CHAPTER 2 THE MUSSAU ISLANDS: NATURAL AND CULTURAL ENVIRONMENTS PATRICK V. KIRCH WITH CARLA CATTERALL * Lying just S of the equator, amidst some of the greatest biogeographical diversity anywhere in the vast Indo- 1 Ej Pacific, the Mussau Islands offered a = ~. remarkable range of habitats, mi- = croenvironments, and resources to their indigenous inhabitants. My aim here is to summarize the salient aspects of this envi- ronmental canvas from the perspective of human ecology. Attention will also be paid to features of the environment with direct relevance to archaeo- logical interpretation. In the second half of the chapter I turn as well to the ethnographic record for Mussau- imperfectly as it is known-for insights this may pro- vide to the prehistorian seeking to recover the longue duree of human history in these islands. GEOGRAPHICAL OVERVIEW The Mussau Islands, part of the St. Matthias Group which also includes Emira (also known as E Mira or Emirau) to the SE (and sometimes includes isolated, tiny Tench Island as well),I are situated at approximately 1400 40' E longitude and 1? 25' S latitude. They form the N point along the arc of the Bismarck Archipelago, and are some 150 km NW of New Hanover Island (Lavongai). The Manus (Admiralty Island) group lies - 200-250 km to the SW. On unusually clear days it is possible to see the high peak of New Hanover from higher elevations on Mussau, but Mussau and Manus are not intervisible. The Mussau group proper (Fig. 2.1, 2.2) is domi- nated by a single large high island, bearing the same name, and by 1 1 other small islands and islets that form a cluster to the SW of the main island (Table 2.1). The largest of these two offshore islands are Eloaua and Emananus (Fig. 2.3), which together with diminutive Eunakuru and Enusagila, make up an atoll-like forma- tion enclosing a lagoon.2 Eloaua and Emananus are both permanently inhabited, and seem to have been continu- ously occupied since the early Lapita period (- 3500 BP). This lengthy period of occupation is reflected in various aspects of their natural history, such as the ex- tent of secondary growth and orchard gardens, and the development of anthropogenic soils on the low-lying calcareous flats. The only airstrip in the group is on Eloaua, an 800-m long packed coral runway constructed by the SDA mission in the early 1970s, capable of land- THE M9 M U S S A U, 2. 65 -E,Mt. Taleanu i S L A N DD - - Roads ,- -Rock ledge or reef awash * Village # 200 meter contour interval 0 1 2 3 4 5 km I I I I I I FIGURE2.1 Map of the Mussau Islands, based on the PNG Topographic Survey 1100,000 map (Sheet 8894, prepared 1980). 29 THE ML 0 0 - DC 0 DC u cn E 0 a) Dc tn a D- a) 0 0 cC~ LX Gl) THE MUSSAU ISLANDS TABLE 2.1 Geographic characteristics of the Mussau Islands. ISLAND NAME GEOLOGICAL TYPE Mussau High island, volcanic core with elevated limestone terraces Eloaua Elevated limestone (makatea) with peripheral unconsolidated sand beach terraces Emananus Elevated limestone (makatea) with peripheral unconsolidated sand beach terraces Boliu Elevated limestone (makatea) with peripheral unconsolidated sand beach terraces Enoanaulu Elevated limestone (makatea) Ebolo Unconsolidated sand cay Emussau Unconsolidated sand cay, possibly on minimally uplifted limestone core Ekaleu Unconsolidated sand cay, possibly on minimally uplifted limestone core Ebanalu Elevated limestone (makatea) Tapatu Unconsolidated sand cay, possibly with minimally uplifted limestone core Eunakuru Unconsolidated sand cay Enusagila Unconsolidated sand cay SIZE MAXIMUM (KM2) ELEVATION (M) 348.0 650 7.5 48 4.0 45 1.0 26 1.0 0.5 0.4 0.3 0.25 0.3 0.3 0.05 58 1.5 2 3 1 7 6 1.5 2 ing small aircraft (4 or 6-seaters) in good weather.3 A deep channel (the Malle Channel) lies between the Eloaua-Emananus "atoll" and the other offshore islets, most of which sit amongst a complex of reefs and lagoons, rich in fish and shellfish resources (Fig. 2.3). Ekaleu Island, at the NW end of the Malle Channel, was the setting for an early German coconut plantation, and is still covered in tall coconut palms, with an aban- doned European-style house that sits atop a large, late prehistoric shell midden.4 Ebanalu, Enoanaulu, and Boliu islands all have small hamlets and gardens, whereas Emussau, Ebolo, and Tapatu are unoccupied. Emussau, however, has a late prehistoric midden site which we test excavated (Site EKS, see Chapter 7). The modern population of the Mussau group num- bers less than 2,000 persons, and is concentrated at sev- eral places on Mussau Island proper, as well as on Eloaua and Emananus. The administrative center is at Palakau at the head of Schadel Bay, and the SDA mission school is located at Boliu on the opposite side of the bay. There is a concrete wharf at Palakau but no airstrip, so that persons flying in or out of the islands must commute between Palakau and Eloaua by canoe or boat. Rough, unpaved roads connect Palakau to communities at Tavol, Etasitel, Roitano, and Lomakanauru on the E and S coasts, and to the logging mill at Alamul on the SW. More remote settlements, such as Tanaliu and Etalat in the N part of Mussau can be reached only on foot or by canoe. There are no roads at all into the interior. NATURAL ENVIRONMENT AND RESOURCES When one has been accustomed to working in other parts of the Pacific-such as Hawai'i or the Society Is- lands-where there is a long tradition of natural sci- ence investigations and hence a rich scholarly literature, it can be exceedingly frustrating to begin work in a group of islands like Mussau, for which the number of natu- ral science citations can be counted on the fingers of one hand. Aside from ourselves, only two organized sci- entific expeditions have ever worked at Mussau, the fa- mous Siidsee Expedition of the Hamburgische Wissenschaftliche Stiftung in 1908 (Thilenius 1927), and the Danish Noona Dan Expedition in 1962 (Wolff 1966), each for relatively short periods. The Siidsee Expedi- tion concentrated largely on ethnography (Nevermann 1933), although they did make some geographical and geological observations, while the Noona Dan Expedi- tion made botanical and zoological collections. As far 31 THE ML FIGURE 2t3 Aerial view of the NW part of Emananus Island, showing one of several lagoonal basins, flanked by coastal mangroves. as I have been able to determine, however, the latter collections have never been fully reported, or analyzed for their biogeographical significance. The only other scientific collecting on Mussau was for birds, by A. F. Eichhorn under the auspices of the great amateur orni- thologist Lord Rothschild, the specimens being worked up by Hartert (1924). There has been no systematic study of the geology, geomorphology, climate, soils, flora, fauna, vegetation patterns, or biogeography of Mussau. Aware of this dearth of geographical and environmen- tal data for the islands, we made many observations of our own during the three field seasons, collecting her- barium voucher specimens of the common secondary forest plants on Eloaua (159 specimens were deposited with the Herbarium Pacificum of the Bernice P. Bishop Museum), obtaining rock samples, and carrying out our own marine ecological survey and mapping (under the direction of Dr. Carla Catterall). The notes that follow are made from these observations, combined with what can be gleaned from more general surveys of PNG and Melanesian natural history. BASAL GEOLOGY Structurally, Mussau lies astride the Emirau-Feni Ridge, the N face of which plunges steeply down into the 6,000 + m deep Manus-Kilinailau Trench (Exon et al. 1986, fig. 2). It is thus on the N sector of a major geo- physical province, sometimes labeled the New Ireland Basin, an "arcuate feature extending 900 km northwest- ward from the Feni Islands in the east to Manus Island in the west" (Exon et al. 1986:39). The basin has a com- plex stratigraphy and geological history commencing with Eocene/early Miocene volcanics, and continuing with the deposition of late Miocene/Pliocene carbon- ates (in some areas with renewed and continuing volcanics). While there are no published studies of Mussau geology, what little information we have is con- sistent with this general New Ireland Basin sequence, and some inferences may be drawn from the limited geological data available for New Ireland, New Hanover, and Manus (Brown 1982; Exon et al. 1986; Francis 1988; Hohnen 1978;Jaques 1980; Stewart and Sandy 1986). The only extant geological map of Mussau is the 1:1,000,000 scale Geology ofPapua New Guinea (1972) published by the Department of Lands, Surveys, and Mines, Port Moresby. At this greatly reduced scale, only two formations are indicated for the Mussau group. The interior portions of Mussau Island proper are shown as consisting of unspecified volcanics of upper Miocene/ Pliocene age (map unit Tm-p ), whereas the outer fringe of Mussau and the smaller offshore islets such as Eloaua E Mt and Emananus are depicted as limestones, marl, and raised coral reefs of Quaternary age (map unit Qs'). This highly generalized picture accords with our own field observations made during reconnaissance surveys to Mussau Island in 1986 and 1988. It was evident to us that, especially on the S, W, and N, Mussau has an exten- sive "apron" of carbonates which overlie the volcanic core. Along the E coast, a number of streams have penetrated this volcanic core, and we were able to col- lect nine samples of igneous rock from these stream beds, derived from the island's interior. These samples fall into two distinct groups: (1) a generically related suite of plutonic (intrusive) diorite-gabbro; and (2) much finer-grained igneous rocks of probable volcanic and hypabyssal origin, with geochemistry of mafic andesite or basalt, and aphanitic textures. The gabbroic plutonic rocks presumably represent the deeply eroded plutonic roots of a local volcanic edifice. In the S part of Mussau Island (for example, the area inland of Lomakunauru) the massive limestone for- mation rises in a series of relatively level plateaux, with abrupt escarpments separating them (Figure 2.5). This suggests that the limestones are of biogenic