citirolia
Cordyline fruticosa
Aleurites moluccana
Broussonetia papyrifera
Cocos nucifera
Indigenous coastal
Thespesia populnea
Calophyllum inophyllum
Cordia subcordata
Barringtonia asiatica
Other indigenous
Pandanus
Hibiscus tiliaceous
Fagraea berteroana
Ficus cf. prolixa
Heliotropium foertherianum
Colubrina asiatica
Family-level identifications and unidentified
Malvaceae
Rubiaceae
Euphorbiaceae
Unidentified
Total (number of fragments)
5
48
2
1
2
2
51
Taxa relative frequencies (%)
Layer I
Layer II
Layer IV
Historical introductions
Polynesian introductions
Indigenous
All family-level, likely to be indigenous, taxa
Unidentified
19
54
17
0
10
0
35
56
6
4
0
17
65
16
2
6
2
1
19
4
1
2
2
2
1
1
5
13
13
5
4
4
1
1
1
1
11
9
3
1
2
10
1
3
3
2
18
2
2
5
15
1
1
1
2
Polynesian introduction in East Polynesia, and was often
planted on sacred sites as well as cultivated for various
socio-economic uses (Butaud et al. 2008: 321-5).
Polynesian-introduced candlenut and M. citifolia continue
to be represented and are joined by Cordyline fruticosa
and the Malay apple (Syzygium malaccense) The banyan
tree is also present, along with Colubrina asiatica, an
indigenous shrub to small tree. The Layer II assemblage
is thus also representative of an anthropogenic vegetation,
being dominated by economic plants, taxa responsive to
disturbance or species useful for their timber, medicinal
properties or symbolic role. In contrast with the older,
underlying assemblage, however, Polynesian introductions
are now slightly more represented both in frequency and
number of taxa.
The sample from Layer I reflects post-European contact
vegetation; the charcoal in this recent sediment
incorporates materials derived from burning the island’s
slopes during the past century or so. The assemblage is
dominated by breadfruit charcoal, but there are also
© 2015 Oceania Publications
Level 13
Layer IV-4
1
7
2
49
1
51
2
5
2
1
49
quantities of the historically introduced mango (Mangifera
indica), guava (Psidium guayava) and Java plum
(Syzygium cumini). Other Polynesian cultivars and the
ruderal H. tiliaceous as well as Pandanus, still common on
Agakauitai, still continue to be represented.
Plant microfossils
Four samples from the sediment column in unit F9 (from
Layer II-1 (38–48 cm), Layer III (60–70 cm), Layer IV-1
(100–110 cm) and Layer IV-4 (130–140 cm) were analysed
for pollen and phytoliths. The samples from II-1 and III
contained sufficient pollen for counting. Fern spores
dominate the pollen assemblages in these upper samples,
particularly monolete psilate spores, reflecting local
(unidentified) ground fern growth. Layer III contains a
large amount of Pandanus tectorius pollen, a species also
common on the degraded Agakauitai hillsides. Coconut
and Casuarina equisetifolia pollen was also found in this
sample; both taxa are Polynesian introductions (Whistler
2009). Pollen of Ipomoea sp. was found in this sample,
16
Human ecodynamics in Mangareva
although it is too degraded to differentiate between that of
Polynesian-introduced sweet potato (I. batatas) and
indigenous species of this genus. Small quantities of
Pandanus and coconut pollen were found in the Layer
IV-4 sample. The hornwort spores found in the Layer II-1
sample suggest local landscape disturbance, as hornworts
are small inconspicuous plants that colonise freshly
exposed soils (Horrocks et al. 2012; Wilmshurst et al.
1999).
Phytoliths were well preserved in all samples from
AGA-3, dominated by grass (chloridoid, panicoid bilobate
and bulliform elongate types) and palm phytoliths,
reflecting either local grass and palm growth, or use of
leaves of these taxa by people in the shelter. Leaf
phytoliths of Polynesian-introduced banana (unequivocal)
were found in the two lowest samples and the uppermost
samples, indicating processing of this crop within the
shelter.
