This item is only available as the following downloads:
1 SUCCESSIONAL DYNAMICS AND SEEDLING REGENERATION IN AMAZONIAN FLOODPLAIN FORESTS By CHRISTINE MARIE LUCAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011
2 2011 Christine Marie Lucas
3 To Mimi for sharing a great love of learning; to Papa for taking me fishing and building me benches to reach things; to Mom and Dad for giving me the great gifts of love and education; and to the many teach ers that opened the world to me
4 ACKNOWLEDGMENTS I thank first my advisor, Emilio Bruna, for his continual support and guidance throughout my experience as a doctoral student. I am grateful for the effort that he has put into my professional development as an ecologist and the confidence that he has continually shown in my work and progress throu ghout this dissertation. I also thank my committee, Katherine Ewel, Kaoru Kitajima, Jack Putz, and Debbie Miller for providing feedback on numerous proposals and chapter drafts. I thank Kaoru for her time and patience in reviewing my study design for see dling experiments and lending me books from her library to help me gain a better understanding of seedling physiology and statistics. I thank Jack for his thoroughness in reviewing all the proposals and manuscripts that I have given him since 2004. I hav e learned a lot about scientific writing from his comments. I thank Debbie for her perspectives from working in wetlands in the Southeastern US and for making t he long drive to Gainesville for committee meet ings In particular, I thank Kathy Ewel for con tributing her advice based on many years of experience in wetland forest ecology and scientific research. Kathy has been a role model for me as a pioneer in the field of wetland forest ecology and as a person that has opened up opportunities for other wom en in ecology. This thesis would not have been possible without the support of numerous colleagues, friends, students, community residents, and field technicians that participated in various aspects of the research Much of the field work of this thesis w as accompanied by Cristiane Nunes Nascimento, then a recent graduate of UF O PA in Santar m. She accompanied me for a year in the field, scribbling data amid the smoke fires to keep the mosquitoes at a tolerable level, as well as provided orientation for the students that worked with us. I would also like to thank Chieno Suemitsu for her
5 support with botanical identification of vrzea species and collaboration in organizing plant identification at INPA Manaus. Chieno facilitated many of my interactions w ith students and was a great mentor for student orientation and botanical field work. Many people accompanied this work i n the field over the course of 6 years working long days with mosquitoes ei From Ilha do Sao Miguel, I had the pleasure of working with Srs. Gote, Blade, Chico, Manoel, Amilton, as well as the women that made my stay in the co mmunity lik e home: Donas Concei o, Cilda, Hilda, and Francisca. In Tapar miri, I thank the family of Sr. Alonso for opening their home and their forest to me: Sr. Gilberto, Dona Nazar, Sr. Barraca, Dona Anita, and Jonhatan, Dunga, Tet, and Van for their help and good humor in the field. I thank the professors and directors of the elementary schools of Santa Maria do Tapar, Tapar miri, Aracampina, and Ilha do So Miguel for inviting me to talk to the students and providing me with an opportunity to learn with t he children by sharing my experience as a scientist with them. The staff and researchers at IPAM have provided support for this research since its initiation in 2004. This research would not have been possible had Toby McGrath not taken the chance of pe rmitting me to conduct an internship with IPAM in th e region in 2003 2004 I enjoyed working alongside the many people that work ed with the Vrzea Project: Marcio, Mauro, Adelson, Vir ginia, Lucilene, Solange, Edy Lopez, and Alcilene. Edy was one of the first people to open his heart to me and I will always remember him fondly for his creative e nergy, laughter, and warmth I am especially grateful to Pervaze Sheikh
6 who placed trust i n me to receive his data in order to continue forest inventories in the vrzea region. I am grateful to have met Mary Menton, who was a mentor and friend in adapting to Brazil and who took me to the field with her and made my first years in Santarm a ple asure. Finally, I would like to thank my husband, Marcelo Crossa, who introduced me to many of the community members that collaborated on this project and who provided much advice and support throughout the design, execution, and completion of the field w ork and the thesis. The community residents of Ilha do So Miguel, Santa Maria do Tapar, Tapar miri, and Aracampina do Ituqui provided invaluable support, insight, and inspiration for much of the research. To Lauro Almeida Sousa Dona Rosa, and Sr. Deco and Dona Doroca for opening their home to me so many times and for being like a family to Marcelo and I for so many years. To Sr. Nildo, Dona Adriana and their four children who often accompanied us into the forest by their house. To Sr. Gilberto, Dona Nazar, and their children for opening their home to Cristiane and I and the joy we had with the children in the evening. To Sr. Santino & Dona Gloria for opening their home to us as well. To Sr. Mello for letting us work in Mata Grande. At the Universid ade Federal Rural da Amazonia (UFRA), I thank the professors for permitting the use of the Seed Laboratory and the drying oven in the Wood Laboratory. Chapter 4 on seed germination was made possible by the orientation and advice of Fatima Mekdee, who had worked 30 years at the Seed Laboratory when I met her in 2007. Many students participated as interns on the seed germination experiment, including Pamella Suely Santa Rosa, Ana Sofia Sousa de Holanda, Jacqueline Bragas, Suelen Dias, and Suelen So usa. They continued to collect data while I went back to the
7 U.S., which enabled us to get germination data over the course of 6 months. Student Suellen Castro Cavalcante was an intern on the data collected in Chapter 3 she helped weigh soil samples an d dry plant material. It was a joy to work with all of these biology students and learn together! I thank all the staff at LBA in Santarm for providing laboratory and work space, in particular Paulo Coutinho, Rodrigo, and Wellington for always being av ailable to help with technological difficulties. I am especially grateful to Troy Beldini for orienting all of the soils analyses, teaching us how to measure soil texture and calculate bulk density. It was so nice to work with such a great team, even on the days when the birds cut off the electricity and we had to weigh samples using the generator! I thank my collaborators at INP A, Florian Wittmann, Jochen Sch ngart, and Maria Piedade, for their contribution to this work. I also thank Pia Parolin for her advice and for welcoming me into the community of researchers working on the ecology of the vrzea Identification of species would not have been possible without the expertise and experience of Jos Ramos, Nelson Rosa, Florian Wittmann, and Michael Hopk ins. I thank the staff at INPA Herbarium and Vrzea Project of INPA & Max Planck for their assist ance and hospitality during my visit in 2009 I am grateful to the cohort of friends that provided support while in Gainesville. The community of friends an d colleagues were one of the main reasons that I was drawn to the Universi ty of Florida. I learned so much from my colleagues at the University of Florida, including the cohort of many brilliant and creative students of the Working Forests in the Tropics Program and the Tropical Conservation and Development Project, and found in them a family away from home. The sharing of
8 knowledge, laughter, ideas, stress coping mechanisms, and food enriched my life while in Gainesville. To those friends that became li ke family while sharing a home: Marisa Tohver, Wendy Lin Bartels, Matt Palumbo, Amy Waterhouse, Kathleen McKee, Margo Stoddard, Marina Londres, and Paula Pinheiro, I am grateful for their support and love and home cooked meals. I thank my dear friends Cam ila Pizano for breathing the fresh air with me on many walks to Lake Alice, and Carla Stefanescu for teaching me how to take better care of myself. I thank my labmates for putting up with me and for being a great source of advice and motivation, in partic ular Paul Gangon and Ian Fiske for their statistical advice, as well as Ernane Vieira Neto, Antnio Aguiar Neto, and Selene Baez I am grateful to have met Brazilian collaborators that h ave been an inspiration: C ris Jurunitz, Jo o Jarenkow, Camila Castanh a, Alexandre Oliveira and Ariadna Lopes I thank the staff of Tropical Conservation and Development Program (TCD), the directors of Working Forests in the Tropics for their investment in my training and education. I thank Caprice, Monica, and Claire in the Wildlife Ecology and Conservation Department that took care of all the paperwork for grants, salaries, enrollment, etc. I thank Doug Daly and Jacquelyn Kallunki at the New York Botanical Garden. I thank Doug for his help with botanical identification and for providing contacts in Brazil in 2002 which initiated my work there, and Jackie for remaining a great resource for botanical identification and for facilitating my visit at NYBG to use the herbarium. I am gra teful for the many teachers and professors that contributed to my learning, including Robert Fritz and Robert Suter at Vassar College. Last but not least, I thank my family for all the support that they have provided during the process of this dissertati on. To Mom and Dad, I am so grateful for all that
9 they have worked hard to provide for me and for their investment in my education and the love and encouragement they provided along the way. To my husband, Marcelo, I am thankful for accompanying me on ma ny trips including to Manaus at 6 months pregnant to identify species in the INPA herbarium, and for being an essential source of support and encouragement from the very beginning of my doctorate to the last year w hen our son Santiago was born. I thank my sister, Kathryn, for being with the family while I was aw ay for so many years and my step father Eddie for being a constant source of support and stability. I thank Grandma Ronnie for her continual support. I thank my dear little Santiago for sleeping (sometimes) while I worked and for being such a source of joy and motivation to finish this dissertation. I am also thankful for my dear grandparents, Mimi and Papa, who provided part of the inspiration for this work, as one s fishing. The smell of fish in the Amazon always made me feel right at home! Mimi shared my love for reading, writing, and learning, as well as a love for nature and people. I thought of them both all the time while I was in the Amazon, and found mys el f at home among many fishermen and loving grandmothers that reminded me of them. Now that I have a son, I am acutely aware of all the love, time, and energy that they have put into my education and development, and for that I am forever grateful.
10 TA BLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 13 LIST OF FIGURES ................................ ................................ ................................ ........ 15 LIST OF ABBREVIATIONS ................................ ................................ ........................... 17 ABSTRACT ................................ ................................ ................................ ................... 18 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 20 Amazonian Floodplain Forests ................................ ................................ ............... 20 Catt le Ranching in Floodplains ................................ ................................ ............... 24 Fishing in Floodplain Forests ................................ ................................ .................. 25 Theoretical Contribution to Plant Community Ecology ................................ ............ 25 Participatory Research in Rural Riverside Communities ................................ ........ 27 Scope of the Dissertation ................................ ................................ ........................ 28 2 HISTORICAL LAND USES IN THE VRZEA OF THE LOWER AMAZON RIVER ................................ ................................ ................................ ..................... 31 Overview ................................ ................................ ................................ ................. 31 Pre Colombian Land Use on the Floodplain ................................ ........................... 32 First Encounter with European Colonists ................................ ................................ 33 Resource Extraction and Exportation: 175 5 1798 ................................ .................. 34 Chocolate from the Amazon Floodplain, 1640 1870 ................................ ............... 35 Botanical Observations by Naturalists, 1843 1850 ................................ ................. 38 Black Gold in the Amazon: Rubber Boom of 1870 1911 ................................ ......... 39 Fibers from the Floodplain: Jute boom of 1931 1990 ................................ ............. 40 Rise in Cattle Ranching: 1970 present ................................ ................................ 41 Summary ................................ ................................ ................................ ................ 43 3 EFFECTS OF SHORT TERM AND PROLONGED SATURATION ON SEED GERMINATION OF AMAZONIAN FLOODPLAIN FOREST SPECIES ................... 46 Overview ................................ ................................ ................................ ................. 46 Background ................................ ................................ ................................ ............. 47 Methods ................................ ................................ ................................ .................. 49 Study Region ................................ ................................ ................................ .... 49 Study Species ................................ ................................ ................................ .. 50 Collection and Experimental Design ................................ ................................ 50
11 Statistical Analyses ................................ ................................ .......................... 52 Results ................................ ................................ ................................ .................... 53 Discussion ................................ ................................ ................................ .............. 54 Effect of Short Term Saturation on Seed Germination ................................ ..... 55 Effects of Long Term Submergence on Seed Germination .............................. 56 Summary ................................ ................................ ................................ .......... 58 4 EFFECTS OF MULTIPLE ST RESSORS ON SEEDLING SURVIVAL AND GROWTH IN A TROPICAL FLOODPLAIN FOREST ................................ ............. 63 Overview ................................ ................................ ................................ ................. 63 Background ................................ ................................ ................................ ............. 63 Methods ................................ ................................ ................................ .................. 67 Study Site ................................ ................................ ................................ ......... 67 Experimental Design ................................ ................................ ........................ 67 Measurement of Environmental Covariates ................................ ...................... 69 Measurement of Survival and Growth ................................ .............................. 70 Tolerance and Resource Allocation Trade offs ................................ ................ 71 Statistical Analyses ................................ ................................ .......................... 72 Results ................................ ................................ ................................ .................... 73 Effect of Flooding and Damage on Annual Seedling Survival .......................... 73 Seasonal Differences in Survival ................................ ................................ ...... 74 Effect of Light on Survival ................................ ................................ ................. 75 Effects of Damage and Flooding on Relative Growth Rates ............................. 76 Effects of Damage and Flooding on Biomass Allocation ................................ .. 76 Effect of Light on Growth and Biomass Allocation ................................ ............ 77 Species Trade offs for Flood, Damage, and Shade Tolerance ......................... 77 Discussion ................................ ................................ ................................ .............. 78 Independence of Effects of Flooding and Damage on Growth and Survival .... 79 Strong Effects of Damage on Growth and Survival ................................ .......... 80 Flood Duration Effect on Seedling Survival ................................ ...................... 82 Species Resource Allocation Trade offs ................................ ........................... 83 Summary ................................ ................................ ................................ .......... 85 5 TREE COMMUNITY DYNAMIC S ACROSS FLOOD AND DISTURBANCE GRADIENTS IN AMAZONIAN FLOODPLAIN FORESTS ................................ ..... 105 Overview ................................ ................................ ................................ ............... 105 Background ................................ ................................ ................................ ........... 106 Methods ................................ ................................ ................................ ................ 108 Study Site ................................ ................................ ................................ ....... 108 Experimental Design ................................ ................................ ...................... 109 Statistical Analyses ................................ ................................ ........................ 112 How do flood level and livestock activity interact to affect seedling density and species density? ................................ ................................ 112 What are the effects of livestock activity and flood level on stand structure and species richness? ................................ ........................... 113
12 What are the effects of livestock activity and flood level on rates of change? ................................ ................................ ................................ 113 How does livestock activity affect species composition? ......................... 114 Results ................................ ................................ ................................ .................. 114 Effects of Flood Level a nd Livestock Activity on Seedlings ............................ 114 Effects of Flood Level and Livestock Activity on Trees ................................ ... 115 Effects of Flood Level and Livestock Activity on Stand Dynamics .................. 116 Effects of Flood Level and Livestock Activity on Species Composition .......... 117 Discussion ................................ ................................ ................................ ............ 118 Seedling Response to Livestock Activity, Flood Level and Light .................... 119 Secondary Succession of Forests ................................ ................................ .. 121 Stand structure and species composition ................................ ................ 121 Floodplain forest stand dynamics ................................ ............................. 124 Successional trajectories of floodplain forests ................................ ......... 125 Implications for Conservation in the Context of Climate Change .................... 126 6 ABOVEGROUND BIOMASS AND CARBON SEQUESTRATION IN SECONDARY FLOODPLAIN FORESTS OF EASTERN AMAZONIA .................. 142 Overview ................................ ................................ ................................ ............... 14 2 Background ................................ ................................ ................................ ........... 142 Methods ................................ ................................ ................................ ................ 145 Study Region ................................ ................................ ................................ .. 145 Study Design ................................ ................................ ................................ .. 146 Statistical Analyses ................................ ................................ ........................ 148 Results ................................ ................................ ................................ .................. 150 Discussion ................................ ................................ ................................ ............ 151 T rends in A boveground Biomass in Amazonian Floodplain Forest ................ 151 Changes in Aboveground Biomass ................................ ................................ 153 Summary ................................ ................................ ................................ ........ 156 7 CONCLUSIONS ................................ ................................ ................................ ... 166 Significance in Ecology ................................ ................................ ......................... 168 Implications for Conservation and Management ................................ ................... 170 APPENDIX A LIST OF WOODY SPECIES IN SANTARM FLOODPLAIN FORESTS .............. 173 B SOILS DATA FOR COMMON GARDEN PLOTS IN CHAPTER 4 ........................ 175 LIST OF REFERENCES ................................ ................................ ............................. 180 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 202
13 LIST OF TABLES Table page 3 1 Seed characteristics of the ten study species in order of wet tolerant and dry tolerant species. ................................ ................................ ........................... 60 4 1 Physiological traits of ten floodplain forest study species. ................................ .. 87 4 2 Survival response to flood level, damage, and light in three time periods: one year, low water season, and flood season ................................ .................. 88 4 3 Summary of generalized mixed model results for the effe cts of damage, flooding, and light on seedling survival. ................................ .............................. 89 4 4 Summary of generalized linear mod el results for the effects of flooding, dama ge, and canopy openness on see dling relative growth rate ...................... 90 4 5 Species traits and surviva l according to damage tolerance, flood tolerance, and shade tolerance of damaged and undamaged seedlings. ........................... 91 4 6 Summary of generalized linear model results for the effects of damage, flood duration, light availability on three growth parameters of seedlings .................... 92 4 7 Summary of results from linear mixed models on the effects of damage, flood duration, and light availabil ity on root:shoot ratios ................................ .............. 93 5 1 Summary of linear mixed model results for the effects of flooding, livestock, activity, and light availab ility on the seedlings ................................ .................. 127 5 2 Summary of linear mixed model results for the effects of flood level and livestock activity on tree stem density, basal area, and species density. .......... 128 5 3 Summary of linear mixed model results for the effects of flood level, livestock activity and their interaction on tree rates of change ................................ ....... 129 5 4 Tree stand structure a nd species richness in the three forest in ventories ........ 130 5 5 Estimated spec ies richness of trees and seedlings across flood level categories a nd average livestock activity ................................ ......................... 131 6 1 Comparison of mean biomass, carbon storage, and carbon sequestration for aboveground woody stems in vrzea forests of the Amazon Basin ................. 158 6 2 Summary of generalized linear model results for the effects of flood level and fo rest age on aboveground woody biomass in and net biomass accumu lation 159 6 3 Summa ry of generalized linear model results for the effects of flood level and cattle activity on a boveground woody biomass and net biomass accumulation 160
14 6 4 Rates of change in aboveground biomass in forest s tands over 9 years by forest age. ................................ ................................ ................................ ........ 161 A 1 Species list for woody species recorded in 2008 inventories, ordered by decreasing Importance Valu e (IV) among trees ................................ ............... 173 B 1 Soil texture and bulk density in the 21 common garden plots of seedlings. ...... 176 B 2 Soil nutrients and organic matter. ................................ ................................ ..... 177
15 LIST OF FIGURES Figure page 1 1 Amazonian floodplain forests of the Santarm region that differ in livestock activity ................................ ................................ ................................ ................ 30 2 1 Jute cultivation on the vrzea of Ilha do So Miguel ................................ ........... 45 3 1 Germination curves for ten study species with log rank test statistics for differences in Kaplan Meier survival curves. ................................ ...................... 61 3 2 Germination curves and log rank test statistics for survival curves of four species ................................ ................................ ................................ ............... 62 4 1 Rainfall and fl ood pulse in the study region. ................................ ....................... 94 4 2 Seedling survival across a gradient of flood levels and light availability for undamaged and damaged s eedlings ................................ ................................ .. 95 4 3 Average leaf number of undamaged seedlings for all species over ti me ............ 96 4 4 Seedling survival acro ss flood duration and d amage treatments for all species ................................ ................................ ................................ ............... 97 4 5 Seedling survival of grouped sp ecies over the seven censuses. ........................ 98 4 6 Seedling survival across a gradient of light availability for all species ................ 99 4 7 Relative growth rates of da maged and undamaged seedlings. ........................ 100 4 8 Root:shoot biomass ratios as a function of flood duration illustrating the interaction between damage and flood duration ................................ .............. 101 4 9 Positive correlation between damage tolerance and flood tolerance for ten s eedling species ................................ ................................ ............................... 102 4 10 Positive correlation between shade tolerance and flood tolerance of ten study species indicated by genus, excluding Hevea ................................ ........ 103 4 11 Trade offs between relative growth rate and survival ................................ ....... 104 5 1 Map of secondary floodplain forests stands in Santarm, Par, Brazil, at the confluence of the Amazon River and the Tapajs River. ................................ .. 132 5 2 Average seedling den sity across a flood gradient. ................................ .......... 133 5 3 Change in seedling density acro ss livestock change trajectories ..................... 134
16 5 4 Number of seedling species in 45 m 2 across a flood gradient .......................... 135 5 5 Trends in forest stand structure across flo od level and livestock activity .......... 136 5 6 Stem and species turnover rates ................................ ................................ ...... 137 5 7 Trends in stem and species turnover acro ss the flood gradient. ....................... 138 5 8 Species accumulation cu rves for seedlings and trees ................................ ...... 139 5 9 Comparisons of relative abundance of the of the 6 most common species of trees, recruits, and seedlings at different flood levels ................................ ....... 140 5 10 Comparisons of relative abundance of the of the 6 most common species of trees, recruits, and seedlings at different livestock activity levels ..................... 141 6 1 Map of the number of consecutive months with rainfall < 100 mm, illustrating the Dry Corridor ................................ ................................ ................................ 162 6 2 Differences in biomass and biomass accumulation across forest age .............. 163 6 3 Annual biom ass accumulation across average cattle impact levels. .............. 164 6 4 Annual b iomass accumulation attributed to tree growth, mortality, an d recruitment ................................ ................................ ................................ ........ 165 B 1 T he relationship between flood level and soil texture i n three floodplain forests ................................ ................................ ................................ .............. 178 B 2 Bulk d ensity at 0 5 cm and 5 10 cm in three floodplain forests ........................ 179
17 LIST OF ABBREVIATION S DBH D iameter at B reast ( 1.3 m ) H eight EMBRAPA Empresa Brasileira de Pesquisa Agropecuaria LBA Large Scale Biosphere Atmosphere Experiment in Amazonia LMM Linear Mixed Model IBGE Brazilian Institute of Geography and Statistics IPAM Instituto de Pesquisa Ambiental da Amaznia INPA Instituto Nacional de Pesquisa Amaznica SD Standard deviation SE Standard error UFRA Universidade Federal Rural da Amaznia UFOPA Universidade Federal do Oeste do Par
18 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy SUCCESSIONAL DYNAMICS AND SEEDLING REGENERATION IN AMAZONIAN FLOODPLAIN FORESTS By Christine Marie Lucas August 2011 Chair: Emilio M. Bruna Major: Wildlife Ecology and Conservation Amazonian floodplain forests are critical ecosystems that sustain the productivity and divers livelihoods of millions of people. To understand how floodplain forests recover from anthropogenic disturbances in the context of severe flood stress, I test ed how environmental stressors and disturbance interact to affect forest recovery at multiple scales in Eastern Amazonian floodplains. First, I review the land use hi story of Amazonian floodplains (Chapter 2). Using laboratory experiments I tested the effects of short term and prolonged saturation on seed germination of ten flood tolerant species (Chapter 3). To understand factors mediating the seedlings I tested how multiple stressors affect growth and mortality in common garden experiments (Chapter 4). To exp lore patterns in forest succ ession I compared changes in the seedling and tree communities across gradients of livestock activity, flooding and forest age (C hapter 5). Finally I estimate d aboveground woody biomass storage and accumulation by floodplain forests (Chapter 6). S eeds of flood tolerant species had diverse strategi es for colonizing floodplains. Mechanical damage to planted seedlings reduced growth and survival during a critical
19 growth window in the low water season. The effects of damage a nd flooding were independent, showing how species persist under the combined effects of multiple stressors Light availability enhanced growth a nd survival of pioneers Among trees, f lood level and forest age were major drivers of dynamics of change in s tand s tructure and species richness. Seedling species richness of seedlings in the forest was mediated by the interaction of light and flood level, showing that shade was an important limiting factor for seedling diversity. I observed no differences in t ree mor tality or recruitment across livestock density levels Biomass accumulation averaged ~5 Mg ha 1 y 1 in forests 15 50 y old, and an average of 70% of biomass g ained by tree growth and recruitment wa s lost to mortality. T hese results suggested that secondary floodplain forests, despite multiple land uses, retain high rates of biomass accumulation dur ing secondary succession. These results suggest how floodplain forests sust ain high productivity and plant diversity, despite many stressors and disturb ances
20 CHAPTER 1 INTRODUCT ION (Mitsch and Gosselink 2000) floodplains are estimated to provide > 25% of terrestrial ecosystem services ( sensu Tockner and Stanford 2002; e.g. disturbance regulation, water supply, and waste treatment) Floodplain fo rests are an integral component of many riparian landscapes, providing a buffer between terrestrial land use activities and aquatic resources and offering habitats for diverse fauna. Riparian floodplain forests are among the most threatened ecosystems, wi regions and severe alteration of hydrology. The fertile soils, abundant natural resources, and access to river transport have made these environments attractive to human settlement and land use Floodpla use activities (Zarin 2004) Europe and North American floodplain forests have e xperienced losses reaching 80%, while tropical and subtropical floodplains currently face high rates of loss and degradation due to land use change and alteration of hydrology (Tockner and Stanford 2002) The decline in tropical floodplain forests threatens not only the ecology of riparian systems, but also livelihoods of millions of people that depend upon floodplain forest resources. Amazonian Floodplain Forests The Amazon River and its adjacent floodplains form the largest and most diverse freshwater system on the g lobe. Total land cover of floodplain forests in the Amazon Basin was estimated at 200,000 km 2 (Junk 1997) but a more recent estimate of floodplain and swamp forest based on remote sensing data estimates 3 29,000 km 2 or
21 4.2% of the total Amazon basin (Saa tchi et al. 2007) Over 10 00 woody species are found in Amazonian floodplain forests, with ca. 31% of these also being found in upland moist tropical forest (Wittmann et al. 2006) Floodplain forests are less diverse than upland forests, with estimates of tree diversity ranging from 53 species ha 1 in the Eastern Amazon (Pires and Koury 1959) to 149 species ha 1 in the Western Amazon (Balslev et al. 1987, Parolin et al. 2004b) Amazonian floodplains constitute a heterogeneous group of forests that vary in stand structure and species composition w ith regards to water chemistry, soil type, hydrology, and geomorphology. Most floodplains fall into two categories distinct in water quality, soil structure, and geological history. White water floodplains ( vrzea ), comprised of nutrient rich sediments o f relatively recent deposition during the Quaternary while black water and clear water floodplains ( igap ) are nutrient poor waters often associated with sandy soils and lower tree diversity (Prance 1979, Furch 1997) These two floodplain forest types differ in species c omposition (Ferreira et al. 2010) and the more productive and diverse vrzea is considered to be of particular interest for regional diversity, nat ural resources, and human livelihoods for a majority of the Amazonian population (Padoch and Steward 2011) water floodi ng by sediment rich waters, vrzea floodplains are distinct in respect to geomorphology, hydrology, species composition, and land use history within the Amazon Basin (Sioli 1984, Junk 1997) Vrzea floodplains are comprised of a mosaic of environments forests, grasslands, floating meadows, and lakes shaped by fluvial
22 dynamics. In the headwaters of the Amazon, rivers are highly dyn amic with high rates of sedimentation and erosion, as well as oxbow lakes that formed by sections of river cut off from the main channel (Pinedo Vasquez 1999) The annual flooding regime in these headwater tributaries varies from polymodal to monomodal, with more rapid rates of change and abrupt fluctuations in river level than the annual floods in the Central and Lower Amazon (Junk 1997) Changes in flood level can reach up to 14 m in these tributaries (Goulding et al. 2003) but flood duration is than in the Central and Lower Amazon (1 4 mo, Nebel et al. 2001c) In the middle and lower stretches of the Ama zon River (excluding the tidal estuary), the flood regime is characterized by a highly predic table monomodal flood pulse that changes in river water level gradually over a 6 month flood season (Junk 1997) There are no oxbow lakes, but rather large seasonal lakes of u p to 30,000 ha (M. Crossa, personal communication, March 25, 2011 ) that connect with rivers during the high water season (Junk 1997) The Amaz on Estuary is characterized by a tidal flood regime with daily fluctuations in river level of 2 4 m, depending on the time of year (Almeida et al. 2004) Species richness and composition vary widely among vrzea forests of different regions and different tributaries within the Amazon Basin (Parolin e t al. 2004b) Basin wide trends in diversity suggest decreasing diversity of vrzea forests along a West to East gradient (Wittmann et al. 2006) Floodplain trees have evolved a suite of physiological and phenological mechanisms for tolerating prolonged periods of waterlogging and/or submergence (Parolin et al. 2004a) Tolerance of anoxia is particularly important at the seedling phase, as seedlings may be submerged for up to 210 days (Parolin 2002) In no other place in the world are tree seedlings known to tolerate such prolonged periods of submergence.
23 The vrzea of the Santarm region is distinguished by extensive areas of grasslands and fl oating meadows in the floodplain that cover up to 80% of the vegetated area. In contrast to the Estuary, Central, and Upper Amazon regions, forests do not dominate the floodplain landscape but are restricted to high elevation levees where the water column ranges from 0 2 m depth. It is these abundant and highly productive grasslands that make the region today so attractive to cattle ranching. In addition, the predictable, monomodal pulsing of relatively shallow floods (7.5 m on average) permits managemen t of cattle on the floodplain, with the use of adjacent uplands for ranching during the high flood season. The major threat to forests is the constant passage of cattle and water buffalo through forests seeking grasslands for grazing. Daily trampling by cattle and water buffalo herds damage understory plants and compact soils, compromising woody plant regeneration. The impacts of cattle in floodplain forests are visually striking, as noted by the openness of the forest understory in areas of high livesto ck traffic (Figure 1 1). Amazonian white water floodplains serve important ecological and economic roles for fisheries (McGrath et al. 1993) agriculture (WinklerPrins 1999) timber (Anderson et al. 1999) and non timber forest products (Fortini et al. 2006) Over 70% of the Amazonian popula tion resides on or near floodplains, and the vrzea supports higher rural population densities than those of uplands. The social, economic, and political context of land use by riverine people in the vrzea has been reviewed in two books (Padoch et al. 1999, Padoch and Steward 2011) Some challenges that riverine landholders in the study region of Santarm fa ce are the communal management of common pool resources such as fish (McGrath et al. 1999a) shifts from artisana l to
24 commercial fishing (McGrath et al. 19 93) migration to uplands following economic downturn of jute (WinklerPrins 2002) and changing land tenure policies (McGrath et al. 2011) Relatively few ecological studies have investigated the impact s of land use transformations on the extent, diversity, species composition, and ecosystem function of Amazonian floodplain forests (Zarin et al. 1998, Zarin et al. 2001, Pereira et al. 2002, Fortini e t al. 2010) Cattle R anching in Floodplains A leading threat to th e conservation and sustainable management of many tropical and sub tropical floodplain s is the intensification of livestock densities While floodplains sustain the impacts of many large grazers, the intensification of activities related to livestock ranc hing has resulted in the degradation and alteration of floodplain habitat in Africa (Hughes 1988) Australia (Gatewood and Cornwell 1976, Robertson and Rowling 2000, Jansen and Robertson 2001) India (Dahdouh Guebas et al. 2006) and South American tropics (Junk and de Cunha 2005) Management of livestock impacts in riparian floodplain systems is known to have been an issue of concern sinc e at least the 1600s in Europe (Klimo and Hager 2001) and more recently in the U.S. (Kauffman and Krueger 1984, Belsky et al. 1999) In the floodplains of the Central and Lower Amazon River, the most recent major cycle of land use transformation is the conversion of floodplains to cattle pasture (Mer tes et al. 1995) Introduced water buffalo and cattle herds are increasing at annual rates of ~13% and 4%, res pectively (Sheikh et al. 2006) Livestock move through forest understories to reach the lush grasslands and meadows of the floodplain. As flood levels rise, cattle may concentrate on higher elevations, which is where floodplain forests occur. L ivestock can trample vegetation in the forest understory and co mpact soil, reducing the density of understory woody plants
25 (Shei kh 2002 ). Over time, the continual effects of dense livestock herds in floodplain otential invasion of grasses W e know of no previous test of the effects of cattle disturbance on seedling regeneration and successional dynamics in floodplain forests over time. Fishing in Floodplain Forests C onservation and man agement of floodplain forests are essential for sust aining Amazonian fisheries (Goulding et al. 1993) Floodplain forests provide food and shelter for diverse aquatic and floodplain biotic communities. In synchrony with the annual flood pulse, floodplain forests yield an abundance of fruits and seeds which are consumed by fishes migrating into the floodplain. Over 200 fish species of the Amazon consume fruits and seeds from flooded forests, including such commercially valuable species as Colossoma macropomum (Araujo Lima and Goulding 1998) In this thesis, I first investigated the seed germination, seedling establishment, and forest dynamics of tree speci es that serve as important food sources for economically valuable fishes that consume their fruits (Lucas 2008) In the study region of Santarm smallholders practice three main resource based livelihood strategies: fishing, farming, and livestock ranching (Almeid a 2004) Fishing is the most commo n strategy, practiced by 95% of families and is the main source of prote in for floodplain families Timber harvesting is no longer a prominent activity in the region of Santarm, as timber has been depleted in the reg ion and wood for construction is purchased from the uplands Theoretical Contribution to Plant Community Ecology The drivers of succession remain a central area of investigation within plant community ecology. Although the dynamics of forest succession h ave been studied for
26 over a century, new questions have emerged: What are the synergistic effects of multiple factors on forest recovery from disturbance? Forested wetlands are excellent systems for investigating how local anthropogenic disturbances inter act with flooding stress to influence forest succession. These ecosystems have adapted to several environmental stresses, most importantly flood induced anoxia, and have also experienced extensive manipulation by human land use practices (Messina and Conner 1998) Flooding stress can interact with land use activities to create a mosaic of plant communities at various stages of succession, a lthough little is known of how flooding and anthropogenic disturbances interact to affect the earliest stages of succession ( e.g., seed dispersal, establishment, seedling growth, and seedling community dynamics). In light of the global focus on climate c hange, there is increasing research on the changes in carbon storage and sequestration in forests over time. The few estimates available for floodplain forests suggest that although total above ground biomass may be lower than upland moist tropical forest s, carbon sequestration by Amazonian floodplain forests is substantial. While successional forests can portray high biomass accumulation rates that offset carbon emissions (Asner et al. 2010) degradation of forests can lead to reductions in biomass and carbon sequestration (Aide et al. 1995, Hughes et al. 1999, Steininger 2000) A large area of floodplain forest in the Eastern Amazon basin is secondary, recovering from deforestation and degradation from multiple land use activities (Anderson et al. 1999, Zarin et al. 2001, Wittmann et al. 2006) The rapid growth rates of early mid successional tree species in secondary floodplain forests could result in high rates of biomass accumulation (Schngart et al. in press) Alternatively, degradation of secondary floodplain forests by land use activities
27 such as logging, extrac tion of fuelwood, and trampling by livestock (Sheikh et al. 2006) could reduce rates of biomass accumulation via high stem mortality and poor recruitment (Chazdon et al. 2007) Participatory Research i n Rural Riverside Communities Scientific research can contribute to sustainable grassroots development of natural resources by integrating user participa tion and capacity building into the research process (Chambers and McBeth 1992, Moller et al. 2004) Participatory research aims to involve local people, communities, and institutions in the research process to shift scientific learni ng from an extractive process to that of exchange and collective learning (Agrawal and Gibson 1999, Klooster 2002, Arnold 2004) There has been a recent increase in participatory appr oaches to scientific research, but furt her evaluation and application of these methods is needed. The participatory approach to ecological research can build adaptive capacity, knowledge, and skills among local people, communities, and researchers to manage and conserve natural resources. Res ource users and communities utilize knowledge and experience to create and adapt development criteria to changing social and ecological circumstances (Armitage 2005). The f ield research for this work was carried out in collaboration with community member s, landholders, students, and governmental and non governmental institutions. The research involved the participat ion of riverine community residents in the research process, from hypothesis formation to the analysis and application of results to resource managemen t. At a regional scale, I collaborated with three floodplain communities and the local branch of EMBRAPA ( Empresa Brasileira de Pesquisa Agropecuaria ) in Santarm to monitor the effects of livestock activity and flood level on seedling community diversity, regeneration strategy, and availability of locally valuable
28 resources. At the lo cal scale, two communities participate d in a series of experiments to test how variation in flood level and cattle activity affect seedling community dynamics, and ultimately, forest succession. The collective learning process during research has the potential to contribute to individual and community capacity to manage and conserve floodplain forest resources Research results were disseminated to communities an d institutions in written and oral forms and evaluated by participants. Beyond basic research, we developed learning activities with local schools to share knowledge about the vrzea ecosystem and the use of its resources, and to share information on scie ntific methods of research. Scope of the Dissertation This dissertation addresses aspects of seedling ecology and forest succession in secondary floodplain forests of the Amazon River. The overall objective of this work is to understand the dynamics of recovery of tropical forests at multiple scales in the c ontext of extreme seasonal floods and chronic disturbance by introduced livestock. I first review the history of land use and forest conversion in Eastern Amazon floodplains since the pre Colonial period (Chapter 2). The remaining Chapters are a series o f studies that examine the effects of various ecological drivers on seed germination, seedling growth and mortality, tree communities, and ecosystem processes. I tested the effect of water saturation and submergence time on seed germination of ten floodpla in forest species (Chapter 3). Once seeds germinate on the floodplain, growth and survival during the first year of establishment is critical. Using common gardens, I test ed the synergistic effects of flood stress shade stress, and mechanical damage on seedling growth and survival during the first year of establishment (Chapter 4). At a broader scale, I used a 4.9 ha network of forest plots to test how flood level and
29 livestock activity affect seedling and tree community dynamics (Chapter 5). These inv entories are a continuation of work initiated by Pervaze Sheikh in 1999. To examine the role of floodplain forests for biomass storage and sequestration, I measured aboveground biomass and biomass change in collaboration with colleagues from the Max Planc k Institute and INPA. We explored trends in biomass sequestration across flood level and forest age (Chapter 6). Together, these chapters address the history of land uses, the driving factors for forest diversity and productivity, and the capacity of for ests to recover amidst many stresses
30 A B Figure 1 1 Amazonian floodplain forests of the Santarm region that differ in livestock activity. A) Forest with heavy cattle activity and (B) light act ivity Photos courtesy of Christine Lucas.
