THE“ m 1293 10070 6153‘,“ L 1" ' ' u....-. _.v‘ ~‘-:_‘ 2.2“ ‘1‘; 3 it of LIU .0. 41.: 3‘3. 1 Michigan State I This is to certify that the thesis entitled STUDIES OF FOREST, AGRICULTURAL, AND SUCCESSIONAL ENVIRONMENTS IN THE UPPER RIO NEGRO REGION OF THE AMAZON BASIN presented by Christopher Uhl has been accepted towards fulfillment of the requirements for Ph- 0- degree in) Botanx Major professor C Date QXDOASL in} \0‘80 0-7639 '- V 2.1“ ‘ I ‘ ,: id“ I. 7. m l, .‘ « ~~"!.',£,'!’, ~ ~ . “ 1 3.55.1; OVERDUE FINES: \ 25¢ per thy per item RETUMING LIBRARY MTERIAL§: Place in book return to hair charge from circulation "co: STUDIES OF FOREST, AGRICULTURAL, AND SUCCESSIONAL ENVIRONMENTS IN THE UPPER RIO NEGRO REGION OF THE AMAZON BASIN By Christopher Uhl A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Botany and Plant Pathology 1980 Christopher Uhl ABSTRACT STUDIES OF FOREST, AGRICULTURAL, AND SUCCESSIONAL ENVIRONMENTS IN THE UPPER RIO NEGRO REGION OF THE AMAZON BASIN By Christopher Uhl Investigations of forest, agricultural, and successional environ- ments were conducted near the town of San Carlos de Rio Negro in the Amazon Basin of Venezuela (lat 1.56 N., long 67.03 W). The forest environment was characterized through studies of species composition, vegetation structure, and regeneration. Three mature tierra firme (i.e. never flooded) forest sites, ranging in size from 0.5 to one hectare, were surveyed. When compared to most other wet tropical tierra firme forests, the San Carlos forest sites had higher densities of trees in the smaller size classes and a lower basal area per hectare. There was an average of 75 tree species per site. Most species were rare, having importance values of less than one. The few relative- ly common species were well represented in all diameter size classes indicating that the sites studied were floristically stable. Gaps were common in the forest and their occurrence is probably beneficial to a persistent sapling and seedling pool. An analysis of the energetics of the local farming system [i.e. slash and burn cultivation of yuca (Manihot esculenta, Crantz.)] and a comparison of productivity between farm crops and secondary vegetation was conducted. Yuca root production was 1.46 metric tons dry weight per hectare in the first year of farm site use but declined to 1.0 metric Christopher Uhl tons per hectare in the second year. Local farmers maintain a ratio of energy output to energy input of 13.9 to one, (measured as the caloric content of the farm site yield vs. calories expended in farming activi- ties). The productivity of secondary vegetation in a cut and burn plot was 109 grams dry weight per m2 2 in the first year of succession and 1446 grams dry weight per m in the second year. Changes ‘hi vegetation structure, species composition, and species dominance were studied over the first 22 months of succession after the cutting and burning of several mature tierra firme forest sites. Only 0.58 individuals per m2 were present four months after burning. The density of secondary woody plants, forbs, and grasses increased sharply from the fourth to the tenth month of succession. There was less change in plant density between the tenth and twenty-second month, but a large change in vegetation height occurred during this time. By 22 months a loose canopy of Cecropia spp. was present at 5 m height. To explain the observed succession, additional field investigations were undertaken. Many of the forest tree species had the ability to sprout when cut. However, burning killed sprouts present on the study sites. Burning also significantly reduced the size of the seed pool. Nevertheless, the first secondary woody colonizers established from seed pool seeds which survived the burn. The first forb and grass colonizers established from seeds dispersed onto the site after the disturbance. The common secondary woody species had bat or bird dispersed seeds. Many forbs and grasses had wind dispersed seeds. Population densities were low during the first months of succession because of the low num- bers of propagules present and because low rainfall during this time was Christopher Uhl not favorable for seedling germination or establishment. There was a rapid increase in population size after about the fourth month of succes- sion because the early colonizers had begun to produce, and disperse locally, large numbers of seeds by that time. There were six distinct microhabitat types present on the post-burn sites. Most species tested showed distinct microhabitat preferences. This dissertation is dedicated to Nan ii ACKNOWLEDGEMENTS I am very thankful to Dr. Peter G. Murphy, my major professor, for his guidance during this study. His confidence in me and interest in my research made my graduate education both enjoyable and productive. In hindsight, I am particularly appreciative of the freedom he gave me in developing this project. I am grateful for the help in the conceptual- ization of my research goals provided by my guidance committee, (both past and present), Drs. F. H. Tschirley, J. H. Fitch, P. A. Werner and G. Schneider. This study would not have been possible without assistance provided through a grant from the National Science Foundation authored by Drs. F. Golley and C. Jordan. Special thanks are extended to Dr. Jordan for his interest in my work and his assistance on numerous technical prob- lems. Working with him has had a profound impact on my development as an ecologist. I am grateful to Dr. Howard Clark and Kathleen Clark for the many hours they have devoted to gathering field data for my ongoing studies when I could not be present in San Carlos. When I was present, their willingness to critique my working hypotheses and their general good humor made my stay both stimulating and pleasant. I am also thankful to Dr. Jerry Hall and to Jan Hall who always received me warmly while they were the resident ecologists in San Carlos. Thanks are likewise extended to Juan Moreno, Getulio Gomez, Pablo Livino, Jose Maquirino and Manuel Silva, the men who worked in the field with me at various times. Their patience with my faltering Spanish and their willingness to help in all tasks, no matter how tedious or bizarre, allowed me to accomplish far more than I would have alone. Special thanks gm) to Pedro Maquirino, whose overall-field competence, intelli- gence, and dependability made his help particularly valuable. I am grateful to Drs. Ernesto Medina and Rafael Herrera of the Instituto Venezolano de Investigaciones Cientificas (IVIC) for the kind reception and frequent assistance they gave me on my visits to their institute. I also wish to thank Gladys Escalante and her co-workers at IVIC for their technical assistance in processing samples. Florencia Montagnini, also at IVIC, helped with many business matters each time I was in Caracas. I wish to extend my appreciation to Dr. Jorge Rabinovitch and Magaly Macedo for programming portions of my data for computer analysis and to the Ecology Center at IVIC for the use of their computer. I would like to thank Joan Yantko at the University of Georgia for her cheery assistance on numerous tedious logistical problems. I am grateful to Dr. Hans Klinge for the frequent encouragement and suggestions he provided when we were in the field together. Thanks also go to Drs. Eberhardt Brunig and Jochen Heuveldop for their many helpful comments regarding the study of forest structure. I would like to acknowledge the assistance Ronald Liesner of the Missouri Botanical Garden has given in providing species determinations for many successional plants. Thanks are extended to Gerry Donnelly, iv Ron Gross, Beth Hutchison, Lyn Loveless and Art Tai who read parts of this thesis and provided many helpful suggestions. I wish to extend thanks to Hanna Priwer for her help with the figures in Part I of this thesis and to William Tregea, Saundra Green, and Gladys Russell for the typing of this thesis. Finally, I would like to thank my wife, Nan. She spent countless hours helping, often creatively, in all phases of this research. Perhaps more importantly, she was patient with me during periods when I was hopelessly preoccupied. TABLE OF CONTENTS Page LIST OF TABLES ........................... ix LIST OF FIGURES .......................... xii GENERAL INTRODUCTION ........................ 1 Organization ......................... 1 Field Site Location and Climate ................ 1 Geology and Soils ....................... 4 Research Goals ........................ 5 LIST OF REFERENCES ......................... 8 PART 1. FOREST COMPOSITION, STRUCTURE, AND REGENERATION INTRODUCTION ............................ 10 METHODS .............................. 15 RESULTS AND DISCUSSION ....................... 18 Composition and Structure of Mature Forest .......... 18 The diversity of tree species .............. 18 Importance values for tree species ............ 21 Density and basal area .................. 26 Density and basal area as a function of A horizon depth ........................ 29 Vertical Structure .................... 33 A summary index of forest composition and structure . . . 39 The successional status of the forest under study . . . . 46 Regeneration within Mature Forest ............... 52 Seedling dynamics .................... 52 Gap and building phases ................. 55 Size class distribution of important tree species . . . . 58 SUMMARY .............................. 60 LIST OF REFERENCES ......................... 62 APPENDIX TABLES .......................... 66 vi PART II. PRODUCTIVITY OF AGRICULTURAL AND SUCCESSIONAL SITES INTRODUCTION ............................ 69 METHODS .............................. 72 Time and Energy Expenditure in Yuca Crop Production ...... 72 Conuco Productivity ..................... 73 Yuca production ..................... 74 Pineapple production ................... 76 Plantain production ................... 77 Cashew production .................... 77 Weed production ..................... 78 Secondary Site Productivity ................. 78 RESULTS AND DISCUSSION ....................... 82 Time and Energy Expenditure in Yuca Crop Production ..... 82 Conuco Productivity ..................... 97 Yuca production ..................... 97 Pineapple production ................... 101 Plantain production .......... ' ......... 102 Cashew production .................... 102 Weed production ..................... 103 Total production ..................... 103 Succession Site Productivity ................. 106 Prospects for Increased Yields ................ 116 SUMMARY .............................. 120 LIST OF REFERENCES ......................... 122 APPENDIX TABLES .......................... 126 PART III. EARLY SUCCESSION STUDIES INTRODUCTION ............................ 130 METHODS .............................. 133 Vegetation Development .................... 133 Vegetation Development Explained ............... 134 Sprouting of cut trees .................. 134 Effects of burning on sprouts and seed pools ....... 134 Seed source of the first colonizers after burning '. . . . 135 Dispersal strategies of secondary species ........ 136 Low population densities during the first months of succession ..................... 136 Rapid population increase after the fourth month of succession ..................... 137 Microhabitat preferences of secondary species ...... 137 vii RESULTS AND DISCUSSION ....................... 140 Vegetation Development .................... 140 Vegetation structure ................... 140 Density ......................... 140 Species richness ..................... 145 Species dominance .................... 148 Questions Posed By the Results of the Vegetation Development Study .................... 153 Vegetation Development Explained ............... 154 Sprouting of cut trees .................. 154 Effects of burning on sprouts and seed pools ....... 156 Seed source of the first colonizers after burning . . . . 160 Dispersal strategies of secondary species ........ 162 Low population densities during the first months of succession ..................... 163 Rapid population increases after the fourth month of succession ................ 166 Microhabitat preferences of secondary species ...... 167 SUMMARY .............................. 188 Vegetation Development .................... 188 Vegetation Development Explained ............... 189 LIST OF REFERENCES ......................... 191 APPENDIX TABLES .......................... 194 EPILOGUE .............................. 198 viii Table A1. A2. LIST OF TABLES PART I Forest tree density by size classes for lowland tierra firme sites at San Carlos and elsewhere .......... Complexity index and the component values thereof for the San Carlos forest and other tierra firme sites ....... Structural characteristics of a 60 year old abandoned farm site and a mature forest site (i.e. Site 1) ...... The height distribution of seedlings based on the survey of 52 one m plots in Site 1 ................ Summary of the changes in seedling numbers over a three year period in 10 permanent I m plots located in Site 1 The percentage of total species present in a geometric series of abundance classes for San Carlos, Site 1, and other lowland tropical tierra firme sites .......... Diversity index values for trees 3 10 cm dbh in Sites 1, 2, 3, and 4 ....................... PART II The time spent, the relative difficulty, and the estimated energy expenditure in conuco field activities during a two crop cycle of conuco-use .................. The time spent, the relative difficulty, and the energy expended in processing 7855 kilograms of yuca roots (i.e. 7855 kilograms is the estimated yuca yield per ha over a two crop cycle of conuco-use) ................ A summary of human energy inputs and yuca crop energy outputs for a two crop cycle of conuco use ......... A summary of crop productivity (grams dry weight per m2 per year) in Conuco Site 1 for years one and two ........ Values for the measurements taken in the estimation of yuca leaf herbivory by chewing insects ........... ix Page 30 45 50 53 54 66 67 83 88 93 98 99 Table 6. 10. 11. 12. A1. A2. A3. Page A comparison of yuca root production for the first crop in three San Carlos conucos. Values are based on the survey of 30 1 x 1 m plots in each site ......... . . . 100 The mean dry weight biomass of weedy species harvested from 27 1 x 1.5 m plots for the period 0-10 months, 11-16 months, and 17-21 months ............ . . . . 104 A sum ary of the standing crop biomass (grams dry weight per m ) for Succession Site 1 at one and two years following the burn ...................... 108 The standing crop biomass of one and two year old succession sites at San Carlos de Rio Negro and elsewhere in the lowland tropics ............... 109 A summary of production for years one and two in Succession Site 1 ...................... 112 A comparison of year one and year two production between Succession Site 1 and Conuco Site 1 .............. 114 A comparison of leaf cover and leaf production in Succession Site 1 and Conuco Site 1 over the two year study period ......................... 115 Human labor input and yuca output for crops one and two of a two crop conuco cycle ................ 126 Regression equations and coefficients of determination used for the estimation of standing crop biomass for crop plants and for Cecropia spp. ................ . 127 Summary of the data used in estimating chewing insect herbivory losses from the leaf litter fraction of leaf production in Succession Site 1 ................ 128 PART III The number of species present in a series of abundance classes in five plant groupings. The abundance classes refer to the number of individuals, of a given species, present in the 27 Succession Eite 1 vegetation survey plots (i.e. a total of 40.5 m ). Values are for surveys conducted at 10 and 22 months after the burning of this site. Only individuals more than 5 cm tall were considered ........................... 151 Theznumber of sprouts and seedlings present in 19 1 m plots in Succession Site 3 three months after the mature forest occupying this site was cut ........... 155 Table Page 3. A comparison between the post-burn seed pool in Succession Site 3 and the seed pool of the tierra firme forest adjacent to this site .............. 159 4. The manner of distribution of root mat, charcoal, bare soil, and slash at 10 months after burning in Succession Site 1. Values are based on the survey of 27 1 x 1.5 m plots ............................. 169 5. Factors related to germination and establishment in seven common secondary species occurring after cut and burn disturbances ..................... 185 Al. The number of individuals per m2 in five plant groupings at 4, 10, 16, and 22 months following the burning of Succession Site 1. Only individuals more than 5 cm tall were considered ..................... 194 A2. The number of individuals per m2 in five plant groupings at 4, 10, 16, and 22 months following the burning of Site 2. Only individuals more than 5 cm tall were considered .......................... 195 A3. The change in species number in six plant groupings over the first 22 months of succession after a burning disturbance. Results are based on the survey of 27 1 x 1.5 m permanent plots located in Succession Site 1. Values in parentheses represent the number of species in each life form not encountered in previous surveys. Only individuals more than 5 cm tall were considered ..... 196 A4. A physical characterization of the six principal micro- habitat types present after the cutting and burning of tierra firme sites in the San Carlos region ......... 197 xi LIST OF FIGURES Figure 1. GENERAL INTRODUCTION A map of Venezuela (bottom inset) with an enlargement of the San Carlos de Rio Negro region. The top inset shows the locations of the study sites relative to each other and to the town of San Carlos. Forest sites are abbreviated FS, succession sites SS, and conuco (i.e. farm) sites CS ........................... PART I An altimetric transect between the Rio Negro and the Casiquiare river showing maximum river level for 1976. The inset at the right upper corner shows a tracing of the microrelief in the lowland area (Figure used with the permission of Herrera) ............... The relationship between cumulative number of species and area for three 0.5 ha tierra firme sites several kilometers apart ...................... The relationship between species nugber and log area for Site 1. Area is expressed in m ............ The number of species in Site 1 occurring in a series importance value classes. Only trees 10 cm or more in dbh were considered ................... The percentage of the total number of Site 1 trees, 1 cm or more in dbh, occurring in a series of five cm diameter classes ........................... A profile diagram of the tierra firme forest, San Carlos de Rio Negro. The diagram represents a strip of forest 50 m long and 7.6 m wide. Only trees over 7.6 m tall are depicted ........................ An idealized profile diagram of the San Carlos de Rio Negro tierra firme forest based on 10 0.1 ha plots ..... The percentage of open or non-canopy space at 1 m vertical intervals in a 7.6 x 50 m strip of forest in Site 1 ..... xii Page 13 28 41 Figure Page 9. 10. 11. The distribution of mid-crown heights for trees 10 cm or more in dbh in Site 1. Mid-crown height = tree height- (crown depth/2) ........................ 43 A profile diagram of a 60 year old second growth forest, San Carlos de Rio Negro. The diagram represents a strip of forest 50 m long by 7.6 m wide. Only trees over 7.6 m high are depicted .................... 49 A map of Site 1 showing the locations of the gap, building, and mature forest phases ................... 57 PART II A summary of the human energy inputs to yuca production for a one ha site over a two crop cycle of conuco-use ........ 91 PART III Three profile diagrams representing the plants present at 3, 10, and 22 months in a permanent I x 20 m transect located in Succession Site 1. Only individuals more than 5 cm tall are illustrated ................ 142 The changes in density, for five plant groupings considered separately, and for all individuals together, over the first 22 months of succession following burning in Succession Site 1. Only individuals more than 5 cm tall were considered as established ........................ 144 Species area curves for Succession Site 1 at 4, 10, 16, and 22 months following the burning of this site. Only indivi- duals more than 5 cm tall were considered ........... 147 The change in the total number of species and in the number of newly encountered species for Succession Site 1 at 4, 10, 16, and 22 months following the burning of this site. Results are based on repeated surveys of 27 1 x 1.5 m permanent plots ................... 150 Monthly rainfall simultaneous with the mean number of new germinations present in monthly surveys of 17 1 x 1.5 m plots in Succession Site 1 for the period from 10 to 23 months following the burning of this site ........... 165 The changes in soil surface temperature over an eight hour period in each of six microhabitat types ........... 172 The changes in soil temperature (2 cm depth) over an eight hour period in each of six microhabitat types ........ 175 xiii Figure 8. 10. The mean amount of water lost over an eight hour daylight period from five evaporation containers placed in each of six microhabitat types ................ The mean number of individuals establishing from seed over a four month period in each of six microhabitat types . . . . The mean number of establishments in each of six microhabitat types for seven common early successional species. An equal number of seeds (from 300 to 600 depending upon the species) were broadcast in each microhabitat type. From five to seven replicates were used for each species . . xiv Page . 177 . 179 . 182 GENERAL INTRODUCTION Organization Since 1974 an international team of scientists has been investigat- ing tropical rain forest structure and function in the upper Rio Negro region of the Amazon Basin (Medina gt 21., 1977). The study is a Man and the Biosphere I Pilot Project under the MAB classification "Ecologi- cal effects of increasing human activities in tropical and subtropical forest ecosystems." The project is headquartered at, the Instituto Venezolano de Investigaciones Cientificas (IVIC) in Caracas, Venezuela. My participation in this project was sponsored through a National Science Foundation grant written by Drs. F. Golley and C. Jordan entitled "Nutrient dynamics of a tropical rain forest and changes in the nutrient cycle due to cutting and burning". Field Site Location and Climate The field site is situated in the Venezuelan state of Amazonas about four kilometers east of the village of San Carlos de Rio Negro, close to the confluence of the Guainia and Casiquiare rivers (lat 1.56 N., long 67.03 W) (Figure 1). San Carlos is 119 m above sea level. The state of Amazonas is one of the most sparsely populated regions of the world, having fewer than 0.05 inhabitants per km2 (Anon., 1973). Human settlements occur mainly along waterways. Most residents of San Carlos are of Indian descent. Figure 1. A map of Venezuela (bottom inset) with an enlargement of the San Carlos de Rio Negro region. The top inset shows the locations of the study sites relative to each other and to the town cfl’ San Carlos. Forest sites are abbreviated FS, suc- cession sites SS, and conuco (i.e. farm) sites CS. L \ SAN CARLOS 31‘3“” souuo I k csa o N: 3 o u 3 a A ”2 ' “2‘“ 2 \\ VI 4 0 m L) \ . \ l I,” I A\ cut,“ \ /" ‘ ._ / \ . \ N SD A O \ 0 SAN CARLOS 2. y' do 6% RIO GRO “\ r‘ a C; 1%. “P ‘v 7 VENEZUELA O ( 4 (r G ’ '7 The climate of the region is equatorial with a 26° C mean daily temperature. The average annual rainfall is 3500 mm with all months usually receiving in excess of 100 mm of rain. There are distinct seasonal trends in rainfall abundance; the driest months occur between November and February (Heuveldop, 1977). The local people concentrate most of their farming and fishing activities in the drier months. Geologyand Soils The configuration of the Amazon Basin dates back to the paleozoic era (Putzer, 1968 cited in Herrera, 1978). The present day basin is delimited by the Guiana shield on the north, the Brazilian shield on the south and the Andes on the west. Since the paleozoic, the basin has been filling in with sediments eroded from the surrounding higher ter- rain. The soils of the San Carlos region were derived from nearby sandstone shields in Guiana (Fittkau gt 31., 1975). The lack of recent geologic activity, the leaching effects of the continually hot wet climate, and the lack of bedrock exposure through erosion over the generally undisturbed flat to rolling landscape has resulted in an extremely oligotrophic environment. (Jordan and Herrera, 1979). In the San Carlos region the terrain consists of a system of rol- ling hills, some rising as much as 50 m above the mean river level (Herrera, 1977). The higher areas are referred to as “tierra firme" (i.e., never flooded). Granite batholiths (inselbergs) and flat topped sandstone "tepuis" are the sole orographic features over this vast rolling plain. The tierra firme soils are oxisols containing a moder- ately developed A horizon overlying lateritic gravel and clay. The soil surface in the tierra firme areas is covered by a mat of fine roots which varies in thickness from 5 to 40 cm. Thirty-seven percent of the total root biomass and fifty-eight percent of the feeder root biomass occurs in this root mat (Stark and Spratt, 1977). The presence of most of the feeder roots above the soil surface (even though the soil is well drained) emphasizes the oligotrophic nature of the environment. Water draining from these tierra firme sites is clear. Between the tierra firme sites are lower areas with soils composed of sandy quartz. These soils have a well developed Bh horizon at one meter depth and have been classified as spodosols (Klinge and Herrera, 1977). The upper 30 cm of these soils consists of humus in various stages of decomposition. Water draining from these lowland areas is stained a yellowish brown, the same color as the Rio Negro. The color results from the low exchange capacity of the spodosols which permits cations and decomposing organic matter in the form of humic acids to become dissolved in the soil water and move into the nearby streams (Medina gt gl., 1977). Areas of limited extent have soils which are distinct from the tierra firme oxisols and lowland site spodosol soils. Research Goals My research has focused on three distinct aspects of tropical ecology: 1) the composition, structure, and regeneration of the tierra firme forests; 2) the productivity of’ agricultural and successional sites; and 3) the dynamics of early succession after man-induced distur- bances. Multiple sites were used for each of these studies. The sites were designated as Forest, Conuco (i.e. farm) and Succession Sites 1, 2, 3, etc. Figure 1 contains a map showing the location of the study sites relative to each other and to the town of San Carlos de Rio Negro. All the sites were within two kilometers of a dirt road running between the towns of San Carlos and Solano. I conducted my research during four three-month stays (occurring at approximately nine month intervals) in the San Carlos region of the Rio Negro during the period from September 1975 to December 1978. The common goal of the U.S. scientists participating in this pro- ject has been to measure nutrient cycles in the forest, agricultural, and early successional environments of the San Carlos region of the Rio Negro. My research provides information needed to elucidate nutrient cycling rates. For example, in my forest study I collected data on the density and size of the forest trees which were used by Dr. C. Jordan and myself (Jordan and Uhl, 1978) to determine the standing crop biomass of the tierra firme forest (a parameter that must be quantified to determine the standing crop of nutrients). My studies on succession dynamics and agricultural energetics have provided data on standing crop biomass, litter production, and insect herbivory, all of which were necessary to evaluate the role of successional communities and crop plants in nutrient cycling. My research was conducted on plots where Dr. R. Herrera concurrently followed changes in soil fertility over time and Dr. C. Jordan monitored changes in the nutrient content of precipi- tation, leached soil water, and stream flow. By pooling our data we have been able to determine rates of nutrient cycling for forest, farm, and successional tierra firme ecosystems. While my research has contributed information necessary to ‘the quantification of nutrient cycles, it went beyond this immediate project goal and addressed questions of fundamental importance regarding tropical forest composition and regeneration, shifting cultivation energetics, and early succession dynamics. LIST OF REFERENCES LIST OF REFERENCES Anon. 1973. Atlas del Territorio Federal Amazonas. CODESUR. 102 pp. Fittkau, E.J., W. Junk, H. Klinge and H. Sioli. 1975. Substrate and vegetation in the Amazon region. Pages 70-93 1g H. Dierschke, ed. Vegetation und substrat. Vaduz, J. Cramer. Herrera, R. 1977. Soil and terrain conditions in the International Amazon Project at San Carlos de Rio Negro, Venezuela: correlation with vegetation types. Pages G 182-187 in Transactions of the International MAB-IUFRO workshop on tropical rain forest ecosystems research, Hamburg-Reinbek. Herrera, R., C. Jordan, H. Klinge and E. Medina. 1978. Amazon ecosys- tems, their structure and functioning with particular emphasis on nutrients. Interciencia 3:223-231. Heuveldop, J. 1977. Erste ergebnisse bestandesmeteorologischer unter- suchungen in regenwald von San Carlos de Rio Negro. Mitt. Bundes- forschungsanstalt fur Forst und Holzwirtschaft, 115:101-106. Jordan, C. and C. Uhl. 1978. Biomass of a "tierra firme" forest of the Amazon Basin. Oecologia Plantarum 13:387-400. Jordan, C. F. and R. Herrera. 1979. Tropical rain forests: are nut- rients really critical? Pages 29-48 in C. Jordan and F. Golley. Nutrient dynamics of a tropical rain forest ecosystem and changes in the nutrient cycles due to cutting and burning. Annual report. Institute of Ecology, Athens, Georgia. Klinge, H. and R. Herrera. 