lable at ScienceDirect

Quaternary Science Reviews 126 (2015) 219e226

Contents lists avai

Quaternary Science Reviews

journal homepage: www.elsevier .com/locate/quascirev

Late Holocene vegetation dynamics and deforestation in Rano Aroi:Implications for Easter Island's ecological and cultural history

Valentí Rull a, *, Núria Ca~nellas-Bolt�a a, b, Olga Margalef c, Alberto S�aez b, Sergi Pla-Rabes c,Santiago Giralt a

a Institute of Earth Sciences Jaume Almera (ICTJA-CSIC), C. Lluís Sol�e i Sabarís s/n, 08028 Barcelona, Spainb Department of Stratigraphy, Paleontology and Marine Geosicences, Universitat de Barcelona, C. Martí i Franqu�es s/n, 08028 Barcelona, Spainc Ecological Research Center and Forestry Applications (CREAF), Autonomous University of Barcelona, Campus Bellaterra, Cerdanyola del Vall�es, 08193Barcelona, Spain

a r t i c l e i n f o

Article history:Received 17 August 2015Received in revised form1 September 2015Accepted 3 September 2015Available online xxx

Keywords:PalaeoecologyPalaeoclimatologyPalaeoenvironmentsLast millenniaHoloceneEaster islandRapa NuiPolynesiaPacific Ocean

* Corresponding author.E-mail address: vrull@ictja.csic.es (V. Rull).

http://dx.doi.org/10.1016/j.quascirev.2015.09.0080277-3791/© 2015 Elsevier Ltd. All rights reserved.

a b s t r a c t

Easter Island (Rapa Nui) has been considered an example of how societies can cause their owndestruction through the overexploitation of natural resources. The flagship of this ecocidal paradigm isthe supposed abrupt, island-wide deforestation that occurred about one millennium ago, a few centuriesafter the arrival of Polynesian settlers to the island. Other hypotheses attribute the forest demise todifferent causes such as fruit consumption by rats or aridity but the occurrence of an abrupt, island-widedeforestation during the last millennium has become paradigmatic in Rapa Nui. We argue that such aview can be questioned, as it is based on the palynological study of incomplete records, owing to theexistence of major sedimentary gaps. Here, we present a multiproxy (pollen, charcoal and geochemistry)study of the Aroi core, the first gap-free sedimentary sequence of the last millennia obtained to date inthe island. Our results show changing vegetation patterns under the action of either climatic oranthropogenic drivers, or both, depending on the time interval considered. Palm forests were present inAroi until the 16th century, when deforestation started, coinciding with fire exacerbation elikely ofhuman origine and a dry climate. This is the latest deforestation event recorded so far in the island andtook place roughly a century before European contact. In comparison to other Easter Island records, thisrecord shows that deforestation was neither simultaneous nor proceeded at the same pace over thewhole island. These findings suggest that Easter Island's deforestation was a heterogeneous process inspace and time, and highlights the relevance of local catchment traits in the island's environmental andland management history.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Easter Island (Rapa Nui in indigenous language) is one of themost remote and enigmatic places on Earth. It is famous for amysterious extinct culture that is manifested as anthropomorphicstone statues called moais. The disappearance of this megalithicculture has been associated with a presumed “ecocidal” socio-ecological collapse triggered by the over-exploitation of naturalresources, a scenario that has been presented as a microcosmicmodel of how humans can destroy their own habitat and leadthemselves to societal breakdown (Flenley and Bahn, 2003;

Diamond, 2005). This view was based on the first palynologicalstudies carried out by Flenley and King (1984) and Flenley et al.(1991) and has received extensive media coverage influencingpublic opinion and also the theoretical study of the evolution ofcivilisations. According to this view, the initial human occupation ofEaster Island occurred relatively recently, likely between AD 800and 1000, and the settlers proceeded from western Polynesia(Vargas et al., 2006) (Fi. 1). The presumed socio-ecological collapsewould have occurred shortly after, between AD 1000 and 1200(Mann et al., 2008). Europeans arrived to the island in 1722 (Flenleyand Bahn, 2003). Another view proposes a similar collapse butdifferent causes. According to Hunt (2006, 2007), the deforestationwould have been caused by the Polynesian rat Rattus exulans, car-ried by humans to the island, which ate the fruit thus preventingforest regeneration. The same authors suggest that the cultural

Fig. 1. Location map. A) Easter Island in the Pacific context. B) Sketch-map of the islandshowing the study site (Rano Aroi) and the other localities mentioned in the text (RanoKao and Rano Raraku). Green areas indicate the Eucalyptus forests planted recently.(For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226220

collapse would have been independent from the ecological catas-trophe and proposed a “genocidal” hypothesis, according to whichEuropeans could have been responsible for the cultural demise viaunknown diseases and slave trading. Other catastrophic hypothe-ses suggested a role for climatic changes related to the Little Ice Age(LIA) or the ENSO variability in the island's deforestation and thesubsequent socio-ecological collapse (e.g. McCall, 1993; Hunter-Anderson, 1998; Nunn, 2000; Junk and Claussen, 2011) but mostof them have been dismissed by the lack of empirical support. Aziziand Flenley (2008) argued that Easter Island forests were insensi-tive to climate changes of greater magnitude and duration, such asthe Last Glacial Maximum (LGM), which makes it unlikely thatminor events, such as the LIA or the ENSO, would have caused theforests' demise. However, recent palaeoecological and palae-oclimatic studies have documented several Holocene droughts thatcould have been involved in recent Easter Island's ecologicalchanges, with or without the influence of human activities (Mannet al., 2008; S�aez et al., 2009; Ca~nellas-Bolt�a et al., 2012, 2013;Margalef et al., 2013, 2014).