reefal orn- gin, with the scarps representing different uplift inter- vals. The main terraces have elevations of approximately 40, 120, and 200 m asl. The offshore islands such as Eloaua are entirely devoid of volcanics, an important point for archaeo- logical investigations, since any volcanic stones (worked or natural) found in our excavations are clearly imports to these islands, even if only from Mussau Island. (A detailed petrographic analysis of manuports from the Lapita sites will be provided in Volume III.) Most of these islands, including Eloaua, Emananus, and Boliu, consist of elevated reef limestones and conglomerates. On Eloaua, at least two major uplift terraces are present, one at an elevation of - 26 m and a higher one (which forms the central, cultivated plateau) at - 46 m asl. These have not been dated, but corals in growth position were evident in various exposures, and I would presume them to be of relatively late Pleistocene age. On other islands, such as Emussau, Ebolo, and Ekaleu, it is uncertain whether the low-lying (- 1.5-3 m asl) unconsolidated marine sands that make up the islands are anchored to an underlying block of limestone, or whether they have merely formed on an exposed reef platform. All of the offshore islands show abundant geo- morphological evidence for a mid-Holocene sea level FIGURE 2.4 The view from Eloaua Island across the Malle Channel, with the large island of Mussau in the distance. Part of Boliu Island is visible to the right. John Male stands near a small dugout canoe, while the stilt posts of a former house protrude from the tidal flats. 34 -IGURE 2.5 Aerial view of the S part of Mussau Island, in the vicinity of Lomakunauru Village, where the terrain ^onsists of a series of elevated limestone terraces. Emussau and Eloaua islands are visible in the left-hand distance. stand at between + 1-1.5 m, in the form of wave-cut, solution notches, and in the deposition of unconsoli- dated calcareous sand aprons and beach terraces at - 2 m asl (current beach terraces, as on Eloaua Island, are only - 1 m asl). Both Eloaua and Boliu islands have narrow "necks" of unconsolidated Holocene sands con- necting blocks of elevated limestone, in each case join- ing what had previously been two separate islets during the former mid-Holocene high-stand. To summarize, the Mussau Group probably origi- nated during the Tertiary through volcanic island-arc formation, as part of the same regional tectonics that formed New Ireland and Manus. The Miocene/Pliocene volcanics were subsequently capped with extensive sub- marine deposits of Quaternary limestones, marls, and conglomerates, which were then elevated through tec- tonic uplift in the late Quaternary. This uplift progres- sion advanced in a series of discrete stages, creating a "stairstep" topography in the S part of Mussau, and on the offshore islets. These latter were the last to be el- evated, presumably during the late Pleistocene. We saw no evidence for Holocene tectonic activity, but geomor- phological indicators of a mid-Holocene (6-3 kyr) higher sea level stand of - + 1.5 m are very clear (see below). This is interpreted as a local reflection of the widespread mid-Holocene high stand thought to have resulted from a combination of eustatic and/or geoidal changes (Dickinson, in press). For the archaeologist, two points are of significance: (1) the limited range of volcanic rocks present in the group, and their restriction to the interior of Mussau Island; and (2) the mid-Holocene sea level changes which had important consequences for shoreline progradation and beach terrace forma- tion in the past 3-4 kyr. CL IATE Situated just 1025' S of the equator, Mussau falls within the Inter-Tropical Convergence Zone @TCZ), probably the most important influence on its climate. The ITCZ is "an organized zone of high cloudiness and rainfall containing deep convective (cumulonimbus) clouds" (McAlpine et al. 1983:15). Thus, while there is limited seasonality with a NW monsoon season from mid-No- vember to the end of March, and a SE trades season from mid-May until the end of September (Reynolds 1972), seasonality is not nearly so pronounced here as in parts of the Bismarck Archipelago situated farther S. The days are hot and humid, the nights only marginally cooler. Afternoon thunderstorms and heavy cloudbursts - ML, THE MUSSAU ISLANDS are common, as the cumulonimbus clouds rise to fre- quently spectacular heights during midday. McAlpine et al. (1983:150-61) propose a climatic classification specific to the PNG area, in which they define eleven distinct climates. Mussau falls within their "Lowland Climate 3" zone, which is "lowland humid." Key aspects of this climate are a mean annual rainfall of - 2,700 mm (range 2,000-3,500 mm), mean annual temperature range of 23-300 C, and a net annual water surplus (precipitation/evaporation ratio 1-2). No specific meteorological data are available for Mussau, but the general climatic conditions of the group may be inferred from other stations in the general re- gion. The two meteorological stations geographically closest to Mussau, both of which fall within the same overall "lowland humid" climatic classification, are Kavieng (New Ireland) and Momote (Manus). For the "standard period" from 1956-70, Kavieng had a mean annual rainfall of 3,282 MM,8 while Momote had a mean of 3,382 mm. Monthly figures at both stations show relatively little seasonal variability (all data are as reported in McAlpine et al. 1983, appendix). Mean annual evapo- ration is 1,369 mm at Kavieng, and 1,673 mm at Momote, indicating a substantial water surplus at both stations. Temperature data are not available for Kavieng, but at Momote the annual mean is 27.2, with mean maximum of 29.9 and mean minimum of 24.40 C. Rela- tive humidity averages 75% at Kavieng, and 77% at Momote. At Momote, mean daily recorded sunshine averages between 4.0 (March) and 6.1 (September) hours per day, indicative of the typically cloudy conditions within the ITCZ. From these data, we may infer that Mussau also has an annual rainfall in excess of 3,000 mm, high positive water balance, mean temperature in the high 200s C, and high relative humidity (roughly 75%). From the perspec- tive of human inhabitants who practice tropical horti- culture, these conditions are virtually ideal for the culti- vation of a wide range of root, tuber, and tree crops, which indeed thrive in Mussau. The limiting factors for horticulture are not climatic (although occasional droughts can cause some damage to gardens9), so much as edaphic and hydrologic. SOILSAND OTHER L4NDFORM FEATURES As with other aspects of Mussau natural environment, there has been no professional study or survey of the islands' soils, and I can only make a few general state- ments based on our limited observations combined with what has been written of soil development in similar climatic and geological conditions elsewhere in PNG (Bleeker 1983). In Mussau, human habitation and land use (primarily for gardening) is confined to the coastal areas and elevated limestone plateaux, so that the un- derlying parent material for soil development consists of reefal limestones, marl, and carbonate conglomer- ates of varying ages and degrees of weathering. Our observations were that soils were relatively shallow, par- ent limestone often lying only 50 cm or even less be- neath the surface, with frequent limestone outcrops. The dominant soils are most probably rendolls, a widespread type of mollisol which forms on limestone in a wet, humid climate with low to moderate seasonality (Bleeker 1983:1 12-14).10 Such soils have "moderate to high fer- tility levels" with relatively moderate to high amounts of available nitrogen and phosphorus, although potas- sium may be more limited (1983:113). On the Eloaua plateau, which is extensively used for swidden garden- ing (with the dominant crops being taro, yam, and sweet potato), these soils seem to produce good yields, but the frequent occurrence of limestone outcrops is a draw- back. A second category of soils is restricted in its distri- bution to the unconsolidated calcareous sand terraces fringing the offshore islands and parts of Mussau Is- land. These soils are anthropogenic, consisting of midden deposits rich in organic materials such as char- coal, shellfish, bone, and other detritus of human occu- pation, with a calcareous sand matrix.'1 The middens which have produced these soils are of both Lapita (as at site ECA) and post-Lapita ages (as at sites EKE and EHN). Well drained and organically enriched, they are prime soils for the cultivation of such root crops as Colocasia andAlocasia aroids, and for introduced manioc (Manihot esculenta). Consequently, they have been exten- sively gardened, and the mechanical action of digging sticks used to make planting holes and to harvest tubers has thoroughly churned the upper 30-40 cm of these soils, with obvious negative consequences for the ar- chaeological record. The geomorphology of the Mussau Islands is strongly controlled by the basal geological structure. A radial drainage pattern dissects the volcanic interior of the main island, and some of the larger streams con- tinue through the surrounding limestones to reach the 35 THE MUSSAU ISLANDS sea, as at the head of Schadel Bay, and along the E coast. In other areas such as the SE tip of Mussau, and on all of the offshore islands, there is no surface drain- age at all, the limestone karst absorbing all rainfall like a sponge. In these areas, water for drinking, cooking, and bathing must be obtained either by rain catchment, or by digging shallow wells near the shoreline to tap the Ghyben-Herzberg aquifer of slightly brackish water (see Wiens 1962:317-26 on such aquifers).12 FLoRA AND VEGETATION PATTERNS The flora of New Guinea and the Bismarck Archipelago is essentially an extension of the East Malesian floristic zone (van Steenis 1950; van Balgooy 1971), a region ex- ceedingly rich in biodiversity. New Guinea itself has been estimated to have on the order of 1,465 plant genera (Paijmans 1976), although not all of these would be present throughout the