SEDIMENTARY FAN ACCUMULATION AT
SITE AGAK-2
Additional evidence for environmental change on
Agakauitai Island comes from a colluvial fan 175 m
south-west of site AGA-3. A trench (AGAK-2) was dug
into this fan by O. Chadwick and N. Porch during a soil
sampling program carried out by our team in 2012. The
trench is 70 m from the present shoreline, at an elevation
of 5 m above sea level, on the surface of a gently sloping
colluvial fan emanating from a steeply rising drainage to
the south, with intermittent flow during rainstorms. The
trench penetrated through 2.33 m of dark brown (7.5 YR
3/4) clay and saprolitic clastics, at which depth a deposit
of calcareous beach sand containing flecks of charcoal,
small quantities of fishbone and a basalt flake was
exposed. This is apparently a low-density midden situated
atop an older beach ridge. Excavating this calcareous
deposit to 2.55 m depth, a circular (∼30 cm diameter),
clay-filled feature was uncovered, evidently a posthole.
The trench area was too small to expose more of the
buried structure.
A sample of dispersed charcoal flecks (unfortunately
too small for identification) was collected from the
calcareous midden deposit immediately adjacent to the
posthole. This was AMS dated (Beta-374441) to 920 ± 30
BP (δ13C/δ12C = −26.9‰), with an age range of calAD
1028–1184 (2σ). If the charcoal derived from burning of
indigenous forest, there is a possibility of a modest in-built
age of perhaps a century, but the date is nonetheless
consistent with initial Polynesian occupation of the
Gambier archipelago in the early first millennium AD
(Kirch et al. 2010). The accumulation of more than 2 m of
colluvium over the island’s original beach flat adds further
evidence for anthropogenic landscape change following
Polynesian colonisation.
Also noteworthy from the AGAK-2 trench was the
presence – at the interface between the clay colluvium and
the underlying calcareous beach sand – of specimens of
endemic endodontid land snails. This assemblage of
terrestrial molluscs (to be reported on in detail elsewhere)
is indicative of the former presence of an endemic biota,
prior to major habitat disturbance.
DISCUSSION AND CONCLUSIONS
The 2012 excavations at AGA-3 revealed a well-stratified
sequence covering a time period from the 13th to 17th
centuries (Layer IV), followed by a depositional hiatus
(Layer III) and a final brief phase of use in the late 18th to
early 19th centuries. This sequence thus spans the period
from early phase of Polynesian occupation in the Gambier
Islands to the time of European contact. Multiple lines of
evidence have been used to assess changes in the
terrestrial ecology of Agakauitai Island, as well as
evidence for exploitation of the inshore marine
environment. While there is no indication that human
activity had any significant impact on nearshore marine
resources, the evidence for anthropogenic transformation
of the island’s terrestrial ecosystem is overwhelming.
The most striking signal of environmental change is the
dramatic reduction in bird bones in the AGA-3 faunal
sequence (Figure 12), mirroring a similar pattern
previously identified at the Onemea (TAR-6) site on
nearby Taravai Island (Kirch et al. 2010). The
zooarchaeological records from both of these sites,
combined with data from Green’s 1959 excavations
(Steadman & Justice 1998), leaves no doubt that prior to
Polynesian arrival, the volcanic islets of the Gambier
group were a major nesting ground for several species of
procellariid seabirds, especially the Tahiti petrel
(Pseudobulweria rostrata) or related species. At AGA-3,
bones of these seabirds are heavily concentrated in the
lower part of Layer IV, dating to the 13th century. Given
the reduced frequency of bones in the upper part of Layer
IV and their paucity in Layer II, it is likely that the
populations of these seabirds had become greatly
diminished after about AD 1300–1400.
Figure 12. The stratigraphic distribution of bird and rat
bones in the AGA-3 site.
© 2015 Oceania Publications
17
Archaeology in Oceania
The opposite trend holds for the Polynesian-introduced
rat, Rattus exulans. The rat is present from the beginning
of the AGA-3 sequence (and is also present through the
TAR-6 sequence on Taravai), but only becomes dominant
in the late Layer II deposit (Figure 12). Our faunal data
also confirm the former presence – in very limited
numbers – of Polynesian-introduced pigs and chickens,
both of which were apparently absent from the island at
the time of initial European contact (Hiroa 1938). There is
no indication that either pigs or chickens were ever raised
in large numbers on Agakauitai.
