M, 55"] ”-5355... 5‘ 0"”. ‘5‘" "03“." (3" -n ‘v "' #355 . ‘5 3;"- ' 5‘" . El ' 5 51:, ‘HI' ' 5' 5"”: 1W -\'n . 15551975.- ‘ 31:33 ”3'45. qu ' '\‘ '36“ ‘ 3:- 33“ fix 3559335 5 (\‘yAL‘ J35??? '53:}; ""14"- '1' .-3 15:3- "3" ‘F-I 3'"; " we: Egg, :9 «533 V125" "‘ ‘ “35' ‘ ... 2‘ ,- ML" $53 " :55“? :31» 39:5“ '5“..% n}. 'l'_ u ' I: f ’ . ' "i ‘ 5 5'” ‘ ‘gfxttg 454 I 5,-5’ L‘. S o . .5 ”3-3 - ' _ :_ Z '3 «‘3 t‘ ‘ “I; (‘55' :3“ ‘25 ”if , w: “73;? A.§&\§? {- éfitigf‘cufi} 3J1; v‘¢\ 4!“ '6'. ‘ v . ._ ‘ at. Y" ‘Q ‘ .3 . ;_.vf 5. 11‘.' 4 ' l. -y‘ 5:9:- 5 :\‘_J.'. ~. We“ ' J “5,3. - ' . o ‘ I ‘ - — .- J . ‘ . —~.— 4.. “*Wfl—sr- 5: 5. | 555.4 _;l I,“ ‘ 1‘ I." :LVJIE if. (Lihkl' :H'V WWHIWWII/Will”MM/WT“ 0064 5451 u ’5 - ‘ ~. £FW 0'4 smart-ve- '1 This is to certify that the dissertation entitled Short-term Growth of an Undisturbed Tropical Moist Forest in the Brazilian Amazon presented by Niro Higuchi has been accepted towards fulfillment of the requirements for PhD degreein Forestry 55/ Major professor Date {/1 31/6? MS U i: an Alfmmm‘n Action/Equal Opportunity Institution 042771 MSU RETURNING MATERIALS: Place in book drop to remove this checkout from LIBRARIES All-(gulls. your record. FINES will be charged if book is returned after the date stamped below. ”UN 1 1 19.0.0. ‘ 17:1" '~¢’ , 1'- } J r- I ‘l I. v 1 I f ' l f‘rfi E-‘k’. “Anya “ 55. mt! 21 “3“}? E . 2%55 SHORT-TERM GROWTH OF AN UNDISTURBED TROPICAL MOIST FOREST IN THE BRAZILIAN AMAZON BY Niro Higuchi A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Forestry 1987 IHEHRACT SHORT-TERM GROWTH or AN uuoxsruaaeo TROPICAL uorsr FOREST IN THE BRAZILIAN amazon. by NIRO HIGUCHI The main objective of this study is to provide basic information for sustained yield management of the tropical moist forest in the Brazilian Amazon. This was done by quantification of diameter distributions, projections of Idiameter distributions and of tree mortality, and by development of short-term growth and yield models. The data for this study were collected from an undisturbed stand located some 90 kilometers north of Manaus, the capital of Amazonas State - Brazil. Three permanent plots were established in 1980 and remeasured in 1985. They are the control plots of a forest management experiment randomly replicated within an area of about 2,000 hectares of pristine Amazonian forest. In each 4-hectare plot (200 by 200 meters), all trees with dbh 25 cm or greater were tagged and their dbhfls were recorded in 1980. In 1985. all tagged trees were remeasured with special attention to ingrowth and mortalityu The number of trees, dbh and basal area of the study area averaged 158 trees/ha. ii 38 cm, and 20 mZ/ha, respectively - in 1980. Among three different hypothesized diameter distribution functions, the Weibull using the percentile approach best fit the observed data. 3 Since age and successive records from long-term permanent plots were not available, the first-order Markov chain approach was used to project diameter distribution and tree mortality and proved to be a realistic alternative. The exponential Lotka growth model was adapted to predict future volume as an alternative for the traditional growth and yield models, and it behaved satisfactorily. The volume for 1990 was also predicted by models based upon the volume estimated in 1985 in relation to the dbh measured in 1980. There was a strong correlation between actual volume and past dbh, but not between past diameter and diameter growth. iii To Inezita and Chico, and Maria my children, my wife - my friends iv ACKNOWLEDGEMENTS I wish to express my gratitudetto Dr. Carl W. Ramm. Chairman of my dissertation committee, for his insight, support, and guidance in the preparation of this work. I also wish to extend my gratitude to Dr. Lee M. James, Dr. Kurt S. Pregitzer and Dr. Peter G. Murphy for serving on my guidance committee and assisting throughout my doctoral program. I would like to extend my acknowledgements to Dr. Phu Nguyen, Mr. T. W. W. Wood, Dr. Jurandyr C. Alencar, Dr. Kurt S. Pregitzer, Dr. Lee M. James, and Dr. Carl W. Ramm. They provided helpful suggestions on an earlier version of specific chapters of this manuscript . I would like to pay special tribute to my wife, my kids and my "pessoal" from Chavantes and Chapeco for their encouragement, patience and supportive "rezas". I am indebted to many people whose friendship was important during the course of this voluntary exile. Thank- you to Antonio & Lucia, Josmar & Fernanda, Carlos, Steve Westin, Robert De Geus, Bill Cole, Don Zak and George Host. I would like to express my sincere gratitude to Luis Maurc>& Fatima Magalhaes for being my proxy in Brazil and for their patient support during this time. Special thanks are due to the "peaozada" of DST (Departamento de Silvicultura Tropical)-Aluizhm5Cabore, Jesus, Barrao, Caroco, Paulista, Armando and other anonymous helpers - who have been my great masters in the forest and particularly for their help during field data collection. I also wish to thank the group of DST's "pica-pan" - Fernando, Antenor, Jurandyr, Magalhaes, Benedito, Noeli and Joaquim - who played an important role during the preparation of this research project. I am also indebted to many people from other departments of INPA for their support. Thank-you to "turma" of administration and to Nakamura. I gratefully acknowledge the support of many staff members of the Department of Forestry of Michigan State University. Finally, my sincere appreciation to my country - Brasil - through CNPq (Conselho Nacional de Desenvolvimento Cientifico e Tecnologi'co) for financial and administrative support, and INPA (Instituto Nacional de Pesquisas da Amazonia) for inspiration. THANK YOU GOD ! vi TITLE . . ABSTRACT DEDICATION ACKNOWLEDG TABLE OF C TABLE EMENT S O O O C ONTENTS O O O 0 LIST OF TABLES . . . . . LIST OF FI CHAPTERS 1. INTRODU Scepe GURES O O O O CTION O O O O O of the Problem OF CONTENTS Statement of the Problem . . . . 2. LITERATURE REVIEW ON THE MANAGEMENT REGENERATION IN THE TROPICAL MOIST FORESTS 2.1. 2.2. 2.3. 2.4. 2.5. 2.5. 2.7. Overview . . . Introduction . . Tropical America Tropical Africa Tropical Asia . Tropical South Pacific . . . Conclusion . . . vii OF NATURAL page ii iv vii xiii 03010101 12 13 16 16 3. THE BRAZILIAN AMAZON . . . . . . . . . 3.1. Introduction . . . . . . . . . . . 3.2. Climate . . . . . . . . . . . . . 3.3. Soils . . . . . . . . . . . . . . 3.4. Vegetation . . . . . . . . . . . . Tropical moist forest on "terra firme" Inundated forests . . . . . . . . "Campina" and "Campinarana" . . . Tropical semi-evergreen forest . . "Cerrado" (Savannas) . . . . . . . 4. DESCRIPTION OF THE STUDY AREA . . . . . . 5. MODELLING THE DIAMETER DISTRIBUTION OF AN UNDISTURBED FOREST STAND IN THE BRAZILIAN AMAZON TROPICAL MOIST FOREST: WEIBULL VERSUS EXPONENTIAL DISTRIBUTION . . . . . . . . 5.1. Introduction . . . . . . . . . . . 5.2. Procedures . . . . . . . . . . . . The data . . . . . . . . . . . . . The diameter distribution functions The application of the functions . 5.3. Discussion of Results . . . . . . 5.4. Conclusion . . . . . . . . . . . . 6. A MARKOV CHAIN APPROACH TO PREDICT MORTALITY AND DIAMETER DISTRIBUTION IN THE BRAZILIAN AMAZON . 6.1. Introduction . . . . . . . . . . . 6. 2. PrOCEdures O O O O O O O O O O O O 20 20 21 23 25 27 30 32 34 35 39 47 47 48 48 49 52 S3 55 68 71 The Data . . . . . . . . . . . . . . . . . 71 The Markov model . . . . . . . . . . . . . 71 The application of the model . . . . . . . 73 6.3. Discussion of Results . . . . . . . . . . 75 6.4. Conclusion . . . . . . . . . . . . . . . . 76 7. SHORT-TERM GROWTH OF UNDISTURBED BRAZILIAN AMAZON TROPICAL MOIST FOREST OF "TERRA FIRME" . . 92 7.1. Introduction . . . . . . . . . . . . . . . 92 7.2. Procedures . . . . . . . . . . . . . . . . 93 The Data . . . . . . . . . . . . . . . . . 94 Model Development . . . . . . . . . . . . 95 7.3. Discussion of Results . . . . . . . . . . 97 7.4. Conclusion . . . . . . . . . . . . . . . . 102 8 0 CONCLUSIONS 0 O O O O O O O O O O O O O O I O O O 110 APPENDIX 0 O O O O O O O O O O O O O O O O O I O O O 113 L I ST OF REFERENCES 0 O O O C O O O O O O O O O O O 1 2 2 ix LIST OF TABLES Page 2.1. 1961 version of TSS - summary of operations . . 18 2.2. Malayan Uniform System (MUS) - summary of activities . . . . . . . . . . . . . . . . . . 19 4.1. Listed species for the NR management project . . . . . . . . . . . . . . . . . . . . 44 5.1. Diameter (cm) descriptive statistics for the study area . . . . . . . . . . . . . . . . . . 56 5.2. Parameter estimates used for diameter distribution - hectare basis . . . . . . . . . 57 5.3. Diameter distribution for all 3 sample plots together (Bacia 3) derived from 3 different methods . . . . . . . . . . . . . . . . . . . . 58 5.4. Diameter distribution for Bloco 1 derived from 3 different methods . . . . . . . . . . . . . . 59 5.5. Diameter distribution for Bloco 2 derived from 3 different methods . . . . . . . . . . . . . . 60 5.6. Diameter distribution for Bloco 4 derived from 3 different methods . . . . . . . . . . . . . . 61 6.1. Bloco 1 - Transition between states during a 5-year period . . . . . . . . . . . . . . . . 78 6.2. Bloco 2 - Transition between states during a 5-year period . . . . . . . .'. . . . . . . . 79 6.3. Bloco 4 - Transition between states during a 5-year period . . . . . . . . . . . . . . . . 80 6.4. Bloco 1 - Transition probability matrix from one state to another during a 5-year period . . 81 6.5. Bloco 2 - Transition probability matrix from one state to another during a S-year period . . 82 6.6. Bloco 4 - Transition probability matrix from one state to another during a 5-year period . . 83 6.7. Bloco 1 Two-step transition probability matrix . . . . . . . . . . . . . . . . . . . . 84 6.8. Bloco 2 - Two-step transition probability matrix . . . . . . . . . . . . . . . . . . . . 85 6.9. Bloco 4 - Two-step transition probability matrix . . . . . . . . . . . . . . . . . . . . 86 6.10. Bloco l - Projection for 1990 . . . . . . . . 87 6.11. Bloco 2 - Projection for 1990 . . . . . . . . 88 6.12. Bloco 4 - Projection for 1990 . . . . . . . . 89 6.13. Summary of one-step transition probability matrix (1985) . . . . . . . . . . . . . . . . 90 6.14. Summary of two-step transition probability matrix - projection for 1990 . . . . . . . . . 91 7.1. Basic distributional characteristics of the data used for individual volume regression equations . . . . . . . . . . . . . . . . . . . 104 7.2. Regression summary for volume estimation models . . . . . . . . . . . . . . . . . . . . 105 7.3. Characteristics of the data used as yield information and yield prediction . . . . . . . 106 xi 7.4. The frequency distribution of the three dominant families by status in 1980, mortality and ingrowth, and by periodic increment classes in cm . . . . . . . . . . . . . . . . . 107 7.5. Regression summary for increment models . . . . 108 7.6. Mean, standard deviation, minimum and maximum for each (a) dbh classes and (b) increment Classes 0 O O O O O O O O O O O I O O O O O O O 109 xii 3.2. 4.1. 4.2. 5.1. 5.3. 5.5. 5.6. LIST OF FIGURES Index map for the Brazilian Amazon vegetation map . . . . . . . . . . . . The vegetation of Brazilian Amazon . . "Ecological Management" Project area . Bacia 3 with 4 experimental blocks . . Bacia 3 - The relationship between the observed and estimated dbh frequencies, the Weibull MLE function . . . . . . . Bacia 3 - The relationship between the observed and estimated dbh frequencies, the Weibull PERC function . . . . . . . Bacia 3 - The relationship between the observed and estimated dbh frequencies, Exponential function . . . . . . . . . Bloco l - The relationship between the observed and estimated dbh frequencies by Exponential, Weibull PERC, and Weibull MLE . . . . . Bloco 2 - The relationship between the observed and estimated dbh frequencies by Exponential, Weibull PERC, and Weibull MLE . . . . . Bloco 4 - The relationship between the observed and estimated dbh frequencies by Exponential, Weibull PERC, and Weibull MLE . . . . . xiii Page 37 38 45 46 62 63 64 65 66 67 CHAPTER 1 INTRODUCTION During the past twenty years, the future of tropical forests has been a matter of international concern. Comprehensive reviews and evaluations are found in Gomez- Pompa et a1. (1972), Budowski (1976a), Leslie (1977), Brunig (1977), Spears (1979), Myers (1982), Myers (1983), Sedjo and Clawson (1983), and Lanly (1983). The discussion is polemic and most concerned scientists have been very pessimistic about the future of tropical forests, especially tropical moist forest (TMF). Nevertheless, there is one facet which all the diverse approaches share to some extent: the TMF ecosystem is very complex and fragile. Therefore, more studies are required for a full understanding and definition of its role for the region. In terms of forest management of TMF, sustained yield based on natural regeneration (NR) has been recommended by most scientists. Nevertheless, success with this silvicultural system is uncommon (Budowski 1976a). While scientists and technicians are discussing the problems of managing the tropical forest, about 20 hectares per minute - an area equivalent to Puerto Rico per month - of tropical forest are being deforestated, according to Murphy (pers. comnh). Myers (1982) pointed out that the principal causes of depauperation and depletion of TM? are timber harvesting followed by slash-and-burn agriculture in Southeast Asia, shifting cultivation in Africa, and cattle ranching in Latin America. In the Brazilian Amazon TMF, about 8 million hectares (approximately 2% of the total area) have been deforestated for the sake of agriculture and cattle ranching programs. By the end of this century more than 2 million hectares will be replaced by artificial lakes for energy generation. In addition, areas open to mineral exploration have also increased significantly. In the face of increasing pressure on the definition of role and vocation of the Brazilian Amazon TMF, in 1979 the Federal Government made a commitment to develop a forest policy for the region. All Amazonian research and educational institutions were engaged to support this policy. In the State of Amazonas two documents were produced at the same time, one by the University of Amazonas (EUA 1979) and another by the National Institute for Research in the Amazon (INPA 1979). Inspired by the worldwide concern on the use of TM? and forest policy, INPA initiated a research project. The project, entitled "Ecological Management of Dry-land (terra- firme)‘Tropical Moist Forest” was approved in 1979 by the Brazilian Federal Government. It was financed by INPA, the Interamerican Development Bank and FINEP (Brazilian Financial Agency for Research). The main objective of this project was to evaluate the impact of forest management practices on the local environment. Basic ecosystem research began in 1976 and the preparation for forest management experiment effectively began in 1980. This dissertation is based on observations of the forest management area over a 5-year period. Only trees with diameter at breast height (dbh) of 25 cm or greater were observed. This study was conceived to provide biological basis for sustained yield management based upon natural regeneration development. ScoEe of the Problem The Ecological Management project is as important to the Brazilian Amazon as the Hubbard Brook Ecosystem Study has been to the mixed-species forest ecosystem of Northeastern United States. In the 2,000-hectare project area, two major research studies have been carried out on ecology and forest management. The areas for each study are referred to as "Bacia 1" or "Bacia Modelo" and "Bacia 3", respectively for ecology studies and forest management experimentation. The initial results of "Bacia Modelo", including a collection of basic ecology research results, were documented by INPA in 1982 (INPA 1982). "Bacia 3" is the area involved in this dissertation. The results of this work will be used to help decision makers in prescribing silvicultural treatments for an experimental area subjected to a commercial timber harvesting. Statement 9; the Problem The present study will investigate three separate topics: quantification of diameter (dbh) distributions, projections of dbh distributions and of tree mortality, and development of short-term growth and yield models for natural unmanaged Amazonian forest. The specific objective of the diameter distribution study is to find out which distribution function best fits the observed data. Three hypothesized models were compared: Weibull by percentile approach, Weibull by maximum likelihood approach, and the exponential distribution functions. The second objective is to test the possibility of using the first-order Markov chain approach to project diameter distributions and to estimate tree mortality. The third objective is to explore alternative ways to model an undisturbed sample of TMF; Besides classical growth and yield models, the Lotka's exponential model was tested. CHAPTER 2 THE MANAGEMENT OF NATURAL REGENERATION IN THE TROPICAL MOIST FORESTS. 2.1. OVERVIEW This chapter reviews the management of tropical moist forests (TMF) using natural regeneration, with or without classical silvicultural systems. A diagnosis of the recent situation of the application and research on natural regeneration management, discussion of methods used in some countries, and perspectives of sustained yield management using natural regeneration are presented. 