volcanism and historical ecology on the willaumez

507

Volcanism and Historical Ecology on the Willaumez Peninsula,Papua New Guinea1

Robin Torrence,2 Vince Neall,3 and W. E. Boyd4

Abstract: The role of natural disasters has been largely overlooked in studies ofSouth Pacific historical ecology. To highlight the importance of rapid-onsetnatural hazards, we focus on the contributions of volcanism in shaping land-scape histories. Results of long-term research in the Willaumez Peninsula onNew Britain in Papua New Guinea illustrate the wide range and complexity ofpotential relationships between volcanic activity and human responses. Despitefrequent severe volcanic impacts, human groups have responded creatively tothese challenges and over time may have developed particular strategies thatcoped with the demands of repeated refuging and recolonization.

In the pioneering book on historical ecol-ogy in the Pacific region, Matthew Spriggsrightly identified the key determinants oflandscape evolution as (1) climate change, (2)natural disasters, and (3) human agency. Likethe majority of the other contributions, hebarely mentioned the second category, whichincluded ‘‘catastrophic events or sequencesof events, such as volcanic eruptions, earth-quakes or hurricanes’’ (Spriggs 1997:80). In-stead, the majority of Pacific scholars havefocused on natural and cultural processesthat played out gradually over relatively longperiods of time. The lack of attention torapid-onset natural hazards is a serious flawin our understanding of historical ecology inthis area because their sudden and violent im-pacts pose a special challenge for human soci-

eties (cf. Cronin et al. 2008:2192–2193).Recent scholarship in archaeology, however,well reflects a growing awareness resultingfrom modern disasters (e.g., Grattan andTorrence 2007:1, Cashman and Giordano2008). Our paper redresses the imbalance ofresearch efforts in Pacific historical ecologyby highlighting the relative roles of naturaland anthropogenic factors in the formationof landscape histories played out in settingssubject to natural disasters. To better illus-trate the general processes, we provide con-crete cases based on our interdisciplinaryresearch on the Willaumez Peninsula in Pa-pua New Guinea.

Although a range of natural disasters maycomprise important elements of local his-torical ecology, we focus here primarily onvolcanic activity because its role has been no-ticeably absent in previous studies of SouthPacific historical ecology. This omission issurprising because volcanism has been an im-portant element in the formation of these en-vironments. In many cases it may also haveseriously restricted the degree to which hu-man agency could shape landscape histories.Many prehistoric human groups experiencedexplosive eruptions, as dramatically illustratedby the burial of well-known South Pacificarchaeological sites under volcanic deposits(e.g., Garanger 1972, Specht et al. 1988,Cronin and Neall 2000, Anson et al. 2005,Bedford et al. 2006). Cultural landscapes im-pacted by repeated volcanic hazards sharegeneral characteristics that set them apart

Pacific Science (2009), vol. 63, no. 4:507–535: 2009 by University of Hawai‘i PressAll rights reserved

1 Funding was provided by Australian ResearchCouncil, Australian Museum, Earthwatch Institute, Aus-tralia and Pacific Foundation, Pacific Biological Founda-tion, New Britain Palm Oil, Ltd. Manuscript accepted 15January 2009.

2 Australian Museum, 6 College Street, Sydney, NewSouth Wales 2010, Australia.

3 Institute of Natural Resources, Massey University,Private Bag 11 222, Palmerston North, New Zealand.

4 School of Environmental Science and Management,Southern Cross University, Lismore, New South Wales2480, Australia.

from more stable settings and those that ex-perience slow climatic change. Extreme levelsof selection operate in these ‘‘catastrophicenvironments,’’ defined by Torrence andDoelman (2007:43) as those that experience‘‘frequent, very severe environmental per-turbations,’’ each of which is serious enoughto cause local extinctions (cf. Hoffmann andParsons 1997:23, Turner and Dale 1998).In these situations human societies are un-able to modify their behavior and remain inplace permanently. Instead, the archaeologi-cal record of catastrophic volcanic settingsis characterized by cycles of abandonmentand (re)colonization (e.g., Sheets et al. 1991,Sheets and McKee 1994, Siebe et al. 1996,Sheets 1999, 2007, Machida and Sugiyama2002, Mastrolorenzo et al. 2002, 2006, Zei-dler and Isaacson 2003, Gaillard et al. 2007).In addition, the repeated occurrence of dis-asters often disrupts natural processes ofsuccession so that the environment is perma-nently maintained in a relatively disturbedand immature state, a condition that in turnmay yield special opportunities for colonizingpopulations. Despite the devastating conse-quences of volcanic activity, the fertile soilsand useful raw materials (e.g., obsidian) pro-vide strong inducements for human settle-ment.

Although settings characterized by highrates of volcanism comprise a substantial riskfor cultural groups, recent scholarship hasshifted from focusing solely on the environ-mental forcing agent to a consideration ofthe vulnerability of societies, which in turn islinked to factors such as social complexity,population, and intensity of land use (Tor-rence and Grattan 2002:5). This new orienta-tion has had its problems, however, becausehumans are conceived as victims rather thanas active agents who creatively shape theirresponses to natural disasters. A better solu-tion is to consider vulnerability, environmen-tal hazards, and recovery as linked processesthat unfold over long periods of time (Grat-tan and Torrence 2007:3). The complexinteractions between violent events, usefulvolcanic products, and human agency com-prise fascinating material for analysis usingthe perspectives of historical ecology.

As a first step in encouraging studies ofhistorical ecology in South Pacific volcanicenvironments, we review the basic types ofhazards commonly found in these settings.Following the general description of eachtype, we draw on case studies from the Will-aumez Peninsula, New Britain, Papua NewGuinea, to illustrate how each component ofvolcanism contributed to the historical ecol-ogy. The case studies are devised to achievetwo goals. First, they illustrate the wide rangeof field and analytical methods that contrib-ute to the reconstruction of historical ecologyin volcanic settings. Second, they provide aspringboard for a general discussion of thepotential role of explosive volcanism in his-torical ecology, not only in the South Pacificarea but also worldwide.

explosive volcanism in the southpacific

In this paper we focus on explosive volcanismthat is typically derived from the plinian styleof volcanic activity rather than the effusivevolcanism classically associated with the Ha-waiian style of eruption. In the South Pacific,plinian volcanism is located at the convergentplate boundaries where subduction is occur-ring. Good examples include the followingvolcanically active areas: the Solomon Islands,Vanuatu, Tonga, the Kermadec arc, and theNorth Island of New Zealand. In the PapuaNew Guinea region a complex of microplatesforms a buffer between movement of thelarge Australian and Pacific plates. In ourstudy area, on the island of New Britain, it isthe Solomon Sea microplate to the south thatis being subducted northward beneath theSouth Bismarck microplate (Bird 2003). Thisis occurring at a rate of about 100 mm peryear in the region of our case study in theWillaumez Peninsula (Wallace et al. 2004).Plate tectonics has led to an array of explosivevolcanism throughout New Britain that re-flects magma generation from differentdepths across the subduction zone that pro-gressively deepens to the north ( Johnson1976, Woodhead et al. 1998). Our researchhas focused on volcanic activity derived pri-marily from two calderas that have been the

508 PACIFIC SCIENCE . October 2009

source of major plinian eruptions in theHolocene. These are the Witori volcaniccenter, 50 km east of the Willaumez isth-mus, and the Dakatau volcanic center at thenorthern tip of the Willaumez Peninsula(Figure 1).

tephra stratigraphy

Although volcanic activity often causes a cul-tural disaster, ironically the same event maybenefit later generations of archaeologists.The spectacular sites of Pompeii (e.g., Si-gurdsson et al. 1985) and Ceren (Sheets1992, 2002) illustrate how whole towns or vil-lages and their contents have been buried andtherefore preserved by deep falls of volcanicash and/or lapilli or pyroclastic-flow deposits.Airborne ash (technically ‘‘airfall tephra’’) canalso cover entire landscapes and freeze themin time. If there are multiple eruptions, a

‘‘layer cake’’ of volcanic deposits interbeddedwith levels of cultural material is formed.This tephra stratigraphy provides the essen-tial backbone for relative dating and enablesarchaeologists, earth scientists, and physicalgeographers to correlate layers with distinc-tive physical and/or chemical characteristicsover large areas (e.g., Sheets et al. 1991,Sheets and McKee 1994, Cronin and Neall2000, Torrence et al. 2000, Lowe et al. 2002,Machida and Sugiyama 2002, Zeidler andIsaacson 2003, Neall et al. 2008, Torrence2008).

Research into the interaction of humansocieties with the catastrophic environmentof the Willaumez Peninsula, Papua NewGuinea, provides a good example of the im-portance of a good tephra stratigraphy. Inthis case interdisciplinary research has built aregional Holocene tephra stratigraphy com-prising nine explosive volcanic eruptions and

Figure 1. Location of study areas on Garua Island and the isthmus region of the Willaumez Peninsula, two volcaniccenters (Dakatau and Witori), obsidian sources (in italics), and other places mentioned in the text.

Volcanism on the Willaumez Peninsula . Torrence et al. 509

identified four minor episodes with a morerestricted distribution (Table 1). These dis-tinctive strata are used to correlate eventsamong archaeological and environmental

contexts spread across the entire region. Asthe result of extensive collaboration among(1) the archaeologists who expose numerousdeep stratigraphic sequences through excava-

TABLE 1

Summary of Volcanic Activity Recorded in the Archaeological Record of Garua Island and the Isthmus Regions in theWillaumez Peninsula

Volcanic Event TypeaB.P. Date

(Based on TL)bB.P. Date (Basedon Radiocarbon)c

HoloceneW-H6 Plinian, subplinian, VEI 4 <500d

W-H5 Phreatomagmatic, VEI 4 <500d

W-H4 Plinian, VEI 4/5 <500d

W-H3 Phreatomagmatic, VEI 4 <500d

W-K4 Phreatomagmatic, plinian,ignimbrite forming, VEI 5

1,310–1,170c

Dk Phreatomagmatic, plinian,ignimbrite forming, VEI 5

1,350–1,270c

W-K3 Plinian, VEI 5 1,740–1,540c

Garbuna Pyroclastic flow ca. 1800e

W-K2 Phreatomagmatic, plinian,ignimbrite forming, VEI 5

3,480–3,160c

Unknown, east of Garbuna Pyroclastic flow 4,200e

W-K1 Plinian, ignimbrite forming, VEI5–6

6,160–5,740c

Kulu tuff Subplinian, phreatomagmatic >W-K1Numundo Maar Subplinian, phreatomagmatic <7500e

PleistoceneTephra H: Kupona na Dari Long period of small dustings of

airfall tephra23,200G 6,100

Tephra G: Kupona na Dari Long period of small dustings ofairfall tephra

Tephra F: Kupona na Dari Long period of small dustings ofairfall tephra

Tephra Lower E: Kupona na Dari Local subplinian or distal plinian 39,8000G 5,200,38,000G 10,4000

