Natural and Cultural History of the Marquesas Islands

Home , Cordia, Cordia subcordata, Fire plough, Francis Raymond Fosberg, Katherine Esau, Metrosideros collina

http://researchspace.auckland.ac.nz

ResearchSpace@Auckland

The digital copy of this thesis is protected by the Copyright Act 1994 (New Zealand).

This thesis may be consulted by you, provided you comply with the provisions of the Act and the following conditions of use:

• Any use you make of these documents or images must be for research or private study purposes only, and you may not make them available to any other person. • Authors control the copyright of their thesis. You will recognise the author's right to be identified as the author of this thesis, and due acknowledgement will be made to the author where appropriate. • You will obtain the author's permission before publishing any material from their thesis.

To request permissions please use the Feedback form on our webpage. http://researchspace.auckland.ac.nz/feedback

General copyright and disclaimer

In addition to the above conditions, authors give their consent for the digital copy of their work to be used subject to the conditions specified on the Library Thesis Consent Form and Deposit Licence.

The role of arboriculture in landscape domestication and agronomic development: A case study from the Marquesas Islands, East Polynesia

Jennifer Marie Huebert

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in Anthropology, The University of Auckland, 2014.

Abstract

Polynesian settlers transformed the native forests of the central Pacific islands into productive economic landscapes. Root crops came to dominate agronomic systems in many areas but arboriculture was the dominant mode of food production in some, and it is not well understood how these different endpoints evolved. In the Marquesas Islands, an economy dominated by Polynesian-introduced tree crops was encountered at western contact. Development of this system was investigated using large wood charcoal assemblages that spanned at least a 600-year period in Marquesan prehistory. Charcoal analysis is uniquely suited to inform on the reflexive processes of socio-economic development and landscape domestication in this setting, providing direct information on past vegetation and the use of arboreal resources. Data were compared from sites in three windward valleys with contrasting geographic and micro-climate conditions. A minimum of 59 taxa were identified including food-producing trees such as breadfruit (Artocarpus altilis), coconut (Cocos nucifera), candlenut (Aleurites moluccana), Pandanus, Terminalia and Tahitian chestnut (Inocarpus fagifer). Although tree crops were cultivated early in the sequence, breadfruit wood was infrequent (or absent) in contexts that dated to before AD 1400. Breadfruit was, however, a frequent occurrence in multiple study locations after 1650. A chronology of landscape domestication was defined for the windward valleys of Nuku Hiva, and the findings of this study suggest that the lush agroforests noted at contact were in place by the mid-17th century. Possible scenarios that could have led to this outcome were considered, and the decision to intensify food production through the cultivation of tree crops in this setting is linked to both the high yields that can be achieved on rugged land and the labour-efficiency of tree cropping. It is further argued variability in climate was a significant factor, and the ability to store fermented breadfruit (mā) for long periods of time would have provided food security if crops repeatedly failed. More generally, it is suggested that arboriculture is an important component of subsistence strategies in environments where climate is highly variable. Comparison with available data from the Society Islands provides some support for this suggestion. Keywords: Marquesas, arboriculture, anthracology, wood charcoal, landscape domestication

Acknowledgements

I am indebted to many people who made this project possible. First and foremost, I owe a debt of gratitude to my primary adviser, Melinda S. Allen. Without her guidance and encouragement this project would not have happened. I also wish to thank Andrew McAlister for advice and camaraderie both in and out of the field, and Sidsel Millerstrom for sharing the charcoal from her Hatiheu excavations. Rod Wallace is thanked for introducing me to the practice of wood charcoal identification and for sharing valuable insights into the interpretation of charcoal assemblages. Simon Holdaway gave valuable feedback in the design of this research project, and has shaped my thinking on archaeological theory over the years. Ongoing advice in the anthracological method, discussions about wood anatomy, and moral support was kindly provided by Emilie Dotte-Sarout. There are many people to thank for assistance with reference materials. Lloyd Donaldson of Scion (New Zealand Forest Research Institute, Ltd.) provided advice on wood anatomy and identification, as well as access to a very nice Scanning Electron Microscope. Gail Murakami gave advice on charcoal identification, including several microphotographs of reference material, and entertained my many questions during an impromptu visit. Catherine Orliac is owed a very gracious thank you for exchanging a sizable reference collection of East Polyneisan woods with Drs. Allen and Wallace some years ago. Kevin Butler also collected wood specimens on Nuku Hiva in 2003 that were an important resource. Deborah Woodcock shared several images of Sapotaceae woods from her personal collection of Hawaiian material. Jessica Lee shared microphotographs of Cordia lutea. A special thanks goes to Michael Thomas of the Rock Herbarium, University of Hawaii, for sharing scanned SEM images of Hawaiian woods taken by the late Charles Lamoureux. Jacqueline Bond of Rotorua expertly sectioned wood for the reference collection. For assistance in collecting botanical specimens in the field, a gracious thanks to Tioka Puhetini, Moetai Huioutu, and Dadou Teikiehuupoko of Nuku Hiva for expert guidance and also for sharing their knowledge of plants. The Teikiehuupoko (Tehina, Roger, and Panui) family of Hatiheu, and Moetai Huioutu and Anne Ragu of Taiohae, were superb hosts during field work. Thanks to Jean-François Butaud and Rhys Gardner for identifying voucher specimens, and to Ewen Cameron (Curator of Botany at the Auckland War Memorial Museum) for adding vouchers from the 2010 field season to the Museum collections. During the past several years I have benefitted greatly from advice,

ii forest walks, and conversations about plants and ethnobotany with Art Whistler and David Eickhoff and I thank them for sharing their knowledge. At the University of Auckland, I wish to thank Briar Sefton for assistance with numerous illustrations and Tim Mackrell for assistance setting up various microscope cameras and for photographing tiny fragments of charcoal for publication. The staff of the Anthropology Department has been very helpful and supportive over the years, and I owe many thanks to many people there. This research was funded in large part by a University of Auckland Doctoral Scholarship, along with grants from the Faculty of Arts Doctoral Research Fund, the Royal Society of New Zealand Skinner Fund, and the University of Auckland Postgraduate Students Association (PGSA). Lastly, I wish to thank my husband, Matthew Crofoot, for his understanding and support in this endeavour.

iii

Table of Contents Chapter 1: Introduction ...... 1 Chapter 2: Theoretical orientation ...... 4 Introduction ...... 4 Studying domesticated landscapes in prehistory ...... 5 Determinist and particularist approaches ...... 6 Ecological approaches...... 6 Socio-natural systems approaches ...... 8 Concepts of the theory ...... 10 Landscape ...... 10 Intensification ...... 11 Risk management ...... 12 Studying scale, pattern, and process in prehistory ...... 13 Theoretical model of landscape domestication ...... 15 Factors influencing outcomes ...... 17 Domesticated landscapes in Polynesian prehistory ...... 19 Pacific production systems ...... 19 Evolution of Pacific production systems ...... 21 Synthesis ...... 23 Case study ...... 24 Chapter 3: Natural and Cultural History of the Marquesas Islands ...... 27 In this chapter ...... 27 Geography, geology, and soils ...... 27 Weather, climate, and water ...... 29 Palaeoclimate and vegetation of the Central East Pacific ...... 30 Forces affecting climate in the region ...... 31 Palaeoclimate data ...... 32 Vegetation zones of the Marquesas ...... 36 Culture history ...... 42 Prior to AD 1400 ...... 43 AD 1400–1600 ...... 44 After AD 1650 ...... 44 Ethnohistory ...... 45 Domestic life ...... 46 Food, food production, and management of landscape ...... 47 Sociopolitical systems ...... 50 Feasting ...... 51 A brief arboreal ethnobotany ...... 51

Summary ...... 55 Chapter 4: Archaeological excavations on Nuku Hiva ...... 57 Introduction ...... 57 Hakaea Beach ...... 58 Overview / Introduction ...... 58 Local setting, geomorphology, vegetation ...... 58 Excavations ...... 59 Stratigraphy, radiocarbon ages, depositional sequence...... 60 Summary ...... 63 Materials selected for analysis ...... 64 Hatiheu sites ...... 65 Pahumano-o-te-tai ...... 66 Overview / Introduction ...... 66 Local setting, geomorphology, vegetation ...... 66 Excavations ...... 67 Stratigraphy, radiocarbon ages, depositional sequence...... 68 Summary ...... 72 Materials selected for analysis ...... 72 Hatiheu inland ...... 7 4 Overview / Introduction ...... 74 Local setting, geomorphology, vegetation ...... 74 Excavations ...... 75 Stratigraphy, radiocarbon ages, depositional sequence...... 77 Summary ...... 80 Materials selected for analysis ...... 81 Anaho sites ...... 82 Teavau'ua...... 82 Overview / Introduction ...... 82 Local setting, geomorphology, vegetation ...... 83 Excavations ...... 84 Stratigraphy, radiocarbon ages, depositional sequence...... 84 Summary ...... 88 Materials selected for analysis ...... 89 Teavau’ua South ...... 91 Overview / Introduction ...... 91 Local setting, geomorphology, vegetation ...... 91 Excavations ...... 92 Stratigraphy, radiocarbon ages, depositional sequence...... 92

Summary ...... 92 Materials selected for analysis ...... 92 Anaho stone structures ...... 93 Overview / Introduction ...... 93 Local setting, geomorphology, vegetation ...... 94 Excavations ...... 94 Stratigraphy, radiocarbon ages, depositional sequence...... 95 Summary ...... 100 Materials selected for analysis ...... 103 Summary ...... 106 Chapter 5: Methodology ...... 107 Introduction ...... 107 Wood structure ...... 108 Charcoal ...... 112 Brief history of the discipline ...... 113 Charcoal studies in Polynesia ...... 115 Formation processes and taphonomy ...... 117 Physical processes ...... 118 Cultural and natural processes ...... 124 Recovery of charcoal ...... 126 Quantitative analysis ...... 130 Evaluating sample adequacy, richness, and diversity ...... 132 Wood and charcoal identification ...... 135 Published reference material ...... 135 Challenges of identifying tropical woods ...... 136 Features destroyed by charring ...... 138 Field and laboratory procedures used in this study ...... 139 Creation of reference collection ...... 139 Field methods ...... 142 Sample selection ...... 143 Identification ...... 144 Sub-sampling ...... 145 Charcoal identification ...... 146 Data recording ...... 148 Summary ...... 148 Chapter 6: Systematic Review and Data Quality Assessment...... 150 In this chapter ...... 150 Taxa identified ...... 150 Systematic review ...... 154

Qualitative data results ...... 158 Artefacts of burning ...... 158 Anatomical anomalies ...... 160 Assessing sample adequacy ...... 161 Taxon accumulation curves ...... 161 Gini concentration indices ...... 166 Data quality assessment ...... 170 Indeterminate material ...... 170 Units for quantitative analysis ...... 171 Taxonomic richness of functional vs. non-functional contexts ...... 180 Sample size and relative abundance ...... 183 Consideration of potential biases ...... 184 Differential charcoal preservation ...... 185 Variability in recovery methods ...... 187 Completeness of reference collection ...... 189 Identifiability of woods ...... 190 Summary ...... 195 Chapter 7: Results by Study Site ...... 197 In this chapter ...... 197 Hakaea ...... 198 Vegetation pattern assessment ...... 205 Hatiheu ...... 209 Pahumano-o-te-tai ...... 209 Inland sites ...... 215 Vegetation pattern assessment ...... 219 Anaho ...... 226 Teavau’ua ...... 226 Teavau’ua South ...... 231 Surface features ...... 234 Vegetation pattern assessment ...... 240 Quantifying change ...... 251 Decline of native trees and shrubs ...... 252 Increase in Polynesian-introduced trees ...... 254 Summary of findings ...... 256 Chapter 8: Landscape domestication in Marquesan prehistory ...... 257 Introduction ...... 257 Processes of landscape domestication ...... 257 Vegetation clearance by fire ...... 258 Coastal vegetation modifications ...... 260

vii

Lowland forest modifications ...... 263 Extirpated or significantly reduced trees and shrubs ...... 265 Cultivation of trees ...... 269 Landscape domestication in the Marquesan prehistoric sequence ...... 280 The role of tree crops in prehistory ...... 286 Tree crops in the Developmental period ...... 286 Tree crops in the Classic period ...... 288 Factors influencing a reliance on tree crops ...... 288 Local conditions and/or historical contingencies ...... 289 Intensification and the physical environment ...... 289 Socio-political systems ...... 291 Climate and risk ...... 291 Preservation potential of breadfruit ...... 292 Comparison to Society Islands ...... 293 Chapter 9: Summary and Concluding Remarks ...... 297 Overview ...... 297 Methodological contributions ...... 300 Substantive implications ...... 301 Future directions ...... 303 References cited ...... 306 Appendix A: Charcoal identifications, absolute counts and weights ...... 333 Appendix B: Wood anatomy reference ...... 350

List of Figures Figure 2.1: A stylised representation of the four ecosystem functions organised into an adaptive cycle ...... 9 Figure 2.2: Exponential scales in space and time ...... 15 Figure 2.3: Conceptual model of land use and environmental change at the household level ...... 16 Figure 2.4: Schematic representation of Pacific land use on a high island ...... 20 Figure 3.1: Map of the Marquesas Islands in Pacific context...... 28 Figure 3.2: North coast of Nuku Hiva, showing landforms and coastline ...... 28 Figure 3.3: Rainfall variability on Nuku Hiva Island ...... 32 Figure 3.4: Climatic response regions of the southwest Pacific ...... 34 Figure 3.5: Some vegetation zones of Nuku Hiva, Marquesas Islands...... 41 Figure 3.6: Revised Marquesan cultural history timeline ...... 42 Figure 4.1: Map of Nuku Hiva ...... 57 Figure 4.2: Hakaea Beach as viewed from the eastern end ...... 59 Figure 4.3: Map of excavation at Hakaea Beach site ...... 60 Figure 4.4: Stratigraphy of SP8...... 61 Figure 4.5: Radiocarbon sequence for Hakaea Beach site...... 62

viii

Figure 4.6: View of Hatiheu Bay from the church cemetery...... 67 Figure 4.7: Map of excavation at Hatiheu churchyard ...... 68 Figure 4.8: Stratigraphy of SP5 (left) and TP1 (right) ...... 70 Figure 4.9: Radiocarbon sequence for Pahumano-o-te-tai site...... 70 Figure 4.10: Schematic drawing of stratigraphy identified in the two tested areas at Pahumano, cross-correlations of strata, and designation of analytic zones ...... 71 Figure 4.11: Hatiheu Valley inland setting near the Kamuihei complex ...... 75 Figure 4.12: Map of western central Hatiheu Valley excavation sites ...... 76 Figure 4.13: Site map showing locations of TU-4 ...... 77 Figure 4.14: Large earth oven near Vaiu’u River in profile ...... 78 Figure 4.15: Radiocarbon sequence for Hatiheu inland sites...... 79 Figure 4.16: View of the Teavau’ua coastal flat looking south-southeast ...... 83 Figure 4.17: Map of excavation at the Teavau’ua site ...... 85 Figure 4.18: A typical earth oven profile (Efe-7) in main occupation layer (IIIb) at Teavau’ua ...... 86 Figure 4.19: Radiocarbon sequence for layers IIIb and IV, Teavau’ua (AHO-1)...... 87 Figure 4.20: Anaho Valley from the overpass to Hatiheu...... 95 Figure 4.21: Map of Anaho stone structures included in the present study ...... 96 Figure 4.22: Plan view of Surface Feature 13 ...... 98 Figure 4.23: Radiocarbon sequence for Anaho stone structures...... 101 Figure 5.1: Cross section of (a) hardwood (Artocarpus altilis), (b) coconut wood, and (c) Cordyline stem ...... 109 Figure 5.2: Three planes of a typical hardwood ...... 109 Figure 5.3: Corewood / juvenile wood (white) and outerwood / mature wood (grey) ...... 111 Figure 5.4: Processes of charcoal formation ...... 113 Figure 5.5: Diagram of taphonomic process of macroscopic plant remains ...... 118 Figure 5.6: Charcoal fragmentation and mass reduction ...... 123 Figure 5.7 Charcoal deposits, archaeological contexts and vegetation representativeness ...... 129 Figure 5.8 Example of a Lorenz curve ...... 134 Figure 5.9 Data entry form used for creation of the wood anatomy database...... 141 Figure 5.10: Data entry form for charcoal identification...... 147 Figure 6.1: Twigs, knots, collapsed pores, fissured rays, vitrification, and anomalous banding noted ...... 159 Figure 6.2: Taxon accumulation curves for Hakaea Beach site...... 162 Figure 6.3: Taxon accumulation curves for Anaho sites...... 163 Figure 6.4: Taxon accumulation curves for Hatiheu Pahumano site ...... 164 Figure 6.5: Lorenz curves for temporal assemblages from six Nuku Hiva study locations...... 168 Figure 6.6: Scatterplot of count-to-weight correlation by location...... 172 Figure 6.7: Scatterplot of count-to-weight correlation by strata, Hakaea Beach Site...... 174 Figure 6.8: Scatterplot of charcoal count-to-weight by recovery method...... 175 Figure 6.9: Scatterplot of count-to-weight correlations by functional context...... 177 Figure 6.10: Scatterplots of charcoal fragment counts-to-weights for select wood taxa and nutshells...... 179 Figure 6.11: Taxonomic richness by functional context type for all sites...... 181 Figure 6.12: Taxonomic richness by functional context type, for each study location...... 183 Figure 7.1: Charcoal relative abundances for Hakaea Beach site...... 199

Figure 7.2: Charcoal relative abundances for Hatiheu Pahumano-o-te-tai site...... 210 Figure 7.3: Charcoal relative abundances for Hatiheu inland sites...... 216 Figure 7.4: Charcoal relative abundances for Anaho coastal flat (AHO-1, Teavau’ua)...... 228 Figure 7.5: Charcoal relative abundances for Anaho Teavau’ua South site...... 232 Figure 7.6: Charcoal relative abundances for sites at Anaho structures...... 237 Figure 8.1: Coastal scene on Nuku Hiva, 1804 ...... 261 Figure 8.2: Temporal process of vegetation transformation in the windward valleys of Nuku Hiva...... 281

List of Tables Table 3.1: Useful trees and shrubs of the Marquesas Islands...... 52 Table 4.1: Radiocarbon determinations for Hakaea Beach site ...... 63 Table 4.2: Cultural features from the Hakaea Beach site selected for study ...... 65 Table 4.3: Radiocarbon determinations for Pahumano-o-te-tai site ...... 71 Table 4.4: Cultural features from the Pahumano-o-te-tai site selected for study ...... 73 Table 4.5: Radiocarbon determinations for Hatiheu inland sites ...... 80 Table 4.6: Cultural features from the Hatiheu inland sites selected for study ...... 81 Table 4.7: Radiocarbon determinations for layers IIIb and IV, Teavau’ua (AHO-1) site...... 88 Table 4.8: Cultural features from the Teavau’ua site selected for study...... 90 Table 4.9: Cultural features from the Teavau’ua south site selected for study...... 93 Table 4.10: Radiocarbon determinations for Anaho stone structures ...... 102 Table 4.11: Cultural features from the Anaho stone structures selected for study...... 104 Table 5.1: Charred wood shrinkage ...... 142 Table 6.1: Summary of sample quantities and fragment weights and counts per area...... 151 Table 6.2: Wood taxa identified in Nuku Hiva samples ...... 152 Table 6.3: Non-wood plant tissues identified in Nuku Hiva samples...... 154 Table 6.4: Systematic review of plant taxa identified in this study...... 155 Table 6.5: Qualitative features of note and genera in which conditions were most frequently observed...... 160 Table 6.6: Percentages of indeterminate wood and unidentifiable material, by weight...... 171 Table 6.7: Kendall tau-b correlation coefficients for count-to-weight by location ...... 172 Table 6.8: Kendall tau-b correlation of charcoal count-to-weight by location and stratigraphic context ...... 173 Table 6.9: Kendall tau-b correlation of charcoal count-to-weight by recovery method ...... 175 Table 6.10: Kendall tau-b correlation coefficients for count-to-weight by functional context ...... 176 Table 6.11: Kendall tau-b correlation of charcoal count-to-weight for most frequently occurring taxa ...... 178 Table 6.12: Relative abundances of Sapindus saponaria in non-feature contexts, by stratum...... 184 Table 6.13: Readily identifiable wood taxa found in this study...... 191 Table 7.1: Ecological zones of trees and shrubs from the Hakaea Beach site...... 206

Table 7.2: Ecological zones of trees and shrubs from the Hatiheu Pahumano-o-te- tai site ...... 220 Table 7.3: Ecological zones of trees and shrubs from the Hatiheu inland sites ...... 221 Table 7.4 Ecological zones of trees and shrubs from the Anaho Teavau’ua (AHO- 1) site ...... 241 Table 7.5: Ecological zones of trees and shrubs from the Anaho Teavau’ua South site ...... 242 Table 7.6: Ecological zones of trees and shrubs from the Anaho north valley surface structure excavations...... 243 Table 7.7: Ecological zones of trees and shrubs from the Anaho south valley surface structures excavations...... 244 Table 7.8: Native trees and shrubs present in contexts before the 14th century AD, but absent from later contexts in the Hakaea Beach assemblage...... 253 Table 7.9: Native trees and shrubs present in contexts before AD 1650, but were absent from late contexts in Anaho north valley assemblages...... 253 Table 7.10: Ubiquity comparisons of Polynesian-introduced taxa at the Hakaea Beach site...... 254 Table 7.11: Ubiquity comparisons of Polynesian-introduced taxa at the Anaho north valley sites...... 255 Table 8.1: Processes of landscape domestication observed in the windward valleys of Nuku Hiva...... 259

Chapter 1: Introduction

As people settled the islands of the central Pacific, they affected widespread, long- lasting changes to the surrounding vegetation and landforms. While they often transformed native forests into productive economic landscapes, local variations in environment, climate, and social systems sometimes led to very different outcomes. Lush landscapes were noted in some locations, while other islands were virtually devoid of all but scrubby vegetation at contact. These outcomes were influenced by a complex interplay of natural and social processes, and consideration of how these contrasting endpoints came about is a major research theme of island archaeology in the Pacific and elsewhere (e.g., Rick et al. 2013). Understanding the interrelated processes of human-induced environmental change, climate change, and agronomic development are an important part of this research. In Polynesia, some important studies of socio-natural ecosystems have focused on addressing the recursive and dynamic relationships between people and their environments (e.g., Allen 1997, 2009a, 2010; Athens et al. 2002; Kirch 2007a, 2011; Rolett 1998; Yen 1989). Two major aspects of this research identified in the literature were the study of landforms and vegetation change. Much of this work has focused on vegetation transformations, with researchers considering what composed the native flora of the islands (e.g., Athens and Ward 1993; Athens et al. 2002; Flenley et al. 1991; Parkes 1994, 1997), what factors are most influential in vegetation change (e.g., Athens 2009; Mieth and Bork 2010; Rolett and Diamond 2004), and how and when these transformations occurred (e.g., Kirch and Ellison 1994; McWethy et al. 2010). A large and important body of related research pertains to the development of agronomic systems in Polynesian prehistory. These studies have focused on evaluating the evidence for traditional cultivation practices (Ballard et al. 2005; Barber 2004; Ladefoged et al. 2010; Leach and Leach 1979; Rosendahl 1994; Yen 1974a), how these practises evolved over time (Allen 2004a; Kirch and Yen 1982; Ladefoged and Graves 2000; Lepofsky 1994; Lepofsky and Kahn 2011; Yen 1985, 1980, 1990), and how they varied in different locations (Barrau 1965; Kirch 1994; Ladefoged and Graves 2008; Ladefoged, Graves, and McCoy 2003). Evaluating processes such as environmental change and the development of food- production systems requires study at both micro- and macro-scales, and the potential of integrative research was effectively demonstrated by the Hawaiʻi Biocomplexity Project (summarised in Kirch et al. 2011). Much of the data used in big-picture studies such as this

1 derives from combining information from surveying, modelling, geomorphological analyses, and palynology, with additional data drawn from archaeological excavation and the analysis of animal remains. Archaeobotanical remains are less often studied, though these materials have significant potential to contribute to this research. Studies of plant materials have informed on the introduction and use of various cultivars (e.g., Allen and Ussher 2013; Coil and Kirch 2005; Ladefoged, Graves, and Coil 2005; McCoy, Graves, and Murakami 2010; Rosendahl and Yen 1971; Weisler and Murakami 1991), cultivation practices (Athens, Ward, and Murakami 1996; Coil 2004; Horrocks and Rechtman 2009; Kirch 1989; Lepofsky 1994; Lepofsky and Kahn 2011), and landscape transformation (e.g., Allen and Murakami 1999; Dotte-Sarout et al. 2013; C. Orliac 2000) in prehistory. These works have also contributed important information to ongoing debates in Pacific Island archaeology regarding the extents of long-distance voyaging and exchange, the process of long term human eco-dynamics, and the chronology and intensity of agronomic development in prehistory. This thesis joins a very small Pacific assemblage of large, systematic wood charcoal studies (Coil 2004; Dotte-Sarout 2010a; Lennstrom and Murakami 2003 and related reports), that build on the aforementioned bodies of work. Polynesian settlers carried with them a suite of plants and could have developed a diverse production system in the Marquesas. Why, then, had it become so specialised by the contact period? When did introduced trees, especially breadfruit, come to dominate the lowland forests? In this thesis, I use archaeobotanical data to evaluate the reflexive processes of vegetation change and agronomic development in several locations on Nuku Hiva, the largest of the Marquesas Islands. Several large archaeological wood charcoal assemblages are examined to construct long-term arboreal vegetation histories in the study locations. There are three specific goals of this research. One, to understand how island environments were domesticated over time; two, to identify temporal trends that reflect changing relationships between human and plant populations; and three, to consider what influences were responsible for shaping the landscapes and agronomic systems that were seen at contact. Several possible scenarios could have led to the development of a production system centred on tree crops, and various alternatives are considered. It is argued that trees are ideal subjects with which to study past cultures in the Pacific because arboreal crops and fuelwood are major components of the traditional economic systems of the islands. While there are considerable challenges to the study of ancient tree use in other regions of the world, these problems are diminished in the islands of Central East Polynesia. The native island floras of this region contain few arboreal taxa

2 that were of economic importance, and humans introduced a limited number of taxa to the archipelagos in prehistory (Whistler 2009). Moreover, a study of arboreal vegetation in the Marquesas is well-suited to addressing the research questions posed by this thesis. Many early foreign visitors remarked on the centrality of breadfruit in the traditional diet, noting that extensive groves of fruit trees were growing to the very margins of the steeply sloping valleys. In the next chapter, the socio-natural systems approach is presented as a theoretical framework that is well-suited to the study of the historical process, landscape domestication, and food-production practices in prehistory. In Chapter Three, the natural and cultural history of the Marquesas is described to provide background for this study, including details of the geography, vegetation, climate, and the traditional culture of the archipelago. The archaeological contexts of the assemblages are presented in Chapter Four, and the methods used to process, identify, and analyse the samples are presented in Chapter Five. A systematic review of the taxa that were identified is then presented in Chapter Six, followed by a data quality assessment in consideration of various formation processes and taphonomic factors that may have affected the assemblage. In Chapter Seven, I present the results by study site. These are discussed in Chapter Eight, where the major findings are employed to evaluate the research questions of this thesis. A summary of the findings are presented in Chapter Nine, and the methodological and substantive contributions of the research are highlighted. To conclude, I identify several directions for future study.

Chapter 2: Theoretical orientation

Introduction Landscape domestication refers to the long and complex interactions between people and the environments they inhabit (Chase 1989; Hynes and Chase 1982; Yen 1989). Humans are seen as primary agents of change, building environments that shape society in a co-evolutionary process. Cultural activities transform places into ‘landscapes’ and significant effort is invested in domesticating entire domains. The manipulation of biological communities—intentional or otherwise—is central to this concept, with human activities creating patterned landscapes that produce many useful plant products. The study of domesticated landscapes, therefore, also encompasses many food- production practices. Taking a landscape approach to studying these activities is useful because traditional subsistence practices in many parts of the world do not fit into the categories of ‘forager’ or ‘farmer’, and often make no clear distinction between domesticated and non-domesticated elements (e.g., Harris and Hillman 1989). This approach refocuses studies of both human-environment interactions and food production practices by deemphasising individual plants, cultivation techniques, and linear trajectories of development, redirecting attention towards plant communities, manipulation strategies, and landscapes (Erickson 2006b). This is particularly useful in Oceanic settings where subsistence practices are so diverse they escape generalisations, productively changing the focus of the research from classification to explanation (Terrell et al. 2003). Trees were, and still are, important components of the domesticated landscapes of the Pacific Islands. Forest modifications have widespread and long-lasting consequences for plants and animals because trees are the keystones of many ecosystems. However, aside from the fact that forests were removed by settlers (directly or indirectly) or burned in swidden cultivation schemes, the role trees played in many prehistoric processes is not well understood. This has sometimes been attributed to difficulties recognising tree cultivation in the distant past. Unlike field systems and water-control technologies, agroforests and tree plantations require few landform modifications. The ethnobotanist Douglas Yen attributes it to a lack of criteria with which to distinguish between domesticated and wild forms, and the undetermined natural distributions of some species (Yen 1985). However, studies of the role of trees in human prehistory can provide empirical data that can be used to evaluate various aspects of socio-natural systems. These data have considerable temporal depth to facilitate the study of long-term social, economic and physiological processes because tree cultivation

4 requires coordinated, long-term interactions of entire social groups. Dotte-Sarout and colleagues (2013) found, for example, that landscape domestication in New Caledonia involved the creation of agroforests instead of large-scale swidden programmes. This study and others like it suggest that archaeological data, especially fossilised plant materials, are well-suited to the study of arboreal food-production systems. In this thesis, I evaluate the aforementioned ideas using a case study from the Marquesas Islands of central East Polynesia. The primary objective of this research is to assess how the lowland forests of the Marquesas were modified in prehistory and to determine when the extensive tree-cropping systems observed by early foreign visitors developed. The question of why an arboreal-based subsistence system was so highly developed in this setting by the late prehistoric period is also addressed. This chapter begins with a review of explanatory frameworks that have been used to study domesticated landscapes in prehistory. The advantage of using a socio-natural systems approach is then discussed and the concepts key to this approach are defined. I concentrate on concepts relevant to the main questions of this thesis, which include landscape, risk-management, and the intensification of food production. The importance of scale, pattern, and process are reviewed, and a conceptual model that will be used to study the transformation of newly occupied landscapes is presented. Social and natural factors that can influence the process of landscape domestication are then outlined, and some consideration is given to how these processes might be recognised in archaeological data. The relevant literature from Polynesia is then reviewed, including several important ideas regarding landscape change and the evolution of agronomic systems in the region. Finally, the Marquesan case study is presented.

Studying domesticated landscapes in prehistory Some have suggested that human domestication of landscapes began with the manipulation of woodlands (Sauer 1952, 21–22), and today the impact is so widespread that no location on Earth remains untouched (Vitousek et al. 1997). The study of relationships between people, environment, and subsistence has a long history and it is of continued interest for both theoretical and practical reasons. A consistent aim of this research has been to determine the relative importance of human and natural factors in global environmental change, but theoretical approaches to the subject have shifted considerably over the past century and are rapidly evolving today. I review the earlier anthropological perceptions of

5 human-environment interactions, and consider how they have shaped current thinking about domesticated landscapes. Several contemporary approaches to the study of human- landscape interactions in prehistory are evaluated, and I discuss the advantages and challenges of taking a socio-natural systems approach.

Determinist and particularist approaches For many centuries, researchers perceived of the environment (colloquially, ‘nature’) as a constraining agent that limited human possibilities. This view, referred to as environmental determinism, encourages unidirectional human-environment relations, progressing in prescribed stages from simple to complex (Moran 2009, 28–30). In the early 20th century, Franz Boas and his students challenged these ideas with a theory called possibilism, or historical particularism (Boas 1920). They considered history to be the main limiting factor that shaped culture and differences in the environment provided a framework that facilitated different possibilities. These ideas were eventually criticised for over- simplifying human-environment relationships and lacking the potential to explain the complexities of cultural diversity (Steward 1972, 35–42). By the mid-20th century the theory of cultural ecology had emerged, introducing an integrated systems approach to studies of human-environment interactions. Steward developed an approach that considered the causal connections between environment, social structures, and resources, and he proposed that resource exploitation had the largest impact on cultural evolution (Steward 1972, 40–1). Despite the fact that it took the interactive relationship between people and the environment into account, a certain amount of environmental determinism still underpinned this theory.

Ecological approaches Subsequent approaches rejected early linear arguments and instead emphasised the existence of feedback loops in human-environment relationships. Ecological approaches were developed at this time by anthropologist Ellen (1982) and geographer and archaeologist Butzer (1982). Ellen held that technology had a large impact on human- environment relations, and advocated developing an emic understanding of indigenous knowledge so as to better understand human-environment systems. Butzer introduced the concept of archaeology as human ecology, advocating a holistic, contextual approach that considered sites, settlements, and subsistence over artefacts. This approach utilised

6 ecosystems theory and concepts from ecology to explore the complex interactions between cultural, biological, and physical factors and processes (Butzer 1982, 7). It predominantly focused on human decision making (i.e., agency) and its impact on the environment, while acknowledging that it can be difficult to directly address aspects of the ecosystem using archaeological data. The integrative treatment of many types of biological materials within this approach, including fossilised plants, is of particular interest because it was the first to explicitly state that plant materials were a “critical record of the reciprocal relationships between people and plants in dynamic ecosystems” (Butzer 1982, 190). Ideas developed by Butzer and Ellen have had a considerable impact on environmental archaeology (Dincauze 2000; Reitz and Shackley 2012), practitioners of which strive to provide descriptions of past environments and explanations of the changing relationships between humans and said environments. Environmental archaeology is largely methodological and descriptive, and practitioners note that a wide gap now exists between it and the field of archaeology as a whole (Reitz and Shackley 2012, 30–1). It has been criticised for lacking high-level theory (Driver 2001, 51) and possibly being redundant (Thomas 2001). Today, many ecological approaches find their roots in the extensive research programme of historical ecology (Balée 1998, 2006; Crumley 1993). Historical ecology has been utilised by social scientists from many disciplines including archaeology (e.g., Anderson 2009; Balée and Erickson 2006; Kirch and Hunt 1997), and supports interdisciplinary research by encouraging the integration of diverse lines of evidence. It has been employed to study the reciprocal relationship between people and landscape by considering their influences on one another over time. Historical ecology uses the core concept of ‘landscape’, rather than ‘environment’ to describe human domains. Landscapes are an artefact of human activity, and they can be used to study the evolution of culture over time (Balée 2006; Crumley 1993). An understanding of temporal scale is critical to this approach, because studies of human-environment interactions must be conducted over times scales that are long enough to encompass both episodes of environmental change and distinctive cultural trends. Historical ecologists work to define useful temporal and spatial concepts, and the practice is considered by some of its founders to be a conceptual toolbox, or a framework for interdisciplinary research, or a point of view rather than a distinct method (Balée 1998, 14; Meyer and Crumley 2012, 111).

Socio-natural systems approaches Socio-natural systems approaches are concerned with the evolution of human- modified landscapes over long temporal scales (Barton et al. 2004; Fisher, Hill, and Feinman 2009b; McGlade 1995). They share many components of the research programme of historical ecology, and draw upon some of the principles of environmental archaeology. Natural and social processes are viewed as dynamic systems that co-evolve and produce long-term landscape changes. Having an awareness of organisational hierarchy and scale is key to evaluating processes at an appropriate level and linking them to human activities. This is a difficult task (Fischer et al. 2009a, 6), but the study of middle range theory has a long history (e.g., Binford 1977, 1980). Constructs such as ‘environment’ and ‘ecosystem’ are rejected as they are considered inseparable from humans and human behaviour, and the nature-culture dichotomy is replaced by a more holistic view—one of interconnectedness. These approaches are variously referred to in the literature using the terms human ecodynamics (McGlade 1995), dynamically coupled human-natural systems (Kirch and Zimmerer 2011), resilience theory (Redman 2005), and long-term socioecology (Barton et al. 2004) among others, and they have been criticised as being overly fragmented and having too many competing options (Fisher, Hill, and Feinman 2009a, 7). According to McGlade (1995, 113–4), one of the first to suggest this type of approach, the human ecodynamics framework he advocates was formulated in response to what had been criticised as representing opposite extremes in human-environmental thinking at the time: one, that functionalist ecological approaches were overly focused on economics, with ‘environment’ considered a resource to be controlled or used; and two, that post-processural approaches over-emphasised the subjective and were too focused on perception of the environment. Interdisciplinary studies of long-term socio-natural systems can bridge the gap between the social and natural sciences and it has been argued that archaeology, with its history of combining the two, is uniquely qualified to do so (e.g., Redman 2005; Van der Leeuw and Redman 2002). Several case studies have demonstrated the potential of this approach to study the relationships of humans and their environments in prehistory (Fisher, Hill, and Feinman 2009b; Kirch 2007b, 2011). It has been noted that certain aspects of socio-natural systems approaches, including their complexity, resilience, ‘fast-intermediate-slow’ variables, variation in process rates, and disturbance regimes (see McGlade and Van der Leeuw 1997) are also components of resilience theory (Redman and Kinzig 2003), which draws heavily on the ecological theory of adaptive change or ‘panarchy’ (Gunderson and Holling 2002, 3–24). Two ideas are

8 central to these approaches: one, that interactions are understood to be recursive rather than linear; and two, that these systems are always in flux rather than in a state of equilibrium. Non-linear models are best suited to the study of socio-natural systems because they incorporate many interlinked processes into feedback loops, such as culture change and environmental perturbation, and the latter can be an important and sometimes critical organising principle (McGlade and van der Leeuw 1997). The dynamics of these systems have been extensively discussed by Gunderson and Holling (2002), who characterise processes in non-equilibrium systems as adaptive cycles. The four main functions of these cycles are: (1) exploitation that occurs upon colonisation, (2) conservation where energy is slowly stored, (3) release where this accumulation is released rapidly, and (4) reorganisation where the system may (or may not) resemble its predecessor (Figure 2.1, left). These cycles occur simultaneously on many levels, and scale changes may occur between the release and reorganisation phases, which are referred to using the terms ‘remember’ and ‘revolt’ (Figure 2.1, right).

Figure 2.1: A stylised representation of the four ecosystem functions organised into an adaptive cycle (left), and remember and revolt cycles (right) (from Gunderson and Holling 2002).

The adaptive cycle can be used as an analytical framework, but further integration of social processes is needed for it to be useful for anthropological applications. This work is purportedly underway (see Folke 2006), but at present the ideas appear to be primarily used to create a conceptual model of dynamic structures and processes (e.g., van der Leeuw 2009, 48). They are useful to the present study for this purpose. An integrative approach was taken by Barton and colleagues (1999, 2004), who argue that understanding taphonomy is crucial to understanding socio-natural landscapes in

9 prehistory. Because landscapes reflect a cumulative and complex record of cultural and natural processes, and are in effect a palimpsest, they argue that the most useful way to increase our understanding is to break down the processes by which landscapes were formed. These processes include deposition, alteration, movement, loss, and accretion, all of which are considered to be as important as initial deposition. In this view, patterns of land use and geomorphology are equally as important as objects, and taphonomic processes are not simply viewed as noise or extraneous information. The archaeological record is viewed holistically as a long-term process. These ideas have influenced the present study. Of particular importance are concerns over the influence of the taphonomic processes, which will be an important component of the analysis.

Concepts of the theory Thus far, the concepts of landscape, intensification, and risk management have been identified as important to the study of domesticated landscapes and food production practices in prehistory. Each of these concepts will be briefly defined as they relate to using a socio-natural systems approach, though it should be noted that these are expansive concepts with an extensive cross-disciplinary literature of their own.

Landscape

Human-occupied spaces were first conceptualised as ‘cultural landscapes’, defined as geographical units long ago by Carl Sauer (1925, 321–33). Though the term landscape has a number of definitions, the one of interest to the present study is that used by historical ecologists. Landscape is a theoretical unit that has a temporal dimension, which is both cultural and evolutionary, but it does not privilege one over the other, and a spatial dimension that refers to regions, territories, or environmental units (Balée and Erickson 2006). Landscape is not to be conflated with the term ‘environment’ or ‘ecosystem’, which are associated with the concept of equilibrium in systems theory. It does not refer solely to land, nature, or space but also encompasses “the world as it is known to those who dwell therein, who inhabit its places and journey along the paths connecting them” (Ingold 1993, 62). Landscapes are viewed as historical and evolutionary, based in human experience (Balée 2006). They are the cumulative result of long-term social and natural processes and as such, studies of landscape can provide ways to examine these processes in prehistory. These ‘socio-ecosystems’ (McGlade 1995) can overlap, making it possible to conceptualise

10 multiple landscape processes operating in the same space and time. The concept of landscape also provides a spatial unit that is easily understood in most disciplines. It is viewed as a unifying concept in archaeology that can assist in considering possibilities in the past (Boyd and Chang 2010, 274). A related concept is that of the ‘domain’. Domains refer to the ecological and conceptual life-spaces of a society. They are defined in space and time as areas where environmental knowledge and strategies of resource manipulation are applied by local social groups (Chase 1989; Hynes and Chase 1982; Stanner 1965, 2). The knowledge, strategies, and actions applied to this sphere have been referred to as ‘domiculture’. Human-plant interactions have several related dimensions in this view (Hynes and Chase 1982, 48). The cultural stance towards the domain is one aspect that can be, alternately, perceived as something other than human or as fully integrated into the world view (e.g., Ellen 1996). Another is the range of plant extraction practices used, which can vary in intensity from the replacement planting of undomesticated plants to the cultivation of organised field systems. A third dimension is the processes used in changing plant distributions and morphology. Together, this tripartite view focuses broadly on human-plant and human-landscape interactions. It effectively replaces the concept of ‘agriculture’, which is especially useful in Pacific settings where many food production systems escape traditional classifications (Terrell et al. 2003; Yen 1989).

Intensification Intensification refers to increased labour per land unit – and/or attempts to extract a higher yield – to produce larger quantities of food from a defined land unit (Boserup 1965; Renfrew 1982). Initially, it was defined as increased productivity achieved through accelerated cropping cycles and capital investment, wherein permanent landscape modifications are constructed to increase yields. The economist Boserup (1965) first defined intensification as increased labour input, and asserted that population growth was the prime mover of agricultural change. Brookfield (1972, 2001) later refined this definition to refer to inputs of capital, labour, and skills against land (a constant) with the primary goal of producing more from a given area, or doing so more securely. Central to this idea is that livestock, trees, and land can all be considered forms of capital (‘landesque capital’) (Brookfield 2001, 184–5). Brookfield introduced the idea that innovation could be a form of intensification, a concept that encompasses genetic and agronomic innovations. Yen (1980,

1985) broadened these ideas by looking at production systems more holistically, and highlighting two aspects of intensification that are not necessarily agricultural: one, that increased gathering of wild plant foods and manipulation of these plant communities was a technological innovation; and two, that removing toxic compounds and/or processing food for long-term storage were post-harvest innovations. Intensification, therefore, is a multi- component process that can be achieved using various individual strategies, such as specialisation, diversification, the expansion of field systems, and the creation of cultivable spaces, or any combination of these techniques (Morrison 1994). Intensification has been the subject of much scholarly debate. Much attention has been paid to the identification of links between populations, resources, and technologies. Studies of causation have been the focus of much of this research, but there is also a need to consider the process of intensification itself (Morrison 1994, 1996, 2007). Morrison points out that food-production practises often involve a mixture of strategies, and new ideas on causation and change may derive from examinations of the individual components. H. Leach (1999) observed that traditional Pacific food-production practices included intensive components, and she questioned our ability to distinguish the archaeological indicators of the intensification process from those of the intensive land use component of subsistence. While it has been acknowledged that these differences can be challenging to tease out of archaeological evidence, prehistoric social structures are equally as important to understand when considering the intensification issue (Brookfield 2001; Kirch 1999).

Risk management Archaeologists are particularly interested in the study of prehistoric variance- minimisation strategies as they relate to food-production practices as a way to better understand how people mitigated risks that were critical to survival. Behavioural ecologists have defined risk as probabilistic variance (Cashdan 1990, 2–3). In studies of food- production systems, risk is equated with the chance of loss or with variance and uncertainty relating to random occurrences or lack of information (Marston 2011). Marston (2011, 116) notes that while the difference between these two definitions is not a significant concern as long as food-production is adequate to meet basic needs, when extreme shortfalls occur they diverge because management of variance becomes important to survival. The practice of risk management can be linked to changing demographic patterns and economic systems – a

12 concern that is exacerbated in circumscribed locations such as islands (Halstead and Jones 1989; Ladefoged and Graves 2000). To anticipate and mediate the consequences of variability, the role played by factors such as frequency, duration, and predictability in various processes needs to be taken into consideration (Halstead and O’Shea 2004). Many strategies have been developed to manage risk, from increasing the mobility of people and goods to innovations in storage. Diversification (e.g., Jones and Halstead 1995; O’Shea 1989) and intensification (e.g., Kirch 2006) are two key ways to cope with risk and often several strategies would be used in combination to achieve these goals. In a study of prehistoric agricultural strategies in leeward Hawaii Island, Allen (2004a) considered the impacts of changing agronomic maximisation and variance-minimisation behaviours. She argued that evaluations of agronomic fitness should take into consideration the shifting dynamics of both practices as they were utilised over time. Markers of risk-management strategies in archaeological materials have been examined through diverse materials and methods (e.g., Halstead and O’Shea 2004; Madsen, Lipo, and Cannon 1999; Marston 2010). Locating indicators of these strategies and quantifying them in a meaningful way has been identified as a particular challenge for archaeologists; while surface and sub-surface features (walls, terracing, boundaries), concentrations of micro- and macro-botanical materials, and changes in sedimentation can be evaluated, some are challenging to date or are uncertain indicators of these particular strategies (H. Leach 1999). With this challenge in mind, it is of particular interest to the present study that faunal and botanical analyses have been said to be well-suited to such an endeavour because their sub-fields are grounded in quantitative measurement and they employ well-defined methods when considering diachronic change (Marston 2011).

Studying scale, pattern, and process in prehistory In the past, landscapes were modified by humans both consciously and unconsciously in ways that gradually benefited society. Yen (1989) observed that the processes of landscape domestication and plant domestication were very similar, with their only discriminating feature being that the latter involves genetic manipulation of a species. The use of fire to modify landscapes, for example, is a significant selective force that operates on vegetation, native fauna, and landforms. But there are diverse pathways to environmental and social change, and these must be investigated in relation to the particular

13 environment, demographics, and social structures of the study location (Kirch 1994, 306–7, 1999). To understand change in prehistory, it is necessary to identify factors that influenced outcomes, and to link patterns with processes in the archaeological record. This investigation begins by considering how best to study pattern, process, and scale in prehistory. When using a socio-natural systems approach, there are both temporal and spatial aspects to the conditions of landscape and agricultural change and these factor into an over- arching concern regarding scale. Relating phenomena across scales has been called “the fundamental conceptual problem in ecology, if not in all of science” (Levin 1992, 1944). Levin reminds us that boundaries researchers draw are arbitrary divisions of continuous assemblages, and it is important to understand the scale of observations because different evolutionary forces operate on different scales. Concepts of scale are intertwined with those of patterning (Hutchinson 1953 cited by Levin 1992, 1944). In essence, patterning is a description of variation, and in order to properly quantify it, scale must be determined. These factors must be understood before one can accurately compare systems and look for processes that are the sources of patterning. Such concerns have been discussed at length by landscape ecologists (e.g., Bell 2012), and ecologists studying vegetation change have noted that system dynamics are best understood in a hierarchical context. This allows for the recognition of different patterns and processes using different scales of observation (e.g., Delcourt, Delcourt, and Webb 1982; Dincauze 2000, 375–81; Whittaker and Fernández- Palacios 2007; Fig. 2.2). Human-modified landscapes are the result of complex decision-making processes. To study patterning in these systems, a multi-scalar approach has been advocated, which has been recommended to study human-forest interactions in particular (VanWey, Ostrom, and Meretsky 2005, 51). Effectively integrating data from different processes to explain the past presents a significant challenge for archaeologists (Boyd and Chang 2010, 275). Kirch and colleagues have demonstrated that a multiscalar approach can be extremely effective when studying processes of landscape and agricultural change (Field et al. 2010; Kirch et al. 2011; Ladefoged and Graves 2008; and others). In their Hawaiʻi Biocomplexity Project (Kirch 2011), macro-scale processes were studied on an island or district scale, and micro- scale processes were studied on a household or landholding scale. Key variables that were modelled and evaluated included environmental variation, agricultural productivity (i.e., crop yields), and human-induced environmental (climate, landforms, vegetation, and soils) change over time. Spatial dynamics were examined at multiple landscape and territory

Figure 2.2: Exponential scales in space and time (from Whittaker and Fernández-Palacios 2007, Fig 2.20).

levels, and social and economic changes were studied at both the island, territory, and household level (Kirch and Zimmerer 2011, 26–8). The researchers noted that this framework was developed specifically for their project, but add that it could be leveraged to study change elsewhere.

Theoretical model of landscape domestication The challenge of understanding how processes identified in one setting can be leveraged to explain change in other locations remains. One way forward is to consider how fine-grained patterns observed in one location can inform research into the properties of established macro-scale processes in others. In the present study, an important task was to consider how to best identify and examine these processes. A useful model of the process of landscape and agricultural change in pristine environments (Figure 2.3), developed to study present-day settlers in the Amazonian forests of Brazil on a household scale, was located in the remote sensing literature (McCracken et al. 1999). These ideas will be leveraged to consider landscape domestication at archaeological timescales, where the model will be redesigned to consider the results of the present study.

Figure 2.3: Conceptual model of land use and environmental change at the household level (from McCracken et al. 1999, Fig. 1).

In this model, household composition and life cycles were linked to the transformation of individual land plots and the local environment, and a chronology of various food-production practises was projected. The idealised process depicted in Figure 2.3 took place over the course of one generation, approximately twenty years, on a single plot of land. The process begins with the assumption that new colonists consist mostly of young families. Five stages are represented in the cycle, each with implications for land use and environmental outcomes. The first stage occurs as small families begin to have young children, and it involves land clearance (deforestation) and cultivation of annual crops (cash crops, in this setting). Households then slowly begin to acquire livestock and perennial crops such as trees and agroforests, which have long maturation dates, require significant labour inputs encourage them to establish, and are limited by water availability and soil

16 conditions. As the household’s second generation reaches maturity, plots are alternately fallow or grazed, annual crop production wanes, and tree crops become an important aspect of production. The authors note that some important variations to this trajectory occur as topography, hydrology, and available labour vary the productivity of the land. Also, not all households are able to afford livestock and perennial crops, nor do they have equal access to markets or trading partners. Many other factors can influence outcomes, and other things need to be taken into consideration when evaluating these processes on archaeological spatial and temporal scales, as will now be discussed.

Factors influencing outcomes Domesticated landscapes were shaped by both social and environmental factors in prehistory. There were climate and edaphic limitations, which may not have favoured certain cultigens. Water availability was also critical. Land area was an important boundary condition, as habitable areas may have been circumscribed physically in areas with very rugged topography, for example, or on small islands, or socially when groups were restricted to certain territories. Food-production practices were very influential. While the various cultigens and techniques used by a society can be attributed to its founders initially, later outcomes can also be attributed to evolving socio-political and socio-economic systems, as well as innovation and demography. Developments in technologies related to the cultivation, processing, and storage of crops may have influenced the cost-benefit of certain cultigens. New cultigens and techniques obtained by trade or exchange may also have a significant impact on subsistence strategies (Yen 1989). Certain cultigens may have been preferred because they had a higher cost-benefit, producing better yields that required less labour input. Others may have become important later, when a group’s primary resources were exhausted. Some may have come to be preferred because they produced culturally valued products. Lastly, access to and control over water resources have been identified as among the most influential aspects of prehistoric socio-economic development (e.g., Barrau 1965; Kirch 1994; Wittfogel 1957). There are also several social processes embedded in production systems that heavily influenced outcomes. One is the production of surpluses for the elites, while the other encompasses production and consumption at the household level (Erickson 2006a). These are sometimes referred to as top-down and bottom-up processes, respectively, and there

17 have been debates as to which had more of a hand in shaping past societies, especially in regards to the intensification process (Erickson 2006a; Janusek and Kolata 2004). Top-down processes encourage production systems to develop under the direction of centralised power structures, sometimes in unsustainable ways. Because archaeological data favours the study of large scale processes such as these, top-down production processes are often studied by prehistorians (e.g., Kirch and Sahlins 1994; Kolata 1996, 2002). Bottom-up processes act as a significant counterpoint, but are not as extensively studied. They involve the actions at the community or kin group level, focusing on the decision-making power of the producers (e.g., farmer or commoner) and their traditional knowledge systems, which are more likely to be conservationist (e.g., Erickson 2006a). There have been efforts made to consider both processes in understanding production systems, and Lepofsky and Kahn (2011) note that “by refocusing our analyses on all players in the production system, a more nuanced understanding of the range of ancient environmental and social interactions emerges”, one that can influence our interpretations of prehistoric process. While complimentary, the research requirements to assess both perspectives are demanding (Janusek and Kolata 2004). Numerous factors have influenced past outcomes, and not all can be studied using archaeological data. With this in mind, several specific elements of vegetation change will be considered in this study: the constituent taxa and the relative contributions of each, their frequency of occurrence, and the bio-geographic distribution of these elements. The first two inform on changes in the species composition of area vegetation, including the introduction of new (‘Polynesian-introduced’) taxa. The latter items inform on the intensity of use and, if biases can be sufficiently accounted for, the ubiquity of woody species in the local area. These data facilitate an assessment of changing vegetation patterns over time. Shifts may indicate changes in human activity and demography, including landscape use, cultivation practices, population size, and social systems. By linking temporal trends noted in the archaeobotanical data with indicators of human behaviour and processes of landscape domestication, the present study will be able to consider the role of tree crops in agronomic systems in this setting, and to study the evolution of the Marquesan arboreal economy. These findings will be used to evaluate which factors have been most influential to the development of a system based on tree cultivation in this setting, and to broadly consider the encompassing process of landscape domestication in tropical Polynesia.

Domesticated landscapes in Polynesian prehistory The archipelagos of the tropical Pacific vary considerably, from small, sandy, relatively flat atolls to rugged high islands that can be anywhere from thousands to millions of years in age. Westward islands are home to a variety of plant and animal taxa originating from Southeast Asia, which attenuate gradually to the east, and eastward locations include elements from the Americas. Because of their relative isolation, the flora and fauna of the eastern archipelagos have a high degree of endemism (Keast and Miller 1996; Mueller- Dombois and Fosberg 1998; Steadman 2006). Polynesian explorers correspondingly encountered both subtly, and at times radically, different biotic and abiotic environments to those of their land of origin as they settled the islands of this region. While marine foods were often plentiful in areas not previously inhabited by humans, terrestrial resources were more restricted. Faunal resources were (except for the westernmost archipelagos) limited to several species of landbirds and bats. Edible plants were very limited in type and palatability, being more useful for utilitarian or medicinal purposes. A number of domesticated plants, including important starchy crops such as breadfruit, bananas, taro, yams, and sweet potato, were introduced throughout the Pacific in prehistory (Barrau 1961; Massal and Barrau 1956; Yen 1985). Native plants of various economic importance, such as coconuts and Pandanus, were supplanted by introduced versions that yielded more desirable products. Overall, the biota introduced by settlers formed a critical component of the landscapes and subsistence systems that developed in Pacific prehistory.

Pacific production systems Traditional subsistence practices in the Pacific encompass a wide variety of techniques and involved the management of entire landscapes (Barrau 1961; Yen 1980, 73; Figure 2.4). There are differences at various levels: individuals within a community may devise different practises, local landscapes may contain several different components, and there are variations across landforms as topography, hydrology, and edaphic conditions change. The suite of practices used in Polynesia include shifting cultivation, intensive dry field cultivation, irrigation, arboriculture, and animal husbandry, which are of varying importance (Kirch 1991a). These components are part of a total solution that forms a mosaic, which needs to be considered in the larger context of landscape. It encompasses the

19 sophisticated maintenance of soil fertility, the management of water, and the balancing of potentially irregular food supplies (Clarke 1994; Kennedy and Clarke 2004).

Figure 2.4: Schematic representation of Pacific land use on a high island (from Barrau 1961, Fig. 11a).1

1 High island of Rarotonga example: 1, ocean; 2, coral calcimorphic soils growing coconuts, sweet potatoes, tomatoes; 3, hydromorphic soils, taro gardens; 4, colluvium clay loam, citrus, and gardens; 5, lateritic soils, scrub, and secondary forest.

Agroforests and multi-cropping can reduce the risk of food shortages, and manipulation of individual plants can create intra-species diversity that produces various crop-yielding plants with different maturation dates. Techniques for long-term storage are also important; some root crops can be stored in situ, nuts can be dried or smoked, bananas and processed starches can be dried, and breadfruit and bananas can be ensiled, the former for very long periods of time (Barrau 1961; Clarke 1994; Yen 1974b). Techniques and tools have been passed down the generations and traditional knowledge systems are sophisticated. One important aspect of cultivation in the Pacific is that it is a social act, and land use and rights of access are the main source of cooperation and conflict in both kinship and political systems (Malinowski 1935). These systems are, at times, extremely complex, encompassing various layers of ownership for land, trees, and even individual branches (Tautain 1897, 541). The cultivation of trees is a tradition of great antiquity, dating back at least as far as the late Pleistocene (Kennedy and Clarke 2004, Table 1). It has played an important role in discussions about the origins of agriculture in the region as a whole (Denham 2004; Gosden 1995). The subsistence practices used by the Lapita people, ancestors of the Polynesians, included a fully developed tree cropping component (see Kirch 1989). Prehistorians have long questioned how and why ancient agroforestry became an important practice in western Oceania. Tree crops may have been an efficient means of producing more food as

20 populations became larger over time, a benefit to both humans and livestock that were fed agricultural produce. Agroforestry can maximise the ‘vertical capacity’ of an agricultural landscape (Kirch and Yen 1982; Latinis 2000), which is often particularly important in island contexts. Yen (1974b, 278) noted that for a minimal labour investment, arboriculture can be integrated into most types of agriculture, acting as stabilising insurance against the seasonal spread of production. The propagation of useful trees can also be labour-efficient when compared to practices that require soil tillage and water management schemes. Tree crops also offers diet variation, not only as the source of fruits and nuts but as the basis for both animal husbandry on the coasts and low islands, and hunting on the high islands. Latinis (2000, 43) has argued that arboricultural practices can provide long-term stability and have evolved in response to highly variable environmental conditions in situations where strategies with shorter-term payoffs would have failed. Tree crop cultivation may have been a prehistoric intensification strategy, a topic that has been explored by archaeobotanists studying the secondary products revolution in the Near East (e.g., Fall, Falconer, and Lines 2002) and the development of ancient agroforestry systems in the Americas (e.g., van der Warker 2005). In Western Oceania, these ideas were pioneered by Yen (1974b) as they pertained to the Solomon Islands, and Kirch and Yen (1982) for the island of Tikopia. However, despite the importance of tree crops in prehistoric subsistence, arboriculture has “perhaps been the most overlooked [form of agricultural intensification] by ethnographers and prehistorians alike” (Kirch 1994, 10).

Evolution of Pacific production systems The development of Polynesian agriculture has been defined as a three-step process that involves adaptation, expansion, and intensification (Kirch 1984; 1994). While this summation describes the big-picture process, distinct variations in tempo and characteristics of change exist. Despite having similar origins, the landscapes and social-political structures that foreign visitors encountered in the early contact period varied (sometimes considerably) from location to location. As previously discussed, both nature and culture played important roles in the evolution of landscapes globally. In Pacific Island settings, the key elements of variation included chance and the founder effect (Yen 1990). Some plants were propagated mainly vegetatively (Yen 1985, 323), producing a genetic bottleneck. Not all elements of the ‘transported landscapes’ of the Pacific were introduced to every location, and there is a

21 notable eastward decline in the inventories of introduced biota. Some items failed to thrive in new settings, or were extirpated at some point in prehistory because they were in direct competition with humans for food resources (e.g., Giovas 2006). Ancestral practices were also fundamental formative factors, as cultivation techniques learnt in the homeland were replicated in new settings. But while methods of landscape transformation and cultivation are similar throughout Oceania, emphasis was not always placed on the same strategies as people adapted to local conditions. Moisture was a key component to how food-production systems evolved in Pacific settings (Barrau 1961, 1965). It has been asserted that an “excess or lack of water probably was one of the main ecological factors which determined Man’s earliest horticultural adaptations to his environment in the Indo-Pacific area” (Barrau 1965, 329). Island size and age, particularly as they related to soil nutrients, were also important abiotic factors (Vitousek 2004). Against this changing backdrop, the decisions people made varied as landscapes changed and populations grew. These decisions sometimes fostered competition for resources, and at other times imposed restrictions on activity. After comparing several case studies extensively, Kirch (1994) concluded that the subsistence strategies that had evolved and the intensification pathways seen in the archaeological record were influenced as much by local environment as social factors, as previously discussed. Closer examination of some of the examples illustrates what is currently understood about the differences between various prehistoric processes and, ultimately, hints at the intricate steps involved in domesticating the landscapes of Polynesia. Studies from Tikopia, Mangaia, Mangareva, and the Hawaiian Islands (Kirch 1994, 2007a) were reviewed to gain an understanding of the evidence available to support these conclusions. In these studies, human arrival on islands is considered marked by sudden increases in the charcoal concentrations in sediment cores and archaeological deposits, and the rapid depression of native avifauna, reef or lagoon fish, and shellfish as people exploited naturally occurring resources. Fossil evidence of other aspects of this process are more limited, but well-defined in theory. These included making environmental modifications remembered from the homeland, and introducing cultigens into the new setting (Yen 1990). Yen notes that a period of experimentation followed, characterised as a process of ‘disintensification’, during which colonists attempted to reassemble their agricultural economy in a new setting. The ebb and flow patterns (Yen 1990, 271) of the process were important reactions to the success or failure of species or technologies in the new environment, and secondary colonisation then involved repeated—and possibly new—

22 introductions. During this period, there was also a need to stockpile food and expand to the extent that future crops were ensured. A further review of these case studies demonstrated that while reconstructions of the process of agricultural development are highly resolved in some areas, they are more generalised and lack detail in others. Much is known about the evolution of these systems in the Hawaiian Islands, for example, where the large landmass, extensively managed field systems, and highly stratified social structure that developed there facilitated elaborately managed productive landscapes (Kirch 1994, 2000). Very different outcomes were noted in the other locations. On the small island of Tikopia, people developed a sustainable and intensive agroforestry system that supported extensive bird life by mimicking the native forest, and coupled this with tightly managed population controls (Kirch and Yen 1982). But on Mangaia and Mangareva (Kirch 2007a, 90–1), techniques such landscape burning and swidden cultivation are thought to have caused extensive erosion. The highly weathered ancient soils of these older islands, coupled with the loss of soil nutrients that had been previously distributed by native birds, may have caused swiddening practises to be abandoned as fields became less productive. As populations grew, cultivation systems expanded and intensified and as serious competition for resources emerged, so did a warrior class. Techniques such as pondfield agriculture were thought to have become more important at this time, and archaeological evidence indicates animals that directly competed for food resources with humans were extirpated.

Synthesis In the foregoing, the general prehistoric patterns of land use in Oceanic settings were described, and several case studies were used to illustrate variations in the key processes of landscape domestication. Certain factors identified in this literature as having influenced Polynesian socio-ecosystem developments have also been highlighted. The need to examine multi-scalar data, including data from fine-grained processes, and to consider the influence of both natural and social factors guided the evaluation of what influenced these outcomes. Several gaps in our understanding of the process of landscape domestication in Central East Polynesia were identified: (1) knowledge of the characteristics of landscape modification and food-production systems during colonisation and settlement; (2) an understanding of the characteristics and tempo of vegetation change at localised scales, especially elements that are poorly represented in pollen spectra; and (3) knowledge of the

23 relative importance of tree crops during any period in prehistory. Addressing these gaps could help address both theoretical and practical questions, such as: What role did tree cultivation play in any of these processes? Is tree-crop production an intensification strategy, or more of a risk-minimisation strategy, in the region? When in the settlement period did distinct cultivation solutions emerge? To what extent did these solutions vary, and what factors were most influential to the outcomes? How did the process of landscape transformation impact other biota and well as social processes? A detailed assessment of how woody vegetation factored into these processes would help to address these questions. An important concern is that intensive arboriculture and the long-term storage of tree crops were innovations in the Marquesas Islands (Kirch 2006, 203–6), and the present study provides an important opportunity to study these elements of the production system.

Case study Our understanding of landscape domestication in Polynesia has been informed by the records of early foreign visitors, who documented a historic endpoint, and previous archaeological and palaeoenvironmental research has largely been focused on reconstructing food-production practices at local and regional scales (e.g., Addison 2006; Allen 2004a; Barrau 1965; Coil 2004; Kirch 1994, 2011; Kirch and Yen 1982; Ladefoged and Graves 2008; Lepofsky 1994). A theoretical model of the prehistoric process has been outlined by McCracken and colleagues (McCracken et al. 1999), within which trajectories of change and factors that influenced variation in outcomes were considered, and Yen (1989, 1990) has described a detailed theoretical model specific to Polynesia. Using these framework and data, several possible scenarios that could have led to the development of a production system centred on tree crops in this location have been hypothesised. Arboreal cultivars may have been an important component of the production system from a very early period and remained so into the contact period, or they may have become more important over time as food production intensified, or it is possible arboriculture became more important in some locations than others due to the influences of local environments or historical contingencies. These alternatives will be considered in this thesis, along with the new information that a study of tree cropping practices contributes to our understanding agricultural development in the region. In undertaking such an analysis, it will be important to evaluate and identify the factors that were most influential to the development of a production system based on arboreal cultivars.

The Marquesas Islands were selected as a case study to evaluate socio-natural landscape processes, intensification, and risk management in prehistory. Historical literature emphasised the prevalence of tree crop cultivation as early as the first European contact in 1595 (Quirós 1904, 28), and over the course of sustained contact it was described as the dominant mode of food production (Ferdon 1993, 89). This outcome was notable in eastern Polynesia, where—although breadfruit was plentiful and well-liked in locations such as Tahiti—herbaceous crops were also common dietary staples (Ferdon 1981, 189–94). However, little is presently known about the character and tempo of changes to the Marquesan food production system in prehistory. Several hypotheses have been put forward regarding how and why it arrived at this endpoint (e.g., Addison 2006, 296–305; Kirch 1991b, 128–9; Millerstrom and Coil 2008, 347) although, as some of these authors admit, the evidence is not substantial and is based on few dates. Archaeological evidence does indicate that Marquesan lifeways went through a major shift some time between AD 1400– 1650 (Allen 2009a, 2010), and changes to subsistence practices may have followed. The present study might provide evidence as to when tree cropping became important, and assist in developing a chronology of agricultural change for the Marquesas. Tree crops are ideal subjects through which to study past human subsistence practices in this archipelago; while there are significant challenges to the study of this material in other regions of the world, these problems are diminished in the islands of central East Polynesia. The native island flora of this region contain few arboreal taxa that were of economic importance, and humans introduced a limited number of taxa in prehistory (Whistler 2009; Table 3.1). Pilot studies of charcoal assemblages from sites on Nuku Hiva produced rich floral assemblages from various temporal contexts (Huebert 2008, 2009; Millerstrom and Coil 2008, 347) suggesting that long-term natural and social change could be studied using these materials. The availability of charcoal assemblages from several different locations facilitates examination of the process of vegetation transformation on several well-defined scales, and localised patterns of change and long- and short-term trends can be studied in several controlled spatial spheres. Factors that could have influenced variability are held constant in these settings. Differences in timing or taxa identified may potentially relate to initial settlement dates, the timing of various plant introductions, and population densities. They may also be environmental, relating to variations in valley size, micro-climate conditions, or hydrology – factors that will be considered in turn. In conclusion, there are several potential contributions of the present research. The Nuku Hiva charcoal assemblages can inform on the processes of human-induced vegetation

25 change and agricultural development. It is of particular interest that these materials have the potential to inform on activities in the early settlement period, providing details about the native vegetation that was modified by settlers and changes in plant environments that corresponded with the rapid avifaunal resource depression that occurred in this period. Currently, these processes are only understood at low resolutions in much of Central East Polynesia. Secondly, large charcoal assemblages from this location may be able to provide a long-term record of human-vegetation interactions on a Pacific high island, and allow for the creation of one of the most well-informed views of the process of landscape domestication in Polynesian prehistory. Thirdly, the results of this study could provide insights into the tree-cropping component of food production systems, and data to evaluate related aspects of intensification and risk-management that are not well understood. To conclude, it should be stated that the present study is focused largely on vegetation reconstructions. It is acknowledged that charcoal assemblages also inform on economic activities such as fuelwood collection and use; these activities are discussed in the course of analysis, but a formal study of resource use is not a part of this thesis.

Chapter 3: Natural and Cultural History of the Marquesas Islands

In this chapter The Marquesas are a group of high islands located in Central East Polynesia. The landscape is rugged and vegetation cover varies from dense tropical forest to open, eroded scrubland. Landscapes at low elevations have been modified over centuries of human interaction with soil, water, and vegetation, and though there is evidence populations were once distributed widely, today most people live in small villages near the coast. Climate systems that affect this location are highly variable and have had a distinct impact on both landscape and society. In this chapter, background information about the setting for this study is presented, including a review of landforms, climate, and vegetation. Climate regimes that affect the archipelago are characterised, and modern meteorological data are summarised along with what is presently known about prehistoric climate in the region. The vegetation zones of the islands are also described with an emphasis on woody flora, to illustrate the range of variation in plant cover. To provide cultural context for the present study, a brief culture history is presented along with an ethnohistory that emphasises traditional uses of plants and foodways. To conclude, a brief arboreal ethnobotany outlines the trees and shrubs that were known to be important in traditional Marquesan culture. This review provides context for the interpretation of archaeological plant materials that is presented in the final part of this thesis.

Geography, geology, and soils The Marquesas Islands are an archipelago of eight volcanic high islands and a number of small islets and seamounts in the central eastern Pacific Ocean near the equator, located at approximately 9°11'S, 139°41'W. They are distant from other landmasses; the Society Islands are 1500 km to the southwest and the Hawaiian Islands are 3500 km to the northeast (Figure 3.1). The islands fall into two geographic groupings: a northern cluster includes Nuku Hiva, Ua Pou, Ua Huka and a number of islets including Eiao; a southern group includes Hiva Oa, Tahuata, Fatu Hiva, and several small islets (Brousse et al. 1978; Decker 1970).

Figure 3.1: Map of the Marquesas Islands in Pacific context.

The larger Marquesan islands are remarkable for their high-elevation ridges, deep valleys, and rugged coastal cliffs that drop sharply into the ocean. Coastal plains are limited to the valley mouths or narrow valley bottoms. There are no fringing reefs because below sea level the islands drop off steeply to the ocean floor, though isolated reefs do occur in a few protected embayments (ORSTOM 1993). The mid-Holocene highstand is thought to not have had a significant impact on palaeoshorelines in this group (Dickinson 2003, 497).

Figure 3.2: North coast of Nuku Hiva, showing landforms and coastline (photo courtesy of Melinda Allen, 2010).

The islands range from 1.8–5.5 ma (mega-annum) in age, similar to major islands of the Hawaiian (0–5.5 ma) and Society (.5–4.2 ma) archipelagos (Clouard and Bonneville 2005). The substrate is highly eroded basalt underlain by many contrasting layers of volcanic material (ORSTOM 1993). Soils in low-lying areas are enriched by eroding volcanic rock and mineral-rich water flowing from upland (Decker 1970, 21–23). Rich soils in well-watered areas, especially in back valleys, support diverse plant life and dense forests. Floodplains in the few elongate canyon-shaped valleys that have sizable permanent streams also support lush vegetation.

Weather, climate, and water The Marquesan climate is mesic-tropical, with mean temperatures ranging from 25° to 27° Celsius (Cauchard and Inchauspe 1978; Decker 1970, 25–27). The southeast Pacific trade winds blow from east to west, and the archipelago lies near the chilly Humboldt current that flows westward from the South American coast. High winds are rare. While annual temperature varies little, rainfall volumes can vary widely based on geography and oscillating climate cycles. High-elevation areas are typically very moist and often shrouded in clouds. Precipitation is strongly influenced by orography. Windward northern and eastern regions of Nuku Hiva (for example) receive up to 1500 mm of rainfall annually, while the western areas are appreciably drier with rainfall as low as 700 mm per annum (ORSTOM 1993). There is a rainy season from January to July, and a drier season in August to December, though accounts vary and it has been noted the seasons are irregular (Adamson 1936, 17). Large islands such as Nuku Hiva and Hiva Oa can receive twice as much rain as the others in the rainy season, and there are distinct differences in precipitation between windward and leeward locations in the dry season (Cauchard and Inchauspe 1978, 75–100; Mueller-Dombois and Fosberg 1998, 446). The Marquesan climate differs from other central Pacific archipelagos because it lies between two wet zones; the rainy season does not parallel the Society or Cook Islands where it begins and ends much earlier (Decker 1970, 42–45; Mueller-Dombois and Fosberg 1998, 387–388). Additionally, the islands are subject to periods of high rainfall and dry conditions. Multi-year droughts can occur during prolonged La Niña phases, causing vegetation to wither and die and crops to fail; famine related to extreme drought was reported by some early visitors and in some oral traditions (Allen 2009a; E. Handy 1923, 7–9; Robarts 1974).

Permanent streams run through most of the larger valleys and intermittent streams are present in many places. The latter, which are often dry for much of the year, can turn into highly erosive torrents during heavy rains (Crook 2007, 96; Decker 1970, 25–29). Robarts (1974, 240) once noted that entire trees washed downstream after multi-day torrents, and I have observed high-altitude waterfalls in Taipivai Valley on Nuku Hiva flow voluminously after heavy rain and streams swell and surge reddish-brown water through low-lying areas carrying sediment, rocks, and branches. Few ponds and swampy areas exist, and many accumulations drain away quickly after heavy rain. Severe tropical storms only occasionally affect the islands of this archipelago, occurring less frequently than the four to eight per century that strike the Society Islands to the southwest (see Parkes and Flenley 1990, 3). Natural disasters affect the archipelago on occasion. The islands are particularly vulnerable to tsunamis. The powerful Aleutian earthquake of 1946 sent waves a half- kilometre inland in places (Decker 1970, 18), depositing sand and debris and exposing some deeply buried cultural deposits (Suggs 1961). On Nuku Hiva, damage from the 2011 Tōhoku earthquake tsunami flooded low-lying coastal areas in several valleys with salt water, killing halointolerant trees and washing sediments from the mouth of the Taipivai river some distance inland. Current data on paleotsumanis affecting the Pacific Islands are not highly resolved. There is some evidence from Hawaii that high-magnitude tsunamis may be visible in the late historical archaeological record (Chagué-Goff et al. 2012), though the authors point out that their study is only a first step in a more comprehensive regional analysis. There is evidence in sedimentary records from Wallis and Futuna that a large tsunami event, which may have reached an area stretching from northern New Zealand to the Marquesas, occurred in the mid-15th century AD (Goff et al. 2010, 2011, 2012).

Palaeoclimate and vegetation of the Central East Pacific While palaeoclimate data provides important information on the environmental conditions experienced by past peoples, prehistoric climate is challenging to study in this region for several reasons. First, primary meteorological data has little temporal depth in the Marquesan archipelago. Prior to the 1960s, records are fragmentary making analysis of all but recent climate trends challenging. Second, there are few catchments on the islands that are suitable for coring, which limits the study of both past sedimentation regimes and broad- scale vegetation change. These conditions also limit the evaluation of catastrophic natural

30 events that may have affected the area in the past. The following discussion provides an orientation to forces that broadly shape climate conditions, and review what is currently understood about past climate regimes in the Central East Pacific. These data are presented to consider the complex interplay of climate and human activity in prehistory.

Forces affecting climate in the region The El Niño/Southern Oscillation (ENSO), a climate system that originates in the Pacific, has far-reaching effects on weather both regionally and worldwide (U.S. CLIVAR Project Office 2013, 1). Regional climate is greatly influenced by these cycles, which are caused by fluctuations in the Pacific’s oceanic-atmospheric system. ENSO originates in tropical latitudes as a result of variations in sea surface temperature, winds along the equator, and other factors, and results in multi-year periods of extended and/or heavy rainfall (El Niño years) and alternating periods of dryness (La Niña years). The severity of events is of particular interest to modern-day climate scientists who are attempting to predict climate cycles that have extreme fluctuations in intensity, an aspect challenging to reconstruct from proxy data (Emile-Geay et al. 2013a). It is also a concern that ENSO cycles in the central and eastern Pacific may be somewhat independent (see U.S. CLIVAR Project Office 2013, 3). Climate in the region is further influenced by longer Interdecadal Pacific Oscillation (IPO) cycles. These cycles shift every 10 to 30 years alternating between cool, dry and warm, wet conditions (Linsley et al. 2008, Table 4). IPO cycles have been correlated with some modern Marquesan climate records, and it was noted that when climate is warmer and wetter during positive cycles, El Niño and La Niña events become more pronounced (Allen 2010). These cycles are especially evident in Nuku Hiva rainfall records from 1983 (an El Niño year) and 1991 (a La Niña year) (Figure 3.3). When the two cycles converge, Allen (2010) proposes that a combined La Niña / negative IPO phase would have negatively affected crops in some Marquesan locations while El Niño / positive IPO convergences may have increased flooding in others. Several other external forcing factors that operate on long time scales, or that occur sporadically, can be powerful forces in climate variability (Jones and Mann 2004, 25–27). Fluctuations in solar irradiance occur on multi-centennial scales and affect climate by cooling the waters of the Eastern Pacific when radiation is high (Emile-Geay et al. 2013b). Other major climatic events can be forced by large volcanic eruptions and other natural

Figure 3.3: Rainfall variability on Nuku Hiva Island (from Allen 2010).

disasters (e.g., Grattan and Torrence 2007; McFadgen 2008). Modelling the reach and effect of these events in prehistory is challenging (Jones and Mann 2004, Fig. 7) but proxy data from a 700-year tree ring sequence taken from 2,222 trees in both tropical and sub-tropical regions has shown that volcanic eruptions may elicit an ENSO-like response in the tropical Pacific (Li et al. 2013).

Palaeoclimate data The foregoing summary of climatic regimes sets the stage for a review of what is currently known about past climate conditions in the central Pacific basin. Proxy records such as tree-ring sequences and ice cores, while used effectively in other areas, are generally under-represented for this region (see Jones and Mann 2004), however these data are frequently updated and amended as sophisticated climate simulations are compared with new empirical data. Reconstructions such as these inform on factors that affect climate change at regional and global scales. Climate data for the region have been reviewed in recent years by Allen (2006), whose major ideas are summarised below alongside more recent findings.

Proxy climate and vegetation data from the Marquesas is limited to the analysis of a several sediment cores. Several cores taken from Hatiheu Valley have produced data with very limited temporal depth (Gourdon 2003), but material taken from one location on Nuku Hiva has been more productive informing on sedimentation regimes, background climate, and modestly on past vegetation during a ca. 750-year sequence (Allen et al. 2011). Allen and colleagues studied three sediment cores from marshes on the Tōvi’i plateau and the Taipivai and Hatiheu valleys, and determined that they developed sometime in the 13th or early 14th century AD, a period after human settlement of the island. Rates of sediment accumulation increased rapidly in the 14th century AD, stabilised, and then accumulated more rapidly again in the mid-17th century AD. Allen and her co-authors conclude that these fluctuations represent an earlier, wetter phase followed by a long drier period, and finally another wet period late in the prehistoric sequence. The pollen spectra of the Tōvi’i core contained mainly ferns and herbaceous plants, a composition that fluctuated only modestly over time, and tree pollen made a minimal contribution to the assemblage. A number of lines of evidence indicate that ENSO warm events intensified in the late Holocene, resulting in more frequent El Niño-like conditions in the eastern equatorial Pacific during this period (Cabarcos, Flores, and Sierro 2014; Moy et al. 2002). Some of the most direct evidence for past regional climate regimes has come from the analysis of oxygen isotopes taken from coral cores on Palmyra Island. These data show that while many parts of the Northern Hemisphere were experiencing cooler temperatures during the Little Ice Age, mean conditions in the central Pacific’s Line Islands were warm and wet from the mid-16th century AD onwards (Cobb et al. 2003). The authors noted that during the mid- 17th century AD, climate fluctuations became extreme in this region. In addition to finding that overall mean climate conditions were warmer and wetter than previously thought, inter- annual climate was extremely variable during this period. These findings are of particular importance to the present study because the Line Islands have been grouped into the same climate sub-zone as the Marquesas Islands (Figure 3.4). Also, Allen and colleagues (2011) have reported that sedimentary regimes noted in the Nuku Hiva cores correspond with the aforementioned wetter (and warmer) climate conditions at Palmyra. They also correspond with increased sedimentation observed at several archaeological sites on the island, including probable wetter conditions after AD 1640 in the windward Anaho and Hakaea Valleys (Allen 2004b, 2009a, 2010; Allen and McAlister 2010).

Figure 3.4: Climatic response regions of the southwest Pacific (from Allen 2010, after Salinger et al. 1995).

Several findings from the Society Islands, high islands closest to the Marquesas, are well-resolved and very informative regarding past climate, vegetation, and human history. First, cores from the high-altitude fresh water Lake Vaihiria on Tahiti yielded an informative pollen sequence that included a rich and diverse assortment of trees, shrubs, and herbs (Parkes 1994; Parkes, Teller, and Flenley 1992). Three distinct phases were identified in a sequence that spanned approximately 500 years BP in the history of the lake. The earliest deposit, which accumulated over approximately 200 years, showed relatively stable deposition with several periods of rapid accumulation. Many primary forest trees were evident, and tree fern pollen was very abundant at this time. The authors reported that several disturbances could be seen in this phase, and remarked that some primary forest species were replaced with secondary species and others disappeared altogether. They suggest that this finding and sedimentary indicators probably relate to large landslide events, and are possibly linked to periods of extreme weather in the 17th century AD. The authors are careful to note that human disturbance may also be a contributing factor, as it is known that people were using other interior areas of Tahiti at this time (e.g., M. Orliac 1997). Later sediments contain evidence of the influence of humans on forest composition, as primary forest trees decrease and secondary arboreal taxa become more frequent, while Pandanus pollen dramatically increased.

In a related study, Parkes (1994,1997) also reported on a pollen and sediment core from Lake Temae, a small brackish body of water in the lowlands of Mo’orea Island. These findings are useful because first, they provide a centennial sequence of past storm events in the region and second, it is demonstrated that these events affected vegetation before human settlement. Parkes’ analysis provides valuable information on the periodicity of ENSO cycles (every 5–7 years), weather events operating on longer cycles (80–90 years), and extremely long-cycle events that had far-reaching effects on vegetation and sedimentation (every 140 years) in the central Pacific between the 4th to 8th centuries AD. It was noted that after several natural disturbances before AD 650, vegetation recovered quickly, while after the late 8th century AD vegetation changes were more widespread and long-lasting, which the author attributed to more intensive human settlement, admitting the exact causes are not yet well understood. Evidence of extreme weather events was also observed in the archaeological work in the ‘Opunohu Valley of Mo’orea (Lepofsky 1994; Lepofsky, Kirch, and Lertzman 1996). There, a rapid increase in sedimentation occurred on the valley floor sometime in the 7th century AD, and Lepofsky determined that this event was probably not caused by human activity as the area was only used for extensive agriculture as early as the 13th century AD after populations increased and expanded inland (Kahn 2011; Lepofsky 1995; Lepofsky and Kahn 2011). To summarise, direct data on prehistoric climate and vegetation of the Marquesas is limited. Some important findings have been presented in recent years that refine our understanding of climate cycles in the region, and factors influencing these regimes are becoming better understood. It has been established climate in the area is highly variable. Though extreme events have been identified long before human occupation, there are findings of importance regarding environmental conditions and vegetation in human prehistory. Perhaps the most important is that from the 16th century AD, conditions in the Marquesan archipelago were warmer, wetter, and potentially stormier than they had been previously. These changes correspond with increases in sedimentation, observed in marshes and archaeological deposits on Nuku Hiva, and there is also some evidence conditions were occasionally very dry. Though it was noted that the Society archipelago is subject to somewhat different climate and weather patterns than the Marquesas, similar shifts may have also been occurring there during this period though in the palaeoenvironmental record, the effects of climate change could not be easily distinguished from changes affected by humans. Overall, it is likely that a combination of both natural and cultural influences impacted landscape and vegetation during the Marquesan prehistoric sequence.

Vegetation zones of the Marquesas The Marquesas Islands host a native flora of approximately 320 species and many genera are composed of a single species, many of which are endemic (Florence and Lorence 1997). Approximately half of the endemic plants in this archipelago are found in high- elevation cloud forests, and high degrees of endemism such as this are characteristic of terrestrial biota in isolated locations (MacArthur 1967). A number of endemic lowland trees are currently under permanent protection by the government of French Polynesia (Meyer 2007), including Lebronnecia kokioides (Malvaceae), Santalum insulare (Santalaceae), and the palm Pelagodoxa henryanum (Arecaceae) (Butaud 2010; Meyer and Butaud 2009). Little is known about the native flora of the Marquesas Islands, and the forests have been highly modified by centuries of human and animal activity. Polynesians domesticated the lowland native forests, and European and American visitors induced further changes through the introduction of ornamental plants, vegetable crops, and grazing animals in the 19th and 20th centuries. Perhaps the largest change has been affected by feral ungulates who graze heavily on native vegetation leaving large areas of land, sometimes in fragile upland areas, in a very degraded state (Decker 1970, 1991; Florence and Lorence 1997). Gregarious, fast-growing shrubs and trees have also replaced both indigenous and Polynesian-introduced flora in some lowland areas (see Meyer 2004), a particular threat to bio-diverse dryland forests (Gillespie et al. 2011). Though plants were collected from this archipelago by foreign visitors as early as 1774 (Nicolson and Fosberg 2004), the Marquesan flora were only first described in detail by del Castillo (1892) and Brown and Brown (1931). In the 1960s, geographer Bryce Decker (1970) extensively described numerous Marquesan plant communities as part of his geography dissertation, and briefer treatments were later given by Hallé (1978) and Florence and Lorence (1997). Some of these sources were synthesised, though unevenly, in a comprehensive volume on the vegetation of the tropical Pacific Islands (Mueller-Dombois and Fosberg 1998). Recently, the flora of the Marquesas has been the subject of intensive study by botanists at the Smithsonian Institution and the Direction de l’environnement de la Polynésie française, work that began many decades ago by F. Raymond Fosberg and Marie- Hélène Sachet, which resulted in numerous plant collections being consolidated into a searchable online database and a forthcoming printed volume (Smithsonian National Museum of Natural History n.d.; Wagner and Lorence 1997). Many Marquesan plants have also been described in a partially published series on the flora of French Polynesia (Florence 1997, 2004). Though the vegetation zones of the Marquesas have been described and

36 grouped somewhat differently by the aforementioned observers over the years, the eight zones outlined by Mueller-Dombois and Fosberg (1998) are the most informed in regional context and the woody vegetation of each is summarised below. Zones pertinent to the present study (marked with an asterisk) are described in greater detail than others, and the dominant taxa in each zone are underlined. Additional information is presented from informal surveys conducted on Nuku Hiva in Anaho and Hatiheu Valleys under the guidance of a village elder (Tioka Puhetini), and Hakaea and several upland locations in the company of Moetai Huioutu in 2010 and 2011. Several zones are illustrated in Figure 3.5. Coastal strand zones* occur at valley mouths immediately behind the high tide line. These zones are very limited in the Marquesas due to the islands’ small and narrow coastal plains. Common trees include Thespesia populnea, Hibiscus tiliaceus, Cordia subcordata, Calophyllum inophyllum, Terminalia catappa, Cocos nucifera, occasionally Guettarda speciosa and Cerbera manghas, and other smaller shrubs and herbs. Most of the aforementioned trees are native, but in some instances modern introductions are also found in these locations. The zone has frequently been remodelled in populated areas and retaining seawalls have sometimes been installed. Xeric sea slope vegetation occurs along the many kilometres of seaward slopes and cliffs that fringe many of the Marquesan islands, extending upwards to elevations of 400 or 500 m in altitude. Today these areas are very degraded and contain a variety of native and introduced plants including Miscanthus grass and weedy herbs, the shrub Cordia lutea, and occasional Casuarina equisetifolia, Pisonia grandis, and Ficus trees. Xerotropical lowland zones* receive less than 2000 mm annual rainfall. These zones extend in leeward areas to 300–400 m elevation, though in places can be found at much higher elevations. This vegetation has been highly modified and patches of native forest are rare. The dry forests of the Marquesas have been divided into two sub-types (Mueller- Dombois and Fosberg 1998). One type occurs from behind the strand to approximately 300 m elevation and is dominated by Pisonia grandis, a large tree that forms a thick and closed canopy. Pisonia is or, more correctly was, common in dry areas of many high islands in the region in the early 20th century. Other large trees found in this zone include Thespesia populnea, Calophyllum inophyllum, and Terminalia with an under-story of native shrubs. Hallé (1974) remarked that this forest type was disappearing under pressure from grazing animals, and at the time of his survey was best visible on small, uninhabited islands such as Mohotani. The second type occurs at low to mid-elevations but can stretch to higher altitudes that receive less than 2000 mm annual rainfall. Vegetation there has been highly

37 modified by feral grazing animals, and remaining patches of native forest are limited. In low to medium elevations of this zone, as around Taiohae on Nuku Hiva, dense monospecific stands of the spindly, introduced Leucaena leucocephala grow in untended areas at the village margins. Florence and Lorence (1997) also report that Cordia lutea, Erythrina variegata, and Casuarina equisetifolia can be principal forest components in some dry lowland locations, but vegetation in dry leeward areas has frequently been reduced to scattered trees and low scrub. In a recent survey of Marquesan dry forests (Gillespie et al. 2011; Terpkosh 2010), which included transects in uninhabited locations on Nuku Hiva and Hiva Oa, forests in this zone were found to frequently and abundantly contain the trees Xylosma suaveolens, Cerbera manghas, Maytenus crenata, and Ficus prolixa. In some areas, Sapindus saponaria, Eugenia reinwardtiana, Thespesia populnea and Morinda citrifolia were frequent. Glochidion marchionicum and Premna serratifolia were also noted. Gillespie and colleagues concluded that species richness in patches of Marquesan native dry forests were low, few endemics were present, and all but one (Morinda citrifolia) of the most frequent and abundantly occurring trees were native to the island group. While their study describes impoverished modern-day dryland forests, there are indications that dry forests of the high islands of tropical Polynesia were richer in the past. Several studies have shown that the flora of these zones were diverse in Hawaiian prehistory (Allen and Murakami 1999, Athens et al. 2002), and the naturalist Joseph Rock (1913) described low-elevation Hawaiian dry forests at the turn of the 20th century as home to a variety of native and endemic trees and shrubs, about half of which are rare or endangered today. Pluviotropical and transition zones* contain mixed mesic and xeric vegetation elements and occur above the cultivated forest remnants. The two zones often inter-grade, and the upper and lower boundaries of the transitional zone can be indistinct. Decker (1991) considers each a separate zone even though his species lists overlap to a great extent, and the two zones were combined by Mueller-Dombois and Fosberg (1998) who note there is no clear distinction between them. Fewer than half of the species found in this zone today are native. In drier areas, some native trees can be abundant including Sapindus saponaria and Xylosma suaveolens, which together can dominate forests in transitional areas (Decker 1970, 24–25), and dense thickets of Leucaena can also be found in some locations. Other trees that can be found in uninhabited valley bottoms of this zone include Ficus, Pandanus, Pisonia, Celtis, and Cordia. In ravines, Thespesia, Erythrina, papaya, Morinda, Psydrax, and Premna can also be found. Transitional forests of the upland slopes are similarly

38 populated, though large trees are absent. Ridge crests in the transition zone also contain similar arboreal elements, but Pandanus and Hibiscus are common along with Glochidion, Psydrax, and coffee, and groves of Casuarina are sometimes found on ridges above the villages. In moister (pluviotropical) areas of this zone, Hibiscus and Aleurites are common, along with Barringtonia, Calophyllum, Inocarpus, Terminalia, Pipturus, and others such as Glochidion, Morinda, Pandanus, and the few remaining sandalwood trees. Decker (1991) notes some strand trees also occur occasionally in interior locations of this zone, including Cordia, Erythrina, Thespesia, Guettarda, Premna and others, transported by human or animal activity. Cultivated forest remnants* are often mesophytic (moist) forests that have been highly modified by prehistoric human activity. These valley forests are almost completely anthropogenic, created by hundreds of years of traditional arboriculture and agriculture. Valley bottom-lands near the sea are, today, home to most modern Marquesan dwellings and gardens where traditional vegetable and fruit crops such as breadfruit, taro, yam, sweet potato, and bananas, and historic introductions including pineapple, mango, and avocado are cultivated (Addison 2006; Decker 1970, 160–64,141–42). Forests found inland of these areas are frequently composed of economic trees, some of which grow in gregarious groupings to the exclusion of other species. Commercial coconut (Cocos nucifera) groves were established in the lowlands of many valleys during the second half of the 19th century and they persist today, though the trees are ageing (Decker 1970, 92–4). Other economic trees of this zone include many noni (Morinda citrifolia), occasional large, old breadfruit (Artocarpus altilis), candlenut (Aleurites moluccana), Tahitian chestnut (Inocarpus fagifer), and Pandanus occurs in sometimes impassible groves. The occasional mountain apple (Syzygium malaccense), vi apple (Spondias cytherea), and clusters of upright bananas (Musa troglodytarum) were also encountered in Hatiheu and Anaho (also see Decker 1970, 124– 29,141–2; Mueller-Dombois and Fosberg 1998, 451). Dense Hibiscus tiliaceus thickets are also present in many parts of this zone. Clusters of Polynesian bamboo (Schizostachyum glaucifolium) and a sterile form of Hibiscus tiliaceus, both traditional cultivars, can be found in scattered locations with the latter occurring extensively on parts of Ua Huka (Decker 1991, 30). Trees introduced in the historic period, such as mango and several species of Syzygium (S. cumini and S. jambos), can form dense stands in this zone. Other historic introductions also thrive in fallow forests, including Cananga odorata (ylang- ylang), Annona muricata (soursop), kapok (Ceiba pentandra), several types of citrus, and avocado trees. Two types of guava (Psidium) are widely dispersed in the lowland forests

39 and though the fruit is sometimes eaten, the tree is an invasive. Many other introduced trees are also found in this zone, including patches of Adenanthera pavonina, whose seedlings carpet the under-story in some areas (fieldnotes 26 March 2010). Mesophytic inland vegetation* includes the upper boundaries of prehistoric cultivation. Back-valley forests (pluviotropical zones, see Decker 1991, 28–9) stretch to the upper boundary of land cultivated by prehistoric Marquesans. Relict agricultural terraces are located in many of these zones. Vegetation is dominated by Hibiscus tiliaceus and Inocarpus fagifer, and sometimes Aleurites moluccana, along with bamboo, Spondias, breadfruit, Morinda, bananas, and Casuarina. Native vegetation that persists, though not abundantly, includes Sapindus, Xylosma, Ficus, Pandanus, Psydrax, Erythrina, and others. Tahitian chestnut (Inocarpus fagifer) trees can form a high, dense canopy in this zone, and Decker (1991, 29) noted that Hibiscus-Inocarpus back valley forests are not diverse as the closed canopy formed by Inocarpus permits very little light below. Exceptions are Coffea arabica, which can thrive in low light conditions, and some larger trees including Aleurites, Terminalia, and Spondias. Colossal banyans (Ficus prolixa), which are spectacular ancient monuments (see Ottino-Garanger 2006), can also be found in this zone. Other native trees and shrubs are present, though patches of native forest are rare. Near back-valley walls land is highly eroded and in many places, scrub, grasses, and ferns are interspersed with very dense and sometimes impassable thickets of Hibiscus tiliaceus. Montane mesic rain forests receive 3000 mm annual rainfall or more, and are located between approximately 400 to 900 m in elevation. These forests are humid much of the time and the low canopy is composed of native trees including Hibiscus, Pandanus, and tree ferns. Cloud and rain forest stretches up high peaks to over 1000 m elevation. These moist forests are often shrouded in clouds, and contain a diverse and largely endemic flora including many types of tree ferns, Metrosideros, Weinmannia, Trimenia, and a host of smaller trees and shrubs. In exposed and often windy locations where cloud cover breaks, the forest canopy opens and shrubs, small trees and low ferns dominate.

Figure 3.5: Some vegetation zones of Nuku Hiva, Marquesas Islands. Clockwise from top, left: a, house garden (Hatiheu); b, open cloud forest (Toovii); c, xeric lowlands (Hakaea); d, cultivated mesic forest (Hatiheu); e, Hibiscus tiliaceus thicket (Anaho); f, coastal strand (Anaho).

Culture history The archipelagos of Central East Polynesia were probably colonised not long after the central Pacific Ocean receded from a mid-Holocene highstand (Dickinson 2003), and archaeologists generally agree that many archipelagos were settled by 1500 years ago or less (Allen and Kahn 2010). Though refined chronologies for the archipelagos of Eastern Polynesia have recently the subject of some debate (e.g., Mulrooney et al. 2011; Wilmshurst et al. 2011), the Marquesan prehistoric sequence is more resolved than others. Chronologies first developed in the mid-20th century have been recently refined (Figure 3.6) after a number of intensive subsurface archaeological programmes (Allen 2004b, 2009a; Allen and McAlister 2010, 2013; Molle and Conte 2011; Rolett 1998; Rolett and Conte 1995; Sinoto 1966, 1970; Suggs 1961). Five phases in Marquesan cultural history are now recognised by archaeologists: Settlement, Developmental, Expansion, Classic, and Historic. For the purposes of the present study, phases have been grouped into three categories: (1) Settlement, Developmental and early Expansion (prior to AD 1400), (2) the later Expansion phase (AD 1400–1650), and (3) the Classic and very early Historic period (after AD 1650 to the early 1800s).