The sequence of macroscopic charcoal from AGA-3
documents a rapid and continual transformation of the
island’s vegetation, from an initial prehistoric phase
dominated by indigenous coastal and lowland taxa
combined with some Polynesian introductions, to a later
prehistoric phase dominated by economic taxa and
indigenous taxa responsive to disturbance, to a
post-contact phase with additional economic introductions
and disturbance tolerant taxa. Supporting evidence comes
from the pollen and phytolith samples, demonstrating a
substantial increase in fernland and in Pandanus (also a
fire-resistant species) in the later prehistoric phase.
Geoarchaeological analysis of the sediments within
AGA-3 suggest increased exposure of the hillslope above
the rock shelter, resulting in increased influx of saprolitic
grains as well as the major event represented by Layer III,
possibly the result of an unusually high energy rainfall
event.
The causes of deforestation on certain Pacific Islands
following human colonisation have been the subject of
some debate (e.g. Athens et al. 2002; Hunt 2007; Rolett &
Diamond 2004), with both direct forest clearance
including the use of fire, and the effects of rat predation on
seeds and seedlings being invoked as key factors – but see
Meyer & Butaud (2009) and Mieth and Bork (2010) for a
contrary view on the role of rats. The sequence from the
AGA-3 Rock Shelter supports a multi-causal model of
dynamic interactions among humans, commensal rats, the
island’s vegetation communities, and populations of
roosting and nesting seabirds. The transformation of
Agakauitai’s plant life began with Polynesian introduction
of a suite of economic plants including coconut, breadfruit,
candlenut and other taxa. Geoarchaeological evidence
from the rock shelter (such as the erosional deposit Layer
III incorporating considerable charcoal) indicates extensive
clearance with the use of fire and heightened erosion over
time. Rats may also have played a secondary role in the
conversion of the indigenous vegetation communities,
although they do not become truly abundant until later in
the sequence.
Equally important, and frequently overlooked in
discussions of island deforestation, is the likely effect of
the dramatic reduction in seabird populations within a
century or two following human colonisation. On
geologically older islands such as those of the Gambier
group, rock-derived nutrients such as phosphorus are
typically depleted (Vitousek 2004). Substantial nutrient
© 2015 Oceania Publications
inputs from nesting seabird populations (through
deposition of their guano) are likely to have played a
major role in maintaining nutrient flows in the Agakauitai
terrestrial ecosystem prior to human arrival – see
Anderson & Polis (1999) and Croll et al. (2005)
regarding seabird nutrient inputs to island ecosystems.
Young et al. (2010) have shown how replacement of
indigenous trees such as Pisonia and Tournefortia with
coconut palms on Palmyra atoll disrupted seabird nesting,
leading in turn to nutrient depletion of the underlying
soils.
We hypothesise that a similar process occurred on
Agakauitai, as the expansion of economic plants (including
coconut) reduced seabird nesting habitats. This, combined
with direct predation on the seabird populations by
humans (and possible secondary predation on
ground-nesting fledgling birds by rats), led to dramatic
reductions in the population of seabirds, in turn
precipitating a significant decline in nutrient inputs. The
decline in Rattus exulans δ15N values over time likewise
appears to signal this disappearance of seabirds on
Agakauitai, either as a record of the disappearance of
seabird from rat diet or of a decline in baseline 15N
availability within the island system as a whole. As
temporal changes in δ13C values do not correspond to
expected patterns for the elimination of seabird from rat
diet, the latter option appears more likely.
In sum, the AGA-3 sequence documents a dramatic
transformation of Agakauitai’s terrestrial ecosystem
following Polynesian colonisation. This transformation
most probably had multiple causes, best explained by a set
of dynamic interactions between the island’s indigenous
vegetation and original seabird populations, and the
colonising humans with their introduced economic plants
and commensal rats. The outcome was a seriously
degraded ecosystem with reduced terrestrial biodiversity,
eroded and nutrient depleted soils and largely deforested
landscape, as witnessed in post-contact times.