31 CHAPTER 2 HISTORICAL LAND USE S IN THE VRZEA OF THE LOWER AMAZON RIVER Overview The white water floodplains along the Amazon River, known as vrzea have a long history of multiple land uses. Pre Colombian history in the vrzea suggests that these nutrient rich floodplains provided important agricultural land and fishing grounds for riverine people and adjacent upland chiefdoms. The floodplains and upland bluffs of the Santarm region in Par state, extending 225 km from the towns of "bidos to Monte Alegre, is rich with archeological evidence of inhabitance on and adjacent to floodplains, including large shell middens and areas of terra preta (Smith 1999) Following European colonization, the floodplain landscape experienced a series of boom and bust cycles in land use, including conversion of fore sts for agriculture, cacao plantations, rubber extraction, jute plantations, and cattle ranching. Throughout these cycles of land use, vrzea forests have served as a valuable resource for fisheries and extraction of timber and non timber forest products. Today, approximately 70% of the Amazonian population live on or adjacent to Amazonian river floodplain systems (IBGE 2007 sensu calculations by Padoch and Steward 2011) and the vrzea supports higher rural populatio n densities than the terra firme (Hiraoka 1995) The proximity of urban centers such as Belm, Santarm, Manaus, Tef, and Iquitos place s increasing demands on vrzea ecosystems for natural resource s. In this chapter, I discuss the historical accounts of land use in the vrzea of the Santarm region since pre Colombian times through the twentieth century.
32 Pre Colombian Land U se on the F loodplain Archeological evidence near the floodplain at the co nfluence of the Tapajs and Amazon Rivers suggests inhabitance of multiple indigenous communities and civilizations. While the natural cycles of erosion and sedimentation of the floodplain may have erased most archeological evidence of pre Colombian indig enous inhabitance on the floodplain, it is speculated that the floodplains had indigenous villages and provided agricultural fields for both floodplain and upland settlements (Denevan 1984, Smith 1999) Among the earliest and best known evidence for large villages near the floodplain is a shell midden of 7 m depth over several hectares wide on an ancient river terrace 50 km downstream of Santarm. Shells, coal, tools, and pottery shards date between 7000 to 8000 y B.P., and suggest permanent settlement along the Santarm floodplains (Roosevelt et al. 1991) The developmental sequence at Santarm sheds li ght on human adaptation to the tropical environment over the millennia, revealing that tropical resources supported the earliest pottery age cultures yet known in the that Amazon floodp lains were extensively exploited for thousands of years and may be more appropriate for development efforts than poor (Roosevelt et al. 1991: 1624) While there has been much scholarly debate regarding the extent and distribution of Pre Colombian indigeno us peoples in Amazonia (Meggers 1971, 1996, Heckenberger et a l. 2003) there is agreement that white water floodplains provided abundant natural resources and nutrient rich soils for farming. Periodic flooding and occasional exceptionally high floods ( e.g., 1819, 1953, 2006, 2009) could have limited dense settlement (Denevan 1984, 1996) Long term settlement on the floodplain of Carmo Island between Santarm and "bidos is suggested by a mound with terra preta
33 and pottery shards (Smith 1999: 27 28) The juxtaposition of rich floodplains and secure, dry uplands, described as the bluff model, was likely advantageous for the development of large, long term settlements and complex societies in the Santarm region by 1000 years BP (Roosevelt 1992, Denevan 1996) The extent of bluff settlements is indicated by the present day distribution of blackened anthrosols, known as terra preta do indio some of which extend into the high e levation vrzea (Denevan 1996, Smith 1999) The distribution of terra preta pottery shards, and shell middens suggest that Tapaj satellite villages extended at least 30 km downstream along the upland bluffs of the south shore of the Amazon River (Smith 1999) The largest terra preta site in the Brazilian Amazon (350 ha) is found in Juruti, on an upland riverside bluff between Santarm and Parentins (Smith 1999) First Encounter with European Colonists The first written account of the large indigenous civilizations in the region Santarm was made by Carvajal in 1541 during Orellana down the Amazon River from Peru (Medina 1934, Palmatary 1965) At a location on the Amazon where the tide was first noticed (between Santarm and Oriximin), Carvajal described gleaming white villages that stretch ed for miles on the river bluff. The land was so populated on the south shore of the Amazon below the Tapajs River that the brigantine stayed to the north shore to avoid conflict and attacks. On June 25, Orellana and his men w ere approached by over 200 canoes of 20 40 men each (Medina 1934, Palmatary 1965) Smith (1999) speculates that in order to support such numbers of warriors, the Tapaj civilization must have been at least 40,000 people. In reference to the floodplain near the Tapajs, after surviving a series of Indian attacks, Carvajal writes
34 observation in June of 1541, when flood levels are high, suggesting that residence on the floodplain was not confined to the dry season. In the 200 years fol activity or economic interest in the Amazon upstream of the estuary There were a few the second written ac count was not until 1640 by the father Cristoval de Acua on the Texeira expedition from Peru to Belm (Markham 1963, Furneaux 1969) Although osity with descriptions of the surrounding vegetation, he does give much detail about indigenous peoples. Acua states that the Tapaj Indians bartered fish, flour, and ducks, which could have come from floodplains nearby (Markham 1963: 124 125) R esource E xtraction and Exportation : 1755 1798 The second half of the eighteenth century in the state of Par defined a new period in the history and land use of the Amazon. The Portuguese took a newfound interest in natural resources from the Lower Amazon, and extraction systems were soon set in place For the floodplain, this signified the implementation and expansion of cacao plantations, initial efforts at systematic agriculture, and exploitation of indigenous people for labor. The Directorate system was a state run program established in 1758 to encourage i ndi genous assimilation, agricultural production, and resource extraction in the Amazon (Anderson 1999) Directorate villages were established in pre existing Aldeas or indigenous villages, with a director and a few white settlers (known as moradores or agregados ). The most important economic activity in Directorate villages in the eighteenth century was forest and riverine collection trips ( negocios do serto ) in search of spices, oils, and other exotic forest products, known as drogas do se rto
35 (Anderson 1999) Non timber f orest products collected included the spice and medicinal herb salsaparrilha used to cure syphilis ( Smilax santarm ensi s ), wild clover ( cravo grosso ), and Brazil nut (largely the upland species, Bertholletia excelsa but also including the floodplain species Lecyth is pisonis ). Various tree barks were collected for caulking materials ( estopa ); tree resins to make tar for s hip construction and repair ( breu ); and tree aromatic oils from floodplain tree species, andiroba (Carapa guianensis) as well as a cotton like stuffing material from the seed pod of the kapok tree s amauma (Ceiba pentandra) (Anderson 1999) Riverine resources included turtle ( Podocnemis expansa) meat salted pirarucu ( Arapaima gigans ), and manatee ( Trichechus inunguis ) meat and oil. The importance of tur tles in Amazonia is summarized as, (Anderson 1999: 54) The overexploitation of the giant river turtle ( Podocnemis expansa) led to its eventual population crash in the 1870s. (Smith 1999) Thirty years after the implementation of the Directorate system, the economic value of Amazonian exports shifted from an emphasis on the harvest of non timber forest products to crops: coffee, cacao, rice, corn, and cotton (Anderson 1999) Cacao, rice, beans and tobacco were among the crops grown on vrzea floodplains. Chocolate from the Amazon F loodplain, 1640 1870 As early as 1640 cacao was recognized as an important commodity from the Eastern Amazon (Palmatary 1965) The value of cacao is reiterated in 1743 as the primary trade commodity of Indians for European fabrics, needles, mirrors, scissors, and combs (Palmatar y 1965) T he cities of Santarm and "bidos (120 km upriver from Santarm ) bec a me widely known as a cacao center of the Amazon. By 1823 cacao
36 (Spix and Martius 1823/1968) The margins of the Amazon and its meandering tributaries ( paranamirims lines of (Penna 1869) The first accounts of the Amazon River from American and European naturalists in the 1800s coincide with the zenith and subsequent decline of the cacao industry. As such, botanical observations of the vrzea around Santarm and "bidos are typical of the following : Had the vegetation of the South bank, along which our course lay, been more interesting, I would n ot have demurred at the delay, for I was unable that caco cultivation is most extensively carried on. The cacoals either reach to the very margin of the river or have an intervening narrow fringe of such weeds, shrubby and herbaceous, that grow co mmonly on inundated river banks (Spruce 1908/1970: 78) Wild cacao grew naturally on the floodplains ( Theobroma spruceana ), but largely domesticated varieties ( Theobroma cacao ) were planted on the floodplain levees, sometimes in groves of thousands of trees. Single plantations with 40,000 planted cacao trees, stretching for kilometers along the riverbank, are reported around the peak of cacao (Penna 1869) The region between Santarm and "bidos appears to have been the number one exporter of cacao in the Amazon, until 1790, when the Furo district farther upstream took the lead (Anderson 1999) Once the Directorate system was terminated in 1798, cacao plantations were transferred to private landholders, by either legal or illegal means. A typical family on the vrzea in the vicinity of "bidos may have managed a cacao plantation of 10,000 to 15,000 trees in 1850, which were harvested for fruits once each year (Bates 1864/1962) Despite the civil strife as sociated with national independence ( e.g.
37 Cabenagem ) and the lack of a coherent policy for land ownership in the Brazilian Amazon, much of the vrzea land, particularly around the old Directorate villages, was held by legal land title or illegal squatters ( posseiros ) (Anderson 1999) Following cacao, the most important export items from the region were tobacco, wild clover, rice, cotton, manioc flour ( fari nha ), and salted pirarucu (a replacement for salted cod). The vrzea was also important for planting corn, cotton, sugar cane, and tobacco (Smith 1879) There are many potential reasons behind the slow dec line of cacao. From the accounts of naturalists, there were inefficiencies in cacao seed production as well as poor management of the plantations (Wallace 1853/1969, Bates 1864/19 62, Spruce 1908/1970) Fruit and seed yield were low in the dark, dank environment of t he cacao understory Cacao seeds, once harvested, were not dried properly and as much as half the crop was thrown out due to fungal infestations of damp seeds (Spruce 1908/1970) By the mid old cacao plantations were reduced to a tenth of their original size due to little effort for manag ing, pruning, or replanting (Penna 1869) Landholders in the vrzea today claim that cacao declined due to a combination of extremely high flood years that ki lled stems and economic demand for alternative forest products such as rubber and jute. The accounts of vast cacao plantations, bluff villages, extraction of riverine resources, and population crash of the giant river turtle in the Lower Amazon leaves the impression that the vrzea in the region of Santarm was heavily cultivated, harvested, and populated in the eighteenth and nineteenth centuries. However, there remains question as to how much of the vrzea forest was converted to cacao plantations and to what extent forests were maintained for forest products such as timber, fruits, and, after 1866, fuelwood for steamships. In 1862, Bates described the vrzea forest between
38 Santarm (River Maica) and the X (Bates 1864/1962: 147) Having passed Sa ntarm and headed to "bidos, Bates noted that the bank of the river was dotted with homes, each surrounded by cacao plantations. Above "bidos the forest along the river (Bates 1864/1962, Orton 1870) where the existence of man is forgotten and only wild paradise resides (Edwards 1847/2004) Whether or not these regions above "bidos and below Monte Alegre did have old growth vrzea forests, it seems that it is at these stretches of river that the landscape Botanical Observations by N aturalists, 1843 1850 The presence and use of vrzea forest species in the nineteenth century indicates that some of the same species are still common, while others have severely declined, pot entially due to increased flood levels. Spix and Martius (1823/1968) observed embauba ( Cecropia latifolia ), an d munguba ( Pseudobombax munguba ), the bark of which is stripped to make cords to pull canoes upstream. Some trees were suffocated by cucurbitaceous vines, a sight still common today. The great kapok tree ( Ceiba pentandra ), known locally as the samauma w as noted as a common and very distinct summer hats or to st uff pillows and mattresses. S amauma is now a very rare sight on the floodplain of the Santarm municipality. Re sidents of the Tapar region today know of two large trees in the entire municipality, and no seedlings have been obse rved in floodplains The abundance of Ceiba pentandra wa s noted repeatedly throughout the nineteenth century, particularly downriver of S a ntarm where cacao plantations we re rarely mentioned and perhaps less extensive (Spix and Martius 182 3/1968, Edwards
39 1847/2004) Another common vrzea species was pau mulato ( Calycophyllum spruceanum (Edwards 1847/2004) Twenty years later, this view has change d dramatically, as pau mulato became known as a prized wood for fuelwood for steamships (Sternberg 1975, Sears 2003) Today, Calycophyllum spruceanum has multiple uses among vrzea residents for fuelwood, construction material, and anti fungal treatment (Sears 2003) Black Gold in the Amazon: Rubber B oom of 1870 1911 The rubber boom had an immense impact on the Amazonian economy, culture, and development, particularly in urban centers such as Sant arm (Smith 1999) Rubber (Smith 1879) The impact of the rubber boom on land use, livelihoods, and forest structure in the vrzea may have been minimal, given the relatively small size and low density of rubber tree stands in comparison to the uplands and sites along Amazon tr ibutaries. The vrzea of the estuary was initially an important producer of rubber, but by 1874 rubber trees in the region suffered losses from over extraction, and rubber tappers were forced further upstream the Amazon to the Tocantins, Madeira, Purus, a nd Negro Rivers (Smith 1879) Despite the abundance of rubber plantations along the Tapajs River near Santarm, and the later development of rubber plantations in nearby Belterra and Fordlandia in the 19 20s, rubber trees in the vrzea of Santarm were not abundant (Penna 1869) Nonetheless, the social and economic changes associated with the rubber boom and b ust had long lasting effects on residents of the region. Cacao plantations, already in decline by 1870, were abandoned by laborers to tap rubber
40 and economic activity, a s cacao declined in the wake of rubber. Rubber also attracted colonists, particularly from the Northeast following the great drought of 1877 1879. When the rubber industry in the Amazon crashed in 1911, many colonists resettled in cities and on the vrze a floodplains (Smith 1999) Many vrzea residents today, including those in Santarm, can trace their roots back to Northeastern settlers that moved to the floodplains in the early 1900s (Smith 1999) Fibers from the Floodplain: J ute b oom of 1931 1990 Jute was a very important cash crop for riverine residents ( ribeirenhos ) in the vrzea of Santarm. Jute ( Corchorus capsularis ) was introduced into the vrzea floodplains of the central and lower Amazon River by the Japanese in 1931. Jute, a native herb of India, was cultivated for its fiber, used for making sacks for coffee and sugar (Smith 1999) The success of jute was attributed to its value as a cash crop for local residents as well as an economi c solution to the poverty that followed the rubber crash in 1911 (Medeiros 1968) Jute plantations in the vrzea of Santar m were planted on old cacao plantations or in newly cleared and burned forests. The plant was harvested post flowering when waters were rising (Medeiros 1968) By 1971 jute reached its peak production in Amazonia approximately 60,000 ha. of floodplain soils were in jute fields, most of which lied between Monte Alegre and Manacapuru (Smith 1999) Soon after, in the early 1980s, jute fibers were replaced by synthetic fibers, and residents witnessed yet another economic crash. Some middle aged vrzea residents recall harve sting jute with their parents at a young age and clearing old cacao plantations to plant jute. Today, jute can still occasionally be observed in small plantations in the Santarm municipality (Figure 2 1 ). One can also view the jute factory in Santarm, which was closed in the 1980s.
41 Rise in Cattle Ranching : 1970 present Although this chapter focuses on the land use and ecology of vrzea forests in the Santarm region, the vrzea is also composed of vast native grasslands. These grasslands support the growing cattle industry, which is a current economic boom in the vrzea of the region. While the economic and ecological impacts of cattle ranching on the floodplain and riverine residents are of major concern today, cattle ranching has had a long history on the floodplains. Cattle ranches in the vrzea were first established in the 1500s on Maraj Island at the mouth of the Amazon River (Smith 2002) In general, cattle ranches were separate from Direc torate villages, and were owned by Portuguese ranchers with African slave work hands (Anderson 1999) By at least 1846, cattle ranches were established o n the floodplains near Santarm (Edwards 1847/2004) [of Santarm ] support large herds of fat cattle, in every way superior to those of Maraj ; and were steamboat s plying upon the river, Santarm beef (: 106Edwards 1847/2004) Given the abundance of grasslands and the quality of beef, one wonde rs why cattle ranching did not become a dominant land use practice sooner, particularly after the arrival of the steamboat in 1866 to facilitate transport. In 1869, 40 landowners had 10,600 head of cattle in the municipality of Santarm (Penna 1869) The majority of these cattle were upstream of Santarm at Lago Sapucu (at the confluence of the Amazon and Trombetas Rivers), Lago Mariapixi, and Lago Grande do Curuai (Penna 1869) where today there remain large cattle ranches (M. Crossa, personal communication, June 1, 2007 ) The extent of cattle ranching in the nin eteenth and early success. For while annual floods nurtured abundant pasture grasses, they could also
42 have devastating impacts on cattle herds. Exceptionally high or rapi dly rising floods killed many cattle, increasing the economic risk associated with investment in cattle herds (Anderson 1999) Cattle also died from dise ase and snake bites, both exacerbated by the onset of the flood season. For example, the great flood of 1859 reduced cattle herds of 5 6,000 head to 100 300 head (Penna 1869) Ranchers also lacked access to boats to transport cattle to the uplands during the flood season. The conversion of floodplains to pasture and the intensification of cattle ranching constitute the most recent major cycle of land use trans formation in the vrzea of the Santarm region (McGrath et al. 1999b) Water buffalo and cat tle herds in the lower Amazon floodplains are increasing at annual rates of 12% and 4%, respectively (Sheikh 2002) Simultaneously, forest cover has decreased to approximately 10% of the area of vegetated floodplain, with the majority of the forested area being secondary growth following deforestation for jute, timber, and agriculture (McGrath 2005, Sheikh 2002). In the floodplains of Santarm, Par, deforestation and intensification of livestock husbandry have created conflicts among community stakeholders regarding management, land rights, and regulatory authority (Sheikh 2002) Conflict and trends in forest conversion are expected to increase with growing development pressures from the urban cente r of Santarm and the paving of BR 163. The capacity of institutions to inform decision makers in appropriate management strategies for the floodplain is hindered by the paucity of data on the ecology, land use change, and stakeholder values of the floodp lains resources Initial research on the impacts of cattle and water buffalo on floodplain forests suggests that expanding unmanaged water buffalo and cattle herds are a potential threat to floodplain forest regeneration (Sheikh 2002) Research by the Brazilian
43 i mpacts may be reduced through the application of management strategies, e.g. the exclusion of livestock by fencing, management of herd size and density, and removal of livestock during the flood season. However, few small scale residents have the capital for the necessary infrastructure. Recent community based initiative for livestock and forest management is demonstrated by local reforestation programs and the construction of fences to exclude livestock from forests that serve as fishing habitat. With f ishing providing 31% of family incomes in floodplain communities of the Lower Amazon, the sustainable management of fishing habitat is a specific goal of community residents and partner institutions (Almeida 2004). Summary The abundant natural resources of the vrzea ecosystem have supported Amazonians for millennia. Despite a number of dynamic changes in land use since European arrival, the ecosystem remains an essential resource as well as a biodiversity reserve for flora and fauna of the Amazon. The vrzea around Santarm has supported multiple cycles of agricultural production, from maize and manioc planted by indigenous peoples, the cacao boom of 1755 1870, rubber plantations of the late 1800s, to the jute boom of 1930 1980. Meanwhile, the floodpla in has consistently provided fishing habitat, non timber forest products, timber and fuelwood, as well as fertile pasture for small scale cattle ranching. The integrity of the vrzea today is challenged by the intensification of cattle and water buffalo h usbandry. The resilience of the system to land use change gives hope that minimal efforts for improved management could create an opportunity for forests and grasslands to maintain their naturally high productivity and diversity. While much of the intern ational attention for conservation and sustainable
44 management focuses on the vast forests of the uplands, it is important to recognize the economic, cultural, and ecological value of the Amazon floodplains. Their history of land use indicates that the vr zea is of immeasurable value for the future of Amazonia.
45 A B Figure 2 1 Jute cultivation on the vrzea of Ilha do S o Miguel A) Drying jute fibers. B) Retting jute plants in the water to separate fibers. Photos courtesy of Christine Lucas.
46 CHAPTER 3 EFFECTS OF SHORT TERM AND PROLONGED SATURATION ON SEED GERMINATION OF AMAZONIAN FLOODPLAIN FOREST SPECIES Overview Seeds in seasonally flooded forests have adapted to submergence, such that some species may require flooding to break seed dormancy. In this study, we test ed seed germination response to time in water among Amazonian floodplain species. To test short term effects, seeds from ten flood tolerant woody species were air dried or placed in water ( i.e. saturated ) for 45 h before germination. While non pioneer species increased germination success after saturation most pioneer species had higher germination success after air dry treatments. Long term saturation ( removing seeds from water at two week intervals over 12 weeks) rev ealed opposite responses among two flood tolerant trees with different dispersal strategies. Whereas wind dispersed Pseudobombax munguba seed germination rates decreased with increasing submergence time, water and fish dispersed Crataeva benthamii seed ge rmination rates increased with submergence time, peaking at 6 wks and maintaining > 80% germination for up to 12 wks. Overall, no species required prolonged saturation to germinate However, five of ten species required short term saturation to initiate g ermination and Crataeva bent hamii provides an example of increased germination success with prolonged submergence. Strategies for avoiding prolonged submergence among study species include d seed buoyancy, delayed fruit maturation, and dispersal mechanism These r esults suggest there are diverse strategies among Amazonian trees for colonizing floodplains
47 Background The alternating wet and dry periods of seasonally inundated wetlands are major driving factors for seedling recruitment (Grime 1979, Lopez 2001) Environmental cues for seed dispers al and germination such as moisture and oxygen availability are mediated by the timing of flood and low water seasons (Leck 1989, Stella et al. 2006) In many riparian wetlands, germination often occurs soon after flood drawdown on recently exposed soils (van der Valk and Davis 1978, Middleton 2000, Stella et al. 2006) Prior to germinati on, seeds can be subjected to variable periods of saturation and hypoxia, which may threaten seed viability (Baskin and Baskin 1998) Seed tolerance of flooding can be advantageous as hydrochory is an important means of dispersal in flooded ecosystems (Middleton 2000, Lopez 2001) Seeds of wetland plants have variable responses to flo oding, including enhanced germination (Martin et al. 1991, Jutila 2001, Cornaglia et al. 2009) reduced germination (Pierce and King 2007, Geissler a nd Gzik 2008) and delayed epi cotyl emergence (Scarano et al. 2003) Some species may require anoxia associated with flooding to break seed dormancy (Jutila 2001) However, such adaptations have been poorly studied in tropical wetlands, despite evidence that some woody species may require exposure to anoxia for germination (Kubitzki and Ziburski 1994) Seasonally flooded forests of the Amazon Basin have an extreme flood regime to which tropical tree species have adapted strategies for seed g ermination and establishment. Floodplain forests on the Amazon River are exposed to prolonged inundation of up to 7 months and average river level fluctuations of up to 10 m (Goulding et al. 2003) Fruit maturation for the majority of arboreal species occurs during the flood season (Kubitzki and Z iburski 1994, Waldoff and Maia 2001) Seeds
48 are subsequently dispersed by vectors including water, fish, wind, and other vertebrates (Goulding 1980, Moegenberg 2002, Lucas 2008) While seeds pass through the digestive tracts of fish in a few days (Silva et al. 2003) they may spend up to several months underwater in hypoxic conditions or (Kubitzki and Ziburski 1994, Lopez 2001) In the case of such extreme flooding conditions, some species may not only tolerate prolonged submergence but also require a period of hypoxia to break seed dormancy. Dormancy prior to hypoxic exposure may be advantageous for timing seed germination with flood recession and the onset of low water growing season. However, such a requirement would limit plant colonization of high elevation forests or micro topographic sites flooded sup ra annually. Despite the importance of seed germination as a bottleneck for establishment on the floodplain, little is known about the germination behavior of the >900 species of Amazonian floodplain forests (Wittmann et al. 2006) Furthermore, few data are available to test the hypothesis that tr opical species display dormancy that is broken by prolonged periods of submergence. Seeds that do not require exposure to anoxia may still require imbibition ( i. e. water absorption) for germination. Many tropical forests species lack dormancy and germinate quickly after imbibition presumably to avoid seed predation and fungal attack (Garwood 1989) Imbibition is critical for germination and early seedling survival for most species (Bradford 1995) and as such is a common treatment to prime recalcitrant se eds (Baskin and Baskin 1998) Imbibition is generally rapid for seeds with permeable seed coats, requiring 10 14 h to saturate tiss ues without resulting in anoxia related seed death (Baskin and Baskin 1998, Meyer et al. 2007) Similar to upland species, Amazonian floodplain forest species, particularly non pioneers, are predicted to be
49 recalcitrant ( i.e. unable to survive desiccation Garwood 1989, Sautu et al. 2006) and as such display enhanced germination after only short term imbibition. The purpose of this study is to test the effects of short term saturation and prolonged submergence on seed germination of floodplain tree s pecies and explore evidence for saturation breaking seed dormancy. I saturation both instances in which seeds are completely submerged or floating on the water surface, depending upon seed buoyancy and time in water Following s aturation treatments, seeds were placed on moist, sterile sandy substrates in temperature controlled germination chambers and monitored for time of radical emergence and full cotyledon expansion. I selected 10 woody species common to secondary floodplain forests on the Lower Amazon River to address the following questions: 1) What are the effects of short term saturation (45 h) vs. long term submergence (2 12 weeks) on seed germination rates and success? 2) Is there evidence for prolonged submergence to b reak seed dormancy among flood tolerant species? M ethods Study Region This study was conducted in floodplain forests within the municipality of Santarm, Par at the confluence of the Amazon and the Tapajs River s as vrzea are composed of grasslands, lakes, stands of giant aroids ( Montri chardia arborescens ), and forested levees (Deneven 1984). The vrzea is characterized by an annual, monomodal flood pulse, averaging 7.5 m in amplitude from 1975 to 2008 and peaking in mid May to mid June (Capitania dos Portos Santarm 2008) Mean annual
50 rainfall is 2100 mm y 1 (Fitzjarrald et al. 2008) with 80% fallin g during the rainy season (January June ; WinklerPrins 1999). Study Species I selected ten woody species common to mid low elevation floodplain forests of the Lower Amazon River (Sheikh 2002) Vitex cymosa, Crataeva benthamii Cordia tetrandra, and Cecropia latiloba are considered early mid secondary species (Worbes and Junk 1999) with foliar cotyledons and low shade tolerance. Vitex cymosa disperses buoyant diaspores during flood drawdown, with four embryos encased in a single buoyant cork like endocarp (Kubitzki and Ziburski 1994) Cordia tetrandra fruits throughout the flood season and disperses seeds with hard bony endocarps and a viscid mesocarp. Pseudobombax munguba is a widely distributed species of late secondary forests that disperses seeds during flood drawdown by kapok fibers that (Worbes and Junk 1999) Gar cinia brasiliensis and Maytenus sp. are shade tolerant mid late successional species with hypogeal storage cotyledons. Tabernaemontana sp. and Casearia aculeata and are shade tolerant forest understory species with dehiscent fleshy fruits. Collection and Experimental D esign Mature fruits of 10 species (hereafter referred to by genus, 3 1) were collected July, 2006. Maturity was re cognized by exocarp coloration or dehisc ence before falling into the river Fruits were transported in plastic bags to the Seed Laboratory at the Universidade Federal Rural da Amaznia (UFRA) in Santarm PA. Within 1 3 days after collection, seeds were e xtracted from fruits by washing with untreated (non chlorinated) well water in 0.5 m mesh size wire
51 strainers and dried in the shade for 1 h. A subsample of seeds was weighed and measured (length, width) with a digital caliper. Mature seeds without obvio us damage were randomly assigned to one of two treatments, 45 h air dried (dry treatment) and 45 h placed in water (wet treatment) in 11 x 11 x 3 cm plastic trays in a n air conditioned room (19 21C). Well water was added to half the trays to 2.5 cm depth to submerge non buoyant species, and changed after 24 h to prevent stagnation. Number of seeds per treatment per species varied between 13 and 100 seeds with two replicates (seed trays) per species (except for 1 replicate for Maytenus and Cordia ), as de termined by seed availability. Following treatments, smaller seeds, separated by species, were placed with sterilized tweezers in 11 x 11 x 3 cm plastic germination trays with a sterilized fine sand substrate. Larger seeded species, Garcinia and Crataev a were placed in 15 x 30 x 4 cm germination trays. Sand substrate had been previously washed and ster ilized in a drying oven at 150 C for 6 h. Prior to seeding, the substrate was saturated with distilled water (Baskin and Baskin 1998) Trays with seeds were placed in a germination chamber at 30C under florescent lighting. Seeds and sand substrate were watered every 1 2 days to av oid desiccation. Germination, defined as radicle emergence, was monitored daily during the first week, then every 2 4 days during a 145 day period (June Decemb er 2006). Germination success wa s the total percent seed germination of combined replicates. G ermination rate wa s the number of seeds germinated per unit time, or days to 50% germination success ( 3 1). Seeds attacked by fung i were washed with a 1% solution of HCl in distilled water for 30 min, and substrate was changed when necessary. Germinants were transferred to a greenhouse after full expansion of cotyledons.