Tephras D1 and D2: Kupona naDari

Subplinian eruption

Tephra Upper C: Kupona na Dari Long period of small dustings ofairfall tephra

Tephra Middle C: Kupona na Dari Long period of small dustings ofairfall tephra

Tephra C: Kupona na Dari Plinian eruptionTephra B: Kupona na Dari Long period of small dustings of

airfall tephraTephra B1: Kupona na Dari Long period of small dustings of

airfall tephraTephra A: Kupona na Dari Series of eruptions in quick

succession culminating in aplinian eruption

36,000G 3,900

a VEI, Volcanic Explosivity Index.b Full dates reported in Torrence et al. (2004b). TL, thermoluminescence.c Based on Bayesian modeling as reported in Petrie and Torrence (2008), except where noted.d Based on unpublished dates and Machida et al. (1996:71, fig. 4).e Based on McKee et al. (2005:17–18, table 1).


tions and locate others through field survey;(2) the geologists, geochemists, and soil scien-tists who interpret and characterize the layers;and (3) dating experts, a highly useful tephro-chronology has been reconstructed based onfive of the volcanic eruptions that impactedthe Willaumez Peninsula during the Holo-cene (e.g., Specht et al. 1991, Machida et al.1996, Torrence et al. 2000, Torrence 2002a,McKee et al. 2005, Neall et al. 2008; cf. Pav-lides 2006) (Figure 2). A Bayesian analysisof radiocarbon dates provides excellent chro-nometric control for the Witori and Daka-tau eruptions, thereby anchoring the tephrastratigraphy in time (Table 1) (Petrie andTorrence 2008). The Pleistocene volcanichistory of the region is much less known be-cause these layers are very deeply buried andtheir diagnostic properties obscured throughextreme weathering in these tropical condi-tions, but studies of several key sections onGarua Island and at the Kupona na Dari siteon the mainland, together with luminescenceand fission track dating, provide a small win-dow into human/volcano interactions for theearliest periods of human settlement (Tor-rence et al. 2004a,b, Lentfer and Torrence2007) (Table 1).

The Holocene tephrochronology in theWillaumez Peninsula comprises the backbonefor dating the archaeological deposits and hasalso been invaluable for studying landscapechanges. Given the great thicknesses of mate-rial deposited near the volcanic center (e.g.,Figure 3), one has to be careful in choosing aregion where the full tephra stratigraphy canbe accessed. The most practical solution is tofocus on regions that were impacted only byairfall tephra (i.e., ‘‘ash’’) with thicknesses of1 m or less, so that archaeological data arewell preserved but still reasonably accessible.The two study regions that we have selected,Garua Island and the isthmus area at thesouthern end of the peninsula, are ideal sitesfor studying the impacts of tephra falls be-cause they are located between the two majorvolcanic centers at Witori and Dakatau (Fig-ure 1).

The history and character of human sociallife and land use during the Holocene hasbeen reconstructed through the interpreta-

tion of materials recovered from ca. 140 testpits (mainly 1 m2) that are distributed widelyacross the two study areas (Figure 4) (Tor-rence 2002a,b, 2004a, Specht and Torrence2007a, Torrence and Doelman 2007). Utiliz-ing expertise on geomorphology, plant mi-crofossils, diatoms, and coral ecology, theresearch team has also reconstructed the his-torical ecology in terms of (1) natural fac-tors, including relative sea level change andvolcanic activity and its consequences, suchas earthquakes, erosion, and tsunamis (e.g.,Boyd and Torrence 1996, Torrence et al.1996, Boyd et al. 1999, 2005, White et al.2002, Jago and Boyd 2005, Neall et al. 2007,2008, Specht and Torrence 2007b), as well as(2) anthropogenic processes of land manage-ment such as burning and plant translocation(e.g., Parr et al. 2001, Boyd et al. 2005, Lent-fer and Torrence 2007). Unfortunately, dueto the lack of contexts that preserve pollen,it has not been possible to adequately addressclimatic change in this research.

volcanic hazards

Geologists generally use the Volcanic Explo-sivity Index (VEI) of Newhall and Self (1982)to rank the scale of volcanic eruptions. Thosein the range of VEI 4 to 8 are classed as cata-clysmic or paroxysmal eruptions involving theeruption of between 1 and 10,000 km3 of py-roclasts to heights of 10 to >25 km (Pyle2000). These eruptions pose very serious con-sequences for human groups, particularly forthose living fairly close to the eruptive center.A good example is the 1991 eruption of Pina-tubo of VEI 6 that forced evacuation of avery large area, much of which is still notable to be safely recolonized, and also affectedglobal temperatures (Newhall and Punong-bayan 1996). Gaillard et al. (2007:225) notedthat the ‘‘eruption and its aftermath causedeconomic losses estimated at a billion US dol-lars, and wreaked havoc in the lives of twomillion people.’’ As shown in Table 1, mostof the Holocene eruptions from our casestudy in the Willaumez Peninsula were VEI5 or slightly higher, therefore posing extremedanger to human inhabitants. Also, given thatthe volcanic events consistently generated se-


vere hazards for human societies over a longperiod of time, we can be in no doubt thatthe Willaumez Peninsula qualifies as a ‘‘cata-strophic environment.’’

It is important to consider both the poten-tial positive as well as negative impacts of vol-canic hazards. Beyond the destruction zone,many of these same catastrophic processes

Figure 2. Holocene tephra stratigraphy in the Willaumez Peninsula.


Figure 3. The human scale provides a good indication of the devastation that must have been caused by this pyroclas-tic flow derived from the W-K2 eruption. This location is ca. 10 km from the Witori caldera.

Figure 4. Location of archaeological excavations on (A) Garua Island and (B) the isthmus region of the WillaumezPeninsula.

also provide new opportunities for humansthrough the creation and remodeling of land-scapes, the production of useful raw ma-terials, and enhancement of soil nutrients(Grattan and Torrence 2007). We considerthe following key factors: (1) seismic activity;(2) tsunamis; (3) explosive eruptions that gen-erate noxious gases, tephras, and lavas; and (4)geothermal activity. With the exception ofgeothermal activity, all of these happen sud-denly, although with various degrees of priorwarning. In addition, the scale and length oftime over which they occur cover a largerange. These natural hazards are markedlydifferent from the low-risk, slow-onset vari-ables that have been well described in studiesof Pacific historical ecology. Having intro-duced the important concepts about volca-nism and summarized our methods, we nowturn to a discussion of the major factors thathave conditioned historical ecology in Pacificvolcanic environments.

seismic activity

Due to the continuing movement of the Aus-tralian and Pacific lithospheric plates, muchof the South Pacific region is characterizedby active seismic activity, with accompanyingearthquakes. Seismic activity varies in magni-tude and frequency depending on the dis-tance of a location from the active edge ofeach lithospheric plate or microplate. Mostearthquakes result from faulting generatedbetween the convergent or strike-slip (trans-current) plate boundaries where plates arecompressed and deformed by the processesof subduction or lateral offset. These earth-quakes can show a large range in magnitudesand may represent some of the largest re-corded. Lesser-magnitude earthquakes andvolcanic tremors occur when magma makesits way toward the surface, and so ultimatelythese may accompany eruptive activity. Thisseismicity is usually localized within 10–50km of the volcanic center.

Turning to our case study, the effectsof seismic activity are evident from manymajor faults visible on contour maps and ae-rial photos of the Willaumez Peninsula. To-gether with large landslides associated with

them, many appear to date within the periodof human occupation (e.g., McKee et al.2005:11–13). Smaller faults that may havebeen associated with one or more earthquakeshave also been identified during fieldwork.Movement along these could have led tocollapses and landslides that had adverseeffects on human settlements and gardens,although they might also have created newclearings in the forest that promoted thegrowth of early colonizing or sun-lovingspecies like wild taros that were beneficial forhuman groups or created opportunities fornew gardens. Although we have not yetobserved clear connections between thesefeatures and human responses, it is worthstressing that, as in the present day, ancientpeople in the Willaumez Peninsula wouldhave regularly experienced earthquakes ofvarious degrees. Some of these would havecaused local modifications and possible dis-ruptions to drainages, forest resources, gar-dens, and settlements.

Within Garua Harbour there is abundantevidence for both localized uplift and subsi-dence caused by relatively recent tectonicactivity (see review in Specht and Torrence[2007b:134–135, plate 6]). Among the mostobvious are large areas of dead corals thathave been uplifted as much as 1 m above thehigh-tide mark since 1973. Also, in anotherarea Torrence and Webb (1992; cf. Boydand Torrence 1996) reported lines of oysterstogether with dead corals that are currentlyabout 1 m above high-water mark. These in-dicate a sudden uplift episode interpreted asrepresenting a large earthquake and are ra-diocarbon dated to ca. 410–185 cal. B.P.(Specht and Torrence 2007b:135, plate 5). Al-though the evidence for this earthquake isbest preserved along the coast, its effectswould probably have been more widespread.Such an event could have caused severe dam-age to houses in the local area that might haveled to fatalities, and, crucially, it might alsohave generated a tsunami.

tsunamis

Tsunamis are a wave or series of waves thatmay suddenly inundate a coastline irrespec-


tive of the meteorological conditions. Theirnonmeteorological origin distinguishes themfrom storm waves. At sea they may travelvery fast, at speeds of 1,000 km/hr, but inshallower waters they lose energy by fric-tional loss, interacting with the seafloor toslow to <65 km/hr. At the same time theybuild in height and may attain >30 m. Theyare most often generated by a displacementof the seafloor caused by tectonic faultingbut are also created by accompanying earth-quakes triggering submarine landslides or bycollapse of volcanic edifices into the sea. Be-cause they often travel over large distances,tsunamis can impact on groups well outsidethe source region, but those living close toactive tectonic environments and volcanoesnear the sea are generally at higher risk. Tsu-namis are a formidable hazard that has notbeen given as much attention by archaeolo-gists as is merited given the strengths of theirimpacts on mortality. A review of historicalrecords concerning tsunamis created by vol-canism found that these were responsible foras much as a quarter of the total deaths asso-ciated with volcanic events (Neall 1996).

In the Willaumez Peninsula case, depositsthat can be ascribed to tsunamis are surpris-ingly rare given the large number of explosivevolcanoes and evidence for earthquakes. Sofar, the only example is on Boduna Islandwithin Garua Harbour (White et al. 2002),but its effects on the local population are un-known. The 30 cm thick paleotsunami de-posit consists of a coarse limestone gravelwith a gray sandy matrix that is sandwichedwithin the well-established regional ash se-quence. Older red brown, clay-rich tephrasunderlie the deposit, and almost 0.5 m of yel-low brown friable loamy tephra overlies it.