2.2. INTRODUCTION There is no doubt of the importance of natural regeneration for the management of TMF's. Very little is known of the response of these forests when subjected to intensive timber-oriented management used in temperate regions (Cheah 1978, Tang 1980, and Rio 1976). Without exception, all countries which contain TMF are still considered as "developing" or "less developed" countries (iJL, a mean GNP/capita about 10% of the North American GNP). Another common characteristic of these countries is the complex floristic composition of their predominantly broadleaf evergreen forests. Historically, natural regeneration management on a sustained basis began with the Malayan Uniform System (MUS) in Malaysia and the Tropical Shelterwood System (TSS) in Nigeria (Fox 1976L.These two systems, modified and improved with the passage of time and experience, have been used extensively'in most tropical countries.lk>be meaningful, natural regeneration management must be regarded as a continuous process of silvicultural treatments to favor economically desirable species. According to Rio (1979), the objective of most treatments is the perpetuation of the existing stands by the replacement of exploited forests without a profound alteration of the characteristic structure of the forest. This review divides the tropical world into tropical America, tropical Africa, tropical Asia, and the tropical south Pacific. The current situation of natural regeneration management and its perspectives are presented separately for each region.‘The term tropical moist forest is based on the Holdridge classification: biotemperature above 24° C and annual precipitation between 2,000 and 4,000 mm. 2.3. TROPICAL AMERICA According to Budowski (1976b), there is no example of mixed TMF in the American tropics being managed on a sustained yield basis. In terms of research, however, Brazil (since 1980) and Suriname (since 1967) have commenced studies to test the possibility of TMF management on a sustained yield basis using natural regeneration. Venezuela started a similar project in the middle of the 19703, but no progress beyond initial field establishment of the experiment and the collection of pre-harvesting data was made. Recently Peru also entered the natural regeneration management era. With the assistance of British and Canadian technical aid programs, the Honduran Forest Service will initiate studies into sustained yield management of the TMF resources using natural regeneration (Wood, pers. comnn). In Costa Rica, sustained yield management has been planned for the Nosara and Parrita river basins, with the assistance of FAO (Food and Agriculture Organization) (Wood, pers. comm.). In Dominica, between 1968 to 1972, an area of approximately 60 hectares was logged and planted with desirable tree species. After about 3 years it was found that this operation was very expensive to maintain, mainly due to the vigorous growth of climbers. Therefore, the option with natural regeneration management was considered (Bell 1976). In Puerto Rico, a timber-management plan was completed in 1966 (Wadsworth 1970). This plan consisted of natural regeneration treatments of 2,700 ha during the next 4 decades. (a) BRAZIL The concept of managing the native forests under a system of sustained yield was introduced by FAO experts in 1958 in Santarem (State of Para) through an agreement with the Brazilian Government. In Manaus (State of Amazonas), researchers at INPA in 1964 initiated studies on enrichment of natural forests, phenology'of tree species, and nursery and plantation management of native and exotic species (Higuchi 1981a). In Santarem natural regeneration research was first carried out fortuitously in 1960 when, after an area was burned for species trial site preparation, copious regeneration of Goupia glabra Aubl. appeared spontaneously. This area is still under observation by researchers of EMBRAPA (Brazilian Enterprise for Agricultural and Animal Husbandry Research) and SUDAM (Superintendency of Development of the Amazon region). Today, besides Goupia glabra Aubl., species such as Vochysia maxima Ducke, Didymopanax morototoni (Aubl.) Decne & P1anch., Manilkara hubggi (Ducke) Standl., and Simaruba amara Aubl. are abundant in an adjacent area. Recent work with natural regeneration management in Santarem is being carried out over blocks of 100 hectares. The forest is harvested with diameter limits of 45 and 55 cm dbh for commercial species after climber cutting and underbrushing. The objectives of this project are to determine the effects of different levels of harvest intensity on the residual stand and regeneration, and to evaluate the growth and yield under natural regeneration management. In Manaus the research with natural regeneration management effectively started in 1980 under an agreement among INPA (National Institute for Research in the Amazon), the Interamerican Development Bank, and FINEP (Brazilian Financial Agency for Research). The main objective of this investigation was to test the possibility of managing the TMF of the region under a system of natural regeneration. A second objective was to use theidata to determine felling cycles along with forecasts of yields by species. Within the experimental blocks (400 by 600 meters), harvesting will be carried out as the main silvicultural treatment. In designated sub-blocks (200 by 200 m) felling intensities will be applied to remove various levels of the basal area of some 40 listed species, 25 cm dbh and above. This project is based on multidisciplinary research involving all departments of INPA (Ecology, Botany, Wood Technology, Pathology, Agriculture, Chemistry and Zoology), which will give scientific support to the Department of Tropical Silviculture. The total area of this project is about 2,000 hectares while the area for silvicultural experimentation is 96 hectares, comprising 4 separate blocks of 24 hectares each. The treatments to be randomized in each block are: (1) control; (2) removal of 25% of exploitable basal area (b.a.); (3) removal of 50% of exploitable b.a.; (4) removal of 75% of exploitable b.a.; (5) removal of 100% of 10 exploitable b.a.; and (6) removal of 50% of exploitable b.a. with enrichment. In each 2 hectare plot a 1 hectare (100 by 100m) permanent sample plot will be established, in which the following studies will be carried out: growth of the residual stand of listed species; recruitment and development of seedlings of the listed species; survival and growth of listed species; growth and mortality of poles and saplings; and studies of increment to determine felling cycles. (b) SURINAME Research into the management of TMF resources was initiated by the Suriname Forest Service in the 19508. The Malayan Uniform System was used but it was discontinued in the early 19603 due to the high costs of silvicultural treatments, the long rotation (70-80 years), and the lack of species with the silvicultural characteristics of the dipterocarps of SE Asia. The need for a management system suited to the conditions of Suriname was met in 1967 under the auspices of the CELOS (Center for Agriculture Research in Suriname). Its objectives were to find an economically and technically feasible method to stimulate the valuable timber species increment after a light harvesting, to improve the regeneration of the valuable species, and to build a forest with sustained yield. Here, light harvesting meant the 11 removal of some 30 trees from the 25-ha experimental area. Besides the classic silvicultural treatments, a refinement was used wherein all non-valuable trees (non-commercial species) were killed with arboricide (2,4,5-T ester, 5% solution in diesel oil) using a‘diameter limit of 20 to 40 cm. In this experiment the liberation treatments were: (1) elimination of competing lianas and non-valuable trees around the leading desirable tree selected on an area of 5 by 5 m; (2) elimination of competing species around the desired species with a diameter criterion (3 to 5 cm dbh), disregarding the location of the selected trees: and (3) elimination of competing species around the desired species in a strip 2 m wide, spaced 12.5 m apart, to provide accessibility. In the sampling area (16 ha), over 1,000 valuable trees larger than 15 cm dbh are being measured yearly. Smaller valuable trees are recorded in a 17.5% subsample using 40 circular plots of 1,000 sq.m each. As a provisional result, de Graaf (1981) reported that the annual volume increment is 2.1 cu.m./ha for valuable trees above 15 cm dbh. According to Johnson (1976), the mean annual growth of the TMF's is about 1 to 3 cu.m./ha in South East Asia, 2 cu.m./ha in Nigeria, and 2.9 to 4.3 cu.m./ha in the Philippines (Dipterocarp forest). Even though there is not too much detail in terms of tree size, in a general sense the forests in Suriname are showing almost the same response to the 12 natural regeneration management as reported elsewhere. 2.4. TROPICAL AFRICA According to Lowe (1978), the tropical shelterwood system (TSS) was a major management preoccupation in Nigeria during the 19503. Altogether about 200,000 ha of forest land were treated under this system. It was intended to obtain sustained or improved yields. The TSS consists of canopy opening to promote survival and growth of seedlings of valuable species. This system has been changed and improved since its introduction, and the last version of TSS in 1961 is presented in Table 2.1. However, TSS has been abandoned in Nigeria, primarily on the economical grounds that it did not make sufficiently intensive use of the land to compete with other forms of land use (Lowe 1978). Nevertheless Rio (1976) pointed out that economically, TSS is more profitable than plantations if the analysis is correctly applied without bias. He related that too often the forest management analyst seems to survey the list of variables and select only those that will contribute positively to the desired end. It seems certain that silvicultural arguments did not contribute to the abandonment of TSS in Nigeria. In Ghana the TSS was tried on an experimental scale between 1948 to 1960. It was found to be unsuitable because of the high maintenance costs and was abandoned (Britwunn 1976). According to this author, the selection system was 13 found to be suitable for Ghana forests although it induced only moderate regeneration. The treatments for this system were: (a) stock survey to map all economic trees with dbh > 66 cm; (b) weeding, cutting and poisoning all climbers and worthless trees which interfere with the development of young economic trees (10 < dbh < 47 cm); and (c) selection of trees to be felled from stock maps. 2.5. TROPICAL ASIA A common characteristic in this region is the significant presence of species of the Dipterocarpaceae. This family contains the most important tropical hardwood timber species. Other important species also occur in this region, exp, teak (Tectona grandis) in Burma, teak and Pinus merkusii in Thailand, Pinus kesiya in the Phillipines, and Pinus merkusii in Indonesia. According to Tang (1980), natural regeneration is the basis for the regeneration of TMF in the region. The silvicultural systems which have been developed for this region are the Philippine Selective Logging System and the Indonesian Selection Felling System for advanced growth, and the Malayan Uniform System (MUS) and Indonesian modified MUS - for seedling regeneration. Table 2.2 presents the sequence of activities necessary for the MUS. The MUS is, in fact, the most popular silvicultural system in tropical Asia. It is mainly used in lowland l4 Dipterocarp forests when.adequate reproduction.i3 already established. There are restrictions in applying it in hill forests where enrichment planting is often necessary (de Graaf 1981L.In West Malaysia about 300,000 hectares have been managed with MUS up to 1976. Cheah (1978) discussed the differences between the new selective felling system and the MUS or the modified MUS. He determined that the first one is more appropriate for dipterocarp forests in Peninsular Malaysia. The selective felling system is a modification of the MUS, consisting of the MUS plus the following operations: pre-felling inventory which includes all trees below and above 15 cm, climber cutting, and marking of residual trees for retention. In Sarawak, the liberation thinning system was introduced in 1975 by the Forest Department to evaluate the effects of different intensities of reduction of stand basal area as an alternative way to manage the natural regeneration (Higuchi 1981b). This system seeks to eliminate only trees which restrain the growth of a selected tree (Hutchinson 1980). Modified MUS and removal of relics (removal of all trees with dbh > 60 cm regardless of species) has also been tested in Sarawak (Lee 1982). In Sabah, the modified MUS was abandoned and replaced in 1971 by the minimum girth system, which retains the basic principles of the MUS (Chai and Udarbp 1977). This new system includes three silvicultural treatments at three different occasions. The first involves climber cutting two 15 years before felling operation to reduce the risks of felling damage» The second combines the natural regeneration inventory by linear sampling of milliacre plot and poison girdling to eliminate competition. The third silvicultural treatment involves the natural regeneration inventory by linear sampling half-chain survey and a liberation treatment. Chai and Udarbp (1977) concluded that the second treatment should be modified to suit the present conditions of logging in Sabah, and they recommended alternative research to reduce logging damage. In Indonesia, since 1972, the Indonesian selective logging system has been used as a means of converting the virgin forest into an enriched managed stand (Soekotjo and Dickmann 1978). This system consists of removal of trees with dbh > 50 cm to favor the growth of residual trees and seedlings of desirable species. Approximately 25 young and healthy overstory trees per hectare are usually left. After 4-5 years, the initial results have shown that the Indonesian system seems to be appropriate for forest management of Indonesian TMF (Soekotjo and Dickmann 1978). In the Philippines, the selective logging system has been used in managing the dipterocarp forests since the 19503” Specifically, this system assures a future crop of timber and forest cover for the protection and conservation of soil and water after the removal of the mature, overmature and defective trees (Virtucio and Torres 1978). According to these authors, the preliminary evaluation of 16 the selective logging has shown positive results for the management of dipterocarp forests. Other countries such as India, Burma and‘Thailand are using the selective felling system to manage their forests (James, pers. comm.). Burma contains 75% of world's stands of natural teak. In India and Thailand, many species of dipterocarp and teak are very important to the country's forest economy. 2.6. TROPICAL SOUTH PACIFIC Natural regeneration management was attempted in Fiji during the 19603. Five years later this project was abandoned (Higuchi 1981b) because the first results were not encouraging. Today the priorities in Fiji are planting Pinus caribaea var. hondurensis and management of Mahogony (Swietenia macrophylla) plantations. In Papua New Guinea, forest plantations seem to be the only long-term alternative for its forests and for the supply of its forest industries (Hilton and Johns 1984). 2.7. CONCLUSION The utilization of natural regeneration as a tool for forest management on a sustained yield basis in the TMF mainly for dipterocarp forests is certain in almost all southeast Asian countries. Although the Tropical Shelterwood System (TSS) was abandoned in Nigeria, there exists a future 17 for natural regeneration as a way to manage the TMF, mainly in well-stocked high forests (Kio 1976L.In South America the first results of research recently established in Suriname and Brazil have shown that natural regeneration management is economically feasible and ecologically acceptable. The greatest obstacles to success with natural regeneration management in tropical countries are the lack of continuity in funding, the inadequacy of the staff, and sometimes political factors. Tang (1980), for example, considers that the success of natural regeneration management depends on the implementation and monitoring phases which can be carried out only with a well-trained staff. According to Fox (1976), all mentioned problems are typical in developing countries, primarily because the anxiety to show progress is more important than anything else. Unfortunately, natural regeneration management requires long periods of time before results are known. It is very important to maintain a cautious approach in using the tropical moist forests because, according to Myers (1983), very little is known about these ecosystems. It will be better to find that we have been vaguely right than certainly wrong. 18 Table 2.1: 1961 version of TSS - summary of operations. YEAR INSTRUCTIONS -5 Op.I Milliacre sampling Op.Ia Demarcation Op.II Climber cutting and cutting uneconomic saplings if advance growth is inadequate Op.III Climber cutting only Op.IV 2nd. milliacre assessment following Op.II 0p.V Poisoning of shade casting trees of lower and middle storeys -4 (if Op.II in year -5, then Op.IV followed by 0p.V) -2 0p.VI Re-demarcation -1 0p.VII Climber cutting 0 Harvesting 8 Op.Ix Re-demarcation Op.X Climber cutting Op.XI Removal of Shelterwood 15 Op.XII Re-demarcation Op.XIII 1/2 chain linear sampling Source: partially reproduced from Lowe (1978). 