ACKNOWLEDGEMENTS
We thank the Service de la Culture et du Patrimoine,
French Polynesia, for permission to conduct archaeological
investigations in the Gambier Islands. The research was
funded by the U. S. National Science Foundation (Grant
BCS-1030049). Mme. Monique Richeton, Mayor of
Rikitea, graciously assisted with arrangements for
accommodation. We thank the CIRAP centre at the
University of French Polynesia for lending equipment and
other assistance. On Taravai, we were hosted by Edouard
and Denise Sanford, who were incredibly gracious. Boat
transport between islands was skilfully provided by the
late M. Tepano Paeamara; Helene Paeamara also helped
our team in numerous ways. We thank Tihoni Reasin,
Raruna Reasin and Tehotu Reasin for assistance during our
stay in Rikitea Village. Rose Guthrie and Jennifer Kahn
assisted in the AGA-3 excavations. Aymeric Hermann
18
Human ecodynamics in Mangareva
obtained the two geological samples from Mt Duff used in
our EDXRF analysis.
REFERENCES
Allen, M.S. and McAlister, A.J. 2013. Early Marquesan
settlement and patterns of interaction: new insights from
Hatiheu Valley, Nuku Hiva Island. Journal of Pacific
Archaeology 4 (1): 90–109.
Ambrose, S.H. 1990. Preparation and characterization of bone
and tooth collagen for isotope analysis. Journal of
Archaeological Science 17: 431–451.
Anderson, A., Conte, E., Kirch, P.V. and Weisler, M. 2003.
Cultural chronology in Mangareva (Gambier Islands), French
Polynesia: Evidence from recent radiocarbon dating. Journal
of the Polynesian Society 112: 119–140.
Anderson, W.B. and Polis, G.A. 1999. Nutrient fluxes from water
to land: Seabirds affect plant nutrient status on Gulf of
California Islands. Oecologia 1183: 324–332.
Athens, J.S., Tuggle, H.D., Ward, J.V. and Welch, D.J. 2002.
Avifaunal extinctions, vegetation change, and Polynesian
impacts in prehistoric Hawaii. Archaeology in Oceania 37:
57–78.
Briggs, A.A., Young, H.S., McCauley, D.J., Hathaway, S.A.,
Dirzo, R. and Fisher, R.N. 2012. Effects of spatial subsidies
and habitat structure on the foraging ecology and size of
geckos. PLoS ONE 7: 1–10.
Bronk Ramsey, C. 2009. Bayesian analysis of radiocarbon dates.
Radiocarbon 51: 337–360.
Brooke, M. de L., O’Connell, T.C., Wingate, D., Madeiros, J.,
Hilton, G.M. and Ratcliffe, N. 2010. Potential for rat
predation to cause decline of the globally threatened
Henderson petrel Pterodroma atrata: Evidence from the field,
stable isotopes, and population modeling. Endangered Species
Research 11: 47–59.
Butaud, J.F. 2013. Gambier. Guide Floristique. Direction de
l’Environnement, Papeete, Tahiti.
Butaud, J.F., Gérard, J. and Guibal, D. 2008. Guide des Arbres
de Polynésie Française: Bois et Utilisations. Au Vent des Îles,
Pape’ete, Tahiti.
Charleux, M., McAlister, A., Mills, P.R. and Lundblad, S.P.
2014. Non-destructive XRF analyses of fine-grained basalts
from Eiao, Marquesas Islands. Journal of Pacific Archaeology
5: 75–89.
Cochereau, P. 1974. Ebauche d’un inventaire faunistique de l’ile
Mangareva (Archipel des Gambier). Cahiers du Pacifique No.
18, Vol. II, 479–532. Fondation Singer-Polignac, Paris.
Conte, E. and Kirch, P.V. (eds). 2004. Archaeological
Investigations in the Mangareva Islands, French Polynesia.
Contributions of the Archaeological Research Facility, No. 62.
University of California, Berkeley, CA.
Cooke, C.M., Jr 1935. Report of C. Montague Cooke, Jr.,
malacologist and leader [of the 1934 Mangarevan Expedition].
In Report of the Director for 1934. Bernice P. Bishop
Museum Bulletin 133. Bishop Museum, Honolulu, HI.
Croll, D.A., Maron, J.L., Estes, J.A., Danner, E.M. and Byrd,
G.V. 2005. Introduced predators transform subarctic islands
from grassland to tundra. Science 307: 1959–1961.
Dean, W.E., Jr. 1974. Determination of carbonate and organic
matter in calcareous sediments and sedimentary rocks by loss
on ignition: Comparison with other methods. Journal of
Sedimentary Petrology 44: 242–248.