1977. Composite root mass from tropaquods under Amazon caatinga in southern Venezuela. Proceedings IV Inter- national Symposium of Tropical Ecology, Panama (in press). Medina, E., R. Herrera, C. Jordan, and H. Klinge. 1977. The Amazon Project of the Venezuelan Institute for Scientific Research. Nature and Resources 13:4-6. Putzer, H. 1968. Uberblick uber die geologische Entwicklung Sudameri- kas. Pages 1-24 in E. J. Fittkau e_t _a_l_., eds. Biogeography and ecology in South A—merica. Vol. 1. As cited in Herrera gt fl” 1978. Stark, N. and M. Spratt. 1977. Root biomass and nutrient storage in rain forest oxisols near San Carlos de Rio Negro. Tropical Ecology 18:1-9. PART I FOREST COMPOSITION, STRUCTURE, AND REGENERATION INTRODUCTION The vegetation types of Amazonia were systematically described by Ducke and Black (1953) and more recently by Prance (1978). Prance detailed eight principle vegetation types: forest on tierra firme (i.e., never flooded), forest. on inundated sites, savanna (n1 tierra firme, savanna (Ni inundated sites, campina (i.e., low vegetation on white sands), montane vegetation, littoral vegetation, and river beach vegetation. Surveys of Amazonia by the Brazilian Projecto RADAM, a radar survey of the entire Amazon basin, and by LANDSAT photographs are providing much needed data on the extent and distribution of these vegetation types. Tierra firme forest, the subject of this study, is by far the most common vegetation type, covering approximately 85% of the Amazonia (Prance, 1978). Some of the more common tree families in Amazonia are the Legumino- sae, Rosaceae, Lecythidaceae, Moraceae, Annonaceae, Burseraceae, Lauraceae, Myristacaceae, Vochysiaceae, Olacaceae, Linaceae, Dichapeta- laceae, Tiliaceae and Palmae (Ducke and Black, 1953). There are still many unclassified plant species. In some areas more than one percent of the collections made may prove to be new species (Goodland and Irwin, 1975). Prance (1972) recently monographed the Chrysobalanaceae plant family and found 22 new species from Amazonia. The central and north- eastern portions of Amazonia are recorded to be considerably richer in species than are the eastern and western outreaches. The northern area of the basin including the Rio Negro region is considered to be the most 10 11 species rich and to have the largest number of endemic species (BrUnig and Klinge, 1975). Continental drift and Pleistocene climate changes have been invoked to explain floristic differences. within Amazonia (Prance, 1978), but soil can also exert influences on species composi- tion (Fittkau gt gl., 1975). BrUnig and Klinge (1975) provide an excellent review of the numer- ous forest inventories that have been conducted in Amazonia. Most of these studies estimate merchantable timber volume only, but a few inves- tigators have examined tierra firme forest structure in detail (Takeuchi, 1961; Rodrigues, 1963). In addition, several workers have examined species richness and abundance patterns in Amazonian tierra firme forest (Black gt gl., 1950; Pires gt gl., 1953; Cain gt gl., 1956). Several vegetation types are found in the rolling topography charac- teristic of the San Carlos region. A topographic profile depicting the edaphic conditions and related vegetation types is shown in Figure 1. The higher tierra firme areas are covered by a mixed forest with a canopy height at 25 to 30 meters. The soils are oxisols. The low areas between these hills contain soils composed of sandy quartz. Altimetric transect studies conducted by Herrera (1977) showed these valleys to be slightly dome-shaped with Amazon caatinga vegetation (sensu Klinge gt gl., 1977), dominated by Micrandra spruceana, occurring on the wettest sites and low savanna-like vegetation (known as "bana") present on the more elevated, better drained domes. A third vegetation type occurs in the lowland sites in areas where the terrain is dissected by deep water holes and small streams. The soil here consists of mixed kaolinite clays and is related to the tierra firme soils. Eperua purpgrea is a strong dominant on these sites. BrUnig gt g1. (1977) have detailed the 12 Figure 1. An altimetric transect between the Rio Negro and the Casiquiare river showing maximum river level for 1976. The inset at the right upper corner shows a tracing of the micro- relief in the lowland area. (Figure used with the permission of Herrera). 13 m (D Q' (In) apnuuv engiolaa 00. ON. 0v. 1 0m. L 0m. Ema—3am scam E 09.225 .5354 s 3.0 2:26... B .33“. 3.52 E— V (w) epnm 14 composition and structure of these three lowland vegetation types. The purpose of this study was to characterize the composition and structure of the tierra firme forest in the San Carlos region. In addition, I investigated certain aspects of forest regeneration in this forest type. 15 METHODS Four tierra firme sites comprising a total area of 2.5 ha were chosen for study (See Figure 1, General Introduction for location of sites). Sites 1, 2, and 3 showed no evidence of recent disturbance and were used for the study of mature forest composition, structure, and regeneration. Site 1 was 1 ha in size and located approximately 4 km east of the town of San Carlos. Studies of nutrient cycling, forest productivity, and forest meteorology have been conducted on this site by other project participants. Sites 2 and 3 were each 0.5 ha in size and were located, respectively, 6 and 7 km east of San Carlos. These three sites were similar in elevation (approximately 100 masl), slope (less than 5°), and soil type (oxisols). According to local inhabitants these sites had never been disturbed by man. Site 4 had been cleared and then farmed for several years approximately 60 years previously. This site was located approximately 1.5 km south of Site 1. Structural character- istics were compared between this old secondary site and the three mature forest sites. In each site all trees 10 cm or more in dbh were mapped and identi- fied either by the local common name or assigned a number. The diameter of each tree was measured. Estimates of tree heights and crown dimen- sions were made by eye with periodic checks on relatively exposed trees using a Haga altimeter. Those trees between 5 and 10 cm dbh were identi- fied, mapped, and measured in three 10 x 100 m transects in Site 1; and 16 trees 1 to 5 cm dbh were similarly surveyed in two 5 x 100 m transects in this site. Initially, I relied heavily on local woodsmen who identified the more common species based on leaf type, trunk shape, presence of latex or other sap, and various characteristics of the inner and outer bark including odor, thickness, color, and texture. Every time a new species was encountered in the transects, a specimen was collected, dried, and placed in a reference collection. I used this reference collection to verify any questionable identifications provided by the woodsmen. Generally their determinations were accurate, although several species were sometimes lumped together under one designation. By constantly collecting specimens and comparing these collections with the reference collection, I was usually able to uncover inconsistencies and make distinctions between similar species. Additional collections of each species were sent to the Venezuelan National Herbarium in Caracas for drying and storage. Unfortunately, I was only able to find fertile specimens of about one-fourth of the tree species present in the study sites. Because it is exceedingly difficult to make identifications of sterile plants from a poorly known flora, the help of Ronald Liesner from the Missouri Botanical Garden was enlisted to collect fertile specimens of tierra firme forest trees and to provide species determinations. Identification of collected specimens is cur- rently under way at Missouri. In addition to the study of tierra firme forest composition and structure, the following three aspects of forest regeneration were investigated in Site 1: (1) regeneration of common tree species, evalu- ated through an analysis of size-class distributions; (2) the extent of 17 three forest phases (gap, building, and mature) (Whitmore, 1975), deter- mined by mapping these phases to scale; and (3) mortality and establish- ment in forest floor seedling communities, investigated through the use of 10 1 m2 permanent plots. In each of these plots all individuals were mapped, identified, and measured for height. These plots were period- ically resurveyed over a three-year period. 18 RESULTS AND DISCUSSION Composition and Structure of Mature Forest The diversity of tree species The high tree species diversity of Amazonian forests is well known (Black gt 11., 1950; Pires gt g_l_., 1953; Cain e_t fl” 1956; Takeuchi, 1961; Rodrigues, 1963; Prance gt gl., 1976). At San Carlos 83 species of trees 10 cm or more in dbh were present in Site 1, 79 in Site 2, and 63 in Site 3. Most species were represented by fewer than four indivi- duals per ha (Appendix Table 1). Several integrative indices have been devised which are sensitive to both species richness and species abun- dance. Two of these, the Simpson's index and the Shannon-Wiener index were calculated for the San Carlos sites (Appendix Table 2). Figure 2 shows that there was a sharp increase in cumulative species number each time a new site was surveyed. Hence, many of the species encountered were only f0und in one of the three sites. This suggests that species 'hi this forest may be clumped in their distribution. In accord with this, Hubbell (1979) found that most tree species in a Costa Rican dry forest were, in fact, clumped in their distribution. In an effort to compare the species richness of the San Carlos tierra firme forest with studies using both larger and smaller plots, I considered the suggestion of Cain and Castro (1959) that communities can be compared with regard to the number of species per unit area if the data are illustrated in semi-log fashion. These authors speculated that 19 Figure 2. The relationship between cumulative number of species and area for three 0.5 ha tierra firme plots several kilometers apart. IZOF 20 o/./. ./ IOO " m 0’. 2 'T s 80 - '/ e / O .5 SITE III E s so - . , - « 3 / 0 O .2 / § 0 SITE III / 0 ° / 20 - SITE I 5 :0 15 Number of IOOO m2 plots 21 a straight line relationship might exist for species number vs. log area for all communities. If' so, direct. comparisons between communities could be accomplished by comparing line slopes or more simply by compar- ing species number/log area values. Figure 3 shows such a semi-log plot for Site 1, San Carlos. Given the size of the area sampled, the points do not describe a straight line. A similar graph presented by Brilnig for several forest types in Sarawak and Borneo likewise shows a non- linear relationship (BrUnig, 1973). If most Amazonian tierra firme tree species are clumped in their distribution (rather than hyper-dispersed), then very large minimum survey areas (e.g. 10 ka) may be necessary for 'intersite comparisons of species richness based on species number/log area values. Importance values for tree species The high species diversity values for the San Carlos tierra firme forest (Appendix Table 2) suggest the absence of numerical dominance by any tree species. A more realistic assessment of the extent of domi- nance from the structural standpoint, however, may be achieved with an index that incorporates measures of basal area and extent of spatial distribution, as well as population size. In this regard, Curtis and McIntosh (1951) proposed that the "importance“ of a population in the community of which it is a component can be expressed as the sum of its relative density, relative dominance, and relative frequency. Impor- tance values were first used in the characterization of tropical vegeta- tion by Cain gt g1. (1956). I computed importance values for all spe- cies in each site. Results for Site 1 are shown in Figure 4. Eighteen species comprise 79.3 percent of the total importance value in this 22 Figure 3. The relationship between species number and log area for Site 1. Area is expressed as m2. 23 _ bl _ O O 4 2 Boga to 39:5: 3:2...an 80° ' 60 - IOO IOOO I0,000 IO Area (log IO) 24 Figure 4. The number of species in Site 1 occurring in a series of importance value classes. Only trees 10 cm or more in dbh were considered. 25 70f Ill/7’4” _ F . O . e m m 320 con 38on no .3522 878 I 7 3-8 s e s s I M , c e N a 3-8 v e 2-2 0 n 9.». m m «Ta 0. a-» m To 26 site. The remaining 20.7 percent of the importance value is shared among 65 species all having importance values of less than 1.7 percent. Expressed in more general terms, three-quarters of the species present in Site 1 constitute only one-fifth of the total importance value. Similar trends were seen in Sites 2 and 3. The highest importance value (as percentage of total) in Site 1 was 19.1 percent for Licania sp. This appears to be an unusually high value for Amazonian tierra firme forest. The highest importance values for Sites 2 and 3 were 11.6 percent and 10.2 percent, respectively. These lower values are more in accord with the importance value high of 7.8 percent reported by Cain gt g_l_. (1956) for a tierra firme forest in Belem, Para. Although Licania sp. had an importance value of 19.1 percent in Site 1, its importance value was 11.6 percent and 6.1 percent in Sites 2 and 3, respectively. Differences in importance values of this magnitude among sites were common and reflect the fact that a species common on one tierra firme site may be rare or even absent in another. Pires and Prance (1977) and Prance gt g1. (1976) working on tierra firme forest in Brazil also found that tree species varied con- siderably in their abundance from place to place within their range of dispersion. Such patterns of abundance cannot be explained until we understand how tropical tree populations are distributed 'hi space. Current studies by Hubbell in Panama are addressing this question. Density and basal area Figure 5 shows the distributions of trees in different diameter classes for San Carlos, Site 1. There is a high density of stems 15 cm in diameter at breast height (dbh) and a paucity of trees more than 50 cm dbh. Forest tree density in tropical lowland regions varies 27 Figure 5. The percentage of the total number of Site 1 trees, 1 cm or more in dbh, occurring in a series of five cm diameter classes. 28 -Etee-ae N 5 l5 '75 ”2 so 32 373 l0l2_ 7/ 9462 12 l l O - 0 <7 CU ebotueoud 99-09 69-99 99-09 69-99 99-09 69-99 99-09 62-93 92-02 6d-9l 9P0| Diameter classes (cm) 29 considerably (Table 1). Reported values range from 347 stems ha"1 in 1 British Guiana (Fanshawe, 1954) to over 800 stems ha- in Armenia Vieja, Ecuador (Grubb gt gl., 1963) for trees 10 cm or more in dbh; and from 1 1 in Armenia 187 stems ha- in Belem (Pires gt gl., 1953) to 466 stems ha- Vieja (Grubb gt gl., 1963) for trees 20 cm or more in dbh. Density of trees 10 to 20 cm dbh is considerably higher for the San Carlos sites than for other lowland tropical forests (Table 1). When trees 20 cm or more in dbh are considered, the San Carlos values are typical of those reported in other rain forest studies; however, when density for trees 40 cm or more in dbh is examined, the San Carlos values are lower than other reported values. The striking difference in density between large and small trees is also evident in the ratio of trees 10 cm or more in dbh to trees 20 cm or more in dbh (Table 1). I am not aware of any other study of lowland tropical tierra firme forest which reports ratios as high as these found in the three San Carlos study sites. The mean basal area for the three San Carlos forest sites was 27.8 m2 ha“1 for trees 10 cm or more in dbh. This figure is somewhat lower than the 33.2 m2 value reported for the tierra firme forest in the Manaus region of Brazil (Takeuchi, 1961) and considerably less than the 40-60 m2 values reported by Holdridge (1972) for lowland wet forest sites in Costa Rica. The low San Carlos basal area values are reflected in the low tierra firme forest biomass of 391 mt ha.1 (Jordan and Uhl, 1978). Density and basal area as a function of A horizon depth The thickness of the sandy A horizon overlying the heavy clay horizon varied from 3 to over 100 cm within Sites 1, 2, and 3. The 30 . NNmN .emewee_oz .e.e .e.c .e.c .u.e NNm moo .eoem epmou .u.: .u.: .u.: .u.c 5mm moccwzcwm .muvm mumou Aevmemfl ..Hm mm fiasco mouwm weave No.N .e.e Nam eNm NNN eczema cocoa .eeeesom mN.N .e.e ewe Noe mew onee> eeeoee< .eoeesow onflmmfl ._;o=eeee .U.C .U.C .U.C .U.C NNO .mzwcmz emmfi ..Hm mm eeeu eter— .U.C .Uec ADV .Uec .me «em—.mm mmmN ..Hm.mm meets eN.N .e.e NNN emN Nee .eepem ommN ..Hm mm eoe_m NN.N .e.c mmN NNN mNe .ee_em ON.m Nm NNN HNm meN eeee .me_eeu gem em.N we eNN see ONm m eoem .mo_eeo cam em.m om NmN New eNN N eoem .mo_eeu eem om.m NH NmN New mmN N oeem .mo_eeu cum 8 mme_u eNNm see so as M see go ON M can go ON w ON M can so ON M eoesom < mmmpu wNwm a U AmvcomHMUOA memuom: coa,msmum eo zuwmcwo .mLm53mme Ucm mo—Lmu cmm gm mmuwm wEka wLmeH Ucmpkop LO$ wmmmmpu wNwm ha hflmmcwt meH #meom .H wPQMH 31 .muopa ozu $0 coo: .muopa once we :mmzaov .muopa omen» we :mszuV .opnmpmm>m mew; mama o: umcu mmpmumucw .u.:AUV oco pcmmocaoc mumu .uwmewomam mmwzcmnpo mmm_:=finv .0 pm 0. NOV eOON .ezeemeea .O.O .O.O .O.O .O.O Nem .eOeOOO OOOOOLO OOOH .meeeeoex ON.N Oe ONN eON ONO .eeeeOOz ONON .eceeneea OO.N .O.O NNN mmm ONO .eeONOO zez AeONeON .Oee_OOOO ON.N NO ONN OeN Ome .eOOeOO O mmeFO ONNO OOO so Oe M OOO eo ON M OOO go ON w ON M OOO so ON M eonOM e mmO_O ONOO O O O Ooeoeoou mcmuuo: Lmalmempm yo zpwmcmo AOOOOOOOOOO .N epOee 32 density of large trees appeared to be related to differences in A hori- zon thickness. I checked for a relationship between A horizon thickness 2 and four elements of forest structure: basal area m- for trees with a dbh of more than 10 cm, number of trees with a dbh of more than 20 cm, number of trees more than 20 m tall, and number of trees 1-5 cm dbh.1 Plots (100 m2) having the thickest A horizon layer (more than 25 cm thick) had a mean basal area significantly greater (at the 0.05 level) than plots (100 m2) with the shallowest (between 5 and 15 cm thick) A horizon depth (t = 3.45, d.f. = 18). The density of trees with a dbh of more than 20 cm was also significantly greater (at the 0.05 level) in the plots with the thicker A horizon (t = 3.26, d.f. = 18). There was no significant difference in the density of small stems or in the den- sity of trees more than 20 m tall between plot groupings. Thus, differences in density of large diameter trees and basal area are associated with A horizon thickness and these differences can be detected on the scale of 100 m2 plots. The lack of an association between the density of trees 1-5 cm dbh and A horizon depth indicates that variations in A horizon thickness do not differentially affect seedling establishment or the survivorship or pole-sized trees. The failure to find an association between A horizon depth and the density of trees more than 20 m tall was at first surprising, because the densi- ty of trees more than 20 cm dbh is associated with A horizon thickness. 1The thickness of the nfineral soil was measured at 344 evenly distri- buted points within Site 1 with a thin metal rod that was inserted into the soil until the lateritic gravel was reached2 Basal area, diameter, and height of trees were measured in ten 100 m plots in the location? having the thickest layer of mineral soil and likewise in ten 100 m plots in the locations having the thinnest layer of mineral soil. A t-test was then used to check for differences in these four structural elements between each group of ten plots. 33 However, many of the trees in the sites studied are more than 20 m tall but have a dbh of less than 20 cm. It may be that trees more than 20 m tall on thin soils have very slow growth rates giving them a high prob- ability of falling over or being knocked over before reaching 20 cm dbh. Vertical structure The most common approach in studying the vertical structure of lowland tropical rain forests has been to construct profile diagrams. By illustrating vegetational features that are difficult to describe in words or numbers, profile diagrams serve as a useful tool in the study and comparison of tropical forest communities. To illustrate the verti- cal structure of the San Carlos tierra firme forest, a standard type (Davis and Richards, 1933) and an idealized type (Holdridge, 1970) of profile diagram were constructed. The Davis and Richards type of diagram is a scale drawing of a 50 x 7.6 m strip of forest based on measurements of the position, height, diameter, and crown dimensions of the constituent trees. This type of diagram has the disadvantage of being quite subjective in that indivi- dual workers rarely choose the same area as being the most representative of the forest type under investigation. Hence, no two diagrams from the same forest will be identical. Figure 6 is a profile diagram for the San Carlos tierra firme forest. Trees range to 30 m in height. The majority of the smaller trees depicted in this diagram are not mature and with time will become part of the upper canopy. Some portions of the diagram show distinct layering (e.g. between the 40 and 50 horizontal marks). In other por- tions of the diagram no layering is evident (e.g. between the 1 and 30 Figure 6. 34 A profile diagram of the tierra firme forest, San Carlos de Rio Negro. The diagram represents a strip of forest 50 n1 long by 7.6 m wide. Only trees over 7.6 m tall are depicted. 35 .5 8:23 On ON A 0. ON .nN .0» (W) mom 36 horizontal marks). It is clear that even within this 50 m strip of forest, vertical structure differs from point to point primarily due to regeneration patterns and species specific differences in the height of mature individuals. The second type of profile diagram constructed was the idealized profile diagram of Holdridge (1970). The idealized profile attempts to represent the total mature structure which has been attained and toward which any immature portions of the forest are developing. Mature indivi- duals are represented in proportion to their density. The resultant diagram is an abstract representation which eliminates the subjectivity of the standard profile. An idealized profile diagram for Site 1 is shown in Figure 7. Although the forest tree species differ in their mature heights, the differences are continuous (i.e., mature-height species clustering or strata are not apparent). In theory, species stratification could result from groups of species whose average mature heights are different enough to form distinct strata with discontinuities between them. However, there is In: convincing evidence from tropical rain forest studies for this type of stratification (Grubb gt g_l_., 1963); nor does this study provide support for species stratification. The stratification of individuals based on total height or canopy height is more controversial and has been frequently addressed in studies of vertical structure. Some workers have presented evidence for the existence of strata in some stands (Grubb gt gl., 1963, Bimbino site), while others have found no evidence for their existence (Schulz, 1960; Paijmans, 1970). In this study, two approaches were employed to test for the existence of stratification of individuals. First, the percen- tage of foliage-free space at 1 m vertical intervals was estimated using Figure 7. 37 An idealized profile diagram of the San Carlos de Rio Negro tierra firme forest based on ten 0.1 ha plots. 90 ‘ F“. ”'4; ' A . / ~~ O _ Oma‘ ”a {m l/ 40f N (W) WPPH IOO 80 70 30 20 I0 Distance (to) 39 the Davis and Richards type profile diagram in Figure 6. Results of this analysis are illustrated in histogram form, Figure 8. As a second test of stratification, midcrown height was calculated for all trees more than 10 cm dbh in Site 1 and the number of trees in one meter mid-crown height classes was determined. Results are illustrated in histogram form in Figure 9. The interpretation of these figures requires a subjective judgment as to whether a peak in the histogram really represents a stratum. Furthermore, as Grubb gt g1. (1963) pointed out, wide height classes will tend to obscure strata while narrow ones will produce more. I have used both 1 and 3 m height intervals for both histograms, and peaks or hollows are not apparent in either case. Although both histograms suggest that strata do not exist on a hectare scale, layering is present on a smaller scale. The occurrence of gaps results in a flush of growth from previously suppressed sapling popula- tions. If heights of these gap scale strata vary throughout the forest, the distribution of heights shown in Figure 9 would be expected. A summary index of forest composition and structure Information on species richness can be combined with data on densi- ty, basal area, and height into a single quantitative expression using Holdridge's complexity index (Holdridge, 1972). In examining tabulated data from several study sites in Costa Rica, Holdridge noted that these four vegetation parameters could be combined multiplicatively to produce a numerical value which agreed with his subjective impression of vegeta- tion physiognomy. The use of indices based on combined characters is not without precedent. Goff and Cottam (1967) suggested that vegetation parameters such as height, basal area, and density could be combined in a variety of ways to form meaningful indices. 40 Figure 8. The percentage of cuxui or non-canopy space at 1 m vertical intervals in a 7.6 x 50 m strip of forest in Site 1. 41 .-. 2:: 3': ..I .:. :2 3;: :- '=‘ a. .. .3 'i :- 5’ :2 I: I. .. .0 .0 .C .. :1: :3 0‘. .. .. 0'? O) O Percentage of open space 01 O Height ( m) 42 Figure 9. The distribution of mid-crown heights for trees 10 cm or more in dbh in Site 1. Mid-crown height = tree height - (crown depth/2). 43 80" b P O 0 6 4. 23233. No .8952 20t Ones» Crown height (m) 44 The complexity index and the component values thereof are shown for the San Carlos forest and three other tropical tierra firme forests in Table 2. The mean complexity index value of 197.3 (based on twenty 10 x 100 m plots) for the three San Carlos sites is only 55 percent to 58 percent of the values reported by Holdridge (1972) for Costa Rican tierra firme rain forest and 75 percent of the value derived from the data of Takeuchi for Manaus (1961). Costa Rican forests have higher complexity index values than the San Carlos forest because they are taller and have a greater basal area per m2, whereas the main component contributing to a: higher complexity index value for the Manaus tierra firme forest is the number of species. San Carlos, with a mean annual temperature of 26°C and a mean yearly rainfall of 3500 mm, falls into Holdridge's moist forest life zone (Holdridge, 1972). Holdridge proposed 270 as a standard complexity index value for the "mature natural moist forest life zone on well drained upland sites without any excessively favorable or restrictive growth factorsfl' The San Carlos value of 197.3 is considerably below this (more similar to Holdridge's lower montane wet forest values). The infertile tierra firme soils of San Carlos (Herrera gt gl., 1978) may constitute a “restrictive growth factor" and thereby account for the low San Carlos complexity index value. As previously noted, the complexity index components (i.e. species number, density, basal area, and height) were quite variable both within and between sites. Holdridge (1972) recommended that at least three 0.1 ha plots be used to calculate the mean value of each complexity index component. To determine if three 0.1 ha plots were sufficient to give a reliable complexity index estimate for the San Carlos forest, four sets 45 .HmmH .m mama .Ocozoxmp nm>gao mmcm mowuoam soc» umuzasouanv .HmmH .mH mama .wcoaoxwh mopnma mecmamwu sage umuzasooamv H.omm e.Hm H.Nm mH.e o.em Numa .mmuwcupo: momm mpmou .mmo N.emm m.Hm n.mm mm.e o.me Nnma .omuwcupo: mowm mumou .mmccmacwm O.ONN 0.0m OO Nm.m 0.0m NOON .Oeosexee m.~mH H.Hm m.N~ mm.N m.Hm Am .N .H mmpmmv moPLmu cmm xoucH coneaz prmcmo mme< gnome: cowumuoO xuwxwpasou mowomam memm .mmpwm weave mecomu eocpo new ummeoe moFLmu cmm an» ace womcmcu mmnpm> ucmcoasou mzp new xwucw hawxwpasoo .N wpnmh 46 of three 0.1 ha plots were analyzed in Site 1. Plot sets chosen for analysis were separated by ten meter wide strips with the first plot randomly selected. Mean complexity index values (for the four sets of plots) ranged from 92.5 to 199.8, s = 44.04 [(64.0 5 150.3 5 236.6) = 95% confidence interval]. Given this variability for complexity index means based on three 0.1 ha plots, it is apparent that reports of the Holdridge complexity index should include a measure of variability. The successional status of the forest under study Throughout this study I have referred to the tierra firme forest on Sites 1, 2, and 3 as "mature." Yet, the high density of stems in the smaller size classes, the low basal area per hectare, and the low canopy height of these forest sites, as compared to literature values for lowland tropical forest structural features, may suggest that the sites studied are late secondary forest still in the building phase of develop- ment. To compare the three tierra firme study sites with secondary forest, I visited 17 abandoned agricultural sites along the Rio Negro with Manuel de Silva and Pedro de Silva, serving as guides. Both are old men who have lived in the San Carlos region all their lives. Most sites consisted of a deteriorated dwelling surrounded by a mosaic of abandoned agricultural plots of different ages ranging from ten years old to more than 60 years old. The sites more than 60 years old were distinguishable because: 1) the canopy was lower; 2) trees of higher diameter classes were absent or when present were of fast growing, soft-wooded species; and 3) fruit trees which were planted at the time of clearing had grown to be canopy trees. An old abandoned site which was typical of the others in structural development was chosen for further study and designated as Site 4. One of the guides recalled his 47 relatives having cleared this site when he was a young boy. On this basis, the site was estimated to be about 60 years old. The soils of this site were similar to those of Sites 1, 2, and 3. Site 4 (0.5 ha) had 68 species of trees 10 cm or more in dbh. Sites 2 and 3 (each also 0.5 ha) had 79 and 63 species of trees present, respectively. Hence, there may be no difference in tree species diver- sity between late secondary sites and the older undisturbed forest sites at San Carlos (Appendix Table 2). Knight (1975) reached a similar conclusion from his studies of old and very old secondary sites in Panama. The Site 4 canopy was at 20 m with many stems loosely packed in the understory space (Figure 10). In contrast, the profile of Site 1 (Figure 6) shows a canopy height at about 25 m with some evidence of local stratification under the tallest trees. The mean height for trees 10 cm or more in dbh was 13.9 m in Site 4 vs 17.2 m in Site 1. The smaller stature of Site 4 is also reflected in the low basal area value for this site (Table 3). The density of both small and large diameter trees was greater in Site 1 than Site 4 (Table 3). It is possible that the more closed, even canopy of the 60 year old site inhibits understory growth, whereas the more broken canopy of the older forest (i.e., Site 1) allows greater light penetration and therefore a denser understory. A second factor of importance may be that forest understory species are slow to colonize disturbed sites. In accord with this idea, an estimated 25.1 percent of the 1-4 cm diameter stems in Site 1 belonged to understory species (i.e., never growing to more than 5 m in height or 5 cm dbh). Only 3 percent of the Site 4 1-4 cm dbh stems were of understory species. Figure 10. 