<%BANNER%>

Population Structure and Seed Production of Carapa guianensis in Three Floodplain Forest Types of the Amazon Estuary

Permanent Link: http://ufdc.ufl.edu/UFE0024811/00001

Material Information

Title: Population Structure and Seed Production of Carapa guianensis in Three Floodplain Forest Types of the Amazon Estuary
Physical Description: 1 online resource (56 p.)
Language: english
Creator: Cunha, Marina
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: amazon, andiroba, brazil, carapa, community, estuary, forest, guianensis, management, ntfp, population, production, seed, structure, timber, tropical, varzea
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Carapa guianensis (andiroba) is one of the most promising Amazonian multiple-use tree species to foster conservation through sustainable management. While the oil extracted from its seeds is one of the most widely used natural remedies in the Amazon, the species has suffered high logging pressure because of its wood of superior quality. In the floodplain forests of the Amazon estuary where C. guianensis plays a crucial socioeconomic role, we addressed two critical questions about C. guianensis ecology: (1) What is the size class structure of C. guianensis within the floodplain forests by forest type? (2) What are C. guianensis seed production rates by forest type, and what main factors explain observed production variation? The study was conducted in collaboration with local forest managers within a Sustainable Development Reserve, where C. guianensis was identified as the most important species in their livelihood systems. C. guianensis population structure varied significantly across forest types, with drastic disparities in seedling, sapling and adult densities. C. guianensis densities ? 10 cm diameter at breast height (dbh) were 28.7 (plus or minus10.3), 23.0 (plus or minus4.2) and 19.5 (plus or minus5.8) trees ha-1 in Baixio (periodically inundated swamp), Restinga (river-side periodically inundated forest) and Terra Preta (upland flooded forest), respectively. Seed production rates varied significantly by forest type, with an average (plus or minus standard deviation) of 2.6 (plus or minus0.4), 4.1 (plus or minus0.4) and 5.5 (plus or minus0.4) kg of viable seeds tree-1 year-1 in Baixio, Restinga and Terra Preta forest types, respectively. Our final mixed model that explained C. guianensis seed production variation included forest type, dbh, year, crown form, crown illumination and interactions as explanatory variables. We found a quadratic relationship between dbh and seed production, and a marked variation in seed production between the two years monitored. In addition, we found evidence that past timber harvests affected commercial timber stock and population-level seed production. On the other hand, preliminary assessment of the total population seed production suggests that, at current harvest levels, the local community is collecting less than 1% of the C. guianensis viable seed produced annually within its forest lands.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Marina Cunha.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Kainer, Karen A.
Local: Co-adviser: Staudhammer, Christina Lynn.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024811:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024811/00001

Material Information

Title: Population Structure and Seed Production of Carapa guianensis in Three Floodplain Forest Types of the Amazon Estuary
Physical Description: 1 online resource (56 p.)
Language: english
Creator: Cunha, Marina
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: amazon, andiroba, brazil, carapa, community, estuary, forest, guianensis, management, ntfp, population, production, seed, structure, timber, tropical, varzea
Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, M.S.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Carapa guianensis (andiroba) is one of the most promising Amazonian multiple-use tree species to foster conservation through sustainable management. While the oil extracted from its seeds is one of the most widely used natural remedies in the Amazon, the species has suffered high logging pressure because of its wood of superior quality. In the floodplain forests of the Amazon estuary where C. guianensis plays a crucial socioeconomic role, we addressed two critical questions about C. guianensis ecology: (1) What is the size class structure of C. guianensis within the floodplain forests by forest type? (2) What are C. guianensis seed production rates by forest type, and what main factors explain observed production variation? The study was conducted in collaboration with local forest managers within a Sustainable Development Reserve, where C. guianensis was identified as the most important species in their livelihood systems. C. guianensis population structure varied significantly across forest types, with drastic disparities in seedling, sapling and adult densities. C. guianensis densities ? 10 cm diameter at breast height (dbh) were 28.7 (plus or minus10.3), 23.0 (plus or minus4.2) and 19.5 (plus or minus5.8) trees ha-1 in Baixio (periodically inundated swamp), Restinga (river-side periodically inundated forest) and Terra Preta (upland flooded forest), respectively. Seed production rates varied significantly by forest type, with an average (plus or minus standard deviation) of 2.6 (plus or minus0.4), 4.1 (plus or minus0.4) and 5.5 (plus or minus0.4) kg of viable seeds tree-1 year-1 in Baixio, Restinga and Terra Preta forest types, respectively. Our final mixed model that explained C. guianensis seed production variation included forest type, dbh, year, crown form, crown illumination and interactions as explanatory variables. We found a quadratic relationship between dbh and seed production, and a marked variation in seed production between the two years monitored. In addition, we found evidence that past timber harvests affected commercial timber stock and population-level seed production. On the other hand, preliminary assessment of the total population seed production suggests that, at current harvest levels, the local community is collecting less than 1% of the C. guianensis viable seed produced annually within its forest lands.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Marina Cunha.
Thesis: Thesis (M.S.)--University of Florida, 2009.
Local: Adviser: Kainer, Karen A.
Local: Co-adviser: Staudhammer, Christina Lynn.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024811:00001


This item has the following downloads:


Full Text

PAGE 1

1 POPULATION STRUCTURE AND SEED PRODUCTION OF Carapa guianensis IN THREE FLOODPLAIN FOREST TYPES OF THE AMAZON ESTUARY By MARINA LONDRES A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Marina Londres

PAGE 3

3 To the local researchers and the Womens Group of the So Joo do Jaburu community, Gurup, Brazil

PAGE 4

4 ACKNOWLEDGMENTS This m asters research represents an interm ediary step and one piece of a project that started two years prior to my arrival at UF. It has been a quite long, intense and challenging journey, and I heartily admit that the participa tion of outstanding co-workers, and other actors, has been vital to the overall success of this endeavor. I thank all the families of the So Joo do Jaburu community, for welcoming the work, for their willingness to cooperate in every possible way, for their ideas, insights and trust in the research process. Among them, I am particularly grateful to our local co -researchers. First, the veteran group, who has worked as volunteers for so long, believing in the promise of the scientific process as a means to bring local improvements and dedicating the best of their intellectual and physical abilities to this research: Andr Santos, Antnio Santos, Domingos Souza, Jos Carlos Costa, Jos Jairton Pi res (Pipo), Manoel Cordoval Souza (Cod), Manoel Souza and Oberaldo Brilhante. Later on, a second generation of local re searchers joined the group, bringing even more life, diversity and tale nts to our research: A ndria Santos, Benedito Souza, Claudionor Lima, Diane Costa, Fernanda Pessoa, Helosa Ferro, Leleco Martins, Mnica Souza, Osias Brilhante, Paulo Nascimento, Raimundo das Graas Souza (Grao), Roseles Guimares, and Sandro Souza. What I have lear ned from these folks, how I have enjoyed working with them, and how much growth and tr ansformation these inter actions have brought to me are just beyond the measurable. Mark Schulze supervised the work since its initial design and implementation. If this participatory research has the merit of being scie ntifically rigorous, Mark is the one responsible for that. To us, he opened door to the fascinati ng world of forest ecology research; he challenged us to do more, and better, and more, and better, to the level at which we never imagined we would be able to attain (and we did). And he he lped us not only as a scientific mentor, but he

PAGE 5

5 created the conditions for this work to keep going at very critical moments when there seemed to be no hope. I also thank Mark for si nking his legs in the mud with us in the early stages of data collection, for being so patient with my initial inexpe rience and for continuing to be engaged along the way. Upon my arrival at the Universi ty of Florida, I had the fortune to work with two incredible co-advisors. Karen Kainer and Christina Sta udhammer gave me thorough advice and support. While enthusiastically engaging in this research, th ey generously shared their precious expertise, and provided me with the tools and guidance to develop as a sc ientist. In addition, Karens passion for community/participatory approaches empowered me to keep up the fight; and Christie sparked a real transformation in my st atistical life it literally moved from hell to heaven. I warmly thank my Imazon team. Valdir Pr imaveras dedication and perseverance were fundamental to the success of da ta collection throughout these years; with a strong background in the regional grass-roots movement, Valdir was also key in our e ndeavor of linking our ecological research with the local reality. Marcelo Galdino had great participation entering loads of data, supervising activities wh ile I was in Florida and helping wi th field data collection. Rodolfo Gadelha did a wonderful job digitizing a massive amount of overwhelmingly detailed participatory maps. Francy Nava displayed incredible energy in helping us set up NTFP markets for the community, after research repercussions started to take pl ace. Edson Vidal had a critical role in the early development of the project, as a supportive and enthusia stic coordinator, and helped us solve tough problems along the way. I do not know how to express my gratitude to Carlos Augusto Ramos. For almost 6 years now, Carlos has been a crucial and wonderful collaborator. The community forest management

PAGE 6

6 groundwork that he established in Gurup created the platform from which our research could flourish. His thoughtful (and successful) approach of linking forest management with social movements in the Amazon estuary region has str ongly inspired and influenced me over this trajectory. Throughout, he has been enhancing the work with his unique perspective, always providing great insight s and warm support. I am ever so thankful to my early advisor Carmen Garcia. The strong guidance and support I received from her were essential to my prof essional establishment in the Amazon, and I still benefit from this investment. Patricia Shanley also had an important ro le in my Amazonian launch. She brought me into the arena, inspired me with her amazing research approach and enriched my work through more recent collaboration. I thank Imazon and FASE-Gurup institutions for the logistical support, constructive interactions and collaborative environments. The European Union, the Tropical Conservation and Development Program at UF, a Rufford Small Grant for Nature Conservation, and the School of Forest Resources and Conservation (U F) provided indispensa ble financial support. I thank my friends (even though I cannot list a ll of them here) for making life so much better, for helping me live through tough moment s, and for encouraging me to keep moving forward despite all the obstacles. Carol Mathias, Julio Cesar da Costa, Carol Mendes, Carlos Anselmo, Teluira Andrade, Carol Bentivi are friends from college ever present in my heart. Bia Ribeiro, Adriano Jerozolimsky, Marco Lentini, Danielle Celentano, Wandreia Baitz, Sheyla Leo are friends from my time in Belm who r eally helped me through this. I thank Marcio Sztutman for the love and care during the years we spent together. Ana Carol Silva, Carla Stefanescu, Raissa Guerra, Ana Eleuterio, Ane Alencar, Leo Pacheco, Paula Pinheiro, Christine Lucas, Laura Schreeg, Lucimar Souza, Paulo Br ando, Amy Duchelle, Tita Alvira, JG Collomb,

PAGE 7

7 Maria DiGiano, Fay Ray, Christie Klimas, Joanna Tucker, Vivian Zeidemann, Geraldo Silva, Morena Maia, Eric Carvalho, Pedro Constantino, Lucas Fortini, Jamie Cotta, Felipe Carvalho, Simone Athayde, Jennifer Arnold, Shoana Humphries, Valrio Gomes helped me survive the intense life abroad over the past two years. Finally, I thank those without whom I would be good for nothing: my Mom, Maria, for her unconditional love and support, fo r helping me to keep strong at any moment and for her liberating guidance in our sear ch for the truth; my Dad, Eduardo, for his immense love and friendship, ever encouraging me to follow my heart, no matter the outcomes, no matter how geographically far it would take me away from hi m; my sister, Flavia, for being an angel in my life; my brother in-law, Gabriel Fernandes; my step-mother Flavia Duarte; and the precious, so loved additions to the family, Letcia, Catarina and Andr.