The recent ecological history of Easter Island is still full of un-certainties that should be addressed by means of new empiricaldata (Rull et al., 2010). These uncertainties have been classified intochronological, ecological and taxonomic (Rull et al., in press). Themain chronological handicaps are the occurrence of a sedimentarygap of multi-millennial magnitude, called here the Critical Sedi-mentary Gap (CSG), and the frequent existence of chronologicalinversions in radiocarbon dates (Butler et al., 2004). From anecological point of view, the actual geographical extent of theassumed ancient palm forests and their taxonomic compositionalso remains elusive (Rull et al., 2010). Taxonomically, the mostrelevant unknown is the identity of the palm that dominated theseforests (Dransfield et al., 1984). From an anthropological perspec-tive, there is also a debate on the potential role of Amerindianpeople in the colonisation and further human occupation of EasterIsland (see Rull et al., 2010, 2013 for a more detailed discussion ofthese topics). The recent finding of a new sedimentary sequence forthe last three millennia that reduced the CSG to a few centuries haschallenged the prevailing catastrophic viewand other paradigmaticideas about Easter Island. The palynological and palaeoclimaticreconstruction of this near-continuous record has suggested that(Ca~nellas-Bolt�a et al. 2013): 1) the first indications of humanpresence, as inferred from the first appearance of the commonweed Verbena litoralis and the onset of fires (as represented by theincrease in charcoal particles) occurred ~450 BC, which is approx-imately 1500 years earlier than expected; 2) the replacement ofpalm-dominated pollen assemblages by grass-dominated pollenassemblages occurred in a gradual, rather than abrupt, fashion overapproximately 2000 years (450 BC to AD 1500); 3) Amerindiancultures might have played a role in the initial settlement of theisland, because V. litoralis is a plant of American origin; 4) lake levelsincreased beginning ca. AD 1200, inferred from changes in diatomcommunities and sedimentary C/N ratio indicating a transition towetter climates; and 5) the small hiatuses detectedmight be due todrought events that prevented peat growth and favoured erosion,affecting vegetation dynamics. These findings should be consideredpreliminary and open to further testing with new empirical data.However, if confirmed, these outcomes have the potential totransform our understanding of the environmental, ecological andhuman history of Easter Island.

Easter Island has three permanent water bodies, two lakes(Rano Raraku and Rano Kao) and a mire (Rano Aroi) (Fig. 1). Mostinferences about Easter Island's palaeoecology and palae-oclimatology come from the Raraku lake, although the Kao lake hasreceived more attention in the last years (Gossen, 2007, 2011;Butler and Flenley, 2010; Horrocks et al., 2102, 2013). The Aroi

mire was included in the first palynological studies but was soondismissed as a useful palaeoecological archive for the last millenniaarguing that recent sediment disturbance by humans in the searchfor water has compromised the completeness and the chronolog-ical coherence of the record (Flenley and King, 1984; Flenley et al.,1991). Before the last millennia, the Aroi palynological recordshowed significant differences from those of Raraku (75 m eleva-tion) and Kao (110 m elevation), where the unidentified palmpollenwas totally dominant (40e80%) suggesting the occurrence ofextensive palm forests in the lowlands. In Aroi (430 m elevation),the pollen diagramwas dominated by grasses (up to 80%) followedby some shrubs, whereas palm pollen was commonly under 10%(Flenley and King, 1984; Flenley et al., 1991). These authors pro-posed that the upper palm forest limit fluctuated around the RanoAroi elevation and this would have determined the observedchanges in composition, with less palms and more shrubs in theforest/grassland ecotone. A short core from Rano Aroi was analyzedfurther by Peteet et al. (2003). It was tentatively dated as repre-senting the end of the Pleistocene (14e10 14C kyr BP), and thepollen data are in agreement with the results of Flenley and King(1984) and Flenley et al. (1991). Recently, the palae-oenvironmental study of Rano Aroi has been revitalised. Indeed,Margalef et al. (2013, 2014) developed multiproxy studies (peatstratigraphy, facies analysis, elemental and isotope geochemistry,XRF, core scanning, pollen and macrofossil analysis) embracing thelast ~70 cal kyr BP and performed a robust palaeoclimatic recon-struction since the Marine Isotope Stage (MIS) 4 to the present. Themain hydrological and environmental changes recorded wereinterpreted in terms of variations in the South Pacific Convergence

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226 221

Zone (SPCZ), Southern Westerlies (SW), storm track and SouthPacific Anticyclone (SPA) locations. These changes resulted to becoherent with the regional climatic variability. The studies byMargalef et al. (2013, 2014) were carried out at a millennial reso-lution and constitute a useful framework to develop further higher-resolution surveys of selected intervals of interest.

This paper reports the results of a multiproxy study of acontinuous (i.e. gap free) Rano Aroi record embracing the lastthree millennia, at multi-decadal resolution. The main aim is toreconstruct vegetation dynamics and to infer the potentialdrivers involved, with emphasis on climatic changes and humanactivities. The three main issues considered in this study(climate, ecology and human impact) are deduced from inde-pendent proxies to avoid circular reasoning. Palaeoclimatictrends are inferred from sedimentology and geochemistry whilepalaeoecological shifts are deduced from pollen analysis, andcharcoal analysis is used to reconstruct changes in fire incidenceas a proxy for human activities. The information obtained willhopefully be useful to address the main unknowns of the EasterIsland's ecological history.

