The Pennsylvania State University

The Graduate School

Intercollege Graduate Degree Program in Ecology

THE ECOLOGY, MANAGEMENT AND MARKETING OF NON-TIMBER

FOREST PRODUCTS IN THE ALTO RIO GUAMÁ INDIGENOUS RESERVE

(EASTERN BRAZILIAN AMAZON)

A Thesis in

Ecology

by

James Campbell Plowden

 2001 James Campbell Plowden

Submitted in Partial Fulfillment of the Requirements for the Degree of

Doctor of Philosophy

December 2001

We approve the thesis of James Campbell Plowden.

Date of Signature

Christopher F. Uhl Professor of Biology Chair of the Intercollege Graduate Degree Program in Ecology Thesis Advisor Chair of Committee

James C. Finley Associate Professor of Forest Resources

Roger Koide Professor of Horticulture Ecology

Stephen M. Smith Professor of Agricultural Economics

iii

ABSTRACT

Indigenous and other forest peoples in the Amazon region have used hundreds of non-timber forest products (NTFPs) for food, medicine, tools, construction and other purposes in their daily lives. As these communities shift from subsistence to more cash-based economies, they are trying to increase their harvest and marketing of some NTFPs as one way to generate extra income. The idea that NTFP harvests can meet these economic goals and reduce deforestation pressure by reducing logging and cash-crop agriculture is politically attractive, but this strategy’s feasibility remains in doubt because the production and market potential of many

NTFPs remains unknown. The need to obtain this sort of information is particularly important in indigenous reserves in the Brazilian Amazon since these comprise the country’s largest category of protected forests, and indigenous people are actively seeking economic means to support community development.

Between 1996 and 2000, I spent a total of two years in the Alto Rio Guamá Indigenous Reserve in eastern Pará state, Brazil studying production ecology of five regionally important NTFPs: oleoresin from copaiba trees (Copaifera spp.), resin from breu trees (Protium spp.), aerial roots from titica vines (Heteropsis spp.), latex from amapá trees (Parahancornia amapa and Couma guianensis), and seed oil from andiroba trees

(Carapa guianensis). My studies were primarily based in the Tembé Indian village of Tekohaw that is a one or two day journey by river and road from the regional center of commerce in Belém. The objectives of NTFP case studies on copaiba, breu and titica were to: 1) quantify the amount of marketable product that could be obtained per plant and per area of forest, 2) identify key factors that influenced variation in these amounts, 3) estimate the amount of harvest that would be possible on a repeated basis and the length of this harvest cycle,

4) estimate the amount of time a harvester needs to spend to find, harvest and process a unit of product and how much income they would earn for this time invested.

Results showed that a variety of factors contributed to the low economic returns that could be obtained by harvesting the NTFPs investigated in this project. Copaiba oleoresin had a relatively high unit price, but the availability of the product was severely restricted by the low density of the trees (<1 tree/ha), the limited percentage (15%) of trees that yielded more than 50 ml of oleoresin when drilled, and a sharp drop-off iv in oleoresin yield in trees that were retapped one year after the initial drilling. While the likelihood of obtaining oil from these trees was greatest in trees in the 55-65 cm Diameter at Breast Height (DBH) group, there was drop-off in oleoresin harvest in even larger trees because most were hollow and had apparently lost their capacity to store oleoresin in enclosed trunk cavities.

Breu trees had a density of 10 trees per ha, and about 40% of trees yielded some resin. The amount of resin increased with tree size and was related to the degree of attack by a bark-boring weevil whose larvae develop inside the resin lumps that form on the trunk. The larvae of a syrphid fly are also found in resin lumps from some types of breu trees, but it is unlikely these flies stimulated resin flow. Breu resin loses about 17% of its weight while drying, but some of this may be due to resin removal by stingless bees. The low market price for the resin makes it difficult for harvesters to regularly earn the Brazilian minimum wage (about $US

3.50 per day in 1999), but since it will take about four years for resin levels to rebound to initial harvest levels, it is a resource that if managed wisely can provide a periodic source of income.

Titica vines are very common in some forests and can be found on as many as several hundred host trees per ha. To meet the demands of wicker furniture makers, harvesters only take relatively thick mature aerial roots. While harvesters may gather 10 to 40 kg of roots per day, the removal of bad pieces, stripping of root cortex, and drying reduces the commercial product to 20% of the harvested weight. This processing of roots can take as much time as the harvesting. Among the products studied, titica offers the best short-term revenue, but since many harvested roots die or fail to regrow after being broken from the main plant, it may take decades for a titica population in an area to return to a harvestable condition.

These case studies showed that it was difficult to find NTFPs that met all the conditions (high density, high yield, low percentage of loss during processing, high price, quick resource renewal) to make their commercial collection worthwhile to Tembé harvesters on an ongoing basis. Better understanding NTFP production capabilities and the factors that influence them will help communities realistically assess how much each resource can contribute to their annual income and how harvesting can be managed to preserve each source population as a long-term resource. An increase in demand for consumer goods, however, will probably necessitate greater investments in intentional NTFP planting and other economic activities. v

TABLE OF CONTENTS

LIST OF FIGURES …………………………………………….……………………………………….. vii

LIST OF TABLES ……………………………………………………………………………………… viii

LIST OF APPENDICES ……………………………………………………………………………….. ix

ACKNOWLEDGEMENTS ……………………………………...... ………………...……………….. x

CHAPTER 1 INTRODUCTION ………………………………………………………………...... 1 INTRODUCTION ………………………………………………………………………………...... 2 SELECTION AND BACKGROUND OF STUDY AREA ………..………………………………… 5 SELECTION OF STUDY SPECIES …………………………………………………………………. 7 REFERENCES ...... ……………………………………………………………………………. 10

CHAPTER 2 THE ECOLOGY OF COPAIBA (COPAIFERA SPP.) OLEORESIN HARVEST IN THE AMAZON …….……………………….……………………………………………………… 13 ABSTRACT .…………………………………………………………………………………………. 14 INTRODUCTION: REVIEW OF COPAIBA PRODUCTION AND USE ….………………………. 16 RESEARCH OBJECTIVES ….………………………………………………………………………. 23 STUDY SITE ……………….………………………………………………………………………... 24 METHODS …………………….……………………………………………………………………... 25 RESULTS ……………………………….……………………………………………………………. 25 DISCUSSION ……………………………………….………………………………………………... 38 SUMMARY AND CONCLUSIONS ………………………….……………………………………... 55 REFERENCES ………………………………………………………………….……………………. 72

CHAPTER 3 THE HARVEST OF BREU RESIN FROM BURSERACEAE TREES IN THE EASTERN BRAZILIAN AMAZON AND THE ROLE OF WEEVILS AND BEES IN ITS FORMATION AND MARKETING ………………….……………………………………………….. 81

ABSTRACT ………………………………………….……………………………………………….. 82 INTRODUCTION …………………………………….………………………………………………. 83 STUDY AREAS ……………………………………………………………………………………… 85 METHODS …………………………………………………………………………………………… 86 RESULTS …………………………………………………………………………………………….. 90 DISCUSSION ………………………………………………………………………………………… 104 CONCLUSIONS ……………………………………………………………………………………… 118 REFERENCES ………………………………………………………………………………………... 142

vi

CHAPTER 4 THE ASSOCIATION OF AN ALIPUMILIO FLY (DIPTERA: SYRPHIDAE) WITH BURSERACEAE TREE RESINS IN THE EASTERN BRAZILIAN AMAZON …………. 153 ABSTRACT …………………………………………………………………………………………... 154 INTRODUCTION …………………………………………………………………………………….. 154 STUDY SITES ………………………………………………………………………………………... 156 MATERIAL AND METHODS ………………………………………………………………………. 157 RESULTS ……………………………………………………………………………………………... 158 DISCUSSION ………………………………………………………………………………………… 160 REFERENCES ……………………………………………………………………………………….. 164

CHAPTER 5 THE ECOLOGY, HARVEST AND MARKETING OF TITICA VINE ROOTS (HETEROPSIS SPP.: ARACEAE) IN THE EASTERN BRAZILIAN AMAZON ………………… 167 ABSTRACT …………………………………………………………………………………………... 168 INTRODUCTION …………………………………………………………………………………….. 169 STUDY AREA ………………………………………………………………………………………... 172 METHODS ……………………………………………………………………………………………. 173 RESULTS ……………………………………………………………………………………………... 178 DISCUSSION ………………………………………………………………………………………… 187 CONCLUSIONS: CHALLENGES FOR SUSTAINABLE TITICA HARVEST ...………………….. 198 REFERENCES ………………………………………………………………………………………... 216

CHAPTER 6 SUMMARY CONCLUSIONS ………………………………………………………… 220 FACTORS THAT AFFECT NON-TIMBER FOREST PRODUCT HARVEST AMOUNTS AND PROFITABILITY …………………………………………………………………………………. 221 RESEARCH IMPLICATIONS FOR THE ROLE OF NON-TIMBER FOREST PRODUCTS IN FOREST COMMUNITY DEVELOPMENT ……………………………………………………... 227 REFERENCES ………………………………………………………………………………………... 232

EPILOGUE ……………………………………………………………………………………………… 241

CHALLENGES AND LESSONS STUDYING NON-TIMBER FOREST PRODUCT ECOLOGY IN AN INDIGENOUS COMMUNITY …………………………………………………………… 242 REFERENCES ……………………………………………………………………………………….. 252

vii

LIST OF FIGURES

FIGURE 2.1 CROSS-SECTION ILLUSTRATION OF SOLID, OLEORESIN BEARING AND HOLLOW COPAIBA TREES ………………………………………………………………………. 67 FIGURE 2.2 PERCENT OF HOLLOW AND FIRE DAMAGED COPAIBA TREES AT TEKOHAW 68 FIGURE 2.3 HEARTWOOD AND HOLLOW PROPORTION OF TRUNK AREA IN COPAIBA TREES AT TEKOHAW SITE ………………………………………………………………………. 69 FIGURE 2.4 OLEORESIN HARVEST FROM COPAIBA TREES AT TEKOHAW BY SIZE CLASS 70 FIGURE 2.5 PATTERNS OF OLEORESIN FLOW FROM COPAIBA TREES AT TEKOHAW ….. 71 FIGURE 3.1 RELATIONSHIP OF DISTANCE FROM VILLAGE TO BREU RESIN HARVEST ... 134 FIGURE 3.2 DISTRIBUTION OF WEEVIL LARVAE HEAD CAPSULE WIDTHS FROM BREU RESIN LUMPS ……………………………………………………………………………………… 135 FIGURE 3.3 ILLUSTRATION OF BREU RESIN LUMP AND ASSOCIATED INSECTS ………... 136 FIGURE 3.4 ESTIMATED GROWTH RATE OF INDIVIDUAL BREU RESIN LUMPS AT TEKOHAW ………………………………………………………………………………………….. 137 FIGURE 3.5 BREU RESIN WEIGHT LOSS IN OPEN OUTSIDE DRYING AT TEKOHAW …….. 138 FIGURE 3.6 BREU RESIN WEIGHT LOSS WITH AND WITHOUT STINGLESS BEES ………... 139 FIGURE 3.7 ESTIMATED FIRST TIME RESIN HARVEST AT DIFFERENT DENSITIES OF BREU TREES ……………………………………………………………………………………….. 140 FIGURE 3.8 ESTIMATED DAILY HARVEST AND SALE VALUE OF DRIED RESIN AT DIFFERENT DENSITIES OF BREU TREES 141 FIGURE 5.1 ILLUSTRATION OF TITICA SEEDLING AND ADULT ON HOST TREE WITH PRINCIPAL PLANT PARTS ……………………………………………………………………….. 203 FIGURE 5.2 ILLUSTRATION OF TITICA ABSORBING ROOT AND PRINCIPAL PARTS …….. 204 FIGURE 5.3 DENSITY OF TITICA HOST TREES AND TITICA ROOTS IN LIGHTLY TO HEAVILY USED FOREST SITES IN THE ALTO RIO GUAMÁ INDIGENOUS RESERVE …... 205 FIGURE 5.4 NUMBER AND TYPE OF TITICA ROOTS PER HOST TREE AT MEDIUM AND LIGHT-USE FOREST SITES AT TEKOHAW AND CAJUEIRA ………………………………… 206 FIGURE 5.5 RELATIONSHIP BETWEEN HOST TREE DBH AND THE NUMBER OF TITICA ROOTS PER HOST TREE AT THE TEKOHAW MEDIUM-USE SITE ………………………….. 207 FIGURE 5.6 DEVELOPMENT TIME AND MORTALITY OF IMMATURE TITICA ROOTS AT THE TEKOHAW MEDIUM-USE SITE ……………………………………………………………. 208 FIGURE 5.7 CONDITION OF TITICA ROOTS 7 MONTHS AFTER CUTTING AT TEKOHAW MEDIUM-USE SITE ………………………………………………………………………………... 209 FIGURE 5.8 LOSS OF TITICA ROOT LENGTH AFTER NODE REMOVAL FROM 37 HOST TREES HARVESTED AT CAJUEIRA …………………………………………………………….. 210 FIGURE 5.9 WEIGHT LOSS DURING PROCESSING OF TITICA ROOTS ……………………… 211 FIGURE 5.10 WEIGHT LOSS OF TITICA ROOTS FROM DRYING AFTER CORTEX REMOVAL 212 FIGURE 5.11 LOCATION OF TITICA HOST TREES HARVESTED IN 4 DAYS BY A COMMERCIAL COLLECTOR AT CAJUEIRA …………………………………………………… 213 FIGURE 5.12 LOCATION AND ORDER OF TITICA HOST TREES HARVESTED BY A COMMERCIAL COLLECTOR AT CAJUEIRA IN ONE DAY …………………………………… 214 FIGURE 5.13 TITICA ROOT STEM STRIPPING RATE BY AGE OF PROCESSOR …………….. 215 viii

LIST OF TABLES

TABLE 2.1 COPAIBA TRUNK DIMENSIONS OF ALL DRILLED, OLEORESIN YIELDING AND HOLLOW TREES AT TEKOHAW...... 58 TABLE 2.2 CORRELATION BETWEEN COPAIBA TREE DIAMETER AND OTHER TRUNK DIMENSIONS AT TEKOHAW STUDY SITE...... 59 TABLE 2.3 HARVEST OF OLEORESIN FROM COPAIBA TREES AT TEKHOHAW STUDY SITE BY TREE TYPE...... 60 TABLE 2.4 HARVEST OF OLEORESIN FROM COPAIBA TREES AT TEKOHAW STUDY SITE BY SIZE CLASS...... 61 TABLE 2.5 CHANGES IN OLEORESIN HARVEST FROM COPAIBA TREES AT TEKOHAW STUDY SITE BY YEAR AND HARVESTING ORDER...... 62 TABLE 2.6 DRILLING TIMES IN COPAIBA TREES AT TEKOHAW...... 63 TABLE 2.7 OLEORESIN PRODUCTION FROM C. MULTIJUGA NEAR MANAUS, BRAZIL 64 TABLE 2.8 COPAIBA OLEORESIN HARVESTING MODEL BY AREA FOR LONG-TERM AND SHORT-TERM PROJECTIONS 65 TABLE 2.9 COPAIBA OLEORESIN HARVESTING MODEL FOR TWO COLLECTION METHODS AT FIRST-TIME HARVEST...... 66 TABLE 3.1 SUMMARY OF RESEARCH OBJECTIVES AND METHODS...... 121 TABLE 3.2 DENSITY OF BREU TREES IN DIFFERENT FOREST TYPES AT TEKOHAW...... 122 TABLE 3.3 BREU RESIN HARVEST PER TREE BY STATUS, TREE TYPE, LIVE TREE SIZE (DBH), FOREST TYPE, AND DISTANCE FROM TEKOHAW ...... 123 TABLE 3.4 DIMENSIONS OF WEEVILS FOUND IN BREU RESIN LUMPS ...... 124 TABLE 3.5 BREU RESIN LUMPS WEIGHT CORRELATION TO WEEVIL LARVAE AND BORE HOLE DIMENSIONS ...... 125 TABLE 3.6 WEEVIL LARVAE DIMENSION CORRELATION TO BORE HOLE NUMBER AND DIMENSIONS...... 126 TABLE 3.7 RESIN LUMP NUMBER AND WEIGHT FOUND ON BREU TREES DURING FOLLOW-UP HARVESTS...... 127 TABLE 3.8 CORRELATION OF TREE SIZE AND HARVEST HISTORY WITH FOLLOW-UP BREU RESIN HARVEST...... 128 TABLE 3.9 PROJECTION OF BREU RESIN YIELD AFTER INITIAL HARVEST ...... 129 TABLE 3.10 BREU RESIN HARVEST MODEL - PART 1. ESTIMATED AVERAGE RESIN PER TREE BY SIZE CLASS ...... 130 TABLE 3.11 BREU RESIN HARVEST MODEL - PART 2. DENSITY OF RESIN YIELDING TREES, RESIN HARVEST AND HARVEST PER VALUE PER HECTARE ...... 131 TABLE 3.12 DENSITY OF BURSERACEAE RESIN YIELDING SPECIES IN BRAZILIAN AMAZON INVENTORIES...... 132 TABLE 3.13 BREU RESIN HARVEST - PART 3. RESIN HARVESTING RATES AND VALUE BY AREA AND TIME ...... 133 TABLE 4.1 BODY DIMENSIONS OF IMMATURE ALIPUMILIO FLIES FROM BREU RESIN AT TEKOHAW 163 TABLE 5.1 SUMMARY OF TITICA STUDY OBJECTIVES, METHODS AND ANALYSIS...... 201 TABLE 6.1 PRODUCTION ECOLOGY SUMMARY FOR CASE STUDY NON-TIMBER FOREST PRODUCTS AND MANIOC...... 231

ix

LIST OF APPENDICES

APPENDIX 2-A COMMON NAMES FOR COPAIFERA TREES IN THE AMAZON REGION 80

APPENDIX 3-A SCIENTIFIC SPECIES AND COMMON NAMES OF RESIN YIELDING SPECIES OF NEOTROPICAL BURSERACEAE...... 150

APPENDIX 5-A GLOSSARY OF TERMS FOR TITICA VINES AND ROOT HARVEST ...... 219

APPENDIX 6-A PRODUCTION ECOLOGY STUDIES ON AMAPÁ LATEX AT TEKOHAW ...... 234

APPENDIX 6-B PRODUCTION ECOLOGY STUDIES ON ANDIROBA OIL AT TEKOHAW ...... 236

APPENDIX 6-C MANIOC AND FARINHA PRODUCTION STUDIES AT TEKOHAW ...... 238

x

ACKNOWLEDGEMENTS

I would like to thank the Rainforest Alliance (Kleinhans Fellowship), National Science Foundation

(Graduate Fellowship), Food, Conservation and Health Foundation, and the Hotchkiss School Students for

Environmental Action for their financial support of this project. I express my deep gratitude to my major adviser Dr. Christopher Uhl who provided expert guidance, moral and financial support to me and this project from beginning to end. I appreciate my committee members Dr. James Finley, Dr. Roger Koide and Dr. Steve

Smith for always being available to help as was my scientific counterpart in Brazil Prof. Francisco de Assis

Oliveira. Other advice, assistance, identification of specimens, and translations were graciously provided by

Dr. William Balée, Dr. Steve Beckerman, Dr. Scott Camazine, Dr. Jason Clay, Sérgio Antonio da Silva, Dr.

Douglas Daly, Dr. James Duke, Dr. B. A. Foote, Maria Graça-Zogbhi, Dr. Joe Kiesecker, Dr. Jean

Langenheim, Judith Lisansky, Dr. James Marden, Dr. Denny Moore, Dr. Charles O’Brien, Dr. William

Overall, Dr. Graham Rotheray, Rafael Rueda, Noêmia Pires de Sales, Dr. Jack Schultz, Dr. Patricia Shanley,

Dr. F. Christian Thompson, Dr. Giorgio Venturieri, Frank von Willert, Museu Goeldi staff botanists, and the entire staff of IMAZON. Tom Nagel did the life-life illustration of the insects associated with the breu resin lump shown in Chapter 3. Frederico Miranda de Oliveira, Francisco Potiguara, José Maria Galvão, Claudionor

Diaz, Jescelino Bessa and other staff members from FUNAI (National Indian Foundation of Brazil) provided crucial logistical support in Belém and the field. Jim Lockman and Edson Vidal from IMAZON kindly arranged access to their research sites at Mojú and Fazenda 7.

I am indebted to the community leaders Lourival Tembé, Veronica Tembé, Muxi Tembé, Waldeci

Tembé, Augustin Tembé, Petroni Ka’apor and the people of Tekohaw and other villages in the Alto Rio

Guamá and Alto Turiaçu Indigenous Reserves for their support of the project and my family. This project would not have been possible without the efforts of Livindo Tembé, Xarope Tembé, Afonso Tembé, Bubute

Tembé, Caroço Tembé, Emídio Tembé, Kapara’i Tembé, Manoel Tembé, Mukuim Tembé, Moreira Tembé,

Reginaldo Tembé, Pachik Tembé, Pedro Tembé, Preto Tembé, Raimundinho Tembé, Santana Tembé, Xina’i xi

Tembé, Xuxa Tembé, Zeca Tembé, Zezinho Tembé, Zu Tembé, Aloisio Munduruku, Caetano Munduruku,

Luçiano Munduruku, Lourival Munduruku, Geraldo Ka’apor, Makú Ka’apor, Xu’i Ka’apor, Godo, Bruce

Hoeft, Haley Mitchell, Delores Coan, Danielle Szabo, Silvana, Cleber, César, Damião Farias, Manoel Lopes

Farias and Elaelson who worked with me in the village and forest between 1996 and 2000.

Special thanks are due to my children Marissa and Luke Plowden for their adventurous spirit and adaptability during our year together in Brazil and for their incredible grace dealing with my absence for almost another whole year when I completed my field work. My parents Theodore and Suzanne Plowden gave my family and me unflagging love and support throughout several challenging years. I dedicate this dissertation to my wife Yuri Kusuda Plowden who has shared six years of the joys and heartaches of this journey and demonstrated more courage, patience, love, perseverance and creative support than one man could hope to see and receive from a partner in a lifetime.

CHAPTER 1

INTRODUCTION 2

INTRODUCTION

My connection to the idea that people living in tropical forests could make a decent living by sustainable harvesting and selling non-timber forest products (NTFPs) began in 1989 when I stayed in a small house in Rio Branco, Brazil formerly occupied by the rubber tapper and union activist Chico Mendes. Only four months before my visit to a forest reserve that now bears his name, he had been standing on his back porch when a local rancher shot and killed him. This man and other large land owners were vehemently opposed to Mendes’ promotion of collective land rights for rubber tappers who had worked in servitude to others in these forests for many decades. The global outrage at his assassination propelled the Brazilian government to move forward with creation of Extractive Reserves throughout their Amazon region. These reserves were intended to help collectors of NTFPs such as rubber, Brazil nuts, and palm hearts improve their standard of living and practice sustainable forest management (Fearnside, 1989; Allegretti, 1998). The political appeal of a concept that could simultaneously reduce rural poverty and conserve endangered rain forests was irresistible. The economic viability of this strategy, however, seemed untested until a paper co-authored by

Charles Peters, Alwyn Gentry and Robert Mendelsohn was published in the journal Nature (Peters et al.,

1989). Their study showed that forest peasants in the Peruvian Amazon could potentially make more money from a hectare of forest in the long run by selling an assortment of NTFPs than selling the rights to cut down the best trees to local loggers.

Case studies of NTFPs from the Amazon, Asia, and Africa (Anderson, 1990; Nepstad and

Schwartzman, 1992; Plotkin and Famolare, 1992) provided additional examples of forest peoples who were making a living selling NTFPs without destroying the forest. Panayotou and Ashton (1992) popularized the idea that NTFP harvesting could be integrated with timber management to generate revenue from the full range of forest resources. Other books followed that provided lists of promising NTFP species and recommended strategies for how to implement NTFP sustainable harvest and marketing (Bursztyn et al, 1993; Clay and

Clement, 1993; Peters, 1994; Warner and Pontual, 1994; Oliveira et al, 1995; SEBRAE, 1996; Wollenberg and

Ingles, 1998). Even NTFP promoters, however, acknowledged that making NTFP marketing the focal point of 3 tropical forest community development and forest conservation cannot succeed unless certain biological, social and economic conditions are met. Long-term commercial harvests of NTFPs are problematic if 1) the resource base is too small to be sustainably harvested, 2) the demand for the raw material is too low to generate a reasonable price for collectors, 3) marketing logistics are too complicated, 4) demand is so high that production from wild populations is replaced by plantation sources or other substitutes or 5) benefits from NTFP sales are not distributed equitably within a community (Browder, 1992; Pendleton, 1992; Survival International., 1992;

Homma, 1993; Dove, 1994; Stiles, 1994).

Some of the most optimistic studies of NTFP economic viability have been done in seasonally flooded

(varzea) forests with dense concentrations of a few economic species in areas that have relatively easy access to urban markets (Peters et al, 1989; Anderson, 1990). It has been harder to demonstrate NTFP commercial viability in upland (terra firme) forests where the natural density of economic species is often low (Peters,

1994) and transporting products to market can be difficult and expensive (Homma, 1993). In theory, a community with a diverse base of NTFPs could avoid some of the pitfalls described above by selling moderate amounts of a variety of products (Clay and Clement, 1996) - an idea being explored in several Extractive

Reserves in Brazil (ELI, 1995; Murrieta and Rueda, 1995).

The goal of developing NTFP marketing in the context of community economic development with forest conservation has so far received little attention in indigenous reserves (Rueda and Lisansky, personal communication 1995). These efforts are especially needed in Brazil where such reserves cover 94.6 million ha - more than three times the size of other protected areas in the country combined (Schwartzman, 1996). The potential conservation value of these areas is enormous, but it cannot be assumed that these values will be maintained in the long term without concentrated attention. Many indigenous lands have been destructively exploited for mineral and timber resources or usurped for conversion to non-Indian pastures and farms. Even where indigenous people have relatively secure control of their territory, traditional land use practices are changing in response to groups’ desire to earn more income (Baksh, 1995; Stearman, 1995). Since increasing production of agricultural cash crops can increase deforestation (Baksh, 1995), developing multiple NTFPs as a source of revenue for indigenous people seems like an attractive option since many groups already use hundreds of rainforest plant and animals species for subsistence purposes (Balée, 1994; Phillips et al, 1994). 4

It cannot be assumed, however, that every NTFP that has been collected for consumption in the village will be worthwhile to harvest for sale. Making this determination will require information about the plant’s basic ecology, handling requirements, sustainable harvest levels (La Frankie, 1994; Peters, 1994; Boot and

Gullison, 1995), economic returns (Godoy et al, 1993; Godoy and Bawa, 1993; Warner and Pontual, 1994; Shanley et al, 1996), and conditions that favor plant regeneration (Martini et al, 1994; Laird, 1995). Studies that combine traditional knowledge with an applied scientific approach could provide valuable information to forest peoples that wish to evaluate the economic viability of specific NTFPs and the role that they can play in community development (Shanley et al, 1996).

The overall aim of my research was to conduct production ecology studies on several economic species of NTFPs in an indigenous community in the Brazilian Amazon. I secondly wished to quantify how much income people were making or could make by harvesting and selling modest amounts of various NTFPs.

The specific objectives of each case study were to: 1) quantify the amount of marketable product that could be obtained per plant and per area of forest, 2) identify key factors that influenced variation in these amounts, 3) estimate the amount of harvest that would be possible on a repeated basis and the length of this harvest cycle, and 4) estimate the amount of time a harvester needs to spend to find, harvest and process a unit of product and how much income they would earn for this time invested. In some cases, I also examined the impact of current harvesting techniques on harvested plants and their populations and suggested changes that might reduce negative impacts. I avoided making a commitment to determine the “sustainable” harvest level since this implied an even more comprehensive approach that would assess what impact a particular harvest level could have on the target population, its dependent fauna and other members of the biotic community (Hall and Bawa,

1993). The results of three NTFP case studies will be presented in Chapters 2 (Copaiba oleoresin), 3 (Breu resin), and 5 (Titica vine roots) of this dissertation. Some of the general conclusions will be discussed in

Chapter 6 – Summary Conclusions.

Apart from the scientific oriented goals described above, I wished to grow from the experience of being a foreign researcher working on an applied ecology research project with a group of indigenous people.

I felt that there was so much I could learn about tropical forest ecology and peoples’ relationship to it by living and working with people who had grown up in the forest. I also believed that I had knowledge, skills and 5 passions I was ready to share that would make a positive contribution to whatever community was willing to take me in. The challenges that I encountered trying to balance the demands of doing high-quality scientific work and being a good citizen in my host community are also discussed in the Epilogue. These experiences provided even more powerful lessons for me than the results of the NTFP case studies themselves, and I hope they will provide food for thought for other researchers.

SELECTION AND BACKGROUND OF STUDY AREA

The Alto Rio Guamá Indigenous Reserve is a 278,000 ha federal indigenous reserve in the Brazilian

Amazon (FUNAI, 1975). It is the main homeland for about 800 Tembé Indians who live in villages along the

Guamá and Gurupi Rivers in eastern Pará state (Sales, 1993). The Tembé use hundreds of forest products for subsistence purposes (Balée, 1987) and have a long history selling a few forest and agricultural products (Arnaud,

1982; CEDI, 1985; Sales, 1994). The high biodiversity and deforestation threat in this and five other contiguous indigenous and biological reserves have made the region a top conservation priority of the entire Amazon region

(Conservation International, 1991; Fearnside and Ferraz, 1995; IBGE, 1995).

In 1993, I visited several Tembé villages along the Guamá River to discuss my desire to study the ecology and market potential of NTFPs in cooperation with an indigenous community. The Tembé leaders were very cordial and frank with me. Their biggest concern was that ranchers, colonists and loggers had illegally occupied large areas of their reserve, and day by day these incursions were getting closer to their villages. They told me a few elders would take me around the small section of forest they still controlled, but they said it would be futile to study its NTFP commercial potential. Invaders had deforested and taken over so much of their territory that they could no longer feed themselves by hunting and fishing, and other forest products were becoming scarce. They urgently needed new sources of income to buy food and basic items, but they most needed help to regain control of their land. I said I would think about how I could assist with the land issue and gratefully accepted their offer to let me tour their forest.

For two days my Tembé guides walked with me through the forest and pointed out more than a dozen trees whose boiled bark treated various ailments. They pulled the plug from a drilled copaiba (Copaifera spp.) 6 tree to collect some liquid oleoresin, showed me how to get drinking water from a machete split vine, and introduced me to the sweet aroma of resin on a fallen breu tree (Protium spp.). After tripping over a log, I discovered that rubbing andiroba (Carapa guianensis Aubl.) seed oil on my bruised shin actually reduced the pain and swelling as well as it kept mosquitoes off my exposed skin. It was exciting to finally get a first hand look at the living plants that were the source of forest remedies I had only previously seen in bottles and string tied bundles in the Belém city folk market. This enthusiasm was tempered by the realization that my hosts had not exaggerated their plight. We reached an area of forest that had been burned by colonists after less than an hour’s walk from the village. The copaiba tree we had taken oil from was the only productive tree the elders knew of that was left in that part of the forest.

On my last night in São Pedro village, I was fortunate enough to witness and briefly take part in a traditional Tembé style celebration. I finally climbed into my hammock late at night after the day’s events and a relentless stomping dance left me exhilarated and exhausted. I drifted to sleep listening to the hypnotic shaking of maracas and a chant about a giant macaw. The tune of this song and the half–learned lyrics kept repeating themselves in my mind long after I came home. I did not know how it would happen, but I had a strong feeling I would hear that chant again.

My next opportunity to connect with the Tembé came three years later in the spring of 1996 when my high school invited me to give a presentation on Earth Day and also agreed to my request to sponsor the visit of an indigenous leader from the Amazon. It seemed like a long-delayed chance to make good on my wish to help the Tembé with their land rights’ concern. This time I made a contact with people working with villages along the Gurupi River. Unlike their cousins in the northern section of the reserve, the Tembé in this area were surrounded by large expanses of primary forest. They had abundant land available for hunting, fishing, farming, and collecting forest products for use in the home and occasionally bringing to market. Gurupi communities were also very aware of the problems caused by non-Indian invaders and were trying to confront them from a position of strength. During a ten day trip to the U.S. in April 1996, the Tembé leader Muxi and I gave presentations to twenty-four government agencies, conservation and human rights groups, schools and churches. He enthusiastically endorsed my proposal to study marketable NTFPs in the southern section of the 7 reserve because villages there were anxious to increase their income to improve health care, education and security of their land. I would be welcome to stay at his village called Tekohaw.

SELECTION OF STUDY SPECIES

The process of selecting candidate species for this research began in the markets of Belém. Located near the mouth of the Amazon River, Belém is the main commercial center for the state of Pará and is the transit point for many Amazonian products en route to the rest of Brazil and foreign ports. I first became familiar with the Ver-o-Peso market that hosts the region’s largest concentration of forest product dealers. The folk market section has primarily medicinal products that include barks, plant seeds and exudates (oil, resin, latex), leaves, and various animal products. Other retail vendors sold an assortment of wild and cultivated fruits, and a wholesale market devoted exclusively to açai palm (Euterpe oleracea Mart.) fruit was held every morning. I later visited the CEASA wholesale produce market where businesses handle large volumes of cultivated fruits and minor amounts of regional forest fruits collected from wild populations and small-scale plantings.

When I arrived at Tekohaw village for the summer of 1996, Tembé leaders told me they had many of the trees whose products I had seen in the market. The challenge was identifying which ones looked most promising as species that I could study and that they could profitably harvest. While their forest had most of the medicinal bark trees being sold in the market, I ruled out these products since dealers seemed to have all the supply of these they needed. Learning how much bark could be removed from a tree without causing it too much damage would be an interesting study, but the ready availability of massive quantities of bark obtained from felled trees would leave little incentive for developing careful harvesting techniques. All of the medicinal leaves sold in the market were cultivated, so these offered no opportunities as marketable forest products.

Some fruits had interesting commercial prospects, but the potential for damage or rotting during uncertain river and overland transport from the reserve to the market was worrisome and gaining reliable data on fruiting patterns could take many years of continuous observation (Wheelwright, 1986). The class of products that initially seemed most feasible from a marketing and research standpoint was plant exudates including the latex 8 from amapá (Parahancornia amapa (Hub.)Ducke), seed oil from andiroba and trunk oleoresin from copaiba.

Their advantage was that they did not spoil in a short time, and dealers seemed willing to buy them with no prior arrangement for $US 3 to 8 per liter with the highest prices offered for copaiba. Copaiba was also the only one of these products that had an international market.

Since the Tembé said that copaiba trees were reasonably abundant in their forest and they had experience collecting its oleoresin, I chose to focus on this tree for the summer of 1996 and did follow-up work with it in subsequent years. The details of this case study are presented in Chapter 2. The initial season of work with copaiba had two concrete results. First it showed me that I thoroughly enjoyed working with these people and their forest, and they seemed to enjoy working with me. I was invited to come back for a longer time and bring my family. Secondly, the research showed that copaiba trees were less common and yielded less oleoresin than the Indians had thought. The next phase of the research would need to include some other species.

We tentatively decided to explore amapá latex , andiroba seed oil, cumaru (Dipteryx odorata L.) seeds and seed oil, açai fruit, bacuri (Platonia insignis Mart.) fruit, and cupuaçu (Theobroma grandiflorum (Willd. ex.

Spreng.)Schum.) fruit as potential new market products. These products were all routinely marketed in the region (Clay and Clement, 1993; Shanley, 1994) and commonly used by the Tembé as food or medicines

(Balée, 1987). The research would also quantify production ecology information on resin produced by breu trees (mostly Protium spp.) and aerial roots from the titica vine (Heteropsis spp.) that were the two most important NTFPs already being sold by the Tembé in the Gurupi region.

I returned to the reserve with my wife and two children in the fall of 1997 to take up this ambitious research agenda. Cumaru was removed as a candidate species quickly because it was relatively rare, trees fruited at different times, and fruit fall was spaced over many months. Bacuri trees were supposedly abundant near the village of Rabo de Mukura, but studying them from a base at Tekohaw was not practical. Although people from Tekohaw called one large section of forest near the village the “cupuzal” in supposed recognition of its high density of cupuaçu trees, our surveys of the area found only two of these trees. Whether forest fires that had passed through there in the early 1980s had killed many of these trees or it was just less naturally common than they thought, it was not a large enough population to study. 9

Of all the marketable fruits, açai palm trees were by far the most abundant and important to the

Tembé. When these trees reached their peak fruiting from October through December, every Tembé family devoted considerable time gathering the hard berries and preparing açai juice. I started production studies in several plots, but the work was thwarted by two main factors. The first was that areas that were far enough from the village to be left alone when fruits were abundant in the beginning of the fruiting season became too attractive for some families not to harvest a few months later when fruits were harder to find. The second problem was that a fire that escaped from a farm field swept through one research plot and caused all the fruit in it to abort. Other challenges conducting ecological research in the heart of a community forest will be discussed in Chapter 6 – Summary Conclusions. Some production ecology work was also conducted with amapá and andiroba. These studies were not complete enough to present as full case studies in this dissertation, but a few highlights from work with these species will be discussed in Chapter 6.

It seems somewhat ironic that my research did not end up focusing on new commercial NTFPs for the

Tembé but studying the two forest products that already earn them the most income – breu resin and titica vine roots. As Chapters 3, 4, and 5 will demonstrate, these efforts that sought to blend indigenous knowledge and beliefs with a rigorous data gathering approach provided the most new insights and hopefully most useful results for ecology as a field of scientific inquiry and for people who are looking for practical means to make a simple living in a forest. 10

REFERENCES

Anderson, Anthony B. 1990. ed. Alternatives to Deforestation: Steps toward Sustainable Use of the Amazon Rain Forest. Columbia University Press, New York.

Baksh, Michael. 1995. Changes in Machigeunga quality of life. pp. 187-205 in Leslie E. Sponsel (ed.) Indigenous Peoples and the Future of Amazonia: An Ecological Anthropology of an Endangered World. The University of Arizona Press, Tuscon.

Balée, William. 1987. A etnobotânica quantitativa dos índios Tembé (Rio Gurupi, Pará). Boletim do Meseu Paraense Emilio Goeldi: Botanica, Agosto (1):29-50.

Balée, William. 1994. Footprints in the Forest: Ka’apor Ethnobotany - the Historical Ecology of Plant Utilization by an Amazonian People. Columbia University Press, New York.

Boot, Rene and Gullison, R.E. 1995. Approaches to developing sustainable extraction systems for tropical forest products. Ecol. Applications 5(4): 896-903.

CEDI. 1985. Tembé. pp. 177-209 in Povos Indígenas no Brasil, No. 8: Sudeste do Pará. Centro Ecumênico de Documentação e Informação, São Paulo.

Clay, Jason W. and Clement, C.R. 1993. Selected species and strategies to enhance income generation from Amazonian forests. FAO Forestry Paper (Final Draft). United Nations Food and Agricultural Organization, Rome.

Conservation International. 1991. Workshop 90: Biological Priorities for Conservation in Amazonia. Conservation International, Washington, D.C.

Dove, Michael R. 1994. Marketing the rainforest: “green” panacea or red herring? Analysis from the East- West Center No. 13, May.

Environmental Law Institute (ELI). 1995. Brazil’s Extractive Reserves: Fundamental Aspects of their Implementation. ELI Research Report. ELI, Washington, D.C.

Fearnside, Philip M. 1989. Extractive reserves in Brazilian Amazonia. BioScience 39(6):387-393.

FUNAI. 1975. Situação Geográfica das Terras do “Posto Indígena Alto Rio Guamá” (map). Ministério do Interior, Fundação Nacional do Índio (FUNAI), 2o Delegacia Regional, Belém, Pará.

IBGE. 1992. Atlas Naçional do Brasil. Fundação Instituto Brasileira do Geografia e Estatistica (IBGE), Rio de Janeiro.

Godoy, Ricardo A. and Bawa, K.S. 1993. The economic value and sustainable harvest of plants and animals from the tropical forest: assumptions, hypotheses, and methods. Economic Botany 47(3):215-219.

Godoy, Ricardo A., Lubowski, R. and Markandaya, A. 1993. A method for the economic valuation of non- timber tropical forest products. Economic Botany 47(3): 220-233. 11

Hall, Pamela and Bawa, K. 1993. Methods to assess the impact of extraction of non-timber tropical forest products on plant populations. Economic Botany 47(3):234-247.

Homma, Alfredo Kingo Oyama. 1993. Extrativismo Vegetal na Amazônia: Limites e Oportunidades. EMBRAPA-SPI, Brasília.

La Frankie, James V. 1994. Population dynamics of some tropical trees that yield non-timber forest products. Economic Botany 48(3): 301-309.

Laird, Sarah. 1995. The natural management of tropical forests for timber and non-timber products. OFI Occasional Papers No. 49. Oxford Forestry Institute, Oxford.

Murrieta, Julio R. and Rueda, Rafael P. 1995. Extractive Reserves. IUCN Forest Conservation Programme, Gland. pp. 133.

Nepstad, Daniel C. and Stephan Schwartzman. 1992. eds. Non-Timber Products from Tropical Forests: Evaluation of a Conservation and Development Strategy. Vol. 9. Advances in Economic Botany. The New York Botanical Garden, Bronx.

Panayotou, Theodore and Peter S. Ashton. 1992. Not by Timber Alone: Economics and Ecology for Sustaining Tropical Forests. Island Press, Washington, D.C.

Pendleton, Linwood H. 1992. Trouble in paradise: practical obstacles to nontimber forestry in Latin America. pp. 252-262 in eds. Plotkin, Mark and Famolare, L. Sustainable Harvest and Marketing of Rain Forest Products. Island Press, Washington, D.C.

Peters, Charles M., Gentry, A.H., and Mendelsohn, R.O. 1989. Valuation of an Amazonian rainforest. Nature 339:655-656.

Peters, Charles M. 1994. Sustainable harvest of non-timber plant resources in tropical moist forest: an ecological primer. Biodiversity Support Program c/o World Wildlife Fund, Washington, D.C.

Plotkin, Mark and Lisa Famolare. 1992. Sustainable Harvest and Marketing of Rain Forest Products. Island Press, Washington, D.C

Rodrigues, Roberto M. 1989. A Flora da Amazônia. CEJUP. pp. 463.

Sales, Noêmia Pires de. 1993. Pressão e Resistência: Os Índios Tembé-Tenetehara do Alto Rio Guamá e a Relação com o Território. União das Escolas Superiores do Pará, Diretoria de Pesquisa, Pós- Graduação e Extensão, Depto. de Ciências Socias. Relatório Final.

Sales, Noêmia Pires de. 1994. Os Tembé no Alto Rio Guamá: Reelaborações Étnicas - Identidade e Território. UNAMA - Universidade da Amazônia, Pró-Reitoria de Pesquisa, Pós-Graduação e Extensão, Depto. de Ciências Socias. Relatório de Pesquisa.

Schwartzman, Steve. 1996. Brazilian indian lands threatened. Environmental Defense Fund electronic mail alert.

Stearman, Allyn MacLean. 1995. Neotropical foraging adaptations and the effects of acculturation on sustainable resource use. pp. 207-224 in Leslie E. Sponsel (ed.) Indigenous Peoples and the Future of Amazonia: An Ecological Anthropology of an Endangered World. The University of Arizona Press, Tuscon. 12

Stiles, Daniel. 1994. Tribals and trade: a strategy for cultural and ecological survival. Ambio 23(2):107-111.

Survival International. 1992. “Rainforest harvest” projects harm, not help, Indian communities. Survival International Press release June 15. London.

Warner, P.D.III. and Pontual, A.C. 1994. Manual de Comercialização de Produtos Florestais. GENESYS (Gênero em Sistemas Econômicos e Socias), The Futures Group, Washington, D.C.

Wheelwright, Nathaniel T. 1986. A seven-year study of individual variation in fruit production in tropical bird- dispersed tree species in the family Lauraceae. pp.19-35 in eds. Estrada, Alejandro and Fleming, T.H. Frugivores and Seed Dispersal. Dr. W. Junk Publishers, Dordrecht

Wollenberg, Eva and Andrew Ingles. 1998. eds. Incomes from the Forest: Methods for the Development and Conservation of Forest Products for Local Communities. Center for International Forestry Research, Bogor.

13

CHAPTER 2

THE ECOLOGY OF COPAIBA (COPAIFERA SPP.) OLEORESIN HARVEST IN

THE AMAZON 14

ABSTRACT

Oleoresin extracted from the trunk of many species of copaiba (Copaifera spp.)(Caesalpiniaceae) trees in the Amazon basin and other parts of South America has been used as a medicine by indigenous people and westerners for more than 400 years. The oleoresin is rich in volatile and non-volatile terpenoid compounds produced in parenchyma cells in many parts of the tree. The material is deposited into resin canals in the sapwood and can accumulate large amounts in cavities when these canals break down in the heartwood. Trees have been harvested by cutting into the trunk with an axe and by drilling into the heartwood. Popular accounts indicate that the best producing species can yield a range of 2.5 to 24 liters of oleoresin per tree with some trees yielding as much as 60 liters at one time. The only previous systematic production study of a copaiba species (C. multijuga Hayne), however, showed that average yields were 0.2 liters per tree during the initial harvest, and that this average steadily approached zero on each of four successive harvests.

I studied copaiba production ecology in the Alto Rio Guamá Indigenous Reserve, Pará, Brazil to evaluate the harvesting economics of this product as a means to increase income for Tembé Indians in the reserve. The study involved drilling three varieties of copaiba trees one meter above ground level and measuring oleoresin yields at the time of the initial harvest and two more times. The average oleoresin yield for the first harvest was 0.073 liters per tree for all 57 trees that were drilled. The average first harvest yield among trees that yielded some resin during the study was 0.230 liters per tree, but this amount rapidly dwindled during successive harvests one and two years later. There were some differences among tree varieties, but trees in the upper-middle size classes (45 - 65 cm DBH) were found to yield the most resin while small trees (<45 cm DBH), very large trees (>65 cm DBH), and hollow trees of any size yielded little or no oleoresin. Internal measurements showed that the size at which oleoresin yield first occurred corresponds well to the size at which heartwood began to form. Resin yield began to drop off in very large trees in which hollowness became increasingly common. Some small trees seemed to be prematurely hollow due to burn damage sustained in an accidental fire that passed through the forest in the 1980s.

The natural low density of copaiba trees (e.g. 0.1 to 2 trees/ha) means that much more time may be needed to initially locate trees than to harvest them. Where density and average production are both low, 15 commercial harvesting may not be viable. Knowledge of specific copaiba features and preferred habitats can improve search efficiency for these trees that are widely dispersed in a diverse forest. Drilling two holes in a tree can take one person 40 minutes and waiting to capture all oleoresin escaping from a large cavity could take hours or days. With large flows, it would be more efficient to leave a collecting device in place and return later. Unless densities, average production, and the number of potential harvests per tree are high, though, copaiba oleoresin extraction will not yield more than a few dollars per day for one person in a 1000 ha area of forest. Plantations of copaiba trees could dramatically increase harvesting efficiency, but it seems unlikely that the price for the oleoresin alone would justify the expense of maintaining these cultivated trees for 70 years or more until resin is potentially extractable in sizeable amounts. The existence of local markets for unprocessed copaiba oleoresin and small businesses that distill and repackage it into several forms, however, shows that favorable economic conditions for harvesting and marketing copaiba from wild populations do exist in some parts of the Amazon. Additional studies are needed to document what species, densities, harvesting techniques, and other conditions make these operations profitable and if they are extracting the raw material on a repeated basis. 16

INTRODUCTION: REVIEW OF COPAIBA PRODUCTION AND USE

OLEORESIN PRODUCTION IN COPAIFERA TREES

Copaiba trees belong to the genus Copaifera in the family Caesalpiniaceae (or Fabaceae sub-family

Caesalpinoideae). The scientific name is derived from a combination of the Tupi Indian word for the resin from the copaiba tree, “copai”, and the Latin verb, “fero”, meaning to bear or produce (Marafioti, 1970; Allen and Allen, 1981). Authors normally attribute 28 to 40 species to this genus that includes both shrubs and trees

(Record and Mell, 1924; Allen and Allen, 1981; Dwyer, 1951). Most Copaifera plants are native to the neotropics with at least 25 species represented in the Amazon region (Gentry, 1993). Another four or five species are found in the African tropics (Record and Mell, 1924; Langenheim, 1973). Neotropical Copaifera species produce liquid oleoresins while material found in African species can harden to a copal like substance

(Anderson, 1955; Langenheim, 1973).

Copaiba oleoresin is produced in the tree’s trunk, stems, and leaves. It consists of large but varying amounts of volatile oils (primarily composed of sesquiterpene hydrocarbons usually including caryophyllene), non-volatile resinous substances and small quantities of acids (Furnemore, 1926; Fernandes, 1949;

Langenheim, 1973; Maia et al. 1978; Leung, 1980; Herres et al, 1986; Duke, 1986; Figliuolo et al., 1987;

Budavari, 1989; Duke and Cellier, 1993; Braga et al., 1998; Monti et al., 1999; Cascon and Gilbert, 2000).

The combination of the sesqui- and diterpenes along with phenolics may be responsible for the plant’s resistance to many insect and microbial attacks (Figliuolo et al., 1987; Feibert and Langenheim, 1988; Macedo and Langenheim, 1989a, 1989b; Langenheim, 1990; Schultes and Raffauf, 1992). The compounds are synthesized in parenchyma cells that line rounded pockets, cysts or elongated canals (Langenheim, 1973). The resin is secreted into these spaces that are formed by schizogeny and/or lysigeny. Schizogeny leads to the separation of cells, which increases their intercellular space to create thin capillaries. Canals up to two meters long have been observed (Correa and Bernal, 1989). In Copaifera, these canals are located in concentric circles in the secondary wood that may delineate seasonal growth (Record and Mell, 1924; Langenheim, 1973; 17

Alencar, 1982; Vetter and Botosso, 1989). In addition to its presence in these schizogenous ducts, the oleoresin sometimes accumulates in lysigenous cavities formed by the breakdown of secretory cells. In the center of the trunk these cavities may enlarge to the point where they contain gallons of oleoresin (Youngken,

1943; Osol and Farrar, 1947; Poucher, 1950).

About twenty Copaifera species from South America are known to be sources of oleoresin. The main commercialized species are: C. reticulata Ducke (80% in the Brazilian Amazon), C. guianensis Desf. (10% in

Brazilian Amazon); C. multijuga Hayne (5% in Brazilian Amazon), C. officinalis L. (3-5% in Brazilian

Amazon), C. langsdorffii Desf., and C. martii Hayne (Allen and Allen, 1981; Sampaio, 1993; MMA/SCA et al., 1998). Oleoresin has also been harvested in the region from: C. bijuga Hayne, C. canime Harms, C. confertiflora Benth., C. coriacea Mart., C. cuneata Tul., C. duckei Dwyer, C. glycycarpa Ducke, C. longicuspis

Ducke, C. longifolia Huber, C. marginata Benth, C. pubiflora Benth, C. spruceana Benth, and C. venezuelana

Pittier and Harms (Osol and Farrar, 1947; Howes, 1949; Dwyer, 1951; Mors and Rizzini, 1966; Allen and

Allen, 1981; Duke, 1986; Sampaio, 1993; Cascon and Gilbert, 2000), and C. camibar Poveda and Sanchez

(Berry et al., 1997). The wide distribution of the genus and its broad utility is reflected in the numerous common names it has throughout the continent (Appendix 2-A).

The chemical composition, color and viscosity of oleoresin varies widely from species to species.

The “Para” variety of copaiba oleoresin comes from several species including C. reticulata, C. multijuga and C. martii (Osol and Farrar, 1947). These are most common in Pará and Amazonas states in Brazil. Since C. duckei is closely related to C. reticulata (Dwyer, 1951), its oleoresin is probably also included in this group.

The oleoresin of these species in general has the highest volatile oil content (35 - 90%) and greatest fluidity among the commonly harvested copaiba trees (Youngken, 1943; Osol and Farrar, 1947). C. reticulata oleoresin emerges from the tree as a clear, colorless and thin liquid, but on exposure to the air it turns a thick yellowish brown and has a disagreeable smell (Le Cointe, 1947; Allen and Allen, 1981; Duke, 1986). C. multijuga is fluid, pale-yellow to transparent and has a light agreeable smell (Le Cointe, 1947; Allen and Allen,

1981) with 58 - 78% essential oils (Furnemore, 1926). A recent analysis found that total sesquiterpenes

(hydrocarbons and oxides) accounted for 80 - 97% of several samples in this species (Cascon and Gilbert,

2000). The botanical characters of the broadly defined species C. langsdorffii are often hard to distinguish 18 from C. reticulata and other Copaifera (Dwyer, 1951). This species including its many varieties is considered the principle source of the more viscous copaiba oleoresin commonly known as the “Maranham” or “Rio de

Janeiro” variety. It has the consistency of olive oil with a range of 30 - 75% volatile oils (Sayre, 1906;

Kraemer, 1907; Furnemore, 1926). The “Maracaibo” variety of copaiba oleoresin principally comes from C. officinalis and the related C. guianensis in the northern Amazon countries of Venezuela, Columbia and the

Guianas (Sayre, 1906; Osol and Farrar, 1947; Dwyer, 1951). It is a thick, dark, and coumarin scented product that has a 40 - 72% range of volatile oils in Surinam and 20 - 40% volatile component in Columbia (Sayre,

1906; Furnemore, 1926; Le Cointe, 1947).

The groupings of Copaifera species that have progressively denser oleoresin are not geographically distributed in the same order. C. langsdorffii with its many variants is found in subtropical dry through wet forest life zones from northern Argentina, through Paraguay and Brazil up into Venezuela and Guyana (Dwyer,

1951; Duke and Cellier, 1993). Of the major oleoresin producing species, it is the species most commonly found in dry woodland areas (USDA, N.S.; Duke and Cellier, 1993). C. reticulata is found primarily in moist tropical forests of Brazil and Bolivia (Dwyer, 1951). It grows best in flood-free areas of forest on sandy soils

(Duke, 1986). Farther to the north C. officinalis and C. multijuga likewise tolerate a variety of dry and wet conditions. C. multijuga can be found in terra firme (non-flooded rainforest), varzea (flooded forests), sandy margins near lakes and streams and even the cerrado (savannah) of central Brazil (Duke, 1986; Sampaio,

1993). It is most often found on extremely nutrient poor soils (Figliuolo et al, 1987). The main home range of

C. officinalis is in Venezuela, but it is also found in the West Indies and the Brazilian state of Bahia (Dwyer,

1951). Its relative C. guianensis is found in Brazil and its range also includes the Guianas, Columbia and

Panama (Dwyer, 1951). Although some leguminous non-woody plants and trees are known for their ability to fix nitrogen through root nodules, Copaifera and other genera in its Detarieae tribe apparently do not possess this capability and still persist well in nutrient poor soils (Figliuolo et al, 1987; Moreira et al., 1992).

COPAIBA ETHNOBOTANY, MODERN USE AND TRADE

Many Amazonian indigenous and traditional forest peoples including the Kaw, Tupi, Puinaves and

Makunas have used this chemically potent oleoresin as a traditional medicine (Record and Mell, 1924; Allen 19 and Allen, 1981; Schultes and Raffauf, 1990; Plotkin et al., 1991). The first western record of these trees was in a 1534 report to Pope Leo X that referred to them as “copei.” A Portuguese monk writing about the natural products of Brazil in 1625 mentioned a drug called “cupayba”. A tree yielding the substance was first described and illustrated in 1648, and the London Pharmacopoeia listed the drug in 1677 (Youngken, 1943).

Linnaeus formally described the genus Copaifera in 1762 (Woodson and Schery, 1951). The oil was included in the United States Pharmacopoeia from 1820 to 1940 when it was admitted to the National Formulary

(Youngken, 1943).

Copaiba was traditionally used in the Amazon region and abroad as a cicatrizant, anti-inflammatory, and a treatment for chronic gonorrhea and skin ailments including herpes, sores, eczema and psoriasis. Internal uses include treatment for a variety of ailments involving mucous membranes such as sore throat, bronchitis, ulcers and various conditions of the genito-urinary tract (including cystitis, leucorrhea, syphilis and urinary incontinence). It is also used as anti-rheumatic, antiseptic, anti-bacterial, diuretic, expectorant, hypotensive agent, laxative, purgative, vermifuge and vulnerary (Youngken, 1943; Uphof, 1968; Leung, 1980; van den

Berg, 1984; Duke, 1986; Basile et al., 1988; Schultes and Raffauf, 1990; Duke and Vásquez, 1994; Viera,

1992; Fleury, 1997). Both C. coriacea and C. multijuga have been noted for various medicinal properties. The species most consistently mentioned for effectiveness in treating tough urinary tract, pulmonary and skin ailments are C. langsdorffii, C. officinalis, and C. reticulata. As mentioned, oleoresins from these trees are thicker than other Copaifera species and have an acrid bitter taste (Le Cointe, 1947; Uphof, 1968; Rodriguez,

1989; Sampaio, 1993). To cut the bitter taste in internal preparations, it is sometimes mixed with andiroba oil

(from Carapa guianensis Aubl.), honey, santal oil, and cubebs (Viera, 1992).

While the advent of anti-biotics decreased foreign demand for copaiba as a treatment for venereal disease (Osol and Farrar, 1947), modern studies have demonstrated the folk medicine’s effectiveness as an anti-bacterial (Verpoorte and Dahl, 1987), anti-oxidant (Desmarchelier et al., 1997) and anti-inflammatory agent (Basile et al., 1988). Today, copaiba oil is still commonly sold in folk medicine markets such as Ver-o-

Peso in Belém. It is also sold in various commercial preparations such as gel caps and included in several

Brazilian brands of “natural” cough syrup (MMA/SCA et al., 1998; Amazon Ervas, N.S.; Lab oratorio São

Lucas, N.S.). 20

Apart from medicinal uses, copaiba oleoresin has been widely used in other applications. The raw material, distilled essential oils, and resinous compounds have made copaiba useful in the preparation of cosmetics, lacquer, paints, varnishes and tracing paper (Youngken, 1943; Fernandes, 1949; Balsam and

Sagarin, 1974; Leung, 1980; Duke, 1986; MMA/SCA et al., 1998). Cosmetic preparations includes soaps, detergents, creams, lotions, massage oil, and perfumes where copaiba oil or resin serves as a fragrance and odor fixative (Poucher, 1950; Leung, 1980; Duke, 1986). Both copaiba oleoresin’s (referred to in the trade as copaiba balsam) and copaiba oil’s ongoing use by cosmetic manufacturers is demonstrated by its inclusion on product lists of eleven U.S. raw material suppliers to this industry (DCI, 1992). Although the term balsam is often used with reference to copaiba, this oleoresin does not fit balsam’s true chemical definition because it lacks the requisite benzoic or cinnamic acid or their esters (Leung, 1980).

The high concentration of sesquiterpene hydrocarbons in C. multijuga (Cascon and Gilbert, 2000) led to the discovery that it could be used directly in diesel engines as a fuel (Record and Hess, 1943; Uphof, 1968;

Wang and Huffman, 1981; Calvin, 1983; Duke and Cellier, 1993). Experimental plantations of the tree were started in the early 1980s near Manaus, Brazil with the idea of producing copaiba oil as an alternative energy source to fossil fuels (Calvin, 1983; MMA/SCA et al., 1998). Even if these could not produce sufficient supplies directly, Calvin (1983) speculated that sesquiterpene producing genes from copaiba might be transferred to a suitable prolific Euphorbiaceous plant to maximize biosynthetic production of these chemicals.

As these plantations approached their 16th year, greater emphasis was being placed on the trees as a source of oleoresin for medicinal and industrial products and wood for pulp and timber (MMA/SCA et al., 1998).

Trade statistics show that Brazil and Venezuela have been the major commercial producers and exporters of copaiba oil. During the late 19th and entire 20th century, the U.S. was the largest importer.

France, Germany, United Kingdom and Japan are other important foreign markets (Duke, 1986; MMA/SCA et al., 1998). Between 1883 and 1920, the U.S. bought at least 15 to 28 tons of the Maracaibo variety of copaiba oil per year from Venezuela (Pittier, 1926; Furnemore, 1926; Dwyer, 1951). U.S. imports picked up shortly before World War II and have fluctuated around 70 to 100 tons per year since (USDA, N.S.; Osol and Farrar,

1947; MMA/SCA et al., 1998). Measuring actual production in the region is difficult, but the Brazilian

Amazon exported about 120 tons to other parts of Brazil and foreign countries in 1994. Amazonas state is the 21 largest exporter accounting for 89 tons per year worth about $US 215,000. Rondônia and Amapa states also exported some of the product (MMA/SCA et al., 1998). The Pará variety is the type of copaiba oil most commonly traded in Brazil (Dwyer, 1951).

OLEORESIN HARVESTING PRACTICES AND YIELD

The first people to use copaiba may have taken advantage of oleoresin that occasionally exudes to the outside of the tree, but all accounts of its use mention two main methods of harvesting the liquid. The first technique involves chopping a large hole or wedge near the base of the tree with an axe. This action may expose an oleoresin-filled cavity and creates a drainage reservoir into which the oleoresin ducts will eventually drain their fluid (USDA, N.S.; Pittier, 1926; Furnemore, 1926; Howes, 1949; Poucher, 1950; Youngken, 1943;

Spruce, 1970). The technique may also kill the tree or at a minimum preclude any potential future harvest

(Pittier, 1926; SEBRAE, 1995).