Figure 3.6: Revised Marquesan cultural history timeline (from Allen 2004b).

Prior to AD 1400 Radiocarbon data from islands in both the northern and southern Marquesas indicate the archipelago was settled before AD 1100, and populations were distributed throughout the islands by AD 1000–1250 (Allen and McAlister 2010). The presence of quantities of fine-grained basalt from the small northern islet of Eiao as well as other local sources indicate people were familiar with a variety of intra-archipelago resources by this time (Allen 2004b; Allen and McAlister 2013). It is likely that the sociopolitical structure of this early period followed the system of Ancestral Polynesian Society, where descent groups were led by a hereditary chief who had secular and religious authority (Kirch 1991b, 140; 1984, 64). During this period, settlers were adapting their lifeways to the distinctive Marquesan topography and climate. While early settlements were somewhat ephemeral and often situated along the coasts (e.g., Sinoto 1979; Suggs 1961), after AD 1300 low stone pavements become common, suggesting more permanent settlements (Anderson and Sinoto 2002; Rolett 1998). The Marquesas Islands were a particularly challenging location to cultivate herbaceous crops and tend livestock. Environmental conditions are unlike other places in Polynesia: expanses of relatively flat land and coastal plains are very limited, valley slopes are steep and arduous to traverse, and climate conditions can be unpredictable. It is evident that sweet potato was being processed in the Marquesas perhaps as early as the 13th century AD (Allen and Ussher 2013). There has also been a suggestion that breadfruit became more important later in this phase, though the postulate is based solely on the occurrence of scraping tools found in archaeological sites on Nuku Hiva (Suggs 1961, 181– 2). Little else is known about the cultivation or use of plant foods at this time, however, and most evidence for prehistoric subsistence has been derived from the analysis of faunal remains. While there has been some debate regarding changes in fishing habits over time (Davidson et al. 1999; Dye 1996; Leach et al. 1997; Rolett 1998; and others), a sharp decline in native bird populations was noted by AD 1300, and some avian species (particularly landbirds) became extirpated or extinct (Steadman 2006, 239–248; Steadman and Rolett 1996). Sedimentary records suggest that forests were reduced and slope erosion was occurring by the end of this period, in the 14th century AD (Allen 2010, 96).

AD 1400–1600 By the mid-15th century AD, Marquesan populations were well established and probably increasing, and this has been identified as a time of significant culture change (Rolett 1998, 247–8). There was a notable increase in pig consumption in the 14th century AD, indicating animal husbandry had expanded (Rolett 1998, 86–9), and it has been suggested that agricultural surpluses were managed by this period to support such activities (Rolett 1998, 254). Ecological stresses appear to have become pronounced at this time. Climate fluctuations became more extreme in this region in the mid-16th century AD, with marked increases in temperature and rainfall, which may have made for generally wetter and stormier conditions in the Marquesas (Allen 2009a). These conditions may have made intra-archipelago travel less frequent, as a sharp drop-off in the use of Eiao basalts was noted at Hanamiai (Rolett 1998, 247), and it also may have reduced off-shore fishing activities (Allen 2010; Dye 1990). Environmental stress has also been linked to the emergence of the distinctive Marquesan socio-political system later in this period, in which the role of hereditary chiefs were diminished and a shifting and competitive system formed (Kirch 1991b, 141; Thomas 1990, 175). In AD 1595, the southern Marquesas were encountered by a Spanish ship carrying settlers, livestock, and crops seeking to found a colony in the Solomon Islands (Quirós 1904, 15–30). People and livestock were disembarked on Tahuata for a week-long respite, and though Quirós only noted that maize was sown, it is possible other plants (and animals) were introduced.

After AD 1650 In the mid-17th century, several powerful El Niño cycles affected the Central Pacific region (Cobb et al. 2003), possibly alternating with very dry periods (Linsley et al. 2008). Allen (2009a, 2010; Aswani and Allen 2009) has argued that exacerbated weather conditions increased sedimentation and shoreline aggradation, and influenced a number of important societal changes including a shift away from the coast, the construction of large stone house foundations, the expansion of religious (me’ae) and community centres (tohua), and further destabilisation of the Marquesan socio-political system. It has been suggested that these highly variable conditions, combined with crop failures and famine, destabilised existing chiefly power structures in a ‘competitive involution’ wherein elaborate, competitive feasting rituals and outright warfare intensified (Kirch 1991b, 135). Crop production may have become intensive and extensive at this time, as there is some evidence

44 irrigated terraces (probably for production of wetland taro) were constructed in this period in well-watered locations (Addison 2006, 733), and fruit-tree groves may have been frequent in some areas (Millerstrom and Coil 2008). It is also presumed that the practice of stockpiling large stores of fermented breadfruit (mā), and the placement of storage pits near residences and behind fortifications, began at this time (Suggs 1961; Thomas 1990). Sustained foreign contact commenced in 1774 when Cook’s second voyage landed on Tahuata in the southern group (Cook 1842, 456–60; G. Forster 2000; J. Forster 1996), and the northern Marquesas were encountered seventeen years later, in 1791, by voyagers who noted them in passing but did not land (Ingraham 1810; Marchand 1810).

Ethnohistory Traditional Marquesan culture was described by foreign visitors at first contact and in the early Historic period, spanning the years AD 1595–1810. Sustained contact began with the visit of Captain James Cook in 1774, and continued sporadically for the next few decades with the occasional ship stopping to replenish supplies (Dening 1980:296–301). By 1797, several foreigners were in residence on the islands and two memoirs (Crook 2007; Robarts 1974) have provided information on many aspects of traditional Marquesan culture and the environment, as well as details about encounters with visiting ships. In 1804, a five- day visit by a Russian vessel outfitted with a naturalist (Langsdorff 1968) provided multiple accounts of the islands from various points of view, and a reflexive description of the aforementioned resident Robarts (Govor 2010; Krusenstern 1813; Lisiansky 1814). Sandalwood traders took an interest in the Marquesas in 1810 and commercial visits became more frequent (Dening 1980, 115). Foreign military interests (e.g., Porter 1822) began to meddle in tribal politics around this time, marking the end of the era. Marquesan lifeways were radically changed from the mid-19th century onward. Introduced diseases took their toll and population declined dramatically well into the early 20th century in one of the most severe cases of depopulation in the Pacific (Rallu 1992). Salvage ethnography aimed at documenting what remained of the traditional culture, including the recording of Marquesan mythology, was not carried out until the late 19th and early 20th centuries (e.g., E. Handy 1923, 1930; von den Steinen 1925). The following observations are, as Ferdon (1993, 125) aptly summarised, a narrow view of the end product of centuries of human occupation.

Domestic life The missionary Thomson (1980) said Marquesan daily life had a proscribed routine. People woke at sunrise and went inland for breadfruit, firewood, and occasionally to retrieve breadfruit paste (mā) from their storage pits. Mā was wrapped in leaves and cooked in an earth oven, then pounded with fresh breadfruit (or taro or banana, and sometimes dressed with coconut milk) to make popoi, a staple dish that was still prepared daily in the early 20th century (E. Handy 1923, 189–195; W. Handy 1965, 25–6). Fish were often eaten raw, and persons of rank occasionally had pork or other prized foods cooked in an earth oven. Other activities for men included drinking kava and sometimes waging war, and tradespeople worked at specialised crafts such as fishing and woodcarving. Stone house foundations (paepae) and other distinctive architecture, including large, paved public assembly areas (tohua) and religious and mortuary structures (me’ae), were used at the time of sustained European contact (Allen 2009a; Linton 1925). Well-to-do Marquesan households consisted of a compound of dwellings, including a sleeping house (fa’e hiamoe), a family cooking shed, and sometimes a small sacred enclosure or platform (E. Handy 1923, 61–7). A separate structure for the exclusive use of men included its own storehouse and cooking facilities. Each family of means had a fenced enclosure that included kava, sugar cane, paper mulberry trees, and root crops that may be vulnerable to rooting pigs, and coconut and breadfruit trees were also planted near houses (Crook 2007, 79–80; J. Forster 1982, 489–91; Porter 1822, 57; Robarts 1974, 127). A fermented breadfruit paste pit (ua mā) (and, perhaps other pits for food storage) was usually located near the dwelling compound (E. Handy 1923, 67, 188; Krusenstern 1813, 1:161). The traditional earthen pit oven, known as an umu, had a central role in everyday life and was especially important in feast preparations (E. Handy 1923). Umu were sometimes permanent stone-lined features located in or near the cooking shed or simple pits dug as they were needed, filled with firewood overlaid with stones which, once hot, were spread around, food placed on top, and closed for cooking (E. Handy 1923; Robarts 1974, 279). Earth oven technology involves the use of heated stones and water to steam-cook food in closed pits, a procedure described in detail by Huebert, Allen, and Wallace (2010). Some foods, such as whole breadfruit, were more commonly roasted directly over an open fire (e.g., Robarts 1974, 251). On occasion, a very large earth oven would be built to roast tī root (Cordyline fruticosa), a sweet delicacy (Crook 2007, 75). Families without property lived as tenants, providing various services to the landowners (Crook 2007, 52; Robarts 1974, 253), while more affluent families sometimes

46 had additional residences on their property at distant locations (Langsdorff 1813, 119). During the main breadfruit harvest, many people relocated to temporary living structures set up near the groves (Robarts 1974, 272). Crook (2007, 147) also noted a few houses and some plantations far up in the hills east of Taiohae, and archaeological evidence of domestic structures near irrigation terracing has also been noted (Addison 2006, 286), suggesting that people may have also resided adjacent to agricultural areas.

Food, food production, and management of landscape While the Marquesas Islands had numerous native birds and fish, which were readily exploited by the early settlers, few native plants provided edible (or tasty) foods. Polynesian settlers introduced domestic animals including pigs, dogs, and chickens to the islands (Rolett 1998, 92–4). Major sources of animal protein included fish, shellfish, sea mammals, native birds, and the introduced pig, as well as dog, chicken, turtle, and rat, the latter two of which are reserved for certain classes (Crook 2007, 71–2; Robarts 1974, 248; Rolett 1998, 98). Carbohydrates and fats derived mostly from imported cultivars, especially breadfruit and coconut which were important components of the traditional diet, along with numerous other fruits, nuts, and vegetables. Robarts (1974, 255) and Krusenstern (1813, 164) noted that they also planted taro (several types were described) and several types of bananas. Trees including breadfruit, banana, coconut, and paper mulberry were planted upon the arrival of children, to provide food and bark for cloth (E. Handy 1923, 79, 181). By the time of sustained Western contact, the people of the Marquesan archipelago had a well-developed arboreal agro-economic system based on tree cropping that provided the appearance of an entirely cultivated landscape (e.g., Crook 2007, 73–4; G. Forster 1777, 27; Krusenstern 1813, 124–5). Visitors in the early 19th century, a period of more regular Western contact, describe the valleys of Nuku Hiva as being carpeted in very tall, mature breadfruit, coconut and banana plants (e.g., G. Forster 2000, 336; Krusenstern 1813, 124–5; Porter 1822, 54). Very little information on cultivation techniques was located in the literature. Housegardens containing fruit trees, and possibly taro and paper mulberry (Broussonetia papyrifera) trees, are mentioned by several early foreign visitors (e.g., Krusenstern 1813, 125; Lisiansky 1814, 73). Stone-face irrigation terraces are present in many well-watered areas, indicating pondfields were once an important food production practice (Addison 2006). Fire was used to clear portions of valleys and lowland slopes for planting (Ferdon 1993, 86; Robarts 1974, 245), and ridges separating valleys were also

47 sometimes cleared this way (Crook 2007, 140) for unknown reasons. A type of swidden technique was observed by Crook (2007, 146), who mentions that ‘Faou trees’ (Hibiscus tiliaceus) and others had been scorched at the roots to prevent further growth, and the spaces between dying trees were planted with paper mulberry, kava, bananas, sugarcane, and some breadfruit that he noted grew very well. Early visitors to the archipelagos of central Polynesia appear to have been enamoured of breadfruit and Porter (1822, 53–54), on visiting Nuku Hiva, noted that by the early 19th century the trees had been so exhaustively described that he had little to add on the subject. Other visitors commented that Marquesan breadfruit was especially large and delicious tasting (G. Forster 2000, 2:338) and in a good season branches might break under the weight of the fruit (Robarts 1974, 271). The numerous varieties found here (Jardin 1857, 318 lists 33 names) are all Artocarpus altilis, a seedless type that is vegetatively propagated by wounding the roots so they produce a new shoot, which can then be removed and transplanted. Genetic research has shown that traditional breadfruit varieties found in East Polynesia are almost completely homogeneous (Ragone 1991, 120; Zerega, Ragone, and Motley 2004), and it follows that these are the result of intensive selection. Different varieties of breadfruit are also appreciated by the varying taste and quality of mā they produce (E. Handy 1923, 189). The importance of breadfruit in this location cannot be over-stated: the Marquesan calendar was based on the main breadfruit crop, counting systems that used units of up to 40–80,000 were devised to keep track of the size of the harvest, and the breadfruit harvests were the main source of wealth and power of the chiefs (Crook 2007, 73; Dening in Robarts 1974, 85; Thomson 1980, 39). Processing the year-end breadfruit harvest required the participation of the entire community, and even small children had tasks to perform (W. Handy 1965, 122–3). Chiefs counted the volume of the harvest, and if this and coconuts were plentiful, he was held in high esteem, and stores were preserved for the tribe in case of hard times (Robarts 1974, 300). Breadfruit provided food for people as well as their livestock, notably pigs (Crook 2007, 72–3). In the Marquesas, breadfruit cropped one major and two minor times per year, seasons that were extended by cultivating trees far up hillslopes where they would fruit somewhat later (Crook 2007, 73–4, 140; Robarts 1974, 271–2). While breadfruit trees are very productive, they are also known to drop fruit when stressed by overly dry or windy conditions (Elevitch 2006), a situation that destroyed many fruits in this location in the 19th century (Robarts 1974, 271). Droughts lasting for several years were known to have severely affected breadfruit yields during this period, leading to

48 famine for those without means or forethought to plan ahead, and fostering competition for scarce resources (Ferdon 1993, 98–100; Robarts 1974, 273–5). Though accounts about the severity of these food shortages may be overwrought (Allen 2010, 99), when drought killed breadfruit trees in Anaho in the late 19th century, people went to great lengths to procure fruit from neighbours in Hatiheu (Stevenson 1987, 49). To ensure steady supply of this dietary staple, breadfruit was fermented in tī leaf- lined storage pits. With regular maintenance, fermented breadfruit (mā) could keep for many decades (Robarts 1974, 273). This technology was not unique to the Marquesas, as it is practised throughout Oceania, but it is one of the only means of long-term food preservation known in the region. While mā for daily use was stored near the house, as previously mentioned, massive tribal storage pits 25 to 30 ft deep and 15 to 20 ft across were located high in the mountains where they were safe from attack (E. Handy 1923, 188– 9). They can still be seen in many places throughout the archipelago and have been of particular interest to prehistorians (e.g., Addison 2006, 94; Kellum 1968, 163–4; Kirch 1991b, 128–9; Millerstrom 2001). Other useful cultivated fruit and nut trees are mentioned in early historic records (e.g., Crook 2007, 73–8), where accounts of abundant coconut and breadfruit are nearly ubiquitous alongside descriptions of extensive banana plantations. Bananas were an important crop and plantations were noted by some of the earliest visitors (G. Forster 1777, 357). Several types of nuts were also described, including a type of large chestnut— Inocarpus fagifer, referred to as ‘ratta’ by Crook (2007, 75)—several of which provided a satisfying meal as well as nourishment in times of hardship (Robarts 1974, 116. 256). A small ‘walnut’ is also mentioned, which is probably Aleurites moluccana, though as a purgative in quantity it is not a principal food (Crook 2007, 76; Marchand 1810, 128; Quirós 1904, 29) and the tropical almond—Terminalia catappa, though T. glabrata is a native Marquesan species that also bears edible nuts—is given brief mention (Crook 2007, 72). The Otaheite apple (Spondias cytherea), a well-known fruit in other parts of East Polynesia, was not noted in the Marquesan archipelago in the late 18th century (Crook 2007, 64–66; G. Forster 2000, 2:27) and it may be an historic introduction to the region, though it was seen by visitors only a few years later (Langsdorff 1813). The mountain apple (Syzygium malaccense), known as kehika, was said to be common here (Crook 2007, 74). In addition to the aforementioned species, food-producing trees and shrubs included noni (Morinda citrifolia), ko (Solanum repandum, Pacific tomato), tī (Cordyline fruticosa), and possibly cashew (Anacardium occidentale) which is mentioned only for Nuku Hiva (Crook 2007, 93)

49 though it is not usually considered a Polynesian introduction to the archipelago. Marchand (1810, 128) also mentions collecting several types of potherbs (water-cress and purslain), and lists a the native name ‘naupata’ for one of these, which may refer to the small native shrub Scaevola taccada. It is also mentioned that in time of scarcity, mā was made with cores, skins and immature breadfruit, all of which are typically discarded, and fau (Hibiscus tiliaceus) leaves and Pandanus nuts were eaten (E. Handy 1923, 201).

Sociopolitical systems At the time of sustained European contact in the late 18th century, when the first detailed observations were made, tribal groups were often residents of a single valley, though not exclusively. The Anaho Valley of Nuku Hiva, for example, was allied with the large neighbouring valley of Hatiheu (Allen 2009a citing Rollin 1929, 69). Traditional Marquesan society was stratified into two classes that were divided based on land ownership. The elite group (akati’a) included an array of chiefs, priests, warriors, and various specialists while the commoners (kikino) were tenants (Crook 2007, 53–5; Robarts 1974, 253). While land, people, and chiefs were considered to have common origins, the Marquesan sociopolitical system at this time was multi-faceted. Authority was not restricted to chiefs, and those of achieved status had the potential to wield a considerable amount of power. Priests, for example, were thought to be influential in mediating droughts (E. Handy 1923, 240). Conflict took many forms: sometimes ritualised battles were waged, while at other times food-producing trees and gardens were destroyed or the losing side was slain. Disagreements over land ownership were one of the main causes of both family quarrels and chiefly battles (Ferdon 1993, 115–22). Land ownership was equated with wealth and power (J. Forster 1996, 233–4; Thomas 1990, 90). People were typically residents on land that was formally owned by the chief, and though they did own their residences and the plants they sowed, first fruits of the harvest were given to the chief who used, redistributed, and stored it for later consumption (E. Handy 1923, 58–9). Control over the products of one’s land was an important aspect of status. When crops failed, fermented breadfruit stores became critically important and large to massive stockpiles were the property of the chief, and sometimes other elites, as silos were also located near residences or religious shrines (Kellum 1968, 82), who controlled them as an important display of political power (Linton 1925; Buck 1938, 207 for Mangareva). Tribal alliances played an important role in who benefited from the stockpiles

50 in times of stress, as food was not shared equally during hardships (Allen 2009a; Robarts 1974). Large stores of mā that could feed entire groups for extended periods were located inland, sometimes in obscure locations (E. Handy 1923, 188–9; Kellum 1968, 163–4; Millerstrom 2001, 247).

Feasting Large feasts (ko’ika) were important events in Marquesan society. They were held at harvest time, to commemorate marriages, births, and deaths, honour special persons, or commemorate completion of a significant task such as construction of a structure. Multiple events could be celebrated at the same time, and feasting could last several days. Competitive feasts were also held to display chiefly power and influence (E. Handy 1923, 213–23). Large quantities food, including entire stocks of pigs, were sometimes slaughtered (Krusenstern 1813, 173; Robarts 1974, 60), and various elaborate breadfruit preparations were reserved for feasts (E. Handy 1923, 190–1; Robarts 1974, 278–9). To prepare or recover from these events, chiefs placed temporary restrictions (kahui or ahui) on resources including fresh breadfruit, taro, coconuts, pigs, and even marine resources (Crook 2007, 93; E. Handy 1923, 59–61).

A brief arboreal ethnobotany The Marquesans had, and many still have, a sophisticated understanding of plant life. There exists a name for every plant part, and a name for each major plant group, including an adjective for each particular member of a group (W. Handy 1965, 168–9). Arboreal products had a number of important uses: fruits and leaves were eaten and used in food preparation, wood provided timber for craft and fuel, and many plant parts were used in medicinal preparations, as ornaments and decorations, and to make cloth and utilitarian objects. To illustrate, a brief introduction to the useful arboreal plants of the Marquesans is presented below. Many of the useful arboreal plants of the islands have been described and illustrated by Butaud (2010), and they are listed along with known traditional uses below in Table 3.1. A limited number of arboreal plants were transported to the Marquesas by the ancient Polynesians, reflecting an overall eastward attenuation of introduced woody species across the region (Whistler 2009) and in total, approximately twenty-six Polynesian-introduced trees, shrubs, and tall woody monocots are presently recognised for this archipelago. While

51 they have many different craft, ornament, and medicinal uses, fourteen of these plants also produce food products.

Table 3.1: Useful trees and shrubs of the Marquesas Islands.

Marquesan and Scientific name Status1 Habit Notes2

English Food Timber Timber Medicinal Ornament names Utilitarian

Aleurites moluccana ʻAma, P tree - + + Nuts used for candlenut illumination; soot from nutshells used as tattoo ink and dye Alyxia stellata Meie E shrub + Bark and leaves popular fragrance plant Artocarpus altilis Mei, breadfruit P tree + + + + Fruits important starchy staple; bark used for rough tapa; sap used as adhesive; leaves used in cooking Barringtonia asiatica Hutu, fish- I tree + Seed extract paralyses poison tree fish Broussonetia papyrifera Ute, paper P tree + Bark is main source of mulberry fine white tapa cloth Calophyllum inophyllum Temanu I? tree + - - Prized timber; seeds for scented oil; medicine Casuarina equisetifolia Toa P tree + - - Prized timber; also dye, medicine Celtis pacifica Vaimanini E* tree - - - Occasional fuel, bark produces sweet sap; minor medicinal uses3 Cocos nucifera Ehi, coconut I, P woody + + + ++ Indigenous and monocot introduced varieties with many uses Cordia subcordata Tou I tree - + - Important timber tree; seeds a minor food Cordyline fruticosa Tī, autī P shrub + + + Roots eaten; leaves had numerous uses Erythrina variegata Kenae, ketae, P tree + - - - Timber used for coral tree lightweight canoes; numerous other uses3 Fagraea berteroana Pua enana E tree + + National flower of the Marquesas; good timber Ficus prolixa Āoa, banyan I tree + Tapa made from young bark; very large trees sacred Gardenia taitensis Tiaʻe P tree + Very fragrant flowers had number of uses Hibiscus rosa-sinensis Koute P shrub + Large, decorative flowers

Marquesan and Scientific name Status1 Habit Notes2

English Food Timber Timber Medicinal Ornament names Utilitarian

Hibiscus tiliaceus Hau I tree + + - + Numerous uses; native status debated in eastern Polynesia; an additional sterile variety (hau he’e) with slender trunks used in construction and craft Inocarpus fagifer Ihi, Tahitian P tree +- - - Large edible nuts used in chestnut traditional Marquesan preparations Morinda citrifolia Noni P tree - + + Fruits were food and medicinal; leaves some medicinal uses Pandanus tectorius Haʻa I, P woody + - - + Indigenous and monocot introduced varieties; many uses Pelagodoxa henryana Enu E woody Very rare and protected monocot today, uses not known Piper methysticum Kava P shrub + Roots used to make mildly narcotic beverage Pipturus argentus Hoka, hona I tree + Very strong fibres made from bark Santalum insulare Puahi, E tree - + Fragrant wood used to sandalwood scent coconut oil Solanum spp. Oupoo (S. P shrubs + + + Fruits were eaten and americanum), some were decorative; hukou (S. ferox), porohiti was a condiment porohiti, for human flesh cannibal cherry (S. viride) Spondias cytherea Vi, Otaheite P tree +- - Fruits eaten; may be a apple 19th century introduction to Marquesas Syzygium malaccense Kehika, P tree +- - Sweet, juicy fruit; bark rose apple extract a purgative Tephrosia purpurea Kohuhu P shrub - + Extract used to poison fish Terminalia catappa Maʻiʻi, tropical P? tree ++ + Timber esteemed for ease almond of workability; nuts a snack food; bark medicinal Terminalia glabrata Maʻiʻi enana E* tree ++ + (as above) Thespesia populnea Miʻo I tree + - + Prized timber; trees sacred Urena lobata Puehu, P shrub + Fibres used for cordage in Hibiscus burr Tahiti

Marquesan and Scientific name Status1 Habit Notes2

English Food Timber Timber Medicinal Ornament names Utilitarian

Waltheria spp. Kaepu P, E* shrub - + W. tomentosa hard wood or sub- used to make lances in shrub Marquesas; stems, leaves, and root bark traditional medicine in Hawaii4 Wikstroemia coriacea Akatea E* shrub + + Some species have poisonous berries, others have medicinal and utilitarian uses 1 I, Indigenous; P, Polynesian introduction; E, Endemic. Extents of endemism are Marquesas Islands unless noted (*), whereby range includes other locations in East Polynesia. 2 Compiled from Brown (1935) and Whistler (2009). 3 From Butaud et al. (2008). 4 From Abbott (1992).

In addition to economic and everyday uses, many of these trees have symbolic and religious importance in the Marquesan belief system. Oral traditions recount the birth of several trees resulting from the coupling of Atea and One-uʻi, including Casuarina, Calophyllum, and Ficus prolixa, associated with masculine activities, and Pandanus, Gardenia taitensis, and others associated with females (E. Handy 1930, 322). Many trees are also part of a genealogy of nature (E. Handy 1930, 345–6) and are considered children of Atea (Vatea) including coconut, breadfruit, Inocarpus, Hibiscus tiliaceus, bamboo, candlenut, Fagraea, Gardenia, Pandanus, Calophyllum, Erythrina, Barringtonia, Cerbera manghas, and Casuarina, many of which the Polynesians introduced to the archipelago. Marquesan legends regarding the origin of fire also mention various trees by name. In a tussle, the god Maui sent an opponent who controlled fire crashing to the ground, dispersing fire to stones and certain trees including Hibiscus tiliaceus, Ficus prolixa, coconut, breadfruit, Premna, Casuarina, Aleurites, and Syzygium malaccense (von den Steinen 1988, 110–23). Some of these trees have known uses in traditional culture, including H. tiliaceus which was used to make the traditional fire plough (E. Handy 1923), and Aleurites nuts were commonly strung on coconut midribs and set alight as a torch (Brown 1935). But other aforementioned trees may have significance related to heat or fire that has since been forgotten, as few references to these uses could be found in the Marquesan literature or from informants.

Certain trees or tree parts also had importance in ritual activities. Some sacred trees or groves were located near me’ae, including Ficus prolixa, which can grow to an immense size, and large Calophyllum, and sometimes Casuarina trees (E. Handy 1923, 119–20). There is also mention in the historic literature of burials in trees: hollowed out logs or old canoes were reused for this purpose, and strangers or persons not claimed by family were buried in hollow trees (Robarts 1974, 56; Thomson 1980, 37–8), while breadfruit and Erythrina woods were sometimes used for coffins (E. Handy 1923, 111). Sticks made of Thespesia populnea were struck together to initiate dance ceremonies and to remove certain tapus (Brown 1935, 178). Branches of Calophyllum were waved as signs of peace (J. Forster 1982, 490), and the wood of Cordia subcordata was prized for use in creating carved drums used for ceremonies and war (E. Handy 1923, 310). It is evident that plants, including many trees and shrubs, were an important part of the Marquesan economic, social, and religious life. This brief ethnobotany provides cultural context for the following study of plants in Marquesan prehistory, as well as a list of the trees and shrubs that were expected in archaeological charcoal assemblages.

Summary Information has been presented on the climate, vegetation, culture history, and ethnobotany of the Marquesas Islands. Significant recent research was summarised along with historic and ethnographic information on the role of woody plants in traditional culture and everyday life. Climate cycles in the Central Pacific were shown to be closely related to ENSO cycles, and these have (and probably had, in prehistoric times) a significant influence on the highly variable moisture regimes of the archipelago. While sedimentary sequences and archaeological sedimentary analyses have provided important data on geomorphological processes that have resulted from human- and climate-induced changes, they have also informed on human-induced vegetation burning practises. Palynological sequences from the archipelago have thus far been only modestly informative on past vegetation change. The major vegetation zones of the archipelago were described, with emphasis on the zones most influenced by prehistoric human activity. Forest cover in the Marquesas is often correlated with elevation, and variations in the composition of vegetation communities are primarily driven by moisture, elevation, and land use patterns in the prehistoric period, but perhaps the most significant impact has been that of feral grazing animals over the past century. The cultural history of the Marquesas was summarised, along with a review of relevant aspects

55 of traditional culture that relate to domestic life, food production, and socio-political systems. Finally, a brief arboreal ethnobotany was presented to provide cultural context to the findings of this study, and to explicitly identify the taxa that are expected in archaeological charcoal assemblages. These details are largely based on information gathered before traditional lifeways were irrevocably changed in the mid-19th century by controlling foreign interests, missionaries, and epidemic diseases.

Chapter 4: Archaeological excavations on Nuku Hiva

Introduction This thesis focuses on the analysis of archaeobotanical material collected from excavations spanning the past two decades of field work conducted by two researchers, Melinda Allen and Sidsel Millerstrom. Allen’s excavations in the valleys of Nuku Hiva are part of a long-term investigation of human ecodynamics in the Marquesas Islands, and the present study draws heavily on materials recovered in this research programme (Allen 2004b, 2009a, 2010; Allen and McAlister 2010, 2013). These assemblages were augmented with material collected by Millerstrom during an extensive analysis of indigenous rock art (Millerstrom 2001). Numerous locations have been investigated in the course of these projects, and sites in three valleys were selected to be part of the present analysis: Hakaea, Hatiheu, and Anaho (Figure 4.1). All are located on the windward side of Nuku Hiva, but they have some variations in size, shape, hydrology, and local climate conditions. In this chapter, I provide details regarding the excavations, radiocarbon chronologies, and subsurface features encountered in these excavations. To contextualise these materials, details of the site setting and the major findings of the research programmes are also summarised.

Figure 4.1: Map of Nuku Hiva with study locations indicted in bold.

Hakaea Beach

Overview / Introduction Situated at the mouth of Hakaea Valley, the beach site is important as one of the most securely dated pre-14th century AD occupation in the Marquesas (Allen and McAlister 2010). In addition to cultural activities, this locale is notable for a range of natural processes, including alluvial activity, tsunami, and storms. The cultural sequence commenced concurrently with the formation of the beach ridge, and in the excavated area it spans approximately three hundred years from the 12th to the 15th centuries AD. Cultural activities of varying intensity are represented, including domestic activities such as food preparation and disposal, and more significant impacts on the landscape are evident by the late 14th century AD. The excavated artefact assemblage from this site was small, composed mainly of fishing gear and tools for pearl-shell manufacture. Faunal materials recovered include fish, shellfish, bird, pig, and dog.

Local setting, geomorphology, vegetation The Hakaea Beach site is located in a narrow, elongated valley that opens onto a 300 m wide coralline sand beach and a narrow and deeply incised bay. Portions of the coastal plain have been reworked over time by high-energy wave action, and there is evidence that occasional high-energy alluvial events have also occurred in the area (Allen and McAlister 2010, Fig. 7). The beach ridge, which has been formed largely by wave action, stretches most of the width of the valley mouth. One or more springs behind the ridge produce fresh water, and a dry stream bed cuts through the western coastal flat. The present-day lowland vegetation behind the beach and the strand zone contain elements of dry forest, transitional forest, and cultivated forest. Mature Calophyllum inophyllum trees are present along the beach ridge, probably planted by residents, and an ageing group of Thespesia populnea grow along a dry, rocky streambed on the western slope abutting the beach. The central flat contains a coconut grove interspersed with useful economic trees (e.g., tamarind, mango, Thespesia, and others). A few hundred metres inland, there are dense thickets of Hibiscus tiliaceus and groups of Sapindus saponaria, Xylosma suaveolens, and small shrubs in open areas. Farther inland, there are mixed elements of native forest that also include Metrosideros collina. Steep slopes to the east and west of the beach are badly eroded with occasional patches of low scrub and trees, but farther inland the valley is more thickly vegetated (A. McAlister, pers. comm. 2014). The beach is used occasionally today; there are

58 few permanent structures and only one resident. A view of the setting is provided in Figure 4.2.

Figure 4.2: Hakaea Beach as viewed from the eastern end (from Allen and McAlister 2010).

Excavations Allen visited Hakaea in 2006 to sample cultural deposits and returned with a field crew the following year to undertake stratigraphically controlled excavations (as described in Allen and McAlister 2010). After probing along several transects, a series of shovel pits were excavated and controlled excavations then targeted significant cultural deposits on the western beach ridge. Seven 1m2 units were opened in an area of about 30m2 (Figure 4.3). Excavations followed natural stratigraphy and features were excavated separately. Sediments were dry screened through 3.2 mm (1/8 inch) mesh in the field with the exception of Layer VII. This layer contained numerous bird bones, and it was wet screened using water from the nearby spring. Charcoal was manually separated from other materials in the University of Auckland archaeology laboratory, a process the author participated in.

Figure 4.3: Map of excavation at Hakaea Beach site (modified from Allen and McAlister 2010).

Stratigraphy, radiocarbon ages, depositional sequence The stratigraphy of this site included nine layers, three of which (III, V, VII) were identified as cultural layers (Figure 4.4). There was ephemeral evidence of site use in other largely sterile layers, and the beach ridge continued to build up during the sequence; not all layers were encountered in all excavation units (see Allen and McAlister 2010). Layer III varied from dark brown loam with fine gravel and charcoal to dark greyish brown sand that ranged from 10 to 30 cm thick. It contained artefacts, shell, and bone, and several widely spaced features, including a post-mould, hearth, oven, and pit, which suggested several areas of domestic activity. A pavement was also suggested in SP-2 by the presence of water- worn pebbles, worked pearl-shell, shellfish, and charcoal. In TP-3 and TP-4, on the summit

60 of the ridge, a hearth in Layer IV (Fe. 4), suggested occasional short-term use of the area between the more sustained occupations of Layer V and III. Layer V varied from dark brown sandy clay loam to light brownish grey fine loamy sand on the beach ridge, 10–15 cm in thickness. This layer contained artefacts, shell, and bone (food remains), increasing in density with depth, a fire feature (Fe. 5), an oven (Fe. 6), two post-moulds, and an accumulation of fire-cracked rock. A culturally sterile layer (VI) of fine sand underlay this layer in several units. Layer VII was a dark brown to black sandy clay loam, 12–15 cm thick in all units. There was abundant charcoal and ash in one unit (TP-6), and quantities of angular gravel. This layer contained shell, bird, and other bone as well as artefacts, though these materials were highly fragmented. A hearth (Fe. 10) and post-mould suggest cultural activities took place in the area during this period.

Figure 4.4: Stratigraphy of SP8. Bucket resting on Layer VIII, the basal cultural layer. Note excavated post-mould along eastern wall of unit (from Allen and McAlister 2008).

Most charcoal samples for radiocarbon dating were collected from in situ deposits during excavation, using a clean trowel. One sample (OZK039) was taken from a washout

61 gulch to the east of the excavation area, and two samples (OZM070, OZM071) were taken from a cleaned profile of a large rubbish pit between TP-7 and SP-2, which had been exposed by the landowner using a mechanical excavator. Of these, ten samples were dated (Figure 4.5). Details of nine samples that represent the main cultural occupations are presented in Table 4.1. In the main, dates were run on short-lived coconut (Cocos nucifera) endocarp. The sample of monocot stem (OZM072) does not appear to have been affected by inbuilt age as it falls within the range established by other samples from Layer V. Barringtonia asiatica (WK-19934) is an indigenous coastal tree that had potential for significant inbuilt age, as some species in the genus can live for 80 to 90 years (Allen and Wallace 2007, Table 2; Elevitch 2006), but the radiocarbon age was also in general agreement with other samples from Layer VII.

Figure 4.5: Radiocarbon sequence for Hakaea Beach site.

Table 4.1: Radiocarbon determinations for Hakaea Beach site. Conventional calAD 2σ Lab No. Unit Layer Provenience Material1 14C age BP2 range3 OZM070 none III Large endocarp, cf. 525±35 1317-1444 rubbish pit, Cocos 77-94 cmbs nucifera OZM073 TP-1 V From screen, Cocos 615±35 1291-1405 86/92 to nucifera 97/104 cmbs endocarp OZM071 none V Large trash endocarp, cf. 600±40 1294-1411 pit, 107-117 Cocos cmbs nucifera OZM072 TP-3 V Small oven Monocot 670±40 1268-1396 (Fe. 6) stem OZK039 none n/a Oven in E Fruit/nut 690±35 1261-1391 profile of endocarp wash-out gulch WK-19934 SP-2 VII Basal Barringtonia 744±33 1221-1293 cultural asiatica layer, 133 cmbs WK-22226 TP-1 VII N profile endocarp, cf. 744±30 1223-1290 Cocos nucifera WK-22228 TP-6 VII Hearth (Fe. endocarp, cf. 746±30 1222-1289 10) in N Cocos profile nucifera WK-22227 TP-5 VII Top of endocarp, cf. 824±30 1164-1264 greasy black Cocos sediment nucifera 1 All material listed was charcoal, identified by Rod Wallace. All analyses listed in table are AMS. OZ prefix indicates Australian Nuclear Science and Technology Organisation, WK indicates Waikato Radiocarbon Laboratory. 2 Data reported by Allen and McAlister (2010). 3 Recalibrated with OxCal v4.23 (Bronk Ramsey 2009) using IntCal13 (Reimer et al. 2013) in an effort to provide uniformly calibrated set of dates for the present analysis.

Summary A small number of artefacts were recovered in these excavations, and fish, shell, dog, pig, and bird bone were also encountered, with the latter being especially abundant in Layer VII. In total, fourteen cultural features were excavated, including hearths, ovens, post- moulds, and several of indeterminate function (Allen and McAlister 2010, Table 1). The earliest cultural occupation encountered at the Hakaea Beach site (Layer VII) indicates that domestic activities, evidenced by a hearth, post-mould, and abundant food remains, took place on sand and/or alluvial wash that had accumulated in the area some time

63 before the 12th or 13th century AD. After a period of sand accumulation, occupation of the area continued ca. AD 1275 to 1400 (Layer V, 1σ range) and evidence of some domestic activities continued. After this, the site was abandoned though infrequent short-term activities did take place during ridge formation (Layer IV), and it was again more permanently occupied in the 14th or 15th century (Layer III). The dense concentration of organic material and evidence of structures built along the stream indicate this occupation was more intense than it had been previously. There is some indication the coastal area was less stable after this period. At the upper boundary of Layer III there was an erosive event, which may have been caused by vegetation clearance, however Allen and McAlister (2010) indicate that this is probably not the only contributor because storm deposition and several large flood events were evident in stratigraphy in the eastern wash-out gully. They further suggest that these conditions may derive from (or have been exacerbated by) the warmer, wetter climate of the Little Ice Age which commenced around this time. After the mid-15th century, the area was abandoned, and it is likely that, as with other parts of the northern coast, people moved inland after this period (Allen 2009a). Several stone structures are present on the rise behind the site, and stone architectural features extend inland.

Materials selected for analysis Fifty-two charcoal samples from nine excavation units (SP-8, TP-1 through TP-8) at the Hakaea Beach site were selected to be part of the present analysis. Though most of this material derives from occupation debris, four cultural features also yielded identifiable charcoal. They include two hearths (Fe. 4 and 10), a small earth oven (Fe. 6), and a fire feature of indeterminate function (Fe. 5). Details regarding provenience, size, and contents are provided in Table 4.2.

Table 4.2: Cultural features from the Hakaea Beach site selected for study (details from Allen and McAlister 2010). Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated number(s) Fe. 4 Hearth TP-3 and TP- Profile: <1/4 charcoal, 5667, 4 lenticular, 50 cm shell, bone - 5668.2 Layer IV length, 10 cm no FCR thick, Plan: 46 by 45 cm, 5 cm thick Fe. 5 Fire TP-4 Profile: Shape <1/4 charcoal, 5671, 5672 feature Layer V uncertain, FCR dimensions not available, Plan: 90 by 30 cm, 20 cm thick overall Fe. 6 Oven TP-3 Profile: Basin, <1/4 many 5677, 5679 Layer V 40 cm length, 20 rounded cm thick, Plan: cobbles, 30 by 25cm, 23 FCR and cm thick charcoal Fe. 10 Hearth TP-6 Profile: Basin, <1/4 charcoal 5739 Layer VII ca. 30 cm length, 8 cm thick, Plan: dimensions not available 1 FCR = Fire cracked rock

Hatiheu sites Hatiheu is a large (approx. 600 ha) amphitheatre-shaped valley on the north coast of Nuku Hiva. It is one of the most well-watered locations on the island (Cauchard and Inchauspe 1978). Boundaries are abrupt, rising to high ridges on all sides, and there are four permanent streams running through the valley. Soils range from a sandy, clayey sediment on the coastal plain to rich alluvial and colluvial deposits further inland (ORSTOM 1993). Stone foundations, platforms, and low stone walls are common features in this location. Relict irrigation terraces can be found throughout the valley; these were thought to have been used for the extensive production of taro (Addison 2006). Hatiheu was an important population and cultural centre in the prehistoric period, and eight ceremonial and/or religious complexes (tohuas) are located there (Millerstrom 2001, 149). Two assemblages derive from excavations in this valley: Pahumano-o-te-tai, a subsurface site near the coast, and units excavated near stone structures in the interior.

Pahumano-o-te-tai

Overview / Introduction Pahumano-o-te-tai is situated on the coastal flat in Hatiheu Valley. This site is notable as another securely dated pre-14th century occupation (Allen and McAlister 2013), and it provides further evidence people were widely established in the archipelago by this time (Allen and McAlister 2010). It is also important because it contained a long-term record of landscape and vegetation change, and because lithic artefacts found at the site indicate the first occupants interacted with people on other parts of the island, and possibly also with other islands in the archipelago as well. The cultural sequence commences between the late 13th to late 14th centuries AD, and spans a six hundred year period. Cultural activities are evident throughout the sequence, including domestic activities such as food preparation and disposal, as well as stone tool manufacture. A repeated sequence of vegetation clearance and occupation, followed by slope destabilisation, is evident in this area. These processes were repeated over a span of several centuries. Artefacts and faunal materials were found at this site, but the latter were infrequent and sometimes exceptionally fragile. Some avian and fish bones from the lowest cultural zone may be naturally deposited.

Local setting, geomorphology, vegetation Hatiheu Bay opens onto a 500 m black sand beach flanked by rocky cliffs and behind a modern stone barrier, and the coastal flat extends approximately 1–1 1/2 km inland. The Pahumano-o-te-tai site is located immediately behind the beach and the main road, within the church grounds of the small village. It lies near a 1 m tall retaining wall of rock, which has been built at the bottom of the escarpment that borders the site. The top of the ridge south of the site provides a full view of the Hatiheu Bay (Figure 4.6). The church cemetery lies on the escarpment, bordered by several coconut stumps, a large mango tree, Morinda citrifolia (noni), frangipani, and other shrubs. Over a low adjoining fence to the east, the escarpment contains a forested area that includes mature coconut trees, low scrub, mango, and Hibiscus trees, and to the west is a banana plantation. At the top of the ridge there are residences with housegardens that includes mature breadfruit trees, bananas, and decorative shrubs. On the coastal flat, permanent streams are located to each side of the excavation area. Mid-20th century tsunami have reworked the coastal plain and destroyed some buildings (summarised by Allen and McAlister 2013). The area has also been subject

66 to major erosive events, as several sharp layer boundaries encountered in excavation (between Zones E and F, and Zones G and H) indicate rapid accumulations of sediment occurred, probably caused by vegetation burning evidenced in the underlying strata.

Figure 4.6: View of Hatiheu Bay from the church cemetery. Excavation area lies between the church and the partially obscured residence on the right (from Allen and McAlister 2013).

Excavations Previous salvage archaeology that took place upon reconstruction of the church yielded an early radiocarbon date (665–1012 calAD 2σ, reported in M. Orliac 2003a). Based on this finding, Allen visited the churchyard in 2011 to conduct further investigations. After excavating a series of shovel tests along a transect perpendicular to the coast, three stratigraphically controlled test units were excavated on the coastal flat (Figure 4.7). An additional three units were then excavated against the ridge face in an undisturbed area to the south of a small residence on the church grounds; excavations in these units were carried out in stages where first the ridge faces were exposed and then small units were excavated into the exposed faces so as not to destabilise the escarpment. Excavations followed the natural stratigraphy, and features were excavated separately. The heavy, clayey sediments at this site were screened using 6.35 mm (1/4 inch) mesh. Charcoal was manually separated from other materials at the time of screening. Several bulk sediment samples were also

67 taken from deposits with the densest concentrations of charcoal, these were processed using water flotation later in the University of Auckland laboratory.

Figure 4.7: Map of excavation at Hatiheu churchyard (from Allen and McAlister 2013).

Stratigraphy, radiocarbon ages, depositional sequence On the ridge face, nine stratigraphic layers were exposed and are described by Allen and McAlister (2013). Three of these (III, V, VII) were prehistoric and III and VII contained especially dense concentrations of charcoal (Figure 4.8). All sediments were clayey. Layer III was a very dark brown and approximately 10 cm thick. A burned log surrounded by red oxidised sediment evidenced in situ burning in this layer. This stratum contained a flake and pig bone. Below this layer was colluvium with no cultural material (Layer IV). Layer V was a dark reddish brown, from 10 to 42 cm thick, increasing in thickness from east to west. A deep pit (not numbered, of undetermined function) filled with cobbles at the base and large rocks at the top was encountered in SP-7 (results of the present study indicate this was a cultural feature, see Chapter Seven). A few basalt flakes, mammal, and bird bone were found in this layer. Below this was another layer of acultural colluvium (VI). Layer VII was a very dark brown, approximately 10 cm thick, with a dense concentration of charcoal; it was not present in SP-7. In SP-6, the layer extended to 115 cmbs, the depth of radiocarbon

68 sample Beta-303439. A small quantity of fish bone and flakes were found in this stratum. These strata extend into the escarpment and appear to increase in thickness, and Allen and McAlister (2013) suggest that they are representative of in situ cultural activities. Two culturally sterile subsoils underlay these sediments. On the coastal flat, four stratigraphic layers were noted, one of which (III) was determined to be prehistoric. The upper layers (I, II) contained historic materials. Layer III was very dark brown, and 30 cm thick in TP-1. Several in situ features were encountered in this layer, including several post-moulds in TP-1, which were closely spaced and of a similar size and depth, possibly indicating a structure. This was underlain by a sterile clay (Layer IV), and several cm into this layer in TP-3 two large flat boulders were encountered. Features in this layer also included ovens, additional post-moulds, a possible hearth or rake- out, and several pits of undetermined function and varying size (see Allen and McAlister 2013, Table 2). Most of the lithic artefacts, including a finished adze, recovered at this site derive from this stratum. Charcoal samples for radiocarbon dating were collected from in situ deposits during excavation using a clean trowel. All materials dated were short-lived Cocos nucifera endocarp. Three samples were dated from units on the ridge profile (TP-5 and 6), and one was taken from a controlled unit on the coastal flat (TP-3) (Figure 4.9). Details of these samples are provided in Table 4.3. Based on the resulting ages, sedimentary characteristics, and artefacts, overall site designations were correlated by Allen and McAlister (2013, Figure 4.10 below). The basal cultural layer in both areas produced internally consistent dates, and is referred to as Zone H. This stratum dates to between the late 13th and late 14th century AD. Layer V of the ridge profile, from between the 14th and 15th century AD, was assigned to Zone F, and Layer III of the ridge profile dating to the late prehistoric period to Zone D. Notably, Zones B to G are not clearly represented on the coastal flat, and may have been removed by historic period constructions related to the church or one or more tidal waves.

Figure 4.8: Stratigraphy of SP5 (left) and TP1 (right) with the Layer II-III boundary and yellowish brown clay lens indicated by arrows (from Allen and McAlister 2013, Figs. 7 and 8).

Figure 4.9: Radiocarbon sequence for Pahumano-o-te-tai site.

Table 4.3: Radiocarbon determinations for Pahumano-o-te-tai site. Analytical Provenience Conventional calAD 2σ Lab No. Unit Material1 zone and sample # 14C age BP2 range3 Beta- Ridge Zone D Layer III Cocos 280 ± 30 1498- 305117 profile 45-55 cmbs nucifera 1795 SP-5 in situ endocarp #6775 Beta- Ridge Zone F Layer V Cocos 560 ± 30 1307- 303438 profile 65-75 cmbs nucifera 1429 SP-5 in situ endocarp #6774 Beta- Ridge Zone H Layer VII Cocos 660 ± 30 1277– 303439 profile 110-115 cmbs nucifera 1393 SP-6 in situ endocarp #6795 Beta- Controlled Zone H Layer III Cocos 690 ± 30 1265- 303440 Units 80 cmbs nucifera 1388 TP-3 Fe. 20 (oven, endocarp in situ) #6841 1 All material identified by the present author. 2 Data reported by Allen and McAlister (2013). 3 Recalibrated with OxCal v4.23 (Bronk Ramsey 2009) using IntCal13 (Reimer et al. 2013) in an effort to provide uniformly calibrated set of dates for the present analysis.

Figure 4.10: Schematic drawing of stratigraphy identified in the two tested areas at Pahumano, cross-correlations of strata, and designation of analytic zones (from Allen and McAlister 2013).

Summary A range of activities are represented at this site. Analysis of the lithics indicated that this was a location for the manufacture and rejuvenation of stone tools. Artefacts recovered included flakes, some with polish, and a finished adze bevel, preforms, and a grindstone fragment, most of which were from Zone H. Geochemical analysis indicated that the material from this zone was sourced from various locations within the archipelago, but mainly Nuku Hiva sources, and this finding is particularly important because it indicates people had discovered and were using a variety of lithic sources early in the cultural sequence (Allen and McAlister 2013, Table 5). Cultural features identified at Pahumano-o-te-tai suggest a structure and domestic activities, such as food preparation and disposal, are evidenced by the presence of a number of ovens and hearths. Radiocarbon dates suggest these activities occurred during the earliest occupation of the site, in the late 13th or late 14th century AD, and continued here throughout the prehistoric sequence. Both wild and domesticated fauna were found at this location. Remains include fish and bird bone, turtle, a shark and a dog tooth, weathered shell, and chiton. Bird bones were frequent in some Zone H contexts, and their concentration along with small fish bones at the lower boundary of this zone size suggests they may be naturally deposited. Initial occupation of this area included clearance of the forest, and the findings of this study indicated the surrounding vegetation was burned repeatedly over the next several centuries. These activities destabilised the soils of the escarpment, and colluvium may have buried and remodelled evidence of earlier occupation below. It was noted that the lowest cultural zone at this site was the earliest evidence of activity in the immediate area, but a longer history of anthropogenic burning in this valley is argued to be evident in a pollen core from an interior location (Allen et al. 2011, Table 1). As with the Hakaea Beach site, these processes may have been exacerbated by increasingly wet and probably stormy conditions, hypothesised to be associated with the Little Ice Age after the mid-15th century AD.

Materials selected for analysis Charcoal from the three main cultural zones were targeted for the present analysis. In total, twenty-one samples from six test units were considered relevant. All samples from the ridge profile, with the exception of those from a large pit in SP-7, were from dispersed

72 sedimentary charcoal. On the coastal flat, samples were taken from both the sedimentary matrix and features, including three ovens (Fe. 2, 10, 20), a hearth or rake-out (Fe. 19), and two pits (Fe. 11, 12). Details regarding provenience, size, and contents are provided in Table 4.4.

Table 4.4: Cultural features from the Pahumano-o-te-tai site selected for study (details from Allen and McAlister 2013). Charcoal Portion Number Type Provenience Dimensions Contents sample excavated number(s) Fe. 2 Oven TP-1 Profile: Ca. 1/2 FCR, 6800 Layer III round, 70 by charcoal, 70 by 15 cm, bone, flake, Plan: basin oxidized base

Fe. 10 Oven TP-2 Profile: Ca. 1/2 FCR, 7147, 7242, Layer II rectangular, charcoal, 7228 90 by 60 by unburned 15 cm, Plan: bone, basin oxidized base

Fe. 11 Deep pit TP-2 Profile: Ca. 1/4 FCR, bone, 6830, 6832, Layer III irregular, 60 cobbles 7227 by 25 by 26 cm, Plan: basin

Fe. 12 Pit TP-2 Profile: Ca. 1/3 Red clay, 6810 Layer III round, 40 by some 15 by 8 cm, charcoal, Plan: basin oxidized base

Fe. 19 Hearth or TP-3 Profile: shape Partial FCR, 6828 rake-out Layer III uncertain, 65 charcoal, by 35 by 20 flakes, bone cm, Plan: shape uncertain

Fe. 20 Oven TP-3 Profile: shape Ca. 1/4 Charcoal, 6841 Layer III uncertain, 65 partially by 50 by 15 oxidised cm, Plan: base basin

Non- Pit SP-7 Profile: shape Partial Small 7224, 7244 numbered Layer V(?) uncertain, 48 cobbles at cm wide by base, large 175 cm deep, rocks above, Plan: u- dispersed shaped charcoal

Hatiheu inland

Overview / Introduction The Teiipoka-Kamuihei-Tahakia complex, which has been recently restored (see Ottino-Garanger 2006), lies in the central Hatiheu valley. The earliest cultural occupation within this complex has been dated to the late 15th to 17th century AD (Ottino-Garanger, Valentin, and Guiot 2003). The area is notable for the range of community activities, including feasts, dancing, and other ceremonies that took place there. The remains of a number of elite residences are present in the surrounds. The central valley also has a high concentration of rock art carvings, which have been the subject of an extensive study by Millerstrom (2001). While investigating the spatial and temporal relationships between rock art and archaeological sites, Millerstrom conducted settlement pattern surveys and test excavations throughout western Hatiheu. Artefact assemblages from these excavations were small, and included mainly stone flakes and a broken adze. Analysis of charcoal samples from this work provided evidence that arboriculture was practised in the area (Millerstrom and Coil 2008). The charcoal assemblage from these excavations was loaned to the author for further study. Of particular interest were unidentified materials, which might contribute to the understanding of vegetation processes and agricultural practices in this valley.

Local setting, geomorphology, vegetation The inland sites included in the present analysis are located in a wedge-shaped high- status area between Vaiu’ua and Puhi’oho streams in the central valley, which encompasses the Kamuihei, Tahakia, and Hikokua tohua. Several hundred metres seaward along the Puhi’oho stream lies an extensive agricultural complex of stone-walled taro pondfields, an area that is swampy and underwater at times (Millerstrom 2001, 227). The land unit is bounded by two permanent watercourses and smaller tributaries criss-cross the area. The modern-day vegetation of the valley interior reflects a moist local climate where an average annual rainfall of 1500 mm has been recorded as one of the highest in the archipelago (Cauchard and Inchauspe 1978), and it includes elements of pluviotropical zones, cultivated forest remnants, and mesophytic inland vegetation. Drier zones can also be found on ridges and slopes, and a tropical dry forest featuring Xylosma suaveolens, Cerbera manghas, Maytenus crenata, Sapindus saponaria and others, is present on the northwestern slopes (Gillespie et al. 2011). Though today the landscape is dominated by coconut trees, extensive tracts of relict gardens and groves can be found throughout the valley. Many mature trees

74 were observed in the inland forests, including large Inocarpus fagifer (Tahitian chestnut), mango, candlenut (Aleurites moluccana) trees, and Hibiscus tiliaceus, all of which form monodominant stands in some areas. Other notable, but less frequently occurring, economic trees were also observed in the valley forests, including Syzygium malaccense, Barringtonia asiatica, and avocado. Several massive Ficus prolixa (banyan) grow near the ceremonial centres, with a diverse pluviotropical under-storey. A view of the setting, depicting the area around the Kamuihei complex, is provided in Figure 4.11.

Figure 4.11: Hatiheu Valley inland setting near the Kamuihei complex; the main road depicted at left (photo courtesy of Melinda Allen, 2010).

Excavations Millerstrom (2001) spent five years in Hatiheu Valley surveying, mapping, and excavating test units near archaeological features during an intensive study of the rock art of the central valley; image boulders in the vicinity of the Kamuihei tohua were frequent and this area was targeted for further study. Excavations in the Mutoka area (Figure 4.12) were conducted at two residential terraces in a cluster of late prehistoric and early historic buildings that surround an unpaved courtyard, and on the Ototemui ridge they were situated within a residential complex that abutted steeply sloping ground. Test units were placed in association with rock art images and related surface architecture, and in total four 1m2 units

75 and two stone-lined pits (pakeho) were excavated. Units were excavated in a combination of arbitrary 10 cm levels and cultural layers, using a trowel and brush. Features were excavated separately. All excavated material was visually inspected, but the sedimentary matrix was not screened. Bulk samples of sediment (18 by 20 cm bags) were retained from each arbitrary level and later wet-screened. The profile of a large earth oven, exposed in a road- cut approximately 200 to 250 m inland from the bay near the Vaiu’ua Stream, was also cleaned and sampled. Charcoal was separated from the clay-rich samples by immersing sediment clumps in water that contained a small amount of sodium hexametaphosphate, which was then wet-sieved through 2 mm mesh (Millerstrom and Coil 2008).

Figure 4.12: Map of western central Hatiheu Valley excavation sites (adapted from Millerstrom and Coil 2008).

Stratigraphy, radiocarbon ages, depositional sequence Test units in the Mutoka area were placed between a raised stone sleeping house structure and a large boulder with carved images (TU-1) and, approximately 20 m from this unit, in front of a large boulder with carved images (TU-2). Millerstrom identified several occupation phases in the structure adjacent to TU-1, noting that a smaller foundation was visible underneath the large boulders. Several fire features pre-dated this older structure, and these features were both encountered in Level 2, including a deep hearth (Efe. 1) and an earth oven (Efe. 2), the former of which had burned hot enough to discolour the adjacent image boulder. Upper sediments in TU-2 were disturbed by coconut and Hibiscus tree roots, and it was suggested that sediment fill in the lowest level (Level 3), which contained black cultural sediment, charcoal fragments, palaeosol, gravel, and cobbles, was construction fill that may have been removed from a nearby (5 m) communal mā storage pit.

Figure 4.13: Site map showing locations of TU-4; label ‘331’ denotes petroglyphs (from Millerstrom and Coil 2008).

Test units on the Ototemui ridge in this area were placed on a steep section of the ridge near two paepae in an isolated complex featuring terraces, stone alignments, walls and enclosures and several large image boulders and polishing stones (TU-4, Figure 4.13) and in a deep, stone-lined pit on a megalithic paepae that is part of a sprawling complex (TU-6).

Cultural sediments encountered at TU-4 were darkly stained and contained flecks of charcoal. Excavation of this unit produced a few flakes, one with polish and one with cortex, and a polishing stone located nearby led Millerstrom to conclude this was a location of stone tool rejuvenation and perhaps manufacture. A 1 cm thick burn layer found in the basal level, situated on culturally sterile subsoil, may represent initial vegetation clearance of the area. Excavation of the stone-lined pit (TU-6) atop the paepae produced various types of sediment, including oven rake-out, red ash, angular stones, a broken adze, and several flakes. The function of pits such as these has not been determined, but there have been suggestions they were used for storage, rubbish disposal, or perhaps sacrifice (Millerstrom 2001, 254–5). The large earth oven feature (not numbered) was buried under approximately a half- metre of overburden (Figure 4.14). The exposed profile showed evidence of in situ burning, demarcated by a layer of red ash on the east and west sides. This feature contained several flakes, charcoal, ash, and oven stones. Millerstrom and Coil (2008) suggest the possibility that the oven may have been re-used over a very long period of time.

Figure 4.14: Large earth oven near Vaiu’u River in profile (from Millerstrom and Coil 2008).

Charcoal was collected from all test units, though not all samples yielded material suitable for radiocarbon dating (Millerstrom 2001, 280). Ultimately, one fragment of charcoal was selected from each of five samples, and the results of these tests are graphed

78 below (Figure 4.15). None of the charcoal was identified, though it should be noted that few preferred short-lived materials such as nut shells or twigs were found in the assemblage. It has been acknowledged that in-built age may have affected these results (Millerstrom and Coil 2008). Of particular concern is the large earth oven, as the sample ages (LLNL-44515 and LLNL-44513) cover a period of time that spans much of the prehistoric sequence. The presence of breadfruit, Thespesia populnea, and Hibiscus in samples from this feature indicate it is possible these potentially long-lived species were dated (see Allen and Huebert 2014). A similar concern was expressed for the age of the oven in TU-1 (LLNL-44512) where these taxa were also identified. Millerstrom and Coil (2008) note that if this date was similarly affected by in-built age, both features in the unit would be of a similar age (i.e., mid-17th century AD at earliest; late prehistoric). The burnt surface in TU-4 (LLNL-44514) also contained Thespesia and Hibiscus. Overall, because historic materials were absent in the dated inland contexts, the authors concluded these deposits are pre-contact; if in-built age was a factor, then they could be late prehistoric.

Figure 4.15: Radiocarbon sequence for Hatiheu inland sites.

Table 4.5: Radiocarbon determinations for Hatiheu inland sites. Conventional calAD 2σ Lab No.1 Unit Depth Provenience Material2 14C age BP3 range4 LLNL- TU-1 40 cmbs Hearth Not known 100±50 1676-1941 44511 Efe. 1 LLNL- TU-1 47 cmbs Oven Not known 340±50 1453-1645 44512 Efe. 2 LLNL- TU-4 30-32 cmbs Burnt Not known 240±50 Post-1489 44514 surface LLNL- Vaiu’ua 120 cmbs Oven Not known 300±60 Post-1447 44515 River (no number, road cut as below) LLNL- Vaiu’ua 144 cmbs Oven Not known 540±60 1296-1448 44513 River (no number, road cut as above) 1 LLNL- prefix indicates Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory. 2 All material dated was charcoal. 3 Data reported by Millerstrom (2001). 4 Recalibrated with OxCal v4.23 (Bronk Ramsey 2009) using IntCal13 (Reimer et al. 2013) in an effort to provide uniformly calibrated set of dates for the present analysis.

Summary A small number of artefacts were recovered from these excavations, consisting mainly of lithic flaking debris. No faunal materials were reported for the seven study sites. In the main, these investigations provided information on the structures they adjoined, with evidence of pre-structural activities, construction, and remodelling noted. The earliest radiocarbon age in this assemblage derives from the large earth oven in the lower valley, which may have been used as early as the 14th century AD, though it may date to as late as the mid-15th century or later if affected by in-built age. While the date of this feature is somewhat uncertain, considering the Pahumano sequence (Allen and McAlister 2013) it is evident people were using the coastal lowlands of Hatiheu before the 15th century AD. Findings from the central valley locations studied by Millerstrom (2001; Millerstrom and Coil 2008) suggest these areas may have been cleared as early as the 15th century AD, and stone foundations were constructed by the 17th century AD. The pollen-sediment core of Allen and colleagues (Allen et al. 2011) supports these findings and suggests clearing activities elsewhere in the Hatiheu inland were even earlier. Millerstrom (2001) notes that the dense distribution of similar stone features and rock art imagery throughout the valley indicate it was densely occupied by the late prehistoric period, and cultural activities continued through the early contact period. It was also argued that the frequency of economic woods, including breadfruit, identified in the charcoal samples suggested that the

80 interior was extensively cultivated by the late prehistoric period (Millerstrom and Coil 2008). The authors further suggest these practises began significantly before many of the stone structures were built, though evidence to support that suggestion was infrequent.

Materials selected for analysis In total, fifteen charcoal samples from four excavation units (TU-1, 2 , 4, and 6) and one profile (large oven, not numbered) in the Hatiheu interior were selected to be part of the present analysis. Much of this material was collected from cultural sediment, but four features also produced charcoal, including two earth ovens (lower valley not numbered, and Efe. 1), one hearth (Efe. 2), and a stone-lined pit (no number). Details regarding provenience, size, and contents of these features are provided in Table 4.6.

Table 4.6: Cultural features from the Hatiheu inland sites selected for study (details from Millerstrom 2001). Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated number(s) None Oven Vaiu’ua River Profile: u- Profile FCR stained 100000, (lower valley) shaped, 170 cm only with ash and 100001, Road cut wide x 90 cm charcoal, 100002 deep (buried by flake 60-70 cm fragments, overburden), oxidized (red Plan: shape not ash) rim on known two sides

Efe. 1 Hearth Mutoka Profile: shape Ca. 1/2 Ash-covered 100013 (mid-valley) not known, 35 (rest FCR, TU-1 by 35 by 18 cm located charcoal Level 2 deep, Plan: below Sfe. round 200)

Efe. 2 Oven Mutoka Profile: shape < 1/2 Loose black 100014 (mid-valley) not known, 80 (remainder soil, angular TU-1 length by 20 cm below pebbles, ash, Level 2 deep, Plan: image basalt flake, shape not known boulder) many FCR

None Stone- Ototemui Profile: to 1.6 m Oven rake- 100003 lined pit (upper valley) rectangular, 12.5 out, adze, (pakeho) TU-6 by 12 by approx several 4-5 m deep, flakes Plan: square 1 FCR = Fire cracked rock

Anaho sites Anaho is a small, amphitheatre shaped valley that lies to the east of Hatiheu. It features a protected bay and coral reef. Along the coast to the south and east, plains are narrow, and land rises to steep valley walls. Over a low ridge to the east, a walking trail connects Anaho to the dunes of Ha’atuatua. There is a permanent spring-fed stream on the northern coastal flat, and an intermittent stream runs through the southern valley. While there is little orographic rain in Anaho, during heavy downpours waterfalls are visible on the steep cliff face of Tukemata to the southeast. The population of Anaho Valley was allied with Hatiheu by the historic period (Allen 2009a, 351–3), and there is little evidence Anaho was an important cultural centre. Cultural activities here date to early in the Marquesan prehistoric sequence (Allen 2014), and the valley was widely used in the prehistoric period. Anaho is notable for an abundance of archaeological remains, which are largely undisturbed due to the few inhabitants and lack of road access. Relict stone-faced agricultural terraces are located near springs and streams (some presently dry) (Addison 2006, 262, 280), and dryland terraces on the northern slopes of the valley (Allen in prep.) suggest food production activities were conducted throughout the valley before contact. Three assemblages derive from excavations in this valley: Teavau’ua (AHO-1) and Teavau’ua South, both subsurface sites on the northern coastal flat, and units excavated near stone structures in the interior.

Teavau'ua

Overview / Introduction Situated in a sheltered area of Anaho Valley, the Teavau’ua (AHO-1) site contained a well-preserved prehistoric stratigraphic sequence and was especially informative regarding domestic activities before the mid-17th century AD (Allen 2004b). This site is notable as it is adjacent to one of the few coral reefs in the archipelago, which would have been an important resource for both food and raw materials in the prehistoric period. Though the cultural sequence at this site commences in the 13th century AD (Allen 2009a and Table 4.7 below), the most intensive occupation dates between the 15th and 17th century AD. A variety of domestic activities are represented including food preparation and disposal, and pearl-shell and stone tool manufacture. The artefact assemblage from this site included numerous fragments of worked pearl-shell and fishhooks in various states of manufacture,

82 and basalt debitage, adzes, and adze preforms. Faunal materials included abundant fish bone and shellfish, and lesser quantities of mammal bone including pig, dog, bird, and turtle.

Figure 4.16: View of the Teavau’ua coastal flat looking south-southeast (from Allen 2004b).

Local setting, geomorphology, vegetation Anaho Valley, which lies immediately to the east of Hatiheu, opens on a deeply incised bay, which features a coral reef in the protected western sector. A narrow coralline sand coastal plain fringes much of the shore, and the coastal plain in the northern valley extends several hundred metres inland. A small semi-permanent stream associated with a spring runs through the southern end of the coastal flat. The study area lies approximately 100 m inland and to the north of this area, under an ageing commercially planted coconut grove (Figure 4.16). The land rises gradually to the west and more steeply to the north. There is some evidence that erosive events, or potentially storm events, have rapidly deposited sand in this area on occasion (Allen 2004b, 158–9). The flora in this part of the valley is of a mixed composition. A dry scrub forest is present on the steep slopes to the north and west of the coastal flat. A mostly closed-canopy moist forest surrounds the area to the west, composed of many Polynesian and historically introduced tree species, and Hibiscus tiliaceus is common. Though some of the northern coastal plain has been cleared, a few large old Casuarina, Calophyllum, and Cordia have

83 been planted in the surrounding area. Clusters of Thespesia, several large Terminalia, and a variety of fruit trees can also be found here. A group of modern buildings lies a short distance to the north of the site and a modern building, overlaying the foundations of a structure associated with the chief of Anaho (Allen 2004b, 159), is present near the shore.

Excavations This site was selected for study because it was hypothesised that the coral reef would have been an important resource for colonists (Allen 2004b, 150). Coring, shovel pits, and controlled excavations were conducted along several transects on the coastal flat in 1997 and 2001, and it was established that subsurface cultural activities were present in an approximately 200 by 300 m area. Allen returned to conduct an areal excavation at the Teavau’ua site in 2003, opening nine alternating 1m2 squares in an area approximately 15m2 (Figure 4.17). Additional units were opened in later seasons closer to the stream. All units were excavated manually in arbitrary 10 cm levels, following natural stratigraphy to sterile sand. Features were excavated separately, and the contents were screened and retained. When possible, the excavated sediments were screened with 3.2 mm (1/8 inch) mesh, otherwise 6.4 mm (1/4 inch) mesh was used. Charcoal was manually separated from the other recovered materials.

Stratigraphy, radiocarbon ages, depositional sequence A sequence of six stratigraphic layers were found at the Teavau’ua site, two of which were determined to be prehistoric. A typical profile from this excavation, illustrating an earth oven feature, is shown in Figure 4.18. The upper cultural layer at this site (IIIa) contained a mixture of historic and traditional materials, which indicate this area was used in the late prehistoric and early historic periods. The main prehistoric occupation at this site was found in Layer IIIb, a darkly stained sand matrix with abundant artefacts, including pearlshell debitage, fishhook blanks, adzes and preforms, flaking debitage, coral abraders, and charcoal. Faunal remains included fish bone, shellfish, some pig and dog bone, and a modest amount of bird bone. Features found in this stratum included several earth ovens, concentrations of charcoal, and rake-out from a hearth or oven; a collection of pebbles may represent a house pavement. Allen (2004b, 168–73) concluded this occupation represents intensive use of the coastal flat. The basal cultural layer at this site, Layer IV, represents

Figure 4.17: Map of excavation at the Teavau’ua site (from Allen et al. 2005). SP-2 (not indicated) is located off of rubbish pit adjacent to site datum.

initial use of the area, possibly on an occasional basis. A post-mould and shallow basin outlined by cobbles and filled with charcoal-stained sand suggests structures were present in the area during this period. Artefacts found in this stratum included worked pearlshell, fishhook blanks and preforms, coral and sea urchin abraders, and basalt debitage. Faunal materials found in Layer IV included low concentrations of fish, bird, dog, and pig bone, and shellfish.

Figure 4.18: A typical earth oven profile (Efe-7) in main occupation layer (IIIb) at Teavau’ua (from Huebert et al. 2010).

Figure 4.19: Radiocarbon sequence for layers IIIb and IV, Teavau’ua (AHO-1).

Most charcoal samples for radiocarbon dating were collected as bulk samples taken in situ from the centre of features or from the wall profiles after excavation was complete. The sample from SP-2 (BETA-108023) was taken from a smaller unit (50cm2) that was situated to investigate an oven feature exposed in the wall of a rubbish pit opened by the landowner. Ten radiocarbon determinations established the age of these occupations (Allen 2004b, 2009a; Allen and McAlister 2010; Petchey et al.2009), and the sequence is illustrated in Figure 4.19. Details of these samples are provided in Table 4.7. Radiocarbon samples from Layer IIIb were in some cases not identified and in other instances were not short-lived materials but the ages are, in the main, internally consistent. Samples from Layer IV include one from preferred material (nutshell, OZI-974), and others from this context overlap the range established by this sample.

Table 4.7: Radiocarbon determinations for layers IIIb and IV, Teavau’ua (AHO-1) site. Conventional calAD 2σ Lab No.1 Unit Layer Provenience Material2 14C age BP3 range4 WK-13833 TP-5 IV n/a Pearl-shell 1169±36 1161-1304 OZI-976 TP-11 IV Exposed Monocot 835±45 1047-1275 profile, #1078 wood OZI-975 TP-10 IV Exposed Unidentified 805±40 1164-1276 profile, #1076 broadleaf wood OZI-974 TP-8 IV Hearth (Efe. Nutshell 730±40 1217-1385 13) charcoal WK-10644 SP-10 IV Ash lens, 48 Wood 635±61 1272-1415 cmbs, #31.1 charcoal BETA- SP-2 IIIb #146.1 Wood 430±80 1323-1646 108023 charcoal WK-10844 SP-40+ IIIb, 60 #34.1 Tournefortia, 395±82 1399-1664 cmbs Terminalia, Ficus sp. WK-10645 SP-45+ IIIb, 27 #30 Wood 379±57 1440-1640 cmbs charcoal WK-10843 SP-20 IIIb, 60- #33.2 cf. 341±50 1453-1645 65 cmbs Tournefortia WK-19116 TP-9 IIIb Efe. 6 (oven) Thespesia 296±34 1488-1661 populnea 1 OZ prefix indicates Australian Nuclear Science and Technology Organisation, WK indicates Waikato Radiocarbon Laboratory, BETA indicates Beta Analytic, Inc. 2 All material listed was charcoal, identified by Rod Wallace. 3 Data reported by Allen (2004, 2009), Allen and McAlister (2010), and Petchey et al. (2009). 4 Recalibrated with OxCal v4.23 (Bronk Ramsey 2009) using IntCal13 (Reimer et al. 2013) in an effort to provide uniformly calibrated set of dates for the present analysis.

Summary The earliest cultural occupation found at the Teavau’ua site (Layer IV) indicated more ephemeral use of the area for fishing, fishhook production, and adze manufacture, the latter exploiting nearby sources of raw material. These activities occurred sometime starting in the 13th century AD. Allen (2004b, 182) notes that this occupation probably does not evidence the earliest use of Anaho, as the faunal assemblage from this stratum contained only small amounts of bird and turtle, species that are typically heavily exploited by early colonists in this archipelago (Rolett 1998, 241–4), and shellfish are of a modest size. Later occupation of the coastal flat (Layer IIIb) occurred sometime in the 15th to 17th century AD. This occupation was more permanent, evidenced by abundant occupation debris and features such as ovens, hearths, and post-moulds. It is notable that this occupation is similar

88 in age to an established occupation in neighbouring Ha’atuatua (Rolett and Conte 1995). After this period, it appears that intensive domestic activities were relocated elsewhere. People probably moved away from the coast at this time, as stone structures a short distance inland of this area have been dated to after the mid-17th century AD (Allen 2009a). It should be noted, however, that people may have continued to occupy the lowlands in this area. Some structures dating to the late prehistoric or early historic period are present near the shore, there is evidence of cultural activity on low slopes adjacent to the coastal flat, and a mixture of traditional and modern cultural materials in the uppermost cultural stratum at this site (Layer IIIa) could indicate the coast was re-occupied in the historic period. Increased alluvial deposition also is indicated by a higher clay content in Layer IIIa, possibly the result of anthropogenic activities or feral herbivore impacts that de-stabilised the surrounding slopes (Allen 2004b, 183).

Materials selected for analysis In total, twenty-seven charcoal samples from ten excavation units (SP-2, TP-8 through TP-16), at the Teavau’ua site were selected to be part of the present study. Some of this material derives from occupation debris. Eight features were also selected: five ovens (Efe. 5, 6, 7, 20, and D-97), two indeterminate fire features (Efe. 8 and 22), and a postmould (Efe. 25). Details regarding provenience, size, and contents are provided in Table 4.8.

Table 4.8: Cultural features from the Teavau’ua site selected for study. Details from Huebert (2009), Huebert et al.(2010), Allen (pers. comm.). Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated numbers Efe. 5 Oven TP-9 n/a n/a n/a 586, 602 Layer IIIb

Efe. 6 Oven TP-9, TP-12 Profile: Basin, c.2/3 Dense 615, Layer IIIb >1 m length, concentration 690, 882 35 cm thick, angular cobbles Plan: 150-160 and charcoal, cm diameter, underlain by shape n/a charcoal and white ash, more cobbles below. Some rocks large, rounded. Bottom edges feature pink oxidized sand. Boulders mark periphery.

Efe. 7 Oven TP-8 Profile: Basin, c. 1/4 Upper boundary 640 Layer IIIb length defined by undetermined, boulder set on 23 cm thick, edge. Angular Plan: 140 cm cobbles, diameter, charcoal shape n/a concentrated near base. Base of feature on one side is pinkish, oxidized soil.

Efe. 8 Indeterminate TP-10 n/a n/a FCR, bone, 1083 Layer IIIb/IV pearl-shell

Efe. 20 Oven TP-14 Profile: Basin, c.1/4 Upper edge 826 Layer IIIb length marked by two undetermined, small boulders, 25 cm thick, suggesting Plan: ~130 cm feature may diameter, have been shape n/a bounded by more of same. Small (10 cm), angular cobbles and charcoal, fishhook head and shank in upper surface of feature.

Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated numbers Efe. 22 Indeterminate TP-15 Plan: approx. n/a Circular area of 845 Layer IV 45 cm dark sand diameter

Efe. 25 Post-mould TP-16 n/a n/a n/a 884 Layer IV

Efe. D-97 Oven SP-2 Profile: basin, uncertain Several angular 139, 146 Layer IIIb length cobbles and uncertain, 25 dense charcoal. cm thick 1 FCR = Fire cracked rock

Teavau’ua South

Overview / Introduction On the south side of the stream is another prehistoric site, simply refereed to here as Teavau’ua South, as details of this excavation are forthcoming (Allen, in prep.). The site is briefly summarised to provide context to the samples used in this study.

Local setting, geomorphology, vegetation The Teavau’ua South site is located approximately 125 m inland from the Anaho coast, and ca. 200 m south of the stream on the narrow coastal flat. This flat, sandy plain lies behind a slight rise at the modern shoreline. The coastal strip extends to the south, and land rises into the forest to the west. Traditional stone house foundations are present to the west and southwest of this area. Vegetation in and around the site consisted mostly of mature coconut palms, many of which have been recently felled, a few Thespesia, and fruit trees. Occasional low thickets of Caeselpinia and Leucaena were present immediately inland of the site. In 2011, the under-storey in this area had been cleared within approximately 50 m of the coastline, and low ornamental shrubs and small trees had been planted at various locations. Several small modern structures have been built in this area in the past, but only concrete foundations remain.

Excavations During the 2011 field season, an area of the Anaho coastal flat south of the Teavau’ua stream was investigated by Allen and a team that included the author. Initially several shovel pits were opened along transects perpendicular to the coast to determine the nature of subsurface deposits in this area, which had not been previously explored. A site estimated at 10 by 15m2 was encountered. Five units (TP-49 through TP-53) were opened in an area approximately 6 by 12m2. All units were excavated following natural stratigraphy in 10 cm levels to sterile. Features were excavated separately. When possible, the excavated sediments were screened with 3.2 mm (1/8 inch) mesh, otherwise 6.4 mm (1/4 inch) mesh was used. Charcoal was routinely separated from other materials at the time of dry screening. In addition, ten 10 L bulk sediment samples from three units (TP-47, 50, 53) were processed by water flotation using a municipal tap near the site, and charcoal was separated from other materials when dry.

Stratigraphy, radiocarbon ages, depositional sequence A sequence of five stratigraphic layers were found at this site, two of which were determined to be prehistoric cultural deposits (Layers III and IV). The sediments in both strata consisted of a brown to yellow-brown sandy loam. Layer IV evidenced a more intensive occupation. The estimated age of the this site is ca. AD 1400 to 1650 (Allen, pers. comm. 2013).

Summary The features, artefacts, and faunal materials found at this site suggest this was an occupation area where food preparation activities and stone tool use and manufacture occurred. These activities are similar to those encountered to the west at the Teavau’ua site (AHO-1). No traditional stone foundations were observed in the immediate vicinity.

Materials selected for analysis Twenty-seven samples were selected from four units at this site (TP-49 through TP- 53). Though most of this material derives from occupation debris, three cultural features also yielded identifiable charcoal, including two features of undetermined function (Efe. 78,

83) and one post-mould (Efe. 79) The contents of these features are currently under review, and details that were available have been provided in Table 4.9.

Table 4.9: Cultural features from the Teavau’ua south site selected for study. Charcoal Portion Number Type Provenience Dimensions Contents sample excavated number(s) Efe. 78 Fire feature TP-51 Profile: ca. ca. 1/4 n/a 7125 Layer III 20 cm deep maximum, Plan: 35 by 45 cm Efe. 79 Post-mould TP-51 n/a n/a n/a 7240 Layer IV Level 6 Efe. 83 Indeterminate TP-53 n/a n/a n/a 7058, 7059 Layer IV Level 4

Anaho stone structures

Overview / Introduction Many types of traditional stone structures are located throughout Anaho Valley, from near the coast to several hundred metres inland, and some are scattered at slightly higher elevations in the southern valley. The sizes and forms of many of these features are consistent with the idea that they were residential structures belonging to a range of commoners and elites (Allen 2008, 2009a). Approximately three hundred structures with varying morphologies were surveyed over a .78km2 area of the valley, and excavations were conducted at select locations in an effort to understand temporal patterns of land use. Structures were determined to be late prehistoric in age, and they are currently being investigated as proxies of socio-political process (Allen 2009a). A variety of domestic activities are represented at these sites, and some deposits underlying the structures provide evidence of prior activity. Artefacts, faunal assemblages, and features were recovered from some of these excavations, though quantities and types of material varied.

Local setting, geomorphology, vegetation Stone structures that were included in the present study are located throughout the valley, from near the coast to the back-valley wall. A number of structures are located in the northern sub-valley near the stream, others are situated in the southern sub-valley. While the coastal sediments in this location are sandy, inland soils are rocky. Micro-climate conditions range from xeric on the northern slopes to mesic in the south, and they are variable throughout the valley. The northern valley is watered by a permanent stream, as previously mentioned. With the exception of structures near the coast, where vegetation has been cleared and replanted in recent years, stone structures are surrounded by mature forests of varying composition. Mature trees were sometimes observed growing on top of the platforms. The forests of Anaho include both native and introduced elements, including useful economic trees such as the occasional breadfruit, Premna, Maytenus, Syzygium malaccense, and citrus trees. Moist lowland forests are composed of numerous thickets of Hibiscus tiliaceus and Pandanus, interspersed with many coconut trees, and occasionally Inocarpus fagifer trees are present in moister areas. Scrub forest on the drier slopes and near ridge tops includes Sapindus saponaria and others. Clusters of gregarious introduced tree species such as mango, candlenut, noni, and Adenanthera pavonina, were also noted in some areas. A panoramic view of the valley setting is provided in Figure 4.20.

Excavations As part of a larger research programme in Anaho, initiated in 2003, Allen has investigated the chronology and function of several architectural features located throughout the valley (Allen 2004b, 2008, 2009a, 2009b, 2009c; Allen and Addison 2002; Allen and McAlister 2004, 2006, 2007, 2008; Allen et al. 2005). Other archaeological work has taken place as part of this programme, including a GIS study of late prehistoric community spatial patterning (Crown 2009), specialised studies of stone and shell tools (Allen and Ussher 2013; McAlister 2011; Ussher 2009), faunal materials (Burt 1999; Cowie 2009; Littleton et al., in prep.), and charcoal (Huebert 2008, 2009; Huebert, Allen, and Wallace 2010). Material included in the present study derives from excavations at raised stepped stone platforms and terraces, simple foundations, pavements, and an earthen storage pit (Allen 2009a, Table 2). The locations of these structures are designated in Figure 4.21. In the case of domestic house foundations, excavation units were often limited in number and size, and aimed mainly determining the age of these structures. At Structure 13, however, an areal

Figure 4.20: Anaho Valley from the overpass to Hatiheu. Note the coral reef and reef break associated with the Teavau’ua stream in the centre of photo.

excavation provided further insights into on-site activities, and at other structures the excavations revealed earlier occupations that were not associated with the surface architecture. In general, the excavations followed natural stratigraphy and features were excavated separately. Charcoal was manually separated from the other materials recovered. When possible, the excavated sediments were screened with 3.2 mm (1/8 inch) mesh (e.g., in sandy sediments), otherwise 6.4 mm (1/4 inch) mesh was used, and all materials recovered during excavation were retained.

Stratigraphy, radiocarbon ages, depositional sequence Test units included in the present analysis were associated with Structures 68, 16, 336, 13, 11, 8, 32, 24 in the north valley, and Structures 2, 232, 242, 245, and 254 in the south valley. While some of these excavations were very productive, sediments in other areas were very rocky and not well developed and few cultural materials were found. A number of subsurface features were encountered in these excavations, including numerous hearths and ovens, as well as other features such as post-moulds, pits, and thin charcoal lenses, some associated with underlying cultural occupations that predate the surface

Figure 4.21: Map of Anaho stone structures included in the present study (adapted from Allen 2009a).

96 architecture. On several occasions, cultural deposits that included charcoal concentrations or oven features were encountered in strata underneath structures. Brief summaries of these excavations are provided below, grouped by locality, as details have been published elsewhere (see references above). Structure 68 is a low stone pavement located on the coastal flat in the north valley (Allen 2004b; Allen and Addison 2002). In the test unit at this location (TP-7), domestic features including a hearth (Efe. 5) were found in association with the structure. Worked pearl-shell, flakes, and bone indicate a number of domestic manufacturing activities took place at this location, and a cache of charred candlenut shells found at this site could indicate tattooing also took place there. A cultural deposit was identified underneath this structure, with probable domestic activities denoted by the presence of an oven (Efe. 6) and a possible post-mould (Efe. 8). Two other structures included in this study are located less than 200 m to the southwest, south of the Teavau’ua stream. They include a partially buried stepped terrace (Structure 16) and a square stepped platform (Structure 13). Two small hearths and a postmould were encountered in the excavation unit at Structure 16 (TP-17), but few artefacts were recovered (Allen 2009a, Fig. 11). Structure 13 was originally investigated by a single test unit, TP-36, placed against the front face of the paepae to evaluate the age of the structure (Allen and McAlister 2007, Fig. 7). A subsequent areal excavation at the front of this structure (test units A10–D14) produced shell, pig, and fish bone, suggesting the area in front of the structure was used for food preparation (Figure 4.22). An earth oven (Efe. 3) and stone-lined pit (Efe. 28) lacking charcoal, possibly used to process breadfruit, were encountered in these excavations. Allen (2009b) suggests this structure was probably a non- elite domestic residence. There is a group of structures to the south of this location that include partially buried house foundations, pits, stone alignments, and stone walls located by a dry streambed (Allen 2009a; Allen and McAlister 2008). Structure 11 at this site is a very large, undoubtedly elite, stepped platform (Allen 2008, 2009a Figs. 8 and 14). Excavations at this structure (TP-23) produced some bone and shell; no features were encountered but there was dispersed charcoal in the underlying stratum. Less than 40 m to the north of this cluster is Structure 336, a small pavement, where three square meters were opened to assess the age of low pavements in relation to raised foundations. Although the excavations have not been published in full, Allen and Ussher (2013) report on a bivalve scraper and associated starch residues from this late prehistoric

97 site. Excavations revealed a second cultural layer underlying the surface pavement (Allen 2009c). TP-42 contained a hearth (Efe. 9) and oven (Efe. 10), along with bone and shell. Small quantities of flakes and worked pearl-shell are indicative of domestic manufacturing activities.

Figure 4.22: Plan view of Surface Feature 13, showing excavation area (shaded). Black arrow indicates location of earth oven (Efe. 3) (from Allen and McAlister 2007, Fig. 7).

Southwest and inland of this area is Structure 24, a stepped platform. The unit excavated at this structure (TP-22) contained no subsurface features (Allen and McAlister 2004). Some unworked shell was found in this excavation. Approximately 100 m southeast lies Structure 32, a stepped platform (Allen 2004b, Figure 2; see also Allen 2009a, Table 2). The test unit excavated at this location (TP-31) did

98 not contain subsurface features, and the few artefacts encountered included mainly lithic flakes. Downhill and seaward of this structure is Structure 8, another stepped platform. Test units included TP-24, placed a short distance from the structure due to an obstruction, and TP-25. A very dense concentration of fish bone and some bird bone was encountered in these units, along with a large earth oven (Efe. 26). An earlier stratum with dispersed charcoal was located below this structure (Allen and McAlister 2004; Allen et al. 2005). Structure 2 is situated along the coast in the south sub-valley (Allen 2008, 2009a Fig. 9, 2009c). This house foundation, which features an alignment of upright tuff stones, is attributed to Queen Kaniho of Anaho, who is thought to have resided here in the early historic period. Three test units were placed adjacent to this structure: one against the north face (TP-44), one against the west face (TP-45) and another a few metres west in front of a small platform (TP-46). The stratigraphy of these units was complex, and seven layers and numerous features were identified, including many post-moulds and a few pits and hearths. The basal cultural layer (Layer IV) pre-dates the structure, and it contained a large quantity of basalt flakes indicating this was probably a tool manufacturing area. Relatively few faunal materials were recovered from the lower strata at this site, but a wealth of fauna were found in the upper strata associated with the structure (Allen 2009c). In the south sub-valley, approximately 300 m inland atop a flat area on the crest of a ridge, are several structures and a small (Structure 232) (Allen 2009a, Fig. 15, Table 2). A test unit (SP-1) was excavated at the edge of this structure. An oven was encountered below and extending underneath the structure, but few cultural materials were recovered. Excavations at other structures in this group suggested to Allen (2009a, 366) that this area was probably used at several different times in the prehistoric and historic periods. Farther back in the south valley, the Te Papa Uka area is situated in a defensible position at the back of the valley wall near the trail to Hatiheu Valley (Allen 2009a; Allen and McAlister 2007). This area includes a group of six or seven stone structures, including Structure 242, 245, and 254. The former is a large, simple platform that is probably not a domestic residence. TP-38 was located between this feature and Structure 245, an earthen pit located behind the larger structures, possibly a facility to store fermented breadfruit. TP- 39 was placed to section the rim and interior of this pit. Few cultural materials and little charcoal were encountered in these units, and a small earth oven (Efe. 29) in TP-38, was the only feature noted. Feature 254 is a large house foundation located prominently at the end of

99 a ridge (Allen 2009a, Fig. 10); the test unit at this location (TP-40) recovered only a few shell and charcoal fragments. Charcoal was collected for radiocarbon dating from all units at these sites, per procedures described for Teavau’ua. All wood charcoal was identified before being submitted for dating. Though many samples did not derive from preferred short-lived materials, the corpus of dates associated with use of the structures were internally consistent and suggested initial use no earlier than the mid-17th century AD (Allen 2009a, 359) and Figure 4.23, below. Strata with cultural materials below the structures provided a wider range of possible ages. Some of these samples contain materials that derive from long-lived tree species, including Cordia subcordata, Terminalia sp., Barringtonia asiatica, and Cocos nucifera wood, all of which can live for nearly a century (Allen and Huebert 2014; Allen and Wallace 2007). While these samples may have yielded ages older than the cultural activities they were intended to date, several samples of nutshell (OZI-977, OZK-037) produced pre-15th century ages, indicating activities were occurring in both the north (Structure 8) and south (Structure 254) areas of the valley at an early date (Table 4.10). Additional evidence of early activities, not discussed herein, were also found at Structure 2 (#6249, 6366, 6359), where one sample (#6351) produced a very early pre-13th century AD age (Allen 2014); the aforementioned materials have been included in the present study.

Summary The results of this study indicate that people were using different areas of Anaho early in the prehistoric period, perhaps as early as the 11th century AD, and activities were not limited to the coastal plain. Cultural activities in the valley were thereafter essentially continuous into the contact period. A significant finding of this long-term research programme was that elaborate stone architecture was constructed late in the prehistoric sequence, after approximately AD 1650 (Allen 2009a). It was also noted that low stone pavements found near larger stone structures were not necessarily transitional forms, as suggested by Suggs (1961), but rather were contemporaneous with raised structures and possibly used for domestic activities such as food preparation. Evidence of remodelling was noted at some structures, indicating they were maintained and modified throughout their histories. It was also evident from the presence of historic artefacts near some foundations —and the memories of local elders—that they were occupied into the 19th and 20th centuries, and some are still occupied today.

100

Figure 4.23: Radiocarbon sequence for Anaho stone structures. Dates for structures (above) and deposits below structures (below).

101

Table 4.10: Radiocarbon determinations for Anaho stone structures. calAD Lab Conventional Structure Locality Unit Provenience Material2 2σ No.1 14C age BP3 range4 Directly associated with structures OZK- None South TP- Layer I #4390 Artocarpus 210±35 After 040 (earthen valley 39 altilis AD storage (Te Papa 1641 pit) Uka) WK- 46 Mid- ST-2 Layer Ib Thespesia 206±33 After 16732 valley populnea and 2 AD undetermined 1643 species WK- 68 North TP-7 Concentration Aleurites Modern After 10646 coast of burn moluccana AD nutshell endocarp 1750 WK- 13 North TP- Layer I #1039 cf. Cordia 178±34 After 16730 stream 20 subcordata AD 1651 WK- 24 North TP- Layer IIa Hibiscus, 153±36 After 13835 valley 22 Sapindus and AD lowland others 1666 WK- 29 Mid- TP- Layer I T. populnea, 125±35 After 16729 valley 21 Sapindus AD saponaria, 1675 undetermined wood OZK- 242 South TP- Oven Efe. 29 Monocot 105±35 After 036 valley 38 #4249 ‘wood’ AD (Te Papa 1680 Uka) WK- 8 North TP- Layer I Sapindus, 100±34 After 16731 valley 24 Artocarpus and AD lowland others 1681 OZK- 16 North TP- Layer I T. populnea 65±45 After 978 stream 17 AD 1682 Below structures OZI- 8 North TP- Layer IIb nutshell 855±45 1043- 977 valley 24 charcoal 1264 lowland WK- 232 Mid- SP-1 Layer II Indeterminate 716±38 1223- 16735 valley 1387 OZI- 232 Mid- SP-1 Layer II C. subcordata 510±40 1318- 980 valley 1450 OZK- 254 South TP- Layer II Nutshell or 735+35 AD 037 valley 40 #4268 fruit endocarp, 1251- (Te Papa cf. coconut 1275 Uka) WK- 11 North TP- Layer II Hibiscus sp. 341±44 AD 13836 stream 23 #1041 1458- 1643

102

calAD Lab Conventional Structure Locality Unit Provenience Material2 2σ No.1 14C age BP3 range4 OZI- 16 North TP- Layer III cf. C. 285±40 AD 979 stream 17 subcordata 1484- 1797 OZK- 11 North TP- Layer II T. populnea 285±35 1490- 035 stream 23 1794 WK- 16 North TP- Layer II cf. 264±38 After 16728 stream 17 Cyclophyllum AD sp. or Eugenia 1490 sp. WK- 32 Mid- TP- Layer II T. populnea, S. 260±34 After 16734 valley 31 saponaria, C. AD subcordata 1491 WK- 68 North TP-7 Oven D-01 T. populnea, 239±48 After 10842 coast under Hibiscus, and AD pavement others 1491 #110.2 WK- 33 Mid- TP- Layer II T. populnea, S. 110±34 After 16733 valley 30 saponaria, H. AD tiliaceus 1680 OZK- 13 North A11 Layer I #4280 Terminalia sp. 125±40 After 038 stream AD 1672 1 WK prefix indicates Waikato Radiocarbon Laboratory, NZ; OZ prefix indicates Australian Nuclear Science and Technology Organisation; BETA prefix indicates Beta Analytic, Inc. 2 Identifications by Rod Wallace. 3 Data reported by Allen (2009). 4 Recalibrated with OxCal v4.23 (Bronk Ramsey 2009) using IntCal13 (Reimer et al. 2013) in an effort to provide uniformly calibrated set of dates for the present analysis.

Materials selected for analysis In total, forty-eight charcoal samples from thirteen stone structures were selected for this analysis. Though much of the charcoal in this assemblage derives from occupation debris, fifteen cultural features also yielded identifiable material. They include seven ovens (Efe. 3, 6, 37, 26, 29, and D-01), three hearths (Efe. 5, 40, and C-01), a possible post-mould (Efe. 8), and four indeterminate fire features (Efe. 7, 28, 59, 61). Details regarding provenience, size, and contents are provided in Table 4.11.

103

Table 4.11: Cultural features from the Anaho stone structures selected for study. Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated number(s) Efe. 3 Oven Sfe. 13 Profile: 3/4 Large number 4201,4218 C14 undetermined of cobbles Layer I shape and concentrated length, 60 cm near top of thick, Plan: feature, circular charcoal which was more dense in lower half of feature

Efe. 5 Hearth Sfe. 68 n/a n/a With stones 105 TP-7 Layer I

Efe. 6 Oven Sfe. 68 n/a n/a n/a 62 SP-6

Efe. 7 Indeterminate Sfe. 68 n/a n/a FCR, flakes 61 fire feature SP-6

Efe. 8 Possible Sfe. 68 n/a n/a n/a None post-mould SP-6 (temporary #9999 assigned)

Efe. 40 Hearth Sfe. 336 n/a n/a n/a 6123 TP-42 Layer III?