PAGE 8

8 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES .........................................................................................................................10LIST OF FIGURES .......................................................................................................................11ABSTRACT ...................................................................................................................... .............12 CHAP TER 1 INTRODUCTION .................................................................................................................. 142 POPULATION STRUCTURE AND SEED PRODUCTION OF Carapa guianensis IN THREE FLOODPLAIN FOREST TYPE S OF THE AMAZON ESTUARY .......................182.1Introduction .................................................................................................................. .182.2Study Species ................................................................................................................192.3Study Site ......................................................................................................................202.4Methods .........................................................................................................................212.4.1Population Structure .......................................................................................... 212.4.2Fruit and Seed Counts ....................................................................................... 222.4.2.1Extensive sample ................................................................................ 222.4.2.2Intensive sample .................................................................................242.4.3Data Analyses ................................................................................................... 252.4.3.1Population structure ............................................................................ 252.4.3.2Seed production .................................................................................. 252.5Results ...........................................................................................................................272.5.1Population Structure .......................................................................................... 272.5.2Population Structure vs. Forest Structure ......................................................... 282.5.3Seed Production ................................................................................................ 292.5.3.1Intensive sample predictive models .................................................... 292.5.3.2Extensive sample mixed model .......................................................... 292.5.3.3Production rates .................................................................................. 302.5.3.4Production variation ............................................................................ 302.6Discussion .................................................................................................................... .312.6.1Population Structure .......................................................................................... 322.6.1.1Densities ............................................................................................. 322.6.1.2Size-class distribution .........................................................................322.6.2Seed Production ................................................................................................ 352.6.2.1Production rates .................................................................................. 352.6.2.2Production variation ............................................................................ 362.6.3Management Implications ................................................................................. 393 CONCLUSION .................................................................................................................... ...48

PAGE 9

9 APPENDIX A FOREST TYPE MAP .............................................................................................................50B DISTRIBUTION OF MODEL RESIDUALS ........................................................................51LIST OF REFERENCES ...............................................................................................................52BIOGRAPHICAL SKETCH .........................................................................................................56

PAGE 10

10 LIST OF TABLES Table page 2-1 Description of four main forest types at So Joo do Jaburu, Gurup, Par, Brazil. ......... 422-2 Descriptive results of the characterization of 1105 C. guianensis trees, according to diameter classes. ............................................................................................................. ...442-3 2007 and 2008 model coefficients (from the intensive sample) ........................................ 452-4 Explanatory variables included in the final model (extensive sample) ............................. 452-5 Mean ( standard deviation) va lues of viable seed production per C. guianensis tree by forest type and percenta ge of non-producer trees. ........................................................ 452-6 Percentage of non-pr oducers by size class. ........................................................................462-7 C. guianensis densities at different study sites. ................................................................. 462-8 C. guianensis estimated total population seed prod uction by forest t ype and by year. ..... 47

PAGE 11

11 LIST OF FIGURES Figure page 2-1 Size class distribution of 1105 C. guianensis trees, by forest type ....................................432-2 Mean ( standard error) C. guianensis individual seed produc tion by forest type and year .......................................................................................................................... ...........462-3 Estimated mean ( standard error) C. guianensis individual seed production by dbh and by forest type ...............................................................................................................47

PAGE 12

12 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science POPULATION STRUCTURE AND SEED PRODUCTION OF Carapa guianensis IN THREE FLOODPLAIN FOREST TYPES OF THE AMAZON ESTUARY By Marina Londres August 2009 Chair: Karen Kainer Co-chair: Christina Staudhammer Major: Forest Resources and Conservation Carapa guianensis (andiroba) is one of the most promising Amazonian multiple-use tree species to foster conservation through sustainabl e management. While the oil extracted from its seeds is one of the most widely used natural remedies in the Amazon, the species has suffered high logging pressure because of its wood of superior quality. In the floodpl ain forests of the Amazon estuary where C. guianensis plays a crucial socioeconomic role, we addressed two critical questions about C. guianensis ecology: (1) What is the size class structure of C. guianensis within the floodplain forests by forest type? (2) What are C. guianensis seed production rates by forest type, and what main factors explain observe d production variation? The study was conducted in collaboration with lo cal forest managers within a Sustainable Development Reserve, where C. guianensis was identified as the most important species in their livelihood systems. C. guianensis population structure varied significantly across forest types, with drastic disparities in seedling, sap ling and adult densities. C. guianensis densities 10 cm diameter at breast height (dbh) were 28.7 (.3), 23.0 (.2) and 19.5 (.8) trees ha-1 in Baixio (periodically inundated swamp), Restinga (river-side periodically inundated forest) and Terra

PAGE 13

13 Preta (upland flooded forest), respectively. Seed production rates varied significantly by forest type, with an average ( standard deviation) of 2.6 (.4), 4.1 (.4) and 5.5 (.4) kg of viable seeds tree-1 year-1 in Baixio Restinga and Terra Preta forest types, respectively. Our final mixed model that explained C. guianensis seed production variation incl uded forest type, dbh, year, crown form, crown illumination and interactions as explanatory variable s. We found a quadratic relationship between dbh and seed production, and a marked variati on in seed production between the two years monitored. In addition, we found evidence that past timber harvests affected commercial timber stock and populatio n-level seed production. On the other hand, preliminary assessment of the to tal population seed production s uggests that, at current harvest levels, the local community is collecting less than 1% of the C. guianensis viable seed produced annually within its forest lands.

PAGE 14

14 CHAPTER 1 INTRODUCTION Carapa guianensis (andiroba) is one of the m ost important Amazonian timber and nontimber tree species. The oil extracted from its se eds has been used by local people medicinally for centuries, and recently has gained the attent ion of pharmaceutical and cosmetics companies. Nonetheless, in recent decades, the species has suffered from high pressure to log its valuable timber, mainly due to increas ed scarcity of mahogany ( Swietenia macrophylla) and Spanish cedar (Cedrela odorata) which belong to the same plant family. Despite its economic and cultural significance, there is a lack of research concerning C. guianensis ecology, especially in the flooded forests of the Amazon estuary where it is most prevalent and densities appear to be greatest. No study has quantified C. guianensis fruit production by forest type thus far. Furthermore, McHargue and Hartshorn (1983) reported that a C. guianensis population at La Selva, Costa Rica, produ ced good seed crops approximately every other year, suggesting th at realistic production estimates must include a wide cross-section of trees measured over a number of consecutive ye ars. In addition, very little is known about C. guianensis population structure across forest types (Klimas et al 2007). The first step in sustainable management of any given species is to determine its population structure and production potential (Peters 1994, 1996). When combin ed with growth or fruit production rates, demographic structure data provide the basis for management decisions (Peters 1994; Bruna & Kress 2002). In general, there is a lack of quantitative information on fruit production and sources of production variation of most tropical fruits, nuts a nd seeds of commercial interest (Ticktin 2004; Kainer et al. 2007), which is neces sary to estimate sustainable harvest levels. In Brazil, most efforts at sustainable forest management have focused on industrial-scale timber harvests. In contrast, little has been done to understand the ecology, management and

PAGE 15

15 production potential of key forest resources within a community context. This is surprising given that forest communities are important stakeholders in managing approximately 25% of the worlds tropical forests (White & Martin 2002). The Amazon estuary region has a long history of local forest use by riverine communities, spread throughout its river and st ream systems. In these areas, isolation and water regime dictate the dynamics of social and economic relations of people living within its fo rests (Queiroz et al. 2004). The rubber-based economy dominated this region for centuries, with some additional income from other non-timber forest products that were exported to Europe. With the end of the Amazonian rubber era in the mid 1950s, a new period began with important socioeconomic consequences: the intensification of timber and palm heart extr action (Oliveira 1991). Nowadays, the economy is centered on extraction of timber and aa ( Euterpe oleracea ) fruit and palm heart. In Gurup a county situated in the easte rn portion of the Amazon basin, in the estuary region a strong grass-roots movement emerged in 1980s and created the Gurup Rural Workers Union. Members of this movement engaged in a successful and wellknown struggle against timber enterprises that were trying to illegally di splace local people from their lands. In the early 1990s, this movement allied with an activist non-governmental or ganization FASE (Federation of Educational and Social Assist ance Organizations), and together they started a process of land titling. Through those efforts, the Itatup-Baqui Sustainable Development Reserve was created, along with other multiple-use reserves. The community So Joo do Jaburu, with an area of approximately 14,000 ha, is within the reserve and divided into parcels managed by appr oximately 50 families. Although they now have their land titled, community members are aware of the fact th at their current resource use

PAGE 16

16 patterns, particularly those base d on tree felling for timber and palm heart, have been generating serious negative impacts on the eco logy and productivity of the fo rest ecosystems on which their livelihoods depend. Consequently, with external NGO incentives, they engaged in a process to more effectively plan their forest management activities. In this process, the potential commercialization of NTFPs (non-timber forest products) arose as a promising alternative to timber and palm heart extraction that could impr ove livelihoods, while generating fewer negative impacts on the forest. In that context, C. guianensis seed oil garnered great community attention due to increasing market demand in th e region. A first step in understanding C. guianensis potential began with the communitys wome ns association, which started working on C. guianensis oil production and prepare d, with FASE support, the C. guianensis Oil Sustainable Management Plan. In 2005, I started collaborati ng with the So Joo do Jabur u community to explore the ecology of tree species that sustain local livelihoods. With the goal of jointly developing guidelines on best forest management practices, community members were engaged in all stages of the research process: setting research priorities, selecting research species (C. guianensis was voted most important), and mapping forest types and species distributions. A subset of the community (local researchers) had greater part icipation, acting as volunteers to collect data, participating in research skill-building activities, and disseminati ng results to their community. Community participation permitted a much larg er sample size of trees and continuous highquality data collection throughout th e year. This research started prior to my graduate studies when I was a professional researcher with the Amazonian Institute of People and the Environment (Imazon). Later, it became the basis of my graduate work.