Dotte-Sarout, E. 2010. “The Ancestor Wood”. Trees, Forests and
Precolonial Kanak Settlement on New Caledonia Grande
Terre. Three volumes including Wood Atlas. Unpublished PhD
thesis, Université Paris I Sorbonne, Paris / Australian National
University, Canberra.
Dye, T.S. and Longenecker, K.R. 2004. Manual of Hawaiian
Fish Remains Identification Based on the Skeletal Reference
Collection of Alan C. Ziegler and Including Otoliths. Society
for Hawaiian Archaeology, Honolulu. HI.
Elevitch, C. (ed.). 2006. Traditional Trees of Pacific Islands:
Their Culture, Environment, and Use. University of Hawai’i
Press, Honolulu, HI.
Emory, K.P. 1939. Archaeology of Mangareva and Neighboring
Atolls. Bernice P. Bishop Museum Bulletin 163. Bishop
Museum, Honolulu, HI.
Emory, K.P., Bonk, W. and Sinoto, Y.H. 1959. Hawaiian
Archaeology: Fishhooks. Bishop Museum Press, Honolulu,
HI.
Flenley, J.R., Teller, J.T., Prentice, M.E., Jackson, J. and Chew,
C. 1991. The late Quaternary vegetational and climatic
history of Easter Island. Journal of Quaternary Science 6:
85–115.
Fukami, T., Wardle, D.A., Bellingham, P.J., Mulder, C.P.H.,
Towns, D.R., Yeates, G.W., Bonner, K.I., Durrett, M.S.,
Grant-Hoffman, M.N. and Williamson, W.M. 2006. Aboveand below-ground impacts of introduced predators in
seabird-dominated island ecosystems. Ecology Letters 9:
1299–1307.
Green, R.C. and Weisler, M.I. 2000. Mangarevan Archaeology:
Interpretations Using New Data and 40 Year Old Excavations
to Establish a Sequence from 1200 to 1900 AD. University of
Otago Studies in Prehistoric Anthropology, No. 19. University
of Otago, Dunedin.
Green, R.C. and Weisler, M.I. 2002. The Mangarevan sequence
and dating of the geographic expansion into Southeast
Polynesia. Asian Perspectives 41: 213–241.
Green, R.C. and Weisler, M.I. 2004. Prehistoric introduction and
extinction of animals in Mangareva, Southeast Pacific.
Archaeology in Oceania 39: 34–41.
Hiroa, T.R. (P. H. Buck) 1938. Ethnology of Mangareva. Bernice
P. Bishop Museum Bulletin 157. Bishop Museum, Honolulu.
HI.
Hogg, A.G., Hua, Q., Blackwell, P.G., Niu, M., Buck, C.E.,
Guilderson, T.P., Heaton, T.J., Palmer, J.G., Reimer, P.J.,
Reimer, W., Turney, C.S.M. and Zimmerman, S.R.H. 2013.
SH Cal 13 Southern Hemisphere calibration 0–50,000 years
BP. Radiocarbon 55: 1889–1903.
Horrocks, M. 2005. A combined procedure for recovering
phytoliths and starch residues from soils, sedimentary deposits
and similar materials. Journal of Archaeological Science 32:
1169–1175.
Horrocks, M., Baisden, W.T., Nieuwoudt, M.K., Flenley, J., Feek,
D., González Nualart, L., Haoa-Cardinali, S. and
Edmunds Gorman, T. 2012. Microfossils of Polynesian
cultigens in lake sediment cores from Rano Kau, Easter
Island. Journal of Paleolimnology 47: 185–204.
Howard, N. and Kirch, P.V. 2004. Zooarchaeological analysis of
faunal assemblages. In E. Conte and P.V. Kirch (eds),
Archaeological Investigations in the Mangareva Islands,
French Polynesia, pp. 106–121. Contributions of the
Archaeological Research Facility, No. 62. University of
California, Berkeley, CA.
Huguenin, B. 1974. La végétation des îles Gambier – relevé
botanique des espèces introduites. Cahiers du Pacifique No.
18, Vol. II, 459–472. Fondation Singer-Polignac, Paris.
Hunt, T.L. 2007. Rethinking Easter Island’s ecological
catastrophe. Journal of Archaeological Science 34:
485–502.
© 2015 Oceania Publications
Archaeology in Oceania
Kahn, J.G., Sinton, J.M., Mills, P.R. and Lundblad, S.P. 2013.