Bismarcks, or on Mussau. Cer- tainly, however, for islands of their size the Mussau group are floristically rich, immediately evident to some- one such as myself who has had prior experience col- lecting plants in the tropical central Pacific. Compared with Niuatoputapu or Futuna in Western Polynesia, where I had made reasonably comprehensive botanical collections as part of my ethnoecological fieldwork, the diversity of plant taxa in Mussau was initially almost overwhelming. There is, of course, no guide to the flora of Mussau, and indeed for the entire Bismarck Archi- pelago the best source is Peekel's Flora of the Bismarck ArchipelagoforNaturalists (Illustrierte Flora des Bismarck- ArchipelsfiirNaturfreunde), authored between 1895 and 1945 by this German priest, but only translated from the manuscript and published recently (Peekel 1984). Peekel worked primarily on New Ireland, but most of the taxa he records are also present in Mussau, and his Flora was an invaluable asset during our 1986 and 1988 expeditions. My prior familiarity with the widespread coastal flora common to the Solomon Islands and West- ern Polynesia, assisted by Peekel's Flora, and where nec- essary by collecting and pressing voucher specimens, allowed us to discern the main vegetation associations present on the offshore islands such as Eloaua and Emananus, which are described below.13 The montane vegetation of the interior of Mussau Island is an en- tirely different matter, one best left to botanical special- ists. VEGETATION PATTERNS OF THE OFFSHORE ISLANDS Figures 2.6 and 2.7 are vegetation maps of Eloaua and Boliu islands, prepared in the field in 1988 by Dana Lepofsky of our project. They are representative of the spatial distribution patterns of the various vegeta- tion associations discussed below. Strand Forest and Littoral Vegetation The immediate strand vegetation of Mussau consists largely of widespread halophytic taxa, many of them pioneering species with floating seeds that are readily dispersed by ocean currents. Growing over beach sand itself one finds the beach morning-glory (Ipomoeapes- caprae), and immediately above the high-water mark such shrubs as Pandanus tectorius, Scaevolafrutescens, and Messerschmidia argentea, as well as various grasses. On low- lying sand cays such as Eunakuru and Enusagila, there are thickets of Pemphis acidula. In many areas, the littoral forest literally overhangs the ocean, the limbs of great Calophyllum inophyllum trees-often festooned with Dendrobium orchids and other epiphytes-shading the water's edge. Along with Calophyllum (whose wood is very important for carving canoes, bowls, and other implements), other common littoral trees are Bamrngtonia asiatica (used for fish poi- son), Terminalia wappa, T samoensis,Pandanustectoius,P. obtusifolia,Hemandiaozigera,Eythrinawriegata, Theia populnea, Cerbera manghas,Ficussp., Guettardaspeciosa, and Cordia subcordata. Hibiscus tiliaceous also grows densely in some areas just behind the strand. In some areas iron- woodtrees (Casuarina equisetifolia) are common, and may form monospecific stands (noted as "ironwood forest" on the maps). Coconut palms (Cocos nucifera) are also frequent immediately along and inland of the strand, in some places as planted groves. Mangrove Swamps Saline mangrove swamps occur at places around the coast of Eloaua, Emananus, Boliu and Ebolo islands, and completely cover Tapatu Island. They are also very extensive around the perimeter of Schadel Bay on Mussau Island, at places nearly 1 km wide.14 Both Rhizophoraapicdata and Brugueragymnorhiza are present, as is Xylocarpus moluccensis, and Cenops tagal. Thickets of Acrostichum fern occur on the inland margins of these swamps where Guettarda trees are found; Nipa palms 36 THE Mt, 2.6 Vegetation map of Eloaua Island, prepared by Dana Lepofsky. are also common in places. These swamps are the habi- tat for particular species of mollusks (e.g., Telescopium, Cerithidea) and for certain crustacea. It is probable that the area in mangroves is now more extensive than it was at the time of the mid-Holocene higher sea-level stand. The Lapita site of EHB on Emananus, and the later prehistoric midden site of Elunguai (EHN) on Eloaua, both have mangrove swamps separating them from the sea, but were probably situated on open beaches at the time they were occupied. Remnant Primary F7orest Little of this association remains on the offshore is- lands, due to several millennia of intensive land use. On Eloaua and Boliu, small areas cling to the steep lime- stone escarpments or in areas where there is no soil (as at the SE tip of Eloaua). Calophyllum sp., Syziginum sp., and Intsia bijuga were dominants, but some Porzetia, Terminalia, and Pandanus dubius were also noted. Late Second Groit/rh These areas consist of mature trees estimated at 40-60 years old. Again, due to intensive land use, there are only small areas with this association. Among the taxa noted here are Calophyllum sp., Intsia bijuga, Syzrgium sp., Pometiapinnata, Ficus spp., Pandanusdubius, and others for which we were only able to obtain native Mussau names. Arborncl/tura! Zones On the offshore islands, substantial tracts on the low- lying calcareous beach terraces have been planted to a range of tree crops, and thus form an anthropogenic vegetation association, an "arboricultural zone." Among the domi- nant trees planted in these zones are breadfruit ( tocarpus altil/s), coconut (Cocos nucifera), the tropical almond (Canancum indicum), Tahitianchestnut (Inocarpusfagiferus), vi-apple (Spondiasdulcis), several species of Pandanus, the Malay-apple (Syzigcum sp.), and many others (Lepofsky 1992). Human habitations are frequently dispersed within 3 7 - ME FIGURE 2.7 Vegetation map of Boliu Island, prepared by Dana Lepofsky. this zone as well (Fig. 2.8). The extensive macroscopic plant remains preserved at the ECA Site on Eloaua re- veal that such arboriculture has been practiced in these islands since at least the initial Lapita period. A full dis- cussion of this topic is deferred to Volume II. Swamp Taro tand Sago Cultiv.7tions In a few low-lying, naturally swampy areas on Eloaua Island, plantings of giant swamp taro (Cyrtosperma chamissonis) and/or sago palm (Metroxylon sp.) have been made. Swldden Cultlv,ations and Second Growth On both Eloaua and Boliu, the major parts of the el- evated limestone plateaux on which the best soils are located are extensively used for shifting cultivations (Fig. 2.9). New gardens are cleared annually; individual fam- ily plots are delineated by logs which are laid out in a rectangular grid. Today the primary crops are dryland taro (Colocasia esculenta) and manioc (Manihotesculenta), but it is certain that prior to the historic-period intro- duction of manioc, yams (Dioscorea spp.) would have been the other important crop. Indeed, a few Dioscorea alata plants are still grown, and feral D. bulbifera andD. nummularia are common throughout the second growth. A minor crop is sugarcane (Saccharum officinarum). Also common throughout this zone is Cordylinefruticosum, the leaves of which are used for parceling food when cooked in the earth oven, and the roots of which are occasion- ally cooked as food. Morinda citrifolia plants are also fre- quent although they do not seem to be purposively cul- tivated. The young second growth in which gardens are repeatedly cut includesMacaranga sp. and much Panda- nus tectornus, along with other species we did not identify. Mangrove A House/village f Primary forest I Late second growth c Shoreline vegetation 4 Second growth and gardens * Arboriculture zone and/or coconut grove * Ironwood forest - | THE ML FIGURE 2.8 Aerial view of the NW coast of Emananus Island, showing habitations dispersed among coconut palms and other economic trees of the arboricultural zone. TERRESTRIAL FAUNAL RESOURCES Without doubt, the most important category of terres- trial (in the broad sense of non-marine) resources in Mussau were birds. Hartert (1924) enumerated 39 spe- cies present in Mussau, based on the collections ob- tained by A. F. Eichhorn for Lord Rothschild, but this can only be a partial list, given the tremendous avifaunal diversity of the Bismarck Archipelago. Indeed, our ar- chaeological faunal assemblages added four species to Hartert's list (Steadman and Kirch 1998). Among the species known to have been taken by the prehistoric inhabitants of the offshore islands (for food, feathers, bone, or other uses) are petrels (Pterodronia sp.), the Brown Booby (Sula leucogaster), Osprey (Pandion haliaetus), Golden Plover (Pluvalis dominica), several species of larids (Sternafoscata, S. hirundo,Anousstolidus,A. minutus), fruit doves and pigeons (Ptilinopus spp., Ducula sp., Caloenas nicobarica), a cockatoo (Cacatus sp.), barn owl (Tyto sp.), and the Glossy Starling (Aplonos metallica). The domestic chicken (Gallus gallus) is also represented in our archaeofaunal collections. A full account of the bird bones recovered from the Mussau sites has been published elsewhere (Steadman and Kirch 1998). Aside from the avifauna, terrestrial animals which would have been of food value to the indigenous in- habitants are more restricted than in New Guinea or the larger islands of the Bismarcks. There are several species of fruit bat or "flying fox" (Pteropus spp.) which were doubtless hunted and provide a sweet meat, as well as fur that is sometimes used for ornamentation. Other taxa of true bats are also present (e.g., the genera Hipposideros, Dobsonia, Nyctimene), but their value as food resources is doubtful. The only marsupial present is the Spotted Cuscus (Phalanger maculatus), which attains a length of - 1 m, including its long tail. Based on evi- dence from New Ireland cave excavations, this is prob- ably a prehistoric introduction into the Bismarcks (Flannery and White 1991). The rat fauna of Mussau is entirely unknown, although some larger species of Rat- tus are certainly present, along with the Pacific Rat (R. exulans). The herpetofauna is also not adequately re- corded, although we observed that various lizards are common (including varanids [Varanidae], geckos [Gekkonidae], skinks [Scincidae], and scale-footed liz- ards [Pygopodidae]), as are snakes (families probably represented in Mussau include Typhlopidae, Boidae, Colubridae, Elapidae, and Hydrophiidae). Frog families likely to be represented include Hylidae and Ranidae. Several species of terrestrial crustacea are quite com- mon in the littoral forests. The hermit crab (Coenobzta cf. olliveri) inhabits dead gastropod shells, and larger in- dividuals are especially fond of making their homes in the shells of Turbo spp.; these hermit crabs are good bait for hook-and-line fishing. Large numbers of land crabs in the genus Cardisoma, which grow to have claw- to-claw spans of 20 cm or more, patrol the under- growth, scurrying about from their subterranean bur- rows. They were said to have been eaten prior to the conversion of the Mussau people to Seventh Day Ad- ventism, but their main relevance to archaeology is as bioturbators (Specht 1985). The Cardisoma burrows are easily 10 cm in diameter, and we found that these pen- -ML FIGURE 2.9 Young taro (Colocasia esculenta) plants growing in a swidden garden on the elevated limestone plateau Df Eloaua Island. etrate as much as 80-100 cm bs, although they do not go below the water table. When prospecting for buried midden deposits in the coastal beach terraces, one is well advised to examine their burrow tips, which fre- quently contain sherds, obsidian flakes, or other artifacts. Most notable of the terrestrial crustacea, however, is the noble Birgus latro, or Coconut Robber Crab. This formidable creature with massive claws and a leg-to-leg span of up to 50 cm is nocturnal, and feeds on coconut meat which it obtains by husking and breaking open the coconut shells. Prized for its succulent flesh, it is rare or endangered in many parts of the Pacific." Owing to the SDA dietary restrictions, the Mussau populations of B. latro have not been subject to human predation in recent years (other than by hungry archaeologists); the abandoned coconut plantation island of Ekaleu is par- ticularly overrun with them. That the Lapita people en- joyed them as well is indicated by carapace fragments recovered from ECA and other sites. MARINE AND LITTORAA ENVIRONMENTAND RESOURCES In Mussau as elsewhere in much of the SW Pacific, the sea was always far more important than the land as a source of animal protein. The offshore islands, and parts of Mussau Island, are surrounded by extensive reefs and lagoons which are prime habitats for a diversity of fish and shellfish, and the open seas and channels can be trolled for game fish. Mussau sits within the zone of highest biogeographic diversity within the vast Indo- Pacific faunal province (Stoddart 1992), and although no detailed inventories have been prepared, one can safely assume that there are several thousand species of edible fish, rays, eels, sharks, mollusks, and crustacea in the Mussau waters. Both the Green Sea Turtle (Chelonia mydas) and the Hawksbill Turtle (Eretmochelys ombnicata) occur in these waters, and there are several genera of porpoises (including Grampus, Tursiops, and Delphints). A THE MUSSAU ISLANDS more extensive discussion of fish and shellfish resources is deferred to Volume II, in which the archaeological faunal materials are analyzed. The dense mangrove swamps rimming Schadel Bay are the prime habitat of Crocodylusporosus, the Estuarine Crocodile, labeled by an expert in herpetology as a "man-eating monster" (Loveridge 1946:45). Adults range in length from 5 up to 10 m, making them easily ca- pable of overpowering "the average man or woman when taken unawares" (1946:46). I was given several accounts of crocodile attacks by my Eloaua friends, with sufficient detail of circumstance to leave no doubt as to their veracity. Curiously, although these crocodiles are clearly substantial "meat packages" whose flesh is quite palatable,'7 not a single crocodile bone was recov- ered from the thousands of excavated vertebrate re- mains in our sites. This suggests that the fear and re- spect accorded these ferocious animals have always outweighed any interest in hunting them for food. In addition to animal foods, the marine environ- ments of Mussau are rich in edible seaweeds, which are regularly harvested by the Mussau people (primarily by women). At least five different species were collected by them, and we were told that baskets of dried sea- weed used to be traded to communities on the main island. As part of our intensive study of prehistoric mol- luskan exploitation carried out in 1988, an environmen- tal survey of the reefs and lagoons surrounding Eloaua, Emananus, Boliu, and Emussau islands was made by Carla Catterall and Mike Ritchie. Dr. Catterall has kindly contributed the following section summarizing the re- sults of this survey. SHALLOW-WATER MARINE HABITATS OF THE OFFSHORE ISLANDS (by Cara Catteral/)'8 There are extensive shallow-water marine areas within the Mussau environmental system, consisting mainly of living coral reefs and associated habitats, which occur in two major sub-systems. The first of these is adjacent to the S part of Mussau Island, and is henceforth referred to as the Southern Mussau system. The second occurs in association with Eloaua and Emananus islands, and forms an elevated coral atoll-like system to the S, to- gether with several smaller islets and cays. This system is referred to henceforth as the Eloaua Atoll system (Fig. 2.10). The latter is separated from the reefs associated with Mussau Island by a deep channel - 1 km wide (the Malle Channel). METHODS USED FOR HABITAT ASSESSMENT A qualitative assessment, description, and map of these shallow-water marine habitats was made during a 15- day period in September, 1988. The first stage of map- ping consisted of delineation of foreshores and of the edges of both shallow-water and deep-water features, based on a detailed map produced by the Royal Austra- lian Survey Corps (Edition 1-AAS, Series T601, Sheet 8894), at 1:100,000 scale (from 1973 aerial photogra- phy, supplemented by patrol reports and hydrographic charts). This map was then transferred to a smaller-scale version on waterproof paper, which was extensively annotated during 3 days spent systematically inspecting shallow-water areas from a Metzler inflatable boat, with frequent pauses to assess the underwater habitat, either by looking over the side with a face-mask, or by snor- keling around less easily resolved areas. During this time, records were made of: water depth; topography; sub- strate (relative percentages of sand, rubble [coral detri- tus], rock, and live coral); type and dominant genera of coral formations present; approximate cover and domi- nant genera of algae or seagrass (low 5-30%, moderate 30-65%, dense 65-100%); and, various other features of biological significance. A detailed map showing the spatial distribution of a large number of habitat categories was then drawn in the field. These categories were then cross-checked, pooled, and rationalized into a more simple and eco- logically meaningful set of generalized habitat types, which were in turn cross-checked against an enlarged aerial photograph of the area. Following map compila- tion, limited quantitative surveys of shellfish density and species composition were also made within selected habitat types. MAP AND DESCRIPTION OF SHALLOW-WATER HABITAT TYPES The variety and spatial arrangement of major shallow- water marine habitat types now present in the Eloaua Atoll system are shown in Figure 2.1 1. Figure 2.12 shows an idealized cross-section through part of an atoll sys- tem to indicate the typical topographic and depth char- acteristics of each habitat type. Figure 2.13 presents quan- 41 THE Mu FIGURE 21 Aerial view of part of the Eloaua Atoll system of reefs and lagoons. The small islet in the left foreground is Enusagila; much of Eloaua Island is visible in the middle distance. The main high island of Mussau looms in the far oistance. titative data on the areal extent of the main habitat types. Such a pattern is typical of tropical Pacific coral reef areas, such as those found in the Pacific atolls and on the outer Great Barrier Reef of Australia. Comparable descriptions of such systems can be found in Wiens (1962). These reefs characteristically contain a high di- versity of habitat types and of species, arranged in a complex spatial pattern, and have developed in a con- text of clear water and low nutrient availability. Where terrestrial runoff from continental areas (or other factors) leads to sedimentation and nutrient en- richment, there are typically a variety of associated changes within the adjacent coral reef ecosystem. These include a differentiation between inshore areas (where mangrove swamps and seagrass beds are more likely to develop) and offshore areas. Within the Mussau envi- ronmental system these processes have led to a pattern in which the Southern Mussau system include consider- able areas of mangrove swamp (mainly N of the mapped area, adjacent to the large, mountainous island of Mussau), and a higher proportion of seagrass beds than the Eloaua Atoll system, where few mangroves oc- cur and the small land areas consist of uplifted coral platforms which provide little nutrient or sediment run- off. The following are descriptions of each of the habi- tats distinguished in Figure 2.11. The tidal range in the Mussau area is usually less than 1 m. Sand or Rubble F/=at (5). Areas 0.5 to 1 m deep at low tide, these habitats are composed of shallow sand or rubble with variable al- gal cover and little else. Three subcategories can be dis- tinguished. S 1 (shallow sand bank) consists of drifts of THE ML E a) 0 cc a) CLi 0 o CO cc Q (D Q- 0 a) c- (_> Qcn 'c Q) n -a Q :ci (i cici c/,c 43 THE MU FIGURE 2.12 An idealized cross-section through part of an characteristics of each habitat type. relatively coarse 100% unvegetated sand. S2 (rubble plain) consists of areas with some sand together with rubble (40-95%); there may also be an algal cover of Caulerpa spp., and/or sparse seagrasses. S3 (sand plain) consists of areas of 100% (or almost) sand cover, some- times of a fine and silty nature, which may have a sparse to dense algal cover of Caulerpa spp. (including C. racemosa and C. cupressoides), and/or a cover of filamentous algae FIGURE 2.13 Areas covered by habitat types within the Eloaua Atoll system, part of the Southern Mussau system, and the entire mapped area (as shown in Figure 2.1 1). 