19 Table 2.2: Malayan Uniform System (MUS) - Summary of activities. ========================================================2= ACTIVITY DESCRIPTION Pre-Felling Except in cases where enumeration data Inventory on trees 39 cm dbh and above is needed for premium determination only. Pre-Felling Treatment of bertam in hill forest only. Treatment Felling Limit All commercial and utilizable species with dbh = 45 cm and above. Tree Marking May or may not be done by forest officers. Directional felling incorporated but essentially for checking completeness of felling only. No marking of residuals for retention. Roading Layout Prescribed specifications and Construction Post-Felling To determine fines on trees unfelled, Inventory royalty on short logs and tops, damage to residuals. Silvicultural To determine correct treatment. Sampling Source: Cheah (1978). CHAPTER 3 THE BRAZILIAN AMAZON 3.1. INTRODUCTION The Amazon region includes the following countries in South America: Brazil with 500 million (mi) hectares (ha), Bolivia (65 mi.ha.), Colombia (62.5 mi.ha.), Peru (61 mi.ha.), Guyana (21.5 mi.ha.), Venezuela (17.5 mi.ha.), Suriname (14.5 mi.ha.L, and French Guyana (9 mi.ha.) (Volatron 1976). The name of this region comes from the Amazon Basin and its main river, the Amazon, which originates on Mt. Huagra in Peru at 5,182 meters above sea level (a.3.l), 195 km from the Pacific shore. According to Palmer (1977) in the first 965 km from its source, the Amazon river drops 4,876 m while in the remaining 5,785 km the fall to sea level is only 306 m. In the Brazilian territory the area of influence of the Amazon Basin includes the following regions: Acre (AC), Rondonia (RO), Amazonas (AM) and Para (PA) states, part of Mato Grosso (MT), Goias (GO) and Maranhao (MA) states, and two federal territories, Roraima (RR) and Amapa (AP). Hereafter, this area will be referred to as the Brazilian Amazon or simply as the Amazon. This area is under geographical and political influence of Amazon Basin, even though it is known that the Amazon forest ecosystem covers 20 21 around 3/5 of this area. The Amazon region corresponds to about 55% of the Brazilian territory, but its population represents only 10% of its total. Fig. 3.1 shows the location of Brazilian Amazon within South America. 3.2. CLIMATE The Brazilian Amazon region is characterized by homogeneity in climate conditions. In the interior of the forest of this region the microclimate is much more equable, especially on the ground itself where no direct sunlight falls (Walter 1979). Coastal and in-land temperatures do not differ greatly. Belem, some 100 km from the sea, has a mean annual temperature of 25° C. Manaus, nearly 1,000 km up- river on the Amazon, has an equivalent of 27° C and Taraqua some 2,000 km in-land has a mean annual temperature of 24.9° C. The maximum temperatures are around 37 to 40° C with a diurnal variation of some 10 degrees. According to Salati & Vose’(1984), however, an important phenomenon to be considered is the "friagem' or cold spells that occur when air masses from the South Polar region hit the central and western parts of Amazon, causing the temperature to fall to about 14° C.‘This phenomenon occurs during the winter in the states of Acre and Rondonia, and in the southern parts of Amazonas state. Rainfall shows greater variability than temperature across the region. There is approximately 3,000 mm annual rainfall on the coast, 3,497 mm at Taraqua (less than 100 km 22 from the limit boundary between Brazil and Colombia), 1,504 mm in Boa Vista (the Capital of Roraima), and 1,670 mm in Conceicao do Araguaia. According to Ranzani (1979) the dominant climatic types (Koppen classification) in the region are Af (coolest month above 18° C and constantly moist) and Aw (coolest month above 18° C and dry period during the winter). Including air moisture regime (presence of dry period with its duration), IBGE (1977) identified five climatic zones: (a) Equatorial very moist without dry period: covers the northwest Amazon (about 30% of Amazonas state) and Belem (the Capital of Para state). (b).Equatorial very moist with short dry period.(less than one month): covers the surrounding areas of type (a) (about 30% AM and 25% of AC). (c) Equatorial moist with dry period ( one to two months): covers the western-center and the southeast of 4Amazon (50% of AC, 30% of AM, 30% of RR, 30% of PAmand 10% of north of MT). (d) Equatorial moist with dry period (three months): covers the southwest and the eastern-center of Amazon (10% of AM, 100% of RD, 70% of PA, 40% of RR, 70% of AP, 10% of G0, 40% of MT and 40% of MA). (e) Tropical semi-moist with dry period (four to five months): covers part of RR and south and southeast of Amazon (30% of RR, 50% of MT, 90% of Goland 60% of MA). 23 3.3. SOILS The soils in the Brazilian Amazon are very old, reaching back as far as the Paleozoic era. Basically the region is composed of a sedimentary basin (Amazon Valley) located between two shields (Guiana and Brazilian). According to IBGE (1977) these two shields are composed of igneous Precambrian and metamorphic rocks from Cambrian- Ordovician, They contain some spots of sediments from the Paleozoic/Mesozoic (60 to 400 million years ago). There are two Paleozoic strips of sediments where Devonian shales predominate, one at the Guiana shield boundaries (east of the 60 degrees of longitude) and another at the Brazilian shield boundaries (east of the 57 degrees of longitude) 30 to 50 km wide (Schubart & Salati 1980). The Amazon Valley is formed by fluvial sediments of coarse texture deposited from the Cretaceous to the Tertiary periods, originated from the erosion of the Precambrian shields (Schubart & Salati 1980). In summary, this is the evolutive process of formation of ”terra firme" (non-flooded ground). Another important formation in the‘Amazon region is the "varzea", or temporarily flooded land. According to Schubart 5 Salati (1980) the ”varzeas" are constituted by the Holocene flood plains of the Solimoes river (Amazon river above Manaus) and.the Amazon as well as their white water tributaries. "Varzeas" are the most recent formation in 24 from the deposition of sediments transported by the rivers (Ranzani 1979). This kind of formation represents only 1.5% of the region, but its high agriculture productivity is significant to the Amazon economy. Ranzani (1979) pointed out that its fertility is not constant, as it depends upon the materials incorporated annually by flooding. According to Cochrane & Sanchez (1980) the following soil orders are found in the Brazilian Amazon: Oxisol "yellow Latosols" (45.5%), Ultisols "red yellow Podzolics" (29.4%), Entisols "azonal, alluvial soils" (14.9%), Alfisols "gray brown Podzolics" (4.1%), Inceptisols "hydromorphics, humic gley soils" (3.3%), Spodosols "Podsols or giant tropical Podzols" (2.2%), Mollisols "Chernozem, humic gley soils" (0.8%), and Vertisols "grumusols" (0.1%). In general, the soils are extremely poor in nutrients and very acid. In fact, almost the entire nutrients amounts required by the forest are contained in the aboveground biomass (Walter 1979). Cochrane 3 Sanchez (1980) pointed out that only about 6% of Amazon has well drained soils with relatively high natural fertility. These soils are found in Altamira (Para state), Porto Velho (the capital of Rondonia state) and Rio Branco (the capital of Acre). Ranzani (1979) stressed that few Amazon soils are suitable for agriculture, grazing or even for reforestation. 25 3.4. VEGETATION Using the Holdridge classification and the climatological observations of IBGE (1977), there are two major life zones in the Brazilian Amazon. These are the tropical moist forest (mean annual biotemperature above 24° C and mean annual precipitation of 2,000 to 4,000 mm) and the tropical dry forest (mean annual biotemperature above 24° C and mean annual precipitation of 1,000 to 2,000 mm). According to Schubart & Salati (1980) about 8% of the Amazon is under secondary vegetation and/or agricultural activitiesu‘Within the tropical moist forest only limited areas on the coast, along the major tributaries of the .Amazon and along the Amazon River have been used for food production (Tosi 1983). The most significant deforestation is located in the tropical dry forest, mainly along the Belem-Brasilia highway, southern portions of MT, and in Rondonia and Acre states. It is well known that the main characteristic of the .Amazon forest is its considerable vegetational diversity, although at first sight it appears to be rather uniform (France 1974). The Amazon region is reported to contain about 6,000 different species of plants, of which one-third are tree species growing to commercial size. The distribution of these trees varies tremendously, particularly in relation to soils and topography. There are many theories to explain this diversity. 26 According in: Prance (1974) the genetic isolation into separate populations after a long dry period in the late Pleistocene and post-Pleistocene was a major factor in the evolution of the species diversity within the lowland forest of Amazon. Schubart & Salati (1980) pointed out that the large number of species and the complexity of their interrelatioships are a function of evolutionary history which can be broadly described by three main categories of factors: proximal (or geographic factors), interactions within.the communities themselves, andtdynamic instability. In spite the complexity and diversity of the Amazon vegetation, a broad classification - based on the Holdridge system plus part of the classification presented by Prance (1974) - will be presented for the two major life zones (Fig. 3.2). 1. Tropical moist forest 1.1. Tropical moist forest on "terra firme" 1.2. Inundated forests: "varzea" (seasonally flooded forest) and "igapo" (permanently water-logged) lu3. Forest on white sand soils or spodosols: "Campina" and "Campina r ana" . 2. Tropical dry forest 2.1. Amazon tropical semi-evergreen forest 2.2. "Cerrado" (Savannas). 27 Tropical moist forest on "terra firme” The superior stratum of this forest type is composed of trees whose heights may vary from 30 to 40 meters. Only a few species can grow above this height. Exceptions are Cedrelinga catenaeformis and Dinizia excelsa with, on some sites, 50 and 60 meters height respectively. For trees with dbh greater than 20 cm, the forest on "terra firme" has a mean commercial volume of 150 to 300 cu.m./ha and a basal area of 20 to 40 sq.m./ha. IBGE (1977), Braga (1979), Silva et a1. (1977), Higuchi et al. (1983a), and four forest inventories carried out by Department of Tropical Silviculture of INPA (National Institute for Research in the Amazon) in different parts of Amazon are the guide for the description of floristic composition of this type of forest. Here the emphasis is only (”1 those species which can characterize specific regions. In a broad sense the following phanerophytes can be considered as typical species of "terra firme": Dinizia excelsa, Bowdichia nitida and Cedrelinga catenaeformis (Leguminosae), Anacardiug gigagtggm (Anacardiaceae), Bertholletia excelsa "Brazilian nut" (Lecythidaceae), Caryocar villosum (Caryocaraceae), Minquartia guianensis (Olacaceae), and two species of Palmae, Oenocarpus bacaba and Astrocaryum mumbaca. The characteristic epiphytes of ”terra firme" are: several species of Phillodendron (Araceae), Clusia insignis and Clusia grandiflora 28 (Guttiferae), several species of Operculina (Convolvulaceae) and Bauhinia macrostachya (Leguminosae). IBGE (1977) divided the ”terra firme" into seven sub- regions to show the characteristic tree species of these areas, in contrast to the previous group of species which is common to all sub-regions. The sub-regions are: (a) Delta of Amazon river: In this area the following species characterize the "terra firme": several species of Parkia, Vatairea guianensis and several species of Ormosia - (Leguminosae), Erisma fuscum and Vochysia guianensis - (Vochysiaceae), several species of Manilkara and Pradosia - (Sapotaceae), and several species of Virola - (Myristicaceae). (b) Northeast Amazon: several species of Micropholis, Ecclinusa, Chrysophyllum and Manilkara - (Sapotaceae), several species of Eperuaz Swartzia, Ormosia, Tachigalia and Inga - (Leguminosae), Goupia glabra (Celastraceae), several species of Iryanthera - (Myristicaceae), and several species of Qualea - (Vochysiaceae). (c) Tocantins & Gurupi rivers: Swietenia macrophylla "Mahogany", Cedrela odorata and 935323 guianensis - (Meliaceae), Hevea brasiliensis - (Euphorbiaceae), Platymiscium duckei, Vouacapoua americana, and several species of Piptadenia and Peltogyne - (Leguminosae), Cordia 29 goeldiana - (Boraginaceae), Mezilaurus itauba - (Lauraceae), several species of Astronium - (Anacardiaceae), and Jacaranda copaia - (Bignoniaceae). (d) Xingu and Tapajos rivers: The floristic composition of this sub-region is almost the same as the sub-region (c). (e) Madeira and Purus rivers: Hymenolobium excelsum, Peltogyne densiflora, several species of E2353; and Elizabetha - (Leguminosae) Swietenia macrophylla and Carapa guianensis - (Meliaceae), Euterpe oleracea - (Palmae), several species of Theobroma - (Sterculiaceae), Cordia goeldiana - (Boraginaceae), Manilkara huberi - (Sapotaceae), Cariniana micrantha - (Lecythidaceae), Hevea brasiliensis. (f) Occidental "Hileia" - Jurua to Brazilian territory limits: several species of Theobroma "Cocoa tree" and numerous palms, and several species of Leguminosae, Myristicaceae, Bombacaceae, Lauraceae, Vochysiaceae and Rubiaceae. (9) Northwestern "Hileia" - Negro to Trombetas river: Leguminosae is the dominant botanical family in this sub- region, mainly species of genera Dimorphandra, Peltogyne, Eperua, Heterostomon and Elizabetha. The genena Dicorynia, Aldina, Macrolobium and Swartzia are endemic:in this sub- region. Other characteristic species are: Carapa guianensis, Cedrela odorata and Cariniana micrantha. 30 (h) Acre: Torresea acreana - (Leguminosae), Hevea brasiliensis, Swietenia macrophylla and several species of Cedrela. Inundated forests This type of forest represents an area of about 7 million hectares, or 1.5% of the Amazon region (Braga 1979). Within this type, the best and the biggest portions are the seasonal "varzea" and tidal "varzea". They are considered very important for the development of the Amazon region because of their soil quality and also because they supply most of the raw material to forest industries. Prance's (1980) key for the classification of inundated forest types was used to describe the vegetation. The author pointed out that the three different types of water (white, black and clear) of Amazon basin are very important to the floristic composition. There are peculiar species for specific water types, mainly due to differences in acidity and nutrient contents. For example, Victoria amazonica is found only in white water. The seven inundated forest types are: (a) Seasonal ”varzea": this type is characterized by a relatively high aboveground biomass and represents the most common type of inundated forests. According to Prance (1980) its herb layer is rich in species of Heliconia (Musaceae) and Costus (Zingiberaceae). The following species can 31 characterize this type of forest.cPrance 1980, Braga 1979, and IBGE 1977): Carapa guianensis, several species of Cecropia - (Moraceae), Ceiba petandra - (Bombacaceae), Couroupita subsessilis - (Lecythidaceae), Euterpe oleracea - (Palmae), Hura crepitans and Piranhea trifoliata - (Euphorbiaceae). (b) Seasonal "igapo" - swamp forest: Usually dominated by sand soils supporting a vegetation much poorer than the seasonal "varzea". According to Braga (1979), the vegetation is very specialized with little specific diversity and very rich in endemism. Characteristic species of this type are: Aldina latifolia - (Leguminosae), several species of Couepia - (Lecythidaceae), some species of Licania - (Chrysobalanaceae), and Macrolobium acaciifolium - (Leguminosae). (c) Mangrove: This type is typical in the estuary of the Amazon. According to Braga (1979) the mangrove type involves an area of about 100,000 hectares with a low and uniform aboveground biomass. This type is characterized by the presence of Avicennia nitida (Verbenaceae), Laguncularia EEEEEQEE (Combretaceae) and BEEEQREQEE ‘1‘. “91.2. (Rhizophoraceae). (d) Tidal "varzea": This type is very similar to the seasonal "varzea" in both species composition and aboveground biomass. Prance (1980) stressed that where the 32 tide is daily, the vegetation is similar to the swamp. Where the spring tide is dominant, is more similar to the seasonal "varzea". The most common palm species are: Mauritia flexuosa, Euterpe oleracea, Raphia taedigera and Manicaria saccifera. Species like Virola surinamensis (Myristicaceae), Ceiba petandra, Mora paraensis, Pithecolobium huberi, Derris liEiEQliiL EXEEBEEE 222222 and lflflé BEEEQQBE ' (Leguminosae), and Tabebuia aquatilis (Bignoniaceae) have also a significant presence in this type of forest. (e) Flood plain: Species from seasonal "varzea" and also from "terra firme" can be found in this forest type. (f) Permanent swamp forest: According to Prance (1980) there are few permanent swamp forests or permanent ”igapo” in the Amazon. This