X-ray fluorescence analysis and intra-island exchange in the
Society Island archipelago (central Eastern Polynesia).
Journal of Archaeological Science 40: 1194–1202.
Kahn, J.G., Nickelsen, C., Stephenson, J., Porch, N.,
Dotte-Sarout, E., Christensen, C.C., Athens, J.S., May, L. and
Kirch, P.V. In press. Late Holocene coastal morphogenesis
and human ecodynamics on Mo’orea, Society Islands (French
Polynesia). The Holocene.
Kirch, P.V. 1984. The Evolution of the Polynesian Chiefdoms.
Cambridge University Press, Cambridge, UK.
Kirch, P.V. 2001. Pigs, humans, and trophic competition on small
Oceanic islands. In A. Anderson and T. Murray (eds),
Australian Archaeologist: Collected Papers in Honour of Jim
Allen, pp. 427–439. Australian National University, Centre for
Archaeological Research and Department of Archaeology and
Natural History, Canberra.
Kirch, P.V. and Conte, E. 2008. Combler une lacune dans la
préhistoire de la Polynésie orientale: Nouvelles données sur
l’archipel des Gambier (Mangareva). Journal de la Société
des Océanistes 128: 91–115.
Kirch, P.V. and Ellison, J. 1994. Palaeoenvironmental evidence
for human colonization of remote Oceanic islands. Antiquity
68: 310–321.
Kirch, P.V., Coil, J., Weisler, M.I., Conte, E. and Anderson, A.J.
2004. Radiocarbon dating and site chronology. In E. Conte
and P.V. Kirch (eds), Archaeological Investigations in the
Mangareva Islands, French Polynesia, pp. 94–105.
Contributions of the Archaeological Research Facility, No. 62.
University of California, Berkeley, CA.
Kirch, P.V., Steadman, D.W., Butler, V.L., Hather, J. and
Weisler, M.I. 1995. Prehistory and human ecology in
Eastern Polynesia: Excavations at Tangatatau rockshelter,
Mangaia, Cook Islands. Archaeology in Oceania 30:
47–65.
Kirch, P.V., Conte, E., Sharp, W. and Nickelsen, C. 2010. The
Onemea Site (Taravai Island, Mangareva) and the human
colonization of Southeastern Polynesia. Archaeology in
Oceania 45: 66–79.
Lacan, F. and Mougin, J.-L. 1974. Les oiseaux de l’archipel des
Gambier. Cahiers du Pacifique No. 18, Vol. II, 533–542.
Fondation Singer-Polignac, Paris.
Lavondès, A. 1971. Poids de pêche polynésiens. Journal de la
Société des Océanistes 33: 341–365.
Leach, F. 1997. A Guide to the Identification of Fish Remains
from New Zealand Archaeological Sites. New Zealand Journal
of Archaeology Special Publications. New Zealand Journal of
Archaeology, Wellington.
Lundblad, S.P., Mills, P.R. and Hon, K. 2008. Analyzing
archaeological basalt using non-destructive energy dispersive
X-ray fluorescence (EDXRF): Effects of post-depositional
chemical weathering and sample size on analytical precision.
Archaeometry 50: 1–11.
Lundblad, S.P., Mills, P.R., Drake-Raue, A. and Kikiloi, S.K.
2010. Non-destructive EDXRF analyses of archaeological
basalts. In M.S. Shackley (ed.), Natural Sciences in
Archaeology Series. X-Ray Fluorescence Spectrometry (XRF)
in Geoarchaeology, pp. 65–80. Springer, New York.
Meyer, J.-Y. and Butaud, J.-F. 2009. The impacts of rats on the
endangered native flora of French Polynesia (Pacific Islands):
Drivers of plant extinction or coup de grace species?
Biological Invasions 11: 1569–1585.
Mieth, A. and Bork, H.-R. 2010. Humans, climate, or introduced
rats – Which is to blame for the woodland destruction on
prehistoric Rapa Nui (Easter Island)? Journal of
Archaeological Science 37: 417–426.
© 2015 Oceania Publications
19
Mizota, C. and Naikatini, A. 2007. Nitrogen isotope composition
of inorganic soil nitrogen and associated vegetation under a
sea bird colony on the Hatana Islands, Rotuma Group, Fiji.
Geochemical Journal 41: 297–301.