10- Eloaua Atoll 5. ~ ~ ~ ~ ~ / D - / U) Other areas /r .. ]* F7 7-7-11 F/7/2 S G F P B,E L Habitat type atoll system to indicate the typical topographic and depth which can vary from a surface film to a thick covering of aggregated algal strands and mats. Seagrass Flat (G). Areas mainly 0.5-1 m deep at low tide, these habitats have a substrate of sand and/or rubble in varying pro- portions, which supports a moderate to high cover of seagrass. The seagrass community in different places is dominated by varying mixtures of Cymodeocea (mainly C. rotundata, but may include C. serrulata), Thalassia hempichii, and the large Enhalusacoroides. Species of the alga Cauleopa (especially C racemosa) and/orHalimeda may also occur in a sparse to dense mixture with the seagrass in Thalassia and Enhalus beds. Among the edible mol- lusk species typically occurring in seagrass flats are Anadara,Acrostemigma, and Gafranium. Coral Mosaic Flat (F). Typical "reef flat" areas are 0.5-1 m deep at low tide and are composed of a spatial mosaic of live (some- times partly dead) coral patches (1-2 m diameter) inter- spersed with areas of sand and rubble. The cover of coral clumps may vary in different areas, from some areas which consist of sand and rubble with as little as 10% live coral cover to areas consisting of sand with a 60-90% cover of coral clumps (microatolls). There may be a sparse to moderate cover of seagrass (mainly Thalassia and Cymodocea) or algae (mainly Caulerpa) on the sand and rubble sections of the coral flat. Impor- tant molluskan taxa found in this habitat include the large bivalves Hippopus hippopus and Tridacna spp., and gastropods of the family Strombidae (Lambis, Strombus). Rock Platform/Reef Crest (P). This is the shallowest habitat type, being mainly - 0.5 m P F Om 0 B I lom w ;: % Lw pm - g!+1 44 THE MUSSAU ISLANDS deep at low tide. The substrate consists mainly of cor- alline rock, formed from dead coral slabs and boul- ders, which may form a relatively smooth plain as a result either of wave action, or the cementing effect of coralline algae. In other areas, the rock platform may be topographically complex, and include large boul- ders thrown up during storm surges, or deeper sec- tions which contain actively growing coral colonies (of- ten Acropora spp.). Sections of some rock platforms may have a thin cover of sand, and algae (especially Caulerpa). Seagrass may be present at sparse to moder- ate densities. Rock platforms are usually present on the outer exposed edges of reefs, and are often better de- veloped in less sheltered situations. This is a prime habi- tat for the Trindacna clams, as well as many other smaller bivalves and gastropods, including Conus spp. Subtidal Basin (B): This category includes basins and slopes of sand inter- spersed with large coral formations ("bommies"), at depths of 1.5-2 m at low tide. Coral cover varies in dif- ferent places, from scattered microatolls (usuallyAcropora or Porites spp.) comprising as little as 5% of total cover, to more densely arranged coral formations constituting up to 50% cover, in a spatial mosaic interspersed with sand patches. The latter resembles a deeper version of some types of coral flat. The sand may be fine and silty, and may have a surface film of filamentous algae, espe- cially in places where the coral cover is low. The subtidal basins are habitat for such mollusks as Chama, Spondylus, andHyotissa. Reef Slope/Edge (E): The outer slopes and drop-offs off coral reefs, which often lie on the seaward side of the reef crest make up this habitat type. This is a transitional zone, which slopes from 1 m at low tide to 10-20 m, with increasing dis- tance from the reef. The zone is topographically very diverse, and is typically biologically very rich. A variety of coral species and formation occur, including many Acropora colonies of varying sizes, and massive Porites coral heads which may reach 2-4 m in diameter and height, reaching almost to the surface from deeper ar- eas. The considerable variation in the nature of the reef slope is associated with the degree of exposure to pre- vailing winds. The slope itself may be rapid or gradual, live coral cover may vary from - 10% up to - 90%, and the remaining substrate may consist of either sand, rock, or both. High rock cover usually occurs in exposed situ- ations, and high sand cover in more sheltered situations such as lagoonal slopes. Lagoon (1): Large areas of deep sand (3-5 m) are enclosed within coral reefs in this habitat type. There may be sparsely scattered (5% cover or less) live or dead coral "bommies," which may support a sparse to dense cover of algae (mainly Caulerpa). The sand cover is often 100%, espe- cially in deeper lagoons, and the sand may be very silty, and may have a surface film of filamentous algae. The different habitat types described above each support different assemblages of plant and animal spe- cies, a fact recognized by the contemporary Eloaua people. For example, they stated that the gastropods Anadara andAcrostenggma occur mainly in seagrass beds, and this was consistent with the results of our quantita- tive faunal survey. The distribution and relative areas of the different habitats are thus a major determinant of the numbers and kinds of shellfish present in the Mussau environmental system. For example, in the present-day Eloaua Atoll system, the lack of mangroves means that the gastropod Telescopium should be virtualy absent from the area, and is only expected to be common near Mussau Island, where there are substantial mangrove stands. Variation in faunal assemblages within and between archaeological sites may result from: (1) underlying dif- ferences in the availability of different faunal taxa, as a consequence of changes in the area covered by differ- ent habitats; (2) differences in human exploitation tech- niques, or cultural selectivity; and (3) direct or indirect effects of exploitation on the population density and age-structure of various species of the shallow water marine fauna. A knowledge of the habitat distribution, and of whether it may have changed over time, permits some prediction of the organisms available to prehis- toric collectors, and may also be used in the interpreta- tion of changes in the midden assemblages. It follows that baseline information useful in the interpretation of midden assemblages includes the answers to the fol- lowing questions: (1) What habitats are present, and where?; (2) what is the relationship between habitat type and density of various faunal taxa?; and (3) how are contemporary habitats likely to differ from those present at the time of archaeological site formation? These are matters that are taken up more fully in Volume II. 45 THE MUSSAU ISLANDS COASTAL GEOMORPHOLOGYAND IMPLICATIONS FOR HOt OCENE SEA-LEVELS One outcome of fieldwork I had carried out in Futuna (Kirch 1981), Tikopia (Kirch and Yen 1982), and Niuatoputapu (Kirch 1988a) during the mid-to-late 1970s was a conviction that the coastal zones of SW Pacific islands were highly dynamic environments dur- ing the Holocene, with major consequences for the ar- chaeological records of these islands. Green (1979a:32) had made the general observation that Lapita sites tended to be situated "on raised coral platforms, ma- rine terraces, and marine sand beaches from which the sea had fairly recently retreated." This was confirmed by my experiences in Tikopia and Niuatoputapu, where the earliest Lapita settlements had been established on beach terraces now both inland of the modern shore- line, and elevated between 2-5 m above the modern tidal range. This geomorphological setting strongly suggested that these Lapita sites had been formed during a period when the relative sea level was somewhat higher than at present, and that there had been a retreat in sea level (and concomitant coastal progradation) during the post- Lapita period (i.e., the last 3 kyr). With this background, I was predisposed to pay careful attention to geomorphological and geoarch- aeological indicators of possible sea level changes dur- ing my fieldwork in the Mussau Islands. At the time we were carrying out the Mussau field- work, a series of geomorphological studies from vari- ous Pacific Islands independently began to yield abun- dant evidence for a mid-Holocene higher sea-level stand over much of the region (e.g., Ash 1987; Bloom 1980, 1983; Chappell 1982, 1987; Isla 1989; McLean 1980; Miyata et al. 1990; Montaggioni and Pirazzoli 1984; Pirazzoli and Montaggioni 1986, 1988; and, Yonekura et al. 1988). Such evidence includes cemented coral plat- forms and degraded microatolls (paleo-low-tide indica- tors), incised shoreline notches (paleo-high-tide indica- tors), and paleobeach ridges and terraces (paleotidal range indicators), many of which can be dated by radio- carbon or other techniques. Continued study of such features, both by geomorphologists and by archaeolo- gists, has now produced a substantial body of evidence confirming such a mid-Holocene high-stand (e.g., Allen 1998; Dickinson 1998, 1999, in press; Dickinson et al. 1994, 1998; Fujimoto et al. 1996; Kirch 1993a). The vertical magnitude of such a high-stand has been vari- ability estimated for different islands and subregions of the Pacific, from as low as + 0.8 m in French Polynesia, to perhaps + 3.0 m in parts of the W Pacific. The timing of this peak highstand is also a matter of considerable discussion, but a broad concensus is emerging that this occured between an envelope of - 6-4 kyr BP. Given the association of numerous Lapita age sites with paleobeach terraces, this high-stand must have persisted until at least - 3 kyr BP. Based on rapid progradation of beach sands that commenced at the To'aga (Samoa) Site around 2 kyr BP, as well as similar evidence from Tikopia and Niutoputapu (Kirch 