PAGE 17

17 Through this collaboration, we addr essed questions that explored C. guianensis ecology, elucidating information critical for guiding su stainable management, while contributing to the broader ecological scientific understanding of tropical tree population structure and seed production. Chapter 2 is presented as an independent chapter for submission to a peer-reviewed journal. Relevant conclusions as well as current and future research strategies are presented in Chapter 3. Finally, because I worked in strong co llaboration with both local stakeholders and scientific mentors, in my thesis I opted to refer this research as our research.

PAGE 18

18 CHAPTER 2 POPULATION STRUCTURE AND SEED PRODUCTION OF Carapa guianensis IN THREE FLOODPL AIN FOREST TYPES OF THE AMAZON ESTUARY 2.1 Introduction The sustainable use of tropical forests is advocat ed as a prom ising strategy that reconciles biodiversity conservation, povert y alleviation and economic deve lopment (Dickinson et al. 1996; Bawa and Seidler 1998; Pearce et al. 2003). The Brazilian government is investing in forestbased socio-economic development: approximately 20 million hectares of sustainable use areas have been created since the 1980s; a recent C oncessions law (law number 11.284) provides a framework for enterprises and communities to ma nage public forests; and current forest policy calls for the creation of 50 milli on hectares of public production forests (SFB 2009). While these initiatives demonstrate political will to move toward sustainable use of this vast forest resource, sound ecological understanding to guide sustainable harvest of economically valuable forest species is often lacking. Carapa guianensis (andiroba) is an excellent model species for such a sustainable-use approach. Because of its broad geographic range, abil ity to thrive in diverse forest types, and its high timber and non-timber values, this tree play s a crucial socio-economic role throughout the Amazon basin. Sharing many wood characteris tics with its relative, bigleaf mahogany ( Swietenia macrophylla) C. guianensis populations are experiencing incr easing logging pressure in both industrial and smallholder contexts This species is also importan t for the medicinal oil extracted from its seeds: a powerful anti-inflammatory and perhaps the most widely used natural remedy in the Amazon (Shanley and Londres, In Press). Additionally, C. guianensis seed oil has been gaining economic relevance, as pharmaceutical and cosmetic companies increasingly purchase seeds and oil extracted by local commun ities. However, little information on C. guianensis population biology is available to guide sustainable management.

PAGE 19

19 To design sustainable systems for timber and non-timber products, species demographics such as growth, reproduction and regeneration need to be assessed on a population level, as well as across forest types. The size distribution of individual densities wi thin a tree species population (or population structure) can provide insight into recrui tment history, shade tolerance, and regeneration potential (Knight 1975; Hartsh orn 1980; Peters 1996). While no substitute for population monitoring, examination of population structures is a valuable first step towards understanding population dynamics. Plant reproduction is a key issu e in sustainability (Peters 199 4), being directly affected when the product harvested is fr uits, seeds or timber; harvest of the latter product affects reproduction because the largest adult trees are pr incipal seed suppliers. The great complexity and variation of fruiting pattern s displayed by tropical trees is a major stumbling block to sustainable resource exploitati on (Janzen 1978; Peters 1996). Ti ming, duration, and intensity of seed production vary substantially by species, aff ecting not only species-level recruitment, but also ecosystem-level dynamics because wildlife populations depend on seed crops (Herrera et al. 1998; Koenig and Knops 2000). Nonetheless, studies of fruiting patterns of tropical trees are still rare (Herrera et al 1998; Peters 1996). Within a sustainable-use rese rve in the floodplain forests ( vrzea ) of the Amazon estuary, we partnered with a local forest -dwelling community to explore C. guianensis ecology. We asked: (1) What is the size class structure of C. guianensis within the floodplain forests by forest type? (2) What are C. guianensis seed production rates by forest type, and what main factors explain observed pr oduction variation? 2.2 Study Species Carapa guianensis Aubl. (Meliaceae family) is a tree sp ecies attaining 30 to 50 m height at maturity. The species has a wide geographic distribution, occurring across the Amazon basin,

PAGE 20

20 Central America, Caribbean Islands and tr opical Africa (Pennington 1981, Plowden 2004). C. guianensis fruit is a four-valved woody capsule, with each valve containing one to four seeds (Pennington 1981; Sampaio 2000). Seeds germinate we ll in different habitats and their buoyancy aids dispersal and establishment in flooded forests such as riparian swamps and floodplains (McHargue and Hartshorn 1983). Indeed, although C. guianensis is found in different habitats such as highlands up to 1400 MASL and dry sites (Leite 1997; Sampaio 2000; Plowden 2004), it is predominant in lowlands and wet areas (L eite 1997; Pennington 1981; McHargue & Hartshorn 1983). Seed dispersal varies by habitat; in floode d forests, seeds are most commonly dispersed by water, while in highland forests, the nutrient-rich seeds are disper sed by scatter-hoarding rodents (McHargue and Hartshor n 1983). In the Brazilian Amazon, the flowering period has been documented to occur from August to Oc tober, and fruit fall be tween March and August (Pereira 1982; Leo and Carvalho 1998; Plowden 2004). 2.3 Study Site Our study was conducted in one of the nine sociopolitical co mmunities of the 64,000 hectare Itatup-Baqui Sustainable Developmen t Reserve (RDS), in Gurup County, Par State, Brazil. The region has a long history of lo cal forest use by riveri ne communities that are spread throughout its river a nd stream systems (Oliveira 1991). The community So Joo do Jaburu is located at Ilha Grande de Gurup, an island in th e mouth of the Amazon River. Gurup is located in the eastern portion of the Amazon basin in the estuary region, and vrzea forests cover most of its land area. These Amazonian vrzea forests arise on geological formations of the current Quaternary period, and their fluvial dynamics are infl uenced by periodic floods of sediment loaded, nutrient rich wh ite-water rivers (Prance 1979; Parolin et al. 2004). The highly dynamic landscape and topographic variation cause s different flooding amplitudes and durations along the flooding gradient, resulting in the formation of seve ral forest types (Kalliola et al.

PAGE 21

21 1991; Parolin et al. 2004; Wittmann et al. 2004). In the Amazon estuary, the flood pulse is linked to the sea tide, and a daily inundation occurs in addition to the annual flooding periodicity (Wittmann et al. 2004). The elevation throughout this re gion is less than 5 MASL and average annual rainfall is approximately 2.5 m (RADAM 1974). 2.4 Methods 2.4.1 Population Structure Forest inventories were conducted from Ja nuary 2005 to September 2006 to provide the following quantitative estim ates of C. guianensis populations in each forest type: (1) density; (2) size class structure; and (3) key individual tree ch aracteristics relevant to fruit production. As a first step, we identified four different forest types in the community of So Joo do JaburuBaixio Restinga, Igap and Terra Preta based on a previous part icipatory mapping exercise conducted with local residents. The main attributes used to differentiate each forest type were tidal influence, species dominance/composition, and edaphic conditions (Table 2-1; Appendix A). Based on this stratification and using the forest type map, six 1-ha (500x20 m) plots were randomly installed within each of the four identif ied forest types, totaling 24 ha inventoried. To minimize sample bias, spatially independent sa mple plots and fairly representative plot distribution across the vrzea landscape were criteria for plot s ite selection. Within each selected plot, we installed a 500-meter reference baseline a nd sampled 10 m on each side of this baseline. The following data were collected for all individuals 10 cm dbh (diameter at breast height): (1) dbh; (2) estimated height; (3) vine load ( 10% of crown covered; 10-6 0% of crown covered; or > 60% of crown covered); (4) crown illumination (no direct light; some overhead or side light; or full overhead light); (5) crown form (perfect complete circle and no major damage; good some damage but at least half-crown; or poor less than half-crown); and (6) canopy position (lower canopy; mid-canopy; upper canopy).

PAGE 22

22 To gather information about individuals < 10 cm dbh, from May to July 2008, we revisited transects in Baixio, Restinga and Terra Preta forests, since the i nventory revealed that C. guianensis only occurred in these 3 of the 4 forest types in So Joo do Jaburu. In each of these 18 transects, we sampled individu als < 2 cm dbh by installing ten nested 7x25 m subplots in each transect. We systematically placed one subplot every 50 meters and flipped a coin to ensure random subplot installation on one side of the tr ansect or the other. Within each subplot, we tagged all individuals between 50 cm height and 2 cm dbh, and recorded all six tree-level variables (as above). We also recorded these si x tree-level variables for all the individuals 2 cm dbh (including individuals 10 cm previously marked) and adde d a seventh forest phase variable (gap vertical hole to the canopy; building phase closed canopy le ss than the mean canopy height of mature forest; and mature forest closed canopy at the average canopy level of mature forest). 2.4.2 Fruit and Seed Counts A different sam pling scheme was used to investigate C. guianensis fruit production. Because floodplain forests of the estuary region ar e subjected to weekly high tides that invade the forest and carry seeds downr iver, we could not count fallen fruits to estimate production. Therefore, we used two sampling regimes in all thr ee forest types: (1) an extensive sample of the total population (507 trees) with monthly crown observations of fruit production; and (2) a smaller intensive sample of 49 fenced-off individuals in which fruit fall was counted bi-weekly. 2.4.2.1 Extensive sample In January and February 2007, we employed st ratified random sa mpling to install 21 line transects (seven transects in each forest type where C. guianensis was present). For each transect, the starting point and angle of or ientation were randomly selecte d, subjected to the constraint that transects had to remain enti rely within a single forest type. Thus, the angle of orientation