type contains very few species, although trees are usually very big and similar to their counterparts of seasonal "varzeaF. The canopy is usually more open than the seasonal "varzea" and the ground is rich in Cyperaceae. "Campina" and "Campinarana" The soil of these two types is almost the same, but their floristic composition and the stand density are different. According to Lisboa (1975), the tropical moist forest on "terra firme" is commonly interrupted by ”islands” with contrasting tree size, structure and physiognomy. Such ”islands” are oommon.in the Rio Negro river basin and in 33 other areas north of the Amazon river, but almost absent in the southern parts of this river. "Campina" and "Campinarana" are evergreen. (a) "Campina":‘According to Braga (1979), this forest type presents a low aboveground biomass with sclerotic vegetation, and covers an area of 3.4 million hectares (0.7% of Amazon). Although the "Campina" soils are excessively drained, acid and poor in nutrients, there is no problem with water availabilityu Lisboa (1975) pointed out that without this characteristic the actual vegetation could be replaced by Gramineae, Cyperaceae and small shrubs. Thee"Campina" floristic composition is variable, but the following species could be considered as characteristic species of this forest type (Braga 1979): Aldina heteroghylla and Ormosia costulata - (Leguminosae), Clusia aff. columnaris (Clusiaceae), Glycoxylon inophyllum (Sapotaceae), Humiria balsamifera (Humiriaceae), Matayba 923 a (Sapindaceae), and Protium heptaphyllum (Burseraceae). According to Lisboa (1975) the epiphytes are abundant in "Campina" because the high intensity of light, e.g., many genera of Orchidaceae (Sauticaria, Octomeria, Rodrfigzia and Maxillaria) and also many species of Bromeliaceae (Aechmea and Tillandsia). (b) "Campinarana" (false "Campina"): In this forest type the trees are larger and the stands are denser in 34 comparison to the "Campina" type. According to Braga (1979), "Campinarana" represents an area of approximately 3 million hectares distributed as small islands in the central Amazon and as bigger portions north of Amazon river (Negro basin). "Campinarana" is also very rich in epiphytes, mainly Hymenophyllaceae and Bryophytae. The following species characterize this forest type (Braga 1979): Aldina discolor, Eperua leucantha and Hymenolobium nitidum - (Leguminosae), Bactris cuspidata (Palmae), Clusia. spathulaefolia (Clusiaceae), Qggm§_ gatigga£_ (Apocynaceae), ‘ggggg rigidifolLa (Euphorbiaceae), Sacoglottis heterocarpa (Humiriaceae) and Scleronema spruceanum (Bombacaceae). Amazon tropical semi-evergreen forest This forest is considered as a transition from Savannas and tropical semi-evergreen to tropical moist forests. It occurs in part of MA, portions of eastern, southern and northern PA, northern MT, almost 90% of Rondonia, portions of AC, small portions at northern and southern AM, a significant portion of the federal territory of Roraima and a small portion of Amapa. In general, according to IBGE (1977), the trees are relatively tall, with medium diameter and under-developed crowns. Lianas are abundant, but epiphytes are almost absent. The species most characteristic of this forest type is Orbignya martiaga (Palmae). Hevea brasiliensis is abundant mainly along the southern tributaries of the Amazon 35 river. In the MA portions and eastern PA the species which characterize this forest type are: Bertholletia excelsa, Ceiba petandra, Vouacapoua americana, Castilloa ulei (Moraceae), Hymenaea courbaril (Leguminosae), Lecythis paraensis (Lecythidaceae), and several species of Palmae, e.g., Oenocarpus bacaba, Maximiliana £2933 and Euterpe oleracea. According to IBGE (1977) the best known portion of Amazon tropical semi-evergreen forest is that in the southern part of PA which partially covers the Brazilian shield. The characteristic species are: Calophyllum Riééilififlfifi (Guttiferae), some species of £9222; Aspidosperma and Moutabea, Apuleia praecoxi Hymenaea stilbocarpa, Lucuna lasiocarpa, Simaruba amara, etc. At the eastern of'the Tapajos river, between Santarem and Belterra, and the northern of the Amazon river, the northern part of PA, the following species are characteristic: Qualea grandiflora and Vochysia ferruginea - (Vochysiaceae), Sclerolobium paniculatum, Dalbergia spruceana and Centrosema venosum - (Leguminosae). "Cerrado" (Savannas) The "Cerrado" trees are relatively short (around 10 meter height) and less abundant than shrubs. Basically there are two strata: the superior which is composed of trees and 36 shrubs, and the inferior which is composed of grasses. The tree stratum is characterized by individuals with crooked stem and branches, thick bark, and thick leaves with rough grained texture with surfaces of 30 by 20 cm. According to IBGE (1977), the characteristic species of "Cerrado" are: Hancornia speciosa (Apocynaceae), Curatella americana (Dilleniaceae), Qagyggar brasiliensis (Caryocaraceae), Salvertia convallariaedora, Kielmeyera coriacea, and Stryphnodendron barbatimao. 37 4 k}? Ii '5 v at a 3 4" s ‘ ’~ 1 m {_ al' wwwvra-wwdarwww D! - Federal District: Brasilia 38 EEEI - CI] ‘gy/ - ‘TIIIA 'llll” llNl-IVIRCIIIW VOIISY SAVAIIA - ‘CIIIAUO" CIASSLAIDS 'VAIIIA” OI "lfihln" COASYAL VIC'YATIBI CD'PLIIII or IBIAIIA. FACIIIIO AID lllflu 'Loollfl GIASSLAIOI COI'LII ur PAIYAWAL CHAPTER 4 DESCRIPTION OF THE STUDY AREA. The data were collected on the control plots of an experiment on natural regeneration management of an uneven- aged mixed stand of the Amazonian forest. This experiment is being carried out by DST (Department of Tropical Silviculture) of INPA.(Nationa1 Institute for Research in the Amazon). The experiment is a branch of the project "Ecological Management of the Dry-land Tropical Moist Forest". This multidisciplinary research involves all departments of INPA: Ecology, Botany, Wood Technology, Plant and Human Patology, Agriculture, Chemistry and Zoology. These departments will give scientific support to DST in its future evaluations of the environmental impact of the forest management. The study area is located within the domain of the Tropical Silviculture Experimental Station of INPA, some 90 kilometers north of Manaus, the capital of Amazonas State, Brazil. The total area of the Station is 23,000 hectares and the project area is approximately 2,000 hectares. The geographical coordinates of the project area are 2° 37' to 2° 38' of south Latitude and 50° 09' to 60° 11' of west longitude. Figure 4.1 shows the location of the study area 39 40 within the Experimental Station. According to Ranzani (1980) the climate is type Am, Koppen classification, warm and moist all year long. The annual rainfall is approximately 2,000 mm without accentuated dry period, even though the wettest period is December to May (Ribeiro 1977). The oxisol soil order "yellow latosols" is predominant in the area. This research was set up only on non-flooded ground, iue., on "terra firme". The soils are extremely poor in nutrients and very acid. The relief is smoothly undulated and it is formed by small plateaus which vary from 500 to 1,000 m in diameter. Most of the experimental treatment areas are located on those plateaus. The vegetation is typical of the Amazonian tropical moist forest on "terra firme". The superior stratum of this forest is composed of trees whose heights vary from 30 to 40 meters. Basically three botanical families dominate the floristic composition of the area, Lecythidaceae, Leguminoseae and Sapotaceae. Individually Micrandropsis scleroxylon W.Rodr. (Euphorbiaceae) and Scleronema micranthum Ducke (Bombacaceae) have an impressive presence in the study area. Several species of Eschweilera , Holopyxidium latifolium R. Knuth, Corytophora alta R. Knuth and Lecythis usitata Miers var. paraensis R. Knuth are the most frequent species of Lecythidaceae. However, Bertholletia excelsa Humb. and Bonpl. "Brazilian nut” 41 (Lecythidaceae) is absent from the area. The most frequent Leguminosae are several species of Inga, Tachigalia, Swartzia, Parkia and Pithecolobium. Within the Sapotaceae the most frequent are several species of Chrysophyllum, Micropholis, Pouteria, Labatia, Ecclinusa, and Manilkara. The floristic composition of the area is presented in the Appendix. The ecological project area is in the Tarumazinho watershed. The project was divided into three parts, referred to as bacia l, bacia 2 and bacia 3. Respectively, these are areas reserved for basic studies, buffer, and harvesting and forest management. Bacia 3 is the basis of this study. Figure 4.2 shows BaciaLB in more detail, Originally this experimental area covered 96 hectares, consisting of 4 blocks (bloco l, bloco 2, bloco 3, and bloco 4) of 24 ha each. After the commercial inventory, bloco 3 was reserved for research on artificial regeneration and, therefore, it was not included in this study. Within each block (400 by 600 m), harvesting will be carried out as the main silvicultural treatment. In designated sub-blocks (200 by 200 m each), different felling intensities will be applied to reduce basal area of some 40 listed species with dbh ; 25 cm. The treatments randomly distributed in each block were: (1) control, (2) removal of 25% of the exploitable basal area (b.a.), (3) removal of 50% of the exploitable b.a., (4) removal of 75% of the exploitable b.a., (5) removal of 100% 42 of the exploitable tha., and (6) removal of 50% of the exploitable txa. with enrichment. In each four-ha sub-block a one-ha plot (100 by 100 m) was established to evaluate the growth of the residual stand of listed species, recruitment and development of seedlings of listed species, survival and growth of listed species, growth and mortality of poles and saplings, and increment evaluation for determining the felling cycles. The listed species for this project are presented in Table 4.1. After the randomization of the blocks, the control sub- blocks were 2, 3 and 5, respectively for blocks 1, 2 and 4. Those sub-blocks, then, were used in this study. Hereafter they will be referred to as bloco l, bloco 2, and bloco 4, and collectively they will be called bacia 3. In 1980, two different inventories were carried out in bacia 3: commercial (complete enumeration of trees with dbh > 25 cm within the experimental blocks), and diagnosis of natural regeneration by sampling. From the commercial inventory (Higuchi et al. 1983a) the following data were obtained: (a) the listed species represent 1/3 of the population, (b) overall means per ha: number of trees = 155, b.a. = 19 sq.m., and volume with bark = 190 qum" (c) block 3 is statistically different from the others in terms of stand stocking and also in terms of floristic composition. From the natural regeneration inventory (Higuchi et a1. 1985) the following summaries were obtained: (a) the 43 stocking index of seedlings averaged 15.6%, (b) the stocking index of poles and saplings averaged 72.8%, and (c) the number of trees smaller than 25 cm dbh and greater than 0.30 m height averaged about 40,000 per hectare. The "milliacre” and "half chain square" methods were used for data collection of the diagnostic inventory, respectively for seedlings (tree species with dbh < 5 cm) and for poles and saplings (5 < dbh < 25 cm). In 1985, all trees tagged in 1980 from the control plots were remeasured. This was done to evaluate the growth of diameter of those trees (increment), to record new trees that moved to the first merchantable dbh class (ingrowth), and to record trees which died during the period 1980-1985 (mortality). 44 Table 4.1: Listed species for the NR management project. Spec1e Family Virola calophylla Warb. Myristicaceae Virola multinervia Ducke Myristicaceae Virola venosa (Bth. ) Warb. Myristicaceae Ocotea cymbarum H. B. K. Lauraceae Dialium guianensis (Aubl. ) Sandw. And1ra micrantha Ducke D1plotropis purpurea (Rich. ) Amsh. Manilkara huberi (Ducke) Standl. Calophyllum angulare A. C. Smith Nectandra rubra (Mez.) C.K. Allen Mezilaurus synandra (Miq.) Kostermans Licaria guianensis Aublet. Platymiscium duckei Huber Caryocar villosum (Aubl.) Pers. Goupia glabra Aubl. Aniba duckei Kostermans Naucleopsis caloneura (Hub.) Ducke Scleronema micrantha Ducke Minquartia guianensis Aubl. Copaifera multijuga Hayne Qualea paraensis Ducke Diniz1a excelsa Ducke P1thecolobium racemosum Ducke Hymenolobium excelsum Ducke Astronium lecointe1 Ducke Clarisia racemosa R. et P. Hymenaea courbaril L. Dipteryx odorata (Aubl.) Willd. Lecyth1s usitata Miers S1maruba amara Aubl. Caryocar pallidum A. C. Smith Erisma fuscum Ducke Holopyxidium latifolium R. Knuth Vouacapoua pallidior Ducke Eschweilera odora (Poepp) Miers Eschweilera longipes (Poit) Miers Anacardium spruceanum Benth. ex Engl. Aniba canellila (H. B. K. ) Mez. Park1a pendula Benth. ex Walp. Corythofora r1mosa Rodr. Cariniana micrantha Ducke Cedrelinga catenaeformis Ducke Peltogyne catingae M. F. _da Silva Bros1mum rubescens Taub. Leg. Papil. Leg. Papil. Leg. Papil. Sapotaceae Guttiferae Lauraceae Lauraceae Lauraceae Leg. Papil. Caryocaraceae Calastraceae Lauraceae Moraceae Bombacaceae Olacaceae Leg. Caesalp. Vochysiaceae Leg. Mimos. Leg. Mimos. Leg. Papil. Anacardiaceae Moraceae Leg. Caesalp. Leg. Papil. Lecythidaceae Simarubaceae Caryocaraceae Vochysiaceae Lecythidaceae Leg. Caesalp. Lecythidaceae Lecythidaceae Anacardiaceae Lauraceae Leg. Mimos. Lecythidaceae Lecythidaceae Leg. Mimos. Leg. Caesalp. Moraceae 4S @- 31w. 3 - :xmagfio E MANEJO [D — A'REA ssnrjo EXPERIMENTAL DE smvncuururu TROPIC I AL @ - AREA nrsznw Badman a: CAMPINA "M I l B. ‘ / ‘1. V l, , crpuc 3 ~\ . ' 5 , ___- , I - 7 a) I (D __,--J 27- I I w - c“; air a a ._ \ (l N UNIVE BR-l74 swim on Sat mun Y Lin»: 5v! do Dunn: we 46 ESTRAOA c i 4' 2 3-7' If; BLOGOS E SUB-BLOCOS N CURSO D‘AGJA —— LIMITES DA AREA --------- LIMITES cos sue - BLOOOS acro ACAMPAMENTO .—..-—~—- ESTRA m. 930 JETA DA CHAPTER 5 MODELLING THE DIAMETER DISTRIBUTION OF AN UNDISTURBED FOREST STAND IN THE BRAZILIAN AMAZON TROPICAL MOIST FOREST: WEIBULL VERSUS EXPONENTIAL DISTRIBUTION 5.1. INTRODUCTION Since total tree height is very difficult to measure accurately, diameter is the most powerful simple tree variable for estimating individual tree volume in the Brazilian Amazon. Therefore, the quantification of diameter distributions is fundamental to understanding the structure of the growing stock and as a baseline for forest management decisions. In‘addition, regardless of the species of tree, Amazonian timber commercialization is commonly based only upon the diameter distribution. Bailey and Dell (1973) and Clutter et al. (1983) gave a comprehensive review of diameter distribution models. According to Clutter et a1. (1983), among various statistical distributions, the Weibull distribution has been used the most to model diameter distributions. These results support Lawrence and Shier (1981), who stated that after the exponential, the Weibull distribution is possibly the most widely used distribution for population dynamics applications. 47 48 The introduction of the Weibull distribution function to problems related to forestry is attributed to Bailey and Dell in 1973 (Zarnoch et al. 1982, Little 1983, Clutter et a1. 1983, and Zarnoch and Dell 1985). Since then, this distribution function has been used extensively for diameter distribution of both even-aged and uneven-aged stands in the USA. The Weibull distribution has not yet been introduced in tropical moist forests, especially in the Brazilian Amazon. There, one of the most common models for diameter distribution is still the exponential (Barros et a1. 1979 and Hosokawa 1981). A comparison was made between the Weibull probability density function and the exponential distribution as diameter distribution models for Amazonian forests. The hypothesized distribution functions were tested to see how well they fit the observed diameters randomly taken from the study area. 5.2. PROCEDURES The Data The data for this study were collected on the research area of Forest Management Project conducted by the Department of Tropical Silviculture of the National Institute for Research in the Amazon (INPA) - described earlier in Chapter 4. 49 The basic descriptive statistics of the study area are presented in Table 5.1. The diameter distribution functions The Weibull probability density function and the exponential were chosen for testing as a diameter distribution model for the experimental area. In this study, the estimators of Weibull parameters were computed using the percentile (Zarnoch and Dell 1985) and maximum likelihood (Cohen 1965) approaches. The estimation of the exponential model parameters was done according tx: Einsensmith (1985). These approaches are described separately. (1) Weibull Maximum Likelihood (MLE) The Weibull distribution, which has the probability density function: f(x) = (c/b)x°"1 exp(-(x)°/b); x20, c>0, b>0 = 0, otherwise has the following likelihood function for a sample of n observations L(xi, ..., xn; c, b) = n(c/b)xi°"l exp(-xi°/b) (1) Taking the logarithm of (l) 50 1n L zln [(c/b)xi°"1 exp (-xi°/b)] {[ln (C/b) + 1n Xic-l ‘ (Xic/b)] l n 1n (c/b) + 2(c-1) 1n x- - (1/b)‘3xiC By differentiation with respect to g and b in turn and equating to zero, the following equations are obtained d ln L/d c n/c + Zln xi - (l/b) zxic 1n xi d In L/d b -(n/b) + (l/b2) r xic = 0 Taking b from (3) b = ( zxi°)/n (4) and substituting in (2) produces: n/c + zln xi - [l/( ixic/n)] zxic 1n xi = 0 n [(l/c) - (xxic ln xi)/zxi°] = - zln x- [inc 1n Xil/[inc] ' (l/C) = (l/n) zln Xi (2) (3) (5) The coefficient 2 can be estimated by any iterative procedures or by a simple trial-and-error approach to equalize both sides of equation (5). The coefficient _b can be estimated by (4). 