The second, more recent method of harvesting copaiba is to drill a 1.5 to 5 cm diameter hole 0.6 to 1.0 meter from the ground into the heartwood center of the tree and collect the resin that drains out over successive hours or days in a container. Once a tapping session is concluded, the hole is tightly plugged with a piece of wood (Pittier, 1926; Calvin, 1983; Duke and Cellier, 1993; SEBRAE, 1995). Harvesters sometimes bore two holes into the trunk. The first hole is drilled less than a meter above the ground; the second one is bored three to seven meters higher - sometimes after the flow has stopped in the bottom hole (Osol and Ferrar, 1947; Mors and Rizzini, 1966; Correa and Bernal, 1989; Duke, 1986; Duke and Cellier, 1993). A variation on the double drilling technique involves breaking off a large branch from the tree (Plowden, in press). Fires are sometimes made at the base of the trunk to hasten resin flow (Duke, 1986; Shanley et al., 1998). A pump was occasionally used to accelerate oleoresin extraction in Venezuela (Pittier, 1926). Several accounts indicate that some species are most often tapped in the dry season to maximize oil yield (Herrera, 1921; Le Cointe,

1947; Rodriguez, 1989; Duke and Cellier, 1993) while other species may yield more in the rainy season

(Dwyer, 1951; Alencar, 1982; SEBRAE, 1995; MMA/SCA et al., 1998). Recommended harvest frequency varies from twice a year (Calvin, 1983) to once every two years (SEBRAE, 1995; MMA/SCA et al., 1998). 22

Different accounts portray widely varying oleoresin yields from copaiba trees. At one extreme are reports that the pressure of the liquid can be so high that it sometimes bursts out of the trunk with a loud explosion (Sayre, 1906; Pittier, 1926). Various authors state that harvesting one tree can produce 40 - 62 liters of oleoresin at a time (Sayre, 1906; Pittier, 1926; Osol and Ferrar, 1947; Poucher, 1950; MMA/SCA et al.,

1998). Other reports indicate that average oleoresin yield per tree is 2.5 liters (SEBRAE, 1995; MMA/SCA et al., 1998), 4.5 liters (Fernandes, 1949), or 20 - 24 liters (Calvin, 1983; Correa and Bernal, 1989; Duke and

Cellier). Even these so-called averages may be based on anecdotal accounts since they do not reference specific studies. They are all much higher than the results of the one previous systematic study of copaiba oleoresin production conducted near Manaus, Brazil (Alencar, 1982). In this study, 82 C. multijuga trees yielded an average of 0.2 liters of oleoresin in the first of five successive tappings. While yields increased for some trees during the second harvest, production dropped to an average 0.03 liters during the final tapping two and a half years after the start of the experimental harvest. The largest amount that ever came out of a tree at one time was 3.5 liters.

Several factors may explain the large discrepancy between the apparent abundant oleoresin in the anecdotal accounts and the modest amounts harvested in the Alencar study. It’s likely that different species in other areas produce more or less than the C. multijuga population near Manaus. Spruce (1970) notes that C. martii trees near Santarem were relatively common but yielded so little oil they were not worth tapping. While

C. multijuga is considered a major oleoresin producer, other populations of this species or other species in more favorable environments could generate more resin. Copaiba trees that were harvested with an axe probably account for some of the highest reported yields; this method would predictably collect more oil than boring one or two holes drilled near the ground. Another factor may be that earlier accounts are based on second-hand stories of high yielding trees that underplay or ignore trees that yield little or no oleoresin.

Local lore indicates that successful harvest of copaiba oleoresin is far from guaranteed. In the eastern

Brazilian Amazon, forest peasants and Indians commonly say that looking up into the canopy before drilling or chopping into a tree will cause the liquid to get sucked up to the top of the tree (Shanley et al., 1998; Plowden, in press). The other stipulation is that a collector (presumably a man) should not have sexual relations with his wife for several days before setting out to harvest copaiba. Folklore also suggests that oleoresin will not leave 23 the tree if a pregnant woman is present beneath it (Shanley et al., 1998). The central dilemma regarding copaiba is that it has been assumed that trees regularly produce oleoresin in live tissues throughout their lives, but there has been tremendously unexplained variability in the success people have had in harvesting this product.

Due to the uncertainties about “average” oleoresin production, it is hard to estimate the number of trees that have been harvested with an axe or a drill to produce more than 100 tons of copaiba oil per year in the Brazilian Amazon alone. During the peak of the trade in the early 20th century, it seems likely that thousands of copaiba trees were destructively harvested with an axe. Since copaiba has a low density that varies from 0.1 to 2 trees per hectare (Ramirez and Arroyo, 1990; Sampaio, 1993; SEBRAE, 1995; MMA/SCA et al., 1998; ter Steege and Zondervan, 2000) harvesting with the destructive harvesting method may have taken a serious toll on copaiba populations (Pittier, 1926; Dodt (1939) cited in Sales, 1994). The vulnerability of Copaifera to logging (Martini et al., 1993) reinforces the caution that will be needed to maintain current populations if adult trees are killed through logging or destructive resin harvesting. Copaiba oleoresin is frequently cited as a potential income generating non-timber forest product (NTFP) for forest peoples in the region (Sampaio, 1993; Warner and Pontual, 1994; SEBRAE, 1995; MMA/SCA, 1998), so there is a strong need to better understand the factors that affect its production and realistically assess harvest potential using minimally destructive methods.

RESEARCH OBJECTIVES

In spite of the long history of copaiba oleoresin use as a medicinal product and cosmetic by indigenous people and modern pharmaceutical companies throughout the Amazon region, there have been almost no systematic efforts to quantify the yield of oleoresin from various types of Copaifera trees. There is even less understanding of what factors related to the tree and its environment influence the amount oleoresin that is produced and stored and how various harvesting techniques and other factors influence the amount that is harvestable. The main goal of this study was to investigate the production ecology of a diverse group of copaiba trees in the Alto Rio Guamá Indigenous Reserve in the eastern Brazilian Amazon and compare results to the Alencar (1982) study and popular accounts. The specific objectives were to combine field observations 24 with low impact harvesting to learn how factors such as tree type, tree size, tree condition (fire damaged, hollow or undamaged), internal trunk component dimensions (amount of sapwood and heartwood), bore hole number, and drilling frequency relate to oleoresin harvest. Since oleoresin flow seemed to vary dramatically between different parts of the tree (sapwood, sound heartwood, rotten or hollowed heartwood), special attention was also paid to elaborating these differences as well to gain a better understanding of the timing and location of oleoresin production and storage in the trees. A secondary objective was to evaluate the harvesting efficiency and economics of copaiba oleoresin extraction with the drilling technique and recommend criteria for optimizing its operation.

STUDY SITE

Between 1996 and 1998, a series of experimental harvests of copaiba trees were conducted in the Alto

Rio Guamá Indigenous Reserve located in eastern Pará state of the Brazilian Amazon. This was part of a larger study to assess the production ecology of a five different commercial NTFPs in the reserve. The main study site was an area of about 500 ha of closed tropical rainforest near the Tembé Indian village of Tekohaw

(020 37.7’ S; 460 33.1’ W) located on the Gurupi River. Underlying soils are on the border of two major soil types classified by the Brazilian system as yellow latossols and red-yellow podzols - the latter called spodosols in the American system (Projeto Radam, 1973; Kalpagé, 1974). The site has three principal forest types: upland (“terra firme”) forest, lowland forest that is occasionally flooded (“baixo”), and seasonally flooded forests near streams and the main river (“igapó”). Near this and other villages, patches of mostly terra firme forest are cleared and burned to plant manioc and other crops (Sales, 1993 & 1994). This land use pattern creates a mosaic of forest in various stages of current farm use, fallowed secondary forest and old forest. Other important aspects of the landscape are swaths of secondary forest that resulted from unintentional forest fires that burned through non-agricultural sections of the area in the early 1980s. The terra firme forest that hosts copaiba is dominated by trees in the Lecythidaceae, Burseraceae, Leguminosae, Sapotaceae, and

Euphorbiaceae families (Balée, 1987).

In the early 20th century, harvesting copaiba oleoresin was one of the principal economic activities of the Tembé in the region (CEDI, 1985; Sales, 1993 & 1994). Itinerant river traders (called “regatões”) were the 25 primary buyers who exchanged basic commodities with the Indians for copaiba oil and other forest products.

The terms of trade were so unfavorable to the Tembé, though, that extended families often moved in response to the exhaustion of local copaiba resources (CEDI, 1985). This predatory harvest became so widespread that one anthropologist who was concerned about the depauperate situation of the tribe in 1939 recommended that the government use “energetic means to stop the complete destruction of the copaiba trees” (Dodt, 1939 cited in Sales, 1994).

METHODS

In the summer of 1996, a four person team of Tembé elders searched the forest in the vicinity of

Tekohaw village with the author for any type of adult (>10 cm DBH) copaiba tree. Searches were conducted primarily in patches of intact or mostly intact terra firme forest sometimes bordered by agricultural fields, young secondary forest, or “baixo” (occasionally flooded) forest in June and July. Searches were concentrated in terra firme areas where the Tembé believed they were most likely to find the widely dispersed trees.

Additional trees located in the area during a systematic inventory of part of the study area were later added to the study.

When trees were located, their diameter was measured at two or more locations within 20 cm of breast height. The measuring points were noted in relation to their distance directly below two nails put in opposite sides of the trunk. This distance was measured on a pair of metal tape measures hung on each nail.

Common tree names were noted according to Tembé classification. Using sight and sound, it was determined to the extent possible if the tree had a large hollow. Other observations included noting the presence of wound scars on the bark that indicated past axe harvesting, recording the presence of burn wounds, and noting if the tree was alive or dead or had any other circumstances that might affect the tree, such as damage from another fallen tree. Tree positions were initially mapped using paces and compass direction; they were later mapped with a GPS unit. Leaf samples were collected when possible to assist with identification. Each tree was marked with a numbered aluminum tag.

The first round of the oleoresin harvest was conducted in the local dry season in July and August,

1996. Depending on tree size and condition, one to three holes were bored toward the center of the trunk with 26 a drill that was 19 mm in diameter and approximately 0.5 meter long. With the exception of two trees (around

28 cm DBH), no tree less than 34 cm DBH was drilled. If more than one hole was bored, subsequent holes were bored about 60o to either side of the initial hole. The first hole was generally made about 1 meter from the ground; other holes were started about 20 cm higher or lower than the first hole to avoid intersecting bore holes. While some trees thought to be hollow were drilled in the first few weeks of the study, obviously hollow trees were not subsequently drilled.

As the drill bore passed into the tree, the distance from the outside of the bark to the following points was determined by inserting a stick into the bore hole when the drill bit was withdrawn and measuring the length of the stick to that point: 1) the beginning of the heartwood (as noticed by the transition from light colored sapwood to dark reddish wood, 2) the place where any oleoresin started flowing out, 3) the beginning and end of any hollow or low density rotten wood, and if reached 4) the transition from heartwood to sapwood on the opposite side of the tree. The angle of the drill with respect to a plane perpendicular to the trunk was also recorded. The cosine of this angle was multiplied times the actual distance the drill bit traveled before reaching these points to estimate the distances from the bark edge to that wood type on a plane perpendicular to the trunk’s vertical orientation. If any oleoresin emerged, it was collected and measured in graduate cylinders or beakers. If no oil came out within several minutes of drilling, a hardwood plug was banged into the drill hole. Observations were also noted for any other liquids or sounds that accompanied the drilling. Some trees in good condition (i.e. not rotten or hollow) that did not produce oil immediately after drilling had the plug briefly removed within the next few days to collect any small amount of oil that may have accumulated.

The second round of harvesting took place in November, 1997, approximately 14 months after the initial drilling. Tree diameters were remeasured and bore-hole plugs were removed to measure oleoresin flow.

No new drilling was conducted except to drill through plugs that broke off during attempts to pull them out.

New plugs were made if necessary to replace original rotted ones. Where oleoresin had leaked onto the trunk in between harvesting events, this was noted. A third round of harvesting was conducted in May, 1998. In this round, trees drilled in 1996 were again checked in the same manner as described for the second round. A few trees not drilled in the first round and other trees subsequently found in the study area were drilled for the first 27 time with the same protocols. This round, the amount of time required to drill to different parts of the tree was also recorded along with the amount of oleoresin that had accumulated at successive five minute intervals.

RESULTS

TREE NUMBERS, TYPES, SIZE, AND CONDITION

The initial search for copaiba trees in the study area in 1996 found 55 copaiba trees. Another 15 trees were found in 1997 and 1998 in the course of doing work with other tree species in the area. A systematic survey of 7 hectares of terra firme forests located 6 copaiba trees (i.e. 0.86 trees per ha). This estimated density, however, may not represent the study area as a whole since only 70 trees were found in an intensively traveled area of two to three hundred hectares.

The Tembé recognize four types of copaiba trees in the nearby forests. The most common and smallest variety was the white copaiba tree (“copai ching”) that had an average DBH of 47.3 ± 2.5 cm (n=43).

It usually had a slightly tapering cylindrical trunk, smooth grey bark and whitish sapwood. Its oleoresin is pale yellow and flows readily. The mid-size copaiba tree in the area is called “copai kuru;” this is best translated as the knobby copaiba tree. These trees had an average DBH of 59.3 ± 4.0 cm (n=14). It also has a grayish cylindrical trunk and light colored sapwood, but it is covered with round eruptions on the bark that are just under 1 cm across and 0.5 cm tall. Its oleoresin is grayish and very viscous. The largest copaiba tree in the area is the red copaiba tree (“copai pirang”) whose trunk flares out into a square base. Red copaiba trees at the site had a mean DBH of 70.2 ± 5.5 cm (n=13). The tree’s bark is usually ridged and brownish red, and its sapwood and medium viscosity oleoresin are also reddish-orange. A fourth type of copaiba tree called “copai hu,” the “large copaiba” tree, is occasionally found close to main rivers. Its characteristics are a blend of those described above. Since only one specimen of this type of tree was found several kilometers from the other study trees in this very different habitat, it was not included in this study.

The common Tembé names for the copaiba trees in their area were applied consistently by different

Indians, but there were variations that made placing some trees into one of these three categories more 28 difficult. Some trees classed as white copaiba trees had a square base like the red copaiba and/or a few knobs typically found on the knobby type. It is not known if the variants of these three types represent separate varieties, separate species, or occasional hybrids between closely related species. Where some doubt existed about how to classify a particular tree, the name designated by the elder during the initial survey has been used.

Attempts were made to gather flowers from study trees to aid in their identification, but since these trees generally only reproduced once every couple of years, collecting the diminutive and short-lived flowers remained an unfulfilled task. A botanist from the Museu Goeldi reported that leaf samples collected from study trees came from at least three possible species of Copaifera, but it was not possible to match species names with the Tembé common names.

A relatively high percentage (21.4%) of the copaiba trees observed in this study had been damaged by fire. The percentage was greatest among the white copaiba trees that were most often found in areas subject to the forest fires of the 1980s. Relatively few knobby and red copaiba trees were burned in this unintentional fire, but several trees were burned or lost in the course of the study in forest areas converted to manioc cultivation.

The other notable feature of the copaiba study trees was that 48% of them had some hollow or rotten wood in their trunk. The percentage of trees with a hollow space increased significantly with tree size class (χ2

= 18.6; df=6; p=0.01) and was significantly higher in fire damaged trees than unburned trees (χ2 = 8.0; df=1; p=0.01)(Figure 2.2). The proportion of tree types with a hollow was highest in red copaiba, but the percentage of hollow trees was not significantly different between the types.

TRUNK DIMENSIONS

During preliminary copaiba drilling, Tembé elders stated that oleoresin was only obtained when the heartwood (“miolo”) of the tree was exposed with a drill or an axe. Consequently this study recorded the dimensions of sapwood, heartwood, oleoresin bearing zones and hollows in each tree and the amount of oleoresin obtained. When drilling into a tree, the transition from white sapwood to dark dark heartwood was also accompanied by an increased difficulty in drilling. The distance from the outer bark to the beginning of 29 heartwood represents the width of the bark plus sapwood. This distance had a mean of 14.2 ± 0.4 cm (Table

2.1; Figure 2.1) and was not significantly related to overall tree size (Table 2.2). Heartwood diameter was estimated by subtracting two times the distance from the bark to the beginning of the heartwood from the total diameter. The ratio of heartwood diameter to overall tree diameter (at drill height) and percentage heartwood area to total cross-sectional area were both highly significant (Table 2.2). In the smallest size class (25 - 35 cm

DBH), heartwood area accounted for about 5% of total area (Figure 2.3). This percentage steadily increased with tree size and exceeded 40% in trees over 75 cm DBH.

When trees yielded oil in the process of drilling, oil flow commenced when the drill had gone an average of 18.7 cm into the trunk (Table 2.1, Figure 2.1). In most cases this point occurred when the drill had passed 4 - 5 cm beyond the start of the heartwood; the distance from the estimated center of the tree also varied widely and averaged 6.6 cm To efficiently capture escaping oil, the drill was withdrawn immediately. With trees that had more than one hole drilled at a time, copious oleoresin flow sometimes emerged from one or more holes (Table 2.3), so oleoresin rich sections were not apparently are all centered in the pith, regularly shaped or connected to each other. Apart from always existing in the heartwood, the unpredictability of oleoresin rich locations is reinforced by the lack of significant correlation between either its starting point, its distance from the tree center or tree diameter (Table 2.2).

Drilling measurements showed that hollows also occurred primarily in heartwood (Figure 2.1), but they may have sometimes occurred in sapwood as well (Table 2.1). When hollows were present, their volumes were quite extensive since 11 out of 14 trees with hollows that were drilled more than once during a harvest had hollows at more than one drilling location. Estimates of these hollow areas showed they occupied a mean of 55% of the heartwood area and 15% of the entire trunk area. The amount of hollow area and ratio of hollow area to total cross sectional area of the trunk were significantly correlated with tree size in trees that had hollows (Table 2.2).

OLEORESIN YIELD WITH COPAIBA TREE DRILLING

When trees were drilled there were a number of possible outcomes once the drill bit made it past the sapwood. In most cases, drilling produced no oleoresin at all (Table 2.3), and the drill frass was solid but dry. 30

In some cases, as drilling proceeded into the heartwood, the frass was highly scented and oily. Drilling further then unleashed a steady outpouring of oleoresin. Most of the oleoresin drained from the presumed cavity came out within the next hour, but depending on the size of the oleoresin rich section, the material continued to slowly drip out for another day or more. A slightly larger percentage of trees yielded a very small amount of resin (≤ 10 ml.) during one harvest, but this sometimes was not discovered until a day later.

The other common occurrence was hitting a tree with a hollow. There was only one case where even a small amount of oleoresin was collected from a tree with a hollow even though intact heartwood was usually encountered before hitting the hollow, and hollows were sometimes confined to a limited section of the tree.

The third condition found was rotten wood. Depending on its condition, these sections were either very easy to drill through or had to be penetrated in stages as crunchy sections were encountered. The drill frass from rotting sections was dry and sometimes powdery. Hollow and rotting sections did not yield oleoresin, but termites often emerged from the bore hole of trees with these conditions. Several times, drilling yielded 0.5 to

1.3 liters of water tinged with tiny bits of oil. Its red color also indicated the presence of phenolics. A small amount of water also emerged prior to the arrival of oleoresin in one knobby copaiba tree.

Another interesting phenomenon that sometimes accompanied drilling was a sucking sound that seemed to be a vacuum attempting to pull material up into the tree. This only occurred after the drill had reached the heartwood. This sound was associated with one each of the following events: before a minor oleoresin flow; after a minor oleoresin flow, no oleoresin flow, and before the drill hit a hollow. One other observation indicated that oleoresin flow in the lower part of the trunk may sometimes be connected with the canopy. Shortly after oleoresin had begun to flow out of the first hole drilled in one tree, a rain storm began.

As the wind periodically got stronger and swayed the upper branches, the resin flow rate temporarily increased as well. When the wind abated, so did the resin flow.

OLEORESIN HARVEST VARIATION BY TREE AND SITE VARIABLE

Oleoresin production among copaiba types showed that a higher percentage of white and red varieties yielded oleoresin more often and in greater amounts than the knobby variety (Table 2.3), but comparing yield in all drilled trees did not reveal significant differences. Looking just at trees that yielded some oleoresin 31

(Oleoresin Harvest Classes 2 &3) accentuates the poorer harvest of the knobby variety, but the preponderance of modest yields in all trees still fails to generate significant differences between these types. The likelihood that the white copaiba tree is a more prolific resin producer is indicated by the fact that its average total harvest and average yield per harvest are higher than both the red and knobby varieties even though the average number of times white trees were harvested and the average number of holes drilled per harvest were lower than the other two types (Table 2.3).

The size of the copaiba tree has a stronger relationship to oleoresin yield than types examined in this study. No tree smaller than 35 cm DBH yielded any oleoresin at all (Oleoresin Harvest Class 1), and no tree smaller than 45 cm DBH yielded more than 50 ml. of oleoresin in any one harvest (Oleoresin Harvest Class

3)(Table 2.4). Oleoresin yield peaked sharply in the 55 - 65 cm DBH size class with a mean total harvest of

305 ml. per tree. This size class also accounted for 63% of the trees that produced >50 ml of oleoresin during any drilling event. Trees greater than 75 cm DBH never yielded more than a trickle of oleoresin flow (Figure

2.4). A One-way AOV comparison of the mean total amounts of oleoresin harvested between size classes of all trees that were drilled had a p-value of 0.071 (F=2.17). The harvest peak in the 55 – 65 cm DBH size class can not be attributed to variation in the order of harvests or number of holes drilled per harvest since these had no significant differences between the size classes (Table 2.4). The dominance of oleoresin harvest by trees that are 45 - 75 cm DBH is reinforced when examining only trees that yielded some oleoresin (Oleoresin

Harvest Classes 2 & 3). The 55 - 65 cm DBH size class again surpasses the others with an average total yield of 566 ml. per tree. Probably due to the small number of oleoresin trees in this sample, however, the differences comparing size class to the production variables are not statistically significant.

Copaiba trees that were hollow yielded very little or no resin when they were drilled. Only 5 out of

27 (18.5%) trees drilled that had even a partial hollow yielded a small amount of oleoresin. The largest amount that came out of any hollow tree at one time was 10 ml. When the distribution of the three oleoresin harvest classes were compared between hollow and non-hollow trees, there was a significant difference (χ2 = 8.5; df=2; p=0.014). The means of the total amount of resin produced by hollow and non-hollow trees during all harvests was significant at the 94% confidence level (AOV: df=56; F=3.67; p=.061) while differences in mean amount of oleoresin per harvest were not significant. Tree size was again shown to be important when resin 32 yields were examined for hollow and non-hollow trees by size class. For the three size classes that include trees between 45 and 75 cm DBH, 53% of the non-hollow trees yielded at least 50 ml. in one harvest and another 20% yielded at least a little oleoresin at some point.

OLEORESIN PRODUCTION VARIATION WITH THE ORDER AND TIMING OF HARVEST

Oleoresin yield sometimes varied according to the order of harvest. The average oleoresin yield for the first harvest was much greater than the second amount that was slightly larger than the third harvest (Table

2.5). These differences, however, were not significant comparing all drilled trees or all trees that yielded some oil (Oleoresin Harvest Classes 2 &3). The first harvest was significantly greater than the second harvest when comparing just those trees (Oleoresin Harvest Class 3) that yielded more than 50 ml. of oleoresin in any one harvest (Mann-Whitney Test; p=.03). This difference could not be attributed to tree size since both groups had a very similar range. The difference did not carry over from the second to the third harvest when average amounts were both very low. The change from the first to second harvest was almost but not always a decrease. In 4 out of 18 oleoresin producing trees that were harvested more than once, the second harvest yielded more than the first. Only one of these four trees, however, yielded more than a very modest increase.

The average amount of oleoresin from the first harvest of trees initially harvested in the dry season

(July and August) of 1996 was smaller than those only harvested in the rainy season (May) of 1998. Since this difference was not significant, no effect of seasonal or annual differences on oleoresin production due to climatic differences can be asserted.

OLEORESIN LEAKAGE

These descriptions of oleoresin harvest may have implied that oleoresin production and storage in these copaiba trees was equivalent to the amount of oleoresin harvested, but it is important to report a few observations that showed that the amount of oleoresin produced, stored and accessible for harvest may be quite different. While oleoresin must be produced in living tree tissues, the drilling study indicated that all major oleoresin flows came from accumulations in the heartwood. The origin of the minor flows that occurred hours, 33 days or months after the initial drilling could not be identified. The leakage of oleoresin onto the trunk from several trees occurred from places other than bore holes, however, this indicated that at least some of this material produced in the sapwood can be released to the outside. First, oleoresin was observed leaking from nail wounds in three trees. Of the two trees that were drilled from this group, one yielded some resin during a harvest, and the other did not. Oleoresin stains were also found around burn wounds of two trees that did not yield any resin when harvested. Another tree was found with copious oleoresin leakage on one side of its buttress root. It was the only case where such leakage was not associated with a visible wound. This large tree had a partial hollow near the side of the root, but unlike most such trees, it did yield a small amount of oleoresin on the opposite side of the tree with heartwood that was intact as far as the drill could reach.

HARVESTING EFFICIENCY AND ECONOMICS

In the summer of 1996, ten days were devoted almost exclusively to searching for copaiba trees.

During this time, three to four men collectively found one to nine trees each day with an average of five trees per day. The lowest numbers occurred during days when forest that was occasionally flooded (“baixo”) was crossed in transit between separate areas of terra firme forest. The overall number showed that copaiba tree density was low even in its highest local concentration. The average number of trees found per day could have been considerably higher if 20 - 30 minutes had not been taken at each tree to measure various tree dimensions and site characteristics. Known trees close to Tekohaw were located quickly. Others were found in patches of terra firme forests that were up to 3.5 km from the village. Actual travel distances to the farthest trees were much longer since there were no straight line trails to those areas of the forest. When the study moved into the harvest phase, one to five trees were drilled each day giving an average of 3.3 trees drilled per day. This number was lower than would be possible doing drilling only because an average of 2.6 other trees per day were found and measured.

Time keeping of some harvesting operations showed that it usually took almost five minutes to drill through the bark and sapwood to reach the heartwood (Table 2.6). Another 20 minutes on average was spent drilling through the heartwood. In spite of its apparent greater hardness, the drilling rate going through heartwood was significantly faster than drilling through sapwood. The time needed to reach a hollow was less 34 than the time needed to reach the oil rich zone in those trees where oleoresin flowed out at the time of the drilling. This matches the shorter average distance from the outside of a tree to a hollow than to an oil rich zone (Table 2.1).

When a good oleoresin flow commenced, it took a considerable amount of time to collect it. There were two principal patterns of oleoresin flow rate out of bore holes (Figure 2.5). In trees that yielded the most resin (>350 ml from one hole), the majority of resin initially come out at a fast steady rate. Once about 93% of the oleoresin rich zone had apparently been drained, the flow rate sharply dropped off to less than 4 ml. per minute. This point was reached within half an hour in holes that ultimately yielded less than one liter, but it took an hour and a half to reach for one hole that produced 1225 ml. With trees that produced smaller amounts of resin, oleoresin slowly trickled out until it stopped.

A waterproof collection system was set up in one of the first trees tapped to capture oleoresin flow for several days. Considering the rapid drop off in flow rate within several hours even with the most copious flows, this practice was not continued since it was time consuming to construct and make repeated visits to trees that were a considerable distance from the village. This strategy for collecting oleoresin that came out quickly after drilling was necessary for research efficiency, but it may not be the best one for optimizing commercial harvest. In sum, it took about 25 minutes to drill one hole to and through the heartwood of an average tree with intact heartwood. It took an average of 16 minutes to discover if a tree was hollow by drilling if the tree was not so hollow that banging on the trunk revealed its condition. Combined drilling and waiting time for oleoresin producing trees varied from half an hour to three hours to capture more than 90% of the material from an oleoresin yielding hole.

ESTIMATES OF COPAIBA OLEORESIN HARVESTING BY AREA AND TIME – IMPLICATIONS FOR HARVESTERS

The efficiency and economic viability of harvesting oleoresin from a wild population of copaiba trees depends on a number of factors. The first is population density (Peters, 1994; Warner and Pontual, 1994).

Since this may range from 0.1 to 2 trees per ha for copaiba trees (MMA/SCA et al., 1998) at a small scale, the number of copaiba trees in an area of forest will have a direct effect on both the size of the potential resource 35 and the efficiency with which that resource may be harvested. In order to predict oleoresin yields and gross revenue on a per tree and per area basis under different conditions, a simple model was constructed using a range of minimum, medium, and maximum values from this and other studies for density, percentage of oleoresin yielding trees in the population, resin yield per tree, the number of times a tree may be harvested during its productive period and the time horizon that is used to assess the output estimates of production and value (Table 2.8).

Density estimates in the model use lower and upper limits for Copaifera trees that ter Steege and

Zondervan (2000) found in Guyana since these represent one of the best data sets from relatively large (100 ha) plot inventories of diverse forest types in the Amazon region. The medium value of 0.3 trees/ha is close to both the mean density of copaiba trees from these plots and the 0.27 trees/ha density figure for the genus in the

Brazilian state of Acre (SEBRAE, 1995). The second key factor in assessing harvest potential and designing an optimal harvesting strategy is to estimate the ratio of productive (or at least potentially productive) trees to the total number of trees by determining which size classes (if any) yielded the largest amount of product

(Peters, 1994). Both the Tekohaw and Manaus copaiba studies showed that only trees in the middle to upper middle size classes of their respective populations were usually worth harvesting (Table 2.4; Alencar, 1982) so the model calculates yields assuming that 15 – 50% of the trees in a population will be in these harvestable size classes.

The two time horizons used for the harvesting by area model are 40 years and 6 years. The longer period corresponds to the estimated amount of a time a tree would have in its prime oleoresin yielding years around Tekohaw or a similar forest. If the diameter growth rate of the trees is 0.5 cm per year, it would take about 40 years for a tree to grow from 45 cm DBH when substantial oleoresin yields first seem possible to 65 cm DBH when the probability of losing oleoresin production and/or storage potential is very high. Amounts of potential resin yield and value are, therefore, based on exploiting the full capacity of a tree during its oleoresin yielding life and estimate harvests under assumptions of minimal harvest potential (one or two harvests) and annual harvests for the entire period. The six year time-line presents a shorter time-span of predicted resin yield where the medium set of criteria allowed for a second and final harvest of productive and the maximum set of criteria allowed for three harvests that occurred on alternate years during the same period. 36

The model shows that under minimum values for overall copaiba tree density, percentage of productive trees in the population, amount of resin harvested per tree, and number of harvests possible per tree, the average yield by area would be less than 1 ml oleoresin per ha per year in both the 40 and 6 year time horizon with a value less than $US 0.01 per ha per year. For a 1000 ha area of intact forest (comparable to the area of forest within easy access of a Tembé village), these equivalent values are about $US 1.58 per year with the 40 year horizon and $US 10.50 per year with the 6 year horizon. The per year values are higher in the shorter time horizon because all of the benefits are reaped in the first and only year the trees are harvested.

Medium estimates for copaiba tree and harvest factors that either match or are slightly better than conditions found at Tekohaw generate oleoresin harvest rates of 6.8 liters per year in the 40 year projection and 45 liters per year in the six year projection from a 1000 ha area. The oleoresin sales value would range from $US 0.07 to $US 2.94 per ha per year with the two time horizons. Maximum estimates are derived from a population with moderate density and the highest proportion of productive trees that biannually yield 2.5 liters of oleoresin per harvest (an amount cited in various popular accounts as the “average” yield of oleoresin per tree). Under these optimistic circumstances, the rate of oleoresin harvest would be 0.9 liters per ha per year worth $US 6.56 per ha per year under both time horizons.

Indians or other forest dwellers traversing the woods to hunt or collect other forest products could easily harvest copaiba trees on an opportunistic basis for many years without approaching the limit of the resource. This mode, however, does not represent the potential upper range of income if the intent was to maximize copaiba oleoresin harvesting for commercial purposes. This model, therefore, assumes that the operation would be divided into a search phase and a harvest phase. In the search phase, a systematic effort would be made to locate all of the adult copaiba trees in a particular area of forest. For ease of calculation and illustrative purposes, this model uses 1000 ha as a base area that would generally be accessible within an hour’s walk from the center of a circular area. During the initial phase, trees would be quickly measured and marked, but little other time would be spent with each tree other than noting if the tree was fire damaged or appeared to be hollow. In the harvest phase, only those trees deemed likely to yield appreciable amounts of oleoresin based on their size and condition would be drilled. Beyond wasting effort, another reason to avoid drilling small trees is that doing so may interfere with normal heartwood development (Hillis, 1987) and 37 increase the chance of early pathogen attack. Each tree would be drilled twice unless the first bore hole showed the tree was hollow. Estimates assume one person does all activities. Drill times could be reduced if two people drilled at the same time, but potential revenues per person would then be reduced.

The second copaiba oleoresin harvesting model (Table 2.9) attempts to estimate the amount of oleoresin that could be harvested and the amount of income earned from selling that product per time invested depending on minimum and maximum assumptions about tree density, resin harvest per tree depending on two different harvesting strategies that are often employed. In both methods, harvesters search for trees and drill any tree two times that he believes will likely yield oleoresin. In the first harvesting strategy (Method A), a harvester collects any oleoresin that comes out of each hole until the flow has almost or completely stopped.

He then plugs the holes and moves on to look for another tree. The model assumes this procedure would be followed every day for six days per week. In the alternate strategy (Method B), the harvester prepares a device to catch oleoresin that avoids collecting rainwater and debris. After drilling trees and determining that resin flow will continue for awhile, he leaves the collection device in place and moves on to drill the next tree. The harvester looks for and drills any productive trees he finds for five days. On the sixth day, all of the trees drilled during the previous two days are revisited to collect accumulated oleoresin and plug the trees. This method makes for more efficient use of time on drilling days, but it requires extra time to return to trees to complete the harvesting.

This model shows that a harvester using Method A would gather 0.13 to 2.36 liters per day under the minimum to maximum conditions and 0.12 to 8.15 liters per day using Method B (Table 2.9). These would translate into $US 0.90 to $16.53 per day with Method A and $US 0.85 to $57.03 per day with Method B.

Under minimum conditions, all of the potential productive copaiba trees in a 1000 ha area could be drilled one time in 70 to 74 days. With higher density and yield values, it would take 230 to 794 days to complete an initial round of harvesting. 38

DISCUSSION

TREE NUMBERS, TYPES, SIZE, AND TAXONOMY

The low density of copaiba trees found in the Tekohaw study area is typical for the genus in other

Amazonian forests. Calculating precise densities of uncommon trees, however, from inventories based on plots that are 1 ha or smaller is not very precise. Ethnobotanical surveys of seven 1 ha plots near the

Tambopata Reserve in Peru found one copaiba tree (Phillips et al., 1994). Balée (1987 & 1994) did similar detailed surveys in one ha forest plots near a Tembé village not far from Tekohaw and seven other areas in the adjoining Alto Turiaçu Indigenous Reserve in Maranhão state and did not find a single copaiba tree. A census of C. pubiflora trees in 277 hectares of forest in Venezuela showed a density of 1.04 trees/ha (Ramirez and

Arroyo, 1990). Broader scale inventories of forests in Guyana showed that all species of Copaifera combined had between 1 and 174 trees per 100 ha in areas where they were found at all, resulting in an average density of

0.36 trees/ha (ter Steege and Zondervan, 2000). Although he did not conduct a systematic inventory, Alencar

(1982) found 82 C. multijuga trees in an extensive search of about 200 ha giving a rough density of 0.41 trees/ha. While the inventory of 7 ha at the Tekohaw site indicated copaiba density was 0.86 trees/ha, the figure for the area as a whole is probably smaller since an intensive search of several hundred hectares of terra firme forest only found 70 copaiba trees.

One notable feature of the copaiba populations in the Tekohaw area was that the mid-range size classes (46% in the 40 – 60 cm DBH size classes) were better represented than the smaller ones (24% in the 20

– 40 cm DBH size classes). Alencar (1982) found a similar size class distribution among C. multijuga trees near Manaus. This lack of trees in the smallest size classes could either mean that younger trees were not found as readily as older ones that were easier to identify and/or that the populations were in phases of relatively poor recruitment of younger trees. This pattern has been observed in demographic studies of other tropical trees (Clark and Clark, 1992) and does not necessarily reflect a lack of propagules. Field observations of fruiting copaiba trees near Tekohaw and Manaus (Alencar, 1981 &1984) show that trees can produce thousands of fruits at a time and that the seeds readily germinate. Like most tropical trees, however, both and 39 seedlings are heavily preyed on by insects and mammals (Alencar, 1984; Ramirez and Arroyo, 1987a, 1987b), and those that survive have special requirements for light conditions to compete with other seedlings (Alencar,

1984; Mostacedo et al., 1998; Ramirez and Arroyo, 1990; Fredericksen et al., 2000). It may just be that the conditions that most favor copaiba establishment occur at long intervals.

For young trees that become established, growth is still slow. The average copaiba tree diameter growth rate at the Tekohaw site was estimated to be 0.65 cm per year. This is greater than the average growth increment of 0.26 cm per year carefully measured in a C. pubiflora population in a drier forest region of

Venezuela (Ramirez and Arroyo, 1990). These growth rates are consistent with the finding that medium age trees (40-70 cm DBH) in the Amazon have lived an average age of 200 years or more (Vetter and Botosso,

1989). This could mean that the trees in the demographic peak of the copaiba populations around Tekohaw were established during conditions that existed more than a century ago.

The problem of identifying Copaifera trees to species in areas of high diversity is a problem both in the field and herbarium. The genus was last described forty years ago by Dwyer (1951) who commented that leaf characteristics alone could not be used to distinguish the species. This finding was echoed by legume taxonomists at both the New York Botanical Garden and the Museu Goeldi in Belém who could not identify the copaiba species from the Tekohaw study area based on leaf samples, although C. reticulata, C. langsdorffii,

C. martii, C. guianensis, and C. duckei were mentioned as possibilities (Sergio Antonio da Silva, personal communication 1998). An analysis of oleoresin samples from several of the study trees, however, did not match the chemical profiles of previously studied species (Vera Cascon, personal communication 1999). For the third time in twenty years, a Brazilian systematist is now taking up the challenge of revising Copaifera taxonomy in the region (Regina Celia, personal communication 2000). Fertile specimens would no doubt help to resolve this dilemma, but since flowering occurs on average once every two to four years (Alencar, 1984 &

1988; Fredericksen et al., 2000; Table 4), the field challenges of collecting these materials and resolving the genus’ taxonomy will require an extremely dedicated effort. 40

TREE CONDITION AND SITE PREFERENCE

The higher percentage of burned white copaiba trees may be connected to site preference since they were located primarily in areas the Tembé described as “areia” (sandy) while the red copaiba trees were found more often in areas described as “barro” (clay) soil. Soil texture measurements of samples from both types, however, showed that each had high percentages of both components (Plowden, unpublished data). Whether or not copaiba type specific site preferences exist or explain differences in their exposure to the forest fire of the 1980s, the expansion of agriculture around Tekohaw led to fire damage and loss of several copaiba study trees and forest habitat. Copaiba does have some resiliency dealing with burn damage, however, since new sprouts were often seen coming out of the lower trunk of trees that had suffered such damage.

The linkage between tree size and hollowness found in this study has been found in other groups of both temperate and tropical trees (Fearnside, 1992; Lindenmayer et al., 1993, 2000; Ball et al., 1999;

Greenberg and Simons, 1999). The occurrence of hollowness also varies by species (Fearnside, 1992;

Lindenmayer et al., 1993; Greenberg and Simons, 1999). In a study of a mixed group of trees near Manaus,

27% of trees greater than 40 cm DBH were hollow (Fearnside, 1992). This is less than the 56% comparable figure for copaiba trees in the Tekohaw study area, but these trees are apparently more resistant than another tropical legume angelim pedra (Dinizia exelsa Ducke) that is almost always hollow (Fearnside, 1992).

The high proportion of hollowness in red copaiba trees (69%) is apparently related to the large number of its trees in the largest size classes (69% were >60 cm DBH). The relatively high proportion of hollow white copaiba trees (38%) likely reflects the relatively high proportion of trees damaged by fire (28%).

In a forest area unaffected by intense fire, the proportion of hollow trees in the smaller size classes would probably be much less than found in this study area. Beside fire, other hollow trees in the study area had been hit by falling trees or were growing with their trunk abutted against another large tree. Many hollow trees had large colonies of termites living in or near them. Other studies have verified that fire, wind induced damage, successional series of fungal pathogens, termites and other insect attacks can all be agents of heartwood rot and hollow formation (Boddy and Rayner, 1981; Lindenmayer et al., 1993, 1997, & 2000; Mattheck et al., 1994;

Greenberg and Simons, 1999). Once tree rot begins, termites are efficient at excavating decayed wood

(Lindenmayer et al., 1993). 41

The only hollow trees found in the smallest size class (20 - 40 cm DBH) of copaiba trees at Tekohaw were all fire damaged, so the onset of hollowness in undamaged trees normally seems to commence in larger trees. It is interesting to note that among the three trees that had apparently been axe harvested in the past (as indicated by a circular area of scar tissue near the base of the trunk that was about 45 cm in diameter), only one of them was hollow. This tree had also been severely fire damaged. These observations showed that axe harvesting does not necessarily lead to early hollowness or mortality.

TRUNK DIMENSIONS

The importance of the relative size of copaiba’s sapwood, heartwood, and hollow sections of a trunk is that they determine the spatial limits for the tree to produce and store oleoresin. Many trees that make copious resin in the cambium quickly release stored material and may make additional amounts to protect living tissues from physical injury or attacks from other organisms (Langenheim, 1973; Phillips and Croteau,

2000). Some trees make phenolic compounds in the tree’s internal zone where senescing sapwood is undergoing its transformation to dead wood. There are a few other trees that store resins in their heartwood, but the site of their synthesis and mechanisms for transferring them to this inner zone are not know (Hillis,

1987; Langenheim, personal communication 2001). The mechanism that links production of oleoresin by

Copaifera parenchyma cells, their deposit in secretory canals, and their accumulation in the heartwood of trees at this study site is not understood. There are other types of Copaifera that apparently yield oleoresin without needing to drill as far into the trunk as the heartwood (Langenheim, personal communication 2001). While oleoresin deposits in the heartwood may stave off microbial attacks, the existence of empty hollows in damaged or older copaiba trees seems to signal an end to the tree’s ability to store substantial amounts of resin.

The finding that copaiba sapwood width is not correlated to tree size once a tree reaches a certain minimum size can be combined with tree growth rate to estimate the age at which a tree may begin to accumulate harvestable quantities of resin. If copaiba diameter increases at an average rate of 0.5 cm per year, the average sapwood width that is 14 cm represents 56 years worth of growth. Since heartwood formation was not observed in the copaiba trees less than 30 cm DBH, the initial point at which resin may begin to accumulate in the tree would take at least this long and probably longer if growth is slower in the younger 42 trees. There are wide variations between trees, but this size for the initiation of heartwood is typical of many

(Hillis, 1987).

Once heartwood formation begins, the close relationship found between heartwood size and overall tree size in these copaiba trees (Table 2.2; Figure 2.3) is common to many other species (Hillis, 1987). While overall tree diameter is very responsive to environmental conditions, the extent of heartwood formation is less affected by these influences and is more determined by genetic factors (Fries, 1998; Ericsson and Fries, 1999).

The conversion of sapwood with live cells and starch reserves to heartwood composed of fully lignified non- living tissues is a natural process in the life of a tree (Hillis, 1987). It involves the active translocation of nitrogen and other mineral nutrients from dying sapwood and its infusion with a range of extractives that include tannins, other phenolics and sometimes terpenoids (Andrews et al. 1999; Bergstrom et al., 1999). The factors that trigger these anatomical, biochemical and physiological changes in the tree’s interior are still largely unknown, but one benefit of the process is that they protect the heartwood for many years from microbial and insect attacks (Bergstrom et al., 1999; Fujita et al., 1999). Keeping heartwood intact is advantageous because trees that surpass a certain threshold of hollowness become vulnerable to breakage

(Mattheck et al., 1994; Lindenmayer et al., 1997). The red color of copaiba’s heartwood indicates it contains substantial phenolics (Hillis, 1987), and the presence of oleoresin in the tree’s interior presumably acts as an additional defense against decay causing organisms.

The key observation about the origin of substantial oleoresin flows is that it was usually well inside the outer limits of the heartwood. This indicates that whatever conditions created these oleoresin rich areas occurred close to the pith and did not exist until a tree had grown well beyond initial heartwood formation.

Since major oleoresin flows were not obtained in trees less than 50 cm DBH (Table 2.4), such trees were probably established for at least 100 years.

Like oil rich zones, tree hollows in these copaiba trees were generally located in the heartwood zone.

As found with other trees, the size of these cavities corresponded well to overall tree size (Fearnside, 1992;

Greenberg and Simons, 1999; Lindenmayer et al., 1999), although the extent to which a hollow consumed a tree’s inner volume varied considerably. Estimates of hollow area size in this study’s copaiba trees (Table 2.1;

Figure 2.3) showed that when they were present, hollows consumed an average of 15% of the total tree cross 43 sectional area and 55% of the heartwood. These are likely underestimates of the true extent of hollowness in all trees, however, since these data were based only on trees that were drilled. A number of other large obviously hollow trees (based on the “hollow” sound produced when banging on the tree) that were not drilled probably contained a higher percentage of hollow space than the areas found in the harvested trees which only revealed their partial hollows upon drilling. These results are comparable to two studies in the Amazon which showed that 20 - 30% of the stem volume of hollow trees was filled with air or light debris from termite activity (Fearnside, 1992). Greenberg and Simons (1999) found that hollows first appeared in turkey oaks

(Quercus laevis Walt.) and sand post oaks (Q. margaretta Ashe) an average of 69 and 106 years after establishment. Rot was found in these respective trees with as few as 35 and 27 growth rings. The average size or age at which hollows began in this study’s copaiba trees could not be calculated, but the smallest size at which a fire damaged tree had a hollow was found was 31.8 cm DBH, while the smallest non-fire damaged tree was 43.2 cm DBH. These trees were at least 60 - 80 years old if it is assumed they grew at an average diameter growth rate of 0.5 cm per year.

COMPARISON OF OLEORESIN HARVEST IN TEKOHAW COPAIBA TREES TO C. MULTIJUGA NEAR MANAUS

This study of several species of copaiba trees near in the eastern Brazilian Amazon had similar results in some aspects to those found by Alencar (1982) who studied oleoresin harvest of 82 C. multijuga trees in the central Brazilian Amazon near Manaus. The most important common finding was that the mean first harvest was greater than subsequent harvests (Table 2.5; Table 2.7). At Manaus the percentage of trees that yielded

>25 ml. per harvest significantly declined from the first to the fifth harvest (Pearson’s Correlation. Coefficient.

= -0.926; p=.024). As found at Tekohaw, the second harvest of several trees was larger than the first, but this effect was not repeated in subsequent harvests where the third, fourth, and fifth harvests from these trees spread about a year apart almost all went to zero. These results suggest the following scenario. The strong initial harvest resulted from draining oleoresin that was produced over many years and had accumulated in cavities in the heartwood. Once this material was removed, these cavities did not fill up again. The second round of harvesting from the same holes sometimes yielded a small amount of resin by draining off material 44 that had dripped into the bore hole from active resin canals in live tissue. This fluid may have built up as a result of normally produced material coming from severed resin ducts. It may also have been the result of enhanced oleoresin production stimulated by the drill wound. The drop off by the third and subsequent harvests might have occurred because woundwood had formed around the drill wound. This type of discolored wood whose physical and chemical properties resemble heartwood (Hillis, 1987; Shigo, 1994) could have sealed off the drill site from active resin ducts in the area. An examination of drilled trees could verify if copaiba sapwood and heartwood react similarly to injuries seen in many other trees. The drop off in third and later harvests might also have occurred if any oleoresin production response to the initial wound was only short-lived.

This phenomenon of rapid harvest drop-off is apparently not shared by all Copaifera species. Apart from many popular accounts that refer to multiple harvests without details, Cascon and Gilbert (2000) report that they removed 300 - 550 ml. of oleoresin from the same holes in ten consecutive tappings at one to four month intervals from the same C. duckei tree in Amapá state, Brazil. While they did not attempt to drain the tree during these harvesting events, the large changes in the ratio of the main oleoresin volatile components during the harvests showed that the material collected in later harvests was not the result of a large previous accumulation.

The second important common finding between the Manaus and Tekohaw copaiba studies was that the mid-size copaiba trees were the best oleoresin producers. Alencar (1982) found a positive but non- significant correlation between tree size and resin harvest, but young trees and very large trees (59 - 78 cm

DBH) yielded little if any oleoresin. It, therefore, seems that the major oleoresin producers in this population were between 40 and 60 cm DBH. This reinforces the idea that the conditions for maximizing oleoresin harvest are somehow related to tree age and condition. Oleoresin tappers of Dipterocarpus alatus Roxb. ex G.

Don in Laos also found that yields peaked in mid-size classes because young trees did not produce much, and older trees were more likely to be hollow (Ankarfjärd and Kegl, 1998). Oleoresin harvest also peaked in the

40-50 and 50-60 cm DBH size classes of Benguet pine (Pinus kesiya Royle ex Gordon) in the Philippines, but in this case the drop off in the largest 60-70 cm DBH size class was not due to hollowness (Orallo and

Veracion, 1984). This is an important finding because it shows that trees past a certain age may actually lose 45 oleoresin production capacity. Rather than viewing hollowness as the cause for the loss of oleoresin production capacity, it may be that the loss of oleoresin production capacity and its defensive properties is a natural result of aging that leads to the increasing likelihood of hollowness in older copaiba trees.

If first harvests primarily tap oleoresin that has accumulated in heartwood cavities over many years, there should be no seasonal effect on its yield. On the other hand, seasonal differences in yield seem quite possible if harvesting primarily drains oleoresin recently deposited in canals from living tissues. C. multijuga yields were higher in harvests conducted in the rainy season than the dry season, although comparison of these group means were not statistically significant (Table 2.5) and harvest order could have been responsible for the higher rainy season yields observed. My study found that first harvest oleoresin yields were higher in a group of trees drilled in the late rainy season in 1998 than ones drilled in the middle of the 1996 dry season. These differences, however, were not statistically significant. Dwyer (1951) reported that oleoresin harvests from C. venezuelana and C. pubiflora in Venezuela also peaked during the rainy season between December and April.

In contrast to these accounts of greater oleoresin harvest in the rainy season are reports that other copaiba species yield (and possibly produce) most oleoresin in the dry season. Peak harvesting of C. officinalis in Venezuela and Peru occurs in the summer months between July and November (Herrera, 1921;

Duke and Cellier, 1993) while C. reticulata trees are most often tapped from August to October (Le Cointe,

1947). Resin production has been shown to be higher in summer months in several pine species perhaps because lower water availability in this season allows a tree to devote more photosynthate to production of defensive chemicals than growth (Lorio and Sommers, 1986; Blanche et al., 1992). In copaiba’s case, peak flowering and fruit production occur during the rainy season (Alencar, 1988) so this might introduce competition for carbon resources during this period. Given the uncertainty about a seasonal effect on oleoresin harvest, the generic recommendation by some Brazilian agencies to harvest copaiba trees in the rainy season

(SEBRAE, 1995; MMA/SCA et al., 1998) seems premature.

There are a few notable differences in the results and conclusions between the Manaus and Tekohaw studies. The first is that the maximum and average oleoresin yield for first time harvest from the C. multijuga trees (3500 ml. max; 212.7 ml. ave.)(Table 2.7) were larger than obtained from the Tekohaw copaiba trees

(2028 ml. max; 72.5 ml. ave.)(Table 2.5). The most obvious possible difference is that different species of 46

Copaifera were being harvested at the two sites, but there is not enough information available yet to make such a comparison. The difference in the size class distribution between the two study areas may account for some of the greater production at Manaus. At that site, 60% of the copaiba trees were in the optimal oleoresin yielding size range while only 49% of the Tekohaw trees were in the 45 - 75 cm DBH range that included their maximum oleoresin yielding trees. Tekohaw copaiba trees that were drilled contained both a higher percentage of small trees that were presumably too young to accumulate much resin and a higher percentage of hollow trees that had lost their capacity to store the material.

One other difference that may be related to a difference in the harvest at the two study areas is the content and viscosity of the oleoresin. At the Tekohaw site the knobby and red copaiba trees both have fairly viscous oleoresins. The white copaiba trees resin was less viscous than the other two, but even it seemed less fluid than samples of “copaiba preta” (presumably C. multijuga) oil observed for sale in the Belém market. As mentioned earlier, there is a wide variation in the percentage of essential oils in copaiba oleoresin at both the intra and inter-specific levels (Sayre, 1906; Kraemer, 1907; Furnemore, 1926; Ossol and Farrar, 1947; Leung,

1980). While no research has examined a potential link between copaiba oleoresin composition and productivity, there is a known inverse relationship between the ratio of volatile to non-volatile chemicals and an oleoresin’s fluidity and viscosity (Langenheim, in press). It has been observed in other oleoresin producing trees that the volatile and other light portions of the material (principally monoterpenes and sesquiterpenes) give it the ability to flow through resin canals and deliver the heavier and potentially more toxic defense compounds (mostly diterpene resin acids) to the site of injury or attack (Langenheim, 1990; Phillips and

Croteau, 1999). Since heavier resins flow more slowly than lighter ones, it is understandable why some harvesters have set a fire at the base of some copaiba (Duke, 1986; Shanley et al., 1998) and some Dipterocarp trees in Asia (Ankarfjärd and Kegl, 1998) to increase the flow rate of some heavy resin producing species.

Many types of terpenoids can be energetically expensive for a tree to produce (Gershenzon, 1994), and the diterpenes found in copaiba might be costlier to make than the simpler sesquiterpenes. If this is true, a tree might be able to produce a higher volume of lighter weight oleoresin than a heavy one for the same investment in carbon and other plant resources. 47

Based on the results of drilling two holes into each tree, Alencar (1982) indicated that all of the available oleoresin may come out of the first hole drilled, and thus implies all of the secretory canals are connected to each other. He acknowledged this runs counter to an assertion of an earlier Brazilian authority

(Pio Corrêa, 1932) who stated that copaiba’s longitudinal secretory canals form long pockets and that the canals of some zones do not connect with those in others. Alencar seems to assume that most of the oleoresin that is tapped comes from material draining from severed active resin ducts so drilling only one bore hole is sufficient to tap this interconnected network of canals. The Tekohaw study was not designed to strictly test the effect of bore hole number on oleoresin yield, but results lend more support to Pio Corrêa’s (1932) original view. There were six trees where the first hole drilled yielded more oleoresin than the second hole, 13 trees where the second hole’s yield was greater than the first and 2 trees where both holes’ yield was the same. In the 13 cases where only one bore hole yielded any resin, the second bore hole was the source of the resin 77% of the time. Other findings stated above support the idea that initial oleoresin harvest primarily taps old accumulations in heartwood cavities, although subsequent harvests may draw from resin that has dripped out of recently severed resin ducts in the cambium and sapwood.

Alencar’s (1982) indication that none of the nine hollow trees at his site yielded any oleoresin points to another potential difference between the studies regarding the degree of compartmentalization of modified wood types in copaiba trees. The average oleoresin yield for hollow trees drilled at Tekohaw was less than 1 ml. per harvest, but 18.5% of such trees did yield a small amount of resin. Heartwood rot and hollowness may consume most of a tree’s heartwood, but at least a minor capacity to produce an oleoresin apparently sometimes remains. The importance of confirming the negative association between hollow trees and oleoresin presence, though, is that it corrects an apparent earlier misconception about this relationship. Spruce

(1970; p. 162), for example, described the mechanism leading to copaiba harvest in the following way: “In old trees the trunk becomes hollow at the core, and there the oil accumulates and is extracted by boring with an auger.”

There are differences in oleoresin harvest levels between the Manaus and Tekohaw copaiba trees, but these are negligible compared to the popular references that claim copaiba trees yield an average of 2.5 to 24 liters per harvest with maximum amounts up to 60 liters. Since these accounts are not based on systematic 48 research, it is impossible to identify specific factors that might be responsible for this large perceived gap.

Tree type, size, condition, and season have been discussed as potential influences in comparing the Manaus and Tekohaw studies. Potential differences that might result from radically different harvesting methods could also influence harvest levels. As mentioned before, axe harvesting would be the surest way to expose oleoresin cavities and canals; it would also be the most destructive. The two height harvesting scheme referenced in popular accounts and mentioned by Tembé elders deserves some consideration. It is an interesting coincidence that the upper height mentioned as the prime spot for drilling (or chopping) a second time is around the area where the heartwood to total cross-sectional area ratio is known to be greater than at the base. Depending on the species, this ratio reaches its peak some 2 - 7 meters above the ground and then tapers inward higher up

(Hillis, 1987; Stokes and Berthier, 2000). Whether this heartwood bulge exists to give the tree some special structural properties or maintain a particular sapwood width optimized for water transmission (Berthier et al.,

2001), piercing it higher up potentially accesses a greater volume of stored oleoresin. Since boring into a tree near ground level sometimes produced a sucking sound that might pull some preformed oleoresin higher up in the trunk, making a second cut higher up could conceivably break the pressure tension in the canals and more easily permit the resin to flow down to the lower hole through the force of gravity. Breaking off an upper limb as was suggested by one harvester near the Tembé reserve could have a similar effect. In this study the observation that resin flow increased from one tree when the wind was blowing hard could also be explained on this principle. A controlled experiment that compared a lower hole harvesting method to a lower and upper hole harvesting strategy would be needed to properly discern any real differences that result from these two techniques.

THE INTERNAL LIFE OF A COPAIBA TREE

The three main questions implicit in this discussion that remain to be answered are what conditions stimulate the production of oleoresin in copaiba trees, what conditions lead to the formation of resin rich cavities, and what conditions lead to the loss of oleoresin formation and storage. As mentioned earlier, anatomical evidence shows that copaiba trees produce longitudinal ducts arranged in concentric rings (Record and Mell, 1924; Langenheim, 1973; Alencar, 1982; Vetter and Botosso, 1989). Each new ring is presumably 49 created once a year in the cambium. While it has been assumed that the parenchyma cells lining these ducts regularly produce resin and deposit it into the canals, there is no direct evidence that this always occurs. The

Tekohaw study has shown that wounding a copaiba tree with a nail, a drill or a fire may lead to the outflow of a small to moderate amount of oleoresin, but in most cases these external injuries do not stimulate any resin flow at all. While insect or pathogen attacks may elicit greater resin production than purely mechanical injury in other oleoresin producing trees (Phillips and Croteau, 1999), their response to any external wound is much more consistent than copaiba’s. While copaiba oleoresin is always found in leaves (Langenheim, personal communication 2001), oleoresin formation in the trunk may be mostly made only when the tree is induced to do so under particular circumstances.

One precondition for copaiba trees having harvestable quantities of oleoresin seems to be a minimum age or size. The smallest tree in the Tekohaw study to yield any resin was 27.5 cm DBH - a tree that is over 50 years old if its diameter grew at average rate of 0.5 cm per year. Given the wide use of copaiba oleoresin as an anti-microbial medicine and the demonstration that leaf resins have anti-fungal properties (Arrhenius and

Langenheim, 1983, 1986), a fungal or bacterial pathogen is a likely candidate for the agent needed to stimulate oleoresin production in the tree. Fungal agents have been found responsible for generation of resins collected from infected Dragon tree (Draceana draco) and gharuwood trees (Aquilaria spp. and Gonystylus bancanus)(Gianno, 1984; Langenheim, in press). Part of the process might be linked to a genetically controlled stimulus.

One speculative scenario for production and storage of oleoresin in copaiba is as follows. When oleoresin is produced but not released to the outside, it may remain in the longitudinal resin ducts. As the tree grows, the older ducts formed in the cambium spend several decades in progressively senescing sapwood. At the sapwood-heartwood transition zone, additional starch resources may be mobilized to create extra amounts of oleoresin before the living cell machinery collapses. It has been shown that heartwood specific phenolic compounds have been created in this zone (Hillis, 1987), but the process for terpenoid resin deposition in the heartwood has yet to be explained (Langenheim, personal communication 2001). As a copaiba tree reaches the

50 cm DBH size (probably after 100 years of age), the oldest resin ducts near the center of the tree seem to disintegrate through lysigeny or some other process and form larger spaces. This might be caused by a specific 50 pathogen or be linked to age related erosion of inner heartwood phenolics. In Scots pine (Pinus sylvestris L.), the concentration of the stilbene pinosylvin decreases from the outer to the inner heartwood, possibly due to oxidation or polymerization of this decay resistant chemical (Bergström et al., 1999). If there is a similar loss of potency in copaiba heartwood phenolics in the oldest section, the release of oleoresin that has been shown to have anti-oxidant properties (Desmarchelier et al., 1997) could provide a second line of defense against microbial and insect attack. No defense is perfect or lasts forever, and it seems that copaiba trees in which hollows have formed have lost their oleoresin defense perhaps because newly formed oleoresin can not be channeled to these areas and older accumulations of resin have drained out through crevices of deteriorating wood. If this scenario is correct, harvesting oleoresin that has been building up in the trunk of a tree for many years that is not replenished could hasten the internal demise of the tree.

It seems possible that drilled copaiba trees that yield with copious flows of water instead of oleoresin may have been infected by some type of “wetwood” infection that surpasses or takes hold after oleoresin defenses have deteriorated. Accumulations of water have been observed in the heartwood of other trees such as Cryptomeria japonica D. Don, although its cause was unknown (Nakada et al., 1999). Extensive work on anatomical, physiological, and biochemical aspects of copaiba resin and canal formation and dissolution would need to be done to assess various aspects of this scenario.