Archaeological evidence for cultural disas-ters due to tsunamis in other areas of theSouth Pacific is also quite scarce, perhaps be-cause in the past, people avoided placing theirsettlements in places they knew were riskybased on past experiences (cf. Davies 2002).The Kurvot site in Vanuatu, where tsunamideposits overlie cultural material, is a notableexception. Galipaud (2002:166) noted thatthis natural disaster may have had a long-term effect on local groups because subse-

quent settlements were shifted away from thedangerous coastal zone. Overall, the situationmay resemble that for the North Pacific dis-cussed by Johnson (2002), Saltonstall andCarver (2002), and Losey (2005), who allstress that societies have been extremely resil-ient in the face of earthquakes and associatedtsunamis and point out that these may evenbe conceived of as beneficial because theyhave led to the creation of new land (but cf.Beget et al. [2008] for a different view).

The impacts of tsunamis are quite limitedin spatial terms, generally within 50–150 m ofthe shoreline (e.g., Davies 2002:30). Becausethe potentially dangerous area is more pre-dictable than most volcanic hazards, groupscan easily avoid the most serious conse-quences simply by placing their settlementsoutside that zone. Along these lines, it is in-teresting that through time, settlements onGarua Island appear to have been increas-ingly focused on inland zones rather than onthe beach (Torrence 2002b), although oldercoastal sites may have been removed by ero-sion following recent uplift.

noxious gases

Highly acidic and noxious gases can beassociated with explosive eruptions. Theycomprise steam, carbon dioxide, and sulfurdioxide that can be emitted in large quanti-ties, and hydrochloric acid, hydrofluoric acid,and ammonia in lesser amounts. All thesegases are harmful if inhaled because theyharm the respiratory system and irritate eyesand skin, sometimes causing acid burning.The effects are likely to be most severe onthose living close to the volcano, particularlyin low-lying areas where gases heavier thanair may accumulate to toxic levels. On larger,high-elevation volcanoes the effects are lessdramatic because the gases are likely to bedissipated by the wind (for an exception seeGrattan et al. [2002, 2007]), but a downdraftcould concentrate gases in a localized area.The most disastrous case of mortality fromnoxious gases was in 1986 when an estimated240,000 metric tonnes of CO2 was suddenlyreleased from Lake Nyos in Cameroon as-phyxiating 1,746 people and 3,000 cattle in


nearby valleys (Neall 1996). Identifying theconsequences of noxious gases in prehistoriccases is only possible when human remainshave been preserved, but their consequencesare still worth considering when reconstruct-ing the potential hazards experienced in thepast.

tephras

Tephras comprise the most common and,generally, the largest-scale hazard createdby explosive eruptions. Depending on theamount of material deposited, which is deter-mined by the magnitude of the eruption,both the pyroclastic flows and airfall tephrathat accompany this class of hazard can haveextremely negative impacts on all life forms,and often they have markedly reshaped thephysical environment. The various types oftephras and their impacts are well illustratedby the Willaumez Peninsula case study.

First, within the vicinity of the Witori andDakatau volcanic centers, pyroclastic flowsfrom a plinian eruption would have totallydestroyed all human life and the plant andanimal resources on which they depended,and would also have drastically altered thephysiography. The scale of the pyroclasticflows in the Willaumez Peninsula variedamong the five large Holocene eruptions,but the majority were restricted to withina ca. 20 km radius of each eruptive center.The W-K2 event was a notable exception. Itgenerated pyroclastic flows tens of metersthick that traveled at great speeds (probably>100 km/hr) as much as 40 km, partially by‘‘skating’’ across Kimbe Bay, although theymay have moved only along a relatively nar-row path (Machida et al. 1996). We expectthat many people would have escaped thedirect effects of this type of hazard becausethere would have been considerable warningin the form of earthquakes. After very largeeruptions, groups would probably have fledthe most highly impacted regions. In con-trast, some of the populations tens of kilo-meters away may have still been vulnerableto the flows.

Second, airfall tephras range in size fromblocks, or volcanic bombs, the size of foot-

balls to tiny particles less than 2 mm that aretermed ‘‘ash’’ by volcanologists. In general,the farther one moves from the source, thesmaller the size of the airfall material, butthe scale of impact also depends on the mag-nitude of the eruption. Destruction of lifeforms from tephra can be caused by immedi-ate impact or, over a broader area, by burial.Airfall tephras by themselves are rarely a di-rect cause of human mortality, except wherethey adversely affect air quality. Very dustyconditions, especially when the tephra con-tains toxic chemicals, can have immediate ad-verse impacts on health leading to death. Inaddition, if poisonous substances adhere tothe surface of airfall tephras that have passedthrough a toxic eruption cloud, they can alsocause harm (Grattan and Gilbertson 1994).

The main effect of tephras on humans isindirect through damage inflicted on essentialresources, such as by fouling water suppliesand destruction of the plant and animal com-munities on which they depend. The degreeof impact depends on the chemical composi-tion of the tephra, the presence of toxins onits surface, its thickness, and local conditionssuch as potential buffering through local veg-etation and geology. For example, tephrawith high amounts of fluorine can cause im-mediate mortality and make areas uninhabit-able for very long periods of time throughcontaminating both water supplies and foodcrops, ultimately impacting on human health(Cronin 2006).

The severity of impacts resulting from air-fall tephra also depends to a large degree onthe climate and the season of the eruption be-cause these set limits on how fast the biotacan regenerate. For example, biological com-munities take much longer to recolonize areasburied by volcanic tephras in Alaska (e.g.,Vanderhoek and Nelson 2007) than in tropi-cal Pacific regions (e.g., Lentfer and Boyd2001), although there will be further variationdepending on how far seeds, spores, and re-colonizing species have to travel (Thornton1996). Based on previous studies, it has beenobserved that thicknesses greater than 50 cmof airfall tephra will strip the tropical forestcanopy and destroy the ground cover to cre-ate a virtual desert. In the 20–50 cm thickness


range trees will be defoliated, but some plantsmay be able to regenerate from buried rootsand suckers or viable wood remaining above-ground. In lesser amounts of tephra, treesmay be able to persist despite some loss ofleaves, but many garden crops will not sur-vive, especially in the absence of rainfall(Blong 1984:316–355, Lentfer and Boyd2001, Torrence 2002a:301–302, Neall et al.2008). The potential for regeneration also de-pends on the chemical content of the tephra,but in all cases it will be deficient in elementsessential for plant growth such as nitrogenand will lack organic matter. Many volcanicsoils are highly favorable for agriculture dueto their textural properties and chemical com-position, but these properties are not sharedby freshly deposited tephra. Soil formationover at least a generation, even in a tropicalsetting, is required to support wild vegetationsufficient to sustain a population or even forsuccessful gardening.

The Willaumez Peninsula case study hasbenefitted from large-scale field mapping ofairfall tephra thicknesses (isopachs) by Ma-chida et al. (1996). The regional picture hasbeen enhanced by stratigraphic informationcollected from test pits in the isthmus studyarea (Figure 4). These data have been usedto construct isopachs for the four HoloceneWitori eruptions in the study area (Figure5), providing an excellent basis for assessingthe volcanic hazard generated by the fivelarge eruptions from the Witori and Dakatauvolcanoes during the Holocene (cf. Boyd etal. 1999).

Archaeological research by Pavlides (2006;pers. comm.) near Yombon in the southernfoothills of the central Whiteman Range andthe Lamogai Plateau found thicknesses of air-fall tephra well over 20 cm from the W-K1and W-K2 eruptions, as much as 150 kmfrom Witori, and smaller thicknesses of theother Witori and Dakatau events. Archaeo-logical data from Garua Island and the Will-aumez isthmus confirm that after the W-K1and W-K2 eruptions, thicknesses of >50 cmof tephra fell in both areas; the isthmus alsoexperienced >50 cm of W-K3 tephra, lesseramounts of W-K4, and only a trace of tephrafrom the Dakatau Dk eruption. Sites on

Garua had thicknesses of up to 1 m of Dkbut only small amounts of W-K3, and W-K4 may be absent (Torrence and Doelman2007: tables 3.2, 3.3).

The large volumes of airfall tephra thatfell on the Willaumez Peninsula largely man-tled the existing landscape, but they had moreprofound effects on coastal environments. Atargeted study of the southeastern side ofthe isthmus based on data on buried corals,sediments, and phytoliths showed how the ac-cumulation of thick layers of Witori tephrasaugmented by material eroded from nearbyslopes considerably extended the coastal plain(Boyd et al. 2005).

Unstable Tephras

Even after an eruption has terminated, deepfalls of tephra compose an important hazard.New hazards are generated by the instabilityof loose, unconsolidated tephra. The debrisis frequently remobilized into lahars thatpose very serious hazards to communities lo-cated downslope from the volcano. Depend-ing on how long after the eruption they areformed, they may still be boiling hot. Becauseit can take several generations for the slopesto stabilize, lahars compose a long-term haz-ard around the base of the volcano, as il-lustrated by the ongoing damage causedthrough flows and flooding near Pinatubo,which erupted in 1991 (Crittenden and Ro-dolfo 2002, Gaillard et al. 2007). In this casethe lahars have created (1) dams that formlarge lakes, but these regularly collapse, caus-ing severe flooding; (2) vast, barren flood-plains composed of coarse unsorted material;and (3) large tidal fans that have destroyedcoastal resources.

After each eruption, remobilization of thetephras in the proximity of Witori Volcanoprobably reoriented drainage patterns, partic-ularly in the floodplain of the Kapiura Riverto the north of the caldera. Massive altera-tions in landscape after the W-K2 eruptionwould have radically altered access routes tothe nearby Mopir obsidian outcrops (Figure1). Together with risks caused by the ongoinginstability of the tephras, it is not surprisingthat obsidian from this source largely disap-


peared from archaeological sites for a consid-erable length of time after that eruption (e.g.,Torrence et al. 1996, Summerhayes et al.1998, Torrence 2004a).

Large-scale remobilization of tephras canalso occur in regions quite distant from thevolcano, leading to landslips, blockages ofwatercourses, and formation of coastal fans,and eventually to the creation of a coastalplain, as probably occurred on both Garua Is-land and on both coasts within the isthmus.The W-K1 tephra appears to have been par-ticularly unstable, perhaps because it fell dur-ing the rainy season. It has a very patchyoccurrence in the archaeological test pits inboth our study areas, but thick redepositedlayers are often identified in low-lying con-texts (e.g., Specht et al. 1988:8–9 [Layer 3],Torrence et al. 1990:461 [ash in photos],Neall et al. 2008), suggesting possible redepo-sition by flooding.