2. Study site

Easter Island (Fig. 1) is located in the eastern Pacific Ocean (27�

070 1600 S, 109� 21’ 59” W) at about 3700 km of the Chilean coastand 2030 km from the nearest oceanic island (Pitcairn). The islandoriginated by the fusion of three volcanic cones (Maunga Terevaka,Kao and Poike), which formed a roughly triangular land mass ofnearly 164 km2 (Fig. 1). In addition to these major volcanoes, theisland is spiked by around 70 other minor satellite cones. The is-land is dominated by the summit of the Terevaka volcano (511 melevation), which has steep slopes at the north and more gentleslopes in the other directions, where most inhabited sites of theisland are located. The climate of Easter Island is subtropical, withan average annual temperature of 21 �C and a range of averagemonthly temperatures between 18 �C in August and 24 �C inJanuary (Mann et al., 2008). The total annual precipitation ishighly variable, ranging between 500 and 1800 mm (average1130 mm); the maximum rainfall occurs from April to June, andthe driest month is October with 70 mm (Azizi and Flenley, 2008).The island is covered by meadows (90%), forests (5%), shrublands(4%), and pioneer and urban vegetation (1%) (Etienne et al., 1982).The meadows are dominated by grasses, mainly Sporobolus indicusand Paspalum scrobiculatum, with Axonopus paschalis as localdominant in the highest sectors of the Maunga Terevaka. The morecommon forests are recent plantations of Eucalyptus spp. (Myrta-ceae) and Dodonaea viscosa (Sapindaceae), both introduced, andthe native Thespesia populnea (Malvaceae). Shrublands are largelydominated by the invader Psidium guajava (Myrtaceae), alsointroduced. Inside the major craters, the only places of the islandwith permanent fresh water, aquatic vegetation is dominated byScirpus californicus (Cyperaceae) and Polygonum acuminatum(Polygonaceae).

Rano Aroi is a mire of ~150 m diameter situated at 430 melevation and located in an ancient Pleistocene volcano crater nearthe highest summit of the island (Fig.1).Water level is controlled bygroundwater inputs subjected to the influence of seasonal varia-tions in precipitation, and human extraction, through the con-struction of an artificial outlet in the 1960s (Herrera and Custodio,2008). The vegetation of the mire is characterized by Scirpus cal-ifornicus, P. acuminatum and the ferns Asplenium polydon, Vittariaelongata and Cyclosorus interruptus; the surrounding area iscovered by grasslands and a small Eucalyptus forest planted duringthe 1960s (Zizka, 1991).

3. Methods

3.1. Coring

The core studied (ARO 08-02) was obtained in 2008 with aRussian corer of 5.5 cm diameter and a 50-cm long chamber. Thecore was retrieved after 8 drives for a total depth of 400 cm. Theupper 75 cm were water, hence, the total peat depth was 325 cm.The cores were sealed, transported to the laboratory and stored at4 �C. This paper concentrates on the uppermost ~2 m of the peatsection representing approximately the last three millennia.

3.2. Dating

Samples for AMS radiocarbon dating consisted of pollen extracts(Vandergoes and Prior, 2003). Dating was performed at BETA An-alytic (USA) and the Poznan Radiocarbon Laboratory (Poland). Theobtained ages were calibrated with CALIB 7.0 using the SHcal13curve (Hogg et al., 2013). The age-depth model was constructedwith CLAM (Blaauw, 2010; Blaauw and Heegaard, 2012). Consid-ering the statistical errors of this age-depth model, dates areexpressed as the mid-point estimation (median) plus/minus the95% confidence interval, using the BC/AD scale. Dates from otherspapers used in the discussion section are expressed in the sameformat as their original references.

3.3. Geochemistry

The core was sampled every 5 cm to determine total carbon,total nitrogen and their stable isotopes. Samples were dried at 60 �Cfor two days, frozen with liquid nitrogen and ground in a ring mill.2.5 mg of the powdered sample was weighed using a microbalanceand packed in tin (Sn) capsules. The concentration and isotopicmeasurements were carried out using a Finnigan Delta PlusEAeCFeIRMS spectrometer.

3.4. Palynology and charcoal analysis

Samples for palynological analysis consisted of 2e4 g of wetsediment. Chemical processing included KOH, HCl and HF di-gestions, and acetolysis, after spiking with Lycopodium tablets(Bennett and Willis, 2001). Pollen counting was conducted until aminimum of 200 pollen grains and the saturation of taxa diversity(Rull, 1987). The pollen sum used to calculate percentages consistedof only pollen grains, excluding those corresponding to aquatic andsemi-aquatic plants. Diagrams were plotted with psimpoll 4.27 andzonation was performed using optimal splitting by informationcontent (Bennett, 1996). Charcoal counts were carried out in thesame samples used for pollen analysis. Charcoal particles below5 mmof maximum length were not considered as a number of themcould be the result of the breakage of larger particles.

4. Results and interpretation

4.1. Chronology

The establishment of a sound age-depth model was a criticalstep in this research. Table 1 shows the details of radiocarbondating and the resulting age model is depicted in Fig. 2. The modelshows a continuous sedimentary trend with no gaps and age in-versions. According to thismodel, the analyzed core contains ~2600years, ranging between approximately 750 BC and AD 1810. Theaverage accumulation rate is 0.7 mm y�1.

Table 1AMS radiocarbon dating: sample details, results and calibration. All samples con-sisted of pollen extracts (Vandergoes and Prior, 2003). Calibration was performedwith Calib 7.1 using the SHcal13 database (Hogg et al., 2013).