Efe. 37 Oven Sfe. 336 n/a n/a n/a 6094 TP-43 Layer III

Efe. 26 Oven Sfe. 8 Profile: 1/3 Large number 1054, TP-24 Shallow basin, of angular 1049 Layer I undetermined cobbles and length, 35 cm dense charcoal thick, Plan: undetermined shape, over 150 cm dia.

Efe. 28 Indeterminate Sfe. 13 n/a n/a Lined with 4074 TP-36 stones, no Layer II charcoal

104

Charcoal Portion Number Type Provenience Dimensions Contents1 sample excavated number(s) Efe. 29 Possible oven Sfe. 242 Profile: Portion Some angular 4237, TP-38 undetermined unspecified rocks and 4245 Layer I shape, Length, charcoal, no 40-45 cm soil oxidation thick, Plan: undetermined shape, approx 75 cm dia.

Efe. 52 Oven Sfe. 2 n/a n/a Few basalt 6356 TP-44 cobbles Level 4 or 6

Efe. 59 Indeterminate Sfe. 2 n/a n/a n/a 6240 TP-46

Efe. 61 Thin charcoal Sfe. 2 n/a n/a n/a 6249, lens TP-44 6359 Level 9 Efe. C- Hearth Sfe. 68 Profile: n/a Charcoal 76 01 TP-7 undetermined concentration Layer I shape, 35+ cm length, 5 cm thick, Plan: undetermined shape, 35+ by 30+ cm dia.

Efe. D- Oven Sfe. 68 Profile: n/a n/a 110, 110.1 01 TP-7 undetermined Layer I shape, 40+ cm length, 15 cm thick, Plan: undetermined shape, 37+ by 40+ cm dia 1 FCR = Fire cracked rock

105

Summary The archaeological investigations from windward Nuku Hiva that have been summarised in this chapter represent at least a six hundred year prehistoric sequence from an early occupation of the archipelago at least by the 12th century AD extending to the late 18th century AD. Sites included in this review were located in three valleys of contrasting catchment sizes, topography, hydrology, and rainfall. Several sites are located in high-status areas or near structures that have been attributed to higher-ranking members of Marquesan society. These include four stratified sites near the coast, and test units at numerous discrete inland settings adjacent to stone structures. Functional contexts of the samples include strata in occupation sites that varied from ephemeral to permanent, and a variety of subsurface features including numerous earth ovens and hearths are represented. Site summaries were presented to provide the functional and temporal contexts of the samples, as most of the excavation reports have been previously published by the principal investigators. This information provides a foundation for the next chapter, where the charcoal assemblages and analytical methods used in this study are presented.

106

Chapter 5: Methodology

Introduction Palaeoecologists and archaeologists have developed an important body of work relating to the reflexive long-term changes that occur in humans and the plant environments they live in. Much of this research has relied on the analysis of pollen, spores, and charcoal concentrations in sediment cores, but studying human-plant interactions at regional scales are challenging because many plants cannot be detected in these records. Carbonised botanical materials from archaeological contexts, on the other hand, provide direct low-level information about human use of plants in the past. In combination, these data sets inform broadly on plant environments and human activity in the distant past, and are useful to address big-picture research questions (e.g., Frawley and O’Connor 2010; Nelle, Dreibrodt, and Dannath 2010; Touflan, Talon, and Walsh 2010). Archaeological wood charcoal analysis is an important component of this research. In this chapter, I first review methods generally used to analyse charcoal from archaeological contexts and then describe the methods specifically used in the present study. Methods archaeologists use to collect and analyse fossil wood charcoal are reviewed in the first section. This review begins with a technical description of wood structure and the charcoalification process. A general background of the practice of archaeological wood charcoal analysis is then presented, and some important Polynesian studies are reviewed. Current methods used to study archaeological wood charcoal assemblages are then explained. Recovery and analysis methodologies have been developed by palaeoethnobotanists to study all types of plant materials from archaeological contexts (recently reviewed by Wright 2010), and methods of analysing charcoal have been further refined by anthracologists to perform vegetation reconstructions and, more recently, to study past human activities (reviewed by Théry-Parisot, Chabal, and Chrzavzez 2010). While these two bodies of literature are not often cross-referenced, it was noted that they share many similar themes and both are incorporated into the following review. In this review, I consider various parameters that affect archaeobotanical assemblages including cultural, natural, and analytical influences, and evaluate the current state of research within the discipline. In the last part of this chapter, methods used in the present study are described, including details of the field methodology, procedures used to assemble the reference collection, and laboratory methods. The rationale used in sample selection and a summary of the sub-sampling procedures are also described.

107

Several terms are defined before delving into a methodological review. First, studies of ancient plant materials are alternately referred to as palaeoethnobotany and archaeobotany, though these two terms have somewhat different meanings. Archaeobotany is the study of plant materials preserved in archaeological contexts, and it encompasses the technical and scientific analyses of these remains. It is considered fundamental to palaeoethnobotany, which is a more interpretive approach focusing on the study of past human and plant interactions (e.g., Ford 1979; Hastorf and Popper 1988; Pearsall 2000). The sub-field of wood charcoal analysis is sometimes referred to as anthracology, a term used primarily by researchers affiliated with the University of Montpellier 2 in France (e.g., Chabal et al. 1999).

Wood structure Wood is a rigid plant tissue used for the transport of water and nutrients. These tissues are incrusted with lignin, and organic compound, which provides structural support for plants to grow into large forms such as shrubs and trees. Many parts of trees including the trunk, branches, roots, and seeds contain woody tissues (Evert and Esau 2006). Many gymnosperms and angiosperms, and some ferns, produce woody tissues. They include the conifers (softwoods) as well as several other gymnosperms, the dicotyledons (hardwoods), and woody monocotyledons (most notably, palms and bamboos). The internal structure of hardwoods, softwoods, and woody palms are different. Both hardwoods and softwoods have similar xylem structures composed of cells that support two systems: an axial system, which primarily transports water and nutrients up and down the stem, and a radial system that conveys these materials from the outside to the centre of the stem. Woody monocotyledons have a much simpler structure composed of vascular bundles that include xylem and phloem, and rigid fibres surrounded by a mass of parenchyma tissues. Transverse sections of these materials are illustrated in Figure 5.1. The axial system of a hardwood tree is composed of various types of cells that have different functions including vessels, axial (longitudinal) parenchyma, and fibres. The radial system is entirely composed of parenchyma cells. Anatomists describe wood structure in a standardised way based on features of these two systems, which are observed at both macro- and microscopic levels. Wood identification typically involves the examination of three planes (Figure 5.2) that are carefully exposed for study by cutting small square blocks with

108

Figure 5.1: Cross section of (a) hardwood (Artocarpus altilis), (b) coconut wood, and (c) Cordyline stem. Scale bar = 500 μm.

a microtome or razor blade. The transverse plane (TS) displays a cross section of vessels, axial parenchyma, and fibres, and a lateral view of ray cells. The tangential longitudinal plane (TLS) reveals a cross section of ray parenchyma cells, and a longitudinal view of the axial cells. The radial plane (RLS or RS) shows a distinctive longitudinal view of ray cells, and an additional longitudinal view of the axial system cells.

Figure 5.2: Three planes of a typical hardwood (from Butterfield and Meylan 1980).

Vessels arrangements in hardwoods are classified into several high-level categories based on vessel porosity that are described as ring-porous, semi-ring porous, or diffuse porous (Wheeler, Baas, and Gasson 1989). Vessel size and grouping arrangements are also distinctive features. Rays are classified by a combination of attributes including width, which can range from one cell wide (uniseriate) to many cells wide (multiseriate), and by cell orientations (procumbent, square, and/or upright) (Kribs 1935). Axial parenchyma patterns are also important for wood classification; they are grouped by degree of association with vessels, by shape, and/or by width when in continuous bands. Many other

109 features are important to wood classification, including minute attributes of cellular anatomy, the presence of mineral inclusions, and presence of various secretory elements. Comparative wood anatomists have demonstrated that closely related taxa share many anatomical features (e.g., Carlquist 2001, 335–350). These distinctions should, in theory, provide a means to identify wood from trees that are closely related, but in practice there are a number of internal and external factors that also influence anatomical variation. Some are briefly reviewed to illustrate the challenges to and limitations of wood identification. Firstly, the rate of wood formation within a tree is influenced by numerous environmental factors. Structural changes may be affected by variations in the local environment including available moisture, temperature, and soil conditions (Metcalfe and Chalk 1950, 2:152–6). For example, vessel diameter and length has been shown to vary in some woods based on the elevation, latitude, or amount of rainfall the tree has been exposed to (e.g., van der Graaff and Baas 1974). Variations in local climate can also influence ring porosity and vessel size, and these differences can be notable in trees that have a wide biogeographic range. For example, those growing in temperate zones tend to produce larger and more frequent vessels early in the growing season. Secondly, sapwood anatomy can vary from heartwood. While sapwood is living tissue, heartwood is not. Oftentimes it darkens in colour as deposits accumulate in the tissues, and ultimately heartwood becomes less permeable (Soerianegara and Lemmens 1993, 26). Wood structure is also influenced by when and where within the tree it was formed. A typical growth pattern is illustrated in Figure 5.3. New wood is formed by the cambium, a thin layer of cells that lie beneath the inner bark. While mature cambium produces outerwood as parts of the tree continues to grow in girth, young cambium produces juvenile wood (sometimes called corewood). This wood is typically formed at the crown and the tips of branches, and is found in the inner 5 to 20 (or more) rings throughout the tree, with hardwoods falling at the lower end of this range. Juvenile wood has notable differences in cell size and arrangement including smaller, shorter vessels that tend to be more frequent, narrow rays with fewer cells than outerwood, and shorter fibres with thinner walls (Bendtsen 1978; Evert and Esau 2006; Lachenbruch, Moore, and Evans 2011; Lev-Yadun 2007; Metcalfe and Chalk 1950, 2:42-43; Rendle 1960). Aside from the aforementioned studies, which are generalised descriptions of wood formation, little specific information could be located regarding juvenile wood.

110

Figure 5.3: Corewood / juvenile wood (white) and outerwood / mature wood (grey). Variability based on (a) position of crown and (b) cambial age (from Lachenbruch et al. 2011, Fig. 5.1).

Another aspect of wood variability to consider are tree rings. This feature is of particular interest in a study of tropical woods, as they are often assumed to be absent, but research has demonstrated this is not always the case. Tree ring porosity is often used to classify woods from temperate regions as patterning can be very distinctive. Rings are formed when the cambium becomes dormant; tree growth slows considerably and a fine line of thin-walled marginal parenchyma is created (Metcalfe and Chalk 1950, 2:44). Dormancy generally occurs when leaves fall from the tree. In temperate climates, this typically occurs as temperatures drop, though it can also be triggered by very dry or wet conditions. In tropical climates the growth of most trees also slows, though researchers have not been able to precisely identify external triggers (Soerianegara and Lemmens 1993, 28–31). Wood anatomists have recently confirmed that tropical trees can produce rings at varying (i.e., not necessarily annual) rates and though they may be visible in some species, in many they are faint and sometimes invisible (Worbes and Fichtler 2010). Therefore, while ring presence may assist in some tropical wood identifications, they are ultimately not reliable features. Finally, tree growth patterns can also react to fungus and insect pests, which can produce complicated modifications to the typical anatomy. Insects can, for example, defoliate trees to the extent that growth slows considerably. Wood that forms around damage, knots, under pressure, or under the influence of strong winds can contain cells that are atypically aligned (e.g., Schweingruber 2007).

111

Overall, certain environmental influences that affect wood anatomy have been more extensively studied than others, and causes of variation may not be well-understood. Though variability is broadly described in textbooks, woods from temperate regions are the most commonly cited examples, with woods of interest to the timber and paper industries being the most extensively documented types.

Charcoal The term charcoal can refer to many types of carbonised plant materials including wood, seeds, nuts, fruits, and parenchymous (storage) tissues. Charcoal is an inert material largely composed of carbon, though it does have varying chemical and physical properties that relate to both the composition of the originating material and the heat source that was applied. These variables include differences in anatomy, degree of and direction of shrinkage, reflectance values, and elemental composition (Braadbaart and Poole 2008). Wood becomes charcoalified when it is exposed to sufficient heat (usually between 300 to 400 degrees Celsius, per Braadbaart and Poole 2008) over time. Charcoal is formed when wood smoulders in the absence of oxygen; it can also form when wood exposed to open flame is incompletely combusted. During the charcoalification process, wood shrinks. As temperatures increase, wood shrinks to a greater extent, it becomes more reflective, and the carbon content increases. With enough heat and time, wood structure disintegrates and charcoal becomes ash (Figure 5.4). The anatomical structure of wood is generally preserved when charcoalified, making taxonomic identifications possible. The structure of woody monocotyledon tissues are less distinctive, though it is straightforward to distinguish between different types such as coconut, Pandanus, and Cordyline. Nut shells, which can be abundant in Pacific archaeological contexts, are often well-preserved (e.g., Yen 1974b, 1982, 1991) , and fracture in predictable ways. Fragments of charred and carbonised parenchymous tissue, which can derive from underground storage tissues and a variety of other plant parts, are sometimes frequent but can be indistinct (but see Hather 2000a). On occasion, the external morphology of underground storage organs or fruit endocarp are retained and can be recognisable (e.g., Coil 2004; Coil and Kirch 2005; Hather and Kirch 1991; Kahn and Ragone 2013; Ladefoged, Graves, and Coil 2005; Rosendahl and Yen 1971; Yen and Head 1993), though these materials are rare.

112

Figure 5.4: Processes of charcoal formation (from Braadbaart and Poole 2008).

Overall, charcoal tends to be durable and it is not subject to the biological processes that would otherwise destroy woody plant parts, but it can disintegrate into very small fragments when trampled, abraded, or subject to repeated wetting and drying cycles. Wood charcoal is the most common material recovered from archaeological contexts by either sieving or flotation (Pearsall 2000, 144). In the Pacific region, charred wood and nut shells form the bulk of archaeobotanical material recovered in excavation (Hather 1992) as few other botanical materials survive.

Brief history of the discipline Archaeological wood charcoal analysis was one of the earliest forms of archaeobotany to be practised, dating to the mid-19th and early 20th centuries (e.g., Fietz 1926; Heer 1865, 38–41). However, the work of Salisbury and Jane (1940) is generally considered the inception of wood charcoal analysis as a specialised discipline. Their study was one of the first to attempt to reconstruct vegetation histories by examining archaeological wood charcoal on a large scale. The authors hypothesised that the proportions of charcoal taxa recovered from domestic fires should equate with the

113 proportions of tree species in the local area in the past. This proposal was rejected by their contemporaries (Godwin and Tansley 1941) because they did not account for cultural factors that influenced fuelwood collection. This classic debate is still widely cited as an example of the challenges inherent in interpreting archaeobotanical assemblages (e.g., Marguerie and Hunot 2007; Marston 2009; Smart and Hoffman 1988). In the mid-20th century, two European researchers made major contributions to the discipline. Cecilia Western analysed wood charcoal from excavations at Jericho, producing the first university thesis on this subject (Western 1971). Some years later, Jean-Louis Vernet of the University Montpellier 2 in France developed a systematic approach to the sampling and analysis of charcoal in both archaeological contexts and other deposits with the aim of reconstructing past vegetation communities. An important component of Vernet’s work was the emphasis on awareness of sample context and the need for systematic charcoal collection. He went on to establish a research unit devoted to the study of bioarchaeology and human ecology, and introduced the term anthracology to describe the discipline. Anthracologists may have a background in archaeology, though they often work closely with palaeobotanical scientists trained in ecology (Vernet 2002), and the basic methods of this practice have been outlined in a handbook (Chabal et al. 1999). Today, there are a growing number of wood charcoal specialists in Europe who call themselves anthracologists, and graduates of the Montpellier school continue to make important methodological and substantive contributions to the discipline (e.g., Chabal 1990, 1992; Dotte-Sarout 2010a; Dufraisse and Martínez 2011; Scheel-Ybert 1998; Théry-Parisot, Chabal, and Costamagno 2010; Théry-Parisot, Chabal, and Chrzavzez 2010; Théry-Parisot et al. 2010). An influential paper on flotation (Struever 1968) and subsequent development of large-scale mechanical flotation equipment in the 1970s led to substantial increases in the quantity of charcoal recovered during archaeological excavation, and the use of reflected light microscopy greatly increased the speed with which these materials could be analysed (Leney and Casteel 1975; Pearsall 2000). Minnis and Ford (1977), for example, produced an early study of prehistoric human-induced landscape change in the American southwest using wood charcoal analysis. Jones (1941) and later Ford (1979), key figures in the development of palaeoethnobotany as a discipline, recommended the use of multiple lines of evidence in the study of vegetation histories, suggesting that wood charcoal must be combined with other techniques such as pollen analysis to provide an informed picture of past vegetation. Ford also supported Godwin and Tansley’s (1941) assertion that fuelwood

114 assemblages are subject to cultural patterning, and that charcoal from domestic contexts can inform on aspects of culture such as social status. Both observations were important and continue to play a role in structuring analyses today. In addition to interpretive study by palaeoethnobotanists, wood charcoal analysis is also used in environmental archaeology (Dincauze 2000, Reitz and Shackley 2012), a discipline that encompasses many different sub-fields including geomorphology, climate, sediments, soils, and faunal analysis. While archaeological wood charcoal analysis is more widely practised today, researchers agree there are methodological and interpretive challenges that must be addressed before the discipline can move forward (session discussions at the 5th International Meeting of Charcoal Analysis, Valencia, September 5–9, 2011). These challenges are due, in part, to the need to better understand charcoal taphonomy. Another concern is that until recently, methods have been developed by researchers studying materials from temperate and Mediterranean climates. This situation is changing, and in recent years several research projects have been focused on the analysis of wood charcoal assemblages from locations such as New Caledonia, Hawaii, Brazil, and central Africa (Coil 2004; Dotte-Sarout 2010a; Hubau et al. 2012; Scheel-Ybert 1998). These researchers concur there is a need to develop specialised protocols for studying charcoal assemblages in species-rich, tropical settings and much work remains to be done to build reference collections that will support further research in these areas.

Charcoal studies in Polynesia While plant materials from archaeological contexts are frequently evaluated by researchers throughout the Americas, Europe, and the Mediterranean, recovery and analysis of archaeobotanical materials and wood charcoal in particular are far less frequent in Oceania. In Polynesia, archaeobotanical research programmes are limited and it follows that analyses of wood charcoal assemblages are not numerous, even though human impact on island vegetation and food-production practises are major research themes (see Kirch and Kahn 2007). Several key figures and studies of Pacific wood charcoal are summarised in the following review, which is largely focused on Polynesia. Researchers Gail Murakami and Heidi Lennstrom produced some of the first wood charcoal studies in the region, and they discussed the methodological challenges involved in identifying and analysing Pacific assemblages (e.g., Graves and Murakami 1993; Lennstrom 1995; Lennstrom and Murakami 2003; Murakami 1983, 1989). In Hawaii, wood charcoal

115 studies have provided valuable information about Polynesian landscapes that have been highly modified by human activity and feral grazing animals (e.g., Allen and Murakami 1999; Dye 2010; Meeker and Murakami 1995; and unpublished reports on Kaho’olawe summarised in Allen 1984). The authors of one study have also examined the social and ritual uses of woods in Hawaiian prehistory (Kolb and Murakami 1994), and several others have informed briefly on prehistoric arboricultural practises including the cultivation of breadfruit and mountain apple (Syzygium malaccense) in the archipelago (e.g., McCoy, Graves, and Murakami 2010; Weisler and Murakami 1991). Coil (2004) combined a study of wood charcoal with phytolith analysis to consider pre-contact landscape modification, vegetation change, and agricultural practises at Kahikinui, Maui and determined that a broad spectrum of economic activities took place in this somewhat marginal and arid landscape. Several researchers have published notable wood charcoal studies on material from other Polynesian archipelagos. An important methodological contribution was made to the discipline when Allen and Wallace (2007; also see McFadgen 1982) demonstrated that inbuilt age was a significant concern for radiocarbon dating charcoal from tropical Polynesian contexts, findings that are presently the subject of continuing refinement (Allen and Huebert 2014; Rieth and Athens 2013). Catherine Orliac has made significant contributions to the study of the prehistoric vegetation of Easter Island by identifying numerous trees and shrubs that were not previously known for this location (e.g., C. Orliac 2000; Orliac and Orliac 2008). Orliac has also conducted wood charcoal research at sites on Tahiti and performed experimental archaeology on a series of earth ovens (e.g., Orliac and Orliac 1980; C. Orliac 1987; M. Orliac 1997; Orliac and Wattez 1989). Charcoal from sites in several valleys on Mo’orea, Society Islands was also examined as part of doctoral research project on prehistoric intensification and the development of agronomic systems in the Society Islands (Lepofsky 1994; Lepofsky and Kahn 2011; Lepofsky, Kirch, and Lertzman 1996). In this study, Lepofsky concluded that severe erosional events occurred when landscape was cleared for cultivation, and diverse native forests were later replaced by an impoverished flora. Finally, there have been several studies of charcoal from sites on Nuku Hiva in the Marquesas Islands. In a pilot study of material from Hatiheu valley (Millerstrom and Coil 2008), it was suggested that arboriculture may have been an important practice there in prehistory. It has also been determined that Marquesan earth oven contexts tend to contain a restricted number of taxa (Huebert 2008; Huebert, Allen, and Wallace 2010). House post charcoal has also been

116 examined from ‘Opunohu Valley with the aim of understanding how various timber species were used in different social contexts (Kahn and Coil 2006). In New Zealand, wood charcoal identifications have been performed for several decades by Rod Wallace. Though these studies have often been focused on waterlogged wood assemblages (e.g., Wallace 1985, 1989, 2000; Wallace and Irwin 2004), charcoal analysis has been used to identify short-lived species for radiocarbon dating and to reassess chronologies defined in prior research (e.g., Wallace and Green 2012). In addition, hundreds of charcoal assemblages have been identified for heritage archaeology projects (Wallace, pers. comm. 2009). Several publications on the prehistoric vegetation at archaeological sites have also been produced as a result of Wallace’s charcoal identifications (e.g., Allen 2005; Boyd et al. 1996; Wallace 1981). In western Oceania, an anthracological research programme was conducted at several sites on New Caledonia (Dotte-Sarout 2010a, 2010b; Dotte-Sarout et al. 2013) and in this study, Dotte-Sarout observed that charcoal assemblages were dominated by anthropogenic taxa. She concluded that deforestation occurred later in the prehistoric sequence than previously thought, and called attention to the challenges of working in an area that has a rich but poorly documented flora. In other studies of sites on the atolls of Kiribati, Di Piazza (1998, 1999) concluded that the charcoal within several old earth ovens included a narrow range of wood taxa and a large volume of Pandanus keys, which were probably the by-product of food processing activities. Finally, in a combined study of pollen and wood charcoal from the Micronesian high island of Kosrae, Athens and colleagues (Athens, Ward, and Murakami 1996) found evidence that native forests were transformed rapidly into agroforests upon colonisation, and this system appears to have persisted throughout the prehistoric sequence.

Formation processes and taphonomy Studies of site formation and charcoal taphonomy consider the complex combination of cultural and natural processes that form and act upon an archaeological deposit (Schiffer 1987, Figure 5.5). The challenges of understanding past human-plant interactions through the study of archaeobotanical remains are commonly acknowledged and to cope with these challenges, practitioners have developed methods to better understand the processes that have acted on plant remains in archaeological contexts (e.g., Chabal et al. 1999; Lee 2012; Pearsall 2000, 244–5; Théry-Parisot, Chabal, and Chrzavzez 2010; Wright 2010). Two main

117 concerns are expressed in this literature: cultural selection and use of various plants and plant parts that are sometimes referred to as cultural transformations, and studies of plant taphonomy and site formation that are considered natural transformations. Some transformations, which anthracologists refer to as filters, are more readily studied than others as the following discussion focused on wood charcoal will illustrate.

Figure 5.5: Diagram of taphonomic process of macroscopic plant remains (from Lee 2012, Fig. 2).

Physical processes Many different formation processes can affect charcoal deposits. Contexts such as charcoal kilns and closed earth ovens can provide ideal conditions for the formation of charcoal but natural processes, including sedimentation regimes and insect or rodent activity, can turbate these deposits and abrade charcoal. Vegetation succession cycles, including processes of re-growth and decay, can affect botanical assemblages thorough turbation by growing tree roots or fungi. These processes are complex and multi-layered, and it has been suggested that experimental (i.e., middle-range) studies, ethnoarchaeology, and geoarchaeology can help us better understand how these processes may have affected

118 archaeological assemblages (Schiffer 1987). Experimental studies on aspects of site formation can assist the researcher to better interpret an assemblage, and while palaeoethnobotanists have recently indicated this research is an avenue that will significantly advance the discipline, they also admit that much work is yet to be done (Wright 2010). It is not surprising, then, that there have been few such experiments involving wood charcoal (but see Goldstein and Shimada 2013; Ludemann 2008; C. Orliac 1987; Orliac and Wattez 1989; Thoms 2008). One research theme regarding formation processes and wood charcoal is the influence of conditions within the combustion environment. In a series of experiments with Polynesian earth ovens, it was concluded that oven temperatures varied based on how many times a pit was re-used, what type and arrangement of oven stones were used, and the species of fuelwood burned (C. Orliac 1987; Orliac and Wattez 1989). Findings of this research led Orliac to comment that the interplay of parameters within this type of combustion environment is complex and production of charcoal can vary. In other studies, charcoal morphology and reflectance measurements have been used to determine whether wood was exposed to high temperatures (Braadbaart and Poole 2008). While such measures may assist in determining what type of fire the wood was burned in, other researchers have disagreed there is any link between these values (McParland et al. 2010). Wood charcoal is more likely to preserve than any other type of plant remain in archaeological sites but the question of what preserves, and why, deserves some attention. Preservation might be influenced by the physical properties of wood, the burning conditions it was subject to, characteristics of the sedimentary matrix, disturbance regimes such as cycles of wetting and drying, and the age of the deposit. Some woods are less affected by post-depositional weathering, and these will be more likely to survive. However, the relevant characteristics—be they species type, wood form, surface area, wood age, degree of moisture at the time of burning, or some combination of the above—have yet to be identified. The importance of understanding the taphonomy of archaeological charcoal has long been recognised, but until recently there have been few advances in this area. While research into the natural processes that influence archaeological charcoal assemblages are currently in an incipient state, progress is being made (e.g., Théry-Parisot, Chabal, and Costamagno 2010; Théry-Parisot, Chabal, and Chrzavzez 2010; Théry-Parisot et al. 2010), with some efforts being made to cooperate with other disciplines where charred fossil wood is studied (Scott and Damblon 2010).

119

Post-depositional processes are one of the major factors that can affect the survival of charred plant materials. Some sedimentary contexts do not favour long-term preservation, and may be subject to varying levels of pressure, wetting and drying, temperature, or abrasion which can render materials too small for identification. Processes such as these may vary across a single site, and an understanding of the sedimentary history of the area is needed to best interpret results. Concerns regarding the depositional environment and its varying effects on charcoal have been long recognised, but they have only recently subject to rigorous study. Different burning temperatures and sedimentary matrices can destabilise the molecular structure of charcoal. Braadbaart and colleagues (Braadbaart, Poole, and van Brussel 2009), for example, have demonstrated that charcoal buried in highly alkaline soil tends to fragment into very small pieces, though if carbonised at higher temperatures (~500 deg C) it is sturdier within these matrices. Salt impregnation can have a similar effect (Lancelotti et al. 2010). Some researchers admit that post-depositional preservation is a very important but difficult parameter to evaluate (Asouti 2002, 245). These factors can be assessed, at least to some extent, through controlled experimentation and most experiments have thus been focused on understanding charcoal formation and fragmentation in isolated laboratory conditions (e.g., Rossen and Olson 1985; Scott 2010). In one experiment (Théry-Parisot, Chabal, and Chrzavzez 2010), it was noted that temperature of burning had some relationship to charcoal density loss, but overall the results were variable and inconclusive and the researchers noted there was a need to better simulate conditions of archaeological contexts in future studies (discussions at 5th International Meeting of Charcoal Analysis, Valencia, September 5–9, 2011). Recently, Chrzavzez and colleagues (2014) reported on a number of charring and fragmentation experiments with several important results. They noted that the force of compression and charring temperature were not correlated with rates of charcoal fragmentation and these values were significantly different for each taxon, and while rates of fragmentation were more variable for small pieces, some taxa displayed very different reactions to compression. It was also noted that the state of preservation of many types of paleobotanical materials including charcoal are very unevenly reported in the literature, which may be due, in part, to the absence of a structured method of classification. Schemes such as these would be useful to guide recording the condition of archaeobotanical materials in a structured and quantifiable way, and these data could then be evaluated to consider many aspects of charcoal taphonomy. They are, however, poorly developed for many paleobotanical

120 materials. For example, there have been recent efforts to classify starch grains, but starch diagenesis is still not well understood (e.g., Collins and Copeland 2011 and related criticism). There have also been attempts to devise classification schemes for waterlogged macro-remains by scoring rates of fragmentation and erosion (e.g., Jones, Tinsley, and Brunning 2007, Table 3). Data such as these are only occasionally reported for wood charcoal assemblages, usually as general comments regarding the assemblage (e.g., Carrion 2006; M. Orliac 1997; Orliac and Orliac 2008). The more established schemes used by palynologists were briefly reviewed to demonstrate what could be achieved if these data were captured systematically for wood charcoal assemblages, as schemes to classify pollen grains have been well-defined. Though there appear to be variations in practice, at a minimum such studies usually include discussion of the condition and percentage of degraded pollen grain in assemblages (e.g., Parkes 1994, 133–4). Some researchers go so far as to include precise guidelines for the categorisation of indeterminable grains by deterioration type, using categories such as corroded, degraded, damaged, and agent of concealment, and link these to processes responsible for their deterioration (Delcourt and Delcourt 1980, Table 1 is a thorough guide to assessing multiple types of damage; Twiddle and Bunting 2010 for current practises). In regards to archaeological assemblages, palynologists have also introduced schemes to evaluate post-depositional biases through consideration of a sophisticated range of assemblage properties (e.g., Bunting and Tipping 2000). Process of site formation and taphonomy are also directly influenced by wood type. It is well known that certain species produce more and/or sturdier charcoal than others, but causal factors are not yet well understood (e.g., Connor and Viljoen 1997). Several important concerns include the burning qualities and rates of charring, the volume of charcoal produced, and the rate of charcoal fragmentation. Density has sometimes been cited as an important factor affecting the burning quality of wood (U.S. National Research Council 1980), and low-density wood is said to be likely to ash more quickly than denser woods. Though some researchers have found density to be directly related to rate of charring in certain tropical woods (Njankouo, Dotreppe, and Franssen 2004), others have argued that dryness and exposed surface area, not wood density, are the primary determinants of the speed and temperature at which certain woods will burn, regardless of species (Chabal et al. 1999, 49). Materials science researchers have suggested that oxygen permeability rates may be a more suitable parameter to consider rates of charring, especially regarding some tropical woods (Hugi et al. 2007). The chemical constituents of wood may

121 also influence charcoal production (Lingens, Windeisen, and Wegener 2005), and extractives such as gums, oils, and other organic and inorganic compounds (Panshin and De Zeeuw 1980) may significantly affect the burning qualities of wood, however outside of the commercial timber industry these items are not well understood. In addition to burning rates, the relationship between wood density and volume of charcoal produced is also not straightforward. Rossen and Olson (1985) observed that density values are not always correlated with the reduction in charcoal weight or volume after charring. In a large (almost 300,000 fragments) experimental study, it was also found that charcoal density loss varied by wood type and density did appear to have some relationship to fragment size, but as previously discussed the conditions of combustion were also a factor (Chrzavzez et al. 2011). Issues of variability in rates of fragmentation and mass loss between different taxa have been major topics of discussion by anthracologists, especially in regards to the over- or under-representation of certain species. Some anthracologists have argued at length that all taxa fragment at the same rate and in the same distribution of fragment size classes, and slight differences can be corrected by following a rigorous sampling protocol and analysing a large quantity of material (e.g., Chabal 1990; Chabal et al. 1999). Chabal analysed a large quantity of samples from archaeological sites in southern France, and after performing a range of statistical tests concluded that fragmentation was random for all taxa in both charcoal fragment size class and fragment quantity. The results of these tests demonstrated that, for wood charcoal, processes of mass reduction (i.e., how much material remains after burning) and rates of fragmentation (i.e., how many pieces remain) occurred independently (Figure 5.6). While this result may be due to the inclusion of outliers, which can make the numbers appear to be more random than they actually are, Chabal argues that without an absolute threshold to define lower or upper limits, no data can be safely rejected. Other researchers present findings in contrast to this conclusion, though some have smaller datasets or deal with very small fragment sizes (e.g., Chrzavzez et al. 2011; Jansen and Nelle 2012, 11; Lancelotti et al. 2010; Willcox 1974, 124; Zalucha 1982, 33–36). Lastly, it was noted that most experimental studies have taken place on woods from temperate climates, and few tropical woods have been the subject of taphonomic studies. Lancelotti and colleagues (2010) have demonstrated that charcoal from tropical dry environments in South Asia tended to create much smaller fragments than assemblages from temperate regions, and one species common to their archaeological assemblages fragmented

122

Figure 5.6: Charcoal fragmentation and mass reduction: (a) species A provides more fragments than B with less mass reduction, (b) species A provides more fragments than B with greater reduction in mass (from Chabal 1992, Fig. 11).

at a notably higher rate than others. In Pacific contexts, it has been observed that low- density tropical woods turn to ash at much lower temperatures than higher density woods in earth ovens (Orliac and Wattez 1989), and it has been demonstrated that high count-to- weight correlations can obscure actual taxonomic differences in fragmentation rates, especially when monocotyledon endocarp fragments such as coconut and Pandanus are part of the assemblage (Di Piazza 1998, 152). In experimental oven studies, C. Orliac (1987) observed that charcoal production varied by wood type; while some woods were reduced entirely to ash (such as Hibiscus tiliaceus), others (Casuarina, Neonauclea) produced a

123 large quantity of charcoal, and still others (such as coconut wood) produced a lesser quantity of charcoal but the resulting fragments were large.

Cultural and natural processes An important cultural consideration regarding formation processes is that of modelling the availability, selection, and use of plant materials by past people. Cultural transformations can be challenging to model, as assumptions must be made about human behaviour and the natural environment in the distant past. These concerns have long been considered by palaeoethnobotanists studying seeds (e.g., Pearsall 2000, 245–6), but have been less often addressed by wood charcoal analysts (but see Asouti 2002, 71–91), and it has been said that anthracologists of the Montpellier School have gone so far as to consider culture processes problematic noise in vegetation reconstructions (Marston 2012). Until recently, little of this specialist literature was focused on studying human activities such as fuelwood collecting or forest management strategies, but the situation is changing. There have recently been calls to recognise that fuelwood is selected in some socio-economic context (Dufraisse 2011), and ethno-archaeological and experimental studies can provide analogous information on plant use that can assist in understanding patterning of wood charcoal data (see Théry-Parisot, Chabal, and Chrzavzez 2010; Théry-Parisot and Henry 2012). Because much archaeological wood charcoal derives from fuel, fuelwood selectivity is a common research theme (e.g., Asouti 2003; Asouti and Austin 2005; Froyd et al. 2010; Gelabert, Asouti, and Martí 2011; Johannessen and Hastorf 1990; Lepofsky, Lyons, and Moss 2003, but examples are numerous). Fuelwood selection involves a combination of cultural and ecological factors, and these processes may operate concurrently at different levels. Practical factors that are frequently cited include selection of hard and long-burning wood, and low smoke production (Smart and Hoffman 1988, 168–9). But it has been suggested that fuelwood was usually collected using the least effort required (referred to as the ‘principle of least effort’), unless preferred materials were abundant and readily procured (Shackleton and Prins 1992), and the anthracologists argue that selection of preferred wood species is a bias over-emphasised by the researcher when in many situations, any dry wood would have sufficed (Chabal et al. 1999, 53–5; Théry-Parisot, Chabal, and Chrzavzez 2010). Other influential factors include the abundance of dry deadwood in an area or a species’ predilection to shed branches naturally (self-pruning)

124

(Ford 1979, 305). The ability of some trees and shrubs to re-colonise disturbed habitats after land clearance, or the speed of their re-growth, may also be a factor in wood exploitation as these can be seen as readily renewable resources (Minnis and Ford 1977). Moreover, the use or avoidance of certain woods may be linked to cultural perceptions the trees or restrictions on use of woods by social status (Kolb and Murakami 1994). The preferences of a group based on cultural identify, including the interplay of group values, fuel needs, and types of available technology, may also operate on selection. The use of ethnographic analogy and archaeological data to model fuelwood acquisition and use has been effectively demonstrated in a study of Andean fuel use over a very long period of time (Johannessen and Hastorf 1990). In this study, it was found that trees cultivated especially for use as fuel were also leveraged as symbols of prosperity, and they were carried long distances in elaborately carved forms to be burned in ceremonial fires. It is evident that much research in the discipline is centred around fuelwood selectivity, collection, and management. Other activities may have introduced wood and other plant parts into charcoal assemblages when they had a secondary use as fuel, which would have been an expedient method of rubbish disposal. These practises could include woodcarving, construction, food preparation, or stripping bark to make cloth or cordage. Branches and bark may also have been collected to prepare extracts for medicinal preparations, dyes, or to remove leaves for thatching. The procurement of wood for construction is also a practice that can introduce wood into a charcoal assemblage, as burned structures and structural elements such as posts are sometimes found in archaeological sites. As previously discussed, charred housepost remains have been used to study variations in timber species used in different social contexts in the Society Islands (Kahn and Coil 2006). The procurement of construction timbers at a location in central Anatolia has also been studied using optimal foraging theories (Marston 2009, 2012), in which resource availability, value, and handling costs were linked with selection of certain timber species. Studies such as these are fewer in number than those regarding fuelwood, but they have demonstrated that when wood type is considered a raw material (similar to stone, bone, or shell used in manufacturing), it can provide important information on resource procurement and social relations in prehistory. Finally, vegetation clearance and burning regimes are also cultural processes that can be studied through the evaluation of wood charcoal assemblages. These activities are evident as burn layers or high concentrations of charcoal in sediment. Deposits such as these are frequently encountered in archaeological excavations and can inform, for example, on

125 native vegetation that was present in newly occupied environments, swiddening or hunting activities, and routine brush clearance. Materials in burn layers are usually evaluated as part of the site history. They are less often the subject of discrete wood charcoal studies, which have been informative on past anthropogenic burning regimes and climate change (e.g., Hubau et al. 2013).

Recovery of charcoal Procedures to recover, process, quantify, and analyse all types of macro-remains have been well developed by the palaeoethnobotanists, and methods designed to recover all types and size classes of material are generally considered adequate to address most research questions (Pearsall 2000, 244–5). Anthracologists have meticulously formulated guidelines for to the recovery and analysis of wood charcoal (see Chabal et al. 1999), and these methods involve systematic flotation schemes, recovery of very small charcoal fragments (as small as 1 mm), and sorting fragments into various size classes. Recently, anthracological procedures have come under scrutiny for being overly formulaic and labour- intensive, and tailored to the assemblages and research questions (i.e., vegetation reconstructions in temperate and Mediterranean climates) of the developers. It has, for example, been demonstrated that some woods may be over-represented in smaller fragment size classes indicating that these materials, which are also more challenging and time- consuming to handle and identify, and these data can skew quantitative results (e.g., Chrzavzez et al. 2011). Anthracologists working with tropical charcoal have commented that these formulae need to be reconsidered for their assemblages (Dotte-Sarout 2010a; Scheel-Ybert 2002), and it has also been suggested that further refinement of the methods could expedite analysis, making wood charcoal analysis feasible in situations where it was previously considered too large an undertaking (Veal 2013). In consideration of the aforementioned concerns, both anthracological methods and palaeoethnobotanical guidelines were evaluated and, as discussed below, several important methodological concerns regarding the recovery and sampling of charcoal were identified. Biases introduced by the researcher during recovery and analysis of materials, sometimes referred to as analytic filters, are also important considerations when interpreting wood charcoal assemblages (Figure 5.5). Macrobotanical field sampling strategies should be based on both the research aims and an understanding of the nature of the archaeological deposit (Jones 1991; Lennstrom and Hastorf 1995; Pearsall 2000, 66–76), and once the

126 research question has been determined, there are three main concerns for the recovery of charcoal: how to choose a sampling protocol, where to take samples, and how many samples to take. Each was considered in turn. Methods to recover charcoal can involve small- or large-scale flotation programmes or rely on hand-picking charcoal from screens. In some situations, charcoal may also be recovered from in situ deposits or bulk sediment samples may be transported to the laboratory for processing. Each method can recover differing amounts of charcoal that might fragment it in different ways. For example, it has been demonstrated that charcoal recovered by water flotation can be excessively fragmented, resulting in smaller sample sizes, and the rates of loss can be variable (Greenlee 1992). Also, the longer it takes to process a float sample (as with clayey sediments) the greater the potential for loss or damage. Different plant parts are variously affected by flotation schemes as well (Wright 2005). Ideally, recovery methods should be consistent, but they may vary due to differences in sedimentary characteristics within a site or other factors beyond the researcher’s control. If more than one recovery method is used, it is important to understand how each has influenced fragmentation rates before attempting quantitative comparisons. A good, general botanical sampling strategy is that of blanket sampling, or collecting material from all units, levels, and features (Pearsall 2000, 66–76). There are, however, situations where this type of sampling is not possible due to time or budget constraints or coordination with other research aims. In these situations, Pearsall notes that a more conservative method is to sample material from a predetermined percentage of contexts. Not all functional contexts should be sampled in the same way, as is recommended for sedimentary analysis in general (Stein 1985, Fig. 1). Awareness of context type is essential to the development of appropriate archaeobotanical field sampling strategies, and the importance of separating results by functional context has long been recognised by archaeobotanists studying wood charcoal assemblages (e.g., Asouti and Austin 2005; Asouti 2002, 244–5; Chabal et al. 1999; Jansen, Mischka, and Nelle 2013; Marston 2012, but there are many). Researchers stress that interpretations are most accurately made by comparing material from similar functional contexts to control for human selectivity and variation in formation processes. Though this concern is often expressed regarding charcoal from kilns, smelting fires, and burned structural material, others have commented that in any situation, understanding the conditions in which charcoal was formed is crucial for correct quantitative interpretation of the archaeobotanical assemblage (Lancelotti et al. 2010). These principles are important for several reasons that have been most clearly described by

127 anthracologists (Figure 5.7). First, samples of dispersed charcoal from occupation debris, and other cultural deposits where charcoal has accumulated for long periods of time, have been shown to inform broadly on the past local vegetation and these materials are considered most suited to vegetation reconstructions (with some caveats, see Théry-Parisot, Chabal, and Chrzavzez 2010). Second, large and concentrated charcoal deposits from burned structures are likely to represent highly selected materials that may have been transported from long distances, and are generally agreed to be a very biased representation of past vegetation. Thirdly, charcoal in domestic heating and cooking contexts can be a very biased representation of local vegetation depending on feature function. For example, wood charcoal collected in situ from earth oven features has been shown to contain a restricted number of dense hardwood species that meet the demanding fuel necessary to heat oven stones (Huebert, Allen, and Wallace 2010), and M. Orliac (1997) has shared a similar observation when discussing an assemblage from the Papeno’o Valley, Tahiti, noting that charcoal from earth ovens was more useful to interpret feature function than to study vegetation change. One important relationship to consider in sampling is that of the sample and the target population (Lee 2012; Smart and Hoffman 1988, 180–85; van der Veen and Fieller 1982; Orton 2000, 41 for broad concerns about archaeological sampling). Samples of the same population are not independent (see Popper 1988, 61), and there are several considerations for both sampling and data analysis. Lennstrom (1995) analysed wood charcoal from feature and non-feature contexts taken from sites in Hawaii to evaluate relationships between samples and their relation to the feature or unit, and their relationship to the site or catchment being studied. Her aim was to characterise gaps between samples and the various target populations, and she concluded that no one sample was representative of the whole at any scale, and different samples from the same population (i.e., one isolated feature) could be remarkably different in both taxonomic richness and the relative abundance of taxa. These findings indicate that when available, multiple samples from a given context should be examined and data combined for analysis. A sampling strategy must also produce enough charcoal to address the research aims. It is also often necessary to establish an upper limit if transportation costs are a factor, as many sites can produce large volumes of charcoal. There is no agreed-upon formula to perform these calculations, and researchers have noted that it depends on the questions that are being asked. Studies of discrete deposits, such as house posts (Kahn and Coil 2006;

128

Figure 5.7 Charcoal deposits, archaeological contexts and vegetation representativeness (adapted from Chabal et al. 1999, Fig. 4).

Wallace 2000), can be based on identification of very little material. In a study of fuel use over a long span of time, where many samples were available, as few as twenty fragments from each sample were identified (Johannessen and Hastorf 1990). In studies that aim to reconstruct vegetation histories, however, a minimum of several hundred fragments per stratigraphic layer has been recommended by anthracologists (Chabal et al. 1999; Keepax 1988, 120–124; Scheel-Ybert 2002, 12–13). The question of how much to sample is linked to the question of how much needs to be identified. For vegetation reconstructions, anthracologists have developed precise values for assemblages from temperate and Mediterranean areas where 150–250 fragments at a minimum are recommended and 400–500 are said to be ideal (Chabal 1992; Chabal et al. 1999; Théry-Parisot, Chabal, and Chrzavzez 2010). Recommendations for other world

129 regions depend on the richness of the local flora, with values for some species-rich tropical areas being slightly higher. In an anthracological study of the neotropics it was demonstrated that counts of 200 fragments, or more preferably 300–400, were ideal (Scheel- Ybert 2002), and in a study of material from New Caledonia, it was determined that samples of around 100 fragments produced acceptable results but samples of 400 fragments were ideal (Dotte-Sarout 2010a, 138–41, Figs. 38–40). It has, however, been recently suggested that the minimum number of fragments to be identified per stratum may be much less for some study areas (e.g., O’Carroll and Mitchell 2012; Veal 2013).

Quantitative analysis Quantifying plant remains, a task that seems straightforward for some archaeological materials, can be challenging for the archaeobotanist as decisions must be made regarding which materials to count. While the task is more involved for those studying seeds and other small plant parts where decisions about what to count are involved (Wright 2010, 51–2), wood can be more straightforward as the basic unit of measurement is fragment count. Some analysts also record the total weight of fragments per taxon and measure fragment size classes. The two main concerns regarding quantifying wood charcoal are whether to use counts or weights as basic units of measurement, and how to compare and contrast quantitative results. Other properties of charcoal could be quantified but these data are seldom collected, as will be discussed at the end of this review. As previously discussed, mass reduction and rates of fragmentation can be variable per taxon. Also, differences in site formation processes also affect the amount and quality of material that survives in a deposit and the ease with which this material can be recovered and identified. It follows that finding the appropriate units, either counts or weights, for quantifying charcoal is a basic task that is continually re-evaluated by charcoal analysts (e.g., Asouti and Austin 2005; Chabal 1990; Coil 2004; Di Piazza 1998; Miller 1985, examples are numerous). I have demonstrated that a basic correlation test between counts and weights of the most frequently occurring taxa in an assemblage can assist in determining which measure will introduce the least bias into further quantitative comparisons (Huebert 2008, 2009). Absolute values such as count and weight are overly influenced by preservation and recovery biases, and are seldom directly used to analyse human-plant relationships (Popper 1988, 60). Instead, quantities are usually evaluated at a low level. Simple presence / absence

130 analysis can be quickly performed by creating a matrix of all taxa and all sites or samples in a study and marking the presence of a given taxon with a tick. This is a very basic method of reporting results for low-level comparisons, useful for quick evaluation of taxonomic richness and cross-comparisons of large data sets. Two other useful low-level quantitative methods used are examination of the relative percentage of taxa within samples and the ubiquity of taxa throughout the assemblage. These two measurements are complimentary and provide different views of the data. Relative percentage values are interdependent, meaning the abundance of one taxon is affected by that of others within a sample. Values are calculated by dividing the sum of counts (or weights) of a given taxon and dividing by the total amount of material in a given sample or stratum, and are displayed as numeric values, pie charts, or bar graphs. These values have been shown to even out variation in density between samples, and effectively inform on the changing use of plant materials over time (Pearsall 2000, 196–9). Ubiquity values are also used by palaeoethnobotanists to study many different types of plant remains and they provide a somewhat more sophisticated method to evaluate taxonomic representation, one that makes low-level quantitative assumptions (Popper 1988, 60–64). Ubiquity values are generated by calculating the frequency of a taxon across analysis units, and dividing the total number of occurrences by the number of proveniences. This type of analysis is most appropriate for reviewing overall trends within each taxon (Hubbard 1980, 53), and it has been employed by wood charcoal analysts (e.g., Asouti and Hather 2001, 25; Smart and Hoffman 1988). It is important to note that ubiquity scores should be used to compare relative importance of the same taxon from samples at a given site, but not to compare and contrast the importance of different taxa (Popper 1988, 60–1). While low-level data analysis provides a means to examine data without regard to quantity of material identified, it is still important to consider that the same preservation and sampling biases that affect other statistical analyses also affect these measures (Kadane 1988, 207; Wright 2010, 52). The interpretation of these values are not straightforward, as they do not relate directly to the amount of wood burned nor to the amounts present in past vegetation communities and they are not comparable between sites (Godwin and Tansley 1941; Smart and Hoffman 1988, 190–2), though some analysts do assume that proportions approximate abundance to an extent (Miller 1985). The issue of whether these measures accurately reflect the parameters of interest involves consideration of fuelwood collection habits in the past, cultural activities that may have taken place, and the duration of occupation. Analysts may consider using very low level comparisons, such as relegating

131 discussion to the most frequently occurring and abundant taxa at a site, if there is cause for concern. Other quantification methods can be used to examine wood charcoal datasets. The creation of rank-ordered taxa lists are a useful way to compare the frequencies of different taxa or plant parts. This method, patterned after statistical ranking methodologies, uses an ordinal scale to arrange results into rank-ordered lists that utilise separate scales defined by the researcher (Popper 1988, 64–66). While ranked taxa have the advantage of being independently evaluated, materials that preserve poorly will still be under-reported in this scheme. The density of charcoal per unit of sediment is also sometimes used by palaeoethnobotanists to consider factors such as intensity of occupation (e.g., Miller 1988, 73–4) or to evaluate different activity areas at a site (e.g., Balme and Beck 2002). When data sets support more involved statistical calculations, such as those from large areal excavations or from systematic excavations across a wide region, these measures can be used to assess differential preservation and comparability of materials from different areas (Lee 2012, with ideas from Orton 2000). Lastly, counts or weights are sometimes used to devise ratios and examine changes in the relative amounts of two items (Miller 1988, 75). These applications are usually used to consider the ratios of different plant parts at a site rather than wood charcoal alone (e.g., Johannessen 1988, 151). Lastly, it is important to note that quantitative comparisons that are typically used to assess changes over time or across spatial areas can be affected by variations in sample sizes. The correlation of samples size with relative abundances of taxa over time, as well as shifts in the number of taxa recovered (taxonomic richness), should be tested. If a weak or nonexistent correlation can be demonstrated, it is valid to evaluate shifts in the relative abundances or richness of taxa between contexts (i.e., strata, features, or sites) (Grayson 1984, 116–30, 131–67).

Evaluating sample adequacy, richness, and diversity The issue of how fully fossil charcoal assemblages can represent past vegetation communities has been addressed by anthracologists and palaeobotanists in some detail (e.g., Chabal et al. 1999; Figueiral and Mosbrugger 2000; Scheel-Ybert 2002). Anthracologists have argued that reliable results can be produced using a strict sampling protocol, and if the resulting assemblage is species-rich, results are reproducible between strata and comparable to palynological data, and they are of a richness similar to the current flora in the study

132 location, a good representation of the past flora can be achieved (Chabal 1992, 220–1). Several key evaluations are used to determine how well fossil assemblages meet these criteria. A taxon accumulation curve or species-area curve is useful to determine sample adequacy. This technique has been borrowed by archaeobiologists from quantitative ecology (Miksicek 1987). Using this method, the frequency with new taxa are encountered during analysis are plotted cumulatively in a simple line chart. Data is plotted per piece or per sample, depending on research aims. When the curve reaches a plateau, it is assumed that the population has been adequately sampled. Several factors can influence the shape of the curve (Lepofsky and Lertzman 2005). First, the sequence with which values are plotted can produce a curve that levels off too quickly if the richest are used first, and the data should be plotted in random order. Second, when more than one statistical population is encountered an otherwise stable curve may spike; for example, when a second ecological community is encountered. Lepofsky and Lertzman note that this method does also have other uses, including assessment of the richness and evenness of samples in a large assemblage. Anthracologists use taxon accumulation curves to assess sample adequacy, where they are referred to as saturation curves (e.g., Chabal et al. 1999; Henry et al. 2012). The method was first developed for use with charcoal assemblages from temperate and Mediterranean areas and later it was used to evaluate charcoal assemblages from the Brazilian tropics, where it was determined that the curves did not tend to reach a plateau due to high species richness and the wide dispersion of many species (Scheel-Ybert 1998, 62). A similar conclusion was also reached in studies of material from New Caledonia (Dotte-Sarout 2010a, 138–9). Diversity measurements summarise taxonomic richness and relative abundance of taxa as a numeric value (Magurran 2004, 72–99). These measures can be useful to consider general trends in archaeobotanical datasets, and can help in locating broad-scale patterns or contrasts between datasets (Popper 1988, 66–9). One method used by anthracologists is to compare the diversity of fossil assemblages to that of extant vegetation communities is through the use of the Gini Coefficient (e.g., Chabal 1992; Chabal et al. 1999, Fig. 13 Pareto curves; Scheel-Ybert 2002). The Gini Coefficient (GC) was originally developed by economists studying the inequality of income distribution. It has also been used by ecologists to study evenness in the dispersion of various plant and animal resources, or various attributes of an organism within a community, and it has been used to study forest

133 structure where it has been shown to be very insensitive to sample size compared to other types of diversity indices (e.g., Lexerød and Eid 2006). The GC is a single value that describes a degree of evenness, where one represents complete inequality and zero represents a very even dispersion. The higher this coefficient, the more uneven the distribution. Values are illustrated in graph called a Lorenz curve, using lines that lie in arcs beneath a diagonal of perfect equality (Figure 5.8). The Gini Coefficient is calculated by measuring the area between this line and the curve (Weiner and Solbrig 1984). To illustrate the use of this coefficient, in vegetation ecology, plots are considered to have low diversity when they have few species and uneven abundances (e.g., GC values are high). They have high diversity if there are many species, and abundances are more even (GC values are low).

Figure 5.8 Example of a Lorenz curve. The area between the curve and line of absolute equality is called the Gini Coefficient; it is a measure of inequality (from Weiner and Solbrig 1984, Fig. 3).

Anthracologists hold that under the ‘Gini-Lorenz law’ (also called Pareto principle, or the 80–20 rule) 80% of the individuals in a population represent 20% of the taxa. Values of 80:20 (expressed as a GC value of .80) are characteristic of temperate vegetation communities which are in a steady state (Poissonet 1968, 1979 in Scheel-Ybert 2002). These values have been shown to be lower in tropical areas (i.e., the higher floristic diversity the lower the value), on the order of 75:25 (Scheel-Ybert 2002 for South American tropics; Dotte-Sarout 2010a, 139–47 for New Caledonia). There are several factors to consider when interpreting Gini Coefficients. First, when sampling is adequate but values are higher than expected, the vegetation communities may be disturbed and thus species-

134 poor, or a cultural preference for certain woods may have introduced a significant bias. Second, values can be too low when sampling is inadequate, thus taxon accumulation curves should be plotted to determine whether a site has been adequately sampled.

Wood and charcoal identification Several different procedures are used to examine wood charcoal. The most common technique used today is the direct examination of freshly fractured surfaces with a high- powered microscope (Leney and Casteel 1975). In some situations, especially when fragments are fragile, charcoal is embedded in paraffin or resin and thin sections are cut for examination with a transmitted light microscope. Scanning Electron Microscopy (SEM) is also sometimes used to examine minute anatomical features at very high resolution. The usefulness of SEM technology in identifying wood charcoal is widely recognised (e.g., Dotte-Sarout 2010a; Orliac and Orliac 2008; Prior and Alvin 1983; Prior and Gasson 1993; Rossen and Olson 1985; Scheel-Ybert 1998), but because of the expense and setup time involved, applications can be limited. The increasing use of benchtop SEM devices are making such studies quicker and more cost-effective. While identification procedures are straightforward, it is usually not possible to identify all botanical materials in an archaeological assemblage (Hather 2000a, 11–12). Wood can be badly degraded by fungal and insect activity before it is burned, and in some cases identifications should be considered tentative because of limitations in materials, references, or experience (R. Wallace, pers. comm., Lepofsky 1994, 269). Several concerns and challenges in the study of tropical wood and charcoal were investigated to consider these limitations in more detail, including the modest number of published reference materials that were located, challenges of identifying diffuse-porous tropical woods, and concerns regarding features that are destroyed or obscured by charring.

Published reference material Charcoal analysts working in Europe, the Americas, and New Zealand benefit from a wide variety of well-established reference collections developed for forestry and commercial purposes, including digitised wood identification keys (e.g., Richter and Dallwitz 2000), wood atlases (e.g., Meylan and Butterfield 1978), and numerous photographs and descriptions of thin sections that can be obtained online (e.g., InsideWood

135

2004; Schoch et al. 2004; Wheeler 2011). The most detailed references contain SEM photographs that illustrate important diagnostic features (e.g., Meylan and Butterfield 1978). Some resources with a regional focus have been published specifically for an archaeological audience (e.g., Friedman 1978; Hather 2000b). Published references for woods of the tropical Pacific Islands are limited. They include SEM images of numerous Hawaiian woods photographed by the late Charles Lamoureux (1985), a text-based description of some Hawaiian woods (Brown 1922), a key for identifying woods used in traditional Hawaiian culture (Lamberton 1955), several anatomical drawings (Brown 1935) and a briefly illustrated volume with anatomical descriptions of 92 wood species found in French Polynesia today (Détienne, Jacquet, and Orliac 1999). Some publications illustrated by wood anatomists also contain useful descriptions or photographs of certain orders or families that are represented in this region (Carlquist 2001, and the IAWA Journal). A rapidly growing atlas of worldwide woods, InsideWood, includes some materials from the tropical Pacific and has the potential to become an important resource (InsideWood 2004; Wheeler 2011).

Challenges of identifying tropical woods Material from areas of high floristic diversity can be challenging to identify with precision even after extensive reference collections have been created (e.g., Dotte-Sarout 2010a; Hubau et al. 2012; Scheel-Ybert 2001). There are several challenges to the identification of tropical woods that have been acknowledged, and because many are diffuse-porous and lack distinct growth or annual rings they can be more challenging to identify than wood from temperate zones (also discussed by Lepofsky 1994, 269). Two relevant concerns regarding identifiability were investigated in more detail, including the absence of diagnostic characteristics and the often unknown (and potentially significant) range of anatomical variability in features that are commonly used to identify wood. Identification of tropical woods can be time consuming. While low-powered lenses can be used for the rapid initial sorting of ring-porous woods (Hoadley 1990; Western 1969 and R. Wallace, pers. comm.), tropical woods must be examined under high magnification and identification often begins with examination the axial arrangement of vessels and parenchyma, vessel diameter, and ray size. Identifications can become challenging when these features are not particularly distinctive. Sometimes very similar taxa can be distinguished by a feature that is infrequent and difficult to locate. A researcher studying

136

Cordia, for example, concluded that the most notable characteristic of the genus was the presence of certain types of crystals (Gottwald 1983), which I have noted do not preserve consistently in archaeological charcoal. A taxon may display wide anatomical variations that overlap with other genera, or have significant intrageneric variations that prohibit a species-level assignment. Wood anatomists have, for example, long understood that Hibiscus tiliaceus wood lacks agreement on some important characteristics typically used for identification (Webber 1934). Anatomical variations can occur between individual trees based on different conditions of the local environment, a particular concern for trees that have a wide biogeographic range. It has, for example, been demonstrated that wood samples from a single individual may not be representative of the range of variation within a population (e.g., Stern and Greene 1958), a topic that is discussed in forestry textbooks (Panshin and De Zeeuw 1980, 240–285). Tropical woods are seldom discussed in these resources, so several specific examples were reviewed. In a study of Hawaiian Metrosideros polymorpha wood (Sastrapradja and Lamoureux 1969), it was determined that only a small number of anatomical characteristics were constant throughout the genus, and they did not include vessels, rays, or axial parenchyma. It has further been found that variations in vessel diameters of Metrosideros can be related to differences in elevation (Fisher et al. 2007). Because anatomical variations such as these can relate to external influences, it is useful to understand the biogeographic range of a species. Some pantropical trees such as Dodonaea viscosa, for example, demonstrate little variation as a result of changes in altitude or latitude but display a wide variation in axial parenchyma arrangements in different geographic locations (Liu and Noshiro 2003). Findings such as these suggest that a comprehensive reference collection should include wood from several parts of a tree and several individuals from different habitats. Some researchers identifying archaeological material from tropical sites have assembled reference collections with these considerations in mind (e.g., Thompson 1994, collection included twigs and roots). The challenges of collecting and processing this quantity of material are a significant concern acknowledged by wood anatomists (e.g., Carlquist 2001, 3–5).

137

Features destroyed by charring Carbonised wood is more challenging to identify than fresh wood because some anatomical characteristics distort or are obliterated during pyrolysis, or they become difficult to distinguish without the use of SEM. The distortion or disappearance of important anatomical characteristics are some of the biggest challenges to identifying archaeological wood charcoal. It has been demonstrated that the temperature and duration of burning are direct influences affecting identifiability (Orvis, Lane, and Horn 2005), though other factors such as the moisture content of wood and amount of decay can also contribute. Experiments have also shown that some tropical woods fracture in shapes that do not facilitate exposure of the necessary three planes used in identification, or the resulting fragments can be very small (<2 mm) (Lancelotti et al. 2010). Many experiments have been conducted to study the changes that occur when wood turns into charcoal (see Scott 2010). These changes can vary for different species, and a few pertinent studies are summarised. An early notable study was conducted for an archaeological audience by Rossen and Olson (1985), who found that pore patterns, ray structure, and tyloses were generally well preserved in their test specimens (all of which were North American taxa), while most types of parenchyma were hardly identifiable after charring. In another study, it was found that oak can shrink over 25% tangentially, and somewhat less in other directions, and most of the features on the vessel walls were obliterated (McGinnes Jr., Kandeel, and Szopa 1971). Scheel-Ybert and colleagues (Gonçalves, Marcati, and Scheel-Ybert 2011), studying wood from tropical Brazilian species, found that overall anatomical structures in charred wood were closely related to those found in fresh wood, and while vessel diameters were reduced, ray widths were only occasionally smaller. And in a study of wood from African savannah trees, it was found that vessel diameters and ray cell sizes were reduced, sometimes by a large percentage, while axial parenchyma cells actually expanded (Prior and Gasson 1993). The latter study was of particular interest, as a species of Terminalia was tested, and it was noted that vessel diameters shrank up to 23%, and it had the most marked expansion of axial parenchyma cells in the study. In addition, some types of fibres can become markedly distorted by charring, inter-vessel pits—a feature common in anatomical keys—can be blocked by extractives such as gums, and crystal inclusions can be destroyed (Prior and Alvin 1983). Calcium oxalate crystals, for example, morph into calcium carbonate above 430 degrees Celsius and can then dissolve under alkaline conditions (Braadbaart, Poole, and Van Brussel 2009; Braadbaart et al. 2012).

138

While many of these changes do not prohibit the identification of charcoal, they do highlight a need to understand the specific changes that may occur during charcoalification. These concerns become important when it is understood that wood anatomy references rely on the measurements of key features such as vessel and cellular pit sizes, axial parenchyma patterns, and (to a lesser extent) presence of various crystals for identifications. A number of charcoal analysts have documented the usefulness of creating charred wood references for the study of archaeological material (Coil 2004; Di Piazza 1998, 150; Dotte-Sarout 2010a; Hubau et al. 2012; Johannessen and Hastorf 1990, 80; Pearsall 2000, 150).

Field and laboratory procedures used in this study In the final section of this chapter, I describe the field and laboratory methods used in the present study. These include creation of the reference collection, field methods that were used to collect charcoal, the sample selection rationale, and the identification and data collection procedures.

Creation of reference collection To create the main reference collection for this research, 105 specimens of wood and other plant parts were collected over the course of two field seasons on Nuku Hiva, and all materials were vouchered and verified by botanists (Rhys Gardner and Jean-Francois Butaud). Ten additional specimens of wood and other plant parts were collected from Mo’orea in the Society Islands, and Auckland markets (the latter included specimens of taro, breadfruit, and banana). The University of Auckland Anthropology Department also already housed a botanical reference collection of 200 wood, fruit, seeds, and various other plant parts from the tropical Pacific including other Marquesan islands, Fiji, Hawaii, Henderson, and the Cook, Austral, and Society Islands. These references have been assembled by Rod Wallace, with contributions from other researchers, over the past twenty years. Three collections in this archive were of particular use in the present study, all of which have been verified by botanists: specimens provided by Catherine Orliac of the Muséum national d’Histoire naturel, Paris, material collected by the palynologist Kevin Butler of Massey University, New Zealand on a prior field expedition with Allen, and woods collected by the botanist Jacques Florence during the Oxford Henderson Expedition (Stephen Waldren, pers. comm. 2012). In combination, these collections include multiple

139 individuals of approximately two dozen species, including five samples of breadfruit (Artocarpus altilis) wood. Many wood specimens from the archive had previously been sectioned, and 120 additional specimens were selected for study. These woods were sectioned, stained with safranin, and permanently mounted onto slides by a technician at Scion (New Zealand Forest Research Institute, Ltd.) in Rotorua. I performed the histological study at the University of Auckland. A photo reference of the anatomical features of each specimen was created, and this reference was archived in the Anthropology Department. Feature descriptions were entered into a relational database (Microsoft SQL Server; Figure 5.9) according to the IAWA (International Association of Wood Anatomists) list of anatomical features used for hardwood identification (Wheeler, Baas, and Gasson 1989), and these data are provided in Appendix B. A charcoal reference was also created from these materials. Specimens were sampled for charring when at least 5mm2 of material was available, and they were seasoned for approximately six months before the procedure. Specimens were wrapped tightly in several layers of aluminium foil and heated outside over a moderate flame until they ceased to smoke. In addition to wood and twig specimens, sections of taro, sweet potato, breadfruit, banana, Pandanus drupes, several types of nuts and nutshells, and sections of coconut petiole and inflorescence were also charred in this manner. Microphotographs were taken of all three planes of the reference charcoal at several different magnifications to create a digital reference. An SEM was also employed to examine a selection of charcoal specimens at very high magnifications. A comprehensive study of changes that occur in wood after charring was beyond the scope of this thesis, but to better understand the changes that may occur in charring some preliminary data on wood shrinkage was collected while creating a portion of the wood charcoal reference (Table 5.1). Taxa selected for this evaluation had a variety of vessel sizes and axial parenchyma arrangements. Length measurements were taken on radial and longitudinal planes before and after charring. It was found that woods shrank most along the radial plane, with Artocarpus shrinking to 59% of the original width, while the longitudinal shrinkage was modest (less than 15%) for most taxa. It was observed that shrinkage was not correlated with wood density. Wood with large-diameter vessels or wide bands of axial parenchyma, such as Artocarpus, Ficus, and Terminalia, tended to shrink to the most, but there were exceptions (e.g., Spondias also has large vessels but shrank very little). Results of this preliminary test indicated that, similar to findings from other charring experiments

140

Figure 5.9 Data entry form used for creation of the wood anatomy database.

141 cited earlier in this chapter, a variety of factors can influence the quality of charcoal produced, and individual attributes were noted to be generally poor predictors of preservation potential.

Table 5.1: Charred wood shrinkage (in mm)1

Taxon / Attribute % radial Wood Wood density % radial Charcoal Charcoal shrinkage shrinkage longitudinal longitudinal longitudinal longitudinal longitudinal Wood radial

Artocarpus altilis 17 17 10 14 0.59 0.82 .27–.49 Casuarina 15 25 14 22 0.93 0.88 .95–1.19 equisetifolia Claoxylon taitense 8 32 6 29 0.75 0.91 .50 Fagraea berteroana 46 13 42 12 0.91 0.92 .58–.72 Ficus prolixa 38 16 30 14 0.79 0.88 .35–.46 Glochidion 22 14 19 12 0.86 0.86 .54–.79 marchionicum Morinda citrifolia 14 15 12 15 0.86 1.00 .52–.64 Premna taitensis 11 7 10 6 0.91 0.86 .60–.75 Spondias cytherea 11 17 11 16 1.00 0.94 .40–.50 Terminalia glabrata 11 27 8 22 0.73 0.81 .40 Terminalia catappa 12 20 10 15 0.83 0.75 .46–.68 1 Unpublished data from Huebert (2009). 2 Density values from Détienne, Jacquet, and Orliac (1999).

Field methods Details of the charcoal recovery methods used at each particular site were summarised in the preceding chapter. While the majority of material evaluated in this study was recovered by dry screening or in situ collections, a portion of the charcoal recovered during the field excavations were processed by water flotation, which is described. While flotation schemes recover more complete archaeobotanical assemblages than dry screening, they are also time-consuming, labour intensive, and subject charcoal to further fragmentation. To manage time and labour constraints, I opted to float sediment from a limited number of test units in the field and to collect bulk samples from others for flotation back in the laboratory. Samples from only three test units at Teavau’ua South in Anaho were processed in the field, and at Pahumano in Hatiheu, where sediments were very clayey, 1 L bulk samples were collected from charcoal-rich layers for flotation in the laboratory.

142

Samples of sediment, collected from throughout arbitrary levels, were processed using a simple bucket flotation technique adapted from procedures formulated for use in the humid tropics of western Oceania (Fairbairn 2005). Each 10 L bucket of sediment was slowly immersed in water drawn from municipal sources, and agitated by hand. Floating residue was poured off into a sieve lined with fine-weave chiffon fabric. Heavy residue was wet-screened though 3.2 mm (1/8 inch) mesh using water poured gently from a tap above, and the recovered material was spread onto a chiffon. Each bundle was tied, tagged, and suspended in a shady spot until completely dry. Samples were then sorted by separating charcoal from other recovered materials, and they were packaged for shipment back to the laboratory. All equipment used in flotation including screens, sieves, chiffons, and trowels were thoroughly washed with fresh water between samples. A flotation log was kept to record details of the provenience, sediment, and characteristics including the volume and weight of each sample.

Sample selection Many hundreds of charcoal samples were available for analysis, and to best meet the requirements of this project a set of criteria were devised to guide sample selection. These decisions were based on considerations of spatial, temporal, and functional context of the samples. Charcoal samples from many different temporal and functional contexts were available. Because one of the overall aims of this research was to reconstruct vegetation histories, recovery of the broadest possible range of taxa was desired. Samples from a variety of functional contexts were selected to meet this goal. Those that reflected an aggregate of domestic activities over a long period of time including debris that accumulated near cooking areas were preferred, as well as concentrations of charcoal from in situ vegetation burning. Many other samples taken from fire features were also selected, however, as shorter-duration deposits such as these can inform on changes in major vegetation components and preferred resources over time (Huebert 2009). Occupation debris was defined as sediment that has accumulated on living surfaces in non-feature contexts. These deposits can contain material from numerous activities, they are likely to have accumulated over the duration of an occupation, and could represent many different activities. Thick, dark deposits of charcoal where artefacts, bone, and shell were lacking were designated burn layers. Some concentrations of charcoal were located below structures; the functional context of these materials is not known, but may result from

143 localised brush clearance or ritual fires burned before construction. These three context types were classified as ‘non-feature’ contexts for analysis, as they are often species-rich and time-averaged sources of information on past local vegetation. Charcoal samples from discrete fire features, such as earth ovens and open hearths, were included in this research, along with several post-moulds and pits of undetermined function. Context types were defined as they have been used elsewhere by Pacific archaeologists (e.g., Allen 1992, 191–96; Orliac and Orliac 1980; Orliac and Wattez 1989). Ovens were defined as fire features with diameters exceeding 70 cm, associated with at least a moderate amount of fire-altered rock and charcoal. In situ firing is sometimes denoted by oxidised perimeters and high heat. These features may have specialised fuel requirements because of the higher energy needed to heat oven stones (Allen 1992; Huebert, Allen, and Wallace 2010). Hearths were defined as smaller, shallower fire features with charcoal that often lack rocks, though they are sometimes stone-lined. Hearths may have been subject to lower firing temperatures than ovens because, for example, less heat is required to grill or roast food than to heat oven stones. Ovens and hearths are subject to somewhat different post-depositional processes: ovens are typically deeper and might later have become traps for debris or used as rubbish dumps (Allen 1992, 1994) while hearths, which can be more ephemeral features, may be more vulnerable to disturbance than pit ovens (Thoms 2009, 577–8). Each may have been reused, but the resulting deposit reflects a palimpsest of events that are functionally equivalent. Together, these context types were classified as ‘features’ for the present analysis and they generally represent shorter-duration events.

Identification Botanical samples from archaeological sites can be large and numerous, especially when large-scale flotation programmes have been employed to collect material. The use of a two-level processing scheme has been suggested as an effective and expedient method to assess patterning in large assemblages (Toll 1988). In this method, material in a sample is divided by size class and the researcher reviews and characterises each fraction within a limited timeframe under low-power magnification. After scanning, decisions are made about which sites or contexts should be fully processed. Procedures such as this allow the researcher to collect data on many more samples, and review many more materials, than simply selecting a random number of samples for a full sort. In the present study, nearly all available samples were scanned before a decision was made to accept or reject them. In

144 scanning, I examined the quantity and size of fragments and the variety of plant parts that were present. Samples that were entirely composed of charcoal too fragile to examine were rejected. Sample size was not a consideration for inclusion, as it was noted that very small samples could contain numerous taxa, and in one instance the largest sample examined was composed of a single taxon. Many samples included in this study were already relatively clean and free of sediment, but the bulk samples collected from two sites (Anaho Structure 336 and Hatiheu Pahumano) were processed in the laboratory. These materials were floated using a simple procedure outlined by Stuijts (2006). Distilled water was used in this process, as some of these materials were radiocarbon samples. After measuring sediment volume, the matrix was gently immersed in distilled water and placed briefly on an agitator, then allowed to sit overnight to loosen the compacted sediments. The light fraction was poured off into a geological sieve and spread out to air dry. The heavy fraction was poured through a 3.2 mm sieve and the retained material was turned out to air dry. After drying, charcoal was separated from both heavy and light fractions. Fine sediment (all material under 3.2 mm) was collected, dried, and put into storage. Initial tests were performed to determine the minimum effective fragment size to identify. Charcoal was carefully poured through nested 6.4 mm (1/4 inch) and 3.2 mm (1/8 inch) geological sieves. Material from each fraction was inspected to assess whether some taxa tended to break into smaller fragments than others. After attempting to identify most of the charcoal from eight samples, the results were reviewed. First, it was noted that in five samples, some taxa—not consistently the same taxa—were found only in the smaller fraction. Second, much of the small material in the 3.2 mm fraction was difficult to handle and tended to crumble upon examination. It was concluded that while there is useful information in charcoal fragments under 6.4 mm in size, material close to 3.2 mm was too small, and subsequently I decided only to evaluate material that was over 4 mm in size.

Sub-sampling After identifying several samples in total, it was decided that a sub-sampling strategy was needed to meet the goals of this research within the allotted timeframe. Devising such a strategy involved determining how much material to identify, and choosing a splitting method. A strategy of examining successive sub-samples to taxonomic redundancy was selected (Leonard 1987). This technique involves splitting a sample into equal portions, and

145 identifying sub-samples in succession until no new taxa are found. To use this method, two decisions had to be made including determining the smallest sample size eligible for splitting, and how many sub-samples to identify to ensure representativeness. To process samples efficiently, it was decided that each sub-sample should contain fewer than 20 fragments. In a previous analysis, I had established that identification of 13– 18% of a sample recovered the most common taxa in an assemblage from this region (Huebert 2008, Table 5), and it was reasoned that a somewhat larger percentage should be identified to meet the goals of the present study. A minimum of 15–25% was placed, and in many cases 25% or more of each sample was identified. To aid in determining when a sample had been adequately analysed, the remaining materials were scanned under a stereomicroscope. No standard method has been adopted for physical task of splitting charcoal samples, and it was noted that practitioners usually select a method with the aim of limiting further fragmentation. Commonly used techniques involve the use of physical sample splitters, the separation of samples into various size classes, or the use of a grid system (Pearsall 2000; van der Veen and Fieller 1982). A riffle box was selected to split samples totalling over 64 fragments, and to reduce abrasion the metal catch trays were lined with a cushion. When samples were small, they were divided by placing fragments sequentially into small containers.

Charcoal identification Identifications were performed using procedures adapted from Leney and Casteel (1975): fragments were snapped crosswise and lengthwise to expose transverse and tangential longitudinal planes, and cleaved along a radial plane using a scalpel. These were pressed into a shallow dish of clean sand for microscopic examination, and the faces were observed using a Zeiss AxioImager epi-illuminated microscope at magnifications of 50– 500X. Potential matches were first deduced using a dichotomous key translated from Détienne and colleagues (1999), before employing the reference collections. Many fragments were photographed during this process to create an archive that would aid subsequent identifications, and any new or unusual features were recorded and photographed at the first occurrence. Some materials were particularly challenging to examine, especially when minute features such as intervessel pit sizes or vessel-ray pit patterns had to be examined, and these materials were pulled for examination with the SEM.

146

Figure 5.10: Data entry form for charcoal identification.

Fragments with distinctive characteristics that could not be positively matched to any taxon in the reference collection were assigned to a numbered category (e.g., Unknown 1, Unknown 2). For each taxon that could not be positively matched, a profile of the features present on all three planes was created. In some cases, anatomical features could not be positively discerned because of the condition of the charcoal or the inability to cleave it cleanly. Fragments that displayed characteristic hardwood cellular patterning, such as vessels or rays, but were too small or indistinct to attempt identification were recorded as

147 indeterminate. Fragments too fragile or distorted for identification were recorded as unidentifiable. All materials were bagged by taxon.

Data recording Results were then tallied and quantitative and qualitative data was entered into a relational database (Figure 5.10). All provenience data from the sample tags were recorded in the database, along with a brief description of the sample condition and the recovery method used. The size fraction was also recorded. Fragment counts for each taxon and each plant part encountered, as well as for all unidentified material in a sample, were then recorded. The total weights of each taxonomic group were recorded, along with any notes made during identification. Qualitative data was also collected during the identification process and recorded in this database. These details were recorded at the taxonomic level, not for each individual fragment. They included notes about the fragment condition, presence of features such as compressed pores, ray fissures, sinuous ray patterning, fungal hyphae, and wide or unusual parenchyma banding.

Summary Though wood charcoal is one of the most commonly occurring remains recovered from archaeological sites, and it has significant potential to inform on past vegetation and human activities, the study of this material is not without challenges. In this chapter, the methods used to study wood charcoal assemblages in general were reviewed, and I then described analysis procedures used in the present study. The history of wood charcoal analysis was reviewed, and an overview of wood anatomy, tree growth, and the formation of charcoal were provided with the aim of contextualising a discussion of the challenges inherent in identifying these materials. A brief history of charcoal analysis in Polynesia was summarised to illustrate the contributions of previous research and demonstrate the potential wood charcoal analysis holds for future study in the region. The methods used to study wood charcoal were then presented and discussed in detail. These topics ranged from consideration of formation processes to cultural, taphonomic, and analytical biases that affect charcoal assemblages. While much has been written about these processes in the palaeoethnobotany and anthracology literature, practitioners admit that the discipline is not yet mature and there is a need for further study

148 in many areas. Approaches to quantifying charcoal were reviewed, and methods anthracologists use to evaluate sample adequacy, richness, and diversity in vegetation reconstructions were explained. It was acknowledged that to adapt these ideas to assemblages from tropical biomes, the methodology requires further refinement. Finally, a review of the methods of identifying tropical wood charcoal has illustrated some challenges that may prevent precise and accurate identifications, and the importance of a charred comparative collection was emphasised. In the last part of this chapter, the field and laboratory methods used in the present study were described. Creation of the comparative collection of fresh wood and charcoal specimens were described in detail. A preliminary study of wood-to-charcoal shrinkage was conducted on some of these materials, and it was found that some aspects of gross anatomy and wood density were poorly correlated with degree of shrinkage, and are therefore poor predictors of preservation potential. Field methods used to recover and process charcoal were described, and laboratory methods were explained in detail. Defining characteristics of functional contexts were provided, and the rationale for sample selection and sub-sampling was explained. Lastly, the process of identification was described along with details of the quantitative and qualitative data that were collected and recorded in the project database.

149

Chapter 6: Systematic Review and Data Quality Assessment

In this chapter The natural history of the Marquesas Islands have been described, and the traditional culture and historic observations of life and landscape in the islands have been presented. Archaeological contexts of the three main study areas were described, and the methodologies used to analyse wood charcoal assemblages were outlined in the preceding chapter. Results of the charcoal identifications are presented in this chapter. Charcoal assemblages, like all archaeobotanical material, must be interpreted after considering the extent of biasing factors that are both natural, cultural, and analytical, and the results are only plotted here after a rigorous assessment of data quality. The chapter begins with a systematic review of the taxa that were identified. This review includes a summary of tree form, status in the Marquesas, and ecological habitat. Many qualitative attributes of the samples and fragments were recorded, and these data are then considered. Next, sample adequacy is assessed by employing two methods that are frequently used by anthracologists. Several data quality assessments are also performed, including an evaluation of the appropriate units for quantitative analysis, an investigation into the need to separate data from feature from non-feature contexts for analysis, and consideration of how variations in sample size have influenced relative abundance values.

Taxa identified In this study, 191 charcoal samples from 123 proveniences (unique test units and strata, or individual features) were analysed. 6525 fragments of charred wood, nutshell, bark, and parenchymous tissue were identified in this study. The total quantity of samples per valley, and corresponding totals for fragment count and weights, are presented in Table 6.1. Absolute counts and weights of the charcoal identifications are provided in Appendix A.

150

Table 6.1: Summary of sample quantities and fragment weights and counts per area. Valley Qty of samples Fragment count Charcoal weight (g) Anaho 102 3343 575.2 Hakaea 52 2373 180.8 Hatiheu 37 809 112.7 Totals 191 6525 868.7

A minimum of 59 distinct taxa were identified in this assemblage (Tables 6.2 and 6.3). All woody materials derive from monocotyledons and angiosperms (hardwoods). Many identifications could be made to the species level because prior to the 19th century, numerous genera found in this study were represented by a single species in the Marquesas. For speciose native genera, assignments were usually made only to genus, denoted by the suffix sp. (singular) or spp. (plural, when more than one species may be present). As previously discussed, numerous factors make the identification of charred wood to species difficult or sometimes impossible: botanists often differentiate species by external morphology and the wood of species within a genus can be indistinguishable, or the wood anatomy of different taxa can sometimes have many shared features, which can become a concern when key differentiators are obliterated in charcoalification. A related concern is the poor documentation of some woods, especially many endemic shrubs and trees, as undocumented taxa do appear to be represented in these assemblages. Most of the charcoal fragments in this assemblage were very small (close to 4 mm, the smallest size examined), small, or medium sized (occasionally up to 4–5 cm). Most was from mature wood, though there were a number of charred twig fragments including, in frequency rank-order: indeterminate hardwood species, Sapindus saponaria, and Thespesia populnea (Figure 6.1a). Only one or two twig fragments of cf. Artocarpus altilis, Celtis pacifica, Premna, Cordia, and several other natives were recovered. The only woody root fragments identified in this assemblage were from Sapindus saponaria and an indeterminate hardwood. Some indeterminate material was knotty, as were some of the unidentifiable charcoal fragments. A few knotty fragments from taxa that have distinctive anatomy, including Thespesia, Sapindus, Celtis, were also identified, while others, including several fragments comparable to Artocarpus and Terminalia, could only be tentatively identified. In addition to wood charcoal, coconut, candlenut, and Pandanus nutshells and other carbonised plant parts were found in this assemblage (Table 6.3). Some fragments of bark and parenchymous tissue were identified, but in most cases these materials were very small and had few distinctive features, with one exception. Charred tissue from Hakaea sample

151

#5721 is a fragment of an underground storage organ. Other materials identified include indeterminate root fragments and several nutshells that could be distinguished from coconut, candlenut, or Pandanus, but were otherwise too fragmentary for identification.

Table 6.2: Wood taxa identified in Nuku Hiva samples.1,2 No. species No. samples % samples Scientific Name Marquesas present present Aleurites moluccana 1 0.5 Allophylus marquesensis 7 3.6 Alphitonia marquesensis 1 0.5 Alstonia cf. costata 2 1.0 Artocarpus altilis 49 25.5 Barringtonia asiatica 2 1.0 Calophyllum inophyllum 11 5.7 Casuarina equisetifolia 1 0.5 Celtis pacifica 26 13.5 Cerbera manghas 3 1.6 Cocos nucifera 32 16.7 Coprosma spp. 6 12 6.3 Cordia subcordata 2 13 6.8 Crossostylis biflora 2 1.0 Cyclophyllum barbatum 14 7.3 Dodonaea viscosa 6 3.1 Erythrina variegata 1 0.5 Guettarda speciosa 9 4.7 Hibiscus spp. 3 39 20.3 Inocarpus fagifer 3 1.6 Maytenus crenata 28 14.6 Melicope spp. 7 3 1.6 Metrosideros collina 2 1.0 Morinda citrifolia 1 0.5 Pandanus tectorius 8 4.2 Pemphis acidula 1 0.5 Phyllanthus cf. marchionicus 3 14 7.3 Pipturus sp. 4 1 0.5 Planchonella spp. 15 7.8 Premna serratifolia 9 4.7 Psydrax odorata 11 5.7 Rubiaceae (indeterminate) ~40 1 0.5

152

No. species No. samples % samples Scientific Name Marquesas present present Santalum insulare 1 0.5 Sapindus saponaria 94 49.0 Terminalia spp. 2 4 2.1 Thespesia populnea 100 52.1 Wikstroemia cf. coriacea 2 1 0.5 Xylosma suaveolens 6 3.1 Unknown02 2 1.0 Unknown09 1 0.5 Unknown11 1 0.5 Unknown13 50 26.0 Unknown15 1 0.5 Unknown17 1 0.5 Unknown18 2 1.0 Unknown19 2 1.0 Unknown21 4 2.1 Unknown22 3 1.6 Unknown23 2 1.0 Unknown24 1 0.5 Unknown25 11 5.7 Unknown26 25 13.0 Unknown27 10 5.2 Unknown28 1 0.5 Unknown29 2 1.0 Unknown30 1 0.5 Unknown31 2 1.0 Unknown32 2 1.0 Unknown33 1 0.5 Unknown34 1 0.5 Indeterminate angiosperm 137 71.4 Indeterminate monocot 55 28.6 Indeterminate 1 0.5 Unidentifiable 85 44.3 1 From Butaud, Gérard, and Guibal (2008), Butaud (2010), and Rensch (2005). 2 Note some identifications are only comparable (cf.) to species occurring in this list.

153

Table 6.3: Non-wood plant tissues identified in Nuku Hiva samples. Scientific Name Plant part No. samples present % samples present Aleurites moluccana endocarp 30 15.6 Cocos nucifera endocarp 106 55.2 Pandanus tectorius drupe 13 6.8 Indeterminate bark 19 9.9 endocarp 3 1.6 parenchymous tissue 13 6.8 unknown 32 16.7

Systematic review A systematic overview of 38 of the taxa identified in this study are presented in Table 6.4. This review includes a summary of plant status, habit, and the ecological context where most often encountered. Traditional uses of the plants were summarised in Chapter Three (Table 3.1), and are discussed in more detail in consideration of the results at each study site. Plant habits are denoted as either arboreal, arboreal or shrub, or shrub according to the most likely form to be found in the study locations. The majority of taxa identified are arboreal (21) and arboreal or shrub (15). Taxa with an arboreal or shrub designation can occur as either small trees or shrubs depending on the local ecological conditions, and it was noted that this classification has a wide size range varying from one to ten metres tall. Only two true shrubs, Dodonaea viscosa and Psydrax odorata, were noted in the assemblage. The typical ecological contexts are indicated as either mesic (moist), xeric (dry), strand (coastal), or mixed. Fourteen are typically found in mesic forests, while eight coastal and five xeric forest taxa were identified. Eleven plants grow in a wide range of habitats. A structured review of these data are conducted for each study location in Chapter Seven. The status of each taxon has been classified as either a Polynesian introduction, indigenous, or endemic. Many taxa found in this study are indigenous, and some are endemic to the Marquesas. Two distinctly different genera from the Sapotaceae family were found in numerous samples; one was identified as Planchonella and the other, Unknown13, more tentatively as Sideroxylon (syn. Nesoluma). More than two taxa may be present in these groups as some anatomical variability was noted. While reference materials for woods of the family were extensive, examples of these two genera were limited to one or two

154

Table 6.4: Systematic review of plant taxa identified in this study.

Marquesan and Ecological Scientific Name Status2 Habit English Name1 Context

Apocynaceae Alstonia costata Polynesian Alstonia, I Arboreal or Mesic atahe (Tahiti) Shrubs Cerbera manghas ʻEva, red-eye sea I Arboreal Mesic mango Arecaceae Cocos nucifera Ehi, coconut P Arboreal Strand

Boraginaceae Cordia subcordata Tou, Island walnut I Arboreal Strand

Calophyllaceae Calophyllum inophyllum Temanu, Indian laurel, P Arboreal Strand portia Cannabaceae Celtis pacifica Vaimanini, Polynesian I Arboreal Mixed celtis Casuarinaceae Casuarina equisetifolia Toa, ironwood P Arboreal Mixed

Celastraceae Maytenus crenata Koinaina E Arboreal or Mixed Shrubs Combretaceae Terminalia spp. Maʻiʻi, tropical almond I, M? Arboreal Mixed

Euphorbiaceae Aleurites moluccana ʻAma, candlenut P Arboreal Mesic

Fabaceae Erythrina variegata Kenae, coral tree I or P Arboreal Xeric Inocarpus fagifer Ihi, Tahitian chestnut P Arboreal Mesic

Lecythidaceae Barringtonia asiatica Hutu, fish-poison tree I Arboreal Strand

Lythraceae Pemphis acidula ʻAʻie (Tahiti) n/k Arboreal Strand

155

Marquesan and Ecological Scientific Name Status2 Habit English Name1 Context

Malvaceae Hibiscus spp. Hau, hau heʻe, koute (3 I, P Arboreal or Mesic types) Shrubs Thespesia populnea Miʻo, Pacific rosewood I Arboreal Strand

Moraceae Artocarpus altilis Mei, breadfruit P Arboreal Mixed

Myrtaceae Metrosideros collina Heua I Arboreal or Mixed Shrubs Pandanaceae Pandanus tectorius Haʻa, screwpine I Arboreal Mixed

Phyllanthaceae Phyllanthus Tevai, hiʻitevai E Arboreal or Mixed marchionicus Shrubs

Rhamnaceae Alphitonia marquesensis Makeʻe E Arboreal Mesic

Rhizophoraceae Crossostylis biflora Mori (Society Islands) I Arboreal Mesic

Rubiaceae Coprosma spp. Mirrorplant, E Arboreal or Mesic hupilo (Hawaii) Shrubs Cyclophyllum barbatum kohenua I Arboreal or Mesic Shrubs Guettarda speciosa Hano, Beach gardenia I Arboreal Strand Morinda citrifolia Noni, Indian mulberry P Arboreal or Mixed Shrubs Psydrax odorata Kohenua I Shrubs Mixed

Rutaceae Melicope spp. Auona E Arboreal or Mesic (M. fatuhivensis), Shrubs alani (Hawaii) Salicaceae Xylosma suaveolens Piʻapiʻau I Arboreal Xeric

156

Marquesan and Ecological Scientific Name Status2 Habit English Name1 Context

Santalaceae Santalum insulare Puahi, sandalwood E Arboreal or Xeric Shrubs Sapindaceae Allophylus marquesensis Puʻuheke I Arboreal or Strand Shrubs Dodonaea viscosa Tiatia I Shrubs Xeric Sapindus saponaria Kokuʻu, soapberry I Arboreal Xeric

Sapotaceae Planchonella spp. (note noted) n/k Arboreal Mesic Unknown13 - n/k Arboreal or Mesic Shrubs Thymelaeaceae Wikstroemia coriacea Akatea I Arboreal or Mixed Shrubs Urticaceae Pipturus sp. Hoka, hona I, E Arboreal or Mesic Shrubs Verbenaceae Premna serratifolia Vaʻovaʻo I Arboreal or Mesic Shrubs 1 From Butaud (2010), Rensch and Whistler (2009), and Whistler (2004, 2009). 2 I, Indigenous; P, Polynesian introduction; E, Endemic; M, Modern introduction; n/k, not known.

individuals. No native plants from this family are currently listed for the island group (Smithsonian National Museum of Natural History n.d.), though both can be found in other archipelagos in the region. At least eight taxa are Polynesian introductions. Many of these are part of the well- known suite of plants transported by early Polynesian settlers (Whistler 2009 and Table 3.1). Some genera include both indigenous and introduced species, or have statuses that are less certain. One example is Hibiscus. The large-flowering shrub (H. rosa-sinensis) and the slender-trunked sterile Hibiscus (H. tiliaceus subsp. tiliaceus ‘Sterilis’) are generally considered to be Polynesian introductions, but the status of the common H. tiliaceus is less certain. Some consider it to be native to the Marquesas (Florence 2004), while others

157 discuss the possibility that it was a Polynesian introduction (Butaud 2010; Whistler 2009). There is some concern that it may not be possible to differentiate between the wood of these three taxa, a topic that will be subsequently discussed, so in the present study these materials were classified under the generic heading. Other species that have uncertain classifications as Polynesian introductions include Terminalia, the tropical almond, which has one indigenous (T. glabrata) and one possibly modern-introduced (T. catappa, Butaud 2010) species in the archipelago. Very little of these materials were identified, and determinations to species level could not be made because of overlapping anatomical variations noted in reference specimens. Erythrina variegata was also identified; the charcoal is soft and light, and was found in only one sample. Erythrina is an uncertain Polynesian introduction in the Marquesas, with sources listing it alternately as indigenous (Smithsonian National Museum of Natural History n.d.) or Polynesian introduced (Butaud 2010).

Qualitative data results Qualitative aspects of the charcoal were recorded at a low level as present or absent on a per taxon, per sample basis. These findings are illustrated in Figure 6.1. Little of this information has been captured for charcoal assemblages in the region, as previously discussed, but it does have potential to inform on human activities. These features can reflect important characteristics of the wood burned (Dufraisse 2006, Fig. 1). Collapsed pores and radial cracks can result from burning green or wet wood, though the presence and extents of these attributes vary by species. Other data recorded included the presence of anomalous features on otherwise typical specimens, which I began to take note of after they were encountered in multiple specimens.

Artefacts of burning Collapsed pores were noted most frequently in fragments of indeterminate material, Thespesia, Sapindus, and Hibiscus (Table 6.5). In some cases these distortions made identifications tentative, as for one or two samples of Artocarpus and Phyllanthus. Ray fissuring was noted in a number of Thespesia, Sapindus, and Hibiscus fragments, as well as some indeterminate material. It was also noted in Artocarpus, Unknown13, Xylosma, and in one or two samples of Phyllanthus, Psydrax, Cordia, and Celtis. In some instances fissuring was so pronounced that identification was impossible.

158

Figure 6.1: Twigs, knots, collapsed pores, fissured rays, vitrification, and anomalous banding noted: a) Thespesia populnea twig (TS), Anaho #110; b) Indeterminate knot (TLS), Anaho #6893; c) Sapindus saponaria collapsed vessels (TS), Anaho #6356; d) Xylosma suaveolens radial cracks (TS), Hakaea #5720; e) Artocarpus altilis vitrification (TS), Hatiheu #7153; f) Sapindus saponaria wide non-parenchymous band (anomaly), Anaho #unspecified.

a. b.

c. d.

e. f.

Any vitrification of charcoal fragments was also noted. Vitrification is the fusion of different anatomical elements that results in an overall brilliant or glassy appearance. It has been said to alter the anatomical structure to the extent that identification is not possible (Théry-Parisot 2001 in McParland et al. 2010), though I have not found this to be true for all material in the assemblage. Distinct stages of the process were observed, especially in Thespesia populnea, as it was noted that some areas of a fragment retained identifiable

159

Table 6.5: Qualitative features of note and genera in which conditions were most frequently observed. Qualitative feature No. occurrences Collapsed pores Indeterminate or unidentifiable 11 Sapindus 5 Thespesia 4 Hibiscus 3

Radial cracks or fissuring Indeterminate or unidentifiable 39 Thespesia 26 Sapindus 15 Hibiscus 11 Artocarpus 4 Unknown13 4 Xylosma 3

Vitrification (anatomical elements appear melted or glassy) Indeterminate or unidentifiable 30 Thespesia 19 Artocarpus 5 Hibiscus 5

characteristics. Vitrification did hamper identification in some instances and many fragments were deemed unidentifiable for this reason, but some fragments of partly vitrified Hibiscus and Artocarpus were also observed. This condition was noted once or twice in Cyclophyllum, Guettarda, Xylosma, Cordia, and some of the numbered unknown taxa. While the exact causes are uncertain, the burning of green wood, seasoned wood, high burning temperatures, and re-burning were not associated with vitrification in several temperate hardwoods recently studied (McParland et al. 2010).