48 A profile diagram of a: 60 year old second growth forest, San Carlos de Rio Negro. The diagram represents a strip of forest 50 m long by 7.6 m wide. Only trees over 7.6 m high are depicted. 49 .5 3.5.20 cm 9. on 8 o. e N - . _ a . .. _. ,, l»: \lj . o. ., ..// > r \ .. r ‘)&/4‘. m. I 8 L mm ow (W) "mow 50 Table 3. Structural characteristics of a 60 year old abandoned farm site and a mature forest site (i.e., Site 1). Characteristic 60 Years Mature Mean height for trees 3 10 cm dbh 13.9 17.2 Basal area per ha of trees 3 1 cm dbh 19.0 34.3 Density per ha of trees 1-4 cm dbh 2,816 9,457 Density per ha of trees 5-9 cm dbh 553 1,012 Density per ha of trees 3 10 cm dbh 518 786 51 Only one of the 17 old secondary sites I visited was indistinguish- able from the surrounding tall forest. One of my guides recalled visit- ing this site 50 years earlier. At that time, the remains of a long abandoned village were barely evident among the forest regrowth. I estimated the site to be 150 to 200 years old. Based on the study of Site 4 and my travels in the region, I am satisfied that Sites 1, 2, and 3 represent the mature phase of forest development for this region. The question as to why the San Carlos forest is comparatively low in stature and basal area when compared to other tropical rain forests of the world still deserves attention. The central Amazon basin is a region of very infertile soils (Herrera gt gl., 1978). Grime (1979) has suggested that under conditions where mineral elements are scarce, natural selection has led to the evolution of plant species which make low demands upon the nutrient stocks of the soil. These species are, in turn, reduced in stature and in potential growth rate. I have already discussed the reduced stature of the San Carlos forest, and results from five years of tierra firme forest tree growth monitoring (Murphy and Jordan, unpublished data) show that dry matter production is also low in this forest (i.e., about 1 mm diameter growth per year for trees more than 10 cm dbh). Hence, I hypothesize that the small stature of the forest is a reflection of the oligotrophic conditions of the central Amazon basin. 52 Regeneration within Mature Forest Seedling dynamics Seedlings are abundant in the tierra firme forest and have high initial survivorship. I defined seedlings as plants less than one cm in dbh but more than two cm tall. The mean density of seedlings in Novem- ber 1976, based on a survey of 52 randomly located 1 m2 plots in Site 1, was 15.2 m-2 (s = 10.1, range 8 to 82). Hence, the estimated number of seedlings per ha is 152,000. The distribution of seedlings in four height classes is shown in Table 4. The difference in seedling number between two size classes can be interpreted as the probability of a seedling surviving from 'the smaller' to the larger size class. For example, the probability of a seedling surviving from the 3 2 cm < 25 cm size class to the 3 150 cm size class is 9 percent (Table 4). I followed survivorship and mortality of tree seedlings through time by conducting four surveys at roughly one year intervals on ten permanent 1 m2 plots located by random methods in Site 1 (Table 5). Mean annual survivorship of marked tree seedlings was 89.3 percent. Most mortality (96 percent of total) occurred in the 2-25 cm height group. Nevertheless, the mean annual percentage of individuals surviv- ing in this height group over the three year study period was 83 percent. The mean annual percentage survivorship for individuals greater than 25 cm tall was 99 percent. However, it appears that an estimated 85 per- cent of the individuals 25 cm or more in height die before reaching a height of 150 cm (Table 4). This suggests that many years are necessary to grow from a height of 25 cm to a height of 150 cm. The height growth of tree seedlings was, in fact, very slow based 53 mm.o m.H N.N m.m cowumw>wu ucmucmpm a OO.O O.N e.e O.N Ns so see: em mmH mmm mmm Opmzuw>wucw Punch so OOH M so OOH v so OO M so OO v so ON M so ON 0 so N M .H mwwm cw OHOPQ E mco mm we >m>csm mcu co women mnempumwm eo cowuznwcumwu azmwm; on» .e wpnmh N 54 eN.O eH OH OO NOH ONNNH NH.m Om NH eOH NOH NNNHH 0.0 O OH ON OOH NN\N -- -- -- eO eeH ON\NH AHOHOHcoz OpcmEcOOHnmpOm Ocpmoo HHmp Eu mN v so N M O>m>czm osesOOOHOsoOO soHOOOO OOHOOoO HOOOH NO osHH .H eoHO OH OooeoOH OOOHO E H Hcmcmscoa OH cw nowsma Lem» omega m sm>o Ocmne:: chHummO :O Ommcmcu mcp No >cm553m .m wHamp N 55 (Mi a one year growth study of 15 tree seedlings between 10 and 20 cm tall and having less than ten leaves. The mean annual stem elongation was 1.57 cm (range 0.5 to 3.5). The mean number of new leaves produced over the one year study period was 1.53 (range 1 to 4), while the mean number of leaves lost was 0.86 (range 0 to 3). The high annual seedling survivorship, low rates of seedling estab- lishment (Table 5), and the slow seedling growth rates suggest a persis- tent ground layer community. Temporal stability of forest floor vegeta- tion in the tropics has been noted by other authors. Went (1973) found 12.2 five to ten year old seedling per m2 in the Amazonian forest near Manaus. In this regard, it may be instructive to think of tropical rain forests as containing a seedling and sapling pool lying in wait for long periods of time until disturbances favor their growth. Gap and buildingphases The most common disturbance that provides favorable conditions for the growth of this seedling and sapling pool is the creation of gaps. Poore (1968), working in central Malaya, found that 9.9 percent of the 2 Hartshorn (1978) found 5.05 rain forest there was in the gap phase. (mean of four plots) of the La Selva, Costa Rica forest to be in the gap phase. I mapped the portions of Site 1 in the gap, building, and mature phases of development; using this map (Figure 11), I determined that 4.8 percent of the San Carlos forest was in the gap phase and 18.4 percent was in the building phase of development. Knowing the proportion of an 2Gap phase is any opening in the forest canopy with seedlings and sap- lings not exceeding 2.7 m tall. Building phase consists of juvenile trees of pole size, but no large trees. The mature phase of the forest has large trees, poles and seedlings intermixed (Whitmore, 1975). 56 Figure 11. A map of Site 1 showing the locations of the gap, build- ing, and mature forest phases. 57 BUILDING PHASE Dilemmas: 58 area in gap phase, it is possible to calculate the forest turnover time. Gap phase persists for about five years in the San Carlos forest. Hence, the turnover time for San Carlos forest is approximately 104 years (100/4.8 x 5 = 104.1). This is comparable to Hartshorn's (1978) calculated turnover time of 118 i 27 years for the La Selva forest in Costa Rica. Size class distribution of important tree species As the turnover time for the San Carlos tierra firme forest appears to be quite short, (i.e. approximately 100 years), it was of interest to know if the species composition would change in the future. Several tropical forest studies have shown the floristic composition of the upper canopy stratum to be different from that of the seedling and sapling strata (Watson, 1937; Aubreville, 1938; Richards, 1939; Jones, 1945). To explain this phenomenon, Aubreville (1938) proposed that the floristic composition of a given area of forest is changing continually, but in a cyclical fashion with groups of species succeeding each other periodically. Support for Aubreville's idea has come mainly from stu- dies in the old world tropics. I analyzed the size class distributions of the common tree species in the San Carlos tierra firme forest (Site 1) to evaluate the applicability of Aubreville's ideas to this forest. Eighteen tree species comprised 79.3 percent of the total impor- tance value for Site 1 (Figure 4). At maturity all of these species have diameters in excess of 20 cm and yet 13 of the 18 species had more than one-half of their individuals in the 1-5 cm dbh class. Because all transects used to sample trees less than 10 cm dbh were located in non-gapped forest, these 13 species appear to be able to regenerate in 59 the shaded understory or at least to persist there for a long time. Furthermore, the representation of each of these species in almost all diameter classes up to 30 cm suggests a temporal continuum of regenera- tion. Three of the 18 most common species in Site 1 had few or no repre- sentatives in the smaller size classes. Most of the individuals of these three species were in the 15-25 cm dbh size class. Although I lack information (Ml the floristic composition of gaps, this type of distribution (i.e. skewed towards the larger size classes), is commonly interpreted to mean that the species in question requires gaps in order to regenerate (Schultz, 1960). The remaining two species had size class distributions which were difficult to interpret because of small sample size. Given that 72 percent of the 18 most common Site 1 tree species were regenerating on site, the abundance pattern of the common tree species will not change appreciably over time on this site. Although this contradicts Aubreville's ideas, it is in accord with the findings of Davis and Richards (1933) for British Guianan mixed forest and with results reported by Schulz (1960) for Surinam. Richards (1952) sug- gested that African forests which appear to have cyclic regeneration may actually be early seral stages. Again, the evidence suggests that in spite of its overall small stature, the San Carlos tierra firme forest may well be representative of one of the region's natural climax vegeta- tion types. 60 SUMMARY Composition, structure, and regeneration were investigated in the tierra firme rain forest of the San Carlos region of the Rio Negro. Species richness of this forest was comparable to other lowland tierra firme forests. Eighty-three tree species 10 cm or more in dbh were found in one l-ha site. The species area curve appeared to level off slightly as 0.1 ha plots were grouped within sites, but when 0.1 ha plots from additional nearby tierra firme sites were added, the cumula- tive number of species increased sharply, suggesting that species in this forest may be clumped in their spatial distribution. Species abundance patterns were similar to other tropical areas, most species being quite rare. The island-like nature of tierra firme sites in the forest landscape offers a unique study opportunity for investigations of plant distribution and evolution. Structural differences were also noted between and within sites. The density and basal area of trees more than 20 cm dbh were greater in locations with a thick A horizon than in locations with a thin A horizon. When compared structurally to other lowland tierra firme forests described in the literature, the San Carlos forest: (1) has a higher number of trees 10 cm or more in dbh, but a smaller number of trees 40 cm or more in dbh; (2) has a basal area value 84 percent that of the tierra firme forest near Manaus, and considerably lower than most other lowland tierra firme forests where values are known; (3) is similar in height (averaging 17.3 m for trees more than 10 cm dbh) to tierra firme forest in the Manaus area but is less 61 tall than lowland forests in other areas where data are avail- able; (4) is similar in vertical structure to other tierra firme for- ests, showing no evidence of stratification of mature species height (idealized profile diagram) nor any evidence of strati- fication of individuals; and (5) has a mean Holdridge Complexity Index of 197.3 which is consi- derably lower than predicted for lowland rain forest without restrictive growth factors. Although the San Carlos forest has relatively high density values for small trees more than 10 cm dbh and relatively low basal area values, this forest is clearly different in vertical structure, canopy height, density, and basal area from old secondary sites, and on this basis is considered to be representative of mature forest for the San Carlos de Rio Negro region. The small stature of this forest is thought to be a reflection of the oligotrophic conditions of the central Amazon Basin. The high seedling survivorship (83 percent mean annual survivorship over a three year period for seedlings 2 to 25 cm tall), low rates of seedling establishment, and slow seedling growth rates suggest a persis- tent forest floor seedling community. Gaps are common in the forest and their occurrence is probably beneficial to the persistent seedling and sapling pool. Most of the important tree species were well represented in all diameter* size' classes indicating 'that ‘the sites studied are floristically stable. LIST OF REFERENCES LIST OF REFERENCES Aubreville, A. 1938. La foret coloniale: les forets de l'Afrique occidentale francaise. Annals Academy Science Colonial, Paris, 9:1-245. Black, G. A.; T. Dobzhansky and C. Pavan. 1950. Some attempts to estimate species diversity and population density of’ trees in Amazonian forests. Botanical Gazette 3:413-425. BrUnig, E. 1973. Species richness and stand diversity in relation to site and succession of forests in Sarawak and Brunei (Borneo). Amazoniana 4:293-320. BrOnig, E. and H. Klinge. 1975. Structure, functioning and productivi- ty in humid tropical forest ecosystems in parts of the neotropics. Mitteilungen der Bundesforschungsanstalt fur Forst und Holzwirt- schaft, Reinbek bei Hamburg. Brijnig, E. G., J. Heuveldop, J. Smith and D. Alder. 1977. Structure and function of a rain forest in the international Amazon ecosystem project: floristic stratification and variation of some features of stand structure and precipitation. Pages 2-31 in Transactions of the international workshop on tropical rain forest ecosystems research, Hamburg-Reinbek. Cain, S. A. and G. M. de Oliveira Castro. 1959. Manual of vegetation analysis. Harper, New York. 325 pp. Cain, S. A.; G. M. de Oliveira Castro; J. M. Pires and N. T. da Silva. 1956. Application of some phytosociological techniques to the Brazilian forest. American Journal of Botany 43:911-941. Curtis, J. T., and R. T. McIntosh. 1951. An upland forest continuum in the prairie-forest border region of Wisconsin. Ecology 32:476-496. Davis, T. W. A. and P. W. Richards. 1933. The vegetation of Moraballi Creek, British Guiana. Journal of Ecology 21:350-384. Ducke, A. and G. A. Black. 1953. Phytogeographical notes on the Brazil- ian Amazon. Annals Academia Brasileira Ciencias 2511-46. Eggeling, W. J. 1947. Observations on the ecology of the Budongo rain forest, Uganda. Journal of Ecology 34:2087. 62 63 Fanshawe, D. B. 1954. Forest types of British Guiana. Caribbean Forester 15:73-111. Fittkau, E. J.; W. Junk; H. Klinge and H. Sioli. 1975. Substrate and vegetation in the Amazon region. p. 70-93. 1g H. Dierschke, ed. Vegetation und substrate. Gramer, Vaduz. Goff, F. G. and G. Cottam. 1967. Gradient analysis: the use of spe- cies and synthetic indices. Ecology 48:793-806. Goodland, F. J. A. and H. S. Irwin. 1975. Amazon jungle: green hell to red desert? Elsevier Scientific Publishing Company, New York. 155 pp. Grime, J.P. 1979. Plant strategies and vegetation processes. John Wiley and Sons, Chichester. Grubb, P. J., J. R. Lloyd; T. D. Pennington and T. C. Whitmore. 1963. A comparison of montane and lowland forests in Equador. I. The forest structure, physiognomy, and floristics. Journal of Ecology 51:576-601. Hartshorn, G. S. 1978. Tree falls and tropical forest. dynamics. p. 617-638. in P. B. Tomlinson and M. H. Zimmermann, eds. Tropical trees as living systems. Cambridge Univ. Press, England. Herrera, R. 1977. Soil and terrain conditions in the International Amazon Project at San Carlos de Rio Negro, Venezuela: correlation with vegetation types. Transactions of the international MAB-IUFRO worksh0p on tropical rain forest ecosystem research, Hamburg- Reinbek, 0182-187. Herrera, R., C. Jordan, H. Klinge and E. Medina. 1978. Amazon ecosys- tems: Their structure and functioning with particular emphasis on nutrients. Interciencia 3:223-231. Holdridge, L. R. 1970. A system for representing structure in tropical forest associations. Pages 8147-150 jg H. T. Odum and R. F. Pigeon, eds., A tropical rain forest. A study of irradiation and ecology at El Verde, Puerto Rico. U.S. Atomic Energy Commission. Oak Ridge, Tennessee. Holdridge, L. R. 1972. Forest environments in tropical life zones: A pilot study. Pergamon Press, New York. Hubbell, S. P. 1979. Tree dispersion, abundance and diversity in a tropical dry forest. Science 203:1299-1309. Jones, E. W. 1945. The structure and reproduction of the virgin forest of the north temperate zone. New Phytologist 44:130-48. Jordan, C. and C. Uhl. 1978. Biomass of a "tierra firme" forest of the Amazon Basin. Oecologia Plantarum 13:387-400. 64 Klinge, H.; E. Medina and R. Herrera. 1977. Studies on the ecology of amazon caatinga forest in southern Venezuela. 1. General features. Acta Cientifica Venezolana 28:270-276. Knight, D.H. 1975. A phytosociological analysis of species rich tropi- cal forest on Barro Colorado Island, Panama. Ecological Monographs 45:259-284. Paijmans, K. 1970. An analysis of four tropical rain forest sites in New Guinea. Journal of Ecology 58:77-101. Pires, J. M.; T. Dobzhansky and G. A. Black. 1953. An estimate of the number of species of trees in the Amazonian forest community. Botanical Gazette 114:467-477. Pires, .L. M. and (L T. Prance. 1977. The Amazon forest: a natural heritage to be preserved. Extinction is forever symposium. The New York Botanical Garden 158-194. Poore, M. E. D. 1968. Studies in Malaysian rain forest. I. The forest on the triassic sediments in Jengka Forest Reserve. Journal of Ecology 56:143-196. Prance, G. T. 1972. Chrysobalanaceae. Flora Neotropica 9:1-410. Prance, G. T. 1978. The origin and evolution of the Amazon flora. Interciencia 3:207-222. Prance, (L T.; W. A. Rodrigues and M. F. da Silva. 1976. Inventario florestal de um hectare de mata de terra firme km 30 da Estrada Manaus-Itacoatiara. Acta Amazonica 6:9-35. Richards, P. W. 1939. Ecological studies on the rain forest of Southern Nigeria. 1. The structure and floristic composition of the primary forest. Journal of Ecology 27:1-67. Richards, P. W. 1952. The tropical rain forest. Cambridge University Press, London. Rodrigues, W. A. 1963. Estudo de 2.6 hectares de mata de terra firme da Serra Navio, Territorio do Amapa. Boletim do Museo Paraense E. Goeldi, Nov. ser. Botanica, No. 19, 22 pp. Schulz, J. P. 1960. Ecological studies on rain forest in northern Surinam. Verhand-konik. nederl. Akad. Wetenschappen, Afd. Natuurk. Tweede reeks, Deel LIII, No. 1, 267 pp. Takeuchi, M. 1961. The structure of the Amazonian vegetation. II. Tropical rain forest. Journal Faculty of Science, University of Tokyo, III, 8:1-26. Watson, J. C. 1937. Age-class representation in virgin forest. Malay Forester 6:146-147. 65 Went, F. W. 1973. Competition among plants. Proceedings of the Nation- al Academy of Science, USA. 70:585-590. Whitmore, T. C. 1975. Tropical rain forests of the Far East. Oxford University Press, London. 66 m: N mmH nun m.H o.N m.m H.m n.mH N.¢N m.m¢ ommw ..Hm um cwmu onsmoaz m; H moa in: in: m.o N.N m.m v.0N m.mN m.mm mmmH ..HM Hm Omcwg Hmccmummu O; H NO --- --- e.O O.e N.O 0.0H O.eO O.NO OmmH ..Hm Oe OoeHO sonm m; H mm N.H in- m.m m.m ¢.m N.MH m.mH m.m¢ mopcmu cmm mete OOHoOOO eON-ONH ONH-OO e0-00 NO-NH OH-O O-e O-N H OOOOeoOO Heoos HOOOH OOOOOHQ wucmcczn< .OOHOO OELOO mcsomu Hmowaocp ucmHon socuo use .H mpwm .moHemu cmm cow OmOOmHu mucmuczam we OOOLOO oweumeoom O :O “commea Omwumam Hmuou No oompcwocoa on» .H mHnmh xOucoaa< 67 xms: \: n m . N xms Huwcaeseu ecu cw Omwumam we geese: mg» n m ewes: m me_ u I .NueH ea OHmeH Eeew Lepumw cewmsm>ceo m u mHNm.m ecm ”Omwumem saw egg we OHmeew>weew we geese: ecu n we ”Omwomem HHm we OHmeew>wecO we geese: Hmuep man u z ”mews: Aw: oneH we w m u z OHmeHv mHNm.m u :Hev .xpweeeseo esp aw w Omwueem we OHmeew>wecw we cewpceeece u we meme: NHwva u H u oamv m.m Nm. H.O o.m mm. mmN mm m; m.o e muwm N.m mm. O.e m.m Om. mmm mm m; m.o m mpwm m.e Om. m.m e.m mm. Own mm m: m.o N mawm o.m mm. e.o O.e Nm. wen mm m; H H mpwm Oewomem .ez u xmez I m xmecH OHmeew>wecH Omwueem m~wm cemumcmwmmo OHmeew>wecH .ez HevxeecH wecmw3uceccmcm m.~emeewm we cmesez we smesez muwm .e ecm .m .N .H Oeuwm cw can so OH M Omega sew mmsz> xmecw hawmcm>wo .N mHemH xwecmee< PART II PRODUCTIVITY OF AGRICULTURAL AND SUCCESSIONAL SITES INTRODUCTION Inhabitants of the Rio Negro region of the Amazon Basin farm by shifting cultivation. The essential characteristics of shifting culti- vation are that an area of forest is cut, burned, and farmed for several years without tilling the soil and then allowed to revert to forest before being cleared to farm again. A typical shifting cultivation cycle 'hi the San Carlos region of Amazonia begins with the cutting of a one-half to two hectare patch of mature tierra firme (i.e., never flooded) forest. The small trees and lianas are cut with machete while the larger trees are felled with an axe. Cutting occurs in September and October (i.e., at the beginning of the dry season). The cut vegetation is left to dry for several months and then burned. Burning causes carbon, nitrogen, and sulfur to be volatilized as oxides. The remaining nutrients are added to the soil as carbonates (e.g., CaCO3, K003) or as phosphates and silicates (Nye and Greenland, 1960). Soon after burning, yuca (Manihot esculenta, Crantz) 2 is planted at a density of one to three stem stocks per m throughout the clearing known locally as a "conuco." Yuca roots are ready for harvest about one year after planting. Yuca harvesting goes on contin- ually after the first year, a 100 to 300 m2 area being harvested every one to two weeks. The harvested areas are replanted to yuca immediately. By 18 months the conuco consists of a mosaic of small plots containing yuca in varying stages of developments Approximately 20 percent of the conuco has crops such as pineapple, plantain, banana, cashew, sugar 69 70 cane, yams, potatoes, or beans planted in combination with yuca. Frequently the conuco is abandoned after the second yuca crop because of declining crop production. 0n better soils conuco use can- tinues for a third cropping period. When it becomes apparent to the cultivator that the plot will only produce one more yuca crop, he cuts a new section of mature tierra firme forest. Sometimes the new plot is immediately adjacent to the plot being abandoned. Occasionally old (i.e., more than 60 years old) second growth stands are chosen for farming, but the low population density of this region precludes the necessity of cropping fallowed land at the present time. This method of agriculture is common throughout the lowland tropics, occurring on a wide array of soil types and practiced by people of widely varying cultures (Nye and Greenland, 1960). Shifting culti- vation practices have been the focus of numerous studies. A biblio- graphy by Conklin (1963) lists over 1300 titles related to all facets of shifting cultivation. In addition to ecologists, the study of shifting cultivation has attracted geographers (Watters, 1960) and anthropolo- gists (Leach, 1959) who have been interested in the cultural aspects of the phenomenon. Nye and Greenland (1960) and UNESCO (1978) provide reviews of the shifting cultivation literature. Three detailed studies of shifting cultivation have been conducted in Venezuela. Watters (1971) working in Northwest Venezuela, investi- gated the technological, sociological, and economic aspects of shifting cultivation and made recommendations for stabilizing and improving conditions of rural land use. His report includes information on farm size, labor inputs, and crop yields. Jolly e_t_ g1. (1961; cited by Petriceks, 1968) conducted a similar study on small farm sites in 71 Northeast Venezuela. Petriceks (1968), in a study encompassing northern and central Venezuela, focused (“1 land tenure practices, kinds of crops planted, yields, population movement, education of farmers, income distribution, and timber exploitation. These three studies provide aui excellent overview of shifting cultivation in all areas of Venezuela except the state of Amazonas where no farm sites were examined. The first part of this paper is devoted to an energetic analysis of shifting cultivation as it is practiced in the San Carlos de Rio Negro region of Amazonia. I compare the energetic returns in the form of food output with the human energy input necessary to raise this food. The energetic efficiency values presented in this paper provide one standard by which land-use planners, agronomists, and others can judge the merits of alternative farming schemes proposed for Amazonia. The second part of this paper compares conuco crop productivity with early successional vegetation productivity. It has been suggested that the rates of net primary production attained by natural plant communities may be indicative of the rates that can be achieved through innovative agricultural systems (Gomez-Pompa gt gl., 1973). If this is true, and if the San Carlos conuco crops are significantly less produc- tive than the natural vegetation, there would be a basis for attempting to design more productive agricultural systems using the structural and life-form dominance patterns of the natural vegetation as a model. The relevance of such a scheme for the San Carlos region of the Amazon Basin is discussed. 72 METHODS Time and Energy Expenditure in Yuca Crop Production I timed conuco field activities and yuca root processing activities in several conucos 'hi the San Carlos vicinity. Field activities were broken down into the following 10 categories: cutting of small stems by machete, felling of large trees by axe, burning, post-burn clearing, yuca stem gathering, stem cutting, stem planting, weeding, crop harvest- ing, and transportation. Each activity was measured on at least two occasions, and all activity times were expressed (Ni a hectare basis. Yuca root processing activities were divided into seven steps: peeling, grating, squeezing, sifting, fire building, baking, and transportation. Activity times for each of these activities were measured by monitoring the yuca root preparation process from beginning to end over two sepa- rate three-day periods. I estimated the human energy expended in these conuco field and root processing activities by separating the activities according to their difficulty into light, intermediate, demanding, and very damanding work groupings. I used calorie expenditure values of 50 kcal per 45.4 kg human body weight per hour for light work, 100 kcal per 45.4 kg per hour for intermediate work, 150 kcal per 45.4 kg per hour for de- manding work, and 250 kcal per 45.4 kg per hour for very demanding work. These values agree with figures supplied by Rose (1938) for work ranging from light (e.g., washing dishes) to heavy (e.g., sawing wood). Men do 73 most of the field work while women do most of the root processing. I estimated that the average adult male weighed 59 kg (basal metabolism approximately 60 kcal per hour) and the average adult female weighed 54 kg (basal metabolism approximately 55 kcal per hour). Basal metabo- lism values are from Carpenter (1948). Using these values, I was able to estimate the total amount of human energy necessary to farm a one hectare conuco and the energy expended in processing a given quantity of yuca roots. Yuca root productivity per hectare for a two-crop cycle of conuco use was estimated based on measurements taken in experimental plots (see yuca production). Knowing the energetic content (in kcal) of the yuca root production, I was able to compare the energetic returns in the form of food with the total work energy input necessary to secure that food, and thereby evaluate the energetic efficiency of the indigenous farming system.1 Conuco Productivity A 60 x 120 m section of mature tierra firme forest was cut in September 1976 and burned in December 1976. This site was used to compare the productivity of the native crop plants with the productivity of the naturally colonizing early successional vegetation for the two year period following burning. The agricultural plot, designated Conuco Site 1 (see Figure 1, General Introduction, for location), was 30 x 50 m 1Generally, conucos are farmed actively far two to three years (i.e., the time necessary to produce two yuca crops). Yuca is by far the most important crop 'hi all conucos of the San Carlos region; 93 percent of the edible crop production in Conuco Site 1 was yuca. Because other crops represent a minor portion of the total crop production, they were not considered in the energetic analysis. 