52 To test the effects of prolonged intervals of submergence on seed viability, Cecropia, Pseudobombax Crataeva and Vitex seeds (diaspores) were placed in separate buckets with untreated well water that was changed daily. At two week intervals, 50 seeds of each species were removed and placed in groups of 25 into 11 x 11 x 3 cm plastic germination boxes with a sterilized fine sand substrate in the germination chamber, as described above. Seeds were removed every two weeks until germination success reached 0% or termination of the study at 12 weeks. Seed germination was monitored every 2 4 days as described above. Pseudobombax Crataeva and Cecropia seeds were also seeded in an open gr eenhouse at UFRA to measure potential differences in germination rate and timing in ambient conditions as opposed to indoor germination chambers. For each species, N=50 seeds were seeded on terra preta mixed with sandy soil in wooden boxes (65 x 40 x 12 c m). Seeds were monitored by the same protocol as in the lab, except germination is defined as cotyledon expansion, as radicle emergence was not visible. Statistical A nalyses Species based comparisons of percent germination were conducted by 2 sample test for equality of proportions (Crawley 2007) To test the significant difference between germination curves in response to soaking and dry treatments, time failure analysis was used where un germi nated seeds are considered right censored data (Traveset et al. 2008) Kaplan Meier survival probability functions were calculated using the survfit() function i n the survival package for R 9.2.0 (Therneau 1999, R Development Team 2006) To compare response curves for seed treatments within each species, log rank test statistics (Mantel and Haenszel 1959) and Gehan Wilcoxon
53 test statistics (Pet o and Peto 1972) were calculated using the survdiff () function in R 9.2.0 (Pyke and Thompson 1986, Therneau 1999) Results F ive of ten species increase d in germination success by 14 100% after 45 hour satu ration versus air drying ( Figure 3 1). Among those five, none of the air dried Crataeva Tabernaemontana and Maytenus seeds germinated, but saturated seeds achieved 69%, 62%, and 100%, germination success, respectively ( Table 3 1). Three of t he t en species ( Pseudobombax Cordia and Vitex ) had higher germination success (3 48% increase) following the air dry treatment. Among Pseudobombax seeds, the difference between germination success in the two treatments was only 3%, but time to 50% germinati on was reduced by 10 days. Finally, Cecropia and Laetia had poor germination success overall ( Table 3 1). Log rank test statistics show ed significant differences in wet vs. dry treatments among all sp ecies except Laetia ( Figure 3 1). Germination rates varied broadly across species, achieving 50% germination in 12 58 d among saturated seeds and 2 84 d among air dry seeds ( Table 3 1). In the prolonged submergence experiment, more Crataeva seeds germina ted with each additional two week period up to 6 weeks, and germination success after up to 12 weeks saturation remained > 80% ( Figure 3 2). Crataeva germination rates increased rapidly from 54 to 5 to 2 days with increasing submergence time of 2, 14, and 28 days, respectively. Pseudobombax seeds displayed similar germination rates and high germination success at > 90% after 0 and 14 d ays in water. In contrast Pseudobombax seeds were unviable (0% germination) after 28 d ays in water Vitex seeds decreased in germination rate and success (51% decrease) after soaking for 2 weeks. Cecropia had
54 low germination success ( < 20% ) across saturation times and no apparent trends with increasing duration in water. Log rank test statistics show ed sig nificant differences among germination curves within all species ( Figure 3 2). In greenhouse conditions with natural sunlight Crataeva seeds achieved 82 100% germination success after 45 h in water Saturation enhanced germination success by 24 32% in c omparison to air drying and, unlike in the germination chamber, air dry seeds germinated. Air dried Pseudobombax seeds increased germination by 12 28% relative to saturated seeds Saturated Cecropia seeds had higher germination success than in germinati on chambers (46% in 14 d), and air dried seeds had low germination success (< 20%). Discussion I found that short term imbibition enhanced germination of five of ten species I evaluated. This response pattern is similar to that of many tropical forest recalcitrant species (Sautu et al. 2006) Long term submergence was not required for germination, but rather enhanced germination of Crataeva and decreased germination of Pseudobombax and Vitex Only one of four flood tolerant species maintained high germi nation success over prolonged submergence, a potential strategy for maintaining seed viability over 1 4 months of submergence by Amazon River floodwaters. Pseudobombax and Vitex highly tolerant of flooding as adults, displayed decreased germination success following prolonged submergence. Those species likely avoid prolonged submergence in hypoxic floodwaters via delayed seed dispersal during flood drawdown and diaspore morphol ogy to promote buoyancy (Wittmann et al. 2007)
55 E ffect of Short Term Saturation o n Seed Germination Short term soaking treatments show ed that imbibition wa s sufficient to initiate seed germination among floodplain forest species. Sub mergence of seeds in oxygenated water is a common treatment for recalcitrant seeds of tropical upland forests (Baskin and Baskin 1998) As dryness is an important limiting factor for moist tropical tree seed germination and early seedling survival (Slot and Poorter 2007) germination is often delayed until seeds imbibe sufficient water above a minimum seed moisture content (Vozzo 2002, S autu et al. 2006) Imbibition is generally rapid for seeds with permeable seed coats, requiring 10 14 h to saturate tissues without resulting in anoxia related seed death (Baskin and Baskin 1998, Meyer et al. 2007) Seed soaking over short periods in oxygenated water, as conducted here likely increase s seed water content, while not inducing dormancy due to hypoxia. Pioneer and shade tolerant species may have respond e d differentially to saturation due to differences in germination requirements. Germination cues for tropical pioneers include irradiance and temperature, cues for the presence of gaps in the forest canopy Surface soils in gaps have low water availability (Marthews et al. 200 8) for which some pioneers have adapted high tolerance of low moisture content (Daws et al. 2007) In Amazonian floodplains, exposed soils high in sediment (>70%) and clay composition (10 33%; unpubl. data), may have low water availability at the surface, making germinatio n on drier soils a potential advantage colonization by early successional species. Seed germination among species with < 30% germination may have been reduced or delayed following saturation due to hypoxia while underwater insufficient irradiance, or lac k of other germination cues such as temperature change (Pearson et al. 2002)
56 Cecropia latiloba may have had exceptionally low germination rates due to low irradiance levels in germination chambers (Vazquez Yanes and Orozco Segovia 1990, Pearson et al. 2002) In a sunlit greenhouse Cecropia had higher germination success. Laetia corymbulosa displayed exceptionally low germination rates (4 7%), and may also require light for germination (Guariguata 2000). In another study, L. corymbulosa germination increased from an average of 6% to 39% after 35 days in water suggesting t hat this species may require prolonged flooding for germination (Wittmann et al. 2007, but see Kubitzki and Ziburski 1994) Our results indicate d a variation in species specific r esponses similar to upland tropical forests and suggest ed that many floodplain species have retained the ability to germinate without exposure to prolonged submergence (Scarano et al. 1997, Lopez 2001) Effects of Long Term Submergence o n Seed Germination Specific seed responses to prolonged submergence provide d an example of different germination s trategies of two common floodplain trees, Crataeva and Pseudobombax Crataeva show ed increasing germination rates and success with increasing submergence time, reaching peak percent germination (98%) after 6 weeks. Its s eed coat softened underwater, and in some cases split open to reveal the cotyledons, suggesting that soaking treatments may break physical barriers to germination. Enhanced germination success of Crataeva seeds after prolonged soaking may be attributed to morphological traits ( e.g., imperm eability) for endur ing long term submergence. In the field, Crataeva trees have non buoyant fruits and seeds that fall into floodwaters where they are consumed by frugivores including fishes (Goulding 1980, Lucas 2008) As such, seeds are underwater for up to 4 months In contrast Pseudobombax seeds perform ed best in air dry treatments. Pseudobombax seeds are
57 wind dispersed seeds that al so land on the water, by kapok fibers Due to rapid flood drawdown (2 6 cm per day; Capitania dos Portos Santarm 2008) during the dispersa l period seed submergence time may be minimal before germinating on exposed floodplain soils Only among Pseudobombax was long term saturation found to inhibit germination. Although hydrochorous and ichthyochorous, Vitex diaspores displayed higher germi nation success after drying. Vitex fruit s matur e during flood drawdown and diaspores are buoyant, which may limit submergence time. Whether species like Crataeva benthamii experience dormancy that is broken by submergence depends upon how dormancy is de fined (Baskin and Baskin 1998) Observations from this study suggest that Crataeva benthamii may have an impermeable seed coat that is slowly made permeable by prolonged submergence. Such traits may be considered a form of morphological dormancy (Baskin and Baskin 1998) Kubitski and Ziburski (1994) demonstrate that seed s of some species of Amazonian floodplains expos ed to an oxygen poor nitrogen atmosphere break physiological seed dormancy. Regardless of the mechanism by which flooding suppresses germination, species germinate d quickly when released from anaerobic conditions. In tropical floodplains, rapid germination maximizes seedling growth during the dry season (Wittmann et al. 2007) Concordantly, study species here all germinate d i n < 3 months. An alternative hypothesis for observed differences between seed germination in short term wet and air dry treatments may be that air dry seeds dehydrated. While the air drying treatment was considered a control for saturation the loss of mo isture to cool air may have adversely affected the germination of recalcitrant seeds. While the
58 reduced germination rates observed in Garcinia and Casearia could reflect desiccation it is questionable whether seed moisture content was reduced to critical levels of 30 70% (Baskin and Baskin 1998) While Garcinia can display decreased viability after partial dehydration (Rodriguez et al. 2000) other studies of Garcinia in ambient conditions support our findings of lower germination without prolonged submergence (Kubitzki and Ziburski 1994) The same study reports 92% germination for Crataeva seeds when removed from fruits, suggesting that the 0% germination of air dried seeds reported here may be a resu lt of dehydration or immaturity. Summary I found differential responses among species to saturation and submergence treatments. Species responses to short term saturation group ed by successional guild, whereby pioneers ha d enhanced germination rates and/or success in dry treatments, but shade tolerant species ha d enhanced germination after 45 h ours of saturation Prolonged submergence responses, however, differ ed among co occurring species of similar successional stages. Seed tolerance of prolonged submersion likely correspond ed with the timing of fruit maturation and dispersal mode. The observed variation in seed response to imbibition and submergence may explain niche differenti ation and diversity in floodplain forests. These findings have implications for both vrzea forest and fisheries restoration efforts. Floodplain forests are a pri ncipal food source for fishes throughout the Amazon Basin (Araujo Lima and Goulding 1997) I contribute to a growing pool of data on germination rates for vrzea forest species (Parolin 2001b, Wittmann et al. 2007) I also tested easi ly applied germination treatments of variable time in water that can be crucial for enhancing the germination success of some species. In particular, I report
59 germination rates for many seeds consumed by frugivorous fishes in the Amazon (e.g., Crataeva Pseudobombax, Cordia, Cecropia ; Goulding 1980, Claro Jr et al. 2004, Lucas 2008) Forest restoration by planting seedlings is one proposed way to increase local populations of Colossoma macropomum and Piaractus brachypomus two commercially valuable fish for the regi on (McGrath et al. 2005) Further studies on seed germination of the > 200 fruiting tree species consumed by Amazon fishes (Goulding et al. 1993) would be useful for forest and fisheries restoration projects.
60 Table 3 1 Seed characteristics of the ten study species in order of wet tolerant (higher germination after saturation) and dry tolerant (high germination after air drying) species. Species Family CM Mass (g) H x w (mm) G wet G dry Logrank test T50 wet T50 dry Wet tolerant species Garcinia brasiliensis Clusiaceae CH 2.93 27.8 x 14.6 1.00 (42) 0.86 (42) 2 =29.1, p<0.0001 58 84 Maytenus sp. Celastraceae CH 0.374 13.4 x 9.75 1.00 (15) 0 (13) 2 =293, p<0.0001 53 Crataeva benthamii * PE 0.290 9.16 x 4.67 0.69 (100) 0 (100) 2 =965, p<0.0001 47 Tabernaemontana sp. Apocynaceae PE 0.047 9.65 x 3.82 0.62 (55) 0 (55) 2 =406, p<0.0001 16 Casearia aculeata PE 0.011 3.80 x 2.83 0.23 (77) 0.04 (76) 2 =66.6, p< 0.0001 Dry tolerant species Pseudobombax munguba Bombacoideae, Malvaceae PE 0.045 5.32 x 4.01 0.97 (100) 1.00 (100) 2 =160, p<0.0001 12 2 Vitex cymosa PE 0.354 10.3 x 6.68 0.52 (42) 0.90 (42) 2 =11.1, p<0.0001 19 7 Cordia tetrandra Boraginaceae PE 0.139 0 (21) 0.24 (21) 2 =53.4, p<0.0001 Cecropia latiloba Urticaceae PE 0.002 2.53 x 1.42 0.18 (100) 0.15 (100) 2 =9.7, p=0.0018 Laetia corymbulosa PE 0.069 3.07 x 2.60 0.04 (100) 0.07 (100) 2 =0, p=0.901 Cotyledon morphology (CM) is indicated as phanerocotylar epigeal (PE) or cryptocotylar hypogeal (CH). Seed size is shown as wet weight (Mass) and height x width. Germination proportion (G) and time (days) to 50% germination (T50) are provided for wet and dry treatments, with sample sizes in parentheses. Log rank test statistics indicate differences in germination curves in time failure analysis. Families are shown according to the Angiosperm Phylogeny Group (Stevens 2001 onwards) Common names in order listed in table: bacuri, ashua, catauari, culho de porco, limorana, munguba, tarum, urua, embauba, and meracoroa. *Garcinia brasiliensis = Rheedia brasiliensis (Tropicos.org 2009) ** Crataeva benthamii = Crateva tapia (Cornejo and Iltis 2008) formerly Capparidaceae or Capparaceae formerly Flacourticaceae formerly Verbenaceae
61 Figure 3 1. Germination curves for ten study species with l og rank test statistics for differences in Kaplan Meier survival curves Species names indicated above (A J). A B C D E F G H I J
62 A B C D Figure 3 2. Germination curves and l og rank test statistics for significant (*** indicates p<0.0001) differences in Kaplan Meier survival curves for four study species (A. Pseudobombax munguba B. Crataeva benthamii C. Cecropia latiloba, D. Vitex cymosa ) subjected to prolonged waterlogging. X axis scale changes for each species, depending upon timing of termination of the experiment.
63 CHAPTER 4 E FFECTS OF MULTIPLE STRESSORS ON SEEDLING SURVIVAL AND GROWTH IN A TROPICAL FLOODPLAIN FOREST Overview Environmental stressors act synergistically to affect plant communities. Species specific responses to stressors mediate species composition and diversity across gradients of stress We investigate whether species show trade offs in stress tolerance or whether a few species emerge as tolerant of multiple stressors. I tested the main and interactive effects of flood duration and mechanical damage on seedling survival and growth of t en woody species of seasonally flooded Amazonian forests. Seedlings were planted in common gardens along gradients of flooding and light availability Half of the seedlings were clipped at 5 cm aboveground, removing ~ 50% total plant biomass. I found tha t damage severely limited growth and survival during a critical growth window in the low water season. After one year, clipping reduced seedling survival by an average of 50% and reduced relative growth rates such that most species failed to recover lost biomass. Flood duration of 3 6 months of submergence had no effect on seedling survival for all but two low flood tolerant species. Damage effects on seedling growth and survival were independent of flood duration. Increased light availability enhanced seedling growth and survival and ameliorated the negative effects of damage, particularly among pioneer species. Tropical flood tolerant species have a broad range of tolerance to mechanical damage, a key limiting factor for seedling persistence in forest s. D amage tolerance and flood tolerance among species was highly correlated suggesting that these combined stresses favor the persistence of a few species These results highlight the importance of examining multiple factors to understand seedling regen eration
64 Background Plants cope with a suite of environmental stresses that act synergistically to affect population dynamics. While the individual effects of stress on plants are well known (Kozlowski and Pallardy 2002) the interactive effects of multiple factors are an increasingly critical aspect of understanding plant communities (Chapin et al. 1987, Tockner et al. 2010) and predicting their responses to anthropogenic disturbance and climate change (Aber et al. 2001, Rhind 2009) Combined stresses can cause severe declines in plant growth (James et al. 2005, Ali et al. 2011) or survival (Niinemets 2010) Alternatively, spec ies adapted to one kind of stress can be tolerant of additional stresses (Myers and Kitajima 2007) Species di fferences in response to multiple stressors ultimately affect species composition and successional change in forests (Butler et al. 2007) The selective pressure of environmental stresses on trees can be greatest on seedlings, a critical bottleneck for species persistence in tropical forests (Grubb 1977, Poorter 2007) Flooding stress is a major driver of population dynamics in riparian and coasta l ecosystems. Research based on temperate species previously suggested that tree seedlings only tolerated short term submergence due to the build up of phytotoxins during anaerobic respiration (Kozlowski 1984) However, Amazonian tree seedl ings in seasonally flooded forests display remarkable tolerance to complete submergence and darkness for < 270 days (Parolin 2002, Parolin 2009) No temperate or boreal tree species are apparently able to withstand complete submergence for such prolonged periods as found in Amazonia (Junk et al. 1989, Mitsch and Gosselink 2000) There is little understanding of how seedlings tolerate such prolonged flooding (Ferreira et al.
65 2009) as well as how flooding interacts with other stresses prevalent in floodplain fore sts to affect seedlings (but see Wittmann and Junk 2003, Parolin et al. 2010) Damage to plants from herbivory, litterfall, and disturbances such as logging is pervasive in many ecosystems (Bond and Midgley 2001) Mechanical damage removes biomass and disrupts apical meristem dominance in tree seedlings, thus placing a stress on plants for additional resources for recovery. Seedlings are though t to be particularly vulnerable to damage as a result of their size, shallow root system, and limited reserves for tissue repair and defense (We isberg et al. 2005, Holdo et al. 2009) Species vary in sprouting ability to sprout as a result of seed size, cotyledon morphology (Green and Juniper 2004) and storage reserves (Canham et al. 1999) Large seeded species have enhanc ed reserves for sprouting after damage, thus increasing their survival in comparison to small seeded species (Harms and Dalling 1997) The sprouting ability of species in tropical floodplain forests is poorly known. The interactive effects of prolonged flooding and damage could have severe effects on seedlings. Seedlings in floodplain forests are affected by myriad damage agents including tr ampling and grazing by herbivores when floods recede (Robertson and Rowling 2000) Flooding stress limits growth by reducing carbon assimilation and potentially losing carbon to rotted roots or prematurely senesced leaves (Kozlowski and Pallardy 1997) As such, flooded seedlings may have reduced reserves available for tissue repair and defense after damage (Waring 1987) Although damage and flooding could decrease seedling survival, regeneration via sprouting is a survival strategy for many flood tolerant species (Frangi and Lugo 1991, Deiller et al. 2003, Ernst and Brooks 2003) Given the resource demands associated with flood tolerance and
66 sprouting, a trade off between flood and damage tolerance and may explain species differences in survival. Light availability acts synergistically with flooding to affect species survival in floodplain ecosystems. A flood /shade tolerance trade off model explains species di stribution in temperate floodplain forests, whereby high flood tolerant species tend to be light demanding and low flood tolerant species are shade tolerant (Hall and Harcombe 1998, Battaglia and Sharitz 2006) Increased light availability increases photosy nthetic activity in the growth season, thus increasing storage reserves and plant height for submergence tolerance in the flood season (Hall and Harcombe 1998, Parolin 2002) This model was based on observations that few shade tolerant species persist in dar k understories of temperate flooded forests (Walk er et al. 1986) Few studies in the tropics examine how light and flood level interact to affect woody plant survival (but see Wittmann and Junk 2003) I examined the main and interactive effects of flood duration, mechanical damage, and light availability on woody seedlings in a seasonally flooded forest of the Lower Amazon River. Stem clipping simulated one type of mechanical damage that occurs in forest understories. Using a series of common gardens along a flood gradient, I addressed three questions: (1) What are the effects of flood duration, light availability, and stem damage on seedling growth and surv ival?; (2) How does variation in flood duration interact with damage and light availability to affect seedling growth and survival?; (3) Are there trade offs in flooding and damage tolerance and flooding and shade tolerance that explain species differences in survival?
67 Methods Study Site This study was conducted in floodplain forests near Santarm, Par, Brazil Amazon Basin, with rainfall of 1800 2000 mm y 1 and 5 consecutive dry months with rainfall below 100 mm (July November; Sombroek 2001) Flood waters of the Am azon River in the region peak on average in mid May, rising 7 8 m and extending up to 40 km from the main channel. Peak flooding lags behind the mid rainy season by 1 3 months (Figure 4 1). Floodplains in the region are a mosaic of natural grasslands, for ested levees, lakes, and dense stands of a giant aroid ( Montrichardia arborescens ), all shaped by the annual flood regime. Many floodplain forests in the region are only ~20 80 years old, having regenerated on abandoned plantations of jute ( Corchorus caps ularis ), cacao plantations ( Theobroma cacao ), and stands of rubber trees ( Hevea brasiliensis ) (WinklerPrins 1999, Sheikh 2002) Cattle and water buffalo ranching are current ly prominent threats to floodplain forests in the region, as animals move through forests to reach grasslands for pasture, trampling soils and seedlings (Sheikh 2002) For this study I selected forests with medium to low cattle impact as indicated by cattle head counts and hoof print density (Sheikh 2002) to avoid highly compacted soils. Forest stands were 100 150 m wide and located on private property on 3 levees 1 5 km a part. Experimental D esign I used a regression design experiment (Cottingham et al. 2005) to measure changes in seedling growth and survival as a function of flood level and mechanical damage in three fl oodplain forests. Using a split plot design, 21 fenced plots of 5 x 5 m
68 (N = 7 per forest) were established in random locations across a flood gradient, and two treatments (damaged and undamaged) to were applied seedlings within plots. Plots were off tra ils and cattle were excluded from plots with barbed wire fences > 0.5 m from the nearest planted seedling. When plots were flooded by the river, I measured water column depth in plots at peak flood levels (15 May 2008) with a weighted line dropped into the water from a canoe. Relative flood level was calculated as the difference between maximum river level (8.36 m a.s.l. in 2008) and water column level (0.7 to 2.5 m in 2008). To calculate flood duration, I used daily river levels from Capitania dos Portos in Santarm to count the number of days for the river to rise from relative flood level of each plot to 8.36 m (peak) and then fall back to the same level Flood duration is the days seedling belowground tissues are waterlogged in saturated soil because the number of days of complete submergence varies with seedling height (see Results). Seedlings were germinated from seeds collected from nearby floodplain forests in the flood season (May July 2007) and grown in shaded, raised nursery beds in a mixture of composted cattle manure, palm fibers and floodplain soil. Ten species (after first mention referred to by genus) were selected that vary in flood tolerance, cotyledon morphology, and seed size (Table 4 1). After expansion of the first true leaves, seedl ings were transplanted to individual bags in a shaded forest understory and watered daily for 2 8 wk. After floodwaters receded, seedlings (N=2268) were transplanted into plots at randomly assigned locations 50 cm apart. Newly planted seedlings had 2 9 tr ue leaves ( 1 within species), mostly due to species differences in the timing of germination.
69 Three weeks after transplanting (October 2007), all seedlings were measured and a subset (N=210; one per species per plot) was harvested to measure initial b iomass. Unharvested seedlings were randomly assigned to a treatment, damaged or scissors (Green and Juniper 2004) and true leaves below 5 cm on the stem were removed. Clipping removed maximum aboveground biomass without removing the dormant buds at cotyledon nodes that provide seed lings with the physiological capability to sprout (Harms and Dalling 1997, del Tredici 2001) Foliar cotyledons (when present) were not removed as this could deplete carbohydrate reserves and reduce phanerocotylar seedling survival (Kitajima 2003) There were five seedlings per species per treatment per plot, except for Guarea guidonia for which there were four seedlings p er plot. Seedlings in both treatment groups were subjected to natural damage during the apical meristem appeared dry and a new dominant shoot sprouted. Measu rement of En vironmental C ovariates Light availability was estimated with hemispherical canopy photos taken at 50 cm aboveground at the center of each plot in the late dry season (November 2007) after most deciduous species had already flushed new leaves. Photos were taken just before sunrise and oriented to magnetic north with a Nikon Coolpix 990 and FC E8 Fisheye lens, with an F stop of 5.8 and automatic speed (Frazer et al. 2001) Photos were analyzed with Gap Light Analyzer Versi on 2.0 to calculate % transient total light [100*(transient diffuse + direct light)/(direct + diffuse radiation incident) as a measurement of light availability to seedlings, accounting for both direct light from overhead gap openings and diffuse light re flected off leaves (Frazer et al. 1999) Soil
70 cores were collected at four points 1.5 m from the plot center in the low water season (January 2008). At each point, litter dep th was measured with a ruler and averaged across the plot. The 7 x 10 cm soil cores were extracted with a PVC tube and rubber mallet and divided into two sections, 0 5 cm and 5 10 cm depth, placed in plastic bags, and weighed in the lab. A 25 g subsample from each core was dried at 105C for 48 h, and used to calculate volumetric soil water content, SWC (1 (Mass dry / Mass wet ), and soil bulk density (1 (SWC*Mass wet core )/Vol core ). Soil pH was measured with an Oyster portable pH kit and standard polymer p H electrode (Model #6015WC, Extech, Waltham, MA) from a 20 g subsample of wet soil mixed with 40 mL of deionized water. To measure soil texture, sand particles were extracted with a sieve (#230: 0.063 mm), silt was decanted, and the clay fraction was calc ulated by subtracting the sand and silt fractions from total dry weight of the sample in the LBA Santarm soil laboratory (Kettler et al. 2001, adapted by Beldini) Soil samples collected in September 2008 were sent to the Embrapa Belm soil laboratory in Brazil for soil nitrogen (organic N, NH 3 & NH 4 ) concentration, using the Kjeldahl method, and extractable soil phosphorous, using the Mehlich method (Embrapa 1997) Measurement of Survival and G rowth I measured seedling survival, height, and diameter at 5 week intervals for seven censuses when seedlings were above water (October 2007 February 2008 and August September 2008) to span a single monomodal flood pulse (Figure 4 1). Because plots emerged from flood waters at different times, the February and August censuses of 2008 were incomplete, excluding 7 and 9 plots, respectively, that were still under water. All seedlings were harvested after one year to measure final above and belowground biomass after drying to a constant weight Relative growth rate (RGR) was calculated
71 as [ln(biomass final ) ln (biomass initial )]/time for the change in total dry plant biomass from initial (October 2007) to final harvest (September 2008), where time = 11.25 months. Initial biomass for damaged seedlings was considered as biomass after damage. Initial biomass of unharvested seedlings was estimated by the linear reg ression equation for stem diameter/2) 2 ; r = 0.84). Tolerance and Resource A llocation T rade offs Trade offs between flood tolerance and shade tolerance as well as flood to lerance and damage tolerance were examined by comparing species survival according to the following definitions of tolerance. Flood tolerance was defined as the proportion of undamaged seedlings surviving the flood season. Damage tolerance was defined as the difference between seedling survival in damaged and undamaged treatments per plot during the low water season. For example, if the difference between survival of damaged and undamaged seedlings of a given species was 5%, it would be considered high d amage tolerant, while if the difference was 80% it would be considered low damage tolerant. A damage / flood tolerance trade off among species was explored by plotting the damage tolerance and flood tolerance of all species. Shade tolerance was defined a s the survival of seedlings in low light levels (9 12% canopy openness) in the low water season. A shade / flood tolerance trade off among species was explored by plotting shade tolerance and flood tolerance levels for all species. Trade offs between gro wth and survival were also explored by comparing RGR and percent survival of damaged and undamaged seedlings for each species (Kitajima 1994, Sack and Grubb 2001)
72 Statistical A nalyses The effects of damage, flooding, light, damage flooding and damage light on the proportion of survival within each plot were tested with logistic weighted generalized linear mixed models (GLMMs) with a binomial error structure and Laplace approximation s for model fitting (Crawley 2007) Models were weighted by sample size using a two vector response variable for survival number of successes (N live) and failures (N dead; Crawley 2007) This approach avoids errors associated with transformation and violation of model assumptions, such as normality and constant variance (Bolker et al. 2009) Forest and plot were included as random effects. As anova tables were not available for GLMMs with binomial err ors in R, I used the anova function with a Chi squared test statistic to test the significance of removing each parameter from the full model (Crawley 2007) The relationship between seed size a nd seedling height of the previous census and seedling survival was tested with binomial models. To explore the effect of flooding and mechanical damage on survival over time, the unequal time intervals between censuses due to flooding violated assumption s for standard survival analysis (Collett 2003) As such, survival is analyzed in three key survival windows: one year (October 2007 September 2008), low water season (October 2007 January 2008), and flood season (January August 2008). To test for differences between low water and flood season survival, I used a GLMM repeated measures model with season, flooding, damage, and light as fixed effects. To test for survival differences between censuses, a similar GLMM was constructed with census month instead of season as a fixed e ffect. Damage treatment was included as a random effect nested within plot in both models. Models for survival of pooled and
73 individual species were created in R 2.9.0 (R Team 2006) using the lme4 package (Bates and Maechler 2009) The effects of damage, flooding, light, damage flooding, and damage light on relative growth rates (RGR) were analyzed with linear mixed models, with plot and f orest as random effects, using the nlme package (Pinheiro et al. 2008) The same models were created for other growth response variables root:shoot ratios, post flood shoot biomass, and seedling height for comparison. To determine if seed size correlated with averag stem diameter/2) 2 ) of sprouts within species, I moment correlation coefficients (Crawley 2007). To compare average initial and final root biomass within species to determine if root biomass of damaged seedli ngs changed over a year, I used two way pair wise t tests. All analyses were conducted using R 2.9.0 (R Team 2006) Results Effect o f Flooding and Damage on Annual Seedling Survival S eedling survival after one year for pooled species was decreased by damage, but no effect s of flood duration or a flood damage interaction were found ( Table 4 2) Damage decreased average seedli ng survival after one year by 29 % (z = 4. 3 p < 0.0001, Figure 4 2), but differences between undamaged and damaged seedling survival varied widely among species (12 75%; Table 4 1). Damage decreased flood season survival by almost half (43%), whereas undamag ed seedlings lost only 34% of individuals (t 9 = 3.7 p = 0.0047). Survival differed by species ( Table 4 2), supporting species specific models ( Table 4 3 ) Although species cotyledon morphology helped explain one year survival ( 2 = 259, p < 0.0001), the effect disappeared when removing the three low flood tolerant species, all with hypogeal cotyledons ( Hevea, Guarea and
74 Ormosia ) Species with large s eed s did not display enhanced survival after damage ( r = 0.33, t = 0.98, p = 0.35). D amage consistently decreased survival within species after one year, with the exception of Coccoloba 2 = 7. 2 p = 0.066) and Hevea for which all individuals died during flooding Seedlings in treatment group experienced natural damage to stems During the low water season, 3% (N=31 of 1117 ) of undamaged seedlings had dry ap ices and 1% had broken stems. In the flood season, an additional 14 of 579 (2%) seedlings displayed dried apices and 31 (5%) broken stems. Flood duration (113 208 days) had no app arent effect on pooled species survival over one year ( Table 4 2) Given an average height of 25 and 12 cm for undamaged and damaged seedlings after one year, the average estimated duration of entire seedling submergence was 100 200 d and 107 200 d respe ctively. During flooding, 5 species senesced leaves ( Vitex, Guarea, Cordia, Hevea Ps eudobombax; Figure 4 3) and 5 were evergreen. I found n o interactive effects of flooding and damage on seedling survival among or within species ( Table 4 2). Within spe cies, flooding had no effect on survival, except for Guarea and Ormosia which show ed decreasing survival with incre asing flood duration, reaching 0% survival at approximately 160 days ( Figure 4 4 Table 4 2). Species differed in response to flooding amon g damaged seedlings ( 2 = 402, p < 0.0001) but not among undamaged seedlings ( 2 = 5.4, p = 0.49). Seasonal Differences in S urvival Survival differed by season (z = 16.6, p < 0.0001) whereby average low water season survival ( November to January; 0.90 % 0.37 SD) was higher than flood season survival ( February August; 0.77 % 0.23 SD ; Figure 4 5 ). While undamaged seedlings suffered decreases in survival only during the flood season ( Figure 4 4 ) damaged
75 seedlings suffered decrease s in survival in both the dry period (November December ) and flood season ( August Table 4 3 ) The drop in damaged seedling survival in December occur red at the end of the peak dry period in the region, following three consecutive months of rainfall < 100 mm (August Novembe r 2007 ; Figure 4 5 ). Despite the potential for low water availability in dry upper soil horizons to decrease damaged seedling survival, t here was no correlation between survival and soil water content in 2 = 1. 3, p = 0.26; range: 21 37% in soils of 3% sand, 79% silt, and 18% clay, on average) Flood duration had a marginally significant effect on flood season survival 2 = 5.8, p = 0.055 ). There was a flood damage interacti ve effect on flood season survival 2 = 5. 6 p = 0.018 ; Fig ure 4 2 ). Within species, Cordia, Mouriri, and Trichilia show ed an interactive effect in which damage enhance d the deleterious effects of prolonged inundation on flood season survival ( Table 4 3 ). Effect of Light on Survival Light, measured as percent can opy openness, was consistently the most important environmental factor for predicting seedling survival. The positive effects of light availability on one year survival were greater for damaged than undamaged seedlings ( Table 4 2, Figure 4 2 ). L ight enha nced the survival of pioneer species with epigeal cotyledons ( e. g. Cordia and Pseudobombax ), but not for most shade tolerant species ( Figure 4 6 ). In addition to canopy openness models were improved by 2 2 = 4.6, p = 0.031, Litter depth was particularly important for explaining low survival among damaged seedlings, which were more vulnerable to burial (esti mate = 0.54, z = 2. 6 p = 0.0098).
76 Effects of Damage a nd Flooding o n Relative Growth Rates Among species, both damage and increasing flood duration reduced RGR (Figure 4 7) There was no apparent interaction between damage and flood duration for pooled s pecies (F = 0.06, p = 0.80) and within most species ( Table 4 4 ) Species had a strong effect on RGR (F = 6.0, p < 0 .0001 ). Within all surviving species except Garcinia damaged seedlings had low RGR, indicating that most species failed to recuperate lost biomass in one year Increasing flood duration decreased RGR among five species ( Table 4 4 ), and fast growing species such as Cordia and Vitex had steeper declines in RGR across the flood gradient than slow growing species ( Garcinia and Trichilia ). Pioneer species ( Pseudobombax, Cordia ) experienced growth spurts immediately after flood drawdown. Effects of Damage and Flooding o n Biomass Allocation Clipping removed an average of 50% (73% SD) of total plant biomass ranging from 19% (7%) in Garcinia to 65% ( 31% SD) in Vitex ( Table 4 5 ) Seedlings responded to damage by sprouting a median of 1 ( range 0 4 ) aboveground shoots, generally from the cotyledon node or leaf node closest to the point of damage. The biomass of newly sprouted shoots 5 weeks after damage did not correlate with seed size (r = 0.19, t 8 = 0.54, p = 0.60) nor after 1 yr (r = 0.4 5 t 6 = 1.2, p = 0.26) indicating that seed size does not explain new shoot biomass The biomass of new shoot s, however, did correlate with initial plant size (r = 0.98, t 347 = 102, p < 0.0001). Damaged seedlings had higher root:shoot ratios one year after damage than undamaged seedlings ( Table 4 6 ) but these effects were significant within only four species ( T able 4 4 ). As this difference was likely a result of the damage treatment and not differential allocation to roots, I compared estimated initial and final root biomass for all species.