Anatomical anomalies The presence of some anomalies were also observed, including several instances of ring porosity in woods not generally known to display this feature. Ring porosity occurred

160 in four of the 100 samples that had Thespesia and three of 94 having Sapindus wood charcoal. Sapindus also occasionally displayed atypical, wide non-parenchymous bands on portions of otherwise typical fragments (Figure 6.1f). These observations necessitated consideration that some taxa may have significant anatomical variations, which can only be understood after examining many hundreds (or thousands) of fragments. Concerns regarding anatomical variability in the woods identified in this study are discussed more fully at the end of this chapter.

Assessing sample adequacy Evaluating the adequacy of sampling in paleovegetation studies is an important practice, especially when research aims are to reconstruct vegetation communities. Two complimentary analyses that are central to anthracological practice were conducted on this data set. The first was an assessment of sample sizes using taxon accumulation curves. The second was an assessment of sample richness, which was achieved by analysing Gini Coefficient values. To reduce the influence of biases introduced by differing cultural and natural processes, the following tests were performed using only charcoal from non-feature (i.e., not ovens, hearths, etc.) contexts.

Taxon accumulation curves Ideal minimum charcoal sampling thresholds have been calculated for a number of global regions, including several tropical areas where from 200 to 400 fragments of charcoal per stratigraphic level have been shown to best represent palaeovegetation communities (Dotte-Sarout 2010a, 138–41, Figs. 38–40 for New Caledonia; Scheel-Ybert 2002 for Brazil). These data are largely unexplored in Polynesian settings, but some relevant data was available from palynological studies in the region. Studies of pollen from several sites in the central east Pacific was of particular interest. In these studies, Parkes (1994, 52–3) determined the ideal quantities of pollen and spores that needed to be identified varied widely due to differences in taxonomic richness at each study site. Counts of between 200 and 400 grains were often sufficient, but some samples were very diverse and identification continued beyond the upper boundary. The foregoing suggest that counts of 200 to 400 fragments of charcoal may be adequate for the Marquesan assemblages. There is a

161 possibility these values are too high, as the Brazilian and New Caledonian assemblages contained up to twice as many taxa as were found in the present study. The following evaluations were performed by plotting taxon accumulation curves for strata at each individual study site where over 100 fragments of charcoal were identified from non-feature contexts. Indeterminate material was included in fragment totals. Data sequences were sorted randomly to ensure the curves did not stabilise artificially, and five sequences were generated for comparison. Results were assessed by considering where plateaus were noted. In some cases there were two plateaus, one that occurred when the more common taxa had been identified, and another when most of the rarer taxa had been found.

Figure 6.2: Taxon accumulation curves for Hakaea Beach site.

At the Hakaea Beach site (Figure 6.2), in the earliest deposit (layer VII) the first plateau was reached at about 100 fragments, when 13–16 of the total 29 taxa were recovered. A second was noted at about 450, by which time the rarer taxa had been encountered. All taxa had been found after 800–900 fragments were identified. Fewer taxa were recovered from the later strata at this site, though the overall sample sizes were also smaller. For layer V material, a plateau was noted in the curve at around 100 fragments,

162 when 15–17 of the 18 taxa that were found in this strata had been encountered. Though the smaller sample size probably had an effect on total richness (a general observation for archaeobiological datasets (Grayson 1984, 132–3), it was noted that the first plateau was reached in this assemblage sooner than it was for layer VII, where Ntaxa=18 at around 150 fragments. This finding could indicate overall species richness was lower by this period. In layer III, the curves begin to level off sooner. By 75–125 fragments, 10–12 of the 14 taxa had been found, and after 225 most had been identified. An overall lower richness value is suggested for this period, especially in comparison to results from layer VII where after 225 fragments, around 20 taxa had already been encountered.

Figure 6.3: Taxon accumulation curves for Anaho sites.

At Anaho, only two strata from different excavations were eligible for analysis (Figure 6.3). These two sites are located only a few hundred metres apart, and they are considered representative of the same vegetation community. Material from the Teavau’ua (AHO-1) site created a curve that did not reach a plateau until approximately 250–300

163 fragments, by which time 11–13 of the 13 taxa found at this site had been recovered. At the Teavau’ua South site, most of the 17 taxa recovered had been encountered after 150–200 fragments were identified. The differences in richness between these two sites may be related to differences in cultural activities that took place at each location. The Teavau’ua AHO-1 site was as an intensively used food-preparation area, and there is evidence that Marquesan cooking fuels were highly selected (Huebert, Allen, and Wallace 2010), probably resulting in a biased assemblage. I noted that when feature contexts were added to the Teavau’ua AHO-1 curve (N=1188), species richness increased by a very small measure (2) and the plateau was reached similarly around 325 fragments, which supports this suggestion.

Figure 6.4: Taxon accumulation curves for Hatiheu Pahumano site: non-feature contexts (above) all contexts (below).

The Pahumano assemblage from Hatiheu Zone H is smaller than the others included this evaluation (N=172). Curves for this sample level off earlier than others, and after 50–75 fragments the most common taxa were encountered (Figure 6.4). A second plateau may be

164 present at about 150 fragments, at which time all 10 taxa had been encountered. However, taxonomic richness for the entire assemblage from the site was actually much higher than this (Ntaxa=19) because two species-rich features were encountered: Efe. 10, an oven, and Efe. 19, a possible hearth or rake-out. A second chart was created using charcoal from all context types for comparison. These curves reached the first remarkable plateau at around 100–125 fragments, at which time 14–16 taxa had been found, and all taxa were seen after 250–300 of the 372 fragments had been identified. The Pahumano results indicate there are several points to consider before undertaking this type of analysis. First, there is some evidence accumulation curves do not provide useful information when assemblages are small. Artificial plateaus may result unless several hundred fragments (or more) are analysed. Second, it may not always be appropriate to reject feature contexts when assessing sample adequacy, especially when there are large differences in richness between the two datasets. This may be a special consideration especially in early contexts, when species richness values can be higher, as will be presently discussed. Three breakpoints were evident in these assemblages. First, the most common taxa were encountered after the first 100–200 fragments. However in some locations, such as the Anaho Teavau’ua South site, as few as 200 fragments were adequate to recover even the rarer taxa. This finding may be related to the smallness of the catchment, perhaps resulting from a range of activities and/or when diverse vegetation was concentrated in a limited area. Second, after 250–450 fragments most of the rarer taxa had been recovered at all sites. Lastly, in the largest assemblage (N=1083) which was the richest of the study, a final plateau was noted at 800–900 fragments, after which the curves stabilised almost completely. There are temporal dimensions to these results that are also important. The early deposit at Hakaea produced a large and diverse charcoal assemblage. A similar pattern was noted for Pahumano, though the assemblage was less diverse. These deposits probably represent an early burn-off of native vegetation. Deposits such as these produced large charcoal assemblages where there existed an opportunity to recover a wide range of taxa. Later cultural deposits tended to produce smaller assemblages. Some of these were rich in taxa, but overall a general trend towards lower diversity was noted later in the sequences (as exemplified at Hakaea). Overall, the results of this analysis suggest that charcoal counts of 250–450 per stratigraphic layer should provide rich samples of past vegetation in Marquesan coastal and

165 lowland settings. This threshold, however, can be variable. The likelihood a deposit may represent an early occupation of the area is also an important consideration, and for these contexts the high end of the range (or higher) is recommended. For smaller catchments, and many of those from late prehistoric contexts when vegetation had been extensively modified, the low end of this range will probably suffice. Finally, this demonstration provided evidence that even though taxonomic richness of native Marquesan forests are probably much lower1 than other species-rich tropical forests, the ideal quantities of charcoal that should be identified for adequate paleovegetation reconstructions were actually quite similar to other tropical forest settings.

Gini concentration indices The next assessment was to consider how completely vegetation communities were represented in the charcoal assemblages. This evaluation has also shown to be complimentary to the previous exercise in assessing sample adequacy (Scheel-Ybert 2002). As previously discussed, anthracologists test representativeness by comparing fossil to extant vegetation communities using the Gini Concentration (GC) index (e.g., Chabal et al. 1999, Fig. 13 Pareto curves). GC values are a measure of diversity as well as evenness: the higher the value, the lower the diversity and less even the distribution in a sampled population. Anthracologists have typically calculated the target values for this type of analysis using data from surveys of modern vegetation. Such data could not be located for any location comparable to the study sites, so benchmark values from studies in other settings were considered for this review. In temperate and Mediterranean regions, as previously discussed, GC values of about .80 are typical for undisturbed vegetation. For continental neotropical forests, a ratio of 25:75 (GC=.75) is typical for modern vegetation (see Scheel- Ybert 1998, 63–72). Scheel-Ybert noted that for archaeological material, values between .72–.78 indicate that sampling had been adequate; values under .71 tended to indicate that not enough material had been sampled (usually samples smaller than 100–150 fragments), and values of .85 or higher were indicative of assemblages with low species diversity. The latter was more typical of disturbed species-poor vegetation communities, and it was further suggested that archaeological assemblages may be similar due to cultural selectivity (i.e., preference for certain woods) (Dotte-Sarout 2010a, 140–1). Dotte determined that GC

1 Exact figures are lacking.

166 values for New Caledonian assemblages were slightly lower at .71–.75 and in some cases they were as low as .64–.68, which may be characteristic of evergreen tropical rainforests (Dotte-Sarout 2010a, 142–3). For Marquesan assemblages, then, it is anticipated that GC values would be in the high range (or higher) because species richness is typically lower than the aforementioned study locations. Gini Coefficients were calculated and Lorenz curves were plotted for temporal contexts in the following locations (Figure 6.5) using the open-source software R (http://www.R-project.org) v2.15 and the ineq library. As with the previous analysis, only material from non-feature contexts were used to reduce the influence of cultural bias. Indeterminate fragments were not included in this analysis. At Hakaea (Figure 6.5a), the oldest cultural stratum had a GC value of .77, which was well within the acceptable range as defined for the species-rich tropics. This is not surprising as the assemblage was, as previously discussed, the largest in the study (N=1083). The sample from layer V was clearly too small (N=115) as the GC value was very low (.51). Layer III produced an acceptable value (.71), though the sample was somewhat small (N=189) being at the low end of the range suggested in the previous analysis. At the Hatiheu sites, all values were low (Figure 6.5b, c). The highest was achieved with the Pahumano assemblage from Zone H, which had a GC of .67. Combined late-period samples from inland sites also had a low value of .58, and it was noted this assemblage was also close to the minimum recommended (N=201). It appears that samples from the Hatiheu sites are too small to represent an accurate picture of past vegetation, however in the previous analysis I suggested that it may not always be appropriate to ignore feature contexts and the Pahumano Zone H assemblage was used as an example as it included several species-rich features. When this evaluation was re-run on the entire Zone H assemblage, which had a much larger sample size (N=286), the GC rose to an acceptable .71. In the Anaho sites (Figure 6.5d, e, f), samples from the Teavau’ua AHO-1 layer IIIb produced a GC value of .77, which was also an acceptable value indicating this sample size was adequate (N=352). Sample sizes for the lower strata at this site, however, were too small as were those from the Teavau’ua South site, with all being well under .6. It was notable that the Teavau’ua South site layer III sample was not small (N=213) and was rich in taxa (N=17), yet still did not appear to be an adequate representation having a GC of only .54. The north valley paepae assemblages showed a similarly low value for both middle and

167

Figure 6.5: Lorenz curves for temporal assemblages from six Nuku Hiva study locations.

A ª(AKAEAª"EACH B ª0AHUMANO

1.0 1.0

III (N=189) G&=.71 D (N=39) G&=.51 V (N=115), G&=.51 E (N=27), G&=.55 0.8 VII (N=817), G&=.77 0.8 H (N=131), G&=.67

0.6 0.6

0.4 0.4 Fragment ct. percentage (L(p)) Fragment ct. percentage (L(p)) 0.2 0.2

0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Taxon percentile (p) Taxon percentile (p)

C ª(ATIHEUª)NLAND D ª4EAVAUgUAª!(/

1.0 1.0

Late contexts (N=201) G&=.58 IIIb (N=352) G&=.77 IV (N=41), G&=.43 0.8 0.8

0.6 0.6

0.4 0.4 Fragment ct. percentage (L(p)) Fragment ct. percentage (L(p)) 0.2 0.2

0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Taxon percentile (p) Taxon percentile (p)

E ª4EAVAUgUAª3OUTH F ª!NAHOª0AEPAEª.ORTHª6ALLEY

1.0 1.0

III (N=213) G&=.54 1−L (N=243) G&=.57 IV (N=44), G&=.35 2−M (N=93), G&=.54 0.8 0.8

0.6 0.6

0.4 0.4 Fragment ct. percentage (L(p)) Fragment ct. percentage (L(p)) 0.2 0.2

0.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0

Taxon percentile (p) Taxon percentile (p)

late periods (.54–.57). The sample from late period contexts in this area was also at the low end of the recommended range (N=243), but diversity was high (Ntaxa=22). Considering the consistency of GC values in several north valley assemblages it is possible that the past

168 vegetation community of this small area was very rich and diverse, but it was noted that values this low were not suggested for even the most species-rich neotropical contexts, so it seems unlikely. Based on these findings, it can be concluded that sample sizes of under 200–300 fragments are often too small to represent the richness of past vegetation in Marquesan coastal and lowland contexts, and samples closer to 400 fragments are usually required to reach a Gini Coefficient of between .71–.78. This analysis has provided some indication that ideal values for the tropical high islands of Polynesia are not necessarily higher than those for other tropical areas, as anticipated, and they actually appear to be rather similar. Ultimately, data from vegetation surveys should be used to calculate the target values. Two additional concerns regarding the use of these indices to study archaeological assemblages were investigated. One is the suggestion that high GC values can result in assemblages where material has been preferentially selected. This was tested using the Anaho Teavau’ua IIIb assemblage, where 1001 fragments of charcoal were available from feature contexts, most of which were earth ovens. The resulting GC value was indeed too high (.85) in comparison to that of non-feature contexts from this stratum (.77). It is evident that in the main, features should be avoided in these evaluations, though not without first considering feature types and richness values. Features may have had more than one function, or had secondary use as a rubbish pit, and they can sometimes be as rich in taxa as sediments at the rest of the site. The second concern is that it is unclear how fossil assemblages can reach GC values similar to those constructed from living vegetation communities, when various woods will be absent due to poor charcoal production or preservation. A limited test bed was created to consider how values of an ideal assemblage (GC=.77, Ntaxa=29) were affected by removal of some taxa. The value was only modestly lower (.02 or less) when up to 8 minor taxa were removed. When one major taxon was removed, this value was somewhat lower (.68), though the total number of fragments was still more than sufficient (over 800). Therefore, while this analysis appears to be robust in when minor vegetation elements are missing, when major forest components have (for example) soft wood or produce poor-quality charcoal the GC values will be low even though sample sizes appear to be adequate. It is evident that consideration must be given to taxa that were expected but not found, and some discussion of the characteristics of those woods must be presented when interpreting results. This may be the case for several Anaho north valley assemblages, including Teavau’ua South Layer III and Teavau’ua Layer IIIb, a topic that will be discussed in the next chapter.

169

Data quality assessment Several aspects of data quality were assessed. The first was sample quality, which was evaluated by examining the percentage of identifiable charcoal in each assemblage. Next, the most appropriate units for quantitative analysis were determined. As discussed in Chapter Five, raw counts or weight are typically not appropriate units for comparative analysis of archaeobotanical material because of their tendency towards unevenness (Pearsall 2000, 194–5), and several factors that may have affected fragmentation rates were evaluated. After this, a formal evaluation of the anthracologists’ recommendation that feature contexts be evaluated separately was undertaken by comparing the taxonomic richness of feature and non-feature contexts. Lastly, the influence of sample size on relative abundances of taxa was evaluated. This assessment highlighted whether sample sizes, rather than the parameters of interest, were being compared in subsequent analyses.

Indeterminate material Two classifications were used for charcoal that could not be identified: indeterminate and unidentifiable. Each informs on a different aspect of the assemblage. Measures of indeterminate material inform on how well the woods in an assemblage tend to produce identifiable charcoal. This classification was used when anatomical features could be discerned but those required for identification had been obliterated in charring. A second distinction, unidentifiable, was used to classify material that was in very poor condition. These fragments were often too fragile, small, badly warped, or degraded to fracture. Both classifications have been used by M. Orliac (2003, 165), who suggests unidentifiable (indéterminable) material could contain soft-wooded species such as Hibiscus, Aleurites, Artocarpus, or Erythrina. Indeterminate charcoal accounts for 14% of material by weight and unidentifiable material accounted for 6% of the total assemblage (Table 6.6). Most indeterminate material was wood, though approximately 125 fragments of nutshell, bark, and parenchyma are included in these figures. Reviewing results from the individual study areas, 15% of the charcoal from Hatiheu samples was indeterminate, while the percentage of these materials from Anaho (9%) was notably smaller, indicating the Anaho assemblage tended to produce charcoal that was more readily identified. Indeterminate material from Hakaea comprised a

170 larger percentage of the charcoal examined (30%), indicating some of the woods in this assemblage produced indistinct charcoal. The Hakaea charcoal was, overall, in good condition as the portion of unidentifiable material from this location totalled only 3% of the assemblage. Material from the other study locations was only in slightly poorer condition. Overall, a larger percentage of charcoal from Anaho and Hatiheu sites were able to be identified, while the Hakaea material was more challenging.

Table 6.6: Percentages of indeterminate wood and unidentifiable material, by weight. Location Indeterminate Unidentifiable Anaho 9% 7% Hatiheu 15% 5% Hakaea 30% 3% All locations 14% 6%

Units for quantitative analysis It was determined that the assemblages in this study do support quantitative comparisons because charcoal was excavated from locations that are in rather close proximity (i.e., same island, same side of the island), have similar deposition histories, and many samples were recovered by teams working under the same principal investigator using the same procedures. To compare assemblages, many archaeobotanists contemplate the issue of whether to use counts or weights as basic units when calculating relative percentages, as previously discussed. The following evaluations have been designed to evaluate this concern and assess the law of fragmentation asserted by anthracologists (i.e., Chabal et al. 1999). In previous studies of charcoal from Nuku Hiva, I determined that weights were the most suitable units for analysis (Huebert 2008, 2009), in part because some of the materials had been previously fractured during selection of radiocarbon samples. Though some of the samples in this assemblage had been subject to the same treatment, the majority were not, and it was determined that the most important need was to consider other factors that had influenced fragmentation to select the most appropriate measure. Five factors were assessed: study location, post-depositional environment, recovery method, functional context, and taxon. Kendall’s tau-b was selected as the most suitable correlation method for these tests. This method is sometimes used for archaeological applications where nonparametric (rank- based) measures of correlation are evaluated (Simek 1989). It was selected because these

171 data are very left-skewed with a number of outliers, and tau-b is robust in regards to extreme and tie values. These analyses were also performed using R and the built-in Kendall method, using counts and weights of each individual sample (i.e., samples from the same contexts were not combined).

Location First, the overall relationship between fragment counts and taxon weights for the entire assemblage was evaluated. The correlation value for all locations combined was strong and highly significant (Table 6.7, last row). This indicates either measure may be suitable when quantitative comparisons are assessed for the entire assemblage. For the individual study locations, the correlation values are also strong for Anaho and Hakaea, but the value for Hatiheu material was more moderate. It was a concern that many charcoal samples from the Hakaea Beach Site were very light and fragments rather small (under 1 cm) which may indicate samples were excessively fragmented. However, an examination of the scatterplot (Figure 6.6) shows that larger, heavier samples were the outliers. Overall, the results of this test do not show large variations in count-to-weight correlations between locations, and for inter-location comparisons either unit should be valid.

Table 6.7: Kendall tau-b correlation coefficients for count-to-weight by location1 Location Correlation coefficient (Tau) Anaho r(100)=.66, p<.001 Hatiheu r(36)=.54, p<.001 Hakaea r(50)=.68, p<.001 All locations r(190)=.62, p<.001 1 Results are considered significant at p=.05.

Figure 6.6: Scatterplot of count-to-weight correlation by location.

172

Post-depositional environment Next, the evenness of fragmentation in key cultural layers was evaluated. This analysis was performed to determine whether charcoal from any given stratum had significantly higher or lower rates of fragmentation than others, possibly due to the changes in sedimentation, soil chemistry, or abrasion. Post-depositional environment, and its effects on charcoal fragmentation that were previously reviewed, are central to this concern because these materials were recovered from several different sedimentary matrices. Coastal sites were buried in mainly coralline sands, while others were located inland where sediments were more rocky.

Table 6.8: Kendall tau-b correlation of charcoal count-to-weight by location and stratigraphic context.1,2 Location and strata Correlation coefficient (Tau) Count [weight in gm] Anaho Teavau’ua AHO-1 IIIb r(15)=.83, p<.001 1188 [213g] Paepae – coastal r(16)=.66, p<.001 873 [146g] Paepae – inland r(28)=.69, p<.001 765 [163g] Teavau’ua South III r(16)=.56, p=.001 292 [29g] Teavau’ua South IV r(6)=.30, p=.3 66 [9g]

Hatiheu Pahumano Zone H r(13)=.66, p<.001 372 [49g]

Hakaea III r(5)=.85, p=.008 249 [20g] V r(11)=.79, p<.001 469 [36g] VII r(19)=.68, p<.001 1114 [91g] 1 Results are considered significant at p=.05. 2 Strata were omitted where fewer than five samples were processed.

Charcoal count-to-weight correlations were found to be strong to very strong for many of the strata at coastal sites (Table 6.8). Anaho AHO-1 layer IIIb, samples from the paepae excavations, the lowest cultural layer at Pahumano, and most of the Hakaea layers appear to be rather evenly fragmented, with the correlation of samples from the former being very strong. Samples from strata at Teavau’ua South (Anaho) were more varied: layer III had a moderate correlation while IV had a weak, but perhaps not significant, value. Material from Hakaea layers III and V had surprisingly strong count-to-weight correlations,

173

Figure 6.7: Scatterplot of count-to-weight correlation by strata, Hakaea Beach Site.

,,, ,9

● ●

●

● ●

● ● ● ●● ● ● ● ●

9 9,, :HLJKW J

●

● ● ● ●

● ● ● ● ● ● ● ● ● ● ● ●● ● ●●● ● ● ● ● ● ● ● ●●● ●●

&RXQW

indicating the assemblage was not excessively fragmented, contrary to the impression that had formed when processing those samples. But Layer VII at the site had a less strong correlation, and an inspection of the scatter plot shows that the samples were composed of small, light fragments (Figure 6.7). This evaluation has demonstrated that quantitative measures are probably unsuitable for inter-site comparisons, and should be undertaken at a low level (perhaps using ranks).

Recovery methods Another expectation was that charcoal from screened samples would be more fragmented than material recovered by other methods. A corollary to this proposal was that charcoal from in situ samples, subject to less abrasion, would be less fragmented. Results show that material from 6.4 mm (1/4 inch) screens had a strong count-to- weight correlation indicating these materials were rather evenly fragmented (Table 6.9). Material from 3.2 mm (1/8 inch) screens had a similar correlation; these samples were mainly composed of a large quantity of smaller, lighter fragments with several exceptions (Figure 6.8). Charcoal from flotation samples had somewhat stronger correlation and though there were only eleven samples recovered by this method, the results were still significant.

174

Charcoal collected in situ, however, had only a moderate count-to-weight correlation. A review of the scatterplot shows that rates of fragmentation for the in situ samples were more variable: while many were composed of small light fragments, around one-third were notably larger by either weight or count. Evaluation of the functional contexts showed that some in situ samples taken from ovens were weighty, while some from other contexts were very light. Overall, it was concluded that recovery method has had only a modest influence on fragmentation rates. For screened samples, either quantitative measure would be appropriate for analysis. However, it is probably necessary to analyse feature samples separately, which is the next evaluation.

Table 6.9: Kendall tau-b correlation of charcoal count-to-weight by recovery method.1,2 Correlation coefficient Recovery method Count [weight in gm] (Tau) 6.4 mm screen r(45)=.68, p<.001 1258 [169g] 3.2 mm screen r(59)=.62, p<.001 3252 [379g] Flotation (water) r(9)=.71, p=.002 298 [21g] In situ sample r(44)=.57, p<.001 1300 [247g] 1 Results are considered significant at p=.05. 2 Some samples were omitted from this test because recovery method was not noted, or fewer than five samples were processed by a given method.

Figure 6.8: Scatterplot of charcoal count-to-weight by recovery method.

175

Functional context The importance of separating charcoal data by functional context was discussed at length in Chapter Five. An inspection of the correlation values per functional context (Table 6.10, Figure 6.9) shows that charcoal from non-feature and oven contexts had strong count- to-weight correlations, while other types were more weakly correlated (though the significance of those values were low). It was determined that either unit of measure should be appropriate when comparing material from non-feature contexts. Feature contexts need to be evaluated separately, however, and it is probably best to further isolate feature contexts by comparing those of like functional classifications (see Huebert, Allen, and Wallace 2010).

Table 6.10: Kendall tau-b correlation coefficients for count-to-weight by functional context1 Location Correlation coefficient (Tau) Non-feature r(128)=.65, p<.001 oven r(31)=.66, p<.001 hearth r(6)=.40, p=.2 pit r(5)=.30, p=.4 post (too few samples) other r(9)=.40, p=.1 1 Results are considered significant at p=.05.

Taxon Finally, the relationship between fragment counts and taxon weights in this assemblage was evaluated to determine whether certain species tended to fragment at higher rates than others. While other researchers have gone so far as to measure and classify individual charcoal fragment sizes (e.g., Di Piazza 1998), a more expedient method was chosen of examining fragment sizes per sample by looking at count-to-weight plots and then identifying general trends. Count-to-weight relationships were evaluated for the four most frequently occurring wood taxa in this assemblage, and for the wood and nut shells of coconut and Pandanus (Table 6.11). While most wood taxa had strong correlation values, breadfruit (Artocarpus) charcoal had a moderate value. This charcoal was often very light and soft, and an examination of the scatterplot illustrates this observation (Figure 6.10). Most breadfruit samples weighed less than 2 g, and many less than 1 g.

176

Figure 6.9: Scatterplot of count-to-weight correlations by functional context.

1. non−feature 2. oven 3. hearth 15 60

10 40 50

20 5 25

0 0 0 0 50 100 150 0 50 100 150 200 0 100 200 4. pit 5. postmould 6. other 4 Weight (g) Weight

10 5.0 3

5 2.5 2

0 0.0 1

0 −5 −2.5

5 10 0 10 20 6 8 10 12 Count

The other three woods evaluated, Sapindus saponaria, Thespesia populnea, and Unknown13 (Sapotaceae), all had strong count-to-weight correlations and therefore even rates of fragmentation, in agreement with the aforementioned law of fragmentation. It was noted throughout this analysis that Thespesia and Sapindus charcoal fragments were often firm and well-preserved. Though intrinsic factors probably influenced the abundance of these two charcoals, properties that are not well understood at this time, their frequency in this study is most likely influenced by some cultural preference for the wood. The same is probably true for Unknown13; this charcoal was often fissured, vitrified, hard, and somewhat dense, being brittle and tending to fracture laminally (i.e., in sheets) without much effort. For these reasons, it was expected that Unknown13 would be excessively fragmented, though it appears that this was offset by the slightly higher weight of the fragments (denoted by points in Figure 6.10 that are often close to, or above, the regression line). Monocot wood was more variable. Coconut had a weak count-to-weight correlation, and this variability could readily be seen in the plot. During analysis, it was noted that this

177

Table 6.11: Kendall tau-b correlation of charcoal count-to-weight for most frequently occurring taxa1 Taxon Correlation coefficient (Tau) Count [weight in gm] Wood Artocarpus altilis r(47)=.54, p<.001 141 [26g] Sapotaceae – Unknown13 r(48)=.68, p<.001 325 [28g] Sapindus saponaria r(92)=.63, p<.001 755 [100g] Thespesia populnea r(98)=.75, p<.001 1429 [311g] Cocos nucifera wood r(30)=.27, p=.05 68 [15g]

Nutshell Aleurites moluccana endocarp r(28)=.55, p<.001 206 [15g] Cocos nucifera endocarp r(104)=.70, p<.001 834 [65g] Pandanus tectorius drupe r(11)=.57, p=.02 32 [3g] 1 Results are considered significant at p=.05.

wood fragmented more variably than the hardwoods. Coconut wood charcoal was at times very dense and difficult to snap. Also, this charcoal was usually firm when in larger chunks but it could be brittle, and when pressure was applied longitudinally it sometimes fractured into many small pieces. Coconut shell had a unexpectedly strong and significant correlation between fragment counts and weights; this was also the largest sample size evaluated. Candlenut (Aleurites) shell fragments are more dense and heavy than other nutshells, but in this study fragments were often small which resulted in a moderate count-to-weight correlation. Some variability in these weights were noted, however, as several fragments weighed more than 300 mg and others less than 40 mg. Pandanus drupe fragments had a very similar, moderate correlation. These fragment sizes appeared to be as variable as other types of nutshell, though there were fewer samples of this material. It was noted that several nutshell caches were represented in these assemblages. While counts or weights are both acceptable units for analysis when looking at the wood and nutshells of many different taxa from this region, it is clear that a ‘law of fragmentation’ does not always apply and there are important exceptions to consider. First, the use of fragment counts would be most appropriate for assemblages that contain Artocarpus because the charcoal is light; it would be under-represented if weights were used. Second, wood from large palms such as coconut does not appear to fragment evenly

178

Figure 6.10: Scatterplots of charcoal fragment counts-to-weights for select wood taxa and nutshells. Axes have been logarithmically transformed to reduce highly skewed distributions.

Aleurites endocarp Coconut endocarp

10.0

0.1

Pandanus drupe Coconut wood

10.0

0.1

Artocarpus Sapindus

Log (Weight/gm) 10.0

0.1

Thespesia y unknown13

10.0

0.1

1 10 100 110100 Log (Count)

and weights can be quite variable; counts would also be a better choice for these materials to deemphasise variations. Lastly, these results provide important data on fragmentation rates for nutshells found in the region. These materials are usually analysed alongside wood charcoal, but there have been suggestions that rates of fragmentation may differ considerably, with nutshells fracturing into fewer, larger fragments than wood, resulting in

179 under-representation of nutshells if counts are used (Di Piazza 1998). This was, however, not a significant concern in the present study for either coconut, candlenut, or Pandanus shells. While some nutshells were slightly heavier than certain woods, they often fragmented into small pieces, and ultimately produced count-to-weight ratios similar to that of wood. Thus the use of either counts or weights may be appropriate for nutshells. This appears to be particularly true for coconut, which was present in over half of the samples in this study. Pandanus drupe and Aleurites endocarp fragments, on the other hand, could be under- or over- represented using weights because they were found to be quite variable in size. It is suggested that when assemblages contain large amounts of either of the latter, an investigation into fragmentation rates should be undertaken to determine the most appropriate measure suited to that assemblage.

Summary To briefly summarise the results of these evaluations, it has been determined that a majority of the assemblage is evenly fragmented. The overall correlation between fragment counts and taxon weights was strong and highly significant, and this indicates either quantification measure may be suitable for comparative analysis, especially when making more general site-by-site comparisons. Recovery methods have had only modest influences on fragmentation rates in this assemblage. More importantly, consideration must be given to the functional contexts of the samples being compared, and it was found that feature contexts should be evaluated separately from non-feature contexts. In several cases, it was demonstrated that charcoal fragmentation rates are similar for the most commonly occurring wood taxa in this assemblage and either measure may be appropriate. However, for breadfruit wood, an important component of the following comparative analysis, and several others, counts would be more appropriate units on which to base further analyses and these were selected for further quantitative analysis.

Taxonomic richness of functional vs. non-functional contexts Different functional contexts represent different types of human activities and as a result, charcoal assemblages found in various functional contexts may be biased in different ways. The following evaluation was designed to better illustrate why results should be separated by functional context, and to evaluate the extent of variability that exists between various types. Per anthracological principles, as discussed in Chapter Five, materials which have accumulated on living surfaces should be the richest as they represent long-term

180

Figure 6.11: Taxonomic richness by functional context type for all sites.

Occupation debris 5

Charcoal concentration 5

Oven 2

Hearth 6

Pit 3.5 Functional context

Postmould 1.5

Stone−lined pit 1

Indeterminate feature 2

0 5 10 15 20 Richness

accumulations from many functions. Charcoal collected from these contexts should be the most informative regarding local vegetation. As a corollary, features such as ovens are expected to contain few taxa because they require specialised fuel in order to heat oven stones to high temperatures. It was expected that less specialised features such as hearths would be more varied, as they do not have such challenging fuel requirements (Huebert, Allen, and Wallace 2010). To evaluate these propositions, the taxonomic richness values of each type of functional context were calculated by counting the total number of taxa present in all samples from each feature (Figure 6.11). Totals include materials marked as comparable (cf.). Feature types in this evaluation are earth ovens, hearths (open fire features), pits, post- moulds, and one large stone-lined pit. Indeterminate features were also considered, but as a separate category. Charcoal from non-feature contexts were classified into two types: occupation debris that accumulated on living surfaces, and isolated concentrations of

181 charcoal found in locations where few indications of intensive human occupation were noted. The latter may derive from localised vegetation clearance, in situ burning of debris, or they may be ephemeral hearths. The entire Nuku Hiva assemblage shows some remarkable differences in richness between various functional contexts. Ovens (N=19) contained a very restricted number of taxa with most containing 1 to 3, though there were several outliers with 8 or more. Indeterminate features and pits are mostly similar, with pits being only slightly richer. Hearths are notably richer with median 6 (sd=2.27, N=7) taxa. In contrast, occupation debris from living surfaces can be very rich with up to 20 taxa, though the median value is 5 (sd=3.95, N=68) taxa. Samples from charcoal concentrations in non-occupation contexts had a median of 5 taxa (sd=2.78, N=12) as well, but tended to be less rich than either hearths or occupation debris. Richness values were then evaluated separately by study location (Figure 6.12), and the differences were also apparent at each site. Ovens at most locations contained a restricted number of taxa (median 2–3) compared to other context types while hearth features were almost always richer. Charcoal from occupation debris, however, was quite variable depending on location. Most of the Anaho sites and Pahumano contained 3–6 taxa in these contexts, while the Anaho Teavau’ua South was slightly richer. At the Hakaea Beach site, however, most context types are richer overall. Charcoal concentrations were only found at the Anaho Structures and Hatiheu Pahumano sites; these contexts tended to be somewhat richer than others at the same location, but they were also more variable. Overall, feature types including earth ovens, post-moulds, and those that were indeterminate contained a very restricted number of taxa compared to general occupation debris at the same site. Hearths had a median richness value similar to that of occupation debris, but most contained a narrower range of taxa. In consideration of the wider implications of this finding, it was noted these results were similar to those of Coil (2004, 341–8) for the Kahikinui site. In that study, samples from stratigraphic levels were the most taxonomically rich (42 samples averaging 6 taxa per assemblage), while feature samples had a more moderate richness (18 samples, averaging 4–5 taxa per assemblage). The similarity of these findings strongly suggests that feature and non-feature contexts should be evaluated separately in Polynesian charcoal assemblages.

182

Figure 6.12: Taxonomic richness by functional context type, for each study location.

Anaho structures Anaho Teavau’ua (AHO−1) Anaho Teavau’ua South

Occupation debris

Charcoal concentration

Oven

Hearth

Pit

Postmould

Stone−lined pit

Indeterminate feature

Hakaea Beach Hatiheu inland Hatiheu Pahumano

Occupation debris

Functional context Charcoal concentration

Oven

Hearth

Pit

Postmould

Stone−lined pit

Indeterminate feature

0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 Richness

Sample size and relative abundance It has been established that, for some strata, sample size has had an effect on species richness in this study. It then also became important to evaluate whether the relative abundances of taxa within a given stratum were also influenced by sample size as these were selected for overall presentation of the results. This problem is often not obvious, especially because the minimum sample size needed to adequately represent a total population is frequently not known (Grayson 1984, 121–130). Abundance percentage values are interdependent, so this relationship was evaluated by considering the most abundant and frequently occurring taxon, Sapindus saponaria, which occurred in 49% of the samples. Only three sites had quantities that made this analysis possible, and even then it was noted that counts were low in many cases (Table 6.12). Some of the results were, however, interesting. At both of the Anaho sites, the rank of counts and relative percentages were inverted, indicating sample sizes were directly influencing relative percentage values. For the Hakaea Beach site, the first ranks were identical by either measure. But one of the count values in second position (a tie) had a

183

Table 6.12: Relative abundances of Sapindus saponaria in non-feature contexts, by stratum. Rank, Relative Stratum Count Rank % count abundance Anaho Teavau’ua AHO-1 IIIb (N=422) 27 1 .06 2 Anaho Teavau’ua AHO-1 IV (N=38) 8 2 .21 1

Anaho Teavau’ua South III (N=227) 14 1 .06 2 Anaho Teavau’ua South IV (N=41) 10 2 .24 1

Hakaea III (N=196) 91 1 .46 1 Hakaea IV (N=234) 27 2 .12 2 Hakaea V (N=150) 9 3 .06 3 Hakaea VI (N=115) 5 4 .04 4 Hakaea VII (N=696) 27 2 .04 4

relative percentage rank that was much lower (rank 4). Thus it appears that samples of a similar size (within approx. 100–150 fragments) produce relative abundance values that are comparable. Samples that are much larger or smaller produce values that may not be directly comparable with others at a site. Grayson (1984, 121) suggests that removal of stratum with small samples can remove this correlation, but it is not really feasible to remove any results from the present study. In consideration of these findings, however, emphasis on changing relative abundances at the sites was reduced and low-level quantification measures were used for inter-site comparisons.

Consideration of potential biases In the following discussion, various taphonomic, recovery, and analytical processes that have affected the assemblage are considered along with how these biases might have affected the results. I especially consider why some materials may be over-represented while others are under-represented or absent from the assemblages. In addition to the following topics, it should also be noted that taxa may also be infrequent or absent from these assemblages for cultural reasons. These may include avoidance of certain materials because they were considered poor fuels for certain purposes, produced toxic smoke, access may have been limited due to territorial boundaries, or the woods had uses restricted to certain socio-economic contexts, all considerations that were broadly discussed in Chapter Five. Specific cultural considerations regarding the presence or absence of taxa in this assemblage are addressed in the presentation of results (Chapter Seven).

184

Differential charcoal preservation It was noted that some taxa occurred more frequently than others in these assemblages. This outcome may be influenced by inherent properties of the wood. Certain taxa may produce more and/or more durable charcoal, while others produce charcoal with poor preservation qualities or turn more readily to ash. In the following discussion, I consider concerns regarding preservation potential. In this study, wood charcoal from Thespesia populnea and Sapindus saponaria were abundant, frequent, and often of good quality. Thespesia charcoal was so abundant that this result is unlikely to be related only to preferential preservation, and probably also relates to the frequency of use. It was encountered in many different fragment sizes and in various conditions from excellent to very poor, sometimes fissuring along the short chains of axial parenchyma or along the rays, or in varying stages of vitrification. Vessels of this taxon are usually not large, fibres are not especially thick, and rays vary from small to large throughout. No particular anatomical feature of Thespesia could be discerned that would make the charcoal especially durable, and some other property of the wood probably influences production of copious amounts of charcoal. Fragments of Sapindus were often firm and in good condition. While the abundance of this taxon probably also reflects its frequent use, some anatomical features appear to be factors that favour the production of good quality charcoal; these include thick-walled fibres and small ray cells, limited ray fissuring, and fissuring that seldom occurs along bands of axial parenchyma. Artocarpus altilis wood occurred frequently, though in lower abundances in this study. Breadfruit wood makes a poor quality fuel (C. Orliac 1991). It may tend to generate large volumes of charcoal as the wood is known to smoulder (Buck 1938), perhaps due the abundance of latex, but overall the charcoal has poor preservation qualities. In the present study, fragments were often soft and spongy, sometimes fracturing into small pieces upon examination. The more fragile nature of breadfruit charcoal can be attributed to the large vessel size and thinner walled cells that occur throughout. Hibiscus wood charcoal was observed in widely varying states of preservation, from light and crumbly to dense and brittle fragments. This wood combusts and turns to ash at a relatively low temperatures (Orliac and Wattez 1989, 71). H. tiliaceus is especially frequent and abundant in disturbed landscapes, it grows quickly, and the wood has many traditional uses. It was likely to have been a common forest element in the past. In consideration of

185 these factors, the uneven presence and quality of Hibiscus charcoal in this study is probably not an accurate reflection of past use or frequency in the local environment. Certain taxa were expected, but not found in this assemblage. Some were valued fuel or food resources, such as Broussonetia papyrifera (paper mulberry), Ficus prolixa (banyan), Pisonia grandis, Spondias cythera (vi apple), and Syzygium malaccense (rose apple). Some of these trees were probably common in forests or home gardens. While many of these woods have distinguishing characteristics that would have made them easy to identify, most produced reference charcoal that was soft, spongy, or delicate and it would not have preserved well in archaeological contexts. Ficus prolixa has distinctive large vessels and banded axial parenchyma arrangements, but it produces very soft charcoal because of its many thin-walled parenchyma cells. Spondias also has very distinctive large round vessels and rays composed of small round cells with large canals. The wood of this large tree is light and soft and it is not durable, as evidenced by the spongy reference charcoal it produced. Broussonetia papyrifera also features large pores and rays composed of small round cells similar to Artocarpus (also a Moraceae). This tree produces charcoal of a quality similar to Artocarpus. The trees were typically harvested to produce bark-cloth when young (J. Forster 1996, 273) and it is likely the small-diameter trunks combusted completely when burned. Pisonia grandis has very distinctive anatomy with large bundles of included phloem that, when burned, would cause large hollow channels to form. It is unlikely that much, if any, of this charcoal would preserve well. Barringtonia asiatica is a common indigenous coastal tree that was expected, but very infrequently found in this assemblage. This wood has distinctive, very large intervessel pits, a feature not common to woods of this region, so it would not have been overlooked. The wood is, however, very light which could have caused the charcoal to preserve poorly. Syzygium malaccense wood was expected in the assemblage, primarily because the fruits were important enough to be mentioned in Marquesan legend (W. Handy 1965, 47) and the trees were observed by several early contact period visitors. Several specimens of Syzygium wood were present in the reference collection, and the charcoal that was created was not of a particularly poor quality. I have identified this wood in samples from other locations on Nuku Hiva (data in prep.), noting that the archaeological fragments were somewhat soft, so it is likely that this material also does not preserve well. Shrubs may also be under-represented in this assemblage, including a number of useful species such as Piper methysticum (kava), Cordyline fruticosa (tī), Solanum viride (cannibal cherry). Wood specimens from these and several other shrubs were present in the

186 reference collection, and the reference charcoal created from shrub wood in this study was generally of good quality. However, most shrubs have small-diameter branches (approximately 1–5 cm), and they may have been more likely to completely combust. Archaeobotanists generally do not focus overmuch on the absence of taxa because charcoal production can vary considerably depending on combustion conditions and properties of the wood. However, it is important to consider these issues when expected taxa are consistently absent. This review stressed the need to consider what may be missing from an assemblage due to certain properties of the wood and the qualities of reference charcoal. Experimental studies, including the creation of a charred reference collection, can assist greatly in this endeavour and the experimental work of C. Orliac (1991, 2003; Orliac and Wattez 1989) provided some important information on the subject. It may be obvious that woods which produce light and soft charcoal will be under-represented in an assemblage but, as noted for Artocarpus, this is not always the case. Similar concerns were expressed regarding shrub charcoal, though it has been found frequently in assemblages from other East Polynesian locations (e.g., Allen and Murakami 1999; Coil 2004). The discovery of many shrubs and small trees in some Rapa Nui assemblages in particular (C. Orliac 2000) could suggest that these materials are frequent in charcoal assemblages when they are dominant vegetation elements, or when no other fuels are available.

Variability in recovery methods Researchers examining faunal materials, artefacts, and lithics have long examined material that had been excavated prior to formulation of their research questions. Botanical data are more rarely studied post-facto, though these materials have potential to inform on many different aspects of past vegetation and human activity. Archaeobotanists and anthracologists argue that collection schemes should be consistent in both sampling scheme and recovery method, and often a full flotation programme is mandated. It is further implied in the literature that research be designed before any field work begins, and that the same person will direct and often carry out the archaeobotanical study. But what of archived materials? And, is absolute consistency in recovery methods really a requirement? In this discussion, I argue that materials recovered by different methods and those taken from archives can be appropriate to address certain types of research questions when it can be determined that biases introduced by variation in recovery methods are minimal.

187

The first concern is regarding variability in recovery methods, as some methods may have introduced excessive fragmentation. A blanket field sampling scheme was used to collect much of the charcoal used in this study, and the majority of charcoal (~70% samples) was recovered from screens during excavations. This recovery method generated a large quantity of material. Exceptions were made when sediments were too compacted for screening, as was some of the Pahumano site, or when many fragile bones were encountered, as with the lower layer at Hakaea. Additionally, samples from each strata at Teavau’ua South were processed by water flotation in the field. Other material was collected in situ as bulk radiocarbon or as bulk sediment samples. The effects of various recovery methods on rates of fragmentation were previously analysed (Table 6.9). To recap those findings, flotation samples tended to be smaller and lighter, but there were only modest differences in fragmentation rates between dry screened and water floated charcoal. In situ samples were somewhat more variable in size and rate of fragmentation than other methods, which may have introduced a modest bias to those samples, but almost half were collected from within feature contexts and those are to be grouped separately for analysis. In addition to excessive fragmentation, there was some concern that different recovery methods may have influenced taxonomic richness values. The literature frequently states that flotation samples tend to recover larger, and ostensibly richer, charcoal assemblages than dry screening methods. It was noted, however, that most dry screened samples in this study produced a rich variety of taxa and range of carbonised plant parts. To test this assumption, the results recovered by both methods were examined from Teavau’ua South. As the entire occupation at this site spanned a period of only around 100 years, and neither stratum contained evidence of widespread burning, large differences in taxonomic richness between contexts at this location were not expected. A review of richness values from this site did show that most dry screened samples recovered 3–5 taxa regardless of screen size, while float samples produced very similar results (2–5 taxa). The difference in these values was small, suggesting that recovery methods had little impact on richness values. One of the flotation samples from this site, however, produced the only Erythrina found in this study, suggesting that more of this especially light and soft charcoal may have been badly abraded or destroyed in screening. To conclude, it does not appear that the various recovery methods used to collect samples in this study have unduly biased rates of fragmentation or taxonomic richness, indicating charcoal recovered from screens using a blanket sampling strategy can be as informative as that recovered by other methods. Because water flotation did recover one

188 fragment of a rare taxon, field flotation should be a component of the sampling strategy when it is feasible. These findings further suggest that point- and bulk-sampled charcoal is also informative, and these samples should not be rejected out-of-hand.

Completeness of reference collection Biases that may have been introduced by limitations of the reference collection were considered next. To assess the diversity of the woody Marquesan flora, a list of 122 trees, shrubs, and some of the woody lianas of the Marquesas was compiled from the botanical references previously discussed. While the list is probably not definitive, it was useful consider the completeness of the reference collection. Many of the plants on this inventory were collected for the present study or they were already part of the University’s wood reference collection. Consideration was also given to materials where anatomical references could be obtained from wood atlases or online databases used by wood anatomists. When considered together, these references were more comprehensive than expected. At the generic level, references could not be located for only about twelve genera, many of which were shrubs or lianas that were rare or difficult to collect. When searching for wood anatomy references in publications and databases, it was noted that shrub woods have not been frequently studied, though there were some exceptions for the Hawaiian flora in the works of Sherwin Carlquist (e.g., Carlquist 2001) and Charles Lamoureux (1985). It was noted that some genera encompass multiple species in the Marquesas. In some cases, samples from various species were available from other reference collections in the departmental archives. These taxa included Terminalia, Syzygium, Glochidion, Macaranga, Premna, Sophora, and Weinmannia, and all were sectioned for study. Wood variations within genera were usually not very great, but some anatomical differences were noted and catalogued in the project database. The reference collection also included redundant samples of several important taxa such as Artocarpus altilis, Hibiscus tiliaceus, and Thespesia populnea among others. Having this depth in the reference collection provided important data on variations in anatomy, with some features measuring outside of ranges suggested by Détienne and colleagues (1999). This variability will be considered in more detail in the following section. An important component of the reference collection was the inclusion of several specimens of twig wood from Aleurites moluccana, Artocarpus altilis, Casuarina

189 equisetifolia, and Thespesia populnea. These references were important as studies of juvenile wood are very limited in the wood anatomy literature. Upon examination of the thin sections, it was noted that some minute features used in identification, notably intervessel pit sizes, were consistent with those measured in specimens of mature wood. This finding provided confidence in identifying material when other anatomical features were not particularly distinctive in the charcoal. For example, it was especially important to understand that candlenut (Aleurites) twigs have very large intervessel pits, similar to those found in the mature wood. Because this anatomical feature was a rarity in the reference collection, and the wood of other potential matches was unlike candlenut wood in other respects, this finding provided assurance that there were very few fragments of candlenut wood recovered in the study. Overall, the reference collection and source materials used in this study were extensive. This evaluation suggested that some of the distinct, but undetermined, taxa found in this study probably derive from trees and shrubs that are rare or extinct in the archipelago today. Some may also derive from woods that make indistinct charcoal, which is the next consideration in this review.

Identifiability of woods In the final part of this methodological discussion, I review biases that were introduced by challenges of taxonomic identification. Some materials were very readily identified at lower magnifications, but for many taxa, the finer anatomical characteristics were examined under high magnification. Even when using an SEM, it was noted that some materials were too similar to make a secure assignment to species, sometimes even to genus. Certain taxa were difficult to fracture which made identifications more challenging. A wide variation in anatomy was observed in some taxa, not only in the reference collection but also in the archaeological material. In some cases, the anatomy of charcoal was not distinctive. Each of these concerns is discussed in turn. First, the ease (or difficulty) with which some taxa could be identified may have introduced some bias in favour of (or against) those species. In particular, the confidence with which very small amounts of material from important economic trees were identified is explicitly described. Wood taxa and nut shells that were very easy to identify are listed in Table 6.13. For brevity, many native species were not selected to be part of this discussion

190 and only key features are discussed here. Fuller anatomical descriptions of reference material can be found in Appendix B.

Table 6.13: Readily identifiable wood taxa found in this study. Preservation as observed in Taxon Key anatomical elements archaeological charcoal specimens Aleurites moluccana Fragments very hard, when Very good preservation of endocarp fractured displays thick-walled structures, fractures cleanly. cells, seed coat (exocarp) has distinctive morphology.

Aleurites moluccana Very large intervessel pits, (Very rare in this assemblage) wood noted even in twig reference.

Artocarpus altilis Large, short vessels with Often good to very good lozenges of axial parenchyma, preservation of features, but ray cells small, round, rays charcoal is usually not dense mostly evenly spaced. and crumbles easily.

Calophyllum inophyllum Large vessels, apotracheal axial Good preservation of parenchyma in long bands; structures. mostly uniseriate rays.

Casuarina equisetifolia Very thick fibres, isolated Charcoal usually very hard. vessels in dendric pattern.

Cocos nucifera endocarp Shiny surface when fractured, Usually very good containing small, isolated preservation of structures, vascular bundles visible under often fractures easily and low magnification. cleanly.

Cocos nucifera ‘wood’ Large vascular bundles with Usually firm and sometimes diffuse, round parenchyma. very hard.

Erythrina variegata Vessels large, wide bands of Very soft and spongy. axial parenchyma with mostly fusiform cells.

Inocarpus fagifer Highly storied uniseriate rays; Relatively good preservation rays very short. of structures.

Pandanus drupe Smooth, concave interior Charcoal is usually firm and surfaces, distinctive exterior difficult to snap. morphology, distinctive cross section.

191

Preservation as observed in Taxon Key anatomical elements archaeological charcoal specimens Sapindus saponaria Axial parenchyma banded, Often very good quality paratracheal, can be seen charcoal, though many unaided; homogenous rays. fragments display sinuous rays and very occasionally have wide non-parenchymous areas.

Thespesia populnea Axial parenchyma diffuse-in- Charcoal usually of very good aggregate; tangential quality, being firm, clean and longitudinal section overall often easy to fracture; pattern very distinctive having fissuring, if present, typically many fusiform or 2-cell long occurs along chains of axial parenchyma. parenchyma.

The most abundant and frequently occurring taxa in this assemblage, Thespesia populnea and Sapindus saponaria, were also the most readily identifiable as the woods have a distinctive combination of anatomical features. Several important economic trees including Artocarpus, Calophyllum, Casuarina, Erythrina, Inocarpus, also have distinctive anatomy and these materials could usually be readily identified. Barringtonia wood has distinctive, large intervessel pits and other features that made it relatively easy to identify, but it was very infrequent in this assemblage. Coconut (Cocos), candlenut (Aleurites), and Pandanus shells fracture in predictable ways and have distinctive cross sections that are unlike other nut shells in the reference collection (such as Cordia and Barringtonia). Wood from the large palms, including taxa such as coconut and Pandanus, are also distinctive from the stems of other woody monocots such as Cordyline, because they typically have very large vascular bundles. Coconut wood charcoal is often hard and fractures in a distinctive way, splintering when snapped. The stems of coconut palms have been well-documented and display a wide range of variation based on location within the tree and age of the wood (Butterfield and Meylan 1980, 48–55). Pandanus wood charcoal is somewhat similar in structure but usually of a poorer quality and in this assemblage it occurred more rarely. There are other tall woody palms native to the Marquesas, including Pelagodoxa henryana, for which reference specimens could not be secured due to the difficulty of sampling endangered plants. This could be a cause for some concern if now- rare palms such as this were once common in lowland settings, as the anatomy may be similar to coconut or Pandanus.

192

Other woods have less distinctive anatomical features, or features that are typically destroyed or obscured by charring, and these materials were probably under-represented or altogether absent in this study. One previously mentioned concern is that many tropical trees have diffuse-porous wood. When other distinctive features of gross anatomy are lacking, it then becomes necessary to rely on minute features for identification and these can be obscured or obliterated when wood is charred. As a result, several important taxa in this assemblage could only be considered comparable, and not assuredly matched, to reference specimens. These include economic trees such as Morinda citrifolia (noni) and Terminalia spp. (tropical almond), the latter of which also has anatomical features that do not favour preservation. Reference charcoal for Morinda, a Polynesian-introduced Rubiaceae, was soft and spongy. This wood features raphide bundles that are well preserved in situ in reference charcoal, and frequent vessel-ray pits that are easy to locate, having very open lumens with scant vesturing. In archaeological specimens, raphide bundles appear to have disintegrated and vessel-ray pit conditions are variable, often being blown out or closed. Terminalia produced firm charcoal, and druses in axial parenchyma were well-preserved and readily seen in reference charcoal. However in archaeological specimens, these crystals often have disintegrated leaving large channels that can cause the wood to become especially fragile. Some of the native shrubs also fall into this category, including a number of taxa in the Rubiaceae family. The Rubiaceae are a large family in this region, and the challenges of distinguishing between some of the genera are important to consider. Wood anatomists have defined two major anatomical types in this family (Jansen et al. 2002). Type 1, which includes taxa found in this region such as Neonauclea, Ixora, Canthium, Psydrax, Coffea, Gardenia, has mostly solitary vessels, narrow rays, and some feature very small-diameter vessels. Type 2 includes Psychotria and Guettarda, a tribe with wide anatomical variations, little axial parenchyma, vessels in short multiples, and slightly wider rays. Within each type, some genera share very similar characteristics and it was noted that the reference materials of some species were very similar when charred. The reference collection contained several different specimens each of Cyclophyllum and Psydrax, for example, with some overlapping ranges of variation noted in vessel size and arrangements, and ray size and composition. Another example is Gardenia, which has mostly small to medium sized mainly isolated vessels, thick fibres, and little to no axial parenchyma. These features are common to several other Type 1 Rubiaceae, thus no distinctive combination of features could be devised to securely distinguish this taxon. In many cases, I found these archaeological

193 specimens very challenging to distinguish and certain taxa in this family were more frequently assigned the comparable (cf.) designation. The presence of woody monocot stems from plants that were not coconut, Pandanus, or Cordyline were also encountered, usually occurring in small quantities. These materials were determined not to be parts of the coconut inflorescence, leaf petiole, or rachis, which were included in the reference collection. They could represent any number of herbaceous plants. Monocot stems, other than from the large palms, were not part of the reference collection and Rod Wallace advised that many are not distinctive. Various types were observed, but very little can be said about these materials. A number of samples also contained amorphous lumps of parenchymous tissue. These were usually small in size, under 2 or 3 mm, and not distinctive in shape or anatomical features. One sample, #5721, contained an approximately 7 mm fragment of a fissured storage organ. This may be a small fragment of Cordyline rhizome tissue, but external morphology was indistinct and there would be little to differentiate it from the storage organs of a number of other plants. Lastly, during the course of this analysis there were indications that several common species appeared to have a wide and undocumented range of anatomical variation, as previously discussed. This is not surprising as the most well-documented tropical woods are those of interest to the timber industry, and in the Marquesas many are modern introductions (see Butaud, Gérard, and Guibal 2008). Variability in wood, as previously discussed, can be influenced by many factors including variation in climate, nutrient availability, water, and more. It is possible that some of the numbered, undetermined taxa found in this study may not be new taxa, and instead may represent intra-specific variations of taxa that are not well documented. One example is Hibiscus tiliaceus. The reference collection included seven specimens of this wood. They displayed a wide range of vessel sizes and arrangements, and ray storying was variable, the latter of which was an important feature used in identification. These findings coincide with the aforementioned observation by anatomists that Hibiscus tiliaceus wood can be quite variable (Webber 1934). A pertinent related concern is that this variability overlaps with that of other species of Hibiscus present in the study area. It therefore cannot be assumed that all Hibiscus charcoal identified in this study derives from H. tiliaceus.

194

Another wood that displayed some anatomical variability is Sapindus saponaria. This taxon has several distinctive anatomical characteristics, and it is unlike other Sapindaceae that occur in this region. On occasion, several interesting anomalies were observed, including wide non-parenchymous banding on fragments that otherwise had attributes very typical of the species. The final example is that of Terminalia, which has one native and one introduced species in the Marquesas. In various other references, including printed references and an online wood anatomy database, it was noted that features such as vessel size and width of axial parenchyma banding could overlap between these species to a greater degree than was present in the reference collection. Based on these findings, I have chosen to identify Terminalia only to the generic level, as even though T. glabrata is native, it is uncertain whether T. catappa was a pre- or post-contact introduction to the islands (Table 3.1). While wood anatomists frequently study more than one specimen when profiling wood anatomy for publication, the difficulty of collecting and the time required to prepare and catalogue multiple thin sections is also acknowledged. Creating taxonomic redundancy in the reference collection is a labour-intensive task, and these challenges are compounded for distant study locations. It does appear to be possible to discern some variability through the examination of large quantities of archaeological material, though this is no substitute for a comprehensive reference collection.

Summary Both qualitative and quantitative data results were presented in the preceding analysis. A large percentage of the charcoal in this assemblage was identifiable, and an additional amount derived from indeterminate wood and other plant parts. At least 59 taxa were identified in this assemblage. Each tree or shrub identified was profiled with a description of the plant habit, typical ecological zone, and status in the Marquesas Islands. Sample adequacy was evaluated using two methods routinely used by anthracologists, and it was determined that charcoal counts of between approximately 250– 450 fragments are required to provide adequate sample richness to reconstruct past vegetation communities in coastal and lowland Marquesan settings. The most reliable data were obtained from samples at the high end of this range, though a more thorough evaluation of target values should be performed with data from modern native vegetation surveys from the region. It was also established that consideration must be given to taxa that

195 were expected but not found, especially in regards to the anatomical characteristics of those woods, as major forest elements may be missing even though fragment counts appear to be adequate. A majority of the assemblages were found to be evenly fragmented. Recovery method has had only a modest influence on rates of fragmentation. It was demonstrated that charcoal fragmentation rates were similar for the most commonly occurring wood taxa and coconut endocarp in this assemblage, but some light and more easily fragmented woods such as breadfruit would be under-represented if weights are used as a basic unit of measurement. Counts were thus selected as the most appropriate for quantitative analysis of the results. It was also demonstrated that feature contexts should be evaluated separately from non-feature contexts, in agreement with an important tenet of the anthracological method (Chabal et al. 1999). Sample sizes in this study varied, and it was demonstrated that these variations do sometimes influence relative abundance values. For this reason, when comparing assemblages, emphasis will be placed on low-level quantification measures. Various taphonomic, recovery, and analytical processes that have affected the assemblage were discussed in some detail. It was argued that charcoal from archives and samples collected before a specific research question has been developed can be very informative, as the assemblages used in this study have proved, as long as materials were collected systematically during excavation. It was found that the reference collection used in this study was extensive, and difficulty in performing taxonomic identifications was probably not related to a lack of reference material. Finally, factors such as differential preservation and identifiability of taxa were considered, as these may have influenced results. While some taxa were probably under-represented in the study due inherently poor preservation qualities, it was noted that important work remains to be done to enhance our understanding processes of site formation and charcoal taphonomy in Pacific Island settings.

196

Chapter 7: Results by Study Site

In this chapter The methods and principles of anthracology have been presented, and the temporal, geographic and cultural context of this study has been established. Methodological challenges to identifying charcoal from the tropical Pacific Islands were evaluated. Aspects of sampling, both in the field and laboratory, have also been reviewed. Wood and nutshell charcoal was identified from numerous functional and temporal contexts, and a rigorous data quality assessment was performed demonstrating this material is suitable to address the main research questions of the thesis. The results for each study site are presented and reviewed in this chapter. To conclude, I quantify two important temporal trends that were observed. Results are presented in two parts. First are the relative abundances within a given strata or temporal zone. These data are based on absolute fragment counts, presented in Appendix A, which were calculated by dividing the total quantity fragments of a given taxa by the total quantity of material found in that stratum. Features have been grouped separately. Values were graphed using psimpoll version 4.27, by K.D. Bennett of Queen’s University Belfast, a program for the plotting and analysis of palaeoecological data. Note that in the graphs presented, bar heights vary in relation to sample size (i.e., tall bars indicate large sample sizes, and vice versa). Vegetation changes were evaluated through examination of temporal shifts in taxa, plant form (trees, shrubs), and status (native, introduced). These data informed on the characteristics of local forests as they were transformed into anthropogenic environments over time. Cultural uses of various plant parts were included in this review to consider how materials may have been introduced into the assemblages. The second component presented for each site is historical plant geographies, also known as phytogeography (Mueller-Dombois and Fosberg 1998, 3), which are used to consider potential changes in ecological conditions over time. Life form and functional attributes of the taxa are discussed, including aspects of plant habitat, succession status, and related plant associates found in the region. Only the key cultural strata were examined in this analysis, though in some cases other strata were considered if they presented new or notable findings. Taxa were coded per the most common habitats where found today in East Polynesian settings (see Decker 1970, 1991; Mueller-Dombois and Fosberg 1998). Many lowland zones are today composed of a mosaic of historically introduced plants and natives,

197 as discussed in Chapter Three, but it is evident that native trees and shrubs were common at lower elevations in the past. Attempts to interpret these results using modern-day observations of plant communities were sometimes challenging, and Rock’s (1913) volume on the indigenous trees of Hawaii was very useful. This reference contains botanically- informed historic observations from an archipelago that shares many of the Marquesan native flora, an exhaustive comparison of which was performed by Decker (1970).

Hakaea As discussed earlier, Hakaea is a small, narrow, elongate valley with a somewhat dry local climate. The beach site represents a very early period of Marquesan occupation; cultural materials found here have been dated to between the 12th and 15th centuries (Allen and McAlister 2010). The site contained three important cultural layers (III, V, VII), with the lower stratum reflecting more modest cultural activity and the uppermost being more pronounced and widespread. Two additional strata contained more ephemeral evidence of cultural activities (IV) or none (VI). Layer IV consisted of small-scale activities that occurred during buildup of the dune. One unit (TP-6) contained lower strata (V–VII) with a boundary that could not be distinguished during excavation. Upon comparing plant taxa identified in the sample from this context (#5721) with those from other levels in the unit, it was determined this sample strongly resembled material from VII and these materials were grouped with layer VII for analysis. Few features were encountered, and they included several hearths, ovens, and post-moulds. Large quantities of charcoal were analysed from the major cultural strata, with over 1000 fragments coming from layer VII, at this site. In the relative abundance chart that follows (Figure 7.1), data are presented in descending order from most recent to oldest temporal context. Richness values (provided in the rightmost column of Figure 7.1) in this assemblage were higher than those of the other study locations. Many native species were identified, and the lowest cultural layer (VII) and layer below (VIII) contain a number of native taxa that do not occur later in the sequence. Though the finding is interesting, the comparably larger sample size from layer VII could have also influenced this result.

198

cf. Alphitoniacf. BarringtoniaCeltis marquesensis cf.pacifica Celtis asiaticaCerbera pacificaCordia manghascf. subcordata Crossostyliscf. CyclophyllumGuettarda bifloracf. Guettarda speciosabarbatumPandanuscf. speciosa Pandanus tectoriusPouteriacf. tectorius Pouteriasp.Sapindus drupe sp. saponaria cf. SapindusSapindusTerminalia saponaria saponariacf. Terminalia sp.Thespesia root cf.sp. populneaThespesiaXylosmaCoprosma populnea sauveolenscf. Coprosma spp.cf. DodonaeaMaytenus spp.Melicope viscosa crenatacf. Melicope spp.(shrub)Phyllanthuscf. spp. Phyllanthuscf. cf. Pipturus marchionicusPremna marchionicus cf.sp. serratifoliaPremnacf. Psydrax serratifoliaWikstroemia unknown13odorata cf. (shrub) coriacea (Sapotaceae)cf. unknown13 (Sapotaceae)Layer

        III

               IV

    IV features

             V

    V features

            VI

                            VII

   VII features

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 20 30 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 20 0 0

Aleurites moluccana endocarp Artocarpuscf. Artocarpus altilisCalophyllumcf. altilisCalophyllumCocos inophyllumcf. nucifera Cocos inophyllumCocos nucifera nucifera endocarp unknown18cf. unknown18 unknown25unknown26 cf. unknown26unknown27unknown28 unknown29 unknown30 unknown31 unknown32 indeterminate indeterminateindeterminate indeterminate barkindeterminate endocarpindeterminate parenchymousindeterminate cf. parenchymous unknown tissue angiosperm tissueindeterminate indeterminateindeterminate cf. angiosperm cf.unidentifiable angiosperm monocot rootFragment ct.Taxonomic richnessLayer

            249 14 III

                  301 22 IV

      73 8 IV features

             197 18 V

        272 8 V features

            167 18 VI

                    1083 29 VII

   31 5 VII features

0 10 20 30 40 50 0 0 0 0 0 0 0 10 20 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 20 30 0 0 0 0 10 20

This assemblage is mostly composed of wood and nutshells from medium to large trees, with some contributions from plants that occur as either small trees or shrubs. It is notable that some forest elements persisted in this area throughout the sequence. The lowest cultural layer includes a large proportion of coconut shell and the wood of a small tree or shrub (Sapotaceae, Unknown13), while the uppermost cultural strata is dominated by charcoal from trees. A few true shrubs were identified, occurring in low abundances in the lower layers and declining to very low levels (or are absent) by layer III. Two taxa that were frequently encountered in this study, in addition to coconut, were Sapindus saponaria and Thespesia populnea. Abundances of this material varied per site and strata, and temporal trends will be reviewed for each assemblage. At the Hakaea Beach site, the trends are noteworthy. Sapindus was present in low abundances until layer III, where it then increased dramatically. This could indicate it became more abundant in the area later in time, however it may also have become a favoured fuel when other options, notably Unknown13 discussed below, had been exhausted. Sapindus did have documented uses as a preferred fuelwood and the charcoal was also abundant in prehistoric earth ovens in Anaho (Huebert 2008, 2009). It was also very common in some valleys in the late 18th century (Crook 2007, 76, 97), and it was considered one of the principal timbers in the early historic period (Thomson 1980, 18). Thespesia was present occasionally, in low abundances, in the lower strata. In the upper layers it then accounted for about 10% of charcoal in layer IV, and somewhat less in layer III, where it occurred in both feature and non-feature contexts. These native coastal trees produced a valued reddish timber with numerous uses, and they were ritually important to the Marquesans (Brown 1935, 178). Thespesia appears to also have been an important fuelwood for earth ovens (Huebert, Allen, and Wallace 2010). Few ovens were found in units opened at this site, however, and none contained this taxon. Overall, the trend for Thespesia probably reflects the depletion of naturally-occurring coastal individuals and later re-establishment of a valuable and culturally important tree. A shift in resource use may be reflected in this trend, from initial selection as an expedient (and good quality) fuelwood to later encouragement or cultivation of timber. In 2010, both trees were encountered frequently within a few hundred metres of the site. A small amount of Guettarda, the beach Gardenia, was identified in layer VII and some features of IV. This is a native coastal tree with large flowers and somewhat dense wood (Whistler, pers. comm. 2009). Though not common in any of the inhabited north coast valleys today, it is known to form gregarious coastal groupings in the region (Butaud,

201

Gérard, and Guibal 2008, 130–1). Though Guettarda were planted around houses in Tahiti in the early contact period (G. Forster 1777, 270), no use or significance in traditional Marquesan culture was noted. These trees, along with Thespesia, Cordia, and Terminalia, probably occurred naturally along the shore and in lowland areas near the coast. Several large and culturally important trees were found in this assemblage. Alphitonia, only tentatively identified, was found in one early sample. This tall, native tree occurs inland in remnant forest patches today. The wood was a valued a canoe timber (Brown 1935, 168). Cordia, primarily a coastal tree, was found in the earlier contexts at this site (V, VII) but not in the later occupation. This native was one of the principal timber trees of the archipelago, it had a number of other important uses, and it also featured in some Polynesian legends (Whistler 2009, 83). Terminalia is another taxon that does not occur in this assemblage after the earliest context. It produces edible nuts that resemble an almond, and the timber was used to make common canoes (Brown 1935, 194–5). Though the range of the native species (T. glabrata) in the past is not well understood, its natural occurrence in coastal areas is probable as T. catappa is a common coastal tree throughout the Pacific. Calophyllum is another large tree found mostly in lowland an coastal locations. Though it has the means to be native, some botanists consider it to be a Polynesian introduction to the Marquesas (Butaud, Gérard, and Guibal 2008; Florence 2004). The wood of this tree was also an important timber used to make canoes, houseposts, and craft items (E. Handy 1923). Early contact-era accounts indicate the trees were subject to tapu restrictions and it was generally forbidden to fell them (Lisiansky 1814, 91). In this assemblage, Calophyllum was only found in the earliest cultural layer, and in later discussion I will return to this finding as the trends at this and other sites seem to indicate it is native. Overall, the disappearance of these important timbers in the uppermost cultural layer at this site suggest a number of large trees were eventually removed from the area. Several other native taxa made small but notable contributions to the assemblage. Celtis and Pandanus, though not abundant, were present in each major cultural strata. Maytenus was also present in many strata, though it did not account for more than 5–6% of the totals. No cultural uses of this taxon were noted. While the former are found in lowland situations today, Pandanus occurs in a wide range of habitats at present and it was observed on cliff faces and high ridges, as well as in lowland thickets, in many different settings. Pandanus trees have numerous cultural uses and the fleshy drupes were eaten (Whistler 2009, 158–67). Barringtonia, a common coastal tree, was tentatively identified only in layer

202

IV. The fruits of this tree were used to create a narcotic to stun fish, and the wood may have had other minor uses (Whistler 2009, 42–44). There were also a number of unknown taxa in this assemblage. While many made a small contribution, perhaps the most interesting was Unknown13, identified as a wood from the Sapotaceae family, probably a species of Sideroxylon (syn. Nesoluma in this region). Sideroxylon was an important fuelwood on Rapa in the early 20th century (Brown 1935, 224), though little is known about its uses or importance in other parts of East Polynesia (Rock 1913, 381). Charcoal from this taxon was the largest contributor to layer VII material, and considering the large sample size this finding suggests it was an important fuelwood. It was also present in the features of this layer, as well as in the underlying deposits. The wood of Unknown13 was found in lesser percentages in subsequent layers, and notably it did not occur in any sample from the uppermost layer (III). This could indicate that supplies of a favoured resource were depleted, a topic that will be discussed in the next chapter. Planchonella was another Sapotaceae identified in this assemblage. It occurred mainly in non-feature contexts, making a small but regular contribution to the total. Both of the aforementioned taxa displayed distinctive characteristics of Sapotaceae wood (Kukachka 1982, and related publications), but fragments sometimes displayed variations (marked cf.) that could indicate additional taxa from this family are present, or that these species have a range of undocumented anatomical variation. These trees have no noted cultural uses in the Marquesas, but some members of the Sapotaceae family have wood that is used to make posts and carved objects in Samoa (Whistler 2004, 156–63). Other natives were only occasional occurrences, and many were only present in layers VII and sometimes V. Premna is a common plant in Marquesan housegardens, and it has a number of traditional medicinal and decorative uses. The wood of this taxon occurred in layer VII, and perhaps in the uppermost layers as well, though it was not frequent in any of these contexts. Its presence in the lowest cultural layer could indicate it was simply a component of the local vegetation, one of few secondary native forest elements encountered in this study. Terminalia also made a very small contribution to the assemblage in layer VII. These are common coastal trees, though the indigenous species (T. glabrata) can also be found in rocky areas several hundred meters above sea level, and around important inland cultural centres (Butaud, Gérard, and Guibal 2008, 269). The wood of Terminalia was used for canoe construction and crafts in the Marquesas (Lisiansky 1814, 90–1; Robarts 1974, 246). The absence of Terminalia may not be remarkable because the wood does not form durable charcoal, as was discussed in Chapter Six. Dodonaea occurs in low abundances in

203 layers VII and V. The wood of this shrub, which is extremely hard, was a component of the traditional fireplough (Brown 1935, 163), but few other cultural uses were noted. Polynesian-introduced taxa in the Hakaea assemblage are not diverse. Coconut shells were large contributors to each layer, indicating they were commonly used as fuel. While coconut trees were probably present in the archipelago before human arrival, varieties with large fruit and higher water content were probably introduced by settlers (Baudouin and Lebrun 2008; Gunn, Baudouin, and Olsen 2011). Coconut wood accounted for a small proportion of this assemblage. Coconut trees are today, and probably were in the past, frequent in lowland locations throughout the archipelago. Aleurites (candlenut) shells were a small but regular part of this assemblage, and a large cache was recovered from an oven in layer V. The kernels were commonly burned for illumination and were used to create tattoo ink (Robarts 1974, 248). Candlenut trees are rare in coastal settings (Elevitch 2006), and the persistent presence of candlenut shells in almost all contexts in this assemblage probably indicates they were transported into the site regularly. Breadfruit (Artocarpus), in addition to producing the main staple carbohydrate in the Marquesan diet, had a variety of other uses. Bark was used to produce tapa cloth for casual wear, the wood was used for small canoes, and the sticky sap had medicinal and utilitarian uses (E. Handy 1923). A small amount of this wood was found in a sample from layer VI, though it was in poor condition and may have originated in the strata above as this was a culturally sterile deposit. Breadfruit wood, then, does not occur with certainty in this assemblage until layer V, where it is present in several non-feature contexts as a little over 5% of the assemblage. In subsequent layers (IV, III), proportions are similar. In layer IV it also occurs in a shallow pit feature eclipsed in excavation (Efe. 4), which was approximately 5 cm thick and more than 25 cm across, containing charcoal and shell. Together these findings suggest breadfruit trees were infrequent, or perhaps not present, in and around the area until the 14th century AD. While rich in taxa and very informative, this was a challenging assemblage to identify. There were more distinct unknown taxa in theses samples than any other study site. These materials may derive from native trees or shrubs that produce indistinct charcoal, or perhaps those that are undocumented. Some material classified in this category may represent ranges of variation in already-identified species. Finally, in the main cultural strata, 13 to 27% of the material could only be classified as charcoal from hardwoods, and 5

204 to 10% of the materials in these strata were in very poor condition and were deemed unidentifiable.

Vegetation pattern assessment The Hakaea Beach site lies in the coastal strand zone of a small valley positioned near the western margin of windward Nuku Hiva. There are several small fresh water springs in the lower valley. A dry streambed, known to run in a torrent after heavy rain, cuts through the coastal flat to the west. Taxa identified in this assemblage can be found in several ecological zones (Table 7.1), and include components of the transitional lowland and coastal strand vegetation. Cordia and Calophyllum are large arboreal elements of coastal strand vegetation in central Polynesia (see Rock 1913, 3–5), and in this assemblage they occur along with common strand trees including Thespesia, Barringtonia, Guettarda, Premna, and others. These large trees and most of their associates were found in the lower layers, but were not present in any samples from the uppermost cultural layer at this site. Most were probably naturally occurring in this location at the time of the early occupation period in the 12th or 13th century. Cordia was found in most of the earlier strata at this site, but was absent from the upper layer. It is native to this archipelago, had numerous uses in traditional Polynesian society, and produced a valued timber. Large specimens of these trees did persist in some Marquesan valleys well into the historic period; a very large specimen, previously mentioned, was seen by Brown along the shore in Hakaui Valley in the early 20th century. These trees could also still be found in out-of-the way coastal areas in Hawaii around this time, as was previously mentioned. Today, Cordia can occasionally be found in this zone, though it is usually present in landscaped areas such as the western coastal flat in Taiohae, though one large individual was seen in a coastal situation in Anaho. Calophyllum only occurs in the earliest strata at this site. The wood of these trees is very durable and was a favoured canoe and craft timber. They are still found on Nuku Hiva today in some coastal situations—several trees were planted behind the beach at Hakaea in 2010—and these large, spreading trees once formed coastal groves in windward locations in Hawaii (Rock 1913, 5– 7). Overall, findings from the lower cultural strata at this site suggest that a diverse native flora, one that included large native trees, was once present in the strand zone.

205

Table 7.1: Ecological zones of trees and shrubs from the Hakaea Beach site.1 Layer / VII V IV III calAD 2σ range2 1164–1293 1268–1411 1317–1444 Strand Barringtonia asiatica  (cf.) Calophyllum inophyllum  Cocos nucifera     Cordia subcordata    Guettarda speciosa  (cf.)  Thespesia populnea    

Mesic Aleurites moluccana     Alphitonia marquesensis  (cf.) Cerbera manghas    Coprosma spp.    Crossostylis biflora  (cf.) Cyclophyllum barbatum  (cf.) Melicope spp.   Pipturus sp.  (cf.) Planchonella sp.    Premna serratifolia   (cf.) Sapotaceae Unknown13   

Xeric Dodonaea viscosa  (cf.)  (cf.) Sapindus saponaria     Xylosma sauveolens 

Mixed Artocarpus altilis    Celtis pacifica     (cf.) Maytenus crenata     Pandanus tectorius    Phyllanthus cf. marchionicus    Psydrax odorata  (cf.)  (cf.)  (cf.) Terminalia spp. 

1 Strand = Coastal strand; Xeric = Xerotropical lowlands, or perhaps dry seaward slopes; Mesic = Pluviotropical lowlands, mesophytic inland vegetation, cultivated forest remains,

206 and riparian zones; Mixed = Transitional lowland zones. Decker (1970, 231–57) indicates there is no clear demarcation between dry and moist zones in many lowland Marquesan locations, and in the transitional zones types are often intermixed (zones per Mueller- Dombois and Fosberg 1998). 2 Reported by Allen and McAlister (2010).

The co-occurrence of Celtis, and sometimes Xylosma, with Sapindus in all layers suggest that a semi-dry open forest (see Florence 1997, 196–99) persisted in lowland Hakaea locations in prehistory, much the same as today. The co-occurrence of Dodonaea and Psydrax in the two lowest layers, and the persistence of Psydrax in layer III, further support this suggestion (see Mueller-Dombois and Fosberg 1998, 497, regarding Hawaii). Some of these taxa, especially Sapindus, had uses as fuelwood in the early contact period and Dodonaea was used to kindle fires, as previously mentioned. The others had minor craft, decorative, or medicinal uses. None of these woods are likely to have been carried into the site from any great distance, and they were probably components of lowland vegetation. Other taxa, including as Pandanus, Phyllanthus, and Maytenus, could be associated with these drier formations, but they also occur in moist areas. Pandanus is especially common on windswept slopes, but they are common trees in many zones. Pandanus drupe fragments were found in this assemblage, which may have been food processing waste, and the wood was a small but regular component of all cultural layers. While the leaves are an important source of thatching and plaiting material, the wood of these trees has few noted cultural uses, but trunks may have been used as posts (Whistler 2009, 166). Phyllanthus (syn. Glochidion) and Maytenus have few cultural uses. It is likely that these taxa also occurred naturally in the area, and were probably components of a transitional lowland forest, for which Decker (1970, 231–57) includes both dry and moist forest elements. A number of species typically found in mesic formations were also noted. Coprosma, Crossostylis, Cyclophyllum, and Melicope are native trees with few cultural uses and most can be found in remnant native forest patches today with the dominant Metrosideros collina. It is interesting that Metrosideros was not identified in the Hakaea assemblage, especially considering that this wood was easily collected (Ref. coll. MQNH10-17) along the eastern road out of the valley, and the charcoal was found in both Hatiheu and Anaho assemblages. Coprosma is usually found in the Marquesas today in windswept areas or moist montane forests (Wagner and Lorence 2011a), though Rock noted that it was once a common element of lowland moist forests in some areas of Hawaii (Rock 1913, 33–4). Its presence may indicate the plants or wood were transported from moist

207 locations. Though it has no known uses in Marquesan culture, Coprosma was once used as fuelwood on Rapa (Brown 1935, 317). Cyclophyllum is also common in moister areas, and in Samoa can be found from the coast to foothill forests (Whistler 2004, 136). Though few cultural uses for these small trees have been noted, the wood is durable and hard. Melicope is also characteristic of moist-wet forests and has few known uses, but the wood was once favoured in Samoa for posts because of its termite resistance (Whistler 2004, 145). Of the aforementioned taxa, Coprosma persists throughout the sequence but the others mentioned are only present in the lowest cultural layer. This trend is notable when combined with the eventual disappearance of Sapotaceae (both Unknown13 and Planchonella), which usually occurs in mesic lowlands elsewhere in the region (Meyer and Butaud 2009). Overall, the presence of numerous trees typically found in mesic environments in the two lower cultural strata at this site may suggest local conditions were moist early in the sequence. Aleurites was present as nutshell fragments, which were found in many samples and all temporal contexts at this site. These Polynesian-introduced trees thrive in moist, interior locations or along watercourses. I only occasionally observed them growing outside of these zones in the Marquesas. But the trees are known to be drought tolerant (Elevitch 2006), and in Hawaii they could once be found in many environments including arid locations along with Psydrax (Rock 1913, 11). It is likely the nuts (candlenut) were imported from other areas as they have several important cultural uses, as previously noted. Premna is another small tree or shrub found typically in mesic areas, but it is also an element of coastal strand vegetation. It is a common feature in housegardens today and it has a number of medicinal and cosmetic uses. Premna is present in both early and late contexts where it may have been naturally occurring, but there is a possibility it was also cultivated in this setting. Trends for the important cultivated species, such as Artocarpus, and large native trees, such as Alphitonia, Cordia, and Calophyllum, also contribute to this review. While the aforementioned large trees only occur in the lower two cultural strata, Artocarpus (breadfruit) wood occurs only with certainty in the middle and uppermost layers. When considered together, these findings indicate a native forest that included several large trees was replaced by useful economic trees in and around this location in under 200, and probably less than 150, years. It is also evident that coconut trees, perhaps a grove, was present in the area and it persisted throughout the sequence as coconut shells are abundant, and a small amount of coconut wood is present, in all layers. Coconut trees were probably indigenous in this coastal setting, and were then cultivated later in the sequence. There is some evidence that ecological conditions may have been moister at Hakaea Beach

208 early in the sequence, but the influence of a larger sample size from layer VII make this a very tentative suggestion.

Hatiheu Hatiheu is a large, broad, and fertile valley located on the windward north coast of Nuku Hiva. Numerous stone architectural foundations and several large ceremonial complexes are located throughout this valley, and it was once an important population centre. Two assemblages were analysed from Hatiheu: material from an occupation near the coast, called Pahumano-o-te-tai, and material from investigations of inland sites, most of which were located near surface architecture. Assemblages from these sites are smaller than the one from Hakaea Beach.

Pahumano-o-te-tai The Pahumano site lies at the base of an inland ridge terminating near the coast, located in the grounds of the village church. Contexts included in this portion of the study date to between the late 13th or 14th century AD and post-contact period (Allen and McAlister 2013). Pahumano contained three cultural zones: D and F are represented by layers III and V in units against a ridge face, and Zone H is represented by layer III on the coastal flat and layer VII on the ridge face. One additional layer in this area, IV, contained a deep pit feature that could not be readily assigned to a particular strata. Allen noted that repeated episodes of burning, erosion, and colluvial activity were evident throughout the sequence, and concluded that in situ domestic activities were evident in these deposits. Zone H samples are larger than the others, which relates directly to the higher concentration of charcoal encountered in this zone. Most of the features analysed at this location, including several ovens, a possible hearth, and several deep pits, are from the lower zone. Results are presented in Figure 7.2 in descending order, from most recent to oldest temporal context. In this assemblage, taxonomic richness values were modest (6 to 7) with the exception of Zone H, where features contained 16 taxa and occupation debris contained 10. A number of native species occur in the lower zones at this site, but they are then absent from the uppermost cultural layer. This trend may correspond to the larger number of fragments examined, and ultimately to the higher charcoal concentrations in Zone H deposits, but it is still noteworthy as it follows a pattern similar to the Hakaea assemblage.

209

cf. Cordiacf. Cyclophyllum subcordataGuettardacf. speciosa barbatumGuettardaPandanus speciosacf. Pandanus tectoriusPouteria drupeSapindus tectorius sp. saponaria SapindusThespesia saponaria populnea root XylosmaAllophylus sauveolensHibiscus marquesensis spp. cf. HibiscusMaytenus Phyllanthusspp. crenatacf. Premna unknown13cf. marchionicus serratifolia (Sapotaceae) Zone

  D

E features

    H

    H features

        

0 0 0 0 0 5 0 0 5 0 0 5 10 15 20 25 30 0 0 5 10 15 0 0 0 5 10 15 0 0 0 0 0 5 10 15 0

E features D H H features F Zone

7 3 6 10 16 Taxonomic richness Taxonomic

49 11 44 172 200 Fragment ct. Fragment

˜ ˜

unidentifiable

0 5 undetermined 0

 

indeterminate unknown indeterminate

0 indeterminate parenchymous tissue parenchymous indeterminate 0

 indeterminate monocot indeterminate 0

 indeterminate bark indeterminate 0

  indeterminate angiosperm indeterminate 0 5 10 15 20 25 30

 cf. unknown21 cf. 0

  unknown19

Pemphis acidula Pemphis  cf. cf. 0

Cocos nucifera Cocos  endocarp

0 5 10 15 20 25

Cocos nucifera Cocos  cf. cf.

Cocos nucifera Cocos 

Casuarina equisetifolia Casuarina 

Calophyllum inophyllum Calophyllum 

0 Artocarpus altilis Artocarpus  0 5 10 15 20 25 30 35 40 The flora identified were wood and nut shells from trees and small trees / shrubs. The lowest cultural zone (H) contained a significant proportion of the wood of a small tree or shrub (Sapotaceae, Unknown13), while most charcoal in the uppermost zone (D) was from native and introduced trees. The dominant taxa of this entire study, Thespesia and Sapindus, follow trends similar to those noted for Hakaea. Thespesia was found in four features from the lowest cultural zone (H), but only accounted for approximately 7% of feature charcoal, and it did not occur outside of feature contexts in this zone. It is absent in Zone F samples, but was then found more abundantly in the upper zone (D). Sapindus made a very modest contribution to this assemblage until the uppermost zone, where it then accounted for one- third of the assemblage. The modest occurrence of these two woods early in the sequence, and their abundance in the uppermost zone, indicate they could have became important fuelwoods at this location when other sources had been depleted. It may also indicate these trees were encouraged in the area later in the sequence, as was suggested for Thespesia at the Hakaea Beach site. In the case of Sapindus, which tends to grow in drier locations, it may also indicate a shift in collection habits. The importance of Sapindus as a fuelwood persisted well into the 20th century, as ethnographers noted its use in a large specialised type of earth oven (E. Handy 1923, 195), and Decker (1970) noted that Sapindus wood was still in demand by bakers in the 1960s. Several other native taxa were found only in the lowest zone (H), and a number of others were found only in the two lower zones (F, H) but not at all in the upper zone. These include Allophylus, Maytenus, Premna, and Phyllanthus, which are small native trees or shrubs. Allophylus is a small tree or shrub of the Sapindaceae family with no known cultural uses in Polynesia (Whistler 2009, 148). Though these trees are present on some Marquesan islands, they are not found on Nuku Hiva today (Smithsonian National Museum of Natural History n.d.). Maytenus also has no known cultural uses but Premna, as previously discussed, was (and is) a common element of house gardens. Phyllanthus (syn. Glochidion) are endemic, and several species occur in the Marquesas. No cultural uses for these trees were noted, and other members of the genus have only minor uses elsewhere (Whistler 2004, 63–4). Other native trees, including Cyclophyllum, Pandanus, and Xylosma, were only found in the lower zones at this site. Cyclophyllum has few uses, but Pandanus, as previously discussed, had a wide range of uses in traditional Polynesian culture. Most of the Pandanus recovered at this site were drupe fragments, which were abundant in Zone H deposits. Xylosma has no known uses in the Marquesas, but the wood is very hard and

212 informants noted it is a preferred fuelwood today (fieldnotes, 3 July 2011). Finally, Hibiscus charcoal was found in both the lower and upper zones at this site in low abundances. In addition to the sprawling, indigenous Hibiscus tiliaceus, several Polynesian-introduced Hibiscus were present in the Marquesas in prehistory, including a species with large showy flowers (H. rosa-sinensis) and another slender variety favoured for craft and construction (H. tiliaceus subsp. tiliaceus ‘Sterilis’), and their woods cannot be securely distinguished. While it seems likely that at least some of this material was H. tiliaceus, which thrives in disturbed areas and has numerous uses throughout the region (Whistler 2009, 132–3), it could also be one of the other varieties considering the area was a locus of domestic activities. Wood of the coastal tree Guettarda (beach gardenia) was also found in the lowest zone. Overall, most of the aforementioned flora were probably naturally occurring in the area, either in the littoral zone seaward of the site or on the adjacent escarpment. Several large and important trees were also found in this assemblage, though they were present in very low abundances and only in the earliest zone (H). These include Cordia, a native coastal tree, and Calophyllum. Calophyllum, as previously mentioned, is considered a Polynesian introduction to these islands, though in this assemblage the trend again suggests it is native. Cordia is a long-lived coastal tree with numerous uses that can reach large proportions in these locations (Brown 1935, Plate 7C). The disappearance of large trees such as these may indicate they were eventually removed from the area, perhaps to make way for other trees or domestic activities. Once again, however, some consideration must be given to the larger sample size from Zone H. Three unknown taxa were also found at this location, none of which were found in contexts later than Zone F. It is remarkable that Unknown13 (a Sapotaceae) occurs abundantly in the early zones but is absent from the upper zone at this site. Unknown19 and Unknown21 follow a similar pattern, though in lesser abundances, though no family affiliation could be determined. These trends suggest a number of natives were eventually removed from the area. Regarding Unknown13, there is also some indication that as this resource became depleted it was replaced by other timbers such as Sapindus, in a trend similar to that seen in the Hakaea assemblage. Only a small number of Polynesian-introduced taxa were found at this site. Breadfruit charcoal occurs in the lowest zone, though it is not abundant and did not occur in any of the features there. It was not found in the middle zone (F), but then accounts for a large proportion of the assemblage in the upper zone (D). Large breadfruit large trees were noted in and around domestic settings throughout the region by many early-contact

213 observers, and they may have been present in a housegarden. However, breadfruit groves were probably present throughout this lush valley by the late prehistoric period, as will be subsequently discussed. Coconut wood is somewhat abundant in the lower zone, but it is absent in the upper zone. Coconut shell is abundant in the lower two zones, but is much less frequent in the upper zone. Together, these findings indicate that coconut trees – perhaps a cultivated grove or naturally-occurring stands seaward of the site – were present in the area and they may have persisted throughout the sequence. Little can be said about the paucity of coconut in Zone D, considering the small sample size of these materials. Casuarina, another Polynesian introduction, occurs only in the upper zone, were it is not abundant. The very hard, reddish wood (sometimes referred to as ironwood) was used to make war implements and other tools, and the tree was a valued dye plant in Polynesia (see Langsdorff 1813, 103; Whistler 2009, 56–7). Casuarina is a considered a good fuelwood today (fieldnotes 3 July 2011), though this use is not well documented in East Polynesian ethnohistories. These very fast-growing trees are common in littoral zones as they tolerate sandy soils and salt spray, though they can be found in a wide range of habitats (Elevitch 2006), and they were probably planted in the local area. No candlenut (Aleurites) shells were identified in the assemblage from this site. This is notable considering the nuts had several important uses, as discussed, and the trees thrive in gregarious groupings in interior locations of this valley today. It is especially interesting considering candlenut shells were frequent in the Hakaea Beach assemblage. Lastly, woody monocot stem and indistinct parenchyma tissues were found only in Zone H. These materials could have originated from a number of smaller plants that were also burned during vegetation clearance. A modest amount of charcoal (11 to 12%) from D and H could not be determined as other than hardwoods, suggesting that there were some woods in the assemblage that produce charcoal with indistinct characteristics. Only 2 to 6% of material from each stratum was unidentifiable, indicating these samples were in rather good condition, a finding that is interesting considering it was often challenging to separate charcoal from the sticky clay sediments at this site. One final observation is in regards to the deep pit feature found in Zone E, excavation unit SP-7. This feature contained many closely-associated large cobbles and several larger fragments of breadfruit charcoal, as well as some Hibiscus and Thespesia. Together, these taxa are more characteristic of Zone D suggesting that the feature may have originated in that layer.

214

Inland sites The Hatiheu inland sites were excavated by Sidsel Millerstrom (2001). Sites were situated in lowland, mid-valley, and upper valley locations. Samples were assigned to three temporal categories for this evaluation; materials assigned to the older phase were taken from a large earth oven located in the lower valley (Vaiu’ua River), while the middle period sample was taken from an oven feature in the mid-valley near a large ceremonial complex (Mutoka). As previously discussed in Chapter Four, it should be noted that there is some potential for in-built age in the radiocarbon samples for this assemblage. All other materials date to the late prehistoric or early historic period and were collected at several middle (Mutoka) and upper valley (Ototemui) locations near stone house foundations and a ceremonial complex. Some this assemblage was the subject of a pilot charcoal study (Millerstrom and Coil 2008). In the relative abundance chart that follows (Figure 7.3), data are grouped by locality and presented in descending chronological order from most recent to oldest temporal context. It should be noted that most contexts date to the late prehistoric period. In this assemblage, taxonomic richness values were modest with non-feature contexts having 8 to 10 taxa, while features contained fewer than 6 and one contained only a single taxon. A number of native trees and shrubs were identified, along with four Polynesian-introduced trees. Of the two dominant taxa in this study, Sapindus accounted for a large proportion of the charcoal recovered from the early Vaiu’ua oven. The presence of this material in a lower valley context is notable, especially in a moist setting such as this located near a watercourse. The trees would not have been growing in the area, and would have been carried to this location from some distance, perhaps from a slope or ridge. Sapindus accounted for a smaller portion (10%) of material from Mutoka (mid-valley) samples, but did not occur within any features.