51 (2) Weibull Percentile (PERC) The Weibull function using the percentile»estimators has the probability density function f(x) (c/b)[(x-a)/b]°‘1 exp {-[(x-a)/b]°}; x3a30,b>0,c>0 0, otherwise The parameters a, b, and g are estimated as follows: 3» I - [xlxn - xzzl/[xl + xn - 2x2] b = “a + X[.63n] ln {[ln (1 - pk)l/[ln (1 - pi)]} where: xi (1 = 1, 2, ... n) = the ith ascendent ordered diameter; pi = 0.16731, and pk = 0.97366. (3) Exponential The parameter estimates of the first order exponential function Y = a*eb"'x can be obtained by the linearization method (or Taylor series). This is an iterative approach using the results of linear least squares in a succession of stages. According to 52 Draper and Smith (1981), the steepest descent and Marquardt's compromise can also be used. Here the linearization method was used to estimate parameters a and E. The application pf the models £9 the data In this context, 5 is the diameter in centimeters measured at breast height (1.30 m) in 1980. The Weibull parameters are defined as: a, the location parameter, which can be the smallest dbh measured, b, the scale parameter, which shows the relative range of values the dbhfs may assume; and c, the shape parameter, which determines the general form of the distribution (Zarnoch et a1. 1982). For MLE estimators, (x - 24.9999)*was used to compute p and 9 based on Cohen's two-parameter Weibull distribution, after assuming a = 25 (the smallest dbh measured). The value 24.9999 was used instead of 25 only to avoid the logarithm of zero, since no significant differences on the general computation was detected. The parameter estimates for all three models, Weibull MLE, Weibull PERC and exponential, are presented in Table 5L2. Estimates are shown for the combined three sample plots (bacia 3) and separately for each sample plot. The Weibull cumulative distribution function was determined by integrating the probability density function for both the MLE and the PERC which provided the probability 53 for each dbh class. Then, the absolute frequency for each dbh class was obtained by the product of the total number of trees per hectare and its probability. The estimated frequency using the exponential distribution was obtained by the simple substitution of each dbh class as independent variable in the equation. To see if the three hypothesized distribution functions fit the data in the sample, the chi-square test was used for goodness of fit (Conover 1980). The null hypothesis was that the distribution function of the observed random variable is the Weibull MLE, or the Weibull PERC, or the exponential; and the alternative hypothesis, otherwise. 5.3. DISCUSSION OF RESULTS The diameter distribution for bacia 3 and for each .sample plot (bloco l, bloco 2 and bloco 4) are presented respectively in Tables 5.3, 5.4, 5.5 and 5.6. Except for the Weibull MLE in bloco l and bloco 2, the remaining computed chi-square's are not significant even for °= .25, i.e., the null hypothesis cannot be rejected. The best fit with the Weibull MLE model occurred in bloco 4 where the highest sample variance (32 = 224.86) was observed. On the other hand, the best fit for the exponential model was in bloco 2, which had with the lowest sample variance (32 = 114.76), and the worst fit was in bloco 4. In this study, the Weibull PERC model was not seriously 54 affected by variation within the sample plot. It was very consistent in fitting the observed data to all three sample plots across a range of diameter Classes. The Weibull MLE model, in contrast, consistently overestimated the frequency of the first dbh class and underestimated the next four classes.‘The Weibull MLE was consistent, after the first dbh class, only in bloco 4. The exponential model, on the other hand, was very consistent where the sample variance was low, and inconsistent with higher variance mainly in estimating the frequency of higher dbh classes. Bailey and Dell (1973) pointed out that c < 1 should occur in all-aged stands of tolerant species. Here the estimate of shape parameter was greater than one, c = 1.02, only for bloco 2. For the other sample plots and for the combined plots the 3's are smaller than one. When the three sample plots are analyzed together on a per hectare basis, representing the entire experimental area (bacia 3), all three hypothesized distribution functions fit the observed data. This is demonstrated by the non- significant chi-square values even for a==.25, although the chi-square of the Weibull MLE model is about ten times greater than the others. Graphically the results of each hypothesized model for the experimental area are presented in Figures 5.1, 5.2 and 5.3, respectively for the Weibull MLE, the Weibull PERC, and the exponential. Figures 5.4, 5.5 and 5.6 represent the relationships between the frequencies of observed dbh 55 classes and the frequencies estimated.by the hypothesized models for each sample plots, respectively for bloco 1, bloco 2 and bloco 4. As expected, except for the Weibull PERC in bloco 2, the curves produced by the Weibull distributions are reversed J because of the c parameter (c<1). These graphics demonstrated that the Weibull PERC produced the lowest dispersion of the observed data around the hypothesized curves for the combined plots and also for individual plots. 5.4. CONCLUSION The results indicate that the Weibull PERC model is the best model of the models tested to quantify the diameter distribution of natural stands in the Brazilian Amazon. The simplicity in estimating the Weibull PERC parameters and its insensibility to the sample variation demonstrated in this work are valuable attributes to be considered. This distribution function can be used with a reliable individual tree volume equation for modelling forest growth and yield for the study area, or for similar areas in other portions of the Amazon. The exponential model also performed adequately in fitting the observed data. However, it may be inconvenient to estimate its parameters by the iterative approach. The ease of this method will depend upon the available computing capabilities. 56 mm.mm m~.m~ me am cu.mm toss-at~> to .eoou mm.v- mn.vas m_.oa_ mm.mk. cutest»: caeuom om.om mm.mm k~.om ea.km eno- ass-nesatc ae.m~— aa.~m cu.ms. 66.6.2 .acaxn: cu.mw cu.mw ee.m~ cu.mw .au.c.: new use To 82 «88 .8 o 4 ocean m guess a noose m sauce muaace .aota mean. 6:3 rue noasuaanuu o>aaemtouou Acootosocaaa "_.m c.6u» 57 mm. mm. 8. mm. Nm 58 8(8H388888888888828881_ 5! '" :ssxsssasstasnese 283;: idgmdéfiddNJJ """" i3 3'336848683838383433818 a! '" 3%35.48338.3§i$...3i3 (QQNBSNWNN"" 'ia 3133373338534333.3.98i3 HlmN—o—o—a : Bl 3'8333388338385838388 8 ElgatdmmVMN—dm a '889888888 888$ $0888 8 I'fidfié‘fiddj ..... ... é ‘esseeeseseweeeeeeee:e Nasetssmsetsss.§§ggg§ mdmfl¥= 5d” man IOBH(X) = center of diameter class in centime up» S9 a 33333339§3858§33..333!8 E :86N88R88“9 SEES .2ggpsesdasg -------- :$ 3 assasasaaaaaassasasaasfi 33 " '“ g (US$288 88”$“8$“8£8“18 H cia'éiéu'uiaioiniéé """" :fi PEIELL ".5 ';63:k~ ”as 8389 H: as: b I SfimOR-IDVNNNfifi—fi [.me mamm—onhmm—o—au... V 154.75 152.3? methods. iRRBRSRSSRSSSSBD D808 immmmmmmmmmmmmmmmmmml In sszsmssstssssggg TDTFI. Tfile 5.4: Diameter distribution for Bloco 1 derived from three different (8400:: Table 5.5: Diameter distribution for Bloco 2 derived from three different ”flak. EHMIFHE IEWELH£ EWUGWML 3' “It!” UEH(X)fl 60 8:88:8568266866332 a: — Iv 133.333.3359832 :s IagwtwdmdeJ"'ig 3i8$33886.§43938't 3 i " m :aaaiaaaaaaaaws=6 3mgmonem—n ' i3 mummvmm :' ‘5 ml 25 (D '5 00 a) a! 25 4 25 O 25 25 m6n51arm mfll ‘RfikfififikSSRRBBRI IDBH(X) = center of diameter class in centimeter. EWIR. film llE metfods. u-‘mm 08HI Table 5.6: Diameter distribution for Bloco 4 derived from three different 61 Elm PERC EF agsssssesesasass 23::1 ~| aaaaaa —: aaaaa ‘aagaig 5! ISBHBISEREQ $33N993§9 !$Qéiéd¥d&éé """" i5 8i628686886682683663:4 6! " '* assassusesessssrmsgs .SaazdfidddeJ """ :8 azsaaaazaataaa.aaaaszw 3!" d 'm ii83$§88414384333833 2 b'sazzmmvmmmé~é """ é 'SKSKSKEBRNBESRROBR R Igggvessssss-g-s -- é 9999999999999999993d Iassvtssssstssssggg g 62 70- {tree/ha 3 11 1 L1 L1 3 N O L L A l A L L ..a 0 L4 . A I -..--___._.,, 10 20 30 40 50 50 70 80 90 100 110 120 DBH Gun) PREDICTION BY WEIBULL MAX. UKEUHOOD Ce-.. Figure 3.1: Bacia 3 - The relationship between the observed and estimated dbh frequencies, using the Weibull MLE function. I tree/Ho 63 "5b"'eb"'7b” so so DBH Own) PREDICTION B'Y WEIBULL PERCENTILE 1% 110 120 Figure 5.2: Bacia 3 - The relationship between the observed and estiaated dbh frequencies, using the Heibull PERC function. f tree/ha 70-1 64 ‘1'er VVY'VVV‘f' r'7111lvv'vrf 102030405000703'0901001101 DBH(cm) PREDICTION BY EXPONENTIAL Figure 5.3: Bacia 3 - The relationship between the observed and estieated dbh frequencies, using the Exponential function. 65 70 s m a. 0-0 M 2 50 \ g .. . "" so 3 ID 0 5 ...... 2,0 ..... 4'0 ..... i ..... *Wghnvaao 08H (cm) (’0 § ...} 3 a can-cu a can-us fl-E H m Ii!- “q H m I 2 “9 ad \ 3 g 3 . g “5 2 “H L ‘Vwé ‘Kué afi afi 0-3 01 40 I 1 20 40 u l 1 DB" (cm) . OBI-I (cm) (a) (C) Figure 5.4: Bloco l - The relationship between the observed and the estieated dbh frequency distributions by (A) Exponential, (B) Heibull PERC, and (C) Heibull MLE functions. 66 : - m 79.5 O-e see-fl ”E 2 E \ a; 8 : b{wj - s ...; aé mi o3 - 6 vvvvvv g IIIII 1% IIIII 3 iiiii “III-I‘vfiusvvv‘k 08H (cm) (A) ‘H : I “In“ ad 2 i 3 ”‘1 of“? 3% zo-f ...: .3 ' 40 I I DBH(un) Figure 5.5: Bloco 2 - The relationship between the observed and the estieated dbh frequency distributions by (A) Exponential, (B) Heibull PERC, and (C) Ueibull MLE functions. 67 70 s m .0 v-O ”is 2 50 \ g 0 "n 29 m , o . N I I OBI-I (cm) (A) 70 a dam! nann— fl' wwlfllun rennin 3‘” 3:» \ \ 9 ‘° 2 in” In w . 1 . 40 I I 40 u I I mfiIknd CBH(ufi) (B) (C) Figure 5.6: Bloco 4 - The relationship between the observed and the estieated dbh frequency distributions by (A) Exponential, (B) Heibull PERC, and (C) Heibull "LE functions. CHAPTER 6 A MARKOV CHAIN APPROACH TO PREDICT MORTALITY AND DIAMETER DISTRIBUTION IN THE BRAZILIAN AMAZON. 6.1. INTRODUCTION Little is known about forest structure and stand dynamics of Amazonian tropical moist forests. Successive records from representative long-term permanent plots practically do not exist. The problem of reconstructing forest history is greatly compounded by the fact that trees can not be reliably aged, species diversity and spatial heterogeneity are high, and fallen logs decay rapidly. It is important to understand and report the natural changes that occur in representative examples of pristine Amazonian forests, because their composition and structure can be altered by man as the demand for tropical timber species increases. The main objective of this chapter is to report S-year changes in the overstory structure of an undisturbed tropical moist forest. This will be done by the transition probabilities of the overstory diameter distribution and mortality of this forest, using a first-order Markov chain. Diameter distribution and tree mortality will be projected 68 69 ahead to 1990 (t+2), based upon a 5-year period of observations completed in 1985 (t+l) and its immediate past in 1980 (t). A first-order Markov chain is a stochastic process in which the transition probabilities during the time interval (t and t+l) depend only upon the state an individual is in at time tior upon the knowledge of the immediate past at til, not upon any previous state (Horn 1975, Chiang 1980, and Bruner and Moser 1973). Shugart (1984) pointed out that the time-invariant nature of each of the transition probabilities is an important characteristic of the Markov approach. Shugart and West (1981) stressed that the importance of understanding forest ecosystems is based not on their age, but on known changes at present. Deterministic models consisting of a single mathematical function (linear trend, polynomial, sinusoids, or exponential growth or decay) have not proven adequate when time series are involved (Morrison 1976). In tropical moist forests, size may be more important than age. One reason for this is that size may be more ecologically informative than age when.it is difficult to make accurate estimates of age (Enright and Ogden 1979). Division of life-cycles into developmental stages may allow prediction of future behavior more accurately than division into true age-classes. Usher (1966) used size attributes instead of age to develop a model for the management of 70 renewable resources. He stressed that an organism which is in i-th class at time t can be in the same class at time £11, or it can be in a next class of that attribute, or it can have died. According to Enright and.Ogden (1979), the transition matrix models in general are suitable for the analysis of many biological problems, mainly in studies related to the forest dynamics. These models have been used intensively in studies of dynamics of populations of plants or animals in many parts of the world. Some examples are: the demography of jack-in- the-pulpit in New York (Bierzychudek 1982); forest dynamics of a population of Araucaria in a tropical rain forest in Papua New Guinea, and Nothofagus in temperate montane forest in New Zealand (Enright and Ogden 1979); termite succession in Ghana (Usher 1979); forest succession in New Jersey (Horn 1975); the application, although without success, of this model in secondary succession in coastal British Columbia (Bellefleur 1981); the discussion of some extensions and application of Hornfls Markov approach for forest dynamics in tropical forests (Acevedo 1981); and the application of Markov model to predict forest stand development (Usher 1966, Usher 1969, Bruner and Moser 1973, Peden et a1. 1973, and Buogiorno and Michie 1980). Alder (1980) also described the transition matrix as a possible tool for analysis of growth and yield data for uneven aged mixed tropical forests. Most of these works include a reasonable review 71 about the theory behind the Markov approach.]} is said to be homogenous with respect to time, or time homogenous, if the transition probability Pij(tvt+1) = Pr lX(t+1)=j|X(t)=il. i.j = 1, 2, ...., m. 72 depends only on the difference between t and t+1, but not on t or t+l separately (Chiang 1980). The computation of this probability can be done as follows. where: nij First, calculate Pij given class i at time 3, individuals in class i at time t. = "ii/"j and r1- = number of individuals in class j.at time t+1, = total number of The transition probability matrix of a Markov chain for a n-state process can be set up as: j=l j=2 j=3 ..... J=m r- “_I 1‘1 P11 P12 P13 -°-°- le i=2 P21 922 923 -°°-- sz P = (pij) i=3 P31 P32 P33 °°°°° 93m 1:” Lfml sz pm3 "°°° pmm [ The probabilities Pij are nonnegative and the sum p11 + p12 + pi3 + see Pim = 1a The transition probability Pij can be of n-step transition probability, pij(n), as the probability that the population goes from state i on one trial to state i 3 trials later.4According to Bruner and Moser (1973), the n- step transition probabilities matrix may be obtained by the 73 equation P‘n) = P“ where P‘“) is the matrix of n-step transition probabilities and Pn is the initial transition matrix raised to the n-th power. In this work, 15 states (i, j = l, 2, 3, u. 15) were established as follow: state 1 = ingrowth (1), states 2 to 14 were defined as dbh classes, from 25 to the generalized class next > 80 cm, in S-cm interval, and state 15 = mortality (M). Ingrowth is defined as those trees not tagged in 1980 which in 1985 reached dbh >, 25 cm. The time interval g and 511 are respectively, 1980 and 1985. Tables 6.1, 6.2 and 6.3 present the transition of the absolute frequency of individuals from the state i to state 1 during a S-year period, respectively for bloco 1, bloco 2, and bloco 4. The state ingrowth does not appear at time 1980 because it means only the movement to the higher dbh class from the generalized dbh < 25 cm class. The probability for transition among states was based on the frequency of trees which either remained in the same class, moved to a higher class, or died during a 5-year period. Tables 6.4, 6.5 and 6.6 present the transitional matrices for blocos 1, 2 and 4, respectively. These tables were set up using their counterparts, Tables 6.1, 6.2 and 74 6.3, as bases for the computation of probabilities. For example, the probabilities for the state 25 cm dbh class for bloco 1 (Table 6.4) were calculated as follows: pq'z = 155/183 = 0.8470, 122,3 = 16/183 = 0.0874, and p2'15 = 12/183 = 0.0656. From all trees in 25 cm dbh class measured in 1980, 84.7% remained in the same class, 8.74% moved to the 30 cm dbh class, and 6.56% died during the period 1980- 1985. The probabilities for other dbh classes and blocks were similarly determined with the respective counterpart tables with absolute frequency distribution. The tw0wstep transition matrix for each block (Tables 6.7, 6.8 and 6.9) were obtained by squaring their counterparts (Tables 6.4, 6.5 and 6.6), respectively for blocos 1, 2 and 4. These tables represent the probability for dbh and mortality distribution after two 5-year periods, i.e., for t_+3, year 1990. The two-step transition matrix is the basis for predicting the distribution of diameter and mortality for the study area in 1990. The eigenvalues (Ai) of the transition matrix of each sample plot were determined according to Anton (1973). The dominant eigenvalue (A1 = 1 since each matrix is non- negative and row sums are 1) and the next largest modulus ( 12) were determined. These values provide the ratio (Al/A2) which, according to Usher (1979), indicates the speed with which the system will approach the ”climax" state. 