PAGE 23

23 was adapted when necessary. Each 50-meter inte rval of the transect baseline was marked to facilitate mapping, and all C. guianensis trees 10 cm dbh within a 100 m strip on each side of the baseline transect were candidates for inclusio n in the sample. To achieve relatively even spatial coverage of the forest type strata and approximately equal representation of stems in each size class, further sampling criteria were developed for each forest type. For example, for the Restinga forest type, based on size class proportions in inventory data, every eighth tree of size class 10 to 30 cm dbh was selected; every third tree of size class 30 to 45 cm was selected; and all trees > 45 cm were selected. Likewise, density variations dictated vari able transect lengths; transects varied from 200 to 800 m and number of trees sampled varied from 20 to 60 per transect. The diameter classes sampled we re 10 to 30, 30 to 45 and 45 to 60 cm dbh for Baixio and Restinga forest types; Terra Preta forest type included additional classes for individuals 60 to 90 cm and 90+, since logging had not eliminated these classes in Terra Preta as it had in Baixio and Restinga forest types. For the entire extens ive sample, we selected a minimum of 40 trees (and a maximum of 60) in each diameter class for each forest type. This method of selection allowed us to develop a model of indi vidual tree production that gave precise estimates over all size classes. For each of the 507 total trees, we assessed the six tree-level variables previously described and conducted monthly monitoring from March to September 2007 and from January to September 2008 to estimate fruit production. First, we estimated the total numbers of fruits in each tree crown by counting all fruits in the visi ble portion of the crown with binoculars. Then, we estimated the proportion of the crown counted to arrive at the extrapolated estimate of the total number of fruits in the entire crown.

PAGE 24

24 2.4.2.2 Intensive sample To verify ex tensive sample estimates, a subs ample of 32 trees was selected and monitored from March to September 2006 (pilot year); we increased the sample size to 49 in December 2006 to January 2007. These 49 trees were monitored from January to September of 2007 and 2008. We employed the same sampling criteria as in the extensive sample, but instead of using transects we randomly selected four to six search areas within each of the three forest types (controlling for fairly representa tive spatial distribution across th e study area), selecting the first three to six trees in each area that met monitori ng search requirements: (1) evidence of fruiting, and (2) a minimum distance of 100 m between selected trees to minimize genetic bias. We monitored 6 trees for each of the following size classes: 30 to 45 and 45 to 60 cm dbh for Baixio and Restinga forest types; and 30 to 45, 45 to 60, 60 to 90 and 90+ cm dbh for the Terra Preta forest type. Information on the six tree-level va riables collected for i nventories and extensive production sampling systems was also collected for the intensive sample. The ground below the crowns of the 49 selected fruiting adults was cleared of underbrush and fenced off using nylon fishing nets. Nets were secu red below-ground at close interval s to prevent seed escape and exclude seed predators (mostly rodents) from consuming seeds once fallen. Each sample tree was monitored bi-weekly to count fruit production and maintain fenc ing. In addition, the same observations of fruit production used on the extensive transects (cro wn fruit counts coupled with extrapolation of the visible pr oportion of the crown) were app lied to the intensive trees, providing a means of relating visual estimates from the intensive sample to those from the extensive method.

PAGE 25

25 2.4.3 Data Analyses 2.4.3.1 Population structure All statistical analyses were performed us ing SAS software (Version 9.2). To determ ine size class structure, trees 10 cm dbh were initially grouped into 10 cm diameter classes, saplings were grouped into a 2 to 10 cm dbh class and seedlings < 2 cm dbh were grouped together (all densities were translated into number per hectare). Next, descriptive analyses were employed to construct dbh histograms by forest type. Differences in population structures by forest type were examined through non-parame tric tests using the SAS procedure PROC LIFETEST (a procedure originally designed to compare survival curves). Wilcoxon and LogRank statistics were calculated to compare the frequency distributions of individuals 2 cm dbh across forest types. Using PROC GLIMMIX, a Poisson regression was performed to assess differences in seedling densities (< 2 cm dbh) across forest types. Chi-square tests of independence were performed to assess if the proportion of trees > 10 cm dbh occupying crown form, crown vine load, crown illumination, forest phase and canopy position was related to forest type. 2.4.3.2 Seed production Prelim inary analyses were performed with the intensive sample (49 fenced trees) to build a predictive equation that estimated seed production per tree. This equation related fruit counts in the crown (number of fruits) to seed counts on the ground (kilogram s of viable seeds per tree). We first summed the total number of viable seeds produced per tree per year for all trees within the intensive sample. For these same trees and ye ars, we then calculated the maximum number of fruits observed in the crown, and found a Pearsons correlation of 0.6 between fruit counts in the crown and seed counts on the ground (P < 0.0001).

PAGE 26

26 Because the study involved repeated measur es (each tree was observed in successive years), a mixed model was then constructed to serve as the predictiv e model (using the SAS procedure PROC MIXED), with total production on the ground (kilograms of viable seeds per tree per year) as a function of ma ximum number of fruits in the crown, year, forest type and the six tree-level variables collected (described in the methods). To ensure that model assumptions were met, natural log-transformations were performed on the response variable. Non-significant variables were dropped sequentially (based on P-va lues), and model results were compared using Akaikes Information Criteria (AIC; Akaike 1973). Model validity was checked through residual analyses. The final predictive model included only year and maximum number of fruits in the crown. To account for the fact that crown counts for both intensive and extensive monitoring started in different months for 2007 and 2008 (March for 2007 versus January for 2008), we estimated separate predictive models for each year: log(kg of viable seeds) = f(maximum number of fruits in the crown). Using the predictive model for each year, we then estimated the predicted log-transformed seed production for each tree in the extensive sa mple (507 trees). The predicted values were subsequently back-transformed to represent original units (kilogram s). Because the seed production model was derived fr om productive trees only, we excluded non-producing trees from the predictive extrapolation and then added them back as zeros to the predicted extensive data set. The predicted production in the extensive sample was then used to estimate a mixed model that related seed produc tion to tree and environmental variab les. Because the distance between trees was substantial in the vast majority of cases, each tree was considered an independent observation. Log-transformation wa s again necessary to meet m odel assumptions, and therefore

PAGE 27

27 we tested the full model using the natural log of viable seeds as the res ponse variable and the following as explanatory variables: year, forest type, the six tree-l evel variables described in the methods and their two-way interac tions. We employed the same mode l selection criteria used for the intensive sample. Finally, we back-transformed the least square means outputs to get original units of production estimates and re lated standard errors to forest type, dbh class, year, and tree variables. Where effects where significant (P < 0.05), we performed pair-wise comparison tests between levels of the effect, maintaining an overall significance level of = 0.05 by using Scheffes multiple comparison test. Since the production sampling was stratified, representing each dbh class equally, model predictions were ad justed when making inference to the entire population; model least square m eans outputs were generated at th e forest type true average dbh in the population to more accurately reflect the relationship between tree size and average seed production in the population. Finally, to extra polate production estimates from sample populations to total production per forest type, we multiplied forest type average production by tree density (trees per ha from i nventory data) and by the area of each forest type within the So Joo do Jaburu community. 2.5 Results 2.5.1 Population Structure Forest inventories revealed that densities of C. guianensis individuals 10 cm dbh were 28.7 (.3), 23.0 (.2) and 19.5 (.8) trees ha-1 in Baixio Restinga and Terra Preta forest types, respectively. Analyzing the shape of C. guianensis diameter distribu tion (all individuals 2 cm dbh), both Log-rank and Wilcoxon tests ev idenced differences by fo rest type (P < 0.0001), and pairwise tests revealed statistically differe nt diameter distributions between the following: Restinga vs. Terra Preta (P < 0.0001), Baixio vs. Terra Preta (P < 0.0001); and Baixio vs. Restinga (P = 0.0415). In addition, the Poisson regressi on indicated that seed ling densities (dbh <

PAGE 28

28 2 cm) differed dramatically. In Baixio we found 22.9 seedlings ha-1, which was significantly different from the other fo rest types (P < 0.0001); in Restinga we found 105 seedlings ha-1, which in turn was also significantly different from Terra Preta (151 seedlings ha-1; P = 0.035). These seedling densities represente d different proportions of the total number of individuals in the populations in each forest type as well: 27%, 71% and 84%, respectively. A visual analysis of the population structure 2 cm dbh (Figure 2-1) illustrates that Restinga and Baixio population structures tended to ha ve a greater proportion of small trees (2 to 20 cm dbh) versus large trees (20 to 40 and 40 cm dbh), and a gradual decrease in tree densities with increasing in size class. In contrast, Terra Preta displayed a flatter dbh distri bution, particularly above 30 cm dbh. Finally, in Baixio and Restinga, we found a lack of indivi duals > 50 cm dbh (0.3 and 0 trees ha-1, respectively), while in Terra Preta, we found 2.8 trees ha-1 in this class. 2.5.2 Population Structure vs. Forest Structure Chi-square tests of independen ce revealed that the distribu tion of individuals > 10 cm dbh by crown form crown vine load, crown illumination, forest phase and canopy position categories were dependent on forest type (P < 0.01). De scriptive statistics demonstrated that in Restinga 74% of trees > 30 cm dbh were in the perf ect crown category, compared to 55% in Baixio and 42% in Terra Preta. Moreover, Terra Preta presented the highest rates of vine infestation (Table 2-2): 42% of trees > 30cm dbh had 10 to 100% of their crowns covered by lianas, compared to 19% in Baixio and 12% in Restinga Also, for trees > 30 cm dbh, Baixio forest type displayed the highest proportion of trees in the full overhead light category (Table 2-2): 82% compared to 61% and 60% in Restinga and Terra Preta forest types, respectively. The overall proportional di stribution of trees among canopy position categories demonstrated that the vast majority of indivi duals > 30 cm dbh were in the mid-canopy (71% for all forest types combined). However, this patte rn masks considerable structural differences

PAGE 29

29 among forest types Terra Preta, having the tallest, most closed canopy, and Baixio the lowest and most open (results from overa ll forest inventory not presente d here). On one end of the spectrum, in Baixio C. guianensis trees > 30 cm dbh were equa lly divided between mid-canopy and upper canopy categories. In contrast, only 8% of trees > 30 cm dbh were in the upper canopy in Terra Preta. Indeed, in Terra Preta only trees > 40 cm dbh were found in the upper canopy. Additionally, trees at the 10 to 30 cm size class appeared to be fairly equally divided between canopy and sub-canopy categories in Baixio and Restinga forest types, while in Terra Preta 70% of trees at the 10 to 30 cm size cl ass were in the sub-canopy category. 2.5.3 Seed Production 2.5.3.1 Intensive sample predictive models The final 2007 and 2008 m ixed models reliably predicted measured ground seed counts for the 49 fenced trees from the estimated maximum number of fruits observed in their crowns (Table 2-3). The resultant predic tive equations were then applied to the extensive sample to predict seed production for this larger sample based on the 507 individual observed crown counts. 2.5.3.2 Extensive sample mixed model The final m ixed model of the 507 extensive sample trees indicated that multiple measured variables explained C. guianensis seed production (Table 2-4). The model demonstrated good fit with the majority of the sample population, but the residual analyses (Appendix B) evidenced lack of fit for trees that produced zero fruits (2 3% of the sample trees in either one or both years). A more complicated model, such as a Zero-altered Gamma (Feuerverger 1979), may be more appropriate for this data, but is beyond the scope of this research.