215

CrossostylisSapindus biflora saponariaThespesia populnea cf. Thespesiacf. AllophylusAlstonia populneaHibiscus marquesensiscf. costata spp. cf. HibiscusMetrosideros spp.Phyllanthuscf. collinaPsydrax cf. marchionicusArtocarpus odorata (shrub) altilis CalophyllumCocosCocos nuciferainophyllumInocarpus nuciferacf. Inocarpus endocarpfagiferunknown33 fagiferunknown34 indeterminate indeterminateangiospermindeterminate unidentifiablebark monocotFrag ct. Taxonomic richnessLocation

             147 10 Mutoka

      21 6 Mutoka hearth

        95 8 Ototemui

 6 1 Ototemui pit

   22 2 Mutoka oven

        38 5 Vaiu’ua oven

0 0 0 20 0 20 40 0 0 0 0 20 40 60 80 0 0 0 0 0 0 20 0 0 0 0 0 0 0 0 0 0 0 0

Thespesia accounted for a large proportion, up to 45%, of samples from both of the oven contexts in this assemblage, which date to earlier periods of activity in the valley (perhaps as early as the 15th century AD, but see discussion in Chapter Four). This taxon also occurred in Ototemui (upper valley) samples, though it was much more abundant in mid-valley locations where it was the largest contributor to the Mutoka assemblage. Thespesia is typically a coastal tree, but both of these sites are located some distance inland, which makes the presence of this wood notable. The Ototemui test units are located in the upper valley almost one km inland, in an isolated complex of stone foundations and enclosures. The Mutoka test units lie approximately 500 m inland and both units in this location had features that contained Thespesia. It is possible that the trees were planted in this area, as they were culturally valued – the sticks and leaves had ceremonial uses (Brown 1935), and Millerstrom indicated this was a chiefly area with several ceremonial structures located nearby. It is also possible Thespesia may have been collected from lower valley settings and transported there. Other native taxa were encountered, but are not frequent in this assemblage. All are small trees or shrubs. Most are only present in non-feature contexts. One is Allophylus, which occurred in a Mutoka sample. This nondescript plant was also found in the Pahumano assemblage, as previously discussed. Another is Alstonia, also found in a Mutoka sample. This is a shrub or small tree with sticky latex with few noted uses. It is occasionally found in open forests on ridges and in scrub, and in the Society Islands it occurs in forests dominated by Metrosideros (Sidiyasa 1998; Whistler 2004, 23). Metrosideros was found in the samples from Ototemui, located several hundred metres away. These trees had a number of uses in traditional Polynesian culture: flowers were decorative, leaves used medicinally and as thatch, and the wood was also used for craft projects and as fuel (Elevitch 2006). In Hawaii, the trees were considered sacred and the wood was used for important figure carvings (Buck 1957, 466–7), though no similar importance was noted for the Marquesas. The co- occurrence of Metrosideros and Alstonia, then, probably indicate that patches of native forest were present in parts of the upper valley in the late prehistoric period. Some trees found in this assemblage are indicators of disturbance. The most abundant of these is Hibiscus, probably much of which is H. tiliaceus, which was a large percentage of the assemblage from mid- and upper valley locations, including feature contexts. This wood was favoured for the widely used fire plough (E. Handy 1923), and it also had numerous other uses as previously discussed. Phyllanthus, found in an upper valley (Ototemui) sample, is a shrub or small tree with few known uses. In combination with

217

Hibiscus, this plant is characteristic of moist valleys (Wagner and Lorence 2011b). Together, these finds indicate that portions of back-valley forests were disturbed, and Hibiscus tiliaceus was probably abundant, by this period. Several Polynesian-introduced trees were also found in this assemblage. Breadfruit (Artocarpus) wood occurs in all locations and temporal contexts. It is a small proportion of charcoal from the older ovens in mid- and lower valley locations, but it accounts for almost 20% of the assemblage from the Mutoka test units. While seldom mentioned as fuelwood, on Mangareva breadfruit logs were left smouldering to sustain cooking fires (Buck 1938), and some of this material may also be prunings or deadfall. Calophyllum, Inocarpus, and coconut shell occur only in later period samples. Calophyllum wood was only found in Mutoka mid-valley contexts. These are large trees that produced an important timber, as previously discussed, and they occur naturally in coastal areas. The presence of this wood at a high-status inland location could suggest the trees were planted near ceremonial structures in this area, as they often marked culturally important features such as me'ae in the Marquesas (Brown 1935). Inocarpus are very large, buttressed trees that bear fruit resembling a large chestnut. They were used to make a traditional dish of grated nuts mixed with coconut milk that was baked in an earth oven (Thomson 1980, 18–19). The young leaves of this tree were eaten as a green vegetable, and larger branches were used as fuel for large ceremonial ovens (E. Handy 1923, 195, 200). Today, Inocarpus are common in some moist inland locations (Smithsonian National Museum of Natural History n.d.) and the wood was easily collected from the study area. The presence of a small amount of Inocarpus wood in two Mutoka proveniences (one a hearth) in this location are thus not surprising. Though there is little evidence these trees were actively cultivated in the Marquesas at contact, they could have been a component of the wooded fruit tree groves mentioned by G. Forster (2000, 336). Coconut shell and wood make a very small contribution to this assemblage. A very small amount of coconut wood was found at the inland Ototemui site, and a small proportion of coconut shell was found in the Mutoka samples. As these materials are abundant at other sites in this study, this finding may indicate that coconuts were less frequently used and/or burned as fuel at inland locations. Finally, it is also notable that no candlenut shells were identified in this assemblage. The nuts were commonly used for illumination, and they had a number of other uses, as discussed. The trees are also frequent in interior settings today. This finding may suggest candlenut trees were infrequent in these

218 areas in prehistory, or perhaps the nuts were seldom used in the high-status study locations, though both proposals seem unlikely. A final observation was made regarding the pakeho, the large (4 to 5 m deep) square stone-lined pit excavated at Ototemui. While these features have been extensively documented in some Marquesan locations, and variations in size and shape have been described (Kellum 1968), as previously mentioned, little is known about their function. Few cultural materials were found in this pit, and little charcoal was recovered. All identifiable material was Hibiscus, which may have been collected from local vegetation and burned in a fire in the pit or atop the platform. It is also possible that these materials are charred debris from brush fires that naturally accumulated in the pit. From 9 to 22% of the charcoal in these samples were indeterminate hardwoods, and a smaller percentage was unidentifiable, with many of the smaller fragments resulting from prior examination. Several concluding comments are made regarding modifications to the data reported by Millerstrom and Coil (2008). While many identifications were confirmed several revisions were made, new taxa were found, and additional material was processed. The tentative (cf.) identification of Claoxylon could not be confirmed and it is now classified as Unknown34. Bauhinia is a modern introduction and unlikely in this assemblage, and these materials have been classified as Unknown35. Material identified as cf. Weinmannia could not be relocated.

Vegetation pattern assessment

Pahumano The Pahumano assemblage is composed of a number of strand vegetation elements throughout the sequence (Table 7.2). Several large trees, including Calophyllum and Cordia, were found in the lower Zone H, along with Thespesia, Guettarda, and Allophylus, all widespread elements of strand vegetation throughout East Polynesia (Mueller-Dombois and Fosberg 1998, 389). As with Hakaea, these large trees and most of their associates were only found in the lower zone at this site, where they made minor contributions to the assemblage and were thus probably naturally occurring in this zone at the time. Allophylus, a small tree or shrub with no known cultural significance, is the only native strand element that appears to have persisted into the middle temporal zone at this site, and it is not present

219

Table 7.2: Ecological zones of trees and shrubs from the Hatiheu Pahumano-o-te-tai site.1

H F D Zone / 1265–1393 1307–1429 1498–1795 calAD 2σ range2

Strand Allophylus marquesensis   Calophyllum inophyllum  Cocos nucifera    Cordia subcordata  (cf.) Guettarda speciosa  Pemphis acidula  (cf.) Thespesia populnea  

Mesic Cyclophyllum barbatum  (cf.)  (cf.) Hibiscus spp.   (cf.) Planchonella sp.  Premna serratifolia  (cf.) Sapotaceae Unknown13  

Xeric Sapindus saponaria   Xylosma suaveolens 

Mixed Artocarpus altilis   Casuarina equisetifolia  Maytenus crenata  Pandanus tectorius   Phyllanthus marchionicus  

1 Strand = Coastal strand; Xeric = Xerotropical lowlands, or perhaps dry seaward slopes; Mesic = Pluviotropical lowlands, mesophytic inland vegetation, cultivated forest remains, and riparian zones; Mixed = Transitional lowland zones. Decker (1970, 231–57) indicates there is no clear demarcation between dry and moist zones in many lowland Marquesan locations, and in the transitional zones types are often intermixed (zones per Mueller- Dombois and Fosberg 1998). 2 Reported by Allen and McAlister (2013).

220

Table 7.3: Ecological zones of trees and shrubs from the Hatiheu inland sites.1 Mid- Upper Lower valley valley Valley Mid-Valley Location / Vaiu’ua Mutoka Ototemui Mutoka calAD 2σ range2 (oven only) 1296– after 1447 1453–1645 After 1489 After 1676 Strand Allophylus marquesensis  (cf.) Calophyllum inophyllum  Cocos nucifera   Thespesia populnea    

Mesic Alstonia cf. costata  Crossostylis biflora  Hibiscus spp.    Inocarpus fagifer 

Xeric Sapindus saponaria  

Mixed Artocarpus altilis     Metrosideros collina  Phyllanthus cf. marchionicus  Psydrax odorata  (cf.)

221 in the upper zone. Thespesia occurs again in the upper zone, where it is then a larger proportion of the assemblage. This trend could indicate it was re-established in the area, or that it became a preferred resource in the late prehistoric period, as it was a culturally valuable tree and timber, as previously discussed. Overall, these results from Zone H indicate a typical strand flora was once present on the coastal flat in this area. The large coastal trees Cordia and Calophyllum may have been removed from the vicinity by the 15th century. Suggestions regarding temporal trends, however, are tentative as the sample sizes from Zones D and F are small. Only a few xeric taxa were noted in any period in this assemblage. The co- occurrence of Sapindus, Xylosma, and Pandanus in the early zone suggest that semi-dry vegetation formations could be found in the area at this time. Xylosma is a small to medium sized tree with few cultural uses, which can be found in a range of ecological conditions, but it is considered characteristic of exposed slopes and semi-dry formations (Brown 1935, 185–8). This vegetation type was most likely found on the escarpment adjacent to the site, or on the ridge that runs inland behind it. Of these taxa, only Sapindus persisted into the upper zone (D). The absence of Xylosma and Pandanus in these samples could indicate this drier plant formation was not present in the vicinity of the site later in the sequence. Sapindus formed a large portion of the upper zone samples. The wood of this tree did have importance as a preferred fuel in the early historic period and it was probably transported to this location, but perhaps not from any great distance as it could have been abundant along the slopes of the ridge. Other native taxa, including Phyllanthus and Maytenus, could be associated with the transitional lowland vegetation zone. Both only occur in Zone H, and had few cultural uses. Of Phyllanthus, the most common native species (P. marchionicus) is widespread in moist valleys, especially disturbed mesic ridge forests (Florence 1997). Pandanus, present in the two lower zones, may be associated with either drier slope vegetation or the transitional zone. The abundance of charred drupe fragments in the rich charcoal deposits from Zone H along the ridge face suggest these trees were a component of vegetation growing on the escarpment. It was also noted that Allophylus, which is a common small tree or shrub found in strand vegetation at other locations, can be found on Tahuata in relict dry to transitional and moist forests, and also along some high ridges (Smithsonian National Museum of Natural History n.d.). In consideration of this information, Allophylus may have been a component of vegetation in several areas in and around the site; as previously noted, it is not reported on Nuku Hiva today.

222

Several trees and small tree / shrubs characteristic of mesic vegetation zones were also found in the early zone context at this site. They include several natives, Cyclophyllum, Planchonella (Sapotaceae), and Unknown13 (also Sapotaceae), all of which had few known cultural uses. Cyclophyllum is common in coastal and foothill forests of Samoa (Whistler 2004, 136), and Sapotaceae are found in moist lowland formations (Meyer and Butaud 2009). These trees and shrubs would have been naturally occurring in this zone, and they suggest a moist vegetation community was also present in the area. This finding is not surprising, considering one of the permanent streams of Hatiheu runs only a few hundred metres to the west. There are some indications of vegetation disturbance in this assemblage. Hibiscus was found in an early zone feature context, and though it should be noted the exact species cannot be determined, if it is H. tiliaceus its presence may indicate that vegetation was cleared in the area prior to occupation of the site. There is some support for this suggestion with the presence of Xylosma in this zone, which can be a pioneer in cleared areas, and Phyllanthus (syn. Glochidion) which can occur in disturbed forests along with Hibiscus (Wagner and Lorence 2011b). Hibiscus also occurs in the uppermost zone at this site, and though it may have become a dominant vegetation component in some parts of the valley by the late prehistoric period, much as it is today, the wood may derive from any number of uses in domestic settings. Trends for the Polynesian-introduced taxa are notable the Pahumano site. Artocarpus is present in only one sample from the lower zone, but it is abundant in samples from the uppermost zone. Coconut wood is only present in the lower zones, but coconut shell is abundant in most contexts. A small amount of Casuarina occurs in the upper zone. The wood of this taxon had a number of craft uses, as discussed, and may have been introduced to the assemblage as a by-product of these activities, though the trees grow in a wide range of conditions and they may simply have been growing near the site. Together, these trends indicate a plant environment that contained a number of useful trees was eventually formed in this location. Charcoal from the uppermost zone, though a small sample, reflects a markedly anthropogenic flora.

Inland sites Samples from the potentially early lower valley earth oven at Vaiu’ua and the mid- valley oven at Mutoka included material from a variety of ecological zones. One of these

223 taxa was Thespesia, a common component of strand vegetation. This wood was a preferred fuel for earth ovens, and considering the sites are some distance inland, the wood was probably transported to these sites for that purpose. As previously discussed, however, Thespesia may also have been planted in the high-status area of the study site. Sapindus was also present in the Vaiu’ua oven feature, and this wood was probably also transported to this location, as it is characteristic of dry formations and would not have been growing in the immediate vicinity. Several other taxa typically found in moist forests were also identified in this feature, and they were probably all elements of the surrounding vegetation. Artocarpus was found in both ovens. As a poor quality fuel, this wood was probably deadfall or prunings from nearby trees. This finding could suggest breadfruit trees were growing in the mid-valley by the 15th century AD, but the possibility of in-built age in the radiocarbon samples for these features make this a tentative suggestion. Most of the material at these sites dates to the late prehistoric period, and they derived from both mid-valley and upper valley contexts. Samples from mid-valley test units contained a number of strand species, including (cf.) Allophylus, Calophyllum, and Thespesia. Calophyllum and Thespesia may have been transported to these sites from more coastal settings, but they may have been planted in this high-status area, perhaps around the ceremonial structures, as previously discussed. Both woods may have been the by-product of craft production, as they were used to make esteemed carved objects, canoes, or canoe parts. Allophylus, as previously noted, is typically a strand species that can also be found in transitional and moist forests, and may have been a native forest element in this location. Hibiscus and Inocarpus thrive in moist interior locations, and were also probably components of the local vegetation. Hibiscus occurred frequently in samples from this area, suggesting secondary thickets of the trees may have been present in the area by the late prehistoric period. Alstonia is a component of native forest with few known uses, and its presence in a mid-valley sample may indicate at least some patches of native forest persisted there in late prehistory. Sapindus was the only taxa that typically occurs in xeric formations that was found in samples from this area. The wood was probably transported to the sites from dry slopes elsewhere in the valley. Three cultivated taxa including coconut, Inocarpus, and breadfruit were also identified in this assemblage. These trees were probably growing in the middle valley, and could be components of the “close tufted wood of fruit-trees” (G. Forster 2000, 2:336) that were extensive in Marquesan valleys at contact. Further inland at Ototemui, a complex of stone walls, terraces, and foundations that lies near the steeply sloping margins of the valley, taxa that were identified reflect a mosaic

224 of disturbed and native vegetation. Metrosideros, Phyllanthus (syn. Glochidion), and Psydrax are all natives that can be found in a variety of habitats. Phyllanthus is widespread in moist valleys with Hibiscus (Wagner and Lorence 2011b), and Metrosideros is a dominant element of remnant forests in many zones, as previously discussed. The latter is probably H. tiliaceus in this setting; the wood was frequent and abundant in these contexts, probably reflecting an environment where thickets of the sprawling trees were common. A small amount of coconut and Thespesia were found in the Ototemui test units. While the trees may have been growing this far inland, it is more likely that the wood and coconuts were transported from lower elevations. Breadfruit wood was also found in several samples at this site. These trees may have also been cultivated in the area, as one of the early missionaries noted that “The Breadfruit tree grows in these places, almost to the summit of the Mountains, & upon the steep sides of the inferior hills” (Crook 2007, 140), and they may have been components of a house garden in this setting.

Synthesis The Hatiheu assemblage includes material from coastal and inland locations in this broad, well-watered valley located on the central windward coast of Nuku Hiva. Numerous small streams are located throughout the middle and upper valley, feeding into three main watercourses that run through the lower valley to the bay. Taxa identified in the coastal and inland study sites were largely strand and mesic forest elements, with several representatives of transitional lowland vegetation zones. Along the coastal flat near Pahumano, a low, open, and somewhat dry vegetation formation was probably present on the escarpment early in the sequence, sometime in the 14th century AD. This is evidenced by the co-occurrence of several taxa characteristic of such formations, and the presence of several small trees or shrubs in the lower temporal zone. A mixture of moist and dry ecological zones were probably also present around this area. It is evident that the assemblage recovered from the lowest cultural strata, which contained a thick deposit of charcoal and rich variety of taxa, resulted from an intense episode of vegetation burning on the ridgeline. Vegetation was burned several more times, and there is some evidence the soils of the escarpment were destabilised. By the 16th century AD, many native plants growing in this coastal area had been replaced by introduced taxa. Considering domestic activities occurred at this site, some of the identified material may have been elements of a house garden. Because the sample size from Zones D

225 and F were small, few changes in ecological conditions over time could be assessed for this area. In the lower valley interior, several elements of moist forest formations were noted. There were also some indicators of disturbance, as well as cultivation of breadfruit, in this zone possibly by the 14th or 15th century AD. In a chiefly area near several ceremonial structures in the middle valley, elements of native vegetation were also identified, along with the wood of several culturally valued trees that were probably planted in and around the area. The presence of cultivars, including Inocarpus and breadfruit, in these samples indicate that food-producing trees were also present in the area by the late prehistoric period (c. AD 1600–1800). These study locations are not far from an extensive agricultural complex of irrigated taro pondfields. Finally, in the upper valley it was noted that elements of native and cultivated vegetation were mixed with an abundance of Hibiscus wood, the latter of which may have been prevalent at moist interior locations such as these, much as it is today. The presence of breadfruit wood in samples from this location suggests that the trees were cultivated in late prehistory far back into Marquesan valleys, a finding that corresponds with early contact accounts.

Anaho Anaho, as discussed, is a small amphitheatre-shaped valley located directly to the east of Hatiheu. Three charcoal assemblages from this valley were analysed: an occupation on the northern coastal flat, called Teavau’ua (AHO-1), a site located several hundred metres southeast of this location referred to as Teavau’ua South, and material from test units placed near stone house foundations that are located throughout the valley. While assemblages from the Teavau’ua AHO-1 and stone structures were large (over 1000 fragments each), the Teavau’ua South collection is smaller.

Teavau’ua The Teavau’ua (AHO-1) site lies on the northern coastal flat, inland of the coral reef in the bay. A small stream runs immediately to the east, and the valley wall rises sharply to the west. This site contained two prehistoric cultural layers: the main occupation in layer IIIb, which dates between the 15th and 17th century AD, and a more ephemeral occupation in layer IV dating to the 13th century AD (Allen 2004b). Allen concluded that this area was

226 a locus of domestic activity, and at least five earth ovens were encountered in excavation. This site produced a large volume of charcoal. Most of the material available for analysis derived from the main occupation in layer IIIb. Results are also presented for layer IV, which included several features, and three samples from a transitional context (layer IIIb/IV) are reported. In the relative abundance chart that follows (Figure 7.4), data are presented in descending order from most recent to oldest temporal context. Taxonomic richness values for this assemblage varied by functional context: 6–13 taxa were recovered from non-feature contexts, while features had only 2–8 taxa. It was notable that the large assemblage from the layer IIIb non-feature contexts produced only a few more taxa than the comparably modest one from layer IV. This finding could indicate that early in the sequence, a richer flora was present in the area, similar to the trends noted at the Hakaea Beach and Hatiheu Pahumano sites. The flora identified in this assemblage were mainly trees, along with several small trees / shrubs and one true shrub (Psydrax odorata). Samples from both the lower and upper strata contain a similar quantity of charcoal from native trees (5–6 taxa). More small tree or shrub taxa were found in the upper layer than in the lower layer. This trend is probably not notable, considering the large difference in sample sizes. The two assemblage dominants in this study, Sapindus and Thespesia, were found in each strata analysed. The relative percentages of Sapindus did not vary widely in the two main cultural strata at this site. It accounted for a modest proportion (approximately 5–15%) of material analysed from non-feature contexts, but accounted for almost 20% of charcoal recovered from features in each strata. Thespesia also accounted for a modest proportion of material from layer IV contexts, both feature and non-feature, but was a large percentage of the assemblage in layer IIIb, where it accounted for almost 50% of non-feature and over 65% of feature contexts. These trends suggest that Sapindus was used as a fuel consistently over time. Thespesia could have become more frequent later in the sequence, though this suggestion is very tentative. The trend is probably related to the number of ovens found in layer IIIb because Thespesia was, as previously discussed, the preferred fuel for earth ovens in the study location. These trees would also have been a component of native vegetation near the coast, while Sapindus would not, which may be why the latter was found in low abundances throughout the sequence.

227

drupe (shrub) marchionicus

Phyllanthus BarringtoniaCeltis cf.pacificaasiatica CeltisCordia pacificacf. subcordata CyclophyllumPandanus barbatumcf. Pouteria tectoriusSapindus sp. saponariacf. Sapindus saponariaThespesia populnea Allophyluscf. Alstonia marquesensisHibiscus costataPhyllanthus spp. cf. unknown13cf. marchionicuscf. unknown13 (Sapotaceae)cf. Psydrax (Sapotaceae) odorataArtocarpus Cocos altilisCocos nucifera nucifera endocarp unknown15unknown21 indeterminate indeterminate bark unknownLayer

                IIIb

       IIIb features

     IIIb−IV

IIIb−IV features

          IV

    IV features

0 0 0 0 0 0 10 0 0 0 10 0 10 0 10 20 30 40 50 60 70 0 0 0 0 0 0 0 0 0 0 0 0 10 20 30 40 0 0 0 0 0

indeterminate angiospermindeterminate unidentifiable monocot Fragment Taxonomicct. richnessLayer

  465 13 IIIb

  723 8 IIIb features

  53 6 IIIb−IV

5 2 IIIb−IV features

  52 7 IV

  37 5 IV features

0 10 20 0 10 0 10

A variety of native taxa were found in this assemblage. They include several large coastal trees, including Barringtonia and Cordia, which were found only in the upper two strata in small proportions. Both have cultural uses; fruits of the former were used to create a narcotic used to stun fish, and the latter was culturally valued and also produced an important timber, as previously discussed. Several other trees, including Cyclophyllum and Pandanus, were found in both strata. Planchonella was infrequently found and only present in upper contexts, while Celtis was only found in the lower layer (IV) contexts. While Cyclophyllum and Planchonella had few known cultural uses, Pandanus had numerous uses. Celtis wood was used occasionally as fuel in this region, other parts of the tree have medicinal uses (Butaud, Gérard, and Guibal 2008, 67–9), and the fruits were once consumed on Rapa (Prebble and Anderson 2012, 170), but no particular uses were documented for the Marquesas. All of the foregoing were probably components of the native forest in this location, as there is little evidence any were cultivated or encouraged in the past. A number of native shrubs or small trees were found in this assemblage. Alstonia (cf.) had few known cultural uses, and was only found in the lower cultural layer. The absence of this taxon and the native trees Barringtonia and Celtis from the much larger layer IIIb assemblage is notable, and could suggest some elements of native vegetation were eventually cleared from the area. The occurrence of Allophylus, Phyllanthus (syn. Glochidion), and Psydrax, however, in layer IIIb indicate that some did persist. While the former had few cultural uses, Psydrax, a shrub, had hard wood that was used for digging sticks in Hawaii (Rock 1913, 437) and houseposts and tools in Samoa (Whistler 2004), and may have been introduced into the assemblage as the by-product of craft carving. The Sapotaceae Unknown13 was found in each strata at this site, though in low abundances. Finally, Hibiscus, which may be an indicator of disturbance or could have been an ornamental cultivar, was present consistently in very small proportions in the upper stratum. While this was a large assemblage, the number of Polynesian-introduced taxa were restricted to Artocarpus and coconut. Coconut wood made a small contribution to the assemblage, and was present in both upper and lower strata. Coconut shell accounted for a larger portion (7–17%) of non-feature contexts in this assemblage and it was also present in several features. Only a very small proportion of breadfruit wood (<1%) was identified in a layer IIIb oven feature. These findings suggest that coconut trees were probably growing in and around the coastal flat, much as they are today, while breadfruit trees were uncommon in the immediate vicinity during this occupation.

230

While a large charcoal assemblage, the taxa identified at this site were not very diverse. A small percentage of this material was from indeterminate hardwoods (approximately 10% from each strata), and a small amount was from indeterminate monocot stems. Less than 15% of the material from each strata was in such poor condition it was deemed unidentifiable.

Teavau’ua South Teavau’ua South lies in the northern lowlands of Anaho Valley, several hundred metres behind the beach and southeast of the stream that runs through the coastal flat. This site contained two pre-contact cultural layers, III and IV, that date to between AD 1400 and 1650. One unit contained a layer (III/IV) with a less distinct boundary, and these results are presented separately. Three features were encountered at this site. Most of the charcoal in this assemblage comes from layer III non-feature contexts, while the other contexts made a more modest contribution to the assemblage. Because a more precise chronology has yet to be determined at this site, discussion of temporal changes in vegetation is very limited in the following review. Taxonomic richness values for this assemblage appear to be correlated with the number of fragments identified (Figure 7.5). Non-feature contexts from layer III were very rich, with 17 taxa identified, while layer IV had only 6. Features contained few taxa (3 to 5), similar to findings from the Teavau’ua (AHO-1) site. The flora identified in this assemblage included mainly trees and a small number of small trees or shrubs. All but one of the arboreal taxa found in the lower layer (IV) are also present in the upper layer (III), which does not suggest large changes to the vegetation occurred between the two occupations. Thespesia was a consistent contribution to most contexts in this assemblages. It was found in some abundance (9–17%), though it was absent from layer IV features. Sapindus was also a modest contributor to the assemblage, being approximately 5% of layer III and 19% of layer IV charcoal from non-feature contexts. It did account for most of the charcoal found in the indeterminate feature in layer III (Efe. 78), which could suggest this feature was an earth oven as Sapindus was also a preferred oven fuel, as previously discussed. Overall, the trends for these two dominant taxa indicate that both were used as fuel at this site in both the earlier and later occupation.

231

Cordia subcordatacf. Cordiacf. Cyclophyllum subcordataErythrinaPandanus variegatacf. barbatum Pandanus tectoriusPouteria Sapindustectorius drupe sp. drupesaponaria ThespesiaCoprosma populneaHibiscus spp. spp.MaytenusPremna crenatacf. Premnaserratifoliaunknown13 serratifolia cf. (Sapotaceae) AleuritesAleuritesArtocarpus moluccana moluccanaCocos altilis nucifera cf.endocarp CocosCocos nucifera nucifera endocarp unknown22unknown23unknown24indeterminateindeterminateindeterminate barkindeterminate parenchymous unknownindeterminate angiospermunidentifiable tissueFragment monocot Taxonomicct. richnessLayer

                          281 17 III

  11 3 III features

    12 4 III/IV

            52 6 IV

      14 5 IV features

0 0 0 0 0 0 0 0 0 20 40 60 0 0 0 20 0 0 0 0 0 0 0 0 0 0 0 20 40 0 0 0 0 0 0 0 0 0 0

Other native taxa found were also identified in this assemblage, including Cordia, Pandanus, and Planchonella. Cordia, a large coastal tree, was found in both layers, though it was more abundant in the lower layer (IV) where it accounted for 13% of the assemblage. A small proportion of Erythrina wood was a unique find in this study, recovered from a non-feature context in layer III. These large trees were seen in coastal situations by very early visitors and the soft, light wood was used for small canoes (Lisiansky 1814, 90–1). Parts of the tree were also used as coffins in the Marquesas (E. Handy 1923, 111), and elsewhere in Polynesia the red seeds had decorative uses (Rock 1913, 191). Pandanus drupe fragments were the only parts of the trees found in this assemblage, and they were not abundant. These materials may have been food waste, though they may also have been introduced after decorative or cosmetic uses (Whistler 2009, 158–67). Planchonella was also present in layer III, but it was not abundant. The trees, as previously discussed, had no known uses in the Marquesas, but in other archipelagos the wood may have been used for construction and craft. All of the above were probably naturally-occurring elements of the lowland and strand forests of Anaho, though trees such as Erythrina may also have been planted. Hibiscus was found in non-feature contexts in layer III, where it comprised 23% of material from that layer. This find may indicate that disturbed, secondary forest was present in the area as early as the 15th century AD, but it may also derive from cultivated types of Hibiscus. If it is the sterile form that has long, slender trunks, the wood may have been imported from some distance for construction or craft use, as it was observed growing at higher elevations today and informants indicated clusters of the trees were typically found some distance upland. Several other native trees or shrubs were found in this assemblage including Coprosma, Cyclophyllum, and Maytenus. Most made up a small percentage of the material from layer III and were probably native forest elements, as they have few known cultural uses. Premna may have been part of this flora, though because it did have value as a medicinal and ornamental tree it may have also been planted in housegardens. Four unknown angiosperm woods were also identified. All made a small contribution to the assemblage (1 to 8%) and occurred in both layer III and IV. They included Unknown13, the Sapotaceae previously discussed, which occurred in a layer IV feature and several layer III non-feature samples. Many of these plants were probably elements of the native forest around the site.

233

Several Polynesian-introduced taxa were also identified in this assemblage, including candlenut (Aleurites) shell and possibly some wood, breadfruit and coconut wood, and coconut shell. Candlenut shell accounted for 6 to 7% of material from layer III non- feature contexts and features from layer IV. A very small amount of wood was tentatively identified as candlenut; this is the only such wood found in the study, and it may indicate that the trees were growing nearby as the wood is not durable and had few uses (Thomson 1980, 20). Breadfruit wood was a small proportion of the material identified from an indeterminate feature in layer IV (Efe. 83). This finding might suggest the trees were growing nearby, though the wood may also be debris from crafting projects as, for example, the inner bark was used to create a rough type of tapa cloth (Marchand 1810, 146; Thomson 1980, 16). Finally, coconut wood was a small part of this assemblage, but coconut shell was present in all layers and functional contexts. It accounted for approximately 10 to 15% of charcoal in non-feature contexts, and was also a large percentage of material found in some features. This findings suggests coconut trees were probably growing near the site, much as they are today. Though not large, this assemblage provided an informative picture of local vegetation in the Anaho lowlands during the middle prehistoric period. The finding of one very rare taxon, Erythrina variegata, was notable. Approximately 8 to 15% of charcoal could not be identified as other than indeterminate hardwoods. Very few monocot stem fragments or parenchymous tissues were noted. Less than 8% of charcoal from each strata was deemed unidentifiable, indicating this assemblage was in somewhat better condition than others in the study.

Surface features The final assemblage in this study is composed of samples from locations distributed throughout Anaho Valley. These test units were placed adjacent to stone house foundations with the general aim of dating the structures (see Allen 2009a). Most date to the late prehistoric period, but in some cases, cultural strata and features that predated the structures were also encountered. Samples from these excavations have been grouped into three temporal categories for analysis: late sites date to after AD 1650, middle prehistoric to AD 1400–1600, and early sites to before AD 1400. Results have been further grouped for discussion by geographic location as present in the north or south valley. A number of features were encountered in these excavations including ovens, hearths, and post-moulds;

234 several concentrations of charcoal of undetermined function were also found below structures. A large quantity of charcoal resulted, and almost 1,000 fragments were analysed from samples from late contexts. Sample sizes from middle and early period contexts were smaller. Taxonomic richness values for this assemblage varied by temporal phase (Figure 7.6), and this trend probably relates to variations in sample size. Material from early contexts and middle prehistoric period contexts were of a similar richness, having 5 to 8 taxa in non-feature contexts, and features from both north and south valley locations had up to 7 taxa. Material from late prehistoric contexts in the north valley was considerably richer, with 16 to 18 taxa occurring in both feature and non-feature contexts. Species richness was, overall, lower in the south valley samples. While this may be related to the smaller assemblage size from this location, it was notable that non-feature contexts from all temporal phases in the south valley had only 3–8 taxa. These findings suggest that a variety of taxa were present throughout the valley, and they persisted in many areas well into the late prehistoric period. Parts of the southern valley, especially inland locations, may have had a more restricted flora later in the sequence, though it should be noted most of the charcoal from these locations was collected from within features, which tend to have lower richness values. It was noted that this assemblage is markedly richer than the similarly sized one from Teavau’ua, which was expected as it covers a much wider geographic area. The flora identified at these sites included a number of trees, small trees / shrubs, and one shrub. Many of these were found in the middle and late period contexts. Several natives, which will be subsequently discussed, were present only in samples from the early or middle periods, suggesting that some taxa may have been removed or reduced in inhabited parts of the valley by the late prehistoric period. Of the assemblage dominants, Thespesia and Sapindus comprise a large portion of the charcoal analysed from these excavations. Thespesia is the most common taxa, accounting for 18 to 39% of the charcoal from non-feature contexts in all temporal groups, and 21 to 43% of the charcoal from middle and late period features. Most of this material was from the north valley, but early south valley contexts also contained some Thespesia at both coastal and inland locations. Sapindus also comprises a large portion of the assemblage. It accounts for 9 to 21% of non-feature contexts and is a moderate to large proportion of material found in many feature contexts from both parts of the valley and coastal and inland locations. The consistent and abundant occurrence of these two taxa indicate both were used frequently as fuel throughout the sequence in Anaho, and the trees

235

Celtis pacificaCordiacf. subcordata CordiaPandanus subcordatacf. Pandanus tectoriusPandanusSapindus tectorius tectorius saponaria drupe Thespesia populnea cf. ThespesiaXylosmacf. populnea XylosmasauveolensCoprosmacf. sauveolens Coprosma spp.Hibiscus spp. spp.Maytenus crenataMetrosideros collina Phyllanthuscf. PhyllanthusPremna cf. marchionicuscf. serratifoliaPsydrax marchionicuscf. Santalum unknown13odorata insulare (shrub) Aleurites(Sapotaceae)Artocarpus moluccana altilis endocarp cf. Artocarpus altilisLocation

               North

           North features

  South

   South features

       North

     North features

     South

 South features

     South

South features

0 0 0 0 0 0 0 0 10 20 30 40 50 60 70 80 0 10 20 30 40 0 0 0 0 0 0 0 10 0 10 20 30 0 0 0 0 0 0 0 0 0 10 20 30 40 0

Calophyllum Cocosinophyllumcf. nucifera CocosCocos nucifera nuciferacf. CocosendocarpInocarpus nuciferacf. Morinda fagifer endocarp unknown02citrifoliaunknown09 unknown11 unknown17 unknown21 unknown29 indeterminate indeterminateindeterminate barkindeterminate endocarpindeterminate parenchymous unknown angiosperm tissue indeterminate monocot unidentifiable Fragment Taxonomicct. richnessLocation

            261 18 North

            879 16 North features

7 3 South

    119 5 South features

 150 6 North

     89 7 North features

  38 5 South

 32 2 South features

    51 8 South

 12 0 South features

0 10 0 0 0 10 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 20 30 40 0 10 20 30 40 50 60 70 80 0 10 20

were probably abundant in the area, though each would have occurred naturally in different areas. The wood of several large native trees were also identified. Cordia was found in some middle and late period contexts from the north valley, in low abundances. This finding suggests that these large coastal trees were still growing in Anaho late in the prehistoric period, a finding that is in contrast to results from other valleys in the study, but in keeping with comments regarding Hawaiian flora where it was noted that Cordia could still be found in out-of-the-way situations in the early 20th century (Rock 1913, 7). Calophyllum was identified in an early south valley context, where it accounted for 17% of a sample. The wood of this tree is notably absent from the larger, late-period assemblage, which may suggest the trees were never common here, and some that did occur along the south valley coast were later removed. Similar to the findings in other locations, there is a suggestion that they are native and not a Polynesian introduction. Other natives found in this assemblage include wood and nut shells of Celtis, Pandanus, and Xylosma, all trees, and shrubs or small trees including Coprosma, Maytenus, Metrosideros, Phyllanthus, Premna, Psydrax, and possibly sandalwood (Santalum). Many of these taxa were not found in abundance, and most accounted for <4% of any context, but a few results were notable. Celtis was found only in north valley contexts, where it was moderately abundant (12%) in features of the middle period, and it was a smaller proportion of material from feature and non-feature contexts of the later period. This tree, as previously discussed, may have had some minor uses, though its importance in the Marquesas is not known. Metrosideros, an important component of many native vegetation communities that had several cultural uses, was a notable portion of one sample from the early period. It was not found in later contexts, which suggests the trees (and perhaps related patches of native lowland forest) were uncommon or absent near residences after the 14th century AD. A small amount of Pandanus wood and nut shells were found in late north valley samples. Other natives with few known uses were also identified in north and south valley contexts, including Xylosma and Coprosma, found in both early and late samples, and several others were identified in middle and/or late contexts. A very small amount of what may be sandalwood (Santalum) charcoal, a small-diameter fragment, was encountered in an early period sample; it is the only such material recovered in this study. Sandalwood trees were valued for their sweet-smelling timber, which was used to scent coconut oil and, in the Marquesas, to embalm the dead (Brown 1935; Whistler 2009). Other native trees and shrubs mentioned were present in small abundances and were probably minor vegetation

238 components. It was notable that many were present in the late period samples, indicating a number of native forest elements persisted in Anaho well into the 17th century AD. Hibiscus was a component of middle period assemblages in both north and south valley locations, at both coastal and lowland sites. This wood did not occur in any early sample. While it may derive from ornamental plantings, its presence in domestic settings at these study sites could also derive from craft projects or other utilitarian uses. As the wood is not esteemed as a fuel and it tends to burn quickly, its presence here in middle period samples (10 to 12% of material from the north valley) may also indicate that H. tiliaceus was present in disturbed lowland situations in Anaho by the 13th or 14th century AD. Seven unknown taxa were also found in this assemblage; most occur in the larger sample from late period contexts, where they account for a small percentage of the assemblages. Most are likely to be minor elements of native vegetation. The Sapotaceae Unknown13 was not found in early contexts, but was encountered in one middle-period sample and several late period contexts, all in coastal settings. The presence of this taxon in late period contexts is interesting, as in other locations Unknown13 was already absent from contexts several hundred years older than this. This finding suggests the trees persisted in parts of Anaho far longer than they did in other locations. Polynesian-introduced taxa found in this assemblage include candlenut shell, breadfruit wood, coconut wood and nutshell, Inocarpus, and cf. Morinda. Breadfruit charcoal was found in both middle and late period samples from both feature and non- feature contexts. It occurred in middle period samples from the northern lowlands (Structure 11, 336) and the southern lowlands (Structure 232), and in many late period samples associated with foundations in the north valley and the south valley. The frequent occurrence of breadfruit wood in late contexts suggests the trees were common throughout the valley by the 17th century, and that they were present in the northern lowlands by the 15th century AD. Aleurites (candlenut) shells were found in late contexts in south and north valley samples. In all situations, they were a very small proportion of the assemblage, and overall, candlenut shells were not frequent at any Anaho study site. This finding is similar to Hatiheu, where none were identified. The meaning of these findings is uncertain. Inocarpus wood was found in a sample from a late-period structure (Structure 8) in the northern valley lowlands. The nuts, which resemble a large chestnut, were used to create a traditional dish, as discussed. These large trees usually thrive in moist inland locations, and they were probably growing around the stream that runs through the coastal flat.

239

Coconut shell fragments were often a small percentage of the material, and though they occurred in many samples, they were not ubiquitous in any phase or location. Shell fragments were not found in early samples, and they were a very small proportion of charcoal from features in middle and late-period contexts. However, in the late period they occurred in almost half of the samples and accounted for 17% of charcoal from non-feature contexts in the north valley. A small amount of coconut wood was also found in an early south valley sample, and later was present in several late contexts where it was a very small percentage of the samples. While coconut (especially shell fragments) was very frequent and abundant in the Hakaea and Pahumano assemblages, it was markedly less so in many of the Anaho contexts. This finding could suggest coconut trees were not frequent or widely distributed throughout the valley, and remained largely an element of coastal vegetation throughout the prehistoric sequence. Cf. Morinda wood was noted in one middle period feature context from the south valley coast (Structure 2). The fruits of this tree, noni, had medicinal uses and were sometimes eaten after roasting, though they have a bitter flavour (Crook 2007, 75). This was the only occurrence of Morinda wood in the study, but its absence should not be over- emphasised as certain anatomical features of the wood required for identification do not appear to preserve well. Though today they are common in many situations, their distribution or abundance in the past remains uncertain. This assemblage was one of the largest in the study. It provided evidence that a variety of native and introduced taxa were present throughout Anaho Valley into the late prehistoric period, and some occurred later here than they did at other study locations. Charcoal from indeterminate hardwoods accounts for 12 to 21% of material from the larger temporal assemblages (late and middle period, north valley contexts), but the proportion was considerably less in feature contexts. Indeterminate monocot stems account for a small percentage of material. Unidentifiable charcoal was less than 7% of late contexts, but somewhat more of middle period samples, indicating much of the charcoal in this assemblage was in relatively good condition.

Vegetation pattern assessment Anaho is a small valley on the central-east windward coast of Nuku Hiva that is home to a variety of microclimates. One permanent stream runs through the northern coastal flat; there are springs in several locations and dry streambeds where water flows out to the

240

Table 7.4 Ecological zones of trees and shrubs from the Anaho Teavau’ua (AHO-1) site.1 Layer / IV IIIb calAD 2σ range2 approx. 1200–1400 approx. 1400–1650 Strand Allophylus marquesensis  Cocos nucifera   Cordia subcordata  Thespesia populnea  

Mesic Alstonia costata  (cf.) Cyclophyllum barbatum  (cf.)  (cf.) Hibiscus spp.  Planchonella sp.  (cf.) Sapotaceae Unknown13  

Xeric Sapindus saponaria  

Mixed Artocarpus altilis  Celtis pacifica  Pandanus tectorius   Phyllanthus cf. marchionicus  Psydrax odorata  (cf.)

241

Table 7.5: Ecological zones of trees and shrubs from the Anaho Teavau’ua South site.1 Layer / IV III Estimated Age2 1400–1650 1400–1650 Strand Cocos nucifera   Cordia subcordata   Thespesia populnea  

Mesic Aleurites moluccana   Coprosma spp.  Cyclophyllum barbatum  (cf.) Hibiscus spp.  Planchonella sp.  Premna sp.  Sapotaceae Unknown13  

Xeric Erythrina variegata  Sapindus saponaria  

Mixed Artocarpus altilis  Maytenus crenata  Pandanus tectorius  

242

Table 7.6: Ecological zones of trees and shrubs from the Anaho north valley surface structure excavations. Found at sites: C=coastal; I=lowlands, mid-valley, or inland .1 (no data) Structure 336; Structures Provenience / Below Structures 68,8,24,32,13,16 calAD 2σ range2 68,11

Before 1400 1400–1650 After 1650 Strand Cocos nucifera I C, I Cordia subcordata I C, I Thespesia populnea C, I C, I

Mesic Aleurites moluccana I Coprosma spp. I (cf.) I Hibiscus spp. C, I C Inocarpus fagifer C Sapotaceae Unknown13 C

Xeric Sapindus saponaria I C, I Xylosma sauveolens I

Mixed Artocarpus altilis I C, I Celtis pacifica C C, I Maytenus crenata I C, I Pandanus tectorius C, I Phyllanthus cf. marchionicus C, I Psydrax odorata C (cf.)

243

Table 7.7: Ecological zones of trees and shrubs from the Anaho south valley surface structures excavations. Found at sites: C=coastal; I=lowlands, mid-valley, or inland .1 Below Structures Structures 232, 2 Structures 242, Provenience / 254, 2 245, 2 calAD 2σ range2 Before 1400 1400–1650 After 1650 Strand Calophyllum inophyllum C Cocos nucifera C, I Thespesia populnea C, I C C

Mesic Aleurites moluccana C, I Coprosma spp. I Hibiscus spp. I Premna sp. I Sapotaceae Unknown13 C

Xeric Santalum insulare C (cf.) Sapindus saponaria C, I C C Xylosma sauveolens C (cf.)

Mixed Artocarpus altilis I I Metrosideros collina I Maytenus crenata I Morinda citrifolia C (cf.)

244 bay from the steep valley walls during heavy rain. Dry slopes and ridges can be found in several locations at the margins of this valley. Taxa identified in charcoal assemblages from this site reflect, accordingly, a mixture of vegetation zones. Both coastal and inland excavations were represented in these materials. Assemblages from Teavau’ua layer IIIb, which dates to the middle prehistoric period, and surface features from the late period in the northern valley are much larger than others from the valley. In the following evaluation, samples recovered from northern sites that date to before AD 1650 are discussed first, and then material from early south valley contexts are reviewed. Samples associated with paepae occupations, which date to after AD 1650, follow and to conclude, temporal trends for the valley as a whole are considered.

Northern sites, before AD 1650 Sites on the northern coastal flat in this discussion include Teavau’ua (AHO-1) and Teavau’ua South, and several deposits found below paepae from Structure 68, 11, and 336. Sites have been grouped into temporal phases that are classified as early, dating to before AD 1400, and the middle prehistoric, dating to AD 1400–1650. A large quantity of material included in this discussion derives from Teavau’ua layer IIIb. Anaho samples from this temporal context contained several coastal strand species including Cordia, coconut (Cocos), and Thespesia. The latter two taxa are well represented in the earliest deposits in the northern valley, including layer IV at Teavau’ua, and all three are present in middle period contexts. Allophylus was also found in the Teavau’ua (AHO-1) middle period context (layer IIIb). All are, as was described for Hatiheu, widespread elements of strand vegetation in the region and were probably components of native coastal vegetation early in the sequence. Erythrina wood was found in one sample from the Teavau’ua South site that dates to the 15th to mid-17th century AD. These large native trees, more typical of strand zones, can be found in interior locations and they are present in both xeric and transitional forests (Decker 1991, 7–8, 24). The wood and seeds have a number of traditional uses and it is therefore uncertain whether this was an element of native coastal vegetation near the site, it had been planted in the area, or the wood had been transported from elsewhere for a craft project. These trees are very rare in Anaho today; the only large specimen known to local residents had been removed in 2011.

245

Only a few taxa that associated mainly with xeric vegetation formations were found in pre-AD 1650 contexts from the north valley. Sapindus was one of these, and was one the most frequently occurring woods in the assemblage. It was present in many samples from both coastal and inland sites in early and middle period contexts. Sapindus trees were probably present on the dry slopes located a short distance to the west and to the north of the coastal flat. The presence of Celtis in both early and middle-period coastal deposits from this area (Teavau’ua layer IV and Structure 68), combined with Sapindus, further suggest that a semi-dry open forest occurred in these areas in the early to middle prehistoric period. Both woods have few uses other than as fuel. Pandanus nutshells occur in some early and middle period contexts from these north valley locations, though they are not abundant. Nutshells may have been food processing waste, though they were perhaps not a particularly important resource, and the trees were most likely a component of native vegetation growing in the area. The co-occurrence of Pandanus with Phyllanthus (syn. Glochidion) and Psydrax, a shrub, in Teavau’ua layer IIIb are indicative of Decker’s transitional zone. The additional occurrence of Maytenus, a small tree or shrub with few known uses, near a structure in the north valley provides further evidence that a mainly low, native forest of this type was present in parts of the northern valley. The occurrence of native trees and shrubs characteristic of mesic vegetation including Coprosma, Cyclophyllum, the Sapotaceae Planchonella and Unknown13, cf. Alstonia, and Hibiscus in many contexts indicate that areas in the northern lowlands were home to a moist vegetation community during this period. Most are small trees and shrubs that were probably growing in the lowlands and coastal flat surrounding the stream, and they indicate a low forest was probably growing in these areas. The presence of Hibiscus, and its frequency in the Teavau’ua South and Structure 11 and 336 deposits, could indicate native forests in this area had been disturbed and recolonised by H. tiliaceus as early as the 15th century AD. Trends for the Polynesian-introduced taxa are consistent at these sites. Candlenut (Aleurites) shells were found in both layers at the Teavau’ua South site, but were not present in other early north valley contexts. The kernels had several important uses, and they may have been imported from other areas, though the trees were probably growing in moist locations in the valley, perhaps near the northern stream or the various freshwater springs. Candlenut trees occur occasionally in the valley today; seedlings were observed growing near a freshwater spring at the back of the south valley in the Vai-o-vai district (fieldnotes,

246

27 March 2010). The presence of these materials in the early assemblages suggests the trees were probably planted in Anaho as early as the 15th century AD. Breadfruit wood was similarly identified in several middle period inland sites from the northern as well as the southern valley, and very small amounts were noted in middle period sites near the coast at Teavau’ua layer IIIb and Teavau’ua South layer IV. These findings suggest breadfruit trees were cultivated in several parts of the valley, probably mainly at inland settings by the 15th century AD.

Southern sites, before AD 1650 Sites in the southern valley that date before AD 1650 include deposits associated with Structure 2 located near the coast, and material from below Structure 232 mid-valley and Structure 254 that lies inland near the back-valley wall. These materials date to the early and middle periods, and the assemblage is small. Several strand species were found in contexts located below the coastal structure, Structure 2, as well as Structure 254 that is located inland against the valley wall at Te Papa Uka. Calophyllum was found in an early coastal context in the south valley (Structure 2). These large, native coastal trees were probably elements of native strand vegetation in the earliest part of the sequence, before AD 1400. Coconut shells and Thespesia were also found in both coastal and inland deposits in the south valley. A small amount of cf. coconut wood may additionally be present in the inland assemblage; while these materials were expected in coastal settings, their presence at an early site far inland is notable. Coconuts may have been transported there from more seaward locations and burned as by-products of food or craft production, similar to findings at Ototemui in Hakaea. Thespesia, similarly, may have been planted this far inland but it is more likely the wood was transported to this location from seaward areas. Several trees characteristic of xeric formations were found in the south valley samples from this period. They include Sapindus and Xylosma, which are characteristic of the vegetation found today along the dry slopes and ridges at the back of the valley (fieldnotes, 27 March 2010). These taxa were also identified in deposits from the coastal site (Structure 2) where Sapindus was frequent, indicating the wood may have been transported from back valley locations or it was collected along the dry hillslope that occurs several hundred metres east of the site, where Sapindus can be found growing today. A small fragment of cf. Santalum insulare (sandalwood) was identified in samples from the

247

Structure 2 at the coast. These trees are extremely rare elements of native vegetation today, as most were harvested for the sandalwood trade, as discussed. The distribution and frequency of sandalwood in prehistory is unknown, though there is some evidence they were rare in other east Polynesian islands by the early contact period (Parkinson 1773, 78). Rock (1913, 127–35) indicated they were once common in some lowland situations in Hawaii, but individuals in native dry forest settings there were very rare. The wood may have been collected in some part of Anaho, though considering its value it also may have been transported from a distance. Elements of mesic vegetation communities are also present in early and middle period contexts from the south valley. These include one instance of Sapotaceae Unknown13 in a middle period sample from the coastal site, and Coprosma and Premna at both inland sites (Structure 232, 254). These trees and shrubs were most likely elements of native vegetation present in the lowland forests of the south valley. The presence of Metrosideros in an early context at the coastal site suggests patches of native forest were still present in the area before AD 1400. The occurrence of some Hibiscus in a south valley inland sample from the middle period could indicate the vegetation of this area was disturbed or cleared by the 15th century AD, though this is a tentative suggestion as the wood may derive from ornamental plantings near habitations.

Paepae excavations, after AD 1650 A large quantity of material in this discussion derives from excavations at stone foundations in the north valley coast and lowlands, and other material is from two features located inland and one along the coast in the southern part of the valley. All contexts date to after AD 1650. Taxa identified in these samples can be found in a variety of habitats, but only a few were elements of dry forest formations. Strand species were found in samples from both coastal and inland sites in the northern part of the valley. These include coconut, Cordia, and Thespesia. Coconut shells occurred frequently in these contexts. A small amount of coconut wood was also identified in coastal locations, indicating coconut trees were probably growing in these places in late prehistory. One exception was Structure 13, located near the northern stream, where coconut shells were absent. Coconut was also absent from late period inland sites in the south valley, though there were few samples from these locations. Cordia was found in two north valley contexts, one near the stream and abundantly in a hearth at the coastal paepae (Structure 68)

248 in this area. These findings suggest Cordia trees probably occurred in the north strand zone, along with Thespesia, and may have extended into lowland areas in late prehistory. Both may have been persistent elements of native coastal forests, though by this period they could also have been planted. Both Sapindus and Xylosma were found in late contexts. The frequency of Sapindus, and the co-occurrence of these two taxa in a north valley context, indicate a xeric vegetation community was probably present on nearby slopes, similar to the finding from early and middle period samples. Celtis occurred in a number of late period samples from the northern valley. Celtis is, as previously discussed, characteristic of semi-dry, open wooded areas alongside Sapindus and Xylosma and it would have been an element of these formations as well. Sapindus was also frequent and abundant in late period samples from structures in the south valley inland, at Te Papa Uka (Structure 242, 245), indicating these trees continued to grow on the dry slopes and ridges around this area. All three taxa can be found in a range of zones, and it is considered that they may also have been present in lowland areas of transitional vegetation. Several elements of mesic vegetation were also found in these samples. They include a number of cultivated and native trees and shrubs. Mesic native forest elements that persisted in this period include Coprosma, Maytenus, and Unknown13 in the northern valley, and Maytenus in the south, all small trees or shrubs with few known cultural uses. Polynesian introductions in this assemblage include Candlenut shells, which were recovered from several inland north valley locations dating to this period. As discussed, the trees may have been planted in moist locations near the stream and farther inland where intermittent streams flow down steep valley walls. The presence of Inocarpus wood at a structure near the coast in the north valley (Structure 8) further suggests that useful trees were planted in moist areas of Anaho by the mid-17th century AD. Breadfruit wood occurred frequently in north and south valley contexts from this period, at both coastal and inland sites, suggesting there were groves in Anaho Valley by the late prehistoric period. Other native taxa were found occasionally in late north valley contexts, co-occurring similarly to earlier deposits from this area. These taxa include Pandanus, Phyllanthus (syn. Glochidion), and Psydrax, all elements of the transitional zone. The former two taxa may have been components of a mesic formation, but with the additional occurrence of Maytenus in a number of samples, there is an indication a low, open forest of mixed composition probably persisted in the northern valley lowlands late in the sequence.

249

Synthesis Trends for the wood of two large trees identified in this assemblage are interesting and in accord with trends from the other coastal excavations. Cordia was found in middle and late contexts from excavations in several coastal locations. These trees may have been planted, but it is also possible this material is deadfall from old individuals that persisted in Anaho into the late prehistoric period, as these trees are long-lived. Calophyllum wood was only found in an early context from the south valley coast, and its absence from later temporal contexts in Anaho is probably notable. It is interesting that, though middle and late period samples from the north valley areas were large, Calophyllum was not found in any sample. This finding could indicate the trees were an infrequent, or possibly absent, component of either the strand or inland forests, or perhaps naturally-occurring trees along the coast had been removed before the 15th century AD. Both Cordia and Calophyllum can be found today in situations near the coast; a large specimen of Cordia was growing near Structure 2 in the south valley and Calophyllum were planted along the beachfront in the north (fieldnotes, 26 March 2010). Coconut was frequent in many different late north valley contexts, and most coastal locations, but it was infrequent in early and middle period contexts near structures. It was very rare in south valley inland sites. These findings suggest that before contact, coconut trees were probably found in the north valley from the coast to some distance inland, where they may have been frequent. In the south valley, however, they were most likely present mainly near the coast. The persistence and frequency of Sapindus throughout the sequence in the Anaho assemblages, and the occasional co-occurrence of Xylosma and Celtis in these contexts, suggests that a semi-dry forest persisted on the slopes and ridges surrounding the valley. These findings further suggest Sapindus trees may have been abundant in these locations, a finding consistent with Decker’s (1991, 24) observation that they are still very common on upland slopes. There are indications that at least one native tree persisted in Anaho later in the sequence than other study locations. Though not common in samples from any period, Unknown13 was only found in two late samples from the northern valley (Structure 24, inland), and its presence there in a post-17th century AD context is the latest occurrence of the wood in this study. Allophylus was, conversely, found in middle period samples (Teavau’ua IIIb) but was absent from late contexts, even though the sample size from late

250 contexts in this valley was large. It is possible that this small, nondescript native tree could have been reduced or removed from the area by the mid-17th century AD. There are also several indications that native vegetation of the valley had been disturbed in northern and southern inland areas by the 15th century AD. One indicator is the increased frequency of Hibiscus in middle prehistoric samples, though it should be noted the sample sizes from early contexts were small. Another notable finding is that the only occurrence of Metrosideros was in an early sample from the south valley coast, which is interesting as this wood was not present in any of the numerous samples from later contexts. As Metrosideros is a primary component of native vegetation formations in the archipelago, this finding could suggest that some of these formations were removed early in the sequence and they did not regenerate at lower elevations. However, the presence of a number of other native trees and shrubs in late period contexts indicate that other elements of native vegetation did persist in parts of Anaho well into the late prehistoric period. Lastly, there is evidence in this assemblage that coconut, candlenut, Inocarpus, and breadfruit trees were cultivated in Anaho by the mid-17th century AD. The occurrence of breadfruit wood in over half of the units near structures that dated to this period, including both south and north valley settings, suggest the trees were probably frequent in the valley by the late prehistory. Breadfruit wood was also found in deposits near several structures dating to the middle period, indicating trees were cultivated here at an earlier time.

Quantifying change Lastly, the individual sequences for each valley were compared to consider broad temporal trends in the data. These trends had the potential to characterise the tempo and extents of prehistoric vegetation change in the windward valleys of Nuku Hiva, and they illustrate how Polynesian plant environments evolved. Because sample sizes from each stratum varied considerably, and this was shown to have impacted relative abundance values, quantifications for this assessment were undertaken at a very low level. Even at this level, it was demonstrated that in some situations vegetation communities were not always well represented in the charcoal assemblages (in accordance with low GC values). At the Hakaea Beach site, layers III and VII were of an adequate size and produced Gini Concentration indices within an acceptable range, and they should support the following analysis. The Anaho Teavau’ua site layer IIIb also produced acceptable values, but others

251 from the north valley lowlands did not, even though sample sizes in many cases were not small. Two assemblages were selected for review: the Hakaea Beach site and material from excavations in the Anaho north valley lowlands. These locations produced the largest number of samples and total fragment counts per temporal context. The Hakaea site contained three main cultural strata in occupations over a period of between 100 to 250 years, that occurred mostly before the 15th century (Allen and McAlister 2010). Trends from this study site provided a view of vegetation change at centennial scales. The Anaho north valley lowland assemblage contained several stratified deposits from sites near the coast, and a large amount of material from the paepae excavations. Radiocarbon ages for these sites spanned a longer period of time, from as early as 13th to 15th century to the late prehistoric, over a period of somewhere between 350–550 years, as well as a shift in habitation from coastal to inland locations after AD 1650 (Allen 2004b, 2009a). Trends noted for this assemblage provided a complimentary view of vegetation change over multi- centennial scales. Two main trends were identified in this assessment. First, it was evident there were declines in the presence of many native trees and shrubs over time. Second, overall increases in the frequencies of Polynesian-introduced trees were noted. Each trend is illustrated and discussed below.

Decline of native trees and shrubs A list of native trees and shrubs that were absent from late contexts were compiled for the Hakaea and Anaho north valley assemblages. To consider their frequency of occurrence at early sites, ubiquity values were calculated for each location (Table 7.8 and 7.9). More than a dozen taxa (Table 7.8) in the Hakaea assemblage were present in the lower strata, which date to before the 14th century, but were not found in any sample from layer III. Eight of these taxa were trees, three were small trees or shrubs, and one was a shrub. Many occurred in only a few samples, and were probably rarer vegetation elements, but Planchonella, Dodonaea, and Unknown25 were present in a quarter of samples, and Unknown13 was a very frequent occurrence in the oldest stratum at this site. All of these taxa were probably common components of native lowland vegetation in the lowland Hakaea Valley.

252

Table 7.8: Native trees and shrubs present in contexts before the 14th century AD, but absent from later contexts in the Hakaea Beach assemblage. Taxon1 Habit Ubiquity in early contexts (N=21) Alphitonia marquesensis Arboreal .05 Calophyllum inophyllum Arboreal .14 Cordia subcordata Arboreal .05 Crossostylis biflora Arboreal .05 Cyclophyllum barbatum Arboreal .19 Guettarda speciosa Arboreal .10 Planchonella sp. Arboreal .29 Terminalia sp. Arboreal .19 Melicope spp. Arboreal or Shrubs .10 Phyllanthus cf. marchionicus Arboreal or Shrubs .10 Sapotaceae (Unknown13) Arboreal or Shrubs .86 Dodonaea viscosa Shrubs .24 1 Plus five additional undetermined taxa.

Table 7.9: Native trees and shrubs present in contexts before AD 1650, but were absent from late contexts in Anaho north valley assemblages. Ubiquity in early and middle Taxon Habit period contexts (N=61) Barringtonia asiatica Arboreal .02 Cyclophyllum barbatum Arboreal .10 Erythrina variegata Arboreal .02 Planchonella sp. Arboreal .07 Allophylus marquesensis Arboreal or Shrubs .03 Alstonia cf. costata Arboreal or Shrubs .02 Premna serratifolia Arboreal or Shrubs .07 1 Plus four additional undetermined taxa.

The Anaho north valley assemblages were combined for this analysis, as they were for the previous assessment. Recall the pre-AD 1650 samples were from the Teavau’ua and Teavau’ua South sites, and several deposits below Structures 11, 68, and 336, while all late samples were associated with structures. Most of the taxa that were only found in pre-AD 1650 contexts (Table 7.9) were probably minor vegetation components. Four were trees and three were small trees or shrubs. Considering the large sample size, Cyclophyllum might have been more frequent, though its ubiquity value is still rather low (.10). At both sites, it is evident that a mixture of native trees and shrub / trees were removed or became minor vegetation components later in the occupation sequences. Both

253

Sapotaceae, Planchonella, and Unknown13 are notable in this regard. Additionally, some taxa that were frequent in early contexts at these locations declined to very low frequencies later on. For example, though Unknown13 did occur in several late Anaho contexts, it was present in very few samples from that period. And Celtis was more frequent in early Hakaea samples, but it occurred on only one late sample. Similar trends were noted for Maytenus, Pandanus, and Psydrax at that site. Though a similar trend was noted in the Anaho material for Pandanus, others mentioned did not follow a similar pattern as with Celtis, for example, which occurred more frequently later in the sequence.

Increase in Polynesian-introduced trees The other major trend noted in this study was an overall increase in the frequencies of charcoal from introduced trees. This trend was evaluated by reviewing ubiquity values for several Polynesian-introduced plants at each study site. Recall that ubiquity values, or frequencies, do not directly represent frequencies of the trees in past landscapes. They are presented to consider trends within a study site, and the values are not directly comparable from one site to another (Smart and Hoffman 1988). Samples were grouped broadly into early, middle, and late categories for this evaluation. At Hakaea, early contexts from Layer VII have a calAD 2σ range of 1164–1293, the middle period from Layer V range is 1268– 1411, and late period from Layer III is 1317–1444. At the Anaho sites, early contexts date to before 1400, middle to between 1400–1650, and late to after 1650.

Table 7.10: Ubiquity comparisons of Polynesian-introduced taxa at the Hakaea Beach site. 1164–1293 1268–1411 1317–1444 Temporal context1 Layer VII Layer V Layer III No. samples 21 13 7 Aleurites moluccana .24 .38 .57 Artocarpus altilis - .23 .71 Calophyllum inophyllum .14 - - Cocos nucifera 1.00 .62 .86 1 Radiocarbon ages reported calAD 2σ range.

At Hakaea, Aleurites, Artocarpus, Calophyllum, and Cocos were examined (Table 7.10). Candlenut shells were a frequent occurrence in this assemblage, present in one- quarter to over one half of samples in all periods. No temporal trend was distinguished for

254 these materials. Breadfruit wood did not occur in any early sample; this is notable as there were a large number of samples from the early period (21). In the middle period, it was present in several samples, and in the late period it was more frequent, occurring in most samples. Calophyllum was only present in early samples, and as previously discussed there is some suggestion it was probably not a Polynesian introduction. Coconut was very frequent in samples from this site, and some of this material was present in most temporal contexts.

Table 7.11: Ubiquity comparisons of Polynesian-introduced taxa at the Anaho north valley sites.

Temporal context1 Before 1400 1400–1650 After 1650

No. samples 7 55 24 Aleurites moluccana - .11 .13 Artocarpus altilis - .09 .42 Cocos nucifera .71 .55 .54 Inocarpus fagifer - - .04 1 Radiocarbon ages reported calAD 2σ range.

At Anaho north valley sites, Aleurites, Artocarpus, Cocos, and Inocarpus were examined (Table 7.11). Candlenut shells were not present in any early sample, though there were fewer from this period (7). They were rarely present in features at this location, but were found in some middle and late period non-feature samples. Breadfruit wood was also not present in any early sample. In the middle period samples, of which there were many, it occurred in only a few samples. In the late period, however, breadfruit was present in almost half of the samples. Coconut displayed no notable trend at this site, and though it was present in many samples throughout the sequence, it was not ubiquitous. Inocarpus was absent from all contexts until the late period, where it was found in one non-feature sample. At both locations, the most notable trend is that of breadfruit, which is absent from all early contexts, but becomes frequent in samples from the later periods at each site. These trends do not appear to have been influenced by differences in the quantity of samples from each period, as it actually goes counter to those values in both locations. It was noted that this cline occurs at different times at each location. At Hakaea, the notable increase in frequency occurs sometime between the late 13th and early 15th centuries AD. At Anaho, the notable shift occurs between the 15th and 17th centuries AD.

255

Summary of findings In this chapter, the results of each study site were presented and discussed, along with a review of cultural uses of the trees that may have introduced these materials into the archaeological record. A phytogeographical assessment of these trees and shrubs were also conducted by synthesising findings for each valley, to consider potential changes in ecological conditions over time. Two major trends noted in the results included a decline in the presence of a number of native trees and shrubs over time, and increases in the frequencies of Polynesian-introduced trees. While the characteristics of change were similar in each study site, it was noted that they occurred at different times. In the final chapter of this thesis, I consider these results and address the main research questions of the study.

256

Chapter 8: Landscape domestication in Marquesan prehistory

Introduction The results of this study have provided information on how the landscapes of Nuku Hiva were changed by Polynesian settlers over approximately 600 years between the 12th and 18th centuries AD. Dynamic changes occurred within forests and human populations during this time. There are indications people reconstructed plant environments in the study locations without substantial loss of vegetation cover, essentially domesticating these landscapes and turning them into productive, forested environments. In this thesis, I sought to investigate factors that were influential to development of an agronomic system based on tree cultivation in prehistory. The analysis of multiple assemblages from several relatively controlled catchments, representing a variety of functional contexts, has yielded a sizeable dataset with which to review this question. In this chapter, I first synthesise the results from individual study locations and review the processes of landscape domestication that were observed. Next, the roles played by tree crops in three prehistoric periods are reviewed in the context of other cultural phenomena, and factors that may have influenced the development of an arbor-centric cultivation system are assessed. Finally, these results are considered within the context of Central East Polynesia and compared with findings from the Society Islands.

Processes of landscape domestication Our current understanding of the processes by which settlers modified their environments in tropical Polynesia was outlined as part of an examination of several case studies in Chapter Two. It was found that while some of parts of the process were well understood, there were several critical gaps in scale and patterning. These include a poorly resolved understanding of the landscape modification and food-production strategies used between the early settlement and late prehistoric intensification periods, the tempo and characteristics of vegetation change at a localised scale, and the relative importance of tree crops in any given prehistoric period. The present study produced important information to address these concerns, as several distinct processes of landscape domestication were observed. Four transformation processes were noted: (1) vegetation clearance through use of fire; (2) modifications to coastal vegetation communities; (3) modifications to inland forests including clearance and replacement of natives; and (4) the development of agroforests and

257 tree plantations. The behaviours these processes represent, and the archaeobotanical indicators that were used to determine them, are summarised in Table 8.1. Each of these processes will now be discussed in detail.

Vegetation clearance by fire The native forests of central East Polynesia once stretched to the coast in many locations. Palynological evidence suggests that some vegetation communities were composed largely of Pandanus, coconut, and other palms, along with trees such as Ficus, Trema, Glochidion, Macaranga and others in certain locations, and an under-storey of low vegetation and ferns (e.g., Athens and Ward 1993; Athens et al. 2002; Ellison 1994; Parkes 1994; Prebble and Wilmshurst 2009; Prebble, Anderson, and Kennett 2012). Some forests were reduced by humans through burning, evidenced by dramatic increases in charcoal particle concentrations in sediment cores, and occasionally by abundant charcoal in basal archaeological deposits. These two sources of evidence provide complimentary information on vegetation communities on two distinctly different temporal and spatial scales. The present study has demonstrated that archaeological charcoal can also offer information about these vegetation formations and, when collected systematically, can do so in detail. Little was known about the composition of the lowland Marquesan vegetation before human settlement, and wood charcoal deriving from burn layers provided detailed information about these formations in several locations. At the Hakaea Beach and Pahumano sites, abundant charcoal was encountered in early cultural deposits at locations near the coast. At Hakaea, the charcoal-rich basal cultural layer dates to the late 12th or 13th century. At Pahumano, repeated burning and slumping of the scarp adjacent to the site produced a series of particularly rich charcoal assemblages. The thick, basal charcoal deposit at this site dates to the 14th century AD. Samples from both locations contained a variety of native taxa that later declined to low levels, or were not seen again. Material from comparable contexts was not available for the Anaho Valley, most likely because the earliest occupations in Anaho have not been located. That said, one charcoal concentration was encountered in a pre-15th century deposit near the southern coast below Structure 2 (Efe. 61, #6249 and 6359). These samples were not very informative, however, as they contained little identifiable charcoal.

258

Table 8.1: Processes of landscape domestication observed in the windward valleys of Nuku Hiva. Archaeobotanical Process Behaviours indicators Thick charcoal deposits Vegetation burning Initial removal of scrub and composed entirely of native native coastal forests taxa

Reduction in frequency and Reduction or removal of abundance of native coastal native vegetation around species that have few known early settlements Expansion of living spaces uses and activity areas near coast Planting of useful trees in First occurrence of homegardens introduced species

Thick charcoal deposits Reduction of lowland forest cover to create swiddening Frequent presence of certain plots native taxa Intensive use of native tree Indicators of forest products (wood, leaves, disturbance, with an increase flowers, fruit) Expansion to inland areas in taxa that thrive in disturbed areas Reduction of seed dispersal mechanisms (i.e., birds, bats) Reduction or disappearance of some native tree and shrub taxa

Increase in frequency and Creation of cultivated forests abundance of introduced for food, construction, craft, taxa, especially food- decorative, medicinal use Development of agroforests producing trees and tree plantations Planting or encouragement Occurrence of taxa in of culturally important trees multiple sites and valleys Intensification of food production

Vegetation burning on Nuku Hiva has a long history. Charcoal concentrations in sediment cores suggest that burning took place in the lowlands from at least the 14th century AD (e.g., Allen et al. 2011), the practice was observed at contact (Crook 2007, 140), and it continues to the present day. Fires such as these are set for a number of reasons, such as to clear brush and make way for construction. Localised firing is also used in swidden cultivation where fast-maturing aroids or yams are typically planted (Kirch 1994, 104). A

259 variant of such practices was observed on leeward Nuku Hiva in the early contact period in an upland location (Crook 2007, 146) where crops such as paper mulberry, kava, plantain, sugar cane, and breadfruit were being cultivated in a burned area, though it was noted that by the early contact period swidden cultivation was infrequent. Fires such as these have been known to escape their intended boundaries and destroy large tracts of vegetation (e.g., Decker 1991, 23; Rolett 2008), and repeated burning has resulted in degraded fern- and grasslands in some areas today. The present study provided only modest evidence of localised burning. In archaeological deposits, these practices can be represented by high concentrations of charcoal in sediment, or as thick burn layers overlain by evidence of cultivation (i.e., fields, agroforests, tree plantations). At Pahumano, the charcoal-rich sediments of Zone H represent the initial removal of native vegetation from the escarpment, and later in this sequence (as early as the 15th century, but possibly late prehistoric) Zones E and D both contain breadfruit wood. While the archaeological evidence indicates this area was a domestic setting early in the occupation sequence, the charcoal suggests that later the escarpment was cultivated to some extent, perhaps as a house garden, and large trees such as breadfruit may have stabilised the highly mobile soils. A charcoal rich burn layer was also found in a late prehistoric context in an upper valley test unit at Hatiheu (TU-4, #100009). This layer contained charcoal from introduced taxa such as breadfruit and Hibiscus. It was underlain by another cultural deposit, which suggests localised burning in an already modified environment, and it may reflect regeneration of an area cultivated by a type of long-fallow swiddening noted in other Oceanic settings (Kirch 1994, 122–3).

Coastal vegetation modifications This study has also provided important information on the evolution of coastal Marquesan vegetation communities in prehistory. Little was known about how the vegetation of these areas was modified by early settlers, or the tempo of these modifications, though it is likely vegetation around early settlements was greatly reduced and replaced with useful vegetation early in the sequence. By the contact period, foreign visitors noted that landscapes near the coast were lush and elaborately constructed environments (Figure 8.1).

260

Figure 8.1: Coastal scene on Nuku Hiva, 1804. Watercolour by Johann Kaspar Horner, a scientist aboard the Krusenstern voyage. Reproduced with permission from the Ethnographic Museum of the University of Zurich, Switzerland. Inv. No. VMZ 820.02.006. Removed for copyright reasons. Image may be requested from: www.musethno.uzh.ch. It can also be viewed on dust jacket of Twelve Days at Nuku Hiva (Govor 2010).

At the Hakaea and Hatiheu Pahumano sites, where early occupations were sampled and a stratified sequence was available, several important aspects of vegetation change were noted. First, it appears these areas were extensively reworked during the first two hundred years of occupation. Second, coconut trees were important components of coastal vegetation both early in the sequence and after settlement. That said, as no interruption in the use of coconuts was noted at these sites, and no temporal shifts in the morphology of coconut shell fragments or wood anatomy could be detected (mostly because of their very small size), it is not possible to determine when or how introduced varieties replaced native varieties. Third, there are strong indications, especially at Hakaea Beach, that Cordia and Calophyllum trees had been reduced in this zone by the 15th century AD. It is unlikely that these trees were completely removed from the coasts, given that several trees were observed growing behind Hakaea Beach in 2010. Lastly, many native plants growing near the coast had been replaced with introduced taxa, including breadfruit and Casuarina, by the 15th or 16th century. This trend included the eventual disappearance in the charcoal record of two Sapotaceae, Planchonella and Unknown13, trees that are not present in the Marquesas today. The decline of Unknown13 wood was especially marked, being nearly ubiquitous in early samples from these sites and absent from all contexts by the end of the sequences.

261

Very few Anaho samples dating to before AD 1400 were available for study, and of these few were informative. One very early sample (#6351, AD 1036–1245 at 2σ as reported by Allen 2014) from a paepae in the south valley (Structure 2) contained Metrosideros and Calophyllum wood. These materials were probably collected from local vegetation, and provide some information on coastal forest formations very early in the Anaho sequence. Coconut trees also appear to have also been an important and continuous component of coastal vegetation in this valley throughout the sequence. The absence of coconut in the paepae test units from some early contexts in this zone may be a sampling issue. These contexts may have also been, both functionally and depositionally, unlikely areas for debris accumulation; most units abutted the stone foundations and some were located on steep slopes. There is some indication that certain native trees, including Metrosideros, were reduced or removed from this zone by the 15th century as these taxa were not found in any of the numerous samples from later contexts. One important difference in the Anaho assemblages is that other native taxa appear to have persisted in the north valley coastal zone until at least the 17th century. These included the shrub Allophylus, which was present in several non-feature contexts in Teavau’ua layer IIIb, Cordia, found in a feature at this location as well as in several paepae units (Structures 13 and 68), and Unknown13, which was found in feature and non-feature contexts at one paepae (Structure 8). These findings suggest that, even after the mid-17th century, native vegetation elements co-existed with introduced trees such as breadfruit and possibly Morinda near the Anaho coasts. It should be noted that Anaho was more extensively sampled than other valleys. One major coastal occupation and a range of test units were sampled throughout the valley, and these excavations yielded a large quantity of charcoal. These materials had the potential to provide a more complete vegetation history for Anaho, and probably contributed to the detection of several native taxa in late-period contexts that were not observed elsewhere. It should also be noted that while the Anaho samples did produce larger volumes of charcoal than other locations, fragment totals were not exponentially higher than Hakaea Beach (for example) and samples did not produce the taxonomically richest assemblages of the study. To summarise, at sites such as Pahumano and Hakaea Beach, rapid and drastic vegetation modifications were noted very early in the sequences. This trend may typify early coastal occupations when they are observed at local scales. After this period, a more gradual transition was noted in the Anaho assemblages. The Anaho materials represent a large spatial area, thus the observed patterns can probably be considered typical of the later

262 phases. This process may have varied in tempo and characteristics in other locations depending on intensity of use and population size, among other factors.

Lowland forest modifications The lowland forests of the Marquesas stretch from behind the coastal zone throughout valley bottoms to the back-valley slopes. Although largely unoccupied and seldom used today, at contact inland areas were home to lush, cultivated forests. It is not known how or when these agroforests developed, or what role native trees played in these landscapes. Palynological studies from Hatiheu cores have informed only modestly on past flora in these contexts (e.g., Allen et al. 2011; Gourdon 2003), and the results of this study have provided data to directly address these concerns. I first consider the exposed dry slopes that can be found near the study sites—around valley margins or on ridges running inland from the coast—and then review changes that occurred in the mesic lowland forests. On the xeric slopes that are located in close proximity to several of the sites in this study, a low, open, native vegetation community was present early in the cultural sequences. This is evidenced by the co-occurrence of frequent taxa such as Sapindus, in combination with Xylosma, Celtis, or Maytenus, trees that are characteristic of dry Marquesan slopes today (Decker 1991, 24). There is overwhelming evidence that Sapindus was a favoured fuelwood, but the trees may also have been frequent in these zones, notably on the slopes and ridges that surround the sites. While Xylosma wood was infrequent at all sites, small amounts of Celtis and Maytenus were present in many samples from Hakaea and the Anaho north valley throughout the sequence. These small trees were probably also common on the xeric slopes, as Florence (1997, 196–9) indicates they occur in semi-dry, open, wooded areas alongside stands of Sapindus and Xylosma. They may also have been occasional components of transitional zone flora (Decker 1991, 24). Guettarda also occurs in early locations close to the coast, including at Hakaea and Pahumano, where it appears to also have been part of a somewhat dry, open native vegetation community. While Sapindus occurs frequently and abundantly in most sites and temporal contexts, wood from Celtis and Guettarda occur less frequently and/or much less abundantly in later in coastal deposits at Hakaea (most notably) and Pahumano. In the Anaho north valley, however, Celtis is still frequent in post-AD 1650 contexts, indicating that though it was reduced in some places it did persist in other areas late in the prehistoric sequence.

263

In the lowland mesic forests, several elements of native vegetation were found in samples from pre-17th century contexts at both Anaho and Hatiheu, indicating that at least some native forest elements persisted in these locations into the late prehistoric period. In Hatiheu, these materials were infrequent and only a few fragments of each were identified. Crossostylis, a native tree found in mesic formations with Metrosideros, was identified in a sample from the lower valley that dated to the 16th or 17th century. Native shrubs such as Allophylus, cf. Alstonia and Phyllanthus, and trees including Metrosideros were identified in samples from the middle and upper Hatiheu valley dating to the late prehistoric or very early historic periods. In the Anaho assemblage, shrubs or small trees native to the mesic lowland forests, such as Coprosma, were found in some of the pre-17th century contexts. Coprosma and Phyllanthus were also found in late prehistoric contexts, indicating some of the smaller elements persisted in these locations. But several important natives, such as Planchonella and possibly Cyclophyllum, do not appear in samples after the middle period, which could indicate that some of these formations were reduced and did not regenerate (and in the case of Planchonella, were reduced to such an extent they were probably extirpated). However, as previously discussed in reference to several other zones, the presence of native trees and shrubs in late period contexts indicate that these formations did persist in parts of Anaho well into the late prehistoric period. There are numerous indicators of disturbance in the lowland forests in these assemblages. Hibiscus tiliaceus is a very common feature of the Marquesan vegetation at many elevations today. It spreads as a result of human activity by regenerating quickly in disturbed areas. The absence of Hibiscus wood in these assemblages was interpreted cautiously because the wood of this tree turns to ash at low temperatures. Conversely, its presence needed to be interpreted with care because the wood of the various taxa that occur there cannot be distinguished at present. In dry areas, this concern could have been addressed by considering its co-occurrence with Xylosma suaveolens, which also emerges in dry areas quickly after disturbance but, as previously mentioned, that wood was infrequent. In the Anaho assemblages, Hibiscus was present in many samples. At the inland Hatiheu sites, it was found in several pre-17th century contexts from the lower and middle valley, and it was frequent and abundant in late prehistoric contexts, mixed with native and cultivated vegetation. It was very infrequent in the Hatiheu Pahumano samples, though it did occur in both early and late contexts. These findings indicate Hibiscus may have been prevalent in moist interior locations in Hatiheu by the 15th or 16th centuries, and was certainly so by the 17th century. At Anaho, Hibiscus was frequent in middle period samples

264

(c. AD 1400–1650) from the coastal and near-coastal locations, but was much less frequent in the inland paepae test units from both the middle and late prehistoric periods. It is possible that in this valley, Hibiscus was not as widespread in prehistory, though this suggestion is made very tentatively. It is also notable that no Hibiscus was identified at the Hakaea Beach site, which may indicate that it was an infrequent occurrence in the lower valleys before the 15th century. Overall, when Hibiscus is evaluated as an indicator of disturbance in moist lowland forest settings, it appears that it only became frequent enough to be important in the charcoal records in some near-coastal locations by AD 1400–1600 and farther inland at some locations after c. 1650. Its uneven occurrence is probably due to the poor preservation qualities of the wood, but it is also a possibility that these fast-growing trees were discouraged around inhabited locations.

Extirpated or significantly reduced trees and shrubs While the demise of sandalwood in the early historic period is well-evidenced, little is known about the prehistoric extinctions and extirpations of other native trees in the Marquesas. The present study has demonstrated that it is possible to recover data from charcoal assemblages to address these concerns. Several taxa identified in this study are rare or endangered in French Polynesia today, or have a range that is more restricted than it seems to have been in the past, and the findings of several Sapotaceae, Metrosideros, Coprosma, and Allophylus are briefly discussed. One of the most notable finds in this study was the frequent and sometimes abundant presence of several members of the Sapotaceae family. Today, no members of this family are known to be native to the Marquesas and several steps were taken to ensure these determinations were accurate. One species endemic to Henderson Island, Sideroxylon (syn. Nesoluma) st-johnaium, was present in the reference collection. The wood anatomy of the Sapotaceae family was also studied: many genera have been documented in a series of publications by the U.S. Forest Products Laboratory (e.g., Kukachka 1982), and a number of specimens from Scion (New Zealand Forest Research Institute, Ltd.) were also examined. Anatomical details of representative archaeological charcoal fragments were examined using an SEM in consultation with wood anatomist Lloyd Donaldson. It was determined that there were two species of interest in this family: Planchonella (some species syn. Pouteria) and Sideroxylon (syn. Nesoluma). Both have representatives in the Society Islands

265

(IUCN 2013; Meyer and Butaud 2009) and the Hawaiian Islands (Smithsonian National Museum of Natural History n.d.) that are in either a critical or endangered state. Planchonella was noted in a number of contexts in this study—most notably Hakaea Beach—and was also identified at Pahumano and several of the Anaho sites. It was probably not a dominant vegetation component at either site. This wood was somewhat more frequent in earlier contexts than later contexts, and persisted in Anaho into the 15th century, but it is absent from all post-1650 contexts. Today, Planchonella taitensis (karaka) can be found from sea level to high elevations on Tahiti, Mo’orea, and Raiatea (also the Tuamotus and Australs), and is found in the largest concentrations in moist and transitional forests, though it is a protected species (Butaud, Gérard, and Guibal 2008). The wood has no known uses in French Polynesia, though Butaud notes in western Oceania it is sometimes used for construction and to carve canoe parts. Planchonella dominates a type of lowland forest in Samoa; several species produce wood that is used for houseposts and carved artefacts (Whistler 2004, 159), and the fruit produced is an important food source for pigeons and bats. I have tentatively identified Unknown13 as a member of Sideroxylon, pending the procurement of reference wood from other members of the genus. The charcoal of this taxon is abundant and nearly ubiquitous in early Hakaea and Pahumano contexts. It occurs infrequently and in much lower numbers throughout the Anaho sequences, which is the only location where it persists after AD 1650. Some Sideroxylon can grow 5–15 m high, may be long-lived, and are found in mesic forests in the Society Islands today (Meyer and Butaud 2009) and in dry forest situations at elevations over 130 m in Hawaii (University of Hawaii 2009). Another member of this family, Sideroxylon st-johnaium, is co-dominant in the plateau forest of Henderson, where it occurs in scrub vegetation and on cliff slopes (Waldren 1998). References to the cultural uses of these woods are scant in the historic literature. J.F.G. Stokes and Brown noted that Sideroxylon polynesicum (also, somewhat confusingly, called karaka) was esteemed as a fuelwood on Rapa in the 1920s (Brown 1935, 224; Prebble and Anderson 2012). It is native to the Cook Islands, several of the Austral Islands, and Hawaii. Rock (1913, 380–1) noted that around the same time, the Hawaiians made no use of any part of the tree (keahi), despite the wood being quite hard and durable. Rock noted that the trees were very prolific, producing large quantities of reddish-purple olive-shaped fruits. The fruits of the Sapotaceae are fleshy and eaten by birds (Pennington 1991, 83). Many mutually beneficial interactions exist between native trees and birds on Pacific

266

Islands, and many trees rely on the actions of animals for seed distribution and pollination (e.g., McConkey and Drake 2002; Meehan, McConkey, and Drake 2002). It is possible plant species such as these may have rapidly declined in prehistory because their main dispersal mechanism, frugiverous birds, were greatly reduced after settlement. The aforementioned trends do correspond with the temporal decline in native bird populations noted at the Hanamiai site in the Southern Marquesas (Rolett 1998). At this location, bird bones were very frequent in the initial occupation (dated to AD 1025–1300), and then declined dramatically over only 200–300 years due to predation by humans, habitat loss, and animal predators (Steadman and Rolett 1996). These findings could suggest that some frugiverous bird extinctions are closely tied to plant extinctions in the archipelago. In broader regional context, few members of the Sapotaceae family are known in French Polynesia. Several are rare and protected due to habitat loss and seed predation by rats (Meyer and Butaud 2009), and over longer time frames by the loss of loss of dispersal agents such as native pigeons and doves (e.g., McConkey and Drake 2002; Steadman and Freifeld 1999). Fossil pollen records have provided very little information on Sapotaceae in central East Polynesia, though it has been reported as a rare element of native vegetation in spectra from coastal sites in Tonga (Fall 2005) and lowland situations on Oahu, Hawaii (Athens et al. 2002). The trees are not mentioned in early contact literature, and there is some indication they were rare or absent in lowland vegetation formations in many parts of the region (with some exceptions for Tongatapu) by the late 18th century as Sapotaceae are poorly covered in the botanical collections of the Forsters (Nicolson and Fosberg 2004 list four plants, none from the Marquesas). Archaeologically, Sapotaceae charcoal has been found at several other Polynesian sites. At Kahikinui, Coil (2004, 361) reported Planchonella in several samples dating to before AD 1650. In the Cooks, several fragments of cf. Planchonella were identified in a pre-15th century context in the Aitutaki excavations, and Allen (1992, 96–7; Allen and Wallace 2007) suggested that the absence of this material from later contexts was notable. When considered together, the findings of this and other studies broadly suggest that Sapotaceae trees were once an important component of the native prehistoric lowland and coastal forests of East Polynesia, but were significantly reduced early in the prehistoric sequence. Several other native trees identified in this study are also of interest because their modern distributions are much more restricted than they appear to have been in the past. Metrosideros collina is rarely found in lowland situations in the Marquesas, usually occurring in remnant patches of native forest. I observed that it is dominant in upland zones

267 where native forests and windswept scrub formations are still present. It was not mentioned by Decker (1991) or Brown (1935) as a component of secondary inland or slope vegetation, indicating these trees were rare even at the upper reaches of habitable land by the 20th century. In the present study, Metrosideros wood was very infrequent, occurring only in an early Anaho south valley context and a late period inland Hatiheu sample. The wood has distinctive anatomy, and when preparing the charred reference collection it was not noted to be particularly friable. These findings may suggest that Metrosideros trees were never a major component of lowland Marquesan forests, although more samples from pre-17th century inland contexts would need to be studied to confirm this. A review of the distribution of Metrosideros trees and the presence of the wood in archaeological assemblages in other central East Polynesian locations would help in considering this suggestion. In the Society Islands today, Metrosideros is concentrated on plateaus, ridges, and moist upper slopes (Butaud, Gérard, and Guibal 2008, 158–61). Butaud and colleagues note that the wood has traditionally been used as fuel, but it was not noted in several prehistoric Society Island charcoal assemblages (M. Orliac 1997; Orliac and Wattez 1989). In Hawaii, however, Metrosideros is a common element of vegetation in many zones and there are remnants of native vegetation in leeward forests (Cuddihy and Stone 1990, 13) and recent lava flows (Rock 1913, 17–9) indicating the trees are not restricted to moist environments. The wood was frequent in charcoal assemblages from dryland locations in the archipelago (Allen and Murakami 1999), indicating that it was probably widely distributed there in prehistory as well. Together, these findings suggest that when the trees were frequent in the local environment, the wood was used a fuel, and they provide additional support for the earlier suggestion that Metrosideros was probably not frequent at lower elevations in (at least some) Marquesan valleys. Another taxa of interest found in the present study was Coprosma, an associate of Metrosideros. Coprosma were rarely mentioned in 20th century floristic surveys of the Marquesan lowlands (Brown 1935, 315; Decker 1970, 91), and it was a poorly understood genus in the study location until recently (Wagner and Lorence 2011a, 110). While Rubiaceae are a large family in this region, the half-dozen members of the Coprosma genus that are native to the Marquesas are primarily small trees or shrubs found at higher elevations in montane rainforests and windswept areas today along with Metrosideros and tree ferns (Mueller-Dombois and Fosberg 1998, 452). Coprosma charcoal was identified throughout the sequence in several lowland contexts from Anaho and Hakaea Beach, though it was not frequent or abundant in any sample. As previously discussed, no traditional uses

268 of the wood or other parts of the tree could be located. The wood was once used as fuel on Rapa (Brown 1935, 317) indicating that it must have some favourable burning qualities, but some New Zealand species have a very unpleasant odour when burned (Taylor 1961). This wood was also found very infrequently in Hawaiian assemblages summarised by Allen and Murakami (1999). Together these findings suggest Coprosma was not a common fuelwood in eastern Polynesia when other options were available. Furthermore, though the charcoal findings are probably not an accurate reflection of its frequency in the past environment, there is some indication it was present at lower Marquesan elevations than it is today. Lastly is the small tree or shrub Allophylus marquesensis. In some East Polynesian locations it is a component of back beach forests (Mueller-Dombois and Fosberg 1998, 389). In the Marquesas, it is found along ridges and in relict dry to transitional mesic forests, but it has not been collected from Nuku Hiva (Smithsonian National Museum of Natural History n.d.). The trees do not appear to have any cultural uses or importance, and Whistler (2004, 148) noted that in Samoa it is an inconspicuous forest element. The wood is distinctive and unlike other Sapindaceae occurring in the area. Allophylus occurs in this study in pre-15th century contexts at Pahumano and pre-1650 contexts at Teavau’ua; at both sites it is absent from later contexts. This trend is particularly notable for Anaho, as the sample size from late contexts there was very large, while at Pahumano the finding may have been influenced by the smaller sample of Zone D. Given that these materials are not abundant or very frequent in any context, findings once again suggest that some native plants were greatly reduced in lowland settings only a few hundred years after the earliest known occupation of the area. The results of this study have provided some interesting information on the prehistoric extinctions and extirpations of some native trees and shrubs. It is evident that several indigenous woody plants were extirpated from the lowlands of Nuku Hiva before contact, and others are have not been noted in the archipelago today. While human activities undoubtedly played an important role in this process, avian extinctions probably also contributed, and floristic datasets such as these provide information with which to consider these trends.

Cultivation of trees ...our way lay through a close tufted wood of fruit-trees, mixed with other sorts, extremely pleasant to us, on account of the thick and cooling shade.

269

Here and there we met with a solitary coco-nut palm, which, far from lifting its royal head with becoming pride, was out-topped and hid by meaner trees... The whole ground, as far as we had gone [inland], was covered with a rich mould, and contained excellent plantations, and groves of various fruit-trees. — G. Forster (2000, 2:336–7), upon visiting Tahuata in 1774

The last aspect of the transformation of lowland vegetation to be discussed is the emergence of cultivated landscapes, in particular the extensive arboreal groves seen at contact. There have been suggestions that tree plantations became extensive in the Marquesas in late prehistory (Millerstrom and Coil 2008); understanding when in fact this transition occurred is key to evaluating broader research questions such as why the Marquesans developed an economy based mainly on arboreal products. In this discussion, I evaluate when and how these systems evolved in three valleys of windward Nuku Hiva. Several potentially cultivated native trees are considered first, followed by the Polynesian introductions of note, including Casuarina, candlenut, coconut, Tahitian chestnut, and breadfruit.

Thespesia Thespesia populnea is a common shore plant in central Polynesia. The fruits are buoyant and easily dispersed by currents, and the pollen has been identified in pre-human contexts from Lake Temae, Mo’orea (Parkes 1994, Figs. 6.17, 6.18; Parkes and Flenley 1990), indicating it has the potential to be native to many of the Central East Polynesian archipelagos. Thespesia is, however, considered a Polynesian introduction to the far corners of the region, including Hawaii (Whistler 2009, 24–6) and Rapa (Butaud, Gérard, and Guibal 2008, 222–3). It is a gregarious tree that can form pure stands in lowland forests of the Marquesas (Hallé 1978, 324–5). It is occasional in inhabited valleys of Nuku Hiva in lowland vegetation, and is planted behind the beachfront in many locations. It is known today primarily as an ornamental and shade tree, and the reddish heartwood is prized throughout the region for carving and craft projects. The wood was once in such demand in the Marquesas that it was reported scarce in some areas due to over-harvesting (Linton 1923, 360) and to ensure a supply, nearly 30 hectares of the trees were planted throughout the Marquesas, Society, and Austral Islands in the 1970s (Butaud, Gérard, and Guibal

270

2008). In traditional Polynesian culture, the trees had ritual importance and featured in several legends (Brown 1935; Elevitch 2006). Thespesia wood was present in many different contexts in this study, indicating it was frequently used in domestic settings and produced copious quantities of charcoal. This material may have been an abundant by-product of craft, construction, or ritual activity burned as rubbish in household fires. A more parsimonious explanation is that it was a common fuelwood in prehistory. This finding is notable because though the wood was also identified in prehistoric earth ovens at several Society Island sites (Orliac and Wattez 1989; M. Orliac 1997), it was not mentioned as a fuel in either the historic or ethnographic literature. The temporal trends for Thespesia wood in this study are then of note, as they reflect changes in intensity of use and may also indicate changing levels of abundance in local vegetation. This material was frequent in Anaho north valley contexts by the 15th century AD and became nearly ubiquitous there after 1650. At the Hakaea Beach site, it was infrequent until the later occupation (c. 15th century), when it becomes nearly ubiquitous. At the Hatiheu sites, however, it is not as frequent and is considerably less abundant at the Pahumano site. Overall, the evidence suggests that Thespesia wood became more important, either as a fuel or craft resource, later in the prehistoric sequence in several valleys for reasons that are not understood. Given that the timber is mentioned in legend, suggesting a certain antiquity, it seems unlikely that naturally-occurring coastal stands of Thespesia would only first be intensively exploited several hundred years after settlement. It is instead suggested that the trees may have been encouraged—or perhaps even cultivated—in some valleys to meet increasing demand over time. These findings refine a suggestion my colleagues and I made in a pilot study of earth oven contexts, where we tentatively noted that anthropogenic impacts may have led to long-term declines of the species (Huebert, Allen, and Wallace 2010, 80–90).

Calophyllum Calophyllum inophyllum are large, slow-growing trees that can reach over 20 m in height and 2 m in diameter. The trees have fruits and flowers with numerous uses in traditional culture and large trees were prized as sources of canoe timber, featuring prominently in Marquesan legend (E. Handy 1930). Several important conclusions can be drawn from the Calophyllum findings in this research.

271

Firstly, the results of the present study directly relate to debates regarding the status of Calophyllum inophyllum in the central East Polynesian archipelagos. The trees are common in coastal settings in many parts of western Oceania, where they are naturally distributed. Guppy (1903) observed that the seeds can float and they may also be distributed by birds, but the fruits are probably too large for most seabirds to ingest (Elevitch 2006). In Polynesia, Whistler (2009) considers these trees to be indigenous to most of the archipelagos except Hawaii, but Fosberg (1975, 79) classified them as indigenous only as far east as the Cook Islands (also see Allen and Huebert 2014). Contrasting ideas such as these are not surprising; only modest fossil evidence has been available to evaluate the status of Calophyllum as the trees are poor pollen producers (Parkes 1994, 103). The pollen was, for example, infrequent in the Lake Temae, Mo’orea cores, with its earliest occurrences dated to a period around or just after the first millennium (Parkes 1994, 135). This might suggest it is a Polynesian introduction to the Society Islands, but it is more likely to be a reflection of the poor pollen-producing qualities of the trees. Archaeological evidence has been more informative, as the wood has distinctive anatomical features and it makes firm, good quality charcoal that has often preserved well, though thus far it has only been reported in late prehistoric contexts in that archipelago (Kahn and Coil 2006). The results of the present study provide some more conclusive evidence regarding the status of Calophyllum in the Marquesas. This wood was found in the earliest deposits at Hakaea Beach and the Anaho south valley coast, in contexts dating to before AD 1300, and it was also found at the Hatiheu Pahumano site in contexts dating to earlier than AD 1400. The presence of this material in early contexts at all three study sites strongly suggests Calophyllum is an indigenous coastal tree in this archipelago. Secondly, the distribution patterns of these trees appear to have shifted throughout the Marquesan sequence, meaning this is another tree that may have been encouraged in some settings later in the prehistoric sequence. In the present study, this wood occurred repeatedly in early coastal contexts at Hakaea, Pahumano, and the Anaho south valley. It occurs in the Hakaea assemblage as early as the late 12th century AD. These findings, when considered together, indicate Calophyllum trees were present along the coasts of northern Nuku Hiva during some of the earliest known occupations, and further suggest the trees were native to Marquesan coasts. Calophyllum is absent from coastal contexts later in these sequences. Because the wood produces good quality charcoal with distinctive anatomy, its absence in these samples was notable. One possible explanation is the trees were removed to accommodate settlements or make way for more useful food-producing trees. Another is

272 that later in the sequence they were used less frequently, perhaps becoming subject to tapu restrictions (as noted in the early contact period by Lisiansky 1814, 91). Finally, there is also some suggestion that Calophyllum trees were planted, or wood was at least used, in high-status settings by the late prehistoric period. The importance of these trees is evident in Marquesan legend, where they are said to flank a house of the gods (E. Handy 1930, 104), and the trees were planted around sacred structures throughout East Polynesia (Whistler 2009, 52). In this study, Calophyllum wood was found to have been burned at a high-status residence near an important ceremonial complex (Kamuehi) in Hatiheu Valley, which may indicate it was growing around a ritually important site. It was also noted that Calophyllum wood was absent from the numerous middle and late period Anaho paepae sites, which could further suggest the wood was infrequently used outside of culturally important contexts by this period.