Distant Effects of Remobilization

The enormous scale of potential landscapetransformation resulting from the remobiliza-tion of deep falls of tephra can be hard toimagine, so we provide an example that isparticularly impressive because the massivechanges occurred at a considerable distancefrom the volcanic center. The data for this re-construction were gathered through intensivemapping of horizons, benefitting from exten-sive modern drainage systems in the oil palmplantations combined with paleoenvironmen-tal cores. Until about 6,000 years B.P. a largeembayment on the west side of the Willau-mez Peninsula (here termed Kulu Bay to dis-tinguish it from the current Riebeck Bay)extended 17 km farther inland from the cur-rent coastline, covering an area of about 200km2 (Figure 6). Coral dated at 6,733G 32years B.P. (Wk-15506) grew along the north-ern shoreline of this bay. After the W-K1eruption between 6,150 and 5,770 cal. B.P.,pumice was washed into the shallow watersof Kulu Bay, transforming the environmentinto a low-lying terrestrial swamp. After theW-K2 eruption, the swampy environmentwas overwhelmed by a massive flood of pum-ice derived from the Kulu-Dagi River, inun-

dating all the lowland. These redepositedbeds are dominated by cross-bedded pumi-ceous fine and medium sands. They vary inplaces to fine pumiceous gravel; toward theupper contact clayey sand may be preserved.Small upward-fining sequences are preservedin some places with fine horizontal lamina-tions.

Close to the point where the river exitedfrom the surrounding hills, the redepositedsurface steepens in gradient, narrows, andforms a prominent terrace on the Tili oilpalm estate (here named the Tili surface).This demonstrates a massive discharge im-mediately after the W-K2 eruption, around3,500–3,200 cal. B.P., unlike any subsequentalluvial event in the region. Our interpreta-tion is that either heavy rains remobilizedW-K2 pumice from the steep slopes of theheadwaters of the Kulu-Dagi River immedi-ately after the eruption (as happened aroundMount Punatubo in 1991) and/or ephemeraldams may have been created from unstabledebris, mobilizing large volumes of pumi-ceous sediment. By extrapolation, our esti-mates for the volume of redeposited materialare between 0.2 and 0.4 million m3. This sed-iment was either carried directly by flood-waters or released by a sudden dam collapseto create a megaflood that radically trans-formed the lower catchment downstream ofwhere it exits the hills. Consequently, muchof the former Kulu Bay was infilled within arelatively short period of time.

A Punctuated Prehistory

More than a decade of research in the Will-aumez Peninsula has produced a clear pictureof the relationship between humans and nat-ural disasters caused directly by airfall tephraor its remobilization. The Holocene volcaniceruptions led to a punctuated population his-tory, characterized by repeated cycles of col-onization and extinction, that when viewedoverall yielded very slow population growth(Torrence and Doelman 2007). Whereas thelong-term influence of volcanism on humanhistory is to reduce the potential for growth,the role of specific disasters in causing short-term cultural change is less clear.


Figure 5. Isopachs of tephra thicknesses in the isthmus study area, Willaumez Peninsula.

Figure 5. (continued)

Given their potential impacts as measuredby their thicknesses and the history of theirredeposition, one would expect that the airfalltephras would have precipitated very seriouscultural disasters after the W-K1 and W-K2eruptions throughout the region stretchingfrom Witori to at least 150 km to the west.The Dakatau Dk eruption would have devas-tated the northern end of the WillaumezPeninsula. After all these events, not onlywere the lands owned by people living inthe study region made uninhabitable, but thehomes and gardens of their neighbors werealso destroyed, making it very difficult for

populations to find refuge for the several gen-erations necessary before resettlement wouldhave been possible. Tephra thicknesses re-sulting from the W-K3 event would alsohave forced abandonment of the two regions,although regeneration would have been muchfaster. W-K4 would have had serious impactsfor the isthmus region but much less forGarua Island.

It is not surprising that the archaeologicalrecord shows local extinctions of human pop-ulations in the study areas following all thevolcanic disasters. Even with Bayesian model-ing, however, it is difficult to reconstruct the

Figure 6. The white line shows the location of the ancient Kulu Bay, which was infilled following the W-K2 eruption.Triangles indicate positions where the Tili surface is currently exposed, and squares mark locations of archaeologicaltest pits. (Base map reproduced with permission from Google Maps/Earth imagery.)


exact periods of abandonment due to the sta-tistical nature of radiocarbon chronology. Ifone uses the modes, it is clear that all theeruptions caused depopulation for long peri-ods of time, and there is a general correlationbetween the scale of the airfall tephra and thelength of abandonment (Table 2) (Petrie andTorrence 2008). This repeated pattern of ex-tinction followed by colonization sets the pa-rameters within which human populationshad the potential to interact with the naturalenvironment.

Clearly, the timing of the colonization inrelation to the state of recovery of the ecosys-tem and the stabilization of sediments wouldhave been critical. Did populations wait untilthe forest had completely regenerated or didsome take advantage of the disaster to makegardens in land cleared as a consequence ofthe tephra falls? There is a hint in our datathat people may have preferentially targetedhill- and ridgetops in the inland regions asthe first places to recolonize, possibly be-cause less tephra was preserved in those local-ities due to immediate posteruptive erosion(Torrence and Doelman 2007:56–59). Foresttrees providing the only local terrestrial foodsources may have survived in those locations.In contrast, the coastal zones would havebeen swampy and unstable due to the outflowof redeposited sediments that probably alsohad a negative effect on marine resources.

As shown in Table 2, when the data areviewed in terms of single eruptions, the cor-relation between strength of the volcanic

forcing agent and human response is not to-tally straightforward, thereby indicating thatcultural factors may have been important,perhaps along the lines of the ‘‘random’’ fac-tors and ‘‘novelties’’ identified in the way nat-ural systems have recovered from disasters(Hoffmann and Parsons 1997, Turner andDale 1998). For instance, the very long pe-riod of abandonment after W-K1 in the isth-mus is difficult to account for solely on thebasis of the volcanic disaster, especially whencompared with W-K2, whose impact washigher but not markedly so. This raises thequestion of whether W-K1 resulted in verywidespread human mortality and exception-ally slow population growth in surroundingregions and/or there was some special reason(e.g., cultural concepts of place) why popula-tions continued to avoid this region (Tor-rence and Doelman 2007:53). Similarly, theseemingly long gap after W-K3 on Garua Is-land does not have a clear relationship withthe potential impact of the event, especiallywhen compared with W-K2 or Dk, and soother factors must have affected the rate andnature of recolonization.

Another interesting point that arises fromthe case study is that the local extinction of apopulation due to a volcanic disaster does notnecessarily lead to the collapse of a culturalsystem, although it may be a contributing fac-tor to cultural change (cf. Manning and Sew-ell 2002 with Allison 2002, Driessen 2002). Inthe case of the Willaumez Peninsula, themanufacture and use of obsidian flaked toolsthat have distinctive stems or tangs, knownas ‘‘stemmed tools’’ (Araho et al. 2002), sur-vived the W-K1 eruption despite the longperiod of abandonment. In contrast, the cul-tural practices associated with them disap-peared after the W-K2 eruption. At thattime Lapita pottery was introduced to theWillaumez Peninsula. It is possible that thedisaster resulting from W-K2 seriously dis-rupted long-distance interaction indicated bythe widespread distribution of stemmed toolsin Melanesia, and this contributed to the ap-pearance of Lapita-style pottery (Torrenceand Swadling 2008).

Our discussion of the impacts of volcanicactivity in New Britain on cultural changehighlights the limitation of our case study in

TABLE 2

Length of Abandonment after Volcanic Eruptions in theWillaumez Peninsula Based on Bayesian Modeling

Reported in Petrie and Torrence (2008)

Isthmus Garua

Eruption

Interval95%HPDa Mode

Interval95%HPD Modes

Post W-K1 1,350–2,000 1710 0–280 110Post W-K2 0–300 150, 160 0–300 135, 145Post W-K3 0–160 55, 70 0–270 95, 110Post Dk — — 0–260 215Post W-K4 0–170 100 — —

a HPD, highest posterior density.


the Willaumez Peninsula. Although the dataprovide an excellent example of the generalcharacteristics of human history that derivefrom a catastrophic volcanic environment,they are not adequate for understanding cul-tural changes because human societies typi-cally operate over much broader scales. Infact, the large scale of social systems and thelength of networks may actually be an out-come of adaptation to the catastrophic envi-ronment itself.

Not all explosive volcanic activity causescultural disasters on as vast a scale as the largeplinian eruptions from Witori and Dakatau.It is worth comparing these with a smalleruption of the Numundo Maar volcano, lo-cated on the northern edge of the isthmus re-gion (McKee et al. 2005:7–8). This smallcrater (ca. 500 m diameter) exploded pyro-clastic surges, scoria, and ash on several occa-sions sometime shortly after ca. 7,500 B.P.Deposits from the most recent eruption haveformed a hard layer of tuff up to 4 m thick atthe maar, thinning to 2 m within 1 km awayand to 0.5 m at 5 km distant from the crater.Such an event would certainly have causedconcern to the local community (repre-sented by artifacts that are stratified underthe tephra), but it seems likely that theywould have been able to readily escape theimmediate danger. Over the longer term, thesmall area covered by the compacted tephrawould not have been able to support vegeta-tion for many years. Only after the W-K1eruption deposited a layer of loose airfalltephra over the top of the tuff, creating a me-dium in which plants could eventually grow,was this relatively small area capable of sus-taining a human population, as evidenced bythe presence of dense archaeological depositsimmediately overlying it. Despite its smallscale, however, this event could neverthelesshave impacted on cultural practice and ideol-ogy and perhaps been incorporated into localcultural memory through oral history andmyth.

useful products

From a human perspective, the most impor-tant benefits of volcanic lavas are their useas raw material for various types of stone

tools. Many highly prized volcanic stonesand glasses were widely exchanged in the Pa-cific region both as raw materials and as fin-ished products. The relatively coarse-grainedstones were often used to make ground andpolished axes and adzes, whereas the fine-grained and glassy materials were flaked toyield sharp edges. Both types were also con-verted into distinctive shapes with impor-tant symbolic meanings that functioned associal currency (e.g., Firth 1959, Torrence2004b, Specht 2005, 2007, Kirch and Kahn2007). Communities living in the vicinity ofthese desirable raw materials had the poten-tial to create social capital by monopolizingaccess to these resources, developing special-ist skills for manufacturing products, and or-ganizing their transport to other regions.Consequently, within the wider Pacific regionmany volcanic stones played important so-cial roles within an island or island groupand also circulated among populations sepa-rated by very large distances (e.g., Collersonand Weisler 2007, Summerhayes 2007).

Throughout the prehistory of the Willau-mez Peninsula the most widely exchangedand valued volcanic stone was obsidian de-rived from three major source regions (Tor-rence et al. 1992) as well as from the Mopirsource situated near the Witori Volcano(Fullagar et al. 1991). A shiny, distinctivestone such as obsidian would certainly appealto the senses and is therefore perfectly suitedfor exchange (Torrence 2005). In the form ofunworked nodules, partly worked cores orpreforms, and as highly worked products rep-resenting skilled craftsmanship, obsidian waspassed among local groups and transportedover enormous distances (e.g., Specht 1981,Summerhayes et al. 1998, Rath and Torrence2003, Torrence 2004a,b, Summerhayes 2007,Torrence and Swadling 2008, Torrence et al.2009). In addition, volcanic lavas (primarilyrhyolite) that were probably sourced locallywere used in the production of ground stoneimplements from the mid-Holocene up to re-cent times (e.g., Specht 2005, 2007). Cur-rently, little is known about whether andhow they were traded, but the finding nearKimbe of a cache of 14 ground stone axeblades suggests that some were accumulatedand traded as valuable objects.