77 Only for Coccoloba, Garcinia and Trichilia were roots significantly bigger after one year (t = 2. 5 p = 0.015; t = 6. 9 p < 0.0001; t = 3. 6 p = 0.0007, respectively). Root:shoot ratios were the only biomass measurement to display a strong damage flood duration interaction (F = 9.6, p = 0. 00 6 ). R oot:shoot ratios among damaged seedlings were greater than those of undamaged seedlings across the gradient of incre asing flood duration ( Figure 4 8 ) A positive correlation between root:shoot ratios and flood duration was observed among five species but the inte ractive effect was observed within one species ( Table 4 7 ). Shoot extension in the two months after flood recedence was also affected by an interacti on between flooding and damage (F = 7. 2 p = 0.016), whereby undamaged seedlings were able to allocate mor e biomass to aboveground shoots at higher elevations than damaged seedlings. Effect of Light on G rowth and Biomass A llocation Light availability increased RGR (F = 11.3, p = 0. 0039 for pooled species; Table 4 4 for within species). However, light did not affect root:shoot ratios, except for two species ( Vitex and Coccoloba, Table 4 7 ) for which shoot biomass was relatively higher with high light availability Neither RGR nor root:shoot ratios were affected by an interaction between light and damage ( Table 4 6 ). Incorporating al l independent environmental variables into a linear mixed model, manual backward stepwise selection revealed that RGR is a function of damage (F = 1 90 p < 0.0001); canopy openness (F = 6 9 p < 0.0001); flood duration (F = 20.9 p = 0.0018); litter depth (F = 10.3, p = 0.012); and nitrogen litter depth (F = 28 .0 p = 0.0007). Species Trade o ffs for Flood, Damage, and Shade Tolerance There was a positive correlation between flood tolerance and damage tolerance among species ( r = 0.84, Figure 4 9). Highly flood and damage tolerant species
78 Coccoloba and Garcinia, both had relatively slow growth rates and high root:shoot ratios. Garcinia was highly damage tolerant and had the lowest percent biomass removed by clipping. A posit ive correlation between shade tolerance and flood tolerance was also found (r = 0.64, Figure 4 10). Species with high survival in low light conditions also displayed high survival after flooding. Coccoloba and Trichilia, two large woody shrubs, were high est in shade and flood tolerance. A negative relationship between survival and growth among undamaged seedlings was observed, such that species with rapid RGR had lower survival rates (Figure 4 11). This trend, however, was largely driven by two extreme s: the rapid RGR / low survival of Ormosia and Guarea and the slow RGR / high survival of Coccoloba. In contrast, damaged seedlings showed the opposite trend, with survival increasing with RGR (Figure 4 11). For example, although Coccoloba was the most t olerant of damage in terms of survival, it ranked 5th in RGR of resprout biomass. Vitex, a damage tolerant and light demanding species, also had high root:shoot ratios and slow RGR as well as early leaf senescence (Figure 4 3). The average RGRs of undamag ed and damaged seedlings within species were not correlated (F=0.15, p=0.70, R2= 0.11), indicating different rates of growth vs. re growth following damage. Discussion The purpose of this study was to explore how flood tolerant seedlings respond to mechani cal damage along gradients of flooding and light availability. I found that damaged seedlings have lower survival than undamaged seedlings in the flood season, but the response of damaged seedlings to flooding is not dependent upon flood duration. The in dependence of the effects of flood duration and damage after one year for survival and growth implies that damage does not affect seedling tolerance to long
79 periods of anoxia and darkness (Parolin 2009) Complete submergence for 3 6 months did not affect seedling survival, exc ept for two species with low flood tolerance. In contrast to flooding, damage substantially reduced seedling survival and growth, such that most species were unable to recover lost biomass within one year of damage. However, more open canopies alleviated the negative effects of damage on growth and survival, particularly among fast growing species. Species varied widely in their response to damage, but I did not find a trade off between damage and flood tolerance nor between shade tolerance and flood tol erance. Rather, a few species emerged as highly tolerant of both prolonged flooding and shade. These results are among the few available to understand how such extremely flood tolerant tree seedlings respond to damage incurred during the critical growth season. Independence of Effects o f Flooding a nd Damage o n Growth a nd Surviv al In contrast to the growing evidence for the importance of interactive effects of stress and disturbance in plant population dynamics, I found little evidence to support the hypothesis that mechanical damage interacts with flood duration to affect seedling growth and survival. Damaged seedlings were more susceptible to death during the flood season, although their susceptibility did not in crease at extremely prolonged flood periods (180 207 d). The time scale over which I measured interactive effects in this study is important. During the flood season, damage interacted with flood duration to affect seedling survival, but this effect was damped across a year, such that there was no effect over one year. Seedling persistence under the sequential perturbations of damage and flooding may be attributed to the 4 to 5 month gap between damage and flooding events. A single damage event during the early growth season allows some recovery of photosynthetic tissue, shoots, and reserves that may later facilitate flood
80 tolerance and diminish submergence time. Floodplain forest seedlings are likely subject to multiple damage events from herbivory, b rowsing, litterfall, and trampling. The effect of repeated damage events throughout the growing season could select for highly damage tolerant species and potentially reduce seedling community diversity. Repeated damage to woody plants by ungulate browsi ng in an Alaskan riparian forest was found to interact with erosion and accretion of flooded areas to alter successional trajectories (Butler et al. 2007) Damaged seedlings may have maintained constant survival rates with extremely prolonged flooding due to the relative increase in root:shoot ratios with increased flood duration (del Tredici 2001) The increase in root:shoot ratio among damaged seedlings is likely a consequence of removal of aboveground biomass for most species, rather than active investment in root biomass a s a response to clipping. Among the species ranking top four in survival, three displayed an increase in allocation to roots among damaged seedlings over one year. While root biomass is not directly correlated with storage reserves (Canham et al. 1999) the relative increase in root biomass may indicate allocation to storage reserves. Future studies should investigate carbohydrate reserve availability in flood tolerant seedlings and allocation to sp routs vs. anaerobic respiration and adventitious root production during flooding. Strong E ffects of D amage on G rowth and S urvival Damage is pervasive in tropical forest understories (Clark and Clark 1989, Alvarez Clare and Kitajima 2009) Although the role of damage in seedling communities is less known for tropical floodplain forests, t his study shows that damage differentially affects species growth and survival (Fine et al. 2004) Given reductions in biomass, height, and the vulnerability of small stems to burial by litter and sediment
81 deposition, damaged seedlings were predicted to have a severe disadvantage in survival compared to undamaged seedlings. Although damage substant ially reduced survival, responses varied widely among species (12 75% survival). Species differences in survival of damaged seedlings did not support the hypothesis that large seeded species with belowground storage cotyledons would have higher survival (Harms and Dalling 1997, Green and J uniper 2004, Baraloto and Forget 2007) Given that damage was applied at 4 9 weeks after germination, seedling dependence on cotyledon reserves for new shoots may be low (Kitajima 1996, Myers and Kitajima 2007) Storage reserves in root and stem tissues after damage could explain species differences in survival (Myers and Kitajima 2007) as species with higher root biomass and slower relative growth rates tended to have higher survival after damage in this study. Oddly, some species with epigeal cotyledons also displayed high root:shoot ratios and low RGR, contrary to the typical allocation patterns among upland tropical seedlings (Paz 2003) relatively low light levels in the forest understory and unobserved additional damage or disease defense in damaged seedlings. Overall, species variation in survival suggests t hat mechanical damage agents, when present, are an important selective force on species composition and distribution in floodplain forests. The growth and survival advantage for seedlings in high light conditions has led to the hypothesis that increased light availability permits increased establishment by flood tolerant species (Battaglia et al. 2000, Parolin 2002) Light enhances resource availability nece ssary for flood tolerance (Hall and Harcombe 1998) and damage recovery (Kabeya and Sakai 2005) In this study, survival and growth after damage
82 were enhanced by light availability, particularly for light demanding species. Moreover, high light availability decreased the difference in survival and growth between damaged and unda maged seedlings of light demanding species. Among mid late successional species with storage cotyledons, the effects of light and damage were independent ( Baraloto and Forget 2007) High light availability makes damaged seedlings less susceptible to death likely via increased carbon reserves for damage recovery (Kabeya and Sakai 2005) However, a canopy openness of 9 19% was still insufficient for seedlings to recover a positive carbon balance in one year after stem loss. There was substantial decrease in survival of damage d seedlings during the peak dry season. Drought is a potentially important factor for floodplain seedling survival (Parolin 2001a, Lopez and Kursar 2007, Parolin et al. 2009) but I were unable to provide direct evidence for limited water availability to seedlings. Following three consecutive months of extremely low and infrequent rainfall (Figure 4 1), damage d seedlings could experience drought stress that reduced survival during the peak dry season. Although variation in soil water content did not explain dry season survival, the low rainfall and low water availability of silt loam soils with 10 33% clay fra ction (Brady and Weil 2000, Guyot et al. 2007) could limit water available to shallow roots. Nonetheless, the dry season decline in survival is confounded by the effect of time, as seedlings are most vulnerable to mortality at e arly stages in development (0 2 months) (Alvarez Clare and Kitajima 2009) Flood Duration Effect on Seedling S urvival Based on species zonation along a flood gradient, variable flooding stress should differentially affect seedling surviva l among species (Wor bes et al. 1992) While species differed in overall survival, I found no effect of increased flood duration on
83 se edling survival except for low flood tolerant species Guarea and Ormosia and the submergence intolerant Hevea For Guarea and Ormosia there was a distinct threshold for survival at ~140 160 d flooding. Flood tolerant seedlings of the Amazonian floodplain species Himatanthus sucuuba survived prolonged flooding by maintain ing high alcohol dehydrogenase (ADH) activity in root tissue, whi ch prevented the build up of toxic acetaldehyde during anaerobic respiration (Ferreira et al. 2009) The lack of an effect of flood duration on seedling survival in most species may simply indicate their high flood tolerance and broad species ranges across the flood gradient. Given the changes in climate and topography on a geological time scale (Hoorne 2006) many species in the floodplain may have adapted to very broad ranges in flood duration. Species R esource Allocation T rade o ffs Species trade offs in resource allocation are thought to be a key mechanism for maintaining species diversity in tropical forests (Wright 2002) Species should display different strategies for survival and growth in response to many stresses for which seedlings have limited reserves to permit tolerance (Grime 1977) If flooding and damage place competitive demands on plant resources, species should display a trade off between damage and flooding tolerance. In this study, ten species show ed no trade offs to support this hypothesis. Rather, a few species emerge as highly tolerant of both flooding and damage. Those species share traits of high root:shoot ratios and low RGR, in concordance with other studies showing that high allocation to belowground reserves is advantageous for both damage and flood tolerance (Kozlowski 1984, del Tredici 2001, Myers and Kitajima 2007) If belowground reserves foster damage and
84 stress tolerance, then such an attribute may represent a single strategy to cope with a combination of s tress and disturbance (Craine 2005) Species differences in surviv al suggest tolerance to multiple stressors among few species. There was little evidence for a flood /shade tolerance trade off whereby species with high flood tolerance have low shade tolerance, while low flood tolerant species have high shade tolerance. Such a trade off has explained why shade tolerant species occur at higher elevations (Battaglia and Sharitz 2006) and why fewer species occur in some shady, low elevation floodplains (Menges and Waller 1983) Our results (Battaglia et al. 2004) and that survival varies widely according to multiple abiotic conditions as well as random events such as litterfall or herbivory. Species also did not sort in survival in the shade according to their physiological shade tolerance (Hall and Harcombe 1998) For example, Coccoloba a species with epigeal cotyledons, had extremely high shade tolerance, while Hevea, with storage cotyledons, had low shade tolerance. Such responses suggest that species sort in survival primarily by flood tolerance, and that high flood tolerance may permit persistence in the shade. In Amazonian floodplain forests, very few species have fast growth rates sufficient to avoid submergence by floods of 1 4 meters depth in forests (Parolin et al. 2002) A tolerance vs. escape model proposes that species at lower elevations must have strategies to tolerate submergence at the seedling phase, while species at higher elevations may escape submergence by long stem growth i n the first growth season (Parolin 2002) Our results support such a model, whereby species at low elevations are tolerant of flooding and shade.
85 A growth survival trade off explain ed species differences in survival among undamaged seedlings. Species should fall along a continuum of fast growing pioneer species with low survival to slow growing, long lived species with high survival (Kita jima 1994, Wright 2002) However, in this study, species were not positioned along the growth survival continuum in order of shade tolerance. Given that survival during flooding is not a function of cotyledon morphology or RGR (Parolin 2002, Battaglia et al. 2004) I might expect species position on the survival growth continuum to be a function of some measureable value of flood toleranc e ( e.g., ADH production). In contrast to undamaged seedlings, damaged seedlings with higher RGR also displayed higher survival. In the case of damaged seedlings, high RGR could increase storage reserves available for survival and recovery after damage (Myers and Kitajima 2007) Summary The goal of this study was to explore the main and interactive effects of fl ooding and mechanical damage on seedling population dynamics and to compare species specific responses. Given the potential frequency of damage events in forest understories damage tolerant species are likely to persist ultimately influencing succession al trajectories and adult species composition (Grubb 1977, Bond and Midgley 2001, Laura nce and Curran 2008) The interactive effect of flooding and damage w as significant only in the flood season. The obser ved seedling responses to damage are thus influenced by the time of measurement, a nd likely by the timing of damage and stress events. T he se results suggest substantial resilience among stress tolerant seedlings in disturbed and stressed environments. Th e vast majority of seedlings in tropical forest s suffer damage before reaching juvenile stages/sizes (Kammesheidt 1998, Scariot 2000) Given the variable intensity of flooding and drought
86 (Schongart and Junk 2007, Marengo et al. 2008) overlapped by disturbance from wind, fire, logging, herbivores, and introduced ungulate s in many tropical floodplain ecosystems (Anderson et al. 1999, Finlayson 2005, Junk and de Cunha 2005) spro uting may be a key strategy for seedling survival (Kozlowski 1992)
87 Table 4 1. Physiological traits of ten floodplain forest study species. Species Family Cot Seed (mg) Survival (%) RGR Root:shoot Undam. Damaged Undam. Damaged Undam. Damaged Coccoloba ovata Polygonaceae PEF 39 94 13 82 21 0.044 0.015 0.84 0.05 0.96 0.07 Ps e u dobombax munguba Malvaceae (Bombacoideae) PEF 45 74 28 29 37 0.102 0.027 0.51 0.11 0.64 0.10 Vitex cymosa Lamiaceae PEF 350 87 17 61 36 0.048 0.017 0.71 0.16 1.07 0.15 Cordia tetrandra Boraginaceae PEF 140 81 29 29 29 0.084 0.018 0.80 0.02 1.14 0.05 Ormosia paraensis Fabaceae P ES 600 14 29 0 0 0.164 0.37 0.06 Mouriri acutiflora Melastomataceae CHS 230 79 22 4 10 0.055 0.063 0.47 0.06 1.05 0.10 Guarea sp Meliaceae CHS 580 11 23 2 7 0.084 0.068 0.82 0.05 0.97 0.37 Trichilia singularis Meliaceae CHS 390 85 14 56 32 0.075 0.006 0.62 0.06 1.16 0.11 Garcinia brasiliensis Clusiaceae CHS 2930 86 20 59 29 0.078 0.031 0.55 0.06 1.19 0.13 Hevea brasiliensis Euphorbiaceae CES 3930 0 0 1 4 Seed size is presented in fresh weight (mg). First year survival (mean & SD) is the per cent of undamaged seedlings and of dam aged seedlings averaged across 21 plots with standard deviations. Abbreviation s for Cotyledon morphology (Cot ): Phanerocotylar epigeal foliar (PEF), Cryptocotyla r hypogeal storage (CHS), Cryptocotylar epigeal storage (CES). Mean relative growth rates over one year [RGR (gg 1 cm 1 )] and a verage root:shoot ratios (mean & SD) harvested at the end of the experiment are shown.
88 Table 4 2. Survival response to flood level, damage, and light in three time periods: one year, low water season, and flood season. Results shown for pooled species and individual species. Species Damage F lood duration Light Flood x damage L ight x damage 2 p 2 p 2 p 2 p 2 p One year 186 <0.0001 3.37 0.19 15.05 0.0005 2.8 0.094 5.58 0.018 Low water season 386 <0.0001 1.00 0.60 10.53 0.0052 0.054 0.82 0.097 0.76 Flood season 172 <0.0001 5.81 0.055 10.74 0.0047 5.56 0.018 2.27 0.13 Coccoloba e 7.19 0.066 0.05 9 .98 3.64 0.16 0.042 0.84 0.14 0.71 Pseudobombax .e 64. 1 <0.0001 0.77 0.68 9.03 0.011 0.26 0.61 1.06 0.30 Vitex e 29.0 <0.0001 4.94 0.085 22. 4 <0.0001 0.86 0.35 8.40 0.0037 Cordia e 78.5 <0.0001 2.10 0.35 6.51 0.039 2.08 0.15 0.81 0.37 Ormosia e 30.9 < 0.0001 8.76 0.013 4.36 0.11 <0.00 1 1.00 <0.00 1 1.00 Mouriri h 142.6 <0.0001 4.45 0.11 1.24 0.54 0.87 0.35 1.11 0.29 Guarea h 8.25 0.041 6.82 0.033 1.00 0.61 0.26 0.61 0.003 0.96 Trichilia h 31.5 <0.0001 1.91 0.38 8.61 0.013 0.91 0.34 1.97 0.16 Garcinia h 21. 6 <0.001 5.61 0.060 1.24 0.54 0.75 0.39 0.24 0.62 Chi squared test statistics with p values for pair wise model comparisons between the full model with and without each parameter removed are shown Cotyledon morphology is displayed as a subscript (e = epigeal, h = hypogeal) for each genus. Hevea is excluded from individual species due to 0% survival after one year
89 Table 4 3 Summary of generalized mixed model results for t he effects of damage, flooding, light, and their interactions on see dling survival in the low water and flood season. Species Damage Flooding Light Flood x damage Light x damage 2 p 2 p 2 p 2 p 2 p LOW WATER SEASON Coccoloba e 4.73 0.19 2.65 0.27 8.82 0.012 0.89 0.34 3.95 0.047 Pseudobom. e 5 6.0 <0.0001 3.77 0.15 5.99 0.050 1.57 0.21 0.028 0.87 Vitex e 13. 4 0.0039 5.79 0.055 15. 2 0.010 2.31 0.12 4.77 0.029 Cordia e 43. 8 <0.0001 5.37 0.068 8.15 0.017 1.37 0.24 1.54 0.21 Ormosia e 181 <0.0001 4.78 0.09 1.97 0.37 0.70 0.40 0.078 0.78 Mouriri h 127 <0.0001 2.47 0.29 2.50 0.29 2.47 0.12 2.17 0.14 Guarea h 50.0 <0.0001 0.34 0.84 2.32 0.31 0.32 0.57 0.0054 0.94 Trichilia h 14.5 0.0023 5.11 0.078 0.61 0.74 0.94 0.33 0.0079 0.93 Garcinia h 1.99 0.58 2.95 0.23 0.52 0.77 0.12 0.73 0.044 0.83 Hevea h 61. 9 <0.0001 2.13 0.34 1.06 0.59 0.55 0.46 0.032 0.86 FLOOD SEASON Coccoloba e 16. 3 <0.0001 3.85 0.15 0.29 0.86 2.46 0.12 0.06 0.80 Pseudobom. e 41. 9 <0.0001 0.86 0.84 7.29 0.063 0.77 0.68 0.35 0.84 Vitex e 32.7 < 0.0001 4.51 0.21 22. 5 <0.0001 0.44 0.80 4.42 0.11 Cordia e 69.4 <0.0001 11.7 0.008 11.4 0.010 10.85 0.004 9.98 0.006 Ormosia e 24.3 <0.0001 6.94 0.074 2.76 0.25 0.050 0.98 0.050 0.98 Mouriri h 68.9 <0.0001 11.7 0.009 2.22 0.53 7.75 0.021 1.88 0.39 Guarea h 6.6 2 0.16 6.85 0.077 3.05 0.38 1.17 0.56 2.38 0.30 Trichilia h 34.1 <0.0001 6.58 0.087 14. 7 0.002 6.51 0.039 5.31 0.070 Garcinia h 20.8 0.0003 8.34 0.039 1.57 0.67 1.04 0.60 1.05 0.59 Chi squared test statistics with p values are shown for pair wise model comparisons between the full model with and without each parameter removed Cotyledon morphology is displayed as a subscript (e = epigeal, h = hypogeal) for each genus. Hevea is excluded from B due to 0% survival after one year.
90 Table 4 4 Summary of generalized linear model results (F tests & p values) for the effects of flooding, damage, and canopy openness (light) on seedling relative growth rate (RGR). Species Damage Flood duration Light Flood x damage Light x damage F p F p F p F p F p ALL 177. 5 <0.0001 10. 9 0.004 11.3 0.004 0.065 0.80 1.48 0.24 Coccoloba e 71.5 <0.0001 3.9 0.065 8.06 0.012 0.010 0.92 0.88 0.36 Pseudobom. e 93.9 <0.0001 4.4 0.053 9.1 0.008 4.69 0.058 1.56 0.24 Vitex e 44. 9 <0.0001 7.5 0.015 8.6 0.010 0.00 1 0.99 1.47 0.24 Cordia e 66. 7 <0.0001 7.2 0.017 6.0 0.03 4.18 0.066 0.002 0.97 Ormosia e 0.051 0.55 0.19 0.70 Mouriri h 28.6 0.033 18. 8 0.001 13.6 0.002 Guarea h 183.2 0.047 1.56 0.34 1.25 0.38 Trichilia h 164.5 <0.0001 6.4 0.023 7.8 0.013 7.05 0.017 0.55 0.47 Garcinia h 56.8 <0.0001 10.8 0.004 6.5 0.022 0.09 0.76 0.009 0.93 Species listed by genera with cotyledon morphology in subscript (e = epigeal, h = hypogeal).
91 Table 4 5 Species traits and survival according to damage tolerance, flood tolerance, and shade tolerance of damaged and undamaged seedlings. Species Habit Guild Elevation % Biomass lost ( SD) Flood tolerance Damage tolerance Shade tolerance Coccoloba Shrub Mid late Low 41 19 97% 92% 67% Vitex Tree Early Low 65 31 95% 85% 49% Trichilia Shrub Mid Low 44 16 89% 85% 53% Garcinia Tree Mid Low 19 7 88% 91% 67% Cordia Tree Early Low 55 31 86% 63% 33% Psuedobombax Tree Mid Low High 52 32 83% 63% 24% Mouriri Tree Mid Low 53 14 81% 47% 37% Guarea Tree Mid late High 52 23 15% 71% 0% Ormosia Tree Mid High 52 18 15% 54% 0% Hevea Tree Introduced High 62 13 0% 21% 0% Damage tolerance = average difference between monthly survival of undamaged and damaged seedlings; flood tolerance = proportion of undamaged seedlings surviving the flood season (February September); and shade tolerance = survival in low light (9 12% canopy openness).
92 Table 4 6. Summary of generalized linear model results (F tests & p values in parentheses) for the e ffects of damage, flood duration, light availability on three growth parameters of seedlings over one year: relative growth rate (RGR), root:shoot ratios, and shoot growth after the flood season, averaged across all species. Species Damage Flood duration Light Flood x damage Light x damage F p F p F p F p F p RGR 177.5 <0.0001 10. 9 0.0045 11.3 0.0039 0.07 0.80 1.48 0.24 Root:shoot 296.4 <0.0001 16.7 0.0009 0.74 0.40 9.6 0.006 0.67 0.43 Post flood growth 26. 5 <0.0001 8.1 0.012 4.08 0.062 7.2 0.016 5.92 0.028
93 Table 4 7 Summary of results from linear mixed models (Chi squared tests and p values in parentheses) on the effects of damage, flood duration, and light availability on r oot:shoot ratio s of seedling species listed by genera with cotyledon morphology in subscript (e = epigeal, h = hypogeal). Species Damage Flood duration Light Flood x damage Light x damage 2 p 2 p 2 p 2 p 2 p ALL 296 <0.0001 16.7 0.0009 0.74 0.40 9.62 0.006 0.67 0.43 Coccoloba e 0.20 0.66 71. 5 <0.0001 8.06 0.012 0.01 0.92 0.88 0.36 Pseudobom. e 9.6 0.015 1.76 0.21 0.011 0.92 5.64 0.045 1.07 0.33 Vitex e 6.17 0.026 13. 3 0.0024 9.11 0.008 4.59 0.058 0.59 0.45 Cordia e 3.33 0.095 1.31 0.27 0.58 0.46 3.03 0.11 1.42 0.26 Ormosia e 0.12 0.76 2.49 0.26 Mouriri h 13.6 0.066 7.62 0.014 1.87 0.19 Guarea h 0.43 0.63 0.47 0.57 4.55 0.17 Trichilia h 23. 4 0.0002 14.6 0.0015 0.84 0.37 3.11 0.097 2.02 0.17 Garcinia h 1.12 0.28 2.74 0.015 2.14 0.049 0.32 0.75 0.093 0.93
94 Figure 4 1. Rainfall and flood pulse in the study region. Mean monthly rainfall from 1982 2008 on the left y axis was provided by the Large Scale Biosphere Atmosphere (LBA) Station in Santarm PA. Change in river level from day 0 (01 October 2007) to day 365 (30 September 200 8) is indicated on the right y axis (Capitania dos Portos Santarm 2008)
95 A B Figure 4 2. Seedling survival across a gradient of flood levels (A) and light availability (B) for undamaged (solid dots) and damaged (white dots) seedlings, averaged across species. Lines are weighted regression model fit to the data for undamaged (solid line) and damaged (dashed line) seedling s
96 Figure 4 3 A verage leaf number of undamaged seedl ings for all species over time Deciduous species are indicated with dashed lines, evergreen species with solid lines. Note the early decline in leaf number for Vitex cymosa
97 A B C D E F G H I Figure 4 4 Seedling s urvival across flood duration and damage treatments for all species (A I) excluding Hevea Points indicate the proportion of individuals surviv ing after one year of undamaged (black dots ) and damaged (white dots ) seedlings in each plot Lines indicate the predi cted survival curves calculated by weighted regression models for undamaged and damaged seedlings.
98 Figure 4 5 Seedling survival (mean s & SE) of grouped speci es over the seven censuses T he time gap d uring flooding ( February August ) is indicated by a zig zag on the x axis. Survival is broken into three period s: low water season ( O ctober to January ), flood season ( F ebruary September ), and o ne year (September) LOW WATER FLOOD 1 yr
99 A B C D E F G H I Figure 4 6 Seedling survival across a gradient of light availability for all species (A J) excluding Hevea Points indicate the proportion of surviv ing individuals after one year of undamaged (black dots ) and damaged (white dots ) seedlings in plots. Lines indic ate the predicted survival curves calculated by weighted regression models for undamaged and damaged seedlings
100 Figure 4 7. R elative growth rates (RGR; means & SD) of damaged and undamaged seedlings. Species on the x axis are ordered by increasing average percent biomass removed after clipping stems at 5 cm aboveground (values shown above bars)
101 Figure 4 8 R oot:shoo t biomass ratios as a function of flood duration and for undamaged (black dots) and damaged (white dots) seedlings illustrating the interaction between damage and flood duration Linear mixed models tested the fixed effects of flooding, damage and light on root:shoot ratios ( Table 4 3)
102 Figure 4 9. Positive correlation between damage tolerance and flood tolerance for ten see dling species, indicated by genus.
103 Figure 4 10. Positive correlation between shade tolerance and flood tolerance of ten study species indicated by genus excluding Hevea
104 Figure 4 11 Trade offs between relative growth rate (RGR) and survival over one year for u ndamaged seedlings (solid line, black dots) and damaged (dash ed line, white dots) of the ten study species indicated by genus abbreviation
105 CHAPTER 5 TREE COMMUNITY DYNAMICS ACROSS FLOOD AND DISTURBANCE GRADIENTS IN AMAZONIAN FLOODPLAIN FORESTS Overview The rates and patterns of forest recovery are mediated by the interplay of disturbance intensity and environmental stress gradients. Amazonian floodpla in forests ( vrzea ) a critical ecosystem for aquatic diversity and productivity face multiple disturbances, but few long term data are available for determining rates of change during secondary succession. In floodplain forests recovering from agricul tural abandonment, I tested the synergistic effects of flood level and introduced livestock activity on change in seedling and tree density an d diversity over a nine year pe riod. Interactions between livestock and flood level affected seedling density, bu t the effects of livestock and flood level on seedling species density were independent. Seedling species density was best described by interactive effects between flood level and light, supporting a flood tolerance / shade tolerance trade off model. Hig h livestock activity in forest understories limited seedling density, and alleviation of this pressure resulted in a net influx of seedlings. Flood level was the major driver of temporal change in stand structure and species richness. I found no effect o f livestock activity on tree recruitment or mortality, nor did I find interactions between livestock and flood level. Overall, livestock disturbance, light availability, and flood level affect the seedling community synergistically. In contrast, forest s tand structure and rates of change in stems and species were largely guided by the main effects of flood level. The recuperation of seedling density following disturbance alleviation may contribute to the persistence of these forests despite intensive lan d use history.
106 Background Floodplain forests are dynamic eco systems exposed to multiple disturbances (Ward and Wiens 2001) forests (Tockner and Stanford 2002) with losses in area of up to 80% in temperate regions and rapid rates of loss in the tropics (Alho et al. 1988, Kingsford 2000) Amazonian white water floodplain forests ( vrzea ) are highly productive forests that comprise approximately 200,000 km 2 (Junk 1997) and con tain >1000 woody species (Wittmann et al. 2004) Tree diversity in the vrzea is remarkable given periods of inundation of up to 7 months and flood heights of up to 14 m (Goulding et al. 2003) Natural floodplain forests are comp o sed of stands at different stages of recovery from periodic erosion and deposition of sediments associated with flooding and river meandering (Salo et al. 1986, Terborgh and Andresen 1998, Godoy et al. 1999) In addition to natural disturbance, floodplain forests have long histories of human land use that also affect successional dynamics. There is a paucity of data about forest stand dynamics of Amazonian floodplain forests (Steege et al. 2003) particularly in the context of anthropogenic disturbance (Zarin et al. 1998, Anderson et al. 1999) The effec ts of multiple interacting factors on forest succession are a major gap in plant community ecology (Chazdon 2003) For est succession the rate and trajectory of change in stand structure and species composition is directed by an interplay of multiple factors including disturbance regimes (White and Pickett 1985) and e nvironmental stress gradients (Grime 1979) Numerous studies from tropical forests have demonstrated how disturbances such as logging, hurricanes, and livestock use affect forest stand structure and species composition (see reviews by Finegan 1996, Gauriguata and Ostertag 2001) Fewer studies have addressed the effects of
107 disturb ance on rates of change ( e.g., stem and species turnover) in successional forests (Chazdon et al. 2007) Environmental s tresses such as flooding and shade also affect successional dynamics, limiting species richness with increasing stress (Grime 1979) In undisturbed floodplain seedling communities, a shade tolerance / flood tolerance model may explain species richness and composition (Hall and Harcombe 1998) Disturbance and environmental stress act synergistica lly to affect forest structure ( e.g., basal area, stem density, and species richness) including seedling communities. For example, hurricanes damaged regeneration less severely in higher parts of the floodplain (Frangi and Lugo 1998) Across a gradient of flooding, trends in basal area vary (Conner et al. 2002) tending to peak at moderate flood levels in Amazonian floodplains (Ferreira 1997, Wittmann et al. 2004) Disturbance events in floodplains such as logging, windstorms, and livestock use could have different effects at different levels within the flood plain. Longer flooding periods in low er elevation forests could limit the time forests are exposed to trampling by livestock, but could also limit the recovery period in the dry season. Livestock impacts are pervasive in tropical floodplains world wide (Robertson and Ro wling 2000, Junk and de Cunha 2005) T he accessibility of water, shade, and forage make understory vegetation in floodplain forests particularly vulnerable to browsing and trampling. While the effects of livestock in forests are most severe for seedli ngs, th e long term effects of livestock activity in riparian forests include poor tree r ecruitment (Robertson and Rowling 2000) exotic invasions (Jansen and Robertson 2001) and conversion to pasture (Pettit et al. 1995) In addition, l ivestock activity over time can increase tree mortality via damage to superficial roots, soil compaction, and browsing
108 (Ramirez Marcial 2003, Mayer et al. 2005) In the Eastern Amazon, cattle and water buffalo move through seasonally flooded forests during the low water season, trampling seedlings and compacting soils (Sheikh 2002) Removal of livestock from floodplains occurs during the rising of annual floods, which then submerge or waterlog seedlings are for up to 6 months per year (Parolin 2002) The combined effect of intense livestock activity and extremely prolonged flooding in vrzea could decrease seedling density and diversity. Sustained limitation of see dling regeneration could lead to low rates of tree recruitment. Alternatively, prolonged flooding could loosen compacted soils compensating for the effects of trampling in the low water season, potentially alleviating the detrimental effects of livestock on seedlings. My goal was to test the main and interactive effects of flood level and livestock activity a chronic disturbance in the forest understory on seedling and adult stand structure and species richness, as well as rates of change over nine ye ars in Amazonian floodplain forests. Among the inventories available for floodplain forests in the Amazon, few provide long term data for monitoring rates of change (for revie w see Wittmann et al. 2006) Using a network of 0.1 ha plots across a flood gradient and varying in livestock activity, I address the following questions : 1) How do flooding and livestock activity affect seedling density and diversity? 2) What are th e effects of livestock activity and flooding on stand structure and species richness? 3) How do livestock affect seedling and tree species composition? Methods Study Site This study was conducted in floodplain forests of the Amazon River near Santarm, ). The Santarm region lies within the
109 Transverse Dry Corridor of the Amazon Basin with rainfall of 1800 2000 mm y 1 and 5 consecutive dry months with rainfall below 100 mm (July November; Sombroek 2001) Seasonal floods in the region peak between May and June, rising to an average of 7.5 m a.s.l. (1975 2008, Capitania dos Portos Santarm 2008) and extending up to 40 km from the main channel. Floodplains in the region are a mosaic of natural grasslands, forested levees, lakes, and giant aroid stands of Montric hardia arborescens Forests occupy ~15% of the area (M. Crossa, personal communication, June 1, 2007 ) and are restricted to higher elevation levees. Forest species composition varies with hydrological geomorphological factors such as water chemistry and flow and soil structure (Godoy et al. 1999) Land use history of floodplain f orests include conversion to cacao plantations ( Theobroma cacao ) and small rubber tree stands ( Hevea brasiliensis ) followed by deforestation for jute ( Corchorus capsularis ) plantations in the 1940 80s (Winklerprins 2006) Forests have regrown on abandoned jute plantations and currently face increasing impacts of livestock grazing in the flood plain (Sheikh et al. 2006) Experimental Design A network of 43 0.1 ha ( 20 x 50 m ) plots was established in 1999 at random locations within forested stands of three floodplain levees (Figure 5 1) Forest stands ranged from 15 50 years since abandonment of jute plantations and were located on private property of local landholders. Plots were classified as light (1), medium (2), or heavy impact (3) of livestock in 1999 and 2008 based on average hoof print density in five 9 m 2 seedl ing plots (0 10 = light, 11 20 = medium, 21 80 pugmarks = heavy) and herd activity as observed in behavioral studies ( Sheikh 2002). Hoof print density was deemed a surrogate variable for herd densities in forests as they remained in soils
110 throughout the low water season (Sheikh 2002) Average livestock activity over nine years was calcul ated from the 1999 and 2008 d ata For graphical purposes, average cattle activity in Figures wa s grouped into three levels (light = light & med light ; medium = medium ; heavy = heavy & med heavy ). I calculated flood level as the average water depth in for ests during the flood season from 1999 to 2008. Water depth was measured at 10 m intervals in forest plots with a weighted line during peak flooding in June 2006 (river level was 8.6 m a.s.l.). To calculate average flood level over 9 years, I subtracted the difference between levels in 2006 and the average for 1999 2008. Flood depth is thus a relative measure of the difference in maximum water column within plots. Forest plots were located on four different levees within the floodplain, two along the ma in Amazon channel and two along different branches of the Amazon River (Figure 5 1). To control for differences in stand structure and species composition due to unmeasured variables such as water chemistry and soil structure, levee was considered a rando m effect in statistical models. To address the effect of flood level light, and livestock activity on seedling communities, five 3 x 3 m subplots were established in random locations in each 0.1 ha plot ( Figure 5 1). Plots were monitored in 1999 for seed ling density only, and in 2008 for seedling de nsity and species composition Seedlings included all woody plants 10 130 cm tall and in 2008 were measured for stem height and identified to species or morphotype Change in seedling density was the differe nce between average densities per plot in 2008 and 1999. Each plot was assigned one of nine livestock change trajectories ( i.e. light to light, light to medium, etc.) for each combination of the three livestock impact levels in 1999 and 2008. Livestock change trajectory was used to test
111 the effect of disturbance alleviation e.g., heavy to light and medium to light impact on the change in seedling densities. I used livestock impact levels in 2008 to test the effects of livestock on seedlings in 2008, assuming that the most recent livestock activity would have the largest influence seedlings in the forest understory. I refer to species per plot to distinguish from estimated species richness calculated from rarefied data (Chazdon 2007). Canopy openness was measured at the center of each subplot in four cardinal directions with a spherical densiometer held at 1.3 m height (Lemmon 1956) Percent grass cover was estimated visually (Kennedy and Addison 1987) To test the effect of flood level and average livestock activity on adult stand structure and rates of change, all t DBH (diameter at 1.3 m height ) were censused during the low water season (November January) of 1999/2000, 2003/04 and 2008 /09 (N > 3 ,000 trees). I measured tree DBH, visually estimated height, and identified trees to spec ies or morpho species I distinguishe d new trees in 2004 and 2008 and deposited at the INPA (Instituto Nacional de Pesquisa Amaznica) herbarium in Manaus, Brazil (collector C.M. Nunes Nascimento, no.1 60) St em density and basal area were calculated for all three censuses. Importance values were calculated for each species as the sum of relative frequency, relative density, and relative dominance of individuals (Brown and Curtis 1952) Rates of change recruitment, mortality, species loss and gain, change in stem density, change in basal area, and change in species richness were calculated for 1999 2003 and 2003 2008. I tested the effects
112 of average livestock impact levels in 1999 2008 on adult stand structure and species composition. Statistical Analyses How do flood level and livestock activity interact to affec t seedling density and species density? Analyses of seedling density were conducted at the sub plot scale, using pugmark density as a predictor variable, and at the plot scale using livestock activity as the predictor variable and average density as the response variable. The effects of hoof print density, relative flood level (water column 0 1.9 m), and percent canopy openness in 2008 on seedling density in subplots were assessed with a linear mixed model (LMM). Seedling density and percent canopy openness were log transformed to achieve normality of model residuals, according to a Shapiro Wilk test. T h e effects of livestock activity (2008), flood level and average percent canopy openness on seedling species density were assessed with LM Ms at the plot scale Due to low species counts (mean = 3.7 2.3) and violation of mode l assumptions the effect of flooding and livestock activity was tested on the sum of all species observed in the five 9 m 2 subplots. Homoscedasticity was tested for all response variables across livestock impact levels with the Bartlett test (Crawley 2007) All models predicting seedling parameters were fit by restricted maximum likelihood (REML), which accounts for unbalanced designs (Bolker et al. 2009) Levee and p lot (nested within levees) were density and species density across livestock activity, as well as the difference in seedling density across livestock change trajectories.
