Prehistoric Giant Swamp Taro (Cyrtosperma Chamissonis) from Henderson Island, Southeast Polynesia1

Prehistoric Giant Swamp Taro (Cyrtosperma chamissonis) from Henderson Island, Southeast Polynesia1

JON G. HATHER 2 AND MARSHALL I. WEISLER 3

ABSTRACT: Subfossilleaf fragments of giant swamp taro (Cyrtosperma cha missonis) were recovered from archaeological contexts dating as early as A.D. 1451 (mean date) on Henderson Island (24 0 22' S, 128 0 19' W), Pitcairn group-a raised limestone (makatea) island isolated at the extreme margin of southeastern Polynesia and the Indo-West Pacific biotic province. Comparison of subfossil specimens and modern reference material from a range of known cultigens under scanning electron microscopy confirms the identification. A period of active interarchipelago voyaging between A.D. 1000 and A.D. 1500 is known from recent summaries of the geochemical analysis of exotic fine grained basalt artifacts from archaeological sites throughout Polynesia. If not an initial colonization, it is during this time that Cyrtosperma should have been introduced prehistorically to most, if not all, of the inhabitable islands of the region, especially those island groups lying to the west of Henderson. Inves tigation of subfossil plant remains adds another dimension to understanding plant distributions, prehistoric crop use, and subsistence practices in the Indo Pacific region.

INVESTIGATION OF THE plants associated with success. Comparable analysis of tropical ar past diet and subsistence economy in the Pa chaeological root and tuber-based econo cific by the direct analysis of the archaeolog mies has, until relatively recently, lacked a ical remains of plant tissues is limited com well-defined methodology and so, apart from pared with that in other parts of the Tropics a few isolated important identifications (see (for example, Central and South America Hather 1992 for a general review), has re and northern Africa). This is largely the re ceived less attention. Consequently, the plant sult of the importance of perishable root and remains associated with past root and tuber tuber foods in Pacific subsistence over that of based economies have also been left un seeds that often preserve well. Methodologies analyzed. In this paper we report the identi for the archaeobotanical analysis of partially fication of the leaf remains of an important or wholly seed-based economies are well es Pacific cultigen, Cyrtosperma chamissonis tablished and have met with considerable (Schott) Merr, from a prehistoric con;6xt and discuss the significance of this idyntification in terms of its past and presel)Ydistribution and its contribution to Pacific archaeobotany. 1 Fieldwork on Henderson Island was supported by Cyrtosperma chamissonis, the giant swamp the Wenner-Gren Foundation for Anthropological Re taro, is an aroid o( considerable dimensions, search (additional sponsors are acknowledged in Weisler cultlvatecCfor· its starchy corm (Figure 1). 1995). Funding to J.G.H. by the Science and Engineer ing Research Council (SERC), U.K. This is contribution Petioles are often spiny and arise directly 62 of the Sir Peter Scott Commemorative Expedition to from the corm at ground level and may reach the Pitcairn Islands. Manuscript accepted I August 1999. over 2 m long. The leaf blades, up to 1 m 2Institute of Archaeology, University College Lon long, are sagittate with pointed inferior lobes, don, 31-34 Gordon Square, London WCIH OPY, U.K. with the midrib on a continuous line with the 3 Department of Anthropology, University of Otago, P.O. Box 56, Dunedin, New Zealand (to whom corre petiole. In this character they are similar to spondence should be addressed). Alocasia but differ from Colocasia. Varieties 149 150 PACIFIC SCIENCE, Volume 54, April 2000

FIGURE I. Giant swamp taro, Cyrtosperma chamissonis, growing in a large excavated pit on Ebon Atoll, southern Marshall Islands. (photo by M. I. Weisler) differ in overall size as well as leaf shape and corms may reach 80 kg in weight (Barrau the spininess of the petiole. Corms commonly 1961:43, Small 1972:66). mature after 2 to 4 yr, when they may weigh The nomenclature of the domesticated up to 15 or 20 kg. Ifleft for 10 or more years, Cyrtosperma is a little confused. Arisaconthis Prehistoric Giant Swamp Taro-RATHER AND WEISLER 151