74 in size and occupied the southern half of the burned site. All conuco work was performed by local workers who were paid for their assistance. Juan Moreno, an experienced local farmer, was em- ployed to supervise the cutting, burning, planting, weeding, and harvest- ing Operations to ensure that each conformed with local practices. This approach allowed ready access to the conuco for measurements and experi- mentation, while ensuring that the site was being operated in a tradi- tional manner. Yuca (Manihot esculenta), pineapple (Ananas sativas), plantain (Musu sp.) and cashew (Anacardium occidentale) were planted in this site starting in January 1977. Productivity was measured differently for each of these four crops. Yuca production Yuca production was estimated through measurements of standing crop biomass, litter production, and herbivory by chewing insects. Standing crop biomass was determined by weighing the above ground (i.e., non- edible) biomass and the below ground (i.e., edible) biomass of all yuca plants ready for harvest during the study period.2 Subsamples of leaves, stems, and roots were taken for dry weight determination. Yuca leaf litter production was estimated by counting leaf scars on 50 randomly located plants at each harvest period. Leaf litter samples 2Large roots were separated from roots less than 1 cm diameter for 50 randomly located yuca plants. The ratio of these two root fractions was determined to allow for the estimation of fine root biomass which the farmers do not bother to remove in the general root harvest. In addi- tion, the above ground portions of each of these 50 plants were sepa- rated into leaf and stem fractions and weighed. Subsamples were taken for dry weight determination. This leaf to stem dry weight ratio was used to estimate the leaf and stem fractions of all subsequently har- vested yuca plants. 75 were taken to determine the mean dry weight per leaf. The number of plants being harvested and the length of the cropping period were then used to estimate yuca leaf litter production. Yuca branch litter pro- duction was estimated prior to each harvest by gathering all downed yuca branches in two 1 x 30 m permanent transects. The collected branch litter was dried and weighed. Yuca production lost to chewing insects was measured by removing the bottom-most leaf (i.e., the oldest leaf) from 75 randomly selected yuca plants at the end of years one and two. The leaves were traced, and the leaf area and area eaten were determined using planimetry. Traced leaf blades were weighed to determine their area to weight rela- tionship. This information was combined with estimates of the total number of leaves produced per year to determine yuca production losses to chewing insects. Yuca production was investigated in two additional one year old conucos designated as Conuco Sites 2 and 3 (see Figure 1, General Intro- duction) to determine if the production values in Conuco Site 1 were representative of other nearby conucos. Both of these conucos were planted for the first time concurrently with Conuco Site 1. The diame- 2 plots were ter and stem length of all yuca individuals in 30 one In measured in each of these two conucos and in Conuco Site 1 at the end of the first year. Root production was then estimated using a regression equation. All yuca production data were expressed on an annual basis to permit comparison with succession site data and on a crop basis to allow for an analysis of agricultural energetics. 76 Pineapple production3 Pineapple fruit production was measured by harvesting and weighing all ripe pineapples at weekly intervals. Periodic subsamples were taken for dry weight determination. Pineapples were occasionally eaten by frugivores. Production losses to frugivores were estimated by multiply- ing the mean weight for non-damaged fruits by the total number of fruits with frugivore damage. Twenty marked pineapple plants were checked every two weeks for litter production. All dead leaves or dead parts of leaves were col- lected, oven dried, and weighed. These data were used to estimate the rate of litter production. At the end of year two the number of pineapple plants in the conuco was counted and each plant was categorized as pre-fruit-bearing, fruit- bearing, or post-fruit-bearing. Five individuals typical of each of these categories were harvested, separated into fruit, leaf, stem, and root portions, and weighed. Sub-samples were taken for dry weight determination. The total density of pineapple plants in each of these three stages and the average weight of a plant in each stage were com- bined with the fruit harvest and litter collection data to estimate pineapple plant production for the two year period. The proportion of the two year pineapple plant production occurring in year one was esti- mated to be one-third of the estimated two-year standing crop (since pineapple plants were only present for six months of year one) plus the 3Local farmers introduce pineapple, plantain, and cashew plants into their conucos throughout the first year of conuco farming. The relative proportions and overall density of these fruit plants are dictated in large part by the availability of plants from neighboring conucos. All fruit plantings were introduced into Conuco Site 1 approximately six months after the burn. 77 litter production during year one minus the estimated pineapple plantlet input weight. Plantain production No plantain fruits were produced during the two year study period. Plantain generally begins to fruit in the second or third year of conuco use and continues fruiting after the conuco has been abandoned. The diameter and height of all plantain plants present at the end of two years were measured. Several plantain plants were then harvested out- side of Conuco Site 1 and used to establish regressions between diameter2 x height and below ground biomass, and diameter2 x height and above ground biomass. These regression equations were then used to estimate the plantain standing crop biomass in Conuco Site 1. To mea- sure plantain litter production, 10 marked plantain plants were surveyed every two weeks, and all dead leaves or dead leaf parts were removed, oven dried, and weighed. The regression estimate of standing crop biomass together with the litter production measurements were used to estimate plantain production for the two-year period. The proportion of the total two-year plantain production occurring in year one was approx- imated in the same manner as described for pineapple. Cashew production Several of the 11 cashew trees present in Conuco Site 1 were begin- ning to flower at the end of two years but none had produced fruit. Leaf standing crap at two years and leaf litter production up to two years were estimated by counting the number of leaves and leaf scars on all cashew trees present in the site at two years. Weight per leaf 78 was determined by harvesting, oven drying, and weighing 50 cashew leaves. The total length of brown (i.e., old, 1.5 to 4 cm diameter) and green (i.e., young, 0.3 to 1.5 cm diameter) stems per plant at two years together with the weight per cm of each stem type was used to estimate the total two year stem production per plant. Root weight, as a percen- tage of above ground weight, was determined by destructive sampling outside of the study plot. The proportion of the two year cashew pro- duction occurring in year one was approximated in the same manner as described for pineapple. Weed production Weeds are common in conucos, particularly after the first year. The harvest method was used to estimate weed production. Twenty-seven 1 x 1.5 m plots were randomly located in Conuco Site 1 ten months after burning. Light weeding which occurred prior to the establishment of these plots was not measured. All non-crop plants in these 27 plots were removed at 10, 16, and 21 months. The harvested plants (i.e., weeds) in each plot were placed in one of five groups: forbs, grasses, secondary woody species, forest tree species, and lianas. The plants in each of these groups were divided into root, stem, and leaf parts. The plant material thus divided was bagged, oven dried, and weighed. This method slightly underestimated weed production because a few forbs and grasses had fruited and begun to decompose by the time the weedings occurred. Secondary Site Productivity Secondary site productivity was measured in Succession Site 1. This site was immediately adjacent to Conuco Site 1 (see Figure I, 79 General Introduction). Measurements of standing crop biomass, litter production, and leaf herbivory by chewing insects were used to determine the productivity of this site. Standing crop biomass at year one was measured by harvesting all plants present in 25 one m2 plots. Plants in each plot were grouped as forbs, grasses, secondary woody plants, forest trees established by sprouting, forest trees established from seed, and lianas. The plant groups present in each plot were then divided into leaf, stem, and root tissue and bagged, oven dried, and weighed. Year two biomass was mea- sured in the same manner for all plants, except those individuals, two meters or more tall, in the genus Cecropia. The root and stem standing crop of these larger Cecropia was estimated by the regression of diameter2 x height on root weight and on stem weight for 13 Cecropia individuals harvested outside of the study site. Cecropia leaves of these harvested plants were placed in three size classes, and the mean weight of leaves in each size class was determined. Then, the stem diameter, height, and number of leaves in each leaf size class were measured for all Cecropia, two meters or more tall, in nine 5 x 5 m plots located in three transects placed in the center of Succession Site 1. The standing crop of the Cecropia individuals present in the nine plots was estimated using the regression and leaf-size-class weight relations. Litter production for the first 10 months of succession was esti- mated by counting leaf scars on all plants present in 25 randomly chosen one m2 plots in Succession Site 1. Plants of the same species as those encountered in the survey plots or of the same general type, in the case of forest trees and lianas, were located outside of Succession Site 1. 8O Twenty-five lower leaves (about to fall) were taken from each plant species and from each plant type for dry weight determination.4 This procedure provided an estimate of mean dry weight per leaf of each plant species and plant type which was then used to calculate litter produc- tion for the first 10 months of succession. Litter production in Succession Site 1, after 10 months, was mea- sured by harvesting 16 randomly located .2 x 2 n1 permanent. plots at weekly intervals. Litter collected from each plot was bagged, oven dried, and weighed. The amount (H: leaf material removed by chewing insects was esti- mated by removing the bottom-most leaf from 30 to 40 plants of the herbaceous, secondary woody, primary tree, and liana plant groupings at the end of years one and two. The total leaf area and area eaten were measured by planimetry for each plant group. Petioles were removed and the leaf blades dried and weighed to determine area to weight relation- ships tkn‘ each plant group. The litter collections at 11, 16, and 23 months, were separated into leaves of herbaceous, secondary"woody, primary tree, and liana species and the percentage of the total leaf litter contributed by each of these plant groups was calculated. The estimates of the total leaf litter production for years one and two, the percentage of the total leaf litter belonging to each of the four plant groupings, the amount eaten by chewing insects for each of the plant groupings, and the leaf area to weight relationship for each plant grouping were used to determine the dry weight of the collected leaf 4Leaves of forest tree species were generally similar in size, belonging to the mesophyll size class of Raunkiaer (1934). Leaves of lianas were more variable in size and shape, but most were also in the mesophyll size class. 81 litter which had been removed by chewing insects prior to leaf fall. The percentage of the standing crop leaf biomass removed by chewing insects for years one and two was considered to be equal to the percen- tage removed by chewing insects from the bottom-most leaves in each of the plant groupings based on the assumption that leaves are only vulner- able to insect attack during the phase of leaf expansion [see Nicholson, Paterson, and Curie (1970) and Rhoades and Cates (1976) for support of this contention]. 82 RESULTS AND DISCUSSION Time and Energy Expenditure in Yuca Crop Production Table 1 shows the time spent, the relative difficulty, and the estimated energy expended in conuco field activities undertaken during a two crop cycle of conuco use. The activities are listed in the chrono- logic order of their occurrence. Forest cutting accounts for 8.0 per- cent of the total human energy devoted to farming (i.e., to both field activities and root processing activities). The understory tree species and saplings of canopy species are cut with a machete. Three experienced conuco farmers, working for 11 hours each, machete-cleared 11,979 m2 or 363 m2 per man-hour. At this rate, 28 man-hours are neces- sary to clear one hectare by machete. Generally, trees more than 10 cm in diameter are cut by axe. The intermediate sized trees (i.e., 10 to 40 cm diameter) are only partially cut. The large trees are then com- pletely felled. As the large trees fall, they carry down with them the partially cut trees. Four adult males required 243 hours to axe-fell 2.47 hectares of mature forest. Thus, the men cleared 102 m2 per hour or one hectare in 98 hours. The cut slash is left to dry for several months. The burn occurs in December or January, the driest period of the year, and is immediate- ly preceded by f0ur to five consecutive rainless days. Using a torch fashioned from tree bark, several farmers start numerous small fires along one side of the clearing and then walk through the clearing to the 83 OOH.mNm eNNH Ocmg Oecu eceumm Oseesesee O ONO.OH OO Oseeee; OOHOH Oswesesee O ONO.OH OO Oseeeoz esoon OOOOsesOO N OOH.HN OO OOOOOOHO Oseon neweemeee OH OOO.HO ONH Oeco eceoem sew cewpmemamea cw newcmmHu mpOO checmeme OH OO0.0e OOH pmm>emc Oeee Hmewm Osweemsme O ONO.OH OO cheemz aeee pOst OcOecmEme O OOH.HN mm acwaemHe ecm chupae seam mua> cheememe m OOH.OH Oe HeeOOcmep ecm OcOcmcme seam mo:> cheemeeu >Lm> O OON.OH OO OcmcmmHu cannuumee checmame >em> H OOO.N O chcszm Oswecmsme >em> HH OHm.mm Om mxm new: meme» eOLmH we chauau mcwecmeee >em> m ONO.OH ON mpmcoms cum: Osmum HHmsO we chaueo OoHOoOOOHO HHOOOH Heseooee LOOO Hoeeuoee OOH>OOo< we NV Hmex Lea meaecv Hmex mEOO .mmzneueceu we mHoxu Oeeo e3» m cheae Omwpw>mgum eHmww eozceo :O eczuweemaxm zmsecm emamewumm on» cam .zupeewwwwe e>wumHme any .ucmam mew“ och .H mHemH 84 other side, igniting slash as they go. These small blazes converge and the whole clearing burns powerfully for 10 to 15 minutes. The fire never advances more than a few meters into the surrounding forest, however. Six man-hours are required to burn one hectare. The work is demanding, but because of its short duration, accounts for only 0.4% of the total human energy input to farming. After the site is burned, yuca stem stocks must be gathered from surrounding conucos and transported to the site. Conucos are usually within 10 minutes walking distance of each other, and farmers are gener- ally willing to give away stem stocks of harvested plants. I estimated that 4800 yuca stems (i.e., the number of stems necessary to plant one hectare at a density of 2.4 stem stocks per m2) could be gathered, bundled into groups of 100, and transported to a new conuco in 40 hours. This estimate provides 30 minutes to gather and bundle 100 stems with 20 minutes for transport. Based ("1 this time estimate and the demanding nature of the work, 1.7 percent of the total conuco energy input is devoted to this activity. Once yuca stems are on the site, they must be cut and planted. The time necessary to cut yuca stems into 20-30 cm stem stocks and insert these into the soil was measured by timing an experienced farmer engaged in these activities. Fourteen stem stocks were cut per minute. To plant the stem stocks, a hole is made in the ground using a stick with a pointed end, and the stocks are inserted into this hole. An average of 7.4 stem stocks were planted per minute. On this basis 83 man-hours are required to cut yuca stems and plant one hectare at a density of 2.4 85 stem stocks per m2.5 These activities account for 7.0 percent of the total human energy input to farming. A considerable amount of work goes into site cleaning and weeding in a two crop conuco cycle. After the burn, time must be spent cutting, piling, and reburning any remaining small slash. The time devoted to this activity (i.e., post-burn clearing) varies, but if the site burns poorly, 100 or more hours per hectare may be required to render it clean enough for planting. Based on my observations, I have assigned 50 hours per hectare as an average amount of time devoted to this activity. About eight months after the site has been cleaned and planted, the first weeding is done. As the life of the conuco is prolonged, weeding occurs with increasing frequency. Weeds are removed by their roots where possible, and the work is demanding. I measured the time it took farmers to weed measured areas of their conucos on two occasions, and 2 per hour. At this rate, 66 found that they weeded approximately 152 m hours are needed to weed one hectare. A thorough weeding combined with trash burning occurs at each yuca harvest. I observed and timed this activity in Conuco Site 1 and in a nearby conuco. Approximately 80 m2 were cleared per hour; hence, an estimated 125 hours are required to clean one hectare after a yuca harvest. Based on these time estimates for the various conuco cleaning and weeding activities and the difficul- ty of these activities, they consume an estimated 16.7 percent of the total conuco energy input. Once the yuca plants are large enough to remove, harvesting of 100 5Density is usually 1 to 3 individuals per m2 at the first yuca planting when stem stocks must be trangported to the site from varying distances, and 2 to 3 individuals per m for subsequent crops when stem stocks are readily available on the site. 86 to 300 m2 sections of conuco occurs every few weeks until the site is abandoned. The plants are removed from the soil by first tugging gently at the base of the stem until the roots are loosened and then pulling firmly. The roots are broken from the base of the stem by hand. The main stems and branches are then broken into 1.5 to two m sections and stacked. A portion of these stems supplies the stem stocks for the next yuca planting. The remaining stems are either given away or rot in place. Labor devoted to the harvesting process accounts for 13.4 per- cent of the total energy input for a two crop cycle of conuco use. Foot travel to and from conucos depends on conuco location relative to dwelling location and on work day length. Most conucos were l/2 to 1 hour walking distance from .San Carlos. On ‘this basis, the average conuco was considered to be 3/4 hours away from the family dwelling. The amount of time a man works in his conuco on a given day varies but is usually from 5 to 10 hours. I regarded eight hours as an average day's work and divided the total field hours worked during a two crop cycle of conuco use by eight to estimate the number of trips taken to the conuco. This number (128.9) was then multiplied by the number of hours necessary to walk to and from this hypothetical conuco (i.e., 1.5) to arrive at an estimated 193 hours spent in transportation. Implements and produce are usually carried on these trips. Walking with light loads requires intermediate amounts of energy (i.e., approximately 100 kcal per 45.4 kg human body weight per hour). Given these time and energy estimates, 6.1 percent of the total conuco energy input occurs as transportation to and from the conuco to do field work. Table 1 shows that 1224 hours are devoted to conuco field activi- ties during a two crop conuco cycle. This field time input corresponds 87 to an estimated 323,150 kcal or an energy input of 264 kcal per work hour. After the yuca roots are harvested, many hours of work are required 6 This is accomplished by to transform them into a palatable form. peeling, grating, and soaking the roots. This procedure provides a favorable environment for linase, a naturally occurring enzyme in the yuca roots, to break down a glucoside in the roots to glucose, acetate, and hydrocyanic acid. The soaking material is then squeezed to remove the water soluble hydrocyanic acid. Table 2 shows the estimated time spent processing an amount of yuca roots equal to that produced by the first two yuca crops on Conuco Site 1. This table divides root processing into seven activities and shows the estimated energy expenditure and difficulty for each activity. Yuca roots are peeled by scraping the skin off with a dull knife. This is light, but time-consuming work; it comprises an estimated 15.6 percent of the total conuco energy input.7 After peeling, the tubers are rendered into a gruel by rubbing them vigorously on a grating board. This moderately demanding work comprises an estimated 9.0 percent of the total energy input to farming. After soaking, the ground roots are squeezed dry and sifted. These operations account for 2.6 percent of the total conuco energy input. 6Bitter yuca is much more common than sweet yuca (a variety of bitter yuca) in the San Carlos area conucos. When questioned, the farmers indicated that the bitter yuca stores better in the soil and is less fed-upon by insects and rodents than sweet yuca. These advantages may outweigh the greater efficiency of preparing sweet yuca for consumption. 7Approximately one-half of the harvested roots are soaked for two to three days before peeling. The skins of these soaked roots are readily removed by hand. Some fermentation takes place during soaking. This fermentation may lead to protein enrichment, although the local people maintain that soaking is done only to improve the flavor of the pro- cessed yuca. 88 eHO.OON OHO.N OOHOH eOOOOesseosO ON OOO.eN NOe OOOOOOLOOOOOOH OOOO_ OH OOO.HN OON OOHOOO use OOOOOO oOOOH O OOO.OH OeH OOOOOOO Ose OOOOHOOO OLOO OOOO_ O HOO.N OO OsOoOOO OOOOH O NeN.O NO OOONOOOOO OOOO Oseesesee OH OOO.eO OON OOOOOOO OOOO osOO_ OO OOe.eO ONO OOHHOOO OOOO szaoOwwwO HHmueH Hmcmpom; sway Hmcaecv zuw>wuo< es OO HOoO osOH HOoO .Hmmzteozeeo we eHoze Oesu e3» m Lm>e m: Lea eHeOO mozx emumswumm on» OO OsmeOeHOx OOHOV Oueee men» we OemLOeHOx OOHO OewOOmoecO :O emeceaxm xmgece ecu ecm .szsowwwwe m>OumHmc on» .uceam mama och .N mHamH 89 The sifted material is sometimes formed into large flat cakes, 50 cm in diameter and 1 cm thick, called "casabe," which are baked over an open fire for several minutes. Alternatively, the sifted material may be stir-fried for 20 to 30 minutes, producing a dry granular cereal called "manioco." Fire building and cooking account for 2.7 percent and 4.5 percent of the total conuco energy input, respectively. Travel to and from the conuco for yuca processing occurs every one to two weeks. Approximately 100 kilograms of yuca roots are processed at one time. I estimated that yuca was processed 72 times based on the fact that the wet weight yield of the two yuca crops produced in Conuco Site 1 was 717.5 grams per m2 or approximately 7175 kg per ha. Each processing requires two round-trips (1.5 hours each) by two adult women. On this basis, 432 hours are devoted to travel for root processing in a two crop cycle. Expressed in energetic terms, 12.4 percent of the total conuco energy input goes to this activity. Figure 1 summarizes the human energy inputs to yuca crop production for a two crop cycle of conuco use in the San Carlos area. An estimated 606,464 kcal are devoted to field work and root processing activities. Slightly more energy (53 percent) is spent in field work than in root processing (47 percent). Harvesting and weeding are the field activi- ties demanding the most energy; in combination they account for 30 percent of the total conuco energy input. The peeling and grating of yuca roots are the most energy consuming processing activities, account- ing for an estimated 25 percent of the total human energy input to conuco agriculture. Total wet weight yuca root production in Conuco Site I for both crops was approximately 7,175 kilograms per hectare (to be discussed 90 Figure 1. A summary of the human energy inputs to yuca crop production for a one ha site over a two crop cycle of conuco-use. 91 Cutting forest 48,496 kcal °/. Burning 2,309 kcal 0.4 % Gathering stem stocks 10,197 kcal 1.7 °/o Planting ‘-' 42,320 kcal 7.0% ; Cleaningand weeding ' 101,589kcal ‘ 16.7% _ Harvesting 81,578 kcal 13.4% Peeling - 94,439 kcal 15.6% Grating and soaking 54.385 kcal 9.0 % Squeezing and sifting 15,793 kcal Fire making and baking 43,728 kcal 7.2 % Traveling 111,630 kcal 18.4 % 92 more fully in the next section). Based on a literature value (Montaldo, 1977), the total energy content of this yield is estimated to be 8,406,345 kcal. A total of 3234 hours was spent in conuco field and root processing activities over the two crop cycle. The estimated energy yield per man hour of work was 2602 kcal (Table 3). Hence, a man farming in San Carlos can satisfy his caloric needs with one hour of work a day. A family of six can receive its caloric needs if the hus- band and wife each farm for approximately two hours a day. Should the family decide to produce yuca in excess of its needs, it can sell the surplus manioco in increments of 12 kilograms for 30 Bolivars (US $7.50). Approximately 18 hours of labor would be necessary to produce these 12 kilograms. The ratio of total energy output in the form of food to total energy input in the form of human labor for a two crop conuco use cycle is 13.9 to one (Table 3). This ratio, computed for crop one separately, is 14.8 to one and for crop two separately is 12.6 to one (Appendix Table 1). The ratio decreased for crop two because of a decrease in yuca production. Farmers sometimes abandon their conucos after harvest- ing the first crop. This is not always a wise strategy from an ener- getics standpoint. For example, post-harvest cleaning, planting, and two weedings are required in the second cropping period of conuco opera- tion. An estimated 340 hours per hectare are needed for these activi- ties. If a new conuco is established instead, the time necessary for cutting, burning, post-burn clearing, stem stock gathering and trans- port, planting, and one weeding will be 371 hours per ha. Hence, estab- lishing a new conuco will require 31 more field hours; this extra work input may not be compensated for by higher production in the new site. 93 Table 3. A summary of human energy inputs and yuca crop energy outputs for a two crop cycle of conuco use. Unit of Parameter Measurement Kcal Total wet weight root production 7175 kg/ha 8,406,345(a) Total field labor spent in food production 1224 hours 323,150 Total processing labor spent in food production 2010 hours 283,314 Manioco yield per man hour work 691 grams(b) 2,602 Ratio of energy out (i.e., yuca yield caloric content) to energy in (i.e., human energy inputs) 13.9 to 1.0 (a)Ca1culated as follows: 7,175 x 1000 = 7,175,000 grams yuca roots 7,175,000 x 0.3116 = 2,235,730 grams dry weight of manioco 2,235,730 x 3.76 = 8,406,345 kcal. One gram dry weight of yuca has 3.76 kcal (Montaldo, 1977). (b)Calculated as follows: .3116 x 7175 = 2235.7 kg dry weight of manioco produced 3234 hours spent in field and processing work 0.691 kilograms of manioco produced per hour human energy input 1224 + 2010 2235.7/3234 94 If the decision is between clearing a new site after the second yuca harvest or continuing to farm the same site for a third cropping period, the energetically correct choice is almost always to abandon, because yuca yields decline markedly and increased time must be spent weeding in the third cropping period. Some native cultivators, however, take pride in being able to farm a site for three or even four cropping periods, even though they sacri- fice labor efficiency in doing so. There may be benefits in extending the period of conuco use. For example, many kinds of fruit trees can be planted, given a longer cropping period. Conuco fruits may be the main source of some essential minerals or vitamins for the local inhabitants. In addition, longer cropping periods may maximize long term yuca produc- tion per land parcel, thus maximizing the percentage of the surrounding land in mature forest at any one time. Maintaining a high percentage of the surrounding land in primary forest may reduce disease, improve hunting, allow ready access to necessary fibers, or accomplish other desired ends. An alternative explanation is that farmers who maintain conucos for more than two cropping periods prefer to do light work (e.g., weeding) for long periods of time rather than heavy work (e.g., forest clearing) for short periods of time. They may maintain this preference even though doing the heavy work would require less overall labor energy input. Several other studies have focused on the energetics of shifting cultivation. Watters (1971) measured the total field labor inputs at four conuco sites in northern Venezuela. Between 65 and 86 days per year were required to do all tasks in the annual cycle for a one hectare conuco. The San Carlos estimate of 64 conuco field work days per ha per 95 crop (see Table I) is in accord with Watters' findings. Norman (1978) using data from Clark and Haswell (1970) gathered in Africa found energy output to input ratios ranging from 7 to 37 (unweighted average 18.5) for cereal crops and from 9 to 54 (unweighted average 29.5) for non- cereal crops. Carneiro (1961) estimated that 9,880,000 kcal of yuca roots were produced per ha per crop by the Kuikuru Indians of central Brazil. This is more than two times the energy content of the San Carlos yuca crop yield. Carneiro did not detail the labor inputs neces- sary to achieve this yield, but he did comment that Kuikuru men only work two hours per day in the field. This suggests that the Kuikuru farming system also has high energy output to input ratios. Finally, Rappaport (1971) reported that the ratio of crop energy output to work energy input was 16.5 to one for shifting cultivators in New Guinea. Although similar to the San Carlos ratio of 13.9 to one, the energy outputs and inputs in Rappaport's study were nearly three times greater than those for San Carlos. Infertile soils are the probable cause of the comparatively low crop yields in the San Carlos region. The high energetic efficiency at the expense of economic efficiency characteristic of shifting cultivation systems is in sharp contrast to mechanized agriculture in developed regions. Pimentel gt g1. (1973) investigated energy outputs and inputs involved in corn production in the United States. Twenty-two hours of human labor were required to grow one hectare of corn in 1970. In contrast, approximately 516 hours of field labor per hectare are necessary per year in San Carlos. However, after determining the energy inputs in the form of machinery, gasoline, fertilizer, seeds, irrigation, herbicides, drying, electrici- ty, and transportation, Pimentel gt g1. estimated that the ratio of 96 kilocalories food produced to kilocalories energy input was 2.82 to one. Most of the energy input was in the form of gasoline and fertilizer, neither of which are used in the San Carlos conucos. Human labor inputs can be reduced to a minimum in the United States because labor saving energy inputs are presently inexpensive there. However, in Amazonia, annual incomes are very low and labor saving inputs are very expensive. Given these conditions, labor-intensive systems of land use are the most sensible for this region. Gross gt g1. (1979), in their study of the agriculture and market participation of four Brazilian Indian societies, found that the degree of participation in the external market economy by the Indian groups was related to their difficulty in making a living from slash and burn agriculture. The San Carlos area farmers must spend 384 hours to pro- duce one million kilocalories of food. Indian groups with efficiencies in this range had become active in the market economy in the Gross gt g1. study. Wage labor and occasional construction of handicrafts are the principal ways San Carlos residents are involved in the market economy. The local people readily suspend their agricultural activities when offered outside employment. However, in the case of the San Carlos region, it is not the degradation of the habitat per se that prompts the movement from shifting cultivation to market activities (as Gross gt g1. suggest), but rather the extremely low fertility soils and low faunal numbers (i.e., the habitat is "degraded" to begin with) that prompts one to seek money as a means of improving his living conditions. When money is available, the residents normally buy guns, machetes, boats, outboard motors, and chickens (i.e., all protein related purchases). 