113 What are the effects of livestock activity and flood level on stand structure and species richness? To test the effects of a verage livestock activity, flood level, and their interactions on adult stand structure, I used repeated measures LMM s fit by restricted maximum likelihood and a n autoregressive correlation structure (Crawley 2007) Models tested the effects of livestock impact, flood level, and their interaction s on stand structure: stem density (stems ha 1 ), basal area (m 2 ha 1 ), and species density ( species count per plot). Levee, plot, and year were treated as random effects. T he effects of flood level and livestock activity on species density were assessed with a linear mixed model where (flood level 2 ) was included as a quadratic term (Crawley 2007) T o test the effects of flooding on estimated species richness, plots were grouped into three flood level categories ( 0 0.60; 0.60 1.30; 1.30 2.0 m ). Species richness values for adult trees and seedlings were calculated using the Chao 1 abundance base d species richness estimator (Chao 1984) in EstimateS Win 8.20 (Colwell 2009) To compare the effects of average livestock activity and flood level category on species richness of adult stems and seedlings, s ample based species rar efaction curves across livestock activity and flood level category were computed by EstimateS using the Coleman method (Coleman 1981) What are the effects of livestock activity and flood level on rates of change? To test the effects of livestock activity, flood level, and their inter action on rates of change in forest stands, I used LMMs fit by restricted maximum likelihood with levee as a random effect. Stem and species turnover rates, change in basal area, and change in stem density were calculated as four parameters for rates of c hange between 1999 and 2008. Recruitment and mortality rates ( i.e. stem turnover rates) were calculated as the
114 relative proportion of trees gained or lost from the total number of live trees in 1999. Species turnover rates were the percents of species lost or gained over nine years (Chazdon 2007). Rates of recruitment, mortality, and percent species loss and gain were logit transformed to achieve normality of model residuals according to Shapiro Wilk test s Differences in rates of change among livestock levels were tested post hoc All analyses were conducted in R 2.9.0 (Team 2009) using the nlme package (Pinheiro et al. 2008) How does livestock activity affect species composition? To compare species composition of seedlings and trees a mong flood levels and livestock impact levels, the Chao abundance based Srenson index of similarity (Chao et al. 2005) was calculated in EstimateS 8.0 (Magurran 2004) for each possible pairwise combination of the 43 plots. In two separate tests for flood level (grouped into 3 categories) and livestock impact level, I used two sample t tests with unequal variances to compare Chao similar ity indices between plots in the same livestock activity category vs. those in different categories Results Effects o f Flood Level a nd Livestock Activity o n Seedli ngs As expected, seedling density decreased with increasing flood level and livestock hoof print density. Furthermore, flooding and livestock activity interacted such that seedling density was low across all flood levels in plots with heavy livestock activity, while seedling density was negatively related to flooding with low livestock activity (Table 5 1, Figure 5 2). Mean seedling density in low impact plots was more than three fold of that in heavy impact plots (2.9 1.9 SD for low impact, 1.3 1.0 SD for medium impact, and 0.7 0.6 SD seedlings m 2 for heavy imp act). Plots changing from
115 heavy or m edium to low impact showed net in creases of seedlings (0.81 1.2 SD and 1.4 1.4 SD seedlings per m 2 respectively), indicating recruitment of seedlings into the understory when livestock activity is reduced ( Figure 5 3). Canopy openness was positively correlated with seedling density. Seedlings were limited to 1.2 1.4 SD plants m 2 in low light environments (2 7% canopy openness) across the flood gradient (Table 5 1). Species density was affected by the interactio n of light and flood level that concur with the flood tolerance / shade tolerance trade off model. At high light levels, species density was constant along the flood gradient at 10 17 species per 0.1 ha plot. At moderate to low light levels, species dens ity declined across the flood gradient (Figure 5 4A). Species density also declined with increasing flood level (F = 3.88, p = 0.056), as did estimated species richness (Figure 5 5) No effect of livestock activity was found on species density (Table 5 2), but rarefaction curves across livestock levels seedlings suggested higher species richness in low impact plots (Figure 5 8). Light availability increased species density and ameliorated negative effects of high flood levels and heavy livestock disturbanc e (Table 5 1; Figure 5 4) Effect s of Flood Level and Livestock A ctivity on Trees Flood level a ffect ed tree species density, but livestock activity did not interact with flooding to alter stand structure. H eavy impact plots had relatively low basal area (2 8.1 8.0 m 2 ha 1 ) and were species poor, reaching ap proximately half of the species richness as found in m edium and low impact plots ( Figure 5 5 ). Contrary to trends in other floodplain forests, basal area increased with flooding (Figure 5 5). No differ ences in stem density between livestock impact levels or flood levels were found ( Table 5 2 ). The relationship between s pecies density and flooding was explained by a quadratic
116 relationship Overall, s pecies density peaked at 15 20 species per 0.1 ha at m oderate flood levels between 0.6 and 1.2 m (100 120 d flooding; Figure 5 5). Effect s o f Flood Level and Livestock Activity on Stand Dynamics Stem and species recruitment into tree size classes were limited by prolonged flooding ( Figure 5 6, Table 5 3 ). In contrast, livestock activity did not affect stem mortality or limi t recruitment of new species ( Table 5 3 ). Rather, livestock activity was associated with low stem mortality, which was lower in heavy impact plots than light and medium impact plots (Tukey HSD p = 0.001 and p = 0.0002, respectively). No differences in recruitment among livestock impact levels was found ( F = 0.63, p = 0.54) and there was no interacti on between flooding and livestock activity, indicating that rates of change in ad ult stands are largely driven by flood level. Average basal area increment over nine years was 0.96 6.9 m 2 ha 1 y 1 Livestock activity did not decrease gains in basal area over time in forest stands ( Table 5 2 ). As expected, young forests showed the highest net increases in basal area, stem density, and species density. From 1999 2008, young forest stands (15 25 y) had a net increase in stem density (12.3 26.8) and basal area (4.6 1.9 SD), averaging 0.51 m 2 ha 1 y 1 In contrast, older stands showed net zero gains in basal area (0.23 6.6 m 2 ha 1 y 1 ) and large reduction in tree density ( 16.9 12.4). In terms of species turnover, young and old forest stands had high increases in species density over time (16.15 11.2% and 14.5 21.7 %, respectively ), but species gain was low in moderate age stands of 30 35 (4.9 8.5 %). Interestingly, species loss was fairly constant across all ages at 10 13.5% ( 12.8 13.7 SD) of total species density in plots. Forest age had no effect on mortality or recruitmen t rates (F = 3.10, p = 0.058 and F = 0.33 p = 0.72, respectively)
117 Effect s o f Flood Level and Livestock Activity on Species Composition A total of 32 families, 63 genera, and 88 woody species occurred in forest stands ( Appendix A ). Species richness of tr ees was lowest in heavy impact plots (Figure 5 5). In accordance with the hypothesis that species richness should increase with decreasing flood stress, estimated species richness of trees was highest at low flood levels (Chao 1 = 65 9.2) in comparison to stands at moderate and high flood levels (Chao 1 = 60.0 3.5 and 38.3 5.4, respectively). T ree and seedling communities were species rich (Chao 1 = 75.7 4.5 & 74.6 4.8 SD respectively) in comparison to adult r ecruits ( Chao 1 = 54.6 3.4 SD ). Plots with the same flood level were more similar in species composition than plots with different flood levels among trees, recruits, and seedlings (t 549 = 7.75 p <0.0001; t 443 = 4.48 p <0.0001; t 517 = 5.40, p<0.0001, respectively). High elevation plots in the floodplain (maximum flood level = 0.02 0.60 m = 2 40 d flooding ) were dominated by L aetia corymbulosa Andira inermis and T riplaris surinamensis trees and T alisia cerasina and T abernaemontana siphilitica in the understory. Mod erately flooded plots ( 0.60 1.3 m = 40 70 d flooding ) were dominated by C ordia tetrandra Pseudobombax munguba and Ficus spp. trees with T siphilitica and Garcinia brasiliensis in the understory. Low elevation plots ( 1.3 1.9 m = 70 100 d underwater) wer e dominated by Cynometra bauhiniifolia C tetrandra and P munguba with T cerasina and Ocotea castanaefolia in the understory Tree species composition changed minimally over the nine year period; however, two low flood tolerant tree species valued fo r timber, Guarea guidonia and Hura crepitans, were lost between 1999 to 2008. P lots with the same livestock activity were more similar in species composition of tree and seedlings than plots of different activity (t 530 = 4.7, p < 0.0001 ; t 530 = 4.8, p <
118 0.0001 respectively ). There was no difference in species similarity for livestock levels among recruits (t 590 = 1.7, p = 0.09). Light and m edium activity plots were dominated by trees of light demanding species while heavy activity plots we re dominate d by Cynometra bauhiniifolia a slow growing mid successional species with low species turnover ( Figure 5 10 ). The understory communities of both light and m edium activity plots were dominated by the common shrub, Tabernaemontana siphilitica and small tre es Talisia cerasina and Trichilia singularis In heavy impact plots, recruits were dominated by C bauhiniifolia (42 % 7% SE ) and ranked 6 th in abundance among seedli ngs (4.7 % 2.2% SE). The understory wa s largely comprised of shrubs, including thorn bearing species, Xylosoma benthamii and Randia armata ( Figure 5 10 ). Discussion Much of the research on Amazonian floodplain forest succession has focused on changes in stand structure and diversity in relation to the flood gradient (Salo et al. 1986, Wittmann et al. 2004) However, floodplain forests are increasingly dominated by secondary forests degraded by logging (Anderson et al. 1999, Zarin et al. 2001) agriculture (Zarin et al. 1998) and livestock ranching (Sheikh et al. 2006) By incorporating land use history as an explanatory variable that acts synergistically with flooding to affect forest succession, I gain a broader understanding of floodp lain forest trajectories of change through time. In Amazonian floodplain forests recovering after abandonment of agriculture, the effects of increasing livestock activity reduces seedling density by one third. In stands receiving heavy livestock activity species richness and basal area were low, but I did not observe an increase in stem mortality and decrease in recruitment with increasing livestock activity. Seedling species count was affected by the interaction of flood level and canopy openness, supp orting the hypothesis that many
119 species require increased light availability for survival at high flood levels (Battaglia and Sharitz 2006) Adult species density peaked in moderately flooded plots, and the complete loss of adult stems of two valuable low flood tolerant timber species was observed ( Guarea guidonia and Hura crepitans ). Recruitment of stems and species into the tree size class was limited by increasing flood levels, demonstrating that the flood regime is important not only for patterns in species richness and composition, but also for rates of change in forest stands through time. Seedling Response to Livestock Activit y, Flood Level and Light Flooding and herbivore activity interact in floodplain ecosystems to affect seedling density (Oesterheld and McNaughton 1991, Butler et al. 2007) C hronic disturbance by livestock in floodplain forests act ed synergistically with floodi ng to severely limit understory plant density in sites with prolonged flooding (>160 d) and intense livestock activity. The negative effects of livestock activity on species counts were apparent when flooding stress was low (50 160 d inundation). Such pa tterns suggest that either the stress of flooding is so severe that it masks the effects of livestock, or that prolonged floods alleviate heavy impacts of livestock. A consequence of the intensification of livestock in forests is the creation of open for ests with sparse vegetation in the underst ory In this study, heavy livestock activity decrease d seedling density to less than one third of that in light activity forests. Reduced seedling density is likely caused by the co n tinual mortality of seedlings by trampling, soil compaction, and browsing by livestock (Kauffman and Krueger 1984) Compacted soils reduce the survival of some floodplain forest species (Sheikh 2002) which can ultimately affect trajectories of succession of floodplain forests. As canopy trees die, the lack of seedlings and young trees in the understory could f avor invasion of
120 grasses in open gaps. However, the expected increase in grass cover in heavy impact plots was not observed indicating that while understories are sparsely populated, there is little replacement by grasses. Canopy openness < 35% in this study may provide insufficient irradiance for grass survival and competition with pioneer woody species (Scholes and Archer 1997, but see Veldman et al. 2009) Species density of seedlings in forest understories was strongly influenced by an interactive effect of light availability and flood level. Contrary to our hypothesis that increased light would favor grass invasion, increasing light availability in the understory increases woody seedling density and species density. At high light levels (10 35% average canopy openness), understories maintain a higher number of species despite prolonged flooding. In turn, a t low light levels (< 5% canopy openness) species density diminishes with increased flooding. Such trends concur with the flood tolerance / shade tolerance trade off hypothesis for seedling diversity in floodplain forests, whereby species with high flood tolerance require higher light availability for growth and survival (Battaglia and Sharitz 2006) Mechanisms for recovery The regenerative capacity of floodplain forests may contribute to their rapid recovery following multiple cycles of deforestation (Smith 1999) In this study, alleviat ion of heavy livestock activity resulted in recovery of seedling densities to those in stands with low activity. The re establishment of understory vegetation could be a result of m ultiple fact ors related to annual flooding that facilitate recovery in floodplains. Firstly, the persistence of resprouts following disturbance could promote rapid recovery of woody vegetation. Secondly, seed dispersal by floodwaters (Moegenberg 2002) fish (Goulding 1980, Kubitzki and Ziburski 1994, Lucas 2008) and
121 birds durin g the flood season overcome spatial/distance based barriers to seed arrival (Holl 1999) Finally, t he rich sediments deposited by floodwaters may replenish nutrients as well as loosen compacted soils, facilitating seed germin ation and seedling growth. T hese factors may alleviate the effects of livestock disturbance on the seedling community and promote recovery of diver se understory vegetation Secondary Succession of Forest s The flood regime in Amazonia is among the most important factors for understanding stand structure as well as rates of change in forests. Flood level predict ed rates of recruitment of new trees and species. Patterns in basal area and species density were also observed across the flood gradient although, contrary to other studies, basal area increase d with flooding and species density peak ed at moderate flood levels. In other studies of neotropica l floodplain forests, basal area peak ed at moderate (Ferreira 2000) or low flood levels (Nebel et al. 2 001b) At high elevations, the increased frequency of late successional species with high density wood can cause basal area to remain cons tant or decline (Wittmann et al. 2004) Species richness should have decreased with increasing flood level, as increasing exposure to an oxic conditions should limit the community to fewer highly flood tolerant species. Stand structure and species composition Basal area in Eastern Amazon floodplain forests reached as high as 81 m 2 ha 1 peaking in moderately flooded forests comprised of Pse udobombax munguba a fast growing late pioneer species in forests 30 35 years since abandonment. Average basal area of forests was 34.2 16.1 SD m 2 ha 1 higher than many other floodplain forests of the Amazon region. For example, in mature species rich floodplain forests of the Western Amazon, basal area ranged from 16.3 to 28.8 m 2 ha 1 peaking in total basal
122 area and net change in basal area at in deeply flooded stands (Ne bel et al. 2001a) Late successional Amazon e stuarine floodplain forests averaged 29.5 m 2 ha 1 (Cattanio et al. 2002) Patterns in basal area in this study were most similar to those of late secondary Ce ntral Amazonia, where basal area of secondary forests reach 48 60 m 2 ha 1 in 50 year old stands dominated by P. munguba at moderate flood levels (Worbes et al. 1992) then decline to 31 38 m 2 ha 1 in late successional and mature stands (Wittmann et al. 2004) Tidally flooded forest stands domi nated by Prioria copaifera in Darien, Panama also have high basal area ranging 31.1 71.1 m 2 ha 1 for stems >10 cm DBH (Grauel 2004) Basal area thus appears to vary within floodplains by a combination of factors including species composition, flood level, and forest age. Overall, floodplain forests have higher basal are a than upland forests, likely due to faster growth on alluvial substrates (Chinea 2002) The floodplain forests of the Santarm region in the Dry Corridor of the Amazon Basin had exceptionally high basal area in comparison to other tropical forests despite their history of deforestation and degradation. The high basal area of stands in this study could be related to the long growi ng season of the region (Sombroek 2001) Although basal area of forests was high, accumulation rates of basal area were relatively low. Change i n basal area varied widely, averaging 0.51 m 2 ha 1 yr 1 in forests of 15 25 years old and 0.025 m 2 ha 1 yr 1 in forests of 40 50 years old. Such accumulation rates are extremely low in comparison to floodplain forests of other regions (Nebe l et al. 2001b) and upland forests (Chazdon et al. 2007) High rates of stem loss (10 30%) contributed to low net gains in basal area. Stem mortality is a
123 function of forest age, as well as unmeasured factors such as exposure to windstorms, bank erosion, and removal of firewood by residents. As expected, es timated species richness of adult stems was limited by flooding. Some studies suggest that species richness declines across the flood gradient, as fewer species are able to tolerate increasing exposure to anoxic conditions (Terborgh and Andresen 1998, Parolin et al. 2004b) However, other studies of Amazonian forests have found that low elevation forests are also species rich (Ferreira 2000, Nebel et al. 2001b) S pecies richness in vrzea forests declines from w est to e ast, with a total of 480 species in the Western Amazon (170 species ha 1 ; Wittmann et al. 2002) 371 species in Central Amazonia, and 133 species in the Lower Amazon Es tuary. This study fills an important gap in species richness of the Dry Corridor in the Lower Amazon region. Although reviews show a gradual decline in species richness on a West East gradient, this study shows that species richness in the Santarm region is exceptionally low (74 76 woody species with stems 10 cm DBH). The drop in observed species richness could be related to m any factors including: 1) intense land use history in the region and multiple cycles of deforestation for cacao plantations, rubber tree planting, and jute cultivation 2) the relatively young age of forests and 3) the relatively dry climate and geological h istory of the region, which accounts for lower species richness in adjacent upland forests (Terborgh and Andresen 1998) Forest age did not correlate with stand structure or rates of change in forests. I expected to find increasing basal area and decreasing stem density with increasing age (Chazdon et al. 2007, Flynn et al. 2010) but no such trends were observed. Within the nine year study period, young forests (15 25 y) had a net increase in basal area and
124 stem density much higher than that of older forests (30 5 0 y). Older forests had no net gain in basal area over nine years, potentially due to the high mortality of pioneer species ( e.g., Cordia tetrandra and Triplaris surinamensis ) in the canopy and the recruitment of mid late successional species. Forest age may not have affected forest structure due to the narrow range in stand age and the interaction of age with flood level to affect basal area and stem density. In deeply flooded forests, mortality > recruitment, but in plots with low flooding, recruitment > mortality, reaching 30 0 stems ha 1 y 1 in young forests. Overall, trends in stand structure across forest age depend ed on flood level. Floodplain forest stand dynamics Stem and species recruitment gradually decreased with increased flood level. Although many studies have shown that increasing flood level decreases recruitment, few studies measure tree recruitment across a fine gradient of flood levels Although tree recruitment rates were similar across livestock activity densities of recruited trees were lower in heavy impact stan ds. T he species recruited into heavy impact plots were thorny shrubs ( Randia armata and Xylosoma benthamii ) that may prevent damage from livestock movement in the understory. Poor recruitment of shade tolerant tree species suggest ed that successional tra jectories of stands with sustained heavy livestock activity could remain species poor over time One group of heavy activity plots had exceptionally high mortality ( ~ 400 stems ha 1 ) likely attributed to the exposure of stands to winds and strong currents along the main channel of the Amazon River. An important finding is the exponential decline in rates of stem and species recruitment across the flood gradient. The gradual decline in species recruitment with increasing flood level suggests that with re latively small differences in flood level and
125 duration, exponentially fewer species are able to survive. Similar results were found for the sapling community in vrzea forests of the Central Amazon (Wittmann and Junk 2003) The average recruitment of new species in a plot over nine years ranged from 8 16% in stands, while species lost ranged from 7 13%. Such rates of species recruitment are greater than those of 25 year old upland forests post livestock abandonment (2 5%), but still lower than for lightly disturbed upland forests with no history of livestock impacts (35 55% over 7 years Chazdon 2007). Successional trajectories of flo odplain forests C hanges in species composition show ed transition of forests from early to late successional stages. E arly successional light demanding species ( Cordia, Cecropia, and Pseudobombax ) were replaced by late secondary species and understory shrubs. of the c anopy species Triplaris surinamensis and understory shrub Tabernaemontana siphilitica were indicators of past degradation or deforestation (F. Wittmann, pers onal communication, April 20, 2010 ). Th e dominant species in plots sustaining high livestock disturbance, Cynometra bauhiniifolia is a slow growing mid successional species with low mortality rates in comparison to early successional species ( e.g., Pseudobombax munguba, Triplaris surinamensis, and Cordia tetrandra ). The establishment of Cynometra dominated stands where heavy livestock impact also occurred may be related to previous land use history and forest management practices prior to livestock introduction No data was collected on fuelw ood harvesting in forest stands nor land use history of burning, which could also affect species trajectories (Mesquita et al. 2001)
126 Implications f or Conservation in the Context o f Climate Change E xtreme flooding events associated with climate change likely have important implications for species composition. In forest stands I observed the loss o f mid successional high vrzea species Guarea guidonia and Hura crepitans that occur in high elevation levees that flood supra annually or annually for short periods (< 50 d, Wittmann et al. 2004) These species are highly valu ed for timber in Amazonia (Marinho et al. in press) The loss of such high elevation species is likely due to the rece nt increase in maximum flood levels from an average of 5.7 m from 1925 1975 to an average peak of 7.5 m from 1975 2008 with exceptional floods of 8.6 m in 2006 and 8.36 m in 2008 (Capitania dos Port os Santarm 2008) Extreme flooding events such as those of 2006, 2008, and 2009 cause mortality of low flood tolerant species that occupy high elevations and contribute substantially to the overall diversity of floodplain forests (Salo et al. 1986, Worbes et al. 1992) As such, the species available for future replacement of canopy trees in floodplain forests are limited not only by intense livestock disturbance but also extreme floodi ng events. Future studies will need to investigate the effects of extreme flooding events on floodplain forest species composition and diversity.
127 Table 5 1 Summary of linear mixed model results (F tests and p values in parentheses ) for the effects of flooding, livestock, activity, and light availability on the seedling community. Flood Livestock Avg. Light Flood livestock Light flood Light livestock 3 way Seedling density 18.76*** (0.0001) 11.4*** (0.0011) 16.7*** (0.0001) 4.31* (0.039) 9.25** (0.0027) 0.004 (0.95) 2.01 (0.16) Average seedling density 9.95** (0.0034) 7.67** (0.0018) 2.92 (0.097) 0.94 (0.40) 6.46 (0.016) 0.42 (0.66) 0.92 (0.41) Average proportion damaged 10.51** (0.0027) 0.37 (0.69) 0.19 (0.67) 0.17 (0.84) 0.0004 (0.98) 2.59 (0.09) 0.21 (0.81) Total species density 3.88 (0.056) 1.08 (0.30) 8.42** (0.006) 0.0034 (0.95) 5.51* (0.024) 1.84 (0.18) 0.0007 (0.98) At the subplot scale the effects of livestock hoof print density, floo d level, and canopy openness (light) on seedling density in 3 x 3 m subplots. At the plot scale the effects of livestock activity, flood level, and ave rage canopy openness on average seedling density, average proportion of seedlings with damage, and species density (number of species in subplots) in 0.1 ha plots. 0.10
128 Table 5 2 Summary of linear mixed model results (F tests an d p values) for the effects of flood level, livestock activity (1999 2008), and their interaction on tree stem density, basal area, and species density Rate of change Flood level Livestock activity Flood livestock Flood 2 Flood 2 x livestock Est. F p F p F p Est. F p F p Stem density 740 0.38 0.54 0.53 0.71 0.23 0.92 Basal area 241 7.74 0.010 2.28 0.088 0.91 0.47 615 5.87 0.023 2.17 0.10 Species density 267 16.8 0.0004 2.22 0.094 0.70 0.60 185 0.54 0.47 2.28 0.088 In models for basal area and species density, flooding was included as a quadratic term. Coefficients (Est.) are included for the effect of flood level. 0.10
129 Table 5 3. Summary of linear mixed model results for the effects of flood level, livestock activity (1999 2008), and their interaction on tree stem mortality and recruitment, species loss and gain, change in stem density, and change in basal area over 9 years Coefficients (Est.) are shown for flood level effects Rate of change Flood level Livestock intensity Flood l ivestock Est. F p F p F p Mortality (stem loss) 0.29 0.004 0.95 5.84 0.001** 1.16 0.35 Recruitment (stem gain) 1.10 10.9 0.002** 0.81 0.52 0.62 0.65 Species gain 0.99 5.32 0.028* 1.88 0.14 0.87 0.49 Species loss 0.18 0.17 0.68 0.78 0.55 0.95 0.45 Change in stem density 39.9 5.67 0.024* 1.03 0.40 1.67 0.18 Change in basal area 20.7 0.002 0.96 0.79 0.54 0.53 0.71 0.10 0.05;
130 Table 5 4 Tree stand structure and species richness (mean & SD) in the three forest inventories, 1999, 2003, and 2008. Year Mean DBH (cm) Stem density (ha 1 ) Basal area (m 2 ha 1 ) Observed species (Mao tau) Species richness (Chao 1) 1999 23.2 13.9 693 206 33.2 16.3 71 2.45 75.7 4.49 2003 23.7 15.0 680 216 38.2 15.6 70 2.03 72.5 2.9 2008 25.6 15.7 571 201.5 37.2 16.1 69 2.37 74.6 5.35
131 Table 5 5 Estimated species richness (Chao 1 means & SD) of trees and seedlings across flood level categories (low = 0 0.60; medium = 0.60 1.3; high = 1.30 2.0 m) and average livestock activity in 1999 2008 for trees and in 2008 for seedlings Flood low Flood medium Flood high Trees 65 9.2 60.5 3.5 38.3 5.4 Seedlings 58.2 6.4 48.5 4.0 32.5 3.2 Livestock light Livestock medium Livestock heavy Trees 64.4 4.7 55.6 4.8 58.5 12.9 Seedlings 65 9.9 64.4 11.2 37.6 3.9
132 Figure 5 1. Map of secondary floodplain forests stands in Santarm Par, Brazil, at the confluence of the Amazon River and the Tapajs River. A) Boxes indicate three locations within which 0.1 ha. plots were established (A: 9 plots; B: 7 plots; C: 27 plots). B) Experimenta l design of 0.1 ha plots with 5 3 x 3 m seedling subplots (indicated in grey). A B
133 A B Figure 5 2. Average seedling density across a flood gradient. Seedling density as a function of A) flood level and livestock activity levels in 2008, and B) flood level and light availability (low = 0 5%, medium = 5 10%, and high = 10 35% canopy openness). Points indicate plot level averages of seedling density.
134 Figure 5 3. Change in seedling dens ity across livestock change trajectories from 1999 to 2008 (means & SD). Bars indicate a given change trajectory from livestock levels in 1999 (x axis) to livestock levels in 2008 (white, grey, and black bars).
135 A B Figure 5 4. Number of seedling species in 45 m 2 across a flood gradient. Species number as a function of A) flood level and light availability (low = 0 5%, medium = 5 10%, and high = 10 35% canopy openness); and B) flood level and livestock activity (light, medium, and heavy) in 2008.
136 A B C D E F Figure 5 5. Trends in forest stand structure across flood level and livestock activity. A) B asal area, B) stem density ha 1 and C) species density 0.1 ha 1 across flood level and livestock activity in 1999 2008. Trends in D) change in basal area, E) change in stem density, and F) change in species density from 1999 to 2008 across flood level and livestock activity. Points indicate values for each of 43 plots for each census in 1999, 2003, and 2008.
137 Figure 5 6. Stem and species turnover rates from 1999 2008 comparing percent of trees lost (mortality) and gained (recruitment) and the number of species lost and gained across flood levels (0 0.6 m = high; 0.61 1.3 m = medium; 1.31 1.9 m = high) in 0.1 ha plots ( means & SD). Differences based on Tukey HSD tests are indicated by lower case letters.
138 A B Figure 5 7. Trends in stem and species turnover across the flood gradient. A) Species lost and gained as a function of flood level. B) Stems lost (mortality) and gained (tree recruitment) as a function of flood level.
139 A B C D Figure 5 8 Species accumulation curves for seedlings and trees A) Species accumulation curves of trees ( 10 cm DBH) across three livestock levels in 1999 2008. B) Speci es accumulation curves of trees across three flood level categories. C) Species accumulation curves for seedlings across three livestock levels in 2008. D) Species accumulation curves for s eedlings across flood level categories. All values based on Coleman rarefaction values with 95% confidence intervals calculated in EstimateS.
140 A B C Figure 5 9 Comparisons of relative abundance of the of the 6 most common species of trees, recruits, and seedlings at different flood levels : A) low flood level B) moderate flood level and C) high flood level
141 A B C Figure 5 10. Comparisons of relative abundance of the 6 most common species of trees, recruits, and seedlings at different levels of livestock activity: A) light B) medium and C) heavy livestock activity
142 CHAPTER 6 ABOVEGROUND BIOMASS AND CARBON SEQUESTRATION IN SECONDARY FLOODPLAIN FORESTS OF EASTERN AMAZONIA Overview Regional carbon budget estimates fo r the Amazon Basin often exclude the contribution of floodplain forests, which comprise ~4% of land area. Biomass accumulation rates in floodplain forests can double those in upland forests, but few estimates are available to confirm such trends across th e Basin. This study fills a gap in ground based data for aboveground woody biomass (AGB) and biomass accumulation rates in Amazonian floodplain forests in the Dry Corridor. We also test how productivity varies within a floodplain by comparing biomass dyna mics across gradients of flood stress, forest age and livestock activity. We estimated average aboveground woody biomass of 15 80 year old forests as ~210 Mg ha 1 Flood level and forest age interacted to affect AGB, which peaked > 250 Mg ha 1 in mid suc cessional forests with moderate flooding (~50 100 days flooding). In a subset of forests monitored over nine years, average aboveground biomass accumulation was ~5 Mg ha 1 y 1 Biomass accumulation rates showed no correlation with flood level, refuting t he hypothesis that productivity should increase with decreasing flood duration. Variation in species composition and livestock activity appeared to explain the slow tree growth and low recruitment that resulted in net carbon losses in some stands. Annual mortality rates averaged 70% of total biomass gains, suggesting relatively high stem mortality. Despite a long history of land use, carbon sequestration by secondary floodplain forests is an important component of local carbon budgets. These data can im prove regional models for C budgets as well as explain local scale variation in biomass across multiple interacting factors.