PAGE 30

30 2.5.3.3 Production rates Based on two-year m easurements of our 507-tr ee sample across three forest types, we found estimates ( standard deviation) of 2.6 (.4), 4.1 (.4) and 5.5 (.4) kg of viable seeds tree-1 year-1 in Baixio Restinga and Terra Preta forest types, respectiv ely, ranging from 0 to 192.5 kg of viable seeds tree-1 year-1 (Table 2-5). Descriptive statisti cs revealed that at least 10% of all trees produced no fruits in any given ye ar, while only 3% of the trees sampled did not produce in either year (Table 2-5). However, al l these consistent non-producers were relatively small individuals (< 30 cm dbh). In spite of the extremely high maximum production observed, only 0.6% and 2.6% of the trees sampled produc ed > 50 kg of viable seeds in 2007 and 2008, respectively. 2.5.3.4 Production variation Variation b y year and by forest type at the population level mean dbh: C. guianensis seed production was higher in 2008 than 2007 (Figur e 2-2). Nonetheless, the amplitude of annual variation was different by forest type. In Terra Preta, seed production was 63% higher in 2008 than in 2007, followed by 49% in Baixio and 30% in Restinga The average individual seed production adjust ed for the diameter distribution for each forest type revealed that, in 2007, Restinga forest type presented significantly higher mean individual seed production rates when compared to Baixio and Terra Preta (P = 0.012 and 0.0017, respectively), which were not signi ficantly different. However, in 2008, Terra Preta mean individual seed production was significantly greater than in Baixio (P = 0.028), but not significantly more than in Restinga Variation by dbh and by forest type at the individual level: Overall, we found a quadratic relationship between seed production and dbh (Figure 23). The model revealed that seed production increased as dbh increased, with an estimated peak in seed production at 82.6

PAGE 31

31 cm dbh, followed by a slight and gradual seed pr oduction decrease. The quadratic relationship, however, was based on observations in Terra Preta only, as this was the only forest type where trees > 60 cm dbh were sampled. All size classe s sampled produced fruits, suggesting that C. guianensis trees 10 cm dbh can be reproductively mature under closed forest. However, even though our data did not illustrate a clear threshol d of minimum production size, nearly a quarter of the trees in the 10 to 20 cm dbh size class did not produce in either year and in any given year approximately one half of these same-sized tr ees produced. In contra st, all trees > 30 cm dbh produced in at least one year, with 95 % producing in both years (Table 2-6). Seed production differences of comparable-sized individuals across fo rest types (Figure 23) indicated that at 10 and 20 cm dbh, Baixio had significantly lower individual production than Restinga and Terra Preta. At 30 cm dbh, Restinga had significantly higher production than Baixio and Terra Preta, which were not significantly di fferent. Above 40 cm dbh, however, Terra Preta had significantly lower individual production than Baixio and Restinga (Figure 2-3). Variation by crown form: No clear pattern of variation by crown form was detected: trees with perfect crowns produced 20% more seeds than trees with good crowns (P = 0.0008). Yet, production of trees with poor crow ns did not differ significantly from that of trees with good crowns. Variation by crown illumination: Trees with higher levels of crown illumination tended to have higher seed producti on. Individual trees exposed to full overhead light produced 18% more than those exposed to some overhead light (P = 0.0003), and 40% more than trees receiving no direct light (P = 0.05). 2.6 Discussion This is the first study to report C. guianensis population biology in the co ntext of the vrzea floodplain forests of the Amazon estuary. We explored C. guianensis population structure

PAGE 32

32 and seed production patterns in different forest ty pes, which in turn, were subjected to different harvest intensities. Overall, forest types presented significantly different C. guianensis population structures and rates of viable seed production. Furthermore, harvest history seems to have compromised not only availability of trees of commercial size, but also seed production potential. 2.6.1 Population Structure 2.6.1.1 Densities The C. guianensis tree d ensities found in all three vrzea forest types were much higher than those reported for upland forests, part icularly in the eastern Amazon, and roughly comparable to those in other flooded fore sts (Table 2-7). Based on these findings, C. guianensis seems to be generally a dominant species; the pa ttern of higher C. guianensis densities in floodplain ( vrzea ) forests than upland ( terra firme ) forests mirrors the broader pattern of lower tree diversity and higher species dominance in vrzea forests (Prance 1979; Anderson 1991). 2.6.1.2 Size-class distribution Both descrip tive analyses and statistical tests demonstrated that C. guianensis dbh distributions differed significantl y among forest types, with mark ed disparities in seedling and sapling size classes. These findings imply that (1) cumulatively, germination and early seedling survival, at least in recent years, has been substantially higher in Terra Preta and Restinga than in Baixio but (2) a high proportion of these seedli ngs do not seem to reach the sapling size classes in Terra Preta and Restinga forest types (Figur e 2-1). A plausible explanation for this disparity is that Baixio is subjected to daily tidal influe nces during the rainy season which coincides with the C. guianensis fruit production period and receding waters carry buoyant C. guianensis seeds downriver. Although the tide also invades the Restinga forest type, it is less frequent than in Baixio Terra Preta is not directly affected by the tide, and consequently, C.

PAGE 33

33 guianensis seeds are not taken away with the water. Alternative hypotheses for the lower seedling densities observed in Baixio include: (1) Early survival is lower for Baixio ; that is, more seedlings die in Baixio before reaching 50 cm height, perhaps killed by rainy season floods; (2) Baixio presents fewer suitable microsites for C. guianensis germination, since the Baixio forest floor is composed of a mosaic of higher patche s where woody plants establish, and lower patches consisting of a muddy substrate, composed onl y of hydrophilic plants and no woody regeneration (Londres, personal observation). As for disparities in transitions from seedling to sapling densities, Janzens (1970) density dependent mortality hypothesis co uld explain differences among forest types. For example, if seedling survival in Baixio (where seedling densities were low) was better than in Restinga and Terra Preta (where seedling densities we re high), Janzens theory hypothesizes that seedlings in these latter forest types shoul d experience greater mortality an d thus result in the observed comparatively lower sapling densities than in Baixio However, we would expect density dependent mortality to influen ce young seedlings (< 50 cm height not measured in this study) more than well-established ones (50 cm height to 2 cm dbh). Another hypothesis could be that seedling growth rates vary by forest type. Given the taller and more closed canopy in Terra Preta resulting in lower unde rstory light levels, we might exp ect slower growth and recruitment rates than in the other forest types. In any case, it will only be possible to understand the dynamics of germination, survival and recr uitment through cross-ye ar population dynamics monitoring. In the other extreme of the si ze-class spectrum, the study reve aled a lack of individuals > 50 cm dbh in Baixio and Restinga forest types. We attribute this difference to human activity rather than a biological response to forest type. Local extractivi sts report that they have been

PAGE 34

34 harvesting C. guianensis for timber more intensively in Baixio and Restinga forest types over the past three decades. While C. guianensis is also logged in Terra Preta, this forest type is less intensively exploited because of the distance to the river the pr imary means of log transport. Indeed, these differences in harvest intens ities in our study site complicate accurate understanding of C. guianensis population processes in different forest types, since harvest history will not only affect the largest size classes but the en tire population structure. Despite population structure differences among forest types, the general shape of the dbh distribution displayed an approximate reverse-J pattern (with the exception of seedlings in Baixio ). This general pattern, in cluding the observed relativel y high seedling and sapling densities and the high proportion of these indivi duals found in low light conditions (see crown illumination categories, Table 2-2) suggest that C. guianensis is a shade tolerant species. In different geographic regions, other C. guianensis studies reported re verse-J dbh distribution patterns, such as Klimas et al. (2007) in th e Brazilian Western Amazon (Acre state) and Henriques and Souza (1989) in terra firme forests at the other Amazonian longitudinal extreme (Maranho state). C. guianensis ability to survive in the forest understory has been attributed to its large seeds that, like many other large-seeded tropical rain forest species, typically lack dormancy. Other studies also have reported that C. guianensis readily establishes and grows under closed canopy (McHargue and Hartshor n 1983; Scarano 2003; Klimas 2007). A negative exponential distribution is typical of species with abundant regeneration when they are capable of germinati on, establishment and survival under mature forest canopy (Knight 1975; Hartshorn 1980; Peters 1996). However, by relating growth and survival rates with the size-class distribution of 216 tr ee populations, Condit et al. ( 1998) concluded that static information about the size dist ribution may not be a good predicto r of future population trends.