Casuarina Casuarina equisetifolia wood was very infrequent in this study. These tall, fast- growing trees have distinctive needle-like leaves that look superficially like conifers. They are common elements of coastal vegetation throughout the Pacific Islands including East Polynesia (Mueller-Dombois and Fosberg 1998, 422). They have also been found in upland situations where they have been planted (Kirch and Yen 1982, 29). In the Marquesas, Casuarina groves have been noted on dry rocky ridge crests up to 500 m in elevation in some places (Hallé 1978), and an early visitor noted seeing them planted near an inland temple (me’ae) in Taiohae, in a grove with breadfruit and coconut (Porter 1822, 109–10). The trees germinate well in locations where vegetation has been burned, and they are frequent today along certain ridges where old, abandoned pathways are located (Decker 1991, 26–7). It should be noted that while they were not observed upland of the study locations, a row of mature trees has been planted behind the northern beach at Anaho. The trees have very hard, heavy wood that is difficult to work, but was carved into important objects such as digging sticks and weapons of war, and the name is synonymous with bravery in the Marquesas and Tahiti (Brown 1935, 16). Although today it is considered a good fuelwood, its use as such is not documented in ethnohistories. These trees also have the ability to fix nitrogen in the soil (Elevitch 2006). Casuarina wood produces dense, hard charcoal that is easy to identify. It was expected to occur with some frequency as a by-product of craft production, or deadfall if it

273 were present in the local areas because it produces copious stem litter (Elevitch 2006). However, it was found in only one sample in this study, at the Pahumano site in a post-AD 1500 context. The charcoal was not found in any Anaho or Hakaea contexts. Given the large number of samples and sites studied, Casuarina was most likely very infrequent in the lowland vegetation of the windward valleys before the late historic period. I have identified the wood in a prehistoric context at Ho’oumi Beach that may be of a similar age (data in prep.), providing additional evidence the trees were present in other Nuku Hiva locations before contact. These findings support the conclusion that Casuarina was a Polynesian introduction to the Marquesas, albeit a late introduction. Conversely, it may have been restricted to leeward areas, as its distribution and frequency appear to have been very limited in windward locations throughout the prehistoric sequence. These findings were briefly considered in a regional context. Casuarina trees have been alternately considered a Polynesian introduction to the archipelagos of East Polynesia, except Hawaii (Whistler 2008), native to much of French Polynesia (Butaud, Gérard, and Guibal 2008), or of uncertain status because the seeds can be dispersed by wind (Florence 2004, 104). The trees are prolific producers of a type of pollen that can travel long distances (Parkes 1994, 109). Palynological evidence does seem to suggest they are a Polynesian introduction to the region. Fossil pollen records from the southern Cook Islands and Mo’orea indicate the trees are a Polynesian introduction to these archipelagos (Parkes 1994), as it does not occur in the spectra until the late first to early second millennium (i.e., after at least the 8th century in Atiu, and 12th in Mo’orea cores). On Mangaia, pollen evidence also suggests Casuarina is a late Polynesian introduction (Ellison 1994). Archaeological evidence also indicates Casuarina was a Polynesian introduction to the region. The wood was identified in an early (1030–1290 calAD 2σ) waterlogged deposit from the Mangarevan coast (M. Orliac 2003b). Orliac considers this assemblage to be alluvial storm debris, with some materials originating from an occupation based on the presence of candlenut and coconut shell fragments. These findings have been compared with the present study, where there was some evidence that Casuarina trees may have been growing in at least two coastal area of Nuku Hiva by the 16th century AD. Although the archaeological evidence is not abundant, when considered together, multiple lines of evidence suggest Casuarina trees were a Polynesian introduction to many of the archipelagos of central East Polynesia. The timing of its introduction may have been irregular, as no clear temporal trend is evident.

274

Aleurites Candlenut (Aleurites moluccana) are tall, spreading trees that grow to 10 m or more. They can occur in gregarious stands, sometimes dominating landscapes in moist areas along rivers and ravines, although they can tolerate drier conditions and a variety of soil types (Elevitch 2006). In the Marquesas, the trees are mainly found in formerly cultivated moist forests (Decker 1991), although I also encountered them along dry roadside slopes in some leeward areas. Their silvery foliage makes the trees conspicuous and easy to spot from a distance. The bark and sap had a variety of uses, the seed kernels were strung together and used as candles, and the nutshells were used to create tattooing ink (Robarts 1974, 248; Whistler 2009). A recurring need for illumination probably accounts for the prevalence of candlenut shells in many archaeological contexts throughout the region (e.g., Allen 2004b; Kahn 2011; M. Orliac 1997, and numerous others). Candlenut (Aleurites moluccana) is another tree that was introduced to virtually all of the Polynesian archipelagos in prehistory (Whistler 2009, 30–2). This has been confirmed in Central East Polynesia by the presence of pollen and macro-remains in prehistoric sedimentary contexts in some locations (e.g., Kirch et al. 1995; M. Orliac 2003b; Prebble 2008, though examples are numerous), and in the Marquesas by the identification of nutshells in a rockshelter deposit at the Vaipikoau cave site on Ua Huka (Kirch 1973) that Sinoto (1966) has correlated with an early occupation at Hane. In the present study, candlenut shells also occurred in some contexts dating to before the 12th or 13th century AD. At Hakaea Beach, they were identified in about half of the samples in all of the major cultural strata. Only a few fragments were identified in each sample, with the exception of a large cache in a middle-period oven. Nutshell fragments were also present in several samples from the Anaho Teavau’ua South, which dates to the 15th century AD, and in small quantities in post-AD 1650 contexts at the paepae excavations. Together these findings provide evidence that Aleurites trees were introduced to the Marquesas in prehistory, and they appear to have been cultivated very early in the sequence. Candlenut shells were absent from all of the Hatiheu sites in this study, which is curious because the moist interior locations in this valley are ideal environments for the trees and they thrive there today. Overall, only one to three small fragments of candlenut shell occurred in most of the samples where they were found. The shells are extremely hard and durable once charred, and for this reason they were probably subject to preferential preservation. The presence of candlenuts in archaeological sediments are an indicator of their regular use in the area at some time in the past. However, one should be cautious before interpreting this to mean the

275 trees were cultivated locally. Because the nuts were frequently used in daily activities, they could have been obtained from some distance away and may have even been a trade commodity, therefore the presence of candlenut shells is not a reliable indicator the trees were growing near the site. There is also some evidence these materials can persist in the landscape for long periods of time. Sharp and colleagues (2010), for example, found that fill used in temple construction in the Society Islands contained candlenut shells that produced radiocarbon dates hundreds of years older than the structures. This finding suggests that candlenut shells may end up in secondary or tertiary contexts completely unrelated to their primary use. When considered together with the results of the present research, in which candlenut shells were found to be absent in places they were expected and sometimes frequent in places where they were not, it is evident that biogeographical interpretations of Aleurites need to be conservatively made. Perhaps more importantly, shell fragments found loose in archaeological sediments are probably a poor choice for radiocarbon dating.

Coconut Coconuts (Cocos nucifera) are one of the most important resources in tropical Oceania. While some coconuts are native to virtually all of the tropical Pacific Islands, as the large seeds are buoyant and can germinate even after long exposure to salt water, some varieties were introduced by settlers throughout region (Gunn, Baudouin, and Olsen 2011). Coconuts are one of the staple foods of Polynesia, providing both coconut meat and water, and they are used to produce a high-quality charcoal in Samoa (Whistler 2009). Coconut trees are mentioned along with breadfruit as the most common trees observed by early- contact era visitors to the Marquesas, which probably indicates they were planted in and around many coastal locations by this period. A mature coconut grove was observed on the western Anaho coastal flat by Robert Louis Stevenson (1987) in 1888, and by the mid- to late-19th century coconut groves were extensive throughout the Marquesan lowlands and produced copra for trade (Decker 1970, 92–3). In this study, coconut shell fragments were nearly ubiquitous in prehistoric samples from coastal and near-coastal settings throughout the sequences. Their frequent occurrence in archaeological contexts probably relates directly to their everyday use in domestic settings and common secondary use as fuel. Coconut shells also produce dense, hard charcoal, but these materials are brittle and tend to produce very small fragments. Coconut wood fragments also occurred occasionally in these contexts. While common in many

276 contexts, coconut was infrequent or altogether absent in samples from inland test units from the Hatiheu Valley and pre-AD 1650 contexts in the paepae test units at Anaho. These findings suggest coconut trees were probably common in many windward coastal and lowland situations throughout the prehistoric sequence, much as they were at contact. However, there is some indication that the trees may not have been frequent farther inland, or perhaps were not frequently used at these locations, an idea that is supported by observations of the younger Forster, who observed that coconuts were a more rare occurrence inland (G. Forster 1777, 24). The elder Forster further noted that (at Vaitahu, Tahuata) “Their houses are surrounded by plantations of breadfruit-Trees & Banana walks & a very few Coconut Trees are among them [emphasis added]ˮ (J. Forster 1982, 3: 490). That said as many accounts indicate, ships were provisioned with coconuts—sometimes in large quantities (e.g., Marchand 1810, 110)—and the trees must have been common in some locations. Coconut trees also may have been mixed in with others throughout the landscape (as noted by Krusenstern 1813, 124), in a manner similar to the highly cultivated landscape of Tikopia (Kirch and Yen 1982, 39), rather than clustered in monospecific groves.

Inocarpus Tahitian chestnut (Inocarpus fagifer) trees are considered to be a Polynesian introduction to the Marquesas. At present, there is little evidence in early-contact literature that Inocarpus trees were actively cultivated in the Marquesas by the historic period, but consumption of the nuts was noted by early visitors (Crook 2007, 75; Quirós 1904, 1:27), and G. Forster encountered a thick grove of the trees on Tahuata (G. Forster 1777, 2:20). These large trees are heavy producers (Elevitch 2006), and an early Marquesan visitor noted that they provided two crops per year (Lisiansky 1814, 91). In traditional Oceanic cultures, virtually all parts of the tree had uses ranging from craft and construction to medicinal preparations and animal fodder, and it also fixes nitrogen, thus enriching the soil (Elevitch 2006). The young leaves of this tree were also eaten as a green vegetable (E. Handy 1923, 200). Mature trees form large buttresses, and the wood is soft and light (Butaud, Gérard, and Guibal 2008, 402–3). Inocarpus was mentioned as a fuelwood by several sources (Elevitch 2006; Whistler 2004, 71–2), and use of large branches in a specialised type of earth oven was noted in the Marquesas by E. Handy (1923, 195). The nuts were observed to be an important food by the earliest Western visitors to Tahuata in 1595, who described them thusly: “Its shape is like a heart, flattened. They eat

277 many, roast and boiled, and leave them on the trees to ripen.” (Quirós 1904, 1:29). This observation corresponds temporally with the findings of this study, where Tahitian chestnut wood was identified in post-AD 1600 contexts. Together, these data provide evidence that Inocarpus were present and probably cultivated in the archipelago by the late prehistoric period. Inocarpus wood was, however, an infrequent occurrence in this study and the absence of this wood from pre-AD 1600 contexts is interesting considering the wood has some known uses as fuel. The anatomy of the wood is very distinctive and would not have been overlooked. It is possible the trees were not introduced to the Marquesas until later in the prehistoric sequence, some time before AD 1595. This suggestion is tentative as there are several reasons Inocarpus may not have been frequent in this assemblage. Firstly, there are only a few sites in this study near the moist locations where the trees are usually found (see W. Handy 1965, 180), such as the interior of Hatiheu and along the Teavau’ua stream in Anaho. However, the wood may have been useful enough to transport to coastal locations for light construction projects and, if that were the case, it would have been expected in some of the earlier deposits. Secondly, mature trees tend to self-prune and have bare trunks up to 4 m, so while the wood is a known fuel in some Pacific locations, branches or deadfall were probably not frequent. Lastly, the wood is very soft and may have completely combusted or produced poor quality charcoal.

Breadfruit Artocarpus altilis (breadfruit) trees are tall, lofty trees with large spreading leaves. Most of the varieties introduced to East Polynesia were sterile and reproduced via cuttings (Ragone 1991, 119–20). In addition to the fruit being a staple food, many parts of the trees were used for thatch, to make tapa cloth and to make small canoes, and the sticky sap had numerous medicinal and utilitarian uses (Whistler 2009). The wood is light and flexible and although it makes charcoal of very uneven quality, the anatomy is distinctive. Breadfruit wood was frequent, though usually not abundant, in the samples. It is probably under-represented in this study because the charcoal is somewhat friable. At the Hakaea Beach site, breadfruit occurs in contexts dating to approximately a century after vegetation clearance. By the end of the sequence at this site, which is potentially less than one hundred years later (c. 14th century AD), it was occurring frequently in the samples. In Hatiheu Valley, there is a similar trend at the Pahumano site. Although there were few late

278 period samples at this location, most contained breadfruit wood, which accounted for almost half of the samples, suggesting that breadfruit trees were probably more frequent near this site by the early 16th century AD. Other late prehistoric contexts from Hatiheu frequently contained breadfruit wood, indicating that by late prehistory the trees were frequent at inland locations. At Anaho, the sequence is much longer and extends into late prehistory. In the north valley, breadfruit wood follows a trend similar to the other study locations, but initially occurs at a later time. It was found very infrequently in pre-AD 1400 contexts and became somewhat more frequent in the middle period, then in post-AD 1650 contexts it was frequent and more abundant. In south valley samples, a similar trend was observed, although there were only a few samples from each temporal period in this location. Overall, these trends do not appear to have been influenced by differences in the quantity of samples analysed from each temporal period because, in some cases, they trend in the opposite direction (i.e., a smaller quantity of samples but increasing presences and abundances). When the findings of these assemblages are considered together, several important conclusions can be drawn. First, it is evident breadfruit was a very early introduction to the Marquesas. It was undoubtedly introduced to the island by the late 13th or early 14th century, and cultivation appears to have been extensive in some locations by the mid-15th century. These findings are notable as, until now, breadfruit had not been identified in any Marquesan contexts dating to before the late prehistoric period (see Millerstrom and Coil 2008). Another potential conclusion is that breadfruit trees were cultivated in windward valleys—probably in domestic areas near the coasts—only one or two centuries after settlement. Breadfruit charcoal was recovered from several early contexts at Hakaea Beach (dated to calAD 2σ 1268–1411) and it became prevalent in later contexts at this site, all of which date to before AD 1450. Scant amounts (1 fragment each) were also found in a Pahumano context dating to AD 1270–1393 and the Anaho Teavau’ua South dating to AD 1430–1480 (both 2σ), although it is worth noting that the earliest occupations of Anaho have not yet been located, indicating breadfruit cultivation was occurring in other locations on the island at this time as well. Contexts included in this study date to between 25–275 years after Allen and McAlister (2010) argue settlements were established throughout the archipelago, AD 1000–1250, and several hundred years after colonisation may have occurred. Because breadfruit wood makes a poor quality fuel, and it has few known uses in traditional Polynesian culture, this finding indicates the trees occurred frequently enough by these times to have produced deadfall or prunings that found their way into domestic fires.

279

Pruning has been practised, at times severely, in the region to gain easier access to fruit (Massal and Barrau 1956, 20). Lastly, breadfruit wood is much more frequent and abundant in samples from the late prehistoric period. A close analysis of the Anaho north valley assemblages, where samples were large enough to support such a conclusion, demonstrates this, as does the Hatiheu inland assemblage, which was smaller but also very informative. These two study locations have produced evidence that this trend was similar at windward locations of the same island that have contrasting catchment sizes and hydrology, and—by the late prehistoric period—social status and population densities.

Landscape domestication in the Marquesan prehistoric sequence The results of this study broadly define a chronology of lowland vegetation change in Marquesan prehistory. Four distinct processes were identified, and each was linked to archaeobotanical indicators and the human activities they represent. Through aspects of landscape domestication, climate, and social change, a picture of the changing human- landscape interactions in the valleys of windward Nuku Hiva emerges from this analysis. The individual catchments were charted separately to consider internal patterning, and then results were plotted on a unified timeline to broadly illustrate how these socio-natural systems evolved (Figure 8.2). It should be noted that the phases depicted in the following figures derive from a cultural history devised by Suggs (1961) and recently refined by Allen (2004b).

Prior to AD 1400 The earliest data analysed in this study derives from the Developmental and early Expansion Periods of Marquesan culture history. Sites from this time period are often located near the coast or watercourses, and there is later evidence of more permanent occupation with the construction of stone pavements, among other things (Allen and McAlister 2010; Rolett 1998; Sinoto 1966). This period is marked by dramatic declines in native avifauna. These changes correspond with a shift in subsistence strategies indicated by an increased reliance on introduced animals (Allen 2004b). By the middle of the Expansion period, deep sea fishing had declined (Rolett 1998, Phase III, 245–7). There is some evidence the use of plant materials became more important at this time.

280

Figure 8.2: Temporal process of vegetation transformation in the windward valleys of Nuku Hiva. Vertical dashed lines are outer boundaries of calAD 2σ ranges for sites in each catchment; black dashed lines indicate evidence was more ephemeral; grey dashed lines indicate other lines of evidence suggest prior activity in the valley.

AD 1100 1300 1500 1700 HAKAEA Vegetation clearance:

Coastal vegetation modifications:

Lowland forest modifications: ?

Cultivation of tree crops:

Settlement Developmental Expansion Classic Early contact

AD 1100 1300 1500 1700 HATIHEU Vegetation clearance:

Coastal vegetation modifications:

Lowland forest modifications:

Cultivation of tree crops:

Settlement Developmental Expansion Classic Early contact

AD 1100 1300 1500 1700 ANAHO Vegetation clearance:

Coastal vegetation modifications:

Lowland forest modifications:

Cultivation of tree crops:

Settlement Developmental Expansion Classic Early contact

AD 1100 1300 1500 1700 OVERALL Vegetation clearance:

Coastal vegetation modifications:

Lowland forest modifications:

Cultivation of tree crops:

Settlement Developmental Expansion Classic Early contact

The native vegetation of these areas appears to have been composed of low, open coastal forests that included large native trees. At several coastal sites (Hakaea, Hatiheu), there are indications that mostly undisturbed native vegetation communities were being

281 modified through the use of fire in this period. Clearance of these areas is linked to the creation of living and cultivation spaces. These activities appear to have been extensive and modifications were long-lasting, and within 200 years the surrounding vegetation had been significantly changed. Wood from large native trees was noted in multiple settings, indicating these trees were once part of the native coastal vegetation in the windward valleys of Nuku Hiva. Materials such as these were being exploited as fuelwood, although several native trees from the Sapotaceae family and Sapindus were the most frequently used woods during this period (when they were locally available). While there is evidence of breadfruit cultivation in coastal settings during this period, it is infrequent and it was probably cultivated on a small scale. Other tree crops such as coconut appear to have been widely cultivated at this time, and were probably more frequent in coastal settings, but not at all locations. These findings could also indicate that house gardens were important components of the early subsistence system. There is evidence people were modifying the inland forests of Anaho, and probably Hatiheu, to some extent at this time, although it is relatively modest. On the xeric slopes and ridges that are located in close proximity to several of the sites in this study, a low, open native vegetation community was present early in the cultural sequences. Fuelwood was also gathered from these formations, and this practice appears to have continued throughout the sequence, especially at Anaho. Candlenut is frequent at one early coastal site (Hakaea), indicating it was being cultivated—probably at locations farther inland—although there is also the possibility it was obtained from other areas through exchange.

AD 1400–1650 The middle period of this study corresponds with most of the Expansion period. This has been characterised as a time of expanding populations, and elaboration in both ceremonial centres and house platforms (Suggs 1961). During this period, the agronomic system was probably expanding. Open-sea fishing continued to decline, animal husbandry continued to expand, and a sharp drop-off in imported stone resources was noted (Rolett 1998, 86–8). These changes occurred against a backdrop of increasing climate variation and, later in this sequence, people began to move inland and build raised house foundations (Allen 2009a). Vegetation of coastal areas had been largely modified by this time. Although some native forest elements persisted, the ones that do not have known cultural uses became less

282 frequent and many native plants growing near the coast had been replaced with introduced taxa. Large native coastal trees were probably reduced or removed from coastal areas by this period, and large trees such as Casuarina occur for the first time (although very infrequently). These findings are linked to increasingly extensive efforts to create productive plant environments in both domestic settings and in the surrounding areas. There may have been some unintended consequences to some of these activities. Deposits at Pahumano indicate that the occasional firing of slopes continued through this period, which appears to have destabilised soils and caused erosion in some areas. The diversity of native taxa in the charcoal assemblages declined by the beginning of this period, and some Sapotaceae were infrequent or not encountered again. However, native forest elements do persist on slopes and ridges (probably in areas that were not burned) and at inland locations. There is evidence Sapindus continued to be frequently exploited as fuelwood where locally available, and Thespesia becomes more frequently used during this period. Thespesia is especially frequent and abundant in domestic food- preparation areas (Teavau’ua) in proximity to earth ovens, suggesting it was probably a preferred oven fuel. These findings may suggest it was encouraged as an important fuel, and also perhaps because it had become more valued for other cultural reasons. There is some indication vegetation communities were increasingly disturbed at both coastal and inland locations by this period. Hibiscus only became frequent enough to be important in the charcoal records at some near-coastal locations, and it may have been more common in moist interior locations in Hatiheu as well at this time. The presence of this and other regrowth species are linked to cultural activities such as forest clearance and regrowth. These processes occurred on multi-year (and perhaps longer) cycles so that trees could become established in the disturbed areas. The presence of some breadfruit wood in these samples may indicate reclamation of once-disturbed areas later in the sequence when they were needed to extend occupation or cultivation spaces, or perhaps for long-fallow swiddening schemes. It is possible some of this material may be ornamental Hibiscus that was cultivated in house gardens, as there is other evidence cultivation continued in domestic areas at this time. Tree crops such as breadfruit continue to contribute to the charcoal spectra in both coastal and inland locations, although they are more frequent in some areas in the Anaho north valley lowlands. Coconut continues to occur frequently and abundantly in some locations, but it is not ubiquitous and is notably infrequent or absent from some

283 assemblages. These trees were probably continuing components of a multi-faceted production system during this period.

After AD 1650 The late prehistoric period of this study corresponds with the Marquesan Classic, a phase marked by several important cultural changes. Elaborate stone architecture and ceremonial centres were built during this period (Millerstrom 2001; Suggs 1961). People moved inland, and several coastal sites including the Teavau’ua site appear to have been abandoned (Allen 2004b, 2009a). Populations grew and it probably became necessary to intensify food production. Extensive irrigated agricultural features are thought to date to this time (Addison 2006). The socio-political system also shifted as climate became increasingly variable (Allen 2010). By this period, vegetation clearance in coastal and lowland locations appears to have been very localised. There is some evidence of vegetation burning at several locations. Steep slopes near the coasts were still being burned (Pahumano), and a concentrated burn layer at an inland Hatiheu site may be evidence of clearance to reclaim overgrown areas for other purposes. The presence of breadfruit wood in the Hatiheu deposit indicates tree crops were being cultivated farther inland by this time. There is distinct evidence that lowland forest modifications were extensive, and the cultivated groves observed in the early contact period appear to have formed in this period. A dramatic increase in the frequency of tree crops, especially breadfruit, was observed in assemblages from this period. Some are noted for the first time, such as Inocarpus. In occupied areas, the native vegetation that remained largely represented those species with cultural uses. Native taxa are even less frequent, although it appears the inland forests of Anaho and Hatiheu did include a number of low, native trees and shrubs, suggesting these areas were not completely dominated by tree crops. Native vegetation elements co-existed near the Anaho coasts with introduced trees such as breadfruit and possibly Morinda, even after the mid-17th century. These findings indicate that the extensively modified coastal and lowland vegetation, which included extensive groves of breadfruit, bananas, and other trees, probably formed after the mid-17th century. It is probable that large fermented breadfruit stores became important at this time. The frequency of Hibiscus in late, inland Hatiheu locations may further indicate that regrowth species were

284 continually encroaching on these groves, making frequent maintenance necessary (clearing of underbrush). Fuelwood assemblages continue to be dominated by Sapindus and Thespesia in this period, with some variations based on local micro-climate conditions (i.e., Sapindus being more common in drier areas), and these materials probably continued to be harvested from locally occurring sources. There are still some indications that Thespesia was encouraged or perhaps cultivated to some extent. Thespesia is also frequent and abundant at the inland Hatiheu sites near these important cultural centres, and the additional occurrence of Calophyllum in these samples suggests culturally valued trees such as these were probably planted around ceremonial and high-status areas.

In the preceding discussion, several different processes of arboreal vegetation transformation that were observed through the study of wood charcoal have been identified and described, and I have linked them to temporally relevant aspects of social and natural systems. The four processes outlined provide a unique perspective on the changing relationships between human and plant populations in Marquesan prehistory. In the earliest contexts studied, vegetation clearance was linked to the creation of living and cultivation spaces and these changes were shown to be rapid and enduring. It is evident that elements of native vegetation were being exploited, perhaps intensively, while introduced trees such as breadfruit were also being cultivated on a small scale. Several hundred years later, as populations were growing and expanding, vegetation of coastal areas had been extensively remodelled and vegetation of inland areas was being modified to a greater extent. In the process of creating these domesticated landscapes, there is evidence that some native elements that may have been favoured fuelwoods had become rare in these areas. In late prehistory, there is evidence tree crops were being cultivated in many locations, including some far inland, and it has been suggested that some native trees (such as Thespesia) may have been being encouraged by this period. Throughout the sequence, there is evidence house gardens included arboreal elements such as breadfruit. The abundance of breadfruit wood in contexts from this period suggests that groves had become extensive, and probably corresponded with increased production of mā. Together, these findings shed light on the long-term dynamics in several aspects of prehistoric socio-ecosystems, including the development of a unique food-production system based on tree cultivation, which is the subject of the final discussion of this thesis.

285

The role of tree crops in prehistory At contact, tree plantations were widespread in Marquesan valleys. Breadfruit was noted to be synonymous with food, which visitors claimed was larger and of a better quality than on any of the other archipelagos (G. Forster 1777, 2:27), and dozens of varieties of seedless breadfruit had been engineered, most likely because they were preferred to create mā (fermented breadfruit paste) (Yen 1991, 87). It is evident that the cultivation of breadfruit had reached a pinnacle in the Marquesas, and this outcome was distinctly different than that at other locations in the East Polynesia. But we know little of how this system evolved, or what factors were most influential to its outcome. Tree crops may have been important from an early period and remained so throughout the sequence (as argued by Athens, Ward, and Murakami 1996 for Kosrae). Alternately, they may have increased in importance over time. It is also possible arboriculture was more important in some settings than others, perhaps as an adaptation to certain conditions in the local environment, or through historical contingencies. In the following discussion, I address these alternatives and consider what this new information on tree crops contributes to our understanding of the timing and pace of transitions in the Marquesan food production system. I then evaluate the main research question of this thesis, which was to identify which factors were most influential to the development of an arboricultural economy. To further demonstrate the importance of these influences, I consider the role of tree-cropping in the agronomic systems of another archipelago of Central East Polynesia.

Tree crops in the Developmental period Various arguments for the early importance of different cultivation methods have been proposed for the Marquesas. Addison (2006) presented a thesis that wetland taro (Colocasia) was an important cultivar early in the Marquesan sequence because it could be cultivated with little effort in naturally-occurring wetlands. This argument focused on the high yield potentials of both wet- and dryland taro, and the main assertion was that cultivation of taro was the most energy-efficient and fast producing component of the Pacific agricultural complex. An alternate scenario that has been proposed is that arboriculture was the foundation of the subsistence economy from a very early date. This suggestion was based on the results of a pilot study of wood charcoal from Hatiheu (Millerstrom and Coil 2008), where authors suggested that by the 14th century AD, breadfruit was established as a dominant crop in the middle Hatiheu Valley. There were,

286 however, few samples from these temporal contexts and some had the potential for inbuilt age, making this scenario an interesting possibility but one in need of further study. The present study provides a large dataset with which to further evaluate these ideas. It was predicted that if tree crops had become very important soon after settlement, a rapid increase in the frequency of tree crop wood would have been expected in all study areas. While this appears to have been the situation at Hakaea, the spatial area those samples represents is limited and conclusions regarding the wider valley cannot be made. In Anaho, the absence of breadfruit wood in pre-15th century contexts from several coastal and inland locations suggests that the trees were not frequent in all locations at this time. The occurrence of a single fragment of breadfruit in the numerous early Pahumano samples provides additional support for a conclusion that tree crops were probably not the dominant mode of production early in the Marquesan sequence. Even though there was little evidence to suggest breadfruit was central to the early production system, there are several indications that tree crops were cultivated very early in the Marquesan sequence, and this is an important finding. It is broadly evident that the cultivation of breadfruit, coconut, and candlenut was well underway in the early settlement period. Candlenut shells occur in some of the earliest contexts in Hakaea and Anaho, and coconut shell occurs in early samples at all locations. Breadfruit appears to have first been cultivated in several coastal settings contemporaneously, occurring within one to two centuries of the earliest occupations. Interestingly, this corresponds with the sharp decline in bird bones in the archaeological record at another Marquesan site (Steadman and Rolett 1996). This trend for breadfruit is most notable at Hakaea and there is some suggestion it was the typical pattern, as Hakaea was the most extensively sampled pre-15th century AD site in the study. However, it is also possible tree crops were cultivated sooner at Hakaea because the local climate was somewhat drier than other windward valleys and there is no permanent stream today. That said, there are presently small springs behind the beach ridge and it is possible the stream flowed more regularly in the past. Irrigated crops would have been less productive than arboreal crops in this setting, and land suitable for swidden cultivation would have been limited. These results provide evidence that tree crops did play an important role in early cultivation schemes, but they were probably only one part of the food-production system during this period (which roughly corresponds with the Developmental Period, Allen 2004b). Microfossil analysis has contributed evidence that a variety of starchy crops such as sweet potato, yam, breadfruit, and taro were being processed in prehistory, some before the

287

15th century (Allen and Ussher 2013). Together, these findings suggest swiddening, irrigated cultivation, and tree cropping were all being practised. More intensive archaeological investigations of relict fields, terraces, and irrigation features will be an important avenue for future research into this process.

Tree crops in the Classic period During the 16th and 17th centuries AD, there are a number of lines of evidence that suggest Marquesan society went through marked changes. Populations in the archipelago had expanded. Corresponding changes to social systems were occurring at this time, resources were stressed, and landforms were undergoing changes due to erosion, sedimentation, and shoreline buildup. These changes occurred along with a more variable climate, alternating wet and dry periods. In these wetter and stormier conditions, there is evidence people moved inland and built raised stone house foundations (Allen 2009a). By the 18th century AD, the unique socio-political system of shifting power structures that characterised Marquesan society at contact are thought to have emerged (Allen 2010). This period is marked by increased competition and conflict. As these pressures increased, there would have been a need to intensify production of the limited cultivable land available. As climate became more variable, there would also have been a need to minimise variance in food production to ensure supplies were available when crops failed. This study has provided new evidence against which to consider the changes that occurred in the food production practices and the Marquesan socio-economic system during the late prehistoric period. It was expected that if tree crops played an important role in the agronomic system at this time, the botanical remains of breadfruit, coconut, Inocarpus, and other taxa would be markedly more frequent, and probably more abundant, in post-17th century AD contexts. It became broadly evident, per the synthesis presented in Chapter Nine, that by this period a number of tree crops were being cultivated in several of the study sites. With the timing of this development established, it is possible to evaluate both environmental and temporally relevant influences that may have influenced this outcome.

Factors influencing a reliance on tree crops In the following discussion, I critically evaluate potential influencing factors with a particular focus on scale, patterning, and historical process. I consider whether this solution evolved in certain valleys in response to unique local conditions or histories, or whether it

288 was a broad adaptation to the Marquesan situation. Considering the complex interplay of topography, climate, and social systems that shaped Marquesan landscapes, a socio-natural systems approach best suits this research question.

Local conditions and/or historical contingencies The dynamics of Marquesan socio-ecosystems had many similarities across the three valleys considered here. All fall within a 15 km radius on the same island, and are broadly similar in their elevation and overall topography, with steep walls and narrow coastal plains. There are some differences in valley sizes and their variety of micro-climate conditions. Hatiheu is large, while Anaho and Hakaea are smaller. The climate of Hatiheu is more uniformly moist and the valley is well-watered, while parts of the other valleys have variable and at times dry conditions. The favourable cultivation conditions and presence of large ceremonial centres in Hatiheu suggest it probably supported a substantial population, and the evidence indicates arboreal cultivation was extensive there by the late prehistoric period. On the Hatiheu coast, breadfruit wood was abundant in three of the four samples from the upper strata at Pahumano, which was also a domestic setting. At inland sites in Hatiheu, this wood occurred in the majority (9/13) of late prehistoric samples from a largely high-status area. There is also evidence breadfruit trees were widespread in Anaho at this time, even though the population was probably smaller and local conditions were more varied. Breadfruit occurred in just under half (14/29) of the paepae samples from Anaho. It was sometimes abundant in sites distributed throughout the valley from the north coast to the south back valley wall. These sites represent a variety of different domestic activities, and include areas home to those of several different social statuses. When considered together, it is evident breadfruit trees were widely cultivated in different functional contexts in the windward valleys by the late prehistoric period. These findings suggest that variations in valley size, micro-climate conditions, probable population sizes, and social status were not factors that defined the extent or intensity of breadfruit cultivation in prehistory.

Intensification and the physical environment The results of the present study contribute information on the role of tree crops in the intensification process, as it is evident that breadfruit was an important component of the food-production system by the 17th century. Four major influences have been identified in

289 the model of agricultural intensification that has been developed for Polynesian prehistory: population pressure, human-induced landscape change as it relates to food productivity, increasing social stratification, and increasing conflict (Kirch 1984, Ch. 7, 1994). In the Marquesas, there is evidence several of these factors played an important role in the process. In this type of environment tree crops such as breadfruit and chestnut are prolific producers, able to provide several crops per year, and they would have been suitable choices with which to intensify production. Tree cultivation maximises the vertical capacity of the agricultural landscape (Kirch and Yen 1982; Latinis 2000, 50), which would have been important in the rugged Marquesan environment where little flat land was available for swidden cultivation, or in locations where a lack of fresh water would have limited irrigation systems. Arboriculture would also have been key to maximising labour efficiency, as established trees require far less maintenance than other techniques. Another advantage of cultivating trees such as breadfruit is that they would produce food to the upper limits of each valley. Historical evidence documents the cultivation of breadfruit at high elevations and on steep hillsides (Crook 2007, 140), indicating people were maximising the cultivable land in Marquesan valleys. The size and shape of the fruit were said to be similar to those grown at lower elevations, but the trees were diminutive and fruited at a later time (Crook 2007, 73–4). This practice would have in effect extended the fruiting seasons, ensuring a supply after the main breadfruit harvests. While this practice was not unique to the Marquesas (Banks 1768, 9 June 1769 and 3 July 1769), the innovation was uniquely suited to the Marquesan landscape, as the extensive area needed to cultivate starchy root crops such as taro and yams would have been limited by the smaller amounts of arable land in each valley. Trees had the additional advantage of stabilising and even enriching soils on slopes and ridges, an effect noted elsewhere in Polynesia (Kirch and Yen 1982, 40–3), which may have influenced their cultivation of the upper reaches of Marquesan valleys. To recap, the physical conditions and the need to extend and intensify food production do not sufficiently explain the increasing cultivation of breadfruit noted in these locations. When one considers that landscapes are the product of cumulative social and natural processes, there are other important influences, including the socio-political process and background climate change, that need to be considered.

290

Socio-political systems Breadfruit probably became important in the Marquesas for reasons other than its high yield and low maintenance benefits, however. In Marquesan society, breadfruit harvests were important social events and the processing of the fermented paste (mā) from the large first harvest of the year was a group effort (E. Handy 1923, 183). At harvest time and on many other occasions, chiefs displayed their wealth by staging large, elaborate feasts. These large, competitive feasts may have had a major influence in the intensification process (Thomas 1990, 101). Fresh breadfruit would have been an important component of these events, and large mā pits have been noted near important ceremonial centres (Millerstrom 2001, 247), suggesting it was probably also on the menu. These feasts would have had demanding production requirements, and the need to supply additional breadfruit for these events as they became more elaborate may have been an important contributor to the centrality of tree crops in the Marquesan agronomic system.

Climate and risk In the central Pacific, the late prehistoric period (after AD 1650) has been characterised as a period of increasing climate fluctuations, with conditions becoming warmer, wetter, and possibly more stormy, alongside highly fluctuating inter-annual variances (Allen 2006 and Chapter Three). An examination of modern climate records demonstrates that pronounced wet and dry cycles are known to occur in the Marquesas. When considered together with the increasingly variable climate conditions in the region in late prehistory, it appears these shifts became more extreme in late prehistory. In addition, rapid sediment accumulations were noted in a core from an upland Nuku Hiva location, leading Allen and colleagues (2011) to suggest that the climate was correspondingly wetter during this period. This situation may have had a significant influence on the evolution of the Marquesan socio-political process (Allen 2010). Extensive and intensive breadfruit cultivation may have also been a response to this increasingly unpredictable climate. Variance-minimisation, rather than productivity increases, has long been considered more important to survival under high-variance conditions (e.g., Allen 2004a; Dunnell 1989; Madsen, Lipo, and Cannon 1999). Tree cultivation would have provided an important stabilising insurance in the food production system (Yen 1974b, 278) under these conditions.

291

The increasingly high-variance climate would have affected the productivity of crops differently, but there would have been risks involved in employing any solution under these conditions. In very dry periods, fruits would have dropped from trees when stressed by a lack of water, and root crops would have failed in some areas when small streams or springs ran dry. Evidence that sedimentation increased in wetter times, as was noted near the Teavau’ua stream (Allen 2009a), indicates that upland locations without sufficient vegetation cover may have destabilised. This situation was typified at Pahumano, where it is evident that the repeated burning of the escarpment caused soils to destabilise, and events such as these would have increased rates of sedimentation in low-lying areas. Significant increases in rainfall would have made maintenance of irrigation systems challenging, and high water flows may have swelled streams and flooded irrigation systems (e.g., Millerstrom (2001) observed a large complex in western Hatiheu was submerged c. 1995). The cultivation of trees would have been a more effective way to minimise variance in the production system under these conditions. Trees are generally able to tolerate periods of drought or inundation better than herbaceous crops. Inocarpus (Tahitian chestnut) can withstand inundation and thrive in moist conditions. Breadfruit trees will drop fruit when stressed and only tolerate inundation for short periods, but they can withstand drought for three to four months (Thaman, Elevitch, and Wilkinson 2000). They also have the ability to regenerate quickly after damage; the roots will quickly produce shoots and large trees can produce fruits in as little as two years after severe loss of limbs (Elevitch 2006). Even so, breadfruit trees were known to die in severe droughts (Stevenson 1987, 49), and other innovations were necessary to ensure a supply of starchy staples would be available while fields and groves were being regenerated.

Preservation potential of breadfruit While it has been demonstrated that tree crops had distinct advantages in the Marquesan geography and climate, the preservation potential of breadfruit was probably a major influence on the development of its centrality as a cultivar. The ability to preserve large quantities of fruit en masse would have been important, as few tropical Pacific crops can be stored for any length of time. This practice has been identified elsewhere as a post- harvest form of intensification (Kirch and Yen 1982, 43–6). It would have allowed virtually all of the crop to be utilised, and provided a steady supply of food throughout the year. Furthermore, the ability to maintain these stores for many generations ensured food would

292 be available if extended droughts or environmental catastrophes ruined crops over multiple seasons. It is thought that these products would have remained edible after one hundred years of storage when properly cared for (E. Handy 1923, 188), which would have evened out variation in crop yield from season to season, as well as mitigated the risks associated with severe crop losses for the chief and his constituents.

Based on the aforementioned evidence, it is possible to conclude that a subsistence system based largely on arboreal cultivars evolved in this setting as an adaptation to a combination of rugged topography and climate change. The timing of the importance of a particular component of the agronomic system has been shown to correspond with increasing climate variance in the mid-17th century AD. Unpredictable conditions brought about by these changes influenced many aspects of Marquesan society, and in turn people made decisions that ultimately created lush, forested, and highly productive landscapes (G. Forster 2000, 2:336–7). Increased rates of erosion and sedimentation were tempered by the widespread cultivation of tree crops. The decision to intensify production of breadfruit was intricately linked to its high yields, labour-efficiency, and perhaps most importantly to its potential for long-term storage. Fermented breadfruit stores were an important component of the socio-political and economic system in terms of both everyday subsistence and as tools of elite power and influence.

Comparison to Society Islands It has been proposed that the great diversity of Polynesian production systems is closely correlated with a number of factors, including different environmental conditions, varying fluctuations in climate cycles, anthropogenic changes to landscapes, and degree of isolation (Kirch 1991a, 113–6). In the preceding analysis, I proposed the main influences in the Marquesas Islands were factors that are unique to the archipelago: extremely rugged topography and increasingly high-variance climate conditions, coupled with the role massive food stores played as an important component of the socio-economic system. In the following discussion, I consider how these factors influenced outcomes in another Central East Polynesian setting. Of particular interest was the assessment of the role of tree crops in the food production system, both in the prehistoric sequence and at contact, in another archipelago. A caveat to the following discussion is that our understanding of the development of agronomic systems, and the contribution of tree crops at any point in

293 prehistory, is not highly resolved in Central East Polynesia. This evaluation relies largely on early contact-era observations. Visitors noted that arboriculture was a highly developed component of the food production system throughout much of central Polynesia at this time, but some researchers have questioned whether these observations were biased, as visitors made sweeping generalisations from aboard their ships, rarely went very far inland, and mainly associated with elites (Lepofsky 1999). The Society Islands lie to the southwest of the Marquesan archipelago. Evidence from the windward high islands of this group, Tahiti and nearby Mo’orea, are considered. The windward Societies have distinct wet and dry seasons, but no extended dry periods (Ferdon 1981, 193). Tropical storms can be severe, but occur only four to eight times per century (Decker 1970, 42–5; Mueller-Dombois and Fosberg 1998, 387–8; Parkes and Flenley 1990, 3). The soils of these islands are fertile, and there are a range of different topographic conditions that include wide coastal plains, which were heavily cultivated. Large well-watered valleys on both islands have been studied by archaeologists, and these studies yielded some results that were useful for comparison. At contact, the agronomic system of Tahiti included tree cultivation, short-fallow swiddening of herbaceous crops, and wetfield cultivation of aroids. Trees were described as an important component of agriculture on the coastal plains (J. Forster 1996, 145). Lepofsky (1999) suggests that coastal groves may have actually been domestic gardens with a well- developed tree-cropping component. She further points out that Europeans seldom visited inland locations, and when they did so, they did not always recognise the complex food- production schemes being practised there. More reliable accounts describe small plantations of yam, taro, mulberry, and kava (Bligh 1937, 423), mosaics of home gardens, and groves of useful trees including coconut, mountain apple, and breadfruit, intermixed with bananas, mulberry, and “various roots” (G. Forster 2000, 2:187) among the houses. High up in the mountains, extensive groves of upright bananas were also cultivated (Banks 1768, 3 July 1769). Lepofsky (1999) notes that while these systems were often characterised as non- intensive and largely based on the cultivation of breadfruit, visitor accounts probably over- represented its importance. It is likely other forms of agriculture were also important and a diverse variety of taxa were cultivated. Several studies contribute to our understanding of agronomic practices in Society Islands prehistory. The evolution of the agronomic system in the ‘Opunohu Valley, Mo’orea has been outlined by Lepofsky (1994; Lepofsky, Kirch, and Lertzman 1996, Fig. 5), who noted that a mixture of techniques were used early in the sequence, including cultivation of

294 coconuts and small-scale swidden gardens. Terrace and barrage cultivation was evident by the 13th century AD. Later in the prehistoric sequence (possibly as early as the 13th century AD, but later in more marginal areas), some early swidden plots on the hillslopes were replaced by agricultural terraces where dryland taro and other annuals were planted. This study provided evidence breadfruit and Inocarpus (and possibly Syzygium) were being cultivated in upper valley locations after the turn of the 15th century AD, though there is some possibility for in-built age in these dates as the radiocarbon samples were run on unidentified charcoal. At this time, the vegetation of upper valley slopes was suggested to be a mixture of early successional vegetation and cultigens, including tree crops, both of which are components of mixed swidden cultivation schemes (Lepofsky 1994, 291–2). It was concluded that swidden cultivation and wetfield cultivation (as barrage irrigation systems) were components of a complex production system in prehistory, and these forms of agriculture were important later in the sequence (Lepofsky 1994, 300–2). Evidence for the continued cultivation of tree crops in prehistory was very modest in this study. There is some suggestion that food-producing trees (other than coconuts) were being cultivated in this setting from around AD 1400, as mentioned, and they were probably continually cultivated into historic period. But the scale of this component of the production system is uncertain. The charcoal assemblages were small, and little can be concluded about the frequency of trees such as Inocarpus, breadfruit, and Syzygium in the landscape at any point in the prehistoric sequence. While evidence for tree crops was absent from later contexts in the ‘Opunohu assemblages (Lepofsky 1994, 301–2), there is other evidence breadfruit was still cultivated on Mo’orea in late prehistory: breadfruit skin was identified in a mid-17th century AD context at a house site in ‘Opunohu (also see Kahn and Ragone 2013), and some breadfruit wood was identified in a late prehistoric to early historic period occupation in the Niuroa Valley, Mo’orea (Lepofsky 1994, 348). Other archaeological evidence of tree cultivation in the Societies derives from a site in the Papeno’o Valley of Tahiti. Though modest, this study demonstrated on the continued cultivation of breadfruit trees there from the 14th century AD through to the contact period (M. Orliac 1997). The site is situated on a promontory adjacent to a rockshelter and a river, near numerous stone foundations that were part of a high-status complex. The findings yielded a large charcoal assemblage, which derived from fire features including numerous earth ovens, which suggests some continuity in the cultivation of breadfruit over time in Papeno’o. A later prehistoric contribution from Inocarpus could indicate that tree crops

295 were consistently cultivated in the surrounding landscape, but there is some suggestion they were not frequent here at any period. Although the prehistoric evidence is modest, it does appear that breadfruit (and several other arboreal taxa) were cultivated earlier in the sequences in these locations, as they were in the Marquesas. While breadfruit wood continued to occur in later contexts, no particular temporal trends were observed. At contact in Tahiti, trees were being cultivated in well-developed home gardens along the coastal plain and at inland locations; reliable accounts indicate tree crops were not a central component, but were instead one part of a diverse agronomic system. While visitors noted that breadfruit was a favourite food of the elites, there appeared to have been several other options between harvests (Ferdon 1981, 182), and although breadfruit was preserved, the infrequent famines were mainly the result of conflict rather than crop failures (Ferdon 1981, 176). These findings suggest that in a setting where a variety of cultivation schemes could be managed, the production system was diverse. Furthermore, it appears that where climate was not highly variable and productivity was more reliable, large food stores did not evolve to become an important tool of political power and influence.

296

Chapter 9: Summary and Concluding Remarks

Overview The processes of landscape domestication differed throughout Polynesia, and the landscapes seen at contact were long-term products of both natural and cultural influences. Topography, soils, and hydrology were important influential factors that effectively set the stage for the transformations that occurred to both landforms and society over more than a millennium. During this time, climate, land, and human activities shifted and changed in a variety of ways as populations grew and people utilised island environments to their limits. It is remarkable that the societies encountered by early European visitors were, in many locations, thriving. People had developed methods to manage the limitations of the land area and resources they had available and, in the Marquesas in particular, to cope with a seemingly unpredictable climate that destroyed crops and affected food shortages every few generations. In this thesis, I set out to better understand how people dealt with these challenges, especially in regards to food production practices as they were refined over centuries (Chapters One and Two). The cultural and natural transformations that occurred over time were varied, complex, and related ultimately to the decisions people made from one generation to the next. As the motivations and needs of society changed, priorities shifted. My aim was to consider how landscapes and society were reflexively domesticated in this process, with a goal of identifying influences that shaped agronomic systems in island environments. It was argued that the study of woody vegetation was an ideal way to investigate the process of landscape domestication, and it would be especially informative in Oceanic settings. It was first demonstrated that tree crops were an important component of Oceanic production systems (Chapter Two). Polynesian settlers carried with them a suite of plants and could have developed diverse agronomic systems. However, in locations such as the Marquesas Islands, tree crops had become of paramount importance by the contact period— a unique outcome in the region. It was a concern that tree crops are the least understood component of prehistoric food-production systems in the region (but see Kennedy and Clarke 2004 for a recent overview of evidence from the southwest Pacific), and fossil wood assemblages had the potential to inform on these systems. The Marquesas were identified as an ideal study setting because the trees and shrubs native to these islands produce little palatable food, and a limited number of trees were introduced there in prehistory.

297

To provide context for this study, the natural and cultural history of the Marquesas was described in Chapter Three. Landforms, weather, hydrology, and vegetation zones were discussed to set the stage for this analysis. An overview of the distinctive configuration of climate cycles that affect this archipelago was presented to illustrate the conditions prehistoric people would have encountered. The possibility that short- and long-cycle climate regimes may have converged in this setting to create extreme wet and dry phases, which may have lasted for years, was discussed. The Marquesan cultural sequence was also presented (Chapter Three). This overview included a summary of the established phases of culture history, from early colonisation to the late prehistoric period. An ethnohistory of Marquesan culture was described as it was observed by early visitors and twentieth century ethnographers, including aspects of everyday domestic life, food-production practices, and management of landscape. The socio-political system and elaborate feasting rituals of the Marquesas were described. Finally, a brief arboreal ethnobotany was presented to inform on the traditional uses of trees and tree products in this setting, and to provide a better understanding of their role in social and ritual contexts. The archaeological investigations from windward Nuku Hiva, which represent at least a six hundred year prehistoric sequence from an early occupation of the archipelago in the 12th century AD (or possibly earlier) to the late 18th century, were summarised in Chapter Four. Study sites were located in three valleys of contrasting catchment size, topography, hydrology, and rainfall. Four stratified sites near the coast, and test units at numerous discrete inland settings adjacent to stone structures, were included in this study. Many are domestic contexts, but several sites were located in an area that has been identified as being home to higher-ranking members of Marquesan society. Functional contexts included occupation debris from living surfaces and a variety of other features, including numerous earth ovens and hearths. Analytical, field, and laboratory methods that were used to study the charcoal assemblages were presented in Chapter Five. Important aspects of studying these materials in archaeological contexts were considered, including formation processes, recovery of materials, and analytical methods. Anthracology was identified as a practice that focuses on the study of wood charcoal, and its usefulness in reconstructing past vegetation in other world regions was presented. It was determined there are several aspects of the method that need further development, including the refinement of anthracological methods for use in tropical regions. I contributed to this need by conducting an extensive review of the quality of data that was collected in the present study, taking into consideration how numerous

298 factors may have affected the charcoal assemblages. The extensiveness of the reference collection and the adequacy of the sampling were assessed, and it was demonstrated that several key assemblages in this study were probably good representatives of local vegetation communities in the past. This finding provided assurance that these data were suitable to address the larger research questions of the thesis. The Marquesan charcoal assemblages informed on vegetation communities in many areas, and were demonstrated to be rich and interesting sources of information on many different aspects of prehistoric life (Chapter Seven). Activities such as the routine collection of fuelwood and the regular clearance or regeneration of vegetation were evident, and processes such as the creation of living and cultivation spaces both near the coast and inland were also identified. These assemblages also informed on food production activities, from the cultivation of house gardens to the eventual construction of extensive arboreal plantations. It was noted that after AD 1650, coastal and lowland vegetation in the windward valleys had been significantly modified, and although a variety of native taxa persisted throughout the sequence, by this time useful economic plants had become an important component of the lowland forests. Introduced food-producing trees such as breadfruit, coconut, to a lesser extent Tahitian chestnut, and those with utilitarian uses such as candlenut were a part of these domesticated landscapes. There is also some suggestion that indigenous trees such as Thespesia populnea may have been encouraged later in the prehistoric period. Throughout the sequence, native trees were an important source of fuelwood, especially Sapindus saponaria, Thespesia, and a Sapotaceae (Unknown13), and there was evidence that some (such as the Sapotaceae) were significantly reduced within the first few centuries of settlement. The earliest assemblages also provided a picture of the native lowland vegetation that was not previously understood, and it was evident that some large native trees such as Calophyllum inophyllum and Cordia subcordata were eventually removed from coastal areas, with the wood of the former occurring later in high-status inland contexts. Four distinct processes of vegetation change were identified in the prehistoric sequence (Chapter Eight). By linking these to human activities at appropriate spatial and temporal scales, it became possible to understand how the larger process of landscape domestication occurred in this setting. These data provided perspective on the changing relationships between human and plant populations in Marquesan prehistory. They also permitted examination of long-term dynamics in several aspects of prehistoric socio-

299 ecosystems, including the development of a unique food-production system that was based on tree cultivation. In the main, there is evidence the Marquesans had created a well developed arboreal- based economic system by the late prehistoric period. The timing of this shift has been shown to correspond with the increasingly variable climate conditions of the mid-17th century AD. Decisions that people made about subsistence ultimately created the lush, forested, and highly productive landscapes seen at contact. The decision to intensify food production through the cultivation of breadfruit can be linked to the high yields that can be achieved on rugged land and the labour-efficiency of tree cropping, but it was argued that probably the most significant factor was the ability to store fermented fruit for very long periods of time. Fermented breadfruit stores played an important role in everyday subsistence, and they also became a key component of socio-political and economic systems in this setting, where they were eventually used to assert elite power and influence. Factors that were identified as influential to the Marquesan outcome were compared with those of the windward Society Islands to consider how the prehistoric processes and outcomes differed. Evidence of climate conditions and prehistoric agronomic sequences in these settings were assessed, although it was noted that the data available to evaluate arboreal components of the latter were modest. It was found that tree crops such as breadfruit were cultivated in the Societies from a comparably early period, but the evidence suggested that the agronomic system remained multi-faceted throughout the sequence. These findings indicate that when a variety of sub-systems could be managed, production systems were diverse, much as they are in other tropical areas today. Furthermore, it is suggested that in locations where climate was not highly variable, preserved food stores did not become important political currency.

Methodological contributions The present study is the first concentrated application of anthracological methods in Polynesian archaeology. These methods are well-developed for use in temperate and Mediterranean settings; others have shown that, with some refinement, they are also useful in tropical areas. In this formal evaluation, it was demonstrated that anthracology is an informative method to study vegetation histories in Polynesian settings. It has been demonstrated that when a methodological approach to charcoal analysis is utilised, vegetation histories can be produced at scales that compliment palynological studies, and

300 elements that are poorly represented in pollen spectra can be evaluated. Important information to further refine the methods for use in Pacific Island settings was provided; innovations were made in analysis, data presentation, and interpretive approaches. There are several specific methodological contributions of this research to the archaeobotanical sub-field of wood charcoal analysis. First, minimum sample sizes needed to effectively reconstruct vegetation in tropical Polynesian contexts were established. Progress was also made in developing an acceptable range of target values to test the reliability of fossil assemblages in these settings. It was further demonstrated that, as anthracologists already emphasise, results from feature and non-feature contexts should be evaluated separately because they inform on different spatial spheres and human activities. While anthracological methods are usually employed to study charcoal assemblages recovered by flotation, I have shown that extensive flotation programmes are not always necessary to produce useful and informative results, and materials recovered by dry screening can also be used effectively to study past vegetation histories. I have further demonstrated that it is possible to use previously collected charcoal to address important research questions, as long as these materials have been collected systematically. Wood charcoal analysts agree there is still much work that needs to be done to improve our understanding of how site formation, taphonomic processes, and taxonomy influence fossil charcoal assemblages. Until these processes are better understood, it was demonstrated that several different methods of data analysis can aid in understanding how archaeological charcoal assemblages have been affected. Namely, it is useful to perform simple statistical correlation tests to consider how various factors might have affected fragmentation, and the results of these tests can aid in determining the most suitable unit for quantitative analysis. The importance of reference collections, and the need to create a charred reference, was also stressed in order to improve our understanding of how various taxa and plant parts respond to charring. All of the aforementioned items are also important in considering what might be missing from an assemblage.

Substantive implications The intricate relationships between forests, people, and food have received much attention in recent decades, especially in the developing world (Hladik et al. 1993, xxi). Environmental and social scientists assert that understanding the complex bio-cultural interactions within tropical forest ecosystems is critical to understanding food production

301 practices on a global scale. Archaeological investigations can make particular contributions to this research because they provide deep time perspectives on human and landscape interactions, and the findings of the present study make several pertinent contributions in this regard. Arboriculture has sometimes been dismissed as difficult to see in the archaeological record (but see Athens, Ward, and Murakami 1996; Kim and Park 2014; Kirch 1989), but it has been demonstrated that arboreal cultivation practices can be studied through the analysis of wood charcoal. The history of breadfruit cultivation in windward Nuku Hiva locations was evident, and the timing of major developments were well resolved. The findings of this study provide evidence that tree cropping has been used as an intensification strategy in the past. These data support assertions that arboriculture is labour-efficient and can increase the productive capacity of landscapes that are otherwise too rugged or circumscribed to support other cultivation schemes. Tree crops such as breadfruit have the added benefit of being able to be stored en masse, allowing virtually all of the crop to be utilised in a ‘post-harvest intensification’ (Kirch and Yen 1982, 43–6). Progress has also been made in understanding the importance of tree-cropping solutions in high-variance climates in prehistory. This was the first empirical dataset used to consider the hypotheses that arboreal-based subsistence systems can be an important risk-management strategy (ideas presented by Latinis 1999, 2000), and it was found that extensive tree cultivation systems can be advantageous in these situations. But in locations such as the Marquesas, where climate fluctuations may have been of long duration, the importance of tree crops also hinged on their potential for long- term storage. By employing a rigorous methodological approach and studying data from multiple sites in controlled catchments, numerous variables in this study were held constant. This approach facilitated the consideration of how differences in local environments and histories may have affected the process of landscape domestication in prehistory, and the outcomes seen at contact. The findings can also be considered alongside evolving ideas about natural and cultural processes in Central East Polynesian prehistory. They have made an important contribution to our understanding of local vegetation histories and the role of humans in the process of landscape domestication in Polynesian prehistory. It was found that people rapidly and irrevocably modified the lowland forests of the study locations, turning them into productive landscapes by replacing forest elements with useful trees, probably without a substantial loss of vegetation cover. Several indigenous forest elements were found to have declined rapidly and appear to have eventually become extirpated in the process, and

302 these findings have notable parallels to the decline of avifauna throughout the region in prehistory. New information has also been provided on the history of both native and introduced woody plants that were useful to the Polynesians. This study has produced evidence that trees such as breadfruit, candlenut, and Tahitian chestnut were introduced to the archipelago in prehistory, which is especially important as palynological data from the islands have informed only modestly on arboreal taxa. The temporal contexts of these findings further suggest that not all plants were introduced at the same time. Human influence in the changing distribution patterns of certain native trees were also observed, and it appears that some were extirpated while others were encouraged. It is also evident that several large trees were indigenous components of coastal vegetation communities and one, Calophyllum inophyllum, whose status has been the subject of some debate, is most likely native to the archipelago. Lastly, a model of the process of landscape domestication in island contexts was outlined, which addresses the concern that few frameworks exist to understand the history of tree use or the perception of forests by humans (Asouti 2002, 12; Matthews and Gosden 1997, 129). By linking temporally relevant aspects of social and natural systems to these processes, it was shown that a model such as this can provide a unique perspective on the changing relationships between human and plant populations in prehistory. Highly resolved data from controlled spatial scales, as were used in this study, contribute to big ideas about the changes that occurred in pristine landscapes when humans colonised and settled the archipelagos of the Pacific Islands. This research aids in understanding the complex dynamics (and feedback mechanisms) that defined socio-natural interactions in prehistory.

Future directions This research has demonstrated that wood charcoal analysis has much to contribute to Pacific archaeology. Continuing study at other sites in the region would contribute greatly to our understanding of human-landscape interactions in prehistory, as well as provide information on the role of tree crops in the evolution of subsistence systems. To these ends, several important directions for future research are noted. First, the cultivation and procurement of fuelwood as an economic resource are often studied through the analysis of wood charcoal. An earlier pilot study demonstrated several important trends in fuel use and management (Huebert, Allen, and Wallace 2010), some of

303 which were confirmed by the present study and others that were revised. A more focused study of Polynesian fuelwood procurement and use is an important future direction of research. While advances have been made in the application of anthracology in the western Pacific and Australia (e.g., Byrne, Dotte-Sarout, and Winton 2013; Dotte-Sarout 2010a), additional work still needs to be done to refine the methods for effective use in Polynesian contexts. Firstly, to support the assessment of sampling adequacy, especially in regards to utilising the Gini Coefficient effectively, data on the modern vegetation communities of the Polynesian islands should be collected. While vegetation surveys have been conducted in some discrete areas (Gillespie et al. 2011), for this research to be most effective a variety of community types—including working agroforests—need to be surveyed. Secondly, studies of formation processes and charcoal taphonomy that are specific to the region would be very useful. Experimental work conducted in the region has focused on the study of charcoal formation and firing temperatures in Tahitian earth ovens (C. Orliac 1987; Orliac and Wattez 1989). Experimental burning in other contexts, such as hearths and vegetation clearance, should also be conducted. More systematic studies, including charring various woods under controlled laboratory conditions, would compliment this research. Further work is also needed to consider how wood charcoal is affected by the different types of sedimentary matrices that are found in the region, including calcareous sand and weathered volcanic basalt. The impact of reduced seed dispersals on forest communities due to rat predation (Athens et al. 2002; Meyer and Butaud 2009) and declining avian populations has been a major research theme in the region, especially regarding frugiverous pigeons, doves, and bats (e.g., McConkey and Drake 2002; Steadman and Freifeld 1999). Because it has been established that avifaunal extinctions occurred at virtually all Pacific Island locations settled by humans (Steadman 2006), charcoal analysis can contribute to these studies of plant- animal mutualisms in Pacific prehistory. It may be challenging to identify which plant species depended on extinct avian seed dispersers, however combining charcoal with zooarchaeolgical datasets would help address this. Collaborations with faunal analysts should prove to be a productive area of research. An important longer-term direction for future research is the development of practical models and useful theories that can be used to interpret wood charcoal datasets. Unlike other bioarchaeological disciplines, such as seed archaeobotany and zooarchaeology, wood charcoal analysis is still developing (see Veal 2013). Charcoal studies have moved

304 from describing background vegetation to making original, substantive contributions to archaeology. It is likely that the potential of wood charcoal analysis will be more fully recognised as more researchers contribute. Though there seems to be little crossover at present, much could be gained from a closer collaboration between anthracologists (who are more methodological) and palaeoethnobotanists (who take a more interpretive approach). Ultimately, these advances will be important to future interdisciplinary collaborations that seek to integrate and interpret diverse datasets. To conclude, this study has demonstrated that charcoal is useful for more than radiocarbon dating, and it is hoped that archaeologists will systematically collect and retain these materials in their site archives.

305

References cited Abbott, Isabella A. 1992. Lā’au Hawaiʻi: Traditional Hawaiian Uses of Plants. Honolulu, Hawaii: Bishop Museum Press. Adamson, Alastair M. 1936. Marquesan Insects: Environment. Bernice P. Bishop Museum Bulletin 139. Honolulu, Hawaii: The Museum. Addison, David J. 2006. ‘Feast or Famine? Predictability, Drought, Density, and Irrigation: The Archaeology of Agriculture in Marquesas Islands Valleys’. Ph.D. diss., University of Hawaii. Allen, Melinda S. 1984. ‘A Review of Archaeobotany and Palaeoethnobotany in Hawaii’. Hawaiian Archaeology 1: 19–30. ———. 1992. ‘Dynamic Landscapes and Human Subsistence: Archaeological Investigations on Aitutaki Island, Southern Cook Islands’. Ph.D. diss., University of Washington. University Microfilms International, Ann Arbor. ———. 1994. ‘The Chronology of Coastal Morphogenesis and Human Settlement on Aitutaki, Southern Cook Islands, Polynesia.’ Radiocarbon 36 (1): 59–71. ———. 1997. ‘Coastal Morphogenesis, Climatic Trends, and Cook Islands Prehistory’. In Historical Ecology in the Pacific Islands: Prehistoric Environmental and Landscape Change, edited by Patrick V. Kirch and Terry L. Hunt, 124–46. New Haven: Yale University Press. ———. 2004a. ‘Bet-Hedging Strategies, Agricultural Change, and Unpredictable Environments: Historical Development of Dryland Agriculture in Kona, Hawaii’. Journal of Anthropological Archaeology 23: 196–224. ———. 2004b. ‘Revisiting and Revising Marquesan Culture History: New Archaeological Investigations at Anaho Bay, Nuku Hiva Island’. Journal of the Polynesian Society 113: 143–96, 224–225 (errata). ———. 2005. ‘Periodicity, Duration and Function of Occupation at Tauroa Point, Northland, New Zealand’. New Zealand Journal of Archaeology 27: 19–62. ———. 2006. ‘New Ideas about Late Holocene Climate Variability in the Central Pacific’. Current Anthropology 47: 521–36. ———. 2008. ‘Environmental Variability and Socio-Political Process in the Marquesas Islands, French Polynesia’. Invited seminar, Stanford University, Dept. Anthropology, Special series,’ Evolutionary Approaches in Anthropology’. ———. 2009a. ‘Domestic Architecture, Socio-Political Process and Climate Variability in the Marquesas Islands, East Polynesia’. Asian Perspectives 48 (2): 342–82. ———. 2009b. ‘Variability in Megalithic Domestic Architecture as a Proxy for Socio- Political Change, Marquesas Islands’, 19th Congress of the Indo-Pacific Prehistory Association, Hanoi, Vietnam, Nov. 29 – Dec. 5, 2009. ———. 2009c. ‘Archaeological Research, Nuku Hiva, Marquesas Islands, French Polynesia: 2008 Annual Report’. ———. 2010. ‘Oscillating Climate and Socio-Political Process: The Case of the Marquesan Chiefdom, Polynesia’. Antiquity 84: 86–102. ———. 2014. ‘Marquesan Colonisation Chronologies and Postcolonisation Interaction: Implications for Hawaiian origins and the ‘Marquesan Homeland’ hypothesis'. Journal of Pacific Archaeology. Manuscript under review. Allen, Melinda S., and David Addison. 2002. ‘Prehistoric Settlement at Anaho Bay, Nuku Hiva, Marquesas Islands: Preliminary Observations’. Archaeology in Oceania 37: 87– 91. Allen, Melinda S., Kevin Butler, John Flenley, and Mark Horrocks. 2011. ‘New Pollen, Sedimentary, and Radiocarbon Records from the Marquesas Islands, East Polynesia: Implications for Archaeological and Palaeoclimate Studies’. The Holocene: 1–12.

306

Allen, Melinda S., and Jennifer M. Huebert. 2014. ‘Short-Lived Plant Materials, Long-Lived Trees, and Polynesian 14C Dating: Considerations for 14C Sample Selection and Documentation’. Radiocarbon 56 (1). Allen, Melinda S., and Jennifer G. Kahn. 2010. ‘Central East Polynesia: An Introduction’. Archaeology in Oceania 45: 49–53. Allen, Melinda S., and Andrew McAlister. 2004. ‘Archaeological Research at Anaho Bay, Nuku Hiva, Marquesas Islands, French Polynesia. 2004 Annual Report. Submitted to the Service de La Culture et Du Patrimoine, Punaauia, French Polynesia.’ ———. 2006. ‘Archaeological Research on Nuku Hiva, Marquesas Islands, French Polynesia. 2006 Annual Report Prepared for Ministère de La Culture de Polynésie Française Service de La Culture et Du Patrimoine.’ ———. 2007. ‘Recent Archaeological Studies in the Northern Marquesas, Nuku Hiva Island. Report to Ministére de La Culture de Polynésie Française Service de La Culture et Du Patrimoine.’ ———. 2008. ‘Archaeological Research on Nuku Hiva, Marquesas Islands, French Polynesia: 2007 Annual Report’. ———. 2010. ‘The Hakaea Beach Site, Marquesan Colonisation, and Models of East Polynesian Settlement’. Archaeology in Oceania 45: 54–65. ———. 2013. ‘Early Marquesan Settlement and Patterns of Interaction: New Insights from Hatiheu Valley, Nuku Hiva Island’. Journal of Pacific Archaeology 4 (1): 90–109. Allen, Melinda S., Andrew McAlister, David Addison, Shankar Aswani, Kevin Butler, and John Flenley. 2005. ‘Archaeological, Marine Ecological, and Palynological Research on Nuku Hiva, Marquesas Islands, French Polynesia: 2003–2004’. Allen, Melinda S., and Gail M. Murakami. 1999. ‘Lanai'i Island's Arid Lowland Vegetation in Late Prehistory’. Pacific Science 53: 88–112. Allen, Melinda S., and Ella Ussher. 2013. ‘Starch Analysis Reveals Prehistoric Plant Translocations and Shell Tool Use, Marquesas Islands, Polynesia’. Journal of Archaeological Science 40 (6): 2799–2812. Allen, Melinda S., and Rod Wallace. 2007. ‘New Evidence from the East Polynesian Gateway: Substantive and Methodological Results from Aitutaki, Southern Cook Islands’. Radiocarbon 49: 1163–79. Anderson, Atholl. 2009. ‘Epilogue: Changing Archaeological Perspectives upon Historical Ecology in the Pacific Islands’. Pacific Science 63: 747–57. Anderson, Atholl, and Yosihiko Sinoto. 2002. ‘New Radiocarbon Ages of Colonization Sites in East Polynesia’. Asian Perspectives 41: 242–57. Asouti, Eleni. 2002. ‘Charcoal Analysis from Çatalhöyük and Pınarbaşı, Two Neolithic Sites in the Konya Plain, South-Central Anatolia Turkey’. Ph.D., University College London . ———. 2003. ‘Woodland Vegetation and Fuel Exploitation at the Prehistoric Campsite of Pınarbaşı, South-Central Anatolia, Turkey: The Evidence from the Wood Charcoal Macro-Remains’. Journal of Archaeological Science 30: 1185–1201. Asouti, Eleni, and Phil Austin. 2005. ‘Reconstructing Woodland Vegetation and Its Exploitation by Past Societies, Based on the Analysis and Interpretations of Archaeological Wood Charcoal Macro-Remains’. Environmental Archaeology 10: 1– 18. Asouti, Eleni, and Jon Hather. 2001. ‘Charcoal Analysis and the Reconstruction of Ancient Woodland Vegetation in the Konya Basin, South-Central Anatolia, Turkey: Results from the Neolithic Site of Çatalhöyük East’. Vegetation History and Archaeobotany 10: 23–32. Aswani, Shankar, and Melinda S. Allen. 2009. ‘A Marquesan Coral Reef (French Polynesia) in Historical Context: An Integrated Socio-Ecological Approach’. Aquantic Conservation: Marine and Freshwater Ecosystems 19: 614–25.

307

Athens, J. Stephen. 2009. ‘Rattus exulans and the Catastrophic Disappearance of Hawaiʻi’s Native Lowland Forest’. Biological Invasions 11: 1489–1501. Athens, J. Stephen, H. David Tuggle, David J. Welch, and Jerome V. Ward. 2002. ‘Avifaunal Extinctions, Vegetation Change, and Polynesian Impacts in Prehistoric Hawaiʻi’. Archaeology in Oceania 37 (2): 57. Athens, J. Stephen, and Jerome V. Ward. 1993. ‘Environmental Change and Prehistoric Polynesian Settlement in Hawaiʻi’. Asian Perspectives 32. Athens, J. Stephen, Jerome V. Ward, and Gail M. Murakami. 1996. ‘Development of an Agroforest on a Micronesian High Island: Prehistoric Kosraean Agriculture’. Antiquity 70. Balée, William L. 1998. ‘Historical Ecology: Premises and Postulates’. In Advances in Historical Ecology, 13–29. New York: Columbia University Press. ———. 2006. ‘The Research Program of Historical Ecology’. Annu. Rev. Anthropol. 35: 75– 98. Balée, William L., and Clark L. Erickson, ed. 2006. Time and Complexity in Historical Ecology: Studies in the Neotropical Lowlands. The Historical Ecology Series. New York: Columbia University Press. Ballard, Chris, Paula Brown, R. Michael Bourke, and Tracy Harwood. 2005. The Sweet Potato in Oceania: A Reappraisal. Vol. 19. University of Sydney. Balme, Jane, and Wendy E. Beck. 2002. ‘Starch and Charcoal: Useful Measures of Activity Areas in Archaeological Rockshelters’. Journal of Archaeological Science 29: 157– 66. Banks, Joseph. 1768. The Endeavour Journal of Sir Joseph Banks. 2 vols. Sydney: State Library of NSW, Project Gutenberg. http://gutenberg.net.au. Barber, Ian. 2004. ‘Crops on the Border: The Growth of Archaeological Knowledge of Polynesian Cultivation in New Zealand’. In Change through Time: 50 Years of New Zealand Archaeology, edited by L. Furey and S. Holdaway, 169–92. New Zealand Archaeological Association Monograph 26. Barrau, Jacques. 1961. Subsistence Agriculture in Polynesia and Micronesia. Honolulu: Bernice P. Bishop Museum. ———. 1965. ‘L’humide et Le Sec: An Essay on Ethnobiological Adaptation to Contrastive Environments in the Indo-Pacific Area’. Journal of the Polynesian Society 74 (3): 329–46. Barton, C. Michael, Joan Bernabeu, J. Emili Aura, and Oreto Garcia. 1999. ‘Land-Use Dynamics and Socioeconomic Change: An Example from the Polop Alto Valley’. American Antiquity 64 (4): 609–34. Barton, C. Michael, Joan Bernabeu, J. Emili Aura, Oreto Garcia, Steven Schmich, and Luis Molina. 2004. ‘Long-Term Socioecology and Contingent Landscapes’. Journal of Archaeological Method and Theory 11 (3): 253–95. Baudouin, Luc, and Patricia Lebrun. 2008. ‘Coconut (Cocos nucifera L.) DNA Studies Support the Hypothesis of an Ancient Austronesian Migration from Southeast Asia to America’. Genetic Resources and Crop Evolution 56 (2): 257–62. Bell, Simon. 2012. Landscape: Pattern, Perception and Process. Abingdon, Oxon: Routledge. Bendtsen, B. Alan. 1978. ‘Properties of Wood from Improved and Intensively Managed Trees.’ Forest Products Journal 28: 61–72. Binford, Lewis R., ed. 1977. For Theory Building in Archaeology. New York: Academic Press. ———. 1980. ‘Willow Smoke and Dogs’ Tails: Hunter-Gatherer Settlement Systems and Archaeological Site Formation’. American Antiquity: 4–20. Bligh, William. 1937. The Log of the Bounty, Being Lieutenant William Bligh’s Log in a Voyage to the South Seas, to Take Breadfruit from the Society Islands to the West

308

Indies. 2 vols. London: Golden Cockerel Press. Boas, Franz. 1920. ‘The Methods of Ethnology’. American Anthropologist 22 (4): 311–21. Boserup, Ester. 1965. The Conditions of Agricultural Growth: The Economics of Agrarian Change under Population Pressure. London: Allen & Unwin. Boyd, W., M. McGlone, A.J. Anderson, and R.T. Wallace. 1996. ‘Late Holocene Vegetation History near Shag Mouth Archaeological Site’. In Shag River Mouth: The Archaeology of an Early Southern Maori Village, edited by A.J. Anderson, B. Allingham, and I.W.G. Smith, 257–75. Canberra: Australian National University. Boyd, William E., and Nigel Chang. 2010. ‘Integrating Social and Environmental Change in Prehistory: A Discussion of the Role of Landscape as a Heuristic in Defining Prehistoric Possibilities in Northeast Thailand’. In Altered Ecologies: Fire, Climate and Human Influence on Terrestrial Landscapes, edited by S.G. Haberle, J. Stevenson, and M. Prebble, 273–98. Canberra, A.C.T.: ANU E-Press. Braadbaart, Freek, and Imogen Poole. 2008. ‘Morphological, Chemical and Physical Changes during Charcoalification of Wood and Its Relevance to Archaeological Contexts’. Journal of Archaeological Science 35 (9): 2434–45. Braadbaart, Freek, Imogen Poole, and A. A. van Brussel. 2009. ‘Preservation Potential of Charcoal in Alkaline Environments: An Experimental Approach and Implications for the Archaeological Record’. Journal of Archaeological Science 36 (8): 1672–79. Braadbaart, Freek, Imogen Poole, Hans D.J. Huisman, and Bertil van Os. 2012. ‘Fuel, Fire and Heat: An Experimental Approach to Highlight the Potential of Studying Ash and Char Remains from Archaeological Contexts’. Journal of Archaeological Science 39 (4): 836–47. Bronk Ramsey, Christopher. 2009. Bayesian analysis of radiocarbon dates. Radiocarbon, 51 (1): 337–360. Brookfield, Harold. 1972. ‘Intensification and Disintensification in Pacific Agriculture: A Theoretical Approach’. Pacific Viewpoint 13 (1): 30–48. ———. 2001. ‘Intensification, and Alternative Approaches to Agricultural Change’. Asia Pacific Viewpoint 42 (2&3): 181–92. Brousse, R., J.-P. Chevalier, M. Denizot, and B. Salvat. 1978. ‘Etude Geomorphologique Des Îles Marquises’. Cachiers Du Pacifique 21: 9–74. Brown, F. B. H. 1922. The Secondary Xylem of Hawaiian Trees. Honolulu, Hawaii: Bishop Museum Press. ———. 1935. Flora of Southeastern Polynesia. Vol. 3: Dicotyledons. Bernice P. Bishop Museum Bulletin 130. Honolulu: Bernice P. Bishop Museum. Brown, F. B. H., and E. D. W. Brown. 1931. Flora of Southeastern Polynesia. 3 vols. Honolulu: Bernice P. Bishop Museum. Buck, Peter H. 1938. Ethnology of Mangareva. Honolulu: Bernice P. Bishop Museum. ———. 1957. Arts and Crafts of Hawaii. Honolulu: Bishop Museum Press. Bunting, M. J., and R. Tipping. 2000. ‘Sorting Dross from Data: Possible Indicators of Post- Depositional Assemblage Biasing in Archaeological Palynology’. In Human Ecodynamics: Proceedings of the Association for Environmental Archaeology Conference 1998 Held at the University of Newcastle upon Tyne, edited by G. Bailey, R. Charles, and N. Winder, 63–69. Oxford, UK: Oxbow. Burt, Deanne. 1999. ‘Prehistoric Marquesan Fishing in Regional Context’. M. A. Thesis, University of Auckland. Butaud, Jean-François. 2010. Nuku Hiva, Ua Huka, Ua Pou: Guide Floristique. Papeete, Tahiti: Direction de l’environnement de la Polynésie française. Butaud, Jean-François, Jean Gérard, and Daniel Guibal. 2008. Guide Des Arbres de Polynésie Française, Bois et Utilisations. Tahiti: Au Vent Des Iles. Butterfield, B. G., and B. A. Meylan. 1980. Three-Dimensional Structure of Wood: An Ultrastructral Approach. London: Chapman and Hall.

309

Butzer, Karl W. 1982. Archaeology as Human Ecology: Method and Theory for a Contextual Approach. Cambridge: Cambridge University Press. Byrne, Chae, Emilie Dotte-Sarout, and Vicky Winton. 2013. ‘Charcoals as Indicators of Ancient Tree and Fuel Strategies: An Application of Anthracology in the Australian Midwest’. Australian Archaeology (77). Cabarcos, Eloy, Jose-Abel Flores, and Francisco J. Sierro. 2014. ‘High-Resolution Productivity Record and Reconstruction of ENSO Dynamics during the Holocene in the Eastern Equatorial Pacific Using Coccolithophores’. The Holocene 24 (2): 176–87. Carlquist, Sherwin J. 2001. Comparative Wood Anatomy: Systematic, Ecological, and Evolutionary Aspects of Dicotyledon Wood. New York: Springer. Carrion, Yolanda. 2006. ‘Tres Montes (Navarra, Spain): Dendrology and Wood Uses in an Arid Environment’. In Charcoal Analysis: New Analytical Tools and Methods for Archaeology: Papers from the Table-Ronde Held in Basel 2004, edited by A. Dufraisse, 1483:83–94. Oxford: Archaeopress. Cashdan, Elizabeth A. 1990. ‘Introduction’. In Risk and Uncertainty in Tribal and Peasant Economies, edited by E. A. Cashdan, 1–16. Boulder: Westview Press. Cauchard, G., and J. Inchauspe. 1978. ‘Climatologie de L’archipel Des Marquises’. Cahiers Du Pacifique 21: 75–106. Chabal, Lucie. 1990. ‘L’étude Paléoécologique de Sites Protohistoriques à Partir Des Charbons de Bois: La Question de L’unité de Mesure. Dénombrements de Fragments Ou Pesées ?’ In Wood and Archaeology. Bois et Archéologie. First European Conference. Louvain-La-Neuve, October 2nd–3rd 1987, PACT No. 22, edited by T. Hackens, A.V. Munaut, and C. Till, 189–205. Strasbourg, PACT. ———. 1992. ‘La Représentativité Paléo-Écologique Des Charbons de Bois Archéologiques Issus Du Bois de Feu’. Bulletin de La Société Botanique de France 139: 213–36. Chabal, Lucie, L. Fabre, J.-F. Terral, and I. Théry-Parisot. 1999. ‘L’ Anthracologie’. In La Botanique, edited by C. Bourquin-Mignot, J.-E. Brochier, and L. Chabal, 43–104. Paris: Errance. Chagué-Goff, C., J. Goff, S.L. Nichol, W. Dudley, A. Zawadzki, J.W. Bennett, S.D. Mooney, et al. 2012. ‘Multi-Proxy Evidence for Trans-Pacific Tsunamis in the Hawaiʻian Islands’. Marine Geology 299–302: 77–89. Chase, A. K. 1989. ‘Domestication and Domiculture in Northern Australia: A Social Perspective’. In Foraging and Farming: The Evolution of Plant Exploitation, edited by David R. Harris and Gordon C. Hillman, 42–54. London: Unwin Hyman. Chrzavzez, Julia, Isabelle Thery-Parisot, Jean-Frédéric Terral, A. Ducom, and G. Fiorucci. 2011. ‘Differential Preservation of Anthracological Material and Mechanical Properties of Wood Charcoal, an Experimental Approach of Fragmentation’. In SAGVNTVM Extra, edited by E. Badal, Y. Carrion, E. Grau, M. Macias, and M. Ntinou, 11:29–30. Saguntum: Papeles Del Laboratorio de Arqueología de Valencia. València, Spain: Universitat de València. Chrzazvez, Julia, Isabelle Théry-Parisot, Gilbert Fiorucci, Jean-Frédéric Terral, and Bernard Thibaut. 2014. ‘Impact of Post-Depositional Processes on Charcoal Fragmentation and Archaeobotanical Implications: Experimental Approach Combining Charcoal Analysis and Biomechanics’. Journal of Archaeological Science 44. Clarke, William C. 1994. ‘Traditional Land Use and Agriculture in the Pacific Islands’. In The Science of Pacific Island Peoples. Suva: University of the South Pacific, 11–38. Suva, Fiji: Institute of Pacific Studies, The University of the South Pacific. Clouard, V., and A. Bonneville. 2005. ‘Ages of Seamounts, Islands, and Plateaus on the Pacific Plate’. Plates, Plumes and Paradigms: 71–90. Cobb, Kim M., Christopher D. Charles, Hai Cheng, and R. Lawrence Edwards. 2003. ‘El Nino/Southern Oscillation and Tropical Pacific Climate during the Last Millennium’. Nature 424: 271–76.

310

Coil, James H. 2004. ‘“The Beauty That Was”: Archaeological Investigations of Ancient Hawaiian Agriculture and Environmental Change in Kahikinui, Maui’. Ph.D. diss., Berkeley: University of California. Coil, James H., and Patrick V. Kirch. 2005. ‘An Ipomean Landscape: Archaeology and the Sweet Potato in Kahikinui, Maui, Hawaiian Islands’. In The Sweet Potato in Oceania: A Reappraisal, edited by Chris Ballard, Paula Brown, R. M. Bourke, and Tracy Harwood, 71–84. Sydney: University of Sydney. Collins, Matthew J., and Les Copeland. 2011. ‘Ancient Starch: Cooked or Just Old?’ Proceedings of the National Academy of Sciences 108 (22): E145–E145. Connor, M.A., and M.H. Viljoen. 1997. ‘Relationships Between Wood Density, Wood Permeability and Charcoal Yield’. In Developments in Thermochemical Biomass Conversion, edited by A. V. Bridgwater and D. G. B. Boocock, 82–96. London; New York: Blackie Academic & Professional. Cook, James. 1842. The Voyages of Captain James Cook. London: William Smith. Cowie, Hannah. 2009. ‘Enamel Hypoplasia in Humans and Their Commensals: Identification of Developmental Stress in the Marquesas Islands’. M. A. Thesis, University of Auckland. Crook, William Pascoe. 2007. An Account of the Marquesas Islands, 1797–1799. Edited by G. Dening, H-M. Le Cleac’h, D. Peacocke, and R. Koenig. Papeete, Tahiti: Haere Po. Crown, Andrea C. 2009. ‘Settlement Proxemics and Late Prehistoric Residential Organisation in Anaho Bay, Marquesas Islands’. M. A. Thesis, University of Auckland. Crumley, Carole L. 1993. ‘Historical Ecology: A Multidimensional Ecological Orientation’. In Historical Ecology : Cultural Knowledge and Changing Landscapes, edited by Carole L. Crumley, 1–16. Santa Fe, N.M.: School of American Research Press; Seattle: Distributed by the University of Washington Press. Cuddihy, L. W., and C. P. Stone. 1990. Alteration of Native Hawaiian Vegetation: Effects of Humans, Their Activities and Introductions. Honolulu, Hawaiʻi: University of Hawaii, Cooperative National Park Resources Studies Unit. Davidson, J. M., K. Fraser, B. F. Leach, and Y. H. Sinoto. 1999. ‘Prehistoric Fishing at Hane, Ua Huka, Marquesas Islands, French Polynesia’. New Zealand Journal of Archaeology 21: 5–28. Decker, Bryce G. 1970. ‘Plants, Man, and Landscape in Marquesan Valleys, French Polynesia’. Ph.D. diss., University of California, Berkeley. ———. 1991. ‘Secondary Plant Cover on Upland Slopes, Marquesas Islands, French Polynesia’. Atoll Research Bulletin 363. Delcourt, Hazel R., Paul A. Delcourt, and Thompson Webb III. 1982. ‘Dynamic Plant Ecology: The Spectrum of Vegetational Change in Space and Time’. Quaternary Science Reviews 1 (3): 153–75. Delcourt, Paul A., and Hazel R. Delcourt. 1980. ‘Pollen Preservation and Quaternary Environmental History in the Southeastern United States’. Palynology 4: 215–31. Del Castillo, Emmanuel Drake. 1892. Flore de la Polynésie française: description des plantes vasculaires qui croissent spontanément ou qui sont généralement cultivées aux Iles de la Société, Marquises, Pomotou, Gambier et Wallis. Paris: G. Masson. Denham, Tim. 2004. ‘The Roots of Agriculture and Arboriculture in New Guinea: Looking beyond Austronesian Expansion, Neolithic Packages and Indigenous Origins’. World Archaeology 36 (4): 610–20. Dening, Greg. 1980. Islands and Beaches: Discourse on a Silent Land: Marquesas, 1774– 1880. Carlton, Vic: Melbourne University Press. Détienne, Pierre, Paulette Jacquet, and Catherine Orliac. 1999. Manuel d’identification des Bois de Polynésie. Montpellier: CIRAD. Di Piazza, Anne. 1998. ‘Archaeobotanical Investigations of an Earth Oven in Kiribati, Gilbert Islands’. Vegetation History and Archaeobotany 7: 149–54.

311

———. 1999. ‘Te Bakoa Site. Two Old Earth Ovens from Nikunau Island (Republic of Kiribati)’. Archaeology in Oceania 34: 40–42. Dickinson, William R. 2003. ‘Impact of Mid-Holocene Hydro-Isostatic Highstand in Regional Sea Level on Habitability of Islands in Pacific Oceania’. Journal of Coastal Research 19: 489–502. Dincauze, Dena F. 2000. Environmental Archaeology: Principles and Practice. Cambridge: Cambridge University Press. Dotte-Sarout, Emilie. 2010a. ‘“The Ancestor Wood”: Trees, Forests and Precolonial Kanak Settlement on New Caledonia Grande Terre: Case Study and Anthracological Approach in the Tiwaka Valley (Northeastern Grande Terre)’. Ph.D. diss., Université de Paris / Sorbonne and Australia National University. ———. 2010b. ‘Perspectives on Pre-Colonial Human Introduction of Woody and Forest Species in the Pacific Islands: Prospective Ethno-Anthracological Research in New Caledonia’. In Selected Papers from the VII International Conference on Easter Island and the Pacific: Migration, Identity, and Cultural Heritage. Gotland University, Sweden August 20–25, 2007, edited by H. Martinsson-Wallin and P. Wallin, 431–54. The Gotland Papers 11. Gotland, Sweden: Gotland University Press. Dotte-Sarout, Emilie, John Ouetcho, Jacques Bolé, David Barret, and Christophe Sand. 2013. ‘Of Charcoals and Forests: Results and Perspectives from an Archaeobotanical Investigation of Precolonial Kanak Settlement Sites in New Caledonia’. In Pacific Archaeology: Documenting the Past 50,000 Years. Papers from the 2011 Lapita Pacific Archaeology Conference., 120–37. University of Otago Studies in Archaeology 25. Otago, New Zealand: University of Otago. Driver, Jonathan. 2001. ‘Environmental Archaeology Is Not Human Palaeoecology’. In Environmental Archaeology: Meaning and Purpose, edited by U. Albarella, 43–53. Dordrecht; Boston: Kluwer Academic Publishers; Norwell, MA. Dufraisse, Alexa. 2006. ‘Charcoal Anatomy Potential, Wood Diameter and Radial Growth’. In Charcoal Analysis: New Analytical Tools and Methods for Archaeology: Papers from the Table-Ronde Held in Basel 2004, edited by Alexa Dufraisse, 47–60. Oxford: Archaeopress. ———. 2011. ‘Interpretation of Firewood Management as a Socio-Ecological Indicator’. In SAGVNTVM Extra, edited by E. Badal, Y. Carrion, E. Grau, M. Macias, and M. Ntinou, 11:179–80. Saguntum: Papeles Del Laboratorio de Arqueología de Valencia. València, Spain: Universitat de València. Dufraisse, Alexa, and M. S. G. Martínez. 2011. ‘Mesurer Les Diamètres Du Bois de Feu En Anthracologie. Outils Dendrométriques et Interprétation Des Données.’ Anthropobotanica 02. Publications Scientifiques du Muséum National d’Histoire Naturelle, Paris. 1–18. Dunnell, Robert C. 1989. ‘Aspects of the Application of Evolutionary Theory in Archaeology’. In Archaeological Thought in America, edited by C. C Lamberg- Karlovsky, 35–49. Cambridge: Cambridge University Press. Dye, Thomas S. 1990. ‘The Causes and Consequences of a Decline in the Prehistoric Marquesan Fishing Industry’. Pacific Production Systems: Approaches to Economic Prehistory (18): 70–84. ———. 1996. ‘Assemblage Definition, Analytic Methods, and Sources of Variability in the Interpretation of Marquesan Subsistence Change’. Asian Perspectives 35 (1): 73–88. ———. 2010. ‘Social Transformation in Old Hawaiʻi: A Bottom-up Approach’. American Antiquity 75 (4): 727–41. Elevitch, Craig R., ed. 2006. Traditional Trees of Pacific Islands: Their Culture, Environment, and Use. Ho’olualoa, Hawaiʻi: Permanent Agriculture Resources. Ellen, Roy F. 1982. Environment, Subsistence and System: The Ecology of Small-Scale Social Formations. Cambridge: Cambridge University Press.

312

———. 1996. ‘The Cognitive Geometry of Nature’. In Nature and Society: Anthropological Perspectives, edited by Phillipe Descola and Gísli Pálsson, 103–24. London; New York: Routledge. Ellison, Joanna C. 1994. ‘Palaeo-Lake and Swamp Stratigraphic Records of Holocene Vegetation and Sea-Level Changes, Mangaia, Cook Islands’. Pacific Science 48 (1): 1–15. Emile-Geay, Julien, Kimberly M. Cobb, Michael E. Mann, and Andrew T. Wittenberg. 2013a. ‘Estimating Central Equatorial Pacific SST Variability over the Past Millennium. Part I: Methodology and Validation’. Journal of Climate 26 (7): 2302–28. ———. 2013b. ‘Estimating Central Equatorial Pacific SST Variability over the Past Millennium. Part II: Reconstructions and Implications’. Journal of Climate 26 (7): 2329–52. Erickson, Clark L. 2006a. ‘Intensification, Political Economy, and the Farming Community; In Defense Of A Bottom-Up Perspective Of The Past.’ In Agricultural Strategies, edited by J. Marcus and C. Stanish, 233–65. Los Angeles: Cotsen Institute. ———. 2006b. ‘The Domesticated Landscapes of the Bolivian Amazon’. In Time and Complexity in Historical Ecology: Studies in the Neotropical Lowlands, edited by W. Balée and C. L. Erickson, 235–78. New York: Columbia University Press. Evert, Ray F., and Katherine Esau. 2006. Esau’s Plant Anatomy: Meristems, Cells, and Tissues of the Plant Body: Their Structure, Function, and Development. Hoboken, N.J.: J. Wiley. Fairbairn, Andrew. 2005. ‘Simple Bucket Flotation and Wet-Sieving in the Wet Tropics. Palaeoworks Technical Papers 4’. Canberra: Palaeoworks, Australian National University. Fall, Patricia L. 2005. ‘Vegetation Change in the Coastal-Lowland Rainforest at Avai’o’vuna Swamp, Vava’u, Kingdom of Tonga’. Quaternary Research 64 (3): 451–59. Fall, Patricia L., Steven E. Falconer, and Lee Lines. 2002. ‘Agricultural Intensification and the Secondary Products Revolution along the Jordan Rift’. Human Ecology 30 (4): 445–82. Ferdon, Edwin N. 1981. Early Tahiti as the Explorers Saw It, 1767–1797. Tucson: University of Arizona Press. ———. 1993. Early Observations of Marquesan Culture, 1595–1813. Tucson: University of Arizona Press. Field, Julie S., Patrick V. Kirch, Kathy Kawelu, and Thegn N. Ladefoged. 2010. ‘Households and Hierarchy: Domestic Modes of Production in Leeward Kohala, Hawaiʻi Island’. The Journal of Island and Coastal Archaeology 5 (1): 52–85. Fietz, A. 1926. ‘Prähistorische Holzkohlen Aus Der Umgebung Brünns (I. Teil)’. Planta 2: 414–23. Figueiral, Isabel, and V. Mosbrugger. 2000. ‘A Review of Charcoal Analysis as a Tool for Assessing Quaternary and Tertiary Environments: Achievements and Limits’. Palaeogeography, Palaeoclimatology, Palaeoecology 164: 397–407. Fisher, Christopher T., James B. Hill, and Gary M. Feinman. 2009a. ‘Introduction: Environmental Studies for Twenty-First-Century Conservation’. In The Archaeology of Environmental Change: Socionatural Legacies of Degradation and Resilience, edited by C. T. Fisher, J. B. Hill, and G. M. Feinman, 1–14. University of Arizona Press. ———, ed. 2009b. The Archaeology of Environmental Change: Socionatural Legacies of Degradation and Resilience. Tucson: University of Arizona Press. Fisher, Jack B., Guillermo Goldstein, Tim J. Jones, and Susan Cordell. 2007. ‘Wood Vessel Diameter Is Related to Elevation and Genotype in the Hawaiian Tree Metrosideros polymorpha (Myrtaceae)’. American Journal of Botany 94: 709–15. Flenley, John R., A.S.M. King, J. Jackson, C. Chew, J. T. Teller, and M. E. Prentice. 1991.