ESTIMATES OF COPAIBA OLEORESIN HARVESTING BY AREA AND TIME – IMPLICATIONS FOR HARVESTERS

The density of copaiba trees, the average amount of oleoresin that can be harvested per tree, and the number of harvests that can be obtained from a tree are the most significant factors that will determine the viability of a commercial copaiba operation since their range of potential values may vary by 15 fold between the minimum and maximum likely values from one area to another. The results of the Manaus (Alencar, 1982) and this study show that potential harvest of oleoresin and revenue are likely to be in the minimum to medium range. At this level commercial harvesting would not be worthwhile. In a case where copaiba trees were relatively abundant and they actually could yield 2.5 liters per tree per harvest, commercial harvesting could generate two to seven times the Brazilian daily minimum wage for someone with access to a large amount of 51 intact forest. While this optimistic scenario has not yet been documented in a specific population, the readily available supply of copaiba oleoresin in various markets indicates that harvesting is commercially viable for at least some people.

The effect of fire on the health of trees and resin production for current and future cohorts is decidedly negative. The presence of burned trees translates into a higher percentage of hollow trees that lowers the ratio of productive to total trees. In areas where people hope to harvest copaiba in the future, those trees, and if possible that area, should be protected from intentional agricultural and accidental fires.

The longer term prospects for copaiba oleoresin supply may depend on how people treat forests in general. The overall density of copaiba trees and the percentage of those trees that yield oleoresin will directly decline in relation to the amount of forest that is burned. How copaiba trees are harvested can also make a large difference in the yield and health of the trees. Axe harvesting may yield the most oleoresin, but this practice is undoubtedly bad for the individual trees (Pittier, 1926; SEBRAE, 1995) and may have been responsible for a region-wide population decline of copaiba trees (Dodt, 1939 cited in Sales, 1994). The lethality of the technique probably depends on the severity of the wound and resiliency of the tree. Oleoresin has been harvested from several species of dipterocarp trees in Southeast Asia for many decades by carefully chopping holes into the trunk and setting the hole on fire (Gianno, 1984, 1990; Ankarfjärd and Kegl, 1998).

Even improper drilling of copaiba trees, however, may kill a tree within three years (SEBRAE, 1995).

The harvesting time model (Table 2.9) demonstrated that the density of copaiba trees also has a strong effect on harvest efficiency. In low density areas, the time required to find potentially productive copaiba trees dwarfs the amount of time devoted to actual harvesting whereas in high density areas, the search factor is almost irrelevant.

Although this variable was not included in the model, search efficiency also varies considerably with a prospective harvester’s ability to correctly identify copaiba trees at a distance since this determines their search rate. Elder Tembé who had a well honed search image for the subtle bark blemishes and canopy form of copaiba trees spent much less time finding them in the midst of a diverse forest than younger men who needed to carefully examine many trees up close to determine their identity. Detailed knowledge or sense about a tree’s habitat preferences also influenced search efficiency. Even though overall copaiba density was 52 low at a large scale at Tekohaw, spending more time looking carefully in promising sites and skirting forest patches where such trees were not generally found probably increased the number of trees found per day.

A comparison of the wait and collect oleoresin resin harvesting strategy (Method A) versus the come back and collect oleoresin strategy (Method B) in the harvesting time model (Table 2.9) showed that these strategies have similar daily harvesting rates under the least favorable circumstances. When trees are more abundant and yield moderate to large amounts of oleoresin, however, it becomes increasingly advantageous to spend the extra time setting up a weather resistant collection device and return later to collect the accumulated material. The advantage of this method is indicated by a greater harvest efficiency (Trees and Resin Harvested per Week), financial return for time invested ($/Day), and the total amount of time that would be needed to harvest all of the trees in a given area (Days to Harvest Area). The best strategy is probably a common sense hybrid of the two methods. If the initial resin flow was weak, it would be worth waiting to collect most of the oleoresin, plug the tree and not devote further time to a follow-up visit. If the initial flow was strong and seemed likely to continue for an hour or longer, setting up a collection device would probably pay off because it would avoid wasting time waiting and allow collection of all the oleoresin - not just the majority that would emerge in several hours.

Where only a few people harvested copaiba in a large area of forest, this activity would be as remunerative as collecting many other forest products. As was seen in the area based model, though, the potential revenue from copaiba harvesting would yield very poor economic returns to harvesters unless conditions were at least as good as described in the medium set of criteria.

COPAIBA OLEORESIN MARKETING

As is true with most extractive products, the middle men fare considerably better than the collectors

(Pendleton, 1992). Small-scale merchants who sell copaiba oleoresin from medicinal plant shops in Belém receive $R 15 - 40 per for a full liter bottle - two to six times more than the $R 7 they normally pay the harvester. These vendors usually sell the liquid in small 20 ml. plastic containers for $R 2 which means the resale price is $R 100 per liter. Companies that purchase larger quantities of copaiba oleoresin sometimes filter the oleoresin to resell some in small containers for $US 2.80 (equivalent to $US 140 - 188 per liter). 53

Most of the product is distilled so the resin and purified oil can be sold separately. The resin is generally packaged in 5 kg sacks whose minimum price is $US 2.60 per kg. The purified essential oil is sold to industrial users in 200 liter barrels for a minimum of $US 10 per liter or packaged in bottles with 50 gel capsules each containing 500mg of oil that sell for $US 12 each - equivalent to $US 480 per liter (MMA/SCA et al., 1998). There are many equipment and labor costs associated with these operations (SEBRAE, 1995;

MMA/SCA et al., 1998), but it is clear that the profit margin of corporate copaiba sellers is still much higher than the collector’s.

Proposals to attract investment in a small copaiba oleoresin processing and packaging factory in a state such as Acre predict that the project could generate a net revenue of about 25% and pay back the initial investment in less than two years (SEBRAE, 1995; MMA/SCA et al., 1998). These proposals are based on the input of 4.4 tons (about 5000 liters) of raw copaiba oleoresin per year. This would be processed and sold as

150 sacks of resin, 3000 liters of purified oil in barrels and 25,000 bottles of capsules. Considering reports that the state of Amazonas annually produces more than 100 tons and Rondônia production had reached 30 tons per year by 1994 (MMA/SCA et al., 1998), this does not seem like an overly ambitious goal. This analysis, however, is based on the unproven assumption that copaiba trees can yield an average of 2.5 liters of oleoresin per year for many years (SEBRAE, 1995) so the production target could theoretically be met by 2000 trees.

Since the proposal recommends that trees are harvested every other year, it really means that 1000 trees would need to be harvested every year with an average yield of 5 liters per tree each time they are tapped. In Acre, copaiba density is believed to be about 0.27 trees per hectare (SEBRAE, 1995) so the operation would need to tap every copaiba tree in a 7407 ha area of forest. If the patterns of oleoresin harvest found at the Manaus

(Alencar, 1982) and the Tekohaw study hold true in Acre, only 25-50% of the trees will yield appreciable amounts of oleoresin, the average yield from these trees will be much less than predicted even for first time harvests, and the average yields will drop dramatically after the initial harvest. If these less favorable conditions prevail, an area 10 to 1000 times greater would be needed to sustain the oleoresin production goals for this one factory. Production studies will be conducted with copaiba trees in Extractive Reserves in Acre to verify whether the optimistic yields included in the factory proposal are justified (Daisy A. P.G. Silva, personal communication 2001). 54

WILL PLANTATIONS BE THE MAIN SOURCE OF COPAIBA IN THE FUTURE?

Studies in other copaiba rich areas in Brazil (Pará, Amazonas, and Rondônia), Venezuela and other countries could help broaden the understanding of the true oleoresin production potential of wild populations.

Alencar (1982) concluded that the wild population of C. multijuga could supply limited amounts of oleoresin for medicinal purposes, but that production levels were insufficient to consider its use as an alternative source of energy to fossil fuels. Early projections were that one acre (0.4 ha) of 100 mature copaiba trees could produce 25 barrels – about 52 liters per tree per year (Wang and Huffman, 1981). It was, therefore, logical that experimental plantations of C. multijuga were initiated in the early 1980s to test its potential for oleoresin, pulp and timber production (MMA/SCA et al., 1998). Related work has shown that copaiba seeds have good germination rates (Alencar, 1981) and have good survival in a plantation setting. After 16 years, trees had reached an average size of 9.2 cm DBH (MMA/SCA et al., 1998). This represents an annual diameter growth rate of 0.575 cm per year. This rate for juveniles growing in full sun is not surprisingly faster than the average equivalent growth rate 0.2595 cm per year for all size classes of a wild population of C. pubiflora (Ramirez and

Arroyo, 1990). Even if the plantation growth rate is steadily maintained, though, it will take 70 - 87 years for these trees to reach the 40 - 50 cm DBH range - the likely minimum size that they could potentially yield substantial amounts of oleoresin. Trees were planted at a 2.5 x 5 meter spacing (MMA/SCA et al., 1998) giving a density of 800 trees per hectare. This plantation will inevitably be thinned over time so if survivors achieve the same crown area as their counterparts in the wild population of C. multijuga (145 square meters at

45 cm DBH - based on data in Alencar, 1982), they will be reduced to a density of 69 trees per hectare. This higher density could dramatically improve harvesting efficiency, but it will take until the middle of this century to find out if a monoculture plantation of these trees will produce more or less oleoresin than their non- cultivated relatives that have been exposed to the full array of biotic influences in a diverse forest.

Even if oleoresin yields are similar or better than found in wild populations, it seems unlikely that the price for copaiba oil will be sufficiently high to economically justify more than 70 years of investment in a copaiba plantation intended solely for oleoresin production. The experimental plantation could, however, serve as a field site for certain types of controlled experimental research that would be difficult to undertake in genetically diverse and dispersed wild populations. While trees are growing it could be worth examining the 55 effect of etherel (based on the phytohormone ethylene) on copaiba trees since it has been shown to stimulate heartwood formation and increase production of various exudates in other trees (Hillis, 1987). Other chemicals and specialized wounding techniques have been specifically used to stimulate resin production in a variety of conifer and hardwood trees (Howes, 1949; Anderson, 1955; Hillis, 1987, Langenheim, in press). Trees that are slated for thinning could also be used to carefully assess the extent that drilling induces wound wood, inhibits heartwood formation and otherwise affects future oleoresin harvest. Other variations on harvesting techniques, such as the high and low drilling scheme, could be tested under more controlled circumstances than would be possible with a wild population.

SUMMARY AND CONCLUSIONS

This study has verified many of the results of Alencar (1982) that wild populations of copaiba trees do not all have the oleoresin yielding capacity that is often described in popular accounts. It has become clearer that trees that are too young, too old, or hollow at any age do not yield large amounts of oleoresin. There may be one set of circumstances that induce some copaiba trees to produce above average amounts of oleoresin in its live tissues during middle age and a different set of circumstances that lead to the creation of resin-filled cavities in the heartwood later in life. While trees in the upper middle size classes are likely to yield the most oleoresin, there is no certainty that all such trees will do so. This study also reinforces the Alencar (1982) study’s result that the first harvest of a copaiba tree is likely to yield more resin than any successive one. This finding, likewise, is contrary to many popular accounts that assert trees can be tapped many times without declining yields. The combination of low density, limited productive trees in a population, low yield per tree, and limited number of potential harvests found in the Tekohaw and Manaus studies lead to very low estimates for potential oleoresin harvest and financial return on a per area basis and per time basis for a harvester. These contrast with proposals that indicate a small factory that processes and repackages copaiba oleoresin for sale in various forms can be reliably supplied by a typical Amazonian terra firme forest and operate with a healthy profit margin. While this study has made some interesting findings about the biological and ecological parameters of copaiba oleoresin production and harvest in one area of the eastern Brazilian Amazon that is rich in Copaifera diversity, additional studies on the physiology of oleoresin production and harvesting protocols 56 will be needed before its findings can be generalized or shown to be anomalies in relation to populations of other Copaifera species in different parts of the Amazon.

Where copaiba trees are found to only yield one or two good oleoresin harvests, local income could be enhanced by culling older harvested trees. Copaiba trees are widely valued for their timber (Chichignoud et al., 1990; Chudnoff, 1984; Martini et al., 1994), but this double use of copaiba trees would need to be done in concert with a well-managed operation that preserved sufficient fruiting trees to avoid negative impacts on its population and other serious pitfalls that frequently accompany logging in the region (Uhl et al., 1991, 1997;

Verissimo et al., 1992; Johns et al., 1996; Barreto et al., 1998). Resin harvest in young trees has been successfully integrated with timber harvest of older trees with longleaf pine (Pinus palustris) and slash pine (P. elliottii) in North America and several dipterocarp species in Southeast Asia (Langenheim, in press).

Copaiba trees that have passed their oleoresin producing prime, however, should not necessarily be thought of as trees that have lost their value to the forest or its people. A variety of invertebrates utilize the small amounts of resin that sometimes ooze onto copaiba bark from natural causes or human related activities.

Several species of Euglossa and Tetragona stingless bees were seen flying close to fresh oleoresin on a copaiba tree trunk. They presumably collected some of the resin for nest building or other purposes (Roubik, 1989;

Nogueiro-Neto, 1997). Adults and several nymphal stages of at least one species of assassin bug

(Apiomerinae: Reduviidae: Hemiptera) were often seen hiding in bark crevices near resin leaks with bits of resin on their forelegs - presumably waiting to capture a bee. This behavior of applying sticky resin to raptorial legs to aid in the capture of stingless bees has been observed in several genera of this group in Costa

Rica and Brazil (Johnson, 1983; Adis, 1984). The Tembé are very familiar with this assassin bug and give it the more distinguished titles “maranu’yw” (Tembé name for a forest spirit) and “don de copaiba” (Portuguese phrase meaning owner or caretaker of the copaiba tree).

Even fire damaged and the oldest trees at Tekohaw and Manaus (Alencar, 1984, 1988) that yielded little or no oleoresin continued to reproduce for many years. Beyond collecting resin, stingless bees also visit copaiba trees to gather pollen from its flowers (Crestana and Kageyama, 1989), and aril coated seeds are consumed by a variety of insects (Ramirez and Arroyo, 1987a, 1987b; Leal and Oliveira, 1998), birds

(Plowden, in press), and mammals (Alencar, 1984). Tembé hunters are well aware of these connections since 57 they searched for toucans in the canopy of fruiting copaiba trees and built “esperas” (an impromptu tree perch made with stout young trees lashed several meters above the ground between two trees) to wait for game animals such as agoutis, pacas, and deer that feed on fallen copaiba fruits at night. This resource is particularly important in the dry season when fewer forest fruits are typically available (Alencar, 1988; Lourival Tembé, personal communication 1996). Even non-reproducing copaiba trees may also have special value to wildlife.

Di Bitetti et al. (2000) showed that C. langsdorffii was one of four species of large trees frequently used as a sleeping site by a group of capuchin monkeys (Cebus apella nigritus). The copaiba and other preferred trees were all tall emergents with large crown diameters and many horizontal branches and forks that provided stable night-time resting places.

The economic prospects of copaiba oleoresin harvesting may not always be attractive in many situations, but it would be inadvisable to use purely economic rationale to argue for their preservation or maintaining the forests that contain them. Copaiba oleoresin will be harvested and used by indigenous and other forest dwellers for their own use whether or not it is worthwhile to sell. The Tembé in particular bemoan the loss of copaiba and other trees valued by their culture to vast areas of forest burned by non-indians that have occupied the Alto Rio Guamá reserve. On the other hand, the desire for sustainable economic development among forest peoples is quite real, so these aspirations will probably need to be met through means other than reversion to the intensive and probably predatory exploitation of copaiba and other similarly scarce extractive forest resources.

58

TABLE 2.1 COPAIBA TRUNK DIMENSIONS OF ALL DRILLED, OLEORESIN YIELDING AND HOLLOW TREES AT TEKOHAW

DIMENSIONS N (#Trees) MEAN ± S.E. (Mean Dia. Drill Height* (cm)) ALL DRILLED TREES Distance to Heartwood (cm) 47 (58.3) 14.2 ± 0.4 Heartwood Diameter (cm) 47 (58.3) 30.0 ± 2.4 Heartwood Area/Total Area % 47 (58.3) 26.0 ± 2.2 OLEORESIN YIELDING TREES Distance to Oil (cm) 8 (57.4) 21.3 ± 2.0 Oil Zone Radius (cm) 8 (57.4) 7.2 ± 2.0 DRILLED HOLLOW TREES Distance to Hollow (cm) 17 (67.3) 18.7 ± 1.8 Hollow Radius (outer edge)(cm) 17 (67.3) 15.0 ± 2.4 Hollow Area/Heartwood Area % 13 (71.4) 55.1 ± 8.6 Hollow Area/Total Area % 13 (71.4) 15.0 ± 2.9 Hollow area estimation method: Where end point of hollow during drilling is on near side of calculated trunk center, hollow area calculated as if beginning and end of hollow represent inner and outer points of a hollow ring in entire cross sectional area of trunk. Where end point of hollow during drilling is on far side of calculated trunk center, hollow area calculated as a circle with a radius based on the average of the beginning and end hollow points. *Dia. Drill Height: Average diameter of tree at heights where it was drilled. Drill height ranged from 0.8 to 1.2 meters above ground level. Due to tapering of the trunk, these diameters were larger than diameters measured at 1.5 meters (DBH). They are included here for reference since dimension distances were also measured at drill height. 59

TABLE 2.2 CORRELATION BETWEEN COPAIBA TREE DIAMETER AND OTHER TRUNK DIMENSIONS AT TEKOHAW STUDY SITE

DIMENSION N (# Trees) Corr. Coeff.1 R-Sq. (%)2 p-value Distance to Heartwood 47 .198 3.9 .181 Heartwood Diameter 47 .933 87.0 .000 Heartwood/Total Area % 47 .755 57.0 .000 Distance to Oil 8 .278 7.7 .505 Oil Zone Radius 8 .245 6.0 .559 Distance to Hollow 17 .363 13.2 .152 Hollow Radius (outer edge) 17 .709 50.3 .001 Hollow Width 13 .691 47.8 .009 Hollow Area 13 .746 55.7 .003 Hollow/Heartwood Area % 13 .202 4.1 .507 Hollow/Total area % 13 .568 32.2 .043 Notes: 1. Pearson’s Correlation Coefficient; 2. Linear regression R-squared value. Correlations compare trunk dimension with tree diameter measured at average drill height (approximately 1 meter above ground).

60

TABLE 2.3 HARVEST OF OLEORESIN FROM COPAIBA TREES AT TEKHOHAW STUDY SITE BY TREE TYPE

WHITE KNOBBY RED All OLEORESIN HARVEST CLASS N = 34 N = 12 N = 11 N = 57 Class 1 (0 ml) 23 (67.7%) 10 (83.3%) 6 (54.6%) 39 (68.4%) Class 2 (1-49 ml max in one harvest) 6 (17.7%) 1 (8.3%) 3 (27.3%) 10 (17.5%) Class 3 (>50 ml max in one harvest) 5 (14.7%) 1 (8.3%) 2 (18.2%) 8 (14.0%) SUMMARY HARVEST DATA (Mean ± S.E.) ALL TREES N = 34 N = 12 N = 11 N = 57 Ave. Total Harvest (ml) 103.3 ± 64.3 22.9 ± 22.5 65.0 ± 43.5 79.1 ± 39.5 Ave. Amount per Harvest (ml) 84.2 ± 60.5 7.7 ± 7.5 47.9 ± 39.1 61.2 ± 36.9 Ave. # Harvests 1.9 ± 0.2 1.8 ± 0.2 2.2 ± 0.2 2.0 ± 0.1 Ave. # Holes per Harvest 1.9 ± 0.1 2.0 ± 0.2 2.1 ± 0.3 2.0 ± 0.1 Ave. # Holes per Harvest with Oil 0.5 ± 0.1 0.2 ± 0.1 0.8 ± 0.3 0.5 ± 0.1 CLASS 2&3 TREES ONLY N = 11 N = 2 N = 5 N = 18 Ave. Total Harvest (ml) 319 ± 188 138.0 ± 133.0 143.0 ± 87.3 250 ± 117 Ave. Amount per Harvest (ml) 260 ± 181 46.2 ± 43.7 105.4 ± 83.0 194 ± 113 Ave. # Harvests 2.0 ± 0.3 2.4 ± 0.5 2.2 ± 0.4 2.2 ± 0.2 Ave. # Holes per Harvest 1.9 ± 0.1 2.5 ± 0.5 2.4 ± 0.4 2.2 ± 0.1 Ave. # Holes per Harvest with Oil 1.5 ± 0.2 1.0 (no S.E.) 1.8 ± 0.2 1.5 ± 0.1

TABLE 2.4 HARVEST OF OLEORESIN FROM COPAIBA TREES AT TEKOHAW STUDY SITE BY SIZE CLASS

DBH (cm) SIZE CLASS 25-35 35-45 45-55 55-65 65-75 >75 All OLEORESIN HARVEST CLASS N = 6 N = 13 N = 11 N = 13 N = 4 N = 10 N = 57 (Number and % of Size Class) Class 1 (0 ml in every harvest) 4 (10.3%) 12 (30.8%) 7 (18.0%) 6 (15.4%) 2 (5.1%) 8 (20.5%) 35 Class 2 (1-49 ml max in one harvest) 2 (20.0%) 1 (10.0%) 2 (20.0) 2 (20.0%) 1 (10.0%) 2 (20.0) 10 Class 3 (>50 ml max in one harvest) 0 (0.0%) 0 (0.0%) 2 (25.0%) 5 (62.5%) 1 (12.5%) 0 (0.0%) 8 SUMMARY HARVEST DATA (Mean ± S.E.) ALL TREES N = 6 N = 13 N = 11 N = 13 N = 4 N = 10 N = 57 Ave. Total Harvest (ml) 2.7 ± 2.0 3.4 ± 3.4 17.4 ± 11.6 305 ± 161 71.0 ± 66.4 0.9 ± 0.6 79.1 ± 39.5 Ave. Amount per Harvest (ml) 0.9 ± 0.7 1.1 ± 1.1 13.8 ± 11.0 247 ± 155 23.7 ± 22.1 0.5 ± 0.3 61.2 ± 36.9 Ave. # Harvests 2.3 ± 0.4 1.9 ± 0.3 1.6 ± 0.3 2.2 ± 0.2 2.3 ± 0.5 1.8 ± 0.2 2.0 ± 0.1 Ave. # Holes per Harvest 1.7 ± 0.2 1.7 ± 0.1 2.0 ± 0.1 2.0 ± 0.2 2.2 ± 0.6 2.1 ± 0.2 2.0 ± 0.1 Ave. # Holes per Harvest with Oil 0.5 ± 0.3 0.1 ± 0.1 0.5 ± 0.2 1.0 ± 0.3 0.8 ± 0.5 0.2 ± 0.1 0.5 ± 0.1 CLASS 2&3 TREES ONLY N = 2 N = 1 N = 4 N = 7 N = 2 N = 2 N = 18 Ave. Total Harvest (ml) 8.0 ± 4.0 44.0 (no S.E.) 47.7 ± 27.6 566 ± 267 142.0 ± 128.0 4.5 ± 0.5 250 ± 117 Ave. Amount per Harvest (ml) 2.7 ± 1.4 14.7 (no S.E.) 38.0 ± 28.3 459 ± 270 47.4 ± 42.6 2.3 ± 0.3 194 ± 113 Ave. # Harvests 3.0 ± 0.0 3.0 (no S.E.) 1.8 ± 0.5 1.7 ± 0.4 3.0 ± 0.0 2.0 ± 0.0 2.2 ± 0.2 Ave. # Holes per Harvest 2.0 ± 0.0 1.3 (no S.E.) 1.9 ± 0.1 2.0 ± 0.0 2.9 ± 1.2 2.5 ± 0.5 2.1 ± 0.1 Ave. # Holes per Harvest with Oil 1.5 ± 0.5 1.0 (no S.E.) 1.3 ± 0.3 1.9 ± 0.3 1.5 ± 0.5 1.0 ± 0.0 1.5 ± 0.2

62

TABLE 2.5 CHANGES IN OLEORESIN HARVEST FROM COPAIBA TREES AT TEKOHAW STUDY SITE BY YEAR AND HARVESTING ORDER

N MEAN ± S.E. MEDIAN MINIMUM MAXIMUM ALL DRILLED TREES FIRST HARVEST 57 72.5 ± 39.3 0 0 2028 (All Trees)(ml) 1996 33 35.1 ± 26.4 0 0 850 1998 24 230 ± 118 0 0 2028 FIRST HARVEST 18 230 ± 118 5.0 0 2028 (Oil Yielding Trees)(ml) 1996 12 96.5 ± 71.1 4.0 0 850 1998 6 496 ± 315 250 2 2028 TREES DRILLED ≥2 TIMES (Oil Yielding Trees) ALL HARVESTS BY ORDER (ml) Harvest 1 12 96.5 ± 71.1 4.0 0 850 Harvest 2 12 19.0 ± 12.1 3.0 0 146 Harvest 3 9 16.3 ± 12.5 2.0 0 114 AMOUNT CHANGE BY Maximum Maximum HARVEST (ml) Reduction Increase Harvest 1 to 2 12 -77.5 ± 74.0 -1.0 -850 144 Harvest 2 to 3 8 -8.6 ± 6.0 -4.5 -44 9

63

TABLE 2.6 DRILLING TIMES IN COPAIBA TREES AT TEKOHAW

DRILLING VARIABLE N (#TREES) MEAN ± S.E. MINIMUM MAXIMUM Time to Reach Heartwood (min./hole) 20 4.70 ± 0.53 1.0 10.0 Time Drilling Heartwood (min./hole) 11 20.05 ± 3.47 5.0 36.0 Time to Reach Hollow (min./hole) 4 11.38 ± 2.45 5.0 16.5 Time to Reach Oleoresin (min./hole) 4 16.12 ± 4.14 10.0 28.0 Time of Oleoresin Flow (min./hole) 5 90.80 ± 34.3 11.5 205.0 Drill Rate in Bark/Sapwood (cm/min.) 17 2.75 ± 0.47 0.0 6.9 Drill Rate in Heartwood (cm/min.) 11 1.17 ± 0.27 0.24 3.5

64

TABLE 2.7 OLEORESIN PRODUCTION FROM C. MULTIJUGA NEAR MANAUS, BRAZIL

HARVEST NUMBER 1 2 3 4 5 HARVEST PERIOD Mar.-Jun. Dec.-Jan. ‘78 Sep.-Nov. ‘78 Sep.-Nov. ‘79 Dec. ‘80 SEASON ‘77 early rainy late dry late dry early rainy AVERAGE RAINFALL mid rainy 188 ml./mo. 75 ml./mo. 75 ml./mo. 188 ml./mo. 288 ml./mo. OLEORESIN HARVEST CLASS N(%) No Oil 28 (34.2) 49 (58.8) 55 (67.1) 59 (72.0) 58 (70.7) Small Oil (<25 ml.) 12 (14.6) * * * * Large Oil (>25 ml.) 28 (34.2) 19 (23.2) 13 (15.9) 9 (11.0) 10 (12.2) Hollow 14 (17.1) 14 (17.1) 14 (17.1) 14 (17.1) 14 (17.1) OLEORESIN HARVEST MEAN (ml./tree) All Trees (n=82) 212.7 119.3 42.0 34.3 34.3 Non Hollow Trees (n=73) 239.0 134.0 47.2 48.7 38.5 Notes: Data from Alencar (1982) Tables 2 & 3, Figure 11)(N=82 Trees) * - Small oleoresin harvest not noted beyond first harvest.

TABLE 2.8 COPAIBA OLEORESIN HARVESTING MODEL BY AREA FOR LONG- TERM AND SHORT-TERM PROJECTIONS

VARIABLE Min. Estimate Med. Estimate Max. Estimate Density (trees/ha) 0.3 0.7 1.5 Total Trees/Area (1000 ha) 300 700 1500 Productive/Total trees 0.15 0.30 0.50 Productive Trees/Area 45 210 750 Resin/Harvest/Tree (l.) 0.2 1 2.5 40 Year Projection (Productive Life of Tree) Number of Harvests in 40 Years 1 2 20 Total Resin (l.)/Tree 0.20 2.00 50.00 Resin (l.)/Tree/Year 0.005 0.05 1.25 $/Tree $ 1.40 $ 14.00 $ 350.00 $/Tree/Year $ 0.04 $ 0.35 $ 8.75 Resin (l.)/ha 0.009 0.420 37.5 Resin (l.)/ha/Year 0.000 0.011 0.938 $/ha $ 0.06 $ 2.94 $ 262.50 $/ha/Year $ 0.002 $ 0.074 $ 6.563 Resin (l.)/Area (1000 ha) 9 420 37500 Resin (l.)/Area (1000 ha)/Year 0.2 10.5 937.5 $/Area (1000 ha) $ 63.00 $ 2940.00 $ 262500.00 $/Area (1000 ha)/Year $ 1.58 $ 73.50 $ 6562.00 6 Year Projection Number of Harvests in 6 Years 1 2 3 Total Resin (l.)/Tree 0.20 2.00 7.50 Resin (l.)/Tree/Year 0.03 0.33 1.25 $/Tree $ 1.40 $ 14.00 $ 52.50 $/Tree/Year $ 0.23 $ 2.33 $ 8.75 Resin (l.)/ha 0.009 0.420 5.625 Resin (l.)/ha/Year 0.000 0.011 0.938 $/ha $ 0.06 $ 2.94 $ 39.38 $/ha/Year $ 0.01 $ 0.49 $ 6.56 Resin (l.)/Area (1000 ha) 9 420 5625 Resin (l.)/Area (1000 ha)/Year 1.5 70.0 937.5 $/Area (1000 ha) $ 63.00 $ 2940.00 $ 39375.00 $/Area (1000 ha)/Year $ 10.50 $ 490.00 $ 6562.50 Notes: All variables in bold are estimates of values based on data from this study and the literature. Other variables are calculations based on those estimates. Medium estimates are based either on typical value in the literature or an average value between minimum and maximum values. Area estimates are based on 1000 hectares (ha). Resin estimates are in liters. Value estimates are based on Brazilian reais ($R) with $7/liter as base price paid to harvesters in 1998. Equivalent value in $US at the time was approximately $R 1 = $US 1. Multi-year projections of values are not calculated with a discount rate or adjusted for inflation. 66

TABLE 2.9 COPAIBA OLEORESIN HARVESTING MODEL FOR TWO COLLECTION METHODS AT FIRST-TIME HARVEST

VARIABLE Min. Estimate Med. Estimate Max. Estimate VARIABLES FOR BOTH METHODS Density (trees/ha) 0.3 0.5 1.5 Productive tree/total tree 0.15 0.30 0.50 Productive trees/ha 0.045 0.21 0.75 Square meters searched/min 500 500 500 Search Time/Productive Tree (min) 444 95 27 Resin/Harvest/Tree (l) 0.2 1.0 2.5 Drill Time/Tree (min) 40 40 40 Site Transit Time/Day (min) 90 90 90 METHOD A (Wait for Resin Flow) Resin Flow Time/Tree (min) 32 110 256 Drill & Flow Time/Tree (min) 72 150 296 Search, Drill, & Flow Time/Tree (min) 516 340 349 Search, Drill & Flow Time/Day (min) 330 330 330 Trees Harvested/Day 0.64 0.97 0.94 Resin Harvested/Day 0.13 0.97 2.36 Trees Harvested/Week 3.8 5.8 5.7 Resin Harvested/Week 0.8 5.8 14.2 Search & Harvest Days/Week 6.0 6.0 6.0 METHOD B (Return for Resin) Monitor Resin Flow Time/Tree (min) 20 20 20 Drill & Monitor Flow Time/Tree (min) 60 60 60 Search, Drill & Monitor Time/Tree (min) 504 155 87 Search, Drill & Monitor Time/Day (min) 330 330 330 Trees Drilled & Monitored/Day (min) 0.65 2.13 3.79 Trees Drilled/Week 3.3 10.6 19.0 Collect Resin Time/Tree (min) 10 10 10 Inter-tree Transit Time (min)* 16 7 4 Resin Collection & Transit Time/Tree (min) 26 17 14 Trees Collected/Day 3.3 10.6 19.0 Trees Harvested/Week 3.3 10.6 19.0 Resin Harvested/Week (l) 0.7 10.6 47.6 Search & Harvest Days/Week 5.4 5.7 5.8 Resin Harvested/Day (l) 0.12 1.88 8.15 SUMMARY COMPARISON OF METHODS $/Day - Method A $ 0.90 $ 6.79 $ 16.53 $/Day - Method B $ 0.85 $ 13.17 $ 57.03 Days to Harvest 1000 ha - Method A 70 216 794 Days to Harvest 1000 ha - Method B 74 112 230 Notes: All variables in bold are estimates of values based on data from this study and the literature. Other variables are calculations based on those estimates. One day is defined as 5.5 hours of work in the field plus 1.5 hours for travel between the forest and the village. One week is defined as six days of work. Both methods invole drilling any potentially productive trees found during the search process. All “productive” trees are drilled two times. Method A harvesters wait to capture resin flow before resuming search. Resin flow time is based on the equation: Flow time (min)=12.2369 + 97.5 * liters oleoresin. Method B harvesters search and drill potentially productive trees for five days. They monitor initial resin flow and set up containers to capture strong flow. They revisit all productive trees on the sixth day to collect accumulated oleoresin. Inter-tree transit time based on average density of productive trees and walking speed of 0.5 m/sec in the forest. Value estimates are based on $R 7.00/liter paid to harvesters in 1998. At the time, $R 1 approximately equaled $US 1. 67

FIGURE 2.1 CROSS-SECTION ILLUSTRATION OF SOLID, OLEORESIN BEARING AND HOLLOW COPAIBA TREES 68

100 Hollow Only Hollow and 75 Fire Damaged Fire 50 Damaged Only

25

0 20 - 40 40 - 60 60 - 80 > 80 (n = 17) (n = 31) (n = 17) (n = 4) PERCENT CONDITION IN SIZE CLASS PERCENT CONDITIONSIZE IN DIAMETER SIZE CLASS (DBH cm.) (n = total # trees in size class)

FIGURE 2.2 PERCENT OF HOLLOW AND FIRE DAMAGED COPAIBA TREES AT TEKOHAW BY SIZE CLASS 69

(Measurements only from Drilled Trees) Hollow Area 40 (n=13 trees)

Heartwood 30 Area (n=47 trees)

20

10

0 % TRUNK AREA AT DRILL HEIGHT DRILL AT AREA % TRUNK 25 - 35 35 - 45 45 - 55 55 - 65 65 - 75 > 75 DIAMETER SIZE CLASS (DBH cm)

FIGURE 2.3 HEARTWOOD AND HOLLOW PROPORTION OF TRUNK AREA IN COPAIBA TREES AT TEKOHAW SITE 70

Average Average 1000 AmountAmount per per HarvestHarvest 750 Amount of First Harvest 500

250

0 MEAN OLEORESIN HARVEST (ml) HARVEST OLEORESIN MEAN 25 - 35 35 - 45 45 - 55 55 - 65 65 - 75 > 75 DIAMETER SIZE CLASS (DBH cm) (For Trees with >0 ml. Harvest; n=18)

FIGURE 2.4 OLEORESIN HARVEST FROM COPAIBA TREES AT TEKOHAW BY SIZE CLASS 71

Tree #(Hole #) 1400 15(2) 1200 58(2) 66(1) 1000 66(2) 67(1) 800 67(2)

600

400

200

AMOUNT OF OLEORESIN (ml) OLEORESIN OF AMOUNT 0 0 25 50 75 100 125 150 175 200 TIME OF OLEORESIN FLOW (min.)

FIGURE 2.5 PATTERNS OF OLEORESIN FLOW FROM COPAIBA TREES AT TEKOHAW 72

REFERENCES

Adis, Joachim. 1984. Eco-entomological observations from the Amazon. V. Feeding habits of Neotropical “bee killers” and “resin bugs” (Apiomerinae: Reduviidae: Hemiptera). Revista Biologica Tropical 32(1):151-153.

Alencar, Jurandyr da Cruz. 1981. “Estudos silviculturais de uma população natural de Copaifera multijuga Hayne (Leguminosae) na Amazônia Central. I. Germinação. Acta Amazonica 11(1): 3-11.

Alencar, Jurandyr da Cruz. 1982. Estudos silviculturais de uma população natural de Copaifera multijuga Hayne (Leguminosae) na Amazônia Central. 2. Produção de óleo-resina. Acta Amazonica 12(1): 75- 89.

Alencar, Jurandyr da Cruz. 1984. Estudos Silviculturais de uma população natural de Copaifera multijuga Hayne (Leguminosae) na Amazônia Central 3. Distribuição espacial da regeneração natural pré- existente. Acta Amazonica 14(1-2): 255-279.

Alencar, Jurandyr da Cruz. 1988. Estudos Silviculturais de uma população natural de Copaifera multijuga Hayne (Leguminosae) na Amazônia Central. 4. Interpretação de dados fenológicos em relação a elementos climáticos. Acta Amazonica 18(3-4): 199-209.

Allen, O.N. and Allen, E.K. 1981. The Leguminosae: A Source Book of Characteristics, Uses and Nodulation, Univ. of Wisconsin Press, Madison.

Amazon Ervas. N.S. (year not specified) Melkatoss. Promotional brochure. Amazon Ervas, Manaus.

Anderson, Arthur B. 1955. Recovery and utilization of tree extractives. Economic Botany 9:108-140.

Andrews, Jeffrey A., Thomas G. Siccama, and Kristina A. Vogt. 1999. The effect of soil nutrient availability on retranslocation of Ca, Mg and K from senescing sapwood in Atlantic white cedar. Plant and Soil 208:117-123.

Ankarfjärd, Renée and Miran Kegl. 1998. Tapping oleoresin from Dipterocarpus alatus (Dipterocarpaceae) in a Lao village. Economic Botany 52(1):7-14.

Anonymous. 1915. Plantes nouvelles ou peu connues de la région amazonienne par A. Ducke. Archivos do Jardim Botanico do Rio de Janeiro.

Arrhenius, S.P. and J.H. Langenheim. 1983. Inhibitory effects of Hymenaea and Copaifera leaf resins on the leaf fungus, Pestalotia. Biochemical Systematics and Ecology 11:361-366.

Arrhenius, S.P. and J.H. Langenheim. 1986. The association of Pestalotia species with members of the leguminous tree genera Hymenaea and Copaifera in the neotropics. Mycologia 78(4):673-676.

Balée, William. 1987. A etnobotânica quantitativa dos índios Tembé (Rio Gurupi, Pará). Boletim do Museu Paraense Emílio Goeldi: Botânica 1:29-50. 73

Ball, I.R., D.B. Lindenmayer, H.P. Possingham. 1999. A tree hollow dynamics simulation model. Forest Ecology and Management 123:179-194.

Balsam, M.S. and Sagarin, E. 1974. eds. Cosmetics: Science and Technology. Vol. 3. John Wiley and Sons, N.Y.

Barreto, P., P. Amaral, E. Vidal, and C. Uhl. 1998. Costs and benefits of forest management for timber production in eastern Amazonia. Forest Ecology and Management 108:9-26.

Basile, A.C., J.A.A. Sertie, P.C.D. Freitas, and A.C. Zanini. 1988. Anti-inflammatory activity of oleoresin from Brazilian Copaifera. Journal of Ethnopharmacology 22: 101-109.

Bergström, Berit, Gabriella Gustafsson, Rolf Gref, and Anders Ericsson. 1999. Seasonal changes of pinosylvin distribution in the sapwood/heartwood boundary of Pinus sylvestris. Trees 14:65-71.

Berry, Paul E., Gerardo Aymard, and Gustavo A. Romero. 1997. Copaifera camibar (Caesalpiniaceae), a useful and locally common, but previously unreported species in Venezuela. BioLlania Edicion Especial 6:275-279.

Berthier, S., A.D. Kokutse, A. Stokes, and T. Fourcaud. 2001. Irregular heartwood formation in Maritime pine (Pinus pinaster Ait): consequences for biomechanical and hydraulic tree functioning. Annals of Botany 87(1):19-25.

Blanche, C.A.., P.L. Lorio Jr., R.A. Sommers, J.D. Hodges, and T.E. Nebeker. 1992. Seasonal cambial growth and development of loblolly pine: xylem formation, inner bark chemistry, resin ducts, and resin flow. Forest Ecology and Management 49:151-165.

Boddy, Lynne and A.D.M. Rayner. 1981. Fungal communities and formation of heartwood wings in attached oak branches undergoing decay. Annals of Botany 47:271-274.

Braga, W.F., C.M. Rezende, O.A.C. Antunes. 1998. Terpenoids from Copaiba cearensis. Phytochemistry 49(1):263-264.

Budavari, Susan. 1989. ed. The Merck Index: an encyclopedia of chemicals, drugs, and biologicals. Merck, Rahway, N.J.

Calvin, M. 1983. New sources for fuel and materials. Science 219: 24- 26.

Cascon, Vera and Benjamin Gilbert. 2000. Characterization of the chemical composition of oleoresins of Copaifera guianensis Desf., Copaifera duckei Dwyer and Copaifera multijuga Hayne. Phytochemistry 55:773-778.

CEDI. 1985. Tembé. pp. 177-209 in Povos Indígenas no Brasil, No. 8: Sudeste do Pará. Centro Ecumênico de Documentação e Informação, São Paulo.

Chichignoud, Michèle, G. Déon, P. Détienne, B. Parant, and P. Vantomme. 1990. Tropical Timber Atlas of Latin America. International Tropical Timber Organization (ITTO), Yokohama and Centre Technique Forestier Tropical (div. CIRAD), Nogent-Sur-Marne Cedex, France.

Chudnoff, Martin. 1984. Tropical Timbers of the World. USDA Forest Service, Agricultural Handbook No. 607. 74

Clark, Deborah A. and D.B. Clark. 1993. Life history diversity of canopy and emergent trees in a neotropical rain forest. Ecological Monographs 62(3):315-344.

Correa, Jaime Enrique Q. and Henry Yesid Bernal. 1989. Especies vegetales promisorias de las paises del Convenio Andrés Bello. Tomo III. Secretaria Ejecutiva del Convenio Andrés Bello (SECAB), Bogotá.

Cortés, Santiago. N.S. (year not specified). Flora de Columbia. Libreria de el Mensajero, Bogota.

DCI. 1992. Raw Material Directory. Drug and Cosmetic Industry. July: 137.

Desmarchelier, C., J. Coussio, and G. Ciccia. 1997. Extracts of Bolivian plants, Copaifera reticulata and Heisteria pallida inhibit in vitro free radical-mediated DNA damage. Phytotherapy Research 11(6):460-462.

Di Bitetti, M.S., E.M.L. Vidal, M.C. Baldovino, and V. Benesovsy. 2000. Sleeping site preferences in tufted capuchin monkeys (Cebus apella nigritus). American Journal of Primatology 50(4):257-274.

Dodt. 1939. Descripção dos Rios Parnahyba e Gurupy. Brasiliana, São Paulo.

Duke, James A. 1986. Copaifera langsdorffii, Copaifera officinalis, and Copaifera reticulata. USDA Files.

Duke, James A. and J.L. Cellier. 1993. CRC Handbook of Alternative Cash Crops. CRC Press, Boca Raton.

Duke, James A. and Rodolfo Vásquez. 1994. Amazonian Ethnobotanical Dictionary. CRC Press, Boca Raton.

Dwyer, J.D. 1951. The Central American, West Indian and South American species of Copaifera Caesalpiniaceae). Brittonia 7(3): 143-172.

Ericsson, T. and A. Fries. 1999. High heritability for heartwood in north Swedish Scots pine. Theoretical and Applied Genetics 98(5):732-735.

Feibert, E.B. and J. H. Langenheim. 1988. Leaf resin variation in Copaifera langsdorffii: relation to irradiance and herbivory. Phytochemistry 27(8): 2527-2532.

Fearnside, Philip M. 1992. Forest biomass in Brazilian Amazônia: comments on the estimate by Brown and Lugo. Interciencia 17(1):19-27.

Fernandes, Jorge. 1949. Sobre o óleo resina de Copaiba e sua Aplicação Industrial. Associação Comercial do Amazonas, Manaus.

Figliuolo, Roberto, S. Naylor, J. Wang, and J. H. Langenheim. 1987. Unusual nonprotein imino acid and its relationship to phenolic and nitrogenous compounds in Copaifera. Phytochemistry 26(12):3255-3259.

Fleury, M. 1997. On medicinal role of Copahu balsam. Acta Botanica Gallica 144(4):473-479.

Fredericksen, T.S., M.J. Justiniano, B. Mostacedo, D. Kennard, and L. McDonald. 2000. Comparative regeneration ecology of three leguminous timber species in a Bolivian tropical dry forest. New Forests 20:45-64.

Fries, A. 1998. Genetic parameters in diallel-crossed Scots pine favor heartwood formation breeding objectives. Canadian Journal of Forest Research 28(6):937-941. 75

Fujita, Masayuki, David R. Gang, Laurence B. Davin, and Norman G. Lewis. 1999. Recombinant pinoresinol- lariciresinol reductases from western red cedar (Thuja plicata) catalyze opposite enantiospecific conversions. The Journal of Biological Chemistry 274(2):618-627.

Furnemore, H. 1926. The Essential Oils. Ernest Benn Ltd., London.

Gentry, Alywn H. 1993. A Field Guide to the Families and Genera of Woody Plants of Northwest South America (Columbia, Ecuador, Peru). The University of Chicago Press, Chicago.

Gershenzon, J. 1994. The cost of plant chemical defense against herbivory: a biochemical perspective. pp. 165- 219 in E.A. Bernays (ed.) Insect Plant Interactions. CRC Press, Boca Raton.

Gianno, Rosemary. 1984. The exploitation of resinous products in a lowland Malayan forest. Wallaceana 43:3- 6.

Gianno, R. 1990. Semelai Culture and Resin Technology. The Connecticut Academy of Arts and Sciences, New Haven.

Greenberg, Cathryn H. and Robert W. Simons. 1999. Age, composition, and stand structure of old-growth oak sites in the Florida High Pine Landscape: implications for ecosystem management and restoration. Natural Areas Journal 19(1):30-40.

Herrera, Fortunato L. 1921. Contribucion a la Flora del Departamento del Cuzco. Universidad del Cuzco, Peru.

Herres, W., Kubczka, K.H., and Schultze W. 1986. HRGC-FTIR investigation on volatile terpenes in Brunke, Ersnst-Joachim (ed.) Progress in Essential Oil Research. Walter de Gruyter, Berlin/New York.

Howes, F. N. 1949. Vegetable Gums and Resins.Chronica Botanica Company, Waltham, MA.

Hillis, W.E. 1987. Heartwood and Tree Exudates. Springer-Verlag, Berlin.

Johns, Jennifer S., Paulo Barreto, and Christopher Uhl. 1996. Logging damage during planned and unplanned logging operations in the eastern Amazon. Forest Ecology and Management 89:59-77.

Johnson, L.K. 1983. Apiomerus pictipes (Reduvio, Chinche Asesina, Assassin Bug). pp. 684-687 in Daniel H. Janzen (ed.) Costa Rican Natural History. The University of Chicago Press, Chicago.

Kalpagé, F.S.C.P. 1974. Tropical Soils: Classification, Fertility and Management. St. Martin’s Press, New York.

Kraemer, Henry. 1907. A Text-Book of Botany and Pharmacognosy. J.B. Lippincott Co., Philadelphia. pp. 840

Laboratório São Lucas. N.S. Oleo de Copaiba (Copaifera officinalis). Promotional leaflet. Laboratório São Lucas, Belém, Pará.

Langenheim, Jean H. 1973. Leguminous resin-producing trees in Africa and South America in Meggers, B.J., Ayensu, E.S., and Duckworth, W.D. eds. Tropical Forest Ecosystems in Africa and South America: A Comparative Review. Smithsonian Institution Press, Washington, D.C.

Langenheim, Jean H. 1990. Plant resins. American Scientist 78: 16-24.

Langenheim, Jean H. in press. Plant Resins: their Value to Plants and Humans. Timber Press, Portland Oregon 76

Le Cointe, Paul. 1947. Amazonia Brasileira III Arvores e Plantas Uteis. Biblioteca Pedagógica Brasileira, Serie 5, Vol. 251. Companhia Editora Nacional, São Paulo.

Leal, I.R. and P.S. Oliveira. 1998. Interactions between fungus-growing ants (Attini), fruits and seeds in cerrado vegetation in southeast Brazil. Biotropica 30(2):170-178.

Leung, Albert L. 1980. Encyclopedia of Common Natural Ingredients used in Food, Drugs and Cosmetics. John Wiley and Sons, New York.

Lindenmayer, D.B., R.B. Cunningham, C.F. Donnelly, M.T. Tanton, and H.A. Nix. 1993. The abundance and development of cavities in Eucalyptus trees: a case study in the montane forests of Victoria, southeastern Australia. Forest Ecology and Management 60:77-104.

Lindenmayer, David B., Ross B. Cunningham, and Christine F. Donnelly. 1997. Decay and collapse of trees with hollows in eastern Australian forests: impacts on arboreal marsupials. Ecological Applications 7(2):625-641.

Lindenmayer, D.B., R.B. Cunningham, M.L. Pope, P. Gibbons, and C.F. Donnelly. 2000. Cavity sizes and types in Australian eucalypts from wet and dry forest types - a simple rule of thumb for estimating size and number of cavities. Forest Ecology and Management 137:139-150.

Lorio, Jr., P.L. and R.A. Sommers. 1986. Evidence of competition for photosynthases between growth processes and oleoresin synthesis in Pinus taeda L. Tree Physiology 2:301-306.

Macedo, C.A. and Langenheim, J.H. 1989(a). A further investigation of leaf sesquiterpene variation in relation to herbivory in two Brazilian populations of Copaifera langsdorffii. Biochemical Systematics and Ecology 17(3): 207-216.

Macedo, C.A. and Langenheim, J.H. 1989(b). Intra- and inter-plant sesquiterpene variability in Copaifera langsdorffii: relation to microlepidopteran herbivory. Biochemical Systematics and Ecology 17(7-8): 551-557.

Maia, J.G. Soares, M.J.C. Varejão, W. Wolter Filho, A.P. Mourão, A.A. Craveiro, and J.W. Alencar. 1978. Estudo químico de óleos essencías, oleaginosas e látices da Amazônia. 1. Composição e oxidação de uma espécie de Copaifera (Nota prévia). Acta Amazonica 8(4):705.

Marafioti, Richard Lynn. 1970. The meaning of generic names of important economic plants. Economic Botany 24:189-207.

Martini, A. M.Z., N.A. Rosa, C. Uhl. 1994. An attempt to predict Amazonian tree species potentially threatened by logging activities. Environmental Conservation 21: 152-162 and unpublished appendices.

Mattheck, C., K. Bethge, and P.W. West. 1994. Breakage of hollow tree stems. Trees Structure and Function 9:47-50.

Pio Corrêa. 1932. Dicionário das plantas úteis do Brasil. New edition. IBDF (Instituto Brasileira para Desenvolvimento de Florestas), Brasilia.

MMA/SCA (Ministério do Meio Ambiente, Secretaria de Coordenação da Amazônia), GTA - Grupo de Trabalho Amazônica and SEBRAE - Serviço Brasileira de Apoio às Micro e Empresas. 1998. Copaíba in Produtos Potenciais da Amazônia. MMA, SCA, SUFRAMA, SEBRAE, Brasília. 77

Monti, H., N. Tiliacos, and R. Faure. 1999. Copaiba oil: isolation and characterization of a new diterpenoid with the dinorlabdane skeleton. Phytochemistry 51(8): 1013-1015.

Moreira, Fatima M. de S., Marlene F. da Silva, and Sérgio M. de Faria. 1992. Occurrence of nodulation in legume species in the Amazon region of Brazil. New Physiologist 121:563-570.

Mors, Walter B. and Carlos T. Rizzini. 1966. Useful Plants of Brazil. Holden-Day Inc., San Francisco.

Mostacedo, Bonifacio C., Todd Fredericksen, and Marisol Toledo. 1998. Respuestas de las plantas a la intensidad de aprovechamento en um bosque semideciduo pluviestacional de la region de Lomerío, Santa Cruz, Bolivia. Revista de la Sociedad Boliviana de Botánica 2(1):75-88.

Nakada, R., Y. Fujisawa, Y. Hirakawa. 1999. Soft X-ray observation of water distribution in the stem of Cryptomeria japonica D-Don I: general description of water distribution. Journal of Wood Science 45(3):188-193.

Nogueiro-Neto, Paulo. 1997. Vida e Criação de Abelhas Indígenas Sem Ferrão. Editora Nogueirapis, São Paulo.

Orallo, Cesar A. and Vicente P. Veracion. 1984. Comparison of four methods of tapping Benguet pine (Pinus kesiya Royle ex Gordon) for oleoresin production in Benguet. Sylvatrop Philippines Forest Research Journal 9(1-2):55-64.

Osol, Arthur and G.E. Farrar. 1947. eds. The Dispensatory of the United States of America, 24th Edition. J.B. Lippincott Co., Philadelphia.

Pendleton, Linwood H. 1992. Trouble in paradise: practical obstacles to nontimber forestry in Latin America. pp. 252-262 in Mark Plotkin and Lisa Famolare (eds.) Sustainable Harvest and Marketing of Rain Forest Products. Island Press, Washington, D.C.

Peters, Charles M. Sustainable Harvest of Non-timber Plant Resources in Tropical Moist Forests: An Ecological Primer. The Biodiversity Support Program, Washington, D.C.

Phillips, O., A.H. Gentry, C. Reynel, P. Wilkin, and C. Gálvez-Durand B. 1994. Quantitative ethnobotany and Amazonian Conservation. Conservation Biology 8(1):225-248.

Phillips, Michael A. and Rodney B. Croteau. 1999. Resin-based defenses in conifers. Trends in Plant Science 4(5):184-191.

Pittier, Henri. 1926. Manual de las Plantas Usuales de Venezuela. Editorial Elite (1939 compendium), Caracas.

Plotkin, Mark J., Boom, B.M. and Allison, M. 1991. Ethnobotany of Aublet’s Histoire des Plantes de la Guiane Françoise (1775). Monographs in Systematic Botany Vol. 35, Missouri Botanical Garden.

Plowden, C. in press. Profile on Copaifera spp. in P. Shanley (ed.) The Management and Marketing of NTFPs: Certification as a Tool to Promote Sustainability. Earthscan (WWF/People and Plants Series).

Poucher, William A. 1950. Perfumes, Cosmetics and Soaps with Special Reference to Synthetics. Vol. 1&2 Chapman and Hall Ltd., New York.

Projeto Radam. 1973. Levantamento de Recursos Naturais and Mapa Exploratório de Solos. Vol. 3. Programa de Integração Nacional. SUDENE. Ministério das Minas e Energia, Departmento Nacional da Produção Mineral, Rio de Janeiro. 78

Ramirez, N. and M.K. Arroyo. 1987(a). Infection, emergence and competitive interactions of seed predators in Copaifera pubiflora Benth. (Leguminosae: Caesalpinioideae) in Venezuela. Acta Cient. Venez. Caracas: Asociacion Venezolana para el Avanca de la Ciencia. 38(2): 216-225.

Ramirez, N. and M.K. Arroyo. 1987(b). Spatial and temporal variations in seed depredation of Copaifera pubiflora Benth. (Leguminosae: Caesalpinioideae) in Venezuela. Biotropica 19(1):32-39.

Ramirez, Nelson and Mary Kalin Arroyo. 1990. Estructura poblacional de Copaifera pubiflora Benth. (Leguminosae; Caesalpiniodeae) en los Altos Llanos Centrales de Venezuela. Biotropica 22(2):124- 132.

Record, Samuel J. and R.W. Hess. 1943. Timbers of the New World. Yale University Press, New Haven.

Record, Samuel J. and Clayton D. Mell. 1924. Timbers of Tropical America. Yale University Press, New Haven.

Rodrigues, Roberto M. 1989. A Flora da Amazônia. CEJUP. Belém.

Roubik, David W. 1989. Ecology and Natural History of Tropical Bees. Cambridge Univ. Press, Cambridge.

Sales, Noêmia Pires de. 1993. Pressão e Resistencia: Os Índios Tembé-Tenetehara do Alto Rio Guamá e a Relação com o Território. União das Escolas Superiores do Pará (UNESPa), Belém.

Sales, Noêmia Pires de. 1994. Os Tembé no Alto Rio Guamá: Reelaborações Étnicas - Identidade e Território. UNAMA - Universidade da Amazônia, Belém.

Sampaio, Paulo de T.B. 1993. Copaíba. pp. 159-164 in Jason W. Clay and C.R. Clement (eds.) Selected Species and Strategies to Enhance Income Generation from Amazonian Forests. United Nations Food and Agricultural Organization, Rome.

Sayre, Lucius E. 1906. A Manual of Organic Materia Medica and Pharmacognosy. P. Blakiston’s Son and Co., Philadelphia.

Schultes, Richard E. and R. F. Raffauf. 1990. The Healing Forest: Medicinal Toxic Plants of the Northwest Amazon. Vol. 2. Dioscorides Press, Portland.

SEBRAE. 1995. Copaíba: Opções de investimento no Acre com produtos florestais não madeireiros. Part of series Produtos Potenciais da Amazônia. Serviço de Apoio às Micro e Pequenas Empresas do Acre (SEBRAE), Rio Branco.

Shanley, Patricia, Margaret Cymerys, and Jurandir Galvão. 1998. Frutíferas da Mata na Vida Amazônica. Patricia Shanley, Belém.

Shigo, Alex L. 1994. Tree Anatomy. Shigo and Trees, Associates. Durham, N.H.

Spruce, Richard. 1970. Notes of a Botanist on the Amazon and Andes. Volume 1. Johnson Reprint Corp., New York.

Stokes, A. and S. Berthier. 2000. Irregular heartwood formation in Pinus pinaster Ait. is related to eccentric, radial, stem growth. Forest Ecology and Management 135(1-3):115-121. 79 ter Steege, Hans and Gerold Zondervan. 2000. A preliminary analysis of large-scale forest inventory data of the Guiana Shield. pp. 35-54 in Hans ter Steege (ed.) Plant Diversity in Guyana. Tropenbos Foundation, Wageningen, the Netherlands.

Uhl, C., A. Verissimo, M. Mattos, A. Brandino and I. Viera. 1991. Social, economic, and ecological consequences of selective logging in an Amazon frontier: the case of Tailandia. Forest Ecology and Management, 46: 243-273.

Uhl, C., P. Barreto, A. Verissimo, E. Vidal, P. Amaral, A.C. Barros, C. Souza Jr., J. Johns and J. Gerwing. 1997. Natural resource management in the Brazilian Amazon. BioScience 47(3):160-168.

Uibarri, E.A. 1997. Fabaceae parte 2 subfam. Caesalpinoideae parte 2 tribu Detariae DC. pp. 10-12 in Flora fanerogamica Argentina. Programa PROFLORA (CONICET) Vol. 33. Cordoba.

USDA (United States Department of Agriculture). N.S. (Year Not specified). Copaiba. Internal leaflet DRP- 26(13) prepared by USDA, Agricultural Research Administration, Bureau of Plant Industry, Division of Drug and Related Plants.

Uphof, J. C. Th. 1968. Dictionary of Economic Plants. Verlag Van J. Cramer, New York.

Vega, Gilberto Maheca, Roberto Rodríguez Soto, and Luis Enrique Acero Duarte. 1984. Estudio Dendrologico de Columbia. Republica de Columbia, Bogota. van den Berg, Maria E. 1984. Plantas Medicinais na Amazônia - Contribuição ao seu Conhecimento Sistemático. Conselho Nacional de Desenvolvimento Científico e Tecnológico/MPEG, Belém.

Verissimo, Adalberto, P. Barreto, M. Mattos, R. Tarifa, and C. Uhl. 1992. Logging impacts and prospects for sustainable forest management in an old Amazonian frontier: the case of Paragominas. Forest Ecology and Management 55:169-199.

Verpoorte, R. and P.P. Dahl. 1987. Medicinal plants of Surinam IV. Antimicrobial activity of some medicinal plants. Journal of Ethnopharmacology 21: 315-318.

Vetter, Roland E. and Paulo C. Botosso. 1989. Remarks on age and growth rate determination of Amazonian trees. IAWA Bulletin 10(2):133-145.

Viera, Lúcio S. 1992. Fitoterapia da Amazônia: Manual das Plantas Medicinais. Editora Agronômica Ceres Ltda, São Paulo.

Wang, Shih-Chi and J.B. Huffman. 1981. Botanochemicals: supplements to petrochemicals. Economic Botany 35(4):369-382.

Warner, P.D. III and Andrea Coutinho Pontual. 1994. Manual de Comercialização de Produtos Florestais. GENESYS, Rio de Janeiro.

Woodson Jr., Robert E. and R.W. Schery. 1951. Flora of Panama (Leguminosae). Annals of the Missouri Botanical Garden 38, p. 27.

Youngken, Heber W. 1943. Text-Book of Pharmacognosy. The Blakiston Co., Philadelphia.