PAGE 35

35 This again illustrates that a complete demographic study of C. guianensis, including cross-year monitoring of growth and mortalit y of all size-classes, would be necessary to attain an accurate understanding of the species popula tion dynamics in a given site. Finally, the relative proportions of individuals within different crown form, crown vine load and canopy position categories reflected st ructural differences among forest types. Terra Preta displayed the lowest propor tion of adult trees > 30 cm dbh in good and perfect crown categories and the highest rates of trees with vine infestation. Moreover, the vast majority of trees in Terra Preta occupied lower canopy positions, when compared to trees in the other two forest types. Complete forest inventories from this site (Lon dres, unpublished) have shown that of the four major forest types, Terra Preta has the tallest and most closed canopy. 2.6.2 Seed Production 2.6.2.1 Production rates Few studies have exam ined C. guianensis seed production and those that have are based on relatively small sample sizes. The most cited C. guianensis seed production study in the literature is based on a 5-tree sample at La Selva, Co sta Rica (McHargue and Hartshorn 1983), where the authors reported an aver age of 80.2 kg of seeds tree-1 year-1. Based on a 10-tree sample size, Ranklin (1978) estimated an average of 31 to 58 kg tree-1 year-1 in Trinidad, and in terra firme forests of the eastern Amazon, Plowden ( 2004) reported an average of 1.2 kg tree-1 year-1 (n = 46). Based on a substantially higher sample size (n = 507), our study revealed averages ranging from 2.6 to 5.5 kg of viable seeds tree-1 year-1. This is the first C. guianensis production study conducted on the floodplain ( vrzea ) forests of the Amazon estuary, with the unique characteristic of tidal influence. Nonethele ss, the comparison of production rates between our study and others conducted in different sites is very likely confounded by the vast disparity in sampling strategies, particularly represen tation of individual trees across each C. guianensis

PAGE 36

36 population. Our sample comprises records on hundreds of individuals under multiple circumstances, including dbh size, forest type and other tree and site conditions. For that reason, we believe that this study pr ovides a reasonable understanding of the species seed production patterns at that site over the timeframe of the i nvestigation. We suggest th at cross-year studies with better population representation would be necessary to understand how C. guianensis seed production patterns vary geographically. 2.6.2.2 Production variation C. guianensis seed production varied m arkedly be tween the two cons ecutive years of investigation. We found, on aver age, a twofold difference in C. guianensis individual seed production from one year to anothe r. This pattern is consistent with the suggestion of McHargue and Hartshorn (1983) that C. guianensis produces good seed crops ev ery other year. Marked annual variation in individual fruit crops is not an uncommon phenomen on. Studies describing annual variation of multiple species found eviden ce that most polycarpic woody plants seem to adhere to alternating supra-a nnual schedules consisting of ei ther high or low production years (Herrera et al. 1998; Koenig and Knops 2000). Ther e is a wide range of proposed explanations for this temporal variation in fruit production of tree species. Among th em, the most common are (1) responses to environmental variables such as rainfall and weather conditions (the resource matching hypothesis) (Kelly 1994), (2) the economy of scale hypothesis (that large reproductive efforts are more efficient than small ones) (J anzen 1978), and (3) switching resources hypothesis (where in successive years plants move resources into rather than away from reproduction [i.e., good years of growth are bad for reproduction and vice-versa]) (Norton and Kelly 1988). However, given that our study only monitored trees for two years, we cannot speculate how and why the temporal fluctuations in C. guianensis seed production patterns took place. Further investigation (including several mo re years of seed production mon itoring) will be crucial to gain

PAGE 37

37 better insights into C. guianensis seed production long-term tem poral fluctuation patterns, as well as of its causes and the consequent effects on the population. The strong correlation between C. guianensis individual seed production and dbh observed in our study seems biol ogically reasonable. Seed produc tion increases uniformly with diameter, reaching an optimal size range of pr oduction, followed by a slight decrease at the largest size class as trees approach senescence (Harper and White 1974). It is possible that in smaller dbh sizes the trees are investing more resources in growth than in reproduction; moreover, it is common that smaller individual s find themselves competitively disadvantaged, especially for light resources. Insufficient studi es on fruit production of tropical trees make it difficult to adequately unders tand the degree to which these patterns vary among species. However, one study on a C. guianensis relative ( Swietenia macrophylla Meliaceae) (Grogan and Galvo 2006) found in one study site that mean annual fruit production by trees > 60 cm dbh was significantly higher than by trees 30 to 60 cm dbh, while in anothe r study site, trees > 90 cm dbh had the lowest observed produc tion rates. Another study on S. macrophylla (Snook et al. 2004) reported that fruit production incr eased with dbh but without a produc tion decrease at the largest dbh classes. On the other hand, a production study of Bertholletia excelsa (Kainer et al. 2007) reported a quadratic curve between nut produc tion and dbh, which is consistent with the C. guianensis dbh vs. production curve found in this study (which is appl icable only to one forest type population with large adults, Terra Preta). Our data illustrated that in all three forest types studied, C. guianensis can be reproductively mature at 10 cm dbh; this finding is consistent neither with Klimas et al (2007), nor with Cloutier at al. (2007) who reported a C. guianensis minimum production size of 20 cm dbh, and 30 cm dbh, respectively.

PAGE 38

38 Site conditions, including nutrient availability, can influence the rate of resource accumulation by long lived plants and thus can change their patterns of reproductive output (Harper and White 1974). Although forest type was a significant variable in explaining seed production in our model, the patterns of vari ation observed at the popul ation level were not consistent between years. One possible explanati on for this inconsistency is the extremely high variation in seed production betw een years, especially for the Terra Preta forest type, which resulted in different relative seed production variation by forest type in a given year. These patterns suggest that C. guianensis seed production varia tion by year might be stronger than the variation by forest type. This fi nding is consistent with Janzen (1978), who reported that seed production by a species of tropical tree is subject ed to a good deal of between habitat variation, often inconsistent from year to year. Terra Preta was the forest type with the overall highest mean seed production per tree when taking into account its averag e size (i.e., production calculated at average dbh; Table 2-5). This top ra nking could be explained by the differences in population structures: Terra Preta had the highest proportion of trees > 45 cm dbh, the size classes with the highest production rates. On the other hand, when looking at differences in seed production of comparable si ze classes by forest type, Terra Preta displayed substantially lower individual production rates for a dult trees > 40 cm dbh (Figure 23). These findings imply that logging activities compromised seed production in Baixio and Restinga ; that is, if the larger trees were not harvested, these two forest types w ould probably display the highest mean seed production rates. We found a significant and positive relations hip between crown illumination and seed production. Trees that receive more sunlight tended to produce more fruits. On the other hand, vine infestation did not display a significant effect in our mode l, whereas other studies have

PAGE 39

39 found vines to negatively affect fruit production (Stevens 1987; Kainer et al. 2007). A possible explanation could be that there are relatively few trees in the population with heavy vine loads and a large fraction of these are large trees, which in turn displa yed the best production rates. In addition, even though our model revealed that ther e was a significant effect of crown form on individual seed production, the natu re of this relationship was not as expected. We observed that trees with perfect crowns did not produce significantly more than trees with poor crowns. Other factors influencing seed producti on (whether measured in our study or not) might be obscuring the real influence of crown form in seed production. For Bertholletia excelsa Wadt et al. (2005) found a positive significant relationship between crown form and Brazil nut production, but crown form was clearly more important in explaining variance of very large trees ( 100 cm dbh) in their study. Further analyses would be appropriate to understand better if, and under what circumstances, crown form influences C. guianensis seed production. 2.6.3 Management Implications Unlike m any other economically valuable tropical tree species, C. guianensis exhibits a set of ecological characteristics that imply great potential for sust ainable harvest: high densities, ability to thrive in diverse forest types, am ple geographic distribution, annual seed production, shade tolerance, and a general tr end to reverse-J size distribution. However, careful consideration of trade-offs and interactions between timbe r and seed production is needed when planning harvests. Likewise, species population dynamics in different forest type s must be considered when designing a landscape-scale mana gement plan. Our study suggests that C. guianensis regeneration and establis hment success varies across forest type s, and that harvest histories have affected both seed production po tential and population structures. Community planning of C. guianensis seed harvest levels and oil production targets must be consistent with population-level seed produc tion. For example, by multip lying the density of

PAGE 40

40 C. guianensis adult trees in each forest type by the ar ea of each forest type (ha) in any given community and by the average seed production tree-1 year-1, we can estimate the total C. guianensis seed production in community lands. In So Joo do Jaburu, production was estimated at 852,303 kg of viable seeds year-1 (Table 2-8). According to a detailed study of C. guianensis seed collection in the same community (L ondres 2004), all of the families together harvested 949 kg of viable seeds in one year; that is, 0.1% of the total estimated annual seed production in the area. This disparity between production and harvest levels strongly suggests that the community could increase seed harvests substantially, since current harvest levels represent a very small fraction of the total C. guianensis population production within the community forest lands. Nevertheless, by eliminating the size classes most productive for seeds, the intensive logging in Baixio and Restinga seems to have reduced the population level fruit production substantially. Even though Terra Preta displayed the lowest produc tion rates at the individual level, the average population produc tion was higher, due to higher densities of large trees. The observed higher mean population production in Terra Preta along with the fact that seedlings in that forest type outnumber trees by a factor of 8 (indicating that most esta blished seedlings do not survive to larger size classe s), implies that an increase in seed harvests would not seriously compromise population persistence. On the othe r hand, the lower observed seedling densities in Baixio may indicate greater vulnerability to seed harv ests. This would suggest that restrictions on C. guianensis seed harvests and investment in enrichment planting in the Baixio forest type merit consideration. Restinga study populations fell in the mi ddle of this population structure gradient, with adult densities and thereby seed production compromised by logging activity, but with abundant advanced regenera tion. Although these differences in static population structure

PAGE 41

41 hint at differences in dynamics, there is a n eed to investigate the underlying causes of the disparities in seedling and sapling densities among forest types. Likewise, we would need growth and mortality data to predict sustainable levels of C. guianensis timber extraction, but even an analys is of the static population structures demonstrates that C. guianensis timber harvests should be pla nned more carefully. The paucity of commercial-sized C. guianensis individuals in Baixio and Restinga demonstrates some vulnerability to intense harvests, even t hough the species possesse s a set of ecological characteristic that allow for sustainable mana gement. We recommend a pause in logging in these forest types to allow population recovery, particularly as s eed oil production appears set to rival or surpass logging as an income-g enerating activity in the study region. The factors influencing seed production vari ation provide some interesting insights concerning C. guianensis seed harvests. The variation by dbh showed that C. guianensis displayed an optimum size of seed production, ab ove 45 cm dbh. Concentrated collection from these trees might increase coll ection efficiency. This size range is also optimal for timber harvests, again highlighting the complexity of designing a sustainable multiple-use management system. The positive and significant effect of li ght intensity in seed production suggests that liberation thinning of trees in key collection ar eas could increase NTFP management potential. Finally, population seed produc tion showed marked annual va riation, which implies that managers may need to adjust harvest targets by year. Nonetheless, th e observed low proportion of seeds harvested in relation to total populati on production strongly sugges ts that sustainable harvest thresholds could be much higher than cu rrent harvest levels wit hout negatively affecting population persistence.