313

‘The Late Quaternary Vegetational and Climatic History of Easter Island’. Journal of Quaternary Science 6 (2): 85–115. Florence, Jacques. 1997. Flore de La Polynésie Française. Vol. 1. Paris: Editions de l’Orstom. ———. 2004. Flore de La Polynésie Française. Vol. 2. Paris: Editions de l’Orstom. Florence, Jacques, and David H. Lorence. 1997. ‘Introduction to the Flora and Vegetation of the Marquesas Islands’. Allertonia 7. Folke, Carl. 2006. ‘Resilience: The Emergence of a Perspective for Social–ecological Systems Analyses’. Global Environmental Change 16 (3): 253–67. Ford, Richard. 1979. ‘Paleoethnobotany in American Archaeology’. In Advances in Archaeological Method and Theory, Vol. 2:285–336. New York: Academic Press. Forster, Georg. 1777. A Voyage Round the World in His Britannic Majesty’s Sloop, Resolution, Commanded by Capt. James Cook, During the Years 1772, 3, 4 and 5. Vol. 2. London: Printed for B. White, Fleet-Street J. Robson, Bond Street; P. Elmsly, Strand; and G. Robinson, Peter-noster Row. ———. 2000. A Voyage Round the World. Vol. 2. University of Hawaii Press. Forster, Johann R. 1982. The Resolution Journal of Johann Reinhold Forster, 1772–1775. Edited by Michael E. Hoare. London: Hakluyt Society. ———. 1996. Observations Made during a Voyage Round the World. Edited by Nicholas Thomas, Harriet Guest, and Michael Dettelbach. Honolulu: University of Hawaiʻi Press. Fosberg, F. Raymond. 1975. ‘Vascular Plants of Aitutaki’. In Almost-Atoll of Aitutaki: Reef Studies in the Cook Islands, South Pacific, edited by D.R. Stoddart and P.E. Gibbs, 74–84. Atoll Research Bulletin 190. Washington, D.C.: Smithsonian Institution. Frawley, Susan, and Sue O’Connor. 2010. ‘A 40,000 Year Wood Charcoal Record from Carpenter’s Gap 1: New Insights into Palaeovegetation Change and Indigenous Foraging Strategies in the Kimberley, Western Australia’. In Altered Ecologies: Fire, Climate and Human Influence on Terrestrial Landscapes, edited by S.G. Haberle, J. Stevenson, and M. Prebble, 32:299–322. Canberra, A.C.T.: ANU E-Press. Friedman, Janet. 1978. Wood Identification by Microscopic Examination: A Guide for the Archaeologist on the Northwest Coast of North America. Heritage record, no. 5. Victoria: British Columbia Provincial Museum. Froyd, Cynthia, Jessica Lee, Atholl Anderson, Simon Haberle, Peter Gasson, and Katherine Willis. 2010. ‘Historic Fuel Wood Use in the Galápagos Islands: Identification of Charred Remains’. Vegetation History and Archaeobotany 19: 207–17. Gelabert, Llorenç Picornell, Eleni Asouti, and Ethel Allué Martí. 2011. ‘The Ethnoarchaeology of Firewood Management in the Fang Villages of Equatorial Guinea, Central Africa: Implications for the Interpretation of Wood Fuel Remains from Archaeological Sites’. Journal of Anthropological Archaeology 30 (3): 375–84. Gillespie, Thomas W., Gunnar Keppel, Stephanie Pau, Jonathan P. Price, Tanguy Jaffré, Jean- Yves Meyer, and Kristin O’Neill. 2011. ‘Floristic Composition and Natural History Characteristics of Dry Forests in the Pacific’. Pacific Science 65 (2): 127–41. Giovas, Christina M. 2006. ‘No Pig Atoll: Island Biogeography and the Extirpation of a Polynesian Domesticate’. Asian Perspectives 45 (1): 69–95. Godwin, H., and A. G. Tansley. 1941. ‘Prehistoric Charcoals as Evidence of Former Vegetation, Soil and Climate’. Journal of Ecology 29: 117–26. Goff, James, Geoffroy Lamarche, Bernard Pelletier, Catherine Chagué-Goff, and Luke Strotz. 2011. ‘Predecessors to the 2009 South Pacific Tsunami in the Wallis and Futuna Archipelago’. Earth-Science Reviews 107 (1–2): 91–106. Goff, James, Bruce G. McFadgen, Catherine Chagué-Goff, and Scott L. Nichol. 2012. ‘Palaeotsunamis and Their Influence on Polynesian Settlement’. The Holocene 22 (9): 1067–69.

314

Goff, James, Steven Pearce, Scott L. Nichol, Catherine Chagué-Goff, Mark Horrocks, and Luke Strotz. 2010. ‘Multi-Proxy Records of Regionally-Sourced Tsunamis, New Zealand’. Geomorphology 118 (3–4): 369–82. Goldstein, D.J., and Izumi Shimada. 2013. ‘An Experimental and Ecological Approach to Modeling Ancient Fuel Use’. In Proceedings of the Fourth International Meeting of Anthracology: Brussels, 8–13 September 2008, Royal Belgian Institute of Natural Sciences, edited by Freddy Damblon, 121–32. BAR International Series 2486. Oxford: Archaeopress. Gonçalves, T.A.P., C.R. Marcati, and R. Scheel-Ybert. 2011. ‘Wood and Charcoal Anatomy in Species of the Brazilian Cerrado: Effect of Carbonization on Wood Structure’. In SAGVNTVM Extra, edited by E. Badal, Y. Carrion, E. Grau, M. Macias, and M. Ntinou, 11:51–52. Saguntum: Papeles Del Laboratorio de Arqueología de Valencia. València, Spain: Universitat de València. Gosden, Chris. 1995. ‘Arboriculture and Agriculture in Coastal Papua New Guinea’. Antiquity 69: 807–17. Gottwald, Helmut. 1983. ‘Wood Anatomical Studies of Boraginaceae (s.l.). I. Cordioideae’. IAWA Journal 4: 161–78. Gourdon, Wilfried. 2003. ‘Flore Pollinique Des Marquises (Polynésie Française): Étude de L’environnement Végétal Marquisien, Indications Paléoclimatologiques et Paléoethnologiques’. IRD (Institut de Recherche pour le Développement) UR055 – Paleotropique et Département de Préhistoire du Muséum national d’Histoire naturelle, USM 204 – « Les Hominidés au Quaternaire ». Govor, Elena. 2010. Twelve Days at Nuku Hiva: Russian Encounters and Mutiny in the South Pacific. Honolulu: University of Hawaiʻi Press. Grattan, John, and Robin Torrence. 2007. Living under the Shadow : Cultural Impacts of Volcanic Eruptions. Walnut Creek, Calif.: Left Coast Press. Graves, Michael W., and Gail M. Muakami. 1993. ‘The Identification of Charcoal from Archaeological Assemblages on Kaho’olawe: Implications for Reconstructing Prehistoric Vegetation’. Hawaii: The Kaho’olawe Island Conveyance Commission. Grayson, Donald K. 1984. Quantitative Zooarchaeology: Topics in the Analysis of Archaeological Faunas. Orlando: Academic Press. Greenlee, Diana M. 1992. ‘Effects of Recovery Techniques and Postdepositional Environment on Archaeological Wood Charcoal Assemblages’. In Deciphering a Shell Midden, 261–82. London: Academic Press. Gunderson, Lance H., and C. S. Holling, ed. 2002. Panarchy: Understanding Transformations in Human and Natural Systems. Washington, D. C.: Island Press. Gunn, B. F, L. Baudouin, and K. M Olsen. 2011. ‘Independent Origins of Cultivated Coconut (Cocos nucifera L.) in the Old World Tropics’. PloS One 6 (6): e21143. Guppy, H. B. 1903. Observations of a Naturalist in the Pacific between 1896 and 1899. Vol. 2. London, New York: Macmillan and Co. Hallé, Francis. 1978. ‘Cahiers Du Pacifique’. Cahiers Du Pacifique 21: 315–57. Halstead, Paul, and Glynis Jones. 1989. ‘Agrarian Ecology in the Greek Islands: Time Stress, Scale and Risk’. The Journal of Hellenic Studies 109: 41–55. Halstead, Paul, and John O’Shea. 2004. Bad Year Economics: Cultural Responses to Risk and Uncertainty. Cambridge: Cambridge University Press. Handy, E. S. C. 1923. The Native Culture in the Marquesas. Honolulu: Bernice Pauahi Bishop Museum. ———. 1930. Marquesan Legends. Honolulu, Hawaii: Bernice Pauahi Bishop Museum. Handy, Willowdean C. 1965. Forever the Land of Men; an Account of a Visit to the Marquesas Islands. New York: Dodd, Mead. Harris, David R., and Gordon C. Hillman, ed. 1989. Foraging and Farming: The Evolution of Plant Exploitation. London: Unwin Hyman.

315

Hastorf, Christine A., and Virginia S. Popper, ed. 1988. Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains. Chicago: University of Chicago Press. Hather, Jon G. 1992. ‘The Archaeobotany of Subsistence in the Pacific’. World Archaeology 24: 70–81. ———. 2000a. Archaeological Parenchyma. London: Archetype. ———. 2000b. Identification of Northern European Woods. London: Archetype Publications. Hather, Jon G., and Patrick V. Kirch. 1991. ‘Prehistoric Sweet Potato (Ipomoea Batatas) from Mangaia Island, Central Polynesia’. Antiquity: 887–93. Heer, Oswald. 1865. Die Pflanzen Der Pfahlbauten. Zürich: Neujahrsblatt Naturforsch. Ges. auf das Jahr. Henry, A., N. Valdeyron, L. Bouby, and I. Thery-Parisot. 2012. ‘History and Evolution of Mesolithic Landscapes in the Haut-Quercy (Lot, France): New Charcoal Data from Archaeological Contexts’. The Holocene 23 (1): 127–36. Hladik, C. M., A. Hladik, O. F. Linares, H. Pagezy, A. Semple, and M. Hadley, ed. 1993. Tropical Forests, People and Food: Biocultural Interactions and Applications to Development. Paris and Carnforth, UK: Unesco Paris and the Parthenon Publishing Group. Hoadley, Bruce. 1990. Identifying Wood: Accurate Results with Simple Tools. Newtown, Connecticut: Taunton Press. Horrocks, Mark, and R. B. Rechtman. 2009. ‘Sweet Potato (Ipomoea Batatas) and Banana (Musa sp.) Microfossils in Deposits from the Kona Field System, Island of Hawaii’. Journal of Archaeological Science 36 (5): 1115–26. Hubau, Wannes, Jan Van den Bulcke, Peter Kitin, Florias Mees, Geert Baert, Dirk Verschuren, Laurent Nsenga, Joris Van Acker, and Hans Beeckman. 2013. ‘Ancient Charcoal as a Natural Archive for Paleofire Regime and Vegetation Change in the Mayumbe, Democratic Republic of the Congo’. Quaternary Research 80 (2): 326–40. Hubau, Wannes, Jan Van den Bulcke, Peter Kitin, Florias Mees, Joris Van Acker, and Hans Beeckman. 2012. ‘Charcoal Identification in Species-Rich Biomes: A Protocol for Central Africa Optimised for the Mayumbe Forest’. Review of Palaeobotany and Palynology 171: 164–78. Hubbard, R. 1980. ‘Development of Agriculture in Europe and the Near East: Evidence from Quantitative Studies’. Economic Botany 34: 51–67. Huebert, Jennifer M. 2008. ‘Archaeobotanical Investigations of Wood Charcoal from Earth Ovens in the Marquesas Islands’. PgDip. diss., University of Auckland. ———. 2009. ‘The Reconstruction of Past Vegetation in the Anaho Valley, Nuku Hiva, Marquesas Islands Using Archaeological Wood Charcoal’. M. A. Thesis, University of Auckland. Huebert, Jennifer M., Melinda S. Allen, and Rod T. Wallace. 2010. ‘Polynesian Earth Ovens and Their Fuels: Interpretation of Wood Charcoal Remains from Anaho Valley, Nuku Hiva, Marquesas Islands’. Journal of the Polynesian Society 119: 61–97. Hugi, Erich, Marcel Wuersch, Walter Risi, and Karim Wakili. 2007. ‘Correlation between Charring Rate and Oxygen Permeability for 12 Different Wood Species’. Journal of Wood Science 53: 71–75. Hutchinson, G. E. 1953. The concept of pattern in ecology. Proceedings of the National Academy of Sciences (USA). 105: 1–12. Hynes, R. A., and A. K. Chase. 1982. ‘Plants, Sites and Domiculture: Aboriginal Influence upon Plant Communities in Cape York Peninsula’. Archaeology in Oceania 17 (1): 38–50. Ingold, Tim. 1993. ‘The Temporality of the Landscape’. World Archaeology 25 (2): 152–74. Ingraham, Joseph. 1810. ‘An Account of a Recent Discovery of Seven Islands in the South Pacific Ocean, by Joseph Ingraham, Citizen of Boston, and Commander of the

316

Brigantine Hope, of Seventy Tons Burthen; of and from This Port, Bound to the N. W. Coast of America. By Permission of the Owners, Copied from the Journal of Said Ingraham and Communicated to the Publick, by the Historical Society.’ In Collections of the Massachusetts Historical Society, for the Year 1793, edited by Massachusetts Historical Society, 20–24. Cornhill, Boston: Munroe & Francis. InsideWood. 2004–onwards. Published on the Internet. Accessed 28 February 2014. http://insidewood.lib.ncsu.edu/search. IUCN. 2013. ‘The IUCN Red List of Threatened Species. Version 2013.1’. Accessed 30 October 2013. http://www.iucnredlist.org. Jansen, Doris, Doris Mischka, and Oliver Nelle. 2013. ‘Wood Usage and Its Influence on the Environment from the Neolithic until the Iron Age: A Case Study of the Graves at Flintbek (Schleswig–Holstein, Northern Germany)’. Vegetation History and Archaeobotany 22 (4): 335–49. Jansen, Doris, and Oliver Nelle. 2012. ‘The Neolithic Woodland - Archaeoanthracology of Six Funnel Beaker Sites in the Lowlands of Germany’. Journal of Archaeological Science. Jansen, S., E. Robbrecht, H. Beeckman, and E. Smets. 2002. ‘A Survey of the Systematic Wood Anatomy of the Rubiaceae’. IAWA Journal 23(1): 1–67. Janusek, John W., and Alan L. Kolata. 2004. ‘Top-down or Bottom-up: Rural Settlement and Raised Field Agriculture in the Lake Titicaca Basin, Bolivia’. Journal of Anthropological Archaeology 23 (4): 404–30. Jardin, Edelston. 1857. ‘Essai Sur l’Histoire Naturelle de l’Archipel de Mendana Ou Des Marquises. 2eme Partie: Botanique.’ In Mémoires de La Société Impériale Des Sciences Naturelles de Cherbourg, 5:289–361. Paris: J.-B. Ballière. Johannessen, Sissel. 1988. ‘Plant Remains and Culture Change: Are Paleoethnobotanical Data Better Than We Think?’ In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 145–66. Chicago: University of Chicago Press. Johannessen, Sissel, and Christine A. Hastorf. 1990. ‘A History of Fuel Management (A.D. 500 to the Present) in the Mantaro Valley, Peru’. Journal of Ethnobiology 10: 61–90. Jones, Glynis, and Paul Halstead. 1995. ‘Maslins, Mixtures and Monocrops: On the Interpretation Ofarchaeobotanical Crop Samples of Heterogeneous Composition’. Journal of Archaeological Science 22 (1): 103–14. Jones, Julie, Heather Tinsley, and Richard Brunning. 2007. ‘Methodologies for Assessment of the State of Preservation of Pollen and Plant Macrofossil Remains in Waterlogged Deposits’. Environmental Archaeology 12 (1): 71–86. Jones, M. K. 1991. ‘Sampling in Palaeoethnobotany’. In Progress in Old World Palaeoethnobotany, edited by van Zeist, W., Wasylikowa, K., and Behre, K.-E, 53– 61. Balkema, Rottendam. Jones, P. D., and M. E. Mann. 2004. ‘Climate over Past Millennia’. Reviews of Geophysics 42 (2), 1–42. Jones, Volney H. 1941. ‘The Nature and Status of Ethnobotany’. Chronica Botanica 6 (10): 219–21. Kadane, Joseph. 1988. ‘Possible Statistical Contributions to Palaeoethnobotany’. In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 206–214. Chicago: University of Chicago Press. Kahn, Jennifer G. 2011. ‘Multi-Phase Construction Sequences and Aggregate Site Complexes of the Prehistoric Windward Society Islands (French Polynesia)’. The Journal of Island and Coastal Archaeology 6 (1): 24. Kahn, Jennifer G., and James H. Coil. 2006. ‘What House Posts Tell Us about Status Difference in Prehistoric Society: An Interpretation of Charcoal Analysis, Sacred

317

Woods and Inter-Site Variability’. Journal of the Polynesian Society 115 (4): 319–52. Kahn, Jennifer G., and Diane Ragone. 2013. ‘Identification of Carbonized Breadfruit (Artocarpus altilis) Skin: Refining Site Function and Site Specialization in the Society Islands, East Polynesia’. Journal of Ethnobiology 33 (2): 237–58. Keast, Allen, and Scott Everett Miller. 1996. The Origin and Evolution of Pacific Islands Biotas, New Guinea to Eastern Polynesia: Patterns and Processes. Amsterdam, The Netherlands: SPB Academic Pub. Keepax, Carole A. 1988. ‘Charcoal Analysis, with Particular Reference to Archaeological Sites in Britain’. Ph.D. diss., University of London. Kellum, Marimari. 1968. ‘Sites and Settlement in Hane Valley, Marquesas’. M. A. Thesis, University of Hawaiʻi. Kennedy, Jean, and William Clarke. 2004. ‘Cultivated Landscapes of the Southwest Pacific’. RMAP Working Papers No. 60. Canberra: Resource Management in Asia-Pacific Program, Research School of Pacific and Asian Studies, Australian National University. Kim, Minkoo, and Jungjae Park. 2014. ‘Vegetation History, Agricultural Intensification, and Changes in Wood-Resource Utilization in Ancient Southwest Korea (1500 BC–AD 700)’. The Holocene 24 (1): 118–29. Kirch, Patrick V. 1973. ‘Prehistoric Subsistence Patterns in the Northern Marquesas Islands, French Polynesia’. Archaeology and Physical Anthropology in Oceania 8: 24–40. ———. 1984. The Evolution of the Polynesian Chiefdoms. New Studies in Archaeology. Cambridge: Cambridge University Press. ———. 1989. ‘Second Millennium B.C. Arboriculture in Melanesia: Archaeological Evidence from the Mussau Islands’. Economic Botany 43: 225–40. ———. 1991a. ‘Polynesian Agricultural Systems’. In Islands, Plants, and Polynesians: An Introduction to Polynesian Ethnobotany: Proceedings of a Symposium, edited by Paul A. Cox and Sandra A. Banack, 113–33. Portland: Dioscorides Press. ———. 1991b. ‘Chiefship and Competitive Involution: The Marquesas Islands of Eastern Polynesia’. In Chiefdoms: Power, Economy, and Ideology, edited by Timothy K. Earle, 119–45. Cambridge: Cambridge University Press. ———. 1994. The Wet and the Dry: Irrigation and Agricultural Intensification in Polynesia. Chicago: University of Chicago Press. ———. 1999. ‘Response to: Intensification in the Pacific: A Critique of the Archaeological Criteria and Their Application’. Current Anthropology 40: 326–28. ———. 2000. On the Road of the Winds: An Archaeological History of the Pacific Islands Before European Contact. Berkeley, Calif.: University of California Press. ———. 2006. ‘Agricultural Intensification: A Polynesian Perspective’. In Agricultural Strategies, edited by J. Marcus and C. Stanish, 191–217. Los Angeles: Cotsen Institute of Archaeology at UCLA. ———. 2007a. ‘Three Islands and an Archipelago: Reciprocal Interactions between Humans and Island Ecosystems in Polynesia’. Earth and Environmental Science Transactions of the Royal Society of Edinburgh 98: 85–99. ———. 2007b. ‘Hawaii as a Model System for Human Ecodynamics’. American Anthropologist 109 (1): 8–26. ———., ed. 2011. Roots of Conflict: Soils, Agriculture, and Sociopolitical Complexity in Ancient Hawaiʻi. Santa Fe: School for Advanced Research Press. Kirch, Patrick V., and Joanna Ellison. 1994. ‘Palaeoenvironmental Evidence for Human Colonization of Remote Oceanic Islands’. Antiquity 68: 310–12. Kirch, Patrick V., and Terry L. Hunt. 1997. Historical Ecology in the Pacific Islands: Prehistoric Environmental and Landscape Change. New Haven: Yale University Press. Kirch, Patrick V., and Jennifer Kahn. 2007. ‘Advances in Polynesian Prehistory: A Review

318

and Assessment of the Past Decade (1993–2004)’. Journal of Archaeological Research 15: 191–238. Kirch, Patrick V., and Marshall D. Sahlins. 1994. Anahulu: Anthropology of History in the Kingdom of Hawaii. 2 Vols. Chicago: University of Chicago Press. Kirch, Patrick V., David W. Steadman, Virginia L. Butler, Jon Hather, and Marshall I. Weisler. 1995. ‘Prehistory and Human Ecology in Eastern Polynesia: Excavations at Tangatatau Rockshelter, Mangaia, Cook Islands’. Archaeology in Oceania 30 (2): 47– 65. Kirch, Patrick V., P. M. Vitousek, O. A. Chadwick, S. Tuljapurkar, T. Ladefoged, S. Hotchkiss, and M. W. Graves. 2011. ‘The Hawaiʻi Biocomplexity Project in Retrospect’. In Roots of Conflict: Soils, Agriculture, and Sociopolitical Complexity in Ancient Hawaiʻi, edited by P. V. Kirch, 163–72. Santa Fe: School for Advanced Research Press. Kirch, Patrick V., and Douglas E. Yen. 1982. Tikopia: The Prehistory and Ecology of a Polynesian Outlier. Honolulu: Bishop Museum Press. Kirch, Patrick V., and K. S. Zimmerer. 2011. ‘Dynamically Coupled Human and Natural Systems: Hawaiʻi as a Model System’. In Roots of Conflict: Soils, Agriculture, and Sociopolitical Complexity in Ancient Hawaiʻi, edited by P. V. Kirch, 3–30. Santa Fe: School for Advanced Research Press. Kolata, Alan L. 1996. Tiwanaku and Its Hinterland: Archaeology and Paleoecology of an Andean Civilization. Vol. 1: Agroecology. Washington: Smithsonian Institution Press. ———. 2002. Tiwanaku and Its Hinterland: Archaeology and Paleoecology of an Andean Civilization. Vol. 2: Urban and Rural Archaeology. Washington: Smithsonian Institution Press. Kolb, M. J., and Gail M. Murakami. 1994. ‘Cultural Dynamics and the Ritual Role of Woods in Pre-Contact Hawaii’. Asian Perspectives 33 (1): 57–78. Kribs, David A. 1935. ‘Salient Lines of Structural Specialization in the Wood Rays of Dicotyledons’. Botanical Gazette 96 (3): 547–57. Krusenstern, Adam Johann von. 1813. Voyage Round the World, in the Years 1803, 1804, 1805, & 1806, by Order of His Imperial Majesty Alexander the First, on Board the Ships Nadeshda and Neva, under the Command of Captain A. J. von Krusenstern of the Imperial Navy. Translated by Richard Belgrave Hoppner. Vol. 1. London: Printed by C. Roworth for J. Murray. Kukachka, B. F. 1982. ‘Wood Anatomy of the Neotropical Sapotaceae. XXXI. Pouteria’. Research Paper FPL 419. Madison, WI: Forest Products Laboratory. Lachenbruch, Barbara, John R. Moore, and Robert Evans. 2011. ‘Radial Variation in Wood Structure and Function in Woody Plants, and Hypotheses for Its Occurrence’. In Size- and Age-Related Changes in Tree Structure and Function, edited by F. C. Meinzer, B. Lachenbruch, T. E. Dawson, 121–64. Dordrecht, Netherlands; New York: Springer. Ladefoged, Thegn N., and Michael W. Graves. 2000. ‘Evolutionary Theory and the Historical Development of Dry-Land Agriculture in North Kohala, Hawaiʻi’. American Antiquity 65 (3): 423–48. ———. 2008. ‘Variable Development of Dryland Agriculture in Hawaiʻi: A Fine-Grained Chronology from the Kohala Field System, Hawaiʻi Island’. Current Anthropology 49 (5): 771–802. Ladefoged, Thegn N., Michael W. Graves, and James H. Coil. 2005. ‘The Introduction of Sweet Potato in Polynesia: Early Remains in Hawaiʻi’. The Journal of the Polynesian Society 114 (4): 359–73. Ladefoged, Thegn N., Michael W. Graves, and Mark D. McCoy. 2003. ‘Archaeological Evidence for Agricultural Development in Kohala, Island of Hawaiʻi’. Journal of Archaeological Science 30 (7): 923–40. Ladefoged, Thegn, Christopher Stevenson, Sonia Haoa, Mara Mulrooney, Cedric Puleston,

319

Peter Vitousek, and Oliver Chadwick. 2010. ‘Soil Nutrient Analysis of Rapa Nui Gardening’. Archaeology in Oceania 45 (2): 80–85. Lamberton, A. R. H. 1955. ‘The Anatomy of Some Woods Utilized by the Ancient Hawaiians’. M. A. Thesis, University of Hawaii, Mānoa. Lamoureux, C. H. 1985. Wood Anatomy – Scanning Electron Microscopy Image Collection. Joseph F. Rock Herbarium, University of Hawaii Museum Consortium, Honolulu, HI http://www.flickr.com/photos/uhmuseum/sets/72157628336981493/. Lancelotti, Carla, Marco Madella, P. Ajithprasad, and Cameron A. Petrie. 2010. ‘Temperature, Compression and Fragmentation: An Experimental Analysis to Assess the Impact of Taphonomic Processes on Charcoal Preservation’. Archaeological and Anthropological Sciences 2 (4): 307–20. Langsdorff, G. H. 1813. Voyages and Travels in Various Parts of the World, during the Years 1803, 1804, 1805, 1806, and 1807. Vol. 1. London: Printed for Henry Colburn. ———. 1968. Voyages and Travels in Various Parts of the World. Amsterdam: Israel. Latinis, D. Kyle. 1999. ‘Subsistence System Diversification in Southeast Asia and the Pacific: Where Does Maluku Fit?’ Ph.D. diss., University of Hawaiʻi. ———. 2000. ‘The Development of Subsistence System Models for Island Southeast Asia and Near Oceania: The Nature and Role of Arboriculture and Arboreal-Based Economies’. World Archaeology 32 (1): 41–67. Leach, Foss, J. Davidson, M. Horwood, and P. Ottino. 1997. ‘The Fishermen of Anapua Rock Shelter, Ua Pou, Marquesas Islands’. Asian Perspectives 36 (1): 51–66. Leach, Foss, and Helen M. Leach. 1979. Prehistoric Man in Palliser Bay. Wellington, N.Z., National Museum of New Zealand. Leach, Helen M. 1999. ‘Intensification in the Pacific: A Critique of the Archaeological Criteria and Their Application’. Current Anthropology 40 (3): 311–39. Lee, Gyoung-Ah. 2012. ‘Taphonomy and Sample Size Estimation in Paleoethnobotany’. Journal of Archaeological Science 39 (3): 648–55. Leney, Lawrence, and Richard W. Casteel. 1975. ‘Simplified Procedure for Examining Charcoal Specimens for Identification’. Journal of Archaeological Science 2: 153–59. Lennstrom, Heidi A. 1995. ‘Assessing Pacific Palaeoethnobotany: Advancing Quantitative Analysis of Hawaiian Archaeobotanical Remains’. Unpublished manuscript. Lennstrom, Heidi A., and Christine A. Hastorf. 1995. ‘Interpretation in Context: Sampling and Analysis in Paleoethnobotany’. American Antiquity 60 (4): 701–21. Lennstrom, Heidi A., and Gail M. Murakami. 2003. ‘Charcoal Samples from Three Environmental Zones’. In Imu, Adzes and Upland Agriculture, edited by L. L. Hartzell, S. A. Lebo, H. A. Lennstrom, S. P. McPherron, and D. I. Olszewski, 259–77. Honolulu: Bishop Museum. Leonard, R.D. 1987. ‘Incremental Sampling in Artifact Analysis’. Journal of Field Archaeology 14 (4): 498–500. Lepofsky, Dana. 1994. ‘Prehistoric Agricultural Intensification in the Society Islands, French Polynesia’. Ph.D. diss., University of California, Berkeley. ———. 1995. ‘A Radiocarbon Chronology for Prehistoric Agriculture in the Society Islands, French Polynesia.’ Radiocarbon 37 (3): 917–30. ———. 1999. ‘Gardens of Eden? An Ethnohistoric Reconstruction of Ma‘ohi (Tahitian) Cultivation’. Ethnohistory 46 (1): 1–29. Lepofsky, Dana, and J. Kahn. 2011. ‘Cultivating an Ecological and Social Balance: Elite Demands and Commoner Knowledge in Ancient Ma‘ohi Agriculture, Society Islands’. American Anthropologist 113 (2): 319–35. Lepofsky, Dana, Patrick V. Kirch, and Kenneth P. Lertzman. 1996. ‘Stratigraphic and Paleobotanical Evidence for Prehistoric Human-Induced Environmental Disturbance on Mo’orea, French Polynesia’. Pacific Science 50 (3): 253–73. Lepofsky, Dana, and Ken Lertzman. 2005. ‘More on Sampling for Richness and Diversity in

320

Archaeobiological Assemblages’. Journal of Ethnobiology 25 (2): 175–88. Lepofsky, Dana, Natasha Lyons, and Madonna L. Moss. 2003. ‘The Use of Driftwood on the North Pacific Coast: An Example from Southeast Alaska’. Journal of Ethnobiology 23 (1): 125–41. Lev-Yadun, Simcha. 2007. ‘Wood Remains from Archaeological Excavations: A Review with a Near Eastern Perspective’. Israel Journal of Earth Sciences 56: 139–62. Levin, Simon A. 1992. ‘The Problem of Pattern and Scale in Ecology: The Robert H. MacArthur Award Lecture’. Ecology 73 (6): 1943–67. Lexerød, Nils L., and Tron Eid. 2006. ‘An Evaluation of Different Diameter Diversity Indices Based on Criteria Related to Forest Management Planning’. Forest Ecology and Management 222 (1–3): 17–28. Li, Jinbao, Shang-Ping Xie, Edward R. Cook, Mariano S. Morales, Duncan A. Christie, Nathaniel C. Johnson, Fahu Chen et al. 2013. ‘El Niño modulations over the past seven centuries.’ Nature Climate Change 3 (9): 822–826. Lingens, A., E. Windeisen, and G. Wegener. 2005. ‘Investigating the Combustion Behaviour of Various Wood Species via Their Fire Gases’. Wood Science and Technology 39 (1): 49–60. Linsley, Braddock K., Peipei Zhang, Alexey Kaplan, Stephen S. Howe, and Gerard M. Wellington. 2008. ‘Interdecadal-Decadal Climate Variability from Multicoral Oxygen Isotope Records in the South Pacific Convergence Zone Region since 1650 A.D.’ Paleoceanography 23: PA2219. Linton, Ralph. 1923. The Material Culture of the Marquesas Islands. Honolulu, Hawaii: Bishop Museum Press. ———. 1925. Archaeology of the Marquesas Islands. Honolulu, Hawaii: The Bernice P. Bishop Museum. Lisiansky, Urey. 1814. A Voyage Round the World, in the Years 1803, 4, 5, & 6. Translation by the author. London: John Booth & Longman, Hurst, Rees, Orme, & Brown. Liu, Jianquan, and Shuichi Noshiro. 2003. ‘Lack of Latitudinal Trends in Wood Anatomy of Dodonaea viscosa (Sapindaceae), a Species with a Worldwide Distribution’. American Journal of Botany 90 (4): 532–39. Ludemann, T. 2008. ‘Experimental Charcoal-Burning with Special Regard to Anthracological Wood Diameter Analysis’. In Charcoals from the Past: Cultural and Palaeoenvironmental Implications edited by G. Fiorentino and D. Magri, 147–58. Oxford: Archaeopress. MacArthur, Robert H. 1967. The Theory of Island Biogeography. Princeton University Press. Madsen, M., C. Lipo, and M. Cannon. 1999. ‘Fitness and Reproductive Trade-Offs in Uncertain Environments: Explaining the Evolution of Cultural Elaboration’. Journal of Anthropological Archaeology 18 (3): 251–81. Magurran, Anne E. 2004. Measuring Biological Diversity. Malden, Ma.: Blackwell Pub. Malinowski, Bronislaw. 1935. Coral Gardens and Their Magic: A Study of the Methods of Tilling the Soil and of Agricultural Rites in the Trobriand Islands. Vol. 1. London: G. Allen & Unwin Ltd. Marchand, Étienne. 1810. ‘A Voyage Around the World’. In A Collection of Voyages and Travels from the Discovery of America to the Commencement of the Nineteenth Century, edited by W. F. Mavor, Vol. XIII:85–174. London: Richard Phillips. Marguerie, Dominique, and Jean-Yves Hunot. 2007. ‘Charcoal Analysis and Dendrology: Data from Archaeological Sites in North-Western France’. Journal of Archaeological Science 34 (9): 1417–33. Marston, John M. 2009. ‘Modeling Wood Acquisition Strategies from Archaeological Charcoal Remains’. Journal of Archaeological Science 36 (10): 2192–2200. ———. 2010. ‘Evaluating Risk, Sustainability, and Decision Making in Agricultural and Land-Use Strategies at Ancient Gordion’. Ph.D. diss., University of California, Los

321

Angeles. ———. 2011. ‘Archaeological Markers of Agricultural Risk Management’. Journal of Anthropological Archaeology 30 (2): 190–205. ———. 2012. ‘Reconstructing the Functional Use of Wood at Phrygian Gordion through Charcoal Analysis’. In The Archaeology of Phrygian Gordion, 47–54. Philadelphia: University of Pennsylvania Museum Press. Massal, Emile, and Jacques Barrau. 1956. Food Plants of the South Sea Islands. Noumea, New Caledonia: South Pacific Commission. Matthews, Peter J., and Chris Gosden. 1997. ‘Plant Remains from Waterlogged Sites in the Arawe Islands, West New Britain Province, Papua New Guinea: Implications for the History of Plant Use and Domestication’. Economic Botany 51 (2): 121–33. McAlister, Andrew. 2011. ‘Methodological Issues in the Geochemical Characterisation and Morphological Analysis of Stone Tools: A Case Study from Nuku Hiva, Marquesas Islands, East Polynesia’. Ph.D. thesis. University of Auckland. McConkey, Kim R., and Donald R. Drake. 2002. ‘25 Extinct Pigeons and Declining Bat Populations: Are Large Seeds Still Being Dispersed in the Tropical Pacific?’ In Seed Dispersal and Frugivory: Ecology, Evolution, and Conservation, edited by D. J. Levey, W. R. Silva, and M. Galetti, 381–96. New York: CABI Pub. McCoy, Mark D., Michael W. Graves, and Gail Murakami. 2010. ‘Introduction of Breadfruit (Artocarpus altilis) to the Hawaiian Islands’. Economic Botany 64 (4): 374–81. McCracken, Stephen D., Eduardo S. Brondfzio, Donald Nelson, Emlllo F. Moran, Andrea D. Siqueira, and Carlos Rodriguez-Pedraza. 1999. ‘Remote Sensing and GIS at Farm Property Level: Demography and Deforestation in the Brazilian Amazon’. Photogrammetric Engineering & Remote Sensing 65 (11): 1311–20. McFadgen, B. G. 1982. “Dating New Zealand Archaeology by Radiocarbon.” New Zealand Journal of Science 25: 379–92. ———. 2008. Hostile Shores: Catastrophic Events in Prehistoric New Zealand and Their Impact on Maori Coastal Communities. Auckland University Press. McGinnes Jr., E.A., S. Kandeel, and P. Szopa. 1971. ‘Some Structural Changes Observed in the Transformation of Wood into Charcoal’. Wood and Fiber Science 3: 77–83. McGlade, James. 1995. ‘Archaeology and the Ecodynamics of Human-Modified Landscapes’. Antiquity 69: 113–32. McGlade, James, and Sander E. van der Leeuw. 1997. ‘Introduction: Archaeology and Non- Linear Dynamics: New Approaches to Long-Term Change’. Time, Process, and Structured Transformation in Archaeology: 1–31. London; New York: Routledge. McParland, Laura C., Margaret E. Collinson, Andrew C. Scott, Gill Campbell, and Robyn Veal. 2010. ‘Is Vitrification in Charcoal a Result of High Temperature Burning of Wood?’ Journal of Archaeological Science 37: 2679–87. McWethy, David B., Cathy Whitlock, Janet M. Wilmshurst, Matt S. McGlone, Mairie Fromont, Xun Li, Ann Dieffenbacher-Krall, William O. Hobbs, Sherilyn C. Fritz, and Edward R. Cook. 2010. ‘Rapid Landscape Transformation in South Island, New Zealand, Following Initial Polynesian Settlement’. Proceedings of the National Academy of Sciences 107 (50): 21343 –48. Meehan, Hayley J., Kim R. McConkey, and Donald R. Drake. 2002. ‘Potential Disruptions to Seed Dispersal Mutualisms in Tonga, Western Polynesia’. Journal of Biogeography 29 (5/6): 695–712. Meeker, Virgil, and Gail M. Murakami. 1995. From Firepits to Charcoal Kilns: Resource Procurement and Land Use In the Upland Kane’ohe Catchment. Honolulu: International Archaeological Research Institute, Inc. (IARII). Metcalfe, C. R., and L. Chalk. 1950. Anatomy of the Dicotyledons: Wood Structure and Conclusion of the General Introduction. Vol. 2. Oxford: Clarendon Press. Meyer, Jean-Yves. 2004. ‘Threat of Invasive Alien Plants to Native Flora and Forest

322

Vegetation of Eastern Polynesia’. Pacific Science 58 (3): 357–75. ———. 2007. ‘Conservation des Forêts Naturelles et Gestion Des Aires Protégées En Polynésie Française (Conserving Natural Forests and Managing Protected Areas in French Polynesia)’. Bois et Forêts Des Tropiques (291): 25–40. Meyer, Jean-Yves, and Jean-François Butaud. 2009. ‘The Impacts of Rats on the Endangered Native Flora of French Polynesia (Pacific Islands): Drivers of Plant Extinction or Coup de Grâce Species?’ Biological Invasions 11: 1569–85. Meyer, W. J, and C. L Crumley. 2012. ‘Historical Ecology: Using What Works to Cross the Divide’. In Atlantic Europe in the First Milennium BC: Crossing the Divide, 109–34. Oxford: Oxford University Press. Meylan, B. A., and B. G. Butterfield. 1978. The Structure of New Zealand Woods. Wellington: Department of Scientific and Industrial Research. Mieth, Andreas, and Hans-Rudolf Bork. 2010. ‘Humans, Climate or Introduced Rats – Which Is to Blame for the Woodland Destruction on Prehistoric Rapa Nui (Easter Island)?’ Journal of Archaeological Science 37 (2): 417–26. Miksicek, Charles. 1987. ‘Formation Processes of the Archaeobotanical Record’. Advances in Archaeological Method and Theory 10:211–47. Miller, Naomi. 1985. ‘Paleoethnobotanical Evidence for Deforestation in Ancient Iran’. Journal of Ethnobiology 5: 1–19. ———. 1988. ‘Ratios in Palaeoethnobotanical Analysis’. In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 72–85. Chicago: University of Chicago Press. Millerstrom, Sidsel N. 2001. ‘Images Carved in Stones and Settlement Patterns Archaeology in Hatiheu Valley, Nuku Hiva, the Marquesas Islands, French Polynesia’. Ph.D. diss., University of California at Berkeley. Millerstrom, Sidsel N., and James H. Coil. 2008. ‘Pre-Contact Arboriculture and Vegetation in the Marquesas Islands, French Polynesia: Charcoal Identification and Radiocarbon Dates from Hatiheu Valley, Nuku Hiva’. Asian Perspectives 47 (2): 330–51. Minnis, P. E., and R.I. Ford. 1977. ‘Analysis of Plant Remains from Chimney Rock Mesa’. In Archaeological Investigations at Chimney Rock Mesa: 1970–72, edited by F. W. Eddy: 81–91. Boulder: Colorado Archaeological Society. Molle, Guillaume, and Eric Conte. 2011. ‘New Perspectives on the Occupation of Hatuana Dune Site, Ua Huka, Marquesas Islands’. Journal of Pacific Archaeology 2 (2). Moran, Emilio F. 2009. Human Adaptability: An Introduction to Ecological Anthropology. Boulder, Colorado: Westview Press. Morrison, Kathleen. 1994. ‘The Intensification of Production: Archaeological Approaches’. Journal of Archaeological Method and Theory 1 (2): 111–59. ———. 1996. ‘Typological Schemes and Agricultural Change: Beyond Boserup in Precolonial South India’. Current Anthropology 37 (4): 583–608. ———. 2007. ‘Rethinking Intensification: Power Relations and Scales of Analysis in Precolonial South India’. In Seeking a Richer Harvest: The Archaeology of Subsistence Intensification, Innovation, and Change, edited by T. L. Thurston and C. T. Fisher, 235–48. New York: Springer Science+Business Media. Moy, Christopher M., Geoffrey O. Seltzer, Donald T. Rodbell, and David M. Anderson. 2002. ‘Variability of El Niño / Southern Oscillation Activity at Millennial Timescales during the Holocene Epoch’. Nature 420 (6912): 162–65. Mueller-Dombois, Dieter, and F. Raymond Fosberg. 1998. Vegetation of the Tropical Pacific Islands. New York: Springer. Mulrooney, Mara A., Simon H. Bickler, Melinda S. Allen, and Thegn N. Ladefoged. 2011. ‘High-Precision Dating of Colonization and Settlement in East Polynesia’. Proceedings of the National Academy of Sciences 108 (23): E192–E194.

323

Murakami, Gail M. 1983. ‘Analysis of Charcoal from Archaeological Contexts’. In Archaeological Investigations of the Mudlane-Waimea-Kawaihae Road Corridor, Island of Hawaiʻi: An Interdisciplinary Study of an Environmental Transect, edited by J. T. Clark and P. V. Kirch. Department of Anthropology Report 83–1. Honolulu: Bernice P. Bishop Museum. ———. 1989. ‘Identification of Charcoal from the Kuolulo Rockshelter’. In Prehistoric Hawaiian Occupation in the Anahulu Valley, O’ahu Island: Excavations in Three Inland Rockshelters, edited by P. V. Kirch, 103–10. Archaeological Research Facility Report No. 47: Dept. of Anthropology, University of California at Berkeley. Nelle, Oliver, Stefan Dreibrodt, and Yasmin Dannath. 2010. ‘Combining Pollen and Charcoal: Evaluating Holocene Vegetation Composition and Dynamics’. Journal of Archaeological Science 37: 2126–35. Nicolson, Dan Henry, and Francis Raymond Fosberg. 2004. The Forsters and the Botany of the Second Cook Expedition (1772–1775). Ruggell, Liechtenstein: A.R.G. Gantner Verlag. Njankouo, Jacques Michel, Jean-Claude Dotreppe, and Jean-Marc Franssen. 2004. ‘Experimental Study of the Charring Rate of Tropical Hardwoods’. Fire and Materials 28: 15–24. O’Carroll, Ellen, and Fraser Mitchell. 2012. ‘Charcoal Sample Guidelines: New Methodological Approaches towards the Quantification and Identification of Charcoal Samples Retrieved from Archaeological Sites’. Saguntum: Papeles Del Laboratorio de Arqueología de Valencia (13): 275–82. O’Shea, John. 1989. ‘The Role of Wild Resources in Small-Scale Agricultural Systems: Tales from the Lakes and the Plains’. In Bad Year Economics: Cultural Responses to Risk and Uncertainty, edited by P. Halstead and J. O’Shea, 57–67. Cambridge; New York: Cambridge University Press. Orliac, Catherine. 1987. ‘Étude fonctionnelle de fours experimentaux à Tahiti’. UA 275 du CNRS. Paris: Laboratoire d’Ethnologie Préhistorique. ———. 1991. ‘Étude du Fonctionnement de Fours Expérimentaux à Tahiti’. UA 275 du CNRS. Paris: Laboratoire d’Ethnologie Préhistorique. ———. 2000. ‘The Woody Vegetation of Easter Island between the Early 14th and the Mid- 17th Centuries A.D.’ In Easter Island Archaeology: Research on Early Rapanui Culture, edited by C. Stevenson and W. Ayres, 211–20. Los Osos: Easter Island Foundation. ———. 2003. ‘Étude Expérimentale du Fonctionnement de Fours Polynésiens à Tahiti’. In Actes Du Colloque de Bourg-En-Brese: Le Feu Domestique et Ses Structures Au Néolithique et Aux Ages Des Métaux, edited by M.C. Frére-Sautot, 209–14. Montagnac: éditions Monique Mergoil, collection préhistoires No.9. Orliac, Catherine, and Michel Orliac. 1980. ‘Les Structures de Combustion et Leur Interprétation Archéologique: Quelques Exemples En Polynésie’. Journal de La Société Des Océanistes 36: 61–76. ———. 2008. ‘Extinct Flora of Easter Island’. In At the Heart of Ancient Societies: French Contributions to Pacific Archaeology, edited by A. DiPiazza, E. Pearthree, and C. Sand, 197–208. Nouméa, New Caledonia: Départment Archéologie (DACC). Orliac, Catherine, and Julia Wattez. 1989. ‘Un Four Polynesien et Son Interpretation Archeologique’. In Nature et Fonction Des Foyers Prehistoriques: Actes Du Colloque International de Nemours, 12-13-14 Mai 1987, 69–75. France: Nemours. Orliac, Michel. 1997. ‘Human Occupation and Environmental Modifications in the Papeno’o Valley, Tahiti’. In Historical Ecology in the Pacific Islands: Prehistoric Environmental and Landscape Change, edited by P. V. Kirch and T. L. Hunt, 201– 309. New Haven: Yale University Press. ———. 2003a. “Annexe: Erosion, Sedimentation et Amenagements Hydrauliques Des Sites

324

Tahakia, Kamuihei et Teiipoka.” In Bilan de La Recherche Archéologique En Polynésie Française, 2001–2002, edited by H. Marchesi, 97–100. Punaauia, Tahiti: Ministère de la culture. ———. 2003b. ‘Un aspect de la flore de Mangareva au XII ème siècle (archipel Gambier, Polynésie française)’. In Archéologie en Océanie insulaire: peuplement, sociétés et paysages, edited by C. Orliac, 150–71. Paris: Artcom. ORSTOM. 1993. Atlas de La Polynésie Française. Paris: Éditions de l’Orstom. Orton, Clive. 2000. Sampling in Archaeology. Cambridge: Cambridge University Press. Orvis, Kenneth H., Chad S. Lane, and Sally P. Horn. 2005. ‘Laboratory Production of Vouchered Reference Charcoal from Small Wood Samples and Non-Woody Plant Tissues’. Palynology 29: 1–11. Ottino-Garanger, Pierre. 2006. Archéologie chez les Taïpi: Hatiheu, un projet partagé aux îles Marquises. Tahiti: Au vent des îles / IRD Éditions. Ottino-Garanger, Pierre, Frédérique Valentin, and Hélène Guiot. 2003. ‘Recherches à Hatiheu, Ile de Nuku Hiva, Archipel Des Iles Marquises. Archéologie, Identité et Développement’. In Bilan de La Recherche Archéologique En Polynésie Française, 2001–2002, edited by Henri Marchesi, 79–100. Punaauia, Tahiti: Ministère de la culture. Panshin, A. J., and Carl De Zeeuw. 1980. Textbook of Wood Technology: Structure, Identification, Properties, and Uses of the Commercial Woods of the United States and Canada. New York: McGraw-Hill. Parkes, Annette. 1994. ‘Holocene Environments and Vegetational Change on Four Polynesian Islands’. Ph.D. diss., Hull, U.K.: Department of Geography, The University of Hull. ———. 1997. ‘Environmental Change and the Impact of Polynesian Colonization: Sedimentary Records from Central Polynesia’. In Historical Ecology in the Pacific Islands, edited by P. V. Kirch and T. L. Hunt, 166–99. New Haven: Yale University Press. Parkes, Annette, and John R. Flenley. 1990. The Hull University Moorea Expedition 1985: Final Report. Hull: Department of Geography, University of Hull. Parkes, Annette, James T. Teller, and John R. Flenley. 1992. ‘Environmental History of the Lake Vaihiria Drainage Basin, Tahiti, French Polynesia’. Journal of Biogeography 19: 431–47. Parkinson, Sydney. 1773. A Journal of a Voyage to the South Seas, in His Majesty’s Ship, the Endeavour. London: printed for Stanfield Parkinson, the editor, and sold by Messrs. Richardson and Urquhart, Evans, Hooper, Murray, Leacroft and Riley. Pearsall, Deborah M. 2000. Paleoethnobotany: A Handbook of Procedures. San Diego: Academic Press. Pennington, Terence Dale. 1991. The Genera of Sapotaceae. Royal Botanic Gardens: Bronx, New York.: Kew & New York Botanical Garden. Petchey, Fiona, Melinda S. Allen, David J. Addison, and Atholl Anderson. 2009. ‘Stability in the South Pacific Surface Marine 14C Reservoir over the Last 750 Years. Evidence from American Samoa, the Southern Cook Islands and the Marquesas’. Journal of Archaeological Science 36: 2234–43. Poissonet, J. 1968. ‘Recherche Sur Les Lois Générales D’équilibre Dans La Composition Floristique Des Formations Herbacées Denses. Premiers Résultats et Hypothèses’. Montpellier, CEPE. ———. 1979. ‘Équilibre et Déséquilibre Des Phytocénoses Herbacées’. In: Dixièmes Journées Du Grenier de Theix, 461–68. INRA. Popper, Virginia S. 1988. ‘Quantitative Measurements in Paleoethnobotany’. In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 53–71. Chicago: University of Chicago Press.

325

Porter, David. 1822. Journal of a Cruise Made to the Pacific Ocean, in the United States Frigate Essex, in the Years 1812, 1813, and 1814. New York: Wiley & Halsted. Prebble, Matiu, and Janet Wilmshurst. 2009. ‘Detecting the Initial Impact of Humans and Introduced Species on Island Environments in Remote Oceania Using Palaeoecology’. Biological Invasions 11: 1529–56. Prebble, Matthew. 2008. ‘No Fruit on That Beautiful Shore: What Plants Were Introduced to the Subtropical Polynesian Islands prior to European Contact?’ In Islands of Inquiry: Colonisation, Seafaring and the Archaeology of Maritime Landscapes, edited by G. Clark, F. Leach, and S. O’Connor, 227–52. Canberra: Australian National University. Prebble, Matthew, and Atholl Anderson. 2012. ‘Palaeobotany and the Early Development of Agriculture on Rapa Island’. In Taking the High Ground: The Archaeology of Rapa, a Fortified Island in Remote East Polynesia, edited by A. Anderson and D. J. Kennett, 167–87. Terra Australis 37. Canberra: Australian National University. Prebble, Matthew, Atholl Anderson, and D. J. Kennett. 2012. ‘Forest Clearance and Agricultural Expansion on Rapa, Austral Archipelago, French Polynesia’. The Holocene 23 (2): 179–96. Prior, Juliet, and K.L. Alvin. 1983. ‘Structural Changes on Charring Woods of Dichrostachys and Salix from Southern Africa’. IAWA Bulletin 4: 197–206. Prior, Juliet, and Peter Gasson. 1993. ‘Anatomical Changes on Charring Six African Hardwoods’. IAWA Journal 14: 77–86. Quirós, Pedro Fernandez de. 1904. The Voyages of Pedro Fernandez de Quiros, 1595 to 1606. Translated by Sir Clements Markham. Vol. 1. London: Hakluyt Society. Ragone, Carol Diane. 1991. ‘Collection, Establishment, and Evaluation of a Germplasm Collection of Pacific Island Breadfruit’. Ph.D. diss., University of Hawaiʻi. Rallu, Jean Louis. 1992. ‘From Decline to Recovery: The Marquesan Population 1886–1945’. Health Transition Review 2 (2): 177–94. Redman, Charles. 2005. ‘Resilience Theory in Archaeology’. American Anthropologist 107: 70–77. Redman, C. L. and A. P. Kinzig. 2003. Resilience of past landscapes: resilience theory, society, and the longue durée. Conservation Ecology 7(1): 14. [online] URL: http://www.consecol.org/vol7/iss1/art14/ Reimer, Paula J., Edouard Bard, Alex Bayliss, J. Warren Beck, Paul G. Blackwell, Christopher Bronk Ramsey, Caitlin E. Buck, et al. 2013. “IntCal13 and Marine13 radiocarbon age calibration curves 0–50,000 years cal BP.” Radiocarbon 55 (4): 1869–1887. Reitz, Elizabeth J., and Myra Shackley. 2012. Environmental Archaeology. New York: Springer. Rendle, B. J. 1960. ‘Juvenile and Adult Wood’. J. Inst. Wood Sci 5: 58–61. Renfrew, Colin. 1982. ‘Polity and Power: Interaction, Intensification and Exploitation’. In An Island Polity: The Archaeology of Exploitation in Melos, edited by C. Renfrew and M. Wagstaff, 264–90. Cambridge: Cambridge University Press. Rensch, Karl Heinz M. 2005. Plant Names of Eastern Polynesia. Mawson, A.C.T.: Archipelago Press. Rensch, Karl Heinz M., and W. Arthur Whistler. 2009. Dictionary of Polynesian Plant Names : Anuta, Austral Islands, Bellona, Cook Islands, Futuna (Wallis & Futuna), Futuna (Vanuatu), Gambier Islands, Hawaiʻi, Kapingamarangi, Leeward Islands, Maori, Marquesas, Mele, Nukuoro, Ouvéa, Pitcairn, Rapa Nui, Rennell, Samoa, Tahiti, Tikopia, Tokelau, Tonga, Tuamotus, Tuvalu, Uvea. Mawson, A.C.T.: Archipelago Press. Richter, H. G., and M. J. Dallwitz. 2000. ‘Commercial Timbers: Descriptions, Illustrations, Identification, and Information Retrieval.’ Accessed 25 June 2009. http://delta- intkey.com.

326

Rick, Torben C., Patrick V. Kirch, Jon M. Erlandson, and Scott M. Fitzpatrick. 2013. ‘Archeology, Deep History, and the Human Transformation of Island Ecosystems’. Anthropocene 4: 33–45. Rieth, Timothy, and J. Steven Athens. 2013. ‘Suggested Best Practices for the Application of Radiocarbon Dating to Hawaiian Archaeology’. Hawaiian Archaeology 13: 3–29. Robarts, Edward. 1974. The Marquesan Journal of Edward Robarts, 1797–1824. Edited by Greg Dening. Canberra: Australian National University Press. Rock, Joseph F.C. 1913. The Indigenous Trees of the Hawaiian Islands. Honolulu: J.F. Rock. Rolett, Barry V. 1998. Hanamiai: Prehistoric Colonization and Cultural Change in the Marquesas Islands, East Polynesia. New Haven: Dept. of Anthropology and The Peabody Museum, Yale University. ———. 2008. ‘Avoiding Collapse: Pre-European Sustainability on Pacific Islands’. Quaternary International 184: 4–10. Rolett, Barry V., and E. Conte. 1995. ‘Renewed Investigation of the Ha’atuatua Dune (Nukuhiva, Marquesas Islands): A Key Site in Polynesian Prehistory’. Journal of the Polynesian Society 104 (2): 195–228. Rolett, Barry V., and Jared Diamond. 2004. ‘Environmental Predictors of Pre-European Deforestation on Pacific Islands’. Nature 431: 443–46. Rollin, L. 1929. Les Iles Marquises. Paris: Société d’Éditions Géographiques, Maritimes et Coloniales. Rosendahl, Paul H. 1994. ‘Aboriginal Hawaiian Structural Remains and Settlement Patterns in the Upland Agricultural Zone at Lapakahi, Island of Hawaii’. Hawaiian Archaeology 3: 14–70. Rosendahl, Paul, and Douglas E. Yen. 1971. ‘Fossil Sweet Potato Remains from Hawaii’. Journal of the Polynesian Society 80 (3): 379–85. Rossen, Jack, and James Olson. 1985. ‘The Controlled Carbonization and Archaeological Analysis of SE U. S. Wood Charcoals’. Journal of Field Archaeology 12: 445–56. Salinger, M. J., R. E. Basher, B. B. Fitzharris, J. E. Hay, P. D. Jones, J. P. MacVeigh, and I. Schmidely-Leleu. 1995. ‘Climate Trends in the South-West Pacific’. International Journal of Climatology 15 (3): 285–302. Salisbury, K. J., and F. W. Jane. 1940. ‘Charcoals from Maiden Castle and Their Significance in Relation to the Vegetation and Climatic Conditions in Prehistoric Times’. Journal of Ecology 28: 310–25. Sastrapradja, D. S., and C. Lamoureux. 1969. ‘Variations in Wood Anatomy of Hawaiian Metrosideros (Myrtaceae)’. Annales Bogorienses V: 1–83. Sauer, Carl Ortwin. 1925. The Morphology of Landscape. University Press. ———. 1952. Agricultural Origins and Dispersals. New York: American Geographical Society. Scheel-Ybert, Rita. 1998. ‘Stabilité De L’Écosystème Sur Le Littoral Sud-Est Du Brésil À L’Holocène Supérieur (5500 – 1400 Ans BP)’. Ph.D. diss., Montpellier, France: Universite Montpellier II. ———. 2001. ‘Man and Vegetation in Southeastern Brazil during the Late Holocene’. Journal of Archaeological Science 28: 471–80. ———. 2002. ‘Evaluation of Sample Reliability in Extant and Fossil Assemblages’. In Charcoal Analysis: Methodological Approaches, Palaeoecological Results and Wood Uses: Proceedings of the Second International Meeting of Anthracology, Paris, September 2000, edited S. Thiébault, 1–7. Oxford: Archaeopress. Schiffer, Michael B. 1987. Formation Processes of the Archaeological Record. Albuquerque, NM : University of New Mexico Press. Schoch, W., I. Heller, F.H. Schweingruber, and F. Kienast. 2004–onwards. ‘Wood Anatomy of Central European Species’. Accessed 7 January 2009. http://www.wsl.ch/land/products/dendro/.

327

Schweingruber, Fritz H. 2007. Wood Structure and Environment. Berlin; New York: Springer. Scott, Andrew C. 2010. ‘Charcoal Recognition, Taphonomy and Uses in Palaeoenvironmental Analysis’. Palaeogeography, Palaeoclimatology, Palaeoecology 291 (1–2): 11–39. Scott, Andrew C., and Freddy Damblon. 2010. ‘Charcoal: Taphonomy and Significance in Geology, Botany and Archaeology’. Palaeogeography, Palaeoclimatology, Palaeoecology 291 (1–2): 1–10. Shackleton, C. M., and F. Prins. 1992. ‘Charcoal Analysis and the “Principle of Least Effort” – A Conceptual Model’. Journal of Archaeological Science 19: 631–37. Sharp, Warren D., Jennifer G. Kahn, Christina M. Polito, and Patrick V. Kirch. 2010. ‘Rapid Evolution of Ritual Architecture in Central Polynesia Indicated by Precise 230Th/U Coral Dating’. Proceedings of the National Academy of Sciences 107 (30): 13234–39. Sidiyasa, Kade. 1998. ‘Taxonomy, Phylogeny, and Wood Anatomy of Alstonia (Apocynaceae)’. Blumea, Suppl 11: 1–230. Simek, Jan F. 1989. ‘Structure and Diversity in Intrasite Spatial Analysis’. In Quantifying Diversity in Archaeology, edited by R. D. Leonard and G. T. Jones, 59–68. Cambridge: Cambridge University Press. Sinoto, Yosihiko. 1966. ‘A Tentative Prehistoric Cultural Sequence in the Northern Marquesas Islands, French Polynesia’. Journal of the Polynesian Society 75: 287–303. ———. 1970. ‘An Archaeologically Based Assessment of the Marquesas Islands as a Dispersal Center in East Polynesia’. In Studies in Oceanic Culture History, edited by R. C. Green and M. Kelly, 05–32. Honolulu, Dept. of Anthropology, Bernice Pauahi Bishop Museum. ———. 1979. ‘The Marquesas’. In The Prehistory of Polynesia, edited by J.D. Jennings, 110–34. Cambridge: Cambridge University Press. Smart, Tristine Lee, and Ellen S. Hoffman. 1988. ‘Environmental Interpretation of Archaeological Charcoal’. In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 167–205. Chicago: University of Chicago Press. Smithsonian National Museum of Natural History. n.d. ‘Flora of the Marquesas Islands’. http://botany.si.edu/pacificislandbiodiversity/marquesasflora/. Data maintained by Warren L. Wagner and David H. Lorence. ———. n.d. ‘Flora of the Hawaiian Islands’. http://botany.si.edu/pacificislandbiodiversity/hawaiianflora/. Data maintained by Warren L. Wagner. Soerianegara, I, and R.H.M.J. Lemmens, ed. 1993. Timber Trees: Major Commercial Timbers. Vol. 1. Plant Resources of South-East Asia 5. Wageningen: Pudoc Scientific Publishers. Stanner, W. E. H. 1965. ‘Aboriginal Territorial Organization: Estate, Range, Domain and Regime’. Oceania 36 (1): 1–26. Steadman, David W. 2006. Extinction and Biogeography of Tropical Pacific Birds. University of Chicago Press. Steadman, David W., and H. B. Freifeld. 1999. ‘The Food Habits of Polynesian Pigeons and Doves: A Systematic and Biogeographic Review’. Ecotropica 5: 13–33. Steadman, David W., and Barry Rolett. 1996. ‘A Chronostratigraphic Analysis of Landbird Extinction on Tahuata, Marquesas Islands’. Journal of Archaeological Science 23 (1): 81–94. Stein, Julie K. 1985. ‘Interpreting Sediments in Cultural Settings’. In Archaeological Sediments in Context, edited by J. K. Stein and W. R. Farrand, 5–19. Maine: Centre for the Study of Early Man. Stern, W.L., and S. Greene. 1958. ‘Some Aspects of Variation in Wood’. Tropical Woods 108: 65–71.

328

Stevenson, Robert Louis. 1987. In the South Seas. London: Hogarth Press. Steward, Julian H. 1972. Theory of Culture Change: The Methodology of Multilinear Evolution. Urbana: University of Illinois Press. Struever, Stuart. 1968. ‘Flotation Techniques for the Recovery of Small-Scale Archaeological Remains’. American Antiquity: 353–62. Stuijts, Ingelise. 2006. ‘Charcoal Sampling Sites and Procedures: Practical Themes from Ireland’. In Charcoal Analysis: New Analytical Tools and Methods for Archaeology: Papers from the Table-Ronde Held in Basel 2004, 25–33. Oxford: Archaeopress. Suggs, Robert C. 1961. The Archaeology of Nuku Hiva, Marquesas Islands, French Polynesia. New York: American Museum of Natural History. Tautain, M. 1897. ‘Notes Sur Les Constructions et Monuments Des Marquises’. Anthropologie 8: 538–58. Taylor, G. Marie. 1961. ‘A Key to the Coprosmas of New Zealand – Part I’. Tuatara 9: 31– 64. Terpkosh, Travis. 2010. ‘Tropical Dry Forests of the Marquesas Islands, French Polynesia’. M. A. Thesis draft, Department of Geography, UCLA. Terrell, John E., John P. Hart, S. Barut, N. Cellinese, A. Curet, T. Denham, C. M. Kusimba, et al. 2003. ‘Domesticated Landscapes: The Subsistence Ecology of Plant and Animal Domestication’. Journal of Archaeological Method and Theory 10 (4): 323–68. Thaman, Randolph R., Craig R. Elevitch, and Kim M. Wilkinson. 2000. Multipurpose Trees for Agroforestry in the Pacific Islands. Agroforestry Guides for Pacific Islands 2. Ho’olualoa, Hawaiʻi: Permanent Agriculture Resources. Théry-Parisot, Isabel, Lucie Chabal, and Sandrine Costamagno, eds. 2010. The Taphonomy of Burned Organic Residues and Combustion Features in Archaeological Contexts Proceedings of the Round Table, May 27–29 2008, CEPAM. Palethnologie 2. Toulouse: Revue Palethnologie. Théry-Parisot, Isabelle, Lucie Chabal, and Julia Chrzavzez. 2010. ‘Anthracology and Taphonomy, from Wood Gathering to Charcoal Analysis. A Review of the Taphonomic Processes Modifying Charcoal Assemblages, in Archaeological Contexts’. Palaeogeography, Palaeoclimatology, Palaeoecology 291 (1–2): 142–53. Théry-Parisot, Isabelle, Lucie Chabal, Maria Ntinou, Laurent Bouby, and Alain Carre. 2010. ‘From Wood to Wood Charcoal: An Experimental Approach to Combustion’. In The Taphonomy of Burned Organic Residues and Combustion Features in Archaeological Contexts Proceedings of the Round Table, May 27–29 2008, CEPAM, 79–91. Palethnologie 2. Toulouse: Revue Palethnologie. Théry-Parisot, Isabelle, and Auréade Henry. 2012. ‘Seasoned or Green? Radial Cracks Analysis as a Method for Identifying the Use of Green Wood as Fuel in Archaeological Charcoal’. Journal of Archaeological Science 39 (2): 381–88. Thomas, Ken. 2001. ‘Environmental Archaeology Is Dead: Long Live Bioarchaeology, Geoarchaeology and Human Palaeoecology’. In Environmental Archaeology: Meaning and Purpose, edited by U. Albarella, 55–58. Dordrecht; Boston: Kluwer Academic Publishers; Norwell, MA. Thomas, Nicholas. 1990. Marquesan Societies: Inequality and Political Transformation in Eastern Polynesia. Oxford: Clarendon Press. Thompson, G. B. 1994. ‘Wood Charcoals from Tropical Sites: A Contribution to Methodology and Interpretation’. In Tropical Archaeobotany: Applications and New Developments, edited by J. G. Hather, 9–33. London: Routledge. Thoms, Alston V. 2008. ‘The Fire Stones Carry: Ethnographic Records and Archaeological Expectations for Hot-Rock Cookery in Western North America’. Journal of Anthropological Archaeology 27: 443–60. ———. 2009. ‘Rocks of Ages: Propagation of Hot-Rock Cookery in Western North America’. Journal of Archaeological Science 36: 573–91.

329

Thomson, Robert. 1980. The Marquesas Islands: Their Description and Early History. Edited by R. D. Craig. 2nd ed. Laie, Hawaii: Institute for Polynesian Studies, Brigham Young University-Hawaii Campus. Toll, Mollie S. 1988. ‘Flotation Sampling: Problems and Some Solutions, with Examples from the American Southwest’. In Current Paleoethnobotany: Analytical Methods and Cultural Interpretations of Archaeological Plant Remains, edited by C. A. Hastorf and V. S. Popper, 36–52. Chicago: University of Chicago Press. Touflan, Philippe, Brigitte Talon, and Kevin Walsh. 2010. ‘Soil Charcoal Analysis: A Reliable Tool for Spatially Precise Studies of Past Forest Dynamics: A Case Study in the French Southern Alps’. The Holocene 20 (1): 45–52. Twiddle, C.L., and M.J. Bunting. 2010. ‘Experimental Investigations into the Preservation of Pollen Grains: A Pilot Study of Four Pollen Types’. Review of Palaeobotany and Palynology 162 (4): 621–30. U.S. CLIVAR Project Office. 2013. ‘U.S. CLIVAR ENSO Diversity Workshop Report’. Report 2013-1. Washington, D.C.: U.S. CLIVAR Project Office. U.S. National Research Council, Advisory Committee on Technology Innovation. 1980. Firewood Crops: Shrub and Tree Species for Energy Production. 2 Vol. Washington D.C.: US Agency for International Develoment. University of Hawaii. 2009. ‘Native Plants Hawaiʻi Project’. Accessed 15 January 2014. http://nativeplants.hawaii.edu/. Ussher, Ella. 2009. ‘Application of Starch Residue Analysis Within Archaeological Research in the Pacific: A Marquesan Case Study of Shell Tools’. M. A. Thesis, University of Auckland. Van der Graaff, N.A., and P. Baas. 1974. ‘Wood Anatomical Variation in Relation to Latitude and Altitude’. Blumea 22: 101–21. Van der Leeuw, Sander E. 2009. ‘What Is an “Environmental Crisis” to an Archaeologist?’ In The Archaeology of Environmental Change: Socionatural Legacies of Degradation and Resilience, edited by C. T. Fisher, J. B. Hill, and G. M. Feinman, 40–61. University of Arizona Press. Van der Leeuw, Sander, and Charles L. Redman. 2002. ‘Placing Archaeology at the Center of Socio-Natural Studies’. American Antiquity 67 (4): 597–605. Van der Veen, M., and N. Fieller. 1982. ‘Sampling Seeds’. Journal of Archaeological Science 9: 287–98. Van der Warker, A. M. 2005. ‘Field Cultivation and Tree Management in Tropical Agriculture: A View from Gulf Coastal Mexico’. World Archaeology 37 (2): 275–89. VanWey, Leah K., Elinor Ostrom, and Vicky Meretsky. 2005. ‘Theories Underlying the Study of Human-Environment Interactions’. In Seeing the Forest and the Trees: Human-Environment Interactions in Forest Ecosystems, edited by Emilio F. Moran and Elinor Ostrom, 23–56. Cambridge, MA: MIT Press. Veal, Robyn. 2013. ‘Fuelling Ancient Mediterranean Cities: A Framework for Charcoal Research’. In The Ancient Mediterranean Environment between Science and History, edited by W.V. Harris, 37–60. Leiden and Boston: Brill. Vernet, Jean-Louis. 2002. ‘Preface’. In Charcoal Analysis: Methodological Approaches, Palaeoecological Results and Wood Uses: Proceedings of the Second International Meeting of Anthracology, Paris, September 2000, edited S. Thiébault, v–vi. Oxford: Archaeopress. Vitousek, Peter M. 2004. Nutrient Cycling and Limitation: Hawaiʻi as a Model System. Princeton University Press. Vitousek, Peter M., Harold A. Mooney, Jane Lubchenco, and Jerry M. Melillo. 1997. ‘Human Domination of Earth’s Ecosystems’. Science 277: 494–99. Von den Steinen, Karl. 1925. Die Marquesaner Und Ihre Kunst : Studien Über Die Entwicklung Primitiver Südseeornamentik Nach Eigenen Reise-Ergebnissen Und Dem

330

Material Der Museen. 3 vols. Berlin: D. Reimer (E. Vohsen). ———. 1988. Von Den Steinen’s Marquesan Myths. Edited by Jennifer Terrell. Translated by Marta Langridge. Canberra: Target Oceania / The Journal of Pacific History. Wagner, Warren, and David H. Lorence. 1997. ‘Studies of Marquesan Vascular Plants: Introduction’. Allertonia 7 (4): 221–25. ———. 2011a. ‘Revision of Coprosma (Rubiaceae, Tribe Anthospermeae) in the Marquesas Islands.’ PhytoKeys (4): 109. ———. 2011b. ‘A Nomenclator of Pacific Oceanic Island Phyllanthus (Phyllanthaceae), Including Glochidion’. PhytoKeys 4: 67. Waldren, Stephen. 1998. ‘Nesoluma st.-Johnianum’. IUCN 2013 Red List of Threatened Species. Version 2013.1. http://www.iucnredlist.org. Wallace, Rod T. 1981. ‘Prehistoric Vegetation at Long Beach, Otago’. New Zealand Archaeological Association Newsletter 24: 154–60. ———. 1985. ‘Studies on the Conservation of Waterlogged Wood in New Zealand’. Ph.D. diss., University of Waikato. ———. 1989. ‘A Preliminary Study of Wood Types Used in Pre-European Maori Wooden Artefacts’. In Saying so Doesn’t Make It so: Papers in Honour of B. Foss Leach, edited by D. G. Sutton. New Zealand Archaeological Association Monograph 17. Dunedin, NZ: New Zealand Archaeological Association. ———. 2000. ‘A Kohika Wharepuni: House Construction Methods of the Late Pre-Contact Maori’. New Zealand Journal of Archaeology 21: 67–86. Wallace, Rod T., and Roger C. Green. 2012. ‘Reassessing the Radiocarbon Chronology of the Maioro Site (R13/1): Northern Waikato, New Zealand’. Journal of the Polynesian Society 121 (1): 75–85. Wallace, Rod T., and Geoff Irwin. 2004. ‘The Wooden Artefacts from Kohika’. In Kohika: The Archaeology of a Late Maori Lake Village in the Ngati Awa Rohe, Bay of Plenty, New Zealand, edited by G.J. Irwin, 83–121. Auckland: Auckland University Press. Webber, I. E. 1934. ‘The Wood of Hibiscus tiliaceus L’. Trop. Woods 37: 14–18. Weiner, Jacob, and Otto T. Solbrig. 1984. ‘The Meaning and Measurement of Size Hierarchies in Plant Populations’. Oecologia 61 (3): 334–36. Weisler, Marshall, and Gail M. Murakami. 1991. ‘The Use of Mountain-Apple (Syzygium malaccense) in a Prehistoric Hawaiian Domestic Structure’. Economic Botany 45: 282–85. Western, Cecilia A. 1969. ‘Wood and Charcoal in Archaeology’. In Science in Archaeology: A Survey of Progress and Research, edited by D. R. Brothwell and E. S. Higgs, 178– 87. London: Thames & Hudson. ———. 1971. ‘The Ecological Interpretation of Ancient Charcoals from Jericho’. Levant 3: 31–40. Wheeler, E.A. 2011. InsideWood – a web resource for hardwood anatomy. IAWA Journal 32 (2): 199–211. Wheeler, E.A., P. Baas, and P.E. Gasson. 1989. ‘IAWA List of Microscopic Features for Hardwood Identification’. IAWA Bulletin 10: 219–332. Whistler, W. Arthur. 2004. Rainforest Trees of Samoa. Honolulu: Isle Botanica. ———. 2009. Plants of the Canoe People. Lawai, Kauaʻi, Hawaiʻi: National Tropical Botanical Garden. Whittaker, Robert J., and José María Fernández-Palacios. 2007. Island Biogeography: Ecology, Evolution, and Conservation. 2nd edition. Oxford; New York, Oxford University Press. Willcox, G. H. 1974. ‘A History of Deforestation as Indicated by Charcoal Analysis of Four Sites in Eastern Anatolia’. Anatolian Studies 24: 117–33. Wilmshurst, Janet M., Terry L. Hunt, Carl P. Lipo, and Atholl J. Anderson. 2011. ‘High- Precision Radiocarbon Dating Shows Recent and Rapid Initial Human Colonization of

331

East Polynesia’. Proceedings of the National Academy of Sciences 108 (5): 1815– 1820. Wittfogel, Karl August. 1957. Oriental Despotism: A Comparative Study of Total Power. New Haven: Yale University Press. Worbes, Martin, and Esther Fichtler. 2010. ‘Wood Anatomy and Tree-Ring Structure and Their Importance for Tropical Dendrochronology’. In Amazonian Floodplain Forests, edited by W. J. Junk, M. T. F. Piedade, F. Wittmann, J. Schöngart, and P. Parolin, 210:329–46. Dordrecht: Springer Netherlands. Wright, Patti J. 2005. ‘Flotation Samples and Some Paleoethnobotanical Implications’. Journal of Archaeological Science 32: 19–26. ———. 2010. ‘Methodological Issues in Paleoethnobotany: A Consideration of Issues, Methods, and Cases’. In Integrating Zooarchaeology and Paleoethnobotany: A Consideration of Issues, Methods, and Cases, edited by A. M. Van Der Warker and T.M. Peres, 37–64. New York: Springer. Yen, Douglas E. 1974a. The Sweet Potato and Oceania: An Essay in Ethnobotany. Honolulu: Bishop Museum Press. ———. 1974b. ‘Arboriculture in the Subsistence of Santa Cruz, Solomon Islands’. Economic Botany 28: 247–84. ———. 1980. ‘Pacific Production Systems’. In South Pacific Agriculture: Choices and Constraints, 73–105. Manila: Asian Development Bank. ———. 1982. ‘The Southeast Solomon Islands Cultural History Programme’. Bulletin of the Indo-Pacific Prehistory Association 3: 52–66. ———. 1985. ‘Wild Plants and Domestication in Pacific Islands’. In Recent Advances in Indo-Pacific Prehistory. Proceedings of the International Symposium Held at Poona, December 19-21, 1978, edited by V. N. Misra and P. Bellwood, 315–26. New Delhi: Oxford & IBH Pub. Co. ———. 1989. ‘The Domestication of the Environment’. In Foraging and Farming: The Evolution of Plant Exploitation, edited by D. R. Harris and G. C. Hillman, 55–75. London: Unwin Hyman. ———. 1990. ‘Environment, Agriculture and the Colonisation of the Pacific’. In Pacific Production Systems: Approaches to Economic Prehistory. Papers from a Symposium at the XV Pacific Science Congress, Dunedin, New Zealand 1983, edited by D. E. Yen and J. M. J. Mummery, 258–77. Canberra [A.C.T]: Dept. of Prehistory, Research School of Pacific Studies, Australian National University. ———. 1991. ‘Polynesian Cultigens and Cultivars: The Questions of Origin’. In Islands, Plants, and Polynesians: An Introduction to Polynesian Ethnobotany, edited by P. A. Cox and S. A. Banack, 67–95. Portland: Dioscorides Press. Yen, Douglas E., and J. Head. 1993. ‘Kumara Remains in Pit O at P5/1288’. In The Archaeology of the Peripheral Pā at Pouerua, Northland, New Zealand, edited by D. G. Sutton, 56–64. Auckland: Auckland University Press. Zalucha, Leonard A. 1982. ‘Methodology in Paleoethnobotany: A Study in Vegetational Reconstruction Dealing with the Mill Creek Culture of Northwestern Iowa’. Ph.D. diss., Madison: University of Wisconsin. Zerega, Nyree J. C., Diane Ragone, and Timothy J. Motley. 2004. “Complex Origins of Breadfruit (Artocarpus Altilis, Moraceae): Implications for Human Migrations in Oceania.” Am. J. Bot. 91: 760–66.

332

Appendix A: Charcoal identifications, absolute counts and weights

333

V/9 VII/4 VII/7 VII/7 VII/7 VII/10 TP‐ VI/6 VI/5 VI/5 V/5 TP‐ IV/3 IV/3 IV/3 III/2 III/1 VII/4 III/3 VII/13 TP‐ VII/6 V/ III/U/G VII/5 III/7 VII/4 VII/ IV/4 VII/8 VI/7 IV/3 III/1 VII/8 V/ VII/ VII/8 V/5 TP‐ VII/10 TP‐ VII/ VII/ IV/2 V/5 TP‐ V/3 TP‐ IV/2 VII/ V/11 TP‐ VII/7 V/5 TP‐ VII/9 VII/9 III/1 IV/3 IV/2 V/4 TP‐ V/5 TP‐ Total V/6 TP‐ V/8 TP‐ Total TP‐ SP TP‐ TP‐ SP SP TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ fragment wt. ‐ TP TP 8 3 SP ‐ ‐ 8 8

1 6 8 3 1

3 3 7 4 7 4 1 6 6 3 6 1 4 7 ‐ #5641 #5737 3 4 4 7 7 4 3 4 4 3 4 3 7

6 5 7 6

7 5 5 4 3 6

3 6 4

#5649 #5648 #5703 #5739 #5741 57 1 #5670 #5637 #5677 #5679 #5715 #5671 #5672 #5720 #5635 #5740

#5705 #5652 #5709 #5628 #5651 #5707.2 58 1 #5680 #5667 #5659 #5722 #5722 #5659 #5668 57 3 #5678 #56682 #5662 #5656 #5664 #5710

‐

#5721 51 3 #5712 52 2 #5728 #5721 53 1 #5730 52 2 #5723 #5726 #99999 #5681 #5695 #5725 #5693 52 1 #5729 #5683 #5643 8

#5731 #5690 #5688

#5640 oven

ct. 1821294141352323753168711811 hearth oven oven oven oven hearth hearth

(Efe.

(Efe. (Efe. (poss.) (poss.) Layer / Level, Unit, Sample No., Feature

(Efe.

Ee4 2 (Efe.4)

(Efe.4) 6)

6) 6)

10) (Efe. (Efe.

)81 5) )73 5) 1 1 2 2 1 1 1 2 1 1 1 18801816801200201700929501701 .8 .7 3980680620220031060.696 1.066 0.083 0.272 0.632 0.678 33.968 1.177 2.785 0.12 0.147 2.935 0.039 0.147 0.022 0.112 1.628 0.108 11.838

011] [0 [0 00]1 [0.03] 00]5 [0.07] [0. [0.23] [0.17] [0.04] 01]2 [0.12] [0.03] [0.14] [0.18] [0.34] 01]1 [0.17] [0.14] [0.06] [0.02] [0.02] [5.61] [4.09]

Aleurites moluccana endocarp 01 2 [0.1] . 11] 17]

01]8 [0.11] cf. Alphitonia marquesensis 1 1 3 1 2 1 2 2 2 2 1

024] [0 [0 [0. [0.06] [0.05] [0.01] [0.11] 00]1 [0.04] [0.05] [0.16] [0.05]

Artocarpus altilis 01 1 [0.1] [0.4] . 24] 34]

00]1 [0.04] [0.05] [0.02] cf. Artocarpus altilis 1

[0.02] cf. Barringtonia asiatica

00]2 [0.06] [0.04] [0.05] Calophyllum inophyllum 1

00]2 [0.04] cf. Calophyllum inophyllum 1 1 1 1 6 1 2 1 8 1 1

007] [0 004] [0 04]7 [0.45] [0 [0 [0.16] [0.04] 00]19 [0.04] 04]5 [0.41] 00]9 3 [0.03] [0.04] 01]2 [0.16] [0.05] 14]9 [1.43] [0.02] Celtis pacifica . . 07] 04] 1 1

00]2 [0.09] [0.06] cf. Celtis pacifica 1 1 1

00]1 [0.03] [0.07] 00]3 [0.02] Cerbera manghas 3 3 5

028] [0 [0 [0.16] 03]38 [0.36] 04]16 [0.46] 04]28 [0.44] [0.08] Cocos nucifera 6 .

28] [1] 1 2 1 1

00]6 [0.03] 00]19 [0.07] 09]10 [0.92] 00]1 [0.06]

cf. Cocos nucifera 01 1 [0.1] 30 30 11 20 38 15 11 48 57 13 10 14 11 7 7 6 6 1 1 4 8 2 1 1 6 2 7 1 16 1 5 4 1

038] [0 42] [0 [0 [0 [2 274] [2 [0. 06]1 [0.66] [0.04] 15]2 [1.52] 01]4 [0.18] [0.57] [0.56] [0.08] [1.56] [0.35] 12]3 [1.22] [1.01] [0.58] [1.09] [0.06] [0.45] 22]2 [2.29] [0.27] [2.87] [0.79] [0.16] [0.35] [0.07] [0.12] [0.15] [0.19] [3.14] [0.86] 16]3 [1.69] 09]1 [0.92] [0.45] [0.65] [0.88] [0.04] 08]1 [0.88] [0.45] [0.61]

Cocos nucifera endocarp [0.1] [0.9] . . .

38] 42] 74] 66] [1]

[0.43] [0.18] 00]1 [0.07] Coprosma spp. 4 4

037] [0 [0 [0.14] [0.12] cf. Coprosma spp. . 37] 1 4

00]1 [0.07] [0.07] [0.14] Cordia subcordata

00]1 [0.08] cf. Crossostylis biflora 11 2 2 1

015] [0 [0 [0.13] 00]3 [0.07] 02]1 [0.22]

cf. Cyclophyllum barbatum [0.5] . 15] 1 3

[0.05] [0.18] [0.13] [0.14] [0.09]

cf. Dodonaea viscosa [0.1] VII/7 Total VII/6 VII/9 VII/5 VII/9 VII/4 VII/8 VII/4 VII/ III/U/G VII/8 VII/4 VII/ III/7 III/1 VII/8 VII/13 TP‐ VI/7 III/3 III/1 VII/7 VII/7 VII/7 VII/10 TP‐ VI/6 VI/5 VI/5 V/5 TP‐ V/5 TP‐ IV/3 IV/3 IV/3 III/2 III/1 VII/10 TP‐ Total VII/ IV/3 VII/ IV/2 VII/ V/ IV/2 IV/4 IV/2 IV/3 V/5 TP‐ V/ V/5 TP‐ V/4 TP‐ V/3 TP‐ V/11 TP‐ V/5 TP‐ V/9 TP‐ V/8 TP‐ V/6 TP‐ TP‐ SP TP‐ TP‐ SP SP TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ fragment wt. ‐ TP TP 8 3 SP ‐ ‐ 8 8

6 1 8 3 3

3 4 7 7 4 1 1 6 1 7 6 4 6 3 ‐ #5641 #5737 4 3 4 4 7 3 7 4 3 3 4 7 4

6 3 7 6 5 4 5 6 5 4 5 3 7 6

#5649 #5648 #5739 #5703 #5741 #5677 #5670 #5679 #5671 #5720 #5715 #5672 #5637 #5635 #5740

#5628 #5707.2 #5709 #5651 #5705 #5652 #5678 #5680 #5667 #5659 #5722 #5664 #5710 #5656 #5668 #5662 #56682 #5722 #5659

2 6

4 ‐

#5721 #5693 #5723 #5729 #5726 #5683 #99999 #5725 #5730 #5681 #5712 #5695 #5728 #5721 #5643 8

#5690 #5731 #5688

#5640 oven

ct. hearth oven oven oven oven hearth hearth

(Efe.

(Efe. (Efe. (poss.) (poss.) Layer / Level, Unit, Sample No., Feature

(Efe.

(Efe.4)

(Efe.4) 6)

6) 6)

10) (Efe. (Efe.

5) 5) 5

01]4 [0.15] .4 .923202203267203901814500216510400106122688100 . 1.01 0.1 0.06 8.801 2.276 0.611 0.091 1.054 1.635 0.022 1.475 0.108 0.349 6.702 0.312 0.202 2.352 0.09 0.149 Guettarda speciosa 525133223381171628281572413 1 1

[0.03] 00]1 [0.06] cf. Guettarda speciosa 10 2 1 2 2 1 3 3 3 1 1 3 3 2 6 1

6 6

[0.14] 01]1 [0.15] 00]3 [0.05] [0.06] [0.32] [0.01] [0.08] 02]1 [0.22] 00]9 [0.05] [0.27] 006] [0 13] [0 [0 [0 02]1 [0.25] [0.04] [0.18] 00]3 [0.05]

Maytenus crenata 03] [0 [0 . . 06] 13] . 3] 2 2

017] [0 [0 [0.03] Melicope spp. . 17]

03]6 [0.31] cf. Melicope spp. 5 4 4

[0.08] [4.83] 039] [0 [0 02]1 [0.27] [1.13] Pandanus tectorius . 39] 3

03]1 [0.35] cf. Pandanus tectorius drupe 1 1

[0.05] [0.02] [0.03] Phyllanthus cf. marchionicus 1 7

[0.12] [1.36] cf. Phyllanthus marchionicus 1

[0.02] cf. Pipturus sp. 1 1 5 4

006] [0 [0 06]2 [0.69] 5 [0.16] 01]3 [0.15] 05]11 [0.58] Planchonella spp. . 06] 2 2 5 2 1

018] [0 [0 [0.14] 02]2 [0.22] [0.06] [0.22] 00]1 [0.06] 01]5 [0.18] cf. Planchonella spp. . 18]

00]1 [0.09] Premna serratifolia 7 1

[0.39] [0.22] cf. Premna serratifolia 16 1 1 2 7 1

[0.05] 00]1 [0.03] 16]2 [1.64] 01]3 [0.15] 03]2 [0.36] 00]1 [0.04] cf. Psydrax odorata 84 12 1 1 7 1 8 1 6 6 2 1 2

[0.05] [0.04] [0.58] [4.38] [0.02] [0.24] [0.02] 048] [0 [0.06] [0 [0.04] [0.73] [0.16] [0.22] [0.17] [0.14] 00]1 [0.03] [0.07] [0.18] [0.24] [0.04] [0.71]

Sapindus saponaria [0.2] . 48]

[0.06] cf. Sapindus saponaria 4

Sapindus saponaria root [0.1] 10 2

[0.84] [0.15] [0.02] Terminalia spp. VII/7 Total VII/6 VII/9 VII/5 VII/9 VII/4 VII/8 VII/4 VII/ III/U/G VII/8 VII/4 VII/ III/7 III/1 VII/8 VII/13 TP‐ VI/7 III/3 III/1 VII/7 VII/7 VII/7 VII/10 TP‐ VI/6 VI/5 VI/5 V/5 TP‐ V/5 TP‐ IV/3 IV/3 IV/3 III/2 III/1 VII/10 TP‐ Total VII/ IV/3 VII/ IV/2 VII/ V/ IV/2 IV/4 IV/2 IV/3 V/5 TP‐ V/ V/5 TP‐ V/4 TP‐ V/3 TP‐ V/11 TP‐ V/5 TP‐ V/9 TP‐ V/8 TP‐ V/6 TP‐ TP‐ SP TP‐ TP‐ SP SP TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ fragment wt. ‐ TP TP 8 3 SP ‐ ‐ 8 8

6 1 8 3 3

3 4 7 7 4 1 1 6 1 7 6 4 6 3 ‐ #5641 #5737 4 3 4 4 7 3 7 4 3 3 4 7 4

6 3 7 6 5 4 5 6 5 4 5 3 7 6

#5649 #5648 #5739 #5703 #5741 #5677 #5670 #5679 #5671 #5720 #5715 #5672 #5637 #5635 #5740

#5628 #5707.2 #5709 #5651 #5705 #5652 #5678 #5680 #5667 #5659 #5722 #5664 #5710 #5656 #5668 #5662 #56682 #5722 #5659

2 6

4 ‐

#5721 #5693 #5723 #5729 #5726 #5683 #99999 #5725 #5730 #5681 #5712 #5695 #5728 #5721 #5643 8

#5690 #5731 #5688

#5640 oven

ct. hearth oven oven oven oven hearth hearth

(Efe.

(Efe. (Efe. (poss.) (poss.) Layer / Level, Unit, Sample No., Feature

(Efe.

(Efe.4)

(Efe.4) 6)

6) 6)

10) (Efe. (Efe.

5) 5) 3

[0.13] .2 .8 .8 .20141.2 .3 .502 .9 .4 .5 .6 .5 .0 .7 .0 .0 0.079 0.101 0.102 0.175 0.008 0.052 0.869 1.253 6.244 1.992 0.22 0.25 0.138 19.325 0.114 0.12 0.089 2.188 0.126 cf. Terminalia spp. 34612323438517721515211221 14 14 4 1 1 3 2 1 6 2 2 2 1 2 2 3

01]1 [0.11] [0.12] 00]1 [0.01] [0.24] 00]7 1 [0.05] [0.01] [0.26] [0.35] [0.08] [0 00]12 [0.05] 049] [0 [0.02] 00]5 [0.05] 02]7 [0.22] 01]8 [0.13] Thespesia populnea . 49] 1

00]70 [0.09] cf. Thespesia populnea 2 2

012] [0 [0 Wikstroemia cf. coriacea . 12] 2

00]1 [0.07] [0.04] Xylosma suaveolens 16 10 34 11 1 1 3 6 6 1 1 2 9 3 1 2 6 5 6 6 5 4

00]1 [0.02] 008] [0 00]1 [0.06] 53] [1 05] [0 [0 [1 [0 [0.01] [0.96] 02]2 [0.25] 02]2 1 1 [0.84] [0.56] [0.22] [0.18] 05]1 [0.57] [0.03] [6.71] [0.06] [0.22] 02]1 [0.28] 02]1 [0.26] 37]3 [3.77] [0.21] [1.29] [0.21] [0.21] [0.55]

unknown13 [0.2] . . . 08] 53] 05] 1

00]1 [0.01] [0.03]

cf. unknown13 01 1 [0.1]

02]1 [0.25] unknown18 5

02]2 [0.22] cf. unknown18 3 3 2 1

056] [0 [0 00]1 [0.03] 00]1 [0.03] [0.06] 2 [0.05] [0.05] 00]7 [0.06] 02]5 [0.27] [0.44] 00]8 [0.03]

unknown25 [0.4] . 56] 23 4 2 2 1 1 1 1 5 1 1 1 5 2

[0.26] 009] [0 007] [0 [0 [0 00]5 [0.04] [0.01] 01]1 [0.03] [0.13] [0.31] [0.12] [1.58] [0.05] [0.25] [0.05] [0.11] [0.29] [0.19] [0.89]

unknown26 [0.9] [0.1] 08 1 [0.8] . .

09] 07] 0 1 [0] 2 5 5

[0.08] 061] [0 [0 [0.14] [0.04] [0.38] cf. unknown26 . 61] 2 2 1 1 1 1 1

007] [0 008] [0 [0 [0 [0.31] 01]1 [0.13] [0.02] [0.06] 00]2 [0.02] [0.06] [0.06] [0.06] unknown27 . . 07] 08]

00]1 [0.05] unknown28

[0.01] unknown29 1

[0.18] unknown30 1 1

002] [0 [0 [0.08] unknown31 . 02] 1

[0.06] [0.04] unknown32 1 1

008] [0 [0 indeterminate . 08] VII/7 Total VII/6 VII/9 VII/5 VII/9 VII/4 VII/8 VII/4 VII/ III/U/G VII/8 VII/4 VII/ III/7 III/1 VII/8 VII/13 TP‐ VI/7 III/3 III/1 VII/7 VII/7 VII/7 VII/10 TP‐ VI/6 VI/5 VI/5 V/5 TP‐ V/5 TP‐ IV/3 IV/3 IV/3 III/2 III/1 VII/10 TP‐ Total VII/ IV/3 VII/ IV/2 VII/ V/ IV/2 IV/4 IV/2 IV/3 V/5 TP‐ V/ V/5 TP‐ V/4 TP‐ V/3 TP‐ V/11 TP‐ V/5 TP‐ V/9 TP‐ V/8 TP‐ V/6 TP‐ TP‐ SP TP‐ TP‐ SP SP TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

TP‐ TP‐ TP‐ TP‐ fragment wt. ‐ TP TP 8 3 SP ‐ ‐ 8 8

6 1 8 3 3

3 4 7 7 4 1 1 6 1 7 6 4 6 3 ‐ #5641 #5737 4 3 4 4 7 3 7 4 3 3 4 7 4

6 3 7 6 5 4 5 6 5 4 5 3 7 6

#5649 #5648 #5739 #5703 #5741 #5677 #5670 #5679 #5671 #5720 #5715 #5672 #5637 #5635 #5740

#5628 #5707.2 #5709 #5651 #5705 #5652 #5678 #5680 #5667 #5659 #5722 #5664 #5710 #5656 #5668 #5662 #56682 #5722 #5659

2 6

4 ‐

#5721 #5693 #5723 #5729 #5726 #5683 #99999 #5725 #5730 #5681 #5712 #5695 #5728 #5721 #5643 8

#5690 #5731 #5688

#5640 oven

ct. hearth oven oven oven oven hearth hearth

(Efe.

(Efe. (Efe. (poss.) (poss.) Layer / Level, Unit, Sample No., Feature

(Efe.

(Efe.4)

(Efe.4) 6)

6) 6)

10) (Efe. (Efe.

5) 5) 1 1 1 1 2 3 5 1 1

00]11 1 [0.02] [0.04] 8 [0.14] 002] [0 [0 [0.06] 03]1 [0.35] [0.07] 00]5 [0.04]

indeterminate bark [0.1] 1 .500 .20081. 04700101 .0 5.635 6.503 0.15 0.021 30.407 16.7 0.058 1.12 0.09 0.85 .