97 Conuco Productivity Table 4 summarizes yuca, pineapple, plantain, and cashew production in Conuco Site 1 for years one and two. Yuca production Nineteen percent of the total yuca production was leaf material. Most of these leaves (72 percent) had fallen as litter prior to yuca root harvest. Leaf production losses due to herbivory by chewing insects accounted for less than 3 percent of the total leaf production (Table 5). Cutting gt g1. (1959) suggested that one reason conucos are abandoned is because of the buildup of insect pests. The percentage of yuca leaf blade eaten on Conuco Site 1 was 2.14 percent at year one and 2.20 percent at year two (Table 5). The low damage by chewing insects and the lack of a difference between years one and two indicates that this component of the herbivore population probably does not reduce the longevity of conucos in the San Carlos region. Stems comprised 51 percent of the dry weight yuca production over the two year period. Stem litter was estimated at 1.6 and 3.4 grams per m2 for years one and two, respectively. Yuca roots made up the remain- ing 30 percent of the total yuca production. The dry weight yuca root production was 146.5 and 100.6 grams per m2 in years one and two, res- pectively. A small fraction of the total root production (9.4 percent) was composed of fine roots too small to be processed for consumption. Table 6 compares yuca root production in Conuco Site 1 with yuca root production in two nearby conucos (i.e., Conuco Sites 2 and 3) at the time of the first yuca crop harvest. Appendix Table 2 contains the regression equation used to estimate root production in Conuco Sites 2 98 Table 4. Azsummary of crop and weed productivity (grams dry weight per m per year) in Conuco Site 1 for years one and two. Production Year 1 Year 2 YUCA: Leavefég) 92.8 82.1 Stems (c) 241.2 234.6 Roots (large) 146.5 100.6 Roots (fine) 14.5 11.1 Total 495.0 428.4 PINEAPPLE: Leaves and stems 36.2 74.0 Roots (d) 0.5 1.0 Fruits --- 18.6 Total 36.7 93.6 PLANTAIN: Leavefe) 0.3 0.5 Stems -1.8 1.9 Roots 0.6 1.2 Total -0.9 3.6 CASHEW: Leaves 1.0 2.1 Stems 0.4 1.1 Roots 0.2 0.6 Total 1.6 3.8 TOTAL CROP PRODUCTION 532.4 529.4 TOTAL EDIBLE HARVEST (Yield) 146.5 119.2 WEED PRODUCTION 30.0 67.9 TOTAL CROP AND WEED PRODUCTION 562.4 597.3 (a)Leaf weights include litter production for each crop. Leaf weights of yuca include an estimate due to loss by chewing insects. Leaf weights of pineapple include the leafy tops of the fruits. (b)All year one crop stem weights are corrected by subtracting the weight of the plantings used in establishing the plants. Stem weights of yuca include stem litter production. (C)Roots must be processed before eating. In root processing the bark and large fibers are removed. Ehe processed root yield is 135.3 and 92.8 grams dry weight per m for years one and two respectively. (d)Pineapple fruit weight includes an estimate of the standing crop fruit weight at two years and an estimate of fruit weight lost to frugivores. (e)High mortality of plantlet inputs caused a negative value. 99 Table 5. Values for the measurements taken in the estimation of yuca leaf herbivory by chewing insects. Year One Year Two Grams total leaf production per m2 91.2 78.7 Grams total leaf blade production per m2 74.8 64.5 Grams total leaf petiole production per m2 16.4 14.1 Percentage of leaf blade eaten 2.14 2.20 Grams leaf blade eaten per m2 1.6 1.4 100 Table 6. A comparison of yuca root production for the first crop in three San Carlos conucos. Values are based on the survey of 30 1 x 1 m plots in each site. Conuco Site 1 Conuco Site 2 Conuco Site 3 Yuca density per m2 1.5 1.8 0.8 Mean root dry weight per m (grams) 170.6 177.4 98.2 Mean root dry weight per plant (grams) 115.3 100.5 116.4 101 and 3. The density of plants was slightly higher in Conuco Site 2 than in Site 1, but the mean dry root weight per m2 was almost identical in these two sites. Conuco Site 3 had an unusually low density of yuca plants because the farmer had trouble securing sufficient stem stocks to plant. Nevertheless, the mean dry root weight per plant was estimated to be within two grams of that measured for Conuco Site 1. Although this is a limited sample, it appears that the first year yuca production value for Conuco Site 1 is similar to yuca production values for other conucos in the San Carlos area. The wet weight yuca production values in Conuco Site 1 were 4.31 metric tons (m.t.) and 2.81 m.t. per ha for years one and two, respec- tively. Doku (1969) reported that yuca yields range from 11 to 34 m.t. per ha in Ghana. The average yield of yuca in Brazil is 12.7 m.t. per ha (Normanha, 1970). Petriceks (1968) estimated the average yuca yield to be seven m.t. per ha for northern Venezuela. The world average yuca production has been recently estimated to be 8.4 m.t. per ha (Kay, 1973). Kay indicates, however, that production can drop to three m.t. per ha on poor soils. The extreme infertility of the San Carlos soils (Herrera gt fl” 1978) probably explains the low yuca production for this region. Pineapple production There were 0.21 pineapple plants per m2 in Conuco Site 1 at two years. Table 4 summarizes pineapple production. The pineapple plantlet input was 26.3 grams dry weight per m2. Leaf and stem production for the two year period was 85 percent of the total plant production. Leaf litter comprised 16 percent of the total leaf and stem production. 102 Estimated yearly root production was negligible, but these figures are undoubtedly underestimates of the true root production. Plant harvest- ing occurred during a dry spell and roots were poorly developed; I have observed that pineapple roots are well developed during rainy times of the year. Fifty-one percent of the pineapple plants had either produced fruit (M“ were bearing fruit at the end of two years. The average wet weight per fruit, without the leafy top, was 1028 grams; the average dry weight per fruit was 180 grams. The total pineapple fruit production, not including the leafy tops, was 18.6 grams dry weight per m2. Of this fruit production, an estimated 2.1 grams per m‘2 was lost to rodents. Plantain production Seven plantain plants were present in Conuco Site 1 at the end of year two, giving a density of one plant per 214 m2. Most of these individuals had produced additional stems vegetatively. Stem inputs were 2.71 grams dry weight per m2. Because 53 percent of the stem plantings died, the input biomass exceeded the estimated year one pro- duction. By two years some of the plantain plants had reached three meters in height, but none had begun flowering. The estimated produc- tion of plantains for year two was 3.6 grams per m2. Cashew production There were nine cashew plants present in Conuco Site 1 at the end 2 of year two, or one plant per 167 m . Six individuals died in the first year. Plantlet inputs were 0.14 grams dry weight per m2. Production 2 was estimated at 1.6 and 3.8 grams dry weight per m for years one and two, respectively (Table 4). No fruits were produced during the two- year study period, although several plants were flowering at the end of 103 year two. Weed production Table 7 summarizes the weed biomass harvested at 10, 16, and 21 months in Conuco Site 1. Weeds comprised 5.6 percent of the total crop plant production in year one and 12.8 percent of the total crop plant 8 . . . Grasses and forbs increased ln biomass and production in year two. density after the first weeding, whereas secondary woody species, forest woody species, and lianas all declined in biomass and density with each successive weeding. Forbs and grasses are favored by weeding because they produce many wind dispersed seeds within a few months of coloniza- tion (i.e., their life cycles are shorter than the interval between weedings). In contrast to forbs and grasses, secondary woody species are weeded out before they can produce seeds locally. After several weedings the seed pool which gives rise to the secondary woody coloni- zers is exhausted, and the rate of long distance seed dispersal is apparently not adequate to allow continual regeneration of these groups. Total production The total dry weight production of crop plants was 532.4 grams per m2 for year one and 529.4 grams per m2 for year two (Table 4). Of this 2 was the yield of edible crop in year one production, 146.5 grams per m and 119.2 grams per m2 was edible crop yield in year two. Year one yield was composed exclusively of yuca roots. Yuca roots made up 84 8Year one weed production (Table 4) was calculated as the sum of the ten month weed harvest and one-third of the 10 to 16 month harvest; year two weed production was computed as the sum of the 21 month weed harvest, one-half of the 21 to 27 month weed harvest, and two-thirds of the 10 to 16 month weed harvest. 104 Table 7. The mean dry weight biomass of weedy species harvested from 27 permanent I x 1.5 m plots for the period 0-10 months, 11-16 months, and 17-21 months. Grams Dry Weight per m2 0-10 months 11-16 months 17-21 months Forbs 4.30 27.23 18.93 Grasses 0.41 9.32 8.74 Secondary woody species 5.83 3.22 1.42 Forest woody species 0.86 0.64 0.43 Lianas 4.40 2.11 0.61 TOTAL 15.80 42.52 30.13 105 percent of the year two yield, pineapple fruits accounting for the remaining 16 percent of the yield. There was a 19 percent decline in overall yield and a 32 percent decline in yuca yield between years one and two. A second year decline in yield is common in tropical agricul- ture. Tondeur (1956), working in the Belgian Congo, reported a 33 percent decline in yuca yield between the first and second crop. Popenoe (1957), in Guatemala, reported that maize yields of the second crop are often reduced to one-half those of the first crop. Among those factors thought to cause tropical farm site yields to decline over time are: 1) herbivores; 2) weeds; 3) erosion; 4) decom- position of the organic material covering the soil; and 5) deterioration of the physical and nutrient status of the soil. The impact of sucking insect herbivores was assessed at San Carlos using radioactive tracers (Montagnini, 1980). Her results indicate that the overall impact of this group is minor. I found no increase in the amount of yuca leaf production removed by chewing insects from year one to year two (Table 5). Hence, herbivory by insects does not appear to be an important causal factor in the second year decline of conuco yields in the San Carlos region. Several authors have associated weeds with declining yields (Joachim and Kandiah, 1944; Watters, 1958). Kang (cited in UNESCO, 1978), working in Nigeria, demonstrated experimentally that loss of soil fertility and weed competition contributed almost equally to the decline of maize yields over time, following clearing. In San Carlos, the presence of weeds did not appear to cause reduced cr0p yields (provided weeding was done regularly). Total weed production for the two year period was only 9 percent of total crop production. 106 There is little apparent erosion during the first year of conuco use in San Carlos because a root mat shields the soil surface from the impact of rain. By the second year, however, much of the root mat has decomposed and erosion is apparent. In some areas of two year old conucos two to three cm of soil are washed away, but one also sees many areas in these conucos where eroded soil has accumulated (e.g., on the upslope side of logs). Given the abundance of slash in conucos, it is probable that most eroded soil remains within the confines of the site. Combined with erosion effects are possible changes in soil surface organic matter, soil structure, and soil fertility. The root mat cover- ing the soil varies in thickness from 5 to 30 cm when conucos are first planted. This root mat represents a source of readily decomposable organic matter which supplies nutrients for crop growth. After two years most of this root mat has decomposed exposing the infertile soil. Declines in conuco productivity parallel root mat decomposition. In contrast, only slight changes in soil structure and soil fertility occur over the first two years of conuco use, and therefore these factors are not thought to be important in causing reduced yields in the San Carlos region (C. Jordan, personal communication). Hence, it appears that the loss of the root mat is the most important factor causing declining crop yields in the San Carlos region. Aside from supplying nutrients, the root mat protects the soil from erosion and inhibits the establishment of weeds (see Part III of this thesis). Succession Site Productivity The productivity of Succession Site 1 was determined through mea- surements of standing crop biomass, litter production, and leaf 107 herbivory by chewing insects. Table 8 summarizes the standing crop biomass of this site at the end of one and two years. The standing crop 2 biomass after two years was 1175.3 grams dry weight per m , an amount more than 17 times greater than the year one value of 65.7 grams per m2. All plant groups increased in biomass from year one to year two. The greatest increase was in the secondary woody plant grouping (i.e., Cecropia spp. and all other secondary woody plants) which comprised 33 percent of the total dry weight biomass after one year and 88 percent of the total dry weight biomass after two years. The percentage of the Succession Site 1 biomass in leaves and stems changed from year one to year two (Table 8). Leaves comprised 48 per- cent of the total year one harvested biomass but only 23 percent of the year two harvested biomass, whereas stems increased from 36 percent of the year one to 59 percent of the year two harvested biomass. Thus the leaf and stem components of the successional vegetation do not increase at uniform rates. Ewel (1971), working in second growth stands in Panama, reported similar trends. Table 9 compares the San Carlos standing crop biomass at one and two years with values reported in other lowland tropical succession studies. The year one San Carlos standing crop value is much lower than the one year values from Guatemala. However, the Guatemalan sites were agricultural fallows which had probably accumulated an abundant seed pool before abandonment. The standing crop biomass value reported by Jordan (1971) for an irradiated montane forest site in Puerto Rico is also greater at one year than the San Carlos value. The two year San Carlos standing crop biomass figure of 1175 grams dry weight per m2 is more similar to literature values (Table 9). 108 Table 8. Azsummary of the standing crop biomass (grams dry weight per m ) for Succession Site 1 at one and two years following the burn. Grams Dry Weight per m2 % of % of Plant Grouping Plant Part Year 1 Total Year 2 Total Forbs Leaves and stems 8.4 12.8 7.9 0.7 Roots 2.2 3.3 7.5 0.6 Grasses Leaves and stems 2.4 3.6 18.6 1.6 Roots 0.5 0.8 4.2 0.4 Forest trees Leaves 5.7 8.6 18.8 1.6 from sprouts Stems 8.4 12.8 23.9 2.0 Forest trees Leaves 0.5 0.8 1.1 0.1 from seed Stems 0.3 0.5 1.3 0.1 Roots 0.0 0.0 0.2 .0 Secondary trees, Leaves 8.7 13.2 172.6 14.7 Cecropia spp. Stems 4.0 6.1 563.8 48.0 Roots 2.2 3.3 177.9 15.1 Secondary woody Leaves 3.1 4.7 34.2 2.9 trees, other Stems 1.9 2.9 75.8 6.4 Roots 1.5 2.3 14.1 1.2 Lianas Leaves 8.8 13.4 19.3 1.6 Stems 3.9 5.9 28.9 2.5 Roots 3.2 4.9 5.2 0.4 Total Biomass 65.7 1175.3 Leaves, % of total 48.0 23.0 Stems, % of total 36.0 59.0 Roots, % of total 15.0 18.0 109 Table 9. The standing crop biomass of one and two year old succession sites at San Carlos de Rio Negro and elsewhere in the lowland tropics. Grams Dry Weight per m2 Site Location One Year Two Year San Carlos, Site 1 66 1175 Columbia, N.W.(a) (Golley gt gl., 1975) -- 1585 Guatemala (a) (Snedaker, 1970) 836 1419 Guatemala (Tergas, 1965) 1165 -- Panama, Site 1 -- 1602 Site 2 -- 1460 (Ewel, 1971) Puerto Rico (Jordan, 1971) 242 761 Africa, Congo (Bartholomew, 1953) -- 1786 (a)0nly above ground biomass was harvested in these studies. 110 Dry weight litter production in Succession Site 1 was 39.5 grams per m2 in year one and 317.2 grams per m2 in year two. Leaves comprised 9891 percent (H: the total litter production over the two year study period; the remaining 1.9 percent was stem litter. Ewel (1968) esti- mated litter production at 460 grams dry weight per m2 for the first year on a post-agricultural site in Guatemala, and Kellman (1970), on a post-agricultural site 'hl the Philippines, estimated litter production at 204 grams dry weight per m2. Reports of litter production for the second year of succession are not available in the literature to my knowledge. An estimated 1.7 and 10.3 grams dry weight per m2 of leaf litter production were lost to chewing insects in years one and two, respec- tively (Appendix Table 3), while a calculated 1.7 (year one) and 9.3 2 of the harvested standing crop leaf (year two) grams dry weight per m biomass were lost to chewing insects. Adding the leaf herbivory losses from the leaf litter and leaf standing crop fractions, an estimated 3.4 2 were removed from Succession Site 1 by chewing grams dry weight per m insects in year one and 19.6 grams dry weight per m2 were removed in year two. These estimates considered separately represent 4.8 percent and 3.3 percent of the total leaf production for years one and two, respectively. None of the four plant groupings had more than eight percent of their total leaf area eaten in either year (Appendix Table 3), but four-fold differences in percentage eaten occurred between some of the plant groupings indicating that specific plant life forms may be fed on differentially by chewing insects. Summing the yearly estimates of standing cropl biomass, litter production, and herbivory by chewing insects gave an estimate of net 111 production (Table 10). Year one net production was 108.6 grams dry weight per m2; production for year two was 1446.4 grams dry weight per m2 or 3.96 grams dry weight per m2 per day. These productivity values are underestimates because root turnover of plants establishing from seed, root production of sprouts, leaf and stem production losses to sucking insects, and root production losses to herbivores were not measured. The measurement of these parameters poses many problems and has seldom been attempted. The ratio of net primary productivity to ecosystem respiration is probably below one in both years one and two because of the decom- position of large amounts of partially burned logs and stumps from the forest previously occupying the site. By about the tenth year most of this wood has decomposed and the site begins to accumulate biomass. There have been no other studies in the lowland tropics, to my knowledge, which estimate the productivity of secondary vegetation after a cutting and burning treatment. Early succession productivity values are available from the temperate zone, but large differences in the length of the growing season and the nature of the pre-fallow distur- bance make interlatitudinal comparisons difficult to interpret. Odum (1960) investigated early successional vegetation in Georgia, USA, and found that production in the first year was about 500 grams dry weight per m‘2 and declined in subsequent years to about 300 grams dry weight per m2. He attributed the high production value for year one to the presence of residual fertilizer. In contrast to Odum's findings, pro- duction in San Carlos was much higher in the second year of succession than in the first year. Several factors are important in explaining the low year one San Carlos production value: 1) the burn kills most tree 112 Table 10. A summary of production for years one and two in Succession Site 1. Production (grams drypweightpper m2) Year One Year Two Standing living biomass 65.7 1175.3 Litter production 39.5 317.2 Herbivory 1055(6) 3.4 19.6 Production, each (b) year separately 108 6 1446.4 Production, both (c) years together 1555.0 (3)0nly herbivory by chewing insects was estimated. (b)Production for year two is equal to total production for both years minus year one production. (C)Production for both years together is equal to: year two standing biomass + litter production for both years + herbivory losses for both years. 113 sprouts, thereby eliminating sprouting as an avenue for quick recovery; 2) several relatively dry months usually follow burning making it diffi- cult for seedlings to establish; 3) seed pools are significantly de- pleted tn/ the burn; and 4) the burn leaves behind a mosaic of surface types, many of which are not suitable for seedling establishment. Each of these factors is discussed in detail in Part III of this thesis. Table 11 compares the production of Succession Site 1 vegetation with Conuco Site 1 crop plant production during years one and two. The conuco was far more productive than the succession site in year one (532 vs. 109 grams dry weight per m2). However, in the second year this trend was reversed, the succession site producing 1446 grams dry weight per m2 vs. 529 grams dry weight per m2 for the conuco. In fact, the succession site production for year two alone was greater than the combined years one and two production in the conuco. A consideration of leaf area index (LAI), leaf production, and leaf retention time helps to explain the difference in productivity between the conuco and the succession site (Table 12). Crop LAI during most of the two year study period was well below one. The yuca leaf area index at one year (i.e., just prior to harvest) was 0.62. This value is probably close to the maximum leaf area cover attainable by yuca on this site. Leaf area index for the other three crop plants was estimated at 0.083 at year one and 0.38 at year two. Leaf area index in the succes- sion site was also less than one during the first year, but increased to 2.79 by the end of year two, giving the secondary vegetation a greater leaf surface available for photosynthesis than the conuco crops during much of the second year. In accord with this, the total leaf production in the succession site over the two-year period was 648 grams dry weight 114 Table 11. A comparison of year one and year two production between Succession Site 1 and Conuco Site 1. Production (grams dry weight per m2) Succession Site 1 Conuco Site 1 Year 1 Year 2 Year 1 Year 2 Production, each year separately 109 1446 532 529 Production, both years together 1555 1061 115 Table 12. A comparison of leaf cover and leaf production in Succession Site 1 and Conuco Site 1 over the two year study period. Secondary Conuco Crops Vegetation Leaf area index 1.0(3) 2.79 Total leaf production over years one and twg (grams dry weight per m ), all crops 289.1 648 Leaf production as a % of total production, all crops 27 42 % of total leaf production (b) measured as standing crop leaves 28 45 (a)This represents the maximum LAI during the two year study period. (b)0nly yuca is considered in this estimate. 116 per m2 \HL. 289.1 grams dry weight per m2 for the conuco crops. These values represent 42 percent of the total primary production in the succession site and 27 percent of the total crop plant production in the conuco site. In addition to producing more leaves, the succession site vegetation retained its leaves longer. Forty-five percent of the total succession leaf production was measured as standing crop leaves, while only 28 percent of the yuca crop leaf production represented standing crop leaves. Hence, three interrelated factors explain, at least in part, why the secondary vegetation was more productive than the conuco crops over the two year study period. These factors are: 1) high LAI values in the succession site in year two compared to the conuco; 2) a greater percentage of total production devoted to leaves in the succes- sion site; and 3) longer leaf retention time for the secondary vegeta- tion. Prospects for Increased Yields The results of this comparative productivity study suggest that agricultural production in the San Carlos region could be increased. The use of legumes, dry mulching, and crop breeding in a planting scheme which resembles the natural succession in structure and life form dom- inance offers, in my opinion, the best hope for increasing yields. Legumes are little used in the San Carlos area. However, the cowpea, lima bean, and winged bean all do well in the wet tropics (Greenland, 1975) and their presence supplies nitrogen to the soil. Experiments done at Turrialba, Costa Rica (North Carolina State Univer- sity, 1973) on the intercropping of beans with yuca showed that yuca presence did not negatively effect bean yields. Similar experiments 117 should be conducted in the San Carlos area. The use of dry mulching, likewise, offers promise for the San Carlos region. Rockwood and Lal (1974), working in Nigeria, found that using crop residues as mulch on traditional no-tillage farm plots in- creased maize yields by as much as 50 percent when compared with plowed plots. The use of a mulch of harvested yuca stems and forest tree leaves could reduce erosion, maintain soil aggregates, conserve soil nutrients, and reduce the establishment of weeds. Crop breeding programs also could have a substantial effect on agricultural productivity in the San Carlos area. Magoon (1970) re- ported that inter-varietal hybridization greatly increased yuca yields in India. Many varieties of yuca are present in the San Carlos area, and some farmers have a good knowledge of varietal differences. Hence, the potential for cooperative work between breeders and knowledgeable farmers is good. The natural succession provides a model for the construction of sustained-yield agro-ecosystems which use legumes, dry mulching, and specially bred crop plants. Sakae Oti, working in Brazil, has designed a farming scheme where all plants present from the earliest successional stage to the well developed forest are of economic value (described in Sioli, 1973). Ewel (1979) is currently developing this concept in Costa Rica by building agro-ecosystems which imitate the structure and function of natural regrowth. The life form dominance patterns of early successional vegetation in San Carlos could, and to some extent already does, serve as a model for agriculture. During the first year of suc- cession, forbs, grasses, secondary woody species, lianas, and primary forest trees are present in one stratum and each contributes about 118 equally to the standing crop biomass of the site (Table 8). To parallel this, conucos can first be planted with forb species (e.g., beans, pineapple), grass species (e.g., upland rice, corn, sugar cane), secon- dary woody species (e.g., yuca, cashew, papaya), liana species (e.g., yams, sweet potatoes), and primary forest tree species (e.g., brazil nuts, rosewood). The optimal density and pattern of plantings can be determined experimentally. By year two the fast growing woody crops form one or more additional strata above the ground layer just as secon- dary woody species do in two year old succession sites. Later, the primary forest tree analogs can assume dominance. This approach to tropical agriculture, by maintaining a continual crop cover, reduces the amount of nutrient loss by leaching, while providing cr0p yields over extended periods of time. The theme of efficiency underlies both the energetic and productiv- ity results presented in this study. The San Carlos conuco farmers receive an energetic return on work energy inputs of 13.9 to one; farm- ing for one hour a day supplies their food energy needs. Introduced farming systems which do not achieve this ratio of return are inferior in energetic efficiency to the native system. Plant nutrients represent a finite resource and agronomic systems should, likewise, be evaluated with respect to the efficiency of their nutrient use (i.e., the degree to which the nutrient capital present within a system is retained within the boundaries of that system over time). In tropical forests, most of the nutrient capital is in biomass (Kira and Shidei, 1967). If large areas of forest are cut, burned, and used intensively for many years, most of the nutrients will be lost. Many hundreds of years may be required to rebuild this nutrient capital. 