80

APPENDIX 2-A COMMON NAMES FOR COPAIFERA TREES IN THE AMAZON REGION

ARGENTINA: timbó-y-atá BOLIVIA: copaibo BRAZIL: Tapajós: copaiba = C. multijuga, copahyba marimary, copahyba angelim; copahiba preta=C. glycycarpa Trombetas: copahyba marimari = C. langsdorffii; copahyba jutahy = C. martii Eastern Pará: copaiba-marimari = C. reticulata, copaiba jutaí Jutai River Falls: copaiba-branca = C. guianensis Maues: copaiba-cuiarana = C. glycycarpa Santarem/Obidos: copaiba-rana, cupa-úba, balsam capivi = C. martii Non-specified region: balsamo de copaiba, cabreuva, copahibera, copahiba, copahyba, c. angelim, c. cuiarana, c. jutahy, c. marimary, c. parda, c. roxa, c. vermelha, copahúva, copaiba, copaibera, copaúva, jaboti meutámeutá, jacaré copahiba, oleo, oleo folha, oleo pardo, olho, pau d’oleo, yébaro CHILE: bálsamo de copaiba, copaiba del Brasil COLUMBIA: trompo puerco = C. guianensis; aceite, arbol de aceite, canime, copaiba, copaibi, kurruma, pata de gallo, queiyane (Kurripaco Indian) (possible species = C. officinalis, C. canime); canime vailuno = C. pubiflora FRENCH GUIANA: panchimouti (possible species = C. officinalis) GUAYANA: balsam, maram (possible species = C. officinalis) PARAGUAY: cupay, kupa’y PERU: copaiba, copaiva, copaiba blanca = C. reticulata SURINAM: apaoewa, hoepelboom, hoepelhout, hoepfroe-hoedoe, hoeproe, koepajoewa, koepawa, pasoemoeti, passiemoetie (possible species = C. officinalis) VENEZUELA: aceite, cabima, cabimbo, cabimo, calenibo, calimbo, copaiba, currucaí, curracay, kurukay, maramo, palo de aceite, palo de aceitillo (possible species = C. officinalis; C. venezuelana NON-SPECIFIED COUNTRY: balsam copaiba, copaiba, copaiva, copahu, Jesuit’s balsam Sources: Anonymous, 1915; Cortés, N.S.; Record and Hess, 1943; Youngken, 1943; Le Cointe, 1947; Dwyer, 1951; Mors and Rizzini, 1966; Spruce, 1970; Leung, 1980; Chudnoff, 1984; Vega et al., 1984; Correa and Bernal, 1989; Chichignoud et al, 1990; Sampaio, 1993; Phillips et al., 1994; Uibarri, 1997.

CHAPTER 3

THE HARVEST OF BREU RESIN FROM BURSERACEAE TREES IN THE

EASTERN BRAZILIAN AMAZON AND THE ROLE OF WEEVILS AND BEES IN

ITS FORMATION AND MARKETING 82

ABSTRACT

Many Brazilian Amazon forest dwellers collect and sell resins from various species of the genus

Protium (family Burseraceae). The resin that is locally called breu is used as a fire starter, incense, medicine and wooden boat caulking material. A study conducted with Tembé Indians in the Alto Rio Guamá Indigenous

Reserve in eastern Pará state revealed that trees yielded an average of 0.8 kg and up to 11 kg of resin during an initial harvest. Resin flow is stimulated by a previously undescribed species of weevil Sternocoelus new species #1 (Coleoptera: Curculionidae). Weevil larvae bore into the inner bark and develop inside resin lumps that form on the bark exterior. This is a novel adaptation by an insect to a terpene-rich compound that is usually a potent defensive agent against pest species. At the Tembé study area and another site in the region, resin was found more often and in larger amounts as tree size increased and tree health deteriorated. Trees in forests that were occasionally flooded had larger resin harvests than ones in upland areas. Resin lump size was closely correlated with the number and size of bore holes made by weevil larvae. Experimental drilling demonstrated that large resin lumps only accumulate when trees are repeatedly bored by beetle larvae. Follow- up harvests at the Tembé site showed that trees that have already been infected by the beetles are more likely to be colonized by additional beetles than uninfected trees. Weevil larvae may take one or two years to reach the pupation stage. A regression model that includes tree diameter, the number of resin lumps harvested during the first harvest, the weight of resin lumps harvested in the first harvest and the time since the first harvest predicts it will take about four years between successive harvests for resin levels to rebuild to the initial harvest level.

Inventories in the Tembé area show that the types of Burseraceae trees likely to yield harvestable resin have a density of 9 to 10 trees/ha. This places the site on the low end of one hectare plots surveyed in various parts of the Brazilian Amazon where densities of such trees can exceed 50 trees/ha. Inventories of 100 ha plots, however, indicate that these high densities may not exist at a larger scale. It is estimated that as density increases from 1 to 50 trees/ha, one experienced collector can harvest an average of 1 to 19 kg of resin per day. Since resin is laid out in the sun to dry, these yields drop by 15-20% before the material is sold for

$US 0.25 to .50 per kg. This weight loss is probably due to the loss of volatile components in the resin, but an 83 experiment comparing drying rates of small resin samples that either permitted or excluded access to stingless bees showed that these stingless bees can remove substantial amounts of resin.

Indigenous communities in the Gurupi River area probably harvest several tons of resin per year.

Since harvesters typically search away from villages, harvesting may have reduced the most accessible supplies of resin. At a landscape scale, however, current harvest levels can probably be maintained because the selling price for the resin and the market demand for it are both relatively low. Phytochemists are now seeking to identify essential oils from breu resin that are marketable to the perfume industry. If such efforts succeed and increase the market demand and price for breu resin, harvesters could earn more but pressures to overharvest the resource would also increase. This study suggests that other researchers and communities involved in resin harvest and medicinal plants in general should further explore possible connections between plant defensive compounds and specific pest agents that may provoke their production. Such agents might better be thought of and potentially managed as a new type of beneficial insect instead of a pest.

INTRODUCTION

Neotropical forest dwellers from the Yucatan to the Amazon have collected resins from trees in the

Protium Burman, Tetragastris Gaertner, and Trattinnickia Willd. genera (family Burseraceae) for subsistence and commercial purposes for hundreds if not thousands of years (Alcorn, 1984a, 1984b; Turner and Miksicek,

1984; Rodrigues, 1989; Neels, 2000). In Brazil, the resin was formerly known as “hard elemi” (Howes, 1949;

Mantell, 1950); the most common current name for both the resin and the trees that produce it is breu

(pronounced “bray-oo”) (Rodrigues, 1989)(Appendix 3-A). The aromatic resin is burned to provide light, start charcoal fires and make ceremonial smoke (Levi-Strauss, 1952; Daly, 1987; Rodrigues, 1989; Balée, 1994).

Various resins from select species are also used as a flavor, scent and a medicine to treat ailments including skin infections and parasites, wound maggots, nasal congestion, rheumatism, toothache, hernias, poor vision, conjunctivitis, respiratory, stomach and menstrual problems (Howes, 1949; Mantell, 1950; Mors and Rizzini,

1966; Reitz, 1981; Secoy and Smith, 1983; Alcorn, 1984(1); Plowman, 1984; van den Berg, 1984; Schultes and Raffauf, 1990; Balée and Daly, 1990; Kainer and Duryea, 1992; Lima et al., 1992; Milliken et al., 1992;

Balée, 1994; Grimes et al., 1994; Boom, 1996; Comerford, 1996; Johnston and Colquhoun, 1996; Milliken and 84

Albert, 1996; Siani et al., 1999(1); Neels, 2000). Sticky forms are used as glues in indigenous handicrafts, glazing for ceramics, and caulking for canoes (Levi-Strauss, 1952; Whitten and Whitten, 1988; Balée and

Daly, 1990; Daly, 1992; Balée, 1994; Grimes et al., 1994; Johnston and Colquhoun, 1996).

Breu resin is found in various qualities. The most common type is a sticky balsam that comes from species in the Protium and Tetragastris genera. It exudes from the tree in a clear to milky white liquid (Daly,

1987; Gentry, 1993). The resin lump that coagulates on the trunk may remain white or turn black depending on the extent of oxidation and polymerization. “Breu branco verdadeiro” (true white breu) is prominently used for caulking wooden boats (Rodrigues, 1989; Balée, 1994). “Breu jauaricica” or “breu preto” (black breu) is darker with greenish blotches and is used more for its aromatic properties (Rodrigues, 1989). While there are at least 35 species in the Protium genus alone in the eastern Amazon, only some of these produce harvestable quantifies of resin (Daly, 1987; Balée, 1987, 1994).

In the eastern Brazilian Amazon, the Tembé and Ka’apor Indians (both Tupi-Guaraní groups) rely on the harvest of breu resin (principally from Protium trees) as one of their main economic products (CEDI, 1985;

Balée, 1987, 1994). The quantity of resin harvested appears to be growing in line with their desire to generate more income. As improvements to local infrastructure make villages more permanent, the local depletion of natural resources such as breu resin becomes more likely. I conducted this study to quantify the amount of breu being removed from the forest, understand some of the underlying mechanisms that influence resin production, estimate the amount of time required for resin to regenerate post-harvest, and make recommendations to guide development of a sustainable harvest mode. It was part of a larger project that evaluated the production ecology of several plants yielding marketable non-timber forest products (NTFPs) in the Alto Rio Guamá Indigenous Reserve (Pará state) and other parts of the region.

The Tembé technique for harvesting breu resin is straightforward. When someone encounters a breu tree with a resin lump on it and wishes to collect it, he pries the lump off with his hand or machete. If the lump is higher up on the trunk, he may knock it loose with a long stick. Lumps of dried resin that have fallen off the trunk are also collected from the ground. Resin collected for personal use is done opportunistically during a hunting trip or searching for other forest products. When collecting breu for sale, the harvester will dedicate one or more days to that task with the goal of filling several large sacks with resin. Upon return to the village, 85 the fresh resin is laid out on bags in the sun for several days to dry and prepare the material for sale. Collectors all reported finding larvae (“tapuru”) in the resin lumps, but none claimed to know what these immature insects turned into or what role if any they played in resin formation (Plowden et al., 2002).

The main objectives of this study were to answer the questions: 1) how much resin be harvested from a population of breu trees and what factors influence variation in this amount, 2) how often can resin be harvested from breu trees and what factors influence this frequency, and 3) how much income can forest people make by harvesting breu resin. Under question one, the role of tree type, tree size, tree condition, population density, size class distribution, forest type, and harvesting variables (distance from the village; interval between successive harvests) were specifically investigated. As the involvement of weevils

(Curculionidae), bees (Apidae) and other invertebrates in this resin system became more apparent, other questions were addressed as subsets of the main questions: 1) what role do weevils play in the formation of harvestable resin, 2) what is the importance of breu trees and resin to these weevils, 3) how much fresh resin is removed by stingless bees. A summary of specific research objectives, surveys and methods, analysis and location of results in the chapter is listed in Table 3.1.

STUDY AREAS

The main study area was in the Alto Rio Guamá Indigenous Reserve located in eastern Pará state of the Brazilian Amazon. It is the principal homeland of the Tembé Indians. Research was carried out in an area of about 500 ha of closed tropical rainforest near the village of Tekohaw (020 37.7’ S; 460 33.1’ W) located on the Gurupi River. Underlying soils are on the border of two major soil types classified by the Brazilian system as yellow latossols and red-yellow podzols - the latter called spodosols in the American system (Projeto

Radam, 1973; Kalpagé, 1974). The site has three main forest types: upland (“terra firme”) forest, lowland forest that is occasionally flooded (“baixo”), and forest near the main river and major streams that is usually flooded during the rainy season. Near this and other villages, forest patches are cleared and burned to plant manioc and other crops (Sales, 1993, 1994). This land use pattern creates a mosaic of forest in various stages of current farm use, fallowed secondary forest and old forest. Other important aspects of the landscape are 86 swaths of secondary forest that resulted from unintentional forest fires that passed through the area in the early

1980s.

Additional observations were made in a 100 ha terra firme forest plot on state road PA150 near the town of Mojú, Pará (020 11.5’ S; 480 49.1’ W). The forests in both study areas are dominated by trees in the

Lecythidaceae, Burseraceae, Leguminosae, and Sapotaceae families (Baleé, 1987; Lockman, personal communication 2000).

METHODS

RESIN HARVEST AMOUNTS PER TREE AND AREA OF FOREST

Breu Tree Density Inventories

Breu tree density at the Tekohaw study site was estimated in two successive belt-transect style inventories. The first one carried out in January, 1998 surveyed trees and site characteristics in two 1000 ha blocks of forest near Tekohaw village. Each block had twelve 500 m long north-south oriented transects originating from the block mid-line that were located through a stratified random selection process. Transects were sub-divided into 25 m long sections. Six out of the twenty sections in each transect were chosen by stratified random sampling for 100% inventory of all trees ≥ 10 cm DBH (at 1.5 m) within 5 meters of the transect. Forest type, common name, and DBH were noted for all inventoried trees. Breu densities were analyzed in 2.55 ha of 102 forest plots (10x25 meters each) that were not recently burned or subject to prolonged annual flooding. The second inventory was carried out in the same blocks in November, 1999 to increase the size of the area surveyed specifically for breu trees. In this study, all breu trees ≥ 10 cm DBH were inventoried in ten 500 m long transect lines in areas not recently burned or subject to prolonged annual flooding. In addition to recording the same type of information gathered in the first inventory, the presence or absence of resin on the trunk was also noted. Inventory results were analyzed to estimate the overall density of 87 resin yielding breu trees, describe the size-class distribution of these trees and estimate the percentage of trees in each size class that had harvestable resin.

Resin Harvest Studies and the Role of Weevils

In June and July 1998 and April 1999, a team of Tembé Indian men searched within 5 km of Tekohaw village on the Pará side of the river for the kinds of breu trees that typically yield resin. The tree’s common name, DBH, general health (live, injured, dead), GPS position, and forest type (“terra firme”, “baixo”) were recorded. Each tree was marked with a numbered aluminum tag, and leaves were collected from some trees for identification purposes. In the first month of the study, all breu trees were marked whether or not they had resin. If resin was present, these lumps were harvested and weighed individually. The number and depth of weevil bore holes found behind each resin lump was then recorded. In the second month, only trees that actually had resin were tagged. All resin lumps were again harvested, but only the number of resin lumps and their aggregate weight were recorded. The length and body width of larvae found in resin lumps were measured. Other arthropods found in resin lumps were classified by general type. Weevil pupae and adults were collected from host trees on the few occasions they were encountered. Results were analyzed to compare resin amounts to tree type, tree size, general tree health, forest type, distance from the village (as a potential indicator of previous harvesting intensity), and extent of weevil activity (number and size of weevils and weevil bore holes). Data on weevils were also analyzed to describe natural history aspects such as the size classes that correspond to larval instars and the relationship between weevil larval and bore hole number and size.

Experimental Drilling

To experimentally compare resin buildup with mechanical wounds similar in shape and size to those made by weevil larvae, 18 breu trees were drilled at the Tekohaw site in November, 1999. The objectives of this test were to measure resin production from one time mechanical boring and explore the relationship between resin production and hole diameter and depth. A hand drill was used to make a series of four holes on the north side of each tree with a 4.2 mm drill bit; a similar series was made on the south side with a 3.4 mm 88 drill bit. Each series consisted of holes drilled to depths of 5, 10, 15, and 20 mm with 40 mm separating adjacent holes. These hole diameters and depths covered the upper range of hole sizes made by beetle larvae.

Each hole’s position was marked underneath with a small circle of red spray paint. The trees were revisited four weeks after drilling to measure resin accumulation.

THE RATE OF BREU RESIN ACCUMULATION

Resin Yields in Follow-up Harvests

Two methods were used to measure the rate that resin accumulates on breu trees. These same methods helped describe host tree colonization preferences of Sternocoelus weevils and the development rate of their larvae in breu resin lumps. In the first method, I conducted follow-up visits to subsets of the trees that had been first harvested or monitored in the summer of 1998. These visits occurred in November 1998, March and July 1999 - approximately 5, 8, and 12 months after the initial harvest. A subset of the trees harvested 8 months after the initial harvest was harvested a third time 4 months later. During these visits, an effort was made to individually remove and separately weigh resin lumps occupied by no more than one weevil larva.

The length, head width, body width, and weight were measured for larvae found in each resin lump.

Resin Lump Growth Rate and Weevil Development Time

In the second method of measuring the rate of resin accumulation, all resin lumps on some trees first harvested in the summer of 1998 were individually marked with a numbered aluminum tag in March 1999.

During this and two other visits to these trees in April and July 1999 a thin gauge electrical wire was wrapped around each resin lump. The wire was then transferred to a piece of paper where a tracing of its perimeter was used to estimate the lump’s basal area. A comparison of resin lump weight to basal area obtained from other harvested lumps was used to estimate the growth rate of these individually marked resin lumps by weight.

The results of both studies were analyzed to compare resin regeneration to factors such as breu tree type, tree size, forest type, time since previous harvest, and amount of resin present at the previous harvest. 89

The most significant factors were combined in a regression model to estimate the amount of time that would be required for resin amounts to rebuild to the average found during the initial harvest. Data from the second study were used to estimate the time required for one weevil to complete its larval (resin forming) stage.

BREU RESIN HARVESTING ECONOMICS AND MARKET

Breu Tree Search and Harvesting Time

During the initial resin harvest at Tekohaw in July, 1998, the time of arrival and departure at each tree with resin was recorded. These data were used to estimate the amount of time needed to find a tree with resin.

Resin Drying Studies and the Role of Stingless Bees

. Resin must be dried prior to selling, but since this process leads to weight loss from volatilization, evaporation and harvest by stingless bees, it has economic implications for commercial harvesters. Resin weight loss during drying was quantified at Tekohaw in July, 1998 by laying out batches of freshly collected resin on plastic sacks for five to eight days outside in the sun. The total weight of each batch and the proportion of different resin types (“branco” and “sarara”) was recorded before drying. At the end of each day, the resin was reweighed. The weighing of individual batches concluded when weight stabilized for two successive days. Results were analyzed to calculate the amount of weight lost during the typical procedure.

A companion experiment was carried out in April, 1999 to estimate the specific effect of stingless bee resin removal on weight loss during the drying process. In this procedure, enclosures that were 25 cm on a side were covered with plastic mesh and placed outside four days in a seven day period. Four enclosures were covered on all sides; four had one side open so they were accessible to bees. One 36g sample of resin taken from the same original breu “sarara” resin lump was placed inside each enclosure. Resin samples were weighed at the beginning and end of every day samples were placed outside. On the second, third and fourth days the samples were outside, the type and number of bees present in each enclosure at one time was recorded in a spot check about every 30 minutes. A few specimens of the two types of bees that visited the enclosures 90 were collected for measurement and identification. Results were analyzed to compare drying rates between resin samples exposed to bee removal and samples that dried without such exposure.

Breu Resin Sales in the Village and Urban Markets

Information on the amounts and values of breu resin collected and sold by Tembé and Ka’apor

Indians was obtained primarily through conversations with one Tembé man who bought resin from Tembé living in or near Tekohaw village. He further cooperated with the research by keeping a log book of resin purchases and sales between July, 1999 and Feb. 2000. I obtained additional information about amounts, values, and end uses of breu sold in urban markets by interviewing the owner of a forest products business in

Belém and managers of several boat repair businesses in the Belém and Icoaraçi that use resin to caulk wooden boats.

Breu Resin Harvest Models

Information on average resin yield by size class of breu trees were combined with estimates of drying rates and values to predict the amount and value of resin that could be harvested from a forest with different densities of breu trees. A second model uses these variables as well as harvesting efficiency and the rate of resin accumulation to predict the daily yield and earnings from breu resin harvesting and the amount of forest that would be needed to employ a breu harvester full-time at different densities of breu trees.

RESULTS

RESIN HARVEST AMOUNTS PER TREE AND AREA OF FOREST

Breu Tree Density Inventories

The Tembé have numerous names in both their language and Portuguese for the wide array of breu trees in their forest (Balée, 1987) that are based on bark, wood, and leaf characteristics. While most 91

Burseraceae species in the area are probably resin producers, there are two major types (that both include several species) that yield harvestable resin on the trunk. These are called “branco” (white) and “sarara.”

According to the Tembé, red “vermelho” breu trees are not a source of such resin. The belt transect inventories conducted at Tekohaw showed that “branco” and “sarara” breu trees had a combined average density of 9.6 trees/ha (Table 3.2). The “branco” and “sarara” types contributed equally to this overall average, but the densities of these two types varied considerably between the two main forest types. “Branco” type trees were more common in the “terra firme” type forest, while “sarara” type trees had a higher density in the wetter

“baixo” type forests.

In the second inventory with the larger sample size the size class distribution of resin yielding breu trees was heavily weighted in the 10-30 cm DBH size classes. Trees that were 10-20 cm DBH accounted for

76% of the trees; 20-30 cm DBH trees garnered 20% of the total, and the 30-40 and 40-50 cm DBH classes each had only 2% of the total trees located.

Resin Harvest by Tree Type

Resin harvest studies at Tekohaw showed that at least seven species of Protium including P. pallidum

Cuatrec., P. trifoliolatum Engl., P. decandrum (Aubl.) Marchand, P. morii Daly, P. giganteum Engl. var. giganteum, P. polybotryum (Turcz.)Engl., and P. glabrescens Swart and Tetragastris panamensis (Engl.)Kuntze had resin lumps on them. P. pallidum and P. trifoliolatum were generally classified as “breu branco.” These trees were found almost exclusively in well-drained “terra firme” type forest. Their resin accumulated in rounded white lumps that turned grayish with age. The other six identified species were considered different varieties of breu “sarara.” “Sarara” trees were found evenly distributed in both “terra firme” and wetter

“baixo” (low) areas of the forest. Their resin lumps tended to be flatter than the “breu branco” lumps. These lumps also blackened to a greater extent and were often covered with dirt and plant-like debris. Resin from

“sarara” type trees is commonly marketed as “breu preto” (black breu).

Results of the initial harvest at Tekohaw showed that the average amount of resin harvested from live and dead trees that had at least one resin lump was 856 ± 100g per tree (Table 3.3). A few trees yielded more than 9 kg of resin, but the median harvest amount of 420g per tree better reflects the fact that many trees 92 yielded relatively small amounts. In the first month of the harvest when data were recorded for trees with and without resin, this average was 562 ± 65g per tree. Sample sizes of known scientific species were too small to compare. A comparison of the main types of resin yielding breu trees showed that “sarara” trees had larger amounts of resin than “branco” trees with a p-value of 0.09 when their means were compared with a One-way

AOV. The number of resin lumps per tree for the two types was almost identical (“branco”: 2.9 ± .2 lumps/tree; “sarara”: 3.2 ± .2 lumps/tree). Several other types of breu trees, principally the “vermelho” (red) type, were relatively common, but since they never had resin lumps they were not included in the harvest survey or inventories.

Resin Harvest by Tree Condition, Size, and Location

The initial resin harvest at Tekohaw showed that tree condition, size and site each explain some variation in resin yield (Table 3.3). The clearest finding was that while only 12 dead trees were found among

183 trees examined, they had an average of twice as much resin on them as live trees (One-way AOV: p=0.01).

Resin yields in general increased with tree size, but the relationship was not a simple one. In the first month of the initial resin harvest when all potential resin yielding trees were examined, the percentage of live breu trees with some resin significantly increased with tree size class (χ2 = 31.91; df=4, p=.010). Only 9% of live trees in the 5 – 9 cm and 20% of live trees in the 10 – 19 DBH size classes had any resin on them. The percentage of live trees that were ≥ 20 cm DBH with resin was 68%. Regression analysis comparing resin amounts found on all live trees in the initial harvest to tree DBH was significant (R2=3.4%; p=0.015; n=171).

Increases in resin amount with size were gradual, though, so means were not significantly different between different size classes (Table 3.3). Non-parametric comparisons, however, showed that the median amounts of resin harvested were significantly greater from trees that were ≥ 20 cm DBH compared to trees in the two smaller size classes.

Resin harvest amounts also varied with the location of the trees at the Tekohaw site. The area searched most intensively had both “terra firme” and “baixo” type forest. Trees in the wetter “baixo” type 93 forest areas that were closest to the streams had almost twice the resin harvest as ones in the drier “terra firme” forest (Table 3.3)(One-way AOV: F=6.39; p=.012).

The other factor that logically influences resin collection success is harvest intensity. Apart from this study, there were no records of harvesting activity. Bark discoloration and bore holes on a trunk indicated where a resin lump was once attached, but it was difficult to tell whether a lump was removed by a harvester or fell off naturally and how long ago it came off. It was, therefore, hypothesized that harvesting intensity was proportional to distance from the village. Using GPS positions, the distance of each tree with resin from

Tekohaw village was calculated and plotted against the amount of resin found on that tree. The average amounts of resin harvested per tree increased with each kilometer (Figure 3.1), but overall such differences were not significant (Table 3.3). It is noteworthy that all finds of 3.5 kg or more of resin per tree were found at least 3.4 km from the village.

The Role of Weevils in the Formation of Breu Resin Lumps

During the resin harvests at Tekohaw, larvae were found in almost all fresh resin lumps. When larvae were present in the second round of resin harvest, there was an average of 1.3 ± 0.1 larva per resin lump

(n=137 lumps). In June of the initial 1998 resin harvest, 93% of resin lumps had weevil bore holes behind them. In a few cases, resin had apparently formed due to a burn or some other non-weevil related wound, but weevil larvae were found in a few lumps not associated with bore holes. The average weight of resin lumps with weevil bore holes behind them was 197 g compared to the 61 g average of resin lumps not associated with bore holes.

An examination of one pupa and seven adults collected from breu trees and resin lumps at Tekohaw showed these larvae were immature beetles that belong primarily or exclusively to a previously undescribed species of bark-boring weevil (Sternocoelus Kuschel new species #1; Curculionidae)(O’Brien, personal communication 2001). Other species of this weevil genus have been identified as stalk borers on manioc

(Manihot spp.) plants (da Costa Lima, 1956; Bastos et al., 1980), but this is apparently the first documented association with a Burseraceae forest tree. One adult specimen of a closely related weevil (Eubulus Kirsch species #1)(O’Brien, personal communication 2001) was also found on one breu tree. 94

Weevil larvae removed from resin lumps in the second round of harvesting had an average head width of 1.8 mm, average body width of 3.6 mm, and average body length of 8.2 mm (Table 3.4). The distribution of head widths (possibly indicating discrete instars) showed two peaks around 1.0 mm and 2.1 mm with a possible third peak at 1.4 mm (Figure 3.2). The weight of larvae measured in the field was 0.12 ± .02 g

(n=16). Measurements of a few pupae and young adults showed that mature weevils generally reach 10 mm in length (including the head) and weighed just over 0.2 g.

Resin associated with weevil attacks on breu trees was usually found in lumps affixed to the lower two meters of tree trunks although some lumps were occasionally seen higher up. Lower lumps were often wedged in a crook between the trunk and a buttress or stilt root. Weevil related lumps were quite distinct from the brittle streaks of resin that formed in machete cuts or small fissures of damaged trees. Lumps were typically 2 to 6 cm thick and averaged 339 ± 59g in the initial Tekohaw harvest. Most distinct lumps had only one developing weevil, but it was sometimes difficult to distinguish separate lumps in “sarara” type trees where resin lumps abutted each other. The weevil larva formed a cavity in the fresh resin that was 3 to 8 mm deep and extended almost to the edge of the resin lump (Figure 3.3). Yellow to greenish spots were sometimes found in fresh white resin that may have been larval fecal material. A few weevil pupae and pre-emergent adults were found in a rounded body-sized tunnel with smooth walls. Adult weevils that were not associated with resin lumps were usually found in the bark of rotting portions of fire-damaged or dead trees.

Weevil larvae apparently feed on the inner bark of breu trees by chewing a simple cylinder shaped hole toward the center of the tree. The mean number of bore holes was 10.1 ± 0.7 per resin lump in live trees

(n=171) in the initial resin harvest. While there was not a significant difference in the number of bore holes between “branco” and “sarara” trees, the average depth of bore holes was significantly greater in “sarara” tree lumps (7.1 ± 0.3 mm; n=81 lumps (776 holes on 19 trees) than in “branco” tree lumps (4.8 ± 0.2mm; n=27 lumps (226 holes on 13 trees) (One-way AOV; F=13.91; p=0.00). In the second resin harvest, there were 3.7 ±

0.2 bore holes per resin lump. A Fully Nested AOV examination of bore hole numbers and diameters measured in those harvests showed that differences among resin lumps accounted for a higher percent of the variation than differences among trees. 95

The best evidence of the link between weevil activity and the formation of harvestable breu resin at

Tekohaw was the relationship between the number and dimensions of weevils, their bore holes and the weight of resin lumps. The number of weevil larvae, sum of weevil head widths, sum of weevil body widths, number of bore holes, sum of bore hole diameters, and sum of bore hole depths all had highly significant correlation (p- values ≤ 0.01) with resin lump weight when compared with Pearson’s Correlation Coefficient and Regression

Analysis during the initial resin harvest (Table 3.5). The number and size of bore holes were somewhat better predictors of resin lump weight than larvae number and dimensions. As seen from the weevil larvae point of view in the initial resin harvest, the number and size of bore holes they chewed were generally better predictors of larvae number and size than the amount of resin generated in the process (Table 3.6).

Experimental Drilling

The drilling experiment was conducted to measure the amount of resin flow generated by mechanical action compared to the resin formed when the same amount of wood is chewed and removed by weevil larvae making typical bore holes. The trees in the drilling experiment included three “branco” type (DBH ave. 23.8 cm) and fifteen “sarara” type trees (DBH ave. 28.7 cm). During the initial resin harvest the “branco” trees in this experiment had an average of 660g per tree collected from them; eleven of the “sarara” trees yielded an average of 1200g per tree. The other trees had no resin previously collected from them. When the trees were revisited four weeks after drilling, little or no resin was found around these drilled holes. These results sharply contrast with resin lumps generated by weevil bore holes where an average of 20 g of resin per hole was found during the initial resin harvest. In 26% of the holes drilled, a tiny amount of resin (estimated to be less than

1g) had dripped out of the hole and dried on the trunk or was still leaking out of the hole. Drill bit size had no effect on the percentage of holes that had any resin coming from the drill hole. The percentage of holes with some resin on the trunk declined with increasing bore hole depth, but the difference was not significant.

Regression analysis showed that tree DBH (R2 = 5.4%; p=0.35) and the weight of resin harvested from the tree during the initial harvest (R2 = 1.8%, p=0.60) were also both poor predictors of the number of holes that had some resin on the trunk in this experiment. 96

THE RATE OF BREU RESIN ACCUMULATION

Resin Yields in Follow-up Harvests

The expectation of follow-up visits to breu trees 4 to 12 months after the initial harvest at Tekohaw was that resin harvests per tree would steadily increase over time. Due to the small number of trees harvested each time in this second round, this trend was not found. (Table 3.7). On trees with any resin taken during the second harvest, the average resin per tree was greatest in the group of trees revisited less than five months after the initial harvest. The mean amount of resin per tree and number of resin lumps per tree harvested eight months after the first harvest were greater than trees harvested more than a year after the initial harvest. A subset of trees harvested for a third time four months after their previous harvest produced the fewest number of resin lumps and the smallest amount of resin.

The time after the initial resin harvest was not by itself a good predictor of resin accumulation in these second round harvests, but tree size was again highly correlated with the number of resin lumps (a measure of weevil infection) and amount of resin taken off the trunk (Table 3.8). Harvest history as measured by the number of resin lumps and resin weight present at the initial harvest were even stronger predictors of these variables for the second harvest than tree size. The strongest correlation was found between the number of resin lumps during the first and second harvests with a correlation coefficient of 0.8 and p-value of 0.000.

While forest type was significantly correlated with resin weight during the initial harvest of breu trees, this factor was not significant in the second harvest round.

Resin Lump Growth Rate and Weevil Development Time

Monitoring the growth rates of individual resin lumps provided additional information on the rate of resin accumulation and the development time of weevil larvae. The wire perimeter technique used to measure resin lump basal area showed that this area and resin lump weight in the study trees were highly correlated

(Regression analysis R2 = 94.6%; p=0.00; n=22 lumps in 5 trees). Resin lump growth per day was estimated by dividing the difference in estimated weight of the lumps in March and July, 1999 by the number of days that lapsed between these two measurements. Regression analysis comparing these growth rates with the final 97 estimated weight (about 12 months after initial harvest) shows that resin growth rate steadily increased with resin lump weight until it reached a maximum around 0.56g/day at a resin lump weight of about 295g (Figure

3.4). Extrapolating the quadratic regression model to the point where the growth rate returns to zero (the point when larval feeding is assumed to stop with the onset of pupation) indicates that maximum resin lump size is around 589 g. An early suspension of the study prevented additional measurements to test this projection. The mean amount of resin per larvae in the initial harvest, however, was 263.6 ± 53.9g so it is likely that the average rate of resin formation sharply declines between 300 and 500 g. This pattern of resin formation indicates that Sternocoelus weevil larvae may spend one, two or possibly three years in a breu resin lump before pupating. The study provided no information on how long the pupa stage lasts before a young adult bores its way out of a resin lump.

Breu Resin Accumulation over Time

Data from the second round of resin harvests and monitoring growth rates of individual resin lumps were used to generate estimates of the time it would take for the average yield of resin per tree from a second round of harvesting breu trees to return to the average of the initial harvest. Tree type, tree size, forest type, initial harvest resin weight, initial harvest resin lump number, and time between first and second harvests were tested in regression models individually and in combination to assess their ability to predict the amount of resin harvested (or estimated to be present) during the second harvest. The model judged to be most useful because of the certainty of its variables and high predictability (p-value = 0.000) included tree size (DBH), weight of resin at first harvest, number of resin lumps at first harvest, and time between first and second harvests. The regression equation was applied to all live trees monitored during the initial harvest to estimate resin amounts that could be harvested from one to eight years after the first harvest (Table 3.9). The model predicts that following a first harvest it would take about four and a half years before a second resin harvest could obtain the same amount. The model assumes that the rate of resin accumulation remains constant regardless of the weevil development time and that no resin yielding trees will die during this period. 98

Other Animals Associated with Resin Lumps

Fly larvae identified as an undescribed species of Alipumilio Shannon (Diptera: Syrphidae)(Rotheray, personal communication 2001) were the principal other insects found associated with breu resin. They were found only in fresh “sarara” breu resin lumps in a section of the lump that was segregated from a weevil larva cavity (Figure 3.3). One distinctive feature of resin containing fly larvae is that maggot movement (and presumably feeding activity) in the resin transformed it from a white sticky cake-like material that breaks apart when pulled to a dark taffy-like substance that stretched. While it seems possible that these fly larvae may occasionally provoke some resin flow in breu trees, it was assumed that this makes a negligible contribution to the amount of resin that is harvested for commercial purposes. The relationship between this fly and the breu resin system are discussed in detail in Chapter 4 of this dissertation.

The most common visitors to the outside of resin lumps were several species of stingless bees

(Apidae: Meliponinae) of the genus Trigona (Venturieri, personal communication 2000). Bees were observed collecting resin from breu trees in the forest at Tekohaw (Figure 3.3) and from piles of resin laid out to dry in

Tekohaw village. They first hovered over the resin, landed on it, and then applied bits of fresh resin to the corbicula hairs on their hind legs. No attempt was made to quantify the amount of resin removed by bees from resin lumps on trees in the forest. The scope of bee collection of resin from piles of human harvested resin being dried in the village is discussed in the section on harvesting economics.

Assassin bugs (Reduviidae: Apiomerinae) were also commonly seen around breu resin. In the forest these bugs were sometimes observed waiting motionless on breu tree trunks near resin lumps (Figure 3.3).

These bugs appeared to have some resin on their forelegs. The presence of these bugs has no major implication for human resin harvesters. Their predation on stingless bees in the forest may reduce the amount of resin that these bees remove by a minor amount.

The other invertebrates associated with breu resin are those that use the chambers of hardened resin lumps left behind by weevils. One type of unidentified ant was the most frequent post-weevil occupant of resin lumps. When dirty old resin collected in the initial harvest was broken apart, these slim ants were sometimes found tending to pupae. Other invertebrates encountered in this process were flatworms, and numerous arthropods including spiders, scorpions, centipedes, millipedes, and earwigs (Overal, personal 99 communication 1998). The presence of these arthropods in dried resin has no major impact on human harvesting, although, harvesters breaking apart dried lumps need to exercise some caution to avoid painful bites and stings from some of these creatures.

BREU RESIN HARVESTING ECONOMICS AND MARKET

Breu Resin Harvesting Rates

Data from arrival and departure at breu trees during the initial harvest at Tekohaw showed that the average time to find any breu tree was about 5 minutes per tree (about 9 minutes to find one with resin). The density of breu trees that might have resin in this forest was estimated to be about 10 trees per ha (4 trees with resin) so an estimated 12 trees could be found in an hour of searching. This means that harvesters search about

1.2 ha of forest per hour. Once a breu tree is found, it takes about four minutes to look for and harvest any available resin. Harvesting was very quick when there are only a few readily accessible resin lumps, but the time was substantially greater when trees had numerous lumps that sometimes required more effort to dislodge.

The combination of search and harvest time meant that about 6.7 trees in 0.7 ha of forest could be harvested per hour.

Resin Drying Studies and the Effect of Stingless Bees

Sixteen batches of resin with initial weights varying from 2 to 9 kg lost the most weight in the first three days they were placed outside and tapered off to near zero from the fourth day on (Figure 3.5). The average mean percent weight loss after five days was 17.0 ± 1.2%. The ultimate weight loss was higher

(although not statistically significant) in batches with the largest percentages of “sarara” type resin in the mix.

Weight loss was assumed to be primarily due to the loss of volatile components of the resin. When resin was still sticky in the first few days of drying, however, stingless bees were frequently seen collecting resin.

In the experiment testing the affect of stingless bees on resin drying rates, resin samples in the partially screened enclosures lost a significantly greater mean percent of their original weight than the fully 100 screened samples (bee treatment: 14.1 ± .3% vs. screen treatment: 3.5 ± .4%; One-way AOV: F=489.13; p=.000)(Figure 3.6). The partially screened enclosures were visited by two varieties of stingless bees. During the first day most or all visits were made by a small yellow bee (mean weight 14.9 ± .7 mg; n=4) called “ipawa ruwer” by the Tembé (Tetragona sp. - now considered another variety of Trigona sp.)(Venturieri, personal communication 2001). This bee was the only type that visited the resin samples until the second hour of the second day of exposure. Larger black bees (mean weight 26.0 ± .6 mg; n=8) that the Tembé call “zawar iruwera” (jaguar bee)(Trigona sp.) then began to enter the open enclosures, and the small bees were never seen again for the remainder of the experiment.

Bees burrowed into moist resin and applied bits of it to the corbicula on their hind legs. They sometimes flew away from one piece and then alighted again to continue gathering resin. One black bee spent six minutes at the enclosure before flying away. The highest number of bees during spot checks was recorded during the second day of exposure when an average of 7 bees per enclosure were seen at one time (maximum of 46 bees during one spot-check in all enclosures combined). The average number of bees present during a spot check declined to 2 bees per enclosure during the third day of exposure (when heavy afternoon rain interrupted exposure time) and rebounded slightly to almost 4 bees per enclosure on the fourth and final day of exposure. The decline over the days of monitoring was, nonetheless, statistically highly significant (One-way

AOV: F=22.31; p=.000). During three days of bee monitoring, the number of bees seen at the open enclosures steadily increased from morning to afternoon observation times.

Breu Resin Sales

Up until 1996 Tembé men who harvested breu resin usually held on to their product until they or someone from the government Indian agency (FUNAI) could bring their sacks to a dealer in the town of

Gurupi or the city of Belém to sell. These opportunities were unpredictable and often involved personal costs for transportation and food. In 1997, two Tembé men in Tekohaw began to buy dried breu resin and titica vines from Tembé and other Indians in the Alto Rio Guamá and neighboring Alto Turiaçu reserves. Harvesters were usually paid $R 0.50 (about $US 0.50) per kg of dried resin. Most chose to receive payment in the form of merchandise acquired by these Indian brokers during their selling trips. 101

A sales logbook kept by the larger buyer in Tekohaw showed that 16 people (or small groups of friends or relatives) sold a total of 1519 kg of breu resin to him between July, 1999 and February, 2000. The mean amount sold by each person or family group was 95 ± 33 kg (median 29 kg; maximum 441 kg). There were 39 transactions during this time with each customer accounting for an average of 2.4 sales (median 1.5; maximum 8) with 39 kg (median 6 kg; maximum 200 kg) per sale.

This one Tembé broker sold one lot of 1000 kg of breu resin in 1998. He sold additional lots of 180,

300, and 770 kg of resin in 1999. When he sold to the dealer in Gurupi, he received 25% of the value of the sale in cash up front and the rest in merchandise. The dealer in Gurupi usually paid $R 1.00/kg for pure batches of “branco” resin or $R 0.70 for pure or mixed batches of “sarara”/”preto”(black) type resin. Dealers in Belém generally offered $R 1.00/kg regardless of resin type and paid 100% of the sales in cash. These buyers, however, usually deferred payment for several weeks or months until they sold the product to a dealer in another city.

The main dealer in Belém had a retail store where he sold small quantities of breu resin and dozens of other medicinal forest products. He sells most of his resin as a wholesaler at $R 3/kg to other dealers who supplied shipyards in larger cities on the northeast coast of Brazil. My conversations with the owners of several boat repair yards in Belém and the nearby port town of Icoaraçi confirmed that the type of fresh resin obtained from Protium and closely related Burseraceae trees in the region is primarily used by commercial boatyards for caulking the sides of wooden boats that sail in salt water. Boatyards in the Belém area that service boats sailing in the Amazon River or brackish estuary used either a petroleum based product

(“betume”) or the amber like “breu pes” to caulk decks and sides. This may come from a hardened

Burseraceae resin, but the source of the material has not been identified. Even though these materials were a little more expensive, the Belém repair people preferred to use them over the Protium derived resins because the “branco” and “preto” types usually needed to be heated and filtered before they could be used due to their abundant gritty impurities. One man also said that heating these resins over fire was more dangerous since overflow could ignite and quickly flare up (Tembé did not mention such problems in their use of breu for caulking canoes and wooden launches). 102

Breu Resin Harvest Models

Data on the density and size structure of breu trees, resin harvest amounts and rates, and marketing factors were used in three models to estimate the amount and value of resin that could be harvested in a given area and time depending on the average density of trees.

The first model (Resin Harvest Model – Part 1:Table 3.10) estimates the amount of resin that would be found on breu trees in a particular size class by multiplying the probability of finding resin on a tree of a given size class by the average amount of resin found on trees of that size class. As was seen in the initial resin harvest, the probability of finding resin on a tree and the amount of resin found on trees that had at least one lump both increased with the size of the tree. These estimates of size class specific averages of resin per tree range from just under 100 g per tree for trees in the 10-19 cm DBH size class to just over 1 kg per tree for trees that are greater than 40 cm DBH.

The second model (Resin Harvest Model - Part 2: Table 3.11) predicts that as breu tree density increases from 1 to 50 trees per ha, the amount of marketable resin that can be collected will vary from 0.3 to

16.5 kg per ha and the value to a harvester will range from $0.14 to $6.85 per ha. While breu tree density was about 10 trees per ha at Tekohaw, this model covers the wider range of densities of potential resin yielding

Burseraceae trees in other parts of the Brazilian Amazon (Table 3.12). These area estimates of yield and value are generated by first multiplying the percentage of each size class in a group of breu trees by the total density of trees to estimate the number of trees of a particular size class that would be present at a specific density.

These estimates of the number of trees of each size class per ha would then be multiplied by the estimated amount of resin per tree at that size class (derived from the first model) to predict the amount of resin that would be found in a hectare of forest from the trees in each size class (Figure 3.7). The model predicts that the two middle size classes (20 – 29 and 30 – 39 cm DBH) would yield the most resin because they are relatively common and have relatively high probabilities of offering large amounts of resin. Although they are numerous, smaller trees would contribute less because they rarely have much resin. Larger trees are almost certain to offer substantial amounts of resin, but they would contribute relatively little to overall harvest because they are relatively rare. 103

The expected amounts of resin from each size class are then totaled to present the expected amount of resin per ha at breu tree densities ranging from 1 to 50 trees per ha. Freshly harvested resin amounts were then multiplied by 0.83 to account for the lost weight that occurs during resin drying (Table 3.11) to predict the amount of marketable resin that could be obtained per ha of forest. Values of resin per ha were obtained by multiplying these dried resin amounts by the $ 0.50 per kg price harvesters are typically paid for this product.

Values would be proportionally higher if they were based on the higher prices involved in retail level transactions.

The third model (Resin Harvest Model - Part 3 (Table 3.13; Figure 3.8) predicts that a harvester could gather 1.6 to 16.8 kg of marketable (dried) resin per day worth $0.78 to $8.37 per day when the density of breu trees ranges from 1 to 50 trees per ha. These estimates are first based on a constant search rate of 200 m2 per minute and a four minute per tree harvesting time. At high densities a harvester would be able to collect resin from a large number of trees each day because they will not need to spend as much time searching for trees as when the density is low. Resin harvest amounts are based on the predicted average resin yield per ha at different densities presented in Table 3.11.

The second part of this model (Table 3.13) projects resin yields (fresh and dried), the number of days it would take one person to completely harvest a 1000 ha block of forest and the potential revenue available in this area at different densities of breu trees. This is an area of forest that is easily accessible within an hour’s walk of a village. At the lowest density (1 tree/ha), an area this size could provide 279 kg of marketable resin worth $140, but all of the available resin could be harvested in six months. At the highest density (50 trees/ha), harvesters could procure almost 14,000 kg of marketable resin worth almost $7000, and it would take one person about 30 months to complete a first round harvest of all resin bearing trees.

The studies of resin rates of accumulation show that breu trees will probably need at least four years between harvests for resin to rebuild to first harvest levels. The final part of the third Resin Harvesting Model predicts that with a four year resin regeneration cycle, some 1752 ha of forest would be needed to sustain one person engaged in a full-time pursuit of breu resin collection even at the 50 breu tree per ha density. This amount would be over 8000 ha if the density was as low as 1 tree per ha. These areas provide baseline 104 estimates that could help assess the viability of ongoing resin harvesting in an area or development of management guidelines.

DISCUSSION

RESIN HARVEST AMOUNTS PER TREE AND AREA OF FOREST

Breu Tree Density Inventories

Inventories in the Brazilian Amazon region show that trees from the Burseraceae family are often the most species rich and numerous with importance values that frequently rank it among the top four families.

While Burseraceae trees are well represented in the forests around Tekohaw, the estimated density of 10 breu trees per hectare seems low in comparison to inventories in other parts of the region where Burseraceae tree density often exceeded 50 trees per ha (see references for Table 3.12). The percentage of those species and trees that would possibly yield harvestable resin is about 50%. Forests in the eastern Brazilian Amazon seemed to have a higher percentage of such trees than forests in the western part of the region where generally non-harvestable species such as Tetragastris altissima were prevalent.

These inventories could ideally be used as a basis for predicting amounts of harvestable resin on a regional level, but it may be problematic to assume that high densities of particular species found in 1 ha inventories would be maintained over a landscape scale. Inventories of 100 ha plots in Guyana showed that

Protium trees were present in 18 out of 23 forest types spread throughout the country (ter Steege and

Zondervan, 2000). When present, the density of this genus varied from 3 to 217 trees per 100 ha with an average density of 0.53 trees per ha.

Resin Harvest by Tree Type

Most species of Burseraceae produce resin in the twigs and bark, (Gentry, 1993), but even among the

35 species of Protium genus trees in the eastern Amazon, only some of these produce harvestable quantifies of 105 resin (Daly, 1987; Balée and Daly, 1990; Balée, 1994). This study confirmed the importance of P. giganteum and P. pallidum as major resin producing species that Balée (1994) found were key commercial sources of resin for the Ka’apor Indians who live in the neighboring Alto Turiaçu Indigenous Reserve. The Ka’apor also collect resins from other species of Protium and Trattinnickia for use as an incense, medicine and flammable material, but it is not known what quantities of these materials are harvested or sold.

At the forest site near the town of Mojú, I found resin accumulation on P. giganteum (classified by local residents as “breu branco”), P. paniculatum Engl., P. guianense (Aubl.) Marchand, P. pilosum (Cuatrec.)

Daly (locally called “breu preto”), Tetragastris altissima (Aubl.) Swart, and Trattinnickia rhoifolia Willd.

(locally called “breu siquereireua” or “sucuruba”). None of the trees identified as P. krukoffii Swart, P. pernavatum Cuatrec., Trattinnickia lawrencei Swart var. bolivianum, or Trattinnickia burserifolia Mart. at

Tekohaw or Mojú had any resin on the trunk.

According to literature about Burseraceae species and resin production, the taxa most likely to yield harvestable amounts of resin are Protium species in the Section Icica Aublet, Section Icicopsis (Engl.)Swart,

Section Sarcoprotium Daly, and Protium trifoliolatum group (Daly, 1987, 1989, 1992; Harley and Daly, 1995) and select other species in Protium and other genera.

While many Burseraceae species have been considered a source of harvested resin (Appendix 3-A), the only estimate of the amount of resin that can be harvested from Burseraceae trees without human intervention is given in an economic valuation of nontimber forest products in Ecuador (Grimes et al., 1994).

This study indicated that 32 P. fibriatum Swart, P. nodulosum Swart, and P. sagotianum Marchand trees in two upland and one alluvial plot could provide an average of 730 g of resin per tree per year. The authors cautioned that this amount was based on ethnobotanical surveys with Indian collectors, not on actual yield studies, and that even experienced collectors did not agree on which species provided harvestable resin.

Resin Harvest by Tree Condition, Size and Location

Tree characteristics associated with resin production in these tropical angiosperm Burseraceae trees correspond well to findings in temperate conifer research. The starkest difference in resin amounts between two groups of trees in this study was between live and dead trees. This is consistent with a lodgepole pine 106

(Pinus contorta Dougl.) study where diseased trees had higher resin flow than healthy ones (Nebeker et al.,

1995). While Sternocoelus weevils’ preferential attacks of weaker trees may further weaken these trees over time, they do not appear to quickly kill their host tree as some temperate wood-boring beetles do. Among live trees at Tekohaw, larger (presumably older) trees were more likely to be attacked by weevils than smaller

(younger) ones, and when they were, they produced more resin per tree (Table 3.3). The positive correlation between tree size and resin flow paralleled results in studies with lodgepole and Scots pine (Pinus sylvestris L.)

(Lieutier et al., 1993; Nebeker et al., 1995). Sternocoelus preference for attacking a specific height section of the host tree’s bole was also found in two studies involving the southern pine beetle (Dendroctonus frontalis

Zimm.)(Coleoptera: Scolytidae) and its loblolly pine (Pinus taeda L.) host (Coulson et al., 1976; Fargo et al.,

1978). Differences in resin composition and flow rate have been documented at different heights in loblolly pine, but these differences have not yet been linked to height preferences in beetle attack distribution (Schmitt et al., 1988;Tisdale and Nebeker, 1992). No observations were made of the Sternocoelus weevils in flight, but their flying capabilities are presumed to be very modest (O’Brien, personal communication 2000).

Accordingly their tendency to colonize lower sections of the trunk may be due to this limitation and an attraction for ovipositing in protected crevices around tree buttress and stilt roots.

Resin harvest amounts were higher at Tekohaw in “baixo” (sometimes seasonally flooded) forest than in more upland “terra firme” type forest. Average tree size in the “baixo” forest type, however, was larger than in the drier areas. It may be that weevil attacks and resin flow were greater in these areas simply because larger trees were found there.

The Role of Weevils in the Formation of Breu Resin Lumps

Bore Hole Patterns and their Implications about the Breu Tree Resin Defense System

Living close to and possibly ingesting resin may give Sternocoelus larvae certain benefits such as physical protection from outside predators, chemical protection against microbes that often attack larvae

(Evans, 1975), and a source of special materials for pheromones (Slansky, 1992). In spite of these potential benefits, though, it is unlikely that the weevils would want to maximize resin flow while feeding since coping 107 with this viscous and terpene rich material would impose unavoidable energy costs. A weevil would, therefore, logically try to maximize nutritive tissue intake while minimizing energy expenditures chewing wood and dealing with fresh resin. The number and size of bore holes and larval development rate, therefore, reflect the interaction between the weevil feeding strategy and the breu resin defense system. One question that arises is whether or not resin production varies more with the type and condition of the tree or with the intensity of the attack. The larger average size of trees might explain why some trees would be attacked more readily than others, but it does not sufficiently explain why resin lumps are bigger on some trees than others.

One interpretation of the greater resin harvests from larger trees is that they are more vulnerable to weevil attacks than smaller trees. The contrary may be true if larger trees produce more resin because weevils need to feed more extensively on them to gain a comparable amount of nutrition. Cross-sections of resin lumps show distinct layers so they must have been formed by numerous feeding and resin flow events. The advantage of chewing the same hole wider and deeper is that chewing energy is presumably only expended in nutrient rich phloem (Coulson and Witter, 1984). One study with Norway (Picea abies (L.)Kar.) and Sitka spruce (P. sitchensis (Bong.) Carr.) trees even showed that the nutritive value of inner bark tissue actually increased in response to wounding (Wainhouse et al., 1998). The downside of staying in the same hole is that resin flow also grows with increases in bore hole size and depth (Table 3.5). It seems that weevils must be able to gauge the benefit-cost ratio for chewing deeper into a given hole. Once this ratio drops below some minimum value it seems better to start a new hole even though this means incurring “start-up” costs chewing through 5 mm of a new section of non-nutritive outer bark that may have its own set of potent structural and chemical defenses.

In addition to variation in bore hole number and size between trees, these differences also exist for resin lumps on the same tree. This may be due to microsite differences in resin structures on the trunk. Breu resin system anatomy has not been closely examined, but studies of a few coniferous trees provide possible explanations. Some conifers have both radial (horizontal) resin ducts in the phloem or cambium and vertical resin ducts in the xylem. The density and size of these ducts can vary seasonally and from place to place on individual trees (Schroeder, 1990; Blanche et al., 1992; Lieutier et al., 1992; Tisdale and Nebeker, 1992; Baier,

1996). Resin viscosity, exudation pressure, or crystallization rate may also vary within and between trees 108

(Baier, 1996). While existing resin ducts (the constitutive defense system) can deliver some preformed resin to a wound site, the wounding process may induce the formation of new resin ducts or additional resin flow in the original ones (Schroeder, 1990; Popp et al., 1991; Lieutier et al., 1992, 1995; Baier, 1996; Ruel et al., 1998;

Tomlin et al., 1998; Christiansen et al., 1999; Phillips and Croteau, 1999; Nagy et al., 2000). Boring a hole into a site with an especially high amount or objectionable quality of resin could, therefore, prompt a

Sternocoelus weevil to start a new hole sooner rather than later.

Several conifer studies have further shown that wounding mostly induces formation of vertical resin ducts and build-up of resin in a vertical orientation (Blanche et al., 1992; Lieutier et al., 1992; Tisdale and

Nebeker., 1992). While it can take six weeks for resin production to become fully functional in newly formed vertical resin ducts (Blanche et al., 1992), induced resin may be more potent than the constitutive type since it can have more numerous and higher concentrations of chemicals that may deter or kill insect attackers and associated fungi (Lieutier et al., 1995; Phillips and Croteau, 1999; Franceschi et al., 2000; Nagy et al., 2000).

These factors may explain why at least one type of bark beetle chews new galleries in its host tree in a horizontal direction (Lieutier et al., 1992). By moving laterally rather than vertically it may be able to keep in front of induced resin defenses. Since Sternocoelus weevil bore holes are also often oriented in a horizontal direction, breu trees may share some resin system features with temperate conifers. Considering these inter- specific, inter-tree, and intra-tree variations in resin production, it is not surprising that resin flow is more closely related to the number of bore holes and sum of bore hole diameters than to larvae size parameters

(Table 3.5).

The occasional finding of weevil larvae in resin lumps not accompanied by bore holes bears some scrutiny. Without a bore hole it seems hard to imagine how the larvae could access inner bark tissue. It also raises the question about how the resin came to be formed on that spot. The first issue suggests that larvae are able to gain some nutritional value directly from the resin. No analysis has yet been done, however, to indicate whether the liquid that exudes from wound sites contains some nutritional sap in addition to the documented oleo-resin constituents. In the case of wounded trees, however, it seems possible larvae could feed on bark tissues or microbes in the sap without making a bore hole. This feeding habit has been well established for the sub-family of fly larvae found on these and other tropical trees discussed in Chapter 4. The observation that 109 one chain saw wounded T. rhoifolia tree at the Mojú site had fewer bore holes per resin lump than unwounded trees demonstrated this possibility.

Weevil Attack Pattern and Life-History

This study showed that usually only one weevil larvae occupied a resin lump with a single chamber.

Since most trees visited in the second round of harvesting only had several resin lumps per tree, it appears that female weevils only oviposited one or at most several eggs at one site. Results also showed that the distribution of attack on host trees by Sternocoelus weevils was not like several pine beetles that depend on mass attack to successfully colonize (and ultimately kill) a host tree (Raffa and Berryman, 1987; Phillips and

Croteau, 1999). The observation that fresh resin lumps on “sarara” trees often form near other fresh or abandoned ones does show that a new weevil may gain some advantage by being near the place where another weevil is or was active. This idea is reinforced by the finding that even trees that had all resin (and presumably all larvae) removed during the initial harvest were recolonized in proportion to the extent they had been colonized before (Table 3.8). The extreme example of this was one tree that had 28 resin lumps removed from it in the first harvest, but eight months later it had already accumulated another 49 resin lumps.

The distribution of weevil larvae head widths indicated two to three peaks. While a beetle body grows steadily throughout the larval stage, the sclerotized head capsule is fixed at the beginning of the molt.

Within a population there is often a consistent progression of sizes of these hard parts that corresponds to each new instar (O’Brien personal communication 2000). If this is true for this species of Sternocoelus, it undergoes two or three instars during its larval stage. Three instars is a typical number for many weevils

(Evans, 1975).

The finding that adult weevils were only found on rotting portions of damaged or dead breu trees has potentially important implications for the population dynamics of this species. If adult weevils do need dead host trees for their stage of the life cycle, this requirement could limit the expansion of the Sternocoelus weevil population if its dispersal range is restricted by poor flying ability. The major forest fire that swept through hundreds of hectares around Tekohaw in the early 1980s burned but did not immediately kill several breu trees found in this study. It seems possible that this fire created more potential adult weevil host trees than would 110 exist under normal rates of tree mortality. As the forest recovers, the number of suitable host trees for adults could decline.

Experimental Drilling

This study’s drilling experiment showed that one time drilling of potential resin producing trees yielded no more than miniscule amounts of resin. Throughout the overall study, resin was sometimes seen in machete cuts made on breu trees by passing Indians, but there was never much resin in these wounds and it was always dry. These observations match those of a tree wounding study with other types of Burseraceae trees in Panama where machete cuts were quickly clogged with resin (Guariguata and Gilbert, 1996). Since weevil generated holes at Tekohaw produced an average of 12.5 g resin per hole (n=246 holes) compared to less than 1 g from the drilled holes, the weevils apparently induce greater resin flows by repeatedly chewing into the breu tree. This is consistent with the observation of layers in resin lumps. Whether weevil stimulation of resin flow is due to mechanical wounding alone or is supplemented by another agent such as a fungus remains to be seen.

Martinez-Habibe (1998) showed that Ticuna Indians in Columbia also harvest resin from various species of Protium that have apparently been formed by weevil attacks. Accounts that other Amazonian natives slashed breu trees with a machete and later collected resin (Mors and Rizzini, 1966; Spruce, 1970;

Daly, 1987), however, indicate that weevils may not be necessary to provoke resin flow in some types of breu trees. Neels (2000) showed that indigenous people in Guatemala successfully harvest resin from Protium copal by scraping out a series of 2 cm wide circles in the bark and reopening the wounds every three days for several months. She found that these trees yielded an average total of 246 g per tree after being wounded once a week for 14 weeks. Resin flow increased every week for the first five weeks and then leveled off producing an average of 17.5 g per tree per collection. Since resin collection is done on a seasonal basis in the region, it was hard to predict if these yields could be maintained throughout the year. While resin collection in the region is generally done on a seasonal basis. There are many other examples of tropical and temperate trees that yield commercial quantities of resin when they repeatedly wounded (Howes, 1949; Langenheim, in press). 111

THE RATE OF BREU RESIN ACCUMULATION

Resin Yields in Follow-up Harvests

The harvest history of breu trees was a stronger predictor of the amount of the next harvest than knowledge of its type, size, and site characteristics. This was shown by strong associations between the number of resin lumps and resin amount in the initial and follow-up harvest of the same trees (Table 3.8).

These relationships are good evidence that the extent to which a tree has been colonized by weevils before will strongly influence the extent to which it will be colonized again. There are three possible explanations for these connections. One is that a tree that was suitable for infection once is simply more likely to be considered suitable as a host tree again to adult weevils searching for a tree to colonize. The second possibility is that infected trees with or without current weevil infestation release more resin volatiles that attract beetles in search of ovipositing sites than ones that have never been infected. The third possibility is that harvesting did not remove all weevil larvae and apparent recolonization was in fact merely a continuation of the growth of larvae that escaped the first harvest.

The low resin per tree and resin per lump values in trees harvested for a third time in this study present a new challenge to explain. Some characteristics of the trees in this group fit well with an expectation of high resin, while others would indicate the opposite. Four months is not a long time to expect high levels of colonization and resin build-up, but other factors may be operating. One possibility is that two rounds of harvesting these specific trees and intensive harvesting in the area may have removed all larvae from the target trees and possibly reduced the weevil population size in the area. While insect numbers are sometimes thought to be almost limitless, it is not inconceivable that an insect highly specialized on a few species of trees in a species rich forest could become vulnerable to intensive harvesting of its larvae.

The other possible explanation for the third harvest’s low resin yield is that seasonality affects resin production. Several studies with temperate conifers have shown that resin production and susceptibility to beetle attacks can vary between seasons and years depending on rainfall levels (Blanche et al., 1992; Lorio et al., 1995; Baier, 1996). Trees suffering from severe drought or other stresses may become more vulnerable to insects or pathogens. On the other hand, trees may actually develop more resistance to attacks in regular dry 112 periods because they can devote more carbon to making defensive compounds than in wetter times when resources may be preferentially devoted to growth and reproductive effort (Dunn and Lorio, 1992; Tisdale and

Nebeker, 1992; Lorio et al., 1995; Nebeker et al., 1995; Baier, 1996; Feeney et al., 1998). This principle may be a factor here since the strong follow-up resin harvest in November 1998 was preceded by a five month dry season period after the initial harvest in June and July. The weak third round harvest in July 1999 was preceded by the last four months of the rainy season following the second harvest in March.