chamissonis Schott and Cyrtosperma edule distribution of this domesticate in the Pacific Schott are now both considered to be synon deserves reexamination. Brief review of the ymous with C. chamissonis. Cyrtosperma evidence for prehistoric long-distance inter merkusii is considered by some also to be archipelago interaction (so-called trade and synonymous with C. chamissonis but by exchange) within Polynesia strongly suggests others to be a separate and nondomesticated that Cyrtosperma should have been an ab possible ancestor of the domesticate (Thomp original introduction throughout the region. son 1982). Others view C. merkusii as a do This additional archaeological information is mesticate itself and a variety of C. chamisso especially important because Whistler (1991), nis in the western Indo-Pacific (purseglove in his recent summary of Polynesian plant 1972). It has been considered that C. cha introductions, reported Cyrtosperma as ab missonis was domesticated in Indonesia originally introduced only as far east as the (Massal and Barrau 1955), and so the two Cook Islands (see also Peters 1994) and that possibilities of C. merkusii being either an it was a modern introduction to the Mar ancestor or an early variety of C. chamissonis quesas and the Society Islands (Whistler are not far apart. In this paper, the plant re 1991: 45, 59). mains identified as Cyrtosperma sp. are con sidered to be C. chamissonis. Archaeological Context Cyrtosperma chamissonis is generally cul tivated on the low Pacific atolls in conjunc The subfossil plant remains were collected tion with coconut, breadfruit, arrowroot, and from the raised limestone (makatea) island pandanus. On such islands C. chamissonis is of Henderson, Pitcairn group (24 0 22' S, favored for its ability to tolerate both occa 128 0 19' W), isolated near the southeastern sional drought and the swampy saline con margin of Polynesia and the Indo-West Pa ditions that characterize the only fertile re cific biotic province (Kay 1984, Benton and gions of such islands. The herbaceous and Spencer 1995). A 5-month archaeological relatively low habitat of aroids is also fa program during 1991-1992 was designed to vored on islands with frequent high winds. investigate how small human groups survived The most important product of C. chamisso on isolated landfalls with impoverished ma nis is the starchy corm, which on some is rine and terrestrial biota, nutrient-poor soils, lands holds a position of considerable prom unpredictable rainfall, and, perhaps most inence in the diet. A secondary use of the important, isolation from high volcanic is crop is, however, the leaves, important in lands with correspondingly richer and more wrapping food for cooking in the earth oven diverse resources (Weisler 1994, 1997a, Wal and for covering the earth oven during dren et al. 1999). Twenty-eight sites were re cooking (Massal and Barrau 1955). corded, consisting mostly of rockshelters and Thompson (1982) described the modern caves, with a few gardening areas (marked by distribution of C. chamissonis in a paper that remnant stands of Cordyline fruticosa, level relates the economic importance of the aroids soil pockets, and stone mounds) and a large in the Pacific to both cultural preference and coastal midden (Weisler 1998a). Measuring ecology. Thompson reported that by far the up to 225 m2 in floor area, all caves and main concentration of C. chamissonis as a rockshelters were situated within 50 m of the major crop is in Micronesia, whereas in both high-tide mark. The deep. recesses of the Melanesia arid .Polynesia,' wnere C. cha shelters and the dry, calcareous sandy depos missonis is reported, it is of minor or only its provided ideal conditions for the preser occasional use. This, Thompson suggested, is vation of plant macrofossils. Thirty-one ra culturally significant, citing Luomala's work diocarbon age determinations from 12 sites in Kiribati (1953, 1974) as an important ex suggest an occupation sequence spanning ample, but is largely determined by environ about 600 yr beginning A.D. 1000 (Weisler mental conditions. The identification of 1995). All sites from which plant macrofossils Cyrtosperma, reported here, suggests that the were recovered are clearly prehistoric. 152 PACIFIC SCIENCE, Volume 54, April 2000

The archaeological deposits were ex rifolia, Musa paradisiaca, Pandanus tectorius, cavated by trowel and sediments sieved and Zingiber officinalis, collected from a through 6.4-rnm mesh (other samples were variety of locations in the Pacific. Leaves of processed with 3.2- and 1.6-rnm screens). modern material were air dried and then Both leaf samples were found in situ. The placed in a drying oven at 100°C. sample from site HEN-6 was found 19 cm below the surface, just above an earth oven (in a layer with dense combustion features) RESULTS and associated with abundant prehistoric ar tifacts, shellfish, and faunal remains includ Much of the archaeological leaf material ing bones of turtle, fish, bird, and rat. Wood seemed well preserved under low-power inci charcoal from deposits immediately below dent light microscopy. Several individual were dated to 380 ± 50 B.P. (before present fragments were up to 6 cm across and, [laboratory sample number Beta-59009]) and though thin and fragile, were pliable. Few calibrated to A.D. 1451-1650 (conventional fragments had the original leaf margin, and age -40 yr for Southern Hemispheric sam the important character of the marginal vein, ples, then calibrated at 2 (J after Stuiver and used to differentiate some taxa, was not Reimer 1993). From HEN-7, the leaf sample available. Under scanning electron micros (11 cm below the surface) was associated copy the leaf surfaces appeared considerably with an earth oven in a layer with prehistoric abraded, and trichomes (hairs), where pres artifacts, shellfish, and bone midden. A wood ent, were visible only as disrupted basal cells. charcoal sample (Beta-62941) provided a Stomata, however, were well preserved and conventional age of 270 ± 50 B.P. calibrated provided an important character upon which to A.D. 1522-1955. Although the latter end of to identify the taxa present. Cyrtosperma the calibration is modern, the archaeological chamissonis was identified from two samples. context is clearly prehistoric where, on Hen Figure 2 shows the similarities in leaf texture derson, this predates the early seventeenth and stomata between the modern reference century. specimen and the archaeological sample.