Moore, P.D., Webb, J.A. and Collinson, M.E. 1991. Pollen
Analysis, 2nd edn. Blackwell Scientific, London.
Orliac, M. 2003. Un aspect de la flore de Mangareva au XIIème
siècle (archipel Gambier, Polynésie française). In C. Orliac
(ed.), Archéologie en Océanie Insulaire: Peuplement, Société
et Paysages, pp. 150–171. Artcom, Paris.
Pestle, W.J. and Colvard, M. 2012. Bone collagen preservation in
the tropics: A case study from ancient Puerto Rico. Journal of
Archaeological Science 39: 2079–2090.
Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.B.,
Ramsey, C.B., Buck, C.E., Cheng, H., Edwards, R.L.,
Friedrich, M., Grootes, P.M., Guilderson, T., Haflidason, H.,
Hajdas, I., Hatté, C., Heaton, T.J., Hoffmann, D.L., Hogg,
A.G., Hughen, K.A., Kaiser, F.K., Kromer, B., Manning,
W.W., Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M.,
Southon, J.R., Staff, R.A., Turney, C.S.M. and van der Plicht,
J. 2013. IntCal 13 and Marine 13 radiocarbon age calibration
curves 0–50,000 years cal BP. Radiocarbon 55: 1869–1887.
Rolett, B. and Diamond, J. 2004. Environmental predictors of
pre-European deforestation on Pacific islands. Nature 431:
443–446.
Rolett, B.V. 1998. Hanamiai: Prehistoric Colonization and
Cultural Change in the Marquesas Islands (East Polynesia).
Yale University Publications in Anthropology, Vol. 81.
Department of Anthropology and The Peabody Museum, New
Haven, CT.
Rolett, B.V., Conte, E., Pearthree, E. and Sinton, J.M. 1997.
Marquesan voyaging: Archaeometric evidence for inter-island
contact. In M.I. Weisler (ed.), Prehistoric Long Distance
Interaction in Oceania, pp. 134–148. New Zealand
Archaeological Association, Auckland.
Salvat, B. and Rives, C. 1991. Coquillages de Polynésie.
Delachaux et Niestlé, Lausanne. 391 pp.
Sealy, J., Johnson, M., Richards, M. and Nehlich, O. 2014.
Comparison of two methods of extracting bone collagen for
stable carbon and nitrogen isotope analysis: Comparing whole
bone demineralization with gelatinization and ultrafiltration.
Journal of Archaeological Science. 47 (1): 64–69. doi:
10.1016/j.jas.2014.04.011
Sinton, J. and Sinoto, Y. 1997. A geochemical database for
Polynesian adze studies. In M. Weisler (ed.), Prehistoric Long
Distance Interaction in Oceania, pp. 194–204. New Zealand
Archaeological Association, Auckland.
Solem, A. 1976. Endodontoid Land Snails from Pacific Islands
(Mollusca: Pulmonata: Sigmurethra). Part I. Family
Endodontidae. Field Museum of Natural History, Chicago.
Spennemann, D.H.R. 1997. Distribution of rat species (Rattus
spp.) on the atolls of the Marshall Islands: Past and present
dispersal. Atoll Research Bulletin 466: 1–18.
Steadman, D.W. and Justice, L.J. 1998. Prehistoric exploitation
of birds on Mangareva, Gambier Islands, French Polynesia.
Man and Culture in Oceania 14: 81–98.
Steadman, D.W., Vargas, C. and Cristino, F. 1994. Stratigraphy,
chronology, and cultural context of an early faunal assemblage
from Easter Island. Asian Perspectives 33: 79–96.
Suggs, R.C. 1961a. Polynesia. Asian Perspectives 5: 88–94.
Suggs, R.C. 1961b. The Archaeology of Nuku Hiva, Marquesas
Islands, French Polynesia. Anthropological Papers of the
American Museum of Natural History, Vol. 49, Part 1.
American Museum of Natural History, New York.
Théry-Parisot, I., Chabal, L. and Chrzavzez, J. 2010.
Anthracology and taphonomy, from wood gathering to
20
charcoal analysis: A review of the taphonomic processes
modifying charcoal assemblages, in archaeological contexts.
Palaeogeography, Palaeoclimatology, Palaeoecology 291:
142–153.