Resin Lump Growth Rate and Weevil Development Time

The resin lump growth study could not precisely measure the amount of time that Sternocoelus weevils spend in the larval stage, but data suggest a minimum of one year and possibly as many as three years.

These weevils have evolved many mechanisms to cope with a noxious resin, but they are probably not immune to its deterrent properties. Even highly specialized pine beetles that use resin elements to make pheromones can succumb to host tree resin defenses (Borden, 1982; Schroeder, 1990; Dunn and Lorio, 1992; Hui and

Zhimo, 1995; Phillips and Croteau, 1999). There is wide variation in the generation time of beetles that may result from the quality of their food resources or periods of dormancy (diapauses) induced by lapses in food availability (Evans, 1975). Some bark beetles in the genus Dendroctonus can complete 4 to 7 generations in one season (Stark, 1982) while other beetles may take five years or more to complete their life cycle (Evans,

1975).

Breu Resin Accumulation over Time and the Impact of Harvesting

Estimating the sustainable harvest rate of a plant-based non-timber forest product generally requires knowing how long one needs to wait to reharvest a product after the previous round of harvesting and what limits may need to be imposed to avoid negative impacts on the plant population’s reproductive success

(Peters, 1994). In the case of breu resin, this task is more complex since resin formation is connected to both the trees’ capacity to produce the material and weevil actions that stimulate it. Harvesting has a potentially greater effect on Sternocoelus weevils than the breu trees themselves because weevil larvae are usually taken along with the resin. 113

This study’s finding that weevil development is very slow has clear implications for the management of breu resin harvest. Systematic harvesting of fresh resin with live larvae can set the clock back on resin production for a considerable amount of time. Repeated harvesting in one area could reduce the Sternocoelus weevil population and the amount of resin that could be harvested. While these weevils prefer to colonize older and weakened breu trees, it is not known if their attacks accelerate their decline. It would be interesting to know if a moderate level of resin harvesting (as generally practiced by the Tembé) actually helps breu trees remain vigorous longer than if weevils were allowed to colonize trees to their full natural extent.

Other Animals that Utilize Resin Lumps

The only other animals found in fresh resin breu lumps beside weevils were larvae of a syrphid fly.

Larvae of this family have been previously found in tree exudates in both tropical and temperate regions

(Thompson, 1969; Ferrar, 1987), but there is only one case where fly activity directly contributed to resin flow

(Hellrigl, 1992)(See Chapter 4 for details).

My observation that stingless bees collected resin from various types of breu trees in the forest at

Tekohaw and Mojú was also noticed by Neels (2000) in her study of P. copal resin harvest in Guatemala. She estimated that these bees (possibly Trigona spp.) removed as much as 25% of the resin that had oozed onto the trunk one week after the trees were wounded in resin harvest experiments. Stingless bee collection of resin from a variety of plants has been noted throughout the tropics (Roubik, 1989; Nogueiro-Neto, 1997). Apart from this study, reduvid assassin bugs have also been seen using resin to facilitate their capture of bees and other insects in Brazil and other parts of Latin America (Adis, 1984; Hogue, 1993).

Some wild mammals also use resin from Burseraceae trees. In Panama, coatis (Nasua narica L.) groom themselves with the aromatic resin of Trattinnickia aspera (Swart), possibly for its pharmaceutical properties (Gompper and Hoylman, 1993). Wild peccaries regularly chewed resin masses on Protium copal trees in Guatemala (Neels, 2000). 114

Should Sternocoelus Weevils be Considered Beneficial Insects?

Weevils are probably the world’s most diverse taxonomic family of organisms with more than 50,000 identified species (Hogue, 1993; Anderson, 1995), but the habits of most are scarcely known. Some weevils and other bark beetles (mostly Scolytidae) have achieved notoriety by colonizing and sometimes killing coniferous and hardwood trees in spite of these trees’ defensive systems of sticky and toxic resins (Storer et al.,

1997; Ruel et al., 1998; Tomlin et al., 1998; Phillips and Croteau, 1999). Various weevils are also considered pests on tropical crops including bananas, rice, coffee, cotton, tea, sugar cane, yams, and palm trees (Lamb,

1974). With the possible exception of an aquatic weevil tested for controlling Eurasian water milfoil (Creed and Sheldon, 1995; Sheldon and Creed, 1995), the term beneficial insect has rarely been applied to weevils or other close herbivorous relatives.

The term beneficial insect is usually reserved for insects that provide pollination services, produce honey, recycle nutrients, prey on pest insects or serve as a direct source of human food, medicine or dye material (Metcalf, 1993). An insect or pathogen that causes economic damage to human agricultural or tree crops is understandably called a pest. The beneficial label, however, should be extended to phytophagous insects whose attacks stimulate production of plant compounds that are in turn used by people, wildlife and other insects including key pollinators as long as such attacks do not significantly hasten the mortality of the host plants..

The Sternocoelus weevil should firstly be considered a beneficial insect because it is responsible for directly generating a significant economic resource for many forest peoples. Secondly, this weevil is providing an indirect benefit to the forest community by stimulating resin collected by stingless bees that pollinate a variety of rainforest trees (Roubik, 1989; Nogueiro-Neto, 1997). Since stingless bees are common pollinators of Protium trees (Daly, 1987), weevil attacks that provoke resin flow may actually provide some benefit to the offended trees. Provoking resin flow may also provide an essential material for the larvae of the syrphid flies that live in the fresh resin of some breu trees and probably also pollinate various rainforest plants (Thompson,

1969; Ferrar, 1987). 115

BREU RESIN HARVESTING ECONOMICS AND MARKET

Breu Resin Harvesting Rates

The amount of time required to find breu trees with resin in a species-rich forest depends on the skill of the searcher and the density of such trees in the forest. It is difficult to quantify the search efficiency of a

“typical” harvester because some individuals had a much stronger intuitive sense of where to find such trees than others. The seasoned harvesters seemed to have a finely tuned search image for breu trees that helped them pick out trees that might have resin at a much greater distance than their less experienced colleagues.

When a harvester found a breu tree of the type that might have resin, he first stopped to examine the trunk and stilt roots for resin stuck to the bark and look around the ground for chunks of dried resin that have fallen from the tree. If resin was found, additional time was spent removing resin lumps from the tree by hand or with a machete. If resin was seen higher up on the trunk, a harvester sometimes cut a long sapling to dislodge the elevated resin lumps. Since time measurements of searching and harvesting were made with a research team, the accuracy of the estimates of these average times could be improved by accompanying actual lone harvesters at work.

Resin Drying Studies and the Effect of Stingless Bees

The loss of weight when harvested breu resin is laid out in the sun is potentially due to three factors: loss of volatile components, evaporation, and removal by stingless bees. Various essential oils have been identified in the resin of several Protium and Tetragastris species (Zoghbi et al., 1993; 1998; Siani et al., 1999a,

1999b) so the release of aromatic substances from fresh resin is to be expected. Water soluble carbohydrates

(gums) have been found mixed with resins in other Burseraceae genera (principally Boswellia and

Commiphora)(Khalid, 1983), but studies of the Protium have apparently not investigated its resin for materials that could evaporate.

Since conventional outdoor drying methods lead to a 15 - 20% weight loss in harvested resin, determining the reasons for these losses could have financial implications for the harvesters. The greater weight loss of resin samples in partially screened enclosures exposed to bee removal than samples in fully 116 screened enclosures showed that bee resin removal was apparently responsible for just as much weight loss as chemical processes. Even if bee removal was the main cause for the difference in weight loss in the resin drying enclosure experiment, it cannot be assumed that bee removal rates from four 36 g samples of resin could be extrapolated to hundreds of kilograms of resin laid out to dry in a single village. Further evidence of large-scale bee removal would be needed before any alterations in drying methods would be justified on economic grounds. Allowing stingless bees full access to harvester resin is certainly providing an important material to one group of the forest’s most ubiquitous group of pollinators.

Breu Resin Sales

It appeared that the total breu resin harvest in Tekohaw and other indigenous communities in the

Gurupi region is several metric tons per year. These communities have access to tens of thousands of hectares of relatively undisturbed forest so the current scale of harvesting is probably sustainable at the landscape level.

Forests near villages and areas close to river access points have probably been harvested most often, so there may be local areas where Sternocoelus weevil populations have been depleted and profitable harvesting will not be possible without a hiatus of many years to permit populations to rebuild. The Tembé and Ka’apor

Indians readily admit they need to spend hours in the forest in more remote areas in order to make a commercial breu resin collecting effort worthwhile since the final product only gives them $0.50/kg. The resin has a long tradition as a medicine and great utility in caulking canoes, but the market demand for it in these uses is limited. The resin’s principal area of supply seems to be in the wet forests of the eastern Amazon, but the fact that boat repair yards in Belém prefer alternatives that are more expensive than Protium derived resin shows that the prospects of more sales to the boat repair market in that area are not good. It is, thus, the lack of demand and consequent low price that limits the amount of harvesting effort more than the limit of the resource.

The price paid for breu resin as a raw material for caulking boats in the eastern Brazilian Amazon is much lower than the $US 9.11 per kg paid for resin from various Protium species in Ecuador (Grimes et al.,

1994). Several groups of Quichua Indians have long used this resin as a ceramic glaze (Whitten, 1976; 117

Whitten and Whitten, 1988), and the collection and sale of the resin for this purpose has made it the most important economic product for the Quijos Quichua in the Napo Region (Grimes et al., 1994).

While resins from various species of Burseraceae have long been used by indigenous people and other forest dwellers as an incense and folk medicine, it seems ironic that the major commercial use of breu resin to caulk wooden boats requires driving off most of the aromatic essential oils. A phytochemist at the Museu

Goeldi in Belém is now analyzing different types of breu resin and other Amazonian plants for their possible uses as scents in the perfume industry (Zoghbi, personal communication 1999). If this screening process is successful, it could increase demand for the resin. This demand might not be sufficient to increase the price as well, but it could at least allow more people to generate modest income from collecting this resin.

Breu Resin Harvest Models

The first two Resin Harvest Models use data on the size class distribution of breu trees at Tekohaw and the average amounts of resin that were found on each size class. While the format of these models may have general value in predicting the amount of resin that could be harvested in other locations, the estimations would need to be adjusted to reflect differences in size class structure and the probability of finding resin on trees in a certain size class. Observations made at the Mojú site indicated that rates of infection by weevil larvae there were considerably lower than found at Tekohaw. Neels (2000) estimated that the annual yield of resin from P. copal trees in Guatemala with a bark wounding harvest method would be about 2.4 kg per ha where the density of productive trees averaged 12.8 trees per ha. This is only slightly less than the 3.3 kg per ha yield of fresh resin that was predicted for an initial harvest at Tekohaw with an estimated density of 10 potentially productive trees per ha.

The third Resin Harvest Model indicates that a harvester can only expect to consistently earn the

Brazilian minimum wage ($R 7.00 per day) by collecting breu resin if he is working in a forest that has at least

25 breu trees per ha that potentially yield harvestable resin. This is based on spending five hours on harvesting activities per day and probably includes several extra hours walking to and from a village or camping spot to a prime collecting area. Some Tembé collectors stated they have filled several sacks of breu resin weighing 20 118 kg each in a day’s hard work, but such events seem to be unusual days of good fortune for the most experienced harvesters and does not reflect the average yield of harvesters in general.

The most sobering result of the third Resin Harvest Model is that incorporating a four to five year hiatus on harvesting the same tree requires that thousands of hectares of forest would need to be available to sustain even a few harvesters on a full-time basis. Since the Tembé and many other forest dwellers rely more on subsistence than commercial activities to support their lifestyle, breu resin harvesting offers a modest but welcome source of cash income to dozens of people rather than full-time paid work for a few. The forests could provide more resin than is currently being harvested, but this surplus seems destined to remain intact unless the demand for the material substantially increased.

CONCLUSIONS

Sustainable harvest of a plant-based non-timber forest product typically depends on managing the harvest within the limits of an individual plant’s capacity to produce the harvested product, its tolerance of the harvesting method, and the capacity of the plant’s population to withstand the harvesting intensity (Peters,

1994). In this case where commercial quantities of breu resin are only produced when trees are attacked by a specialized bark-boring weevil, sustainable harvest of the product will also require maintenance of the weevil population. Routine resin harvesting removes many weevil larvae so intensive harvesting could eventually reduce local weevil populations to the point where breu tree colonization and resin formation decline.

Resin harvesting, though, may actually help maintain the health of breu trees and prolong their resin producing lives. Individual weevil attacks do not kill their host trees, but the accumulation of weevil larvae and their bore holes may increase the exposure of the trees to bacterial and fungal pathogens. Over time weevil attacks could hasten the demise of the tree. While this study shows that resin harvesting does not stop recolonization, it may slow down the deterioration of the tree associated with weevil attacks.

Some important evolutionary and ecological questions remain to be answered in future studies. How widespread is the association between the Sternocoelus or other weevils and Burseraceae trees in other parts of the Amazon? Other researchers have noted the relationship between unidentified weevils and resin producing

Burseraceae trees in Amazonas states of both Brazil and Columbia (Zoghbi, personal communication 1998; 119

Martinez-Habibe, 1998). Further inquiries about weevil life history and the specificity of its relationships with host trees throughout the region could make an important contribution to the evolution of weevil-host plant associations (Anderson, 1993). There is nothing known yet concerning the role that any fungi or other microbe might play in stimulating resin flow or weevil adaptation to its resin-rich environment. It is clear that the weevils’ provocation of resin flow provides a material that is important to many forest dwellers in their every day lives and in commerce. It would be also be valuable to explore the extent to which these resins are critical materials for other invertebrates such as syrphid flies and stingless bees that provide pollination services to the forest. The breu trees might not agree, but from a human perspective this new species of Sternocoelus, the

“resin” weevil, might arguably be considered a new sort of beneficial insect.

Given the numerous factors that restrict breu resin harvest potential within close proximity of the major villages near the Gurupi River, it is not surprising that most people who collect this resin for sale go upriver to reach larger areas of intact forest. This opens up a much larger area of forest to exploitation, but there are still limits to this. Collectors generally do not travel farther than several hours into the forest away from the river and a large portion of the reserve has been occupied and/or deforested by non-Indian colonists, ranchers and loggers (Sales, 1993, 1994). In theory, overharvesting has the potential to reduce long-term harvesting potential by reducing the population of weevils that stimulate resin flow. One procedure that could avoid this possibility would be to restrict resin harvest to dried lumps that no longer contain living larvae. The non-systematic manner in which harvesters search the forest, however, probably provides a strong buffer against severe impacts on the weevils. Harvesters have a strong tendency to seek out medium and large trees that are most likely to yield large amounts of resin. This practice undoubtedly leaves many smaller trees with some weevils on them as a de facto beetle population reserve.

While depletion of the weevil population may not be an immediate concern, the increasing demand for cash income by the Tembé could reduce the amount of breu resin available in the most accessible areas of forest. Even with healthy beetle populations, this study predicts that it will take at least four years for resin levels to build up to initial harvest levels. The current relatively low market price and demand for resin may provide a sufficient barrier to overharvesting. Success in developing the resin as a new source of fragrances for the perfume industry could expand the market and economic benefits for resin collectors; it could also push 120 local harvesting into the non-sustainable realm. If harvests increase, it would probably be advisable for communities to institute a more proactive form of managing resin harvests. One scheme would be to make areas off-limits to harvesting for a certain number of years after an intensive harvest. Another management tool that could be explored is transplanting small resin lumps from trees with multiple lumps to suitable uncolonized trees. This technique of “seeding” a tree with an insect that produces a commercial product has worked well with lac insects responsible for producing natural shellac (Sharma et al., 1999).

In conclusion this study has demonstrated that the harvest potential of this NTFP that is considered to be a plant defensive compound is directly influenced by the host tree’s relationship to a herbivorous weevil.

There are no doubt many other useful resins and defense compounds produced by plants in response to attacks by insects or microbial pathogens. Investigating these relationships could address many outstanding coevolutionary questions and help develop strategies to sustainably manage such resources.

121

TABLE 3.1 SUMMARY OF RESEARCH OBJECTIVES AND METHODS

OBJECTIVE SURVEY/METHOD MEASUREMENTS ANALYSIS RESULTS Density of Breu Trees Belt transect inventories: 7.3 Tree Type (branco or sarara Trees/ha by Tree type, Tree Table 3.2 & Size Class Structure ha inventoried in 10x500 m only); Tree Size (DBH ≥10 DBH, Forest Type plots in 1/98 and 11/99 in cm); Forest Type (terra firme forest 1 km from Tekohaw or baixo) Amount of Resin on Initial Harvest Survey: Resin Tree Type; Tree DBH; Tree Comparison of Resin Table 3.3; Breu Trees during was harvested from all breu Status (Dead or Alive); Forest Amount/tree to Tree Type, Fig. 3.1 Initial Harvest trees in two month (6-7/98) Type; GPS reading; Number Tree DBH, Tree Status; Forest survey of forest <4 km from resin lumps; weight of resin Type, and Distance from Tekohaw lumps; Tekohaw Sternocoelus Weevils Initial Harvest Survey Length, width, weight of Dimensions of weevils and Table 3.4, and their Role in Breu weevil larvae, pupae & adults head width distribution pattern Fig. 3.2 Resin Yield Number and size (width) of Correlation of resin lump Table 3.5 weevils in resin lump; number weight to weevil number and and size (width, depth) of dimensions and to bore hole weevil bore holes behind resin number and dimensions lump; weight of resin lump Number and size (width) of Correlation of weevil number Table 3.6 weevils in resin lump; number and dimensions to bore hole and size (width, depth) of number and dimensions bore holes behind resin lump Resin Yield by Breu Tree Drilling Tree DBH; Presence or Percentage of holes with resin; Results in Drilling Breu Trees Experiment: 18 trees drilled absence of resin in drill holes comparison of Tree DBH to text with bore hole dimensions one month after drilling percentage with resin similar to weevils (11/99) Rates of Resin Follow-up Harvests: subsets Number and weight of resin Comparison of number of Table 3.7 Accumulation on Breu of trees were reharvested in lumps (all trees) resin lumps and resin weight Trees 11/98, 3/99, and 7/99 to time since previous harvest Resin Lump Growth Study: In Area of individual resin lumps Resin lump area to weight Fig. 3.4 3/99, new resin lumps were on tagged trees measured in ratio measured from harvested tagged on 16 breu trees last 3/99 & 7/99. Area and weight lumps. Daily resin growth harvested in 6-7/98. of resin lumps from 5 trees in rate (g/day) estimated for Follow-up Harvests tagged resin lumps Follow-up Harvests & Resin Number of resin lumps and Correlation between resin Table 3.8 Lump Growth Study weight per tree; Tree DBH; lump number & weight at first Time since previous harvest & second harvest and DBH Regression Model Estimates Table 3.9 for Resin Accumulation Resin Weight Loss Resin Drying Study: 16 Resin batches weighed daily Percentage resin weight loss Fig. 3.5 during Outside Drying batches observed (7/98) until stabilized during outside drying Resin Weight Loss Resin Drying in Open and Resin sample weights at end Resin weight loss with and Fig. 3.6 due to Stingless Bees Closed Enclosures (4/99) of each day without bee removal Percentage of Breu Initial Harvest Survey: In 6/98 Tree Size (DBH); Presence or Probability of resin by Tree Table 3.10 Trees with Resin all breu trees found in forest absence of resin on trunk Size were checked for resin Amount and Value of Interviews with Resin Buyers Amount of resin bought and Average price per kg of dried Results in Resin Sales price paid resin text Resin Amount and Initial Harvest Survey, Resin Same as previously described Amount and value of resin Table 3.11 Value per Hectare Drying Study, Interviews harvest per hectare of forest Resin Amount and Initial Harvest Survey, Resin Same as previously described Amount and value of resin Table 3.12 Value per Time Drying Study, Interviews harvest per day and renewable Harvested harvesting cycle 122

TABLE 3.2 DENSITY OF BREU TREES IN DIFFERENT FOREST TYPES AT TEKOHAW

Area Branco Tree Sarara Tree Total Tree Surveyed Number and Number and Number and FOREST (ha) Density Density Density TYPE No. (%) No. (%) Tr./ha No. (%) Tr./ha No. (%) Tr./ha Terra Firme 6.25 33 5.3 24 3.8 57 9.1 (86%) (94%) (69%) (81%) Baixo 1.05 2 1.9 11 10.5 13 12.4 (14%) (6%) (31%) (19%) Total 7.30 35 4.8 35 4.8 70 9.6 Combined Results of Two Inventories of Resin Yielding Breu Tree Types (≥ 10 cm DBH) in Tekohaw Survey Blocks

123

TABLE 3.3 BREU RESIN HARVEST PER TREE BY STATUS, TREE TYPE, LIVE TREE SIZE (DBH), FOREST TYPE, AND DISTANCE FROM TEKOHAW

CATEGORY Number of Mean Tree Resin Harvest (g) Resin Harvest (g) Trees DBH(cm) Mean ± S.E. Median

OVERALL 183 28.4 856 ± 100 420 STATUS Live 171 28.3 789 ± 85a 380 Dead 12 29.2 1808 ± 923b 930 TREE TYPE Branco 72 24.5 646 ± 135 a 325 a Sarara 111 30.8 992 ± 139 a 508 a LIVE TREE DBH 5 - 9 cm 2 7.2 38 ± 28 a 38 a 10-19 cm 33 16.3 437 ± 133 a 200 a 20-29 cm 71 25.1 766 ± 145 a 340 a 30-39 cm 46 34.9 1009 ± 176 a 480 a 40-49 cm 14 44.5 905 ± 228 a 640 a 50-69 cm 5 54.9 1394 ± 507 a 1120 a FOREST TYPE Terra Firme 110 25.6 650 ± 98 a 315 a Baixo 71 32.8 1167 ± 204b 540b DISTANCE 1.0-1.9 km 53 25.8 669 ± 109 a 320 a 2.0-2.9 km 29 27.0 756 ± 127 a 540 a 3.0-4.2 km 99 30.2 992 ± 171 a 422 a

All categories include live and dead trees with resin except DBH size classes. Sample sizes in some categories are different due to missing data from some trees in that category. Distance is the distance between the tree and Tekohaw village. Different letters in mean column indicate that pairwise p-values in AOV comparisons of means were ≥ 0.05. Different letters in Median column indicate that p-values of pairwise comparisons were ≤ 0.05 with Mann-Whitney Test and Mood’s Median Test had a p-value ≤ 0.05 in comparing all medians for that category. 124

TABLE 3.4 DIMENSIONS OF WEEVILS FOUND IN BREU RESIN LUMPS

N Mean ± S.E. Larvae Head Width (mm) 160 1.84±.04 Body Width (mm) 158 3.58±.11 Body Length (mm) 158 8.15±.27 Pupae1 Head Width (mm) 2 1.80±.25 Body Width (mm) 2 4.85±.69 Body Length (mm) 2 9.55±2.35 Body Weight (g) 2 0.12±.06 New Adult2 Pronotum Len. (mm) 3 3.67±.19 Pronotum Wid. (mm) 3 4.93±.12 Elytra Length (mm) 3 8.47±.38 Elytra Width (mm) 3 6.68±.22 Body Length (mm) 3 11.82±.23 Body Weight (g) 3 0.22±.01 Measures done on specimens stored in alcohol. Linear dimensions done with calipers examining specimens under dissecting microscope and measured to nearest 0.01 mm. Weight measurements done with electronic scale to nearest 0.0001 g. 1. Includes one pupa and one animal in transition between pupa and adult condition. 2. Includes two males and one female. Female was larger in all dimensions.

125

TABLE 3.5 BREU RESIN LUMPS WEIGHT CORRELATION TO WEEVIL LARVAE AND BORE HOLE DIMENSIONS

Correlations with Resin Lump Weight WEEVIL LARVAE No. Corr. (per resin lump with larvae) Lumps Coef.a R2 (% )b Number of Larvae 137 .456** 20.8** Head Width Sum (mm) 136 .352** 12.4** Head Width Maximum (mm) 136 .261** 6.8** Body Width Sum (mm) 129 .456** 20.8** Body Width Maximum (mm) 129 .346** 12.0** BORE HOLES (per lump with bore holes) Number of Bore Holes 245 .540** 29.2** Diameter Maximum (mm) 244 .355** 12.6** Diameter Sum (mm) 244 .488** 23.8** Depth Sum (mm) 33 .632** 40.0** Volume Sum (mm3) 33 .671** 45.0** a - Pearson’s Correlation Coefficient ρ (rho) b - Regression Analysis R2 Value expressed as percentage **- Correlation and Regression Values significant at p ≤ 0.01 126

TABLE 3.6 WEEVIL LARVAE DIMENSION CORRELATION TO BORE HOLE NUMBER AND DIMENSIONS

Bore-Hole Number Bore-Hole Dia. Sum Bore-Hole Dia. Max. WEEVIL LARVAE Corr.Cf.a R2 (%)b Corr.Cf. a R2 (%)b Corr.Cf. a R2 (%)b Number of Larvae .311** 9.7** .336** 11.3** .133 1.8 n=129 resin lumps Sum of Head Widths .444** 19.7** .512** 26.2** .330** 10.9** (mm) n=127 resin lumps Maximum Head Width .425** 18.1** .500** 25.0** .544** 29.6** (mm) n=127 resin lumps Sum of Body Widths .583** 33.9** .634** 40.2** .401** 16.0** (mm) n=120 resin lumps Maximum Body Width .535** 28.6** .616** 37.9** .589** 34.7** (mm) n=120 resin lumps a - Pearson’s Correlation Coefficient b - Regression Analysis R2 Value expressed as percentage. **- Correlation and Regression Values significant at p ≤ 0.01. 127

TABLE 3.7 RESIN LUMP NUMBER AND WEIGHT FOUND ON BREU TREES DURING FOLLOW-UP HARVESTS

Second Harvest Month Nov. 1998 Mar. 1999 July 1999 Time Since Last Harvest 139 - 147 Days 241 - 272 Days 364 - 388 Days N trees 11 26 12 Ave. Resin/tree (g) 134.0 ± 50.6 164.9 ± 49.5 57.5 ± 32.4 Ave. Lump/tree 3.3 ± 0.9 5.2 ± 1.9 2.3 ± 0.6 Ave. Resin/lump (g) 40.9 ± 6.9 31.8 ± 2.7 25.3 ± 6.9

128

TABLE 3.8 CORRELATION OF TREE SIZE AND HARVEST HISTORY WITH FOLLOW-UP BREU RESIN HARVEST

Tree DBH (cm) Resin (g)/Tree in Resin Lumps/Tree in First Harvest First Harvest Corr. Cf.a R2 (%)b Corr. Cf.a R2 (%)b Corr. Cf.a R2 (%)b Resin (g)/Tree in .407** 16.6** .548** 30.0** .716** 51.3** Second Harvest Resin Lumps/Tree .440** 19.4** .476** 22.7** .802** 64.3** in Second Harvest a - Pearson’s Correlation Coefficient; b - Regression Analysis R2 Value expressed as percentage **Correlation and Regression Values significant at p ≤ 0.01 Analysis includes 49 trees with resin and 17 trees without resin during follow-up harvest.

129

TABLE 3.9 PROJECTION OF BREU RESIN YIELD AFTER INITIAL HARVEST

First Time of Second Harvest Harvest and Estimated Resin Yield (g) per Tree (Mean ± S.E.) Resin (g) Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 per Tree1 All Trees 562±65 153±12 278±23 403±35 528±46 653±58 777±69 902±81 Trees with Resin 789±85 215±14 390±28 566±42 741±56 916±71 1091±85 1266±99 Regression Model: Resin Wt/tree in Second Harvest = 28.4 + 0.000381*WtH1*Days + 0.000017*DBH2 *Days + 0.0443*Lumps*Days R2=32.6; p=0.000 WtH1=Resin weight collected from tree during initial harvest; Lumps=Number of resin lumps collected from tree during initial harvest; DBH2 = square of DBH (cm) at 1.5 m.; Days=Number of days after initial harvest Based on Regression Model from Second Round Resin Harvests and Resin Lump Growth Study (n=82 trees) applied to All Live Trees (n=240) and Live Trees with Resin (n=171) Monitored in Initial Harvest Note: 1) Mean ± S.E. 130

TABLE 3.10 BREU RESIN HARVEST MODEL - PART 1. ESTIMATED AVERAGE RESIN PER TREE BY SIZE CLASS

AVE. RESIN PER EST. RESIN PER BREU TREE PROBABILITY OF TREE FOR TREES TREE FOR ALL SIZE CLASS RESIN WITH RESIN (Kg) TREES (Kg) 10-19 cm DBH 0.20 0.44 0.09 20-29 cm DBH 0.62 0.77 0.47 30-39 cm DBH 0.80 1.01 0.81 >40 cm DBH 1.00 1.03 1.03

131

TABLE 3.11 BREU RESIN HARVEST MODEL - PART 2. DENSITY OF RESIN YIELDING TREES, RESIN HARVEST AND HARVEST PER VALUE PER HECTARE

ESTIMATED BREU TREES PER SIZE CLASS AT DIFFERENT DENSITIES (Trees/ha) BREU TREE SIZE CLASS 1 Tree/ha 5 Trees/ha 10 Trees/ha 25 Trees/ha 50 Trees/ha SIZE CLASS PERCENT 10-19 cm DBH 0.55 .55 2.75 5.5 13.75 27.50 20-29 cm DBH 0.26 .26 1.30 2.6 6.50 13.00 30-39 cm DBH 0.14 .14 0.70 1.4 3.50 7.00 >40 cm DBH 0.05 .05 0.25 0.5 1.25 2.50 AVE. RESIN ESTIMATED BREU RESIN AMOUNT AND VALUE PER HA (Kg/tree) AT DIFFERENT DENSITIES (Kg Resin/ha) 10-19 cm DBH 0.09 0.05 0.24 0.48 1.20 2.40 20-29 cm DBH 0.47 0.12 0.59 1.18 2.94 5.88 30-39 cm DBH 0.81 0.11 0.57 1.13 2.83 5.65 >40 cm DBH 1.03 0.05 0.26 0.52 1.29 2.59 AREA FRESH 0.34 1.65 3.30 8.26 16.51 RESIN (Kg /ha) AREA DRIED 0.28 1.37 2.74 6.85 13.71 RESIN (Kg/ha) AREA VALUE 0.14 0.69 1.37 3.43 6.85 ($R/ha) Equations for Variables 1) Kg Resin per Size Class at Given Density = Resin per Tree at Size Class (Table 3.9) x Number of Trees in Size Class at Given Density = Kg Resin per Size Class at given Density 2) Area Fresh Resin = Total Kg Fresh Resin at Given Density for All Size Classes Combined = Kg Fresh Resin/ha 3) Area Dried Resin = Kg Fresh Resin/ha x 83% = Kg Dried Resin/ha 4) Area Value (for harvester) = Kg Dried Resin/ha x $R 0.50/Kg = $R/ha 132

TABLE 3.12 DENSITY OF BURSERACEAE RESIN YIELDING SPECIES IN BRAZILIAN AMAZON INVENTORIES

Ha NUMBER OF SPECIES DENSITY(trees/ha) Ref REGION SURVEYED All Species Resin Species1 All Species Resin Species1 EAST (Maranhão, 4 20 12 80.5 57.8 1 Pará, and Amapa 1 10 7 72.0 52.0 1.1 1 7 5 68.0 46.0 2 4 17 10 41.5 30.8 3 1 6 3 56.0 30.0 4 42 9 7 12.3 10.8 5 1 2 1 11.0 10.0 6 1 6 2 35.0 5.0 7 1 4 1 7.0 1.0 8 6 7 3 64.3 0.0 9 (3) 4 1 10 (3) 3 2 11 CENTRAL (Manaus, AM) 70 48 21 12 WEST (Rondônia) 1 6 4 70.0 9.0 13 1 1 1 5.0 5.0 14 0.52 1 1 4.0 4.0 15 2 5 1 25.0 1.5 16 6 8 2 20.8 0.7 17 1 1 0 60.0 0.0 18 Site References: Balée, 1994 (1,5); Balée, 1986(1.1); da Silva and Rosa, 1989(2, 6); Almeida et al., 1993(3); Dantas et al., 1980(4); da Silva et al., 1987(7); Salomão et al., 1988(8); Salomão, 1991(9); Lopes, 1993 (10); Mori et al., 1989(11); Rankin-de-Mérona et al., 1992(12); Salomão and Lisboa, 1988(13); Maciel and Lisboa, 1989(14); Lisboa, 1989 (15,18); Absy et al., 1987(16,17). Other Notes: 1. Includes species known to be a source of harvestable resin or close relatives of these (Daly, 1987; Harley and Daly, 1995): Protium altsonii, P. carnosum, P. crassipetalum, P. decandrum, P. giganteum, P. guianense, P. heptaphyllum, P. morii, P. opacum, P. pallidum, P. paniculatum, P. pilosum, P. polybotryum, P. robustum, P. rubrum, P. sagotianum, P. tenuifolium, P. trifoliolatum, Tetragastris panamensis, Crepidospermum goudotianum. 2. Secondary forest site; others are assumed to be old-growth forest 3. Non-plot survey 133

TABLE 3.13 BREU RESIN HARVEST - PART 3. RESIN HARVESTING RATES AND VALUE BY AREA AND TIME

DENSITY (Trees/ha) 1 5 10 25 50 SEARCH AND HARVEST (Min./Tree) 54 14 9 6 5 TREE HARVEST RATE (Trees/Day) 5.6 21.4 33.3 50.0 60.0 AREA HARVEST RATE (Ha/Day) 5.6 4.3 3.3 2.0 1.2 RESIN HARVEST RATE (Kg/Day) 1.9 7.2 11.2 16.8 20.2 DRIED RESIN YIELD (Kg/Day) 1.6 6.0 9.3 14.0 16.8 VALUE OF EFFORT ($/Day) 0.78 3.00 4.65 7.00 8.37 AMOUNT, VALUE, TIME PER HARVEST IN 1000 ha Fresh Resin (Kg) 336 1681 3363 8406 16813 Dried Resin (Kg) 279 1395 2791 6977 13955 Collecting Days (5 hours per day) 180 233 300 500 833 Value($R) 140 698 1395 3489 6977 AREA NEEDED FOR ONE PERSON DAILY HARVEST (ha) 4 Year Harvest Cycle 8111 6257 4867 2920 1752 5 Year Harvest Cycle 10139 7821 6083 3650 2190 Equations for Variables Area Search Rate = 1 min./ 200 m2 x 10,000 m2/ha = 50 Min./ha Tree Search Rate = 50 Min./ha x 1 ha/Number of Trees = y Min./Tree Trees Harvest Time = 4 min./tree Tree Harvest Rate = 5 hrs. (300 min.)/y min./tree = Trees Harvested/Day Area Harvest Rate = Trees Harvested/Day x 1 ha/Number of Trees = Area Harvested/Day Resin Harvest Rate = Resin Harvested/ha (Table 3.10) x Area Harvested/Day = Kg Resin Harvested/Day Dried Resin Yield = Kg Resin Harvested/Day x 83% = Kg Dry Resin Harvested/Day Value of Effort = Kg Dry Resin Harvested/Day x $R 0.50/Kg = $ Earned/Day ($R ≈ $US in 1998) Area Needed for Continuous Daily Harvest = Area Harvest Rate x No. Years between Harvests from Same Trees

134

FITTED LINE REGRESSION PLOT R-Sq = 1.3%; p=0.124; 10000 N=181 Trees with Resin

3-4 Km. 2-3 Km. 992 g./tree 5000 1-2 Km. 756 g./tree 669 g./tree

0

1000 2000 3000 4000 RESIN HARVEST PERRESIN HARVEST TREE (g) DISTANCE FROM TEKOHAW VILLAGE (m)

FIGURE 3.1 RELATIONSHIP OF DISTANCE FROM VILLAGE TO BREU RESIN HARVEST 135

14

12

10

8

6

4

2

0 PERCENT IN EACHWIDTH CLASS 1 2 3 HEAD CAPSULE WIDTH (mm.) (n = 160 larvae from Tekohaw)

FIGURE 3.2 DISTRIBUTION OF WEEVIL LARVAE HEAD CAPSULE WIDTHS FROM BREU RESIN LUMPS 136

a. Sternocoelus weevil larva feeding on breu tree in bore hole adjacent to chamber in resin lump b. Alipumilio fly larvae in resin lump c. Trigona stingless bee collecting fresh resin d. Reduvid assassin bug stalking bees and other insects attracted to fresh resin

FIGURE 3.3 ILLUSTRATION OF BREU RESIN LUMP AND ASSOCIATED INSECTS 137

Growth rates based on estimated weight differences in resin lumps measured at 9 and 12 months after initial resin harvest. N=53 Resin Lumps on 16 Trees. 0.9 Fitted Line Regression Plot 0.8 (Quadratic Model) 0.7 R-Sq = 59.7 %

0.6

0.5

0.4

0.3

0.2

Estimated ResinEstimated Lump Growth Rate(g/day) 0.1 0.0

0 100 200 300 Estimated Resin Lump Weight (g) at 12 Months after Initial Resin Harvest

FIGURE 3.4 ESTIMATED GROWTH RATE OF INDIVIDUAL BREU RESIN LUMPS AT TEKOHAW 138

Interval Plot: Mean plus 95% Confidence Interval

18

13

8 CUMULATIVE WEIGHT LOSS (%) LOSS WEIGHT CUMULATIVE 12345 DAYS EXPOSED OUTSIDE (N = 16 Batches of Resin)

FIGURE 3.5 BREU RESIN WEIGHT LOSS IN OPEN OUTSIDE DRYING AT TEKOHAW 139

Interval Plot: Mean plus 95% Confidence Interval

14 Closed to Bees Open to Bees

9

4 CUMULATIVE WEIGHT LOSS (%) Day 1 Day 3 Day 4 DAYS IN SCREENED ENCLOSURES N= 4 Samples Breu "Sarara" per Treatment

FIGURE 3.6 BREU RESIN WEIGHT LOSS WITH AND WITHOUT STINGLESS BEES 140

DBH SIZE Mean Contribution of Resin per CLASS ( cm ) 15 Hectare by each DBH Size Class - > 39 Based on Resin Harvest and Size 30-39 Class Population Data at Tekohaw 10 20-29 10-19

5 BREU RESIN PERBREU RESIN (kg)HECTARE

0

1 5 10 25 50 BREU TREES PER HECTARE

FIGURE 3.7 ESTIMATED FIRST TIME RESIN HARVEST AT DIFFERENT DENSITIES OF BREU TREES 141

Based on Resin Harvest Amounts and Collection Rates at Tekohaw

KG DRIED BREU RESIN 15.00 HARVESTED PER DAY

12.50

10.00

7.50 $R EARNED PER DAY 5.00 HARVESTING RESIN 2.50

0 DRIED RESIN (Kg) /// REVENUE ($R) /// REVENUE (Kg) RESIN DRIED 0 10 20 30 40 50 BREU TREES PER HECTARE

FIGURE 3.8 ESTIMATED DAILY HARVEST AND SALE VALUE OF DRIED RESIN AT DIFFERENT DENSITIES OF BREU TREES 142

REFERENCES

Absy, Maria Lúcia, Ghillean T. Prance, and Edelcílio Marques Barbosa. 1987. Inventário florístico de floresta natural na área da estrada Cuiabá-Porto Velho (BR-364). Acta Amazonica 16/17:Sup. 85-121 (1986/87).

Adis, Joachim. 1984. Eco-entomological observations from the Amazon. V. Feeding habits of neotropical “bee killers” and “resin bugs” (Apiomerinae: Reduviidae: Hemiptera). Revista Biologica Tropical 32(1):151-153.

Alcorn, Janis B. 1984(1). Huastec Mayan Ethnobotany. University of Texas Press, Austin.

Alcorn, Janis B. 1984(2). Development policy, forests and peasant farms: reflections on Huastec-managed forests’ contributions to commercial production and resource conservation. Economic Botany 38(4):389-406.

Almeida, Samuel S., Pedro L.B. Lisboa, and Antônio Sérgio L. Silva. 1993. Diversidade florística de uma comunidade arbórea na estação científica “Ferreira Penna,” em Caxiuanã (Pará). Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 9(1):93-128.

Anderson, Robert S. 1993. Weevils and plants: phylogenetic versus ecological mediation of evolution of host plant associations in Curculioninae (Coleoptera: Curculionidae). Memoirs of the Entomological Society of Canada 165:197-232.

Anderson, Robert S. 1995. An evolutionary perspective on diversity in Curculionidea. Memoir of the Entomological Society of Washington 14:103-114.

Baier, P. 1996. Defense reactions of Norway spruce (Picea abies Karst.) to controlled attacks of Ips typographus (L.)(Col., Scolytidae) in relation to tree parameters. Journal of Applied Entomology 120:587-593.

Balée, William and Douglas C. Daly. 1990. Resin classification by the Ka’apor Indians. Advances in Economic Botany 8:24-34.

Balée, William. 1986. Análise preliminar de inventário florestal e a etnobotânica Ka’apor (Maranhão). Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 2(2):141-167.

Balée, William. 1987. A etnobotânica quantitativa dos índios Tembé (Rio Gurupi, Pará). Boletim do Museu Paraense Emílio Goeldi: Botânica 1:29-50.

Balée, William L. 1994. Footprints of the Forest: Ka’apor Ethnobotany - the Historical Ecology of Plant Utilization by an Amazonian People. Columbia University Press, New York.

Bastos, José A. M., Raimundo W. de Figueiredo, and Flávio de Wimar The. 1980. Broca das estacas de maniçoba, Sternocoelus manihoti (MSHL) (Col.: Curculionidae) em casa de vegetação do centro de ciências agrárias, Fortaleza, Ceará, Brasil. Fitossanidade, Fortaleza 4(1):9-11. 143

Blanche, C.A., P.L. Lorio Jr., R.A. Sommers, J.D. Hodges, and T.E. Nebeker. 1992. Seasonal cambial growth and development of loblolly pine: xylem formation, inner bark chemistry, resin ducts, and resin flow. Forest Ecology and Management 49:151-165.

Boom, Brian M. 1996. Ethnobotany of the Chácabo Indians, Beni, Bolivia. 2nd Ed. The New York Botanical Garden, New York.

Borden, J.H. 1982. Aggregation pheromones. Pp. 74-138 in Jeffry B. Mitton and Kareen B. Sturgeon (eds.) Bark Beetles in North American Conifers, University of Texas Press, Austin.

CEDI. 1985. Tembé. pp. 177-209 in Povos Indígenas no Brasil, No. 8: Sudeste do Pará. Centro Ecumênico de Documentação e Informação, São Paulo.

Christiansen, E., P. Krokene, A.A. Berryman, V.R. Franceschi, T. Krekling, F. Lieutier, A. Lönneborg and H. Solheim. 1999. Mechanical injury and fungal infection induce acquired resistance in Norway spruce. Tree Physiology 19:399-403.

Comerford, Simon C. 1996. Medicinal plants of two Mayan healers from San Andrés, Petén, Guatemala. Economic Botany 50(3):327-336.

Correa, Jaime Enrique Q. and Henry Yesid Bernal. 1989. Especies vegetales promisorias de las paises del Convenio Andrés Bello. Tomo III. Secretaria Ejecutiva del Convenio Andrés Bello (SECAB), Bogotá.

Cortés, Santiago. (Year not specified) Flora de Columbia. Libreria de el Mensajero, Bogota.

Coulson, R.N., A.M. Mayyasi, J.L. Folz, F.P. Hain, and W.C. Martin. 1976. Resource utilization by the southern pine beetle, Dendroctonus frontalis (Coleoptera: Scolytidae). Canadian Entomologist 108:353-362.

Coulson, Robert N. and John A. Witter. 1984. Forest Entomology: Ecology and Management. John Wiley and Sons, New York.

Creed, Robert P. and Sallie P. Sheldon. 1995. Weevils and watermilfoil: did a North American herbivore cause the decline of an exotic plant? Ecological Applications 54(4):1113-1121. da Costa Lima, A. 1956. Insetos do Brasil. Vol. 10, Chapter 29, Coleópteros. Escola Naçional de Agronomia, Série Didática No. 12. da Silva, Manoela F.F., Nelson A. Rosa, and Jorge Oliveira. 1987. Estudos botânicos na área do Projeto Ferro Carajás. 5. Aspectos florísticos da mata do Rio Gelado, Pará. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 3(1):1-20. da Silva, Manoela F.F. and Nelson A. Rosa. 1989. Análise do estrato arbóreo da vegetação sobre jazidas de cobre na Serra dos Carajás, PA. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 5(2):175- 205.

Daly, Douglas Charles de Burgh. 1987. A Taxonomic Revision of Protium (Burseraceae) in Eastern Amazonia and the Guianas. PhD dissertation from The City University of New York.

Daly, Douglas C. 1989. Studies in neotropical Burseraceae. II. Generic limits in new world Protieae and Canarieae. Brittonia 41(1):17-27. 144

Daly, Douglas C. 1992. New taxa and combinations in Protium Burm. f. Studies in neotropical Burseraceae VI. Brittonia 44(3):280-299.

Dantas, Mário, Irenice Alves Rodrigues, and Nina R. M. Müller. 1980. Estudos fito-ecológicos do tropico úmido Brasileiro: Aspectos fitossociológicos de mata sobre Latossolo Amarelo em Capitão Poço, Pará. EMBRAPA, Belém. Boletim de Pesquisa No. 9.

Dunn, James P. and Peter L. Lorio, Jr. 1992. Effects of bark girdling on carbohydrate supply and resistance of loblolly pine to southern pine beetle (Dendroctonus frontalis Zimm.) attack. Forest Ecology and Management 50:317-330.

Evans, Gly. 1975. The Life of Beetles. George Allen and Publishers, London.

Fargo, W.S., R.N. Coulson, P.E. Pulley, D.N. Pope, and C.L. Kelley. 1978. Spatial and temporal patterns of within-tree colonization by Dendroctonus frontalis (Coleoptera: Scolytidae). Canadian Entomologist 110:1213-1232.

Feeney, Shelly R., Thomas E. Kolb, W. Wallace Covington, and Michael R. Wagner. 1998. Influence of thinning and burning restoration treatments on presettlement ponderosa pine at the Gus Pearson Natural Area. Canadian Journal of Forest Research 28:1295-1306.

Ferrar, P. 1987. A Guide to the Breeding Habits and Immature Stages of Diptera Cyclorrhapha, Part 1. Entomonograph, Volume 8. E.J. Brill/Scandinavian Science Press, Leiden.

Franceschi, Vincent R., Paal Krokene, Trygve Krekling and Erik Christiansen. 2000. Phloem parenchyma cells are involved in local and distant defense responses to fungal inoculation or bark-beetle attack in Norway spruce (Pinaceae). American Journal of Botany 87(3):314-326.

Gentry, Alwyn H. 1993. A Field Guide to the Families and Genera of Woody Plants of Northwest South America. The University of Chicago Press, Chicago.

Gompper, Matthew E. and Anne M. Hoylman. 1993. Grooming with Trattinnickia resin: possible pharmaceutical plant use by coatis in Panama. Journal of Tropical Ecology 9:533-540.

Grimes, A., S. Loomis, P. Jahnige, M. Burnham, K. Onthank, R. Alarcón, W. P. Cuenca, C. C. Martinez, D. Neill, M. Balick, B. Bennett, and R. Mendelsohn. 1994. Valuing the rain forest: the economic value of nontimber forest products in Ecuador. Ambio 23(7):405-410.

Guariguata, Manuel E. and Gregory S. Gilbert. 1996. Interspecific variation in rates of trunk wound closure in a Panamanian lowland forest. Biotropica 28(1):23-29.

Harley, M.M. and D.C. Daly. 1995. Burseraceae Kunth. Protieae March. em. Engl. World Pollen and Spore Flora 20:1-44.

Hellrigl, Von K. 1992. Die fichtenharzfliege Cheilosia morio Zett. (Dipt. Syrphidae) als physiologischer schädling an fichten in Südtirol. Anz. Schädlingskde., Pflazenschutz, Umweltschutz 65: 33-36.

Hogue, Charles L. 1993. Latin American Insects and Entomology. Univ. of California Press, Berkeley.

Howes, F.N. 1949. Vegetable Gums and Resins. Chronica Botanica Company, Waltham, MA.

Hui, Ye and Zhao Zhimo. 1995. Life table of Tomicus piniperda (L.)(Col., Scolytidae) and its analysis. Journal of Applied Entomology 119:145-148. 145

Johnston, Mark and Araminta Colquhoun. 1996. Preliminary ethnobotanical survey of kurupukari:an Amerindian settlement of central Guyana. Economic Botany 50(2):182-194.

Kainer, Karen A. and Mary L. Duryea. 1992. Tapping women’s knowledge: plant resource use in extractive reserves, Acre, Brazil. 1992. Economic Botany 46(4):408-425.

Kalpagé, F.S.C.P. 1974. Tropical Soils: Classification, Fertility and Management. St. Martin’s Press, New York.

Khalid, Sami A. 1983. Chemistry of the Burseraceae. pp.281-299 in eds. P.G. Waterman and M.F. Grundon. Chemistry and Chemotaxonomy of the Rutales. Academic Press, New York.

Lamb, K.P. 1974. Economic Entomology in the Tropics. Academic Press, New York.

Langenheim, Jean H. 1990. Plant resins. American Scientist 78:16-24.

Langenheim, Jean H. In press. Plant Resins: their Value to Plants and Humans. Timber Press, Portland.

Levi-Strauss, Claude. 1952. The use of wild plants in tropical South America. Economic Botany 52:252-270.

Lieutier, F., G. Vouland, M. Pettinetti, J. Garcia, P. Romary and A. Yart. 1992. Defense reactions of Norway spruce (Picea abies Karst.) to artificial insertion of Dendroctonus micans Kug.(Col. Scolytidae). 1992. Journal of Applied Entomology 114:174-186.

Lieutier, François, Jacques Garcia, Paul Romary, Annie Yart, Hervé Jactel, and Daniel. 1993. Inter-tree variability in the induced defense reaction of Scots pine to single inoculations by Ophiostoma brunneo-ciliatum, a bark-beetle-associated fungus. Forest Ecology and Management 59:257-270.

Lieutier, F., J. Garcia, A. Yart and P. Romary. 1995. Wound reactions of Scots pine (Pinus sylvestris L.) to attacks by Tomicus piniperda L. and Ips sexdentatus Boern. (Col., Scolytidae). Journal of Applied Entomology 119:591-600.

Lima, E.D., O.F. Gompertz, M.D. Paulo, and A.M. Giesbrecht. 1992. Invitro antifungal activity of essential oils against clinical isolates of dermatophytes. Revista de Microbiologia 23(4):235-238.

Lisboa, Pedro L.B. 1989. Estudo florístico da vegetação arbórea de uma floresta secundária, em Rondônia. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 5(2):145-162.

Lopes, José do Carmo Alves. 1993. Demografia e flutuações temporais da regeneração natural após uma exploração florestal: flona do Tapajós - PA. Master’s Thesis Univ. of São Paulo, Piracicaba, SP.

Lorio, Peter L. Jr., Frederick M. Stephen and Timothy D. Paine. 1995. Environment and ontogeny modify loblolly pine response to induced acute water deficits and bark beetle attack. Forest Ecology and Management 73:97-110.

Maciel, Ubirajara N. and Pedro L.B. Lisboa. 1989. Estudo florístico de 1 hectare de mata de terra firme no km 15 da Rodovia Presidente Médeci-Costa Marques (RO-429), Rondônia. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 5(1):25-37.

Mantell, C.L. 1950. The natural hard resins-their botany, sources and utilization. Economic Botany 4:203-242. 146

Martinez-Habibe, Maria Cristina. 1998. Estudio Preliminar de la Familia Burseraceae en Columbia con Enfasis en la Etnobotanica Ticuna de San Martin de Amacayacu (Amazonas). Senior Biology Thesis of National University of Columbia, Bogotá.

Metcalf, Robert L. 1993. Destructive and Useful Insects; Their Habits and Control. McGraw Hill, Inc. New York.

Milliken, William, Robert P. Miller, Sharon R. Pollard, and Elisa V. Wandelli. 1992. The Ethnobotany of the Waimiri Atroari Indians of Brazil. Royal Botanic Gardens, Kew.

Milliken, William and Bruce Albert. 1996. The use of medicinal plants by the Yanomami Indians of Brazil. Economic Botany 50(1):10-25.

Mori, Scott A., Benedito V. Rabelo, Chih-Hua Tsou, and Douglas Daly. 1989. Composition and structure of an eastern Amazonian forest at Camaipi, Amapa, Brazil. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 5(1):3-18.

Mors, Walter B. and C.T. Rizzini. 1966. Useful Plants of Brazil. Holden-Day Inc., San Francisco.

Nagy, Nina E., Vincent R. Franceschi, Halvor Solheim, Trygve Krekling and Erik Christiansen. 2000. Wound- induced traumatic resin duct development in stems of Norway spruce (Pinaceae): anatomy and cytochemical traits. American Journal of Botany 87(3):302-313.

Nebeker, T.E., R.F. Schmitz and R.A. Tisdale. 1995. Comparison of oleoresin flow in relation to wound size, growth rates, and disease status of lodgepole pine. Canadian Journal of Botany 73:370-375.

Neels, Sabine. 2000. Yield, Sustainable Harvest and Cultural Uses of Resin from the Copal Tree (Protium copal; Burseraceae) in the Carmelita Community Forest Concession, Petén, Guatemala. M.S. Thesis, Dept. of Forestry, University of British Columbia, Vancouver.

Nogueiro-Neto, Paulo. 1997. Vida e Criação de Abelhas Indígenas Sem Ferrão. Editora Nogueirapis, São Paulo.

Paula, José Elias de and José Luiz de Hamburgo Alves. 1997. Madeiras Nativas.

Peters, Charles M. 1994. Sustainable Harvest of Non-timber Plant Resources in Tropical Moist Forest: an Ecological Primer. The Biodiversity Support Program, Washington, D.C.

Phillips, Michael A. and Rodney B. Croteau. 1999. Resin-based defenses in conifers. Trends in Plant Science 4(5):184-191.

Pinedo-Vasquez, Miguel, Daniel Zarin, Peter Jipp, and Jomber Chota-Inuma. 1990. Use-values of tree species in a communal forest reserve in northeast Peru. Conservation Biology 4(4):405-416.

Pittier, Henri. 1926. Manual de las Plantas Usuales de Venezuela. Editorial Elite (1939 compendium), Caracas.

Plowden, Campbell, Christopher Uhl and Francisco Assis de Oliveira. 2002. Breu resin harvest by Tembé Indians and its dependence on a bark-boring beetle. in J.R. Stepp, F.S. Wyndham, and R.K. Zarger (eds.) Ethnobiology and Biocultural Diversity. University of Georgia Press.

Plowman, T. 1984. The ethnobotany of coca (Erythroxylum spp., Erythroxylaceae). Advances in Economic Botany 1: 62-111. 147

Popp, Michael P., Jon D. Johnson and Thomas L. Massey. 1991. Stimulation of resin flow in slash and loblolly pine by bark beetle vectored fungi. Canadian Journal of Forest Research 21:1124-1126.

Projeto Radam. 1973. Levantamento de Recursos Naturais and Mapa Exploratório de Solos. Vol. 3. Programa de Integração Nacional. SUDENE. Ministério das Minas e Energia, Departmento Nacional da Produção Mineral, Rio de Janeiro.

Raffa, K.F., and A.A. Berryman. 1987. Interacting selective pressures in conifer-bark beetle systems: A basis for reciprocal adaptations? American Naturalist 129:234-262.

Rankin-de-Mérona, Judy M., Ghillean T. Prance, Roger W. Hutchings, Marlene Freitas da Silva, William A. Rodrigues, and Marie E. Uehling. 1992. Preliminary results of a large-scale tree inventory of upland rain forest in the Central Amazon. Acta Amazonica 22(4):493-534.

Reitz, Raulino. 1981. Flora Ilustrada Catarinense: Burseráceas. Conselho Nacional de Desenvolvimento Científico e Tecnológico - CNPq. Itajaí.

Rodrigues, Roberto Martins. 1989. A Flora da Amazônia. CEJUP, Belém.

Roubik, David W. 1989. Ecology and Natural History of Tropical Bees. Cambridge Univ. Press, Cambridge.

Ruel, Jonathan J., Matthew P. Ayres and Peter L. Lorio, Jr. 1998. Loblolly pine responds to mechanical wounding with increased resin flow. Canadian Journal of Forest Research 28:596-602.

Sales, Noêmia Pires de. 1993. Pressão e Resistencia: Os Índios Tembé-Tenetehara do Alto Rio Guamá e a Relação com o Território. União das Escolas Superiores do Pará (UNESPa), Belém.

Sales, Noêmia Pires de. 1994. Os Tembé no Alto Rio Guamá: Reelaborações Étnicas - Identidade e Território. UNAMA - Universidade da Amazônia, Belém.

Salomão, Rafael de Paiva and Pedro L.B. Lisboa. 1988. Análise ecológica da vegetação de uma floresta pluvial tropical de terra firme, Rondônia. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 4(2):195- 233.

Salomão, Rafael de Paiva, Manoela F.F. da Silva, and Nelson Araújo Rosa. 1988. Inventário ecológico em floresta pluvial tropical de terra firme, Serra Norte, Carajás, Pará. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 4(1):1-46.

Salomão, Rafael de Paiva. 1991. Uso de parcelas permanentes para estudos da vegetação da floresta tropical úmida. I. Município de Marabá, Pará. Boletim do Museu Paraense Emílio Goeldi, sér. Botânica 7(2):543-604.

Schmitt, J.J., T.E. Nebeker, C.A. Blanche and J.D. Hodges. 1988. Physical properties and monoterpene composition of xylem oleoresin along the bole of Pinus taeda in relation to southern pine beetle attack distribution. Canadian Journal of Botany 66:156-160.

Schroeder, L.M. 1990. Duct resin flow in Scots pine in relation to the attack of the bark beetle Tomicus piniperda (L.)(Col., Scolytidae). Journal of Applied Entomology 109:105-112.

Schultes, Richard E. and R.F. Raffauf. 1990. The Healing Forest: Medicinal Toxic Plants of the Northwest Amazon. Vol. 2. Dioscorides Press, Portland. 148

Secoy, D.M. and A.E. Smith. 1983. Use of plants in control of agricultural and domestic pests. Economic Botany 37(1):28-57.

Sharma, K. Krishan, A. Bhattacharya and S.N. Sushil. 1999. Indian lac insect, Kerria lacca, as an important source of honeydew. Bee World 80(3):115-118.

Sheldon, Sallie P. and Robert P. Creed, Jr. 1995. Use of a native insect as a biological control for an introduced weed. Ecological Applications 54(4):1122-1132.

Siani, A.C., M.F.S. Ramos, O. Menezes de Lima Jr., R. Ribeiro dos Santos, E. Fernandez-Ferreira, R.O.A. Soares, E.C. Rosas, G.S. Susunaga, A.C. Guimarães, M.G.B. Zoghbi, M.G.M.O. Henriques. 1999(1). Evaluation of anti-inflammatory-related activity of essential oils from the leaves and resin of species of Protium. Journal of Ethnopharmacology 66(1):57-69.

Siani, Antonio C., Mônica F.S. Ramos, Anderson C. Guimarães, Glória S. Susunaga and Maria das G.B. Zoghbi. 1999(2). Volatile constituents from oleoresin of Protium heptaphyllum (Aubl.) March. Journal of Essential Oil Research 11:72-74.

Slansky, Frank Jr. 1992. Allelochemical-nutrient interactions in herbivore nutritional ecology. Pp. 135-174 in eds. Gerald A. Rosenthal and May R. Berenbaum. Herbivores: their Interactions with Secondary Plant Metabolites. Academic Press, Inc., San Diego.

Spruce, Richard. 1970. Notes of a Botanist on the Amazon and Andes. Volume 2. Johnson Reprint Corp., New York.

Stark, R.W. 1982. Generalized ecology and life cycle of bark beetles. Pp. 21-45 in eds. Jeffrey B. Mitton and Kareen B. Sturgeon. Bark Beetles in North American Conifers. University of Texas Press, Austin.

Storer, Andrew J., David Wainhouse and Martin R. Speight. 1997. The effect of larval aggregation behavior on larval growth of the spruce bark beetle Dendroctonus micans. Ecological Entomology 22: 109-115. ter Steege, Hans and Gerold Zondervan. 2000. A preliminary analysis of large-scale forest inventory data of the Guiana Shield. pp. 35-54 in Hans ter Steege (ed.) Plant Diversity in Guyana. Tropenbos Foundation, Wageningen, the Netherlands.

Thompson, F. Christian. 1969. Contribution to a Generic Revision of the Neotropical Milesinae (Diptera: Syrphidae). PhD Dissertation in Entomology, University of Massachusetts.

Tisdale, Robert A. and Evan Nebeker. 1992. Resin flow as a function of height along the bole of loblolly pine. Canadian Journal of Botany 70:2509-2511.

Tomlin, Elizabeth S., Rene I. Alfaro, John H. Borden and Fangliang He. 1998. Histological response of resistant and susceptible white spruce to simulated white pine weevil damage.

Turner, B.L.II and Charles H. Miksicek. 1984. Economic plant species associated with prehistoric agriculture in the Maya lowlands. Economic Botany 38(2):179-193. van den Berg, Maria E. 1984. Ver-o-Peso: The ethnobotany of an Amazonian market. Advances in Economic Botany 1: 140-149.

Vega, Gilberto M., Roberto R. Soto, and Luis E.A. Duarte. 1984. Estudio Dendrologico de Columbia. Republica de Columbia, Bogotá. 149

Wainhouse, D., R. Ashburner, E. Ward and R. Boswell. 1998. The effect of lignin and bark wounding on susceptibility of spruce trees to Dendroctonus micans. Journal of Chemical Ecology 24(9):1551-1561.

Whitten, Norman E., Jr. 1976. Sacha Runa: Ethnicity and Adaptation of Ecuadorian Jungle Quichua. University of Illinois Press, Urbana.