MATERIALS AND METHODS DISCUSSION Plant remains were systematically col The distribution of any plant has a tem lected for archaeobotanical analysis from the poral dimension that may be of considerable Henderson excavations. Leaf fragments from importance to anyone studying the evolution a variety of contexts were collected by hand of a specific plant or flora. To those studying and carefully packed for detailed laboratory the origin and dispersal of plants used by analysis. The leaf fragments were examined people, the temporal dimension is of particu under low-power incident light microscopy, lar importance and forms the basis of many and small fragments of both the abaxial and theoretical discussions of archaeological top adaxial surfaces were mounted for examina ics such as economy, migration, sedentism, tion by scanning electron microscopy. Stan and agricultural origins. Cyrtosperma is not dardbotanical texts were used to comple considered to be a crop of widespread im- ment data g-athered froill inodem- ieTerence -- portance; indeed, if -fenos·fo. be·-grown as a material in identifying the archaeological re major staple only in regions where there is mains. The modern reference material com environmental discrimination against other prised the following taxa: Alocasia macro crops, with perhaps cultural preference a rrhiza, Artocarpus altilis, Asplenium nidus, factor of varying significance. The impor Cocos nucifera, Colocasia esculenta, Cordy tance of the archaeological remains of Cyr line fruticosa, Cucuma longa, Cyrtosperma tosperma is in indicating a marginal agrarian chamissonis, Hibiscus tiliaceus, Morinda cit- environment where conditions were such that

'''~~''''''r " ...... " .....,..... ~ ~, ..",,..., '..~ ". ''''_' ,.,.,,,.-~ ... ,.. .""...... Prehistoric Giant Swamp Taro-HATHER AND WEISLER 153

FIGURE 2. Scanning electron micrographs of Cyrtosperma chamissonis. (1) Surface ofmodern leaf (scale bar = 75 Ilm); (2) detail of stomata of modern leaf (scale bar = 10 Ilm); (3) surface of archaeological leaf (scale bar = 75 Ilm); (4) detail ofstomata of archaeological leaf (scale bar = 10 Ilffi). (photos by J. G. Hather) other crops could not be grown. The practice Archaeological evidence for long-distance, of using Cyrtosperma as a crop in such con postcolonization voyaging within Polynesia ditions, so as to make productive otherwise has been summarized recently (Weisler 1997b, marginal islands, is uncommon in Polynesia, 1998b) and adds an important independent although well known in Micronesia (Weisler line of data for interpreting aboriginal plant 1999). As reported by Thompson (1982 : 193), dispersals in the region. Figure 3 presents the the modern distribution of Cyrtosperma prehistoric interaction spheres as inferred places its farthest east occurrence aboll! 2000 from. the distr!}mtion of. exotic fine-grained kin northwest of Henderson (128 0 18' W, basalt adze material. The largest solid line 24 0 22' S) on Makatea Island, Tuamotu Ar delimits the area with artifacts originating chipelago (148°l5'W, 15 0 50' S). As such, from Tutuila, Samoa-an interaction sphere the presence of Cyrtosperma on Henderson operating perhaps as early as 2000 yr ago gives, perhaps, an indication of agrarian (Best et al. 1992, Clark et al. 1997: 79). East practices in the Pacific that may now be only of the Cook Islands, all interaction spheres locally isolated, but may have been of con date to between A.D. 1000 and A.D. 1500 and siderable importance in the past. suggest a time of prolonged long-distance 154 PACIFIC SCIENCE, Volume 54, April 2000

/<":~"" 20' :' Hawaiian Is..••• Mariana Is. .' ...... ' ...... Caroline Is. . :. Marshall Is. • '. Yap Is... .: '. .: .• ,• ./ ••••••• /' Polynesian Triangle