Obsidian was first transported and possiblyexchanged as soon as people colonized theWillaumez Peninsula, and it continued to bemoved over various geographical scales upuntil the recent past (Specht 1981, White1996). Because obsidian exchange has per-sisted perhaps as long as 40,000 years, despitethe discontinuous settlement history of theWillaumez Peninsula, it seems likely that thiskind of cultural behavior might represent aform of adaptation to this catastrophic en-vironment. The social links forged by theexchange of obsidian (and possibly also ofground stone tools as well as many perishablematerials) could have provided access to cru-cial places of refuge and other forms of as-sistance for people who often suffered thedisastrous consequences of volcanic activity(Torrence 2004b, Torrence and Doelman2007:52–53). For this reason, the pattern ofseeking out trading partners in other placeswas possibly maintained throughout the hu-man history of the region.

geothermal activity

The most constant volcanic features in theWillaumez Peninsula are various forms ofgeothermal activity such as hot springs, boil-ing pools of mud, and geysers. Currently,the most active areas are at Pangalu and Tala-sea, on Boduna Island, along the east coastnear the village of Patanga, at the top ofMount Garbuna, and scattered around thesouthern foot of Mount Garbuna (Hemingand Smith 1969). Although the position ofthese geothermal areas is relatively stable,and, in general, they have had few detrimen-tal effects on human life or resources, newareas arise from time to time and can renderpatches of forest or fields unsuitable for use.In 2005 and 2008 the geothermal area on thesummit of Mount Garbuna experienced mi-nor explosions of steam-bearing reworkedsediments that lasted several weeks (Smithso-nian Institution 2005), but so far there havenot been serious negative consequences forhuman life or essential resources.

Geothermal activity in the WillaumezPeninsula has possibly produced as many pos-itive as negative human impacts. These areasattract megapodes, flightless birds that lay

their eggs in the warm soils. Both eggs andbirds provide a predictable and rich sourceof food, although overexploitation is a pos-sibility. Colorful soils formed near the hotsprings at Talasea and near Patanga wereused as pigments and traded widely in thepast (Specht 1981). The hot water can alsobe useful as a medium for cooking.

anthropogenic factors

Given the scale and frequency of environ-mental changes in volcanic environments andthe resulting necessity for abandonment, ref-uging, and recolonization, human populationsare frequently faced with simple, depauperateecosystems. In such cases one might questionthe potential for the types of anthropogenicfactors so often stressed in previous studiesof the historical ecology of South Pacific is-lands (e.g., human subsistence and land-usepractices leading to extirpation, extinction,deforestation, and erosion [e.g., Anderson1989, Enright and Gosden 1992, Kirch andHunt 1997, Steadman 2006]).

Boyd and Torrence (1996) reported theresults of a stratigraphic study of 30 sectionslargely from archaeological excavations onGarua Island supplemented by a study ofcoastal geomorphology (e.g., raised coralsand fossilized beaches) on surrounding islandsand the mainland. This was aimed at examin-ing the possibility that prehistoric gardeningpractices led to periods of erosion as hadbeen observed in other Pacific regions (e.g.,Enright and Gosden 1992, Gosden andWebb 1994, Spriggs 1997). They noted thatmajor episodes of erosion require the pres-ence of exposed soils that can be mobilizedby slope wash, slope creep, and runoff, andthat human land clearance is only one poten-tial cause. The destruction of vegetation dueto the emplacement of airfall tephra is an-other factor that should be considered.

The Holocene sedimentary record onGarua Island holds remarkably little evidencefor soil erosion. Boyd and Torrence (1996)described seven periods of erosion, identifiedwith the assistance of the regional tephrastratigraphy, based on an analysis of 30 sec-tions from archaeological sites, but only oneof these, dating to the twentieth century, was


convincingly associated with human activity.Only the earliest period of erosion could beaccounted for by regional sea-level changes,three events were closely tied to volcanicactivity in the form of airfall tephras, and thecause of two others was more likely due totectonic factors than to human land-use prac-tices.

Turning to evidence from plant microfos-sils, analyses of phytolith and starch assem-blages taken from excavations on GaruaIsland and the Willaumez isthmus region(Boyd et al. 2005, Lentfer and Torrence2007) indicate that humans were using fire tocreate breaks in the forest at least by the earlyHolocene (and possibly before that). By themiddle Holocene this practice prevented nat-ural succession following volcanic eruptions.Based on the presence of phytoliths, a seriesof economically useful plants was introducedthrough time beginning in the period afterthe W-K1 eruption. Both the plant microfos-sils and studies of tool use show that therewas little if any changed land use from beforeto after theW-K2 eruption (i.e., pre- and con-temporary with Lapita pottery) (Kononenko2007, Lentfer and Torrence 2007), contraryto previous predictions for an increase in theintensity of land use through time (Torrenceet al. 2000, Torrence 2002a). Overall, therewas a very slow increase in the intensificationof landscape clearance after earliest settle-ment, but humans did not make a major im-pact on natural vegetation until after the lastmajor volcanic eruption (Dk/W-K4) roughly1,310–1,170 years ago, when there was aquantum leap in the degree of human inter-ference in the successional sequence (Boydet al. 2005, Lentfer and Torrence 2007).Population levels, as monitored by the rateof deposition of obsidian artifacts, also followthe same pattern of very slow increase untilafter the major Witori and Dakatau eruptionshad ceased (Torrence and Doelman 2007:56).

Pulling all the threads together (erosion,plant macrofossils, nature and intensity oftool use, and population levels), it is clearthat the high frequency of severe volcanicdisasters restricted the influence of humanimpacts on the historical ecology of the Will-aumez Peninsula in comparison with other

more stable environments described, for ex-ample, in Kirch and Hunt (1997). The volca-nic disasters intervened relatively frequentlyto grossly remodel the landscape and resetthe ecosystem back to an early stage of suc-cession, so perhaps there was less possibilityfor severe disruption of the ecosystem by hu-man land practices, because the ecosystemshad already been disturbed by volcanic activ-ity. On the other hand, the volcanic eruptionsmay also have opened up new opportunitiesfor populations. If human groups returned tothe region during the early stages of succes-sion, they would not have been forced to clearprimary rain forest to establish gardens. Fol-lowing proposals by Denevan (2001, 2004)for the history of agriculture in Amazonia, itmight have made sense for the first colonizersto develop techniques to maintain the soil fer-tility of their plots and/or to restrict regrowthin a swidden system of cultivation to sec-ondary forest rather than adopt a wide-scalesystem of shifting cultivation that requiredclearance of primary forest.

memory

Many human societies retain the memories ofprevious volcanic disasters through oral tradi-tions, myths, and religious practices. Thesemay have played a beneficial role by pro-viding the impetus to avoid dangerous areas,by alerting people about the onset of volca-nic hazards, and by encouraging evacuationbefore the explosive phase of the eruptionbegan, as has been observed with recent vol-canic events elsewhere in the South Pacificregion (e.g., Lentfer and Boyd 2001, Croninand Cashman 2007, Gaillard et al. 2007; cf.Chester and Duncan 2007).

Based on case studies in other areas of theworld, we would also predict that volcanicdisasters might be creatively incorporatedinto social practice and memory, especiallywithin ritual and ideology (Grattan and Tor-rence 2007:10–11) (e.g., examples in Blong1982, Plunket and Urunuela 1998, Elson etal. 2002, Chester and Duncan 2007, Dillian2007, Holmberg 2007). To date there is verylittle evidence about how people in the Will-aumez Peninsula conceived of volcanic activ-ity and if they incorporated these concepts


into their daily lives. A broken Lapita pot re-covered from a modern spring in a locationnear geothermic activity (Specht and Tor-rence 2007a:90) may suggest ritual activity.Obsidian artifacts commonly found near hotsprings and geysers in this region may also in-dicate that these areas held special signifi-cance. Finally, it is worth noting that becausethe important Kutau/Bao obsidian source wasformed by an eruption that occurred after theregion was first colonized, populations mayhave associated the obsidian with a powerfuland frightening event. The ascription of spe-cial meaning to this stone may therefore helpexplain why material from this source andnot others derived from the Willaumez Pen-insula was most highly favored throughoutprehistory and was the only obsidian widelydispersed outside the immediate region (Tor-rence et al. 2004a,b, Summerhayes 2007; cf.Dillian 2007).

On the contrary, it could be argued thatcatastrophic environments provide little op-portunity for memory because of high mor-talities, long periods of abandonment, theremodeling of the landscape, and the removalof potential landmarks (Torrence and Doel-man 2007:53–55). It is possible that somememories were retained among societies inthe Willaumez Peninsula despite catastrophicenvironmental changes. For instance, the re-colonization of inland regions was very rapidin some areas, suggesting that people had de-liberately returned to familiar places. One ex-ample is Lapita-style pottery, found on thehills that had once surrounded the Kulu Bay,although this tidal embayment, heavily fa-vored by populations in earlier periods, hadbeen radically altered and was now an inlandswamp. Although these hills might simplybe convenient spots for placing dwellings,the deposition of small quantities of potteryaway from the coastal settings more typicalof that time period might also signify the spe-cial significance of these places (Specht andTorrence 2007a:89–90).

historical ecology of catastrophicvolcanic environments

In a review of human responses to volcanicactivity in Central America, Sheets (2007:85;

cf. 1999) concluded that ‘‘egalitarian soci-eties with low levels of built environments,minimal reliance on intensive agricultureand domesticated staples, and low populationdensities exhibited the greatest resilience tosudden massive volcanic stresses.’’ Based onour study of a single region, we propose thatSheets’ argument is the wrong way around,probably because of his focus on single erup-tions rather than exploring the long-term in-teractions between catastrophic environmentsand cultural groups. In other words, thesmall, mobile populations with flexible subsis-tence and social strategies and long-distancesocial ties that Sheets described are actuallythe outcome of evolution within the highlyunstable volcanic environments both in NewBritain and in Central America. The relativefrequency of disasters has acted to keep pop-ulations low because these regions are pe-riodically abandoned, but people have alsofound creative ways to maintain themselveswithin the region after a volcanic event (e.g.,through using social links to provide refuges).In addition, they have adopted subsistenceand settlement patterns that enable them toreturn quickly, for example through targetingareas least affected by tephra falls.