119 In contrast, the natural successional vegetation conserves available nutrients and re-establishes mechanisms to cycle them (Ewel, 1979). Agricultural systems which mimic the natural successional systems in structure and life form dominance may be able to quickly take up a large portion of the nutrients from the cut forest and cycle these nutrients internally. A land ethic based on the efficiency of resource use, whether the resource be energy or nutrients, will ultimately improve the quality of life in the San Carlos area, as well as assure the conservation of rain forest resources in this oligotrophic region of the Amazon Basin. 120 SUMMARY An analysis of the energetics of conuco agriculture (i.e., shifting cultivation) and a comparison of productivity between conuco crops and secondary vegetation were undertaken in the Rio Negro region of the Amazon basin. Although relatively unproductive, conuco agriculture is much more energy efficient than mechanized agriculture. The ratio (H: energy output in the form of processed food to energy input in the form of human labor was 13.9 to one for a two crop conuco-use cycle. Slightly more energy was devoted to field activities than to processing activi- ties. The field activities demanding the most energy were harvesting and weeding; in combination they accounted for 30 percent of the total conuco energy input. The peeling and grating of yuca roots were the most energy consuming processing activities accounting for an estimated 24 percent of the total conuco energy input. The return per unit kilo- calorie work input decreased for the second cropping period because yields declined and increasing time was spent in weeding. Conucos in the San Carlos area are planted predominantly in yuca (Manihot esculenta, Crantz.). Yuca wet weight production was 4.31 m.t. per ha in year one and declined to 2.81 m.t. per ha in the second year of conuco use. The infertile soils of the region are probably the major reason for these low yields. No increase in damage by chewing insects was noted from year one to year two. The standing crop biomass of the one year old San Carlos succession 121 site vegetation was much lower than other reported values; however, biomass increased greatly in the second year and was similar to litera- ture values for two year old tropical regrowth sites. Leaf herbivory of the secondary vegetation by chewing insects was low in both years, but may differ among plant life forms. The conuco crops were more productive than the adjacent succession site vegetation during the first year after cutting and burning, but in the second year the secondary vegetation was more than two times as productive as the conuco crops. This year two difference in production is probably (hue to the higher leaf area index values, the greater re- source allocation to leaves, and the longer leaf retention time of the secondary vegetation. Attempts to increase crop yields should seek both to preserve the extremely favorable food energy output to work energy input ratios characteristic of the native farming system and to minimize the loss of accumulated nutrient capital present on the site after burning. Farm practices which imitate the structure and life form dominance patterns characteristic of early successional vegetation merit atten- tion. Planting the crop equivalents of forbs, grasses, secondary woody species, lianas, and primary farest species in a man-devised agricul- tural succession, combined with an increased emphasis on legumes, dry mulching, and crop breeding may represent the best hope for raising agricultural yields in this region. LIST OF REFERENCES 122 LIST OF REFERENCES Bartholomew, W. V., J. Meyer, and H. Laudelout. 1953. Mineral nutrient immobilization under forest and grass fallow in the Yamgambi (Belgian Congo) region. Publication Institut National Etude Agron- omy Congo Belge, Serie Science Nombre 57. Carneiro, R. L. 1961. Slash and burn cultivation among the Kinkuru and its implications for cultural development in the Amazon basin. Pages 47-68 in J. Wilbert, ed. The evolution of horticultural systems in native South America. Anthropologica, Caracas. Carpenter, T. M. 1948. Factors and formulas for computing respiratory exchange and biological transformations of energy. Carnegie Insti- tute, New York. Clark, C., and it Haswell. 1970. The economics of subsistence agri- culture. MacMillan, London, 4th edition. 245 pp. Conklin, H. C. 1963. The study of shifting cultivation. Studies and monographs, VI. Union Panamericana, Washington, D. C. Cutting, C. V., R. A. Wood, P. Brown, and H. B. Ambrose. 1959. Assess- ment of fertility status and the maintenance of productivity of soils in Nyasaland. Paper no. 69. Third Interafrican Soils Con- ference, Dalaba. Doku, E. V. 1969. Cassava in Ghana. Ghana University Press. 57 pp. Ewel, J. 1968. Dynamics of litter accumulation under forest succession in eastern Guatemala lowlands. M.S. thesis. University of Florida, Gainesville. 85 pp. Ewel, J. 1971. Biomass changes in early tropical succession. Turrialba 21:110-112. Ewel, J. 1979. Crossroads for ‘tropical biology. Mosaic 10:10-18. Golley, F. 8., J. Ewel, and G. I. Child. 1975. Vegetation biomass of five ecosystems irl northwestern Columbia. Tropical Ecology 17:16-22. Gomez-Pompa, A. and others. 1973. Recovery of tropical ecosystems. Pages 113-138 in E. G. Farnworth and F. B. Golley eds. Fragile Ecosystems. spFinger Verlag, New York. Greenland, D. J. 1975. Bringing the green revolution to the shifting cultivator. Science 190:841-844. 123 Gross, D. R., G. Eiten, N. M. Flowers, F. M. Levi, M. L. Ritter, D. W. Werner. 1979. Ecology and acculturation among native peo- ples of central Brazil. Science 30:1043-1050. Herrera, R., C. F. Jordan, H. Klinge, and E. Medina. 1978. Amazon ecosystems: their structure and functioning with particular empha- sis on nutrients. Interciencia 3:223-231. Joachim, A. W. R., and S. Kandiah. 1944. The effect of shifting (Chena) cultivation and subsequent regeneration of vegetation on soil composition and structure. Tropical Agriculturist. 104:3-11. As cited in Nye and Greenland, 1960. Jolly, A. L., A. S. de Lozada, and H. C. Conklin. 1961. Las fincas pequenas. Pages 114-171 12 Corporacion Venezolana de Guyana. Reconocimiento Agropecuario Forestal del Oriente de la Guazana Venezolana. Consejo de Bienestar Rural, Caracas. Jordan, C. F. 1971. Productivity of a tropical rain forest and its relation to a world pattern of energy shortage. Journal of Ecology 59:127-142. Kay, 0. E. 1973. Root crops. The Tropical Products Institute Crop and Product Digest. No. 2. Kellman, M. C. 1970. Secondary plant succession in tropical montane Mindanao. Publication BG/Z. Australian National University, Canberra, Australia. 174 pp. Kira, T. and T. Shidei. 1967. Primary production and turnover of organic matter in different forest ecosystems of the western Paci- fic. Japanese Journal of Ecology 17:70-87. Leach, E. R. 1959. Some economic advantages of shifting cultivation. Pages 64-65 jg Proceedings Ninth Pacific Science Congress of the Pacific Science Association. Bangkok, Thailand. Magoon, M. L. 1970. Problems and prospects in the genetic improvement of cassava in India. Pages 58-61 jg Proceedings Second Inter- national Symposium on Tropical Root Crops. Honolulu, Hawaii. Montagnini, F. 1980. Influencia de los insectos sobre el ciclo de nutrientes y productividad en los ecosistemas de conucos de San Carlos de Rio Negro, Territorio Federal Amazonas. Masters thesis. Instituto Venezolano de Investigaciones Cientificas, Caracas, Venezuela. Montaldo, A. 1977. Cultivo de raices y tuberculos tropicales. Insti- tuto Interamericano de Ciencias Agricolas OEA. San Jose, Costa Rica. 26 pp. 124 Nicholson, 1. A., I. S. Paterson, and A. Currie. 1970. A study of vegetational dynamics: selection by sheep and cattle in Nardus pasture. Pages 73-98 E A. Watson, ed. Animal populations in relation to their food resources. Blackwell, London. Norman, M. J. T. 1978. Energy inputs and outputs of subsistence crap- ping systems in the tropics. Agra-Ecosystems 4:355-366. Normanha, E. S. 1970. General aspects of cassava root production in Brazil. Pages 61-63 12 Proceedings Second International Symposium on Tropical Root Crops. Honolulu, Hawaii. North Carolina State University. 1973. Annual report: Agronomic economic research on tropical soils. Soil Science Department. Raleigh, N.C. 190 pp. Nye, P. H., and D. J. Greenland. 1960. The soil under shifting culti- vation. 156 pp. Commonwealth Bureau Soil Science Technical Com- munication No. 51. Harpenden. Odum, E. 1960. Organic production and turnover in old field succes- sion. Ecology 40:34-49. Petriceks, J. 1968. Shifting cultivation in Venezuela. Ph.D. disser- tation. SUNY College of Forestry, Syracuse. 334 pp. Pimentel, D., L. E. Hurd, A. C. Bellotti, M. J. Forster, I. N. Oka, 0. D. Sholes, and R. J. Whitman. 1973. Food production and the energy crisis. Science 182:443-449. Popenoe, H. 1957. The influence of the shifting cultivation cycle on soil properties in Central America. Pages 72-77 _i_g Proceedings North Pacific Science Congress, Bangkok. Rappaport, R. A. 1971. The flow of energy in an agricultural society. Scientific American 225:116-132. Raunkiaer, C. 1934. The life-forms of plants and statistical plant geography. Oxford, London. Rhoades, D. F. and R. G. Cates. 1976. Toward a general theory of plant anti-herbivore chemistry. in J. Wallace, ed. Recent advances in phytochemistry, Vol. 10, B'iBChemical interactions between plants and insects. Plenum, New York. Rockwood, W. G., and R. Lal. 1974. Mulch tillage: a technique for soil and water conservation in the tropics. Span 17:77-79. Rose, M. S. 1938. The f0undations of nutrition. The MacMillan Co., New York. Table IX, pp. 606-607. 125 Sanchez, P., C. Seubert, E. Tyler, C. Valverde, C. Lopez, M. Nurena, and M. Wade. 1974. Investigaciones en manejo de suelos tropicales en Yurimaguas, selva baja del Peru. Paper presented at Seminario de Sistemas de Agricultura 'Tropical. Lima, Peru. June 1-8, 1974. Sioli, H. 1973. Recent human activities in the Brazilian Amazon region and their ecological effects. Pages 321-334 i_n B. J. Meggers gt g1. eds. Tropical forest ecosystems in Africa and South America: a comparative review. Smithsonian Institution. Washington, D. C. 350 pp. Snedaker, S. 1970. Ecological studies on tropical moist forest succes- sion in eastern lowland Guatemala. Ph.D. dissertation. University of Florida, Gainesville. 131 pp. Tergas, L. E. 1965. Correlation of nutrient availability in soil and uptake by native vegetation. M.S. thesis. University of Florida, Gainesville. 64 pp. Tondeur, M. G. 1956. L'agriculture nomade au Congo Belge. Pages 15-108 j_n Organisation de Nations Unies pour l'Alimentation et l'Agriculture. L'Agriculture Nomade. Congo Belge, Cote-d'ivoire. Vol. 1. Rome. As cited in Nye and Greenland, 1960. UNESCO, UNEP, FAO. 1978. Tropical forest ecosystems. Unipub. New York. Watters, R. F. 1958. Cultivation in Old Samoa. Economic Geography 34:338-351. Watters, R. F. 1960. The nature of shifting cultivation. Pacific Viewpoint 1:59-99. Watters, R. F. 1971. Shifting cultivation in Latin America. FAO Forestry development paper no. 17. FAO, Rome. 305 pp. APPENDIX TABLES 126 Appendix Table 1. Human labor input and yuca output for crops one and two of a two crop conuco cycle. Energy Inputs and Outputs Crop One of Crop Two of Two Crop Cycle Two Crop Cycle Activity (kcal) (kcal) Cutting and burning 70,049 0 Field work 88,972 127,468 Travel (for field work) 19,132 17,529 Processing 130,060 78,285 Travel (for processing) 46,800 28,169 Total energy inputs 355,013 251,451 Yuca production energy output 5,247,668 3,158,677 Energy outzEnergy in 14.8:1 12.6:1 127 OO. H.e u xNH. n O .paz wees .O> “Lowe; x waO .OOO mwmecomu em. H.OeN + xmm. u z .Hmz seam .O> pcawmc x waO .Oam mwmecumu NO. 0.000 + xO.NN n O .OO3 OOOOOO-3OHOO .O> OOOHOO x NOOO OOOOOOHO OO. N.OO + XO.N n O .OO: waeH .Os osOOOO x NOOO swaosape OO. O.HNOH + XO.Oe n O .OO3 seem .Os oOOOOs x NOOO OOOOOOHO ON. 0.0H + xOO. n z .Hmz wmeH .O> tems:: wmmO mu:> OO. O.NH + xew. n O .umz seam .m> cpmcep seam mua> OO. mm + xHH. n x .Hmz wees .O> cuocmH seam x waO muz> Ne eewpmeem OwcmcewamHmm ucmHO .OOO mwmecomu New ecm Oucmwa Oecu Lew OOmEewa Oece newecmpm we cowpmewpmm esp New new: cewpmcwsceume we mpcmquwweeo ecm Ocewumaee cewmmmcamm .N OHOON xwesoaae 128 OuomOcH OcHzmce >9 um>eemc NE\NO ON. On OO. O 1 New. \mO. e :eumec: pme Om: Hmcp H we cewpomcw u new. 0 mOO. u O.H mce cmmz cw cowpeaeeta NepawH Ozeeomngm; emumswumm u NE\NO mO. en O. mm x NHH. O .Opommcw chzmzu ea OOeH cowpezeeea LOHHOH wmeH Ozememetmc Hemeceomm wee cmmz esp Oewpzaeeo cw new: OOOHO cowpmHonmo ecu eemeuee m>mc H .mems mam: Ompmewumm omega 3e; eumNHOOHHw ehHmv 0.0H N.H m.O H.O O.N N.e 0.0 0.0 mmemwO N.N 0.0 0.0 H.O 0.0 O.H Omega amuse; O.O O.H O.N O.O O.NO N.ON 38: 1:25:88 N.H 0.0 0.0 0.0 O.e N.HH maemumecm: NONmeceumm N Nmm> H Nmm> N Nmm> H Nmm> N Nmm> H Nmm> chOOeNO ucmHO cmumm NE Lem OcOOsz mmc< wmmO HHmO Nmppwg Omar; OsON sea :38 OO ON :38 NO ON .H mqu cewmmmoeam cw cewaozeeca wmeH we :ewpumcw NeppwH wmmH esp secw OmOOeH Hee>wecmc uemmcw chzmce chpmswpme cw mom: meme ecu we HgmsEOO .m mHemH xwecmqa< PART III EARLY SUCCESSION STUDIES INTRODUCTION The lack of detailed succession studies in the lowland trapics and the importance of such studies was pointed out by Richards (1952) and more recently by Whitmore (1975). Many tropical succession studies have focused on species listings and general descriptions of different successional stages (Kenoyer, 1929; Stevenson, 1929; Holdridge gt gl., 1947; Fosberg, 1950; Grieg-Smith, 1952; Allen, 1956). Several recent investigators (Budowski, 1961; Blum, 1968; Kellman, 1970; Snedaker, 1970) have attempted to piece together succession in tropical lowland areas by looking at chferent-aged stands at one point in time. This approach assumes that the age of a site is the most important determinant of site floristics and structure, but this may not be the case. For example, Kellman (1970), working in the lNfilippines, found that above ground biomass varied by as much as five times in successional plots of the same age but with different histories of agricultural use prior to abandonment. A second approach to the study of succession is to initiate specific types of disturbances and then to study the subsequent recovery through time using permanent plots. Symington (1933), working in Malaysia, was the first to use this experimental-type approach in studies of tropical succession. Using this method, the investigator can evaluate the effects of different factors (i.e., the treatments used) on succession. The importance of such factors as the type of disturbance, the size of the disturbance, the duration of the disturbance, and the seasonal occurrence 130 131 of disturbance can be evaluated using this approach. Other factors such as the flora in the vicinity of the disturbance, the species specific dispersal characteristics of this flora, and the type and abundance of seed dispersal vectors available in the vicinity of the disturbance are more difficult to handle experimentally but certainly merit study as their influence on succession may be great. In this study I have combined a structural and floristic descrip- tion of early succession after cutting and burning in the San Carlos de Rio Negro region of the Amazon Basin with experiments designed to elucidate the factors which control this succession. To explain the observed succession, answers to seven questions were sought. The questions were: 1. 00 cut forest trees have the ability to sprout? 2. What effect does burning after cutting have on sprouts and on seed pools? 3. Do the first secondary colonizers after the burn originate from wind dispersed seeds or seed pool seeds? 4. What are the dispersal strategies employed by secondary species? 5. Why are population densities so low during the first months of succession? 6. Why ck) population densities increase rapidly after about the fourth month of succession? 7. Can the spatial distribution of Species colonizing early succession sites be related to any physical factors? Detailed field observations and experiments were used to answer each of these questions. 132 I specifically chose to investigate the succession after forest cutting and burning because almost all man-induced disturbances in the Amazon Basin involve forest cutting and burning. 133 METHODS Three tierra firme sites, designated as Succession Sites 1, 2, and 3 were used in this study. Figure 1 (General Introduction) gives the locations of these sites relative to the town of San Carlos de Rio Negro. The vegetation on these three sites prior to cutting was mature forest. Forest cutting in the Rio Negro region usually occurs at the beginning of the relatively dry months; burning occurs about three months after cutting. Succession Sites 1 and 2 were cut in September 1976 and burned in December 1976; Succession Site 3 was cut in August 1978 and burned in November 1978. The three sites were similar in size (0.25 ha) and slope (less than 5°). Vegetation Development Structural changes in the early successional vegetation in Suc- cession Site 1 were depicted using profile diagrams. All plants present in a 1 x 20 m permanent transect were mapped, measured, and drawn to scale at 3, 10, and 22 months following the burning of this site. Changes in plant density, species number, and species dominance were studied in Succession Sites 1 and 2 using 27 l x 1.5 m randomly located plots in each site. Each plot was surveyed at 4, 10, 16, and 22 months. In the first survey period, all established plants (defined to be greater than 5 cm tall) were identified (by scientific name where possible), measured for height, mapped, and marked with plastic tooth- picks. At each subsequent survey period all new establishments were identified, measured, and marked and all mortality of previously marked 134 individuals was recorded. Vegetation Development Explained Sprouting of cut trees The effect of a cutting treatment on succession was investigated in 2 permanent plots. Nine plots were Succession Site 3 using 19 one m located in areas of heavy slash cover; five plots in areas of moderate slash cover; and the remaining five plots in areas of light slash cover. This partitioning was in accord with the nature of slash distribution. One week before the site was burned all individuals that had established since the forest was cut (i.e., three months previously) were counted, classified as sprouts or seedlings, and tagged in each of these 19 plots. Effects of burning on sprouts and seed pools Peak burn temperatures were measured in each of the 19 one m2 permanent vegetation study plots in Succession Site 3 using 10 temper- ature sensitive lacquers (Tempilaq, Omega Engineering, Inc., Stamford, Connecticut), having distinct melting points ranging from 38 to 593°C. These lacquers were painted on asbestos boards. Each board was set vertically with approximately 10 cm in the soil and 20 cm protruding above the soil surface. The effect of differences in peak burn temperature on post-burn sprout and seedling establishment was investigated by resurveying the 19 one m2 permanent vegetation study plots in Succession Site 3 three months after the burn. All pre-burn sprouts and seedlings (i.e., marked) and new sprouts and seedlings in these plots were recorded. 135 The effect of burning on seed pools in Site 3 was evaluated by comparing the number of seed germinations in soil samples taken from the burned site and the non-burned mature forest adjacent to the site. Each sample was 20 x 20 x 5 cm deep. One half (i.e., 10) of the burn site samples were located in areas which had experienced high burn tempera- tures; the other half were located in areas which were only lightly burned. All samples were spread on vermiculite or sterilized sand and set 1.7 m above the ground in a sunny location on large tables. Trays were watered daily. Contamination by wind and animal-borne seeds was checked through surveys (Hi five trays, containing soil which had been oven dried at 250°C for five hours, placed among the sample trays under study. Trays were surveyed for new germinations three times a week for two months, after which time the soil in each tray was turned over and mixed. Surveys continued until no further' germinations ‘were found. Seedlings were identified whenever possible. Seed source of the first colonizers after burning To evaluate the relative contribution of the seed pool surviving the burn vs. post-burn seed dispersal in the establishment of plant cover, six sets of 2 x 2 m paired plots were established in Succession Site 3 immediately after the burn. Methyl-bromide was applied to one plot ‘hl each pair to kill the seed pool. All plots were covered with moderate amounts of slash to simulate the post-burn slash cover of the area. A survey of plant density and composition was made in all plots four months after the treatments were initiated. The degree of differ- ence ‘hl recovery between plots not receiving the methyl-bromide treat- ment and those receiving it was regarded as a measure of the importance of the seed pool in recovery. To check that methyl-bromide effectively 136 exterminated the seed pool, potted seeds of seven of the more common succession species, representing all successional life forms, were subjected ix) the methyl-bromide treatment. Seeds of these same seven species, not exposed to methyl-bromide, were planted in pots containing methyl-bromide treated soil to verify that the treated soil would not negativehy affect the germination and establishment of post-treatment dispersed seeds. Dispersal strategies of secondary species The time necessary for plants to colonize cut and burn succession sites aunt the species composition of these sites is influenced by the dispersal strategies of the surrounding flora. These strategies were studied through observations of adaptations for mechanical dispersal in fruits and seeds and through vector capture studies. Bats were mist netted on two consecutive evenings using seven nets in the vicinity of Succession Site 1. Birds were mist netted on two consecutive mornings in the same location where bat netting occurred. Seeds in the feces of the bat and bird species caught in the nets were germinated and identi- fied. Low population densities duringthe first months of succession The effect of short droughts (defined as two or more days without rain) on seed germination and seedling establishment was studied in Succession Site 1 using 17 randomly located 1 x 1.5 m plots. All new seedlings in these 17 plots were marked at approximately monthly inter- vals using plastic, color-coded toothpicks beginning 10 months after the burning of this site and continuing until the twenty-third month of succession. Attempts were made to correlate monthly rainfall totals 137 with the number of new germinations present in each monthly survey and with seedling mortality and seedling establishment. Rapid population increase after the fourth month of succession Germination and establishment of seeds produced within the succes- sion site (rather than of seeds dispersed from afar) was thought to account for the rapid population increase after the fourth month of succession. To evaluate the magnitude of short distance (i.e., within- site) seed dispersal vs. long distance (i.e., intersite) seed dispersal, 30 seed traps, each with an area of 154 cm2, were placed in Succession Site 1 sixteen months after the burning of this site. Succession Site 1 had a complete cover of vegetation at that time. The traps were set at 2 m intervals along two transects. Thirty seed traps were simulta- neously placed in a nearby freshly burned site (i.e., no vegetation present). The difference in seed numbers caught in seed traps in these two sites provided a measure of the relative importance of local vs. long distance seed dispersal. Microhabitat preferences of secondary species Successful establishment from seed frequently depends on the phys- ical characteristics of the microhabitat in which the seed lands. In the San Carlos region, root mat, charred wood, and bare soil are the princi- pal surfaces available for colonization after burning. These three surface types are sometimes in combination with slash giving a total of six post-burn microhabitats (i.e., root mat, charred wood, bare soil, root mat with slash, charred wood with slash, and bare soil with slash). The percentage cover of the root mat, charred wood, and bare soil sur- faces and of slash was estimated in each of the 27 permanent vegetation 138 study plots located in Succession Site I. Preferential colonization of these microhabitat types by the natural flora was measured by construct- ing the six microhabitats in blocks in Succession Site 2. Each treat- ment block was located non-randomly in areas with root mat present. The root mat was removed for the two exposed soil and two charred-wood treatments. Freshly made charcoal (charcoal was used instead of charred wood to avoid seed contamination) was crushed to roughly 1 cm square chunks and placed in a 2 cm thick layer in the charred wood (hereafter referred to as charcoal) treatment. Seven 30 cm long by 3 cm diameter poles were placed in each of the three treatments receiving slash in each block. Fifteen blocks were constructed in this manner with each treatment type 30 x 30 cm in size. Two months after initiating this experiment, the height and identification of all establishments in each treatment type were recorded. Results were analyzed using a two-way analysis of variance. To determine microhabitat preferences at the species level, seeds of seven of the more common secondary species (one grass, two forb, and four secondary woody species) were broadcast on additional sets of blocks containing these six nficrohabitat types. Five to seven blocks were used for each of the seven speciestested. Approximately 300 seeds were placed in each treatment type within each block. Seeds were counted or measured volumetrically to ensure that uniform numbers were broadcast. The height and number of establishments in each treatment were recorded at roughly one month intervals for two to four months after the experiments were initiated. Physical characteristics which often affect seedling establishment were also investigated in the six treatment types. Measurements of 139 surface and soil temperature were taken at hourly intervals from 8:00 AM to 5:00 PM in each microhabitat type on three rainless days. Soil evaporation was measured by removing plugs of soil (with adjacent root mat or charcoal when present) from each treatment type and placing each plug in an aluminum container. The soil in each can was wetted to field capacity, weighed, and replaced in the field. At dusk the cans were collected and reweighed. Subtracting evening weight from morning weight gave soil water evaporation for the day for each of the microhabitats. This procedure was repeated on three consecutive days. 140 RESULTS AND DISCUSSION Veggtation Development Vegetation structure Figure 1 shows three profile diagrams representing the plants present at 3, 10, and 22 months in a permanent I x 20 m transect located in Succession Site 1. These three profiles are typical of the sequence of structural development seen throughout Succession Site 1 and in other cut and burn sites in the San Carlos area. Density was low at three months but had increased sharply by the 10 month survey. There was little change in density between the 10 and 22 month surveys. However, a large change in vegetation height occurred between these survey periods. By 22 months a loose canopy of Cecropia was present, and vegetation cover had increased from the 10 month value of 20 to 70 percent. Boerboom (1974) reported a similar development of vegetation structure (N1 well drained soils after cutting and burning in Surinam. Density Figure 2 illustrates the changes in density in different groupings of plant species over the first 22 months of succession in Succession Site 1. The first colonizers appeared five weeks after the burn. The total plant density at four months was 0.58 individuals per m2. All of these plants were less than 25 cm tall. Forest tree species accounted for 41 percent of all individuals present at this time. By 10 months herbaceous and woody secondary species had increased rapidly relative to forest tree species and accounted for 74 percent of the total 140 141 Figure 1. Three profile diagrams representing the plants present at 3, 10, and 22 months in a permanent I x 20 m transect located in Succession Site 1. Only individuals more than 5 cm tall are illustrated. J. a- 3 MONTHS :- Lon 142 (N) 446139 O 3..- N l a. (29? (fig: e t- ? . I 36¢ D r9! ’ ' O- - ' O - 7.5;: M- as. be m k; w " ‘ ' . m 2': I ', '. c— r- t- 2 § 8 , \ O 3.44- N ' o-I . N 'r.‘ "5‘— . ' i U 1' ’9‘ o] . . 4 ;. R «haul-rtye1,___._..__.__ ._ '- ‘fl . "H \ fix tin: ' '4." ,' ' - ' r - ‘_ fifczfi‘f?" g h l£" A ' Li: I 3.; ‘ ' ‘1 3 3 ’ ‘ 9‘91! .' ' N ’ '2 1 -\‘\K‘ -—'- --N—— -- ,.. ‘Ia su— ‘ a, v ‘ \\nnumu\r - O ' P" 11 up | ‘ -"-'-'-‘-3 - 33.: “(s '- ‘ ' t ' a . . - . . . — .r;4§11_____ :39f37.n A) . s ‘ Q ' . I‘ '- -‘_. 't I u ‘ 9 :‘n ‘ ‘ r -‘ ““ ——--- ,_ .I‘. 2‘? '1-\‘ z \ ,t \ ,_ ~ . g 1 n» 'lwr ,L t\\‘ - c ‘ r I I . CL._.- __ .4: \- g m N .— O . ‘ 7 Distance (I) Figure 2. 