Bismarck ; Kiribati. .: •••• ~~~~'::>.~/O!l)O~ ~""'~':TuvaIU''.,~./ MarqUe:~'I>" 0' ~w Bntaln', .. ~ ":'. okelau '. ~ "" ~'.. :. .... sa~a ~ •••••• santaCruz, ,,: .,..~ ' .. ".TU~rT)Otu~s '. ; FiJI#,.·: Cook Is. .. -"':'~"''''', ". Vanuatu _ .:-' Socia 'is .... ).,~. .': .... ~':. .~) ;Tonga . ~ j .~\ .... 20° ;: .. ". New Celedonia'"'" .' •. : •••••:....:.. . ~~tcalm Is. group ... • • Auslral Is.:'.·' M .\') • .. angareva, R N" / HendefSon Island apa UI ;,- .. Cyrtospenna 'p. ."- ;{\.. :~.~~: . ... r.¥ ------40' (~zealand •••••••••••••••••••••••••• 100llkm --=== •150'E ".:. _...... 170"W 150" 130" 110'

FIGURE 3. The Polynesian triangle and prehistoric interaction spheres delimited by the spatial and temporal dis tribution of exotic, fine-grained basalt adze material (after Weisler 1998b). The Australs and Tuamotus undoubtedly played a vital part in interisland contacts in central Polynesia, but detailed data are currently lacking. If not intro duced at initial colonization, it is most likely that most, if not all, cultigens were distributed throughout Polynesia during the period ofpostsettlement voyaging, A.D. 1000-1500. Consequently, the modern distribution of Cyrtosperma does not reflect its prehistoric occurrence.

voyaging. For example, we know that adze discouraged its growth. Having identified rock from a source on Eiao, Marquesas, was Cyrtosperma from two prehistoric contexts transferred to the Societies and from Eiao to on Henderson Island-arguably one of the Mangareva during the late twelfth to early most isolated landfalls-we believe that there fifteenth centuries (Weisler 1998b). Along is little reason to accept that Cyrtosperma with stone tools it is highly likely that culti was a modern introduction on any inhabit gens and other commodities were carried as able island within tropical Polynesia. well. Although Brown, in his botanical sur Archaeobotanical investigation in the Pa vey of the Marquesas in the 1920s, found cific is in its infancy and it is only by con that Cyrtosperma was rare, it was "well tinued, often-isolated analyses that the range .... Jm()wnJQ the_okl i!iliabit~I!t~ ... [and] it and type of archaeological plant remains would seem that in ancient times it was cul~ .. may be -idintl£ecf ana- evaluated (Hather tivated" (1931: 131). From the archaeologi 1991, Hather and Kirch 1991, Hather and cal evidence for prehistoric interaction, it Ellison 1995). The use of large leaves in the seems likely that Cyrtosperma was an ab subsistence economy of the Indo-Pacific original introduction across Polynesia, except makes them invaluable as archaeological re New Zealand and Easter Island, where cli mains in interpreting the spread of crop mate and/or cultural preference may have plants and .. agrarian practices in the region.

,..... ,.", , , .. ,. ~, ,. ., . ., ~ ...., ,." Prehistoric Giant Swamp Taro-HATHER AND WEISLER 155