Sample Lab code Depth (cm) 14C yr BP Cal yr BP (2s)

ARO 08-02 02-25/26 Poz-45907 76 190 ± 30 134e285ARO 08-02 02-35 Poz-31998 85 270 ± 30 272e325ARO 08-02 03 9/10 Beta-397173 110 760 ± 30 635e690ARO 08-02 03-25 Poz-31999 125 1050 ± 30 898e963ARO 08-02 04-3/4 Beta-397174 154 1730 ± 30 1533e1629ARO 08-02 04-25 Poz-32000 175 2265 ± 30 2155e2329ARO 08-02 04-34 Beta-197175 184 2550 ± 30 2463e2742

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226222

4.2. Vegetation dynamics and fire

The pollen diagram is dominated by palms (Arecaceae), com-posites (Asteraceae) and grasses (Poaceae) (Fig. 3). There is a con-spicuous difference between the lower part of the diagram, wheremost pollen types remain at similar percentages, and the upper-most section, where all types exhibit fluctuations. To facilitateinterpretation, the diagram was subdivided into four phases.

Phase I (750 ± 130 BC to AD 1250 ± 60) shows a remarkableconstancy in pollen assemblages and, therefore, in vegetation pat-terns through time. The dominant pollen types are palms andcomposites, both with values of approximately 20e40%, suggestingthe occurrence of a palm woodland with compositae shrubs in theunderstory. Grasses also showed constant and persistent values(50e80%) suggesting that grasslands were important, likelydominant, regionally. Plants typical of the present-day local mirevegetation (Cyperaceae, Polygonum and ferns) were almostcompletely absent. Within this uniformity, a recurrent patternemerges in the form of two sequences of palm-Asteraceae-grasspeaks likely representing processes of landscape opening frompalm-dominated to grassland-dominated landscapes. These se-quences lasted several centuries each and occurred between ca.300 ± 80 BC and AD 50 ± 70 (LO1) and between ca. AD 600 ± 65 andAD 1100 ± 60 (LO2), respectively.

Fig. 2. Age-depth model using the best-fit option (linear interpolation). Two temporalscales (Cal years BP and AD/BC) are used to facilitate correlations. The solid line rep-resents the mid-point age estimate (median) and the gray band is the 95% confidenceinterval for interpolated ages.

Phase II (AD 1250 ± 60 to AD 1520 ± 85) began with a con-spicuous increase in Sophora (up to 20%), suggesting that this shrubstarted to be important in the understory of the palm woodland tothe detriment of Asteraceae, which underwent a significant decline.Sophora peaked around AD 1340 ± 70 and disappeared at AD1490 ± 80, coinciding with a significant increase in palms. Thepollen assemblage of this phase is suggestive of a progressivedensification of palm woodlands to more closed forests at theexpense of the understory shrubs, a situation that reached amaximum at AD 1520 ± 85. The elements of the local mire vege-tation also underwent a remarkable increase coinciding withwoodland densification.

Phase III (AD 1520 ± 85 to AD 1620 ± 110) was characterized bythe abrupt decline and further disappearance of palm and Asteraceaepollen, suggesting rapid deforestation coinciding with the equallyabrupt decline of mire vegetation and the maximum amount ofgrasses (90%), which is interpreted as the establishment of presentlandscapes dominated by grasslands and devoid of the former palmforests. The palynological expression of this deforestation (LO3) issimilar to the former phases of landscape opening (LO1 and LO2) butis significantly faster. It is noteworthy that the intermediate Aster-aceae peak is lacking, suggesting the synchronous elimination of thecanopy trees and the understory shrubs. The coincidence of thisforest demise with a conspicuous increase in charcoal suggests thatfire likely played a leading role in deforestation.

Present-day landscapes, totally dominated by grasslandswithout any representatives of the former woodland vegetation,were established during phase IV (AD 1620 ± 110 to AD 1810 ± 130).A peak in fire incidencewas recorded at AD 1660 ± 130, followed bya rapid decline until the end of the diagram.

4.3. Climatic/environmental trends

The more significant variations in geochemical proxies were twoshifts coinciding with the pollen sequences of landscape openingLO1 and LO2. In both sequences, a consistent pattern of C/N decline,related mainly to an increase in nitrogen, coupled with increases ind15N and d13C, is manifest (Fig. 4). A decline in the C/N ratio and anincrease in d15N are compatible with lower water levels, stagnation,and eventually drying, along with changes in the peat-formingvegetation. In these anoxic conditions, when the nitrogen supply ishigh, denitrification processes are favored and denitrifying bacteriaselect the more labile 14N, which is lost via degasification, thuscausing 15N enrichment (Houlton and Bai, 2009; McLauchlan et al.,2013). The increase in d13C could be due, in part, to the general in-crease of C4 grasses (�16 to �10‰) to the detriment of palms andother C3 plants and algae (�33 to �24‰) (O'Leary, 1988). Patterns ofvariation inpollen and geochemistry are consistentwith dry climatesduring LO1 and LO2, as manifested in low water levels and theestablishment of open vegetation at the expense of palmwoodlands.Wetter climates would have returned during Phase II, as suggestedby d13C values around �25‰, indicating the local dominance of C3plants, including palms, ferns and aquatic species, and d15N valuesclose to zero, which is compatible with increased nitrogen fixationby cyanobacteria due to higher water levels. During Phase IV,geochemical proxies show a trend similar to the lower half of LO1and LO2, suggesting the initiation of a third shift to a drier climate.