143 Background Tropical forests are an important sink for global carbon (Phillips et al. 1998, Baker et al. 2004) Tropical rivers and wetlands, however, can be impor tant sources of carbon via net annual losses from out gassing of CO2 (Rich ey et al. 2002) Floodplain forests are highly productive ecosystems at the transition zone between upland forests and rivers (Junk et al. 1989) Despite comprising only approximately 4% of land area of the Amazon Basin (Saatchi et al. 2007) floodplain forests can sequester carbon at 2 3 times the rate of upland tropical forests (Nebel et al. 2001a, Schngart et al. 2010) Such high rates of product ion could help offset net annual carbon losses, but few data are available to incorporate seasonally flooded forests into carbon budget models (Table 6 1; Baker et al. 2004, Saatchi et al. 2007) Further ground based data for biomass and biomass accumulation rates are needed to understand the role of floodplain forests in source sink relationships of carbon in tropical watersheds. Tropical forest biomass and carbon sequestra tion are affected by regional scale factors such as dry season duration and extreme climate events such as El Nio/La Nia Southern Oscillation (Malhi et al. 2004) Regional models for forest biomass suggest that dry regions have lower basal area, which can lead to lower estimations for biomass (Saatchi et al. 2007) Upland tropical forest stands can experience net biomass losses due to increased stem mortality associated with drought (Condit et al. 1995, Nepstad et al. 2007) Trees in the floodplain, however, are unlikely to experience water stress in drought years due to the proximity of the water table, even during exceptionally dry years (Parolin et al. 2009) Seasonally flooded forests should show opposite trends, with increases in biomass accumulation during drought (Schngart et al. 2010) The decrease i n flood levels in drought years can increase the terrestrial
144 phase for tree growth floodplains, and thus increase diameter increment (Schongart et al. 2004) The Dry Corridor of low rainfall and prolonged dry seasons in the Amazon Basin has been useful for predicting the effects of drought on upland forest biomass (Phillips et al. 2009) Data for floodplain forest biomass and carbon sequestration rates in the Dry Corridor are lacking, making it difficult to understand how floodplain forests differ ac ross regions. Stand biomass and biomass accumulation rates vary with many local abiotic factors, including environmental stresses such as flooding. The response of forest productivity to flood stress can follow many trends, including that predicted by th e subsidy stress model (Odum et al. 1979) The model suggests that productivity is higher in forests with short term seasonal floods than in forests with s tagnant water and long term flooding. In stands with short term seasonal flooding, flooding is not considered a stress to the ecosystem but a subsidy that enhances productivity in comparison to upland forests. This model was developed from observations i n cypress swamp forests of the southeastern US (Conner and Day 1976) An alternati ve hypothesis, which has received more support from empirical data (Megonigal et al. 1997) suggests that the physiological stress of anoxia supersedes the subsidies to production during flooding (Mitsch a nd Rust 1984) This model also predicts decreasing productivity with increased flooding, but with productivity rates equal to or lower than that of upland forests. A large area of floodplain forest in the Eastern Amazon is secondary, recovering from de forestation from timber extraction and agriculture (Anderson et al. 1999, Zarin et al. 2001, Wittmann et al. 2006) The rapid growth rates of early mid successional tree
145 species in secondary floodplain forests could result in high rates of biomass accumulation (Schngart et al. 2010) However, secondary floodplain forests also face continual degradation by logging, fuelwood extraction, and trampling by livestock (Sheikh et al. 2006) These local land use activities could reduce biomass accumulation rates via high stem mortality and poor recruitment (Chazdon et al. 2007) The goal of this study was to provide long term data on aboveground woody biomass (AGB) and aboveground biomass accumulation rates in secondary floodplain fore sts in the Dry Corridor of the Amazon Basin. I measure d AGB and biomass accumulation in a network of secondary floodplain forest plots in the Lower Amazon. I address ed the following questions: 1) What are AGB and biomass accumulation rates for secondary floodplain forests in the Dry Corridor and how do they compare to floodplain forests of other regions?, 2) How do AGB and biomass accumulation vary across flood level and forest age?, 3) How do biomass accumulation rates vary across different livestock int ensities? Methods Study R egion The Amazon Dry Corridor has relatively low rainfall (1200 2200 mm yr 1 ) and prolonged dry seasons of 3 6 months < 100 mm month 1 (Figure 6 1; Sombroek 2001) Patches of dry forest and savanna within the corridor are relicts of drier periods within the Last Glacial Maximum (Anhuf et al. 2006) The Santarm region of Par state in Brazil lies within the Dry Corridor and has an average rainfall of 2180 mm yr 1 and 5 consecutive dry months < 100 mm (July December; Fitzjarrald et al. 2008) In the region, seasonal floods of the Amazon River peak between May and June, rising to an average of 7.5 m a.s.l. (1975 2008, Capitani a dos Portos Santarm
146 2008) and extending up to 40 km from the main channel. Floodplains in the region are a mosaic of natural grasslands, forested levees, lakes, and giant aroid stands of Montrichardia arborescens Forests occupy approximately 15% of the area (M. Crossa, personal communication, June 1, 2007), restricted to higher elevation levees within the floodplain. Historical land uses in floodplain forests include the planting of cacao ( Theobroma cacao ) and rubber tree s ( Hevea brasiliensis ) in t he 1800s to early 1900s, followed by deforestation for jute ( Corchorus capsularis ) plantations in 1940 1990 (Winklerprins 2006) Forests have regrown on abandoned jute plantations but currently face increasing activity by cattle and water buffalo in the floodplain (Sheikh et al. 2006) Study Design A network of 49 plots of 20 x 50 m was established in three levees within the floodplains of the Santarm municipality by P.S. in 1999 to evaluate the effects of cattle and water buffalo on vrzea forest sta nd dynamics. The plots were located in three communities of floodplain residents previously associated with the Vrzea Project of IPAM Instituto de Pesquisa Ambiental da Amaznia as well as on land owned by EMBRAPA This network is among the few semi permanent system of plots monitoring secondary floodplain forests in the Amazon. By 2008, 43 plots remained due to losses by deforestation (N=2), bank erosion (N=3), as well as changes in willingness to participate by landholders (N=1). In 2008, 6 new plots were added to the network to incorporate older forest used to estimate forest biomass in 2008, and 43 plots were used to measure change in aboveground biomass over the nine year study period. Forest age was determined based on informal interviews with landholders in 1999 and 2008. All t rees DBH (diameter at 1.3 m height) were measured at the end of the dry season in the low
147 water period (December January) in 199 9. Plots were censused again at the same time of year in 2003 and 2008 by C.L Plots were classified as light (1), medium (2), or heavy impact (3) of livestock in 1999 and 2008 based on average hoof print density in five 9 m 2 seedling plots (0 10 = light, 11 20 = medium, 21 80 pugmarks = heavy) and herd activity as observed in b ehavioral studies ( Sheikh 2002). Hoof print density was deemed a surrogate variable for herd densities in forests as they remained in soils throughout the low water season (Sheikh 2002) Average livestock activity over nine years was calculated from the 1999 and 2008 d ata (light, med light, medium, med heavy, and heavy ). I calculated flo od level as the average water depth in forests during the flood season from 1999 to 2008. Water depth was measured at 10 m intervals in forest plots with a weighted line during peak flooding in June 2006 (river level was 8.6 m a.s.l.). To calculate avera ge flood level over 9 years, I subtracted the difference between levels in 2006 and the avera ge for 1999 2008. Flood depth wa s thus a relative measure of the difference in maximum water column within plots. Forest plots were located on four different levees within the floodplain, two along the main Amazon channel and two along different branch es of the Amazon River (Figure 6 1) To control for differences in stand structure and species composition due to unmeasured variables such as water chemistry and soil structure, levee was considered a random effect in statistical models. To estimate ab oveground biomass and net biomass a ccumulation I measured forest inventories in 1999, 2003, and 2008. Tree h eight was estimated visually for each tree, with a sub sample of standing trees measured with a meter tape t o test for accuracy. W ood density values w ere obtained from previously
14 8 published values from species in floodplain forests of the Central and Western Amazon (Wittmann 2006) I calculated total aboveground woody biomass and biomass accumulation rates based on growth intervals in 1999 2004 and 2004 2008 Voucher spec imens were collected from a subsample of trees in collaboration with the University of Western Par of Santarm (UFOPA) and deposited at the INPA Manaus and UF OPA herbaria. Permission to collect non threatened species and deliver to the se herbaria was grant ed by SISBIO the federal bureau for the within country collection of biological material in Brazil (Authorization # 18177 1 ). Statistical Analyses Aboveground biomass of trees was calculated using allometric regression equations based on wood density, diam eter, and height. As allometric equations for trees in floodplain forests have not been calibrated, I used models from terra firme tropical forests developed by Cannell (Cannell 1984) as well as models for tropical dry forest and tropical moist forest (Chave et al. 2005) Tropical dry forest allometric equations were developed for trees with severe water stress > 5 mo, and thus may be more adequate for flooded trees with a perio d of slow growth during annual floods (Sch ngart et al. 2010) Of the 74 species recorded wood densit y was available for 43 species (Wittmann et al. 2011) W ood densities for 17 of the remaining species were calculated by genus level estimates for species occurring in vrzea ; 6 species were calculated by genus level estimates from trees in terra firme tropical forests; and 8 species, including unknown morphospecies were estimated by the average wood density of vrzea trees in Central and Western Amazonia (Schngart et al. 2010) Carbon was assumed to contribute 50% of biomass (Clark et al. 200 1, Saatchi et al. 2007)
149 (Cannell 1984) Dry tropical forest model (Chave et. al 2005) Moist tropical forest model (Chave et. al 2005) Trends in biomass accumulation and loss were observed at two time intervals (1999 2003 and 2003 2008) over a period of nine years The 43 sites with repeated measurements of DBH were grouped into three age categories to compare biomass trends in stand s of different age : 15 25 (N=6); 30 35 (N=21); 45 50 (N=16) years since agricultural abandonment. Forest biomass was compared across flood levels and forest age categories with repeated measures Linear Mixed Models fit by restricted maximum likelihood and an autoregressive correlation structure (Crawley 2007) Biomass was log transformed to achieve normality of model residuals, according to Shapiro Wilk tests. Due to the polynomial relationship between flood level and biomass, the quadratic term for flooding was included as a quadratic term in models (Crawley 2007) I accounted for heterogeneity in variance of biomass within age categ ories by using the VarIdent variance structure in Generalized Linear Models (Zuur et al. 2009) Average annual change in biomass was compared across flood levels and forest age categories with Linear Mixed Models fit by restricted maximum likelihood Differences in biomass and net biomass change between forest ag e categories were tested post hoc
150 HSD tests. All analyses were conducted in R 2.9.0 (R Dev elopment Core Team 2009) (Pinheiro et al. 2008) Results I identified 2568 of 2852 individuals in the 2008 inventory to species, an additional 216 to genus, and 36 to family; 32 trees were unknown Wood density among trees averaged 0.57 0.16 SD g/cm 3 and was highest in forests 30 35 y old (Tukey HSD, p < 0.05). Average AGB of forests 15 80 y old was 204 97 Mg ha 1 tropical forest model; 220 105 Mg ha 1 according to the Dry tropical forest model; and 211 87 Mg ha 1 according to moist tropical forest model, with no differences among model estimates (Tukey HSD, p < 0.05; Figure 6 2A). As such, I hereafter refer to values reported by the moist tropical forests as it provid ed an intermediate estimate for biomass. Across forest ages, AGB peaked at 268 25 Mg ha 1 in forest plots of 26 35 y, then tapered off at 197 8 to 213 9 Mg ha 1 in 80 yr old forest plots (Figure 6 3B). Although no main effect of age was found in gene ralized linear models, there were interactive effects of forest age with flooding and the quadratic term for flooding (Table 6 2). Across a flood gradient of 0 to 1.9 m depth, AGB peaked at moderate flood levels (0.7 1.3 m or approximately 100 130 d of f looding). AGB was predicted by a negative quadratic relationship in relation to flood level (Figure 6 2B; Table 6 2) Plots with exceptionally high AGB of 500 570 Mg ha 1 had large trees 71 149 cm DBH ( e.g. Ficus insipida, Triplaris surinamensis ). Net ab oveground biomass accumulation averaged 4.9 13.8 SD Mg ha 1 y 1 over nine years. Biomass accumulation ranged from 10.6 Mg ha 1 y 1 in young forests (15 25
151 y) to 2.8 Mg ha 1 y 1 in late successional forests of 40 50 y old. There were no effects of fores t age or flood level on biomass change (Figure 6 2B; Table 6 2). Net change in AGB was not correlated with flood level (t 41 = 1.28, p = 0.2 1 r = 0.20) Plots sustaining heavy cattle activity over time displayed negative changes in AGB over time (Figure 6 3). Heavily impacted stands had low biomass gains in tree growth, but no difference with other plots in terms of biomass lost via mortality or biomass gained via recruitment (Figure 6 3). There was no main effect of average cattle impact on bio mass accumulation (Table 6 3 ), but forests with sustained heavy cattle impacts had lower biomass accumulation than from forest stands with less impact (Tukey HSD < 0.05; Figure 6 3). By measuring the components of net biomass change, I found that 70% of bi omass gained was lost to stem mortality on an annual basis. Annual loss in AGB due to stem mortality was 11.0 9.3 Mg ha 1 y 1 or 2.8 2.8 % y 1 of total AGB. The contribution of recruited stems to annual gains in AGB averaged 1.6 1.2 Mg ha 1 y 1 or 0.22 0.30 % y 1 (Table 6 4). The remaining gains in AGB (98%) were attributed to diameter and height growth of standing trees. Discussion T rends in A boveground Biomass in A mazonian Floodplain Forest Secondary floodplain forests were a net carbon sink despite histories of intensive land use. To our knowledge, I provide the first estimates available for aboveground Corridor. I calculated mean aboveground biomass of forests 15 80 years old as ~210 Mg ha 1 comparable to that of less degraded forests of the Central Amazon. Amazonian floodplain forests, although characterized by seasonal or tidal flooding,
152 differed widely in aboveground biomass (Johnson et al. 2000) Such differences may be due to a combination of factors, including hydrological regime, climate, and land use history. Net annual rates of biomass accumulation were lower than those of floodplain forests of similar ages (Table 6 4) but similar to rates in upland forests. Flood level was a key predictor of forest biomass. Forests reached maximum biomass at moderately flooded sites (see also Mitsch and Ewel 1979, Ferreira 1997) I expected biomass to peak in high elevation fores ts with low flood levels, as a longer dry period would increase the growth season. The low AGB in high elevation forests may be due to die off of large, low flood tolerant species in high flood years ( e.g., Hura crepitans, Guarea guidonia ); harvesting of t rees for fuelwood ( e.g., Calycophyllum spruceanum); and the exposure of high elevation levees to wind and soil erosion. In the Central Amazon, stands of similar species composition and land use history had peak biomass in mid high elevation stands flooded 1 2 mo y 1 (Schngart 2003) However, in the Western Amazon floodplain forests of Peru, biomass was highest in low elevation tauhampa forest flooded 4 mo y 1 (Nebel et al. 2001a) Comparison of biomass trends across the flood gradient is complicated by within site variation in soil moisture availability, n utrient status, flood regime, and species composition, which all affect biomass. While the subsidy stress model predicts high biomass in stands with short term periodic flooding, I contribute to growing empirical evidence that the relationship between flo oding stress and forest biomass is site specific (Mitsch and Rust 1984, Megonigal et al. 1997) Forest age is a primary predictor of forest biomass, but its effect can be difficult to detect over short time periods (Steininger 2000) The effect of forest age on AGB was
153 dependent upon flood level, whereby differences between old and young forests were only det ected at moderate flood levels. Contrary to the common trend of increasing biomass with age, I found the highest AGB in forests of 30 35 years old. Similarly, a study of Central Amazon floodplain forests showed that AGB peaked at 30 50 y and declined in mature stands (Schngart et al. 2010) Upland secondary forests approach mature stand biomass as early as 30 years post agricultural abandonment (Letcher and Chazdon 2009) ; 72 years (Hughes et al. 1999) ; or > 80 years (Saldarriaga et al. 1988) The high biomass in younger floodplain forests may be caused by phases of succession charac terized by dense stands of fast growing long lived pioneers, Pseudobombax munguba (Schngart et al. 2010) and Triplaris surinamensis (this study). I have little evidence to suggest that variation in land use intensity had visible effects on forest biomass. Biomass in forests sustaining inten se livestock activity over time had relatively lower biomass than other stands. While browsing and trampling by livestock can decrease riparian vegetation biomass (Robertson and Rowling 2000, Kauffman et al. 2004) the low biomass in high disturbance sites may be related to other factors such as soil texture, geomorphology, and species composition. In the Amazon estuarine floodplain forests recoveri ng from abandoned jute and banana cultivation, AGB of forests 50 60 y old is double the values for the Santarm region (Johnson 1999) The fertility of floodplain forest soils and annual cycles of erosion and deposition are potential mechanisms by which these forests rapidly recover su ch large amounts of carbon in AGB following disturbance by small scale agriculture. Changes in Aboveground B iomass Estimated carbon sequestration by aboveground woody biomass was approximately half that of the Central and Western Amazon vrzea forests ( Tab le 6 1)
154 I expected higher annual biomass accumulation rates than other floodplain forests given the long dry season and high nutrient status of soils (Johnson et al. 2000) Nonetheless, early successional forests show a net influx of ~5 Mg C ha 1 yr 1 which is relatively high considering rates of C sequestration by AGB in early successional upland forests of 6 Mg ha 1 yr 1 (Steininger 2000) Late successional forests > 4 0 y old show C sequestration rates ( 1.40 0.62 Mg C ha yr ) equal to that of old growth upland forests in the Santarm region (Rice et al. 2004) Based on these comparisons, I conclude that while these forests have suffered multiple phases of deforestation and have low biodiversity, they may play an important role for C budgets in the region. Net change in aboveground biomass m ay have been low due to high losses to stem mortality of 70% of biomass gained. Similar results were found in a Western Ama zon secondary forest where gross biomass accumulation was 9.7 Mg ha 1 yr 1 of which half was lost annual to stem mortality (Valencia et al. 2009) Mortality in floodplain forests may be related to the death of pioneer species as a natural phase of succession. I observed high mortality of pioneer stems ( e.g., Cordia tetrandr a, Triplaris surinamensis ) in plots subjected to windstorms and strong currents along the main channel of the Amazon River. Mortality may also be a result of anthropogenic disturbances from long term cattle and water buffalo activity within forests, as we ll as removal of stems by firewood harvesting. There was no correlation between flood level and biomass accumulation to support a subsidy stress model for woody production in Amazonian floodplain forests. As found in river floodplain systems in temperate regions (Megonigal et al. 1997) woody biomass production was highly variable across all flood levels. Such variation
155 may be attributed to differences in species composition or land use. Biomass accumulation in secondary forests is known to be a function of many factors, including stand age, length of the growt h season, soil moisture availability, nutrient status, disturbance size and disturbance intensity as well as species composition (Johnson et al. 2000) For floodplain forests, I find supporting evidence for decreasing biomass accumulation with forest age, with wide variance within stand age classes. Despite the shorter growth season for tree s at lower elevations with sometimes slow flowing waters, I found no evidence to support lower biomass accumulation in low elevation forest stands. Disturbance by large herbivores can limit biomass production in forests (Robertson and Rowling 2000) but I did not find evidence to support this conclusion. Rather in this study the net loss of biomass in stands with heavy cattle impacts was a result of low tree growth rates, and not high tree mortality or low recruitment (Figure 6 4). Cattle and water buffalo could decrease tree growth rates via soil compaction a nd root damage, but I have no data to show such an effect. Growth rates are instead likely to be low due to the fact that 5 out of the 6 plots maintaining heavy cattle impacts are dominated by Cynometra bauhiniifolia a slow growing dense wood (WD=0.81) o f mid late successional stages. Given species composition and forest age, I would expect biomass accumulation to be low. Negative change in biomass appears to be attributed to the fact that biomass loss from stem mortality exceeds gains from stem growth and recruitment. Livestock could play a role in causing low recruitment rates. Finally, it is important to consider the potential sources of error in biomass calculations (Phillips et al. 2000) Among the largest sources of error for biomass
156 calculation is the selection of the allometric model for biomass estimation (Chave et al. 2004) The allometric equations applied here were developed for terra firme forests, as no equations for vrzea forests are available. Trees of the same species are shorter in the vrzea than in uplands (Worbes 1997) and thus models based on tapering of trunks in uplands may overest imate biomass for vrzea species. Although I incorporated height and wood density to increase accuracy, error for biomass estimates using such terms in models may still reach 14% (Nelson et al. 1999) I also suspect that there were overestimates of large trees with DBH > 30 (Nogueira et al. 2006) resulting in biomass estimates of 450 570 Mg ha 1 for some stands. Allometric equations based on trees from floodplain forests are clear ly necessary. In addition, errors in height estimation can be an important source of error in tree measurement component of biomass estimation (Chave et al. 2004) Finally, repeated measurements of DBH are less accurate than dend rometer bands for measurements of diameter growth. In addition, bark thickness, termite nests, vines, knots, etc. can change DBH measurements. Summary F loodplain forests may play a small but important role in carbon sequestration i n the car bon budgets f or Amazonian forests and aquatic habitats Our results suggest that secondary floodplain forests annually sequester substantial amounts of carbon I found substantial losses in biomass due to stem mortality, a potential result of both anthropogenic disturbance s ( e.g., livestock ranching, deforestation, and fuelwood harvesting) and natural disturbances ( e.g., extreme floods, windstorms). While total aboveground biomass peaked at moderate flood levels, annual biomass production was not related to flood level, as predicted by the subsidy stress model. Extreme climat e events such as drought and large floods, as well as land conversion for cattle pasture
157 are among the major threats to forests in the Amazon today (Malhi et al. 2008) Continual monitoring of floodplain forest response to such events will be important for understanding their resilience to climat e change and anthropogenic dist urbance and their role in regional carbon budgets for the Amazon.
158 Table 6 1. Comparison of mean biomass, carbon storage, and carbon sequestration for aboveground woody stems in vrzea forests of the Amazon B asin (adop ted from Shngart et al 2010 ). Location in vrzea Forest age (sample size) Biomass (Mg ha 1 ) C sequestration by AWB (Mg C ha 1 yr 1 ) Lower Amazon estuary 1 (Ilha Maraj) 193 18 Lower Amazon 12 15 (1) 77 21 estuary 2 20 30 (3) 247 (73) (Macap) 50 60 (3) 355 12.4 Lower Amazon 3 15 25 (6) 148 62 5.3 1.6 (Santarm, this study) 30 35 (21) 274 112 2.4 8.1 45 50 (16) 188 67 1.4 6.3 80 (3) 204 13 Central Amazon 4 44 (1) 258 3.7 (Manaus) 80 (7) 279 3.6 Central Amazon 5 7 18 3 2.81 0.43 (Mamirau) 20 117 9 8.45 0.49 50 261 10 7.17 0.58 120 230 9 3.74 0.16 240 239 11 2.73 0.13 Peruvian Amazon 6 (Loreto) unknown 344 487 7.43 1.5 Bolivia & Peru 7 old growth 1 95 357 0.49 0.1 Estimated regional 8 average for vrzea various 248 23 Means and SD are shown, with the exception of data from the Lower Ama zon estuary (2), which are SE. 1 (Almeida et al. 2004) 4 plots of 1 ha in Ilha Maraj and on the Rivers Par and Xingu in the tidal estuary; 2 (Johnson 1999) 3 1 ha plots 50 60 y old and 4 plots of 10 x 10 m 6 30 y old in Macap in the tidal estuary of the Amazon. 3 4 (Worbes 1997) Based on inventory data between 1981 1991 from 6 stands of vrzea on the Ilha Montrichardia near Manaus on the Amazon River. 5 (Schngart et al. 2010) cm DBH. 6 (Nebel et al. 2001 a) : 9 forest plots of 1 ha in the Loreto Dept. of Peru; trees > 10 cm DBH. 7 (Malhi et al. 2006) : Review of 6 plots in Bolivia and Peru, excluding those from Nebel et al 2001. 8 (Saatchi et al. 2007) : Remote Sensing model estimate based on a meta analysis including plots in floodplain forest in the Western Amazon (Peru, Bolivia, Ecuador, Colombia) and the Eastern Amazon at the mouth of the Amazon River (Maraj Island, Brazil).
159 Table 6 2. Summary of generalized linear m odel results for the effects of relative flood level, forest age, and their interactions on aboveground woody biomass in 2008 and net bi omass accumulation from 1999 2008. Coefficient estimates are shown for continuous variables. Estimate SE F value p value Aboveground woody biomass Forest age 6.64 0.62 0.01 0.92 Flooding 4.99 1.17 24.0 <0.0001 Flooding 2 1.51 0.45 13.78 0.0006 Age Flooding 1.36 0.63 4.64 0.037 Age Flooding 2 0.39 0.24 15.05 0.0004 Net biomass accumulation Forest age 0.83 0.44 Flooding 63.9 351.6 0.48 0.50 Forest age Flooding 0.54 0.59
160 Table 6 3. Summary of generalized linear m odel results for the effects of relative flood level, cattle activity and their interactions on aboveground woody biomass in 2008 and net biomass accumulation from 1999 2008 Coefficient estimates are shown for continuous variables. Estimate SE F value p valu e Aboveground woody biomass Cattle activity 0.87 0.53 1 .0 0 0.32 Flooding 1.65 1.61 37.6 <0.0001 Flooding 2 1.15 0.95 23. 0 <0.0001 Cattle Flooding 1.88 0.85 4.86 0.03 Cattle Flooding 2 1.04 0.49 0.06 0.81 Net biomass accumulation Cattle activity -0.95 0.45 Flooding 1685 3142 0.61 0.44 Cattle activity Flooding 0.30 0.88
161 Table 6 4 Rates of change in aboveground biomass in forest stands over 9 years (means & SD) by forest age Forest age category 15 25 y 30 35 y 40 50 y Mortality rate (Mg ha 1 yr 1 ) 5.4 5.8 11.5 8.0 12.5 9.3 Total gain (Mg ha 1 yr 1 ) 15.5 5.1 16.1 14.3 15.1 9.5 Growth rate (Mg ha 1 yr 1 ) 13.9 5.4 15.7 14.0 14.5 9.4 Recruitment rate (Mg ha 1 yr 1 ) 1.65 0.47 0.57 0.78 0.67 0.73 Net biomass change (Mg ha 1 yr 1 ) 10.6 3.3 4.89 16.3 2.79 12.7 Relative mortality rate (% yr 1 ) 1.8 1.3 2.5 1.6 3.6 4.0 Relative recruitment rate (% yr 1 ) 0.59 0.33 0.11 0.16 0.22 0.34 Relative growth rate (% yr 1 ) 4.6 1.3 2.6 2.0 3.2 2.0
162 Figure 6 1. Map of the number of consecutive months with rainfall < 100 mm, illustrating the Dry Corridor crossing the Amazon River at 52 58W. Bold points indicate locations where ground data for AWB of vrzea forest has been published ( Table 6 1 ); star indicates study location. Map reprinted by permission from Sombroek, W. 2001. Spatial and temporal patterns of Amazon rainfall Consequences for the planning of agricultural occupation and the protection of primary forests. Ambio 30 : 388 396.
163 A B C D Figure 6 2. Differences in biomass and biomass accumulation across forest age. A) Aboveground biomass across forest age for three model estimates (Cannell, dry forest, and moist forest). B) Annual biomass accumulation across forest age. C) Aboveground biomass across livestock activity levels. D) Annual biomass accumulation across livestock activity levels. Significant differences between means (with SD) to the p < 0.05 level from Tukey HSD tests are shown by letters above bars.
164 F igure 6 3 Annual biomass accumulation (means & SE) across average cattle impact levels. Lower case letters show significant differences between means within each model estimate (Tukey HSD tests, p < 0.05).
165 Figure 6 4 Annual biomass accumulation (means & SD) attributed to tree growth, mortality, and recruitment from 1999 2008 Lower case letters indicate differences between means according to Tukey HSD tests (p < 0.05).
166 CHAPTER 7 CONCLU SION S In this research I explored the effects of ecological drivers on regeneration and succession of secondary floodplai n forests. My results broaden the understanding of seedling ecology in tropical floodplain forests and may be applied to conservation and management of floodplain ecosystems in Amazonia. In Chapter 2 I aimed to emphasize the repeated cycles of land use for agriculture and forest products in Amazonian floodplain forests via a literature search of historic accounts of the vrzea economy and landscape from Pre Colombian period through the present da y Su ch a history may explain the lack of old growth stands or no impact zones in floodplain forests of the Santarm region with which to compare forests with livestock impacts and history of jute plantations The repeated recovery of floodplain forests from m ultiple cycles of deforestation suggest remarkable resiliency against disturbance. Although all the forests that were inventoried for this research had been cut for agriculture or silviculture or had been disturbed by livestock, I identified 88 woody spec ies in fewer than 5 ha of forest. Subsequent c hapters helped to explain how floodplain tree species recover from such disturbances. In Chapter 3, I reported germination rates of vrzea seeds and found evidence for species specific responses to saturatio n treatments. My results supported the broader conclusion that the vrzea is ecologically diverse, with tree species that differ by their phenology and disperse seeds via multiple vectors. I highlighted two common vrzea species with very different germi nation syndromes Crataeva benthamii which has fish dispersed seeds that tolerate many months of submersion, and Pseudobombax munguba which have wind dispersed seeds with low tolerance of submersion.
167 In Chapter 4, I tested the effects of three stressors flood duration, mechanical damage, and shade on seedling density and species richness, using a common garden experiment. Mechanical damage to stems of planted seedlings decreased survival by ~50% but did not interact with flood duration Species varied widely in their ability to resprout after mechanical damage. Seedlings experienced two major declines in survival over the course of a year: first, among damaged seedlings, during the dry season and, second, a t after flooding. This suggested that damage d seedlings may experien ce drought stress in the dry period a finding observed in other floodplains as well (Lopez and Kursar 2007) L ight availability in the understory emerged as a key factor for seedling growth and survival during the first year of establishment. In Chapter 5 I monitored the dynamics of succession i n floodplain forests of different livestock activity levels and age. F lood l evel, ranging from 0 to 2 m depth, wa s a major driver of forest stand structure as well as species and stem recruitment rates Impacts of livestock activ ity on stand structure and rates of change were undetectable, but increasing livestock density did decrease seedling density and number of species. Light availability and flood level interacted to affect seedling species density, such that seedling species density remained high across the flood gradient with high light availability. Overall, this study show ed that for all the factors that can affect the structure of seedling communities and forests, flooding emerges as the most important for explaining tem poral trends. Using the same database from floodplain forests in the Santarm region I calculated average aboveground woody biomass as 212 Mg ha 1 with a mean ra te of carbon sequestration by aboveground biomass of 2.5 Mg C ha 1 y 1 (Chapter 6) My
168 resul ts suggest that despite land use histories of deforestation for agriculture and livestock ranching, secondary forests sequester a substantial amount of carbon on an annual basis. Heavily impact ed forest stands displayed net carbon loss, a result of slow g rowth rates of certain tree species. Significance in Ecology My dissertation research applied key ecological concepts in plant community ecology to a critical ecosystem, tropical floodplain forests. A major gap in community ecology is how many factors wor k synergistically to affect tropical forest succession My results show ed how these interactive effects may be evident only i n certain phases of tree life history For example, w hile livestock activity and flood level interacted to limit seedling densit y these effects did not scale up to tree s Within the growing field of research on tropical seedling ecology, an increasing number of studies are recognizing the importance of seedling recovery from damage as a mechanism for persistence in the forest und erst ory. Prior to this study, I kne w of no published data on the ability to resprout among floodplain forest seedlings of the Amazon. I found that species vary bro a dly in their ability to survive and grow after mechanical damage. Surprisingly, experimen tal removal of ~50% of seedling biomass had no effect on species flood tolerance. This finding should open the door to further physiological study as to how floodplain seedlings are able to tolerate such prolonged exposure to anoxia, even when plants have lost all foliar tissue. Trade off models are an important concept within plant strategy theory to explain species distributions (Gilbert et al. 2006) These models are founded on the principle that resources are limited, and thus there are trade offs in the allocation of resources to growth, su rvival, and reproduction. I fou nd little evidence for a flood /shade tolerance
169 trade off to explain species survival patterns for Amazonian floodplain forest seedlings. Rather, a few species emerge as tolerant of many stressors, i.e. flooding, shade, and mechanical damage. My f indings suggest an escape vs. tolerance model for seedling establishment on the floodplain, whereby seedlings at low elevations are tolerant of prolonged submergence as well as other stresses (Parolin 2002) Such results show h ow dynamics of regeneration in tropical floodplains can differ from those of temperate floodplain forests (Battaglia and Sharitz 2006) Perhaps the largest growing area within ecology today is the effect of human activities on climate change and ecosystem processes related to carbon budgets. I provide d the first available data for carbon storage and sequestration in secondary Amazonian floodplain forests of the Dry Corridor region. I found that despite the multiple cycles of land use history, low diversity, and livestock impacts, that forests average ~210 Mg ha 1 with annual C sequestration rates of 2.5 Mg C ha 1 y 1 These values wer e comparable to other secondary forests and confirm the important rol e of successional forests for C sequestration In summary, u nderstanding how plant communities are established and change through time and space continues to be a major focus of research in plant community ecology. This dissertation res earch contribute s to the understanding of secondary forest dynamics in response to the interactive effects of stress and disturbance, lending insight as to how the stochastic effects of disturbance change seedling communities (Chase and Leib old 2003) In addition, I contribute d to the understanding of carbon uptake by fast growing floodplain forest species and the role of forested wetlands in the Amazon carbon budget. Ground based data on forest biomass a nd tree species
170 composition contr ibute valuable information to the debate on the role of secondary forests and wetlands in carbon sequestration. From one of the few semi permanent networks of floodplain forest inventory plots in the Amazon, the results on biomass accumulation, biod iversi ty, and regeneration fill a gap in our knowledge of tropical floodplain forest ecology in the Dry Corridor of the Amazon. These results also allow for a more accurate basin wide understanding of ecological and diversity gradients from the estuary to the W est Amazon. Implications for Conservation and Management The vrzea forests of the Amazon face diverse challenges and opportunities for conservation and management of floodplain resources. Each region is distinct in the resources harvested as well as th e socio political and cultural contexts that permit conservation planning and sustainable management to occur. For example, in the floodplains of the Amazon estuary, communities currently face challenges as to how to manage tidal floodplain forests for sm all scale timber (Anderson et al. 1995) One opportunity f or diminishing the demand on forests for timber is the harvesting of non timber forest products such as latex, aa fruits ( Euterpe oleracea ) and andiroba oil ( Carapa guianensis ; Fortini et al. 2006) In the floodplains of the Lower Amaz on in the region of Santarm, I hav e reviewed challenges to forest conservation such as the increasing intensity of cattle and water buffalo (Sheikh 2002) as well as small scale deforestation for agriculture (WinklerPrins 1999) An opportunity for conserving forests and improving their management is the use of floodplain forests for local fisheries. Approximately 120 species of fish in the Amazon consum e fruits from floodp lain forest trees (Horn et al. in press) of which 150 woody species serve as food resources (Goulding et al. 1993). In the Santarm region
171 alone, 27 woody flowering species of 22 families were found in the guts of two characin species Colossoma macropom um and Piaractus brachypomus highly valued both commercially and for local consumption (Lucas 2008) The eco nomic value of these fish that consume forest fruits during the flood season and r ely on forests as nur series is one of the reasons why floodplain forests in the Santarm region are still intact. As such, the sustainable management of the local fishing industry may provide an opportunity for conservation of floodplain forests in the region. Many species investigated in this research were identified as important food sources for Amazon fishes Informa tion on the regeneration of these species seed germination rates, conditions for seedling survival, recruitment of adult stems are directly applicable to understanding the status of fish fruit resources and designing plans for sustainable management. In addition, these data may be ap plied to tree planting pr ojects in the region to recuperate fish abundance in some communities. One local initiative exists in the Santarm region for community based tree planting project on floodplains degraded by livestock use (McGrath et al. 2005) At a broader scale this research provides an example of the synergistic effects of major threats in the Amazon today (Malhi et al. 2008) including the effects of flood ing and drought stress on forest re generation and the mechanisms by which cattle affect forest succession. In monitoring forest change across a period of exceptionall y high floods (2006 a nd 2008), I found the local ext irpation of two low flood tolerant timber species, Guarea guidonia and Hura crepitans In comparing forest stand structure today to that described by elderly residents and naturalists (Chapter 2), I found that other valuable species such as Carapa guianensis and Ceiba pentandra are also
172 virtually lost for the region. Landholders that have taken initiatives to enrich floodplain forests with valued specie s report that large floods kill planted seedlings and saplings of Carapa guianensis Increased extreme events such as drought and high floods are predicted by climate change scenarios f or the Amazon B asin (Malhi and Wright 2004) As such, it will become exceedingly important to understand how these supra annual events affect floodplain forest resources.