Their particular use in cooking-to wrap interdisciplinary approach. N. Z. Ar food for placing in an earth oven or for stor chaeol. Assoc. Monogr. 21. age, to line or cover an earth oven, or for HATHER, J. G. 1991. The identification of serving cooked food-positions leaf material charred archaeological remains of vegeta in a broad range of preservational contexts. tive parenchymous tissue. J. Archaeol. Sci. Both leaf remains described here were found 18: 661-675. associated with earth ovens, lending weight ---, ED. 1992. Tropical archaeobotany: to this argument. The types of remains pres Applications and new developments. ent in the agrarian past ofEurope, Southwest Routledge, London. Asia, North Africa, and Mesoamerica are so HATHER, J. G., and J. C. ELLISON. 1995. considerably different from those in the Indo Plant macrofossil analysis from lake cores: Pacific region that many of the tried and An example from Polynesia. Pages 191 tested techniques and methods of interpreta 202 in A. J. Barham and R. I. Macphail, tion are not applicable. We should not let this eds. Archaeological sediments and soils: fact delay our pursuit of understanding past Analysis, interpretation and management. plant uses in the Indo-Pacific region. Institute of Archaeology, University Col lege, London. HATHER, J. G., and P. V. KIRCH. 1991. Pre historic sweet potato (Ipomoea hatatas) ACKNOWLEDGMENTS from Mangaia Island, Central Polynesia. M.I.W. thanks Steve Christian and family Antiquity 65: 887-893. of Pitcairn Island for their wonderful hospi KAY, A. 1984. Patterns of speciation in the tality and Roger Green for supplying copies Indo-West Pacific. Pages 15-31 in F. J. of an Auckland University Ph.D. thesis. Radovsky, P. H. Raven, and S. H. Somer, Funding to J.G.H. by the Science and En eds. Biogeography of the tropical Pacific: gineering Research Council (SERC), U.K. is Proceedings of a symposium. Bernice P. also gratefully acknowledged. Bishop Mus. Spec. Publ. 72. LUOMALA, K. 1953. Ethnobotany of the Gil bert Islands. Bernice P. Bishop Mus. Bull. 213: 1-129. LITERATURE CITED ---. 1974. The Cyrtosperma systemic BARRAu, J. 1961. Subsistence agriculture in pattern: Aspects of production in the Gil Polynesia and Micronesia. Bernice P. bert Islands. J. Polynesian Soc. 83: 14-34. Bishop Mus. Bull. 223 : 1-94. MASSAL, E., and J. BARRAu. 1955. Pacific BENTON, T., AND T. SPENCER, EDS. 1995. The subsistence crops-taros. South Pac. Pitcairn Islands: Biogeography, ecology Comm. Q. Bull. April 1955: 17-22. and prehistory. Academic Press, London. PETERS, C. 1994. Human settlement and BEST, S., P. SHEPPARD, R. C. GREEN, and R. landscape change on Rarotonga, Southern PARKER. 1992. Necromancing the stone: Cook Islands. Ph.D. thesis, University of Archaeologists and adzes in American Auckland. Samoa. J. Polynesian Soc. 101 :45-85. PuRsEGLOVE, J. W. 1972. Tropical crops: BROWN, F. B. H. 1931. Flora of southeastern Monocotyledons I and II. Longman, Polynesia. I, Monocotyledons. Bernice P. London. Bishop Mus. Bull. 84: 1-194. SMALL, C. A. 1972. Atoll agriculture in the CLARK, J. T., E. WRIGHT, and D. J. HER Gilberts and Ellice Islands. Department of DRICH. 1997. Interactions within and be Agriculture, Tarawa. yond the Samoan archipelago: Evidence STUIVER, M., and P. J. REIMER. 1993. Ex from basaltic rock geochemistry. Pages tended 14C data base and revised Calib 3.0 68-84 in M. I. Weisler, ed. Prehistoric 14C age calibration program. Radiocar long-distance interaction in Oceania: An bon 35: 215-230. 156 PACIFIC SCIENCE, Volume 54, April 2000

THOMPSON, S. 1982. Cyrtosperma chamissonis ---, ED. 1997b. Prehistoric long-distance (Araceae): Ecology, distribution, and eco interaction in Oceania: An interdisciplin nomic importance in the South Pacific. J. ary approach. N. Z. Archaeol. Assoc. Agric. Trop. Bot. Appl. 29: 185-203. Manogr. 21. WALDREN, S., M. 1. WEISLER, J. G. HATHER, ---. 1998a. Issues in the colonization and and D. MORROW. 1999. The non-native settlement of Polynesian islands. Pages plant species of Henderson Island, South 79-94 in P. Vargas, ed. Easter Island and central Pacific Ocean. Atoll. Res. Bull. East Polynesian prehistory. Easter Island 463. Studies Institute, FAU, University of WEISLER, M. 1. 1994. The settlement of mar Chile, Santiago. ginal Polynesia: New evidence from Hen ---. 1998b. Hard evidence for prehistoric derson Island. J. Field Archaeol. 21: 83 interaction in Polynesia. Curf. Anthropol. 102. 39: 521-532. ---. 1995. Henderson Island prehistory: ---. 1999. The antiquity of aroid pit Colonization and extinction on a remote agriculture and significance of buried A Polynesian island. BioI. J. Linn. Soc. horizons on Pacific atolls. Geoarchae 56: 377-404. ology 14 (7): (in press). ---. 1997a. Prehistoric long-distance in WmsTLER, W. A. 1991. Polynesian plant in teraction at the margins of Polynesia. troductions. Pages 41-66 in P. A. Cox and Pages 149-172 in M. 1. Weisler, ed. Pre S. A. Banack, eds. Islands, plants, and historic long-distance interaction in Polynesians: An introduction to Polyne Oceania: An interdisciplinary approach. sian ethnobotany. Dioscorides Press, N. Z. Archaeol. Assoc. Monogr. 21. Portland, Oregon.