Although, when considered on the timescale of thousands of years, human societieshave found ways to persist in volcanic envi-ronments, it is questionable to what extentand in what ways they can make a substantialand long-lasting impact on landscapes thatare so radically altered periodically. It is inthe realm of anthropogenic factors that thehistorical ecology of volcanic environments,and particularly those with frequent large-scale eruptions, may differ from other typesof settings. There is much more potentialfor human activity to disrupt mature ecosys-tems than those characterized by early stagesof succession, because the latter are primarilymade up of colonizing species that regeneratequickly. In fact, it is quite difficult to extractthe signal of human modifications from thoseof an ecosystem in the process of recoveringfrom a volcanic eruption (Lentfer and Tor-rence 2007). It would be very interesting infuture research to compare and contrast thelong-term history of vegetation changes dueto human agency in active volcanic environ-


ments versus those in more stable island envi-ronments in the South Pacific.

The Willaumez Peninsula case study raisesfurther issues that would benefit from com-parative studies both within the Pacific regionand further afield (cf. Torrence and Grattan2002:11–15). First, a broader range of studiesabout how humans have coped with, adaptedto, and taken advantage of the products ofvolcanic activity is required to see if thereare common strategies that have been devel-oped across volcanic environments. To beginwith, comparative studies of cultural disas-ters should examine variation in the strengthand frequency of the environmental forcingagencies and identify the factors that havehad the most impact. Second, better infor-mation is required concerning strategies thatenable groups to survive disasters and, in par-ticular, how they find refuge for long periodsof time outside the afflicted region. We needto know how far afield people successfullyfind refuge and what social strategies areused to broker access to resources. Are vic-tims limited to areas where they had alreadyestablished social links, perhaps through ex-change? From a methodological point ofview, we need to know how to identify the ar-rival and subsequent impacts of an influx ofrefugees (e.g., Lilley 2004a,b).

Another important issue demanding re-search is (re)colonization. It would be in-teresting to know if there are similarities inthe way people have (re)colonized (1) land-scapes after volcanic activity versus (2) islandsnever previously inhabited. Ideally, we wouldlike to understand the relationships amongpopulation size (both of the source popula-tion and the founding population), subsis-tence strategies, the length of time a regionis abandoned, and the speed with which (re)-colonization takes place.

It seems possible that over a long pe-riod, populations resident in regions withactive volcanism may have devised particularstrategies that cope well with frequent cyclesof abandonment and recolonization. Thesemight include many of those noted for theWillaumez Peninsula (e.g., high mobility,flexible social forms and subsistence systems,the maintenance of long-distance social ties,

the rapid adoption of new technologies, andother forms of cultural behavior). If so, thenthese groups might be preadapted for colo-nizing places that have never been settled be-fore. In this light, the presence of obsidianfrom the Willaumez Peninsula in most ofthe earliest Lapita sites in Remote Oceania(Summerhayes 2007) may be more informa-tive than just signaling the source of the pop-ulations.

conclusions

Returning to our criticism of Spriggs’ (1997)summary of the major factors that haveshaped the historical ecology in the SouthPacific region, it should now be clear thatnatural disasters, and especially volcanic haz-ards, demand more serious study than in paststudies. Since the influential review of histor-ical ecology edited by Kirch and Hunt (1997),modern disasters, such as the eruption of Pi-natubo and the Aceh tsunami, have no doubtraised the profile of these powerful forces andprobably influenced recent archaeologicalscholarship (e.g., summary in Grattan andTorrence [2007]), although hints about theimportance of volcanism in the South Pacifichad been in the literature for some time (e.g.,Garanger 1972).

In this paper we have used case studiesfrom interdisciplinary research in the Willau-mez Peninsula to illustrate the wide range ofprocesses that impact on human groups livingwithin active volcanic environments. We havetried to show that whereas these dynamic set-tings frequently produce cultural disasters,human groups have also responded creativelyto these challenges. No environment is com-pletely stable. All South Pacific islands are af-fected to some extent by climatic change, andmany experience rapid-onset natural disas-ters, such as cyclones. We have argued thatthe factors that set active volcanic environ-ments apart from other settings are the fre-quency of the hazards and the very largescale of their impacts. The next logical stepwould be to generalize our work even furtherby assessing the role of different kinds anddegrees of (in)stability in the historical ecol-ogy of the South Pacific region.


acknowledgments

We thank the following institutions in PapuaNew Guinea and their staff for long-termsupport of our research: National Museumand Art Gallery, National Research Institute,University of Papua New Guinea, West NewBritain Provincial Cultural Center, KimbeBay Shipping Agencies, New Britain PalmOil, Ltd., Walindi Plantation and Resort,Mahonia Na Dari Research Station. Numer-ous colleagues have provided comments,ideas, and various forms of assistance overthe years. We especially thank StephenAthens, Hugh Davies, Trudy Doelman, Rich-ard Fullagar, Matthew Irwin, Peter Jackson,Carol Lentfer, Hiroshi Machida, ChrisMcKee, Ken Mulvaney, John Namuno, Tan-ya O’Neill, Jeff Parr, Christina Pavlides,Ed Rhodes, Glenn Summerhayes, MichaelTherin, John Webb, and, above all, JimSpecht and Peter White. Finally, we acknowl-edge the invaluable support and friendship ofthe local communities where we have workedand the sterling efforts of our Papua NewGuinean and international volunteers.

Literature Cited

Allison, P. 2002. Recurring tremors: Thecontinuing impact of the AD 79 eruptionof Mt Vesuvius. Pages 107–125 in R. Tor-rence and J. Grattan, eds. Natural disastersand cultural change. Routledge, London.

Anderson, A. 1989. Prodigious birds: Moasand moa-hunting in prehistoric New Zea-land. Cambridge University Press, Cam-bridge.

Anson, D., R. Walter, and R. C. Green. 2005.A revised and redated event phase se-quence for the Reber-Rakival Lapita site,Watom Island, East New Britain Province,Papua New Guinea. Univ. Otago Stud.Prehist. Anthropol. 20.

Araho, N., R. Torrence, and J. P. White.2002. Valuable and useful: Mid-Holocenestemmed obsidian artefacts from WestNew Britain, Papua New Guinea. Proc.Prehist. Soc. 68:61–81.

Bedford, S., M. Spriggs, and R. Regenvanu.2006. The Teouma Lapita site and the

early settlement of the Pacific. Antiquity80:812–828.

Beget, J., C. Gardner, and K. Davis. 2008.Volcanic tsunamis and prehistoric culturaltransitions in Cook Inlet, Alaska. J. Volca-nol. Geotherm. Res. 176:377–386.

Bird, P. 2003. An updated digital model ofplate tectonics. Geochem. Geophys. Geo-sys. 4:1027, doi:10.1029/2001GC000252.

Blong, R. 1982. The time of darkness: Locallegends and volcanic reality in Papua NewGuinea. University of Washington Press,Seattle.

———. 1984. Volcanic hazards: A source-book on the effects of eruptions. AcademicPress, Sydney.

Boyd, W., C. Lentfer, and G. Luker. 1999.Environmental impacts of major cata-strophic Holocene volcanic eruptions inNew Britain, PNG: A preliminary modelfor palaeoenvironmental change. Pages361–372 in J. Kesby, J. Stanley, F.McLean, and L. Olive, eds. Geodiversity:Readings in Australian geography at theclose of the 20th century. Australian De-fence Force Academy, Canberra.

Boyd, W., C. Lentfer, and J. Parr. 2005. In-teractions between human activity, volca-nic eruptions and vegetation during theHolocene at Garua and Numundo, WestNew Britain, PNG. Quat. Res. 64:384–398.

Boyd, W., and R. Torrence. 1996. Periodicerosion and human land use on Garua Is-land, PNG: A progress report. Pages265–274 in S. Ulm, I. Lilley, and A. Ross,eds. Australian Archaeology ‘95: Proceed-ings of the 1995 Australian ArchaeologicalAssociation Annual Conference, Tempus6. University of Queensland, St. Lucia.

Cashman, K. V., and G. Giordano, eds. 2008.Volcanos and human history. J. Volcanol.Geotherm. Res. 176:325–438.

Chester, D., and A. Duncan. 2007. Geomy-thology, theodicy, and the continuing rele-vance of religious worldviews on responsesto volcanic eruptions. Pages 203–224 inJ. Grattan and R. Torrence, eds. Underthe shadow: The cultural impacts of volca-nic eruptions. Left Coast Press, WalnutCreek, California.


Collerson, K., and M. Weisler. 2007. Stoneadze composition and the extent of ancientPolynesian voyaging and trade. Science(Washington, D.C.) 28:1907–1911.

Crittenden, K., and K. Rodolfo. 2002. Ba-color town and Pinatubo Volcano, Philip-pines: Coping with recurrent lahardisaster. Pages 43–65 in R. Torrence andJ. Grattan, eds. Natural disasters and cul-tural change. Routledge, London.

Cronin, S. 2006. Impacts of gas and ash erup-tions on pastoral and subsistence agricul-ture. Pages 19–20 in Abstracts volume,Fourth Conference, Cities on Volcanoes,International Association for Volcanologyand Chemistry of the Earth’s Interior,Quito, Ecuador.

Cronin, S., and K. Cashman. 2007. Volcanicoral traditions in hazard assessment andmitigation. Pages 175–202 in J. Grattanand R. Torrence, eds. Under the shadow:The cultural impacts of volcanic eruptions.Left Coast Press, Walnut Creek, Califor-nia.

Cronin, S., and V. Neall. 2000. Impacts ofvolcanism on pre-European inhabitants ofTaveuni, Fiji. Bull. Volcanol. 62:199–213.

Cronin, S., K. Nemeth, and V. Neall. 2008.Volcanism and archaeology. Pages 2185–2196 in D. Pearsall, ed. Encyclopedia ofarchaeology. Vol. 3. Academic Press, NewYork.

Davies, H. 2002. Tsunamis and the coastalcommunities of Papua New Guinea. Pages28–42 in R. Torrence and J. Grattan, eds.Natural disasters and cultural change.Routledge, London.

Denevan, W. M. 2001. Cultivated landscapesof native Amazonia and the Andes. OxfordUniversity Press, Oxford.

———. 2004. Semi-intensive pre-Europeancultivation and the origins of anthropo-genic dark earths in Amazonia. Pages135–143 in B. Glaser and W. I. Woods,eds. Amazonian dark earths: Explorationsin space and time. Springer, London.

Dillian, C. 2007. Archaeology of fire andglass: Cultural adoption of Glass Moun-tain in obsidian. Pages 253–273 in J. Grat-tan and R. Torrence, eds. Under theshadow: The cultural impacts of volcanic

eruptions. Left Coast Press, WalnutCreek, California.

Driessen, J. 2002. Towards an archaeology ofcrisis: Defining the long-term impact ofthe Bronze Age Santorini eruption. Pages250–263 in R. Torrence and J. Grattan,eds. Natural disasters and cultural change.Routledge, London.

Elson, M., M. Ort, J. Hesse, and W. Duffield.2002. Lava, corn, and ritual in the north-ern Southwest. Am. Antiq. 67:119–135.

Enright, N., and C. Gosden. 1992. Unsta-ble archipelagos: South-west Pacific envi-ronment and prehistory since 30,000 B.P.Pages 160–198 in J. Dodson, ed. Thenaıve lands: Prehistory and environmentalchange in Australia and the Southwest Pa-cific. Longman, Melbourne.