1993a), sea-level retreat to the modern stand was certainly in train by this time.19 Ex- planations for this dynamic sequence of changing Ho- locene sea-levels will presumably need to combine glo- bal eustatic changes with regional hydro-isostatic changes, such as those resulting from thermotectonic subsidence, local volcanism, lithospheric flexure, or other geotectonic processes (Dickinson, in press). In Chapter 4, I present detailed geoarchaeological evidence for a dynamic coastal environment of Eloaua Island during the period that the ECA, ECB, and EHB Lapita sites were formed, including older shorelines and paleobeach terraces. This evidence is corroborated by two kinds of general geomorphological observations made during our Mussau fieldwork, which can be briefly described here. (1) The first is the presence of low-ly- ing aprons of unconsolidated calcareous beach sands which form the modern coasts of much of Eloaua, Emananus, and other offshore islands in the Mussau group today. The berms of these beach terraces are typi- cally - 1 m above the modern mean tide level. These unconsolidated sand aprons are frequently the setting for modern Mussau villages and hamlets, but show no evidence of archaeological deposits older than - 1 kyr BP. (2) In a number of islet localities, these low-lying sandy aprons abut, on their inland perimeters, with paleobeach terraces that are typically - 1 m higher than the seaward aprons, and thus - 2 m higher than modem mean sea level. It is on such paleobeach terraces, espe- cially that in the central part of Eloaua Island (Fig. 2.14), that archaeological deposits of Lapita age are found, such as the ECA and EHB sites (see Chapter 4). It is our contention that these slightly elevated paleobeach terraces formed as a result of the mid-Holocene high- stand, and that the seaward, low-lying sandy aprons de- veloped during the subsequent phase of sea-level re- treat to the modern stand. (3) The third indicator of a 46 F ML higher sea-level stand consists of incised shoreline notches typically found along the cliffed limestone coasts where the sandy terraces and aprons are absent. As has been documented for other limestone coasts (e.g., Pirazzoli 1996; Stoddard 1965; Yonekura et al. 1988), such notches form through a combination of corro- sion and bioerosion, from wave action at high tide, and from solution. An example of such an incised shoreline notch is shown in Figure 2.15. I obtained measured profiles of five such incised shoreline notches, one at Esapinoaka Island (a small is- let N of Boliu Island), two on Boliu Island (the first at Talinganitavao Point on the NW side of Boliu, the sec- ond at Etangatumanusa), and two on Eloaua Island. The cross-sections of these profiles are depicted in Fig- ure 2.16. In all three cases, it is clear that the maximum extent of incision lies - 1-1.5 m higher than modern mean high water. In four cases, a "double notch" is present, with the lower notch being actively incised to- day while the upper notch is partially coated with sta- lactites and flow stone, indicating no active incision. Thus, these upper incisions indicate a period of higher rela- tive sea level. Given the absence of evidence for local tectonic uplift, the most plausible interpretation is that they formed as a consequence of the mid-Holocene high-stand, which therefore must have been - + 1 m. This evidence for local coastal dynamism has im- plications both for the archaeological record of Mussau, FIGURE 2.14 Geological map of Eloaua Island, showing the central areas of uplifted limestone with their sharp escarpments, and the distribution of late Holocene paleobeach terraces of unconsolidated marine sands. and for interpretations of marine ecology and human subsistence patterns during the past few thousand years. Archaeologically, understanding the temporal pattern of changing sea levels-from a mid-Holocene high-stand followed by a fall in relative sea-level and rapid coastal progradation-is essential to interpretating the forma- tion processes at Lapita sites such as ECA and others. As will be seen in Chapter 5, it is also critical to under- standing why there are no pre-Lapita archaeological de- posits in the coastal EKQ rockshelter on Mussau Is- land. But these coastal processes have also affected the areal extent and character of the marine microhabitats of the Mussau offshore islands, with major implications for the distribution and extent of human-exploited re- sources, including mollusks, crustaceans, seaweeds, and inshore fishes. These are issues that will be taken up in greater detailed in Volume II of this series. THE ETHNOGRAPHIC RECORD The ethnographic record for Mussau is almost as irn- poverished as that for the islands' natural history. The primary source is Nevermann's 250-page volume (1933) in the monumental Ergebnosseder Siisee-Expedition series, at first glance a seemingly comprehensive ethnographic account. However, the expedition was in the Mussau group only from 9 August to 26 September, 1908, and ethnographic inquiries were made using an interpreter ; Uplifted reef escarpment Unconsolidated a calcareous sands L imetn Plateau I T- I I >X- I'. '''D \' I ,'' . THE ML FIGURE 2.15 A typical incised shoreline notch along the coast of Boliu Island, Jason Tyler stands with the stadia rod on the currently eroding solution platform. Note the actively incising notch, relating to modern sea level, at -+0.5 m, and the higher notch at -+1 .5 m, relating to the mid-Holocene high-stand. with Pidgin English as the medium of translation, a far from satisfactory method. Not surprisingly, a large part of Nevermann's volume deals with material culture, and those sections on social organization, religion, and so forth are superficial and of questionable accuracy. None- theless, the expedition achieved a remarkable amount in such a short stay, and provides a record which is not now possible to replicate. Other early German sources include Parkinson (1901, 1905, and 1907) who made brief visits to Mussau in 1900, Danneil (1901), and Meyer (1900). In 1925, E. W. Pearson Chinnery, Government Anthropologist to the Territory of New Guinea, spent shghtly more than two months in the St. Matthias Group, and published a set of ethnographic "notes" only slightly less extensive than those of Nevermann (Chinnery 1925)." Although the majority of his text deals with Emira rather than Mussau, he does provide additional useful information. I am not aware of any ethnographic fieldwork subsequent to Chinnery's survey. There is good reason to suppose-on the evidence both of archaeology and of linguistics-that Mussau has been continuously inhabited since the mid-second millennium BC, 2 and that the population presently in- habiting the islands has to a large extent descended from those who occupied the Lapita villages on Eloaua and Emananus. This is not to downplay the important cul- tural changes which have occurred between ca. 1500 BC and the "ethnographic present," or to suggest that there have not been significant external relations and quite possibly additional immigration. However, a continu- ous historical sequence linking the Lapita period in Mussau with the ethnographic record is significant, be- cause it opens the possibility for a "direct historical ap- proach" within Near Oceania, and not just in Western Polynesia where continuity from Lapita to Polynesian cultures is well established (Kirch and Green 2001). THE MUSSAU ISLANDS 4 0A Stalactites ~~~Sol ution pitted limestone~-" v 0 3 - 2 - 1 - Om- Stalactite Older notch Active notch Sharply'.' % Solution pitted... % & solutions4 .17 lime stone .' pte I I I I I I I I I I I I - - I I I 1 I 1 I I I 1 I 1 I Om 1 2 3 4 5 6 7 8 9 10 1 1 1 2 0 m 1 2 3 4 5 6 7 8 9 10 1 1 1 2 ?~~~~~~~~. Stalactites .. .. * Solutio ~ ~ .. .. .. Y limeston acite 2 3 - 2 - 1 - m - Stalactites Older notch 2 Om 3 - FIGURE 2.16 Measured profiles of five incised shoreline notches. The star and arrows indicate approximate modern mean sea level, and the tidal range. (A) Eloaua Island, lagoon coast; (B) Eloaua Island, lagoon coast; (C) Boliu Island, Talingaitavao Point; (D) Boliu Island, Etangatumanusa; (E) Esapinoaka Islet. 2-~ 1 - Om- Older notch Active notch T..~ ~~~~~~-- -- 3 - 2 - 1 - Om- 3 - 2 - 1 - Om - Om 3 4 I I I I I 49 1 1 2 3 Om - Mu FIGURE 217 Views of a traditional hamlet at Palakau and house at Enai, taken during the Sudsee Expedition of 1908 (from Nevermann 1933, plate 10). Note that the Palakau hamlet is set among large economic trees of the arboricultural zone, with small house gardens adjacent. THE MUSSAU LANGUAGEAND ITS EXTERNAL AFFINITIES The language of Mussau has never been comprehen- sively studied, although Chinnery (1925) collected a short vocabulary and Nevermann's (1933) monograph con- tains many terms. Blust (1984), a trained linguist who specializes in Austronesian languages, was able to work with several Mussau speakers resident in Manus, and obtained a more extensive 570-word list, which allowed him to make some specific assessments of synchronic and diachronic phonology. Blust's work was sufficient to establish that Mussau is an Austronesian language, and part of the Oceanic subgroup. Blust (1978) felt that there was little basis for subgrouping Mussau with the languages of the Admiralty group (Manus), nor does it THE Mu FIGURE 2.18 A contemporary village on Eloaua Island, showing a mix of sago-leaf thatched houses, and houses with rough-cut planking and corrugated metal roofs. Like its predecessor shown in Figure 2.1 7, this hamlet is situated in the arboricultural zone. seem to fit well with any of the other first-order sub- groups which Ross (1988) defined for the Near Oce- anic region. It is clearly not a part of the New Ireland group of languages. Ross (1988, 1989) was uncertain about whether Mussau comprised a separate subgroup or should be linked with the Admiralty cluster, although he suggests that Mussau may represent, like the Admi- ralty group, a separate first-order subgroup within the Oceanic cluster. In his most recent overview of Austronesian subgrouping, Ross says the following: ... the Admiralties family