173 APPENDIX A LIST OF WOODY SPECIE S IN SANTARM FLOODP LAIN FORESTS Table A 1. Species list for woody species recorded in 2008 inventories ordered by decreasing Importance Value (IV) among tree Species Family IV Pseudobombax munguba Malvaceae (Bombacoideae) 23.07 Ficus insipida Moraceae 21.38 Laetia corymbulosa Salicaceae 20.46 Triplaris surinamensis Polygonaceae 19.83 Cynometra bauhiniifolia Fabaceae (Caesalpinioideae) 16.83 Cordia tetrandra Boraginaceae 15.96 Andira inermis Fabaceae (Faboideae) 15.12 Calycophyllum spruceanum Rubiaceae 10.45 Luehea cymulosa Malvaceae (Tilioideae) 9.51 Sapium glandulosum Euphorbiaceae 8.85 Crataeva benthamii Brassicaceae 8.75 Ocotea sp. Lauraceae 7.00 Spondias mombin Anacardiaceae 6.93 Vitex cymosa Lamiaceae 6.42 Albizia inundata Fabaceae (Mimosoideae) 6.32 Neea sp. Nyctaginaceae 5.40 Swartzia leptopetala Fabaceae (Faboideae) 4.77 Toulicia guianensis Sapindaceae 4.39 unknown 4.04 Fabaceae 1 Fabaceae 3.99 Coccoloba ovata Polygonaceae 3.92 Pterocarpus amazonum Fabaceae (Faboideae) 3.78 Eugenia sp. Myrtaceae 3.68 Albizia niopoides Fabaceae (Mimosoideae) 3.53 Inga cayennensis Fabaceae (Mimosoideae) 3.49 Lonchocarpus sp. Fabaceae (Faboideae) 3.42 Ficus sp. Moraceae 3.40 Myrtaceae 1 Myrtaceae 3.23 Genipa americana Rubiaceae 3.19 Piranhea trifoliata Euphorbiaceae 3.16 Pouteria glomerata Sapotaceae 3.11 Hura crepitans Euphorbiaceae 3.07 Gustavia augusta Lecythidaceae 2.56 Talisia cerasina Sapindaceae 2.40 Garcinia brasiliensis Clusiaceae 2.29 Lecythis pisonis Lecythidaceae 2.26 Zygia cataraensis Fabaceae (Mimosoideae) 2.19
174 Table A 1. Continued. Species Family IV Xylosma benthamii Fabaceae (Mimosoideae) 2.16 Sorocea duckei Moraceae 1.99 Crescentia amazonica Bignoniaceae 1.51 Bothriospora corymbosa Rubiaceae 1.37 Nectandra amazonum Lauraceae 1.33 Cupania latifolia Sapindaceae 1.22 Banara cf. nitida Flacourtiaceae 1.19 Simaba orinocensis Simaroubaceae 1.18 Zanthozylum compactum Rutaceae 1.15 Casearea aculeata Salicaceae 1.14 Trichilia singularis Meliaceae 1.13 Guazuma ulmifolia Malvaceae (Byttnerioideae) 1.07 Xylopia calophylla Annonaceae 1.06 Maytenus guianensis Celastraceae 0.95 Crudia amazonica Fabaceae (Caesalpinioideae) 0.94 Hevea brasiliensis Euphorbiaceae 0.91 Myrcia cf. deflexa Myrtaceae 0.89 Connarus incomptus Connaraceae 0.74 Lecointea amazonica Fabaceae (Faboideae) 0.70 Tabebuia barbata Bignoniaceae 0.60 Solanum sp. Solanaceae 0.59 Inga sp. 1 Fabaceae (Mimosoideae) 0.54 Annonaceae 1 Annonaceae 0.38 Inga stenoptera Fabaceae (Mimosoideae) 0.38 Cassia grandis Fabaceae (Caesalpinioideae) 0.36 Pouteria aff. sagotiana Sapotaceae 0.29 Homalium racemosum Salicaceae 0.29 Cecropia latiloba Urticaeae 0.28 Mouriri guianensis Melastomataceae 0.28 Inga sp. 3 Fabaceae (Mimosoideae) 0.28 Inga sp. 2 Fabaceae (Mimosoideae) 0.27 Faramea sp. Rubiaceae 0.27
175 APPENDIX B SOILS DATA FOR COMMON GARDEN PLOTS IN CHAPTER 4 The soil texture results agree d with other data collected in agricultural fields in the vrzea in the same region: Soils were SILT LOAM soils, approximately 75% sil t, with very little sand (0.4 13%), and clay (10 33%). Sand content was highest at the high eleva tion plots (12 13%) and dropped abruptly with decreasing elevati on, while clay content increased gradually as elevation decreased (Figure A 1 A,C ). There seemed to be no trend in silt con tent across elevation (Figur e A 1B the highest clay content, confirming local resident from 4.5 to 5.9, with the oldest and pH values. Bulk density ranged from 0.9 to 1.3 g/cm 3 normal for silt loam soils. Despite the fact that and water buffalo, the forest was also the lowest elevatio n, which may have compensate d for bulk density. As plots were selected off of cattle trails, soil bulk density was not expected to be above 1.4 g/cm 3 Soils were collected in the early rainy season January 23 27, 2008.
176 Table B 1. Soil textu re and bulk density in the 21 common garden plots of seedlings. Forest Age Plot Flood (m) BD 0 5 cm BD 5 10 cm pH % S and % S ilt % C lay Doroca 80 1 1.2 1.05 0.98 4.6 0.9% 88% 11% Doroca 80 2 1.7 0.77 0.90 4.5 2.2% 83% 15% Doroca 80 3 1.7 1.13 1.26 4.9 2.6% 82% 15% Doroca 80 4 1. 8 1.06 1.00 4.7 0.5% 84% 16% Doroca 80 5 2.3 0.90 1.16 4.9 0.3% 77% 23% Doroca 80 6 2.5 0.94 1.11 5.0 0.8% 78% 21% Doroca 80 7 2.5 0.86 1.16 5.2 0.4% 79% 20% Doroca Means 0.96 1.08 4.8 1.1% 82% 17% Grande 35 1 1.8 1.01 1.33 5.7 2.2% 75% 23% Grande 35 2 1. 8 1.02 1.25 5.6 2.0% 78% 21% Grande 35 3 2. 1 0.96 1.12 4.7 0.6% 83% 16% Grande 35 4 2. 2 0.96 1.22 5.1 2.0% 66% 32% Grande 35 5 2.3 1.14 1.27 5.7 2.1% 78% 20% Grande 35 6 2.5 1.01 1.32 5.2 1.8% 76% 22% Grande 35 7 2. 8 0.88 1.22 5.0 3.3% 78% 19% Mata Grande Means 1.00 1.25 5.3 2.0% 76% 22% Santino 80 1 1.0 1.27 1.40 5.2 12.6% 72% 16% Santino 80 2 1.2 1.13 1.20 5.0 5.2% 81% 13% Santino 80 3 1. 2 1.17 1.33 5.5 13.6% 74% 12% Santino 80 4 1.5 1.00 1.40 5.0 3.3% 82% 14% Santino 80 5 1. 8 0.82 1.27 5.2 0.9% 79% 20% Santino 80 6 1.3 1.21 1.35 5.9 2.3% 80% 17% Santino 80 7 1.2 1.01 1.37 5.6 7.7% 83% 10% Santino 80 8 1.6 0.77 1.06 5.5 1.4% 82% 17% Santino Means 1.05 1.30 5.4 5.9% 79% 15%
177 Table B 2. Soil nutrients and organic matter. Forest Age Plot Flood (m) OM P (mg/g) N (mg/g) Ca Ca+Mg Al Cu Mn Fe Zn Doroca 80 1 1.2 18.3 25.3 28.1 4.7 6.4 2.3 6.2 79.7 1146 10.2 Doroca 80 2 1.7 24.3 28.9 5.5 7.1 2.1 Doroca 80 3 1.7 18.3 5.6 125.0 4.6 6.8 3.2 6.6 50.6 1033 6.3 Doroca 80 4 1.75 35.6 25.1 31.8 5.4 7.7 2.1 6.9 75.2 1055 10.5 Doroca 80 5 2.3 11.8 33.2 7.0 9.8 2.1 Doroca 80 6 2.5 22.3 13.3 35.3 6.5 8.8 2.2 6.3 46.0 662 6.6 Doroca 80 7 2.5 23.7 39.1 5.6 7.5 2.3 Doroca Means 23.6 18.4 45.9 5.6 7.7 2.3 6.5 62.9 974 8.4 Grande 35 1 1.8 18.0 101.9 4.2 5.6 1.1 Grande 35 2 1.75 32.5 24.6 174.3 5.5 7.4 0.9 6.6 72.5 886 7.3 Grande 35 3 2.05 25.1 42.1 6.0 7.3 2.4 Grande 35 4 2.15 22.3 6.6 66.6 8.6 11.3 1.8 9.5 63.9 1552 9.0 Grande 35 5 2.3 17.9 101.8 5.6 7.5 1.7 Grande 35 6 2.5 6.8 84.9 6.6 9.2 1.9 Grande 35 7 2.75 35.8 6.4 34.7 7.3 10.0 1.9 8.0 49.0 891 8.0 Mata Grande Means 30.2 15.0 86.6 6.3 8.3 1.7 8.0 61.8 1110 8.1 Santino 80 1 1.0 32.8 35.2 4.6 6.0 1.0 Santino 80 2 1.2 25.5 35.5 5.0 6.5 1.4 Santino 80 3 1.15 20.3 33.4 28.3 4.4 5.6 1.0 6.9 54.7 960 9.0 Santino 80 4 1.5 27.1 36.3 5.2 6.9 1.0 Santino 80 5 1.75 16.5 15.5 40.4 7.5 10.0 1.5 9.3 89.9 984 13.9 Santino 80 6 1.3 27.1 39.7 5.6 7.4 1.2 Santino 80 7 1.2 37.2 29.5 34.5 12.4 14.0 0.0 4.5 89.7 569 8.6 Santino 80 8 1.6 19.3 78.5 5.7 7.7 1.7 Santino Means 24.6 25.3 41.9 6.5 8.3 1.1 6.9 78.1 838 10.5 Values for o rganic matter (OM) and nutrient levels were provided by the Soils Laboratory of EMBRAPA in Belm, Brazil.
178 A B C Figure B 1. T he relationship between flood level and soil texture in three floodplain forest s.
179 A B C Figure B 2. Bulk density at 0 5 cm ( diamonds) and 5 10 cm ( squares) in three floodplain forests, Mata Doroca (A), Mata Grande (B), and Mata Sant/Gil (C).
180 LIST OF REFERENCES Aber, J., R. P. Neilson, S. McNulty, J. M. Lenihan, D. Bachelet, and R. J. Drapek. 2001. Forest processes and global environmental change: Predicting the effects of individual and multiple stressors. Bioscience 51 :735 751. Agrawal, A. and C. C. Gibson. 199 9. Enchantment and disenchantment: the role of community in natural resource conservation. World Development 27 :629 649. Aide, T. M., J. K. Zimmerman, L. Herrera, M. Rosario, and M. Serrano. 1995. Forest recovery in abandoned tropical pastures in Puerto Rico. Forest Ecology And Management 77 :77 86. Alho, C. J. R., T. E. Lacher, and H. C. Goncalves. 1988. Environmental degrad ation in the Pantanal ecosystem in Brazil, the world's largest wetland is being threatened by human activities. Bioscience 38 :164 171. Ali, S., F. Zeng, S. Cai, B. Qiu, and G. Zhang. 2011. The interaction of salinity and chromium in the influence of barl ey growth and oxidative stress. Plant Soil and Environment 57 :153 159. Almeida, O. 2004. Fisheries Management in the Brazilian Amazon. Ph.D. dissertation. Imperial College of Science, Technology and Medicine, London, England. Almeida, S. S., D. D. Amaral, and A. S. L. Silva. 2004. Floristic analysis and structure of tidal flooded forests in the Amazonian estuary. Acta Amazonica 34 :513 524. Alvarez Clare, S. and K. Kitajima. 2009. Susceptibility of tree seedlings to biotic and abiotic hazards in the understo ry of a moist tropical forest in Panama. Biotropica 41 :47 56. Anderson, A. B., P. Magee, A. Gely, and M. A. G. Jardim. 1995. Forest management patterns in the floodplain of the Amazon estuary. Conservation Biology 9 :47 61. Anderson, A. B., I. Mousasticoshv ily, and D. S. Macedo. 1999. Logging of Virola surinamensis in the Amazon floodplain: Impacts and alternatives. Pages 119 133 in C. Padoch, J. M. Ayres, M. Pinedo Vasquez, and A. Henderson, editors. Vrzea: diversity, development, and conservation of Amazo nia's whitewater floodplains. The New York Botanical Garden Press, Bronx, NY. Anderson, R. L. 1999. Colonization as Exploitation in the Amazon Rain Forest 1758 1910. University Press of Florida, Gainesville. Anhuf, D., M. P. Ledru, H. Behling, F. W. D. C. Jr., R. C. Cordeiro, T. V. d. Hammen, I. Karmann, J. A. Marengo, P. E. D. Oliveira, L. Pessenda, A. Siffedine, A. L. Albuquerque, and P. L. D. S. Dias. 2006. Paleo environmental change in Amazonian and African rainforest during the LGM. Palaeogeography, Pa laeoclimatology, Palaeoecology 239 :510 527.
181 Araujo Lima, C. and M. Goulding. 1997. So Fruitful a Fish: Ecology, Conservation, and Aquaculture of the Amazon's Tambaqui. Columbia University Press, New York. Araujo Lima, C. and M. Goulding. 1998. Os Frutos do Tambaqui: Ecologia, Conservao e Cultivo na Amaznia. Sociedade Civil Mamirau MCT CNPq, Tef. Arnold, J. S. 2004. What is the value of participation?: building social capital to support sustainable management of Tohono O'Odham Rangelands. Master's The sis. Graduate College. University of Arizona, Tucson. Asner, G. P., G. V. N. Powell, J. Mascaro, D. E. Knapp, J. K. Clark, J. Jacobson, T. Kennedy Bowdoin, A. Balaji, G. Paez Acosta, E. Victoria, L. Secada, M. Valqui, and R. F. Hughes. 2010. High resolutio n forest carbon stocks and emissions in the Amazon. Proceedings of the National Academy of Sciences of the United States of America 107 :16738 16742. Baker, T. R., O. L. Phillips, Y. Malhi, S. Almeida, L. Arroyo, A. Di Fiore, T. Erwin, N. Higuchi, T. J. Killeen, S. G. Laurance, W. F. Laurance, S. L. Lewis, A. Monteagudo, D. A. Neill, P. N. Vargas, N. C. A. Pitman, J. N. M. Silva, and R. V. Martinez. 200 4. Increasing biomass in Amazonian forest plots. Philosophical Transactions of the Royal Society of London Series B Biological Sciences 359 :353 365. Balslev, H., J. Luteyn, B. Ollgaard, and L. B. Holm Nielsen. 1987. Composition and structure of adjacent un flooded and floodplain forest in Amazonian Ecuador. Opera Botanica:37 58. Baraloto, C. and P. M. Forget. 2007. Seed size, seedling morphology, and response to deep shade and damage in neotropical rain forest trees. American Journal of Botany 94 :901 911. Ba skin, C. C. and J. M. Baskin. 1998. Seeds: Ecology, Biogeography, and Evolution of Dormancy and Germination. Academic Press, San Diego, CA. Bates, D. and M. Maechler. 2009. lme4: Linear mixed effects models using S4 classes. R package version 0.999375 31. http://CRAN.R project.org/package=lme4 Bates, H. W. 1864/1962. The Naturalist on the River Amazons. University of California Press, Berkeley. Battaglia, L. L., B. S. Collins, and R. R. Sharitz. 2004. Do published tolerance ratings and dispersal factors predict species distributions in bottomland hardwoods? Forest Ecology and Management 198 :15 30. Battaglia, L. L., S. A. Fore, and R. R. Sharitz. 2000. Seedling emergence, survival and size in relation to light and water availability in two bottomland hardwood species. Journal of Ecology 88 :1041 1050.
182 Battaglia, L. L. and R. R. Sharitz. 2006. Responses of floodplain forest species to spatially condensed gradients: a test of the flood shade tolerance tradeo ff hypothesis. Oecologia 147 :108 118. Belsky, A. J., A. Matzke, and S. Uselman. 1999. Survey of livestock influences on stream and riparian ecosystems in the western United States. Journal of Soil and Water Conservation 54 :419 431. Bolker, B. M., M. E. Bro oks, C. J. Clark, S. W. Geange, J. R. Poulsen, M. H. H. Stevens, and J. S. S. White. 2009. Generalized linear mixed models: a practical guide for ecology and evolution. Trends in Ecology & Evolution 24 :127 135. Bond, W. J. and J. J. Midgley. 2001. Ecology of sprouting in woody plants: the persistence niche. Trends in Ecology & Evolution 16 :45 51. Bradford, K. J. 1995. Water relations in seed germination. Pages 351 396 in J. Kigel and G. Galili, editors. Seed Development and Germination. Marcel Dekker, Inc., New York. Brady, N. C. and R. R. Weil. 2000. Elements of the Nature and Properties of Soils. Prentice Hall, Inc., Upper Saddle River, NJ. Brown, R. T. and J. T. Curtis. 1952. The upland conifer hardwood forests of northern Wisconsin. Ecol Monogr 22 :217 234. Butler, L. G., K. Kielland, T. S. Rupp, and T. A. Hanley. 2007. Interactive controls of herbivory and fluvial dynamics on landscape vegetation patterns on the Tanana River floodplain, interior Alaska. Journal of Biogeography 34 :1622 1631. Canham, C. D., R. K. Kobe, E. F. Latty, and R. L. Chazdon. 1999. Interspecific and intraspecific variation in tree seedling survival: effects of allocation to roots versus carbohydrate reserves. Oecologia 121 :1 11. Cannell, M. G. R. 1984. Woody biomass of forest stands. Forest Ecology And Management 8 :299 312. Capitania dos Portos Santarm 2008. Observaces pluviomtricas do Rio Tapajs Leitura da rgua. Cattanio, J. H., A. B. Anderson, and M. S. Carvalho. 2002. Floristic composition and topographic variation in a tidal floodplain forest in the Amazon Estuary. Revista Brasileira de Botanica 25 (4) 419 430. Chambers, R. and M. K. McBeth. 1992. Community enc ouragement: Returning to the basis for community development. Journal of the Community Development Society 23 :20 38.
183 Chao, A. 1984. Non parametric estimation of the number of classes in a population. Scandinavian Journal of Statistics 11 :265 270. Chao, A., R. L. Chazdon, R. K. Colwell, and T. J. Shen. 2005. A new statistical approach for assessing similarity of species composition with incidence and abundance data. m3m Monografias Tercer Milenio 4 :85 96. Chapin, F. S., A. J. Bloom, C. B. Field, and R. H. Wa ring. 1987. Plant responses to multiple environmental factors. Bioscience 37 :49 57. Chase, J. M. and M. A. Leibold. 2003. Ecological Niches: Linking Classical and Contemporary Approaches. University of Chicago Press, Chicago. Chave, J., C. Andalo, S. Brown M. A. Cairns, J. Q. Chambers, D. Eamus, H. Folster, F. Fromard, N. Higuchi, T. Kira, J. P. Lescure, B. W. Nelson, H. Ogawa, H. Puig, B. Riera, and T. Yamakura. 2005. Tree allometry and improved estimation of carbon stocks and balance in tropical forests. Oecologia 145 :87 99. Chave, J., R. Condit, S. Aguilar, A. Hernandez, S. Lao, and R. Perez. 2004. Error propagation and scaling for tropical forest biomass estimates. Philosophical Transactions of the Royal Society of London Series B Biological Sciences 35 9 :409 420. Chazdon, R. L. 2003. Tropical forest recovery: Legacies of human impact and natural disturbances. Perspectives in Plant Ecology Evolution and Systematics 6 :51 71. Chazdon, R. L., S. G. Letcher, M. van Breugel, M. Martinez Ramos, F. Bongers, and B. Finegan. 2007. Rates of change in tree communities of secondary Neotropical forests following major disturbances. Philosophical Transactions of the Royal Society B Biological Sciences 362 :273 289. Chinea, J. D. 2002. Tropical forest succession on abando ned farms in the Humacao Municipality of eastern Puerto Rico. Forest Ecology and Management 167 :195 207. Clark, D. A., S. Brown, D. W. Kicklighter, J. Q. Chambers, J. R. Thomlinson, and J. Ni. 2001. Measuring net primary production in forests: Concepts and field methods. Ecological Applications 11 :356 370. Clark, D. B. and D. A. Clark. 1989. The role of physical damage in the seedling mortality regime of a neotropical rainforest. Oikos 55 Claro Jr, L., E. Ferreira, J. Zuanon, and C. Araujo Lima. 2004. O ef eito da floresta alagada na alimentao de peixes onvoros em lagos de vrzea da Amaznia Central, Brasil. Acta Amazonica 34 :133 137. Coleman, B. D. 1981. On random placement and species area relations. Mathematical Biosciences 54 :191 215.
184 Collett, D. 2003 Modeling Survival Data in Medical Research, 2 nd Edition. Chapman and Hall, Washington DC. Colwell, R. K. 2009. EstimateS: Statistical estimation of species richness and shared species from samples. Version 8.2. User's Guide and application published at: http://purl.oclc.org/estimates Condit, R., S. P. Hubbell, and R. B. Foster. 1995. Mortality rates of 205 neotropical tree and shrub species and the impact of a severe drought. Ecological Monographs 65 :419 439. Conner, W. H. and J. W. Day. 1976. Productivity and composition of a baldcypress water tupelo site and a bottomland hardwood site in a Louisiana swamp. American Journal of Botany 63 :1354 1364. Conner, W. H., L. Mihalia, and J. Wolfe. 2002. Tree community s tructure and changes from 1987 to 1999 in three Louisiana and three South Carolina forested wetlands. Wetlands 22 :58 70. Cornaglia, P. S., G. E. Schrauf, and V. A. Deregibus. 2009. Flooding and grazing promote germination and seedling establishment in the perennial grass Paspalum dilatatum Austral Ecology 34 :343 350. Cornejo, X. and H. H. Iltis. 2008. A revision of the American species of the genus Crateva (Capparaceae). Harvard Papers in Botany 13 :121 135. Cottingham, K. L., J. T. Lennon, and B. L. Brown. 2005. Knowing when to draw the line: designing more informative ecological experiments. Fronteirs in Ecology and the Environment 3 :145 152. Craine, J. M. 2005. Reconciling plant strategy theories of Grime a nd Tilman. Journal of Ecology 93 :1041 1052. Crawley, M. J. 2007. The R Book. John Wiley & Sons, Ltd., West Sussex, England. Dahdouh Guebas, F., D. Vrancken, T. Ravishankar, and N. Koedam. 2006. Short term mangrove browsing by feral water buffalo: conflict between natural resources, wildlife and subsistence interests? Environmental Conservation 33 :157 163. Daws, M. I., S. Bolton, D. Burslem, N. C. Garwood, and C. E. Mullins. 2007. Loss of desiccation tolerance during germination in neo tropical pioneer seeds : Implications for seed mortality and germination characteristics. Seed Science Research 17 :273 281. Deiller, A. F., J. M. N. Walter, and M. Tremolieres. 2003. Regeneration strategies in a temperate hardwood floodplain forest of the Upper Rhine: sexual ver sus vegetative reproduction of woody species. Forest Ecology and Management 180 :215 225.
185 del Tredici, P. 2001. Sprouting in temperate trees: a morphological and ecological review. Botanical Review 67 :121 140. Denevan, W. M. 1984. Ecological heterogeneity a nd horizontal zonation of agriculture in the Amazon floodplain. Pages 311 336 in M. Schmink and C. H. Wood, editors. Frontier Expansion in Amazonia. University of Florida Press, Gainesville. Denevan, W. M. 1996. A bluff model of riverine settlement in preh istoric Amazonia. Annals of the Association of American Geographers 86 :654 681. Edwards, W. H. 1847/2004. Voyage Up the River Amazon. The Narrative Press, Santa Barbara. Embrapa. 1997. Manual de Mtodos de Anlise de Solo, 2 a edio. Empresa Brasileira de Pesquisa Agropecuria Centro Nacional de Pesquisa de Solos, Rio de Janeiro. Ernst, K. A. and J. R. Brooks. 2003. Prolonged flooding decreased stem density, tree size and shifted composition towards clonal species in a central Florida hardwood swamp. Forest Ecology and Management 173 :261 279. Ferreira, C. S., M. T. F. Piedade, A. C Franco, J. F. Carvalho Goncalves, and W. J. Junk. 2009. Adaptive strategies to tolerate prolonged flooding in seedlings of floodplain and upland populations of Himatanthus sucuuba, a Central Amazon tree. Aquatic Botany 90 :246 252. Ferreira, L. V. 1997. E ffects of the duration of flooding on species richness and floristic composition in three hectares in the Jau National Park in floodplain forests in Central Amazonia. Biodiversity and Conservation 6 :1353 1363. Ferreira, L. V. 2000. Effects of flooding dura tion on species richness, floristic composition and forest structure in river margin habitat in Amazonian blackwater floodplain forests: implications for future design of protected areas. Biodiversity and Conservation 9 :1 14. Ferreira, L. V., S. S. Almeida and P. Parolin. 2010. Amazonian white and blackwater floodplain forests in Brazil: Large differences on a small scale. Ecotropica 16 :31 41. Fine, P. V. A., I. Mesones, and P. D. Coley. 2004. Herbivores promote habitat specialization by trees in amazonia n forests. Science 305 :663 665. Finegan, B. 1996. Pattern and process in neotropical secondary rain forests: The first 100 years of succession. Trends in Ecology & Evolution 11 :119 124. Finlayson, C. M. 2005. Plant ecology of Australia's tropical floodplai n wetlands: A review. Annals of Botany 96 :541 555.
186 Fitzjarrald, D. R., R. K. Sakai, O. L. L. Moraes, R. C. de Oliveira, O. C. Acevedo, M. J. Czikowsky, and T. Beldini. 2008. Spatial and temporal rainfall variability near the Amazon Tapajos confluence. Jour nal of Geophysical Research Biogeosciences 113 :Art. No. G00B11. Flynn, D. F. B., M. Uriarte, T. Crk, J. B. Pascarella, J. K. Zimmerman, T. M. Aide, and M. A. C. Ortiz. 2010. Hurricane disturbance alters secondary forest recovery in Puerto Rico. Biotropica 42 :149 157. Fortini, L. B., E. M. Bruna, D. J. Zarin, S. S. Vasconcelos, and I. S. Miranda. 2010. Altered resource availability and the population dynamics of tree species in Amazonian secondary forests. Oecologia 162 :923 934. Fortini, L. B., F. G. Rabelo, and D. J. Zarin. 2006. Mixed potential for sustainable forest use in the tidal floodplain of the Amazon River. Forest Ecology and Management 231 :78 85. Frangi, J. L. and A. E. Lugo. 1991. Hurricane Damage to a Flood Plain Forest in the Luquillo Mountains of Puerto Rico. Biotropica 23 :324 335. Frangi, J. L. and A. E. Lugo. 1998. A flood plain palm forest in the Luquillo Mountains of Puerto Rico five years after Hurricane Hugo. Biotropica 30 :339 348. Frazer, G. W., C. D. Canham, and K. P. Lertzman. 1999. Gap Light Analyzer (GLA), Version 2.0: Imaging software to extract canopy structure and gap light transmission indices from true colour fisheye photographs, users manual and program documentation. Copyright 1999: Simon Fraser University, Burnaby, British Co lumbia, and the Institute of Ecosystem Studies, Millbrook, New York. Frazer, G. W., R. A. Fournier, J. A. Trofymow, and R. J. Hall. 2001. A comparison of digital and film fisheye photography for analysis of forest canopy structure and gap light transmissio n. Agricultural and Forest Meteorology 109 :249 263. Furch, K. 1997. Chemistry of vrzea and igap soils and nutrient inventory of their floodplain forests. Pages 47 68 in W. Junk, editor. The Central Amazon Floodplain: Ecology of a Pulsing System. Springer, Berlin. Furneaux, R. 1969. The Amazon. Western Printing Services Ltd, Bristol. Garwood, N. C. 1989. Tropical soil seed banks: A review. in M. A. Leck, editor. Ecology of Soil Seed Banks. Academic Press, San Diego. Gatewood, S. and G. Cornwell. 1976. An analysis of cattle ranching in Kissimmee River basin. Tallahassee, FL. Gauriguata, M. R. and R. Ostertag. 2001. Neotropical secondary forest succession: changes in structur al and functional characteristics. Forest Ecology and Management 148 :185 206.
187 Geissler, K. and A. Gzik. 2008. The impact of flooding and drought on seeds of Cnidium dubium, Gratiola officinalis, and Juncus atratus three endangered perennial river corridor plants of Central European lowlands. Aquatic Botany 89 :283 291. Gilbert, B., S. J. Wright, H. C. Muller Landau, K. Kitajima, and A. Hernandez. 2006. Life history trade offs in tropical trees and lianas. Ecology 8 7 :1281 1288. Godoy, J. R., G. Petts, and J. Salo. 1999. Riparian flooded forests of the Orinoco and Amazon Basins: A comparative review. Biodiversity and Conservation 8 :551 586. Goulding, M. 1980. The Fishes and the Forest: Explorations in Amazonian natura l history. University of California Press, Berkeley. Goulding, M., R. Barthem, and E. Ferreira. 2003. The Smithsonian Atlas of the Amazon. Smithsonian Books, Washington. Goulding, M., N. J. H. Smith, and D. J. Mahar. 1993. Floods of Fortune: Ecology and ec onomy along the Amazon. Columbia University Press, New York. Grauel, W. T. 2004. Ecology and Management of Wetland Forests Dominated by Prioria copaifera in Darien, Panama University of Florida, Gainesville, FL. Green, P. T. and P. A. Juniper. 2004. Seed m ass, seedling herbivory and the reserve effect in tropical rainforest seedlings. Functional Ecology 18 :539 547. Grime, J. P. 1977. Evidence for Existence of 3 Primary Strategies in Plants and Its Relevance to Ecological and Evolutionary Theory. American Na turalist 111 :1169 1194. Grime, J. P. 1979. Plant strategies and vegetation processes. John Wiley & Sons, New York. Grubb, P. J. 1977. Maintenance of species richness in plant communities Importance of regeneration niche. Biological Reviews of the Cambrid ge Philosophical Society 52 :107 145. Guyot, J. L., J. M. Jouanneau, L. Soares, G. R. Boaventura, N. Maillet, and C. Lagane. 2007. Clay mineral composition of river sediments in the Amazon Basin. Catena 71 :340 356. Hall, R. B. W. and P. A. Harcombe. 1998. F looding alters apparent position of floodplain saplings on a light gradient. Ecology 79 :847 855. Harms, K. E. and J. W. Dalling. 1997. Damage and herbivory tolerance through resprouting as an advantage of large seed size in tropical trees and lianas. Journ al of Tropical Ecology 13 :617 621.
188 Heckenberger, M. J., A. Kuikuro, U. T. Kuikuro, J. C. Russell, M. Schmidt, C. Fausto, and B. Franchetto. 2003. Amazonia 1492: Pristine forest or cultural parkland? Science 301 :1710 1714. Hiraoka, M. 1995. Land use changes in the Amazon estuary. Global Environmental Change Human and Policy Dimensions 5 :323 336. Holdo, R. M., R. D. Holt, and J. M. Fryxell. 2009. Grazers, browsers, and fire influence the extent and spatial pattern of tree cover in the Serengeti. Ecological Ap plications 19 :95 109. Holl, K. D. 1999. Factors limiting tropical rain forest regeneration in abandoned pasture: Seed rain, seed germination, microclimate, and soil. Biotropica 31 :229 242. Hoorne, C. 2006. The birth of the mighty Amazon. Scientific America n 294 :52 59. Hughes, F. M. R. 1988. The ecology of African floodplain forests in semi arid and arid zones A review. Journal of Biogeography 15 :127 140. Hughes, R. F., J. B. Kauffman, and V. J. Jaramillo. 1999. Biomass, carbon, and nutrient dynamics of se condary forests in a humid tropical region of Mexico. Ecology 80 :1892 1907. James, J. J., R. L. Tiller, and J. H. Richards. 2005. Multiple resources limit plant growth and function in a saline alkaline desert community. Journal of Ecology 93 :113 126. Jansen, A. and A. I. Robertson. 2001. Relationships between livestock management and the ecological condition of riparian habitats along an Australian floodplain river. Journal of Applied Ecology 38 :63 75. Johnson, C. M. 1999. Post disturbance recovery of aboveground biomass and nutrients in tropical, temperate, and boreal forests. University of Pennsylvania, Philadelphia, Pennsylvania. Johnson, C. M., D. J. Zarin, and A. H. Johnson. 2000. Post disturbance aboveground biomass accumulation in global secondar y forests. Ecology 81 :1395 1401. Junk, W. J. 1997. The central Amazon floodplain: ecology of a pulsing system. Springer, New York. Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river floodplain systems. Proceedings of the In ternational Large River Symposium, Canadian Special Publications in Fisheries and Aquatic Sciences 106 :110 127. Junk, W. J. and C. N. de Cunha. 2005. Pantanal: a large South American wetland at a crossroads. Ecological Engineering 24 :391 401.