Thibault, J.-C. and Cibois, A. 2012. From early Polynesian
settlements to the present: Bird extinctions in the Gambier
Islands. Pacific Science 66: 271–281.
Vitousek, P. 2004. Nutrient Cycling and Limitation: Hawai’i
as a Model System. Princeton University Press, Princeton,
NJ.
Waugh, S., Champeau, J., Cranwell, S. and Faulquier, L. 2013.
Seabirds of the Gambier Archipelago, French Polynesia, in
2010. Marine Ornithology 41: 7–12.
Weisler, M.I. 1996. An archaeological survey of Mangareva:
Implications for regional settlement models and interaction
studies. Man and Culture in Oceania 12: 61–85.
Weisler, M.I. 1997. Prehistoric long-distance interaction at the
margins of Oceania. In M.I. Weisler (ed.), Prehistoric Long
Distance Interaction in Oceania, pp. 149–172. New Zealand
Archaeological Association, Auckland.
Weisler, M.I. and Green, R.C. 2001. Holistic approaches to
interaction studies: A Polynesian example. In M. Jones and P.
Sheppard (eds), Australasian Connections and New
Directions: Proceedings of the 7th Australasian Archaeometry
Conference, pp. 413–447. Research Papers in Anthropology
and Linguistics, No. 5. University of Auckland, Auckland.
Weisler, M.I. and Green, R.C. 2013. Fishing strategies in regional
context: An analysis of fish bones from five sites excavated in
1959. Journal of Pacific Archaeology 4: 73–89.
Weisler, M.I., Conte, E. and Kirch, P.V. 2004. Material culture
and geochemical sourcing of basalt artifacts. In E. Conte and
P.V. Kirch (eds), Archaeological Investigations in the
Mangareva Islands, French Polynesia, pp. 128–148.
Contributions of the Archaeological Research Facility, No. 62.
University of California, Berkeley, CA.
Welch, A.J., Wiley, A.E., James, H.F., Ostrom, P.H., Stafford,
T.W. and Fleischer, R.C. 2012. Ancient DNA reveals genetic
stability despite demographic decline: 3,000 years of
population history in the endemic Hawaiian petrel. Molecular
Biology and Evolution 29: 3729–3740.
Wheeler, E., Baas, P. and Gasson, P. 1989. IAWA list of
microscopic features for hardwood identification. IAWA
Journal 10: 219–332.
Human ecodynamics in Mangareva
Whistler, W.A. 2009. Plants of the Canoe People: An
Ethnobotanical Voyage through Polynesia. National Tropical
Botanical Garden, Lawai, Kauaʻi, HI.
Wilmshurst, J.M., Eden, D.E. and Froggatt, P.C. 1999. Late
Holocene forest disturbance in Gisborne, New Zealand: A
comparison of terrestrial and marine pollen records. New
Zealand Journal of Botany 37: 523–540.
Wilson, J. 1799. A Missionary Voyage to the Southern Pacific
Ocean, Performed in the Years 1796, 1797, 1798, in the Ship
Duff. T. Chapman, London.
Worthy, T.W. and Tennyson, A.J.D. 2004. Avifaunal
assemblages from Nenega-iti and Onemea. In E. Conte
and P.V. Kirch (eds), Archaeological Investigations in the
Mangareva Islands (Gambier Archipelago), French
Polynesia, pp. 122–127. Archaeological Research
Facility Contribution No. 62. University of California,
Berkeley, CA.
Young, H.S., McCauley, D.J., Dundar, R.B. and Dirzo, R. 2010.
Plant cause ecosystem nutrient depletion via the interruption
of bird-derived spatial subsidies. Proceedings of the National
Academy of Sciences of the United States of America 107:
2072–2077.
Zimmerman, E.C. 1936. Cryptorrhynchinae of Henderson,
Pitcairn, and Mangareva Islands (Coleoptera, Curculionidae).
Occasional Papers of the Bernice P. Bishop Museum 12 (20):
3–8.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the
online version of this paper on the publisher’s website:
Table S1. Results of Rattus exulans bone collagen δ13C and
δ15N analysis.
Table S2. Basalt adzes from site AGA-3.
Table S3. Stratigraphic distribution of fish remains in site
AGA-3.
Table S4. Stratigraphic distribution of marine molluscs in
site AGA-3.
© 2015 Oceania Publications
Peter Mills