5. Discussion and conclusions

5.1. Vegetation and climate prior to polynesian settlement(750 ± 130 BC to AD 1250 ± 60)

The constancy of vegetation documented in Phase I coincideswith the results of previous palynological analyses, which extended

Fig. 3. Percentage pollen diagram of the core studied. Charcoal is expressed in concentration units (particles per gram of sediment). Solid lines express x10 exaggeration. LO e

Landscape Opening events.

Fig. 4. Geochemical parameters measured at the core studied. Smoothing was carried out by LOESS (Local Regression Scatterplot Smoothing). The pollen phases (I to IV) and theLandscape Opening events (LO1 to LO3) are indicated.

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226 223

this constancy to approximately 30,000 years BP (Flenley and King,1984; Flenley et al., 1991; Azizi and Flenley, 2008). The main dif-ference is that in these previous works, Rano Aroi was considered tohold ecotonal vegetation corresponding to the upper forestboundary. We believe that we do not have any conclusive evidence

for this and interpret our pollen assemblages as indicative of a palmwoodland with a shrubby understory dominated by compositae, asituation that is common in other Pacific islands (Vargas et al.,2011). According to regional paleoclimatic reconstructions ofEastern Polynesia (Nunn, 2007), climates during Phase I werewarm

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226224

and dry, characterized by severe droughts, reduced storminess andlow interannual variability (Fig. 5). Based on vegetation shifts, LO1and LO2 are strong candidates for drier intervals, and this inter-pretation is supported by biogeochemical proxies. Sequence LO1occurred before the Preclassic Maya Collapse of Central America, aperiod when the Pacific sea level was 1e2 m higher than the pre-sent. Since that time, sea level has fallen gradually, but it remainedhigher than at present until the end of Phase I. Sequence LO2 wascoeval with the Classic Maya Collapse (~AD 900) attributed to anumber of prolonged droughts that contributed to the final demiseof this civilization (Haugh et al., 2003). In the Pacific islands, theLO2 period was characterized by sea levels higher than today andincreased food production thanks to the development of irrigationpractices and terracing (Nunn, 2007).

5.2. Post-settlement, climatic crisis and coastal abandonment (AD1250 ± 60 to AD 1620 ± 110)

According to archaeological evidence, Polynesian colonizersarrived at Easter Island at the end of Phase I or the beginning ofPhase II, although the dates of arrival range from AD 800 to AD1300, depending on the authors (Flenley and Bahn, 2003; Hunt andLipo, 2006; Vargas et al., 2006; Wilmshurst et al., 2011). This was aperiod of intensified long-distance navigation and new settlements,especially in the eastern Pacific, where many islands and archi-pelagos were likely colonized by Polynesians during this time(Fig. 5) (Nunn, 2007). At Easter Island, the end of Phase I was a timeof societal prosperity and open coastal settlement, known as AhuMoai, when the construction of the emblematic moais reached itsmaximum (Flenley and Bahn, 2003). However, Phase II witnessed adifferent scenario. Our interpretation of wetter climates duringPhase II also coincides with regional trends. The onset of this phasewas coeval with a regional lowstand phase and a change in theclimate to a cooler and wetter climate with increased storminessdue to ENSO intensification, which has been described as the “1300

Fig. 5. Correlations between the results obtained in this paper and the regional reconstrucrecords that have been documented for Easter Island are also included. RWP e Roman WarmAge. The boundary between the DACP and the MWP is indicated by a dashed line because itshas been located at AD 950 while in the Pacific Basin, the beginning of the MWP has been

event” (Nunn, 2007). During the second part of Phase II, the onset ofpalm forest densification at Aroi coincided with another regionalclimate shift to the cold and dry Little Ice Age (LIA) climate. The sealevel remained lower than today and inter-island navigation wassignificantly reduced. Archaeological reconstructions of Easter Is-land show that coastal settlements began to be abandoned andreplaced by inland and upland habitats (Nunn, 2007). Societalconflicts started to be frequent, but the Ahu moai period continuedduring the first part of Phase II.

5.3. Deforestation and European contact (AD 1620 ± 110 to AD1810 ± 130)

The onset of deforestation (Phase III) coincided with thebeginning of the Huri moai phase, characterized by moai toppling,societal conflict and cave settlement. The significant role of fire indeforestation suggests that humans could have been involved.Drier climates could have facilitated the initiation and spread of fireby increasing the ignitability and vegetation flammability, but cli-matic dryness alone is unlikely to have been the cause for defor-estation, as former dry shifts (LO1 and LO2) were not sufficient toinitiate fires and/or remove forests. This is the most recent well-dated deforestation event recorded so far at Easter Island, and itstarted between AD 1435 and AD 1605 (considering the 95% con-fidence interval of the age-depth model), more than a centurybefore European contact (AD 1722). Therefore, if humans wereinvolved in deforestation by fire, Rapanuis are the most likelycandidates, as Europeans had not yet arrived at the island. It isnoteworthy that the Aroi forests were in place six centuries afterthe dates proposed by the ecocidal theory for island deforestation,which agrees with the predictions of the genocidal view (Rainbird,2002; Peiser, 2005). However, as discussed below, the Aroi defor-estation took place well before European contact. Using the char-coal record as a proxy for human activity in the Aroi catchment, wenote that intense landscape disturbance did not began until AD

tions for the Eastern Pacific during the last millennium. The well dated deforestationPeriod, DACP e Dark Ages Cold period, MWP e Medieval Warm Period, LIA e Little Ice

chronological uncertainty depends on the region. For example, in Europe, this boundarydated to AD 750. EI e Easter Island, C e Classic, PC e Preclassic.