PAGE 42

42 Table 2-1. Description of four main forest types at So Joo do Jaburu, Gurup, Par, Brazil. Forest type Inundation pattern Tidal influence Location of forest zone with respect to the river Harvest history Baixio Periodic and seasonal Almost daily during the rainy season 30 to50 m from the sides of the river, drained by a dense stream system Intensively harvested for timber Restinga Periodic and seasonal During the full moon of the rainy season At riverside Intensively harvested for timber Igap Periodic and seasonal Flooded nine months a year, with water level varying with tide 80 to 3000 m from the sides of the river Intensively harvested for timber and palm heart Terra Preta Seasonal No direct tidal influence Starts 500 m upstream from main stem of Jaburu river Lightly harvested for timber, intensively harvested for palm heart

PAGE 43

43 0.0 20.0 40.0 60.0 80.0 100.0 120.0 140.0 160.0 < 22 10 10 2020 3030 4040 5050 6060 100Density (trees ha 1)Diameter class (cm) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 10 2020 3030 4040 5050 6060 100Density (trees ha 1)Diameter class (cm) Baixio Restinga Terra Preta Figure 2-1. Size class distribution of 1105 C. guianensis trees, by forest type. The inset figure represents the same data, but enhances visu al differences between the larger diameter classes by forest type.

PAGE 44

44 Table 2-2. Descriptive results of the characterization of 1105 C. guianensis trees, according to diameter classes. Forest type Diameter class N* Density (trees ha-1) Crown forma (%) Crown vine load b (%) Crown illumination c (%) Forest phase d (%) Canopy position e (%) Per G Po 10 10-60 >60 F S N M B G Up Mid Sub Baixio < 2 cm 24 22.9 58 38 4 100 0 0 0 50 50 42 58 0 2-10 cm 200 33.3 48 43 9 86 11 3 1 44 55 38 57 6 2 7 91 10-30 cm 150 25.0 41 50 9 79 17 5 28 54 18 34 62 4 4 51 45 > 30 cm 22 3.7 55 45 0 82 14 5 82 18 0 41 55 5 50 50 0 Total 396 84.9 47 46 8 84 12 4 16 47 37 37 59 5 7 34 59 Restinga < 2 cm 111 105.7 70 26 4 98 2 0 6 22 72 42 54 4 2-10 cm 126 21.0 46 46 8 85 13 2 3 33 63 46 48 6 0 5 95 10-30 cm 107 17.8 49 39 12 83 12 5 23 56 22 33 55 12 3 58 40 > 30 cm 31 5.2 74 23 3 87 6 6 61 35 3 35 61 3 32 61 6 Total 375 149.7 56 36 7 89 9 3 14 36 49 40 53 7 6 39 55 Terra Preta < 2 cm 159 151.4 42 47 11 98 2 0 2 45 53 56 43 1 2-10 cm 58 9.7 53 37 11 96 4 0 4 39 58 61 37 2 0 5 95 10-30 cm 67 11.2 58 30 12 79 15 6 15 63 22 54 42 4 0 30 70 > 30 cm 50 8.3 42 40 18 58 30 12 60 36 4 58 40 2 8 86 6 Total 334 180.6 47 41 12 88 9 3 14 46 40 57 41 2 3 42 55 All Forest Types < 2 cm 294 123.7 54 38 8 98 2 0 3 37 60 50 49 2 2-10 cm 384 25.7 48 43 9 87 10 2 2 40 58 44 51 5 1 6 93 10-30 cm 324 19.7 47 42 11 80 15 5 24 56 20 38 56 7 3 49 48 > 30 cm 103 6.4 54 36 10 72 19 9 65 32 3 48 50 3 24 71 5 Total 1105 103.8 50 41 9 87 10 3 15 43 42 44 52 5 6 38 57 Individuals < 2 cm dbh are sub-sampled within inventory plots. a Crown form: Per, perfect comp lete circle; G, good half-crown; Po, poor less than half-crown. b Percentage of the crown covered with vines. c Crown illumination: N, no direct light; S, some overhead or sidelight; F, full overhead light. d Forest phase: M, mature phase; B, building phase; G, canopy gap. e Canopy position: Up, upper-canopy; Mid, mid-canopy; Sub, sub-canopy.

PAGE 45

45 Table 2-3. 2007 and 2008 model coefficients (f rom the intensive samp le). The dependent variable of each model is the log of total kg of viable seed production tree-1 year-1. Year Effect Estimate Standard error DF t Value Pr > |t| 2007 Intercept 1.194 0.205 43 5.82 <.0001 max_fruits* 0.051 0.012 43 4.27 0.0001 2008 Intercept 0.823 0.271 47 3.04 0.0038 max_fruits* 0.047 0.008 47 5.69 <.0001 *Maximum number of fruits estimated from crown counts observations. Table 2-4. Explanatory variables included in the final model (ext ensive sample). The dependent variable of the model is the log of total kg of viable seed production tree-1 year-1. Variable DF F value Pr > F Forest type 2 7.33 0.0007 Dbh 1 141.01 <.0001 Dbh2 1 21.56 <.0001 Crown form 2 7.29 0.0008 Crown illumination 2 7.08 0.0009 Year 1 139.68 <.0001 Forest type x year 2 13.06 <.0001 Forest type x dbh 2 9.42 <.0001 Table 2-5. Mean ( standard deviation) values of viable seed production per C. guianensis tree by forest type and percenta ge of non-producer trees. Forest Type Viable seeds (kg)* Viable seeds 2007 year (kg)* Viable seeds 2008 year (kg)* Nonproducers 2007 (%) Nonproducers 2008 (%) Consistent non producers (%) Baixio 2.6 0.4 1.8 0.4 3.6 0.4 18 15 7 Restinga 4.1 0.4 3.4 0.4 4.9 0.4 5 15 2 Terra Preta 5.5 0.4 3.3 0.4 8.8 0.4 15 3 1 All 13 10 3 *Average production of each forest type based on average dbh in population structure sample.

PAGE 46

46 0 1 2 3 4 5 6 7 8 9 10 BaixioRestingaTerra PretaAverage Kg of viable seeds tree 1Forest Type 2007 2008 Figure 2-2. Mean ( standard error) C. guianensis individual seed produc tion by forest type and year. Estimates were adjusted to represent the average dbh of C. guianensis populations at each forest type (based on inventory data). Table 2-6. Percentage of non-producers by size class. Size-class Non-producers 2007 (%) Non-producers 2008 (%) Consistent nonproducers (%) 10 20 42 61 23 20 30 19 5 1 30 5 0 0 Table 2-7. C. guianensis densities at diffe rent study sites. C. guianensis density Forest type Region Source 28.9, 23.0 and 19.5* Floodplain forests ( Baixio Restinga and Terra Preta forest types, respectively) Amazon estuary Londres (this study) 25.7 and 14.6* occasionally inundated and in terra firme respectively Western Amazonia Klimas et al. (2007) 6.7 and 5.6* occasio nally inundated and in terra firme respectively Eastern Amazonia Plowden (2004) 62* Occasionally inundated Costa Rica McHargue and Hartshorn (1983) 2.5** Terra firme Eastern Amazonia Cloutier et al. (2007) trees > 10 cm dbh ha-1. ** trees > 30 cm dbh ha-1. P < 0.0001 P = 0.04 P < 0.0001

PAGE 47

47 0 2 4 6 8 10 12 14 16 18 20 22 24 102030405060708090100Average Kg of viable seeds tree 1Diameter (cm) Baixio Restinga Terra Preta Figure 2-3. Estimated mean ( standard error) C. guianensis individual seed production by dbh and by forest type. The letters represent si gnificant differences in seed production among forest types at =0.05 (Scheffes multiple comparison test). Table 2-8. C. guianensis estimated total population seed prod uction by forest t ype and by year. Forest type Density* (trees ha-1) Area (ha) Estimated number of trees >10 cm dbh Mean production* (kg viable seeds tree-1) Estimated population production (kg of viable seeds) Baixio 28.7 4190 120,250 2.6 312,650 Restinga 23.0 1758 40,441 4.1 165,807 Terra Preta 19.5 3486 67,972 5.5 373,846 Total 852,303 *Individuals 10 cm dbh. a b b a b b a a b a b a a b a a b a

PAGE 48

48 CHAPTER 3 CONCLUSION Despite the increasing attention paid to sustai nable use of tropical fore sts for timber and NTFP resources, ecological understanding of the key commercial tree species is very limited. The objective of our stud y was to investigate C. guianensis population stru cture and seed production patterns in three distinct fo rest types of the flooded forests ( vrzea ) of the Amazon estuary region. We have found that both population structures and fruit production vary by forest type, suggesting that management decisions, as well as further ecological research on C. guianensis should acknowledge habitat variation, especially throughout the vrzea landscape of the Amazon estuary. Yet, we also found evidence that, to a certain extent, some differences in C. guianensis ecological variables measured across forest types are consequences of logging legacies. In addition, our analysis of the factors explaining C. guianensis seed production va riation provided interesting insights, such as the marked vari ation by year, the optimum dbh size for seed production and the increased seed production by crown illumination. Although we still need more complete population monitoring to design sustainable management systems, the results presented provide a valuable first step in guidi ng sustainable management. While timber harvests appeared to have had dr astic consequences for C. guianensis populations, we found evidence that even a substantial increase in C. guianensis seed harvests will not compromise population persistence. Finally, as previously state d, a long term population dynamics study is needed not only to better understand C. guianensis population trends, but also to pr ecisely assess sustainable harvest limits. C. guianensis population monitoring is currentl y underway. At the same study site (encompassing the three forest types) and usi ng our permanent samples, we are currently

PAGE 49

49 conducting the third year of fruit production meas urements and the second year of population monitoring (including growth and mortality data). In collaboration with our local research partners at the So Joo do Jaburu co mmunity, we plan to continue this C. guianensis population dynamics investigation until at least 2012.

PAGE 50

50 APPENDIX A FOREST TYPE MAP Figure A-1. Forest type map of the community of So Joo do Jaburu, Gurup, Par state, Brazil. The map was produced by local resi dents in participatory mapping workshops and combined satellite imagery interpreta tion (Radar Jers-1 a nd Landsat 7) with traditional ecological knowledge.