and the St Matthias group, are each clearly defined by a set of innovations, and Proto- Admiralty has been reconstmcted... The St Matthias group may yet prove to be specifically associated with the Admiralties family; it is not closely relat[ed] to its southern neighbors on New Ireland (1995:89). Ross (pers. comm., 1988) also has looked at the ques- tion of whether there are any affinities with Micronesian languages, and "noted that the very features that link the Admiralties and Mussau also link both with Micronesia." However, the evidence is not sufficient to say whether these features are retentions, shared inno- vations, or mutual borrowings. If, after further linguistic study, Mussau does prove to be a first-order (or even relatively high-order) sub- group of Oceanic, this has significant implications for culture history, given the putative linkage between the earliest Lapita phase in Near Oceania and the emplace- ment of the Proto-Oceanic dialect linkage." Alterna- tively, if it proves to be related to the Admiralty cluster, this could imply that there was originally a dialect link- age between Manus and Mussau, which would be en- tirely consistent with the archaeological evidence for material exchange between these island groups. SOCIAL ORGANIZATION According to Chinnery (1925:205-6) Mussau society was divided into two exogamous groups, named E Veli (with the valusu pigeon as its totem), and Saitalai (with the sava eagle as its totem). (We were given identical informa- tion by Ave Male in 1985.) Chinneryprovides a lengthy list of subgroups under each of these main exogamous divisions. Inheritance was supposedly matrilineal (Chinnery 1925:129). Chinnery uses the term "chief' to describe the political leaders (the Mussau term is vau or vaum), but does not say whether these positions were hereditary. Mussau never seems to have been densely settled, with population concentrated in small pockets here and there. Parkinson (1907) thought the total population to 51 THE MUSSAU ISLANDS be not more than 1,000, while Nevermann (1933:38) based on his partial census of several villages, doubled that estimate. Still, a population of only 2,000 in a group of islands with a total land of area of 363 km2 gives an approximate density of only - 5.5 persons/km2, a rather low number in comparison with other parts of Oceania.24 SETTLEMENTPATTERNS, LAND USE, AND SUBSISTENCE The settlement pattern observed in the early German colonial period seems to have been one of dispersed hamlets or small villages comprised of clusters of dwell- ings, situated near the shore, and generally set among groves of fruit and nut trees, not unlike the situation today. Chinnery (1925:205) gives the term masiliki for these hamlets. Referring to Emira, he adds that "for- merly, it was said, large chief's houses, with carved posts, occurred" (1925:147). Nevermann's photos of a ham- let and house in Enai on Mussau Island are reproduced here as Figure 2.17. These can be contrasted with the contemporary settlement pattern and house types, as seen in Figure 2.18. Parkinson (1907) described the Mussau huts (called ale) as "very prinmitive," with Pandanus-leaf thatched roofs just high enough to allow one to stand upright. There was no special sleeping place, but in each "there was a hearth and beside it a little heap of stones the size of a fist which are probably made hot for the purpose of preparing food." Taro (ast), bananas (uri), and breadfruit (ulu) were the main starch staples, according to Parkinson (1907), the first two crops grown in "extensive gardens" near the villages. He notes that coconuts were sparse, espe- cially on the offshore islands, but that there were more of them on the SE side of Mussau Island. He saw nei- ther dogs nor fowls; pigs were present but not numer- ous on the small islands, but raised "in great numbers" on the larger island. Nevermann (1933:83-107) elabo- rates considerably on Parkinson's observations, although again much of his text deals with material culture. He confirms taro and banana as main staples, but also lists a significant number of the tree crops by native name (presumably the botanical identifications were unknown to him), including tauno (Pometiapinnata), i (Inocarpus fagiferus), aitabagi (Terminalia whitmoret), natu (Burckella obokta), malai (Spondiasdukcis), ta (probablyDranomelon dao), alinaca (Barrmngtonia magnfica), and ieri (Pandanus kaernbachii).35 One aspect of arboriculture which was evident in 1908, but which is now completely absent due to the impact of missionization, is the cultivation of betel nut (Areca catechu). The swidden cultivations are also described, garden clearing being men's work, while the planting, weeding, and harvesting was done bywomen. According to Nevermann (1933) the main animal food was fish, including sharks and bonito, but other flesh foods noted include cuscus, dolphins, birds, wood maggots, grasshoppers, mussels, snails, and octopus. Pigs were raised on both Mussau and Emira, and wild boar hunted as well; pigs are said to have been eaten prima- rily for festivals, when many were consumed. MATERIAL CULTURE Nevermann (1933) describes and illustrates the mate- rial culture of Mussau quite extensively, and there is no need to repeat here a long list of objects. I will, how- ever, comment on some items that are reflected in the archaeological record; Figure 2.19 reproduces some of Nevermann's illustrations of these specimens. Of household equipment, Parkinson (1907) noted wooden bowls and coconut shell water containers. Co- conut scrapers consisted of a wooden stool-like arrange- ment with a Cardium shell lashed to it, of which Nevermann (1933:103-4) illustrates several variant ex- amples. Of some interest, because we recovered archaeo- logical examples from Site ECA, is a taro peeling knife ('taroschabemuschel') made from pearlshell with a ground cutting edge (1933:106, fig. 52); such implements are still used on Eloaua today. Parkinson (1907) also mentions a kind of short pestle made of Tridacna shell. Parkinson (1907) saw several kinds of fishing nets (he gives the terms uben and kea for these), but says no fishhooks were noticed. Nevermann (1933, fig. 31), however, illustrates a few crude one-piece hooks of tor- toise shell. Nevermann illustrates several kinds of fish nets, poles, and spears, and some rather crude angling hooks made of tortoise shell. According to Parkinson (1907), all of the "axes" (adzes, called iama) were made of Terebra shell, and he explicitly notes that no adzes of stone or Tridacna were seen. Nevermann (1933:117-20) concurs with this state- ment, noting that TrMiacna adzes were, however, observed on Tench Island. The apparent absence of Tridacna adzes 52 THE MUSSAU ISLANDS B C F E IF II I I .1 I1 H G FIGURE 2.19 Mussau artifacts collected by the SCudsee Expedition of 1908: (a) taro-peeling knife of pearl shell; (b) Triton-shell trumpet; (c) obsidian 'knives'; (d) bone spatula; (e) Trochus-shell armband; (f) coconut grater; (g, h) hafted adzes with shell blades. A D 53 I I. Q . 04 t  THE MUSSAU ISLANDS in Mussau is of note, because they do appear in the archaeological record, althoughin late period sites Terebra adzes are more common. Parkinson (1907) reported that the Mussau men wore no clothing, but that some of them covered the glans penis with a Cypraea shell, "in exactly the same way as they do in Manus." He gives the Mussau name for this shell as bule. Nevermann (1933, fig. 22) illus- trated a number of these shells, which had incised or carved designs. Parkinson also mentions "rather crudely ground armrings" (mare) of Trochus shell, worn by both men and women, along with "small tortoise shell rings, in the nostrils." The armrings are again illustrated by Nevermann (1933, fig. 15), and we recovered examples of these from post-Lapita period sites in Mussau. Of particular note is the use of a backstrap loom on Emira (but evidently not in Mussau), a technology most likely introduced from Micronesia. (The loom is well illus- trated by Nevermann [1933, plate 9].) The main woven objects were a narrow belt worn for dancing, a loin cov- ering worn by women, and a large mat for wrapping the dead (Chinnery 1925:196). Given the archaeological record of obsidian import- ing and use in Mussau, Nevermann's comments (1933:116) on obsidian "knives" (obsidianmesser) are of some interest. He observed that obsidian was used for shaving and for skinning animals, and gives a Mussau word, palane, for these flakes. He says that the Mussau people obtained these flakes by searching in the bush or at the edge of the bush, particularly along the E coast of Mussau Island (at Enai, Etasitel, and Etalat), and that not all locations where obsidian could be found were known to everyone.26 Since we know that obsidian does not occur geologically in Mussau, this strongly suggests that the Mussau people in Nevermann's time obtained flakes by scouring the sites of old villages or hamlets. This in turn raises the possibility that obsidian found in some archaeological contexts (especially in the post- Lapita period) could well have been recycled from ear- lier occupation sites, and need not imply direct impor- tation from an external source. TPADE AND EXCHANGE Given the extensive archaeological record for long-dis- tance (external) exchange, especially in the Lapita pe- riod, the ethnographic record of exchange or trade is of some interest. Parkinson (1907) commented that ca- noes he saw on the offshore islands were small, and not adapted to open water voyaging; he estimates that the largest might hold eight or ten people. But in a later (1905) visit to the E coast of Mussau, he saw "big, well made canoes which I had not observed on my previ- ous visit. These magnificent barks, some of them up to twenty-four meters in length, are carefully and not inar- tistically carved and painted at both ends" (1907). These he estimates could hold 30 to 40 people. Despite the presence of these canoes, however, he says that the is- landers are "not seafarers, and probably never leave their coasts." However, Parkinson goes on to note that on Emussau he was shown a spear