189 Jutila, H. M. 2001. Effect of flooding and draw down disturbance on germination from a seashore meadow seed bank. Journal of Vegetation Science 12 :729 738. Kabeya, D. and S. Sakai. 2005. The relative importance of carbohydrate and nitrogen for the resprouting ability o f Quercus crispula seedlings. Annals of Botany 96 :479 488. Kammesheidt, L. 1998. The role of tree sprouts in the restorations of stand structure and species diversity in tropical moist forest after slash and burn agriculture in Eastern Paraguay. Plant Ecol ogy 139 :155 165. Kauffman, J. B. and W. C. Krueger. 1984. Livestock impacts on riparian ecosystems and streamside management implications... A review. Journal of Range Management 37 :430 438. Kauffman, J. B., A. S. Thorpe, and E. N. J. Brookshire. 2004. Liv estock exclusion and belowground ecosystem responses in riparian meadows of Eastern Oregon. Ecological Applications 14 :1671 1679. Kennedy, K. A. and P. A. Addison. 1987. Some considerations for the use of visual estimates of plant cover in biomonitoring. J ournal of Ecology 75 :151 157. Kettler, T. A., J. W. Doran, and T. L. Gilbert. 2001. Simplified method for soil particle size determination to accompany soil quality analyses. Soil Science Society of America Journal 65 :849 852. Kingsford, R. T. 2000. Ecolog ical impacts of dams, water diversions and river management on floodplain wetlands in Australia. Austral Ecology 25 :109 127. Kitajima, K. 1994. Relative Importance of Photosynthetic Traits and Allocation Patterns as Correlates of Seedling Shade Tolerance o f 13 Tropical Trees. Oecologia 98 :419 428. Kitajima, K. 1996. Cotyledon functional morphology and patterns of seed reserve utilization by neotropical tree seedlings. Pages 193 210 in M. D. Swaine, editor. Ecology of Tropical Forest Tree Seedlings. Partheno n Press, New York. Kitajima, K. 2003. Impact of cotyledon and leaf removal on seedling survival in three tree species with contrasting cotyledon functions. Biotropica 35 :429 434. Klimo, E. and H. Hager. 2001. The Floodplain Forests in Europe: Current Situa tion and Perspectives. Brill, Leiden. Klooster, D. J. 2002. Toward adaptive community forest management: Integrating local forest knowledge with scientific forestry. Economic Geography 78 :43 70. Kozlowski, T. T. 1984. Plant responses to flooding of soil. B ioscience 34 :162 167.
190 Kozlowski, T. T. 1992. Carbohydrate sources and sinks in woody plants. Botanical Review 58 :107 222. Kozlowski, T. T. and S. G. Pallardy. 1997. Growth control in woody plants. Academic Press, San Diego, CA. Kozlowski, T. T. and S. G. Pallardy. 2002. Acclimation and adaptive responses of woody plants to environmental stresses. Botanical Review 68 :270 334. Kubitzki, K. and A. Ziburski. 1994. Seed dispersal in flood plain forests of Amazonia. Biotropica 26 :30 43 Laurance, W. F. and T. J. Curran. 2008. Impacts of wind disturbance on fragmented tropical forests: A review and synthesis. Austral Ecology 33 :399 408. Leck, M. A. 1989. Wetland seed banks. Pages 283 305 in M. A. Leck, T. Parker, and R. L. Simpson, edito rs. Ecology of Soil Seed Banks. Academic Press, San Diego, California. Lemmon, P. E. 1956. A spherical densiometer for estimating forest overstory density. Forest Sci 2 :314 320. Letcher, S. G. and R. L. Chazdon. 2009. Rapid recovery of biomass, species ric hness, and species composition in a forest chronosequence in Northeastern Costa Rica. Biotropica 41 :608 617. Lopez, O. R. 2001. Seed flotation and postflooding germination in tropical terra firme and seasonally flooded forest species. Funcional Ecology 15 : 763 771. Lopez, O. R. and T. A. Kursar. 2007. Interannual variation in rainfall, drought stress and seedling mortality may mediate monodominance in tropical flooded forests. Oecologia 154 :35 43. Lucas, C. M. 2008. Within flood season variation in fruit consumption and seed dispersal by two characin fishes of the Amazon. Biotropica 40 :581 589. Malhi, Y., T. R. Baker, O. L. Phillips, S. Almeida, et al 2004. The above ground coarse wood productivity of 104 Neotropical forest plots. Global Change Biology 10 :563 591. Malhi, Y., J. T. Roberts, R. A. Betts, T. J. Killeen, W. H. Li, and C. A. Nobre. 2008. Climate change, deforestation, and the fate of the Amazon. Science 319 :169 172. Malhi, Y., D. Wood, T R. Baker, J. Wright, et al 2006. The regional variation of aboveground live biomass in old growth Amazonian forests. Global Change Biology 12 :1107 1138.
191 Malhi, Y. and J. Wright. 2004. Spatial patterns and recent trends in the climate of tropical rainfor est regions. Philosophical Transactions of the Royal Society of London Series B Biological Sciences 359 :311 329. Mantel, N. and W. Haenszel. 1959. Statistical aspects of data from retrospective studies of disease. Journal of the National Cancer Institute 2 2 :719 748. Marengo, J. A., C. A. Nobre, J. Tomasella, M. D. Oyama, G. S. De Oliveira, R. De Oliveira, H. Camargo, L. M. Alves, and I. F. Brown. 2008. The drought of Amazonia in 2005. Journal of Climate 21 :495 516. Marinho, T. A. S., M. T. F. Piedade, and F Wittmann. in press. Distribution and population structure of four Central Amazonian high vrzea timber species. Wetlands Ecology and Management, DOI 10.1007/s11273 010 9186 y. Markham, C. R. 1963. Expeditions into the Valley of the Amazons, 1539, 1540, 1 639. Burt Franklin, Publisher, New York. Marthews, T. R., D. Burslem, S. R. Paton, F. Yanyuez, and C. E. Mullins. 2008. Soil drying in a tropical forest: Three distinct environments controlled by gap size. Ecological Modelling 216 :369 384. Martin, B. A., S F. Cerwick, and L. D. Reding. 1991. Physiological basis for inhibition of maize seed germination by flooding. Crop Science 31 :1052 1057. Mayer, A. C., B. L. Estermann, V. Stockli, and M. Kreuzer. 2005. Experimental determination of the effects of cattle stocking density and grazing period on forest regeneration on a subalpine wood pasture. Animal Research 54 :153 171. McGrath, D., S. P. Gama, A. Cardoso, O. Almeida, and J. H. Benatti. 2011. Integrating comanagement and land tenure policies for the sustainable management of the Lower Amazon Floodplain. Pages 199 136 in M. Pinedo Vasquez, M. L. Ruffino, C. Padoch, and E. Brondizio, editors. The Amazon Vrzea: The Decade Past and the Decade Ahead. Springer, New York. McGrath, D. G., O. T. Almeida, M. Crossa, A. Cardoso, and M. Cunha. 2005. Working towards community based ecosystem management of the Lower Amazon floodplain. PLEC News and Views 6 :3 10. McGrath, D. G., F. d. Castro, E. Cmara, and C. Futemma. 1999a. Community management of floodplain lakes and the sustainable development of Amazonian fisheries. in C. Padoch, J. M. Ayres, M. Pinedo Vasquez, and A. Henderson, edit ors. Vrzea: Diversity, Development, and Conservation of Amazonia's Whitewater Floodplains. New York Botanical Garden Press, Bronx. McGrath, D. G., F. d. Castro, E. Cmara, and C. Futemma. 1999b. Community management of floodplain lakes and the sustainable development of Amazonian fisheries. Pages 59 82 in C. Padoch, J. M. Ayres, M. Pinedo Vasquez, and A.
192 Henderson, editors. Vrzea: Diversity, Development, and Conservation of Amazonia's Whitewater Floodplains. New York Botanical Garden Press, Bronx. McGrath D. G., F. Decastro, C. Futemma, B. D. Deamaral, and J. Calabria. 1993. Fisheries and the evolution of resource management on the Lower Amazon Floodplain. Human Ecology 21 :167 195. Medeiros, S. V. 1968. Juta: Aspectos Sociais Agrcolas e Economicas. Secre taria de Pruduo em colaborao com o Banco do Estado do Amazonas S.A., Estado do Amazonas. Medina, J. T. 1934. The Discovery of the Amazon: According to the Account of Friar Gaspar de Carvajal and other Documents. American Geographical Society, New York. Meggers, B. J. 1971. Man and culture in a conterfeit paradise. Aldine, Chicago. Meggers, B. J. 1996. Paleoindians in the Brazilian Amazon. Science 274 :1823 1823. Megonigal, J. P., W. H. Conner, S. Kroeger, and R. R. Sharitz. 1997. Aboveground production i n Southeastern floodplain forests: A test of the subsidy stress hypothesis. Ecology 78 :370 384. Menges, E. S. and D. M. Waller. 1983. Plant strategies in relation to elevation and light in floodplain herbs. American Naturalist 122 :454 473. Mertes, L. A. K. D. L. Daniel, J. M. Melack, B. Nelson, L. A. Martinelli, and B. R. Forsberg. 1995. Spatial patterns of hydrology, geomorphology, and vegetation on the floodplain of the Amazon River in Brazil from a Remote Sensing perspective. Geomorphology 13 :215 232. M esquita, R. C. G., K. Ickes, G. Ganade, and G. B. Williamson. 2001. Alternative successional pathways in the Amazon Basin. Journal of Ecology 89 :528 537. Messina, M. G. and W. H. Conner, editors. 1998. Southern Forested Wetlands: Ecology and Management. Le wis Publishers, Boca Raton. Meyer, C. J., E. Steudle, and C. A. Peterson. 2007. Patterns and kinetics of water uptake by soybean seeds. Journal of Experimental Botany 58 :717 732. Middleton, B. 2000. Hydrochory, seed banks, and regeneration dynamics along the landscape boundaries of a forested wetland. Plant Ecology 146 :169 184. Mitsch, W. J. and K. C. Ewel. 1979. Comparative biomass and growth of cypress in Florida wetlands. Americ an Midland Naturalist 101 :417 426. Mitsch, W. J. and J. G. Gosselink. 2000. Wetlands, Third Edition. John Wiley & Sons, Inc., New York.
193 Mitsch, W. J. and W. G. Rust. 1984. Tree growth responses to flooding in a bottomland forest in Northeastern Illinois. F orest Science 30 :499 510. Moegenberg, S. M. 2002. Spatial and temporal variation in hydrochory in Amazonian floodplain forest. Biotropica 34 :606 612. Moller, H., F. Berkes, P. O. Lyver, and M. Kisliogliu. 2004. Combining science and traditional ecological knowledge: Monitoring populations for co managemenT. Ecology and Society 9 :2. Myers, J. A. and K. Kitajima. 2007. Carbohydrate storage enhances seedling shade and stress tolerance in a neotropical forest. Journal of Ecology 95 :383 395. Nebel, G., J. Dragst ed, and A. S. Vega. 2001a. Litter fall, biomass and net primary production in flood plain forests in the Peruvian Amazon. Forest Ecology and Management 150 :93 102. Nebel, G., L. P. Kvist, J. K. Vanclay, H. Christensen, L. Freitas, and J. Ruiz. 2001b. Structure and floristic composition of flood plain forests in the Peruvian Amazon I. Overstorey. Forest Ecology and Management 150 :27 57. Nebel, G., L. P. Kvist, J. K. V anclay, and H. Vidaurre. 2001c. Forest dynamics in flood plain forests in the Peruvian Amazon: effects of disturbance and implications for management. Forest Ecology and Management 150 :79 92. Nelson, B. W., R. Mesquita, J. L. G. Pereira, S. G. A. de Souza, G. T. Batista, and L. B. Couto. 1999. Allometric regressions for improved estimate of secondary forest biomass in the central Amazon. Forest Ecology And Management 117 :149 167. Nepstad, D., I. Tohver, D. Ray, P. Moutinho, and G. Cardinot. 2007. Mortality of large trees and lianas following experimental drought in an Amazon forest. Ecology 88 :2259 2269. Niinemets, U. 2010. Responses of forest trees to single and multiple environmental stresses from seedlings to mature plants: Past stress history, stress int eractions, tolerance and acclimation. Forest Ecology and Management 260 :1623 1639. Nogueira, E. M., B. W. Nelson, and P. A. Fearnside. 2006. Volume and biomass of trees in central Amazonia: influence of irregularly shaped and hollow trunks. Forest Ecology And Management 227 :14 21. Odum, E. P., J. T. Finn, and E. H. Franz. 1979. Perturbation theory and the subsidy stress gradient. Bioscience 29 :349 352. Oesterheld, M. and S. J. McNaughton. 1991. Interactive effect of flooding and grazing on the growth of Ser engeti grasses. Oecologia 88 :153 156.
194 Orton, J. M. A. 1870. The Andes and the Amazon; or, Across the Continent of South America. Harper & Brothers, Pulbishers, New York. Padoch, C., J. M. Ayres, M. Pinedo Vasquez, and A. Henderson. 1999. Vrzea: Diversity, Development, and Conservation of Amazonia's Whitewater Floodplains. New York Botanical Garden Press, New York. Padoch, C. and A. Steward. 2011. The Vrzea : The Decade Past. in M. Pinedo Vasquez, M. L. Ruffino, C. Padoch, and E. S. Brondzio, editors. The A mazon Vrzea: The Decade Past and the Decade Ahead. Springer, New York. Palmatary, H. C. 1965. The River of the Amazons: Its Discovery and Early Exploration 1500 1743. Carlton Press, New York. Parolin, P. 2001a. Morphological and physiological adjustments to waterlogging and drought in seedlings of Amazonian floodplain trees. Oecologia 128 :326 335. Parolin, P. 2001b. Seed germination and early establishment of 12 tree species from nutrient rich and nutrient poor Central Amazonian floodplains. Aquatic Botany 70 :89 103. Parolin, P. 2002. Submergence tolerance vs. escape from submergence: two strategies of seedling establishment in Amazonian floodplains. Environmental and Experimental Botany 48 :177 186. Parolin, P. 2009. Submerged in darkness: adaptations to pr olonged submergence by woody species of the Amazonian floodplains. Annals of Botany 103 :359 376. Parolin, P., O. De Simone, K. Haase, D. Waldhoff, S. Rottenberger, U. Kuhn, J. Kesselmeier, B. Kleiss, W. Schmidt, M. T. F. Piedade, and W. J. Junk. 2004a. Cen tral Amazonian floodplain forests: Tree adaptations in a pulsing system. Botanical Review 70 :357 380. Parolin, P., L. V. Ferreira, A. Albernaz, and S. S. Almeida. 2004b. Tree species distribution in varzea forests of Brazilian Amazonia. Folia Geobotanica 3 9 :371 383. Parolin, P., C. Lucas, M. T. F. Piedade, and F. Wittmann. 2010. Drought responses of flood tolerant trees in Amazonian floodplains. Annals of Botany 105 :129 139. Parolin, P., C. M. Lucas, F. Wittmann, and M. T. F. Piedade. 2009. Drought response s of flood tolerant trees in Amazonian floodplains. Annals of Botany:doi:10.1093/aob/mcp1258, available online at www.aob.oxfordjournals.org Parolin, P., A. C. Oliveira, M. T. F. Piedade, F. Wittmann, and W. J. Junk. 2002. Pioneer trees in Amazonian floodplains: Three key species form monospecific stands in different habitats. Folia Geobotanica 37 :225 238.
195 Paz, H. 2003. Root/shoot allocation an d root architecture in seedlings: Variation among forest sites, microhabitats, and ecological groups. Biotropica 35 :318 332. Pearson, T. R. H., D. Burslem, C. E. Mullins, and J. W. Dalling. 2002. Germination ecology of neotropical pioneers: Interacting eff ects of environmental conditions and seed size. Ecology 83 :2798 2807. Penna, D. S. F. 1869. A Regio Occidental da Provincia do Par: Resenhas estatisticas das comarcas de Obidos e Santarem. Diario de Belm, Par. Pereira, V. F. G., R. G. Congalton, and D. J. Zarin. 2002. Spatial and temporal analysis of a tidal floodplain landscape Arnapi, Brazil Using geographic information systems and remote sensing. Photogrammetric Engineering and Remote Sensing 68 :463 472. Peto, R. and J. Peto. 1972. Asymptotically efficient rank invariant procedures. Journal of the Royal Statistical Society, Series A 135 :185 207. Pettit, N. E., R. H. Froend, and P. G. Ladd. 1995. Grazing in remnant woodland vegetation Changes in species composition and life form groups. Journal o f Vegetation Science 6 :121 130. Phillips, D. L., S. L. Brown, P. E. Schroeder, and R. A. Birdsey. 2000. Toward error analysis of large scale forest carbon budgets. Global Ecology and Biogeography 9 :305 313. Phillips, O. L., L. Aragao, S. L. Lewis, J. B. Fi sher, et al 2009. Drought sensitivity of the Amazon rainforest. Science 323 :1344 1347. Phillips, O. L., Y. Malhi, N. Higuchi, W. F. Laurance, P. V. Nunez, R. M. Vasquez, S. G. Laurance, L. V. Ferreira, M. Stern, S. Brown, and J. Grace. 1998. Changes in th e carbon balance of tropical forests: Evidence from long term plots. Science 282 :439 442. Pierce, A. R. and S. L. King. 2007. The effects of flooding aid sedimentation on seed germination of two bottomland hardwood tree species. Wetlands 27 :588 594. Pinedo Vasquez, M. 1999. Changes in soil formation and vegetation on silt bars and backslopes of levees following intensive production of rice and jute. Pages 301 311 in C. Padoch, J. M. Ayres, M. Pinedo Vasquez, and A. Henderson, editors. Varzea: Diversity, Dev elopment, and Conservation of Amazonia's Whitewater Floodplains. New York Botanical Garden Press, New York. Pinheiro, J., D. Bates, S. DebRoy, D. Sarkar, and R Core T eam. 2008. nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1 90. Pire s, J. M. and H. M. Koury. 1959. Estudo de um trecho de mata de vrzea prximo a Belm. Boletn Goeldi 36 :3 44.
196 Poorter, L. 2007. Are species adapted to their regeneration niche, adult niche, or both? American Naturalist 169 :433 442. Prance, G. T. 1979. Not es on the vegetation of Amazonia III. The terminology of Amazonian forest types subject to inundation. Brittonia 31 Pyke, D. A. and J. N. Thompson. 1986. Statistical analysis of survival and removal rate experiments. Ecology 67 :240 245. R Core Development Team 2006. R: A language and environment for statistical computing. R Foundation for Statistical Computing,Vienna, Austria. ISBN 3 900051 07 0, URL http://www.R project.org R Core Development Team 2009. R: A langu age and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Ramirez Marcial, N. 2003. Survival and growth of tree seedlings in anthropogenically disturbed Mexican montane rain forests. Journal of Vegetation Scien ce 14 :881 890. Rhind, S. M. 2009. Anthropogenic pollutants: a threat to ecosystem sustainability? Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364 :3391 3401. Rice, A. H., E. H. Pyle, S. R. Saleska, L. Hutyra, M. Palace, M. Keller, P. B. de Camargo, K. Portilho, D. F. Marques, and S. C. Wofsy. 2004. Carbon balance and vegetation dynamics in an old growth Amazonian forest. Ecological Applications 14 :S55 S71. Richey, J. E., J. M. Melack, A. K. Aufdenkampe, V. M. Ball ester, and L. L. Hess. 2002. Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO2. Nature 416 :617 620. Robertson, A. I. and R. W. Rowling. 2000. Effects of livestock on riparian zone vegetation in an Australian drylan d river. Regulated Rivers Research & Management 16 :527 541. Rodriguez, M. D., A. Orozco Segovia, M. E. Sanchez Coronado, and C. Vazquez Yanes. 2000. Seed germination of six mature neotropical rain forest species in response to dehydration. Tree Physiology 20 :693 699. Roosevelt, A. C. 1992. Secrets of the forest: an archaeologist reappraises the past and future of Amazonia. The Sciences 32 :22 28. Roosevelt, A. C., R. A. Housley, M. I. Dasilveira, S. Maranca, and R. Johnson. 1991. 8th Millennium pottery from a prehistoric shell midden in the Brazilian Amazon. Science 254 :1621 1624.
197 Saatchi, S. S., R. A. Houghton, R. Alvala, J. V. Soares, and Y. Yu. 2007. Distribution of aboveground live biomass in the Amazon basin. Global Change Biology 13 :816 837. Sack, L. and P. J. Grubb. 2001. Why do species of woody seedlings change rank in relative growth rate between low and high irradiance? Functional Ecology 15 :145 154. Saldarriaga, J. G., D. C. West, M. L. Tharp, and C. Uhl. 1988. Long term chronosequence of forest succession in the Upper Rio Negro of Colombia and Venezuela. Journal of Ecology 76 :938 958. Salo, J., R. Kalliola, I. Hakkinen, Y. Makinen, P. Nie mela, M. Puhakka, and P. D. Coley. 1986. River dynamics and the diversity of Amazon lowland forest. Nature 322 :254 258. Sautu, A., J. M. Baskin, C. C. Baskin, and R. Condit. 2006. Studies on the seed biology of 100 native species of trees in a seasonal moist tropical forest, Panama, Central America. Forest Ecology and Management 234 :245 263. Scarano, F. R., T. S. Pereira, a nd 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. Scarano, F. R., K. T. Ribeiro, L. F. D. De Moraes, and H. C. De Lima. 1997. Plant esta blishment on flooded and unflooded patches of a freshwater swamp forest in southeastern Brazil. Journal of Tropical Ecology 13 :793 803. Scariot, A. 2000. Seedling mortality by litterfall in Amazonian forest fragments. Biotropica 32 :662 669. Scholes, R. J. and S. R. Archer. 1997. Tree grass interactions in savannas. Annual Review of Ecology and Systematics 28 :517 544. Schngart, J. 2003. Dendrochronologische Untersuchungen in berschwemmungswlden der vrzea Zentralamazoniens. Georg August Universitt, Gtti ngen. Schongart, J. and W. J. Junk. 2007. Forecasting the flood pulse in Central Amazonia by ENSO indices. Journal Of Hydrology 335 :124 132. Schongart, J., W. J. Junk, M. T. F. Piedade, J. M. Ayres, A. Huttermann, and M. Worbes. 2004. Teleconnection betwee n tree growth in the Amazonian floodplains and the El Nino Southern Oscillation effect. Global Change Biology 10 :683 692. Schngart, J., F. Wittmanss, and M. Worbes. 2010. Biomass and net primary production of Central Amazonian floodplain forests. Pages 34 7 388 in W. J. Junk, M. T. F. Piedade, F. Wittmann, J. Schngart, and P. Parolin, editors. Amazonian Floodplain
198 Forests: Ecophysiology, Biodiversity, and Sustainable Management. Springer, New York. Schngart, J., F. Wittmanss, and M. Worbes. in press. Biom ass and Net Primary Production of Central Amazonian Floodplain Forests. Sears, R. R. 2003. New Forestry on the Floodplain: The Ecology and Management of Calycophyllum spruceanum (Rubiaceae) on the Amazon Landscape. Ph.D. Columbia University, New York. Shei kh, P. A. 2002. The impacts of water buffalo and cattle ranching on the Lower Amazon floodplain: An ecological and socio economic comparison. Ph.D. thesis. The Pennsylvania State University. Sheikh, P. A., F. D. Merry, and D. G. McGrath. 2006. Water buffal o and cattle ranching in the Lower Amazon Basin: Comparisons and conflicts. Agricultural Systems 87 :313 330. Silva, J. A. M. d., M. Pereira, and M. I. de Oliveira Pereira. 2003. Fruits and seeds consumed by tambaqui ( Colossoma macropomum Cuvier 1818) inco rporated in the diets: gastrointestinal tract digestibility and transit velocity. Revista Brasileira De Zootecnia Brazilian Journal of Animal Science 32 :1815 1824. Sioli, H. 1984. The Amazon and its main affluents: hydrography, morphology of the river cour ses, and river types. Pages 127 165 in H. Sioli, editor. The Amazon Limnology and landscape ecology of a mighty tropical river and its basin. W. Junk, Dordrecht. Slot, M. and L. Poorter. 2007. Diversity of tropical tree seedling responses to drought. Bio tropica 39 :683 690. Smith, H. H. 1879. Brazil: the Amazons and the Coast. Charles Scribner's Sons, New York. Smith, N. J. H. 1999. The Amazon River Forest: A Natural History of Plants, Animals, and People. Oxford University Press, New York. Smith, N. J. H. 2002. Amazon Sweet Sea. University of Texas Press, Austin. Sombroek, W. 2001. Spatial and temporal patterns of Amazon rainfall Consequences for the planning of agricultural occupation and the protection of primary forests. Ambio 30 :388 396. Spix, J. B. v. and C. F. P. v. Martius. 1823/1968. Viagem pelo Brasil, 1817 1820: Excertos e Ilustraes. Edies Melhoramentos, So Paulo. Spruce, R. 1908/1970. Notes of a Botanist on the Amazon and the Andes, Volume 1. Johnson Reprint Corporation, New York.
199 Steege, H. t., N. Pitman, D. Sabatier, H. Castellanos, et al 2003. A spatial model of tree alpha diversity and tree density for the Amazon. Biodiversity and Conservation 12 :2255 2277. Steininger, M. K. 2000. Secondary forest structure and biomass following short and extended land use in central and southern Amazonia. Journal of Tropical Ecology 16 :689 708. Stella, J. C., J. J. Battles, B. K. Orr, and J. R. McBride. 2006. Synchrony of se ed dispersal, hydrology and local climate in a semi arid river reach in California. Ecosystems 9 :1200 1214. Sternberg, H. O. R. 1975. The Amazon River of Brazil. Steiner, Wiesbaden. Stevens, P. F. 2001 onwards. Angiosperm Phylogeny Website. Version 9, June 2008 [and more or less continuously updated since]. http://www.mobot.org/MOBOT/research/APweb/ Terborgh, J. and E. Andresen. 1998. The composition of Amazonian forests: patterns at local and region al scales. Journal of Tropical Ecology 14 :645 664. Therneau, T. M. 1999. A package for survival in R. Tockner, K., M. Pusch, D. Borchardt, and M. S. Lorang. 2010. Multiple stressors in coupled river floodplain ecosystems. Freshwater Biology 55 :135 151. Toc kner, K. and J. A. Stanford. 2002. Riverine flood plains: present state and future trends. Environmental Conservation 29 :308 330. Traveset, A., J. Rodriguez Perez, and B. Pias. 2008. Seed trait changes in dispersers' guts and consequences for germination a nd seedling growth. Ecology 89 :95 106. Tropicos.org. 2009. Missouri Botanical Garden. 28 Jul 2009 < http://www.tropicos.org/Name/7800401> Valencia, R., R. Condit, H. C. Muller Landau, C. Hernandez, and H. Navarrete. 2009. Dissecting biomass dynamics in a large Amazonian forest plot. Journal of Tropical Ecology 25 :473 482. van der Valk, A. G. and C. B. Davis. 1978. The role of seed banks in the vegetation dynamics of prarie glacial marshes. Ecology 59 :322 335. Vazquez Yanes, C. and A. Orozco Segovia. 1990. Ecological significance of light controlled seed germination in 2 contrasting tropical habitats. Oecologia 83 :171 175.
200 Veldman, J. W., B. Mostacedo, M. Pena Claros, and F. E. Putz. 2009. Selective loggin g and fire as drivers of alien grass invasion in a Bolivian tropical dry forest. Forest Ecology and Management 258 :1643 1649. Vozzo, J. A. 2002. Tropical Tree Seed Manual. U.S. Department of Agriculture, Forest Service, Washington DC. Waldoff, D. and L. M. A. Maia. 2001. Production and chemical composition of fruit. Pages 392 415 in W. Junk, editor. The Central Amazon Floodplain: Actual use and options for a Sustainable Management. Backhuys Publishers, Leiden. Walker, L. R., J. C. Zasada, and F. S. Chapin. 1986. The role of life history processes in primary succession on an Alaskan floodplain. Ecology 67 :1243 1253. Wallace, A. R. 1853/1969. A Narrative of Travels on the Amazon and Rio Negro, with an account of t he native tribes, and observations on the climate, geology, and natural history of the Amazon Valley. Haskell House Publishers Ltd., New York. Ward, J. V. and J. A. Wiens. 2001. Ecotones of riverine ecosystems: Role and typology, spatio temporal dynamics, and river regulation. Ecohydrology and Hydrobiology 1 :25 36. Waring, R. H. 1987. Characteristics of trees predisposed to die. Bioscience 37 :569 574. Weisberg, P. J., F. Bonavia, and H. Bugmann. 2005. Modeling the interacting effects of browsing and shading on mountain forest tree regeneration ( Picea abies ). Ecological Modelling 185 :213 230. White, P. S. and S. T. A. Pickett. 1985. The Ecology of Natural Disturbance and Patch Dynamics. Academic Press, Inc., Orlando. WinklerPrins, A. 2002. Seasonal floodplain upland migration along the lower Amazon River. Geographical Review 92 :415 431. Winklerprins, A. 2006. Jute cultivation in the lower Amazon, 1940 1990: an ethnographic account from Santarem, Para, Brazil. Journal of Historical Geography 32 :818 838. Winkler Prins, A. M. G. A. 1999. Between the Floods: Soils and agriculture on the Lower Amazon Floodplain, Brazil. Ph.D. University of Wisconsin, Madison. Wittmann, A. D., M. T. F. Piedade, P. Parolin, and F. Wittmann. 2007. Germination in four low varzea tree spe cies of Central Amazonia. Aquatic Botany 86 :197 203. Wittmann, F. 2006. Database of the MAUA Project, Manaus. Wittmann, F., D. Anhuf, and W. Junk. 2002. Tree species distribution and community structure of central Amazonian vrzea forests by remote sensing techniques. Journal of Tropical Ecology 18 :805 820.
201 Wittmann, F. and W. J. Junk. 2003. Sapling communities in Amazonian white water forests. Journal of Biogeography 30 :1533 1544. Wittmann, F., W. J. Junk, and M. T. F. Piedade. 2004. The varzea forests in Amazonia: flooding and the highly dynamic geomorphology interact with natural forest succession. Forest Ecology and Management 196 :199 212. Wittmann, F., J. Schoengart, J. M. Brito A. Oliveira Wittmann, P. Parolin, M. T. F. Piedade, W. J. Junk, and J. L. Guillaumet. 2011. Manual of tree species in central Amazonian white water floodplains: Taxonomy, ecology, and use. Instituto Nacional de Pesquisas da Amazonia, Manaus. Wittmann, F. J. Schongart, J. C. Montero, T. Motzer, W. J. Junk, M. T. F. Piedade, H. L. Queiroz, and M. Worbes. 2006. Tree species composition and diversity gradients in white water forests across the Amazon Basin. Journal of Biogeography 33 :1334 1347. Worbes, M. 19 97. The forest ecosystem of the floodplains. Pages 223 266 in W. J. Junk, editor. The Central Amazon Floodplain: Ecology of a Pulsing System. Springer, Berlin. Worbes, M. and W. J. Junk. 1999. How old are tropical trees? The persistence of a myth. Iawa Jou rnal 20 :255 260. Worbes, M., H. Klinge, J. D. Revilla, and C. Martius. 1992. On the dynamics, floristic subdivision and geographical distribution of vrzea forests in Central Amazonia. Journal of Vegetation Science 3 :553 564. Wright, S. J. 2002. Plant dive rsity in tropical forests: a review of mechanisms of species coexistence. Oecologia 130 :1 14. Zarin, D. J. 2004. Working forests: Concepts & realities. in D. J. Zarin, J. R. R. Alavalapati, F. E. Putz, and M. Schmink, editors. Working Forests in the America n Tropics: Conservation through Sustainable Management? Columbia University Press, New York. Zarin, D. J., A. L. Duchesne, and M. Hiraoka. 1998. Shifting cultivation on the tidal floodplains of Amazonia: impacts on soil nutrient status. Agroforestry System s 41 :307 311. Zarin, D. J., V. F. G. Pereira, H. Raffles, F. G. Rabelo, M. Pinedo Vasquez, and R. G. Congalton. 2001. Landscape change in tidal floodplains near the mouth of the Amazon River. Forest Ecology and Management 154 :383 393. Zuur, A. F., I. E. N., N. J. Walker, A. A. Saveliev, and G. M. Smith. 2009. Mixed Effects Models and Extensions in Ecology with R. Springer, New York.
202 BIOGRAPHICAL SKETCH Christine Lucas was born in Salem, Massachusetts in 1978 in a typical New England snow storm. She grew up on the North Shore of Massachusetts in Beverly as the eldest of two sisters and moved inland to Framingham, MA, where she attended Framingham High School. Christine received her BA in Biology with a minor in Anthropology in 2001 from Vassar Colle ge in the Hudson Valley of NY. She then moved to New York City to work at the New York Botanical Garden, an experience which introduced her to colleagues in Brazil. In 2003 she was awarded the Clark Fellowship from Vassar College to conduct research on f ruit eating fish and floodplain forest stand dynamics in collaboration with IPAM in Santarm, Brazil. Her interaction with floodplain residents and the vrzea ecosystem provided the inspiration for her to return as a Ph.D. student. Her experience as a me ntor for biology students in Brazil has inspired her to seek out teaching as a profession. Christine joined the Wildlife Ecology and the Working Forests of the Tropics P rogram Christine received her Ph.D. from the University o f Florida in the summer of 2011. Christine currently lives in Montevideo, Uruguay and plans to seek new research projects concerning the ecology, conservation, and management of wetland forests wo rld wide, while continuing to build capacity among young ecologists in the U.S. and Latin America. In her spare time she enjoys doing botanical illustration and dancing.