Table 2Temporal differences between the deforestation events recorded at Easter Island using pollen analysis. Average deforestation rates have been estimated indirectly as thedecline of palm pollen percentage per century (%C�1). Less reliable estimates are bracketed.

Deforestation event Onset End Time (y) Rates (%C�1) References

Rano Aroi AD1520 AD1620 100 73 This paperRano Raraku 450BC AD1530 1980 7 Ca~nellas-Bolt�a et al. (2013)Rano Kao-K1 (AD50) (AD100) (50) (160) Butler and Flenley (2010)Rano Kao-K2 (AD1350) (AD1800) (450) (18) Butler and Flenley (2010)

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226 225

1520 ± 85, peaked at AD 1660 ± 130 and then declined, ending byAD 1810 ± 130 (Phase IV). This is consistent with recent archaeo-logical studies showing that the intensity of land use decreasedsubstantially in some areas of the island, especially in the uplands,slightly before European contact (Stevenson and Haoa, 2008;Stevenson et al., 2015). The age range of maximum fire incidence(AD 1660 ± 130, or AD 1530 to AD 1790) overlaps with Europeancontact (AD 1722), which prevents the dismissal of a link betweenfire and post-contact human activities during Phase IV. However, asnoted above, deforestation of the Aroi catchment seems to havebeen already completed before the arrival of the first Europeans.

5.4. Comparison with other deforestation records

The Aroi scenario cannot be extrapolated to the whole island,especially to coastal lowlands. In these lowlands, the situation wasradically different. In addition to the core used in this study (ARO08-02), the only well dated and fairly continuous sequence of thelast millennia with a comparable resolution is RAR 08, from thelowland crater lake Rano Raraku (75 m elevation) (Fig. 1). This re-cord shows that, at Raraku, deforestation began at ~450 BC andprogressed in a gradual fashion for more than two millennia untilits end at ~AD 1530 (Ca~nellas-Bolt�a et al., 2013). As in Aroi, firesseem to have played a key role in this process. Human presencewasdeduced from the occurrence of the weed V. litoralis, the firstappearance of which coincided with the first fires and the onset ofdeforestation. Therefore, deforestation began ~2000 years beforeand ended more than 100 years sooner in Raraku than in Aroi(Table 2). Regarding rates, deforestation was abrupt at Aroi (<100years) and gradual at Raraku, where it lastedmore than 2000 years.Comparatively, deforestation proceeded ~14 times faster in RanoAroi. Another lowland crater lake with palynological records isRano Kao (110 m elevation) (Fig. 1), but the occurrence of frequentage inversions in its sediments has prevented the attainment of asequence comparable to Aroi or Raraku for the last millennia(Horrocks et al., 2012). However, two deforestation events associ-ated with increases in charcoal have been reported in the Kao re-cord; the first (K1) between ~AD50 and ~AD100 and the second(K2) between ~AD 1350 and ~AD 1800 (Butler and Flenley, 2010).Because of dating issues, it is difficult to estimate the rates at whichthese events occurred, but K1 was likely more abrupt than K2(Table 2). Remarkably, Rano Kao palm forests totally recovered afterthe K1 deforestation, which is the only case of forest regenerationafter fire recorded so far on the island during the last millennia. Aconstant in all these deforestations is that they are accompanied byincreases in charcoal, indicating fire exacerbation and, likely, hu-man disturbance.

The comparison of the palynological records of Aroi, Raraku andKao clearly documents the heterogeneous spatio-temporal defor-estation patterns across Easter Island during the last two millennia.This suggests the occurrence of heterogeneous patterns of envi-ronmental change and/or local management by humans, which isin agreement with previous archaeological evidence (Stevensonand Haoa, 2008; Stevenson et al., 2015). It follows that paleoeco-logical evidence from Easter Island should be analyzed under a

mostly local perspective, likely restricted to the lake/mire catch-ment, and generalizations should be based on an overall consid-eration of all the available evidence over the whole island, ratherthan on the island-wide extrapolation of local observations (Rullet al., 2010). Local factors such as rainfall, elevation, area, latitude,age and isolation have been found to be relevant for understandingdeforestation patterns across the whole assemblage of Pacificislands (Rolett and Diamond, 2004). Intra-island environmentaland cultural variability, however, has not been sufficientlyaddressed but it may be relevant, even in an island as small asEaster Island, where the Aroi, Raraku and Kao catchments areremarkably heterogeneous in terms of elevation, size, topography,hydrology, microclimate, soils, vegetation and human activity(Stevenson and Haoa, 2008; Rull et al., 2010; Stevenson et al., 2015).Accepting such heterogeneity would help reconcile accurate butapparently inconsistent empirical observations instead of havingthem used to erect contradictory hypotheses on the tempo andmode of Easter Island deforestation, particularly those which as-sume one single abrupt and island-wide deforestation event as apremise.

Acknowledgements

This research was funded by the Spanish Ministry of Educationand Science projects LAVOLTER (CGL2004-00683) and GEOBILA(CGL2007-60932/BTE). Fieldwork permits were facilitated byCONAF (Corporaci�on Nacional Forestal, Chile).

References

Azizi, G., Flenley, J.R., 2008. The last glacial maximum climatic conditions on EasterIsland. Quat. Int. 184, 166e176.

Bennett, K.D., 1996. Determination of the number of zones in a biostratigraphicalsequence. New Phytol. 132, 155e170.