75 6.3. DISCUSSION OF RESULTS The projections for 1990 of number of survivors from 1980, the frequency distribution of dbh classes, and the mortality by dbh classes, respectively for blocos 1, 2 and 4 are presented in Tables 6.10, 6.11 and 6.12. These projections were determined based on the product of the two- step transition matrix and the initial values of each state. Volume stocking in 1990 can be estimated by applying a reliable individual tree volume equation to the projected diameter distributions. The frequency of individuals per dbh classes is available for each sample plot. The plot ratios (Al/A2) were 1.15, 1.00 and 1.00, respectively for blocos l, 2 and 4. The mean ratio, 1.05, suggests that the studied area will approach the "climax" state slower than the two systems discussed by Usher (1979), mixed hardwoods in Connecticut (ratio 1.34) and in New Jersey (ratio 1.57). This result makes sense if compared with the distribution of changes in dbh classes and mortality which occurred over a 5-year period. Using Table 6.13, which is the summary'of one-step transition matrix, the mean estimates of the probability of changes and mortality per plot are respectively 0.1205 and 0.0918. This means that 12.05% of the total number of trees in a plot changed dbh classes, and that 9.18% died during a 5-year period. In an absolute basis, using the mean number of trees/plot = 631, 76 trees changed classes, 58 died, and 23 (the mean ingrowth/plot) grew into the measurable dbh 76 classes. Thus, these results suggest that the studied area is not a static population. The projection for 1990, based on the summary presented in Table 6.14, also does not show any trends that this population is not changing. The average of the probability of changes of dbh classes increased to 0.1895 and mortality to 0.1717. 6.4. CONCLUSION In the study area, the average rates of mortality and ingrowth, during a 5-year period of observation, are respectively 9.18% and 3.72% in relation to the total of initial number of trees recorded in 1980. There is no evidence that the probability of mortality increases as dbh increase. The same trend is observed for changes in dbh classes (the movement from one class to another), i.e, the changes are occurring independently of the diameter size.4As this study dealt with only the control plots, it will be very interesting to compare these plots with other experimental plots to see how effective were the silvicultural treatments to change the rates of mortality and ingrowth. The ratio (Al/A2) leads to the conclusion that the population under investigation is not static, that changes are still taking place, and that the rates of ingrowth and death are not perfectly balanced. However, it is also 77 necessary to keep in mind that this population is truncated by size, i.e., only trees with dbh ; 25 cm were involved. The Markov approach has a lot of potential. It can be used as a baseline to project the mortality and diameter distribution, or at least to predict the direction of future trends, for forest management purposes in natural stands of the Brazilian Amazon. It provides a general insight into the nature of the dyamics of a sample of pristine Amazonian forest which, consequently, will be very helpful to assist decision makers in exploring and understanding the Amazonian forest issues. In 1990 this procedure will be repeated. Then, the Markov chain approach will be evaluated and, if necessary, refined based upon a 10-year of observatirun If valid, the projection ahead to year 2000 will be possible. 78 .QOLIN coal saunas Jul—n any .vmm u Amu__o»ta¢o : u mew u mam—\Dnaou new saw u Acuaotmcav a u mew u cam—\Dosaa mew Nm N N— m m dd mu mN mm mm kw MN mN— um— vn u N «a Q n m v N N Mn N u o— N— N a m cm v wN —m m m mN Vt w m mm em 1 a v M? as m g NN Nm mm— Nu NN ma— mm— N— mm and mN mN algae t axsc DGA ms ON mw Om mm Om mv 0* mm am mN .uoatoa iguana a mcmtau “no reasons a» any canan can note coauancnt» a _ ocean nsdow ~i~naseesmssemsggi a n u—.m sunlh 79 .aOLaN cues secede Julun Axv .9... n 35:38: : .. m8 0 82:33 vi 36 u “5.8.6:: o .. So n 82:33 nu ma mN Nm am can ‘§ I ID V 1‘ N In \D N‘DO'INY v-0 V‘ $ Iii—0 N— u MN O— agasssaaa**~* afiflzaxtocxmkfimmowmmommvavmmommwuq Nvu NN— ON mm— Nb— ON ~08029803020§§.l§ 'd 62.3.. bamum a mafia .3 into...- 3 3. 33.. .8 8.... 53.20.: .. ~ 808 N6 .33 80 .0968 coal saunas {'3 2C .58 u .323th z .. 93 n 82:33 as Bm u xfiaobcc _ .. owe n 82:33 mNm NW _ mu m a mu m «N um m me as ON— m— mg ... wxac m— N a mu 8A w w ms m m ON m— u a mu mm m a h 8 N N N 9 mm mm m m NN am 8 N N NN me mm Q a me 9 ea 5 w «h mm AN" N— h 8— a a" N a N— B— N VN ..N u an .3fizax2aakmkokmmoommommvovmmomfian gaggfifig 63%.. Latin a 9.7.3 AC .652- 3 so 33.. .8 of... 83.23., .. 4 803 6.6 .33. 81 .0958 else seals. 1...; any 34 t fix...— #10: Inc. «Neg. Nah. 8A SMN. 8w. m5 gm. gm. Oh and. mwho. Nmmh. mm 53—. mag. 8mm. 8 Mann. NW8. 3 an. mg. 3. am '8‘. 8:. gm. m1 menu. 82... «who. mmmm. De. 3. mNuo. "NQN. mg. a 3:. mwhu. VNnN. 8 $8. v5. 23. mN 34 u 0‘ :axtamkmkonmmommmommvovmmflmN—n {IIIIII’IO iiiIIIi .37.... ...-mum - mar-d A? ...:qu 3 2o 33? .....o 8.... $.31. 3.3.1.3.... 8.305.... u a 808 3.6 .33. 82 .aOLaM case season Jenna nay aaao._ : auuo.a 5x0: anae._ aux aaao._ mu comm. comm. an aamw. anon. aamw. mm aamw. cams om meaw. . mmc_. wen» mm 4_~o mak_ cams. _ am mama. mmko. mean. me name. mmmw. none. 84 «mos. mawa. _mam. mm mane. mane. wmv_. mask. on. have ammo. mama. mu aaau.. a a z axis new ms as no am am an ac av an an m~ _ n .uoat-u t-cmum a meatnu Aw. turban» a» has can». .co cote xatanc mum—an-notu sausage-t» u u coo—a "m.u can-p 83 60.60 See access v1.3 25 oaaa._ : aaca.a 5x1: __~_. ammo. ammo ask auao.~ mu aaao.a on «was. some. «can. mu unwa. owns. no mean. mama. ~u_a. mm mwea. Ammo. vans. om ammo. vcma. sums mv mama. mmm~ cums. av 88. 3.3. ~96. mm memo. _mma. cams. on mmaa. noes. «mun. amaa. nu ouao._ a a : ax.c oak mm as no um mm on mv so an an mw a a .uoaton tucmrm a acorns any reason» on “as canon can «one xatanc mum—«nanotn someone-t» u v cacao no.6 can-p 84 .8...“ a: 800% «1.3 CO 3.— 3.— XKN. MNZ. Rum. Nmma. mg. mem. gN. Sun. E. has. mes 3. MN". mem. g. ~20. mqu. N—mh. mend. wNno. an. a NmQN. Dunc. and Red. 0‘8. 30. 9:. gm. mg. kg. 88. NNNf. 3. Chan g. 83. EN. £8. mau. vn—D. m8“. shun. $8. ‘58. Eva. ~0809930302k§§= 0!. : axon new as on no no mm on me so an an m~ _ n III‘IIII IIII‘O ItlIIIIIII’lIIII!III II .xatanc mAMLMAuaoLa consonants acuuuoaa oz» u _ cacao "5.6 obey» 6.8: Bloco 2 - Ihe hub-step transition probdailitg satrix. Tble so 5 40 45 50 55 60 j l a on deli—Odd 0000000000 .238 .0268 .0015 .6611 .1810 .02“ .4577 .334 . .7533 .7251 .1607 .0144 .mas .5953 .515 .09!) ”8889988882883: 00 blank spaces eean zeros. 86 .aOLafl csaa assess Jesus any s: *3 mvNu. Numb. mmmu. mmmw. Oman. Nwmu. ummm. hw—D. mmmN. mNum. vao. mDDN. unmm. hmoa. VN—u. vv—h. mmuo. vmmn. NMNN. : wxac DNA ms nu mm wmmn. om mm am we as mm on scan. MNua. gnaw. mmum. mN . n 4x0: aux ms as no no mm on m4 :4 mm on m~ a n .xatanc muzaanqnota caauancatu acuunozu us» I 4 ocean "m.w cannh Table 6.10: Bloco l - Projection For 1990. aortalitg dies. distr. survivors 0 trees 5 in 1980 g 87 um what!) mom «I 0451 .- (:83k889883882 I MVWMONNNO—O 12 17 ! 8889981888828 3 ‘~08099808020§§. 102.19 543.62 l1 I8 ,_. 1:3 Tdale 6.11: Bloco 2 - Projection for 1990. Otrees din. distr. nrflwn: nddfig 1 1g) 11'! 88 REBERSBKRRBBB afiggmvw—o— 3i ownmmuvo—ooo d" d I awasassaeeas g o h wvmwmé d' §§:;“9"" a No onumwv ..| Exam" 5333§38888333 8 sassszsss~444 — fl" Qisgggaansas'~* g§§00023**~* g I~na099003022§§.!§ Tble 6.12: Bloco 4 - Wojection for 1990. eortal itg din. distr. survivors Stress in 1% 89 I ! NghflNlflN-OdOON '2 88118893288888 g ‘~0809930302R§§.!§ 90 in 1905. bloco 4 bloco 2 bloco l 1 Table 6.13: Sueearg of one-step transition probability eatrix - H 0656 1111 0365 0741 1364 0960 0 1667 1530 5000 0 0714 3 classes '888998888288 l '3“: 91 Nmmu . ”N: . 3N— . memN. muNN. _NDN. seas eNRJ mmemé snmmJ o—mod wwaN thsN ~43 mmmN 83 o a echN MN: 8A a a o 0 who. 33.. mn o c a mum... g. o on meNu . meg . mN—m. ame. 3mm. 28. mm vva. a mN3. Bum. meg. mNmn . 8 menu. mmma. mew-m. 2.2. g. uwNN. mm QDWN. as: . mNm— . mesh. we: . menu . om 3N". mNum. :3. .62. NmVN. 2.0—. we ONsN. mwNN. Nuw— . Nag. k: . 2b— . 9. 1mm". ~NN~. mug. RON. a. mum... mm mms: . when. we: . ummN. Emu . mmMN. 8 nmnN. 8: . Name. «ms— . m8— . humu . N t amt-£0 : emcee-o .- smelt". lease—o e 0003 N 0003 u 0003 cg 830-8.... .. 3.33 33.888 833.93 aggro...» to Fla-am "1.6 .33 CHAPTER 7 SHORT-TERM GROWTH OF UNDISTURBED BRAZILIAN AMAZON TROPICAL MOIST FOREST OF "TERRA FIRME" 7 . 1 . INTRODUCTION Based upon the available literature about growth and yield studies, the mixed uneven-aged stands of the Amazonian forest are condemned to stay where they have always been. These forests are not an attractive forest investment because little is known of the past growth and future growth potential. Age and site index, two fundamental variables used in developing even the simplest growth and yield models (Sullivan and Clutter 1972, Ferguson and Leech 1978, Alder 1980, Smith 1983, and Clutter et al. 1983) are not available. In addition, long-term successive measurements on permanent sample plots are non-existent. The main objective of this work is to give'a starting point for growth and yield studies in the Brazilian Amazon based on a 5-year period of observation. A major constraint is that age, site index and total height of trees are not available or even practical to obtain. Therefore, only diameters measured in 1980 and their 5-year increments will be used in an attempt to project growth and yield. Another objective is to avoid the problems that the tropical 92 93 countries of southeast Asia have faced in terms of divulgation of their experiences with growth and yield studies. In Peninsular Malaysia, for example, although growth and yield studies were established back in the 1950's, almost thirty years later few analyses have been reported (Tang and Mohd 1981). Revilla (1981) also pointed out that the growth and yield studies reported in Malaysia and Philippines do not reflect the abundance of growth data available. 7.2. PROCEDURES This work focused on two separate aspects of growth and yield, the current volume estimation for individual trees and prediction of future volume. In individual current volume estimation, the emphasis was on the selection of the best model to estimate the 1980 and 1985 merchantable volumes based on either single-entry (dbh as independent variable) or double-entry (dbh and height as independent variables) regression equations. The variable height (H) from double entry models was estimated from diameter-height equations. Several regression equation models were tested. The selection of the best model for volume estimation was based on the Furnival index (Furnival 1961) - adjusted standard error of estimate (SEE) used to compare logarithm equations with non-logarithm equations - residual analysis and the coefficient of determination (R2). In yield information and prediction, the emphasis was 94 on the development of individual volume growth models based on diameter or basal area increments, the development of a model for volume of 1985 as a function of the diameter or volume measured in 1980, and the use of the exponential Lotka's growth model for volume prediction on a hectare basis. The data The data used to develop models for current individual volume estimations came from Higuchi and Ramm (1985). For this study, trees with dbh < 20 cm were excluded leaving a total of 654 cases. Table 7.1 presents the basic distributional characteristics of the data. For yield information and prediction, the data came from the three four-hectare permanent plots described in detail in Chapter 4. I The quantitative information for all three sample plots are summarized in Table 7.3. These data refer to 52 botanical families found in the study area, including about 350 different tree species with dbh ; 25 cm. Three families (Lecythidaceae, Leguminosae and Sapotaceae) contributed 50% of the total number of trees. Table 7.4 presents the distribution of frequencies of the three dominant families in terms of status in 1980, mortality (M), ingrowth (I), and various classes of periodic increment (PI). 95 Model development From the available literature, models were selected which matched the variables available for the study area. For individual tree volume estimation, the following models were tested: (a) Single-entry models (Loetsch et a1. 1973) v = a + b*02 (1) v = a + b*D + c*02 (2) log V = a + b*log D (3) log V = a + b*log D + c*(l/D) (4) (b) Diameter/Height models b*D + cm2 + d*D3 (Clutter 1963) + H=a log H a + b*(l/D) (Loetsch et al. 1973) l/H II A! + b*(l/D) (Rai 1979) (7) log H a + b*log D (Schreuder et al. 1979) (c) Double-entry models (Loetsch et al. 1973) log V = a + b*log D + c*log H or V = a*Db*H° v = a + b*DZ*H (10) For all models: log denotes logarithm to base 10. (5) (5) (8) (9) 96 D denotes diameter at breast height (dbh) outside bark in centimeters (cm). It is measured at 1.3 m above ground level. H denotes merchantable height in meters (m), i.e., the length of stem from the ground to the crown. V denotes merchantable volume in cubic meters (cu.m.). For individual yield prediction, the following models were tested: (a) Increment d0 = a + b*D + c*02 (West 1980) (11) dBA = a + b*D + cm2 (West 1980) (12) db = a + b*(D - 25)2 (West 1981) (13) Where: do = periodic diameter increment in cm. dBA = periodic basal area increment in squared meters (sq.m.). These three models were weighted using the inverse of the estimated sample variance as weight for each diameter class. The weighted models will be equations (14), (15) and (16). (b) Volume in 1985 = f (1980 vol. or 1980 dbh) V(85) a + b*D(80) + c*(D(80))2 (17) 0(85) a + b*D(80) (Soekotjo 1981) (18) 97 Where: V(85) = volume estimated in 1985 in cu.m. D(85) dbh measured in 1985 in cm. dbh measured in 1980 in cm. 0(80) (c) Lotka equation (Pielou 1977) Adapting this model to predict volume growth produced V(t) = V(0)*ert (19) Where: V(t) = volume at time t (for t = l, 2, n 5-year periods) in cu.m./ha. V(0) = volume at time 0 (1980) in cu.m./ha. r = b - d = the intrinsic rate of natural increase. b= [I + Increment]/V(t) = ingrowth (I) and increment rate (flow-in quantity). d=M/V(t)= mortality (M) rate(flow-out quantity). 7.3. DISCUSSION OF RESULTS Only 1/3 of the total of species belonging to the three dominant families are considered commercial species by local markets in Manaus. There exists no occurrence of the two most valuable species for exportation of the Amazonian forests, Swietenia macrophylla King (Mahogany) and Cedrela odorata L" The individual dbh periodic increment (PI) of the study area averaged 1.06 cm, equivalent to 0.21 cm/year. This mean PI was estimated based on a population from which 31.3% did 98 not have any increment at all (Table 7.4). The mean dbh for each increment class is also presented in Table 7.4. Note that the maximum mean increment occurred in trees with 40.9 cm dbh, and that zero-increment occurred in trees with 42.5 cm dbh. More than 80% of trees had a PI less than 2 cm. The average periodic annual increment (PAI), 0.21 cm/year, can be compared with the long-term PAI of 0.10 to 0.12 cm/yr obtained in Puerto Rico, Maricao and Luquillo forests (Weaver 1982), and with the PAI of 0.22 to 0.48 cm/yr from the southern Ontario hardwood forest (West 1979). The PAI for ”pau-rosa” (Aniba duckei Kostermans) at Ducke Reserve, about 20 km north of Manaus, was 0.38 cm/year (Alencar and Araujo 1981). In southeast Asia, however, the dbh PAI's for virgin or managed forests are at least twice as large as the PAI of the study area (Miller 1981 and Tang and Mohd 1981). Although the PAI is positive, the stand stocking decreased during the period 1980-1985 in terms of number of trees, basal area and volume (Table 7.3). This is explained by a mortality rate which was twice the ingrowth rate. Another explanation for the decreases is the mean dbh's for mortality and ingrowth, which were 39 cm and 26.3 cm, respectively (Table 7.3). The regression models for individual trees volume estimation were developed using the ordinary least squares method. The regression summary for these models is presented in Table 7.2. 99 The diameter/height models did not perform adequately and, therefore, they were not used. Probably the reason for failure in fitting the diameter/height curve is because merchantable height was used instead of total height. All proposed models used total height. In Rai's (1979) work, for example, a R2 = 0.956 was obtained, while the highest R2 of this work was 0.154. To estimate the current volume of 1980 and 1985, the following equation was used log V = -3.4033 + 2.2673*log D (3) This equation was chosen because it presented an appropriate residual distribution, had an acceptable R2 and SEE, and because it was as precise as the equation (4) with three coefficients. Before developing the proposed regression equations for increment and growth' studies, a contingency table was developed to test the differences in probabilities among sample plots (Conover 1980).