143 The changes in density, for five plant groupings considered separately, and for all individuals together, over the first 22 months of succession following burning in Succession Site 1. Only individuals more than 5 cm tall were considered as established. ltDlVlUJALS PER M2 HUBER OF 144 TOTAL INDIVIDUALS SECONDARY MY / a... GRASSES .FmEST TREES ’ LIANAS 4 10 15 22 KNIT-ls AFTER EWING 145 individuals present. The increase in total density after 10 months was slower. Forbs increased until 16 months and then decreased, while grass and secondary woody species numbers continued to increase throughout the 2 at 22 months study period. There was a total of 7.88 individuals per m with secondary species (i.e., forbs, grasses, and secondary woody plants) comprising 80 percent of all individuals. Forest tree species accounted for only 10 percent of the total individuals present at 22 months. The detailed results of these density surveys are presented in Appendix Table 1. Similar to Succession Site 1, the density at four months in Site 2 was low (i.e., 0.35 individuals per m2 ). The density increased to 9.33 individuals per m2 at 10 months primarily because of heavy colonization by the monocarpic herbaceous composite, Eupatorium cerasifolium. By 16 months most individuals of this species had fruited and died causing a drop in total density to 3.81 individuals per m2. Trends in the other plant groupings were similar to those recorded in Succession Site 1 (see Appendix Table 2). Other sites that I have observed show a pattern of density change similar to that described for Succession Site 1, although variation in detail such as that described for Succession Site 2 does OCCUY‘. Species richness Species area curves (Figure 3), based on the survey of the 27 1 x 1.5 m vegetation survey plots in Succession Site 1, were constructed four times over the 22 month study period. At four months the species area curve had a low slope but did not appear to be leveling off. Seventy-one percent of the species sampled at that time were forest tree 146 Figure 3. Species area curves for Succession Site 1 at 4, 10, 16, and 22 months following the burning of this site. Only individuals more than 5 cm tall were considered. Cmnve ween OF SPECIES ’45- 147 22 mourns l6 mums 10 ms Alel-ls L” I 1 I IS PLOTS (1 x 1.5 M) 25 148 or liana species. Because of the high diversity of the forest flora this curve would probably have continued with its positive slope through many more sample plots. The 10, 16, and 22 month curves increased sharply for the first five plots sampled and more gradually as the remaining plots were considered. Appendix Table 3 summarizes the changes in species number with time in each of six plant groupings (i e., forbs, grasses, secondary woody, lianas, forest trees from sprouts, and forest trees from seed). All groupings show the same trend of increasing rapidly in species number up to 10 months and slowly thereafter. The sharp increase in total species number between the four and ten month survey parallels the abrupt increase in total plant den- sity recorded for this time period (Figure 2). Once a species was encountered in a survey, it generally appeared in subsequent surveys. For example, of the 46 species present in the 10 month survey, 40 were still present at 22 months. Five of the six species not present were of forest species represented by only one individual in the 10 month survey. Hence, the early successional flora (i.e., up to two years) may be best regarded as one unit or association rather than a series of associations which replace one another. Figure 4 indicates that most of the early successional flora in Succession Site 1 had been sampled by 22 months. Species dominance At the 10 month survey, one-half of the 46 species present in the 27 vegetation study plots (i.e., a combined area of 40.5m2) in Succes- sion Site 1 were represented by only one individual (Table 1). Only four species had more than eight individuals present. Eupatorium cerasifolium, £1 herbaceous composite, was the only species with more 149 Figure 4. The change in the total number of species and in the number of newly encountered species for Succession Site 1 at 4, 10, 16, and 22 months following the burning of this site. Results are based on repeated surveys of 27 1 x 1.5 m perma- nent plots. NUMBER OF SPECIES 40 - 150 TOTAL SPECIES PREVIOUSLY UNENCOUNTERED SPECIES MONTHS 151 H H vOrOO N N NOINH e H N H OHrO O N O N H 0-0 NH O O O O elN ON HH N O N O H OOost NN HOV H H NOINH O N H OHIO N H H 0-0 NH O O O N H erN ON O e e N e H Osesez OH HOV HmpeH moot» OmcewO Oeeez OoOOmeo Once; HOHmOuw>wecw we LerOOV HOoceO Ocmeceeom OoOOmHO oucmeczn< Oowooam we Noeezz .eocoewmceo ocoz HHmu 20 O cmcu owes mHmae uw>OecO OHOO .oqu Owe» we Oswecze one Nopwm mcpees NN ecm OH eouuzeceo OOo>NOO Lew on OozHe> .H e 0.0e we moem Howe» e ..o.wO Owe—a >o>csm cowpmuomo> H oqu eewmmooezm NN on» cw ucoOocm .mowooOO co>OO m we .OHmaew>wecO we tongue on» on Lowoc OoOOmHe oecmecanm och .OchazecO ecmHO o>ww cw OoOOmHe oocmueanm we mowcom m cw pcomoca mowooam we Logan: och .H oHemH 152 than 16 individuals present in the 40.5 m2 sample area. Lack of numeri- cal dominance was particularly evident among grasses and primary forest trees, no species having more than four individuals present. At 22 months, many species (i.e., 42 percent) were still represent- ed by only one individual in the sample plots (Table 1). All liana and forest tree species had less than nine individuals present at that time. Other plant groupings showed evidence of numerical dominance, however. For example, Eupatorium cerasifolium had more than one individual pres- ent per m2. Two grass species each had more than nine individuals 2 present in the 40.5 m sample area. The secondary woody genera Cecropia and Solanum had also increased in density. Viewed in terms of biomass, Cecropia spp. was the overwhelmingly dominant genus at 22 months, ac- counting for approximately 78 percent of the total Site 1 living phyto- mass (see Part II, Table 8). The gradual increase in Cecropia spp. dominance over the first 22 months of succession was accompanied by an increase ‘hl species richness within the site. This finding suggests that the clear dominance of Cecropia spp. at 22 months did not affect species packing (given the 0.25 ha scale used in this study). If this is true, it would not conform to Grimes' (1979) conclusion that increas- ing dominance is associated with decreasing species richness. Throughout the lowland American tropics, succession after burning or short fallows appears to lead quickly to a stage dominated by secon- dary woody species, frequently in the genus Cecropia. Kenoyer (1929), working (H1 Barro Colorado Island in the Panama Canal Zone, noted that grasses, sedges, and herbs dominated the first months of succession followed by large-leaved monocots at one year and secondary woody spe- cies in the genera Cecropia, Trema, Ochroma, and Cordia at two years. 153 Snedaker (1970) noted the frequent occurrence of monodominant stands of Heliconia spp. and Calathea spp. in one year old fallows in Guatemala. These monocultures were generally overtopped by rapidly growing soft- woods such as Cecropia spp. and Cochlospermum spp. in the second year of succession. Many investigations in temperate areas have also described the first stages of succession after forest fires (Skutch, 1929; McMinn, 1951; Lynton, 1955; Lutz, 1956; Cremer and Mount, 1965). Bryophytes, ferns, herbs, and woody sprouts are the most commonly reported dominant vegetation elements in these studies. Hence, with the exception of bryophytes, which are infrequent, all the floristic elements of temper- ate postburn succession are conspicuous in the San Carlos recovery. Questions Posed by the Results of the yggetation Development Study To fully understand the factors which control succession in the upper Rio Negro region of the Amazon basin, I found it necessary to buttress my descriptive studies of succession with further field obser- vations and experiments. Answers to the following seven questions were sought: 1. 00 cut forest trees have the ability to sprout? 2. What effect does burning after cutting have on sprouts and on seed pools? 3. Do the first secondary colonizers after the burn originate from wind dispersal seeds or seed pool seeds? 4. What are the dispersal strategies employed by secondary spe- cies? 154 5. Why are population densities so low during the first months of succession? 6. Why (k) population densities increase rapidly after about the fourth month of succession? 7. Can the spatial distribution of species colonizing early suc- cession sites be related to any physical factors? Vegetation Development Explained Sprouting of cut trees The importance of stump and root sprouting in recovery following the cutting of tropical forests has frequently been mentioned (Ducke and Black, 1953; Emerson, 1953; Sarukhan, 1964; Jordan, 1969; Snedaker, 1970). Sprouting is the most common means of regeneration after forest cutting in the San Carlos region as well. Succession Site 3 had 7.3 individuals present per In2 three months after the mature forest occupy- ing this Site was cut. Eighty-seven percent of these individuals were of sprout origin (Table 2). Forest ground layer herbs and tree seedlings (probably present but not damaged during cutting) comprised the remain- ing 13 percent (i.e., 18 plants) of the individuals present in the plots. No secondary plants were encountered, indicating that cutting may inhibit their establishment. Succession Sites 1 and 2 were also dominated by tree sprouts three months after forest cutting. In most cases, it was easy to distinguish sprouts from seed- established individuals once the plant was excavated because sprouts usually retained evidence of scar tissue where the stem had been cut by machete, as well as a characteristic stem curvature below the scar tissue where a lateral bud had grown around the cut region of the stem. 155 Table 2. The number of sprouts and seedlings present in 19 1 m2 plots in Succession Site 3 three months after the mature forest occupying this site was cut. Sprouts Seedlings Total Individuals 121 18 Mean per m2 6.37 0.95 Percentage of Total 87 13 156 Two additional cut-only tierra firme sites, one and two years old, were observed in the San Carlos vicinity. Tree sprouts were the domi- nate component of the vegetation in both of these sites. The two year site had regenerated to a height of 5 to 6 meters. Hence, in answer to the first question posed, cutting of the forest in this region stimulates sprouting. This proclivity toward sprouting by forest trees simplifies the succession after cutting. There is only one seral stage; tree sprouts dominate the early months of succession and simply grow larger. Effects of burning on sprouts and seed pools Succession after the burning of a cut forest is distinct from the characteristic succession following cutting, suggesting that burning is very important in influencing succession. The burns in the San Carlos region can be very hot. The temperature reached at a given distance above or below ground in Succession Site 3 was projected to be midway between the temperature sensitivities of the pair of lacquers, one of which melted and the other of which did not melt. On this basis, the approximated mean peak burn temperature was 392°C at 7.5 cm above ground (range 148° to 593°C; s.d. = 157.0); 138° at the soil surface (range 76° to 310°C; s.d. = 62.4); and 100°C at one cm below the soil surface (range 48° to 199°C; s.d. = 44.6). The wide ranges and high standard deviations at each of these measurement points indicates that burn effects ("1 cut sites are not spatially uniform. In accord with this, burn measurements in areas of light slash cover had surface temperatures ranging from 76° to 124°C. Surface temperatures in areas of heavy slash cover ranged from 226° to 310°C. 157 Literature reports of peak temperatures for slash burns differ greatly because the investigated fires have occurred under varied weath- er conditions and with different quantities and qualities of fuel. Brinkmann and Vieira (1971), measuring the burn temperatures of a cut second growth forest in the Manaus region of Brazil, reported surface temperatures ranging from 100° to 150°C and below ground temperatures (at 2 cm depth) ranging from 95° to 125°C. Hotter burns have been recorded in the temperate zone. For example, Isaac and Hopkins (1937) recorded temperatures of 1005°C above the forest floor and 320°C at 2.5 cm below the soil surface in coniferous forest slash fires. There was no apparent correlation (all correlation coefficients were less than 0.3) between burn peak temperatures at 10 to 20 cm above ground, or 0 to 1 cm below ground (as measured in the 19 one In2 vegeta- tion study plots) and sprout or seedling density in these plots. Temperatures in all of the 19 plots may have been high enough to kill sprouts. Plant tissue is damaged or killed when subjected to tempera- tures in excess of 54°C for several minutes (Hare, 1961). All the Succession Site 3 burn measurement plots experienced temperatures great- er than 124°C at 20 cm above the ground. Indeed, of the 121 sprouts present in the 19 one m2 vegetation survey plots (Table 2), only one survived the burn. Furthermore, very few new sprout establishments (i.e., 0.63 per m2) were present four months after burning. Hence, the burn effectively eliminated sprouting as a quick means of recovery. The reason for the absence of any apparent correlation between burn intensity and the density of colonizers establishing from seed was difficult to understand, but investigations on the effects of burning on seed pool mortality provided some insights to this question. The mean 158 number of germinations per soil sample from lightly burned areas in Succession Site 3 was 7.3 (n == 10) vs. a mean of 5.3 (n = 10) germi- nations per sample for areas which were heavily burned in this site. This difference was not significant (t = 0.69). A possible explanation for this result (and for the lack of a correlation between burn inten- sity and seedling establishment) is that the threshold temperatures beyond which seed mortality occurs (i.e., below 100°C for the seven most common secondary species in the San Carlos area; Uhl, unpublished data) were reached throughout Succession Site 3 at the soil surface, but that this threshold was not reached for buried seeds in either lightly or heavily burned areas. In accord with this, 14 of the 19 burn measure- ment plots had temperatures in excess of 124° at the soil surface, but only 2 of the 19 plots reached temperatures as high as 124°C at 1.5 cm below the soil surface. Table £3 summarizes the seed pool germination results from Succes- sion Site 3 and from the tierra firme forest adjacent to this site. The forest was regarded as a control from which the effects of burning on seed pools could be evaluated. There was a significant difference (t = 7.37; a = 0.001) between the tierra firme forest seed pool and the postburn Succession Site 3 seed pool (samples from lightly burned and intensively burned plots pooled) indicating that burning does deplete seed pools. Succession Site 3, sampled immediately after the burn, had .an estimated germinable seed pool of 157 seeds per m2 . Twenty-two percent of the seedlings did not grow to an identifiable stage. Cecropia spp. seedlings were the most frequent, comprising 78 percent of all identi- fied individuals. Only two forest trees (Ficus sp.) and one herb 159 Table 3. A comparison between the post-burn seed pool in Succession of the tierra firme forest adjacent Site 3 and the seed pool to this site. Post-Burn Forest Site Site Number of samples 20 17 Mean number of germinations per sample(a) 6.3 30.1 Mean number of germinations per m2 157 752 Percentage of total germinations growing to an identifiable stage 78 71 Percentage of total identified germinations, forest species 4 4 Total number of forest species 1 1 Percentage of total identified germinations, secondary species 96 96 Percentage of total identified germinations, Cecropia spp. 78 90 Total number of secondary species 7 6 (3)0nly two contaminations (i.e., 0.4 per sample) occurred in the control trays during the three month observation period. 160 (Euphorbiaceae) were present. The remaining individuals belonged to six species of secondary woody plants and one species of liana. The tierra firme forest had an estimated germinable seed pool of 752 seeds per m2, Table 3. Cecropia spp. comprised 90 percent of all identified germinations. Four percent of the identified germinations were of forest trees, all of one species. The remaining individuals belonged to f0ur species of secondary woody plants, three species of lianas, and one forb species. Guevara and Gomez-Pompa (1972) found 175 to 689 germinable seeds per m2 in lowland primary forest in Veracruz, Mexico. Most of these seeds, likewise, were of secondary species. And Keay (1960), working in Nigeria, found approximately 100 germinable seeds per m2 in forest soils there. Eighty-four percent of the germina- tions in Keay's study were of secondary species; 15 percent were herbs and lianas and only 1 percent were forest trees. Hence, in answer to the second question posed regarding burn ef- fects on early succession, burns in the San Carlos area can be very hot, reaching above ground temperatures in excess of 593°C and soil temper- atures (at one cm depth) in excess of 173°C. Most pre-burn sprouts are killed by the burn, and sprout establishment is low after the burn. Seed pools, likewise, are significantly reduced by burning. Seed source of the first colonizers after burnipg Seed pool germination studies (Table 3) showed that, although burning reduces the size of the seed pool, there is still a substantial germinable seed pool (i.e., approximately 157 seeds per m2) present following burning. The results from the methyl-bromide seed pool exter- mination experiment indicate that this seed pool is important in the post-burn establishment of some plant species. 161 A mean of 4.0 woody individuals (i.e., 1.0 plant per m2) had estab- lished from seed four months after the burn in the non-methyl-bromide plots vs. 0.2 (i.e., 0.05 plants per m2) in the methyl-bromide plots. This difference was significant at the 0.05 level (paired t statistic = 3.0; d.f. = 5). Fifty-seven percent of these woody establishments were in the genus Cecropia; the remaining individuals were of other secondary woody species and of forest species. Hence, it appears likely that the first woody colonizers of cut and burn disturbances in this region originate from seed pool seeds. Grasses were present in both treatment types after four months. Means of 6.0 and 2.2 grass individuals per treatment plot were recorded in the methyl-bromide and the non-methyl-bromide treatments, respective- ly. This difference was not statistically significant (paired t = 0.57; d.f. = 5) providing evidence that grasses rely on post-burn seed disper- sal for their initial establishment of Site populations rather than on seeds present in the seed pool. This conclusion is supported by the forest seed pool germination study (Table 3) in which no grass estab- lishments were recorded. Forbs did not establish in either treatment suggesting that they, too, are dependent on post-burn dispersal. Controls verified that seeds were killed by methyl-bromide and that seeds would germinate and grow in methyl-bromide treated soil. These results provide an answer to the third question posed by providing evidence that post-burn dispersal is an important seed source for the initial establishment of forbs and grasses, but that the first secondary and primary woody colonizers of cut and burned sites in this region originate from seeds present in the seed pool. 162 ngpersal strategies of secondary species Although most seeds produced by secondary plants appear to remain close to the parent plant, the wide distributions of many tropical weedy species indicate that means of long distance (i.e., inter-site) disper- sal also exist for secondary plants (Kellman and Adams, 1970). Many of the common forb and grass species in the San Carlos region post-burn succession have wind-dispersed seeds. In contrast, the common secondary woody species have berries (e.g., ytggtg spp., Solanum spp., and Clidemia spp.) or multiple fruits (e.g., Cecropia spp.) and rely on animal vectors for dispersal. Without the aid of vectors, these fruits fall to the ground below the parent plant. Three species of bats, Vampyros helleri, Artibeus sp. and a species in the Sturniratildae were mist netted and found through feces germination tests to be transporting copious seeds of Cecropia ficifolia while one bat species, Grollia perspicillata was transporting seeds of Vismia lauriformis. Five species of birds were also mist netted. The three most common species in the nets (i.e., represented by 4 to 10 individuals each) were trans- porting hundreds of seeds of Solanum stramonifolium and Clidemia sericea. Based on this limited survey, it appears that the four domi- nant secondary woody plant species (i.e., in terms of numbers and bio- mass) rely on bats or birds for dispersal. The role of bats and birds in the dispersal of secondary plants was noted by Beard (1945) in his successional studies on St. Vincent Island in the West Indies. More recently, Vasquez-Yanes gt g1. (1975) stressed the importance of bats in seed dispersal in the humid tropics, and Trejo (1976), studying seed dispersal in 106 frugivorous bird species in Mexico, found most bird species to be dispersing seeds of secondary 163 species. In answer to the fourth question posed, effective means of between site seed movement (e.g., mechanical or with the aid of animal vectors) appear to be incorporated into the seed dispersal strategies of most secondary species of the San Carlos region. Low population densities during the first months of succession Monthly rainfall is generally greater than 100 mm for all months in the San Carlos region, but a relatively dry season occurs from about October to March of each year. This period of lower rainfall is also a thma of low seed germination and low seedling establishment in early successional sites. Figure 5 illustrates monthly rainfall simultaneous with the mean number of new germinations present in monthly surveys of 17 1 x 1.5 m plots in Succession Site 1 for the period from 10 to 23 months following the burning of this site. A total of 4,332 germina- tions were marked during this 13 month survey. The lowest number of germinations (0.31 m2) and the lowest rainfall (101 mm) were in January 1978. The rainfall high, 524 mm, was for the month of June at which time the mean number of new germinations present (26.5 per m2) also peaked. The correlation coefficient between monthly rainfall and the mean number of germinations present in monthly surveys for the 13 month period was .80. Lower numbers of germinations during the drier months resulted in lower numbers of establishments at that time. In the four month period from December 1977 through March 1978 (i.e., the 12th to the 16th month of succession), seven individuals became established in the 17 plots under survey. A total of 593 mm of rain fell during this four month period. In contrast, 1763 mm of rain fell in the next four months and 164 Figure 5. Monthly rainfall simultaneous with the mean number of new germinations present in monthly surveys of 17 1 x 1.5 m plots in Succession Site 1 for the period from 10 to 23 months following the burning of this site. 165 --- RAINFALL --- SEEDLING NUMBER R =.80 O -30 C" A 600- [$1 2 2 / E Z v \\ (D m _J I, 0‘ m g u. I \ I: 2 E I \ 2 C < \ 0 Z O: \ U) (I) \ I11 4 .\ 10 F131 :0 200 \ :0 \ I \ I \ ZN I \ l l 5;. 4 l 1 OCT JAN APR JUL OCT 1977 1978 166 25 individuals established. The monthly rainfall for this eight month period was positively correlated with the number of seedlings marked each month which subsequently established (r = .87). As these results indicate, seedling mortality can be reduced arti- ficially by watering. In a separate experiment, 75 naturally colonizing seedlings in Open, bare soil plots were watered and surveyed daily for 14 days and compared to 75 control seedlings (i.e., non-watered). Mortality from desiccation was 25 percent in the non-watered plots vs. 5 percent in the watered plots (significant at the 0.05 level, paired t-test). These results provide an answer to the fifth question posed by showing that periods of relatively low rainfall reduce the rate of seed germination and seedling survivorship in cut and burn sites. Aside from the period of reduced rainfall from the 12th to the 16th month of suc- cession in Succession Site 1, rainfall was also relatively low during the first three months of succession in this site (and in Succession Site 2 and 3 as well). The relatively low rainfall coincident with the first months of succession, in part, explains the slow initial coloni- zation of these Sites. Rapid population increase after the fourth month of succession The initial succession site colonizers, regardless of origin, are at very low densities during the first months of succession after burn- ing (Figure 2). The first forb and grass colonizers on Succession Site 1 flowered and fruited from the fourth to the sixth month after burning, thereby supplying seeds locally which gave rise to comparative- ly dense populations of forbs and grasses. These second populations grew more slowly, flowering and fruiting toward the end of the first 167 year of succession. Most of the seeds produced by these plants landed below the parent plant. Many of these seeds germinated but few indivi- duals matured. Some individuals of Egpatorium cerasifolium persisted as seedlings for up to five months before dying. The few individuals that did mature from this third generation fruited at about 16 to 20 months. Further establishment of forbs and grasses after this time was res- tricted to exposed locations. A similar pattern was observed in Succession Site 1 for secondary woody species. It was only after the sparsely distributed early coloni- zers began to flower and fruit (i.e., at about 10 months) that high population densities of secondary woody species' seedlings were recorded for this site. The magnitude of within-Site seed dispersal as compared to between-site seed dispersal is reflected in the results of the seed trap study. No seeds were diSpersed on to the seed traps in the freshly burned (non-vegetated) site over the two week sample period; whereas 91 seeds were captured in the seed traps placed in the 16 month old succes- 2 were dispersed sion site indicating that approximately 197 seeds per m within this site over the two week sample period. Hence, in answer to the sixth question posed, my results, based on field observations and a limited seed trapping survey, indicate that within-site seed dispersal is important in explaining the rapid increase in population size ob- served after the fourth month of succession. Microhabitatppreferences of secondary species Root mat, charred wood, and bare soil surfaces and these three surface types combined with slash comprise the principal microhabitats available for plant colonization in cut and burn succession sites in the San Carlos region. Root mat surfaces (i.e., a 5 to 30 cm thick mesh of 168 0.1 to 2 mm diameter roots produced by the forest trees previously occupying the site) can comprise up to 80 percent of the total surface cover in lightly burned sites. Mineral soil overlain with charred wood occurs in areas where burns are intense and large Slash abundant. Bare soil surfaces occur where the root mat is thin and dry enough to burn completely. The percentage of bare soil surface increases with time because the root mat decomposes exposing the bare soil, and the charred wood is washed away in heavy rains leaving bare soil below. In the later stages of succession the root mat is reformed over the bare soil sur- face. The mean percentage cover of slash and of these three surface types, as estimated in the 27 1 x 1.5 m vegetation study plots in Suc- cession Site 1 at 10 months, were quite similar (Table 4). All three surface types and slash are clumped in distribution (Table 4). Slash clumping is the result of the felling technique practiced by local farmers whereby the toppling of large trees is used to sweep over tens of partially cut smaller trees. This procedure produces large concen- trations of slash (e.g., 100 to 900 m2) in some areas with essentially open area. hi between. Burning is most intense in areas of high slash concentration. These areas also have the highest charred wood surface cover and log cover after burning. Plants establishing from seed in the 27 plots were also clumped in their distribution (sz/x = 3.9, t = 10.2***). The number of seed estab- lished plants was weakly correlated with both percent charred wood surface cover (r = .40) and percent slash cover (r = .51). Establish- ment was not correlated with percent root mat surface (r = .10) and was 169 .oazp oomwczm we :ewpmewpmo cheeHoocO eczecm on» :e Owauocwe OOOOH Om: cOmHO Opewa oEeO ew omamoon NOOH Hmeco we: Ooee OoOOO Lo>eu oomwcsm ooccp ocu we saw ocHHeO .HOO. o O OeoeoweswawwHOO eoaezwu xxxO.¢O 0.0H N.NOO N.OO No>ee cmmHO eoOEOHo xxxO.eO N.NN 0.0NO HOvN.ON oemwLOO HOeO ocmO eoOEOHo xxx0.00 O.NH N.Omv O.NN oemwLOO tee: uowwmcu eoOEOHe Hmvxxxn.mm 0.0N 0.00HH. 0.00 oumw.5O me poem sowosaweomwO o XN O O HOO I N N emoz .OHeHO E O.H x H NN we Oo>e=m one ce eomme ocm Oosz> .H oHOO :ewmmouozm cw chcwan Louwm Ocpcee OH Hm OOmHO ecm .Hwem ocme .eeez eocceco .ume pee; we cowpsewcpmwe we coccme on» .e oHaOH 170 somewhat negatively correlated with percent bare soil surface (r == - .41). Based on these distribution and correlation analyses, it appeared that locations with charred wood and slash were favored sites for estab- lishment. I was able to determine, in more definitive terms, the microhabitat preferences of colonizing species and the physical characteristics of the nficrohabitats available for colonization by constructing multiple sets of the six post-burn microhabitat types and then measuring their physical differences and the rate at which they were colonized. With regard to physical factors, these six microhabitats differ markedly in their temperature and evaporation characteristics. Figure 6 shows the changes in surface temperature for each of the microhabitats during a typical day having a maximum air temperature of 33°C. Surface temperatures increased until 1:00 to 3:00 PM in all microhabitats and then declined steadily. The charcoal surface heated most rapidly and reached the highest temperature for any microhabitat (i.e. 54°C). The bare soil surface, a cream colored sand, was heated less than the char- coal and root mat surfaces. Accordingly, reflected light from the bare soil surface, measured in luxes, was 7.6 percent of open site light on an overcast day vs. 3.9 percent and 2.4 percent of open site light, respectively, for the root mat and charcoal surfaces. Slash presence shaded the soil surface thereby reducing surface temperatures. Light was reduced to 33 percent of open site light among the nested wood poles in the three slash treatments. The bare soil surface had the lowest temperature of the slash treatments with the root mat and charcoal slash surfaces loosely grouped at slightly higher temperatures. 