Bennett, K.D., Willis, K.J., 2001. Pollen. In: Birks, H.J.B., Last, W.M., Smol, J.P. (Eds.),Tracking Environmental Change Using Lake Sediments, Terrestrial Algal andSiliceous Indicators, vol. 3. Kluwer, Dordrecht, pp. 5e32.

Blaauw, M., 2010. Methods and code for ‘classical’ age-modeling of radiocarbonsequences. Quat. Geochronol. 5, 512e518.

Blaauw, M., Heegaard, E., 2012. Estimation of age-depth relationships. In:Birks, H.J.B., Juggins, S., Lotter, A., Smol, J.P. (Eds.), Tracking EnvironmentalChange Using Lake Sediments, Data Handling and Statistical Techniques, vol. 5,pp. 379e413.

Butler, K.R., Flenley, J.R., 2010. The Rano Kau 2 pollen diagram: palaeoecologyrevealed. Rapa Nui J. 24, 5e10.

Butler, K., Prior, C.A., Flenley, J.R., 2004. Anomalous radiocarbon dates from EasterIsland. Radiocarbon 46, 395e405.

Ca~nellas-Bolt�a, N., Rull, V., S�aez, A., Margalef, O., Giralt, S., Pueyo, J.J., Birks, H.H.,Birks, H.J.B., Pla-Rabes, S., 2012. Macrofossils in Raraku Lake (Easter Island)integrated with sedimentary and geochemical records: towards a paleoeco-logical synthesis. Quat. Sci. Rev. 34, 113e126.

Ca~nellas-Bolt�a, N., Rull, V., S�aez, A., Margalef, O., Bao, R., Pla-Rabes, S., Blaauw, M.,Valero-Garc�es, B., Giralt, S., 2013. Vegetation changes and human settlement ofEaster Island during the last millennia: a multiproxy study of the Lake Rarakusediments. Quat. Sci. Rev. 72, 36e48.

Diamond, J., 2005. Collapse. How Societies Choose to Fail or Survive. Allen Lane,London.

Dransfield, J., Flenley, J.R., King, S.M., Harkness, D.D., Rapu, S., 1984. A recentlyextinct palm from Easter Island. Nature 312, 750e752.

Etienne, M., Michea, G., Díaz, E., 1982. Flora, vegetaci�on y potential pastoral de la Islade Pascua. Boletín T�ecnico n� 47. Universidad de Chile, Facultad de CienciasAgrarias, Veterinarias y Forestales, Santiago.

Flenley, J.R., Bahn, P.G., 2003. The Enigmas of Easter Island. Oxford University Press,

V. Rull et al. / Quaternary Science Reviews 126 (2015) 219e226226

Oxford.Flenley, J.R., King, S.M., 1984. Late quaternary pollen records from Easter Island.

Nature 307, 47e50.Flenley, J.R., King, S.M., Jackson, J., Chew, C., Teller, J.T., Prentice, M.E., 1991. The late

quaternary vegetational and climatic history of Easter Island. J. Quat. Sci. 6,85e115.

Gossen, C., 2007. The mystery lies in the Scirpus. Rapa Nui J. 21, 105e110.Gossen, C., 2011. Deforestation, Drought and Humans: New Discoveries of the Late

Quaternary Paleoenvironment of Rapa Nui (Easter Island). PhD dissertation.Portland State University, USA.

Haug, G.H., Günther, D., Peterson, L.C., Sigman, D.M., Hughen, K.A., Aeschlimann, B.,2003. Climate and the collapse of Maya civilization. Science 299, 1731e1735.

Herrera, C., Custodio, E., 2008. Conceptual hydrogeological model of volcanic EasterIsland (Chile) after chemical and isotopic surveys. Hydrogeol. J. 16, 1329e1348.

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, R.W., Turney, C.S.M., Zimmerman, S.R.H., 2013.SHCal13 southern hemisphere calibration, 0-50,000 Years cal BP. Radiocarbon55, 1889e1903.

Horrocks, M., Baisden, W.T., Nieuwoudt, M.K., Flenley, J., Feek, D., Gonz�alezNualart, L., Haoa- Cardinali, S., Edmunds Gorman, T., 2012. Microfossils ofPolynesian cultigens in lake sediments cores from Rano Kau, Easter Island.J. Paleolimnol. 47, 185e204.

Horrocks, M., Marra, M., Baisden, W.T., Flenley, J., Feek, D., Gonz�alez, L., Haoa-Caardinali, S., Gorman, T.E., 2013. Pollen, phytoliths, arthropods and high-resolution 14C sampling from Rano Kau, Easter Island: evidence for late qua-ternary environments, ant (Formicidae) distributions and human activity.J. Paleolimnol. 50, 417e432.

Houlton, B.Z., Bai, E., 2009. Imprint of denitrifying bacteria on the global terrestrialbiosphere. Proc. Natl. Acad. Sci. U. S. A. 106, 21713e21716.

Hunt, T.L., 2006. Rethinking the fall of Easter Island. Am. Sci. 94, 412e419.Hunt, T.L., 2007. Rethinking Easter Island's ecological catastrophe. J. Archaeol. Sci.

34, 485e502.Hunt, T.L., Lipo, C.P., 2006. Late colonization of easter island. Science 311,

1603e1606.Hunter-Anderson, R.L., 1998. Human vs. climatic impacts at Rapa Nui, did people

really cut down all those trees? Easter Island in Pacific context, South SeasSymposium. In: Stevenson, C.M., Lee, G., Morin, F.J. (Eds.), Proceedings of theFourth International Conference on Easter Island and East Polynesia, Albu-rquerque, USA, pp. 95e99.