Whitten, Dorothea and Norman E. Whitten, Jr. 1988. From Myth to Creation. University of Chicago Press, Urban.

Zoghbi, Maria das G.B., Emidio V.L. da Cunha and Wilson Wolter Filho. 1993. Essential oil of Protium unifoliatum (Burseraceae). Acta Amazonica 23(1):15-16.

Zoghbi, Maria das G.B., Jane B.G. Siqueira, Esther L.A. Wolter, and Orlando L.P. Júnior. 1994. Constituíntes químicos de Protium paniculatum (Burseraceae). Acta Amazonica 24 (1-2): 59-62.

Zoghbi, Maria das G.B., José G.S. Maia and Arnaldo I.R. Luz. 1995. Volatile constituents from leaves and stems of Protium heptaphyllum (Aubl.) March. Journal of Essential Oil Research 7:541-543.

Zoghbi, M.G.B., E.H.A. Andrade, A.S. Santos, A.I.R. Luz, and J.G.S. Maia. 1998. Volatile constituents of the resins from Protium subserratum (Engl.) Engl. and Tetragastris panamensis (Engl.) Kuntz. 150

APPENDIX 3-A SCIENTIFIC SPECIES AND COMMON NAMES OF RESIN YIELDING SPECIES OF NEOTROPICAL BURSERACEAE

TAXA REGION/GROUP COMMON NAMES REF. PROTIUM P. altsonii Sandw. (includes syn. P. Venezuela catamajaca, chipoi-yek 8 paraense) Guyana tsepur 8 Surinam tiengi-monnie 8 Brazil (Amapa) breu mescla 8 Brazil (Amazonas) breu branco 8 Brazil (Ka’apor) ara-kanei’y, kandeiape’y, 1,8,3 kanei’y pitag Brazil (Tembé) hykata’ywching 2 P. apiculatum Swart Brazil breu grande 13 P. aracouchini (Aubl.)March Brazil icica 12 Brazil (Tembé) hykata’ywpihun 2 P. brasiliense (Spreng.)Engl. Brazil icica 12 P. carana Marchand* Columbia caraña 7 Venezuela caraña 18 Brazil icica 12 P. crassipetalum Cuatrec. Amazon (Kuripako) ma-mee-ree’-ma 18 Columbia (Amazon) caraño 19 P. decandrum (Aubl.)Marchand Brazil breu, breu vermelho 13 Brazil (Ka’apor) hikeuteu ‘y pihun, 3, 1 kanei’y pitag, yawar-’a’y Brazil (Tembé) hykata’ywra 2 P. elemifera* Columbia elemi, tacamahaca 7 P. fibriatum Swart Ecuador (Quijos Quichua) shillquillo 5 P. giganteum Engl. Columbia (Amaz/Piapoco) kirreri 19 P. giganteum Engl. var. crassifolium Fr. Guiana encens blanc, encens gris 8 P. giganteum var. giganteum Brazil (Pará) breu branco, breu manga, 8 breu preto, breu vermelho Brazil (Ka’apor) breu branco, kandeiaka’y 3,8,1 hikeuteu ‘yw chig, kandei-y tuwyr Guyana (Arawak) haiwa 10 Brazil icica 12 P. heptaphyllum (Aubl.) Marchand ssp. Panama caraña, copal, elemi, 6 heptaphyllum (includes P. tacamahaca) sasafras Amazon (Tanimuka) hee-ta-ma-ka 18 Amazon (Kubeo) o-mo-ta’ 18 Columbia anime, anime blanco, brea, 6,19, caraña, guacamayo, 18,7 guacharaco, incienso, pergamin, tacamahaco Venezuela currucai, tacamahaco 18, 6 Guyana (Arawak,Carib,Warrau) haiowa, sipipio, shipu, 12 Brazil almecega, icica 16, 9, breu branco do campo 13,12 Brazil (Ka’apor) ara-kanei’y, hikeuteu’y, 3 breu branco verdadeiro, kanei’y pitag, yawar-’a-’y 151

Appendix 3-A (Continued) TAXA REGION/GROUP COMMON NAMES REF. P. icicariba (DC.) Marchand Venezuela guacamayo 18 Brazil Brazilian elemi, 15, 9 almecega, jauaricica, brea branca, breu branco Venezuela tacamahaco 18 P. llanorum Cuatrec. Columbia (Amazon) hugucal, incienso 19 P. macrophyllum H.B.K. Columbia anime, cunday, 19, 7 guacharaco hediondo P. nervosum Cuatrec. Columbia anime 19 P. niloi Pires Brazil (Tembé) kiriwa’yw 2 P. nodulosum Swart Ecuador (Quijos Quichua) shillquillo 5 P. opacum Swart ssp.opacum Brazil breu preto 13 Peru 17 P. pallidum Cuatrec. Brazil (Ka’apor) breu branco, 3,1 hikeuteu ‘yw chig, kanei-’y tuwyr, yawara-’a’cu’y Brazil (Tembé) hykata’ywching 2 P. paniculatum Engl. var. paniculatum Peru copal, shininga arana 8 Brazil almecega 16 P. pilosum (Cuatrec.)Daly Fr. Guiana encens 8 Brazil (Pará) breu branco 8 P. polybotryum (Turcz.)Engl. ssp. Columbia (Amazon) caraño 19 polybotryum Brazil (Ka’apor) hikeuteu-iran-’yw 3, 1 kanei-ape’y, yawar-’a-’y Brazil (Tembé) kiriwa’yw 2 Fr.Guiana encens blanc, wet-sali 8 Fr.Guiana (Paramaka) moni 8 Surinam rode bast tingimoni, 8 ajawabali, joelliballi beleroe Surinam (Carib) peraka, tamoene poeleka 8 Brazil (Pará) breu branco 8 Brazil (Ka’apor) akakandei’y, kandeiape’y 8 P. puncticulatum J.F. Macbr. Peru 17 P. robustum (Swart) D. Porter Peru 17 P. sagotianum Marchand (includes syn. Brazil breu branco 13 P. insigne) Brazil (Ka’apor) kanei-aka’y, 1,3 yawara’a’cu’y Brazil (Tembé) hykata’ywching 2 Columbia currucay, urrucay, caraño 7,21,19 Ecuador (Quijos Quichua) shillquillo 5 Venezuela cabimbo 18 P. spruceanum (Benth.)Engl. Columbia (Amazon) incienso 19 Brazil (Ka’apor) kandei’y tuwyr, 8,1 kanei-’y tuwyr Brazil (Tembé) hykata’ywching 2 P. tenuifolium Engl. Brazil breu, breu preto 13 Brazil (Ka’apor) kanei-ape’y, yawar-’a-’y 1,3 Brazil (Tembé) hykata’ywching 2 P. trifoliolatum Engl. Brazil (Tembé) hykata’ywrang 2 P. unifoliolatum Spruce ex Engl. Brazil (Yanomami) warapé kohi 14 152

Appendix 3-A (Continued) TAXA REGION/GROUP COMMON NAMES REF. Protium spp. Central America caraño, copal, fontole, 4 fosforito, pom, tontol Ecuador anime blanco 4 Columbia anime, caraño, currucay 4 Columbia (Amazon) breo, copal, de-ep, tovake 19 tamamuri, tatamoco Columbia (Amaz/Kubeo) taloake 19 Columbia (Amaz/Kurripaco, Yeral kakaine, keterre, tabaiba 19 Columbia (Andina) carano, incienso 19 Columbia (Caribe) indio desnudo 19 Venezuela anime, azucarito, caraño 4 Guyana haiawa, kurokay, porokay 4 Fr. Guiana encens blanco, tinguimoni, 4 encens gris, encens rojo Brazil almecega, aruru 4 Bolivia caraño 4 TETRAGASTRIS T. altissima (Aubl.)Swart Peru 17 Columbia guacamayo, tacamahaca 7 Brazil breu manga, sali 13 Brazil (Ka’apor) breu manga, waruwa’y 3 Brazil (Tembé) iwapepirangyw 2 T. mucronata (Rusby) Swart Columbia (Amazon) copal 19 T. panamensis (Engl.)Kuntze Columbia (Amazon) caraño 19 Columbia (Amaz/Piapoco) kirrei 19 Brazil barrotinho, breu preto,sali 13 Brazil (Ka’apor) breu preto, papara-’yg, 3 waruwa-ywa-pitag-’y TRATTINNICKIA T. aspera (Standl.)Swart Columbia (Amazon) caraño, echicorai-guna 19 Columbia (Andina, Pacific/Cholo caraño, ampó 19 T. burserifolia Mart. Brazil amescla, 13 breu sucuruba branco Brazil (Ka’apor) kyryhu-’y, kyryh’y-’y 1,3 T. rhoifolia Willd. Brazil breu sucuruba 13 Brazil (Ka’apor) breu mescla, kyryhu’-y, 3 kyryh’y-’y Brazil (Tembé) kiriwa’yw 2 Trattinnickia spp. Columbia (Andina) caraño 19 Columbia (East Llanos) caraño 19 Brazil amesclao 4 BURSERACEAE (non-specified) Peru caraña, caraño, copal 8 Surinam salie, tiengi-monnie 8 Brazil (west Amaz) breu sicantá, sicantá 8 Brazil (central/south) almecegueira, almesca, 8 almisca, almiscar, amescla, icicariba Bolivia isígale, isigo 8 References: 1)Balée and Daly,1990; 2)Balée,1987; 3)Balée,1994; 4)Chichignoud et al.,1990; 5)Grimes et al., 1994; 6)Correa and Bernal,1989; 7)Cortes, n.s., and Renner et al.,1990; 8)Daly,1987; 9)Howes,1949; 10)Johnston and Colquhoun,1996; 11) Levi-Strauss,1952; 12)Martini et al.,1993; 13)Milliken and Albert,1996; 14)Mors and Rizzini,1966; 15)Paula and Alves, 1997; 16)Pinedo-Vasquez et al.,1990; 17)Pittier, 1926; 18)Shultes and Raffauf, 1990; 19)Vega et al., 1984. Notes: * - questionable species

CHAPTER 4

THE ASSOCIATION OF AN ALIPUMILIO FLY (DIPTERA: SYRPHIDAE) WITH

BURSERACEAE TREE RESINS IN THE EASTERN BRAZILIAN AMAZON 154

ABSTRACT

There are numerous species of flies (Diptera) whose larvae develop in plant saps or rotting exudates.

There are relatively few examples of immature flies that have adapted to resin flows of coniferous or angiosperm trees. Most of these cases have been members of primitive syrphid genera (Syrphidae) including

Cheilosia in the temperate region and Alipumilio in the neotropics. A recent study of resin harvested by

Tembé Indians in the eastern Brazilian Amazon has revealed a potentially new species of Alipumilio that develops in resin masses on various species of Burseraceae trees. These resin flows are primarily stimulated by the actions of a previously undescribed species of bark-boring weevil in the genus Sternocoelus

(Coleoptera: Curculionidae). Resin lumps are sometimes also associated with severe physical tree wounds.

This fly larva’s morphological features, presence and movement in the resin mass, and unsuccessful rearing apart from fresh resin indicate that this Alipumilio species may be consuming bacterial and fungal spores as well as sap materials coming out of tree wounds. While Sternocoelus weevils are found in resin lumps in a range of Protium and other Burseraceae tree species in the region, Alipumilio larvae were only found in the resin of some of these species. The study speculates that its absence on some species may be due to chemical properties of resin types that inhibit build-up of micro-organisms that nourish this kind of larvae.

If this distinction is borne out, the presence or absence of mycophagous syrphid larvae could serve as an initial indicator of certain medicinal properties of resins in this diverse group of trees found throughout the Amazon.

INTRODUCTION

Almost 6,000 species of syrphid flies (Diptera: Syrphidae) commonly called flower or hover flies have been described so far (Vockeroth and Thompson, 1987; Sommaggio, 1999). Adult syrphids are often mimics of stinging Hymenoptera, and many are important pollinators (Vockeroth and Thompson, 1987;

Teskey, 1976; Speight, 1978; Ervik and Feil, 1997). Larvae of the family have diverse feeding patterns that include mycophagous, phytophagous, saprophagous and entomophagous habits (Vockeroth and Thompson, 155

1987; Ferrar, 1987; Rotheray and Gilbert, 1999). They are typically found under the bark, tree exudates, decaying sapwood of living trees, dead rotting trees, or under fallen leaves. Their diet may include bacterial or fungal spores, sap, cambial tissue, decaying wood, aphids or immature ants (Hartley, 1961; Teskey, 1976;

Perry and Stubbs, 1978; Gilbert, 1986; Ferrar, 1987; Foote, 1987; Vockeroth and Thompson, 1987; Rotheray and Gilbert, 1999). There are well documented cases where beetles (Coleoptera) have overcome resin defenses of coniferous trees (Raffa and Berryman, 1987; Phillips and Croteau, 1999), but apart from Cecidomyiidae gall midges, few flies have apparently been able to circumvent the potent physical and chemical deterrents posed by plant resins (Teskey, 1976; Pyenson, 1980; Coulson and Witter, 1984; Thompson, 1985).

Cladistic analysis disputes the division of Syrphidae into three monophyletic sub-families (Rotheray and Gilbert, 1999), but closely related genera typically grouped in the primitive Eristalinae or Milesinae provide the best examples of flies that have adapted to resin-based tree defenses (Rotheray and Gilbert, 1999;

Rotheray et al. 2000). Hellrigl (1992) described finding larvae of the spruce resin fly Cheilosia morio Zett. in resin lumps on the trunks of spruce trees in northern Alps region of Italy. While these resin lumps resembled ones made by attacks of the wood-boring beetle Dendroctonus micans Kug., observations indicated that the flies were directly responsible for the resin flow in these trees. In Washington State, U.S.A., Burke (1905 cited in Teskey, 1976 and Rotheray, 1990) observed that Cheilosia alaskensis Hunter fed on the sap and cambium of hemlock (Tsuga heterophylla (Rafin.)Sarg.) trees by scraping cavities next to the sapwood. In the neotropics, some larval specimens from the Alipumilio Shannon genus were reared in the pitch of Araucaria australis trees in southern Brazil (Thompson, 1972). Larvae of Alipumilio femoratus Shannon were reared in the resinous sap of a tree in the Ecuadorian Amazon (Rotheray et al., 2000; Whitten and Thompson, personal communication). Only a dozen specimens of Alipumilio found from Argentina to Mexico have so far accounted for nine species, so the diversity of this genus and examples of resin resistant lifestyles is probably much greater than is reflected in its poor representation in collections and field research (Thompson, 1972).

A recent study of resin harvested from the trunk of various Burseraceae trees in the Alto Rio Guamá

Indigenous Reserve in Pará state, Brazil (Plowden et al., 2002) has revealed another and probably new species of Alipumilio (Rotheray, personal communication 2001) whose larvae shares the unusual habit of developing in tree resin. Similar larvae were found in another eastern Brazilian Amazon site near the Mojú River. The 156

Burseraceae trees and the resin that exudes from bark wounds as a clear to milky white liquid (Daly, 1987;

Gentry, 1993) are most commonly known in the Brazilian Amazon as breu (Rodrigues, 1989). The aromatic and sticky balsam resins of breu trees have been collected by forest dwellers throughout the Amazon for subsistence and commercial purposes including incense, flavors, medicinal products, illumination, handicraft glues, and caulking material for wooden boats (Mors and Rizzini, 1966; Reitz, 1981; Plowman, 1984; van den

Berg, 1984; Balée and Daly, 1990; Schultes and Raffauf, 1990; Daly, 1992; Lima et al., 1992; Balée, 1994;

Boom, 1996; Siani et al., 1999a). Stingless bees and other invertebrates also utilize the resin for various purposes (Chapter 3).

The larvae of a previously undescribed species of bark-boring weevil Sternocoelus Kuschel (O’Brien, personal communication 2000) are primarily responsible for the build-up of resin on at least a dozen species of trees in the Burseraceae genera Protium Burman, Tetragastris Gaertner, and Trattinnickia Willd. at several sites in the eastern Brazilian Amazon (Chapter 3). Like other phloem boring beetles, this weevil’s larva first uses its mandibles to chew through the tree’s outer bark to access more nutrient rich inner bark. It makes simple cylinder shaped bore holes in the cambium that average 0.7 cm deep and sever resin ducts interspersed in this tissue. As resin drips out of the wound onto the trunk and begins to solidify, the larva burrows into the malleable resin and shapes a chamber inside the resin lump. Fly larvae were also sometimes found in resin lumps associated with weevil larvae. This report will present information about the natural history of this

Alipumilio resin fly and its potential use as an indicator of resin anti-microbial potency.

STUDY SITES

The main study area (Tekohaw site) was an area of about 500 ha of closed tropical rainforest near the

Tembé Indian village of Tekohaw (020 37.7’ S; 460 33.1’ W.). This village is located on the Gurupi River in the Alto Rio Guamá Indigenous Reserve (eastern Pará state, Brazil). A second study area (Mojú site) in eastern Pará was located in a similar forest on state road PA150 near the town of Mojú (020 11.5’ S; 480 49.1’

W.) Both forests are dominated by trees in the Lecythidaceae, Leguminosae, Sapotaceae, and Burseraceae families (Baleé, unpublished data; Lockman, personal communication). 157

MATERIAL AND METHODS

In June and July 1998, a team of Tembé Indian men located 250 breu trees within 5 km of the village of Tekohaw. The tree’s common name as well as tree size, tree status (live, injured, dead), forest type (“terra firme” and “baixo”) were recorded. Each tree was marked with a numbered aluminum tag, and leaves were collected from a sample of trees to assist with identification. Measurements were made on the number and weight of resin lumps during this initial and follow-up harvests in November 1998, March and July 1999. The number of insect larvae found in resin lumps was measured in a sample of trees. Once the larvae were distinguished as weevil and fly larvae, they were separated into these categories and counted separately.

Lengths, widths and weights of fly larvae stored in alcohol were later measured in the lab since precise instruments were not available to do so in the field.). Type and paratype specimens will be deposited in Brazil.

A similar protocol was used to measure resin lumps and tree-insect relationships on trees at the Mojú site in March 2000. Resin lumps were harvested from several species of unwounded Protium and Tetragastris trees and one Trattinnickia rhoifolia Willd. tree that had a large amount of resin exuding from a chainsaw wound. Leaf specimens were collected from a sample of Burseraceae trees at the site including all those that had resin to identify host tree species of the weevil and fly.

Since only a small fraction of immature tropical syrphids have been identified (Thompson, 1987), several attempts were made to raise the Alipumilio larvae. During the initial attempts, fly larvae in resin lumps collected at the Tekohaw site were placed in mesh cages and jars with mesh tops. Fly larvae living in resin collected from the Trattinnickia tree at the Mojú site were placed in glass vials with medium used for rearing

Drosophila fruit flies. 158

RESULTS

RESIN LUMPS AND HOST TREE PREFERENCES

The initial resin harvest at Tekohaw showed that dead trees had significantly larger amounts of resin on them than live trees. Large trees were also colonized more often by Sternocoelus weevils than small trees

(Chapter 3). Leaf specimens identified by botanists at the Museu Goeldi in Belém revealed that at least eight species of Protium including P. pallidum Cuatrec, P. trifoliatum Engl., P. decandrum (Aublet) Marchand, P. morii Daly, P. giganteum Engl. var. giganteum, P. polybotryum (Turcz.) Engl., and P. glabrescens Swart. and

Tetragastris panamensis (Engl.) O Kuntze had resin lumps at the Tekohaw site. P. pallidum and P. trifoliatum were generally classified as “breu branco” (white breu) by local Tembé while the other six species were considered different varieties of “breu sarara.” While Sternocoelus weevil larvae were found in resin lumps in both types of trees, Alipumilio fly larvae were only found in resin lumps on “sarara” type trees.

The Mojú site also had resin accumulation on P. giganteum as well as on P. paniculatum Engl.,

Tetragastris altissima (Aubl.) Swart., and Trattinnickia rhoifolia Willd. The last species was only encountered with resin on the trunk of a tree wounded with a chainsaw. Weevil larvae were again found in resin lumps on all of these types of trees. Fly larvae, however, were only found in the resin of the wounded Trattinnickia tree.

None of the trees identified as P. krukoffii Swart., P. guianense (Aublet) Marchand, P. pernavatum Cuatrec, P. pilosum (Cuatrec) Daly, Trattinnickia lawrencei var. bolivianum Swart., or Trattinnickia burseraefolia Mart. at either site had any resin on the trunk.

CHARACTER OF RESIN LUMPS

Resin associated with weevil attacks on breu trees was usually found in round lumps affixed to the lower part of tree trunks although some lumps were occasionally seen much higher up. Lumps were often wedged in the crook between the trunk and a buttress or stilt root. Lumps were typically 2 to 6 cm thick and averaged 339 ± 59 g in the initial Tekohaw harvest. The weevil larva forms a cavity in the fresh resin that is 3 to 8 mm deep and extends almost to the inner edge of the resin lump (Figure 3.3). Fly larvae were found in 159 resin lumps both with and without weevil larvae at both sites. When resin lumps contained both kinds, however, the flies were in a separate section of the lump that usually occupied no more than 25% of the lump.

In cases where flies were found in resin without weevils, some lumps had bore holes behind them indicating prior weevil activity.

One distinctive feature of a resin mass containing fly larvae was that it had a dark color and taffy-like consistency that stretched out when pulled apart. This was different from the cake-like quality of resin associated with weevils that was sticky but readily broke into separate pieces. While weevil larvae were usually found quiescent in a resin chamber or occasionally in a bore hole, maggots were almost always observed moving in their section of resin. This activity that presumably accompanies feeding, therefore, seemed responsible for the transformation of resin consistency. Dried resin lumps further revealed the different feeding strategies of the immature resin weevil and fly. Resin lumps occupied solely by weevils had distinct layers indicating different resin flow events. These layers logically resulted from successive boring in the cambium during the weevil’s slow larval development. Resin sections that were inhabited by fly larvae, however, dried in a well-mixed matrix interspersed with narrow meandering tunnels.

NUMBER, SIZE, AND IDENTITY OF FLY LARVAE

During the initial harvest at Tekohaw, the identity of the fly larvae was not known, so precise counts are not available. Finding all fly larvae in resin lumps was also more difficult than locating immature weevils since the maggots are immersed in a mass of sticky resin. During the second round of resin harvests at

Tekohaw, 9 out of 38 (24%) of resin lumps that had weevil or fly larvae in them had at least one fly larva in them, and 6 of these (16%) had only fly larvae. This measure overestimates the number of resin lumps associated only with flies since only 15 lumps out of 910 examined (1.6%) did not have weevil bore holes behind them. In the chainsaw damaged tree at Mojú, 5 out of 22 (23%) resin lumps with larvae had only fly larvae in them, but only 3 of these lumps had no weevil bore holes behind them.

At both sites, two “types” of fly larvae were found. Both types had similar mouth hooks, but one

“type” that had three pairs of spiracular slits was considerably smaller than the “type” that had numerous spiracular slits radially oriented around the ecdysial scar (Table 4.1). Rotheray (personal communication 160

2001) believes that these two “types” of larvae probably represent Stage 2 and 3 instars of the same species of

Alipumilio. The average body width of the larvae changed from 1.2 to 3.0 mm and the average length went from 5.1 to 7.7 mm in the shift from Stage 2 to Stage 3 larvae. The average wet weight during these stages was 4.2 and 37.1 mg respectively, but these figures should probably be increased to account for weight loss that occurs during storage in alcohol. Rotheray further noted that the form of the thorax and the head of this larva is very similar to A. femoratus Shannon. The third stage larva has a number of interesting and novel features, such as a very flattened anal segment, different arrangement of spiracular openings to A. femoratus and has an almost complete band of spicules around the segments (Rotheray, personal communication 2001).

The number of Stage 3 larvae present in a single resin lump was rarely more than one at the Tekohaw site. When Stage 2 larvae were present, however, there were usually several and as many as fifty in one lump.

This pattern was repeated at Mojú where there was rarely more than one third stage larva present in one resin lump, but two to twenty-five second stage larvae were present in some lumps in the wounded Trattinnickia tree.

None of the attempts to rear these breu resin flies was successful. Over a period of months, the resin lumps collected at Tekohaw that were placed in cages and jars became covered with mold and eventually dried out before any mature flies emerged. Fly larvae placed in vials with Drosophila medium also did not survive to maturity (Overal, personal communication 2000).

DISCUSSION

The finding that Sternocoelus weevil attacks and resin amounts were highest in large and dead trees is consistent with earlier research on temperate coniferous trees (Lieutier et al., 1993, Nebeker et al., 1995).

While the Alipumilio fly is sometimes capable of colonizing trees without these weevils, it seems that in most cases this fly relies on the weevil to first bore into a tree before it can access a food source.

Figuring out the diet and feeding method of this Alipumilio fly during its immature stages is one of the principal natural history questions raised by this investigation. The answers are central to determining if the fly plays some direct role in provoking the flow of resin that is harvested by people. As noted, several 161 species of closely related Cheilosia larvae, feed directly on cambial tissues so they are not limited to feeding on tree sap that is exuding due to other causes (Rotheray, 1990; Hellrigl, 1992; Rotheray and Gilbert, 1999). An examination of the mouth parts of A. femoratus, however, did not produce conclusive evidence regarding the diet of this species. It has well-developed mouth hooks and other characteristics that indicate it could feed on bacteria, yeasts and other micro-organisms in its host sap or it might be a phytophage that feeds on cambial tissues where the sap exudes (Rotheray et al. 2000). Its overall morphological characteristics place it in the most basal position of the Syrphidae in proximity to the Eumerus and Merodon genera that are primarily considered mycophagous (Rotheray and Gilbert, 1999; Rotheray et al., 2000). This study seems to affirm the probability that Alipumilio does not feed directly on the cambium since it was exclusively found moving about in resin masses. Their frequent movement in the resin is consistent with the filter-feeding mode of mycophagous larvae living in semi-liquid mediums (Rotheray and Gilbert, 1999).

There have been several studies of the volatile components of Burseraceae resins (Zoghbi et al., 1993,

1994, 1995, 1998; Siani et al., 1999a, 1999b), but studies have not yet analyzed these materials for the presence of microbes or nutrients that might come out of wounds along with the resin. The failure of the initial attempts to rear these fly larvae on small lumps of resin removed from the tree or the Drosophila medium suggest the maggots require a fresh source of organic material from the host tree in order to complete their development. It is possible that bacteria or fungus attracted to the wound site or brought in by the weevils need to build up to a certain level before these larvae can meet their nutritional needs. In order to successfully rear several other types of immature Syrphid flies, Rotheray (personal communication 2001) needed to keep them supplied with an ongoing supply of fresh exudate from their typical host tree.

The Alipumilio larva’s discriminating diet raises an interesting question about the variation in host- tree preferences between the weevil and the fly. Weevils attacked many but not all species of Burseraceae in the same forest while Alipumilio were apparently more selective in their host preferences. The notable difference was the presence of fly larvae at Tekohaw in “sarara” type resin lumps and its total absence in

“branco” type resin lumps at Tekohaw and Mojú. While “branco” and “sarara” type breu trees represented various taxonomic species, “branco” type resin lumps all formed as white masses that slowly turned gray.

Fresh “sarara” type resin also dripped out of the tree as a milky white liquid, but its resin lumps quickly 162 darkened and were almost black when dry. “Sarara” lumps were also covered with unidentified gritty and small plant-like material.

These visible differences between “branco” and “sarara” type resin are likely due to differences in their chemical composition. Several studies have found significant variation in the volatile components of resins from various Protium species (Zoghbi et al., 1993, 1994, 1995, 1998; Siani et al., 1999a, 1999b) that may impart their resins with specific medicinal properties. P. heptaphyllum (Aublet) Marchand is one type of

“branco” breu tree whose resin is used by native peoples in Northeastern Brazil as a treatment for infectious skin diseases, although extracted essential oils did not show great effectiveness against several dermatophytes in one test (Lima et al., 1992). Another pharmacological test did demonstrate its potency as an anti- inflammatory agent (Siani et al. 1999a).

A comparison of the chemistry and microbial communities of various types of breu resin could help explain whether Alipumilio larvae are not found in the resin of certain host trees because it has physical or chemical properties that are inhospitable to them or because the resin simply lacks sufficient microbes to nourish them. It is interesting to note that “branco” resin sometimes earns a higher market price than the

“sarara” type so it has some medicinal or other properties that has made it more attractive to human consumers than flies.

Sommaggio (1999) has suggested that the large diversity and differences in environmental requirements in syrphid flies make them well suited as bioindicators. As this case demonstrates, the presence or absence of one probable mycophagous syrphid larvae in the resin of a group of closely related trees could be a first-level indicator of the anti-microbial potency of each tree species’ defense compounds. Other mycophagous larvae might serve a similar purpose in analogous situations. TABLE 4.1 BODY DIMENSIONS OF IMMATURE ALIPUMILIO FLIES FROM BREU RESIN AT TEKOHAW

Stage 2 Larvae (n = 9) Stage 3 Larvae (n=49) Dimension Mean ± S.E. Maximum Mean ± S.E. Maximum Body width (mm) 1.21 ± 0.09 1.53 2.98 ± 0.12 4.10 Body length (mm) 5.12 ± 0.40 6.80 7.66 ± 0.27 12.74 Spir.Plate width (mm)1 0.39 ± 0.02 0.47 1.05 ± 0.04 1.40 Weight (mg)2 4.24 ± 0.81 8.10 37.05 ± 3.29 100.70 1 - Posterior Spiracular Plate 2 - Wet weight of larvae preserved in 70% alcohol. 164

REFERENCES

Balée, William and Douglas C. Daly. 1990. Resin classification by the Ka’apor Indians. Advances in Economic Botany 8:24-34.

Balée, William. 1994. Footprints of the Forest: Ka’apor Ethnobotany - the Historical Ecology of Plant Utilization by an Amazonian People. Columbia University Press, New York.

Boom, Brian M. 1996. Ethnobotany of the Chácabo Indians, Beni, Bolivia. 2nd Ed. The New York Botanical Garden, New York.

Coulson, Robert N. and John A. Witter. 1984. Forest Entomology: Ecology and Management. John Wiley and Sons, New York.

Daly, Douglas Charles de Burgh. 1987. A Taxonomic Revision of Protium (Burseraceae) in Eastern Amazonia and the Guianas. PhD dissertation from The City University of New York.

Daly, Douglas C. 1992. New taxa and combinations in Protium Burm. f. Studies in neotropical Burseraceae VI. Brittonia 44(3):280-299.

Ervik, F. and J.P. Feil. 1997. Reproductive biology of the monoecious understory palm Prestoea schultzeana in Amazonian Ecuador. Biotropica 29(3):309-317.

Ferrar, P. 1987. A Guide to the Breeding Habits and Immature Stages of Diptera Cyclorrhapha, Part 1. Entomonograph, Volume 8. E.J. Brill/Scandinavian Science Press, Leiden.

Foote, B.A. 1987. Syrphidae (Syrphoideae). pp. 792-795 in Frederick W. Stehr (ed.) Immature Insects, Volume 2. Kendall/Hunt Publishing Company, Dubuque.

Gilbert, Francis S. 1986. Hoverflies. Cambridge University Press, Cambridge.

Gentry, Alwyn H. 1993. A Field Guide to the Families and Genera of Woody Plants of Northwest South America. The University of Chicago Press, Chicago.

Hartley, J.C. 1961. A taxonomic account of the larvae of some British Syrphidae. Proceedings of the Zoological Society of London 136:505-573.

Hellrigl, Von K. 1992. Die fichtenharzfliege Cheilosia morio Zett. (Dipt. Syrphidae) als physiologischer schädling an fichten in Südtirol. Anz. Schädlingskde., Pflazenschutz, Umweltschutz 65: 33-36.

Lieutier, François, Jacques Garcia, Paul Romary, Annie Yart, Hervé Jactel, and Daniel Sauvard. 1993. Inter- tree variability in the induced defense reaction of Scots pine to single inoculations by Ophiostoma brunneo-ciliatum, a bark-beetle-associated fungus. Forest Ecology and Management 59:257-270.

Lima, E.D., O.F. Gompertz, M.D. Paulo, and A.M. Giesbrecht.1992. Invitro antifungal activity of essential oils against clinical isolates of dermatophytes. Revista de Microbiologia 23(4):235-238.

Mors, Walter B. and C.T. Rizzini. 1966. Useful Plants of Brazil. Holden-Day Inc., San Francisco. 165

Nebeker, T.E., R.F. Schmitz and R.A. Tisdale. 1995. Comparison of oleoresin flow in relation to wound size, growth rates, and disease status of lodgepole pine. Canadian Journal of Botany 73:370-375.

Perry, Ivan and Alan E. Stubbs. 1978. pp. 65-73 in eds. Alan Stubbs and Peter Chandler. A Dipterist’s Handbook. The Amateur Entomologist, Volume 15.

Phillips, Michael A. and Rodney B. Croteau. 1999. Resin-based defenses in conifers. Trends in Plant Science 4(5):184-191.

Plowden, Campbell, Christopher Uhl and Francisco Assis de Oliveira. 2002. Breu resin harvest by Tembé Indians and its dependence on a bark-boring beetle. in J.R. Stepp, F.S. Wyndham, and R.K. Zarger (eds.) Ethnobiology and Biocultural Diversity. University of Georgia Press.

Plowman, T 1984. The ethnobotany of coca (Erythroxylum spp., Erythroxylaceae). Advances in Economic Botany 1: 62-111.

Pyenson, Louis L. 1980. Fundamentals of Entomology and Plant Pathology. AVI Publishing Company, Westport.

Sommaggio, D. 1999. Syrphidae: can they be used as environmental bioindicators? Agriculture Ecosystems and Environment 74(1-3):343-356.

Rotheray, Graham E. 1990. The relationship between feeding mode and morphology in Cheilosia larvae (Diptera, Syrphidae). Journal of Natural History 24:7-19.

Rotheray, Graham and Francis Gilbert. 1999. Phylogeny of palaearctic Syrphidae (Diptera):evidence from larval stages. Zoological Journal of the Linnean Society 127:1-112.

Rotheray, Graham E., M. Angeles Marcos-Garcia, E. Geoffrey Hancock, and Francis S. Gilbert. 2000. The systematic position of Alipumilio and Nausigaster based on early stages (Diptera, Syrphidae). Studia Dipterologica 7(1):133-144.

Schultes, Richard E. and R.F. Raffauf. 1990. The Healing Forest: Medicinal Toxic Plants of the Northwest Amazon. Vol. 2. Dioscorides Press, Portland.

Siani, A.C., M.F.S. Ramos, O. Menezes-de-Lima Jr., R. Ribeiro-dos-Santos, E. Fernandez-Ferreira, R.O.A. Soares, E.C. Rosas, G.S. Susunaga, A.C. Guimarães, M.G.B. Zoghbi, M.G.M.O. Henriques. 1999a. Evaluation of anti-inflammatory-related activity of essential oils from the leaves and resin of species of Protium. Journal of Ethnopharmacology 66(1):57-69.

Siani, Antonio C., Mônica F.S. Ramos, Anderson C. Guimarães, Glória S. Susunaga and Maria das G.B. Zoghbi. 1999b. Volatile constituents from oleoresin of Protium heptaphyllum (Aubl.) March. Journal of Essential Oil Research 11:72-74.

Sommaggio, D. 1999. Syrphidae: can they be used as environmental bioindicators? Agriculture Ecosystems and Environment 74 (1-3): 343-356.

Speight, Martin C.D. 1978. Flower-visiting flies. pp. 229-236 in eds. Alan Stubbs and Peter Chandler. A Dipterist’s Handbook. The Amateur Entomologist, Volume 15.

Städler, Erich. 1992. Behavioral responses of insects to plant secondary compounds. pp. 45-88 in eds. 174 in eds. Gerald A. Rosenthal and May R. Berenbaum. Herbivores: their Interactions with Secondary Plant Metabolites. Academic Press, Inc., San Diego. 166

Teskey, H.J. Diptera Larvae Associated with Trees in North America. Memoirs of the Entomological Society of Canada, No. 100. The Entomological Society of Canada, Ottawa.

Thompson, F. Christian. 1972. A contribution to a generic revision of the neotropical Milesinae (Diptera: Syrphidae). Arquivos de Zoologia 23(2):73-215.

Thompson, Lynne C. 1985. Insect pests of forests. pp. 509-551 in Robert E. Pfadt (ed.) Fundamentals of Applied Entomology. Collier Macmillan, New York. van den Berg, Maria E. 1984. Ver-o-Peso: The ethnobotany of an Amazonian market. Advances in Economic Botany 1: 140-149.

Vockeroth, J.R. and F.C. Thompson. 1987. Syrphidae. pp. 713-743 in J.F. McAlpine (ed.) Manual of Nearctic Diptera, Volume 2. Research Branch Agriculture Canada Monograph No. 28.

Zoghbi, Maria das G.B., Emidio V.L. da Cunha, and Wilson Wolter Filho. 1993. Essential oil of Protium unifoliatum (Burseraceae). Acta Amazonica 23(1):15-16.

Zoghbi, Maria das G.B., Jane B.G. Siqueira, Esther L.A. Wolter, and Orlando L.P. Júnior. 1994. Constituíntes químicos de Protium paniculatum (Burseraceae). Acta Amazonica 24 (1-2): 59-62.

Zoghbi, Maria das G.B., José G.S. Maia, and Arnaldo I.R. Luz. 1995. Volatile constituents from leaves and stems of Protium heptaphyllum (Aubl.) March. Journal of Essential Oil Research 7:541-543.

Zoghbi, Maria das G.B., E.H.A. Andrade, A.S. Santos, A.I.R. Luz, and J.G.S. Maia. 1998. Volatile constituents of the resins from Protium subserratum (Engl.) Engl. and Tetragastris panamensis (Engl.) Kuntz. CHAPTER 5

THE ECOLOGY, HARVEST AND MARKETING OF TITICA VINE ROOTS

(HETEROPSIS SPP.: ARACEAE) IN THE EASTERN BRAZILIAN AMAZON ABSTRACT

The aerial roots from various species of the hemiepiphytic vine Heteropsis (Araceae) locally known as “titica” have long been used in the Amazon region as a strong flexible material in construction and making handicrafts. In recent decades commercial demand for this product has increased for use in making wicker furniture. I conducted studies in the Alto Rio Guamá Indigenous Reserve and other sites in the eastern

Brazilian Amazon to describe and quantify several aspects of titica ecology and root biology (density, host tree size and type, root abundance, root growth and maturation rate), root harvesting, and transformation of the roots into a commercial product. I collected information on these topics by doing inventories of titica vines, accompanying harvesters, experimentally cutting roots, and interviewing titica dealers and furniture makers in urban areas.

The average density of host trees with titica roots in relatively undisturbed upland “terra firme” forest areas with moderate to light titica harvesting was 371 trees per ha with 115 of these having commercially harvestable roots. In these areas there was an average of 0.9 commercially harvestable roots and 2.4 non- harvestable roots per host tree generating a titica root density of 1332 roots per ha with 26% of these being commercially harvestable. Titica plants did not show a positive preference for particular types of host trees, but the number of titica roots per host tree did increase with host tree size. The vines were virtually absent from forest areas subject to periodic flooding. Titica was also rarely found on pioneer tree species in patches of secondary forest recovering from a fire that occurred almost twenty years ago.

Commercial quality titica roots need to be relatively thick (usually ≥ 3.5 mm) and have long (≥ 2 m) sections without nodes (bulges). In order to prepare roots for sale, nodes, bad pieces, and the root cortex were removed. Once the processed roots dried, their weight was reduced to 19% of the initial amount harvested in the forest. Since harvesters remove some nodes from roots immediately after pulling them from host trees, the weight of roots they end up selling is about 30% of the amount they bring out of the forest. Harvesters typically collected 10 to 40 kg of roots per day in the forest and depending on the amount and extent of help from other family members, they spent an extra day processing the roots from each day of harvesting. Since 169 dealers in the region usually pay about $1.00 per kg of processed roots, a harvester collecting 35 kg of roots per day will earn about $US 5.25 per day for the 7 days he spent traveling and harvesting roots and the 3 days he and his family spent harvesting and processing the roots. The revenue of first time harvests of titica roots was about $US 10 per ha.

My experimental cutting of mature titica roots showed that harvesting may have a severe impact on root survival. In two treatments where at least half of mature titica roots per host tree were cut about 4 m above the ground, an average of 63% cut roots died and only 16% showed some regrowth after seven months.

The roots that did start to grow back had an average of 1.7 new roots arising from them, but each cut root had only one commercial quality root to replace it. These commercial quality roots were growing at an average of

220 cm per year. The high mortality and slow regrowth following root harvest indicate that many decades may be needed in between intensive titica harvests in one area. Options for managing titica harvests include reducing the percentage of roots removed per plant, setting up blocks of forest that would be harvested in succession, and in some cases integrating titica harvests into long-term timber cutting cycles.

INTRODUCTION

Indigenous people and other forest dwellers in the Amazon region have long harvested Heteropsis

(Araceae) vines for use as a lashing material in construction and making handicrafts (Boom and Moestl, 1990;

Oliveira et al., 1991; Milliken et al., 1992; Balée, 1994; Paz y Miño C. et al., 1995). The strong flexible aerial roots of these hemiepiphytic plants have also been sold to make wicker (“rattan”) furniture, baskets and other woven articles in Brazil, Guyana, Venezuela, and Peru (Pio Corrêa, 1931; Madison, 1979a; Rodrigues, 1989;

Whitehead and Godoy, 1991; Balée, 1994; Berry et al., 1995; Hoffman, 1997; Troy and Harte, 1998; Bown,

2000). As deforestation and over-harvesting reduce supplies of rattan from Asian palm species, the demand for the roots of Heteropsis and other suitable vines from the Amazon seems to be increasing in both South

American domestic and international markets (Whitehead and Godoy, 1991; Hoffman, 1997; Troy and Harte,

1998). Like many non-timber forest products (NTFPs) from the region there is relatively little information 170 about the amounts and impacts of current harvesting and how much could be available if the harvest was managed on a sustainable basis.

There are about 13 species of Heteropsis plants (Croat, 1988; Gentry, 1993), mostly found in the

Amazon region. The species that have been most utilized are H. flexuosa (Kunth) Bunting, H. longispathacea

Engl., H. spruceana Schott, and H. oblongifolia Kunth (Pio Corrêa, 1931; Boom and Moestl, 1990; Oliveira et al., 1991; Balée, 1994; Berry et al., 1995; Paz y Miño C. et al., 1995, Troy and Harte, 1998). While H. jenmanii Oliver has also been widely referenced in the literature, this name is now properly considered a synonym for H. flexuosa (Bunting, 1979). The vernacular name most often used for the commonly harvested species in the Brazilian Amazon is “cipó titica” (titica vine)(Pio Corrêa, 1931; Rodrigues, 1989; Oliveira et al.,

1991; Balée, 1994, Troy and Harte, 1998).

Titica plants are one type of Araceae vines with a hemiepiphytic habit (sensu Putz and Holbrook,

1986; Appendix 5-A) found in primary tropical moist forests (Hoffman, 1987; Croat, 1988; Bown, 2000).

Unlike “true” epiphytes, the seeds of this secondary type of hemiepiphyte germinate on the forest floor and reach reproductive status in a host tree. Young seedlings appear to use a skototropic mechanism to grow along the forest floor toward shady spots in search of a support structure (Hoffman, 1975) as described for a related

Monstera vine (Strong and Ray, 1975). This support is usually provided by a host tree, but rocks have also served as a climbing substrate for Heteropsis (Croat, 1988). When a seedling reaches a potential host, it switches to a light seeking mode and grows up the trunk with the aid of fine adventitious anchoring (climber) roots that adhere to the bark (Wilder, 1992; Appendix 5-A; Figure 5.1). The juvenile stage is reached when the stem reaches the lower part of the host tree canopy (typically where the first main branch is located) and produces its first lateral branches with larger leaves (Ray, 1992; Hoffman, 1997; Figure 5.1). As the titica plant approaches adulthood, the plant’s original connection with the soil is broken when the lower stem senesces. Around this time, a second type of adventitious root called an absorbing root (Appendix 5-A) sprouts from the main titica stem or a lower branch and grows downward until it enters the ground and branches out to give the plant a renewed supply of water and nutrients from the soil (Kelly, 1985; Putz and

Holbrook, 1986; Croat, 1988; Benzing, 1990; Wilder, 1992; Hoffman, 1997, Bown, 2000; Figure 5.1). For the remainder of this chapter, a “titica root” will refer to the aerial portion of an absorbing root unless otherwise 171 specified. Titica plants can also reproduce vegetatively via a flagellar shoot that can arise from a main stem on a standing or fallen host tree. It may reclimb the same host tree or grow scandently on the forest floor in search of a new one (Hoffman, 1997).

A titica root usually descends in close contact with the host tree where it wraps around the trunk or is intertwined with other titica roots or roots of other vines. Occasionally a root that arises from a lateral branch hanging out from the host tree descends directly to the ground. Hoffman (1997) calls these two types of roots

“trunk roots” and “drop roots” (Appendix 5-A; Figure 5.1). In the process of growing down, a root apical meristem will often abort or be injured and form a node that is a pronounced bulge about 1 cm long (Wilder,

1989; Hoffman, 1997; Figure 5.2). In many cases, a replacement root (Appendix 5-A) of similar width will sprout above this node and continue the downward growth (Wilder, 1989, Wilder and Johansen, 1992).

While Heteropsis is an herbaceous monocot that is technically not woody (Putz, 1984a), its roots and stem are tough. Heteropsis roots are distinguished from other equally strong vines and are particularly attractive for commercial purposes because the dark epidermis and cortex can be easily peeled away to leave a smooth, blond, and rot-resistant inner stele packed with long strong fibers. This core can be further split along the vine’s length into slender pieces that retain strength and flexibility (Rodrigues, 1989; Whitehead and

Godoy, 1991; Potiguara and Nascimento, 1994; Hoffman, 1997).

The plant’s presumed abundance in primary forests and resilient growth has led to optimistic projections about titica harvest revenues for forest-based communities (Godoy and Whitehead, 1991; Paz y

Miño C. et al., 1995; Troy and Harte, 1998). Two recent studies, however, indicate that severed roots often die and harvesting all roots often kills the whole plant (Hoffman, 1997; Durigan, 1998). In Guyana, harvesting impacts are reduced by limiting harvest to the relatively small number of long drop roots that hang free from the host tree trunk (Hoffman, 1997). Most harvesters in Brazil, though, apparently pull down most mature roots (Troy and Harte, 1998; Durigan, 2000). There is, therefore, a need for more information about the basic biology of titica vines, harvesting intensity and economics, and regeneration potential under different harvesting regimes.

As part of my research to evaluate the ecological and economic aspects of several NTFPs in the Alto

Rio Guamá Indigenous Reserve, I conducted observations and field experiments to describe the harvest and 172 commercialization of titica vines in this reserve and regional markets. The main objectives of these titica studies were to: 1) quantify the density of titica plants and roots on a per area basis, 2) quantify the number and size of titica roots per host tree, 3) describe the relationship between titica plants and host tree diameter and type, 4) estimate the maturation time and natural mortality of titica roots, 5) quantify the impact of harvesting on root mortality and regrowth, 6) quantify the length and weight of roots harvested and transformation during commercial processing, 7) document the daily yield of commercial titica harvesting and the time involved in collecting and processing roots, and 8) describe the structure and economics of titica harvesting and marketing.

I will finally discuss how different management scenarios could address some of the problems apparently caused by current harvesting practices in the eastern Brazilian Amazon.

STUDY AREA

The main study area for observing titica ecology and harvest was the Alto Rio Guamá Indigenous

Reserve in eastern Pará state, Brazil. This 278,000 ha area of closed tropical rainforest is the principal homeland of the Tembé Indians. Studies were carried out near the village of Tekohaw on the Gurupi River and the village of Cajueira on the Uraím River in the southern part of the reserve. Additional observations on plants were made in similar forests on Fazenda 7, a private property near the city of Paragominas, Pará.

Dealers and furniture makers that handled titica roots were interviewed in the town of Gurupi on the Gurupi

River in Maranhão, and Belém, the capital city of Pará. 173

METHODS

TITICA BIOLOGY AND POTENTIAL HARVEST OF AERIAL ROOTS

Density of Titica Host Trees and Roots and Number of Roots per Host Tree

The density of host trees with titica plants, the number of roots per unit area, and the number of roots per host tree was estimated during November and December 1999 in three inventories at Tekohaw and

Cajueira. These sites represented a gradient of forest use and disturbance. The first inventory at Tekohaw was conducted in an area of forest about 1 km from the village that was heavily used by many of the village’s 25 household members in trips to farm fields, hunt, or collect forest products for personal use. The area was a mosaic of mostly intact “terra firme” type forest and secondary forest that had burned about 20 years ago.

Titica plants were surveyed in 1.7 ha divided into 48 unburned and 19 burned 10x25 m (250 m2) plots that were located through stratifed random sampling along 500 m long belt transects. The second Tekohaw inventory was conducted in an area of forest about 3 km from the village. This area was next to a commonly used trail, but use of forest resources was less extensive there than the site closer to the village. The third inventory site was located about 2 km from the village of Cajueira. This intact forest was lightly used by people from the small village of Cajueira, but it was becoming a major site for commercial titica collection by people from other villages in the reserve. The second and third inventories were each conducted in 0.6 ha of forest divided into unburned 10x25 m plots located by stratified random sampling of six plots each in four 250 m long belt transects.

The information collected during these surveys was the number of host trees in every plot that had either a titica root or climbing stem. Climbing stems generally represented titica seedlings or juveniles

(Appendix 5-A) while roots represented adult titica plants. The number and type (described below) of roots that were present at 1.5 m above the ground were recorded for each host tree. Since roots generally did not branch out until they got close to the ground, this number approximated the number of roots attached to the 174 main plant stem in the lower canopy of the host tree. All live roots were classified as immature or mature

(Appendix 5-A). Immature roots had soft outer layers and may or may not have entered the ground. All mature roots had entered the ground and had fully hardened outer layers. At the medium-use site (3 km from

Tekohaw) and light-use site (at Cajueira), experienced collectors and I further classified mature roots as commercial or non-commercial. Commercial roots were generally at least 3.5 mm wide and had relatively few nodes visible at ground level. Dead roots were detected by their degraded state and were easily broken when pulled. These were not counted in surveys.

The average number of host trees with titica plants and the average number of titica roots of different types per 250 m2 plot at each site was multiplied by 40 to generate estimates of titica host tree, titica root density and titica climbing stem density per ha. Data on host trees with titica roots were combined from the medium-use (Tekohaw) and light-use (Cajueira) sites to analyze the average number and percentage of commercial roots per host tree.

Comparative Dimensions of Commercial and Non-commercial Mature Roots

The differences in the width and abundance of nodes on mature commercial and non-commercial roots were quantified by harvesting the mature roots from all 73 host trees that had commercial roots in 0.5 ha

(20 plots) at the Cajueira inventory site. Prior to measurements, experienced harvesters classified each root as commercial (n=112 roots), non-commercial due to thinness (n=113 roots), or non-commercial due to too many nodes (n=11 roots). The width of each root was measured at one meter intervals. When the measuring point would have fallen on a node, the measurement was taken 4 cm below it. The number of nodes was counted on every root and divided by the root length to calculate the average number of nodes per meter.

The Relationship Between Titica Plants and their Host Tree Type and Diameter

Host Tree Type and Diversity

The common Tembé or Portuguese name (morphospecies) of the host tree bearing titica plants was recorded in the three inventories described above to gauge the diversity of these hosts. The number of host 175 trees belonging to each morphospecies was divided by the total number of morphospecies at the heavily used site at Tekohaw to calculate the percentage that each morphospecies contributed to the total. These percentages were ranked from most common to least common and compared to a similar abundance ranking of morphospecies of all trees (≥10 cm DBH) that were inventoried in 0.9 ha of the same belt transects (36 plots x

250 m2 per plot). Botanical specimens were not collected from host trees, so this comparison could not be made on the basis of scientific species.

Host Tree Diameter

The DBH (diameter at 1.5 m) of host trees bearing titica plants was recorded in 0.25 ha (10 plots) in the heavily used site at Tekohaw. The DBH size class of host trees ≥10 cm was compared with the size class distribution of all trees ≥10 cm DBH inventoried in 0.2 ha (8 plots) of the same belt transects. Data were analyzed to compare the size class distribution of titica host trees to all trees. The average density of host trees

(≥10 cm DBH) with titica roots was divided by the average density of all trees with ≥10 cm to estimate the percentage of trees in the area that served as hosts for adult titica plants.

The DBH of host trees with at least one mature titica root was recorded in all 0.6 ha (24 plots) surveyed in the medium-use site at Tekohaw. The number of climbing stems, number and type of roots was compared with the host tree DBH to examine the relationship between host tree size and titica plant development.

Rate of Root Maturation and Natural Mortality

The amount of time that roots spend in various phases was investigated by individually marking 56 immature roots (already in the ground) on 31 host trees and 159 mature roots on 39 host trees in the medium- use site at Tekohaw. Each root type was spray painted a different color in December, 1998. The status of these roots was recorded as immature, mature or dead in follow-up observations in March, April and July,

1999 – 76, 134, and 210 days respectively after the initial observations. Results were analyzed to determine the maturation and mortality rate of immature roots and the mortality rate of mature roots. 176

TITICA HARVEST IMPACT, YIELD AND SHORT-TERM ECONOMICS

Impact of Harvesting on Root Mortality and Regrowth

To measure the effect of harvesting on titica roots and test the hypothesis that the amount of roots cut per plant could affect the mortality and regrowth of these roots, mature aerial roots were cut from two treatment groups of host trees at the medium-use site at Tekohaw in December, 1998. In the first treatment, 37 host trees had half (or half plus one) of all commercial quality titica roots present at 4 m above the ground

(accessed by ladder) cut with pruning shears approximately 4 cm below the nearest node. In the second treatment with 56 host trees, all of the commercial roots were cut in the same way. Host trees were assigned to treatment groups so there was approximately an equal average number of roots per tree in both groups before any roots were cut. Nodes of the severed roots were spray painted to facilitate future observations. Roots were cut rather than pulled to facilitate follow-up observations of regrowth.

Host trees were revisited 3 and 7 months later to record the status of these painted cut roots as dead or alive. If living roots had any regrowth within a meter of the point where the root was cut, the number, lengths, and widths of newly sprouted roots were also recorded. The percentage of roots that were dead, alive, and alive with growth per host tree were calculated and compared for both treatment groups. The mortality rate of uncut mature roots observed at ground level (described in the previous section) provided information on the natural rate of mature root mortality. Where regrowth was observed on cut roots, replacement roots ≥3.5 mm wide were classified as commercial quality. Regrowth data at 7 months was multiplied by 1.7 to estimate the regrowth length of all roots and commercial quality roots per year.

Root Transformation During Commercial Preparation

The first stage of preparing commercial quality titica roots for sale was breaking and cutting harvested roots into stems that are free of nodes and other pieces that are split or damaged. I measured the effect that this process had on root length, weight and the number of root stems per host tree by processing the commercial roots harvested from 37 host trees in 0.25 ha (10 plots) at the Cajueira site (half of the host trees measuring dimensions of commercial and non-commercial roots described above). 177

Following the removal of nodes and bad pieces, the outer layers of titica roots need to be stripped off to prepare them for sale. Since roots in eastern Pará are sold by weight, I measured the loss in weight that accompanies each transformation between harvested and saleable roots. I conducted this study by harvesting and processing all the commercial roots from 20 host trees with at least two commercial roots found near the medium-use inventory site at Tekohaw. The roots from each host tree were weighed immediately after harvest, after node and bad piece removal, and after cortex removal. Total length of roots was also measured before and after node removal. The stripped root stems were then stored in the shade and reweighed every day for nine days until there was no further weight loss from evaporation. The average percentage weight loss of roots per host was calculated for each stage of commercial preparation.

Harvester Daily Yield and Time Involved in Collecting and Processing Titica Roots

Data on harvesting yield and rates were gathered by accompanying one man collecting titica vine roots at Cajueira for five days. After pulling down a root, this commercial collector broke roots into pieces that had only one node at the end so root stems could be easily tied in a bundle at the top where the nodes were bunched together. The number of host trees and root stems harvested per host tree were recorded each day.

The number of commercial roots collected per host tree and the GPS position of each host tree were noted for four days. GPS data was analyzed to describe the harvester’s search pattern and to estimate the average distance between host trees and area searched. The time of arrival and departure at each host tree was recorded for the first two days and analyzed to estimate the average search and harvest time per host tree. All observations were made by two researchers so the normal working pace of the harvester was not interrupted.

The amount of time required to process roots was measured for a sample of roots harvested at

Cajueira. Observations included the starting number of stems, and starting and finishing weight of each bundle of root stems. Since titica root processing is often a family endeavor, the age of the person stripping the roots was recorded along with the number for each bundle of root stems. Results were analyzed to determine the average number and grams of root stems processed per minute and the relationship between age of processor and the processing rate. 178

Titica Vine Market Structure and Economics

In November 1998, I interviewed two dealers in the town of Gurupi and the city of Belém who buy large quantities of titica vine roots and other forest products and the managers of four businesses in Belém that make artesenal style furniture with titica roots and other plant materials. These managers were asked to comment on sources for titica and the amount of titica roots they bought or consumed per year and per finished product and amounts of money related to these purchases and sales.

To assess commercial activity in the reserve, I requested a Tembé Indian forest product buyer in

Tekohaw to maintain a log book that recorded the name of the collector and amounts of titica roots he bought from Tembé and Ka’apor Indians in the area. He provided information on transactions that occurred between

July, 1999 and January, 2000.

RESULTS

TITICA BIOLOGY AND POTENTIAL HARVEST OF AERIAL ROOTS

Density of Titica Host Trees and Roots and Number of Roots per Host Tree

Inventories at Tekohaw and Cajueira showed that as the intensity of site use increased, the density of host trees with titica plants and the number of titica roots per ha decreased. The most heavily used forest plots close to Tekohaw had the lowest density of host trees in these three inventories of unburned forest sites with any titica roots (mean 143 ± 22 per ha). Comparing this number to the density of all trees ≥ 10 cm DBH in the same plots indicated that as many as 48% of these trees had a titica plant with at least one root. The lightly used site at Cajueira had the highest titica host tree density (mean 453 ± 32)(Figure 5.3). The density of host trees with commercial quality roots was 67 ± 15 per ha at the medium-use site at Tekohaw (24% of host tree density with any roots) and 163 ± 14 per ha at Cajueira (36% of host tree density with any roots). The effects 179 of the 20-year-old fire were still evident at the Tekohaw site since there was an average of only 2 trees per ha in the burned plots in the heavy-use area that hosted any form of titica plant.

The number of titica roots per ha in the unburned plots at the heavy, medium and light uses sites reflected the same pattern as the host tree density. The lightly used Cajueira site had the highest density with an estimated 1748 ± 142 titica roots per ha while the heavily used site at Tekohaw had the lowest with 544 ±

84 roots per ha (Figure 5.3). The percentage of these roots that was commercial quality at Cajueira was 32% while only 15% of the roots at the medium-use Tekohaw site could be harvested for sale.

The estimated density of climbing stems per hectare was almost identical at the medium-use site at

Tekohaw and light use site at Cajueira and Tekohaw with a combined average of 267 climbing stems per ha.

Evidence of seedlings at the heavy-use site at Tekohaw was very low since there was only an average of 19 climbing stems per ha found there.

Analysis of roots per host tree showed minimal differences between the medium and lightly used sites at Tekohaw and Cajueira so these data were combined to measure the average number and type of titica roots per host tree in the reserve. There was an average of 3.6 ± 0.2 roots in host trees with any type of root with an average 1.0 ± 0.1 commercial quality roots per host tree (Figure 5.4). On host trees with at least one commercial quality root, the average number of roots was 5.8 ± 0.4 with an average of 3.1 ± 0.3 commercial quality roots per host tree.

Comparative Dimensions of Commercial and Non-commercial Mature Roots

The average width of commercial quality roots harvested from the Cajueira inventory site was 5.4 ±

0.1 mm. This was significantly greater than the roots considered non-commercial because they were too thin that averaged 2.5 ± 0.1 mm (One-way AOV: F=586.4; p=0.000) and not different from roots considered unfit for sale because they had too many nodes. There was an average of 1.2 ± 0.1 nodes per meter on commercial quality roots. This was significantly less than the combined average of both categories of roots considered non-commercial that had 2.1 ± 0.1 nodes per meter (One-way AOV: F=50.8; p=0.000). 180

The Relationship Between Titica Plants and their Host Tree Type and Diameter

Host Tree Type and Diversity

The number of morphospecies that served as host trees for titica plants was the least at the heavily used site at Tekohaw (29 morphospecies from 183 trees in 1.2 ha inventoried), intermediate at the heavy-use site at Cajueira (44 morphospecies from 364 trees in 0.6 ha inventoried) and greatest at the medium-use site at

Cajueira (52 morphospecies from 224 trees in 0.6 ha inventoried). The imprecision of combining common tree names into morphospecies groups means that strict numerical comparisons of these numbers are not appropriate. The results in general showed that titica plants do not show any strong preferences for the type of host tree they colonize.

The approximate match between the relative abundance of all trees and the relatively abundance of titica host trees was demonstrated in the comparison of host tree morphospecies at the heavily used Tekohaw site and the inventory of all trees (≥10 cm DBH) in the same transects. These results revealed that five of the six most common host trees for titica (caçador (Lecythis spp.), tiriba (Eschweilera spp.), macucu (Licania spp.), breu (Protium spp.), and faveira (Parkia spp.)) were also among the eight most common trees in the general tree survey. Trees that were in the ten most common types found in the general survey that were almost never recorded as titica host trees were embauba (Cecropia spp.), açai (Euterpe spp.), inga (Inga spp.), lacre (Vismia spp.) and marupá (Simaruba spp.).