from Manus, "a sign that there is occasional communication with these West- ern neighbors" (1907). Apparently the Emussau people told him of five such visits from Manus people, noting that they did not like to receive these visitors because they were "warlike and quarrelsome." Parkinson also relates a vague account of immigration "from the north," and noting the presence of a Micronesian type loom in Emira, speculated on contacts from the Micronesian region. Nevermann (1933:163-78) treats "trade and com- merce" in somewhat greater detail, although much of his account deals with canoe technology. He concludes that their canoes were not suitable for long sea trips (although this may simply reflect his European bias), and concludes that this is the reason that a regular trade was never established with New Hanover, despite inter- visibility of these island groups. Nevermann goes on to discuss local trade within the Mussau group, and be- tween Mussau and Emira (1933:174-77). He notes that on the main island most communication between vil- lages was along the beach or by canoe, as the paths were rough and limited. A main article of trade between the S part of Mussau, where reefs are extensive, and the NE coastal villages was dried seaweed, which the women from the S villages would pack in baskets. Similarly, the Malakat people in the N traded calabashes and strings of wooden beads to those in the S (Enai). Other trade goods mentioned are pigs, pig tusks, and Dasyurus teeth for necklaces, red dye obtained from root (turmeric?), taro corms, bananas, and fish. The Emira people traded their loom-woven fiber belts, skirts, and mats to Mussau, as well as carved spears. Chinnery reports that "in the old days of trade with St. Matthias [Mussau], the people of E Mira distributed large numbers of pais [belts] and kaijo [loin coverings] in the way of exchange for the 54 THE MUSSAU ISLANDS taro, pig, and betel nut brought by the St. Matthias trad- ers" (1925:201). Nevermann implies that there were quite formal exchange relationships between four vil- lages of the E coast of Mussau and counterpart vil- lages in Emir. Regarding longer-distance relationships with other island groups, Nevermann (1933:175-76) repeats some of Parkinson's account regarding stories of a Manus chief (Po Sin) who spent some time in Mussau before returning to Manus by way of New Hanover. He also recounts another story of a voyage by some people of Eboliu Island, who around 1883 sailed to Manus to present a festival.27 Since Nevermann had convinced himself that the obsidian flakes found on Mussau were of local origin, he discounted the possibility that they had been derived through external trade, although he notes that this was Hellwig's theory. To sum up the limited data on trade and exchange available from the early German colonial period, there was quite extensive and regularized trade among the various villages and hamlets of the Mussau group itself, and between certain communities in Mussau and Emira. There are sufficient hints at connections with Manus to also make it clear that the Manus group was well known to the Mussau people, and that either Manus people came to Mussau on occasion or vice versa (or both). However, such contacts seem to have been more irregu- lar and infrequent. Curiously, despite the fact that New Hanover is visible from Mussau in clear weather, there seems to have been no contact between these groups at all. NOTES TO CHAPTER2 I There has been some inconsistency in the application of the name "St. Matthias." Some authors (e.g., Nevermann 1933) apply it to the combination of Mussau, Emira, and Tench, which is the sense used here as well. Others (e.g., Chinnery 1925) use it as a synonym for Mussau. The name was given by Dampier who sighted the group on February 25, 1700, but did not land. 2 I say "atoll-like" because strictly speaking they do not fit the definition of an atoll, since they incorporate elevated reefs (makatea) as well as unconsolidated sands (motu); moreover, the origin of this atoll-like structure appears to be entirely as a result of tectonic island-arc uplift, rather than through sub- sidence. I The Eloaua airstrip is not an officially maintained airfield, and has no regular maintenance staff, radio, or navigational facilities. In rainy weather landing on the field can be danger- ous due to a thick mat of slick algae which has grown over the crushed coral, making the plane's brakes largely ineffec- tive! More than once we watched as the "scheduled" weekly Talair plane overflew the runway, wagged its wings to indi- cate the pilot's displeasure at the look of things, and turned back for Kavieng without landing. 'I had hoped to excavate this shell midden during the 1988 season, but could not secure permission from the landhold- ers. 5 Hohnen (1978, fig. 25), although he does not deal with Mussau specifically, depicts it as beingpart of a Quatemaryvolcanic arc including the Tabar and Lihir islands. I am dubious of this, because the volcanics of the latter are clearly much younger than that of Mussau, and Tabar and Lihir also lack the massive limestone capping that surrounds Mussau Island. It thus seems more likely that Mussau had its origins in the same Tertiary volcanic arc with New Ireland and Manus. This is apparently also the view of Exon et al. (1986:49). Only actual geological fieldwork, of course, can resolve this matter. 6 These samples were petrographically studied in thin-section by Prof. William Dickinson, and will be discussed further in relation to archaeological manuports in Volume Im. 7 R. Herzenberg (in Thilenius 1927:440) also reported on a single rock sample collected by the Siidsee Expedition in Mussau in 1908, which he examined in thin section: "Das Gestein ist epidotisierter Porphyr." 8 Rainfall at the Kavieng Station (New Ireland) has a median of 3,078 mm, with lowest and highest recorded annual fig- ures of 1,631 and 4,865 mm respectively. 'In 1988, the islands had been suffering from several months of drought when we arrived in early September. The effects were immediately evident as the plane flew low over Eloaua, and we could discern the parched condition of the second- 55 THE MUSSAU ISLANDS ary forest on the limestone plateau. In the gardens, the greatest impact was on the drylandtaro (Colocasia esculenta) crop, while sweet potatoes and manioc suffered much less. 0 Bleeker (1983:113) notes that on "coral terraces" rendolls may grade into rhodudalfs and rhodustalfs, and these soil types may also be present in Mussau. 11 These soils are essentially the same as those described from the larger islets of Micronesian atolls, under the rubric "Arno loamy sand" (Wiens 1962:345-46), in which there is a highly organic layer, black to very dark gray when wet, which de- rives its coloration in large part from anthropogenic input of charcoal and ash. 1 While living on Eloaua, we used such a well to obtain water for bathing and washing our kitchen implements; it was, how- ever, slightly too brackish for our drinking needs. 13 In this work I acknowledge the help of Dana Lepofsky, who did most of the plant collecting and pressing in 1988, and whose vegetation surveys of Eloaua and Boliu islands were especially useful in preparing this overview. 14 The Schadel Bay mangrove swamps are particularly noted as the territory of the large salt-water crocodile, Crocodylus porosus, and as such are shunned by Mussau people. 1" For an account of these creatures, and their ability to climb coconut palms and thus obtain the nuts, see O'Brien (1922), one of the classics of South Seas lore. 16 To the initiated, the rich, oily contents of the digestive pouch (despite its rather disgusting appearance) are even more deli- cious than the meat of claws and legs. One must be cautious, however, that in catching or cooking a crab one does not puncture the gut which runs through this pouch, or the con- tents will be spoiled. 1' This I personally verified in a Chinese restaurant in Port Moresby which serves crocodile meat sauteed in black bean sauce. The particular crocodiles are raised commercially for their flesh and skins, but I have no reason to think that the wild flesh would be any less tasty. 18 Catterall thanks Mike Ritchie for help in all aspects of the fieldwork, and Melissa Giese and Kenn Tews for assistance with the production of the habitat map. 19 Fujimoto et al. (1996) have argued this mid-Holocene high stand was followed by a -2 m low stand between - 3-1 kyr BP 201 am grateful to Holly McEldowney for securing for me a xerox copy of this rare publication, which is so obscure that it is not even listed in Taylor's comprehensivePaczficBibliography. 21 Whether the group had a pre-Lapita population, as has now been demonstrated for Manus as well as New Ireland and New Britain, remains an un-answered question. 22 There are evidently some differences between the speech of Mussau and that of Emira, but whether these are only slight dialectical differences as Chinnery (1925) maintained, is not clearly established. Isolated Tench evidently has its own distinct language. 231 I use the term "dialect linkage" here in the sense defined by Ross (1995:46); others have used the term "dialect chain". 24 In this regard, Mussau reflects a widespread Near Oceanic pattern of low population density, which contrasts strikingly with the situation in Remote Oceania. As I argue at greater length elsewhere (Kirch 2000), these differences may, at least in part, reflect the effects of endemic malaria and other infectious and debilitating disease in Near Oceania, which have continu- ally suppressed population growth. 25 The Mussau names are as given by Nevermann (1933:83), but without his complicated system of diacritical marks. I have supplied the scientific binomials based on Lepofsky (1992, table 1). 26 "Auf St. Matthias finden sie sich bei Enai, Etasitel, und Etalat vor allem am Buschrande und auf Emir an einigen Uferstellen und mitten im Busch" (1933:116). 27 "Um 1883 fuhren Eboliu-Leute nach der Erzahlung des Dolmetschers Makiroar nach der Groflen Admiralitatsinel, um dort zu einem Fest zugegen zu sein" (1933:175). 56