Junk, C., Claussen, M., 2011. Simulated climate variability in the region of Rapa Nuiduring the last millennium. Clim. Past 7, 381e395.

Mann, D., Edwards, J., Chase, J., Beck, W., Reanier, R., Mass, M., Finney, B., Loret, J.,2008. Drought, vegetation change, and human history on Rapa Nui (Isla dePascua, Easter Island). Quat. Res. 69, 16e28.

Margalef, O., Ca~nellas-Bolt�a, N., Pla-Rabes, S., Giralt, S., Pueyo, J.J., Joosten, H.,Rull, V., Buchaca, T., Hern�andez, A., Valero-Garc�es, B.L., Moreno, A., S�aez, A.,2013. A 70,000 year geochemical and palaeoecological record of climatic andenvironmental change from Rano Aroi peatland (Easter Island). Glob. PlanetChange 108, 72e84.

Margalef, O., Martínez-Cortizas, A., Kylander, M., Pla-Rabes, S., Ca~nellas-Bolt�a, N.,Pueyo, J.J., S�aez, A., Valero-Garc�es, B., Giralt, S., 2014. Environmental processesin Rano Aroi (Easter Island) peat geochemistry forced by climate variability

during the last 70 kyr. Palaeogeogr. Palaeoclimatol. Palaeoecol. 414, 438e450.McCall, G., 1993. Little ice age, some speculations for Rapanui. Rapa Nui J. 7, 65e70.McLauchlan, K.K., Williams, J.J., Craine, J.M., Jeffers, E.S., 2013. Changes in nitrogen

cycling during the Holocene epoch. Nature 495, 352e355.Nunn, P.D., 2000. Environmental catastrophe in the Pacific islands around A.D. 1300.

Geoarchaeology 15, 715e740.Nunn, P.D., 2007. Climate, Environment and Society in the Pacific During the Last

Millennium. Elsevier, Amsterdam.O'Leary, M.H., 1988. Carbon isotopes in photosynthesis. BioScience 38, 328e336.Peiser, B., 2005. From genocide to ecocide, the rape of Rapa Nui. Energ Environ. 16,

513e539.Peteet, D., Beck, W., Ortiz, J., O'Connell, S., Kurdyla, D., Mann, D., 2003. Rapid

vegetational and sediment change from rano aroi crater, easter island. Easterisland. In: Loret, J., Tanacredi, J.T. (Eds.), Easter Island. Scientific Exploration intothe World's Environmental Problems in Microcosm. Kluwer Academic/PlenumPubl., New York, pp. 81e92.

Rainbird, P., 2002. A message for our future? The Rapa Nui ecodisaster and Pacificisland environments. World Archaeol. 33, 436e451.

Rull, V., 1987. A note on pollen counting in paeoecology. Pollen Spores 29, 471e480.Rull, V., Ca~nellas-Bolt�a, N., S�aez, A., Giralt, S., Pla, S., Margalef, O., 2010. Paleoecology

of Easter Island: evidence and uncertainties. Earth Sci. Rev. 99, 50e60.Rull, V., Ca~nellas-Bolt�a, N., S�aez, A., Margalef, O., Bao, R., Pla-Rabes, S., Valero-

Garc�es, B., Giralt, S., 2013. Challenging Easter Island's collapse: the need forarchaeological-paleoecological synergies. Front. Ecol. Evol. 1 http://dx.doi.org/10.3389/fevo.2013.00003.

Rull, V., Ca~nellas-Bolt�a, N., S�aez, A., Giralt, S., Pla-Rabes, S., Margalef, O., 2015.Paleoecology and human history of Easter Island (southern Pacific Ocean):some problems and potential solutions. In: Proc. Workshop “Palaeoenviron-ment and the Development of Early Societies”. German Archaeological Institute(in press).

S�aez, A., Valero-Garc�es, B., Giralt, S., Moreno, A., Bao, R., Pueyo, J.J., Hern�andez, A.,Casas, D., 2009. Glacial to holocene climate changes in the SE Pacific. The RarakuLake sedimentary record (Easter island, 27�S). Quat. Sci. Rev. 28, 2743e2759.

Stevenson, C.M., Haoa, S., 2008. Prehistoric Rapa Nui: Landscape and SettlementArchaeology at Hanga Ho’onu. Easter Island Foundation, Los Osos.

Stevenson, C.M., Puleston, C.O., Vitousek, P.M., Chadwick, O.A., Haoa, S.,Ladefoged, T.N., 2015. Variation in Rapa Nui (Easter Island) land use indicatesproduction and population peaks prior to European contact. Proc. Natl. Acad.Sci. U. S. A. 112, 1025e1030.

Vandergoes, M.J., Prior, C.A., 2003. AMS dating of pollen concentratesda method-ological study of late quaternary sediments from South Westland, New Zealand.Radiocarbon 45, 479e491.

Vargas, P., Cristino, C., Izaurieta, R., 2006. 1000 a~nos en Rapa Nui. Arqueología delasentamiento. Universitaria, Santiago.

Vargas, R., Reif, A., Faúndez, M.J., 2011. The forests of Robinson Crusoe Island, Chile:an endemism hotspot in danger. Bosque 32, 155e164.

Wilmshurst, J.M., Hunt, T.L., Lipo, C.P., Anderson, A.J., 2011. High-precision radio-carbon dating shows recent and rapid initial human colonization of East Pol-ynesia. Proc. Natl. Acad. Sci. U. S. A. 108, 1815e1820.

Zizka, G., 1991. Flowering plants of easter island. Palmarum Hortus FrancofurtensisSci. Rep. 3, 1e108.