‘This test was carried out to test the possibility of pooling the sample plots. In this case let the probability of a randomly selected value from the i-th bloco being classified in the j-th category be denoted by Pij' for i = l, 2, 3, and j = l to 5. The hypotheses were: HO: All of the probabilities in the same sample plot are equal to each other (i.e., plj = p2j = p3j for all j). 100 Bl: At least two of the probabilities in the same sample plot are not equal to each other. The chi-square test for differences in probabilities was carried.out based on the tabulated data from Table 7.3 for number of trees and the mean diameter for each sample plot. For contingency tables, the rows (j) were constituted of different categories (status in 1980, status in 1985, ingrowth, mortality, and periodic increment), and the columns (i) by the sample plots (bloco l, bloco 2 and bloco 4). For both number of trees and mean diameter, the null hypotheses could not be rejected. All sample plots were combined therefore for further development of the growth and yield models. . All increment models developed had very poor fits, as demonstrated by the low R2 values and high 8888 in Table 7.5. Model fit was not improved through the use of weighted least squares. Based on these results S-year periodic increments should not be used as a baseline for projection of growth and yield of the Amazonian forests. A possible explanation of this result is shown in Table 7.6a, which contains the mean, standard deviation, minimum and maximum increment.by dbh c1asses.‘The same procedure was used for dbh when increment classes were considered (Table 7.6b). The contingency table was used to test the differences in probabilities for mean, standard deviation and maximum increment and dbh. There were no differences among these values for both diameter classes and increment classes. 101 Statistically, this means that the mean increment is not significantly different between dbh classes, and that the mean dbh is equal for all increment classes. In contrast, the two models, equations (17) and (18), developed for individual volume and diameter growth performed very well. The explanation for this successful fit can be found in the explanation for the failure of increment models. In model (16), for example, the objective was to study the relationship between the dbh measured in 1985 and the same dbh measured in 1980. As the increments were very small, non-negative and non-significant, the dependent and independent variables were approximately equal. In both equations the regression coefficients were highly significant. With these models one may now predict the individual volume or dbh growth for another period of time, for 1990, based on the dbh measured in 1985. In 1990, these models can be validated and refined using, then, a lO-year period. Finally, fitting Lotka's model (19) to the data produced V(t) = v(o)*e(-0.0347*t) The intrinsic rate of increase, r, was obtained on a hectare basis based on the data from Table 7.3 as follows: 0' ll (3.8197 + 12.8010)/ 273.6415 = 0.0617 D: II 26.1019/273.6415 = 0.0954 102 r = 0.0617 - 0.0954 = - 0.0347 Using this equation, the estimated volume in 1985 is V(1) = v(o)*e('0.0347*1) V(1) = 274.7256 cu.m./ha and for 1990 V(2) = V(0)*e('0.0347*2) V(2) = 265.3458 cu.m./ha Based on the intrinsic rate of natural changes in a 5- year period, the yield estimation for 1985 is very close to the observed yield (V(85) = 273.642 cu.m). The projection for 1990 also looks acceptable. This means that Lotkafis model seems very promising to predict the future volume yield. However, a longer period of observation is necessary to fully validate this model. 7.4. CONCLUSION In the study area the dbh increment for individual trees during the period 1980-1985 averaged 1.06 cm. The mortality rate was twice the ingrowth rate, and the stand stocking decreased around 4% during this period. Two models, V(85) = a + b*D(80) + c*(o(80))2 and Lotka's model, could be used to estimate individual tree volume for the next period with an acceptable reliability. 103 In 1990, then, these models can be retested, validated, and refined, if necessary, using a lO-year period of observation. There was no indication that dbh could be used to predict either the merchantable height or short-term diameter/basal area increment. Traditional growth and yield models can not be applied to the Brazilian Amazon forests since age and site index are not available» The only alternative remaining seems to be the use of simpler models based on stand structure monitoring at successive occasions. The findings of this study suggest that the growth and yield studies are not possible only on temperate forests, but they are also feasible on an undisturbed tropical moist forest in the Amazon. To achieve this goal, however, the best estimation of the current volume and the dependability of the permanent sample plots must be pursued. 104 STAN , ~26 mead mum...” ..m... 1.58 239 98.5..." 8~.m .8...” «3.9 «no .3 «:92 8:68 898 $12 «3.: 4mm 08 is .9... ......m... ”mam ,6. ...... mg... (11.3%; And-u 0:83 ”53!.qu 8mg... 0.50) 13331.; .5... no! 33 .5 co 833133.20 _6fifiahuau 3.8 :d .3: 105 .3630me 8:65.... 0 n. pl .0 .n .0 u...- .xoucm 33:15“. u E .3030. .....o 6...... PIE! .1. mum ....ouaacmcfiou «0 9.3344000 0 NE: 88...” «.86. 2.: 886 .38.? 98.: 8E... Th e + «a u + a a + o u z 333 ~82. Id 3&6 Sand o 9: a + o u 2 mo— Bmod £86 «.3... mmEd 83.: Ex: a + - .... .5 86; 3.86 9.: 39...... 88.. 3x: a + o u z T: floats-83 3o 32.: 396 Ned Rand- 88.. 38.? ex: 0 + a m: a + o u a no. 82.: 32.: mm... Road 89.? a 9: .... + a u 9 m3 386 335 83 28.: $3.? named «a o + a a + a u ... :86 :36 85 98.: 38.9. ~60 119 0.7483 1.5173 0.0 7.2 overall 1733 0.9573 1.1400 0.0 8.0 (b) DBH by periodic increment (PI) classes. DBH (cm) PI CLASSEStcm) N MEAN STD.DEV. MIN MAX 0.0-0.49 672 40.6786 16.3201 25.0 116.0 0.5-0.99 303 33.7789 9.3157 25.0 92.0 1.0-1.49 312 35.5929 10.3683 25.0 94.0 1.5-1.99 136 35.6618 8.9577 25.0 62.0 2.0-2.49 130 38.4615 11.4764 25.0 76.0 2.5-2.99 55 39.0727 11.5002 25.0 73.0 >3.00 125 38.7280 12.7459 25.0 91.0 overall 1733 37.8697 13.5655 25.0 116.0 CHAPTER 8 CONCLUSIONS The forest resources of the study area are declining with respect to number of trees per unit area, growing stock volume and basal area. The dominant latent root, 1.: 0.97. supports the previous affirmation. According to Enright and Ogden (1979). when the intrinsic rate of natural increase (r) is equal to zero. I. is equal 1, i.e., the population under investigation is perfectly balanced or the birth rate is equal to the mortality rate; when r > 0, A > 1, the population is increasing and the birth rate is greater than the mortality rate; and when r < 0, A < l, the population is declining or the birth rate is smaller than the mortality rate. Another reason for the decline is that mortality and ingrowth are not balanced.'These components averaged, during a 5-year period of observations, 9.18% and 3.72% respectively, in relation to the total recorded in 1980. Mortality appears to be independent of diameter size. However, the survivors are growing at a mean rate of 0.21 cm/year in dbh. The ratio Al/Az averaged 1.05, suggesting that this population will approach the ”climax” state slower than at 110 111 least two mixed temperate hardwood systems, one in Connecticut and another in New Jersey. In terms of utilizable volume stock, the forest of the study area,(contro1 plots of NR experiment) is relatively poor. The total volume including bark averaged 284 cu.m/ha for all tree species with dbh >, 25 cm. From this total, only about 40% fulfills the minimum size requirements (dbh > 40 cm) for Amazonian forest industries, and from this only about 30% qualifies as economically desirable species. Thus, the remaining utilizable volume is no more than 35 cu.m/ha. This is much less than half of the average volume reported for growing stock of North American temperate forests, Brazilian temperate forests, or even tropical moist forests in SE Asia. In contrast, the area is very important in species richness. About 350 different tree species from 52 different botanical families were identified. Lecythidaceae, Leguminosae and Sapotaceae are the dominant families contributing more than 50% of all trees tallied. Among the three hypothesized diameter distribution functions tested in this study, the Weibull PERC (distribution whose parameters were computed using the percentile approach) showed the best fit for the observed data. The Weibull MLE (the maximum likelihood approach) and the exponential distributions were very sensitive to variation within sample plots. When the Weibull shape parameter, 0, equals 1, an exponential distribution results, 112 but even in this case, the Weibull PERC fitted the observed data as precisely as the exponential function. In addition, the Weibull PERC function is very simple in estimating its parameters; it does not require sophisticated computer capabilities. The first-order Markov chain analysis allowed projection of the overstory mortality and frequency distribution by dbh classes. Since age and successive records from long-term permanent plots are not available, the Markov approach is a realistic alternative to predict the direction of future trends in the study area. Traditional growth and yield models cannot be applied to the study area since age and site index are not available. The 5-year increments did not show any indications that they can be correlated with dbh. From this study, Lotka's model appears to be a powerful alternative to replace the traditional growth and yield models. Another alternative is the equation, V(85) = a + b* 0(80) + c*(D(80)f2, which presented a very good fit for the observed data. All mathematical models developed in this study have as output the abstract model for a management strategy to be implemented in the real world. In 1990, these models should validated and, if necessary, refined based then on a 10-year period of observations. If valid, they will be very helpful to exercise the simulation as a means of determining model time response for a longer period. APPENDIX APPENDIX Floristic composition of bacia 3 by botanical family (Developed by Department of Botany of INPA) 1. ANACARDIACEAE Anacardium spruceanum Benth ex Engl. Astronium - l sni(*) Tapira retusa Ducke 2. ANONACEAE Anaxagorea - l sni Anona ambota Aubl. Bocageopsxs multiflora (Mart.) R.E. Fr. Bocageopsis - 1 sni nguetia flagelaris Huber Duguetii - l sni Ephe ranthus amazonicus R.E. Fries Ephedranthus - l sni Quatteria olivacea R.E. Fries Guatteria - l sni ggeudoxandra cariaceae R.E. Fries ggllinia insignia R.E. Fries var. pallida R.E. Fries Unono Sis - l sni Xylopia benthami R.E. Fries Xylopia - l sni 3 . APOCYNACEAE Ambelania acida Aubl. Anacampta - l sni Aspidosperma album (Vahl.) R. Ben. Aspidosperma obscurinervius Azamb. AspidOsperma carapanauba Pichon Aspidosperma - l sni gouma macrocarpa Barb. Rodr. geissospermum ar enteum R. Rodr. Himatanthus sucqua (Spruce) Woodson 4. ARALIACEAE Didymopanax morototoni (Aubl.) Decne. & Planch. * sni = species not identified for determined genus. 113 9. 114 BOMBACACEAE Bombacopsis - 2 sni's gatostemma milanezii Paula Nov. §cleronema micranthum Ducke Scleronema - l sni BIGNONIACEAE Jacaranda copaia D. Don. Jacaranda - l sni Tabebuia serratifolia (D. Don.) Nichols. BORAGINACEAE Cordia - l sni BURSERACEAE gemicrepidospermum rhoifolium (Bth.) Swart. grotium aracouchili (Aubl.) March. grotium heptaphyllum (Aubl.) March. grotium subserratum Engler Protium - 4 sni's Tetra astris unifoliata (Engl.) Cuatr. Tetragastris - 2 snI's Trattinickia - l sni CARYOCARACEAE garygcar pallidum A.C. Smith Caryocar villosum (Aubl.) Pers. 10. CELASTRACEAE Goupia glabra Aubl. 11. CHRYSOBALANACEAE gouepia leptostachya Benth. ex Hook Coue ia - 1 sni Birtella glandulosa Spreng. Licania al a (Ben.) Cuatr. canescens R. Ben. Licania Licania Licania Licania fiicania Licania Licania Eicania Licania Licania gracilipes Taub. heteromorphg hypoleuca Benth. kunthiana Hook f. latifolia Benth. ex Hook micrantha Miq. oblongifalia Standl. reticulata Prance - l sni Parinari montane Aubl. 12. COMBRETACEAE Buchenavia parvifolia Ducke Buchenavia - 2 sni s 13. 14. 15. 16. 17. 18. 19. 20. 115 CONNARACEAE Connarus - l sni DICHAPETALACEAE Tapura amazonica DUCKEODENDRACEAE Duckeodendron cestroides Ruhlm EBENACEAE Diospyros bullata A.C. Smith ELAEOCARPACEAE Sloanea - l sni ERYTHROXYLACEAE Erythroxylum - lsni EUPHORBIACEAE Anomalocalyx - l sni Qpnceveiba guianensis Aubl. Qroton lanjouwensis Jablonski Croton - l sni erpetes variabilis Vittien Qavarretia - 1 sni glycidendron amazonicum Ducke Hevea guianensis Aubl. Mabea caudata Pax. ex K. Holhm. Mabea - l sni Micrandra rossiana R.E. Schultes Micrandra siphonioides Bth. Micrandropsis scleroxylon W. Rodr. gausandra macropetala Ducke Pera - 1 sni ngonophora schomburgkiana Miers. ex Bth. FLACOURTIACEAE gasearia combaymensig Tul. gasearia ulmifolia Vahl. ex Von. Casearia - l sni Carpotroche - 1 sni Laetia procera (Poepp.) Eichl. ____,_ Ryania - l sni 21. GUTTIFERAE Qalophyllum brasiliense Camb. Carai a - l sni Clusia - l sni Bavetiopsis - l sni Moronobea coccinea Aubl. Moronobea pulchra Ducke Rheedia 2 - sni‘s S¥mphonia globulifera Linn V smia uckei Maguire 22. 23. 24. 25. 116 Vismia guianensis (Aubl. ) Choisy Tovomita - l sni HIPPOCRATEACEAE Salacia - 1 sni HUMIRIACEAE Duckesia verrucosa (Ducke) Cuatr. Endopleura uchi (Huber) Cuatr. Humiria balsam1fera (Aubl. ) St. Hill Sacoglottis ceratocarpa Ducke Sacaglottis - 1 sni Vantanea macrocarpa Ducke Vantanea parviflora Lam. Vantanea - l sni ICACINACEAE Emmotum - 2 sni's Poraqueiba - l sni LAURACEAE Aniba canelilla (B. B. K. ) Mez. Aniba duckei Kosterm. AniEa ferrea Kubitzki Aniba rosaedora Ducke Aniba terminalis Ducke Aniba - 1 sni Endlicheria - 4 sni's Licaria canela (Meissn. ) Kosterm. Licaria guianensis Aublet Licaria r1gida Kost. Licaria - 3 sni' s Mezilaurus decurrens (Ducke) Kost. Mezilaurus synandra (Mez.) Kosterm. Mezilaurus - 1 sni Nectandra rubra (Mez.) C.K. Allen Nectandra - l sni Qcotea canaliculata Mez. Ocotea neesiana (Miq.) Kosterm. Ocotea - 9 sni's 26. LECYTHIDACEAE Cariniana decandra Ducke Car1niana micrantha Ducke Corytophora alta Knuth Corytgphora rimosa Rodr. Couratari - 1 sni _schweilera fracta R. Knuth §schweilera odora (Poepp.) Miers. Eschweilera - 8 sni's Gustavia au usta L. GustavIa elliptica Mori Holopyx1dium latifolium (A. c. Smith) R. Knuth. Lecythis usitata Miers var. paraensis R. Knuth. 117 27. LEGUMINOSAE CAESALPINIODEAE Aldina hetero h 11a Spruce ex Bth. Bocoa viriaiE§Ota (Ducke) Cowan Cassia rubriflora Ducke Copaifera mu1tijug_ Hayne Elizabetha bicoior Ducke Elizabetha princeps Schomb. ex Bth. Elizabetha - 1 sni Eperua bijug_ Mart. ex Benth. var. glabriflora Ducke §p§gua duckeana Cowan Eperua schomburgkiana Benth. Hymenaea pArvifo1ia Huber ngenaea - 4 sni' s Macrolobium limbatum Spr. ex Benth. Macrolobium microcalyx Ducke Peltogyne cat1ng§g subsp. labra (W. Rodr. ) M. F.Si1va Peltogyne paniculata subsp. pgn1cu1ata Benth. Swartzia ing1folia Ducke Swartzia pgnacoco (Aubl. ) Cowan Swartzia polyghy11g D.C. §wartz1a recurva Poepp. & Endl. Swartzia reticglata Ducke Swartzia ulei Harms Swartzia - 3 sni's §g1erolobium - 1 sni Vouacapoua pallidior Ducke Tachigal1a gyrmecophilla (Ducke) Ducke TAch1gAIia pAn1cu1ata Aubl. TachigAIia - 1 sni 28. LEGUMINOSAE MIMOSOIDEAE Dimorphandra parviflora Spr. ex Bth. Diniz1a excelsa Ducke Enterolobium schomburgkii Benth. Hymenolobium - 1 sni Inga aff.brevia1ata Ducke Inga paraensis Ducke Inga cayennensis Benth. Inga - 4 sn1' s Parkia multiju a Bth. Parkia opposIE olia Spr. ex Bth. Parkia gendula Benth. ex Walp. Parkia - 2 an 's Piptaaenia pailostachya (D. C. ) Bth. Piptaden1a suaveolens Miq. g1gtadenia - l sni Pithecolobium racemosum Ducke Pithecolobium - 2 sni‘s Sgryphnodendron racemiferum (Ducke) W. Rodr. Stryphnodendron - 1 sni 29. LEGUMINOSAE PAPILIONOIDEAE Andira parviflora Ducke 30. 31. 32. 33. 34. 118 Andira unifoliata Ducke Andira - 1 sni DIpteryx alata Vogel Digterxx ma nifica Ducke D1pteryx oaorata (AubL ) Willd. D1Qteryx gppositifolia (Aubl. ) Willd. Dipteryx polyphylla (Ducke) Hub. fiigterxx - 1 sni gymenolobium sericeum Ducke gymenolobium cf. pulcherrimum Ducke gymenolobium - 1 sni Qrmosia sm1thii Rudd. Diplotrop1s - 1 sni Platymiscium duckei Huber LINACEAE Roucheria callophylla Planch MALPIGHIACEAE Byrsonima stipulacea Adr. Juss. B rsonima - 1 sni Pteranara arborea Ducke MELASTOMATACEAE Bellucia grossularioides (L. ) Triana Miconia elaeagnoides Cogn. Micon1a re e111 Cogn. Mouriria angu11costa Morley MourirIa - 1 sn1 MELIACEAE Guarea - 2 sni's Trichilia - 2 sni's- MORACEAE Brosimum guianensis Aubl. Brosimum pgtabile Ducke Brosimum pgr1nar1oides Ducke subep. parinarioides Brosimum utile (H. B. K. ) Pittier Brosimu rubescens Taub. Cecropia scyadophylla Mart. var. juranyana Snethlage C1aris1a racemosa R. et P. Cousagoua - 1 sni F cus clusiaefolia Schott F1cus guianensis Desv. Helicosty11s - 1 sni Maggira calophylla (P.A.E.) C.C. Berg. Magu1ra sclerophylla (Ducke) C. C. Berg. Nauc1e0psis caloneura (Bub. ) Ducke Naucleqpsis glabra Spruce ex Baill NAucleops1s macroghxlla Miq. Perebea mollis (P. E. ) Huber subsp. mollis Perebea mollis (P. S. C. ) Huber Pourouma ovata Trecul. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 119 Pseudolmedia - 1 sni Sorocea - 1 sni MYRISTICACEAE gpgpsoneura ulei Warb. lgyanthera - 1 sni Osteophloeum lat s ermum (A.D.C.) warb. Virola calophylla Mgf. girola carinata (Bth.) Warb. ViroIa elon ata (Bth.) Warb. Virola c . m1chelii Heckel girola multinervia Ducke girola pavonis (A.D.C.) Smith girola venosa Warb. Virola venosa (Benth.) Warb. MYRTACEAE Eu enia - 3 sni's M rc1a ma na Legrand Mxrc1a a ax (Rich.) D.C. MONIMIACEAE Siparuna dicipiens (Tul.) A.D.C. NYCTAGINACEAE Neea cf. altissima P. et E. Neea - 2 sni‘s OCHNACEAE Ouratea discophora Ducke Ouratea - 1 sni OLACACEAE Agtandra - 1 ani Qhaunochiton - 2 sni's Heisteria acumitetg (3.3.) Engl. EEISteria barbata Cuatr. . Heisteria - 2 sni's ginquartia guianensis Aubl. Ptychopetalum olacoides Benth. PROTEACEAE Rougala - 1 sni QUIINACEAE Quiina abovata Tul. Quiina - 1 sni Tonto ia guianensis Aubl. RHABDENDRACEAE Rhabdodendron amazonicum (Spr. ex Bth.) Hub. RBIZOPHORACEAE Anisophyllea manausensis Pirea 5 W. Rodr. 120 Sterigmapetalum obovatum Kuhlmann 4S. RUBIACEAE Amaioua - 1 sni Duroia fusifera Hook f. ex K. Schum Duroia - 1 sni Elaeagia - 1 sni Faramea - 1 sni Perainandusa - 2 sni' s PTlicourea anisoloba M. Arg. Palicourea cf. 1q9g1flora (Aubl. ) A. Rich. PTgamea - 1 sni Psychotria prancei Steyermark Remijia - 3 sni' s 46. SAPINDACEAE Mata ba - 1 sni M1cropholis - 1 sni Talis1a - 1 sni Toulicia - 1 ani 47. SAPOTACEAE Achrouteria Egmifera Eyma Achrouter1a - 2 sn1 s Chrysophyllum oppositum (Ducke) Ducke Chrysophyllum anomalum Piree Q1plocem venezuelana Aubr. Ecclinusa Bacuri AfiBr. et Pellegr. Ecclinusa ucugu1rana Aubr. 5 Pellegr. Ecclinusa - 2 sn1 s Pranchetella platyphylla (A. C. Sm.) Aubr. Pranchetella - 1 sni Glycoxylon pedicellatum (Ducke) Ducke Lafiétia - 4 sni' s Manilkara amazonica (Huber) Standley Manilkara huberi (Ducke) Chev. Manilkara cavalcantei Pires et Rodr. Manilkara surinamenEis (Miq.) Dubard Micropholis truncifiora Ducke Micropholis guyanensis Pierre Micropholis venulosa Pierre Micropholis rosaainha-brava Aubr. et Pellegr. Microphol1s mensalis (BTehni) Aubr. Microphol1s - S sni' s Myrtiluma eu eniifolia (Pierre) Baill Neoxxthecec uaantha (Sandw.) Aubr. Pouteria ggyanensis Aubl. L. O. A. Teixeira 82 Pradosia verticillata Ducke gr1eure11a manaosens1s Aubr. Pseudolabatié - 1 sni RafilkoferelIa - 1 sni Ra ala spuria (Ducke) Aubr. R1charde11a manaoseneis Aubr. et Pellegr. Rlchardella macrophylla (Lum.) Aubr. 48. 49. SO. 51. 52. '53. 54. 121 szxgiopsis oppositifolia Ducke SarcauIis brasiliensis (A.D.C.) Eyma SIHARUBACEAE §imaruba amara Aubl. Simaba guianensis Aubl. subsp. guianensis SImaEa cuspidata Spruce STERCULIACEAE §tercu1ia speciosa K. Schum. Sterculia - 1 sni TheoEroma sylvestris Aubl. ex Mart. STYRACACEAE Stxrax - 1 sni TILIACEAE Apeiba echinata Gaertn. Apeiba burehelii Sprague Luehea - 1 ani VERBENACEAE Vitex triflora Vahl. VIOLACEAE geonia glycicarpa Ruiz et Pav. ginorea guianensis Aubl. var. subintegrifolia Rinorea racemosa (Mart. et Zucc.) O. Ktze. égphirshox sufinamensis Eichl. VOCHYSIACEAE ggisma bicolor Ducke Erisma fuscum Ducke Qualea clavata Staflen Qualea paraensis Ducke QuaIea cassiquiarens1§ (Spr.) Warm. anlea labourianana Paula Qualea brevipediceilata Staflen Vochysia obiflénsie (355.) Ducke LIST 01? REFERENCES LIST OF REFERENCES Acevedo, M.F. 1981. On Born's Markovian Model of Forest Dynamics With Particular Reference to Tropical Forests. Theoretical Population Biology 19:230-250. Alder, D. 1979. A Distance-independent Tree Model for Exotic Conifer Plantations in East Africa. Forest Sci. 25:59-71. Alder, D. 1980. Forest Volume Estimation and Yield Prediction. Vt2 - Yield Prediction. FAD Forestry Paper 22/2. 194p. Alencar, J3C. and qu. Araujo. 1981. Incremento Anual do Pau-rosa (Aniba duckei Kostermans) Lauraceae, em Floresta Tropical Umida Primaria. Acta Amazonica 11(3):547-552. Anderson, T.W. and INA. Goodman. 1957. Statistical Inference About Markov Chains. Annals of Mathematical Statistics 28:89-110. Anton, B. 1973. Elementary Linear Algebra. John Wiley and Sons, Inc. New York. 296p. Bailey, R.L. and T.R. Dell. 1973. Quantifying Diameter Distributions With the Weibull Function. Forest Sci. 19:97-104. Barros, P.L.C., S.A. Machado, D. Burger, and J.D.P. Siqueira. 1979. Comparacao de Modelos Descritivos da Distribuicao Diametrica em uma Floresta Tropical. Floresta lO(2):l9-32. Bell, 6.8. 1976. Notes on Natural Regeneration in Dominica. Windward Islands. Commonw. For. Rev. 55(1)27-36. Bellefleur, P. 1981. Markov Models of Forest-type Secondary Succession in Coastal British Columbia. Can. J. For. RES. 11:18-29. Bierzychudek, P. 1982. The Demography of Jack-in-the- pulpit, a Forest Perennial That Changes Sex. Ecol. Monographs 52(4)::335-351. Braga, IhI.S. 1979. Subdivisao Fitogeografica, Tipos de Vegetacao, Conservacao e Inventario Floristico da 122 123 Floresta Amazonica. Acta Amazonica 9(4):53-80. Britwum, S.P.K. 1976. Natural and Artificial Regeneration Practices in the High Forest of Ghana. Ghana Forestry Journal 2:45-49. Bruner, 8.0. and J.W. Moser Jr. 1973. A Markov Chain Approach to the Prediction of Diameter Distributions in Uneven-aged Forest Stands. Can. J. For. Res. :409- 417. Brunig, E.F. 1977. The Tropical Rain Forest - A Wasted Asset or an Essential Biosphere Resource? Ambio 6(4). Budowski, G. 1976a. Why Save Tropical Rain Forests? Some Arguments for Campaigning Conservationists. Amazoniana 5(4):529-538. Budowski, G. 1976b. Los recursos naturales en el desarrolo del Tropico Humedo. Reunion sobre desarrolo regional del Tropico Americano. IICA-Tropicos. Belem, Para, Brazil. Buogiorno, J. and B.C. Michie. 1980. A Matrix Model of Uneven-Aged Forest Management. Forest Sci. 26:609-625. Chai, D.N.P. and M.P. Udarbp. 1977. The Effectiveness of Current Silvicultural Practice in Sabah. The Malaysian Forester 40(1):27-35. Cheah, Inc. 1978. Forest regeneration and development options in Peninsular Malaysia today. The Malaysian Forester 41(2):l71-175. Chiang, CAL. 1980. An Introduction to Stochastic Processes and Their Applications. Robert E. Krieger Publ. Co. Huntington, New York. Clutter, £LL. 1963. Compatible Growth and Yield Models for Loblolly Pine. Forest Sci. 9:354-371. Clutter, J.L., J.C. Fortson, L.V. Pienaar, G.B. Brister, and R.L. Bailey. 1983. Timber Management: A Quantitative Approach. John Wiley and Sons, Inc. New York. 333p. Cochrane, T.T. and P.A. Sanchez. 1980. Land Resources, Soils and Their Management in the Amazon Region: A State of Knowledge Report. In: Amazonia - Agriculture and Land Use Research - proceedings of the International Conference at CIAT in Cali, Colombia. Cohen, AHC. 1965. Maximum Likelihood Estimation in the 124 Weibull Distribution Based on Complete and on Censored Samples. Technometrics 7(4):579-588. Conover, WuJ. 1980. Practical Nonparametric Statistics. John Wiley and Sons, Inc. New York. 2ed. 493p. De Graaf, N.R. 1981. Sustained Timber Production in the South American Tropical Rainforest. Paper prepared for the workshop on Management of low fertility acid soils of the Humid American Tropics in Paramaribo. (copy). Departamento de Silvicultura Tropical. 1982. Inventario Florestal no Rio Trombetas (INPA/SBELL/ALCOA). Technical Report. 49p. Departamento de Silvicultura Tropical. 1983. Inventario Florestal da UBE de Santa Izabel. Technical Report. 35p. Departamento de Silvicultura Tropical. 1983. Inventario Florestal do Projeto Rio Arinos. Technical Report. 64p. Departamento de Silvicultura Tropical. 1983. Inventario Florestal da UHE de Balbina. Technical Report. lllp. Draper, NzR. and B. Smith. 1981. Applied Regression Analysis. John Wiley and Sons, Inc. New York. 2ed. 709p. Einsesmith, S.P. 1985. PLOTIT: User's Guide. Enright, N. and J. Ogden. 1979. Applications of Transition Matrix Models in Forest Dynamics: Araucaria in Papua New Guinea and Nothofa us in New ZeTland. Australian Journal of Ecology 4:3-g3. Ferguson, I.S. and J3W. Leech. 1978. Generalized Least Squares Estimation of Yield Functions. Forest Sci. 24: 27-42. Fox, J.E.D. 1976. Constraints on the natural regeneration of tropical moist forest. Forest Ecology Management 1:37-65. FUA - Fundacao Universidade do Amazonas. 1979. Proposta de Politica Florestal para a Amazonia Brasileira. Metro Cubico - Livros e Revistas Tecnicas Ltda. Manaus, Am. 72p. Furnival, G.M. 1961. An Index for Comparing Equations Used in Constructing Volume Tables. Forest Sci. 7:337-361. Gomez-Pompa, A., c. Vazquez-Yanes, and S. Guevara. 1972. 125 The Tropical Rain Forest: A Nonrenewable Resource. Science 177:762-765. Grossman, S.I. and J.E. Turner. 1974. Mathematics for the Biological Sciences. Macmillan Publishing Co., Inc. New York. 512p. Higuchi, N. 1981a. A silvicultura no INPA. Supl. Acta Amazonica 11(1):99-107. Higuchi, N. 1981b. Relatorio sobre o Encontro da IUFRO nas Filipinas e ”study-tour” por Filipinas, Sarawak e Fiji. Report not published. Higuchi, N., F.C.S. Jardim, J. Santos, A.P. Barbosa, and T.W.W. Wood. 1983. Inventario Florestal Comercial da Bacia 3. Acta Amazonica (in press). Higuchi, N., F.C.S. Jardim, J.Santos, and J.C. Alencar. 1985. Bacia 3 - Inventario Diagnostico da Regeneracao Natural. Acta Amazonica 15(1-2):199-233. Higuchi, N. and C.W. Ramm. 1985. Developing Bole Wood Volume Equations for a Group of Tree Species of Central Amazon (Brazil). Commonw. For. Rev. 64(1):33- 41. Hilton, R.G.B. and R.J. Johns. 1984. The future of forestry in Papua New Guinea. Commonw. For. Rev. 63(2):103-106. Horn, H.S. 1975. Markovian Properties of Forest Succession. In: Ecology and Evolution of Communities (M. Cody and J. Diamond, eds.) pp. 196-211. Harvard University Press. Cambridge, Mass. Hosokawa, R.T. 1981. Manejo de Florestas Tropicais Umidas em Regime de Rendimento Sustentado. UFPr. Technical Report. Hutchinson, I.D. 1980. Liberation Thinning: A Tool in the Management of Mixed Dipterocarp Forest in Sarawak. UNDP/FAO Field Document 8 5. Kuching. IBGE. 1977. Geografia do Brasil. Sergraf-IBGE. Vol. 1 and 4. INPA - Instituto Nacional de Pesquisas na Amazonia. 1979. Estrategias Para a Politica Florestal na Amazonia Brasileira. Suplemento Acta Amazonica 9(4). 216p. INPA - Instituto Nacional de Pesquisas na Amazonia. 1982. Bacia Modelo: Caracterizacao Ecologica de uma Bacia Hidrografica Experimental na Amazonia Central. Suplemento Acta Amazonica 12(3). 114p. 126 James, L.M. (pers. comm.). Notes of FOR 454 - Forestry in International Development. Winter/1984. Johnson, N.E. 1976. Biological opportunities and risks associated with fast-growing plantation in the Tropics. Journal of Forestry, April:206-211. Kio, P.R.o. 1976. What future for natural regeneration of tropical high forest? An appraisal with examples from Nigeria and Uganda. Commonwealth Forestry Review 55(4):309-318. Kio, P.R.O. 1979. Management Strategies in the Natural Tropical High Forest. Forest Ecology and Management 2:207-220. Lanly, J.P. 1983. Assessment of the Forest Resources of the Tropics. Forestry Abstracts (Review Article) 44(6):287- 318. Lawrence, K.D. and D.R. Shier. 1981. A Comparison of Least Squares and Least Absolute Deviation Regression Models for Estimating Weibull Parameters. Commun. Statist. - Simula. Computa. B10(3):315-326. Lee, H.S. 1982. Silvicultural Management in Sarawak. The Malaysian Forester 45(2):485-496. Leslie, A. 1977. Where Contradictory Theory and Practice Co-exist. Unasylva 29(15):2-18. Little, S.N. 1983. Weibull Diameter Distributions for Mixed Stands of Western Conifers. Can. J. For. Res. 13: 85- 88. Lisboa, P.L. 1975. Observacoes Sobre a Vegetacao das Campinas Amazonicas. Acta Amazonica 5(3):211-223. Loetsch, F., F. Zohrer, and K.E. Haller. 1973. Forest Inventory. BLV Verlagsgesellschaft mbH, Munchen. v.2. Lowe, R.G. 1978. Experience with the Tropical Shelterwood System on regeneration in natural forest in Nigeria. Forest Ecology Management 1:193-212. Miller, T.B. 1981. Growth and Yield of Logged-over Mixed Dipterocarp Forest in East Kalimantan. The Malayasian For. 44(2):419-424. Morrison, D.F. 1976. Multivariate Statistical Methods. McGraw-Hill Inc. 415p. Murphy. P.G. (pers. comm.). Notes of HOT 441 - Geographical 127 Plant Ecology. Winter/1985. Myers, N. 1982. Depletion of Tropical Moist Forests: A Comparative Review of Rates and Causes in Three Main Regions. Acta Amazonica 12(4):745-758. Myers, N. 1983. Tropical moist forests: over-exploited and under-utilized? Forest Ecology Management 6:59- 79. Palmer, J.R. 1977. Forestry in Brazil - Amazonia. Commonw. For. Rev. 56(2):115-l29. Peden, L.M., J.S. Williams, and W.E. Frayer. 1973. A Markov Model for Stand Projection. Forest Sci. 19:303-314. Pielou, ELC. 1977. Mathematical Ecology. John Wiley and Sons. New York. Prance, G.T. 1974. Phytogeographic Support for the Theory of Pleistocene Forest Refuges in the Amazon Basin, Based on Evidence from Distribution Patterns in Caryocaraceae, Chrysobalanaceae, Dichapetalaceae and Lecythidaceae. Acta Amazonica 3(3):5-28. Prance, G.T. 1980. A Terminologia dos Tipos de Florestas Amazonicas Sujeitas a Inundacao. Acta Amazonica 10(3): 495-504. Rai, S.N. 1979. Diameter/Height, and Diameter/Girth Relationship of Some Rain Forest Tree Species of Karnataka - India. The Malayasian For. 42(1):S3-58. Ranzani, G. 1979. Recursos Pedologicos da Amazonia. Acta Amazonica 9(4):23-35. Ranzani, G. 1980. Identificacao e Caracterizacao de Alguns Solos da EEST. Acta Amazonica 10(1):7-41. Revilla,.AJL 1981. A Review of Growth and Yield Studies on the Tropical Mixed Forests in the Asia Region. Paper for the IUFRO meeting on ”Growth and Yield Studies on the Tropical Mixed Forests”, in Los Banos, the Philippines. Ribeiro, M.N.G. 1976. Aspectos Climatologicos de Manaus. Acta Amazonica 6(2):229-233. Salati, E. and P.B. Vose. 1984. Amazon Basin: A System in Equilibrium. Science 225:129-138. Schubart, H.O.R. and E. Salati. 1980. Natural Resources for Land Use in the Amazon: The Natural Systems. In: Amazonia -- Agriculture and Land Use Research 128 Proceedings. Cali, Colombia. 211-259. Schreuder, H.T., W.L. Hafley, and P.A. Bennett. 1979. Yield Predictions for Unthinned Natural Slash Pine Stands. Forest Sci. 25:25-30. Sedjo, R.A. and M. Clawson. 1983. How Serious is Tropical Deforestation. Journal of Forestry Dec.:792-794. Shugart,‘HJL 1984. A Theory of Forest Dynamics: The Ecological Forest Succession Models. Springer-Verlag Inc. New York. 278p. Shugart, H.H. and D.C. West. 1981. Long-term Dynamics of Forest Ecosystems. Am. Scientist 69:647-652. Silva, M.P., P.L. Lisboa, and R.C.L. Lisboa. Nomes Vulgares de Plantas Amazonicas. INPA. 222p. Smith, V.G. 1983. Compatible Basal Area Growth and Yield Models Consistent With Forest Growth Theory. Forest Sci. 29:279-288 Soekotjo and 0.1. Dickmann. 1978. The Effect of Indonesian Selective Logging on Natural Regeneration in the East Kalimantan Rain Forest. Paper for Eighth World Forestry Congress in Jakarta, Indonesia. Soekotjo. 1981. Diameter Growth of Residual Stands in Logged Over Areas in East Kalimantan Tropical Rain Forest Indonesia. PhD Dissertation Submitted to Department of Forestry - Michigan State University. Spears, J.S. 1979. Can the Wet Tropical Forest Survive? Commonw. For. Rev. 58(3):165-180. Sullivan, AAD. and JuL. Clutter. 1972. A Simultaneous Growth and Yield Model for Loblolly Pine. Forest Sci. 18:76-86. Tang, H.T. 1980. Factors affecting regeneration methods for Tropical high forests in South East Asia. The Malaysian Forester 43(4):469-480. Tang, H.T. and W.R. Mohd. 1981. Growth and Yield Studies in Inland Mixed Indigenous Forests in Peninsular Malayasia. Paper for IUFRO meeting "Growth and Yield Studies on the Tropical Mixed Forests“, in Los Banos, the Philippines. Tosi, J. 1983. Provisional Ecological Map of the Republic of Brazil. Institute of Tropical Forestry. San Jose, Costa Rica. 129 Usher, M.B. 1966. A Matrix Approach to the Management of Renewable Resources, With Special Reference to Selection Forests. Journal of Applied Ecology 3(2): 355-367. Usher, M.M. 1969. A Matrix Model for Forest Management. Biometrics June:309-315. Usher, M.B. 1979. Markovian Approaches to Ecological Succession. Journal of Animal Ecology 48:413-426. Virtucio, F.D. and M.G. Torres. 1978. Status of Management and Utilization of Forest Resources in the Philippines. The Malaysian Forester 41(2):149-166. Volatron, B. 1976. La Mise en Valeur des Richesses Forestieres en Amazon Bresilienne et en Colombie. Revue Bois et Forets des Tropiques 165(1):59-76. Wadsworth, F.H. 1970. Review of Past Research in the Luquillo Mountains of Puerto Rico. In: A Tropical Rain Forest. Odum, H.T. and R.F. Pigeon (eds). pp. B-33 to 8.46. Walter, H. 1973. Vegetation of the Earth and Ecological Systems of the Geo-biosphere. Springer-Verlag New York Inc. 2ed. 274p. Weaver, PJL 1982. Tree Diameter Increment in Upper Montane Forest of Puerto Rico. Turrialba 32(2)119-122. West, P.W. 1980. Use of Diameter Increment and Basal Area Increment in Tree Growth Studies. Can. J. For. Res. 10:71-77. West, P.W. 1981. Simulation of Diameter Growth and Mortality in Regrowth Eucalypt Forest of Southern Tasmania. Forest Sci. 27:603-616. Wood, T.W.W. 1985. (Personal communication). British Aid Program with Honduran Forest Service. Tegucigalpa, Honduras. Zarnoch, S.J., C.W. Ramm, V.J. Rudolph, and M.W. Day. 1982. The Effects of Red Pine Thinning Regimes on Diameter Distribution Fitted to Weibull Function. MSU Agricultural Experiment Station East Lansing. RI-423. 11p. Zarnoch, S.J. and T.R. Dell. 1985. An Evaluation of Percentile and Maximum Likelihood Estimators of Weibull Parameters. Forest Sci. 31:260-268. "1111111111111111“