02] 0 1 [0] 732 333129 159 94 2 1 383 33 1 25 3 17 2

00]8 [0.08] 00]1 [0.01] indeterminate endocarp 4 6 1 2 2 3

[0.05] [0.64] 00]6 [0.02] [0.03] [0.06] 1 [0.04] 00]9 [0.01] [0.06]

indeterminate parenchymous tissue 02 1 [0.2] 1 1

006] [0 [0 cf. indeterminate parenchymous tissue . 06] 3 2 2 3 1 1 1 1 1 1 1 5 1

2 2

01]9 [0.11] 5 [0.48] 1 [0.02] 007] [0 [0 00]23 [0.09] 00]2 [0.02] 02]20 [0.29] 00]8 [0.04] 00]3 [0.01] [0.03] 00]5 [0.02] 00]1 [0.03] 03]34 [0.36] 00]16 [0.01] 00]21 [0.03]

indeterminate wood 01] [0 [0 . 07] . 1] 14 14 10 10 24 15 11 12 12 11 10 14 14 14 1 2 1 1 6 2 4 3 6 8 1 4 4

09]2 [0.94] [1 [0 04]4 [0.41] 4 2 [1.68] [0.28] 127] [1 35] [0 02]1 98 [0.22] [0.89] [0.54] 09]8 [0.99] 00]6 [0.07] [0.33] 1 [0.03] [3.29] 00]2 [0.03] 01]1 [0.14] [0.09] [2.54] 00]1 [0.03] 06]1 [0.64] [0.41] 02]2 [0.24] [0.36] [0.15] [0.15] [0.95] 1 [0.04] 57]4 [5.72] [1.03] [0.11] 00]12 [0.03] [1.93] 08]2 [0.88] 05]1 [0.59] 01]1 [0.12] 01]2 [0.15] 01]13 11 [0.11] [0.09] [0.36]

indeterminate angiosperm [0 01 2 [0.1] 06] [0 03 17 [0.3] 01 2 [0.1] [0.5] 04 2 [0.4] 02 1 [0.2] . . 27] 35] . 6]

00]2 [0.02] cf. indeterminate angiosperm

01]2 [0.15] cf. indeterminate angiosperm root 9 9 1 1 2 2 4 2 1 6 1 1 7 3 2 5

036] [0 009] [0 09] [0 [0 [0 [0 00]6 [0.06] 02]1 [0.21] 16 [0.06] 00]4 [0.08] 4 [0.49] 00]6 [0.02] 14]8 [1.42] 00]3 [0.03] 04]6 [0.41] 00]4 [0.01] [0.04] 00]2 [0.04] 09]8 [0.94] 01]2 [0.11] [0.23] 01]7 [0.16] 00]14 [0.05] 00]1 [0.01] 01]5 [0.18] 00]5 [0.04] 00]45 [0.06]

indeterminate monocot 01 1 [0.1] 01 10 [0.1] 07 4 [0.7] [0.2] 01 82 [0.1] 01 9 [0.1] . . . 36] 09] 09] 15 1 1 5 3 5 5

007] [0 [0 01]65 [0.18] 06]66 [0.66] 18 [0.13] 02]75 [0.21] 00]50 [0.06] 157 [0.43] 01]24 [0.19] 14 [0.13] 00]66 [0.08] 00]18 [0.01] 01]5 [0.12] 04]55 [0.42] 03]12 [0.33] 00]83 [0.04] 00]27 [0.08] 02]85 [0.21] 21 [0.02] 01]132 [0.11] 02]31 [0.24] 03]96 [0.35] 00]108 [0.05] 02]112 [0.21] 00]10 [0.02] 03]25 [0.37] 00]17 [0.01]

unidentifiable 01] [0 [0 01 38 [0.1] 01 80 [0.1] 02 51 [0.2] 02 96 [0.2] 01 103 [0.1] 01 33 [0.1] . 07] . 1] 63 63 86 86 34 34 24 18 26 180.831 7

2 2 2

1 [10.54] [11.94]

[3 [8 [1 302] [3 886] [8 73] [1 [1.98] [1.03] [9.53] [2.62] [3.73] [3.94] [6.49] [3.85] [2.12] [8.66] [8.82] [0.11] [0.08] [1.32] [4.86] [0.45] [2.84] [1.46] [0.15] [2.12] [0.24] [7.81] [2.34] [5.76] [0.49] [4.09] [9.99] [1.16] [1.39] [0.48] [0.53] [7.43] [1.51] [4.11] [1.14] [0.45] [1.47] [6.49] [5.91] [1.55] [1.12] [0.53] [0.45] [1.21] [1.94] 35

2373 Frag ct. [wt.] . . .

02] 86] 73] [2] [7] H III H H H III D H III H VII H VII H III E D H III H VII E Total H III Total H III F H III H III H III F H III V V IV IV? III III

III III

/ /

/ / /SP‐ / / /3 / /4 / /4

/ / / / /

fragment wt. SP SP / / /

/ SP SP TP TP‐ TP TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

SP SP SP SP TP‐ TP‐ TP‐ ‐ ‐ ‐ 5 6 ‐ ‐ 6 7 5 ‐ ‐ ‐ 3 2 3 2 2 2 3 2 2 1 ‐

6 6 5 75 1 #7159 #7229 7 #7224

#6794 #6830

3 2 #7227 #6810 #6828 #7147

#6800 75 10 #7153 #6832 #7242

#7244 #6841 #6841

#7228 75 1 extension #7155 #7221 #7226 79 1 #7293

t 62051141195331011161121312ct.

oven

deep

deep oven oven deep deep pit hearth oven oven

deep oven

(Efe. #7151

pit

pit(no (Efe (Efe. (E (E (Efe. pit pit

pit

(Efe. or fe fe

(Efe.

12) (Efe. (Efe.

. . . (no rake

Zone, Layer / Level, Unit, Sample No., Feature 20) 20) 10) 2) 10)

10) )5 #)

11)

)4 #)

11) 11) ‐ out

(Efe.

19)

2 1 2

[0.18] [0.11] 00]1 [0.09] [0.38] Allophylus marquesensis .61.412 .622 52 .503 . .210 .900 .10113 .40.22 0.04 1.36 0.1 0.01 0.04 2.29 1.01 1.42 0.3 0.34 0.35 15.27 2.21 0.36 1.23 14.54 0.76

[12.26]

08]1 [0.89] 00]1 [0.03] [1.36] Artocarpus altilis 1

02]1 [0.25] 03]5 [0.35] [0.33]

Calophyllum inophyllum [0.3]

03]2 [0.36] Casuarina equisetifolia 3 3 4

[0.42] [0 03]3 [0.31] 06]13 [0.68] 00]11 [0.05] [0.75] Cocos nucifera . 42] 4 1

03]34 [0.33] 00]21 [0.02] cf. Cocos nucifera 17 5 5 1 1 1 5 2 2

1 [0.27] [0 00]1 1 [0.08] [0.07] [0.75] 23]9 [2.31] 02]1 [0.21] 31]1 [3.17] [0.13] [0.42] [0.22] [0.68] 05]1 [0.54] [3.52]

Cocos nucifera endocarp 28 3 [2.8] [0.1] . 27]

[0.34] cf. Cordia subcordata 1

[0.13] [0.14] 00]1 [0.03] cf. Cyclophyllum barbatum 7 2

[0.45] [0.37]

Guettarda speciosa [0.6]

09]1 [0.91] [0.06] [0.04] cf. Guettarda speciosa 14 2

[1.99]

cf. Hibiscus spp. [0.3]

00]1 [0.04] cf. Hibiscus spp. 1

[0.01] Maytenus crenata

00]13 [0.07] 00]1 [0.03] Pandanus tectorius drupe

[1.36] cf. Pandanus tectorius drupe

[0.04] cf. Pemphis acidula 1

[0.14] [0.08] Phyllanthus cf. marchionicus Total Total H VII H III E D H H H III D H VII H III E H VII H III H III F F H III H III H III H III H III H III V V IV IV? III III

III III

/ /

/ /SP‐ /

/3 / / /4 / / / / / / / /4

fragment wt. SP SP / / /

/ SP SP TP TP‐ TP TP‐ TP‐ TP‐ TP‐ TP‐ TP‐ TP‐

SP SP SP SP TP‐ TP‐ TP‐ ‐ ‐ ‐ 6 5 ‐ ‐ 6 7 5 ‐ ‐ ‐ 3 2 3 1 2 2 3 2 2 2 ‐

6 6 5 #7229 #7159 7 #7224

#6794 #6830

2 3 #6800

#6810 #6828 #7227 #7147

2 #7153 #6832 #7242

#7244 #6841 #6841

extension #7228 #7155 #7221 #7226 #7293

ct.

oven

deep

deep oven oven deep oven pit hearth deep oven

deep oven

(Efe. #7151

pit

pit(no (Efe. (Efe (Efe. (Efe. (Efe. pit pit

pit

(Efe. or

(Efe.

12) (Efe. (Efe.

. (no rake

Zone, Layer / Level, Unit, Sample No., Feature 20) 20) 2) 10) 10)

10) #)

11)

11) 11) ‐ out

(Efe.

19)

06]1 [0.67] Planchonella spp. .701 .900 . .771 .604 .504 . .610 .90.69 0.09 1.01 5.56 1.3 0.43 0.05 0.49 0.66 7.14 0.27 7.2 0.09 6.39 0.14 0.67 41221242696213137310115 1

01]4 [0.14] cf. Premna serratifolia 15 1 1 1 2 1

05]1 [0.59] 7 [0.07] 50]1 [5.05] [0.31] 00]24 [0.06] 01]4 [0.17] [0.14] Sapindus saponaria 1

[0.09] Sapindus saponaria root 2 7 2

[0.35] [0.02] [0.32] [5.15] 04]1 [0.44] 02]1 [0.22]

Thespesia populnea 07 3 [0.7]

[0.11] [0.16] Xylosma suaveolens 1 7 7 7 6 9 7

00]1 [0.08] [1.19] [1 00]5 [0.05] 08]1 [0.85] 09]1 [0.98] 04]1 [0.46] 06]1 [0.67] 04]14 [0.45] 10]2 [1.02] 13]3 [1.39] unknown13 . 19]

00]2 [0.07] 05]2 [0.59] unknown19 2

[0.49] cf. unknown21

00]3 [0.05] indeterminate bark

04]5 [0.43] indeterminate parenchymous tissue 1 1 2 1

[0.09] [0 00]5 [0.05] 04]1 [0.41] 00]2 [0.08] 04]11 [0.41] 02]14 [0.26] indeterminate unknown . 09] 13 1 1 2 2

[0.08] [0 02]30 [0.26] 04]6 [0.47] 04]1 [0.46] 02]34 [0.23] 03]8 [0.34] 02]11 [0.22] 00]12 [0.06] 08]54 [0.86] [0.52] 01]12 [0.14] 01]13 [0.11] 05]6 [0.59] 11]2 [1.12]

indeterminate angiosperm 01 10 [0.1] . 08] 2 2

[0.29] [0 02]4 [0.26] [0.46] indeterminate monocot . 29] 1

00]3 [0.09] undetermined 4 3 3

00]19 [0.09] 00]43 [0.08] 02]68 [0.22]

unidentifiable [0.2] [0 01 119 [0.1] . 2] 22 22 4 3

[12.26] [10.46]

8 1

[2 [2.54] [2.13] [0.83] [0.53] [1.47] [6.03] [5.48] [2.84] [1.49] [6.03] [1.66] [1.96] [1.25] [8.07] 74.03

Frag ct. [wt.] [0.2] [0.6] [2.2] [5.9] [0.1] . 480 54] Total Total Mutoka TU‐1II/3#100013hearth Vaiu'ua Mutoka TU‐1 #100014oven Mutoka TU‐2 #100005 Mutoka TU‐2 1#100004 Mutoka TU‐1III #100015 Mutoka TU‐1 3#100012 Vaiu'ua Ototemui Mutoka TU‐2 3#100006 Mutoka TU‐1 3#100011 Ototemui Ototemui Ototemui Ototemui Vaiu'ua wt. fragment River

River River TU‐4 1#100007 TU‐4 3#100009 TU‐4 2#100008 TU‐6 #100003 TU‐4 basal scattered 144cmbs #100002 120 ct. cmbs #100001 #100010 #100000

Stone Area, Unit, Level, Sample #, Feature ‐lined oven oven oven pit

[0.1] cf. Allophylus marquesensis . .251 .40103 . .803 .200 .30.21 0.23 0.03 1.22 0.35 6.08 0.3 0.33 0.1 0.54 5.13 0.12 0.1 134831415735233 3 01]9 [0.12] Alstonia costata 1 [0.07] 00]1 2 [0.09] 8 5 1 4 5 6 [0.22] [0.04] [1.05] [0.64] 04]2 [0.43] [1.58] 7 [0.4] [0.61] Artocarpus altilis [0.03] [0.51] Calophyllum inophyllum 1

[0.1] Cocos nucifera 1 2 [0.14] [0.09] 1 [0.1] Cocos nucifera endocarp

1 [0.3] Crossostylis biflora 10 2 [0.09] 7 1 6 9 [0.66] 9 3 [0.07] 01]1 5 [0.16] [0.86] [1.52] [1.14] [1.07] 4 1 [0.01] [0.4] cf. Hibiscus spp. [0.1] 2 [0.31] [0.04] cf. Hibiscus spp. 4 02]2 [0.22] Inocarpus fagifer 1 [1]

[0.03] cf. Inocarpus fagifer 3 [0.23] Metrosideros collina 3 [0.21] Phyllanthus marchionicus Total Total Mutoka TU‐1 #100014oven Vaiu'ua Mutoka TU‐1II/3#100013hearth Mutoka TU‐1 3#100012 Mutoka TU‐1 3#100011 Vaiu'ua Ototemui Ototemui Ototemui Ototemui Ototemui Mutoka TU‐2 3#100006 Mutoka TU‐2 #100005 Mutoka TU‐2 1#100004 Mutoka TU‐1III #100015 Vaiu'ua wt. fragment River River River TU‐6 #100003 TU‐4 basal TU‐4 3#100009 TU‐4 2#100008 TU‐4 1#100007 scattered 144cmbs #100002 120cmbs #100001 ct. #100010 #100000

Stone Area, Unit, Level, Sample #, Feature ‐lined oven oven oven pit

1 [0.14] cf. Psydrax odorata .413 .400 .405 .168 . 4.9 1.4 6.85 0.11 0.58 0.04 0.05 8.54 1.34 0.14 12775223348430 3 [0.21] 04]21 9 [0.41] 05]14 6 [0.53] 01]1 9 [0.19] Sapindus saponaria 10 [1.13] 10 [1.39] 1 [0.02] 1 [0.01] 1 [0.06] 4 [1.65] 8 [0.99] [2.37] [0.47] 00]1 [0.05] 4

Thespesia populnea [0.4] 1 [0.02] [0.03] cf. Thespesia populnea 2 [0.04] unknown33 3 [0.58] unknown34 00]2 2 [0.01] 1

01 3 [0.17] [0.1] indeterminate bark 9 1 8 3 [0.84] 2 6 2 4 2 3 3 [0.87] [0.03] [0.44] [0.22] [0.45] [0.08] [1.57] [0.25] [0.01] 15]4 [1.56]

[0.36] indeterminate angiosperm

00]6 1 [0.08] [1.4] indeterminate monocot 18 1 1 1 1 6 20]47 [2.03] 01]16 [0.17] 09]21 [0.95] 00]32 [0.07] 01]16 [0.17] 13]22 [1.33] 1

01 32 [0.1] unidentifiable 11 42 [3.54] 21 17 12 20 8 [6.34] [0.91] [0.71] [4.24] [1.91] 6 [0.2] [2.34] [0.27] [2.16] [2.87] [4.96] [3.38] [1.37] [3.38] [0.11]

38.69 Frag ct. [wt.] 329 IV/5 TP-16#858 IIIb/ SP-2#146oven (Efe.D97) IV/5 TP-16 #884 postmould (Efe. 25) (Efe. IV/5 TP-16#884postmould IV/5 TP-14#797 IV/ TP-15 #845 undetermined (Efe. 22) TP-15#845undetermined(Efe. IV/ IIIb-IV/5 TP-8#611 IV/ TP-16#1077 IV/ IV/ TP-11#1078 IV/ IIIb-IV/4 TP-15#814 TP-10#1083IIIb-IV/ undetermined(8) 6) IIIb/6 TP-9#615oven(Efe. IIIb/6 TP-12#690ove IIIb/3 TP-16#840 IIIb/3 TP-15#796 D97) IIIb/ SP-2#139oven (Efe. 6) IIIb/ TP-9#882oven(Efe. 5) IIIb/ TP-9#602oven(Efe. 5) IIIb/ TP-9#586oven(Efe. 7) IIIb/ TP-8#640oven(Efe. 20) IIIb/ TP-14#826oven(Efe. IIIb/ TP-12#1072 Total fragmentTotal ct. IV/5 TP-13#731 IIIb/6 TP-10#716 IIIb/5 TP-9#583 IIIb/5 TP-10#1082 IIIb/4 TP-8#601 IIIb/4 TP-13#726 Total wt. Total n (Efe. 6) (Efe. n Layer/Level, Unit, Sample No., Feature 1 [0.04] 1 [0.18]

.1 .506 .700 .606 .0 .206827 .10170110290.496 0.209 0.131 0.177 0.31 2.78 0.678 0.02 6.509 0.65 0.26 0.02 0.07 0.66 0.15 0.217 Allophylus marquesensis 5282 5 1 3 1 2 22 8 2 75 2 1 1 1 1 1 2 1 [0.15] cf. Alstonia costata 1 [0.66] Artocarpus altilis 1 [0.07] Barringtonia asiatica 1 [0.02] Celtis paciﬁca 1 [0.26] cf. Celtis paciﬁca 01]1[0.12] 1 [0.15] 05 1[0.07] 1 [0.5] Cocos nucifera 19 [1.01] 17 [1.99] 2 [0.07] 2 [0.21] 3 [0.27] 7 [0.95] 3 [0.24] 1 [0.04] 1 [0.02] 5 [0.46] 1 [0.07] 5 [0.3] 7 [0.7] Cocos nucifera endocarp 2 [0.02] Cordia subcordata 1 [0.08] 5 [0.51] 1 [0.05] 00]1[0.21] 1 [0.04] cf. Cyclophyllum barbatum 1 [0.31] 1 [0.03] 1 [0.05] 4 [0.27] 1 [0.77] 3 [0.14]

10 [1] Hibiscus spp. 1 [0.09] 1 [0.22] Pandanus tectorius drupe 01]3[.3 1[0.21] 3[0.13] 1 [0.18] Phyllanthus cf. marchionicus

cf. Phyllanthus marchionicus

cf. Planchonella spp. 1 [0.17] 2 2 [0.27] [0.05] cf. Psydrax odorata IV/ IIIb‐IV/5 TP8#611 IIIb‐IV/4 TP15#814 IIIb‐IV/ IIIb/6 TP‐9#615 oven(Efe. IIIb/6 TP‐12#690 oven(Efe. IIIb/6 TP‐10#716 IIIb/5 TP‐9#583 IIIb/5 TP‐10#1082 IIIb/4 TP‐8#601 IIIb/4 TP‐13#726 IIIb/3 TP‐16#840 IV/ IIIb/3 TP IIIb/ SP‐2#146 oven(Efe. IIIb/ SP‐2#139 oven(Efe. IIIb/ TP‐9#882 oven(Efe. IIIb/ TP‐9#602 oven(Efe. IIIb/ TP‐9#586 oven(Efe. IIIb/ TP‐8#640 oven(Efe. IIIb/ TP‐14#826 oven(Efe. IIIb/ TP‐12#1072 IV/ IV/5 TP‐16#884 postmould IV/5 TP‐16#858 IV/5 TP‐14#797 IV/5 TP‐13#731 Total Total TP‐15#845 undetermined(Efe. TP‐11#1078 TP‐16#1077 wt. fragment TP‐10#1083 ‐15 #796 ct. undetermined(8) Layer/Level, Unit, Sample No., Feature D97) D97) 6) 5) 5) 7) 6) 20) (Efe. 6) 25) 22) 27 [2.68] 17 [2.94] 16 [2.36] 20 [5.18] 64 [7.06] 2 [0.05] 7 [0.45] 1 [0.03] 5 [1.06] 3 1 [0.02] 1 [0.04] 2 [0.32] 9 [0.53] 02]1 3 [0.28] 1 [0.06] 2 [0.69] 91600 5.5 .6 .526501703201 .710 13.557 1.08 8.17 0.19 0.322 0.187 2.605 0.05 0.868 159.052 0.08 29.126 [0.38] 5 [4.4] 8 [0.6] Sapindus saponaria 9 3 11858 126 9 83 5 8 1 31 1 9 738 2 194 00]4 1 [0.04] 00]19 [0.04] cf. Sapindus saponaria 193 112 88 [32.34] 98 [9.49] 11 55 58 18 34 15 [81.01] 1.9 1 [11.59] 8 [0.43] 1 1 2 [0.34] 1 2 [0.39] 3 1 7

[8.14] [0.04] [0.06] [1.54] 1 [6.12] [0.05] [0.26] [0.26] 00]1 [0.01] [0.88] [1.29] 20]4 [2.06] [0.51] 7 [1.15]

[1.1] Thespesia populnea 2 [0.12] [0.02] [0.04] [0.64] [0.05] unknown13 1 [0.05] cf. unknown13 31

[2.6] unknown15 1 01]1 [0.19] unknown21 1 [0.06] [0.06] 6

[0.2] indeterminate bark 3 1 00]4 [0.04] 00]5 [0.05] 1

[0.1] indeterminate unknown 15 00]1 2 [0.06] 00]5 1 [0.05] 4 [0.39] 9 1 2 [0.09] 4 [0.68] 6 [1.94] 7 [1.21] 1 2 [0.19] 1 [0.11] 2 [0.52] 7 6 [1.04] 05]1 [0.54] [0.06] 00]2 [0.09] [0.08] [0.28] [0.24] 4 [0.2]

[0.4] indeterminate angiosperm [0.55] 01]8 [0.17] [0.06]

[0.3] indeterminate monocot 45 12 16]209 18 [1.62] 15 4 15]93 5 [1.56] 10]29 6 [1.08] 1 1 3 3 03]25 5 [0.33] 47]150 [4.75] 04]40 [0.44] 09]120 [0.94] 05]88 [0.57] 17]227 [1.77] 00]14 [0.05] 03]10 38 [0.33] [0.11] 00]28 [0.01] unidentifiable 10 10 27 55 13 10 15 18 18 63 [14.48] [11.48] [22.42] [86.24] 8 [0.43] 9 228.61 [1.06] [11.5] [0.46] [1.72]

[5.13] [2.04] [0.46] [2.15] [2.92] [2.12] [33.9] [3.65] [6.49] [0.54] [4.73] [2.32] 5 [0.3] 3

[4.07] Frag ct. [wt.] [3.2] [1.2] [3.1] [0.5] 1335 IV/8 TP‐51 #7241 IV/7 TP‐50 #6909 IV/4 TP‐53 #7077 III/n/a III/IV/6 TP‐49 #6925 III/6 TP‐50 #7237 IV/6 TP‐51#7240 postmould IV/6 TP‐51 #7132 IV/5 TP‐53 #7235 III/3 TP‐51 #6944 III/3 TP‐50 #6877 III/4 TP‐49 #7188 III/3 TP‐53 #7100 III/4 TP‐50 #6880 III/5 TP‐51 #7122 III/5 TP‐50 #7086 IV/4 TP‐53#7059undetermined(Efe. IV/4 TP‐53 #7058undetermined(Efe. III/3 TP‐49 #7186 III/2 TP‐53 #7079 III/4 TP‐50 #6881 III/5 TP‐49 #6900 III/5 TP‐50 #6893 III/5 TP‐50 #6892 Total Total III/ TP‐51#7125 III/4 TP‐51 #7113 III/4 TP‐50 #7071 wt. fragment TP‐53 #7104 ct. undetermined(Efe.

Layer/Level, Unit, Sample No., Feature (Efe. 79) 78) 83) 83)

1 00]2 [0.05] cf. Aleurites moluccana .517 .330 .736 .411 .335 .98.07 0.19 3.54 0.03 1.18 0.24 3.61 0.07 3.04 0.03 1.71 0.05 71814 5465 4 15 1 8 4 49 1 8 1 17 1 00]1 1 [0.05] 1 1 11 1 [0.23] [0.06] [0.11] [0.06]

Aleurites moluccana endocarp [1.2]

[0.03] Artocarpus altilis 1 [1.93] 1 3 1 2 [0.27] [0.24] [0.34]

[0.26] Cocos nucifera 1 [0.07] cf. Cocos nucifera 4 [0.19] 1 6 [0.35] 2 7 [0.27] 2 4 [0.22] 1 1 [0.39] 3 [0.11] 1 [0.12] 1 [0.08] 6 5 3 1 [0.06] [0.15] [0.13] [0.31] 1 [0.1]

[0.19]

[0.63] [0.27] [0.04] Cocos nucifera endocarp 4 [0.24] cf. Coprosma 1 6 [0.12] 1 [0.96]

[0.1] Cordia subcordata 1

[0.03] cf. Cordia subcordata 14 1 [0.58] [2.96] cf. Cyclophyllum barbatum 01]1 4 [0.19] Erythrina variegata 15 13 7 [1.22] 1 [0.07] 1 [0.04] 9 15 [1.18] [2.13] [1.56] [0.07] 3 [1.7] [0.1] Hibiscus spp. III/5 TP III/5 TP‐50 #6892 III/5 TP‐49 #6900 III/4 TP‐51 #7113 III/4 TP‐50 #7071 III/4 TP‐50 #6881 III/4 TP‐50 #6880 III/4 TP‐49 #7188 III/3 TP‐53 #7100 III/3 TP‐51 #6944 III/3 TP‐50 #6877 III/3 TP‐49 #7186 III/2 TP‐53 #7079 Total IV/8 TP‐51 #7241 IV/7 TP‐50 #6909 IV/6 TP‐51#7240 postmould IV/6 TP‐51 #7132 IV/5 TP‐53 #7235 IV/4 TP‐53#7059undetermined(Efe. IV/4 TP‐53 #7058undetermined(Efe. IV/4 TP‐53 #7077 III/n/a III/IV/6 TP‐49 #6925 III/6 TP‐50 #7237 III/5 TP‐51 #7122 III/5 TP‐50 #7086 III/ TP‐51#7125 Total fragment wt. TP‐53 #7104 ‐50 #6893 ct. undetermined(Efe.

Layer/Level, Unit, Sample No., Feature (Efe. 79) 78) 83) 83) 1 [0.04] 1 [0.11] 1

[0.1] Maytenus crenata .504 .105 .702 .62404 . .10.05 0.31 3.9 0.49 2.4 2.66 0.23 0.07 0.54 0.21 0.47 0.25 33 542 4 15 5 33 33 2 2 7 4 6 3 1 [0.13] 4 [0.12] 1 [0.22] Pandanus tectorius drupe 1 [0.09] 1[0.02] 1 [0.03] 1 [0.04] 1 [0.05] cf. Pandanus tectorius drupe 00]1[0.03] 1 [0.07] 5 [0.45] Planchonella spp. 1 [0.04] Premna serratifolia 1 [0.13] 01 03]2[.6 2[0.24] 2[0.16] 2[0.32] 1 [0.1] cf. Premna sp. 01]2[.9 1[0.21] 2[0.09] 2 [0.12] 3[0.31] 2 [0.09] 00]1[0.01] 3 [0.06] 3[0.05] 3[0.19] 1[0.02] 2[0.09] 2 [0.16] 2[0.04] 1 [0.03] 7[0.78] 2 [0.06] 1[0.06] 1 [0.06] 1 [0.07] 1[0.05] 6 [0.56] 1[0.03] 2 [0.08] 1[0.06] 9 [1.05] Sapindus saponaria 2 [0.17] 2 [0.07] 3 [0.18] 3 [0.3] Thespesia populnea 1 [0.02] unknown13 8 [2.15] 4 [1.56] unknown22 1 [0.26] unknown23 2 [0.05] unknown24 III/5 TP III/5 TP‐50 #6892 III/5 TP‐49 #6900 III/4 TP‐51 #7113 III/4 TP‐50 #7071 III/4 TP‐50 #6881 III/4 TP‐50 #6880 III/4 TP‐49 #7188 III/3 TP‐53 #7100 III/3 TP‐51 #6944 III/3 TP‐50 #6877 III/3 TP‐49 #7186 III/2 TP‐53 #7079 Total IV/8 TP‐51 #7241 IV/7 TP‐50 #6909 IV/6 TP‐51#7240 postmould IV/6 TP‐51 #7132 IV/5 TP‐53 #7235 IV/4 TP‐53#7059undetermined(Efe. IV/4 TP‐53 #7058undetermined(Efe. IV/4 TP‐53 #7077 III/n/a III/IV/6 TP‐49 #6925 III/6 TP‐50 #7237 III/5 TP‐51 #7122 III/5 TP‐50 #7086 III/ TP‐51#7125 Total fragment wt. TP‐53 #7104 ‐50 #6893 ct. undetermined(Efe.

Layer/Level, Unit, Sample No., Feature (Efe. 79) 78) 83) 83)

3 [0.1] indeterminate bark . .200 . .10.96 0.01 3.9 0.01 0.02 0.1 31151123 1 [0.02] indeterminate parenchymous tissue 00]2 1 [0.01] indeterminate unknown 07]1 11 [0.78] 1 [0.12] 2 [0.23] 2 2 [0.02] 1 [0.07] 5 [1.17] 1 [0.01] 2 6 [0.42] 3 [0.39] 2 [0.07] 1 [0.1] [0.06] [0.07] [0.09] 5 [0.1] 5 [0.2] indeterminate angiosperm

00]1 [0.01] indeterminate monocot 00]5 1 [0.04] 18 1 [0.02] 2 02]15 1 5 [0.26] 02]9 4 [0.24] 3 2 2 1 00]5 [0.07] 00]9 [0.05] 01]25 [0.14] 00]23 [0.03] 00]29 [0.09] 00]18 [0.01] 36 [0.01] unidentifiable 12 [1.52] 11 [1.23] 10 17 12 12 13 28 11 1 [1.93] 7 [0.27] 5 [0.27] 7 6 25 [4.3] [2.71] [0.42] [0.11] [0.63] [0.53] [0.42] 1 [0.1] [1.04] [2.65] [2.06] [0.28] [0.94] [2.49] [1.86] [2.63] [2.48] [1.38] [1.06] [0.49] [2.14] 38.34

[2.4] Frag ct. [wt.] 370 Late, Early, South Late, Late, Total Late, Early, South Early, South Early, South Early, South Early, South Middle, Middle, Middle, Middle, Middle, Middle, Middle, Late, Late, Total Late, Late, Late, Late, Late, Late, Late, Early, South Late, Late, Late, Middle, Late, Late, Late, Late, Late, Late, Middle, Late, Late, Late, Late, Late, Late, Late, Middle, Middle, Middle, South Northcoast,Sfe.8 TP‐24I/3#1054 Northcoast,Sfe.68 TP‐7I/4 #110.1oven(Efe. Northcoast,Sfe.68 TP‐7I/4 #110oven(Efe.

Northcoast,Sfe.68 TP‐7I/3 #105hearth(Efe. Northcoast,Sfe.68 TP‐7I/4#76hearth(Efe. South Northcoast,Sfe.8 TP‐25II/8#1081 Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, Northmid South Northcoast,Sfe.8 TP‐24IIb/‐ #1080 Northcoast,Sfe.8 TP‐24I/6#1049 Northstream, Northmid Northmid South South Northstream, South Northstream, Northmid Northstream, wt. fragment North stream, Northlowland, Northcoast,Sfe.68 SP‐6 ‐/‐ #9999poss. Northcoast,Sfe.68 SP‐6 ‐/‐ #61firefeature(Efe. Northcoast,Sfe.68 TP‐7/SP6 ‐‐ #62oven(Efe. Northlowland, Northlowland, Northlowland, Northlowland, South South South inland, inland, inland, inland, coast,Sfe. inland, inland, coast,Sfe. coast,Sfe. coast,Sfe. coast,Sfe. inland, coast,Sfe. coast,Sfe. lowland, lowland, ct. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. 2 TP‐46 ‐/‐ #6240undetermined(Efe. 2 TP‐45VII/‐ #6351 2 TP‐44IVorVI/ 2 TP‐44VI?/ 2 TP‐44 ‐/9 #6249thin Sfe. 242 TP‐38I/‐ #4237oven(Efe. 242 TP‐38I/‐ #4245oven(Efe. 245 TP‐39ext.I/2#4421 245 TP‐39ext.I/3#439 Sfe. Sfe. Sfe.11 TP‐23II/4#1033 Sfe. Sfe. Sfe. Sfe. 254 TP‐40II/5#425 254 TP‐40II/6#441 254 TP‐40II/6#425 13 C14I/‐ #4201oven(Efe. 13 TP‐36I/2#4066 13 TP‐36II/3#4074 13 A10 13 B11 ‐/3#4525 13 C10 ‐/3#4606 16 TP‐17II/6#971 13 C14I/‐ #4218oven(Efe. 16 TP‐17II/7#976 16 TP‐17III/9#987 16 TP‐17I/4#947 16 TP‐17 2 TP‐44 ‐/4or6#6356oven(Efe. 2 TP‐46IV/VI/ 24 TP‐22II/4#1008 24 TP‐22II/5#1012 32 TP‐31I/‐ #3030 24 TP‐22II/6#1021 232 SP‐1I/3#3085 232 SP‐1I/II/4#308 336 TP‐42III/6#611 336 TP‐42III?/6#6123hearth(Efe. 11 TP‐23II/‐ #1079 336 TP‐42 ‐ /‐ #6094oven(Efe. ‐/3 #4171 II/5 #961 ‐ #6359 thin ‐ #6368 ‐ #6366 Temporal phase, Structure (Sfe.), oven(Efe. oven(Efe.

6 6 4 Sample No., Unit, Layer/Level, Feature undetermined(Efe. 0 ccoallens 7 ccoallens 2 postmould C‐01) D‐01) 26) 26) 3) 3) 5) D‐01) 29) 29) (Efe. 7) 6) (Efe. 52) 37) (Efe. 59) 61) 61) 40) 28) 8) 1 1 2 3

[0.35] 01]1 [0.11] [0.47] [0.53] .5 .1 .7021.0 .515953100 .1 .3 .9 .7 .5 .7 .9 1.053 1.291 0.072 2.959 0.078 6.596 0.237 1.612 0.04 5.361 1.589 3.25 13.101 0.2 0.57 3.815 1.457 Aleurites moluccana endocarp 0798 51844 941 17 14 4 39 3 48 4 8 1 55 4 3 82 9 7 30 7 01]1 2 [0.19] 1 [0.09] 2 8 1 [0.11] 1 1 1 4 1 [0.16] 2 2 [0.3] 2 1 [0.02] [0.01]

[0.21] [0.16] [0.22] [0.05] [0.06] [1.93]

Artocarpus altilis [0.1] [0.2] 3 [0.08] 2 [0.32] 1 [0.12] [0.05] cf. Artocarpus altilis

9 [0.2] Calophyllum inophyllum 54 [6.63] 4 [1.08] 7 [2.19] 4 3 5 3 1 [0.14] 1

10]2 [0.26] [1.09] [1.37] [0.24]

[0.1] Celtis pacifica 1 [0.34] [2.91] Cocos nucifera 1 [0.83] 3 07]5 [0.76] cf. Cocos nucifera 19 [0.74] 1 [0.28] 7 1 5 1 1 [0.12] 5 2 3 [0.28] 4 1 [0.5] [0.52]

08]1 [0.77] [0.82] [0.12] [0.57] [0.25] [0.33] [0.06] Cocos nucifera endocarp

[0.04] cf. Cocos nucifera endocarp 1 [0.12] 1 2 4

[0.52] [0.49] [0.48] Coprosma spp. 1 [0.14] 3

[0.1] cf. Coprosma spp. 47 [6.5] 1

[0.1] Cordia subcordata 00]15 2 [0.03] 1 [0.05] cf. Cordia subcordata 6 [0.74] 10 [0.4] 2 3 [0.24] 1 [0.3] 2 [0.69] [0.49]

[0.1] Hibiscus spp. 4 [0.07] Inocarpus fagifer 1 3 [0.38] 1 [0.03] 1 [0.08] 2 5 1 [0.05] [0.22] [0.44]

Maytenus crenata [0.1] 17

[1.05] Metrosideros collina Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Middle, Late, Late, Late, Middle, Middle, Middle, Middle, Middle, Late, Total Early, South Early, South Early, South Middle, Middle, Total Early, South Early, South Early, South Early, South Late, Middle, Late, Middle, Middle, Middle, Northmid Northcoast,Sfe.8 TP‐25II/8#1081 Northmid Northcoast,Sfe.8 TP‐24IIb/‐ #1080 Northmid Northcoast,Sfe.8 TP‐24I/6#1049 Northmid Northstream, Northstream, Northstream, Northcoast,Sfe.8 TP‐24I/3#1054 Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, South South Northcoast,Sfe.68 TP‐7I/4 #110.1oven(Efe. Northcoast,Sfe.68 TP‐7I/4 #110oven(Efe. Northcoast,Sfe.68 TP‐7I/3 #105hearth(Efe. Northstream, Northstream, South Northcoast,Sfe.68 TP‐7I/4#76hearth(Efe. South South fragment wt. Northlowland, Northlowland, Northcoast,Sfe.68 SP‐6 ‐/‐ #9999poss. Northcoast,Sfe.68 SP‐6 ‐/‐ #61firefeature(Efe. Northcoast,Sfe.68 TP‐7/SP6 ‐‐ #62oven(Efe. Northlowland, Northlowland, South South South South North lowland, coast,Sfe. inland, inland, inland, inland, inland, inland, coast,Sfe. inland, coast,Sfe. coast,Sfe. coast,Sfe. coast,Sfe. coast,Sfe. lowland, lowland, ct. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe.13 A10 Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. 2 TP‐46 ‐/‐ #6240undetermined(Efe. 2 TP‐44 ‐/9 #6249thin 2 TP‐45VII/‐ #6351 2 TP‐44IVorVI/ 2 TP‐44VI?/ 242 TP‐38I/‐ #4237oven(Efe. 242 TP‐38I/‐ #4245oven(Efe. 245 TP‐39ext.I/2#4421 245 TP‐39ext.I/3#439 Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. 254 TP‐40II/6#441 254 TP‐40II/6#425 254 13 B11 ‐/3#4525 13 C10 ‐/3#4606 13 C14I/‐ #4201oven(Efe. 13 TP‐36I/2#4066 13 TP‐36II/3#4074 13 C14I/‐ #4218oven(Efe. 16 TP‐17II/5#961 16 TP‐17II/6#971 16 TP‐17II/7#976 16 TP‐17III/9#987 16 TP‐17I/4#947 2 TP‐44 ‐/4or6#6356oven(Efe. 2 TP‐46IV/VI/ 24 TP‐22II/4#1008 24 TP‐22II/5#1012 24 TP‐22II/6#1021 32 TP‐31I/‐ #3030 232 SP‐1I/3#3085 232 SP‐1I/II/4#308 336 TP‐42III?/6#6123hearth(Efe. 11 TP‐23II/‐ #1079 11 TP‐23II/4#1033 336 TP‐42 ‐ /‐ #6094oven(Efe. 336 TP‐42III/6#611 TP‐40II/5 #425 ‐/3 #4171 ‐ #6359 thin ‐ #6368 ‐ #6366 Temporal phase, Structure (Sfe.), oven(Efe. oven(Efe.

.8 .3 .0 .654900400 0.01 1 0.09 0.034 5.459 0.26 0.101 5.234 0.185 cf. Morinda citrifolia 811613359131512 1 5 1 3 1 509 313 1 6 1 1 28 2 2 3 2 1 [0.13] 2 [5.1] Pandanus tectorius

2 [0.1] cf. Pandanus tectorius 2 [0.26] Pandanus tectorius drupe 24 [5.29] 1 [0.07] 1 [0.01] 2 [0.09] Phyllanthus cf. marchionicus 1 [0.03] cf. Phyllanthus marchionicus 1 [0.09] Premna serratifolia

cf. Psydrax odorata 6 [1] 1 [0.01] cf. Santalum insulare 79 [19.67] 3[02]140[24.59] 73 [10.22] 9[.1 4[0.08] 19 [0.51] 4[.3 8[2.19] 14 [2.33] 0[.6 6[0.16] 20 [0.86] 3[.8 7[0.68] 13 [1.98] 2[.5 1[0.05] 12 [0.85] 2 [1.06] 1 [0.09] 29 [0.88] 5 [0.16] 5 [0.28] 11[1.3] 4[0.28] 7 [1.71] 1 [0.08] 13[1.98] 2 [0.27] 1 [0.07] 19 [3.9] 2[2.29] 1 [0.62] 01]5[0.38] 2 [0.15] 8 [1.06] 2 [0.16] 15 [17.06] 7 [2.97] 07]6[0.79] 1[0.11] 21[4.89] 1 [0.78] 9[2.56] 9 [0.95] 2[0.11] 3 [0.26] 2 [0.14] 51.841 04 2 [0.12] 3 [0.4] 2 [0.3] Sapindus saponaria 80 [10.66] 82 [53.72] 34 [3.81] 3.0 .101900 .4 .1 0.019 1.715 0.447 0.01 0.119 0.11 131.008 10 [0.5] 5 [0.39] 1 [0.11] 1 [0.27] 4 [0.41] 5 [0.52] 2 [0.23] Thespesia populnea

cf. Thespesia populnea 3 [0.12] Xylosma suaveolens 1 [0.01] cf. Xylosma suaveolens 01]1[1.72] 1 [0.14] 4 [0.31] unknown02

unknown09 2 [0.02] unknown11 Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Late, Middle, Late, Late, Late, Middle, Middle, Middle, Middle, Middle, Late, Total Early, South Early, South Early, South Middle, Middle, Total Early, South Early, South Early, South Early, South Late, Middle, Late, Middle, Middle, Middle, Northmid Northcoast,Sfe.8 TP‐25II/8#1081 Northmid Northcoast,Sfe.8 TP‐24IIb/‐ #1080 Northmid Northcoast,Sfe.8 TP‐24I/6#1049 Northmid Northstream, Northstream, Northstream, Northcoast,Sfe.8 TP‐24I/3#1054 Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, Northstream, South South Northcoast,Sfe.68 TP‐7I/4 #110.1oven(Efe. Northcoast,Sfe.68 TP‐7I/4 #110oven(Efe. Northcoast,Sfe.68 TP‐7I/3 #105hearth(Efe. Northstream, Northstream, South Northcoast,Sfe.68 TP‐7I/4#76hearth(Efe. South South fragment wt. Northlowland, Northlowland, Northcoast,Sfe.68 SP‐6 ‐/‐ #9999poss. Northcoast,Sfe.68 SP‐6 ‐/‐ #61firefeature(Efe. Northcoast,Sfe.68 TP‐7/SP6 ‐‐ #62oven(Efe. Northlowland, Northlowland, South South South South North lowland, coast,Sfe. inland, inland, inland, inland, inland, inland, coast,Sfe. inland, coast,Sfe. coast,Sfe. coast,Sfe. coast,Sfe. coast,Sfe. lowland, lowland, ct. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. ‐valley, Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe.13 A10 Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. 2 TP‐46 ‐/‐ #6240undetermined(Efe. 2 TP‐44 ‐/9 #6249thin 2 TP‐45VII/‐ #6351 2 TP‐44IVorVI/ 2 TP‐44VI?/ 242 TP‐38I/‐ #4237oven(Efe. 242 TP‐38I/‐ #4245oven(Efe. 245 TP‐39ext.I/2#4421 245 TP‐39ext.I/3#439 Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. Sfe. 254 TP‐40II/6#441 254 TP‐40II/6#425 254 13 B11 ‐/3#4525 13 C10 ‐/3#4606 13 C14I/‐ #4201oven(Efe. 13 TP‐36I/2#4066 13 TP‐36II/3#4074 13 C14I/‐ #4218oven(Efe. 16 TP‐17II/5#961 16 TP‐17II/6#971 16 TP‐17II/7#976 16 TP‐17III/9#987 16 TP‐17I/4#947 2 TP‐44 ‐/4or6#6356oven(Efe. 2 TP‐46IV/VI/ 24 TP‐22II/4#1008 24 TP‐22II/5#1012 24 TP‐22II/6#1021 32 TP‐31I/‐ #3030 232 SP‐1I/3#3085 232 SP‐1I/II/4#308 336 TP‐42III?/6#6123hearth(Efe. 11 TP‐23II/‐ #1079 11 TP‐23II/4#1033 336 TP‐42 ‐ /‐ #6094oven(Efe. 336 TP‐42III/6#611 TP‐40II/5 #425 ‐/3 #4171 ‐ #6359 thin ‐ #6368 ‐ #6366 Temporal phase, Structure (Sfe.), oven(Efe. oven(Efe.

.4 .305202 .0 .40010271.6 03827.343 20.308 17.269 0.257 0.051 0.14 0.606 0.24 0.502 0.13 0.445 unknown13 41613115143133124 1 [0.13] unknown17 3

[0.5] unknown21 3 [0] 1 [0.24] unknown29 1 2 [0.43] [0.18] indeterminate bark 1 [0.14] indeterminate endocarp 1 [0.05] indeterminate parenchymous tissue 1 2 [0.16] 1 [0.02] 1 [0.03] [0.05] indeterminate unknown 23 [0.64] 41]1 10 [4.13] 14]1 14 [1.42] 01]1 10 [0.12] 5 [0.53] 2 5 [1.67] 2 [0.26] 1 [0.04] 1 [0.07] 6 [1.01] 2 [0.01] 4 [0.51] 3 [0.28] 01]101 3 [0.17] 2 1 [0.23] 2 [0.07] 4 [0.63] 13]1 9 [1.38] 5 05]4 3 [0.53] 2 1 [0.07] 4 01]1 7 [0.17] 6 6 [0.31] [1.15] [0.17] 2 [0.5] [0.29] 08]2 [0.82] 01]4 [0.11] indeterminate angiosperm 1 [0.03] 1 [0.05] 1 [0.01] 6 [0.41] 1 5 04]15 [0.42] 00]7 [0.06] 00]5 [0.06] [0.27] [0.04] 1.]7 [16.9] 00]10 [0.07] [0.01] [0.84] [0.36] [0.28]

[0.5] indeterminate monocot 25 16 8 11 3 01]37 2 [0.13] 68 7 [0.72] 1 1.8 32 [15.88] 01]19 1 [0.19] 6 03]91 [0.37] 05]137 [0.59] 27]26 [2.78] [0.41] 08]282 [0.83] 07]15 [0.74] 11]63 [1.17] 09]133 [0.95] 03]33 [0.32] 02]24 [0.25] 20]65 [2.02] unidentifiable 82 [53.72] 101 23 10 10 [1.37] 22 11 24 37 26 14 19 12 [2.82] 37 61 28 5 [1.38] 6 [3.62] 4 9 8 9 8 [21.13] 5 [0.52] 2 [0.22] 5 2 [0.85] 4 3 6 6 [39.22] 5 6 [16.69] [14.73] [19.23] [48.28] 308.21 3 [0.32] [0.88] [0.61] [2.98] [2.06] [5.65] [1.12] [4.25] [1.67] [2.38] [1.91] [0.28] [6.29] [3.95] [7.87] [1.93] [0.03] [0.41] [0.85] [1.28] [4.09] [1.92] [0.38] [0.43] [1.61] [2.02]

[1.38] [2.49] [1.34] [2.93] [1.28] [2.64] [0.2]

1638 Frag ct. [wt.] [15] Appendix B: Wood anatomy reference

Anatomical descriptions of the woods sectioned for the present study are described in the following section. As the reference library contains well over 300 specimens, only a select number of taxa have been described.

350

Anacardiaceae Anacardium occidentale L. Reference Coll. Num.: 301 (Collector No. 51) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Druses present

Radial Section Ray cell composition type: Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial groups 2-5. Most vessels 75-115 µm diameter, 15-18 per mm. Elements ~300 µm length. 12-15 Rays per mm. Rays all uniseriate. Druses often present in vessels with tyloses.

351

Anacardiaceae Mangifera indica L. Reference Coll. Num.: 171 (Collector No. TPVL1-1) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - winged-aliform Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present Styloids and / or elongate crystals present

Radial Section Ray cell composition type: All ray cells procumbent Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: <= 900 µm

352

Notes: Vessels solitary or in groups of 2-3, 100-155 µm diameter. Mostly widely spaced, but one area dense where vessels occur in a wide range of sizes and axial parenchyma is more frequent. Distribution ranges 9 to 26 vessels per mm (as per previous statement). Some axial parenchyma strands may be more than 4 cells long. Vessel elements difficult to examine because of tyloses, but about 200-250 µm length.10-12 rays per mm. Crystals in some ray square and upright cells. Fibre pits minute but common.

Anacardiaceae Rhus taitensis Guill. Reference Coll. Num.: 60 (Collector No. 14) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Large - >= 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

353

Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels soliatary or in radial chains of 2-3; Two diameter classes: 50-80 µm and 100-150 µm diameter (the larger are most common); 7-9 per mm. Vessel elements 390-460 µm length. About 5 rays per mm; mostly 2-3 seriate though occasionally uniseriate. Crystals occur in upright ray cells. Fibre pits minute and infrequent.

Anacardiaceae Spondias cytherea Sonn. Reference Coll. Num.: 318 (Collector No. 68) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - aliform (shape unspecified) Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Large - >= 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: <= 4 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? X Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

354

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in groups of 2. Diameters (135)-167-215 µm; circular to oval in shape; 7-8 per mm; elements often short, 250- 400 µm. Axial parenchyma is not common, occurs in a thin sheath around vessels or very occasionally in an aliform shape. 4-5 Rays per mm; (3)-4-10-(12) cells wide and sometimes over 1mm tall; most have 1 row of uprights at end or occasionally 2. Fibres are infrequently septate. Crystals occasional in ray cells. Some type of deposit in vessels and some ray cells. Simple fibre pits, frequent in radial walls.

Annonaceae Annona muricata L. Reference Coll. Num.: 303 (Collector No. 53) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma reticulate Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

355

Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary or in radial or tangential chains of 2-3; diameters (74)-100-130 µm, very occasionally smaller, usually very rounded in shape; elements 265-389 µm length; 15-20 per mm. Some type of deposit seen in vessels and axial parenchyma, but only visible in RS of this specimen. Usually 5 rays per mm, (3) 4-7 cells wide, 600-1000 µm; widths rather consistent. The shorter rays are vaguely storied.

Annonaceae Cananga odorata (Lam.) Hook. f. & Thomson Reference Coll. Num.: 289 (Collector No. 44) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: >= 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

356

Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells procumbent Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length: 900-1600 µm

Notes: Vessels solitary or in radial groups of 2-(3), distributed evenly; many (150) 200-235 µm diameter; approx. 5 per mm. Elements 310-384 µm length. Rays (2-3) 4-6 cells wide, almost exactly 4 per mm.

Apocynaceae Alstonia costata (G.Forst.) R.Br. Reference Coll. Num.: No specimen in collection. Details from Sidiyasa (1998). Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: >= 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings?

357

Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Minute - <= 4 µm Vessel pits vestured? X Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Description for heavy Alstonia (Sidiyasa 1998): Vessels 72-149 per mm, often chained, 41-59 µm diameter. Vessel elements 620-905 µm length. 10-13 rays per mm, thin, 400-660 µm height. Intervessel pits minute, 3-4 µm. Source notes rays more or less similar all species: 1-3(4) wide with 1-15 cell ends, can often be over 1mm tall. Crystals, if occur, found in axial parenchyma and ray cells. Fibre pits very few. Never tyloses.

Apocynaceae Alyxia stellata (J.R.Forst. & G. Forst.) Roem.&Schult. Reference Coll. Num.: 19 (Collector No. 94) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: >= 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

358

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length: 900-1600 µm

Notes: Vessels 35-50 µm diameter; 90-120 vessels per mm; elements 500-600 µm long. Noted some intervessel pits in a reticulate pattern. Ray of 2 types: widths uniseriate and 3-4 wide with long uniseriate ends. Multiseriate portions under 300 µm, but can be up to 1000 when ends included. 12-15 rays per mm. Fibre pits are very common and frequent.

Apocynaceae Cerbera manghas L. Reference Coll. Num.: 267 (Collector No. 22) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? X Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma?

359

Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Multiseriate portion wide as uniseriate portions Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length: 900-1600 µm

Notes: Vessels solitary and radially grouped 2 - 6 or more; diameters around 100 µm with some slightly smaller or larger; 15-20 per mm; elements 540-570 µm length. Axial parenchyma not seen in all sections, but when present occurs in short or very long bands 1-2-(3) cells wide. Not always associated with the vessels. About 7 Rays per mm, most are uniseriate but occasionally 2 cells wide. Vessel-ray pits in upright cells are gash-like.

Apocynaceae Cerbera manghas L. Reference Coll. Num.: 161 (Collector No. TVCF-2) Plant Part: woody stem

360

Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? X Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Multiseriate portion wide as uniseriate portions Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length: 900-1600 µm

Notes: Vessels (100)-120-150 µm diameter, often chained 4+ radially. Axial parenchyma in occasional bands 1-3 cells wide and occasionally vasicentric in sleeves of 1 or 2 cells; also occasionally in dispersed chainettes. Rays short and mostly uniseriate, often with slender biseriate portions.

Apocynaceae Ochrosia oppositifolium (Lam.) K.Schum. Reference Coll. Num.: 49 (Collector No. 109) Plant Part: woody stem

361

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? X Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels often isolated and sometimes in groups of 2 (3); diameters range 75-90 µm, though occasionally vessels are as small as 35-40 µm; about 40 per mm, sometimes more frequent. Vessel elements 550-765 (900) µm length. Most axial

362

parenchyma strands 5-8 cells long. 5-7 Rays per mm; 2-3 seriate with long uniseriate ends or (less frequently) uniseriate; crystals frequent in ray parenchyma.

Apocynaceae Rauvolvia nukuhivensis (Fosberg & Sachet) Reference Coll. Num.: 377 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

363

Notes: Vessels mostly solitary and of 2 size classes, larger 100-125 µm diameter, smaller 80-90 µm; about 10-12 per mm but in some areas much more frequent. In some areas vessel deposits are common and completely fill the cell; elements 540-650 µm length. Rays mostly 2 cells wide with long uniseriate ends; interconnected rays; about 10 per mm. Prismatic crystals common in chambered axial parenchyma. Fibre pits bordered, common on both radial and tangential walls.

Aquifoliaceae Ilex anomala Hook. & Arn. Reference Coll. Num.: 41 (Collector No.93) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel clusters common Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? X Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? X Included phloem?

Tangential Longitudinal Section Vessel element length: >= 800 µm Vessel perforation plates: Scalariform perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits scalariform Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: <= 4 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells?

364

Notes: Frequent vessels often in radial or tangential groups of 2-3, or at times 6+; diameters range from (45)-50-60 µm; 50 - 70 per mm. Tyloses are only occasional. Vessel elements very long, 880-900 µm. Intervessel pits occur both scalariform and opposite, and bars are very frequent. Low rays are mostly upright cells, high rays are procumbent with 2-4 rows of square and uprights. Fibre pits are abundant. Vessel-ray pits occur in two types.

Araliaceae Meryta sp. Reference Coll. Num.: 47 (Collector No. 86) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? X Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Scalariform perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits scalariform Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions:

365

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels isolated or in chains or clusters of 2-(3); many 55-85 µm diamter, though some can be smaller; vessel elements have wide range from 450-550 µm to 900 µm length. Tyloses are very infrequent. Intervessel pits scalariform with bars widely spaced, 3-4 each. About 5 rays per mm; 3-5 seriate and though some cells in each are procumbent, the types are mixed sometimes with many uprights at ends.

Araliaceae Polyscias verrucosa (Seem.) Lowry & G.M.Plunkett Reference Coll. Num.: 59 (Collector Nos. 193, 91, 53) Plant Part: woody stem

Transverse Section Growth rings? X Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel clusters common Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

366

Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in clusters or radial chains of 2-4-(5-6); diameters range widely from 45-120 µm and are evenly distributed in size and grouping, though a marginal band of smaller-diameter vessels is apparent in this section. 26-35 vessels per mm; elements 383-425 µm, sometimes longer. Intervessel pits are mixed scalariform and opposite in the same vessel. 7-8 Rays per mm; of 2 sizes: usually very short uniseriate rays composed of upright cells, and taller 2-3 seriate rays with 1 or more upright or square at ends.

Arecaceae Cocos nucifera L. Reference Coll. Num.: 91 (Collector No. none) Plant Part: woody stem Notes: Vascular bundles ~650-800 µm diameter. Usually 2 or 3 large vessels per bundle, occasionally a few additional smaller; (1)-2- (3) per mm. Surrounded by parenchyma cells of even size and distribution.

Arecaceae Cocos nucifera L. Reference Coll. Num.: 286 (Collector No. 41) Plant Part: petiole Notes: Vascular bundles of 3 sizes: largest ~ 800 µm diameter (most frequent) with other clustered vessels 50-150 µm in diameter. Medium size bundles 350-380 µm, with several clustered vessels of 80-90 um. Smallest are 50-100 µm and have no vessels. Surrounded by parenchymous tissue. 1 large, 2 medium, and 3-5 small bundles per mm.

Arecaceae Cocos nucifera L. Reference Coll. Num.: 287 (Collector No. 42) Plant Part: inflorescence Notes: Frequent (2-3 per mm) and closely spaces vascular bundles 500- 600 µm diameter, with 2 or 3 clustered vessels 80-90 µm diameter, sometimes occurring with a number of smaller vessels. 367

Surrounded by parenchymous tissues composed of cells that are of a regular size and shape. Some bundles are compressed.

Asparagaceae Cordyline fruticosa (L.) A. Chev. Reference Coll. Num.: 94 (Collector No. none) Plant Part: woody stem Notes: Secondary vascular bundles composed of 10-12 cells, ~250 mm wide, oval to round and very frequent in TS, arranged in rows with 1-2 parenchyma rows between each.

Asteraceae Fitchia taitensis Nadeaud Reference Coll. Num.: 35 (Collector No. 332) Plant Part: woody stem very small diameter wood

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells?

368

Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary or in radial chains of 2-3, diameters 65-89 µm (smaller towards the pith); 330-350 µm length. 4 rays per mm, most 3-4 cells wide.

Bignoniaceae Crescentia cujete L. Reference Coll. Num.: 317 (Collector No. 67) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - confluent Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays and / or axial elements irregularly storied Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells procumbent Body ray cells procumbent with one row of upright and / or square marginal cells

369

Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Most vessels 75-100 µm in diameter; occasionally chained radially 2-(3); 15-20 per mm; elements 100-185 µm in length. About 10 rays per mm; most bi-seriate; most rays composed of procumbent cells, but some have one square at end. Vessel ray pits similar to intervessel pits. Axial parenchyma and vessel elements somewhat storied. Fibre pits infrequent but appear bordered.

Boraginaceae Cordia subcordata Lam. Reference Coll. Num.: 305 (Collector No.55) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - lozenge-aliform Paratracheal axial parenchyma - winged-aliform Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present Crystal sand present

370

Radial Section Ray cell composition type: Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels single or clustered 2-3; 90-155 µm in diameter though some smaller (~60um); 10-15 per mm, elements 250-300 µm length. Rays (1-2)-3-4 cells wide with a mixture of cell types throughout, brown crystal sand deposits distinctive and easy to see in all sections.

Calophyllaceae Calophyllum inophyllum L. Reference Coll. Num.: 246 (Collector No. 1) Plant Part: woody stem upper trunk wood

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessels in diagonal and / or radial pattern Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: Vessels of two distinct diameter classes Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: <= 5 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals?

371

Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessels of two sizes: under 150 µm and almost uniformly 190- 200 µm; element lengths 300-445 µm. Ray cells mostly uniseriate, though occasionally 2 cells wide in places; composed of procumbent cells bordered by one row of square cells (sometimes 3-4 rows); occasionally square cells occur in the middle of a ray. Ray parenchyma frequently filled with some type of deposit.

Cannabaceae Celtis pacifica Planch. Reference Coll. Num.: 89 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

372

Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial chains of 2-5-(6); diameters (40)-50- 75 µm; sizes in chains tend to taper or larger vessels cap chains of smaller vessels; 10-16-(25) per mm; elements 261-278 µm length, sometimes much shorter. Axial parenchyma bands 3-4 cells wide, connecting the vessels. About 8 rays per mm; (2)-3-4 cells wide, interconnected at times by uniseriate portions on one or both ends. Crystals in many upright ray cells.

Casuarinaceae Casuarina equisetifolia Forst. Reference Coll. Num.: 282 (Collector No. 37) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type:

373

Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Scalariform perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells procumbent Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length: 900-1600 µm

Notes: Vessels 160-180 µm diameter and in places, very frequent; elements 200-300 µm in length. Perforation plates often appear simple. Rays 200-330 µm high; most 2-(3) cells wide with uniseriate ends. Distinctive homogenous rays - all cells procumbent. Fibre pits very common.

Celastraceae Maytenus crenata (Forst. f.) Lobr.-Callen Reference Coll. Num.: 172 (Collector No. ANVL1-2) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: >= 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? X

374

Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Minute - <= 4 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels mostly solitary; (28)-33-45 µm diameter, very rounded and evenly distributed; some with deposits; about 100 or more per mm; elements 350-500 µm length. Bands of thin walled fibres noted, about 4-5 cells wide and very regularly spaced resembling parenchyma banding. About 15 rays per mm, most uniseriate and very long, with occasional bi-seriate portions. Fibres have thick walls that are heavily pitted and thin walls with simple pits.

Combretaceae Terminalia catappa L. Reference Coll. Num.: 316 (Collector No. 66) Plant Part: woody stem

375

Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - winged-aliform Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Druses present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: May be vaguely ring-porous. Vessels solitary or radial by 2-3(4); diameters range from 85-125-135 µm; 4-6 per mm, more when vessels are smaller; elements 340-350 µm though often much shorter. About 5 rays per mm, most 2-3 cells wide. Occasional large druses in enlarged axial parenchyma cells.

Combretaceae Terminalia glabrata Forst. f. Reference Coll. Num.: 73 (Collector No. 116) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping:

376

Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: <= 5 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - winged-aliform Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Druses present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial or cluster of 2-4-(5); diameters range 108-163um; 4-5-(6-9) per mm; element lengths 273-329, sometimes shorter. About 7 rays per mm; most 2-3 cells wide. Fibres very occasionally septate. Druses occasional in enlarged axial parenchyma.

Cunoniaceae Weinmannia marquesana G. Forst. Reference Coll. Num.: 325 (Collector No.75)

Plant Part: woody stem

377

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? X Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Scalariform perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits scalariform Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels mostly solitary or occasionally by 2; about 68-80 per mm. Elements from 570-600-810 µm length; sometimes have tyloses and/or some type of deposit in vessels. Intervessel pits opposite to sclariform. Most ray parenchyma filled with a dark substance. Chambered axial parenchma frequently have crystals. 7-8 Rays per mm, of 2 sizes: uniseriate and 2-3 seriate with ends 2-3-or much longer.

378

Euphorbiaceae Aleurites moluccana Willd. Reference Coll. Num.: 310 (Collector No.60) Plant Part: woody stem

Transverse Section Growth rings?

Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm >= 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: <= 5 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: >= 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Large - >= 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

379

Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial chains (or divided clusters, occasionally of tapering sizes) of 2-4; diameters (126)-180-215 µm, sometimes smaller but mostly large and often very round in shape; 2-3(4-5) per mm; elements (475)-866-970 µm length. About 9 rays per mm; most 2-(3) cells wide with ends of varying lengths; some interconnected by uniseriate portions. Crystals occasional in (sometimes chambered) axial parenchyma and some ray cells. Intervessel pits are extremely large, some almost 20 µm.

Fabaceae Adenanthera pavonina L. Reference Coll. Num.: 260 (Collector No.15) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - lozenge-aliform Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate

380

Mineral inclusions: Druses present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessel diameter (80)=110-145 µm; length (350) 425-550 µm. Intervessel pits ~5 µm. Fibres 515-600 µm, some filled with prismatic crystals. Ray height 370-475-(770) µm. Occasionally some type of deposit in vessels.

Fabaceae Caesalpinia bonduc (L.) Reference Coll. Num.: 259 (Collector No.14) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - winged-aliform Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

381

Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present Crystals of other shapes (mostly small)

Radial Section Ray cell composition type: Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Cross section quite variable. Bands of axial parenchyma are sporadic but distinct when they do occur, paratracheal parenchyma occurs in many different types. Vessels (85)-100- 150 µm diameter, about half the time chained 2-6 radially (chained vessels usually under 80 µm diameter); elements 225- 365 µm length. Rays typically 1-3 cells wide and range from 370 to well over 1700 µm long. Axial parenchyma sometimes have numerous (8 or more) chambers each filled with a small oblong crystal. Ray cells are composed of rather large procumbent cells with occasional square cells (1-many) throughout, occasionally containing rhomboid crystals.

Fabaceae Erythrina variegata L. Reference Coll. Num.: 311 (Collector No.61) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: >= 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: <= 5 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm

382

Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? X Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Axial parenchyma and / or vessel elements storied Rays per mm: <= 4 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Notes: Vessels 230-260 µm diameter, mostly isolated or joined very closely (2); very widely spaced having less than 3-4 per mm; elements 200 µm or less in length. Fibres occasionally have prismatic crystals.

Fabaceae Indigofera suffruticosa Mill. Reference Coll. Num.: 270 (Collector No.25) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - unilateral paratracheal Fibre wall thickness: Fibres thin- to thick-walled

383

Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? X Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Notes: Vessels solitary or in radial groups of 2, 3 (4, 5); diameters in two size classes, 25-30 µm and 40-80 µm; elements very short, 80-115 µm. Some type of vessel deposits, but infrequent. About 10 rays per mm; high rays 250-280 µm length, low rays smaller.

Fabaceae Inocarpus fagifer Fosb. Reference Coll. Num.: 168 (Collector No. HTF1-3) Plant Part: woody stem

384

Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: All rays storied Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Notes: Vessels widely spaced and often solitary, sometimes joined by 2-3 radially or in clusters; diameters (65)-89-140 µm, 5-7 per mm; elements 194-248 µm length. Paratracheal bands of axial parenchyma almost all 3-4 cells wide, plus small sheaths of parenchyma surround many vessels. Rays almost exclusively uinseriate but very occasionally bi-seriate. Crystals occasional in strands of chambered parenchyma cells.

Fabaceae Sophora tetraptera J. Mill. Reference Coll. Num.: 65 (Collector No. 195) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous

385

Vessel arrangement: Vessels in dendritic pattern Vessel grouping: Vessel clusters common Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? X Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Axial parenchyma and / or vessel elements storied Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels often clustered and arranged in a dendric pattern; diameters from 33-55um, those in clusters can be smaller; 40-50 per mm; elements 138-172 µm length. About 5 rays per mm; of 2 sizes: 1-2 cells wide and short, and 5-6 cells wide and 325-525 µm tall. Sheath cells only occasional. Crystals occasional in parenchyma. Axial parenchyma is storied.

Lecythidaceae Barringtonia asiatica Kurtz Reference Coll. Num.: 307 (Collector No. 57) Plant Part: woody stem

386

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - confluent Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Large - >= 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays and / or axial elements irregularly storied Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Multiseriate portion wide as uniseriate portions Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: >= 1600 µm

Notes: Vessels 100-150 µm diameter, isolated or in radial groupings by 2 or 3; elements 460-470 µm length. Intervessel pits 12-14 µm. Large rays are 1750-2000 in height. Crystals not frequent.

387

Loganiaceae Fagraea berteroana A. Gray Reference Coll. Num.: 162 (Collector No. TVCF-3) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Large - >= 10 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: <= 900 µm

Notes: Vessel diameter 110-125, occasionally smaller; sometimes joined by 2; 5-10 per mm; elements 400-500 µm length. Rays are occluded by deposits making them very hard to examine.

388

Around 12-14 rays per mm, can be over 1mm in length. Fibres occasionally, very subtly, with septa.

Loranthaceae Decaisnina forsteriana (Schult. & Schult. f.) Barlow Reference Coll. Num.: 328 (Collector No. 78) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel clusters common Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? X Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

389

Notes: Vessel diameters 70-90 µm, single or in small clusters 2-(3); 15- 20 per mm; elements 130-170 µm or shorter. About 5-7 rays per mm; some are 3-4 seriate and tall, some are uni- or bi-seriate and shorter. Sheath cells are only occasional. Fibre pits are common.

Lythraceae Pemphis acidula J.R.Forst. & G.Forst. Reference Coll. Num.: 52 (Collector No. 215) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? X Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening?

390

Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial chains of 2-5; 73-85 (123) µm in diameter; elements around 280 µm length, some shorter; 24-28 per mm. Most vessels blocked with tyloses and some dark deposit fills many cells. Rays mostly uniseriate, only occasionally 2-3 wide.

Malvaceae Commersonia bartramia Merr. Reference Coll. Num.: 27 (Collector No. 27) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square

391

Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or frequently in chains (sometimes clusters) of 2-4(5); diameters 87-122um, sometimes smaller; about 15-20 per mm; elements 323-425 µm length. Rays of 2 sizes: uniseriate with occasional bi-seriate portions, and 3-4 seriate with long ends. 12-15 per mm. Crystals occasional to frequent in ray cells, sometimes multiple per cell.

Malvaceae Hibiscus tiliaceus L. Reference Coll. Num.: 292 (Collector No. 47) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? X Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Low rays storied, high rays non-storied Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm

392

Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Multiseriate portion wide as uniseriate portions Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions: Druses present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessels often large, and often clustered or chained 3-5; sometimes cells are very irregular and/or quite small (under 50 µm); most 95-190 µm diameter; 10-15 per mm; elements 390- 450 µm length. Low rays are 3-4 cells wide with one row uprights at end, high rays are of very mixed composition and can extend to over 1mm in length; about 5 per mm; storying notable; some high ray cells are enlarged with druses. Parenchyma strands usually 4 cells long. Fibre pits mainly on the radial walls.

Malvaceae Thespesia populnea Soland. Reference Coll. Num.: 283 (Collector No. 38) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

393

Vessel pits vestured? Axial parenchyma cell type: Fusiform parenchyma cells Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Low rays storied, high rays non-storied Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Rays with procumbent, square and upright cells mixed throughout the ray Ray tile cells? X Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in radial groups or clusters of 2-4-(5); some clusters have large cells; as few as as 7-9 per mm and up to 11 per mm; diameters 89-120 µm, sometimes smaller and occasionally a bit larger; elements 232-322 µm length. Axial parenchyma in short chainettes 1 cell wide; in TLS, fusiform and 2 cells wide. Intervessel pits ~7 µm. 6-8 rays per mm; mostly bi- seriate and storied, but occasional longer rays are 3-4 wide, having a mix of cell types throughout. Crystals can be frequent in square and upright ray cells. Fibre pits small but frequent, on radial and tangential walls.

Moraceae Artocarpus altilis (Park.) Fosb. Reference Coll. Num.: 277 (Collector No. 32) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: >= 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - lozenge-aliform Paratracheal axial parenchyma - confluent

394

Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? X Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Large vessels 220-260 µm diameter; often solitary, but also in clusters and chains of 2-3-4; elements 400-450 µm length. Mostly 5 rays per mm; of 2 sizes, uniseriate and 2-4 wide, larger rays 375-600 µm in length.

Moraceae Artocarpus altilis (Park.) Fosb. Reference Coll. Num.: 276 (Collector No. 31) Plant Part: woody stem small diameter twig Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits?

395

Axial parenchyma arrangements: Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - lozenge-aliform Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: <= 4 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? X Fibre types: Non-Septate Mineral inclusions:

Notes: Twig with pith. Vessels 116-150 µm diameter. Intervessel pits ~9 µm. Rays small, 1-2-(3) wide, and short.

Moraceae Broussonetia papyrifera L. Reference Coll. Num.: 83 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? X Vessel porosity: ring porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X

396

Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Axial canals in short tangential lines Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? X Fibre types: Non-Septate Mineral inclusions: Silica bodies present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: <= 900 µm

Notes: Vessel diameter average 100-150 µm, but large earlywood are 250+ um; smaller groups with diameters 40-46 µm; about 8 per mm. Axial parenchyma not frequent. Majority of rays are 3-5 cells wide. Small radial vessel clusters, surrounded by axial parenchyma, distinguish this taxon from Artocarpus.

Moraceae Ficus prolixa G. Forst. Reference Coll. Num.: 306 (Collector No. 56) Plant Part: woody stem

397

Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? X Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessels (150)-225-260 µm diameter, widely dispersed by about 3 per mm; element length 420 µm. Axial parenchyma bands 3-6 cells wide and frequent. About 5-7 rays per mm; most 2-3 cells wide, but some are about 6. Sheath cells only occasional. Prismatic crystals very occasional, in ray cells.

Myrsinaceae Myrsine hosakae St. John Reference Coll. Num.: 48 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: <= 50 µm

398

Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? X Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Minute - <= 4 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Fibres storied Rays per mm: <= 4 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Larger rays commonly 10-seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Notes: Vessels solitary or in radial chains of 2-3-4-(5); diameters (42)- 50-(61) µm, around 60 per mm; elements 230-270 µm length. Low rays 2-6 cells wide, high rays can be over 20. Prismatic crystals only occasional in ray parenchyma.

Myrtaceae Eugenia reinwardtiana (Blume) D. C. Reference Coll. Num.: 97 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more)

399

Vessel diameter: <= 50 µm Vessels of two distinct diameter classes Vessels of two distinct diameter classes, wood not ring-porous X Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary and small diameter, of 2 sizes: about 15 µm and beteween 25-36 µm; 80-100 per mm; elements 383-440 µm length. Rays 1-2-(3) seriate with uniseriate ends, about 10-12 per mm; uniseriate rays all square or upright cells, multiseriate rays proc with several rows of uprights. Crystals occasional in chambered axial parenchyma.

Myrtaceae Metrosideros collina (J.R. & G. Forst.) A. Gray Reference Coll. Num.: 262 (Collector No. 17)

400

Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary and of two sizes: 40-50 µm and 18-20 µm diameter (smaller are called vascular tracheids); about 80 per

401

mm; elements 375-450 µm length but occasionally shorter. 12- 15 rays per mm, comprised of short bi-seriate sections with often very long uniseriate ends. Crystals are frequent in chambered axial parenchyma cells, and axial parenchyma strands are long.

Myrtaceae Syzygium malaccense (L.) Merr. & Perry Reference Coll. Num.: 321 (Collector No. 71) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present Crystal sand present

402

Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels 125-170 µm diameter, often chained radially 2-4; ~15 per mm; length 340-375 µm. Tyloses very abundant. 6-9 rays per mm; deposits within ray parenchyma are very frequent. Axial parenchyma bands mostly 4-6 cells wide. Rhomboid crystals frequent in chambered axial parenchyma. Vessel-ray pits scalariform-gash like, but very difficult to see. Fibre pits not frequent, and mostly on radial walls.

Nyctaginaceae Pisonia grandis R. Br. Reference Coll. Num.: 122 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel clusters common Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem? X

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Fusiform parenchyma cells Vascular / vasicentric tracheids present? If rays storied, structure type: Axial parenchyma and / or vessel elements storied Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Raphides present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells

403

Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary or in radial chains (sometimes clusters) of 2-5; widely spaced groups occurring directly above included phloem; 50-92 µm diameter, 6-9 per mm; elements 180-190 µm length, sometimes much shorter. Rays 4-5 per mm. Most rays 1-2-(3) cells wide and very short.

Pandanaceae Pandanus sp. Reference Coll. Num.: 12 (Collector No. 89-3) Notes: Vascular bundles 600-900 µm diameter; usually 2 larger vessels in each about 75 µm diameter, with a few other smaller in diameter (20-25 µm).

Phyllanthaceae Phyllanthus marchionicus (F. Br.) Reference Coll. Num.: 251 (Collector No. 6) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Ray width: Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Mineral inclusions: Styloids and / or elongate crystals present Crystals of other shapes (mostly small)

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessel diameters between 70-100 µm; 580-850 µm length with many on the shorter end of this range. Elongate and irregular crystals occur in some ray cells, the former occasionally in enlarged cells. Enlarged, rounded vessel-ray pits.

Piperaceae Piper methysticum J. R. Forst. & G. Forst. Reference Coll. Num.: 143 (Collector No. none) Plant Part: Woody stem Notes: Vessels solitary or in clusters of 2-3; diameters (64)-80-86 µm; 8-9 per mm; elements 230-370 µm length. Rays very wide. Axial elements all storied.

Pittosporaceae Pittosporum taitense Putt. Reference Coll. Num.: 53 (Collector No. 81) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel clusters common Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates 405

Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? X If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Larger rays commonly 10-seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or (more often) in clusters; very frequent; diameters range from (32-45)-77-85 µm; 80-90 per mm; elements 350-480 µm length. 4-6 rays per mm, of 2 types: 1-2 seriate and very short (not as frequent) and 5-6 cells wide and taller with upright cells at the ends. Large crystals in some enlarged upright ray cells. Fibre pits very small, but abundant in RS.

Rhamnaceae Alphitonia zizyphoides F. Br. Reference Coll. Num.: 18 (Collector No. 31) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands

406

Fibre wall thickness: Fibres very thin-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Styloids and / or elongate crystals present

Notes: Vessels (85)-100-150 µm diameter, sometimes in lines or clusters 3-5; 10-15 vessels per mm; elements 300-550 µm length. Marginal parenchyma banding very noticeable. Most rays 2-seriate, some are uniseriate and quite small; ray height (130) 300-350 µm; 6 - 8 per mm.

Rhizophoraceae Crossostylis biflora Forst. Reference Coll. Num.: 29 (Collector No. 70) Plant Part: woody stem

407

Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: >= 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: <= 4 / mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Larger rays commonly 10-seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: >= 1600 µm

Notes: Vessels frequent, diameters (70)-110-130 µm, often in radial or tangential groups of 2-3; approx 25-30 per mm; elements 500- 750-900 µm in length. Larger rays 20+ seriate, smaller rays 4-6 seriate. Vessel-ray pits abundant. Intervessel pits can also be somewhat opposite in appearance - distribution and size not even. Ray parenchyma contain some type of deposit (not crystalline).

Rubiaceae Coprosma taitensis A. Gray Reference Coll. Num.: 28 (Collector No. 83) Plant Part: woody stem

Transverse Section Growth rings?

408

Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: <= 4 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length: 900-1600 µm

Notes: Vessels 65-85 µm diameter; often solitary but sometimes chained radially 2-3-(4); just over 40 per mm; vessel length ~580 µm. Axial parenchyma often diffuse or in short lines, evenly scattered throughout. Rays are widely spaced and long (sometimes over 1mm); occasionally shorter. Fibre pits are distinctly bordered.

409

Rubiaceae Cyclophyllum barbatum (G.Forst.) N.Hallé&J.Florence Reference Coll. Num.: 330 (Collector No. 80) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Multiseriate portion wide as uniseriate portions Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Notes: Vessels solitary, 40-50 µm diameter; elements 500-600 µm length; 70-80 per mm. Most rays 2 cells wide, some uniseriate, 10-12 per mm.

410

Rubiaceae Gardenia taitensis DC. Reference Coll. Num.: 315 (Collector No. 65) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels almost all solitary but occasionally by 2; 45-55 µm diameter; elements 430-560 µm length, sometimes shorter. Most rays biseriate with long uniseriate ends. Vessel ray pits are horizontal / scalariform. Fibre pits very common.

411

Rubiaceae Guettarda speciosa L. Reference Coll. Num.: 6 (Collector No. 89-9) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Multiseriate portion wide as uniseriate portions Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length: 900-1600 µm

412

Notes: Vessels (50) 80-100 in diameter; elements 550-690 µm length. Intervessel pits between 6-8 µm in diameter. No parenchyma. Most rays 3-(4) seriate, narrow to 1-2 large cells that are almost same width; often interconnected and can be quite long.

Rubiaceae Ixora fragrans A. Gray Reference Coll. Num.: 114 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: <= 50 µm 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

413

Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels almost exclusively solitary; diameters (44)-55-79 µm; most filled with deposits, some with brown silaceous deposits; around 50-55 per mm; elements 350-480 µm length. Axial parenchyma bands about 3 cells wide, of varying lengths very short to very long, and generally not associated with vessels. 13- 18 rays per mm; 1-3 cells wide, some with uniseriate ends.

Rubiaceae Morinda citrifolia L. Reference Coll. Num.: 117 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? X Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Raphides present

414

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary or in chains or clusters of 2-3-(4); of two sizes: larger 110-125 µm, smaller 50-55 µm diameter; 15-20 per mm; elements 585-595 µm in length but can also be quite short (400 µm or less). 6-9 rays per mm; 1-3(4) cells wide. Raphides are 55-92 µm long. Most rays of procumbent cells with 1-2-3-(4+) rows of square or upright at ends.

Rubiaceae Neonauclea forsteri Merr. Reference Coll. Num.: 50 (Collector No. 13, 292) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: >= 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Rays exclusively uniseriate

415

Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells procumbent Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Vessels often solitary, occasionally by 2; 100-150 µm diameter, about 12 per mm; elements 980-1100 µm length. Rays uniseriate with cells square or upright, sometimes 2 cells wide with very long uniseriate ends.

Rubiaceae Psychotria tahitensis Drake Reference Coll. Num.: 144 (Collector No. 71) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

416

Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Mineral inclusions: Raphides present

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels 1-5-(6-7) in radial chains; 37-53 µm diameter; 80-100 per mm; elements 479-540 µm length. Vessel deposits occasionally block the lumens. About 10 rays per mm, of 2 sizes: 1-2 seriate having upright or square cells, and 2-3 seriate with with procumbent cells in centre and 2-4 rows of uprights at end. Upright ray cells occasionally enlarged and filled with raphides.

Rubiaceae Psydrax odorata (G.Forst.) A.C.Sm. & S.P.Darwin Reference Coll. Num.: 269 (Collector No. 24) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

417

Vessel pits vestured? X Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Eight (5-8) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels mostly solitary, occasionally in radial groups of 2; diameters (40)-70-85 µm; 40-45 vessels per mm; elements 600- 700 length though sometimes shorter or longer. Rays mostly bi- seriate, some uniseriate, 10-12 per mm; most having procumbent cells with many square and uprights at ends. Uniseriate rays are composed of all square and upright cells.

Rutaceae Melicope revoluta J.Florence Reference Coll. Num.: 46 (Collector No. 194) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse-in-aggregates Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

418

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Silica bodies present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels solitary, often chained radially 2-3-(4,5,6); 70-100 µm diameter; elements 250-327 µm length. Most rays 2-3-seriate and very long, but occasionally narrower and shorter. Silica in the parenchyma. Vessel-ray pits appear to be similar to intervessel in shape and size. Intervessel pits 4-5 µm.

Salicaceae Xylosma suaveolens G. Forst. Reference Coll. Num.: 250 (Collector No. 5) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type:

419

Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Minute - <= 4 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Notes: Vessels solitary or often in radial chains of 3-5-(6); 30-35 µm in diameter; about 80-100 per mm; elements 540-700 µm long. 20- 23 rays per mm; many bi-seriate with very long ends and sometimes interconnected, others are uniseriate. Crystals very frequent in ray parenchyma. Vessel-ray pits may be 2 types.

Santalaceae Santalum insulare Bert. ex A.DC. Reference Coll. Num.: 319 (Collector No. 69) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates

420

Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Vessels almost exclusively solitary; diameters (75) 90-110-140 (148) µm; about 40 per mm; elements 195-320 µm long, somtimes quite a bit shorter. About 6 rays per mm; mostly bi- seriate, ranging from very short (50 µm or so) to about 300 µm tall. Crystals in chambered axial parenchyma. Most vessel-ray pits are similar to intervessel, but occasionally enlarged and circular in upright ray cells.

Sapindaceae Allophylus marquesensis F. Br. Reference Coll. Num.: 17 (Collector No. 181) Plant Part: woody stem

Transverse Section Growth rings? X Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes,

421

wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? X Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 100 µm - 300 µm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Raphides present

Notes: Vessels 60 to 90 µm diameter, sometimes chained 3-7 radially; 10-20 per mm; element length 165-275 µm. Occasionally vessels are occluded by some type of resin deposit. Rays 1 - 2 seriate only,12-16 per mm; heights 150-200 µm. Most ray parenchyma filled with some type of residue. Notable for bands of alternating thick and thin fibres, which vaguely resemble parenchyma bands in TS.

Sapindaceae Dodonaea viscosa Jacq. Reference Coll. Num.: No tropical specimen, description from InsideWood (2004). Plant Part: woody stem

Transverse Section Growth rings?

422

Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: <= 50 µm 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? X Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Paratracheal axial parenchyma - scanty paratracheal Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma bands more than three cells wide Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? X Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells procumbent Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Prismatic crystals in long chains of 30 or more in chambered axial parenchyma; 2 seriate rays relatively common. Parenchyma rarely banded.

423

Sapindaceae Pometia pinnata J.R.Forst. & G.Forst. Reference Coll. Num.: 268 (Collector No. 23) Plant Part: woody stem

Transverse Section Growth rings? X Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - aliform (shape unspecified) Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Minute - <= 4 µm Vessel pits vestured? Axial parenchyma cell type: Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: All ray cells upright and / or square Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Vessels mostly solitary, sometimes in groups of 2; diameters (70-80) 100-116 µm; 10-14 vessels per mm; tyloses occasional; elements 360-385 µm length. About 10 rays per mm, all

424

uniseriate and most are short; cells all square with one (or more) rows of upright at ends. Prismatic crystals occur in both types of ray cells.

Sapindaceae Sapindus saponaria L. Reference Coll. Num.: 249 (Collector No. 4) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - winged-aliform Paratracheal axial parenchyma - confluent Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? X Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Two cells per parenchyma strand Four (3-4) cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

425

Fibre length:

Notes: Vessels solitary or in radial or clustered groups of 2-4; diameters (60) 85-100 µm. 7-15 per mm; elements 200-235 µm length, sometimes shorter. Helical thickenings sometimes so thick it is difficult to examine intervessel pits. 7-9 rays per mm, most 2-3 seriate, heights vary widely with some over 500um. Crystals frequent in axial parenchyma.

Sapotaceae Planchonella sp. Reference Coll. Num.: No specimen. Description of features of Hawaiian P. sandwicense, Brown (1922) and Planchonella sp., Lamoureux (1985).

Transverse Section Growth rings? Vessel porosity: Wood diffuse-porous Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: 50 - 100 µm 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions: Silica bodies present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells All ray cells upright and / or square Ray tile cells?

426

Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessels mostly chained radially 4-6+; diameters mostly under 120 µm; element lengths ~300-450 µm. Axial parenchyma bands narrow, only a few cells wide. Rays very small; 1-2 cells wide; 15 or more per mm; up to 40 cells high; composed of procumbent (and occasionally square cells) with more than one row of upright cells; silica deposits in ray cells. Fibres very thick-walled. Intervessel pits 4-5 µm and not vestured. Vessel- ray pits enlarged and rounded or slash-like.

Sapotaceae Sideroxylon st.-johnianum (H. J. Lam & Meeuse) Smedmark & Anderb. Reference Coll. Num.: 119 (Collector No. none) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessels in diagonal and / or radial pattern Vessel grouping: Vessel diameter: <= 50 µm 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? X Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

427

Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Vessel-ray pits with much reduced borders to apparently simple: pits horizontal (scalariform, gash-like) to vertical (palisade) Vessel-ray pits restricted to marginal rows Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Many vessels solitary, some grouped by 2; 30-35 per mm; diameters range from (38)-50-80 µm; elements 350-478 µm length; approx. one-third are filled with a deposit. Axial parenchyma bands 3-4 cells wide, regularly spaced and not necessarily assoc. with vessels. 8-11 rays per mm, most bi- seriate with long ends; some interconnected. Crystals frequent in axial parenchyma.

Sapotaceae Unknown13 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels per square millimetre 40 - 100 vessels per square millimetre Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

428

Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm >= 12 /mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels frequent; solitary or in radial groups of 2-3-4 about half of the time; diameters 80-120 µm; elements short. Bands of axial parenchyma 2-3-(4) cells wide and spaced at regular intervals. Rays are very small, 2-(3) seriate in centre with many uniseriate cells at ends; 8-10-(12+) per mm. Fibre walls very thick. Crystals frequent, probably in chambered axial parenchyma strands. Tyloses very frequent in vessels; could be scerified but similar-looking structures may have been formed by other deposits. Intervessel pits about 5 µm. In charcoal, thick fibres often have completely closed lumens and certain features appear to have fused making examination difficult at times.

Solanaceae Solanum viride G. Forst. ex Spreng. Reference Coll. Num.: 323 (Collector No. 73) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma?

429

Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 300 µm - 500 µm Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels solitary or in radial chains of 4-6(+); diameters range from 40-67 µm; about 50-60 vessels per mm; elements 285-350 µm length, occasionally very short. About 10 rays per mm; most uniseriate, very occasionally bi-seriate and only for short portion of the ray. Ray structure in this specimen is very simple (mostly uniseriate) and pith is present. Note, specimen was a small shrub.

Thymelaeaceae Wikstroemia coriacea Seem. Reference Coll. Num.: 329 (Collector No. 79) Plant Part: woody stem

430

Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - winged-aliform Paratracheal axial parenchyma - confluent Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? X Axial parenchyma cell type: Two cells per parenchyma strand Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Vessels solitary but often in clusters of 2-6; diameters 46-60 µm, sometimes a bit smaller when clusters are large; about 15-20 per mm; elements 235-260 µm length, sometimes a bit shorter or longer. 5-7 rays per mm; most 2-4 cells wide.

Urticaceae Pipturus argenteus (G.Forst.) Wedd. Reference Coll. Num.: 14 (Collector No. 89-12) Plant Part: woody stem

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes,

431

wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? X Intervessel pit size: Alternate: Medium - 7 - 10 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) > 1 mm Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Ray tile cells? Vessel-ray pit type: Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length: 900-1600 µm

Notes: Vessels solitary or occasional in radial groups of 2; (100)-145- 175 µm diameter, but very occasional small vessels about 60 µm; 5-6 per mm; elements 220-290 µm length. About 5 rays per mm; (2)- 3-6 cells wide. Some type of deposit in ray cells.

Verbenaceae Premna serratifolia L. Reference Coll. Num.: 278 (Collector No. 33) Plant Part: woody stem

Transverse Section Growth rings? X Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes,

432

wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? X Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions: Acicular crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibres with simple to minutely bordered pits Fibre length:

Notes: Distinct growth rings, some areas diffuse and others vaguely semi-ring porous. Vessels solitary or in radial groups of 2-4, diameters can be quite small (30-48um) but most are 50-100 µm; about 25 per mm, but sometimes many more and sometimes less. Tyloses are frequent and abundant, obscuring vessel elements. Rays 6-7 per mm; 2-3-(4) cells wide, occasionally interconnected; many cells have numerous acicular crystals that are easy to see in RS. Most fibres are septate; pits are simple and somewhat frequent on the radial walls.

Unkown11 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous

433

Vessel arrangement: Vessel grouping: Vessel diameter: Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma bands more than three cells wide Fibre wall thickness: Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels of average diameter, singly or in groups of 2. Very wide bands of axial parenchyma. Most rays uniseriate, sometimes bi- seriate.

Unkown15 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm

434

Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels mostly in chains or clusters 2-3(4-7); widely spaced in places, some groups of small vessels; elements 150-200 µm length. Rays 2-3-4- cells wide, can sometimes be long; no uniseriate rays; >20 per mm; most procumbent with upright ends; occasionally square cells mid-ray. Parenchyma noted, but faint - could be paratracheal.

Unkown17 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous

435

Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 20 - 40 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Fibre wall thickness: Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Non-Septate Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels solitary or by 2-4; 50-100 µm diameter. Rays 2-4-(5-6) wide, procumbent cells very small and uniform in size; 1-2 upright cells at ends. Fibres septate. Vessel-ray pit mostly blown out, but may be rounded. Intervessel pits 5-6 µm.

Unkown18 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous

436

Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) 500 µm - 1 mm Ray width: Rays exclusively uniseriate Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Body ray cells procumbent with over 4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels 1-2-3-(4-8) in chains and clusters, 50-87 µm diameter, 200-300 length. Alternate intervessel pits 6-7 µm. Rays of 2 sizes: uniseriate and 3-4 wide with long ends, frequent, of mixed cell types. Axial parenchyma associated with vessels. Fibre pits common on tangential and radial walls. Rays can be very long, but many not over 1mm.

Unkown19 Plant Part: Charred wood (archaeological material)

437

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Axial parenchyma absent or extremely rare Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres very thin-walled Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: 4-12 / mm Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Notes: Vessels solitary or by 2-3; elements around 600 µm length. Intervessel pits 5-6 µm. Fibres have pits on tangential walls.

438

Unkown21 Plant Part: Charred wood (archaeological material)

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) Ray width: Rays exclusively uniseriate Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Notes: Vessels solitary and radial chains (2)-3; 50-85 µm diameter, widely spaced. Intervessel pits 5 µm. Vessel-ray pits very rounded but similar size. Rays 3-4 cells wide with 2-4 uprights at ends, also uniseriate. Fibre pits very frequent on the radial walls.

439

Unkown22 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 100 - 200 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - aliform (shape unspecified) Paratracheal axial parenchyma - lozenge-aliform Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: 350 - 800 µm Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels large 130-160 µm diameter, single or by 2; about 12 per mm; elements about 500 µm long, sometimes shorter. Distinct lozenges of axial parenchyma. Fibre walls thick but not closed. Intervessel pits distinctly oval to rounded, 6-7 µm. Vessel-ray pits similar. Rays hard to examine closely but most 2-3 cells wide, procumbent with one row square at ends. Might be some type of deposit in ray cells.

440

Unkown24 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: <= 50 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 40 - 100 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Apotracheal axial parenchyma - diffuse-in-aggregates Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions: Styloids and / or elongate crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels very frequent, 50-70 µm wide, sometimes joined axially or radially. Parenchyma scattered singly and diffuse in chainettes. Sometimes filled with an oblong crystal. Rays mostly bi-seriate with ends of a few cells. Intervessel pits very small (4- 5 µm).

441

Unkown25 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: Vessel arrangement: Vessel grouping: Vessel diameter: Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Paratracheal axial parenchyma - aliform (shape unspecified) Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Fibres storied Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre pits common in both radial and tang. walls Fibre length:

Notes: Vessels often in radial groups or chains. Thick walled fibres, Axial parenchyma in close lozenges surrounding vessels and in

442

occasional bands few cells wide. Rays have 1 row square cells at end - very neat and tidy in radial section. Most notable for distinctve, very common fibre pits and storying of most elements in tangential longitudinal section. Intervessel pits 6 µm, arranged in diagonal channels. Vessel-ray pits similar.

Unkown26 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - vasicentric Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Banded parenchyma - Axial parenchyma in marginal or in seemingly marginal bands Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) Ray width: Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening?

443

Fibre pit type: Fibre length:

Notes: Vessels of many sizes 50-100 µm, solitary or joined in groups and clusters; elements often short (under 200 µm). Axial parenchyma occasionally assoc. with vessels and in sporadic bands few cells wide. Intervessel pits 6 µm but can be larger (to 7-8 µm), Vessel-ray pits similar. Rays of 2 sizes: uniseriate and 4-5-(6). TLS can be very dark and murky, or very shiny and clear. Ray cells sometimes have crumbly brown deposits. Radial section shiny, easy to examine ray cells, which are square or procumbent with 3-4-5 cells upright at ends.

Unkown27 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessels exclusively solitary (90% or more) Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Rays exclusively uniseriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: All ray cells upright and / or square Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells

444

Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Very shiny cross section, parenchyma vitrified, rays uniseriate. Crumby whitish deposits in some cells in tangential longitudinal section. Something coating most vessel walls. Intervessel pits ~5 µm. Fibre pits mainly on the radial walls, large and frequent. In radial section, vessel pits are 5-6 µm. Rays mostly uniseriate, or bi-seriate in places but very narrow. Rays of mixed procumbent and square cells. Vessel-ray pits may be rounded in some places.

Unkown28 Plant Part: Charred wood (archaeological material)

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: All ray cells procumbent

445

Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels 2-3, in chains and clusters, widely spaced, ~75 µm diameter, ~275 µm length. Axial parenchyma bands confluent and more than 3 cells wide in places; subtle in cross section. Rays frequent, tidy, most 2-seriate; lots of crystals; most composed of even, procumbent cells but occasionally square, and some ends have row of square cells. Neat and shiny in radial section. Vessel-ray pits similar, small.

Unkown29 Plant Part: Charred wood (archaeological material)

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types: Septate Non-Septate

446

Mineral inclusions:

Radial Section Ray cell composition type: Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels chained 2-5+, small diameter, very clean in cross section. Rays slender, frequent 2-(3) and some very long. Rays mainly have procumbent cells with a few square cells at ends. Fibres thick.

Unkown30 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Apotracheal axial parenchyma - diffuse Paratracheal axial parenchyma - scanty paratracheal Fibre wall thickness: Fibres thin- to thick-walled Fibres very thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Larger rays commonly 4 - to 10 seriate Ray sheath cells? Radial canals? Laticifer or tanniferous tubes? Fibre types:

447

Mineral inclusions: Prismatic crystals present

Radial Section Ray cell composition type: Body ray cells procumbent with one row of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Vessel-ray pits with much reduced borders to apparently simple: pits rounded or angular Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Most notable for vessels chained radially 2-4-(5). Cross section very smooth and shiny, axial parenchyma mostly diffuse, chambered with crystals in TLS. Rays 3-4-(5) wide and not very long, vaguely storied. Ray cells procumbent with one row square at ends, square cells have rounded pits. Intervessel pits ~5 µm.

Unkown31 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: diffuse porous Vessel arrangement: Vessel grouping: Vessel diameter: 50 - 100 µm Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: 5 - 20 vessels Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Paratracheal axial parenchyma - scanty paratracheal Banded parenchyma - Axial parenchyma in narrow bands or lines up to three cells wide Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: <= 350 µm Vessel perforation plates: Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits alternate Intervessel pits polygonal shape? Intervessel pit size: Alternate: Small - 4 - 7 µm Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? Ray height: (if 2 sizes, the larger rays) Ray width: Ray width 1 to 3 cells Ray sheath cells? X

448

Radial canals? Laticifer or tanniferous tubes? Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Vessel-ray pits with distinct borders; similar to intervessel pits in size and shape throughout the ray cell Fibre helical thickening? Fibre pit type: Fibre length:

Notes: Vessels 50-100 µm, widely dispersed, often by 2-3. Axial parenchyma faintly visible around vessels and irregularly winged, also in irregular radial lines.

Unkown32 Plant Part: Charred wood (archaeological material)

Transverse Section Growth rings? Vessel porosity: Vessel arrangement: Vessel grouping: Vessels in radial multiples of 4 or more common Vessel diameter: Vessels of two distinct diameter classes, wood not ring-porous Vessels per sq mm: Vessels with solitary outlines? Vessel tyloses? Other vessel deposits? Axial parenchyma arrangements: Fibre wall thickness: Fibres thin- to thick-walled Axial canal type: Fibre bands like parenchyma? Included phloem?

Tangential Longitudinal Section Vessel element length: Vessel perforation plates: Simple perforation plates Vessel helical thickenings? Intervessel pit arrangement: Intervessel pits polygonal shape? Intervessel pit size: Vessel pits vestured? Axial parenchyma cell type: Vascular / vasicentric tracheids present? If rays storied, structure type: Rays per mm: Rays of two sizes? X Ray height: (if 2 sizes, the larger rays) Ray width: Rays exclusively uniseriate Larger rays commonly 4 - to 10 seriate Ray sheath cells? X Radial canals? Laticifer or tanniferous tubes?

449

Fibre types: Mineral inclusions:

Radial Section Ray cell composition type: Body ray cells procumbent with mostly 2-4 rows of upright and / or square marginal cells Ray tile cells? Vessel-ray pit type: Fibre helical thickening? Fibre pit type: Fibres with distinctly bordered pits Fibre length:

Notes: Most notable for frequently chained vessels in groups of 4-6.

450