Host Tree Diameter

The survey of titica host trees at the heavily used site at Tekohaw showed that 21.2% of host trees with titica roots are ≤10 cm DBH (min. 4.2 cm). Most host trees were in the 10 to 19 cm DBH range (40.4%) with 17.3%, 13.5%, and 7.7% found respectively in the 20-29, 30-39, and ≥40 cm DBH size classes. No data were available on the proportion of all trees in the smallest size class, but a comparison of the percentages in each of the size classes ≥10 cm DBH between titica host trees (n=43 host trees) and all trees (n=85 trees) in the same area was highly correlated (Pearson’s Correlation Coefficient: ρ=.976; p=.024) so in general adult titica plants had no special association with the diameter of their host trees. 181

The titica inventory at the medium-use site at Tekohaw showed that the average number of all titica roots and commercial roots increased with the diameter of host trees with any type of titica root (Figure 5.5).

The mean number of roots on host trees with 4-9 cm DBH was significantly less than the average in all larger size classes (One-way AOV; p ≤ 0.05). The greatest value was found in 30-39 cm DBH host trees (5.3 ± 1.0 roots per host). The association between large host trees and commercial titica roots was even stronger than with roots of all sorts since there was a mean of 2.5 ± 0.7 commercial roots per host that was significantly greater than found in all three of the smaller size classes of trees <30 cm DBH (One-way AOV: p ≤

0.05)(Figure 5.5).

The relationship between host tree DBH and titica plants with climbing stems is the reverse of the one with adult plants with roots. The inventory showed that 84% of the 122 host trees with titica climbing stems were <10 cm DBH. Less than 6% of all host trees with a titica climbing stem were ≥20 cm DBH.

Rate of Root Maturation and Natural Mortality

Immature aerial roots that had penetrated the ground in December 1998, steadily reached maturity during the three follow-up observations of titica roots at Tekohaw (Figure 4.6). Seven months after the initial observations, 86% of the roots that were immature at the start of the observations had developed the hard exterior and strength characteristic of mature feeder roots. About 9% of them died before they reached maturity. Roots that reached maturity had a higher life expectancy since only 2 out of 159 (1.3%) of mature roots died during the seven months of these observations.

TITICA HARVEST IMPACT, YIELD AND SHORT-TERM ECONOMICS

Impact of Harvesting on Root Mortality and Regrowth

Three months after titica roots were cut at the Tekohaw medium-use site, 33 % of cut roots had died on host trees where half the commercial roots were cut; the mortality of cut roots was 45% in host trees where all commercial roots were cut, although this difference was not statistically significant. Seven months after 182 cutting, the mortality of cut roots was close to 63% for both groups (Figure 5.7). It was not possible to observe effects of cutting on the whole plant from the 4 meter elevation. No mortality of unsevered roots was observed in the first treatment group where only half of the commercial quality vines had been cut.

Three months after root cutting, only 2 out of 224 roots had a new root growing. Seven months after cutting, regrowth was observed more often in the 100% root cut treatment (Figure 5.7), but the difference was not significant. Overall regrowth was seen in both treatment groups combined on 23 cut roots (16.9 ± 3.7% of cut roots) from 22% of host trees that had roots cut. When this occurred there was an average of 1.7 new roots per root cut (maximum 4) that usually grew out from the nearest or second to nearest node above the cut.

These new roots sometimes diverged into two roots within centimeters of their origin from the main root.

Judging by their diameter, 59.5 ± 9.4% of these new roots were commercial width. It is not known, however, what percentage of these new roots would ultimately make it to the ground and mature to commercial quality.

If linear growth of new roots produced after old root cutting is averaged over all host trees that had roots cut, the average cut root will regrow an estimated 19.7 ± 5.3 cm per root per year. If only host trees with regrowing cut roots are considered, this same figure is 91.7 ± 17.1 cm per root per year. For host trees as a whole, the estimated amount of regrowth from cut roots is 76.6 ± 22.3 cm per host tree per year when all host trees are considered; it is 356.1 ± 77.0 cm per host tree per year for host trees with at least one cut root that sprouted a new root. For all of these regrowth measurements, the amounts were higher for the treatment where all commercial roots were cut, but these differences were not statistically significant.

Root Transformation During Commercial Preparation

After nodes and unsellable pieces were removed from commercial roots harvested from the Cajueira survey plots, the number of stems produced was about twice the number of roots harvested per host tree

(Figure 5.8). The total length of commercial roots harvested per host tree was 13 meters at the Cajueira survey site; it was 22 meters per host tree from the 20 host trees harvested for the product transformation observations at the Tekohaw sample harvest. After initial processing in the Cajueira survey and Tekohaw harvest sample, these total lengths were reduced by 27 to 30%. 183

The weight of root stems after node and bad-piece removal was 31% less than the starting weight of harvested roots from these two samples. After the cortex was removed from root stems derived from the host trees harvested at Tekohaw, the weight immediately dropped from an average of 420 ± 60g per host tree to 183

± 26 (Figure 5.9), a further weight loss of 56%. In the days that followed, the uncovered stele lost additional weight, presumably due to evaporation of water. After eight days, the average weight of the cleaned root stems per host tree had declined to 119 ± 17 g.. A daily charting of this process showed that the biggest losses occurred in the first five days. After that the percentage weight loss due to drying leveled off at 35.1 ± 1.1%

(Figure 5.10). Comparing the weight of roots harvested from the host trees to that of the saleable root stems that have had their nodes and cortex removed and dried, only 19.4% of the original weight remained. Since harvesters typically remove some nodes and bad-pieces immediately after roots are harvested, the ratio of root weight brought out of the forest to the ultimate weight of roots sold is probably about 30%.

Harvester Daily Yield and Time Involved in Collecting and Processing Titica Roots

Accompanying a harvester collecting titica roots for sale at Cajueira provided a variety of measures of harvesting yields, rates and effort. During five consecutive days of harvesting, he collected roots from an average of 44.8 ± 4.1 host trees per day (n=5 days). This led to a daily average of 192.0 ± 12.4 roots collected per day on the final four days (harvested roots were not counted on the first day). Roots gathered during the entire five-day period had some nodes removed in the forest and were broken into 2435 root stems that weighed 167.5 kg producing daily averages of 487 ± 19 stems per day and 33.5 ± 1.4 kg of root stems per day .

The average time he spent between two harvested trees was 1.2 ± 0.2 minutes per tree (n=78 host trees in the first two days). Using the straight line distance between two harvested trees (calculated by the difference in GPS readings), this means that 39.1 ± 4.4 meters were searched per minute (n=50 host trees), an estimated search rate of 26.1 ± 4.9 min. to search one ha of forest if the search path was 10 meters wide. The average time spent at a tree was 4.9 ± 0.4 minutes per tree (n=84 host trees in the first two days). This time included cutting roots to be harvested at the base, pulling the descending roots from the tree, breaking them into stems and tying them into a bundle. As harvesting progressed, multiple bundles were tied together into a

“cabeça” – head. Finally several “cabeças” were united in a large bundle called a “feixa” that weighed 15 to 184

20 kg. Depending on the search pattern, a completed “feixa” was sometimes left by a trail for a time and picked up later so the harvester would not have to carry this load from tree to tree. It was then usually picked up on the way back to the village. This harvester’s usual goal was to collect two “feixas” per day. Older men usually had a more modest goal of one small “feixa” (10 to 15 kg) per day while strong men intent on making as much money as possible said they sometimes harvested three to four “feixas” per day if they were in an area with lots of titica.

The average distance between successive host trees was 59.8 ± 3.8 meters (134 host trees in 4 days).

A map of the host tree positions visited for four days shows that the harvester searched in one general direction throughout the day (Figure 5.11), but when he found an area that seemed to have many host trees with titica in close proximity, he wandered around that area for awhile before returning to the general search pattern direction. An examination of the order that host trees were visited on the first day (Figure 5.12) shows that his movements included frequent backtracking to look for trees with large numbers of titica roots. If the harvester’s average search path was 10 meters wide, the density of host trees with commercial quality titica roots he was collecting from was 29.2 ± 2.3 host trees per ha. The density of commercial roots he was harvesting was effectively 122.3 ± 15.2 roots per ha. These are both substantially less than the densities of commercial roots and host trees with commercial roots estimated from the survey of nearby transects at

Cajueira and the Tekohaw medium-use site.

Roots collected by the harvester at Cajueira were brought back to Tekohaw where his family all participated in cutting off the top node from each root stem and stripping off the cortex. Processing the product of the father’s five days of solo collecting activity took about two and a half days for the family with two to seven members working at any point in time. The average rate for processing these root stems was 0.96

± .04 stems per minute (roots from 147 host trees). By initial weight, stems were stripped at the average rate of

64.8 ± 2.8 g per minute. The rate of root processing was significantly correlated with the age of the person stripping off the cortex (Pearson’s Correlation Coefficient: ρ=.192; p=0.02). Children that were 10 years and under were given smaller root bundles to process, and they averaged 0.35 to 0.92 stems per minute at this task while people who were 13 years and over all stripped just over one stem per minute (Figure 5.13) although these differences were only statistically significant between groups of children 8 years and under and people 185 who were 10 years and older. If one skilled adult had processed all these roots at the rate of 1 stem per minute, it would have taken them just over 40 hours – about eight hours per day for five days.

Titica Vine Market Structure and Economics

Between 1997 and 1999, the prices that harvesters were paid for cleaned titica roots varied from $R1 to $R2 per kg. This represented an equivalent amount of $US in late 1997 and gradually declined to prices that were 50 – 60% of these values. Before 1998, titica roots collected in the Tembé reserve were periodically sold to a small dealer in a river town or the large city of Belém. The advantage of selling to the river dealer was that he immediately paid for the vine roots in cash, but he usually paid about 25% less than the dealers in the city. The downside of getting a higher price from these city dealers was that they deferred full payment until they sold them to other customers. These large city dealers typically only bought batches of hundreds of kilograms of cleaned roots at one time so collectors had to wait until a large amount had been gathered before a sale was possible.

From 1998 on, two Tembé men in the village of Tekohaw began buying titica roots and breu resin

(from Protium spp. Chapter 3) from Tembé and Ka’apor Indians in the area. Harvesters received cash or credit toward merchandise that the Tembé middlemen bought when the roots were sold. Between July 1999 and Jan.

2000, the major Tembé buyer in Tekohaw bought 1754 kg of processed titica roots from 19 different collectors. Conversations with Tembé Indians indicated they usually devoted several weeks to commercial harvest of titica roots per year. These efforts usually yielded 10 to 45 kg of roots (with some nodes removed) per day that yielded an estimated 3 to 15 kg of sellable root stems per day of collecting.

The dealers from Gurupi and Belém sell some titica to other dealers outside of Belém, but most consumers are local artesenal style furniture makers. These businesses are all small. In the smallest, the owner along with a few other people do most of the work. The largest one had five regular employees, but up to 60 people would be hired on a temporary basis if the owner secured a large order. All stores had ready-made items for sale, but most of the business seemed to depend on custom orders..

The mainstay of these businesses was making vine furniture, but to varying extents all of the shops also made wooden or metal furniture where vines are either a marginal component or absent. Titica vine was 186 the principal vine used in these enterprises and was the dominant material in many types of chairs and sofas. It is sometimes mixed with other vines from the region and cultivated “vime” vines cultivated in southern Brazil.

A small to medium chair consumed three or four kg of titica with a large one using up five to eight kg. A large basket can consume one kg of titica while leftovers from larger pieces are used to make small baskets. Most of these items were made only from titica root stems only and are usually sold in their natural color. These stems were also used to wrap joints in furniture made with other principal materials such as other vines or metal.

Titica can be split and soaked in caustic soda to make it extra pliable. It was also sometimes painted or stained..

Most of the furniture makers said that they bought titica from three or four regular suppliers who came by with the material in hand since deals were rarely brokered in advance. These suppliers came from both Pará and the adjoining state of Amapá. The smallest shop used 50 to 100 kg of titica per month; the larger ones used 100 to 200 kg per month. A few bought only a month’s supply at a time, while others laid in a stock of 500 to 1000 kg. The prices paid for cleaned titica roots ranged from $US 1 to 4 per kg; $US 2 per kg was a rough average. The larger scale dealers paid lower prices for buying larger quantities at one time.

Wicker furniture making is a labor-intensive enterprise. A large chair generally took one experienced person one and half to two days complete; a complicated piece could take three days. Since workers were paid about two minimum salaries ($US 12 per day), the cost inputs for a large chair were about $US 10 for material and $US 24 for labor for a total of $US 34. A regular large chair usually sold for $US 100 to 180. A very fancy one could fetch from $US 500 to 1000, but such sales are rare. These dealers sold all or most of their products to customers in Belém. Larger dealers did some business with buyers in São Paulo or outside Brazil

(U.S., Japan, France, and Belgium). One of the two larger dealers said that business was not as good as it had been in the past. He used to have 20 - 25 regular employees and had reduced his staff to five. It took these people several years to learn their craft well so he was not training any more.

In sum, the market for titica roots in Belém was fairly strong, but like many commodities the price paid to harvesters fluctuated by 100% within several months. It seemed that the maximum price, however, would never exceed the price for a readily available cultivated fiber from southern Brazil. 187

DISCUSSION

TITICA BIOLOGY AND POTENTIAL HARVEST OF AERIAL ROOTS

Density of Titica Host Trees and Roots and Number of Roots per Host Tree

Estimates of the density of trees with adult titica plants at Cajueira and Tekohaw varied from 143 to

453 host trees per ha. This range is higher than H. flexuosa host tree densities in Guyana that varied from 61 to

232 host trees per ha (based on surveys in 0.1 ha plots)(Hoffman, 1997). This Guyana population colonized an average of 9 to 41% of all trees ≥ 10 cm DBH in various habitat types, lower than the 48% colonization estimate for the heavily used site at Tekohaw. The lower H. flexuosa host tree density in Guyana may be partially due to the large abundance of greenheart trees (Chlorocardium rodiei) that were rarely colonized by the vine (Hoffman, 1997). Titica densities from both the Tembé reserve and Guyana study were dramatically larger than the 1.4 to 5.3 titica host tree per ha figure reported by Durigan (1998) for the Jaú Park in the central

Brazilian Amazon.

From a harvester’s point of view, the density of roots per hectare matters more than the density of host trees alone. Given the superior densities of titica commercial stems at Cajueira compared to Tekohaw, it is not surprising that Tembé men from Tekohaw were willing to spend a whole day traveling by boat to Cajueira when they wished to gather titica vine roots for sale. While both Tembé sites had a higher density of host trees with titica than in Guyana, the Guyana sites had an estimated 997 to 1175 mature roots per ha that matched figures for the Tembé sites because titica plants in Guyana had a higher average number of mature roots per tree (Hoffman, 1997).

The other noteworthy aspect of the density measurements was the near absence of titica host trees in forest areas near Tekohaw that were burned in the early 1980s. The causes for this probably relate to past direct effects of fire, the physiology of titica and the characteristics of the trees most commonly found in the secondary forest of these areas. It is apparent from the relationship between titica root number and host tree 188 size and Hoffman’s (1997) observations of H. flexuosa’s slow growth rate that it takes many decades for a titica plant to progress from the juvenile stage of climbing a host tree to the time that aerial roots descend from a main stem high up on the trunk and mature in the ground. A fire that happened 20 years ago that was severe enough to kill most rainforest trees would have also killed their attached vines so the lack of titica plants with numerous roots in these burn zones is easily understood. While numerous healthy titica plants are close enough to recolonize the recovering forest, pioneer trees such as Cecropia, Vismia, and Simaruba that are most common in this area’s secondary forest almost never had titica plants on them. This may be due to specific vine avoidance traits of these trees or their mere location in light rich environments that are beyond the temperature and humidity tolerances of titica plants. López-Portillo et al. (2000) have noted that aroid hemiepiphytes die quickly when suddenly exposed to high light environments possibly because their hydraulic architecture is not adapted to this condition. Whatever specific reasons lie behind the low density of titica plants in secondary forests, the progressive degradation of Amazonian forests by logging, agricultural clearing and fire has stark implications for titica populations. When a forest is degraded, titica density will decline along with its role as an economic resource for a long time.

While a few host trees with titica plants at Cajueira and Tekohaw had as many as 15 absorbing roots on them, the mean number of three mature roots per host tree in the Tembé reserve is considerably less than the

7.7 roots per host tree (calculated from data in Durigan, 1998) in the Jaú National Park in Amazonas state,

Brazil and the 6.6 to 10.5 roots per host tree averages found by Hoffman (1997) in various forest types in

Guyana. The difference between the eastern and central Brazilian Amazon sites may be due in part to

Durigan’s (1988) selection of host trees by the nearest-neighbor method (Durigan, 1998) rather than censusing all titica host trees in a plot since this method was intentionally designed to mimic a harvester selection process. Accompanying a harvester collecting titica roots in Cajueira showed a mean 6.9 roots per host tree average in an area of forest immediately adjacent to the survey plots with less than half that number. It was apparent that this harvester intentionally avoided spending time collecting roots from host trees that offered few or no commercial quality roots. The comparison between his root per host tree average and the figure obtained from the nearby inventory shows that caution is needed in estimating the actual abundance of a resource based solely on data derived from commercial collection. The large difference in the titica roots per 189 host tree average found in the inventories conducted in Guyana and the Tembé sites in Brazil indicate that differences in forest composition, structure, or climate can affect the productivity of titica plants.

Comparative Dimensions of Commercial and Non-commercial Mature Roots

The average length of individual roots harvested at the Cajueira site (4.7 meters) was considerably shorter than the 10.1 meter average found by Hoffman (1997) in Guyana. This variation is probably due more to different harvesting practices in Guyana and the Tembé site than to biological factors. At the Tembé sites, most or all mature titica roots that are considered thick enough to sell are harvested from host trees except when they clearly have a large number of nodes on them. Most of these are trunk roots (Figure 5.12; Appendix

5-A) that wrap around the host tree trunk while descending and become tangled with other titica roots or stems of other locally common climbing vines such as “tarakuwa” (tentatively Philodendron grandiflorum (Jacq.)

Schott (Balée, 1994) and timbo-açu (tentatively Evodianthus spp.(Balée, 1994; Troy and Harte, 1998).

In Guyana, harvesters generally only harvest drop roots that grow straight down from a lateral branch

(Hoffman, 1997). These roots consequently have fewer nodes than trunk roots, and yield longer roots when pulled and longer sections of node free sections. This marketing preference for longer stems presumably increases the number of host trees that are harvested in Guyana since Hoffman (1997) found that the number of trunk roots was often ten times greater than the number of commercial (“drop”) roots per host tree and 73% of all host trees had only trunk roots. The reason there may be relatively few drop roots is that an absorbing root held on a branch far from the main stem would increase the risk of breaking off the branch whose main purpose is to hold leaves out to gather sunlight. As Ray (1986) points out, the branch already has to make a trade off between the length and width of a branch and the weight of leaves it can support. Accordingly holding an aerial root farther from the trunk would force a reduction in leaf mass with no apparent gain. 190

The Relationship Between Titica Plants and their Host Tree Type and Diameter

Host Tree Type and Diversity

The wide diversity of host trees for titica plants at both Cajueira and Tekohaw indicate that this vine does not preferentially select certain taxa of trees to serve as climbing substrates. Hoffman (1997) had the same result in Guyana where more than 42 identified tree taxa served as hosts for H. flexuosa. It was the apparent lack of colonization on several common trees that was more noteworthy in the titica survey of the heavily used site at Tekohaw.

Kelly (1985) found that most host trees for a variety of vascular epiphytes in Jamaican tropical forest had single straight trunks for most of their height, somewhat narrow crowns and thin, smooth or narrowly fissured bark. Other features that may discourage climbing vines on pioneer and other trees are rapid growth, flexible trunks, large leaves, exfoliating and chemical emitting bark, and the presence of protective insects

(Putz, 1984a, 1984b; Hegarty, 1991; Talley et al., 1996). Three of the common species in the heavily used and partially burned site at Tekohaw that apparently lacked titica vines were pioneer trees in the Cecropia, Vismia, and Simaruba genera. Putz (1984a,b) and Janzen (1969) have both noted that Cecropia is often associated with

Azteca ants that vigorously defend their host trees from any plants that contact their host trees. The lack of

Euterpe palms serving as titica hosts at Tekohaw paralleled the apparent absence of palms as a substrate for H. flexuosa in Guyana (Hoffman, 1997). Other authors indicated that some palms are relatively free of vines due to their large long compound leaves that are shed as units (Putz, 1984a; Hegarty, 1991). The tendency for

Euterpe palms to be located near small streams in terra firme forests may have also contributed to their unsuitability as a titica host tree since Heteropsis seems to prefer well-drained soil habitats (Hoffman, 1997).

Host Tree Diameter

Results of the surveys in Tekohaw indicated that for trees ≥10 cm DBH there was no special relationship between tree DBH and the probability that it was colonized by a titica plant. This is what would be expected for a vine that ascends its host with adventitious roots (Putz, 1984b; Hegarty, 1991). The finding that the number of titica roots increases with host tree size reinforces the finding of Durigan (1999) who found 191 a significant positive correlation between host tree DBH and the number of mature roots on it. This relationship may not hold true for the largest trees since the number of titica roots per host tree peaked in the

30 - 39 cm DBH size class and slightly declined (although not significantly so) in the host trees ≥40 cm DBH.

Hoffman (1997) noted that very few emergent trees in the Guyana rainforest had H. flexuosa vines in them.

While titica plants do not apparently have positive preferences for host tree species, they do show strong preference for locating their main stem and branches in the upper bole and lowest branching parts of their host trees (Hoffman, 1997). Vertical specialization and stratification of epiphytes in their host trees has been observed in other neotropical forests and apparently relates to each species’ preferred conditions for light and moisture (Putz, 1984b; Kelly, 1985; Todzia, 1986; ter Steege and Cornelissen, 1989). The location of titica’s main stem and foliage below the host tree’s upper canopy provides it with intermediate levels of light and moisture (Hoffman, 1997). It seems possible that if host trees become emergent, titica plants could arrive in an environment that exceeds their tolerance for direct sunlight.

The Tekohaw inventory finding that 84% of titica plants with climbing stems (presumably seedlings or juveniles) were found on trees ≤10 cm DBH shows that only examining host trees with adult titica plants may omit important observations about this vine’s establishment phase. Putz (1984b) found that certain twining vines can only climb host trees ≤10 cm DBH, but it seems this structural limitation would not apply to titica seedlings because they use anchor roots to adhere to host trees. Additional studies are, therefore, needed to further examine if Heteropsis climbing stems seek out or have better survival on small host trees and what might explain this facet of their natural history.

Rate of Root Maturation and Natural Mortality

As was done at Tekohaw, Hoffman (1997) tracked the development of immature titica roots. Six months after initial observations, he found that 52% of these roots had matured, 33% remained the same and

9% had died. This mortality rate was identical to that found at Tekohaw. The maturation rate was less than the

86% figure found at Tekohaw after seven months, but both studies indicate that immature roots probably take six to nine months to mature once they have taken root in the ground. This maturation process observable above ground is accompanied below ground by the extension and ramification of the root away from the host 192 tree just below the surface (unpublished data from Fazenda 7 observations; Wilder, 1992). While 9% or slightly more of these immature roots may die during this stage, the low mortality rate of fully established mature roots found at Tekohaw showed that they live for many years.

TITICA ROOT HARVEST YIELD AND SHORT-TERM ECONOMICS

Impact of Harvesting on Root Mortality and Regrowth

Documenting the effect of conventional harvesting on the affected roots is a logistically challenging proposition because there is no way to predict where a root will break when pulled from below and they strongly resist being broken by bending. Hoffman (1997) determined that some roots break off near their attachment point, but many others did not. Thus to simulate the effect of root breakage, he cut about 150 commercial roots in 61 host tree canopies. Six months after cutting, 17% of the severed roots had died. The percentage of roots that died in this experiment was considerably less than the 63% average mortality found in the cutting experiment at Tekohaw seven months after cutting. While the average number of roots cut per host tree in the Guyana experiment was similar to the average number cut at Tekohaw (2.5 in Guyana vs. 1.8 and

2.9 in the two Tekohaw treatments), the Guyana treatment only cut an average of 30% of the commercial roots per host tree versus 50% and 100% cut at Tekohaw. It is also possible that the Guyana plants suffered less because they were cut in the February in the middle of the rainy season while the titica roots were cut at

Tekohaw in December when the rainy season was just beginning.

The Tekohaw cutting experiment did not make observations on the effect of harvesting or root cutting on the whole titica plant, but Hoffman (1997) noted that when 100% of the roots were cut the whole plant died.

The plant showed signed of serious stress including branch dieback and yellow leaves when just over 80% of roots were cut, sometimes showed a stressed response when 50-80% of roots were cut, and rarely showed signs of stress if less than 50% of roots were cut. Durigan (1998) did follow-up observations of titica plants following conventional root harvesting and found that 15% of the titica plants died within two months if all commercial roots were pulled down. One year after harvesting, the whole plant mortality rate had risen to an average of 46% (range 17 to 83 % at four sites). 193

In the early phase of their establishment on a host tree, Heteropsis and other closely related secondary epiphytes secure a supply of water and nutrients from the ground through their original climbing stem. There may be an intermediate phase when this connection with the ground has deteriorated and the stem is only receiving modest nourishment via the anchoring roots. In the closely related Monstera genus, the size and shape of leaves changes with progressive life stages in ways that suggest responses to shifting light conditions and transpiration demands (Ray, 1990, 1992). Heteropsis maintains a monopodial growth pattern throughout its development (Hoffman, 1997) and does not go through dramatic heteroblastic leaf development as

Monstera does, but its leaves can go through minor changes in shape as it matures (Croat, 1988). Finally when absorbing roots succeed in becoming established in the ground, the plant’s access to a more abundant source of water and nutrients in the soil is firmly renewed. Its leaves can grow larger with the expectation that their transpiration potential will not be overwhelmed by warmer and drier conditions they are exposed to higher on the trunk.

The root cutting experiments done with Heteropsis in Guyana and Tekohaw and similar tests conducted with a Monstera vine (Simmonds, 1950) show that a sudden severing of these conduits can lead to stress or death of the plant. An anatomical study of the roots of three species of Cyclanthaceae that have similar life histories to some Araceae vines showed that absorbing roots often have ten times as many treacheary and sieve elements and nonperiferal xylem and phloem fascicles (bundles) as anchor roots (Wilder and Johansen, 1992). Absorbing roots may be especially important to secondary hemiepiphytic aroids since they deliver water and solutes from the ground to the main stem (Wilder, 1992) and may help the plant store water and reverse embolisms in the xylem (López-Portillo et al., 2000). The sudden loss of the thickest absorbing roots would, therefore, leave any remaining thin absorbing roots and anchor roots ill-equipped to fulfill these critical functions for the plant by themselves.

The most obvious potential impact of conventional root harvesting from titica plants would be that the large amount of force applied by a harvester to break thick aerial roots would dislodge the main stem of the plant before the root gave way. In practice this rarely happens. I only saw a harvester pull down the main stem of a titica plant one time in the course of witnessing hundreds of plants being harvested. Likewise 194

Hoffman (1997) and Durigan (1998) only saw main stems pulled loose by harvesters a few times during their repeated observations of titica root collection.

Hoffman’s (1997) cutting experiment with H. flexuosa in Guyana (1997) found that 33% of cut roots had some regrowth six months later while 44% were still alive but had no regrowth. This percentage of regrowth was twice as great as found in the Tekohaw cutting study where an average of 17% cut roots had some new roots growing near the cut portion. The average number of new roots sprouted per cut root with regrowth in the Guyana study (Hoffman, 1997) was 1.7 – identical to the finding at Tekohaw, but these new roots were generally smaller than the root that had been cut. My finding that about 60% of the new roots were still commercial size seven months after being cut conforms well with studies of Cyclanthaceae vines that showed only one replacement root (Figure 5.2; Appendix 5-A) arises from an aborted apice (forming a node) or above the excised point of a parental root. This replacement root is generally a little narrower than the parental root, but it is thicker than other non-replacement first and higher-order roots that may also emerge from parental roots (Wilder, 1989, 1992).

Hoffman (1997) measured the length of new roots that grew from H. flexuosa roots six months after they were cut and found that each cut root with regrowth produced an average of 20 cm of new roots per month or 240 cm per year if this rate held steady. Since an average of 2.5 roots were cut per tree and 33% of those had regrowth, the average amount of titica root regrowth per host tree would be about 198 cm per year. This was again a more robust response than found at Tekohaw where the estimated growth of all new titica roots on all host trees where roots were cut was about 77 cm per host tree per year. Only 60% of this growth (46 cm per year) would contribute to the production of new commercial quality roots. Hoffman (1997) found that immature absorbing roots growing down under natural circumstances grew 26 cm per month – 30% faster than the new roots emerging from cut roots so longer term studies are needed to properly estimate the regrowth of roots from harvested titica plants.

One tempting conclusion of this comparison between regrowth rates of cut titica roots in Guyana and

Tekohaw is that cutting a smaller percentage of roots in the Guyana experiment left those plants with more resources to survive the cutting and regrow more often and faster. The fact that the Guyana plants also had a higher number of roots per host tree to begin with indicates that genetic and/or environmental differences may 195 be just as responsible for their greater resilience to root cutting as much as differences in the cutting treatments.

Harvesting impacts need to be further explored with a wider range of treatment conditions at the same site or multiple sites to better address this issue.

Root Transformation During Commercial Preparation

The Tembé practice of removing the cortex from titica root stems before sale is required by almost all regional dealers. This is distinct from the practice in Guyana where harvesters generally sell bundles of H. flexuosa with the cortex intact. Removing the cortex is done to give the product more aesthetic appeal and flexibility in making furniture and handicrafts. Doing so, however, leads to a significant transformation in the weight of the roots collected in the forest (Figure 5.9) and may leave the cleaned roots prone to mildew

(Hoffman, 1997). When used as a lashing material in house construction, Tembé leave the titica root cortex on because removing it takes considerable time and it has greater durability in that form when exposed to the elements.

Harvester Daily Yield and Time Involved in Collecting and Processing Titica Roots

In examining the data on search times and distance between host trees it should be remembered that the harvester spotted and sometimes inspected other trees with titica roots that were not harvested.

Consequently the search time between host trees with any titica plant would be less than the value reported.

The actual distance between all host trees would also be less than reported, but the distance walked between the host trees harvested would be greater than this value since the search pattern between trees was often circuitous. With their well-honed search images for titica root, some Tembé harvesters can no doubt spot a vine laden tree from a considerable distance, but the dense forest mandates that most host trees that are harvested are spotted from close proximity. If 10 meters is a reasonable estimate of the harvester’s reliable search path width, his perception of the forest is that it has only 29 titica host trees and 122 roots per ha.

Assuming that the total titica host tree and root densities in the forest area being searched by the harvester and the adjacent survey plots is similar, the lower densities estimated from harvester data are at least partially explained by his greater selectivity of host trees and roots. It may also mean that the harvester is not locating 196 many titica plants that are hidden on the far side of host trees or otherwise obscured from view. While harvesting most of the commercial quality (i.e. thick) absorbing roots from selected host trees may have negative consequences for the harvested titica plants (Hoffman, 1997; Durigan, 1998), the probability that harvesters do not systematically harvest the most robust roots from all titica plants in an area means that a sizeable number are left unharvested – at least until another harvester searches the area. Accompanying the commercial harvester at Cajueira showed, however, that he quickly moved through an area that had been recently harvested (as evidenced by cut and broken roots around a host tree) until he found another area that seemed to have a high concentration of unharvested plants.

The involvement of family members in removing the nodes and cortex from titica roots is a practice that has economic and social importance for the indigenous harvesters. Although young Tembé children are not generally expected to contribute directly to their family’s sustenance, helping out with titica root stripping is one activity where young children can directly help generate income for their family and save adults time to do other activities that they are uniquely capable of doing. Measurements of titica processing rates show that by the time children get to be 10 years old, they have attained almost the same level of root stripping efficiency as their elders.

The social and economic aspects of titica harvesting are quite different in Guyana where the market readily accepts titica roots with the cortex on (Hoffman, 1997). Although dealers there seem willing to pay more for roots that have been stripped, the market (and perhaps social) structure) does not support this.

Titica Vine Market Structure and Economics

While titica harvesting is the most important source of income for many Tembé and other forest peoples in Brazil (Durigan, 1998), the hard work is poorly remunerated. As an example, the harvester whose efforts were closely monitored at Cajueira spent two days traveling there and back from Tekohaw and five days harvesting to bring back about 168 kg of rough root stems. He and his family spent another three days removing, nodes, bad parts and the cortex. Once these cleaned stems had dried for a week, they were left with about 33 kg of roots to sell for about $ 1.00 per kg. This means he made about $33.00 for ten days of his time

(not including his family’s labor), about $US 3.30 on average per day – just under the Brazilian minimum 197 wage at the time. These earnings are lower than are generally obtained by harvesters in Guyana who can make

$US 5 to 10 per day in the forest (not including travel time and expenses) by collecting one or two bundles (an estimated 10 to 20 kg) of prime quality roots and selling them without having to remove the cortex (Hoffman,

1997).

A simple average value of titica harvest by area can be estimated by assuming there are 100 host trees per ha with commercial vines and that each of these trees will yield 0.5 kg of titica roots. This means that each hectare will yield 50 kg of harvestable unprocessed roots – or 10 kg of commercial roots. These will sell for about $ 10. This is a better per hectare value than can be obtained from other NTFPs examined in this study, but it should be remembered that these harvest levels are derived from harvests from forests that have not been harvested for a long time and would not be able to be harvested again at this level for many years to come.

Considering the high mark up between the cost of the raw material and the finished product, it is tempting to contemplate the extra income Tembé or other harvesters could earn if they made and possibly sold titica based furniture themselves since they possess an aptitude for working with this material. There are several challenges that could make this a problematic venture. First, it is much easier to transport large batches of straight vines than bulky furniture on small boats or vehicles. Second, the furniture dealers in Belém only buy vine roots - not finished product. Troy and Harte (1998) provide additional details about the wicker furniture businesses established in Belém, which indicate it might be possible for rural artisans to increase sales of handicrafts made from titica and other vines, but there would be significant marketing challenges. Indians or other harvesters would need to rent their own space in the city which would incur considerable costs and marketing expertise. Since much of the local business is based on custom orders, there would also need to be an effective means to link customers with artisans to discuss customer needs. Current communication systems do not favor this. The one option for expanding the profitability to harvesters in the short-term would be for them to explore markets for the vines in more distant cities such as Fortaleza where furniture making is also apparently prominent. It is possible that the higher prices paid for delivering debarked roots to these cities would justify the extra cost and logistics of bringing them beyond Gurupi or Belém. 198

CONCLUSIONS: CHALLENGES FOR SUSTAINABLE TITICA HARVEST

Like many NTFPs, harvesting of titica vines presents a challenging dilemma. It is a plant that has proven useful to Amazonian forest peoples as a vital material in building structures and making other objects for every day use. The vines are abundant in many primary forests, and minor root harvesting for personal use would not make an appreciable dent in their populations. The increased demand for titica roots as a prime material for making wicker furniture has created a very welcome source of cash income for forest peoples, but this has also placed a much greater pressure on titica resources in the forest. In Guyana, commercial harvest of

Heteropsis is generally limited to drop roots that are only found on 27% of host trees and then only comprise an average of 31% of the mature roots. By contrast Tembé Indians and other Brazilian titica harvesters generally take most thick trunk and drop roots they encounter. The fact that individual harvesters do not find every host tree in an area and spare non-commercially viable roots provides some buffer for titica populations in areas that are lightly harvested. It is evident, however, that commercial harvesting near the largest villages in the reserve is no longer viable, so these activities have moved to smaller villages and more remote areas. A great deal of forest interior will remain unharvested because it is not practical to carry heavy vine bundles more than a few kilometers to the nearest river so the species itself is not endangered. Without some form of proactive management of titica harvesting, though, the most accessible areas will probably be exhausted as a commercial source of vines in relatively few years.

Considering his measurements and analysis of H. flexuosa stem growth, root growth and maturation rates, Hoffman (1997) estimates it takes 20 to 66 years for one titica plant to go from the climbing stem stage to the point when the first absorbing root reaches maturation in the ground. This long time horizon suggests three possible management strategies. Since there are strong indications that current harvesting practices in the eastern Brazilian Amazon have a deleterious effect on titica plants, one option would be to encourage or require harvesters to reduce the number or percentage of roots taken from host trees. In order to frame this recommendation on a scientific basis, longer term research will be needed to establish the rate at which harvested roots grow back and the rate at which new absorbing roots are formed on branches from the main stem. This information could help determine the number and frequency at which absorbing roots could be 199 harvested without jeopardizing the whole plant’s well-being. Hoffman (1997) recommends that even the modest harvesting rate in Guyana should be reduced by half (to about 1.25 drop roots per host tree) to reduce harsh impacts on individual plants and preserve a future harvest opportunity. Implementing a more conservative harvesting practice where access to primary forest is not strictly controlled would be challenging.

If one harvester pulled down less than all available commercial roots, the next person to come along would be inclined to take the rest. It would take understanding and commitment among a group of harvesters in an area of common property (such as an indigenous or extractive reserve) to abide by such limits.

Traditional beliefs about natural history of the forest can often provide the basis for sound management of resources without the need to interject an outside scientific interpretation. In the case of titica, however, some of these beliefs may make adoption of stricter harvesting norms more difficult. Tembé men who have lived in the forest their whole lives stated that the slender stems of titica that had only ascended several meters up a host tree were a different type of titica than the plants whose roots they harvested. Given their secondary hemiepiphytic habit and the possibility that these original climbing stems disintegrate before aerial roots of an adult plant descend, it is understandable why the two life stages would not be connected.

Another belief of some Tembé and other Amazonian indigenous groups is that the aerial roots of titica and other common secondary hemiepiphytic vines are generated from the extending legs of different species of dead ants (Lenko and Papavero, 1979). These stories principally come from rich mythology concerning the painful stinging tucangueira ant (Paraponera clavata). The link between these and other ants and the descending roots of some vines may derive from the observation that after ants infected and killed by

Cordyceps and Hirsutella fungal parasites exhibit stringy mycelia that sprout from the ant’s corpse and resemble young brown roots (Nelson and Papavero, 1979; Luis Diego Gomez, personal communication 2000).

Tembé understand the biological significance of fruits and are intimately familiar with ones that serve as sources of food for people and terrestrial game animals. The reproductive structures of titica, however, are generally not familiar even to men who frequently harvest the roots. This may be because the fruits are usually out of sight and are likely dispersed in the canopy by animals such as bats and primates (Madison, 1979b;

Vieira and Izar, 1999). It may be arrogant to challenge traditional beliefs, but it seems that harvesters who appreciate the long time span and stages involved in vegetative growth and reproduction of titica plants would 200 be more open to developing alternate harvesting practices that would preserve an important economic and subsistence resource.

The second management option would be to treat titica harvests more like a logging operation where the expectation is that most harvested plants will die and future harvests will depend on waiting for the target species to regenerate to harvestable size. If this system was adopted, harvesting zones could be established where roots could be collected on a rotating basis – possibly with zone limits established after pre-harvest inventories. The area would remain closed to further titica root collection until periodic monitoring indicated that the population and commercial root density had sufficiently recovered. Some of the same challenges described for management option one would also exist for this system, but it might be easier to enforce within a community than a tree by tree harvesting standard. Enforcing protection of forest areas from outsiders is another problem that has already reduced the titica resource base in the Tembé reserve. As forest clearing progresses in the eastern Brazilian Amazon, the invasion of indigenous and other protected areas by settlers and loggers has become increasingly acute. The Brazilian Natural Resource Agency (IBAMA) has fined non-

Indians for harvesting titica vines (and timber) in the Alto Rio Guamá Reserve, but such enforcement actions are rare in relation to the extent of such illegal harvests (Potiguara, personal communication 1998).

The third option for managing titica could operate in situations where logging or forest clearance was a certainty. If it was known that a significant amount of a forest canopy was going to be opened by some activity, it seems likely that titica plants would not survive well in such open environments. In the case of light individual tree removal, additional research could show what disturbance thresholds impact titica plants on remaining trees. Where major canopy removal was anticipated, it would make sense to fully harvest roots of titica plants in these areas before the area was logged or cleared. Fruits of reproductive titica plants could be collected when possible and used for germination tests. It would be logical to test planting titica in both intact forest and advanced secondary forests where non-pioneer species had become established and would be allowed to grow for the foreseeable future. If Hoffman’s (1997) lower-end estimate of titica maturation rates proves to be correct, integrating titica vines and root harvest into a long-term timber rotation (50 years or longer) might be ecologically feasible. It’s hard to say at this point if the demand for titica will get to the point where developing more sophisticated silvicultural approaches would be economically viable. 201

TABLE 5.1 SUMMARY OF TITICA STUDY OBJECTIVES, METHODS AND ANALYSIS

TITICA BIOLOGY AND POTENTIAL HARVEST OF AERIAL ROOTS OBJECTIVE SITE AND SAMPLE MEASUREMENTS ANALYSIS RESULTS SIZE PRESENTED Density of Titica Host Tekohaw Heavy-use 1) Number of host trees 1) Multiply average Figure 5.3 and text Trees and Roots Site (1.2 ha unburned; with a titica root per number of host trees 0.5 ha burned) 250m2 plot per plot by 40 to estimate density per ha Tekohaw Medium-use 2) Number of roots Site (0.6 ha) (immature, mature, 2) Multiply average commercial) per host tree sum of roots per plot by Cajueira Light-use Site 40 to estimate density (0.6 ha) per ha Number and Type of Tekohaw Medium-use Number of roots Compute combined Figure 5.4 and text Titica Roots per Host Site (171 host trees) (immature, mature, mean number of roots Tree commercial) per host tree of different types per Cajueira Light-use Site host tree (306 host trees) Dimensions of Cajueira Light-use Site Length, width and number Compute and compare text Different Root Types (73 host trees) of nodes per root average dimensions for commercial and non- commercial roots Titica Relation to Host Tekohaw Heavy-use Host tree common name; Compare relative text Tree Type Site (titica survey 1.7 Common names of all abundance of titica host ha; all tree survey 0.9 trees ≥10 cm DBH trees with all trees in ha) the area Tekohaw Medium & Common names of titica Estimate number of text Cajueira Light-use host trees in density host tree Sites inventories morphospecies per site Relation of Tree DBH Tekohaw Heavy-use 1) Titica host tree DBH Compare % of host text to Titica Colonization Site (titica survey 0.3 2) All tree DBH trees in each DBH size ha; all tree survey 0.2 class to % in all tree ha) survey Relation of Host Tree Tekohaw Medium-use 1) Titica host tree DBH Compute mean number Figure 5.5 and text DBH to Titica Site (143 host trees) 2) Number of climbing of total roots and Climbing Stem and stems and roots per host commercial roots per Root Number tree host tree Root Maturation and Tekohaw Medium-use Track color coded Calculate % of Figure 5.6 and text Natural Mortality Site (Immature Roots immature and mature immature roots that die on 31 host trees; roots for 7 months and or mature and % of Mature Roots on 39 note transitions to mature roots that die host trees) maturity or death 202

TABLE 5.1 (Continued)

TITICA HARVEST IMPACT, YIELD AND SHORT-TERM ECONOMICS OBJECTIVE SITE AND SAMPLE MEASUREMENTS ANALYSIS RESULTS SIZE PRESENTED Effect of Two Cutting Tekohaw Medium-use 1) Monitor condition of 1) Compare % cut roots Figure 5.7 and text Treatments on Root Site (37 host trees with roots (dead, live, live with dead, alive, and alive Survival and Regrowth 50% commercial roots regrowth) 7 months after with regrowth between cut; 56 host trees with cutting. treatments. 100% commercial 2) Measure number and 2) Estimate regrowth roots cut)) width of new roots arising rate of commercial from cut roots. quality roots. Root Length and Cajueira Light-use Site 1) Number and length of Calculate average % Figure 5.8 and text Weight Loss in (37 host trees) roots per host tree before loss in length and Commercial Processing node removal increase in root stem 2) Number and length of number that root stems per host tree accompanies initial after node removal commercial processing Tekohaw Medium-use Weight of roots after node Percentage weight loss Figures 5.9 & 5.10 Site (20 host trees) removal, cortex removal after each stage of and text and each day of drying processing Harvester Daily Yield Accompanied 1) Host trees, whole roots 1) Calculate average Figures 5.11 & and Search Pattern Comercial Harvester and root stems collected number of host tree 5.12 and text near Cajueira for 5 per day visited and roots days 2) Host tree GPS position collected per day 2) Calculate average distance between harvested host trees 3) Plot host tree search pattern Root Processing Rate Observed 9 family 1) Starting and stopping 1) Calculate overall Figure 5.13 and members processing time for processing each average of root stems text titica roots at Tekohaw bundle of roots and root weight 2) Initial and final weight processed per minute and stem number of roots 2) Compare root per processed bundle processing rate by age 3) Age of processor of processor Value of Titica Roots to Inventories at 1) Calculate value of text Harvester Tekohaw and Cajueira titica roots per ha 2) Calculate value of titica harvest per day collecting and processing Value of Titica Roots to Interviews with forest 1) Buying and selling Compare value of titica text Forest Product Dealers product dealers and prices for titica roots to collector with and Wicker Furniture furniture makers in 2) Weight of titica roots in dealers and furniture Makers Gurupi and Belém common furniture pieces makers

203

a. Absorbing Trunk Root b. Absorbing Drop Root c. Absorbing Underground Root d. Main Climbing Stem with Anchor Roots e. Lateral Branch with Leaves and Fruit f. Senescing End of Climbing Stem g. Seedling Climbing Stem

FIGURE 5.1 ILLUSTRATION OF TITICA SEEDLING AND ADULT ON HOST TREE WITH PRINCIPAL PLANT PARTS 204

a. Absorbing Root Node b. Absorbing Replacement Root c. Non-replacement Root d. Aborted Absorbing Root Apical Meristem

FIGURE 5.2 ILLUSTRATION OF TITICA ABSORBING ROOT AND PRINCIPAL PARTS 205

Interval Plot: Mean plus 95% Confidence Interval

Tekohaw 1800 Heavy Use Site (1.2 ha Surveyed) TITICA ROOTS 1500 PER HA Tekohaw Medium Use Site 1200 (0.6 ha Surveyed)

Cajueira 900 HOST TREES PER HA Light Use Site DENSITY WITH TITICA ROOTS (0.6 ha Surveyed) 600

300

0

Heavy Medium Light Heavy M edium Light INTENSITY OF PREVIOUS SITE USE

FIGURE 5.3 DENSITY OF TITICA HOST TREES AND TITICA ROOTS IN LIGHTLY TO HEAVILY USED FOREST SITES IN THE ALTO RIO GUAMÁ INDIGENOUS RESERVE 206

7 ROOT TYPE 6 53% Commercial Mature Commercial 5 Mature Non- Commercial 4 27% Commercial Immature

3

2

1 MEAN NUMBER OF NUMBER MEAN PERROOTS TREE HOST

0 HOST TREES HOST TREES WITH AT WITH ANY LEAST ONE TITICA TITICA ROOT COMMERCIAL ROOT (N=477 Host Trees) (N=150 Host Trees)

FIGURE 5.4 NUMBER AND TYPE OF TITICA ROOTS PER HOST TREE AT MEDIUM AND LIGHT-USE FOREST SITES AT TEKOHAW AND CAJUEIRA 207

Interval Plot: Mean plus 95% Confidence Interval

7

6

5 All Roots

4

3

2

1

0 Commercial Roots TITICA ROOTS PER HOST TREE HOST PER ROOTS TITICA 4 - 9 10 - 19 20 - 29 30 - 39 >39 HOST TREE DBH SIZE CLASS (cm) (143 Host Trees with Titica Roots)

FIGURE 5.5 RELATIONSHIP BETWEEN HOST TREE DBH AND THE NUMBER OF TITICA ROOTS PER HOST TREE AT THE TEKOHAW MEDIUM-USE SITE 208

IMMATURE TITICA ROOT DEVELOPMENT PHASE Live Immature Dead Immature New Live Mature 100

75

50

25 PERCENT OF ROOTS IN DEVELOPMENT PHASE 0

Start 76 134 210 DAYS AFTER START OF OBSERVATION N = 56 Immature Titica Roots from 31 Host Trees at Tekohaw

FIGURE 5.6 DEVELOPMENT TIME AND MORTALITY OF IMMATURE TITICA ROOTS AT THE TEKOHAW MEDIUM-USE SITE 209

CUTTING 50% Commercial Roots Cut (37 Host T rees) TREATMENTS 100% Commercial Roots Cut (56 Host Trees)

70 Mean + 95% Confidence Interval

60

50

40

30

20

10

0 MEAN PERCENT RESPONSE OF RESPONSE PERCENT MEAN TREE PER HOST ROOTS TITICA 50% 100% 50% 100% 50% 100% CUT ROOTS CUT ROOTS CUT ROOTS DEAD ALIVE WITH REGROWTH

FIGURE 5.7 CONDITION OF TITICA ROOTS 7 MONTHS AFTER CUTTING AT TEKOHAW MEDIUM-USE SITE 210

ONE HOST TREE AT CAJUEIRA YIELDED AN AVERAGE OF 2.9 COMMERCIAL QUALITY TITICA ROOTS WITH NODES 5 (M ean Total length: 13.6 m)

4 (M ean Total length: 8.9 m)

3

2

1 LENGTH PERLENGTH TITICA ROOT (m) STEM AND ROOT 0 THESE ROOTS LOST A MEAN OF 30% TOTAL LENGTH WHEN NODES WERE REMOVED AND YIELDED 5.8 ROOT STEMS

FIGURE 5.8 LOSS OF TITICA ROOT LENGTH AFTER NODE REMOVAL FROM 37 HOST TREES HARVESTED AT CAJUEIRA 211

Interval Plot: M ean plus 95% Confidence Interval 740 Commercial Titica Roots Harvested from 20 Host 640 Trees at Tekohaw Medium Use Site 540

440

340

240 MEAN WEIGHT OF WEIGHT MEAN ROOTS PER TREE (g)

140

Roots pulled Nodes & bad Cortex 8 days after from tree pieces removed removed cortex removed STAGE OF ROOT PROCESSING

FIGURE 5.9 WEIGHT LOSS DURING PROCESSING OF TITICA ROOTS 212

Interval Plot: Mean plus 95% Confidence Interval

30

20 PERCENT WEIGHT LOSS PERCENT ROOTS STRIPPED OF 10

1 2 3 4 5 6 7 8 DAYS OF ROOT DRYING WITHOUT CORTEX Commercial Titica Roots Harvested from 20 Host Trees at Tekohaw Medium Use Site

FIGURE 5.10 WEIGHT LOSS OF TITICA ROOTS FROM DRYING AFTER CORTEX REMOVAL 213

4500 Day 1 Day 2 4400 Day 3 4300 Day 4 4200 4100 4000 3900 3800 3700 3600 NORTH AXIS POSITION (meters) POSITION AXIS NORTH 100 200 300 400 500 600 700 800 900 EAST AXIS POSITION (meters)

FIGURE 5.11 LOCATION OF TITICA HOST TREES HARVESTED IN 4 DAYS BY A COMMERCIAL COLLECTOR AT CAJUEIRA 214

4000 32 31 27 28 25 26 29 20 19 24 3900 22 30 23 18

17 21

3800 10 15 5 13 9 12 16 7 14 1 2 8 11 3700 3 6

NORTH AXIS POSITION (meters) POSITION AXIS NORTH 4

400 500 600 700 EAST AXIS POSITION (meters)

FIGURE 5.12 LOCATION AND ORDER OF TITICA HOST TREES HARVESTED BY A COMMERCIAL COLLECTOR AT CAJUEIRA ON ONE DAY 215

Interval Plot: M ean plus 95% Confidence Interval 1.5 Titica Roots Collected by One Harvester at Cajueira and Processed by his Family at Tekohaw

1.0

0.5

ROOT STEMS STRIPPED PER MINUTE ROOT STEMS 4-6 7-8 10 13 25 42-44 n=(2)7 n=(2)12 n=(1)24 n=(2)59 n=(1)7 n=(2)36 AGE OF TITICA PROCESSOR (Years) N=(Number of people)Number of Root Bunches

FIGURE 5.13 TITICA ROOT STEM STRIPPING RATE BY AGE OF PROCESSOR 216

REFERENCES

Balée, William. 1994. Footprints of the Forest: Ka’apor Ethnobotany – the Historical Ecology of Plant Utilization of Plant Utilization by an Amazonian People. Columbia University Press.

Benzing, David H. 1990. Vascular Epiphytes: General Biology and Related Biota. Cambridge University Press, Cambridge.

Berry, Paul. E., Bruce K. Holst, and Kay Yatskievych. 1995. eds. Flora of the Venezuelan Guayana, Vol. 1. Missouri Botanical Garden, St. Louis and Timber Press, Portland.

Boom, Brian M. and Sylvia Moestl. 1990. Ethnobotanical notes of José M. Cruxent from the Franco- Venezuelan expedition to the headwaters of the Orinoco River, 1951-1952. Economic Botany 44:416-419.

Bown, Deni. 2000. Aroids: Plants of the Arum Family. Timber Press, Portland.

Bunting, G.S. 1979. Sinopsis de las Araceae de Venezuela. Revista de la Faculdad de Agronomia 10:139- 290.

Croat, Thomas B. Ecology and life forms of Araceae. Aroideana 11(3):4-55.

Durigan, Carlos C. 1998. Biologia e Extrativismo do Cipó-Titica (Heteropsis spp. – Araceae) – Estudo para Avaliação dos Impactos da Coleta sobre a Vegetação de Terra-Firme no Parque Nacional do Jaú. M.S. Thesis, Instituto Nacional de Pesquisas da Amazônia (INPA) and Universidade do Amazonas (UA).

Gentry, Alwyn H. 1993. A Field Guide to the Families and Genera of Woody Plants of Northwest South America (Columbia, Ecuador, Peru) with Supplementary Notes on Herbaceous Taxa. The University of Chicago Press, Chicago.

Hegarty, Elwyn E. 1991. Vine-host interactions. pp. 357-375 in eds. Francis E. Putz and Harold A. Mooney. The Biology of Vines. Cambridge University Press, Cambridge.

Hoffman, Bruce. 1997. The Biology and Use of Nibbi Heteropsis Flexuosa (Araceae): The Source of an Aerial Root Fiber Product in Guyana. M.S. Thesis, Florida International University, Miami.

Janzen, Daniel. 1969. Allelopathy by myrmecophytes: the ant Azteca as an allelopathic agent of Cecropia. Ecology 50:147-153.

Kelly, Daniel L. 1985. Epiphytes and climbers of a Jamaican rain forest: vertical distribution, life forms and life histories. Journal of Biogeography 12:223-241.

Lenko, Karol and Nelson Papavero. 1979. Insectos no Folclore. Artes e Ciências Humanos, São Paulo.

López-Portillo, J., F.W. Ewers, G. Angeles, and J.B. Fisher. 2000. Hydraulic architecture of Monstera acuminata: evolutionary consequences of the hemiepiphytic growth form. New Phytologist 145:289-299. 217

Madison, Michael. 1979a. Notes on some aroids along the Rio Negro, Brazil. Aroideana 2(3):67-77.

Madison, Michael. 1979b. Protection of developing seeds in Neotropical Araceae. Aroideana 2(2):51-67.

Milliken, William, Robert P. Miller, Sharon R. Pollard, and Elisa V. Wandelli. 1992. The Ethnobotany of the Waimiri Atroari Indians of Brazil. Royal Botanic Gardens, Kew.

Oliveira, Jorge, Samuel S. de Almeida, Raimunda Vilhena-Potyguara, and Luiz C.B. Lobato. 1991. Espécies vegetais produtoras de fibras utilizadas por comunidades Amazônicas. Boletim do Museu Paraense Emílio Goeldi sér. Botânica 7(2):393-428.

Paz y Miño C., Guillermo, Henrick Balslev, and Renato Valencia. 1995. Useful lianas of the Siona-Secoya Indians from Amazonian Ecuador. Economic Botany 49(3):269-275.

Pio Corrêa, M. 1931. Dicionário das Plantas Úteis do Brasil. Vol. 2. Ministério de Agricultura, Rio de Janeiro.

Potiguara, Raimunda C. de V. and Manoel E. do Nascimento. 1994. Contribuição à anatomia dos órgãos vegetativos de Heteropsis jenmanii Oliv. (Araceae). Boletim do Museu Paraense Emílio Goeldi sér. Botânica 10(2):237-247.

Putz, Francis E. 1984a. How trees avoid and shed lianas. Biotropica 16:19-23.

Putz, Francis E. 1984b. The natural history of lianas on Barro Colorado Island, Panama. Ecology 65(6):1713-1724.

Putz, Francis E. and N. Michele Holbrook. 1986. Notes on the natural history of hemiepiphytes. Selbyana 9:61-69.

Ray, Thomas S. 1986. Growth correlations within the segment in the Araceae. American Journal of Botany 73(7):993-1001.

Ray, Thomas S. 1990. Metamorphosis in the Araceae. American Journal of Botany 77(2):1599-1609.

Ray, Thomas S. 1992. Foraging behavior in tropical herbaceous climbers (Araceae). Journal of Ecology 80:189-203.

Rodrigues, Roberto M. 1989. A Flora da Amazônia. CEJUP, Belém.

Simmonds, N.W. 1950. Notes on the biology of the Araceae of Trinidad. Journal of Ecology 38:277-286.

Strong, Donald R., Jr. and Thomas S. Ray, Jr. 1975. Host tree location behavior of a tropical vine (Monstera gigantea) by skototropism. Science 190:804-806.

Talley, Sharon M., William N. Setzer, and Betsy R. Jackes. 1996. Host associations of two adventitious- root-climbing vines in a north Queensland tropical rain forest. Biotropica 28(3):356-366. ter Steege, H. and J.H.C. Cornelissen. 1989. Distribution and ecology of vascular epiphytes in lowland rain forest of Guyana. Biotropica 21(4):331-339.

Todzia, Carol. 1986. Growth habits, host tree species, and density of hemiepiphytes on Barro Colorado Island, Panama. Biotropica 18(1):22-27. 218

Troy, Austin and Alexis Harte. 1998. The Liana Project: Traditional Arts, Conservation, and Economic Development in the Amazon. Rainforest Action Network, San Francisco.

Viera, E.M. and P. Izar. 1999. Interaction between aroids and arboreal mammals in the Brazilian Atlantic rainforest. Plant Ecology 145(1):75-82.

Whitehead, B.W. and R. Godoy. 1991. The extraction of rattan-like lianas in the new world tropics: a possible prototype for sustainable forest management. Agroforestry Systems 16:247-255.

Wilder, George J. 1989. Morphology of Dianthoveus cremnophilus (Cyclanthaceae). Canadian Journal of Botany 67:2450-2464.

Wilder, George J. and Jeffrey Johansen. 1992. Comparative anatomy of absorbing roots and anchoring roots in three species of Cyclanthaceae (Monocotyledoneae). Canadian Journal of Botany 70:2384-2404.

Wilder, George J. 1992. Comparative morphology and anatomy of absorbing roots and anchoring roots in three species of Cyclanthaceae. Canadian Journal of Botany 70:38-48. APPENDIX 5-A GLOSSARY OF TERMS FOR TITICA VINES AND ROOT HARVEST

PLANT GROWTH HABITS CLIMBER/VINE – A plant that is rooted in the ground but needs support for a weak stem (Kelly, 1985). TRUE EPIPHYTE – A plant whose growth is restricted to growth on trees, shrubs, or rocks (2). HEMIEPIPHYTE – A plant whose natural life cycle includes both an epiphytic and terrestrial phase. It excludes woody lianas and vines that germinate terrestrially and maintain lifetime contact with the ground (3, 5). PRIMARY HEMIEPIPHYTE – Plants that start as epiphytes in a tree and later become rooted in the ground (5). SECONDARY HEMIEPIPHYTE – A plant that is hybrid of a climber and an epiphyte because it germinates in the ground, ascends a tree and then loses contact with the ground. Some plants including titica produce adventitious roots that later reenter the soil (2, 5).

GROWTH STAGES SEEDLING – A plant with one apical shoot meristem on a seedling shoot that grows along the forest floor. It reaches a more advanced stage when it finds a host tree and a climbing stem begins to ascend it (4, 6). JUVENILE – A plant whose climbing stem has produced lateral leafy branches without reproductive structures (4). ADULT – A plant with multiple leafy lateral branches capable of producing a terminal inflorescence. Absorbing roots have arisen from the main climbing stem or lateral branches and descended into the ground. Flagellar shoots may also descend from branches and reclimb the same host tree or grow along the ground in search of a new one (4, 6).

CLASSIFICATION OF ROOTS ANCHORING ROOT/CLIMBING ROOT – A short and narrow aerial root whose primary function is to secure a climbing stem, main plant stem or lateral branches to the host tree (6, 7). ABSORBING ROOT/FEEDER ROOT – A long adventitious root that grows from a climbing stem or lateral branch toward the ground. Its main function is to provide water and nutrition to the main stem once it has entered the ground (6, 7). MATURE ROOT – An absorbing root that has entered the ground and presumably has extensive growth in the soil. Its aerial portion has a hardened outer layer (1) IMMATURE ROOT – An absorbing root that is still descending or has not yet grown extensively in the soil. When a finger nail is scraped over the outer layer of its aerial portion, a greenish color is revealed underneath (1). REPLACEMENT ROOT – A root that grows back from an anchoring or absorbing root when the original root apical meristem aborts or is damaged through mechanical injury or herbivory. Replacement roots continue growing in the direction of the parent root and are about the same thickness (4, 6). NONREPLACEMENT ROOT – A root that grows off of any portion of an anchoring or absorbing root. They generally grow in a different direction and are narrower than the parent root (6). ROOT NODE – A bulge in an absorbing root that occurs just above the point of an aborted apical meristem or injury. Replacement roots frequently emerge just above a node (4) ROOT INTERNODE – The portion of an absorbing root between two successive nodes (4). TRUNK ROOT – An absorbing root that descends in close contact with the host tree trunk and other vines (4). DROP ROOT – An absorbing roots that grows straight down from its attachment point on the plant stem (4). COMMERCIAL ROOT – A mature absorbing root that is fit for sale to wicker furniture makers. The criteria varies regionally, but such roots are judged to be relatively thick (≥ 3.5 mm) and have relatively few nodes (1). COMMERCIAL STEM – A section of a mature absorbing root that has been separated from nodes and in some cases stripped of its outer layers (epidermis and cortex). In the eastern Brazilian Amazon, such stems are generally at least 3.0 mm thick and at least 1 meter long although most are considerably longer (1).