Firth, R. 1959. Ritual adzes in Tikopia. Pages149–159 in J. Freeman and W. Geddes,eds. Anthropology in the South Seas.Thomas Avery and Sons, Plymouth.

Fullagar, R., G. Summerhayes, B. Ivuyo, andJ. Specht. 1991. Obsidian sources at Mo-pir, West New Britain Province, PapuaNew Guinea. Archaeol. Oceania 26:110–114.

Gaillard, J.-C., F. Delfin, E. Dizon, V. Paz, E.Ramos, C. Remotigue, K. Rodlfo, F. Sirin-gan, J. Soria, and J. Umbal. 2007. Plan-ning for the future: A multidisciplinaryapproach to reconstructing the Buag epi-sode of Mt. Pinatubo, Philippines. Pages225–252 in J. Grattan and R. Torrence,eds. Under the shadow: The cultural im-pacts of volcanic eruptions. Left CoastPress, Walnut Creek, California.

Galipaud, J.-C. 2002. Under the volcano: Ni-Vanuatu and their environment. Pages162–171 in R. Torrence and J. Grattan,eds. Natural disasters and cultural change.Routledge, London.

Garanger, J. 1972. Archeologie des Nouvelle-Hebrides: Contribution a la connaissancedes iles du center. ORSTOM, Paris.

Gosden, C., and J. Webb. 1994. The creationof a Papua New Guinean landscape: Ar-chaeological and geomorphological evi-dence. J. Field Archaeol. 21:29–51.

Grattan, J., M. Brayshay, and R. Schut-tenhlm. 2002. ‘‘The end is nigh’’? Social


and environmental responses to volcanicgas pollution. Pages 87–106 in R. Tor-rence and J. Grattan, eds. Natural disastersand cultural change. Routledge, London.

Grattan, J., and D. Gilbertson. 1994. Acid-loading from Icelandic tephra falling onacidified ecosystems as the key to set-tlement abandonment and environmentalstress in northern and western Britain. J.Archaeol. Sci. 21:851–859.

Grattan, J., S. Michnowicz, and R. Rabartin.2007. The long shadow: Understandingthe influence of the Laki Fissure eruptionon human mortality in Europe. Pages175–202 in J. Grattan and R. Torrence,eds. Under the shadow: The cultural im-pacts of volcanic eruptions. Left CoastPress, Walnut Creek, California.

Grattan, J., and R. Torrence. 2007. Beyondgloom and doom: The long-term conse-quences of volcanic disasters. Pages 1–18in J. Grattan and R. Torrence, eds. Underthe shadow: The cultural impacts of volca-nic eruptions. Left Coast Press, WalnutCreek, California.

Heming, R. F., and I. E. Smith. 1969. Noteson the thermal fields of Talasea, Pan-galu, and Kasiloi, New Britain, TPNG.Bur. Miner. Resour. Geol. Geophys. Rep.(Canberra) 1969/115.

Hoffmann, A., and P. Parsons. 1997. Extremeenvironmental change and evolution.Cambridge University Press, Cambridge.

Holmberg, K. 2007. Beyond the catastrophe:The volcanic landscape of Baru, westernPanama. Pages 274–298 in J. Grattan andR. Torrence, eds. Under the shadow: Thecultural impacts of volcanic eruptions. LeftCoast Press, Walnut Creek, California.

Jago, L., and W. Boyd. 2005. How a wettropical rainforest copes with repeated vol-canic destruction. Quat. Res. 63:399–406.

Johnson, L. L. 2002. Natural disasters andcultural change in the Shumagin Islands.Pages 193–203 in R. Torrence and J.Grattan, eds. Natural disasters and culturalchange. Routledge, London.

Johnson, R. W. 1976. Late Cainozoic volcan-ism and plate tectonics at the southernmargin of the Bismark Sea, Papua NewGuinea. Pages 101–116 in R. W. Johnson,

ed. Volcanism in Australia. Elsevier, Am-sterdam.

Kirch, P., and T. Hunt. 1997. Historical ecol-ogy in the Pacific islands: Prehistoric en-vironmental and landscape change. YaleUniversity Press, New Haven, Connecti-cut.

Kirch, P., and J. Kahn. 2007. Advances inPolynesian prehistory: A review and as-sessment of the last decade (1993–2004).J. Archaeol. Res. 15:191–238.

Kononenko, N. 2007. The contribution ofuse-wear/residue studies of obsidian forunderstanding changes in settlement andsubsistence patterns in West New Britain,Papua New Guinea. Bull. Indo-Pac. Pre-hist. Assoc. 27:135–143.

Lentfer, C., and W. Boyd. 2001. Mauntenpaia. Volcanoes, people and environment:The 1994 Rabaul eruption. SouthernCross University Press, Lismore.

Lentfer, C., and R. Torrence. 2007. Holo-cene volcanic activity, vegetation suc-cession and ancient human land use:Unraveling the interactions on Garua Is-land, Papua New Guinea, Rev. Palaeobot.Palynol. 143:83–105.

Lilley, I. 2004a. Diaspora and identity in ar-chaeology: Moving beyond the Black At-lantic. Pages 287–312 in L. Meskell andR. Preucel, eds. A companion to social ar-chaeology. Blackwell, Oxford.

———. 2004b. Trade and culture historyacross the Vitiaz Strait, Papua NewGuinea. Pages 89–96 in V. Attenbrow andR. Fullagar, eds. A Pacific odyssey: Ar-chaeology and anthropology in the west-ern Pacific. Papers in honour of JimSpecht. Rec. Aust. Mus. Suppl. 29.

Losey, R. J. 2005. Earthquakes and tsunamisas elements of environmental disturbanceon the Northwest Coast of North Amer-ica. J. Anthropol. Archaeol. 24:101–116.

Lowe, D. J., R. M. Newnham, and J. D.McCraw. 2002. Volcanism and earlyMaori society in New Zealand. Pages126–161 in R. Torrence and J. Grattan,eds. Natural disasters and cultural change.Routledge, London.

Machida, H., R. Blong, J. Specht, R. Tor-rence, H. Moriwaki, Y. Hayakawa, B.


Talai, D. Lolok, and C. Pain. 1996. Holo-cene explosive eruptions of Witori andDatataua caldera volcanoes in West NewBritain, Papua New Guinea. Quat. Int.34–36:65–78.

Machida, H., and S. Sugiyama. 2002. The im-pact of the Kikai-Akahoya explosive erup-tions on human societies. Pages 313–325in R. Torrence and J. Grattan, eds. Natur-al disasters and cultural change. Routledge,London.

Manning, S., and D. Sewell. 2002. Volcanoesand history: A significant relationship?The case of Santorini. Pages 264–291 inR. Torrence and J. Grattan, eds. Naturaldisasters and cultural change. Routledge,London.

Mastrolorenzo, G., D. Palladino, G. Vecchio,and J. Taddeucci. 2002. The 472 AD Pol-lena eruption of the Somma-Vesuvius(Italy) and its environmental impact at theend of the Roman Empire. J. Volcanol.Geotherm. Res. 113:19–36.

Mastrolorenzo, G., P. Petrone, L. Pap-palardo, and M. Sheridan. 2006. TheAvellino 3780-yr-B.P. catastrophe as aworst-case scenario for a future eruptionat Vesuvius. Proc. Natl. Acad. Sci. U.S.A.103:4366–4370.

McKee, C., H. Patia, J. Kuduon, and R. Tor-rence. 2005. Volcanic hazard assessmentof the Krummel-Garbuna-Welcker volca-nic complex, southern Willaumez Penin-sula, WNB, Papua New Guinea. Geol.Surv. Papua New Guinea Rep. 2005/4.

Neall, V. 1996. Hydrological disasters as-sociated with volcanoes. Pages 395–425in V. Singh, ed. Hydrology of disasters.Kluwer Academic Publishers, Dordrecht,The Netherlands.

Neall, V., R. Torrence, T. O’Neill, M.Turner, B. Hayward, and W. Boyd. 2007.Reconstructing the impacts of volcanicdisasters on Holocene paleoenvironmentsand human land-use: Willaumez isthmus,New Britain, Papua New Guinea. Quat.Int. 167–168:302–303.

Neall, V., R. Wallace, and R. Torrence. 2008.The volcanic environment for 40,000 yearsof human occupation on the Willaumezisthmus, West New Britain, Papua New

Guinea. J. Volcanol. Geotherm. Res. 176:330–343. doi:10.1016/j.jvolgeores.2008.01.037

Newhall, C., and S. Punongbayan, eds. 1996.Fire and mud: Eruptions and lahars ofMount Pinatubo, Philippines. Universityof Washington Press and Philippine Insti-tute of Volcanology and Seismology, Seat-tle and Quezon City.

Newhall, C., and S. Self. 1982. The volcanicexplosivity index (VEI): An estimate of ex-plosive magnitude for historical volcanism.J. Geophys. Res. 87:1231–1238.

Parr, J., C. Lentfer, and W. Boyd. 2001. Spa-tial analysis of phytolith assemblages at anarchaeological site in West New Britain,Papua New Guinea. Pages 125–134 in G.Clark, A. Anderson, T. Vunidilo, eds. Thearchaeology of Lapita dispersal in Oce-ania. Terra Australia 17. Pandanus Books,Canberra.

Pavlides, C. 2006. Life before Lapita: Newdevelopments in Melanesia’s long-termhistory. Pages 205–227 in I. Lilley, ed. Ar-chaeology of Oceania: Australia and thePacific islands. Blackwell, Oxford.

Petrie, C., and R. Torrence. 2008. Assessingthe effects of volcanic disasters on humansettlement in the Willaumez Peninsula,Papua New Guinea: A Bayesian approach.Holocene 18:729–744.

Plunket, P., and G. Urunuela. 1998. Preclas-sic household patterns preserved undervolcanic ash at Tetimpa, Puebla, Mexico.Lat. Am. Antiq. 9:287–309.

Pyle, D. 2000. Sizes of volcanic eruptions.Pages 263–269 in H. Sigurdsson, ed. En-cyclopedia of volcanoes. Academic Press,San Diego.

Rath, P., and R. Torrence. 2003. Produc-ing value: Stemmed tools from GaruaIsland, Papua New Guinea. Aust. Ar-chaeol. 57:119–127.

Saltonstall, P., and G. A. Carver. 2002. Earth-quakes, subsidence, prehistoric site attri-tion and the archaeological record: A viewfrom the Settlement Point site, KodiakArchipelago, Alaska. Pages 172–192 inR. Torrence and J. Grattan, eds. Naturaldisasters and cultural change. Routledge,London.

Sheets, P. 1992. The Ceren site: A prehistoric


village buried by volcanic ash in CentralAmerica. Harcourt Brace, Fort Worth.

———. 1999. The effects of explosive volca-nism on ancient egalitarian, ranked, andstratified societies in Middle America.Pages 36–58 in A. Oliver-Smith and S.Hoffman, eds. The angry earth: Disasterin anthropological perspective. Routledge,New York.