171 Figure 6. The changes in soil surface temperature over an eight hour period in each of six microhabitat types. surface temperature (cl Soil 55 45 172 Root mat with slash Bare soil Root mat Charcoal / \\ \ ' A. \ / ..o. "..... °... \‘\ .¥ \ l/ ‘ , Charcoal with slash Bare soil with alaah I ll 12 2 4 Ti me of day 173 Soil temperature differences (2 cm depth) were more muted for the six microhabitats, but again the three slash and the three non-slash treatments were clusterd (Figure 7). The soil temperatures (H: the charcoal and root mat nficrohabitats were slightly lower than the soil temperatures ikn‘ the bare soil microhabitat; these surfaces apparently insulate the soil from direct solar radiation. No clear differences in soil temperature were present among the three surfaces with slash. Soil evaporation (Figure 8) was greatest from the bare soil micro- habitat followed by the root mat surface type and then the charcoal surface type; this order was repeated in these three microhabitat types with slash present. All but the charcoal vs. bare soil with slash and the root mat with slash vs. charcoal with slash treatment comparisons were found to be significantly different from each other at the 0.05 level [Duncan's multiple range test (DMR)] based (N1 a two-way ANOVA. There is a close correspondence between the soil evaporation ranking by treatment type and the soil temperature ranking. Appendix 'Table 4 provides a summary of the physical characteristics of the six microhabi- tat types. The existence of clear differences in temperature and soil evapor- ation among the six microhabitats suggested that differences in germi- nation and establishment would occur for seeds dispersed (Hi to these microhabitat types. Figure 9 shows the results of the natural colon- ization of these six microhabitat types over a three month period based on the survey of 15 six treatment blocks. The charcoal surface with slash had a significantly greater number of establishments (p < 0.05, DMR) than the other treatments, and the root mat with slash treatment had significantly more establishments (p < 0.05, DMR) than the three 174 Figure 7. The changes in soil temperature (2 cm depth) over an eight hour period in each of six microhabitat types. Soil temperature (c) 45 25 Bare soil Root mat 175 Charcoal Root mat with slash Bare soil with slash 12 2 4 Time of day 176 Figure 8. The mean amount of water lost over an eight hour daylight period from five evaporation containers placed in each of six microhabitat types. ““““““ W c aaaaaaa \\\\\\\\\\\\\\\\\\\ 3' W 178 Figure 9. The mean number of individuals establishing from seed over a four month period in each of six microhabitat types. 179 . ....... W “mm... W A 180 non-slash treatments. The examination of Figures 8 and 9 simultaneously suggests that evaporation and establishment are inversely related. Where evaporation was greatest (i.e., bare soil and root mat treat- ments), establishment was lowest. These results, then, provide experi- mental evidence that the distribution of plants in early succession sites could be strongly influenced by the distribution of surface types. The results of this microhabitat colonization experiment did not necessarily reflect the microhabitat preferences of all secondary spe- cies in the region, because only three secondary species colonized the treatment blocks during the experiment. More detailed information on species microhabitat preferences was gained by placing seeds of second- ary species directly on the six microhabitat types and then checking for establishment differences. Seven of the more common secondary species were tested in this way for their microhabitat establishment prefer- ences. Figure 10 illustrates microhabitat differences in establishment for each species. The four secondary woody species: Cecropia ficifolia, Vismia lauriformis, Solanum stramonifolium, and Clidemia sericea and the forb, Eupatorium cerasifolium had their highest establishment in the slash treatments (Figure 10). Q. ficifolia had significantly greater (p < 0.05, DMR) establishment in the bare soil with slash and charcoal with slash treatments than in all other treatments. V. lauriformis, like 9. ficifolia, had its greatest establishment in the bare soil with slash and charcoal with slash treatment (significantly greater p < 0.1 DMR than in all other treatments except the root mat with slash treatment). §. stramonifolium had significantly' greater' establishment (p < 0.05, DMR) in the charcoal with slash treatment than in all other treatments Figure 10. 181 The mean number of establishments in each of six microhabitat types for seven common early successional species. An equal number of seeds (from 300 to 600 depending upon the species) were broadcast in each microhabitat type. From five to seven replicates were used for each species. 182 Cecropia ficifolia 90" Ionum stramonifolium Vism' louritormis mhzwzzmfimfi. § ...... ................. .. ............... ......... “on-nonououooeoeu u .............. .. .........I....O.....H.H. ........... OOOOOOOOOOOOOOOOOOO Eupatorium msitolium Ponicum loxum Oh . 0 4 60' 2 Clidehio sericea mm .3 132:2 60 Borrerio Iotitolio 15" 183 except the bare soil with slash treatment. 9. sericea, likewise, showed it highest establishment in the charcoal with slash treatment, having significantly greater establishment (p < 0.05, DMR) there than in all other treatments except the root mat with slash treatment. The forb, E. cerasifolium, had its best establishment (p < 0.1, DMR) in the bare soil with slash and charcoal with slash treatments. In contrast to these species, the grass, Panicum laxum, showed a clear preference for the bare soil microhabitat, having significantly greater establishment there (p < 0.01, DMR) than in all other treatments (Figure 10). Very few individuals of this species established in the three slash treatments. The second forb species tested, Borreria latifolia, showed 'the least microhabitat preference of all species tested (Figure 10); no significant differences between treatments were detectable at the 0.1 level (DMR) for this species. Other investigators have found species distributions to be corre- lated with microhabitat distributions [e.g., Bratton (1976) in temperate deciduous forest herbs and Grime (1979) in temperate pastures on shallow soils over fissured limestone]. Ultimately, there were differences in the number of establishments in the six microhabitat types for a given species because of differential germination of seeds and/or differential mortality of seedlings. Differences in germination among the field treatments used may have resulted from different frequencies of favor- able germination microsites. The moisture, temperature, and surface roughness of the six treatments were different, and these factors proba- bly determined the number of safe germination sites. Harper and Benton (1966), in an experiment where they placed seeds of three species on 184 sinter plates with water under different tensions, showed that seed germination, for the species tested, was related to the percentage of the total seed surface area in contact with the sinter plate and the evaporating power of the atmosphere. Differences in seed surface area for the species used in the San Carlos experiment are reflected in differences in seed weight for these species which sometimes exceeded an order of magnitude (Table 5). Once seeds germinate, their establishment differs depending on the microhabitat they are in. For example, establishment was very low in the root mat treatment for all species tested. Seeds entrapped in the root mat may germinate several cm above the soil surface under favorable moisture conditions but die before the hypocotyl penetrates the soil below. Those seeds that sift through the root mat and germinate in contact with the soil may not have sufficient starch reserves to grow through the root mat to the lighted environment above. It is also possible that germination or growth inhibitors are present in the dead roots comprising the root mat. Ooyama (1954) found that extracts of unburned litter inhibited germination and seedling growth for some tree species. The seven species tested also had low establishment in exposed charcoal microhabitats, but establishment was generally high on the charcoal surface with slash. Charcoal acts to increase surface temper- ature, but reduces soil temperature and evaporation. Charcoal can also increase soil pH (Tyron, 1948). I frequently noted wilted and dried seedlings in the charcoal treatments. The establishments that do occur in this treatment probably result from seeds landing in moist micro- sites, germinating, and then growing quickly during relatively cool wet 185 Table 5. Factors related to germination and establishment in seven common secondary species occurring after cut and burn disturbances. Factor Species Dry Weight (mg) per Hypocotyl Length (cm) Seed (n = 160) before Cotyledon Emergence (n = 15) Cecropia ficifolia 0.5 0.9 Vismia lauriformis __(a) 0.7 Solanum stramonifolium 1.0 4.1 Clidemia sericea 0.05 0.2 Borreria latifolia 0.7 2.6 Eupatorium cerasifolium 0.05 0.2 Panicum laxum 0.1 0.5 (a)Data not available. 186 periods. Establishment was high in the bare soil with slash microhabitat for all secondary species tested presumably because the short hypocotyls of these species can enter the soil directly while receiving protection from direct solar radiation. Establishment was low in the exposed bare soil microhabitat for all species tested except the grass, 2. laxum. The extremely wet period (there was rain on 21 consecutive days follow- ing the broadcasting on the grass seeds) during which this experiment was conducted probably influenced the outcome. Seeds of this same grass species were re-broadcast in each of the six microhabitat types during drier weather with almost the opposite results. At this time, there were over 100 establishments in the bare soil with slash treatment and only three establishments in the exposed bare soil treatment; establish- ments were also higher in the charcoal with slash treatment than in the charcoal treatment. No establishments occurred in the root mat treat- ments. Heat probably promotes germination in this species. During rainy periods the wet slash microsites may be too cool for establish- ment. During drier periods, however, these slash microsites are warmer, yet moist, and stimulate germination. Hence, during rainy periods some species may be able to colonize exposed microhabitats where they are unable to establish during drier periods. I hypothesized that species with heavier seeds (larger endosperm) and long hypocotyl lengths before cotyledon emergence would be at an advantage in the exposed microhabitats, and that species with lighter seeds and shorter hypocotyles would be restricted to the slash micro- habitats. The two species with the lightest seeds and shortest hypocotyles (C. sericea and E. cerasifolium) were largely restricted to 187 the slash microhabitat as predicted (Table 5). However, S. stramonifolium, the species with the heaviest seeds and longest hypocotyles, did poorly in the exposed microhabitats. g. latifolia (the species with the second heaviest seeds) had good establishment in these non-slash plots. g. latifolia seedlings which were germinated in petri plates appeared to have contractile roots. This attribute may contri- bute to the success of this species in the exposed microhabitats. The seventh question that I sought to answer in this investigation was: can the spatial distribution of species colonizing early succes- sion sites be related to any physical factors? The results of these experiments strongly indicate that, given ample seeds, the distribution of secondary species after a cut and burn disturbance is predictable from a knowledge of the microhabitat distribution within the site. 0f the three surface types, the bare soil was the most preferred and the root mat the least preferred for establishment. Secondary woody species had their highest establishment in the slash microhabitats; whereas the grass species studied established best in the exposed microhabitats. One of the forb species tested showed no microhabitat preference and the other forb studied established best in the slash microhabitats. A knowledge of seed germination triggers, seed weight, and seedling mor- phology (e.g., hypocotyl length) may help in predicting which microhabi- tat a species may best be able to colonize. It is further suggested that microhabitat preferences may differ for some species depending on whether dispersal onto a given microhabitat occurs during a rainy or a dry period. 188 SUMMARY Vegetation Development An investigation of changes in vegetation structure, plant density, and species richness was conducted during the first 22 months of succes- sion after the cutting and burning of mature tierra firme forest sites in the Rio Negro region of the Amazon Basin. Very few plants estab- lished during the first four months after burning. The density of secondary woody plants, grasses, and forbs increased sharply from the fourth to the tenth month following burning. There was less change in plant density between the tenth and twenty-second month, but a large change in vegetation height occurred during this time. By the 22nd month of succession a loose canopy of Cecropia was present, light at two meters above the ground had been reduced to 26 percent of full sunlight, and the percentage vegetation cover had increased from the ten month value of 20 percent to 70 percent. Species number increased rapidly from the fourth to the tenth month. Once a species was encountered in a survey it was generally present in subsequent surveys. By 22 months, most of the early succes- sional flora present in the site under study had been sampled using 27 1 x 1.5 n1 permanent plots. This early successional flona is perhaps best regarded as a single association. Species dominance was not apparent during the first year of succes- sion, only one species having more than 16 individuals present in a 40.5 m2 sample area (i.e., 27 plots), but by 22 months Cecropia spp. was the 189 dominant genus on the sites (accounting for 78 percent of the total living phytomass in Site 1). The clear dominance of Cecropia at 22 months did not affect species packing on the 0.25 ha scale used in this study. Vegetation Development Explained Many tree species of the tierra firme forest are stimulated to sprout when they are cut. Hence, after the forest is cut there is a quick, single seral stage return to forest via sprouting. Burning after cutting changes the nature of succession dramatical- ly. Burn temperatures are high enough to kill most sprouts and to reduce, significantly, the size of the seed pool. The first secondary woody colonizers after burning were shown to arise from seed pool seeds, while the first forb and grass colonizers arose from seeds dispersed onto the site after the burn. The common secondary woody species had bat and bird dispersed seeds while many grass and forb species had seeds with adaptations for wind dispersal. Population densities were low during the first months of succession because of the low number of propagules present and because relatively low rainfall during this period reduced the rate of seed germination and seedling survival. Population numbers increased sharply after the fourth month of succession because many of the initial colonizers began to set seed locally at this time. There were six distinct microhabitat types present on post-burn sites. These microhabitat types were different from each other in soil temperature and soil evaporation. Most species tested showed clear microhabitat. preferences. Hence, the actual spatial distribution (If 190 species within secondary sites was somewhat predictable from a knowledge of the distribution of these microhabitat types. LIST OF REFERENCES 191 LIST OF REFERENCES Allen, P. H. 1956. The Rain Forests of Golfo Dulce. University of Florida Press, Gainesville. 417 pp. Beard, J. S. 1945. The progress of plant succession on the Soufriere of St. Vincent. Journal of Ecology 33:1-9. Blum, K. E. 1968. Contributions toward an understanding of vegeta- tional development in the Pacific lowlands of Panama. Ph.D. thesis. Florida State University, Tallahassee. 119 pp. Boerboom, J. it A. 1974. Succession studies in the humid tropical lowland of Surinam. Pages 349-357 in Proceedings of the First International Congress of Ecology, The Hague. Bratton, S. P. 1976. Resource division in an understory herb commun- ity: response to temporal and microtopographic gradients. American Naturalist 110:679-693. Brinkman, W. L. F., and A. N. Vieira. 1971. The effect of burning on germination of seeds at different soil depths of various tropical tree species. Turrialba 21:72-82. Budowski, G. 1961. Studies on fOrest succession in Costa Rica and Panama. Ph.D. thesis. Yale University, New Haven. 189 pp. Cremer, K. W., and A. B. Mount. 1965. Early stages of plant succes- sion following the complete felling and burning of Eucalyptus regnans fbrest in the Florentine valley, Tasmania. Australian Journal of Botany 13:303-322. Ducke, A., and G. A. Black. 1953. Phytogeographical notes on the Brazilian Amazon. Annais Academia Brasileira Ciencias 25:1-46. Emerson, R. A. 1953. A preliminary survey of the nfilpa system of maize culture as practiced by the Maya Indians of the northern part of the Yucatan peninsula. Annals of the Missouri Botanical Garden 40:51-62. Fosberg, F. R. 1950. Ecological notes on the upper Amazon. Ecology 31:650-653. Grieg-Smith, P. 1952. Ecological observations in degraded and secon- dary forest in Trinidad, British West Indies. Journal of Ecology 40:283-330. Grime, J. P. 1979. Plant strategies and vegetation processes. John Wiley and Sons, Chichester. Guevara, S., and A. Gomez-Pompa. 1972. Seeds from surface soils in a tropical region of Veracruz, Mexico. Journal of the Arnold Arboretum 53:312-335. 192 Hare, R. C. 1961. Heat effects on living plants. U.S. Forest Ser- vice Occasional Paper 183. 32 pp. Harper, .L L., and it. A. Benton. 1966. The behavior of seeds in soil, Part 2: The germination of seeds on the surface of a water supplying substrate. Journal of Ecology 54:151-166. Holdridge, L. R. and others. 1947. The forests of western and cen- tral Equador. USDA Forest Service, Washington, D. C. 134 pp. Isaac, L. A., and H. G. Hopkins. 1937. The forest soil of the Douglas fir region and the changes wrought upon it by logging and slash burning. Ecology 18: 264-279. Jordan, (L F. 1969. Secondary succession ‘Hl the irradiated area. Pages 65- 86 in C. F. Jordan and G. E. Drewry, The rain forest project annual report. Puerto Rico Nuclear Center, Puerto Rico. Keay, R. W. J. 1960. Seeds in forest soil. Nigerian Forestry Infor- mation Bulletin, new series, no. 4. 4 pp. Kellman, M. C. 1970. Secondary plant succession in tropical montane Mindanao. Publication BG/2. Australian National University, Canberra, Australia. 174 pp. Kellman, M. C., and C. 0. Adams. 1970. Milpa weeds of Cayo District, Belize (British Honduras). Canadian Geographer 14:323-343. Kenoyer, L. A. 1929. General and successional ecology of the lowland tropical rain forest at Barro Colorado Island, Panama. Ecology 10:201-222. Lutz, H. J. 1956. The ecological effects of forest fires in the interior of Alaska. U.S. Department of Agriculture Technical Bulletin 1133. Lynton, M. J. 1955. Observations on the origin and early development of a plant community following a forest fire. Forestry Chronicle 31:154-161. McMinn, R. G. 1951. The vegetation of a burn near Blaney Lake, B.C. Ecology 32:135-140. Ooyama, N. 1954. The growth-inhibiting substances contained in the leaf litter of trees. Journal of the Japanese Forestry Society 36:38-41. Richards, P. W. 1952. The tropical rain forest. University Press, Cambridge. 450 pp. Sarukhan, J. 1964. Estudio sucesional de una area talada de Tuxtepec, Oaxaca. Publicacion Especial Instituto Nacional Investigaciones Forestal, Mexico 3:107-172. 193 Skutch, A. F. 1929. Early stages. of" plant succession following forest fires. Ecology 10:177-190. Snedaker, S. 1970. Ecological studies on tropical moist forest succession in eastern lowland Guatemala. Ph.D. thesis. Univer- sity of Florida, Gainesville. 131 pp. Stevenson, D. 1929. Types of forest growth in British Honduras. Tropical Woods 14:20-25. Symington, C. F. 1933. The study of secondary growth on rain forest sites in Malaya. Malayan Forest 2:107-117. Trejo, L. 1976. Diseminacion de semillas por aves in los Tuxtlas, Veracruz. Pages 447-470 in A. Gomez-Pompa and others, eds. Regeneracion de selvas. Compania Editorial Continental, Mexico. Tyron, E. H. 1948. Effect of charcoal on certain physical, chemical, and biological properties of forest soils. Ecological Monographs 18:81-115. Vasquez-Yanes, C., A. Orosco, G. Francoise, and L. Trejo. 1975. Some observations on seed dispersal by bats in a tropical humid re- gion. Biotropica 7:73-76. Whitmore, T. C. 1975. Tropical rain forests of the Far East. Oxford University Press, London. 278 pp. APPENDIX TABLES 194 mm.n oe.m em.e mm.o o<~¢h mm.o em.o mm.o ¢N.o meega “emcee «5.0 m¢.o nm.o mH.o mmcmwo mm.~ OH.N mn.a oH.o zeeez zueeceeem mm.H em.o o~.o oo.o memmego oo.~ m¢.~ m¢.H mo.o mane; meecoz NN mcecoz ea mzecoz eH usage: 4 ecwasoae ecm_a we Lemlwpmeew>wecH we geesez .eegeewmcee wee: Ppep so m ence egos mpeeew> -weew zpco .H eewm :ewmmeeeem we newcgee ecu mcwzePFee mcpcee Nu new .mH .OH .¢ we mmcmeeegm acmpe e>w$ cw NE Lee mpmeem>wecp we geese: och .H e_nme xwucaaa< 195 ek.e He.m mm.m mm.o ewecH ee geesez .eeeeewmceu eee: Ppea so m can» eees e_meew>wecw >_co .N euwm cewmmeeeem Ne mcweeee eca acmzeppep mcpces NN ece .mH .OH .e we mmcweeeem acmpe e>we cm N E Lee mpeeew>wecw we geese: ezh .N e.nme x_uceaa< 196 3v mm 33 mm 33 me D .Ewew 2: N 8v N ANV N o Aeeem Eeww eezm: neeumev meewp umewem AmV 4H Ame NH Ame NH m Anewesoeam we necmww lempmev ween» amewem 3v 3 AC 3 A3 3 w mace: AHV ea an ma va HH m xeeez Nemeceeem AHV m ANV m Avv e o memmmeo :v m A3 3 An: N N mewee mzecoz NN mnecoz ea mgecoz OH meecoz e ecwasoee e=m_a meweeem we weesez .eeeeewmcee ewe; wwep so m cane ewes eweeew>wecw cho .m>e>eem meew>eee cw eeeepceeece we: Eeew ewww some cw meweeem we geese: ecu pcemeeeew memezeceeee cw mezwe> .H eawm :ewmmeeoem aw eeoeeew muewe ucecmswee E m.H x H NN we >e>eem ecu ce eemee ewe muwemem .eeceeeeemwe mcwceee ece mcwppee m wepwm :ewmmeeeem we mcpces NN pmeww ecu we>e macweeeea acewe xwm aw geese: meweeem cw emcece ecw .m e_eew xweceee< 197 .meueuwweew e>ww oe; peuwoeceeows seem .A.s.e ooum op .e.e ooumv oewwee zen ece e wow mucesewemees :e oemee ewe wee—£3v pceseeamees e.>eo ece co oemee ewe meweueeeeeeu eueweemoem oce eeewwee cuee wow meewe>Aev -- e.~ -- -- e.w Aogmw_ eueewu we a may Scamp neeeepwem o.mm O.OOH o.mm o.mm O.OOH Aecewwcsm .wsw we a meo ecew_ weeewo mo.o no.0 eo.o mo.o mH.o Aevaemv eeewwee we Nae wee cewueweee>m N.om N.Nm w.om m.mN m.mm eeeueweeseu eeeww3m teem unawwxeo ceez N.Hm m.o¢ N.Nm m.om “.mm Aeveeeueeeeeeu eoewwem ucmwwzeo ceez cmewm oce weeeeezu cmewm oce cmewm oce wwem cexew muceeeeemeez weeewecu we: poem wwem eeem ewem meezw peeweeceeewz .cewmew meweeu :em ecu cw meawm eseww eeeewu we mcwceee oce mcwuuee ecu weawe ucemeee meezp peuwoezewews weewUwae xwm ecu we cowperweueeeecu weewmzce < .¢ ewoew xwoceee< 198 EPILOGUE This thesis has focused on forest, agricultural, and successional environments in the upper Rio Negro region of the Amazon Basin in Vene- zuela. Each study could be extended to provide a fuller understanding of the tropical environments investigated. The description of forest composition and structure (Part I) gave a static view of the tierra firme forest with a suggestion of processes provided tn! seedling establishment and mortality studies, gap mapping, and size-class distribution analysis. A full understanding of the tree population dynamics of the Amazonian forests can only be lachieved through long-term studies of large permanent plots. In addition, de- tailed investigations of dormancy, seed germination requirements, seed- ling growth requirements, breeding systems, and dispersal mechanisms need to be undertaken for the dominant species in Amazon forest communi- ties. Very little is known of the ecology of these species. No doubt, there are many different ecological strategies which provide solutions to the problems posed by a rain forest existence. Finally, we need to begin studying below ground forest structure and function. Certainly many of the approaches used to characterize forest structure above ground could be modified for use in describing below ground forest structure. Root morphospecies can often be distinguished on the basis of color, branching, bark, and exudate. It is important for our under- standing of forest functioning to know the rooting depth of trees, the degree of root species packing, and the amount (if any) of root 199 stratification. Parts II and III of this thesis focused on disturbed environments, both farming and successional. The Food and Agricultural Organization (FAO) of the United Nations estimates that about 40 percent of the tropical forests of the world have been destroyed in the last 150 years. There may be only small patches of this once massive belt of forest left by the end of this century. Meanwhile, populations continue to increase very rapidly in tropical countries. The annual increase in population size for Latin America is 2.7 percent; this is the highest growth rate for any major geographic area. Yet the volume of food production in Latin America is barely keeping ahead of population growth. As their population increases, South American countries are looking increasingly to the Amazon Basin as a source of crop and grazing land. Already Brazil has embarked on an ambitious program to resettle farmers from more arid regions into Amazonia. Before attempts are made to farm Amazonia on a large scale, the native agricultural system should be studied in detail. The indigenous farmers have a broad knowledge of farming based on generations of natu- ral experimentation. Agronomist and land-use planners have much to learn from these native farmers. Much applied research on a local scale is also needed. In the entire Amazon Basin (one-half the size of the United States) there are only ten experimental stations, some with only one researcher with as much as a high school education. Yet there is a great diversity of ecosystem types in the Amazon Basin. Annually flooded varzea forest occupies 1-2 percent of Amazonia. Flooding results in the deposition of nutrient-rich silt while eliminating pests and weeds. With the use of 200 dikes and annual crops, these areas offer high agricultural potential. In addition, isolated areas of high soil fertility are scattered among the senile oxisols and could be farmed intensively. There are also islands of savanna within the forest which could be effectively used for cattle grazing. Perhaps the best potential for development lies with the waterways. There are 20,000 miles of navigable rivers present in Amazonia and approximately 2,000 fish species present (the world's richest fish fauna). Improved fishing techniques and fish culture could probably multiply fish production several fold. The area with the least potential for' profitable lagricultural development is the tierra firme forest on nutrient-poor soils. However, the vast extent of this forest type demands that ways of use also be investigated. The killing of large non-economic trees in situ will increase the growth of species desirable for their fruit, latex, resin, oils, pharmaceuticals, etc. Canopy harvesting for tree species rich in protein should also be investigated. We need to operate from the premise that the continued clearing of Amazonian forests is inevitable as long as the environmental conse- quences of forest clearing are unclear. Hence, we need to design our research along applied lines. Studies of the effects of forest clearing on local climate should receive high priority. Succession studies should include investigations of the effects of small vs. large scale disturbance and bulldozer scraping on recovery. Life history studies of secondary species should be emphasized as much as studies of primary forest species. Through an understanding of the biology of the impor- tant species will emerge an understanding of observed processes. 201 The International Amazon Project at San Carlos de Rio Negro is providing one example of how basic research with an emphasis on applica- tion could help to shape developmental policy. Workers with a wide range of expertise have investigated the structure and function of the natural forest ecosystem and then measured the effects of disturbance on ecosystem functioning. The results are providing planners with a con- crete indication 01: the effects of human disturbance in the Rio Negro region. Attention now needs to be focused on ways in which this envi- ronment can be used without being irreparably destroyed. By restricting themselves to esoteric questions, biologists surren- der all power that they might have to temper development or to ensure that development be pursued wisely. Presenting government. decision makers with a list of 50 rare species, some of which are biologically interesting or aesthetically pleasing, will have no effect on a policy maker in a country struggling under the burden of rapid population growth, food and energy shortages, and runaway inflation. If, on the other hand, decision makers are provided with information on the ecolog- ical consequences of different development plans, they can use this information in land-use policy decisions. At the very least our obli- gation is to supply relevant facts, not emotional entreaties.