References: 1) Author Description; 2) Croat, 1988; 3) Gentry, 1991; 4) Hoffman, 1997; 5) Putz and Holbrook, 1986; 6) Ray, 1992; 7) Wilder, 1992; 8) Wilder and Johansen, 1992

CHAPTER 6

SUMMARY CONCLUSIONS 221

FACTORS THAT AFFECT NON-TIMBER FOREST PRODUCT HARVEST AMOUNTS AND PROFITABILITY

The first and second objectives of this study were to quantify the amount of marketable non-timber forest product (NTFP) that could be obtained per plant and per area of forest and identify key factors that influenced variation in these amounts. The case studies showed that the size of the resource base and available product were most influenced by the density of the plant, plant size, land use practices (e.g. fire), and sometimes plant interactions with insects or microbes. Once products were harvested, the amount of saleable product that an area of forest would yield depended on the conversion efficiency of the raw product to commercial product.

The third objective of the study was to estimate the amount of harvest that would be possible on a repeated basis and the length of this harvest cycle. The case studies and work done with other NTFPs in the study area showed that some plants could be harvested several times a year while other plants might only yield marketable quantities of product once in their life. The amounts varied depending on what material was removed from the plant and how fast it regenerated.

The fourth main study objective was to estimate the amount of time a harvester needed to spend to find, harvest and process a unit of product and how much income they would earn for the time invested. The case studies found that search time was directly related to plant density and the proportion of harvestable plants within the population. Products that were obtained from the outside of trees (resin lumps on tree trunks, vine roots on host trees, seeds on the ground) were collected much more quickly than products that required drilling or scraping trees to procure interior products (latex and oleoresin in heartwood). The time devoted to procuring raw materials was much greater if collectors needed to travel far from their villages to find concentrated populations of harvestable plants. The degree of processing raw products into commercial products varied extensively between products and took as much or more time for some products as the time required to find and harvest the product in the forest. The amount of income that harvesters could earn from each NTFP of course also depended on the demand for the product and its market price. Among the products examined, there was a 10 ten fold difference between the unit price of the least expensive to the most expensive, but the most important sources of revenue for

Tembé NTFP harvesters were from products that were the most abundant – not the most expensive. 222

SIZE OF RESOURCE BASE AND AVAILABLE PRODUCT

The most important factors that influenced the size of the resource base were the density of the plant and the percentage of those plants that offered harvestable product. Among the three case studies in this thesis, titica vines were by far the most abundant since they were found on several hundred host trees per ha, and at least 50 plants per ha offered harvestable aerial roots (Table 6.1). This number was two orders of magnitude greater than copaiba trees whose density was close to 0.5 trees per ha. The density of the case study trees varied with habitat changes in the Tekohaw study area. Titica vines, and to a somewhat lesser extent, copaiba trees were found almost entirely in upland “terra firme” sites. Breu trees were found almost equally in “terra firme” and wetter

“baixo” type forests, but different species were found more commonly in one habitat or the other.

Larger and presumably older individual plants consistently had a harvestable product more often and offered larger amounts per plant among the NTFP case studies. The percentage of the size range that offered harvestable quantities of product from each NFTP, however, varied considerably. Copaiba trees had a fairly narrow size range of trees from which oleoresin was reliably obtained, so only 15% of the trees could be considered productive from a harvester’s point of view. In contrast, the probability of finding resin on breu trees steadily increased with their size, and overall 40% of them offered harvestable product. My studies conducted with trees that produced “amapá” latex showed that Couma guianensis Aublet (called “black amapá” by the

Tembé) was an exception to this pattern. The amount of latex collected from these trees had no correlation to tree size throughout the five week period of their harvest (Appendix 6-A).

A marketable product sold by one name may come from a variety of species. This was most apparent with Burseraceae trees where some types of trees had considerable resin while others had none. Among those species that were commonly harvested, however, there was no significant average difference in resin yield.

Additional research on specific species rather than loosely defined groups will probably reveal more definitive differences. The difficulty of trusting common names to reflect distinct taxa was also apparent with amapá latex.

The Tembé used the term amapá to refer to both Parahancornia amapa (Huber) Ducke and C. guianensis trees in the Apocyanaceae family and used their latex interchangeably as a medicinal tonic. The latex from Brosimum potabile Ducke(family Moraceae) was also frequently sold in Belém and other parts of the Amazon as “amapá”

(van den Berg, 1984). 223

Land use practices and site history also had a strong influence on the size and productive potential of the

NTFP resource base. A fire that passed through the Gurupi section of the reserve in the early 1980s burned extensive swaths of forest within several kilometers of the village. Almost 20 years later, though, titica vines are still almost absent from these areas that are dominated by pioneer tree species. Some copaiba trees that were burned in the fire survived, but their wounds apparently led to an early hollow condition and consequent loss of oleoresin storage capacity. The Tembé are also losing part of their resource base for wild populations of NTFPs as forests are cleared for agriculture and adjacent forest areas are burned by fires escaping from farm plot preparation. As forest burning related to logging, ranching and agriculture increases throughout the Amazon

(Nepstad et al., 1999), there could be a corresponding direct and indirect depletion of the region’s NTFP resource base.

Changing settlement patterns among the Tembé could also affect their access to both commercial and subsistence types of NTFPs. Some villages in the Gurupi area have received additions to their physical infrastructure in the form of running water systems and construction of health clinics and school buildings. More people have been attracted to these villages and they are taking on a more permanent character than in the past

(Potiguara, personal communication 1999). The apparent result of this pattern is that more forest close to the village has been converted to agriculture and people need to go farther to hunt, fish, and collect forest products. It was already noted in the case studies that breu and titica harvesting trips were conducted hours away from

Tekohaw. People also reported that close to the village it was getting harder to find adequate supplies of the palm frond “ubim” (probably Geonoma baculifera Kunth. (Balée, 1994)) used as the primary roof thatch material and long stems of the herbaceous “guarimã” plant (probably Ishnosiphon spp.) used for making baskets. In response to the shortage and relatively short useful life of “ubim” (two to three years), more people were cutting down trees to make wooden shingle roofs.

As people in the village buy more electronic goods and gas stoves, the need for more batteries and fuel increases. This requires more trips to the city necessitating greater gasoline and diesel fuel consumption and more frequent boat engine repairs. All of these things, of course, cost more money that leads to greater pressure on natural resources. Such depletions and shifts in resource use that accompany acculturation have been noted with other indigenous groups (Robinson and Redford, 1991; Baksh, 1995; Stearman, 1995). This issue will no doubt 224 pose a mounting challenge for the Tembé and other people who wish to maintain access to the full range of their traditional wild plant and animal resources while they integrate more aspects of the non-Indian world into their lifestyle. The chances that expanding sales of NTFPs in the raw form will be able to meet the increasing consumer desires of people in this situation seem very slim.

The amount of raw material per plant is the starting point for assessing an NTFP’s productive potential, but the amount of material that can be sold after the raw material is collected is what counts in determining its actual yield per area of forest. Converting raw product to a saleable product can also significantly add to the overall handling time for a particular product. Some exudates, such as copaiba and amapá latex, can be sold in essentially the same form they come out of the tree. Breu resin had a modest conversion loss of 17% in the drying process, while saleable titica roots that were reduced to node-free segments and stripped of their cortex and dried only retained 20% of the original weight (Commercial Product/Raw Product Table 6.1). The most extreme example of processing loss among the products investigated in this study was the preparation of oil from andiroba

(Carapa guianensis Aublet) seeds. In one sample prepared at Tekohaw it took an estimated 14.4 kg of seeds to produce one liter of oil showing a conversion ratio of 3% by weight (Appendix 6-B). Even a commercial manufacturer of andiroba oil in Belém reported he only obtained a 10% yield in his mechanized process (Morães, personal communication 2000).

REGENERATION RATE OF THE RESOURCE AFTER HARVEST

The sale of NTFPs may generate less revenue per hectare of forest than other land uses such as logging in the short-term but be more profitable in the long-term because the harvest cycle is shorter (Peters et al., 1989).

Fruits can be potentially collected every year with large harvests available on a more irregular basis from trees that have a mast fruiting pattern (one year of peak fruit production out of many years with little or no fruit production).

The NTFPs examined in this study showed that the time a harvester would need to wait between successive harvests to achieve a steady harvest level varies tremendously with the product being harvested.

Harvesting latex from amapá trees at Tekohaw showed that some products can be obtained from the same trees on a relatively frequent basis, but there are limits to the amount of material a tree can produce in a short period of time. C. guianensis and P. amapa trees both yielded latex twice a week for five weeks in 1999, but in 225 both species the average amount collected per tree dropped off significantly by the fifth week of harvesting

(Appendix 6-A). A similar series of harvests conducted on the same trees in 1998 showed that latex harvest levels had returned to full capacity after a year of rest. Experience with rubber tree (Hevea brasiliensis) latex extraction has shown that an improper harvesting regimen can reduce a tree’s productive capacity (Oliveira, 1998) so additional studies with amapá latex species are needed to determine what method and frequency of cutting are tolerated by them.

Quantifying the appropriate harvest cycle with breu trees was complicated because resin build-up was determined by the growth rate and colonization patterns of Sternocoelus weevils as well as the resin generation capacity of the trees. While resin could be harvested from the same tree once or twice a year, the study estimated that initial harvest levels could only be repeated if trees were harvested once every four to five years. There are probably many other NTFPs based on plant defensive compounds whose production and availability are based on insect and microbial attacks that have not yet been described.

The renewable harvest frequency for titica vines will probably be measured in decades rather than years.

This study in the Tembé reserve reinforced results of other researchers (Hoffman, 1997; Durigan, 1998) who found that some severed roots do grow back slowly, but many roots and possibly many plants die as a result of harvesting. It is, therefore, possible that an area that has been intensively harvested will not offer commercial densities of titica vines again until young plants grow up and have mature roots fixed in the ground. In such cases, calculating NTFP yields and devising sound harvesting strategies should look to timber management for models that assume harvested plants will die and that product for the next round of harvesting will need to come from the next generation of plants.

The copaiba case study demonstrated another NTFP where the first harvest from individual plants could be the only harvest even if harvesting does not mortally wound the tree. This study confirmed results of research on C. multijuga (Alencar, 1982) that showed successive harvests from the same trees conducted six months to a year after the initial one yielded little or no oleoresin. There are many published statements that copaiba trees yield large amounts of oleoresin on a repeated basis, but these cases have yet to be scientifically documented.

Managing NTFP species like copaiba could also benefit from long-term timber management models, but it should 226 not be assumed that copaiba trees that have lost oleoresin yielding capacity have exhausted their ecological role in their community in part because they can remain reproductively active for many more years.

HARVESTING AND PROCESSING TIME

Harvesters will obviously make more money for each day invested in an NTFP enterprise by minimizing time spent finding plants, harvesting the raw material and processing it into a marketable form. While copaiba oleoresin requires no processing once it is obtained, finding trees in a low density population, drilling into the heartwood of prospective trees, and possibly returning several days later to collect the product all involve a lot of time. Selling breu is relatively easy because trees with resin are moderately abundant, resin is usually quickly removed from the bark and processing only requires minimal drying outside. Titica vines are the most common

NTFP in the Tembé forest area, but stripping the cortex from the roots requires as much time as harvesting them

(Processing Day/Harvesting Day Table 6.1). This low conversion ratio of freshly harvested roots to clean root stems and the large amount of time required to carry out this processing made the yield of marketable titica product for one day’s work (Commercial Product per Day Harvesting & Processing Table 6.1) much lower than for breu resin. Preparing andiroba oil was the most labor consuming NTFP enterprise I observed in this study. It first required finding and visiting fruiting trees every day to gather seeds that had fallen before too many rotted or were consumed by wildlife. Secondly, seeds needed to be boiled, dried, dehulled, mashed and put out in the sun daily for three to six weeks to leach the oil from them (Appendix 6-B).

MARKET PRICE AND DEMAND

This study confirmed the finding of many previous studies that the price offered to harvesters for raw or even semi-processed NTFPs is very low in relation to the amount of time that must be devoted to procuring any of them. The sobering result of the production ecology studies and low market prices of NTFPs examined in this study was that collectors needed ideal and in some cases unattainable circumstances to make commercial harvesting of these products worthwhile to them. Titica was the only plant that offered the possibility for harvesters working in areas where roots had not been intensively harvested in recent years to earn as much as the 227

Brazilian minimum wage (about $US 3.50 per day in 1999). It was, therefore, not surprising that this was the most important economic NTFP for the Indians in the Gurupi Region. Some Tembé men considered commercial breu collection worth their time, while others did not. Copaiba tree density and/or oleoresin yield per tree would have to be ten times higher than current levels for a person at Tekohaw to make minimum wage collecting this oleoresin. People who tried to make money selling andiroba oil would be the hardest pressed to make the endeavor worthwhile. Given the $US 1.50 per liter paid for the raw oil and a 10% conversion ratio, a harvester collecting and making the oil would need to collect about 47 kg of seeds per day to make minimum wage of $US

3.50 per day. Collectors who sell seeds directly to an oil press factory have the same collection burden since they are paid only $US 0.07 per kg for raw seeds. This would have been impossible at Tekohaw in 1999 when harvesters visiting 47 trees every day collected only 0.6 kg of seeds per day in the peak month of the fruiting season (Appendix 6-B).

I did not gather information on the amount of agricultural products sold by the Tembé, but informal interviews indicated that titica sales were a larger source of income for most people than farinha (manioc flour), rice, or beans. Most families only planted enough manioc to meet their family’s anticipated food needs for the year and only sold farinha if they had a surplus and the price was at least $US 7.50 per sack. At this price, they would earn the equivalent of $US 2.17 per person per day for their efforts making this labor-intensive crop

(Appendix 6-C). This was lower than several of the NTFP case study species (Table 6.1), but the value of the agricultural product per hectare of land was far greater than any of these NTFPs in the short-term or long-term.

The regional office of the Brazilian Federal Indian Agency (FUNAI) was very supportive of my efforts to identify additional marketable NTFPs for the Tembé, but they have prioritized their technical assistance to help villages increase their production of agricultural cash crops such as farinha, rice and beans (Galvão, personal communication 1999).

RESEARCH IMPLICATIONS FOR THE ROLE OF NON-TIMBER FOREST PRODUCTS IN FOREST COMMUNITY DEVELOPMENT

The fact that products such as copaiba oleoresin, andiroba oil, amapá latex, breu resin and titica roots are readily available in regional and some national markets shows that the Tembé and other harvesters are willing to 228 collect them in spite of their low price. It seems that people living in the forest need or want money so badly that they are willing to accept these prices because they perceive they have no better alternative. The Tembé do so rather grudgingly because they are aware that middle-men and retailers are re-selling these products in the same or slightly altered forms for four to more than 100 times as much as they were paid for them. While the need to redress this imbalance in compensation to harvesters in the NTFP marketing chain has been repeatedly stated

(Clay and Clement, 1993; Warner and Pontual, 1994), I saw no easy ways that the Tembé could sell their products to retail vendors or end consumers or take on value-added processing, and they do not control enough of the resource base to simply demand higher prices.

It is clear that the Tembé are going to expand their production of agricultural cash crops as a primary means of increasing village income. The question then remains whether or not NTFPs can provide other economic resources. Continuing to sell breu resin and titica vine roots at current levels provides a modest but at least fairly dependable source of cash to buy certain staples. The other product that can be taken directly from the forest that seems to offer possible economic reward is fruit from açai (Euterpe oleracea) palm tree. In early 2000, an açai fruit buyer sent a boat up the Gurupi River to buy several hundred kilograms of the berries. The Tembé could conceivably take these fruits to market themselves, but the collecting and delivery would have to be highly organized because the berries need to be ripe when picked and reach the market within 24 hours before they spoil.

It will always be easier for people living in flooded varzea forest islands near Belém to harvest and sell açai fruit bunches from nearby trees and bring them to market only one to two hours away than it will be for the Tembé or anyone who lives in place that requires a 12 – 16 hour boat ride and 4 hours by bus or truck to get to market. One potential problem with such a commercial venture is that it could lead to a competition between people wishing to harvest açai for consumption in the home and those wishing to collect it for sale. The fruits are abundant during the peaks of the fruiting season, but in off-peak times, people travel farther from their home village in order to collect the berries.

While NTFPs are sometimes thought of exclusively as products from wild populations, many commonly marketed ones, particularly fruits, are the result of intentional planting. It is common for Tembé to plant cajú

(Anacardium occidentale L.)(cashew) trees in the midst of their manioc fields. After the manioc field has gone through one or two cycles of planting, it is left fallow, but the cajú trees continue to grow amidst the pioneer trees 229 that take over the plot. The masterful timing of this agroforestry practice is that the juicy fruits ripen in these fallow plots late in the dry season when people are laboring in the hot sun in the new fields and need liquid refreshment. The nuts are roasted outside the village so people will not be unnecessarily exposed to the toxic fumes that emanate from their hull. The Tembé man who has been buying breu resin and titica roots started buying cajú nuts from villagers since he found a dealer willing to buy this product in batches of several hundred kilograms (Kapara’i Tembé, personal communication 2000). There are many other trees and herbaceous plants that Tembé plant in their backyard gardens and fields for personal use, so there are probably other plants with a marketable product such as cupuaçu that could also be integrated into an agroforestry system.

The major source of income for many Tembé in Tekohaw is the sale of handicrafts (Mitchell, personal communication 1999). These items are not always thought of as NTFPs, but since they are made with a diverse array of plant materials including seeds, leaves, roots, fibers, wood, and resins and animal materials including feathers, fur, and hide they should qualify even though they occasionally include synthetic materials such as plastic beads and fishing line. One typical Tembé man might toil for three days in the forest and three more days in the village to process 100 kg of titica roots and make $20 for his effort while another man might spend much less time to collect one kg of the same roots and turn them into a basket that could also earn him $20. Skilled

Tembé women can collect one bunch of nuts from a tucumã palm tree (Astrocaryum vulgare Mart.), carve tiny pieces into the shapes of turtles, frogs, and fish, and string them into necklaces that earn them as much as selling a

60 kg sack of farinha (manioc flour).

The limitation of handicraft sales as source of income for community development is that it mostly benefits a few people who are creative artisans and assertive entrepreneurs. Many Tembé who churned out mediocre items were frustrated when the government shop in Belém would not even pay them the bottom price because it already had a hundred items just like it. The handicraft business was unusually good for a year when an

American intermediary took responsibility for buying top-quality items in Tembé villages and personally shipped them to a socially-motivated wholesale buyer in San Francisco. This channel fell apart after the American left

Brazil because it proved too difficult to maintain good communication between leaders in the village, the new intermediary in Belém and the buyer in the U.S. 230

The Tembé and some other Amazonian communities have so far ruled out logging as a means to generate community income. Where logging will occur in well-planned and executed forest management settings, there may be ways that NTFP harvests could be integrated into such plans to increase forest product revenue before any cutting occurs. This study has shown that there are cases when an NTFP can only be profitably harvested from a tree one time or only once during a long period of regeneration. In the case of copaiba trees that only yield a significant amount of oleoresin one time, cutting down a tree could be postponed at least until it reaches the size class of maximum oleoresin production (55 – 65 cm DBH in the trees at Tekohaw) and has its oleoresin removed.

Depending on the circumstances, NTFPs may or may not be able to offer significant economic opportunities for forest communities. The most important role for these products may not be the amount of revenue they generate for a community, but the amount of money they save people who would otherwise need to buy food and other items from stores. People are able to survive in remote parts of the jungle without much cash if they can procure game, fish, fruit, medicines and materials to make shelter, tools and other basic necessities from the forest and streams around them. Where forests are destroyed and these NTFP resources are lost, forest peoples are put into a double bind. They first lose a direct source of food and other materials that they would then need to buy, and in the process of losing the forest they have also lost many of the resources that allowed them to make any money at all. The Tembé communities living along the Guamá River provide a striking contrast to the ones living along the Gurupi. In the Guamá area, the Tembé now have to grow cash crops so they can buy meat from the city. So much of their forest has been burned by non-Indians that hunting game animals is almost impossible, and fishing is rarely worthwhile. The Gurupi Tembé are far from rich in monetary standards, but they directly meet most dietary and other material needs with a diverse variety of fish, wildlife, agricultural and forest products.

They readily acknowledge that their strength as a culture depends on access to all of the plants and animals they use and share a home with in the forest. It’s not surprising that preserving the forest from destruction is their major concern. 231

TABLE 6.1 PRODUCTION ECOLOGY SUMMARY FOR CASE STUDY NON- TIMBER FOREST PRODUCTS AND MANIOC

VARIABLE COPAIBA BREU TITICA MANIOC (Prod. in liters) (Prod. in kg) (Prod. in kg) (Prod. in kg) Density (Harvested Plants/ha) 0.5 10 50 6032 Raw Product/plant 0.1 0.5 0.5* 3.3 Raw Product/ha 0.05 5.0 25.0 17743 Plants Harvested/day 5 30 50 X Area (ha) Harvested/day 5 3 1.5 0.02** Raw Product Harvested/day 0.25 15.0 37.5 376.2 Comm.Prod./Raw Product 100% 83% 20% 23% Comm. Prod./ha 0.05 4.15 5.0 4216 Comm. Prod./Day Harvesting 0.25 12.45 7.50 X Processing Day/Harvesting Day 0 0 1 4.1 Comm. Prod./Day Harv.& Process. 0.25 12.45 3.75 17.43 $US/Comm. Product 3.50 0.25 1.00 0.13 $US/Day Harv.&Process. 0.88 3.11 3.75 2.17 $US/ha (1st Harvest) 0.18 1.04 5.00 527.00 Harvests/30 Years 1 - 10 6 – 7.5 1 - 2 1 - 3 $US/ha (30 Years) 0.18 – 1.75 6.24 – 7.80 5.00 – 10.00 527 – 1581 Values are considered “typical” based on findings of this study and reports in the literature. They could be higher or lower in different regions. Notes: * Raw Product/plant for titica refers to weight of harvestable roots per host tree with the vine. ** - Harvest of manioc includes time devoted to forest clearing, burning, planting, weeding and harvesting roots X – variable not applicable to manioc 232

REFERENCES

Albuquerque, Milton de. 1969. A Mandioca na Amazônia. SUDAM, Belém.

Albuquerque, Milton de and Eloisa Maria Ramos Cardoso. 1980. A Mandioca no Trópico Úmido. EMBRAPA/CPATU, Belém.

Alencar, Jurandyr da Cruz . 1982. Estudos silviculturais de uma população natural de Copaifera multijuga Hayne (Leguminosae) na Amazônia Central. 2. Produção de óleo-resina. Acta Amazonica 12(1): 75- 89.

Baksh, Michael. 1995. Changes in Machigeunga quality of life. pp. 187-205 in Leslie E. Sponsel (ed.) Indigenous Peoples and the Future of Amazonia: An Ecological Anthropology of an Endangered World. The University of Arizona Press, Tuscon.

Balée, William. 1994. Footprints in the Forest: Ka’apor Ethnobotany - the Historical Ecology of Plant Utilization by an Amazonian People. Columbia University Press, New York.

Carvalho, José Edmar Urano de., Francisco José Câmara Figueirêdo, and Carlos Hans Müller. 1996. Comportamento ortodoxo em sementes de sorva, Couma utilis. Boletim de Pesquisa 168, EMBRAPA, Belém.

Clay, Jason W. and Clement, C.R. 1993. Selected species and strategies to enhance income generation from Amazonian forests. FAO Forestry Paper (Final Draft). United Nations Food and Agricultural Organization, Rome.

Duke, James. 1970. Ethnobotanical observations on the Chocó Indians. Economic Botany 24:344-366.

Durigan, Carlos C. 1998. Biologia e Extrativismo do Cipó-Titica (Heteropsis spp. – Araceae) – Estudo para Avaliação dos Impactos da Coleta sobre a Vegetação de Terra-Firme no Parque Nacional do Jaú. M.S. Thesis, Instituto Nacional de Pesquisas da Amazônia (INPA) and Universidade do Amazonas (UA).

Grimes, A., S. Loomis, P. Jahnige, M. Burnham, K. Onthank, R. Alarcón, W. P. Cuenca, C. C. Martinez, D. Neill, M. Balick, B. Bennett, and R. Mendelsohn. 1994. Valuing the rain forest: the economic value of nontimber forest products in Ecuador. Ambio 23(7):405-410.

Hoffman, Bruce. 1997. The Biology and Use of Nibbi Heteropsis Flexuosa (Araceae): The Source of an Aerial Root Fiber Product in Guyana. M.S. Thesis, Florida International University, Miami.

Lancaster, P.A., J.S. Ingram, M.Y. Lim and D.G. Coursey. 1982. Traditional cassava-based foods: survey of processing techniques. Economic Botany 36(1):12-45.

Nepstad, Daniel C., Adriana G. Moreira, and Ane A. Alencar. 1999. Flames in the Rain Forest: Origins, Impacts and Alternatives to Amazonian Fire. The Pilot Program to Conserve the Brazilian Rain Forest, Brasilia.

Oliveira, Ronaldo L. 1998. Extrativismo e meio ambiente: conclusões de um estudo sobre a relação do seringueiro com o meio ambiente. pp. 93-117 in Alfredo Kingo Oyama Homma (ed.), Amazônia: Meio Ambiente e Desenvolvimento Agrícola. EMBRAPA-SPI, Brasília. 233

Peters, Charles M. 1994. Sustainable harvest of non-timber plant resources in tropical moist forest: an ecological primer. Biodiversity Support Program c/o World Wildlife Fund, Washington, D.C.

Mors, Walter B. and Carlos T. Rizzini. 1966. Useful Plants of Brazil. Holden-Day Inc., San Francisco.

Robinson, John G. and Kent H. Redford. 1991. eds. Neotropical Wildlife Use and Conservation. The University of Chicago Press, Chicago.

Stearman, Allyn MacLean. 1995. Neotropical foraging adaptations and the effects of acculturation on sustainable resource use. pp. 207-224 in Leslie E. Sponsel (ed.) Indigenous Peoples and the Future of Amazonia: An Ecological Anthropology of an Endangered World. The University of Arizona Press, Tuscon. van den Berg, Maria E. 1984. Ver-o-Peso: The ethnobotany of an Amazonian market. Advances in Economic Botany 1: 140-149.

Warner, P.D.III. and Pontual, A.C. 1994. Manual de Comercialização de Produtos Florestais. GENESYS (Gênero em Sistemas Econômicos e Socias), The Futures Group, Washington, D.C. 234

APPENDIX 6-A PRODUCTION ECOLOGY STUDIES ON AMAPÁ LATEX AT TEKOHAW

INTRODUCTION

In 1998, I conducted a pilot harvesting study on 17 trees near Tekohaw that the Tembé classified as

“amapá branca” (white amapá), “amapá vermelha” (red amapá) and “amapá preta” (black amapá). An examination of leaf specimens from each study tree revealed that all white and red amapá trees were

Parahancornia amapa (family Apocyanaceae) – the species most commonly assigned the amapá common name

(van den Berg, 1984). The tree called black amapá by the Tembé was Couma guianensis Aubl. – another species in the Apocyanaceae family. Other species in this genus have been well-known producers of latex used for industrial purposes, chewing gum and medicinal purposes (Mors and Rizzini, 1966; Duke, 1970;

Carvalho et al., 1996). All types of these trees were occasionally harvested by some people from the village to obtain the latex used as a general tonic, particularly by women feeling weak after childbirth. The traditional method of harvesting is making a diagonal gash in the tree with two narrowly spaced cuts of a machete. The latex is then collected in a container that is held or fixed at the bottom of the gash. In order to standardize latex harvest studies for this tree, I used a short handled tool with a curved sharpened point (“faca de borracha”- rubber knife) that is typically used for collecting latex from rubber trees (Hevea brasiliensis).

METHODS

In the pilot study, all 17 amapá trees ≥30 cm DBH that were found during the search for copaiba trees in 1998 were included in the study. It included eight C. guianensis trees (average DBH 38.1 ± 2.5 cm) and nine P. amapa trees (average DBH 40.8 ± 2.5). Each tree was first cut with a rubber knife at a 45 degree angle to a length that was one-third of its circumference at a height approximately 1.2 m above the ground. Each cut was approximately one cm wide and one cm deep. This depth was sufficient to remove the bark and expose the latex producing tissues and vessels. The next five cuts were made four cm above the previous cut. The sixth cut was made on the opposite side of the tree, and then subsequent cuts were made above it. Trees were 235 harvested twice a week for five weeks. Latex was collected in graduate cylinders until latex flow diminished to a drip rate of ≤1 ml in a five minute interval. The study showed that C. guianensis trees yielded more latex than P. amapa trees, but results lacked some precision because there was considerable variation in the width and depth of cuts made by different Indians harvesting the latex.

The amapá latex harvest study was repeated on the same trees in April and May, 1999 with three modifications. Each cut was 25 cm every time a tree was harvested, and all cuts were made by the same harvester. The pilot study showed that on average 98% of the latex that emerged from a cut did so in the first

45 minutes; the 1999 study only collected latex from each tree after cutting for this amount of time. This study was also conducted for five weeks with twice-weekly harvesting.

RESULTS

The results of the 1999 amapá harvest study showed distinct patterns of resin production and yield between the two species of trees. The C. guianensis trees yielded an average of 100 ± 27 ml per tree during the first cut. This average steadily declined to 22 ± 6 ml per harvest as new cuts were made every three or four days above the previous one. The yield rebounded to 75 ± 19 ml per tree when a new series of cuts was begun to the side of the first ones, but it, too, steadily declined to 21 ± 8 ml per tree during the tenth and final harvest.

The P. amapa trees yielded 19 ± 6 ml during the first cut, but this average dipped and rose to a peak of 43 ± 12 ml per tree during the sixth round of harvesting (week 3 of the study) and then declined to an average yield of

28 ± 6 ml per tree in the tenth and final harvest. The Couma harvest pattern indicates that latex that had already been produced was simply being drained away in successive rounds of harvesting. The Parahancornia harvest showed that harvesting may have at least temporarily stimulated latex production.

The relationship between latex harvest and the size of the tree was also different between these two species. Regression analysis that compared tree diameter (DBH) of C. guianensis trees to the amount of latex harvested from them fell far short of significance (R2 = 0.2; p = 0.687) while the same comparison for P. amapa trees did show a slight but significant correlation (R2 = 6.4; p = 0.26). 236

APPENDIX 6-B PRODUCTION ECOLOGY STUDIES ON ANDIROBA OIL AT TEKOHAW

METHODS

I studied andiroba (Carapa guianensis) fruit production and seed oil processing between March and

July of 1999. During this time, Tembé colleagues and I visited 47 trees in the “baixo” (forest subject to occasional seasonal flooding) and “terra firme” forest around Tekohaw. These trees had a diameter range of

20.9 to 70.5 cm DBH with an average DBH of 38.1 ± 1.6 cm. We visited study trees every other day in March when fruits were just beginning to mature on some trees. Between April and July when fruits matured and fell, trees were visited every day. Mature andiroba fruits have a hard outer pod with an average of nine seeds per fruit that each weighs an average of 16.7 ± 0.1 g (n = 2134 seeds). During these visits we collected all fruits

(including loose seeds and pieces of the outer pod) found underneath the canopy of every tree. Each seed was weighed and classified as bad (rotten or largely consumed by insect larvae) or good (lack of these “negative” qualities). An attempt was made to match loose seeds with loose pod pieces so the total number of fruits (and seeds) that fell from each tree per day could be estimated. Teeth marks on some seeds indicated that many were partially or fully consumed by wildlife including agoutis, pacas, and brockett deer.

RESULTS

Results of this study showed that these andiroba trees produced an estimated average of 8.0 ± 1.8 fruits per tree, 73.1 ± 16.2 seeds per tree, and 1222 ± 271 g of seeds per tree during the five months of observations. We collected an average of 807 ± 179 g of seeds per tree (66% of the estimated total weight of seeds that fell) with an average of 573 ± 127 g of seeds per tree (71% of seeds that were found) that were considered to be in “good” condition. The greatest number of seeds found was in June and July. In July we found an average of 616 g of seeds per day from all 47 trees combined (13.1 g of seed per tree). Observations of the trees indicated that very few trees had any fruits left in the canopy by the end of the harvesting period. 237

Seeds collected in this harvest were processed into oil by one Tembé woman using her traditional method. Her method was to boil the seeds, store them in the shade until some oil appeared, break open the seed’s outer cover, scoop out the seed meat, mash the seed meat, and then placed the mashed seed meat on a tilted metal surface in the sun with a container at the bottom of the apparatus to catch the oil that slowly emerges. Depending on the size of the seed batch and weather conditions, it took three to six weeks to process the oil from three to nine kg batches of seeds. In one sample, 2569 g of seeds yielded 178 ml of oil that weighed 93.3 g. This means that 14.4 kg of seeds would be needed to produce one liter of oil (6.9% conversion ratio). The conversion from weight of seeds to weight of oil from was 3.6%. The process of preparing oil took so long that I was unable to get results from five other batches of seeds that were still being processed when my time in the field ended.

238

APPENDIX 6-C MANIOC AND FARINHA PRODUCTION STUDIES AT TEKOHAW

METHODS

Like many Amazonian people, the Tembé grow manioc (Manihot esculenta Crantz) and convert the roots to farinha flour that is their most important staple food and occasional cash crop. Since the varieties of manioc that often grow best in poor soils have lethal levels of cyanide, roots need to be detoxified before the product is safe to eat (Lancaster et al., 1982). I did a study in Tekohaw to measure the production of manioc roots, the conversion ratio of processing the roots to farinha, the time involved in the different stages of handling the plant and final product, and the economics of the process. I wished to compare the value of this crop on a per hectare basis and revenue per day basis with the NTFPs examined in this study.

I accompanied Tembé through each of the five phases of manioc and farinha production: field clearing, field burning, manioc planting, manioc weeding, manioc root harvesting, and root processing to farinha. Fields are created in patches of either secondary or primary forest by cutting down most or all large trees (usually those ≥ 10 cm DBH) in the middle of the dry season (August or September). Based on interviews with 25 household heads in Tekohaw, each family cleared an average of 1.03 ± 0.10 ha of forest in

1997. Since groups of men typically take turn clearing each other’s fields, I measured clearing time by marking out five 25 x 25 meter (0.0625 ha) plots in areas destined for clearing and measured the amount of time it took five men using axes and machetes to clear each plot. The average amount was divided by five to estimate the clearing rate per person. I watched field burning, manioc planting, and manioc weeding several times and made an estimate of the amount of time involved in traveling to the site and conducting these operations per hectare based on conversations with the families involved and my own observations.

Depending on the variety of manioc, roots are first ready for harvest between six and fifteen months after planting. On the harvest day, people need to travel to their field, pull up manioc plants, cut off the roots, carry the roots to a stream to soak (to initiate cyanide leaching), and return to the village. I timed one family 239 harvesting manioc roots from a 20 x 20 meter section (0.04 ha) of their field and approximated travel times based on my observations and conversations with several families.

I measured manioc productivity by accompanying owners to 13 different fields. All but one of these fields had been planted 15 months earlier and weeded three to four months after planting. Harvesting was done by pulling up plants in a contiguous series of 2 x 2 meter plots in a harvest area designated by the field owner.

The number of manioc stems and weight of manioc roots was recorded for each plot. The study team harvested 12 to 40 plots per field per harvest. The number of plots sampled per field in one day depended on how much manioc the field owner needed to produce a two to three week supply of farinha for his family. The average area harvested per field was 119.4 ± 13.0 m2 (0.01 ha).

Farinha production day begins three to five days after manioc roots have been soaked in a stream when the owner determines the roots have reached the proper degree of softness. He then gathers fire wood and brings it to the shelter that stores the “forno” – the ceramic oven used to cook the manioc in a 1.5 to 2 meter wide round steel pan. The soft manioc roots are taken out of the water and have their outer shells removed. Root interiors are then grated and mashed and have hard pieces removed. The mass is then placed in a “tipiti” (a five-foot long flexible tube woven from local reeds) or other type of press to squeeze out as much free water as possible. The strained mass is then placed in the pan of the hot forno to drive off any remaining cyanide and produce the dry crunchy farinha flour. I quantified the conversion ratio and time required for a family group to process manioc roots to farinha flour on five occasions.

RESULTS

Results of yield studies showed that the Tembé fields produced an average of 17743 ± 1892 kg manioc roots/ha (n = 13 fields). This average places it in the “regular” yield range (15 to 20 metric tons/ha) for manioc harvests in the Amazon where root production between 20 and 25 tons per hectare is considered

“good” and yield less than 15 tons per ha is rated “low” (Albuquerque, 1969). It took an average of 4.21 ±

0.26 kg of manioc roots to make 1 kg of farinha (n=5) so the Tembé conversion rate of 23.7% for this process was lower than the 32 to 35% range cited in other Amazon manioc studies (Albuquerque, 1969 and

Albuquerque and Cardoso, 1980). Using the Tembé conversion rate, the fields around Tekohaw produced an 240 average of 4216 ± 450 kg of farinha per ha. Farinha is typically sold in 60 kg sacks that sold for about $R 15 per sack ($R 10 to $R 20 range) between 1997 and 1999 (Galvão, personal communication 1999). Tembé fields were, therefore, producing an average of 70.3 sacks of farinha/ha worth an estimated $R 1054/ha.

The time required to clear a forest with hand tools averaged 0.033 ± .006 ha per hour per person with the assumption that two hours were spent in transit to the field. Using the farinha per ha average value mentioned above, the clearing rate averaged 0.47 ± 0.09 hours per person per sack of farinha produced. This rate was 45% longer in the two plots that contained undisturbed primary forest than in the three plots of 20- year-old secondary forest because one large tree took much longer to bring down than many smaller pioneer trees. Since 1999, the village has had greater access to chainsaws so the clearing rate will be much faster.

Field plots were burned two to three months after they were cleared. Manioc stems were planted several weeks after clearance. When new manioc plants were three to four months old, the fields were weeded.

Each of these operations required the owner and several friends or family members to spend about one day to go to and from the site and carry out the required task. I estimate that each of these operations added about the same amount of time needed to make the final product (0.5 hour per person per sack of farinha) as did field clearing activities. Results of manioc harvesting showed that it required about 3.2 hours per person per sack of farinha produced. Processing manioc roots into farinha was by far the most time consuming phase of the entire process since it took an average of 19.4 ± 4.6 hours per person per sack of farinha produced.

Combining the five elements of farinha production showed that a total of 24.1 hours of time per person was needed to produce one 60 kg sack of the product. This means a Tembé farmer could make $R

2.87, $R 4.34, or $R 5.81 per seven hour day as the selling price for a sack of farinha increased from $R 10 to

$R 15 to $R 20. While the Tembé have not quantified the value of their time in these terms, they clearly carefully assess their opportunity costs in deciding whether or not to sell farinha (and NTFPs as well) since most men were not willing to sell it unless the price was at least $R 15 per sack. One man who made a high- quality farinha wouldn’t sell a sack for less than $R 18. Tembé farinha sellers reaped the full sales price of their product because the government Indian agency (FUNAI) usually took responsibility for arranging its sale and transportation. FUNAI was also trying to make farinha production more efficient by providing electric manioc grinders and new farinha roasting ovens. EPILOGUE

242

CHALLENGES AND LESSONS STUDYING NON-TIMBER FOREST PRODUCT ECOLOGY IN AN INDIGENOUS COMMUNITY

Learning how to do publishable quality scientific work is not easy under any circumstances; learning how to do it in a different culture, a different environment and speaking a different language makes the learning curve even steeper. Scientific journals, though, tend to present the successful parts of researchers’ work, and relatively few researchers write about the non-scientific challenges or failures they faced. As I stated in the introduction to this dissertation, my experiences in the field taught me many things that I cannot condense into a case study table, so I will present some of these thoughts here since I believe they, too, might be valuable to other researchers.

OFFICIAL CHALLENGES

The official challenges of doing NTFP research in Brazil and many countries begin long before one sets foot in the forest. Getting a research visa is the first major hurdle. This first requires finding a “scientific counterpart” who will vouch for the integrity of your work and financial solvency. I was fortunate to find a professor at the agricultural university (FCAP) in Belém to take on this role of handling paperwork with the federal government. Getting the visa next required the approval of my research proposal by the national research council (CNPq), natural resources agency (IBAMA), the Indian agency (FUNAI), three independent scientists, and Minister of Technology. This took “only” nine months because I spent hundreds of dollars in phone calls to Brasília pleading with various officials to move my application on to the next person. Patience and persistence are required to succeed.

HEALTH CHALLENGES

I got every immunization required or recommended ahead of time and brought a small pharmacy’s worth of prescription and over-the-counter drugs with me into the field. “Where There is No Doctor” (Werner et al., 1992) was my family’s most well-read book during our time in the village. I drank water that had been 243 passed through a ceramic filter, applied andiroba oil to my skin to ward off biting bugs during the day and slept in a hammock surrounded by mosquito netting at night. In spite of these precautions, there were other irritants and more serious maladies we had to regularly confront.

After getting badly infected bug bites during my first trip, my family adopted an evening first aid ritual, we took turns treating bites and scrapes with alcohol, Mercurochrome, and anti-biotic ointment. Tembé neighbors showed us how to pluck burrowing fleas from tender toe areas with an orange tree spine before they festered. One day my wife and daughter removed several hundred tiny ticks from my body after I had stumbled into a nest of them in the forest. Other nasty encounters with invertebrates included wasp stings, ant bites, and accidentally squishing a venomous hairy caterpillar on my arm. Tall rubber boots saved my field crew and me several times from poisonous snake bites, and I was lucky not to get malaria when many others in the village did. My body was not always strong enough or well adapted to the environment, though, so I periodically got sick with a fever or malaise of unknown origin that caused me to miss days in the field. I left the village for a whole month once when bad headaches and pain in my joints became an unrelenting part of my day. I did not completely recuperate until I came home for an extended period of time. I loved this field work as much as anything I’ve ever done, but taking care of my health in this climate required a lot of vigilance.

LOGISTIC, COMMUNICATION AND TECHNICAL CHALLENGES

Doing research in primary tropical forests almost by definition requires working in remote places because most places that are easy to get to have been significantly altered. Traveling from Belém to my field site at Tekohaw was almost always a saga. Scheduling a trip to the reserve took anywhere from two days to two weeks. Once a journey began, it first involved a four hour bus ride from Belém to the town of Gurupi at the intersection of the main Belém highway with the Gurupi River. One often had to spend one or two nights in a rickety store room until a government or Tembé boat arrived and was provisioned for the trip up river.

Traveling in an open motorboat involved a 6 to 10 hour trip to the village if the motor did not break and gasoline did not run out on the way. Going by covered motor launch saved one from exposure to rain, but the trade-off was spending 12 to 24 hours in transit with the constant noise and fumes of a diesel engine. Travel 244 during the dry season was particularly tricky since boats had to navigate through rocky rapids, and several boats capsized in the process. Tekohaw now has an airstrip that can handle small planes, but chartering a private plane for this one hour trip costs about $800, so I took whatever boat was available and did my best to protect equipment for whatever might happen on the way.

Once I made it into the village, it was not easy to communicate with anyone outside it. The life line of Tekohaw to other villages and FUNAI is a ham radio. There were daily check-ins to share important news, pass along requests for transportation, and arrange meetings. Once a month or so I could send out a letter with someone going to the city or receive mail brought in by the FUNAI agent. I looked into a cell phone with international capabilities but the licensing procedure and costs were prohibitive. In my second year I periodically listened to BBC International, but in general once I was in the reserve I pretty much had to accept that I was not going to have any contact with the rest of the world for six to eight weeks. This was a relaxing blessing in many ways and tough in others.

My brand of ecological field work fortunately did not rely on a lot of sophisticated equipment. I brought standard 50 meter measuring tapes, tree diameter tapes, compasses, tree tags and GPS units from the

U.S., and I easily got basic tools like machetes, hammers, drills in Belém hardware stores. Accidental machete cuts through measuring tapes were remedied with duct tape. Setting up my house as a lab and office that could function at night was a bigger challenge. Three unbreakable solar panels worked flawlessly for years, but I went through three expensive car batteries before I finally worked out (with the help of a digital voltmeter) how to budget my energy consumption for a couple of compact fluorescent light bulbs, a laptop computer and portable printer. I did without the computer for the first few months since its screen was damaged in the first boat trip to the reserve, but once it was repaired, it became a vital tool for creating data collection sheets, entering data, and writing reports in the field.

BLENDING INDIGENOUS KNOWLEDGE WITH SCIENTIFIC INQUIRY

I intentionally wanted to study NTFPs with indigenous people because I believed that I could learn a lot more about these plants working with people who were familiar with them than if I simply set off with a tape measure into a pristine area of forest by myself. It took less than a few days to confirm that my Tembé 245 colleagues knew more about the forest they had grown up in than I could learn in my lifetime, but I was eager to learn as much as I could from them. It took a good bit longer to realize that their understanding and beliefs did not always translate well into my desire to assign numerical values to some things or describe natural history relationships in traditional biological terms. One lesson was learning how to gauge the response to a question about the abundance of a particular plant or how much product it typically yielded. Beginning with the copaiba work, the Tembé elders I first worked with assured me that we would find lots of trees and told me about times when they had filled large cans with oleoresin. It was rather sobering to them and me when we found that it took several weeks of daily searching just to find 50 trees and that we could only gather a few liters of oleoresin after drilling most of them.

Another important lesson I learned working in the villages was that the level of skills and knowledge varied dramatically between individuals. Some people were very adept at finding certain kinds of trees, but not others. Elders could generally identify hundreds of types of trees but were unable to read, write or count higher than ten. Younger men generally knew relatively few trees, but a few were quick to learn how to read instruments and reliably record the measurements on data sheets. Research teams for any given NTFP almost always had people with a blend of these skills, so the research could provide opportunities for cross- generational learning and appreciation.

When I started asking about breu, Tembé and Ka’apor collectors told me about the “tapuru” (larvae) they found in the lumps and related stories about filling entire sacks of resin from single trees. As the research progressed, it became apparent that the Indians’ observation of the “tapuru” was fundamental to the resin’s formation, and we proceeded to probe this relationship together in great detail. On the harvest side, it again became apparent that the huge yields people described were the result of their best days, not their normal ones.

This reinforced the recommendation by Grimes et al. (1994) that estimates of NTFP harvest amounts obtained through ethnobotanical inquiry be followed up with actual production studies in order to develop reliable quantitative estimates.

The most culturally challenging situation I confronted related to work in the field involved the regeneration of titica vine roots. As mentioned in the case study, one Tembé elder told me that the slender titica vines climbing some host trees were a different kind of vine than the ones whose aerial roots were 246 harvested. Other Tembé told me that these aerial roots were created when the legs of certain dead ants become elongated. It was a believable story to some since they had not seen flowers or fruits of the part of the plant that is largely out of site in the host tree’s upper trunk. I was skeptical about the Indian version of titica regeneration, but I tried to keep an open mind. Before I found out that titica was a secondary hemiepiphyte that actually starts its life on the ground, I was wondering if ants might play some role in the germination of titica seeds in the canopy. When I later harvested several whole plants at a different site, it was immediately apparent that the roots were attached to the branches off the main stem of the titica plant and that normal reproductive structures were held on the branches’ tips. I regret that I did not harvest some plants with the

Tembé since the plant could have provided us with a concrete example to discuss what I felt awkward saying based on book knowledge. My hope is that harvesters who have a better understanding of the plant’s regeneration mechanisms will be better able to regulate the harvest to preserve the resource for the future.

EXPECTATIONS OF A RESEARCHER

Working with the Tembé provided me an opportunity to learn about the forest in ways I never could have done on my own and gave me some of the best and worst experiences of my life. Many of these successes and disappointments relate to the fact that my relationship with the community went far beyond my role as a researcher studying NTFPs. As mentioned in Chapter 1, my initial contact with Tembé from the

Gurupi River villages occurred in the U.S. when I escorted a leader on a 10-day long tour of the U.S. that gave him many opportunities to discuss his people’s needs in the areas of health care, education and land rights. I was welcomed to Tekohaw not just as someone who would study the economic potential of a few plants but as someone who would be an active participant in many aspects of village life and concerns. For the better part of four years, I was known to everyone in the community as “Zui’zu” (white frog). I was determined not to be another scientist who came into the community, gathered whatever data and materials I wanted and left without giving much in return.

I tried to help direct resources to the Tembé community in a variety of ways. The most supportive people were from the State College Friends Meeting that formed the Tembé Indian Support Committee. This committee secured several large donations from one individual and did fundraising projects that included: 247 young children making and selling buttons they made with Indian designs, older children making dinner and offering babysitting services, and all children selling flower bulbs. I bought and traded for Tembé handicrafts in the village and sold them at the Friends Meeting and at slide shows I did about the Tembé in the U.S. I arranged for a Tembé leader from Canindé village to attend an Amazonian indigenous conference in New York and recruited five people from the U.S. to attend a traditional festival in the reserve in return for a donation to the community. I obtained donations of clothing, hand tools, medical and school supplies, soccer balls and sports equipment for six Indian villages along the Gurupi River. Funds raised in these projects were used to finish construction of a large wooden launch, buy a new motorboat and diesel engine, repair two gasoline engines, produce a book of 200 traditional Tembé chants in Tembé and Portuguese, and possibly finance a major land rights project.

While living in the village I tried to contribute to the community by bringing in ample quantities of fish hooks, fishing line and ammunition for trade, loaned out my tools, fixed broken water pipes, tended sick children, took family photos, and gave evening classes in math, geography, science, and environmental advocacy. When I left the village I donated my solar panels, furniture and remaining first aid and paper supplies to the schools and infirmaries of three different villages.

Trying to get my research done and be a good employer presented many challenges. The first summer in Tekohaw the elder Tembé worked with me without formal wages, but at the end of the summer they each made a list of items they wanted from the city in compensation for their time. I initially liked this system because it felt more like a cooperative venture than a business arrangement, but it was not fair because a few people who worked little wanted more than others whose daily efforts were integral to the project. When I returned in 1997 I paid everybody who worked with me the same amount for each day they worked and raised this amount every year. This system in general worked well, but there was social pressure to hire as many people as possible. This was difficult because my funds were limited, and I mostly needed people who either knew forest trees very well, could read and write numbers well or both. At one point I created a special project to measure manioc productivity more as a way of paying some extra people than because the information was vital to my research goals. 248

I strove for an elusive balance between firmness and flexibility in dealing with research needs versus other needs of the village and my helpers. I got very little work done in the fall of 1997 because most men were away from the village cutting boundary lines for the reserve. Shortly after they came back they all got engaged in the 10-day long St. Benedict’s Festival. I had to plead with the community to release a few people to me or I would not be able to accomplish anything. A year and a half later the situation was reversed. I sometimes had four crews of people going out on the same day to get production data on andiroba, breu, amapá and copaiba. It seemed wonderful until the chief came to me and asked me to release some of these men so they could work on the community garden and building projects. It became apparent that while many people appreciated the opportunity to work, the community was suffering because their desire to do work for the community for free had begun to wane. Some of these people at least willingly made donations to a travel fund for the leader when there was an important meeting for him to attend in Belém or Brasília.

There is no doubt that taking on such a broad range of research topics had other consequences for my work and relationships. By trying to gather production data about so many plants during a two-year period, I inevitably sacrificed gaining a deeper understanding of a few of them. Over time, I became more of a field manager than an ecologist slowly building my insights through day by day observations in the field. Part of this seemed unavoidable since there were many days when I felt too weak or sick to go in the forest, but this mode gradually altered my relationship with my Tembé field assistants. I knew next to nothing about the forest, but when I faced the mud, bugs, poisonous snakes, rain and shared one can of sardines with them for lunch I earned their respect. When I stayed in the village more often during the day, I had fewer chances to personally share their successes and travails in the forest. My interactions with them became more centered around checking data sheets at the end of the day and paying them minimum wage every few weeks. Our relationship had uncomfortably shifted from friends and colleagues to employer and employees.

At the end of the project it was clear that the expectations the community had of me (and that I had for myself) were more than I could possibly meet. By hiring people from one family in Tekohaw, other families sometimes felt slighted if I did not hire someone from theirs. Since I lived in Tekohaw it was natural that I hired mostly people from there and provided other types of assistance to that community. This made some people in the other Tembé villages upset that I was not paying more attention to them. Since my field work 249 was carried out exclusively in Tembé villages and fundraising projects brought benefits mostly to them, the

Ka’apor Indian villages on the other side of the Gurupi River began to feel increasingly angry that I was unjustly ignoring them. I was aware that such feelings could arise and tried to deal with it by hiring representatives from several villages to work in projects at Tekohaw, but it was difficult for people with families to stay away from their home villages for long. I did one well-received side project in Cajueira village, but it was not possible to do so in others without fracturing ongoing research at Tekohaw. Boats and boat equipment were theoretically used by all of the villages, but over time, each village wanted their own craft. I tried to channel funds earmarked for a land right’s project to a committee of village leaders, but the delay in releasing funds due to inter-village politics led to recriminations against me by individual villages that wanted me to release the funds directly to them.

I did one round of visits to all of the Gurupi villages to explain the goals of the research project and did regular updates at Tekohaw. It was evident, however, that even at Tekohaw the level of understanding of all of these studies was not very high. The simplest explanation was that I was studying plants that might lead to additional sources of income, and I used the most creative ways I could think of to convey quantitative concepts to people who knew little or no math. When the research began to show that the new products being investigated were not very promising in this respect, it became harder to explain to most people why continuing such studies was still important or relevant to them. Information on some study trees was lost because people removed numbered tags or harvested trees for themselves in between research harvests. Some people who worked with me took a genuine interest in probing topics like the relationship between the breu resin and the weevil larvae that lived in it, but many young men worked with me mostly because it was a relatively easy and interesting way to make some money. My family was warmly embraced with no reservations, but I believe most of the village accepted me as a friendly foreigner with quirky interests and was content with my presence as long as I provided some tangible benefits to the community on a regular basis.

The increasing interest that people had in making money seemed to fuel speculation that I hoped to get rich taking photos, recording chants and collecting plants around the village. These issues were periodically discussed, but it became increasingly difficult to work without unspoken suspicions and jealousies clouding many interactions. By the time I returned to Brazil in the spring of 2000 to wrap-up field work for my 250 dissertation and to present some preliminary research results, the cord of trust between several villages and me had badly frayed. I left with my head high but a very heavy heart.

Indigenous people have an understandable expectation that researchers they welcome to their communities will carry out their work with full consultation, conduct themselves with integrity and fairly compensate people who work with them and the villages that host them. These were my expectations going into this project, but it is apparent I needed to communicate more openly and realistically about these subjects more often. My dialogue with leaders sometimes worked and sometimes did not because I had to operate in the context of intra- and inter-village politics that were often beyond my understanding or control. I could speak Portuguese well enough to do my work in the forest and conduct daily life in the village, but at times I was simply unable to communicate effectively about complex and sometimes volatile emotional and cultural issues because I couldn’t use my native language.

I admit that for a time I enjoyed the feeling that I could do fascinating work in the forest and make contributions to the community in so many ways. The needs of the people seemed deep and diverse, and they stimulated creative energies to help however I could. Trying to be a researcher, medic, teacher, employer, banker, fundraiser, archivist, handicraft buyer, hardware supplier, father and husband were too many roles to keep up for the long haul. My inability to balance these roles led me to make mistakes that hurt my relationship with some members of the community. The cumulative weight of these roles almost stretched the patience of my family and my health to their breaking points.

The issue that I will ponder long and hard as a result of my time with the Tembé is the role of money in the community. I was initially attracted to studying the economic potential of NTFPs because I thought that increasing revenue would help community improve health care, education and support forest protection. Over time these beliefs steadily eroded. Most people wanted “modern” medicines because they had lost some of the knowledge or faith to use rainforest plants, but they didn’t want to spend their money on these things because they expected the government to supply them. The Tembé wanted to have Tembé teachers in their schools, but they consistently failed to honor commitments to compensate the local people who devoted long hours to teaching the village children. Young leaders were unanimously praised for their efforts to defend their land rights, but raising donations from the community to send these people to important meetings was a perennial 251 struggle. People had not abandoned their desire to help the welfare of their community, but they clearly chose to spend most or all of what little money they made on material goods for their family. The appeal of cooking food with a clean flame from a gas stove instead of cutting fire wood, making charcoal and living in a soot- filled house is obvious. People needed cash to purchase certain foods and materials like salt and fish hooks to survive. Given my affluence as a middle-class American, it would be hypocritical to criticize anyone for wanting a new shirt or CD player, but I left with no illusions that increasing village income would necessarily improve the Tembé quality of life, preserve its traditional culture or save its forest. Money can sometimes play a vital role in supporting such objectives, but mechanisms need to be in place in the community to handle such funds in a positive way. My feeling was that the struggle to accomplish these goals was being waged from people’s hearts and didn’t depend on how much money they had or the size of their radio.

I hope I will have the chance to work with another community in the Amazon. There is an important role that researchers can play to promote forest conservation and support rural community development. They can continue to study the complexities of nature and help develop ways for people to coexist with the full diversity of a tropical ecosystem in a world where social and economic conditions are rapidly changing. I would ideally like to take on this kind of challenge again as part of an integrated group of researchers and development practitioners. I do not wish to fall back into a narrowly defined role of a “specialist,” but I realize that I can’t do all the things that I’d like to do working alone. I would seek to do my part in an inter- disciplinary team well without apology for what I can’t do, and maintain as much humility as possible. 252

REFERENCES

Grimes, A., S. Loomis, P. Jahnige, M. Burnham, K. Onthank, R. Alarcón, W. P. Cuenca, C. C. Martinez, D. Neill, M. Balick, B. Bennett, and R. Mendelsohn. 1994. Valuing the rain forest: the economic value of nontimber forest products in Ecuador. Ambio 23(7):405-410.

Werner, David, Carol Thuman, and Jane Maxwell. 1992. Where There is No Doctor: A Village Health Care Handbook. Hesperian Foundation, Berkeley.

VITA JAMES CAMPBELL PLOWDEN

EDUCATION: 1975: University of Pennsylvania, BA Biology 1993 - 1994: University of Maryland, one year grad. Non-degree; one year in Environmental Science Master's program 2001: Penn State University, PhD Ecology

WORK HISTORY: 2000 (8/00-8/02): Penn State University, University Park, PA; Intercollege Graduate Degree Program in Ecology Program; Acting Head (7/01-8/02); Research Assistant (1/01-6/01); Biology Dept. Instructor/Graduate Teaching Assistant (8/00-12/00; 8/01-12/01).

1995 (3/95 - 7/95): Environmental Investigation Agency, Washington, DC; Wildlife Trade Investigator

1994 (1/94 - 1/95): Univ. of Maryland, College Park, MD; Natural Resources Management Program, Graduate Teaching Assistant and Program Assistant

1993 (1/88 - 4/93): Greenpeace International, Washington, DC; Tropical Forests Campaign, International Coordinator/Special Projects Manager; International Whale Campaign Coordinator

1985 (1/85 - 12/87): The Humane Society of the US, Washington, DC; Whale Campaign Coordinator

1982 (1/82 - 12/84): Greenpeace USA, Seattle, WA; National Wildlife Coordinator

1979 (5/79 - 5/80; 9/80 - 12/81): Greenpeace International, Tokyo, Japan; Japan Project Coordinator

RECENT PUBLICATIONS Plowden, C., C. Uhl, and F.A.de Oliveira (in press) Breu resin harvest by Tembé Indians and its dependence on a bark- boring beetle. in J.R. Stepp, F.S. Wyndham, and R.K. Zarger (eds.) 2002. Ethnobiology and Biocultural Diversity. University of Georgia Press. Plowden, C. (in press) Politics and progress: challenges to sustainable harvest of forest products by Tembé Indians in the Brazilian Amazon - Response to editorial by Cunningham and Shanley. People and Plants Handbook special issue Managing Resources: Community-based Conservation. Plowden, C. (in press). Profiles on Carapa guianensis, Parahancornia amapa, Protium spp., and Copaifera spp. in P. Shanley, (ed.) Tapping the Green Markets: Management and Certification of Non-timber Forest Products. Earthscan Publications, London. Plowden, C. and D. Bowles. 1997. The illegal market in tiger parts in northern Sumatra, Indonesia. Oryx 31(1): 59-66.

AWARDS AND GRANTS: 2001: Penn State Univ. J. Brian Horton Award in Ecology 2000: Penn State Univ. Biology Department Honorable Mention for Best Teaching Assistant Award 2000: Penn State Univ. Ecology Program Professional Meeting Travel Award 1997: Rainforest Alliance Kleinhans Fellowship (2 years) 1997: Food, Conservation and Health Foundation Grant 1995: National Science Foundation Graduate Fellowship (3 years) 1993: The Hotchkiss School Alumni Association Community Service Award

VOLUNTEER ACTIVITIES: 1999 - 2001: Ecology Graduate Student Organization (of Penn State) - Co-founder and President 1997 - 2001: State College Friends Meeting, State College, PA - Co-clerk Tembé Indian Support Committee 1985 - 1993: Tropical Ecosystem Research and Rescue Alliance International, Washington,. D.C. - Founder/President