PAGE 51

51 APPENDIX B DISTRIBUTION OF MODEL RESIDUALS Residual-3 -2 -1 0 1 2 3 Predicted-1 0 1 2 3 4 Figure B-1. Extensive sample mi xed model residual values vers us predicted values. Linear trend in some residuals indicates modera te lack of fit in non-producing trees.

PAGE 52

52 LIST OF REFERENCES Akaike, H. 1973. Information theory as an extens ion of the m aximum likelihood principle. Pages 267 in B. N. Pretrov, Csaki, F.F., editor. Proceedings of the Second International Symposium on Information Theory. Ak ademiai Kiado, Budapest, Hungary. Anderson, A. 1991. Forest management strategies by rural inhabitants in the Amazon estuary Pages 351-360 in A. Gomes Pompa, T. C. Whitm ore, and M. Hadley, editors. Rain Forest Regeneration and Management Unesco and The Parthenon Publishing Group, Paris. Bawa, K. S., and R. Seidler. 1998. Natural Forest Management and Conservation of Biodiversity in Tropical Forests. Conservation Biology 12:46-55. Bruna, E. M., and W. J. Kress. 2002. Habitat Fr agmentation and the Dem ographic Structure of an Amazonian Understory Herb (Heliconia acuminata ). The Journal of the Society for Conservation Biology 16:1256-1266. Cloutier, D., M. Kanashiro, A. Y. Ciampi, a nd D. J. Schoen. 2007. Impact of selective logging on inbreeding and gene dispersal in an Amazonian tree population of Carapa guianensis Aubl. Molecular Ecology 16:797-809. Condit, R., R. Sukumar, S. P. Hubbell, and R. B. Foster. 1998. Predicting population trends from size distributions: A direct test in a tr opical tree community. American Naturalist 152:495-509. Dickinson, M. B., J. C. Dickinson, and F. E. Putz. 1996. Natural forest management as a conservation tool in the tropics: divergent views on possibilities and alternatives. Commonwealth Forestry Review 75:309-315. Feuerverger, A. 1979. On Some Methods of Anal ysis for Weather Experiments. Biometrika 66:655-658. Grogan, J., and J. Galvo. 2006. Factors Limitin g Post-logging Seedling Regeneration by Bigleaf Mahogany ( Swietenia macrophylla ) in Southeastern Amazonia, Brazil, and Implications for Sustainable Management. Biotropica 38:219-228. Harper, J. L., and J. White. 1974. The Demography of Plants. Annual Review of Ecology and Systematics 5:419-463. Hartshorn, G. S. 1980. Neot ropical Forest Dynami cs. Biotropica 12:23-30. Henriques, R. P. B., and E. Sousa. 1989. Popul ation-Structure, Dispersion and Microhabitat Regeneration of Carapa-Guianensis in Northeastern Brazil. Biotropica 21:204-209. Herrera, C. M., P. Jordano, J. Guitian, and A. Traveset. 1998. Annual Variability in Seed Production by Woody Plants and the Masting Concept: Reassessment of Principles and Relationship to Pollination and Seed Disp ersal. The American Naturalist 152:576-594.

PAGE 53

53 Janzen, D. H. 1970. Herbivores and the Number of Tree Species in Tropical Forests. The American Natura list 104:501-528. Janzen, D. H. 1978. Seeding patterns of tropical trees. Pages 83-128 in P. B. Tomlinson and M. H. Zimmerman, editors. Tropical trees as livi ng systems. Cambridge University Press, Cambridge, UK. Kainer, K. A., L. H. O. Wadt, and C. L. St audhammer. 2007. Explaining va riation in Brazil nut fruit production. Forest Ecol ogy and Management 250:244-255. Kalliola, R., J. Salo, M. Puhakka, and M. Rajasilta. 1991. New Site Formation and Colonizing Vegetation in Primary Succession on the Western Amazon Floodplains. Journal of Ecology 79:877-901. Kelly, D. 1994. The evolutionary ecology of ma st seeding. Trends in Ecology & Evolution 9:465-470. Klimas, C. A., K. A. Kainer, and L. H. O. Wadt. 2007. Popul ation structure of Carapa guianensis in two forest types in the southweste rn Brazilian Amazon. Forest Ecology and Management 250:256-265. Knight, D. H. 1975. A Phytosociological Analysis of Species-Rich Tropical Forest on Barro Colorado Island, Panama. Ecol ogical Monographs 45:259-284. Koenig, W. D., and J. M. H. Knops. 2000. Patte rns of Annual Seed Production by Northern Hemisphere Trees: A Global Perspective. The American Naturalist 155:59-69. Leo, N. V. M., and J. O. P. Carvalho. 1998. Fe nologia Reprodutiva de 25 Especies Arbreas da Amaznia. EMBRAPA. Leite, M. C. L. 1997. Ecologia de Carapa guianensis Aublet (Meliaceae) Andiroba. UFPA/Museu Paraense Emilio Goeldi, Belm, PA, Brazil. Londres, M. 2004. Diagnstico Participativo do Sistema Produtivo da Andiroba ( Carapa sp.) e Estudo Botnico da Espcie, nas Comunida des So Joo do Jaburu e Fortaleza do Jaburu, Municpio de Gurup, PA. Universi dade de So Paulo, Piracicaba. McHargue, L. A., and G. S. Hartshorn. 1983. Seed and Seedling Ecology of Carapa guianensis. Turrialba 33:399-404. Norton, D. A., and D. Kelly. 1988. Mast Seeding Over 33 Years by Dacrydium cupressinum Lamb. (rimu) (Podocarpaceae) in New Zealand: The Importance of Economies of Scale. Functional Ecology 2:399-408. Oliveira, P. H. B. 1991. Ribeiri nhos e Roceiros: gnese, subordin ao e resistncia camponesa em Gurup, Par. Universidade de So Paulo, So Paulo.

PAGE 54

54 Parolin, P., L. Valle Ferreira, A. L. K. M. Albernaz, and S. S. Almeida. 2004. Tree species distribution in vrzea forests of Brazilia n Amazonia. Folia Geobotanica 39:371-383. Pearce, D., F. E. Putz, and J. K. Vanclay. 2003. Sustainable forestry in th e tropics: panacea or folly? Forest Ecology and Management 172:229-247. Pennington, T. D. 1981. Meliaceae. Pages 470 in Flor a Neotropica, New York Botanical Garden Pr. Department, New York. Pereira, A. P. 1982. Ensaio em viveiro florestal e frutificao de algumas espcies Amaznicas. Pages 1135-1138 in Anais do Congresso Nacional sobre Essencias Nativas, So Paulo, Brazil. Peters, C. M. 1994. Sustainable Harvest of Non-tim ber Plant Resources in Tropical Moist Forest: An Ecological Primer. New York Botanical Garden, New York. Peters, C. M. 1996. The Ecology and Management of Non-Timber Forest Resources. The World Bank, Washington DC. Plowden, C. 2004. The ecology and harvest of andir oba seeds for oil produc tion in the Brazilian Amazon. Conservation and Society 2:251. Prance, G. T. 1979. Notes on the Vegetation of Amazonia III. The Terminology of Amazonian Forest Types Subject to Inundation. Brittonia 31:26-38. Queiroz, S. T., G.; Alves, P. 2004. Diagnstico Scio-econmico-ambiental da Reserva de Desenvolvimento Sustentvel do Itatup-Ba qui, Gurup Par. FASE Projeto Demonstrativo Gurup, Belm. RADAM. 1974. Projeto RADAM BRASIL, Rio de Ja neiro. Departamento Mineral de Produo Nacional. Ranklin, J. M. 1978. The Influence of Seed Predation and Plant Competition on Three Species Abundances in Two Adjacent Tropical Rain Forest Communities in Trinidad, West Indies. Ann Arbor: University of Michigan. Sampaio, P. d. T. B. 2000. Andiroba ( Carapa guianensis ). in J. W. Clay, Sampaio, P. de T.B., Clements, C.R., editor. Biodiversidade Amazni ca: exemplos e estratgias de utilizao. Programa de Desenvolvimento Empresarial e Tecnolgico, Manaus. SFB. 2009. Servio Florestal Brasileiro. Concesso Florestal. Avaiable at http://www.mma.gov.br/index.php?ido=conteudo monta&idEstrutura=95&idMenu=5731 Acessed in June 11, 2009. Scarano, F. R., T. S. Pereira, and G. Rocas 2003. Seed germination during floatation and seedling growth of Carapa guianensis, a tree from flood-prone forests of the Amazon. Plant Ecology 168:291-296.

PAGE 55

55 Shanley, P., and M. Londres. In Press. Andiroba in P. Shanley, G. Medina, and M. Cymerys, editors. Fruit Trees and Useful Plants in Amazonian Lives. FAO Rome. Snook, L. K., L. Cmara-Cabrales, and M. J. Kelty. 2005. Six years of fruit production by mahogany trees ( Swietenia macrophylla King): patterns of varia tion and implications for sustainability. Forest Ecology and Management 206:221-235. Stevens, G. C. 1987. Lianas as Structural Parasites: The Bursera simaruba example. Ecology 68:77-81. Ticktin, T. 2004. The ecological implications of harvesting non-timber forest products. Journal of Applied Ecology 41:11-21. Wadt, L. H. O., K. A. Kainer, and D. A. P. Gomes-Silva. 2005. Popul ation structure and nut yield of a Bertholletia excelsa stand in Southwestern Amazonia. Forest Ecology and Management 211:371-384. White, A., and A. Martin. 2002. Who owns the worl ds forests? Forest Trends, Washington, DC. Wittmann, F., W. J. Junk, and M. T. F. Piedad e. 2004. The vrzea forests in Amazonia: flooding and the highly dynamic geomorphology interact with natural forest succession. Forest Ecology and Management 196:199-212.

PAGE 56

56 BIOGRAPHICAL SKETCH Marina Londres was born in So Paulo, and grew up on a farm in Rio de Janeiro State, Brazil. She received her Bachelor's Degree in Fo restry at the University of So Paulo in 2004. Over the past 6 years, she has been working as a forest ecologist with riverine and indigenous populations in Amazonia, with the goal of ge nerating ecological information necessary to elaborate guidelines for be st management practices.