———, ed. 2002. Before the volcano erupted:The ancient Ceren village in CentralAmerica. University of Texas Press, Aus-tin.

———. 2007. People and volcanoes in theZapotitan Valley, El Salvador. Pages 67–89 in J. Grattan and R. Torrence, eds.Under the shadow: The cultural impactsof volcanic eruptions. Left Coast Press,Walnut Creek, California.

Sheets, P., J. Hoopes, W. Melson, B. McKee,T. Sever, M. Mueller, M. Cheanult, and J.Bradley. 1991. Prehistory and volcanism inthe Arenal area, Costa Rica. J. Field Ar-chaeol. 18:445–465.

Sheets, P., and B. McKee, eds. 1994. Archae-ology, volcanism, and remote sensing inthe Arenal region, Costa Rica. Universityof Texas Press, Austin.

Siebe, C., M. Abrams, J. Macıas, and J. Oben-holzner. 1996. Repeated volcanic disastersin prehispanic time at Popocatepetl, cen-tral Mexico: Past key to the future? Geol-ogy 24:399–402.

Sigurdsson, H., S. Carey, W. Cornell, and T.Pescatore. 1985. The eruption of Vesuviusin AD 79. Natl. Geogr. Res. 1:332–387.

Smithsonian Institution. 2005. Garbuna (Pa-pua New Guinea): First historically wit-nessed eruption in October 2005. Bull.Global Volcano Net. 30 (11).

Specht, J. 1981. Obsidian sources at Talasea,West New Britain, Papua New Guinea. J.Polynesian Soc. 90:337–356.

———. 2005. Stone axe blades and valuablesin New Britain, Papua New Guinea. Pages15–22 in C. Gross, H. Lyons, and D.Counts, eds. A polymath anthropologist:Essays in honour of Ann Chowning. Re-search in Anthropology and LinguisticsMonograph 6. Department of Anthropol-ogy, University of Auckland, Auckland.

———. 2007. Shaft-hole stone implementsof New Britain, Papua New Guinea. Pages263–275 in A. Anderson, K. Green, and F.Leach, eds. Vastly ingenious: The archae-ology of Pacific material culture. OtagoUniversity Press, Dunedin.

Specht, J., R. Fullagar, and R. Torrence.1991. What was the significance of Lapitapottery at Talasea? Bull. Indo-Pac. Prehist.Assoc. 11:281–294.

Specht, J., R. Fullagar, R. Torrence, and N.Baker. 1988. Prehistoric obsidian exchangein Melanesia: A perspective from the Tala-sea sources. Aust. Archaeol. 27:3–16.

Specht, J., and R. Torrence. 2007a. Lapita allover: Land-use on the Willaumez Penin-sula, Papua New Guinea. Pages 71–96 inS. Bedford, C. Sand, and S. Connaugh-ton, eds. Oceanic explorations: Lapita andwestern Pacific settlement. Terra Australis26. ANU E-Press, Canberra. http://epress.anu.edu.au/ta26_citation.html.

———. 2007b. Pottery of the Talasea area,West New Britain Province. In J. Spechtand V. Attenbrow, eds. Archaeologicalstudies of the middle and late Holo-cene, Papua New Guinea, Part IV.Tech. Rep. Aust. Mus. 20:131–196. www.australianmuseum.net.au/pdf/publications/1476_complete.pdf.

Spriggs, M. 1997. Landscape catastrophe andlandscape enhancement: Are either or bothtrue in the Pacific? Pages 80–104 in P.Kirch and T. Hunt, eds. Historical ecologyin the Pacific islands: Prehistoric environ-mental and landscape change. Yale Uni-versity Press, New Haven, Connecticut.

Steadman, D. 2006. Extinction and biogeog-raphy of tropical Pacific birds. Universityof Chicago Press, Chicago.

Summerhayes, G. 2007. Island Melanesianpasts: A view from archaeology. Pages 10–35 in J. Friedlaender, ed. Genes, languageand culture history in the Southwest Pa-cific. Oxford University Press, Oxford.

Summerhayes, G. R., J. R. Bird, R. Fullagar,C. Gosden, J. Specht, and R. Torrence.1998. Application of PIXE-PIGME toarchaeological analysis of changing pat-terns of obsidian use in West New Britain,Papua New Guinea. Pages 129–158 in


S. Shackley, ed. Archaeological obsidianstudies. Plenum, New York.

Thornton, I. 1996. Krakatau: The destructionand reassembly of an island ecosystem.Harvard University Press, Cambridge.

Torrence, R. 2002a. What makes a disaster?A long-term view of volcanic eruptionsand human responses in Papua NewGuinea. Pages 292–310 in R. Torrenceand J. Grattan, eds. Natural disasters andcultural change. Routledge, London.

———. 2002b. Cultural landscapes on GaruaIsland, PNG. Antiquity 76:766–776.

———. 2004a. Now you see it, now youdon’t: Changing obsidian source use inthe Willaumez Peninsula, Papua NewGuinea. Pages 115–125 in J. Cherry, C.Scarre, and S. Shennan, eds. Explainingsocial change: Studies in honour of ColinRenfrew. McDonald Institute for Archaeo-logical Research, Cambridge.

———. 2004b. Pre-Lapita valuables in IslandMelanesia. Pages 163–172 in V. Atten-brow and R. Fullagar, eds. A Pacific odys-sey: Archaeology and anthropology in thewestern Pacific. Papers in honour of JimSpecht. Rec. Aust. Mus. Suppl. 29.

———. 2005. Valued stone: How so? Pages357–372 in I. Macfarlane, R. Paton, andM. Mountain, eds. Many exchanges: Ar-chaeology, history, community and thework of Isabel McBryde. Aborig. Hist.Mono. 11. Australian National University,Canberra.

———. 2008. Punctuated landscapes: Crea-ting cultural places in volcanically activeenvironments. Pages 333–343 in B. Davidand J. Thomas, eds. Handbook of land-scape archaeology. Left Coast Press, Wal-nut Creek, California.

Torrence, R., R. Bonetti, A. Guglielmetti,A. Manzoni, and M. Oddone. 2004a. Im-portance of source availability and acces-sibility: A case study from Papua NewGuinea. Mediterr. Archaeol. Archaeom.4:53–65.

Torrence, R., and T. Doelman. 2007. Prob-lems of scale: Evaluating the effects ofvolcanic disasters on cultural change inthe Willaumez Peninsula, Papua NewGuinea. Pages 42–66 in J. Grattan and R.

Torrence, eds. Under the shadow: Thecultural impacts of volcanic eruptions.Left Coast Press, Walnut Creek, Califor-nia.

Torrence, R., and J. Grattan. 2002. Trends inthe archaeology of disasters. Pages 1–18 inR. Torrence and J. Grattan, eds. Naturaldisasters and cultural change, Routledge,London.

Torrence, R., V. Neall, T. Doelman, E.Rhodes, C. McKee, H. Davies, R. Bonetti,A. Guglielmetti, M. Manzoni, M. Oddone,J. Parr, and C. Wallace. 2004b. Pleistocenecolonization of the Bismarck Archipelago:New evidence from West New Britain.Archaeol. Oceania 39:101–130.

Torrence, R., C. Pavlides, P. Jackson, and J.Webb. 2000. Volcanic disasters and cul-tural discontinuities in the Holocene ofWest New Britain, Papua New Guinea.Geol. Soc. Lond. Spec. Mem. 171:225–244.

Torrence, R., J. Specht, and R. Fullagar.1990. Pompeiis in the Pacific. Aust. Nat.Hist. 23:457–463.

Torrence, R., J. Specht, R. Fullagar, and J. R.Bird. 1992. From Pleistocene to present:Obsidian sources in West New BritainProvince, Papua New Guinea. Rec. Aust.Mus. 42:83–98.

Torrence, R., J. Specht, R. Fullagar, and G.Summerhayes. 1996. Which obsidian isworth it? Pages 211–224 in J. M. David-son, G. Irwin, B. F. Leach, A. Pawley, andD. Brown, eds. Oceanic culture history: Es-says in honour of Roger Green. New Zea-land Archaeological Association, Dunedin.

Torrence, R., and P. Swadling. 2008. Socialnetworks and the spread of Lapita? Anti-quity 82:600–616.

Torrence, R., P. Swadling, W. Ambrose, N.Kononenko, P. Rath, and M. Glascock.2009. Obsidian stemmed tools and Mid-Holocene interaction. Asian Perspect.48:118–147.

Torrence, R., and J. Webb. 1992. Report onarchaeological research on Garua Island,West New Britain Province, PNG. July–August 1992. Available at National Re-search Institute, Port Moresby, PapuaNew Guinea.


Turner, M., and V. Dale. 1998. Comparinglarge, infrequent disturbances: What havewe learned? Ecosystems 1:493–496.

Vanderhoek, R., and R. Nelson. 2007. Eco-logical roadblocks on a constrained land-scape: The cultural effects of catastrophicHolocene volcanism on the Alaska Penin-sula, Southwest Alaska. Pages 153–174 inJ. Grattan and R. Torrence, eds. Underthe shadow: The cultural impacts of vol-canic eruptions. Left Coast Press, WalnutCreek, California.

Wallace, L. M., C. Stevens, E. Silver, R.McCaffrey, W. Loratung, S. Hasiata, R.Stanaway, R. Curley, R. Rosa, and J.Taugaloidi. 2004. GPS and seismologicalconstraints on active tectonics and arc-continent collision in Papua New Guinea:Implications for mechanics of microplaterotations in a plate boundary zone. J.Geophys. Res., Earth Sci. (online: 109,B05404. doi: 10.1029/2003JB002481).

White, J. P. 1996. Rocks in the head: Think-ing about the distribution of obsidian in

Near Oceania. Pages 199–209 in J. M. Da-vidson, G. Irwin, B. F. Leach, A. Pawley,and D. Brown, eds. Oceanic culture his-tory: Essays in honour of Roger Green.New Zealand Archaeological Association,Dunedin.

White, J., C. Coroneos, V. Neall, W. Boyd,and R. Torrence. 2002. FEA site, BodunaIsland: Further investigations. Pages 101–107 in S. Bedford, D. Burley, and C.Sand, eds. Fifty years in the field. N. Z.Archaeol. Assoc. Mongr. 25, Auckland.

Woodhead, J. D., S. E. Eggins, and R. W.Johnson. 1998. Magma genesis in theNew Britain island arc: Further insightsinto melting and mass transfer processes.J. Petrol. 39:1641–1668.

Zeidler, J., and J. Isaacson. 2003. Settlementprocess and historical contingency in thewestern Ecuadorian Formative. Pages69–123 in J. Raymond and R. Burger,eds. Archaeology of Formative Ecuador.Dumbarton Oaks Research Library andCollection, Washington, D.C.


volcanism and historical ecology on the willaumez

Documents